Routine Tonsil Pathology Rarely Identifies Unexpected Malignancy: A Safety-Net Hospital Experience.

Goldberg, Zachary N; Pillai, Anjali; Sure, Asritha; Grillone, Gregory A; Zhao, Qing; Platt, Michael P · Laryngoscope · 2026

retrospective_cohort · Level III

Where this comes from

Abstract

Routine pathological examination of tonsillectomy specimens has declined as occult malignancy is rare, but its value in safety-net settings is unstudied. We characterized preoperative clinical assessment as a triage instrument, and occult malignancy incidence, in this setting. This retrospective cohort study included all patients undergoing tonsillectomy for any indication at a large US safety-net hospital from January 2022 to June 2025. Pathology reports were screened for malignancy; malignant cases underwent chart review for demographics, Charlson Comorbidity Index (CCI), and preoperative assessment, stratified as expected (biopsy-proven), suspected (clinical or radiographic concern without tissue diagnosis), or unexpected (no preoperative concern). Chi-square and Mann-Whitney U tests compared malignancy type and CCI across strata. Of 769 specimens (307 pediatric, 462 adult), no pediatric malignancies were identified (0%; 95% CI, 0%-1.19%). Twenty adult malignancies occurred: 13 squamous cell carcinoma, 7 hematologic. As a head and neck oncology referral center, the cohort includes tonsillectomies for known or suspected malignancy; 19 of 20 (95%) were expected or suspected preoperatively. One mantle cell lymphoma was unexpected, comprising 0.22% of adult specimens (1/462; 95% CI, 0.005%-1.20%) and 0.23% of 443 resected without preoperative concern (0.006%-1.25%). Preoperative risk category was not associated with malignancy type (p = 0.41) or CCI (p = 0.37). No pediatric malignancies were identified, and 95% of adult malignancies were preoperatively anticipated. Despite the elevated cancer risk profile of safety-net populations, preoperative clinical assessment performed well as a triage tool, supporting risk-stratified pathology protocols. Incidence estimates were imprecise at this sample size.