Predictors of recurrence and implications for overall survival in borderline resectable pancreatic cancer patients treated with total neoadjuvant therapy: A retrospective cohort study.

Abstract

e16781 Background: Borderline resectable pancreatic ductal adenocarcinoma (BR PDAC) is a subset of pancreatic cancer with unique prognostic implications. A multimodality, neoadjuvant therapy (NAT) approach has known benefits such as downstaging tumors, reducing the risk of a positive margin, and treating micrometastatic disease. Patient selection and therapy sequencing are controversial owing to the high risk of recurrence. Therefore, we aimed to identify factors predicting recurrence and overall survival (OS) in BR PDAC patients undergoing NAT. Methods: We identified BR PDAC patients treated with NAT followed by surgery at Moffitt Cancer Center between 2008-2015. We evaluated clinical, demographic, and perioperative factors to identify predictors of recurrence and OS. Statistical analyses were performed with univariate and multivariable Cox regression models. Results: 117 patients with BR PDAC who received NAT were evaluated; 53 (45%) were female and 64 (55%) were male. Of those, 91 (78%) received gemcitabine/xeloda/taxotere and the remainder received other gemcitabine or 5FU regimens. 107 (91%) patients received neoadjuvant radiation, mainly 5-day dose painted SBRT. Post- NAT CA 19-9 normalized in 39 (33%) and 11 (9%) additional patients normalized after surgery. Pathologic treatment effect was appreciated in 85 (73%) patients and pathologic complete response (pCR) was seen in 18 (15%). 95 (81%) patients initiated adjuvant therapy. In multivariable analysis, the strongest predictor of recurrence was higher log(10) post-operative CA 19-9 (HR 2.29; 95% CI, 1.40-3.75). Time from initiation of NAT to surgery ≥ 5 months also predicted recurrence (HR 2.08; 95% CI, 1.16-3.72). In OS analysis, 71 patients had recurrence after multimodality treatment; 61 (86%) died and 10 (14%) were alive. In multivariable analysis, the strongest predictors of prolonged OS from recurrence were female gender (HR 0.47; 95% CI, 0.26-0.84) and presence of a treatment effect (HR 0.37; 95% CI, 0.15-0.93). Conclusions: We observed treatment effects and pCR comparable to other institutions, supporting a multimodality approach to BR PDAC. Higher post-operative CA 19-9 and time to surgery ≥ 5 months from initiation of NAT were associated with an increased risk of recurrence. These data suggest that timely receipt of surgery after completion of NAT may impact recurrence and OS in BR PDAC.