Abstract
Cigarette smoking is associated with adverse cardiac outcomes including incident heart failure (HF). However, key components of potential pathways from smoking to HF have not been evaluated in older adults. In a community-based study we studied cross-sectional associations of smoking with blood and imaging biomarkers reflecting mechanisms of cardiac disease. Serial nested, multivariable Cox models were used to determine associations of smoking with HF, and to assess the influence of biochemical and functional (cardiac strain) phenotypes on these associations. Compared to never smokers, smokers had higher levels of inflammation (CRP and IL6), cardiomyocyte injury (hscTnT), myocardial “stress”/fibrosis (sST2, galectin3); and worse LV systolic and diastolic function. In models adjusting for age, sex, and race (DEMO), and for clinical factors potentially in the causal pathway (CLIN), smoking exposures were associated with CRP and IL6, sST2, hscTnT, and with NTproBNP (in whites). In DEMO adjusted models cumulative burden of smoking was associated with worse LV systolic strain. Current and former smoking were associated with HF in DEMO models (HR 1.41, 95% CI 1.22-1.64 and HR 1.14, 95% CI 1.03-1.25, respectively), and with current smoking after CLIN adjustment. Adjustment for time-varying myocardial infarction (MI), inflammation, cardiac strain, hscTnT, sST2, galectin3 did not materially alter the associations. Smoking was associated with HF with preserved, as well as, decreased ejection fraction. In conclusion, in older adults, smoking is associated with multiple blood and imaging biomarker measures of pathophysiology previously linked to HF, and to incident HF even after adjustment for clinical intermediates.
Keywords: aging, smoking, heart failure, biomarkers, cardiac strain
INTRODUCTION
The leading cause of smoking-attributable mortality is cardiovascular disease, exceeding lung cancer and COPD (1). While smoking is associated with coronary artery disease, even in its absence, smoking remains a risk factor for heart failure (HF) (2-5). The mechanisms underlying the association of smoking with HF are incompletely understood. Research has shown an association of smoking with decreased systolic and diastolic LV function, and cardiac interstial fibrosis (2,6-8). In population studies, smoking has been associated with inflammation (9), increased N terminal pro-brain natriuretic peptide (NTproBNP) (10), and cardiac troponin T (hscTnT) (11), all known predictors of incident HF (12-15). A nexus between LV function and biomarkers associated with HF (hscTnT, NTproBNP) and fibrosis (soluble suppression of tumorigenicity 2 biomarker - sST2) is suggested by data (16) showing baseline associations with LV longitudinal strain (LVLS) and LA reservoir strain (LARS). However, to our knowledge there are no studies in older adults which have evaluated potential major components of the pathway from smoking exposure to HF – i.e. inflammation, myocardial injury/repair, subclinical systolic and diastolic dysfunction. We leveraged data from the Cardiovascular Health Study on smoking exposures, blood biomarkers reflecting mechanisms of cardiac disease, as well as sensitive echocardiographic strain measures of cardiac function to study these relationships (Figure 1). In this study, we considered blood and imaging biomarkers as smoking outcomes, as well as potential intermediates in the relationship between smoking and HF.
Figure 1.
Conceptual illustration showing potential relationships between cigarette smoking, intermediate pathobiologic pathways, and heart failure.
METHODS
CHS is a multicenter, prospective, observational cohort study of cardiovascular disease (CVD) in older adults. Participants (N=5201) included those initially enrolled in 1989-90 (original cohort), and an African-American supplemental cohort (N=687) enrolled in 1992-93. The methodology and design of the CHS have been reported previously (17, 18) and are summarized in Supplemental Methods. Participants with a history of CHF at baseline were excluded from analysis.
Smoking information (cigarette smoking only) was obtained via questionnaire at baseline (1989-90) for the initial cohort, and at year 5 (1992-3) for the supplemental cohort. The questionnaire requested information on current and former cigarette smoking included estimates on total years of smoking, age at starting and stopping smoking, average number of cigarettes smoked per day, as well as passive exposure to smoking. Participants who smoked >100 cigarettes or five packs in their lifetime were defined as “ever” smokers; those who smoked cigarettes within the last 30 days were classified as “current” smokers, those who responded positively to having smoked > 100 cigarettes in their lifetime, but had not smoked within 30 days were classified as “former” smokers. “Passive” smokers were neither current nor former smokers, but responded positively to the question on living with others who did smoke regularly. The cumulative exposure to cigarette smoking (pack-years) was calculated from number of years of smoking and estimated number of cigarettes smoked per day. “Never” smokers were those who had no history of current, former or passive smoking.
Methods to assess HF in CHS have been reported previously (19,20).
The design of the echocardiography protocol used in CHS has been described in detail elsewhere (21) and in the Supplemental Methods. To obtain measurements of LVLS, LV end-diastolic strain rate (LVEDSR) and LARS, between 2017- 2018 archived echocardiograms originally recorded during the 1989-90 baseline visit for the original cohort on videotape in analog format were digitized as previously described (16). LV longitudinal strain was measured in 4,015; LA reservoir strain in 3,918, and LV end-diastolic strain rate in 3,929 participants. Details of the strain analyses are described in the Supplemental Methods.
NT-proBNP , hscTnT, sST2 and Gal 3 were measured in serum collected in 1989-90 (original cohort), 1992-93(both cohorts) , and 1994-5 (supplemental cohort) and stored at −70C to −80C. Inflammation biomarkers ( interleukin 6 -IL6, C-reactive protein (CRP), and fibrinogen (fbg) - were measured from blood samples at the time of the baseline examination. NTproBNP was measured in 3098; sST2 in 2,761; hscTnT in 3,017; galectin 3 in 2,736, hsCRP in 5,532, IL6 5,131, and fibrinogen 5,519 participants .The methods for assays have been previously reported (13.14, 23, 25-27) and are further described in the Supplemental Methods. Of note,asessment of IL-6 utilized the Quantakine immunoassay (R&D systems) whereby levels are lower than in commonly used assays. The assay was not harmonized with other assays, which makes its levels not directly comparable with others.
To determine the association of smoking exposures with incident HF we computed the HF incidence rates with Poisson regression with offset to accommodate differential follow-up overall, and by smoking categories. The primary exposures were smoking status with current, former, passive, and never smokers categories, and pack-years of smoking. We used pack-years continuously and also categorized them into four categories as zero, followed by tertiles for those with a positive number of pack-years.
Kaplan- Meier cumulative survival curves were constructed. Cox hazards models were used to estimated hazard ratios of incident HF by smoking exposure. Adjustments for covariates were done using several nested models (M). These were: M0- age, male gender, black race; M1: M0 + height, weight, systolic BP, diastolic BP, history of hypertension, diabetes, stroke, TIA, intermittent claudication, GFR (cystatin C), myocardial infarction, serum total cholesterol, HDL cholesterol, LDL cholesterol, serum insulin, serum glucose; M2: M1 + ln c-reactive protein, fibrinogen, interleukin 6; M3: M2 + LVLS, LVEDSR, and LARS; M4: M3 + main effects and interactions between echo reader and strain measure quality score to control for the effect of reader and quality of strain data; M5: M2 + ln transformed biomarkers ( hscTnT, NTproBNP, gal 3, sST2); M6: M4 + ln transformed biomarkers ( hscTnT, NTproBNP, galectin 3, sST2) in the original cohort, since strain measures were restricted to the original cohort. In a secondary analysis, we added time varying myocardial infarction to Model M1 to test the hypothesis that interval acute myocardial infarction affected the relationship between smoking exposures and incident HF. In other secondary analyses, we used cause-specific Cox regression to evaluate association of smoking exposures with HF functional phenotypes – HF with preserved ejection fraction (HFPEF) and HF with reduced ejection fraction (HFREF). We considered M0 as the main model, with M1 and subsequent models used to determine the strength of associations after accounting for factors potentially in the causal pathway between smoking and outcomes.
Cross-sectional associations between smoking exposure and baseline blood biomarkers (CRP, IL6, fbg, hscTnT, NTproBNP, sST2, gal3) in both cohorts, and with LVLS, LVEDSR, and LARS ( original cohort), were assessed by linear regression analyses using nested, incrementally-adjusted models as described above.
Due to heavy tails, biomarkers CRP, hscTnT, NTproBNP, sST2, and galectin3 were In transformed in all analyses. We report baseline characteristics as means ± standard deviation for continous variables and counts (percentages) for categorical variables. We compared the characteristics of the participants across smoking status by using t-tests for continuous variables and chi-square tests for categorical variables, with never smokers being the reference group.
We also assessed adjusted associations between smoking exposures and CRP, IL6, fbg, hscTnT, NTproBNP, sST2, gal3. Cross-sectional analyses were performed using models M0, M1, and M2. As a sensitivity analysis, we used relative risk estimation in models for binary biomarkers, specifically sST2 >35 ng/ml, NTproBNP >190 ng/ml, galectin −3 >17.8 ng/ml and hscTnT >13 ng/ml. For hsc-TnT, subjects with levels below the limit of detection (3 ng/ml) were given imputed values at 2.99 ng/ml.
Further we modeled with linear regression the interval change in biomarkers (i.e.1989-90 to 1992-93 for the initial cohort; 1992-93 to 1994 -1995 for the supplemental cohort). In these models we added the baseline biomarkers among covariates in each model.
In the original cohort, we modeled with cross-sectional linear regression LVLS, LVEDSR and LARS. We used covariates from models M0, M1 and M2 augmented with main effects and interactions between echo reader and strain measure quality score to control for the effect of reader and quality of CM data. As a sensitivity analysis we analyzed separate models by race.Analyses were performed in R [ref R Core Team (2020). R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria.URL https://www.R-project.org/.]
RESULTS
Many baseline characteristics differed by smoking status (Table 1). Compared to never smokers, current smokers were younger, smaller, less likely to be black, had higher alcohol use, higher cystatin C, lower GFR, lower HDL cholesterol, lower systolic and diastolic BP, less history of hypertension, higher CRP (as well as IL6, fibrinogen, white blood cell count, and sCD14) and higher hemoglobin levels.
Table 1 –
Descriptive Characteristics by Smoking Category
| Total (5607) | Never (2396) | Former (2331) | Current (674) | Passive (206) | |
|---|---|---|---|---|---|
| Age (yrs) | 72.7 ±5.6 | 73.7 ±6.0 | 72.4 ±5.2 * | 70.8 ±4.6* | 71.7 ±5.2 |
| Height (cm) | 164.9 ±9.5 | 162.6 ±9.0 | 167.4 ±9.4 * | 164.9 ±9.1* | 163.56 ±9.9 |
| Weight (kg) | 72.6 ±14.6 | 71.1 ±14.7 | 75.3 ±14.2 * | 68.9 ±14.0* | 72.4 ±15.4 |
| BMI (kg/m2) | 26.65 ±4.67 | 26.84 ±4.92* | 26.82 ±4.3 | 25.3 ±4.55 | 27.05 ±5.27 |
| Female (%) | 3248 (57.9%) | 1705 (71.2%) | 995 (42.7%)* | 409 (60.7%)* | 139 (67.5%) |
| Black (%) | 861 (15.4) | 397 (16.6) | 305 (13.1) * | 137 (20.3)§ | 22 (10.7)§ |
| GFR (ml/min) | 78.3 ±19.5 | 79.0 ±19.4 | 78.4 ±19.4 | 75.8 ±19.6* | 77.3 ±20.0 |
| LDL (mg/dl) | 130.4 ±35.0 | 132.7 ± 35.0 | 129 ±35.0 § | 129.5 ±37.7 | 133.9 ±35.6 |
| HDL (mg/dl) | 54.4 ±15.8 | 55.8 ±15.5 | 53.2 ±15.8 * | 53.9 ±16.4§ | 54.8 ±16.1 |
| SBP (mmHg) | 136.6 ±21.6 | 138.5 ±21.8 | 135.6 ±21.4 * | 133.6 ±21.6* | 135.3 ±21.3 § |
| DBP (mmHg) | 70.9 ±11.3 | 71.2 ±11.5 | 70.8 ±11.2 | 70.0 ±11.2 § | 71.8 ±11.0 |
| Glucose (mg/dl) | 110.6 ±36.3 | 109.9±37.1 | 111.5 ±34.6 | 108.5 ±38.6 | 115.3 ±37.6§ |
| Hgb (mg/dl) | 14.0± 1.4 | 13.8 ± 1.4 | 14.2± 1.4* | 14.3± 1.3* | 14.0± 1.7 |
| Edu> 12yrs (%) | 2448 (43.8) | 1007(42.2) | 1108(47.6%)* | 263 (39.3%) | 70 (34.3%) |
| Alcohol (%) | 587 (10) | 138 (5.8) | 325 (14) * | 109 (16.2) * | 15 (7.4) |
| Diabetes (%) | 868(15.6) | 356(15.1) | 382 (16.5) | 85 (12.7) | 45 (22.1)§ |
| Estrogen (%) | 400 (7.1) | 191 (8.0) | 152 (6.5) | 43 (6.4) | 14 (6.8) |
| Beta B1 (%) | 695 (12.4%) | 295 (12.3%) | 306 (13.1%) | 65 (9.7%) | 29 (14.1%) |
| ACEI (%) | 333 (5.9%) | 143 (6.0%) | 156 (6.7%) | 29 (4.3%) | 5 (2.4%) |
| CHD (%) | 979 (17.5) | 360 (15.0) | 474 (20.3)* | 98 (14.5) | 47 (22.8)§ |
| Stroke (%) | 215(3.8) | 78 (3.3) | 104 (4.5)§ | 27 (4.0) | 6 (2.9) |
| TIA (%) | 143 (2.6) | 45(1.9) | 76 (3.3)§ | 18 (2.7) | 4 (1.9) |
| CKD (%) | 1462 (29.8) | 633 (30) | 598 (29.5) | 178 (29.7) | 53 (29.8) |
| HTN (%) | 3269 (58.4) | 1467 (61.3) | 1335 (57.3)§ | 358 (53.1)* | 109 (53.2) |
| COPD (%) | 722 (17.1) | 120 (6.6) | 396 (22.0) * | 196 (43.5) * | 10 (6.6) |
ACEI = angiotensin converting enzyme inhibitor medication; Beta BL= beta blocker medication; CHD = coronary heart disease; CKD = chronic kidney disease; DBP = diastolic blood pressure; Edu = years of education ; estrogen = number of participants receiving estrogen supplementation; GFR = glomerular filtration rate (cystatin C); HDL = high density lipoprotein; HTN = hypertension; LDL = low density lipoprotein; SBP = systolic blood pressure; TIA = transient ischemic attack.
p < 0.05
p < 0.001 smoking category vs “never”.
Cardiac strain and “cardiac” biomarkers (i.e.NTproBNP, hscTnT, sST2, and gal3) values also differed according to smoking status (Table 2). Former smokers had lower LVLS and LVEDSR. Former and current smokers had lower NTproBNP levels than never smokers. Former smokers had higher levels of hscTnT and sST2. All smoking categories had higher CRP levels.
Table 2 -.
Cardiac Strain and Blood Biomarkers by Smoking Category
| Smoking | Total | Never | Former | Current | Passive |
|---|---|---|---|---|---|
| LVLS (%) | 14.22 ±3.74 | 14.35±3.74 | 14.03±3.69* | 14.34±3.82 | 14.42±4.03 |
| LARS (%) | 40.61±0.15 | 40.28±15.49 | 40.77±15.51 | 41.6±15.44 | 39.49±14.94 |
| LVEDSR (1/sec) | 0.65±0.24 | 0.66±0.24 | 0.64±0.23* | 0.68±0.25 | 0.64 ±0.20 |
| NTproBNP (pg/mL) | 240.3± 675.6 | 253.7 ±730.2 | 240.5 ±714.6§ | 193.8 ±276.4* | 219.4 ±282.0 |
| hscTnT (pg/mL) | 9.0 ± 27.0 | 8.6 ±18.7 | 9.9 ±37.0* | 7.8 ±8.1 | 8.23 ±8.0 |
| sST2 (ng/mL) | 25.3±10.3 | 24.7±8.8 | 25.8 ±10.7§ | 25.1 ±13.3 | 26.2 ±10.2 |
| Gal3 (ng/mL) | 16.6 ±6.5 | 16.9 ±6.7 | 16.3 ±6.1* | 16.3 ±6.4 | 17.2 ±8.4 |
| ↑NTproBNP | 1291 (30%) | 604 (32.1%) | 492 (27.5%)§ | 141 (29.4%) | 54 (34.2%) |
| ↑hscTnT | 692 (16.4%) | 281 (15.2%) | 307 (17.6%) | 75 (16%) | 29 (19%) |
| ↑sST2 | 495 (12.7%) | 194 (11.4%) | 227 (13.9%)* | 51 (11.6%) | 23 (16.7%) |
| ↑Gal3 | 1254 (32.4%) | 573 (33.9%) | 499 (30.9%) | 132 (30.2%) | 50 (37%) |
| Fibrinogen (mg/dL) | 323 ± 67 | 322 ± 65 | 318 ± 67§ | 340 ± 69* | 319 ± 61 |
| IL6 (pg/ml) | 2.17 ± 1.90 | 2.09 ± 2.05 | 2.16 ± 1.74 | 2.44 ± 1.65* | 2.18 ± 1.63 |
| CRP (mg/dL) | 4.66 ±8.14 | 4.25 ±7.48 | 4.74 ±8.87§ | 5.86 ±8.3* | 4.53 ±5.71§ |
| sCD14 (ng/ml) | 1636±357 | 1636±350 | 1515±347* | 1699±400§ | 1674±362 |
| TNFaR1 (pg/ml) | 1544±655 | 1537±653 | 1554±670 | 1532±623 | 1559±602 |
| WBC (x 1000) | 6.29±2.10 | 6.00±1.68 | 6.28±2.20§ | 7.28±2.40§ | 6.47±3.02* |
CRP = C-reactive protein; Gal 3 = galectin 3; IL6 = interleukin 6;. LARS = left atrial reservoir strain; LVEDSR = left ventricular end-diastolic strain rate; LVLS – left ventricular longitudinal strain; NTproBNP = N terminal brain natriuretic protein ; sST2 = soluble suppression of tumorigenicity 2 biomarker. TNFaR1 = tumor necrosis factor receptor 1; WBC = white blood cell count.x ↑=increased above pre-specified partition values.
p ≤ 0.05
p ≤ 0.01 smoking category vs. never smoking (ln transformed values used for pairwise comparisons of NTproBNP, hscTnT, sST2, Gal3, and CRP).
In models adjusted for demographics and for risk factors no smoking exposure was associated with any strain measure (Supplemental Table 1). Higher cumulative burden of smoking (pack years > 1 SD) was inversely associated (β −0.151, −0.266, −0.036, p= 0.01) with LV systolic strain in a demographics adjusted model.
In risk-factor adjusted models (Supplemental Table 2a) current smoking was strongly associated with CRP and less strongly associated with former and passive smoking, as well as with cumulative smoking burden. IL6 was significantly associated (Supplemental Table 2b) with all smoking exposures. Fibrinogen (Supplemental Table 2c) was significantly associated with current smoking and cumulative smoking burden.
For the whole cohort, NTproBNP was not associated with any smoking exposure (Supplemental Table 3). In a secondary analysis comparing NTproBNP in whites (n =3623) and blacks (n= 684), average lnNTproBNP was lower (p<0.001) in blacks (4.50 ± 1.24) than in whites (4.77 ±1.14). In whites, multiple smoking exposures were significantly associated with NTproBNP levels, and with increased NTproBNP, as well as with interval change in NTproBNP. Specifically, in demographics adjusted models of white participants, there was association of NTproBNP with cumulative smoking (β 0.08, CI 0.01, 0.15; p = 0.03), increased NTproBNP (>190 pg/ml) with current smoking (RR 1.18; CI 1.00, 1.40; p = 0.05); and interval increase in NTproBNP levels with current smoking (β 0.11; CI 0.0,0.21; p = 0.05). In risk factor adjusted models, increased NTproBNP in white men was associated with current smoking (RR 1.37; CL 1.06, 1.81; p = 0.02). There were no significant association of any measure ofNTproBNP with any smoking exposure in black participants.
In multivariable models adjusted for risk factors and for inflammation markers, sST levels as well as increased sST2, were associated with passive smoking, and with cumulative burden of smoking (Supplemental Table 4,5).
Increased hscTnT was associated (Supplemental Table 5) with passive smoking in a demographics adjusted model, and with current smoking in a risk factor adjusted model. Galectin 3 was not associated with any smoking measure in adjusted models.
At the end of follow-up (6/30/2014), 5613 participants were alive and eligible for longitudinal analysis. Over a median follow-up of 13.6 yrs (mean 13.8 yrs, maximum 25.1 yrs), there were 2,018 adjudicated incident cases of HF, with a total incidence rate of 2.92 per 100 person-years (Figure 2, Supplemental Table 7). Incidence rate of HF was greater in men than women, greater in smokers than never smokers, comparable in current and former smokers, and lower in passive smokers. Greater cumulative smoking exposure (pack-years) was associated with a higher incidence of HF (Figure 3, supplemental Table 8).
Figure 2.
Kaplan Meir survival plot - heart failure free interval according to type of smoking exposure.
Figure 3.
Kaplan Meir survival plot –heart failure free interval according to cumulative smoking exposure.
In serial Cox models (Tables 3, 4), incident HF was significantly associated in a demographics adjusted model with former smoking (HR 1.14, 1.029- 1.25) and current (HR 1.41, CI 1.2-1.6), and in risk-factor adjusted models with current smoking (HR 1.57, CI 1.34-1.83. In a secondary analysis adding time-varying acute myocardial infarction to model M1, the association of incident HF with current smoking was not attenuated (HR 1.62, CI 1.37-1.91). Significant associations of current smoking with incident HF persisted (HR 1.62, CI 1.26-2.08) in additional models also controlling for inflammation, cardiac strain measures, and pre-specified biomarkers of interest ((hscTnT, NTproBNP, sST2, gal3). Cumulative smoking burden was significantly associated (Table 4) with HF in demographics-adjusted models, as well as risk factor, and risk factor + inflammation-adjusted models.
Table 3.
Cox Models of Smoking Exposures and Heart Failure
| HR | 95%CI | p-val | HR | 95% CI | p-val | ||
|---|---|---|---|---|---|---|---|
| M0 | M3 | ||||||
| Former M0 | 1.138 | (1.032,1.253) | 0.009 | Former M3 | 1.134 | (0.998,1.288) | 0.053 |
| Current M0 | 1.412 | (1.218,1.636) | <0.001 | Current M3 | 1.577 | (1.29,1.927) | <0.001 |
| M1 | lvls | 0.958 | (0.936,0.98) | <0.001 | |||
| Former M1 | 1.096 | (0.991,1.211) | 0.074 | lars | 0.988 | (0.984,0.992) | <0.001 |
| Current M1 | 1.58 | (1.355,1.843) | <0.001 | lvedsr | 0.919 | (0.644,1.312) | 0.643 |
| M2 | M4 | ||||||
| Former M2 | 1.107 | (0.996,1.23) | 0.06 | Former M4 | 1.132 | (0.996,1.287) | 0.058 |
| Current M2 | 1.521 | (1.294,1.788) | <0.001 | Current M4 | 1.593 | (1.301,1.95) | <0.001 |
| fib | 1.001 | (1,1.001) | 0.194 | M5 | |||
| log(crp)) | 1.129 | (1.062,1.2) | <0.001 | Former M5 | 1.1 | (0.969,1.25) | 0.141 |
| il6 | 1.032 | (1.011,1.055) | 0.003 | Current M5 | 1.5 | (1.231,1.827) | <0.001 |
| M0 | M2 | ||||||
| Ever | 1.189 | (1.085,1.302) | <0.001 | Ever | 1.177 | (1.066,1.3) | 0.001 |
| M1 | fib | 1.001 | (1,1.002) | 0.132 | |||
| Ever | 1.175 | (1.069,1.291) | 0.001 | (log(crp) | 1.136 | (1.069,1.208) | <0.001 |
Top part for smoking status of current and former (versus never), bottom part for ever (current or former) versus never. Nested models: M0 – age, gender, black race; M1-M0 + risk factors; M2-M1 + ln CRP, fbg, ln IL-6; M3-M2+LVLS, LARS, LVEDSR; M4- M3 + echo reader interactions, image quality; M5- M2 +ln transformed biomarkers; M6-M4+ln transformed biomarkers original cohort. crp= C-reactive protein, fib= fibrinogen, lars= LA reservoir strain, lvedsr = LV end diastolic strain rate; lvls = LV longitudinal strain.
Table 4.
Cox Models of Cumulative Smoking Exposure (pack years) and Heart Failure
| HR | 95%CI | p-val | |
|---|---|---|---|
| M0 | |||
| pkyrsSD M0 | 1.203 | (1.142,1.267) | <0.001 |
| M1 | |||
| pkyrsSD M1 | 1.156 | (1.094,1.221) | <0.001 |
| M2 | |||
| pkyrsSD M2 | 1.139 | (1.075,1.207) | <0.001 |
| fib | 1.001 | (1,1.002) | 0.196 |
| I(log(crp)) | 1.157 | (1.074,1.245) | <0.001 |
| il6 | 1.038 | (1.005,1.071) | 0.024 |
| M3 | |||
| pkyrsSD M3 | 1.132 | (1.065,1.203) | <0.001 |
| lvls | 0.955 | (0.932,0.978) | <0.001 |
| lals | 0.989 | (0.984,0.993) | <0.001 |
| lvedsr | 0.923 | (0.642,1.326) | 0.665 |
| M4 | |||
| pkyrsSD M4 | 1.134 | (1.066,1.206) | <0.001 |
pkyrsSD = standard deviation of smoking pack years (SD=26.6). Cox model abbreviations per table 3.
Secondary analyses of associations of smoking with HF phenotype are shown in Table 5. There were 650 subjects who developed HFPEF, 456 HFREF, and a larger group (912) in whom systolic LV function at the point-of-care for the incident HF event could not be identified.
Table 5:
Association of Smoking with Incident HFPEF vs. HFREF
| HFPEF | HR | 95%CI | p-val | HFREF | HR | 95% CI | p-val |
|---|---|---|---|---|---|---|---|
| M0 | M0 | ||||||
| Former M0 | 1.212 | (1.024,1.433) | 0.025 | Former M0 | 0.989 | (0.807,1.213) | 0.916 |
| Current M0 | 1.087 | (0.817,1.448) | 0.566 | Current M0 | 1.333 | (0.981,1.811) | 0.066 |
| M1 | M1 | ||||||
| Former M1 | 1.188 | (0.999,1.413) | 0.052 | Former M1 | 0.946 | (0.766,1.169) | 0.608 |
| Current M1 | 1.184 | (0.879,1.596) | 0.267 | Current M1 | 1.478 | (1.074,2.035) | 0.017 |
| M2 | M2 | ||||||
| Former M2 | 1.204 | (1.004,1.443) | 0.045 | Former M2 | 1.00 | (0.799,1.252) | 0.999 |
| Current M2 | 1.222 | (0.9,1.658) | 0.198 | Current M2 | 1.398 | (0.992,1.969) | 0.055 |
Nested Models (M0-M2) per Table 3. HFPEF = heart failure with preserved ejection fraction; HFREF = HF with reduced ejection fraction.
Former smoking was significantly associated with incident HFPEF in minimally-adjusted models, as well as risk factor, and risk factor + inflammation-adjusted models. Current smoking was significantly associated with HFREF in the risk-factor adjusted model, and marginally with the risk factor + inflammation adjusted model. Differences in HFPEF and HFREF associations with smoking were not statistically significant. Cumulative smoking exposure was significantly associated (Supplemental Table 9) with both HFPEF and HFREF in demographics, risk-factor adjusted models, as well as in models additionally adjusted for cardiac strain and prespecified biomarkers. In participants with HFREF, there was additional association with cumulative smoking in a model adjusted for inflammation in addition to risk factors. There were no significant race or gender interactions of associations of smoking with incident HF.
In risk-factor adjusted models (Table 3), inflammation biomarkers (CRP, IL6) were both significantly associated with incident HF in current smokers, but only slightly attenuated the association of current smoking with HF . Although LVLS and LARS were both inversely significantly associated with incident HF in current smokers, addition of LVLS, LARS and LVEDSR to risk factor and inflammation-adjusted models did not attenuate the association of smoking with incident HF. Finally, addition of prespecified biomarkers (hscTnT, gal 3, NTproBNP) of interest did not attenuate the association of current smoking with incident HF in models including strain.
DISCUSSION
The principal findings of this study are that in older community-based individuals_cigarette smoking was associated with incident HF in current and former smokers. The risk for HF included both those with HFPEF and HFREF, and was affected by the cumulative burden of smoking exposure. Moreover, smoking was associated with strain measures of subclinical LV systolic and diastolic dysfunction, and with blood biomarker evidence of inflammation (CRP, IL6), as well as cardiomyocyte injury (hscTnT) and diastolic overload (NTproBNP) – all known predictors of clinically evident HF. Former smoking showed a more modest association with HF than current smoking, one that became borderline non-significant after adjustments for covariates, many of which may represent the effects of smoking and hence are partial causal mediators of the smoking-HF relationship.
Adjustments for the proposed biochemical and echo strain intermediates did not meaningfully influence the associations of current smoking and of cumulative smoking with HF. This suggests that the long-term burden of smoking and its impact on HF risk, was not captured by one-time assessment of those measures. Alternatively, pathophysiolic processes, other than those measured by these biomarkers, may be in the causal pathways between smoking and HF.
Notably, there were racial differences in associations of smoking with NTproBNP. NTproBNP levels were lower in black than in white participants, and association of NTproBNP with smoking was only found in whites.
The present study is unique in that it provides novel information on the association of multiple smoking exposures with incident HF, including HFPEF and HFREF, in a large free-living cohort of older individuals, and explores linkages with cardiac strain on echocardiography and blood biomarkers of pathobiology relevant to HF.
While most studies have found that cigarette smoking poses a risk for HF, there variation in the magnitude of risk for HF (RR 1.00 to 2.35), and some have shown no association (25). There are sparse data on risks of current, former, as well as passive smoking for HF in older, community dwelling individuals. Moreover, few studies have examined associations of smoking with imaging and blood biomarkers of biologic pathways that are known to be linked to the clinical expression of HF, and determined their relevance to incident HF in smokers.
Our study results are consistent with those of other cohort studies (2,4,5) in finding associations of current smoking with HF, but differ from Health ABC (4) in finding an association of HF with former smoking, independent of pack-years of exposure, as well from MESA (5) where former smoking was not associated with HF.
There are few well-powered studies which have evaluated the influence of cigarette smoking on blood and imaging biomarkers of cardiac function and myocardial injury. Associations of smoking with elevations of troponin and NTproBNP over pre-specified values have been reported in ARIC (11), while failure to find association of smoking with troponin in MESA (29) may reflect use of an assay method with lower sensitivity.
In a recent UK Biobank study (30), MRI evaluation of 204 participants found lower LV systolic strain in current smokers than non-smokers. In the Jackson Heart Study (2), there was an association of smoking with lower LV circumferential strain on MRI, and with higher BNP levels. Our study extends those results in finding associations of smoking with LVLS as well as with LVEDSR assessed by archival speckle tracking echocardiography. Moreover, we found an association of smoking with blood biomarkers other than natriuretic peptides, i.e. hscTnT and sST2. However, while the Jackson Heart Study (2) found a significant association of BNP with smoking exposures in a study cohort comprised exclusively of black participants, we found no significant associations of smoking with NTproBNP in black participants. Reasons for this discrepancy are uncertain, but may represent lack of statistical power, and possibly the use of different natriuretic peptides. Specifically, the Jackson Heart Study evaluated 4129 black and by design no white participants. In contrast, the majority of the CHS cohort was white, with only 861 black participants.
Smoking exposures were limited to self-report, and baseline assessments of exposure were not updated throughout the 25 year follow-up period. Hence, the effects of changes in smoking exposure on outcomes which may have occurred during follow up were not considered. However, the use of exposure data at baseline answers a particular question, of interest to clinicians, about the risk of future HF based on what is known now. When smoking exposure status is updated, changes in smoking status are highly likely to reflect unmeasured changes in health status, and assumptions that these changes are not associated with other risk factors are likely to be incorrect.
Blood and imaging biomarkers were not obtained at multiple intervals throughout the duration of follow-up. Hence, single, or even two serial measurements, are likely “snapshots” which do may not represent total exposure to pathophysiologic components intended to be represented by these biomarkers.
The methods used for assessment of cardiac strain were designed to allow use of archived research videotapes obtained prior to contemporary digital echocardiography. Hence the values in this study are not intended to be used for clinical diagnosis, nor compared to studies done with different methodology. However, for a well powered population study as is the case here, the methods used for analysis of archived tapes have been well-validated for research purposes (22). Functional and structural measures of the right ventricle and atrium, or of LV stiffness, were not available, and hence their influence on the relationships between smoking and HF could not be assessed.
Frozen blood samples were thawed and analyzed for the biomarkers approximately 20 years after blood draw and freezing. While we cannot confirm stability of these substances over two decades, per the Arrhenius equation (31), minimal degradation would be expected to occur in samples frozen at −70°C to −80°C.
Analyses were not adjusted for multiple comparisons; hence, the likelihood of a type 1 error may have been increased. Nonetheless, the findings are consistent with our prespecified hypotheses that there are associations of smoking exposures with blood and imaging biomarkers reflecting mechanisms of cardiac disease, and with incident heart failure (31). While the present study evaluated imaging and blood biomarkers as representing subclinical outcomes of smoking exposures, it does not determine if these biomarkers mediated the association of smoking with incident heart failure.
In summary, even in older individuals, the risks of cigarette smoking for HF persist in current and former smokers. Smoking is associated with adverse changes in imaging and blood biomarkers of pathophysiology consistent with stage B HF. Race may affect associations of smoking with incident HF.
Supplementary Material
Acknowledgments
This research was supported by contracts HHSN268201200036C, HHSN268200800007C, HHSN268201800001C, N01HC55222, N01HC85079, N01HC85080, N01HC85081, N01HC85082, N01HC85083, N01HC85086, 75N92021D00006, and grants U01HL080295 and U01HL130114 from the National Heart, Lung, and Blood Institute (NHLBI), with additional contribution from the National Institute of Neurological Disorders and Stroke (NINDS). Additional support was provided by R01AG023629 from the National Institute on Aging (NIA). A full list of principal CHS investigators and institutions can be found at CHS-NHLBI.org.
Footnotes
The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health. None of the authors have relevant relationships with industry.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Declaration of interests
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.
References
- 1.Carter BD, Abnet CC, Feskanich D, Freedman ND, Hartge P, Lewis CE, Ockene JK, Prentice RL, Speizer FE, Thun MJ, Jacobs EJ. Smoking and mortality--beyond established causes. N Engl J Med 2015; 372(7):631–640. [DOI] [PubMed] [Google Scholar]
- 2.Kamimura D, Cain LR, Mentz RJ, White WB, Blaha MJ, DeFilippis AP, Fox ER, Rodriguez CJ, Keith RJ, Benjamin EJ, Butler J, Bhatnagar A, Robertson RM, Winniford MD, Correa A, Hall ME. Cigarette Smoking and Incident Heart Failure: Insights From the Jackson Heart Study. Circulation 2018;137(24):2572–2582. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Ho JE, Lyass A, Lee DS, Vasan RS, Kannel WB, Larson MG, Levy D. Predictors of new-onset heart failure: differences in preserved versus reduced ejection fraction. Circ Heart Fail 2013; 6(2):279–286. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Gopal DM, Kalogeropoulos AP, Georgiopoulou VV, Smith AL, Bauer DC, Newman AB, Kim L, Bibbins-Domingo K, Tindle H, Harris TB, Tang WW, Kritchevsky SB, Butler J. Cigarette smoking exposure and heart failure risk in older adults: the Health, Aging, and Body Composition Study. Am Heart J 2012;164(2):236–242. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Watson M, Dardari Z, Kianoush S, Hall ME, DeFilippis AP, Keith RJ,Benjamin EJ, Rodriguez CJ, Bhatnagar A, Lima JA, Butler J, Blaha MJ, Rifai MA. Relation Between Cigarette Smoking and Heart Failure (from the Multiethnic Study of Atherosclerosis). Am J Cardiol 2019;123(12):1972–1977. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Gulel O, Soylu K, Yazici M, Demircan S, Durna K, Sahin M. Longitudinal diastolic myocardial functions are affected by chronic smoking in young healthy people: a study of color tissue Doppler imaging. Echocardiography 2007;24(5):494–498. [DOI] [PubMed] [Google Scholar]
- 7.Farsalinos K, Tsiapras D, Kyrzopoulos S, Voudris V. Acute and chronic effects of smoking on myocardial function in healthy heavy smokers: a study of Doppler flow, Doppler tissue velocity, and two-dimensional speckle tracking echocardiography. Echocardiography 2013;30(3):285–292. [DOI] [PubMed] [Google Scholar]
- 8.Ahmed SS, Moschos CB, Lyons MM, Oldewurtel HA, Coumbis RJ, Regan TJ. Cardiovascular effects of long-term cigarette smoking and nicotine administration. Am J Cardiol 1976;37(1):33–40. [DOI] [PubMed] [Google Scholar]
- 9.McEvoy JW, Blaha MJ, DeFilippis AP, Lima JA, Bluemke DA, Hundley WG, Min JK, Shaw LJ, Lloyd-Jones DM , Barr RG, Budoff MJ, Blumenthal RS, Nasir K. Cigarette smoking and cardiovascular events: role of inflammation and subclinical atherosclerosis from the MultiEthnic Study of Atherosclerosis. Arterioscler Thromb Vasc Biol 2015;35(3):700–709. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Nadruz W Jr, Claggett B, Gonçalves A, Querejeta-Roca G, Fernandes-Silva MM, Shah AM, Cheng S, Tanaka H, Heiss G, Kitzman DW, Solomon SD. Smoking and Cardiac Structure and Function in the Elderly: The ARIC Study (Atherosclerosis Risk in Communities). Circ Cardiovasc Imaging 2016;9(9): e004950. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Nadruz W Jr, Gonçalves A, Claggett B, Querejeta Roca G, Shah AM, Cheng S, Heiss G, Ballantyne CM, Solomon SD. Influence of cigarette smoking on cardiac biomarkers: the Atherosclerosis Risk in Communities (ARIC) Study. Eur J Heart Fail 2016;18(6):629–637. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Glick D, deFilippi CR, Christenson R, Gottdiener JS, Seliger SL. Long-term trajectory of two unique cardiac biomarkers and subsequent left ventricular structural pathology and risk of incident heart failure in community-dwelling older adults at low baseline risk. JACC Heart Fail 2013;1(4):353–360. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.deFilippi CR, de Lemos JA, Christenson RH, Gottdiener JS, Kop WJ, Zhan M, Seliger SL. Association of serial measures of cardiac troponin T using a sensitive assay with incident heart failure and cardiovascular mortality in older adults. JAMA 2010;304(22):2494–2502. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.deFilippi CR, Christenson RH, Gottdiener JS, Kop WJ, Seliger SL. Dynamic cardiovascular risk assessment in elderly people. The role of repeated N-terminal pro-B-type natriuretic peptide testing. J Am Coll Cardiol 2010;55(5):441–450. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.de Boer RA, Nayor M, deFilippi CR, Enserro D, Bhambhani V, Kizer JR, Blaha MJ, Brouwers FP, Cushman M, Lima JAC, Bahrami H, van der Harst P, Wang TJ, Gansevoort RT, Fox CS, Gaggin HK, Kop WJ, Liu K, Vasan RS, Psaty BM, Lee DS, Hillege HL, Bartz TM, Benjamin EJ, Chan C, Allison M, Gardin JM, Januzzi JL Jr, Shah SJ, Levy D, Herrington DM, Larson MG, van Gilst WH, Gottdiener JS, Bertoni AG, Ho JE. Association of Cardiovascular Biomarkers With Incident Heart Failure With Preserved and Reduced Ejection Fraction. JAMA Cardiol 2018;3(3):215–224. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Gottdiener JS, Seliger S, deFilippi C, Christenson R, Baldridge AS, Kizer JR, Psaty BM, Shah SJ. Relation of Biomarkers of Cardiac Injury, Stress, and Fibrosis With Cardiac Mechanics in Patients ≥ 65 Years of Age. Am J Cardiol 2020;136:156–163. [DOI] [PubMed] [Google Scholar]
- 17.Fried LP, Borhani NO, Enright P, Furberg CD, Gardin JM, Kronmal RA, Kuller LH, Manolio TA, Mittelmark MB, Newman A, et al. The Cardiovascular Health Study: design and rationale. Ann Epidemiol 1991;1(3):263–276. [DOI] [PubMed] [Google Scholar]
- 18.Arnold AM, Psaty BM, Kuller LH, Burke GL, Manolio TA, Fried LP, Robbins JA, Kronmal RA. Incidence of cardiovascular disease in older Americans: the cardiovascular health study. J Am Geriatr Soc 2005;53(2):211–218. [DOI] [PubMed] [Google Scholar]
- 19.Psaty BM, Kuller LH, Bild D, Burke GL, Kittner SJ, Mittelmark M, Price TR, Rautaharju PM, Robbins J. Methods of assessing prevalent cardiovascular disease in the Cardiovascular Health Study. Ann Epidemiol 1995;5(4):270–277. [DOI] [PubMed] [Google Scholar]
- 20.Ives DG, Fitzpatrick AL, Bild DE, Psaty BM, Kuller LH, Crowley PM, Cruise RG, Theroux S. Surveillance and ascertainment of cardiovascular events. The Cardiovascular Health Study. Ann Epidemiol 1995;5(4):278–285. [DOI] [PubMed] [Google Scholar]
- 21.Gardin JM, Wong ND, Bommer W, Klopfenstein HS, Smith VE, Tabatznik B, Siscovick D, Lobodzinski S, Anton-Culver H, Manolio TA. Echocardiographic design of a multicenter investigation of free-living elderly subjects: the Cardiovascular Health Study. J Am Soc Echocardiogr 1992;5(1):63–72. [DOI] [PubMed] [Google Scholar]
- 22.Aguilar FG, Selvaraj S, Martinez EE, Katz DH, Beussink L, Kim KY, Ping J, Rasmussen-Torvik L, Goyal A, Sha J, Irvin MR, Arnett DK, Shah SJ. Archeological Echocardiography: Digitization and Speckle Tracking Analysis of Archival Echocardiograms in the HyperGEN Study. Echocardiography 2016;33(3):386–397. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Dieplinger B, Januzzi JL Jr., Steinmair M, Gabriel C, Poelz W, Haltmayer M, Mueller T. Analytical and clinical evaluation of a novel high-sensitivity assay for measurement of soluble ST2 in human plasma--the Presage ST2 assay. Clin Chim Acta 2009;409(1-2):33–40. [DOI] [PubMed] [Google Scholar]
- 24.Christenson RH, Duh SH, Wu AH, Smith A, Abel G, deFilippi CR, Wang S, Adourian A, Adiletto C, Gardiner P. Multi-center determination of galectin-3 assay performance characteristics: Anatomy of a novel assay for use in heart failure. Clin Biochem 2010;43(7-8):683–690. [DOI] [PubMed] [Google Scholar]
- 25.Cushman M, Cornell ES, Howard PR, et al. Laboratory methods and quality assurance in the Cardiovascular Health Study. Clin Chem 1995;41:264–270. [PubMed] [Google Scholar]
- 26.Macy EM, Hayes TE, Tracy RP. Variability in the measurement of C-reactive protein in healthy subjects: implications for reference intervals and epidemiological applications. Clin Chem 1997;43:52–58. [PubMed] [Google Scholar]
- 27.Jenny NS, Tracy RP, Ogg MS, et al. In the elderly, interleukin-6 plasma levels and the - 174G>C polymorphism are associated with the development of cardiovascular disease. Arterioscler Thromb Vasc Biol 2002;22:2066–2071. [DOI] [PubMed] [Google Scholar]
- 28.Aune D, Schlesinger S, Norat T, Riboli E. Tobacco smoking and the risk of heart failure: A systematic review and meta-analysis of prospective studies. Eur J Prev Cardiol 2019;26(3):279–288. [DOI] [PubMed] [Google Scholar]
- 29.Al Rifai M, DeFilippis AP, McEvoy JW, Hall ME, Acien AN, Jones MR, Keith R, Magid HS, Rodriguez CJ, Barr GR, Benjamin EJ, Robertson RM, Bhatnagar A, Blaha MJ. The relationship between smoking intensity and subclinical cardiovascular injury: The Multi-Ethnic Study of Atherosclerosis (MESA). Atherosclerosis 2017;258:119–130. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.Hendriks T, van Dijk R, Alsabaan NA, van der Harst P. Active Tobacco Smoking Impairs Cardiac Systolic Function. Sci Rep 2020;10(1):6608. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31.Jerne NK, Perry WL. The stability of biological standards. Bull World Health Organ 1956;14(1):167–182. [PMC free article] [PubMed] [Google Scholar]
- 32.Rothman KJ. No adjustments are needed for multiple comparisons. Epidemiology 1990;1(1):43–46. [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.