An updated systematic review and meta-analysis: correlation between circulating plasminogen activator inhibitor-1 levels and metabolic dysfunction-associated steatotic liver disease

Abstract

Background Metabolic dysfunction-associated steatotic liver disease (MASLD)—formerly known as non-alcoholic fatty liver disease—is the leading cause of chronic liver disease globally. However, early diagnosis and specific treatment have not yet been achieved. Plasminogen activator inhibitor-1 (PAI-1) is closely related to MASLD and may be an important indicator for its diagnosis and treatment. This updated meta-analysis aimed to comprehensively explore the relationship between circulating PAI-1 levels and MASLD. Methods Article retrieval was carried out in seven provenances (PubMed, Cochrane Library, EMBASE, CNKI, WANFANG, Clinical Trials Database, and Grey Literature Database) using free text words and MeSH terms. The meta-analysis was performed using RevMan 5.3 and Stata 12. Outcomes were presented as standardised mean difference (SMD) with 95% confidence intervals (CI). Results A total of 23 studies were included in the meta-analysis. The levels of circulating PAI-1 in patients with MASLD were significantly higher than those in healthy controls (SMD = 1.44, 95% CI [1.06, 1.83]). Subgroup and meta-regression analyses based on key factors, such as area, age, and BMI, were performed to explore the sources of heterogeneity. In the BMI subgroup analysis, the participants were divided into two groups. While the subgroup results were inconclusive, the meta-regression suggested that the area (i.e. the geographic region of the study population) may be a major contributor to heterogeneity (p < 0.001). Conclusions Patients with MASLD have significantly high circulating PAI-1 levels; however, this correlation may vary in different regions, thereby providing a reference for further research on MASLD.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12876-025-04148-8.

Background

Non-alcoholic fatty liver disease (NAFLD) is a fatty degeneration of the liver, which is usually classified into benign non-alcoholic fatty liver (NAFL) and non-alcoholic steatohepatitis (NASH), with or without liver fibrosis [1, 2]. Metabolic dysfunction-associated steatotic liver disease (MASLD) [3] has recently been proposed as a new nomenclature to replace non-alcoholic fatty liver disease (NAFLD), emphasising its strong association with obesity, type 2 diabetes mellitus (T2DM), and metabolic syndrome (MetS) [4–6]. MASLD closely aligns with another previously introduced term, metabolic dysfunction-associated fatty liver disease (MAFLD), both of which highlight metabolic dysfunction as a key driver of liver disease. Currently, the quantitative assessment of MASLD is performed based on hepatic inflammatory activity and fibrosis [7]. Research suggests that it already affects one in four people worldwide. The prevalence of MASLD, the most widespread chronic liver disease, increases annually [8]. Furthermore, this disease may progress to cirrhosis and hepatocellular carcinoma (HCC); however, the lack of specific treatments and early diagnosis has not been addressed. Despite being the diagnostic gold standard, liver biopsy faces limitations in clinical adoption owing to its invasive nature, sampling variability (reflecting regional pathological differences within the liver), and substantial economic burden [9, 10]. These constraints frequently result in delayed diagnoses and missed windows for early-stage interventions. Therefore, it is imperative to explore noninvasive biomarkers for the early diagnosis and assessment of MASLD.

Plasminogen activator inhibitor-1 (PAI-1), a serine protease inhibitor, is the most important inhibitor of tissue plasminogen and urokinase-type plasminogen activators. PAI-1 was first observed in the conditioned medium of human endothelial cells. Multiple factors regulate PAI-1 levels [11, 12], which are statistically associated with cardiovascular diseases, MetS, organ fibrosis, obesity, and T2DM [13]. PAI-1 activation inhibits fibrinolysis resulting in enhanced fibrinogenesis [14]. Increased fibrin levels promote the progression of liver disease and fibrosis. Animal studies suggest that reducing PAI-1 production significantly reverses hepatic steatosis and lipid accumulation [15].

A correlation between elevated circulating PAI-1 levels and MASLD development in humans has been reported, suggesting that PAI-1 is a promising pharmacological target for the treatment of MASLD. However, the consistency and relevance of the relationship between circulating PAI-1 levels and MASLD requires further discussion because of variations in sample sources. The purpose of this updated systematic review and meta-analysis was to comprehensively investigate the relationship between PAI-1 and MASLD and to identify more evidence for PAI-1 as a potential biomarker and therapeutic target for MASLD.

Methods

This systematic review strictly complied with the Cochrane Handbook for Systematic Review of Interventions [16] and was conducted in line with the required entries of the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) statement [17] (Supplemental file 1). The registration ID of the protocol in PROSPERO is CRD 420,223,312,285 (Supplemental file 2).

Article retrieval

Two researchers (QCL and YQZ) individually accessed articles related to MASLD and PAI-1 from seven databases: PubMed, Cochrane Library, EMBASE, CNKI, WANFANG, Clinical Trials Database (clinicaltrials.gov), and Grey Literature Database (opengrey.eu). A combination of free-text words and Medical Subject Headings (MeSH) terms was used to form a retrieval formula. The specific retrieval strategies are presented in Supplemental file 3. The references of the retrieved articles were also critically scrutinised to obtain potentially suitable theses. If the complete text or associated data were unavailable, the corresponding author was contacted via email. The literature collection step covered each database from its inception to 27 December 2023 without language restrictions.

Article screening

Two reviewers (QCL and YTR) screened the articles in a double-blind manner. Titles and abstracts were first scrutinised according to our screening provisions. Final determinations were made after reviewing the full texts of the initially included studies. Any disagreement was resolved by a senior reviewer (SL).

Studies adhering to the following inclusion criteria were selected: (1) adult patients aged ≥ 18 years who were clinically diagnosed with MASLD (MAFLD, NAFLD, NAFL, NASH, or simple steatosis) [18]; (2) healthy individuals without fatty liver disease or metabolic illnesses as controls; (3) measurement of circulating PAI-1 levels in either the serum or plasma of case and control groups; and (4) case-control or cohort studies.

Articles containing any of the following were eliminated: (1) secondary fatty liver associated with other causes, such as excessive drinking and other liver diseases; (2) patients with hepatic fibrosis associated with MASLD; (3) PAI-1 levels that were not measured or were measured in liver tissue, not in serum or plasma; (4) articles for which sufficient data were not obtained even after contacting the authors; and (5) duplicate publications or studies conducted on the same cohort of patients.

Data extraction

Two researchers (QCL and YTR) extracted the data separately and reached a consensus on any issue through third-party discussions (SL). The following information was gathered: (1) basic information on the incorporated literature (first author’s name, publication date, country, and Newcastle–Ottawa Scale [NOS] score); (2) the conditions of the groups (group size, age, sex, body mass index [BMI], homoeostatic model assessment for insulin resistance [HOMA-IR], and underlying diseases); (3) diagnostic methods for fatty liver; (4) circulating PAI-1 levels in serum or plasma; and (5) methods of detecting circulating PAI-1 levels.

Quality assessment

Two reviewers (QCL and YTR) used the NOS, which consists of three aspects—selection, comparability, and exposure—to assess the quality of the literature [19]. A maximum of one star per entry in the selection and exposure sections, and a maximum of two stars in the comparability section, out of a total of nine stars, were assigned to each article. A high NOS score indicated the scientific rationality of the experimental design of the included studies. Meanwhile, the GRADE scale was used to verify the evidence certainty (https://gdt.gradepro.org).

Statistical analysis

Statistical analyses were performed using RevMan 5.3 and Stata 12 (Stata Corporation, College Station, TX, USA). Before analysis, the data was transformed to convert the median [interquartile range] to mean ± standard deviation (SD) [20, 21]. All results are represented using the standardised mean difference (SMD) and 95% confidence interval (CI). For heterogeneity, the I² test, Cochran’s Q-test, and Galbraith’s figure were used. A p-value above 0.1 represented low heterogeneity. Based on this, the fixed-effects model was selected for data calculation, and the random-effects model was used when heterogeneity was striking (P < 0.1) [22, 23]. To identify potential sources of heterogeneity, subgroup analyses were performed according to the area (i.e. geographic region of the study population), age, BMI, sex, HOMA-IR score, disease severity, NOS score, study unit, and MAFLD diagnostic methods. In parallel, meta-regression analyses were conducted using area, age, BMI, HOMA-IR, NOS score, and unit as covariates. In the subgroup analysis of BMI, participants were divided into two groups [24]: overweight (BMI ≥ 25 kg/m²) and obese (BMI ≥ 30 kg/m²). Publication bias was evaluated using funnel charts and Egger’s test. If publication bias existed, its effect on the results was verified using trimming and filling. Sensitivity analysis was performed to determine the stability of the results.

Result

Study selection

The search was performed according to the PRISMA guidelines, and seven databases were scanned by two investigators; consequently, 708 articles were obtained (Fig. 1). Of these, 581 were screened after duplicate articles were eliminated. Finally, 23 articles [25–47] were incorporated into the quantitative analysis (1,672 MASLD patients and 1,291 controls), covering six countries from four continents (i.e. Brazil, China, Egypt, Germany, Italy, and Turkey). All the included studies were case-control studies. The average age and BMI of MASLD patients were 29.8–58.8 years and 25.2–42.0 kg/m2, respectively. The average age and BMI of the control population were 27.0–56.5 years, and 18.8–32.8 kg/m2, respectively. Additional information on these 23 articles is presented in Table 1.

Fig. 1.

Flowchart of study inclusions and exclusions

Table 1.

Baseline characteristics of studies included in the meta-analysis

Author	Year	Country		MAFLD group			Control group			Detection Method
				Sex(F/M)	Age	BMI (km/m2)	Sex(F/M)	Age	BMI (km/m2)
Abdel-Razik et al. [23]	2009	Egypt		65/29	48.4 ± 14.3	27.3 ± 3.5	64/30	48 ± 13.6	23.8 ± 1.1	ELISA kit
Alvares-da-Silva et al. [24]	2014	Brazil		14/8	58.5 ± 6.5	31.7 ± 4.4	9/11	51.2 ± 9.3	25.1 ± 2.7	/
Bai et al. [25]	2014	China	NAFLD	3/10	45 ± 8	26.0 ± 1.9	0/13	49 ± 7	20.7 ± 2.4	ELISA kit
			NAFLD+hypertriglyceridemia	2/9	38 ± 8	27.3 ± 3.8
Bilgir et al. [26]	2014	Turkey		32/21	58.8 ± 5.9	32.0 ± 4.1	26/19	56.2 ± 6.1	32.8 ± 3	ELISA kit
Cao et al. [27]	2010	China		20/31	43.6 ± 10.6	27.8 ± 3.6	9/11	41.3 ± 10.3	22.6 ± 1.6	ELISA kit
Chang et al. [28]	2015	China		78/132	44.1 ± 12.7	26.5 ± 4	156/264	45.1 ± 16.9	23.3 ± 2.9	ELISA kit
Chen et al. [29]	2015	China	NAFLD	3/15	43.8 ± 7.4	26.0 ± 2.0	4/18	44.2 ± 8.1	20.6 ± 2.7	ELISA kit
			NAFLD+hypertriglyceridemia	3/17	42.5 ± 6.9	27.2 ± 3.6
Deng et al. [30]	2005	China		33/57	47.8 ± 8.5	25.8 ± 4.3	12/16	45.4 ± 10.7	/	chromogenic substrate assay
Deng et al. [31]	2005	China		27/43	47.2 ± 9.6	/	12/16	46.8 ± 10.6	/	chromogenic substrate assay
Fang et al. [32]	2010	China		24/41	55 ± 9	26.5 ± 2.7	24/41	55 ± 9	23.3 ± 3.1	ELISA kit
Jiang et al. [33]	2020	China		17/21	51.1 ± 12.2	26.4 ± 3	20/30	53.9 ± 10.8	22.2 ± 1.4	ELISA kit
Kargili et al. [34]	2010	Turkey		15/13	47.2 ± 12.1	25.2 ± 2	27/6	50 ± 9.9	27.1 ± 1	/
Li et al. [35]	2014	China		0/50	46 ± 4	26.5 ± 2.9	0/50	46 ± 4	22.3 ± 2.6	chromogenic substrate assay
Nier et al.[36]	2020	Germany	mild NAFLD	9/11	51.5 ± 2.4	29.2 ± 0.8	10/4	47.4 ± 1.2	23.3 ± 0.7	ELISA kit
			moderate NAFLD	11/20	47.9 ± 2.4	31.3 ± 0.8
			severe NAFLD	10/2	52.2 ± 4.2	35.7 ± 2.1
Targher et al. [37]	2008	Italy		0/45	47 ± 2	26.5 ± 2	0/45	47 ± 2	24.5 ± 1	chromogenic substrate assay
Thuy et al. [38]	2008	Germany		3/9	55 ± 4	27.8 ± 0.7	4/2	47 ± 7	22.5 ± 1.2	chromogenic substrate assay
Tian et al. [39]	2020	China		43/39	29.8 ± 6.8	42 ± 9.3	13/12	27 ± 2.4	18.8 ± 3.9	ELISA kit
Volynets et al. [40]	2012	Germany		11/9	41.9 ± 2.3	33.1 ± 1.9	7/3	39.6 ± 3.9	23.1 ± 1	chromogenic substrate assay
Wang et al. [41]	2023	China		55/211	42 ± 10.4	26.4 ± 2.7	79/20	40 ± 9.6	22.1 ± 2.8	ELISA kit
Xu et al. [42]	2011	China		11/37	55 ± 10.7	25.7 ± 2.8	11/40	56.5 ± 11	23.8 ± 3.6	chromogenic substrate assay
Yang et al. [43]	2023	China	mild NAFLD	/	49.6 ± 8.9	25.7 ± 2.3	/	50.9 ± 9.0	19.6 ± 3.8	ELISA kit
			moderate NAFLD	/	50.7 ± 8.8	27.9 ± 2.6
Yener et al. [44]	2007	Turkey		13/14	47.9 ± 7.1	30.5 ± 5.5	11/7	42.2 ± 14.5	24.3 ± 4	ELISA kit
Zhu et al. [45]	2015	China		29/57	53 ± 13.2	26.2 ± 3.3	29/57	53 ± 13.1	22.9 ± 2.9	Millipore’s MILLIPLEX MAP Human Adipokine Magnetic Bead Panel 1 kit

Abbreviations: M Male, F Female, BMI Body mass index, MAFLD Metabolic associated fatty liver disease, ELISA Enzyme‑linked immune‑sorbent assay

^aRanked by beginning letter of the first author

^bData are presented as the mean and SD or as the count, as appropriate.

^cCirculating PAI-1 data were not obtained in 25 patients due to the absence of available blood samples.

Quality assessment

Using NOS to assess the quality of 23 studies yielded an average score of 6.65 (Table 2). Seven articles (26, 32, 33, 38, 40, 42, 46) were assigned a score of 5 rating, and the rest were rated higher than 5. The GRADE system results indicated that the overall evidence for this study was of low certainty (see Supplemental file 4), since all the studies were observational; the substantial level of heterogeneity also contributed to this, thus lowering the reliability of the findings.

Table 2.

The Newcastle–Ottawa scale (NOS) score of included articles

Author	Year	Selection				Comparability	Exposure			Total	Average
		Adequate definition	Representativeness	Selection of Controls	Definition of Controls		Ascertainment of exposure	Same method	Non-response rate
Abdel-Razik et al. [23]	2009	1	1	0	1	2	1	1	1	8	6.65
Alvares-da-Silva et al. [24]	2014	1	0	0	1	0	1	1	1	5
Bai et al. [25]	2014	1	1	1	1	0	1	1	1	7
Bilgir et al.[26]	2014	1	1	0	1	0	1	1	1	6
Cao et al. [27]	2010	1	1	1	1	2	1	1	1	9
Chang et al. [28]	2015	1	1	1	1	2	1	1	1	9
Chen et al. [29]	2015	1	1	1	1	0	1	1	1	7
Deng et al. [30]	2005	1	0	0	1	0	1	1	1	5
Deng et al. [31]	2005	1	0	0	1	0	1	1	1	5
Fang et al. [32]	2010	1	1	1	1	0	1	1	1	7
Jiang et al. [33]	2020	1	1	1	1	2	1	1	1	9
Kargil. et al. [34]	2010	1	1	1	1	1	0	1	1	7
Li et al. [35]	2014	1	0	0	1	1	1	1	1	6
Nier et al. [36]	2020	1	0	0	1	0	1	1	1	5
Targher et al. [37]	2008	1	1	0	1	2	1	1	1	8
Thuy et al. [38]	2008	1	0	0	1	0	1	1	1	5
Tian et al. [39]	2020	1	1	1	1	0	1	1	1	7
Volynets et al. [40]	2012	1	0	0	1	0	1	1	1	5
Wang et al. [41]	2023	1	0	1	1	0	1	1	1	6
Xu et al. [42]	2011	1	0	1	1	0	1	1	1	6
Yang.et, al. [43]	2023	1	0	1	1	0	1	1	1	6
Yener et al. [44]	2007	1	0	0	1	0	1	1	1	5
Zhu et al. [45]	2015	1	1	1	1	2	1	1	1	9

Correlation between PAI-1 levels and MASLD

The results of the meta-analysis of the 23 studies are shown in Fig. 2. Owing to the high heterogeneity among the studies (p < 0.00001, I² = 94%), a random-effects model was used for this meta-analysis. In this model, circulating PAI-1 levels in patients with MASLD were significantly higher than those in their healthy counterparts, with an SMD of 1.44.

Fig. 2.

Forest plot of circulating PAI-1 levels between MASLD and the healthy control group (Random-Effects Model, SMD)

[1.06, 1.83]. Owing to differences in the methods of measuring PAI-1 levels, different studies used different units to represent circulating PAI-1 levels: 12, 8, and 3 articles expressed the levels in ng/mL, AU/mL, and no units (this was clarified with the authors through email), respectively. Marked heterogeneity was also observed in the Galbraith diagram (Fig. 3). Most, but not all, articles were within reasonable limits. Therefore, subgroup and meta-regression analyses were necessary to explore the sources of heterogeneity.

Fig. 3.

Galbraith Test result of circulating PAI-1 levels between MASLD and the healthy control group

Subgroup analysis

The results of subgroup analyses are shown in Table 3. Owing to the notable heterogeneity, the random effects model was selected in this subgroup analysis. Subgroup analysis was then performed based on the area, age, BMI, HOMA-IR, severity, NOS score, unit, and diagnostic methods for MASLD (a forest plot of subgroups is shown in Supplemental file 5). All subgroup analyses showed that the circulating PAI-1 levels were significantly higher in patients with MASLD than in healthy participants. There are some caveats to this subgroup analysis. First, subgroup analysis by area showed that the 95% CI of the two groups did not overlap (Asian: SMD = 0.87 [0.62, 1.12]; other population groups: SMD = 3.80 [2.13, 5.46]); thus, the circulating PAI-1 levels of patients with MASLD in other areas were significantly higher than those of Asian patients. Second, because of the difficulty in unifying units, this meta-analysis grouped studies using ng/mL as one group and studies using other units as another. Unfortunately, none of the subgroup analyses yielded meaningful results.

Table 3.

Subgroup analysis of the Circulating plasminogen activator inhibitor-1 levels in metabolic-associated fatty liver disease patients compared with controls

Subgroups	Classification	Studies	SMD (95%CI)	I2(%)	p for heterogeneity
Area	Asian	17	0.87 [0.62, 1.12]	86%	< 0.00001
	Others	6	3.80 [2.13, 5.46]	96%	< 0.00001
Total		23	1.44 [1.06, 1.73]	94%	< 0.00001
Age	age ≤ 50	16	1.41 [0.93, 1.88]	95%	< 0.00001
	age > 50	7	1.50 [0.84, 2.16]	90%	< 0.00001
Total		23	1.44 [1.06, 1.83]	94%	< 0.00001
BMI	25 kg/m2 ≤ BMI < 30 kg/m2	16	1.31 [0.90, 1.73]	94%	< 0.00001
	BMI ≥ 30 kg/m2	6	2.31 [1.02, 3.60]	96%	< 0.00001
Total		22	1.49 [1.09, 1.89]	95%	< 0.00001
Sex	Male percentage ≤ 0.5	6	1.81 [0.47, 3.15]	97%	< 0.00001
	Male percentage > 0.5	16	1.39 [1.01, 1.77]	92%	< 0.00001
		22	1.50 [1.10, 1.90]	94%	< 0.00001
HOMA-IR	HOMA-IR ≤ 5	8	1.45 [0.74, 2.16]	96%	< 0.00001
	HOMA-IR > 5	3	2.83 [0.10, 5.57]	97%	< 0.00001
Total		11	1.73 [1.06, 2.39]	96%	< 0.00001
severity	mild MAFLD group	4	1.49 [0.38, 2.59]	92%	< 0.00001
	moderate to severe MAFLD group	4	2.71 [1.01, 4.40]	96%	< 0.00001
Total		4	2.07 [1.17, 2.96]	94%	< 0.00001
NOS score	NOS score ≤ 5	7	2.56 [1.36, 3.76]	95%	< 0.00001
	NOS score > 5	16	1.17 [0.76, 1.57]	95%	< 0.00001
Total		23	1.44 [1.06, 1.83]	94%	< 0.00001
Unit	the unit (ng/ml)	12	1.11 [0.62, 1.60]	94%	< 0.00001
	the unit not (ng/ml)	11	1.93 [1.27, 2.60]	95%	< 0.00001
Total		23	1.44 [1.06, 1.83]	94%	< 0.00001
diagnosis methods of MAFLD	biopsy-proven	5	1.83 [0.71, 2.94]	93%	< 0.00001
	CT	2	0.62 [0.11, 1.14]	74%	0.05
	ultrasonography	9	1.63 [1.63, 2.21]	96%	< 0.00001
Total		16	1.52 [1.09, 1.95]	95%	< 0.00001

Abbreviations: BMI Body mass index, CI Confidence interval, CT Computed tomography, HOMA-IR Homeostasis model assessment of insulin resistance, NOS Newcastle-Ottawa Scale, SMD Standardised mean difference

Metaregression

To further explore the sources of heterogeneity, meta-regression analysis was performed (Table 4). Based on these results, the geographical area may be the cause of heterogeneity (p < 0.001), BMI (p = 0.252), age (p = 0.636), and HOMA-IR (p = 0.324).

Table 4.

Meta-regression of the Circulating plasminogen activator inhibitor-1 levels and metabolic-associated fatty liver disease

Covariates	No. Studies	Coefficient	Standard error	t	P	95%CI
Area	23	2.766303	0.5645066	4.90	< 0.001	[1.592347, 3.940259]
BMI	22	0.9848863	0.8343378	1.18	0.252	[−0.7555119, 2.725285]
Age	23	0.3835935	0.798549	0.48	0.636	[−1.27708, 2.044267]
HOMA-IR	11	1.353521	1.298344	1.04	0.324	[−1.583538, 4.29058]
NOS Score	23	−1.438933	0.742237	−1.94	0.066	[−2.982499, 0.1046335]
Unit	23	1.004103	0.6962657	1.44	0.164	[−0.4438606, 2.452067]

Abbreviations: BMI Body mass index, CI Confidence interval, HOMA-IR Homeostasis model assessment of insulin resistance, NOS Newcastle-Ottawa Scale

The NOS scores (p = 0.066) and units (p = 0.164) were not sources of heterogeneity. All meta-regression analyses are presented in Supplemental file 6.

Sensitivity analysis

Individual studies were sequentially removed to analyse the sensitivity of the meta-analysis (Fig. 4). The results showed that none of the articles significantly impacted the results, indicating good stability.

Fig. 4.

Sensitivity analysis plot of circulating PAI-1 levels between MASLD and the healthy control group

Publication bias

Publication bias assessed using Egger’s test is shown in Fig. 5. These results confirmed the existence of a publication bias (p < 0.1). The values changed after verification using the trim-and-filling method (before trimming, SMD = 1.490 [1.094, 1.886]; after filling, SMD = 1.180 [0.753, 1.608]). Although the SMD decreased, the results were not reversed, indicating the stability of the study. The funnel plot demonstrated publication bias (Fig. 6).

Fig. 5.

Egger’s publication bias plot of circulating PAI-1 levels between MASLD and the healthy control group

Fig. 6.

Funnel plot of circulating PAI-1 levels between MASLD and the healthy control group Supporting information

Discussion

Many risk factors are contributing to the increasing prevalence of MASLD. MASLD has been shown to have a mutual influence on metabolic abnormalities, such as obesity, MetS, and T2DM [48]. It can progress to liver cirrhosis and HCC, grievously damaging health, meaning it can be life-threatening [49]. Therefore, early noninvasive diagnostic technology and systematic treatment of MASLD are required to meet medical needs. The mechanisms of inflammation, oxidative damage, and apoptosis caused by liver insulin resistance, lipid accumulation, and crosslinking mitochondrial dysfunction are generally recognised [50]. Some regulatory proteins, such as cytokeratin-18 fragment [51], irisin [52] and leptin [53], participate in the inflammatory response of the liver and play a role in MASLD progression. An interaction between PAI-1 and MASLD was also observed.

Chang et al. observed that circulating PAI-1 levels were significantly increased in patients with MASLD. The increase in insulin and its precursors owing to insulin resistance contributes to the high expression of PAI-1 in hepatocytes [54]. Chen et al. reported that the PAI-1 promoter region contains reaction elements related to metabolic or inflammatory pathways, such as tumour necrosis factor, transforming growth factor, and very low-density lipoproteins [55, 56]. It has a proinflammatory effect on the liver, and its potential mechanism is related to an increase in fibrin content. Increased fibrin deposits in the liver’s extracellular matrix trigger inflammation via multiple mechanisms, including by slowing down blood flow, causing hypoxia of hepatocytes, or transmitting inflammatory signals through its receptors, including integrin αIIbβ3/αMβ2/αvβ3 [57]. A close relationship between circulating PAI-1 levels and endotoxin expression in the liver has been previously confirmed. An increase in endotoxins induces liver lipid accumulation and promotes steatohepatitis by activating the coagulation system [58]. Thus, PAI-1 is closely associated with MASLD, and is expected to be a target for MASLD treatment in the future.

Alsharoh et al. [59] conducted a meta-analysis of 11 studies that examined circulating PAI-1 levels in patients with NAFL and NASH and performed a subgroup analysis of diagnostic methods for NAFL. According to the new definition of MASLD, this meta-analysis combined circulating PAI-1 levels in patients with and without steatohepatitis for quantitative analysis. The diagnostic methods of MASLD, baseline patient data (e.g. BMI, age, and HOMA-IR, sex ratio), and the quality of the included articles were analysed to provide more details about the connection between MASLD and PAI-1. Additional databases (CNKI, Wanfang, Clinical Trials Database, and Grey Literature Database) were added to the literature retrieval, and the number of studies for meta-analysis was expanded from 11 to 23 to ensure that the conclusions would be more powerful and representative. Among the included studies, four selected in the review by Alsharoh et al. were excluded from the current meta-analysis. The reasons were as follows: (1) the studies included patients with subclinical coronary artery calcification as an underlying disease [60]; (2) they were cross-sectional studies without a healthy control group [61, 62]; and (3) the diagnosis of NAFLD and measurement of circulating PAI-1 levels for analysis were not performed at the same time point and were too long apart [63]. The overall results showed that circulating PAI-1 levels in patients with MASLD were significantly elevated compared to those in healthy groups. This conclusion is consistent with the results of a previous meta-analysis on NAFLD. However, Alsharoh et al. did not observe a significant increase in PAI-1 levels in patients with NASH, which is inconsistent with the results of the current review. Furthermore, sensitivity analysis was conducted to confirm the stability and consistency of the results.

Simultaneously, a subgroup analysis was performed from other perspectives (area, age, BMI, sex ratio, HOMA-IR, severity of MASLD, unit, NOS score, and diagnostic methods of MASLD) to explore the source of high heterogeneity. These factors were then included in the univariate and multivariate meta-regression analyses to further verify the potential source of heterogeneity. Hence, geographical area was found to be the potential cause of heterogeneity (p < 0.05), which is consistent with previous studies that showed regional differences in MASLD [64]; the outcome of the meta-regression further confirmed this finding. Evidence indicates that PAI-1 levels vary across populations, possibly due to genetic predispositions, lifestyle factors, or regional dietary patterns [65–67]. For example, Asian populations often exhibit higher PAI-1 levels than Western cohorts, which may reflect differences in metabolic profiles or MASLD severity [68]. These findings underscore the importance of considering ethnicity in interpreting PAI-1 as a MASLD biomarker, as well as the need for region-specific diagnostic thresholds to improve clinical applicability. Articles have reported differences in noninvasive diagnosis of MASLD between men and women, such as Crudele et al. [69] and Lonardo et al. [70]. However, it is regrettable that in the subgroup analysis based on the ‘male percentage’, no such differences were observed. This discrepancy may stem from the relatively balanced sex distribution across included studies. Likewise, studies have established a strong correlation between alcohol intake and PAI-1 levels [71]. Notably, within the framework of the MASLD, a specific subgroup, MetALD, includes individuals with alcohol consumption. Nonetheless, the majority of studies in our analysis adhered to the traditional NAFLD definition, which excludes individuals with daily alcohol consumption exceeding 20 g. This exclusion has resulted in a paucity of detailed data on alcohol intake, thereby limiting our ability to investigate its impact on PAI-1 levels, particularly within the MetALD subgroup. Future research should aim to elucidate the influence of alcohol consumption on PAI-1 levels in the context of MASLD to provide a more comprehensive understanding of this metabolic interplay.

This updated review further confirmed that circulating PAI-1 levels are associated with MASLD. This report lays the foundation for the exploration of noninvasive biomarkers for MASLD. The discovery of this relationship can help doctors identify potential patients with MASLD as early as possible in clinical practice, thus enabling timely therapy. As Asians accounted for the vast majority of the study population, these conclusions may be more applicable to Asian populations. More clinical studies from other regions are needed to confirm the stability of the relationship between PAI-1 and MASLD.

Advantages and limitations

This study analysed the correlation between circulating PAI-1 levels and MASLD and provided evidence for the clinical exploration of noninvasive markers to diagnose MASLD. This study also provides clues for clinical research on the therapeutic targets of MASLD. Seven databases covering six countries on four continents were searched to ensure the representativeness and popularisation of the conclusions. Detailed subgroup and meta-regression analyses were conducted to determine the underlying factors that produced the heterogeneity, and reasonable explanations and inferences were made. The sensitivity analysis also showed that the results were stable.

This meta-analysis has certain limitations. First, this meta-analysis confirmed that the levels of PAI-1 were closely related to MASLD. However, because the articles included were observational studies, the causal relationship between PAI-1 levels and MASLD could not be further explained. Second, some confounding factors of the participants were not completely adjusted for, such as alcohol consumption, hyperlipidaemia, and diabetes, which may influence circulating PAI-1 levels and contribute to the risk of bias. Owing to inconsistencies in the methods for measuring PAI-1, including the use of ELISA kits and chromogenic substrate detection, different studies have measured different units of circulating PAI-1, causing great variations in the values. Therefore, SMD was calculated to report the combined results. Additionally, circulating PAI-1 levels reportedly show an upward trend in the later stages of hepatic steatosis, with an increase in steatosis grade, lobular inflammation, and fibrosis stage [72]; however, this was not observed in the subgroup analysis of severity in our study. This may be because the number of included reports on severity was limited to only three articles. Finally, since all included patients with NAFLD had a BMI of ≥ 25 kg/m², analysis of normal-weight individuals was not possible in this study. Large-scale prospective cohort studies are warranted to further explore the correlation between circulating PAI-1 levels and MASLD progression.

Conclusion

This meta-analysis of 23 case-control studies demonstrated a potential association between circulating PAI-1 levels and MASLD, which may serve as an auxiliary diagnostic marker for this condition. Circulating PAI-1 levels were notably higher in patients with MASLD than in healthy controls. Regression analysis suggested that area and NOS scores were probable sources of heterogeneity. This meta-analysis will facilitate the development of clinical studies and the exploration of convenient noninvasive diagnostic biomarkers and effective therapeutic targets for circulating PAI-1 levels and MASLD. In addition, owing to the very low GRADE scores, our study can only act as a reference for the relationship between circulation PAI-1 level and MASLD. The role of PAI-1 in MASLD pathogenesis needs further comprehensive studies to substantiate the conclusions of this meta-analysis.

Abbreviations

CI

Confidence intervals

GRADE

Grading of recommendation assessment, development, and evaluation

HCC

Hepatocellular carcinoma

HOMA-IR

Homeostatic model assessment for insulin resistance

MAFLD

Metabolic dysfunction-associated steatotic liver disease

MASLD

Metabolic-associated fatty liver disease

MetS

Metabolic syndrome

NAFL

Non-alcoholic fatty liver

NAFLD

Non-alcoholic fatty liver disease

NASH

Non-alcoholic steatohepatitis

NOS

Newcastle-Ottawa Scale

PAI-1

Plasminogen activator inhibitor-1

PRISMA

Preferred Reporting Items for Systematic Reviews and Meta-Analyses

SMD

Sstandardised mean difference

T2DM

Type 2 diabetes mellitus

Authors’ contributions

Miaojuan Wang, Jie Hu and Shan Liu: Study design and revision of the article; Qicong Li and Yuting Ruan: Data collection, performing the analysis and drafting the article; Yuqing Zhu: study design, data collection; Yani Ke: Quality assessment of the included studies and revision of the article and revision of the article; Yingying Cai, Chenglu Shen, Qin Zhang, Shuaihang Chen and Kaihan Wu: Quality assessment of the included studies. All authors approved the final manuscript.

Funding

This study was supported by the Natural Science Foundation of Zhejiang Province, China (LQ19H290001), the research project of Zhejiang Chinese Medicine University (2021JKZKTS042B) and health science and Technology Project of Zhejiang Province (2022KY921).

Data availability

All data generated or analysed during this study are included in this published article and its supplementary information files.

Declarations

Ethics approval and consent to participate

This study was done in accordance with ethics guidelines of Helsinki. The protocol for this study was registered in the Prospective Register of Systematic Reviews (PROSPERO) website (https://www.crd.york.ac.uk/prospero/) with the following ID: CRD 420223312285.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.