Clonal Hematopoiesis and Major Adverse Cardiac Events in People With HIV: Insights From the REPRIEVE Trial.
Nov 6, 2025DOI: 10.1161/atvbaha.125.322896Journal ArticleAHA
Written by: Liying XueRomit BhattacharyaMd Mesbah UddinTetsushi NakaoRoger ZouAniruddh PatelSara HaidermotaAbhishek NiroulaVictoria ViscosiDarina PostupakaAarushi BhatnagarPhoebe FinneranRachel BernardoMarissa R DiggsKathleen V FitchSarah M ChuSara McCallumJudith S CurrierCarl J FichtenbaumCarlos D MalvestuttoJudith A AbergGerald S BloomfieldHeather J RibaudoMarkella V ZanniPeter LibbyWhitney HornsbyMichael T LuPamela S DouglasSteven K GrinspoonPradeep Natarajan
Find these authors on Roon
AP
Dr. Aniruddh Patel
General Cardiology · MGH
PN
Dr. Pradeep Natarajan
Preventive Cardiology · MGH
Introduction
The REPRIEVE (Randomized Trial to Prevent Vascular Events in HIV) study is a large, international randomized trial demonstrating a 36% reduction in major adverse cardiovascular events (MACE) in people with HIV (PWH) with low-to-moderate predicted risk of cardiovascular events assigned to pitavastatin compared to placebo1,2. PWH have long been known to be at 50–100% increased relative risk of atherosclerotic cardiovascular disease (ASCVD)3–5, even with undetectable viral loads. Furthermore, a large component of this risk is independent of traditional cardiovascular risk factors such as dyslipidemia, hypertension, obesity, and smoking and partly attributed to residual inflammatory risk mechanisms6,7. The context of the REPRIEVE trial provides a unique opportunity to examine risk conferred by non-traditional risk factors given the population of stably treated PWH with low-to-moderate ASCVD risk.One of the postulated mechanisms of residual inflammation in PWH relates to clonal hematopoiesis of indeterminate potential (CHIP) - an emerging risk factor for both cardiovascular disease and leukemia8,9. CHIP refers to the age-related presence of leukemogenic somatic mutations (typically in DNMT3A, TET2, ASXL1, JAK2, and TP53 among other genes) in hematopoietic stem cells (HSCs) with resultant clonal expansion detectable in peripheral blood, but without overt hematologic abnormalities. These somatic mutations can accrue and expand throughout the lifespan due to exposures and intrinsic features. CHIP is common, reaching ~10% prevalence by age 70, is associated with increased ASCVD event risk10 and has been found to have substantially increased prevalence among PWH across the age spectrum11.Whether CHIP contributes to the increased ASCVD risk among PWH is not yet known. CHIP is more common among older PWH relative to those without HIV11–13. Using blood-based whole exome sequencing in REPRIEVE, we previously showed that traditional CVD risk factors and HIV-specific risk factors were associated with heightened CHIP prevalence also among early middle-aged PWH14. With the completion of the main REPRIEVE trial and availability of new high-coverage sequencing for a targeted CHIP panel, providing increased accuracy and new insights into smaller clones in the most common CHIP genes at a variant allele frequency (VAF) of 1% and lower15, we re-evaluated CHIP prevalence and, for the first time, examined its associations with MACE and its subcomponents.This analysis used both whole exome sequencing (WES) and targeted-sequencing CHIP identification and outcomes data from the REPRIEVE trial to (1) verify the association between the prevalence of CHIP and HIV-specific risk factors, and (2) analyze the relationship between CHIP and incident MACE.
Methods
Because of the sensitive nature of the data collected for this study, requests to access the dataset from qualified researchers trained in human subject confidentiality protocols may be sent to the REPRIEVE trial investigators at kfitch@partners.org.
PWH aged 40 to 75 years without known ASCVD, on stable antiretroviral therapy (ART), not receiving statins, and with low-to-moderate estimated ASCVD risk (as calculated with the 2013 American College of Cardiology / American Heart Association (ACC/AHA) Pooled Cohort Equations [PCE]) were enrolled in the REPRIEVE trial and underwent double-blind randomization to pitavastatin versus placebo, as previously described16,17. Baseline risk factors were assessed at the time of enrollment in the trial, including baseline blood DNA collected among participants enrolled at Advancing Clinical Trials Globally (ACTG) network sites. Enrollment at these sites occurred from April 2015 to February 2019. The trial closed due to the observed efficacy of pitavastatin in April 2023 with participants then followed until a final visit through August 2023. Participants’ countries of residence were grouped using the World Health Organization (WHO) Global Burden of Disease (GBD) scheme as previously described18. The Coordinating Centers and sites obtained approval from the institutional review board and other applicable regulatory entities. All participants received study information, including discussion of risks and benefits, and signed the approved declaration of informed consent. The Massachusetts General Hospital Institutional Review Board approved the present secondary data analyses.
At enrollment, participants provided clinical histories and blood samples, which were assayed for standard laboratory investigations. Data on demographic parameters, medical history, lifestyle behaviors, and HIV-specific parameters were collected as part of REPRIEVE, as previously described1. The trial collected information on the current ART regimen, duration of ART use, and previous exposure to selected medications (abacavir [ABC], tenofovir disoproxil fumarate [TDF], thymidine analogs [zidovudine, stavudine, didanosine], and protease inhibitors [PIs]). Current regimens were characterized based on a nucleoside reverse transcriptase inhibitors (NRTI) backbone with either non-nucleoside reverse transcriptase inhibitors (NNRTI), integrase strand transfer inhibitors (INSTI), or PIs, or as other combinations.
The primary outcome for the REPRIEVE trial was the first occurrence of a MACE - a composite outcome including cardiovascular (CV) death, myocardial infarction (MI), hospitalization for unstable angina, stroke, transient ischemic attack (TIA), peripheral arterial disease (PAD), revascularization of a coronary, carotid, or peripheral artery, or death from an undetermined cause, as previously described1. Hard MACE was defined as the composite of MI, stroke, or known CV death. All suspected MACE outcomes were independently adjudicated by members of the Thrombolysis in Myocardial Infarction (TIMI) group who were unaware of trial-group assignments, as previously described1. The first occurrence of the components of MACE was additionally made available and individually assessed for relationship with CHIP.
Targeted CHIP sequencing assay was performed on the whole blood samples from 4490 REPRIEVE participants by Vanderbilt University Medical Center with Illumina NovaSeq 6000. A panel of 22 CHIP genes was prioritized and enriched in the DNA libraries, aiming for a high depth of coverage ~2000X in these common CHIP gene regions, robustly detecting variants at 1% VAF and lower15. For downstream analysis, a single sample was randomly selected for each of the 45 individuals with multiple sequenced samples, as all duplicated samples showed consistent CHIP detection results.
Somatic mutations were identified from targeted CHIP assay data using GATK Mutect219 software in the Terra platform (https://app.terra.bio/), as previously described20,21. A panel-of-normal (the 1000 Genomes PON) and Genome Aggregation Database (gnomAD) were used to filter germline variants from the putative somatic mutation calls. Mutect2 calls were further filtered, and variants were retained if (1) total depth of coverage (DP) ≥ 100; (2) number of reads supporting the alternate allele (AD) ≥10; (3) ≥ 2 reads in both forward and reverse direction supporting the alternate allele; (4) not germline variants or recurrent sequencing artifacts. CHIP calls from WES were retained if (1) total DP ≥20, (2) AD ≥3, (3) ≥1 reads in both forward and reverse directions supporting the alternate allele, (4) VAF≥0.02 and gnomAD allele frequency <0.001 (for details, see Bhattacharya, R. et al. 2024).
Descriptive statistics are reported using means with standard deviation for continuous variables, and absolute counts with percentages are used for categorical variables. Sociodemographic and clinical variables, CD4 nadir, and ART duration are assessed for their association with the prevalence of CHIP with binomial logistic regression, and log2 normalized continuous maximum CHIP variant allele frequency per participant with linear regressions, adjusting for age, natal sex, and smoking status. Binomial logistic regression models were also used to evaluate the association of MACE with the presence of CHIP, diabetes status was additionally adjusted in these models apart from the abovementioned variables.
All exploratory analyses were adjusted for multiple comparisons using the Benjamini-Hochberg method with a false discovery rate (FDR) of 0.05. All statistical analyses were performed using R version 4.1.0.
Results
Of the total 7769 participants from the REPRIEVE study, 4837 were recruited from ACTG sites and 4490 of them were genotyped with targeted CHIP panel sequencing (Supplemental Table 1; Supplemental Figure 1). The median age of the sequenced cohort was 50 (IQR, 45–55) years, 1653 (36.8%) were female at birth, 2039 (45.4%) were Black, and 1845 (41.0%) were US-based High income sites. The median 10-year Atherosclerotic Cardiovascular Disease risk score by the Pooled Cohort Equations was 3.9% (IQR, 1.7–6.6%), 899 (20.0%) of the participants had CD4 nadir ≥350 cells/per mm3, 3032 (67.4%) had CD4 count >500 cells/per mm3. Additional details on the participants’ characteristics are provided in Table 1.
We defined all clones as any CH driver mutations detected at the full VAF spectrum (range 0.4%-72.0%), small clones as CH driver mutations with VAF <2%, CHIP as variants with VAF ≥2%, and large CHIP as variants with VAF ≥10% consistent with prior literature. Among 4490 individuals, 837 (18.6%) were identified to have CH clones, including 385 (8.6%) individuals with CHIP and 61 individuals (1.4%) with large CHIP. A detailed list of CHIP variants and their corresponding VAFs is provided in Supplemental Table 2. The most frequent CHIP driver mutations were in DNMT3A, TET2, ZNF318, ASXL1, and TP53 (Figure 1A). Among 385 individuals with CHIP variants, 330 (85.7%) carried a single clone, 47 (12.2%) carried two clones, 4 (1.0%) carried three clones, and 1 carried 4 (0.3%) clones (Figure 1B).
CHIP prevalence demonstrated a significant positive association with age across all VAF thresholds (Figure 1C). Each additional year of age increased the odds of CHIP by 8.5% (OR=1.085; P=1.04×10−23) and with large CHIP by 10.1% (OR=1.101; P=4.02×10−7). The group of current and former smokers showed 42% higher odds of CHIP compared to the non-smoker group (OR=1.407; P=2.55×10−3). No significant associations were observed between other clinical and sociodemographic variables and CHIP prevalence (Supplemental Table 3), consistent with previous findings from WES in the REPRIEVE cohort14. Targeted sequencing enhanced the detection of smaller clones (VAF <2%) while also identifying CHIP (VAF ≥2%), resulting in a mean VAF of 3.46% across all detected clones (Figure 1D) compared to a mean VAF of 11.98% in WES14.
In the analysis of 4490 paired samples with both WES and targeted sequencing, targeted sequencing demonstrated a higher prevalence of CHIP driver variants in all the common CHIP genes (Supplemental Figure 2A, 2B). A total of 176 overlapping variants were identified by both targeted sequencing and WES in the same individuals, exhibiting a strong correlation in VAF (Pearson’s r =0.96, P=1.4×10−93) (Supplemental Figure 2C, 2D).
There were 865 unique variants detected by targeted sequencing and 49 unique variants detected by WES. Of the 49 WES-specific variants, 25 were located within the regions covered by targeted sequencing15 (Supplemental Figure 3), and had low alternative allele depth, preventing them from passing the Mutect2 pipeline filtering criteria described in the Method section. A list detailing WES-specific CHIP variants, their location, alternative allele depth and filtering status in the targeted sequencing results are provided in Supplemental Table 4.
Generalized linear regression analyses were conducted to evaluate the associations between the prevalence of sequencing-specific variants and age. No significant evidence supported an association between WES-specific variants and age (OR=1.03; P=0.227). In contrast, targeted sequencing-specific variants, regardless of VAF (OR=1.08, P<2.2×10−16), and those with VAF ≥2% (OR=1.08, P=3.4×10−14) showed a significant positive correlation with age (Supplemental Figure 2E) supporting fidelity of genotyping and quality control procedures. The enhanced sensitivity of targeted sequencing resulted in a higher prevalence of CHIP mutations than previous reports using WES (Supplemental Figure 2F, Supplemental Table 5).
After adjusting for age, natal sex, and smoking status, the prevalence of all clones, CHIP, and large CHIP did not show a statistically significant association with lower CD4 nadir groups compared to the reference group of 350+ cells per mm3 (Figure 2A, Supplemental Table 6). However, when the largest clones’ VAF per individual were analyzed, regression analysis revealed significantly higher VAFs in the 200–349 cells/mm3 (P=6.0×10−4) and <50 cells per mm3 CD4 nadir group (P=1.8×10−2). This association remained significant in a sensitivity analysis adjusting for global burden of disease regions as a known confounding variable (P200–349=2.9×10−3, P<50=1.7×10−3). Although the CD4 nadir groups of 50–199 cells per mm3 also exhibited increased VAFs compared to the reference group, these associations lost statistical significance after adjusting for multiple hypotheses (Figure 2B, Supplemental Table 7). Association tests regarding the duration of ART indicated no significant relationship with CHIP prevalence or the maximum VAF per individual (Figure 2A, 2B).
Among the 4490 sequenced individuals, 142 (3.2%) experienced any primary MACE outcome. Of these, 10 individuals (7.0%) had CHIP, and 4 individuals (2.8%) had large CHIP. Among the 4,348 (96.8%) who did not experience any MACE, 366 (8.4%) had CHIP and 56 (1.3%) had large CHIP. The occurrence of the first primary MACE and its subcategory events were modeled as the outcome of the presence and burden of CH clones. After adjusting for age, natal sex, smoking status, and diabetes, the first primary MACE was significantly lower among individuals with clones of any VAF (OR=0.40; 95% CI, 0.23–0.69; P=1.06×10−3); this effect appeared to be driven by CHIP with VAF<2% (OR=0.38, 95% CI 0.15–0.60, FDR-adjusted P-value=0.04). However, no significant association was found with the prevalence of CHIP with VAF≥2% or large CHIP with VAF≥10% (Supplemental Figure 4A, Table 2).
For all the major adverse cardiovascular events defined in the previous REPRIEVE publication1, the event rate of the MACE subcategories was relatively low within the cohort. Among individuals with CHIP, 5 had hard MACE, 1 had CV death, 2 had peripheral arterial ischemia, 3 had cardiac catheterization or revascularization, 1 had peripheral arterial revascularization, 1 had stroke, and 3 had MI (Supplemental Table 8). Given the number of individuals exhibiting both MACE events and CHIP clones is limited, no significant association was observed between the prevalence of CHIP and MACE events (Figure 3A, Supplemental Table 9). Despite these low event rates, large CHIP (VAF≥10% was found to be associated with events.
Large CHIP (VAF≥10%) was associated with overall MACE (OR=2.92, 95% CI 1.02–8.33, p=4.12×10−2), where 4.4% (4 out of 91) of all individuals with MACE had large CHIP. Out of the 37 individuals with MI, 3 (8.1%) had large CHIP – large CHIP was associated with MI (OR=5.45; 95% CI, 1.59–18.65; P=6.38×10−3). Similarly, among the 30 individuals who underwent cardiac catheterization or revascularization, 3 (10.0%) had a large CHIP (OR=7.96; 95% CI, 2.28–27.74; P=1.13×10−3) (Figure 3A, Supplemental Table 9). With adjustment for FDR, cardiac catheterization or revascularization was the only remaining significant increased outcome with the presence of large CHIP. When the largest VAF per individual was modeled as the continuous independent variable, associations of a 5% increase in the odds of MI for an increase of 1% VAF (OR=1.05; 95% CI, 1.01–1.09; P=9.2×10−3), and a 6% increase in the odds of cardiac catheterization and revascularization for an increase of 1% VAF (OR=1.06; 95% CI, 1.02–1.10; P=2.0×10−3) were found. Hard MACE and stroke were not significantly increased with the higher VAF (Figure 3B, Supplemental Table 10).
All the individuals with any primary MACE events and large CHIP harbored a driver mutation in DNMT3A. The presence of large DNMT3A CHIP was associated with increased odds of first cardiac catheterization and revascularization (OR=14.90; 95% CI, 4.19–53.21; P=3.05×10−5), and of first MI (OR=9.72; 95% CI, 2.78–34.04; P=3.73×10−4). Large DNMT3A CHIP also showed an increase in odds of the combined first primary MACE (OR=3.23; 95% CI, 1.11–9.40; P=0.032) and first hard MACE (OR=5.57; 95% CI, 1.90–16.34; P=1.78×10−3), but they were not statistically significant after FDR adjustment (Table 2, Supplemental Table 11).
No significant association was found for the interaction between pitavastatin and the presence of CHIP on the occurrence of first major adverse cardiovascular events (all clones Pinteraction = 0.37; small clones Pinteraction = 0.22; CHIP Pinteraction = 0.11; large CHIP Pinteraction = 0.86) (Supplemental Table 12). This clinical effect of pitavastatin on MACE for PWH was similar for those with versus without CHIP. No specific baseline ART regiment was significantly associated with the presence of CHIP after FDR adjustment (Supplemental Table 13).
Discussion
To assess the presence of leukemogenic driver mutations across a broad spectrum of clonal fractions among PWH, we used high-coverage targeted sequencing in participants of the REPRIEVE trial. While top genes were similar to the general population, we noted a relative enrichment of PPM1D, TP53, and ZNF318. Our findings suggest large CHIP is associated with increased odds of specific MACE outcomes, such as MI and cardiac catheterization or revascularization, but not with the broader primary MACE outcome, including other CV deaths and strokes. This association was similar among those treated with pitavastatin or placebo. We observed an increase in VAF of CHIP driver mutations in individuals with CD4 nadir levels of 200–349 cells/mm³, suggesting a possible association between clone size and HIV-specific risk factors. These results have several implications described as follows.
First, despite overall heightened CHIP prevalence among PWH, we did not demonstrate a significant positive association between CHIP and the broad composite primary MACE outcome. Exploratory analyses indicated a significant negative association between the prevalence of all CH clones and the first occurrence of primary MACE, which appeared to be driven by the statistically significant lower risk of MACE among individuals with small clones with a VAF<2%. It remains unclear whether this reflects a true biological effect, and further investigation is needed to validate this finding among PWH and the general population. The fact that the REPRIEVE trial was also very successful in preventing events, and that there was a small overlap between individuals with MACE and CHIP may have also limited statistical power. The selection of PWH at low-to-moderate calculated ASCVD risk may have inadvertently also led to selection bias that mitigated risk of MACE among individuals with CHIP, as we previously observed24. This may speak to the fact that even among individuals with inflammatory conditions like chronic HIV infection and CHIP, well-managed traditional risk factor profiles and healthy lifestyles may indeed mitigate risk – particularly when the clone size is small.
By contrast, large CHIP (VAF≥10%) was found to be associated with MACE and several of its components. Large CHIP was associated with a 3-fold increased odds of MACE, a 5-fold increased odds of MI, and an 8-fold increased odds of first cardiac catheterization or revascularization. This was especially true with large DNMT3A clones, which was associated with a 10-fold increased odds of first occurrence of MI and 15-fold increased odds of cardiac catheterization/revascularization. We did not examine associations with specific types of MI, but given the stronger observed with the revascularization outcome, this may be an interesting area of further exploration. Analyses within previous randomized trials also reported the presence of CHIP was significantly associated with first MI and coronary revascularization in the non-HIV general population22. However, these trials had a larger proportion of participants with MACE events given enrichment for older age and cardiovascular risk factors by study design. Our study had a mean age of 49.9 years (±6.4), with only 6.3% of individuals with age over 60 years (Supplemental Table 14), which limits the ability to observe the exponential increase in CHIP prevalence typically seen in older populations with HIV11. Furthermore, by design, participants in REPRIEVE had relatively few cardiovascular risk factors highlighting the need for concomitant atherogenic factors for cardiovascular events in the context of CHIP apart from healthy diet and lifestyle23,24. Given the relatively low number of individuals having both MACE events and large CHIP, we must be cautious to avoid over-interpretation of these study findings linking MACE and revascularization to large CHIP among PWH.
Second, our results suggested a higher CH VAF distribution, rather than prevalence of CHIP, in lower CD4 nadir groups and PWH with the first occurrence of MI and cardiac catheterization or revascularization events, providing insights beyond the historical arbitrary threshold of 2% VAF for CHIP. Several previous studies reported that the presence of CHIP is associated with lower CD nadir in PWH12,25. As previously discussed, our cohort had a much younger age distribution, such an association was not observed with sufficient statistical evidence. Unlike our previous WES findings, where the lowest CD4 nadir count (<50 cells/mm3) was most strongly associated with CHIP prevalence14, this study revealed a stronger link between CD4 nadir levels of 200–349 cells/mm3 and clone size. This suggests that the threshold at which CD4 nadir significantly influences CH expansion may be higher than previously reported. Clone size at the full VAF spectrum was reported to be significantly larger in PWH compared to unaffected population controls26, and our study further supported such findings by showing that larger CH mutation size was linked with lower CD4 nadir as the HIV-specific risk factor. Smaller clones may represent early hematopoietic changes, and our results suggest that they could reflect the effects of immune suppression at a younger age. Similarly, smaller clones may be an early risk marker for MACE. Though smaller clones may not directly affect MACE events, they could serve as an early marker of both immune dysfunction and cardiovascular risk in PWH at a younger age.
Third, this study represents the first comparison of paired WES and a high-throughput targeted CHIP panel sequencing samples analysis. Targeted sequencing demonstrated superior sensitivity and accuracy in detecting CHIP clones by showing around five times more high-confident CH mutations compared to WES14, and enables the detection of small clones with VAF <2% increased the power of modeling VAF as a continuous variable to investigate the effects of CHIP expansion on clinical outcomes. However, as it is limited to a particular CHIP panel that covers the top ~95% of the CHIP driver genes15, some clones detected by WES14 could not be validated in our study because they were outside of the covered regions.
This study needs to be interpreted in the context of its limitations. First, given the relatively young age of the cohort, CHIP prevalence and cardiovascular event rates were low. Prior studies have shown the greater prevalence of CHIP among PWH to be most pronounced in later middle age. Second, all the individuals with large CHIP and MACE events harbored DNMT3A mutations. While this enabled a significant association between large DNMT3A CHIP and MI - consistent with prior reports suggesting DNMT3A-driven hypomethylation may promote cardiovascular outcomes 8,27 - the overall low non-DNMT3A CHIP prevalence limited our ability to conduct other gene-specific analyses. Lastly, bulk next-generation sequencing precluded the ability to characterize hematopoietic cellular subsets that may be differentially impacted by CHIP mutations, which could be explored in future dedicated studies.
In conclusion, our study provided rigorous CH detection with high coverage targeted sequencing in PWH, showing the presence of large CHIP is associated with the first occurrence of a subset of MACE outcomes, namely MI and cardiac catheterization or revascularization, in a young primary cardiovascular prevention cohort of PWH, with only low-moderate traditional cardiovascular risk.
PN
Dr. Pradeep Natarajan
Cardiology · Preventive CardiologyMassachusetts General Hospitalpradeep
