Sex-Difference of Associations Between Cigarette Smoking and Myocardial Fibrosis: The Multi-Ethnic Study of Atherosclerosis.
Feb 26, 2026DOI: 10.1161/jaha.125.044664Journal ArticleAHA
Written by: Elie W AklRalph ZeitounOmar ChehabHaiou LiVinithra VaradarajanColin O WuAlain G BertoniKarol E WatsonDavid A BluemkeBharath Ambale-VenkateshJoão A C Lima
The data that support the findings of this study are available from the corresponding author upon reasonable request. The MESA (Multi‐Ethnic Study of Atherosclerosis) is a prospective, observational cohort study that included 6814 men and women with no prior history of CVD from 4 racial and ethnic groups (White, Black, Hispanic/Latino, Chinese) and recruited from 6 states over 2 years between 2000 and 2002. A detailed study design has been described previously.11 The institutional review boards approved the study protocol of participating institutions, and each participant signed informed consent before recruitment.11 Baseline examination (exam 1) was conducted between 2000 and 2002 and then participants were followed up and data were collected over 4 subsequent follow‐up visits (exam 2 [2002–2004], exam 3 [2004–2005], exam 4 [2004–2005], and exam 5 [2010–2012]). A complete‐case analysis was performed; participants with missing data for the primary exposure (smoking) or outcomes (CMR markers) were excluded from the final analytical cohort. MESA participants who have smoking data available at exam 1 and exam 5 and have received a cardiac magnetic resonance imaging (MRI) at exam 5 with available native T1 time (ms) or ECV (%) were included. Exam 5 was selected as the analytic visit because native T1 and ECV measures were obtained exclusively during this exam. Those with MI or congestive heart failure by exam 5 were excluded (Figure S1).Cigarette smoking data in MESA were collected through a questionnaire administered at each exam. Smoking status was assessed through the following questions: (1) “Have you smoked at least 100 cigarettes in your lifetime?,” (2) “Have you smoked during the past 30 days?” If the participant answered No to the first question, then they are classified as “never smoker.” If they answered Yes to the first question, then based on the second question they are classified as either “current smoker” if they answered Yes or “former smoker” if they answered No. Smoking intensity was assessed based on the following question “On average, about how many cigarettes a day do/did you smoke?”For each participant with at least 3 available data measurements regarding the average number of cigarettes smoked per day at exam 1, exam 5, and any of the remaining exams (2–4), the cumulative smoking exposure between exam 1 and exam 5 was calculated. To explore the cumulative exposures and time trends of smoking, we considered the age‐based time‐trend analysis described in Wu and Tian.12 Because the data on the number of cigarettes per day are right skewed and zero inflated, we calculated subject‐specific trajectories of smoking exposure using the linear interpolation introduced in Wu and Tian.12 The area under the subject‐specific trajectories for the i‐th participant's cigarettes per day for the past 10 years is calculated by AUCAi5−10,Ai5=∫Ai5−10Ai5X^itdt, where Ai5 is the age of i‐th participant at exam 5. The data were analyzed using R version 4.2.3.
To measure temporal changes in smoking exposure, we categorized participants based on their smoking status at exam 1 and exam 5 into 4 categories: never smokers at exams 1 and 5 are labeled as “never smoker”; former smokers at exams 1 and 5 are labeled “former smoker for >10 years”; former smokers at exam 5 and never or current smokers at exam 1 are labeled “former smoker for <10 years”; and current smokers at exams 1 and 5 are labeled as “smoker for >10 years.”
The cardiac MRI protocol was described previously.13 In brief, CMR was performed using 1.5 T scanners (Avanto and Espree, Siemens Medical Systems; Signa LX, GE Healthcare) with a 6‐channel anterior phased‐array torso coil and posterior coil elements. Interstitial myocardial fibrosis was evaluated using the single breath‐hold modified Look‐Locker inversion recovery sequence. Pre‐ and postcontrast injection mid‐left ventricular short‐axis T1 maps were obtained. QMass (Research version, Leiden, Netherlands) was used to analyze the images. A region of interest was manually delineated around the core myocardium on each T1 map (pre‐ and postcontrast) to determine the myocardial T1 time. If a myocardial scar was present, it was included in the region of interest. Greater diffuse myocardial fibrosis has been linked to lower postcontrast T1 times. The partition coefficient was derived by plotting the myocardium's 1/T1 timings against the blood pool and measuring the slope of the resulting linear regression line. ECV=partition coefficient×(1–hematocrit/100) was used to compute the ECV. Considering the limited availability of hematocrit laboratory values, we used a synthetic ECV based on the blood's longitudinal relaxation rate. This has been shown to have a high correlation to conventional ECV and CVD outcomes.14
Clinical characteristics and demographics from the evaluation during exam 5 were used for analysis. Categorical variables are summarized as counts and relative frequencies, and continuous variables were summarized as mean±SD. Differences between men and women in baseline characteristics were evaluated using standardized mean differences. Linear regression was used to evaluate the association between native T1 and ECV and each smoking status at exam 5, reported average number of cigarettes smoked per day at exam 5, 10‐year cumulative smoking exposure between exam 1 and exam 5, and smoking habit changes between exam 1 and exam 5. Model 1 covariates include age, sex, and race. Model 2 includes Model 1 covariates in addition to highest level of education, body mass index, systolic blood pressure, use of antihypertensive medications, low‐density lipoprotein level, use of lipid‐lowering medications, diabetes status, and estimated glomerular filtration rate. Model 3 includes Model 2 covariates in addition to mean Agatston coronary artery calcium score and the presence of late gadolinium enhancement by cardiac MRI. These confounders are known to influence the development of myocardial fibrosis and affect parametric mapping as previously described in multiple studies.15, 16 All analyses were first conducted with sex included as an independent covariate. When significant interaction was identified, sex‐stratified models were constructed to display sex‐specific associations. Statistical significance was defined as P<0.05. Interaction with sex was tested for all analyses, and subsequent stratification by sex was performed when interaction was found to be significant. Unstandardized beta coefficients are presented. Statistical analyses were performed using SPSS Statistics version 28.
RESULTS
MESA recruited 6814 participants at baseline. Among these participants, 410 with MI or congestive heart failure or with no follow‐up data on MI and congestive heart failure by exam 5 were excluded. We further excluded 4286 participants who did not have an adequate cardiac MRI for T1 mapping. This resulted in a final analytical cohort of 2118 participants (53% women, mean age 68.6±9 years) who had cigarette smoking data and had undergone CMR analysis for myocardial fibrosis. Of those, 1309 participants had available ECV data, and 2118 participants had available native T1 data. The mean native T1 time was 971±46 ms and the mean ECV was 26.8±2.8%. Participants with a history of MI or congestive heart failure by the time of CMR acquisition (at exam 5) were excluded from the analysis. The baseline characteristics of the study population stratified by sex are provided in Table 1.
Demographics of Participants at Exam 5
Continuous variables are expressed as mean±SD and categorical variables as n (%). Positive values indicate higher means or proportions in men, negative values higher in women. The absolute SMD quantifies the magnitude of difference: ≈0.1 small, ≈0.3 moderate, and ≥0.5 large.
BP indicates blood pressure; and SMD, standardized mean difference.
Tables 2 and 3 show that current smoking status at exam 5 was significantly associated with higher ECV and native T1 time, respectively, independent of demographic characteristics and traditional cardiovascular risk factors. There was a significant interaction between sex and current smoking status for ECV and with both former and current smoking status for native T1 time. After stratifying by sex, current smoking status was independently associated with a 2% increase in ECV% and a ~30 ms increase in native T1 time in women but not in men in the unadjusted and fully adjusted analysis (P<0.001 for both). There was no significant association between former smoking status and ECV or native T1 time in male or female participants.
Multivariable Linear Regression Models for Association and Sex Interaction of Cigarette Smoking Status With ECV (%) at Exam 5
Model 1: adjusted for age, sex, race, highest level of education Sex is not considered as a covariate after stratification. Model 2: adjusted for Model 1 covariates + body mass index, systolic blood pressure, use of antihypertensive medications, low‐density lipoprotein level, high‐density lipoprotein level, use of lipid‐lowering medications, fasting glucose level, diabetes status, glomerular filtration rate. Model 3: adjusted for Model 2 covariates + mean Agatson coronary artery calcium score + presence of late gadolinium enhancement.
ECV indicates extracellular volume.
Significant interaction with sex; P=0.029.
Multivariable Linear Regression Models for Association and Sex Interaction of Cigarette Smoking Status With Native T1 Time (ms) at Exam 5
Model 1: adjusted for age, sex, race, highest level of education Sex is not considered as a covariate after stratification. Model 2: adjusted for Model 1 covariates + body mass index, systolic blood pressure, use of antihypertensive medications, low‐density lipoprotein level, high‐density lipoprotein level, use of lipid‐lowering medications, fasting glucose level, diabetes status, glomerular filtration rate. Model 3: adjusted for Model 2 covariates + mean Agatson coronary artery calcium score+presence of late gadolinium enhancement.
Significant interaction with sex.† P=0.009.‡ P=0.006.
Tables 4 and 5 show that a higher reported average number of cigarettes per day at exam 5 was significantly associated with higher ECV (β=0.1; P<0.001) and native T1 time (β=1; P=0.021), respectively, independent of demographic characteristics and traditional cardiovascular risk factors. Higher calculated mean number of cigarettes per day was also significantly associated with higher ECV (β=0.1; P<0.001) and native T1 time (β=0.5; P=0.022). There was a significant interaction between sex and both the reported average number of cigarettes per day at exam 5 and the calculated mean number of cigarettes per day for ECV; however, the interaction with sex was significant only with the calculated mean number of cigarettes per day for native T1 time. When stratified by sex, both the reported average number of cigarettes per day at exam 5 (β=0.1; P<0.001) and the calculated mean number of cigarettes per day (β=0.1; P<0.001) were significantly associated with higher ECV, and the calculated mean number of cigarettes per day was associated with higher native T1 (β=1; P=0.006) in women but not in men (Figure 1). This model predicts that female participants smoking 1 pack (20 cigarettes) per day would have an ECV ~2% points higher than a nonsmoking woman, an effect magnitude similar to our finding for “current smoker” as a category.
Multivariable Linear Regression Models for Association and Sex Interaction of Average Number of Cigarettes Smoked Per Day (at Exam 5, and Calculated Average Between Exams 1 and 5) and ECV (%)
Model 1: adjusted for age, sex, race, highest level of education Sex is not considered as a covariate after stratification. Model 2: adjusted for Model 1 covariates + body mass index, systolic blood pressure, use of antihypertensive medications, low‐density lipoprotein level, high‐density lipoprotein level, use of lipid‐lowering medications, fasting glucose level, diabetes status, glomerular filtration rate. Model 3: adjusted for Model 2 covariates+mean Agatson coronary artery calcium score+presence of late gadolinium.
ECV indicates extracellular volume.
Significant interaction with sex.† P=0.016.‡ P<0.001.
Multivariable Linear Regression Models for Association and Sex Interaction of Average Number of Cigarettes Smoked Per Day (at Exam 5, and Calculated Average Between Exams 1 and 5) and Native T1 Time (ms)
Model 1: adjusted for age, sex, race, highest level of education Sex is not considered as a covariate after stratification. Model 2: adjusted for Model 1 covariates+body mass index, systolic blood pressure, use of antihypertensive medications, low‐density lipoprotein level, high‐density lipoprotein level, use of lipid‐lowering medications, fasting glucose level, diabetes status, glomerular filtration rate. Model 3: adjusted for Model 2 covariates+mean Agatson coronary artery calcium score+presence of late gadolinium enhancement.
Significant interaction with sex; P=0.007.
Partial regression plots demonstrating sex‐stratified association between the calculated average number of daily cigarettes and markers of interstitial myocardial fibrosis by cardiac magnetic resonance imaging after controlling for demographic characteristics and traditional cardiac risk factor. ECV indicates extracellular volume.
As shown in Tables 6 and 7, being a smoker for >10 years was associated with a 2.3% increase in ECV % (P<0.001) and 29 ms increase in native T1 time (P<0.001), in women but not in men. However, smoking cessation for >10 years was not associated with increase in ECV or native T1 time. Moreover, smoking cessation for <10 years was significantly associated with a 0.7% increase in ECV (P=0.012) but no change in native T1 time and the former association is lost when stratified by sex.
Multivariable Linear Regression Models for Association and Sex Interaction of Change in Cigarette Smoking Status Between Exams 1 and 5 and ECV (%)
Model 1: adjusted for age, sex, race, highest level of education Sex is not considered as a covariate after stratification. Model 2: adjusted for Model 1 covariates+body mass index, systolic blood pressure, use of antihypertensive medications, low‐density lipoprotein level, high‐density lipoprotein level, use of lipid‐lowering medications, fasting glucose level, diabetes status, glomerular filtration rate. Model 3: adjusted for Model 2 covariates+mean Agatson coronary artery calcium score+presence of late gadolinium enhancement.
ECV indicates extracellular volume.
Significant interaction with sex; P=0.031.
Multivariable Linear Regression Models for Association and Sex Interaction of Change in Cigarette Smoking Status Between Exams 1 and 5 and Native T1 (ms)
Model 1: adjusted for age, sex, race, highest level of education Sex is not considered as a covariate after stratification. Model 2: adjusted for Model 1 covariates + body mass index, systolic blood pressure, use of antihypertensive medications, low‐density lipoprotein level, high‐density lipoprotein level, use of lipid‐lowering medications, fasting glucose level, diabetes status, glomerular filtration rate. Model 3: adjusted for Model 2 covariates+mean Agatson coronary artery calcium score+presence of late gadolinium enhancement.
Significant interaction with sex.
P=0.023.‡ P=0.012.
DISCUSSION
In this study, we found that current smoking status in women but not in men is associated with higher markers of myocardial fibrosis by CMR in a population with no prior CVD events. In addition, our analysis revealed a significant dose‐dependent effect between smoking burden at the time of the cardiac MRI and cumulative smoking exposure over a 10‐year period before CMR acquisition and the prevalence of myocardial fibrosis in female participants. Importantly also, we found that the association between smoking and myocardial fibrosis was lost for participants who stopped smoking, particularly if cessation occurred >10 years before fibrosis assessment. These findings align with emerging evidence of sex‐specific cardiovascular responses to smoking and underscores the utility of advanced imaging in detecting subclinical myocardial changes.
Few studies have investigated the relationship between smoking and interstitial myocardial fibrosis. In a cross‐sectional study of 381 patients without known CVD, current smoking status was associated with elevated ECV when compared with former and nonsmokers (0.25% [95% CI, 0.24%–0.26%] versus 0.24% [95% CI, 0.24%–0.26%], P<0.01).17 Another study of 239 patients with type 2 diabetes showed that active smoking was associated with a 2% increase in ECV when compared with nonsmokers.18 Our results corroborate these findings, as we demonstrated that current smoking status was associated with greater ECV and longer native T1 time. However, when stratified by sex we found that this association was mainly driven by a ~2% ECV increase among female participants who were current smokers when compared with nonsmoker MESA participants.
Several studies have established a significant association between smoking intensity and increased risk of CVD mortality.19, 20, 21, 22 Using the MESA cohort, Nance et al. showed that the average intensity of smoking at baseline is associated with worse CVD outcomes in current smokers.23 Furthermore, research has shown that even relatively low levels of fine particulate exposure from cigarette smoke are sufficient to induce adverse biological responses and increase the risk of CVD mortality. The exposure–response relationship between CVD mortality and fine particulate matter is relatively steep at low levels of exposure and flattens out at higher exposures.24 Our results showed similar dose‐dependent relationship between smoking exposure and myocardial fibrosis by CMR, however, without plateauing at higher exposure levels. Increasing smoking intensity reflected by the reported number of cigarettes smoked per day was associated with higher markers of fibrosis. Moreover, higher cumulative exposure of smoking over a 10‐year period before CMR acquisition was associated with higher fibrosis markers.
Smoking cessation is widely recommended to reduce CVD burden. However, the timeline for the reduction of cardiovascular risk after quitting smoking in comparison to never smokers remain undefined. The Atherosclerotic CVD Risk Estimator Plus states that the CVD risk of former smokers becomes similar to those who never smoked after 5 years.25 Duncan et al. showed that among heavy smokers (≥20 pack‐years), those who successfully quit smoking exhibited a significantly reduced risk of CVD within the first 5 years following cessation, in comparison to their counterparts who continued smoking (hazard ratio, 0.61). In addition, it took ~10 to 15 years following smoking cessation for former heavy smokers to no longer exhibit a statistically significant increase in CVD risk when compared with individuals who had never smoked.26 Moreover, in a recent study, Ding et al. showed that smoking cessation was significantly associated with lower risk of heart failure; however the residual risk persisted for a few decades after smoking cessation.27 Our findings support these results, as we showed that the association between smoking and interstitial fibrosis by CMR was no longer significant in participants who stopped smoking for >10 years before fibrosis assessment.
Several studies have consistently demonstrated that at baseline, women exhibit higher native T1 values and ECV than men, suggesting inherent physiological differences in myocardial structure and function between the sexes. One study found that women have higher myocardial perfusion, blood volume, and ECV compared with men, both at rest and during stress conditions, indicating that sex is an independent contributor to these parameters beyond other physiological factors.28 Another study highlighted that female sex is associated with higher native T1 values, even after adjusting for variables such as age, heart rate, and left ventricular mass index, further supporting the notion of intrinsic sex‐based disparities in myocardial tissue composition.29 Previous studies have assessed the sex‐differentiated effect of smoking on the development of CVD. The Copenhagen Heart Study found that smoking was associated with a higher risk for all acute coronary syndromes in women when compared with men.30 Similarly, it has been shown that female smokers are more likely to develop ST‐segment–elevation MI when compared with male smokers aged <65 years.31, 32 In our study, we show that smoking‐induced interstitial myocardial fibrosis is more likely to be present among women when compared with men.
The mechanisms contributing to a greater risk of CVD in female smokers compared with their male counterparts are not well understood. However, there appear to be significant sex‐related differences with regard to susceptibility to heart disease.33 In a recent study, Chehab et al. showed that sex hormone imbalances were associated with higher markers of fibrosis by CMR in men. Moreover, higher levels of total testosterone was inversely associated with interstitial myocardial fibrosis in postmenopausal women after controlling for smoking status at baseline.34 In postmenopausal women, the loss of estrogen and a shift toward android fat distribution further potentiate smoking's fibrotic effects, a mechanism less relevant in men, who lack this hormonal axis.35 In addition, smoking has been shown to differentially increase proinflammatory cytokines without enhancement in an inflammatory response in women compared with men36 which is known to be associated with increase interstitial myocardial fibrosis.37 It is worth noting that given that the vast majority of women in this study were postmenopausal, these findings should be interpreted as reflecting the postmenopausal female population rather than all women.
Previous studies have suggested, that women are more predisposed to concentric cardiac remodeling compared with men, who are more predisposed to eccentric cardiac remodeling.38 Although premenopausal women are much less prone to develop CVD than men of similar age, this advantage no longer applies after menopause.39 The mechanisms contributing to the greater risk of concentric cardiac remodeling in women remain incompletely understood, but there appears to be important sexual dimorphisms in ventricular‐vascular structure and function that develop with aging which may predispose to concentric remodeling in women.40 In this regard, it has been suggested that increases in the activities of proteolytic enzymes such as matrix metalloproteinases, calpains, cathepsins, and caspases contribute to the process of cardiac remodeling.41
Despite being an established risk factor for CVD, cigarette smoking interacts with known risk factors like physical inactivity and obesity, as well as several other cardiovascular pathologic processes to greatly increase the risk for worse cardiovascular outcomes. In our study, cigarette smoking is an important marker of myocardial fibrosis among women which remained in fully adjusted model, when compared with male smokers.
Finally, although the absolute magnitude of the association between cigarette smoking and myocardial fibrosis, as measured by ECV and native T1 time, appears to be modest (eg, a 0.1% increase per cigarette smoked per day or ~2% per pack per day), the critical finding lies in the marked sex‐based difference in this relationship. Our results demonstrate that smoking is significantly associated with increased myocardial fibrosis markers in women, but not in men, underscoring a potential sex‐specific vulnerability. This divergence, further illustrated in the Figure 1 through the differential trends in ECV and native T1, highlights the underlying differences in the pathophysiology of CVD between sexes. These findings support the growing body of evidence suggesting that women may have heightened myocardial sensitivity to external stressors such as smoking, which may partly explain their distinct clinical presentation and progression of CVD.
LIMITATIONS
Given the cross‐sectional study design, a causal relationship between smoking and myocardial fibrosis cannot be established. The assessment of smoking status relied on self‐reported data, which are subject to recall bias and social desirability bias. Furthermore, the linear interpolation method used in this article assumes smooth changes in smoking intensity between exams, which may not fully reflect actual smoking behavior patterns. Despite adjustment for multiple potential confounders, residual confounding cannot be excluded, particularly from variables such as menopausal status and health behaviors (eg, physical activity) that were not uniformly available at exam 5.
More than 95% of the female participants in this study were postmenopausal; thus, the observed associations between smoking and myocardial fibrosis in women primarily apply to postmenopausal women and may not generalize to premenopausal populations. In addition, native T1 and ECV may be influenced by other myocardial pathologies. We could not control for viral infections, which are known to increase interstitial myocardial fibrosis, as such data were available only at exam 1. Similarly, C‐reactive protein data were available only at baseline. Information on noncombustible tobacco use, such as vaping or smokeless tobacco, was not collected at the time of data acquisition, and therefore the present findings cannot be generalized to these exposures.
CONCLUSIONS
In this multiethnic, population‐based study, cigarette smoking is associated with higher CMR measures of interstitial myocardial fibrosis (reflected by higher ECV and longer native T1 time), in women but not in men. This indicates differences in the pathophysiology of CVDs by sex. However, additional studies are required to further understand such important relationships.
Sources of Funding
This research was supported by contracts 75N92020D00001, HHSN268201500003I, N01‐HC‐95159, 75N92020D00005, N01‐HC‐95160, 75N92020D00002, N01‐HC‐95161, 75N92020D00003, N01‐HC‐95162, 75N92020D00006, N01‐HC‐95163, 75N92020D00004, N01‐HC‐95164, 75N92020D00007, N01‐HC‐95165, N01‐HC‐95166, N01‐HC‐95167, N01‐HC‐95168 and N01‐HC‐95169 from the National Heart, Lung, and Blood Institute, and by grants UL1‐TR‐000040, UL1‐TR‐001079, and UL1‐TR‐001420 from the National Center for Advancing Translational Sciences.
Disclosures
None.
Supporting information
Figure S1
VV
Dr. Vinithra Varadarajan
Internal Medicine · Hospital MedicineAnne Arundel Medical Centervinithra