Association of infection frequency and incident clonal hematopoiesis

Abstract Background: Clonal hematopoiesis (CH) is an age-related phenomenon in which hematopoietic stem cells acquire somatic mutations that confer relative fitness advantage, resulting in clonal expansion. CH is a malignancy precursor state: people with CH have a 4-fold increased relative risk of hematologic malignancy. Moreover, individuals with CH carry a 2.5-fold increased relative risk of all-cause mortality in addition to risk for cardiovascular disease and other diseases of aging. Clinical epidemiology studies have defined the genetic landscape of CH; collectively, these studies reveal that DNMT3A and TET2 mutations are the most common. CH is not strictly a product of genetics as many studies demonstrate clonal expansion secondary to inflammatory stimuli such as infection. We have previously shown in a competitive murine transplant model that chronic mycobacterial infection is a driver of DNMT3A CH. Others have shown that microbial gut translocation drives TET2 clonal expansion. Clinical studies corroborate that infection is a powerful driver of infection: people living with HIV have an increased prevalence of CH when compared to individuals without HIV. Yet, the relationship between infection frequency and risk of subsequent CH remains unknown. Objective: To determine the association between infection frequency and incident CH. Study Design/Methods: This closed prospective cohort study leveraged the ongoing Atherosclerosis Risk in Communities (ARIC) study. Hospital documented infections were identified using ICD codes recorded in hospital discharge summaries. Infections were typically distinguished by site (e.g. pneumonia, urinary tract infection) whereas pathogen identification (e.g. Staphylococcus, Salmonella) was not available. Self-reported infections were captured in standardized interviews. Whole exome sequencing was done at two time points (separated by ~20 years) from which CH status and variant allele frequency (none: 2%, small: 2-10%, large: 10%) were determined. ARIC participants who underwent sequencing at both time points were eligible for inclusion. ARIC participants with CH at the first sequencing time point were excluded from this analysis. Additionally, ARIC participants with malignancy at any time point were excluded from this study. We stratified our cohort by number of infectious challenges and used logistic regression models to estimate the odds ratio (OR) of incident CH. We classified incident CH as any CH, DNMT3A CH, TET2 CH, or non-DNMT3A CH (including TET2 and other less common mutations in this cohort such as ASXL1 and TP53). All analyses were adjusted for age, sex, race, socioeconomic status, substance (tobacco and alcohol) use, and co-morbidities (cardiovascular disease and diabetes). Results: 3,367 of 15,792 ARIC participants met inclusion criteria for this analysis. 59.1% were female, 40.9% were male; 24% were Black, and 76% were White. Mean age was 55.3 years at the first sequencing time point. Of the participants, 19.7% had incident CH, 6.9% had large CH, and 5.2% had large non-DNMT3A CH. We observed a stepwise positive association between frequency of hospital documented infections (0, 1, 2, or ≥ 3+) for TET2 and other non-DNMT3A CH across clone size. Participants with ≥ 3 hospital documented infections had increased odds of incident CH (OR 1.41, p = 0.03), especially large CH (OR 1.83, P = 0.008) and large non-DNMT3A CH (OR 1.81; p = 0.02). Most study participants reported at least one cold over the study period; we observed a stepwise positive trend between increased frequency of colds and odds of incident CH across clone size. Those with ≥ 11 self-reported colds over the study period were more likely to develop CH (OR 1.24, p = 0.1) as compared to those with 11 colds. We observed no association between any specific infection type and incident CH in our analyses of hospital documented and self-reported infections. Conclusions: This study is the first to demonstrate a positive association between infection frequency and incident CH in a general population, highlighting a modifiable risk factor for CH. Individuals with more frequent infections may have increased CH risk, especially large and non-DNMT3A CH. The finding that CH may result from more frequent infectious challenges could inform evolving guidelines for CH surveillance. Further work is required to describe the mutation-specific impact underlying this observed relationship.