Biological and lifestyle predictors of survival in head and neck cancer.

Student thesis: Doctoral ThesisDoctor of Philosophy (PhD)


Head and neck cancer (HNC) is a heterogeneous disease with variable clinical outcomes. Several biomarkers for HNC prognosis have been identified and incorporated into clinical prediction models, but the ability of these models to predict outcome could be improved.

Using self-report and peripheral blood-based DNA methylation (DNAm) predictors, I examined whether lifestyle and life-course exposures (smoking, alcohol-drinking, body mass index (BMI), education and biological aging) predict all-cause mortality and should be considered for inclusion in prognostic models. Additionally, I examined the association of circulating metabolites with all-cause mortality.

I found that current smokers with HNC were twice as likely to die during follow-up than never-smokers (HR=2.0; 95% CI: 1.4, 3.0), after controlling for established prognostic predictors. In people with oropharyngeal cancer (OPC), DNAm-based predictors of smoking were associated with all-cause mortality. Based on a DNAm predictor that included 2,623 CpG sites (Joehanes-Bonferroni) and on methylation levels at a single site (cg05575921) within the aryl-hydrocarbon receptor repressor (AHRR) gene, the change in risk of death (all-cause) per unit increase in standardised DNAm score was approximately two-fold, where 1 unit increase corresponds to the average increase in the measured methylation score experienced by current smokers compared with never smokers (HR=1.89 [95% CI:1.06, 3.47] and 1.92 [95% CI:1.03, 2.33]).

Two DNAm-age measures, comprising a set of 1,030 CpGs (AgeAccelGrim) and 71 CpGs (IEAAHannum), were associated with all-cause mortality in OPC. AgeAccelGrim had the largest magnitude of effect in fully adjusted models: each SD increase in AgeAccelGrim resulted in a 39% increased all-cause mortality risk (HR=1.39; 95% CI:1.06, 1.83). The addition of AgeAccelGrim to a standard clinical model slightly improved mortality risk prediction at 3-years (AUC: 0.80 vs. 0.77; p-value for difference=0.069).

In adjusted models, acetate and creatinine were associated with mortality in OPC (HRs per SD increase in metabolic trait =1.30 [95% CI:1.11, 1.51] and 0.68 [95% CI: 0.53, 0.89], respectively).

This thesis demonstrates the potential of lifestyle, epigenetic and metabolomic measures to enhance survival prediction in people with HNC, though these findings need to be replicated in independent clinical cohorts.
Date of Award23 Jun 2020
Original languageEnglish
Awarding Institution
  • The University of Bristol
SupervisorDiana Dos Santos Ferreira (Supervisor), Suzanne M Ingle (Supervisor), Andy Ness (Supervisor), Richard M Martin (Supervisor) & Margaret T May (Supervisor)

Cite this