Adiponectin and visfatin concentrations in children treated with valproic acid


Address correspondence to Markus Rauchenzauner, Department of Pediatrics IV, Division of Neuropediatrics, Medical University Innsbruck, Anichstrasse 35, 6020 Innsbruck, Austria. E-mail:


Chronic antiepileptic therapy with valproic acid (VPA) is associated with increased body weight and insulin resistance in adults and children. Attempts to determine the underlying pathophysiologic mechanisms have failed. Adipocytokines have recently been defined as a link between glucose and fat metabolism. We herein demonstrate that VPA-associated overweight is accompanied by lower adiponectin and higher leptin concentrations in children. The absence of any relationship with visfatin concentration does not suggest a role of this novel insulin-mimetic hormone in VPA-associated metabolic alterations. Therefore, adiponectin and leptin but not visfatin may be considered as potential regulators of glucose and fat metabolism during VPA-therapy.

Valproic acid (VPA) is a short-chain, branched fatty acid with a broad spectrum of antiepileptic and anticonvulsant activity (Davis et al., 1994). Despite being well-tolerated overall and having a low incidence of serious side effects, VPA can induce weight gain (Biton et al., 2003), Nonalcoholic fatty liver disease (NAFLD) (Luef et al., 2004) and insulin resistance (IR) (Pylvanen et al., 2006). However, attempts to determine the underlying pathophysiologic mechanisms of these metabolic disturbances have been unsuccessful.

Adipose tissue is generally considered as an endocrine organ which releases various factors into the circulation. The so-called adipocytokines (e.g., adiponectin, visfatin) have recently been defined as a link between glucose and fat metabolism (Tilg & Moschen, 2006). Adiponectin concentrations are negatively correlated with body-mass-index (BMI), IR and cardiac events (Chandran et al., 2003, Tilg & Moschen, 2006). In children, adiponectin concentrations appear to be an independent predictor for the metabolic syndrome (MS) and may play a role in the pathophysiology of the disorder (Korner et al., 2007). In adults, patients who gained weight during VPA treatment showed significantly lower adiponectin concentrations than patients who did not gain weight (Greco et al., 2005). In contrast, the novel adipocytokine visfatin was described as a protein, able to mimic insulin functions and to activate the intracellular signalling cascade for insulin (Fukuhara et al., 2005). The molecule is mainly expressed in visceral fat, liver, muscle and bone marrow. Recently, it was suggested that the elevation of plasma visfatin concentrations in obese children may be independently involved in the development of MS and precede abnormal glucose tolerance (Haider et al., 2006).

The aim of this study was to evaluate the influence of VPA treatment on parameters of glucose homeostasis, novel insulin resistance-related adipocytokines (adiponectin, visfatin), body composition, and serum lipids in children with epilepsy.

Materials and Methods


One hundred forty-two outpatient children >6 years with antiepileptic drug (AED) monotherapy (>6 months) were included. The first group includes 84 patients receiving VPA as their sole treatment. The second group (serving as controls) consists of 58 children matched for age and duration of therapy with AED monotherapy known not to influence body weight, glucose homeostasis and adipocytokines (lamotrigine (LTG) [n = 12], sulthiame (STM) [n = 12], levetiracetam (LEV) [n = 8] and oxcarbazepine (OXC) [n = 26]) (Biton, 2006). Epilepsy was defined according to the Guidelines of the International League Against Epilepsy (1989). All children had normal cerebral computer tomograms (CT) and/or magnetic resonance images (MRI). The main exclusion criteria were as follows: children known to suffer from any disease possibly affecting adipocytokines or lipid metabolism, children with muscular or bone diseases, with genetic syndromes, major congenital malformations and cancer or other neurological diseases. The study protocol was approved by the local ethics committee. Age- and sex-specific standard deviation scores (SDS) for height, weight and BMI were calculated according to German reference data (Kromeyer-Hauschild et al., 2001). The homeostasis model assessment index for IR (HOMA-IR) was calculated according to the following equation: insulin (μU/m) × (glucose (mmol/L)/22.5)). Children with a BMI above the 85th centile were considered overweight (Haider et al., 2006). Body-impedance analysis (BIA) was performed according to the standard tetrapolar procedure using a Human-Im BIA (DS Medigroup, Milan, Italy) as described previously (Pecoraro et al., 2003).

Chemical analysis

Serum leptin was measured using an enzyme-linked immunosorbent assay kit (ELISA) (R&D Systems, Wiesbaden, Germany). Adiponectin was determined with Human Adiponectin RIA Kit (Linco Research, MO, U.S.A.) and visfatin with a specific enzyme-linked immunosorbent assay (EIA) (Phoenix Peptides, Karlsruhe, Germany).


Since most variables were not normally distributed, nonparametric tests were used. Between groups analysis was performed using the Mann–Whitney U-test. The Wilcoxon signed ranks test was applied to determine significance of repeated measures. Spearman's σ was used to analyze associations between adipocytokines, body fat, insulin and HOMA-IR-index. A p-value of less than 0.05 was considered as statistically significant.


In the VPA-treated group, 37 children (44%) were considered overweight (BMI >85th centile), and in the control group only seven children (12%) exceeded the 85th centile.

VPA-treated children had significantly higher BMI SDS, percent body fat, HOMA-IR, insulin and leptin concentrations (p ≤ 0.039) as well as a tendency for higher triglyceride levels (p = 0.053) than children treated with LTG, STM, LEV or OXC (Table 1). In VPA-treated children, correlation analyses revealed significant negative correlations of adiponectin with BMI SDS (r =−0.529, p < 0.001), body fat (r =−0.377, p = 0.001), triglycerides (r =−284, p = 0.014) and leptin concentrations (r =−0.299, p = 0.010) as well as a significant positive correlation with HDL-cholesterol (r = 0.385, p = 0.005). Furthermore, visfatin concentrations showed a significant positive correlation with HDL-cholesterol (r = 0.374, p = 0.001) and a significant negative correlation with triglycerides (r =−0.250, p = 0.030) in children treated with VPA. In controls, no correlation between adiponectin or visfatin with markers of obesity was seen. Correlation analyses between any of the tested parameters with age, sex, duration of therapy or serum concentrations revealed no significant relationship in patients and controls. Additional analyses were performed based on the BMI of the children. Comparison of overweight VPA-treated children (BMI >85th centile, n = 37) with lean VPA-treated children (BMI <85th centile, n = 47) showed significantly higher BMI SDS (median [range] 1.61 [1.02–3.38] vs. −0.34 [−2.80 to 0.94]), percentage body fat (29.0 [15.8–48.1] vs. 17.1 [4.5–28.7]%, each p < 0.001), HOMA-IR (3.8 [0.5–9.4] vs. 2.3 [0.4–12.3], p = 0.031), LDL-cholesterol (100.0 [47.0–166.0] vs. 84.0 [26.0–137.0] mg/dl, p = 0.011) and leptin concentrations (37.7 [10.5–88.5] vs. 11.2 [1.3–84.1] ng/ml, p < 0.001) as well as lower HDL-cholesterol (57.0 [47.0–80.0] vs. 65.5 [34.0–113.0] mg/dl, p = 0.013) and adiponectin levels (20.10 [7.51–42.79] vs. 29.35 [13.15–62.43]μg/ml, p < 0.001, Fig. 1) in overweight children. Comparison of lean children (BMI <85th centile) undergoing VPA monotherapy versus controls revealed higher adiponectin (p = 0.041) and leptin concentrations (p < 0.001), higher percentage body fat (p = 0.001) and insulin levels (p = 0.046) as well as lower LDL-cholesterol (p = 0.009) in lean VPA-treated children (Table 2).

Table 1.  Clinical characteristics and laboratory findings
 VPA group (n = 84)Controls (n = 58)p value
  1. n, number of subjects.

  2. Values are median (range).

  3. SDS, standard deviation score; BMI, body mass index; HDL-C, high-density lipoprotein cholesterol; LDL-C, low-density lipoprotein cholesterol.

Age (years)12.8 (6.0–18.6)  12.3 (6.0–18.3)  0.993
Sex (M/F)36/4838/20 
BMI SDS 0.7 (−2.8 to 3.4) −0.4 (−3.1 to 2.8)<0.001
Body fat (%)23.0 (4.5–48.1)  11.2 (3.4–28.0)  <0.001
Glucose (mg/dl)85.0 (57.0–134.0)85.0 (59.0–107.0)0.958
Insulin (mg/dl)13.1 (2.1–48.2)  8.2 (1.3–28.2) 0.039
HOMA-IR2.7 (0.4–12.3) 1.9 (0.3–4.8)0.015
Total cholesterol (mg/dl)159.0 (100.0–223.0)160.0 (98.0–241.0) 0.149
HDL-C (mg/dl)59.0 (34.0–113.0)62.0 (29.0–92.0) 0.805
LDL-C (mg/dl)90.0 (26.0–166.0)93.0 (52.0–170.0)0.115
Triglycerides (mg/dl)96.5 (37.0–191.0)77.0 (24.0–270.0)0.053
Leptin (ng/ml)22.6 (1.3–88.5)  4.5 (0.7–32.5)  <0.001
Adiponectin (μg/ml)25.7 (7.5–62.4)  25.4 (8.2–58.8)  0.678
Visfatin (ng/ml)79.4 (35.7–650.9)82.3 (49.5–796.4)0.759
Figure 1.

Serum adiponectin concentrations in overweight vs. lean VPA-treated children. The box plots show median, interquartile values, and 10th and 90th percentiles.

Table 2.  Anthropometric data, adipocytokines and glucose homeostasis in lean patients
 Lean VPA patients (n = 47)Lean controls (n = 51)p value
  1. n, number of subjects.

  2. Values are median (range).

  3. SDS, standard deviation score; BMI, body mass index; HDL-C, high-density lipoprotein cholesterol; LDL-C, low-density lipoprotein cholesterol; VPA, valproic acid; HOMA-IR, homeostasis model assessment index for insulin resistance.

BMI SDS −0.4 (−2.9 to 1.7)−0.4 (−3.1 to 2.8)0.438
Body fat (%)17.1 (4.5–28.7)  10.9 (3.4–28.0)  0.001
Glucose (mg/dl)84.0 (57.0–134.0)85.0 (59.0–107.0)0.584
Insulin (mg/dl)10.6 (2.1–37.3)   8.4 (1.3–28.2)  0.046
HOMA-IR2.3 (0.4–12.3) 1.9 (0.3–6.7)  0.078
Total cholesterol (mg/dl)152.5 (100.0–220.0) 162.0 (98.0–241.0)  0.055
HDL-C (mg/dl)65.5 (34.0–113.0) 62.0 (29.0–92.0)  0.446
LDL-C (mg/dl)84.0 (26.0–137.0)96.0 (52.0–170.0)0.009
Triglycerides (mg/dl)90.5 (37.0–181.0)78.0 (24.0–270.0)0.325
Leptin (ng/ml)11.2 (1.3–84.1)  4.1 (0.7–32.5) <0.001
Adiponectin (μg/ml)29.4 (13.2–62.4) 26.4 (8.2–58.8)  0.041
Visfatin (ng/ml)87.3 (43.7–331.2)85.3 (49.5–796.4)0.881


The present findings reveal a distinct relationship between VPA-therapy, overweight and the adipocytokine axis. Furthermore, this is the first study showing significantly lower adiponectin and HDL-cholesterol concentrations as well as higher leptin, HOMA-IR, percentage body fat and LDL-cholesterol in overweight children undergoing VPA-therapy compared to lean VPA-treated children. In adults, only one study addressed this issue showing higher leptin and insulin levels as well as lower adiponectin concentrations in patients who gained weight during VPA-therapy (Greco et al., 2005).

In obese children, the prevalence of the MS is reported to be 30%, and a substantial amount of literature supports a central role of adipocytokines in the pathogenesis of the disorder (Stefan et al., 2002). Adipocytokines (e.g., adiponectin, leptin, visfatin) are adipocyte-derived hormones regulating metabolism and considered as key regulators of glucose and fat homeostasis (Tilg & Moschen, 2006). Adiponectin has been postulated to play an important role in the modulation of insulin-sensitivity and is significantly inversely associated with obesity (Chandran et al., 2003). In both adults and children, hypoadiponectinemia appears to be a strong predictor for MS (Korner et al., 2007). Interestingly, although directly correlated with body weight and highly associated with the amount of adipose tissue, leptin is also correlated with IR independently of fat mass, suggesting that hyperleptinemia acts as an independent component of the metabolic syndrome. VPA therapy is well known to be highly associated with weight gain, high serum insulin and leptin as well as low adiponectin concentrations (Biton et al., 2003; Luef et al., 2004; Pylvanen et al., 2006). However, the question of whether hyperleptinemia, hypoadiponectinemia and hyperinsulinemia are among the causes or consequences of VPA-associated obesity has not been addressed. From a metabolic perspective, the low adiponectin levels in overweight VPA-treated patients observed in our study are in accordance with prior reports showing lower concentrations in patients with obesity and type 2 diabetes, providing a biological link between overweight and overweight-related disorders (Chandran et al., 2003). Hypoadiponectinemia in association with hyperleptinemia, high LDL-cholesterol levels and high body fat reflect a greater cardiovascular burden of these children later in life.

Subsequently, glucose homeostasis and fat metabolism in lean VPA-treated children was investigated. Interestingly, lean VPA-treated children revealed higher adiponectin levels, higher percentage body fat, leptin and insulin concentrations as well as lower LDL-cholesterol than did lean controls. There are three conceivable mechanisms for higher adiponectin concentrations despite similar body weight in lean VPA-treated children: First, higher adiponectin concentrations may simply reflect, to some extent, a counter regulatory mechanism to upregulation of insulin resistance-inducing factors (e.g., leptin, body fat). Second, since adiponectin interacts with specific receptors (adipoR1 and adipoR2), direct modulation or feedback mechanisms influencing the adiponectin receptor system may also be involved in adiponectin mRNA expression. Third, an increasing body of literature has demonstrated the importance of fat distribution and especially the contribution of visceral fat to the development of MS and cardiovascular disease (Fujioka et al., 1987; Yamashita et al., 1996). The negative correlation of adiponectin levels and visceral fat is considered stronger than between adiponectin and subcutaneous fat (Matsuzawa, 2006). As a consequence, hypoadiponectinemia induced by visceral fat accumulation particularly reflects a strong risk factor for the development of obesity associated disorders (e.g., diabetes mellitus, hypertension and cardiac events). Therefore, a favorable fat distribution (besides absolute fat mass) might additionally contribute to higher adiponectin concentrations in lean VPA-treated children.

A growing body of evidence indicates a role of visfatin in glucose homeostasis independently of other parameters (Haider et al., 2006). The plasma levels of visfatin correlate significantly with the amount of visceral fat and blood glucose levels and are increased in diabetes mellitus and the polycystic ovary syndrome, both negatively associated with insulin sensitivity (Chen et al., 2007). In the present study, serum visfatin concentrations did not differ between the two groups as well as between overweight and lean VPA-treated subjects. This seems reasonable, as there was no evidence for high visceral fat accumulation and also glucose levels did not differ between groups.

The main limitation of this study is the lack of a medication-free control group. This has been overcome by using an age-matched disease control group receiving AED monotherapy not modulating body-weight and fat metabolism. Another possible confounding factor may be the unequal sex distribution between groups. However, sex-specific analyses (data not shown) revealed similar results.

In summary, VPA-associated overweight is accompanied by lower adiponectin and higher leptin concentrations in children. The absence of any relationship with visfatin concentrations does not suggest a role of this novel insulin-mimetic hormone in VPA-associated alterations of glucose metabolism. Therefore, adiponectin and leptin but not visfatin may be considered as potential regulators of glucose and fat metabolism during VPA-therapy.


We confirm that we have read the Journal's position on issues involved in ethical publication and affirm that this report is consistent with these guidelines.

Conflicts of interest statement: The authors report no conflicts of interest.