HPV Gene Expression Signatures, Prognosis in Cervical Cancers

Curtis Pickering, PhD, is assistant professor of head and neck cancer at the University of Texas MD Anderson Cancer Center in Houston. He coauthored a new study, published in JCI Insight, describing candidate prognostic gene expression signatures for human papillomavirus (HPV)-positive oropharyngeal and cervical cancers. The team’s goal was to validate a gene panel that can identify lower-risk patients for whom treatment can be safely de-escalated, reducing potential treatment toxicities.

Cancer Network: Why do patients with HPV-negative head and neck cancer tend to have shorter survival times than those with HPV-positive cancers? Is that well-understood?

Dr. Pickering: HPV-positive head and neck tumors have longer survival times because they respond better to radiation-based therapy than HPV-negative tumors. It is not completely understood why HPV causes a better response to radiation-based therapy, although many hypotheses have been proposed.

Cancer Network: What led you and your colleagues to suspect that gene expression profiling might identify molecular subtypes of HPV-positive cancers with prognostic significance? (Or was the search for gene expression signatures driven primarily by the need for prognostic biomarkers to help inform treatment decisionmaking for patients HPV-positive cancers?)

Dr. Pickering: Gene expression profiling is a powerful tool that gives a snapshot of the functions going on in a tumor. It can often be used to identify different subtypes of tumors with unique characteristics. Our initial goal was to use gene expression profiling to understand variations among HPV-positive tumors and how that might associate with HPV function.

Previous genomic studies, like TCGA (The Cancer Genome Atlas), indicated that there were different amounts of HPV gene expression among HPV-positive tumors. This suggested to us that there would be different levels of HPV function among HPV-positive tumors. Since HPV status is associated with better treatment response we hypothesized that the level of HPV function might also be associated with treatment response. For example, the tumors with highest HPV function might have the best responses.

We approached this hypothesis with the goal of understanding the biology of HPV function and treatment response, and with the hope that it would lead to new treatment options for the patients with poor response. We did not start by searching for a biomarker. Fortunately, the gene expression signature we identified is also a powerful prognostic biomarker. This has led to two main future directions for the project; develop a clinical biomarker assay for risk stratification, and identify novel therapeutic targets for the tumors with poor prognosis.

Cancer Network: Using RNA sequencing data from 80 oropharyngeal cancers in The Cancer Genome Atlas, your team identified 582 genes that differentiated high-risk HPV, low-risk HPV, and HPV-negative. Then you were able to narrow that down to 38 genes that can differentiate between high- and low-risk HPV-positive tumors, including two viral genes not previously linked to tumor progression. What were those two genes and what did your team’s cell-line experiments suggest about their function and clinical implications?

Dr. Pickering: The HPV genes E1 and E1^E4 were identified as differentially expressed between the high and low risk groups. In cell line experiments we found a strong association between E1^E4 expression and response to radiation treatment; with the cell lines expressing low or no E1^E4 being more resistant to the radiation treatment. This is consistent with the finding that high risk HPV-positive tumors express lower levels of E1^E4. Additionally, when we stratified cell lines by their E1^E4 expression we identified differential expression of similar genes and pathways to what was observed in the human tumors. We are now in the process of functionally testing whether E1^E4 can modulate cell line aggressiveness and radiation sensitivity.

Cancer Network: Your team retrospectively validated the gene signature using two patient cohorts: patients with oropharyngeal cancer and cervical cancers. How did the panel perform with those two cohorts?

Dr. Pickering: In both validation cohorts the biomarker signature stratified patients into two groups with significantly different survival outcomes. This stratification was statistically significant on multivariate analysis, suggesting that it is independent of other major clinical factors.

Cancer Network: What’s next?

Dr. Pickering: We are working to translate these findings into a robust clinical biomarker assay amenable for use on routine clinical specimens. This is not a trivial task because clinical specimens are often small and not collected to preserve the RNA that we are detecting, but it should still be feasible with currently available technology. Once the assay is optimized, we will test it on additional retrospective cohorts of samples and eventually in prospective samples, ideally as part of ongoing clinical trials.