Therapeutic Targeting of Hypoxia in Xenograft Models of Head and Neck Cancer

Abstract

Purpose: Human papilloma virus-negative head and neck squamous cell carcinoma (HNSCC) is associated with high levels of tumour hypoxia, poor patient prognosis and very little clinical progress over the last few decades. Therapeutic targeting of tumour hypoxia in HNSCC may be possible through the use of hypoxia activated prodrugs (HAPs). HAPs have yet to be successful in clinical trials despite good preclinical antitumour activity, likely due to a lack of patient stratification strategies for tumour hypoxia and appropriate preclinical models. The aim of this master's thesis was to investigate tumour hypoxia and expression of candidate genes of HAP sensitivity in patient derived xenograft (PDX) and cell line xenograft (CLX) models of HNSCC as well as to determine the sensitivity of CLX models to selected HAPs. Experimental Design: The hypoxic fraction of six novel PDX models and six CLX models of HNSCC was assessed through pimonidazole (pimo) immunohistochemistry. Evaluation of Toustrup gene signature classification and expression of POR, SLFN11 and MKI67 were carried out by whole-transcriptome RNA sequencing of seven CLX models and NanoString analysis of nine second generation (P2) PDX models that were also compared to their original (P0) clinical specimens. The ex vivo anticancer activity of DNA cytotoxins PR-104 and TH- 302, the DNA-PK inhibitor SN38023 and nitro-CBI compounds SN30726 and SN38737 were assessed as single agents or in combination with 13 Gy radiation (RAD) through excision assays carried out on two CLX models (UT-SCC-54C and UT-SCC-74B). Results: CLX models displayed significantly higher pimo-positive hypoxic fractions (HF) (5.6% ± 0.7%) than PDX models of comparable size (4.8% ± 0.4%, P<0.005). Significant correlation between Toustrup classification and pimo-positive HF (P<0.005) was found in PDX models analysed alone, but not when combined with CLX models which displayed lower Toustrup scores regardless of greater pimo-positive HF compared to PDXs. The Toustrup hypoxia signature, POR and MKI67 gene expression showed good correlation between P2 PDX and P0 clinical tumours, although this was not the case with SLFN11 expression, which was 5.2-fold lower in P2 PDX tumours. In the excision assays, PR-104, TH-302 and SN38737 displayed antitumour activity as single agents, while in combination with 13Gy RAD, antitumour activity was seen with PR-104, TH-302, SN38737 and SN38023. Conclusion: The characterisation of PDX and CLX models of HNSCC outlined potential differences between these models in terms of tumour hypoxia and gene expression. However, due to the similar hypoxia, gene expression and HAP sensitivity in UT-SCC-54C and UT-SCC- 74B, it was not possible to evaluate the effects of tumour hypoxia and gene expression on HAP sensitivity. Nevertheless, ongoing excision assays in other CLX models will allow for the future evaluation of HAP sensitivity based on the models characterised in this project.