Introduction/Background In advanced ovarian cancer (AOC) optimal cytoreductive surgery, <1 cm visible disease (VD), is associated with improved survival. Survival rates in patients with a suboptimal cytoreduction are equivocal. Surgery can be extensive and associated with significant morbidity and mortality. Tumour resectability and patient co-morbidity affect treatment planning. Pre-operative predictive models may provide an objective measure to aid this decision-making process. This study aimed to externally validate the ability of two pre-operative predictive models (Sudan et al 2014, 2017) to determine the likelihood of suboptimal cytoreductive surgery (>1 cm VD) and any residual disease in the treatment of AOC in a London teaching hospital.

Methodology Between January 2018- June 2020, 236 patients were treated for AOC in a London Teaching Hospital. 145 had cytoreductive surgery. 6 had incomplete records and were excluded. Suidan et al (2014, 2017) model’s resectability score 1 (RS1) (suboptimal cytoreduction) and resectability score 2 (RS2) (any residual disease) were used to score patients against clinical and radiological criteria. Receiver operating characteristic (ROC) curve analysis was used to determine the accuracy of models.

Results The optimal cytoreductive surgery rate was 88.28% (n=128). 80.69% (n=117) had no visible disease. Both RS1 and RS2 models predicted surgical outcomes. RS1 AUC 0.862 (95% CI: 0.8189 to 0.9067, P<0.0001), RS2 AUC 0.869 (95% CI 0.8263 to 0.9126, P<0.0001).

Conclusion In our centre, Suidan et al’s RS1 and RS2 models were able to predict cytoreductive outcomes. Predictive models may help determine patient suitability for cytoreductive surgery in AOC treatment.