Prognostic impact of lymph node count features in total laryngectomy for advanced laryngeal squamous cell carcinoma.

INTRODUCTION

The optimal surgical management in advanced laryngeal squamous cell carcinoma (ALSCC) is still under debate. The extent of neck dissection as well as the nodal involvement affect survival metrics in head and neck cancer (HNSCC) patients. Despite pN status, other parameters like nodal yield (NY) or lymph node ratio (LNR) have been investigated before. There are data showing that log odds of positive lymph nodes (LODDS) are a good survival prognosticator in HNSCC in general but specific data on ALSCC is missing. This study aims to assess the prognostic value of lymph node count features on survival in ALSCC.

METHODS

We conducted a retrospective patient chart review on curative intent laryngectomy and bilateral neck dissection for ALSCC between 2009 and 2024 at a tertiary care center. Investigated lymph node count features besides NY included lymph node burden (LNB = number of positive lymph nodes), LNR (= LNB/NY) and the LODDS = log ((LNB + 0.5) / (NY-LNB + 0.5)). Univariate survival analysis was performed using the log-rank testing and Kaplan-Meier curves. The R maxstat package was utilized for an optimized cut-off point determination for cohort risk stratification.

RESULTS

We included 56 patients who underwent laryngectomy and bilateral neck dissection in our department. Survival analysis revealed a 5-year OS of 51% and median OS of 60.7 months. The LODDS ranged from -2.48 to 0.37 with a mean value of -1.68 ± 0.50. The cut-off at -1.94 for LODDS showed a 5-year DSF of 33% vs. 61% with a HR of 0.27 (p = 0.005) and the optimized cut-off of -1.55 showed significant differences in 5-year OS (69 vs. 17%, HR: 0.29, p = 0.003). LODDS indicated the highest concordance indices for both DFS and OS compared to LNB and LNR.

CONCLUSION

We propose LODDS to serve as a superior prognosticator compared to LNB and LNR concerning DFS and OS in TL for ALSCC. LODDS values of -1.94 for DFS and -1.55 for OS appear as suitable thresholds for risk stratification.