Assessment of Outcomes by Intention-to-Treat Comparison for Locally Advanced Pancreatic Cancer: A Population-Derived Cohort Study.

Abstract

The overall treatment response among patients with locally advanced pancreatic cancer (LAPC) is poorly understood as most studies report solely on resected patients. We aimed to investigate the outcomes in patients with LAPC as an intention-to-treat-analysis from the time of diagnosis from a complete source population. An observational cohort study in a population-defined region within a universal healthcare system. All consecutive patients discussed at multi-disciplinary tumour board (MDT), aged ≥ 18 years and diagnosed with LAPC were included. Exposure was set as recommended treatment by MDT (i.e. upfront surgery, neoadjuvant therapy, palliative treatment or best supportive care). Outcome measures were overall survival analysed by Kaplan-Meier survival estimates and multivariable analyses using logistic regression for odds ratios (OR) and Cox proportional hazard analysis for hazard ratios (HR). In total, 8803 MDT events (6055 unique patients) with pancreatic disease were held during the study period. Some 1436 (24%) had pancreatic cancer, of which 162 (11%) had LAPC and 134 met the population-defined criteria. In overall survival analyses, the patients who were recommended neoadjuvant therapy (± surgery) demonstrated no significant difference to palliative chemotherapy (median 11.0 months vs. 11.8 months; p = 0.226). In multivariable analysis, adjusted OR for overall survival comparing the treatment groups was 0.27 (95% CI 0.02-3.29, p = 0.306) and Cox proportional HR 0.96 (95% CI 0.58-1.59, p = 0.865). In patients with LAPC, survival was not statistically different between those recommended for attempt at neoadjuvant (± surgery) compared with those recommended palliative chemotherapy. The findings suggest that conversion/downstaging chemotherapy is successful in only a select few.