Association of clonal haematopoiesis and diabetes with outcomes after autologous haematopoietic cell transplantation

Diabetes mellitus (DM) is a common and serious comorbidity in cancer survivors,1 including those undergoing autologous haematopoietic cell transplantation (HCT). In this high-risk population, DM has been linked to increased susceptibility to infection, treatment-related toxicity, and higher post-transplant non-relapse mortality (NRM).2, 3 One emerging biomarker that may inform risk stratification is clonal haematopoiesis (CH), defined by the clonal expansion of haematopoietic stem cells carrying somatic mutations in leukemogenic genes (e.g., DNMT3A, TET2, ASXL1).4, 5 In non-cancer populations, CH has been associated with an up to 75% higher risk of DM, particularly among individuals with TET2 and ASXL1 mutations.6 Recent preclinical studies demonstrate that CH mutations worsen glucose intolerance, impair β-cell function, and promote systemic inflammation, supporting a biologically plausible link between CH and metabolic disease.6, 7 The relevance of this association in the HCT population, especially in relation to pre-existing metabolic disorders, has not been fully investigated. To address this gap, we evaluated the association between CH and DM and evaluated their combined relationship with NRM after HCT. This retrospective cohort study included 1931 consecutive patients who underwent a first HCT for lymphoma or plasma cell dyscrasias at City of Hope between 2010 and 2016. Targeted pre-HCT DNA sequencing was performed at a sequencing depth of 1000× to detect CH variants (defined as variant allele frequency VAF ≥2%). CH variant annotation and calling were performed following established protocols as previously described.8 Data on patient demographics, BMI (categorised as non-overweight [1 mutation. The most frequently mutated genes were DNMT3A (36.6%), TET2 (15.2%), PPM1D (13.2%), and ASXL1 (8.0%); Figure S1. Patients with CH had significantly higher prevalence of DM compared to those without CH (22.1% vs. 16.3%, p = 0.009; Table S1). In multivariable analysis, CH was independently and significantly associated with increased odds of DM (OR = 1.5, 95% CI = 1.1–2.0). Among CH genes, ASXL1 demonstrated the strongest association: 28.6% of patients with ASXL1 mutations had DM versus 16.3% of patients without CH (p = 0.05; Table 1). Subgroup analyses revealed that 42.9% of Asian patients with DM and normal BMI had CH, whereas in other racial/ethnic groups, DM was more commonly observed among patients with overweight/obese BMI. The 5-year cumulative incidence of NRM was 3.8% (95% CI = 3.0%–4.8%). NRM was significantly higher among patients with CH (9.9% vs. 2.3%, p < 0.001) or DM (7.8% vs. 3.1%, p < 0.001). Patients with CH and DM had the highest NRM (17.6%). In Fine-Gray competing risk models, co-occurrence of CH and DM conferred a 15.5-fold risk of NRM (95% CI = 7.3–33.1) compared with patients with neither condition; Table S2 and Figure 1. This pattern persisted even when therapy-related myeloid neoplasm (t-MN) was included as a competing risk. Finally, patients with CH, DM, and normal BMI demonstrated the greatest vulnerability, with a 26.8-fold increased risk of NRM (95% CI = 9.7–74.1; Table S3). This study provides novel evidence that CH and DM independently increase the risk of adverse health outcomes, including NRM, after autologous HCT. The strong association between CH and DM among individuals with ASXL1 mutations reinforces mechanistic hypotheses linking specific CH driver mutations to metabolic dysfunction through chronic inflammation and immune dysregulation.6, 7 These findings extend emerging preclinical data demonstrating that ASXL1-mutant haematopoietic cells can impair glucose homeostasis and drive systemic inflammatory responses.6 A striking finding was that the adverse association of co-occurring CH and DM was greatest in patients with normal-range BMI. This challenges the prevailing paradigm that higher BMI is the principal determinant of metabolic risk,11 instead suggesting alternative pathways of vulnerability. The results highlight the emerging concept that sarcopenia, rather than excess adiposity, may interact with CH-associated inflammation to heighten risk after HCT.2, 12 They also raise the question of how body composition influences drug metabolism and pharmacokinetics, with the possibility of increased treatment-related toxicity among patients with poorer nutritional status. This highlights future opportunities to optimise chemotherapy dosing strategies using more precise measures of body composition.13, 14 By demonstrating increased risk among patients traditionally considered low-risk by BMI alone, our findings expose an important blind spot in current survivorship frameworks. These insights advance the field by proposing a multidimensional model for pre-HCT risk stratification. Incorporating CH status into routine evaluation, particularly for patients with DM or other features of metabolic syndrome, could enhance identification of individuals at heightened risk of poor outcomes. Objective assessments of body composition (e.g., imaging, bioimpedance analysis) may further improve detection of sarcopenia and unrecognised metabolic compromise.2 Such approaches are particularly innovative in the HCT setting, where current risk algorithms rarely integrate molecular or metabolic biomarkers. The feasibility of incorporating CH testing into standard pre-HCT workups makes this a highly translatable strategy. Beyond risk identification, our findings point to new opportunities for prevention and intervention. Structured exercise programs, metabolic therapies, and anti-inflammatory agents may hold promise for patients with CH and DM, especially in the pre-HCT period. Enhanced post-HCT monitoring of glycaemic control represents another pragmatic strategy. Together, these strategies underscore the translational potential of integrating molecular biology, metabolic health, and survivorship care. The intersection of CH and metabolic disease is likely to become increasingly relevant as HCT populations age and the prevalence of both conditions rises. Our findings raise important questions about the long-term metabolic trajectory of HCT survivors with CH. Prospective studies are needed to evaluate longitudinal changes in glucose metabolism, insulin sensitivity, and systemic inflammation among these individuals. Leveraging dynamic clinical tools such as continuous glucose monitoring, inflammatory biomarker profiling, and polygenic risk scores may enable earlier detection of metabolic decline and guide timely interventions. Several limitations of our study warrant consideration. We lacked granular data on DM duration, severity, and glycaemic control, and objective measures of muscle mass or insulin sensitivity. The limited cohort size and relatively low number of NRM events precluded meaningful gene-specific analyses. The observational design precludes causal inference. Nonetheless, the biological plausibility of our findings, reinforced by emerging mechanistic studies, and their consistency across subgroups provide a compelling rationale for further investigation. Validation in prospective, multi-institutional cohorts will be essential to confirm these results. In conclusion, this study demonstrates that CH and DM each contribute to increased NRM after autologous HCT, with compounded risk among patients not overweight by BMI. Moving forward, integrated clinical tools that combine CH status, metabolic profiling, and physiologic assessments such as sarcopenia screening may enable more personalised and effective risk stratification. As HCT populations become older and more medically complex, redefining survivorship care through this lens may ultimately improve long-term outcomes and quality of life. Rusha Bhandari, June-Wha Rhee, and Saro H. Armenian conceptualised the study, interpreted the data, wrote the initial manuscript, and reviewed and edited the manuscript. Sitong Chen analysed the data and reviewed and edited the manuscript. Raju Pillai provided laboratory resources and reviewed and edited the manuscript. Alysia Bosworth, Artem Oganesyan, Liezl Atencio, Caitlyn Estrada, Mareen Kassabian, and Lanie Lindenfeld abstracted data and reviewed and edited the manuscript. Scott Goldsmith, Michael Rosenzweig, Alex F. Herrera, Matthew G. Mei, Ryotaro Nakamura, Rama Natarajan, and Stephen J. Forman interpreted the data and reviewed and edited the manuscript. F. Lennie Wong conceptualised the study, analysed the data, interpreted the data, and reviewed and edited the manuscript. Saro H. Armenian is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis. The authors would like to acknowledge the work provided by the Leadership and Staff of the CoH Center for Informatics most notably Research Informatics, and the utilization of the POSEIDON data exploration, visualization, and analysis platform including the Honest Broker process. V Foundation for Cancer Research (grant no. DT2019-006 to Saro H. Armenian). Research reported in this publication was supported in part by the National Cancer Institute of the National Institutes of Health under Award Number K12CA001727 (Mortimer) and by the TREC Training Workshop R25CA203650 (Irwin). The authors declare no conflicts of interest. Alex F. Herrera reports research funding from Bristol Myers Squibb, Genentech, Merck, Pfizer; consultancy from Bristol Myers Squibb, Genentech, Merck, AstraZeneca, Takeda, Lilly, Genmab, Pfizer, Abbvie, and Allogene Therapeutics. No disclosures were reported by the other authors. The data that support the findings of this study are available from the corresponding author upon reasonable request. Table S1. Demographic and clinical characteristics of patients. Table S2. Five-year cumulative incidence of non-relapse mortality by CH and diabetes status. Table S3. Five-year cumulative incidence and risk of non-relapse mortality by CH, DM, and BMI categories. Figure S1. Characteristics of clonal haematopoiesis (CH) mutations by diabetes status. Please note: The publisher is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.