The Typology of Metabolic Syndrome in the Transition to Adulthood

Abstract

Background:

Metabolic syndrome (MetS) is a clustering of risks associated with cardiometabolic disease in adults. Obesity is considered the major etiologic factor. However, unlike obesity, the natural history of MetS as adolescents transition to adulthood is unknown.

Objective:

The purpose of this study was to characterize the typology of MetS as adolescents transition to young adulthood and to explore determinants of that typology.

Design/Participants:

A total of 458 participants from a school-based longitudinal cohort study of baseline 5th to 12th graders were followed for 9 years.

Methods:

Based on the presence or absence of MetS at study visits (year [Y] 1, Y4, Y8, and Y10), a MetS typology was defined, and its characteristics were explored using multinomial regression modeling.

Results:

Both obesity and MetS increased (obesity from 21.0% to 33.4% and MetS from 2.8% to 17.9%). MetS typology was as follows: never, 76.9%; incident, 16.4%; unstable/remitted, 5.7%; and persistent, 1.1%. Of Y1 MetS-positive cases, 61.5% remitted, as did 36.4% of Y4 MetS-positive cases and 25% of Y8 MetS-positive cases. Most incident cases (56.0%, n = 42) occurred in Y10; only 12% (n = 9) occurred in Y4. Obesity increased the odds of MetS (incident: odds ratio [OR] = 4.42, 95% confidence interval [CI] = 2.23–8.76; unstable/remitted: OR = 7.79, 95% CI = 3.12–19.41; persistent: OR = 31.36, 95% CI = 2.99–328.98). In addition, changes in body mass index over the study were associated with persistent (OR = 1.27, 95% CI = 1.03–1.56) and incident MetS (OR = 1.49, 95% CI = 1.31–1.71), but not unstable/remitted MetS (OR = 1.09, 95% CI = 0.99–1.19). Of note, body mass index increased for 77% of those with unstable/remitted MetS, including 90% (n = 9/10) of persistently obese youth with unstable/remitted MetS.

Conclusions:

During the transition to adulthood, the diagnosis of MetS is highly unstable and fluctuates even among those who are obese and gaining weight.

Adolescence is a critical period for determining adult cardiometabolic risk. The development of atherosclerosis, as evidenced by the presence of fatty streaks, begins in adolescence in association with increased adiposity, dyslipidemia, and hypertension (1). Moreover, adolescent obesity, hyperglycemia, and hypertension have been shown to increase the risk of premature death in adulthood (2). Given these associations and the increased prevalence of obesity, including severe obesity, among adolescents (3–5), identifying youth at risk for adult cardiometabolic disease and intervening to reduce risk is an increasingly urgent mandate for pediatricians. However, the optimal strategy for identifying at-risk youth remains unclear, particularly given the normative developmental changes in physiology during the second decade of life (6, 7).

Among adults, metabolic syndrome (MetS), defined as the coexistence of at least 3 of 5 sentinel cardiovascular risk factors (abdominal obesity, dysglycemia, hypertriglyceridemia, decreased HDL, and hypertension) is used as an indicator of increased cardiometabolic risk (8–10). For adolescents, however, the utility of MetS as a diagnostic concept remains uncertain (11, 12). There is no consensus as to its definition among 10 to 16 year olds, and no organization has endorsed a definition for use in children <10 years of age (11, 13, 14). Regardless of definition, multiple studies of adolescents, including a previously reported 3-year follow-up of the present cohort (7), have shown instability in the diagnosis of MetS, calling its utility into question for this developmental period (7, 15–17). Similarly, studies assessing MetS in adolescence and again in adulthood have shown only modest predictive utility of adolescent MetS in adult MetS or cardiometabolic disease (18, 19). To date, no studies have serially evaluated MetS during the adolescent transition and assessed its stability in relationship to demographics or changes in weight status in a community population over a longer time horizon that reflects the transition to adulthood.

The purpose of the current investigation is to address this gap in the literature. Using data from the Princeton School District (PSD) Study, a school-based longitudinal cohort study from a Midwestern school district with 4 waves of data collection in 10 years, we define a typology of MetS as these adolescents aged into young adulthood and then determine correlates of this typology.

Materials and Methods

These data are drawn from phase 2 of the PSD Study, a longitudinal cohort study that followed a cohort of 5th to 12th graders from a single school district (94% non-Hispanic black and white) in the Greater Cincinnati area from 2001 to 2011 (20, 21). Approval was granted by the institutional review boards at all participating institutions, and written parental consent and student assent (students <18 years old) or written student consent (students ≥18 years old) was obtained. Study visits were performed at baseline (year [Y] 1) as well as at Y4, Y8, and Y10. Phase 2 of the PSD Study included 801 of 1191 eligible phase 1 participants. Among these, 510 self-identified as non-Hispanic black or white, were nondiabetic, and participated in Y10. Among these 510, we were able to ascertain MetS status at each study visit for n = 458. These 458 phase 2 participants comprised the current study. There were no differences in parent education or in family history of diabetes between the phase 2 participants included in this study and those excluded. Those who were included were more likely to be female, black, and obese at baseline. Because obese black females are a high-risk group for type 2 diabetes mellitus (T2DM) in adolescence, we see these differences in participation as strengths of the study, even if they limit generalizability.

Measures

A physical examination to measure weight, height, waist circumference, and blood pressure (BP) and phlebotomy after a ≥10-hour fast were performed at each study visit. Details were described previously (7, 22). Subjects were categorized as prepubertal, pubertal (ie, peripubertal), or postpubertal using a validated algorithm based on the relationship between sex steroid concentrations and detailed physical examination performed as part of the National Growth and Health Study (23, 24) and with the use of the estradiol concentration and the presence or absence of menarche for 2 years in females or free testosterone concentrations and axillary hair assessment in males by trained staff (20). Pubertal stage was assessed in Y1 and Y4. At Y4, 97.2% of the cohort was postpubertal; at Y8 and Y10 all subjects were ≥18 years old and were presumed to be postpubertal. Laboratory assays included lipids (high-density lipoprotein [HDL] and triglyceride) measured by standard procedures following the Centers for Disease Control and Prevention (CDC)/National Heart, Lung, and Blood Institute (NHLBI) lipid standardization protocol, and glucose was measured by an enzymatic method using a Hitachi 704 automatic chemistry analyzer. Body mass index (BMI) (kilograms per meter squared) was derived from measured height and weight. Although BMI z scores were calculated for subjects <18 years old, these were not used in longitudinal analyses because previous research has shown that the change in absolute BMI is a more valid measure over time than the change in BMI z score, with the latter measure being highly dependent on baseline BMI (25, 26). Weight categories (normal weight, overweight, or obese) for subjects <18 years were determined using sex-specific growth charts from the CDC (27), with normal weight defined as BMI <85th percentile, overweight defined as BMI ≥85th percentile but <95th percentile, and obese defined as BMI ≥95th percentile. Regardless of age, subjects were considered to be overweight if BMI is ≥25 kg/m² but <30 kg/m², and obese if BMI is ≥30 kg/m² per CDC definitions for adults. The CDC adult definitions were also applied to subjects aged ≥18 years old.

MetS and MetS typology definitions

MetS was defined using the 2005 criteria from the American Heart Association (AHA)/NHLBI (7, 28). Adult AHA/NHLBI criteria were chosen for use at all time points. We applied the adult criteria across all waves because studies have shown that the adult criteria have greater stability than the “adapted” pediatric percentile-based criteria (7, 19), and this definition is recommended by the International Diabetes Federation for those aged ≥16 years old (14). Nearly one quarter (24%) of the cohort was ≥16 year old at Y1 and two thirds (66%) were ≥16 years old at Y4.

Based on MetS status (positive or negative) at each assessment, a MetS typology was determined as follows: never MetS (MetS-negative at all assessments), unstable/remitted MetS (MetS-positive at baseline but MetS-negative by Y10, or fluctuating MetS status throughout the follow-up period), incident MetS (MetS-negative at baseline, developed MetS during follow-up and, once present, the MetS persisted through to Y10), and persistent MetS (MetS-positive at all four assessments). Of note, those who were MetS-negative from Y1 to Y8 and developed MetS at Y10 are considered as incident cases.

Data analysis

Data were analyzed using SPSS (SPSS Inc). Because most variables were not normally distributed, summary statistics for continuous variables are presented as median (interquartile range [IQR]), and between-group comparisons were performed using the Kruskal-Wallis ANOVA for continuous variables and χ² for categorical variables. The change in BMI over the follow-up period was calculated by subtracting Y1 BMI from Y10 BMI. Multinomial logistic regression was used to determine independent correlates of the MetS typology with “never MetS” as the reference group. Two models were assessed: model 1, including baseline age, sex, race/ethnicity, duration of follow-up, baseline pubertal status, and baseline obesity; and model 2, including all factors in model 1 as well as change in BMI. Pairwise interactions between sex, race/ethnicity, baseline obesity, baseline pubertal status, and BMI change were tested and, if significant, were included in model 2. Odds ratios (ORs) and 95% confidence intervals (CIs) are reported from the multivariable models.

Results

The characteristics of the cohort are shown in Table 1. Pubertal development was completed in 53.3% of the cohort at study entry and in 97.2% of the cohort at Y4. Mean age at baseline was 14.3 ± 2.1 years (range, 10.2–19.3 years). At Y10, mean age was 23.4+2.2 years (range, 19.0–29.1 years). Mean length of follow-up was 9.1 ± 0.3 years. During this 10-year period, antihypertension treatment was initiated in 1 subject; no subjects were treated for dysglycemia, diabetes, or hypertriglyceridemia.

Table 1.

Baseline Characteristics of the Cohort

	Entire Cohort (n = 458)	Change in Weight Categorya
		No Change (n = 247, 53.9%)	Change (n = 211, 46.1%)	P Valueb
Female sex	264 (57.6)	148 (59.9)	116 (55.0)	.29
Non-Hispanic black	222 (48.5)	125 (50.6)	97 (46.0)	.32
Family history of diabetes	308 (67.2)	168 (68.0)	140 (66.4)	.70
Baseline pubertal status				.03
Prepubertal	38 (8.3)	28 (11.3)	10 (4.8)
Pubertal	175 (38.3)	87 (35.2)	88 (41.9)
Postpubertal	244 (53.3)	132 (53.4)	112 (53.3)
Weight status
Y1				<.001
Normal weight	260 (56.8)	165 (66.8)	95 (45.0)
Overweight	102 (22.3)	17 (6.9)	85 (40.3)
Obese	96 (21.0)	65 (26.3)	31 (14.7)
Y10				<.001
Normal weight	204 (44.5)	165 (66.8)	39 (18.5)
Overweight	101 (22.1)	17 (6.9)	84 (39.8)
Obese	153 (33.4)	65 (26.3)	88 (41.7)
Baseline anthropomorphics
Weight, kg	60.2 (50.5–73.2)	56.8 (45.3–74.2)	63.1 (56.1–72.7)	.001
Height, cm	162 (157–170)	162 (155–169)	164 (159–171)	.21
BMI, kg/m2	22.4 (19.6–25.9)	20.4 (18.3–27.7)	23.3 (21.8–25.6)	<.001
BMI, z score	0.82 (0.13–1.56)	0.26 (0.21–1.74)	1.10 (0.67–1.51)	<.001

Data are n (%) or median (IQR).

^aChange in weight category over follow-up. Those with no change in weight category were either lean at every assessment (Y1, Y4, Y8, and Y10), overweight at every assessment, or obese at every assessment.

^bP value for comparison between Change vs No Change in Weight Category using Kruskal-Wallis ANOVA for continuous variables and χ² for categorical variables.

At baseline, 21.0% of the subjects were obese; obesity increased over the study period to 33.4% (Figure 1). Table 1 also presents baseline characteristics by whether the subject had a change in weight status over the study period. More than half of the subjects (53.9%, n = 247) had no change in weight status, including 26.3% of whom had persistent obesity and 6.9% who were persistently overweight. Change in weight category was not associated with age, sex, race/ethnicity, or family history of diabetes.

Figure 1.

Prevalence of obesity (gray bars) and MetS (black bars) at each assessment.

MetS prevalence rose from 2.8% at Y1 to 17.9% at Y10 (Figure 1). Slightly less than one quarter of the subjects (23.1%) were MetS-positive at 1 or more study visits. MetS was quite unstable: 61.5% of Y1 MetS-positive cases remitted during follow-up as did 36.4% of Y4 MetS-positive cases and 25% of Y8 MetS-positive cases. Further, 42 incident MetS cases were found in Y10. Thus, the stability of these “new” cases is not known.

The stability of each component of MetS between Y1 and Y10 is shown in Figure 2. The HDL criterion showed the greatest degree of instability, with 42% of the cohort having unstable or remittent low HDL during follow-up. The percentage of subjects with unstable/remitted status for other criteria were as follows: elevated glucose, 8%; elevated waist circumference, 10%; elevated triglyceride, 11%; and elevated BP, 9%.

Figure 2.

Description of the typology of each MetS component. Each component is represented by a bar which is subdivided to represent the proportion for each category: Black, persistent abnormal; gray, incident abnormal; dotted, unstable/remitted; and white, never abnormal. Overall MetS typology is shown in the right-most bar for reference. TGL, triglycerides.

MetS typology was as follows: never, 76.9%; incident, 16.4%; unstable/remitted, 5.7%; and persistent, 1.1%. Thus, among the 106 who were MetS-positive at least once, 70.8% had incident MetS, 24.5% had unstable/remitted MetS, and 4.7% had persistent MetS. Baseline characteristics of the 4 MetS typology groups are shown in Table 2. Age, length of follow-up, sex, race/ethnicity, family history of diabetes, and baseline pubertal status were not significantly different among MetS typology groups. Adiposity measures were the only factors significantly associated with MetS typology in bivariate analyses. BMI increase, which reflects, in part, the expected progression of weight gain during adolescence, was significantly different between MetS typology groups (P < .001 for overall trend). Pairwise comparisons indicated that the incident MetS group had a significantly greater BMI increase (+8.2 kg/m²; IQR, 5.3–10.7 kg/m²) than either the never MetS group (+3.4 kg/m²; IQR, 1.3–6.1 kg/m²) or the unstable/remitted group (+5.2 kg/m²; IQR, 0.4–9.3 kg/m²). No other pairwise comparisons were statistically significant. Although the unstable MetS group showed less BMI gain than the incident group, 77% of the unstable/remitted group had BMI gains from Y1 to Y10. Of 65 subjects who had persistent obesity throughout the entire 10-year period, 15.4% (n = 10) had unstable/remitted MetS. Only 1 of these obese subjects with unstable/remitted typology had decreasing BMI over follow-up, with the other 9 subjects showing BMI gain (median gain, 7.9 kg/m²; IQR, 5.2–10.8 kg/m²). Of note, the BMI change in the persistent MetS group (+9.3 kg/m²; IQR, 2.4–14.8 kg/m²) was higher than that for all other groups but was not significantly different, perhaps due to the small number in this group. BMI changes in the MetS typology groups by baseline obesity status are shown in Figure 3, demonstrating that BMI changed across all MetS typology groups regardless of baseline weight status. Statistical comparisons within each baseline weight status group were not performed because of the limited sample sizes.

Table 2.

MetS Typology Characteristics

	MetS Typology				P Value
	Never MetS (n = 352, 76.9%)	Unstable/Remitted MetS (n = 26, 5.7%)	Incident MetS (n = 75, 16.4%)	Persistent MetS (n = 5, 1.1%)
Female	205 (58.2)	18 (69.2)	37 (49.3)	4 (80.0)	.20
Non-Hispanic black	174 (49.4)	13 (50.0)	33 (44.0)	2 (40.0)	.83
Family history of diabetes	235 (66.8)	18 (69.2)	51 (68.0)	4 (80.0)	.93
Baseline pubertal status					.11
Prepubertal	35 (10.0)	2 (2.7)	0 (0)	1 (20)
Pubertal	137 (39.0)	25 (33.3)	12 (46.2)	1 (20)
Postpubertal	179 (51.0)	48 (64.0)	14 (53.8)	3 (60)
Weight status
Y1					<.001
Normal weight	226 (64.2)	8 (30.8)	25 (33.3)	1 (20.0)
Overweight	76 (21.6)	5 (19.2)	21 (28.0)	0 (0)
Obese	50 (14.2)	13 (50.0)	29 (38.7)	4 (80.0)
Y4					<.001
Normal weight	234 (66.5)	6 (23.1)	17 (22.7)	1 (20)
Overweight	73 (20.7)	6 (23.1)	23 (30.7)	0 (0)
Obese	45 (12.8)	14 (53.8)	35 (46.7)	4 (80)
Y8					<.001
Normal weight	209 (59.4)	4 (15.4)	10 (13.3)	0 (0)
Overweight	80 (22.7)	7 (26.9)	16 (21.3)	1 (20)
Obese	63 (17.9)	15 (57.7)	49 (65.3)	4 (80)
Y10					<.001
Normal weight	194 (55.1)	5 (19.2)	5 (6.7)	0 (0)
Overweight	85 (24.1)	5 (19.2)	10 (13.3)	1 (20.0)
Obese	73 (20.7)	16 (61.5)	60 (80.0)	4 (80.0)
Baseline age, y	14.2 (12.4–15.8)	14.9 (13.5–15.5)	14.9 (13.2–16.4)	15.6 (11.3–16.8)	.08
Length of follow-up, y	9.1 (8.9–9.4)	8.9 (8.7–9.2)	9.0 (8.9–9.5)	9.1 (9.0–9.8)	.18
Baseline BMI z score	0.64 (−0.02 to 1.30)	1.65 (0.89–2.06)	1.45 (0.67–2.01)	2.37 (1.39–2.38)	<.001
BMI, kg/m2
Y1	21.4 (18.9–24.5)	26.5 (22.7–31.2)	24.8 (22.3–29.7)	31.9 (26.3–39.0)	<.001
Y4	22.6 (20.7–26.3)	29.5 (24.4–34.2)	27.8 (25.0–33.6)	38.4 (29.4–41.8)	<.001
Y8	23.9 (21.6–27.8)	32.8 (26.9–36.3)	32.2 (28.0–37.9)	40.3 (30.4–49.6)	<.001
Y10	24.4 (22.0–29.1)	33.3 (26.2–38.3)	33.2 (30.4–39.5)	38.6 (33.4–50.3)	<.001
Change in BMI from Y10 to Y1	3.4 (1.3–6.1)	5.2 (0.4–9.3)	8.2 (5.3–10.7)	9.3 (2.4–14.8)	<.001

Data are n (%) or median (IQR).

^aP value for comparison between MetS typology groups using Kruskal-Wallis ANOVA for continuous variables and χ² for categorical variables.

Figure 3.

Change in BMI during follow-up (BMI at Y10 − BMI at Y1) in each MetS typology group, by baseline weight status. ○, outliers; *, extreme outliers. No statistical comparisons between groups were performed because of limited sample sizes in some subgroups.

Multinomial regression: MetS typology correlates

Results of the multinomial logistic regression modeling for the cohort are shown in Table 3. Non-Hispanic black race/ethnicity was protective against incident MetS compared with never MetS but was not significantly associated with persistent or unstable/remitted MetS. Obesity at baseline significantly increased the odds for persistent MetS, unstable/remitted MetS, and incident MetS. The increase in BMI from Y1 to Y10 also significantly increased the odds of persistent MetS and incident MetS, whereas this association did not reach statistical significance for unstable/remitted MetS. There was a significant interaction between pubertal status and BMI change in the incident MetS group, such that, given the same increase in BMI, the risk for incident MetS was lower in those who were postpubertal compared with those who were peripubertal at baseline. In models stratified by puberty status, the OR for incident MetS for each unit change in BMI was 1.22 (95% CI, 1.12–1.33) for those who were postpubertal and 1.49 (95% CI, 1.30–1.72) for those who were peripubertal.

Table 3.

Independent Correlates of MetS Typology Group From Multinomial Regression Modeling

	OR (95% CIs)a
	Persistent MetS		Unstable/Remitted MetS		Incident MetS
	Model 1	Model 2	Model 1	Model 2	Model 1	Model 2b
Baseline age, y	1.11 (0.59–2.09)	1.27 (0.62–2.60)	1.38 (1.05–1.83)	1.45 (1.08–1.93)	1.13 (0.95–1.35)	1.31 (1.06–1.62)
Duration of follow-up, y	3.09 (0.26–36.35)	3.10 (0.27–36.00)	0.22 (0.05–0.94)	0.23 (0.05–0.97)	0.88 (0.40–1.90)	0.85 (0.36–2.03)
Female	3.73 (0.36–38.69)	2.39 (0.21–27.84)	2.16 (0.86–5.45)	2.10 (0.83–5.33)	0.74 (0.43–1.26)	0.56 (0.30–1.03)
Non-Hispanic black	0.29 (0.04–2.11)	0.20 (0.03–1.68)	0.59 (0.24–1.44)	0.54 (0.22–1.32)	0.57 (0.33–0.98)	0.37 (0.19–0.70)
Postpubertal at baseline	0.66 (0.04–10.24)	0.74 (0.03–16.51)	0.46 (0.14–1.50)	0.46 (0.14–1.50)	1.20 (0.56–2.56)	4.91 (1.16–20.78)
Baseline obese	39.00 (3.74–407.00)	31.36 (2.99–328.98)	7.36 (2.98–18.17)	7.79 (3.12–19.41)	4.33 (2.40–7.81)	4.42 (2.23–8.76)
Change in BMI, kg/m2		1.27 (1.03–1.56)		1.09 (0.99–1.19)		1.49 (1.31–1.71)

^aNever MetS is the reference group.

^bA significant interaction between pubertal status and BMI change was found for the incident MetS group (OR = 0.82, 95% CI = 0.70–0.95; the peripubertal group is the reference group). The ORs and 95% CIs for model 2 for this group are adjusted for this pubertal status × BMI change interaction.

Discussion

This decade-long study of middle- and high-school students in a community setting followed into young adulthood provides novel data regarding the typology of MetS and its relation to weight status during the transition from adolescence to young adulthood. Of MetS cases diagnosed in adolescence, the majority do not persist into young adulthood. Rather, the increasing prevalence of MetS over time is due primarily to incident cases at any one point in time, not persistence of cases within an individual over time. Further, although we demonstrated that BMI gain was associated with incident MetS, we also showed increasing BMI in the vast majority of individuals with unstable/remitted MetS, even among those with persistent obesity. These findings suggest a dissociation between weight status and MetS stability.

With regard to the stability of MetS in adolescence, prior work with the PSD Study cohort had demonstrated that only approximately one half of MetS-positive adolescents retained the diagnosis over a 3-year period (7). We now extend these findings over a longer duration, tripling the follow-up time to 9 years. With increasing follow-up, we have demonstrated that nearly two thirds of Y1 MetS-positive cases remitted on subsequent evaluation. Other studies have also demonstrated significant instability in the diagnosis of pediatric MetS in both very short-term (16) and longer-term stability studies (12, 16). Importantly, these studies used percentile-based definitions of MetS for adolescents <16 years old, and some of the “instability” in MetS could be interpreted as being due to changing definitions over time. In our study, in which our baseline minimum age was 10 years, an accepted age for diagnosing MetS per the International Diabetes Federation (14), we used a consistent definition at all ages and demonstrated a similar rate of MetS instability over long-term follow-up. Taken together, our data and those from other cohorts suggest that a diagnosis of MetS in adolescence has low utility for forecasting future MetS, particularly in the absence of significant weight gain in adolescence. Its predictive value is no better than a coin-flip. Furthermore the instability of MetS and its fluctuating status, which we found even in the context of increasing weight gain, raise significant caution about its use in adolescence. Although weight gain and changes in BMI are expected during adolescence, we found that BMI also increased among those who were obese at baseline, in whom weight stability or weight loss would be ideal, and across all MetS types. These findings refute a relationship between MetS remission and BMI reduction; rather they show that BMI may increase even as MetS is “lost.” We also demonstrated an interaction between pubertal status and change in BMI for incident MetS, which suggests that weight gain during puberty may be particularly important in the development of MetS during the adolescent years. However, because such a high proportion of those with incident MetS developed MetS in Y10 (44%) which may therefore be unstable, this interaction effect should be considered preliminary.

From a clinical perspective, given the instability of MetS overall and in the context of ongoing BMI gain, use of MetS as a diagnostic entity may misclassify adolescents who will not have a high cardiometabolic risk in adulthood, and, more importantly, may provide false reassurance to those who lose the diagnosis but continue to have increasing BMI. Indeed, whereas abundant evidence shows that childhood and adolescent obesity are harbingers of significant cardiovascular morbidity and mortality in adulthood (2, 29–31), few studies support a lasting adult health impact of a pediatric diagnosis of MetS. In a combined analysis of the Bogalusa Heart Study and the Young Finns Study, Magnussen et al (12) found that individuals with pediatric MetS resolving by adulthood showed no increased risk for T2DM or increased carotid intima-media thickness (cIMT) compared with those never having MetS. As expected, those with persistent MetS from childhood to adulthood showed significant increases in the risk of T2DM and elevated cIMT (12), but, importantly, analysis from the same group has demonstrated that the strength of BMI alone to predict adult cIMT and T2DM is similar to the predictive value of MetS (19). These data reinforce the importance of obesity itself in conferring cardiometabolic risk, as does the increasing recognition of obesity as a disease, including the American Medical Association's recent policy decision to consider obesity as such, and highlight the cardiometabolic risks of obesity itself, obviating the need for a separate measure of risk clustering in pediatrics.

Although the utility of MetS as a diagnostic concept in pediatrics is questionable, the importance of assessing the individual components of MetS and providing risk counseling accordingly is well established. Current NHLBI guidelines recommend routine universal screening for lipids in adolescents (once between ages 9 and 11 years and once between 18 and 21 years), with additional screening for those with risk factors and routine biyearly screening of fasting glucose in overweight adolescents with risk factors (32, 33). Previous work has shown that obesity and lipids strongly track from adolescence to adulthood, with more modest stability in measures of BP and glycemia (34–39). Our analysis demonstrates some instability in the components of MetS, in particular, in HDL and triglycerides, which probably drives instability in MetS itself. Unlike the composite diagnosis of MetS, however, component risk factors can be targeted and directly treated. These are important risks to manage in adolescence, because autopsy studies have established that they are associated with the early development and the extent of atherosclerotic lesions in children and young adults (1, 40–42). Our data do not diminish the importance of identifying and treating these cardiovascular risk factors in adolescence, but rather question the utility of clustering these factors into an entity of pediatric MetS, which is no more stable than its individual components, has not shown a clear predictive value for adult disease, and may distract from a focus on body weight and individual metabolic risks.

The current study has important limitations. We do not have multiday measures of blood glucose or BP at each wave of data collection, both of which are known to have day-to-day variation. In addition, by design, only overweight/obese subjects had BP measurements at the first assessment (Y1), which may have led to a slight underestimation of the prevalence of MetS at baseline. However, because MetS is extremely rare in normal-weight subjects, this misclassification risk is low. We also do not have sophisticated measures such as dual energy x-ray absorptiometry scanning, which would provide a better indication of adiposity than BMI and possibly waist circumference. However, because detailed adiposity measures are also not generally available in the clinical setting, we believe that the use of BMI and waist circumference maximizes the clinical relevance of the data. We did not calculate BMI z scores at Y8 and Y10 because a majority of the cohort (64% and 88%, respectively) was >20.5 years old, precluding valid calculation of BMI z score and, therefore, use of BMI z score in longitudinal analyses. However, studies have shown that BMI is a valid measure of adiposity in longitudinal studies of adolescents and provides more interpretable findings than BMI z score (25, 26). The cohort comprised only non-Hispanic blacks and whites; therefore, we cannot be sure that the data are applicable to other racial/ethnic populations. Medication use was recorded at each visit, but we cannot exclude the possibility of intermittent, unreported medication use between visits contributing to the instability of MetS. Finally, the current cohort of 458 subjects, drawn from phase 2 of the PSD Study, had specific inclusion criteria, and were selected on the basis of having available data regarding MetS at all 4 assessments. This may limit generalizability. Despite these limitations, the length of follow-up and comprehensive measures over multiple points in time in a large biracial cohort assessed in the community setting are important strengths of these data.

In sum, this study demonstrates not only substantial instability in MetS but also a fluctuation in MetS status even among those gaining weight as adolescents transition to adulthood. Weight status and the individual components of MetS in adolescence are well-established targets for improving adult health, whereas MetS as a “unifying” entity may have little use in the pediatric age group.