Development and validation of a prognostic predictive model of pulmonary spindle cell carcinoma from the surveillance, epidemiology and end results database

Abstract

Background: Pulmonary spindle cell carcinoma (PSCC) is a rare type of non-small cell lung cancer (NSCLC). The prognostic influent factors and therapeutic methods of PSCC are unclear, for there are only some case reports or small samples' analysis. This study aims to find prognosis related factors of PSCC, develop and validate a nomogram to predict their survival probability.

Methods: The Surveillance, Epidemiology, and End Results (SEER) 18 Registries database (2000-2018) was searched to study PSCC. According to diagnosed time, data was divided into primary cohort (2000-2015) and validation cohort (2016-2018), both followed until December 31 2018. Chosen by Least Absolute Shrinkage and Selection Operator (LASSO) regression, age, sex, stage, surgery, chemotherapy, N, size and history of malignancy were taken out as predictive variables. The primary cohort was used to develop a nomogram to predict 1-, 3- and 5-year overall survival (OS) probability, and be validated by the validation cohort using concordance index (C-index) and calibration curves. Both cohorts were used to conduct a Cox regression to find the influential factors on OS of PSCC.

Results: The nomogram shows a good concordance and discrimination on the prediction of OS, both internal (n=457 and C-index is 0.79) and external validation (n=100 and C-index is 0.76). The median survival time of PSCC is 4 months, with 20.1% OS possibility in 5 years. Multivariate analysis identified patients of older age [hazard ratio (HR), 1.02; 95% confidence interval (CI): 1.01-1.04], larger size of neoplasm (HR, 1.01; 95% CI: 1.01-1.01), M1 (HR, 2.96; 95% CI: 2.17-4.04), N2 (HR, 2.55; 95% CI: 1.81-3.59) or N3 (HR, 2.99; 95% CI: 1.58-5.66), regional stages (HR, 2.11; 95% CI: 1.29-3.44) and distant stages (HR, 6.17; 95% CI: 3.83-9.94) had a lower OS possibility, while surgery (HR, 0.39; 95% CI: 0.28-0.53) and history of malignancy (HR, 0.68; 95% CI: 0.48-0.98) was protective factors for PSCC. PSCC survived longer with surgery performed instead of chemotherapy or radiotherapy.

Conclusions: Patients of PSCC have a poor prognosis, and using the nomogram developed by this study can predict their 1-, 3- and 5-year OS probability. Surgery is a better choice for PSCC and more studies are necessary to find potential treatment like targeted therapy, programmed death-1 (PD-1) and programmed death ligand 1 (PD-L1).

Keywords: Pulmonary spindle cell carcinoma (PSCC); Surveillance, Epidemiology, and End Results (SEER); prognosis; pulmonary sarcomatoid carcinoma (PSC).

Figure 1

Flow chart of nomogram developed and comparing the overall survival probabilities of pulmonary sarcomatoid carcinoma. SEER, The Surveillance, Epidemiology, and End Results; PSCC, pulmonary spindle cell carcinoma; LASSO, Least Absolute Shrinkage and Selection Operator.

Figure 2

The survival analysis of PSC. (A) The Kaplan-Meier plot shows different OS of PSC, P<0.0001. (B) The forest plot shows hazard ratio among PSC. OS, overall survival; PSC, pulmonary sarcomatoid carcinoma.

Figure 3

Nomogram to predict the 1-, 3- and 5-year OS of patients with PSCC. OS, overall survival; PSCC, pulmonary spindle cell carcinoma.

Figure 4

C-index trend of nomogram with 1,000 bootstrap samples from primary cohort. C-index, concordance index.

Figure 5

Calibration curves both primary cohort and validation cohorts. (A) Calibration curve of nomogram of 1-year OS in PSCC primary cohort. (B) Calibration curve of nomogram of 3-year OS in PSCC primary cohort. (C) Calibration curve of nomogram of 5-year OS in PSCC primary cohort. (D) Calibration curve of nomogram of 1-year OS in PSCC validation cohort. OS, overall survival; PSCC, pulmonary spindle cell carcinoma.

Figure 6

Multivariate analysis of pulmonary spindle cell carcinoma with Cox regression. *, P<0.05.

Figure 7

K-M curves of PSCC with or without therapies. (A) K-M curve of all PSCC patients. (B) K-M curve of PSCC patients based on whether received chemotherapy or not. (C) K-M curve of PSCC patients based on whether received radiotherapy or not. (D) K-M curve of PSCC patients based on whether received surgery or not. K-M curve, Kaplan-Meier curve; PSCC, pulmonary spindle cell carcinoma.