Introduction

Excessive weight, defined as a body mass index (BMI) ≥ 25 (weight in kilograms divided by the square of the height in meters), is associated with an increased risk of metabolic complications such as dyslipidemia, hypertension, and diabetes [1–3]. However, not all overweight and obese subjects present an at-risk metabolic profile. Indeed, recent evidence suggests that a considerable number of overweight and obese subjects are free of metabolic complications [4, 5]. The prevalence of this metabolically healthy obese/overweight (MHOO) phenotype varied between 18 and 44 % [4]. Little is known about the magnitude of the prospective manifestation of cardiometabolic complications in the MHOO phenotype [4–8]. Furthermore, it is unknown whether weight or WC gain or both are predictors for a shift from the MHOO to the MAOO phenotype.

A healthy lifestyle is associated with a favorable cardiometabolic risk profile [9–12]. Therefore, it is likely that a healthy diet, not smoking, and/or physical activity counterbalance the adverse effects of excessive weight. So far, evidence on the impact of behavioral factors on the MHOO phenotype is scarce [13–16].

We undertook this study with the aim to determine 10-year changes in anthropometric variables and lifestyle associated with the transition of the MHOO to MAOO phenotype in a representative Spanish population.

Subjects and methods

Study population

Data were obtained from a population-based survey conducted in Girona (Spain) in 2000 and 2009. The baseline survey in 2000 examined a random population-based sample of 3058 men and women aged 25–74 years (participation rate: 71.0 %). In 2009, the 2,715 noninstitutionalized participants still residing in the catchment area were invited for re-examination and 2,181 attended (participation rate of follow-up: 80.3 %). For the purpose of the present study, participants with a baseline BMI ≥ 25.0 were included for cross-sectional analysis (n = 1,445). Prospective analysis was performed with the baseline MHOO phenotype group (n = 301). All measurements were performed at baseline and at follow-up. Participants were duly informed and signed their consent to participate in the study. The project was approved by the local ethics committee (CEIC—PSMAR, Barcelona, Spain).

Definition of the MHOO phenotype

In the present study, we were interested in the predictive value of lifestyle on the maintenance of metabolic health in subjects with abnormal fat accumulation. The WHO had defined overweight and obesity as “abnormal or excessive fat accumulation that presents a risk to health” [17]. Therefore, we decided to combine the two categories characterized by excessive accumulation of fat in adipose tissue in our analysis. The outcome of interest was the transition of the MHOO to the MAOO phenotype. For this purpose, we defined the MHOO phenotype as participants with a BMI ≥ 25.0 and free from the following common cardiometabolic risk factors: hypercholesterolemia, hypertriglyceridemia, diabetes, hypertension, and low HDL and high LDL cholesterol levels as defined in laboratory measurements. In contrast, the presence of one or more of these conditions defined the MAOO phenotype.

Laboratory measurements

Blood was withdrawn after 10–14 h fasting. Serum-sample aliquots were stored at −80 °C. Glycemia was measured in an aliquot of serum. Total cholesterol and triglyceride concentrations were determined enzymatically (Roche Diagnostics, Basel, Switzerland). High-density lipoprotein (HDL) cholesterol was measured after precipitation of apoprotein B-containing lipoproteins with phosphotungstic-Mg++ (Boehringher, Mannheim, Germany). Analyses were performed in a Cobas Mira Plus autoanalyzer (Roche Diagnostics, Basel, Switzerland). Quality control was performed with External Quality Assessment-WHO Lipid Program (World Health Organization, Prague, Czech Republic) and Monitrol-Quality Control Program (Baxter Diagnostics, Dudingen, Switzerland). Low-density lipoprotein (LDL) cholesterol was calculated by the Friedewald equation whenever triglycerides were <300 mg/dL. Cut-points for abnormal lipid levels were total cholesterol ≥ 240 mg/dL, LDL ≥ 160 mg/dL, triglycerides ≥ 200 mg/dL, and HDL ≤ 40 mg/dL for men and HDL ≤ 50 mg/dL for women [18]. Diabetes was defined as use of insulin or hypoglycemic agents or fasting blood glucose >125 mg/dL or previous diagnosis of diabetes by a physician [19].

Blood pressure measurement

Blood pressure was measured with a periodically calibrated mercury sphygmomanometer. A cuff adapted to upper arm perimeter (young, adult, and obese) was selected for each participant. Measurements were performed after a 5-minute rest. Two measurements were taken, and the lower value was recorded for the study. Participants were considered hypertensive if previously diagnosed by a physician, under treatment, or presenting systolic blood pressure (SBP) ≥ 140 or diastolic blood pressure (DBP) ≥ 90 mmHg [18].

Anthropometrics

A calibrated precision scale was used for weight measurement. Readings were rounded up to 200 g. Height was measured in the standing position and rounded up to the nearest 0.5 cm. Weight was divided by height squared (kg/m2) to establish the BMI.

Measurement of waist circumference was performed midway between the lowest rib and the iliac crest in the horizontal positions. The measurement, taken in cm, was rounded to the nearest 0.5 cm.

Dietary assessment

Food consumption was determined using a validated [20, 21] food frequency questionnaire (FFQ) administered by a trained interviewer at baseline and at follow-up. In a 166-item food list including alcoholic and nonalcoholic beverages, participants indicated their usual consumption and chose from 10 frequency categories ranging from never or less than once per month to 6 or more times per day.

The Mediterranean-like dietary score (MLDS) was calculated according to the tertile distribution of energy-adjusted food intake [21]. For cereals, fruits, vegetables, legumes, fish, olive oil, low-fat dairy products, and nuts, the lowest tertile is coded as 1, medium as 2, and highest as 3. For meat (including red meat and sausages), sugar-sweetened carbonated beverages and added sugars, pastries, and fast food, the score was inverted, with the highest tertile coded as 1 and lowest as 3.

Other variables

Leisure-time physical activity (LTPA) was measured by the validated Minnesota LTPA questionnaire administered by a trained interviewer [22, 23]. Measurements of smoking habits and demographic and socioeconomic variables were obtained from structured standard questionnaires administered by trained personnel. Participants were categorized as nonsmokers (never smokers and ex-smokers with more than 1 year of smoking cessation) and current smokers. Maximum education level attained was elicited and recorded for analysis as primary school versus secondary school or university.

Lifestyle changes

We hypothesized that changes or maintenance of several lifestyle variables together will have an additive effect on the transition away from the MHOO phenotype. For this analysis, we created a joint variable composed of secular trends of diet quality, LTPA, and smoking. Unhealthy behaviors were coded 0 and defined as follows: low LTPA (below median) at baseline and follow-up; low adherence to the MLDS (below median) at baseline and follow-up; smoking at baseline and follow-up; high LTPA at baseline and low LTPA at follow-up; high adherence to the MLDS and low adherence to the MLDS at follow-up. Healthy behaviors were coded 1 and defined as follows: high LTPA (above median) at baseline and follow-up; high adherence to the MLDS (above median) at baseline and follow-up; nonsmoking at baseline and follow-up; low LTPA at baseline and high LTPA at follow-up; low adherence to the MLDS and high adherence to the MLDS at follow-up; smoking at baseline and cessation of smoking at follow-up.

The sum of scores of the dichotomized lifestyle variables LTPA, smoking, and diet quality yielded healthy lifestyle index (HLI) scores ranging from 0 to 7.

Statistical analysis

Differences in continuous variables were compared using the Student t test. The chi-square test was used for categorical variables. Multiple logistic regression models were fitted to determine the confounder-controlled association of metabolic phenotypes and changes in BMI, WC, WHtR, and lifestyle. Additionally, we compared the magnitude of the associations between metabolic phenotypes and changes in these anthropometric measures. For this purpose, anthropometric variables were standardized as z values. To explore effect modification according to sex, lifestyle, and baseline weight and WC, we modeled interaction terms for sex/weight change, sex/WC change, sex/HLI index, sex/BMI change, and sex/WHtR change. Differences were considered significant if p < 0.05. Statistical analysis was performed using SPSS version 18.0. (SPSS Inc. Chicago, IL).

Results

The prevalence of the MHOO phenotype was 20.8 (24.2 % in overweight and 16.1 % in obese participants) at baseline. At 10-year follow-up, 49.3 % of the initial MHOO phenotype had transitioned to the MAOO phenotype (48.4 % in overweight and 57.0 % in obese participants).

No significant interactions between sex/weight change, sex/WC change, sex/HLI index, sex/BMI change, and sex/WHtR change were observed.

At baseline, the MHOO phenotype group (n = 301) was younger, more highly educated, smoke more, adhered to a less healthy diet, and had higher energy intake, lower BMI, lower WC, and lower WHtR than the MAOO phenotype group (n = 1,144) (Table 1). Participants who remained MHOO after 10 years of follow-up (n = 153) were younger and had a higher energy intake at baseline compared with those who became MAOO (n = 148) (Table 1). At follow-up, higher WC and WHtR were found in participants who became MAOO. Blood lipids, fasting glucose, and systolic and diastolic blood pressures were significantly different between the MHOO and MAOO phenotype at all measurement points with the exception of follow-up diastolic blood pressure of participants who remained MHOO and those who became MAOO after 10 years of follow-up (Table 1). The incidence rate of diabetes in the initial MHOO phenotype was 9.3 %.

Table 1 Characteristics of overweight/obese study participants according to metabolically healthy overweight/obese (MHOO) and metabolically abnormal overweight/obese (MAOO) phenotype
Full size table

At follow-up, 16.4 % (n = 187) of the metabolically abnormal healthy overweight/obese phenotype [19.4 % overweight (n = 137) and 11.5 % obese (n = 50)] at baseline had transitioned to the MHOO phenotype. This phenotype was significantly younger, with lower BMI and WC at baseline than the stable MAOO phenotype. Stratified analysis by overweight and obesity categories revealed the same results.

A central question of the present study was to determine the effect of changes in BMI, WHtR, WC, and lifestyle variables on the shift from the MHOO to MAOO phenotype. For this purpose, we focused our analysis on participants who presented the MHOO phenotype at baseline. An increase in BMI, WC, and WHtR was a positive predictor for the transition to the MAOO phenotype in sex- and age-adjusted models (Table 2). Further adjusting for the HLI, educational level, alcohol consumption, and energy intake at baseline did not meaningfully affect the direction and magnitude of the associations. The magnitude of the associations was similar among anthropometric measures.

Table 2 Odds ratio (OR) and 95 % CI of the transition from metabolically healthy to abnormal overweight/obese phenotype and changes in body mass index (BMI), waist circumferences (WC), and waist to height ratio (WHtR)
Full size table

An increase in 1 unit of the HLI score was associated with a 33.0 % lower risk of transition to the MAOO phenotype (Table 3). This association was independent of baseline values of educational level, alcohol consumption, smoking, energy intake, BMI, WC, and WHtR (Table 3). None of the possible combinations of lifestyle changes, with the exception of the concomitant presence of all three healthy lifestyle changes, were significantly associated with the transition of the MHOO to the MAOO phenotype (Table 4).

Table 3 Odds ratio (OR) and 95 % CI of the transition from metabolically healthy to abnormal overweight/obese phenotype and changes in the healthy lifestyle index
Full size table
Table 4 Odds ratio (OR) and 95 % CIa of the transition from metabolically healthy to abnormal overweight/obese phenotype and changes in each possible combination of lifestyle behaviors
Full size table

Age- and sex-adjusted linear regression models, stratified by the overweight and obese phenotypes, revealed that a 1 point increase in the HLI was associated with a 39 % [OR 0.61 (CI 0.39;0.94), p = 0.026] and 18 % [OR 0.82 (CI 0.46;1.44), p = 0.494] decrease in risk in the transition from the healthy to the unhealthy metabolically overweight and obese phenotype, respectively.

Changes in BMI, WC, and WHtR were predictive, albeit not significant in all cases, for the transition from the healthy to the unhealthy metabolically overweight [∆BMI = OR 1.35 (CI 1.11;1.47), p = 0.002; ∆WC = OR 1.05 (CI 1.00; 1.10), p = 0.066; WHtR = OR 2.53 (CI 1.10; 5.83), p = 0.029] and obese phenotype [∆BMI = OR 1.05 (CI 0.90;1.23), p = 0.525; ∆WC = OR 1.05 (CI 0.99;1.12), p = 0.11; WHtR = 2.47 (CI 0.87;7.00), p = 0.089]. All models were adjusted for sex, age, and the corresponding anthropometric variable at baseline.

Discussion

An increase in surrogate measures of general and abdominal adiposity was positively associated with the shift from the MHOO to the MAOO phenotype. These associations were not independent of weight gain. In contrast, a high HLI score was negatively associated with this transition.

The prevalence of the MHOO phenotype was 20.8 % in this overweight and obese population, which is considerably less than that reported by another large population-based study [24]. However, this finding is not surprising because prevalence rates of the MHOO phenotype vary greatly between studies, mainly due to the lack of a unique definition of the phenotype [4, 13]. Additionally, most of the studies included small, nonrepresentative populations, making it difficult to compare results.

It has been shown that the MHOO phenotype is more prevalent in younger people, [24, 25] which is in line with results from this study. Moreover, our prospective analysis revealed that a shift of the MHOO phenotype is more probable in older age groups. Of course, this finding is not surprising if we consider the strong positive association between age and cardiometabolic risk [26]. On the other hand, little is known about socioeconomic and lifestyle determinants of the MHOO phenotype. Wildman et al. [24] reported a higher degree of physical activity and a higher educational level in the MHOO phenotype. However, this finding was not confirmed by other authors [27]. Obese black women with the metabolically healthy obese (MHO) phenotype were more educated and had lower unemployment than their metabolically unhealthy obese peers [27]. The MHOO phenotype in the present study was characterized by a higher educational level, a higher smoking prevalence, and a less healthy diet. However, these associations were strongly age dependent. Furthermore, none of these variables significantly predicted the transition from the MHOO to MAOO phenotype. This finding is in line with a recently published report showing no significant prospective association between baseline physical activity, alcohol consumption, or smoking and the transition to the metabolically unhealthy obese phenotype [16]. A few longitudinal studies have addressed the impact on health outcomes of obesity without metabolic complications [5–8, 16]. The evidence from these studies is inconsistent; therefore, whether healthy obese individuals are at lower risk of degenerative changes remains an open question. A recent publication by Appleton et al. [16] was the first to report on the association between the stability of the MHO phenotype and incidence of diabetes and cardiovascular disease (CVD). They found that the risk of developing diabetes, but not CVD, was significantly higher for the transition of the MHO phenotype to the metabolically unhealthy obese phenotype compared with the stable or improved MHO. It is important to note that controlling for cardiovascular risk factors strongly attenuates the impact of overweight and obesity on cardiovascular events [28].

A key question is whether the MHOO phenotype is time-related or represents a constant characteristic. In this study, nearly half of the initially MHOO subjects shifted to the MAOO phenotype after 10 years. Recently published data show a positive association between WC and the MHOO phenotype at population scale [25]. Indeed, WC increase, and weight gain has been associated with the presentation of cardiometabolic abnormalities [29–31]. Therefore, it is reasonable to assume that gaining weight and increasing WC will increase the risk of a shift from the MHOO to the MAOO phenotype. Our results showed that an increase in 1 standard deviation of BMI, WC, and WHtR increased the risk of the transition to the MAOO phenotype by 49, 41, and 43 %, respectively. This increase in general and abdominal adiposity predicted the transition to the MAOO phenotype independently of lifestyle changes.

Both maintaining a healthy lifestyle and making healthful lifestyle changes have been associated with a favorable cardiometabolic profile [9–12]. In this study, the adherence to a healthy diet and not smoking were nonsignificant predictors for the stability of the MHOO phenotype. The strongest although not significant association was observed for diet quality measured by the MLDS. A large body of literature demonstrates protective effects of the Mediterranean diet on cardiovascular health in different populations [32–35]. An interesting finding of this study is the additive effect of healthy lifestyle factors on cardiovascular risk factors in subjects at cardiometabolic risk. A previous report from the European Prospective Investigation into Cancer and Nutrition (EPIC)-Potsdam cohort demonstrated progressive decline of the incidence of chronic disease as the number of healthy lifestyle factors increased [9], in line with findings from other cohort studies indicating a favorable and additive effect of lifestyle variables on disease outcomes [10–12]. Kantartzis et al. [36] reported a significant increase in insulin sensitivity in obese insulin resistant but not in MHO patients after a 9 months lifestyle intervention program. However, follow-up insulin sensitivity was nearly twice as high in the MHO group compared with insulin-resistant obese group. In the present study, we found an independent association of anthropometric measures of adiposity and lifestyle changes on the transition of the MHOO phenotype. This finding indicates that anthropometric and lifestyle changes are linked to this transition by different pathophysiological pathways.

The underlying mechanisms linked to the preservation of a favorable cardiometabolic profile in at-risk individuals are poorly understood. It has been shown that visceral fat mass accumulation is strongly linked with a cardiometabolic risk profile. Furthermore, excessive fat mass is associated with a chronic inflammatory state and rise in oxidative stress, which are in turn mediators for the development of cardiometabolic disorders [37, 38]. However, not all individuals with elevated fat mass suffer from a chronic inflammatory state and a rise in oxidative stress [39, 40]. In this context, it is of interest to note that intervention and observational studies provide evidence for a protective role of diet and physical activity on inflammation and oxidative stress [41–43]. The impact of ectopic fat for the metabolic status of the obese has been highlighted by Stefan and colleagues [44]. They found that lower ectopic fat accumulation in the liver and skeletal muscle is predictive for the MHO phenotype.

The strengths of this study are the population-based design and the measurement of anthropometric, cardiovascular risk, and lifestyle variables at baseline and at 10-year follow-up. Furthermore, dietary and LTPA data were recorded using validated questionnaires. One limitation of this study is the relatively modest sample size that makes it impossible to perform stratified analysis by BMI class.

In conclusion, our results indicate that an increase in anthropometric measures of adiposity heightens the risk of transition from the MHOO phenotype to the MAOO phenotype. In contrast, a healthy lifestyle increases the probability of maintaining a favorable cardiometabolic profile and thus counterbalances the risk of a transition away from the MHOO phenotype.