Validation of 8th edition AJCC staging system of small cell lung cancer in a North American cohort
Date
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
Background The TNM staging system is a critical tool for predicting cancer prognosis. The 8th edition of this system, developed by the Union for International Cancer Control (UICC) and the American Joint Committee on Cancer (AJCC) for non-small cell lung cancer, was adopted for small cell lung cancer (SCLC) without modifications. The AJCC/ UICC staging committee acknowledged that the dataset used for the 8th edition staging system had limited global representation and some inconsistencies in nomenclature and treatment. Therefore, they recommended further validation of the 8th edition for SCLC in independent cohorts. The objective of this thesis study was to compare the prognostic accuracy of the 8th editions of the small-cell lung cancer (SCLC) staging system by UICC/AJCC in a large North American cohort with the 7th editions. Methods This retrospective cohort study used the North American Surveillance, Epidemiology, and End Results Registries (SEER) research plus a database with custom fields to extract eligible cases submitted by November 2022. The study included all histologically/cytologically diagnosed small-cell lung cancer (SCLC) patients from 2018 to 2020, staged according to the 8th edition of the system in the SEER database, based on predetermined eligibility and exclusion criteria. Data on demographic, disease characteristics, treatments, and outcomes were collected. Kaplan-Meier curves, Cox regression models and time-dependent receiver operating characteristics (ROC) were used to compare the discriminatory ability and prognostic performance of the 7th and 8th editions. Overall survival were estimated using the Kaplan-Meier method, with the differences assessed using a log-rank test. Cox regression were used to examine the factors associated with overall survival, including the formal comparisons between adjacent stage categories. The C-index were used to evaluate the overall performance of the 7th and 8th edition staging systems. Results A cohort of 11,212 cases of SCLC diagnosed between 2018 to 2020 meeting inclusion and exclusion criteria was identified. Of these, 11,192 (99.82%) were recorded as 8041/3 (small cell carcinoma, NOS), and the remaining 20 cases (0.18%) were recoded as oat cell carcinoma. In this dataset only 419 (3.74%) patients were treated surgically, 5,316 (47.41%) cases received external beam radiation treatment, and 8,090 (72.15%) cases received systemic treatment. Both unadjusted and adjusted model (adjusted for age, sex and race) based on 8th edition staging indicated that a higher M stage was associated with worse overall survival (p<0.001) and worse hazard compared to baseline M stage The pairwise hazard ratio for M1a versus M0 was 2.44 (95% CI 2.24 vs 2.67), and M1c versus M1b was 1.28 (95% CI 1.20 vs 1.37), suggesting good discrimination between these stages. Harrell’s concordance score was 0.6126 for the 7th edition M stage and 0.6211 for the 8th edition M stage, suggesting a slightly better or similar fit for the 8th edition. In model adjusted for age, sex and race, the time-dependent AUC for the 7th edition M stage was 0.6962, compared to 0.6985 for the 8th edition M stage suggesting a similar model fit for both editions. Both unadjusted and adjusted model (adjusted for age, sex and race) based on the8th edition indicated that higher stage groups predicted worse OS (p=<0.001). Except for stage group 2B, hazards progressive worsened with higher stage groups. Harrell’s concordance score was 0.6239 for the 7th edition stage group and 0.6301 for the 8th edition stage group. In model adjusted for age, sex and race, the time-dependent AUC for the 7th edition stage group was 0.7065 and the AUC for the 8th edition stage group was 0.7096, suggesting a comparable model fit, with 8th edition being marginally better. Significance The present study provides an independent validation and evaluation of the 8th edition of the staging system for small-cell lung cancer in a North American cohort. These findings showed that the 8th edition offers better discrimination for various M stages and overall stage groups. It showed a marginally better C-index and time-dependent AUC compared to previous edition. This external validation reinforces the credibility of the updated staging system and its enhanced utility in managing SCLC. We confirmed that the 8th edition M1b and M1c show significant discrimination from each other, with M1c stage indicating a worse prognosis. This study provides supportive evidence for these changes. Additionally, we found significant discrimination between higher stage groups and a slightly better fit for 8th edition stage group compared to the 7th edition, further supporting these changes. However, there was no significant difference in the prognostic ability of the early stages 1A, 1B and 2A. This highlights the need for further evaluation of these early stages in independent cohorts, which could aid in future refinement to the staging system. The present study provides supportive evidence for 8th edition SCLC changes and evidence that could be useful for future editions of staging system revision in SCLC.