Distinct characteristics of lymphoid and myeloid clonal hematopoiesis in Word Trade Center first responders

Introduction

In the aftermath of the 9/11 terrorist attacks on the World Trade Center (WTC), over 91,000 individuals were involved in the rescue and recovery efforts, clean-up of debris and restoration of essential services (Rescue & Recovery Workers - 9/11 Health). These included first responders such as firefighters, police officers, and paramedics, in addition to operating engineers, steel workers, railway tunnel workers, telecommunications workers, sanitation workers, medical examiners, and volunteers. Many had no prior training in civil disaster response (Herbert et al. 2006). Due to their unprecedented exposure to a complex mix of known or suspected airborne carcinogens, there were justified concerns regarding their elevated cancer risk. WTC responders were exposed to benzene, formaldehyde, asbestos, silica, cement dust, glass fibers, heavy metals, polycyclic aromatic hydrocarbons, and polychlorinated biphenyls, polychlorinated dibenzofurans, and dioxins (Lioy and Georgopoulos 2006). Exposures to these agents may increase cancer risk. However, the association between exposure to this complex mix and the development of hematologic malignancies is not well characterized.

Recently, it has been discovered that a condition called clonal hematopoiesis (CH) elevates risk for hematological neoplasm, cytopenia, cardiovascular disease (CVD), infection and all-cause mortality (Dawoud et al. 2021; Jaiswal et al. 2014, 2017; Jaiswal and Ebert 2019; Niroula et al. 2021; Sperling et al. 2017; Zekavat et al. 2021). Individuals with CH harbor clonal genomic mutations in their blood cells associated with certain hematologic malignancies, yet with no detectable hematologic disorders nor unexplained persistent cytopenia (Heuser et al. 2016). A subset of CH, termed clonal hematopoiesis of indeterminate potential (CHIP), involves having a clonal population of blood cells that carry a point mutation or short insertion/deletion with a variant allele fraction (VAF) ≥ 2% in a gene recurrently mutated in blood cancers. CHIP onset is strongly associated with age, reflecting the gradual accumulation of somatic mutations and quiescent clones which accompany cellular senescence (Genovese et al. 2014). CHIP has been shown to occur in the myeloid lineage (M-CHIP) (Abelson et al. 2018; Desai et al. 2018), and studies have mostly focused on genes known to be recurrently mutated in myeloid malignancies (Bick et al. 2020b; Jaiswal et al. 2017). However, CHIP can also occur in the lymphoid lineage (L-CHIP), and contribute to the risk of lymphoid malignancies, an area that is less documented (Agathangelidis et al. 2018; Condoluci and Rossi 2018; Niroula et al. 2021; Singh et al. 2020; von Beck et al. 2023; Weigert et al. 2012).

Here, we studied the prevalence of M- and L-CHIP (M/L-CHIP) in 350 WTC responders using ultra deep (250X) whole-exome sequencing (WES). Our study has several unique aspects. First, we studied participants from the WTC Health Program General Responders Cohort (GRC), which is a relatively diverse group of workers with respect to their occupational titles. Second, we leveraged the wealth of additional data we collected on these responders, including demographics (gender, race/ethnicity), WTC exposure, blood counts (including various immune cell subsets), smoking history, body mass index (BMI), mental, cognitive and general health characteristics, as well as HLA alleles derived from the WES datasets, to study their association with M/L-CHIP. Third, utilizing WES enabled us to study the full spectrum of both M/L-CHIP mutations without being a priori restricted to a particular gene-set. This is critical, as the literature and the definition of what constitutes CHIP genes and mutations continues to evolve, our raw data files will continue providing a rich resource for future investigations. Overall, our results demonstrate that deep sequencing of M/L-CHIP mutations, integrated with other metrics, enables a more personalized risk-adapted assessment for CHIP associated etiologies in persons exposed to airborne carcinogens. We anticipate that the study findings will inform the development of CHIP-related diagnostic measures which can be incorporated into screening programs for cancer and other inflammation related conditions.

Methods

Results

To study the prevalence of CHIP mutations in WTC responders, we performed deep WES at 250X on 350 banked blood samples from 345 participants (Figure 1A). The participants were aged 48-90 years (median, 59 years), with no previous hematologic malignancy at enrollment, with demographics detailed in Table 1A. We applied rigorous QC metrics and filters (Figure 1B, Supplementary Table S1), to reveal M/L-CHIP prevalence, which we then associated with age, ancestry, exposure, HLA zygosity, and other clinical, laboratory, mental and cognitive data. Finally, we compared CHIP prevalence in the WTC cohort to 293 unexposed controls (demographics in Table 1B) from the New York City area.

• Download figure
• Open in new tab

Figure 1. Study design and pipeline.

A) Study outline - To understand L/M-CHIP prevalence in WTC debris exposed first responders, we collected blood from participants, extracted DNA, and performed WES on 350 samples. Next, we aligned the exomes to the human genome and called somatic variants using Mutect2 to identify L/-M-CHIP carriers, the specific CHIP mutations and VAF. Additionally, we compared the prevalence of L/-M-CHIP in WTC debris exposed first responders to healthy controls from the New York City area. Finally, we associated CHIP prevalence with respect to age, ancestry, exposure, cognitive/metal data, HLA zygosity, blood counts and other clinical variables. B) WES data analysis pipeline to identify M/L-CHIP mutations.

Prevalence and characteristics of L/M-CHIP mutations

After sample QC, 16.2% (56 participants) of the 345 WTC participants harbored 71 M-CHIP mutations in13 genes, while 21.4% (74 participants) harbored 85 L-CHIP mutations in 43 genes (Supplementary Table S2). Overall, 34.2% (118/345) of the WTC participants harbored at least one M- or L-CHIP mutation. We observed greater clonal complexity, identified by the presence of more than one mutation in 7.0% (24/345) of WTC participants, where 3.5% (12 participants) carried both mutation types.

The majority (>80%) of the participants who harbored M/L-CHIP mutations carried only a single mutation (Figure 2-A, D). Of the M-CHIP mutations, 39% were non-synonymous, 30% were stop-gain, 23% were frameshift deletions and the rest were frameshift insertions and splicing (Figure 2B). For L-CHIP, 87% were non-synonymous, 9% were stop-gain and the rest were frameshift indels (Figure 2E). The top M-CHIP genes were DNMT3A, TET2, PPM1D and ASXL1 (Figure 2C). The top L-CHIP genes were EEF1A1, DDX11, KMT2D, ATM and FAT2 (Figure 2F). Figure 3A shows the VAF distribution of the M/L-CHIP mutations. The highest VAF was in TET2 mutations (Figure 3C).

• Download figure
• Open in new tab

Figure 2. Characteristics of M-CHIP and L-CHIP mutations in 345 WTC-debris-exposed first-responders.

A) Number of participants with 1,2, 3 and 4 M-CHIP mutations, B) Number of different types of M-CHIP mutations, C) Number of M-CHIP mutations observed in each gene, D) Number of participants with 1,2, 3 and 4 L-CHIP mutations, E) Number of different types of M-CHIP mutations, and F) Number of L-CHIP mutations observed in top genes (#Mutations > 2).

• Download figure
• Open in new tab

Figure 3. Variant allele fraction (VAF) of M-CHIP and L-CHIP mutations.

A) Histogram of the VAF of the M-CHIP mutations, B) Histogram of the VAF of the M-CHIP mutations, C) Variant allele fraction (VAF) distribution of mutations in the top M-CHIP genes D) VAF distribution of mutations in the top L-CHIP genes. Based on VAF, TET2 has the largest M-CHIP clones, while EEF1A1 and DDX11 have the largest L-CHIP clones.

Additional factors on CHIP prevalence

We studied the associations between M- and/or L-CHIP mutations and i) clinical data - age, gender, race, smoking status, cardiovascular diseases (CVD), stroke, body mass index (BMI), Montreal Cognitive Assessment (MoCA), PTSD Checklist (PCL); ii) laboratory data (blood and lipid counts) and iii) HLA zygosity (class I and class II alleles) (Supplementary Table S3, Figure 4). Further, we studied the associations driven by the top mutated M-CHIP (DNMT3A, TET2, PPM1D) and L-CHIP (EEF1A1, DDX11) genes. Figure 4 summarizes the factors with at least one significant association in the univariate analysis. We did not observe associations in L-CHIP positive WTC cases with the factors considered in this study. In Supplementary Table S3 we list all associations with the participant characteristics we considered.

• Download figure
• Open in new tab

Figure 4: CHIP-Phenotype associations.

Heatmaps showing significant associations between overall CHIP, M-CHIP, M-CHIP driven by DNMT3A, TET2, PPM1D mutations and L-CHIP driven by EEF1A1, DDX11 mutations with clinical variables, blood parameters and HLA zygosity. Blue and red correspond to significant (p ≤ 0.05) negative (OR < 1) and positive (OR > 1) correlation. White represents absence of significant correlation. Asterisk represents significant associations (p ≤ 0.05) in the multivariate regression analysis.

Clinical data

Several studies (Bick et al. 2020b; Kar et al. 2022) have observed that older age, smoking and other clinical characteristics are associated with CHIP. We report the distribution of M/L-CHIP in the WTC cohort with respect to age in Table 2. Consistent with literature on CHIP and aging (Bick et al. 2020b; Niroula et al. 2021), participants with CHIP mutations, specifically M-CHIP mutations and the top M-CHIP genes, DNMT3A and TET2, were associated with older age compared to CHIP negative (individuals without M/L-CHIP mutations) cases. Age remained a significant factor in the multivariate logistic regression model too. The median age of participants with CHIP, M-CHIP, DNMT3A, TET2 were 61, 62, 66.5 and 64 years, respectively. Whereas the median age of CHIP negative participants was 59 years. Former smokers were 44.6% of the WTC cohort, and a smoking history was associated with M-CHIP in both uni- and multivariate regression models. M-CHIP mutations and particularly DNMT3A were also associated with lower BMI compared to CHIP negative participants. Finally with regards to the mental health and cognitive characteristics, participants with DDX11 mutations (a top L-CHIP gene), were associated with higher PCL scores and lower MoCA scores. MoCA scores remained significant in the multivariate logistic regression model. The median MoCA score for participants with DDX11 mutation was 22, which was within the range indicative of mild cognitive impairment.

View this table:

• View inline
• View popup
• Download powerpoint

Table 2. Prevalence of M-CHIP and L-CHIP mutations

Laboratory data

Previous studies have revealed associations of M-CHIP with myeloid cell parameters (e.g. platelet, red blood cell, neutrophil and monocyte counts), and L-CHIP with elevated lymphocyte counts (Niroula et al. 2021). In the WTC cohort, M-CHIP positive cases had lower platelet counts (uni- and multivariate regression models) as compared to CHIP negative cases. Participants with the most frequently observed M-CHIP mutations, DNMT3A and TET2, were associated with lower absolute lymphocyte counts. TET2 mutation carriers were additionally associated with lower RBC counts, and higher segmented neutrophils. At the same time, PPM1D mutation carriers were associated with lower platelet counts and higher mean corpuscular hemoglobin (MCH). Further, DDX11 mutation carriers were associated with higher absolute lymphocytes, mean corpuscular volume (MCV), lymphocyte monocyte ratio (LMR), and lower segmented neutrophils.

HLA zygosity

We tested potential major histocompatibility complex (MHC) determinants that might be associated with CHIP positivity, and observed an association with a higher proportion of HLA-DMB homozygosity, which remained significant in the multivariate logistic regression model. At the same time, PPM1D mutation status was associated with a higher proportion of DQA1 homozygosity. Finally, participants with the top L-CHIP mutation gene, EEF1A1, were associated with a lower proportion of DPA1 homozygosity.

M/L-CHIP prevalence was higher in WTC cohort versus unexposed controls

To understand the impact of WTC debris-exposure on CHIP prevalence, we compared the WTC cohort to 293 healthy unexposed controls from New York area (Nathan et al. 2023). After downsampling, in the WTC cohort the prevalence of detectable M-CHIP mutations decreased to 7.5% (26/345) and L-CHIP to 9.9% (34/345). Yet, for unexposed controls, M-CHIP prevalence was 3.1% (9/293) and L-CHIP 2.0% (6/293). Grouping the participants into age strata, both CHIP mutation types were still generally more prevalent in the WTC cohort (Figure 5A and B, Supplementary Table S4). L-CHIP mutations were statistically significantly more prevalent in WTC participants ≤ 55 years old in comparison to unexposed controls (p=0.04).

• Download figure
• Open in new tab

Figure 5. Prevalence and characteristics of M-CHIP and L-CHIP mutations in 345 WTC- debris-exposed individuals (blue) and 293 unexposed controls (red).

A) % Prevalence of M-CHIP mutations in the cohorts as a function of age B) % Prevalence of L-CHIP mutations in the cohorts as a function of age C) Genes identified with M-CHIP mutations D) Genes identified with L-CHIP mutations

Discussion

CHIP is an early marker of risk for numerous maladies, including cardiovascular diseases (CVD, both ischemic heart disease and stroke), hematologic malignancies, severe COVID-19 outcomes, several solid cancers, infection, and all-cause mortality (Bhattacharya et al. 2022; Bick et al. 2020a; Bolton et al. 2021; Jaiswal et al. 2014, 2017; Jaiswal and Ebert 2019). There is also increasing appreciation that some CHIP mutations are tied to inflammatory consequences such as anemia (Girelli and Busti 2019). While CHIP presence is not a prerequisite for progression to malignancy, it significantly increases risk. The transformation rate from M-CHIP to blood cancer is estimated around 0.5-1.0% per year, roughly 13 times greater than the incidence in the general population (Genovese et al. 2014; Heuser et al. 2016). L-CHIP was also associated with higher incidence of lymphoid malignancy (Hazard Ratio ∼4) (Niroula et al. 2021). Here, we report M and/or L-CHIP mutation prevalence and patterns in a population of WTC responders. This is arguably the first study to document L-CHIP mutations within a WTC cohort and within a cohort of occupationally exposed individuals in general. In addition, we uniquely leveraged a rich dataset of clinical, laboratory, mental, cognitive and HLA zygosity information to study their associations with CHIP.

Despite the severe toxic inhalation exposures of WTC responders, to date, only one study has assessed their CHIP prevalence. Focused on FDNY first responders, this study used targeted sequencing on 237 genes known to be recurrently mutated in hematologic malignancies, and reported elevated CHIP in comparison to unexposed controls (Jasra et al. 2022). However, increased cancer mortality has also been reported for WTC responders who were not part of the FDNY (Li et al. 2023). Indeed, the combined WTC population (pooled from FDNY, WTC Health Registry and GRC) has a documented elevated risk for melanoma, as well as cancers of the prostate, thyroid and tonsil compared to the general population (Li et al. 2022), and WTC GRC cohort observed a modest elevation of hematologic malignancies (Shapiro et al. 2019). Here, our deep WEX of the WTC GRC, which includes a more diverse group of first responders than the FDNY, revealed a high prevalence of both M-CHIP (16.2%) and L-CHIP (21.4%) mutations, with an overall prevalence of 34.2%. While the elevation in M-CHIP with respect to WTC debris exposure in the WTC GRC was consistent with the previous report on targeted sequencing of samples from WTC FDNY participants, the genes, mutations and filtering criteria were different between the studies, thus preventing direct comparisons.

Currently, there are no standard definitions of CHIP, and the continual identification of specific driver mutations and their subsequent inclusion (and exclusion) in CHIP gene lists changes the definition of the condition. For this reason, the 2022 World Health Organization definition of CHIP is not gene or variant specific, only referring to somatic mutations of myeloid malignancy- associated genes detected in the blood or bone marrow at a VAF of ≥ 2% (≥4% for X-linked gene mutations in males) in individuals without a diagnosed hematologic disorder or unexplained cytopenia (Khoury et al. 2022; Steensma et al. 2015). As a result of this evolving landscape, different research studies report results on different variants and genes, without an international consensus on how CHIP should be defined (Bick et al. 2020b; Niroula et al. 2021). Furthermore, there are no established ‘gold standard’ analysis pipelines devoted to CHIP analysis. For example, while a VAF of ≥2 % is recommended, there are no guidelines on what the corresponding read depth would be, as more mutations with a true VAF of 2% will be missed at a read depth of 30X than 250X. In addition, the filters to remove technical artifacts vary between studies. Hence, both the biological and technical definitions of CHIP vary between studies, making comparisons between studies and with population level controls challenging.

Because of the differences described above, instead of directly comparing CHIP prevalence in our study with population level cohorts (ie. UK BioBank and TOPMED), we analyzed an unexposed control cohort using the same analytical pipeline (Figure 1B). Furthermore, we downsampled the WTC GRC findings to match sequencing depth of the control cohort. We observed that both M- and L-CHIP were generally more prevalent in WTC responders than unexposed controls. This suggests that WTC exposure or other characteristics of WTC responders, such as chronic stress following the WTC experience (Rogers et al. 2022) is associated with elevated M/L-CHIP, especially for those 55 and younger.

Our findings on age-associated increase in M-CHIP prevalence in the WTC cohort were consistent with its well-documented association with age in other population-level cohorts (Bick et al. 2020b; Niroula et al. 2021). These studies have documented that lifestyle factors like smoking history, BMI were also positively associated with M-CHIP prevalence (Komic et al. 2023). While we observed a similar association for smokers in the WTC cohort, a larger cohort size is needed to understand the association of BMI with M-CHIP in this population.

Genes with mutations most commonly associated with M-CHIP in the literature are involved in epigenetic regulation (TET2, ASXL1, DNMT3A, and IDH2), RNA splicing (SF3B1, SRSF2, U2AF1), cell signaling (JAK2 and NRAS), and DNA repair and cell cycle regulation (TP53) (Heuser et al. 2016; Sperling et al. 2017; Steensma et al. 2015). Similar to these trends, we observed M-CHIP mutations in most of these genes (see Figure 2C). For L-CHIP mutations, prior studies have reported an even distribution across a larger number of genes (Niroula et al. 2021). While we did observe some of the previously reported L-CHIP gene mutations, 21% of L- CHIP mutations were in EEF1A1 and 15% in DDX11 (see Figure 2F). This signal was driven by three putative mutations, DDX11:P368S, EEF1A1:E293K and EE1F1A:V315L, which have known associations with hematologic malignancies (Ma et al. 2022; Papaemmanuil et al. 2013, 2016; Tyner et al. 2018). L-CHIP mutations in EEF1A1 and DDX11 remained significantly more prevalent in the WTC cohort in comparison to unexposed controls as well. Note that, EEF1A1 and DDX11 were not among the genes considered in the previous targeted sequencing WTC- FDNY study (Jasra et al. 2022).

As the WTC responder population ages, multiple studies are reporting an increased risk for neurocognitive and motor dysfunction that resembles neurodegenerative diseases, in addition to the presence of cortical atrophy and cognitive impairment at midlife within this population. These risks have been associated not only with physical exposures at the WTC site, but also with chronic post-traumatic stress disorder (PTSD) (Clouston et al. 2022). We therefore investigated the association of the available mental and cognitive data on the WTC cohort with CHIP, and observed an association for L-CHIP DDX11 gene with higher PCL scores (indicative of PTSD severity) and lower MoCA scores (indicative of mild cognitive impairment). DDX11 is a helicase that participates in cellular processes that alter RNA secondary structure, and helicase dysfunction has been implicated in multiple syndromes (Uchiumi et al. 2015), including neurodegeneration (Lovell et al. 2000).

As CHIP can associate with inflammatory diseases, we leveraged our WEX data to further perform Human Leukocyte Antigen (HLA) typing (both classes I and II) and test the diversity in the HLA genes in association with M/L-CHIP, in the context of the heterozygote advantage hypothesis. This hypothesis posits that individuals who harbor heterozygous genotypes at the HLA genes are able to display a greater variety of antigenic peptides than those with homozygous genotypes at the HLA genes, leading to an immune response to a broader range of antigens (Doherty and Zinkernagel 1975; Pagliuca et al. 2022; Penn et al. 2002). Testing for homozygosity in Class I and Class II HLA genes with CHIP, overall CHIP prevalence was statistically significantly associated with higher proportion of HLA-DMB homozygosity. We also observed a weaker signal for PPM1D mutations with higher proportion of HLA-DQA1 homozygosity; and EEF1A1 mutations with a lower proportion of HLA-DPA1 homozygosity. These are HLA Class II genes expressed in antigen presenting cells, including lymphocytes, and future studies are needed to understand the interplay between homozygosity in these genes, immune deficiencies and cancer (Ogobuiro et al. 2023; Pagliuca et al. 2022; Planelles et al. 2006).

In exploratory analyses, we compared the association of blood counts data in WTC responders with the UK BioBank population. Specifically, we investigated M/L-CHIP with myeloid parameters (e.g. platelet, red blood cell, neutrophil and monocyte counts), and lymphocyte counts (Niroula et al. 2021). In the WTC cohort, consistent with the population-level UK Biobank cohort, TET2 mutations were negatively associated with lymphocyte counts (Kar et al. 2022) and PPM1D mutations were negatively associated with platelet counts (Kamphuis et al. 2023). However, most associations we observed between M-CHIP mutations and blood count were not statistically significant in a multivariate regression analysis, when age and other factors were taken into account. We also observed opposite trends from the UK Biobank cohort in the association of M-CHIP with platelet counts, and TET2 mutations with segmented neutrophil counts (though UK Biobank uses neutrophil counts). To understand whether this discrepancy is due to WTC-debris exposure, further investigation is warranted.

Recent studies on the UK BioBank report that in individuals without detectable CHIP at the time of lab measurements, abnormal complete blood count (CBC) labs can still be predictive of risk for future CHIP (Gu et al. 2023; Weeks et al. 2023), and that those that harbor abnormal myeloid blood cell parameters and CHIP are at the highest risk for developing myeloid malignancies. Similarly, those with elevated lymphocyte counts were reported to be at elevated risk for lymphoid malignancies, and those additionally with L-CHIP mutations at highest risk.

These reports suggest that the WTC participants with elevated myeloid and/or lymphoid blood counts, together with CHIP mutations, may be at highest risk for future malignancies. Currently, there is interest in treating CHIP patients with activated inflammatory pathways with IL-1β inhibitor and NLRP3 inhibitors in addition to IRAK1 inhibitors (Kanagal-Shamanna et al. 2024; Wang et al. 2021). Thus, it may be appropriate to investigate such interventions in WTC responders at the highest risk to prevent or delay overt myeloid neoplasm, CVD, major bleeding, and infection.

Overall, our study focused on the impact of exposure from 9/11 attacks on WTC first responders, and yet has implications beyond this population. While the WTC exposure included a complex mix of carcinogenic compounds specific to the site, exposure itself was a main determinant of risk for CHIP. This warrants future risk studies on populations exposed to debris (both from building fires and collapses) from modern warfare in populated cities.

Study findings will inform future research for risk stratification for those at highest risk, who may need to undergo more frequent screening and monitoring to prevent or delay cancer development. Despite a higher relative risk of malignancies in CHIP positive individuals, the absolute incidence remains low in the CHIP positive (Heuser et al. 2016). Yet, to identify individuals at highest risk for cancers, genetic information on CHIP can additionally be combined with demographic, clinical and laboratory data to build risk models. Efforts towards this end are already starting to gain traction (Weeks et al. 2023). We anticipate that with the establishment of CHIP clinics at many academic centers, longitudinal monitoring of CHIP individuals will inform improved risk models. Those at highest risk can then be followed, have risk assessment and eventually participate in clinical trials in CHIP clinics.

This study should be considered in the context of its limitations. Here, we compared the prevalence of CHIP in WTC rescue and recovery workers to unexposed controls. While the control participants were from the New York Area, they were not first responders. Previous studies on CHIP have additionally considered mosaic chromosomal alterations (mCAs) in the peripheral blood of individuals. We were unable investigate the prevalence of mCAs, as WES typically does not include enough heterozygous sites to allow their detection at low cell fractions (Loh et al. 2020). The study size is another weakness; the limited number of study participants makes it harder to detect modest effects. Thus, we cannot distinguish between no evidence for association due to limited power and an actual lack of association (e.g. between age and L- CHIP). This is especially important in understanding the differences in the association trends for M-CHIP positive cases with clinical and lab data in the WTC cohort versus the UK BioBank. Finally, the multiplicity of analyses we conducted may lead to results which are statistically significant at the 5% level, yet due to chance (false positives): in this respect, our results need confirmation in independent populations.

Code Availability

All analyses were performed utilizing standard publicly available software. Any specific analysis code details are available from the authors upon request.

Authors’ contributions

P.B and Z.H.G. conceived and designed the study. M.E.S., P.K., R.J.K., and Z.H.G. wrote the manuscript. B.J.L. and X.Y. recruited participants and handled sample and data collection. Z.H.G led sample sequencing at Azenta Inc. M.E.S. performed sequence analyses and P.K. performed statistical analyses. All authors were involved in the interpretation of the results. B.J.L., J.M. and P.B. edited the manuscript. Z.H.G. supervised the study. All authors approved the final manuscript.

Supplementary Tables

Supplementary Table S1. List of myeloid associated (M-CHIP) mutations and lymphoid associated (L-CHIP) mutations considered in this study.

Supplementary Table S2. List of M/L-CHIP mutations identified among the 345 WTC samples.

Supplementary Table S3. CHIP-Phenotype associations.