The short-term outcomes of primary tumor resection with synchronous metastases from pancreas neuroendocrine tumors: an analysis of the ACS-NSQIP registry

Abstract

Background: The incidence of pancreatic neuroendocrine tumors (PNET) is rising worldwide, with many patients developing metastatic disease. Resection of the primary tumor despite of unresectable metastases is suggested to improve long-term oncologic outcomes. Assessments of the morbidity of this approach are scarce and limited to small, single-institutional studies. We sought to determine the patient characteristics and short-term outcomes of pancreatectomy for metastatic (M1) PNET, and compare them to non-metastatic (M0) PNET. The comparison was undertaken to provide a reference with which providers are familiar – i.e. outcomes of pancreatectomy for M0 disease, to better appreciate the burden of complications with M1. Methods: We conducted a retrospective cohort study using the ACS-NSQIP targeted pancreatectomy registry. We included all adults undergoing pancreatectomy for PNET between 2014-2016, with complete staging information. M1 and M0 PNET groups were created. Primary outcomes were 30-day major morbidity, mortality, and prolonged length of stay (>75th percentile). The characteristics and outcomes of M1 and M0 groups were compared using Pearson Chi square, Fisher exact test, and Student t test, as appropriate. A propensity score was created with the following potential confounders defined a priori: age, sex, body mass index, ASA classification, T and N stage, surgical procedure, open approach, and functional histology. The association between M1 and outcomes was adjusted for the propensity score using logistic binomial regression. Results: Of 1,381 included patients, 148 (12%) had M1 disease. Patients with M1 were more likely to be younger, have lower hematocrit, higher ASA class, lower body mass index, and have pre-operative weight loss, than M0 patients (all p < 0.05). T1/2 (75.7% vs. 28.8%, p < 0.01), node positive (72.3% vs. 25.9%, p < 0.01), and functional (29.7% vs. 22.1%, p = 0.04) tumors were more common in the M1 group. There were no differences in the type of pancreatectomy (distal pancreatectomy 60.1% Vs. 57.3%, p = 0.46), but there were more open resections (78.4% vs. 55.5%, p < 0.01) and longer median operating time (322.5 vs. 261 minutes, p < 0.01) for M1. No significant difference was detected in overall 30-day major morbidity (38.5% vs. 39.7%, p = 0.77) and mortality (2.7% vs. 1.0%, p = 0.06) between groups. Prolonged length of stay was more common with M1 (35.1% vs. 25.8%, p = 0.01). Adjusted regression analysis revealed no independent association between M1 and 30-day morbidity (relative risk 0.90, 95% confidence interval: 0.71–1.14), or prolonged length of stay (relative risk 2.89, 95% confidence interval: 0.78–10.7). Regression was not conducted on mortality due to too few events. Conclusion: In a multi-institutional analysis of prospectively collected data, resection of the primary tumor for M1 PNET was not associated with different post-operative morbidity and mortality than for M0 disease. The knowledge of short-term outcomes provided herein is important to weigh perioperative risks against the potential benefits of PNET primary tumor resection in the setting of M1 disease. This can support both decision-making and patient counselling.