European Heart Journal (2016) 37, 2756–2764 CLINICAL RESEARCH

doi:10.1093/eurheartj/ehv738 Coronary artery disease

Impact of active smoking on myocardial infarction

severity in reperfused ST-segment elevation
myocardial infarction patients: the smoker’s
paradox revisited

Downloaded from https://academic.oup.com/eurheartj/article/37/36/2756/2197391 by guest on 20 October 2021

Rolf Symons 1*†, Pier Giorgio Masci 2†, Marco Francone 3, Piet Claus 1, Andrea Barison 4,
Iacopo Carbone 3, Luciano Agati 3, Nicola Galea 3, Stefan Janssens 5, and Jan Bogaert 1
1
Department of Imaging and Pathology, Medical Imaging Research Centre, University Hospitals, Herestraat 49, 3000 Leuven, Belgium; 2Centre for Cardiovascular Magnetic Resonance,
Lausanne University Hospital, Lausanne, Switzerland; 3Department of Radiological, Oncological, and Pathological Sciences, La Sapienza University, Rome, Italy; 4MRI Department,
Fondazione G. Monasterio CNR-Regione Toscana, Pisa, Italy; and 5Department of Cardiovascular Diseases, University Hospitals Leuven, Leuven, Belgium

Received 24 June 2015; revised 14 November 2015; accepted 10 December 2015; online publish-ahead-of-print 24 January 2016

See page 2765 for the editorial comment on this article (doi:10.1093/eurheartj/ehw111)

Aims To investigate the influence of cardiovascular risk factors (CV-RFs) on infarct severity and post-infarction left ventricu-
lar (LV) remodelling in acutely reperfused ST-segment elevation myocardial infarction (STEMI) patients studied with
cardiovascular magnetic resonance (CMR).
.....................................................................................................................................................................................
Methods Four-hundred seventy-one patients were included in the study. Baseline CMR was performed at 4 + 1 days after STEMI
and results to assess area-at-risk, infarct size (IS), myocardial salvage index (MSI), microvascular obstruction (MVO), intramyocar-
dial haemorrhage (IMH), LV volumes, and function. Cardiovascular magnetic resonance was repeated 4 months after
STEMI (n ¼ 383) to assess adverse LV remodelling (increase of LV end-diastolic volume .20% between baseline and
follow-up). Smoking was associated with IMH at baseline even after correction for other factors associated with ischae-
mia-reperfusion injury including MVO, IS, and MSI (OR: 2.17, 95% CI: 1.17– 4.00, P ¼ 0.01). Unexpectedly, smoking was
an independent protective predictor against adverse LV remodelling (OR: 0.43, 95% CI: 0.24–0.77, P ¼ 0.005), consist-
ent with the ‘smoker’s paradox’. However, the presence of IMH at baseline abolished the paradoxical, beneficial effects
of smoking with respect to IS, baseline LV function, and post-infarction LV remodelling. No association between other
CV-RFs, infarct severity, or post-infarction LV remodelling was observed.
.....................................................................................................................................................................................
Conclusion In patients with reperfused STEMI, smoking is strongly and independently associated with IMH at baseline. Nonetheless,
consistent with the ‘smoker’s paradox’, smoking was an independent predictor of more favourable post-infarction LV
remodelling. However, the paradoxical beneficial effects of smoking were lost in patients with IMH.
-----------------------------------------------------------------------------------------------------------------------------------------------------------
Keywords Myocardial infarction † Left ventricular remodelling † Cardiovascular risk factors † Smoker’s paradox

circulation via endothelial dysfunction, enhanced inflammation and

Introduction
Traditional cardiovascular risk factors (CV-RFs) including high chol-
esterol, arterial hypertension, family history of ischaemic heart dis-
ease, smoking, diabetes, and obesity predict coronary artery disease
at the level of the epicardial vessels and microvasculature.1 – 3 For
many years preceding an acute coronary syndrome, these CV-RFs
exert detrimental effects at different levels of the coronary
oxidative stress, predisposing patients to acute ischaemic complica-
tions.3 However, the influence of CV-RFs on the severity of ischae-
mic myocardial damage at the time of acute ST-segment elevation
myocardial infarction (STEMI) and, thereafter on post-infarction
left ventricular (LV) remodelling, has not been explored yet.
With the advent of cardiovascular magnetic resonance (CMR)
imaging, comprehensive information on myocardial infarction

* Corresponding author. Medical Imaging Research Centre, University Hospitals Leuven, Herestraat 49, 3000, Leuven, Belgium. Tel: +32 472249304, Email: rolf.symons@uzleuven.be
†
Both first authors have equally contributed to this manuscript.
Published on behalf of the European Society of Cardiology. All rights reserved. & The Author 2016. For permissions please email: journals.permissions@oup.com.
The smoker’s paradox revisited
severity can be obtained in a single exam, including the extent of the
area-at-risk (AAR), infarct size (IS), myocardial salvage index (MSI),
microvascular obstruction (MVO), and intramyocardial haemor-
rhage (IMH), along with an accurate assessment of LV morphology
and function.4 – 6 Intramyocardial haemorrhage is a marker of irre-
versible damage of the coronary microvasculature, and is a strong
and independent predictor of adverse LV remodelling and clinical
outcome in patients with STEMI.7 – 9 Our aim was to determine
the influence of CV-RFs on the pattern of acute ischaemic damage
and post-infarction LV remodelling in a large prospective cohort of
patients with STEMI treated by primary percutaneous coronary
intervention (pPCI), using comprehensive CMR at a baseline early
post-infarction phase and after 4-month follow-up.
Methods
Study design
This is a multicentre, prospective, longitudinal study conducted in a co-
hort of acute STEMI patients studied with CMR. Between May 2005 and
May 2014, 512 acute STEMI patients from three tertiary referral centres
[244 at University Hospitals Leuven, Leuven, Belgium (Centre A), 157 at
La Sapienza University Hospital, Rome, Italy (Centre B), and 111 at Fon-
dazione G. Monasterio, Pisa, Italy (Centre C)] were screened for study
enrolment. Patients were included if they were older than 18 years, met
ECG criteria for acute STEMI and were successfully treated by pPCI
within 12 h of symptom onset (i.e. chest pain). Exclusion criteria in-
cluded pulmonary oedema and/or cardiogenic shock persisting after
the first week of hospitalization, prior coronary revascularization or
myocardial infarction, claustrophobia and glomerular filtration rate
,30 mL/min. The study complied with the Declaration of Helsinki.
Local ethic review boards approved the protocol, and written informed
consent was obtained from all patients.
Coronary artery disease risk factor
assessment
In each centre, a cardiologist blinded to CMR results, assessed CV-RFs
profile using a structured interview. Patients who smoked regularly dur-
ing the previous 12 months were classified as smokers.10 Diabetes was
considered present if patients had history of diabetes and they were on
Figure 1 (A) Short-axis T2-weighted image at baseline shows high signal intensity in the left ventricular lateral wall corresponding to the
area-at-risk (arrows). Central hypointense area corresponds with intramyocardial haemorrhage (asterisk). (B) Short-axis late gadolinium enhance-
ment image at baseline shows high signal intensity in the left ventricular lateral wall corresponding to an extensive area of transmural necrosis
(arrows). Central hypointense area corresponds with microvascular obstruction (asterisk). (C) Short-axis late gadolinium enhancement image
at follow-up shows important wall thinning and remodelling of the lateral wall (arrows).
2757
insulin or oral hypoglycaemic agents.11 Hypertension was defined as his-
tory of high blood pressure in association with antihypertensive medical
therapy,12 or evidence of high systolic (.140 mmHg, or .135 mmHg
in diabetic patients) or diastolic (.90 mmHg, or .85 mmHg in diabetic
patients) blood pressure during hospitalization. Patients were identified
as having dyslipidaemia if this condition was certified by the treating
physician, if patients were on lipid-lowering medications or if low-
density lipoprotein cholesterol levels exceeded 160 mg/dL
(4.14 mmol/L).13 Family history of coronary artery disease was defined
as the presence of cardiovascular disease in a first-degree relative before
the age of 55 years for men or 65 years for women. Obesity was defined
as a BMI ≥ 30.14 At follow-up, patients were classified according to their
risk profile at baseline. Clinical management, medical treatment, and
Downloaded from https://academic.oup.com/eurheartj/article/37/36/2756/2197391 by guest on 20 October 2021
pPCI were performed according to the most recent ESC guidelines
for acute STEMI.15
Cardiovascular magnetic resonance protocol
All CMR studies were performed on 1.5T units, Intera-CV (Philips, Best,
The Netherlands) (Centre A), Avanto (Siemens, Erlangen, Germany)
(Centre B), and CVi (GE Healthcare, Milwaukee, WI, USA) (Centre
C), using dedicated cardiac software, phased-array surface receiver
coils, and electrocardiogram triggering. Breath-hold steady-state free-
precession cine CMR was performed in cardiac vertical and horizontal
long-axis and in short-axis orientation. Area-at-risk and IMH were de-
termined using breath-hold black-blood T2-weighted short inversion-
time inversion-recovery fast spin-echo sequence in cardiac short axis.
Post-contrast breath-hold T1-weighted 2D (Avanto/CVi) or 3D (Intera)
inversion-recovery segmented gradient-echo sequence was used
to detect MVO and myocardial necrosis. Late gadolinium enhance-
ment (LGE) imaging was performed 10 – 20 min following contrast
administration.
Image analysis
All CMR studies were analysed by experienced observers blinded to the
clinical data (JB, MF, PGM). T2-weighted images were used to quantify
AAR and to determine the presence of IMH in the ischaemic myocar-
dium. In the LV myocardial wall supplied by the infarct-related artery,
myocardial tissue with signal intensity (SI) 2 SDs above the mean signal
obtained in the remote non-infarcted myocardium was considered
AAR. The contours of automatically detected AAR were manually
adapted to exclude increased SI from the adjacent blood pool
2758 R. Symons et al.

(‘slow flow’) or artefacts related to cardiac motion or respiration. Intra-
myocardial haemorrhage was defined as a ≥1 mL hypointense area in
the core of the AAR having mean SI , 2SD the SI of the periphery of
the AAR. Intramyocardial haemorrhage was considered part of myocar-
dial oedema for calculating AAR.7 On LGE images, infarcted myocar-
dium was identified as the myocardium with SI . 5SD exceeding the
mean SI of remote myocardium. The contours of automatically detected
infarcted myocardium were manually adapted. Microvascular obstruc-
tion was defined as the hypoenhanced region within the infarcted myo-
cardium. If present, MVO was considered part of the infarcted
myocardium (Figure 1). Myocardial salvage index was defined as the dif-
ference between AAR and IS divided by AAR extent.4 Cardiovascular
magnetic resonance was used to calculate LV end-diastolic (EDV), end-
Statistical analysis
Continuous variables were expressed as mean + SD or median (25th–
75th percentiles) as appropriate. Categorical variables were expressed
as frequency with percentage. Student’s independent sample t- or the
Mann – Whitney test was used as appropriate to compare continuous
variable differences. A comparison between categorical variables was
performed by x 2 test. Pearson’s correlation coefficient (r) was used
to test correlation between continuous variables. Student’s paired sam-
ple t- or Wilcoxon’s test was used as appropriate to compare continu-
ous variable differences between baseline and follow-up. Adverse LV
remodelling was defined as an increase in LV-end-diastolic volume index
(EDVi) .20% between baseline and follow-up.16,17 Univariable logistic
regression analysis was utilized to assess the association between base-

Downloaded from https://academic.oup.com/eurheartj/article/37/36/2756/2197391 by guest on 20 October 2021

systolic (ESV) and stroke (SV) volumes, ejection fraction (EF), and mass. line variables and CV-RFs with IMH occurrence at baseline CMR and ad-
Left ventricular volumes and mass were normalized to BSA, IS was nor- verse LV remodelling. Variables with P , 0.10 at univariable analysis
malized to LV mass (IS%). were included as covariates in multivariable logistic regression analysis.

Table 1 Baseline characteristics in non-smokers vs. smokers

Variable All Non-smokers Smokers P-value

(n 5 471) (n 5 244) (n 5 227)
...............................................................................................................................................................................
Age (years) 60.0 + 11.6 64.4 + 11.3 55.3 + 9.9 ,0.001
BMI (kg/m2) 26.9 + 3.9 27.0 + 4.0 26.9 + 3.9 0.75
...............................................................................................................................................................................
CV-RFs n (%)
Male gender 390 (83) 195 (80) 195 (86) 0.09
Diabetes 70 (15) 45 (18) 25 (11) 0.02
Hypertension 215 (46) 119 (49) 96 (42) 0.16
Dyslipidaemia 282 (60) 152 (62) 130 (57) 0.27
Family history of CAD 232 (49) 119 (49) 113 (50) 0.83
Obesity 87 (19) 43 (18) 44 (19) 0.65
...............................................................................................................................................................................
Medical treatment at baseline n (%)
ACEi/ARB 53 (11) 34 (14) 19 (8) 0.06
b-blocker 48 (10) 23 (9) 25 (11) 0.54
Statin 50 (11) 33 (14) 17 (8) 0.24
Antiplatelet agent 37 (8) 26 (11) 11 (5) 0.26
Oral hypoglycaemic drug 20 (4) 14 (6) 6 (3) 0.10
Insulin 12 (3) 9 (3) 3 (1) 0.14
...............................................................................................................................................................................
Blood pressure on admission (mmHg)
Systolic 131.8 + 22.6 132.0 + 22.0 131.6 + 23.4 0.86
Diastolic 78.1 + 12.9 77.8 + 13.0 78.5 + 12.7 0.57
Peak troponin I (mg/L) 50.0 (13.0– 111.9) 53.9 (10.5– 117.3) 44.1 (16.0–99.2) 0.82
Time to pPCI (min) 196 (140–295) 210 (150– 300) 189 (138– 282) 0.14
...............................................................................................................................................................................
Infarct-related vessel, n (%)
RCA 141 (30) 73 (30) 68 (30) 0.84
LAD 236 (50) 127 (52) 109 (48) 0.38
LCx 94 (20) 44 (18) 50 (22) 0.28
TIMI 0– 1 flow grade before pPCI, n (%) 286 (72) 138 (68) 148 (76) 0.11
TIMI 3 flow grade after pPCI, n (%) 351 (88) 181 (90) 170 (87) 0.38
SYNTAX score 16.9 + 8.9 17.5 + 9.3 16.1 + 8.3 0.11
RCA dominance, n (%) 364 (92) 182 (90) 182 (93) 0.28

ACEi, angiotensin-converting enzyme inhibitor; ARB, angiotensin-II inhibitor; CV-RFs, cardiovascular risk factors; LAD, left anterior descending coronary artery; LCx, left
circumflex coronary artery; pPCI, primary percutaneous coronary intervention; RCA, right coronary artery; TIMI, thrombolysis in myocardial infarction; Time to pPCI, time from
symptom onset to pPCI; SYNTAX, synergy between PCI with TAXUS and cardiac surgery; TIMI-grade, SYNTAX score and coronary dominance in 398 patients (non-current
smoking ¼ 202; current smoking ¼ 196).
The smoker’s paradox revisited 2759

Considering the association between active smoking and IMH occur-

rence, the interaction of these variables on adverse LV remodelling
was tested. Considering the strong correlation between AAR and IS
(r ¼ 0.84, P , 0.001) and between left ventricular ejection fraction
(LV-EF) and LV-ESVi (r ¼ 20.76, P , 0.001), these covariates were in-
troduced separately in the multivariable models. All statistical analyses
were conducted with SPSS (version 17.0, SPSS, Chicago, Illinois).
P-values of ,0.05 were considered to indicate statistical significance.
All reported P-values are two tailed.

Results
Study population

Downloaded from https://academic.oup.com/eurheartj/article/37/36/2756/2197391 by guest on 20 October 2021

Baseline characteristics are summarized in Table 1. Of 512 eligible
patients, 471 were included in the final analysis at baseline and
383 (81%) at follow-up (Figure 2). All patients were successfully trea-
ted with pPCI within 12 h of symptom onset (median: 196 min,
25th – 75th percentiles: 140 – 295 min). Cardiovascular magnetic
resonance was performed as soon as reasonably achievable, with re-
spect to clinical status and organizational feasibility. Nearly half of
the patients were smokers (48%), and they were approximately
one decade younger than non-smokers and had less diabetes.

Cardiovascular risk factors and infarct

severity
Intramyocardial haemorrhage was detected in 118 patients (25%).
Smokers had larger LV-EDVi, SVi, and higher incidence of IMH com-
pared with non-smokers (31 vs. 20%, P ¼ 0.005). At follow-up,
LV-EF increased in smokers and non-smokers due to a slight but sig-
nificant increase in LV-EDVi (Table 2). With the exception of smok-
ing, no association was observed between CV-RFs and LV-EF, IS,
MVO, IMH, or LV volumes. Indeed, smoking was strongly and inde-
pendently associated with occurrence of IMH (see Supplementary
material online).
Stratification of patients based on the presence or absence of
IMH is shown in Table 3 and Figure 3. In the non-IMH group, smokers
had smaller IS, despite similar AAR, greater LV-EF and LV-SVi than
non-smokers. At follow-up, LV-EF and LV-SVi improved in smokers
and non-smokers. However, in smokers this improvement was at-
tributed to a decrease of LV-ESVi. Conversely, in non-smokers
the improvements of LV-EF and LV-SVi were related to an increase Figure 2 Patient flow chart.
of LV-EDVi rather than to a reduction of LV-ESVi. These findings de-
note that in the non-IMH group, smokers experienced more favour-
able post-infarction LV remodelling than non-smokers. (P ¼ 0.46). At multivariable logistic regression analysis, smoking
In the IMH group, no significant differences were found between remained a protective factor against adverse LV remodelling. This
non-smokers and smokers with respect to IS, LV volumes, and result was confirmed when AAR replaced infarct size in the multi-
LV-EF at baseline. At follow-up, LV-EF did not improve in smokers variable model (Table 5).
and non-smokers despite a significant increase of LV volumes, de- Smoking was protective against adverse LV remodelling at uni-
noting adverse post-infarction LV remodelling in all patients, regard- variable and multivariable analysis in patients without IMH. On the
less of their smoking status. contrary, in patients with IMH no association was observed between
smoking and adverse LV remodelling (Table 6).
The impact of smoking on post-infarction
left ventricular remodelling
At univariable logistic regression, non-smoking, higher peak tropo-
Discussion
nin level, larger AAR and infarct size, lower MSI, and MVO were as- In this multicentre, prospective study conducted in a cohort
sociated with adverse LV remodelling (Table 4). No interaction was of acutely reperfused STEMI patients investigated with serial
observed between smoking and IMH on adverse LV remodelling CMR, we found that smokers had a higher incidence of IMH than
2760 R. Symons et al.

Table 2 Cardiovascular magnetic resonance data at baseline and follow-up

Variable All Non-smokers Smokers P-value

Baseline (n 5 471) (n 5 244) (n 5 227)
Follow-up (n 5 383) (n 5 194) (n 5 189)
...............................................................................................................................................................................
AAR (g)
Baseline 34.3 + 21.7 34.9 + 21.2 33.7 + 22.3 0.55
IS%
Baseline 16.7 + 12.8 17.4 + 13.2 15.9 + 12.4 0.22
Follow-up 11.0 + 8.9 11.6 + 9.3 10.4 + 8.6 0.20
P-value ,0.001 ,0.001 ,0.001

Downloaded from https://academic.oup.com/eurheartj/article/37/36/2756/2197391 by guest on 20 October 2021

MSI (%)
Baseline 42.4 + 26.7 42.4 + 26.9 42.3 + 26.6 0.96
MVO, n (%)
Baseline 212 (45) 104 (43) 108 (48) 0.28
IMH, n (%)
Baseline 118 (25) 48 (20) 70 (31) 0.005
LV-EDVi (mL/m2)
Baseline 78.3 + 18.5 76.2 + 18.0 80.6 + 18.8 0.01
Follow-up 82.5 + 21.4 81.8 + 20.5 83.2 + 22.4 0.53
P-value ,0.001 ,0.001 0.04
LV-ESVi (mL/m2)
Baseline 40.1 + 14.5 39.4 + 14.3 40.8 + 14.7 0.30
Follow-up 40.7 + 16.8 40.7 + 16.5 40.7 + 17.1 0.98
P-value 0.54 0.07 0.45
LV-SVi (mL/m2)
Baseline 38.2 + 9.4 36.7 + 8.8 39.7 + 9.8 ,0.001
Follow-up 41.8 + 10.6 41.1 + 10.3 42.4 + 10.9 0.23
P-value ,0.001 ,0.001 ,0.001
LV-EF (%)
Baseline 49.5 + 9.9 49.0 + 9.8 50.1 + 9.9 0.20
Follow-up 51.8 + 10.3 51.4 + 10.4 52.2 + 10.2 0.47
P-value ,0.001 ,0.001 ,0.001
LV-Mass (g/m2)
Baseline 64.4 + 15.4 63.4 + 15.4 65.4 + 15.2 0.16
Follow-up 56.7 + 12.2 56.5 + 12.2 56.9 + 12.2 0.76
P-value ,0.001 ,0.001 ,0.001

AAR, area-at-risk; EDVi, end-diastolic volume index; EF, ejection fraction; ESVi, end-systolic volume index; IMH, intramyocardial haemorrhage; IS, infarct size; LV, left ventricular;
MSI, myocardial salvage index; MVO, microvascular obstruction; SVi, stroke-volume index.

non-smokers, even though the former was 1 decade younger and
had less co-morbidities than non-smokers. The association between
smoking and IMH was confirmed by multivariable analysis after cor-
rection for factors implicated in ischaemia-reperfusion injury, includ-
ing IS, MVO, and MSI. Despite the independent association between
smoking and the development of IMH, smoking was an independent
predictor of better post-infarction LV remodelling at follow-up. Fur-
thermore, by dichotomizing the study population based on the pres-
ence or absence of IMH, we observed that in the subgroup without
IMH smokers had lower IS, better LV systolic performance, and
more favourable post-infarction LV remodelling compared with
non-smokers. These paradoxical beneficial effects of smoking
were abolished in patients with IMH.
In agreement with literature, we found that smokers were about
one decade younger and had a more favourable CV-RFs profile than
non-smokers. The better CV-RFs profile of smokers may explain
the result of univariable analysis, which showed a protective effect
of smoking against adverse LV remodelling. However, after adjust-
ment for other factors implicated in post-infarction LV remodelling,
such as IS and MSI, smoking remained an independent, paradoxical
protective factor. These findings are consistent with the ‘smoker’s
paradox’, a previously reported, but not consistently observed, as-
sociation between smoking and better short-term outcomes after
acute infarction.18 The ‘smoker’s paradox’ has been first described
.25 years ago, and continues to stir discussions in studies of
patients with acute myocardial infarction.19,20
Our study is the first to demonstrate an independent and strong
association between smoking and IMH occurrence in a large pro-
spective cohort of STEMI patients. Intramyocardial haemorrhage
is a marker of severe ischaemia-reperfusion injury caused by
The smoker’s paradox revisited 2761

Table 3 Cardiovascular magnetic resonance data at baseline and follow-up in the subgroups with and without
intramyocardial haemorrhage

No IMH Smokers Non-smokers P-value

Baseline/follow-up (n 5 157/134) (n 5 196/156)
...............................................................................................................................................................................
Age (years) 57 + 10 64 + 11 ,0.001
Diabetes, n (%) 14 (9) 34 (17) 0.02
AAR (g) 28.2 + 18.8 31.0 + 19.1 0.46
MVO, n (%) 43 (27) 65 (33) 0.44
IS% baseline 11.5 + 8.9 14.7 + 11.3 0.003
IS% follow-up 7.5 + 5.9 10.4 + 8.6 0.001

Downloaded from https://academic.oup.com/eurheartj/article/37/36/2756/2197391 by guest on 20 October 2021

P-value ,0.001 ,0.001
LV-EDVi (mL/m2) baseline 77.6 + 16.4 74.0 + 16.3 0.15
LV-EDVi (mL/m2) follow-up 78.0 + 17.0 78.6 + 18.4 0.78
P-value 0.75 ,0.001
LV-ESVi (mL/m2) baseline 37.7 + 12.5 37.4 + 12.7 0.70
LV-ESVi (mL/m2) follow-up 35.9 + 13.1 37.2 + 13.3 0.41
P-value 0.04 0.72
LV-SVi (mL/m2) baseline 40.0 + 8.8 36.6 + 8.4 0.001
LV-SVi (mL/m2) follow-up 42.0 + 9.5 41.3 + 10.5 0.61
P-value 0.01 ,0.001
LV-EF (mL/m2) baseline 51.9 + 9.0 50.1 + 9.2 0.03
LV-EF (mL/m2) follow-up 54.9 + 9.4 53.5 + 9.5 0.21
P-value ,0.001 ,0.001
LV-Mass (mL/m2) baseline 62.6 + 13.5 60.4 + 11.9 0.52
LV-Mass (mL/m2) follow-up 55.7 + 11.4 55.9 + 12.5 0.87
P-value ,0.001 ,0.001
...............................................................................................................................................................................
IMH Smokers Non-smokers P-value
Baseline/follow-up (n 5 70/55) (n 5 48/38)
...............................................................................................................................................................................
Age (years) 55 + 8 64 + 11 ,0.001
Diabetes, n (%) 11 (16) 11 (23) 0.32
AAR (g) 48.1 + 23.0 50.5 + 22.5 0.58
MVO, n (%) 65 (93) 39 (81) 0.06
IS% baseline 25.8 + 13.2 28.1 + 14.6 0.37
IS% follow-up 18.0 + 9.8 17.1 + 10.3 0.74
P-value ,0.001 ,0.001
LV-EDVi (mL/m2) baseline 88.9 + 18.6 84.2 + 18.4 0.21
LV-EDVi (mL/m2) follow-up 95.8 + 28.4 95.2 + 23.5 0.92
P-value 0.01 ,0.001
LV-ESVi (mL/m2) baseline 49.7 + 15.2 47.3 + 13.7 0.55
LV-ESVi (mL/m2) follow-up 52.2 + 20.0 55.1 + 20.5 0.50
P-value 0.29 0.001
LV-SVi (mL/m2) baseline 39.1 + 8.4 36.3 + 9.7 0.13
LV-SVi (mL/m2) follow-up 43.5 + 13.8 40.0 + 9.4 0.15
P-value 0.002 0.03
LV-EF (mL/m2) baseline 44.9 + 9.2 43.4 + 8.7 0.37
LVEF (mL/m2) follow-up 45.6 + 9.2 42.9 + 9.8 0.18
P-value 0.13 0.30
LV-Mass (mL/m2) baseline 70.1 + 15.4 68.1 + 15.5 0.49
LV-Mass (mL/m2) follow-up 55.9 + 13.7 59.1 + 10.9 0.76
P-value ,0.001 ,0.001

Abbreviations as reported in previous tables.

2762 R. Symons et al.

Downloaded from https://academic.oup.com/eurheartj/article/37/36/2756/2197391 by guest on 20 October 2021

Figure 3 Differences in left ventricular ejection fraction (%) and IS% in the intramyocardial haemorrhage and non-intramyocardial haemorrhage
group between smokers and non-smokers.

post-infarction IMH and the detrimental effects of smoking on the

Table 4 Univariable analysis for determinants of microvasculature, it is conceivable that the coronary microcircula-
adverse left ventricular remodelling tion in smokers was less resistant to ischaemia-reperfusion injury,
explaining the strong and independent association between smoking
Characteristics OR (95% CI) P-value
................................................................................ and IMH (Figure 4).
Age (per year) 1.01 (0.99–1.03) 0.44 Interestingly, our results suggest a different impact of smoking on
BMI (per kg/m2) 0.55 (0.15–2.02) 0.37 acute ischaemic damage and post-infarction remodelling based on
Male gender (yes vs. no) 0.59 (0.32–1.09) 0.09 the presence or absence of IMH. In the subgroup of patients without
Smoking (yes vs. no) 0.44 (0.26–0.73) 0.002 IMH at baseline CMR, smokers had lower IS and better LV systolic
Diabetes (yes vs. no) 0.56 (0.24–1.28) 0.17 performance compared with non-smokers. During follow-up, smo-
Hypertension (yes vs. no) 0.90 (0.54–1.49) 0.68 kers showed an improvement in LV function due to a reduction in
Dyslipidaemia (yes vs. no) 0.85 (0.52–1.41) 0.53 LV-ESVi without any concomitant variation in LV-EDVI, denoting a
Familial history of CAD (yes vs. no) 1.15 (0.70–1.88) 0.59 favourable pattern of ventricular remodelling. Conversely, in non-
Obesity (yes vs. no) 0.96 (0.51–1.80) 0.90 smokers, the improvement of LV-EF occurred at the expense of
Peak Troponin (per mg/L) 1.00 (1.00–1.01) 0.01 an increase of LV-EDVi without any reduction of LV-ESVi.
Time to pPCI (per min) 1.00 (1.00–1.00) 0.95 On the other hand, in patients with IMH, smokers and non-
AAR (per g) 1.02 (1.00–1.03) 0.007 smokers showed comparable IS, LV volumes, and function. During
IS (per %) 50.75 (7.89–326.42) ,0.001 follow-up, smokers and non-smokers showed an increase in LV vo-
MSI (per %) 0.23 (0.08–0.64) 0.005 lumes without improving LV-EF. These results were confirmed by
MVO (yes vs. no) 1.70 (1.03–2.80) 0.04 univariable and multivariable analyses, which showed that in the
IMH (yes vs. no) 1.64 (0.95–2.84) 0.08 non-IMH subgroup smoking was an independent protective predictor
b-Blocker (yes vs. no) 1.59 (0.76–3.32) 0.22 against adverse LV remodelling, whereas no association was found
ACEi/ARB (yes vs. no) 1.27 (0.62–2.61) 0.51 between smoking and post-infarction LV remodelling in the IMH
Statin (yes vs. no) 1.03 (0.45–2.39) 0.94 subgroup. Thus, the development of profound structural damage
SYNTAX score (per point) 1.03 (0.87–1.16) 0.24 of the coronary microvasculature at early post-infarction phase
RCA dominance (yes vs. no) 0.78 (0.34–1.83) 0.57 abolished the potential paradoxical protective effects of smoking
with respect to IS and post-infarction LV remodelling. These
Abbreviations as reported in previous tables. findings, if confirmed by larger prospective studies, warrant a
more vigorous strategy for mitigating the hazardous effects of
ischaemia-reperfusion in smokers with STEMI. Remarkably, in the
profound structural damage of the coronary microvasculature, IMH subgroup the severity of acute ischaemic damage, LV dys-
which leads to extravasation of erythrocytes.21 The hazardous ef- function and post-infarction remodelling was comparable between
fects of smoking on the microcirculation due to increased oxidative smokers and non-smokers, although smokers were about one
stress and prothrombotic state are well established.3 Given the cen- decade younger and had a more favourable CV-RFs profile than
tral role of the coronary microcirculation in the development of non-smokers.
The smoker’s paradox revisited 2763

Table 5 Multivariable analysis for determinants of adverse left ventricular remodelling

Model-a P-value Model-b P-value

OR (95% CI) OR (95% CI)
...............................................................................................................................................................................
Male gender (yes vs. no) 0.49 (0.24– 1.01) 0.05 0.43 (0.21– 0.88) 0.02
Smoking (yes vs. no) 0.43 (0.24– 0.77) 0.005 0.42 (0.23– 0.75) 0.003
Peak troponin (per mg/L) 1.00 (1.00– 1.00) 0.36 1.00 (1.00– 1.00) 0.36
AAR (per g) N/A N/A 1.01 (1.00– 1.03) 0.15
IS (per %) 11.87 (0.58– 241.40) 0.11 N/A N/A
MSI (per %) 0.56 (0.14– 2.26) 0.41 0.30 (0.09– 1.03) 0.06
MVO (yes vs. no) 1.11 (0.54– 2.26) 0.87 1.14 (0.56– 2.30) 0.72

Downloaded from https://academic.oup.com/eurheartj/article/37/36/2756/2197391 by guest on 20 October 2021

IMH (yes vs. no) 1.07 (0.50– 2.29) 0.87 1.08 (0.51– 2.32) 0.84

Abbreviations as reported in previous tables.

Table 6 Univariable and multivariable analyses for adverse left ventricular remodelling based on the presence or
absence of intramyocardial haemorrhage at baseline

Variable Overall Smokers Non-smokers OR (95% CI) P-value

...............................................................................................................................................................................
No-IMH (n ¼ 290)
Adverse remodelling 53/290 (18%) 14/134 (10%) 39/156 (25%) 0.35 (0.18– 0.67) 0.002
0.38 (0.19– 0.76)a 0.006
IMH (n ¼ 93)
Adverse remodelling 25/93 (27%) 12/55 (22%) 13/38 (34%) 0.54 (0.21– 1.36) 0.19

a
After correction for male gender, peak of troponin, IS, and MSI; abbreviations as in previous tables.

Although the mechanisms explaining the loss of paradoxical
beneficial effects of smoking in the IMH subgroup may be attributed
to deleterious consequences of an impaired myocardial reperfu-
sion,22 those associated with reduced IS and more favourable post-
infarction LV remodelling in smokers compared with non-smokers
in the non-IMH subgroup remain largely speculative. Carbon mon-
oxide (CO) might be advocated as a potential beneficial mediator in
smokers without IMH23 (Figure 4). Smokers have higher blood con-
centration of CO-haemoglobin than non-smokers, and CO is an im-
portant biological regulator with antioxidant, anti-inflammatory,
anti-apoptotic, anti-thrombotic, and vasoregulatory properties.
However, the lack of information on blood concentration of CO-
haemoglobin and smoking habit during follow-up render further in-
terpretation of our findings elusive, and further studies are war-
ranted to better understand the potential mechanisms of the
‘smoker’s paradox’.
correlations of the presence and extent of IMH. Recent studies,
however, have validated the accuracy of CMR in the diagnosis
of IMH. 21 Third, we used T2-weighted sequences to detect
IMH. Recent studies showed T2*-weighted sequences to be
more sensitive in detecting IMH.24 Thus, our study may slightly
underestimate the prevalence of IMH. Fourth, patients with pul-
monary oedema and/or cardiogenic shock persisting after the
first week from the acute event were excluded from the study.
Thus, our data cannot be generalized to STEMI patients with
persistent haemodynamic instability. Fifth, we did not assess
modifications in CV-RFs at follow-up. Therefore, we cannot
determine a potential effect of smoking cessation on LV remod-
elling after 4 months. Finally, since the study participants were
predominantly Caucasian, our results cannot be generalized to
other ethnic groups.

Study limitations
This study was conducted at three different tertiary care centres
using different vendor CMR units. A similar study protocol was
used in all centres to maximise homogeneity in data acquisition
(see Supplementary material online). Our study explored the as-
sociation between CV-RFs, acute ischaemic myocardial damage,
and post-infarction remodelling, which has not been reported
in the literature. Accordingly, no preliminary statistical power
analysis was performed. Second, we lacked pathological
Conclusion
In STEMI patients, smoking is associated with IMH independently of
other factors implicated in ischaemia-reperfusion injury. Nonethe-
less, smoking was an independent protective predictor against ad-
verse LV remodelling during short-term follow-up, consistent with
the ‘smoker’s paradox’. While this phenomenon was clearly observ-
able in patients without IMH, ‘the smoker’s paradox’ was abolished
in patients developing IMH.
2764 R. Symons et al.

Downloaded from https://academic.oup.com/eurheartj/article/37/36/2756/2197391 by guest on 20 October 2021

Figure 4 Smoking increases the oxidative stress to the endothelium by the formation of free radicals and inflammatory cytokines, which leads to
a decrease of nitric oxide bioavailability and, thereby, to endothelial dysfunction. This may potentially render the endothelium more susceptible to
ischaemia-reperfusion injury favouring the development of intramyocardial haemorrhage. Potential cardioprotective mechanisms mediated by in-
haled carbon monoxide in smokers involve an increase in cyclic guanosine monophosphate levels and enhanced activity of big-conductance Ca2+-
activated potassium channels (K+ Ca2+), ultimately reducing intracellular Ca
2+
concentrations and promoting vasodilatation. Additionally, carbon
monoxide enhances the activity of mitochondrial ATP-dependent potassium channels (KATP) which prevents mitochondrial Ca2+ overload, a crit-
ical trigger for ischaemia-reperfusion injury. For a more detailed description of the potential cardioprotective mechanisms of carbon monoxide
see Ref. 23. CO, carbon monoxide; cGMP, cyclic guanosine monophosphate.

Supplementary material
Supplementary material is available at European Heart Journal online.
Authors’ contributions
R.S., P.G.M., P.C. performed statistical analysis; P.G.M., M.F., S.J., J.B.
handled funding and supervision; R.S., P.G.M., M.F., A.B., I.C., L.A.,
N.G., J.B. acquired the data; R.S., P.G.M., M.F., J.B. conceived and de-
signed the research; R.S., P.G.M., S.J., J.B. drafted the manuscript; R.S.,
P.G.M., M.F., P.C., A.B., I.C., L.A., N.G., S.J., J.B. made critical revision
of the manuscript for key intellectual content.
Conflict of interest: none declared.
References
1. Dawber TR, Meadors GF, Moore FE. Epidemiologic approaches to heart disease:
the Framingham study. Am J Public Health 1951;41:279 –286.
2. Kannel WB, Feinleib M, McNamara PM, Garrison RJ, Castelli WP. An investigation
of coronary heart disease in families: the Framingham offspring study. Am J Epidemiol
1979;110:281 –290.
3. Camici PG, Crea F. Coronary microvascular dysfunction. N Engl J Med 2007;356:
830 –840.
4. Masci PG, Ganame J, Strata E, Desmet W, Aquaro GD, Dymarkowski S, Valenti V,
Janssens S, Lombardi M, Van de Werf F, L’Abbate A, Bogaert J. Myocardial salvage
by CMR correlates with LV remodelling and early ST-segment resolution in acute
myocardial infarction. JACC Cardiovasc Imaging 2010;3:45– 51.
5. Eitel I, Desch S, Fuernau G, Schulz-Menger J, Messroghli D, Dietz R. Prognostic
significance and determinants of myocardial salvage assessed by cardiovascular
magnetic resonance. J Am Coll Cardiol 2008;51:1581 –1587.
6. Ganame J, Messalli G, Dymarkowski S, Rademakers FE, Desmet W, Van de Werf F,
Bogaert J. Impact of myocardial haemorrhage on left ventricular function and re-
modelling in patients with reperfused acute myocardial infarction. Eur Heart J
2009;30:1440 – 1449.
7. Symons R, Masci PG, Goetschalckx K, Doulaptsis K, Janssens S, Bogaert J. Effect of
infarct severity on regional and global left ventricular remodelling in patients with
successfully reperfused ST segment elevation myocardial infarction. Radiology 2015;
274:93 –102.
8. Eitel I, Kubusch K, Strohm O, Desch S, Mikami Y, de Waha S, Gutberlet M,
Schuler G, Friedrich MG, Thiele H. Prognostic value and determinants of a hy-
pointense infarct core in T2-weighted cardiac magnetic resonance in acute reper-
fused ST-elevation myocardial infarction. Circ Cardiovasc Imaging 2011;4:
354 – 362.
9. Mather AN, Fairbairn TA, Ball SG, Greenwood JP, Plein S. Reperfusion hemor-
rhage as determined by cardiovascular MRI is a predictor of adverse left
ventricular remodelling and markers of late arrhythmic risk. Heart 2011;97:
453 – 459.
The smoker’s paradox revisited
10. Wilson PWF, D’Agostino RB, Levy D, Belanger AM, Silbershatz H, Kannel WB.
Prediction of coronary heart disease using risk factor categories. Circulation 1998;
97:1837 – 1847.
11. Definition, diagnosis and classification of diabetes mellitus and its complications.
Part 1. Diagnosis and classification of diabetes mellitus. Geneva: World Health
Organization, 1999.
12. Joint National Committee on Prevention, Detection, Evaluation, and Treatment of
High Blood Pressure. National Heart, Lung, and Blood Institute; National High
Blood Pressure Education Program Coordinating Committee. Hypertension 2003;
42:1206 –1252.
13. Third Report of the National Cholesterol Education Program (NCEP) Expert Panel
on Detection, Evaluation, and Treatment of High Blood Cholesterol in Adults
(Adult Treatment Panel III): final report. Circulation 2002;106:3143 – 3421.
14. National Institutes of Health, National Heart, Lung and Blood Institute. Clinical
guidelines on the identification, evaluation, and treatment of overweight and obes-
ity in adults: the evidence report. Obes Res 1998;6(Suppl. 2):S51 –S210.
15. Task Force on the Management of ST-Segment Elevation Acute Myocardial Infarc-
tion of the European Society of Cardiology (ESC), Steg PG, James SK, Atar D,
Badano LP, Blömstrom-Lundqvist C, Borger MA, Di Mario C, Dickstein K,
Ducrocq G, Fernandez-Aviles F, Gershlick AH, Giannuzzi P, Halvorsen S,
Huber K, Juni P, Kastrati A, Knuuti J, Lenzen MJ, Mahaffey KW, Valgimigli M, van
’t Hof A, Widimsky P, Zahger D. ESC Guidelines for the management of acute myo-
cardial infarction in patients presenting with ST-segment elevation. Eur Heart J
2012;33:2569 – 2619.
16. Carrick D, Haig C, Rauhalammi S, Ahmed N, Mordi I, McEntegart M, Petrie MC,
Eteiba H, Lindsay M, Watkins S, Hood S, Davie A, Mahrous A, Sattar N,
Welsh P, Tzemos N, Radjenovic A, Ford I, Oldroyd KG, Berry C. Pathophysiology
2764a
of LV remodelling in survivors of STEMI: inflammation, remote myocardium, and
prognosis. JACC Cardiovasc Imaging 2015;8:779–789.
17. van Kranenburg M, Magro M, Thiele H, de Waha S, Eitel I, Cochet A, Cottin Y,
Atar D, Buser P, Wu E, Lee D, Bodi V, Klug G, Metzler B, Delewi R,
Bernhardt P, Rottbauer W, Boersma E, Zijlstra F, van Geuns RJ. Prognostic value
of microvascular obstruction and infarct size, as measured by CMR in STEMI
patients. J Am Coll Cardiol Img 2014;7:930 – 939.
18. Kelly TL, Gilpin E, Ahnve S, Henning H, Ross J Jr. Smoking status at the time of acute
myocardial infarction and subsequent prognosis. Am Heart J 1985;110:535 –541.
19. Gupta T, Kolte D, Khera S, Aronow WS, Palaniswamy C, Mujib M, Jain D, Sule S,
Ahmed A, Iwai S, Eugenio P, Lessner S, Frishman WH, Panza JA, Fonarow GC. Re-
lation of smoking status to outcomes after cardiopulmonary resuscitation for in-
hospital cardiac arrest. Am J Cardiol 2014;114:169 –174.
20. Verouden NJ, Haeck JD, Kuijt WJ, Meuwissen M, Koch KT, Henriques JP, Baan J,
Vis MM, Piek JJ, Tijssen JG, de Winter RJ. Clinical and angiographic predictors of
ST-segment recovery after pPCI. Am J Cardiol 2010;105:1692 –1697.
Downloaded from https://academic.oup.com/eurheartj/article/37/36/2756/2197391 by guest on 20 October 2021
21. Basso C, Corbetti F, Silva C, Abudureheman A, Lacognata C, Cacciavillani L,
Tarantini G, Marra MP, Ramondo A, Thiene G, Iliceto S. Morphologic validation
of reperfused hemorrhagic myocardial infarction by cardiovascular magnetic
resonance. Am J Cardiol 2007;100:1322 –1327.
22. Eitel I, de Waha S, Wöhrle J, Fuernau G, Lurz P, Pauschinger M, Desch S, Schuler G,
Thiele H. Comprehensive prognosis assessment by CMR imaging after ST-segment
elevation myocardial infarction. J Am Coll Cardiol 2014;64:1217 –1226.
23. Motterlini R, Otterbein LE. The therapeutic potential of carbon monoxide. Nat Rev
Drug Discov 2010;9:728–743.
24. Kali A, Tang RL, Kumar A, Min JK, Dharmakumar R. Detection of acute reperfu-
sion myocardial hemorrhage with cardiac MR imaging. Radiology 2013;269:
387 – 395.