Mayo Clinic Institutional Review Board approved the current study. Written consent was obtained from all participants. The study was conducted in accordance with the guidelines of the Declaration of Helsinki.
Study population
In this study, 188 patients who underwent invasive coronary endothelial function testing at Mayo Clinic, Rochester, Minnesota, from 1993 to 2015 were enrolled. Patients presenting with chest pain without history of cardiovascular intervention, myocardial infarction, heart failure, known structural cardiac diseases or evidence of obstructive coronary artery disease on cardiac angiography were selected for physiological assessment of microcirculatory endothelial function by assessment of the change in diameter and blood flow of the coronary artery in response to administration of intracoronary graded infusion of acetylcholine (13, 14). Based on the results, patients were categorized into two groups; patients with (n=123) and without (n=65) endothelial dysfunction.
Invasive coronary functional testing
The study protocol has been described in detail elsewhere (15, 16). In brief, patients presenting to the catheterization laboratory with non-obstructive coronary artery (< 40% stenosis) underwent invasive coronary functional testing using an intracoronary Doppler guidewire. A Doppler guidewire (0.014-inch FloWire, Philips/Volcano Inc) was advanced 2-3 mm distal to the tip of 2.2 F coronary-infusion catheter (Ultrafuse, SciMed Life System) positioned into the mid-portion of the left anterior descending artery (LAD). In all patients, acetylcholine was selectively infused into the LAD at concentrations of 10−6, 10−5, and 10−4 mol/L over 3 minutes at each concentration. Doppler measurements and coronary angiography were obtained after each infusion. CAD was measured in segment 5 mm distal to the tip of the Doppler wire. CBF was calculated from the Doppler-derived time velocity integral and vessel diameter, as previously described (17, 18). Epicardial endothelial dysfunction was defined as a decrease in CAD of > 20% in response to acetylcholine as compared to baseline; microvascular endothelial dysfunction was defined as a maximal percentage increase in CBF in response to acetylcholine < 50% as compared to baseline (13, 16). We defined coronary endothelial dysfunction as either epicardial or microvascular endothelial dysfunction or both.
Clinical and biochemical data
For all participants prior to undergoing diagnostic tests, demographic and clinical characteristics were obtained through history and physical examination, as previously described (19, 20). Data regarding age, sex, body mass index (BMI), smoking status (never/previous/current), hypertension, diabetes and hyperlipidemia were collected. Diabetes was defined as a positive history of diabetes and/or consumption of antidiabetic medications; hypertension was defined as positive history of hypertension and/or consumption of antihypertensive medications; hyperlipidemia was defined as positive history of serum lipid profile out of the normal range and/or usage of lipid lowering medications. Venous blood samples for routine biochemical tests (complete blood count, glucose, creatinine, lipid profile) were obtained after overnight fasting before the procedure. Buffy coat and plasma samples extracted from whole blood were used for collection of DNA and measurement of cytokine expression levels, respectively. Using echocardiography, the percentage of left ventricular ejection fraction was determined and then compared between two study groups. A standardized questionnaire was administered to the patients to record occurrence of MACE and hematologic malignancies in average follow-up of 12.2 ± 4.3 years, as was previously described (19, 21). Review of medical records was performed blindly by an independent investigator.
Detection of clonal hematopoiesis by targeted capture assays
DNA was extracted from buffy coat samples following the procedures in Qiagen’s Puregene kit. We sequenced the entire coding regions of 35 genes using a customized 150Kb Agilent SureSelect panel. Samples were paired-end sequenced (150 bp reads), using Illumina HiSeq 4000 sequencer with 96 samples per lane of flow cell. The median coverage depth per sample across the 35 genes was >1000X per nucleotide, allowing the detection of mutations with variant allelic fraction (VAF) as low as 1%. Raw variants were annotated using GATK Variant Annotator for variant quality, and Biological Reference Repository (BioR) was used for variant annotation (22). Variants with a Mapping Quality <20, read depth <10X, or found in <1% of reads were removed. Additionally, sequencing artifacts found in homopolymers were excluded. Finally, variants of significant interest were visually inspected using Integrative Genomics Viewer (IGV) (23).
Primary bioinformatics analysis
The paired reads of targeted sequencing are mapped to human genome reference (Hg38) using BWA (http://bio-bwa.sourceforge.net/bwa.shtml), duplicate reads are marked using picard (https://broadinstitute.github.io/picard/), mutations are called using GATK (https://gatk.broadinstitute.org/hc/en-us) by three steps: base recalibration, haplotype caller, and Variant Quality Score Recalibration (VQSR). Finally, mutations are annotated with databases Clinvar (http://www.clinvar.com/), dbSNP (https://www.ncbi.nlm.nih.gov/pmc/articles/PMC29783/) and 1000 genome (https://www.internationalgenome.org/).
Definitions related to Clonal Hematopoiesis Mutations
Variants were classified as CHIP related if they had a variant allele frequency of >2%, and exhibited a minor allele frequency of ≤0.1% in the Exome Aggregation Consortium Project (6), a database of known non-somatic variants (ExAC, http://exac.broadinstitute.org/), and were present in the Catalogue of Somatic Mutations in Cancer (COSMIC, https://cancer.sanger.ac.uk/cosmic) database. Additionally, if CHIP mutations were cited in COSMIC as somatic and identified in a hematologic malignancy (known to be pathogenic), we labeled them CH-PD (clonal hematopoiesis with a putative driver). Variants below minor allele threshold, and not present in COSMIC, were defined as a variant of uncertain significance (VUS).
Cytokine level analysis
Plasma samples from 178 patients were tested for the levels of IL-6 and IL-8 cytokines, by the Cytokine Human Magnetic Kit (cat: LHC0001M, ThermoFisher Scientific) (24). Ten patients were excluded due to missing plasma samples. The bead mix was prepared according to the manufacturer’s protocol and samples analyzed in a multiplex array using ProcartaPlex magnetic beads via a Luminex® 200™ instrument (Austin, TX). All plasma samples were run in duplicate. The data were analyzed using the instrument specific software, xPONENT®. Statistical comparisons were tabulated using the non-parametric Welch’s t test via Prism Software.
Statistical analysis
In current study, continuous variable are expressed as mean ± standard deviation (SD). For comparison of quantitative variables between two groups, student t-test was performed for normally distributed variables and Mann-Whitney U test (or Kruskal-Wallis test) for abnormal distributed data. Categorical variable are expressed as frequencies (%) and analyzed between different groups by the Chi-square test. Univariate and multivariate logistic regression analyses were performed to evaluate the independent association between CHIP and increased risk of MACE as well as association of existence of ASXL1, DNMT3A and TET2 mutations with MACE. For evaluation of association between levels of IL-6 and IL-8 and mutations in ASXL1, DNMT3A and TET2, all study participants were categorized into three groups: individuals with normal endothelial function, individuals with endothelial dysfunction, but without mutations in ASXL1, DNMT3A and TET2 genes and individuals with endothelial dysfunction and mutations in ASXL1, DNMT3A and TET2 genes. These genes were selected from CHIP and VUS groups. Results of regression model were reported as odd ratios (ORs) and 95% confidence interval (95% CIs). A P -value < 0.05 was assumed for statistical significance. All analyses were perform using SPSS software version 25.0 (SPSS Inc., Chicago, IL, USA), GraphPad Prism 6.07 (GraphPad Software, La Jolla, California, USA), and JMP 9 software (SAS Institute, Inc., Cary, NC).