Sami Alsulami, Alexander Rühle, Eugene Yu, Jie Su, Brian O'Sullivan, Ali Hosni, John N Waldron, Ilan Weinreb, Stephen Smith, John de Almeida, David Goldstein, Christopher Mkl Yao, Andrew McPartlin, Andrew Bayley, Scott V Bratman, Ezra Hahn, Andrew Hope, John Kim, Nauman Malik, Enrique Sanz-Garcia, Anna Spreafico, Li Tong, Wei Xu, Eric Bartlett, Shao Hui Huang
Similar outcomes between major and minor SGCs support their unification in TNM9. While pTNM9 classification improves prognostic discrimination over pTNM8, cN-category performance remains suboptimal, likely reflecting moderate radiologic-pathologic concordance in LN assessment.
OBJECTIVES: To validate the performance of pathological and clinical UICC 9th edition/AJCC Version 9 (TNM9) classifications in salivary gland carcinoma (SGC).
METHODS: Patients with newly diagnosed major or minor SGC treated between 2006 and 2021 were retrospectively reviewed. Cases were restaged using pathological and clinical TNM8 and TNM9 criteria to compare 5-year disease-free survival (DFS) across both stage classifications. TNM9 restaging incorporated re-review of pre-treatment CT/MRI and pathology to assess lymph node (LN) number and extranodal extension (ENE).
RESULTS: Among 402 patients (282 major, 120 minor SGCs), DFS did not differ between major and minor tumors (p = 0.322). Median follow-up was 5.4 years. Under TNM8, pN categories showed paradoxical 5-year DFS for N1 vs N2a (58 % vs 67 %), although stage stratification remained acceptable (I/II/III/IVA-B: 97 %/88 %/64 %/40 %). For TNM9 assessment, radiologic-pathologic concordance for LN count was moderate (Kappa = 0.600). Imaging-detected ENE demonstrated high specificity (99 %) but low sensitivity (30 %) for pathologic ENE. TNM9 improved DFS separation by pN category (N0/N1/N2: 83 %/55 %/27 %) and stage (I/II/IIIA/IIIB: 97 %/88 %/61 %/33 %), with a slightly higher C-index (0.822 vs 0.817). TNM8 cN categories showed limited discrimination for DFS between cN1 vs cN2a (53 % vs 50 %), while TNM9 cN did not clearly distinguish cN1 from cN2 (36 % vs 42 %). Nonetheless, cTNM8 (I/II/III/IVA-B: 94 %/87 %/62 %/47 %) and cTNM9 stage (I/II/IIIA/IIIB: 94 %/87 %/58 %/40 %) both showed acceptable discrimination in DFS, with comparable C-indices (0.804 vs 0.808).
CONCLUSIONS: Similar outcomes between major and minor SGCs support their unification in TNM9. While pTNM9 classification improves prognostic discrimination over pTNM8, cN-category performance remains suboptimal, likely reflecting moderate radiologic-pathologic concordance in LN assessment.