Clonal hematopoiesis has long been defined by the presence of known driver mutations such as DNMT3A or TET2. But recent single-cell whole-genome studies have shown that most clonal expansions in aging blood occur without these canonical drivers. Around this work in the archive are studies that use machine learning to predict disease from routine blood tests, reconstruct the genealogy of tumor clones, and describe evolution as a patchwork quilt. Our paper stands apart: it does not predict disease or build a tree, but instead uncovers the inherited architecture of the mutation accumulation process itself. This is a step back from the clinic to the foundation, but a step that explains why some people accumulate clones faster than others.
The authors took whole-genome blood sequences from 791,067 people across two biobanks: 481,079 from UK Biobank and 309,988 from All of Us. Instead of searching for driver mutations, they counted 'passenger' mutations: single somatic variants that appear in only one person and are almost certainly neutral. To filter out technical noise, they kept only variants with support from 4-8 reads out of 30x coverage, which corresponds to a clone making up a quarter or half of blood cells. Then, using regression, they removed the influence of age, sex, and smoking, yielding a quantitative trait called 'passenger mutation burden.' For this trait, they performed a genome-wide association study (GWAS) in 481,079 UK Biobank participants, a rare-variant analysis (RVAS), and analyses stratified by sex and ancestry. Candidate genes were prioritized using an empirical Bayesian approach that combines variant pathogenicity, data on open chromatin in blood stem cells, and colocalization with expression loci.
The meta-analysis of the two cohorts revealed 81 loci associated with passenger mutation burden at a significance level of P < 5×10⁻⁸. Of these, 39 overlapped with known clonal hematopoiesis loci, including TERT, TCL1A, ATM, CHEK2, and BCL2, while 42 were new. Among the new ones are NFE2, the master transcription factor for erythroid and megakaryocytic maturation; MSANTD2, a regulator of stem cell adhesion; PTPN7, a phosphatase in the MAPK pathway; and CD163, a macrophage receptor linked to anti-inflammatory polarization. Rare-variant analysis identified 11 candidate genes, including MBD2, a protein that reads DNA methylation marks. Deleterious variants in MBD2 increased mutation burden (β = 0.353, P = 10⁻¹⁸·⁵¹). This closes the methylation axis: DNMT3A writes marks, TET2 erases them, and MBD2 interprets them. Stratification by sex yielded 14 new loci, including the female-specific RMND5B and the male-specific OLIG3. Ancestry analysis in All of Us revealed an African locus KAZN/TMEM51 and an East Asian locus PTPRN2 with a large effect (β = 0.538). The observed mutation burden was more strongly associated with hematologic malignancies and mortality than polygenic risk, and spatial analysis uncovered geographic clustering not explained by income, air quality, or chemotherapy.
This work differs from neighboring studies by using a quantitative passenger mutation trait instead of the binary presence of drivers. This allowed the discovery of more loci with a smaller sample size than previous driver-centric GWAS. For the first time, it shows convergence of common and rare germline variants with somatic drivers on the same biological axes: DNA methylation, RNA splicing, DNA repair, and mTORC1 signaling. The discovery of MBD2 as a rare germline factor extends the methylation axis from 'writing' and 'erasing' enzymes to 'reading' proteins. The spatial heterogeneity that persists after accounting for known factors points to unmeasured environmental exposures that may be as important as genetics.
The field may move toward integrating richer exposure data: radiation, fine particulate matter, access to medical care. As non-European samples in biobanks grow, it will become possible to test how well the discovered loci transfer to other populations. Single-cell technologies such as ReDeeM will allow linking germline variants to specific stem cell subpopulations and their clonal dynamics. Combining polygenic risk and observed mutation burden into a single tool for stratifying acute myeloid leukemia risk is also promising.
The results will influence the genetic epidemiology of blood aging and the development of risk stratification tools for hematologic malignancies. They also provide new targets for experimental validation in stem cell biology.
Next steps include functional validation of the new candidate genes in cellular models of hematopoiesis and extending phenome-wide associations to non-European groups. It is also necessary to identify the specific environmental factors responsible for the geographic clustering.
In the field of clonal hematopoiesis, it remains unresolved which mechanisms give clones a selective advantage without canonical driver mutations. This work directly links that problem to germline variation in epigenetic regulators and signaling pathways, offering specific genes for testing. Another unresolved question is why clonal hematopoiesis manifests differently in men and women and across populations; stratified analyses provide the first clues but require larger samples.
The MBD2 gene identified in this work encodes a protein that reads methylation marks on DNA. If DNMT3A is a writer and TET2 an eraser, then MBD2 is a reader that interprets what is written. Deleterious variants in this 'reader' increase mutation burden just as somatic mutations in the 'writer' or 'eraser' do.
The idea that neutral 'passenger' mutations can serve as clocks ticking out the age of a clone echoes the concept of molecular clocks from Greg Egan's novel Permutation City, where cell age is determined by the accumulation of random changes.
Not published on arXiv. Here are its page, the PDF and the formalization: you can check it yourself, not only read our account.