Concordance between independent and investigator assessment of disease-free survival (DFS) in the APACT trial

4618 Background: APACT was a phase III trial of adjuvant nab-paclitaxel + gemcitabine ( nab-P + Gem) vs Gem alone in patients with resected pancreatic cancer (PC) and the first adjuvant PC trial to use independently assessed DFS as the primary endpoint (DFS by investigator review was a prespecified sensitivity analysis). We examined concordance between independent and investigator DFS review. Methods: For the independent assessment, reviewers determined recurrence by computed tomography or magnetic resonance imaging but were blinded to treatment and clinical data. Investigator-assessed DFS was based on all available data. Concordance was summarized by κ statistics. Patients who did not have recurrence or were alive were censored at the last tumor assessment date with disease-free status or the randomization date if the last tumor assessment with disease-free status was missing. Patients who received new anticancer therapy or cancer-related surgery prior to recurrence or death were censored at the date of last tumor assessment with disease-free status prior to the start of new anticancer therapy or cancer-related surgery or the randomization date if the last tumor assessment date with disease-free status prior to the start of subsequent new anticancer therapy or cancer-related surgery was missing. All censoring rules were the same for analysis of DFS by independent and investigator review. Results: Median DFS by independent review was 19.4 ( nab-P + Gem) vs 18.8 (Gem) months (hazard ratio [HR] 0.88; 95% CI, 0.73 - 1.06; P = 0.18); median investigator-assessed DFS was 16.6 ( nab-P + Gem) vs 13.7 (Gem) months (HR 0.82; 95% CI, 0.69 - 0.97; nominal P = 0.017). Moderate concordance was found between independent- and investigator-assessed DFS (Table); similar results were observed in the nab-P + Gem (concordance, 78%; κ coefficient, 0.56) and Gem alone (concordance, 76%; κ coefficient, 0.53) arms. Conclusions: The results reflect the complexities of defining the recurrence timepoint accurately and suggest that radiological review in the absence of clinical context is suboptimal for recurrence detection in resected PC. These findings may inform future clinical trial design. Registration: EudraCT (2013-003398-91); ClinicalTrials.gov (NCT01964430). Clinical trial information: NCT01964430 . [Table: see text]