Association Between Smoking Status and Outcomes in ST-Segment Elevation and Non-ST-Segment Elevation Myocardial Infarction Patients Undergoing Percutaneous Coronary Intervention


 Smoking is one of the leading risk factors for cardiovascular diseases, including ischemic heart disease and hypertension. However, in acute myocardial infarction (AMI) patients, smoking has been associated with better clinical outcomes, a phenomenon termed the “smoker’s paradox.” Given the known detrimental effects of smoking on the cardiovascular system, it has been proposed that the beneficial effects of smoking on outcomes is due to age differences between smokers and non-smokers and is therefore a smoker’s pseudoparadox. The aim of this study was to evaluate the association between smoking status and clinical outcomes in ST-segment elevation (STEMI) and non-STEMI (NSTEMI) patients treated by percutaneous coronary intervention (PCI), using a national multi-ethnic Asian registry. In unadjusted analyses, current smokers had better clinical outcomes following STEMI and NSTEMI. However, after adjusting for age, the protective effect of smoking was lost, confirming a smoker’s pseudoparadox. Interestingly, although current smokers had increased risk for recurrent MI within 1 year after PCI in both STEMI and NSTEMI patients, there was no increase in mortality. In summary, we confirm the existence of a smoker’s pseudoparadox in a multi-ethnic Asian cohort of STEMI and NSTEMI patients and report increased risk of recurrent MI, but not mortality, in smokers.


Introduction
Cardiovascular disease (CVD) is the leading cause of death worldwide accounting for about 30% of deaths 1 . CVD is an imminent health threat as the number of global deaths from CVD has been rapidly increasing due to the aging population [2][3][4] . Among the various CVDs, acute myocardial infarction (AMI) results in a signi cant 30-day mortality of between 3-14% 5 .
Smoking is one of the strongest risk factors for cardiovascular disease including AMI 6 . However, paradoxically, a number of clinical studies have documented that smokers might have a better prognosis following AMI events as compared to non-smokers in both patients with ST-segment elevation (STEMI) and non-ST-segment elevation myocardial infarction (NSTEMI) 7,8 . To explain this seemingly counterintuitive bene cial effect of smoking on AMI, a number of hypotheses have been proposed; 1) preconditioning of cardiomyocytes, 2) cellular reprograming from necrosis to apoptosis 9 , and 3) reduced impact of platelets 10 . Of these, the preconditioning phenomenon has been most supported due to the established role of preconditioning on the regeneration of cardiomyocytes 11 . The size of the infarct is strongly associated with patient outcomes, such as all-cause mortality and hospitalization after STEMI 12 .
Based on this observation and the possibility that cigarette smoking may mimic a transient preconditioning stimulus, it has been proposed that smoking leads to a better outcome via preconditioning in cardiomyocytes and, therefore, decreases the size of infarction 13,14 .
The results of previously published observational studies demonstrating bene cial effects of smoking and post-AMI outcomes have been challenged 7,8 . In both of these reports, the average age of smokers was about 9 years younger than that of non-smokers. Given the fact that aging is one of the risk factors of AMI and the age difference was not adjusted in the studies, it is possible that the observed "protection" was presumably due to younger age of the smokers rather than smoking itself 15 . Therefore, the role of cigarette smoking on the outcomes of STEMI and NSTEMI patient remains controversial and has implications on public health. As such, to further clarify this issue, a national population-based multiethnic Asian acute myocardial infarction registry was used to evaluate the associations between smoking status and clinical outcomes in STEMI and NSTEMI patients treated by percutaneous coronary intervention.

Data Collection
For this study, data from a national registry, the Singapore Myocardial Infarction Registry (SMIR), were utilized. The institutional review board granted an exemption for conducting this study without need for informed consent (SingHealth Centralised Institutional Review Board Reference No: 2016/2480) as this study involved analysis of a dataset without identi ers. The research was conducted in accordance with the Declaration of Helsinki. The SMIR collects epidemiology and clinical data of AMI cases diagnosed in the public and private hospitals in Singapore, in addition to certi ed out-of-hospital AMI deaths [16][17][18] . Reporting of AMI cases to this registry is mandated by law in accordance with the National Registry of Diseases Act. The registry data included patient medical claim listings, hospital in-patient discharge summaries and cardiac biomarker listings from hospital laboratories. To identify the cases of AMI, the International Classi cation of Diseases, Ninth Revision, Clinical Modi cation (ICD-9-CM) code 410 was used for the data collected between 2007 and 2011. For any data that were collected from 2012 onwards, ICD-10 (Australian Modi cation) codes I21 and I22 were used to identify AMI. The AMI cases were further classi ed into STEMI and NSTEMI based on diagnosis documented by the clinicians in the medical records. The following criteria were used to de ne STEMI: 1) chest pain for 20 minutes, 2) signi cant STsegment elevation, and 3) positive for cardiac biomarkers. To ensure that the data were captured in an accurate and consistent manner across all hospitals over the years, annual internal audit was performed. Among the STEMI and NSTEMI cases in January 2007 to December 2015, information on the patients who received PCI was extracted and utilized for analysis. Only patients who received PCI were included as the patients with a Type 1 MI were the focus of this study.

Clinical Outcomes
The primary clinical outcome of this study was all-cause mortality at 1-year post-presentation. Mortality data were obtained from the Death Registry of Ministry of Home Affairs and were merged with the SMIR data. Secondary outcomes included all-cause mortality at 30 days post-presentation, and the rst episode of MI occurring within 1 year after the index PCI. The analyses were strati ed by the type of AMI (STEMI and NSTEMI) and patients were compared based on their smoking status (never smoked, former smoker, and current smoker). Smoking data were self-reported by the patients or their family based on documentation in the medical records.

Statistical Analysis
Categorical variables were expressed as frequency and percentages, while continuous variables were expressed as median and interquartile range. Missing data were excluded from the analyses through case deletion without imputation to maintain data in its original form. Univariable and multivariable cox regression were performed to determine the hazards ratios (HR) of having the primary and secondary outcomes. Speci cally for the secondary outcome of recurrent AMI, competing risk from non-AMI deaths was adjusted using the Fine-Gray proportional hazards model 19 .
This study did not require informed consent according to the exemption granted by the institutional review board as this was a study using deidenti ed data (SingHealth CIRB Reference No: 2016/2480).
This study followed the principles of the Declaration of Helsinki. The statistician of the study had access to anonymized individual data, while the other co-authors had access to the analyzed aggregated data.
All analyses were done using Stata SE Version 13 (StataCorp. 2013. Stata Statistical Software: Release 13. College Station, TX: StataCorp LP). All statistical tests were 2-tailed and results were deemed to be statistically signi cant if p<0.05.

Study Population
A total of 21,261 AMI patients (12,307 STEMI and 8,954 NSTEMI) who received PCI from January 2007 to December 2015 were included in this study. The SMIR population was multi-ethnic with about three-fths being Chinese, one-fth Malays and one-fth Indian (Tables 1 and 2). The STEMI group comprised 4,549 (37%) never smoked, 1,703 (14%) former smokers, and 6,055 (49%) current smokers ( Table 1). The median age of patients who never smoked or were former smokers were similar, but the median age of current smokers was signi cantly less by 8 years. The NSTEMI group comprised 3,902 (44%) never smoked, 1,766 (20%) former smokers, and 3,286 (37%) current smokers ( Table 2). The median age of current smokers with NSTEMI patients was about 9 years less than the never smoked and former smokers. For both STEMI and NSTEMI, former and current smokers were predominantly male (> 90%), when compared to patients who have never smoked (50-60%) (Tables 1 and 2). The female patients with STEMI had a worse 1-year-mortality (HR 1.33, 95% CI 1.05-1.69), suggesting that gender was a potential confounding factor. Therefore, gender was included for adjustment when the multivariate regression was performed.  In the STEMI group, the unadjusted HRs for death within 1 year were signi cantly lower for current smokers (HR 0.50), when compared to never smokers (Table 3). For death within 30 days, the unadjusted HRs were signi cantly lower in both former (HR 0.68) and current (HR 0.50) smokers, when compared to never smokers (Table 3). However, after adjustment for potential confounders, this protective effect was no longer observed (Table 3). Similar 30-day and 1-year mortality results were observed when we strati ed the analysis by ethnicity (Table 3). Notably, although the unadjusted HRs for recurrent MI within 1 year did not differ signi cantly according to smoking status, the adjusted HR for current smokers (HR 1.39) was signi cantly higher than never smokers, especially among Chinese patients, suggesting that current smokers have increased risk of recurrent MI following STEMI.

Discussion
In this national registry-based study of a multi-ethnic cohort of AMI patients treated by PCI, we found that former and current smokers had a decreased unadjusted HR for both 30-day and 1-year mortality, when compared to never smokers, but this protective effect of smoking was not present after adjustment of potential confounding factors, suggesting the existence of a smokers' pseudoparadox on mortality for both STEMI and NSTEMI. We found an increased adjusted HR for recurrent MI within 1 year among current smokers in the STEMI group and the Chinese STEMI sub-group, when compared to never smokers. Similarly, we found that the adjusted risk for recurrent MI within 1 year were increased among both former and current smokers in the NSTEMI group, when compared to never smokers.
The smoker's paradox was rst described in 1968 by Weinblatt et al. as an unexpected result because the smokers had a lower 1 month mortality after their MI events as compared to non-smokers in the report 20 . This paradoxical result was observed in another independent study by Tamsin Lisa et al. where the smokers had a lower prevalence for AMI and other CVDs, such as hypertension, congestive heart failure and angina pectoris when logistic regression was performed 21 . In this paper, the authors noted that smokers were about 10 years younger than the non-smokers and this paradoxical result was presumably due to the age difference, given that age is one of the most important factors for developing CVDs and predisposes to AMI events via generation of reactive oxygen species (ROS), cellular senescence, and epigenetic changes 22 .
Many of the former studies which attempted to elucidate the association of smoking with clinical outcomes of AMI had intrinsic limitations. For example, one of the earlier studies which included clinical data from the three hospitals reported the protective effect of smoking on the outcomes of acute STEMI 8 .
However, this study did not have age adjustment despite the signi cant difference in age between smokers and non-smokers which makes data interpretation di cult. Similarly, in another study where national Malay data were utilized, the study reported a positive correlation between smoking and better outcomes for both STEMI and NSTEMI patients 7 . This paper was one of the few studies which utilized the national dataset and reported both STEMI and NSTEMI separately. However, this study again did not include adjustment for age even though there was about 9 years of difference in age between smokers and non-smokers, suggesting that the lack of age adjustment could possibly impact the data interpretation. Therefore, the abovementioned epidemiological studies which demonstrated the protective association of smoking with the outcomes of AMI may be confounded by the lack of statistical adjustment for age.
Two recent papers have reported the detrimental effect of smoking on the AMI outcomes. In the national dataset of Chinese cohort, Gao et al. found that without the proper adjustments for age and the number of cigarettes smoked, the smoking history was associated with a better outcome in the MI patients 23 .
However, when the number of cigarettes smoked and age are considered, the same dataset showed a signi cantly negative correlation of smoking with clinical outcomes in the MI patients. Similarly, independent groups reported that smoking was associated with an increased risk of all-cause mortality and heart failure from meta-analysis of PCI trials 24,25 . However, these two studies had some limitations as the former study did not subclassify AMI patients into STEMI and NSTEMI and was conducted only in patients with Chinese ethnicity. Similarly, the latter study analyzed STEMI patients who participated in clinical trials, suggesting that the patient population might not be generalizable. Our present report contributes further to the eld by elucidating the detrimental effect of smoking on clinical outcomes of multi-ethnic STEMI and NSTEMI patients as our study utilized data of a national registry with appropriate statistical adjustment for potential confounders.
In our dataset, it was consistently observed that both STEMI and NSTEMI smokers were at higher risk of recurrent MI within 1 year. This result is consistent with the former reports by Redfors and Gao et al as they also reported an increased risk of recurrent MI by smoking 23,25 . Interestingly, despite the increase in recurrent MI, an increased all-cause mortality was not observed in our data analysis. This was an unexpected result as other groups observed both increased recurrent MI and all-cause mortality 23 and the recurrent MI is one of the strongest predictors for an increased all-cause morality 26,27 . This phenomenon could be due to other factors, such as advances in the health care system or the short travel time between the patients' houses and the hospitals in Singapore due to urbanization which would lead to better mortality [28][29][30][31][32] . Another interesting result we found was that the former Malay smokers had a better clinical outcome in terms of death within 30 days after NSTEMI events with a hazard ratio of 0.30. This result may indicate that for this speci c group, there were confounding factors other than age.
One of the notable ndings of our study was that the detrimental association between smoking. The clinical outcomes were observed in the overall and Chinese group for both STEMI and NSTEMI, but not in the Malay and Indian groups (Tables 3 and 4). This result suggests that the differential effect of smoking on the prognosis of STEMI and NSTEMI may be in uenced by ethnicity. Similarly, other groups previously reported that the risk of smoking on AMI events was dependent on the types of small nucleotide polymorphism (SNP) on certain genes, such as Paraoxonase (PON1) and rs1122608 33,34 . In other words, these SNPs may determine whether smoking increases the susceptibility of subjects to AMI. The SNPs that are be associated with the worse prognosis by smoking in our cohort and the mechanisms by which smoking interacts with these SNPs need to be experimentally identi ed and validated in the future.
There is indirect evidence suggesting the detrimental role of smoking on AMI. For example, cases of AMI signi cantly decreased after the introduction of national smoking bans in multiple countries, suggesting that smoking increases the risk of AMI [35][36][37] . Moreover, the percentage of smokers among AMI patients was signi cantly greater as compared to non-AMI patients 38 . Considering the fact that cigarette smoking is attributable to 15% of the CVD cases and mortality 39 and increases the risk of MI, smoking should be strictly avoided. Additionally, there are still unanswered questions on whether second-hand smoking or cigarette alternatives like e-cigarettes also lead to poorer outcomes in MI patients.
The smoker's paradox has been observed in a number of diseases, such as idiopathic pulmonary brosis (IPF) 40 , ischemic stroke 41-43 , Alzheimer's disease [44][45][46] , and Parkinson's disease 40,47−49 . Smoking is associated with lower incidence of IPF 50 . This favorable association between smoking and IPF is generally accepted as pseudo-protection of smoking caused by tolerance issues; non-smokers with asymptomatic lung diseases who previously attempted to smoke could not tolerate smoking and retrained themselves from this habit as their lungs are more sensitive to smokes. This could lead to a lower percentage of smokers among IPF patients compared with non-IPF 50 . Similarly, the positive association between smoking and the prognosis of the patients with ischemic stroke is considered pseudo-protection due to the differences in clinical baselines, such as age [41][42][43] . Interestingly, the clinical papers which reported the protective correlation of smoking with Parkinson' and Alzheimer's disease were case-control studies where the cases were age, gender, and race matched, suggesting that the protection was not due to the baseline mismatch 40,44−49 . It has been proposed that the positive association is the result of the neuro-protective effect of nicotine, but this hypothesis requires further experimental validation [51][52][53] . However, this cardioprotective effect of nicotine on AMI was not observed in the patients with nicotine replacement therapy as reported in multiple independent studies 22,54,55 .
A limitation of the study is that the SMIR database does not contain information on the duration of smoking and the number of cigarettes the smokers consume. Additionally, the SMIR does not include the information on how long it has been after the former smokers quit their smoking behavior. For that reason, patients who identi ed as former smokers were classi ed as such unlike the WHO de nition which requires 12 months of cessation to qualify one as a former smoker 56 . Hence, the results from our study may not be directly comparable with other studies due to the different de nition. Furthermore, due to the lack of information on the number of cigarettes smoked, we could not examine for the presence of any potential dose-response relationship between smoking and outcomes.

Conclusion
In summary, we found that smokers seemingly had better clinical outcomes (30-day and 1-year mortalities) after STEMI or NSTEMI. However, upon adjustment, the seemingly bene cial effects of smoking on mortality disappeared and the risk of recurrent MI within 1-year was signi cant higher in STEMI and NSTEMI smokers, con rming the presence of a smokers' pseudoparadox for mortality. This data demonstrates that the previously reported protective effect of smoking was actually pseudoprotective and smoking worsens clinical outcomes in both STEMI and NSTEMI patients. Declarations