Background

The NIMRAD phase III trial randomised 338 patients with head and squamous cell carcinoma (HNSCC) who were unsuitable for concurrent chemotherapy to radiotherapy alone or with hypoxia-modifying nimorazole. The trial aimed to validate prospectively a hypoxia-associated gene signature as a biomarker for treatment selection. The purpose here was to report our success rate in generating hypoxia signature scores for the trial and to investigate associations between scores and clinicopathological factors.

Method

Formalin-fixed paraffin embedded (FFPE) diagnostic biopsy samples were collected prospectively in the NIMRAD trial. RNA was extracted from 3 x 20 µm sections and the expression levels of 26 signature genes and five endogenous controls were analysed using customised TaqMan Low Density Array (TLDA) cards. The median of the median expression of the 26 genes was used to stratify patients into low- or high-hypoxia groups. Additional tissue sections were stained for p16 using immunohistochemistry (IHC) and HPV using in situ hybridisation (ISH). Primary tumour site and TNM staging data were collected from patient pathology reports.

Results

Biopsies were obtained from 313/338 NIMRAD patients. Sufficient RNA for downstream analysis was obtained for 286/313 patients representing an overall success rate in generating hypoxia scores of 85%. P-16 IHC and HPV-ISH data were generated for 284 and 274 patients, respectively. We found a 96% concordance between p16 IHC and HPV ISH staining. Tumours originating in the larynx (64%), or hypopharynx (59%) had higher percentages of hypoxia-high scores than those originating in the oropharynx (44%) (p=0.011). 70% of oropharynx were p16 positive compared with 2% hypopharynx and 6% larynx tumours. Oropharynx p16 positive tumours had higher percentages of hypoxia-low scores (63%) compared with oropharynx tumours testing negative for p16 (p=0.004). Most p16 negative tumours were classified as hypoxia-high irrespective of the primary tumour site (60% oropharynx, 60% hypopharynx, 67% larynx). Larger more invasive tumours (T3 vs T1-T2) were more likely to be hypoxia-high (63% vs 33-39%, p=0.002).

Conclusion

(1) It is possible to generate hypoxia signature scores prospectively within a UK randomised trial. (2) Tumour hypoxia varies with HNSCC sub-site, HPV status and T stage. 

Impact statement

Tumour hypoxia biomarker to aid patient treatment selection.