Background: A proportion of women with ovarian cancer, even those with advanced-stage disease, will experience long-term survival post diagnosis. Mixture cure modeling provides insight into factors related to both short-term recurrence and death as well as long-term survival. The goals of this study were to estimate the long-term survival rate, and to evaluate associations between patient characteristics and the long-term survival rate and landmark 5- and 10- year survival, in a sample of newly-diagnosed ovarian cancer patients.

Methods: The Tempus EMR dataset contains patients from academic & community oncology centers in the U.S. This study included adult women with a new primary diagnosis of ovarian, fallopian tube, or peritoneal cancer between 2000 and 2020. Women treated with a poly-ADP ribose polymerase inhibitor (PARPi), pembrolizumab, or nivolumab were excluded (n = 947; final sample size = 6,598). Mixture cure models were used to estimate the long-term survival rate (probability of living as long as an age- and gender-matched person without ovarian cancer), overall and stratified by patient demographic & clinical characteristics. Relative importance of variables associated with long-term survival was evaluated using logistic regression models with 5- and 10-year landmark outcomes.

Results: Younger age, earlier stage, lower grade, favorable ECOG score, and platinum sensitivity were associated with better overall survival (OS). The long-term survival rate in our model was 25.0% (95% CI: 21.5-28.9). The long-term survival rate was higher for age < 65 vs. age 75+ [29.0% (24.8-33.6) vs. 10.5% (2.9-31.6)], early vs. advanced stage disease [62.0% (41.4-78.9) vs. 17.8% (14.1-22.2)], grade 1-2 vs. 3-4 [31.4% (12.1-60.3) vs. 20.4% (14.6-27.8)], ECOG 0-2 vs. 3-4 [26.1% (21.4-31.3) vs. 2.6% (0.05- 56.5)], and platinum sensitive vs. platinum resistant patients [47.5% (43.1-51.9) vs. 9.4% (7.0-12.5)]. Analyzing 5- and 10-year landmarks [n = 3,944 (59.8%), n = 3,107 (47.1%), respectively] in multivariable logistic regression models, platinum sensitivity had the strongest association, followed by cancer stage.

Conclusions: Understanding differences in long-term survival based on patient characteristics is important for clinical decision-making and personalizing treatment. Overall, one quarter of ovarian cancer patients were long-term survivors; long-term survivors were more likely to be younger and have earlier stage disease, low grade disease, better performance status, or sensitivity to platinum-based therapy. These results also provide a baseline to evaluate long-term survival in patients on newer therapies such as PARPi when sufficient follow-up data are available.

VIEW THE PUBLICATION