Association between vitamin D deficiency and longitudinal risk of head and neck cancer: a multi-institutional...

Vitamin D is implicated in cancer biology through its regulation of cellular proliferation, apoptosis, and immune surveillance; however, evidence linking vitamin D deficiency (VDD) to head and neck cancer (HNC) risk remains limited by small sample sizes and inadequate control for confounders.

Using the TriNetX Global Collaborative Network, we conducted a retrospective cohort study comparing adults with laboratory-confirmed VDD (25-hydroxyvitamin D < 20 ng/mL) against controls (≥30 ng/mL). After applying the exclusion criteria and 1:1 propensity score matching, 108,010 patients per group were retained. In a separate exploratory dose-response analysis, patients with vitamin D insufficiency (20.0-29.9 ng/mL) were compared with the same sufficiency threshold (≥30 ng/mL). A 180-day landmark period was applied to mitigate reverse causation, with follow-up extending to 10 years. The primary outcome was incident HNC; the secondary outcomes included laryngeal, oral, and other HNC subtypes. Positive (osteoporotic fracture) and negative (appendicitis) control outcomes were assessed to evaluate internal validity.

VDD was associated with a significantly higher risk of incident HNC (HR 1.56, 95% CI 1.43-1.70, p < 0.001). Consistent associations were observed across anatomical subtypes: laryngeal cancer (HR 1.72, p < 0.001), oral cancer (HR 1.53, p < 0.001), and other HNCs (HR 1.46, p < 0.001). Vitamin D insufficiency was examined separately in 140,247 matched pairs and was associated with a smaller increase in HNC risk than VDD (HR 1.21, p < 0.001), supporting a dose-response pattern. Control outcomes validated the study design, with osteoporotic fracture confirming the expected positive association (HR 1.58, p < 0.001) and appendicitis demonstrating no spurious signal (HR 1.06, p = 0.226). The association was prominent in patients aged >50 years but absent in younger adults (p for interaction <0.001). Sensitivity analyses with extended landmark periods and alternative cohort definitions yielded consistent results.

In this observational study, VDD was significantly associated with an increased risk of HNC, with consistency across subtypes and a dose-response gradient. However, the observational design precludes causal inference, and residual confounding cannot be fully excluded. Prospective studies with detailed lifestyle data are needed to confirm these findings.