应用竞争风险模型探索宫颈癌患者预后的影响因素

Determination of prognostic factors of cervical cancer patients : Application of a competing risk model

目的 :探讨竞争风险场合下影响宫颈癌患者预后的因素。方法:选取美国医疗保险监测、流行病学和最终结果(Surveillance,Epidemiology and End Results,SEER)数据库中1988—2008年华人宫颈癌患者的数据。兴趣事件为患者死于宫颈癌,竞争事件为死于其他疾病(如心脏病或其他肿瘤等)。在考虑竞争风险时,运用累积风险模型计算不同诊断年龄、国际妇产科联盟(International Federation of Gynecology and Obstetrics,FIGO)分期、淋巴结转移程度及放疗情况下兴趣事件的累积发生率,组间比较采用Gray检验;然后,运用Fine-Gray模型进行多因素回归分析。结果 :共有585例宫颈癌患者纳入分析,其中50例患者发生兴趣事件,43例患者发生竞争事件。经Gray检验发现,在不同初诊年龄、FIGO分期、淋巴结转移程度和放疗情况下兴趣事件的累积发生率差异均有统计学意义(P值均<0.001);多因素回归分析显示,在Fine-Gray模型中,FIGO分期与淋巴结转移程度是宫颈癌患者预后的独立危险因素(P值均<0.05)。在竞争风险存在时,原位癌(0期)、Ⅱ期和Ⅲ期患者死于宫颈癌的相对危险度是Ⅰ期患者的0.022倍、2.34倍和3.57倍,而区域(近端)淋巴结转移和远端淋巴结转移的患者与无淋巴转移的患者相比死于宫颈癌的风险更高(相对危险度分别为2.497和3.518)。结论 :竞争风险模型下,宫颈癌患者的FIGO分期越高,淋巴转移越严重,则预后越差。当存在竞争风险时,应合理运用分析方法和模型来进行数据分析。

Objective： To explore the prognostic factors of cervical cancer patients under the condition of competing risks. Methods： The data of Chinese cervical cancer patients （1988-2008） were collected from Surveillance, Epidemiology and End Results （SEER）database. It was marked as interesting event that the patients died of cervical cancer, and the competing event was defined as death from other reasons （such as myocardial infarction, other cancers, etc.）. In considering the competing risks, the cumulative risk model was used to estimate cumulative incidence of interesting event according to age at diagnosis, stage by International Federation of Gynecology and Obstetrics （FIGO） criteria, degree of lymph node metastasis, and radiotherapy or not. The differences between the two groups were compared by Gray test, and the multivariate regression analysis was performed by Fine-Gray model.Results： A total of 585 cases of cervical cancer were included in this analysis, in which 50 cases occurred interesting event （dead of cervical cancer）, and 43 cases occurred competing events （dead from other diseases）. The results of Gray test showed that cumulative incidence rates of interesting events were statistically different among different ages at diagnosis, FIGO stages and lymph node metastasis degrees, as well as radiotherapy or not （all P 〈 0.001）. The multivariate regression analysis showed that FIGO stage and lymph node metastasis degree were independent risks of the prognosis of cervical cancer in Fine-Gray model （both P 〈 0.05）. In the presence of competing risks, the relative risk of death of cervical cancer in patients at stage 0 （carcinoma in situ）, stage II and stage III were 0.022, 2.34 and 3.57 times of stage I ; while the risks of death of cervical cancer in the patients with regional metastatic lymph nodes （proximal） （relative risk = 2.497） and distant metastatic lymph nodes （relative risk = 3.51 8）were higher than that in the patients without lymph node metastasis.Conclusion： In consideration of competing risks, the patients of higher FIGO stage and higher degree of lymph node metastasis have poorer prognosis. In the presence of competing risks, more reasonable analysis method and model should be used to analyze the data.