Impact of donor-specific antibodies on immune rejection rates in liver transplantation: A systematic review and meta-analysis

Abstract

BACKGROUND

Liver transplantation is the definitive treatment for end-stage liver disease. While the liver is considered relatively immunotolerant, the clinical impact of donor-specific antibodies (DSAs) remains controversial. Some evidence suggests DSAs contribute to antibody-mediated rejection and graft injury, whereas other data indicate minimal effects on outcomes. This uncertainty complicates post-transplant management and risk assessment. We hypothesized that the presence of DSAs, particularly those with high mean fluorescence intensity, is significantly associated with increased rates of immune rejection in liver transplant recipients.

AIM

To investigate the effect of DSAs on immune rejection rates in patients receiving liver transplants.

METHODS

A systematic review and meta-analysis were conducted using PubMed, Excerpta Medica, and Cochrane databases (January 2010 to March 2025). Out of 1550 records, 10 studies meeting inclusion criteria were selected. Data on demographics, mean fluorescence intensity, and rejection outcomes were extracted. Statistical analysis utilized pooled odds ratios with 95% confidence intervals, sensitivity analysis to assess study variability, and meta-regression to evaluate the impact of mean fluorescence intensity levels, patient age, and sex on rejection risk.

RESULTS

The pooled analysis showed no significant association between the presence of donor-specific antibodies and antibody-mediated rejection (odds ratio = 0.528; 95% confidence interval: 0.227-1.228). However, meta-regression demonstrated that increased mean fluorescence intensity levels significantly correlated with higher rejection risk. Sensitivity analysis indicated that excluding one specific study altered the statistical significance of the primary outcome, highlighting heterogeneity. An inverse association was observed between Model for End-Stage Liver Disease scores and rejection risk. Factors such as DSA class ratio, patient age, and sex had minimal impact on outcomes.

CONCLUSION

While donor-specific antibodies alone do not consistently predict rejection, high mean fluorescence intensity levels serve as indicators of adverse outcomes. Standardized detection and personalized immunosuppression may optimize graft survival.

Key Words: Donor-specific antibodies; Liver transplantation; Antibody-mediated rejection; Mean fluorescence intensity; Meta-analysis; Graft survival; Systematic review

Core Tip: This meta-analysis evaluates the impact of donor-specific antibodies (DSAs) on immune rejection following liver transplantation. While the presence of DSAs alone may not consistently predict rejection, quantitative measures - specifically high mean fluorescence intensity levels - are significantly correlated with increased rejection risk. These findings suggest that standardized, quantitative DSA monitoring, rather than simple qualitative detection, is essential for guiding personalized immunosuppressive protocols and optimizing long-term graft survival.

INTRODUCTION

Liver transplantation (LT) is the definitive treatment for patients with end-stage liver diseases, including cirrhosis, acute liver failure and hepatocellular carcinoma, providing notable improvements of survival and quality of life[1]. The global demand for LT continues to exceed organ availability. In recent studies, it is estimated that the prevalence of patients awaiting a liver transplant range from 15 to 50 per 1 million people in various regions, reflecting a significant unmet clinical need[1-3]. However, despite advancements in surgical techniques and immunosuppressive therapies - including calcineurin inhibitors (e.g., tacrolimus and cyclosporine), antiproliferative agents (e.g., mycophenolate mofetil), mammalian target of rapamycin inhibitors, and corticosteroids - immune-mediated graft rejection remains a notable challenge that threatens graft survival and patient outcomes[4,5]. Immune rejection in LT is categorized into chronic rejection, antibody-mediated rejection (AMR) and acute cellular rejection[6,7]. Previous studies and clinical management focused on T-cell-mediated mechanisms underlying acute cellular rejection[8,9]. However, emerging evidence highlights the role of donor-specific antibodies (DSAs), which are antibodies directed against donor human leukocyte antigen (HLA) molecules. DSAs lead to AMR, a phenomenon recognized for its association with chronic graft injury and fibrosis[10].

Unlike the established predictive value of DSAs for graft loss and acute rejection in kidney and heart transplants, their role in LT is controversial. The unique immunotolerant microenvironment of the liver, which is attributed to its regenerative ability, Kupffer cell activity and secretion of soluble HLA antigens, is hypothesized to reduce DSA-mediated injury[11-13]. However, this paradigm is challenged by studies demonstrating that preformed or de novo DSAs are linked to early allograft dysfunction, biliary complications and chronic rejection characterized by ductopenia and vasculopathy[11]. For example, pre-transplant DSAs are associated with acute AMR in approximately 1% of all early (< 90 days) liver allograft failures, a rate that rises to 10% among DSA-positive patients with idiopathic early graft loss. Furthermore, the presence of persistent post-transplant DSAs is associated with an increased risk of chronic rejection, particularly in the setting of less intensive immunosuppression[14]. These findings suggest that the tolerance of the liver may depend on the characteristics of DSAs such as mean fluorescence intensity (MFI), complement fixation capacity and duration of exposure to DSAs[14-16].

Heterogeneity in study design, DSA detection methodologies and outcome definitions confounds the literature on DSAs in LT. Previous studies rely on complement-dependent cytotoxicity assays, which lack sensitivity compared with contemporary solid-phase assays (such as Luminex)[17]. The latter enables the precise detection of antibody specificity and strength[17]. This technological evolution identifies nuanced associations between DSAs and subclinical inflammation; however, inconsistencies persist. While a number of cohorts report notable associations between high-MFI DSAs and rejection[18-20], no independent impact is found in other studies, highlighting the influence of confounding variables such as immunosuppression protocols and HLA mismatches[11,21-23]. Furthermore, the 2016 Banff Working Group highlighted the diagnostic challenges of AMR in liver grafts, emphasizing the need for standardized histopathological criteria[24].

Despite growing interest in the clinical relevance of DSAs, a comprehensive synthesis of the evidence is yet to be elucidated. Existing reviews are often limited by outdated data or a narrow focus on specific subtypes of DSAs, leaving questions unresolved, such as whether DSAs consistently increase the risk of rejection, or whether their effect is modulated by immunological and clinical variables. This meta-analysis aimed to quantify the impact of DSAs on immune rejection rates following LT by integrating data from both observational and interventional studies. Furthermore, the objective of this study was to potentially support risk stratification, inform post-transplant monitoring protocols and optimize immunosuppressive strategies by clarifying dose-response associations, temporal dynamics and relevant risk modifiers. An understanding of the DSA-rejection association may be required to advance personalized medicine in transplant hepatology and to improve long-term outcomes and the allocation of donor organs.

MATERIALS AND METHODS

Present study was designed to explore the impact of DSAs on immune rejection rates in LT. The procedural steps of this meta-analysis were aligned with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses recommendations to ensure transparency and rigor[25].

Search strategy

A multi-platform systematic search was executed using major electronic databases, including PubMed, EMBASE, and the Cochrane Library, covering the period up to March 2025. No linguistic constraints were applied during this retrieval phase. The Medical Subject Heading terms and keywords used were “liver transplantation”, “donor-specific antibodies”, “DSA”, “rejection”, “graft survival” and “immune-mediated injury”. To ensure comprehensive coverage, we also conducted a secondary screening of reference lists from pertinent publications.

Inclusion and exclusion criteria

Studies were considered eligible for inclusion in this study if they met the following criteria: (1) Liver transplant recipients were aged ≥ 18 years; (2) DSAs (whether preformed or de novo) were present in the exposure group; (3) DSAs were absent in the comparison group; (4) Outcomes such as acute rejection, chronic rejection, graft survival, or AMR were reported; and (5) The study design was a randomized controlled trial (RCT), prospective cohort, or retrospective cohort.

Studies were excluded if: (1) The study focused solely on pediatric LT; (2) The publication was a case report or review article that lacked original data; or (3) Sufficient data for extraction were not available.

Data extraction and quality assessment

Eligibility was confirmed by full-text evaluation after the abstracts and titles were checked for relevancy by two independent experts. Any inconsistencies were settled via discussion or consultation with a third expert. Data regarding the immunosuppression protocols, DSA detection techniques, MFI thresholds, patient demographics, study features and reported results were extracted. Study quality was assessed using the Cochrane risk of bias tool for randomized controlled trials and the Newcastle-Ottawa Scale for observational studies, including selection, performance, detection, attrition, and reporting biases.

Statistical analysis

Statistical analyses were carried out in Comprehensive Meta-Analysis software (version 4; Biostat Inc., Englewood, NJ, United States). A random-effects model was applied to accommodate between-study heterogeneity, and odds ratios (ORs) with 95% confidence intervals (CIs) were estimated for dichotomous outcomes. Robustness of the results was verified using leave-one-out sensitivity analyses, in which each study was removed in turn to assess its influence on the overall conclusions.

Heterogeneity was quantified via the prediction interval statistic. Publication bias was appraised utilizing the visualization of funnel plots, which was supplemented by both Begg and Mazumdar’s rank correlation test and Egger’s regression test. The trim-and-fill approach was additionally employed to adjust for potential publication bias by estimating and incorporating hypothetical missing studies. Meta-regression was conducted to examine the association between DSA MFI levels and rejection risk. P < 0.05 was considered to indicate a statistically significant difference.

RESULTS

Study selection

The process used to select studies involved a comprehensive review of the literature on DSAs in LT. Initially, a total of 1550 studies were identified through database searches and the screening of references. Following deduplication and relevance screening of titles and abstracts, 340 studies were taken forward for full-text evaluation. Subsequently, 330 studies were excluded according to the predefined eligibility criteria. Ultimately, 10 studies[26-35] were included in the final analysis, which provided a robust dataset for evaluating the incidence, detection methods, thresholds, immunosuppression protocols and outcomes associated with DSAs in LT (Figure 1).

Figure 1 Preferred Reporting Items for Systematic Reviews and Meta-analyses flowchart depicting the study selection process. An illustration of the Preferred Reporting Items for Systematic Reviews and Meta-analyses flowchart outlining the selection process of the studies used in the study. The number of records identified through database searches, screened, assessed for eligibility and included in the final analysis are detailed. Reasons for exclusion at each stage are provided.

Characteristics of included studies

Table 1 indicated the findings regarding the DSAs in LT, incidence, detection methods, MFI thresholds, immunosuppressive protocols and clinical outcomes from the 10 studies that were included in this analysis. These findings suggested that DSAs were related to an elevated risk of AMR and acute rejection, although the impact on patient and graft survival varied between the different studies.

Table 1 Summary of studies on donor-specific antibodies in liver transplantation.

Ref.	Study characteristics	Number of patients	Detection method	MFI threshold	Immunosuppression	Outcomes
Del Bello et al[26], 2015	Incidence of de novo DSAs	152	Luminex	> 1000	CNI, MMF and steroids	In total, 14% of patients developed DSAs. Among these DSA-positive patients, 43% (equivalent to 6% of the total cohort) developed AMR
Del Bello et al[27], 2020	Preformed DSAs and outcomes	1788	Luminex (One Lambda, Immucor)	1000-5000	Tac, CNI, mycophenolic acid, mTOR inhibitors and steroids	High MFI DSAs associated with rejection
Kaneku et al[28], 2013	Role of de novo DSAs	749	Luminex	> 5000	CyA/Tac/SRL/MMF	Graft loss was predicted in 8.1% of all patients who developed de novo DSAs
Iacob et al[29], 2015	Humoral response and graft failure	174	Luminex	> 5000 (class II)	CNI-based	In total, 19.5% of patients with DSAs who had class II antibodies and C4d positivity experienced graft failure
O'Leary et al[31], 2014	DSAs and fibrosis in HCV	507	Luminex	> 5000	CNI, MMF and SRL	The presence of DSAs was associated with an increased risk of fibrosis and mortality
San Segundo et al[32], 2016	Long-term de novo DSAs	28	Luminex and C1q	> 3500	CNI-based	DQ DSAs and fibrosis
Song et al[33], 2015	DSAs in LDLT	219	Enzyme-linked immunosorbent assay	> 3500	Tac and basiliximab	Among DSA-positive patients, 14.6% had multiple DSAs, which was associate with a poor outcome
Taner et al[34], 2012	DSA course of 1 year	90	Luminex	> 2000	Triple therapy	Preformed DSAs were detected in 22.2% of patients, but they had a limited impact on graft outcomes
Vandevoorde et al[35], 2018	DSA prevalence and risk	297	Immucor LSA	≥ 1500	CNI, MMF and steroids	Preformed DSAs were present in 4.7% of patients and showed a limited clinical impact
Ogawa et al[30], 2023	DSAs in LDLT with desensitization	37	FlowPRA and Luminex	≥ 1000	Tac, MMF and Rituximab	After 1-year, 100% of patients survived

DSA: Donor-specific antibody; HCV: Hepatitis C virus; LDLT: Living donor liver transplantation; MFI: Mean fluorescence intensity; CNI: Calcineurin inhibitor; MMF: Mycophenolate mofetil; Tac: Tacrolimus; mTOR: Mammalian target of rapamycin; CyA: Cyclosporine A; SRL: Sirolimus; AMR: Antibody-mediated rejection; DQ DSAs: Donor-specific antibodies targeting human leukocyte antigen-DQ antigens.

Primary meta-analysis: Association between DSAs and AMR

However, as revealed in Figure 2A, the pooled analysis demonstrated no statistically significant association between DSAs and AMR, with an OR of 0.528 (P = 0.138) and a 95%CI: 0.227-1.228, which included the null value (OR = 1). Moreover, the prediction interval (0.046-6.060) highlighted uncertainty and heterogeneity across the included studies. This suggested that the clinical relevance of DSAs in LT may be influenced by a complex interplay of confounding variables, such as differences in immunosuppressive regimens, DSA characteristics (e.g., MFI and complement-binding capacity), and patient demographics.

Figure 2 Forest plot. A: A forest plot of the meta-analysis showing odds ratios and 95% confidence intervals for the association between donor-specific antibodies and antibody-mediated rejection. Odds ratios and 95% confidence intervals for each study, as well as the pooled odds ratio and prediction interval from the meta-analysis demonstrated in a forest plot; B: Leave-one-out sensitivity analysis of the association between donor-specific antibodies and antibody-mediated rejection. Odds ratios and 95% confidence intervals for each study, as well as the pooled odds ratio with one study removed from a leave-one-out meta-analysis. DSA: Donor-specific antibody; CI: Confidence interval.

Sensitivity analysis

A leave-one-out sensitivity analysis (Figure 2B) investigated the association between DSAs and AMR in LT. The findings demonstrated that omitting any single study did not change the overall finding, except for the study by Ogawa et al[30]. The exclusion of this specific study resulted in statistical significance (P = 0.026), suggesting that the overall finding was sensitive to the inclusion of this particular study and highlighting the potential impact of its unique characteristics or population on the pooled result.

Prediction interval analysis

As shown in Figure 3, the prediction interval analysis indicated that the mean effect size is 0.53 with a 95%CI: 0.23-1.23. This suggested that DSAs were not significantly related to rejection in LT on average, as the CI included 1 (the null effect). However, the prediction interval, which accounts for variability across studies and estimates the range of true effects in 95% of comparable populations, spanned from 0.05 to 6.06. This wide interval reflected substantial heterogeneity and uncertainty, indicating that DSAs may have a protective effect in a number of populations, while DSAs may increase the risk of rejection in other populations. This variability highlighted the need for further investigation into study-specific factors, such as the characteristics of DSAs, immunosuppressive protocols and the demographics of patients, which may influence the outcomes.

Figure 3 Distribution of true effect sizes for the association between donor-specific antibodies and antibody-mediated rejection. Mean effect size was 0.53 with a 95% confidence interval of 0.23-1.23. The true effect size in 95% of all comparable populations fell in the interval of 0.05-6.06.

Assessment of publication bias

The funnel plot (Figure 4) appeared symmetrical, indicating minimal small-study effects. Begg and Mazumdar’s test yielded a Kendall’s tau of 0.027 with a two-tailed P = 0.916, which suggested no statistically significant publication bias. Furthermore, Egger’s regression intercept was -0.350 (two-tailed P = 0.786), which supported the absence of significant asymmetry. These results indicated that the findings of this meta-analysis were unlikely to be influenced by publication bias.

Figure 4 Funnel plot assessing publication bias in studies investigating the association between donor-specific antibodies and antibody-mediated rejection. Possibility of publication bias in the meta-analysis was assessed using a funnel plot. The standard error was plotted against the log odds ratio for each study.

Furthermore, Duval and Tweedie’s trim-and-fill analysis identified one potentially missing study on the left side of the funnel plot. After inputting the missing study[31] using a random-effects model, the pooled OR shifted from 0.527 (95%CI: 0.226-1.228) to 0.388 (95%CI: 0.154-0.975). While the adjusted CI excluded 1, suggesting a possible association between DSAs and rejection. This result warrants cautious interpretation given the wide intervals and heterogeneity across the included studies.

Meta-regression analysis

The results of the meta-regression analysis (Figure 5) evaluated the impact of three covariates, sex (%), Model for End-Stage Liver Disease (MELD) score and age, on the log OR for rejection in LT. The overall model was statistically significant (Q = 11.53; df = 3; P = 0.009), indicating that at least one covariate significantly resulted in variance between the studies. Among the covariates, it was revealed that the MELD score had a significant negative association with rejection risk (coefficient: -0.328; P = 0.021), suggesting that increases in the MELD scores were associated with a lower risk of rejection. It was revealed that sex (%) had a non-significant trend (coefficient: -0.074; P = 0.079), indicating that sex did not influence rejection rates. Additionally, it was revealed that age did not have a significant association (coefficient: 0.121; P = 0.534). The model explained 65% of the total variance between studies (R² analog 0.65), with residual heterogeneity remaining moderate (I² = 46.28%; τ² = 0.8354). These findings indicated that the MELD score was a significant moderator of rejection risk, while sex and age did not appear to contribute to the variability in the outcomes across studies.

Figure 5 Meta-regression analyses of log odds ratio on covariates. Meta-regression analyses were used to investigate the association between log odds ratios and various covariates. A: Regression of log odds ratio on age; B: Regression of log odds ratio on sex; C: Regression of log odds ratio on Model for End-Stage Liver Disease score. Each plot includes a regression line (red line) with 95% confidence intervals (green lines). MELD: Model for End-Stage Liver Disease.

The complementary analysis (Figure 6) demonstrated that the total variance in true effects was 2.409, with 65% of this variance (1.574) explained by the model and 35% (0.835) remaining unexplained. The R² value of 0.65 indicated that the covariates included in the model accounted for a substantial proportion of the variability in the effect sizes across the studies. This suggested that the predictors used in the regression analysis, such as MELD score, sex or age, contributed to explaining the differences in the rejection rates among liver transplant recipients. However, the unexplained variance highlighted that additional factor (e.g., used medications and their dose) that were not included in the model may influence outcomes, which required further investigation.

Figure 6 Proportion of variance in true effects using the model. Proportion of total variance in true effects (2.409) that was explained (1.574) and not explained (0.835) by the model. The R² quantified the proportion of variance explained by the model. R²: Coefficient of determination.

Furthermore, the meta-regression analysis revealed a significant positive association between the MFI of DSAs and rejection risk in LT (coefficient: 0.0014; P = 0.0138; Figure 7A), suggesting that higher MFI levels were associated with increases in the risk of immune rejection. With 68% of the variance between studies explained (R² = 0.68), MFI may be a predictor of rejection, although the remaining heterogeneity (I² = 66.88%) indicated additional influencing factors such as immunosuppressive protocols and patient-specific variables. The scatterplot regression confirms this trend (Figure 7A), but confidence bands highlight uncertainty at extreme MFI values, emphasizing the need for larger studies to refine predictive thresholds.

Figure 7 Meta-regression analyses of log odds ratio on covariates. Meta-regression analyses were used to investigate the association between log odds ratios and various covariates. A: Regression of log odds ratio on mean fluorescence intensity; B: Regression of log odds ratio on donor-specific antibodies class ratio (class I/II). Each plot includes a regression line (red line) with 95% confidence intervals (green lines). MFI: Mean fluorescence intensity; DSA: Donor-specific antibody.

In the studies included in this analysis, heterogeneous MFI thresholds were used to define DSA positivity, ranging from ≥ 1000 to ≥ 5000. This variability could be influenced by differences in institutional protocols, Luminex assay calibration and clinical risk tolerance. Therefore, instead of applying a uniform cut-off, the MFI was modeled as a continuous variable in the meta-regression, so that this diversity was included and the dose-dependent effects were assessed. The meta-regression analysis results suggested that the DSA class ratio (I/II) was not a significant predictor of the log OR, which was demonstrated by P = 0.7959 and a non-significant coefficient (0.2854) (Figure 7B). Furthermore, the wide CI indicated uncertainty in the estimated effect. The high I² value (72.03%) suggested heterogeneity among the included studies, which suggested that additional factors may contribute to the observed variability. The R² analog value of 0.00 indicated that the model did not explain the variance between the studies.

Quality assessment

Based on the Newcastle-Ottawa Scale (NOS) criteria (Table 2), in the 10 studies, the exposures were determined (DSA status via validated immunological assays) and it ensured that outcomes were assessed with standard clinical or histopathological criteria, which fulfilled the NOS requirements for outcome quality. Differences in total NOS scores were mostly due to study design features (in particular, the presence of comparison groups and statistical control of confounders) instead of deficiencies in outcome ascertainment or follow-up. Studies that had more rigorous designs (such as prospective data collection, larger sample sizes or matched/adjusted comparisons) achieved a higher NOS total score, which indicated a higher methodological quality and a lower risk of bias in the findings of the study.

Table 2 Newcastle-Ottawa scale quality assessment of key studies on donor-specific antibodies in adult liver transplantation.

Ref.	Selection (0-4)	Comparability (0-2)	Outcome (0-3)	Total (0-9)
Del Bello et al[26], 2015	4	1	3	8
Del Bello et al[27], 2020	4	2	3	9
Kaneku et al[28], 2013	4	2	3	9
Iacob et al[29], 2015	4	1	3	8
Ogawa et al[30], 2023	4	2	3	9
O'Leary et al[31], 2014	4	1	3	8
San Segundo et al[32], 2016	4	1	3	8
Song et al[33], 2015	3	0	3	6
Taner et al[34], 2012	4	0	3	7
Vandevoorde et al[35], 2018	4	0	3	7

DISCUSSION

Although the liver is historically regarded as an “immune-privileged” organ owing to mechanisms such as the release of soluble HLA antigens, the presence of tolerogenic immune cells and its remarkable regenerative capacity; our findings indicate that high-MFI DSA remains a risk factor in LT. The main final pathway for DSA-mediated damage is chronic activation of liver sinusoidal endothelial cells (LSECs) mediated by DSA and the subsequent cascade it triggers. Mechanically, upon binding to HLA molecules on the surface of LSECs, DSA not only may activate the complement cascade but also, more critically, can recruit and activate natural killer cells via Fc receptors (e.g., CD16), initiating antibody-dependent cell-mediated cytotoxicity and causing direct endothelial injury[6]. This recurrent injury and inflammation drive LSECs toward a pro-fibrotic phenotype, leading to the sustained secretion of cytokines such as transforming growth factor-β, which ultimately initiates the process of perisinusoidal fibrosis[6]. While no statistically significant association between DSAs and AMR was found in the pooled analysis of this study, the wide prediction interval highlighted heterogeneity and clinical uncertainty between studies.

This ambiguity is reflected in the findings of Song et al[33] and Taner et al[34], which report a minimal impact of DSAs on short-term rejection rates. In contrast, O'Leary et al[31] and Kaneku et al[28] reported notable associations between persistent or high-MFI DSAs and chronic rejection or graft fibrosis. Additionally, a previous meta-analysis by Beyzaei et al[11] confirms that de novo DSAs are associated with a notably increased risk of graft loss and chronic rejection compared to transplant recipients without de novo DSAs (OR approximately of 3.6 and OR approximately of 6.4, respectively). These conflicting results suggest that the impact of DSAs is not uniform, but rather modulated by characteristics such as antibody class, intensity (MFI) and the immune environment of the patient.

The absence of a statistically significant association between DSAs and AMR in this pooled analysis may be explained by several biological and methodological factors. Firstly, the liver is known to have immunotolerant properties, such as Kupffer cell activity, secretion of soluble HLA and dual blood supply, that attenuate the pathological effects of DSAs, unlike those in other solid organ transplants[36]. Secondly, not all DSAs are equally pathogenic; and their impact depends on characteristics such as class (I vs II), MFI level and complement-binding capacity[37]. The studies included in this analysis varied in how DSAs were measured and classified (e.g., thresholds ranging from 1000 MFI to 5000 MFI), which may have reduced the pooled association. Thirdly, the definition and histopathological diagnosis of AMR in LT are not standardized, contributing to variability across studies. Finally, heterogeneity in immunosuppression regimens may have confounded results; and some patients with DSAs have received intensified therapy (such as Ogawa et al[30] and San Segundo et al[32]), which reduces the manifestation of rejection. These factors may have accounted for the absence of a statistically significant pooled effect despite underlying biological plausibility.

In this meta-regression analysis, a significant positive association between MFI and rejection risk was observed, which suggested that high-MFI DSAs (often interpreted as proxies for higher antibody burden or binding affinity) posed a greater threat to graft integrity compared to DSAs with lower MFI values. However, the MFI thresholds used among the included studies in the present meta-analysis varied. These discrepancies may have influenced how the DSA-associated rejection was detected and classified, and could partly explain the heterogeneity observed in the pooled findings of the study. While high thresholds (such as MFI ≥ 5000) were adopted in a number of studies to reflect clinically meaningful sensitization, more lenient or inclusive cut-off values (e.g., MFI ≥ 3000-4000) were used in other studies, which classified a broader range of antibody levels as positive. In the study, MFI was analyzed as a continuous variable in order to accommodate for these differences, which enabled an interpretation of the association between DSA strength and rejection risk. This aligns with the results reported by Goto et al[38] and O'Leary et al[31], in which DSAs with MFI > 10000 are associated with higher rates of fibrosis and chronic rejection compared to patients with lower-MFI or no DSAs. Furthermore, the study by Goto et al[38], demonstrates that pediatric living-donor liver transplant recipients with post-transplant DSA MFI values ≥ 9378 have notable graft fibrosis progression, assessed as meta-analysis of histological data in viral hepatitis ≥ F2 (a histologic grading system where F2 indicates significant fibrosis). By contrast, the presence of low-MFI DSAs may be insufficient to cause clinically meaningful injury, potentially explaining the non-significant findings in pooled estimates.

In the meta-regression analysis, the DSA class ratio (class I/II) did not significantly predict rejection risk. While class I DSAs are often transient post-transplant, class II, particularly HLA-DQ antibodies, are more persistent and are implicated in chronic injury[27,39]. Specifically, class II DSAs bind to MHC class II molecules, which are expressed in endothelial cells, and activate the classical complement pathway to cause persistent vascular endothelial injury, which in turn triggers microvascular occlusion and tissue fibrosis through the damage repair cycle. Although, our findings did not demonstrate a statistically significant difference between class I and class II DSA regarding rejection, this finding may be “false negative”. Some studies suggest that DSAs-mediated rejection may be attributed to different IgG subclasses (e.g., IgG3 has stronger complement-binding ability and IgG4 is often devoid of complement-fixing ability) and the specific sites of HLA mismatch[6,40]. However, these studies in our meta-regression analysis did not report these classifications, respectively. Therefore, in the future, we ought to refine more immunological dimensions (such as IgG subclasses and specific sites of HLA mismatch) so that to achieve more accurate risk stratification and analyze DSA heterogeneity.

Heterogeneity among the studies included in the analysis may also have reflected the variations in the DSA detection methods. While earlier investigations rely on complement-dependent cytotoxicity assays (such as Ogawa et al[30]), the Luminex-based solid-phase assays are utilized in recent reports (San Segundo et al[32] and Taner et al[34]). Especially, Ogawa et al[30] uses both methods to detect DSA, which may be a key factor for the extremely high heterogeneity. These methodological differences, combined with non-uniform MFI thresholds and timing of DSA measurements, have likely contributed to inconsistent results across the literature and the meta-analysis. For example, San Segundo et al[32] defined the threshold for positivity as > 3500 MFI. But, Taner et al[34] utilized a positive threshold as ≥ 2000 MFI. The guidelines from the 2016 Banff Working Group highlight the need for assay standardization and histopathologic criteria in order to reliably diagnose AMR in liver allografts[24].

Moreover, the meta-regression showed that higher MELD scores were inversely associated with rejection. The observed inverse association between the MELD score and rejection risk may reflect biologically and clinically plausible mechanisms. Patients with higher MELD scores, which indicate more advanced liver disease and that patients are immunocompromised, are often managed with more intensive immunosuppressive regimens (e.g., higher tacrolimus troughs or basiliximab induction in the early-stage post-transplant) to prevent complications. This may reduce both the cellular and antibody-mediated graft rejection[41]. Additionally, severe liver dysfunction is associated with immune dysregulation, including reduced circulating lymphocyte subsets and altered T-helper/T-regulatory cell ratios[42]. These immunophenotypic changes may decrease alloimmune responsiveness, further diminishing rejection risk. However, the downside is an increased susceptibility to infections and sepsis in recipients with a high-MELD score. This is demonstrated in cohorts with hepatocellular carcinoma in which higher MELD scores are associated with post-transplant septic shock and mortality[43]. Clinically, this suggests that MELD stratification may guide personalized immunosuppression: Patients with a high-MELD score may benefit from a lower level of immunosuppression in order to balance infection risk while preserving graft integrity, whereas patients with a low-MELD score may require vigilant rejection surveillance. Future studies should confirm these mechanistic insights using immune profiling and controlled immunosuppression trials based on MELD strata.

The sensitivity analysis revealed that the exclusion of Ogawa et al’s study[30] significantly altered the pooled risk estimates. The Ogawa et al’s study[30] focused on living donor liver transplant recipients. Notably, in this study, an aggressive desensitization protocol was implemented (including rituximab and plasma exchange) for patients with high-titer DSA. These high-risk patients had a survival rate of 100% after surgery. Without treatment, they often have a poor prognosis. Although the inclusion of these successful cases in the meta-analysis would reduce the overall hazard ratio, this change conveys the core message that DSA does have a risk, but this risk can be modified by interventions.

These findings suggest that post-transplant monitoring should have a more nuanced, personalized approach. While routine DSA surveillance is standard in renal and cardiac transplantation[44,45], its role in liver recipients is still being debated. The results of the study suggested that the quantitative assessment of DSA strength (such as the use of MFI values), instead of the binary DSA status, may offer superior risk stratification. For example, identifying de novo DSAs may prompt early biopsy or personalized immunosuppression adjustments, as is already practiced in kidney transplant protocols[46]. Current expert consensus supports monitoring DSA in solid organ transplant recipients, with MFI thresholds - commonly ranging from 1000 to 1500 or higher - used to define antibody positivity and guide clinical decision-making, particularly in kidney transplantation[47].

Despite these insights, several limitations of the study should be paid attention to. The studies included in the study had heterogeneity in terms of their designs, sample sizes, DSA detection platforms and immunosuppressive regimens. The sensitivity of the findings to a single study emphasized the fragility of the pooled estimate and larger[30], prospective studies should be used in future experiments. Furthermore, the retrospective nature of a number of the included studies limited the causal inference and introduced the risk of residual confounding factors, such as differences in immunosuppressive regimens or patient comorbidities. Additionally, the overall sample size of the included studies was limited. Only 10 studies met the inclusion criteria. This small sample size may have reduced the statistical power of the meta-analysis, limited the generalizability of the findings and contributed to the wide prediction interval observed. Moreover, variability in pediatric vs adult cohorts requires further investigation. Among 67 pediatric liver transplant recipients, a study by Melere et al’s reports[48] a notably lower rejection-free survival at both 12 (76% vs 100%) and 24 (58% vs 95%) months in those with de novo DSAs, indicating a nearly ten-fold higher rejection risk compared with those that are DSA-negative. Additionally, categorical subgroup analyses (such as stratification by DSA type or MFI level) were not made in the study due to the limited number of eligible studies and the sparse distribution of relevant covariates across the included studies. Subgroups with less than three studies are considered unreliable for meta-analytic comparison due to the increase in the risk of false-positive or false-negative findings. Therefore, meta-regression analyses were conducted, which were more appropriate due to the limited number of studies included. Through such analysis, the continuous modeling of covariate effects could be completed. This approach enabled the study to investigate the potential moderators, such as MFI, MELD score and DSA class ratio, while the statistical power was maintained and the arbitrary dichotomization of variables was avoided. However, the lack of traditional subgroup analyses limited interpretation, and it might be possible to make more robust stratified analyses through future meta-analyses with larger study pools. Finally, the lack of consensus regarding the MFI thresholds for defining clinically notable DSAs across studies may have limited comparability and complicated the pooled interpretation. Future studies should aim to harmonize DSA reporting standards in order to facilitate more robust cross-study comparisons.

Future research should prioritize standardization of DSA assays, rigorous reporting of MFI thresholds and inclusion of immunological and clinical covariates, such as HLA mismatches, antibody subclasses and complement-binding capacity. The integration of molecular biomarkers, such as gene expression profiles or circulating cell-free DNA, may further improve the precision of rejection risk prediction.

Based on the findings of this study, DSA monitoring in liver transplant recipients is recommended to include both qualitative (presence/absence) and quantitative (MFI-based) assessments using standardized Luminex-based platforms. And some studies suggest that MFI thresholds of > 2000 may require follow-up, while levels > 10000 should prompt consideration of early biopsy or immunosuppression adjustment[24,27,28,31]. Meanwhile, it is suggested that routine monitoring ought to be prioritized for high-risk groups (such as those with high panel reactive antibody, those undergoing re-transplantation, or those with specific HLA mismatches). Besides, for patients with low MFI DSA and stable liver function, an event-driven approach is more appropriate. DSA testing should be performed by specific clinical events, such as worse liver functions, acute cellular rejection or evidence of microvascular inflammation. Additionally, class II DSAs and persistent or rising MFI values should be considered as higher risk and may justify preemptive clinical action. Harmonized reporting of DSA characteristics, including MFI, complement-binding status (such as C1q) and temporal dynamics should be carried out, in order to facilitate risk stratification and individualized management protocols in future clinical practices.

CONCLUSION

This study demonstrated that the impact of DSAs on immune rejection after LT was context-dependent. While the overall association with AMR was not statistically significant, elevated antibody intensity, as reflected by higher MFIs, was significantly associated with an increased risk of rejection. Moreover, patients with higher MELD scores appeared less susceptible to rejection, likely due to more intensive immunosuppression. These findings suggested the clinical relevance of incorporating quantitative antibody profiling into post-transplant monitoring, in order to enable risk stratification and individualized immunosuppressive management. Furthermore, the results of this study suggested that standardized, MFI-guided DSA monitoring protocols should be developed, in order to optimize post-transplant immunosuppressive strategies.