Gevaert and Colleagues Discover a Distinct Cancer Subtype and a Biomarker for It

Olivier Gevaert, PhD

In an article published online in EBioMedicine* in March 2017, Olivier Gevaert, PhD (assistant professor, biomedical informatics research) and his coauthors showed a novel subtype of head and neck squamous cell carcinoma with important implications for treatment. 

Smoking is the primary cause of head and neck squamous cell carcinoma (HNSCC), although about ten years ago researchers found that some patients without a history of smoking had HNSCC caused by the human papilloma virus (HPV). Gevaert and his colleagues have found an atypical type of HNSCC in patients who are both non-smokers and HPV negative. “One characteristic of this subtype,” says Gevaert, “is that none of the risk factors we associate with head and neck squamous cell carcinoma are present in these patients: no history of smoking, no history of alcohol consumption, no HPV.”

HNSCCs occur more often in men than women, but in this atypical subtype there is a higher percentage of women (43 percent) than in other subtypes. They also tend to be slightly older.

Therapy for these cancers is a significant issue. Gevaert explains: “the current treatment strategy for head and neck cancer is to first determine whether patients are HPV positive (HPV+) or HPV negative (HPV-). HPV status is the only factor used to stratify HNSCC treatment in the clinic. There are no biomarkers available to stratify HPV negative HNSCCs into clinically relevant subgroups; therefore, HPV negative HNSCC is treated as a single clinical entity. No treatment with a clear understanding of the biology has been described.”

There were several novel elements to this study. First was the methodology.

“We studied head and neck squamous cell cancer using epigenomic data, looking at changes of the DNA by a process that is called DNA methylation. The method that we used, which is called MethylMix, we developed previously in a pan-cancer setting, in which we used it to define subtypes in multiple cancers from The Cancer Genome Atlas (TCGA). Somewhat surprisingly, we found that using DNA methylation data is more robust than other methods, for example gene expression data. One reason for this is that DNA methylation is in healthy tissues reflecting cell type, so it is closer to the origin of a tumor.

“The other advantage of using DNA methylation data is that because it is a DNA-based assay, it is more stable than RNA, and it has more practical advantages. DNA can be stored easier and for a long time in paraffin. It is possible to use archival samples. Also, DNA-based analysis is easier to be developed into a robust assay.”

“During many conversations with clinicians treating HNSCC, they told us that they felt that non-smoking, HPV- HNSCCs, particularly those occurring within the oral cavity in females, represent a distinct clinical entity. However, no molecular evidence existed to substantiate this, nor did any biomarkers exist to diagnose this subtype. In our study, we found four HPV- subtypes, one of which matches this subtype. Now we have found a biomarker to subtype the HPV- head and neck cancers, which is the DNA methylation pattern.” 

John Sunwoo and I are looking at cultures from these patients and are thinking about applying a class of drugs called Smac mimetics, and seeking a response.

Now that they know how to subtype these patients, Gevaert and his colleagues are pursuing two additional activities in parallel, one focused on diagnosis and the other on treatment.

“First, we want an assay that can be used clinically, making it very easy to classify patients into one of the five subgroups. We are quite confident that designing the assay will be straightforward because the signals that we’re seeing in the data are very strong and stable also in validation cohorts. We don’t expect any scientific challenges; it’s more of a technical challenge since we want to use as few genes as possible so that the assay will be cheap. In addition, the fewer genes we use, the more reproducible the assay is going to be.”

“Second, based on the biology that we describe in the article, we have an immune story in this subtype. This subtype is highly inflamed, and based on expression of known biomarkers, it can be more susceptible to immunotherapy than other HPV- HNSCC subtypes. But there’s also enrichment of mutations in CASP8, which is a gene that’s involved in controlling apoptosis or cell death. Because that gene is being turned off, apoptosis is being turned off. We want to reactivate the apoptosis pathway so that the cancer cells will die. John Sunwoo, MD (associate professor of otolaryngology - head and neck surgery), and I are looking at cultures from these patients and are thinking about applying a class of drugs called Smac mimetics, and seeking a response.”

*Brennan K, Koenig JL, Gentles AJ, Sunwoo JB, Gevaert O. Identification of an atypical etiological head and neck carcinoma subtype featuring the CpG island methylator phenotype. EBioMed 2017 DOI: