Nomogram for Prediction of Survival of Postoperative Small Cell Lung Cancer Patients:An Analysis Based on SEER

Hui Pan,Yalei Zhang,Dakai Xiao,Zhihua Guo,Jinxin Zhang,Jiaxi He
DOI: https://doi.org/10.3969/j.issn.1674-0904.2019.06.008
2019-01-01
Abstract:To identify potential prognostic factors of postoperative small cell lung cancer (SCLC) patients and establish an effective nomogram for prediction of survival outcomes. Methods: Patients who had been diagnosed as SCLC from 2004 to 2012 in the Surveillance, Epidemiology, and End Result database ( SEER) were identified and collected. Kaplan-Meier method was used to estimate the overall survival (OS) in the non-surgery group and the surgery group. Cox regression was performed to identify independ-ent prognostic factors. Four models were established using different staging system and compared by concordance index (C-index) and calibration curve. Two methods were utilized to conduct the comparison. Results: A total of 45226 SCLC patients were identified. Af-ter applying the exclusion criteria, 867 postoperative patients were included and analyzed. In multivariate analysis, prognostic factors were age, sex, surgery, radiation sequence, tumor diameter, tumor extension, T stage, N stage, number of lymph node examination, grade of pathological differentiation and metastasis, respectively. Independent covariates were selected using Akaike's information cri-terion. Nomogram was then formulated based on results of multivariate analysis. C-index of the nomogram incorporating tumor size and extension was 0. 706, which was higher than other conventional classifications such as AJCC TNM Classification (0. 700), VALSG (0. 665) and IASLC (0. 667) staging systems. Validation of the nomogram demonstrated that it had an ideal predictive accuracy. Conclusion: Operation is associated with a better survival of SCLC patients. Tumor size and extension are important independent prog-nostic factors. Nomogram incorporating tumor size, extension and other variables are ideal prognostic prediction tools for OS of postop-erative SCLC patients, which has better predictive accuracy than conventional classifications.
What problem does this paper attempt to address?