Objective: To explore the survival factors and construct a prognostic index (PI) for oral squamous cell carcinoma (OSCC). Methods: From January 2004 to June 2016, a total of 634 patients with pathologically confirmed OSCC were recruited in a hospital of Fujian. The clinical and follow-up data of all the patients with pathologically confirmed OSCC were collected to identify the factors influencing the prognosis of OSCC. All the patients were randomly divided into two groups: modeling group (modeling dataset, n=318) and validation group (validation dataset, n=316). Randomization was carried out by using computer-generated random numbers. In the modeling dataset, survival rates were calculated using Kaplan-Meier method and compared using the log-rank test. Cox regression model was used to estimate the hazard ratio (HRs) and 95% confidence intervals (CIs) of prognosis factors. An PI for OSCC patients prognostic prediction model was developed based on β value of each significant variable obtained from the multivariate Cox regression model. Using the tertile analysis, patients were divided into high-risk group, moderate-risk group, and low-risk group according to the PI, the Akaike information criterion (AIC) and Harrell's c-statistic (C index) were used to evaluated the model's predictability. Results: Results from the multivariate Cox regression model indicated that aged ≥55 years (HR=2.22, 95%CI: 1.45-3.39), poor oral hygiene (HR=2.12, 95%CI: 1.27-3.54), first diagnosis of lymph node metastasis (HR=5.78, 95%CI: 3.60-9.27), TNM stage Ⅲ-Ⅳ (stage Ⅰ as reference) (HR=2.43, 95%CI: 1.10-5.37) and poor differentiation (well differentiation as reference) (HR=2.53, 95%CI: 1.60-4.01) were the risk factors influencing the prognosis of OSCC. The PI model had a high predictability in modeling group and validation group (AIC and C index were 1 205.80, 0.700 2 and 1 150.47, 0.737 3). Conclusion: Age, poor oral hygiene, first diagnosis of lymph node metastasis, TNM stage and histological grade were factors associated with the prognosis of OSCC, and the PI model has a certain significance in the clinical treatment of OSCC.
目的： 探讨福建省口腔鳞状细胞癌（OSCC）患者预后影响因素并构建预后指数（PI）。 方法： 收集福建医科大学附属第一医院口腔颌面外科2004年1月至2016年6月经病理确诊634例OSCC患者的临床资料，并进行随访，将患者随机分为建立模型组（建模组，318例）和验证模型组（验证组，316例）。在建模组，运用Kaplan-Meier法计算生存率，log-rank检验比较生存率差别，采用Cox比例风险回归模型探讨OSCC患者预后影响因素并计算死亡风险比值（HR）及其95%CI，以模型组中有意义的预测变量的β值计算PI。运用三分位法将患者分为高危组、中危组和低危组，并在验证组中采用赤池信息准则（AIC）和Harrell一致性指数（Harrell’s c-index，C）检验模型的预测效能。 结果： Cox比例风险回归模型分析发现，OSCC患者中年龄≥55岁的HR值（95%CI）为2.22（1.45~3.39）；口腔卫生较差的HR值（95%CI）为2.12（1.27~3.54）；首诊淋巴结转移的HR值（95%CI）为5.78（3.60~9.27）；TNM分期为Ⅲ~Ⅳ期（Ⅰ期为参照）的HR值（95%CI）为2.43（1.10~5.37）；组织分化程度为低分化（高分化为参照）的HR值（95%CI）为2.53（1.60~4.01）。建模组和验证组中预后模型的预测效能良好（AIC和C值分别为1 205.80、0.700 2和1 150.47、0.737 3）。 结论： 年龄、口腔卫生、首诊断淋巴结转移、TNM分期和组织病理学分级是OSCC预后影响因素，构建PI模型可有效指导临床治疗。.
Keywords: Cox regression; Oral carcinoma; Prognostic index.