Sunitinib-associated Pancreatic Atrophy in Patients with Gastrointestinal Stromal Tumor: A Toxicity with Prognostic Implications Detected at Imaging.

Purpose To evaluate the effect and clinical importance of sunitinib on pancreatic volume in patients with gastrointestinal stromal tumor (GIST). Materials and Methods This retrospective study was approved by the institutional review board and compliant with HIPAA. The requirement to obtain informed consent was waived. The authors evaluated 65 patients with GIST treated with oral sunitinib and a control group of 30 patients with GIST who did not receive any therapy (mean age: 56 years [range, 29-75 years] vs 60 years [range, 27-78 years], respectively; P = .11) seen at their institution from January 2002 through December 2008. Segmented pancreatic volumes of study and control groups were measured with computed tomography by using commercial software by two independent readers who were blinded to study group and the timing of the scan at a median of 6.2 and 6.1 months, respectively. Pre- and posttreatment volumes (Wilcoxon signed rank test) and rate of volume change per month (Wilcoxon rank sum test) were compared. Interobserver agreement was calculated. Associations and prognostic importance of pancreatic atrophy were studied by using multivariate linear regression and Cox proportional analysis, respectively. Results Both readers recorded significant pancreatic volume loss in the study group (respective median pre- and posttreatment volume: 76.1 cm(3) and 58.4 cm(3) for reader 1 and 67.7 cm(3) and 59.0 cm(3) for reader 2; P < .0001 for both) but not in the control group (respective median pre- and posttreatment volume: 79.9 cm(3) and 83.8 cm(3) for reader 1 [P = .43] and 79.9 cm(3) and 84.8 cm(3) for reader 2 [P = .50]). The rate of volume loss per month was greater in the study group than in the control group (reader 1: -2.1% vs -0.1%, respectively, P = .003; reader 2: -2.0% vs -0.3%, P < .0001). Twenty-three of the 65 patients who received sunitinib (35%) showed at least 3% pancreatic volume loss per month, compared with only one of the 30 patients in the control group (3%). The concordance correlation coefficient for pre- and posttreatment measurements was 0.83 (95% confidence interval [CI]: 0.75, 0.89) and 0.91 (95% CI: 0.86, 0.94), respectively, and the mean relative difference was 0.04% and 1.2%. Sunitinib treatment was independently associated with pancreatic atrophy (P = .03; risk estimate: 2.64; 95% CI: 0.17, 5.11). At Cox proportional analysis within the study group, more than 3%, more than 5%, and more than 7% loss per month were independently associated with worse survival (P < .001, hazard ratio: >1.00 for all). Conclusion Pancreatic atrophy is a sunitinib-associated toxicity detected at imaging. It may be a clinically important biomarker because a higher rate of pancreatic atrophy was independently associated with shorter survival. (©) RSNA, 2016.