Impact of low-carbohydrate diet on serum levels of leptin and adiponectin levels: a systematic review and meta-analysis in adult

Abstract

Background

Various studies have evaluated the effects of low-carbohydrate diet (LCD) on serum concentrations of adipokines. Although the association between LCD and serum levels of leptin and adiponectin has been studied extensively, the results were not consistent.

Objective

The purpose of this study was to systematically evaluate the effect of LCD on serum levels of leptin and adiponectin.

Design

Electronic databases were retrieved in PubMed, Embase, Scopus and Web of Science to search relevant publications. Pooled standard mean difference (SMD) with 95% confidence interval (CI) was calculated by the random-effect model. Cochrane Q test and I² statistic were used to test heterogeneity. Subgroup analysis and meta-regression were applied to assess possible sources of heterogeneity.

Results

A total of thirty-five articles were included in final analysis. Meta-analysis results revealed no statistical association between LCD and adiponectin concentration (WMD: 0.32 ng/ml, 95% CI: - 0.02, 0.66, p=0.062). Subgroup analysis showed that LCD increased adiponectin concentration in subjects under 45 years old and in studies with long term duration intervention. Also, did not observe a significant effect from the LCD on serum concentration of leptin (WMD: - 0.77 ng/ml, 95% CI: -3.15, 1.61, P=0.409). Subgroup analysis did not show any new information. The results of this study did not support the evidence for the positive effects of LCD on serum leptin and adiponectin levels.

Introduction

Obesity is one of the major health problems in many countries and it is estimated that about one-third of people worldwide are obese and overweight[1–3]. As well, obesity-related causes are responsible for more than 2.8 million death each year [2]. Obesity is important risk factor for progression of several chronic disease including type 2 diabetes, cardiovascular disease, and hypertension [2, 4]. Adipocyte-derived hormones, known as adipokines, are important determinants of insulin resistance and involved in various metabolic processes[5]. Insulin-sensitizing hormone, adiponectin, proinflammatory hormones, tumor necrosis factor (TNF)-α, and leptin are main adipokines. Leptin and adiponectin are consider as a main adipokines which have associated with adiposity [6, 7]. Individuals with obesity exhibit a dysregulation of adipokines. Unlike leptin, an increase in visceral adipose tissue reduces the serum concentration of adiponectin. Also, a decrease in adiponectin level is associated with the etiology of insulin resistance, diabetes mellitus, hypertension, ischemic heart disease and atherosclerotic disease [6, 8–11]. On the other hand, leptin exert pleiotropic effects on different tissues and involved in different metabolic functions such as appetite, energy expenditure, insulin sensitivity, fat distribution, and lipid and glucose metabolism [12–14]. Prior studies were showed the favorable effects of weight loss on adipocytokines changes. Among the different dietary interventions for weight loss, carbohydrate restriction is one of the most common strategy [4]. A low-carbohydrate diet is the particular interest to subjects with overweight or obesity and patients with diabetes, dyslipidemia and some other chronic disorders [5, 6, 15–17]. Also, some previous studies have reported that a low carbohydrate diet has significant effects on adipokine concentration [18, 19]. However, the results of existing studies are inconsistent. So, this study aimed to evaluate the effects of low-carbohydrate diet on serum levels of leptin and adiponectin.

Methods

The present meta-analysis was conducted based on guidance provided by the Cochrane Handbook for Systematic Reviews of Interventions and the Preferred Reporting Items for Systematic reviews and Meta-Analyses (PRISMA) guidelines [20].

Literature search

Systematically search was performed by two independent researchers (MR and GKH) in PubMed, Embase, Scopus and Web of Science up to the July 2019. The search items included: [Key terms for low-carbohydrate diet] AND [Key terms for outcomes] AND [Key terms for study design]. Also, hand-scanning was performed in Google Scholar, Cochrane Library and reference lists of related papers to find additional references. The search process was systematically performed without any language, age, sex or date of publication restrictions. Also, we used the “*” term to increase the search sensitivity. We used from the EndNote V9 software for manuscript inclusion and duplicate removing.

Inclusion and exclusion criteria

We considered studies to be eligible if they met the following criteria: (1) controlled trial in subjects treated with low carbohydrate diet or control diet and (2) evaluated adiponectin and leptin as an outcome. Studies were excluded if participants were children (<18 years) or if they had a concomitant intervention for which effects could not be separated; were of less than 2 weeks’ duration; did not report the baseline and final values of outcome variables (or changes) in both intervention or control group; other type of article such as letter observational or review study and studies with non-English writing. Moreover, we excluded unpublished documents and grey literature, such as conference papers, dissertations, and patents.

Data extraction

The titles and abstracts and quality of the included studies were evaluated by two independent investigators and unrelated articles were deleted and then the full-text versions of eligible studies were retrieved for further evaluation. In the event of any dispute between the two reviewers, the matter was resolved through consultation with a third party. The main outcome sought in the studies was the mean change between the baseline and final adiponecin and leptin concentrations. After selection of the eligible studies, following data were extracted: first author name, publication year, sample size, sex distribution, country, study design, mean age and weight of participants, intervention duration, percent of carbohydrate in diet, type of control diet, net changes ([mean post-treatment] – [mean pre-treatment]) and standard deviation (SD) for changes (√ ([(SD pre)2 + (SD post)2] – [2r × SD pre × SD post]), with assuming coefficient correlation (r) = 0.5). Moreover, SD was obtained from standard error (SE) using following formula: (SD = SE × √n). In case of any unclear data about outcomes, an E-mail was sent to the corresponding author.

Risk of bias assessment

We used from the Cochrane Handbook recommendations for evaluation risk of bias [21]. Two investigators were evaluated the quality of trials independently in five categories: adequate sequence generation; allocation concealment; blinding of the outcome assessors; handling of missing data (intention-to-treat or per protocol analysis); selective outcome reporting. The nature of the included studies were open intervention and blinding was not applicable. It has been used from the “Low”, “High” and “Unclear” terms to reveal the status of each domain. RevMan V5.3 software (Cochrane Collaboration, Oxford, UK) was used to draw any relevant figures.

Statistical analysis

We used from the random-effects model to calculate the pool standardized mean difference (SMD) and 95% confidence intervals (CI). Also, for evaluation of statistical heterogeneity among the trials, Cochran Q test was used, and inconsistency was examined by the I² test. Moreover, I² statistic - with values of 25%, 50% and 75% were regarded as low, moderate and high, respectively - and P were considered to detect potential heterogeneity among studies [22]. To identify the potential sources of heterogeneity, stratified analyses were performed according to the following indicators: BMI, age, carbohydrate percent, intervention duration and study quality (low or high). Funnel plots were used to visually inspect for the presence of publication bias. In addition, for further investigation of publication bias, Begg’s rank correlation and Egger’s linear regression tests were used. All analyses were carried out using Stata, version 12 SE (Stata Crop, College Station, TX, USA). P-values <0.05 were considered statistically significant.

Results

Study selection

We found a total of 321 records from our initial search in databases. 109 duplicates were removed. Then, after screening based on title and abstract, 125 articles were excluded because they were not match with our criteria (unrelated title and abstract [n=89], animal studies [n=23], and review studies [n=13]). Next, 87 suitable articles were candidate for full-text assessment. Among these, 52 articles were excluded due to not reported required information. Finally, 35 eligible studies were included to our systematic review and meta-analysis. Figure 1, shows a summary of the study selection.

Fig. 1.

Flow diagram illustrating the study selection process for meta-analysis and systematic review

Characteristics of the studies

Characteristics of the 35 studies included in the current meta-analysis are shown in Table 1. These studies included a total of 2327 participants including 1240 participants in the intervention group and 1087 in the control group. All the included studies were randomized, controlled trials with double-blind design. Of the 35 studies, 30 were parallel and the other 5 were cross-over. Of the 35 studies, 27 studies conducted on obese subjects and 8 studies used from participants with chronic disorders (patients with T2DM, osteoarthritis, breast cancer, metabolic syndrome and NAFLD). The duration and carbohydrate percent varied between trials. The total daily carbohydrate percent changed from 10 to 45% between studies. The duration changed from 4 to 52 weeks.

Table 1.

Characteristic of included studies in meta-analysis

Author	Publication year	Country	Study Design	Participant	Sample size and Sex	Trial Duration (Week)	Means Age Mean ± SD		Means BMI Mean ± SD		Intervention	Sample Size
							LCD	CG	LCD	CG	CHO percent	LCD	CG
Brehm. BJ	2003	USA	Parallel	Obese subjects	53: 53 F	24	44.22±6.83	43.10±8.56	33.73±1.83	34.04±1.83	30	26	27
Arvidsson. E	2004	Sweden	Parallel	Obese subjects	40: 40 F	10	35.3±2.0	35.1+-1.6	37.6±1.1	36.6+-0.9	38	20	20
Seshadri, P	2005	USA	parallel	T2DM	75: 14 M, 61 F	24	54±9	54±10	43.73±6.03	43.06±6.44	31	40	35
Cardillo. S	2006	USA	parallel	Obese subjects	53: 44 M, 9 F	144	54.34±10.23	55.17±10.14	43.2+-7.2	43.4+-6.7	39	26	27
de Luis. DA	2007	Spain	Parallel	Obese subjects	90: 27 M, 63 F	12	43±14.3	42.1±16	35.8±5.5	35.6±5.7	29	47	43
Summer SS	2007	USA	Parallel	Obese subjects	42: 42 F	24	44.2±6.84	43.1±8.56	33.12±1.78	34.04±1.84	22	20	30
De Luis. DA	2008	Spain	Parallel	Obese subjects	26: 8 M, 18 F	12	46.7 ± 13.6	46.7 ± 13.6	36.1±4.7	36.1±5.8	38	12	14
Keogh JB	2008	Australia	Parallel	Obese subjects	70: 30 M, 40 F	8	50.5±8.1	49.4±8.2	33.5±4.1	33.9±4.1	20	37	33
De Luis. DA	2009	Spain	Parallel	Obese subjects	21: 4 M, 17 F	12	45.8 ± 14.2	46.3 ± 9.7	36.7± 4.7	36.8±6.4	38	12	9
Ratliff J	2009	USA	Parallel	obesity	28: 28 M	12	40-70	40-70	25-37	25-37	20	13	15
Al-Sarraj T	2009	UAE	Parallel	Metabolic Syndrome	39: 14 M, 25 F	12	18–50	18–50	38.7 ± 7.6	33.56 ±6.0	20	20	19
Luis DA	2009	Spain	Parallel	overweight and obese	94: 22 M, 72 F	8	43.5 ± 11.6	44.7 ± 16.6	35.9+6.0	35.4+5.9	38	54	40
Bradley. U	2010	USA	parallel	Overweight subjects	27: 9 M, 15 F	8	37.1 ± 8.9	40.5 ± 10.4	34.5 ± 4.2	32.8 ± 3.0	20	12	12
Vetter ML	2010	USA	parallel	T2DM	79: 70 M, 9 F	24	60.8±10.3	58.6±9.2	38.2±6.0	36.1±4.6	27	37	42
Belobrajdic DP	2010	Australia	parallel	Overweight subjects	76: 76 M	12	51±1.0	51.4+1.3	33.2+0.7	31.8+0.6	36	34	42
Sartor F	2010	UK	parallel	Obese subjects	19: 5 M, 14 F	4	41.5 ± 14.3	37.6 + 10.8	32.3 + 3.8	32.5 + 4.7	35	9	10
Wycherley TP	2010	Australia	Parallel	Obese subjects	49: 17 M, 32 F	52	49.9±1.7	50.2±1.4	33.5±0.8	33.9±0.8	20	26	23
Haufe S	2011	Germany	Parallel	overweight and obese	170: 35 M, 135 F	24	48.4± 9.6	49.23 ± 10.11	29.32± 1.76	30.44± 1.67	18	84	86
Kreider. RB	2011	USA	cross-over	Obese subjects	32: 17 M, 15 F	4	30.3 ± 5.7	30.3 ± 5.7	34.4 ± 4.9	34.6 ± 5.3	40	32	32
Rolland C	2011	UK	Parallel	Obese subjects	72: 30 M, 42 F	36	45.8±13.8	41.9±6.5	40.6±3.7	46.7±9.0	20	38	34
Ebbeling CB	2012	USA	crossover	Obese subjects	21: 13 M, 8 F	4	30.3 ±5.7	30.3 ±5.7	34.4±4.9	34.4±4.9	10	21	21
Rajaie, S	2012	Iran	cross-over	Metabolic syndrome	39: 39 F	24	42.4±7.2	42.4±7.2	31.7+5.0	31.7+5.0	43	39	39
Kitabchi AE	2013	USA	Parallel	Obese subjects	32: 32 F	24	35.9±2.1	35.4±2.0	41.3+1.8	37.0+1.5	40	14	18
Numao S	2013	Japan	cross-over	healthy	11: 11 M	4	32.7±4.1	32.7±4.1	21.6±0.5	21.6±0.5	19	11	11
Ruth MR	2013	USA	Parallel	Obese subjects	55: 48 M, 7 F	12	43.5±11.5	41.5±12.8	37.1±4.6	35.9±4.8	10	29	26
Llanos AA	2014	USA	Parallel	obese humans	79: 79 F	52	30-60	30-60	30.1+2.6	30.1+2.6	40	38	41
Thompson HJ	2014	USA	parallel	breast cancer	137: 137 F	24	55.2 ± 8.9	54.5 ± 9.2	29.4 ± 2.5	28.2 ± 2.4	32	65	72
Hu T	2015	USA	Parallel	obese adults	148: 17 M, 131 F	48	45.8±9.9	47.8±10.4	35.2±3.8	35.6±4.5	34	75	73
Youssef M	2015	Qatar	Parallel	overweight	12: 12 F	6	44.5±9.3	43.1±8.4	29.1±3.9	27.9±3.5	30	6	6
de Luis DA	2015	Spain	Parallel	obese adults	331: 85 M, 246 F	36	50.5+13.1	49.9+12.0	35.4+5.3	35.1+6.1	34	168	163
Kani AH	2017	Iran	Parallel	NAFLD	30: 14 M, 16 F	8	49.3 ± 3.5	45.6 ± 2.6	28.5±5.2	27.8±4.1	45	15	15
Ebbeling CB	2018	USA	Parallel	obesity	111: 32 M, 79 F	20	37.1 ±13.3	39.8± 15.1	32.0± 4.8	31.7± 4.3	20	57	54
Ohlsson B	2018	Sweden	cross-over	T2DM	30: 13 M, 17 F	12	57.5 ± 8.2	57.5 ± 8.2	29.9 ± 4.1	29.9 ± 4.1	42	30	30
Aller R	2019	Spain	parallel	obese adults	122: 70 M, 52 F	36	39.8±5	37.6±7	34.8+5.1	34.9+5.1	33	65	57
Strath LJ	2016	USA	Parallel	Osteoarthritis	14: 6 M, 8 F	12	71 + 3.12	72.33 + 1.97	35.64 + 7.35	29.65 + 4.48	20	8	6

IG, intervention group; CG, control group; NR, not reported; F, Female; M, Male; NR, not reported

Quality assessment

Thirty- three studies mentioned lower risk of bias related with random sequence generation and in two studies it had unclear. Nineteen studies had lower risk of bias related with allocation concealment, seven studies had higher risk and nine studies had unclear status related to the allocation concealment. About Blinding of participants, all of the studies had higher risk of bias except three studies which had unclear status. Eight studies had mentioned lower risk for blinding of outcome assessment, six trials had unclear-risk status and 21 trials had higher risk of bias. In twenty-six included trials it has been shown a lower risk, seven trials had unclear and two trials had higher risks for incomplete data outcome assessment. All of the included studies had lower risk of bias for reporting bias except six studies which had unclear status. Also, all of the trials had “other source of bias” except three studies with higher risk and five studies with unclear status. Details of risk of bias assessment are described in Fig. 2.

Fig. 2.

The methodological quality of included studies on effect of low-carbohydrate diet on serum concentration of leptin and adiponectin based on review authors’ judgments about each risk of bias item for each included study

Meta-analysis results

Effect of low-carbohydrate diet on adiponectin level

The effect of LCD on adiponectin level has been investigated in 25 trials with 26 effect sizes including 1289 participants (687 interventions and 602 controls). Pooled effect size from the random-effects model demonstrated that LCD vs. control diet could not significantly increase adiponectin concentration (WMD: 0.32 ng/ml, 95% CI: - 0.02, 0.66, p=0.062) with moderate heterogeneity (I² = 34.8%, p=0.0.04) (Fig. 3). We stratified studies based on participant’s age (<45 years and ≥45 years), BMI (>35 kg/m² and ≤35), trial duration (<6 months / ≥6 months) and carbohydrate percent (>30% and <30%). Subgroup analysis showed that LCD increased adiponectin concentration in subjects under 45 years old (WMD: 0.36 ng/ml, 95% CI: 0.03, 0.68, P=0.03), long term trials (WMD: 0.5 ng/ml, 95% CI: 0.07, 0.94, P=0.023) (Table 2).

Fig. 3.

Forest plot for the effect of low-carbohydrate diet on serum adiponectin concentration, expressed as mean differences between the intervention and the control diets

Table 2.

Results of subgroup-analysis for effect of low carbohydrate diet on adiponectin and leptin levels

	No. of effect sizes	WMD (95% CI)	P within1	I2 (%)
Subgroup analyses for adiponectin and low-carbohydrate diet
Duration of follow up
Less than 6 months	12	0.02 (-0.6, 0.64)	0.95	37
6 months and more	14	0.5 (0.07, 0.94)	0.023	36.7
Participant’s age
<45 years	11	0.36 (0.03, 0.68)	0.03	22.3
≥45 years	15	0.25 (-0.5, 1)	0.51	47.7
BMI
≤35 kg/m2	9	0.48 (-0.03, 0.99)	0.06	26.3
>35 kg/m2	17	0.21 (-0.29, 0.71)	0.41	44.3
CHO percent
>30%	12	0.49(-0.23, 0.75)	0.37	48.2
<30%	14	0.09(-0.84, 1.02)	0.44	39.7
Subgroup analyses for leptin and low-carbohydrate diet
Duration of follow up
Less than 6 months	16	0.04 (-2.92, 3.01)	0.97	70.6
6 months and more	12	-1.61 (-5.31, 2.09)	0. 39	85.6
Participant’s age
<45 years	13	1.98(-1.57, 5.54)	0.274	67.2
≥45 years	15	-2.68(-5.68,0.32)	0.08	83.7
BMI
≤35 kg/m2	19	-0.64 (-3.39, 2.11)	0.46	77.9
>35 kg/m2	9	-1.07 (-5.30, 3.16)	0.49	79.6
CHO percent
<30%		-1.33(-4.14, 1.48)	0.274	65.6
>30%		-0.3(-4.11,3.51)	0.08	86.9

Effect of low-carbohydrate diet on leptin level

Twenty-five with 28 effect size including a total of 1434 participants (802 interventions and 632 controls) indicated LCD compared to the control diet could not significantly changed leptin concentration (WMD: - 0.77 ng/ml, 95% CI: -3.15, 1.61, P=0.409). There was low heterogeneity between studies (I² = 80.5, P <0.001) (Fig. 4). Subgroup analysis demonstrated that there were no significant differences in the effects of LCD on leptin levels based on participant’s age (<45 years and ≥45 years), BMI (>35 kg/m² and ≤35), quality (high or low), trial duration (<6 months / ≥6 months), carbohydrate percent (>30% and <30%).

Fig. 4.

Forest plot for the effect of low-carbohydrate diet on serum leptin concentration, expressed as mean differences between the intervention and the control diets

Publication bias

Publication bias assessment was performed based on visual inspection of funnel plot and Egger’s linear regression test. Results showed that there wasn’t any publication bias for adiponectin (P=0.33) (Fig. 5) and leptin (P=0.151) (Fig. 6).

Fig. 5.

Funnel plots detailing publication bias in the selected studies of the relation between consumption of low-carbohydrate diet and circulating adiponectin

Fig. 6.

Funnel plots detailing publication bias in the selected studies of the relation between consumption of low-carbohydrate diet and circulating leptin

Sensitivity analysis

The sensitivity analysis also conducted to assess the impact of each individual study on the pooled effect size by removing each study in turn. The sensitivity analysis showed that the result was not significantly influenced by any of studies interlarded with adiponectin and leptin levels.

Discussion

Adipokines, a group of molecules secreted by the adipose tissue and fat mass, exert a wide range of biologic effects in body [23]. Their connection to different aspects of human health is rather a new topic and has attracted many researchers’ interests so far. Adiponectin and leptin are two of main adipokines and recent researches found an important relation between abnormal levels of them and diseases such as type 2 diabetes mellitus, cardiovascular disease and some cancers [24, 25]. Higher levels of adiponectin play a conservative role against these diseases whereas higher leptin concentrations play an adverse role [26]. It is clear now that life style modification such as calorie restriction and adhering to a healthy diet have a drastic effect in promoting a healthy adipokine profile [24].

To our knowledge, the present meta-analysis is the first study that evaluated the effects of low carbohydrate diet on serum adiponectin and leptin concentrations. Our results found insufficient evidence to support the effect of LCD on adiponectin and leptin levels. However, there was moderate heterogeneity between studies examined adiponectin on account of age and trial duration. The subgroup analysis indicated a noticeable increase in adiponectin level occurred in subjects under 45 years old and long-term trials.

In consistent with our results, other studies found no substantial effect of LCD on both adipokines [27–30]. These studies suggested that the composition of macronutrients has no significant effect on adiponectin increment or leptin decline and just pointed out that restricted calorie accompanied by a lost in fat mass and weight can lead to an improvement in these adopokines concentrations [27–30]. However, Summer et al. indicated that adiponectin increased only in LCD whereas low fat diet and weight lost had not such effect. They were mentioned that short term of the study can be a cause of this result and they may have seen no differences between diets if they had conducted a longer investigation [31]. These findings raise the possibility that the regulation of adiponectin and leptin synthesis and the rate of release of these adipokines responds differentially to specific constituents of the diet and that this response could mediate distinct metabolic effects of weight loss.

Another important factor is the type of adiponectin complexes. Adiponectin exists in serum in 3 forms: low molecular weight (LMW), high molecular weight (HMW) and medium molecular weight (MMW) [32, 33]. A recent research, which examined the changes of adiponectin forms in response to a diet-induced weight loss indicated that HMW and MMW complexes increased whereas the total amount did not change significantly [33].

In the evaluated studies, total adiponectine was measured and this can be a possible explanation of why no relationships were seen in our meta-analysis.

No associations were found between adhering to LCD and leptin concentration in our study. Further subgroup analysis revealed no changes in results. Previous studies indicated that LC diet reduces leptin concentration in a greater amount compared to low fat and conventional diet [30, 34, 35]. It is suggested that this effect can be due to caloric restriction, a decrease in adipocyte glucose metabolism, lower insulin level in plasma or due to an increase in leptin sensitivity [29, 36, 37]. On the other hand, one of the reasons for the change in leptin levels in some studies is weight loss, which has a significant effect on its concentration. However, we excluded studies that examined the simultaneous effect of weight loss and LCD on leptin and adiponectin levels and we evaluated the independent effect of LCD.

The secretion of leptin and adiponectin from adipocytes is directly linked to glucose metabolism and to the equilibrium between pro-and anti-inflammatory cytokines. Leptin production and secretion are stimulated in response to increased concentrations of insulin and inflammatory cytokines. Reduced insulin reaction and decline in inflammatory cytokines, as seen in animal models and obese humans when using high protein and LCD diets, could thus decrease the post-prandial leptin release [38–40]. Some studies have also shown that diets high in carbohydrates and fats increase leptin resistance, and limiting the carbohydrates and fats that occur in high-protein diets improves leptin resistance [41, 42]. Previous studies found that LCD may effect on leptin only in short term and there is no significant difference between LCD and LFD in longer time [5, 43]. It has been suggested that weight loss can occur more effective by LCD than LF diet particularly in short term intervention and this can be an explanation for differences observed between LCD and other diets. As well, decline in leptin concentrations did not sustain more than 1 year in LCD diet with weight regain and it returned to the baseline levels after 36 months [31].

On the other hand, circulating adiponectin levels are inversely related to insulin concentrations [44, 45]. One of the possible reasons why LCD was effective for change in serum levels of adiponectin might be dietary composition of protein. The low-carbohydrate diet is accompanied by high protein [46, 47]. It has been reported that higher protein intake is associated with higher adiponectin concentration [48]. It has been reported in some previous studies that increase in the serum levels of adiponectin after adherence from the LCD or high protein diets could thus be the result of weight loss, rather than alterations in macronutrient composition [45, 49]. However, the results of some other studies contradict these findings [50]. Besides, weight lost repeatedly shown that contributes in lowering leptin levels. Several studies mentioned that the weight lost that occurred as a result of LCD is responsible for this effect not the macronutrient composition of diet [27, 28, 30]. Previous studies found that LCD may effect on leptin only in short term and there is no significant difference between LCD and LFD in longer time [5, 43]. Some studies suggest that weight loss can occur more effective by LCD than LF diet particularly in short term and this can be an explanation for differences observed between LCD and other diets. As well, decline in leptin concentrations did not sustain more than 1 year in LCD diet with weight regain and it returned to the baseline levels after 36 months [6, 31]. Similar results were also indicated by another study in which leptin levels raised after a decrease in serum of people who regained weight after 5 months adhering LCD [43]. This can also illustrate that LCD is not an independent factor in modifying leptin levels and may exert its effect by affecting weight loss.

The present article was performed based on comprehensive and systematic search to found all relevant papers. Investigators conducted the publication bias assessment using Egger’s test and Begg’s test and there was no evidence for publication bias. Because leptin and adiponectin have inflammatory and anti-inflammatory effects, respectively, the results of this study will help clinician who prescribe macronutrient-restricted diets.

Following items can be considered the limitations for our study: Firstly, the quality scores of included studies varied from low to high and had a limited number of high-quality studies. One of the main reasons why these studies did not receive full quality scores were unlikely to use blinding and concealment of allocation. Second, although all participants in the evaluated studies received a low-carb diet, the percentage of carbohydrate intake varied from study to study ranged from 5 to 20% of daily energy from carbohydrates. The carbohydrate content may directly lead to high heterogeneity and affect the summarized results.

Conclusions

In conclusion, based on the high-quality studies we not found any significant effects from the LCD intervention on leptin and adiponectin concentration. In view of the current condition in the implementation of LCD, we conclude that a variety of questions need to be addressed in the future. One of the important issues was the duration of the interventions. As mentioned, in some studies with short intervention time, significant results were seen, while longer intervention time caused the results to be insignificant. Perhaps one of the reasons for this observation is the dietary compliance from LCD in the studies with long duration time. Future studies should improve the dietary compliance and focus on the long-term efficacy of LCD.

Abbreviations

CI

confidence intervals

HMW

high molecular weight

LMW

low molecular weight

SE

standard error

WMD

weighted mean difference

Authors’ contributions

NM, RT, ARA, MR: conducted the research; RT and JH: analyzed the data; NM, MR, AR and ARA: wrote the manuscript; MR and NM: had primary responsibility for the final content; and all authors: designed the research, and read and approved the final manuscript.

Funding

This study was financially supported by Zanjan University of Medical Sciences, Zanjan, Iran. The funder had no role in the undertaking, data analyses, or reporting of this systematic review.

Data availability

All data generated or analyzed during this study are included in this published article.

Declarations

Consent for publication

Not applicable.

Ethics approval and consent to participate

Not applicable.

Conflict of interest

The authors declare that they have no competing interests.