5
GENETICS Is it time to sound an alarm about false-positive cell-free DNA testing for fetal aneuploidy? Michael T. Mennuti, MD; Athena M. Cherry, PhD; Jennifer J. D. Morrissette, PhD; Lorraine Dugoff, MD C ell-free DNA (cfDNA) testing on maternal blood samples recently has been introduced in obstetrics prac- tice as a method for fetal aneuploidy screening. 1 This testing examines cfDNA fragments circulating in the maternal plasma that originate primarily from cells of the mother and to a lesser extent from placental cells. The DNA fragments from particular chromosomes are iden- tied by their nucleic acid sequence with a process that is known as massively parallel shotgun sequencing. 2,3 Assuming that the maternal cells are euploid, quantitative analysis of the cfDNA frag- ments is used to predict whether the cells of the pregnancy have the normal or abnormal copy number of specic chro- mosomes. Recently, one laboratory has introduced the use of parental single- nucleotide polymorphisms as a means of predicting the copy number of specic chromosomes in the placental cells. 4 Validation studies in high-risk pa- tients who have undergone invasive diagnostic testing have shown a sensi- tivity of >98% for detection of trisomy 21 and trisomy 18, with remarkably low false-positive rates (<0.5%) but somewhat lower sensitivity for trisomy 13. 5-10 Laboratories that developed the tests and clinical investigators who studied the test performance propose that positive results should be conrmed by invasive prenatal diagnosis before important parental decisions are made regarding the pregnancy. This recom- mendation is important because the positive predictive value of the test is imperfect (ie, some portion of preg- nancies with an abnormal result will not be affected). Although the test methods demonstrate excellent discrimination between euploid pregnancies and those with trisomy 13, 18, or 21, the necessity of establishing a quantitative cutoff in- evitably will result in some false-positive and false-negative results. We are aware of only one report of prospective performance of this testing in routine clinical practice. 11 Recently, 8 cases in which there were discordant results between an abnormal cfDNA test and the normal cytogenetic testing of the pregnancy have come to our attention. The cfDNA testing in these cases was not performed in the same laboratory. Massively parallel shotgun sequencing was used to identify the chromosomal origin of cfDNA fragments in all 8 cases. The discordance between the cfDNA testing and the cytogenetic tests dramatically underscores the impor- tance of offering invasive diagnostic testing after an abnormal cfDNA test result. We describe these cases to call attention to the urgent need for studies to understand the possible causes of discordant results so that this informa- tion may be used in the development of clinical guidelines for evaluation of similar cases. Case 1 A 40-year-old woman (G3P1011) had rst- and second-trimester serum inte- grated screening with adjusted risks for Down syndrome of 1:2200, trisomy 18 of 1:10,000, and open neural tube defect of 1:6000. The patient requested a cfDNA From the Department of Obstetrics and Gynecology (Drs Mennuti and Dugoff), Prenatal Cytogenetic Laboratory (Dr Mennuti); Department of Pathology and Laboratory Medicine and Clinical Cancer Cytogenetics (Dr Morrissette), Perelman School of Medicine, University of Pennsylvania, Philadelphia PA; and the Department of Pathology and Cytogenetics Laboratory, Stanford Hospital and Clinics, Stanford University School of Medicine, Palo Alto, CA (Dr Cherry). Received Dec. 10, 2012; revised March 5, 2013; accepted March 19, 2013. The authors report no conict of interest. Reprints not available from the authors. 0002-9378/$36.00 ª 2013 Mosby, Inc. All rights reserved. http://dx.doi.org/10.1016/j.ajog.2013.03.027 Testing cell-free DNA (cfDNA) in maternal blood samples has been shown to have very high sensitivity for the detection of fetal aneuploidy with very low false-positive results in high-risk patients who undergo invasive prenatal diagnosis. Recent observation in clinical practice of several cases of positive cfDNA tests for trisomy 18 and trisomy 13, which were not confirmed by cytogenetic testing of the pregnancy, may reflect a limitation of the positive predictive value of this quantitative testing, particularly when it is used to detect rare aneuploidies. Analysis of a larger number of false-positive cases is needed to evaluate whether these observations reflect the positive predictive value that should be expected. Infrequently, mechanisms (such as low percentage mosaicism or confined placental mosaicism) might also lead to positive cfDNA testing that is not concordant with standard prenatal cytogenetic diagnosis. The need to explore these and other possible causes of false-positive cfDNA testing is exemplified by 2 of these cases. Additional evaluation of cfDNA testing in clinical practice and a mechanism for the systematic reporting of false-positive and false-negative cases will be important before this test is offered widely to the general population of low-risk obstetric patients. In the meantime, incorporating information about the positive predictive value in pretest counseling and in clinical laboratory reports is recommended. These experiences reinforce the importance of offering invasive testing to confirm cfDNA results before parental decision-making. Key words: cfDNA, chromosome, fetus, positive predictive value, testing MONTH 2013 American Journal of Obstetrics & Gynecology 1 Clinical Opinion www. AJOG.org

Is it time to sound an alarm about false-positive cell-free DNA testing for fetal aneuploidy?

Embed Size (px)

Citation preview

Clinical Opinion www.AJOG.org

GENETICS

Is it time to sound an alarm about false-positive cell-free DNAtesting for fetal aneuploidy?Michael T. Mennuti, MD; Athena M. Cherry, PhD; Jennifer J. D. Morrissette, PhD; Lorraine Dugoff, MD

ell-free DNA (cfDNA) testing on

Testing cell-free DNA (cfDNA) in maternal blood samples has been shown to have veryhigh sensitivity for the detection of fetal aneuploidy with very low false-positive results inhigh-risk patients who undergo invasive prenatal diagnosis. Recent observation in clinicalpractice of several cases of positive cfDNA tests for trisomy 18 and trisomy 13, whichwere not confirmed by cytogenetic testing of the pregnancy, may reflect a limitation of thepositive predictive value of this quantitative testing, particularly when it is used to detectrare aneuploidies. Analysis of a larger number of false-positive cases is needed toevaluate whether these observations reflect the positive predictive value that should beexpected. Infrequently, mechanisms (such as low percentage mosaicism or confinedplacental mosaicism) might also lead to positive cfDNA testing that is not concordant withstandard prenatal cytogenetic diagnosis. The need to explore these and other possiblecauses of false-positive cfDNA testing is exemplified by 2 of these cases. Additionalevaluation of cfDNA testing in clinical practice and a mechanism for the systematicreporting of false-positive and false-negative cases will be important before this test isoffered widely to the general population of low-risk obstetric patients. In the meantime,incorporating information about the positive predictive value in pretest counseling and inclinical laboratory reports is recommended. These experiences reinforce the importanceof offering invasive testing to confirm cfDNA results before parental decision-making.

Key words: cfDNA, chromosome, fetus, positive predictive value, testing

C maternal blood samples recentlyhas been introduced in obstetrics prac-tice as a method for fetal aneuploidyscreening.1 This testing examines cfDNAfragments circulating in the maternalplasma that originate primarily fromcells of the mother and to a lesser extentfrom placental cells. The DNA fragmentsfrom particular chromosomes are iden-tified by their nucleic acid sequencewith a process that is known as massivelyparallel shotgun sequencing.2,3 Assumingthat the maternal cells are euploid,quantitative analysis of the cfDNA frag-ments is used to predict whether the cellsof the pregnancy have the normal orabnormal copy number of specific chro-mosomes. Recently, one laboratory hasintroduced the use of parental single-nucleotide polymorphisms as a meansof predicting the copy number of specificchromosomes in the placental cells.4

Validation studies in high-risk pa-tients who have undergone invasivediagnostic testing have shown a sensi-tivity of >98% for detection of trisomy21 and trisomy 18, with remarkablylow false-positive rates (<0.5%) butsomewhat lower sensitivity for trisomy

From the Department of Obstetrics andGynecology (Drs Mennuti and Dugoff), PrenatalCytogenetic Laboratory (Dr Mennuti);Department of Pathology and LaboratoryMedicine and Clinical Cancer Cytogenetics(Dr Morrissette), Perelman School of Medicine,University of Pennsylvania, Philadelphia PA; andthe Department of Pathology and CytogeneticsLaboratory, Stanford Hospital and Clinics,Stanford University School of Medicine,Palo Alto, CA (Dr Cherry).

ReceivedDec. 10, 2012; revisedMarch 5, 2013;accepted March 19, 2013.

The authors report no conflict of interest.

Reprints not available from the authors.

0002-9378/$36.00ª 2013 Mosby, Inc. All rights reserved.http://dx.doi.org/10.1016/j.ajog.2013.03.027

13.5-10 Laboratories that developed thetests and clinical investigators whostudied the test performance proposethat positive results should be confirmedby invasive prenatal diagnosis beforeimportant parental decisions are maderegarding the pregnancy. This recom-mendation is important because thepositive predictive value of the test isimperfect (ie, some portion of preg-nancies with an abnormal result will notbe affected). Although the test methodsdemonstrate excellent discriminationbetween euploid pregnancies and thosewith trisomy 13, 18, or 21, the necessityof establishing a quantitative cutoff in-evitably will result in some false-positiveand false-negative results.We are aware of only one report of

prospective performance of this testingin routine clinical practice.11 Recently, 8cases in which there were discordantresults between an abnormal cfDNA testand the normal cytogenetic testing of thepregnancy have come to our attention.

MONTH 2013

The cfDNA testing in these cases was notperformed in the same laboratory.Massively parallel shotgun sequencingwas used to identify the chromosomalorigin of cfDNA fragments in all 8cases. The discordance between thecfDNA testing and the cytogenetic testsdramatically underscores the impor-tance of offering invasive diagnostictesting after an abnormal cfDNA testresult. We describe these cases to callattention to the urgent need for studiesto understand the possible causes ofdiscordant results so that this informa-tion may be used in the development ofclinical guidelines for evaluation ofsimilar cases.

Case 1A 40-year-old woman (G3P1011) hadfirst- and second-trimester serum inte-grated screening with adjusted risks forDown syndrome of 1:2200, trisomy 18 of1:10,000, and open neural tube defect of1:6000. The patient requested a cfDNA

American Journal of Obstetrics & Gynecology 1

Clinical Opinion Genetics www.AJOG.org

test at 16 weeks 5 days’ gestation. Theresult was reported as negative for tri-somy 21 but showed an increased rep-resentation of chromosome 18 material.At 18.5 weeks’ gestation, the patient hada normal fetal anatomic survey andopted for amniocentesis. Amniocentesiswas reported to demonstrate a 46,XYchromosome constitution in 15 meta-phases from 12 colonies. Studies of DNAfrom parental blood samples and fromthe amniotic cells confirmed biparentalinheritance at 3 informative loci onchromosome 18. Ultrasound scans thatwere performed at 27.6 and 34.6 weeks’gestation showed normal interval fetalgrowth. At 41 weeks’ gestation, the pa-tient delivered a healthy male infantwho weighed 6 lbs 11 oz. The infant hada normal clinical examination and anuneventful hospital course and was dis-charged at 2 days of age. Samples forchromosome testing were not obtainedon the infant or placenta.

Case 2A 34-year-old woman (G2P0010) with amaternal age of 35 years at her projectedestimated delivery date opted for cfDNAtesting at 11 weeks’ gestation. The resultwas negative for trisomy 21 but showedan increased representation of chromo-some 18 material. Amniocentesis thatwas performed at 16 weeks’ gestationrevealed a normal 46,XY chromosomeanalysis, and the amniotic fluid alpha-fetoprotein was normal. Fetal anatomicsurvey by ultrasound scanning was nor-mal. The patient has not yet delivered atthe time of this report.

Case 3A 20-year-old primigravid woman hadfirst-trimester aneuploidy screeningwith adjusted risk for trisomy 21 of1:10,000 and trisomy 18 of 1:10,000.The first-trimester screening includedbeta human chorionic gonadotropin(hCG), pregnancy-associated plasmaproteineA (PAPP-A), maternal age, andnuchal translucency (NT). No commentregarding visualization of the nasal bonewas provided. The patient subsequentlyhad second-trimester maternal serumalpha-protein with adjusted risk of openneural tube defect of 1:1200. A fetal

2 American Journal of Obstetrics & Gynecology M

anatomic survey was normal at 19.3weeks’ gestation, and fetal biometrylagged 6 days behind the first-trimesterultrasound scan. At 23.3 weeks’ gesta-tion, there was a progressive lag in fetalgrowth that was noted by ultrasoundscanning with an estimate fetal weightat the 12th percentile. A hypoplasticnasal bone and echogenic bowel werenoted. The patient declined amniocen-tesis for chromosome analysis but optedto have cfDNA testing. The cfDNAshowed an increased representation ofchromosome 18 material. Amniocen-tesis was performed, and fluorescencein situ hybridization (FISH) was nega-tive for trisomy 18. At 28.4 weeks’ ges-tation, reversed end diastolic velocity onumbilical artery Doppler velocimetryprompted delivery of a male infantwho weighed 570 g (<5th percentilefor the gestational age) with an other-wise normal physical examination thatwas consistent with his gestational age.Karyotypes and single-nucleotide poly-morphism microarrays from the am-niocentesis and neonatal blood sampleswere reported as normal male, 46,XY.

Case 4A 41-year-old woman (G6P3023) hadcfDNA testing performed at 11 weeks’gestation as her initial screening foraneuploidy. Her cfDNA result reporteda >99% risk for trisomy 18. Subse-quently, first-trimester screening by betahCG, PAPP-A, and NT was performedand resulted in a risk of 1:720 for trisomy21 and 1:20,000 for trisomy 18. Infor-mation regarding the nasal bone was notprovided. A chorionic villus biopsy wasperformed at 13.1 weeks’ gestation. Bothdirect and long-term cultures showed anormal male karyotype (46,XY). Noabnormalities were noted on subsequentultrasound scanning, and the pregnancyis continuing.

Case 5A 35-year-old woman (G1P0) had first-trimester screening by beta hCG andNTmeasurement at 11 weeks’ gestationthat reported a risk of 1:87 for trisomy 21(hCG 1.48 multiples of the median;PAPP-A, 0.36 multiples of the median;NT, 1.6 mm). Information regarding the

ONTH 2013

nasal bone was not provided. The cfDNAtesting result was positive for trisomy 18.Amniocentesis was performed at 16weeks’ gestation. FISH on amniotic fluidcells was normal, as was chromosomeanalysis of cultured amniotic cells thatshowed a normal female karyotype(46,XX). Subsequent ultrasound scan-ning showed normal interval growth,and the fetal anatomic evaluation ap-peared normal. The pregnancy is con-tinuing at the time of this report.

Case 6A 37-year-old woman (G2P1001) optedfor cfDNA testing at 12.6 weeks’ gesta-tion. Ultrasound scanning showed asingleton pregnancy with a crown-rumplength measurement that was consistentwith the estimated delivery date and anuchal translucency measurement of1.6 mm. The cfDNA showed an in-creased representation of chromosome13 material. Amniocentesis was per-formed at 15 weeks’ gestation. Sixty-sixcells from 15 different colonies and anadditional 34 cells from amniotic cellsubcultures were examined. All 100 cellsthat were examined were found to have anormal 46,XX chromosome constitu-tion. The patient declined uniparentaldisomy testing and a microarray of theamniotic cells. Ultrasound scanning at18 weeks’ gestation showed appropriatefetal growth and normal fetal anatomy.Ultrasound assessment at 28 weeks’gestation is planned.

Case 7A 34-year-old (35 years old at herestimated delivery date) woman (G2P1)had cfDNA testing performed at 12.3weeks’ gestation. The result was reportedas positive for trisomy 13. Chorionicvilli sampling (CVS) was performedat 13.3 weeks’ gestation. FISH was posi-tive for trisomy 13 and male gender in50 cells that were examined. The patientelected to terminate the pregnancy.The karyotype from the CVS culturesubsequently was reported as 46XY in40 cells from 3 independent cultures.Single-nucleotide polymorphism micro-array analysis was reported as consistentwith a normal male. Examination ofa sample of tissue that was submitted

www.AJOG.org Genetics Clinical Opinion

for chromosome testing after termina-tion of the pregnancy showed chorionicvilli and decidua. No identifiable fetaltissue was available for study in thisportion of the sample. Interphase FISHshowed 2 signals for chromosome 13;a 46,XY chromosome constitution wasnoted in all 100 metaphases that wereexamined.

Case 8A 34-year-old woman (G2P1) hadcfDNA testing that was positive for tri-somy 13. Ultrasound scanning showed asingle intrauterine pregnancy that wasconsistent with a gestational age of 13.6weeks and a nuchal translucency thatmeasured 1.7 mm. Anatomic evaluationof the fetus was normal for the firsttrimester, with the exception of whatappeared to be mild dilation of therenal pelves. Direct preparation on CVSshowed 46,XX,þ13,der(13;13)(q10;q10)in 5 of 5metaphases that were examined.This cytogenetic finding would be ex-pected to result in a cfDNA test resultthat was positive for trisomy 13 and atrisomy 13 phenotype in the fetus. Twoindependent long-term cultures fromthe CVS showed a 46,XX chromosomeconstitution in 15 of 15 metaphases thatwere examined. An additional 50 meta-phases that were examined from theCVS cultures were euploid. Further re-view of the direct preparation slidesshowed the derivative (13;13) trans-location chromosome in all metaphasesthat were examined. Amniocentesis wasperformed at 15 weeks’ gestation. All 9primary colonies that were examinedhad a normal 46,XX chromosome con-stitution. An additional 41 metaphasefrom an amniotic cell culture flaskalso had a 46,XX chromosome con-stitution. Fetal anatomic evaluation byultrasound scanning at the time ofamniocentesis appeared normal. Thepregnancy is continuing at the time ofthis report, and ultrasound follow-up isplanned.

CommentAlthough many of the patients that wedescribed above have not yet deliveredtheir pregnancies, for the purposes ofthis discussionwe will refer to all them as

having received false-positive cfDNA testresults inasmuch as the results were notconfirmed by prenatal chromosomeanalysis, which has been the standardagainst which the testing has been vali-dated clinically. Given the quantitativenature of the interpretation of cfDNAtesting for aneuploidy detection and thelow prevalence of trisomy 18 and tri-somy 13, it should not be surprising toclinicians that they occasionally mayencounter false-positive test results,perhaps without ever having experi-enced a true positive test result in prac-tice. Without knowing the number oftests that were performed by the labo-ratories, it is not possible to estimatewhether the number of cases with posi-tive cfDNA test results in which theprenatal chromosome testing was notconcordant exceed the number of false-positive cfDNA test results that wereexpected, based on the publishedstudies. If false-positive test results, inlarge part, are due to the quantitativenature of the test interpretation, thenone would expect that careful analysismight demonstrate that a group of suchcases would have test results closer to thecut-off for normal than might beobserved for a group with true positivetest results. Because it has been shownthat the discrimination between normaland aneuploid pregnancies improveswith increasing “fetal fraction” ofcfDNA, it might also be relevant to knowthe relative “fetal fraction” in a group ofcases with false-positive test results. Aninformative analysis of false-positive testresults would require a much largercollection of cases with quantitative datathat is not generally available in thelaboratory reports that are issued to cli-nicians. A rate of false-positive test re-sults that is not higher than expected anda careful study that would demonstratethat these occurrences are most likely afunction of the quantitative nature of thetest interpretation would reaffirm theimportance of always offering invasivetesting when there is an abnormalcfDNA test result. The identification ofthe limitations of the positive predictivevalue as the likely cause of testing resultssuch as those observed in cases 1-6would also reduce concern about sample

MONTH 2013

errors or rare biologic mechanisms thatresult in false-positive test results. If, onthe other hand, the false-positive testresults do not appear to be based onthe quantitative test interpretation, thenother explanations should be sought.These could be important for the futureclinical application of the test, forcounseling, and for the assessment ofpatients with positive cfDNA results.

Cases 7 and 8 raise important ques-tions about other low probability bio-logicmechanisms for some false-positivecfDNA results. Inasmuch as the DNAthat was measured in this testing is ofplacental origin, mosaicism for aneu-ploidy in the placenta is a possibleexplanation. Mosaicism in the placentacan reflect mosaicism throughout thepregnancy, which includes the fetus, or itcan be limited to the placenta (known asconfined placental mosaicism [CPM]).Other, much less likely biologic causes offalse-positive test results might includethe contribution of a vanishing aneu-ploid twin to the cfDNA in the maternalplasma, a copy number duplicationof portion of a fetal chromosome toosmall to be detected by the standardcytogenetic testing, or a low percentagematernal mosaicism for an autosomalaneuploidy that may be either constitu-tional or acquired. Concern aboutmaternal mosaicism for sex chromo-some aneuploidy is likely to becomerelevant because gender and sex chro-mosome aneuploidy are reported in-creasingly by the laboratories that areperforming this testing.

Should clinicians and cytogeneticistsbe concerned that clinically relevant fetalor placental mosaicism may be the causeof what initially appears to be false-positive cfDNA testing results? If so,when the initial analysis of the confir-matory testing demonstrates all normalcells, it would seem prudent to increasethe number of colonies or cells to beexamined to increase the confidencethat there is not clinically relevantmosaicism. Guidelines for the “extensivework-up for mosaicism” generally areused when only a portion of cells in anamniotic fluid or CVS are aneuploid.12

These same guidelines might be usedwhen CVS or amniotic fluid samples are

American Journal of Obstetrics & Gynecology 3

Clinical Opinion Genetics www.AJOG.org

being tested in such cases, even thoughaneuploid metaphases were not ob-served in the initial standard evaluation.

Because CPMhas been associated withfetal growth restriction in some reports,serial ultrasound scanning to evaluatefetal growth might also be consideredif CPM is suspected.13-15 CPMmay occurbecause of missegregation of chro-mosomes in mitosis of a normal cell,which results in a lineage of daughtercells receiving 45 or 47 chromosomes.Importantly, CPMmay also be the resultof “trisomy rescue” in which a cell with47 chromosomes loses an extra chro-mosome during mitosis and a lineage ofcells with a normal number of chromo-somes results. Trisomy rescue poses anincreased risk of uniparental disomy,which is a condition in which bothmembers of a chromosome pair areinherited from one parent. Althoughuniparental disomy is an infrequentoccurrence, it may result in phenotypicproblems if there are imprinted genes onthe specific chromosome that is involvedor if there is a recessive mutation thatbecomes homozygous as a result ofthe uniparental disomy.16 Even thoughimprinted genes have not been identifiedon chromosomes 13 and 18, testing foruniparental disomy might be consideredselectively in cases that are associatedwith false-positive cfDNA testing. Thefinding of uniparental disomy in suchcases might be taken as indirect evidenceof “trisomy rescue” that is associatedwith mosaicism as the cause of theabnormal cfDNA result.

Invasive prenatal diagnostic testingmay also demonstrate “pseudomosai-cism,” which is most often considered anin vitro phenomenon in the laboratory,rather than a reflection of the true fetalor placental chromosome constitution.This perplexing problem occurs less fre-quently with amniocentesis than it doeswith CVS. For this reason the findingof mosaicism in CVS often leads to arecommendation for amniocentesis inan attempt to resolve the issue. Althoughlow percentage mosaicism cannot beexcluded with certainty, an adequateamniotic fluid study with no abnormalcells coupled with a normal ultra-sound scanning is generally somewhat

4 American Journal of Obstetrics & Gynecology M

reassuring regarding clinically significantmosaicism. On the other hand, confir-mation of mosaicism by amniocentesisis often considered to reflect the fetalstatus. Thus, if CPM is identified asa relatively important cause of false-positive cfDNA testing for certain aneu-ploidies, one might question whetheramniocentesis would be preferred overCVS for diagnostic confirmation in thesecircumstances. Such a suggestion wouldhave the undesirable effect of losing thebenefit of obtaining early test results forthese patients. Thus, we believe thatsuggesting that amniocentesis might be abetter confirmatory test should not beconsidered without scientific evidenceto support a clear association of CPMwith specific false-positive cfDNA testingresults.For several decades, screening for

aneuploidy with the use of maternalserum markers and fetal ultrasoundscanning has focused on increasingsensitivity while minimizing the numberof invasive diagnostic tests.17 Over thistime, the positive predictive value (ie, theproportion of positive test results thatare true positives) has improved sub-stantially but remains less than ideal.Obstetricians know this because theyspend considerable time helping patientsunderstand that a positive screening testdoes not necessarily mean an affectedfetus and that invasive testing by chro-mosome analysis and possibly micro-array is needed to make or to exclude adiagnosis of aneuploidy with a high de-gree of confidence. Why then might webe alarmed by the observation of false-positive cfDNA results? The answermay be obvious. The high sensitivityof cfDNA testing and the low false-positive rate has led some investigatorsto speculate about the possibility ofcfDNA replacing diagnostic testing inthe future.18,19 In a competitive envi-ronment, the laboratories have extendedthe testing to the rarer autosomal aneu-ploidies and recently to sex chromosomeaneuploidies. Scientific publications,marketing materials, and clinical labo-ratory reports emphasize the highsensitivity of the testing but do not statethe positive predictive value. It seemsthat, in the excitement about this

ONTH 2013

important advance, clinicians may alsohave lost sight of the fact that, as wetest for disorders with lower and lowerprevalence, the balance between truepositives and false positives shifts. Thus,a sense of alarm on learning about thesecases may simply be a result of expecta-tions that have been created for us.

We believe that it is unfortunate thatthe laboratories that offer cfDNA testinghave not requested that clinicians reportfalse-positive or false-negative resultsto them and that they do not providequantitative laboratory results that couldbe used in the retrospective analysis ofthese cases. Although the positive pre-dictive value of cfDNA based on data inthe published validation studies appearsbetter than the other currently availablemethods for aneuploidy screening, it isimperfect and will be influenced by theprevalence of the disease. A publishedstudy of routine clinical use in a largepopulation and a study of stored samplesin a general population that has aneu-ploidy screening in the first trimesterhave shown detection rates that aresimilar to those in high-risk pop-ulations.11,20 It seems likely that thesetests will be applied to lower risk pop-ulations in the future. If so, we shouldexpect that a larger proportion of posi-tive results will be false positives becauseof the low prevalence of these rareproblems in the general population. Forthis reason, we believe that data fromclinical practice regarding performancecharacteristics of this testing is neededurgently to avoid having the confidenceof physicians and patients in this newtesting undermined by additional anec-dotal experiences.

We discuss low probability biologicexplanations and raise hypotheticalquestions to underscore the need forcoordinated clinical and laboratory in-vestigation of these cases. The results ofsuch investigations will be important toinform the pretest and posttest coun-seling by clinicians who offer cfDNAtesting. Also, the development of clinicalpractice guidelines for the evaluationof cases of potential false-positive cfDNAtesting awaits evidence on which tobase recommendations. In the mean-time, clinicians and cytogeneticists will

www.AJOG.org Genetics Clinical Opinion

necessarily individualize evaluation ofthese patients.

We propose that the establishment ofa registry for the purpose of gatheringinformation about false-positive andfalse-negative cfDNA testing would be animportant step to address the questionsthat we raise. Only more systematicallygathered data, rather than anecdotalexperiences, will help us decide whetherto be alarmed by these experiences.Finally, rather than being alarmed bythese reports, clinicians should use thisexperience to remind themselves todiscuss the possibility of false-positivetest results during pretest counselingfor screening, which would includecfDNA testing, and as reinforcement tofollow the recommendation to offerdiagnostic tests to patients who have apositive screening test.5 -

ACKNOWLEDGMENTS

We acknowledge Robert Debbs, DO, KristinKinsler, DO, and Harish Sehdev, MD, forproviding information about their patients.

REFERENCES

1. Chiu RW, Lo YM. Clinical applications ofmaternal plasma fetal DNA analysis: translatingthe fruits of 15 years of research. Clin Chem LabMed 2013;51:197-2.2. Chiu RW, Chan KC, Gao Y, et al. Noninvasiveprenatal diagnosis of fetal chromosomal aneu-ploidy bymassively parallel genomic sequencing

of DNA in maternal plasma. Proc Natl Acad SciUSA 2008;105:20458-63.3. Fan HC, Blumenfeld YJ, Chitkara U,Hudgins L, Quake SR. Noninvasive diagnosis offetal aneuploidy by shotgun sequencing DNAfrom maternal blood. Proc Natl Acad Sci USA2008;105:16266-71.4. Zimmerman B, Hill M, Gemelos G, et al.Noninvasive prenatal aneuploidy testing ofchromosome 13, 18, 21, X, and Y, using tar-geted sequencing of polymorphic loci. PrenatDiagn 2012;32:1233-41.5. American College of Obstetricians andGynecologists. Noninvasive prenatal testing forfetal aneuploidy; ACOG committee opinion no.545. Washington, DC: The College; 2012.6. Palomaki GE, Deciu C, Kloza EM, et al. DNAsequencing of maternal plasma reliably identifiestrisomy 18 and trisomy 13 as well as Downsyndrome: an international collaborative study.Genet Med 2012;14:296-305.7. Ashoor G, Syngelaki A, Wagner M, Birdir C,Nicolaides KH. Chromosome-selective se-quencing of maternal plasma cell-free DNA forfirst-trimester detection of trisomy 21 and trisomy18. Am J Obstet Gynecol 2012;206:322.e1-5.8. Sparks AB, Struble CA, Wang ET, Song K,Oliphant A. Noninvasive prenatal detection andselective analysis of cell-free DNA obtained frommaternal blood: evaluation for trisomy 21 andtrisomy 18. Am J Obstet Gynecol 2012;206:319.e1-9.9. Bianchi DW, Platt LD, Goldberg JD,Abuhamad AZ, Sehnert AJ, Rava RP. Genome-wide fetal aneuploidy detection by maternalplasma DNA sequencing. Obstet Gynecol2012;119:890-901.10. Norton ME, Brar H, Weiss J, et al. Non-invasive chromosomal evaluation (NICE) study:results of a multicenter prospective cohort studyfor detection of fetal trisomy 21 and trisomy 18.Am J Obstet Gynecol 2012;207:137.e1-8.

MONTH 2013

11. Dan S, Wang W, Ren J, et al. Clinicalapplication of massively parallel sequencing-based prenatal noninvasive fetal trisomytest for trisomies 21 and 18 in 11 105 preg-nancies with mixed risk factors. Prenat Diagn2012;9:1-8.12. Hsu LY, Benn PA. Revised guidelines for thediagnosis of mosaicism in amniocytes. PrenatDiagn 1999;19:1081-2.13. Lestou VS, Kalousek DK. Confinedplacental mosaicism and intrauterine fetalgrowth. Arch Dis Child Fetal Neonatal Ed1998;79:223-6.14. Wolstenholme J, Rooney DE, Davison EV.Confined placental mosaicism, IUGR, andadverse pregnancy outcome: a controlledretrospective U.K. collaborative survey. PrenatDiagn 1994;14:345-61.15. Stipoljev F, Latin V, Kos M, Miskovic B,Kurjak A. Correlation of confined placentalmosaicism with fetal intrauterine growth retar-dation: a case control study of placentas atdelivery. Fetal Diagn Ther 2001;16:4-9.16. Kotzot D. Prenatal testing for uniparentaldisomy: indications and clinical relevance.Ultrasound Obstet Gynecol 2008;31:100-5.17. American College of Obstetricians andGynecologists. First-trimester screening for fetalaneuploidy; ACOG committee opinion no. 296.Washington, DC: The College; 2004.18. Simpson JL. Is cell-free fetal DNA frommaternal blood finally ready for prime time?Obstet Gynecol 2012;119:883-5.19. Wright C, Quake SR, Bianchi D, Wald NJ.Community Corner: Opening the Pandora’s boxof prenatal genetic testing. Nat Med 2011;17:250-1.20. Nicolaides KH, Syngelaki A, Ashoor G,Birdir C, Touzet G. Noninvasive prenatal testingfor fetal trisomies in a routinely screened first-trimester population. Am J Obstet Gynecol2012;207:374.e1-6.

American Journal of Obstetrics & Gynecology 5