High lymph node ratio adds prognostic information beyond extranodal extension in oral cavity cancer

In 661 node-positive oral cavity cancer patients, lymph node ratio ≥0.10 independently predicted worse survival, with the highest risk when extranodal extension coexisted.

KEY POINTS

  • This retrospective secondary analysis used 661 node-positive patients from the randomized phase III OCAT trial (NCT00193843), conducted at Tata Memorial Centre, Mumbai. The parent trial compared postoperative 60 Gy in 30 fractions over six weeks, accelerated RT delivering the same dose over five weeks, and 60 Gy with concurrent cisplatin-based chemoradiotherapy. Patients were randomized across the three treatment arms, but lymph node ratio (LNR) itself was analyzed retrospectively rather than through a randomized biomarker-guided strategy.
  • LNR was defined as the number of metastatic lymph nodes divided by the total number dissected, with ≥0.10 prespecified as high LNR. Among the 661 patients, 250 (37.8%) had high LNR, while 487 (73.7%) had extranodal extension (ENE). Median dissected lymph node count was 31, with a median of two positive nodes. The cohort was predominantly high risk, with 91.1% having pT3–4 disease, and median follow-up reached 91 months.
  • High LNR independently predicted poorer outcomes after adjustment for ENE and established pathological factors. Multivariable analyses showed worse overall survival (HR 1.43, 95% CI 1.15–1.78; p=0.001), recurrence-free survival (HR 1.48, 95% CI 1.18–1.85; p=0.001), and locoregional control (HR 1.40, 95% CI 1.06–1.85; p=0.018). ENE remained independently prognostic for all three endpoints, indicating that quantitative nodal burden and extranodal invasion provide complementary prognostic information.
  • Combined LNR and ENE stratification identified markedly different five-year survival rates. Five-year OS was 55.1% for low LNR/ENE-negative, 45.1% for low LNR/ENE-positive, 52.5% for high LNR/ENE-negative, and 30.8% for high LNR/ENE-positive disease. Compared with low LNR/ENE-negative patients, the high LNR/ENE-positive group had HR 2.08 (95% CI 1.47–2.95; p<0.001). Importantly, high LNR without ENE was not significantly associated with worse survival in this stratified comparison (HR 1.12; p=0.65).
  • Adding LNR improved prognostic model discrimination only modestly. Harrell's C-index increased from 0.618 to 0.633, an absolute improvement of 0.015. Likelihood-ratio testing supported improved model fit (χ²=10.34; p=0.0015), and the optimism-corrected C-index was 0.621. However, decision-curve analysis showed largely overlapping net-benefit curves, suggesting that statistical improvement does not necessarily translate into substantial clinical decision-making benefit.
  • The prognostic association was consistent across alternative thresholds and adjuvant treatments. Sensitivity analyses supported LNR cutoffs of 0.07, 0.10, and 0.15, with 0.10 providing the strongest discrimination in this cohort. No significant interaction was identified between LNR and randomized treatment arm (p=0.43), and the formal LNR-by-ENE interaction was also nonsignificant. Therefore, the combined groups provide prognostic stratification rather than evidence that high LNR predicts benefit from intensified chemoradiotherapy.
  • Clinical implementation requires external validation. The analysis was retrospective, despite prospectively collected trial data, and used AJCC seventh-edition staging. The unusually high ENE prevalence and predominantly advanced oral cavity tumors may limit generalizability to lower-risk populations. No biomarker-directed treatment allocation was tested, and the study does not establish that increasing RT dose, changing elective nodal volumes, or adding systemic therapy improves outcomes specifically in patients with high LNR.

CLINICAL TAKEAWAY

Lymph node ratio provides additional prognostic information beyond extranodal extension and conventional staging after surgery for node-positive oral cavity squamous cell carcinoma. Combining LNR ≥0.10 with ENE identifies a particularly high-risk subgroup, although improvement in overall model discrimination is modest. LNR may refine postoperative counseling and trial stratification, but is not yet a validated basis for changing adjuvant treatment intensity.

SOURCE

Radiotherapy and Oncology

Browse more research Suggest a correction