Clonal hematopoiesis of indeterminate potential and type 2 diabetes mellitus among patients with STEMI: from a prospective cohort study combing bidirectional Mendelian randomization

Abstract

Aim

Both clonal hematopoiesis of indeterminate potential (CHIP) and type 2 diabetes mellitus (T2DM) are conditions closely associated with advancing age. This study delves into the possible implications and prognostic significance of CHIP and T2DM in patients diagnosed with ST-segment elevation myocardial infarction (STEMI).

Methods

Deep-targeted sequencing employing a unique molecular identifier (UMI) for the analysis of 42 CHIP mutations—achieving an impressive mean depth of coverage at 1000 × —was conducted on a cohort of 1430 patients diagnosed with acute myocardial infarction (473 patients with T2DM and 930 non-DM subjects). Variant allele fraction ≥ 2.0% indicated the presence of CHIP mutations. The association between CHIP and T2DM was evaluated by the comparison of (i) the prevalence of CHIP mutations among individuals with diabetes versus those without, (ii) the clinical characteristics delineated by CHIP mutations within the cohort of diabetic patients and (iii) the prognostic significance and correlation of CHIP mutations with mortality rates in T2DM subjects. Furthermore, a two-sample bidirectional Mendelian randomization study was performed using genetic instruments from the genome-wide association study for TET2 mutation CH from the UK Biobank (UKB) (2041 cases,173,918 controls) to investigate the causal relationship with T2DM from the FinnGen consortium (65,085 cases and 335,112 controls), and vice versa.

Results

(i) Most commonly CHIP mutations exhibiting a variant allele fraction of ≥ 2.0% were identified in 50/473 (10.6%) patients with T2DM, demonstrating a greater prevalence compared to non-DM subjects [69/930 (7.4%); P < 0.05] across various age groups. (ii) After multivariable adjustment, the mortality of any CHIP mutations were 2.03-fold higher in DM [adjusted hazard ratio (HR) 2.03; 95% confidence interval (CI) 1.07–3.84, P < 0.05]. (iii) In gene-specific analyses, TET2 somatic mutation presented the highest association with mortality among T2DM (adjusted HR 5.24; 95% CI 2.02–13.61, P = 0.001). ASXL1 CHIP mutation which displayed a striking correlation with cardiac death (HR: 3.14; 95% CI 1.24–7.93; P < 0.05) with consistent associations observed among T2DM subgroup (HR: 4.51; 95% CI 1.30–15.6; P < 0.05). (iv) The correlation between PCSK9 and the Tet2-CHIP mutation was observed in both the T2DM cohort (correlation = 0.1215, P = 0.011) and the overall enrolled cohort (correlation = 0.0578, P = 0.0382). (v) Bidirectional Mendelian randomization studies indicated that the development of T2DM increases the propensity for CHIP. However, CHIP does not subsequently accelerate the onset of T2DM.

Conclusion

CHIP mutations, particularly TET2, are more prevalent in patients with T2DM compared to individuals without diabetes. The presence of these mutations is associated with adverse clinical outcomes, notably increased mortality rates. Moreover, bidirectional Mendelian randomization analyses provide supporting evidence for a potential causal relationship between TET2-related CHIP and the development of T2DM.

Graphical abstract

Central Illustration: The association between clonal hematopoiesis of indeterminate potential (CHIP) and type 2 diabetes mellitus (T2DM): The prevalence of CHIP is notably higher in individuals with T2DM, as demonstrated in a prospective study within an Asian cohort of acute myocardial infarction (AMI). Furthermore, the predictive value of CHIP as a marker for poor clinical prognosis in T2DM has been assessed in this study. Mendelian randomization studies suggest that the development of T2DM may increase the propensity for CHIP, as indicated by findings from the UK Biobank and FinnGen consortium. T2DM, type 2 diabetes mellitus; CHIP, clonal haematopoiesis of indeterminate potential.

Key question

Clonal hematopoiesis of indeterminate potential (CHIP) has been identified as a significant cardiovascular risk factor. Nonetheless, the relationship between CHIP and type 2 diabetes mellitus (T2DM) remains to be elucidated.

Key finding

CHIP was found to be 1.43fold more prevalent in patients with T2DM compared to those without T2DM who were admitted for acutemyocardial infarction (AMI). The mortality rate associated with any mutations in CHIP was observed to be 2.03 fold higher amongi ndividuals with T2DM. Specifically, CHIP mutations predominantly involving the Tet2 gene are more common in T2DM patientsthan in non T2DM subjects, and their presence is linked to poorer clinical outcomes, particularly regarding mortality. Moreover,bidir ectional Mendelian randomization analyses further support a potential causal relationship between Tet2 CHIP and T2DM.

Take home message

CHIP seems to play an important role throughout the disease course of T2DM. Understanding the role of CHIP in T2DM could haveclinical implications for risk stratification and patient care.

Introduction

It is well established that the risk of developing Type 2 Diabetes Mellitus (T2DM) escalates with advancing age, attributable to a decline in insulin sensitivity, chronic inflammation, and deterioration in both pancreatic beta-cell function and mass as one ages [1]. Large-scale next-generation sequencing investigations of blood-derived DNA have unveiled the presence of an expansion of somatic mutations acquired within progenitor hematopoietic stem cells, observed in individuals who appear to be healthy [2]. This phenomenon, referred to as clonal hematopoiesis of indeterminate potential (CHIP), proliferates exponentially with advancing age, ultimately impacting over 10% of individuals aged 70 years and older [3]. The epigenetic regulators DNMT3A, TET2, ASXL1, JAK2, SF3B1/SRSF2, and PPM1D/TP53 epitomize the six most frequently mutated genes intricately associated with CHIP in CVD [2]. Consequently, in our study, we identified these aforementioned genes as the most commonly mutations linked to CHIP. The emergence of CHIP mutations has been correlated with an approximately twofold increase in the risk of developing coronary heart disease (CHD) [3].

CHIP has recently emerged as a significant cardiovascular risk factor [4]. Individuals afflicted with CHIP exhibit a markedly heightened susceptibility to accelerated atherosclerosis, premature cerebrovascular accidents, myocardial infarction, and progressive ischaemic heart failure (HF) compared to their unaffected counterparts [5,6,7]. The underlying pathogenesis has been elucidated as an exacerbated inflammatory response or impaired immune function involving blood cells bearing CHIP mutations, coupled with an intensified activity of immune cells characterized by augmented peripheral secretion of chemokines and cytokines [8,9,10].

Recent comprehensive meta-analyses [11] have indeed highlighted the correlation between CHIP and an elevated incidence of T2DM. Furthermore, mutations associated with CHIP, particularly those found within genes linked to CHD and mortality, have also been implicated in T2DM suggesting a shared pathophysiology related to aging. Nevertheless, the correlation between CHIP and T2DM necessitates further validation within an independent cohort, particularly among individuals diagnosed with acute coronary syndrome. Furthermore, a comprehensive exploration of the clinical characteristics in patients with T2DM, stratified by the presence of CHIP, is imperative.

Consequently, we undertook an investigation into the following: (i) the prevalence of CHIP mutations in individuals diagnosed with T2DM compared to those devoid of T2DM within an Acute Myocardial Infarction (AMI) cohort, (ii) the associated clinical outcomes in patients with T2DM stratified by CHIP status, and (iii) the causal relationship between TET2 mutation-associated CHIP and T2DM through a two-sample bidirectional Mendelian randomization (MR) framework. In this prospective cohort study involving a Chinese AMI population, coupled with analyses from the UK Biobank and FinnGen consortium, we explored the potential contributions of CHIP to both the onset and progression of T2DM.

Methods

Study design, participants and sample collection

The overall scheme of the study is shown in Fig. 1. The longitudinal and prospective study known as China, Risk, Genetics, Archiving, and Monograph (CRGAM) represents a seminal population-based investigation involving patients aged 18 years and older who are diagnosed with ST-Elevation Myocardial Infarction (STEMI) and undergoing primary Percutaneous Coronary Intervention (PCI). This inquiry was initially conceived with the intent to formulate cardiovascular risk scores predicated upon traditional risk factors while simultaneously assessing whether the integration of genetic or biomarker variations could augment these predictive models. Executed within the esteemed Fuwai Hospital in Beijing—a renowned national tertiary care center concentrating on cardiovascular afflictions—this research has diligently enrolled patients who underwent emergency coronary angiography subsequent to an AMI diagnosis between March 2017 and January 2020. Comprehensive accounts of the project's aims along with intricate details pertaining to cohort descriptions have been documented in prior publications [12].

Fig. 1

An overall scheme of the study. T2DM, type 2 diabetes mellitus; CHIP, clonal hematopoiesis of indeterminate potential; VAF, Variant allele fraction; SNP, single nucleotide polymorphisms

The diagnosis and classification of AMI were conducted in strict adherence to contemporary guidelines and universally acknowledged definitions. This encompassed evaluations rooted in clinical presentations, distinctive findings from electrocardiographic assessments, dynamic fluctuations in cardiac enzyme levels, and corroborative imaging evidence. The present cohort constituted all patients diagnosed with STEMI or non-STEMI, who were subsequently stratified for conservative medical management, PCI, or coronary artery bypass grafting (CABG), predicated on the results derived from coronary angiography along with their individual clinical statuses. With regard to the eligibility criteria, adult patients were incorporated into the study under two primary conditions: (1) they had undergone primary percutaneous coronary intervention (PCI), which encompasses stent implantation, thrombus aspiration, and balloon dilation within the coronary arteries; and (2) they provided written informed consent. Conversely, individuals were excluded from analysis if they met any of the following conditions: (1) those who declined participation; (2) those who could not be traced in follow-up communications; (3)those who lacking the measured CHIP parameters; (4) candidates exhibiting a past or present history of malignancies—in conjunction with moderate or severe valvular heart diseases—previous cardiac surgical interventions (with exceptions made only for those subjected to thoracoscopic ablation), as well as instances of moderate liver dysfunction and chronic kidney disease characterized by an estimated glomerular filtration rate below 30 mL/min/1.73 m² have been systematically omitted from inclusion in order to mitigate potential confounding influences between CHIP and other neoplasms or persistent illnesses. This methodological rigor aims solely at elucidating the intrinsic association between CHIP and T2DM.

Demographic data, medication history, and laboratory test results were comprehensively examined across all study populations. The high-sensitivity C-reactive protein (hs-CRP), fasting glucose, creatinine etc. measurement was meticulously extracted from the electronic medical records, reflecting the most recent available test result pertinent to the date when the study sample was collected. Comorbidities were identified through a triad of methodologies: self-reporting by participants, detailed examination of electronic medical records, or clinical diagnoses based on established measured values. The study protocol was rigorously aligned with the ethical principles articulated in the Declaration of Helsinki and secured endorsement from our institution’s ethics committee (No. 2017-866). Written informed consent was duly obtained from all participants involved.

Outcomes and follow-up

The principal outcome of the present analysis was mortality from all causes. Routine follow-up assessments were conducted following patient discharge. At the conclusion of the first month, patients were invited to return to the medical facility for comprehensive evaluations; for those unable to attend in person, a follow-up was executed via telephone interviews conducted by staff members at the institute's information center, utilizing a standardized questionnaire independent of external influence. Subsequently, all patients were scheduled for additional phone interviews at both six and twelve months post-discharge. For individuals who survived beyond one year, subsequent follow-ups would occur on an annual basis. The research team received monthly updates regarding follow-up data, and all reported adverse events underwent systematic assessment by a dedicated group of physicians.

Deep targeted sequencing of clonal hematopoiesis mutations

The peripheral blood sample was collected following the attainment of written informed consent. Subsequently, DNA extraction was conducted on peripheral white blood cells to facilitate the targeted sequencing of 42 genes associated with CHIP (refer to Supplementary Data, Table S1).We meticulously crafted a bespoke panel (Agilent, USA) encompassing a comprehensive array of hematopoietic genes and variant loci, informed by insights gleaned from prior research [3]. The ensuing targeted sequencing was executed by an esteemed commercial entity (Tianhao, China), adhering to the detailed procedural framework delineated below. DNA extraction was conducted using whole blood samples procured from the radial or femoral artery prior to heparinization and PCI. The integrity of genomic DNA was rigorously evaluated via agarose gel electrophoresis; distinct bands devoid of trailing indicated the absence of degradation. Subsequently, both the concentration and quality of the extracted DNA were quantified using Nanodrop 2000/Qubit instrumentation. The requisite specifications included: a concentration threshold of ≥ 20 ng/μL, a total yield of ≥ 1 μg, alongside an OD260/280 ratio ranging between 1.8 and 2.0. These parameters served to guarantee elevated purity levels for the DNA sample, ensuring it was free from any protein or RNA contamination. The isolated DNA then underwent several pivotal steps: size selection, purification, fragmentation, end repair with inherent adenylation at both termini, followed by ligation with unique molecular identifier (UMI) sequencing adapters to facilitate library construction for sequencing purposes. Upon completion of this intricate assembly process, PCR amplification ensued in concert with hybridization capture enrichment aimed at specific genomic regions utilizing an appropriately tailored probe array. Following hybridization diligence, the resultant library underwent thorough washing and purification before encountering subsequent PCR amplification processes culminating in the final utilization-ready DNA library designated for sequencing endeavors. Rigorous quality control assessments were implemented to affirm that all libraries conformed steadfastly to established sequencing benchmarks prior to proceeding on to illumination courtesy of the Illumina Nova platform in paired-end mode leveraging dual reads at 2 × 150 bp that yielded corresponding FASTQ files. In aggregate across all analyzed specimens (n = 1403), we attained an extraordinary average target mean depth coverage amounting to 14,219 × while observing maximum and minimum mean depths reaching remarkable figures of 36,507 × and 4579 × respectively. The Picard software adeptly facilitated the analysis of alignment results derived from single-stranded consensus sequences, following meticulous UMI correction for each sample. The ANNOVAR tool was employed to systematically compare all identified single nucleotide variants (SNVs) and insertions/deletions (InDels) against the most recent population databases, alongside functional and disease-related repositories. This comprehensive analysis appraised the frequency of variants, their functional attributes, conservation status, and associated pathogenicity concerning these SNV/InDel alterations. Variant allele fraction (VAF) ≥ 2.0% indicated the presence of CHIP mutations [13].

Selection of data sources of the two-sample bidirectional MR

All data utilized in the bidirectional MR of this study were ethically obtained and are publicly available. The study followed the Strengthening the Reporting of Observational Studies in Epidemiology Using Mendelian Randomization (STROBE-MR) statement to ensure complete and transparent reporting [14]. Furthermore, a two-sample bidirectional Mendelian randomization study was performed using genetic instruments from the genome-wide association study for TET2 mutation CHIP from the UK Biobank (UKB) [15] (2041 European ancestry cases,173,918 European ancestry controls) (http://ftp.ebi.ac.uk/pub/databases/gwas/summary_statistics/GCST90102001-GCST90103000/GCST90102620) to investigate the causal relationship with T2DM from the FinnGen consortium R10 release data (65,085 cases and 335,112 controls) (finngen_R10_T2D), and vice versa, yielding a step change in our understanding of CH pathogenesis and T2DM.

Statistical analysis

Continuous variables were articulated as mean ± standard deviation, while categorical variables were expressed as counts (%). Differences in clinical characteristics and the prevalence of CHIP across groups were analyzed using Student’s t-test or Mann–Whitney test for continuous variables, and Pearson’s chi-square test or Fisher’s exact test for categorical variables, as appropriate. We meticulously evaluated the risk associated with clinical outcomes in relation to the presence of CHIP mutations, employing Cox proportional hazards regression models. Our findings are articulated through hazard ratios (HRs) accompanied by 95% confidence intervals (CIs). Model 1 was adjusted for age, sex, smoking status, hypertension, history of stroke, myocardial infarction, peripheral artery atherosclerosis, and prior PCI. Model 2 further integrated adjustments for history of CABG, height, weight, and TIMI score in addition to those accounted for in Model 1; the selected confounders were chosen based on previously described criteria [16,17,18]. Kaplan–Meier survival curves were constructed to delineate the incidence rates of all-cause mortality among groups stratified by any significant or commonly CHIP mutations (VAF ≥ 2%). Discrepancy rates of cumulative events were compared utilizing the log-rank test.

To validate the reliability of our findings, we conducted the following analyses: (i) a comparative investigation of CHIP prevalence in T2DM patients against an age-matched cohort of non-T2DM individuals and (ii) an examination of CHIP prevalence within T2DM subjects juxtaposed with that observed in non-T2DM counterparts using various definitions for positive CHIP (including any mutations, commonly CHIP mutations). Furthermore, to elucidate the independent association between CHIP and mortality within the context of T2DM, we refined our assessment by adjusting Model 2 to account for factors such as age, gender, and pre-existing arterial disease history.

In the two-sample bidirectional MR study, we estimated the causal relationship of TET2 mutation CHIP from the UK Biobank with T2DM from the FinnGen consortium, and vice versa. Several MR techniques, which include inverse variance weighting (IVW), MR-PRESSO, weighted median, simple mode and weighted mode approaches, were employed to investigate the causal relationship between the exposures and the outcomes. Cochran's Q-test was performed to quantify the heterogeneity of instrumental variables. The MR-Egger regression intercept and MR pleiotropy residual sum and outlier test (MR-PRESSO) were conducte to test potential horizontal pleiotropy. Leave-one-out sensitivity analyses were used to assess the stabilities of single nucleotide polymorphisms (SNPs). Instrumental variables were selected based on the following criteria: (1) significant genetic variation in the genome (P < 5 × 10^–6) was chosen as a possible instrument; (2) r² < 0.001 and kb = 10,000 to avoid linkage disequilibrium; and (3) F-statistic [19] > 10. A two-sided P < 0.05 was considered statistically significant. All analyses were performed using R Programming Language X64 4.4.1 (R Foundation for Statistical Computing, Vienna, Austria) with R packages “TwoSampleMR”, “MRPRESSO”, and “PheWAS”.

Results

Baseline characteristics

A comparison of the demographic and clinical characteristics of both non-T2DM and T2DM cohort (n = 473) stratified by CHIP status in the prospective AMI cohort (n = 1403) is presented in Table 1. The baseline characteristics of patients with T2DM and those without T2DM in the prospective AMI cohort are detailed herein (see Supplementary Data, Table S2). The mean age was found to be 59.6 ± 12.7 years for subjects devoid of T2DM compared to a slightly older average of 62.1 ± 11.3 years among T2DM patients, exhibiting statistical significance (P < 0.001). Moreover, the male representation was notably higher across both cohorts, with a striking disparity: 83.5% among non-T2DM individuals versus 75.7% within the T2DM group (P < 0.001). Additionally, patients suffering from T2DM exhibited a significantly elevated prevalence of comorbidities relative to their non-diabetic counterparts, particularly concerning hypertension—70.4% as opposed to 61.0% (P < 0.001)—and hyperlipidemia—93.4% compared to an impressive figure of 88.3% in the absence of diabetes (P = 0.002). Patients diagnosed with T2DM exhibited a significantly greater prevalence of comorbidities in comparison to their non-T2DM counterparts. This disparity was particularly pronounced concerning hypertension, where the rates were 70.4% versus 61.0% (P < 0.001), and hyperlipidemia, which presented an alarming prevalence of 93.4% compared to 88.3% (P = 0.002). Moreover, individuals suffering from T2DM demonstrated an elevated incidence of mortality (P = 0.042), along with heightened occurrences of cardiac death (P = 0.042) and ischemic stroke (P < 0.001) when juxtaposed against those without this metabolic disorder. Clinical features according to the presence of clonal haematopoiesis of indeterminate potential (variant allele fraction ≥ 2.0%) in all enrolled cohort (see Supplementary data, Table S3). Clinical features according to the presence of clonal haematopoiesis of indeterminate potential (variant allele fraction ≥ 2.0%) in T2DM (see Supplementary data, Table S4).

Table 1 Clinical features according to the presence of clonal hematopoiesis of indeterminate potential (variant allele fraction ≥ 2.0%) in non-T2DM and T2DM cohorts

The prevalence of clonal hematopoiesis of indeterminate potential in type 2 diabetes mellitus

The comparative analysis of the prevalence of CHIP mutations among non-T2DM individuals and patients suffering from diabetes mellitus is illustrated in Fig. 2. Overall, the most frequently mutated genes associated with CHIP, exhibiting a VAF of at least 2.0%, were identified in 7.4% (69/930) of non-T2DM subjects and in 10.6% (50/473) of T2DM patients (P < 0.05). Details regarding the specific CHIP mutations detected in both non-T2DM individuals and T2DM patients can be found in Supplementary Materials online, specifically Material S2. When stratified by age, both non-T2DM and DM cohorts exhibited an increasing prevalence of CHIP mutations correlating with advancing age—a consistent overall trend. Additionally, it was noted that individuals afflicted with T2DM displayed a higher incidence of commonly CHIP mutations relative to their non-T2DM counterparts across all age brackets, with the exception of those aged between 65 and 69 years (see Fig. 2A). The comparative analysis of the prevalence of CHIP mutations among survival subjects versus patients succumbing to all causes of death is elucidated in Fig. 2B. Notably, a total of 42 CHIP-associated gene mutations exhibiting a VAF of no less than 2.0% were identified in 13.8% (176/1271) of survival subjects, contrasting with an occurrence rate of 22.0% (29/132) among patients facing mortality from various causes (P = 0.012). Individuals harboring the most frequently occurring CHIP along with somatic mutations in DNMT3A, TET2, or ASXL1—collectively designated as DTA—exhibit a significantly elevated incidence of mortality within the entire enrolled cohort (P < 0.05; refer to Fig. 2B). This pattern persists similarly within the T2DM cohort (Fig. 2C). However, this trend appears markedly attenuated among the non-T2DM group (Fig. 2D).

Fig. 2

The frequency of clonal haematopoiesis of indeterminate potential mutations (variant allele fraction ≥ 2.0%) in non-T2DM subjects versus patients with T2DM according to A age. Additionally, we analyzed the occurrence of specific types of CHIP mutations among surviving subjects versus patients succumbing to various causes of mortality across the entire enrolled cohort (B), the T2DM subgroup (C), and the non-T2DM demographic (D). T2DM type 2 diabetes mellitus; CHIP clonal haematopoiesis of indeterminate potential; VAF variant allele fraction

The comparison of the baseline characteristics between age-matched cohorts of non-T2DM individuals (n = 814) and those with T2DM (n = 407) is elaborated upon in Supplementary data online, Table S5. Notably, the prevalence of hypertension was found to be markedly elevated in T2DM patients at 70.5% compared to 60.1% in their non-diabetic counterparts; similarly, hyperlipidemia rates stood at a significant 93.1% versus 88.7%, alongside a higher incidence of PCI history, recorded at 20.6% against 15.4%. Moreover, lipid profiles revealed that T2DM individuals exhibited augmented levels when contrasted with non-T2DM subjects; specifically, triglycerides measured an average of 1.9 mmol/L as opposed to 1.6 mmol/L in the latter group, while fatty acid-binding protein-4 (FABP4) concentrations were notably higher as well—8.6 ng/mL compared to just 7.5 ng/mL.

Furthermore, Plot illustrating the odds ratio of PCSK9 (Sfig1A) and WBC (Sfig1B) in association with the TET2 CHIP mutation, assessed through multivariate logistic regression analysis within a cohort of individuals with T2DM which adjusted for age, sex, smoking, hypertension, hyperlipidemia, height, weight (Supplementary data, Fig. S1). We have measured the inflammatory index including high sensitive C-reactive protein (hs-CRP), proprotein convertase subtilisin/kexin type 9 (PCSK9), Lipoxin A4 (LXA4), Antimicrobial peptide 37 (LL-37), White blood cells (WBC). The correlation of CHIP status and inflammatory markers among all enrolled and T2DM cohort has been evaluated. PCSK9 was correlated with Tet2-CHIP mutation among T2DM (correlation = 0.1215, p = 0.011) and all-enrolled (correlation = 0.0578, p = 0.0382) cohort (Supplementary data, Table S6).

Clonal haematopoiesis of indeterminate potential in type 2 diabetes mellitus: a higher risk of mortality

We have presented the multivariable-adjusted associations of CHIP with clinical outcomes during follow-up among all enrolled cohort and T2DM participants (Table 2). (i) Common vs no CHIP mutation was associated with a multivariable adjusted HR of 1.68 (95% CI 1.04–2.73; P < 0.05) for all-cause death among all enrolled cohort. Hazards associated with TET2, or ASXL1 CHIP were greater for all-cause death (HR: 1.97; 95% CI 1.11–3.50; P < 0.05) or cardiac death (HR: 2.39; 95% CI 1.12–5.09; P < 0.05) when examined separately. Among the six gene-specific subtypes of CHIP scrutinized, pronounced and consistently significant associations emerged prominently with large TET2 CHIP regarding all-cause mortality (HR: 2.12; 95% CI 1.08–4.18; P < 0.05), alongside ASXL1 CHIP mutation which displayed a striking correlation with cardiac death (HR: 3.14; 95% CI 1.24–7.93; P < 0.05) with consistent associations observed among T2DM subgroup (HR: 4.51; 95% CI 1.30–15.6; P < 0.05). (Table 2). (ii) The hazards associated with CHIP mutations were found to be more pronounced in the T2DM cohort compared to the overall enrolled cohort. In T2DM sub-analyses, Any vs no CHIP was associated with a multivariable adjusted HR of 2.07 (95% CI 1.09–3.93; P < 0.05) for the all-cause mortality, with consistent associations observed for common CHIP mutation (HR: 2.40; 95% CI 1.18–4.88; P < 0.05), DNMT3A, TET2,or ASXL1 (HR: 1.51; 95% CI 1.20–5.25; P < 0.05) and TET2,or ASXL1 (HR: 4.09; 95% CI 1.87–8.96; P < 0.001) when examined separately. TET CHIP associated strongest with outcomes in T2DM sub-analyses (HR: 4.62; 95% CI 1.78–11.9; P < 0.01) (Table 2).

Table 2 Multivariable-adjusted associations of CHIP with clinical outcomes during follow-up

Following comprehensive multivariable adjustments for confounding variables— including age, sex, smoking status, hypertension, history of stroke, history of MI, history of peripheral artery atherosclerosis, as well as prior PCI and CABG, alongside height, weight and the TIMI score (Fig. 3 Model 2)—it was observed that CHIP mutations were consistently more prevalent in patients with T2DM compared to their non-T2DM cohort [adjusted hazards ratio (HR) 2.03; 95% CI 1.07–1.74, P = 0.030]. Upon examination of specific CHIP mutations, it emerged that TET2 somatic mutations exhibited the most pronounced correlation with mortality rates among T2DM patients (adjusted HR 5.24; 95% CI 2.02–13.61, P = 0.001) followed closely by TET2 or ASXL1 (TA)-CHIP mutations (adjusted HR 4.06; 95% CI 1.85–8.91, P < 0.001) (Fig. 3A). The prevalence of CHIP mutations in T2DM patients was consistently higher than in non-T2DM subjects: 14.3% versus 16.0% for any CHIP mutations, and 2.3% versus 3.9% for TET2 mutation (Fig. 3B). The higher HR of any CHIP mutations in age-matched T2DM cohort was maintained after confounders were adjusted (adjusted HR 2.21; 95% CI 1.14–4.26, P = 0.019).

Fig. 3

Multivariable logistic regression analyses of clonal hematopoiesis of indeterminate potential prevalence (variant allele fraction ≥ 2.0%) in T2DM compared with non-T2DM in A the entire study cohort, B age-matched cohort. Model 1: adjusted for age, sex, smoking, hypertension, stroke, history of MI, peripheral artery atherosclerosis, history of PCI; Model 2: adjusted for age, sex, smoking, hypertension, stroke, history of MI, peripheral artery atherosclerosis, history of PCI, history of CABG, height, weight, TIMI score. T2DM type 2 diabetes mellitus; CHIP clonal hematopoiesis of indeterminate potential; CI confidence interval; HR hazard ratio

During a median follow-up of 3.96 years, there were 40 (10%) mortality events (cardiac death, 26 events; recurrence MI, 14 events; heart failure, 13 events; revascularization, 74 events; and ischemic stroke 36 events) in T2DM without CHIP and 15 (20.5%) mortality events (cardiac death, 7 events; recurrence MI, 4 events; heart failure, 3 events; revascularization, 12 events; and ischemic stroke46 events) in T2DM with CHIP. The Kaplan–Meier curves vividly illustrate cumulative mortality rates across this median follow-up duration, stratified by the presence or absence of CHIP mutations—specifically distinguishing between those with any form of CHIP mutation and those without—in our entire enrolled cohort comprising both T2DM patients and non-T2DM participants, as displayed in Fig. 4. Notably, within this comprehensive cohort, significant disparities emerged when categorizing the subjects based on their CHIP status: a comparison between any CHIP versus no CHIP revealed striking differences encapsulated by a log-rank P value of 0.0092 as illustrated in Fig. 4A; similarly, differentiating between commonly present versus rarely occurring CHIPs yielded even more pronounced distinctions represented by a log-rank P value less than 0.0001 as shown in Fig. 4B. Furthermore, among the T2DM cohort, significant differences were observed when comparing individuals with any CHIP to those without CHIP (log-rank P = 0.0075 for trend) (Fig. 4C), as well as between commonly occurring CHIP and non-commonly occurring CHIP (log-rank P = 0.0038 for trend) (Fig. 4D). However, these trends weakened in individuals without T2DM. The presence of increasing CHIP mutations was not associated with a higher cumulative incidence of mortality over time (log-rank P = 0.28 for trend) (Fig. 4E), and this association diminished further when categorized into commonly occurring versus non-commonly occurring CHIP mutations (log-rank P = 0.014 for trend) (Fig. 4F). Kaplan–Meier estimation of the outcome events including mortality, cardiac death, heart failure, recurrence of MI, revascularization, and ischemic stroke, across the entire enrolled cohort (Fig. S2A1–4) as well as within T2DM (Fig. S2B1–19) and non-T2DM (Fig. S2C1–4) subgroups stratified by CHIP status (encompassing any CHIP, commonly observed CHIP variants, and large CHIP mutations with VAF ≥ 10.0%) is presented in the Supplementary data online, Fig. S2.

Fig. 4

Kaplan–Meier estimation of mortality, meticulously stratified by the presence of any CHIP without adjusted for covariates across the entire enrolled cohort (A), as well as in patients with Type 2 Diabetes Mellitus (T2DM) (C) and non-T2DM participants (E). Additionally, an analysis focusing on common CHIP occurrences within the overall enrolled cohort (B), T2DM patients (D), and non-T2DM participants (F). T2DM type 2 diabetes mellitus; CHIP clonal hematopoiesis of indeterminate potential

MR studies indicate that T2DM causes TET2-CHIP mutations

We conducted a two-sample bidirectional MR study to elucidate whether the acquisition of TET2-related Clonal Hematopoiesis exerts a causal influence on Type 2 Diabetes Mellitus (T2DM), and conversely, whether T2DM impacts CHIP. Bidirectional Mendelian randomization studies corroborate the notion that the progression of T2DM enhances susceptibility to developing TET2-CHIP. The interplay between T2DM and CHIP was rigorously examined within the MR framework to ascertain causality. A succinct overview of genetic instruments derived from genome-wide association studies pertaining to overall CHIP mutations, as well as specific mutations in DNMT3A and TET2, alongside both small and large CHIP mutations sourced from the UK Biobank, is presented in Supplementary Table S7. Notably, we identified instrumental variables (IVs) from an independent GWAS focused on CH mutations.

A total of 368 independent SNPs were meticulously selected as genetic instruments for T2DM, with each SNP exhibiting an F-statistic exceeding the threshold of 10 (P < 5*10–6, r² < 0.001, window size = 1000 kb) (refer to the Supplementary Table S8 and supplementary material for the mutation list). The IVW analysis revealed a significant causal relationship between T2DM exposure and TET2-CHIP mutations (IVW methods P = 0.0053 OR 1.0012 95% CI 1.0004–1.0021) (Table 3). Furthermore, Mendelian randomization estimates of the effect of T2DM on overall CHIP mutation, DNMT3A CHIP mutations, small CHIP mutations and large CHIP mutations are shown in Supplementary Table S9. Furthermore, there was no indication of heterogeneity among the genetic instruments employed in this study (MR Egger, Q = 287.3529, P = 0.9938; IVW, Q = 291.2631, P = 0.9911) (see Supplementary Table S10). Additionally, we have evaluated horizontal pleiotropy utilizing the MR-PRESSO [20] method. The results of the MR analysis provided the causal estimate, standard deviation (SD) for both the raw analyses. The analysis of outlier-corrected yields produced no results, indicating the absence of any outliers. The outcomes of the MR-PRESSO global test include a list of observed residual sum of squares (RSSobs) along with empirical P-value (P value). A global test p-value of MR-PRESSO greater than 0.05 indicates (P value = 0.995) there is absence of horizontal pleiotropy (see Supplementary Table S11). Nevertheless, it is noteworthy that our investigation did not uncover any association between CHIP mutation exposure and outcomes related to T2DM (refer to Supplementary Table S12 and Fig. S3).

Table 3 Mendelian randomization estimates of the effect of T2DM on TET2-CHIP mutation

Discussion

In this comprehensive study, we meticulously assessed the correlation between CHIP and T2DM in a prospective manner. We also delved into the prognostic significance of CHIP mutations and their impact on mortality within T2DM patients by employing deep-targeted sequencing methodologies to scrutinize CHIP mutations among an AMI cohort. Furthermore, we investigated the causal relationship linking T2DM with CHIP mutations through data derived from both the UK Biobank and the FinnGen consortium (Central Illustration). The principal discoveries from this prospective cohort study involving patients with T2DM are delineated as follows: (i) The prevalence of CHIP mutations, exhibiting a variant allele fraction of ≥ 2.0%, is observed to be 1.4-fold more frequent in T2DM patients compared to their non-T2DM counterparts across age demographics within the AMI cohort. (ii) Mortality associated with any CHIP mutations reveals a staggering increase, being 2.03-fold greater in individuals with diabetes mellitus. Notably, gene-specific analyses demonstrate that somatic mutations in TET2 are significantly correlated with mortality among T2DM patients, presenting an astounding 5.24-fold elevation post multivariable adjustments. (iii) A substantive association between CHIP and mortality was identified during a median follow-up period of 3.96 years for those suffering from T2DM. (iv) Bidirectional Mendelian randomization studies corroborate the notion that the progression of T2DM enhances susceptibility to developing TET2-CHIP; however, conversely, such alterations do not precipitate or expedite the onset of diabetes itself.

Clonal hematopoiesis of indeterminate potential is a condition intricately associated with advancing age; it remains relatively uncommon among younger individuals but exhibits significant prevalence among the elderly populace. Indeed, as many as 10% of adults over the age of 70 harbor clones that are large enough to be detected [3, 4]. A strong confounding of age aside, the most frequently mutated genes identified in acute myocardial infarction were DNMT3A and TET2; these genes are also implicated in various cardiovascular disease and thrombosis disease [5, 6, 15]. Consequently, we meticulously adjusted for variables including hypertension, smoking, stroke history, prior myocardial infarction, peripheral artery atherosclerosis, as well as histories of PCI and CABG. Following comprehensive multivariable adjustments for confounding variables,we observed that CHIP mutations were consistently more prevalent in patients with T2DM compared to their non-T2DM cohort. The prevalence of CHIP mutations in T2DM patients was consistently higher than in non-T2DM subjects. The higher HR of any CHIP mutations in age-matched T2DM cohort was maintained after confounders were adjusted. Clonal hematopoiesis is articulated as a phenomenon wherein a solitary hematopoietic stem/progenitor cell (HSPC) acquires a selective advantage over an extensive spectrum of HSPCs. When associated with somatic mutations in genes linked to myeloid malignancies, such as TET2-mediated clonal hematopoiesis of indeterminate potential, commonly referred to as CHIP, it signifies an augmented risk for both hematological malignancies and cardiovascular diseases [21]. CHIP has recently emerged as a novel cardiovascular risk factor and serves as a harbinger for adverse clinical events such as atherosclerotic cardiovascular disease, chronic heart failure, and severe degenerative aortic valve stenosis [22,23,24,25]. In order to elucidate whether the acquisition of TET2-related CHIP exerts a causal influence on T2DM, we conducted a two-sample bidirectional MR study and found that the progression of T2DM enhances susceptibility to developing TET2-CHIP. Prior investigations have revealed that the prevalence of clonal expansion due to somatic mutations in hematopoietic stem cells escalates markedly with advancing age, implicating CHIP in aging-related chronic diseases [13]. However, there exists relatively scant data delving into the nuanced relationship between CHIP and T2DM, despite its global impact on substantial populations and its anticipated rise—marking it as an increasingly significant healthcare burden [26]. Intriguingly, recent European cohort studies provide compelling evidence linking CHIP to T2DM. In particular, research undertaken by Deirdre K. et al. [11] analyzed genetic data from approximately 17,637 participants encompassing six cohorts through meta-analysis. Bonnefond et al. [27] similarly noted heightened prevalence of clonal mosaicism among patients afflicted with established T2D compared to those without; however, the cross-sectional design of their study hindered elucidation of the temporal directionality inherent in this association. Although the aforementioned studies independently substantiate a plausible connection between CHIP and T2DM, further validation remains imperative to fortify their conclusions—such endeavors might include investigation across diverse non-European ancestries or employing deeper targeted sequencing coupled with stricter definitions of CHIP. In this study, we present valuable independent replication of the recently identified association between CHIP and T2DM within a comparably sized East Asian population. This was achieved through deep targeted sequencing achieving an average depth of coverage exceeding 1000 × alongside elevated variant allele frequency thresholds (VAF ≥ 2.0%).

We conducted a comprehensive analysis revealing that TET2 mutations among clones exhibited the highest correlation with mortality risks in individuals diagnosed with T2DM within our AMI cohort, as determined through gene-specific evaluations. We identified the most pronounced risks in individuals harboring large clones with mutations in TET2. Furthermore, CHIP driven by mutations in the protein-coding gene ALXA4 demonstrated a notable association with cardiac-cause mortality; intriguingly, this correlation was unexpectedly robust when examined alongside inflammatory biomarkers such as PCSK9. Future endeavors to replicate this association and elucidate the underlying mechanisms are imperative. In stark contrast, DNMT3A CHIP—the predominant variant observed within the general population—was not significantly linked to the primary outcome. This finding resonates with observational data suggesting that DNMT3A CHIP correlates less strongly with CAD compared to other CHIP variants like TET2. Nonetheless, prior research conducted on smaller cohorts has indicated that mutations in DNMT3A may indeed signify a graver prognosis in contexts such as ST-segment elevation myocardial infarction and chronic ischemic heart failure. Furthermore, DNMT3A CHIP, recognized as the most prevalent form of CHIP within the general population, was also significantly correlated with our principal outcomes. Moreover, CHIP stemming from mutations in genes responsible for DNA damage repair—specifically TET2 or ASXL1—was found to be associated with all-cause mortality. While this observation aligns with extant evidence indicating that DNMT3A CHIP exhibits a comparatively weaker association with CAD relative to other CHIP variants like TET2, prior studies involving smaller cohorts have similarly suggested that mutations in DNMT3A may herald a more dire prognosis during episodes of ST-segment elevation myocardial infarction and chronic ischemic heart failure [28,29,30]. Recent experimental findings further suggest that alterations in DNMT3A expedite fibrosis progression; thus it stands to reason that the mechanisms by which CHIP exacerbates disease manifestations may diverge across various cardiovascular disease subtypes (for instance: atherothrombosis versus heart failure) [31].

Previous researches have fostered the intriguing hypothesis that age-related chronic ailments—encompassing both CVD and non-CVD conditions—may be substantially influenced by the emergence of somatic mutations that give rise to clonal hematopoiesis [32, 33]. The plausible interaction between CHIP and T2DM is likely mediated by a potential convergence in pathophysiological mechanisms, magnified inflammatory responses, and immune dysregulation. Mutations within the TET2 gene are recognized for their capacity to amplify inflammation among cardiac macrophages, as well as enhance the expression of key mediators such as interleukin (IL)-1β—NLRP3 inflammasome, IL-6, IL-8, along with atherogenic chemokines [34, 35]. Furthermore, deficiencies in DNMT3A have been implicated in the modulation of mast cell reactions to both acute and chronic stimuli; they engage heightened activity levels resulting in amplified production of IL-6, tumor necrosis factor, and IL-13 when exposed to immunoglobulin stimulation in vitro [36, 37]. In conclusion, our observations reveal that the presence of CHIP—after controlling for age and other risk factors associated with T2DM—is correlated with an elevated risk of developing T2DM over time. This association is particularly pronounced for CHIP variants involving TET2 which are previously linked to atherogenic pathologies.

The interconnection between T2DM and CVD is well-documented, as both conditions share a multitude of upstream risk factors such as advanced age, obesity, smoking habits, dietary choices, and various lifestyle determinants. Consequently, it is entirely plausible that an accumulating burden of somatic clonal expansion concerning specific genetic variants triggers the emergence of both health issues. Individuals harboring mutations associated with somatic clonal expansion in genes like DNMT3A, TET2, ASXL1, JAK2, and TP53 face up to double the risk of developing incident Coronary Heart Disease, alongside elevated coronary artery calcification scores compared to non-carriers [1, 38, 39]. Moreover, research utilizing animal models has demonstrated that hematopoietic clonal expansion related to TET2 leads to significantly larger atherosclerotic lesions when juxtaposed with control subjects. Experimental studies in mice have further elucidated how loss-of-function mutations in the TET2 gene within bone marrow cells exacerbate insulin resistance [26, 40] linked to obesity. This effect may be mediated by heightened expression levels of IL-1b facilitated through CHIP, particularly within white adipose tissue. Additional mechanistic investigations underscore the notion that clonal expansion could catalyze atherosclerosis via various local and systemic inflammatory pathways [32, 34, 41, 42]; thus positing inflammation as one potential mechanism underpinning CHIP’s role as a catalyst for age-related chronic ailments including CHD and T2DM. The exacerbated inflammation precipitated by CHIP mutations may instigate alterations in atrial substrate through structural modifications along with electrophysiological and autonomic remodeling—thereby ultimately fueling the onset and progression of CVD. Indeed, Ninni et al.[43] proposed that carriers of CHIP exhibited augmented levels of activated circulating monocytes characterized by gene expressions pertinent to activation or chemotaxis alongside an increase in inflammatory macrophages derived from monocytes—illustrating a more pronounced inflammatory response overall. This study substantiates that traditional blood markers indicative of inflammation fail to comprehensively encompass the cardiovascular risks associated with CHIP and may serve only limited purposes in discerning individuals predisposed to CHIP. Indiscriminate screening for CHIP across general populations is neither feasible nor currently endorsed; however, targeting demographies at heightened risk of developing CHIP could prove more cost-effective. Further investigation is imperative to identify both clinical and germline genetic determinants that signal markedly elevated risk levels, thereby facilitating informed diagnosis of CHIP—an endeavor poised to gain significance as sequencing costs diminish and evidence-based precision therapies focused on CHIP emerge.

By employing a more profound targeted sequencing approach for CHIP mutations, characterized by an average depth of coverage exceeding 1000 ×, and adopting a more rigorous definition of CHIP (with a VAF threshold set at ≥ 2.0%) than previous investigations, this study represents, to the best of our knowledge, the inaugural research conducted within East Asian populations that elucidates an association between CHIP and T2DM. We meticulously examined the poor prognostic outcomes associated with T2DM through the lens of CHIP mutation presence and posited a compelling hypothesis suggesting that T2DM may play a contributory role in the emergence of CHIP mutations.

Conclusion

Our findings illuminate the notion that CHIP embodies a heterogeneous array of gene-specific phenotypes, rather than serving as a mere dichotomous risk factor. The investigators discerned that participants harboring TET2-CHIP experienced a more pronounced increase in major adverse cardiovascular events associated with T2DM, in comparison to individuals bearing other CHIP subtypes or those devoid of any CHIP manifestation. Our results fortify the concept that scrutinizing CHIP phenotypes at both the genetic and mutational tier may pave the way for precision medicine tailored specifically for T2DM patients, akin to the genetic risk stratification prevalent in hematologic malignancies. Individuals possessing TET2 or other higher-risk CHIP subtypes—such as DNMT3A and ASXL1—may represent an optimal cohort for prospective clinical trials employing CHIP-guided novel therapeutics within AMI contexts among those suffering from T2DM.

Limitation

Our study has several limitations. Firstly, given that we defined the presence of CHIP mutations through targeted sequencing of the 42 most commonly identified genes within the GENIE cohort; it is plausible that alternative methodologies employing broader DNA-sequencing approaches may yield differing evaluations regarding CHIP prevalence [44]. Nevertheless, it should be noted that various gene sets can interpret the presence of CHIP mutations differently; thus, the impending beta version of the fifth World Health Organization diagnostic criteria aims to establish a robust consensus in this arena. Secondarily, one must consider potential selection bias inherent in our hospital-based recruitment methodology. Thirdly, while the initial segment of our investigation focuses on a STEMI cohort derived from a Chinese population—wherein we analyze the prevalence rates of CHIP and T2DM—the subsequent analysis employs data extracted from both UK Biobank and FinnGen cohorts, which are predominantly composed of European populations. This disparity raises concerns about ethnic representation and genetic diversity between these two datasets, thereby potentially influencing both generalizability and consistency in our findings. Lastly, it is imperative to interpret our results with caution as they arise from an exploratory analysis; consequently, they necessitate further statistical validation to substantiate their applicability.