The extent of lymph node dissection is not associated with disease-free survival following lobar or sublobar resection: Results from Cancer and Leukemia Group B 140503 (Alliance)

Study Design and Participants

Cancer and Leukemia Group B (CALGB, now Alliance for Clinical Trials) 140,503 was a multicenter, international, randomized, noninferiority phase III trial in patients with NSCLC 2 cm or less clinically staged as T1aN0 based on the seventh edition of the TNM staging system. Patients were registered for the trial if they met the preoperative eligibility criteria and randomized after meeting intraoperative eligibility criteria. Preoperative eligibility criteria were previously reported.⁷ Briefly, patients were registered for the trial if they had on computed tomography (CT) scanning a peripheral lung nodule with a solid component 2 cm or less in size that was presumed or confirmed to be NSCLC. Patients with pure ground-glass opacities or pathologically confirmed N1 or N2 disease were not eligible. Intraoperative eligibility criteria included histological confirmation of NSCLC (if not already obtained) and confirmation of N0 status by frozen-section examination (for right-sided tumors: node levels 4, 7, and 10; for left-sided tumors: node levels 5 or 6, 7, and 10). Nodes previously sampled by mediastinoscopy or endobronchial ultrasound within 6 weeks of the definitive surgical procedure did not need to be resampled.

Trial Oversight

This trial was registered with ClinicalTrials.gov, Number NCT00499330. It was conducted according to the principles of the Declaration of Helsinki and the International Council for Harmonisation Good Clinical Practice guidelines. The National Cancer Institute Central Institutional Review Board and the Institutional Review Board at each participating institution approved the protocol (#0709009395, date: February 11, 2025). All patients provided written informed consent before trial enrollment. Since its activation, CALGB/Alliance 140,503 was monitored by the Alliance Data and Safety Monitoring Board (DSMB) twice annually. The first and third authors (N.A. and X.W.) developed the trial design, and for the current analysis, the first 3 authors had full access to the raw data and analyzed the data.

The Alliance Statistical and Data Management Center and the study chairperson reviewed the data following Alliance policies to ensure data quality. Based on interim analyses conducted up to November 2021 and a validation analysis in March 2022, the Alliance DSMB unanimously recommended releasing the data and terminating further DSMB monitoring of the trial.

Randomization and Procedures

Eligible patients were preregistered for the trial using the OPEN registration system, a web-based system for patients’ enrollment into National Cancer Institute–sponsored cooperative group clinical trials. After intraoperative eligibility was confirmed, as described above, patients underwent randomization (1:1) to LR or SLR, based on a permuted-block randomization scheme with stratification for radiographic tumor size (<1 cm, 1-1.5 cm vs > 1.5-2.0 cm), histology (squamous cell carcinoma, adenocarcinoma, and other), and smoking status (never, former, and current). The type of SLR (wedge resection or segmentectomy) and the choice of surgical approach (thoracotomy vs video-assisted thoracoscopic surgery or robotic-assisted surgery) was at the surgeon’s discretion. After surgery, patients underwent CT scans every 6 months for 1 year and then annually for a minimum of 5 years. In a subsequent trial amendment, surveillance imaging was changed to CT scans every 6 months for 2 years and then annually for 5 additional years (7 years after surgery).

Extent of Lymphadenectomy

The extent of LAD in CALGB 140503, beyond the simple sampling (S) required for randomization, was left to the discretion of the surgeon. LAD was categorized in the trial’s case report forms (CRFs) as S, systematic sampling (SS), or complete lymph node dissection (CLND). S was defined as sampling 2 N2 stations and 1 N1 station (the minimum requirement for randomization). SS refers to sampling more than 2 N2 stations and 1 N1 station. Complete (radical) mediastinal node dissection is a near-complete excision of all nodal tissue in the upper, middle, and lower mediastinal stations. For this report, the extent of LAD in individual patients was extracted from the CRFs as entered by the surgeons or their designees. Source documents were not reviewed. The number of N2 and N1 nodal stations dissected in each case was also abstracted from the CRFs.

End Points

The primary end point of this post hoc analysis was to determine the impact of the extent of LAD on DFS. Secondary end points included the determination of intergroup differences in OS, RFS, and the incidence of disease recurrence. DFS was defined as the time between randomization and disease recurrence or death of all causes, whichever came first. Patients who did not experience a DFS event were censored at the time of the last follow-up. OS was defined as the time between randomization and death from any cause. Patients who were alive were censored at the time of the last follow-up. RFS was defined as the time between randomization and any disease recurrence or death from the index lung cancer. Patients who did not experience recurrence were censored at the time of death or the time of the last clinical follow-up. We have previously reported the incidence of locoregional recurrence based on the extent of parenchymal resection. In this article, we report the incidence of nodal recurrence, including isolated hilar nodal recurrence, mediastinal nodal recurrence with or without associated hilar disease, supraclavicular nodal recurrence, and any systemic recurrence. Major surgical complications graded according to the Common Terminology Criteria for Adverse Events version 4.0 were abstracted for all 3 groups from the case report forms.

Statistical Analysis

This post hoc analysis of CALGB 140503 was focused on differences in survival and recurrence based on the extent of LAD. Baseline characteristics were compared using the Kruskal–Wallis test for continuous variables, chi-square test for most categorical variables, and Fisher exact test for N-stage due to low (<5) expected counts. Survival end points were characterized using the Kaplan–Meier estimator and compared by the stratified log-rank test, with tumor size, histology, and smoking status as stratification factors. Between-group differences in incidence of recurrence rates were compared by the chi-square test. Multivariable analyses were performed using Cox proportional hazards regression models to evaluate the effects of baseline predictors on RFS. Hazard ratios (HRs) were compiled with 95% CIs.

All reported P values are 2-sided and not adjusted for multiplicity. All statistical analyses were conducted by the Alliance study statisticians and statistical programmers, and the data were locked down on June 21, 2022. Data management and statistical analysis were done in SAS 9.4, and graphs were generated in R 4.4.2.

Disease-Free Survival

After a median follow-up of 7 years, DFS, the primary end point of this study, was similar across all 3 groups of LAD. Five-year DFS was 62.3% after CLND, 65.7% after SS, and 61.2% after S (Figure 2, A). Additionally, when the SS and S groups were merged, there was no difference in 5-year DFS between patients in the merged group compared with those who underwent CLND (Figure 2, B). Given these results, all subsequent DFS analyses were performed using the merged SS and S groups, which are subsequently collectively referred to as the nodal sampling group. There was no difference in DFS between the LR and SLR arms of the trial based on the extent of LAD (Figure 2, C and D). Five-year DFS among patients who had CLND was 65.7% (95% CI, 56.4-76.6) after LR and 58.5% (95% CI, 48.2-71.1) after SLR (P = .530). Five-year DFS in patients who had S/SS was 63.5% (95% CI, 57.6-70.0) after LR and 65.1% (95% CI, 59.2-71.6) after SLR.

Additionally, within each arm of the trial (LR and SLR), there was no difference in DFS based on the extent of nodal dissection. After LR, 5-year DFS was 65.7% after CLND and 63.5% after nodal sampling (SS + S) (Figure 2, E). Likewise, after SLR, 5-year DFS was not significantly different between patients who had CLND and those in the nodal sampling group (Figure 2, F). For OS, all analyses mirrored those for DFS (Figure E1, A-F).

Recurrence-Free Survival

For the entire cohort, RFS was similar across the 3 LAD groups (Figure 3, A). Five-year RFS was 70.9% after CNLD, 72.8% after SS, and 65.4% after S (stratified log-rank P = .158). Additionally, when the SS and S groups were merged, there was no difference in 5-year RFS between patients in the merged sampling groups and those who underwent CLND (Figure 3, B). Given these results, all subsequent RFS analyses were performed using the merged SS and S groups. There was no difference in RFS after CLND or nodal sampling based on the extent of parenchymal resection (Figure 3, C and D). Additionally, after LR, the 5-year RFS was 72.5% after CLND and 70.8% after nodal sampling (Figure 3, E). Five-year RFS after SLR was 68.9% after CLND and 70.2% after nodal sampling (Figure 3, F). Among patients who received SLR and CLND, 5-year RFS was 69.9% after wedge resection and 67.4% after segmentectomy (Figure 4, A). After nodal sampling, the 5-year RFS was 69.1% for wedge resection and 68.7% for segmentectomy (Figure 4, B).

In a multivariable exploratory analysis of variables potentially associated with RFS, including age, sex, tumor location, histologic type, smoking history, tumor size, and Eastern Cooperative Oncology Group performance status score, only a history of current smoking was significantly associated with worse RFS (HR 1.49 P = .009) (Figure 5). Notably, neither the extent of LAD nor the extent of parenchymal resection was a significant predictor of RFS. However, there was a trend toward worse RFS associated with tumors larger than 1.5 cm (HR, 1.33, P = .052).

Recurrence Patterns

Table 2 shows recurrence patterns across all 3 groups of LAD. The majority of recurrences were systemic. There was no difference between groups in the incidence of isolated hilar nodal recurrence, mediastinal nodal recurrence with or without hilar disease, and supraclavicular nodal recurrence. Isolated hilar nodal recurrence occurred in 7.7% of patients after CLND, 4% of patients after SS, and 5.7% of patients after S (P = .221). Mediastinal nodal recurrence with or without hilar recurrence occurred in 7.7% of patients after CLND, 8.3% of patients after SS, and 7% of patients after S (P = .828)

Limitations

This study has several limitations. First, the comparative effectiveness of nodal sampling versus dissection was not a prespecified end point of the CALGB 140503 trial. Therefore, the results of this retrospective analysis may be confounded by bias from hidden, unrecognized variables. Second, the protocol did not specifically describe the technical aspects of nodal dissection or sampling beyond what was required for randomization. Given the predictable heterogeneity in surgical practices, we are not able to definitively allocate patients to sampling or dissection beyond the surgeon’s attestation to that effect in the case report forms. Notwithstanding these limitations, the study results add to an accumulating body of evidence from randomized trials in lung cancer and other solid tumors including melanoma, breast, colorectal, and early gastric cancers,^3,4,14-17 showing that in selected cohorts of patients with a high likelihood of absence of nodal metastases, extended resection of negative nodes is not associated with improved survival or reduced recurrence of disease.