Effect of vitamin D supplementation on body composition, lipid profile, and glycemic indices in patients with obesity-associated metabolic syndrome: a systematic review and meta-analysis

Abstract

Background

Vitamin D deficiency is common among individuals with obesity and metabolic disorders. Evidence on its effect on metabolic markers remains inconclusive. This systematic review and meta-analysis aimed to evaluate the impact of vitamin D supplementation on lipid profile, glycemic control, and anthropometric indices.

Methods

A comprehensive search of PubMed, Embase, Cochrane Library, Web of Science, and China Knowledge Network was conducted from inception to May 2024. Randomized controlled trials (RCTs) evaluating oral vitamin D supplementation in adults with overweight, obesity, or metabolic disorders were included. Weighted mean differences (WMDs) and 95% confidence intervals (CIs) were calculated using a random-effects model.

Results

Twenty-five RCTs (30 arms) were included. Overall, vitamin D supplementation did not significantly affect triglycerides (TG), total cholesterol (TC), HDL-C, LDL-C, fasting plasma glucose (FPG), insulin, HOMA-IR, HbA1c, waist circumference (WC), or body weight (p < 0.05). Subgroup analyses showed significant effects in certain populations, such as increased HDL-C in individuals with diabetes and elevated LDL-C in younger adults and males. A significant reduction in BMI was observed after adjusting for publication bias (p < 0.05). Heterogeneity varied across outcomes, and risk of bias was generally low, although some studies had unclear reporting.

Conclusion

Vitamin D supplementation has limited effects on metabolic and anthropometric markers in adults with obesity or related metabolic disorders. Certain subgroups may benefit, warranting further targeted research.

Introduction

Metabolic disorders such as metabolic syndrome and type 2 diabetes are reaching epidemic proportions worldwide. Metabolic syndrome characterized by central obesity, dysglycemia, hypertension and dyslipidemia affects roughly one-quarter of the global population [1] (and about 40% of adults in some countries [2]. These clusters of metabolic risk factors greatly increase the risk of cardiovascular disease and diabetes. The underlying drivers include excess adiposity and insulin resistance, which promote chronic inflammation and atherogenic lipid profiles [3]. Against this backdrop, novel interventions to improve glycemic control, body weight and lipid metabolism are urgently sought.

Vitamin D, a fat-soluble secosteroid obtained from skin synthesis or diet, is well known for its role in calcium homeostasis, but it also has diverse “extraskeletal” actions [4]. The active form 1,25-dihydroxyvitamin D binds to vitamin D receptors (VDR) expressed in many tissues including pancreatic β-cells, skeletal muscle, adipocytes and immune cells [5]. Through endocrine and autocrine mechanisms, vitamin D influences cell proliferation, immune modulation and metabolic pathways. In particular, vitamin D has been shown to exert anti-inflammatory, anti-fibrotic and insulin-sensitizing effects [6–8]. Interest in these roles has surged because vitamin D insufficiency is widespread, nearly one-third of people globally have suboptimal levels [9, 10]. Accordingly, investigators have explored whether improving vitamin D status might favorably affect metabolic regulation.

Correlative studies indeed link low 25-hydroxyvitamin D [25(OH)D] with worse metabolic health. Cross-sectional analyses show an inverse relationship between serum 25(OH)D and body mass index (BMI), insulin resistance (HOMA-IR) and hemoglobin A1c [11]. A meta-analysis of observational studies found that higher 25(OH)D is associated with substantially lower risk of type 2 diabetes (≈ 55% risk reduction) and metabolic syndrome (≈ 51% reduction) [12]. Mechanistically, vitamin D has been proposed to upregulate insulin receptor expression and lipoprotein lipase activity, and to modulate the renin–angiotensin system and adipokine secretion [1]. Several randomized trials and meta-analyses report modest metabolic benefits of vitamin D supplementation. A recent meta-analysis of 39 trials (n ≈ 3000) found that vitamin D significantly lowered fasting plasma glucose (FPG), HbA1c, HOMA-IR and fasting insulin compared to placebo [13]. However, other studies have found little or no benefit. A meta-analyses of weight outcomes have generally found no significant effect of vitamin D on BMI [14]. These inconsistent findings may stem from heterogeneity among trials. Studies have varied widely in patient population (obese vs. non-obese, diabetic vs. pre-diabetic), baseline vitamin D status, supplement dose (IU and interval), and duration of follow-up. Co-interventions (e.g. calcium) and study quality may also modulate outcomes. Thus, the overall evidence for metabolic effects of vitamin D remains equivocal.

Given these mixed data and the lack of consensus, we undertook a systematic review and meta-analysis of randomized trials to clarify the effects of vitamin D supplementation on metabolic health. Specifically, we focus on adult patients with metabolic disorders (such as metabolic syndrome, obesity/overweight or diabetes) and assess key outcomes of interest: glycated hemoglobin (A1c), FPG, fasting insulin and HOMA-IR (insulin resistance), lipid profile (high- and low-density lipoprotein (HDL and LDL), total cholesterol (TC), triglycerides (TG)), and obesity markers (BMI). By synthesizing all available evidence with no date restrictions, this analysis aims to determine whether vitamin D supplementation produces consistent improvements in glycemic control, lipid metabolism, or obesity indices in this high-risk population.

Methods

Protocol and registration

This systematic review and meta-analysis was conducted in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA)-2020 guidelines [15].

Eligibility criteria

We included randomized controlled trials (RCTs) that met the following criteria based on PICOS format: Population: Adults (≥ 18 years) with overweight, obesity, or metabolic disorders (e.g., type 2 diabetes mellitus [T2DM], prediabetes, non-alcoholic fatty liver disease [NAFLD], or polycystic ovary syndrome [PCOS]); Intervention: Oral vitamin D supplementation (any form, dose, or duration), either as a single agent or in combination with lifestyle interventions; Comparison: Placebo or no supplementation; Outcomes: At least one of the following: TG, TC, HDL-C, LDL-C, FPG, fasting insulin, HOMA-IR, HbA1c, BMI, WC, or body weight; Study Design: Randomized controlled trials with either parallel or crossover designs.

We excluded animal studies, observational designs, non-English publications, and trials without relevant outcome data.

Data sources and search strategy

A systematic literature search was conducted in PubMed, Embase, Cochrane Library, Web of Science, and China Knowledge Network was conducted from inception to May 2024. Search terms included combinations of: vitamin D, cholecalciferol, ergocalciferol, supplementation, randomized controlled trial, obesity, diabetes, lipids, glucose, BMI, and related terms. Reference lists of included articles and relevant reviews were manually screened for additional studies.

Study selection

All retrieved articles were imported into EndNote for duplicate removal. Two independent reviewers (LG and YW) screened titles and abstracts, followed by full-text assessments for eligibility. Discrepancies were resolved through discussion or consultation with a third reviewer (LH).

Data extraction

Data were independently extracted by two reviewers using a standardized form. Extracted data included: first author, publication year, country, sample size, participant characteristics (age, sex, BMI, and health status), intervention details (vitamin D form, dose, and duration), control group characteristics, and reported outcomes (baseline and post-intervention means and SDs). Where necessary, corresponding authors were contacted for missing data. In cases where the SD was missing but the standard error (SE) or confidence intervals (CI) were available, we derived the SD using the formula SD = SE×$\surd n$ [16].

Risk of bias assessment

Risk of bias for each included study was assessed using the Cochrane Collaboration’s Risk of Bias tool [17], covering seven domains: random sequence generation, allocation concealment, blinding of participants/personnel, blinding of outcome assessment, incomplete outcome data, selective reporting, and other sources of bias. Each domain was rated as “low”, “unclear”, or “high” risk.

Data synthesis and statistical analysis

Data were pooled using random-effects models to calculate weighted mean differences (WMDs) with 95% confidence intervals (CIs). Heterogeneity was assessed using the I² statistic, with thresholds of 25%, 50%, and 75% representing low, moderate, and high heterogeneity, respectively. Subgroup analyses were performed based on age, sex, BMI, health condition, vitamin D dose, study duration, and sample size. Sensitivity analyses were conducted by omitting one study at a time. Publication bias was evaluated using Egger’s [18] and Begg’s tests [19], and funnel plots. When bias was suspected, trim-and-fill analyses were conducted to adjust the pooled effect estimates. All statistical analyses were conducted using STATA version 17.0 (StataCorp, College Station, TX, USA). A two-sided p-value < 0.05 was considered statistically significant.

Results

Characteristics of included studies

The PRISMA flowchart is illustrated in Fig. 1. A total of 25 RCTs comprising 30 arms were included in the meta-analysis. The studies were conducted across diverse geographic regions, including the USA (n = 6) [20–25], Norway (n = 3) [26–28], Iran (n = 3) [29–31], Italy (n = 2) [32, 33], Denmark (n = 2) [34, 35], and one study each from Austria [36], China [37], Germany [38], Greece [39], India [40], Japan [41], Jordan [42], Australia [43], and the United Kingdom [44] (Table 1). The study populations included individuals with various metabolic conditions such as type 2 diabetes mellitus (T2DM), prediabetes, non-alcoholic fatty liver disease (NAFLD), polycystic ovary syndrome (PCOS), and vitamin D deficiency, as well as overweight or obese adults. The sample sizes varied widely, ranging from 14 to 2423 participants, and trial durations ranged from 4 to 48 weeks. Mean age of participants ranged from 23.76 to 65 years, and the baseline BMI values spanned from 21.9 to 42.7 kg/m², indicating the inclusion of both overweight and obese populations. Most studies enrolled both sexes, although a subset focused exclusively on either female (n = 10) or male (n = 1). All interventions involved vitamin D supplementation, most commonly in the form of cholecalciferol (vitamin D3), with daily doses ranging from 400 IU to 20,000 IU. Some trials combined vitamin D with lifestyle interventions such as weight loss or weight maintenance programs.

Fig. 1.

Study selection flow diagram

Table 1.

Characteristic of included studies in meta-analysis

Studies	Country	Study design	Participant	Sex	Sample size		Trial duration (Week)	Means age		Means BMI		Intervention
					IG	CG		IG	CG	IG	CG	Intervention type	Dose (IU/day)
Angellotti et al. (2020)	USA	RCT. PC.DB	T2DM	F/M	59	55	48	60.1 ± 8.4	60.3 ± 8.5	30.9	30.9	Vitamin D 3	4000
Barchetta et al. (2016)	Italy	RCT. PC.DB	T2D patients with NAFLD	F/M	26	29	24	57.4 ± 10.7	59.8 ± 9.1	29.3 ± 4.4	30.8 ± 4.5	Cholecalciferol	2000
Desouza et al. (2022)	USA	RCT. PC.DB	Patients with prediabetes	F/M	1211	1212	192	59.6 ± 9.9	60.4 ± 10	32.1 ± 4.5	32.1 ± 4.5	Vitamin D 3	4000
Jorde et al. (2010, A)	Norway	RCT. PC.DB	Overweight or obese subjects	F/M	114	56	48	47.5 ± 11.4	47.5 ± 11.4	34.7 ± 3.8	34.7 ± 3.8	Vitamin D (cholecalciferol, vitamin D3)	5714
Jorde et al. (2010, B)	Norway	RCT. PC.DB	Overweight or obese subjects	F/M	104	56	12	47.5 ± 11.4	47.5 ± 11.4	34.7 ± 3.8	34.7 ± 3.8	Vitamin D (cholecalciferol, vitamin D3)	2875
Kampmann et al. (2014)	Denmark	RCT. PC.DB	Patients with type 2 diabetes and vitamin D insufficiency	F/M	7	8	12	61.6 ± 4.4	57 ± 4.5	35.3 ± 2.9	32.4 ± 2	Colecalciferol	6533
Lerchbaum et al. (2019)	Austria	RCT. PC.DB	Healthy men	M	96	96	12	40 ± 17.03	47 ± 17.03	26.8 ± 4.81	26.4 ± 4.37	Vitamin D 3	2875
Lotfi-Dizaji et al. (2018)	Iran	RCT. PC.DB	Obese subjects with vitamin D deficiency	F/M	22	22	12	35.18 ± 7	34.9 ± 10.37	35.1 ± 3.61	36.11 ± 3.37	vitamin D 3	7142
Mai et al. (2017)	Italy	RCT. PC.SB	Severe obesity	F/M	12	12	4	38 ± 8.31	37 ± 10.39	42.7 ± 4.50	39.8 ± 3.11	Cholecalciferol	20,000
Mason et al. (2014)	USA	RCT. PC.DB	Overweight/obese women	F	94	93	48	60.3 ± 5.3	59 ± 4.7	32.3 ± 5.5	32.5 ± 6.1	Vitamin D 3	2000
Meng et al. (2020)	USA	RCT. PC.DB	T2DM	F/M	56	58	24	65 ± 9	65 ± 9	NR	NR	Vitamin D 3	4000
Mitri et al. (2011)	USA	RCT. PC.DB	Adults at high risk of diabetes	F/M	22	22	16	57 ± 9.38	59 ± 9.38	32 ± 4.69	32 ± 4.69	Cholecalciferol	2000
Sukumar et al. (2015, A)	USA	RCT. PC.DB	Overweight/obese older wome	F	19	20	6	57 ± 6	57 ± 6	31 ± 4	31 ± 4	Vitamin D 3 + weight loss	2500
Sukumar et al. (2015, B)	USA	RCT. PC.DB	Overweight/obese older wome	F	18	19	6	57 ± 6	57 ± 6	31 ± 4	31 ± 4	vitamin D 3 + weight maintenance	2500
Wen et al. (2024)	China	RCT. PC.DB	Polycystic ovary syndrome	F	28	29	12	26.7 ± 9.3	25.3 ± 8.6	25.2 ± 7.4	24.9 ± 7.9	Vitamin D	2000
Xenos et al. (2022)	Greece	RCT. PC.DB	Obese caucasians with vitamin d deficiency or insufficiency	F/M	76	49	12	40.14 ± 8.84	42.96 ± 11.07	35.26 ± 7.33	34.00 ± 6.37	Vitamin D 3 + weight loss	3000
Al-Bayyari et al. (2018)	Jordan	RCT. PC.DB	Overweight reproductive women	F	50	48	8	23.76 ± 4.38	23.31 ± 5.19	27.28 ± 1.90	26.85 ± 1.66	Vitamin D 3	7142
Khosravi et al. (2018)	Iran	RCT. PC.DB	Overweight and obese women	F	26	27	6	29.1 ± 9.6	26.9 ± 9.1	28 ± 2.7	27.8 ± 2.6	Vitamin D	7142
Kjærgaard et al. (2012)	Norway	RCT. PC.DB	People with vitamin D insufficiency	F/M	120	110	24	53.4 ± 10.3	53.3 ± 10.1	27.5 ± 4	28 ± 4.2	Vitamin D 3	5714
Mousa et al. (2017)	Australia	RCT. PC.DB	Overweight or obese adults	F/M	28	26	16	30.5 ± 7.40	29.5 ± 11.85	30.2 ± 4.51	29.8 ± 3.77	Cholecalciferol	4893
Nagpal et al. (2009)	India	RCT. PC.DB	Healthy, middle-aged, centrally obese men	M	35	36	6	42.4 ± 6.6	45.0 ± 9.2	26.7 ± 4.54	26.0 ± 3.46	Vitamin D 3	7619
Salehpour et al. (2012)	Iran	RCT. PC.DB	Healthy overweight and obese women	F	39	38	12	38 ± 7	37 ± 8	30.1 ± 3.9	29.5 ± 4.4	Cholecalciferol	1000
Sneve et al. (2008, A)	Norway	RCT. PC.DB	Overweight and obese subjects	F/M	116	56	48	46.4 ± 11.3	48.9 ± 11	35 ± 4.1	35.1 ± 3.8	Cholecalciferol + calcium	5714
Sneve et al. (2008, B)	Norway	RCT. PC.DB	Overweight and obese subjects	F/M	106	56	48	47.6 ± 11.9	48.9 ± 11	34.4 ± 3.9	35.1 ± 3.8	Cholecalciferol + calcium	2857
Sun et al. (2016)	Japan	RCT. PC.DB	Healthy adults	F/M	40	41	48	45	43	21.9 ± 2.52	22.2 ± 2.56	Vitamin D	420
Wamberg et al. (2013)	Denmark	RCT. PC.DB	Obese adults with low vitamin D levels	F/M	22	21	26	39.5 ± 8.0	41.2 ± 6.8	36.1 ± 3.4	35 ± 3.2	Cholecalciferol	7000
Wood et al. (2012, A)	United Kingdom	RCT. PC.DB	Healthy postmenopausal women	F	84	46	48	63.5 ± 1.9	63.9 ± 2.3	26.6 ± 4.2	26.6 ± 4.4	Vitamin D 3	400
Wood et al. (2012, B)	United Kingdom	RCT. PC.DB	Healthy postmenopausal women	F	90	45	48	64.1 ± 2.3	63.9 ± 2.3	26.8 ± 4.2	26.6 ± 4.5	Vitamin D 3	1000
Zittermann et al. (2009)	Germany	RCT. PC.DB	Healthy overweight subjects	F/M	82	83	48	47.4 ± 10.3	48.8 ± 10.1	101.9 ± 16.1	96.1 ± 15.0	Vitamin D	3320

IG, intervention group; CG, control group; DB, double-blinded; SB, single-blinded; PC, placebo-controlled; R, randomized; NR, not reported; F, Female; M, Male; NR, not reported

Quality assessment of included studies

The risk of bias was assessed across seven domains (Table 2). All studies were rated as low risk for random sequence generation and allocation concealment, indicating robust randomization procedures and appropriate concealment strategies. However, selective reporting was judged as unclear or high risk in 8 studies, and other sources of bias were found in 21 trials, most commonly due to potential conflicts of interest, short durations, or small sample sizes. Blinding of participants, personnel, and outcome assessment was consistently well-reported, with nearly all studies rated as low risk in these domains. Incomplete outcome data were also generally well managed across studies. Overall, the majority of the included RCTs were considered to have a low risk of bias, suggesting a moderate to high methodological quality of evidence in this meta-analysis.

Table 2.

Quality assessment

Studies	Random sequence generation	Allocation concealment	Selective reporting	Other sources of bias	Blinding (participants and personnel)	Blinding (outcome assessment)	Incomplete outcome data
Angellotti et al. (2020)	L	L	U	H	L	L	L
Barchetta et al. (2016)	L	L	L	H	L	L	L
Desouza et al. (2022)	L	L	L	H	L	L	L
Jorde et al. (2010, A)	L	L	L	L	L	L	L
Jorde et al. (2010, B)	L	L	L	L	L	L	L
Kampmann et al. (2014)	L	L	H	H	L	L	L
Lerchbaum et al. (2019)	L	L	U	H	L	L	L
Lotfi-Dizaji et al. (2018)	L	L	L	L	L	L	L
Mai et al. (2017)	L	L	H	H	L	H	L
Mason et al. (2014)	L	L	L	L	L	L	L
Meng et al. (2020)	L	L	H	H	L	L	L
Mitri et al. (2011)	L	L	H	H	L	L	L
Sukumar et al. (2015, A)	L	L	H	H	L	L	L
Sukumar et al. (2015, B)	L	L	H	H	L	L	L
Wen et al. (2024)	L	L	H	H	L	L	L
Xenos et al. (2022)	L	L	L	U	L	L	L
Al-Bayyari et al. (2018)	L	L	L	L	L	L	L
Khosravi et al. (2018)	L	L	H	H	L	L	L
Kjærgaard et al. (2012)	L	L	U	H	L	L	L
Mousa et al. (2017)	L	L	L	L	L	L	L
Nagpal et al. (2009)	L	L	L	H	L	L	L
Salehpour et al. (2012)	L	L	L	L	L	L	L
Sneve et al. (2008, A)	L	L	L	H	L	L	L
Sneve et al. (2008, B)	L	L	L	H	L	L	L
Sun et al. (2016)	L	L	L	H	L	L	L
Wamberg et al. (2013)	L	L	L	H	L	L	L
Wood et al. (2012, A)	L	L	H	H	L	L	L
Wood et al. (2012, B)	L	L	H	H	L	L	L
Zittermann et al. (2009)	L	L	L	H	L	L	L

Effect of vitamin D on TG

The overall pooled analysis indicated that vitamin D supplementation did not significantly affect TG levels (WMD: –2.36 mg/dL; 95% CI –6.24 to 1.51; I² = 53.2%, p = 0.232) (Table 3, Fig. 2A). Subgroup analysis revealed a significant reduction in TG among participants aged ≤ 50 years, whereas no significant effect was found in those > 50 years. No statistically significant effects were observed across subgroups based on sex, intervention duration, health condition, sample size, dosage, or BMI category (Table 3). Sensitivity analysis confirmed the robustness of findings. Egger’s and Begg’s tests showed no evidence of small-study effects (p = 0.232 and p = 0.913, respectively), although the asymmetry was found in funnel plot (Fig. 2B), indicating potential publication bias. Trim-and-fill analysis added two imputed studies and slightly strengthened the estimated effect (WMD: –3.60 mg/dL; 95% CI –7.81, 0.61; p = 0.093), but the result remained non-significant.

Table 3.

Subgroup analyses for the effects of vitamin D supplementation on lipid profile

	NO	WMD (95% CI)	I2 (%)	P-heterogeneity
Vitamin D on TG
Overall	14	− 2.36 (− 6.24, 1.51)	53.2	0.232
Age (year)
≤ 50	9	− 5.93 (− 11.40, − 0.47)	27.5	0.033
> 50	5	0.86 (− 4.18, 5.90)	47.5	0.738
Sex
Female	3	− 0.70 (− 5.24, 3.84)	0.0	0.762
Male	2	0.75 (− 21.16, 22.66)	64.5	0.946
Both	9	− 3.11 (− 8.81, 2.60)	62.9	0.286
Intervention duration (week)
≤ 15	4	− 2.06 (− 14.18, 10.05)	35.8	0.738
˃15	10	− 2.17 (− 6.45, 2.11)	61	0.320
Sample size
≤ 100	6	4.68 (− 10.14, 19.50)	62.7	0.536
> 100	8	− 2.80 (− 6.90, 1.30)	36.0	0.180
Dosage (IU)
< 5000	10	− 3.38 (− 7.50, 0.74)	61.4	0.108
≥ 5000	4	6.42 (− 4.76, 17.59)	0.0	0.260
Health conditions
Diabetic	3	11.35 (− 7.85, 30.55)	72.4	0.247
Overweight/obese	5	− 8.32 (− 20.87, 4.23)	46.9	0.194
Healthy	3	− 4.94 (− 10.86, 0.98)	30.4	0.102
Obese with low vitamin D deficiency	1	0.00 (− 39.46, 39.46)	−	1.000
Postmenopausal	2	− 1.03 (− 5.62, 3.57)	0.0	0.661
BMI
≤ 30	7	− 1.05 (− 6.34, 4.24)	60.3	0.697
˃30	7	− 4.83 (− 13.29, 3.63)	46.1	0.263
Vitamin D on TC
Overall	14	1.81 (− 1.36, 4.98)	72.0	0.262
Age(year)
≤ 50	7	2.83 (− 2.32, 7.98)	71.5	0.281
> 50	7	1.05 (− 2.69, 4.78)	43.0	0.583
Sex
Female	3	− 0.30 (− 7.12, 6.51)	55.8	0.930
Male	2	8.93 (2.19, 15.67)	0.0	0.009
Both	9	1.41 (− 2.50, 5.31)	76.5	0.480
Intervention duration (week)
≤ 15	5	7.41 (2.88, 11.93)	0.0	0.0001
˃15	9	− 0.10 (− 3.59, 3.40)	75.4	0.956
Sample size
≤ 100	7	1.26 (− 4.36, 6.87)	49.7	0.661
> 100	7	2.29 (− 0.93, 5.50)	50.3	0.163
Dosage (IU)
< 5000	9	2.04 (− 1.72, 5.79)	80.7	0.288
≥ 5000	5	1.24 (− 4.61, 7.08)	17.4	0.679
Health conditions
Diabetic	4	3.48 (1.53, 5.44)	0.0	< 0.0001
Overweight/obese	4	4.19 (− 0.23, 8.61)	0.0	0.063
Healthy	3	3.13 (− 7.35, 13.61)	84.4	0.558
Obese with low vitamin D deficiency	1	− 7.72 (− 21.15, 5.71)	−	0.260
Postmenopausal	2	− 3.12 (− 7.52, 1.28)	0.0	0.165
BMI
≤ 30	7	1.92 (− 3.02, 6.86)	71.3	0.447
˃30	7	3.12 (1.32, 4.93)	0.0	0.001
Vitamin D on HDL-C
Overall	16	0.02 (− 0.92, 0.96)	53.5	0.970
Age(year)
≤ 50	9	− 0.14 (− 1.40, 1.12)	43.8	0.827
> 50	7	0.37 (− 0.83, 1.58)	39.8	0.543
Sex
Female	3	0.34 (− 2.05, 2.73)	51.6	0.781
Male	2	− 0.26 (− 5.84, 5.32)	82.4	0.928
Both	12	0.11 (− 0.91, 1.14)	47.0	0.826
Intervention duration (week)
≤ 15	5	0.56 (− 2.05, 3.17)	59.0	0.675
˃15	11	− 0.07 (− 1.10, 0.96)	54.4	0.888
Sample size
≤ 100	7	− 0.13 (− 2.18, 1.92)	46.2	0.902
> 100	9	0.15 (− 0.87, 1.17)	52.8	0.776
Dosage (IU)
< 5000	11	− 0.11 (− 1.16, 0.94)	58.4	0.841
≥ 5000	5	0.54 (− 2.09, 3.16)	49.6	0.689
Health conditions
Diabetic	5	1.35 (0.57, 2.13)	4.5	0.0001
Overweight/obese	5	0.51 (− 0.68, 1.71)	0.0	0.400
Healthy	3	− 0.79 (− 3.46, 1.88)	65.6	0.563
Obese with low vitamin D deficiency	1	− 3.86 (− 8.88, 1.16)	−	0.132
Postmenopausal	2	− 0.64 (− 2.26, 0.98)	0.0	0.970
BMI
≤ 30	7	− 0.49 (− 2.04, 1.06)	49	0.534
˃30	8	0.77 (− 0.09, 1.64)	22.7	0.081
NR	1	0.0 (− 5.39, 5.39)	−	1.000
Vitamin D on LDL-C
Overall	15	2.18 (− 0.39, 4.74)	52.1	0.096
Age(year)
≤ 50	8	4.76 (0.45, 9.08)	56.6	0.031
> 50	7	0.11 (− 2.32, 2.53)	22.7	0.932
Sex
Female	3	− 0.21 (− 4.85, 4.43)	46.4	0.930
Male	2	10.64 (5.10, 16.18)	0.0	< 0.001
Both	10	1.55 (− 1.15, 4.24)	28.3	0.260
Intervention duration (week)
≤ 15	5	5.59 (− 0.54, 11.73)	62.8	0.074
˃15	10	0.66 (− 1.21, 2.52)	11.5	0.490
Sample size
≤ 100	6	0.83 (− 4.39, 6.06)	41.5	0.754
> 100	9	2.77 (− 0.32, 5.87)	60.5	0.079
Dosage (IU)
< 5000	11	2.47 (− 0.45, 5.40)	55.4	0.097
≥ 5000	4	1.18 (− 5.45, 7.80)	55.0	0.728
Health conditions
Diabetic	5	1.51 (− 0.67, 3.69)	3.7	0.174
Overweight/obese	5	4.44 (0.10, 8.78)	29.6	0.045
Healthy	3	5.00 (− 5.44, 15.43)	80.6	0.348
Postmenopausal	2	− 2.09 (− 5.58, 1.40)	0.0	0.240
BMI
≤ 30	7	2.74 (− 2.15, 7.62)	68.8	0.273
˃ 30	7	1.96 (− 1.27, 5.18)	39.1	0.234
NR	1	3.86 (− 12.44, 20.16)	−	0.643

Fig. 2.

Forest plot (A) and funnel plot (B) detailing mean difference and 95% confidence intervals (CIs) of the effects of vitamin D supplementation on TG levels

Effect of vitamin D on TC

Vitamin D supplementation had no significant effect on TC overall (WMD: 1.81 mg/dL; 95% CI –1.36 to 4.98; I² = 72.0%, p = 0.262) (Table 3, Fig. 3A), confirmed by sensitivity analysis. However, in subgroup analysis by sex, a significant increase in TC was observed among males, individuals with diabetes, and participants with BMI > 30. Furthermore, short-term interventions (≤ 15 weeks) were associated with a significant increase in TC, while longer durations showed no effect. No significant effects were found in other subgroups (Table 3). Egger’s (p = 0.243) and Begg’s tests (p = 0.661) did not indicate publication bias. However, an asymmetry was found in funnel plot (Fig. 3B). After applying the trim-and-fill method, one study was added, and the pooled estimate remained non-significant (WMD: 1.38 mg/dL; 95% CI –1.74, 4.49; p = 0.386).

Fig. 3.

Forest plot (A) and funnel plot (B) detailing mean difference and 95% confidence intervals (CIs) of the effects of Vitamin D supplementation on TC levels

Effect of vitamin D on HDL-C

Vitamin D supplementation had no overall effect on HDL-C levels (WMD: 0.02 mg/dL; 95% CI –0.92 to 0.96; I² = 53.5%, p = 0.970) (Table 3, Fig. 4A). Same results were found following sensitivity analysis. In subgroup analyses, a significant increase in HDL-C was observed only in individuals with diabetes. No significant effects were detected across other subgroups stratified by age, sex, intervention duration, sample size, dosage, or BMI (Table 3). Publication bias was also not detected based on Begg’s test (p = 0.893). However, Egger’s (p = 0.022) and funnel plot found an evidence for publication bias (Fig. 4B). Trim and fill analysis did not change the result on HDL-C (WMD: 0.02 mg/dL; 95% CI –0.92 to 0.96; p > 0.05).

Fig. 4.

Forest plot (A) and funnel plot (B) detailing mean difference and 95% confidence intervals (CIs) of the effects of vitamin D supplementation on HDL-C levels

Effect of vitamin D on LDL-C

Overall, vitamin D supplementation showed a non-significant trend toward increasing LDL-C (WMD: 2.18 mg/dL; 95% CI –0.39 to 4.74; I² = 52.1%, p = 0.096) (Table 3, Fig. 5A), confirmed by sensitivity analysis. Significant increases in LDL-C were found among participants aged ≤ 50 years and in males. Additionally, vitamin D significantly increased LDL-C in overweight/obese individuals. Other subgroup analyses, including by dosage, intervention duration, and BMI, did not show statistically significant effects (Table 3). Egger’s test did not indicate publication bias (p = 0.652), as well as Begg's test (p = 0.767). However, an asymmetry in funnel plot suggested a potential publication bias (Fig. 5B). After applying the trim-and-fill method, three studies were added, and the pooled estimate remained non-significant (WMD: 0.45 mg/dL; 95% CI –2.35, 3.26; p = 0.751).

Fig. 5.

Forest plot (A) and funnel plot (B) detailing mean difference and 95% confidence intervals (CIs) of the effects of vitamin D supplementation on LDL-C levels

Effect of vitamin D on FPG

The WMD was 0.13 mg/dL (95% CI − 2.18 to 2.43; p = 0.913; I² = 75.4%, p < 0.001), showing no significant effect of vitamin D supplementation on FPG (Fig. 6A). Sensitivity analysis indicated the results were robust with no single study significantly altering the pooled effect. Begg’s test (p = 0.276) and Egger’s test (p = 0.612) showed no evidence of publication bias. Funnel plot appeared symmetrical, supporting absence of bias (Fig. 6B).

Fig. 6.

Forest plot (A) and funnel plot (B) detailing mean difference and 95% confidence intervals (CIs) of the effects of vitamin D supplementation on FPG levels

Effect of vitamin D on insulin

The pooled WMD was 0.31 µIU/mL (95% CI − 0.43 to 1.05; p = 0.415; I² = 21.2%, p = 0.255), indicating no significant change in fasting insulin levels (Fig. 7). Sensitivity analysis confirmed the stability of the results with no influential outliers. Begg’s test (p = 1.000) suggested no publication bias.

Fig. 7.

Forest plot detailing mean difference and 95% confidence intervals (CIs) of the effects of vitamin D supplementation on insulin

Effect of vitamin D on HOMA-IR

The WMD was − 0.07 (95% CI − 0.35 to 0.20; p = 0.599; I² = 38.7%, p = 0.076), showing no significant effect on insulin resistance (Fig. 8A). Sensitivity analysis showed consistent results without significant impact from any individual study. Egger’s test (p = 0.495) and Begg’s test (p = 0.760) did not indicate publication bias; the funnel plot was visually symmetrical (Fig. 8B).

Fig. 8.

Forest plot (A) and funnel plot (B) detailing mean difference and 95% confidence intervals (CIs) of the effects of vitamin D supplementation on HOMA-IR levels

Effect of vitamin D on A1c

Vitamin D supplementation resulted in a WMD of − 0.002% (95% CI − 0.023 to 0.019; p = 0.871; I² = 12.2%, p = 0.335), indicating no significant change (Fig. 9). Sensitivity analysis confirmed robustness with no study exerting undue influence. Begg’s test (p = 1.000) supported lack of publication bias.

Fig. 9.

Forest plot detailing mean difference and 95% confidence intervals (CIs) of the effects of vitamin D supplementation on A1C

Effect of vitamin D on BMI

The random-effects model showed a non-significant WMD of − 0.28 kg/m² (95% CI − 0.65 to 0.10; p = 0.147; I² = 86.6%, p < 0.001) (Fig. 10A). Sensitivity analysis demonstrated stable findings without any influential studies. Begg’s test (p = 0.084) and Egger’s test (p = 0.730) indicated no strong publication bias, although funnel plots showed slight asymmetry (Fig. 10B), which was addressed by trim-and-fill. After trim-and-fill adjustment for publication bias, the WMD was − 0.37 (95% CI − 0.71 to − 0.04; p = 0.031).

Fig. 10.

Forest plot (A) and funnel plot (B) detailing mean difference and 95% confidence intervals (CIs) of the effects of vitamin D supplementation on BMI levels

Effect of vitamin D on WC

The pooled WMD was 0.07 cm (95% CI − 1.00 to 1.13; p = 0.904; I² = 45.6%, p = 0.076), indicating no significant effect (Fig. 11). Sensitivity analysis confirmed robustness of the results. Begg’s test (p = 0.902) indicated no publication bias.

Fig. 11.

Forest plot detailing mean difference and 95% confidence intervals (CIs) of the effects of vitamin D supplementation on WC

Effect of vitamin D on body weight

The overall WMD for body weight was − 0.11 kg (95% CI − 0.56 to 0.34; I² = 20.7%, p = 0.229), showing no significant effect of vitamin D supplementation on body weight (Fig. 12A). Leave-one-out sensitivity analysis demonstrated that omission of any single study did not substantially change the overall effect size. Begg’s test showed no evidence of publication bias (p = 0.622). However, Egger’s test indicated potential publication bias with a significant intercept (p = 0.035). This suggests some asymmetry in the funnel plot that may reflect small-study effects or bias (Fig. 12B). However, trim and fill analysis did not change the findings (WMD was − 0.11 kg (95% CI − 0.56 to 0.34)).

Fig. 12.

Forest plot (A) and funnel plot (B) detailing mean difference and 95% confidence intervals (CIs) of the effects of vitamin D supplementation on weight levels

Discussion

This meta-analysis of 25 RCTs with 30 intervention arms evaluated the impact of vitamin D supplementation on lipid profile, glycemic indices, and anthropometric parameters in overweight or obese individuals and those with metabolic disorders. Overall, the findings revealed no statistically significant effects of vitamin D supplementation on TG, TC, HDL-C, LDL-C, FPG, insulin levels, insulin resistance (HOMA-IR), A1c, WC, or body weight. However, subgroup analyses suggested potential benefits in specific populations; for instance, TG reduction in younger adults (≤ 50 years) and HDL-C improvement among diabetics. A slight but statistically significant reduction in BMI was observed only after adjusting for potential publication bias. These findings suggest that while vitamin D supplementation alone may not substantially alter cardiometabolic risk factors across all populations, it might exert modest or context-specific effects depending on age, sex, metabolic condition, or baseline BMI.

The results of this meta-analysis were marked by considerable heterogeneity across several outcomes. For instance, high between-study heterogeneity was observed in the effects on TC, FPG, and BMI, indicating substantial inconsistency in the effect sizes reported by individual trials. Several factors likely contributed to this heterogeneity. First, the included trials varied widely in sample size (ranging from 14 to 2423 participants), duration (4 to 48 weeks), health status of participants (T2DM, PCOS, NAFLD, overweight/obesity, vitamin D deficiency), and vitamin D dosage (400 to 20,000 IU/day). Additionally, most studies used vitamin D3 (cholecalciferol), but differences in formulation, delivery method, and adherence may have influenced outcomes. Despite these differences, the majority of included studies were assessed as having low risk of bias, particularly in domains related to random sequence generation, allocation concealment, blinding, and completeness of outcome data. This suggests that the internal validity of individual trials was generally sound. The presence of publication bias was suspected based on funnel plot asymmetry for several outcomes (e.g., HDL-C, TG, BMI), although this was not always confirmed by Egger’s or Begg’s tests. The trim-and-fill analysis led to a slight change in effect size in some cases, most notably a statistically significant reduction in BMI after adjustment. Collectively, while the included studies were methodologically moderate-to-strong, the substantial heterogeneity and potential publication bias should temper the interpretation of null or subgroup-specific findings.

Regarding TG, our result is consistent with a meta-analysis in metabolic syndrome patients, which reported no effect of vitamin D on TG [45]. In contrast, Mahboub et al. found a significant increase in TG (MD ≈ + 28–31 mg/dL) with both low- and high-dose vitamin D [46]. On the other hand, some RCT meta-analyses have found vitamin D modestly lowers TG [47]. Vitamin D supplementation had no significant effect on TC and HDL-C. Both Qi et al. and Mahboub et al. found no change in TC, LDL-C, and HDL-C with vitamin D [45, 46]. An umbrella meta-analysis similarly found no WMD change in TC, LDL-C, and HDL-C [47]. Observational data link low vitamin D with higher TC, LDL-C, and lower HDL-C [48], but RCTs of vitamin D generally report minimal changes. Regarding glycemic indices, our results were inconsistent with prior findings [13, 45, 49]. Heterogeneity was very high in these studies, reflecting differences in dose, duration, and baseline vitamin D/diabetes status. Sensitivity analyses (subgroup and leave-one-out) indicate this FPG reduction is driven by trials with vitamin D–deficient participants or higher doses/duration [13, 36, 49]. We observed a very modest reduction in BMI with vitamin D following trim and fill analysis. An umbrella review of meta-analyses found supplementation lowered BMI by ~ 0.11 kg/m² [50]. Overall body weight showed no significant change with vitamin D supplementation. Both umbrella evidence [50] and individual trials report non-significant weight differences. For instance, calcitriol treatment in PCOS patients did not alter weight or BMI after 3 months [51].

Few trials directly compare vitamin D forms for metabolic outcomes. In a large trial of adults at diabetes risk, both ergocalciferol (D₂) and cholecalciferol (D₃) raised 25(OH)D levels (to ~ 54 nmol/L and ~ 84 nmol/L respectively) but had no effect on HbA1c or lipid levels [52]. That study observed a similar non-response for glycemic markers with D₂ or D₃ over 4 months [52]. Calcitriol (the active 1,25(OH)₂D) has also been tested. In a PCOS trial, daily calcitriol (0.5 µg) had no significant impact on weight, BMI or insulin resistance, whereas metformin did [51]. There is no clear evidence that D₂, D₃ or calcitriol differ in their metabolic impact beyond their potency in raising 25(OH)D. In practice, D₃ is often preferred clinically for repletion because of its higher efficacy in increasing serum levels [53]. All forms operate via VDR-mediated pathways; differences in outcomes likely hinge more on achieved 25(OH)D or 1,25(OH)₂D levels rather than the compound itself [54].

Mechanistically, vitamin D increases lipoprotein lipase (LPL) expression in muscle and adipose tissue [55]. It also upregulates peroxisome proliferator-activated receptor alpha (PPAR-α), boosting fatty acid β-oxidation [56]. Furthermore, vitamin D reduces parathyroid (PTH) levels, which can otherwise suppress lipolytic activity [57]. In macrophages, VDR activation enhances reverse cholesterol transport and HDL formation, indirectly promoting TG utilization [58]. Insulin-sensitizing and anti-inflammatory effects of vitamin D may also improve lipid handling. Vitamin D augments insulin release via Ca²⁺ influx [59]. It also upregulates insulin receptor expression in muscle and adipose tissue, and increases the expression of PPAR-γ which promotes insulin synthesis [60]. Conversely, some studies have reported that vitamin D supplementation leads to increases in LDL cholesterol levels. One proposed mechanism is that vitamin D enhances intestinal calcium absorption, which may reduce the formation of insoluble calcium-fatty acid complexes, leading to increased absorption of saturated fats and subsequent elevation in LDL cholesterol [61]. Furthermore, role of vitamin D in regulating enzymes such as HMG-CoA reductase, a key enzyme in cholesterol synthesis, suggests that vitamin D status could influence endogenous cholesterol production [62]. These complex and sometimes opposing mechanisms highlight the need for further research to elucidate the conditions under which vitamin D supplementation may beneficially or adversely affect lipid profiles. Factors such as baseline vitamin D status, dosage, duration of supplementation, and individual metabolic conditions likely play significant roles in determining the lipid response to vitamin D.

This meta-analysis has several notable strengths. First, it is one of the most comprehensive syntheses to date, incorporating data from 25 RCTs and 30 intervention arms across diverse populations with various metabolic conditions, including obesity, T2DM, NAFLD, and PCOS. Second, we conducted detailed subgroup analyses to explore sources of heterogeneity, including age, sex, BMI, health condition, vitamin D dosage, and study duration, which provides greater insight into population-specific effects. Third, the overall methodological quality of the included studies was moderate to high. In addition, rigorous sensitivity analyses and publication bias assessments were performed to ensure the robustness and reliability of the findings. However, the study has some limitations that should be acknowledged. The most prominent is the considerable heterogeneity observed across many outcomes, such as BMI, TC, and FPG, which limits the ability to draw definitive conclusions. This variability likely arises from differences in baseline vitamin D status, participant characteristics, underlying metabolic conditions, intervention dosages, and study durations. Although subgroup analyses addressed some of these factors, residual confounding cannot be excluded. Another limitation is that most included trials focused on individuals with vitamin D deficiency, restricting the generalizability of results to vitamin D–sufficient populations. Furthermore, while most studies used cholecalciferol (vitamin D₃), variation in vitamin D forms and delivery methods may have influenced the outcomes but could not be analyzed separately due to limited data. Lastly, despite a broad search strategy, the possibility of missed unpublished or non-English language trials remains, potentially introducing publication bias.

Conclusion

In summary, this meta-analysis found that vitamin D supplementation did not produce significant improvements in lipid profile, glycemic markers, or anthropometric indices in adults with various metabolic conditions. While minor subgroup-specific effects were observed such as increased TC and LDL-C in males and modest BMI reduction after publication bias adjustment, these findings were not consistent across the broader population. Future research should focus on identifying population groups that may benefit most from vitamin D supplementation and elucidate the mechanistic pathways underlying its metabolic effects.

Author contributions

LG, LH and YW came up with the idea for the study and designed the search strategy. YW and ShG searched databases, while NK, and LG were responsible for selecting the studies and extracting data, carried out statistical analyses. LG, LH and YW wrote the initial draft of the manuscript. The manuscript was then critically revised by YW, and LH, and all authors approved the final version.

Funding

This study was supported by grants from Natural Science Foundation of Zhejiang Province (LY24H200001), Health Bureau Foundation of Zhejiang Province (2022KY520, 2022KY610 and 2024KY1104), Chinese Medicine Science and Technology Program of Zhejiang Province (2023ZL268) and Basic Scientific Research Funds of Department of Education of Zhejiang Province (KYQN202110).

Availability of data and materials

The original data used during the current study can be obtained by contacting the corresponding author.

Declarations

Ethics approval and consent to participate

Not applicable.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.