Molecular Imaging in Early Phase Oncology Trials Lessons Learned Potential Role for SNM Network Susan Galbraith MB BChir PhD Vice-President Oncology Discovery Medicine & Clinical Biomarkers Bristol-Myers Squibb 1
Use of imaging in oncology drug development Pre-clinical/Phase 1 Tissue Distribution/ Pharmacokinetics Is there a pharmcodynamic effect in tumor tissue? Dose response / time course of response Sequence/schedule selection Phase I/II Can FDG-PET/ FLT-PET/Apoptosis PET imaging provide earlier/more robust evidence of anti- tumor efficacy than measurement of tumor size? Phase II/III Can tumor response be predicted by target expression/ activation? 2
Considerations for use of Imaging in Early Phase Trials For decision driving data (rather than just pretty pictures) generally cohorts >3 needed for imaging studies (n depends on repeatability & effect size of interest) Select appropriate imaging modality based on understanding of biology Ensure consistent implementation of imaging protocol at all sites Centralized data analysis 3
Implications Technology used - relatively established vs cutting edge Definition of every stage of imaging process Repeatability studies needed before measurement of treatment effect SDV as detailed as for clinical aspects of study Site selection Use of consensus guidelines for methodology 4
Current Status Repeatability (FDG-PET) Single expert center literature data n=16 SUV (Standard Uptake Value), wcv = 9% (J Nucl Med 1999 Weber et al) Compares well with repeatability of CT volume measurements, but mainly thoracic malignancies -?how representative of most centers/ tumor types BMS experience multicenter 8 sites (CA182003 trial) n=62, metastatic colorectal cancer 2 baseline scans within 1 week SUVmax, wcv 16%, 95% repeatability coefficient (-34%, 52%) 5
CA182-003 Phase I study of brivanib and cetuximab in advanced metastatic gastrointestinal malignancies Screen brivanib x 1 Washout brivanib (daily) + cetuximab (weekly) -14 C1D1 C1D8 C1D15 C2D1 24h-PK PD/markers 24h-PK PD/markers PK (predose) PD/markers PK(predose) PD/markers FDG-PET performed twice in one week at screening, C1D15-22 and C2D28 6
Repeatability Results All patients fasted for at least 4 hrs prior to scan Plasma Glucose Mean 103 mg/dl (range 50-154 ng/dl) 8 diabetic subjects 1 had plasma glucose of 209 ng/dl data excluded 2 other subjects without glucose data excluded QA failures (1 center performed central read) 3 subjects - missing/elevated glucose 1 subject - tracer extravasation 3 subjects - no scans submitted 2 subjects - unreadable CDs 6 subjects - DICOM inconsistencies 7
Effect of Correction for Lean Body Mass and Plasma Glucose on Repeatability 8
Effect of Compliance with Timing of Scan Scan Time Relative to FDG Dose 50-70 min <50 or >70 min Between Scan Difference in Timing 15 min > 15 min N 25 21 34 12 % Difference in Basleine SUVmax Mean (SD) 10.6 (7.3) 16.1 (18.2) 11.9 (9.5) 16.6 (21.4) 9
Percent Difference in Baseline Scans - Variation by Site 10
Bland Altman Plot - Difference vs Mean for SUVmax 11
Repeatability Results Parameter SUVmax - All Mean % difference 1.4 (-3.9,( 7.0) wcv 16.4% 95% Repeatability Coefficient -34.3, 52.3% SUVmax - QA 1.8 (-3.3,( 7.2) 13.5% -29.5, 41.9% SUVmean 0 (-4.6,( 4.8) 12.3% -27.4, 37.8% SUVpeak 1.1 (-3.7,( 6.1) 12.7% -28.1, 39.1% 12
Conclusions for FDG-PET repeatability Other SUV parameters do not have significantly better repeatability than SUVmax Correction for lean body mass not required body weight or BSA is sufficient Plasma glucose correction not required Compliance with recommendations on interval between injection and scan varied between sites No statistically significant effect on repeatability but trend to worse repeatability at less compliant sites We should reconsider definitions for Metabolic Response 25% decrease is insufficient Technology is robust enough for multi-center trial environment 13
CA182003 Brivanib and Erbitux Site 0013 Subject number 00005 Baseline Scan 3 Scan 4 FDG-PET 14
CA182003 Anatomic Response (WHO) 15
CA182003 Anatomic / Metabolic Response Metabolic responses measured by FDG-PET on Day 56 in evaluable patients with CRC receiving brivanib 800 mg Best Anatomical Response % Change from Baseline (SUVmax) Patient *median PFS 269 d (177+ - 449+) ** Similar results were seen at Day 15 16
How might SNM Network help in implementing imaging in drug development? Allow cross reference to common IND for novel tracers Improve routine QA at sites phantom program Improve awareness about importance of compliance with imaging procedures Remove need for site qualification for every study?? Build dataset of lessons learned Update consensus guidelines/imaging procedure manuals based on experience 17
Acknowledgements Linda Velasquez BMS Clinical Scientist CA182003 Wendy Hayes, Patrick Chow BMS Imaging Group Georgia Kollia BMS statistician Tony Shields, Otto Hoekstra (FDG-PET experts) Investigators on CA182003 Derek Jonker, Lilian Siu,, Anthony al Khoueiry,, John Marshall, Chris Garrett, Pat Lorusso,, J Buter,, Pierre Major Patients who volunteered 18