No Smoke without Fire?

This article refers to ‘Influence of cigarette smoking on cardiac biomarkers: the Atherosclerosis Risk in Communities (ARIC) Study’, by W. Nadruz et al., published in this issue on pages XXX–XXX. Cigarette smoking is undeniably a major cause of preventable death and, as a powerful risk factor for coronary heart disease, heart failure and cardiovascular death, much of the morbidity and mortality attributable to smoking is cardiovascular in nature.1, 2 Accelerated atheroma is a major pathway for smoking-related cardiovascular harm.3 However, smoking-induced abnormalities in cardiac and vascular structure and function may long precede overt symptomatic occlusive macrovascular disease suggesting the presence of other important mechanisms.4-6 Nadruz et al.7 have sought to add to current understanding by inspecting the relationship between smoking and two markers of cardiac duress, plasma N-terminal pro-brain natriuretic peptide (NT-proBNP) and high-sensitivity troponin T (hsTnT), in the large Atherosclerosis Risk in Communities (ARIC) study. Data from 9649 ARIC participants free of prevalent CAD or heart failure were considered in cross-sectional analyses and, of these, a subgroup of 3151 provided longitudinal data from follow-up over 15 years. Although both markers were significantly related to cumulative pack years of smoking, on cross-sectional analysis raw unadjusted data revealed no difference between levels of either marker when comparing current with never-smokers. However, with adjustment for age and gender NT-proBNP levels were significantly higher in current smokers and remained so with additional adjustment for a further 10 relevant covariates. This illustrates the power of a large, well-annotated cohort in enabling dissection of genuine relationships in the presence of multiple potentially obscuring confounders. In contrast to NT-proBNP, hsTnT remained apparently unrelated to smoking status despite similar multivariable analysis. The authors hypothesized this apparent contradiction might reflect a close association of elevated hsTnT with death thus preventing any obvious association of hsTnT with smoking status at any single point in time. Effectively, the composition of surviving current smokers may be biased towards those smokers constitutionally equipped to tolerate smoking without cardiac injury and therefore without elevation of hsTnT. This theory finds support from the follow-up phase of the report. Smokers with elevated hsTnT (i.e. >14 pg/mL) incurred by far the highest cardiovascular mortality and morbidity in the subsequent 15 years. Rates of death and heart failure in current smokers with raised hsTnT at baseline were both 4-fold that in current smokers without elevated hsTnT and rates of 9- and 11-fold, respectively, were observed in never smokers without elevated troponin. In contrast the corresponding ratios for NT-proBNP were twofold and threefold and fourfold and fivefold, respectively. Ominously, current smokers at baseline were more likely than former or never-smokers to exhibit increased marker levels at follow up. The report helps integrate two well-known relationships: the clear increased risks associated with smoking and the well-proven prognostic power of both NT-proBNP and hsTnT for adverse cardiovascular clinical outcomes.1-3, 8-10 The size of the ARIC sample, together with the details of its annotation and the duration of follow-up have allowed explanation of a misleading apparent lack of relationship between smoking and hsTnT on inspection of cross-sectional data, as observed in both the current report and previous reports (including a previous report from ARIC).8, 9, 11 Does this help with understanding the mechanisms underpinning smoking-related harm? The authors invoke two concepts: that hsTnT reflects myocardial injury and cell death while NT-proBNP reports on myocardial haemodynamic load. At face value this might point to smoking-related myocyte destruction partly underpinning the sinister association between raised hsTnT and mortality. Loading of the ventricle, reflected by NT-proBNP, may be present but less important. However, this leaves the underlying mechanism of myocyte injury unexplained. Does elevation of hsTnT reflect acceleration of otherwise occult epicardial coronary disease, microvascular dysfunction stemming from smoking-induced endothelial dysfunction, direct toxic actions of smoking on the myocardium, or some combination of these candidate effects? Data from studies of this kind reveal associations but cannot provide full mechanistic insights and cannot provide definitive evidence of cause and effect. Interpretation of modestly elevated marker levels in asymptomatic ambulant populations, as observed in this study, is not straight-forward. By definition, the late generation highly sensitive cardiac troponin assays allow detection of circulating cardiac troponin T or I in most normal people. Therefore, the presence of cardiac troponin in plasma is normal and much of it does not reflect pathological processes. At what level plasma troponins begin to be influenced more by abnormal myocyte drop-out than by the ill-defined normal turnover processes dictating plasma troponin in health is poorly understood. Levels above the 99 percentile of the reference range in apparently healthy people are suggestive of abnormality but simply applying this threshold offers no information about mechanism and, in any case, by definition 1% of healthy people will exhibit plasma troponin above this threshold.12 Similarly, inter-subject variation in, and marginal elevation of, plasma NT-proBNP concentrations in asymptomatic ambulant subjects is not purely a reflection of cardiac haemodynamic load. The signal to noise ratio when relating transmural cardiac distending pressures to plasma NT-proBNP is far lower in this epidemiological setting than in acutely symptomatic populations presenting to the emergency department. We know sex, body mass index, kidney function, age, and genetic variants in the BNP gene all exert significant effects upon plasma NT-proBNP at a population level.13-15 The relative contributions of these factors to circulating NT-proBNP will be higher in non-acute settings. In addition, BNP/NT-proBNP synthesis and release can be triggered in response to ischaemia, neurohormones (endothelin 1 and angiotensin 2), and cytokines in the absence of elevated intracardiac pressures. The relevance of these (and other as yet unknown influences) must be considered rather than uncritically ascribing all variations in plasma natriuretic peptides within community-based populations to differences in cardiac haemodynamic load. Evidence of toxic effects from smoking, potentially independent of acceleration of macrovascular atheromatous disease, comprises data from preclinical models and observations in asymptomatic human populations. A recent review summarizes preclinical data on candidate mechanisms for the direct cardiotoxicity of cigarette smoke.6 In rats, chronic exposure to cigarette smoke leads to enlargement of left-sided cardiac chambers, myocardial hypertrophy, and reductions in left ventricular function in the absence of occlusive coronary artery disease. Ultrastructural changes include myofilament disorganization and loss and mitochondrial derangement. Impairment of cardiac structure and function induced by experimental myocardial infarction or ventricular overload is exacerbated by cigarette smoke. The underlying mechanisms appear to include increased generation of reactive oxygen species and activation of cytokine-mediated pro-inflammatory pathways.6 The extent to which findings in animal models can be extrapolated to human smokers is uncertain. With respect to human data, Rosen et al.4 reported smoking to be related to reduced regional left ventricular function in asymptomatic subjects participating in the Multi-Ethnic Study of Atherosclerosis (MESA). In 1184 subjects (aged 45–84 years) undergoing tagged cardiac magnetic resonance scanning, smoking was found to be independently related to reduced peak systolic circumferential strain of left ventricular segments corresponding to territories perfused by left anterior descending and right coronary arteries. A dose–response relationship to cumulative pack-years of smoking was noted. Dalen et al.5 reported echocardiographic left ventricular strain in 1266 participants free of hypertension, diabetes, and known cardiovascular disease. Several risk factors including smoking were found to be related to subtle impairment of cardiac function. Smoking was independently related to reduced left ventricular strain specifically in women and to reduced right ventricular function in the cohort as a whole. These cohort-based data certainly indicate that smoking is associated with differences in cardiac function long before overt symptomatic disease evolves. However, the same uncertainty with respect to underlying mechanisms inherent in the current report from Nadruz et al.7 applies equally to these previous reports. Are these findings indicative of early ischaemic damage or are other mechanisms involved? Neither Rosen et al.4 nor Dalen et al.5 report any data on circulating biomarkers. Reversibility of smoking-induced changes in biomarker levels might support a dynamic mechanism rather than permanent structural ischaemic injury. Otsuka et al.16 studied a worksite population of 969 men aged 44 ± 6 years and found a subtle elevation in plasma NT-proBNP in current compared with never smokers (21.7 ± 2.3 pg/ml vs. 17.9 ± 2.1 pg/ml, P < 0.001). This modest but significant intergroup difference persisted after adjustment for age, obesity, heart rate, hypertension, dyslipidaemia, fasting glucose, diabetes, left ventricular hypertrophy, estimated glomerular filtration rate, alcohol intake, and exercise levels. Current smokers were three times more likely to have elevated plasma NT-proBNP (designated in this paper as >54.5 pg/mL) than never smokers. Interestingly, former smokers had lower levels of elevated NT-proBNP than current smokers and there was a modest negative correlation between plasma NT-proBNP and duration of smoking cessation. Cardiac troponin levels were not reported. These findings are suggestive of, but not proof of, reversibility of marker elevations with smoking cessation. The study lacked serial measurements and was not accompanied by any data on cardiac structure and function. Series reporting elevated cardiac biomarkers without smoking data and other reports of smoking-related impairment of cardiac function in asymptomatic smokers without biomarker data—all without mechanistic information—provide a rich field of enquiry for further research. It would be informative to gather concurrent and serial cardiac imaging and biomarker data at varying intervals before, during, and after initiation and later cessation of smoking in subjects with and without cardiac disease. The addition of information on the epicardial coronary arteries would be valuable. The reservoir of patients (smokers and non-smokers) passing through cardiac catheter laboratories around the world who are found to have angiographically normal coronary arteries may constitute a source of appropriate research participants. These resources would allow the design and execution of studies for ascertainment and interrogation of concurrent smoking-induced changes in cardiac function and biomarker release, their inter-relationships and reversibility, and their association with, or independence from, fixed epicardial coronary artery disease. At the level of clinical application it already seems clear that elevation of hsTnT identifies smokers at especially high cardiovascular risk, which may inform counselling and smoking cessation strategies. Conflict of interest: none declared.