Metabolic dysfunction-associated fatty liver and antiviral response in chronic hepatitis B: A systematic review and meta-analysis

Abstract

BACKGROUND

Metabolic dysfunction-associated fatty liver disease (MAFLD) frequently coexists with chronic hepatitis B (CHB). However, its association with antiviral treatment responses and long-term clinical outcomes in patients with CHB remains unclear.

AIM

To systematically evaluate the impact of MAFLD on antiviral treatment responses and long-term clinical outcomes in patients with CHB.

METHODS

This systematic review and meta-analysis was conducted in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses guidelines. PubMed, the Cochrane Library, EMBASE, and Web of Science were searched for studies published up to November 30, 2025. Meta-analyses were performed to assess associations between MAFLD and antiviral treatment responses or clinical outcomes. Subgroup analyses were conducted according to antiviral regimen and MAFLD diagnostic approach, and leave-one-out sensitivity analyses were performed to evaluate robustness.

RESULTS

Thirty-two studies were included, of which 20 evaluated antiviral treatment efficacy and 13 assessed clinical outcomes in patients with CHB and concomitant MAFLD (one study reported both treatment responses and clinical outcomes). MAFLD was associated with a higher rate of hepatitis B surface antigen seroclearance at 48 weeks [odds ratio (OR) = 1.41, 95%CI: 1.28-1.55], although this association was attenuated at 96 weeks (OR = 1.44, 95%CI: 0.96-2.16). No significant association was observed between MAFLD and hepatitis B e antigen seroclearance at any time-point. MAFLD was consistently associated with lower rates of alanine aminotransferase normalization and hepatitis B virus DNA clearance throughout the treatment duration. MAFLD was associated with a significantly increased risk of hepatocellular carcinoma (HCC) despite antiviral therapy (OR = 1.50, 95%CI: 1.01-2.24). Subgroup analyses according to the antiviral regimen revealed distinct patterns, partly explaining the heterogeneity in the overall analyses. Subgroup analyses suggested that the antiviral regimen and MAFLD diagnostic approach partly explained the heterogeneity, particularly for hepatitis B virus DNA clearance and HCC.

CONCLUSION

MAFLD was associated with lower virological and biochemical response rates, higher odds of hepatitis B surface antigen seroclearance at week 48, particularly among patients receiving nucleos(t)ide analogue therapy, and higher odds of HCC in patients with CHB, but not with hepatitis B e antigen seroclearance. These findings support metabolic risk assessment in CHB management and highlight the need for standardized MAFLD/metabolic dysfunction-associated steatotic liver disease definitions.

Key Words: Metabolic dysfunction-associated fatty liver disease; Chronic hepatitis B; Antiviral therapy; Meta-analysis; Systematic review; Clinical outcomes

Core Tip: Metabolic dysfunction-associated fatty liver disease (MAFLD) frequently coexists with chronic hepatitis B, but its clinical implications remain unclear. This meta-analysis showed that MAFLD was associated with lower rates of hepatitis B virus DNA clearance and alanine aminotransferase normalization, with no significant difference in hepatitis B e antigen seroclearance, and treatment duration-specific and regimen-specific differences in hepatitis B surface antigen seroclearance. MAFLD was also associated with higher odds of hepatocellular carcinoma, although this finding was influenced by diagnostic approach and showed limited robustness. These results support incorporating metabolic risk assessment into the long-term management of chronic hepatitis B.

INTRODUCTION

Hepatitis B virus (HBV) infection remains a major global public health challenge. Approximately one third of the global population has been exposed to HBV, with nearly 300 million individuals living with chronic infection that accounts for more than one million deaths annually worldwide[1]. Although antiviral therapy can effectively suppress viral replication, a considerable proportion of patients with chronic hepatitis B (CHB) still progress to advanced liver disease, including liver fibrosis, cirrhosis, and hepatocellular carcinoma (HCC). These observations highlight the importance of identifying non-viral factors that may influence the treatment response and long-term clinical outcomes in patients with CHB.

Metabolic dysfunction-associated fatty liver disease (MAFLD), previously referred to as nonalcoholic fatty liver disease (NAFLD), has emerged as one of the most prevalent chronic liver diseases worldwide. MAFLD is defined by the presence of hepatic steatosis in combination with cardiometabolic risk factors and reflects the central role of metabolic dysfunction in fatty liver disease[2]. Although MAFLD represents an evolution of disease nomenclature emphasizing metabolic dysfunction, accumulating evidence suggests that the clinical and pathophysiological differences between MAFLD and NAFLD are modest. Therefore, findings derived from studies using NAFLD-based definitions are generally considered applicable to MAFLD[3]. With the global increase in obesity and type 2 diabetes, the prevalence of MAFLD continues to increase. Consequently, the coexistence of CHB and MAFLD has become increasingly common in clinical practice, with a reported prevalence ranging from 20% to 40%[4]. Concurrence of viral hepatitis and metabolic dysfunction can complicate disease assessment, risk stratification, and therapeutic management.

However, the relationship between MAFLD and HBV infection remains complex and controversial. Mechanistic and clinical studies suggest bidirectional interactions, whereby hepatic steatosis and metabolic dysfunction may influence the intrahepatic immune milieu, fibrogenesis, and viral activity, while HBV-related factors and antiviral therapy may also affect hepatic lipid metabolism[5]. Consistent with this biological complexity, clinical studies have reported conflicting findings regarding the effect of MAFLD on antiviral responses and long-term outcomes in patients with CHB. Some studies suggest that hepatic steatosis is associated with poorer antiviral responses and increased risks of HCC, whereas others report suppressed HBV replication, higher rates of hepatitis B surface antigen (HBsAg) seroclearance, and lower risks of adverse clinical outcomes[4,6,7]. These discrepancies may reflect differences in study design, definitions of steatotic liver disease, antiviral regimens, and baseline patient characteristics.

Given the growing prevalence of metabolic liver disease among patients with CHB and the heterogeneity of existing evidence, a comprehensive synthesis of the available data is needed. Therefore, this study aimed to evaluate the effect of MAFLD on antiviral treatment responses and clinical outcomes in patients with CHB.

MATERIALS AND METHODS

Study design and registration

This systematic review and meta-analysis was conducted in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses guidelines. The study protocol was prospectively registered in the PROSPERO database (CRD420261280156).

Search strategy

Two investigators (Liu J and Bi YZ) independently searched PubMed, EMBASE, Web of Science, and the Cochrane Library from database inception to November 30, 2025. The search strategy was developed using the PICOS framework and combined Medical Subject Headings and relevant free-text terms related to CHB and metabolic fatty liver disease (e.g., “chronic hepatitis B”, “fatty liver”, “nonalcoholic fatty liver disease”, and “metabolic dysfunction-associated steatotic liver disease”), together with terms related to antiviral therapy and clinical outcomes [e.g., “nucleos(t)ide analogues”, “interferon”, and “hepatocellular carcinoma”] (Supplementary Table 1). The reference lists of the relevant articles were also screened to identify additional eligible studies.

Eligibility criteria

Studies were included if they met the following criteria: (1) Adult patients (≥ 18 years) with CHB, with or without MAFLD, receiving antiviral therapy [nucleos(t)ide analogues (NAs) and/or interferon (IFN)]; (2) Clear comparison between CHB with MAFLD and CHB alone; (3) Reporting at least one predefined outcome, including antiviral treatment responses [HBV DNA clearance, alanine aminotransferase (ALT) normalization, or hepatitis B e antigen (HBeAg)/HBsAg seroclearance] or clinical outcomes (liver fibrosis, cirrhosis, HCC, or mortality); and (4) Randomized controlled trials or observational studies (prospective or retrospective cohort, or cross-sectional). Only studies published in English were included in this meta-analysis.

Studies were excluded if they were reviews, editorials, case reports, conference abstracts, or studies involving animals or pediatric populations; included patients with other chronic liver diseases, HIV coinfection, or pregnancy-related liver disease; or lacked sufficient data for quantitative synthesis.

Study selection and data extraction

Study selection and data extraction were performed independently by two investigators (Liu J and Bi YZ), and disagreements were resolved thrpugh discussion with a third reviewer (Xin YN). After removal of duplicates, titles and abstracts were screened according to predefined eligibility criteria, followed by full-text assessment of potentially relevant studies. Data were extracted using a standardized form, including study characteristics (first author, country, and publication year), participant characteristics (sample size, age, MAFLD diagnostic method, and antiviral regimen), and study design. When necessary, numerical data were extracted from figures using GetData Graph Digitizer (version 2.26) or obtained by contacting the corresponding authors.

Risk of bias and quality assessment

The methodological quality of the included studies was independently assessed by two reviewers using the Newcastle-Ottawa Scale (NOS). The NOS evaluates study quality across three domains: Selection, comparability, and outcome/exposure assessment, with total scores ranging from 0 to 9. Studies scoring 7-9, 4-6, and 1-3 were considered high-quality, moderate-quality, and low-quality, respectively.

Statistical analysis

Meta-analyses were performed using Review Manager (version 5.4) and Stata (version 18.0). Odds ratios (ORs) with 95%CIs were used as summary effect estimates. Statistical heterogeneity was assessed using the I² statistic. A fixed-effects model was applied when heterogeneity was low to moderate (I² ≤ 50%); otherwise, a random-effects model was used. Prespecified subgroup analyses were conducted according to antiviral regimen, and additional subgroup analyses were performed according to the MAFLD diagnostic method to explore the potential sources of heterogeneity. Leave-one-out sensitivity analyses were conducted by sequentially excluding each study to evaluate the robustness of the pooled estimates. Publication bias was evaluated using funnel plots and Egger’s or Begg’s tests when appropriate.

RESULTS

Study selection

In total 1469 records were identified in the initial database search, of which 403 duplicates were removed. After screening titles, abstracts, and full texts according to the predefined eligibility criteria, 32 studies were included in the final meta-analysis, comprising 253336 patients with CHB, including those with concomitant MAFLD. The study selection process is illustrated in the Preferred Reporting Items for Systematic Reviews and Meta-Analyses flow diagram (Figure 1, Supplementary Table 2).

Figure 1  Flow chart of the study identification, screening and inclusion processes.

Study characteristics

Among the 32 included studies, 7 were prospective cohort studies and 25 were retrospective cohort studies. Total 20 studies[8-27] reported outcomes related to antiviral treatment efficacy, and 13 studies[21,28-39] evaluated clinical outcomes, with one study[21] reporting both. The studies were conducted across multiple regions, including China, South Korea, and the United States. Detailed study characteristics are summarized in Table 1[8-39]. Baseline characteristics were generally comparable between patients with CHB and MAFLD and those with CHB alone.

Table 1 Characteristics and methodological quality of studies included in the meta-analysis, mean ± SD.

Ref.	Study	Country	Sample size	Age		MAFLD diagnostics	Medication	Male		Female		NOS score
				CHB	CHB + MAFLD			CHB	CHB + MAFLD	CHB	CHB + MAFLD
Yendewa et al[8], 2025	RCS	United States	25460	53.90 ± 15.00	53.40 ± 13.10	Radiological	NAs	7599	7338	5006	5218	8
Yang et al[9], 2025	RCS	China	186	38.06 ± 12.76	-	Liver biopsy	NAs	85	57	33	11	8
Ren et al[10], 2025	RCS	China	320	46.21 ± 11.43	45.57 ± 10.61	FibroScan	NAs + PEG-IFN	106	76	89	49	8
Li et al[11], 2025	RCS	China	158	40.00 ± 8.00	39.00 ± 8.00	Ultrasound or FibroScan	NAs + PEG-IFN	75	60	16	7	8
Chau et al[12], 2025	RCS	A large international cohort	1736	49.50 ± 13.00	49.60 ± 12.00	Radiological	NAs	586	581	282	287	8
Li et al[13], 2024	RCS	China	120	29.50 ± 5.60	31.00 ± 7.10	Ultrasound and elasticity	NAs	53	37	27	3	6
Huang et al[14], 2024	RCS	China	4084	48.00 ± 14.83	51.65 ± 11.14	Ultrasound	NAs + PEG-IFN	1605	594	1592	293	8
Chen et al[15], 2024	RCS	China	120	39.00 ± 11.10	41.70 ± 10.40	Liver biopsy	NAs	28	48	33	11	8
Zhang et al[16], 2023	RCS	China	267	38.05 ± 12.71	37.51 ± 10.75	Liver biopsy	NAs	111	75	53	28	7
Li et al[17], 2021	RCS	United States and China	2158	49.69 ± 10.01	49.68 ± 9.13	Ultrasound or CT	NAs	768	763	311	316	8
Liang et al[18], 2020	PCS	China	226	29.50 ± 5.47	-	Liver biopsy	PEG-IFN-2a	82	95	37	12	6
Li et al[19], 2020	RCS	United States	555	45.50 ± 14.60	47.70 ± 12.50	Imaging or histology	NAs	211	126	157	61	7
Chen et al[20], 2020	RCS	China	196	36.20 ± 10.80	42.70 ± 10.80	Liver biopsy	NAs	61	81	33	21	8
Kim et al[21], 2019	RCS	Korea	334	50.40 ± 10.24	49.70 ± 10.70	FibroScan	NAs	119	91	69	55	6
Zhu et al[22], 2016	RCS	China	125	39.61 ± 10.87	39.26 ± 10.39	Ultrasound	NAs	41	42	23	19	7
Liu et al[23], 2016	PCS	China	60	37.35 ± 8.49	37.70 ± 7.90	CT ratios of liver/spleen	NAs	12	23	8	17	6
Gong et al[24], 2015	PCS	China	89	25.10 ± 7.80	30.30 ± 7.40	Liver biopsy	PEG-IFN-2a	38	20	20	11	7
Karaman et al[25], 2013	RCS	Turkey	119	40.03 ± 13.80	43.80 ± 11.70	Liver biopsy	NAs	-	-	-	-	6
Jin et al[26], 2012	PCS	China	213	39.55 ± 7.83	39.56 ± 11.87	Ultrasound	NAs	85	32	63	33	8
Ateş et al[27], 2011	PCS	Turkey	84	35.20 ± 8.90	50.50 ± 8.70	Liver biopsy	PEG-IFN	46	14	19	5	6
Amano et al[28], 2025	RCS	Japan	164	-	-	According to the international expert consensus	NAs	-	-	-	-	6
Huang et al[29], 2023	RCS	China	10546	47.00 ± 14.83	52.00 ± 11.87	Ultrasound	NAs + PEG-IFN	3880	1505	4454	707	8
Yun et al[30], 2022	RCS	Korea	197346	53.30 ± 8.90	53.00 ± 8.90	Fatty liver index ≥ 30	NAs + PEG-IFN	60068	51335	71129	14814	7
Yu et al[31], 2022	PCS	China	1453	50.31 ± 9.46	50.33 ± 9.80	Ultrasound	NAs + PEG-IFN	-	-	-	-	7
Ko et al[32], 2022	CSS, RCS	Canada	1178	46.2 ± 1.64	-	Ultrasound	NAs + PEG-IFN	289	387	262	237	6
van Kleef et al[33], 2021	RCS	Canada, the Netherlands	1076	-	-	Liver biopsy	NAs + PEG-IFN	470	237	310	59	8
Mak et al[34], 2021	PCS	China	1914	55.95 ± 10.91	55.11 ± 11.73	FibroScan	NAs	559	563	398	394	8
Lim et al[35], 2020	RCS	Singapore	289	-	-	Liver biopsy	NAs + PEG-IFN	68	141	36	44	8
Choi et al[36], 2020	RCS	Canada, the Netherlands	1089	37.60 ± 13.00	43.90 ± 12.80	Liver biopsy	NAs + PEG-IFN	569	149	335	36	6
Cho et al[37], 2020	RCS	Korea	826	54.00 ± 8.80	52.00 ± 9.40	Ultrasound	NAs	333	171	233	89	8
Peleg et al[38], 2019	RCS	Israeli	524	42.80 ± 11.42	50.73 ± 10.87	Liver biopsy	NAs + PEG-IFN	173	142	110	99	8
Lee et al[39], 2019	RCS	Korea	321	40.30 ± 11.93	43.94 ± 11.35	Liver biopsy	NAs	146	50	105	20	8

PCS: Prospective cohort study; RCS: Retrospective cohort study; CSS: Cross-sectional study; CHB: Chronic hepatitis B; MAFLD: Metabolic dysfunction-associated fatty liver disease; NAs: Nucleos(t)ide analogues; PEG-IFN: Pegylated interferon; NOS: Newcastle-Ottawa Scale; CT: Computed tomography.

Risk of bias assessment

The risk of bias of the included studies was independently assessed by two reviewers using the NOS. Overall, all included studies were considered to have an acceptable risk of bias, with quality assessment results summarized in Table 1[8-39].

Results of the meta-analysis

HBsAg seroclearance rate: Eight studies[8,10,11,14,16,17,20,24] reported HBsAg seroclearance, including seven studies with 48-week data and five with 96-week data. At 48 weeks (Figure 2A)[8,10,11,14,17,20,24], patients with CHB and concomitant MAFLD had a significantly higher HBsAg seroclearance rate than those with CHB alone (OR = 1.41, 95%CI: 1.28-1.55, P < 0.00001), with low heterogeneity (I² = 43%). At 96 weeks (Figure 2B)[8,14,16,17,24], a similar trend was observed but did not reach statistical significance (OR = 1.44, 95%CI: 0.96-2.16, P = 0.08), with substantial heterogeneity (I² = 71%).

Figure 2 Forest plots comparing hepatitis B surface antigen seroclearance rates between patients with chronic hepatitis B alone and those with chronic hepatitis B and metabolic dysfunction-associated fatty liver disease. A: 48 weeks; B: 96 weeks, C: 48 weeks, nucleos(t)ide analogues therapy; D: 48 weeks, interferon-based therapy; E: 96 weeks, nucleos(t)ide analogues therapy; F: 96 weeks, interferon-based therapy. Pooled effects are expressed as odds ratios with 95%CIs. MAFLD: Metabolic dysfunction-associated fatty liver disease; CHB: Chronic hepatitis B.

HBeAg seroclearance rate: Nine studies[8,11,16,20-24,26] evaluated HBeAg seroclearance. No significant association between MAFLD and HBeAg seroclearance was observed at 24 weeks (Figure 3A, OR = 0.90, 95%CI: 0.52-1.59, P = 0.73)[11,16,23,26], 48 weeks (Figure 3B; OR = 0.99, 95%CI: 0.61-1.60, P = 0.97)[8,11,16,20-22,24,26], or 96 weeks (Figure 3C; OR = 0.96, 95%CI: 0.60-1.54, P = 0.88)[8,16,20,22,24,26]. Moderate heterogeneity was observed at 48 weeks (I² = 67%) and 96 weeks (I² = 68%).

Figure 3 Forest plots comparing hepatitis B e antigen seroclearance rates between patients with chronic hepatitis B alone and those with chronic hepatitis B and metabolic dysfunction-associated fatty liver disease. A: 24 weeks; B: 48 weeks; C: 96 weeks; D: 48 weeks nucleos(t)ide analogues therapy; E: 48 weeks, interferon-based therapy. Pooled effects are expressed as odds ratios with 95%CIs. MAFLD: Metabolic dysfunction-associated fatty liver disease; CHB: Chronic hepatitis B.

ALT normalization rate: Ten studies[11-13,16,18,19,22-24,26] reported ALT normalization rates. Compared with patients with CHB alone, those with MAFLD had significantly lower ALT normalization rates at 24 weeks (Figure 4A, OR = 0.45, 95%CI: 0.28-0.72, P = 0.0009)[11,16,18,22,23,26], 48 weeks (Figure 4B; OR = 0.51, 95%CI: 0.34-0.75, P = 0.0008)[11-13,16,18,19,22,24,26], and 96 weeks (Figure 4C; OR = 0.47, 95%CI: 0.31-0.72, P = 0.0005)[12,13,16,19,22,24,26]. Moderate heterogeneity was observed across studies (I² range: 55%-74%).

Figure 4 Forest plots comparing alanine aminotransferase normalization rates between patients with chronic hepatitis B alone and those with chronic hepatitis B and metabolic dysfunction-associated fatty liver disease. A: 24 weeks; B: 48 weeks; C: 96 weeks; D: 24 weeks nucleos(t)ide analogues therapy; E: 24 weeks, interferon-based therapy; F: 48 weeks nucleos(t)ide analogues therapy; G: 48 weeks, interferon-based therapy. Pooled effects are expressed as odds ratios with 95%CIs. MAFLD: Metabolic dysfunction-associated fatty liver disease; CHB: Chronic hepatitis B.

HBV-DNA clearance rate: Fifteen studies[9,12,13,15,16,18-27] reported HBV DNA clearance rates. MAFLD was associated with significantly lower rates of HBV DNA clearance at 24 weeks (Figure 5A, OR = 0.57, 95%CI: 0.44-0.75, P < 0.0001)[15,16,18,22,23,26], 48 weeks (Figure 5B, OR = 0.71, 95%CI: 0.53-0.93, P = 0.01)[9,12,13,15,16,18-22,24-27], 96 weeks (Figure 5C, OR = 0.60, 95%CI: 0.40-0.89, P = 0.01)[12,13,16,19,22,24,26]. Heterogeneity was low at 24 weeks (I² = 0%) but substantial at later time points (48 weeks: I² = 69%; 96 weeks: I² = 67%).

Figure 5 Forest plots comparing hepatitis B virus DNA clearance rates between patients with chronic hepatitis B alone and those with chronic hepatitis B and metabolic dysfunction-associated fatty liver disease. A: 24 weeks; B: 48 weeks; C: 96 weeks; D: 48 weeks nucleos(t)ide analogues therapy; E: 48 weeks, interferon-based therapy. Pooled effects are expressed as odds ratios with 95%CIs. MAFLD: Metabolic dysfunction-associated fatty liver disease; CHB: Chronic hepatitis B.

Risk of HCC

Thirteen studies[21,28-39] reported the risk of HCC with a minimum follow-up of five years. Pooled analysis showed that patients with CHB and MAFLD had higher odds of HCC than those with CHB alone (OR = 1.50, 95%CI: 1.01-2.24, P = 0.04). Substantial heterogeneity was observed across studies (I² = 87%) (Figure 6)[21,28-39].

Figure 6 Forest plot comparing the risk of hepatocellular carcinoma between patients with chronic hepatitis B alone and those with chronic hepatitis B and metabolic dysfunction-associated fatty liver disease. Pooled effects are expressed as odds ratios with 95%CIs. MAFLD: Metabolic dysfunction-associated fatty liver disease; CHB: Chronic hepatitis B.

Subgroup analyses stratified by antiviral regimen

To explore potential sources of heterogeneity, prespecified subgroup analyses were conducted according to antiviral regimen, including NAs-based and IFN-based therapies.

HBsAg seroclearance rate

Among patients receiving NAs therapy, MAFLD was associated with a significantly higher HBsAg seroclearance rate at 48 weeks (Figure 2C; OR = 1.39, 95%CI: 1.26-1.53, P < 0.00001; I² = 0%)[8,17,20]. In the IFN-treated subgroup, the association at 48 weeks was not statistically significant (Figure 2D; OR = 1.62, 95%CI: 0.77-3.41, P = 0.20; I² = 63%)[10,11,14,24]. At 96 weeks, MAFLD remained significantly associated with HBsAg seroclearance in the NAs-treated subgroup (Figure 2E; OR = 1.11, 95%CI: 1.02-1.21, P = 0.02; I² = 38%)[8,16,17], whereas a stronger association was observed in the IFN-treated subgroup (Figure 2F; OR = 2.06, 95%CI: 1.42-2.98, P = 0.0001; I² = 0%)[14,24].

HBeAg seroclearance rate

Subgroup analyses at 48 weeks revealed no significant association between MAFLD and HBeAg seroclearance in either treatment group. The pooled OR was 1.06 (95%CI: 0.61-1.84, P = 0.83; I² = 69%) in the NAs-treated subgroup (Figure 3D)[8,16,20-22,26] and 0.79 (95%CI: 0.39-1.58, P = 0.50; I² = 0%) in the IFN-treated subgroup (Figure 3E)[11,24].

ALT normalization rate

At 24 weeks, MAFLD was associated with significantly lower ALT normalization rates in both subgroups (NAs therapy, Figure 4D: OR = 0.65, 95%CI: 0.44-0.94, P = 0.02; IFN therapy, Figure 4E: OR = 0.25, 95%CI: 0.15-0.41, P < 0.00001), with no heterogeneity (I² = 0%)[11,16,18,22,23,26]. Consistent results were observed at 48 weeks, with reduced ALT normalization in both NAs-treated (Figure 4F, OR = 0.76, 95%CI: 0.64-0.89, P = 0.001; I² = 27%)[12,13,16,19,22,26] and IFN-treated patients (Figure 4G, OR = 0.40, 95%CI: 0.16-0.99, P = 0.05; I² = 79%)[11,18,24].

HBV-DNA clearance rate

At 48 weeks, MAFLD was associated with reduced HBV DNA clearance with both antiviral regimens. The pooled OR was 0.74 (95%CI: 0.54-1.00, P = 0.05; I² = 71%) in the NA-treated subgroup (Figure 5D)[9,12,13,15,16,19-22,25,26] and 0.54 (95%CI: 0.35-0.85, P = 0.008; I² = 23%) in the IFN-treated subgroup (Figure 5E)[18,24,27].

Subgroup analyses stratified by MAFLD diagnostic method

For subgroup analyses according to the diagnostic approach used to identify MAFLD, the included studies were classified into three categories: (1) Ultrasound-based assessment, including conventional ultrasonography and FibroScan-based assessment; (2) Histology-based assessment, including liver biopsy only; and (3) Other imaging/clinical assessment, including all remaining diagnostic methods.

Overall, the diagnostic approach did not significantly modify the association between MAFLD and HBsAg seroclearance at 48 weeks or HBeAg seroclearance at 48 weeks and 96 weeks. For HBsAg seroclearance at 48 weeks (Supplementary Figure 1)[8,10,11,14,17,20,24], the pooled ORs were 1.58 (95%CI: 0.98-2.55), 1.03 (95%CI: 0.33-3.18), and 2.26 (95%CI: 0.69-7.36) in the ultrasound-based, histology-based, and other imaging/clinical assessment subgroups, respectively, with no significant subgroup difference. Similarly, MAFLD was not significantly associated with HBeAg seroclearance in either the ultrasound-based or histology-based subgroup at 48 weeks or 96 weeks (Supplementary Figures 2 and 3)[8,11,16,20-22,24,26]. For ALT normalization at 48 weeks (Supplementary Figure 4)[11-13,16,18,19,22,24,26], MAFLD tended to be associated with lower normalization rates across diagnostic subgroups, with the clearest association observed in the ultrasound-based subgroup (OR = 0.46, 95%CI: 0.32-0.67). Significant subgroup differences were observed for HBV DNA clearance at 48 weeks and HCC risk. MAFLD was associated with reduced HBV DNA clearance in the ultrasound-based subgroup (OR = 0.56, 95%CI: 0.41-0.76), whereas this association was attenuated in the histology-based subgroup and absent in the other imaging/clinical assessment subgroup (Supplementary Figure 5)[21,28-39]. For HCC, increased risks were observed in the histology-based subgroup (OR = 3.28, 95%CI: 2.26-4.74) and the other imaging/clinical assessment subgroup (OR = 2.01, 95%CI: 1.94-2.09), but not in the ultrasound-based subgroup (OR = 0.81, 95%CI: 0.47-1.38) (Supplementary Figure 6)[21,28-39].

Publication bias and sensitivity analysis

Publication bias was evaluated using funnel plots and Egger’s regression tests where applicable. Most analyses showed symmetric funnel plots and non-significant Egger test results (all P > 0.05), suggesting no substantial publication bias. However, funnel plot asymmetry was observed for HBeAg seroclearance at 48 weeks and 96 weeks, ALT normalization at 96 weeks, and HCC risk, indicating potential small-study effects in these analyses. These findings should therefore be interpreted with caution.

Leave-one-out sensitivity analyses were further performed to evaluate the robustness of the pooled estimates. The results for HBsAg seroclearance at 48 weeks, HBeAg seroclearance at 48 weeks and 96 weeks, ALT normalization at 48 weeks and 96 weeks, and HBV DNA response at 48 weeks and 96 weeks remained generally consistent with the primary analyses after sequential exclusion of individual studies, suggesting that these findings were not substantially influenced by any single study. However, for HBsAg seroclearance at 96 weeks, exclusion of the largest study (Yendewa et al[8], 2025) resulted in a change in the statistical significance of the pooled estimate, indicating that this result may be influenced by this large-scale study. Similarly, the statistical significance of the pooled HCC estimate was not retained after the exclusion of certain individual studies, indicating limited robustness of the HCC finding. These results should therefore be interpreted cautiously.

DISCUSSION

In this systematic review and meta-analysis, we evaluated the associations between MAFLD and antiviral treatment responses and clinical outcomes in patients with CHB. Across 32 studies including 253336 patients, concomitant MAFLD was associated with lower rates of HBV DNA clearance and ALT normalization during antiviral therapy, no significant difference in HBeAg seroclearance, treatment duration- and regimen-specific differences in HBsAg seroclearance, and higher odds of HCC during long-term follow-up. These findings suggest that metabolic dysfunction may have clinical relevance beyond its coexistence with CHB, particularly in the interpretation of treatment response and long-term risk assessment. Because the included studies used different methods to identify MAFLD, we further performed subgroup analyses according to diagnostic approach. The diagnostic approach did not materially alter the findings for HBsAg seroclearance at 48 weeks or HBeAg seroclearance at 48 weeks and 96 weeks, whereas it appeared to account for part of the heterogeneity observed in ALT normalization, HBV DNA clearance, and HCC risk. Leave-one-out sensitivity analyses also supported the overall stability of most treatment-response outcomes, but the estimates for HBsAg seroclearance at 96 weeks and HCC risk were less robust after exclusion of certain influential studies. Therefore, these findings should be interpreted as clinically relevant associations rather than evidence of a direct causal effect of MAFLD on antiviral efficacy or hepatocarcinogenesis.

Several mechanisms may underlie the observed associations between MAFLD and virological and biochemical outcomes; however, these outcomes should be interpreted separately. MAFLD is characterized by insulin resistance, lipotoxicity, and chronic low-grade inflammation, which may contribute to persistent hepatocellular injury even when HBV replication is adequately suppressed. Therefore, the lower ALT normalization rate observed in patients with concomitant MAFLD may reflect ongoing metabolic liver injury, including steatosis-related or steatohepatitis-related inflammation, rather than inadequate antiviral efficacy[5,40]. As a nonspecific marker of hepatocellular injury, ALT normalization should not be regarded as a direct surrogate for virological response. In the diagnostic-method subgroup analysis, MAFLD tended to be associated with lower ALT normalization rates across all diagnostic subgroups, with the clearest association observed in the ultrasound-based subgroup. Similar but less precise trends were observed in the histology-based and other imaging/clinical assessment subgroups, supporting the view that impaired biochemical normalization is not solely attributable to a single diagnostic approach. By contrast, the lower rate of HBV DNA clearance represents a virological association that may involve different mechanisms. Metabolic liver injury, particularly in the presence of steatohepatitis, may alter the intrahepatic immune microenvironment and antiviral immune signaling[41]. In our subgroup analyses stratified by antiviral regimen, the inverse association between MAFLD and HBV DNA clearance appeared more pronounced among patients receiving IFN-based therapy, suggesting a possible interaction between metabolic dysfunction and immune-mediated treatment response. This interpretation is consistent with previous reports of an association between hepatic steatosis, metabolic abnormalities, and reduced IFN responsiveness in chronic viral hepatitis[5,10,11,16]. Notably, subgroup analyses according to MAFLD diagnostic approach showed significant differences for HBV DNA clearance at 48 weeks. The association was most evident in the ultrasound-based subgroup, was attenuated in the histology-based subgroup, and was not observed in the other imaging/clinical assessment subgroup. These findings suggest that diagnostic approach may partly explain heterogeneity in virological response, although differences in baseline metabolic burden, viral activity, antiviral regimen, and outcome definitions across studies may also have contributed to the observed variability[40,42].

In contrast to virological and biochemical responses, our analysis showed no significant association between MAFLD and HBeAg seroclearance across all evaluated time points and antiviral regimens. This finding is consistent with previous reports and suggests that HBeAg seroclearance is primarily driven by viral replication dynamics and host adaptive immune responses rather than by metabolic liver injury[6,11,20]. Accordingly, existing clinical predictors of HBeAg seroclearance during antiviral therapy largely emphasize virological and immune-related parameters, such as baseline HBeAg levels and hepatitis B core antibody titers, rather than metabolic characteristics[43]. Importantly, the absence of an association persisted after stratification by antiviral regimen and was also consistent across diagnostic-method subgroups at both 48 weeks and 96 weeks. Leave-one-out sensitivity analyses further showed that the pooled estimates for HBeAg seroclearance at 48 weeks and 96 weeks remained generally stable after sequential exclusion of individual studies. These results strengthen the interpretation that MAFLD is not meaningfully associated with HBeAg seroclearance in patients with CHB receiving antiviral therapy.

In contrast, MAFLD was associated with a higher likelihood of HBsAg seroclearance during antiviral therapy in patients with CHB. Although this association was evident at 48 weeks, it became less consistent with longer follow-up. Stratification by antiviral regimen revealed distinct patterns, suggesting that therapeutic strategy may influence the relationship between MAFLD and HBsAg clearance[44]. Several mechanisms may contribute to this observation. Hepatic steatosis and metabolic stress can alter the intrahepatic immune microenvironment and promote oxidative stress, potentially enhancing antiviral immune activity[29]. Moreover, alterations in lipid metabolism may suppress HBV replication and reduce the transcriptional activity of covalently closed circular DNA, thereby decreasing viral antigen production[20]. Furthermore, metabolic stress states may facilitate the activation of envelope-specific CD4⁺ T cells, thereby promoting B-cell activation and anti-HBs antibody production, which could contribute to outcomes associated with functional cure[45]. Regimen-specific analyses further suggested different temporal dynamics. The association between MAFLD and HBsAg seroclearance was more apparent during the early phase of NAs therapy but attenuated with prolonged treatment, whereas in IFN-based therapy the association emerged at later time points, possibly reflecting delayed immune-mediated effects. These treatment-specific differences may also partly explain the heterogeneity observed in longer follow-up analyses. Simultaneously, subgroup analyses according to MAFLD diagnostic approaches showed consistent trends across different methods, and no significant differences were observed between subgroups, suggesting that the diagnostic approach may not substantially modify the association at this time point. The corresponding 95% prediction interval ranged from 0.65 to 3.07, indicating that the magnitude of the association may vary across different populations. At 96 weeks, the pooled analysis showed a similar direction of effect but did not reach statistical significance. The 95% prediction interval (0.40-5.20) further suggested considerable uncertainty in the estimated effect across populations. Leave-one-out sensitivity analysis indicated that this finding was influenced by the largest cohort (Yendewa et al[8], 2025); after exclusion of this study, the association became consistent with the 48-week analysis. These findings suggest that MAFLD may be associated with HBsAg seroclearance during antiviral therapy, although the strength of this association may vary according to follow-up duration, cohort characteristics, and study population. Further studies with longer follow-up and well-characterized populations are needed to clarify this relationship.

In contrast to some previous reports[29,31,33], our meta-analysis demonstrated that MAFLD was associated with a significantly increased long-term risk of HCC in patients with CHB despite antiviral therapy. This association identifies metabolic dysfunction as a potential risk marker beyond viral suppression, but does not establish MAFLD as a causal determinant of hepatocarcinogenesis. Several mechanisms may contribute to this increased risk. First, MAFLD is characterized by lipotoxicity, oxidative stress, and chronic low-grade inflammation, which promote genomic instability, hepatocyte turnover, and fibrogenesis – processes that can drive carcinogenesis independently of viral replication[46-48]. In addition, metabolic dysfunction may accelerate fibrosis progression and reshape the hepatic tumor microenvironment, facilitating malignant transformation even when HBV DNA is effectively suppressed[33,49]. Importantly, Emerging evidence also suggests potential interactions between MAFLD-related pathways and HBV-specific oncogenic mechanisms, including residual covalently closed circular DNA activity and viral integration[6,50]. The diagnostic-method subgroup analysis also helped explain heterogeneity in the HCC outcome. Increased HCC risk was observed mainly in histology-based studies and in studies using other imaging/clinical assessment, whereas no significant association was found in the ultrasound-based subgroup. This difference may reflect variation in disease severity and case ascertainment, as liver biopsy may capture more advanced metabolic liver injury, including steatohepatitis and fibrosis, while clinical or composite criteria may identify patients with a greater systemic metabolic burden. By contrast, ultrasound-based assessment mainly detects hepatic steatosis and is less informative for inflammatory activity or fibrosis severity. Moreover, leave-one-out sensitivity analysis showed that the statistical significance of the pooled HCC estimate was not retained after exclusion of certain individual studies, indicating limited robustness. These findings support continued HCC surveillance and metabolic risk assessment in patients with CHB and concomitant MAFLD, but the magnitude of HCC risk should be interpreted cautiously and validated in future prospective studies.

Limitations and strengths

Several limitations of this study should be acknowledged. First, most included studies were observational, which precludes causal inference and may be subject to residual confounding, particularly related to metabolic comorbidities and fibrosis stage. Accordingly, the pooled estimates should be interpreted as associations rather than evidence that MAFLD directly impairs antiviral response or causes HCC. Second, the definition and ascertainment of MAFLD varied across studies, reflecting the evolution of diagnostic criteria and the use of different assessment methods. Although subgroup analyses were performed according to the diagnostic approach used, residual heterogeneity could not be completely eliminated. Moreover, because alcohol consumption was not uniformly reported or adjusted for across the included studies, potential confounding by alcohol-related liver injury cannot be excluded, particularly in studies using MAFLD-based definitions that do not necessarily exclude alcohol intake. Third, several virological and methodological factors, including baseline HBV DNA levels, HBeAg status, quantitative HBsAg levels, outcome definitions, and detection thresholds, have been inconsistently reported across studies. These differences may have influenced the pooled estimates for antiviral treatment responses and limited the comparability of individual studies. Fourth, the robustness of some pooled estimates was limited, particularly for outcomes based on fewer studies or those influenced by large cohorts. Finally, some analyses suggested possible small-study effects. However, the interpretation of publication bias tests is limited when the number of studies is small and when substantial heterogeneity or rare outcomes are present. Despite these limitations, this meta-analysis has several strengths. The large sample size and multinational representation enhance the robustness and generalizability of the findings. Moreover, this study evaluated a broad range of antiviral treatment responses and long-term clinical outcomes. Subgroup analyses stratified by antiviral regimen and MAFLD diagnostic approach helped explore the sources of heterogeneity and improved the interpretability of the pooled estimates. Leave-one-out sensitivity analyses also distinguished relatively stable findings from those requiringed more cautious interpretation.

CONCLUSION

In conclusion, this meta-analysis suggests that MAFLD is associated with lower rates of HBV DNA clearance and ALT normalization, no significant difference in HBeAg seroclearance, and treatment duration-specific and regimen-specific differences in HBsAg seroclearance among patients with CHB receiving antiviral therapy. MAFLD was also associated with higher odds of HCC during long-term follow-up, although this finding showed limited robustness in the sensitivity analyses and varied according to diagnostic approach. These findings support the integration of metabolic risk assessment into the management of patients with CHB, while highlighting the need for standardized MAFLD/metabolic dysfunction-associated steatotic liver disease definitions and prospective studies with adequate adjustment for fibrosis stage and metabolic comorbidities.