Abstract
BACKGROUND: Not all patients with advanced pancreatic cancer (PC) profit from 2nd-line chemotherapy. We evaluated predictive factors that are routinely collected during clinical care with the aim to support an informed and shared decision. METHODS: In a prospective study across 35 German sites, 151 patients with PC previously treated with gemcitabine/nab-paclitaxel were enrolled and 146 patients received biweekly nanoliposomal irinotecan/5-fluorouracil/FA. We investigated whether time-to-treatment-failure of 1st-line (TTF1) predicts 2nd-line treatment outcome. Patients were stratified into three equal cohorts based on TTF1. Primary endpoint was TTF2, with secondary endpoints including overall survival (OS) and growth modulation index (GMI). RESULTS: Median TTF2 was 3.71 months (95% CI 2.50-4.11). Median OS was 7.72 months (95% CI 6.11-9.00). TTF1 did not predict TTF2 or OS (HR 0.93, 95% CI 0.58-1.47. Baseline parameters significantly associated with TTF2 and OS included neutrophil count, CRP levels, and liver metastases, whereas ECOG performance score (PS) was primarily associated with OS and to a lesser extent with TTF2. During treatment, patients with a CA 19-9 reduction of ≥25% had significantly improved TTF2 and OS (p < 0.001). CONCLUSION: TTF1 is not predictive of TTF2 or OS. Therefore, 2nd-line treatment should not be withheld irrespective of duration of TTF1. CA 19-9 dynamics can be used to predict further benefit to some extent. CLINICAL TRIAL REGISTRATION: EudraCT: 2016-005147-17; ClinicalTrials.gov: NCT03468335.