Adiponectin and Lipid Profiles Compared with Insulins in Relation to Early Growth of British South Asian and European Children: The Manchester Children's Growth and Vascular Health Study

Abstract

Context:

Adiponectin, high-density lipoprotein cholesterol (HDL-C) and insulin concentrations may be important in the pathophysiology of cardiovascular disease.

Objective:

We tested the hypothesis that serum adiponectin rather than insulin differs from early life, between South Asians and Europeans, with a potentially key role in excess cardiovascular risk characteristic of adult South Asians.

Design and Participants:

We conducted a longitudinal study of 215 British-born children of European (n = 138) and South Asian (n = 77) origin, from birth to 3 yr.

Main Outcome Measure:

Serum adiponectin, insulin, proinsulin and HDL-C concentrations were assessed in relation to ethnic group and growth in anthropometric variables from 0–3 yr of age.

Results:

Serum adiponectin was lower in South Asian children, despite their smaller size, notable at age 3–6 months (9.5 vs. 11.8 mg/liter; P = 0.04), with no ethnic differences in serum lipids or insulin or proinsulin. In mixed-effects longitudinal models for HDL-C, determinants were adiponectin (P = 0.034), age (P < 0.001), and body mass index (P < 0.001) but not ethnicity. None of these or growth variables affected either insulin or proinsulin. In a fully adjusted mixed-effects longitudinal model including age, sex, insulin, and proinsulin, the independent determinants of serum adiponectin were height [21.3 (95% confidence interval = 31.7–10.8 cm lower, for every 1 mmol/liter increase in adiponectin, P < 0.001], HDL-C [2.8 (1.3–4.2) mmol/liter higher, P < 0.0001], body mass index (lower, P = 0.03), and South Asian ethnicity (lower, P = 0.01).

Conclusions:

These British South Asian-origin infants have lower serum adiponectin but no differences in HDL-C or insulin molecules. In South Asians, factors affecting adiponectin metabolism in early life, rather than insulin resistance, likely determine later excess cardiovascular risk.

People of Indian subcontinental origin (South Asians) have an elevated risk of diabetes, coronary heart disease, and general cardiovascular disease compared with European or most other ethnic groups globally, only partially accounted for by established risk factors (1–6). Some of this disparity has been attributed to an apparent underlying predisposition for central obesity (7, 8), raised triglycerides, and low high-density lipoprotein cholesterol (HDL-C) (1). The adipokine adiponectin, a close associate of the latter risk factor, is lower in adult South Asian populations (9) and may be important in the pathophysiology of susceptibility to diabetes and cardiovascular disease.

The atheroprotective properties of adiponectin in adults are well established, particularly its antiinflammatory (10, 11) and insulin-sensitizing effects (12, 13). Experimental data suggest an important role for adiponectin in improving insulin sensitivity and glucose tolerance (14), notably preceding their decline in vivo (13, 14). Adiponectin is positively and independently associated with HDL-C (15–18), and its effects may be mediated through a lowering of hepatic lipase activity (19) and decreasing HDL-C catabolism (20). Whether hypoadiponectinemia contributes to the low HDL-C observed in South Asians is unclear. This is an important question because HDL-C is inversely associated with increased coronary heart disease risk independent of high low-density lipoprotein cholesterol (LDL-C) and triglycerides (21).

It is not clear whether British-born South Asians share the same risk patterns as their counterparts born in the Indian subcontinent. Similarly, the much-highlighted role of apparent hyperinsulinemia as a key pathway for developing cardiovascular disease, as opposed to diabetes, remains controversial (22, 23). Comparison of risk factors in ethnic groups reared in similar environments may help shed light on the etiology of both the disease and ethnic risk patterns. Work in India suggests that the South Asian phenotype of central adiposity and glucose intolerance may already be present at birth (24, 25); however, there seem to be no prospective data comparing metabolic risk factors, particularly adiponectin and lipids, in South Asian infants and children born in the United Kingdom. We previously observed that lipid levels were similar, whereas adiponectin was slightly lower in a small sample of British-born South Asian and European newborns (26). Here we present data, from a larger sample of our mixed longitudinal cohort of these South Asian and European infants and children, which investigate the determinants of serum adiponectin and its role in the characteristic low HDL-C and similar determinants of serum insulin in South Asians, followed up to 3 yr from birth. We test the hypothesis that South Asian ethnicity will be associated with higher insulin only after lower concentrations of adiponectin, and the latter will also determine, in part, these populations' typical lower HDL-C levels and excess cardiovascular and diabetes risk.

Subjects and Methods

Participants

From a series of mothers screened in pregnancy in Manchester (UK) and who agreed to allow their babies into the study, 215 infants and children of European and South Asian (mostly Pakistani) origin (138 and 77, respectively) were measured at birth and then at one or more of the time points 3, 6, 12, 24, and 36 months (participants seen at two or more time points n = 209) with 673 total observations for measures of anthropometry (length, weight, and subscapular and triceps skinfold measurements). More infants of South Asian origin had unemployed fathers (32%) compared with those of European babies (18%).

Blood samples, fasting as best as possible (>3 h), were measured for serum lipids, adiponectin, insulin, and proinsulin. Ethnicity was defined as having at least three grandparents of respective origins, cross-referenced against self-defined groups from Britain's 2001 census categories. Carefully standardized measures of neonatal and child anthropometry were taken by trained research nurses at birth and postnatally. All children were from healthy pregnancies without fetal distress and maternal complications. (Pregnancies with complications such as preeclampsia and gestational diabetes mellitus were excluded from the cohort.) The Central Manchester Ethics Committee approved the study, and written informed parental consent was obtained before enrollment.

Infant and child anthropometry

Birth measures were carried out at delivery in the Manchester Royal Infirmary and further follow-up measures at the Wellcome Trust Clinical Research Facility in Manchester, alongside and affiliated to the hospital. An infant electronic scale was used to obtain weight at birth and during infancy (grams) and in children using sitting electronic scales (kilograms). Birth length (centimeters) was obtained using a standardized infant plastic length board and measured supine in infants and standing in children using a fixed stadiometer. Suprailiac, subscapular, and triceps skinfold measurements were obtained using Harpenden skinfold calipers (BATY International, Burgess Hill, West Sussex, UK) and recorded to the nearest 0.1 mm. Quality assurance checks showed that interobserver difference in skinfold measurements averaged 10%, and 88% of triceps and 93% of subscapular duplicates were within 10% of each other. Subscapular to tricep ratios were calculated as a measure of truncal to peripheral adiposity. A trained research nurse carried out all child anthropometric measurements, and the mean of two measures was used in the analysis.

Adiponectin assay

Serum total adiponectin was measured using the DuoSet ELISA development system (R&D Systems, Minneapolis, MN). The lower limit of detection was 1 μg/liter, and the intraassay and interassay coefficients of variation were 6.8 and 10.2%, respectively.

Insulin and proinsulin assay

Both insulin and proinsulin were measured by monoclonal-based ELISA (Diagenics Ltd., Bletchley UK). The lower limit of detection for insulin was less than 6 pmol/liter and for proinsulin was less than 0.5 pmol/liter with intra- and interassay coefficients of variation of less than 5 and 6%, respectively. There is less than 0.1% cross-reactivity between these monoclonal assays.

Cholesterol assay

Serum total cholesterol and triglycerides were measured by the cholesterol esterase/cholesterol oxidase/4-aminoantipyreneperoxidase (enzymic CHOD/PAP) and lipoprotein lipase/glycerokinase/glycerol-3-phosphate oxidase/peroxidase (enzymic GPO/PAP) methods, respectively, on a Cobas Mira S analyzer (ABX Diagnostics, Shefford, UK), and all reagents were obtained from the same source. HDL-C was measured by a second-generation homogeneous method using polyethylene glycol-modified enzymes (Roche Diagnostics, Lewes, UK). LDL-C was calculated using the Friedewald formula. Calculated LDL-C of less than 0.1 mmol/liter was set as the detection limit for cholesterol of 0.1 mmol/liter.

Statistical methods and data analysis

Statistical analysis was carried out in Stata version 10.2 (Stata Corp., College Station, TX). Data are presented as arithmetic means and sd or geometric means and 95% confidence intervals (CI). Skewed data (insulin and proinsulin) were Ln transformed before analyses. ANOVA were used to test for statistically significant differences between ethnic groups. Body mass index (BMI)-adjusted means for subscapular thickness were estimated from linear regressions at each time point using the Stata module adjmean. Random effects generalized least squares models (Stata module xtreg) were used assess the effect of time and ethnicity on the metabolic variables.

We used general linear mixed-effects regression models (Stata module xtmixed) to estimate longitudinally the effect of infant and child age, sex, BMI, length (or height), and ethnicity as independent variables on levels of lipid (HDL-C), insulin/proinsulin, and adiponectin as dependent variables. The general linear mixed model (mixed model) can be used to describe nonlinear relationships across time in a longitudinal dataset with multiple missing data points. The se of the regression estimates from the models is adjusted to account for participants who contribute data at several time points. The models take into account that the observations on these subjects are not independent.

Results

Anthropometric characteristics show that birth weight (mean difference = −0.377 kg, 95% CI = −0.52 to −0.23 kg; P < 0.001) and length (−1.57 cm, 95% CI = −2.23 to −0.91 cm; P < 0.001) were significantly lower in South Asian babies, however, with no comparative reduction in skinfold size. A similar excess (P < 0.05) of subscapular skinfold in relation to weight was also apparent in South Asians at birth and 3, 6, and 24 months (Tables 1 and 2). In separate analyses, BMI-adjusted subscapular skinfolds were significantly greater in South Asian children at 3 months (6.2 vs. 7.1 mm; F = 7.0; P = 0.01) and 24 months (5.9 vs. 5.2 mm; F = 7.7; P = 0.008) but not at birth.

Table 1.

Anthropometric characteristics at birth: the Manchester Children's Growth and Vascular Health study

	European (138) (males = 89)	South Asian (77) (males = 50)
Birth weight (g)	3402 (478)	3025 (477)a
Birth length (cm)	50.9 (2.1)	49.4 (2.3)a
Ponderal index (kg/m3)	25.6 (2.6)	25.0 (2.7)
BMI (kg/m2)	13.1 (1.33)	12.34 (1.32)a
Tricep (mm)	4.2 (0.9)	3.9 (0.9)
Subscapular (mm)	4.1 (0.9)	3.9 (1.0)
Sub/tricep ratio	0.98 (0.16)	0.99 (0.17)

The data presented are from infants at birth and are reported as mean (sd). ANOVA was used to test for ethnic differences.

^aP < 0.001.

Table 2.

Infant and child anthropometric characteristics: the Manchester Children's Growth and Vascular Health study

	3 months, n = 45		6 months, n = 24		12 months, n = 67		24 months, n = 66		36 months, n = 59
	E (22)	SA (17)	E (17)	SA (7)	E (43)	SA (24)	E (44)	SA (22)	E (37)	SA (22)
Weight (kg)	6.4 (0.7)	6.3 (0.9)	8.1 (0.8)	7.5 (0.6)	10.3 (1.1)	9.3b (1.3)	13.0 (1.3)	11.7b (1.6)	15.5 (1.9)	14.0b (1.9)
Length (cm)	62.2 (2.7)	61.9 (3.4)	68 (3.5)	66.5 (2.8)	77.5 (3.3)	76.0 (3.4)	87.5 (2.8)	85.9a (3.3)	96.6 (3.7)	94.9 (3.1)
Ponderal index (kg/m3)	26.6 (2.8)	26.9 (3.4)	25.4 (2.6)	25.4(2.2)	22.3 (2.2)	21.3 (2.1)	19.3 (1.6)	18.4a (2.1)	17.2 (1.4)	16.2a (1.2)
BMI (kg/m2)	16.5 (1.5)	16.5 (1.7)	17.3 (1.3)	16.9 (1.0)	17.2 (1.4)	16.1a (1.6)	16.9 (1.3)	15.8b (1.7)	16.6 (1.4)	15.3b (1.4)
Subscapular (mm)	6.3 (1.5)	7.3b (1.3)	6.5 (1.0)	6.9 (1.2)	6.5 (1.7)	6.1 (1.5)	5.5 (0.8)	5.8 (1.9)	6.1 (1.9)	5.5 (1.6)
Triceps (mm)	6.5 (2.2)	6.6 (2.2)	7.9 (3.0)	7.8 (3.3)	6.9 (2.2)	6.4 (2.3)	7.0 (2.0)	6.3 (1.6)	8.7 (1.9)	7.3a (2.1)
Subscapular/triceps ratio	1.10 (0.28)	1.17 (0.30)	0.90 (0.30)	1.00 (0.32)	0.99 (0.25)	1.02 (0.29)	0.86 (0.25)	0.91 (0.20)	0.71 (0.29)	0.74 (0.15)

Data are mean mean (sd). ANOVA was used to test for ethnic differences. E, European origin; SA, South Asian origin.

^aP < 0.05.

^bP < 0.01.

Using random-effects generalized least squares models, mean lipoproteins concentrations were similar between ethnic groups (Table 3). Significant reductions (P < 0.0001, independent of ethnicity) from birth were found for triglycerides, HDL-C, and LDL-C as well as Ln insulin and Ln proinsulin. There was no significant change in total cholesterol over time. Serum adiponectin declined over time (P < 0.0001) with the average at each time point lower in South Asians but not statistically significant (P = 0.058).

Table 3.

Infant and child metabolic characteristics: the Manchester Children's Growth and Vascular Health study

	3 and 6 months, n = 69		12 months, n = 67		24 months, n = 66		36 months, n = 59
	E (45)	SA (24)	E (43)	SA (24)	E (44)	SA (22)	E (37)	SA (22)
Adiponectin (mg/liter)	11.8 (4.5)	9.5a (3.6)	7.0 (2.1)	6.8 (2.1)	6.0 (1.3)	5.8 (2.3)	5.2 (1.8)	5.1 (1.8)
Triglyceride (mmol/liter)	1.6 (0.7)	1.3a (0.6)	1.4 (0.8)	1.4 (0.9)	1.2 (0.5)	1.1 (0.4)	0.8 (0.3)	0.9 (0.3)
Cholesterol (mmol/liter
Total	3.9 (0.7)	4.1 (0.9)	4.1 (1.1)	3.6 (1.2)	4.1 (0.7)	4.3 (0.8)	4.1 (0.8)	4.3 (0.6)
HDL	1.2 (0.3)	1.2 (0.4)	1.2 (0.4)	1.1 (0.3)	1.3 (0.3)	1.3 (0.3)	1.4 (0.3)	1.4 (0.3)
LDL	2.0 (0.6)	2.3 (0.9)	2.3 (0.9)	1.9 (1.0)	2.3 (0.6)	2.6 (0.8)	2.4 (0.7)	2.5 (0.5)
Insulinb (pmol/liter)	6.5 (4.1, 10.4)	5.1 (3.3, 7.7)	3.2 (2.3, 4.5)	2.5 (1.6, 3.9)	3.1 (2.2, 4.1)	3.5 (2.7, 4.7)	3.6 (1.6, 7.9)	5.2 (3.0, 9.0)
Proinsulinb (pmol/liter)	10.9 (7.8,15.1)	10.6 (7.9, 14.3)	4.5 (3.1, 6.6)	5.2 (3.2, 8.5)	5.0 (3.9, 6.5)	4.9 (3.1, 7.6)	4.5 (2.9, 7.1)	4.6 (2.5, 8.8)

Data are mean (sd) or geometric mean (95% CI). E, European origin; SA, South Asian origin.

^aP < 0.05.

^bLn transformed before analyses.

Determinants of serum HDL-C, triglycerides, serum insulin, and proinsulin

In an age-, ethnicity-, and sex-adjusted model, adiponectin was positively associated with HDL-C, length independent of weight, and BMI (Table 4, model 1). There were no ethnic differences among the children at 3 months through to 3 yr, in either insulin or proinsulin (Table 3). Adiponectin was not a determinant of either insulin or proinsulin in these models. In the adjusted analyses, none of the growth variables or ethnicity was associated with an increment in serum insulin (Table 4, model 3), whereas for proinsulin, length and age were positive predictors (Table 4, model 4).

Table 4.

Mixed longitudinal regression analyses of determinants of HDL-C (model 1), triglycerides (model 2), insulin (model 3), and proinsulin (model 4) in infants and children of South Asian and European origin: the Manchester Children's Growth and Vascular Health study

	Dependent variables
	HDL-C (model 1)		Triglycerides (model 2)		Ln insulin (model 3)		Ln Proinsulin (model 4)
	β	95% CI	β	95% CI	β	95% CI	β	95% CI
Explanatory variables
Age (months)	0.02	−0.09 to 0.14	−0.32a	−0.57 to −0.08	−0.84a	−1.52 to −0.16	0.30	−0.07 to 0.68
Sexd	0.02	−0.07 to 0.11	−0.08	−0.26 to 0.10	0.09	−0.42 to 0.60	−0.13	−4.1 to 0.43
BMI (kg/m2)	0.04b	0.02 to 0.07	0.50	0.00 to 0.10	0.01	−0.17 to 0.19	−0.01	−0.10 to 0.09
Length/height (cm)	1.31b	0.37 to 2.25	1.11	−0.80 to 3.03	4.39	−1.61 to 10.40	−4.07a	−7.32 to −0.82
Ethnicitye	0.05	−0.06 to 0.15	0.01	−0.19 to 0.20	0.30	−0.28 to 0.89	−0.19	−0.51 to 0.12
Adiponectin (mg/liter)	0.04c	0.02 to 0.05	−0.01	−0.04 to 0.02	−0.004	−0.09 to 0.08	0.01	−0.03 to 0.06

Data are β-coefficient and 95% CI. Insulin and proinsulin are Ln transformed before inclusion as dependent variables in models. P values are for independent variables in model values.

^aP < 0.05.

^bP < 0.01.

^cP < 0.001.

^dFemales are reference group in model.

^eEuropeans are reference group in model.

Determinants of serum adiponectin

To explore the determinants of serum adiponectin in relation to ethnicity, lipids, and anthropometry, data from all time points were analyzed together in an age- and sex-adjusted mixed longitudinal regression analysis. In initial analyses adjusted for age, sex, and ethnicity, birth characteristics including triglycerides, LDL-C, and skinfold measures (subscapular, triceps, and subscapular to triceps ratio) were not associated with adiponectin. In a separate model adjusted for sex and age, South Asian ethnicity was negatively associated with adiponectin (β = −1.24; 95% CI = −2.18 to −0.30; P = 0.01). Further adjustment for BMI and skinfolds did not attenuate the ethnic effect. In the final age- and sex-adjusted model (Table 5), length, BMI, HDL-C, and ethnicity were significantly and independently (of insulin and proinsulin) associated with serum total adiponectin. Including triglycerides or LDL-C in this model in separate analyses (data not shown) did not materially change the results.

Table 5.

Mixed longitudinal regression analysis of determinants of serum adiponectin: the Manchester Children's Growth and Vascular Health study

Adiponectin	β	95% CI	P value
Length/height (per cm)	−21.26	−31.74 to −10.78	<0.001
HDL-C (per mmol/liter)	2.77	1.35 to 4.18	<0.001
Ethnicitya	−1.36	−2.44 to −0.27	0.014
BMI (kg/m2)	−0.35	−0.68 to −0.03	0.035
Ln proinsulin (per unit mmol/liter)	0.18	−0.37 to 0.72	0.518
Ln insulin (per unit mmol/liter)	−0.07	−0.36 to 0.22	0.623
Sexb	0.20	−0.76 to 1.17	0.683
Age (month)	0.23	−1.02 to 1.48	0.720

Data are β-coefficients and 95% CI.

^aEuropeans are reference group in model.

^bFemales are reference group in model.

Discussion

As with our previous findings in newborns (26), there were no ethnic differences here in lipid profile; however, overall, adiponectin concentrations were lower in South Asian children. Consistent with data in adults (27), South Asian ethnicity was a highly significant and negative predictor of adiponectin, even after adjustment for BMI, truncal to peripheral ratio, and skinfold thickness.

Brown adipocytes express adiponectin and are thought to be the source of abundant levels of adiponectin at birth (28). This may be relevant to the ethnic difference here in adiponectin, but brown fat could not be measured here, and the suggestion remains speculative. Teleologically, the requirement for brown adipose tissue to prevent neonatal hypothermia would be less in the South Asian as opposed to the North European environment.

Similar models for insulin and proinsulin showed that ethnicity, age, sex, skinfolds, adiponectin, and BMI were not associated with serum levels of these hormones (Table 4). Higher levels of insulin have been observed at birth in Indian babies born in India (25) as compared with those in European origin babies (in the UK). Whincup et al. (29) observed higher insulin levels in 10-yr-old South Asian children compared with Europeans in Britain, demonstrating that the metabolic environment may already be impaired. However, our data suggest no such excess in early life, despite these relatively small numbers, which should still be adequate for comparison over time. Neither insulin nor proinsulin concentrations differed by ethnic group here, nor did prediction models over the first 3 yr of life show body mass or central fat estimates from skinfolds to be related to insulin. The proinsulin result suggested a powerful inverse effect of length or height so that taller children had lower levels. The implication of this is unclear but may reflect better general growth resulting in lower circulating proinsulin. The infants and follow-up children here were all studied in the same setting. Experimental data also suggest that low circulating adiponectin levels precede insulin resistance (30, 31). Mice lacking adiponectin display severe hepatic insulin resistance and develop glucose intolerance in response to a high-fat diet (14).

Adiponectin is an important contributor to peroxisome proliferator-activated receptor-γ-mediated improvements in glucose tolerance through mechanisms that involve the AMP-activated protein kinase pathway (14). Consistently, and including a systematic review (32), adiponectin is an independent predictor of type 2 diabetes in adult Pima Indians (33) and South Asians (34). One hypothesis to be drawn is that low adiponectin precedes and therefore may be driving the low HDL-C as well as insulin resistance and thus hyperinsulinemia in South Asians from early life. What leads to these low HDL levels remains unknown, although perhaps fat oxidation is a key next step. Despite lower birth weight and length, skinfolds, particularly subscapular to triceps ratios as indirect indicators of central fat tendency here did not differ at birth, resulting in a disproportionately higher ratio of subscapular and tricep skinfold thickness relative to weight at birth in South Asians compared with European babies. A higher subscapular to weight ratio was also apparent in South Asians at 3, 6, and 24 months. Weight and length had caught up with the European-origin infants by 3 months, still with no skinfold difference. How good these indices are for central fat estimation at these ages is unknown.

There were no ethnic differences in HDL-C, demonstrating that ethnic lipid disparities observed in adults are not yet present in children of this age, at least not in those reared in similar environments as here. Thus, any later differences are probably induced by lifestyle factors. Lower HDL-C has been related to diet (35) and low physical activity (36), and the dyslipidemia consistently observed in South Asian adults has been ascribed to increased insulin resistance (37). However, again, the data found here suggest, to the contrary, that lower adiponectin precedes both lipid and any insulin changes. The apparent later insulin resistance is thus a secondary result, not a primary cause, of the dyslipidemia. Low levels of circulating plasma HDL-C appear to be a result of increased catabolism as opposed to decreased production and triglyceride enrichment of HDL-C plays a key role in HDL-C catabolism rates (38). in vivo studies demonstrated increased apolipoprotein A-I catabolic rates in hypertriglyceridemic individuals with no major changes in production. The enhanced remodeling of HDL-C due to cholesterol ester transfer protein-mediated triglyceride enrichment and increased hepatic lipase activity in insulin resistance is thought to be responsible for the increased HDL-C catabolism (39).

Although the altered lipoprotein lipase activity may directly explain the association of low adiponectin with increased triglyceride levels, the effect of down-regulation of lipoprotein lipase would be to limit the release of essential HDL-C components, such as the cholesterol apolipoproteins and phospholipids from the surface of triglyceride-rich lipoproteins. Whatever the explanation for the effect of adiponectin on lipoprotein metabolism in adults, it is clear from our study that at birth and in early infancy, whereas adiponectin levels are already lower in South Asians, other factors must be of more overwhelming importance in determining lipoprotein levels. The most obvious factor yet to emerge is a difference in, and perhaps distribution of, body weight, particularly in visceral adipose tissue, which would increase hepatic very-low-density lipoprotein production and perhaps make adiponectin a more important determinant of triglyceride and HDL-C levels.

Study limitations

Limitations of our study include relatively small numbers of South Asian infants and their longitudinal follow-up and, by definition, the limits of interpreting blood levels for physiological changes arising in tissues. Yet blood levels are all that can be studied longitudinally in children. The sample of infants studied may not be representative of South Asian infants born in the United Kingdom or even of those of Pakistani origin predominant in the local population, because the study was a follow-up of a local maternal screening program with dropout in and after pregnancy. However, anthropometric characteristics, particularly birth size of these South Asian babies are similar to those previously found. To our knowledge, these are the first British data available in this ethnic group for lipids or insulins over the first 3 yr of life. The lack of difference in serum insulin at this age may be due to insufficient numbers to detect a difference as well as differences in fasting times. The physiological benefits of adiponectin depend on the distribution of its various isoforms and its high molecular weight multimer (HMW) has emerged as the most significant in relation to its insulin-sensitizing and antiinflammatory actions (40, 41). Studies suggest that low levels of HMW adiponectin may underlie the links between low blood adiponectin and type 2 diabetes (40, 42), this variation has specifically been found in South Asian women (42, 43). Whether HMW adiponectin is reduced in our young population of British South Asian children remains to be determined.

Ongoing follow up of our cohort and investigation into the change in adiponectin, its HMW, and insulin levels with lipid profiles, in relation to lifestyle factors, will provide further information on the determinants and consequences of the early metabolic environment and how these mediate later disparities in cardiovascular and diabetes risk.