肺大细胞神经内分泌癌临床特征分析及预后模型建立

Clinical characteristics of pulmonary large cell neuroendocrine carcinoma and a nomogram model for predicting its prognosis

目的基于监测、流行病学和结果数据库(Surveillance, Epidemiology, and End Results,SEER),分析肺大细胞神经内分泌癌(pulmonary large cell neuroendocrine carcinoma, PLCNEC)的临床特征和患者生存情况,建立预后模型。方法从SEER数据库中收集2004-2013年诊断为PLCNEC的患者1 656例,描述其基本临床特征,通过Kaplan-Meier法绘制生存曲线并进行生存分析。使用Cox比例风险模型探索与总生存期(overall survival,OS)相关的临床特征,并建立可视化的预测患者预后的列线图(nomogram)模型,通过重复抽样对模型进行验证。结果 1 656例PLCNEC患者中,男性(55.37%)和诊断年龄≥65岁(52.05%)居多,以白种人(84.12%)为主;40.94%做过手术治疗,37.80%做过放疗,51.93%做过化疗。生存分析发现:患者中位生存期为11个月,1年生存率为46.5%。Cox比例风险模型多因素分析发现:女性、黑种人及其他种族、手术治疗、放疗及化疗为保护因素,而年龄≥65岁,肿瘤长径增大,肿瘤向邻近组织、区域淋巴结及远处扩散为危害因素,肿瘤部位对生存影响无明显差异。利用上述与OS相关的临床特征,建立可视化列线图模型,可通过基本临床特征来预测PLCNEC患者的1年生存率。重复抽样验证发现该模型1年生存率预测值与实际值一致性较好(C-index为0.76)。结论 PLCNEC患者中位OS为11个月,1年生存率为46.5%;性别、诊断年龄、种族、基线肿瘤长径、肿瘤扩散、手术治疗、放射治疗及化疗等临床特征与PLCNEC生存预后相关;成功建立可视化列线图模型,可预测PLCNEC患者1年生存率。

Objective To analyze the clinical characteristics and survival of patients with pulmonary large cell neuroendocrine carcinoma(PLCNEC) and construct a nomogram model for predicting its prognosis. Methods We searched the Surveillance, Epidemiology, and End Results(SEER) database to collect the data of patients diagnosed with PLCNEC between 2004 and 2013. Kaplan-Meier method was used for survival analysis of the patients. A Cox proportional hazard model was used to identify the factors affecting the overall survival(OS), and the significant factors were used to construct a nomogram model for predicting the patients’ 1-year survival probability. The discrimination power of this model was assessed based on the C-index, and the calibration was evaluated with a bootstrap procedure. Results A total of 1 656 cases of PLCNEC with complete record of the clinical characteristics were collected. Of these patients, 55.37% were male, 52.05% were above 65 years of age at diagnosis, and 84.12% were white;51.93% of the patients received chemotherapy, 40.94% had surgeries, and 37.80% had radiotherapy. Kaplan-Meier survival analysis showed that the median OS of patients with PLCNEC was 11 months, and their 1-year survival rate was 46.5%. Multivariate analysis identified a female gender, black race, surgery, radiotherapy and chemotherapy as the protective factors, while an age ≥65 years, an increased tumor size, invasion of adjacent tissues, lymph node involvement and distant metastasis were the risk factors for poor prognosis;the primary tumor sites did not significantly affect the survival outcomes of the patients. We established a nomogram model for predicting the 1-year survival probability of the patients based on the significant factors identified in the multivariate analysis. The C-index of this prediction model was 0.76, and assessment of calibration showed that the 1-year survival rate predicted by this model was highly consistent with the observed 1-year survival rate of the patients. Conclusion According to the data from SEER, patients with PLCNEC have a median OS of 11 months and a 1-year survival rate of 46.5%. Gender, age, race, tumor size, tumor growth and metastasis and interventions are all associated with the OS of the patients. The nomogram model we established based on the clinical characteristics of PLCNEC is capable of predicting the 1-year survival probability of the patients.