PURPOSE: To investigate the use of xenograft models in a novel gene signature validation method using gene expression microarrays. MATERIALS AND METHODS: Gene expression profiles of ten human Head and Neck squamous cell carcinomas (HNSCCs) were obtained. Several published prognostic gene expression signatures were evaluated within this set. These consisted of different radiotherapy relevant signatures (i.e. for hypoxia, proliferation and 'stemness'). Signatures were correlated with various endpoints that have been determined in the ten different xenograft models. These include immunohistochemical measures for hypoxia and proliferation, volume doubling time (VDT) and local tumour control after fractionated irradiation or after single dose irradiation under clamp hypoxia. RESULTS: We found several significant correlations between the published gene expression signatures and tumour parameters. Several signatures, like the proliferation and wound signature correlated with BrdU labelling index. Further a 'stemness'-related gene signature showed a strong negative correlation with hypoxic fraction. CONCLUSIONS: Simultaneous assessment of immunohistochemistry, in vivo tumour properties and gene expression profiling in a comprehensive set of xenograft models can be used to validate and potentially infer biological information about prognostic gene signatures.
PURPOSE: To investigate the use of xenograft models in a novel gene signature validation method using gene expression microarrays. MATERIALS AND METHODS: Gene expression profiles of ten human Head and Neck squamous cell carcinomas (HNSCCs) were obtained. Several published prognostic gene expression signatures were evaluated within this set. These consisted of different radiotherapy relevant signatures (i.e. for hypoxia, proliferation and 'stemness'). Signatures were correlated with various endpoints that have been determined in the ten different xenograft models. These include immunohistochemical measures for hypoxia and proliferation, volume doubling time (VDT) and local tumour control after fractionated irradiation or after single dose irradiation under clamp hypoxia. RESULTS: We found several significant correlations between the published gene expression signatures and tumour parameters. Several signatures, like the proliferation and wound signature correlated with BrdU labelling index. Further a 'stemness'-related gene signature showed a strong negative correlation with hypoxic fraction. CONCLUSIONS: Simultaneous assessment of immunohistochemistry, in vivo tumour properties and gene expression profiling in a comprehensive set of xenograft models can be used to validate and potentially infer biological information about prognostic gene signatures.