Study identification and selection
We identified in an initial search 12363 articles. After eliminating 4333 duplicate studies, 8030 were screened by title and abstract, 7232 were eliminated because not meet eligibility criteria. A total of 798 articles were reviewed in full text. Of those, 21 articles met the inclusion criteria and were included in the review and meta-analysis [22–42]. (Figure 1).
Characteristic of selected studies
A total of 5516 girls and 5629 boys were included in the 21 studies selected (Table 1). The articles were published in English language between 1999 and 2021, from Europe (n = 11), America (n = 4), Oceania (n = 3), Asia (n = 3). Sample sizes ranged from 26 to 4633 subjects; ten studies had a sample size greater than 200 subjects. Study designs were as follows: cross-sectional (n = 12), and cohort (n = 9). The blood leptin was assessed in plasma (n = 8) or serum (n = 12), one study did not report the type of blood sample used.[29] The measurement methods of leptin were: ELISA (n = 5), RIA (n = 15) and Milliplex MAP Multiplex assay (n = 1). BFP was measured by plethysmography (n= 3), bioelectrical impedance (n= 5), dual-energy X-ray absorptiometry (n= 5), skinfolds (n= 7) and TOBEC (n= 1). BMI was reported as kg/m2 (n= 10), standardized BMI (n= 1), BMI Z-score (n= 4), and in six studies with newborns and children in the first year of life, weight and height were reported.
Table 1. Characteristic of studies selected.
Studies by age groups
|
Study design
|
Country
|
Sample Size by Sex
|
Age (years)
|
BMI/BMI z,score/BMI sds/ Weight
|
Body fat percentage (BFP)
|
Circulating leptin
|
Girls
n (%)
|
Boys
n (%)
|
Girls
mean(±SE)
|
Boys mean(±SE)
|
Girls
|
Boys
|
Girls
means (±SE)
|
Boys
means (±SE)
|
Method
|
Girls
(ng/mL)
|
Boys
(ng/mL)
|
Blood Sample
|
Method
|
Newborns
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
Okereke (2002)[22]
|
Cross sectional
|
EEUU
|
32 (41)
|
46(59)
|
-
|
-
|
3.2 (0. 41)a
|
3.5 (0.5)a
|
11.25 (3.98)
|
11.9(4.4)
|
TOBEC
|
16 (13.8)
|
12.7 (12.8)
|
Serum
|
RIA
|
Javaid (2005)[23]
|
Cohort
|
UK
|
50 (43)
|
67(57)
|
-
|
-
|
3.2 (0.5)a
|
3.51 (0.5)a
|
15.9 (13.6;19.4) ¶
|
14.5 (12;16) ¶
|
DXA
|
9.1(5.9;13.9) ¶
|
7.4 (4.2;11.4)¶
|
Serum
|
RIA
|
Euclydes (2018)[24]
|
Cohort
|
Brazil
|
62 (59.6)
|
42(40.4)
|
-
|
-
|
3.4 (0.4)a
|
3.34 (0.4)a
|
9.8 (4.1)
|
8.2 (3.6)
|
PG
|
29 (26.9)
|
22.8 (25.1)
|
Plasma
|
ELISA
|
Benhalima(2021)[25]
|
Cohort
|
Belgium
|
233 (48)
|
251 (52)
|
-
|
-
|
-0.1 ± 0.9†
|
-0.2 ± 0.9†
|
39 (0.0)
|
39(0.0)
|
Skinfolds
|
13.9(9.4;20) ¶
|
7.6 (4.4;14.2)¶
|
Plasma
|
RIA
|
0.25 to 0.5 years
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
Estampador (2014)[26]
|
Cohort
|
Sweden
|
15 (48)
|
16 (52)
|
0.32(0.03)
|
0.32(0.03)
|
6.2 (1.07) a
|
6.9 (0.69) a
|
25.6 (0.79)
|
27.1(0.25)
|
PG
|
4.9 (2.9)
|
4 (1.6)
|
Plasma
|
RIA
|
de Fluiter (2021)[27]
|
Cohort
|
Netherland
|
138 (46)
|
159 (54)
|
0.25
|
0.25
|
5.7 (5.1;6.2) a
|
6.2 (5.7;6.6) a
|
23.3 (19.6;26.5)¶
|
22.1(19.0;24.6)¶
|
PG
|
1.7(1.5;1.9) ¶
|
1.3(1.1;1.5) ¶
|
Plasma
|
MAP
|
138 (46)
|
159 (54)
|
0.5
|
0.5
|
7.3 (6.8–7.8)
|
7.9 (7.3;8.4)
|
25.1 (21.6;28.9)¶
|
23.5 (20.0;26.7)¶
|
PG
|
0.9(0.6;1.1) ¶
|
0.8(0.7;1.0) ¶
|
Plasma
|
MAP
|
3 to 5.9 years
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
Erhardt (2014)[28]
|
Cohort
|
Europe ( eight countries))
|
47 (53)
|
42 (47)
|
3 to 3.9❢
|
3;3.9❢
|
16.2 (1)
|
16.3 (0.9)
|
16.8 (2.9)
|
15.6 (2.5)
|
Skinfolds
|
2.0(1.5;3.8) ¶
|
1.5(1.1;2.5) ¶
|
Serum
|
ELISA
|
33 (50)
|
33 (50)
|
4 to 4.9❢
|
4;4.9❢
|
15.8 (0.8)
|
16 (0.8)
|
16.6 (2.5)
|
15.1 (2.3)
|
Skinfolds
|
1.9(1.5;2.7) ¶
|
1.3(1.0;1.7) ¶
|
Serum
|
ELISA
|
32(53)
|
28(47)
|
5 to 5.9❢
|
5;5.9❢
|
15.6 (0.9)
|
15.7 (0.9)
|
16.5 (3.2)
|
14.5 (3.2)
|
Skinfolds
|
1.9(1.5;2.7)¶
|
1.2(0.9;2.4) ¶
|
Serum
|
ELISA
|
Jáuregui (2020)[29]
|
Cross sectional
|
Mexico
|
197(49)
|
203(51)
|
4.7(0.5)
|
4.8(0.6)
|
0.2(1) †
|
0.2 (1.1) †
|
25 (5.7)
|
22.8 (5.8)
|
BIA
|
3.5(3.1)
|
2.6(2.7)
|
-
|
ELISA
|
6 to 7.9 years
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
Garnett (2004)[30]
|
Cohort
|
Australia
|
137 (54)
|
118 (46)
|
7.8(0.6)
|
7.9(0.6)
|
16.9(2.3)
|
16.7 (2.5)
|
23.6 (8.4)
|
17.4 (8.6)
|
DXA
|
3.4(3.2;3.6)❡
|
2.2(2.0;2.4)❡
|
Serum
|
RIA
|
Kim (2011)[32]
|
Cross sectional
|
Korea
|
229 (50)
|
231 (50)
|
7.8(0.5)
|
7.9(0.5)
|
15.5(1.4)
|
16 (1.2)
|
16.7 (4.1)
|
13.8 (4.1)
|
BIA
|
3.4 (2.1)
|
2.5(1.8)
|
Serum
|
RIA
|
Metcalf (2011)[31]
|
Cohort
|
UK
|
158(56)
|
122(44)
|
6.9(0.3)
|
6.9(0.3)
|
16.4(2.8)
|
15.7 (2.1)
|
21.9 (12.8) ℑ
|
14.4 (8.0) ℑ
|
DXA
|
4 (5.2) ℑ
|
2.6 (2.7) ℑ
|
Serum
|
RIA
|
148(55)
|
121(45)
|
7.8(0.3)
|
7.9(0.3)
|
16.9(3.1)
|
15.9 (2.3)
|
23.3 (14.5) ℑ
|
14.8 (9.8) ℑ
|
DXA
|
4.4 (6.6) ℑ
|
2.6 (3.1) ℑ
|
Serum
|
RIA
|
Jeffery (2012)[33]
|
Cohort
|
UK
|
101
|
134
|
7.0
|
7.0
|
0.54(0.22) ‡
|
0.21 (0.18) ‡
|
30 (1.72) §
|
26 (1.2) §
|
Skinfolds
|
4.6 (1.3) §
|
2.7 (0.6) §
|
Serum
|
RIA
|
Erhardt (2014)[28]
|
Cohort
|
Europe ( eight countries)
|
41(48)
|
45(52)
|
6 to 6.9❢
|
6; 6.9❢
|
15.8(1.1)
|
15.8 (1)
|
16.9 (2.7)
|
13.6 (2.9)
|
Skinfolds
|
2.7 (1.5;3.7) ¶
|
1.3 (0.8;2.5)¶
|
Serum
|
ELISA
|
72 (53)
|
65 (47)
|
7 to 7.9❢
|
7; 7.9❢
|
15.9(1)
|
15.9 (1.2)
|
16.5 (3.1)
|
14.5 (3.5)
|
Skinfolds
|
2.0(1.4;3.8)¶
|
1.7 (1.0;3.3)¶
|
Serum
|
ELISA
|
Vitery (2020)[34]
|
Cross sectional
|
Colombia
|
56 (51)
|
54 (49)
|
7.9 (1.2)
|
7.8 (1.3)
|
16.4(1.8)
|
16.1 (1.2)
|
22.4 (4.1)
|
18.4 (3.5)
|
Skinfolds
|
6.9 (5.0)
|
3.3 (3.7)
|
Plasma
|
ELISA
|
Haapala (2021)[35]
|
Cohort
|
Finland
|
192 (49)
|
198 (51)
|
7.6 (0.4)
|
7.7 (0.4)
|
-0.2(1.1) †
|
-0.2 (1.1) †
|
22.2 (7.4)
|
17.2 (7.7)
|
DXA
|
5.9 (4.2)
|
4.2 (3)
|
Plasma
|
RIA
|
8 to 10 years
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
Byrnes (1999)[36]
|
Cross sectional
|
Australia
|
30(51)
|
29(49)
|
8.6 (0.2)
|
8. 5(0.3)
|
0.5†
|
0.3†
|
25.9 (1.1)
|
18.4 (1.2) §
|
BIA
|
11.5 (2.2) §
|
6.5 (1) §
|
Plasma
|
RIA
|
Arrowsmith (2002)[37]
|
Cohort
|
Australia
|
12(46)
|
14(54)
|
7.9 (0.8)
|
8.3 (0.8)
|
16.6(2.2)
|
18.3 (3.7)
|
19.9 (4.5)
|
19.2 (5.8)
|
Skinfolds
|
6.2 (3.7)
|
8.3 (5.6)
|
Plasma
|
RIA
|
Celi(2005)[38]
|
Cohort
|
Italy
|
395
|
447
|
8.5(7.2;9.6)❡
|
9.4(8.3;10.6)❡
|
20.6(16.1;24.1)
|
21.5(16.1;24.8)
|
35.2 (11.1)
|
31.7 (11.7)
|
Skinfolds
|
11.9(5.2;20.6)❡
|
9.9(3.8;18.4)❡
|
Serum
|
RIA
|
Dencker (2006)[39]
|
Cross sectional
|
Sweden
|
79(46)
|
91(54)
|
9.8 (0.6)
|
10 (0.6)
|
17.4(2.7)
|
17.6 (2.6)
|
21.9 (8.4)
|
16.1 (8.3)
|
DXA
|
5.3 (4.8)
|
3.2 (4.2)
|
Serum
|
RIA
|
Yamborisut (2009)[40]
|
Cross sectional
|
Thailand
|
28 (55)
|
23 (45)
|
8.1 (0.9)
|
8.3 (0.9)
|
-0.11†
|
-0.35†
|
16.9 (5.6)
|
12.6 (4.7)
|
BIA
|
6 (2.6)
|
3.7 (2.4)
|
Serum
|
RIA
|
Metcalf (2011)[31]
|
Cohort
|
UK
|
148 (56)
|
114 (44)
|
8.9 (0.3)
|
8.9(0.3)
|
17.9(3.7)
|
16.4 (2.8)
|
26.4 (13.4) ℑ
|
16.6 (12.7) ℑ
|
DXA
|
6.7 (7.1) ℑ
|
3.8 (3.8) ℑ
|
Serum
|
RIA
|
147 (56)
|
115 (44)
|
9.9 (0.3)
|
9.9 (0.3)
|
18.4 (5)
|
17 (3.1)
|
26.8 (14)ℑ
|
18.9 (15.3) ℑ
|
DXA
|
8.2 (12.7) ℑ
|
4.2 (6.1) ℑ
|
Serum
|
RIA
|
Jeffery (2012)[33]
|
Cohort
|
UK
|
101 (43)
|
134 (57)
|
8.0
|
8.0
|
0.53‡
|
0.28‡
|
30.1 (2.1) §
|
26.6 (1.7) §
|
Skinfolds
|
5.3 (1.7)
|
3.2 (1.2)
|
Serum
|
RIA
|
101 (49)
|
134 (51)
|
9.0
|
9.0
|
0.61‡
|
0.38‡
|
32.3 (1.8) §
|
27.8 (1.5) §
|
Skinfolds
|
7.2 (1.9)
|
4.5 (1.1)
|
Serum
|
RIA
|
101 (58)
|
134(42)
|
10.0
|
10.0
|
0.62‡
|
0.43‡
|
33 (1.8) §
|
2.9 (1.6) §
|
Skinfolds
|
8..8 (2.2)
|
5.3 (1.4)
|
Serum
|
RIA
|
Nightingale (2013)[41]
|
Cross sectional
|
UK
|
2237(48)
|
2396(52)
|
9 to 10❢
|
9;10❢
|
18.6(18.4;18.7)❡
|
18.3(18.1;18.4)❡
|
29.9(29.4;30.4)❡
|
27.2(26.7;27.7)❡
|
Skinfolds
|
11.5 (11;1)❡
|
7.2 (6.9;7.5)❡
|
Serum
|
RIA
|
Thillan (2021)[42]
|
Cross sectional
|
Sri Lanka
|
84 (51)
|
80 (49)
|
9.1 (0.3)
|
9.2 (0.3)
|
14.9(13.7;16.3) ¶
|
14.6 (13.8;17) ¶
|
20.7(15.5;27.1)¶
|
16.2(12.6;24.1)¶
|
BIA
|
3.7(2.1; 6.6)¶
|
2.2 (0.9;5.3)¶
|
Plasma
|
ELISA
|
BMI: body mass index (kg/m2); DXA: dual-energy X-ray absorptiometry scan; TOBEC: total body electrical conductivity; BIA: bioelectrical impedance analysis, PG:Plethysmography. SD: standard deviation; Weight in Kg and SDa , BMI z-score †; BMI sds ‡; standard error §; Median (interquartile range) ¶; 95% confidence interval ❡; Age Range ❢; Median (range) ℑ
Association of sex with leptin levels
The overall effect of sex in the children was associated with higher levels of blood leptin in girls (MD= 1.69 ng/mL, 95%CI: 1.21 to 2.17, I2 = 95.3%). Newborns girls had 3.6 ng/mL (95%-CI: 1.79 to 5.4, I2 = 17%, P = 0.31,) higher serum leptin levels than boys. Between 0.25 to 0.5 yrs, boys had 0.24ng/mL lower leptin but no was significant (95%-CI: -0.01 to 0.49 ng/mL, I2 = 97%, P <0.01). The group 3-5.9 yrs, the leptin was 0.8 ng/mL (95%-CI: 0.52 to 1.07 ng/mL , I2 = 0%, P = 0.74 ) higher in girls, while that the ages of 6-7.9 yrs and 8-10 yrs the blood leptin was respectively of 1.71ng/mL (95%-CI: 0.94 to 2.49 ng/mL, I2 = 82.3%, P<0.01,) and 2.53 ng/mL (95%-CI: 1.80 to 3.26 ng/mL, I2 = 72.3%, P<0.01) higher in girls ( Figure 2).
Association of BFP with sex
BFP showed to be higher in girls in the overall effect (MD= 3.44%, 95%CI: 2.55 to 4.33, I2 = 87.3%) Meta-analysis by age showed a BFP lower in boys except in the newborn group where the association was not significant (95%-CI: -0.26 to 3.15, I2 = 80%, P = <0.01). After contacting the authors, we omitted the study of Behalima et al. [25] from this group, because the reports of the mean and SD were considered biologically not plausible (BFP for boys: 39% ± 0 and girls: 39% ± 0). Boys in the group 0.25-0.5 yrs had 1.56 % (95%-CI: 0.79 to 2.33 %, I2 = 50%, P =0.14) lower BFP. For the groups of 3-5.9 yrs, was 1.69% (95%-CI: 1.08 to 2.29 %, I2 = 0%, P =0.6); to 6-7.9 yrs was 4.71% (95%-CI: 3.38 to 6.05 % , I2 = 89.9%, P <0.01) and to the group 8-10 yrs 4.60% (95%-CI: 3.08 to 6.13 % , I2 = 82.4%, P <0.01) higher in girls in all cases (Figure 3).
Sensitivity Analysis
Although there was high heterogeneity in the levels of leptin in the groups of 0.25 to 0.5 yrs and 6 to 7.9 yrs (I 2 >75%), most studies showed blood leptin and BFP higher in girls than in boys across all age groups. In the group of newborns the omitting of the study of Benhalima (2021)[25], showed no significant difference on leptin blood cord (MD= 2.09, 95% CI:-0.40 to 4.58, p=0.100, I 2 =0.0%).In the group of 8-10 yrs, were identified as outliers the studies of Arrowsmith (2002)[37] and Nightingale(2013)[41], the removal of these studies resulted in an increase of the effect (MD= 2.59 ng/mL, 95% CI: 1.62 to 3.32 ng/mL, p < 0.0001, I2=71.6%). The omitting only of the study of Nightingale (2013) [41] resulted in a considerable reduction of heterogeneity (MD= 2.19 ng/mL, 95% CI: 1.72 to 2.66 ng/mL, p < 0.0001, I2=15.8%).
For BFP we found high heterogeneity in the age groups of newborns, 6-7.9 yrs, and 8-10 yrs (I 2 >75%). In the newborn group the omitting the study of Javaid (2005) [23] resulted in a non-significant effect of sex on fat (MD = 0.6, 95% CI: –0.97 to 2.17%, p= 0.452, I2= 71.6%). A similar result was found when was remove the study of Euclydes (2018) [24] (MD= 1.3195% CI: -1.32 to 3.93%, p=0.330, I2=90%). In the group 6-7.9 yrs old, we identified as outliers the studies of Metcalf (2011) [31] and Erhardt (2014) [28], the remotion of these two studies increased the effect of sex (MD= 5.20, 95% CI: 3.58 to 6.83, p < 0.0001, I2= 92.0% ). In the group of 8-10 yrs old, the omitting of one of the groups of Metcalf (2011) [31] resulted in a reduction of heterogeneity (MD= 3.84, 95% CI: 2.81 to 4.87%, p < 0.0001, I2= 53.6%). (Tables S1 and S2)
Sensibility analysis indicated that the overall statistical significance did not change when any single study or several outliers were omitted, except in the group of newborns. Therefore, the results of this meta-analysis were deemed to be relatively reliable and credible.
Subgroup Analyses of Plasma Leptin Levels and BFP
To examine the influence of certain characteristics of the studies selected on blood leptin and BFP we carry out a subgroup analysis on the overall effect. Considering the study design we found that the cross-sectional studies (MD= 1.94, 95%CI=1.05 to 2.82, I2 = 90.8%, p< 0.01) shown a higher difference by sex that cohort studies (MD= 1.55, 95%CI= 1.01 to 1.74 ng/mL, I2 = 95.5%, p< 0.01).The results indicated both for the measurements of leptin by ELISA (MD=0.78 ng/mL, 95%CI= 0.39 to 1.16 ng/mL, I2 = 89.7%, p< 0.01) and RIA (MD=2.34 ng/mL, 95%CI=1.74 to 2.94 ng/mL, I2 = 89.1%, p< 0.01) the girls had higher leptin. Lower ELISA values can be explained by the fact that most of the studies that used this technique were in children under 6 yrs of age. The subgroup analysis by each region showed also higher leptin blood in girls in each group. (Table S3). Studies with high quality (n=17) shown that girls had higher leptin (MD=1.90 ng/mL, 95%CI=1.38 to 2.41 ng/mL, I2 = 86.7%, p< 0.01) while the studies with low quality (n=4) didn't show this effect (MD=0.78 ng/mL, 95%CI=-0.21 to 1.77ng/mL, I2 = 94.7%, p< 0.01).
According to the design study, BFP was higher in girls, with a larger effect in studies type cohort (MD= 3.89%, 95%IC= 2.68 to 5.09%, p< 0.0001, I2 = 90.5%, p< 0.01) that cross-sectional (MD=2.79%, 95%IC=1.85 to 3.72, p< 0.0002, I2 = 68.1%, p< 0.01). Independent of the method of measure and geographic region, boys had lower BPF and while the quality assessments indicated for BFP similar effect in the studies with low and high quality. (Table S4).
Publication bias
The funnel plot for blood leptin and BFP by age group were symmetrical with most of the studies located on the top left of the diagram (Figures S1, S2).