Clinical predictors of bevacizumab-associated gastrointestinal perforation

Objectives: Bevacizumab is a generally well-tolerated drug, but bevacizumab-associated gastrointestinal perforations (BAP) occur in 0 to 15% of patients with ovarian carcinoma. Our goal was to evaluate the clinical predictors of BAP in order to identify factors, which may preclude patients from receiving treatment. Methods: We conducted a review of patients with recurrent epithelial ovarian carcinoma treated with bevacizumab between 2006 and 2009. Demographic and treatment data were collected for statistical analysis. Results: Eighty-two patients were identified; perforation occurred in 8 (9.76%). Among patients with perforation, a significantly higher incidence of prior bowel surgeries (p = 0.0008) and prior bowel obstruction or ileus (p < 0.0001) were found compared to non-perforated patients. The median age at onset of bevacizumab in the perforated group was 3 years younger (60 vs. 63 years, p = 0.61). The incidence of thromboembolic events, GI comorbidities, number of prior chemotherapies, and body mass index were similar between the groups. None of the patients in the perforated group developed grade 3 or 4 hypertension, compared to a 32.4% incidence among the non-perforated patients (p = 0.09). Upon multivariate analysis, when controlled for age greater or less than 60, prior bowel surgery, obstruction/ileus, and grade 3 or 4 hypertension, only the presence of obstruction/ileus was noted to be a significant predictor of perforation (p = 0.04). Conclusions: Predicting BAP remains a challenge. Bowel obstruction or ileus appears to be associated with increased risk of BAP.