Pretreatment diffusion weighted imaging for clinical outcome assessment in patients undergoing definitive chemoradiation for pancreatic adenocarcinoma.

Abstract

432 Background: Factors predicting patterns of failure for pancreatic ductal adenocarcinoma (PDA) are not well-established. Diffusion-weighted MRI (DWI) is useful in predicting recurrence for other gastrointestinal malignancies. The purpose of this study was to evaluate for correlation between tumor DWI parameters and clinical outcomes following chemoradiotherapy (CRT) for pancreatic adenocarcinoma. Methods: From 2009 to 2013, 27 patients with locally advanced (n=14), borderline resectable (n=12) or resectable (n=1) PDA underwent CRT. All patients received upfront FOLFIRINOX or gemcitabine-based chemotherapy prior to CRT, and gemcitabine (median weekly dose 800 mg/m2) concurrent with radiotherapy. DWI was obtained during radiotherapy treatment planning, and apparent diffusion coefficient (ADC) maps were generated to allow determination of median tumor ADC. Tumor-directed CRT was administered with respiratory gated intensity modulated radiation therapy, 55Gy in 25 fractions without elective nodal coverage. Patients were followed by imaging every 2-3 months. Log rank test was used to estimate an optimal ADC cut-point of 1.4*10-3mm2/sec based on distant metastasis free survival (DMFS) and overall survival (OS). Analysis of local recurrence was not performed due to limited events. The log-rank test was used to evaluate differences in outcome between groups based on ADC classification. Results: The median time to last follow-up from completion of CRT was 9.7 months for all patients and 11 months among living patients. 5 patients underwent resection following CRT. The 1-year DMFS for patients with median ADC<1.4*10-3mm2/sec was 79.5 percent, compared to 47.6 percent for those with median ADC >1.4. The 1-year OS for patients with median ADC<1.4 was 100 percent, compared to 46.3 percent for those with median ADC >1.4. Both DMFS (p=.041) and OS (p=.003) were significantly improved on log-rank test for ADC<1.4 compared to ADC>1.4. Conclusions: A correlation was observed between DWI parameters and clinical outcomes for PDA. Further prospective study should be explored to validate DWI as a prognostic imaging biomarker.