Predicting recurrence after radiotherapy in head and neck cancer

Abstract

Head and neck squamous cell carcinoma (HNSCC) is the sixth most common cancer worldwide. Radiotherapy is a mainstay of treatment, either alone for early stage tumors or combined with chemotherapy for late stage tumors. An overall 5-year survival rate of around 50% for HNSCC demonstrates that treatment is often unsuccessful. Prediction of outcome is, therefore, aimed at sparing patients from ineffective and toxic treatments on the one hand, and indicating more successful treatment modalities on the other. Both functional and genetic assays have been developed to predict intrinsic radiosensitivity, hypoxia, and repopulation rate. Few, however, have shown consistent correlations with outcome across multiple studies. Messenger RNA and microRNA profiling show promise for predicting hypoxia, whereas epidermal growth factor receptor expression combined with other measures of tumor differentiation grade shows promise for predicting repopulation rate. Intrinsic radiosensitivity assays have not proven useful to date, although development of repair protein foci assays indicates promise from preclinical studies. Assays for cancer stem cell content have shown promise in several clinical studies. In addition, 2 assays showing robustness as predictors for outcome in HNSCC are human papilloma virus status and epidermal growth factor receptor expression. Neither these nor stem cell assays, however, can as yet reliably indicate alternative and better treatments for poor prognosis patients. It would be of great value to have assays that predict the benefit for an individual from combining new molecularly targeted agents with radiotherapy to increase response, in particular those that exploit tumor mutations to provide tumor specificity. Predictive assays are being developed for detecting defects in repair pathways for single- and double-strand DNA breaks, which should allow selection of drugs targeting the appropriate backup pathway, thus exploiting the concept of synthetic lethality. This is one of the most promising areas for prediction, both currently and in the future.