Skip to main content
Premium Trial:

Request an Annual Quote

Head and Neck Cancer Characteristics, Subtypes Refined With Single-Cell Transcriptomes

NEW YORK (GenomeWeb) – A team from the US and Israel has taken a single-cell transcriptomic approach to characterizing and comparing head and neck squamous cell carcinoma (HNSCC) tumors from 18 individuals. 

As they reported online today in Cell, researchers did single-cell RNA sequencing on more than 5,900 individual cells from 18 untreated primary HNSCC tumors and five matched HNSCC lymph node metastases. They also used exome sequencing and SNaPshot targeted genotyping to profile somatic mutations and chromosomal alterations in bulk tumor samples.

When they combined the bulk tumor and single-cell transcriptome profiles, the researchers gained new insights into both the malignant and stromal cell subtypes present in the head and neck cancers they focused on. The results also suggested that cancerous cells carrying the signature of a so-called partial epithelial-to-mesenchymal transition (p-EMT) associated with tumor spread and metastasis tend to be found on the growing outskirts — or "leading edge" — of primary tumors.

"Our results provide insight into the HNSCC ecosystem and define stromal interactions and a p-EMT program associated with metastasis," senior author Bradley Bernstein, a pathology researcher affiliated with Massachusetts General Hospital and the Broad Institute, and his co-authors wrote.

The researchers used the SMART-Seq2 protocol and Illumina NextSeq 500 instrument to do scRNA-seq on 5,902 individuals cells from fresh oral cavity HNSCC biopsy samples from 18 individuals. Along with the primary tumor samples, the cells came from lymph node metastases from five of the HNSCC patients. Using Illumina instruments, they also did exome sequencing and targeted SNaPshot sequencing on fresh and formalin-fixed paraffin-embedded tumor samples, respectively.

Based on inferred copy number profiles, epithelial cell marker expression, and cell expression clusters, the team identified 2,215 malignant cells and 3,363 non-malignant cells in single-cell transcriptome set. The latter cells fell into eight clusters representing T cells, B/plasma cells, macrophages, mast cells, endothelial cells, fibroblasts, and myocyte cells. Some of those non-malignant cell types could be further clustered into cellular subsets, particularly fibroblasts and immune T cells.

When they looked at clustering of the malignant cells, however, the researchers found that they typically grouped alongside other individual cells from the same tumor, consistent with the presence of pronounced between-individual tumor variability. The differentially expressed genes fell in pathways contributing to everything from cell stress and hypoxia to detoxification, drug metabolism, immune activation, and epithelial differentiation.

Within individual tumors, malignant cells with the p-EMT — which are suspected of having the ability to metastasize — typically showed up at the edges of tumors in association with cancer-associated fibroblast cells beyond the tumor's boundaries, the team reported. Results from in vitro experiments on sorted cells from HNSCC cell lines with higher- or lower-than-usual p-EMT signature expression supported the notion that p-EMT cells tend toward invasiveness, perhaps prompted in part by microenvironment cell interactions.

For the most part, malignant cell patterns in metastatic tumors were comparable to those in corresponding primary tumors, though the researchers cautioned that they profiled a modest number of matched primary tumor-lymph node metastasis tumor pairs.

Incorporating the new transcriptome data, particularly for a handful of tumors with the highest proportion of sequenced malignant cells, they found that malignant cells from the current study typically fell in just three of the four HNSCC expression subtypes previously reported for the Cancer Genome Atlas project.

On the other hand, many of the non-malignant cells from the stroma and immune cell collection had expression profiles similar to the mesenchymal subtype of HNSCC, prompting the authors to suggest that "the mesenchymal TCGA subtype reflects high stromal representation in bulk samples, rather than a distinct malignant cell program."

Based on their follow-up analyses, they argued that "HNSCC tumors may be refined into three subtypes of malignant cells… with the previously described mesenchymal subtype reflecting malignant-basal tumors with a large stromal component."