Inflammatory Reprogramming of the Tumor Microenvironment By Clonal Hematopoiesis Is Associated with Clinical Outcomes in Solid Cancer

Background: Clonal hematopoiesis (CH) drives systemic inflammation and chronic disease with aging. In solid tumor patients, CH is common and has context-dependent effects on survival outcomes. CH mutations promote both pro- and anti-tumor immune phenotypes in solid tumor models, although implications in human cancer are still unclear. We present a multi-omics characterization of CH in the tumor microenvironment (TME) to evaluate the biomarker potential of CH in oncology.

Methods: This study used the data of 1,550 patients from the Clinical Proteomics Tumor Analysis Consortium (CPTAC) and 8,927 patients from The Cancer Genome Atlas (TCGA) cohorts. Patients were treatment-naïve and mainly had local disease. Somatic variants in peripheral blood and tumor whole exome sequencing (WES) were detected using GATK-Mutect2, and CH status was ascertained as described previously (PMID: 36652671). Cox proportional hazard models controlled for age, sex, tumor type, metastatic status, and smoking were used to assess overall survival (OS). Differential expression analysis was conducted using DESeq2 and limma for tumor bulk RNA-seq and mass spectrometry proteomics data, followed by gene set enrichment analysis. Immune cell abundance was estimated using CIBERSORTx.

Results: 349 CH mutations were identified in 18.3% of patients in CPTAC, and CH was strongly associated with age (p=1.6x10^-11). CH mutations were most frequent in epigenetic regulators DNMT3A and TET2 (37.8%, n=132; 20.6%, n=72). In 103 patients with CH, the mutation called in peripheral blood WES was also present in tumor WES (CH-Tum). CH-Tum was associated with higher tumor immune infiltrate (p= 0.003) and peripheral blood VAF ≥10% (p=8.6x10^-10). In TCGA, the prevalence of CH was 8.74% (n=780/8,927); however, CH detection was related to WES depth. CPTAC had significantly higher average depth than TCGA (248x vs 94x, p=0), and higher coverage of the most common CH drivers (DNMT3A 215x vs. 77x, p=0; TET2 312x vs 44x, p=0).

In CPTAC, CH-Tum, but not CH, was associated with worse OS (CH-Tum HR = 1.74 [1.13-2.69]; CH HR = 1.12 [0.83-1.50]). CH-Tum was also associated with a reduced likelihood of patients being classified as tumor free at follow up (OR = 0.39 [0.19-0.82]). Across cancers, CH-Tum was associated with elevated expression of inflammatory markers like IL1A and the S100 alarmins, enrichment of HALLMARK pathways including IL6-JAK-STAT signalling, angiogenesis, and the epithelial mesenchymal transition (EMT) in tumor RNA-seq. Furthermore, gene sets predictive of immune checkpoint inhibitor response were upregulated in tumors from patients with CH. CH-Tum correlated with higher macrophage, neutrophil, and regulatory T cell signatures within the tumor.

Gene expression changes and clinical outcomes related to CH varied between cancer types. Glioblastoma multiforme (GBM) stood out with prominent enrichment of numerous inflammatory pathways and macrophage infiltration, accompanied by worse OS with CH-Tum (HR = 3.63 [1.21-10.9]). There was a nominal trend towards GBM cases with CH-Tum adopting the aggressive, immune-rich mesenchymal phenotype (OR=6.55 [0.70-61.1]), as well as significant enrichment of glioma stem cell signatures. Even within the mesenchymal GBM cases, CH-Tum was associated with enrichment of inflammatory signalling, angiogenesis, and mitogenic signalling, trending strongly towards worse OS in multivariate analysis (HR=3.82 [0.99-14.7]).

Conclusion: CH is common in treatment-naïve solid cancer patients, and the infiltration of CH-mutant leukocytes into the TME is associated with poor prognosis across cancers. CH-Tum is correlated with more pronounced inflammatory dysregulation, altered immune infiltrates, and cancer hallmarks like angiogenesis and EMT, suggesting that clonal burden in the blood may not be the sole determinant of non-hematologic outcomes with CH. Rather, the degree of mutant cell infiltration of the tumor – or any tissue – may be a paradigm for understanding the connection between CH and chronic disease. Heterogeneity of outcomes across different cancer types also supports a role for the local microenvironment in dictating the effects of CH, as suggested by the exacerbation of immune-rich, mesenchymal GBM phenotypes observed with CH-Tum. As more is uncovered about CH in the solid tumor context, CH-Tum presents a promising new biomarker for improving outcomes in the era of immuno-oncology.