knynk1107@yuhs.ac

To construct a CT based Fagotti scoring system without staging laparoscopy by analyzing correlation between the laparoscopic findings and the corresponding CT findings in advanced ovarian cancer patients.

The pre-operative CT and staging laparoscopic records based on Fagotti score system of 157 patients with stage III/IV ovarian cancer were reviewed, who underwent debulking surgery between 2010 and 2018. Ten CT features known as predictor of sub-optimal resection were evaluated by two independent radiologists who were blinded to the laparoscopy and the surgical outcome. Each imaging features were matched with relevant laparoscopic parameters by Spearman correlation between them. Variable selection and Model construction was performed by logistic regression with a least absolute shrinkage and selection operator (LASSO) method. Final CT-based scoring system was internally validated using 5-fold cross validation.

Among the 157 patients, 120 (76.4%) was rated predictive index value (PIV, sum of scores) >= 8 on staging laparoscopy, who assigned to non-resectable group initially. Complete/optimal cytoreduction was achieved in 23 (63.5%)/37(100%) among the remaining 37 patients (PIV < 8), respectively. Table 1 shows regression coefficient between CT features and laparoscopic parameters as result of LASSO regression modeling. The ROC analysis showed that the area under the curve(AUC) was 0.7234 (95% CI 0.6225~0.8243) (Fig.2).

Central tumor burden such as mesenteric diseases and paraaortic lymphadenopathy and upper abdominal spread including diaphragm and gastro-transverse-splenic (GTS) space involvement on preoperative CT was identified distinct prediction factor for high PIV. The CT based PIV prediction model may be useful for patient stratification in the era of staging laparoscopy.

Although the achievement of complete cytoreduction was known as the important prognostic factor of advanced ovarian cancer, there is no standardized model for predicting surgical outcome.