Obesity, metabolic syndrome and cardiovascular prognosis: from the Partners coronary computed tomography angiography registry

Atherosclerosis, obesity, metabolic syndrome (MetS), and diabetes mellitus are closely linked and constitute major health problems worldwide. Not only has the prevalence of obesity been increasing to epidemic proportions in the United States, but obesity is also causally related to most of the major cardiovascular (CV) risk factors, including high blood pressure, dyslipidemia and insulin resistance leading to metabolic syndrome and type-2 diabetes mellitus [1]. Due to accelerated systemic atherosclerosis and resultant high cardiovascular event rates among patients with type-2 diabetes, the World Health Organization has called for increased preventive efforts to stem the tide of increasing prevalence of type-2 diabetes, which occurs in association with obesity [2]. Beyond risk factor changes, obesity also increases the risk of future cardiovascular events. Nevertheless, some research has pointed out that certain groups of obese patients may fare better, the so-called “obesity paradox” [3], and likewise have questioned the importance of obesity without metabolic abnormalities (“metabolically healthy obesity”) on the development of future adverse CV events [1].

The metabolic syndrome has been used to identify individuals who have a cluster of risk factors, which together pose a higher prevalence of CAD and increased risk of subsequent clinical cardiovascular disease (CVD). The increased CVD risk appears to be related to the risk factor clustering and insulin resistance associated with the metabolic syndrome rather than simply to obesity [4‐7]. Moreover, individuals with MetS have a higher risk of CAD when coronary artery calcium (CAC) is increased [8].

The independent association between obesity and cardiovascular disease, particularly coronary artery disease remains controversial. A recent meta-analysis found that compared with metabolically healthy normal-weight individuals, obese individuals are at increased risk for adverse long-term CVD events even in the absence of metabolic abnormalities, suggesting that there is no healthy pattern of increased weight [9]. However, previous studies also demonstrated that metabolic health was the main determinant risk of acute myocardial infarction (MI) [1] or CV outcomes, not obesity itself [10].

The presence of coronary artery atherosclerosis is an established marker of future CV risk in symptomatic individuals [11]. However, the association of those findings according to obesity and presence of MetS and their association with clinical outcomes has not been rigorously studied. Thus, in the present study we sought to investigate the interplay of BMI, MetS and CAD detected by coronary computed tomography and their association with future CV events.

Our study has demonstrated that metabolic risk modifies the association of BMI with increasing prevalence, extent, and severity of CAD. Metabolic risk among BMI categories also modified the risk of incident cardiovascular events. These findings suggest that although BMI serves as a marker of CAD risk when considered in isolation, the hazard associated with BMI is mostly mediated by the presence of other metabolic risk factors. Our group has previously demonstrated an incremental increase in CAD burden and future adverse events with incrementally worse metabolic health [7]; the current study extends upon the prior analysis by modeling CV prognosis non-linearly across the full range of BMI. Although most obese individuals have a cluster of CM risk factors, those without this pattern have a significantly lower probability of CAD and future CV events. Therefore, one must carefully evaluate all potential risk factors when considering the presence or absence of an “obesity paradox” since not all obesity phenotypes have a homogenous prevalence of CAD or risk of adverse CHD events.

In prior studies, a variety of authors have suggested evidence of an “obesity paradox,” yet others have discredited this finding as an effect of residual confounding. For example, Lavie et al. [3] have noted that since heart failure is a catabolic state, obese patients have improved survival relative to normal weight, which may simply demonstrate residual confounding due to more advanced disease state. Similar findings of “obesity paradox” have been reported with hypertensive patients, although this may relate to different pathophysiology and co-morbid conditions among the obese versus normal weight with hypertension [3]. Despite such counter-intuitive reports, a large pooled analysis of 19 prospective studies that excluded smokers and those with known cancer or heart disease identified no evidence of obesity paradox [34]. In this study, a J-shaped curve for obesity in association with all-cause death was noted and the healthiest BMI was noted to be 20–24.9 kg/m². Each incremental 5 point increase in BMI was associated with a hazard ratio of 1.31 (95% CI, 1.29–1.33) [34]. Similarly, the Prospective Studies Collaboration identified the healthiest BMI as 22–25 kg/m² for the lowest association with cardiovascular death with a hazard ratio per 5 point increase of 1.32 (95% CI, 1.29–1.36) [35]. An important large study of 13,874 patients undergoing coronary CTA reported greater prevalence, extent and severity of CAD in overweight and obese individuals. BMI was independently associated with a higher risk of non-fatal MI. Although BMI was not associated independently with all-cause death, this study was limited by linear modeling of BMI as a risk factor without consideration of modeling effects of extreme BMI < 20 kg/m² or BMI > 40 kg/m² [36].

An emerging phenotype of “metabolically healthy obesity (MHO)” has been recognized, although the literature to date is not comprehensive. A recent systematic review on the topic of metabolically healthy obesity identified 15 cohort and 5 cross-sectional studies that defined metabolically healthy obesity using either lack of insulin resistance or lack of MetS [37]. This review noted that only two of seven cohort studies that evaluated all-cause death found a significantly increased risk, one of seven evaluating CV deaths, and 3 of 9 evaluating incident CVD. However, 5 of 9 evaluating incident CVD demonstrated a consistent trend for MHO to have an increased clinical risk relative to metabolically healthy normal weight, while only 1 of 9 studies showed no association whatsoever. Thus, although the available studies to date are not numerous, the majority of cohort studies published have not consistently proven that metabolically healthy obesity is not associated with increased adverse clinical outcomes when compared to metabolically healthy normal weight individuals (in contrast to metabolically unhealthy obesity which does convey worse prognosis). However, 5 cross-sectional studies suggested a slight increase in subclinical atherosclerosis among those with MHO, which suggests that with a longer follow-up and larger sample size a small increased risk might be identified even for metabolically healthy obesity. A study looking at progression of plaque with coronary CTA reported that MetS was an independent predictor of progression of coronary artery stenosis or development of vulnerable plaque after accounting for traditional CM risk factors including BMI, HR = 1.47, (95% CI 1.01–2.15, p = 0.045) [38].

Scientists have debated the relative merits of regional adiposity (e.g., abdominal or central obesity) versus absolute weight metrics, such as body mass index. Although most widely applied criteria for MetS diagnosis place greater emphasis on central adiposity, this measure is rarely measured clinically and thus the WHO and others [25, 26], often substitute BMI as a reasonable, albeit imperfect, surrogate. Nevertheless, several studies have demonstrated a relevance of central adiposity. For example, a large screening cross-sectional study of 21,335 middle-aged Korean men identified that abdominal adiposity was associated with CAC, although ultrasound evidence of non-alcoholic fatty liver disease when compared to abdominal obesity, demonstrated an even stronger association for CAC [39]. On the other hand, in a smaller Asian screening cross-sectional study of 3157 subjects who underwent CT, visceral adiposity was associated with CAD in univariable analysis but not by multivariable analysis after adjustment for age, gender, dyslipidemia, diabetes mellitus, and the ratio of visceral to subcutaneous fat [40]. Yet another interesting study stratified 2078 normal weight subjects (18.5 ≤ BMI < 25 kg/m²) who underwent CTA according to percentage of body fat. Individuals with the highest tertile of body fat, even at normal weight (“normal weight obesity”) exhibited increased prevalence of CTA non-calcified coronary plaques (21.6 vs. 14.5%, p = 0.039), more abnormal aortic pulse wave velocity, and increasingly abnormal cardiometabolic risk factors [41]. Thus, although not routinely measured in most clinics, several reports have demonstrated a value of evaluating central adiposity.

Currently over half of American adults are overweight or obese. Furthermore, in contrast to previous decades there is a shift toward a greater proportion of morbid versus milder obesity [3]. Some encouraging recent data suggest a plateau in the prevalence of American obesity, yet the prevalence worldwide has continued to increase in almost every corner of the globe, with certain regions particularly afflicted, such as parts of India [42]. For example, the worldwide prevalence of obesity has doubled from 1980 to 2008, such that over 1.4 billion adults are now overweight and a 0.5 billion are obese. Even more concerning is the crisis of childhood obesity with 43 million preschoolers [43] worldwide overweight, who may begin to have adverse cardiovascular structural and metabolic effects even in childhood and early teen years [44]. Obese individuals are at increased risk of diabetes, heart disease, stroke, sleep apnea, and other chronic conditions and consume increased healthcare resources relative to those of normal weight.

In spite of the debate about an “obesity paradox,” this often represents a comparison to unhealthy thinner patients in an end-stage disease state (as in CHF), residual confounding (such as when smoking is not adequately accounted for), or overly adjusting for co-linear variables associated with obesity (such as dyslipidemia, dysglycemia, or hypertension). Because of the important health effects of obesity as demonstrated by our study and others, for conditions including CAD, obstructive sleep apnea, heart failure, stroke, or death, successful weight loss interventions are of great importance. Although guidelines endorse exercise and healthy as first line recommendations, long term success through lifestyle interventions remain elusive for many patients. Recently, the 3 year follow-up results of the STAMPEDE trial, which compared bariatric surgery to non-surgical intervention for obese patients with uncontrolled type 2 diabetes mellitus and noted persisting improvements in metabolic health and weight loss after for surgery [45]. Thus, identifying patients with MetS, which may be treatable and in some cases reversible through lifestyle or surgical intervention, could benefit lifetime cardiovascular risk.

Contemporary interest to improve risk stratification of patients with either MetS or diabetes mellitus have led to limited studies of screening populations. For example, although major guidelines define diabetes mellitus as high-risk for CAD events, one study screened 98 asymptomatic patients with diabetes using CAC, CTA, and carotid ultrasound and identified 55 (56%) who had no detectable CAD and could potentially be re-classified from high to low CV risk. Sixteen subjects (including three with CAC = 0) were found to have obstructive CAD by CTA, but no clinical outcomes were reported for this cross-sectional study [46]. Similarly, Ryu et al. screened 755 patients with MetS by CTA and identified an increasing extent and burden of CAD according to the number of MetS risk factors, among which abdominal obesity and hypertension had the strongest effect [47]. A small study of 39 Japanese men also noted that those with versus without MetS had higher prevalence of CAD (31.3 versus 4.3%, p = 0.033) and lower serum adiponectin levels (4.5 ± 0.6 versus 6.4 ± 0.6 μg/mL, p = 0.014) [48]. Another study that screened 1000 asymptomatic Korean patients with diabetes found that 78% had no detectable CAD by CTA. Although 22% had some plaque that may modify preventive clinical decision making, the incidence of downstream adverse clinical events was low over 17 months follow-up and consisted entirely of 1 unstable angina and 14 coronary revascularizations [49]. The FACTOR-64 study, which randomized asymptomatic diabetics to screening by CTA or standard therapy also failed to show any significant difference in downstream events, owing to the unexpectedly low number of events and the use of preventive therapies in both arms (>70% on statins) [50]. Thus, although screening can further characterize the otherwise homogenous risk categorization of patients with diabetes from simply “high risk,” the impact of such testing upon the otherwise low clinical event incidence among asymptomatic patients remains uncertain.

Notwithstanding these results, our study must be interpreted in the context of its inherent limitations. First, a retrospective design in a tertiary referral center may result in selection bias. Similarly, the patients were clinically referred for a coronary CTA, which may increase the prevalence of CAD, and clinical risk compared to the general population. Also, we did not have available measures of central adiposity, such as waist circumference. In spite of these limitations, we demonstrate that although obesity is associated with CAD prevalence, extent, severity and prognosis much of the risk can be explained by cardiometabolic health within obesity subcategories.

Compared to individuals with normal BMI (20–24.9 kg/m²), there was an increased prevalence, extent, and severity of CAD for those with BMI > 25 kg/m². BMI by univariable analysis demonstrated a J-shaped curve with prognosis, though this association was significantly modified after adjusting for metabolic risk factors. BMI was a marker for MetS, which was more significantly associated with prognosis.