Molecular Imaging using Positron Emission Tomography:

Assessment of (neuro-)receptor changes with PET

Ronald Boellaardr.boellaard@vumc.nl

Even voorstellen (mini CV)• Ronald Boellaard• Huidige functie: klinisch fysicus en UHD bij de afdeling

Nucleaire Geneeskunde, VUmc, A’dam

• Vooropleiding:- VWO (Gym-β), 1987- Exp.Natuurkunde (en Biologie), 1994- AIO/promovendus op het NKI (afdeling RT) , 1998- opleiding klin.fys. Op VUmc, 2001- klin.fys./UHD op VUmc – tot heden

• Klinische of Medische Fysica = toegepaste fysica

Presentation• General introduction NM and PET• Physics and principles of PET

- general introduction- overview of (neuro-receptor) tracers- positron emission and coincidence detection

• PET pharmacokinetic analysis- principles of kinetic modelling- generation of parametric images

• SPM example of assessment of (neuro-) receptor change

EmissionTomography

Physiology Imaging

Biochemistry Quantification

Pharmacokinetics Flexibility

NM & Positron Emission Tomography

The spectrum of medical imaging Jones, 1996

Structure/anatomy X-ray/CT/MRI

Physiology US, SPECT, PET, MRI/S

Metabolism PET, MRS

Drug distribution PET

Molecular pathways PET

Molecular targets PET, SPECT

Clinical Applications Clinical Applications • Oncology

• Cardiology

• Neurology / Psychiatry

• Pneumology

• Nephrology

......

• Oncology

• Cardiology

• Neurology / Psychiatry

• Pneumology

• Nephrology

......

Very basic principle of nuclear medicine and PET

• Inject radiopharmaceutical (single photon or positron emitter labelled to a drug)

• Use gamma or PET camera to:- evaluate distribution of radiopharmaceutical at some time after injection

- evaluatie uptake, retention and washout of radiopharmaceutical = dynamic or kinetic information

I. Qualitative analysis of PET studies“qualitative/visual inspection”

Examples of FDG whole body scans

Purpose: staging, unknown primary

II. Semi-quantitative analysis of PET studies“standard uptake values (SUV)”

SUV is the uptake of a radiopharmaceutical, normalised to the injected doseand body weight (or lean body mass or body surface area etc)

regions of interest analysis: Average uptake (Bq/cc) in e.g. tumor

Purpose: diagnosis (benign/malignant), prognosis, response monitoring, definition of RT treatment volumes,…

CTI / Siemens HR+ PET scanner RDS 111 15O-cyclotron

Department of Nuclear Medicine and PET Research

location ‘hospital’

Department of Nuclear Medicine and PET Research

location ‘Radionuclide Centre’

HRRT PET scannerGMP lab with 6 hot cells

The High Resolution Research Tomograph (HRRT) PET scanner

HRRTCPS Research

• 8 panel detector heads

• 60.000 LSO crystals

• 1 crystal = 2.1 x 2.1 x 7.5 mm

• 1 billion lines of response

• Cs-137 singles transmission

• 3D only, no septa

• Only 10 scanners in the world (up to now 4 operational)

[11C]-Verapamil for imaging Pgp (Blood Brain Barrier Research)

mdr1a(-/-)/1b(-/-) KO mouse

mdr1a+/+/1b(+/+)WT mouse

Figure A: HR+, 7 mm resolution

Figure B: HRRT, 2.5 mm resolution

Figure A: HR+, 7 mm resolution

Figure B: HRRT, 2.5 mm resolution

0.0

5000.0

10000.0

15000.0

20000.0

25000.0

30000.0

[Bq/

cc]

0.0

5000.0

10000.0

15000.0

20000.0

25000.0

30000.0

[Bq/

cc]

Figure A: HR+, 7 mm resolution

Figure B: HRRT, 2.5 mm resolution

HRRT upcoming protocols: Clinical Comparison with HR+:

A STUDY IN NORMAL SUBJECTS USING THE TRACERS [11C]RACLOPRIDE, [11C]FLUMAZENIL AND [18F]FP-b-CIT.

HRRT upcoming protocols: Clinical Comparison with HR+:

A STUDY IN NORMAL SUBJECTS USING THE TRACERS [11C]RACLOPRIDE, [11C]FLUMAZENIL AND [18F]FP-b-CIT.

Isotope production

Nuclear reactions t1/2

18F (p,n) 110 min

11C (p,a) 20 min

13N (p,a) 10 min

15O (p,n) 2 min

GMP- LAB

Current Tracers [11C][11C]Flumazenil

central type benzodiazepine receptor

(R)-[11C]PK11195 activated microglia

[11C]Raclopride D2/D3

(R) -[11C]Verapamil PgP in BBB

[11C]R116301 NK1 receptor

[11C] PIB amyloid

A B

Current Tracers [18F]

[18F]FP-CITdopamine transporter

[18F]MPPF5HT1a receptor

[18F]FDDNPamyloid

[18F]FLTproliferation

[18F]Prolineaminoacid

[18F]FDG glucose metabolism

Current Tracers[15O]

[15O]H2Operfusion

[15O]O2 oxygen consumption

[15O]CO blood volume

OXYGEN EXTRACTION FRACTION

CBF CMRO OEFCBF CMRO OEF

Presentation• General introduction NM and PET• Physics and principles of PET

- principles- overview of (neuro-receptor) tracers- positron emission and coincidence detection

• PET pharmacokinetic analysis- principles of kinetic modelling- generation of parametric images

• SPM example of assessment of (neuro-) receptor change

Positron emission

511 keV fotonen

positron annihilates with electron

Annihilation produces 2 photons of 511 keV which are sent out in opposite directions

Positron emission detection

Positron emission tomography is based on the simultaneous (coincidence) detection of both annihilation photons

PET

radio-nuclide: positron emitter -> 2 photons

acquisition: coincidence-detection

coincidence processor

PET image reconstruction

ProjectionsProjections

ImageImage

ReconstructionReconstruction

PET scanner acquires projection

reconstruction of activity distribution in patient

PET Image reconstruction

FilteredBackprojection

IterativeReconstruction

Results patients (2)Example images, early frame, poor statistics, ‘fully converged’

FBP NAW-OSEM WLS-nn SP-OS-(W)LS

Presentation• General introduction NM and PET• Physics and principles of PET

- principles- overview of (neuro-receptor) tracers- positron emission and coincidence detection

• PET pharmacokinetic analysis- principles of kinetic modelling- generation of parametric images

• SPM example of assessment of (neuro-) receptor change

Tracer Kinetic Modelling

Tracer Model:

Purpose:

Method:

Mathematical description of thefate of the tracer in the humanbody, in particular in the organunder study

To quantify functional entitiesgiven the distribution ofRadioactivity (over time)

Divide possible distribution oftracer in a limited number ofdiscrete compartments

Brain

FDG uptake as function of time

T=0

T=60min

pharmacokinetic modelling

Dynamic scanParametric image representingbinding of tracer in tissue

Purpose: generation of image representing distribution of PET pharmacokinetic parameter: glucose consumption, DNA synthesis, perfusion etc etc.

Uptake, retention and washout of radiopharmaceutical

• Used radiopharm. (=tracer)

• Supply of tracer in arterial blood (= input function)

• “Physiology” of tumor/organ, which can be quantified using a PET-pharmacokinetic model

Shape and amplitude of time activity curve depends on:

Dynamic PET scanspharmacokinetic analysis

• dynamic scans consist of 20 to 40 sequential acquisitions during a 60 min period

• dynamic scans provide info on the variation of the activity(=pharmaceutical) in an organ/tumor as function of time

• dyn. scans are made to study and quantify the “functional or physiological” behaviour of the organ of interest (glucose and oxygen consumption, blood flow, blood volume, neuroreceptor density)

PET scan

Bolus injector

Bolustoediening bij dyn. (Ex) scans

Loodpot met activiteit

Veneuze inspuiting

Bloodsampler

detectorpompwaste

Analysis of dynamic PET scansInput function

0

50

100

150

200

250

0 2 4 6 8 10

Tijd (min)

Blo

od

ac

tiv

ity

(k

Bq

/cc

) 1022 keV

511 keV

manual samples

Input function also needs to be corrected for metabolites and plasma/blood ratio’s

Blood FreeBound

(or metabolizedor trapped)

Example of Two Tissue Compartment Model

Tissue

PET

Analyse van dynamische PET scanskinetische analyse

0.0E+00

1.0E+04

2.0E+04

3.0E+04

4.0E+04

0 20 40 60 80Tijd (min)

Ac

t.co

nc.

(Bq

/cc

)

0

50

100

150

200

250

0 2 4 6 8 10

Tijd (min)

Blo

od

ac

tiv

ity

(k

Bq

/cc) 1022 keV

511 keV

manual samples

K1

k2

k3

k4Cf CbCa

Quantitative value of apharmacokineticparameter, such as:-glucose comsumption-Perfusion-DNA synthesis-Hypoxia

Overview of ‘common’ pharmacokinetic models

Plasma input models

• Single tissue compartment model (1TC-R)

• Single tissue compartment model (1TC-Ir)

• Irreversible two tissue compartment model (2TC-Ir)

• Reversible two tissue compartment model (2TC-R)

Reference tissue input models

• Simplified reference tissue model

• Full reference tissue model

Reversible single tissue compartment model with plasma input

Blood

Tissue

PET

K1

k2

K1=E x F, E=extraction and F=flow=perfusionVd= K1/k2 = volume of distribution

Irreversible single tissue compartment model with plasma input

Blood

Tissue

PET

K1

K1=E x F, E=extraction and F=flow=perfusion

Irreversible two tissue compartment model with plasma input

k3Blood

Tissue

PET

K1

k2

K1=E x F, E=extraction and F=flow=perfusionKi= K1 x k3/(k2+k3)

FreeBound/

metabolized/trapped

Reversible two tissue compartment model with plasma input

Blood

Tissue

PET

K1

k2

K1=E x F, E=extraction and F=flow=perfusionBP=k3/k4 (sum of specific and ‘slow’ non-specific bindingVd= K1/k2 x (1+BP)

Free Bound

k3

k4

Reference tissue models

A reference tissue time activity curve (TAC) is used as input in stead of plasma input

R1=K1/k2=K1’/k2’=relative flow distributionBP=k3/k4=‘specific’ binding

Presentation

• Physics and principles of PET- general introduction- overview of (neuro-receptor) tracers- positron emission and coincidence detection

• PET pharmacokinetic analysis- principles of kinetic modelling- generation of parametric images

• SPM example of assessment of (neuro-) receptor change

Parametric pharmacokinetic modelling

Dynamic scanParametric image representingbinding of tracer in tissue

Purpose: generation of image representing distribution of PET pharmacokinetic parameter: glucose consumption, DNA synthesis, perfusion etc etc.

PET pharmacokinetic parametric methods

• Parametric=pixelwise=voxelwise, i.e. calculation/modeling is performed per pixel/voxel

• A 3D PET image (volume) consists of ~106 voxels

• Ergo, parametric methods need to be fast

• Most parametric methods use ‘tricks’ to gain computational speed (linearisation,basis function method, (multi-) linear plots)

• Parametric methods are fast calculations performed for each voxel (independently).

Parametric kinetic modelling(1) basis function and linear methods

Blood flow model example

Cb, CpK1

k2

Ct

CtkCpKdt

dCt21

2 solutions for differential equation:- convolution:

- linearization:

Theory

CtkCbKdt

dCt21

tkpt eCKC 2

1 tVF

pdeCF )/(

tpt CkCkC 21

Theory(Basis function method)

bbtVF

bbROI CVeCFVC d )/()1(

tVFb

deC )/(

Theory(Basis function method)

1. Determine F and Va for each basis function using linear least

squares fitting (GLM)

2. Calculate sum of weighted squared difference (Xsqr) for each basis function, F and Va

3. Minimum amongst Xsqr provides ‘best fit’ for F, Va and basis function (=F/Vd)

aatVF

aaROI CVeCFVC d )/()1(

Theory(linearization, linear least squares)

soeeeeROI

ROI

V

k

k

tCbtCttCp

tCbtCttCp

tC

tC

2

11111

)()()(

.....................

)()()(

)(

.......

)(

Y = X (+ ) =X-1Y in theory, but not possible due to noise

LS solution (GLM):=[XTX]-1XTY

Results – Clinical evaluationExamples of parametric CBF images –

various method

BFM GLLS LLS

Examples of parametric images

B C D

E F G

A

A=LoganC=Ichise 1D=Ichise 2E=Ref.LoganF=RPM1G=RPM2

Each voxel value represents the value for a pharmacokinetic parameter (Vd or BP)

Presentation• General introduction NM and PET• Physics and principles of PET

- principles- overview of (neuro-receptor) tracers- positron emission and coincidence detection

• PET pharmacokinetic analysis- principles of kinetic modelling- generation of parametric images

• SPM example of assessment of (neuro-) receptor change

Example of use of parametric PET data for SPM analysis

PET studiesDynamic [11C](R)-PK11195 PET studies of 10 young and 10 elderly healthy control subjects.

Scans were acquired in 3D mode using an HR+ scanner (Siemens). A neuro-insert was used for additional shielding for outside field of view activity.

Kinetic modellingParametric binding potential (BP) images were generated using Ichise linearisation of the simplified reference tissue models using a cerebellum time activity curve as reference tissue input.

Example of PK11195 BP image

Difference between anatomical VOI and SPM

Yellow = Thalamus (& pulvinar) VOI defined on MRIRed = SPM VOI

Effect of VOI method on observed changes in PK11195 binding

-0.05

0.05

0.15

0.25

ANA (A1) PVE (A2) BP>0 (D1) BP-Man (D2) SPM* p>0.01(D3)

VOI method

BP

Young

Old

SPM based on parametric PET data

• SPM might be used to more precisely locate areas of interest and to avoid that VOI are “contaminated” with regions without change.

• Data driven VOI provide a higher sensitivity for assessing (changes in) receptor binding.

• A drawback of data driven VOI, however, is that they depend on the data being used. Both sample size and statistical quality will affect size and shape of the VOI.

• Consequently, data driven VOI strategies may be less reproducible across studies and subjects than anatomically based VOI.