Acta Diabetologica

, Volume 52, Issue 1, pp 167–174

Levels of adiponectin and leptin at onset of type 1 diabetes have changed over time in children and adolescents


    • Steno Diabetes Center
  • Stefanie Eising
    • Hørup Lægehus
  • David Michael Hougaard
    • Center for Neonatal Screening, Department of Clinical Biochemistry, Immunology and GeneticsStatens Serum Institut
  • Henrik Bindesbøl Mortensen
    • Herlev Hospital, Department of PediatricsUniversity of Copenhagen
  • Kristin Skogstrand
    • Center for Neonatal Screening, Department of Clinical Biochemistry, Immunology and GeneticsStatens Serum Institut
  • Flemming Pociot
    • Herlev Hospital, Department of PediatricsUniversity of Copenhagen
  • Jesper Johannesen
    • Herlev Hospital, Department of PediatricsUniversity of Copenhagen
  • Jannet Svensson
    • Herlev Hospital, Department of PediatricsUniversity of Copenhagen
Original Article

DOI: 10.1007/s00592-014-0630-y

Cite this article as:
Safai, N., Eising, S., Hougaard, D.M. et al. Acta Diabetol (2015) 52: 167. doi:10.1007/s00592-014-0630-y


Adiponectin and leptin are proteins secreted by the adipose tissue and have an influence on insulin sensitivity and on inflammatory markers. Altered levels could play a part in the pathogenesis of type 1 diabetes mellitus. We determined adiponectin and leptin levels over a nine-year period in children with type 1 diabetes mellitus (T1D) in relation to the increasing incidence of T1D, and studied the impact of patient status, age, gender and body mass index (BMI). Data were derived from a population-based registry of diabetic children (DanDiabKids) from 1997 to 2005. Children with newly diagnosed T1D (n = 482) were included, and healthy siblings (n = 479) were chosen as a control group. Leptin levels were significantly higher in recent years (in both patients and siblings), whereas for adiponectin, the levels were lower in recent years in the patient group. Leptin levels were lower in children with T1D (RR 0.74, p = 0.003) and in males (RR 0.52, p < 0.001) and increasing with age in both groups. For adiponectin, there was a negative association between level and age in patients. Both adipokines showed a significant correlation with BMI and lower levels in children with blood samples taken within the first 2 days after initiation of insulin treatment. There has been a change in leptin and adiponectin levels in children with or without T1D from 1997 to 2005. This is not explained by changes in BMI and may reflect changes in other factors like diet or physical activity.


Diabetes in childhoodPathophysiologyAdipose tissue biologyAdipocytokines


The incidence of type 1 diabetes mellitus (T1D) has been increasing worldwide for approximately 40 years, with an average increase of 3 % per year [1]. The aetiology is not entirely known, but T1D results from decreased insulin production by the pancreatic β cells. β cell deficiency is due to autoimmune process or, in some cases, due to idiopathic destruction. There is increasing evidence that the adipose tissue and immunologic processes are closely linked. Adipokines are proteins secreted by the adipose tissue. Leptin and adiponectin are two of the adipokines that have been studied and found to play an important role in regulation of lipid and carbohydrate metabolism. Adiponectin is exclusively synthesised in white adipose tissue, and its levels are paradoxically lower in obese than in lean humans [2, 3]. Adiponectin increases insulin sensitivity in both the liver and skeletal muscle and has potent immuno-suppressive properties, as it induces the production of the anti-inflammatory mediators IL-10 and IL-1RA in primary human monocytes, monocyte-derived macrophages and dendritic cells [4]. Low adiponectin levels are associated with an increased risk of atherosclerotic disease like coronary artery disease [57]. Leptin decreases appetite, increases energy expenditure, suppresses insulin synthesis and secretion and increases insulin sensitivity [8]. Furthermore, recent research shows that leptin regulates the immune response, both innate and adaptive response [9]. Changes in these two proteins could reflect environmental changes, e.g. obesity, diet, physical activity or infections.

In the present study, our aim was therefore to compare these two adipokines in newly diagnosed children with T1D with a group of siblings and test for an altered adipokine phenotype over time. The relatively large dataset allows us to further elucidate the influence of gender, age and season.

Materials and methods

Data for this study were derived from a population-based registry of diabetic children, initiated in 1996 (DanDiabKids) and contains information on more than 4,000 newly diagnosed cases of T1D aged 0–15 years. There is an associated biological bank comprising blood samples from approximately 75 % of all of the children and includes their first degree relatives including more than 1,500 siblings <15 years of age. For this study, a random sample of approximately 500 cases with blood samples, taken less than three months after onset, were chosen. The study population has been described in previous publications [10, 11]. The date of onset was defined as the date of the first insulin injection. Diabetes duration was measured in months. Five hundred siblings, with similar age, season and sample year representing the entire study period, were chosen as controls. Blood samples were taken within one month from diagnosis in 90 % of the samples in both patients and siblings from the same family. The siblings were included if they have a sibling diagnosed with diabetes before the age of 15.

There were 482 patients in the study of which 255 (53 %) were males; of the 479 siblings, 266 (56 %) were males. Eighteen out of 482 cases were immigrants, 13 had unreported ethnicity and the remainder reported being of Danish origin. The ethnicity of siblings is not reported but, given that only siblings with the same father and mother as the patients were included, the distribution of ethnicity is assumed the same in the sibling group. The sample included 203 complete sibships with at least one case and one non-case, comprising 434 children in total (for details see [10]). In children with diabetes, the mean age at blood sampling was 9.83 years; the range was 1.0–18.3 years; in siblings, the mean age was 10.2 and the range 0.4–16.0 years of age. We did not have information about their hormonal status.

Body mass index (BMI) was calculated as weight in kilograms/height in meters2. The BMI Z score was calculated based on Danish reference values [12]. We were able to calculate BMI on 280 patients, where we had available BMI data ±30 days from blood sample. We assume that there is no significant change in their BMI within that short period, the weight measured at onset in the database is the weight at discharge and therefore not influenced by dehydration.

Serum samples were stored at −80 °C during the entire study period from 1997 to 2005 and only thawed for analysis once.


Leptin was analysed using the commercially available high-capacity Luminex xMAP technology, developed in-house as described in Skogstrand et al. [13]. Adiponectin was analysed in a 3-plex analysis of serum samples that were further diluted because of the high concentration of the analyte. The assay detects an epitope in the globular domain of human adiponectin, and cross-reactivity tests have been made without any reactions. The working range for the Luminex assay was defined as the concentration range for the analyte within which the coefficient of variation (standard deviation of repeated measurements divided by the mean) was <20 %. The mean intra-assay variation and inter-assay variation were 11 and 17 %, respectively.

There were no values below the detection limit. The highest concentration in the working range was for leptin 4,000 pg/ml.

Antibodies against glutamic acid decarboxylase (GAD65A) and insulinoma-associated antigen-2A (IA-2AA) were analysed as previously described [10, 14] (Clinical Research Center, Lund University, Malmö, Sweden). Antibody positivity was defined as a value above 31 U/ml for GAD65A and 5 U/ml for IA-2AA.

Statistical methods

To meet the assumption of the statistical model (normally distributed residuals), the levels of the biomarkers were log-transformed before analysis. Therefore, the results are reported as relative change instead of absolute differences.

Log-adipokine levels were modelled based on a multiple linear regression analysis for the effects of gender, age at sampling, case status (case/sibling), date of sampling and season of sampling (time of year). The season of sampling was divided into four seasons: winter (December, January, February); spring (March, April, May); summer (June, July, August); and autumn (September, October, November). BMI was only available in a subsample of patients, and therefore, it was not included in the models. The following models were used:

Model 1a:
$$ \log ({\text{adipokine}}) = \beta_{0} + \beta_{1} *{\text{sex}} + \beta_{2} *{\text{age}} + \beta_{3} *{\text{case}}\_{\text{status}} + \beta_{4} *{\text{sam - date}} + \beta_{5} *{\text{sea}} + \beta_{6} *{\text{fam - id}} $$
Model 1b:
$$ \log ({\text{adipokine}}) = \beta_{0} + \beta_{1} *{\text{sex}} + \beta_{2} *{\text{age}} + \beta_{3} *{\text{case}}\_{\text{status}} + \beta_{4} *{\text{sam - date}} + \beta_{5} *{\text{auto}} + \beta_{6} *{\text{sea}} + \beta_{7} *{\text{fam - id}} $$
Model 2:
$$ \log ({\text{adipokine}}) = a*{\text{sex}} + b*{\text{age}} + c*{\text{case}}\_{\text{status}} + d*{\text{assay}} + e*{\text{sam - date}} + f*{\text{sea}} $$
Sea is the variation of four seasons and sam-date the date of blood sample. Age and sampling date represent continuous variables, whereas gender and case status are character variables. Auto is the autoantibody status (positivity for GAD65 or IA-2 autoantibodies). A given estimate of, e.g., gender is the difference between children with the same patient status, age and sample date and in model with autoantibodies the same autoantibody status.

Since cases and siblings share genes, a fixed effect of family was included as a proxy measure of the genetic profile. There are no adjustments for multiple testing, as the goal of this study was to estimate a small set of a priori-defined effects.

The study was performed in accordance with the Helsinki II Declaration and was approved by Danish Ethical Committee H-KA-20070009. All patients, their parents or guardians gave informed consent.


Changes over time

We found a tendency towards lower adiponectin levels in both patients and siblings in recent years, but only significantly in patients RR 0.98 (p = 0.009) (Fig. 1), while a significant year-to-year increase in leptin levels was seen in recent years in both patients and siblings with RR 1.1 (95 % confidence interval (CI) 1.09–1.14) p < 0.0001 (Fig. 1). In the subpopulation where BMI was available, we have adjusted for BMI. This did not alter the results. There was no significant seasonal variation in leptin or adiponectin.
Fig. 1

Changes in adipokines with period and age. Adiponectin decreases significantly with age in both patients and siblings. Leptin levels increase with age. Adiponectin has a tendency of decreasing through the years, but this is only significant in patients. A significant increase is seen in leptin levels from 1997 to 2005. Stars and dotted line patients, black dots and full line siblings

Influence of age

Adiponectin levels decreased by age RR 0.97 (95 % CI 0.96–0.98) p < 0.0001 in the joined analysis and when looking at each group separately (p < 0.0001) (Fig. 1).

Leptin levels increased significantly with age in both siblings RR 1.07 (95 % CI 1.04–1.10) and children with T1D RR 1.11 (95 % CI 1.09–1.14) p < 0.0001 (Fig. 1).

Patients versus siblings

In our study, differences in adiponectin levels between case and siblings were not demonstrable, but we found leptin levels to be significantly lower in patients RR 0.74 (95 % CI 0.60–0.90) p = 0.003 (Fig. 2).
Fig. 2

Adipokines according to case status and gender. This figure shows the adipokine levels in siblings and patients in the two upper diagrams and levels according to gender in the two lower diagrams. The bold black line is the median, and the thin grey line marks the 10th and 90th percentile. No difference is seen in adiponectin levels between patients and siblings. Leptin levels are significantly lower in the case group (p = 0.003). No significant difference is observed between males and females in adiponectin levels. Leptin levels are significantly higher in females than in males (p < 0.0001)

Changes during the first three months after diagnosis

To explore the influence of insulin treatment and diabetes duration, patients were tested for differences according to the blood samples taken on day 0, 1, 2, 3, 4, 5, 6–14 and beyond 2 weeks after onset. For both adipokines, we found significantly lower levels in those tested the first two days compared to those children with samples taken later (p < 0.0001). Leptin levels being 68 % lower and adiponectin 23 % lower in children with samples taken early.

Influence of gender

No significant difference in adiponectin concentrations was observed between females and males in both groups (Fig. 2).

Males had significantly lower levels of leptin than females RR 0.52 (95 % CI 0.47–0.59) p < 0.0001 (Fig. 2).

Influence of BMI

In 280 patients, BMI measures within one month of the blood sampling were available. In those individuals, the mean BMI z score was −0.12 (SD 1.46) and ranged from −4.57 to 4.02.

Adiponectin showed a negative correlation with BMI z score RR 0.92 (95 % CI 0.89–0.96) p < 0.001, whereas leptin showed the opposite, a significant positive correlation with BMI z score RR 1.41 (95 %CI 1.30–1.53) p < 0.001(Fig. 3). These relations were attenuated when adjusted for age.
Fig. 3

Adipokines and BMI z score. This figure shows that adiponectin has a negative correlation with BMI z score, p = 0.002, and leptin has a positive correlation with BMI z score, p < 0.001. Stars and dotted line males, black dots and full line females

There were no signs of an increase in BMI over the time period (Fig. 4).
Fig. 4

BMI z score changes by age and period. No significant changes are seen in BMI z score with increasing age nor over the time period 1998–2006. Stars and dotted line females, black dots and full line males

Influence of autoantibodies

In children negative for autoantibodies, there was a higher level of leptin in the univariate analysis, but it became insignificant in the adjusted model. Adiponectin showed no difference according to autoantibodies.


The fact that there is an increasing incidence of T1D has led us search for responsible associated factors. We found increased leptin levels and decreased adiponectin levels in children, which in parallel with the rise of T1D may reflect several factors associated with diabetes such as insulin sensitivity, immune activity, obesity, physical activity or diet. All factors are interrelated and part of an environmental change that is still ongoing.

Paediatric obesity has become more common in the USA and Western Europe [1519]. Recent data from the USA show a high prevalence of overweight in both sexes aged 9 through 19 years (33.2 %) and obesity (18.2 %) [20]. It is known that obesity increases the risk for type 2 diabetes, hypertension and hyperlipidaemia during childhood and adolescence, but its potential role in T1D development is not clarified. Our study, as in many others, has shown a significant relationship between an increasing BMI and an increase in leptin levels [2123] and a decrease in adiponectin levels [22, 24]. This could indicate that obesity may be the environmental factor to study further. However, in our cohort, no increase in BMI over time was identified and a Danish study did not find any increase in the prevalence of overweight and obesity in preschool children over a decade from 1992 to 2001 [25]. Still changes in adipokines were demonstrated indicating other factors may influence the regulation of these hormones than fat mass or reflect that BMI is not a precise measure of obesity.

Could adipokines reflect physical activity (PA)? PA is known to increase sympathetic nervous system activity, which might inhibit leptin secretion. If obese children are less physically active, there will be less negative feedback on leptin secretion leading to higher leptin levels. Alternatively, they might be less sensitive to leptin and therefore have upregulated levels. However, in healthy 5-year-old Pima Indian children, a positive correlation between PA and leptin has been reported, even after adjusting for per cent body fat [26]. A new study suggests that leptin represents an obesity-related signal from the periphery to the brain altering brain activity and locomotion leading to less physical activity in mice with high levels of leptin [27]. A Norwegian study showed that adiponectin levels increased significantly in obese adolescent after 3 months of high-intensity physical activity [28]. Most cross-sectional studies show a positive correlation between PA and plasma adiponectin levels, and moderate-to-high intensity exercise has the greatest impact on adiponectin levels in adults [29].

The association between adipokines and diet is conflicting. The intake of fibre, vegetables and food with a low glycaemic index has shown a positive association with adiponectin levels in men with type 2 diabetes [30] and an inverse association with leptin levels in a group of young Japanese women [31], others show no association [30]. A recent review found no association between intake of saturated fatty acids and leptin or adiponectin levels [32].

In an attempt to illuminate the influence of timing, the present study analysed diabetes duration as a categorical variable. This revealed lower levels of both leptin and adiponectin the first few days after onset in accordance with others [3335]. Some of it may have been influenced by dehydration, ketoacidosis and the catabolic state that they were in, as well as insulin treatment itself. Very few studies have recorded samples before insulin treatment [33, 35, 36]. But when comparing the adiponectin levels in our paediatric T1D subjects (the majority were tested within 1 month of diagnosis) with their healthy siblings, we did not find any difference, which is consistent with other studies [22, 34]. One study showed an increase after a month and normalisation after 4 months [34]. Others have found elevated adiponectin concentrations in children with T1D compared with a normal healthy control group [21, 24, 37, 38]. This may reflect differences in insulin levels or timing since elevated adiponectin levels have been found in those who had T1D longer than 0.5 year [21, 24] when beta cell function is deteriorating. The lower level of leptin can be explained by a lower insulin level in our patients with T1D or higher levels of soluble leptin receptors [39]. Adipokines and insulin are highly correlated through the adipoinsular axis. Insulin exerts a stimulatory effect on leptin secretion, and leptin downregulates insulin synthesis and secretion [8]. Others have found no significant correlation [36] or, like us, lower levels of leptin 4.4 years after onset [38].

Insulin also downregulates adiponectin gene in adipocytes, and therefore, insulin deficiency may promote secretion of adiponectin. In accordance with this, another study found that patients with low adiponectin levels 1 month after diagnosis (when beta cell function is present) had a higher probability of entering remission [40].

The decrease in adiponectin levels with age in both the healthy siblings and patients is in accordance with a decrease with pubertal development [41]. We confirm no gender difference in adiponectin levels [21, 22]. Contrary to some [34] but in accordance with previous studies in both adults and children [39], we found higher levels of leptin in females.

Our study has a number of advantages: first, the patients represented a random sample from a population-based registry of nearly 2,200 diabetic children diagnosed in Denmark from 1997 to 2005. They are representative of all newly diagnosed children in Denmark during that period, during which time an increase in incidence of 3.4 % is demonstrated [42].

Second, the siblings in the study carry an increased genetic risk of diabetes, compared to the background population, and may have experienced the same infections and lived under the same hygiene conditions as the probands, which implies that the differences found are likely to be underestimated.

Limitations of our study include the fact that the samples have been stored for up to ten years (1997–2005) at minus −80 °C degrees prior to analyses. There are currently no studies of the possible storage decay with repeated measures of adipokine levels in the same serum samples taken years apart. Very few studies actually report storage time or period of sampling. The increase by sampling date (or decrease by storage time) seems to be a steady change per year, and since our results are in both directions, decay seems to be a minor problem. Another limitation is the fact that adiponectin level was not divided in high molecular weight and low molecular weight adiponectin levels, which would have been more specific. Moreover, as previously mentioned, we do not have the BMI on all of the patients or the siblings and therefore have not been able to adjust for it in the whole study population. This should be taken into account in further studies.

We conclude that there is an increase in leptin levels over the years in both healthy and diabetic children and a decrease in adiponectin levels in patients leading to a more pro-inflammatory phenotype at onset. The changes in adipokines are not explained by increasing BMI, and therefore, other environmental factors affecting adipokines such as diet and physical activity should be investigated.


The DSBD Biobank is funded by Grants from the Danish Medical Research Council (271-07-0657) and the Danish Diabetes Association. This study was funded by the Aase and Ejnar Danielsens Foundation, Copenhagen Municipality and the Danish Diabetes Association.

Conflict of interest

Narges Safai, Stefanie Eising, David Michael Hougaard, Henrik Bindesbøl Mortensen, Kristin Skogstrand, Flemming Pociot, Jesper Johannesen and Jannet Svensson declare that they have no conflict of interest.

Human and animal rights

All procedures followed were in accordance with the ethical standards of the responsible committee on human experimentation (institutional and national) and with the Helsinki Declaration of 1975, as revised in 2008.

Informed consent

Informed consent was obtained from all patients before being included in the study.

Copyright information

© Springer-Verlag Italia 2014