Upload
others
View
3
Download
0
Embed Size (px)
Citation preview
Technology Assessment Program
Report No. 10
Positron Emission Tomography
Descriptive Analysis of Experience with PET in VA
A Systematic Review Update of FDG-PET as a DiagnosticTest in Cancer and Alzheimer’s Disease
Authors: Elizabeth Adams, R.R.T., M.P.H., Management & Program AnalystKaren Flynn, D.D.S., M.S., Manager, MDRC Technology Assessment Program
Contributors: Elaine Alligood, M.L.S., MDRC Information Center LibrarianJennifer Cheslog, MDRC Data Assistant
Report Date: December 1998
MTA 98-032
The Health Services Research and Development Service (HSR&D) is a programwithin the Veterans Health Administration's Office of Research and Development.HSR&D provides expertise in health services research, a field that examines theeffects of organization, financing and management on a wide range of problems inhealth care delivery, quality of care, access, cost and patient outcomes. Itsprograms span the continuum of health care research and delivery, from basicresearch to the dissemination of research results, and ultimately to the applicationof these findings to clinical, managerial and policy decisions.
Technology Assessment ProgramManagement Decision and Research Center (152M)Health Services Research and Development Service
Office of Research and DevelopmentVA Medical Center
150 South Huntington AvenueBoston, MA 02130
Tel: (617) 278-4469 FTS: (700) 839-4469 Fax: (617) [email protected]
Released July 1999
December 1998
MTA98-032 MDRC Technology Assessment Program - PET Update - Page i
ACKNOWLEDGEMENTS
The contributions of the following reviewers are gratefully acknowledged. The MDRC takes fullresponsibility for the views expressed herein. Participation as a reviewer does not imply endorsement.
Valerie A. Lawrence, M.D. Medical Consultant, MDRC TA ProgramAssociate Professor, Dept of MedicineUniversity of Texas Health Science CenterAudie L. Murphy VA Medical CenterSan Antonio, Texas
Dr. Matthias Perleth, MPH Dept. of Epidemiology, Social Medicine and HealthSystem ResearchHannover Medical SchoolHannover Germany
David Hailey, M.D. Director, Health Technology Assessment UnitAlberta Heritage Foundation for Medical ResearchEdmonton, Alberta Canada
Paula Corabian Research Associate, Health Technology AssessmentUnitAlberta Heritage Foundation for Medical ResearchEdmonton, Alberta Canada
Mitchell Burken, M.D. Medical OfficerCoverage and Analysis GroupOffice of Clinical Standards and QualityHealth Care Financing AdministrationBaltimore, Maryland
Ron Milhorn Senior Health Insurance SpecialistCoverage and Analysis GroupOffice of Clinical Standards and QualityHealth Care Financing AdministrationBaltimore, Maryland
Captain R.K. Leedham, Jr., R.Ph., Chief Project Management StaffM.S., BCNP Division of Medical Imaging and
RadiopharmaceuticalsCenter for Drug Evaluation and ResearchFood and Drug Administration
David Pfister, M.D Associate Attending PhysicianMemorial Sloan-Kettering HospitalAssociate Professor of MedicineCornell University Medical CollegeNew York, New York
M. Elizabeth Glenn, M.D. RadiologistWomen's Health Center
December 1998
MTA98-032 MDRC Technology Assessment Program - PET Update - Page ii
Memphis, Tennessee
Lisa Hammond, M.D. Clinical Investigator, Institute for DrugDevelopmentCancer Therapy and Research CenterSan Antonio Cancer InstituteStaff OncologistAudie L. Murphy VA Medical CenterSan Antonio, Texas
James K. Stoller, M.D. Vice Chairman, Division of MedicineHead, Section of Respiratory TherapyDept. of Pulmonary and Critical Care MedicineThe Cleveland Clinic FoundationCleveland, Ohio
Dawn Provenzale, M.D. Director, GI Outcomes ResearchDuke University Medical CenterSenior Research AssociateCenter for Health Services Research in Primary CareVA Medical CenterDurham, North Carolina
Ladislav Volicer, M.D., Ph.D. Professor of Pharmacology and PsychiatryBoston University School of MedicineClinical Director/GRECCVA Medical CenterBedford, Massachusetts
The MDRC Technology Assessment Program wishes to thank Julie Lowery, Ph.D., Bonnie Bootsmiller,Ph.D., and Andrew Behler, M.P.H. of the VA Center for Practice Management and Outcomes Research inAnn Arbor for contributing VA registry data to the report.
The MDRC Technology Assessment Program also wishes to thank Maria Fonseca, Diane Hanks,Matthew Eberle, Kevin Rys, and the staff of the Information Dissemination Program for their help withthe report.
December 1998
MTA98-032 MDRC Technology Assessment Program - PET Update - Page iii
Positron Emission Tomography
PREFACE
The Veterans Health Administration (VHA) has 10 positron emission tomography (PET) imagingfacilities and shares ownership and operations with some of its academic affiliates and one with theDepartment of Defense. Significant resource commitments are associated with the acquisition andoperation of these facilities.
In 1996, the MDRC Technology Assessment Program produced a technology assessment report inresponse to a request from the Office of the Under Secretary for Health for information on VHA’sexperience with PET. The Advisory Committee for the project provided guidance on the scope andcontent of the report. The assessment reported the results of: 1) systematic reviews of clinicalapplications of PET using 2-[F-18]-2-deoxy-D-glucose (FDG) in selected cancers (head and neck, lungcancer staging, solitary pulmonary nodules, breast, and colorectal) and Alzheimer’s disease,representing conditions of importance to the veteran population, and 2) surveys of and site visits to VHAPET Centers on PET utilization, center operations, and research activities.
The MDRC found that research into the clinical utility of PET for the selected oncology conditions was inits preliminary stages. Methodological weaknesses in the published literature seriously limited thevalidity of the available evidence on the accuracy of PET as a diagnostic test, and PET’s contribution toimproving outcomes had not been systematically assessed. The lack of epidemiological information inthese studies made extrapolation of study results to defined VHA populations, and subsequent planningfor these populations, difficult.
PET is an accurate diagnostic test for dementia of the Alzheimer’s type. Studies to determine whetherthis accuracy extends to confirmed Alzheimer’s disease are under way in Europe. Nonetheless, lack ofvalid estimates of the positive predictive value of PET, parallel developments in other tests, and limitedtreatment options for Alzheimer’s disease argue for continued use of PET primarily as a research tool.Accordingly, the evidence as of September 1996 did not support widespread incorporation of PETstudies into routine diagnostic strategies for the applications included in the assessment.
The site visits and surveys confirmed that VHA has made a substantial resource commitment to its PETfacilities and that VHA researchers regard PET as an important research tool. Site investigatorsidentified a wide range of research and clinical activities in VHA PET centers, but noted that theseactivities remained largely uncoordinated. The MDRC concluded that VHA should maximize the valueof its existing commitment, rather than establish additional PET centers. This could include:
q coordinating activities of VHA PET facilities and their academic affiliates to comply with FDAregulations, to identify research areas of interest to VHA, and to design multi-center studies of highmethodologic quality;
q implementing a VHA PET registry for systematic data collection and for tracking the utility of PET inselected conditions;
q supporting rigorous, prospectively designed clinical research that expands the body of PET literaturein a methodologically sound manner; and
q submitting currently unpublished data from studies of high methodological quality for peer review.
December 1998
MTA98-032 MDRC Technology Assessment Program - PET Update - Page iv
Positron Emission Tomography1998 Update
EXECUTIVE SUMMARY
PurposeAfter the delivery of the original assessment report, the Under Secretary for Health directed theOffice of Patient Care Services to implement the assessment recommendations. VHA PETcenters collaborated on the design of the implementation process, which included initiating amulti-center VHA PET registry, supporting prospective research, and this updated systematicreview.
To produce this report the MDRC Technology Assessment (TA) Program surveyed VHA PETfacilities, used registry data, and conducted systematic reviews of the published PET literaturefrom September 1996 through December 1998 for selected cancers and Alzheimer’s disease.This report includes studies using positron emitting coincidence imaging with theradiopharmaceutical FDG to study cellular glucose metabolism.
BackgroundPET is a minimally invasive nuclear medicine imaging modality that uses the principle ofcoincidence detection to measure biochemical processes within tissues. PET may complementor supplant other imaging modalities, such as radiography, computed tomography (CT), ormagnetic resonance imaging (MRI), which rely on predominantly anatomic definitions ofdisease.
Conventional positron emission coincidence imaging is accomplished using cameras specificallydesigned, or “dedicated,” for imaging positron-emitting radioisotopes. Dual-headed gammacameras are being adapted for coincidence imaging positron emitters (called “camera-basedPET”) as a lower cost and more accessible alternative to dedicated PET. Both PET systems havewhole body scanning capability.
Key Findings
Cost and ReimbursementA dedicated PET system costs from $800,000 to $2.5 million, and a cyclotron costs from $1.2million to $1.7 million, in addition to the costs of installation, construction, and operation.Camera-based PET systems sell for about $850,000. Annual operating costs vary considerably.The charge for a PET scan will depend on these cost factors, as well as the clinical indication, theradiopharmaceutical used, and caseload.
Effective January 1, 1998 Medicare began offering interim provisional coverage for FDG-PETscans using either dedicated or camera-based PET for characterizing solitary pulmonary nodulesand initial staging of suspected metastatic non-small cell lung cancer. On or after July 1, 1999Medicare expanded coverage to include detecting and localizing recurrent colorectal cancer witha rising carcinoembryonic antigen, staging and characterizing Hodgkin’s and non-Hodgkin’slymphoma in place of a gallium scan or lymphangiogram, and identifying metastases inmelanoma recurrence in place of gallium studies.
December 1998
MTA98-032 MDRC Technology Assessment Program - PET Update - Page v
The national average payment is $1,980 per scan, excluding the professional component. HCFAwill collect and analyze claims data and data from other sources to determine the medicaleffectiveness of PET in managing these conditions, after which HCFA will decide the extent towhich it should modify the coverage policy.
RegulationRecent changes in FDA regulation now permit PET imaging facilities that manufactureradiopharmaceuticals on-site to continue in accordance with the positron emission compoundingstandards and the official monographs of the United States Pharmacopoeia. FDA has eitherapproved or cleared for marketing both PET systems to image radionuclides in the body.
Experience in VHA• VHA continues its moratorium on adding more dedicated PET facilities within the system.
Many VA medical centers are modifying dual-headed gamma cameras for coincidencedetection.
• A survey of active funded research at VHA PET sites underscores the importance of PET asa basic research tool. Most of the research is in neurology and cardiology and is funded by arange of private and public VA and non-VA sources.
• There has been an increase in the number of diagnostic PET scans, particularly in oncology.Lung cancer staging was the most common oncology indication among VHA PET sites inFY 1998.
• VHA is maximizing its investment in PET by developing a PET registry to collect criticalpatient information, funding rigorous, prospectively designed clinical research, and trackingthe published peer-reviewed PET literature available in the public domain.
• The MDRC TA Program is coordinating a joint project with other members of theInternational Network of Agencies for Health Technology Assessment (INAHTA) to producea report on the use of PET among countries represented by INAHTA members.
Evidence of effectivenessThe existing evidence argues against routine clinical use of PET for diagnosing Alzheimer’sdisease until more effective treatments and risk modification interventions for Alzheimer’sdisease are developed, and until meaningful and robust predictive values are obtained from anongoing European multicenter PET study. The systematic reviews indicate that the datasupporting the use of either dedicated or camera-based PET system with FDG in managingpatients with selected cancers are deficient.
• The evidence for using camera-based PET in oncology is limited to one small preliminarystudy in the tertiary-care setting, comparing camera-based PET to dedicated PET using nosuitable reference standard. Accordingly, it did not meet the inclusion criteria for thisreview.
• Included studies assessed dedicated PET as a complement to or replacement for anatomicimaging modalities, as a noninvasive alternative to invasive procedures, or as a method forincreasing the diagnostic certainty for performing an invasive procedure. Studies focused onthe technical feasibility of using dedicated PET and on defining diagnostic accuracy in thetertiary care setting.
December 1998
MTA98-032 MDRC Technology Assessment Program - PET Update - Page vi
• Studies generally enrolled highly selected patients and failed to adequately describe theprevious work up or the size or composition of the referral base from which the patientsample was drawn. All had at least one of the methodologic biases often found in diagnosticimaging test evaluations, and their presence will tend to inflate estimates of diagnosticaccuracy. Methods for defining disease on PET imaging have not been standardized and maylimit the generalizability of findings across institutions.
• The few studies reporting the influence of PET on changes in diagnostic certainty and/ortreatment planning were usually retrospective case series that were not originally designed todocument these changes and were not systematically conducted or reported as such. Someauthors used likelihood ratios and predictive values to define PET’s clinical usefulness, butproper interpretation of these estimates is conditioned on what was known about the patientbefore the test and on deriving PET results independently of other test results. None of thestudies met both conditions, and the influence of PET on diagnostic certainty and subsequenttreatment planning could not be determined.
Conclusions/Recommendationsq VHA continues its commitment to delivering high quality patient care and to rational
resource management through its support of VHA PET centers, carefully appraising the PETliterature to identify areas in need of research, and funding rigorous, prospective clinicalresearch.
q The prevailing evidence does not support the use of either dedicated or modified camera-based PET as a diagnostic test for the applications in this review. The TA Program identifiedseveral methodologically rigorous studies of other diagnostic imaging modalities that couldserve as models for designing higher quality PET research.
q Systematic reviews from other technology assessment agencies, which used methods similarto VHA’s, derived similar conclusions. As in VHA, patients with cancer constitute aconsiderable burden to the health systems represented by these agencies, and there is growingsupport for assessing either PET modality in the work up of these patients. Accordingly,agencies identified the uses for PET in oncology, particularly staging non-small cell lungcancer, as major topics for research.
q Several cooperative trials, including a VHA Cooperative Study of PET in solitary pulmonarynodules, are ongoing or planned. Clinicians should await the results of these efforts beforeincorporating PET into routine diagnostic strategies.
q Individuals interested in clinical PET would benefit from an accessible central repositorycontaining information on existing and proposed rigorously designed cooperative trials ofPET. This source could help guide the diffusion of PET into clinical care, as its usefulnessand contribution to improved patient outcomes are appropriately evaluated.
December 1998
MTA98-032 MDRC Technology Assessment Program - PET Update - Page vii
TABLE OF CONTENTS
I. INTRODUCTION ............................................................................................................. 1
II. DESCRIPTION OF TECHNOLOGY ............................................................................... 1A. Instrumentation............................................................................................................. 1B. Radiopharmaceutical.................................................................................................... 2C. Data Analysis................................................................................................................ 2D. Potential Roles for PET................................................................................................ 3
III. REGULATION AND REIMBURSEMENT..................................................................... 4
A. The Food and Drug Administration (FDA).................................................................. 4B. Health Care Financing Administration (HCFA) and Medicare.................................... 4
IV. ACCESS AND COST........................................................................................................ 6
V. EXPERIENCE IN VHA..................................................................................................... 6 VI. METHODS FOR THE SYSTEMATIC REVIEW ............................................................ 9
VII. APPRAISAL OF THE LITERATURE............................................................................ 11
A. Data Synthesis ............................................................................................................ 12
VIII. PUBLISHED FINDINGS ................................................................................................ 12A. Head and Neck Cancer............................................................................................... 13B. Breast Cancer.............................................................................................................. 19C. Non-Small Cell Lung Cancer..................................................................................... 26D. Solitary Pulmonary Nodules ...................................................................................... 35E. Colorectal Cancer ....................................................................................................... 40F. Alzheimer’s Disease ................................................................................................... 46
IX. ONGOING CLINICAL STUDIES AND ON-LINE RESOURCES ............................... 51
X. OTHER SYSTEMATIC REVIEWS OF PET................................................................. 52
XI. CONCLUSIONS.............................................................................................................. 54A. Experience in VA....................................................................................................... 54B. Systematic reviews..................................................................................................... 54
XII. REFERENCES ..............................................................................................................R-1
XIII. EPILOGUE.................................................................................................................... E-1
XIV. APPENDIX 1: Methods For The Systematic Review.................................................A1-1
XV. APPENDIX 2: Models Of High Quality Efficacy Studies of Diagnostic ImagingTechnologies ................................................................................................................A2-1
December 1998
MTA98-032 MDRC Technology Assessment Program - PET Update - Page viii
XVI. APPENDIX 3: Active Funded Research at VHA PET Facilities as ofOctober 1, 1998...........................................................................................................A3-1
XVII. APPENDIX 4: Data Abstraction Tables of Included Diagnostic Efficacy Studies of FDG-PET in Cancer .............................................................................................................A4-1
XVIII. APPENDIX 5: Technology Assessments of PET Produced by OtherOrganizations ............................................................................................................. A5-1
LIST OF TABLES AND FIGURESTable 1: Pricing of New PET Scan Indications Approved by HCFA................................................................................5Table 2: VHA PET Facilities and Sharing Partners .............................................................................................................7Table 3: Diagnostic-specific Utilization Data Across VHA PET Facilities for FY 1998..............................................8Table 4: Systematic Review Protocol...................................................................................................................................10Table 5: Summary of the Technical Efficacy of Camera-based PET in 31 Patients with 109 Lesions.....................12Table 6: Characteristics of Diagnostic Accuracy Studies of FDG-PET of Patients with Head and
Neck Cancer...............................................................................................................................................................16Table 7: Summary of Diagnostic Accuracy Studies of FDG-PET in Head and Neck Cancer....................................18Table 8: Characteristics of Prospective Studies of Axillary Lymph Node (N) Staging with FDG-PET in
Patients with Potentially Operable Breast Cancer...............................................................................................21Table 9: Characteristics of Studies using FDG-PET to Stage Recurrent Disease and Metastases in Patients
with Breast Cancer....................................................................................................................................................22Table 10: Summary of Diagnostic Accuracy Studies of PET and Alternatives in Breast Cancer................................25Table 11: Characteristics of Prospective Studies of Mediastinal Lymph Node (N) Staging with FDG-PET
In Patients with Potentially Operable NSCLC.....................................................................................................28Table 12: Characteristics of Prospective Studies of Distant Metastases (M) Staging with FDG-PET in
Patients with NSCLC...............................................................................................................................................30Table 13: Summary of Diagnostic Accuracy Studies of PET and Alternatives in Staging Lung Cancer...................33Table 14: Characteristics of Studies Using FDG-PET of Patients with Radiographically Indeterminate
Solitary Pulmonary Nodules ...................................................................................................................................37Table 15: Summary of the Diagnostic Accuracy and Diagnostic Thinking Efficacy Studies of PET in
Indeterminate Solitary Pulmonary Nodules (SPN) .............................................................................................39Table 16: Characteristics of Studies of Pre-operative Staging with FDG-PET in Patients with Suspected
Recurrent Colorectal Cancer...................................................................................................................................43Table 17: Summary of Diagnostic Accuracy Studies of FDG-PET in Colorectal Cancer.............................................45Table 18: Summary of Recent Technical Efficacy Studies using FDG-PET in Alzheimer’s Disease........................49Table 19: Active NIH Trials of FDG-PET in Selected Cancers and Alzheimer’s Disease...........................................51Table 20: Methodologic Quality of Diagnostic Accuracy Studies of FDG PET in Selected Cancers.......................E-2
December 1998
MTA98-032 MDRC Technology Assessment Program - PET Update - Page 1
I. INTRODUCTION
VHA is committed to improved quality of care and outcomes for veterans and to rationalresource management. As health care decision making transitions from a rationale based onresources and opinions to a rationale based on evidence from research, VHA uses technologyassessment (TA) processes and information to guide evidence-based decisions. Health ServicesResearch and Development Service, through the Management Decision and Research Center(MDRC), produces and disseminates TA information in the form of systematic reviews of theliterature. VHA uses these reviews to support clinical policy and focus on areas in need offurther research.
For example, after delivery of the original MDRC PET technology assessment (Flynn, 1996), theUnder Secretary for Health directed the Office of Patient Care Services to implement theassessment findings and recommendations. As a result, VHA continued its moratorium onadding more dedicated PET scanners to its system. A new VHA cooperative study incorporatedstudy design suggestions from the initial assessment. VHA PET Center Directors wereinstrumental in designing the implementation strategies, which included initiating a multi-centerVHA PET registry, completing a rigorous single-site outcome study, and updating the 1996MDRC PET systematic review.
In this update, the MDRC used evidence-based medicine frameworks and methodology toproduce systematic reviews of the peer-reviewed PET literature from September 1996 throughDecember 1998. It reviews the performance of dedicated PET systems and gamma camerasystems with coincidence detection capabilities in selected cancers of the head and neck, breast,and colo-rectum, lung cancer staging, solitary pulmonary nodules, as well as Alzheimer’sdisease. The report also contains:
• clinical and research experience across VHA PET facilities;• VHA implementation strategies for recommendations made in the first report;• ongoing multi-site clinical trials of PET for the indications reviewed in the report;• findings and recommendations from reviews of PET conducted by other technology
assessment agencies; and• a description of an international collaboration studying PET use among countries represented
by the collaboration.
II. DESCRIPTION OF TECHNOLOGY
A. Instrumentation
Positron Emission Tomography (PET) is a minimally invasive nuclear medicine imagingmodality that uses radiopharmaceuticals to capture and measure biochemical processeswithin tissues. PET, like other nuclear medicine techniques, defines disease in terms ofquantifiably abnormal regional chemistry. PET may complement other imagingmodalities, such as radiography, computed tomography (CT), or magnetic resonanceimaging (MRI), which rely on predominantly anatomic definitions of disease.
PET imaging employs radioactive isotopes that decay by emitting a positively chargedelectron, called a positron, from the nucleus. The positron collides with a negatively
December 1998
MTA98-032 MDRC Technology Assessment Program - PET Update - Page 2
charged electron resulting in two high energy (511 keV) photons that travel in oppositedirections. PET uses the principle of coincidence detection to form the raw image. Thatis, radiation detectors are arranged in a ring around the patient to allow for simultaneous(coincidence) detection of the two photons. The exact site of origin is recorded, and across-sectional image is displayed.
Dedicated PET systems are optimized for high energy dual photon coincidence detection.Two modified forms of single photon emission computed tomography (SPECT) are nowavailable for imaging positron emitters and may be a less costly alternative to dedicatedPET (Jarrit and Acton, 1996):
• dual-headed SPECT cameras adapted for coincidence detection, called “camera-based” PET, or
• multi-headed SPECT cameras adapted with special collimators for high energy(511keV) photon absorption.
Both Jarrit and Acton (1996) and Coleman (1997) emphasized that neither modifiedSPECT system is optimized for clinical use, particularly in oncology. Lower sensitivityrestricts their use to studies using isotopes with longer half-lives, and performance andcost data comparing either system to dedicated PET are limited. These authors cautionagainst the premature use of these systems, which could be detrimental to the futureacceptance of both dedicated PET and modified PET systems.
In light of recent federal regulatory changes (See Section 111-Regulation andReimbursement) this report will address only dual-headed gamma cameras adaptedfor coincidence imaging (“camera-based” PET) and dedicated PET systems.
B. Radiopharmaceutical
The most widely used radiopharmaceutical in PET imaging is the cyclotron-producedFDG. FDG is a D-glucose analog used to study cellular glucose metabolism. Sincemany diagnostic PET studies rely on FDG, its availability is critical to a facility thatwishes to conduct clinical studies using either dedicated or camera-based PET systems.
C. Data analysis
PET and other nuclear medicine image patterns represent spatial and temporalarrangements of the physiological or biochemical process under investigation. There aremany ways to detect and compare these patterns such as visual analysis of metabolicpatterns, region of interest (ROI) analysis where the regions are hand-drawn or placed(sometimes with coregistration with anatomic images), and neural networks. PET datamay be managed by using absolute metabolic values or by normalizing to a referencevalue to generate metabolic ratios.
D. Potential roles for PET
Flynn (1996) summarized the general rationale for the use of PET in oncology. PET maydetect abnormalities in tissue biochemical and physiological processes caused by manyforms of cancer. Reliance on tumor histology and anatomy limits the oncologist’s tools
December 1998
MTA98-032 MDRC Technology Assessment Program - PET Update - Page 3
for selecting optimal treatment, and adding metabolic data from PET may expand theoncologist’s ability to optimize treatment. Finally, monitoring metabolic responses totreatment could allow early redirection of therapy. Several potential applications for PETin oncology were noted:
• Detecting tumors (which may employ coregistration techniques that combine PETand anatomic imaging into a single image);
• Staging (particularly using whole-body imaging methods) although there is a lowerlimit to the size of metastases that can be detected by PET;
• Detecting local disease recurrence, since anatomically-based imaging is often limitedby the effects of treatment;
• Predicting tumor response to chemotherapy; and• Monitoring treatment.
Studies of Alzheimer’s disease and other neurologic and psychiatric conditions predatestudies of PET for other diagnostic applications and are prevalent in the PET literature.PET allows qualitative and quantitative evaluation of cerebral physiology and explorationof the biochemical bases for clinical diseases. FDG PET brain studies have been used formany research and clinical purposes related to the central nervous system. These include(Hoffman, 1993):
• defining the magnitude and distribution of normal local cerebral glucose metabolism,and the effects of age and sex on metabolism;
• locating seizure foci in patients with partial complex seizures who are potentialsurgical candidates for temporal lobectomy;
• assessing brain tumors for degree of malignancy at diagnosis, persistent postoperative tumor, differentiating high- from low-grade tumors and radiation necrosisfrom persistent tumor;
• evaluating schizophrenia, affective disorders, obsessive-compulsive disorder;• studying cerebral metabolism in cerebrovascular disease; and• defining regions of altered glucose metabolism in various forms of dementia such as
Alzheimer’s disease, Pick’s disease, and Huntington’s disease.
Expanded roles for PET in selected applications will be discussed in Section VIIIPublished Findings for each application.
December 1998
MTA98-032 MDRC Technology Assessment Program - PET Update - Page 4
III. REGULATION AND REIMBURSEMENT
A. The Food and Drug Administration (FDA)
FDA has either approved or cleared for marketing dedicated PET scanners and coincidentimaging gamma cameras to image radionuclides in the body. To date, the FDA hasapproved two PET radiopharmaceuticals for clinical use:
• Rubidium (82Rb), limited to rest alone or rest with pharmacologic stress PET scans, isused for noninvasive imaging of the perfusion of the heart for the diagnosis andmanagement of patients with known or suspected coronary artery disease.
• FDG indicated for identifying regions of abnormal glucose hypometabolismassociated with foci of epileptic seizure. Approval for use is restricted to TheMethodist Medical Center in Peoria, Illinois.
In the Food and Drug Modernization Act, which was signed into law on November 21,1997, Congress directed the FDA to develop new approval procedures and appropriatecurrent good manufacturing practice requirements for PET drug products. FDA may notrequire the submission of new or abbreviated new drug applications for PET drugproducts, which are not adulterated, for a period of 4 years after the date of enactment ofthe Modernization Act or for 2 years after FDA develops the new procedures, whicheveris longer. FDA has begun developing these procedures.
In the meantime, PET drug products may be manufactured for clinical use providing theyare produced in accordance with the positron emission compounding standards and theofficial monographs of the United States Pharmacopoeia. These standards are to assurethat PET drug products are safe and have the identity, strength, quality, and purity thatthey are represented to possess.
B. Health Care Financing Administration (HCFA) and Medicare
A health technology review conducted by the Center for Practice and TechnologyAssessment (formerly the Office of Health Technology Assessment), Agency for HealthCare Policy and Research (1998) provided the basis for Medicare’s first coverage policyfor PET scans performed on or after March 14, 1995 (HCFA, AB972760):
• PET scans using Rubidium (82Rb), done at rest or with pharmacological stress, fornoninvasive imaging of the perfusion of the heart for the diagnosis and managementof patients with known or suspected coronary artery disease. Coverage is limited toPET scans used in place of SPECT or following an inconclusive SPECT scan, whichprovide information deemed necessary to determine treatment intervention.
In an agreement with the Chairman of the Senate Appropriations Committee in late 1997,the Secretary of Health and Human Services committed to expanding Medicare coverageof PET scans on an interim basis to include diagnosing solitary pulmonary nodules andinitial lung cancer staging (Stevens, 1997). Effective January 1, 1998, FDG-PET scanswill be covered when performed using either dedicated or camera-based PET system toimage radionuclides in the body for the following conditions (HCFA, 3b4120):
December 1998
MTA98-032 MDRC Technology Assessment Program - PET Update - Page 5
• characterizing solitary pulmonary nodules (SPNs) for the primary purpose ofdetermining the likelihood of malignancy to plan treatment. Coverage is limited toclaims that include evidence of the initial detection of a primary lung tumor, usuallyby CT.
• initial staging of suspected metastatic non-small cell lung cancer in thoracic(mediastinal) lymph nodes in patients with pathologically confirmed primary lungtumor, but whose extent of disease has not yet been established. Coverage is limitedto claims that include evidence of confirmed primary tumor, concurrent CT, andfollow-up lymph node biopsy.
The use of routine biopsy following a negative PET scan is considered inappropriate inthese conditions, and payment for biopsy will be denied unless the claim is supported byevidence explaining the medical necessity of the biopsy.
After an expedited review of scientific information presented at a town hall meeting inJanuary 20-21, 1999, HCFA agreed to expand coverage for PET scans performed on orafter July 1, 1999 to diagnose and manage the following three indications:
• detecting and localizing recurrent colorectal cancer with rising carcinoembryonicantigen (CEA);
• staging and characterizing both Hodgkin’s and non-Hodgkin’s lymphoma in place ofa gallium scan or lymphangiogram; and
• identifying metastases in melanoma recurrence in place of gallium studies prior tosurgery.
Table 1: Pricing of New PET Scan Indications Approved by HCFA*
HCPCSCodes Description
National Average Paymentfor Technical Component**
G0125 PET lung imaging of solitary pulmonary nodules usingFDG, following CT
$1,980
G0126PET lung imaging for initial staging of solitarypulmonary nodules using FDG, following CT or ofpathologically diagnosed non-small cell lung cancer
$1,980
G0163 PET, whole body, for recurrence of colorectal orcolorectal metastatic cancer
$1,980
G0164 PET, whole body, for staging and characterizinglymphoma $1,980
G0165 PET, whole body, for recurrence of melanoma ormelanoma metastatic cancer $1,980
*From www.hcfa.gov/pubforms/14%5Fcar/3b4120.htm**technical component only, including payment for radiotracer, using revenue code 404. Claims for professional componentshould user modifier 26.
Medicare coverage is conditioned on the ability of PET to affect the management andtreatment of patients with these cancers. HCFA will collect and analyze claims data, anddata from other sources, to determine the medical effectiveness of PET in managing theseconditions. After sufficient claims data have been collected, HCFA will decide the extentto which it should modify the coverage policy.
December 1998
MTA98-032 MDRC Technology Assessment Program - PET Update - Page 6
IV. ACCESS AND COST
The Institute for Clinical PET (1999) reports that there are nearly 147 facilities with coincidencedetection capability in the United States. There are 10 dedicated PET facilities in the VHAsystem, making VHA one of the largest owners of dedicated PET scanners by any single healthsystem in the world.
ECRI (1996) reports that the cost of a PET scanner ranges from $800,000 to $2.5 million,excluding costs associated with installation, construction, and operation, and a cyclotron costsfrom $1.2 million to $1.7 million. Annual operating costs vary considerably and may includepersonnel salaries, scanner and cyclotron supplies, service and maintenance contracts, equipmentamortization, and other indirect costs. Ultimately, what a PET facility charges for a PET scanwill depend on these factors, as well as the clinical indication, the radiopharmaceutical used, andcaseload (Flynn, 1996).
Currently, there is a moratorium on adding PET facilities in VHA. Many VHA medical centerswithout access to PET facilities are adapting gamma cameras for coincidence imaging. The costof upgrading dual-headed gamma cameras for coincidence imaging is approximately $250,000;dual-headed gamma cameras without the upgrade sells for about $600,000 (ECRI, 1996).
V. EXPERIENCE IN VHA
Table 2 lists VHA PET (dedicated) sites and their sharing partners. In all but two sites, both thecamera and cyclotron are in the same location. However, ownership of the camera and cyclotronvaries across sites (Flynn, 1996). All sites have access to FDG.
December 1998
MTA98-032 MDRC Technology Assessment Program - PET Update - Page 7
Table 2: VHA PET Facilities and Sharing Partners
VHA PET Facility VISN Facility Location Sharing PartnerVA Connecticut Health Care SystemVAMC West Haven, Connecticut 1 VAMC Yale University
VA West New York Health Care SystemVAMC Buffalo, New York
2 VAMC(cyclotron at sharing partner)
State University of New York atBuffalo
VA Pittsburgh Health Care System VAMCPittsburgh, Pennsylvania 4 Sharing Partner UPMC Health Systems- Presbyterian
Richard L. RoudabushVAMC Indianapolis, Indiana
11 Sharing partner Indiana University
VAMC Ann Arbor, Michigan 11 Sharing Partner University of Michigan Ann Arbor
VAMC Minneapolis, Minnesota 13 VAMC None
St. Louis VA Medical CenterSt. Louis, Missouri 15 Sharing Partner St. Louis University
VA South Texas Health Care System VAMCSan Antonio, Texas 17 Sharing Partner-UTHSC University of Texas Health Science
Center
VA Palo Alto Health Care SystemVAMC Palo Alto, California 21
VAMC(no cyclotron, FDG purchased fromprivate source)
None
VA Greater Los Angeles Healthcare SystemVAMC West Los Angeles, California 22 VAMC Individual investigators
Research continues to constitute considerable activity conducted at VHA PET facilities. AllVHA PET facilities were surveyed for a list of active funded research at their site. The results ofthis survey are listed in Appendix III. Most are multi-year studies with funding from a range ofprivate and public VA and non-VA sponsors. The majority of funded PET research is for thestudy of neurologic conditions, followed by studies in cardiology.
The VA HSR&D Center for Practice Management and Outcomes Research, Office of Researchand Development, provided FY 1998 utilization data from the VHA PET registry for theconditions in this report (See Table 3). Of the subjects that had radiopharmaceutical dataavailable, nearly 70% were scanned using FDG, representing the radiopharmaceutical most oftenused across VHA PET sites.
Given the significant burden lung cancer represents in both the veteran and general populations,not surprisingly lung cancer was the major oncology diagnosis among VHA PET sites in FY1998. Alzheimer’s disease, colorectal cancer, and head and neck cancer have roughly equivalentnumbers of veteran and non-veterans scanned, whereas non-veterans comprise a higher portionof subjects with breast cancer, as expected. The distribution of veterans and non-veterans withinand across diagnoses may change as evidence of PET’s clinical utility is clarified, or ifreimbursement policies in either the public or private sector are altered.
December 1998
MTA98-032 MDRC Technology Assessment Program - PET Update - Page 8
Table 3: Diagnostic-specific Utilization Data Across VHA PET Facilities forFY 1998
Diagnosis # Veterans # Non-veterans Total (% of all neurology subjects)Alzheimer’s disease 11 6 17 (3.4%)
Diagnosis # Veterans # Non-veterans Total (% of all oncology subjects)Lung cancer 246 192 438 (29.4%)*
Colorectal cancer 63 80 143 (9.5%)**
Breast cancer 1 34 35 (2.3%)
Head & neck cancer 58 52 110 (7.4%)
*excludes 8 patients with unknown veteran status**excludes 2 patients with unknown veteran status
In the 1996 assessment, the TA Program recommended that VHA maximize the value derivedfrom its existing commitment, rather than invest in additional PET centers, and suggested waysin which PET activities could be coordinated across the VHA system (See Preface). Since then,several suggestions have been implemented:
Develop and maintain a VHA PET registry
The VHA Office of Patient Care Services is providing recurring funding to the HSR&DCenter for Practice Management and Outcomes Research in Ann Arbor, Michigan todevelop and maintain a VHA PET registry. The Center is collecting annual facilityutilization data and subject-specific data from all VHA PET facilities.
Support rigorous, prospectively designed clinical research
• The VHA Office of Patient Care Services is providing funding to the VHACooperative Studies Center and to the PET Center in West Haven, Connecticut tocomplete an outcome analysis. The study addresses clinical utility, cost, utilization ofother diagnostic studies, and the impact of PET on treatment planning.
• VHA Cooperative Studies Program is funding a multi-year cooperative trial toevaluate the clinical utility of PET in characterizing solitary pulmonary nodules (SeeAppendix III, St. Louis). The Palo Alto Cooperative Studies Coordinating Center ismonitoring the study. Six VHA PET sites and four non-VHA PET sites with VAaffiliation are participating. Patient accrual started in August, 1998.
Results from these studies should clarify the evidence on the utility of FDG-PET in themanagement of patients with selected clinical conditions.
Conduct regular updates of the PET literature
The VHA Office of Patient Care Services also agreed to fund regular systematic reviewupdates of the 1996 MDRC PET Technology Assessment.
December 1998
MTA98-032 MDRC Technology Assessment Program - PET Update - Page 9
VI. METHODS FOR THE SYSTEMATIC REVIEW
Information about the value of PET scanning in selected cancers and Alzheimer’s disease wasobtained by conducting a systematic review of the published literature. A systematic review usesa scientific approach to limit bias and to improve the accuracy of conclusions based on theavailable data. A systematic review addresses a focused clinical question, uses appropriate andexplicit criteria to select studies for inclusion, conducts a comprehensive search, and appraisesthe validity of the individual studies in a reproducible manner. With respect to the diagnostictest literature, the point of a systematic review can be to examine the ultimate value or benefitderived from the test (Guyatt, 1995).
The MDRC uses a review protocol to guide the inclusion, analysis, and summary of evidence forthis review (See Table 4 and Appendix 1). The protocol uses three analytic frameworks toappraise the literature, ensuring that studies are evaluated in a consistent, reproducible manner,and that studies included in the report conform to established scientific standards. Theseframeworks are critical to understanding the report analysis, conclusions, and recommendations.
Assign to Fryback and Thornbury hierarchical model of diagnostic efficacy
Fryback and Thornbury (1991) note that the localized view of the goal of diagnosticradiology would be that it provides the best images and the most accurate diagnosespossible. A more global view recognizes diagnostic radiology as part of a larger systemof medical care whose goal is to treat patients effectively and efficiently. Viewed in thislarger context, even high-quality images may not contribute to improved care in someinstances, and images of lesser quality may be of great value in others.
Fryback and Thornbury (1991; 1992) present an evolving hierarchical model forassessing the efficacy of diagnostic imaging procedures. Their model, with a list of thetypes of measures that appear in the literature at each level in the hierarchy, is presentedin Appendix I. Using this model, it is possible to follow the development of a diagnostictechnology and to align current research efforts with a particular level of development.
Assess the quality of individual studies of diagnostic tests using evidence-basedmedicine criteria
This assessment has adopted evidence-based medicine criteria as a requirement forassignment of studies to the “diagnostic accuracy” level of the hierarchy. These criteriawill be applied to individual studies in the report. If the criteria are not met, the studywill generally be considered insufficiently rigorous to provide the basis for patient caredecisions. However, such studies often provide useful information on the technicalcharacteristics of a diagnostic test or may provide information necessary to subsequentdiagnostic accuracy studies.
Evaluate the strength of the evidence supporting a causal link between the use of thetechnology and improved outcomes of care
Recommendations about the use of a technology should be linked to the quality of theavailable evidence, which ultimately depends on the strength of the evidence. Thestrength of the evidence relates to the overall research design and to the quality of the
December 1998
MTA98-032 MDRC Technology Assessment Program - PET Update - Page 10
implementation and analysis, i.e. how well bias and confounding factors are controlled inthe design and conduct of a study. Attributes that strengthen the validity of findingsinclude: randomized (vs. nonrandomized), controlled (vs. uncontrolled), blinded (vs.unblinded), prospective (vs. retrospective), large (vs. small), multi-site (vs. single site),and contemporaneous (vs. historical) controls.
Table 4: Systematic Review Protocol
1) Conduct search of MEDLINE and other databases. Also search end references from retrieved articles and listings of English language, public domaintechnology assessments.
2) Apply inclusion criteria to search:• English language articles reporting primary data and published in a peer review journal (not abstracts)• studies > 12 human subjects (not animal studies) with the disease of interest• studies using dedicated PET systems or gamma camera systems adapted with 511 keV coincidence imaging capability• studies using the radiopharmaceutical 2-[18F]fluoro-2-D-glucose (FDG)• study not duplicated or superseded by subsequent study with the same purpose from the same institution• study design and methods clearly described (i.e. sufficient information to judge comparability of case and control groups, details of imaging
protocol, whether visual or quantitative analysis of PET data used, or type of PET quantitative data analysis used)
3) Retrieve full text articles meeting inclusion criteria.
4) Review full text articles and assign to level of Fryback and Thornbury (1991) diagnostic efficacy hierarchy.
5) To assess methodologic quality, apply evidence based medicine criteria to studies of diagnostic tests:• clearly identified comparison groups, ≥ 1 of which is free of the target disorder.• either an objective diagnostic standard (e.g. a machine-produced laboratory result) or a contemporary clinical diagnostic standard (e.g. a
venogram for deep venous thrombosis) with demonstrably reproducible criteria for any subjectively interpreted component (e.g., report of better-than-chance agreement among interpreters).
• interpretation of the test without knowledge of the diagnostic standard result (no test review bias).• interpretation of the diagnostic standard without knowledge of the test result (no diagnostic review bias).
6) To further refine judgment of methodological quality, grade diagnostic accuracy or thinking efficacy studies:Grade A - Studies with broad generalizability to a variety of patients and no significant flaws in research methodsGrade B - Studies with a narrower spectrum of generalizability, and with only a few flaws that are well described (and impact on conclusions can be
assessed)Grade C - Studies with several methods flaws, small sample sizes, incomplete reporting or retrospective studies of diagnostic accuracyGrade D- Studies with multiple flaws in methods, no credible reference standard for diagnosis, evidence of work up, test review, or diagnostic review
bias, or opinions without substantiating data
7) Evaluate quality of studies at each efficacy level; conduct meta analyses if appropriate.8) Rank the evidence for the degree to which it supports a causal link between technology use and improved outcomes.
Modifications made to the grading system accounted for the degree to which bias could bereasonably minimized in the study design, given the nature of the clinical work up. More
December 1998
MTA98-032 MDRC Technology Assessment Program - PET Update - Page 11
common methods for minimizing the effects of bias are described in Appendix I. If the studyprovides evidence that the investigators reduced the effects of bias, the methodologic qualitygrade was advanced to the next highest level.
It should be noted that inclusion criteria could influence report findings. The inclusion criteriachosen for this report permit review of the best evidence available on the clinical use of FDGPET scans for selected conditions. These generally represent larger controlled studies publishedin the peer-reviewed literature. A limitation of this analysis is the potential language bias owingto including only English language articles. Thus, the reader should keep in mind that thefindings and recommendations are based only on evidence that meets criteria for inclusion in thereport.
VII. APPRAISAL OF THE LITERATURE
For this update, titles and abstracts of 474 references were screened. Sixty-four references weredetermined to be relevant, and their full text articles were reviewed for potential inclusion in thesystematic review. Additional articles were retrieved to provide background materials about thetechnology and selected clinical applications.
Forty-seven articles from the database searches and from end references of initially retrievedarticles met the inclusion criteria for review. Each included study was classified according toclinical condition and assigned to a diagnostic efficacy level as follows:
Efficacy level*
Head
& N
eck
Brea
st
Lung
sta
ging
SPN
Colo
rect
al
Alzh
eim
er’s
Technical 4 4 7 1 2 8Diagnostic accuracy 3 6 7 2 3 0Diagnostic thinking ? ?Therapeutic ? ?Patient outcomeSocietal
* Adapted from Fryback and Thornbury, 1991? Anecdotal data also presented in diagnostic accuracy studies.
In all oncology areas, higher levels of studies in the diagnostic test hierarchy supersededtechnical efficacy (feasibility) studies, represented the best evidence on the efficacy of FDGPET, and were summarized for this review. Technical efficacy studies are listed in thereferences. In Alzheimer’s disease, only technical efficacy studies met the inclusion criteria forreview.
All but one of the included studies were single-site studies classified as case series (Level Vevidence), representing a relatively weak study design that does not provide strong evidence ofeffectiveness. Case series contain useful information about the clinical course and prognosis ofpatients, can suggest relationships between interventions and outcomes, and can generate ideasfor further research. All studies used patients with no disease or with benign disease as internalcontrols.
December 1998
MTA98-032 MDRC Technology Assessment Program - PET Update - Page 12
All included studies used dedicated PET systems. The TA Program identified only onepreliminary study using camera-based PET in oncology (Shreve, 1998). These authors comparedblinded readings of camera-based PET images, using attenuation-corrected dedicated PET as thestandard of reference, in 31 patients with known or suspected tumors. Accordingly, it did notmeet criteria for inclusion in this review. The results are summarized below.
Table 5: Summary of the Technical Efficacy of Camera-based PETin 31 Patients with 109 Lesions
Site Short-axis diameter (cm)Range, mean
# lesions detected oncamera-based PET
# lesions detected ondedicated PET
Lung 0.9-4.0, 2.7 13 14
Mediastinum 0.6-1.3, 1.0 5 15
Mediastinum 1.5-3.5, 2.2 15 16
Axilla 1.2-1.5, 1.3 5 9
Head and neck 1.1-2.4, 1.7 5 7
Abdomen 1.2-6.3, 2.8 6 26
Skeleton Not available, could not bedetermined 11 22
The authors concluded that camera-based FDG PET could depict many of the lesions depictedwith dedicated PET. Detection of lesions using camera-based PET was greatest in the lung andpoorest in the abdomen and in all sites, excluding the lungs, for tumors generally less than 1.5cm in short-axis diameter. The results of this preliminary study require valid estimates ofdiagnostic accuracy and marginal value using an appropriate reference standard in order toestablish camera-based PET as a diagnostic tool.
A. Data Synthesis
This report presents a qualitative overview to synthesize the best available evidence. Aquantitative synthesis (meta-analysis) was not attempted. The methodologicalweaknesses of case series, combined with present differences in design and analysisamong the eligible studies, argued against the validity and usefulness of pooling studyresults (Eysenck, 1994).
VIII. PUBLISHED FINDINGS
Background information on each clinical condition such as risk factors, diagnosis, alternativediagnostic modalities, staging, treatment and survival was described in detail in the first MDRCPET report (Flynn, 1996), and will not be presented here. A brief synopsis of updatedepidemiological information and an account of the potential role(s) for PET are presented foreach condition in addition to critical evaluation of the literature.
Epidemiological information for oncology conditions in this report is supplied by the AmericanCancer Society (American Cancer Society, 1998). Data on the veteran population are providedby the 1997 Annual Report of the Secretary of Veterans Affairs (West, 1998).
December 1998
MTA98-032 MDRC Technology Assessment Program - PET Update - Page 13
Results are presented according to the potential role of PET in the management of each disease.Full data abstraction tables of the best evidence of PET for each cancer section are found inAppendix IV.
A. Head and Neck Cancer
This report will define head and neck cancer as the common squamous cell carcinomas ofthe oral cavity, nasal cavity and paranasal sinuses, pharynx, and larynx. Skin, brain,thyroid, and salivary gland tumors and the rare tumors of other histopathologic types(sarcomas and lymphomas) that can have primary sites in the head and neck will not bediscussed.
Approximately 41,400 new cases of head and neck cancer (3% of all incident cases of alltypes of cancer) and 12,300 deaths (2% of all cancer-related deaths) attributed to headand neck cancer are estimated for the United States in 1998. Within Veterans HealthAdministration malignant neoplasms of the lip, oral cavity, and pharynx (not larynx)accounted for 2,259 total discharges (0.3% of all discharges), with an average length ofstay of 18.5 days, in FY 1997.
Nearly one-third of patients with head and neck cancer has lower stage, confined diseaseat diagnosis. Most of the remaining patients have locally or regionally advanced diseaseincluding spread to lymph nodes in the neck. Less frequent is head and neck cancer thathas metastasized beyond the neck region (e.g., brain, lung, bone, or liver), at initialdiagnosis. Accordingly, standard therapy emphasizes local and regional approaches(surgery, radiation therapy, or combination) with curative intent.
Chemotherapy is increasingly being added to standard therapy to improve the outcome ofpatients with locally advanced disease (PDQ®; 1999). For resectable diseaseneoadjuvant chemotherapy is incorporated into many organ preservation strategies toshrink tumors preoperatively and may improve locoregional control. Organ preservationapproaches using concomitant chemotherapy with radiation are advocated in patientswith unresectable disease.
Diagnostic tests are used at several points in the initial work up and treatment of head andneck cancer. These include delineating disease at the primary site (including locatingunknown primary), identifying early nodal metastases, monitoring results of treatment,and identifying persistent and recurrent disease. CT and MRI have improved detection ofoccult cervical metastases for patients with head and neck cancer and subsequentmanagement of patients at high risk of cervical metastases.
However, improvements are still needed to define the primary site and in the other pointsin the work up mentioned above. The ability to assess response to chemotherapy-radiation organ preservation approaches is becoming increasingly more important, sincesurgical excision would be indicated in the event of treatment failure. The functionalinformation on glucose metabolism in head and neck tumors supplied by FDG PET couldbe clinically useful.
Table 6 depicts the study elements and Table 7 summarizes the data and quality ofindividual studies of PET using FDG in head and neck cancer.
December 1998
MTA98-032 MDRC Technology Assessment Program - PET Update - Page 14
Detecting unknown primaries in patients with metastatic cervical nodes
Braams (1997), a small technical feasibility study, detected unknown primaries in13 patients with various histologic types of cervical metastases (see referencelist). They performed whole-body PET followed by endoscopy, after physicalexam and MRI and/or CT of the head and neck area failed to detect the primarytumor. PET identified the primary tumor in four (30%) patients and missed onesmall tumor (4mm) in another. Follow up over 18 to 30 months revealed noprimary lesion in the remaining eight patients. The authors suggested that PETmay be useful in guiding endoscopic exam and in identifying the primary site todirect more appropriate treatment.
Detecting primary disease
The MDRC Technology Assessment Program was unable to locate any PETstudies that met evidenced-based criteria for diagnosing primary disease.
Detecting cervical node metastases
Two studies in Table 6 met some of the evidence-based medicine criteria fordiagnostic test evaluations. Wong (1997) evaluated 16 patients, who had neckdissections, from a consecutive case series of 54 patients with known primarydisease or with suspected recurrence or residual disease. Data suggestcomparable performance of PET to anatomic imaging and improved performanceover clinical exam across patients with a range of stages, but a test of statisticalsignificance was not reported. In a small number of patients with occult nodal(N0) disease, PET did not perform as well as in patients with more advanceddisease. In addition to small sample size in the subgroup analyses, several aspectsof the study design were either unclear or not reported making the efficacy of PETdifficult to determine.
In a retrospective evaluation of 14 patients with N0 disease on clinical exam,Myers (1998) reported a trend of increased accuracy of PET, although notstatistically significant, over CT. PET combined with CT showed even greaterimprovement. Data were analyzed by dissected side and not by patient, andimportant study design elements were not reported.
December 1998
MTA98-032 MDRC Technology Assessment Program - PET Update - Page 15
Monitoring treatment response
Lowe (1997) presented preliminary data on 28 consecutive patients with advancedhead and neck cancer, who were enrolled in a neoadjuvant organ-preservationprotocol, to assess PET in evaluating tumor response to chemotherapy. Themethods were reasonably well described, and the study met all evidence-basedmedicine criteria for diagnostic test evaluations. The data suggest good faceaccuracy of PET in distinguishing complete response from residual disease. Wideconfidence intervals reflect a small study size, and no comparison data werepresented.
The authors commented that while a positive PET scan may be indicative ofresidual tumor and warrant repeat tissue sampling or resection, a negative PETscan may also call for tissue sampling to rule out false negative results. They alsostated that PET may be used in situations when sampling bias is more likely, forexample, difficult access, questionable post-therapy biopsy results, or normal,reepithelialized appearance of the tumor site post-therapy.
Detecting recurrent disease
Wong and associates (1997) assessed PET prospectively for detecting bothprimary site recurrence in 12 patients and nodal recurrence in 13 post-treatmentpatients. PET showed high sensitivity in detecting recurrence at the primarysite, but they presented no comparison data. For detecting nodal recurrence,PET was more sensitive than CT or MRI, was equal to clinical exam, and hadsuperior specificity to both anatomic imaging and clinical exam.
December 1998
MTA98-032 MDRC Technology Assessment Program - PET Update - Page 16
Table 6: Characteristics of Diagnostic Accuracy Studies of FDG-PET inPatients with Head and Neck Cancer
Studycharacteristics Lowe et al. (1997) Wong et al. (1997) Myers et al. (1998)
Perspective ? prospective prospective retrospective
Patient source Consecutive patients betweenDecember 1994 and May 1996with head and neck cancer:• 28 with stage III/IV who were
participating in a neoadjuvantorgan-preservation protocolusing Taxol and carboplatin
54 consecutive patients whopresented to head and neckclinics at two hospitals31 with primary disease (TI=2T2=10 T3=9 T4=10), 23 withsuspected recurrence or residualdisease)• 16 had neck dissections
116 patients diagnosed withhead and neck cancer, ofwhich 72 had biopsy-provenSCC and 26 underwent neckdissections:• 14 patients with N0 disease
(24 total neck dissections)on clinical exam
Extent of disease(# patients)
Stage III=3Stage IV =25
N0=8N1=4N2a=2N2b=2
Stage I=1Stage II=8Stage III=2Stage IV=3
Benignconditions 6 patients with pathologic
complete response None reported None reported
PET criteria forpositive result 1,2,or 3 on a 4 point scale Not reported Not reported
Contrast CTcriteria forpositive node
N/A Standard size and morphologicalcriteria used to assess nodaldisease on CT/MRI
Not reported
Interpretation • Blinded visual consensususing a before and aftercomparison format
• 4-point scale• two readers
Not reported Not reported
Gold standarddetermination(# patients)
Pathologic complete responseor residual disease based onpost therapy biopsies obtainedafter PET blinded to PET data(28)
• independent biopsy (16)• All suspicious areas of
aerodigestive tract werebiopsied
Histopathology for number ofnodes, presence ofmalignancy, and extracapsularspread (14)
Data analysis By patient By patient By dissection
Summary/Discussion
Since the 1996 MDRC PET report seven additional studies (three of diagnosticaccuracy) of PET in head and neck cancer were published, met the inclusioncriteria, and were reviewed. Evaluations of PET in head and neck cancer havefocused mainly on detecting cervical node metastases in patients with knownprimaries, diagnosing disease recurrence, and monitoring response to treatment.
PET has potential uses at several points in the diagnosis and management of headand neck cancer patients. An early step in defining these uses is obtainingestimates of diagnostic accuracy. Only Lowe and associates (1997) met allevidence-based medicine criteria for diagnostic test evaluations, and the methodswere reasonably well described. The two other studies did not report blinding oftest interpreters and had other methodologic limitations, which affect the validityof the results, and it was unclear whether PET was used in addition to, or as asubstitute for, other tests. All of the studies in Table 7 received low methodologicquality scores due to presence of significant bias, insufficient reporting and/orsmall sample sizes. The diagnostic accuracy estimates from these studies shouldbe interpreted cautiously.
December 1998
MTA98-032 MDRC Technology Assessment Program - PET Update - Page 17
Information from a whole-body PET scan could have important treatmentimplications for patients with head and neck cancer. For example, identifying theprimary tumor site not detected by other modalities could alter treatmentplanning. If the primary is from the head and neck, it is potentially curable withsurgery and/or radiation therapy, whereas if the primary is located elsewhere, lesstoxic palliative treatment can be given. While there is a lower limit to the size oftumor that can be detected by PET, if validated in larger, rigorous studies, moreaccurate staging with PET could result in more appropriate treatment.
Minn et al (1997) (see reference list) assessed the feasibility of FDG uptake topredict cancer aggressiveness and survival. The results from 37 patients withprimarily advanced Stage III/IV disease suggested a correlation between FDGuptake and prognostic significance on univariate analysis but not on multivariateanalysis. Using FDG uptake to identify high-risk patients who would benefitfrom post-treatment surveillance requires further comparative study. Nonetheless,the wide range of primary sites and stages of head and neck cancer and theassociated wide range of site-specific treatment and outcomes would complicatesuch evaluations of PET.
Accurate diagnosis of disease recurrence is critical to the treating clinician. Withthe addition of chemotherapy to many organ-sparing protocols, the ability toaccurately assess nonsurgical treatment failure becomes increasingly moreimportant to judicious surgical salvage. For patients who become symptomatic orwho develop a mass during post-therapy surveillance, PET must be able todistinguish recurrence from treatment-related inflammation or fibrosis.
Goodwin (1998) suggested ways to improve such evaluations of PET that mayprovide more useful data to the treating physician. A prospective study of thesepatients, rather than a retrospective study of patients who had PET for variousreasons and at various times after treatment, would more appropriately address theclinical issue. Pretreating patients with steroids or antibiotics to reduceinflammation might enhance the positive predictive value of PET. Otherconsiderations include cost-effectiveness and capturing individual patient history,such as the timing of signs and symptoms after completion of therapy.
Controlled, prospective, blinded studies are needed to define the utility ofPET (either dedicated or camera-based systems) relative to other imagingmodalities in patients with head and neck cancer. Multiple sites may beneeded to accrue a sufficient number of patients. Results from this updatedliterature review confirm the conclusions and recommendations from thefirst report (see Preface).
December 1998
MTA98-032 MDRC Technology Assessment Program - PET Update - Page 18
Met
hodo
logi
cQ
ualit
y G
rade
D D D C D
Stud
y De
sign
Lim
itatio
ns
S,R,
W,t
S,r,W
,t,d
S,R,
W,t
S,R
S,R,
W,t
blinding — — — + —
Goldstandard H H H H H
Evid
ence
-bas
edM
edic
ine
Crite
ria
Comparisongroup + + + + +
Othe
r
Clini
cal e
xam
Se=5
8%
PET
+ CT
Se=8
6%Sp
=100
%PP
V=10
0%NP
V=91
%Ac
c=95
%
Clini
cal e
xam
Se=1
00%
Sp=6
0%
MRI
CT
CT +
MRI
Se=6
7%
Se=5
7%Sp
=90%
PPV=
80%
NPV=
75%
Acc=
76%
*
CT +
MRI
Se=7
5%Sp
=80%
Ope
ratin
g Ch
arac
teris
tics
PET
Se=6
7%
Se=7
8%Sp
=100
%PP
V=10
0%NP
V=88
%Ac
c=92
%**
(P=0
.11)
Se=1
00%
Se=9
0% (7
7-10
0%)
Sp=8
3% (5
3-10
0%)
PPV=
95%
NPV=
71%
Accu
arac
y=89
%
Se=1
00%
Sp=1
00%
N
12 p
ositiv
e ca
ses
4 ne
gativ
e ca
ses
9 po
sitive
disse
ction
s15
neg
ative
disse
ction
s
10 p
ositiv
e ca
ses
2 ne
gativ
e ca
ses
21 p
ositiv
e ca
ses
6 ne
gativ
e ca
ses
8 po
sitive
cas
es5
nega
tive
case
s
Stud
y
Won
g 19
97
Mye
rs 1
998
Won
g 19
97
Lowe
199
7
Won
g 19
97
Tabl
e 7:
Sum
mar
y of
Dia
gnos
tic A
ccur
acy
Stu
dies
of F
DG
-PE
T in
Hea
d an
d N
eck
Can
cer
H =
histo
logy;
F =
Follo
w up
; S =
sm
all s
ize;
R =
refe
rral b
ias;
W =
wor
k up
bias
; T
= te
st re
view
bias;
D =
diag
nosti
c re
view
bias
(upp
er c
ase
indica
tes
signif
icant
limita
tion;
lowe
r cas
e ind
icate
slim
itatio
n m
inim
ized
by s
tudy
des
ign,
pre
senc
e of
bia
s un
clear
, or s
mal
l effe
ct o
n op
erat
ing
char
acte
ristic
s)
Rol
e
Dete
ctin
gno
dal
met
asta
ses
Dete
ctin
glo
cal
recu
rrenc
e
Dete
ctin
gno
dal
recu
rrenc
e
December 1998
MTA98-032 MDRC Technology Assessment Program - PET Update - Page 19
B. Breast cancer
The American Cancer Society estimates 180,300 new cases (178,700 women and 1,600men) of breast cancer will be diagnosed in 1998 in the United States. After a 4% per yearincrease in the 1980s, breast cancer incidence rates have leveled off in recent years toabout 110 cases per 100,000. An estimated 43,500 women and 400 men will die ofbreast cancer in 1998, making breast cancer the second major cause of cancer death inwomen. Mortality rates continue to decline, particularly in younger women, likely due toearlier detection and improved treatment.
In FY 1997, there were 1.2 million female veterans (4.8% of all veterans) living in theUnited States, and the percentage of females in the veteran population is expected toincrease. In accordance with the Women Veterans Health Program Act of 1992, HealthServices Research and Development supports research to increase outreach and access tohealth care and to explore health issues that affect many women, including breast cancer(Feussner, 1997). VHA has also established the Mammography Quality Standards Officeand has made available a nationwide toll-free mammography information line (888-492-7844) to expand mammography services to female veterans.
Potential applications for PET in breast cancer management were defined previously(Flynn, 1996):
• Non-surgical evaluation of breast disease;• Staging recurrent disease;• Quantifying tumor glycolytic rate as a prognostic factor;• Monitoring response to therapy;• Patient selection for axillary dissection and for preoperative therapy;• Screening in subgroups of women (eg, those with breast implants, with prior breast
radiotherapy, multiple breast masses and history of negative biopsy results, orseverely fibrocystic breasts).
Table 10 summarizes the data and quality of individual studies of PET using FDG inbreast cancer. Only studies of dedicated PET for non-surgical diagnosis of breastdisease, patient selection for axillary dissection, and staging recurrent/metastatic diseasemet the inclusion criteria for this review. Three studies evaluated quantitative indices ofFDG uptake as an indicator of prognosis. These studies were classified as technicalefficacy due to their preliminary nature and will be discussed in the Summary/Discussionsection.
Defining unknown primary disease
Palmedo (1997) prospectively compared PET to scintimammography (SMM)using 99mTc MIBI in the pre-surgical evaluation of 20 patients with 22 suspiciousprimary lesions detected by clinical exam or mammography. The mean lesionsize was 29mm (range 8-53mm), of which only 3 patients had lesions smaller than9mm. Quantitative analysis of tracer uptake was also performed to characterizedisease, but no cut-off value was defined prospectively. Anecdotal data suggestedthat PET was superior to SMM in detecting axillary lymph involvement, but
December 1998
MTA98-032 MDRC Technology Assessment Program - PET Update - Page 20
neither test could determine extent of disease. The authors stressed that themenstrual cycle and age, which can alter MIBI uptake and FDG uptake,respectively, in normal tissue and the methods used to calculate FDG uptakecould affect test accuracy.
Detecting axillary lymph node involvement
The three studies in Table 8 met the inclusion criteria for review. Utech (1996),Crippa (1998), and Adler (1997) compared PET to axillary lymph node dissection(ALND) in patients with either suspected or confirmed breast cancer who werescheduled for axillary staging. Therapeutic decisions at surgery were based onclinical and routine imaging results, including mammography. PET was added inthe test sequence after the routine work up as a potential noninvasive method forstaging the axilla, the rationale being that a negative PET scan might obviate theneed for ALND in selected patients and, thus, decrease the morbidity and costsassociated with the procedure.
All were prospective studies, but only Crippa (1998) reported a consecutiveseries. The evidence for the use of PET in staging the axilla is confined to a selectgroup of patients with a high prevalence of malignancy and few benignconditions. The extent of axillary disease, reported in two studies, was limited topatients with metastases to ipsilateral axillary nodes. Crippa (1998) providedlimited evidence from small subgroups on the ability of PET to determine extentof disease, which is an important prognostic indicator; not surprisingly, PETsensitivity improved with more advanced disease.
Two studies used multiple readers to interpret PET images, but neither studyassessed interobserver variability. Of note, Adler (1997) used a higher dose oftracer and longer scanning times than were used in other studies. All studiesreported some evidence of blinding to the gold standard, but none met strictevidence-based criteria for blinding. Patient and disease characteristics, studydesign elements, and units of analysis varied across studies, and many studydesign elements were incompletely described or not reported, making the validityof these results difficult to assess.
December 1998
MTA98-032 MDRC Technology Assessment Program - PET Update - Page 21
Table 8: Characteristics of Prospective Studies of Axillary Lymph Node (N)Staging With FDG-PET in Patients with Potentially Operable BreastCancer
Note: All studies included primary tumors of mixed histologies, primarily invasive ductal carcinoma.
Study Characteristics Utech et al. (1996) Crippa et al. (1998) Adler et al. (1997)Patient source 124 patients with newly
diagnosed and histologicallyproven breast cancer prior totherapy• 64 patients with
metastatic nodes• 60 w/ surgically negative
axilla• ? consecutive series
68 consecutive patients (72 totalaxilla) with palpable breast nodulesscheduled for surgery based onclinical andmammography/ultrasound results• 61 had ALND• no ALND in patients with benign
lesions (8) and in situ ductalcarcinoma (3)
From a larger prospectivestudy of PET, 50 patients with52 axillary dissections who metinclusion criteria:• age ≥ 30 years• ≥ 2 ALND within 3 mo. Of
PET scan• ≥ 10 nodes dissected• ability to fast ≥ 4 hours• ?consecutive series
Exclusion criteria(# patients)
Hyperglycemic patients None reported • History of ipsilateral axillarylymph node dissection
• Preoperative systemictherapy
• Primary tumor < 5mm• Uninterpretable PET scan
(2)Benign conditions ofbreast (#patients)
None • proliferative dysplasia withoutatypica (6)
• focal inflammation (2)
None
Primary tumor size(mean, range)
Reported as:<1cm=16>1cm=49>2cm=30>3cm=29
2.0 cm, 0.4-6.7cm Reported as:T0=1T1=31T2=17T3=3
Prevalence of confirmedN metastases(# positive patients/totalpatients)
44/124=35% 27/61=44% 20/52=38% (by axilla)
Extent of N metastases(# patients)
N0=79N1=43N2=2• one with bilateral disease
N0=36 (# axilla)N1a=21N1b=13N2=2
Not reported
Axillary node size Not reported Not reported Range <0.1cm-2.5cm
PET criteria for positivenode discrete focal uptake >
background focal uptake > surrounding tissue)increased FDG uptake andscan quality; scores ≥ 3=positive on a 5-point scale
Interpretation • 3 radiologists + 1 nuclearmedicine
• blinded to all data exceptprimary tumor
• # readers not reported• blinded to histopathology, but to
other information not reported
• two readers• independent, blinded to all
but axilla side• discrepancies resolved by
consensusGold standarddetermination(# patients)
• histology (104)• histology + follow up (20)• extensive nodal sampling
(average #/patient=19,range 7-46)
• histology (61)• Extensive nodal sampling
(average # /axilla=21, range 12-38)
• histology (50)• extensive nodal sampling
(average #/patient=17,range not reported)
Data analysis By patient By axilla By axilla
ALND=axillary lymph node dissection
Detecting recurrence and metastases
The two studies in Table 9 presented the best evidence on the use of PET to stagerecurrent disease and metastases in breast cancer patients.
December 1998
MTA98-032 MDRC Technology Assessment Program - PET Update - Page 22
Table 9: Characteristics of Studies Using FDG PET to Stage Recurrent Diseaseand Metastases in Patients with Breast Cancer
Note: Both were retrospective studies.
Study Characteristics Bender et al. (1997) Moon et al. (1998)Patient source 75 patients with suspected recurrent or with
metastatic disease in undecided or equivocalcases• Includes results from CT/MRI• 63 patients had both PET and CT/MRI
data available for comparison• ?consecutive series
57 female patients (83 lesion sites) with a clinicalsuspicion of recurrence not resolved byconventional imaging:• who underwent primary surgery with or without
adjuvant chemo- or radiation therapy and• who were referred to the UCLA PET center from
October 1990 to October 1995• ?consecutive series
Exclusion criteria(# patients)
None reported • patients who underwent chemo-or radiationtherapy within 3 mo before PET
• lesions that were biopsied• lesions diagnosed with known disease
Benign conditions ofbreast (#patients)
None (# sites)• seroma (1)• muscle uptake (5)• thyroiditis (1)• radiation pneumonitis (1)• blood pool of great vessels (2)• osteoarthritis (1)• intestine (1)• unknown (6)
Primary tumor histology Well-differentiated ductal carcinoma (46)Infiltrating lobular carcinoma (10)
Not reported
Prevalence of confirmedlocal recurrence (#patients)
14/63=22% 29/57=51%
Prevalence of confirmedN metastases(# positive patients/totalpatients)
17/63=27% 8/26=31% (reported by lesion site)
Extent of M metastases(# patients)
• Bone (15)• Lung (5)• Liver (2)
• Bone (16)• Lung/Chest wall (7)• Liver (2)
PET criteria for positivelesion
4 point qualitative scale (intense, moderate,low, none)• Positivity criteria not defined
5 point qualitative scale• scores ≥ 3=positive
CT/MRI criteria forpositive lesion not defined N/A
Interpretation • 2 readers• independent• not blinded to other data
• 3 readers, discrepancies resolved by 4th reader• independent• blinded to histology but aware of suspicion of
metastasesGold standarddetermination(# patients)
• histology (71)• follow up (4)
• histology• lesion morphology on 2 or more conventional
imaging studies• ≥ 6 months of clinical and radiographic follow up
after PET
Data analysis By patient By patient and by lesion
Both studies were retrospective case series of patients with suspected recurrenceand/or metastases and equivocal findings after conventional imaging. PET wasused as a complement to conventional imaging. It was unclear whether thepatients in these studies represented consecutive case series. It should be notedthat Bender (1997) presented data on 75 patients, but only 63 patients hadinformation on both PET and CT/MRI for direct comparison. Few benign
December 1998
MTA98-032 MDRC Technology Assessment Program - PET Update - Page 23
conditions were represented in either study. This may be an artifact of the workup, and the benign cases were likely identified prior to inclusion. Both studieshad a higher proportion of patients with metastases to the bone than to lung and/or chest wall, or liver. It was difficult to compare other characteristics of thepatient population across studies due to incomplete reporting or variations in theunits of analysis.
Both studies used qualitative scales to define lesions on imaging and multiplereaders to interpret the images. Moon (1998) presented some data oninterobserver variability. Moon (1998) met most of the evidence-based medicinecriteria for blinding, but Bender (1997) did not blind interpreters to other data.
Summary/Discussion
PET has several potential uses in the management of patients with breast cancer.Since 1996, four technical efficacy and six diagnostic accuracy efficacy studieswere published that met inclusion criteria for the review, representing the bestevidence supporting the use of PET in breast cancer management to date. No newstudies were identified that assessed the role of PET in evaluating response totreatment or screening in subgroups of women, such as women with radiodensebreasts or breast implants.
The evidence on the ability of PET to detect unknown primary disease for thisreport is limited to one small study comprising a select group with a highprevalence of malignancy and few patients with small primary lesions less than1cm. Limitations in study design and reporting suggest the preliminary nature ofthis study. The results should be confirmed in a larger group of patients with arange of tumor sizes, benign conditions and stages of disease. Newer PET modelswith higher resolution and availability of new dedicated breast PET scanners mayimprove detection of smaller lesions (Wahl, 1998).
The current best evidence, derived exclusively from case series of patients with ahigh prevalence of malignancy and with few benign conditions, does not supportthe routine use of PET as the initial test in patient selection for ALND. At facevalue, the operating characteristics from these studies suggest that PET has arelatively high sensitivity with a lower positive predictive value and acorrespondingly lower specificity with a higher negative predictive value ascompared to ALND. PET also yielded a fair number of false positives, many ofwhich could not be explained. Some of the more recent studies are larger, butmethodologic biases and incomplete reporting justified low methodologic qualityscores.
Variations in the characteristics of the study populations, scanning techniques,and in the units of analysis may affect the generalizability of these results,particularly to mammographically tested populations, which typically have alower prevalence of malignancy. Predictive values and other estimates ofdiagnostic accuracy should be interpreted with caution.
December 1998
MTA98-032 MDRC Technology Assessment Program - PET Update - Page 24
ALND with histopathology of dissected nodes supplies critical information totreatment management, is currently recommended by the NCI for most patientswith Stage 1 or higher disease, but is associated with significant morbidity.Relative to other studies of screening and treatment options, published PET datato date are based on small numbers of patients. Moreover, the lower boundary ofresolution limits the ability of current PET modalities to detect tumors less than1cm in diameter. The consequences of false negative PET results in the absenceof ALND in patients for whom effective treatment is available should be avoided.
The potential for PET to visualize the internal mammary nodes (potentially N3disease) has been reported (Wahl, 1998). An NCI-sponsored multi-center trial isevaluating the accuracy of PET in staging the axilla and will include patients withN3 disease (See Section IX). Clinicians should await the results of this studybefore incorporating PET into routine clinical practice.
Likewise, the evidence on use of PET in detecting recurrent disease andmetastases and defining unknown breast disease is in its early stages. PET wastypically part of a testing sequence, but the marginal value of PET in the work upof these patients remains to be determined. The authors emphasized, and the TAProgram concurs with, the need for further studies to assess the clinical impact ofPET in the management of recurrent breast cancer.
Utech (1996), Crippa (1998), and Oshida (1998) (See technical efficacy list inReference Section) presented some evidence on the feasibility of usingquantitative FDG PET uptake by either the primary tumor or axillary lymph nodesas a prognostic indicator. Any attempt to correlate PET data with survivalrequires knowledge of the underlying characteristics of the study population andsufficient follow up time to track survival (Laupacis, 1994). The range of diseasestages and corresponding treatment options would further confound the results.Large, rigorous studies are needed to define the utility of PET as a prognostic test.
Controlled, prospective, blinded studies are needed to define the utility ofPET (either dedicated or camera-based systems) relative to other imagingmodalities in patients with breast cancer. Multiple sites may be needed toaccrue a sufficient number of patients. Results from this updated literaturereview confirm the conclusions and recommendations from the first report(see Preface).
December 1998
MTA98-032 MDRC Technology Assessment Program - PET Update - Page 25
Met
hodo
logi
cQ
ualit
y G
rade
C C C C D C
Stud
y De
sign
Lim
itatio
ns
S,r,t
,d
r,d
S,r,t
,d
S,d,
r
S,r,w
,T,d
S,R
,t,d
blinding + + + + +
Goldstandard H H
+ F H H H+F
H+F
Evid
ence
-bas
edM
edic
ine
Crite
ria
Comparisongroup + + + + + +
CT/M
RI
Se=4
6-91
%Sp
=95-
98%
PPV=
88-9
1%NP
V=86
-98%
Acc=
87-9
7%
SMM
Se=9
2%Sp
=86%
Ope
ratin
g Ch
arac
teris
tics
PET
Se=9
2%Sp
=86%
Se=1
00%
Sp=7
5%PP
V=69
%NP
V=10
0%Ac
c=84
%
Se=8
5%Sp
=91%
PPV=
85%
NPV=
91%
Acc=
89%
(ove
rall v
alues
repo
rted)
Se=9
5%Sp
=66%
PPV=
63%
NPV=
95%
Acc=
77%
(ove
rall
valu
es)
Se=7
3-10
0%Sp
=93-
96%
PPV=
85-8
8%NP
V=92
-100
%Ac
c=90
-97%
Se=9
3%Sp
=61-
79%
PPV=
82%
NPV=
92%
(ove
rall
valu
es)
N
13 m
alig
nant
lesio
ns7
benig
n les
ions
in 20
cas
es
44 p
ositiv
e ca
ses
80 n
egat
ive c
ases
27 p
ositiv
e ax
illa45
neg
ative
axil
la
in 61
cas
es
20 p
ositiv
e ax
illa32
neg
ative
axil
la
in 50
cas
es
54 p
ositiv
e ca
ses
9 ne
gativ
e ca
ses
29 p
ositiv
e ca
ses
28 n
egat
ive c
ases
Stud
y
Palm
edo
1997
Utec
h 19
96
Crip
pa 1
998
Adle
r 199
7
Bend
er 1
997
Moo
n 19
98
Tabl
e 10
: S
umm
ary
of D
iagn
ostic
Acc
urac
y S
tudi
es o
f PE
T an
d A
ltern
ativ
es in
Bre
ast C
ance
r
H =
histo
logy;
F =
Follo
w up
; S =
sm
all s
ize;
R =
refe
rral b
ias;
W =
wor
k up
bias
; T
= te
st re
view
bias;
D =
diag
nosti
c re
view
bias
(upp
er c
ase
indica
tes
signif
icant
limita
tion;
lowe
rca
se in
dica
tes
limita
tion
min
imize
d by
stu
dy d
esig
n, p
rese
nce
of b
ias
uncle
ar, o
r sm
all e
ffect
on
oper
atin
g ch
arac
teris
tics)
Rol
e
Defin
ing
unkn
own
prim
ary
dise
ase
Dete
ctin
g ax
illary
node
invo
lvem
ent
Det
ectin
gre
curre
nce
ordi
stan
tm
etas
tase
s
SMM
= sc
intim
amm
ogra
phy w
ith 99
m Tc-
MIB
I
December 1998
MTA98-032 MDRC Technology Assessment Program - PET Update - Page 26
C. Non-Small Cell Lung Cancer
Bronchogenic carcinoma, classified as either small cell or non-small cell, comprises 95%of all primary lung cancers. This section will address only non-small cell varieties, asthey constitute the majority (75%) of all bronchogenic carcinomas and, when localized,have the potential for cure with surgical resection.
Bronchogenic carcinoma is the leading cause of cancer death in the United States. In1998 the American Cancer Society estimates 171,500 new cases of lung cancer and160,100 deaths from lung cancer. Malignant neoplasms of the bronchus and lungaccounted for 9,730 discharges (1.5% of all discharges) with an average length of stay of13.8 days within the Veterans Health Administration in FY 1997.
Non-small cell lung cancers (NSCLC) include adenocarcinoma (includingbronchioalveolar), squamous (or epidermoid) cell carcinoma, and large cell (includinglarge cell anaplastic) carcinoma. While 5-15% of NSCLCs are incidental findings on achest x-ray, the vast majority of patients have symptomatic, advanced disease at clinicalpresentation.
Initial diagnosis is based on complete history, physical exam, and chest x-ray. If canceris suspected, then staging is needed to assess the extent of local and distant disease.Stage of disease is the primary predictor of response to treatment and one of theimportant predictors of survival.
CT is the preferred diagnostic imaging test and is used at several points in themanagement of a patient with lung cancer: 1) to stage disease; 2) to evaluate treatmentresponse; and 3) to differentiate recurrent disease from fibrosis. Use of other diagnosticimaging technologies to stage lung cancer is circumscribed largely because of technicallimitations, availability, and cost.
CT provides morphologic (typically size) detail of the disease site. Accordingly, diseasestatus of mediastinal lymph nodes are classified according to size, with nodes greater than1 cm in diameter generally indicative of malignancy. This can be problematic, becausebenign lymph nodes may appear enlarged and micrometastases may appear normal onCT. Consequently, biopsy confirmation of the primary site and metastases is required todetermine the most appropriate treatment.
More accurate noninvasive methods for staging NSCLC are needed to minimize the useof invasive procedures for diagnosis and monitoring treatment response. To this end, themetabolic information provided by a PET scan may be useful. Several roles for PET instaging lung cancer have been identified in the literature:
• Defining unknown primary disease;• Detecting hilar and mediastinal metastases;• Detecting distant metastases;• Defining recurrence from fibrosis;• Analyzing tumor biology;• Monitoring response to therapy;
December 1998
MTA98-032 MDRC Technology Assessment Program - PET Update - Page 27
• Predicting tumor response by measuring uptake of chemotherapeutic agents.
Tables 11 and 12 depict study characteristics and Table 13 summarizes the data andquality of individual diagnostic accuracy studies of FDG-PET in NSCLC that met theinclusion criteria for this review. Scores were further refined with pluses and minuses toreflect the degree to which investigators minimized the effect of these biases ondiagnostic accuracy results.
Defining unknown primary disease
Two studies met the inclusion criteria for the report. Guhlman (1997) andHagberg (1997) are relatively small retrospective surgical series with a highprevalence of malignancy in their respective cohorts. Both evaluated PET in thetest sequence after CT, but only Guhlman (1997) measured PET independently ofother tests in all patients. Neither study presented data comparing PET to CTalone. Both studies received low methodologic quality grades due to incompletereporting of methods and significant biases in study design, which may inflateestimates of diagnostic accuracy.
Detecting hilar/mediastinal adenopathy
Recent evidence on the use of PET in NSCLC emphasizes its staging potential.Six studies meeting the inclusion criteria presented evidence on the diagnosticaccuracy of PET in nodal (N) staging and are listed in Table 13. All enrolledpatients had suspected or biopsy-proven lung cancer. Data analyses included onlybiopsy-verified cases, implying a strong presence of work up bias across allstudies. All studies assessed the role of PET independently of CT in the work up;Vansteenkiste (1997) also assessed PET as an adjunct to CT.
Guhlman (1997) and Hagberg (1997) were small retrospective studies withseveral methodologic flaws. The remaining four studies were reported asprospective evaluations of PET. Ambiguous descriptions of study methodologycall into question the true, real-time prospective nature of three of them (Steinert,1997; Vansteenkiste, 1997; Sasaki, 1996). Of these three, Sasaki (1996) was themost methodologically flawed.
Bury (1997) presented the largest and the only discernibly true prospectiveevaluation of PET in staging patients with NSCLC. Steinert (1997) andVansteenkiste (1997) also presented notable attributes. These three studiesrepresent the strongest evidence on the use of PET in N staging patients withNSCLC and are presented in Table 11 for comparison.
December 1998
MTA98-032 MDRC Technology Assessment Program - PET Update - Page 28
Table 11: Characteristics of Prospective Studies of Mediastinal Lymph Node(N) Staging With FDG-PET in Patients with Potentially OperableNSCLC
Note: All studies included mixed histologies, primarily squamous cell and adenocarcinoma.
StudyCharacteristics Bury et al. (1997) Steinert et al. (1997) Vansteenkiste et al. (1997)Patient source 141 consecutive patients who
presented between 9/94-10/96with new or suspected NSCLCbased on sputum cytology,needle biopsy, or flexiblebronchoscopy• 109 enrolled
62 surgical candidates withsuspected or proven NSCLCwho had PET between 2/94 and3/96• 47 enrolled
Unknown # patients who presentedbetween 9/95-4/96 with suspected orconfirmed NSCLC and who hadstandard M staging• 50 enrolled
Exclusioncriteria(# patients)
• poor physiologic status (22)• poor compliance or no
definitive diagnosis (11)
• prior neoadjuvant therapy• diabetes• inadequate CT (2)• distant metastases (8)• inadequate sampling (5)
• inoperable due to distantmetastases
• diabetes• treatment with oral corticosteroids• ischemic cardiomyopathy• direct mediastinal invasion of
primary tumor• obvious bulky mediastinal
adenopathiesPrevalence ofconfirmed Nmetastases(#N1-N3/#patients)
34/66=52% 29/47=62% 15/50=30%
Extent of Nmetastases(# patients)
N0=32N1=20N2=10N3=4
N0=18N1=16N2=7N3=6
N0=35N2=15
Benignconditions
• nonspecific inflammation=2• pneumonia=1• multinodular goiter=1• localized FDG uptake in
hepatic-splenic angle ofcolon=1
none reported none reported
PET criteria forpositive node
• moderate uptake: > 2Xuptake in contralateral orreference region
• intense uptake: markedlyhigher than reference region
• FDG uptake ≥ FDG uptake inbrain
• nodular appearance
Grades 4 and 5 on a 5-pointsemiquantitative scale
Contrast CTcriteria forpositive node
short axis diameter > 10 mm • short axis diameter > 10 mmexcept:
• upper paratracheal nodes >7mm short axis diameter
• infracarinal station > 11 mmshort axis diameter
maximal cross-sectional diameter ≥1.5 cm
Interpretation • independent, blind• consensus by 2 radiologists
and 2 nuclear medicine
• independent, blind• 1 radiology reader• 1 nuclear medicine reader
• independent, blind• one chest physician, one
radiologist• 2 nuclear medicine readers
Gold standarddetermination(# patients)
• histology frommediastinoscopy (5),thoracotomy (51), both (10)
• radiologic follow up basedon CT or PET
• all accessible nodes atsurgery sampled
• extensive nodal sampling atthoracotomy of all identifiablenodes regardless of size onimaging
• mediastinoscopy (22) and/orthoracotomy (18)
• nodal sampling atmediastinoscopy (47) and atthoracotomy (49), fine needleaspiration (1)
• extent of sampling not reported
Data analysis correlated by patient correlated by nodal station correlated by patient
December 1998
MTA98-032 MDRC Technology Assessment Program - PET Update - Page 29
Variations in study characteristics and units of analyses contributed to the rangeof reported estimates of diagnostic accuracy and differences in quality scoresacross studies. All studies had a significant degree of work up bias, whichcontributed to their low quality scores. All conducted varying degrees of nodalsampling, a means for minimizing diagnostic review bias, but the extent ofsampling varied and was not reported with sufficient detail to enable the reader toquantify the effect of this bias on diagnostic accuracy. Bury (1997) andVansteenkiste (1997) utilized multiple readers for blinded, independent imageinterpretation, but neither assessed interobserver variability.
Bury (1997) provided the strongest evidence to date on the diagnostic accuracy ofPET in N staging NSCLC. A comparison of PET to CT yielded comparableaccuracy estimates. The authors presented data on the impact of PET inmodifying treatment, but no methods for systematic assessment were described.Bias in the stated methods and in incomplete reporting of other critical designelements hindered evaluation of study validity in the other studies. None of thestudies assessed the incremental value of PET in the work up of NSCLC.
Detecting distant metastases
Studies in Table 12 met the inclusion criteria for review. Erasmus (1997)reported on 27 patients diagnosed with bronchogenic carcinoma and adrenalmasses detected by CT. Adrenal masses are common in patients with NSCLC,but in the absence of other extrathoracic metastases, they are likely to be benign.Diagnosis of many adrenal masses remains indeterminate after standard anatomicimaging (CT or MRI), and a biopsy is required before treatment can be planned.The rationale for using PET in this case is to improve the noninvasive diagnosticaccuracy, thus reducing the need for biopsy. Patients with normal FDG uptake inthe adrenals and no evidence of distant metastases might be considered eligiblefor curative resection.
The findings suggest that, as an adjunct to CT, PET can discern malignant frombenign adrenal masses using both visual and semiquantitative analyses. Resultsfrom this small preliminary study would need to be confirmed in larger,prospective studies to ascertain valid estimates of diagnostic accuracy and theadded value of PET in diagnosing adrenal masses in these patients.
Bury (1997) present the strongest evidence to date on the use of PET for Mstaging NSCLC. They compared PET independently to conventional imaging(chest CT, abdominal CT, and bone scintigraphy) for M staging 109 patients withnew or suspected NSCLC. The results suggest modest improvements insensitivity and negative predictive value for PET over conventional imaging. Theauthors reported that PET correctly changed M stage, as determined byconventional imaging, in 14% of the cases and modified therapy in 20% of thepatients, but the methods for assessing these changes were not described.
December 1998
MTA98-032 MDRC Technology Assessment Program - PET Update - Page 30
Table 12: Characteristics of Prospective Studies of Distant Metastases (M)Staging With FDG-PET in Patients with NSCLC
StudyCharacteristics Bury et al. (1997) Erasmus et al. (1997)Patient source 141 consecutive patients with new or suspected
NSCLC who had PET and conventional imagingbetween September 1994 and October 1996:• 109 patients enrolled in study• 39 patients with 59 sites of confirmed distant
metastases
Unknown # consecutive cases presenting tothoracic surgery, oncology, or pulmonarybetween January 1993 and January 1996 with adiagnosis of bronchogenic carcinoma and anadrenal mass detected by CT• 27 patients with 33 adrenal masses enrolled in
study
Exclusion criteria(# patients)
• Poor physiologic status (22)• Poor compliance or no definitive diagnosis (11)
• Inability to obtain informed consent• Poor clinical status• Death
Patientcharacteristics
• 77 men, 32 women• mean age= 64 yrs (44-83 yrs)
• 19 men, 8 women• mean age= 57 yrs. (39-76 yrs)
Characteristics ofmetastases (#patients)
• NSCLC (109)• Mean diameter not reported
• NSCLC (24); Small cell (3)• Bilateral masses (6)• Mean diameter=3 cm (1-9cm)
Prevalence ofconfirmed distantmetastases
39 pts /109 pts=36% 23 sites /33 sites=70%
Locations of distantmetastases (# sites)
• Adrenal glands(10)• Nonregional lymph nodes (6)• Lung (10); Bone (13); Liver (18)• Pleura (1); Soft tissue (1)
Adrenal glands (27)
Benign conditions(# sites)
• Nonspecific inflammation (2)• Pneumonia (1)• Multinodular goiter (1)• Localized FDG uptake in hepato-splenic angle
of colon (1)
Not reported
PET criteria forpositive metastases
• Moderate uptake: > 2X uptake in contralateralor reference region
• intense uptake: markedly higher thanreference region
Positive activity= activity > background
CT criteria forpositive metastases
• Nodule characteristics not defined• Presence of clinical disease (symptomatic
patient, progression on imaging, abnormalbiochemistry) 6 months after imaging negativeimaging
Visual detection of mass, characteristics notdefined
Interpretation • Independent, blinded to all data excepthistology of primary tumor
• Consensus by 2 radiologists and 2 nuclearmedicine
• Independent, blinded to clinical and biopsyfindings
• 3 readers
Gold standarddetermination(# patients or sites)
• Biopsy (21)• Clinical and radiologic follow up (88)
• Percutaneous needle biopsy (11)• Growth characteristics on sequential CT
studies (16)• CT attenuation values < 10H (6)
Data analysis Correlated by patient correlated by site
December 1998
MTA98-032 MDRC Technology Assessment Program - PET Update - Page 31
Summary/Discussion
Early studies of PET suggested several potential uses for PET in managingNSCLC (Flynn, 1996). Positive trends in Medicare and private sector coveragepolicies for PET in lung cancer staging continue to fuel interest in the use ofdedicated and camera-based PET as diagnostic tools. Since the first report, theTA Program identified 14 additional studies (7 of diagnostic accuracy) usingdedicated PET, which met the inclusion criteria for this report. There were threeareas in which potential uses for PET in NSCLC were studied: defining unknownprimary disease, detecting nodal metastases, and detecting distant metastaticdisease.
The best evidence on the diagnostic accuracy of PET in staging NSCLC suggestscomparable accuracy of PET to CT in nodal staging and slightly better sensitivity,negative predictive value, and accuracy of PET over conventional imaging instaging distant metastases (Bury, 1997). Significant methodological biases,incomplete reporting of critical design elements, and variations in studycharacteristics (e.g., lack of uniform criteria for defining positive results on PET)limit the validity of the included studies and warranted low methodologic qualityscores.
Appropriate use of the reference standard, or the “truth measure”, is among themost challenging aspects of these studies to assess. Diagnostic review bias isoften introduced, as biopsy sampling is rarely carried out independently ofimaging results (e.g., it would be impractical to blind the surgeon to imaging).Bury (1997) minimized the effect of diagnostic review bias in nodal staging byconducting extensive nodal sampling and in distant staging by confirming diseasestatus in all subjects using radiologic or clinical follow up or other confirmatorytests.
Imaging results are often used to determine which patients receive biopsyverification of mediastinal involvement (work up bias). To improve N stagingaccuracy several investigators advocated complementing the sensitivity of CTwith the high negative predictive value of PET. They reasoned that a negativePET scan following a positive or indeterminate CT scan would excludemediastinal metastases with a high degree of certainty and might obviate the needfor invasive mediastinal evaluation (e.g., mediastinoscopy).
The best evidence for PET’s N staging potential is confined to biopsy verifiedcases who had suspicious nodes on imaging. The size criteria for characterizingdisease on CT and the lower detectable limit of resolution with PET maymisclassify small tumor involvement, resulting in understaging. Failure toconfirm disease status through follow-up in patients with negative CT or PETresults may miss false negative results; failure to include the results in the analysiswould result in inflated sensitivity and negative predictive values. Accurate,robust negative predictive values from studies that reduce the effect of work upbias are critical to determining the utility of PET in mediastinal staging.
December 1998
MTA98-032 MDRC Technology Assessment Program - PET Update - Page 32
Methodologically rigorous evaluations of diagnostic imaging, which reduced oraccounted for the effects of methodologic biases on diagnostic accuracy, havebeen published (See Appendix II). In particular, Webb (1991) of the RadiologicDiagnostic Oncology Group (RDOG) provides an excellent model for evaluatingdiagnostic imaging in staging NSCLC. From patient enrollment to data analysisthis rigorous evaluation offers extensive, detailed techniques for limiting themany biases inherent in diagnostic imaging studies. Incorporating study designelements from this model would strengthen the current best evidence for stagingNSCLC using PET.
The value of diagnostic PET cannot be determined solely on improved accuracyover existing modalities. PET must demonstrate changes in diagnostic certaintyand/or treatment planning or lower overall costs of patient management to justifyits role in the work up. It can be argued that the metabolic information from PETmay complement the information provided by conventional anatomic imaging andimprove staging accuracy. More accurate staging may lead to more appropriatetreatment planning. Studies included in this review reported anecdotal evidenceof changes in treatment planning attributable to PET, but the impact of PET ontreatment management was not systematically assessed, or reported as such.Furthermore, the range of stages and histologies of NSCLC and the associatedrange of treatments and outcomes would confound the effect of PET on outcomesof treatment, many of which are under investigation.
The TA Program concludes that the prevailing evidence does not support theroutine use of either dedicated or camera-based PET in lung cancer staging.Data from rigorous, prospective clinical trials are needed to determine theadded value of PET in the work up of NSCLC. Methodologically rigorousstudies of diagnostic imaging have been published in the peer-reviewedliterature. These studies may serve as models for guiding design of futurePET research. Review of the more recent evidence confirms the conclusionsfrom the first report.
December 1998
MTA98-032 MDRC Technology Assessment Program - PET Update - Page 33
Met
hodo
logi
cQ
ualit
y G
rade
C D D+ D-
D- D D D-
Stud
y De
sign
Lim
itatio
ns
S,r,w
,d
S,r,w
,T,d
s,r,W
,t,d
S,R,
W,d
S,r,W
,d
S,r,W
,d
S,r,W
,t,D
S,R,
W,T
,d
blinding + — + + + + +
goldstandard H H H H H H H H
Evid
ence
-bas
ed M
edic
ine
Crit
eria comparison
group + + + + + + + +
CT a
lone
(95%
CI)
Not r
epor
ted
Not r
epor
ted
Se=7
9%Sp
=71%
PPV=
75%
NPV=
76%
Acc=
75%
Se=5
0% (2
7-73
%)
Sp=7
5% (4
3-95
%)
Acc=
59%
(41-
76%
)
Se=5
6%Sp
=100
%
Se=5
7%Sp
=94%
PPV=
76%
NPV=
87%
Acc=
85%
Se=6
7%Sp
=63%
PPV=
43%
NPV=
81%
Acc=
64%
Se=6
5%Sp
=87%
(P<0
.05)
PPV=
61%
NPV=
89%
Acc=
82%
(P<0
.05)
PET
+ CT
Se=9
3%Sp
=70%
Se=9
3%Sp
=97%
PPV=
93%
NPV=
97%
Acc=
96%
Ope
ratin
g Ch
arac
teris
tics
PET
alon
e (9
5%CI
)
Se=9
4%Sp
=86%
Acc=
91%
Se=8
9% (7
2-96
%)
Sp=8
7% (7
1-97
%)
PPV=
89%
(72-
96%
)NP
V=87
% (7
1-96
%)
Acc=
88%
Se=8
0% (5
6-94
%)
Sp=1
00%
(73-
100%
)Ac
c=87
% (7
1-96
%) (
p<.0
2)
Se=6
7%Sp
=100
%
Se=8
9% (P
=0.0
066)
Sp=9
9%PP
V=96
%NP
V=97
%Ac
c=97
%
Se=6
7%Sp
=97%
PPV=
91%
NPV=
87%
Acc=
88%
Se=7
6%Sp
=98%
(P<0
.05)
PPV=
93%
NPV=
93%
Acc=
93%
(P<0
.05)
N 32 m
alig
nant
case
s14
ben
ign
case
s
44 p
ositiv
e no
dules
10 n
eg n
odule
s(in
49
patie
nts)
34 p
ositiv
e ca
ses
32 n
egat
ive c
ases
20 p
ositiv
e ca
ses
12 n
egat
ive c
ases
9 po
sitive
nod
es9
nega
tive
node
s(in
18
patie
nts
with
N2 d
iseas
e on
ly)
28 p
ositiv
e no
dal
statio
ns84
neg
ative
nod
alsta
tions
(in 4
7 pa
tient
s)
15 p
ositiv
e ca
ses
35 n
egat
ive c
ases
17 p
ositiv
e re
gions
54 n
egat
ivere
gion
s(in
unk
nown
#pa
tient
s)
Stud
y
Guh
lman
199
7
Hagb
erg
1997
Bury
199
7
Guh
lman
199
7
Hagb
erg
1997
Stei
nert
1997
Vans
teen
kiste
1997
Sasa
ki 19
96
Tabl
e 13
: Sum
mar
y of
Dia
gnos
tic A
ccur
acy
Stu
dies
of P
ET
and
Alte
rnat
ives
in S
tagi
ng L
ung
Can
cer
H =
histo
logy;
F =
Follo
w up
; S =
sm
all s
ize;
R =
refe
rral b
ias;
W =
wor
k up
bias
; T
= te
st re
view
bias;
D =
diag
nosti
c re
view
bias
(upp
er c
ase
indica
tes
signif
icant
limita
tion;
lowe
r cas
e ind
icate
s lim
itatio
n m
inim
ized
bystu
dy d
esign
, pre
senc
e of
bias
unc
lear,
or s
mall
effe
ct on
ope
ratin
g ch
arac
teris
tics)
Rol
e
(Som
e as
sess
edm
ultipl
e ro
les)
Defin
ing
unkn
own
prim
ary
dise
ase
Dete
ctin
gm
edia
stin
al/h
ilar
aden
opat
hy
December 1998
MTA98-032 MDRC Technology Assessment Program - PET Update - Page 34
Met
hodo
logi
cQ
ualit
y G
rade
B/C C
Stud
y De
sign
Lim
itatio
ns
r,t,
S,r,d
blinding + +
goldstandard H
+ F
hist
olog
y +
CT fo
llow
up
Evid
ence
-bas
ed M
edic
ine
Crite
ria
comparisongroup
+ +in
tern
al
CT a
lone
(95%
CI)
(con
vent
iona
l im
agin
g*)
Se=8
2%Sp
=89%
PPV=
80%
NPV=
89%
Acc=
86%
PET
+ CT
Ope
ratin
g Ch
arac
teris
tics
PET
alon
e (9
5%CI
)
Se=1
00%
(91-
100%
)Sp
=94%
(86-
98%
)PP
V=90
% (7
8-97
%)
NPV=
100%
(95-
100%
)Ac
c=96
% (9
0-98
%)
Se=1
00%
Sp=8
0%
N 39 p
ositiv
e ca
ses
70 n
egat
ive c
ases
23 m
alig
nant
lesio
ns10
ben
ign le
sions
Stud
y
Bury
199
7
Eras
mus
199
7
Tabl
e 13
(con
t.): S
umm
ary
of D
iagn
ostic
Acc
urac
y S
tudi
es o
f PE
T an
d A
ltern
ativ
es in
Sta
ging
Lun
g C
ance
r
H =
histo
logy;
F =
Follo
w up
; S =
sm
all s
ize;
R =
refe
rral b
ias;
W =
wor
k up
bias
; T
= te
st re
view
bias;
D =
diag
nosti
c re
view
bias
(upp
er c
ase
indica
tes
signif
icant
limita
tion;
lowe
r cas
e ind
icate
s lim
itatio
n m
inim
ized
bystu
dy d
esign
, pre
senc
e of
bias
unc
lear,
or s
mall
effe
ct on
ope
ratin
g ch
arac
teris
tics)
Rol
e
(Som
e as
sess
edm
ultipl
e ro
les)
Dete
cting
dist
ant
met
asta
ses
* Bur
y et
al 1
997
defin
ed c
onve
ntion
al im
aging
as
ches
t CT,
abd
omina
l CT,
and
bon
e sc
intigr
aphy
December 1998
MTA98-032 MDRC Technology Assessment Program - PET Update - Page 35
D. Solitary Pulmonary Nodules
Background information on solitary pulmonary nodules (SPN) is supplied by Lillingtonand Caskey (1993). A SPN is a single spherical lesion within the lung not associatedwith hilar enlargement or atelectasis and with a diameter generally less than 4.0 cm. TheAmerican Cancer Society reports that SPNs represent approximately 15% of all lungcancer diagnosed and estimates 25,725 new cases of malignant SPNs in the United Statesin 1998.
The differential diagnoses of a SPN include many malignant and benign processes. Themost common malignant forms are bronchogenic carcinomas. Reported prevalence ofmalignant SPNs range from less than 5% to greater than 70% because of differences inthe spectrum and severity of disease within each reported patient series. A malignantSPN represents a clinical stage I lesion, which is potentially curable with resection.Infectious granulomas represent the majority of benign processes and are causedpredominately by coccidiomycosis, histoplasmosis, and tuberculosis.
The following risk factors directly correlate with the probability of cancer in patients witha SPN: 1) patient’s age; 2) smoking history; 3) antecedent malignancy; 4) stability oflesion size on chest x-ray for 2 years; 5) absence of benign patterns of calcification withinthe nodule; and 6) nodule morphology (size and edge characteristics on CT). Thebaseline prevalence of malignancy in the study population may suggest the likelihood ofa malignant SPN. Exposure to benign diseases such as tuberculosis or a history ofresidence in areas endemic for coccidiomycosis or histoplasmosis will suggest a lesserlikelihood, but not rule out, malignancy.
Following clinical exam and chest radiography, the standard radiologic method of choicefor evaluating SPNs is CT. CT provides information on the location and morphology ofthe nodule and can be used to guide biopsy procedures. Iodinated contrast material andhigh resolution CT densitometry may be used to enhance the differential diagnosis.However, limitations in the use of CT have been reported. Many SPNs are classified as“indeterminate” after CT and warrant invasive biopsy confirmation to determine theappropriate therapeutic course.
FDG PET has been proposed as a potential solution for improving the noninvasivedifferential diagnosis of SPNs, thereby reducing the need for higher risk invasive biopsysampling and the associated morbidity and costs. Current evidence from this reviewsupports the complementary use of PET after CT in the work up of patients with nodulediameters less than 3 cm or 4 cm, i.e., those nodules most likely to be indeterminate.
Table 14 displays the attributes of each study to highlight the variations in study qualityand in criteria relevant to the applicability of the results. Table 15 summarizes the dataand quality of individual diagnostic accuracy studies of FDG PET in SPNs.
Characterizing indeterminate solitary pulmonary nodules
December 1998
MTA98-032 MDRC Technology Assessment Program - PET Update - Page 36
Two studies met the inclusion criteria for this report. Dewan (1997) conducted aretrospective single-site study of indeterminate SPNs in 52 consecutive patients,who underwent PET between April 1990 and February 1994. They comparedPET with and without standard criteria (clinical and radiologic data) usinglikelihood ratios1 in Bayesian analysis to predict the probability of cancer in aSPN. Using sensitivity and specificity derived from this patient group, theauthors determined that PET alone was the best predictor of cancer.
However, biases in study design and violation of the assumption of conditionalindependence between tests in the testing sequence, a requirement of Bayesiananalysis, preclude drawing definitive conclusions regarding the accuracy of PETand its contribution to diagnostic certainty in these patients. Moreover, the impactof PET on treatment planning was not assessed. It is also important to note thatmany of these patients may have been included in studies assessed in the 1996report.
Lowe (1998) conducted a multi-site study of radiologically indeterminate SPNs in105 consecutive patients, who underwent imaging between October 1993 andAugust 1994. The study population included a broader range of benignconditions and nodule sizes compared with other published studies for thisindication, reflecting the advantages of multi-site design. The authors presented avery detailed description of their blinding procedures and were the onlyinvestigators to calculate interobserver variability in visual analysis. From thestated methods, it is unclear whether they collected patient data in a “real-time”prospective fashion or retrospectively from surgical series.
These authors calculated likelihood ratios overall and for each subgroup. Thelikelihood of cancer was consistently higher using quantitative analysis overvisual analysis. Except for specificity in SPNs ≤ 3cm in diameter, there were nosignificant differences between visual and quantitative analyses in the otherdiagnostic accuracy measures across subgroups. Small sample sizes in thesubgroups likely contributed to the failure to detect any significant differences.Interobserver variability was very low (kappa=0.95), indicating goodreproducibility of image interpretation.
1 Likelihood ratio, expressed as Sensitivity/1-Specificity, is a measure of accuracy that indicates by how much adiagnostic test result will raise or lower the pretest probability of disease, thereby increasing the certainty about apositive or negative diagnosis.
December 1998
MTA98-032 MDRC Technology Assessment Program - PET Update - Page 37
Table 14: Characteristics of Studies Using FDG-PET of Patients withRadiographically Indeterminate Solitary Pulmonary Nodules
StudyCharacteristics Dewan et al. (1997) Lowe et al. (1998)Perspective Retrospective Prospective (?not real-time)
Patient source
52 consecutive patients who underwent PETbetween April 1990 and February 1994• included 3 with extrathoracic malignancy
Multisite study of 89 of 105 consecutive patients whounderwent imaging between October 1993 and August1994
Exclusion criteria(# patients)
• Cavitary or calcified nodules• Nodule size > 3cm• Age ≤ 30 years• # patients not reported
• no definitive histologic confirmation (8)• 4 not classified as radiographically indeterminate SPN
(4)• no available CT scans (2)• nodule size < 0.7cm or > 4.0cm on CT(? # pts.)
Patient demographics
• 43 men (83%)• mean age ± SD=63.6±11.3 years• 41(79%) current smokers• 52% ≥ 20 cigs/day
• 61 men (69%)• mean age ± SD=63±9.5 years• smoking status not reported
Prevalence ofmalignancy 37/52=71% 60/89=67%
Nodule size in cm(%malig. pts. vs.%benign pts.)
≤ 1.0= 19% vs. 47%1.1-2.0=51% vs. 40%2.1-3.0=30% vs. 13%
0.7-1.5= 25% vs. 66%1.6-3.0=60% vs. 24%3.1-4.0=15% vs. 10%
Nodule Morphology(%malig. pts vs. %benign pts.)
Edge characteristics reported:• Sharp, smooth=14% vs.20%• Lobulated=30% vs. 40%• Slightly irregular w/ few spiculations=38% vs.
33%• Grossly irregular and spiculated=19% vs. 7%
Not reported
Benign conditions(#pts.)
• histoplasma granuloma with active inflammation(2)
• other conditions not reported
• granuloma (7), coccidiomycosis (4), benign cellulardebris (4), nonspecific inflammation (3), necrotizinggranuloma (3)
• fibrosis (1), hemangioma (1), aspergillosis (1),metaplasia (1)
PET criteria forpositive node
focal FDG uptake > surrounding lung tissue, butmore than mild intensity
• focal uptake > mediastinal blood pool structures(qualitative)
• SUV> 2.5 (semiquantitative)CT criteria for noduleedge
based on 4-type scale to reflect degree ofspiculation and irregularity not specified to image interpreters
Interpretation of PET
• qualitative• 1 reader blinded to histology• blinding to clinical and radiologic information
varied
• semiquantitative using SUV• independent qualitative analysis using 2 readers
blinded to clinical , imaging, and histopathologic datareached by consensus
• readers interpreted studies with which they were notinvolved to ensure blinding
• interobserver variability calculated
Interpretation of CT• independent• 2 readers blinded to clinical diagnosis• consensus reading
• independent interpretation by > 1 reader blinded toclinical, PET, or histopathologic results
• qualitative interpretation as benign or indeterminateGold standarddetermination(# patients)
thoracotomy (36), mediastinoscopy (3),bronchoscopy (3), needle lung biopsy (9), follow-up imaging for > 2 yrs (1)
TTNA (29) or surgery (60)
Data analysis By patient By patient
TTNA=Transthoracic Needle Aspiration
December 1998
MTA98-032 MDRC Technology Assessment Program - PET Update - Page 38
Summary/discussion
Since the 1996 report, three additional studies using dedicated PET in diagnosingsolitary pulmonary nodules met the inclusion criteria for review. One was atechnical feasibility study, and two were of diagnostic accuracy assessing PET inthe test sequence after CT but prior to any histologic confirmation of disease.Both had significant biases in study design that warranted low methodologicquality scores and call for caution in generalizing these results to otherpopulations.
Most false negative results reported in the PET literature are caused by smallnodules with diameters commonly <1 cm that approach the resolution limits ofthe camera. Both studies reported false negatives comprising a variety of non-small cell cancers with diameters ranging from 1 cm to 2.5 cm. Moreover, theimpact of PET on treatment planning, particularly the decision to proceed tosurgery, was not systematically assessed.
One of the deficiencies outlined in the first report is the relatively low number ofpatients and a correspondingly narrow spectrum of benign conditions representedin the study base. Lowe (1998) presented the largest and only multi-site study ofPET in diagnosing SPNs. Multi-site trials have the advantage of recruiting largernumbers of patients with a comprehensive array of malignant and benignconditions that are needed to apply the results to other populations. The detaileddescription of the blinding procedures used in the study may serve as a model forfuture studies of PET.
Both studies derived likelihood ratios (LR) to quantify the importance of the PETresults in the work up of SPNs. As with predictive values, LRs are more usefulaccuracy measures to a clinician than sensitivity and specificity. LRs are used tocalculate the probability of disease given a test result. They are independent ofdisease prevalence in most circumstances, but differences in case mix andmethodologic biases can influence their validity (Gurney, 1993).
For example, the prevalence of malignancy in SPNs is lower in communityhospitals than in most surgical series or in tertiary care facilities, where most PETscanners are found. Areas that experience a higher prevalence of particularbenign conditions may encounter more false positive results on PET. A studywith too few patients with benign nodules may overestimate specificity andinflate the negative LR; presence of methodologic biases may overestimatesensitivity and inflate the positive LR. In both studies the inclusion criteriafavored a higher proportion of patients with malignancies and with too few benignconditions to offset the influence on specificity. Thus, rigorous study of a largernumber and range of patients with a mix of diseases is needed to derive validlikelihood ratios for PET in patients with SPNs.
December 1998
MTA98-032 MDRC Technology Assessment Program - PET Update - Page 39
Met
hodo
logi
cQ
ualit
y G
rade
D D
Stud
y De
sign
Lim
itatio
ns
S,R,
W,T
,d
s,r,W
,d
blinding
varie
d
+(fo
r visu
alan
alys
ison
ly)
goldstandard H
+ F
H
Evid
ence
-bas
ed M
edic
ine
Crite
ria
comparisongroup
+
+
PET
afte
r CT
Sem
iqua
ntita
tive
anal
ysis
(95%
CI)
Se=9
2% (8
2-10
0%)
Sp=9
0% (7
9-10
0%)
Acc=
91%
LRma
l=9.
0LR
ben=
0.09
Ove
rall v
alues
repo
rted
Ope
ratin
g Ch
arac
teris
tics
PET
afte
r CT
Visu
al a
naly
sis
(95%
CI)
Se=9
5%Sp
=87%
Acc=
92%
LRma
l=7.
11 (6
.36-
7.96
)LR
ben=
0.06
(0.0
5-0.
07)
Ove
rall v
alues
repo
rted
Se=9
8% (9
5-10
0%)
Sp=6
9% (5
7-81
%)
Acc=
89%
LRma
l=3.
0LR
ben=
0.02
Ove
rall v
alues
repo
rted
N 32 m
alig
nant
case
s14
ben
ign
case
s
60 m
alig
nant
case
s29
ben
ign
case
s
Stud
y
Dewa
n 19
97
Lowe
199
8
Tabl
e 15
: S
umm
ary
of th
e D
iagn
ostic
Acc
urac
y an
d D
iagn
ostic
Thi
nkin
g E
ffic
acy
Stu
dies
of P
ET
in In
dete
rmin
ate
S
olita
ry P
ulm
onar
y N
odul
es (S
PN
)
H =
histo
logy;
F =
Follo
w up
; S =
sm
all s
ize;
R =
refe
rral b
ias;
W =
wor
k up
bias
; T
= te
st re
view
bias;
D =
diag
nosti
c re
view
bias
(upp
er c
ase
indica
tes
signif
icant
limita
tion;
lowe
r cas
e ind
icate
s lim
itatio
nm
inim
ized
by s
tudy
des
ign,
pre
senc
e of
bia
s un
clear
, or s
mal
l effe
ct o
n op
erat
ing
char
acte
ristic
s)
Rol
e
Defin
ing
inde
term
inat
eSP
N
December 1998
MTA98-032 MDRC Technology Assessment Program - PET Update - Page 40
Once valid LRs are derived, they may be used to estimate the odds that a patienthas a cancer, given the PET result. Any attempt to use LRs in evaluating the oddsof cancer after PET requires: 1) knowledge of the odds of cancer before PET, and2) that the PET results were derived independent of the other test results. Inneither study were both conditions satisfied, and the influence of PET ondiagnostic certainty and subsequent treatment planning could not be determined.
Rigorous studies of patients comprising a range of pre-PET probabilities ofmalignancies are needed to assess the diagnostic accuracy and contributionof either dedicated or camera-based PET to the work up of solitarypulmonary nodules. Multiple sites may be needed to accrue a sufficientnumber and array of patients. Results from this review update confirm theconclusions and recommendations from the first report.
The Cooperative Studies Program of the VHA Office of Research andDevelopment has funded a multi-year cooperative study to determine theefficacy of FDG-PET in defining solitary pulmonary nodules (See SectionVIII). Results from this study should address the shortcomings of theexisting literature.
E. Colorectal Cancer
Colorectal cancer is the third leading cause of death among men and women, representinga significant public health problem in the United States. Colorectal cancers account forapproximately 11% of new cancer diagnoses. Death rates from colorectal cancer havefallen 25% for women and 13% for men during the past 20 years, reflecting a decreasingincidence of new cancer cases and increasing survival rates.
An estimated 131,600 cases and 56,500 deaths are attributable to colorectal cancer in theUnited States in 1998. An estimated 1 million veterans over the age of 50 will developcolorectal cancer over the remainder of their lives and nearly 433,000 will die from it(Wingo, 1995; Brown, 1996). Within the Veterans Health Administration, malignantneoplasms of the digestive organs and peritoneum (which include colorectal cancer)accounted for 8,280 discharges (1.2% of all discharges) with an average length of stay of15.7 days in FY 1997.
Winawer (1997) reported the following risk factors for colorectal cancer: age over 50years; a history of adenomatous polyps; a history of curative intent resection of colorectalcancer; inflammatory bowel disease; and familial colorectal cancer, adenomatouspolyposis, or hereditary nonpolyposis colorectal cancer.
Nationally, the estimated relative five-year survival rate among veterans is approximately40%, substantially lower than estimates from the general population of 62% (colon) and59% (rectum). In VA, the Office of Research and Development (ORD)’s EpidemiologicResearch and Information Center in Durham, North Carolina is conducting a four-yearinitiative to identify factors that may explain the worsened prognosis among veterans,and that may be responsive to intervention (Provenzale, 1998). ORD is also conducting alarge prospective study of risk factors and/or detection of altered cell proliferation for
December 1998
MTA98-032 MDRC Technology Assessment Program - PET Update - Page 41
large colonic adenomas in asymptomatic subjects; the results will have importantimplications for colon cancer screening (Lieberman, 1998).
Data on management of colorectal cancer are from the National Cancer Institute’sPhysician Desk Query (PDQ) system retrieved in October 1998. The most prevalenthistologic type of colorectal cancer is adenocarcinoma. Metastases to the liver,abdominal cavity, and extra-abdominal areas at initial diagnosis are common, as isrecurrent disease after surgical resection of the primary tumor. Prognosis andmanagement depends on the depth of tumor penetration into the bowel wall and thepresence of both regional lymph node involvement and distant metastases (staging).
Surgery is the primary therapy for colorectal cancer, and for cancers that have notmetastasized, it is frequently curative. Many patients with confined recurrent disease orwith metastases limited to the liver or lungs may also be amenable to resection.However, the high rate of recurrence and a troubling overall five-year survival rate callfor more appropriate selection of patients who may benefit from surgical resection. Themorbidity and costs associated with surgery for patients who do not have genuinelyresectable recurrent tumor could be avoided by improved methods of tumor detection.
Stotland (1997) reviewed several imaging modalities commonly used to stage anddiagnose colorectal cancer. The most common modalities include CT, MRI, endoscopicultrasonography (EUS), and transabdominal ultrasonography. The popularity of EUS, inparticular, has grown in recent years for its ability to image the depth of tumorpenetration into the bowel wall and regional lymph node involvement. MR endorectalcoils or ultrasound probes may be used to image rectal lesions. However, all structuralimaging modalities are circumscribed in their ability to determine the presence and extentof disease and disease recurrence. Information from newer modalities, such asintraoperative ultrasonography, immunoscintigraphy, arterioportography, and PET, mayincrease the accuracy of staging and detecting recurrence.
Potential roles for PET in colorectal management have been identified in the literature:
• Pre-operative staging, including diagnosing presence and extent of liver metastases,and;
• Post-operative monitoring of recurrent disease.
Five studies met the inclusion criteria for review. Of these, two were technical efficacystudies and are listed in the reference section. Table 16 lists the characteristics of tworetrospective case series and one prospective case series of diagnostic accuracy, andTable 17 summarizes the data and quality, representing the best evidence for the use ofPET in managing patients with colorectal cancer. All studies presented some anecdotalevidence of therapeutic efficacy.
Preoperative staging of colorectal cancer
The TA Program identified one small uncontrolled, unblinded technical feasibilitystudy of PET for staging initial primary colorectal cancer (Abdel-Nabi, 1998).
December 1998
MTA98-032 MDRC Technology Assessment Program - PET Update - Page 42
No diagnostic efficacy studies of staging primary colorectal carcinomas usingPET were identified for review.
Four relatively small case series presented evidence on the use of PET in patientswith suspected recurrent colorectal cancer, of which Ruhlmann (1997) was aretrospective technical feasibility study. The three remaining case series arediagnostic accuracy studies. Ogunbiyi (1997) and Flanagan (1998) areretrospective analyses from the same institution with overlapping studypopulations. Ogunbiyi (1997) studied 58 patients with a high suspicion forrecurrence, including some with advanced primary disease, based on clinicalsymptoms, elevated plasma carcinoembryonic antigen (CEA) concentration,and/or CT findings. Flanagan (1998) assessed the ability of PET to detectrecurrence in 22 asymptomatic patients with a post-operative elevated CEAconcentration and normal clinical and radiologic findings.
Delbeke (1997) presented the only prospective comparison of PET to CT and CTarterial portography (CTAP) in detecting liver and extrahepatic metastases in 52patients with suspected recurrent colorectal cancer. This is likely a continuationof an earlier, smaller study from the same institution (Vitola, 1996), which wasreviewed in the previous 1996 MDRC technology assessment.
In all studies PET was performed as an adjunct to the routine clinical andradiologic work up, but the initial work up was not described in detail. Currentevidence suggests that, when PET is added to the work up, there is improvedsensitivity in distinguishing recurrence from post-surgical changes anddocumenting the presence and extent of liver and more distant metastases.However, the methodologic shortcomings in these studies limit the validity ofthese estimates. Predictive values may be subject to considerable referral biasowing to the high suspicion for malignancy in the study population. Lack ofdocumentation of disease severity and underlying condition of the liver,completeness of the work up prior to PET, and blinding further hindersassessment of these results.
December 1998
MTA98-032 MDRC Technology Assessment Program - PET Update - Page 43
Table 16: Characteristics of Studies of Pre-operative Staging With FDG-PET inPatients with Suspected Recurrent Colorectal Cancer
StudyCharacteristics Delbeke et al. (1997) Ogunbiyi et al. (1997) Flanagan et al. (1998)
Perspective Prospective Retrospective Retrospective
Patient source 52 patients presented on 61occasions with suspectedrecurrent carcinoma• Consecutive series
58 patients who had PETbetween 1/91 and 1/95 withsuspected recurrent (n=47) oradvanced primary (n=11)disease
• ? Consecutive series
22 of 128 patients with history ofcolorectal cancer, who underwentPET from 6/93 to 6/96
• ? Consecutive series
Inclusion criteria • Elevated CEA levels orabnormal CT
• Abdominal CT (n=48);CTAP (n=40); or both(n=29)
• High clinical suspicion andequivocal or positive CTfindings (n=39)
• Elevated CEA levels withnormal CT (n=19)
• Normal CEA levels after initialresection
• Plasma CEA level > 5.0 ng/ml(mean 25 ng/ml), normalimaging studies, endoscopy,and physical exam on routinefollow-up
Patientcharacteristics
• 31 men, 21 women• Mean age 63 ± 11 yrs
• 33 men, 25 women• Mean age 60 yrs. (23-81 yrs)
• 17 men, 5 women• Ages 17-84• Primary site: colon (9), rectum
(10), rectosigmoid (2), appendix(1)
Extent of disease(#patients)
• Liver metastases (45)• Extrahepatic disease (26,
including 16 with liver mets)
• Primary disease or localrecurrence (21)
• Liver metastases (23)• Extrahepatic metastases (20)
• Stage B (10)• Stages C (5), C1 (2), C2 (3),• Stage D (2)
Benignconditions(# patients)
• Normal liver (7)• Post-surgical site (8)• Local fibrosis (2)• Resolving abscess (1),
hepatic cyst (1), hematoma(1)
Not reported in reproducibledetail
Not reported
PET criteria forpositive site
• Not specified for qualitativePET
• Cut-off not specified forsemiquantitative analysis
• Malignancy=FDG uptakemoderately or markedlyintense;
• Benign=no or mild uptake, orif abnormality identified onother imaging for which nocorresponding abnormalitywas present on PET
Not specified
Contrast CTcriteria forpositive site
• Not specified for surgicalcases
• In nonsurgical cases, anincrease in lesion volume >20% on serial scans
Not specified Not specified
CTAP criteria forpositive site
• Not specified for surgicalcases
• In nonsurgical cases, anincrease in lesion volume >20% on serial scans
N/A N/A
Interpretation • 2 readers for PET , 2readers for CT and CTAP,
• Independent, qualitativePET blinded to otherimaging results
• Semiquantitative PET SURcalculations excludedlesions < 1 cm in diameter
• Qualitative PET interpretedwith access to CT results
• Two readers• CT interpreted in “routine
clinical fashion”
• Qualitative PET scansinterpreted with access to CTresults
• Consensus of at least tworeaders
• CT interpreted in “routine clinicalfashion”
Gold standarddetermination(# patients)
• Clinical or radiologic followup (n=17)
• Histopathology obtainedsurgically (n=44)
• Percutaneous fine needleaspiration (n=2)
• Nonresected lesions=surgical exam andintraoperative ultrasound(unknown #)
• Surgery, histology, or both(n=40);
• Clinical and radiologic followup (n=16); autopsyreports(n=2)
• All patients followed for atleast 12 months after PET oruntil death
• Pathology (n= 9)• All patients had radiologic and
clinical follow up ≥ 6 months
Data analysis By lesion site By patient By patient
December 1998
MTA98-032 MDRC Technology Assessment Program - PET Update - Page 44
Each study presented some evidence on changes in patient managementattributable to PET, but the methods for assessment were not reported. Theevidence suggests that adding PET to the work up may help optimize treatment(e.g., improve patient selection for curative surgery) by documenting the presenceor absence of hepatic or more distant metastases. These data would need to beconfirmed in much larger prospective studies designed to systematically assessthe incremental value of PET against the many other available imaging modalitiesused in the work up of colorectal cancer.
Postoperative monitoring recurrent disease
The TA Program did not identify any studies in the published literature thataddressed the role of PET in routine postoperative monitoring of patients forrecurrent disease.
Summary/Discussion
Since the first report, five additional studies using dedicated PET in themanagement of colorectal cancer met the inclusion criteria for review. The bestevidence to support the use of PET in colorectal cancers are three reported caseseries of diagnostic accuracy, of which two were retrospective studies from thesame institution with overlapping study populations. All assessed the ability ofPET as an adjunct to CT and other diagnostic tests to stage potentially operablepatients with a high suspicion of recurrent disease; the one prospective case seriesalso included patients with advanced primary disease. No diagnostic accuracystudies of PET to stage early, primary disease were identified.
Current evidence suggests that to further define recurrent disease, PET added afterCT may offer improved sensitivity over CT alone. The absolute sensitivity ofimaging modalities in detecting hepatic and more distant metastases is difficult todetermine (Stark, 1987). Work-up bias is present when results from PET and/orother imaging tests under evaluation are used to direct biopsies to confirmsuspicious liver lesions or to direct the choice of the most appropriate referencemeasure. Biopsy resection, while not entirely perfect, is a very accurate referencemeasure.
All authors attempted to offset work up bias by confirming disease in unresectedpatients using less perfect truth measures, such as clinical and radiologic follow-up, surgical exam and palpation, and intraoperative ultrasound. Although usingthese truth measures may not adequately identify the number of false negatives,they are reasonable alternatives and are preferred over nothing. The extent towhich work up bias can be eliminated in this clinical setting is limited.
All of these studies had significant methodologic biases and insufficient reportingof fundamental design elements that preclude definitive assessment of studyvalidity. The accuracy estimates from these studies should be interpreted withcaution.
December 1998
MTA98-032 MDRC Technology Assessment Program - PET Update - Page 45
Met
hodo
logi
cQ
ualit
y G
rade
D D D D D
Stud
y De
sign
Lim
itatio
ns
S,R,
w,T
,d
S,r,T
,D
S,r,W
,t,d
S,R,
w,T
,d
S,r,w
,t,d
blinding
parti
al
parti
al
goldstandard
Surg
, H &
F
H &
F
Surg
, int
ra-
oper
ativ
eUS
, H &
F
Surg
, H, F
& au
tops
y
H &
F
Evid
ence
-bas
ed M
edic
ine
Crit
eria
comparison group + + + + +
CT a
lone
Se=5
7%Sp
=81%
PPV=
71%
NPV=
70%
Acc=
70%
Se=8
1%Sp
=60%
Acc=
78%
Se=7
4%Sp
=86%
PPV=
77%
NPV=
83%
Acc=
81%
Se=7
4%
CTA
P
Se=9
7%Sp
=5%
Acc=
80%
Ope
ratin
g Ch
arac
teris
tics
PET
afte
r CT
Se=9
0% (P
=0.0
08)
Sp=1
00%
PPV=
100%
NPV=
93%
Acc=
96%
Se=1
00%
Sp=7
1%PP
V=89
%NP
V=10
0%
Se=9
1%Sp
=95%
Acc=
92%
Se=9
6% (P
=0.0
2)Sp
=100
%PP
V=10
0%NP
V=97
%Ac
c=98
%
Se=1
00%
N 21 re
curre
nt c
ases
26 n
o re
curre
nce
15 re
curre
nt c
ases
7 no
recu
rrenc
es
104
mal
igna
nt le
sions
23 b
enign
lesio
ns
in 45
pat
ients
23 c
ases
with
dise
ase
35 n
o di
seas
e
34 m
alig
nant
lesio
ns5
benig
n les
ions
in 26
pat
ients
Stud
y
Ogu
nbiyi
(199
7)
Flan
agan
(199
8)
Delb
eke
(199
7)
Ogu
nbiyi
(199
7)
Delb
eke
et a
l.(1
997)
Tabl
e 17
: S
umm
ary
of D
iagn
ostic
Acc
urac
y S
tudi
es o
f FD
G-P
ET
in C
olor
ecta
l Can
cer
H =
histo
logy;
F =
Follo
w-up
; S =
sm
all s
ize;
R =
refe
rral b
ias;
W =
wor
k up
bias
; T
= te
st re
view
bias;
D =
diag
nosti
c re
view
bias
(upp
er c
ase
indica
tes
signif
icant
limita
tion;
lowe
r cas
e ind
icate
slim
itatio
n m
inim
ized
by s
tudy
des
ign,
pre
senc
e of
bia
s un
clear
, or s
mal
l effe
ct o
n op
erat
ing
char
acte
ristic
s)
Rol
e
Diag
nosin
g lo
cal
recu
rrenc
e
Dete
cting
liver
met
asta
ses
Dete
ctin
gex
trahe
patic
met
asta
ses
CTAP
=CT
with
arte
rial p
orto
grap
hy
December 1998
MTA98-032 MDRC Technology Assessment Program - PET Update - Page 46
All discussed changes in therapeutic management attributable to PET, but themethods for evaluation, details of the work up, or documentation of diseaseseverity among the cases were not described. To suggest that PET improves thepre-operative staging process for selecting more appropriate patients for resectionbased on the existing evidence is ill-advised.
The TA Program did not identify any studies evaluating the efficacy of PET inpost-operative monitoring. There is no consensus on the benefit of routineintensive follow-up after primary treatment, and the timing, frequency, type, andindications for post-operative follow-up using imaging are not standardized(Stotland, 1997). Any evaluation of PET in this role would be in the context ofuncertain benefits of such monitoring.
Appendix II lists two particularly relevant studies for staging colorectal cancerand could serve as models for future PET research. Notable design features arehighlighted. Zerhouni (1996) of the Radiology Diagnostic Oncology Groupconducted a large, multi-site trial to compare the relative accuracies of CT andMRI in staging primary colorectal cancer. Stark (1987) compared CT and MRI todetect liver metastases, an important aspect of staging colorectal cancer patients.Studies of PET that incorporate these features with the comparable level of detailwould provide more robust data on which to more confidently judge the addedvalue of PET in the work up of colorectal cancer.
The TA Program concludes that the prevailing evidence does not support theroutine use of either dedicated or camera-based PET in the management ofcolorectal cancer. Larger, prospective studies of diagnostic accuracy andsubsequent therapeutic efficacy of PET in the work up are needed.Methodologically rigorous studies of diagnostic imaging have been publishedthat may serve as models for guiding design of future PET research. Reviewof the recent evidence confirms the conclusions from the first report.
F. Alzheimer’s Disease
This section briefly summarizes Alzheimer’s disease (AD) and presents updatedepidemiological information and results of a systematic review of the literatureevaluating PET using FDG as a diagnostic test in AD. Appendix 8 of the MDRCtechnology assessment report on PET (Flynn, 1996) provides an expanded discussion ofthe disease, diagnosis, treatment, methodological and ethical considerations, andalternative neuroimaging technologies and other relevant diagnostic tests used in AD.
Unless otherwise noted, epidemiological information is from a consensus statement of theAmerican Association for Geriatric Psychiatry, the Alzheimer’s Association, and theAmerican Geriatrics Society (Small, 1997). AD, a progressive neurodegenerativedisorder, is the most common form of dementia and affects an estimated 4 million peoplein the United States. AD is characterized by steady irreversible decline in cognition,functioning, and behavior with sparing of motor and sensory functions until later stages.The rate of progression is variable, but duration of illness from diagnosis to death isapproximately 10 years.
December 1998
MTA98-032 MDRC Technology Assessment Program - PET Update - Page 47
The reported prevalence of AD is approximately 6-8% of all persons 65 years or older. Itdoubles every 5 years after the age of 60 years, so that about 30% of the population olderthan 85 years will have AD. By the next century, an estimated 600,000 veterans withsevere dementia will require long-term institutional care (ORD Impacts, 1997). Thedirect and indirect costs for care of AD patients in the United States approach $100billion annually. The true costs of AD to society is likely much more, as economicassessments frequently underestimate the economic and emotional burden imposed on thecaregivers as well as the patients.
Hendrie (1998) recently summarized the achievements in understanding genetic andnongenetic risk factors associated with AD. Genetic risk factors account for about 2% ofall AD cases. Both causative (mutations on chromosomes 1, 12, 14, and 21) andassociative genes (APOE-4 allele2 on chromosome 19) for AD have been identified. InVA, ORD researchers are: 1) studying genetic and environmental factors that contributeto delayed onset of AD in subjects with chromosome 1 mutations (ORD, 1997), and 2)are following subjects with the APOE-4 allele at higher risk for developing AD to betterdetect and characterize early stages of this disease (Bondi, 1997).
Diagnostic tests that detect the presence of the APOE-4 allele for apolipoprotein E, aserum lipoprotein involved in cholesterol transport, are under investigation, but expertsdiffer on its usefulness. Since the APOE-4 allele is found in many elderly personswithout AD and is not always found in patients with AD, the Working Group of theAmerican Medical Genetics/American Society of Human Genetics concluded thatpredictive testing of APOE-4 for AD should not be done.
The only nongenetic risk factors consistently associated with risk for AD are age andfamily history. Other possible risk factors with a predominately positive associationinclude low education, depression, estrogen-replacement therapy, nonsteroidal anti-inflammatory drugs (NSAIDs). Female gender, head injury, hypothyroidism and, to alesser extent, insulin-dependent diabetes, aluminum exposure and smoking areinconsistently associated with an increased risk for AD. Clinical trials examining the roleof estrogen, NSAIDs, and vitamin E in AD are reportedly underway.
The primary role of diagnostic testing is the differential diagnosis of AD from otherreversible or treatable dementias. A definitive diagnosis is based on a typical clinicalpicture and histopathologic sampling of brain tissue at autopsy. In the absence ofhistologic confirmation, patients with probable AD are often referred to as havingdementia of the Alzheimer’s type (DAT). Two distinct sets of antemortem clinicalcriteria from the following may be used to characterize patients with DAT:
• (NINCDS/ADRDA)--National Institute of Neurologic and Communication Disordersand Stroke and the Alzheimer’s Disease and Related Disorders Association
• (DSM-IIIR or the more recent DSM-IV)--Diagnostic and Statistical Manual forMental Disorders, American Psychiatric Association.
2 In Mendelian genetics, an allele is any alternative form of a gene at a given locus. An allele may express adominant, a recessive, or an intermediate trait.
December 1998
MTA98-032 MDRC Technology Assessment Program - PET Update - Page 48
While advanced stage AD is usually easier to diagnose, early stage disease can beproblematic. There is no cure for AD, but psychosocial techniques for behavioralproblems associated with dementia and drug therapies for cognitive impairment havebeen developed, which can improve quality of life. HSR&D researchers found that twoapproaches improve quality of care and reduce costs associated with caring for ADpatients: 1) simulated presence therapy, which uses selected memories through taperecorded conversations to manage problem behaviors in AD patients (Camberg, 1999);and 2) hospice care for managing AD patients with advanced dementia (Volicer, 1994).
New therapy aimed at slowing disease progression is also available. Since it is mosteffective if given at the earliest stages of AD, there is a need for obtaining earlier andmore accurate antemortem diagnoses. Such information would also help patients andtheir families better prepare for future challenges. Functional imaging technologies suchas PET and SPECT have been used to improve diagnostic certainty and to provideinformation on the pathophysiologic basis of AD.
Eight studies of technical efficacy using only dedicated PET scanners met the inclusioncriteria for review. The TA Program was unable to identify published PET studies athigher levels of the Fryback and Thornbury diagnostic efficacy hierarchy. The followingtable summarizes information from these studies. All studies used FDG-PET to studyregional cerebral glucose metabolic rates; Ishii (1997) also measured cerebellar glucosemetabolic rates.
Evidence from recent technical efficacy studies shows a growing interest in the use ofPET to better understand the biological mechanisms of neurodegenerative disease. Theresearch suggests a link between cognitive function, functional imaging data, and theneurobiology of dementia. There is also increasing emphasis in these studies onimproving methods for detecting early stage AD by improving the measurement ofregional brain function. More precisely defined neuroanatomical atlases and methods ofanalysis may help explain the underlying pathophysiology of AD and the differencesbetween diseases and disease progression.
Results from Imamura (1997) and Vander Borght (1997) underscore the limitations inexisting knowledge using PET to diagnose AD. That is, while the temporal and parietalmetabolic patterns often differentiate AD from other causes of dementia, AD also sharesfunctional imaging features with other causes.
December 1998
MTA98-032 MDRC Technology Assessment Program - PET Update - Page 49
Table 18: Summary of Recent Technical Efficacy Studies Using FDG PET inAlzheimer’s Disease
Study Objective Findings suggest…Desgranges et al.(1998)
N = 19
To study the neuronal basis for memoryimpairment in AD using Tulving’s hierarchicalmodel of memory systems and PETmeasurement of resting regional cerebralglucose utilization
• Their methodology for mapping neuronal substrates of cognitiveimpairment are valid and useful.
Higuchi et al. (1997)
N = 20
To examine regional cerebral glucosemetabolism using PET in AD patients withdefined genetic risk factors (APOE-4, ACT,and PS-1 genotypes)
• APOE-4 does not adversely affect the AD process or preserve brainmetabolism after clinical onset of AD.
• ACT gene has deleterious effects on cerebral glucose metabolismduring the clinical stages of AD.
• Differences in cerebral regions are influenced by the two genes.• Inheritance pattern of the two alleles may explain divergent patterns
of progression in AD.
Imamura et al. (1997)
N = 38
To study regional cerebral glucosemetabolism in AD vs. dementia with Lewybodies (DLB)
• There are differences in regional glucose hypometabolism consistentwith the pathological and neurochemical differences between DLBand AD.
• FDG-PET may help in the clinical discrimination between DLB andAD.
Ishii et al. (1997)
N = 81
To study regional cerebral and cerebellarglucose metabolic rates in AD
• There is a significant cerebellar glucose metabolic reduction in severeAD with no apparent cerebellar atrophy.
• AD is a global degenerative brain disease in which degeneration iscorrelated with severity.
• Method of analysis using normalization of regional glucose metabolicdata to cerebellar values may be liable to err in severe AD patients.
Pietrini et al. (1997)
N = 16
To study regional glucose metabolism understress using an audiovisual paradigm innondemented adults with trisomy 21 Down’ssyndrome
• There are no differences in metabolism at rest.• In older subjects had significantly lower glucose metabolic rates in the
parietal and temporal cortical areas.• A stress test paradigm can detect metabolic abnormalities in the
preclinical stages of AD.
Stein et al. (1998)
N = 50
Using a template of Brodmann areas derivedfrom whole brain histological section atlas toanalyze glucose metabolic rates in ADpatients
• Vulnerability is greatest in cortical areas that are in closer synapticcontact with limbic areas.
• Integrating statistical techniques of brain imaging intoneuroanatomical atlases and incorporating fine-tuned calibration ofneuroanatomical studies into brain-imaging analyses, may increasecorrelation of findings and a more complete characterization of thepathophysiology of AD.
Vander Borght et al.(1997)
N = 27
To study regional cerebral glucosemetabolism in AD vs. Parkinson’s diseasewith dementia (PDD)
• AD and PDD may share common features in the patterns ofmetabolic alterations and also presence of regional metabolicdifferences in the visual cortex and in the medical temporal cortex.
• These differences may help explain different degrees andcombinations of disease specific underlying pathological andneurochemical processes.
Yamaguchi et al. (1997)
N = 23
To study regional glucose metabolism inhippocampal atrophy in AD
• Morphologic asymmetry of the hippocampus and a metabolicasymmetry of the temporoparieto-occipital were correlated.
• These asymmetries are present in early stage AD.
Summary/Discussion
Recent evidence exploits functional imaging technologies such as PET forpathophysiologic information that may be applied toward earlier preclinicaldiagnoses of AD. Jagust (1996) highlighted the importance and the complexitiesof obtaining earlier and more accurate diagnoses of AD:
• Earlier diagnosis is important for understanding the biological mechanisms ofAD;
December 1998
MTA98-032 MDRC Technology Assessment Program - PET Update - Page 50
• Clinically, early diagnosis becomes more critical, as treatments becomeavailable;
• Information from early diagnoses may enable forecasting which elderlypersons who experience memory lapses will develop dementia;
• Normal aging processes can complicate early diagnosis; and• Research should also assess factors key to the production of disease
symptoms.
The best evidence demonstrating the accuracy of FDG PET in diagnosingAlzheimer’s disease is from four published studies reviewed by Flynn (1996).They are listed in the Alzheimer’s disease references (Section XI). Althoughthese studies reported good diagnostic accuracy for PET in AD, the diagnosticutility of PET remains controversial:
• While each set of clinical criteria has different associated sensitivity,specificity, and likelihood ratios, careful application of the clinical criteriadoes appear to identify most cases of treatable dementia.
• Sources of bias attributed to the spectrum and severity of disease, the use ofclinical criteria as the gold standard, and the choice of clinical criteria(NINCDS/ADRDA versus DSM-IIIR or DSM-IV) may have influenceddiagnostic accuracy estimates in these studies.
• Few studies applied PET prospectively to large numbers of patients with aspectrum of dementia and disease severity, which would be necessary todefine the positive predictive value of PET as a diagnostic test, and followedthem until death.
Flynn (1996) reported that a cooperative group of European PET centers isconducting such a study. The study will include patients with NINCDS/ADRDA“possible” AD, the patients in whom there is the greatest uncertainty regardingdiagnosis and for whom a more accurate test would most contribute to posttestcertainty.
Small (1997) suggested that improved diagnostic information to patients and theirfamilies may allow families to better prepare for the challenges ahead and thatearly and accurate diagnosis may prevent the use of costly medical resources.The TA Program was unable to locate any studies of PET that assessed the impactof PET on the costs associated with caring for patients with AD.
Flynn (1996) concluded that existing evidence argues against routine clinicaluse of PET for diagnosing AD until more effective treatments and riskmodification interventions for AD are developed, and until meaningful androbust predictive values are obtained from an ongoing European multicenterPET study.
December 1998
MTA98-032 MDRC Technology Assessment Program - PET Update - Page 51
The value of improved diagnostic information to AD patients and theirfamilies should not be dismissed; however, this value should be quantified inthe context of accessibility and accuracy of alternative imaging technologiesand of phenotypically or genetically defined subsets of AD. In the absence ofeffective treatments for AD, an accurate diagnostic test may be neededprimarily in research for epidemiologic studies and evaluations of potentialtherapies.
IX. ONGOING CLINICAL STUDIES AND ON-LINE RESOURCES
Several on-line sources provide useful information about ongoing clinical trials:
• CenterWatch Clinical Trials Listing Service [http://www.centerwatch.com]
• NIH Clinical Research Studies [http://clinicalstudies.info.nih.gov/]
• NCI cancerTrials PDQ database search [http://cancertrials.nci.nih.gov/]
These on-line sources were searched in November 1998 for active clinical trials studying theefficacy of FDG PET. Thirty-eight active protocols using FDG PET were retrieved, of which thefollowing six protocols are assessing diagnostic PET for the conditions reviewed in this report.
Table 19: Active NIH Trials of FDG PET in Selected Cancers and Alzheimer’s Disease
PROTOCOL COMMENTS
NCI-94-C-0151Diagnostic study of PET in patients with stage II-IV or recurrent breast cancer
Single-site
Sponsor- NCI
Start date 1994Active accrual for at least 3 years
MSKCC-97046, NCI-G97-1308Comparison of positron emitter Iodine I124 Iododeoxyuridine with fludeoxyglucose F 18 (F-18-Fluoro-2-Deoxy-(D)-Glucose) as a tracer for glycolysis on scans and in tumor samples in patients with advanced breastcancer
Single site
Sponsor- local funding *
Start date (1997)Active accrual for about 1 year
NCI-97-C-0068Phase II study of Anti-CEA antibody immunoscintigraphy and PET in the localization of recurrent colorectalcarcinoma in patients with rising serum CEA levels in the absence of imageable disease by conventionalmodalities
Single site
Sponsor- NCI
Start date (1997)Active accrual for 3 years
MSKCC-96079, NCI-G97-1334Phase II/III Diagnostic Study of Whole Body PET to measure the response to induction chemotherapy ofpotentially resectable lung and esophageal carcinomas
Single site
Sponsor- local funding *
Start date (1997)Active accrual open
NCI-98-C-0163The use of PET and MRI to assess the effects of anti-neoplastic therapy on tumor associated vasculature
Unknown
Sponsor- NCI
Start date 1998Accrual pending
81-N-0010Study of regional cerebral utilization of glucose in organic dementia and Down syndrome by the Laboratory ofNeurosciences of The National Institute on Aging
Unknown
Sponsor-National Institute of NeurologicalDisorders and Stroke (NINDS)
Start date 1981Active accrual
* Personal communication: Dr. Steven Larson, Memorial Sloan Kettering Cancer Center, New York
Since there is no central repository for locating active clinical trials of PET, these sources maynot provide a complete listing of all multi-site studies evaluating PET as a clinical test.
December 1998
MTA98-032 MDRC Technology Assessment Program - PET Update - Page 52
Consequently, individuals actively involved in the use and evaluation of PET were queried fortheir knowledge of other relevant cooperative trials.
• NCI is funding a multi-center trial of FDG PET in staging breast cancer. The primary goalis to assess the accuracy of PET for detecting the presence, absence, and extent of axillarynodal metastases in women with newly diagnosed breast cancer; a secondary endpoint willevaluate PET for detecting internal mammary nodal disease as a prognostic indicator(personal communication: Dr. Barry Siegel, Washington University, St. Louis, Missouri).
• NCI is sponsoring a new cooperative group within the American College of Surgeons calledthe American College of Surgeons Oncology Group (ACoSOG) (NIH, 1998). The ACoSOGwill design and conduct cooperative trials in surgical oncology. The primary goal of theACoSOG is to evaluate surgical approaches for diagnosis and treatment of patients withmalignant solid tumors. Patients with the most common cancers of the breast, lung, andcolo-rectum will be studied initially. Completion of two protocols comparing theincremental value of PET to conventional staging in potentially operable patients with lungcancer and esophageal cancer is imminent (personal communication: Dr. Barry Siegel).
• The Southwest Oncology Group (SWOG) is developing a companion study within a PhaseIII cooperative trial comparing surgery and pre-operative chemotherapy for patients withlung cancer. The companion study will evaluate PET in assessing tumor response tochemotherapy. Both studies will be activated in 1999 (personal communication: SuzanMyers, SWOG).
X. OTHER SYSTEMATIC REVIEWS OF PET
Since 1996 several organizations have conducted assessments to support evidence-basedrecommendations for the use of PET as a diagnostic test (See Appendix V). The majority ofassessments were qualitative systematic reviews of dedicated PET used in neurology to diagnoseand manage patients with medically refractory partial seizures, central nervous system tumors,and cerebrovascular disease. Recent systematic reviews reflect an increasing interest in PET andin other positron imaging modalities to manage patients with non-central nervous systemcancers, emphasizing staging non-small cell lung cancer.
For the indications in this review, the findings of assessments with either full text or abstracts inEnglish in the public domain, or otherwise available to the MDRC, are summarized below:
• There is general agreement that the evidence on FDG-PET for diagnosing, staging ormonitoring treatment of primary cancers outside the lung is not firmly established.
• There is general agreement that the effect of PET on the management of patients withprimary lung cancers is not known.
The Agencia de Evaluación de Tecnologías Sanitarias (AETS) in Spain, the Committee forEvaluation and Diffusion of Innovative Technologies (CEDIT) in France, and the NHSHealth Technology Assessment Programme (NHS HTAP) in the United Kingdomrecommend comparative studies of effectiveness and of the diagnostic contribution of
December 1998
MTA98-032 MDRC Technology Assessment Program - PET Update - Page 53
dedicated PET (and, in some cases, coincidence imaging gamma cameras) in patients withlung cancer.
• Assessment findings and recommendations are mixed regarding the use of PET todiagnose and stage non-small cell lung cancer and solitary pulmonary nodules (SPNs).
Two agencies, AETS and the NHS HTAP, used VA review methods and frameworks toupdate and/or expand the first VA PET report (Flynn, 1996). Both reports confirmed VA’soriginal findings that the evidence for the diagnostic efficacy of PET in managing patientswith lung cancer was insufficient. Blue Cross/Blue Shield Association found that FDG-PETimaging meet their quality assessment criteria for staging mediastinal lymph nodes andcharacterizing radiographically indeterminate SPNs, provided the test results could changemedical management (HCFA, 1997).
An ECRI quantitative analysis determined that for both lung cancer indications PET is cost-effective when used to confirm resectability, but that PET is not cost-effective when usedearlier in the diagnostic algorithm. A SPN strategy using CT for initial diagnosis, needlebiopsy to confirm positive results, and PET to confirm negative results attained the greatestlife expectancy (Mitchell, 1998).
There are several possible reasons for the discrepancies across these assessments. Variationsin criteria for including published studies and for judging the quality of the included studies,in analytical methods, in the rationale for the assessment, and in the focus of the report arelikely causes. Often, assessments must be purchased or may require language translation tobe systematically evaluated. For this review, the MDRC considered information availableonly in the public domain in English or with English translation. Proprietary or non-translated reports may have derived different conclusions. Valid comparisons of technologyassessments that address similar topics are critical to health care organizations wishing toestablish policies based on the best available evidence.
Increasingly, agencies are using quantitative analyses, (e.g., decision analyses, meta-analyses, and cost-effectiveness analyses) to quantify the utility of clinical PET. Manyanalyses extrapolate existing diagnostic accuracy estimates to population impact, or poolaccuracy results from multiple studies. It is important to note that the validity of the studiesthat are the source of these estimates is an essential consideration when evaluating therobustness of the results (Petitti, 1994).
• Until recently, agencies considered only dedicated PET scanners, but now are asked toreview other positron imaging modalities.
An expert panel at CEDIT considered coincidence imaging gamma cameras and dedicatedPET in their recommendations. The NHS HTAP report will include evaluations of partialring PET, coincidence imaging gamma cameras, and collimated 511 keV imaging.
PET is a topic for a joint project of the International Network of Agencies for Health TechnologyAssessment (INAHTA), to which the TA Program belongs. The TA Program is coordinating theproject with members from Spain and the Agency for Health Care Policy and Research. Memberagencies are collaborating to synthesize their assessments of clinical PET applications into a
December 1998
MTA98-032 MDRC Technology Assessment Program - PET Update - Page 54
single, broadly applicable document. The report will also include a description of the evolutionof PET use in the United States and current indications and coverage policies of PET amongcountries represented by INAHTA members. The report will be available in 1999 on theINAHTA web site at [http://www.inahta.org].
XI. CONCLUSIONS
A. Experience in VA
VHA continues to make a substantial resource commitment to its PET imaging facilities.This commitment has the potential to help support two parts of VHA’s mission: researchand clinical care. The medical community regards PET as an important basic researchtool. A survey of active funded research at VHA PET sites underscores this importance,with the vast majority of basic research activity in neurology and cardiology. VHA ismaximizing its investment in PET by supporting high quality outcomes research andsystematic collection of utilization data.
All VHA PET sites have access to FDG, enabling them to conduct glucose metabolicstudies for various clinical applications. The number of PET oncology studies conductedacross VHA PET facilities from FY 1994 to FY 1998 has nearly quadrupled, likelyreflecting the positive changes in Medicare and private sector reimbursement andchanges in practitioners’ attitudes. Since VHA continues its moratorium on addingdedicated PET centers to its system, many VA medical centers without access todedicated PET scanners are adapting existing dual-headed gamma cameras forcoincidence detection.
B. Systematic reviews
The prevailing evidence does not support the use of either dedicated or gamma camerasmodified for coincidence detection (camera-based PET) as a diagnostic test for theapplications in this review. All studies were subject to considerable bias, which will haveresulted in overestimating accuracy and clinical value. Several studies presentedanecdotal data on the influence of PET on changing diagnostic certainty and treatmentplanning, but the methods for assessing these changes were not described, and thesystematic nature could not be determined.
Caution must be exercised to not apply accuracy estimates from dedicated PET tocamera-based PET systems. Whereas dedicated PET scanners are limited primarily totertiary care institutions, dual-headed gamma camera systems are more widely employed.Technical differences between the two systems and potential differences in the studypopulations represented across different health facilities emphasize the need for large,rigorous studies of diagnostic efficacy to define the clinical role of camera-based PET.
The TA Program identified several methodologically rigorous studies of other diagnosticimaging modalities that could serve as models for designing future PET research(Appendix II). Incorporating aspects from these studies would correct the methodologicshortcomings of the existing literature and strengthen the evidence on which to basefuture patient care decisions.
December 1998
MTA98-032 MDRC Technology Assessment Program - PET Update - Page 55
Qualitative systematic reviews produced by other technology assessment agencies, whichused methods similar to the VA PET report, reached similar conclusions. Most agenciesagree that the effect of positron imaging on managing patients with cancer needs furtherstudy. Several cooperative trials and other data collection efforts are ongoing or are beingproposed that may address many unanswered questions regarding the utility of FDG PETin the work up of patients with cancer and Alzheimer’s disease. Clinicians should awaitthe results of these efforts before incorporating PET into routine diagnosticstrategies. Nonetheless, variations across studies in study populations, imagingprotocols, threshold values, and formulae for calculating quantitative uptake values maylimit the generalizability of the findings to other institutions and populations. Review ofrecent evidence confirms the conclusions from the original VA PET assessment(Flynn, 1996).
Information on some of the cooperative trials can be accessed through on-line datasources. Advocates of clinical PET and decision makers interested in its clinicalutility would benefit from an accessible central repository containing information onexisting and proposed rigorously designed cooperative trials of PET. This sourcecould help guide the diffusion of PET into clinical care, as its usefulness and contributionto improved patient outcomes are appropriately evaluated.
December 1998
MTA98-032 MDRC Technology Assessment Program - PET Update - Page R - 1
XII. REFERENCES
Background
Positron Emission Tomography. In: ECRI, Ed. Health Technology Forecast. PlymouthMeeting, PA: ECRI; 1996.
Food and Drug Modernization Act of 1997. s830. 1997, 105th Congress: Washington, DC.
Health Care Financing Administration. Health Care Financing Administration. (1999).Carriers manual part 3. Chapter IV- Claims review and adjudication procedures. [Web site].Available: http://www.hcfa.gov/pubforms/14%5Fcar/3b4120.htm, [July 20, 1999].
Illumina Interactive, Boston, MA. (1998). CenterWatch Clinical Trials Listing ServiceTM.[Web Site]. Available: http://www.centerwatch.com/, [May 27, 1998].
18F-Labeled 2-Deoxy-2-Fluoro-D-Glucose Positron-Emission Tomography Scans for theLocalization of the Epileptogenic Foci. Health Technology Assessment Number 12. AHCPRPub. No. 98-0044. 1998, U.S. Department for Health and Human Services. Agency for HealthCare Policy and Research: Rockville, MD.
American Cancer Society. Cancer facts & figures - 1998. 1998, American Cancer Society:Atlanta, GA.
Balch CM, Singletary SE, Bland KI. Clinical decision-making in early breast cancer [seecomments]. Ann Surg 1993; 217(3): 207-225.
Begg CB. Biases in the assessment of diagnostic tests. Stat Med 1987; 6(4): 411-423.
Bombardieri E, Crippa F, Maffioli L, Greco M. Nuclear medicine techniques for the study ofbreast cancer. Eur J Nucl Med 1997; 24(7): 809-824.
Bondi MW. U.S. Department of Veterans Affairs. Medical Research. (April 5, 1997).Cognitive and Imaging Studies of Older Adults with the ApoE-E4 Allele. [Web site]. Available:http://www.va.gov/research/files/VA011308.HTM, [April 28, 1999].
Brown J. Annual report of the Secretary of Veterans Affairs. 1997, Department of VeteransAffairs: Washington, DC.
Buscombe JR, Cwikla JB, Thakrar DS, Hilson AJ. Scintigraphic imaging of breast cancer: areview. Nucl Med Commun 1997; 18(8): 698-709.
Camberg L, Woods P, Ooi WL, Hurley A, Volicer L, Ashley J, et al. Evaluation of simulatedpresence: a personalized approach to enhance well-being in persons with Alzheimer's disease. JAm Geriatr Soc 1999; 47(4): 446-452.
Coleman RE. Camera-based PET: the best is yet to come [editorial]. J Nucl Med 1997; 38(11):1796-1797.
December 1998
MTA98-032 MDRC Technology Assessment Program - PET Update - Page R - 2
Cook DJ, Guyatt G, Laupacis A, Sackett DL. Rules of evidence and clinical recommendationson the use of antithrombotic agents [published erratum appears in Chest 1994 Feb;105(2):647].Chest 1992; 102(4 Suppl): 305S-311S.
ECRI. Diagnostic imaging. Health Technology Forecast 1996; 7-16.
Eysenck HJ. Meta-analysis and its problems. Bmj 1994; 309(6957): 789-792.
Fahey FH. State of the art in emission tomography equipment. Radiographics 1996; 16(2): 409-420.
Feussner JR. Refining Research Priorities: New Initiatives Meeting Veterans Needs. 1997,Veterans Health Administration. Office of Research & Development: Washington, DC.
Flynn K, Adams E, Anderson D, et. al. Positron emission tomography: descriptive analysis ofexperience with PET in VA and systematic reviews: FDG-PET as a diagnostic test for cancerand Alzheimer's disease. 1996, U.S. Department of Veterans Affairs. Management Decision &Research Center: Washington, DC.
Food and Drug Administration. Revocation of certain guidance documents on Positron EmissionTomography drug products. Federal Register 1997; 62(244): 66636.
Food and Drug Administration. Revocation of regulation on Positron Emission Tomographydrug products. Federal Register 1997; 62(244): 66522.
Fryback DG, Thornbury JR. The efficacy of diagnostic imaging. Med Decis Making 1991;11(2): 88-94.
Goodwin WJ. PET and recurrent squamous cell carcinoma of the head and neck: a surgeon'sview. AJNR: Am J Neuroradiol 1998; 19(7): 1197.
Gurney JW. Determining the likelihood of malignancy in solitary pulmonary nodules withBayesian analysis. Part I. Theory. Radiology 1993; 186(2): 405-413.
Guyatt GH, Sackett DL, Sinclair JC, Hayward R, Cook DJ, Cook RJ. Users' guides to themedical literature. IX. A method for grading health care recommendations. Evidence-BasedMedicine Working Group [published erratum appears in JAMA 1996 Apr 24;275(16):1232].JAMA 1995; 274(22): 1800-1804.
Haynes RB, Sackett D. Purpose and procedure (abbreviated). Evidence-based Medicine 1995; 12.
December 1998
MTA98-032 MDRC Technology Assessment Program - PET Update - Page R - 3
Health Care Financing Administration. Health Care Financing Administration. (1998).Transmittal No. AB-97-27. Implementing Instructions - PET Scans for Characterizing SolitaryPulmonary Nodules or Staging Lung Cancer Performed on or After January 1, 1998. [WebSite]. Available: httP//www.hcfa.org/pubforms/transmit/AB972760.htm/, [January 13, 1999].
Health Care Financing Administration. Health Care Financing Administration. (1998).Transmittal No. A-98-9. Positron Emission Tomography(PET) Scans-ACTION. [Web Site].Available: http://www.hcfa.gov/pubforms/transmit/a98960.htm/, [March 23].
Hendrie HC. Epidemiology of dementia and Alzheimer's disease. Am J Geriatr Psychiatry1998; 6(2 Suppl 1): S3-18.
Hoffman JM, Hanson MW, Coleman RE. Clinical positron emission tomography imaging.Radiology Clinics of North America 1993; 31 935.
Ibrahim MA. Epidemiology and health policy. Rockville, Md.: Aspen Systems Corp; 1985.
Institute for Clinical PET. Institute for Clinical PET, Foothill Ranch, CA. (Feb. 2, 1998). U.S.PET Centers. [Web Site]. Available: http://www.icppet.org, [May 3, 1999].
International Network of Agencies for Health Technology Assessment. International Network ofAgencies for Health Technology Assessment. (1999). Homepage. [Web site]. Available:http://www.inahta.org/, [April 28, 1999].
Ioannidis JP, Cappelleri JC, Lau J. Issues in comparisons between meta-analyses and largetrials. JAMA 1998; 279(14): 1089-1093.
Jagust WJ. Functional imaging patterns in Alzheimer's disease. Relationships to neurobiology.Ann N Y Acad Sci 1996; 777 30-36.
Jarritt PH, Acton PD. PET imaging using gamma camera systems: a review. Nucl MedCommun 1996; 17(9): 758-766.
Kent DL, Larson EB. Magnetic resonance imaging of the brain and spine. Is clinical efficacyestablished after the first decade? [published erratum appears in Ann Intern Med 1988 Sep1;109(5):438]. Ann Intern Med 1988; 108(3): 402-424.
Kent DL, Larson EB. Disease, level of impact, and quality of research methods. Threedimensions of clinical efficacy assessment applied to magnetic resonance imaging. Invest Radiol1992; 27(3): 245-254.
Kent DL, Haynor DR, Longstreth WT, Larson EB. The clinical efficacy of magnetic resonanceimaging in neuroimaging [see comments]. Ann Intern Med 1994; 120(10): 856-871.
Larson S. Personal communication. November 19, 1998.Laupacis A, Wells G, Richardson WS, Tugwell P. Users' guides to the medical literature. V.How to use an article about prognosis. Evidence-Based Medicine Working Group. JAMA 1994;272(3): 234-237.
December 1998
MTA98-032 MDRC Technology Assessment Program - PET Update - Page R - 4
Lieberman DA. U.S. Department of Veterans Affairs. Cooperatives Studies Program. (1998).Prospective Evaluation of Risk Factors for Large Colonic Adenomas in Asymptomatic Subjects.[Web Site]. Available: http://www.va.gov/research/files/VA011064.HTM, [April 28, 1999].
Lillington GA, CI. C. Evaluation and management of solitary and multiple pulmonary nodules.Clinics in Chest Medicine 1993; 14(1): 111-119.
Maublant J. Scintigraphic imaging of breast tumors. Eur J Radiol 1997; 24(1): 2-10.
Mitchell MD, Turkelson C, Doggett D. Cost-effectiveness analysis of PET in lung cancerdiagnosis and staging. Annual Meeting of the International Society for Technology Assessmentin Health Care, Abstracts. Vol. 14; 33, 1998.
Mushlin AI, Detsky AS, Phelps CE, O'Connor PW, Kido DK, Kucharczyk W. The accuracy ofmagnetic resonance imaging in patients with suspected multiple sclerosis. The Rochester-Toronto Magnetic Resonance Imaging Study Group. JAMA 1993; 269(24): 3146-3151.
National Institute of Health. Warren Grant Magnuson Clinical Center, NIH. (1997). NIHClinical research studies protocol database. [Web Site]. Available:http://clinicalstudies.info.nih.gov/, [July 7, 1997].
National Institutes of Health. National Cancer Institute. National Institutes of Health,. (March27, 1998). American College of Surgeons and National Cancer Institute Announce Funding ofClinical Trials Project. [Web site]. Available: http://www.nih.gov/news/pr/mar98/nci-27.htm, [April 28, 1999].
PDQ. National Cancer Institute. (1998). PDQ clinical trials search. [Web Site]. Available:http://cancernet.nci.nih.gov/prot/patsrch.shtml/, [December 1998].
PDQ. National Cancer Institute. (1999). PDQ Treatment Health Professionals: Lip and oralcavity cancer. [Web Site]. Available: http://cancernet.nci.nih.gov/, [April 30, 1999].
PDQ. National Cancer Institute. (1999). PDQ Treatment Health Professionals: Colon Cancer.[Web Site]. Available: http://cancernet.nci.nih.gov/, [April 30, 1999].
PDQ. National Cancer Institute. (1999). PDQ Treatment Health Professionals: LaryngealCancer. [Web Site]. Available: http://cancernet.nci.nih.gov/, [April 30, 1999].
PDQ. National Cancer Institute. (1999). PDQ Treatment Health Professionals:Hypopharyngeal cancer. [Web Site]. Available: http://cancernet.nci.nih.gov/, [April 30,1999].PDQ. National Cancer Institute. (April, 1999). PDQ Treatment Health Professionals:Oropharyngeal cancer. [Web Site]. Available: http://cancernet.nci.nih.gov/, [April 30, 1999].
Petitti DB. Meta-Analysis, Decision-Analysis, and Cost-Effectiveness Analysis: methods forquantitative synthesis in medicine. New York, NY: Oxford University Press; 1994.
Provenzale D. Colorectal Cancer: Risk Factors for Advanced Disease - Comparison of Stage atDiagnosis with SEER Cancer Statistics. Veterans Affairs Health Services Research &
December 1998
MTA98-032 MDRC Technology Assessment Program - PET Update - Page R - 5
Development 17th Annual Meeting, Health Services Research at the Interface. SubmittedAbstracts. 131-132, February 24-26, 1999.
Rifkin MD, Zerhouni EA, Gatsonis CA, Quint LE, Paushter DM, Epstein JI. Comparison ofmagnetic resonance imaging and ultrasonography in staging early prostate cancer. Results of amulti-institutional cooperative trial [see comments]. N Engl J Med 1990; 323(10): 621-626.
Sackett D. Clinical epidemiology: a basic science for clinical medicine. Boston: Little Brown;1991.
Shreve PD, Steventon RS, Deters EC, Kison PV, Gross MD, Wahl RL. Oncologic diagnosiswith 2-[fluorine-18]fluoro-2-deoxy-D-glucose imaging: dual-head coincidence gamma cameraversus positron emission tomographic scanner. Radiology 1998; 207(2): 431-437.
Small GW, Rabins PV, Barry PP, Buckholtz NS, DeKosky ST, Ferris SH, et al. Diagnosis andtreatment of Alzheimer disease and related disorders. Consensus statement of the AmericanAssociation for Geriatric Psychiatry, the Alzheimer's Association, and the American GeriatricsSociety [see comments]. JAMA 1997; 278(16): 1363-1371.
Stark DD, Wittenberg J, Butch RJ, Ferrucci JT. Hepatic metastases: randomized, controlledcomparison of detection with MR imaging and CT. Radiology 1987; 165(2): 399-406.
Stevens T. (1997). News Release 11/6/97- Sen. Stevens announces new Medicarereimbursement policy for PET scans used for screening and staging of lung cancer. [PressRelease]. Available: http://www.dana.org/dana/pr_110697.html/, [December 17, 1997].
Stotland BR, Siegelman ES, Morris JB, Kochman ML. Preoperative and postoperative imagingfor colorectal cancer. Hematology/Oncology Clinics of North America 1997; 11(4): 635-654.
Thornbury JR, Kido DK, Mushlin AI, Phelps CE, Mooney C, Fryback DG. Increasing thescientific quality of clinical efficacy studies of magnetic resonance imaging. Invest Radiol 1991;26(9): 829-835.
Thornbury JR, Fryback DG. Technology assessment--an American view. Eur J Radiol 1992;14(2): 147-156.
Thornbury JR, Fryback DG, Turski PA, Javid MJ, McDonald JV, Beinlich BR. Disk-causednerve compression in patients with acute low-back pain: diagnosis with MR, CT myelography,and plain CT [published erratum appears in Radiology 1993 Jun;187(3):880] [see comments].Radiology 1993; 186(3): 731-738.
U.S. Department of Health and Human Services. Health Care Financing Administration.Medicare Program; Open Town Hall Meeting to Discuss the Positron Emission Tomography.Federal Register 1999; 64(1): 170-171.
U.S. Department of Health and Human Services. Health Care Financing Administration. U.S.Department of Health and Human Services. Health Care Financing Administration. (April 25,1999). Coverage Issues Manual - Diagnostic Services. [Web site]. Available: [April 28, 1999].
December 1998
MTA98-032 MDRC Technology Assessment Program - PET Update - Page R - 6
U.S. Department of Health and Human Services. Health Care Financing Administration. (March8, 1999). Positron Emission Tomography (PET) Scan Coverage Expanded. [Web site].Available: http://www.hcfa.gov/news/pr1999/n990308.htm, [May 3, 1999].
U.S. Department of Health and Human Services. Health Care Financing Administration.Technology Advisory Committee. (September 29, 1997). Technology Advisory CommitteeMinutes, August 5 and 6, 1997. [Web site]. Available:http://www.hcfa.gov/events/0897tmin.htm, [May 3, 1999].
U.S. Department of Veterans Affairs. Veterans Health Administration. Office of Research andDevelopment. Improving Health Care for Veterans: 1997 Annual Report. 1998, U.S.Department of Veterans Affairs. Veterans Health Administration. Office of Research andDevelopment: Washington, D.C.
U.S. Secretary of Veterans Affairs. Annual Report Fiscal Year 1995. 1996, U.S. Department ofVeterans Affairs: Washington, D.C.
Vitola JV, Delbeke D, Sandler MP, Campbell MG, Powers TA, Wright Jk. Positron emissiontomography to stage suspected metastatic colorectal carcinoma to the liver. Am J Surg 1996;171(1): 21-26.
Volicer L, Collard A, Hurley A, Bishop C, Kern D, Karon S. Impact of special care unit forpatients with advanced Alzheimer's disease on patients' discomfort and costs [see comments]. JAm Geriatr Soc 1994; 42(6): 597-603.
Wahl RL. Overview of the current status of PET in breast cancer imaging. Quarterly Journal ofNuclear Medicine 1998; 42(1): 1-7.
Webb WR, Gatsonis C, Zerhouni EA, Heelan RT, Glazer GM, Francis IR. CT and MR imagingin staging non-small cell bronchogenic carcinoma: report of the Radiologic Diagnostic OncologyGroup. Radiology 1991; 178(3): 705-713.
Winawer SJ, Fletcher RH, Miller L, Godlee F, Stolar MH, Mulrow CD, et al. Colorectal cancerscreening: clinical guidelines and rationale [see comments] [published errata appear inGastroenterology 1997 Mar;112(3):1060 and 1998 Mar;114(3):625]. Gastroenterology 1997;112(2): 594-642.
Wingo P, Tong T, Bolden S. Cancer Statistics 1995. CA Cancer J Clin 1996; 65 5-27.
Zerhouni EA, Rutter C, Hamilton SR, Balfe DM, Megibow AJ, Francis IR, et al. CT and MRimaging in the staging of colorectal carcinoma: report of the Radiology Diagnostic OncologyGroup II. Radiology 1996; 200(2): 443-451.
Head and NeckIncluded Studies
Diagnostic Accuracy
December 1998
MTA98-032 MDRC Technology Assessment Program - PET Update - Page R - 7
Lowe VJ, Dunphy FR, Varvares M, Kim H, Wittry M, Dunphy CH, et al. Evaluation ofchemotherapy response in patients with advanced head and neck cancer using [F-18]fluorodeoxyglucose positron emission tomography. Head Neck 1997; 19(8): 666-674.
Myers LL, Wax MK, Nabi H, Simpson GT, Lamonica D. Positron emission tomographyin the evaluation of the N0 neck. Laryngoscope 1998; 108(2): 232-236.
Wong WL, Chevretton EB, McGurk M, Hussain K, Davis J, Beaney R, et al. Aprospective study of PET-FDG imaging for the assessment of head and neck squamouscell carcinoma. Clin Otolaryngol 1997; 22(3): 209-214.
Technical EfficacyBraams JW, Pruim J, Kole AC, Nikkels PG, Vaalburg W, Vermey A, et al. Detection ofunknown primary head and neck tumors by positron emission tomography. Int J OralMaxillofac Surg 1997; 26(2): 112-115.
Kuhn GD, Reisser C, DimitrakopoulouStrauss A, Oberdorfer F, Strauss LG. PET studiesof perfusion and glucose metabolism in patients with untreated head and neck tumours.Onkologie 1997; 20(3): 226-230.
Minn H, Lapela M, Klemi PJ, Grenman R, Leskinen S, Lindholm P, et al. Prediction ofsurvival with fluorine-18-fluorodeoxyglucose and PET in head and neck cancer. JournalOf Nuclear Medicine 1997; 38(12): 1907-1911.
Sakamoto H, Nakai Y, Ohashi Y, Okamura T, Ochi H. Positron emission tomographicimaging of head and neck lesions. Eur Arch Otorhinolaryngol Suppl 1997; 1 pS123-6:
December 1998
MTA98-032 MDRC Technology Assessment Program - PET Update - Page R - 8
Excluded StudiesChanglai SP, Kao CH, Chieng PU. 18F-2-fluoro-2-deoxy-D-glucose positron emissiontomography of head and neck in patients with nasopharyngeal carcinomas. OncologyReports 1997; 4(6): 1331-1334.
McGuirt WF, Greven K, Williams D, 3rd, Keyes JW, Jr., Watson N, Cappellari JO, et al.PET scanning in head and neck oncology: a review. Head Neck 1998; 20(3): 208-215.
Breast CancerIncluded Studies
Diagnostic AccuracyAdler LP, Faulhaber PF, Schnur KC, Al-Kasi NL, Shenk RR. Axillary lymph nodemetastases: Screening with (F-18)2-deoxy-2-fluoro- D-glucose (FDG) PET. Radiology1997; 203(2): 323-327.
Bender H, Kirst J, Palmedo H, Schomburg A, Wagner U, Ruhlmann J, et al. Value of 18fluoro-deoxyglucose positron emission tomography in the staging of recurrent breastcarcinoma. Anticancer Res 1997; 17(3b): 1687-1692.
Crippa F, Agresti R, Seregni E, Greco M, Pascali C, Bogni A, et al. Prospectiveevaluation of fluorine-18-FDG PET in presurgical staging of the axilla in breast cancer. JNucl Med 1998; 39(1): 4-8.
Moon DH, Maddahi J, Silverman DH, Glaspy JA, Phelps ME, Hoh CK. Accuracy ofwhole-body fluorine-18-FDG PET for the detection of recurrent or metastatic breastcarcinoma. J Nucl Med 1998; 39(3): 431-435.
Palmedo H, Bender H, Grunwald F, Mallmann P, Zamora P, Krebs D, et al. Comparisonof fluorine-18 fluorodeoxyglucose positron emission tomography and technetium-99mmethoxyisobutylisonitrile scintimammography in the detection of breast tumours. Eur JNucl Med 1997; 24(9): 1138-1145.
Utech CI, Young CS, Winter PF. Prospective evaluation of fluorine-18fluorodeoxyglucose positron emission tomography in breast cancer for staging of theaxilla related to surgery and immunocytochemistry. Eur J Nucl Med 1996; 23(12): 1588-1593.
Technical EfficacyAvril N, Bense S, Ziegler SI, Dose J, Weber W, Laubenbacher C, et al. Breast imagingwith fluorine-18-FDG PET: quantitative image analysis. J Nucl Med 1997; 38(8): 1186-1191.
Noh DY, Yun IJ, Kim JS, Kang HS, Lee DS, Chung JK, et al. Diagnostic value ofpositron emission tomography for detecting breast cancer. World J Surg 1998; 22(3):223-227; discussion 227-228.Oshida M, Uno K, Suzuki M, Nagashima T, Hashimoto H, Yagata H, et al. Predictingthe prognoses of breast carcinoma patients with positron emission tomography using 2-deoxy-2-fluoro[18F]-D-glucose. Cancer 1998; 82(11): 2227-2234.
December 1998
MTA98-032 MDRC Technology Assessment Program - PET Update - Page R - 9
Torizuka T, Zasadny KR, Recker B, Wahl RL. Untreated primary lung and breastcancers: correlation between F-18 FDG kinetic rate constants and findings of in vitrostudies. Radiology 1998; 207(3): 767-774.
Excluded StudiesCrippa F, Agresti R, Donne VD, Pascali C, Bogni A, Chiesa C, et al. The contribution ofpositron emission tomography (PET) with 18F-fluorodeoxyglucose (FDG) in thepreoperative detection of axillary metastases of breast cancer: the experience of the NationalCancer Institute of Milan. Tumori 1997; 83(2): 542-543.
Holle LH, Trampert L, Lung-Kurt S, Villena-Heinsen CE, Puschel W, Schmidt S, et al.Investigations of breast tumors with fluorine-18-fluorodeoxyglucose and SPECT. J NuclMed 1996; 37(4): 615-622.
Katzenellenbogen JA, Welch MJ, Dehdashti F. The development of estrogen and progestinradiopharmaceuticals for imaging breast cancer. Anticancer Res 1997; 17(3b): 1573-1576.
Kole AC, Nieweg OE, Pruim J, Paans AM, Plukker JT, Hoekstra HJ, et al. Standardizeduptake value and quantification of metabolism for breast cancer imaging with FDG and L-[1-11C]tyrosine PET. J Nucl Med 1997; 38(5): 692-696.
Petren-Mallmin M, Andreasson I, Ljunggren O, Ahlstrom H, Bergh J, Antoni G, et al.Skeletal metastases from breast cancer: uptake of 18F-fluoride measured with positronemission tomography in correlation with CT. Skeletal Radiol 1998; 27(2): 72-76.
Lung CancerIncluded Studies
Diagnostic AccuracyBury T, Dowlati A, Paulus P, Corhay JL, Hustinx R, Ghaye B, et al. Whole-body(18)FDG positron emission tomography in the staging of non-small cell lung cancer.European Respiratory Journal 1997; 10(11): 2529-2534.
Erasmus JJ, Patz EF, Jr., McAdams HP, Murray JG, Herndon J, Coleman RE, et al.Evaluation of adrenal masses in patients with bronchogenic carcinoma using 18F-fluorodeoxyglucose positron emission tomography [see comments]. AJR Am JRoentgenol 1997; 168(5): 1357-1360.
Guhlmann A, Storck M, Kotzerke J, Moog F, Sunder-Plassmann L, Reske SN. Lymphnode staging in non-small cell lung cancer: evaluation by [18F] FDG positron emissiontomography (PET). Thorax 1997; 52(5): 438-441.Sasaki M, Ichiya Y, Kuwabara Y, Akashi Y, Yoshida T, Fukumura T, et al. Theusefulness of FDG positron emission tomography for the detection of mediastinal lymphnode metastases in patients with non-small cell lung cancer: a comparative study with X-ray computed tomography. Eur J Nucl Med 1996; 23(7): 741-747.
Vansteenkiste JF, Stroobants SG, De Leyn PR, Dupont PJ, Verschakelen JA, NackaertsKL, et al. Mediastinal lymph node staging with FDG-PET scan in patients with
December 1998
MTA98-032 MDRC Technology Assessment Program - PET Update - Page R - 10
potentially operable non-small cell lung cancer: a prospective analysis of 50 cases.Leuven Lung Cancer Group. Chest 1997; 112(6): 1480-1486.
Technical EfficacyBury T, Paulus P, Dowlati A, Corhay JL, Rigo P, Radermecker MF. Evaluation ofpleural diseases with FDG-PET imaging: Preliminary report. Thorax 1997; 52(2): 187-189.
Ferlin G, Rubello D, Chierichetti F, Zanco P, Bergamin R, Trento P, et al. The role offluorine-18-deoxyglucose (FDG) positron emission tomography (PET) whole body scan(WBS) in the staging and follow-up of cancer patients: Our first experience. Tumori1997; 83(3): 679-684.
Higashi K, Nishikawa T, Seki H, Oguchi M, Nambu Y, Ueda Y, et al. Comparison offluorine-18-FDG PET and thallium-201 SPECT in evaluation of lung cancer. Journal OfNuclear Medicine 1998; 39(1): 9-15.
Kim BT, Kim Y, Lee KS, Yoon SB, Cheon EM, Kwon OJ, et al. Localized form ofbronchioloalveolar carcinoma: FDG PET findings. AJR Am J Roentgenol 1998; 170(4):935-939.
Nettelbladt OS, Sundin AE, Valind SO, Gustafsson GR, Lamberg K, Langstrom B, et al.Combined fluorine-18-FDG and carbon-11-methionine PET for diagnosis of tumors inlung and mediastinum. J Nucl Med 1998; 39(4): 640-647.
Torizuka T, Zasadny KR, Recker B, Wahl RL. Untreated primary lung and breastcancers: correlation between F-18 FDG kinetic rate constants and findings of in vitrostudies. Radiology 1998; 207(3): 767-774.
Wang H, Maurea S, Mainolfi C, Fiore F, Gravina A, Panico MR, et al. Tc-99m MIBIscintigraphy in patients with lung cancer. Comparison with CT and fluorine-18 FDG PETimaging. Clin Nucl Med 1997; 22(4): 243-249.
Excluded StudiesErasmus JJ, McAdams HP, Patz EF, Jr., Coleman RE, Ahuja V, Goodman PC. Evaluation ofprimary pulmonary carcinoid tumors using FDG PET. AJR Am J Roentgenol 1998; 170(5):1369-1373.
Erdi YE, Mawlawi O, Larson SM, Imbriaco M, Yeung H, Finn R, et al. Segmentation oflung lesion volume by adaptive positron emission tomography image thresholding. Cancer1997; 80(12 Suppl): 2505-2509.
SPNIncluded Studies
Diagnostic AccuracyDewan NA, Shehan CJ, Reeb SD, Gobar LS, Scott WJ, Ryschon K. Likelihood ofmalignancy in a solitary pulmonary nodule: comparison of Bayesian analysis and resultsof FDG-PET scan. Chest 1997; 112(2): 416-422.
December 1998
MTA98-032 MDRC Technology Assessment Program - PET Update - Page R - 11
Lowe VJ, Fletcher JW, Gobar L, Lawson M, Kirchner P, Valk P, et al. Prospectiveinvestigation of positron emission tomography in lung nodules. J Clin Oncol 1998;16(3): 1075-1084.
Excluded StudiesLowe VJ, Duhaylongsod FG, Patz EF, Delong DM, Hoffman JM, Wolfe WG, et al.Pulmonary abnormalities and PET data analysis: a retrospective study [see comments].Radiology 1997; 202(2): 435-439.
Colorectal CancerIncluded Studies
Diagnostic AccuracyDelbeke D, Vitola JV, Sandler MP, Arildsen RC, Powers TA, Wright JK, Jr., et al.Staging recurrent metastatic colorectal carcinoma with PET. J Nucl Med 1997; 38(8):1196-1201.
Flanagan FL, Dehdashti F, Ogunbiyi OA, Kodner IJ, Siegel BA. Utility of FDG-PET forinvestigating unexplained plasma CEA elevation in patients with colorectal cancer [seecomments]. Ann Surg 1998; 227(3): 319-323.
Ogunbiyi OA, Flanagan FL, Dehdashti F, Siegel BA, Trask DD, Birnbaum EH, et al.Detection of recurrent and metastatic colorectal cancer: Comparison of position emissiontomography and computed tomography. Annals of Surgical Oncology 1997; 4(8): 613-620.
Technical EfficacyAbdel-Nabi H, Doerr RJ, Lamonica DM, Cronin VR, Galantowicz PJ, Carbone GM, etal. Staging of primary colorectal carcinomas with fluorine-18 fluorodeoxyglucosewhole-body PET: correlation with histopathologic and CT findings. Radiology 1998;206(3): 755-760.
Ruhlmann J, Schomburg A, Bender H, Oehr P, Robertz-Vaupel GM, Vaupel H, et al.Fluorodeoxyglucose whole-body positron emission tomography in colorectal cancerpatients studied in routine daily practice. Dis Colon Rectum 1997; 40(10): 1195-1204.
Alzheimer’s DiseaseIncluded Studies
Diagnostic Accuracy (reviewed in 1996 report)Burdette JH, Minoshima S, Vander Borght T, Tran DD, Kuhl DE. Alzheimer disease:improved visual interpretation of PET images by using three-dimensional stereotaxicsurface projections. Radiology 1996; 198(3): 837-843.
Herholz K, Perani D, Salmon E, Franck G, Fazio F, Heiss WD, et al. Comparability ofFDG PET studies in probable Alzheimer's disease. J Nucl Med 1993; 34(9): 1460-1466.
December 1998
MTA98-032 MDRC Technology Assessment Program - PET Update - Page R - 12
Kippenhan JS, Barker WW, Nagel J, Grady C, Duara R. Neural-network classification ofnormal and Alzheimer's disease subjects using high-resolution and low-resolution PETcameras [see comments]. J Nucl Med 1994; 35(1): 7-15.
Salmon E, Sadzot B, Maquet P, Degueldre C, Lemaire C, Rigo P, et al. Differentialdiagnosis of Alzheimer's disease with PET. J Nucl Med 1994; 35(3): 391-398.
Technical EfficacyDesgranges B, Baron JC, de la Sayette V, Petit-Taboue MC, Benali K, Landeau B, et al.The neural substrates of memory systems impairment in Alzheimer's disease: A PETstudy of resting brain glucose utilization. Brain 1998; 121(Pt 4): 611-631.
Higuchi M, Arai H, Nakagawa T, Higuchi S, Muramatsu T, Matsushita S, et al. Regionalcerebral glucose utilization is modulated by the dosage of apolipoprotein E type 4 alleleand alpha1-antichymotrypsin type A allele in Alzheimer's disease. Neuroreport 1997;8(12): 2639-2643.
Imamura T, Ishii K, Sasaki M, Kitagaki H, Yamaji S, Hirono N, et al. Regional cerebralglucose metabolism in dementia with Lewy bodies and Alzheimer's disease: Acomparative study using positron emission tomography. Neuroscience Letters 1997;235(1-2): 49-52.
Ishii K, Sasaki M, Kitagaki H, Yamaji S, Sakamoto S, Matsuda K, et al. Reduction ofcerebellar glucose metabolism in advanced Alzheimer's disease. J Nucl Med 1997; 38(6):925-928.
Pietrini P, Dani A, Furey ML, Alexander GE, Freo U, Grady CL, et al. Low glucosemetabolism during brain stimulation in older Down's syndrome subjects at risk forAlzheimer's disease prior to dementia. Am J Psychiatry 1997; 154(8): 1063-1069.
Stein DJ, Buchsbaum MS, Hof PR, Siegel BV, Jr., Shihabuddin L. Greater metabolicrate decreases in hippocampal formation and proisocortex than in neocortex inAlzheimer's disease. Neuropsychobiology 1998; 37(1): 10-19.
Vander Borght T, Minoshima S, Giordani B, Foster NL, Frey KA, Berent S, et al.Cerebral metabolic differences in Parkinson's and Alzheimer's diseases matched fordementia severity. J Nucl Med 1997; 38(5): 797-802.
Yamaguchi S, Meguro K, Itoh M, Hayasaka C, Shimada M, Yamazaki H, et al.Decreased cortical glucose metabolism correlates with hippocampal atrophy inAlzheimer's disease as shown by MRI and PET. J Neurol Neurosurg Psychiatry 1997;62(6): 596-600.
Excluded StudiesAlbin RL, Minoshima S, D'Amato CJ, Frey KA, Kuhl DA, Sima AA. Fluoro-deoxyglucosepositron emission tomography in diffuse Lewy body disease. Neurology 1996; 47(2): 462-466.
December 1998
MTA98-032 MDRC Technology Assessment Program - PET Update - Page R - 13
Bergman H, Chertkow H, Wolfson C, Stern J, Rush C, Whitehead V, et al. HM-PAO(CERETEC) SPECT brain scanning in the diagnosis of Alzheimer's disease. Journal Of theAmerican Geriatrics Society 1997; 45(1): 15-20.
de Leon MJ, McRae T, Rusinek H, Convit A, De Santi S, Tarshish C, et al. Cortisol reduceshippocampal glucose metabolism in normal elderly, but not in Alzheimer's disease. J ClinEndocrinol Metab 1997; 82(10): 3251-3259.
Kondoh Y, Nagata K, Sasaki H, Hatazawa J. Dynamic FDG-PET study in probableAlzheimer's disease. Ann N Y Acad Sci 1997; 826: 406-409.
Marcus DL, Freedman ML. Decreased brain glucose metabolism in microvessels frompatients with Alzheimer's disease. Ann N Y Acad Sci 1997; 826: 248-253.
Mega MS, Chen SS, Thompson PM, Woods RP, Karaca TJ, Tiwari A, et al. Mappinghistology to metabolism: coregistration of stained whole-brain sections to premortem PET inAlzheimer's disease. Neuroimage 1997; 5(2): 147-153.
Meguro K, Yamaguchi S, Itoh M, Fujiwara T, Yamadori A. Striatal dopamine metabolismcorrelated with frontotemporal glucose utilization in Alzheimer's disease: a double-tracerPET study. Neurology 1997; 49(4): 941-945.
Minoshima S, Giordani B, Berent S, Frey KA, Foster NL, Kuhl DE. Metabolic reduction inthe posterior cingulate cortex in very early Alzheimer's disease. Ann Neurol 1997; 42(1): 85-94.
December 1998
MTA98-032 MDRC Technology Assessment Program - PET Update - Page E - 1
XIII. EPILOGUE
On January 28, 1999 the TA Program conducted a final update of the literature by searching theliterature published from July 6, 1998 through December 31, 1998 using the same search andappraisal strategies described in Appendix 1. Titles and abstracts of 346 citations were screened.Forty-one were determined to be relevant, and their full text articles were reviewed for potentialinclusion in the review.
Thirty articles from the database searches and from end references of initially retrieved articlesmet inclusion criteria for review. Each included study was classified according to clinicalcondition and assigned to a diagnostic efficacy level as follows:
Efficacy level*H
ead
& N
eck
Brea
st
Lung
sta
ging
SPN
Col
orec
tal
Alzh
eim
er’s
Technical 3 1 6 0 0 7*Diagnostic accuracy 5 1 5** 1 0 0Diagnostic thinking †TherapeuticPatient outcome 1Societal*includes 6 overlapping studies from same institution**includes 3 overlapping studies from same institution (Vansteenkiste, 1998a,b,c)†diagnostic thinking data and diagnostic accuracy data provided from one study (Vansteenkiste, 1998a)
All of the studies represented are single-site case series. All studies used dedicated PET systems.PET was usually added in the work up to complement anatomic imaging data, and most wereretrospective analyses.
As in the main report, recent studies of FDG PET in Alzheimer’s disease explore therelationships between regional glucose metabolism and cognitive function and are classified astechnical efficacy studies. Several studies of diagnostic PET in oncology met inclusion forreview and could be classified at higher levels of diagnostic efficacy. Five studies in lung cancerstaging, three from the same institution (Vansteenkiste, 1998a; 1998b; 1998c) were continuationsof studies reviewed in the main report with overlapping study populations (Bury, 1998; Weder,1998)
The diagnostic accuracy studies were further appraised for study quality and content. None of thestudies met strict evidence-based medicine criteria for evaluations of diagnostic tests, as theextent of blinding was either not clearly reported or was incomplete. However, two met most ofthe criteria and had reasonably well reported and designed studies, despite their small sizes(Smith, 1998; Präuer, 1998). All studies used patients with no metastases or with benigndiseases as internal controls, and all reported using an objective gold standard. Expanded criteriafor methodologic quality of diagnostic accuracy studies used by the American College ofPhysicians yielded the following quality scores:
December 1998
MTA98-032 MDRC Technology Assessment Program - PET Update - Page E - 2
Table 20: Methodologic Quality of Diagnostic Accuracy Studies of FDG PET inSelected Cancers
MethodologicQuality Grade*
Hea
d &
Nec
k
Brea
st
Lung
sta
ging
SPN
Col
orec
tal
ABC 1 3 1D 5 2
Studies received overall quality scores of “D” if the presence of referral bias and methodologicbiases related to the association between test interpretation and gold standard diagnosis were notminimized in the study. Studies received a “C” because of small study sizes, incompletereporting of critical study design elements, and/or a study design that minimized the effect ofmethodologic biases. Several asserted the potential for PET to directly affect patientmanagement, but this was not systematically assessed in any study.
Two studies were classified either as diagnostic thinking efficacy (Vansteenkiste, 1998a) orpatient outcome efficacy (Gambhir, 1998). Expanding on their study reviewed in the mainreport, Vansteenkiste (1998a) used ROC analysis with PET to calculate optimal accuracy andlikelihood ratios (LR) for estimating the probability of nodal metastases in 690 lymph nodestations in 68 patients with non-small cell lung cancer. For their study population, a cut-off SUVof 4.40 provided optimal accuracy. Based on these data, the authors suggested that positive LRsfor SUVs <3.5 or >4.5 offered high diagnostic value and recommended the following:
• The high negative predictive value of mediastinal CT+PET is sufficient to exclude N2/N3disease, to exclude malignancy in individual node stations and, therefore, to omit invasivemediastinal staging.
• Despite the high positive predictive value of CT+PET, mediastinoscopy is still advised inpatients with a positive mediastinal PET to ensure that no patient with N0 or N1 disease isdenied curative resection based on a false positive PET.
LRs can vary with severity of disease in the case mix and positivity criteria (different thresholdvalues) used for interpretation of both imaging tests. There were few benign conditions that maycontribute to false positive diagnoses on CT and PET, and only four patients had confirmed N3disease. The authors calculated positive LRs for both CT and quantitative PET but did not reportthe probability of nodal metastases before CT. In the absence of knowing the pre-test probabilityof malignancy, LRs are inconclusive for assessing the impact of the test on diagnosis ortreatment planning, and these findings should be interpreted cautiously.
Gambhir (1998) conducted a cost-effectiveness analysis to compare various strategies fordiagnosing and managing SPNs. Expanding on a decision analysis by Cummings (1986), theauthors incorporated PET into a CT-based strategy for patients with noncalcified solitarypulmonary nodules < 3cm in diameter. They concluded that a CT-plus-PET strategy was themost cost-effective over a wide range of pre-CT probabilities of malignancy (0.12 to 0.69), andoffered cost savings over the CT-alone strategy ranging from $91 to $2,200 per patient.
December 1998
MTA98-032 MDRC Technology Assessment Program - PET Update - Page E - 3
The assumptions upon which the analysis is based may affect the stability of the conclusions.PET sensitivity and specificity estimates were based on data from one abstract and biasedestimates from three peer-reviewed studies, which were reviewed in the first VA PET report(Flynn, 1996). The model did not account for the possibility of an indeterminate PET scan.Payment and charge data used in the analysis may not adequately reflect true costs or besufficiently comprehensive to reflect the true work-up of these patients.
The MDRC agrees with the authors’ statement that “this analysis is not a substitute for clinicaltrials, but a guide to the design of clinical trials.” The MDRC does not agree with the authors’statement that “there is significant savings when using a PET-based strategy. This warrants amore widespread dissemination of the technology.” Given the preliminary nature of theassumptions, a more widespread dissemination of the technology based on the results of thiscost-effectiveness analysis would be premature.
Conclusion
Recent studies from 1998 do not provide conclusive evidence to support the use of PET in thework up of patients with the cancers assessed in this report. Prospective, rigorously designedstudies with a sufficient spectrum of patients are needed to assess the incremental value of PETin these patients. The impact of PET results on treatment planning has been alleged, but furtherresearch designed to assess impact on treatment management and associated costs is needed. Thefindings from recent 1998 studies confirm the conclusions and recommendations in the mainreport.
December 1998
MTA98-032 MDRC Technology Assessment Program - PET Update - Page E - 4
References
BackgroundCummings SR, Lillington GA, Richard RJ. Managing solitary pulmonary nodules. The choiceof strategy is a "close call." Am Rev Respir Dis 1986; 134(3): 453-460.
Laupacis A, Wells G, Richardson WS, Tugwell P. Users' guides to the medical literature. V.How to use an article about prognosis. Evidence-Based Medicine Working Group. JAMA 1994;272(3): 234-237.
Phelps CE. Good technologies gone bad: how and why the cost-effectiveness of a medicalintervention changes for different populations. Med Decis Making 1997; 17(1): 107-117.
Head and NeckIncluded Studies
Diagnostic AccuracyAdams S, Baum RP, Stuckensen T, Bitter K, Hor G. Prospective comparison of 18F-FDG PET with conventional imaging modalities (CT, MRI, US) in lymph node stagingof head and neck cancer. Eur J Nucl Med 1998; 25(9): 1255-1260.
Fischbein NJ, Aassar OS, Caputo GR, Kaplan MJ, Singer MI, Price DC, et al. Clinicalutility of positron emission tomography with F-18-fluorodeoxyglucose in detectingresidual/recurrent squamous cell carcinoma of the head and neck. American Journal OfNeuroradiology 1998; 19(7): 1189-1196.
Kao CH, ChangLai SP, Chieng PU, Yen RF, Yen TC. Detection of recurrent orpersistent nasopharyngeal carcinomas after radiotherapy with 18-fluoro-2-deoxyglucosepositron emission tomography and comparison with computed tomography. J Clin Oncol1998; 16(11): 3550-3555.
Manolidis S, Donald PJ, Volk P, Pounds TR. The use of positron emission tomographyscanning in occult and recurrent head and neck cancer. Acta Otolaryngol Suppl 1998;534(11): 1-11.
Paulus P, Sambon A, Vivegnis D, Hustinx R, Moreau P, Collignon J, et al. 18FDG-PETfor the assessment of primary head and neck tumors: clinical, computed tomography, andhistopathological correlation in 38 patients. Laryngoscope 1998; 108(10): 1578-1583.
Technical EfficacyCollins BT, Gardner LJ, Verma AK, Lowe VJ, Dunphy FR, Boyd JH. Correlation of fineneedle aspiration biopsy and fluoride-18 fluorodeoxyglucose positron emissiontomography in the assessment of locally recurrent and metastatic head and neckneoplasia. Acta Cytol 1998; 42(6): 1325-1329.
Kole AC, Nieweg OE, Pruim J, Hoekstra HJ, Koops HS, Roodenburg JL, et al. Detectionof unknown occult primary tumors using positron emission tomography. Cancer 1998;82(6): 1160-1166.
December 1998
MTA98-032 MDRC Technology Assessment Program - PET Update - Page E - 5
Uematsu H, Sadato N, Yonekura Y, Tsuchida T, Nakamura S, Sugimoto K, et al.Coregistration of FDG PET and MRI of the head and neck using normal distribution ofFDG. J Nucl Med 1998; 39(12): 2121-2127.
Excluded StudiesDavis JP, Maisey MN, Chevretton EB. Positron emission tomography--a useful imagingtechnique for otolaryngology, head and neck surgery? [editorial]. J Laryngol Otol 1998;112(2): 125-127.
Breast CancerIncluded Studies
Diagnostic AccuracySmith IC, Ogston KN, Whitford P, Smith FW, Sharp P, Norton M, et al. Staging of theaxilla in breast cancer: accurate in vivo assessment using positron emission tomographywith 2-(fluorine-18)-fluoro-2-deoxy-D-glucose. Ann Surg 1998; 228(2): 220-227.
Technical EfficacyCook GJ, Houston S, Rubens R, Maisey MN, Fogelman I. Detection of bone metastasesin breast cancer by 18FDG PET: differing metabolic activity in osteoblastic andosteolytic lesions. J Clin Oncol 1998; 16(10): 3375-3379.
Excluded StudiesDehdashti F, Flanagan FL, Mortimer JE, Katzenellenbogen JA, Welch MJ, Siegel BA.Positron emission tomographic assessment of "metabolic flare" to predict response ofmetastatic breast cancer to antiestrogen therapy. Eur J Nucl Med 1999; 26(1): 51-56.
Flanagan FL, Dehdashti F, Siegel BA. PET in breast cancer. Seminars in Nuclear Medicine1998.
Greco M, Agresti R, Giovanazzi R. Impact of the diagnostic methods on the therapeuticstrategies. Q J Nucl Med 1998; 42(1): 66-80.
Paganelli G, De Cicco C, Cremonesi M, Prisco G, Calza P, Luini A, et al. Optimizedsentinel node scintigraphy in breast cancer. Q J Nucl Med 1998; 42(1): 49-53.
Lung CancerIncluded Studies
Diagnostic AccuracyBury T, Barreto A, Daenen F, Barthelemy N, Ghaye B, Rigo P. Fluorine-18deoxyglucose positron emission tomography for the detection of bone metastases inpatients with non-small cell lung cancer. Eur J Nucl Med 1998; 25(9): 1244-1247.
Vansteenkiste JF, Stroobants SG, De Leyn PR, Dupont PJ, Verbeken EK, Potential useof FDG-PET scan after induction chemotherapy in surgically staged IIIa-N-2 non-small-cell lung cancer: A prospective pilot study. Annals Of Oncology 1998; 9(11): 1193-1198.
Vansteenkiste JF, Stroobants SG, Dupont PJ, De Leyn PR, De Wever WF, Verbeken EK,et al. FDG-PET scan in potentially operable non-small cell lung cancer: do
December 1998
MTA98-032 MDRC Technology Assessment Program - PET Update - Page E - 6
anatometabolic PET-CT fusion images improve the localisation of regional lymph nodemetastases? The Leuven Lung Cancer Group. Eur J Nucl Med 1998; 25(11): 1495-1501.
Vansteenkiste JF, Stroobants SG, De Leyn PR, Dupont PJ, Bogaert J, Maes A, et al.Lymph node staging in non-small-cell lung cancer with FDG-PET scan: a prospectivestudy on 690 lymph node stations from 68 patients. J Clin Oncol 1998; 16(6): 2142-2149.
Weder W, Schmid RA, Bruchhaus H, Hillinger S, von Schulthess GK, Steinert HC.Detection of extrathoracic metastases by positron emission tomography in lung cancer.Ann Thorac Surg 1998; 66(3): 886-892; discussion 892-883.
Technical EfficacyAhuja V, Coleman RE, Herndon J, Patz EF, Jr. The prognostic significance offluorodeoxyglucose positron emission tomography imaging for patients with non-smallcell lung carcinoma. Cancer 1998; 83(5): 918-924.
Gupta N, Gill H, Graeber G, Bishop H, Hurst J, Stephens T. Dynamic positron emissiontomography with F-18 fluorodeoxyglucose imaging in differentiation of benign frommalignant lung/mediastinal lesions. Chest 1998; 114(4): 1105-1111.
Higashi K, Ueda Y, Seki H, Yuasa K, Oguchi M, Noguchi T, et al. Fluorine-18-FDGPET imaging is negative in bronchioloalveolar lung carcinoma. J Nucl Med 1998; 39(6):1016-1020.
Imran MB, Kubota K, Yamada S, Fukuda H, Yamada K, Fujiwara T, et al. Lesion-to-background ratio in nonattenuation-corrected whole-body FDG PET images. J Nucl Med1998; 39(7): 1219-1223.
Kiffer JD, Berlangieri SU, Scott AM, Quong G, Feigen M, Schumer W, et al. Thecontribution of 18F-fluoro-2-deoxy-glucose positron emission tomographic imaging toradiotherapy planning in lung cancer. Lung Cancer 1998; 19(3): 167-177.
Kutlu CA, Pastorino U, Maisey M, Goldstraw P. Selective use of PET scan in thepreoperative staging of NSCLC. Lung Cancer 1998; 21(3): 177-184.
Excluded StudiesErasmus JJ, McAdams HP, Patz EF, Jr., Coleman RE, Ahuja V, Goodman PC. Evaluation ofprimary pulmonary carcinoid tumors using FDG PET. AJR Am J Roentgenol 1998; 170(5):1369-1373.
Laymon CM, Turkington TG, Coleman RE. Attenuation effects in gamma-cameracoincidence imaging. IEEE Transactions on Nuclear Science 1998; 45(6): 3115-3121.
Lowe VJ, Naunheim KS. Positron emission tomography in lung cancer. Ann Thorac Surg1998; 65(6): 1821-1829.
December 1998
MTA98-032 MDRC Technology Assessment Program - PET Update - Page E - 7
SPNIncluded Studies
Diagnostic AccuracyPrauer HW, Weber WA, Romer W, Treumann T, Ziegler SI, Schwaiger M. Controlledprospective study of positron emission tomography using the glucose analogue [18f]fluorodeoxyglucose in the evaluation of pulmonary nodules. Br J Surg 1998; 85(11):1506-1511.
Patient Outcome EfficacyGambhir SS, Shepherd JE, Shah BD, Hart E, Hoh CK, Valk PE, et al. Analyticaldecision model for the cost-effective management of solitary pulmonary nodules. J ClinOncol 1998; 16(6): 2113-2125.
Colorectal CancerExcluded Studies
Miraldi F, Vesselle H, Faulhaber PF, Adler LP, Leisure GP. Elimination of artifactualaccumulation of FDG in PET imaging of colorectal cancer. Clin Nucl Med 1998; 23(1):3-7.
Alzheimer’s DiseaseIncluded Studies
Technical EfficacyHirono N, Mori E, Ishii K, Imamura T, Shimomura T, Tanimukai S, et al. Regionalmetabolism: associations with dyscalculia in Alzheimer's disease. J Neurol NeurosurgPsychiatry 1998; 65(6): 913-916.
Hirono N, Mori E, Ishii K, Ikejiri Y, Imamura T, Shimomura T, et al. Hypofunction inthe posterior cingulate gyrus correlates with disorientation for time and place inAlzheimer's disease. J Neurol Neurosurg Psychiatry 1998; 64(4): 552-554.
Hirono N, Mori E, Ishii K, Ikejiri Y, Imamura T, Shimomura T, et al. Regionalhypometabolism related to language disturbance in Alzheimer's disease. Dement GeriatrCogn Disord 1998; 9(2): 68-73.
Hirono N, Mori E, Ishii K, Ikejiri Y, Imamura T, Shimomura T, et al. Frontal lobehypometabolism and depression in Alzheimer's disease. Neurology 1998; 50(2): 380-383.
Ibanez V, Pietrini P, Alexander GE, Furey ML, Teichberg D, Rajapakse JC, et al.Regional glucose metabolic abnormalities are not the result of atrophy in Alzheimer'sdisease. Neurology 1998; 50(6): 1585-1593.
Ishii K, Sasaki M, Yamaji S, Sakamoto S, Kitagaki H, Mori E. Relatively preservedhippocampal glucose metabolism in mild Alzheimer's disease. Dement Geriatr CognDisord 1998; 9(6): 317-322.
December 1998
MTA98-032 MDRC Technology Assessment Program - PET Update - Page E - 8
Ishii K, Sakamoto S, Sasaki M, Kitagaki H, Yamaji S, Hashimoto M, et al. Cerebralglucose metabolism in patients with frontotemporal dementia. J Nucl Med 1998; 39(11):1875-1878.
Excluded StudiesAlavi A, Clark C, Fazekas F. Cerebral ischemia and Alzheimer's disease: critical role of PETand implications for therapeutic intervention. J Nucl Med 1998; 39(8): 1363-1365.
Frey KA, Minoshima S, Kuhl DE. Neurochemical imaging of Alzheimer's disease and otherdegenerative dementias. Q J Nucl Med 1998; 42(3): 166-178.
Small GW, Leiter F. Neuroimaging for diagnosis of dementia. J Clin Psychiatry 1998;59(11): 4-7.
December 1998
MTA98-032 MDRC Technology Assessment Program - PET Update - Page A1-1
XIV. APPENDIX 1
Methods for the Systematic Review
The MDRC performed a systematic review of the published literature to address the diagnosticefficacy of PET in selected cancer applications and Alzheimer’s disease. A systematic reviewdiffers from a traditional narrative literature review in that it uses a rigorous scientific approachto limit bias and to improve the accuracy of conclusions based on the available data (Guyatt,1995). A systematic review addresses a focused clinical question, uses appropriate and explicitcriteria to select studies for inclusion, conducts a comprehensive search, and appraises thevalidity of the individual studies in a reproducible manner.
Consistent with established methods for conducting a systematic review, the MDRC developedcriteria to select studies for inclusion, conducted a comprehensive search, and appraised thevalidity of the individual studies in a reproducible fashion using the analytic frameworkspresented below.
Search Strategy
An update of the literature was carried out by thoroughly searching the literature published fromSeptember 1996 through July 6, 1998. MEDLINE, HealthSTAR, EMBASE, CurrentContents, and BIOSIS were searched using a range of descriptors: tomography, emissioncomputed; positron emission tomography; gamma camera; PET; and other synonyms. Thesewere combined with the descriptors for Alzheimer’s, colorectal neoplasms, breast neoplasms,head and neck neoplasms, and lung neoplasms. Over 400 citations were retrieved.
Inclusion Criteria
All published studies included in this report met the following inclusion criteria:
• English language articles reporting primary data and published in a peer reviewjournal (not abstracts);
• studies > 12 human subjects (not animal studies) with the disease of interest;• studies using positron emission transverse tomography or positron emission
coincidence imaging;• studies using the radiopharmaceutical 2-[18F]fluoro-2-D-glucose (FDG);• study not duplicated or superseded by later study with the same purpose from the
same institution; and• study design and methods clearly described (i.e. sufficient information to judge
comparability of case and control groups, details of imaging protocol, whether visualor quantitative analysis of PET data used, or type of PET quantitative data analysisused).
Methodologic standards for studies
December 1998
MTA98-032 MDRC Technology Assessment Program - PET Update - Page A1-2
The purpose of appraising the literature using clearly defined methodologic criteria is to ensurethat studies are evaluated in a consistent, reproducible manner, and that studies included in thereport conform to established scientific standards. Studies reviewed for possible inclusion in thisreport were classified according to the strength of the evidence they provided, and the strongestavailable evidence for each application was summarized. The strength of a study is based on theoverall research design and on the quality of the implementation and analysis. The methodologicstandards and the types of studies to which they were applied are summarized below. Thestandards are also discussed in the MDRC report Assessing Diagnostic Technologies (Flynn,1996).
1. Assign to level of diagnostic efficacy hierarchy
Accurate estimation of the characteristics of a diagnostic test is one of the early steps inthe assessment of that test. However, a complete assessment requires further research.
Fryback and Thornbury (1991) note that the localized view of the goal of diagnosticradiology would be that it provides the best images and the most accurate diagnosespossible. A more global view recognizes diagnostic radiology as part of a larger systemof medical care whose goal is to treat patients effectively and efficiently. Viewed in thislarger context, even high-quality images may not contribute to improved care in someinstances, and images of lesser quality may be of great value in others. The point of thesystematic view may be to examine the ultimate value or benefit that is derived from anyparticular diagnostic examination.
Fryback and Thornbury (1991; 1992) present the most recent manifestation of anevolving hierarchical model for assessing the efficacy of diagnostic imaging procedures.Their model, with a list of the types of measures that appear in the literature at each levelin the hierarchy, is presented in the next table. The table progresses from the micro, orlocal level, at which the concern is the physical imaging process itself, to the societalefficacy level. The model stipulates that for a procedure to be efficacious at a higherlevel in the hierarchy it must be efficacious at the lower levels, but the reverse is not true;this asymmetry is often lost in research reports at Levels 1 and 2. Using this model, it ispossible to follow the development of a diagnostic technology, and to align currentresearch efforts with a particular level of development.
December 1998
MTA98-032 MDRC Technology Assessment Program - PET Update - Page A1-3
Com
men
ts
• Ph
ysica
l par
amet
ers
desc
ribin
g te
chni
cal im
agin
g qu
ality
• Jo
int fu
nctio
n of
imag
es a
nd o
bser
ver
• Al
so a
func
tion
of c
linici
an w
ho re
ques
ts d
iagn
ostic
pro
cedu
re, s
ince
sel
ectio
nco
ntro
ls sp
ecific
ity o
f tes
t in
clini
cal p
ract
ice a
nd s
ensit
ivity
to th
e ex
tent
that
itva
ries
with
the
spec
trum
of t
he d
iseas
e
• In
ducin
g ch
ange
in c
linici
ans’
diag
nost
ic th
inkin
g is
a ne
cess
ary
prer
equi
site
toim
pact
on p
atien
ts.•
Clin
ician
s m
ay v
alue
resu
lts w
hich
reas
sure
them
, but
whi
ch d
o no
t cha
nge
treat
men
t dec
ision
s.•
Empi
rical
met
hods
to m
easu
re c
hang
e in
pre
test
dia
gnos
tic p
roba
biliti
es a
ssum
edby
clin
ician
s ar
e pr
obab
ly be
st fo
r det
erm
inin
g th
e ab
senc
e of
dia
gnos
tic th
inkin
gef
ficac
y, ra
ther
than
est
imat
ing
the
mag
nitu
de o
f cha
nge
in d
iagn
ostic
thin
king
due
to im
agin
g in
form
atio
n.•
Imag
ing
exam
inat
ion
resu
lt m
ay in
fluen
ce c
linici
an’s
diag
nost
ic th
inkin
g, b
ut h
as n
oim
pact
on p
atien
t tre
atm
ent.
• In
situ
atio
ns w
here
RCT
s of
dec
ision
mak
ing
with
and
with
out t
he im
agin
gin
form
atio
n ca
nnot
be
perfo
rmed
eth
ically
or b
ecau
se o
f the
mom
entu
m fo
r usin
g a
parti
cula
r pro
cedu
re, a
skin
g Le
vel 4
que
stio
ns m
ay b
e on
ly ef
ficac
y st
udy
poss
ible
.•
Inte
grat
ing n
egat
ive in
form
ation
abo
ut a
test
from
Lev
el 3
and
4 stu
dies
may
help
to d
irect
clin
ical u
se a
way
from
imag
ing
test
s th
at a
re n
ot u
sefu
l or h
ave
been
supp
lante
d by
oth
er te
sts.
• De
finitiv
e an
swer
re e
fficac
y wi
th re
spec
t to
patie
nt o
utco
me
requ
ires
RCT
(invo
lving
with
hold
ing
test
from
som
e pa
tient
s).
• RC
Ts m
ay b
e as
socia
ted
with
form
idab
le s
tatis
tical
, em
piric
al, a
nd e
thica
lpr
oble
ms
and
are
just
ified
only
in c
aref
ully
sele
cted
situ
atio
ns.
• W
eake
r evid
ence
may
be
deriv
ed fr
om c
ase
cont
rol s
tudie
s or
cas
e se
ries.
• In
depe
nden
t con
tribu
tion
of im
agin
g to
pat
ient
out
com
e m
ay b
e sm
all,
requ
iring
very
larg
e sa
mpl
e siz
es.
• De
cisio
n an
alyt
ic ap
proa
ch c
an b
e al
tern
ative
to R
CT, b
ut th
e an
alys
es m
ay s
uffe
rfro
m th
e sa
me
bias
es a
s th
eir s
econ
dary
dat
a so
urce
s.•
Decis
ion
anal
yses
can
hig
hlig
ht c
ritica
l pie
ces
of in
form
atio
n an
d gu
ide
futu
rest
udie
s.
• Ec
onom
ic ev
alua
tions
of e
volvi
ng te
chno
logi
es d
o no
t pro
vide
defin
itive
answ
ers,
since
val
ues
and
judg
men
ts p
lay
a sig
nific
ant r
ole
in in
terp
reta
tion
of re
sults
.•
Cost
utilit
y an
alyse
s im
ply a
t lea
st Le
vel 5
effic
acy
data
or m
odels
.
Typi
cal M
easu
res
of A
naly
sis
• Re
solu
tion
of lin
e pa
irs•
Mod
ulat
ion
trans
fer f
unct
ion
• G
ray
scal
e ra
nge
• Am
ount
of m
ottle
• Sh
arpn
ess
• Yi
eld
of a
bnor
mal
or n
orm
al d
iagn
oses
in a
cas
e se
ries
• Di
agno
stic
accu
racy
(% o
f cor
rect
dia
gnos
es in
cas
e se
ries)
• Se
nsitiv
ity a
nd s
pecif
icity
in a
def
ined
clin
ical s
ettin
g•
Mea
sure
s of
RO
C cu
rve
heig
ht (d
’) or
are
a un
der t
he c
urve
Az
• Nu
mbe
r (%
) of c
ases
in s
erie
s in
whi
ch im
age
judg
ed "h
elpf
ul" t
o m
akin
g di
agno
sis•
Entro
py c
hang
e in
diff
eren
tial d
iagn
osis
prob
abilit
y di
strib
utio
n•
Diffe
renc
e in
clin
ician
s’ su
bjec
tivel
y es
timat
ed d
iagn
osis
prob
abilit
ies
pre-
to p
ostte
stin
form
atio
n•
Empir
ical s
ubjec
tive
log-lik
eliho
od ra
tio fo
r tes
t pos
itive
and
nega
tive
in a
case
ser
ies
• Nu
mbe
r (%
) of t
imes
imag
es ju
dged
hel
pful
in p
lann
ing
man
agem
ent o
f the
pat
ient
in a
case
serie
s•
% o
f tim
es m
edica
l pro
cedu
re a
void
ed d
ue to
imag
e in
form
atio
n•
% o
f tim
es th
erap
y pl
anne
d pr
etes
t cha
nged
afte
r im
agin
g in
form
atio
n wa
s ob
tain
ed(re
trosp
ectiv
ely
infe
rred
from
pat
ient
reco
rds)
• %
of t
imes
clin
ician
s pr
ospe
ctive
ly st
ated
ther
apeu
tic c
hoice
s ch
ange
d af
ter t
est
info
rmat
ion
• %
of p
atie
nts
impr
oved
with
test
com
pare
d wi
th n
o te
st•
Mor
bidity
(or p
roce
dure
s) a
voide
d wi
th te
st•
Chan
ge in
qua
lity-a
djuste
d life
exp
ecta
ncy
• Ex
pect
ed v
alue
of t
est i
nfor
mat
ion
in Q
ALYS
• Co
st pe
r QAL
Y sa
ved
with
imag
ing in
form
ation
• Co
st-b
enef
it ana
lysis
from
socie
tal v
iewp
oint
• Co
st-e
ffect
ivene
ss a
nalys
is fro
m s
ocie
tal v
iewp
oint
• Co
st-u
tility
ana
lysis
from
soc
ieta
l vie
wpoi
nt
A Hi
erar
chic
al M
odel
of E
ffica
cy fo
r Dia
gnos
tic Im
agin
g
Leve
l
1. T
echn
ical e
ffica
cy
2. D
iagn
ostic
acc
urac
y ef
ficac
y
3. D
iagn
ostic
thin
king
effic
acy
4. T
hera
peut
ic ef
ficac
y
5. P
atien
t out
com
e ef
ficac
y
6. S
ocie
tal e
fficac
y
Adap
ted
from
Fry
back
and
Tho
rnbu
ry, 1
991
Abbr
evia
tions
: RC
T, ra
ndom
ized
clini
cal t
rial
ROC,
rece
iver o
pera
ting
char
acte
ristic
s
Q
ALY,
qua
lity a
djus
ted
life y
ear
December 1998
MTA98-032 MDRC Technology Assessment Program - PET Update - Page A1-4
2. Assess the quality of individual studies of diagnostic tests
Criteria for assessing the quality of a diagnostic test evaluation have been defined for usein evidence-based medicine (Haynes and Sackett, 1995). These criteria, listed below,will be applied to individual studies in the report. If the criteria are not met, the studywill generally be considered insufficiently rigorous to provide the basis for patient caredecisions. However, such studies often provide useful information on the technicalcharacteristics of a diagnostic test, or may provide information necessary to subsequentdiagnostic accuracy studies.
Evidence-based medicine criteria for evaluating studies of diagnosis
§ Clearly identified comparison groups, of which ≥ 1is free of the target disorder. § Either an objective diagnostic standard (e.g., a machine-produced laboratory result) or a contemporary clinical diagnostic
standard (e.g., a venogram for deep venous thrombosis) with demonstrably reproducible criteria for any subjectivelyinterpreted component (e.g:, report of better-than-chance agreement among interpreters).
§ Interpretation of the test without knowledge of the diagnostic standard result (no test review bias). § Interpretation of the diagnostic standard without knowledge of the test result (no diagnostic review bias).
Haynes and Sackett, 1995
Documentation of test accuracy does not translate into documentation that the test isclinically useful. Sensitivity and specificity, while not as dependent on prevalence ofdisease as predictive values, can be biased by differences in patient mix in the studypopulation and the patients on whom the test will be used in clinical practice (Sackett etal. 1991). A published study that does not supply valid information needed to calculateposttest probability of disease (i.e., predictive values or likelihood ratios) would not assistclinicians in interpreting its results, or taking action based on those results.
Evidence-based criteria provide a broad quality screen for clinicians who arecontemplating using a test in their own patients. A somewhat more detailed set of qualitycriteria, that expand on those of evidence-based medicine, have been used by theAmerican College of Physicians in evaluations of the literature on magnetic resonanceimaging (Kent et al., 1994; Kent and Larson, 1992; Kent and Larson, 1988). Thesecriteria were applied to studies of diagnostic accuracy and diagnostic thinkingefficacy.
December 1998
MTA98-032 MDRC Technology Assessment Program - PET Update - Page A1-5
Methodologic quality of diagnostic accuracy studies
Grade CriteriaA Studies with broad generalizability to a variety of patients and no significant flaws in research
methods• ≥ 35 patients with disease and ≥ 35 patients without disease (since such numbers yield 95% CIswhose lower bound excludes 0.90 if Se = 1)
• patients drawn from a clinically relevant sample (not filtered to include only severe disease) whoseclinical symptoms are completely described
• diagnoses defined by an appropriate reference standard• PET studies technically of high quality and evaluated independently of the reference diagnosis
B Studies with a narrower spectrum of generalizability, and with only a few flaws that are welldescribed (and impact on conclusions can be assessed)• ≥ 35 cases with and without disease• more limited spectrum of patients, typically reflecting referral bias of university centers (more severeillness)• free of other methods flaws that promote interaction between test result and disease determination• prospective study still required
C Studies with several flaws in methods• small sample sizes• incomplete reporting• retrospective studies of diagnostic accuracy
D Studies with multiple flaws in methods• no credible reference standard for diagnosis• test result and determination of final diagnosis not independent (diagnostic review and/or test reviewbias)• source of patient cohort could not be determined or was obviously influenced by the test result(work-up bias)• opinions without substantiating data
Studies that assess the efficacy of diagnostic tests, particularly estimates of sensitivityand specificity, are susceptible to a variety of biases (Begg, 1987). Thornbury et al.(1991) described five aspects of research methodology that may influence accuracyestimates. Insufficient sample size may result in failure to detect differences betweenimaging modalities, if in fact they do exist, and may provide imprecise estimates ofimaging accuracy.
Differences among patient populations in the spectrum of disease presentation (case mix)and severity result in referral bias. The spectrum of patients needed to assess adiagnostic test will depend on the clinical situation. For example, at initial presentationof abnormality the spectrum should also include patients with no abnormality as well aspatients with abnormalities that may be confused with malignancy. For diagnosingrecurrent disease the spectrum should include patients with recurrence, patients with norecurrence, and patients with treatment changes that may be confused with malignancyon testing. A wider spectrum of patients would be needed to assess a test when there is ahigh prevalence of benign conditions (eg. SPN), whereas a test could be assessed in anarrower spectrum of patients with higher prevalence cancers.
Biases related to the appropriate use of a diagnostic reference standard are work up bias,test review bias, and diagnostic review bias. Presence of referral bias and referencestandard methodologic biases result in overestimation of true positive rates andunderestimation of false positive and negative rates.
December 1998
MTA98-032 MDRC Technology Assessment Program - PET Update - Page A1-6
Considerable activity in the diagnostic testing literature is focusing on developing studydesigns and analytic techniques to correct for, or minimize the effect of, these biases.Some of the more common methods for limiting their influence on diagnostic accuracyestimates are presented below:
Biases in Studies of Diagnostic Imaging Tests
Type of bias Techniques to minimize bias Comments
Referral/spectrum
the influence of spectrum andseverity of disease (case mix)on test characteristics
• referral sources from a variety of medicalpractice settings in which potential patientsubjects are first encountered
• clearly defined referral• define patient groups based on physician’s pre-
test probability estimate of disease• adequate subgroup sizes
⇒ gives sufficient number and mixof patients needed to definepredictive values
⇒ can determine generalizability ofstudy results to own population
⇒ allows subgroup analysis ofdiagnostic accuracy estimates
Work-up/verification
• results from imaging testdetermine the choice ofpatient verified by the goldstandard, or
• study is restricted to biopsyverified cases
• all patients have all competing tests• prospective study in which all patients receive
definitive verification of disease status• sufficient follow-up time• retrospective adjustments• algebraic correction involving regression of
empirical disease frequencies against theprobability of disease as determined in apredictive model
⇒ magnitude of the bias is relatedto association between selectionfor verification and test result
⇒ maximizes diagnostic certainty⇒ require test results and
covariate data from the sourcepopulation and verified sample
Test review
imaging test interpretation is notindependent of final diagnosis,clinical information or results ofcomparison test
• randomized, blinded, independent interpretationof imaging test
• readings with and without clinical information• allow sufficient time between readings• standardize diagnostic terms and degrees of
abnormality• document impact of uninterpretable results• use multiple readers and determine
interobserver variability and methods forresolving differences
⇒ can determine effect of clinicalinformation on diagnosticprobability estimates
⇒ frequency of uninterpretability isan important consideration inthe cost-effectiveness of a test
Diagnostic review/incorporation
gold standard diagnosis is notindependent of imaging testresults
• extensive nodal sampling regardless of imagingresults
• expert interdisciplinary panel to review patientinformation and revise diagnostic and probabilityestimates incrementally
⇒ blinding practitioner to imagingmay be impractical , but effect ofbias can be minimized
⇒ panel process optimizes thefinal diagnosis in cases in whichbiopsy result is and is notavailable
Adapted from Begg (1987), Thornbury et al. (1991), and Webb et al. (1991)
3. Evaluate the strength of the evidence supporting a causal link between the useof the technology and improved outcomes of care
The third analytic framework for the literature review will rank the available evidence forthe degree to which it supports a causal link between the use of the technology andimproved outcomes. Recommendations about the use of a technology should be linked tothe quality of the available evidence, with the strength of the evidence dependent on thequality of the available evidence.
Several models for this framework exist that are based on well-established scientificprinciples of study design. Flynn (1996) used the model below by Cook (1992) tosummarize the relative strengths associated with various study designs and to rank the
December 1998
MTA98-032 MDRC Technology Assessment Program - PET Update - Page A1-7
persuasiveness of their findings between the use of the technology and the outcome ofinterest:
Classifications of study designs and levels of evidence(when high quality meta analyses/overviews are not available)
Level Description
I Randomized trials with low false-positive (alpha) and low false-negative (beta) errors(high power)
• positive trial with statistically significant treatment effect (low alpha error)• negative trial that was large enough to exclude the possibility of a clinically important
benefit (low beta error/high power; i.e. had a narrow confidence interval around thetreatment effect, the lower end of which was greater than the minimum clinically importantbenefit)
• meta analysis can be used to generate a pooled estimate of treatment efficacy across allhigh quality, relevant studies and can reveal any inconsistencies in results
II Randomized trials with high false-positive (alpha) and/or high false negative (beta)
errors (low power) • trial with interesting positive trend that is not statistically significant (high alpha error)• negative trial but possibility of a clinically important benefit (high beta error/low power; i.e.
very wide confidence intervals around the treatment effect)• small positive trials with wide confidence intervals around the treatment effect, making it
difficult to judge the magnitude of the effect• when Level II studies are pooled (through quantitative meta analysis), the aggregate effects
may provide Level I evidence
III Nonrandomized concurrent cohort comparisons between contemporaneous patientswho did and did not (through refusal, noncompliance, contraindication, local practice,oversight, etc.) receive treatment • results subject to biases• Level III data can be subjected to meta analysis, but the result would not shift these data to
another Level, and is not usually recommended
IV Nonrandomized historical cohort comparison between current patients who didreceive treatment (as a result of local policy) and former patients (from the sameinstitution or from the literature) who did not (since at another time or in anotherinstitution different treatment policies prevailed) • results subject to biases, including those that result from inappropriate comparisons over
time and space
V Case series without control subjects • may contain useful information about clinical course and prognosis but can only hint at
efficacy
Source: Cook et al. (1992)
Ibrahim (1987) presented a similar framework to display the continuum of study designsand their causal implications.
December 1998
MTA98-032 MDRC Technology Assessment Program - PET Update - Page A1-8
Continuum of study designs and their causal implications
Level* Study design Inference/strength of evidence
IRandomized controlled trials (RCT)Community randomized trialsSystematic reviews of RCTs
Firm
II Prospective cohort Moderately firm
III Before-after with controlsHistorical cohort Highly suggestive
IV Case-control Moderately suggestive
VTime seriesEcologic correlationsCross-sectional
Suggestive
VIAnecdoteClinical hunchesCase history
Speculative
Adapted from Ibrahim, (1985).*For simplicity, the numerical order was reversed for this review to align with the levels found in the previous table.
Levels IV, V, and VI are observational (nonexperimental) studies. Observational studies aresubject to many forms of bias, which can diminish the accuracy of their findings. They do notprovide strong evidence linking interventions with the observed outcomes; however, they can beuseful for generating hypotheses for future research. Levels II and III are considered quasi-experimental designs. They are commonly used in health care and provide stronger evidencethan can be obtained from observational studies. Level I studies are true experimental studiesand provide the most persuasive evidence for linking interventions with the observed outcomes.
Both frameworks will be used to appraise the strength of the evidence that links use of PET withdesired outcomes, particularly to effect change in diagnosis and treatment management.
December 1998
MTA98-032 MDRC Technology Assessment Program - PET Update - Page A2-1
XV. APPENDIX 2
Models of High Quality Efficacy Studies of Diagnostic Imaging Technologies
Study Highlights of study design
Mushlin (1993)
MRI vs. CT in patients withsuspected multiple sclerosis
§ multi-site study with well-defined referral sources and filters, included patients with an uncertaindiagnosis, representing those in whom the tests might be used
§ sufficient sample size§ all patients receive all tests under evaluation§ independent, blinded image interpretation§ varying degrees of abnormality on the images were noted to permit calculation of receiver-operating
characteristics (ROC) analysis and likelihood ratios for summary comparisons§ sufficient follow-up to permit reasonable diagnostic certainty§ use of technology that is representative of what is available and widely used in most medical
communitiesStark (1987)
MRI vs. CT in patients diagnosedwith liver metastases
§ included patients with and without disease, and patients with benign disease commonly confused withmetastases
§ independent, blinded interpretation of each test and gold standard diagnosis§ used ROC analysis to permit comparison of tests over a range of confidence levels and diagnostic
thresholdsWebb (1991)
MRI vs. CT to determine extent ofdisease in patients with non-smallcell bronchogenic carcinoma
§ multi-site study with a detailed description of the filter through which patients entering into the studywere passed (to reduce referral bias)
§ data dichotomized to analyze lower and advanced stage disease§ blinded, independent interpretation of test results and interobserver variability calculated§ independent pathologic data available for all patients analyzed§ use of standardized forms for data analysis§ extensive nodal sampling not limited to abnormal results on imaging§ assessed influence of sampling procedure on results
Rifkin (1990)
MRI vs. transrectalultrasonography to determineextent of disease in surgicalcandidates with probablelocalized prostate cancer
§ large consecutive case series and a multi-site study§ used standardized forms for data analysis§ blinded, independent interpretation of test results using a five-point grading scale appropriate for ROC
analysis§ lesions identified on diagnostic imaging were matched with pathological findings using a computer
algorithm
Thornbury (1993)
MRI vs. plain CT vs. CTmyelography in patients withacute low-back pain and radicularpain
§ patients with a range of probability of disease were included, based on initial clinical diagnosis beforeimaging
§ sample size sufficient to provide reasonable statistical power§ MRI and one of the two CT tests were performed in all patients§ follow-up time sufficient to permit reasonable diagnostic certainty§ randomized, unpaired blinded interpretation of all tests§ use of an expert interdisciplinary panel to determine true diagnosis§ data collection provided information for use in a cost-effectiveness analysis
Zerhouni (1996)
CT vs. MRI in staging colorectalcarcinoma
§ multi-institutional study with well defined and described study population and referral filter§ all subjects received either histopathologic, follow-up verification, or corrected for work up bias using
technique of Gray et al (1984)§ well-defined positivity criteria§ blind, independent interpretation of each test compared to joint interpretation§ standardized surgical form for data collection of extent of disease for gold standard determination§ extensive quality control procedures to monitor data collection and compliance§ data analysis stratified based on pre-test knowledge of disease
December 1998
MTA98-032 MDRC Technology Assessment Program - PET Update - Page A3-1
XVI. APPENDIX 3
Active Funded Research at VHA PET Facilities as of October 1, 1998
Site Study Title/Number Funding/Sponsor Start/CompletionDates
St. Louis 18F-Fluorodeoxyglucose (FDG) PositronEmission Tomography (PET) Imaging in theManagement of Patients with SolitaryPulmonary Nodules (CSP 27)
$2,306,632 – funded by VHA ORDCooperative Studies Program
1998/5 yearproject
Neurobehavioral Correlates of Mental StressIschemia (R01 HL59619-01A1)
$1,300,000 - NIH National Heart,Lung and Blood Institute
1998-2001
Psychological, CNS and MyocardialMechanisms in Mental Stress Ischemia
$374,000 - Merit Review Award 1998-2000
CNS Correlates of Mental Stress InducedMyocardial Ischemia in Women
$100,000 - Charles A. DanaFoundation, Neuroscience ResearchProgram on Brain-Body Interaction
Starts 1998,duration 3 years
Study to Determine the Effect of Atorvastatinon the Progression of Atherosclerosis
$210,000 - Parke-DavisPharmaceutical Research
1998-1999 (6-month project)
Impact of PET on Patient Care Algorithm $50,000 - funded by VHA Office ofPatient Care Services
1998-1999
PET Measurement of Cerebral Blood FlowCorrelates of Memory in Posttraumatic StressDisorder
$421,094 - Career DevelopmentAward
10/1/97-9/30/00
PET Measurement of Hippocampal Function(Memory) in Depression
$56,500 - National Alliance forResearch in Schizophrenia andDepression, Young Investigator Award
7/1/97-6/30/99
Cerebral Metabolic Correlates of AMPT-induced Depressive Relapse
$306,000 7/1/96-6/30/99
PET Measurement of Cerebral Blood FlowCorrelates of Traumatic Memory in PTSD
$850,000 per year ContinuingRenewal
Hippocampal Function in Gulf War Combat-related PTSD
$299,400 7/1/98-6/30/02
Hippocampus in Women with Abuse-relatedPTSD
$967,000 - NIMH 1/1/99-12/30/02
PET Measurement of BenzodiazepineReceptor in Anxiety
$850,000 per year - National Centerfor Posttraumatic Stress DisorderGrant
ContinuingRenewal
PET Measurement of Cerebral Blood FlowCorrelates of Conditioned Fear
$850,000 - National Center forPosttraumatic Stress Disorder Grant
ContinuingRenewal
Transmyocardial Laser Revascularization inChronic Canine Model of Ischemia
$80,000 - United States SurgicalCorp.
10/96-12/98
Dynamic SPECT BMIPP Imaging comparisonwith Perfusion and FDG Accumulation
$149,400 - Nihon Mediphysics 3/96-6/99
West Haven
PET Neuroreceptor Imaging (Serotonin-2Aand Serotonin-1A)
• $100,000 - National Institute ofMental Health Clinical ResearchCenter
• $55,000 - VA SchizophreniaResearch Center
• 10/1/96-9/30/01
• 10/1/94-12/31/99
Quantitative Assessment of FunctionalConnectivity in the Hereditary Ataxias (PO1NS33718)
$87,720 - Sponsored by NIH/NINDS 1/1/95-12/31/99
Spatial and Temporal Patterns in FunctionalNeuroimaging (P20 MH57180)
$1,113,418 - Sponsored by NIH $9/30/96-9/29/01
Correlation of Cholinergic Reserve andCognitive Function with Positron EmissionTomography (LOI-96-001)
$106,446 - With the Alzheimer’sAssociation
10/15/96-10/14/98
Motor Cortex and the Control of DynamicForce
$75,500 - Merit Review Award by VA 11/1/96-10/30/01
Minneapolis
Functional MRI of Human Motor Cortex(5RO1 NS32437-02)
$150,178 - Sponsored by NIH.NINDS 4/1/95-3/30/98
December 1998
MTA98-032 MDRC Technology Assessment Program - PET Update - Page A3-2
Site Study Title/Number Funding/Sponsor Start/CompletionDates
Functional reorganization with cortical motorareas
$33,000 - Funded by Charles A. DanaFoundation
1/1/95-12/31/98
Neural mechanisms of drawing movementsunder different load conditions
$73,300 - Funded by the NationalScience Foundation
4/1/97-3/31/00
Optimizing 3D Iterative Reconstructions forPET (R29 NS33721)
$71,369 - Sponsored by NINDS 12/1/94-11/30/99
Regional FDG Uptake in Stunned vsHibernating Myocardium (R29 HL52157)
$78,012 - Sponsored by NIH/NHLBI 2/1/96-1/31/01
Quantitative Magnetic ResonanceAssessment of Microvascular Dysfunction(R01 HL58876)
$194,475 - Sponsored by NIH 9/1/97-8/31/00
Functional Anatomy of Human Cognition $99,000 - VA Merit Review Award 10/1/95-9/30/99PET studies of Lexical Processing inSchizophrenia
$30,000 - Young Investigator Awardfrom NARSAD
7/1/96-6/30/98
Lexical Processing in the DifferentialDiagnosis of Mania from Depression
$12,151 - Funded by MinnesotaMedical Foundation
4/1/98-3/31/99
PET Imaging of Hunger and Satiety $38,704 - Minnesota Obesity Center 8/1/96-7/31/97Hippocampal and Memory Dysfunction inNormal Aging
$29,700 - Alzheimer’s DiseaseAssociation
7/1/96-12/31/97
Positron Emission Tomographic Study ofTinnitus and Auditory Plasticity
$46,125 - American TinnitusAssociation
6/1/96-10/30/97
Positron Emission Tomographic Studies ofthe Auditory System
Jane H. Cummings Foundation 6/1/97
A Comparison of Cerebral Blood Flow inMigraineurs During Headache, HeadacheFree, and Treatment Periods
$114,300 - Department of Defense Start 7/1/95duration of twoyears
PET Studies of Temporal Mandibular JointPain
$20,000 - State University of NewYork
Start 6./1/97duration of oneyear
Glucose Transport in Stunned andHibernating Myocardium
$105,000 - New York State Affiliate,American Heart Association
7/1/97-6/30/00
Chronic Alterations in Glucose Transport inHibernating and Stunned Myocardium
$277,800 - American HeartAssociation
7/1/96-6/30/01
Chronic Adaptations to Myocardial Ischemia $1,120,447 - NIH and National HeartBlood and Lung Institute
PET Studies of Tinnitus and Hearing Loss $1,272,652 - NIH and NationalInstitute on Deafness andCommunicative Disorders
Starts 1/98duration of 5 years
Buffalo
PET Imaging subproject $48,240 - NIH and National Instituteof Aging
Fluoxetine Effects on Mood, Cognition &Metabolism
$507,446 - National Institute ofMental Health
Ends 8/31/98
Anterior Cingulate Metabolism in Depression $99,992 - NARSAD Ends 9/14/98Multimethodological Studies in CognitiveNeuroscience
$85,440 - Blue List Neurobiology Ends 12/31/98
The Role of PET in Conjunction with MaximalExercise Stress in Assessment of ChronicStable Coronary Artery Disease
$25,000 - Dupont Pharmaceuticals,Inc.
Ends 01/01/99
The Effects of Prozac Treatment on Mood,Cognition and Brain Glucose Metabolism inPatients with Primary Unipolar Depression
$49,940 - Eli Lilly and Co. Ends 01/01/99
PET/TMS Mapping of the Neural Circuitry ofDevelopmental Stuttering
$100,000 - Dan Foundation Ends 12/31/99
Interactive Effects of Mood and CognitionChallenges on Anterior Cingulate Function inRemitted Depression
$60,000 - NARSAD YoungInvestigator Award
Ends 06/30/00
Hunger for Air Study $140,000 - Mathers Foundation Ends 06/30/00
San Antonio
Investigating the Neural Bases of ChronicStuttering
$435,231 - NIH Ends 11/30/01
Indianapolis Role of Hemodynamics in In-Vivo InsulinResistance (R01 DK 42469)
$207,453 - sponsored by NIH 7/1/95-6/30/00
December 1998
MTA98-032 MDRC Technology Assessment Program - PET Update - Page A3-3
Site Study Title/Number Funding/Sponsor Start/CompletionDates
SCOR in Sudden Cardiac Death (P50 DK52323)
$258,274 - sponsored by NIH 1/1/95-12/31/99
PET Imaging in the Surgical Management ofMelanoma
$127,918 - Sponsored by NIH 4/1/97-3/31/01
Effect of NIDDM on Glucose Transport intoSkeletal Muscle
Not available 1998
The Effect of Troglitazone, Metformin, andSulfonylurea on Insulin-stimulated GlucoseTransport and Phosphorylation, OxidativeEnzyme Capacity and Muscle Composition inNIDDM
Not available Ongoing
Echocardiographic Assessment of MyocardialViability in patients with Impaired LeftVentricular Function
Not available 1998
Pittsburgh
The Role of PET Scanning in Staging thePatient with Intrathoracic Malignancies: Non-small Cell Lung Cancer
Not available 1998
Pre-frontal Dysfunction in Frontal LobeEpilepsy
VA Merit Review
Psychiatric and Behavioral Disturbances inAlzheimer’s Disease
NIMH
The Study of Cognitive Processes in NormalIndividuals: Activation Studies of the NormalHuman Frontal Lobe
Mathers Charitable Foundation
Effect of Smoking on Coronary Blood FlowReserve and Attenuation Effect on CoronaryVasodilator Response of Nitroglycerine
California Tobacco Institute
Perception and Modulation of VisceralSensations
NIH and Astra Pharmaceuticals
Central Nervous System Processing ofSensory Information in Irritable BowelSyndrome (IBS) and Fibromyalgia
CAP
Functional Electrical Stimulation on SpinalCord Injured Patients
VA PM&R R&D
Evaluation of Limb Blood Flow with 15O-H20PET
VA PM&R R&D
15O-H 20 Scanning in Schizophrenia;Assessing Training-Related Improvement
Stanley Foundation and/or NARSADYoung Investigator Award
Brain Metabolic Changes with CigaretteCraving
California Tobacco institute
PET-FDG Imaging of Opioid DependentSubjects
NIDA
Pathogenesis of Symptomatic vs. SilentMyocardial Ischemia
Not Available
West LosAngeles
Assessment of Myocardial Viability Using PETto Determine Benefit for Revascularization
Not Available
Michigan Alzheimer’s Disease ResearchCenter
NIA
PET study of Biochemistry and Metabolism ofCNS
NIND&S
Forebrain Mechanisms of Pain and Analgesia $300,000 - VA Merit AwardForebrain Responses to Chronic Pain and ItsTreatment
NICH&HD
Concomitant Chemotherapy and Radiation forOrgan Preservation in Patients with Advanced(Stage III, IV) Laryngeal Cancer
University of Mich./VA
Ann Arbor
Combined Hormone Replacement Therapyand Myocardial Blood Flow
VA
December 1998
MTA98-032 MDRC Technology Assessment Program - PET Update - Page A3-4
Site Study Title/Number Funding/Sponsor Start/CompletionDates
Effect of Conjugated Equine Estrogen andMicronized Progesterone on Coronary ArteryEndothelial Function as Assessed by PositronEmission Tomography
VA
Limbic Blood Flow & Opiate Receptor PET inPosttraumatic Stress Disorder
$288,500 - VA Merit Award
Paroxysmal Dystonia-Choreoathetosis NIND&SPET Studies of Dopaminergic Neurons inChronic Severe Alcoholism
NIAA&A
Metabolic Imaging of Renal Masses withPositron Emission Tomography
VA
Metabolic Imaging of Pancreatic Disease withPositron Emission Tomography
University of Mich./VA
Imaging of Intermediary Metabolism inNeoplasia using C-11 Acetate PET
VA
December 1998
MTA98-032 MDRC Technology Assessment Program - PET Update - Page A4-1
Resu
lts/C
omm
ents
De
tect
ing
dise
ase
recu
rren
ce (w
ith 9
5% C
I) (2
1 po
sitiv
e ca
ses,
6 n
egat
ive
case
s)PE
T: S
e=90
% (7
7-10
0%);
Sp=8
3% (5
3-10
0%);
PPV=
95%
; NPV
=71%
; Acc
urac
y=89
%;
LR+=
5.43
; LR-
=0.1
1
Au
thor
s’ c
omm
ents
• PE
T m
ay b
e us
ed in
situ
atio
ns w
hen
sam
plin
g bi
as is
mor
e lik
ely
eg. d
ifficu
lt ac
cess
,qu
estio
nabl
e po
st-th
erap
y bi
opsy
resu
lts, o
r a n
orm
al, r
eepi
thel
ializ
ed a
ppea
ranc
e of
tum
orsit
e po
st-th
erap
y•
PET
scan
sho
uld
be o
btai
ned
befo
re b
iops
y to
avo
id p
ossib
le c
onfu
sing
effe
cts
of p
ost
biop
sy in
flam
mat
ion
or w
ait 5
-7 d
ays
afte
r nee
dle
biop
sy o
r 6 w
eeks
afte
r sur
gica
l res
ectio
n•
Posit
ive P
ET s
can
may
be
indi
cativ
e of
resid
ual t
umor
and
war
rant
repe
at ti
ssue
sam
plin
gor
rese
ctio
n•
Nega
tive
PET
scan
may
also
nec
essit
ate
tissu
e co
nfirm
atio
n to
rule
out
false
neg
ative
resu
lts
Patie
nts/
Met
hods
Purp
ose
To e
valu
ate
(?pr
ospe
ctive
ly) c
hem
othe
rapy
resp
onse
usin
g PE
T in
pat
ient
s wi
th a
dvan
ced
head
and
nec
k ca
ncer
Case
s28
con
secu
tive
patie
nts
with
Sta
ge II
I/IV
head
and
nec
k ca
ncer
who
wer
e pa
rticip
atin
g in
ane
oadj
uvan
t org
an-p
rese
rvat
ion
prot
ocol
usin
g ta
xol a
nd c
arbo
plat
in
Met
hods
• PE
T sc
ans
and
tissu
e bi
opsy
per
form
ed o
n al
l pat
ient
s be
fore
and
afte
r (1-
2 we
eks)
chem
othe
rapy
• Ti
ssue
obt
ained
afte
r 2 c
ours
es if
clin
ical r
espo
nse
by c
linica
l exa
m a
nd C
T wa
s <
50%
as
dete
rmin
ed b
y ch
ange
in p
rimar
y tu
mor
size
; and
afte
r 3 c
ours
es if
resp
onse
was
> 5
0%•
Blind
ed v
isual
cons
ensu
s an
alysis
of P
ET b
y tw
o re
ader
s us
ing a
bef
ore
and
afte
rco
mpa
rison
form
at o
n on
e pa
ge p
er p
atien
t usin
g a
4-po
int s
cale
• RO
Is m
easu
red
and
SURs
cal
cula
ted
corre
cted
for b
ody
weig
ht w
hile
blin
ded
to p
atho
logy
resu
lts•
Post
ther
apy
biop
sies
obta
ined
afte
r PET
blin
ded
to P
ET d
ata
• Pa
tient
s cla
ssifie
d as
pat
holo
gic
com
plet
e re
spon
se (P
CR) o
r res
idua
l dise
ase
(RD)
bas
edon
bio
psy
resu
lts
Lim
itatio
ns o
f stu
dy d
esig
n•
Smal
l sam
ple
size
• Sh
ort f
ollo
w up
tim
e•
No c
ompa
rison
dat
a pr
esen
ted
XV
II.
AP
PE
ND
IX 4
: D
ata
Abs
trac
tion
Tab
les
of I
nclu
ded
Dia
gnos
tic
Eff
icac
y St
udie
s of
FD
G-P
ET
in C
ance
r
Dia
gn
ost
ic E
ffic
acy
of
FD
G P
ET
in H
ead
an
d N
eck
Can
cer
Stud
y
Lowe
et a
l. (1
997)
(St.
Loui
s He
alth
Scie
nces
Cen
ter,
Miss
ouri)
December 1998
MTA98-032 MDRC Technology Assessment Program - PET Update - Page A4-2
De
tect
ing
know
n pr
imar
y di
seas
e (3
1 po
sitiv
e ca
ses,
0 n
egat
ive
case
s)Al
l 31
prim
ary
mali
gnan
t tum
ors
were
det
ecte
d by
PET
as
hype
rmet
aboli
c ar
eas
Dete
ctin
g no
dal m
etas
tase
s (1
2 po
sitiv
e ca
ses,
4 n
egat
ive
case
s)PE
T: S
e=67
%CR
/MRI
: Se
=67%
Clin
ical e
xam
: Se
=58%
• 7
of 1
2 pa
tient
s wi
th in
volve
d no
des
had
clinic
ally
obvio
us n
odal
met
asta
ses;
all w
ere
ident
ified
by P
ET•
PET
ident
ified
1 of
5 p
atien
ts wi
th o
ccult
nod
al dis
ease
De
tect
ing
loca
l rec
urre
nce
(10
patie
nts
with
recu
rren
ces,
2 n
on re
curr
ence
s)PE
T co
rrectl
y ide
ntifie
d pr
esen
ce o
r abs
ence
of d
iseas
e in
all 1
2 pa
tient
sSe
=100
%
Dete
ctin
g no
dal r
ecur
renc
e (8
pos
itive
pat
ient
s, 5
neg
ativ
e pa
tient
s)PE
T: S
e=10
0%; S
p=10
0%CR
/MRI
: Se
=75%
; Sp=
80%
Clin
ical e
xam
: Se
=100
%; S
p=60
%
O
ther
find
ings
• Ab
norm
al u
ptak
e un
rela
ted
to m
alig
nanc
y wa
s ca
used
by
oste
omye
litis
of th
e m
andi
ble
follo
wing
den
tal e
xtra
ctio
n an
d su
rger
y pe
rform
ed w
ithin
2 m
onth
s of
PET
imag
ing.
Find
ings
did
not
affe
ct cl
inica
l man
agem
ent
• Ef
fect
s of
pos
t-bio
psy
infla
mm
atio
n on
imag
ing
uncle
ar.
Purp
oses
To p
rosp
ectiv
ely
eval
uate
prim
ary
dise
ase,
ear
ly no
dal m
etas
tase
s, a
nd re
curre
nt d
iseas
e in
patie
nts
head
and
nec
k ca
ncer
Ca
ses
54 c
onse
cutiv
e pa
tient
s wh
o pr
esen
ted
to h
ead
and
neck
clin
ics a
t bot
h ho
spita
ls fo
ras
sess
men
t of s
quam
ous
cell c
arcin
oma
(31
with
prim
ary
dise
ase,
23
with
sus
pect
edre
curre
nce
or re
sidua
l dise
ase)
• of
whic
h 16
had
nec
k dis
secti
ons
(N0=
8; N
1=4;
N2a
=2; N
2b=2
)
M
etho
ds•
All p
atie
nts
exam
ined
und
er a
nest
hesia
and
clin
ically
sta
ged
befo
re im
agin
g•
All s
uspi
cious
are
as o
f aer
odig
estiv
e tra
ct w
ere
biop
sied
• Al
l pat
ients
had
PET
and
anat
omic
imag
ing:
CT (n
=37)
; MRI
(n=1
3);
CT+M
RI (n
=4)
• Cl
inica
l ass
essm
ent,
CT, M
RI, P
ET e
valu
atio
n, a
nd h
istol
ogy
each
per
form
edin
depe
nden
tly•
Stan
dard
size
and
mor
phol
ogica
l crit
eria
use
d to
ass
ess
noda
l dise
ase
on C
T/M
RI
Li
mita
tions
of s
tudy
des
ign
• Sm
all s
ampl
e siz
es in
sub
grou
p an
alys
es•
High
sus
picio
n of
mal
igna
ncy
(refe
rral b
ias)
• O
nly
biop
sy v
erifie
d ca
ses
anal
yzed
(wor
k up
bia
s)•
Crite
ria fo
r pos
itive
PET
not d
efin
ed (p
oten
tial t
est r
evie
w bi
as)
• Bl
indi
ng u
ncle
ar (p
oten
tial t
est r
evie
w bi
as)
• PE
T da
ta a
nalys
is no
t ful
ly de
scrib
ed•
Ord
er o
f biop
sy a
nd im
aging
not
des
cribe
d.
Won
g et
al.
(199
7)(C
linica
l PET
Cent
re o
f Guy
’san
d St
. Tho
mas
’Ho
spita
ls, L
ondo
n,U
K)
December 1998
MTA98-032 MDRC Technology Assessment Program - PET Update - Page A4-3
Resu
lts/C
omm
ents
Dete
ctin
g no
dal m
etas
tase
s (9
pos
itive
dis
sect
ions
, 15
nega
tive
diss
ectio
ns)
PET:
Se=
78%
; Sp=
100%
; PPV
=100
%; N
PV=8
8%; a
ccur
acy=
92%
CT:
Se=5
7%; S
p=90
%; P
PV=8
0%; N
PV=7
5%; a
ccur
acy=
76%
*tren
d in
incr
ease
d ac
cura
cy o
f PET
ove
r CT
(P =
0.1
1)PE
T +
CT:
Se=8
6%; S
p=10
0%; P
PV=1
00%
; NPV
=91%
; acc
urac
y=95
%
Oth
er fi
ndin
gs•
PET
accu
rate
ly de
tect
ed p
rese
nce
or a
bsen
ce o
f cer
vical
met
asta
ses
in a
ll 8 p
atie
nts
with
SCC
of th
e or
al c
avity
• In
5 p
atien
ts wi
th e
ither
car
cinom
a of
the
orop
hary
nx o
r hyo
phar
ynx
PET
corre
ctly
iden
tified
cer
vical
met
asta
ses
in tw
o of
four
pat
ient
s wi
th n
eck
met
asta
ses
Patie
nts/
Met
hods
Purp
ose
To a
sses
s re
trosp
ectiv
ely th
e cli
nical
effe
ctive
ness
of P
ET in
the
evalu
ation
of N
0 sta
ged
neck
patie
nts
with
squ
amou
s ce
ll can
cer (
SCC)
of t
he u
pper
aer
odig
estiv
e tra
ct
Case
s14
pat
ient
s wi
th N
0 di
seas
e (2
4 to
tal n
eck
diss
ectio
ns) o
n cli
nica
l exa
m:
• St
age
I=1;
Sta
ge II
=8; S
tage
III=
2; S
tage
IV=3
from
a la
rger
stu
dy o
f 116
con
secu
tive
patie
nts
diagn
osed
with
hea
d an
d ne
ck c
ance
r, of
whic
h:•
72 h
ad b
iops
y-pr
oven
SCC
• 26
und
erwe
nt n
eck
disse
ction
s
Met
hods
• Al
l pat
ients
had
com
plete
exa
m c
onsis
ting
of P
ET a
nd p
anen
dosc
opy
• Ni
ne p
atie
nts h
ad p
reop
erat
ive C
T•
All p
atie
nts
had
mod
ified
radi
cal n
eck
diss
ectio
ns w
ith re
mov
al o
f lev
els
1 to
V•
Path
olog
ic sp
ecim
ens
exam
ined
for n
umbe
r of n
odes
, pre
senc
e of
mal
igna
ncy,
and
extra
caps
ular
spre
ad.
• PE
T sc
ans
corre
late
d wi
th p
atho
logi
c re
sults
, site
of p
rimar
y tu
mor
, and
CT
Lim
itatio
ns o
f stu
dy d
esig
n•
Small
num
ber o
f cas
es•
Imag
ing
test
s in
fluen
ced
sele
ctio
n of
pat
ient
s fo
r sur
gery
and
nod
al s
ampl
ing
not d
escr
ibed
(wor
kup
bia
s)•
Thre
shold
s fo
r cha
racte
rizing
dise
ase
on im
aging
not
repo
rted
(pot
entia
l tes
t rev
iew b
ias)
• In
depe
nden
t blin
d ev
aluat
ion o
f tes
ts an
d go
ld sta
ndar
d no
t rep
orte
d (p
oten
tial t
est r
eview
bias
and
diag
nost
ic re
view
bias
)
Stud
y
Mye
rs e
t al.
(199
8)(S
UNY,
VA
Med
ical C
ente
r,Bu
ffalo
, NY)
December 1998
MTA98-032 MDRC Technology Assessment Program - PET Update - Page A4-4
Resu
lts/C
omm
ents
Dete
ctin
g ax
illar
y ly
mph
nod
e in
volv
emen
t (44
pos
itive
cas
es, 8
0ne
gativ
e ca
ses)
PET:
Se=
100%
; Sp=
75%
%; P
PV=6
9%; N
PV=1
00%
• Re
ason
s fo
r fal
se p
ositiv
e re
sults
und
eter
min
ed
Oth
er fi
ndin
gs•
Wea
k co
rrela
tion
betw
een
DUR
of m
etas
tatic
axil
lary
lym
ph n
odes
and
tum
or s
ize a
nd b
etwe
en S
PF•
No c
orre
lation
bet
ween
DUR
and
tum
or g
rade
or h
istop
atho
logy
• Co
rrela
tion
betw
een
DUR
and
horm
one
rece
ptor
s wa
s un
dete
rmin
ed
Auth
ors’
com
men
ts•
False
pos
itive
findi
ngs
may
ben
efit
from
com
plem
enta
ry u
se o
flym
phos
cintig
raph
y•
Conf
irmat
ion
is ne
eded
to d
eter
min
e if
PET
shou
ld b
e co
nsid
ered
the
initia
l tes
t in
axilla
ry ly
mph
nod
es
Patie
nts/
Met
hods
Purp
ose
To s
tudy
PET
for s
tagi
ng a
xilla
ry ly
mph
nod
e m
etas
tase
s in
bre
ast c
ance
r
Case
s12
4 pa
tient
s wi
th n
ewly
diag
nose
d an
d hi
stol
ogica
lly p
rove
n br
east
can
cer w
ho w
ere
stud
ied
with
PET
prio
r to
ther
apy:
• St
age
I=65
; Sta
ge II
a=30
; Sta
ge II
b=23
; Sta
ge II
Ia=6
• Tu
mor
size
: < 1
cm=1
6; >
1cm
=49;
>2c
m=3
0; >
3cm
=29
• M
ixed
type
s- 8
2% in
vasiv
e du
ctal
car
cinom
a•
Hype
rglyc
emic
patie
nts
exclu
ded
• Ax
illary
sta
tus:
10
patie
nts
with
pos
itive
lymph
nod
es o
n cli
nica
l exa
m, 4
pat
ient
s on
mam
mog
raph
y
Met
hods
• Al
l pat
ients
had
ER a
nd P
R as
says
, DNA
flow
cyto
met
ry, S
PF, P
ET•
DUR
calcu
lated
for p
rimar
y an
d ax
illary
nod
e up
take
: DU
R 1-
3 fo
r met
asta
ses
and
≥ 3
for
prim
ary t
umor
• Q
ualita
tive
PET
read
as
posit
ive if
disc
rete
foca
l upt
ake
> ba
ckgr
ound
• Im
ages
read
by
expe
rienc
ed ra
diolog
ists,
final
read
by
nucle
ar m
ed p
hysic
ian fr
om h
ard
copy
and
video
mon
itor b
linde
d to
lym
ph n
ode
stat
us; r
eade
r awa
re o
f prim
ary
carc
inom
a•
All p
atie
nts
had
leve
l I d
issec
tion,
som
e ha
d Le
vel I
I, no
ne h
ad le
vel I
II•
Aver
age
# di
ssec
ted
node
s=20
(ran
ge 7
-39)
for t
rue
posit
ives;
16
(rang
e 7-
36) f
or tr
uene
gativ
es; 2
0 (ra
nge
9-46
) for
false
pos
itives
• Q
ualita
tive
PET
com
pare
d to
pat
holog
y•
20 p
atie
nts
with
false
pos
itive
resu
lts fo
llowe
d fo
r 1-2
yea
rs fo
r rec
urre
nce
• DU
R co
rrela
ted
with
tum
or s
ize, g
rade
and
hist
opat
holo
gy, S
PF, D
NA p
loid
y, a
nd h
orm
one
rece
ptor
s
Lim
itatio
ns o
f stu
dy d
esig
n•
Sour
ce p
opul
atio
n un
clear
; ?co
nsec
utive
ser
ies
(pot
entia
l ref
erra
l bia
s)•
Ord
er o
f tes
ting
uncle
ar•
Influ
ence
of P
ET re
sults
on
biops
y pr
oced
ure
uncle
ar (p
oten
tial d
iagno
stic
revie
w bia
s bu
tm
inim
ized
by e
xten
sive
noda
l sam
plin
g)
Dia
gn
ost
ic A
ccu
racy
Eff
icac
y S
tud
ies
of
FD
G P
ET
in
Bre
ast
Can
cer
Stud
y
Utec
h et
al.
(199
6)(U
niv.
of I
llinoi
s Co
llege
of
Med
icine
and
Dow
nsta
teCl
inica
l PET
Cen
ter,
Peor
ia,
Illino
is)
December 1998
MTA98-032 MDRC Technology Assessment Program - PET Update - Page A4-5
Resu
lts/C
omm
ents
Dete
ctin
g ax
illar
y ly
mph
nod
e in
volv
emen
t (20
pos
itive
axi
lla, 3
2 ne
gativ
e ax
illa)
over
all
PET:
Se=
95%
; Sp=
66%
; PPV
=63%
; NPV
=95%
• fa
lse p
ositiv
es c
ause
d by
ext
ensiv
e sin
us h
istio
cyto
sis a
nd fa
t rep
lace
men
t, m
ildpla
smac
ytosis
, and
hem
oside
rin-la
den
mac
roph
ages
• fa
lse n
egat
ive lik
ely
due
to p
atie
nt’s
larg
e siz
e
Oth
er fi
ndin
gs•
Auth
ors
foun
d no
diffe
renc
es in
resu
lts b
etwe
en tw
o sc
anne
rs
Auth
ors’
com
men
ts•
PET
is ac
cept
able
for u
se a
s a
scre
enin
g te
st•
axilla
ry d
issec
tion
shou
ld b
e pe
rform
ed in
pat
ient
s wi
th p
ositiv
e PE
T sc
ans
toco
nfirm
met
asta
ses,
to d
eter
min
e th
e nu
mbe
r of l
ymph
nod
es in
volve
d, a
nd fo
rlo
cal c
ontro
l•
estim
ates
of $
120,
000
in c
ost s
avin
gs a
nd d
ecre
ased
mor
bidi
ty w
ould
hav
eoc
curre
d in
the
22 p
atie
nts
with
neg
ative
PET
sca
ns, b
ut n
o fo
llow
up d
ata
pres
ente
d to
con
firm
• in
clusio
n of
add
itiona
l inte
rpre
tive
crite
ria s
uch
as th
e pr
esen
ce o
f bila
tera
l axil
lary
activ
ity, i
nten
sity
and
exte
nt o
f acti
vity
may
help
impr
ove
spec
ificity
of a
xillar
y PE
Tin
the
futu
re
Stud
y inc
ludes
20
patie
nts
revie
wed
in 19
96 M
DRC
PET
repo
rt
Patie
nts/
Met
hods
Purp
ose
to p
rosp
ectiv
ely
eval
uate
FDG
-PET
as
a sc
reen
ing
test
for a
xilla
ry ly
mph
nod
em
etas
tase
s in
brea
st ca
ncer
Case
s50
pat
ient
s wi
th 5
2 ax
illary
diss
ectio
ns (2
pat
ient
s wi
th b
ilate
ral d
iseas
e)wh
o m
et th
e fo
llowi
ng in
clusio
n cr
iteria
:•
Age
≥ 30
yea
rs•
Ope
rable
bre
ast c
ance
r•
At le
ast l
evel
2 ax
illary
lym
ph n
ode
disse
ction
to b
e pe
rform
ed w
ithin
3 m
onth
s of
PET
• M
inim
um o
f 10
lymph
nod
es d
issec
ted
• Ab
ility
to fa
st fo
r at l
east
4 h
ours
Exclu
sion
crite
ria w
ere:
• Hi
stor
y of
ipsil
ater
al a
xilla
ry ly
mph
nod
e di
ssec
tion
• Pr
eope
rativ
e sy
stem
ic th
erap
y•
Prim
ary
tum
or <
5m
m•
Unin
terp
reta
ble
PET
scan
(2)
Fina
l prim
ary
tum
or s
tagi
ng:
T0=1
; T1=
31; T
2=17
; T3=
3
Met
hods
• Tr
ansm
ission
and
em
ission
PET
sca
ns o
btain
ed o
n tw
o sc
anne
rs (S
cand
itron
ixSP
3000
and
CTI
ECA
T EX
ACT)
• PE
T sc
ans
revie
wed
by tw
o in
depe
nden
t rea
ders
blin
ded
to a
ll inf
orm
atio
n ot
her
than
axil
la s
ide
• PE
T sc
ans
grad
ed o
n a
5-po
int L
ikert
scal
e fo
r pre
senc
e of
incr
ease
d FD
G u
ptak
ean
d sc
an q
uality
; disc
repa
ncie
s re
solve
d by
con
sens
us; s
core
s ≥
3 ar
e po
sitive
• Hi
stop
atho
logy
obt
aine
d; a
ll lym
ph n
odes
diss
ecte
d (a
vera
ge #
/pat
ient
=17)
• O
pera
ting
char
acte
ristic
s of
bot
h sc
anne
rs c
ompa
red
Lim
itatio
ns o
f Stu
dy D
esig
n•
Patie
nt s
ourc
e un
clear
; ?co
nsec
utive
ser
ies
(pot
entia
l ref
erra
l bia
s)•
Asso
ciatio
n be
twee
n te
st re
sult
and
gold
sta
ndar
d de
term
inat
ion
uncle
ar (p
oten
tial
diag
nost
ic re
view
bias
min
imize
d by
ext
ensiv
e no
dal s
ampl
ing)
• No
follo
w-up
dat
a pr
esen
ted
• Au
thor
s us
ed h
igher
dos
e of
FDG
and
long
er s
cann
ing ti
mes
than
in o
ther
stu
dies
Stud
y
Adle
r et a
l. (1
997)
(Uni
vers
ity H
ospi
tals
and
Case
Wes
tern
Res
erve
Uni
vers
ity,
Clev
elan
d, O
hio)
December 1998
MTA98-032 MDRC Technology Assessment Program - PET Update - Page A4-6
Resu
lts/C
omm
ents
Dete
ctin
g ax
illar
y no
de in
volv
emen
tN0
dise
ase
(10
posit
ive a
xilla
, 26
nega
tive
axilla
)PE
T: S
e=70
%; S
p=92
%; a
ccur
acy=
86%
N1a
dise
ase
(8 p
ositiv
e ax
illa, 1
3 ne
gativ
e ax
illa)
PET:
Se=
87.5
%; S
p=10
0%; a
ccur
acy=
95%
N1b-
2 (9
pos
itive
axilla
, 6 n
egat
ive a
xilla
)PE
T: S
e=10
0%; S
p=67
%; a
ccur
acy=
87%
over
all (
27 p
ositiv
e ax
illa, 4
5 ne
gativ
e ax
illa)
PET:
Se=
85%
; Sp=
91%
; acc
urac
y=89
%; P
PV=8
5%; N
PV=9
1%•
false
pos
itives
due
to v
ascu
lar u
ptak
e an
d un
dete
rmine
d ca
uses
• fa
lse n
egat
ives
due
to m
icros
copi
c an
d un
expl
aine
d m
acro
scop
ic in
volve
men
t,m
ean
size=
6mm
(5-8
mm
)
Oth
er fi
ndin
gs•
Med
ian
SUV
in c
arcin
omas
with
axil
lary
met
asta
ses
(4.6
) was
hig
her t
han
that
in ca
rcino
mas
with
out a
xillar
y m
etas
tase
s (2
.9),
but t
here
was
a s
ignific
ant
over
lap
betw
een
the
two
grou
ps (i
nter
quar
tile ra
nges
= 2.
7-7.
2 an
d 1.
9-4.
5,re
spec
tivel
y)•
ROC
anal
ysis
show
ed b
est c
utof
f val
ue o
f SUV
(2.9
) was
ass
ocia
ted
with
aSe
=74%
and
Sp=
58%
• An
y ch
ange
in re
ading
of R
OC
curv
e wa
s no
t use
ful in
pat
ient m
anag
emen
tre
ferre
d fo
r ALN
D•
Auth
ors
prop
ose
usin
g PE
T in
pat
ient
s wi
th v
ery
low
prob
abilit
y of
axil
lary
met
asta
ses
(t1a)
, in
whom
axil
lary
sur
gery
may
be
avoi
ded,
to m
onito
rre
lapse
s•
SUV
value
of p
rimar
y no
t a p
rogn
ostic
indic
ator
of a
xillar
y sp
read
Patie
nts/
Met
hods
Purp
ose
to e
valua
te p
rosp
ectiv
ely n
oninv
asive
sta
ging
of a
xillar
y no
des
using
PET
for m
etas
tase
s
Case
s68
con
secu
tive
patie
nts
with
pal
pabl
e br
east
nod
ules
(uni
late
ral d
iseas
e=64
; bila
tera
l=4)
who
were
sch
edul
ed fo
r bre
ast s
urge
ry w
/ or w
/o A
LND
base
d on
clin
ical a
nd in
stru
men
tal
(mam
mog
raph
y an
d/or
ultr
ason
ogra
phy)
resu
lts•
61 w
ith A
LND
with
72
tota
l axil
la s
ampl
ed: N
0=36
; N1a
=21;
N1b
=13;
N2=
2•
11 w
ith n
o AL
ND w
ere
class
ified
as n
egat
ive•
tota
l # b
reas
t nod
ules=
81 (7
3 m
align
ancie
s, 8
benig
n): T
1=45
; T2=
30; T
3=2;
T4=
4-
63%
of m
alig
nanc
ies
were
infilt
ratin
g du
ctal
car
cinom
as-
benig
n co
nditio
ns w
ere
proli
fera
tive
dysp
lasia
with
out a
typica
or f
ocal
inflam
mat
ion-
aver
age
size
= 20
mm
(4m
m-6
7mm
)
Met
hods
• PE
T em
ission
and
tran
smiss
ion s
cans
per
form
ed 1
to 7
day
s be
fore
sur
gery
• PE
T vis
ual in
terp
reta
tion
blind
ed to
hist
opat
holog
y; loc
alize
d up
take
> s
urro
undin
gtis
sue
class
ified
as p
ositiv
e•
Mea
n SU
Vs o
f bre
ast c
arcin
oma
were
cal
cula
ted
• RO
C an
alys
is pe
rform
ed u
sing
SUVs
bet
ween
two
grou
ps to
ass
ess
SUV
as a
prog
nost
ic in
dica
tor
• Av
erag
e #
diss
ecte
d no
des/
axilla
=21
(rang
e 12
-38)
• No
trea
tmen
t dec
ision
s m
ade
on th
e ba
sis o
f PET
Lim
itatio
ns o
f stu
dy d
esig
n•
High
inde
x of
sus
picio
n of
mal
igna
ncy
(refe
rral b
ias,
min
imize
d by
usin
g co
nsec
utive
serie
s)•
Blin
ding
of r
eade
rs to
oth
er c
linica
l info
rmat
ion
not r
epor
ted
(pot
entia
l tes
t rev
iew
bias
)•
Diag
nosti
c re
view
bias
mini
mize
d by
exte
nsive
nod
al sa
mpli
ng•
PET
used
in te
st se
quen
ce, i
ncre
men
tal v
alue
not a
sses
sed
Stud
y
Crip
pa e
t al.
(199
8)(N
ation
al Ca
ncer
Insti
tute
,M
ilan,
Ital
y)
December 1998
MTA98-032 MDRC Technology Assessment Program - PET Update - Page A4-7
Resu
lts/C
omm
ents
Dete
ctin
g un
know
n pr
imar
y (1
3 m
alig
nant
tum
ors,
7 b
enig
n tu
mor
s)PE
T: S
e=92
%; S
p=86
%SM
M:
Se=9
2%; S
p=86
%•
calcu
lation
s ex
clude
d 2
recu
rrenc
es•
false
pos
itives
due
to fi
broa
deno
ma
• fa
lse n
egat
ives
due
to lo
cal r
ecur
renc
es w
ith d
iam
eter
s <
9mm
Axill
ary
node
invo
lvem
ent
(5 p
ositi
ve p
atie
nts,
7 n
egat
ive
patie
nts)
• PE
T co
rrectl
y de
tecte
d ax
illary
invo
lvem
ent i
n all
12
patie
nts
• (3
0 po
sitive
nod
es)
• PE
T de
tecte
d 9
of 3
0 no
des
• SM
M d
etec
ted
8 of
30
node
s
Qua
ntita
tive
anal
ysis
PET:
mea
n SU
V=2.
57 (0
.3-6
.2 w
ith m
edia
n SU
V=1.
6)SM
M:
mea
n TN
R=1.
97 (1
.42-
3.1
with
med
ian
TNR=
1.8)
Auth
ors’
com
men
ts•
Men
stru
al c
ycle
may
alte
r MIB
I upt
ake
in n
orm
al ti
ssue
• Di
ffuse
FDG
upt
ake
in n
orm
al tis
sue
decli
nes
with
age
• Th
resh
olds
and
var
iatio
ns in
SUV
cal
cula
tions
can
impa
ct te
st c
hara
cter
istics
• Bo
th te
sts c
ould
dete
ct ax
illary
nod
e inv
olvem
ent b
ut n
ot e
xtent
of d
iseas
e•
Larg
er c
ohor
t nee
ded
to c
onfir
m re
sults
Patie
nts/
Met
hods
Purp
ose
to c
ompa
re p
rosp
ectiv
ely
the
diag
nost
ic ac
cura
cy o
f FDG
PET
vs.
scin
timam
mog
raph
y(S
MM
) (pla
nar a
nd s
ingle-
phot
on e
miss
ion to
mog
raph
y) u
sing
99m T
c MIB
I
Case
s20
pat
ients
with
22
susp
iciou
s pr
imar
y les
ions
dete
cted
by P
E or
mam
mog
raph
ysc
hedu
led fo
r exc
ision
al bio
psy
• 14
pat
ient
s wi
th 1
5 m
alig
nant
prim
ary
lesio
ns (i
nclu
ding
2 lo
cal r
ecur
renc
es);
mea
nsiz
e 29
mm
(ran
ge 8
-53m
m);
3 pa
tient
s wi
th tu
mor
s ≤
8mm
• 5
patie
nts
with
30
axilla
ry n
ode
met
asta
ses
(all d
iam
eter
s ≥
12m
m)
Met
hods
• SM
M fo
llowe
d by
PET
sca
ns o
f bre
asts
and
axilla
ry re
gions
wer
e pe
rform
ed >
24
hrs
apar
t in
all p
atie
nts
durin
g th
e we
ek p
rior t
o su
rger
y•
ROIs
ana
lyzed
, M
IBI T
NR o
n pl
anar
imag
es a
nd S
UV F
DG u
ptak
e ca
lcula
ted
• M
amm
ogra
ms
used
for s
cintig
raph
ic lo
caliz
atio
n•
Foca
l upt
ake
class
ified
as n
orm
al o
r abn
orm
al•
Inde
pend
ent,
blin
ded
inte
rpre
tatio
n of
SM
M a
nd P
ET b
y tw
o nu
clear
med
icine
phys
ician
s•
PET
and
SMM
resu
lts e
ach
com
pare
d wi
th h
istop
atho
logy
Lim
itatio
ns o
f stu
dy d
esig
n•
Smal
l size
• Hi
gh in
dex
of s
uspi
cion
for m
alig
nanc
y (p
oten
tial r
efer
ral b
ias)
• Ex
tent
of b
lindi
ng to
oth
er c
linica
l info
rmat
ion
not c
lear
(pot
entia
l tes
t rev
iew
bias
)•
Asso
ciatio
n be
twee
n te
st re
sult
and
gold
sta
ndar
d de
term
inat
ion
uncle
ar (p
oten
tial
diag
nost
ic re
view
bias
)•
Incr
emen
tal v
alue
of t
est u
sed
in w
ork
up n
ot a
sses
sed
Stud
y
Palm
edo
et a
l. (1
997)
(Uni
vers
ity o
f Bon
n, G
erm
any)
December 1998
MTA98-032 MDRC Technology Assessment Program - PET Update - Page A4-8
Resu
lts/C
omm
ents
Dire
ct v
isua
l com
paris
on (6
3 pa
tient
s av
aila
ble
for d
irect
com
paris
on)
Loca
l rec
urre
nce
(15
posi
tive
lesi
ons,
48
nega
tive
lesi
ons)
PET:
Se=
73%
; Sp=
96%
; PPV
=85%
; NPV
=92%
; Acc
=90%
CT/M
RI: S
e=91
%; S
p=98
%; P
PV=9
1%; N
PV=9
8%; A
cc=9
7%
Lym
ph n
odes
(22
posi
tive
lesi
ons,
41
nega
tive
lesi
ons)
PET:
Se=
95%
; Sp=
93%
; PPV
=88%
; NPV
=97%
; Acc
=94%
CT/M
RI: S
e=74
%; S
p=95
%; P
PV=8
9%; N
PV=8
7%; A
cc=8
8%(d
iscre
panc
ies
in to
tal n
umbe
rs o
f PET
lesio
ns v
s. to
tal n
umbe
r of C
T/M
RI le
sions
)
Bone
(13
posi
tive
lesi
ons,
50
nega
tive
lesi
ons)
PET:
Se=
100%
; Sp=
96%
; PPV
=87%
; NPV
=100
%; A
cc=9
7%CT
/MRI
: Se=
46%
; Sp=
98%
; PPV
=88%
; NPV
=86%
; Acc
=87%
Lung
(6 p
ositi
ve s
ites,
57
nega
tive
site
s; in
5 p
atie
nts)
• PE
T ha
d 2
false
pos
itive*
and
1 fa
lse n
egat
ive re
sults
• CT
/MRI
had
2 fa
lse p
ositiv
e an
d 1
false
neg
ative
resu
lts•
Reas
ons
for C
T/M
RI fa
lse p
ositiv
es n
ot re
porte
d
Live
r (2
posi
tive
site
s, 7
3 ne
gativ
e si
tes;
in 2
pat
ient
s)•
PET
had
one
false
pos
itive*
resu
lt an
d no
false
neg
ative
resu
lts•
CT/M
RI h
ad 1
false
pos
itive
and
1 fa
lse n
egat
ive re
sult
• *d
ue to
arti
fact
wro
ngly
inte
rpre
ted
durin
g th
e le
arni
ng p
hase
of t
he fa
cility
Auth
ors’
com
men
ts•
PET
and
CT/M
RI h
ad s
imila
r res
ults
re lu
ng a
nd liv
er m
etas
tase
s•
CT/M
RI id
entifi
ed m
ore
loca
l rec
urre
nces
cor
rect
ly•
PET
iden
tified
1/3
mor
e pa
tient
s wi
th ly
mph
nod
e m
etas
tase
s, s
ugge
stin
g th
e us
eof
PET
ear
ly in
resta
ging
of b
reas
t can
cer.
• Se
miqu
antita
tive
analy
sis a
nd tu
mor
app
eara
nce
info
may
dec
reas
e nu
mbe
r of
false
pos
itive
resu
lts•
Resu
lts s
ugge
st th
e im
porta
nce
of P
ET a
s a
com
plem
ent t
o m
orph
olog
ic te
sts
inth
e st
agin
g of
recu
rrenc
e•
PET
may
also
play
a ro
le in
whole
bod
y sta
ging
of h
igh ri
sk p
atien
ts•
Furth
er s
tudi
es a
re n
eede
d to
ass
ess
the
clini
cal im
pact
of P
ET in
the
man
agem
ent o
f rec
urre
nt b
reas
t can
cer a
nd it
s co
nseq
uenc
e on
ove
rall s
urviv
al
Patie
nts/
Met
hods
Purp
ose
To a
sses
s th
e fe
asib
ility
of P
ET in
sta
ging
recu
rrent
bre
ast c
arcin
oma
Case
s75
pat
ient
s wi
th s
uspe
cted
recu
rrent
or w
ith m
etas
tatic
dise
ase
in u
ndec
ided
or
equiv
ocal
case
s•
prim
ary
tum
or h
istol
ogy:
wel
l-diff
eren
tiate
d du
ctal
car
cinom
a (n
=46)
; inf
iltrat
ing
lobu
lar c
arcin
oma
(n=1
0)•
all p
atien
ts ha
d PE
T; 6
3 pa
tient
s ha
d PE
T an
d CT
/MRI
of w
hich:
- 14
pts
with
con
firm
ed lo
cal r
ecur
renc
e; 1
7 pt
s w/
con
firm
ed ly
mph
nod
e m
ets
• no
recu
rrenc
e (N
=15)
; loc
al re
curre
nce
(N=2
0); l
ymph
nod
e in
volve
men
t (N=
28);
dist
ant m
etas
tase
s (N
=22)
Met
hods
• Al
l pat
ients
had
rout
ine w
ork-
up c
onsis
ting
of p
hysic
al ex
am, a
xillar
y lym
ph n
ode
US, o
ptio
nal t
hora
x/ab
dom
en C
T an
d/or
MRI
, bon
e sc
intig
raph
y an
d se
rum
tum
orm
arke
rs•
All h
ad h
istol
ogic
conf
irmat
ion
(sur
gery
or b
iops
y), e
xcep
t 4 b
y fo
llow-
up•
PET
eval
uate
d qu
alita
tivel
y us
ing
a 4-
poin
t sca
le (i
nten
se, m
oder
ate,
low,
non
e)•
Two
read
ers
of P
ET n
ot b
linde
d to
ava
ilabl
e da
ta•
PET
com
pare
d ind
epen
dent
ly to
eac
h of
the
stand
ard
imag
ing m
odali
ties
• Al
l pat
ient
s fo
llowe
d fo
r at l
east
6 m
onth
s (1
-2 v
isits
)
Lim
itatio
ns o
f stu
dy d
esig
n•
Small
stu
dy s
ize in
sub
grou
p an
alyse
s•
High
inde
x of
sus
picio
n fo
r mal
igna
ncy
(pot
entia
l ref
erra
l bia
s)•
Patie
nt in
clusio
n ba
sed
in pa
rt on
resu
lts o
f CT/
MRI
(wor
k up
bias
but
mini
mize
d by
all p
atie
nts
rece
iving
def
initiv
e di
seas
e ve
rifica
tion)
• In
terp
reta
tion
of P
ET re
sults
not
blin
ded
to o
ther
clin
ical a
nd im
agin
g da
ta; p
ositiv
itycr
iteria
not
def
ined
(tes
t rev
iew
bias
)•
Asso
ciatio
n be
twee
n te
st re
sult
and
gold
sta
ndar
d de
term
inat
ion
uncle
ar (p
oten
tial
diag
nost
ic re
view
bias
)
Stud
y
Bend
er e
t al.
(199
7)(U
nive
rsity
of B
onn,
Ger
man
y)
December 1998
MTA98-032 MDRC Technology Assessment Program - PET Update - Page A4-9
Resu
lts/C
omm
ents
Ove
rall
diag
nost
ic a
ccur
acy
(29
posi
tive
case
s, 2
8 ne
gativ
e ca
ses)
Scor
es ≥
4 d
efin
ed a
s po
sitiv
ePE
T: S
e=93
%; S
p=79
%; P
PV=8
2%; N
PV=9
2%Sc
ores
≥ 3
def
ined
as
posi
tive
PET:
Se=
93%
; Sp=
61%
; PPV
=82%
; NPV
=92%
Ove
rall
diag
nost
ic a
ccur
acy
(41
posi
tive
lesi
ons,
39
nega
tive
lesi
ons)
Scor
es ≥
4 d
efin
ed a
s po
sitiv
ePE
T: S
e=85
%; S
p=79
%; P
PV=8
1%; N
PV=8
4%Sc
ores
≥ 3
def
ined
as
posi
tive
PET:
Se=
90%
; Sp=
54%
; PPV
=67%
; NPV
=84%
ROC
anal
ysis
Az=0
.91
for p
atie
nt d
etec
tion;
Az=
0.88
for l
esion
det
ectio
n
Inte
robs
erve
r var
iabi
lity
• In
48%
of p
atie
nts,
sco
res
of a
ll 3 o
bser
vers
wer
e th
e sa
me
• In
38%
of p
atien
ts, a
sco
re fr
om o
ne re
ader
dev
iated
one
sco
re g
rade
from
the
othe
r 2 re
ader
s•
In 1
4% o
f pat
ients,
a s
core
from
1 re
ader
dev
iated
mor
e th
an 1
sco
re g
rade
from
the
othe
rre
ader
s
Oth
er fi
ndin
gs•
Bone
met
asta
ses
had
a la
rger
pro
porti
on o
f fal
se-n
egat
ive le
sions
than
oth
er m
alig
nant
site
swh
en s
core
s of
4 o
r 5 w
ere
rega
rded
as
posit
ive; B
one
Se=6
9% (1
1 of
16)
vs.
Non
-bon
eSe
=96%
(24
of 2
5) (p
<0.0
5)•
False
neg
ative
lesio
ns (s
core
d ≥
4) in
clude
d 5
bone
met
asta
ses
and
one
small
bre
ast s
ite, o
fwh
ich 3
bon
e an
d on
e br
east
lesion
sho
wed
mild
upt
ake
(sco
red
2-3)
, one
lesio
n wa
s co
nfirm
edpo
sitive
on
follo
w up
PET
sca
n•
Whe
n sc
ores
of ≥
3 w
ere
rega
rded
as
posit
ive, l
ymph
nod
e sit
es h
ad m
ore
false
pos
itives
than
othe
r site
s; L
ymph
Sp=
13%
vs.
Oth
er S
p=79
% (p
<0.0
5); w
ith s
core
s ≥
4, L
ymph
Sp=
60%
vs.
Oth
er S
p=92
% (p
<0.0
01)
• PP
V=62
% fo
r les
ions
with
a s
core
of 4
and
PPV
=90%
for l
esio
ns w
ith a
sco
re o
f 5•
False
pos
itive
lesio
ns (s
core
d ≥
3) w
ere
attri
bute
d to
mus
cle u
ptak
e, in
flam
mat
ion,
and
phys
iolo
gica
l and
arti
fact
ual F
DG u
ptak
e, a
nd u
nkno
wn c
ause
s•
Char
acte
ristic
s of
5 p
atien
ts wi
th fa
sting
bloo
d glu
cose
leve
ls >
110
mg/
dl: 4
diab
etics
, 2re
ceive
d ins
ulin,
3 w
ere
true
nega
tive,
one
true
pos
itive.
Auth
ors’
com
men
ts•
Mor
e st
rict a
ttent
ion
to p
atie
nt p
repa
ratio
n, re
cogn
ition
of a
rtifa
ctua
l upt
akes
, and
info
rmat
ion
oncli
nica
l hist
ory
(re in
flam
mat
ory
dise
ase)
will
impr
ove
spec
ificity
of P
ET•
Lim
itatio
ns o
f stu
dy: s
tudy
bia
sed
towa
rd p
ositiv
e le
sions
and
mor
e di
fficul
t cas
es, n
ot a
llre
gion
s we
re p
rosp
ectiv
ely
exam
ined
with
oth
er c
onve
ntio
nal im
agin
g st
udie
s•
Atte
nuat
ion
corre
ctio
n wo
uld
prov
ide
a m
ore
accu
rate
repr
esen
tatio
n of
trac
er d
istrib
utio
n•
A pr
ospe
ctive
stu
dy is
nee
ded
to fu
rther
ass
ess
the
role
of P
ET in
pos
t-sur
gical
brea
st ca
ncer
man
agem
ent
Patie
nts/
Met
hods
Purp
ose
To re
trosp
ectiv
ely
eval
uate
the
diag
nost
ic ac
cura
cy o
f PET
in p
atie
nts
with
susp
ecte
d re
curre
nt o
r met
asta
tic b
reas
t can
cer
Case
s57
fem
ale
patie
nts
with
83
refe
renc
e sit
es (2
9 wi
th d
iseas
e, 2
8 no
recu
rrenc
eor
met
asta
ses)
• wh
o un
derw
ent p
rimar
y su
rger
y wi
th o
r with
out a
djuv
ant c
hem
o- o
rra
diat
ion
ther
apy a
nd•
who
were
refe
rred
to th
e UC
LA P
ET c
ente
r fro
m O
ctob
er 1
990
to O
ctob
er19
95 (m
ean
time
inter
val b
etwe
en d
iagno
sis a
nd P
ET s
can=
4 yr
s. ra
nge
1m
o. T
o 17
yr 9
mo)
• wh
o ha
d a
clinic
al su
spici
on o
f dise
ase
recu
rrenc
e no
t res
olved
by
conv
entio
nal i
mag
ing
exclu
ded
case
s:•
patie
nts
who
unde
rwen
t che
mo-
or ra
diat
ion
ther
apy
with
in 3
mo
befo
rePE
T•
lesio
ns th
at w
ere
biop
sied
• les
ions
diagn
osed
with
kno
wn d
iseas
e
Met
hods
• Al
l pat
ients
unde
rwen
t hist
ory
and
PE, m
ultipl
e lab
s an
d im
aging
tests
• Di
agno
stic
conf
irmat
ion b
ased
on
biops
y, les
ion m
orph
ology
for t
umor
on
2 or
mor
e im
agin
g st
udie
s, a
nd fo
r at l
east
6 m
onth
s cli
nica
l and
radi
ogra
phic
follo
w up
• PE
T ab
norm
alitie
s th
at re
solve
d wi
thou
t tre
atm
ent w
ere
cons
ider
ed to
be
false
-pos
itive
resu
lts•
ECAT
931
and
ECA
T 96
1 we
re u
sed
• In
depe
nden
t visu
al in
spec
tion
of P
ET b
y 3
read
ers
info
rmed
clin
ical
susp
icion
of m
etas
tase
s, b
ut b
linde
d to
gol
d st
anda
rd; d
iscre
panc
ies
reso
lved
by 4th
read
er a
ware
of a
disc
repa
ncy
but n
ot o
f spe
cifics
• PE
T im
ages
sco
red
from
1 (d
efin
itely
nega
tive)
to 5
(def
inite
ly po
sitive
)•
Lesio
n sit
e de
fined
as
any
abno
rmali
ty su
gges
ting
the
poss
ibly
of b
reas
tre
curre
nce
or m
etas
tase
s eit
her c
linica
lly o
r on
imag
ing, t
here
fore
,an
alysis
was
bias
ed to
ward
pos
itive
lesion
s•
Analy
sis b
y pa
tient
and
by
lesion
Lim
itatio
ns o
f stu
dy d
esig
n•
Onl
y m
ore
diffic
ult c
ases
inclu
ded
and
susp
iciou
s sit
es a
sses
sed
(refe
rral
bias
)•
Not a
ll reg
ions
wer
e pr
ospe
ctive
ly ex
amin
ed w
ith o
ther
con
vent
iona
lim
agin
g te
sts
• Pa
rtial
blin
ding
(pot
entia
l tes
t rev
iew
bias
)•
Blin
ding
of g
old
stan
dard
dia
gnos
is (c
onfir
mat
ion)
to P
ET re
sults
unc
lear
(pot
entia
l dia
gnos
tic re
view
bias
)
Stud
y
Moo
n et
al.
(199
8)(U
CLA,
Los
Ang
eles
, Cal
iforn
iaan
d Un
ivers
ity o
f Ulsa
n, S
eoul
,Ko
rea)
December 1998
MTA98-032 MDRC Technology Assessment Program - PET Update - Page A4-10
Resu
lts/C
omm
ents
Defin
ing
know
n pr
imar
y di
seas
e (1
09 c
ases
)•
all p
rimar
y tu
mor
s sh
owed
incr
ease
d fo
cal F
DG u
ptak
e; in
tens
e in
101
case
s, m
oder
ate
in 8
cas
es•
no c
orre
latio
n be
twee
n hi
stop
atho
logy
and
FDG
upt
ake
Med
iast
inal
invo
lvem
ent (
34 p
ositi
ve c
ases
, 32
nega
tive
case
s)PE
T: S
e=89
% (9
5% C
I: 72
-96%
); Sp
=87%
(95%
CI:
71-9
7%);
PPV=
89%
(
95%
CI:7
2-96
%);
NPV=
87%
(95%
CI:
71-9
6%);
accu
racy
=88%
CT:
Se=7
9%; S
p=71
%; P
PV=7
5%; N
PV=7
6%; a
ccur
acy=
75%
(no
95%
CI re
porte
d)•
disag
reem
ent b
etwe
en P
ET a
nd C
T in
29 c
ases
(44%
); co
rrect
chan
ges
by P
ET=2
2 ca
ses
(33%
); co
rrect
cha
nges
by
CT=7
(11%
)
Dist
ant m
etas
tase
s (3
9 po
sitiv
e ca
ses,
70
nega
tive
case
s)PE
T: S
e=10
0% (9
5% C
I: 9
1-10
0%);
Sp=9
4% (9
5% C
I: 86
-98%
);PP
V=90
% (9
5% C
I: 7
8-97
%);
NPV=
100%
(95%
CI:
95-
100%
);ac
cura
cy=9
6% (9
5% C
I: 9
0-98
%)
CI: S
e=82
%; S
p=89
%; P
PV=8
0%; N
PV=8
9%; a
ccur
acy=
86%
(95%
CI
not r
epor
ted)
• m
oder
ate
FDG
upt
ake
in 7
of 8
cas
es w
ere
< 2
cm.
• PE
T fa
lse p
ositiv
es c
ause
d by
non
spec
ific in
flam
mat
ion
in a
xilla
rylym
ph n
ode,
pne
umon
ia s
eque
lae,
ben
ign
mul
tinod
ular
goi
ter,
anat
omica
l misi
dent
ificat
ion
• PE
T ha
d no
false
pos
itive
FDG
upt
ake
in a
dren
al g
land
s•
PET
corre
ctly
chan
ged
M s
tage
, as
dete
rmine
d by
CI,
in 15
cas
es(1
4%)
Chan
ges
in th
erap
eutic
stra
tegy
• PE
T m
odifie
d th
erap
y in
27
patie
nts
(20%
) (1
0 to
cur
ative
sur
gery
, 8to
a m
ore
cura
tive
appr
oach
with
che
mo-
and
/or r
adia
tion
ther
apy,
9to
mor
e pa
lliativ
e ap
proa
ch)
• no
pat
ient
follo
w up
dat
a re
porte
d
Oth
er fi
ndin
gs•
auth
ors
foun
d PE
T m
ore
usef
ul th
an C
I in
eval
uatin
g ad
rena
l mas
ses
• fa
lse p
ositiv
e PE
T in
axilla
ry s
ite p
roba
bly c
ause
d by
extr
avas
ation
of
ante
cubi
tal v
ein
durin
g FD
G in
ject
ion
• lac
k of
ana
tom
ical m
arke
rs lim
ited
prec
ise lo
caliz
ation
of s
ome
PET
findin
gs•
CI +
PET
cou
ld in
crea
se th
e ac
cura
cy o
f det
ectio
n•
pros
pect
ive c
ompa
rison
nee
ded
to c
ompa
re P
ET w
ith b
one
scan
ning
Patie
nts/
Met
hods
Purp
ose
to p
rosp
ectiv
ely
com
pare
the
accu
racy
of F
DG-P
ET a
nd c
onve
ntio
nal im
agin
g (C
I) fo
r sta
ging
NSCL
C
Case
s14
1 co
nsec
utive
pat
ient
s be
twee
n 9/
94-1
0/96
with
new
ly di
agno
sed
NSCL
C ba
sed
onsp
utum
cyt
olog
y, n
eedl
e bi
opsy
, or f
lexib
le b
ronc
hosc
opy:
• ex
clusio
n cr
iteria
inclu
ded
poor
phy
siolo
gica
l sta
tus
(n=2
1) a
nd in
appr
opria
te fo
llow
up(n
=11)
• of
whi
ch 1
09 w
ere
enro
lled
in th
e st
udy:
• 77
men
, 32
wom
en; m
ean
age=
64 (4
4-83
)•
squa
mou
s ce
ll=50
; ade
nosq
uam
ous
cell=
8; a
deno
carc
inom
a=46
; und
iffere
ntia
ted
larg
ece
ll=5
• st
age
I=32
; sta
ge II
=8; s
tage
IIIA
=22;
stag
e III
B=8;
stag
e IV
=39;
N0=
32; N
1=20
; N2=
10;
N3=4
; T4=
4•
beni
gn c
ondi
tions
= n
onsp
ecific
infla
mm
atio
n, p
neum
onia
, mul
tinod
ular
goi
ter,
loca
lized
FDG
upt
ake
in h
epat
ic-sp
leni
c an
gle
of c
olon
• 66
cas
es w
ith s
uspe
cted
med
iast
inal
invo
lvem
ent o
n CT
or P
ET h
ad b
iops
y co
nfirm
atio
n
Met
hods
• all
pat
ients
had
CI b
efor
e PE
T; C
I= c
hest
and
abdo
mina
l CT
scan
ning
and
bone
scin
tigra
phy
• su
spici
ous
lesion
s on
bon
e sc
intigr
aphy
con
firm
ed b
y bo
ne ra
diogr
aphy
• co
ntra
st C
T po
sitive
crit
eria
> 1
0mm
on
shor
t axis
• PE
T da
ta a
nalyz
ed b
y vis
ual in
terp
reta
tion;
pos
itive
resu
lts=m
oder
ate
(abo
ut tw
ice th
eac
tivity
in c
ontra
late
ral o
r ref
eren
ce re
gion
) and
inte
nse
(mar
kedl
y hi
gher
than
refe
renc
eac
tivity
) FDG
upt
ake
• PE
T an
d CI
inte
rpre
ted
sepa
rate
ly by
2 n
uclea
r med
icine
read
ers
and
2 ra
diolog
y re
ader
sbl
inde
d to
N a
nd M
hist
olog
y bu
t not
blin
ded
to h
istol
ogy
of p
rimar
y tu
mor
• co
nfirm
atio
n of
sus
pect
ed m
edia
stin
al in
volve
men
t or d
istan
t met
asta
ses
on C
I or P
ETdo
ne w
ithin
21
days
of i
mag
ing
• N
stag
ing
conf
irmed
by
exte
nsive
nod
al s
ampl
ing
• M
sta
ging
conf
irmed
by
biops
y (2
1)or
clin
ical a
nd/o
r rad
iolog
ic fo
llow-
up (8
8); a
bsen
ce o
fde
mon
stra
ted
met
asta
ses
6 m
onth
s af
ter n
egat
ive im
agin
g co
nsid
ered
neg
ative
for
met
asta
ses
• sta
tistic
al an
alysis
by
patie
nt
Lim
itatio
ns o
f stu
dy d
esig
n•
choic
e of
N s
tage
coh
ort i
nflue
nced
by
imag
ing te
sts; o
nly b
iopsy
ver
ified
case
s inc
luded
in N
coh
ort (
work
up
bias
min
imize
d by
all p
atie
nts
havin
g bo
th te
sts)
• re
ader
s no
t blin
ded
to p
rimar
y tu
mor
hist
olog
y (p
artia
l tes
t rev
iew
bias
)•
stron
g as
socia
tion
betw
een
test
resu
lts a
nd d
eter
mina
tion
of g
old s
tand
ard
(diag
nosti
cre
view
bias
min
imize
d by
ext
ensiv
e no
dal s
ampl
ing
in N
sta
ging
• inc
omple
te re
porti
ng o
f met
hods
for e
valua
ting
chan
ges
in tre
atm
ent s
trate
gy
Dia
gn
ost
ic A
ccu
racy
Eff
icac
y o
f F
DG
PE
T in
Lu
ng
Can
cer
Stud
y
Bury
et a
l. (1
997)
(CHU
Liè
ge, B
elgi
um)
December 1998
MTA98-032 MDRC Technology Assessment Program - PET Update - Page A4-11
Resu
lts/C
omm
ents
Defin
ing
adre
nal d
isea
se (2
3 m
alig
nant
lesi
ons,
10
beni
gn le
sion
s)PE
T vis
ual a
nalys
is: S
e=10
0%; S
p=80
%SU
R an
alys
is: m
alig
nant
lesio
ns m
ean=
6.28
± 2
.5 v
s. be
nign
lesion
sm
ean=
1.77
± 0
.89
(p <
0.0
001)
O
ther
find
ings
• Ch
arac
teris
tics
of m
alig
nant
mas
ses:
- m
ean
diam
eter
= 4
cm
- n=
6 ne
w on
follo
w up
CT
(mea
n tim
e=4
mon
ths)
, n=1
0 gr
owth
chan
ges
on C
T; n
=7 b
y bio
psy
- ch
ange
s in
adr
enal
mas
ses
on C
T co
nsist
ent w
ith c
hang
es in
thor
ax•
Char
acte
ristic
s of
ben
ign
mas
ses:
- n=
4 by
biop
sy; n
=5 w
ith fe
atur
es <
10
H on
CT;
n=1
with
ben
ignfe
atur
es o
n CT
Patie
nts/
Met
hods
Purp
ose
To a
sses
s PE
T in
diffe
rent
iatin
g be
nign
from
met
asta
tic a
dren
al m
asse
s in
pat
ient
s wi
thbr
onch
ogen
ic ca
rcin
oma
Ca
ses
27 c
onse
cutiv
e ca
ses
with
33
tota
l lesio
ns (2
3 m
alig
nant
, 10
beni
gn) p
rese
ntin
g to
thor
acic
surg
ery,
onco
logy,
or p
ulmon
ary
betw
een
Janu
ary
1993
and
Jan
uary
199
6 wi
th b
ronc
hoge
nicca
rcino
ma
and
an a
dren
al m
ass
dete
cted
by C
TCh
arac
teris
tics:
• 19
men
, 8 w
omen
; mea
n ag
e 57
yrs
(ran
ge 3
9-76
)•
24 w
ith N
SCLC
, 3 s
mal
l cel
l; bi
late
ral m
asse
s in
6 p
atie
nts
• m
ean
diam
eter
of a
dren
al m
asse
s=3
cm (r
ange
1-9
cm)
M
etho
ds•
FDG
PET
per
form
ed a
fter C
T•
Inde
pend
ent i
nter
pret
atio
n of
adr
enal
act
ivity
by
3 re
ader
s bl
inde
d to
clin
ical a
ndpa
thol
ogic
findi
ngs
and
othe
r im
agin
g te
st•
Posit
ive a
ctivi
ty=
activ
ity >
bac
kgro
und;
neg
ative
act
ivity
= ac
tivity
≤ b
ackg
roun
d•
ROI a
nd S
UR d
eter
mine
d bli
nded
to b
iopsy
resu
lts•
Conf
irmat
ion
of a
dren
al m
asse
s by
:-
perc
utan
eous
nee
dle b
iopsy
(n=1
1) w
ithin
a m
ean
of 5
day
s be
fore
PET
(n=9
) and
afte
r(n
=2),
- gr
owth
cha
racte
ristic
s on
follo
w up
(mea
n=4
mon
ths)
CT
(n=1
6), a
nd-
CT H
ouns
field
uni
t mea
sure
men
t < 1
0H d
iagn
ostic
of a
ben
ign
lesio
n (n
=6)
Li
mita
tions
of s
tudy
des
ign
• Hi
gh p
roba
bility
of m
alig
nanc
y a
nd b
enig
n co
nditio
ns n
ot d
epict
ed (p
oten
tial r
efer
ral b
ias)
• As
socia
tion
betw
een
PET
resu
lts a
nd c
hoice
of c
onfir
mat
ion
met
hod
uncle
ar (p
oten
tial
diag
nost
ic re
view
bias
)•
Incr
emen
tal v
alue
of P
ET in
test
seq
uenc
e no
t det
erm
ined
Stud
y
Eras
mus
et a
l. (1
997)
(Duk
e Un
iv. M
edica
l Cen
ter,
Durh
am, N
C)
December 1998
MTA98-032 MDRC Technology Assessment Program - PET Update - Page A4-12
Resu
lts/C
omm
ents
Defin
ing
unkn
own
prim
ary
tum
or (3
2 m
alig
nant
cas
es, 1
4 be
nign
case
s)PE
T: S
e=94
%; S
p=86
%; a
ccur
acy=
91%
CT:
data
not
repo
rted
• PE
T fa
lse p
ositiv
es c
ause
d by
an
aspe
rgillo
ma
with
act
iveinf
lamm
ation
and
a fl
orid
absc
ess
• PE
T fa
lse n
egat
ives
caus
ed b
y a
1 cm
intra
pulm
onar
yad
enoc
arcin
oma
met
asta
sis a
nd b
ronc
hioa
lveol
ar c
arcin
oma
Med
iast
inal
/Hila
r inv
olve
men
t (w
ith 9
5% C
I) (2
0 po
sitiv
e ca
ses,
12
nega
tive
case
s)PE
T: S
e=80
% (5
6%-9
4%);
Sp=1
00%
(73%
-100
%);
accu
racy
=87%
(71%
-96%
)•
PET
accu
racy
(# p
atie
nts)
: N0=
12/1
2; N
1=3/
5; N
2=9/
11; N
3=4/
4CT
: Se=
50%
(27%
-73%
); Sp
=75%
(43%
-95%
); ac
cura
cy=5
9% (4
1%-
76%
) (p
<.02
for o
vera
ll acc
urac
y)•
CT a
ccur
acy
(# p
atie
nts)
: N0=
9/12
; N1=
2/5;
N2=
6/11
; N3=
2/4
• CT
false
pos
itive
node
s ca
used
by
nons
pecif
ic in
flam
mat
ion
inclu
ding
sarc
oidos
is, in
flam
mat
ory
pseu
dotu
mor
, and
pne
umon
ia•
PET
false
pos
itive
hila
r nod
es c
ause
d by
non
spec
ific in
flam
mat
ion
• PE
T dif
fere
ntiat
ed N
1/N2
dise
ase
from
N3
disea
se in
4 p
atien
ts, b
uton
ly 2
of 4
with
CT
Oth
er fi
ndin
gs•
cura
tive
rese
ctio
n wo
uld
have
bee
n av
oide
d in
2 p
atie
nts
with
N3
disea
se o
n PE
T bu
t not
with
CT
• tu
mor
invo
lvem
ent o
f per
ibron
chial
hila
r nod
es a
nd m
edias
tinal
node
sad
jace
nt to
the
bron
chus
may
be
enha
nced
with
ana
tom
etab
olic
imag
ing
• PE
T +
CT m
ay d
ecre
ase
the
need
for i
nvas
ive d
iagno
stic
proc
edur
essu
ch a
s m
edia
stin
osco
py
Patie
nts/
Met
hods
Purp
ose
to e
valua
te re
trosp
ectiv
ely th
e ac
cura
cy o
f FDG
PET
in th
orac
ic lym
ph n
ode
stagin
g in
patie
nts w
ith N
SCLC
Case
s46
con
secu
tive
patie
nts
(32
case
s, 1
4 be
nign
pro
cess
es) w
ho u
nder
went
thor
acot
omy
for
lung
tum
ors
from
199
4 to
8/9
5:•
41 m
en, 5
wom
en; m
ean
age=
56.7
yrs
(24-
78)
• sq
uam
ous
cell (
n=19
); ad
enoc
arcin
oma
(n=7
); la
rge
cell (
n=6)
• T1
=3; T
2=12
; T3=
10; T
4=7;
N0=
12; N
1=5;
N2=
11; N
3=4
• be
nign
con
ditio
ns:
pneu
mon
ia (n
=4);
tube
rcul
osis
(n=3
); on
e ea
ch o
f flo
rid a
bsce
ss,
aspe
rgillo
ma,
ham
arto
ma,
ane
urys
m o
f sub
clavia
n ar
tery
, lun
g fib
rosis
, inf
lam
mat
ory
pseu
dotu
mor
of w
hich
32
mal
igna
nt c
ases
und
erwe
nt fu
rther
med
iast
inal
eva
luat
ion
Met
hods
• all
pat
ients
unde
rwen
t con
trast
CT o
f che
st pr
ior to
PET
3 w
eeks
bef
ore
surg
ery
• po
sitive
nod
e on
CT
defin
ed a
s >
10m
m in
sho
rt ax
is di
amet
er•
blind
, ind
epen
dent
inte
rpre
tatio
n of
CT
by tw
o ex
perie
nced
radio
logist
s to
clin
ical a
nd P
ETfin
dings
• bl
ind,
inde
pend
ent v
isual
inte
rpre
tatio
n of
PET
by
2 ex
perie
nced
nuc
lear
med
icine
phys
ician
s•
hist
opat
holo
gy a
nd T
N cla
ssific
atio
n co
nfirm
ed s
urgi
cally
on
patie
nts
with
prim
ary
lung
canc
er•
surg
eon
cond
ucte
d th
orou
gh d
issec
tion
of m
edia
stin
al n
odes
, dat
a on
ext
ent n
ot re
porte
d•
PET
and
CT re
sults
map
ped
and
com
pare
d to
hist
olog
ic fin
ding
s•
statis
tical
analy
sis re
porte
d by
pat
ient
Lim
itatio
ns o
f stu
dy d
esig
n•
retro
spec
tive
stud
y of
sur
gica
l ser
ies—
high
pro
babi
lity o
f mal
igna
ncy
(pot
entia
l ref
erra
lbi
as)
• sm
all s
ampl
e siz
e lim
its s
ubgr
oup
anal
yses
• on
ly bi
opsy
ver
ified
case
s an
alyz
ed (w
ork-
up b
ias)
• as
socia
tion
betw
een
test
resu
lts a
nd b
iops
y co
nfirm
atio
n un
clear
(pot
entia
l dia
gnos
ticre
view
bias
, min
imize
d by
ext
ensiv
e no
dal s
ampl
ing)
Stud
y
Guh
lman
et a
l. (1
997)
(Uni
vers
ity o
f Ulm
, Ger
man
y)
December 1998
MTA98-032 MDRC Technology Assessment Program - PET Update - Page A4-13
Resu
lts/C
omm
ents
Defin
ing
unkn
own
prim
ary
tum
or (4
4 po
sitiv
e no
des,
10
nega
tive
node
s)PE
T: S
e=93
%; S
p=70
%CT
: no
dat
a re
porte
d•
all f
alse
pos
itives
cau
sed
by g
ranu
lom
as•
3 fa
lse n
egat
ives
cau
sed
by p
oor q
uality
PET
sca
ns (h
yper
glyc
emia
at th
e tim
e of
sca
n, n
o at
tenu
atio
n co
rrect
ion,
r ou
tdat
ed s
cann
er);
one
false
neg
ative
due
to re
nal c
ell c
arcin
oma
unex
plai
ned
Med
iast
inal
invo
lvem
ent—
N2 d
isea
se o
nly
(9 p
ositi
ve n
odes
, 9ne
gativ
e no
des)
PET:
Se=
67%
; Sp=
100
%CT
: Se
=56%
; Sp=
100%
• sm
all n
umbe
rs, s
elec
tion
and
verif
icatio
n bi
as m
ay c
ontri
bute
to la
ckof
sig
nific
ance
bet
ween
PET
and
CT
Oth
er fi
ndin
gs/C
omm
ents
• qu
antita
tive
analy
sis m
ay a
llow
mor
e ac
cura
te d
iffere
ntiat
ion o
fdi
seas
e•
bala
nced
disc
ussio
n of
the
influ
ence
of s
tudy
size
, sel
ectio
n bi
as a
ndve
rifica
tion
bias
es o
n re
sults
• fu
rther
stu
dy n
eede
d to
clar
ify ro
le of
PET
in s
tagin
g•
cost
-effe
ctive
ness
stu
dies
nee
ded
prio
r to
advo
catin
g th
e ro
utin
e us
eof
PET
in m
anag
emen
t of N
SCLS
Patie
nts/
Met
hods
Purp
ose
• to
eva
luat
e re
trosp
ectiv
ely
PET
in c
hara
cter
izing
pul
mon
ary
nodu
les
and
stag
ing
bron
chog
enic
carc
inom
a•
to c
ompa
re C
T wi
th P
ET fo
r dia
gnos
ing
N2 d
iseas
e
Case
s49
con
secu
tive
patie
nts
pres
entin
g be
twee
n 9/
94 a
nd 3
/96
(31
mali
gnan
t cas
es, 1
8 be
nign
case
s) w
ith 5
4 pu
lmon
ary
nodu
les (4
4 po
sitive
nod
ules,
10 n
egat
ive n
odule
s):
• 45
men
, 4 w
omen
; mea
n ag
e=63
(37-
85)
• sq
uam
ous=
15 n
odul
es; a
deno
carc
inom
a=16
; lar
gece
ll=3;
aden
osqu
amou
s=3;
bron
choa
lveol
ar=2
; aty
pica
l car
cinoi
d=1;
sm
all c
ell=
1; re
nal
cell=
2;m
alig
nant
mel
anom
a=1
• be
nign
con
ditio
ns: g
ranu
lom
a=4;
ham
arto
ma=
3; n
ecro
tic ti
ssue
=2; f
unga
l bal
l=1
• ex
clusio
n cr
iteria
: in
dete
rmin
ate
PET
scan
(2) a
nd in
adeq
uate
hist
opat
holo
gic
info
rmat
ion
of m
edia
stin
um (1
1)•
18 o
f 31
mali
gnan
t cas
es h
ad c
omple
te P
ET, C
T, a
nd h
istop
atho
logy
infor
mat
ion a
ndwe
re in
clude
d in
analy
sis o
f med
iastin
al no
des
(N2
only)
Met
hods
• CT
per
form
ed a
nd in
terp
rete
d pr
ior to
PET
• CT
inte
rpre
ted
by o
ne in
vest
igat
or; p
ositiv
e m
edia
stin
al ly
mph
nod
es >
1 c
m in
sho
rt ax
isdi
amet
er•
initia
l PET
sca
ns v
isual
ly in
terp
rete
d by
one
inve
stig
ator
not
blin
ded
to C
XR o
r CT
findin
gs•
blin
ded
med
iast
inal
PET
imag
es re
read
by
two
inve
stig
ator
s•
PET
FDG
upt
ake
class
ified
as p
ositiv
e, n
egat
ive o
r ind
eter
min
ate—
no d
iffere
nce
betw
een
initia
l rea
d an
d re
read
, but
met
hods
for c
ompa
rison
not
des
crib
ed•
all p
ulmon
ary
nodu
les c
onfir
med
by
histo
- or c
ytopa
tholo
gy; e
xtent
of n
odal
sam
pling
not
repo
rted
• da
ta a
nalyz
ed b
y no
de
Lim
itatio
ns o
f stu
dy d
esig
n•
smal
l num
ber o
f sub
ject
s•
retro
spec
tive
desig
n--p
atie
nt s
ourc
e an
d filt
ers
uncle
ar (p
oten
tial r
efer
ral b
ias)
• in
fluen
ce o
f im
agin
g te
sts
on s
elec
tion
of s
urgi
cal c
andi
date
s un
clear
; N s
tage
coh
ort
rest
ricte
d to
bio
psy
verif
ied
case
s (w
ork
up b
ias)
• as
socia
tion
of te
st re
sults
and
det
erm
inat
ion
of g
old
stan
dard
unc
lear
(pot
entia
l dia
gnos
ticre
view
bias
)•
inte
rpre
tatio
n of
prim
ary
tum
or n
ot b
linde
d (te
st re
view
bias
)
Stud
y
Hagb
erg
et a
l. (1
997)
(VA
Palo
Alto
Hea
lth C
are
Syste
man
d St
anfo
rd U
nive
rsity
Sch
ool o
fM
edici
ne)
December 1998
MTA98-032 MDRC Technology Assessment Program - PET Update - Page A4-14
Resu
lts/C
omm
ents
Dete
ctin
g m
edia
stin
al ly
mph
nod
e m
etas
tase
s (2
8 po
sitiv
e no
des,
84 n
egat
ive
node
s)PE
T: S
e=89
%*;
Sp=9
9%; P
PV=
96%
; NPV
=97%
; acc
urac
y=97
%CT
: Se=
57%
*; Sp
=94%
; PPV
=76%
; NPV
=87%
; acc
urac
y=85
%*
(P=.
0066
)
(n=4
7 pa
tient
s)•
PET:
45/
47 (9
6%) c
orre
ctly
stage
d vs
. CT:
37/
47 (7
9%) (
P=.0
1341
)
Oth
er fi
ndin
gs•
no s
tatis
tical
ly sig
nific
ant d
iffer
ence
bet
ween
sta
ging
with
conv
entio
nal C
T an
d wi
th s
pira
l CT
• co
rrela
tion
of n
odal
sta
tion
on im
agin
g wi
th s
urgi
cal r
esul
ts m
atch
edex
cept
for o
ne c
ase
Patie
nts/
Met
hods
Purp
ose
to c
ompa
re p
rosp
ectiv
ely
(??)
the
accu
racy
of F
DG P
ET w
ith C
T in
sta
ging
NSC
LC
Case
s62
sur
gical
cand
idate
s wi
th s
uspe
cted
or p
rove
n NS
CLC
who
had
PET
betw
een
2/94
and
3/96
and
who
had
no
prior
neo
adjuv
ant t
hera
py o
r diab
etes
• ex
clusio
n cr
iteria
: in
adeq
uate
CT=
2; d
istan
t met
asta
ses=
8; in
adeq
uate
nod
al s
ampl
ing=
5•
ther
efor
e, 4
7 pa
tient
s wi
th s
uspe
cted
or c
onfir
med
NSC
LC o
f mixe
d ty
pes
rem
aine
d in
the
study
• sq
uam
ous=
24; a
deno
carc
inom
a=17
; lar
ge c
ell=
6•
29 (6
2%) w
ith n
odal
met
asta
ses;
N0-
N1=3
4; N
2=7;
N3=
6
Met
hods
• CT
, em
issio
n an
d tra
nsm
issio
n PE
T ob
tain
ed o
n al
l pat
ient
s•
blin
d, in
depe
nden
t int
erpr
etat
ion
of P
ET a
nd C
T sc
ans
befo
re s
urgi
cal s
tagi
ng•
CT p
ositiv
e cr
iteria
=nod
es >
10m
m in
sho
rt ax
is dia
met
er, e
xcep
t upp
er p
arat
rach
eal
statio
ns >
7m
m o
r inf
raca
rinal
statio
n >
11m
m•
pres
ence
and
site
of m
edia
stin
al a
nd tr
ache
obro
nchi
al n
odes
reco
rded
acc
ordi
ng to
ATS
lymph
nod
e st
atio
n m
appi
ng s
yste
m; e
xten
t of l
ymph
nod
e m
etas
tase
s cla
ssifie
d ac
cord
ing
to A
JC lu
ng c
ance
r sta
ging
sys
tem
• all
pat
ients
unde
rwen
t exte
nded
sur
gical
lymph
nod
e sta
ging
rega
rdles
s of
PET
or C
Tfin
ding
s or
nod
al s
ize•
PET
and
CT a
vaila
ble
to s
urge
on d
urin
g su
rger
y•
patie
nts
unde
rgoi
ng le
ft th
orac
otom
y ha
d lim
ited
sam
plin
g of
med
iast
inal
lym
ph n
odes
; car
ewa
s ta
ken
to re
sect
all p
reop
erat
ively
stag
ed p
ositiv
e lym
ph n
odes
• PE
T an
d CT
cor
rela
ted
with
hist
opat
holo
gic
resu
lts
Lim
itatio
ns o
f stu
dy d
esig
n•
Real
-tim
e pr
ospe
ctive
des
ign
uncle
ar•
Patie
nt s
ourc
e un
clear
and
hig
h pr
obab
ility
of m
alig
nanc
y in
sur
gica
l ser
ies
(pot
entia
lre
ferra
l bia
s)•
Onl
y bi
opsy
ver
ified
case
s an
alyz
ed (w
ork
up b
ias)
• St
rong
cor
relat
ion b
etwe
en im
aging
resu
lts a
nd b
iopsy
con
firm
ation
(diag
nosti
c re
view
bias
) min
imize
d by
nod
al s
ampl
ing
Stud
y
Stei
nert
et a
l. (1
997)
(Uni
vers
ity H
ospi
tal,
Zuric
h,Sw
itzer
land
)
December 1998
MTA98-032 MDRC Technology Assessment Program - PET Update - Page A4-15
Resu
lts/C
omm
ents
Dete
ctin
g m
edia
stin
al in
volv
emen
t (15
pos
itive
cas
es, 3
5 ne
gativ
eca
ses)
PET:
Se=
67%
; Sp=
97%
; acc
urac
y=88
%; P
PV=9
1%; N
PV=8
7%CT
: Se
=67%
; Sp=
63%
; acc
urac
y=64
%; P
PV=4
3%; N
PV=8
1%PE
T +
CT:
Se=
93%
; Sp=
97%
; acc
urac
y=96
%; P
PV=9
3%; N
PV=9
7%•
of 1
8 dis
cord
ant r
esult
s, CT
was
cor
rect
in 1/
18, a
nd P
ET w
as c
orre
ctin
17/
18 (p
=0.0
04)
Auth
ors’
com
men
ts•
med
iast
inos
copy
cou
ld b
e om
itted
in p
atie
nts
with
nor
mal
CT
and
PET
or in
pat
ients
with
abn
orm
al CT
but
nor
mal
PET
• al
l pat
ient
s wi
th a
bnor
mal
med
iast
inal
PET
sho
uld
still
proc
eed
toin
vasiv
e m
edia
stin
al s
tagi
ng, t
o be
sur
e th
at n
o pa
tient
with
N0
or N
1di
seas
e is
deni
ed th
e ch
ance
of c
ure
by d
irect
sur
gica
l res
ectio
n•
in th
is st
udy
the
need
for i
nvas
ive m
edia
stin
al s
tagi
ng c
ould
be
redu
ced
to 1
3 or
50
patie
nts,
resu
lting
in im
porta
nt s
aving
s in
oper
atio
n tim
e•
if ac
cura
cy o
f med
iast
inal
PET
will
be c
onfir
med
in fu
ture
dat
a, it
islik
ely
that
PET
will
subs
tant
ially
cha
nge
curre
nt c
linica
l pra
ctice
of
stagin
g of
NSC
LC
Patie
nts/
Met
hods
Purp
ose
to c
ompa
re p
rosp
ectiv
ely??
CT,
PET
, and
PET
+ C
T in
stagin
g m
edias
tinal
lymph
nod
es in
patie
nts w
ith N
SCLS
Case
sUn
know
n #
of p
atie
nts
with
sus
pect
ed o
r bio
psy
prov
en N
SCLC
who
wer
e po
tent
ially
oper
able
afte
r sta
ndar
d st
agin
g fo
r dist
ant m
etas
tase
s•
exclu
sion
crite
ria:
diab
etes
; tre
atm
ent w
ith o
ral c
ortic
oste
roid
s; is
chem
ic ca
rdio
myo
path
y;m
edias
tinal
invas
ion o
f prim
ary
tum
or; o
bviou
s bu
lky m
etas
tase
s•
reco
rds
of 5
0 pa
tient
s tre
ated
bet
ween
9/9
5 an
d 4/
96 w
ere
anal
yzed
- sq
uam
ous=
32; a
deno
carc
inom
a=10
; lar
ge c
ell=
8-
T1==
3;T2
=32;
T3=
15; N
0=35
; N2=
15
Met
hods
• in
terp
reta
tion
of c
ontra
st C
T by
two
inte
rpre
ters
blin
ded
to b
ronc
hosc
opic
or p
atho
logi
cfin
ding
s; p
ositiv
e no
de ≥
15
mm
long
axis
dia
met
er•
SUV
PET
imag
es in
terp
rete
d bli
nded
to c
linica
l, CT
, and
pat
holog
ic da
ta b
y tw
oin
depe
nden
t rea
ders
; five
poi
nt s
emiq
uant
itativ
e sc
ale
used
1-5
; pos
itive
node
= 4
or 5
• CT
+ P
ET v
isual
ly in
terp
rete
d by
two
read
ers
blin
ded
to p
atho
logi
c da
ta•
surg
ical s
tagi
ng d
one
by m
edia
stin
osco
py a
nd in
traop
erat
ive s
tagi
ng in
cas
e of
rese
ctio
n•
CT, P
ET, a
nd s
urgi
cal s
tagi
ng c
arrie
d ou
t with
in o
ne m
onth
• M
LN m
ap u
sed
for i
mag
ing
and
surg
ical s
tagi
ng•
data
ana
lyzed
by
patie
nt
Lim
itatio
ns o
f stu
dy d
esig
n•
Smal
l sam
ple
size
• Li
mite
d pa
tient
dat
a; h
igh
prob
abilit
y of
mal
igna
ncy
(pot
entia
l ref
erra
l bia
s)•
pros
pect
ive d
esig
n un
clear
• on
ly bi
opsy
ver
ified
case
s an
alyz
ed (w
ork
up b
ias)
• as
socia
tion
betw
een
test
resu
lts a
nd d
eter
min
atio
n of
gol
d st
anda
rd u
ncle
ar (d
iagn
ostic
revie
w bi
as)
• ex
tent
of n
odal
sam
plin
g no
t rep
orte
d•
blin
ding
to c
linica
l dat
a on
visu
al in
terp
reta
tion
not r
epor
ted
(pot
entia
l tes
t rev
iew
bias
)•
met
hods
for a
sses
sing
chan
ges
in tr
eatm
ent n
ot re
porte
d
Stud
y
Vans
teen
kiste
et a
l. (1
997)
(UHG
, Leu
ven,
Bel
gium
)
December 1998
MTA98-032 MDRC Technology Assessment Program - PET Update - Page A4-16
Resu
lts/C
omm
ents
Dete
ctio
n of
kno
wn
prim
ary
(29
lesi
ons)
PET
FDG
upt
ake:
9.1±
4.6
(TM
R±S
D)CT
: 43
.3±1
8.5m
m (m
ean
± SD
)
Med
iast
inal
lym
ph n
ode
met
asta
ses
(17
posi
tive
regi
ons,
54
nega
tive
regi
ons)
PET:
Se=
76%
; Sp=
98%
*; Ac
cura
cy=9
3%**
; PPV
=93%
; NPV
=93%
CT:
Se=6
5%; S
p=87
%*;
Accu
racy
=82%
**; P
PV=6
1%; N
PV=8
9%*,*
* (P<
0.05
)•
All P
ET fa
lse n
egat
ives
< 7m
m in
sho
rt ax
is di
amet
er d
ue to
par
tial
volu
me
effe
ct•
smal
lest
true
pos
itive
on P
ET w
as 7
mm
in s
hort
axis
diam
eter
• CT
false
pos
itives
cau
sed
by n
on-s
pecif
ic in
flam
mat
ory
chan
ges
and
an e
nlar
ged
trach
eobr
onch
ial ly
mph
nod
e of
unr
epor
ted
caus
e•
PET
false
pos
itive
caus
ed b
y an
enla
rged
trac
heob
ronc
hial ly
mph
node
of u
nrep
orte
d ca
use
Auth
ors’
Com
men
ts•
volu
ntar
y an
d in
volu
ntar
y m
ovem
ent c
an c
ontri
bute
toun
dere
stim
atio
n of
FDG
upt
ake
• m
edias
tinal
evalu
ation
may
hav
e be
en lim
ited
by P
ET fi
eld o
f view
• us
e of
qua
ntita
tive
anal
ysis
usin
g PE
T is
limite
d in
lym
ph n
ode
eval
uatio
n du
e its
com
plex
ity, t
he in
fluen
ce o
f the
par
tial v
olum
eef
fect,
and
limite
d vis
ual id
entifi
catio
n of
lym
ph n
odes
• au
thor
s su
gges
t PET
as
a co
mpl
emen
tary
dia
gnos
tic m
etho
d wi
thCT
; im
prov
emen
ts in
tech
nica
l and
qua
ntita
tive
met
hods
sho
uld
impr
ove
the
diag
nost
ic ab
ility
of P
ET
Patie
nts/
Met
hods
Purp
ose
to c
ompa
re ?
pros
pect
ively
FDG
PET
with
CT
in th
e de
tect
ion
of m
edia
stin
al ly
mph
nod
em
etas
tase
s
Case
s29
new
ly di
agno
sed
patie
nts
with
NSC
LC o
f mixe
d ty
pes
who
had
unde
rgon
e su
rger
y an
dwh
o ha
d pa
thol
ogic
conf
irmat
ion
of d
iseas
e•
aden
ocar
cinom
a=18
; squ
amou
s=9;
ade
nosq
uam
ous=
1; la
rge
cell=
1•
N0-N
1=17
; N2=
11; N
3=1
• 13
2 ou
t of 2
61 m
edias
tinal
lymph
nod
es w
ere
surg
ically
rese
cted
and
histo
path
ology
conf
irmed
of
- 71
regi
ons
had
CT, P
ET, a
nd h
istop
atho
logi
c in
form
atio
n an
d we
re in
clude
d in
the
stud
y
Met
hods
• m
edias
tinal
lymph
nod
es c
lassif
ied in
to n
ine re
gions
bas
ed o
n m
appin
g pr
opos
ed b
y th
eJa
pan
Lung
Can
cer S
ocie
ty•
FDG
upt
ake
mea
sure
d by
TM
R of
the
prim
ary
tum
or•
visua
l inte
rpre
tatio
n of
PET
imag
es p
erfo
rmed
by
3 nu
clear
med
icine
read
ers
• PE
T po
sitive
crit
eria
wer
e FD
G u
ptak
e in
nod
es>
that
in o
ther
med
iast
inal
stru
ctur
es•
cont
rast
CT
inte
rpre
ted
by 2
radi
olog
ists
• CT
pos
itive
node
s ≥1
0 m
m o
n sh
ort a
xis d
iam
eter
• go
ld s
tand
ard
biop
sy o
btai
ned
surg
ically
• da
ta a
nalyz
ed b
y no
dal r
egio
n
Lim
itatio
ns o
f stu
dy d
esig
n•
not r
eal-t
ime
pros
pect
ive d
esig
n (re
ferra
l bia
s)•
regio
ns in
clude
d in
analy
sis re
pres
ent a
n un
know
n nu
mbe
r of p
atien
ts•
only
biop
sy v
erifie
d ca
ses
anal
yzed
(wor
k up
bia
s)•
blin
ding
of r
eade
rs n
ot re
porte
d (te
st re
view
bias
)•
asso
ciatio
n be
twee
n te
st re
sults
and
det
erm
inat
ion
of g
old
stan
dard
unc
lear
, and
ext
ent
of n
odal
sam
plin
g no
t des
crib
ed (p
oten
tial d
iagn
ostic
revie
w bi
as)
Stud
y
Sasa
ki et
al.
(199
6)(K
yush
u Un
ivers
ity, F
ukuo
ka,
Japa
n)
December 1998
MTA98-032 MDRC Technology Assessment Program - PET Update - Page A4-17
Resu
lts/C
omm
ents
Diag
nost
ic A
ccur
acy
(37
mal
igna
ncie
s, 1
5 be
nign
)O
vera
llPE
T+CT
: Se
=95%
; Sp=
87%
; acc
urac
y=92
%no
dule
s ≤
1.5
cmPE
T+CT
: Se
=83%
; Sp=
100%
nodu
les >
1.5
cmPE
T+CT
: Se
=100
%; S
p=67
%
• 2
false
pos
itives
due
to h
istop
lasm
a gr
anul
oma
with
act
ive in
flam
mat
ion
• 2
false
neg
ative
s we
re 1
cm
sca
r ade
noca
rcin
oma
and
aden
ocar
cinom
a
Diag
nost
ic T
hink
ing
Effic
acy
Baye
s’ T
heor
em•
LR fo
r mal
igna
nt S
PN w
ith a
bnor
mal
PET
=7.1
1 (9
5% C
I, 6.
36 to
7.9
6)•
LR fo
r mal
igna
nt S
PN w
ith n
orm
al P
ET=0
.06
(95%
CI,
0.05
to 0
.07)
ROC
curv
e an
alys
isPE
T alo
ne w
as th
e be
st pr
edict
or o
f mali
gnan
cy a
t diffe
rent
leve
ls of
pCA
,th
e st
anda
rd c
riter
ia th
e wo
rst,
and
stan
dard
crit
eria
+ P
ET w
asin
term
edia
te
Patie
nts/
Met
hods
Purp
ose
to c
ompa
re th
e pr
obab
ility
of c
ance
r (pC
A) in
a S
PN u
sing
stand
ard
crite
ria w
ith B
ayes
ian A
nalys
is an
dPE
T (re
trosp
ectiv
e an
alysis
)
Case
s52
con
secu
tive
patie
nts
(37
mal
igna
nt c
ases
, 15
beni
gn c
ases
) who
met
the
follo
wing
sel
ectio
n cr
iteria
:•
unde
rwen
t PET
imag
ing b
etwe
en A
pril 1
990
and
Febr
uary
199
4•
nonc
alcif
ied,
non
cavit
ary
SPN
base
d on
CXR
and
CT
class
ified
as in
dete
rmin
ate
• ag
e >
30 y
ears
• no
dule
size
≤ 3
cm
• gr
oup
includ
ed 3
pat
ients
with
extr
atho
racic
mali
gnan
cy a
nd o
ne p
atien
t with
sta
ble n
odule
for >
2 y
rsRe
porte
d pa
tient
cha
ract
erist
ics:
• 43
men
, 9 w
omen
; mea
n ag
e=63
.6 y
rs ±
11.
3yrs
; 79%
cur
rent
sm
oker
s of
whi
ch 5
2% s
mok
ed ≥
20
cigs/d
ay•
Edge
cha
racte
ristic
s (%
mali
gnan
t cas
es v
s. %
ben
ign c
ases
): S
harp
, sm
ooth
=14%
vs.2
0%;
Lobu
late
d=30
% v
s. 4
0%; S
light
ly irr
egul
ar w
/ few
spi
cula
tions
=38%
vs.
33%
; Gro
ssly
irreg
ular
and
spicu
late
d=19
% v
s. 7
%
Met
hods
• PE
T pe
rform
ed w
ithin
2-4
wee
ks a
fter C
T; C
T de
nsito
met
ry n
ot p
erfo
rmed
• hi
stol
ogic
diag
nosis
obt
aine
d by
thor
acot
omy,
med
iast
inos
copy
, bro
ncho
scop
y, o
r nee
dle
lung
bio
psy
• qu
alita
tive
PET
scan
s re
ad b
y on
e re
ader
blin
ded
to h
istol
ogy;
clin
ical a
nd ra
diol
ogic
data
ava
ilabl
e to
the
read
er v
aried
, but
size
and
loca
tion
was
know
n in
all p
atien
ts•
beni
gn P
ET=n
o fo
cal F
DG u
ptak
e; m
alig
nant
PET
=foc
al F
DG a
ccum
ulat
ion
grea
ter t
han
surro
undi
ngtis
sue
but m
ore
than
mild
• no
dule
edg
e on
CT
inte
rpre
ted
inde
pend
ently
by
2 pu
lmon
olog
ists
usin
g 4
type
cla
ssific
atio
n sy
stem
blind
ed to
clin
ical d
iagno
sis; d
iscre
pant
inte
rpre
tatio
ns re
ache
d by
con
sens
us•
odds
-likel
ihoo
d of
mal
igna
ncy
estim
ated
usin
g Ba
yes
Theo
rem
• st
anda
rd c
riter
ia fo
r pro
babi
lity o
f can
cer (
pCA)
bas
ed o
n pa
tient
’s ag
e, s
mok
ing
hist
ory,
hist
ory
of p
rior
mal
igna
ncy,
nod
ule
size
and
edge
, and
pre
senc
e of
cal
cifica
tion
• pC
A of
sta
ndar
d cr
iteria
com
pare
d to
sta
ndar
d cr
iteria
+ P
ET a
nd P
ET a
lone
Lim
itatio
ns o
f Stu
dy D
esig
n•
retro
spec
tive
desig
n•
all p
atien
ts ha
s inv
asive
biop
sy d
eter
mina
tion,
imply
ing a
high
inde
x of
sus
picion
for m
align
ancy
(refe
rral b
ias)
• so
urce
of p
atie
nt c
ohor
t inf
luen
ced
by te
st re
sults
(wor
k up
bia
s)•
asso
ciatio
n be
twee
n te
st re
sults
and
det
erm
inat
ion
of g
old
stan
dard
unc
lear
(pot
entia
l dia
gnos
tic re
view
bias
)•
blin
ding
of c
linica
l and
radi
olog
ic in
form
atio
n va
ried
(test
revie
w bi
as)
• PE
T an
d ot
her t
ests
not i
ndep
ende
nt, a
requ
irem
ent o
f Bay
es’ T
heor
em•
pre-
PET
prob
abilit
y of
can
cer i
n pa
tient
s un
know
n
Dia
gn
ost
ic A
ccu
racy
an
d D
iag
no
stic
Th
inki
ng
Eff
icac
y S
tud
ies
of
FD
G P
ET
in S
olit
ary
Pu
lmo
nar
y N
od
ule
s
Stud
y
Dewa
n et
al.
(199
7)(C
reig
hton
Uni
vers
ity a
ndVA
MC
Om
aha,
NE)
December 1998
MTA98-032 MDRC Technology Assessment Program - PET Update - Page A4-18
Resu
lts/C
omm
ents
Defin
ing
SPN
(60
mal
igna
nt c
ases
, 29
beni
gn c
ases
) rep
orte
d w
ith 9
5% C
IO
vera
ll (≤
4cm
)SU
V: S
e=92
% (8
2-10
0%);
Sp=9
0% (7
9-10
0%);
Acc=
91%
; LR+
=9.0
; LR-
=0.0
9Vi
sual:
Se=
98%
(95-
100%
); Sp
=69%
(57-
81%
); Ac
c=89
%; L
R+=3
.0; L
R-=0
.02
• M
align
ant m
ean
SUV
± 1
SD=6
.9 ±
3.9
vs.
benig
n =1
.7 ±
1.0
(P <
0.0
01)
≤ 1.
5cm
(15
posit
ives,
19
nega
tives
)SU
V: S
e=80
% (6
0-10
0%);
Sp=9
5% (8
5-10
0%);
Acc=
88%
; LR+
=15.
0; L
R-=0
.2Vi
sual
: Se=
100%
(100
-100
%);
Sp=7
4% (5
5-93
%);
Acc=
85%
; LR+
=4.0
; LR-
=0.0
>1.5
cm (4
5 po
sitive
s, 1
0 ne
gativ
es)
SUV:
Se=
96%
(90-
100%
); Sp
=80%
(55-
100%
); Ac
c=93
%; L
R+=5
.0; L
R-=0
.06
Visu
al: S
e=98
% (9
4-10
0%);
Sp=6
0% (4
6-74
%);
Acc=
91%
; LR+
=2.0
; LR-
=0.0
4≤
3.0c
m (5
1 po
sitive
s, 2
6 ne
gativ
es)
SUV:
Se=
90%
(82-
98%
); Sp
=92%
(85-
99%
); Ac
c=91
%; L
R+=1
2.0;
LR-
=0.1
Visu
al: S
e=98
% (9
4-10
0%);
Sp=6
9% (5
6-82
%);
Acc=
88%
; LR+
=3.0
; LR-
=0.0
3
Oth
er fi
ndin
gs•
Κ=0
.95;
2/8
9 di
scre
pant
cas
es c
ause
d by
gra
nulo
ma
and
acut
e in
flam
mat
ion
• be
nign
con
ditio
ns:
gran
ulom
a (7
), co
ccid
iom
ycos
is (4
), be
nign
cel
lula
r deb
ris(4
), no
nspe
cific
infla
mm
atio
n (3
), ne
crot
izing
gra
nulo
ma
(3),
fibro
sis (1
),he
man
giom
a (1
), as
perg
illosis
(1),
met
apla
sia (1
)•
mal
igna
ncie
s: N
SCLC
(50)
, mel
anom
a (5
), Ho
dgkin
’s lym
phom
a (1
), sm
all-c
ell
(1),
mal
igna
nt n
eura
l tum
or (1
), m
alig
nant
car
cinoi
d (1
), co
lon
canc
er (1
)•
SUV
false
neg
ative
s (5
): 2
.0cm
bro
nchi
oalve
olar
can
cer,
1.5c
m s
quam
ous
cell
canc
er, 1
.0cm
mel
anom
a m
ets,
2.5
cm s
quam
ous
cell c
ance
r in
a pa
tient
with
bloo
d gl
ucos
e=34
1; a
ll fal
se n
egat
ives
in R
upp
er lo
be•
Visu
al fa
lse n
egat
ives
(1):
2.0c
m b
ronc
hioa
lveol
ar c
ance
r•
SUV
and
visua
l fals
e po
sitive
s (3
): g
ranu
loma,
nec
rotiz
ing g
ranu
loma,
necr
otizi
ng g
ranu
loma
with
hist
oplas
mos
is; 2
in R
upp
er lo
be, 1
in l
uppe
r lob
e•
Seru
m g
luco
se v
alue
s on
61
patie
nts,
mea
n ±
SD=9
9 ±
56 m
g/dL
; 2/7
diab
etics
had
elev
ated
glu
cose
leve
ls wi
th 1
false
pos
itive
and
2 fa
lse n
egat
ive re
sults
• In
4 c
ases
CXR
did
not i
dent
ify th
e SP
N
Auth
ors’
com
men
ts•
Resu
lts fr
om v
isual
ana
lysis
mig
ht b
e m
ore
help
ful t
han
quan
titativ
e an
alys
is fo
rsm
all n
odul
es (≤
1.5
cm) a
nd in
cas
es in
whic
h ele
vate
d glu
cose
leve
ls ar
eun
avoid
able
to re
duce
the
num
ber o
f fals
e ne
gativ
e ca
ses
• G
iven
the
mult
isite
nat
ure
of th
e stu
dy a
nd m
ultipl
e ra
diolog
ists
used
to in
terp
ret
the
films
acro
ss s
ites,
study
pop
ulatio
n is
repr
esen
tativ
e of
the
prop
ortio
n of
inde
term
inat
e SP
Ns fr
om th
e sit
es in
clude
d
Patie
nts/
Met
hods
Purp
ose
To p
rosp
ectiv
ely e
valua
te th
e dia
gnos
tic a
ccur
acy
of F
DG-P
ET im
aging
in e
valua
ting
SPNs
Case
s89
of 1
05 c
onse
cutiv
e pa
tient
s wh
o m
et th
e fo
llowi
ng in
clusio
n cr
iteria
:•
Imag
ing
perfo
rmed
bet
ween
Oct
ober
199
3 an
d Au
gust
199
4•
SPNs
0.7
cm-4
.0cm
in s
ize v
isual
ized
on C
T•
Cons
ider
ed in
dete
rmin
ate
for m
alig
nanc
y by
CXR
and
CT
and
clini
cal d
ata
• De
finitiv
e pa
thol
ogic
conf
irmat
ion
by T
TNA
(n=2
9) o
r sur
gery
(n=6
0)Ex
clude
d pa
tient
s:
8 no
def
initiv
e pa
thol
ogy;
4 d
efin
itely
beni
gn S
PN o
n CX
R/CT
; 2 w
/o C
TRe
porte
d pa
tient
cha
ract
erist
ics:
• 61
men
, 28
wom
en; m
ean
age=
63
± 9.
5 yr
s
Met
hods
• Im
agin
g pe
rform
ed p
rior t
o tre
atm
ent o
f SPN
• AP
and
late
ral C
XR a
nd C
T of
at l
east
ches
t and
adr
enals
obt
ained
; thin
-sec
tion
trans
axial
imag
es a
nd IV
con
trast
used
in s
ome
studie
s•
Inde
pend
ent q
ualita
tive
inte
rpre
tatio
n of
CXR
and
CT
by re
ader
s at
2 p
artic
ipat
ing
sites
oth
erth
an w
here
the
stud
ies
were
per
form
ed, b
linde
d to
clin
ical,
PET,
and
gol
d st
anda
rd re
sults
• Se
miqu
antita
tive
anal
ysis
(SUV
) per
form
ed; S
UV>
2.5
= m
alig
nant
• In
depe
nden
t visu
al an
alysis
of P
ET b
y 2
read
ers
(of 3
ava
ilable
read
ers)
blin
ded
to c
linica
l,CX
R/CT
, and
gold
sta
ndar
d re
sults
in e
ach
case
; foc
al up
take
> m
edias
tinal
blood
poo
lst
ruct
ures
= m
alig
nant
• PE
T co
mpa
red
to h
istol
ogy;
Se,
Sp,
acc
urac
y, a
nd L
Rs c
alcu
late
d•
Κ v
alue
ass
esse
d fo
r int
erob
serv
er v
aria
bility
Lim
itatio
ns o
f stu
dy d
esig
n•
smal
l sam
ple
size
in s
ubgr
oup
anal
yses
• ?
real
-tim
e pr
ospe
ctive
dat
a co
llect
ion
or r
etro
spec
tive
data
col
lect
ion
from
sur
gica
l ser
ies;
cond
itiona
l inde
pend
ence
am
ong
test
s un
clear
• hi
gh p
reva
lenc
e of
mal
igna
ncy
in s
tudy
pop
ulat
ion
(refe
rral b
ias)
• in
clusio
n cr
iterio
n of
bio
psy
verif
icatio
n bi
ased
towa
rd p
atie
nts
with
hig
h pr
obab
ility
of m
alig
nanc
y(w
ork-
up b
ias)
• bli
nding
of s
urge
ons
to P
ET re
sults
not
alw
ays
cond
ucte
d fo
r eth
ical r
easo
ns (d
iagno
stic
revie
wbi
as)
• pr
e-PE
T pr
obab
ility
of c
ance
r in
patie
nts
unkn
own
Note
: So
me
sites
(eg.
Cre
ight
on U
nive
rsity
Med
ical C
ente
r, O
mah
a, N
ebra
ska)
may
hav
e in
clude
dda
ta fr
om p
revio
usly
publ
ished
pat
ient
ser
ies,
whi
ch w
ere
revie
wed
in th
e 19
96 M
DRC
PET
asse
ssm
ent.
Stud
y
Lowe
et a
l. (1
998)
(mul
ti-sit
e st
udy
from
9U.
S. s
ites)
December 1998
MTA98-032 MDRC Technology Assessment Program - PET Update - Page A4-19
Resu
lts/C
omm
ents
Dete
ctin
g re
curr
ence
s ov
eral
l** (5
5 pa
tient
s w
ith re
curr
ence
s, 6
with
sca
r)PE
T: S
e=98
%; S
p=83
%CT
: in
suffi
cient
dat
a to
cal
cula
te re
sults
CTAP
: ins
uffic
ient d
ata
to c
alcula
te re
sults
Dete
ctin
g liv
er le
sion
s** (
104
mal
igna
nt le
sion
s, 2
3 be
nign
lesi
ons)
PE
T: S
e=91
%; S
p=96
%; a
ccur
acy=
92%
CT:
Se=8
1%; S
p=60
%; a
ccur
acy=
78%
CTAP
: Se=
97%
; Sp=
5%; a
ccur
acy=
80%
Exclu
ding
les
ions
< 1
cm
(18
mal
igna
nt, 5
ben
ign)
PET:
Se=
99%
; acc
urac
y=98
%CT
: Se=
87%
; acc
urac
y=83
%CT
AP:
Se=9
7%; a
ccur
acy=
80%
Oth
er fi
ndin
gs•
If on
ly hi
stol
ogica
lly p
rove
n le
sions
wer
e in
clude
d, te
st c
hara
cter
istics
rem
aine
d wi
thin
1%
of a
bove
val
ues,
but
no
data
ava
ilabl
e to
repl
icate
calcu
latio
ns•
Accu
rate
diffe
rent
iatio
n of
pos
tsur
gica
l cha
nges
from
mal
igna
nt re
curre
nce:
PET=
12/
14 si
tes;
CT=
7/11
site
s; C
TAP=
5/11
site
s
Dete
ctin
g ex
trahe
patic
lesi
ons*
*(34
mal
igna
nt le
sion
s, 5
ben
ign
lesi
ons)
PET:
Se=
100%
CT:
Se=7
4%
Qua
ntita
tive
anal
ysis
of h
epat
ic le
sion
s**
• SU
R m
align
ant =
8.1
± 4
.1 v
s. SU
R be
nign
= 2.
0 ±
1.0
(p<
0.00
01)
• Fo
r ext
rahe
patic
lesio
ns th
e SU
R wa
s le
ss h
elpf
ul th
an C
T in
diffe
rent
iatin
gbo
wel u
ptak
e fro
m m
etas
tase
s
Ther
apeu
tic e
ffica
cy**
• PE
T he
lped
to p
lan s
urge
ry b
y ide
ntify
ing s
ite o
f rec
urre
nce
in 10
% o
f pat
ients
(n=6
)•
PET
helpe
d to
avo
id un
nece
ssar
y su
rger
y in
18%
of p
atien
ts (n
=11)
• Im
pact
of fa
lse p
ositiv
e an
d fa
lse n
egat
ive P
ET s
cans
on
patie
nt m
anag
emen
twa
s no
t rep
orte
d
**No
te:
PET
utilit
y wa
s ev
alua
ted
com
plem
enta
ry to
dia
gnos
tic te
sts
done
ear
lier
in th
e wo
rk u
p.
Patie
nts/
Met
hods
Purp
oses
Pros
pect
ive a
sses
smen
t:•
To a
sses
s th
e ac
cura
cy o
f FDG
PET
vs.
CT v
s. CT
arte
rial p
orto
grap
hy (C
TAP)
in d
etec
ting
liver
met
asta
ses
• To
ass
ess
the
accu
racy
of F
DG P
ET v
s. CT
in d
etec
ting
extra
hepa
tic m
etas
tase
s•
To e
valu
ate
the
impa
ct o
n m
anag
emen
t of p
atie
nts
with
recu
rrent
col
orec
tal c
arcin
oma
(retro
spec
tive)
Case
s52
con
secu
tive
patie
nts
pres
ente
d on
61
occa
sions
for e
valu
atio
n of
sus
pect
ed re
curre
nt c
arcin
oma
base
d el
evat
ed C
EA le
vels
or a
bnor
mal
find
ings
on
CT (i
nclu
des
9 re
peat
pat
ient
s)•
45 h
ad liv
er m
etas
tase
s, inc
luding
16
with
con
com
itant
extr
ahep
atic
disea
se, 1
0 ha
d ex
trahe
patic
dise
ase
only
• To
tal li
ver l
esio
ns:
104
mal
igna
nt, 2
3 be
nign
(0.
3 cm
-6 c
m in
size
)•
Tota
l ext
rahe
patic
lesio
ns:
34 m
alig
nant
, 5 b
enig
n•
Beni
gn c
ondi
tions
: No
rmal
liver
(7),
Post
-sur
gica
l site
(8),
Loca
l fib
rosis
(2),
Reso
lving
abs
cess
(1),
hepa
tic c
yst (
1), h
emat
oma
(1)
• 31
men
, 21
wom
en; M
ean
age
63 ±
11
yrs
Met
hods
• 40
pat
ient
s un
derw
ent a
bdom
inal
CT;
CT
porto
grap
hy=4
0; o
r bot
h=29
• PE
T, C
T, a
nd C
T po
rtogr
aphy
(bot
h wi
th c
ontra
st) p
erfo
rmed
with
in 2
mon
ths
of e
ach
othe
r•
Patie
nts
with
abn
orm
al PE
T sc
ans
in ex
tra-a
bdom
inal a
reas
had
add
itiona
l CT
scan
of t
hat r
egion
• PE
T vis
ually
inte
rpre
ted,
and
ana
lyzed
sem
iqua
ntita
tivel
y us
ing
SUR
corre
cted
for b
ody
weig
ht,
by tw
o nu
clear
med
icine
phy
sicia
ns; S
UR c
alcu
latio
ns e
xclu
ded
lesio
ns <
1 c
m in
dia
met
er•
CT a
nd C
T po
rtogr
aphy
inte
rpre
ted
inde
pend
ently
by
two
expe
rienc
ed ra
diol
ogist
s•
All r
eade
rs b
linde
d to
oth
er im
aging
resu
lts•
Dise
ase
conf
irmed
with
clin
ical o
r rad
iolo
gic
follo
w up
(n=1
7) o
r hist
opat
holo
gy o
btai
ned
surg
ically
(n=4
4) ,
exce
pt fo
r two
lesio
ns th
at w
ere
exam
ined
afte
r per
cuta
neou
s fin
e ne
edle
asp
iratio
n•
Surg
ical e
xam
and
intra
oper
ative
ultr
asou
nd u
sed
to c
onfir
m n
onre
secte
d liv
er le
sions
• Re
curre
nce
defin
ed p
atho
logi
cally
or b
y su
spec
ted
recu
rrenc
e on
imag
ing
• Ch
ange
s in
patie
nt m
anag
emen
t ret
rosp
ectiv
ely re
viewe
d wi
th th
e su
rgeo
ns
Lim
itatio
ns o
f Stu
dy D
esig
n•
High
pre
vale
nce
of m
alig
nanc
y an
d un
clear
pat
ient
sou
rce
(pot
entia
l ref
erra
l bia
s)•
Stro
ng a
ssoc
iation
bet
ween
imag
ing re
sults
and
cho
ice o
f pat
ient c
ohor
t; CT
of e
xtrah
epat
icar
eas
depe
nden
t on
PET
resu
lts (w
ork
up b
ias, m
inim
ized
by fo
llow
up o
f all p
atien
ts)•
Crite
ria fo
r pos
itive
test
and
cut
-off
for s
emiq
uant
itativ
e an
alys
is no
t rep
orte
d (p
oten
tial t
est r
evie
wbi
as)
• Bl
indi
ng to
oth
er c
linica
l info
rmat
ion
not r
epor
ted
(pot
entia
l tes
t rev
iew
bias
)•
Asso
ciatio
n be
twee
n im
agin
g te
st re
sults
and
gol
d st
anda
rd d
eter
min
atio
n un
clear
(pot
entia
ldi
agno
stic
revie
w bi
as)
• De
tails
of m
etho
ds fo
r eva
luat
ing
ther
apeu
tic e
ffica
cy n
ot re
porte
d
Dia
gn
ost
ic A
ccu
racy
an
d T
her
apeu
tic
Eff
icac
y S
tud
ies
of
FD
G P
ET
in C
olo
rect
al C
ance
r
Stud
y
Delb
eke
et a
l. (1
997)
(Van
derb
ilt Un
ivers
ityM
edica
l Cen
ter,
Nash
ville
, Ten
ness
ee)
December 1998
MTA98-032 MDRC Technology Assessment Program - PET Update - Page A4-20
Resu
lts/C
omm
ents
Defin
ing
loca
l pel
vic
recu
rren
ce (2
1 di
seas
e, 2
6 no
dis
ease
)PE
T+CT
: Se
=90%
*; Sp
=100
%; P
PV=1
00%
; NPV
=93%
; Acc
=96%
CT:
Se=5
7%*;
Sp=8
1%; P
PV=7
1%; N
PV=7
0%; A
cc=7
0%*(
P =
0.00
8)•
PET
corre
ctly
iden
tified
pre
senc
e of
dise
ase
in a
ll pat
ient
s wi
th tr
ue p
ositiv
e CT
findin
gs•
PET
was
usef
ul in
diffe
rent
iatin
g po
stop
erat
ive fi
bros
is fro
m re
curre
nce
in 6
patie
nts
with
pos
itive
CT s
cans
• PE
T co
nfirm
ed d
iseas
e in
4 p
atie
nts
with
equ
ivoca
l CT
findi
ngs
• 2
false
neg
ative
s on
bot
h PE
T an
d CT
wer
e dif
fuse
mes
orec
tal a
nd a
nasto
mot
ichi
stol
ogie
s pr
oven
by
trans
rect
al U
S-gu
ided
bio
psie
s.
Defin
ing
hepa
tic m
etas
tase
s (2
3 di
seas
e, 3
5 no
dis
ease
)PE
T+CT
: Se
=96%
*; Sp
=100
%; P
PV=1
00%
; NPV
=97%
; Acc
=98%
CT:
Se=7
4%*;
Sp=8
6%; P
PV=7
7%; N
PV=8
3%; A
cc=8
1%*(
P =
0.02
)•
PET
iden
tified
all 5
pat
ient
s wi
th s
olita
ry m
etas
tase
s, C
T id
entifi
ed 2
pat
ient
swi
th so
litary
met
s•
PET
ident
ified
17/1
8 pa
tient
s an
d CT
iden
tified
10/
18 p
atien
ts wi
th m
ultipl
eles
ions
• 0n
e fa
lse n
egat
ive o
n bo
th C
T an
d PE
T fo
und
to b
e m
ultip
le s
uper
ficia
l hep
atic
lesio
ns u
p to
3 c
m in
dia
met
er
Defin
ing
extra
hepa
tic m
etas
tase
s (2
0 di
seas
e, 3
8 no
dis
ease
)PE
T id
entifi
ed e
xtra
-hep
atic
met
asta
ses
in 2
1 sit
es in
20
patie
nts,
of w
hich
9les
ions
were
miss
ed o
n CT
or C
XR__
____
____
____
____
____
____
____
____
____
____
____
____
____
____
____
__
Th
erap
eutic
effi
cacy
PE
T inf
luenc
ed c
linica
l man
agem
ent i
n 47
% (1
0/21
) pat
ients
with
loca
l rec
urre
ntdi
seas
e, 4
3% (1
0/23
) with
hep
atic
met
asta
ses,
and
38%
(8/2
0) w
ith e
xtra
hepa
ticm
etas
tase
s
Patie
nts/
Met
hods
Purp
ose
Retro
spec
tive
asse
ssm
ent:
• To
eva
luat
e PE
T ve
rsus
CT
for s
tagi
ng re
curre
nt a
nd m
etas
tatic
col
orec
tal c
ance
rs•
To a
sses
s th
e im
pact
of P
ET o
n cli
nica
l man
agem
ent o
f pat
ient
s wi
th c
olor
ecta
l can
cer
Case
s58
pat
ients
had
PET
betw
een
1/91
and
1/9
5 fo
r eva
luatio
n of
sus
pecte
d re
curre
nt (n
=47)
or
adva
nced
prim
ary
(n=1
1) d
iseas
e:
• ba
sed
on h
igh
clini
cal s
uspi
cion
and
equi
voca
l or p
ositiv
e CT
find
ings
(n=3
9) o
r clin
ical
susp
icion
alon
e, in
cludin
g ra
ised
CEA
levels
with
nor
mal
CT (n
=19)
• 33
men
, 25
wom
en; m
ean
age
60 y
rs. (
23-8
1 yr
s)•
beni
gn c
ondi
tions
not
repo
rted
in re
prod
ucib
le d
etai
l
Met
hods
• Al
l pat
ient
s un
derw
ent c
olon
osco
py a
nd c
ontra
st C
T of
che
st, a
bdom
en a
nd p
elvis
with
in 4
wks
prio
r to
PET
• CT
inte
rpre
ted
for e
xtent
of l
ocal
pelvi
c re
curre
nce
and
pres
ence
of m
etas
tase
s•
Qua
litativ
e PE
T in
terp
rete
d by
two
read
ers
with
acc
ess
to C
T re
sults
• M
alig
nanc
y=FD
G u
ptak
e m
oder
atel
y or
mar
kedl
y in
tens
e; b
enig
n=no
or m
ild u
ptak
e, o
r if
abno
rmal
ity id
entifi
ed o
n ot
her i
mag
ing
for w
hich
no
corre
spon
ding
abn
orm
ality
was
pre
sent
on
PET
• G
old
stan
dard
= su
rger
y, h
istol
ogy,
or b
oth
(n=4
0); c
linica
l and
radi
olog
ic fo
llow
up (n
=16)
;au
tops
y re
ports
(n=2
), an
d tre
atm
ent o
utco
mes
• Al
l pat
ients
follo
wed
for a
t lea
st 12
mon
ths
afte
r PET
or u
ntil d
eath
• Im
pact
of P
ET o
n pa
tient
man
agem
ent w
as a
sses
sed;
pos
itive
impa
ct=a
ltera
tion
in c
linica
lde
cisio
ns w
ith P
ET re
sults
Lim
itatio
ns o
f stu
dy d
esig
n•
? Co
nsec
utive
ser
ies•
Retro
spec
tive
anal
ysis
• Hi
gh p
reva
lenc
e of
mal
igna
ncy
(pot
entia
l ref
erra
l bia
s)•
Stro
ng a
ssoc
iatio
n be
twee
n te
st re
sults
and
cho
ice o
f pat
ient
coh
ort (
work
-up
bias
, min
imize
d by
follo
w up
of a
ll sub
jects
for a
t lea
st 12
mon
ths
afte
r pet
or u
ntil d
eath
)•
Crite
ria fo
r pos
itive
resu
lt on
imag
ing
not r
epor
ted
• Bl
indi
ng n
ot re
porte
d; c
t res
ults
ava
ilabl
e to
pet
read
ers
(test
revie
w bi
as)
• In
crem
enta
l valu
e of
pet
not
ass
esse
d•
Asso
ciatio
n be
twee
n te
st re
sults
and
gol
d st
anda
rd d
eter
min
atio
n un
clear
(pot
entia
l dia
gnos
ticre
view
bias
)•
Met
hods
for a
sses
sing
ther
apeu
tic e
fficac
y no
t rep
orte
d
Stud
y
Ogu
nbiyi
et a
l. (1
997)
(Was
hing
ton
Univ.
Scho
ol o
f Med
icine
, St.
Loui
s, M
issou
ri)
December 1998
MTA98-032 MDRC Technology Assessment Program - PET Update - Page A4-21
Resu
lts/C
omm
ents
Dete
ctin
g re
curre
nt d
isea
se (1
5 re
curre
nce,
7 n
o re
curre
nce)
PET:
Se=
100%
; Sp=
71%
; PPV
=89%
; NPV
=100
%•
2 fa
lse p
ositiv
es d
ue to
asy
mm
etric
acti
vity
in bo
wel a
nd b
ladde
r dive
rticu
lum,
and
incre
ased
upt
ake
in do
me
of liv
er in
a p
atien
t in
whom
a p
oor q
uality
PET
scan
was
pro
duce
d du
e to
larg
e pa
tient
size
____
____
____
____
____
____
____
____
____
____
____
____
____
____
____
____
Th
erap
eutic
effi
cacy
• G
uided
by
the
PET
resu
lts, c
urat
ive s
urge
ry w
as a
ttem
pted
in o
nly 4
or 1
5pa
tient
s wi
th d
iseas
e•
Neith
er fa
lse p
ositiv
es o
n P
ET re
sulte
d in
mism
anag
emen
t; bo
th p
atie
nts
had
equi
voca
l fin
ding
s, a
nd re
ferri
ng p
hysic
ians
opt
ed fo
r add
itiona
l rad
iolo
gic
and
follo
w up
stu
dies
• Al
l 5 p
atie
nts
with
neg
ative
PET
sca
ns w
ere
alive
and
dise
ase
free
9-24
mon
ths
afte
r PET
; 2 p
atie
nts
had
nega
tive
biop
sy o
f ana
stom
otic
site,
oth
er 3
patie
nts
had
no d
iseas
e pr
ogre
ssion
on
follo
w up
*Not
e: o
verla
ppin
g pa
tient
pop
ulat
ions
with
pre
vious
stu
dy
Patie
nts/
Met
hods
Pu
rpos
eTo
retro
spec
tivel
y as
sess
PET
in p
atie
nts
with
une
xpla
ined
risin
g ca
rcin
oem
bryo
nic
antig
en (C
EA)
leve
ls af
ter t
reat
men
t of c
olor
ecta
l can
cer
Ca
ses
22 o
f 128
pat
ients
with
a h
istor
y of
colo
recta
l car
cinom
a wh
o un
derw
ent P
ET fr
om 6
/93
to 6
/96,
were
enr
olle
d an
d we
re p
oten
tial c
andi
date
s fo
r exp
lora
tory
lapa
roto
my:
• all
had
plas
ma
CEA
level
> 5.
0 ng
/ml (
mea
n 25
ng/
ml),
nor
mal
imag
ing s
tudie
s, en
dosc
opy,
and
phys
ical e
xam
on
rout
ine fo
llow-
up•
all p
atien
ts ha
d no
rmal
CEA
levels
afte
r res
ectio
n of
their
prim
ary
tum
ors
• 17
men
, 5 w
omen
; age
s 17
-84
• Pr
imar
y sit
e: c
olon
(9),
rect
um (1
0), r
ecto
sigm
oid
(2),
appe
ndix
(1)
• St
age
B (1
0); S
tage
s C
(5),
C1 (2
), C2
(3);
Stag
e D
(2)
M
etho
ds•
Patie
nts
with
hist
ory
of re
ctal o
r rec
tosig
moid
car
cinom
a ha
d co
ntra
st CT
of c
hest,
abd
omen
,an
d pe
lvis
• Pa
tient
s wi
th h
istor
y of
col
on c
ance
r had
con
trast
CT
of a
bdom
en a
nd p
elvis
• CT
sca
ns p
erfo
rmed
≤ 4
wee
ks b
efor
e PE
T•
CT in
terp
rete
d in
“rou
tine
clini
cal f
ashi
on”
• PE
T int
erpr
eted
qua
litativ
ely in
“rou
tine
clinic
al fa
shion
”, inc
luding
cor
relat
ing w
ith C
T, a
nd b
yco
nsen
sus
of a
t lea
st tw
o re
ader
s•
PET
used
in tr
eatm
ent m
anag
emen
t at t
he d
iscre
tion
of th
e re
ferri
ng s
urge
on•
PET
corre
late
d wi
th h
istol
ogy,
long
term
radi
olog
ic an
d cli
nica
l fol
low-
up ≥
6 m
onth
s•
PET
true
posit
ive=c
onfir
mat
ion
by b
iops
y or
obv
ious
dise
ase
site
on fo
llow
up im
agin
g di
rect
edby
PET
and
with
in 6
mon
ths
of P
ET•
PET
true
nega
tive=
con
firm
ation
by
biops
y or
no
abno
rmali
ty ve
rified
by
othe
r im
aging
or
clinic
al fo
llow-
up w
ithin
6 m
onth
s of
PET
Li
mita
tions
of s
tudy
des
ign
• Re
trosp
ectiv
e an
alys
is•
High
pro
babi
lity o
f mal
igna
ncy
(pot
entia
l ref
erra
l bia
s)•
Met
hods
for i
mag
e in
terp
reta
tion
uncle
ar a
nd re
ader
s no
t blin
ded
to C
T(te
st re
view
bias
)•
Blin
ding
of P
ET re
sults
and
refe
renc
e st
anda
rd n
ot re
porte
d; s
trong
cor
rela
tion
betw
een
test
resu
lts a
nd g
old
stan
dard
det
erm
inat
ion
(dia
gnos
tic re
view
bias
)•
Met
hods
for s
yste
mat
ic as
sess
men
t of t
hera
peut
ic ef
ficac
y no
t rep
orte
d
Stud
y
Flan
agan
et a
l. (1
998)
(Was
hing
ton
Univ.
Scho
ol o
f Med
icine
, St.
Loui
s, M
issou
ri)