This analysis used UK Biobank data to examine age-related hematopoietic clones and mosaic sex chromosome loss. The full sample comprised 450,587 participants, of whom 46,324 had plasma proteomics measurements available. Analyses were performed under UK Biobank application 170256; UK Biobank operates with ethical approval and all participants provided informed consent.
The investigators evaluated three age-associated hematopoietic alterations: clonal hematopoiesis of indeterminate potential (CHIP), mosaic loss of chromosome Y (mLOY) and mosaic loss of chromosome X (mLOX). All three alterations demonstrated an age dependence, meaning prevalence increased with advancing chronological age. Despite this shared relationship with ageing, the alterations diverged in downstream systemic signatures and disease associations in the studied cohort.
Although CHIP, mLOY and mLOX are all age-associated, the study reports distinct spectra of incident disease tied to each alteration. Specifically, non-DNMT3A CHIP — that is, CHIP excluding DNMT3A-mutant clones — associated with a broad multisystem disease burden across organs and systems. In contrast, mLOY exhibited a narrower disease profile, with stronger associations reported for respiratory, musculoskeletal and cardiovascular conditions. By contrast, mLOX lacked broad age-related disease associations in these analyses.
These observations indicate that the presence of a hematopoietic clone or sex chromosome mosaicism does not simply reflect a generic ageing phenotype but instead may confer alteration-specific vulnerabilities to particular disease categories.
The study mapped clone burden (measures of mosaicism or size of CHIP clones) to circulating proteomic programs using the plasma proteomics subset. Distinct proteomic signatures were identified for each alteration type:
mLOY associated with proteins and pathways involved in neutrophil degranulation and extracellular matrix remodeling. This program aligns with inflammatory and tissue-remodeling biology that could plausibly link to the observed respiratory and musculoskeletal associations.
Non-DNMT3A CHIP linked to proteomic patterns reflecting myeloid immune regulation. These proteomic changes are consistent with altered innate immune signaling and myeloid cell functional programs tied to multisystem disease vulnerability.
mLOX unexpectedly associated with proteomic signatures indicative of cytotoxic lymphocyte and NK-cell responses. This finding suggests a distinct immunologic footprint for X chromosome mosaic loss that differs from the myeloid-dominant signals seen with CHIP and mLOY.
Proteomic mapping therefore differentiated alteration-specific systemic programs rather than a single, shared proteomic ageing signature.
To interrogate potential causal relationships between the proteins identified and disease outcomes, the authors applied Mendelian randomization analyses. These analyses provided support for selected protein–disease relationships uncovered in the proteomic mapping. The report indicates Mendelian randomization supported some of the hypothesized links between circulating proteins associated with particular clones and relevant disease risks, reinforcing the notion that proteomic signatures may reflect mechanistic pathways connecting specific hematopoietic alterations to disease.
The source did not provide exhaustive detail on which specific proteins or which disease endpoints were supported by Mendelian randomization in the summary text available here; those specifics were not reported in the excerpt.
Using a large UK Biobank sample with plasma proteomics in a subset, the authors conclude that age-related hematopoietic clones and mosaic sex chromosome loss are not interchangeable markers of biological ageing. Instead, CHIP (particularly non-DNMT3A CHIP), mLOY, and mLOX each define distinct systemic proteomic programs and associate with different patterns of disease vulnerability. These findings imply that clinical and epidemiologic assessment of age-related hematopoietic alterations should consider alteration-specific biology when estimating disease risk or exploring potential targeted interventions.
Ethical and data availability notes: the analyses used UK Biobank resources under application 170256; UK Biobank approval and participant consent were documented. The preprint posted on medRxiv is dated August 31, 2026. The authors declared no competing interests in the source record.