Department of Obstetrics and Gynecology, Nagoya University Graduate School of Medicine, Tsurumai-Cho 65, Showa-Ku, Nagoya, 466-8550, Japan.
Department of Obstetrics and Gynecology, Toyohashi Municipal Hospital, Toyohashi, Japan.
Int J Clin Oncol. 2023 Dec;28(12):1680-1689. doi: 10.1007/s10147-023-02417-8. Epub 2023 Oct 7.
This study aimed to explore the prognostic value of mean platelet volume (MPV) in patients with ovarian clear cell carcinoma (OCCC) and evaluate the predictive performance of a random forest model incorporating MPV and other key clinicopathological factors.
A total of 204 patients with OCCC treated between January 2004 and December 2019 were retrospectively analyzed. Clinicopathological characteristics and preoperative laboratory data were collected, and survival outcomes were evaluated using the Kaplan-Meier method and Cox proportional hazards models. An optimal MPV cutoff was determined by receiver operating characteristic (ROC) curve analysis. A random forest model was then constructed using the identified independent prognostic factors, and its predictive performance was evaluated.
The ROC analysis identified 9.3 fL as the MPV cutoff value for predicting 2-year survival. The MPV-low group had lower 5-year overall survival and progression-free survival rates than the MPV-high group (p = 0.003 and p = 0.034, respectively). High MPV emerged as an independent prognostic factor (p = 0.006). The random forest model, incorporating the FIGO stage, residual tumors, peritoneal cytology, and MPV, demonstrated robust predictive performance (area under the curve: 0.905).
MPV is a promising prognostic indicator in OCCC. Lower MPV correlated with worse survival rates, advocating its potential utility in refining patient management strategies. The commendable predictive performance of the random forest model, integrating MPV and other significant prognostic factors, suggests a pathway toward enhanced survival prediction, thereby warranting further research.
本研究旨在探讨平均血小板体积（MPV）在卵巢透明细胞癌（OCCC）患者中的预后价值，并评估纳入 MPV 和其他关键临床病理因素的随机森林模型的预测性能。
回顾性分析了 204 例 2004 年 1 月至 2019 年 12 月期间接受治疗的 OCCC 患者。收集了临床病理特征和术前实验室数据，并采用 Kaplan-Meier 方法和 Cox 比例风险模型评估生存结局。通过受试者工作特征（ROC）曲线分析确定最佳 MPV 截断值。然后，使用确定的独立预后因素构建随机森林模型，并评估其预测性能。
ROC 分析确定 9.3 fL 为预测 2 年生存率的 MPV 截断值。MPV 低组的 5 年总生存率和无进展生存率均低于 MPV 高组（p=0.003 和 p=0.034）。高 MPV 是独立的预后因素（p=0.006）。纳入国际妇产科联盟（FIGO）分期、残余肿瘤、腹膜细胞学和 MPV 的随机森林模型显示出强大的预测性能（曲线下面积：0.905）。
MPV 是 OCCC 有前途的预后指标。较低的 MPV 与较差的生存率相关，这表明其在细化患者管理策略方面具有潜在的应用价值。随机森林模型结合 MPV 和其他重要预后因素的卓越预测性能表明，可能有一条途径可以提高生存预测，因此值得进一步研究。
