Abstract
Context:
Although the health risks of obesity compared to normal weight have been well studied, the cumulative risk associated with chronic obesity remains unknown. Specifically, debate continues about the importance of recommending weight loss for those with metabolically healthy obesity.
Objective:
We hypothesized that relatively greater severity and longer duration of obesity are associated with greater incident metabolic syndrome.
Design, Setting, Participants, and Measures:
Using repeated measures logistic regression with random effects, we investigated the association of time-varying obesity severity and duration with incident metabolic syndrome in 2,748 Multi-Ethnic Study of Atherosclerosis participants with obesity (body mass index ≥30 kg/m2) at any visit. Obesity duration was defined as the cumulative number of visits with measured obesity and obesity severity by the World Health Organization levels I–III based on body mass index. Metabolic syndrome was defined using Adult Treatment Panel III criteria modified to exclude waist circumference.
Results:
Higher obesity severity (level II odds ratio [OR], 1.32 [95% confidence interval, 1.09–1.60]; level III OR, 1.63 [1.25–2.14] vs level I) and duration (by number of visits: two visits OR, 4.43 [3.54–5.53]; three visits OR, 5.29 [4.21–6.63]; four visits OR, 5.73 [4.52–7.27]; five visits OR, 6.15 [4.19–9.03] vs one visit duration of obesity) were both associated with a higher odds of incident metabolic syndrome.
Conclusion:
Both duration and severity of obesity are positively associated with incident metabolic syndrome, suggesting that metabolically healthy obesity is a transient state in the pathway to cardiometabolic disease. Weight loss should be recommended to all individuals with obesity, including those who are currently defined as metabolically healthy.
Obesity severity and duration are associated with incident metabolic syndrome, suggesting metabolically healthy obesity is transient and additional exposure to obesity leads to metabolic syndrome.
Increasing interest in discriminating between different levels of obesity-related cardiometabolic risk has led to the concept of metabolically healthy obesity. Although definitions of metabolically healthy obesity vary in the literature, it is commonly defined as obesity in the absence of metabolic syndrome. While there is a growing body of research about this group, it remains unclear how to differentiate those with persistent metabolically health obesity from those who will later develop the comorbidities traditionally associated with excess adiposity. Although several studies have shown that those who are obese but metabolically healthy have intermediate risk of type 2 diabetes (1), cardiovascular disease, and mortality compared to other obese subgroups (2–5), the question of whether this group is really “healthy” is pressing but undetermined. While evidence consistently suggests that higher obesity severity is associated with increased risk for diabetes and cardiovascular disease even within the metabolically healthy obesity category (6, 7), almost no studies have accounted for the conflation of severity with duration or determined whether those with longer duration of obesity are more likely to transition to metabolic syndrome. Answering these particular research questions is important because they provide insight into the correct approach for responding to metabolically healthy obesity, both clinically as well as from a public health perspective.
Additional evidence that metabolically healthy obesity may not be maintainable comes from several studies indicating that the prevalence of metabolically healthy obesity declines with age (2, 8, 9). To our knowledge, only one study provides evidence that metabolically healthy obesity is a temporally intermediate stage on the pathway to cardiometabolic risk (2), and only one study reports the relationship of obesity severity and duration with cardiovascular risk in individuals who are defined as having metabolically healthy obesity (10).
Past studies have indicated that increased duration of obesity leads to increased cardiometabolic risks (10–15), but almost none has discriminated between the risks from obesity duration and severity (10). This is of particular interest with regards to metabolically healthy obesity because the length and extent of obesity progression may have important implications for the metabolic health of individuals at different stages along this pathway to cardiometabolic dysfunction. Within this framework, we hypothesized that relatively greater severity and longer duration of obesity will be associated with greater incident metabolic syndrome in the Multi-Ethnic Study of Atherosclerosis (MESA). In contrast to the idea that those with metabolically healthy obesity are somehow resistant to the negative health effects of obesity, evidence for this hypothesis would reinforce the understanding that cumulative exposure to obesity determines the progression from metabolically healthy obesity to metabolic syndrome.
Materials and Methods
Study population
MESA is a longitudinal cohort study started in 2000 with the recruitment of 6814 participants from diverse racial/ethnic backgrounds aged 45 and older and free from cardiovascular disease from six sites across the United States (16). Participants were actively followed every 2 years for five visits. Inclusion for the present analysis was determined by obesity status and the goal of conducting analysis separately for prevalent and incident metabolic syndrome. Inclusion criteria for our primary analysis included having obesity measured at any point during the MESA follow-up. A full description of participant entry and exit into analytic subgroups is explained in the statistical analyses section. Age, sex, race/ethnicity (Caucasian, Asian, African American, and Hispanic), income, education, and obesity at age 40 were collected by self-report at baseline. Institutional review board approval was provided by multiple institutional review boards participating in the MESA study, and all participants provided written informed consent.
Measurement of obesity severity and duration
Body mass index (BMI) was assessed at every visit in MESA and calculated using height and weight measured twice using a standardized protocol and averaged (16). We defined obesity as BMI greater than 30 kg/m2. We defined obesity severity both continuously, using BMI and categorically using the World Health Organization BMI cut-points (level I, 30.00–34.99 kg/m2; level II, 35.00–39.99 kg/m2; and level III: ≥40.00 kg/m2) (17). We defined obesity duration as the cumulative number of visits with obesity since the visit of first measured obesity in MESA. For example, if a participant is obese at visits 2, 3, and 5, but not at visits 1 and 4, then at visit 5 their final obesity duration would be three visits of obesity. For this study, we analyzed all obesity variables as time varying.
Definition of metabolic syndrome
We used modified Adult Treatment Panel III criteria to define metabolic syndrome (18) as two or more of the following criteria: 1) high triglycerides: triglyceride level greater than 150 mg/dl (1.6935 mmol/liter); 2) low high-density lipoprotein (HDL) cholesterol: HDL cholesterol less than 40 mg/dl (1.0344 mmol/liter) in men and less than 50 mg/dl (1.293 mmol/liter) in women; 3) high blood pressure (BP): systolic BP greater than 130 mm Hg or diastolic BP greater than 85 mm Hg or use of BP medications; 4) high fasting glucose: Fasting glucose higher than 110 mg/dl (6.105 mmol/liter) or diagnosis of diabetes defined by the 2003 American Diabetes Association fasting criteria algorithm.
As secondary outcomes, we also assessed dichotomous individual metabolic syndrome components listed previously.
Statistical analysis
We described the participant characteristics at baseline by calculating mean and SE by metabolic syndrome status across follow-up. We determined the prevalence of metabolic syndrome by obesity severity and duration for each MESA visit. We estimated the association of overall obesity compared to normal weight with prevalent and incident metabolic syndrome.
To investigate the association between cumulative exposure to obesity and metabolic syndrome in participants with metabolically healthy obesity, we started by excluding participants who never had measured obesity during follow-up or who had metabolic syndrome before measured obesity (n = 4066). We used a single logistic regression model, adjusted for age, sex, and race/ethnicity, to assess the relationship of highest obesity severity recorded at any visit and total obesity duration across all follow-up with metabolic syndrome measured at any visit.
For our primary analysis, we made use of all the data available in MESA in repeated measures analysis for those participants with metabolically healthy obesity at any point during MESA follow-up. MESA participants entered the risk set for repeated measures analysis at the first visit where they had measured obesity. We used repeated measures logistic models with time-varying exposures to estimate the association of obesity severity and duration with prevalent and incident metabolic syndrome separately. For analyses of incident metabolic syndrome, we excluded all participants with measured metabolic syndrome at visit 1 (n = 1474). To account for correlation between repeated measures, we used generalized estimating equations for binary responses with a logit link for the population-averaged estimates. Incorporating random effects to account for individual heterogeneity, we used mixed effects logistic models. Based on exploration of the longitudinal structure of the data, we used an exchangeable correlation structure and because the estimates for time were not significant, except for visit 5, we included a dummy variable for visit 5 in our models. We formally tested for an interaction between obesity duration and severity. We adjusted for confounding by age, sex, and race/ethnicity and tested whether the inclusion of random effects significantly improved the model fit using likelihood ratios. Based on prior literature and our a priori hypotheses, we formally tested for effect modification by sex (female vs male), race/ethnicity (Asian, African American, and Hispanic vs Caucasian), and age (<70 vs ≥70 years). Finally, we conducted our primary analysis separately for each of the metabolic syndrome components, as well as for high low-density lipoprotein (LDL). All analyses were conducted using Stata 11 (19).
Sensitivity analysis
We determined whether our results were sensitive to different definitions of obesity exposure. Some participants with reported obesity were reported as nonobese at subsequent visits (n = 710), so we also tested whether intermittent obesity was associated with a better metabolic health profile compared to continuous obesity duration in subgroup analysis. Similarly, some participants exhibited an intermittent pattern of metabolic syndrome across follow-up (n = 2127), so we investigated whether intermittent obesity helped explain intermittent metabolic syndrome. To account for misclassification of obesity duration, we investigated whether our results differed when the sample was restricted to those with incident obesity; that is, those who developed obesity after visit 1. We also tested whether self-reported obesity status at age 40, as a potential indicator of duration misclassification, influenced study results. Finally, because there is not currently a standard definition of metabolically healthy obesity (20–22), we used the original Adult Treatment Panel III criteria in sensitivity analysis to assess whether our results differed by the inclusion of waist circumference of larger than 102 cm in men and larger than 88 cm in women in the metabolic syndrome definition.
Role of the funding source
The study sponsors had no role in the study design; collection, analysis, or interpretation of the data; the writing of the report; or the decision to publish.
Results
The MESA cohort through visit 5 consisted of 29 528 BMI observations from 6814 MESA participants, with an average of 4.3 observations per person. Participants with metabolic syndrome by the end of follow-up were older, had lower income and education, and had a higher average BMI at baseline compared to those without metabolic syndrome (Table 1). Among participants who were obese for all five visits (n = 1040), only 24% (n = 251) were free of metabolic syndrome throughout follow-up. They were younger (mean age 58 years), more likely to be female (63%) and Hispanic (46% compared to 34% Caucasian, 0% Asian, and 20% African American), and had a higher BMI (mean of 35.0 kg/m2) at baseline compared to other participants.
Table 1.
Baseline Characteristics (Mean (se) or %) for Participants in MESA by Metabolic Syndrome Status at Completed Follow-Up
| Characteristic at Visit 1 | Metabolic Syndrome Status at Visit 5 |
All (n = 6814) | ||
|---|---|---|---|---|
| None (n = 2877) | Intermittenta (n = 2127) | Consistentb (n = 1810) | ||
| Age (y) | 61.5 (0.20) | 61.5 (0.23) | 63.6 (0.22) | 62.2 (0.12) |
| Sex (% female) | 53.4 (0.93) | 49.4 (1.18) | 55.0 (1.08) | 52.8 (0.60) |
| Race/ethnicity | ||||
| White (%) | 44.7 (0.93) | 38.1 (1.14) | 30.4 (1.00) | 38.5 (0.59) |
| Asian (%) | 10.9 (0.58) | 14.0 (0.82) | 11.1 (0.68) | 11.8 (0.39) |
| African American (%) | 27.9 (0.84) | 25.2 (1.02) | 29.8 (1.00) | 27.8 (0.54) |
| Hispanic (%) | 16.5 (0.69) | 22.8 (0.99) | 28.7 (0.98) | 22.0 (0.50) |
| Gross gamily income (% ≥$35 000) | 62.5 (0.92) | 55.2 (1.19) | 45.9 (1.11) | 55.4 (0.61) |
| Education (% completed high school) | 87.3 (0.62) | 82.5 (0.89) | 74.3 (0.95) | 82.0 (0.47) |
| BMI (kg/m2) | 26.6 (0.09) | 28.8 (0.12) | 30.4 (0.12) | 28.3 (0.07) |
| Obesity level | ||||
| Not obese | 79.7 (0.75) | 64.7 (1.12) | 54.3 (1.08) | 67.8 (0.57) |
| I | 14.2 (0.65) | 23.8 (1.00) | 27.8 (0.97) | 21.0 (0.49) |
| II | 4.31 (0.38) | 7.90 (0.63) | 11.7 (0.70) | 7.56 (0.32) |
| III | 1.84 (0.25) | 3.59 (0.44) | 6.25 (0.53) | 3.68 (0.23) |
| High triglycerides (%) | 6.60 (0.46) | 44.0 (1.17) | 48.1 (1.08) | 29.5 (0.55) |
| Low HDL (%) | 10.7 (0.58) | 51.0 (1.18) | 59.0 (1.07) | 36.5 (0.58) |
| High BP (%) | 25.5 (0.81) | 51.4 (1.18) | 65.4 (1.03) | 44.9 (0.60) |
| High fasting glucose (%) | 1.77 (0.25) | 14.5 (0.83) | 40.2 (1.06) | 17.1 (0.46) |
Metabolic syndrome is defined as 2 or more of the following: Triglyceride level ≥150 mg/dl; HDL cholesterol <40 mg/dl in men and <50 mg/dl in women; systolic BP ≥130 mm Hg or diastolic blood pressure ≥85 mm Hg or use of BP medications; rasting glucose ≥110 mg/dl or diagnoses of diabetes. Bold indicates where intermittent or consistent metabolic syndrome estimates differ statistically from the “none” category at the P < .05 level.
Intermittent metabolic syndrome is defined by participants with measured metabolic syndrome who did not meet the criteria for metabolic syndrome at a subsequent visit.
Consistent metabolic syndrome is defined by participants with metabolic syndrome at every attended visit after the first measurement of metabolic syndrome.
The overall prevalence of metabolic syndrome in MESA participants by the end of visit 5 was 68% (3937 of 6814), with prevalence of metabolic syndrome components in the following declining order: high BP, low HDL, high triglycerides, and high fasting glucose. The prevalence was 48% (1956 of 4066) for those without any measured obesity across completed follow-up. For participants with obesity at any visit, the prevalence of metabolic syndrome was 72% (1981 of 2748) across follow-up. The prevalence of metabolic syndrome at completed follow-up was monotonically higher by obesity severity with a prevalence of 69% for level I (1142 of 1656), 76% for level II (532 of 700), and 78% for level III (307 of 392). Of those without metabolic syndrome at baseline, 32% overall (1362 of 4211), 26% of those without obesity (739 of 2849), and 44% of those with obesity (595 of 1362) transitioned to metabolic syndrome during follow-up. The incidence and prevalence of metabolic syndrome were generally higher for higher obesity severity and duration at each MESA visit (Table 2 and Supplemental Table 1).
Table 2.
Metabolic Syndrome Incidence by Obesity Severity and Duration for Each MESA Visit Among Those With Metabolically Healthy Obesity
| Obesity Exposure | Visit 1 (n = 1032) | Visit 2 (n = 965) | Visit 3 (n = 942) | Visit 4 (n = 946) | Visit 5 (n = 824) |
|---|---|---|---|---|---|
| Severity (WHO obesity level) | |||||
| I | NA | 20.4 (17–23) | 25.6 (22–30) | 24.5 (21–27) | 23.5 (20–27) |
| II | NA | 27.4 (21–33) | 32.6 (26–39) | 32.2 (26–38) | 28.5 (22–35) |
| III | NA | 29.5 (21–38) | 34.0 (24–44) | 37.1 (28–46) | 37.4 (28–47) |
| P for trenda | .009 | .022 | .002 | .003 | |
| Duration (number of visits with obesity)b | |||||
| 1 | NA | 21.1 (16–26) | 20.9 (15–27) | 25.4 (18–33) | 25.2 (18–32) |
| 2 | NA | 22.3 (19–25) | 25.5 (19–32) | 23.3 (18–31) | 19.8 (12–27) |
| 3 | NA | NA | 27.2 (24–30) | 26.0 (20–32) | 23.5 (14–32) |
| 4 | NA | NA | NA | 28.2 (25–32) | 23.7 (17–30) |
| 5 | NA | NA | NA | NA | 27.7 (24–32) |
| P for trenda | .69 | .092 | .27 | .20 |
Abbreviations: NA, not available; WHO, World Health Organization.
Nonparametric test for trend.
Duration is calculated using obesity measured during MESA visits.
Associations of overall obesity with metabolic syndrome
Across the complete follow-up, overall obesity was associated with more than a doubling of the odds of incident metabolic syndrome compared to normal weight (odds ratio [OR], 2.26 [95% confidence interval, 1.94–2.63]). A similar estimate for overall obesity was observed with a repeated measures analysis (2.01 [1.88–2.15]), and with the mixed effects model (OR, 3.02 [2.57–3.55]). The mixed effects model also provided evidence of significant individual heterogeneity (ρ = 0.47 [0.44–0.50]). Estimates for prevalent metabolic syndrome were similar (data not shown).
Associations of obesity severity and duration with metabolic syndrome
Across the complete follow-up, a higher obesity severity and total obesity duration were strongly and monotonically associated with greater odds of incident metabolic syndrome by the end of follow-up (Table 3). In repeated measures analysis, including random effects estimates in the model, significantly improved model fit (P < .001) compared to the simpler population averaged model. Combined with the SD of the random intercepts (1.62 log odds) and the significant variation in metabolic syndrome incidence attributable to differences between subjects (61%), this improvement indicates that there is significant individual heterogeneity in metabolic syndrome risk. Although some specific estimates for obesity severity were not significant, all linear tests for trend for severity were significant at the P < .05 level (Table 3). All ORs for obesity duration, compared to the lowest group, were positive and significant (Table 3). Results were similar for prevalent metabolic syndrome (Supplemental Table 2).
Table 3.
Incident Metabolic Syndrome ORs and 95% Confidence Intervals From Logistic Regression by Obesity Exposure Status Among MESA Participants With Measured Obesity at Any Visit
| Obesity Exposure | Model |
||
|---|---|---|---|
| Across Completed Follow-Up | Repeated Measuresa | Mixed Effectsb | |
| Cumulative Exposure to Obesity (n = 1362) | |||
| Severity (obesity level) | |||
| I | 1.0 (Ref) | 1.0 (Ref) | 1.0 (Ref) |
| II | 1.27 (0.94–1.71) | 1.32 (1.09–1.60) | 1.46 (1.07–2.00) |
| III | 1.44 (1.10–2.33) | 1.63 (1.25–2.14) | 2.06 (1.28–3.29) |
| P for trend | .010 | <.001 | .001 |
| Duration (number of visits with obesity) | |||
| 1 | 1.0 (Ref) | 1.0 (Ref) | 1.0 (Ref) |
| 2 | 1.80 (1.04–3.13) | 4.43 (3.54–5.53) | 10.23 (7.15–14.6) |
| 3 | 1.92 (1.13–3.24) | 5.29 (4.21–6.63) | 13.85 (9.50–20.2) |
| 4 | 2.53 (1.60–4.01) | 5.73 (4.52–7.27) | 16.10 (10.8–24.1) |
| 5 | 2.96 (1.93–4.54) | 6.15 (4.19–9.03) | 18.04 (9.51–34.2) |
| P for trend | <.001 | <.001 | <.001 |
| Random intercept variance (σ2) | NA | NA | 1.62 (1.38–1.86) |
| Intraclass correlation (ρ) | NA | NA | 0.61 (0.55–0.66) |
All models include age, sex, race/ethnicity, and repeated measures models include an indicator variable for visit 5. Obesity level I: BMI 30.0–34.9; level II: BMI 35.0–39.9; and level III: BMI 40.0+. Metabolic syndrome is defined as 2 or more of the following: Triglyceride level ≥150 mg/dl; HDL cholesterol <40 mg/dl in men and <50 mg/dl in women; systolic BP ≥130 mm Hg or diastolic BP ≥8 5mm Hg or use of blood pressure medications; rasting glucose ≥110 mg/dl or diagnoses of diabetes. Bold indicates estimates differ statistically from the reference category at the P < .05 level.
Abbreviations: NA, not available; Ref, reference.
Population averaged estimates from generalized estimating equations model.
Fixed and random effects estimates from mixed effects model.
Analysis of BMI severity as a continuous variable produced similar results to those using World Health Organization categories. Adjusting for obesity duration, each 1 kg/m2 higher BMI value is associated with a 66% higher odds of prevalent metabolic syndrome (OR, 1.66 [1.27–2.17]). The odds of incident metabolic syndrome were 1.05 (1.01–1.09) times higher for every unit of BMI.
Associations of obesity with metabolic syndrome components
Estimates for incident metabolic syndrome components were mostly consistent with those for metabolic syndrome (Table 4). The inverse relationship of obesity severity and duration with incident high LDL may indicate that LDL and metabolic syndrome are markers for different aspects of metabolic health. Estimates for prevalent metabolic syndrome components are mostly similar (Supplemental Table 3), although the significant inverse relationship of obesity severity and duration with prevalent high triglycerides contradict the results for incident metabolic syndrome.
Table 4.
ORs and 95% Confidence Intervals for Incident Metabolic Syndrome Components and High LDL by Time-Varying Obesity Severity and Duration in MESA
| Obesity Exposure | High BP (n = 1288) | High Fasting Glucose (n = 2095) | Low HDL (n = 1470) | High Triglycerides (n = 1778) | High LDLa (n = 1452) |
|---|---|---|---|---|---|
| Severity (WHO obesity level) | |||||
| I | 1.0 (Ref)b | 1.0 (Ref)b | 1.0 (Ref) | 1.0 (Ref) | 1.0 (Ref)b |
| II | 1.30 (1.07–1.59) | 1.60 (1.33–1.93) | 0.92 (0.74–1.14) | 0.92 (0.74–1.15) | 0.89 (0.70–1.12) |
| III | 1.66 (1.25–2.21 | 1.86 (1.43–2.43) | 1.15 (0.86–1.54) | 0.98 (0.72–1.34) | 0.69 (0.49–0.98) |
| Duration (number of visits with obesity) | |||||
| 1 | 1.0 (Ref)b | 1.0 (Ref)b | 1.0 (Ref) | 1.0 (Ref)b | 1.0 (Ref)b |
| 2 | 4.15 (3.25–5.30) | 7.11 (5.21–9.71) | 4.79 (3.71–6.17) | 4.07 (3.17–5.23) | 4.99 (3.67–6.80) |
| 3 | 5.85 (4.58–7.49) | 7.70 (5.62–10.6) | 5.52 (4.26–7.16) | 4.02 (3.10–5.20) | 5.39 (3.94–7.39) |
| 4 | 8.14 (6.33–10.5) | 12.74 (9.28–17.5) | 5.09 (3.87–6.69) | 3.72 (2.83–4.90) | 4.82 (3.46–6.70) |
| 5 | 5.25 (3.69–7.48) | 16.97 (10.6–27.0) | 5.83 (3.65–9.31) | 4.31 (2.54–7.31) | 4.66 (2.75–7.91) |
All models use repeated measures logistic regression including age, sex, race/ethnicity, and an indicator variable for visit 5. Obesity level I: BMI 30.0–34.9; level II: BMI 35.0–39.9; and level III: BMI 40.0+. Bold indicates estimates differ statistically from the reference category at the P < .05 level.
Abbreviations: Ref, reference; WHO, World Health Organization.
High LDL is defined as LDL>130 or use of lipid-lowering medication.
Nonparametric test for trend is significant (P < .05).
Heterogeneity and sensitivity analysis
There was significant interaction for obesity severity and duration with incident metabolic syndrome by age, sex, and race/ethnicity (P < .05) (Supplemental Tables 4 and 5). There was significant interaction between severity and duration (P < .001). Associations were stronger for those younger than age 70 years compared to those older than 70 years of age, women compared to men, and Hispanic participants compared to White participants. Sample sizes for the Asian subgroup were too small to produce convergent models. Estimates for prevalent metabolic syndrome were broadly similar, with somewhat less evidence of heterogeneity (data not shown). For prevalent metabolic syndrome, there was no significant interaction between severity and duration (P = .10).
Employing different definitions of obesity exposure, including the exclusion of participants with prevalent obesity (data not shown), assessment of misclassification by using obesity at age 40 (Supplemental Tables 4 and 5), and assessment of intermittent obesity across resulted in similar findings. Intermittent obesity was associated with incident metabolic syndrome (OR, 2.32 [1.82–2.92]), and ORs for intermittent obesity were not significantly different (P = .52) than for consistent obesity (OR, 2.52 [2.14–2.96]). Similarly, using the conventional definition of metabolic syndrome that includes waist circumference did not alter the estimates (Supplemental Tables 4 and 5).
Discussion
In MESA, higher obesity severity and duration were associated with prevalent metabolic syndrome, and both higher obesity severity and duration were associated with increased odds of incident metabolic syndrome among participants with obesity, supporting our a priori hypothesis. There was also significant individual heterogeneity in metabolic syndrome risk, separate from other explanatory factors. These associations were robust to multiple subgroup analysis. Similarly, higher severity and duration were associated with increased risk for metabolic syndrome components.
In the context of the current controversy about whether metabolically healthy obesity is truly a healthy state and whether weight loss should be recommended for this group, our results provide novel evidence that metabolically healthy obesity is a transient state on the pathway to metabolic syndrome and possibly later cardiometabolic disease. Further, our results suggest that exposure to obesity itself may be a primary cause of progression from a healthy state to metabolic risk. Although some literature suggests that duration of obesity is associated with cardiovascular disease (10–15), evidence is lacking on the separate roles that obesity severity and duration play in metabolic health and the patterns of transition between healthy and unhealthy states. Additional evidence about these pathways may help explain why those with metabolically healthy obesity are at intermediate risk for cardiometabolic outcomes between normal weight individuals and individuals in other obesity subgroups.
Our results indicate that resilience to the metabolic effects of obesity is individually heterogeneous. Although several mechanisms for metabolic resilience have been speculated, including inflammation and dimorphic body fat distributions (23), it remains unknown why obese individuals progress to metabolic dysfunction at different rates and whether it is possible to avoid metabolic syndrome indefinitely in the presence of obesity. Consistent with the prior literature (2, 7, 10, 24–26), our findings suggest that additional exposure to obesity may lead to metabolic syndrome in most individuals over time. Appleton et al found that one-third of North West Adelaide Health Study participants with metabolically healthy obesity transitioned to being metabolically “at-risk” over 4 years, and point out the difficulty in comparing across studies due to the lack of a standard definition of metabolic syndrome (2). Mørkedal et al found metabolic health, defined by modified International Diabetes Federation criteria, was more strongly associated with acute myocardial infarction than obesity, but the reverse for incident heart failure suggesting that the influence of obesity and metabolic health may vary by outcome (2). Although studies such as that by Hamer et al showing that 44.5% of those with metabolically healthy obesity transition to metabolic syndrome over 8 years provide additional support for the well-established idea that obesity in general is associated with metabolic syndrome, they do not provide any specific explanation for why some participants transition and others do not (24, 25). These studies all highlight the crucial need for replication and additional study to understand the complex role of cumulative obesity exposure in cardiometabolic health and to refine the concept of metabolically healthy obesity.
Studies that have investigated determinants of metabolically healthy obesity predominantly show that those who transition to metabolic syndrome are already less healthy at baseline compared to those who remain metabolically healthy (7, 27, 28); however, this evidence is circular and misses the originating cause of these differences. Similarly, while Hwang et al show that visceral fat level at baseline is associated with transition to metabolic syndrome, differences in visceral fat are likely determined by previous obesity severity and duration (26). Although it is generally assumed that metabolically healthy obesity is a transient condition and that obesity plays a role in this transition, to our knowledge this is the first study to explicitly test these assumptions. Similarly, those with metabolically health obesity have been shown to be at intermediate risk for diabetes and cardiovascular disease between those with normal weight and those with metabolic syndrome (1–5), and our results suggest that differing levels of obesity exposure may explain these differences. Our findings implicate obesity exposure as the mechanism for both initiation of metabolic risk as well as progression of that risk, and further support primary prevention of obesity for reductions in risk along the entire pathway from metabolic syndrome to cardiovascular disease.
The primary limitation of our study is that the total duration of obesity for participants who were obese at visit 1 is unknown and may vary; however, we addressed this issue in a number of ways in sensitivity analyses. Another limitation is that, in MESA, some participants experience metabolic syndrome intermittently, and we cannot be sure that we excluded all participants with prevalent metabolic syndrome before they experienced obesity. Similarly, the discrete nature of the data is a potential limitation because we cannot ensure that obesity precedes metabolic syndrome for those participants who have both measured for the first time at the same visit. The existence of intermittent metabolic syndrome poses additional complications in understanding the risk associated with metabolic syndrome and the role obesity plays in that risk. By definition a composite of cardiovascular risk factors, the potential for each of the metabolic syndrome components to vary over time further complicates this issue. Additional study is needed in this area. More generally, the definition of metabolic syndrome itself is being debated and the use of other definitions might produce different results. Many definitions of metabolic syndrome do not include LDL cholesterol; however, lipid-lowering medications targeted at LDL may reduce triglyceride levels or increase HDL levels leading to misclassification of metabolic syndrome in those with high LDL. Although these results are consistent with a growing body of work showing that associations with obesity are stronger in those younger than age 70 compared with older adults, the mechanisms behind these findings remain unknown (10). Finally, our main results and exploratory data analysis indicate that the underlying odds for visit 5 are lower than those for other visits in MESA, limiting our ability to fully investigate the association with obesity duration across the full follow-up period. Some evidence of differential dropout by metabolic syndrome status at visit 5 and the potential for competing risks or survival bias indicates the importance of replicating these results in larger studies with longer follow-up.
The main strength of this work is the ability to use repeated measures analysis to investigate the association of obesity severity and duration separately with metabolic syndrome across 10 years of follow-up. Access to obesity and metabolic syndrome measured clinically in a standardized way as well as the ability to investigate both metabolic syndrome prevalence and incidence allow for a targeted investigation into the nature of metabolically healthy obesity. Greater depth in investigating intermittent obesity and intermittent metabolic syndrome allows for additional insights. Finally, the depth of data and diversity of the MESA cohort allow for substantial subgroup and sensitivity analyses to support the main findings.
In MESA, obesity severity and duration were strongly associated with progression from metabolically healthy obesity to incident metabolic syndrome and almost one-half of participants with obesity transitioned to metabolic syndrome across 10 years of follow-up. Among those with the longest duration of obesity, only 24% remained free of metabolic syndrome. These results suggest that continued exposure to obesity does not result in a stable metabolically healthy state. Further, these results suggest that a primary cause of this transition is exposure to obesity itself, and not any special resilience to the effects of obesity. Weight loss should be recommended to all individuals with obesity, including those who are currently metabolically healthy.
Acknowledgments
The authors thank the other investigators, the staff, and the participants of the MESA study for their valuable contributions. A full list of participating MESA investigators and institutions can be found at http://www.mesa-nhlbi.org.
This work was supported by National Heart, Lung, and Blood Institute grants (T32HL079891 and 5T32HL007261 to M.M.-C.); National Heart, Lung, and Blood Institute contracts (HHSN268201500003I, N01-HC-95159, N01-HC-95160, N01-HC-95161, N01-HC-95162, N01-HC-95163, N01-HC-95164, N01-HC-95165, N01-HC-95166, N01-HC-95167, N01-HC-95168, and N01-HC-95169); and National Center for Research Resources grants (UL1-TR-000040 and UL1-TR-001079).
Disclosure Summary: The authors have nothing to disclose.
Footnotes
- BMI
- body mass index
- BP
- blood pressure
- HDL
- high-density lipoprotein
- LDL
- low-density lipoprotein
- MESA
- Multi-Ethnic Study of Atherosclerosis
- OR
- odds ratio.
References
- 1. Bell JA, Kivimaki M, Hamer M. Metabolically healthy obesity and risk of incident type 2 diabetes: a meta-analysis of prospective cohort studies. Obesity Rev. 2014;15:504–515. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2. Appleton SL, Seaborn CJ, Visvanathan R, et al. Diabetes and cardiovascular disease outcomes in the metabolically healthy obese phenotype: a cohort study. Diabetes Care. 2013;36:2388–2394. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3. Fan J, Song Y, Chen Y, Hui R, Zhang W. Combined effect of obesity and cardio-metabolic abnormality on the risk of cardiovascular disease: a meta-analysis of prospective cohort studies. Int J Cardiol. 2013;168:4761–4768. [DOI] [PubMed] [Google Scholar]
- 4. Kramer CK, Zinman B, Retnakaran R. Are metabolically healthy overweight and obesity benign conditions? A systematic review and meta-analysis. Ann Intern Med. 2013;159:758–769. [DOI] [PubMed] [Google Scholar]
- 5. Ortega FB, Lee DC, Katzmarzyk PT, et al. The intriguing metabolically healthy but obese phenotype: cardiovascular prognosis and role of fitness. Eur Heart J. 2013;34:389–397. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6. Aung K, Lorenzo C, Hinojosa MA, Haffner SM. Risk of developing diabetes and cardiovascular disease in metabolically unhealthy normal-weight and metabolically healthy obese individuals. J Clin Endocrinol Metab. 2014;99:462–468. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7. Schroder H, Ramos R, Baena-Dıez JM, et al. Determinants of the transition from a cardiometabolic normal to abnormal overweight/obese phenotype in a Spanish population. Eu J Nutr. 2014;531:1345–1352. [DOI] [PubMed] [Google Scholar]
- 8. van Vliet-Ostaptchouk JV, Nuotio ML, Slagter SN, et al. The prevalence of metabolic syndrome and metabolically healthy obesity in Europe: a collaborative analysis of ten large cohort studies. BMC Endocr Disord. 2014;14:9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9. Wildman RP, Muntner P, Reynolds K, et al. The obese without cardiometabolic risk factor clustering and the normal weight with cardiometabolic risk factor clustering: prevalence and correlates of 2 phenotypes among the US population (NHANES 1999–2004). Arch Intern Med. 2008;168:1617–1624. [DOI] [PubMed] [Google Scholar]
- 10. Mørkedal B, Vatten LJ, Romundstad PR, Laugsand LE, Janszky I. Risk of Myocardial Infarction and Heart Failure Among Metabolically Healthy But Obese Individuals: HUNT (Nord-Trøndelag Health Study), Norway. J Am Coll Cardiol. 2014;63:1071–1078. [DOI] [PubMed] [Google Scholar]
- 11. Abdullah A, Stoelwinder J, Shortreed S, et al. The duration of obesity and the risk of type 2 diabetes. Public Health Nutr. 2011;14:119–126. [DOI] [PubMed] [Google Scholar]
- 12. Abdullah A, Wolfe R, Stoelwinder JU, et al. The number of years lived with obesity and the risk of all-cause and cause-specific mortality. Int J Epidemiol. 2011;40:985–996. [DOI] [PubMed] [Google Scholar]
- 13. de Koning L, Hu FB. Commentary: obesity-years–a new metric to measure health effects of obesity. Int J Epidemiol. 2011;40:996–997. [DOI] [PubMed] [Google Scholar]
- 14. Reis JP, Loria CM, Lewis CE, et al. Association between duration of overall and abdominal obesity beginning in young adulthood and coronary artery calcification in middle age. JAMA. 2013;310:280–288. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15. The NS, Richardson AS, Gordon-Larsen P. Timing and duration of obesity in relation to diabetes: findings from an ethnically diverse, nationally representative sample. Diabetes Care. 2013;36:865–872. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16. Bild DE, Bluemke DA, Burke GL, et al. Multi-ethnic study of atherosclerosis: objectives and desig. Am J Epidemiol. 2002;56:871–881. [DOI] [PubMed] [Google Scholar]
- 17. World Health Organization (WHO) and Food and Agriculture Organization of the United Nations (FAO). (2016, 8/9/2016). World Health Organization BMI classification. Retrieved September 8, 2016. http://apps.who.int/bmi/index.jsp?introPage=intro_3.html.
- 18. Expert Panel on Detection Evaluation and Treatment of High Blood Cholesterol in Adults 2001 Executive Summary of The 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). JAMA. 285:2486–2497. [DOI] [PubMed] [Google Scholar]
- 19. StataCorp 2009 Stata Statistical Software: Release 11. College Station, TX: StataCorp LP [Google Scholar]
- 20. Durward CM, Hartman TJ, Nickols-Richardson SM. All-cause mortality risk of metabolically healthy obese individuals in NHANES III. J Obesity. 2012;2012:460321. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21. Martinez-Larrad MT, Corbaton Anchuelo A, Del Prado N, Ibarra Rueda JM, Gabriel R, Serrano-Rios M. Profile of individuals who are metabolically healthy obese using different definition criteria. A population-based analysis in the Spanish population. PloS one. 2014;9:e106641. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22. Yoo HK, Choi EY, Park EW, Cheong YS, Bae RA. Comparison of metabolic characteristics of Metabolically Healthy but Obese (MHO) middle-aged men according to different criteria. Kor J Family Med. 2013;34:19–26. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23. Bluher S, Schwarz P. Metabolically healthy obesity from childhood to adulthood - does weight status alone matter? Metab Clin Exp. 2014;63:1084–1092. [DOI] [PubMed] [Google Scholar]
- 24. Bell JA, Hamer M, Batty GD, Singh-Manoux A, Sabia S, Kivimäki M. Incidence of metabolic risk factors among healthy obese adults: 20-year follow-up. J Am Coll Cardiol. 2015;66:871–872. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25. Hamer M, Bell J, Sabia S, Batty G, Kivimäki M. Stability of metabolically healthy obesity over 8 years: the English Longitudinal Study of Ageing. Eur J Endocrinol. 2015;173:703–708. [DOI] [PubMed] [Google Scholar]
- 26. Hwang Y-C, Hayashi T, Fujimoto WY, et al. Visceral abdominal fat accumulation predicts the conversion of metabolically healthy obese subjects to an unhealthy phenotype. Int J Obesity. 2015;39:1365–1370. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27. Lopez-Garcia E, Guallar-Castillon P, Leon-Munoz L, Rodriguez-Artalejo F. Prevalence and determinants of metabolically healthy obesity in Spain. Atherosclerosis. 2013;231:152–157. [DOI] [PubMed] [Google Scholar]
- 28. Peppa M, Koliaki C, Papaefstathiou A, et al. Body composition determinants of metabolic phenotypes of obesity in nonobese and obese postmenopausal women. Obesity (Silver Spring, MD). 2013;21:1807–1814. [DOI] [PubMed] [Google Scholar]