Prognostic Thresholds for Lymph Node Metastasis in Medullary Thyroid Cancer: A Restricted Cubic Splines Analysis

<b><i>Objectives:</i></b> The current American Joint Committee on Cancer medullary thyroid cancer (MTC) staging system qualitatively stratifies lymph node (LN) status based on involved LN compartments; however, American Thyroid Association guidelines note that quantitative assessment of LN metastases "should be incorporated." Several studies have proposed LN ratio (LNR) and number of positive LNs as prognostic parameters. We (1) assess whether there are prognostically significant LN thresholds, (2) estimate their association with MTC-specific mortality, and (3) appraise the identified thresholds using an institutional database. <b><i>Methods:</i></b> In this retrospective cohort analysis, MTC patients were abstracted from the Surveillance, Epidemiology, and End Results database (2004-20). Cox models with restricted cubic splines assessed the functional relationship of LNR and positive LN count with MTC-specific mortality. Thresholds were estimated using Markov Chain Monte Carlo and bootstrapping. Multivariable models estimated the association of the thresholds with mortality. The testing cohort comprised 149 patients with MTC at a single institution (1996-2025). <b><i>Results:</i></b> There were 2709 patients in the derivation cohort; 2098 (77.4%) had LNs examined. Mean patient age was 54.1 years, 59.1% were female, and 69.6% were non-Hispanic White. Mean tumor size was 23.5 mm; 52.7% of patients with LNs examined had ≥1 positive LN. The 5-year MTC-specific survival was 93.3%. Threshold values of 7.8 positive LNs and a LNR of 13.8% were identified (nonlinearity <i>p</i> < 0.001 for both). Adjusted analyses revealed that ≥8 positive LNs were associated with a significantly increased hazard of MTC-specific mortality (hazard ratio [HR] 1.54, confidence interval [CI]: 1.09-2.17, <i>p</i> = 0.014, model area under the curve [AUC] 86.7%); a LNR ≥14% was associated with a significantly increased mortality hazard (HR: 3.308, CI: 2.096-5.222, <i>p</i> < 0.001, model AUC: 87.9%). The thresholds were significantly associated with recurrence-free survival in the testing cohort: 5-year recurrence-free survival was 56.5% (CI: 39.5 - 70.4) for patients with ≥8 positive LNs and 90.5% (CI: 82.5 - 94.9) for <8 (log-rank <i>p</i> < 0.001); it was 67.9% (CI: 55 - 77.9) for patients with LNR ≥14% and 92.8% (CI: 83.6 - 97) for LNR <14 (log-rank <i>p</i> < 0.001). <b><i>Conclusions:</i></b> Using population-level data, we identified robust LN thresholds associated with MTC-specific mortality. Compared with a threshold of 8 positive LNs, a LNR threshold ≥14% was associated with a greater increase in hazard of MTC-specific mortality.