Site and extent of distant metastasis at diagnosis as used to stratify survival in head and neck cancer: A population-based analysis.

Abstract

e18041 Background: Distant metastases in head and neck cancer (HNC) are associated with poor prognosis. Population-level data defining the prognostic impact of specific metastatic sites and multisite dissemination are limited. We compared overall survival by metastatic site and extent. Methods: Using SEER 17 Registries (2010–2017), we conducted a retrospective analysis of patients with HNC, classifying distant metastatic involvement by site (bone, lung, liver, brain). The study followed a prespecified protocol and was exempt from institutional review board oversight. Patients were categorized as having isolated single-site or multisite metastatic disease. Overall survival was estimated using Kaplan–Meier methods with log-rank testing. Multivariable Cox proportional hazards models adjusted for age, sex, and year of diagnosis were used to assess diagnosis-time prognostic associations. Analyses were conducted using complete-case methods. Results: Among 7,491 patients with head and neck cancer, 1,082 (14.4%) had distant metastases to bone, lung, liver, or brain. Lung metastases were most common (n=789), followed by bone (n=323), liver (n=180), and brain (n=31). Patients without distant metastases had a median overall survival (OS) of 28 months. Metastases involved a single organ in 81.0% (876/1,082) and were multisite in 19.0% (206/1,082), most frequently with liver metastases (63.9%) and least frequently with lung metastases (20.5%). Among patients with isolated metastatic disease, median OS differed by site: brain-only 3 months, bone-only 8 months, lung-only 9 months, and liver-only 11 months (log-rank p<1×10⁻¹⁶). For bone, lung, and liver metastases, multisite disease was associated with worse survival (bone: 8 vs 5 months, p=0.0055; lung: 9 vs 4 months, p=1.77×10⁻⁷; liver: 11 vs 5 months, p=0.0036), while brain metastases had poor outcomes regardless of extent (p=0.59). In multivariable Cox regression, metastatic site remained independently associated with OS. Compared with non-metastatic disease, brain metastases conferred the highest mortality risk (HR 3.92, 95% CI 2.74–5.59), followed by lung (HR 2.17, 95% CI 2.00–2.35), bone (HR 1.82, 95% CI 1.61–2.07), and liver metastases (HR 1.56, 95% CI 1.32–1.84) (all p<0.001). Conclusions: In this large, population-based SEER cohort, survival in metastatic head and neck cancer at diagnosis varied systematically by both metastatic site and extent of dissemination, with effects persisting after multivariable adjustment. Brain metastases defined a rare but uniformly high-risk presentation, while bone, lung, and liver metastases exhibited substantial prognostic heterogeneity that was further modified by the presence of multisite disease. These findings establish diagnosis-time metastatic site and extent as complementary determinants of survival.