Quantitative In­vivo Imaging of the Impact of Cancer Therapy on the Normal Pediatric Brain

Wilburn E. Reddick, Ph.D.

Diagnostic Image and Signal Processing Laboratory Division of Translational Imaging Research

Department of Radiological Sciences

Wilburn E. Reddick, Ph.D.

Diagnostic Image and Signal Processing Laboratory Division of Translational Imaging Research

Department of Radiological Sciences

The Clinical Problem

AcAcuuttee lylymmphoblastphoblasticic leuleukkememiaia (ALL)(ALL) isis tthehe mmosostt ccomommmonon cchildhood childhood cancancerer –– AAffffeecctintingg 22,4,40000 chchililddrerenn aannnnuuaallllyy inin tthhee USUS

–– YYoouunngg aaggee aatt ddiagiagnnoosisis as anndd hhigighh susurrvviivvaal rl raatete

BrainBrain TTumumorsors areare tthehe mmosostt ccomommmonon ssolidolid ttumumorsors ooff cchildhoodhildhood –– AAffffeecctintingg 33,1,11100 chchililddrerenn aannnnuuaallllyy inin tthhee USUS

–– MMoostst cocommmonmon ccaauusese ooff cacanncceerr relarelatetedd ddeeaathth inin chchililddrreenn

–– HiHigghh ratratee ooff sesevveerere mmoorrbbidityidity RJ Gilbertson Source: CBTRUS and SEER

1930

1950

1970

1990

100-ALL

25-

0-

75- CNS

50-

% survival

Increasing Importance of Neurotoxicity

RJ Gilbertson Source: CBTRUS and SEER

Independent Research Program Probing Substrates of Neurotoxicity

Basic Research Focus: Development of innovative algorithms and methods to quantify the structure and integrity of cerebral white matter in vivo

Clinical Research Focus: Use non­invasive imaging technology to quantify neurostructural changes resulting from radiological or pharmacological insult

Ultimate Goal: To assist in the development of therapy that would prevent, mediate, or intervene to minimize impact of neurotoxicity in survivors of pediatric cancer

Translational Imaging Research

Basic Research Image Registration and Fusion RF Correction Segmentation Volume of Interest Analyses

Diffusion and Perfusion

Clinical Research (BT) Historical Background

Most Recent Results

Ongoing Studies

Clinical Research (ALL) Most Recent Results

Ongoing Studies

3D Affine Registration

Within an examination

Between examinations

2000

Num

ber of

Pixels 1600

1200

800

400

0

white gray

3000 3500 4000 4500 5000 5500 6000

Dose (cGy)

Fusion of RT Dose with Segmented MR

TE Merchant

Bias Field Correction (in plane)

(Ji et al. MRM [in prep], 2005)

PD In

tens

ity (a

.u.)

1400

1300 Pre

Post 1200

1100

1000 0 5 10 15 20

Section Number

Bias Field Correction (between planes)

Kohonen Self­Organizing Map (Segmentation)

Output

Input

Learning Algorithm

, ( 2

j − weight , Δ weight i j = (neigh iter )) [input i j ] 2 2⎡− (x + y )⎤

neigh iter ) = η ∗exp 2 ⎥( ⎢ 2 ∗σ⎢ ⎥⎣ ⎦

iter

maxη = .0 005 iter

T1 T2 PD FLAIR ⎛ iter ⎞

maxSignal Intensity σ = 3 ⎜⎜0 4 . iter ⎟⎟⎝ ⎠

(Reddick et al. IEEE­TMI, 1997)

SOM of Normal Examination

T1 T2 PD FLAIR

Intra­class correlations for N =14 White matter ri = 0.91 (p < 0.01) SOM Gray matter ri = 0.95 (p < 0.01) CSF ri = 0.98 (p < 0.01)

(Reddick et al. MRM, 2002)

SOM of Abnormal ExaminationT1 T2 PD FLAIR

SOM

(Reddick et al. MRM, 2002)

Additional Refinements

FLAIR SOM­02 SOM­03 SOM­04

Kappa measure of agreement (N = 15) Obs 1 0.651 0.653 0.744 Obs 2 0.602 0.615 0.699

(Glass et al. MRM, 2004)

60 Gray

400 80

350 70

Gray 300

White Volum

e (m

L) 250

200

150 White

CSF

50

40

Volum

e (m

L)

CSF 100

50

0 0 0 1 2 3 4 5 6 0 1 2 3 4 5 6

Time Since RT (years) Time Since RT (years)

10

20

30

Index vs Expanded Sampling

Expand Coverage

3D Visualization

ZY Shan (Mulhern et al. Lancet Onc, 2004)

Prefrontal

Frontal Parietal / Mid­Temporal Parietal / Occipital Right Left

Regional Analysis

(Mulhern et al. JINS, 2004)

Quantifying White Matter Integrity

RJ Ogg FLAIR SOM ADC FA

Quantifying White Matter Perfusion

RJ Ogg T2 CBV

Translational Imaging Research

Basic Research Image Registration and Fusion RF Correction Segmentation Volume of Interest Analyses

Diffusion and Perfusion

Clinical Research (BT) Historical Background

Most Recent Results

Ongoing Studies

Clinical Research (ALL) Most Recent Results

Ongoing Studies

Why Normal­Appearing White Matter?

Two age­matched groups treated for brain tumors of the Posterior Fossa

Variable MB (N=18) LGA (N=18) Significance FSIQ 82.0 + 10.9 92.9 + 15.7 P=0.026

ICV 82.5 + 5.4 85.2 + 6.0 NS White 21.4 + 4.4 24.7 + 5.7 P=0.008 Gray 52.6 + 5.1 54.3 + 6.1 NS CSF 8.1 + 4.0 6.1 + 4.5 NS

RK Mulhern (Reddick et al. MRI, 1998)

Inform

ation Raw

Score

Normal Population 25

20

15

10

5

0 0 1 2 3 4 5 6 7 8 9 10 11 12

Time Since CSI (Years)

(Expected)

MB patients (obtained) N=52

A New Understanding of Decreasing IQ

(Palmer et al. JCO, 2001)

Linking Therapy & Neurocognitive Deficits

Cross sectional study of Medulloblastoma survivors (N=42)

Age at irradiation significantly associated with FSIQ (R 2 = 0.170; P = 0.006; controlled for time since irradiation)

Mediational model: ~70% of association explained by Normal Appearing White Matter

(Mulhern et al. JCO, 2001)

Developmental Model

RT

Math

NAWM IQAttention

Age @ RT

Time Since RT

Reading

Spelling

P < .001

P < .001

P < .001

P < .001

N = 40 Median 5.7 yrs post RT

Moderators

Model explains: ~ 60% of variance in reading ~ 60% of variance in spelling ~ 80% of variance in math

(Reddick et al. Cancer, 2003)

Most Recent Results

Longitudinal study of 324 MR exams from 52 subjects treated for Medulloblastoma

All received 36 Gy CSI 19 had shunts placed Median age @ irradiation 8.3 yrs (3.4 to 20.0 yrs) Median time since irradiation 2.5 yrs (­0.2 to 7.9 yrs)

Cross­sectional study of a subset of 19 patients age similar to controls and without shunts

Single most recent MR Age at examination 13.0 ± 3.1 yrs

26 healthy sibling controls imaged once Age at examination 12.6 ± 3.4 yrs

A Gajjar

Volum

e (cc)

Younger at RT Older at RT

30 30 WM no shunt

25 25

20 WM shunt 20

WM no shunt

WM shunt

CSF shunt

CSF no shunt

6 7 8 9 10 11

Patient Age (years) Volum

e (cc)

CSF shunt 15 15

10

5

0

10 CSF no shunt

5

0

12 13 14 15 16 17

Patient Age (years)

Longitudinal Brain Volume Development

(Reddick et al. Neuro Onc, 2005)

­

­

­

­

5

0

5

10

NAWM

Volum

e (m

L)

Younger at RT Older at RT 10

0 HC

HC

MB

NAWM

Volum

e (m

L)

5

MB

5

10 10 6 7 8 9 10 11 12 12 13 14 15 16 17 18

Patient Age (years) Patient Age (years)

Longitudinal Brain Volume Development

(Reddick et al. Neuro Onc, 2005)

NAWM

(ml)

40 HC

30

20 MB

10 p<0.001

0 5 10 15 20

Age @ MR (years)

Longitudinal Brain Volume Development

(Reddick et al. Neuro Onc, 2005)

Translational Imaging Research

Basic Research Image Registration and Fusion RF Correction Segmentation Volume of Interest Analyses

Diffusion and Perfusion

Clinical Research (BT) Historical Background

Most Recent Results

Ongoing Studies

Clinical Research (ALL) Most Recent Results

Ongoing Studies

Most Recent Results

Longitudinal study of 164 MR exams from 45 subjects treated for ALL on Total 14

Low Risk Standard / High Risk Number of Subjects Post 1 IV­MTX 21 23 Post 4 IV­MTX 20 21 Post 7 IV­MTX 21 21 End of Therapy 20 17

Gender Male 10 11 Female 12 12

Age at Diagnosis (years) 5.0 ± 2.7 9.2 ± 4.8

C­H Pui

0

20

40

60

80

100

Obs

erved Prevalen

ce of L

E

100 p = 0.011 p = 0.011 87 86 p < 0.001 p = 0.011

80

5 20 45 132

Week on Protocol

67

Predicted

proba

bility of

LE

5 20 45 132

Week on Protocol

p < 0.001 67 60 60 59 57 55 60

47 40

40

17 17 15 14 20

0

41

Prevalence of LE Standard / High­Risk

Low­Risk

(Reddick et al. AJNR [in press], 2005)

Transient vs. Persistent

Week 5 20 45 132 240 KJ Helton

Propo

rtion WM

affe

cted

35% p = 0.002 p = 0.046 p = 0.010

30%

25%

p = 0.099 p = 0.019 20%

Standard / High­Risk 15% Low­Risk

10%

5%

0%

5 20 45 132

Week on Protocol

Extent of LE

(Reddick et al. AJNR [in review], 2005)

= =

= =

50

175 50 p 0.090 p 0.007

150

40p = 0.063 p = 0.065 125 p 0.007 p 0.083

5 20 45 132

Week on Protocol

T2 increase (m

s)

5 20 45 132

Week on Protocol

100

75

30

20

10 25

0 0

T1 increase

(ms)

Intensity of LE Standard / High­Risk

Low­Risk

(Reddick et al. AJNR [in review], 2005)

300

T1 increase

(ms)

250

200

150

100

50

0 10 20 30 40 50

T2 increase (ms)

Relationship Between Intensity Measures

(Reddick et al. AJNR [in review], 2005)

Translational Imaging Research Summary

BasBasiicc RResesearearcchh DevDeveellopopeded esesssententiiaall nnovovelel iimmageage procprocesesssiingng ccapaapabbiilliittiieses whiwhicchh were owere optiptimmiizzeded ffor sor speciapecialliizzeded cclliininiccaall resresearcearch ah apppplliiccatatiioonnss

ContContiinnueue to dto devevelelopop iinnnonovvaatitivve ae allgorgoriithmthmss anandd mmethethods to qods to quanuantitiffyy ththe stre strucucture anture andd iintntegriegrittyy ofof ccerebralerebral whiwhitte me matteatterr iin vivn vivoo

CCliniclinicalal RResesearcearchh UsUseded nonon­inn­invvasasiivvee iimmagiagingng technotechnollogogyy toto quaquantintiffyy neuneurrosostructructuraltural cchhangangeses resresulultitingng ffrroomm radiradiolologogiiccaall oror phapharrmmacacolologogiiccalal iinsnsuullt at andnd rrelelateated thd thesese che changangeses toto neuneurrococogognniititivve pe perferforormmancancee

Translational Imaging Research Summary

Building on extensive experience with MB New studies designed to combine radiation dosimetry maps with MR imaging measures of perfusion and diffusion

Investigate the integrity of white matter microvasculature and axonal myelin

Changes in these measures is hypothesized to precede more global changes in cerebral white matter volume

120 subjects with 1560 MR exams

Translational Imaging Research Summary

Building on preliminary experience with ALL Ongoing ALL study designed to test hypotheses that early changes in MR imaging measures are:

predictive of later white matter changes proportionate to exposure to HDMTX related to CSF and plasma homocysteine predictive of treatment­induced neurocognitivedeficits and diminished quality of life insurvivors

300 subjects with 1200 MR examinations

Acknowledgements

DDepartepartmmentent ofof RRadiologicadiologicalal SScciencienceess •• DDr.r. RRobertobert OggOgg DiDivv TTrrans Ians Immagag ResRes

•• DDr.r. ZZuyuyaoao ShanShan DiDivv TTrrans Ians Immagag ResRes

•• DDr.r. KaKatthleen Heltonhleen Helton DiDivv DDiiaag Img Imagiagingng

•• DDr.r. TThomhomasas MMercerchhantant DiDivv RRadad OOncnc

IInsnsttiittututionalional CCollaboratollaboratorsors •• DDr.r. AmAmarar GajjarGajjar DeptDept ofof HemHem/O/Oncnc

•• DDr.r. CChing­Hhing­Honon PuiPui DeptDept ofof HemHem/O/Oncnc

•• DDr.r. RRayaymmonondd MMulhernulhern DiDivv BBehehavav MMeded

Acknowledgements

•• JJohnohn GlaGlassss •• RRhonda Simmhonda Simmonsons •• GregGreg BernBernsstteinein •• KimKimberlberlyy JJohnsohnsonon •• BrianBrian TTayaylorlor •• JJohnohn SSttagiagicchh •• DDr.r. QingQing JJii •• JJinesineshh JJainain •• TTraravvisis MMilleriller

SSiig Pg Prrococ SSupuperverviissoror

SSiig Pg Prrococ RReses AAssssiisstantantt

SSiig Pg Prrococ RReses AAssssiisstantantt

SSiig Pg Prrococ RReses AAssssiisstantantt

PPOOEE SStudtudentent

PPrrojojecect Let Leaaderder

SSofoftware Etware Engngiineneerer

SSofoftware Etware Engngiineneerer

SSyysstemtemss AAdmdmiininisstratortrator