Is CA19-9 Response Following Chemoradiation Therapy for Borderline Resectable Pancreatic Cancer an Early Surrogate for Eventual Survival Outcomes?

Abstract

Early surrogate markers of progression-free survival (PFS) and overall survival (OS) in pancreatic cancer patients who receive chemoradiation therapy (CRT) have been poorly studied. The current study was conducted to determine whether tumor marker CA19-9 following CRT is an independent prognostic biomarker in borderline resectable pancreatic cancers. Patients who had borderline resectable pancreatic cancer per the M.D. Anderson classification and received CRT before potentially undergoing surgery between August 2006 and August 2012 were identified. Patients with pre-CRT Ca19-9 levels that were unrecorded (n = 26), ≤50 (n = 38) or associated with a bilirubin >2 (n = 13) were excluded from analysis as were those with unrecorded post-CRT CA19-9 levels (n = 6). Ninety-three patients met the study's inclusion criteria. Seventy-seven (83%) underwent induction chemotherapy before consolidative CRT. Median radiation dose was 50.4 Gy (range, 30 to 70 Gy). Concurrent chemotherapy during CRT was 5-fluorouracil (1.1%), gemcitabine-based (22.6%), or capecitabine-based (66.7%). Univariate and multivariate statistical methods were used to determine significance of the post-CRT changes in CA19-9 on PFS and OS outcomes calculated from start of treatment. Median follow-up was 14.0 months (range, 1.8 to 76 months). Estimated median OS and PFS were 16.5 and 8.3 months, respectively. On univariate analysis, post-CRT normalization of CA19-9 (≤50) was a significant prognostic factor for both OS and PFS. Median OS for patients whose CA19-9 normalized was 27.4 months compared to 15.5 months for those whose CA19-9 did not normalize (p = 0.042). Corresponding median PFS durations were 11.2 and 7.0 months (p = 0.018), respectively. The most critical determinant of OS and PFS on univariate analysis was selection of patients for surgery following CRT. Median OS was 30.8 months for patients who underwent surgery compared to 14.9 months for those who did not (p < 0.001). On multivariate analysis, post-CRT CA19-9 was no longer an independent predictor of OS or PFS. The biggest determinant of this outcome was the selection of patients for surgery following CRT with a hazard ratio of 0.33 (95% CI, 0.18-0.62, p = 0.001) for OS. An association was noted between the likelihood of undergoing surgery and CA19-9 normalization (64.3% of surgical patients, 35.7% of non-surgical patients, p = 0.002). While post-treatment normalization of CA19-9 may predict PFS or OS outcomes among borderline resectable pancreatic cancer patients treated with CRT, this effect probably influences the decision to undergo surgery, which then overshadows the impact on PFS and OS.