Inflammatory Biomarkers and Oral Health Disorders as Predictors of Head and Neck Cancer: A Retrospective Longitudinal Study

Abstract

Head and neck cancers (HNCs) are often diagnosed late, leading to poor prognosis. Chronic inflammation, particularly periodontitis, has been linked to carcinogenesis, but systemic inflammatory markers remain underexplored. This study was the first to examine whether elevated C-reactive protein (CRP) can serve as a cost-effective adjunct in HNC risk assessment, alongside oral health indicators. A retrospective cohort study analysed 23,742 hospital records (4833 patients, 2015-2022) from the University Hospital of Debrecen. HNC cases were identified using ICD-10 codes, with CRP and periodontitis as key predictors. Kaplan-Meier survival analysis, log-rank tests, and Weibull regression were used to assess risk, with model performance evaluated via AIC/BIC and ROC curves. Periodontitis was significantly associated with HNC (HR 5.99 [1.96-18.30]), while elevated CRP (>15 mg/L) independently increased risk (HR 4.16 [1.45-12.00]). Females had a significantly lower risk than males (HR 0.06 [0.01-0.50]). CRP may serve as a cost-effective, easily accessible biomarker for early HNC detection when combined with oral health screening. Integrating systemic inflammation markers into HNC risk assessment models could potentially improve early diagnosis in high-risk populations.

Keywords: C-reactive protein; CRP; eGFR; estimated glomerular filtration rate; head and neck cancers; malignant; oral cavity cancer; oral squamous cell carcinoma.

Figure 1

Cumulative hazard plots by periodontitis, CRP levels, gender, and dental developmental disorders. Note: Cumulative hazard functions illustrating the relationships between time to head and neck cancer diagnosis and key variables: (A) periodontitis (yes vs. no), (B) C-reactive protein levels (>15 mg/L vs. ≤15 mg/L), (C) gender (male vs. female), and (D) dental developmental (DD) disorders (present vs. absent). Hazard functions are stratified by categories with 95% confidence intervals (shaded areas), calculated using the Nelson–Aalen estimator. CRP, C-reactive protein; DD disorders, dental developmental disorders.

Figure 2

Cumulative hazard plots by embedded and impacted teeth, dental caries, disease of hard tissue of teeth, and disorders of teeth and supporting structures. Note: Cumulative hazard functions depicting the association between time to head and neck cancer diagnosis and key dental conditions: (A) embedded and impacted teeth (present vs. absent), (B) dental caries (present vs. absent), (C) disease of hard tissue (DHT) of teeth (present vs. absent), and (D) disorders of teeth and supporting structures (DTSS) (present vs. absent). Hazard functions are stratified by condition categories with 95% confidence intervals (shaded areas), calculated using the Nelson–Aalen estimator. DHT, disease of hard tissue; DTSS, disorders of teeth and supporting structures.

Figure 3

Receiver operating characteristic (ROC) curve of the Weibull regression model for predicting head and neck cancer. Note: The ROC curve shows an AUC of 0.8646, demonstrating excellent model discrimination.

Figure 4

Kaplan–Meier survival curves stratified by significant predictors. Kaplan–Meier survival curves depicting observed and Weibull model-predicted probabilities over time. Stratification is based on significant covariates: (A) impacted teeth, (B) periodontitis, (C) C-reactive protein (CRP) levels, and (D) gender.