Lymph Node Ratio Enhances Predictive Value for Treatment Outcomes in Patients with Non-Small Cell Lung Cancer Undergoing Surgery: A Retrospective Cohort Study

Lung cancer is a leading cause of cancer-related mortality worldwide, with non-small cell lung cancer (NSCLC) accounting for about 85% of cases. Radical resection with lymph node dissection is the standard of care for resectable NSCLC, and pathologic nodal (pN) stage is a major prognostic factor. However, the optimal extent of lymph node evaluation remains debated across guidelines (e.g., IASLC/ESTS recommend at least 6 examined lymph nodes, while Chinese guidelines suggest at least 12). Prior studies have linked a higher number of examined lymph nodes with improved survival, yet surgical quality can affect the adequacy of nodal staging. The lymph node ratio (LNR), the ratio of metastatic to examined lymph nodes, has shown prognostic value in multiple cancers. This study investigates whether LNR provides superior prognostic stratification compared to conventional pN staging in surgically treated NSCLC and proposes a revised pN (r-pN) classification that integrates both measures.

The paper contextualizes variability in recommended lymph node evaluation (IASLC/ESTS ≥6 nodes vs. Chinese guidelines ≥12) and conflicting thresholds for adequate nodal assessment (10 or 16 nodes) reported in prior studies. Evidence indicates more examined lymph nodes correlate with better prognosis, but discrepancies in surgical quality may impair the accuracy of pN staging. LNR has been validated as a prognostic factor in colorectal, breast, and gastric cancers and has emerging support in NSCLC. These prior findings motivate evaluating LNR against pN and exploring a combined classification to better capture nodal disease burden and examination adequacy.

Data source: SEER databases (SEER*Stat v8.3.6) with de-identified, population-based cancer registry data from the United States. Study design: Retrospective cohort. Population: NSCLC patients diagnosed 2004–2015 meeting inclusion criteria: histologically confirmed adenocarcinoma or squamous cell carcinoma; age ≥18; initial therapy surgery; lymph node examination performed; definitive TNM staging per AJCC 7th edition; non-M1 and non-N0; no chemotherapy or radiotherapy. Variables extracted: age, sex, race, primary tumor site, grade, histology, T stage, N stage, number of resected lymph nodes, number of positive lymph nodes. Endpoints: Primary—cancer-specific survival (CSS), from diagnosis to death due to lung cancer; Secondary—overall survival (OS), from diagnosis to death from any cause. LNR stratification: LNR calculated as positive nodes divided by examined nodes; optimal cut-offs determined using X-tile software, which evaluates all possible divisions with log-rank statistics, selecting the split with maximal chi-square. Identified cut-offs were 0.19 and 0.73, defining LNR1 (<0.19), LNR2 (0.19–0.73), LNR3 (>0.73). Revised pN (r-pN): Created by integrating LNR stages with AJCC 7th pN categories to refine nodal risk stratification. Statistical analysis: Group comparisons with chi-square or Fisher’s exact tests. Survival analyses via Kaplan–Meier with log-rank tests for overall and pairwise comparisons (Benjamini–Hochberg adjustment). Multivariable proportional hazards models used to compare predictive performance across pN, LNR, and r-pN using Harrell’s C-index (discrimination) and Akaike information criterion (AIC; calibration). Software: SPSS v26.0 and R v4.2.2. Two-sided P < 0.05 considered statistically significant.

- Cohort selection: From 383,271 initially identified cases, 7,792 surgically treated NSCLC patients met inclusion criteria. Baseline characteristics are provided for age (≤60: 28.3%), sex (female: 46.5%), race (White: 82.7%), site (upper lobe: 54.1%), grade (I/II: 51.2%), histology (adenocarcinoma: 65.7%), and T stage (T2: 56.6%). - LNR cut-offs: X-tile identified two cut-offs at 0.19 and 0.73, defining LNR1 (<0.19), LNR2 (0.19–0.73), LNR3 (>0.73). - Cancer-specific survival (CSS): Median CSS by LNR—LNR1: 71 months; LNR2: 41 months; LNR3: 17 months. Compared to LNR1, LNR2 HR 1.46 (95% CI 1.36–1.57; P < 0.001) and LNR3 HR 2.85 (95% CI 2.58–3.15; P < 0.001). By pN—pN1: 62 months; pN2: 33 months; pN3: 14 months; pN2 vs pN1 HR 1.61 (95% CI 1.51–1.73; P < 0.001), pN3 vs pN1 HR 3.21 (95% CI 2.63–3.92; P < 0.001). - Overall survival (OS): Median OS by LNR—LNR1: 50 months; LNR2: 35 months; LNR3: 16 months. Compared to LNR1, LNR2 HR 1.36 (95% CI 1.27–1.45; P < 0.001) and LNR3 HR 2.60 (95% CI 2.37–2.85; P < 0.001). By pN—pN1: 46 months; pN2: 29 months; pN3: 13 months; pN2 vs pN1 HR 1.50 (95% CI 1.41–1.60; P < 0.001), pN3 vs pN1 HR 2.81 (95% CI 2.33–3.41; P < 0.001). - Predictive performance: For CSS, LNR outperformed pN (AIC 59,919 vs. 60,040; C-index 0.648 vs. 0.640). For OS, LNR also outperformed pN (AIC 72,957 vs. 73,100; C-index 0.642 vs. 0.633). The r-pN classification showed the best performance: CSS AIC 59,814; C-index 0.656. OS AIC 72,872; C-index 0.648. - Risk stratification: r-pN effectively separated patients into low-, medium-, and high-risk groups for both CSS and OS (log-rank P < 0.001).

The study addressed the need for improved nodal prognostication in NSCLC by evaluating LNR against conventional pN staging. Because pN reflects only the anatomical location of nodal disease and is sensitive to variability in surgical and pathological assessment, it may not capture true disease burden. LNR incorporates both the extent of nodal metastasis and the thoroughness of nodal examination, thereby mitigating staging variability. Findings demonstrated that LNR better discriminates CSS and OS than pN, and integrating both measures into an r-pN system further improved model fit and discrimination. This enhanced stratification has practical implications: it may refine postoperative risk assessment and guide adjuvant therapy decisions more precisely for surgically treated NSCLC patients.

LNR is a strong, simple, and reproducible prognostic factor that outperforms conventional pN staging for predicting CSS and OS in NSCLC patients undergoing surgery. The proposed r-pN classification that integrates LNR with pN further improves prognostic accuracy and may inform more individualized adjuvant treatment strategies. Future work should include prospective validation, exploration of integration with the latest TNM editions and molecular markers, and assessment of clinical utility in guiding adjuvant therapy across diverse practice settings.

This was a retrospective analysis of SEER data, which may be subject to selection bias and unmeasured confounding. Nodal assessment quality likely varied across institutions and surgeons, potentially influencing both examined and positive node counts. Inclusion was limited to patients who underwent surgery without chemotherapy or radiotherapy, which may affect generalizability to broader NSCLC populations. TNM staging used the AJCC 7th edition, and LNR cut-offs derived via X-tile require external validation.