Objective: To determine the utility of a clinical calculator to redefine prognosis and need for chemotherapy among patients with early-stage high-risk epithelial ovarian cancer.
Methods: Data were abstracted for stage I-II, high-risk ovarian cancer from the National Cancer Database from years 2005 to 2015. Based on demographic, pathologic, surgical, and laboratory characteristics, a clinical score was developed using Cox regression. Propensity score weighting was used to adjust for differences between patients who did and did not receive chemotherapy.
Results: Of 8188 patients with early-stage high-risk ovarian cancer, 6915 (84%) did and 1273 (16%) did not receive chemotherapy. A clinical calculator was created utilizing age, stage, histology, grade, tumor size, number of pelvic and paraaortic lymph nodes examined, the presence of malignant ascites, and CA125. The calculator divided patients into low, moderate, and high-risk groups with 5-year OS (overall survival) of 92%, 82%, and 66%, and 10-year OS of 85%, 67%, and 44%, respectively. Chemotherapy improved 5-year OS and 10-year OS in the high-risk group (56% to 73%; p < 0.001, 34% to 48%; p < 0.001). The moderate risk group had improved 5-year OS (80% to 85%; p = 0.01) but not 10-year OS (66% to 66%; p = 0.13). Chemotherapy did not improve 5-year or 10-year OS in low-risk patients (93% to 92%, p = 1.0, 86% to 84%, p = 0.99).
Conclusions: The prognosis among high-risk early-stage ovarian cancer patients is heterogeneous. This calculator may aid in patient-centered counseling regarding potential treatment benefits.
Keywords: Chemotherapy; Machine learning; Ovarian cancer; Personalized medicine; Risk score; Treatment.
Published by Elsevier Inc.