NEW YORK – Researchers affiliated with the National Cancer Institute's Clinical Proteomic Tumor Analysis Consortium (CPTAC) have identified a set of immune subtypes spanning 10 different cancer types.
The findings, published last week in Cell, indicate that similar immune mechanisms may be at work across a broad range of cancers, raising the possibility that they share common therapeutic targets and could be susceptible to the same treatments, said Francesca Petralia, assistant professor at the Icahn School of Medicine at Mount Sinai and first author of the paper.
The study looked at 1,056 treatment-naïve tumor samples from 10 different cancer types: breast cancer (113 samples), clear cell renal cell carcinoma (103), colon cancer (96), glioblastoma (99), head and neck squamous carcinoma (110), lung squamous carcinoma (108), lung adenocarcinoma (110), ovarian cancer (82), pancreatic ductal adenocarcinoma (140), and uterine cancer (95). These samples had previously undergone various analyses by CPTAC researchers, including whole-genome sequencing, RNA-sequencing, proteomics, and phosphoproteomics.
The researchers analyzed these data in various ways, aimed at characterizing the immune landscape of the different tumors.
Using the BayesDeBulk algorithm, for example, they used bulk gene expression and proteomic data to estimate the proportion of different cell types in the tumors' microenvironments, providing a measure of immune infiltration.
Combining previously observed immune signatures with genetic and proteomic data, the researchers placed the different tumors into immune subtypes, identifying seven subtypes across all samples. They then explored the relationship between these immune subtypes and molecular features including the mutation profiles of genes commonly mutated in cancer, gene copy number variation, germline variants, DNA methylation, kinase activity, and kinase-based transcription factor regulation.
Pei Wang, professor of genetics and genomic sciences at Mount Sinai and the senior author on the study, said that this novel combination of data types and analysis approaches enabled new insights into tumor immunity.
Compared to previous work, "we not only included more data types like proteomic [and] phosphoproteomic [data], but we also utilized different data analysis tools," she said. "We not only used pathway signature [analysis] but also did the deconvolution analysis to derive the cell type composition of the tumor microenvironment and used all that information to derive the subtypes."
Wang suggested that in addition to indicating commonalities across cancer types, the team's subtypes revealed "a bit more heterogeneity within each individual tumor."
One of the most notable of the seven immune subtypes the researchers observed was the CD8-/IFNG+ subtype, Petralia said, which is characterized by a depletion of T cells but an activation of interferon gamma signaling, which is involved in the regulation of T cells and other components of the immune system.
"It means that upstream interferon gamma signaling is activated, but for some reason the T cells are not there," she said.
While previous work has identified tumor subtypes with activated interferon gamma signaling, she noted, these studies did not distinguish between tumors in which T cells were depleted versus enriched.
"By leveraging proteomics and gene expression data, we were basically able to separate the tumors that have interferon gamma activation and lymphocytes from tumors that [have interferon gamma activation but] don't have lymphocytes," she said.
The results suggest that the tumor is evading the immune system, Petralia added. "There is some evasive mechanism in place that prevents the immune cells from being around the tumor."
Also notable was the CD8+/IFNG+ immune subtype, in which interferon gamma signaling is activated and T cells are enriched. The researchers observed this subtype, which they linked to high tumor immunogenicity, across all 10 tumor types.
To determine if this CD8+/IFNG+ subtype is, in fact, linked to response to immunotherapy, the CPTAC team looked at data from a Phase III clinical trial in which 425 non-small cell lung cancer patients were treated with Genentech's immune checkpoint inhibitor Tecentriq (atezolizumab). Using an immune subtype prediction model they developed using CPTAC RNA-seq data, they were able to assign these patients to the different immune subtypes and identified 75 as CD8+/IFNG+. These patients "showed significantly better" progression-free survival when treated with Tecentriq than those with other subtypes, the authors noted. This subtype-based difference in PFS did not appear in a separate group of 355 patients in the same trial who were treated with the chemotherapy docetaxel.
"We were able to show, basically, that [patients with the CD8+/IFNG+ subtypes] had better results in the immunotherapy trial but not in the chemotherapy trial," Petralia said.
"That is really our goal," Wang added, "to use these subtypes to identify patients who could be more responsive to treatment."
She and her colleagues are now investigating whether the subtypes they identified in the study could help predict response to immunotherapies in melanoma. She noted that while immunotherapy is a first-line treatment for this cancer, around half of patients don't respond.
Wang said the researchers also hope to expand their analyses to other cancer types to determine if they observe the same subtypes. She noted that they have analyzed a separate lung cancer dataset and found that they could replicate their results.
The researchers also used phosphoproteomics data to explore kinases and signaling pathways linked to tumor immunogenicity, identifying protein targets that might help convert non-responding or "cold" tumors to highly responsive or "hot" tumors.
"There are still some complicated mechanisms [underpinning tumor immunology] that we don't fully understand," Wang said. "That is why right now, I don't think clinically we have a very good tool to predict which patients will respond well to checkpoint immunotherapy. That is why we want to characterize these immune subtypes. We really want to know for each of them what the mechanisms are that prevent a tumor from responding to treatment."