Head and Neck Cancer Genes Predictive of Radioresistance and Detriment to Local Control

Abstract

Radiotherapy (RT) is an essential treatment modality in head and neck cancer (HNC). Analysis of gene expression profiles in relation to in vitro cell survival and in vivo tumor response after RT may help identify predictive biomarkers and delineate mechanisms responsible for tumor radioresistance. Our goal is to identify genes that predict for both in vitro survival in HNC cell lines and locoregional recurrence (LRR) after RT in patients with HNC. After institutional review board approval, patient data was abstracted from a prospectively maintained institutional database and clinical chart review. Spearman’s rank correlation was calculated between expression and survival fraction after 2 Gy of RT (SF2) to cancer cell lines in HNC (GSE51370), representing a radioresistant gene expression profile in vitro. All microarray chips were normalized using the iterative rank-order normalization method. In patients treated with RT, Cox proportional hazard regression analysis was used to correlate gene expression and time to LRR. Genes significantly associated (P<0.05) with both LRR and cell line survival after RT were evaluated, to identify overlapping genes with similar expression profiles for both radioresistance and disease recurrence. A total of 11 HNC cell lines and 69 patients were analyzed. There were a number of genes associated with SF2 (n = 641) and LRR (n = 1591), but only 25 genes with a similar overlapping expression profile, either significantly decreased (n = 9) or increased (n = 16) in the 2 sets. The genes whose decreased expression was associated with a higher risk of LRR and radioresistance in HNC cell lines included: a metastasis/apoptosis suppressor (SDPR), transcriptional repressor (HES1), and an oxidative stress related checkpoint inhibitor (LCMT2). Less characterized genes identified were RCBTB1, LOC10099634, LOC202181, SCIN, SEL1L3, and LINC00324. The increase in gene expression that was associated with a higher risk of LRR included genes previously described as inhibitors of apoptosis (SLC30A2, EEF1A2, MAPK3, and NLRX1), associated with carcinogenesis (SGCG, EEF1A2, CDH15, NLRX1, and KLK14), and predictive of poor clinical outcome (EEF1A2, GAPDHS, KLK14, PAF1, and CCKAR) in various malignancies (i.e. breast, prostate, gastrointestinal, ovary, muscle, lung, and brain). Less characterized genes with an increased expression include:KRTAP5-8, GP8, ERVH48-1, TMEM38B, LOC339803, and CACNG1. The highest risk of LRR was CCKAR (HR 6.6, P = 0.02), which has been commonly associated with more aggressive gastrointestinal malignancies. This study has identified a number of genes associated with radioresistance and LRR in HNC. Many of these genes have been previously described in other malignancies and may be potential biomarkers and therapeutic targets in HNC. Further in vitro validation is required particularly for the uncharacterized genes identified.