H. A. Choudry1, Y. Shuai2, J. F. Pingpank1, M. P. Holtzman1, S. S. Ahrendt1, H. L. Jones1, L. Ramalingam1, A. H. Zureikat1, H. J. Zeh1, D. L. Bartlett1 1University Of Pittsburgh Medical Center,Surgical Oncology,Pittsburgh, PA, USA 2University Of Pittsburgh Cancer Institute,Biostatistics Facility,Pittsburgh, PA, USA
Introduction: Postoperative morbidity may negatively impact cancer-related outcomes by inducing a pro-tumorigenic environment and preventing the timely initiation of postoperative systemic therapy. We hypothesized that postoperative morbidity would predict cancer-related survival, independent of tumor histology, grade, extent of disease, and other comorbidities.
Methods: We addressed our hypothesis by using a prospective database of 1296 patients with peritoneal metastases undergoing complex surgical resection with high postoperative morbidity and long-term cancer-related mortality rates. We graded all postoperative morbidity using the Clavien-Dindo grading system. Kaplan-Meier method was used to estimate survival. Multivariate analyses identified associations with survival and postoperative morbidity.
Results: Cytoreductive surgery and hyperthermic intraperitoneal chemoperfusion was performed for peritoneal metastases from cancers of the appendix (50%), colorectum (30%), ovary (8%) and mesothelioma (12%). Tumor burden assessed by median peritoneal carcinomatosis index (PCI) was 16 and optimal cytoreduction (residual tumor < 2.5mm) was achieved in 93% of patients. Major postoperative morbidity (Clavien-Dindo grades 3-5) occurred in 24% of patients and long-term cancer-related mortality was 53%, after a median follow-up of 55 months. Median progression-free survival and overall survival calculated from surgery were 15 and 39 months, respectively. In a multivariate Cox proportional hazards model, major postoperative morbidity (Clavien-Dindo grades 3/4) was an independent negative predictor of survival (HR 1.4) along with non-appendiceal primary histology, higher tumor grade, higher PCI, incomplete cytoreduction, higher age-adjusted Charlson comorbidity index, and recurrent symptomatic disease at presentation. Patients with grades 3/4 postoperative morbidity were 1.6/2.5 times more likely to die of their cancer than those with no post-operative complications. Using multivariate logistic regression model, independent predictors of major postoperative morbidity included higher preoperative ASA (American Society of Anesthesiologists) physical status classification, longer operative time, higher PCI, and non-appendiceal primary histology.
Conclusion: In our experience, postoperative morbidity independently predicted cancer-related survival, regardless of comorbidities, tumor type, extent, grade, and completeness of surgery. Future work will focus on mechanism underlying this phenomenon. Moreover, the extent of surgical resection required to clear the disease played a dominant role in predicting occurrence of postoperative morbidity. Ongoing studies will address optimization of selection criteria and perioperative management strategies that may reduce postoperative morbidity in such patients that frequently require lengthy procedures and multi-visceral resections.