Effects of curcumin in patients with non-alcoholic fatty liver disease: A systematic review and meta-analysis

Abstract

Background:

Curcumin is an anti-inflammatory that is proposed to have a positive impact on patients with non-alcoholic fatty liver disease (NAFLD). We aim to assess the effects of curcumin in patients with NAFLD.

Methods:

Clinical trials from PubMed, Scopus, the Web of Science, and Cochrane CENTRAL with variables alanine transferase, aspartate transaminase, alkaline phosphatase, glycated hemoglobin (HBA1c), BMI, waist circumference, total cholesterol, total glycerides, high-density lipoproteins, and low-density lipoproteins were included. Homogeneous and heterogeneous were analyzed under a fixed-effects model and the random-effects model, respectively.

Results:

Fourteen clinical trials found that curcumin has no statistically significant effect on alanine transferase (MD = −2.20 [−6.03, 1.63], p = 0.26], aspartate transaminase (MD = 1.37 [−4.56, 1.81], p = 0.4), alkaline phosphatase (MD = 3.06 [−15.85, 9.73], p = 0.64), glycated hemoglobin (HBA1c), (MD = −0.06 [−0.13, 0.02], p = 0.16], and BMI (MD = 0.04 [−0.38, 0.46], p = 0.86). Curcumin reduced the waist circumference (MD = −4.87 [−8.50, −1.25], p = 0.008). Lipid profile parameters were not significant, except the total glycerides (MD = −13.22 [−24.19, −2.24], p = 0.02).

Conclusions:

Curcumin significantly reduces total glycerides and waist circumference in NAFLD.

Lay summary: Curcumin is the main active component of turmeric. It has been thought to have anti-inflammatory, antioxidant, anti-diabetic, anti-hyperlipidemia, immune-modulatory, reno-protective, anti-cancer, hepato-protective, antimicrobial, and anti-fibrotic properties. Our study focuses on the effects of curcumin on liver enzymes and metabolic factors in fatty liver disease.

Introduction

Non-alcoholic fatty liver disease (NAFLD) has risen significantly over the past 30 years, becoming the most prevalent chronic liver disease worldwide and the second cause of liver transplantation worldwide (1–3). NAFLD is highly prevalent across nearly all countries, especially in developed countries, 25%–30%, while in developing countries, 6%–35% (4,5). Not only this, but its prevalence is constantly increasing as a result of the rising epidemiology of diabetes and obesity (6). NAFLD is characterized by excessive lipid deposition in liver cells due to causes apart from alcohol drinking, including drugs and hypothyroidism (7,8). The pathogenesis of NAFLD is not yet fully understood; however, its pathogenesis could be defined by the complex interaction between multiple factors, such as genetic, hormonal, and nutritional factors (9–11).

Moreover, it is associated with other liver disorders such as cirrhosis and hepatocellular carcinoma (HCC), in addition to cardiovascular disease and diabetes (12,13). The management for NAFLD in the present involves diet and lifestyle modifications for achieving weight loss as well as management of existing metabolic risk factors and medications (14–16). Unfortunately, a high percentage of patients fail to be strict with these lifestyle modifications (17,18). Thus there is an urge to develop a safe and effective drug for NAFLD.

On the other hand, over the past years, developing evidence showed that investigators are working to find effective auxiliary treatments as natural alternatives to chemical drugs. Curcumin is one of these natural alternatives that has been taken into consideration concerning this matter (19,20). Curcumin has a polyphenol structure and is the main active component of turmeric (21). Not only is curcumin vastly used in food as a spice, but it also has many therapeutic benefits, as it acts as an anti-inflammatory, antioxidant, anti-diabetic, anti-hyperlipidemia, immune-modulatory, reno-protective, anti-cancer, hepato-protective, hypoglycemic, antimicrobial, and anti-fibrotic (20,22–24). Curcumin treats NAFLD through a series of mechanisms. Firstly, it clears the liver cells from fat and decreases the synthesis of triglycerides by inhibiting the enzyme HMG COA reductase (25,26). Secondly, it increases the activation of cholesterol-7alpha-hydroxylase as well as reduces the absorption of cholesterol from the intestines (26).

The hepato-protective activity of curcumin for the treatment of NAFLD has been evaluated in multiple clinical trials. Nonetheless, the results are not decisive. While some studies were in favour of the beneficial effects of curcumin (3,27–32), others did not show any significant effects (29,33–37).

Therefore, it was important to conduct this review to help to establish a clearer picture of the effects of curcumin on NAFLD.

Methods

This meta-analysis was performed following the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) (38) and the guidelines reported in the Cochrane Handbook for Systematic Reviews of Interventions (39).

Literature search

In our search, we used four databases: Web of Science, SCOPUS, Cochrane CENTRAL, and PubMed, from inception until May 2023 (Online Appendix 1).

Eligibility criteria

All the studies involved in our analysis should have the following criteria (PICO):

• (I)Population: Patients with nonalcoholic fatty liver disease
• (II)Intervention: Curcumin regardless of the route of administration
• (III)Comparator: Placebo
• (IV)Outcomes: alanine aminotransferase (ALT) (IU/L), aspartate aminotransferase (AST) (IU/L), and alkaline phosphatase (ALP) (U/L) as primary outcomes; the secondary outcomes were gamma-glutamyl transferase (GGT) (IU/L), HbA1c, BMI (kg/m2), LDL (mg/dL), HDL (mg/dL), total cholesterol (TC) (mg/dL), and triglyceride (TG) (mg/dL)
• (V)Study design: We used only randomized clinical trials (RCTs); the exclusion criteria were (1) non-RCTs, (2) studies that had no data or measures for our outcomes, and (3) the trials that we could not find their full text

Screening of results

We exported our results of the search using Endnote X8.0.1 (Build 1044), with the exclusion of duplicates automatically by computer. After that, in two steps, we manually screened the studies: first, the abstract and title screening, then the full-text screening of the included studies.

Data extraction and analysis

After the step of screening, we got the data from the selected studies and categorized the data into two main groups: (1) baseline and demographic data of patients in the included studies, which was age, sample size, gender, BMI, level of ALP, level of ALT, level of AST, level of GGT, HbA1c, TG, and TC; and (2) data for analysis, including outcome values of ALP, ALT, AST, GGT, TC, TG, BMI, and HbA1c. In addition to the previous two categories, the data about the seven domains assessing the risk of bias were extracted according to Cochrane's risk of bias (40).

Data analysis

We performed our analysis with Review Manager Software (RevMan 5.4.1). We analyzed continuous outcomes using mean difference (MD) and standard deviations (SD), and relative 95% CIs under the inverse variance method. Two tests prove the inconsistency among studies, the I-square test (I²) and the p-value of the Chi-square test. The outcomes that have I² > 50%, p < 0.1 were known to be heterogeneous, while outcomes that have I² < 50%, p > 0.1 were known to be homogeneous, according to the Cochrane Handbook. We analyzed the homogenous data by a fixed-effects model, while the heterogeneous outcomes were analyzed by the random-effects model.

Quality assessment

We followed the Grading of Recommendations, Development, and Evaluation (GRADE) guidelines (41). Only the controlled trials were included, and we evaluated the quality of this systematic review and meta-analysis by the Grading of Recommendations Assessment observational evidence. According to the Cochrane risk of bias (ROB) tool, we assessed the ROB for the included studies. The tool depends on seven domains for assessment of the risk of bias: (1) randomization, (2) blinding allocation of the patients into each group, (3) blinding of patients and personnel, (4) attrition bias, (5) selection bias, (6) awareness of the outcome assessor, (7) other bias. Then we extract the total risk of bias for the studies.

Results

Summary of included studies

Figure 1 shows a PRISMA flow diagram of our literature search. In our study, we did an analysis of 975 patients from 16 studies (3,21,27,30–32,42–51); 494 patients received curcumin, and 481 patients were allocated to the control group. The mean age of the participants in the treatment group was 39.8 ± 9.1 years, while that of the control group was 44.5 ± 10. Tables 1–3 show a summary of the participants, their demographic data, BMI, ALP, ALT, AST, TC, TG, HDL, LDL, dose, and duration.

Figure 1:

PRISMA flow diagram of our literature search

Table 1:

A detailed summary of the included participants’ demographic data

Study ID	Sample size		Age				Gender				Alanine aminotransferase (ALT)				Aspartate aminotransferase (AST)
			Curcumin		Control		Curcumin		Control		Curcumin		Control		Curcumin		Control
	Curcumin	Placebo	MEAN	SD	MEAN	SD	Males	Total	Males	Total	MEAN	SD	MEAN	SD	MEAN	SD	MEAN	SD
Chashminam2019	25	20	46.56	11.25	37.75	14.4	13	25	14	20	IU/L 50..08	36.35	40.6	20.44	IU/L35.16	19.5	24.6	11.14
Mirhafez2021	35	37	45	11.1	43.1	11.6	20	35	22	37	nr	nr	nr	nr	nr	nr	nr	nr
Cicero2019 (56D)	40	40	54	3	53	5	18	40	19	40	21	6	23	5	23	7	24	7
Hariri2020	23	22	40.95	12.24	40.06	13.69	14	23	12	47	IU/L45.59	28.79	38.91	20.86	32.01	16.43	26.66	10.65
Moradi2020	11	11	nr	nr	nr	nr	nr	nr	nr	nr	nr	nr	nr	nr	nr	nr	nr	nr
Panahi2019	35	35	46.63	13.07	47.51	14.49	20	35	19	35	IU/L50.23	27.21	38.88	25.26	IU/L34.26	21.48	31.82	13.96
Saberi-Karimian2020	27	28	nr	nr	nr	nr	nr	nr	nr	nr	IU/L30	15.12	30.4	16.14	IU/L24.7	6.9	24.55	8.41
Jazayeri-Tehran2019	42	42	41.8	5.6	42.5	6.2	23	42	23	42	IU/L 42.8	11.6	42.1	8.2	IU/L 28.4	6.7	27.6	7.8
Mirhafez2019	33	32	44.8	11.14	40.7	11.83	18	33	19	32	47.66	35.2	40.07	19.82	32.45	18.6	25.18	11.11
Panahi2016	50	52	44.98	12.59	47.21	10.29	30	50	36	52	nr	nr	nr	nr	nr	nr	nr	nr
Panahi2017	50	52	44.98	12.59	47.21	10.29	22	44	27	43	IU/L 35.46	22.97	36.81	24.32	27.63	11.35	27.44	10.01
Rahmani2016	40	40	46.37	11.57	48.95	9.78	19	40	19	40	39.07	19.79	30.35	13.97	28.88	10.6	32.05	17.64
Sadati2018	25	21	46.64	11.7	45.33	11.47	11	25	12	21	nr	nr	nr	nr	nr	nr	nr	nr
Sadati2019	27	23	46.19	11.5	45.13	10.9	13	27	14	23	IU/L 26.05	15.09	28.02	13.06	IU/L 17.65	9.95	16.23	5.66
Sadati2019 (2)	25	25	46.19	11.5	45.1	10.9	13	25	14	25	IU/L 26.54	15.46	28.02	13.06	IU/L 17.78	9.56	16.23	5.66
Ghaffari 2018	21	21	42.57	6.9	40.4	9.3	11	21	8	21	23	27.2	23.6	30.4	24	11.5	24.3	13.6

Table 3:

Detailed summary of the included TC, TG, HDL, and LDL

Study ID	Total cholesterol (TC)				Triglyceride (TG)				HDL				LDL
	Curcumin		Control		Curcumin		Control		Curcumin		Control		Curcumin		Control
	MEAN	SD	MEAN	SD	MEAN	SD	MEAN	SD	MEAN	SD	MEAN	SD	MEAN	SD	MEAN	SD
Chashminam2019	211.8	37.3	196.7	36.8	151.84	77.3	145.05	70.6	45.84	9.05	46.12	12.48	117.38	30.2	108.17	25.76
Mirhafez2021	202.8	37.2	194	36.2	142	50.6	136.5	44.7	44.8	9.6	45.6	10.6	111.2	29.5	105.6	25.2
Cicero2019 (56D)	193	15	195	16	185	21	181	22	40	3	41	4	116	11	118	12
Hariri2020	nr	nr	nr	nr	nr	nr	nr	nr	nr	nr	nr	nr	nr	nr	nr	nr
Moradi2020	nr	nr	nr	nr	nr	nr	nr	nr	nr	nr	nr	nr	nr	nr	nr	nr
Panahi2019	196	40.5	180.44	33.6	152.33	59.26	132.42	63.15	46.09	14.38	44.29	11.65	113.12	36.8	104.99	31.8
Saberi-Karimian2020	44.96	48.33	190.14	44	156.04	65.85	147.37	85.267	43.11	10.14	45	11.9	107.18	39.48	114.37	43.55
Jazayeri-Tehran2019	212.9	18.9	211.8	21.4	175.9	70.3	181.2	65.6	41.8	5.6	42.7	5.7	135.6	17.6	133	20.7
Mirhafez2019	204.21	37.68	183.64	29.14	130.46	87.27	139.19	58.51	41.43	13.86	45	10.55	138.4	40..39	108.71	24.14
Panahi2016	199.57	40.8	184.06	46.85	151.41	75.61	150.07	68.46	46.82	9.84	47.44	10.6	1130.6	33.45	108.94	46.22
Panahi2017	nr	nr	nr	nr	nr	nr	nr	nr	nr	nr	nr	nr	nr	nr	nr	nr
Rahmani2016	198.59	41.76	187.78	32.95	199.68	91.46	160.2	61.94	44.26	11.83	42.62	6.67	107.06	31.36	115.57	22.3
Sadati2018	nr	nr	nr	nr	nr	nr	nr	nr	nr	nr	nr	nr	nr	nr	nr	nr
Sadati2019	183.59	30.59	192.88	33.13	164.55	75.6	149.47	78.07	39.9	0.27	40.13	0.21	110.77	28.71	122.85	34.616
Sadati2019 (2)	nr	nr	nr	nr	nr	nr	nr	nr	nr	nr	nr	nr	nr	nr	nr	nr
Ghaffari 2018	nr	nr	nr	nr	nr	nr	nr	nr	nr	nr	nr	nr	nr	nr	nr	nr

Table 2:

Detailed summary of the included ALP, BMI, WC, dose, and duration

Study ID	Alkaline phosphatase (AP)				BMI				Waist circumference (cm)				Dose, mg	Duration
	Curcumin		Control		Curcumin		Control		Curcumin		Control
	MEAN	SD	MEAN	SD	MEAN	SD	MEAN	SD	MEAN	SD	MEAN	SD
Chashminam2019	IU/L 206	72.25	184.95	52.59	30.03	3.5	28.08	4.2	nr	nr	nr	nr	250	8
Mirhafez2021	nr	nr	nr	nr	30.8	5.1	29.2	4.2	nr	nr	nr	nr	250	8
Cicero2019 (56D)	28	9	27	9	27.1	1.8	26.9	1.9	94	7	93	8	200	8
Hariri2020	nr	nr	nr	nr	30.59	5.91	28.87	3.59	102.83	14.72	98.41	8.9	250	8
Moradi2020	IU/L376.36	88.45	320.09	81.99	27.48	1.26	27.03	0.56	nr	nr	nr	nr	80	12
Panahi2019	IU/L196.167	38.5	178.5	55.2	nr	nr	nr	nr	nr	nr	nr	nr	500	12
Saberi-Karimian2020	nr	nr	nr	nr	83.25	14.41	85.09	12.19	103.42	9.52	105.73	9.73	500	8
Jazayeri-Tehran2019	nr	nr	nr	nr	30.67	2.14	30.75	2.35	105.4	6.2	103.8	6.7	80	12
Mirhafez2019	nr	nr	nr	nr	30.06	5.76	27.72	5.97		nr	nr	nr	250	8
Panahi2016	nr	nr	nr	nr	28.97	3.42	29.07	3.48	97.07	11.04	101.9	11.43	1,000	8
Panahi2017	IU/L 172.52	40.65	170.77	40.8	28.79	3.42	29.07	3.48	97.07	11.04	101.9	11.43	1,000	8
Rahmani2016	nr	nr	nr	nr	30.84	4.45	31.35	5.67	nr	nr	nr	nr	500	8
Sadati2018	nr	nr	nr	nr	32.57	4.64	32.19	5.22	102.32	9.08	101.43	8.56	1,500	12
Sadati2019	nr	nr	nr	nr	32.3	4.55	32.38	5.02	102.19	8.78	103.28	8.83	1,500	12
Sadati2019 (2)	nr	nr	nr	nr	32.3	4.55	32.38	5.02	102.19	8.78	103.28	8.83	1,500	12
Ghaffari 2018	nr	nr	nr	nr	31.8	4.9	32.9	4.81	nr	nr	nr	nr	3,000	12

Results of risk of bias assessment

Overall, the included studies had a low risk of bias, as assessed using the Cochrane tool. All studies were at low risk of bias for randomization. For allocation concealment, 10 studies (3,31,32,42–48) reported adequate allocation concealment and were therefore at low risk of bias. Five studies (27,30,49–51) reported no allocation concealment, and one study (21) reported inadequate data. Most of the included studies were blinded, with five studies (27,30,48,50,51) not blinded by participants and personnel, and two studies (21,49) reporting inadequate data. Eight studies (3,32,42–47) were at low risk of blinding of outcome assessment. Six studies (27,30,48–51) were not blinded to outcome assessment, and two studies (21,31) reported inadequate data. The other domains of the Cochrane tool were at low risk of bias, except for two studies (31,47) that reported inadequate data for selective reporting, and three studies (27,32,44) that reported inadequate data for incomplete outcomes. A summary of the risk of bias of the included studies is presented in Figure 2.

Figure 2:

Risk of bias assessment

Analysis of outcomes

ALT

The ALT outcome was reported by 14 studies (3,21,47–49,51,30–32,42–46). We divided the trials into six subgroups according to the dose of Curcumin. The first subgroup (200 mg or less) shows that there was a significant decrease in ALT level in the curcumin group (MD = −4.82 [−8.39, −1.26]), (p < 0.01). The analysis was heterogeneous (p = 0.08); I² = 60%. The second subgroup (250 mg) shows that there was no significant difference between both groups (MD = −1.99 [−4.37, 0.38]), (p = 0.1). The analysis was homogeneous (p = 0.47); I² = 0%. The third subgroup (500 mg) shows that there was no significant difference between both groups (MD = 1.24 [−3.38, 5.87]), (p = 0.6). The analysis was homogeneous (p = 0.35); I² = 4%. The fourth subgroup (1,000 mg) shows a decrease in the ALT level in the Curcumin group (MD = −16.48 [−24.59, −8.37]) (p < .01). The fifth subgroup (1,500 mg) shows that there was no significant difference between both groups (MD = −0.58 [−4.16, 2.99]), (p = 0.75). The analysis was homogeneous (p = 0.9); I² = 0%. The sixth subgroup (3,000 mg) shows that there was no significant difference between both groups (MD = −3.00 [−10.29, 4.29]), (p = 0.42), Figure 3.

Figure 3:

Forest plot for the analysis of ALT outcome

AST

The AST outcome was reported by 14 studies (3,21,47–49,51,30–32,42–46). We divided the trials into six subgroups according to the dose of curcumin. The first subgroup (200 mg or less) shows that there was no significant difference between both groups (MD = −2.62 [−6.07, 0.82]), (p = 0.14). The analysis was heterogeneous (p = 0.03); I² = 70%. The second subgroup (250 mg) shows that there was no significant difference between both groups (MD = 0.32 [−2.65, 3.29]), (p = 0.83). The analysis was homogeneous (p = 0.54); I² = 0%. The third subgroup (500 mg) shows that there was no significant difference between both groups (MD = −3.05 [−9.62, 3.51]), (p = 0.36). The analysis was heterogeneous (p = 0.009); I² = 79%. The fourth subgroup (1,000 mg) shows a decrease in the AST level in the curcumin group (MD = −10.55 [−14.85, −6.25]), (p < 0.01). The fifth subgroup (1,500 mg) shows that there was no significant difference between both groups (MD = 2.07 [−0.29, 4.43]), (p = 0.09). The analysis was homogeneous (p = 0.94); I² = 0%. The sixth subgroup (3,000 mg) shows that there was no significant difference between both groups (MD = 0.10 [−4.35, 4.55]), (p = 0.96), Figure 4.

Figure 4:

Forest plot for the analysis of AST outcome

ALP

Five studies (3,30,47–49) reported the ALP outcome. The combined effect estimate revealed no significant difference between curcumin and placebo (MD = −1.42 [−12.81, 9.96]), (p = 0.82). Pooled analysis was homogeneous (p = 0.13); I² = 44%, Figure 5.

Figure 5:

Forest plot for the analysis of ALP outcome

HBA1c

Four studies (27,32,49,50) reported this outcome. The overall MD did not favour any of the two groups (MD = −0.37 [−0.86, 0.13]), (p = 0.14), and pooled analysis was heterogeneous (p = 0.001) (I² = 93%); Figure 6A–B. We solved the heterogeneity by the exclusion of Rahmani et al. (21). The overall MD did not favour any of the two groups (MD = −0.01 [−0.17, 0.14]), (p = 0.86), and pooled analysis was homogeneous (p = 0.18) (I² = 42%).

Figure 6A:

Forest plot for the analysis of HbA1c outcome

BMI

The BMI outcome was reported by 13 studies (3,21,48,50,51,30–32,43–47). The analysis did not reveal any significant difference between both groups (MD = −0.07 [−0.48, 0.35]), (p = 0.76). Pooled analysis was homogeneous (p = 0.73) (I² = 0%), Figure 7.

Figure 6B:

Forest plot for the analysis of HbA1c outcome

Figure 7:

Forest plot for the analysis of BMI outcome

WC

Six studies (27,32,43,44,46,51) reported this outcome. We divided the trials into two subgroups according to the dose of curcumin. The first subgroup (500 mg or less) shows that there was no significant difference between both groups (MD = −1.68 [−3.49, 0.13]), (p = 0.07). The analysis was homogeneous (p = 0.34); I² = 10%. The second subgroup (1 g or more) revealed that curcumin significantly reduces waist circumference compared with placebo (MD = −4.87 [−8.50, −1.25], p = 0.008). Pooled analysis was homogeneous (p = 0.32) (I² = 0%), Figure S1.

TC

Ten studies (3,21,27,31,32,42–44,47,49) reported this outcome. We divided the trials into five subgroups according to the dose of curcumin. The first subgroup (200 mg or less) shows a decrease in the level of TC in the curcumin group (MD = −6.37 [−11.96, −0.79]), (p = 0.03). The analysis was homogeneous (p = 0.31); I² = 3%. The second subgroup (250 mg) shows that there was no significant difference between both groups (MD = 10.44 [−0.39, 21.26]), (p = 0.06). The analysis was homogeneous (p = 0.88); I² = 0%. The third subgroup (500 mg) shows that there was no significant difference between both groups (MD = −2.62 [−23.71, 18.47]), (p = 0.81). The analysis was heterogeneous (p = 0.02); I² = 74%. The fourth subgroup (1,000 mg) shows a decrease in the TC level in the curcumin group (MD = −41.08 [−62.59, −19.57]), (p < 0.01). The fifth subgroup (1,500 mg) shows that there was no significant difference between both groups (MD = −17.46 [−37.03, 2.11]), (p = 0.08), Figure S2.

TG

Ten studies (3,21,27,31,32,42–44,47,49) reported this outcome. The overall MD did not favour any of the two groups (MD = −6.60 [−18.48, 5.29]), (p = 0.28); pooled analysis was heterogeneous (p = 0.05) (I² = 47%), Figures S3A–B. We solved the heterogeneity by the exclusion of Mirhafez et al. (31). The overall MD did not favour any of the two groups (MD = −9.86 [−21.15, 1.44]), (p = 0.09), and pooled analysis was homogeneous (p = 0.12) (I² = 38%).

HDL

Ten studies (3,21,27,31,32,42–44,47,49) reported this outcome. The overall MD did not favour any of the two groups (MD = 0.14 [−2.23, 2.51]), (p = 0.91); pooled analysis was heterogeneous (p = 0.001) (I² = 76%), Figures S4A–B. We solved the heterogeneity by the exclusion of Jazayeri-Tehran et al. (32). The overall MD did not favour any of the two groups (MD = −0.45 [−1.95, 1.05]), (p = 0.56), and pooled analysis was homogeneous (p = 0.22) (I² = 25%).

LDL

Ten studies (3,21,27,31,32,42–44,47,49) reported this outcome. We divided the trials into five subgroups according to the dose of curcumin. The first subgroup (200 mg or less) shows a decrease in the level of LDL in the curcumin group (MD = −6.80 [−12.25, −1.35]), (p = 0.01). The analysis was homogeneous (p = 0.25); I² = 24%. The second subgroup (250 mg) shows a decrease in the level of LDL in the curcumin group (MD = 11.45 [1.53, 21.38]) (p = 0.02). The analysis was homogeneous (p = 0.97); I² = 0%. The third subgroup (500 mg) shows that there was no significant difference between both groups (MD = −5.32 [−35.09, 24.45]), (p = 0.73). The analysis was heterogeneous (p = 0.001); I² = 92%. The fourth subgroup (1,000 mg) shows a decrease in the LDL level in the curcumin group (MD = −29.60 [−44.78, −14.42]), (p < .01). The fifth subgroup (1,500 mg) shows that there was no significant difference between both groups (MD = −10.80 [−29.31, 7.71]), (p = 0.25), Figure S5.

Discussion

In this meta-analysis, we included 10 clinical trials with 2,000 patients divided into curcumin and placebo subgroups. We found no significant difference between the curcumin and placebo subgroups regarding the values of ALT, AST, ALP, HbA1c, BMI, total cholesterol, HDL, and LDL. On the other hand, curcumin significantly reduced total glycerides and waist circumference compared with the placebo subgroups.

Regarding the ALT, AST, and HDL levels, our results showed inconsistency with those of Jalali et al. (52). They conducted a meta-analysis on nine RCTs, evaluating the effect of curcumin supplementation on liver enzymes, blood lipid profiles, glycemic indices, and anthropometric parameters. They found that curcumin significantly reduced the ALT and AST levels compared to the placebo group in seven trials (p = 0.049, p = 0.032, respectively). Also, it reduced the serum LDL concentration but not the TG or HDL levels. Regarding the glycemic indices, FBS and HOMA-IR were reduced greatly in the curcumin group. But HbA1c showed no significant change, which shows consistency with our results. As for anthropometric parameters, waist circumference decreased significantly in the curcumin group but not the body weight or the BMI.

Besides, they performed a meta-regression analysis finding that the outcomes did not affect by the dosage or the duration of curcumin supplementations. On the other hand, they found curcumin significantly reduced the liver enzymes in specific subgroups in their subgroup analysis.

Curcumin has a variety of benefits, including anti-inflammatory, antioxidant, anti-diabetic, anti-hyperlipidemia, immune-modulatory, reno-protective, anti-cancer, hepato-protective, hypoglycemic, antimicrobial, and anti-fibrotic activities (20,22–24). In NAFLD, oxidative stress and reactive oxygen species (ROS) play a significant role in the pathogenesis of NAFLD, hence the importance of anti-oxidant usage (53). It controls the inflammatory process by targeting certain molecules. This includes growth factors, transcription factors, cytokines, and enzymes in addition to regulation of proliferation and apoptosis (54). Tumour necrosis factor α (TNF-α) represents the major mediator of the inflammatory process, which is regulated by (NF)-κB. Curcumin suppresses (NF)-κB activation by various stimuli, resulting in the reduction of inflammatory and oxidative stress (54,55). Besides, curcumin inhibits the expression of growth-promoting genes. This includes, for example, the expression of TNF-α), interleukin 1β (IL-1β), and interleukin 6 (IL-6) (56,57).

Panahi et al. in 2016 (58) studied the serum cytokine concentrations in metabolic syndrome patients. In their study, curcumin significantly reduced the concentration of pro-inflammatory cytokines (TNF-a, IL-6, TGF-b, and MCP-1) that play a major role in the inflammatory process of the disease, compared to placebo (p < 0.001). In addition, Soni and Kuttan (59) found that in 10 healthy individuals, curcumin lowered the total serum cholesterol and lipid peroxide in addition to increasing serum HDL.

In 2017 in their RCT, Panahi et al. (30) found that short-term curcumin supplementation has beneficial effects on NAFLD patients. They used a 1,000 mg/day dosage for 8 weeks in combination with lifestyle modifications. It reduced biochemical markers levels; the transaminases ALT and AST compared to baseline readings in the curcumin group compared to the placebo group had elevated levels. Regarding ultrasonographic findings, 75% of the curcumin group showed improvement compared to 4.7% of the placebo group. These results show inconsistency with our results. However, this may be due to the small sample of the trial compared to the combined results of our meta-analysis.

Mirhafez et al. (31) conducted a double-blinded RCT about curcumin's different effects; their results show consistency with our results. Regarding BMI, weight, and FBS, curcumin and placebo groups show no significant difference between both groups.

As for the lipids profile (HDL and LDL), the results were better for curcumin compared with placebo in between-groups analysis, although the difference was not significant compared to the baseline levels. Also, liver enzymes (ALT and AST) show no significant difference between both groups. On the other hand, the curcumin group showed a significant reduction in leptin levels and significant elevation in adiponectin levels compared to the placebo group. Besides, it reduced the leptin serum levels and leptin:adiponectin ratio compared to placebo.

The included trials show certain limitations in interpreting the results. First was the small sample size of clinical trials. Second, the duration of the studies was too short to establish a valuable conclusion. Third was the risk of bias results. Fourth, most of the studies were conducted in limited settings, which cannot give a generalizable result. Finally, there are not enough data about the dosage or content of the agent used and no available data about the severity of the NAFLD state.

But our meta-analysis showed different strength points; we included only the randomized controlled trials. We solved any heterogenicity of the results using sensitivity analysis. We conducted this meta-analysis in adherence to the Cochrane Handbook.

Conclusion

Curcumin showed no significant reduction of ALT, AST, ALP, HbA1c, BMI, total cholesterol, HDL, and LDL in NAFLD compared to placebo. However, it significantly reduced total glycerides and waist circumference compared with the placebo subgroup. It may become a promising agent in NAFLD treatment, but it needs further studies.

Contributions:

Conceptualization, A Malik; Data curation, M Malik; Formal analysis, A Malik; Writing—original draft, M Malik; Writing—­review & editing, M Malik.

Ethics Approval:

N/A

Informed Consent:

N/A

Registry and the Registration No. of the Study/Trial:

N/A

Data Accessibility:

Data will be available to any researcher who contacts the corresponding author.

Funding:

N/A

Disclosures:

The authors have no conflicts of interest to disclose.

Peer Review:

This manuscript was peer reviewed.

Animal Studies:

N/A