NEW YORK – A research team from the Chinese Academy of Medical Sciences and Peking Union Medical College as well as other centers in China has identified four molecular subtypes of esophageal squamous cell carcinoma (ESCC), including one linked to better-than-usual response to immune checkpoint blockade immunotherapy.
"These results emphasize the clinical value of unbiased molecular classification based on multiomics data and have the potential to further improve the understanding and treatment of ESCC," senior authors Qimin Zhan, Yongping Cui, and Yuchen Jiao, along with their colleagues, wrote in Cancer Cell on Thursday.
Using whole-genome sequencing, exome sequencing, whole-genome bisulfite sequencing, RNA and small RNA sequencing, and proteomic profiling, the researchers assessed matched tumor and normal samples from 155 ESCC cases, putting together an "ESCC Genome and Epigenome Atlas" from the profiles. In the process, they identified molecular features that slotted the tumors into four subtypes, dubbed the cell cycle activation (CCA), NRF2 oncogenic activation (NRFA), immune suppression, and immune modulation subtypes.
Because one of these ESCC subtypes appeared to have particularly pronounced response to checkpoint immunotherapies such as anti-PD-1 treatment, the team went on to establish a classifier model for finding tumors in this immune-related ESCC subtype, focusing on more than two dozen tumor features that appeared to coincide with favorable checkpoint immunotherapy response and survival.
"This comprehensive profiling led to the development of a multiomics-based molecular classification of ESCC, and its potential to predict response to immune therapy was validated in an independent cohort from a clinical trial with a single-agent anti-PD-1 antibody," the authors explained.
With genome, gene expression, small RNA, DNA methylation, and clinical features found in the initial set of 155 ESCC cases, together with proteomic profiles for 73 of these ESCC cases, the investigators tallied mutation patterns, copy number profiles, and mutational signatures, as well as methylation, transcriptional, and proteomic shifts in the ESCC tumors, using these multiomic features to cluster the tumors into four subtypes.
Within the CCA subtype, for example, the team described tumors marked by homozygous CDKN2A/B deletions, amplification of the CCND1 gene, and hypermethylation features consistent with a so-called esophageal CpG island methylator phenotype (E-CIMP).
The researchers' patient-derived organoid experiments suggested that tumors in that subtype may be primed to respond to the type of CDK4/6 inhibitors used to target some breast cancers. On the other hand, they found that NRFA subtype tumors characterized by E-CIMP in combination with NFE2L2, KEAP1, or CUL3 mutations, SOX2 gene amplifications, and enhanced NFE2L2 and SOX2 expression may be more amenable to treatment with NRF2 inhibitors.
Another cluster contained tumors with scant mutations, higher-than-usual ERBB2 expression, and tumor infiltration by immune B cells and specific natural killer cells — an immune suppression (IS) subtype with potential for responding to checkpoint immunotherapy in combination with targeted ERBB2 therapy.
Finally, the team explained, ESCC tumors with high tumor infiltration by CD8-positive T cells, natural killer cells, and macrophages but E-CIMP-negative features, were placed in an immune modulation (IM) subtype associated with response to immune checkpoint blockade immunotherapy.
"IS and IM cases were highly immune infiltrated but differed in the type and distribution of immune cells," the authors reported, noting that "IM cases showed better response to immune checkpoint blockade therapy than other subtypes in a clinical trial."
To search for possible IM subtype markers, the team analyzed gene expression, exome sequence, and methylation profiles for 43 ESCC cases from a Phase I safety and tolerability trial of a single-agent anti-PD-1 treatment known as SHR-1210, sponsored by Chinese pharmaceutical company Jiangsu Hengrui.
In that set of cases, the investigators initially identified 48 molecular features ranging from expression profiles and copy number variants to differentially methylated regions that appeared to coincide with the IM subtype.
After whittling that set down to 28 IM-related multiomic features, they used the model to test a subset of 17 patients receiving PD-1-targeting immunotherapy. There, the 28-feature placed seven of the cases in the IM subtype and another 10 cases into other subtypes.
Along with increased overall survival times in the immunotherapy-treated ESCC patients with tumors classified as IM subtype, they found that the 28 selected features predicted immunotherapy treatment response with 90 percent specificity and nearly 86 percent sensitivity.
When the team used the IM features model to classify another 81 ESCC cases profiled for the Cancer Genome Atlas project, meanwhile, it found that 17 cases with IM tumors had other features found in the subtype, including enhanced natural killer cell and macrophage immune cell infiltration and higher-than-usual overall immune scores.
"[D]ue to the high cost for multiomics sequencing, clinical implementation of the multiomics IM classifier will require development of new assays to efficiently profile the 28 features from [formalin-fixed, paraffin-embedded] samples," the authors noted, adding that additional assay development and validation would be needed to take the proposed IM classifier into the clinic in the future.