Assessment of Heterogeneous Treatment Response in Patients with Multiple Solid Tumors
P Scully1*, M Vanderhoek2, S Perlman1, G Liu1, R Jeraj1, (1) University of Wisconsin, Madison, WI, (2) Henry Ford Hospital, Detroit, MIWE-C-BRA-7 Wednesday 10:30:00 AM - 12:30:00 PM Room: Ballroom A
Purpose: Imaging biomarkers of treatment response have been defined for individual tumors, but when multiple tumors are present the evaluation is not straightforward. This study investigates the intra-patient heterogeneity of tumor response in patients with multiple metastatic lesions.
Methods: Twelve patients with two or more metastatic solid tumors of various histology were treated with molecular targeted therapy. Each patient received [18F]FLT PET/CT scans pre-, mid-, and post-treatment, and tumors were manually segmented on each scan. SUVmean, SUVmax, SUVpeak, SUVtotal, and PET-defined volume were calculated for each tumor. Response was defined as the percent change relative to baseline for a single tumor, and heterogeneity was quantified as the range of response across all tumors within a single patient. This was calculated for both mid/pre and post/pre response, yielding 120 total measurements of response heterogeneity.
Results: For all response measures, substantial intra-patient heterogeneity in tumor response was observed. The average inter-tumor range of response was highest for volume and SUVtotal (45%) and lowest for SUVmean (25%). Heterogeneity greater than 40% was observed in 35% of cases (42/120), and in 10% of cases (12/120) heterogeneity greater than 75% was observed. Interestingly, in 18% of cases (21/120), tumors exhibited opposite trends in response, with one increasing and one decreasing at least 20% relative to baseline. Volume and SUVtotal are most sensitive to ROI definition, which represents the largest source of uncertainty for response calculation in this study.
Conclusions: Heterogeneous treatment response was observed in patients with multiple tumors, in some cases resulting in ambiguous trends. Accordingly, assessment of a single lesion is insufficient to fully characterize such patients' response to therapy. Future research will investigate how to apply imaging biomarkers of treatment response for this population, as this study suggests either new methods for interpretation or entirely new metrics are required.