Usefulness of computed tomography in predicting cytoreductive surgical outcomes for ovarian cancer.

PURPOSE

The objective of this study was to identify features of preoperative computed tomography (CT) scans that can best predict outcomes of primary cytoreductive surgery in ovarian cancer patients.

METHODS

Preoperative CT scans of 98 patients were evaluated retrospectively. Multiple logistic regression analysis was used to develop two models.

RESULTS

Although optimal surgical reduction was attempted in 98 patients, 12 had suboptimal results. Having tumor implants on the small or large bowel mesenteries (any size) or at other sites (cutoff index: ≥ 1 cm) was found to be significant (p < 0.001) for predicting a suboptimal cytoreduction outcome. Two predictive models were created using multiple logistic regression analysis; both consider diffuse peritoneal thickening (DPT), infrarenal para-aortic or pelvic lymph node involvement, a bowel encasement tumor (≥ 2 cm), and any tumor implants in the cul-de-sac as significant. Model 1 adds consideration to any tumors in the pelvic or retroperitoneum and has an accuracy of 90.8% for predicting a suboptimal surgery. Model 2 (accuracy of 93.9%) adds to the core of predictors the presence of tumor implants on the bowel mesenteries (≥ 2 cm), omental caking (≥ 2 cm), and ascites fluid.

CONCLUSION

Using specific CT findings from patients with ovarian cancer, we have devised two predictive models that have an accuracy of greater than 90% for predicting whether cytoreductive surgery will completely remove all tumor tissue, which should greatly aid in the differential decision-making as to whether to attempt cytoreductive surgery first, or to advance directly to neoadjuvant chemotherapy.