Exposure to Volatile Organic Compounds is associated with Hypertension in Black Adults: The Jackson Heart Study

Abstract

Background

The prevalence of hypertension is higher in Black adults than among White and Hispanic adults. Nevertheless, reasons underlying the higher rates of hypertension in the Black population remain unclear but may relate to exposure to environmental chemicals such as volatile organic compounds (VOCs).

Methods

We evaluated the associations of blood pressure (BP) and hypertension with VOC exposure in non-smokers and smokers in a subgroup of the Jackson Heart Study (JHS), consisting of 778 never smokers and 416 age- and sex-matched current smokers. We measured urinary metabolites of 17 VOCs by mass spectrometry.

Results

After adjusting for covariates, we found that in non-smokers, metabolites of acrolein and crotonaldehyde were associated with a 1.6 mm Hg (95%CI: 0.4, 2.7; p = 0.007) and a 0.8 mm Hg (95%CI: 0.01, 1.6; p = 0.049) higher systolic BP, and the styrene metabolite was associated with a 0.4 mm Hg (95%CI: 0.09, 0.8, p = 0.02) higher diastolic BP. Current smokers had 2.8 mm Hg (95% CI 0.5, 5.1) higher systolic BP. They were at higher risk of hypertension (relative risk = 1.2; 95% CI, 1.1, 1.4), and had higher urinary levels of several VOC metabolites. Individuals who smoke had higher levels of the urinary metabolites of acrolein, 1,3-butadiene, and crotonaldehyde and were associated with higher systolic BP. The associations were stronger in participants who were <60 years of age and male. Using Bayesian kernel machine regression to assess the effects of multiple VOC exposures, we found that the relationship between VOCs and hypertension in non-smokers was driven primarily by acrolein and styrene in non-smokers, and crotonaldehyde in smokers.

Conclusions

Hypertension in Black individuals may be attributed, in part, to VOC exposure from the environment or tobacco smoke.

The Jackson Heart Study of Cardiovascular Disease Among African Americans (JHS) https://clinicaltrials.gov/ct2/show/NCT00005485 NCT00005485

GRAPHICAL ABSTRACT

INTRODUCTION

The Black population carries a disproportionate burden of hypertension,¹ and the rate of hypertension-related CVD among Blacks in the US is higher than that in any other race/ethnic group in the world. Hypertension accounts for 50% of the racial difference in mortality between Black and White populations² and it is directly related to disparities in stroke, heart failure, and peripheral artery disease among Black populations.³ While no specific genetic susceptibilities have been identified, interactions between several genetic, biological, and social factors are thought to contribute to the high vulnerability of Black individuals to hypertension.^{2, 3} Among environmental vulnerabilities, exposure to pollutants may represent an additional risk for increased cardiovascular disease (CVD) risk in minority and vulnerable populations. Extensive work has shown that nearly all major emission source sectors disproportionately affect people of color⁴ and that such socioeconomic and ethnic disparities have persisted despite overall decrease in particulate air pollution in the US.⁵

Although components of polluted air such as particulate matter, ozone, and nitrogen oxides have been extensively studied, other common constituents of ambient and indoor air such as volatile organic compounds (VOCs) have received less attention. VOCs are ubiquitous components of urban air, present in automobile exhausts, industrial emissions, as well as emissions from household products such as cleaners, disinfectants, and paints.⁶ In humans, some of the highest levels of exposure occur due to smoking because aerosols of both combustible^{7, 8} and electronic^{8, 9} cigarettes contain high levels of VOCs. Regardless of specific sources, exposure to VOCs such as acrolein¹⁰ has been associated with increased thrombosis and CVD risk in individuals with moderate to high CVD risk. In animal models, exposure to acrolein induces systemic dyslipidemia¹¹ and platelet activation¹² and exacerbates atherosclerosis.¹³ We have recently reported that in humans, exposure to VOCs such as ethylbenzene, styrene, xylene,¹⁴ and benzene¹⁵ is negatively associated with the circulating levels of angiogenic cells in the peripheral blood, and that exposures to acrolein, 1,3-butadiene, or crotonaldehyde are associated with endothelial dysfunction or elevated risk of hypertension.¹⁶ Nevertheless, the extent of VOC exposure and its contribution to hypertension in the general Black population has not been assessed. Therefore, the current study was designed to estimate VOC exposure in the well characterized Jackson Heart Study (JHS) cohort of Black participants in order to identify the extent to which VOC exposure affects BP and hypertension in those who do or do not smoke.

METHODS

Study Population

The JHS is a community-based, observational study of CVD in 35–84 years of age Black individuals from the Jackson area of Mississippi.¹⁷ Participants who met enrollment criteria gave written informed consent. The Institutional Review Board at the University of Mississippi Medical Center approved the study. To create a nested, matched 2:1 subsample, we selected 796 never smokers and 398 current smokers from the baseline exam (visit 1, 2000–2004; n=5,306). Participants were matched based on age (± 5 years), and sex. Clinical measures, spot urine samples, and questionnaires were collected at Visit 1. Those with missing urine measurements or uncertain tobacco exposure status were excluded from the study.

Smoking Variables

Self-reported smoking status was quantified by tobacco questionnaire item 3 (TOBA3), ‘Do you now smoke cigarettes?’, or non-smokers, who responded no to TOBA1 ‘Have you smoked at least 400 cigarettes in your lifetime?’ To capture tobacco users in addition to cigarette users, and minimize cigarette use misclassification, we confirmed smoking status based on creatinine-corrected levels of urinary cotinine of >40ng/mg.¹⁸ The specificity of the urinary cotinine assay is 98%, whereas self-reported smoking status has a specificity of 85%.¹⁹ Reclassification of self-reported smoking status by urinary cotinine levels increased the number of smokers from 398 to 416 and decreased the number of non-smokers to 778 (Supplemental Table S1). To estimate cigarette per day (CPD), we used TOBA5, ‘How many cigarettes do you smoke per day?’ To quantify pack-years, or cumulative smoking, we used the product of TOBA5/20 cigarettes in a pack and the age at when the participant first smoked cigarettes (TOBA2).

Urinary VOC Metabolites

To estimate VOC exposure, we measured urinary levels of 24 metabolites generated from 17 parent VOCs using UPLC-MS as described before.^{7, 20} In brief, we spiked diluted urine with isotopically labeled internal standards, and calculated the concentration of each analyte using a 9-point calibration curve with an R²>0.99. Metabolites below the limit of detection (LOD) were imputed with the LOD/√2. Eight metabolites were below LOD in >50% of the samples and were removed from further analysis. All urinary VOC metabolites (ng/mL) were adjusted for individual creatinine (mg/dL) and reported as nanogram of VOC metabolite per milligrams of creatinine (ng/mg creatinine). See Supplemental Table S2 for VOC parent compounds, short and long names, and frequency of samples measured above the LOD.

BP and Hypertension Variables

During the clinic examination at Visit 1, sitting systolic and diastolic BP (SBP, DBP) was measured twice using a Hawksley random zero sphygmomanometer. Measurements were taken at least 1 minute apart, averaged, and then calibrated to a semiautomatic oscillometric device as described.²¹ Hypertension status was defined as SBP ≥140 mm Hg or DBP ≥90 mm Hg, or self-reported use of BP lowering medication.¹⁷

Covariates

Age, sex, alcohol use, self-reported history of stroke, coronary heart disease (CHD), CVD, anthropometric data such as body mass index (BMI, kg/m²), psychosocial data such as individual education, and health insurance coverage were collected during Visit 1. Venous blood samples were collected after >12 h of fasting. Low- and high-density lipoprotein (LDL, HDL), triglycerides, total cholesterol, fasting plasma glucose (FPG), and acetylated hemoglobin (HbA1c), were assessed using standard laboratory techniques. Diabetes mellitus (DM) was defined as FPG ≥126 or HbA1c ≥ 6.5 or self-reported DM medication use. Estimated glomerular filtration rate (eGFR) was calculated by using the modification of diet in renal disease study equation.²² Because fine particulate matter (diameter ≤2.5 μg/m³, PM_2.5) is an established CVD risk factor,²³ we estimated PM_2.5 exposure on the day of enrollment for adjustment in our models. To estimate daily mean values of PM_2.5, we averaged daily measurements from all EPA monitors in the Jackson area and matched them to the participant enrollment date.

Statistical Analysis

To characterize the study population, we compared continuous and categorical variables in non-smokers and smokers using t-tests and χ² tests. If continuous variables were not normally distributed, we used the non-parametric method, the Kruskal-Wallis test, for two-group comparisons and reported median [25^th, 75^th percentile] instead of mean (standard deviation, SD). We utilized linear models to estimate associations between VOCs and smoking status with BP. To estimate associations between VOC metabolite levels and smoking we used generalized linear models with the gamma distribution and log link function. Modified Poisson regression with robust standard errors was used to test whether VOCs and smoking were associated with hypertension.²⁴ Model adjustments were chosen a priori based on literature review and directed acyclic graph for BP and hypertension (Supplemental Figure S1).^{25, 26} Adjustments included age, sex, BMI, physical activity, education, eGFR, cholesterol to HDL ratio, triglycerides, BP medications, diabetes, PM_2.5, and smoking status. We conducted a mediation analysis to assess whether the effects of smoking on BP were mediated by VOCs. We flexibly modeled the VOC metabolites as a mixture as an exploratory analysis in non-smokers and smokers using Bayesian Kernel Machine Regression (BKMR), a kernel-regression based method that characterizes the exposure response function of multiple predictors on a health outcome while other predictors are fixed to a specific percentile.^{27, 28} In two separate analyses, 1) we loaded each urinary VOC metabolite into the kernel as a single component and 2) grouped urinary VOC metabolites that had the same parent compound in hierarchical BKMR and compared posterior inclusion probabilities (PIPs) to determine the metabolites having the most effect on the outcome. Then, we selected a priori the VOC metabolites with the highest PIPs from the previous kernels, as well as the single-pollutant models, and loaded them into the kernel and ran the BKMR analysis.

We conducted several sensitivity analyses. For participants on BP medication, we imputed underlying blood pressure values and compared them to measured blood pressure by adding a 10 mm Hg or 5 mm Hg constant to SBP or DBP, respectively.²⁹ These models were not adjusted for BP medication. Additionally, we repeated our analysis of hypertension using the most recent hypertension guidelines (SBP ≥130 mm Hg or DBP ≥80 mm Hg, or self-reported use of BP lowering medication). To assess other potential confounders, we added cotinine, CPD, income, and whether individuals had any type of health insurance (private health insurance, Medicaid, Medicare, or veteran’s affairs insurance) to our models. We corrected for multiple tests by adjusting p-values to control the false discovery rate at 5%, the expected proportion of false discoveries among the rejected hypotheses.³⁰ Sensitivity analyses results can be found in the supplemental information.

Values were reported as mm Hg, Relative Risk (RR) of hypertension, or percent difference in outcome per interquartile range (IQR) of urinary VOC metabolite level (ng/mg) with 95% confidence intervals (CI) and p-values. Data acquisition, merging, data cleaning, and graphical displays were conducted using tidyverse.³¹ Statistical analysis was performed in R (version 3.1.3)³² using the process function³³ for mediation analysis, and the BKMR package.²⁸

RESULTS

Characteristics of the participants by smoking status are shown in Table 1. In comparison with smokers, non-smokers had higher BMI, consumed less alcoholic drinks per week, and reported higher levels of physical activity. Smokers were less likely to have health insurance and had less education than non-smokers. Although BP and hypertension were not significantly different between the two groups, smokers had higher levels of albumin and eGFR than non-smokers. Moreover, participants in the non-smoking group were prescribed less BP medications, had lower self-reported history of stroke, CHD, and CVD, and higher levels of triglycerides.

Table 1.

Characteristics of the nested sample from the JHS cohort, by smoking status. Urine samples were collected from 1194 female and male participants of the Jackson Heart Study consisting of 778 non-smokers (urinary cotinine ≤40ng/mg) and 416 age- and sex-matched current smokers (urinary cotinine >40ng/mg). Values are reported as mean (SD) for normally distributed, median [25^th, 75^th percentile] for nonnormally distributed continuous variables or n (%) for categorical variables.

Variable	Overall	Non-smokers	Smokers	P-Value
n	1194	778	416
Age (years)	51.4 ± 11.1	51.3 ± 11.1	51.7 ± 10.9	0.591
Sex (% Male)	595 (49.8)	381 (49.0)	214 (51.4)	0.452
BMI (kg/m2)	31.3 ± 7.1	32.1 ± 7.0	29.7 ± 7.0	<0.001
Average drinks per week	2.8 ± 7.7	1.2 ± 4.2	5.8 ±11.2	<0.001
Physical Activity				<0.001
Poor Health	572 (47.9)	311 (40.0)	261 (62.7)
Intermediate Health	395 (33.1)	288 (37.0)	107 (25.7)
Ideal Health	227 (19.0)	179 (23.0)	48 (11.5)
Education				<0.001
Less than high school	161 (13.5)	70 (9.0)	91 (21.9)
High school graduate/GED	237 (19.8)	143 (18.4)	94 (22.6)
More than high school	796 (66.7)	565 (72.6)	231 (55.5)
Income				<0.001
Poor	147 (14.7)	69 (10.7)	78 (22.2)
Lower-Middle	223 (22.4)	130 (20.1)	93 (26.5)
Upper-Middle	298 (29.9)	203 (31.4)	95 (27.1)
Affluent	329 (33.0)	244 (37.8)	85 (24.2)
Insured	994 (83.6)	667 (86.2)	327 (78.8)	0.001
Albumin to Creatinine Ratio	0.05 ± 0.3	0.04 ± 0.2	0.08 ± 0.4	0.048
eGFR (mL/min/1.73 m2)	89.4 ± 17.6	87.3 ± 17.1	93.4 ± 17.8	<0.001
SBP (mm Hg)	123.8 [115.6, 135.8]	122.9 [115.6, 133.9]	125.7 [114.9, 138.3]	0.115
DBP (mm Hg)	77.2 ± 8.7	77.6 ± 8.7	76.6 ± 8.7	0.071
Hypertension	606 (50.8)	397 (51.0)	209 (50.2)	0.842
BP Medications	505 (42.8)	352 (45.9)	153 (37.1)	0.005
DM	218 (18.3)	144 (18.5)	74 (17.8)	0.834
HbA1c (%)	47 [39, 56]	48 [39, 56]	47 [39, 56]	0.514
FPG (mg/dL)	89 [65, 125]	87 [64, 119]	95 [67, 138]	0.004
DM Medications	140 (11.8)	95 (12.3)	45 (10.9)	0.534
LDL (mg/dL)	124.0 ± 35.0	123.7± 33.6	124.7 ± 37.8	0.67
HDL (mg/dL)	47 [39, 56]	48 [39, 56]	47 [39, 56]	0.514
Triglycerides (mg/dL)	89 [65, 125]	87 [64, 119]	95 [67, 138]	0.004
Total Cholesterol (mg/dL)	194.2 ± 37.6	192.9 ± 35.7	196.8± 40.9	0.111
Stroke	41 (3.4)	18 (2.3)	23 (5.5)	0.006
CHD	89 (7.5)	37 (4.8)	52 (12.5)	<0.001
CVD	120 (10.1)	52 (6.7)	68 (16.3)	<0.001
PM2.5 (Mg/m3)	11.2 [8.9, 15.1]	11.2 [8.9, 15.2]	11.2 [9.1, 14.9]	0.887
Cotinine (ng/mg)	3.4 [1.1, 470]	1.4 [0.9, 3.2]	773 [411, 1,337]	<0.001

Urinary levels of VOC metabolites were significantly higher in participants who smoke than those that do not smoke, with the exception of BPMA, a metabolite derived from 1-bromopropane, and the toluene metabolite – BMA (Table 2). The mean value of the nicotine metabolite, ANB was much higher in smokers than non-smokers; however, due to large variability, the difference did not attain statistical significance. Nonetheless, as expected, the level of other metabolites of nicotine – ANTB, 3HC and cotinine were much higher in smokers than non-smokers.

Table 2.

Urinary VOC metabolite levels (ng/mg creatinine) overall and by smoking status. Urine samples were collected from 1194 JHS participants, consisting of 778 non-smokers (urinary cotinine ≤40ng/mg) and 416 age- and sex-matched current smokers (urinary cotinine >40ng/mg). Values are reported as median [25^th, 75^th percentile.

	Overall	Non-smokers	Smokers	p
n	1,194	778	416
NIC	8.6 [4.3, 161.2]	5.1 [3.6, 8.5]	469.2 [126.9, 1,263.4]	<0.001
COT	3.4 [1.1, 470.2]	1.4 [0.8, 3.2]	773.0 [410.6, 1,337.4]	<0.001
ANB	7.6 [4.4, 13.5]	5.9 [3.5, 9.6]	12.7 [8.1, 19.6]	<0.001
ANTB	1.4 [0.6, 3.8]	0.9 [0.5, 1.8]	5.4 [2.1, 9.6]	<0.001
3HC	23.7 [7.5, 1,754.8]	10.9 [3.7, 22.4]	3,266 [1,617, 5,628]	<0.001
CEMA	113.9 [68.2, 210.3]	85.5 [56.9, 132.5]	239.3 [138.5, 367.4]	<0.001
3HPMA	236.9 [141.3, 588.9]	172.2 [117.9, 252.9]	895.3 [393.9, 1,626.0]	<0.001
AAMA	53.7 [33.1, 98.2]	40.4 [28.1, 62.1]	104.7 [64.5, 144.6]	<0.001
CYMA	1.4 [0.6, 68.2]	0.8 [0.4, 1.4]	121.6 [60.5, 206.0]	<0.001
MU	123.0 [74.4, 209.2]	115.8 [71.1, 193.3]	141.4 [84.2, 230.6]	0.001
BPMA	9.9 [4.6, 18.5]	9.6 [4.5, 18.0]	10.1 [4.8, 19.7]	0.588
DHBMA	350.3 [250.6, 470.0]	305.3 [229.7, 404.1]	437.0 [334.3, 611.8]	<0.001
MHBMA3	7.5 [3.9, 23.3]	4.8 [3.2, 7.8]	41.2 [19.1, 67.4]	<0.001
HPMMA	152.4 [102.2, 402.2]	117.2 [92.4, 160.6]	639 [247.4, 1,124.0]	<0.001
AMCC	98.3 [55.0, 238.2]	67.8 [43.1, 103.0]	332.6 [190.7, 510.5]	<0.001
PGA	176.1 [129.5, 262.9]	148.6 [115.9, 193.9]	285.5 [209.6, 381.2]	<0.001
2HPMA	28.4 [16.7, 51.7]	20.7 [14.3, 33.9]	50.0 [31.2, 76.9]	<0.001
MA	148.5 [102.4, 232.9]	121.2 [90.4, 161.3]	251.3 [180.6, 353.1]	<0.001
BMA	6.0 [3.8, 11.4]	5.9 [3.8, 11.3]	6.2 [3.8, 11.6]	0.624
2MHA	13.5 [3.6, 37.0]	6.7 [1.5, 18.3]	40.5 [18.6, 71.1]	<0.001
34MHA	181.9 [105.0, 434.8]	128.1 [89.8, 202.7]	495.7 [302.3, 796.9]	<0.001

In the entire nested sample, higher levels of the metabolites of acrolein (CEMA, 3HPMA), acrylonitrile (CYMA), 1,3-butadiene (MHBMA3), crotonaldehyde (HPMMA), and xylene (34MHA) were associated with higher SBP (Supplemental Figure S2, Table S3). No significant associations were observed with DBP or hypertension. These effects were modified by age and sex where individuals <60 years and males had stronger associations than females (Supplemental Figure S3).

In non-smokers, we found that the urinary levels of the metabolites of acrolein and crotonaldehyde were associated with SBP. There was a 1.6 mm Hg (95%CI: 0.4, 2.7; p = 0.007) and a 0.8 mm Hg (95%CI: 0.01, 1.6; p = 0.049) higher SBP per IQR of acrolein metabolites CEMA and 3HPMA, respectively, and a 1.0 mm Hg (95%CI: 0.08, 0.8; p = 0.02) higher SBP per IQR of HPMMA. Additionally, the styrene metabolite (MA) was associated with 0.4 mm Hg (95%CI: 0.09, 0.8, p = 0.02) higher DBP (Figure 1, Table S3).

Figure 1. Associations between urinary VOC metabolites and BP and hypertension by 656 smoking status in participants of the Jackson Heart Study.

Urine samples were collected 657 from 1194 female and male participants of the Jackson Heart Study consisting of 778 non-658 smokers (urinary cotinine ≤40ng/mg) and 416 age- and sex-matched current smokers (urinary 659 cotinine >40ng/mg). Panels show differences in systolic (A) and diastolic (B) blood pressure (mm 660 Hg), and RR of hypertension (C), as well as 95% CI per IQR for each VOC metabolite shown on 661 the Y axis by smoking status: n=778 nonsmokers (open circles) and n=416 smokers (filled 662 circles). The levels of VOC metabolites (ng/mL) were normalized to the levels of creatinine 663 (mg/mL) in the urine (ng/mg creatinine). Models are adjusted for age, sex, BMI, physical activity, 664 education, eGFR, cholesterol:HDL, triglycerides, diabetes mellitus, blood pressure medications, 665 and ambient PM_2.5 levels on the day of enrollment.

In smokers, we found that the urinary levels of metabolites of acrylonitrile, 1,3-butadiene, and crotonaldehyde were associated with SBP. There was a 2.5 mm Hg (95% CI: 0.1, 5.0; p = 0.05) higher SBP per IQR of acrylonitrile metabolite, CYMA. There was a 3.9 mm Hg (95% CI: 1.3, 6.4; p = 0.003) higher SBP per IQR of 1,3-butadiene metabolite, MHBMA3. Finally, the crotonaldehyde metabolite was associated with higher SBP independent of smoking status with a 3.4 mm Hg higher (95% CI: 0.8, 5.9; p = 0.01) SBP per IQR of HPMMA among smokers. No significant associations were found with DBP or hypertension.

Smoking status was associated with both SBP and hypertension (Figure 2, Table S5). In comparison with non-smokers, smokers had 2.8 mm Hg (CI: 0.5, 5.1, p = 0.02) higher SBP and a 1.2-fold greater risk of hypertension (CI: 1.1, 1.4; p = <0.001). Additionally, the highest quartile of cotinine was associated with a 5.3 mm Hg (CI: 2.3, 8.3; p = <0.001) higher SBP and 1.3 greater risk (CI: 1.1, 1.5; p = 0.048) of hypertension. No associations were observed with CPD or pack-years (Figure S4).

Figure 2. Associations between smoking and BP and hypertension.

From the Jackson Heart Study 1194 female and male participants were randomly selected consisting of 778 non-smokers (urinary cotinine ≤40ng/mg) and 416 age- and sex-matched current smokers (urinary cotinine >40ng/mg). Panels show differences in systolic (A) and diastolic (B) blood pressure (mm Hg), and RR of hypertension (C), as well as 95% CI as compared to non-smokers (Smoker,defined by urinary cotinine >40 ng/mg), or quartile of urinary cotinine (ng/mg creatinine, Cotinine Q2-Q4), or cigarettes per day (CPD) on the Y axis. Models were adjusted for age, sex, BMI, physical activity, education, eGFR, cholesterol:HDL, triglycerides, diabetes mellitus, blood pressure medications, and ambient levels of PM_2.5 on the day of enrollment.

Smoking status, urinary cotinine, CPD, and pack-years were associated with higher urinary levels of all VOC metabolites, except metabolites of benzene (MU), 1-bromopropane (BPMA), and toluene (BMA) as shown in Figure S5 (and Tables S5 and S6). The acrylonitrile metabolite (CYMA) was the most robust indicator of tobacco exposure with a high correlation with cotinine (ρ = 0.8) and a large association with smoking status (Figure S5, Table S6). Strong associations were observed between urinary cotinine levels and CPD with CEMA, 3HPMA, CYMA, MHBMA3, HPPMA, AMCC, 2MHA, and 3,4MHA (Figure S5, Table S6 and S7). Associations were generally stronger for urinary cotinine levels than CPD. Levels of urinary VOC metabolites were not as strongly associated with pack-years, which is consistent with VOC metabolites being a measure of acute, rather than chronic exposure to cigarette smoke.

To better understand the effect of individual VOC metabolites as well as mixtures of VOC metabolites, we used BKMR to explore the effect of VOCs in non-smokers and smokers. For SBP among non-smokers, there was no significant relationship with the overall VOC mixture and SBP, however, the acrolein metabolite, CEMA, had the highest PIP of 0.5322, indicating CEMA as the primary driver of higher SBP among non-smokers (Figure S7). For DBP among non-smokers, we found a positive, though not significant relationship between the VOC mixture and DBP, primarily driven by the styrene metabolite (MA) with a PIP of 0.1421, supporting previous results for higher DBP in non-smokers (Figure S8). For hypertension among non-smokers, there was a negative, relationship with the VOC mixture and risk of hypertension, driven by the xylene metabolite, 34MHA (Figure S9). In smokers, we found a positive relationship between the VOC mixture and SBP, primarily driven by acrolein metabolite (3HPMA) with a PIP of 0.2961 (Figure S10). For hypertension, crotonaldehyde metabolite, HPMMA, had the highest PIP, 0.1722. There was a positive, though not significant relationship, between the VOC mixture and hypertension (Figure S12). There were no clear associations between the VOC mixture and DBP among smokers (Figure S11).

In our mediation analysis (Table S8). we found that crotonaldehyde significantly mediated the relationship between smoking and SBP, with a total effect of 2.8 mm Hg (Bootstrap 95% CI: 0.5, 5.1), direct estimate of 0.5 mm Hg (Bootstrap 95% CI: −2.3, 3.2) and an indirect effect estimate of 2.3 mm Hg (Bootstrap 95% CI: 0.9, 3.9). Finally, we conducted several sensitivity analyses. Imputing underlying blood pressure slightly strengthened estimates, but changes were less than 10%. After correcting for multiple tests, the association between the crotonaldehyde metabolite, HPMMA, and SBP remained significant. Sensitivity analyses that further adjusted for income, health insurance, or diabetes medications did not change effect estimates more than 10%. See supplemental information for sensitivity analyses results (Supplemental Figures S13-S19).

DISCUSSION

In this study of Black participants, we found that levels of the urinary metabolites of VOCs such as acrolein, crotonaldehyde, and styrene were associated with elevated blood pressure and this association was significant in both non-smokers and smokers, suggesting that regardless of the source, VOC exposures may be significant contributors to the risk of hypertension. Additionally, we found that current smoking was associated with elevated SBP and higher risk of hypertension. In smokers, metabolites derived from acrolein, acrylonitrile, 1,3-butadiene, and crotonaldehyde were associated with higher SBP, particularly in individuals who were males and younger than 60 years. Mediation and BKMR analyses supported these results. Taken together, these observations suggest that exposure to VOCs, regardless of their source of origin, could increase blood pressure, and that some of the vascular effects of smoking may be attributable, in part, to VOCs found in tobacco smoke.

High BP is a major modifiable risk factor for CVD. Interventions to reduce BP include BP lowering medications, quitting smoking or e-cigarette use, and other diet and physical activity interventions. Our study indicates that VOC exposure is associated with high BP and hypertension risk, independent of smoking status, implicating other sources of VOCs contributing to increased CVD risk. Air pollution is one of the largest global environmental threats, responsible for approximately 9 million deaths worldwide.³⁴ Exposure to air pollutant, PM_2.5, is associated with greater risk of hypertension and high BP.³⁵ The emphasis on PM_2.5 has largely been motivated by the availability of monitoring data, however, in addition to particulate matter, polluted air also contains a number of gaseous or vaporous co-pollutants such as VOCs. Our study of VOC metabolites reports effect sizes ranging from 0.4 mm Hg higher SBP for styrene metabolite, MA, to 1.6 mm Hg for acrolein metabolite, CEMA. Compared to our study, SBP effect sizes are much higher than many of those reported for PM_2.5 (0.53 mm Hg; 95% CI 0.26–0.8) in meta-analyses and indicate significant contribution to the burden of CVD.^{23, 35}

Previous work has shown inconsistent associations between smoking and BP. Although smoking acutely increases BP, smoking cessation is not accompanied by a decrease in BP.³⁶ Moreover, data from large studies such as the Framingham Heart Study³⁷ and NHANES³⁸ show an inverse relationship between smoking and BP. Furthermore, in a recent meta-analysis of 141,317 individuals with European ancestry, smoking heaviness was associated with increased heart rate but not BP.³⁹ However, most of these findings are based on data from non-Black populations. In contrast to these findings, in our study of Black participants, we found a strong association between smoking and hypertension, and with higher SBP. Reasons for such discrepancies are unclear but may relate to different populations examined and different methodological approaches. The robust association between smoking and hypertension seen in our study may be because we studied, for the first time, a large sample of Black individuals living in a similar environment, and/or because we closely matched smokers to non-smokers on age and sex, allowing us to make direct comparisons between nearly equal sized population (2:1 ratio of non-smokers and smokers), a comparison not afforded by previous studies. However, we cannot rule out the possibility that the association between smoking and hypertension is particular to our study population and/or their basal risk of hypertension.

Previous work has linked the elevated risk of hypertension in Black populations to psychosocial factors as well as excessive sodium and low potassium intake, excessive alcohol intake, and obesity.⁴⁰ Our results suggest that in addition to these factors, VOC exposure (from environmental sources or tobacco smoke) may also be a significant contributor to hypertension risk in Black individuals. While obesity is associated with greater risk for hypertension, and lower rates of blood pressure in smokers have been linked to lower BMI^{37, 38} results from the JHS study show that even though smokers had lower BMI, they have elevated SBP and higher hypertension risk than non-smokers. Thus, the effects of smoking on blood pressure, at least in the present study, seems to be independent of obesity.

We found a specific association of several VOCs such acrolein, 1,3-butadiene, and crotonaldehyde with systolic, but not diastolic, BP which may be reflective of the high sensitivity of the Black population for systolic hypertension. Systolic hypertension is usually the result of arterial thickening and stiffening caused either by atherosclerosis, medial degradation, or impaired endothelial-mediated vasodilation,⁴¹ processes that are highly sensitive to VOCs.^{16, 42, 43} Previous work in animal models has shown that exposure to VOCs such as acrolein can induce endothelial injury, vascular dysfunction, dyslipidemia, and accelerate the progression of atherosclerotic lesions,^{11, 44, 45} suggesting that exposure to VOCs may represent a novel unrecognized CVD risk factor and that vascular injury induced by smoking, may be in part, attributable to VOCs present in tobacco smoke.

Our observations that VOC metabolites of acrolein and crotonaldehyde were associated with increased systolic blood pressure in non-smokers indicate that environmental sources of VOCs may also contribute to increased CVD risk, supporting our previous analysis of VOC metabolites and vascular dysfunction in non-smokers in Louisville, KY.¹⁶ In the Louisville Healthy Heart Study, the acrolein metabolite, 3HPMA, was associated with higher SBP, with effect estimates higher in Black participants compared to White participants. These differences may represent a racial disparity in environmental exposures; however, this is not a disparity we were able to explore in the all-Black Jackson Heart Study. Although both acrolein metabolites, CEMA and 3HPMA, were associated with higher BP, CEMA was the more significant metabolite in this population.

VOCs are primarily detoxified via glutathione-linked pathway, which generates N-acetyl cysteine-S-conjugates that are excreted in urine. Some VOCs, such as benzene, undergo phase I metabolism, catalyzed by cytochrome P450. Therefore, genotype and metabolic clearance of VOCs can vary by age, sex, race, diet, water intake, and eGFR. Despite such sources of variability, we found that metabolites of acrolein, acrylonitrile, 1,3-butadiene, crotonaldehyde, and xylene were associated with increased SBP. Of these compounds, acrolein and 1,3-butadiene have previously been referenced as two of the most harmful components of cigarette smoke, though the contribution of these chemicals as CVD risk factors has not yet been widely accepted in tobacco regulation. However, there is evidence that exposure to acrolein increases CVD risk by increasing BP in both normotensive and hypertensive rats,⁴⁶ suppressing endothelial nitric oxide synthase activation,⁴⁷ attenuating endothelial cell migration,⁴⁸ blocking vascular endothelial growth factor,⁴⁴ and reducing circulating levels of angiogenic cells, specifically in participants with hypertension.^{10, 14} Some studies have also reported that exposure to 1,3-butadiene increases CVD risk, particularly increased odds of high BP in normotensive pregnant women, decreased endothelial function and increased DBP in non-smokers, and increased arteriosclerotic heart disease in exposed workers.^{49, 50} Additional data from animal models will be helpful in further substantiating the links uncovered by our study.

Exposure to crotonaldehyde was significantly associated with SBP, independent of smoking status. Crotonaldehyde is an unsaturated aldehyde and an atherogenic compound.⁵¹ Acute and chronic exposure to crotonaldehyde in mice decreases DBP, attenuates acetylcholine induced aortic relaxation,⁵² and causes increased vasotoxicity in aorta and superior mesenteric artery.⁴³ However, exposure to crotonaldehyde has not previously been reported to be associated with SBP in humans. The unsaturated aldehyde is produced endogenously from lipid peroxidation as well as 1,3-butadiene metabolism. It is also present in vehicle exhaust and cigarette smoke, which may represent major sources of human exposure. In addition to crotonaldehyde, exposure to acrolein and styrene was associated with BP in non-smokers which suggests that exposure to even low environmental levels of these VOCs could contribute to higher BP.

Overall, the results of our study suggest that exposure to VOCs increases the risk of hypertension. However, because the JHS is made-up of all-Black participants, results may not be generalizable to other race/ethnic groups. Moreover, due to the cross-sectional design of our study, we could not infer causality or rule out residual confounding. Finally, because the half-lives of the VOC metabolites vary due to differences in times of exposure, there may be additional variability in the data leading to exposure misclassification.

Despite these limitations, our study has many strengths. These include a nested study design of age- and sex-matched group of smokers and non-smokers based on cotinine values. We used mediation analyses, and both single and multi-pollutant models to characterize this relationship in non-smokers and smokers using the novel BKMR method. Nevertheless, further work is required to corroborate our findings. Longitudinal studies are required to identify whether exposure to VOCs leads to an increase in BP and incident hypertension, and whether the effects of VOCs differ among individuals of different race/ethnicities. Additionally, more comprehensive models are needed with the growing evidence of the association between high blood pressure and covarying exposures, such as road traffic noise^53–55 and other emerging pollutants.

PERSPECTIVES.

VOCs are frequent components of ambient and indoor air; however, it remains unclear whether exposure to VOC represent a significant CVD risk. In this study, we found that VOC exposure in Black individuals is significantly associated with elevated blood pressure and higher risk of hypertension. Although derived from a cross-sectional sample, these findings may explain at least in part, the burden of hypertension and heart disease in Black populations. Thus, decreasing exposures to VOCs such as acrolein, 1,3-butadiene, crotonaldehyde and styrene present either in ambient air or in tobacco smoke could significantly improve public health.

HIGHLIGHTS.

• Exposure to VOCs is associated with higher risk of hypertension.

• VOCs present in the ambient air or tobacco smoke are associated with increased SBP.

• VOC exposure may contribute to hypertension in Black populations, particularly among those who smoke.

• Reduction of VOC exposure may improve cardiovascular health.

NOVELTY AND RELEVANCE.

What is New?

• Exposure to VOCs is associated with higher risk of hypertension.

• VOCs present in the ambient air or tobacco smoke are associated with increased SBP.

What is Relevant?

• VOC exposure may contribute to hypertension in Black populations, particularly among those who smoke.

Clinical/Pathophysiological Implication

• Reduction of VOC exposure may improve cardiovascular health.

NONSTANDARD ABBREVIATIONS AND ACRONYMS

CVD

Cardiovascular Disease

BP

Blood pressure

SBP

Systolic Blood Pressure

DBP

Diastolic Blood Pressure

HTN

Hypertension

VOCs

Volatile Organic Compounds

JHS

Jackson Heart Study

CARDIA

Coronary Artery Risk Development in Young Adults

NHANES

National Health and Nutrition Examination Survey

TOBA

Tobacco Questionnaire

CPD

Cigarettes per day

PKYR

Pack-years

PM2.5

Fine particulate matter of ≤2.5μm in diameter

CHD

Coronary heart disease

BMI

Body mass index (kg/m2)

LDL

low density lipoprotein

HDL

high density lipoprotein

eGFR

estimate glomerular filtration rate

FPG

fasting plasma glucose

Hba1c

Acetylated hemoglobin

EPA

U.S. Environmental Protection Agency

GLM

Generalized Linear Model

BKMR

Bayesian Kernel Machine Regression

IQR

Interquartile Range

RR

Relative Risk

CI

Confidence interval

P

p-value

PIP

Posterior inclusion probability