Validating gene signatures in locally advanced HNSCC patients treated by PORT-C and in xenografts

Purpose or Objective:

To personalize radiotherapy of locally advanced head and neck squamous cell carcinoma (HNSCC), gene signatures have been developed that are related to processes involved in radioresistance of HNSCC. To date, most of these signatures were developed for patients who have received primary radiotherapy (pRCTx). The aim of our study was (I) to apply existing gene signatures related to different radio-biological processes for patients treated with postoperative radiochemotherapy (PORT-C) and (II) to validate these signatures in xenograft models.

Material and Methods:

This study is based on two cohorts: (i) 128 patients with locally advanced HVP16 DNA-negative HNSCC who received PORT-C in a multicentre retrospective study of the German Cancer Consortium Radiation Oncology Group (DKTK-ROG) between 2005 and 2011 and (ii) 60 mice bearing xenografts of 10 cell lines of human squamous cell carcinoma. Gene signatures related to cancer stem cell (CSC) markers [1], DNA repair [2], hypoxia [3], radiosensitivity [4] and epithelial mesenchymal transition (EMT) [5] were selected. Whole transcriptome analysis was performed using the HTA 2.0 Array (Affymetrix). Based on the gene signatures, risk groups were generated as described in the original publications (i) for the PORT-C cohort and (ii) for the xenograft data. The primary endpoint for the PORT-C cohort was loco-regional control (LRC), while the endpoint for the xenograft data was the dose to control 50% of the tumours (TCD50). The endpoints were evaluated byCox regression and the Mann-Whitney-U test, respectively.

Results:

All signatures were able to stratify patients treated with PORT-C into risk groups with a significant difference in LRC (figure A) or showed a statistical trend (radiosensitivity signature). This was confirmed in multivariable Cox regression including age and T stage. In the xenograft models, the gene signatures based on cancer stem cell markers, hypoxia and radiosensitivity showed a significant association with TCD50 (figure B).The correlations between these three gene signatures were weak (R<0.2).

Conclusion:

In our study, we successfully transferred gene signatures that were developed for patients with locally advanced HNSCC treated mainly by pRCTx to a patient cohort treated by PORT-C. Furthermore, cancer stem cell markers, hypoxia-associated genes and a radiosensitivity signature were able to stratify xenografts with respect to the TCD50. Since these signatures were weakly correlated, they may be considered as independent and robust biomarkers for future personalized radiotherapy of patients with locally advanced HNSCC.

References:

[1] Linge et al. Radiother Oncol 121: 364 (2016).
[2] Shen et al. Oncol Rep 83: 3403 (2017).
[3] Toustrup et al. Cancer Res 71: 5923 (2011).
[4] Kim et al. BMC Genomics; 13: 348 (2012).
[5] Chung et al. Cancer Res; 66: 8210 (2006).