Publications Search - Abstract View

Title: A model for predicting clinical outcome in patients with human papillomavirus-positive tonsillar and base of tongue cancer.
Authors: Tertipis N,  Hammar U,  Näsman A,  Vlastos A,  Nordfors C,  Grün N,  Ährlund-Richter A,  Sivars L,  Haeggblom L,  Marklund L,  Hammarstedt-Nordenvall L,  Chaturvedi AK,  Munck-Wikland E,  Ramqvist T,  Bottai M,  Dalianis T
Journal: Eur J Cancer
Date: 2015 Aug
Branches: IIB
PubMed ID: 26025766
PMC ID: not available
Abstract: AIM: To combine clinical and molecular markers into an algorithm for predicting outcome for individual patients with human papillomavirus (HPV) DNA/p16(INK4a) positive tonsillar and base of tongue squamous cell carcinoma (TSCC and BOTSCC). BACKGROUND: Head-neck cancer treatment has become more intensified, comprising not only surgery and radiotherapy, but also induction/concomitant chemotherapy and targeted therapy. With less treatment, 3-year disease free survival (DFS) is 80% for HPV-positive TSCC and BOTSCC. An 85-100% 3-year DFS is observed for HPV(+) TSCC and BOTSCC with absence of HLA class I, or CD44 expression, or high CD8(+) tumour-infiltrating lymphocyte (TIL) counts suggesting that therapy could be tapered for many if patients could be identified individually. PATIENTS AND METHODS: Patients treated curatively, with HPV DNA/p16(INK4a) positive tumours examined for HLA class I and II, CD44 and CD8(+)TILs, were included. An L1-regularised logistic regression was used to evaluate the effect of the biomarker data, age, stage, diagnosis, smoking and treatment on 3-year risk of death or relapse on a training cohort of 197 patients diagnosed 2000-2007 and validated on a cohort of 118 patients diagnosed 2008-2011. RESULTS: The variables finally included in the model were HLA class I, CD8(+) TILs, age, stage and diagnosis (TSCC or BOTSCC). The model showed acceptable discrimination and calibration. The discriminative ability of the model did not diminish after validation (AUC=0.77). CONCLUSION: To our knowledge, this is the first model to utilise information from several markers to predict an individual probability of clinical outcome for patients with HPV DNA/p16(INK4a) positive tumours.