Work

Intraoral Subsite, Histopathological Subtype, and Completeness of Pathological Characterization as Determinants of Cancer‐Specific Mortality in Minor Salivary Gland Carcinoma: A Population‐Based Competing‐Risks Analysis of 5820 Tumors

Year: 2026
Type: article
Study design: Observational Study
Abstract: OBJECTIVES: To quantify how completely intraoral minor salivary gland carcinomas are characterized pathologically across subsites, and whether subsite and subtype predict cancer-specific death after accounting for characterization, competing mortality, and stage. MATERIALS AND METHODS: SEER cohort (17 registries, 2000-2023) of 5820 tumors in 5777 patients. Cause-specific Cox, Fine-Gray, and Lunn-McNeil models adjusted for age, sex, race, subtype, grade, stage, era, and treatment were refitted without treatment and after excluding unclassified histology and unrecorded grade. Cause-of-death coding was checked by relative survival. RESULTS: Unclassified histology ranged from 8.7% (retromolar area) to 57.5% (mobile tongue) and unrecorded grade from 23.1% to 61.8% (both p < 0.001); unclassified histology was itself prognostic (hazard ratio 2.95, 95% CI 2.37-3.68). Ten-year cancer mortality ranged from 10.1% (buccal mucosa) to 35.1% (mobile tongue; p < 0.001); adjusted hazards were 2.03 (1.60-2.57) for mobile tongue, 2.01 (1.58-2.55) for floor of mouth, and 1.70 (1.26-2.30) for gingiva/alveolus. After excluding unclassified histology and unrecorded grade, floor of mouth (2.20, 1.48-3.26) and gingiva/alveolus (2.53, 1.47-4.35) persisted, whereas the mobile-tongue estimate became imprecise (1.61, 0.86-3.01). Other-cause mortality overtook cancer mortality at 9.4 years (13.2 years without a prior malignancy); 27.1% of other-cause deaths were second malignancies. Conditional 5-year cancer mortality in 10-year survivors was 12.7% for adenoid cystic and 1.2% for mucoepidermoid carcinoma. Machine learning gave no meaningful gain over logistic regression. CONCLUSIONS: Completeness of pathological characterization in registry data is patterned by anatomy and strongly prognostic, so anatomical prognosis and registry documentation are entangled; it is not a measure of diagnostic quality. Floor of mouth and gingiva/alveolus are robust subsite determinants of cancer death; the mobile-tongue excess cannot be separated from incomplete characterization. Competing mortality is substantial, largely from second malignancy. Findings may help inform risk-adapted surveillance, but cancer-specific mortality alone cannot define its intensity.
Language: English
Subfield: Surgery
Field: Medicine
Open Access status: gold
Estimated APC paid: $3,440
OpenAlex ID: W7234894933
FWCI:
Cites: 29
Cited by:
Related to: 10