Re: Cheng YF, Hung WH, Chen HC, et al. Comparison of Treatment Strategies for Patients With Clinical Stage T1–3/N2 Lung Cancer. J Natl Compr Canc Netw 2020;18(2):143–150.
We would like to comment on the recent published article by Cheng et al1 entitled “Comparison of Treatment Strategies for Patients With Clinical Stage T1-3/N2 Lung Cancer.” Cheng et al compared the survival of patients with cT1–3N2 lung cancer who underwent different treatment strategies. They found that multimodal treatments tended to provide better 5-year overall survival than surgery alone or concurrent chemoradiotherapy, in which lobectomy was considered as the optimal types of resection. However, we have some concerns.
First, as shown in Figure 2 and 3, there are intersections between Kaplan-Meier curves. For instance, therapy group 2 and 3 overlapped in Figure 2A–C, and pneumonectomy and lobectomy as well as others overlapped in Figure 3. Therefore, a time-dependent Cox regression model should have been used2,3 rather than the routine Cox proportional hazards regression model. The authors should clarify whether the results were changed in the time-dependent model.
Second, as shown in Figure 3, patients with cT2N2 or cT3N2 disease who underwent surgery accounted for a minority in the entire cohort. Because the treatment strategies were heterogeneous (there were 5 different treatment modalities), it is obvious that different types of resection can hardly have comparability once subgroup analysis was made. In other words, the authors could not answer whether lobectomy was always the optimal choice for patients with lung cancer with the same clinical stage who underwent the same treatment modality. Additionally, there was no information concerning surgical margins and pulmonary function in the SEER database from 2010 through 2015,4 which is a major limitation to comparing surgery type in this study.
Lastly, the authors presented results of a multivariate analysis in Table 3, which included both surgery types and therapy groups. However, there could be collinearity between different surgical procedures and different treatment modality as covariates. For instance, for patients with cT1N2 lung cancer who underwent neoadjuvant chemoradiotherapy and experienced a complete response, sublobar resection may be performed; however, for those who did not undergo neoadjuvant therapy, lobectomy was routinely performed. The authors should address this issue to make readers better understand the results.
In summary, there are several inherent limitations of this study. The authors should make more explanation concerning the aforementioned problems.
Cheng YF, Hung WH, Chen HC, et al. Comparison of treatment strategies for patients with clinical stage T1-3/N2 lung cancer. J Natl Compr Canc Netw 2020;18:143–150.
Fisher LD, Lin DY, Time-dependent covariates in the Cox proportional-hazards regression model. Annu Rev Public Health 1999;20;145–157.
Moolgavkar SH, Chang ET, Watson HN, Lau EC. An assessment of the Cox proportional hazards regression model for epidemiologic studies. Risk Anal 2018;38:777–794.
National Cancer Institute. Surveillance, Epidemiology, and End Results. Radiation/Chemotherapy Databases (1975–2016). Accessed July 6, 2020. Available at: https://seer.cancer.gov/data/treatment.html