Given that only a limited number of CRC patients benefit from ICI therapy, it is essential that biomarkers of ICI efficacy be identified in order to guide appropriate treatment planning. At present, however, predicting ICI responses remains a major clinical challenge. Herein, we utilized ssGSEA scores for 29 immune-related gene sets in order to separate CRC samples from five datasets into three subtypes (Immunity_L, Immunity_M, and Immunity_H) via a hierarchical clustering approach.
Tumor immune subtyping can offer important insights into the TIME in CRC patients and may help guide immunotherapeutic interventions in these individuals. We found that samples of the Immunity_H subtype exhibited significantly higher immune and stromal scores as well as significantly lower tumor purity. These Immunity_H tumors were also “hot” from an immunological perspective, with higher levels of DC infiltration, macrophage infiltration, IFN signaling, inflammation, cytolytic activity, and CTLA4/CD274 expression. These Immunity_H patients therefore had more favorable clinical outcomes, and may be more responsive to certain immunotherapeutic interventions.
MSI-H tumors have been shown to be ideal candidates for treatment with specific immunotherapeutic agents. To date, there have been multiple trials conducted in MSI-H/dMMR metastatic CRC patients with previously-treated disease (KEYNOTE-164[39], KEYNOTE-158[40], and CheckMate-142[9]), and these achieved objective response rates (ORRs) of 28% − 52%, with 24% − 59% progression-free survival (PFS) and 72% − 76% overall survival (OS). While promising, these patients are a minority of the CRC patient population, with most patients having pMMR/microsatellite stable (MSS) tumors that are not responsive to immunotherapy. As such, novel treatment strategies are urgently needed to increase the immunogenicity of tumors in these patients. We found that MSI patient samples primarily clustered in the Immnity_H group, suggesting that patients meeting both of these criteria would likely be the most responsive to immunotherapy treatment.
Combination immunotherapy treatments have been suggested to prolong survival in melanoma patients with BRAF mutations [41], but further study is needed to understanding the relationship between specific mutations and anti-PD1 responses in CRC patients. In this study, we found that almost all tumors exhibiting KRAS mutations were MSS and clustered in the Immunity_L subtype, whereas the majority of tumors with BRAF V600E mutations were MSI-H tumors in the Immunity_H group. BRAFV600E are known to be more common in MSI-H tumors (38.9%) relative to MSI-low tumors (9.3%) [42]. Liao et al. found that oncogenic KRAS mutations can drive and immunosuppressive gene expression program via repressing the expression of IRF2, increasing CRC tumor resistance to ICI treatment [23]. This suggests that a combination of KRAS mutations and an Immunity_L gene expression profile are associated with poor immunotherapy responses, although further trials will be needed to validate this hypothesis.
TMB can be analyzed to predict the responsiveness of MSI-H metastatic CRC tumors to ICI therapy [36]. One study found that CRC patients with high TMB had a significantly longer OS than did patients with low TMB (hazard ratio [HR], 0.73 [95% CI, 0.57 to 0.95]; P = 0.02)[43]. High TMB samples in the present study included the majority of samples in the Immunity_H group as well as some Immunity_M samples, with those in the former group being likely to be responsive to immunotherapy.
We identified LCK, GNGT2, CD3G, CCR4, and CCR5 as hub genes in our PPI network. CCR5 is able to promote macrophage polarization towards a more robust anti-tumor phenotype in those with hepatic CRC metastases [44], while LCK and CD3G can promote memory CD4 + T cell responses in the CRC TIME [45]. We also found that gene expression profiles associated with T cell activation, lymphocyte differentiation, and leukocyte cell-cell adhesion were significantly enriched in Immunity_H samples relative to Immunity_L samples, as were chemokine signaling, CAMS, and the hematopoietic cell lineage. This suggests that tumors of the Immunity_H subtype were the most immunogenic.