Viewing RCC with a DNA methylation lens ENHANCES understanding of ICI resistance

Summary

Clear cell renal cell carcinoma tumors with an enhancer demethylator phenotype (TED) harbor a worse prognosis and derive less clinical benefit from immunotherapy. The TED phenotype may help predict immunotherapy resistance.

In this issue of Clinical Cancer Research, Lu and colleagues define a subgroup of clear-cell renal cell carcinoma (ccRCC) tumors with high enhancer demethylation (TED+) that have a poor prognosis and associate with resistance to immune checkpoint inhibition (ICI) (1). While their study is not the first to examine DNA methylation patterns in ccRCC, prior work has focused primarily on the heterogeneity of DNA methylation between ccRCC tumors or the correlation between DNA methylation patterns with specific genomic alterations (2,3). For example, the TCGA found that SETD2 mutations were associated with DNA hypomethylation at non-promoter regions (2), which might have been expected given that SETD2 regulates the histone mark, H3K36me3, that recruits the DNA methyltransferase, DNMT3A. Moreover, in an unbiased manner, Sato et al. defined 3 clusters of ccRCC tumors with high, intermediate, and low levels of DNA methylation with the hypermethylated genes in the high methylation group being enriched for genes regulated by polycomb repressive complex 2 (PRC2) (3).

Lu and colleagues also asked how global DNA methylation patterns differ between ccRCC tumors (1). However, they put a twist on prior work. Instead of starting with the most variably methylated probes across tumors they first established a set of CpG sites that are demethylated in cancer (versus normal kidney) by selecting for probes with high methylation levels in normal tissues that also have variable methylation levels in tumors. Through unsupervised clustering of the TCGA-KIRC cohort using this set of probes, they found three groups: hypomethylator, semimethylator, and hypermethylator that have distinct differences in survival. These findings were validated in an independent cohort (GES61441). Focusing on the hypomethylator phenotype, the group noted that in these tumors, the hypomethylated regions are significantly enriched in enhancers, which play key roles in transcriptional regulation. Reclustering of the TCGA cohort using only demethylated CpG probes located in enhancers, the authors identified two subgroups stratified by high and low levels of enhancer demethylation, TED+ and TED- respectively. TED+ tumors impressively associate with worse prognosis, more advanced stage, and were also enriched in tumors with sarcomatoid and rhabdoid histologic differentiation (Fig. 1).

Fig. 1. Clinical, immune, and histologic associations of Tumors with high Enhancer Demethylation (TED+).

ccRCC tumors can be divided into subgroups stratified by high and low levels of enhancer demethylation, TED+ and TED- respectively. TED+ tumors are associated with worse prognosis, high level of immune infiltration, and enriched with sarcomatoid and rhabdoid histology. (Adapted from an image created with BioRender.com.)

They noted that the differentially methylated regions unique to the hypomethylated (TED+) tumors mapped to genes involved in immune regulatory pathways. Further investigation showed that TED+ tumors have higher MeTIL (methylation of tumor infiltrating lymphocytes) scores (4), which has been shown to be a highly sensitive measure and representative of TIL content of tumors as well as hypomethylation of probes involved in the interferon gamma response and IL-6/JAK/STAT HALLMARK pathways. TED+ tumors also had higher RNA expression levels of immune checkpoints but also higher levels of immune suppressive cells such as regulatory T cells (Tregs) and M2 macrophages. TED+ tumors had increased stromal infiltration of both myofibroblastic cancer associated fibroblasts and inflammatory CAFs (Fig. 1). In keeping with these findings, the group noted that TED+ tumors have a higher TIDE (Tumor Immune Dysfunction and Exclusion) score, which has been shown to represent T cell dysfunction in immune infiltrated tumors (5). Finally, while TED+ and TED- tumors had equivalent tumor mutational burdens (TMB), genomically, TED+ tumors had a lower proportion of tumors with PBRM1 mutation (than TED-) but a higher proportion of tumors with 9p21.3 deletion. In aggregate, these immunogenomic features demonstrate that TED+ tumors are more T cell inflamed yet actively immune suppressed as well.

The immunogenomic characteristics of TED+ tumors would predict for immunotherapy resistance. Lu and colleagues explored this hypothesis using a small set of samples of patients treated with the combination of ipilimumab and nivolumab (n=46) from the BIONIKK trial, a biomarker-driven, open-label, non-comparative, randomized, phase 2 trial of nivolumab, nivolumab–ipilimumab, and VEGFR-tyrosine kinase inhibitors (6). Only five of the 46 patients were classified as TED+. Nonetheless, they found that patients with TED+ tumors had significantly worse PFS, time to second treatment, and OS compared to TED- tumors (n=41). In a second cohort of tumors treated with sunitinib, the Beuselinck cohort (n=102), no difference was seen in OS, PFS, and response (to sunitinib) in TED+ versus TED- tumors, suggesting that the association of TED+ tumors with immunotherapy resistance is not merely a generalized predisposition to treatment resistance.

In an attempt to understand how the TED+ phenotype associated with immunotherapy resistance the authors examined a list of 875 demethylated probes in the TED+ tumors and found that one of these probes, cg23602092, is located in the promoter shore of Ten-Eleven Translocation 1 (TET1), a methylcytosine dioxygenase that catalyzes the conversion of 5-methylcytosine (5-mC) to 5-hydroxymethlcytosine (5-hmC), the initial step in active DNA demethylation. cg23602092 was significantly hypomethylated in TED+ tumors in all three of the cohorts examined (TCGA, GES61441, and BIONIKK). The authors hypothesize therefore that promoter hypermethylation upregulates TET1 expression leading to increased enhancer demethylation and the TED+ phenotype.

The work by Lu and colleagues is significant in that it provides a potentially novel predictive factor, TED+, for ICI resistance. Current predictors of ICI response or resistance in RCC have largely focused on genomic DNA alterations (i.e. PBRM1), characteristics of the immune microenvironment (i.e. immune gene signatures or T cell receptor clonality), or regulation of endogenous retroviruses (7–13) rather than epigenetic characteristics of a tumor. It will be important to discern the value added of the TED+ phenotype over these other immunogenomic characteristics that are more commonly being incorporated as biomarkers in clinical trials.

A couple of key unanswered questions remain from their work. First, is the enhancer hypomethylator phenotype derived from ccRCC tumor cells or from non-tumor cell autonomous components of the microenvironment? It is notable that the TED+ tumors contain a higher level of immune cells as assessed by multiple orthogonal methods. Moreover, some of the putative transcription factors identified to be driving a TED+ phenotype (based on upregulated regulon activity) are primarily found in immune cells (i.e. IRF4, POU2F2). Ultimately however, understanding which cell types are contributing to the TED+ enhancer demethylator phenotype while interesting, may not be necessary to understand prior its implementation as a predictive assay for ICI resistance.

Second, if the enhancer hypomethylation changes are tumor cell autonomous, what might be driving this phenotype? While the authors put forth a very plausible explanation (increased TET1 expression due to promoter hypomethylation), since TED+ tumors have global hypomethylation, the TET1 promoter may merely be one of many demethylated CpG sites and not necessarily causal. It is also notable that loss of 9p21.3 is highly associated with the TED+ phenotype. The 9p21.3 locus not only contains the CDKN2A/B locus but also the MTAP gene that encodes MTAP (S-methyl-5’-thioadenosine phosphorylase), which is an enzyme critical for the purine/methionine salvage pathway and its inactivation has been shown to alter DNA methylation levels (14). Therefore, while the enhancer hypomethylator phenotype may certainly be driven by TET1 overexpression, follow up functional studies will be critical to firmly establish this link.

The authors should be commended, as report of the TED+ phenotype is novel, aligns with the notion that VHL deficiency drives enhancer activation in ccRCC, and its potential utility in predicting resistance to ICI is especially exciting (15). Future work should assess the value added of the TED+ phenotype over the predictive information captured through other common immunogenomic assessments such as TMB, immune gene signatures, or TCR clonality. Hopefully, the TED+ phenotype adds independent information that enhances and further refines our ability to offer the right treatment to the right patient.