TCGA Data on Head and Neck Squamous Cell Carcinoma Suggest Therapy-Specific Implications of Intratumor Heterogeneity

Abstract

To determine whether the relation of intratumor genetic heterogeneity to outcome in head and neck squamous cell carcinoma (HNSCC) differs depending on the therapy received. Three hundred and five initial HNSCC cases from The Cancer Genome Atlas (TCGA) showed a strong relation of high heterogeneity to shorter overall survival (OS) in HNSCC (PLoS Medicine 12: e1001786, 2015), but were insufficient to determine whether the relation to OS depended on therapy type. More complete TCGA data now allow examination of this issue. Clinical and whole-exome sequencing (WES) data on 528 HNSCC cases in TCGA were obtained from the NCI Genome Data Center and the Broad Institute. Clinical data were reviewed to determine if initial therapy included radiation or chemotherapy as primary or adjuvant therapy. Intratumor genetic heterogeneity was assessed by a modification of the MATH measure (Oral Oncology 49: 211, 2013) that improved handling of differing tumor-cell fractions among samples. Cox multiple regression of OS included age, year of diagnosis, smoking history, anatomic subsite, N and T classifications, HPV status, therapy, MATH, and another WES-derived classification. The interaction of MATH with therapy was evaluated to address the project's objective. Three hundred and ninety-three TCGA HNSCC cases (144 deaths) had sufficient data. The interaction of MATH with therapy was significantly related to OS (P = .016). With other clinical variables accounted for, the longest OS was seen for patients with low-MATH tumors who received chemoradiation (baseline for hazard ratio, HR). Intermediate OS was seen for patients with high-MATH tumors receiving chemoradiation (HR, 2.3; 95% CI, 1.1-5.2); patients with low-MATH tumors receiving no adjuvant therapy (HR, 1.7; CI, 0.7-4.2); and patients receiving adjuvant radiation without chemotherapy, with low (HR, 1.6; CI, 0.7-4.1) or high (HR, 1.7; CI, 0.7-4.1) MATH. The shortest OS was for patients with high-MATH tumors not receiving adjuvant therapy (HR=6.6; CI, 2.8-20). This first report that the relation of intratumor genetic heterogeneity to OS depends on therapy, although based on retrospective analysis and statistical control of other variables, has provocative implications that deserve prospective study. The results suggest that patients with high intratumor heterogeneity might benefit from radiation even when clinical considerations suggest that adjuvant therapy can be omitted. The results also suggest, however, that such patients might not benefit from the addition of chemotherapy, and thus could be spared the morbidity of complications from combination therapy. Intratumor heterogeneity should be evaluated in controlled trials that compare adjuvant radiation against chemoradiation following surgery for HNSCC.