Routine Baseline Variables Fail to Reliably Predict Mucinous Borderline Ovarian Recurrence

Routine clinical variables gathered at the time of primary surgery fail to provide clinically actionable risk stratification for patients with mucinous-dominant borderline ovarian tumors, even though disease recurrences frequently emerge many years after initial surgical treatment.

Postoperative surveillance protocols for borderline ovarian tumors classified with mucinous and related institutional histology labels remain heterogeneous across oncology practice. This uncertainty regarding optimal follow-up is particularly prominent following organ-sparing surgical interventions, leaving open the clinical question of whether routine baseline factors can reliably identify patients requiring intensified or extended monitoring.

To investigate recurrence patterns and the utility of prognostic modeling, researchers evaluated a retrospective cohort of 178 patients treated at a single center between 2010 and 2022. The index operation was defined as the primary tumor-removing surgery. One patient who never achieved a postoperative disease-free state was excluded.

In the study protocol, a first recurrence required objective demonstration of a new ovarian lesion following postoperative imaging that confirmed no evidence of disease, alongside subsequent histological confirmation. Across the cohort, 27 of 178 patients, or 15.2 percent, developed a pathology-confirmed first recurrence, and seven of those individuals later developed a second recurrence, representing 34 total recurrence episodes. The median time to first recurrence was 41.0 months, with an interquartile range of 20.5 to 66.0 months. Ten first recurrences were documented more than five years following index surgery, including three that manifested at ten years or later.

To evaluate whether routinely collected patient characteristics could anticipate recurrence, investigators examined patient age, FIGO stage IC or higher, and the use of broad organ-sparing surgery. These clinical parameters were assessed using ridge-penalized Cox proportional hazards regression and an alternative random survival forest machine learning model under repeated nested cross-validation.

Model discrimination across the cohort remained limited. The concordance index was 0.719 with a 95 percent confidence interval of 0.612 to 0.807 for penalized Cox regression, compared with 0.742 and a 95 percent confidence interval of 0.655 to 0.816 for the random survival forest, corresponding to an insignificant paired difference of 0.023 with a 95 percent confidence interval of -0.102 to 0.136. When evaluated at 36 months, the area under the receiver operating characteristic curve was 0.666 for the Cox model and 0.580 for the random survival forest. Among 73 patients with known status at 36 months, average precision was 0.373 for Cox regression and 0.207 for the random survival forest.

In addition to modest discrimination statistics, neither predictive approach demonstrated stable clinical net benefit, and the random survival forest failed to show a dependable advantage over standard penalized regression. The researchers concluded that while late recurrences underscore ongoing clinical uncertainty regarding the appropriate duration and intensity of postoperative monitoring after organ-sparing procedures, these models remain strictly exploratory and cannot be used to establish follow-up schedules or alter routine clinical care.

Subject of Research: Postoperative recurrence timing and predictive modeling in mucinous-dominant borderline ovarian tumors

Article Title: Recurrence timing and exploratory prediction after surgery for mucinous-dominant borderline ovarian tumors: implications for postoperative surveillance in a single-center retrospective cohort

Article References: Lin, J., Liu, Y., Chen, P., Li, Y., & Jin, R. (2026). Recurrence timing and exploratory prediction after surgery for mucinous-dominant borderline ovarian tumors: implications for postoperative surveillance in a single-center retrospective cohort. BMC Women's Health. https://doi.org/10.1186/s12905-026-05007-8

Image Credits: AI Generated

DOI: 10.1186/s12905-026-05007-8

Keywords: mucinous borderline ovarian tumors, borderline ovarian tumors, postoperative surveillance, organ-sparing surgery, tumor recurrence, recurrence timing, late recurrence, ridge-penalized Cox regression, random survival forest, FIGO stage IC, nested cross-validation, concordance index, clinical net benefit, retrospective cohort study, pathological confirmation, single-center study, index surgery