Association of smoking and physical inactivity with MRI derived changes in cardiac function and structure in cardiovascular healthy subjects

We aimed to investigate the association of smoking and physical exercise on ventricular function and structure, determined by cardiac magnetic resonance imaging (CMR), in subjects without known cardiovascular diseases. A total of 381 participants (median age 57 years) of the Cooperative Health Research in the Region of Augsburg (KORA) FF4 cohort underwent CMR. The participants’ smoking and sporting habits were measured by a questionnaire. Physical inactivity was associated with a reduction of left ventricular ejection fraction (LV-EF), stroke volume, early diastolic peak filling rate and peak ejection rate of the left ventricle as well as right ventricular stroke volume. LV-EF was reduced in subjects with almost no physical activity compared to subjects with regular physical activity (68.4%, 95%CI 66.8–70.1% vs. 70.8%, 95%CI 69.2–72.3%, p < 0,05). Smokers had lower right ventricular end-diastolic volumes (80.6 ml/m², 95%CI 76.7–84.5 ml/m²; never-smokers: 85.5 ml/m², 95%CI 82.6–88.3 ml/m²; p < 0.05) but higher extracellular volume fractions (ECV) and fibrosis volumes (34.3 ml, 95%CI 32.5–36.0 ml, vs. 31.0 ml, 95%CI 29.6–32.3 ml, p < 0.01). We conclude that asymptomatic individuals without known cardiovascular diseases show differences in cardiac function and structure depending on their physical activity and smoking habits. This underlines the importance of prevention and health education.

Association of smoking status with cardiac function and structure. There were no significant associations between smoking status and left ventricular function (Table 5). However, there was a tendency of increased LV mass in current smokers (145.2 ml 95%-CI 139.8-150.7 ml) compared to never-smokers (139.6 ml, 95%-CI 135.5-143.6 ml) (p = 0.102).

Discussion
In this cross-sectional study, we examined the association of physical inactivity and smoking habits on cardiac function and structure as assessed by CMR in subjects without known cardiovascular disease. Parameters for left and right ventricular function were within normal limits for the vast majority of participants. However, we found that physical inactivity is associated with a reduction of LV and RV systolic and LV diastolic function. Furthermore, smoking was associated with a reduction of RV-EDV and with elevated ECV and fibrosis volumes.
These findings are in line with results of the Multi-Ethnic Study of Atherosclerosis (MESA) 16 . In MESA as well as in our study, LV-SV increased with exercise. In contrast to MESA, we have also found a positive effect on LV-EF. Potentially these results could be explained by differences in the assessment of physical activity or differences in the ethnicities of the study (the KORA population being completely Caucasian). Furthermore, we used a 3-Tesla MRI system, which offers an increased signal-to-noise-ratio and contrast-to-noise-ratio compared to 1.5-Tesla 22 . www.nature.com/scientificreports www.nature.com/scientificreports/ Levy et al. showed that older and untrained subjects had a lower PFR1 23 . This correlates with our results, where subjects with the highest physical activity per week had the highest PFR1. In our study subjects with more than 2 hours of physical activity per week had the highest PER, as was already demonstrated by Stratton et al. 24 , where untrained men had lower PER.
Our study shows that physical activity has a significant positive effect on RV-SV. Regular physical activity also positively influenced RV-EDV and RV-EF, but these results were not significant. This is in line with the results of Aaron et al. in the MESA trial 15 . The higher RV-SV could be caused by hypertrophy of cardiomyocytes due to physical activity, causing higher left and right SV 25 .
In former studies, subjects with higher physical activity rates showed no difference in cardiac fibrosis volumes 26 , but more LGE 27 , while in our study physical activity had no significant effect on fibrosis volumes, ECV or LGE. As physical activity influences cardiac function and hypertrophy, fibrosis and ECV could also be affected, whereas presence of LGE should not differ between subjects with more or less physical activity. These partly differing results need more research in the future.
We observed a trend towards increased left ventricular mass in smokers, a result that is in line with MESA and other studies 16,28 . In contrast, the LARGE study found a smaller left ventricular mass in smokers, maybe due to the special cohort of the LARGE study, which comprised only healthy young men 29 .
Analogous to the MESA study, RV-EDV was lower in smokers [30][31][32] . Cigarette smoke includes several toxic substances that cause aging of cardiomyocytes and thus reduced compliance of the right ventricle 33 . Smoking can lead to chronic obstructive pulmonary disease, which causes hyperinflation of the lungs 32 . Both mechanisms can thus reduce RV filling.
We were able to demonstrate that current smokers had significantly higher ECV and fibrosis volumes compared to non-smokers, which can be explained by oxidative stress and inflammation causing cardiac remodeling in smokers 34 . In MESA, smokers also had more LGE 35 , whereas in our study smokers had less LGE, but this result was not significant. It is known that smokers are more vulnerable concerning coronary heart disease so it could be expected that smokers and ex-smokers show more LGE than never-smokers. This, too, needs further research with larger cohorts.
Limitations and strengths of the study. As this is a cross-sectional study, we can only show correlations; it is not possible to draw conclusions about causations. As opposed to the LARGE study (young white males) or the MESA study (4 racial/ethnic groups) our subjects were aged 38-72, caucassian, and were approximately 50%  www.nature.com/scientificreports www.nature.com/scientificreports/ females. As a result, our cohort is better suited for the population in Germany than the LARGE or MESA cohorts. Our results are in general in line with the results of other large cohort studies examining subclinical cardiovascular changes depending on risk factors.
Smoking status was assessed via questionnaire, so there was no biochemical confirmation of smoking status. Whereas subjects with known cardiovascular diseases were excluded from the KORA study, participants with known pulmonary diseases that could affect the right ventricle, were not excluded. Another limitation is that the subjects' amount of physical activity was based on an interview and not validated via ergometry, resting heart rate or electronic activity trackers like wristbands.
The differences between the examined groups were small and most probably have no clinical significance but may represent early manifestations of potentially progressive disease.
Most studies about the impact of cardiac risk factors used echocardiography, for example the Framingham Study. Echocardiography is a widely available method, it is fast, cost-effective and does not use radiation. Unfortunately, echocardiography is more dependent on the patient's constitution, depending on the patient's acoustic windows, and has quite a high inter-observer variability 36 . Especially when measuring LV-EF using the Simpson method, echocardiography has its limitations because frequently the endocardial contours are often not clearly visible when not using contrast agents 17 . Whereas cardiac magnetic resonance imaging (CMR) has lower inter-observer variabilities, is less dependent on acoustic windows and shows better contrast between    www.nature.com/scientificreports www.nature.com/scientificreports/ endocardium and blood in order to properly measure endocardial contours to measure left ventricular volumes and calculate left ventricular ejection fraction 18,37 . For left ventricular hypertrophy, CMR has higher sensitivity rates than echocardiography 38 . Furthermore, it was shown that CMR has higher reproducibility rates in right ventricular parameters 39,40 . CMR can therefore be called the standard of reference for detecting the smallest, but still significant changes in cardiac function and mass parameters 19 . conclusions Using CMR, our study showed that in subjects without a history of cardiovascular diseases, parameters of left and right ventricular function increase with physical activity and decrease with smoking. In addition, we were found increases in ECV and fibrosis volumes in smokers. These results underline once more the importance of prevention and health education in order to prevent subclinical and in hereafter symptomatic cardiovascular changes. The current study population was drawn from a sub-study that aimed at examining MRI-derived subclinical cardiovascular disease burden in diabetics, pre-diabetics and controls without a history of cardiovascular disease 42 . For this study, subjects were recruited from the KORA FF4 follow-up. Exclusion criteria were: age >72 years, subjects with validated/self-reported stroke, myocardial infarction or revascularization, subjects with a cardiac pacemaker or implantable defibrillator, cerebral aneurysm clip, neural stimulator, any type of ear implant, ocular foreign body (e.g. metal shavings), any implanted device (e.g. insulin pump, drug infusion device), pregnant or breast feeding female subjects, claustrophobia, known allergy against gadolinium compounds and serum creatinine ≥1.3 mg/dl. The inclusion criteria were: willingness to undergo whole-body MRI and qualification in the diabetics, prediabetics or control group according to the World Health Organization criteria 43 . In total, 400 subjects aged 38-72 years were included in this study. The study was approved by the institutional review board of the Ludwig-Maximilians-University, Munich, and all participants provided written informed consent 42 . All methods were performed in accordance with the relevant guidelines and regulations.

Methods
MRi examination and evaluation. All images were obtained with the same 3-Tesla MR system (Magnetom Skyra, Siemens Healthcare, Erlangen Germany) in the Institute of Clinical Radiology of the Ludwig-Maximilian University Hospital in Munich. Imaging was performed using an 18-channel body coil in combination with the table-mounted spine matrix coil. The MRI examination was performed within three months after the FF4 visit at the study center. The protocol comprised imaging of the neurocranium, the carotid vessels, liver and whole-body fat imaging as well as cardiac imaging. Total acquisition time was approximately one hour. The full protocol is shown in Supplementary Table S1. cardiac function analysis. Left ventricular function was assessed using cine-steady state free precession sequences which were acquired in four chamber view and short axis stack with 10 slices and 25 phases using the following parameters: slice thickness 8 mm, in-plane voxel size 1.5 × 1.5 mm², field of view 297 × 360 mm², matrix 240 × 160, repetition time 29.97 ms, echo time 1.46 ms, flip angle 63°.
Evaluation of cardiac function parameters was performed using a dedicated software package (cvi42, Version 4. 1.5(190), Circle Cardiovascular Imaging Inc., Calgary, Canada). The left ventricular short-axis and four chamber view images were analyzed semi-automatically. The reading was performed according to standardized post-processing guidelines of the Society for Cardiovascular Magnetic Resonance 44 by two readers blinded to any information regarding the subjects' risk factors and clinical status. After automatic contouring of the endo-and epicardial contours, the outlines were corrected manually, if necessary. The papillary muscles were excluded from the myocardial mass and included in the ventricular volumes. The position of end-diastolic and end-systolic phase were chosen automatically by the software. The derived parameters were: end-systolic volume (LV-ESV), end-diastolic volume (LV-EDV), stroke volume (LV-SV), ejection fraction (LV-EF), systolic and diastolic myocardial mass, cardiac output, and left ventricular wall thickness. An example of the LV contouring is given in Fig. 2. Furthermore, left ventricular filling and ejection rates were determined based on the results of the left ventricular  www.nature.com/scientificreports www.nature.com/scientificreports/ volumetry using an in-house software (pyHeart). Measured parameters were: the peak ejection rate (PER), the peak filling rate of the early diastole (PFR1), the peak filling rate of the late diastole (PFR2) and the delay time between early and late diastolic fillings.
Right ventricular volumes and function were analyzed by manual tracking of the endocardial contour on short-axis and manual detection of the apex and tricuspid valve on four chamber view images according to current guidelines 44 . The largest and smallest right ventricular volume were defined as end-systolic and end-diastolic phase. Papillary muscles were included in the ventricular volume. Parameters included: end-diastolic (RV-EDV) and end-systolic volume (RV-ESV), stroke volume (RV-SV), cardiac output, ejection fraction (RV-EF).
Late gadolinium enhancement (LGE) sequences were acquired 10 minutes after administration of gadopentate dimeglumine (0,2 mmol/kg, Gadovist, Bayer Healthcare, Berlin, Germany) as Fast Low Angle Shot (FLASH) inversion recovery sequences with slice thickness 8 mm, FOV 300 × 360 mm, Matrix 256 × 140, TR 700-1000 ms, TE 1.55 ms and FA 20-55°. Analysis of late gadolinium enhancement (LGE) was performed by two experienced readers using the same software also used for the cardiac function analysis (cvi42; Circle Cardiovascular Imaging). In case of discrepancy, a consensus reading was performed. The presence and distribution pattern (subendocardial, midmyocardial, and epicardial) of LGE was documented using the AHA 17-segment model 45 .
For measuring myocardial fibrosis, T1 ECG-gated steady-state free-precession-based modified Look-Locker inversion-recovery (MOLLI) sequences with 5(3)3 pattern (acquiring 5 images after the first inversion, followed by a 3 heartbeat pause and then again 3 images after the second inversion) were acquired pre-and 10 minutes post contrast. These sequences were acquired on short axis at the mid-ventricular and basal short-axis plane (segments 1-12) with the following parameters:: Slice thickness 8 mm, spatial resolution: 1.5 × 1.5 mm2, acquired voxel size: 2.25 × 1.5 mm2, FOV: 323 × 380 mm using a 256 × 144 mm matrix, TE: 1.1, TI: 100-3500 with a 35° flip angle. T1 relaxation times were calculated per segment (1-12 of the 17 segments of AHA classification). The inner and outer contour of the left ventricular myocardium was segmented following recommendations 46 to omit influence of surrounding fat or blood by two blinded readers. Segments with obvious artifacts or presence of LGE were discarded. Another region of interest was placed in the blood volume. ECV as a measure of the amount of extracellular matrix was calculated from the T1 relaxation times pre and post contrast by taking into account the haematocrit. Fibrosis volume was derived using the following formula:  www.nature.com/scientificreports www.nature.com/scientificreports/ Assessment of smoking habits, physical activity and covariates. Smoking habits and physical activity of the subjects were measured in a standardized interview during the FF4 follow-up at the KORA study center. Physical activity was scaled as follows: Physical activity regularly more than 2 hours per week (group 1), regularly one hour per week (group 2), irregularly one hour per week (group 3), almost no activity (group 4). For smoking the scaling was: smoker, ex-smoker, never-smoker. Subjects were classified as smokers if they reported current regular or sporadic cigarette smoking.
Several co-variates were measured in the KORA FF4 study center by standardized interview, basic health examinations and laboratory analyses. They comprised: age (years), sex (male, female), body mass index (BMI, calculated as weight divided by squared height, kg/m²), waist circumference, body surface area, systolic and diastolic blood pressure, diabetes status (non-diabetics, pre-diabetics, diabetics), intake of antihypertensive, lipid-lowering or antidiabetic medication. Laboratory measurements including HbA1c, glucose, total cholesterol, high and low density lipoprotein and triglycerides were performed according to KORA standards 47 . Blood pressure was measured three times at the right arm of seated participants after an at least five-minute resting period. The mean of the second and third measurement was used for the analyses. Arterial hypertension was defined as increased systolic blood pressure (≥140 mmHg) and/or increased diastolic blood pressure (≥90 mmHg) or use of antihypertensive medication under awareness of having hypertension according to 2003 World Health Organization / International Society of Hypertension criteria 48 .
Statistical analyses. Baseline characteristics of the study sample were separately described for women and men by median and interquartile range for continuous variables and absolute numbers and percent values for categorical variables.
Associations of physical inactivity and smoking (risk factors = independent variables) with left and right ventricular structure and function parameters (outcomes = dependent variables) were investigated by calculating adjusted predicted outcome means or proportions with 95% confidence intervals by using average co-variable values after separated linear or logistic regression models for each risk factor and each outcome. Adjustments were made for age, sex, body mass index, systolic and diastolic blood pressure as well as diabetes mellitus. Physical inactivity and smoking status were treated as factor variables and each predicted mean was compared with its reference category of regular 2 hours activity per week or never smoking status, respectively. The p-value from t-test of the corresponding β-coefficient from the linear regression model was used to evaluate outcome differences between risk factor groups. Normal distributions of predicted residuals were tested visually. In addition, observed differences of LV-EF between physical activity categories were presented by box plots providing group medians, interquartile ranges and lower and upper adjacent values. All analyses were additionally adjusted for sampling weights considering differences in age, sex and diabetes status between the study sample (n = 400) and the entire KORA cohort (n = 2279, median age = 60 years, 48% men, 15% participants with diabetes) yielding no substantially changed findings. A p-value of <0.05 was considered statistically significant. Statistical analyses were performed using Stata 14.1 (Stata Corporation, College Station, TX, U.S.A.).

Data availability
All data are available on the KORA.PASST platform, accessible by contacting the Helmholtz Centre, Munich. Additional information is shown in the appendix.