Our preclinical study of sunitinib (SU) in combination with conformal large single fraction radiation in an orthotopic murine brain tumor model, discovered that changes in apparent diffusion coefficient (ADC), AUC and K_trans were promising imaging biomarkers that could predict response to SU as well as combined SU and radiation.¹ Based on our preclinical findings, we designed a prospective phase I trial of SU and radiosurgery (SRS) for brain metastases that incorporated translational investigation of these imaging biomarkers.

PURPOSE – To determining whether changes in diffusion (ADC) and dynamic contrast-enhanced MRI (AUC, K_trans) metrics were predictive for response to SU and SRS in our phase I study population of patients with brain metastases.

METHODS – Within a phase I trial that evaluated a 4-week course of SU at escalating doses in combination with SRS, delivered on day 7 of SU, multi-parametric MR images were acquired at baseline (Day -7 - Day 0, before SU), Day 7 (7 days of SU, before SRS), Day 9 (2 days after SRS) and 1 month (completion of SU). The following sequences were acquired at each time point: T2-weighted imaging, quantitative T1 (3D FLASH), DWI (echo-planar imaging with TR/TE 7700/110; diffusion gradient encoding in 3 orthogonal directions; b-values= 0, 150, 1000, 1800; FOV, 200 x 200 mm; matrix size 128 X 128; slice thickness 3 mm; number of signals acquired 3), DCE-MRI (3D FLASH with TR/TE 4.8/1.86; flip angles = 30deg; FOV 220 x 200 mm; matrix size 174 x 192 pixels; slice thickness 1.5 mm), and T1-gad (3D MPrage).

Kinetic analysis for DCE-MRI utilized a voxel-based approach (temporal dynamic analysis, TDA)² and assumed a 2-compartment Modified Tofts model to estimate K_trans, K_ep, V_e and iAUC. The tumors were segmented semi-automatically using common thresholds at each time point on T1-gad. Histogram analysis of the TDA results on each ROI was completed. ADC maps were produced with in-house software (Matlab) and calculated for each voxel by fitting the mono-exponential model equation, $$$$$, to 4-point plots of signal intensity (S) by using a linear least square fit algorithm. Tumour contours from T1-gad images were transferred to the ADC maps. Mean and median ADC was calculated for each tumor. Only patients with both baseline and serial follow-up imaging data were evaluated for imaging biomarker data analysis. Changes in tumor volumes (T1-gad), DCE-MRI metrics (iAUC, K_trans) and diffusion weighted imaging metrics (ADC) were evaluated in comparison to baseline measurements (taken at day -7). Correlation results were calculated by Pearson correlation and significance tests generating p-values were performed by Whitney Rank Sum test in SigmaPlot.

RESULTS – Biomarker analysis was completed on 13 tumors from 10 patients accrued in the phase I trial (13 ADC and 11 DCE-MRI). When all histologies were grouped together, correlations were weak between tumor response (defined as volume reduction from baseline to day 20) and changes in perfusion metrics (K_trans and AUC) from baseline to day 0 (K_trans -0.27, AUC -0.32) or day 2 (K_trans -0.44, AUC -0.41). However, RCC tumors had a greater response compared with other histologies after 7 days SU reaching significance for AUC (p=0.045) but not K_trans (p=0.41). Similarly, ADC response after 7 days SU, measured as ADC(day0)/ADC(day-7), was 0.79 for RCC tumors compared with 1.00 for other histologies (p=0.008). At this early time point after 7 days SU, there was a strong correlation between ADC response and tumor volume change from baseline (R²=0.77). After SRS, the difference between tumor types was no longer evident at day 2 (AUC p=0.195 K_trans, p=0.29) or at day 20 (AUC p=0.126 K_trans, p=0.089)

CONCLUSION – Differential early imaging responses to sunitinib were detected for RCC brain metastases using diffusion (ADC) and DCE metrics (AUC). As sunitinib is an established effective treatment for RCC,³ these early changes in ADC and AUC may show promise as predictive biomarkers of response to sunitinib. Following radiosurgery, a focal ablative large dose radiation treatment, this differential response in ADC and AUC was no longer seen between RCC and other histologies.