The problem: Most patients with metastatic non-small-cell lung cancer (NSCLC) do not respond to immune checkpoint inhibitors (ICI). Current biomarkers like PD-L1 expression and tumor mutational burden (TMB) are helpful but not reliable enough on their own to predict who will benefit.
The question: Researchers from Dana-Farber, Memorial Sloan Kettering, and Gustave Roussy examined whether mutations in DNMT3A, a DNA methylation enzyme, could serve as a new predictive biomarker for ICI response.
Why epigenetics: DNMT3A adds methyl groups to DNA, influencing which genes are silenced or expressed. Loss-of-function mutations in DNMT3A reduce this activity and may reshape how immune cells recognize and attack tumor cells.
Study scope: The team analyzed 1,539 NSCLC patients across three leading cancer centers, with comprehensive tumor genomic profiling and follow-up on immunotherapy outcomes.
Patient cohorts: A discovery cohort of 747 patients from Dana-Farber and a validation cohort of 792 from Memorial Sloan Kettering and Gustave Roussy were included. All had received PD-(L)1 inhibitors, alone or combined with CTLA4 inhibitors.
Genomic profiling: Tumors were sequenced using platforms covering 277-447 cancer genes (OncoPanel), 341-468 genes (MSK-IMPACT), or FoundationOne CDx. All DNMT3A loss-of-function mutations were classified as deleterious.
Clonality assessment: To rule out contamination from clonal hematopoiesis (a blood cell phenomenon), the team used INCOMMON to distinguish tumor-derived clonal DNMT3A mutations from subclonal ones that might arise from aging blood cells.
Transcriptomic analysis: NSCLC cell lines and TCGA tumor data were used to link DNMT3A expression levels to immune gene expression patterns, including immune checkpoint, antigen presentation, and inflammatory signaling pathways.
Response rates: In the combined cohort, patients with DNMT3A mutations had a 41.7% objective response rate versus 21.5% for wild-type patients. The difference was statistically significant (P less than 0.001).
Survival benefit: DNMT3A-mutant patients had longer median progression-free survival (9.2 vs. 2.9 months, hazard ratio 0.61) and overall survival (29.6 vs. 13.3 months, hazard ratio 0.66) compared to wild-type patients.
Independent predictor: Multivariable Cox regression confirmed DNMT3A mutation as an independent predictor of improved survival after adjusting for age, sex, smoking status, PD-L1 expression, TMB, and histology.
Predictive not prognostic: Among patients who did not receive immunotherapy, DNMT3A mutation status had no impact on overall survival, indicating the association is specifically tied to immunotherapy response rather than general tumor biology.
Reduced expression: In NSCLC cell lines with pathogenic DNMT3A mutations, both mRNA and protein levels of DNMT3A were significantly lower, consistent with loss-of-function mutations triggering mRNA decay.
Immune gene activation: Tumors with low DNMT3A expression showed higher activity of immune pathways including PD-1 signaling, interferon-gamma (INF-gamma) response, TNF-alpha, and MHC class II antigen presentation, compared to high-DNMT3A tumors.
Immune desert vs. enriched: Tumors with high DNMT3A expression were more likely to be classified as immune deserts with low immune cell infiltration, while low-DNMT3A tumors were more immune-enriched, suggesting DNMT3A suppresses the immune microenvironment.
HLA gene expression: Low DNMT3A expression correlated with higher expression of HLA class II genes and PD-L1, both key to immune recognition, providing a mechanistic link to why these tumors respond better to checkpoint inhibitors.
The concern: DNMT3A mutations are also common in clonal hematopoiesis of indeterminate potential (CHIP), a benign aging-related process in blood cells. Tumor-only sequencing might capture these blood-derived mutations rather than true cancer mutations.
Clonality testing: Using the INCOMMON algorithm, 2.3% of patients had subclonal DNMT3A mutations likely representing CHIP. These were older patients, consistent with the known age-related nature of CHIP.
Cancer-specific signal confirmed: Even after removing subclonal mutations, clonal DNMT3A mutations retained their association with improved immunotherapy outcomes, confirming the biomarker signal is genuinely tumor-derived.
Comparison with TET2: TET2, another gene frequently mutated in CHIP, was not enriched among immunotherapy responders, suggesting the DNMT3A finding is specific and not a general CHIP-related artifact.
New biomarker candidate: DNMT3A mutation status could be added to genomic profiling panels to help identify patients most likely to benefit from single-agent immunotherapy, complementing PD-L1 and TMB testing.
TMB independence: Causal mediation analysis showed the DNMT3A benefit is independent of TMB, meaning these are two separate predictive signals that provide additive information when combined.
Therapeutic target: Since DNMT3A loss changes the immune landscape, DNMT inhibitors currently in clinical development might be used to recreate this effect in patients without mutations, potentially sensitizing tumors to immunotherapy.
Combination strategies: The finding that DNMT3A loss upregulates MHC-I expression in response to interferon suggests a rationale for combining DNMT inhibitors with checkpoint blockade to broaden the population that benefits.
Retrospective design: All cohorts were retrospective, which introduces potential selection biases and limits the ability to draw causal conclusions. Prospective validation is needed before clinical implementation.
Rare mutation frequency: DNMT3A mutations occur in only 4.7% of NSCLC cases, making subgroup analyses statistically underpowered and limiting the clinical reach of this biomarker in an unselected population.
Lack of functional validation: The transcriptomic associations do not yet establish the precise molecular mechanisms by which DNMT3A loss reshapes the immune microenvironment. Cell-based and animal model studies are needed.
Paired blood sequencing needed: The DFCI and Gustave Roussy cohorts lacked paired blood sequencing to definitively identify CHIP. Routine tumor-blood paired sequencing would improve accuracy of DNMT3A mutation classification in future studies.