High-titer autoantibodies are a hallmark of autoimmune liver diseases. Specifically, antinuclear antibodies (ANA) and anti-mitochondrial antibodies (AMA) are commonly found in primary biliary cholangitis (PBC), where they can directly attack bile duct epithelial cells, leading to chronic cholestasis and fibrosis (9). Anti-liver-kidney microsomal type 1 antibodies (anti-LKM1) are associated with autoimmune hepatitis (AIH), and their presence suggests T cell and B cell-mediated hepatocyte damage. ScRNA-seq is being applied to directly dissect the pathological role of B cells in autoimmune liver disease. For example, a study integrating multi omics methods performed scRNA-seq on immune cells of AIH patients, which not only confirmed the overall expansion of B cell population in the peripheral blood of patients, but more importantly, identified a specific AIF1+ B cell subset that was significantly enriched in AIH. This study directly correlated specific B cell subsets (AIF1+ B cells) with the disease status of AIH through single-cell resolution data, providing a new entry point for understanding the cellular immune abnormalities behind the production of autoantibodies. This demonstrates the powerful ability of scRNA-seq in revealing the complex immune microenvironment of the autoimmune liver, especially the heterogeneity of B cells (10).

In patients with chronic viral hepatitis, such as hepatitis C, the detection rate of autoantibodies is significantly elevated and correlates with disease activity and treatment response (11). Studies indicate that approximately 20-30% of patients with chronic hepatitis C (CHC) have detectable ANA or anti-LKM1 in their serum, suggesting these antibodies may be directly induced by HCV triggering an autoimmune response, rather than being solely caused by interferon therapy (12). Similarly, chronic HBV infection can lead to liver fibrosis and cirrhosis, during which autoantibodies produced by B cells may exacerbate liver injury and disease progression.

In-depth investigation of B cell function reveals that in chronic hepatitis B (CHB) or CHC, CD19⁺ memory B cells, as key players in hepatic immune memory, are enriched within the liver and participate directly in viral clearance by secreting antiviral antibodies (e.g., anti-HBV or anti-HCV antibodies)(Table 1). It is important to note that the long-term survival of these cells depends on specific signals in the liver microenvironment such as BAFF and TACI, and their functional state is closely related to the efficiency of the patient’s immune response (13). Studies combining scRNA-seq and B cell receptor (BCR) sequencing have found that clonal expansion of CD19⁺ memory B cells is directly linked to dynamic changes in antibody levels in patients with chronic hepatitis (14). Furthermore, dendritic cells (DCs), acting as a critical bridge between innate and adaptive immunity, also play a central role in CLDs. In chronic hepatitis, CD11c⁺ DCs recruit and activate T cells and B cells by expressing the chemokine CXCL13 and costimulatory molecules like CD80/CD86, thereby guiding B cells to migrate into tertiary lymphoid structures (TLS) to promote the production of antiviral antibodies. This mechanism is crucial for maintaining liver immune homeostasis (15). TLS are transient immune cell aggregates formed in non-lymphoid tissues, participating in local immune responses. They are composed of B cells, T cells, dendritic cells, and stromal cells, functioning similarly to lymph nodes, and are commonly found in chronic inflammation, infection, or tumor microenvironments.

Analysis of T cell populations shows that in patients with PBC, the CD4⁺ T cell subset shifts, with the expansion of pro-inflammatory Th1-like cells closely associated with bile duct epithelial cell injury(Table 1). These Th1-like cells are primarily confined to the liver tissue, and their numbers positively correlate with disease severity. In fact, multiple T cell subsets exhibit enhanced inflammatory responses, particularly pronounced in Th1-like cells, indicating their major pro-inflammatory role in PBC pathogenesis. Notably, although the number of regulatory T cells (Tregs) increases in PBC patients, their functional genes display a paradoxical activation state, involving these cells in both immunosuppressive and immune-activating processes. This results in Tregs being unable to effectively perform their traditional immunosuppressive functions, suggesting that Treg dysfunction may be a key factor in the breakdown of immune tolerance (9). Additionally, liver sinusoidal endothelial cells (LSECs), upon stimulation by liver injury, upregulate MHC-II molecules and present antigens to CD4+ T cells. This process indicates that LSECs possess antigen-presenting capabilities and can activate autoreactive T cells, thereby promoting disease progression (16). On the other hand, in HCC, infiltrating γδ T cells exhibit LAG3-dependent dysfunction. Overexpression of LAG3 may contribute to immunometabolic dysregulation by suppressing the metabolic phenotype of effector T cells (17).

In recent years, scRNA-seq has provided a high-resolution perspective for analyzing the function and mechanisms of specific T cell subsets in CLDs, revealing the central roles of various gene-positive T cell populations at different disease stages. In chronic viral hepatitis (e.g., CHB or CHC), IFNγ-positive CD8+ T cells are the core effector cells of antiviral immunity (Table 1). They directly kill infected hepatocytes by secreting IFNγ and induce the activation of type I interferon signaling pathways (e.g., ISG genes). However, persistent viral infection leads some IFNγ+ CD8+ T cells into a functionally exhausted state (e.g., PD-1+ Tim-3+), resulting in significantly reduced secretory capacity, thereby promoting viral persistence and liver injury (18). Combining scRNA-seq with T cell receptor (TCR) sequencing has confirmed that their clonal expansion is directly associated with the response rate to antiviral therapy, providing molecular markers for monitoring treatment efficacy (19). Furthermore, RNA velocity and lineage tracing tools have become indispensable for dissecting the dynamic differentiation trajectories of hepatic immune cells. By modeling the splicing dynamics of pre-mRNA and mature mRNA, RNA velocity predicts the future functional state of cells, revealing the transition of naive T cells to exhausted T cells in chronic viral hepatitis or the differentiation of monocytes into pro-fibrotic SAMs during liver fibrosis. Combined with TCR/BCR clonal tracing, these tools have identified clonally related T cell subsets that transition between anti-inflammatory and pro-inflammatory states across disease stages, providing a dynamic perspective on immune cell plasticity in CLDs. For instance, lineage tracing has confirmed that CXCR6+ NKT cells originate from a common precursor pool and switch their functional phenotype in response to persistent liver injury, validating their dual role in fibrosis progression (20). Similarly noteworthy, in the progression of liver fibrosis in CHB patients, CXCL13-positive T cells such as T follicular helper (Tfh) cells and memory T cells guide T cell migration into the liver via secretion of CXCL13 and expression of its receptor CXCR5 (Table 1). These cells produce high levels of HBV-specific interferon (IFN)-γ and IL-21, possess stronger antiviral capacity, promote the formation of local immune memory and the durability of antiviral immunity, and may indirectly influence the production of anti-fibrotic antibodies by recruiting B cells to form germinal centers (21).Overall, these findings not only deepen the understanding of immune mechanisms in CLDs but also provide a theoretical basis for immunotherapy strategies targeting specific T cell subsets (e.g., PD-L1+ TAMs, IL-17A+ γδ T cells, IFNγ+ CD8+ T cells, CXCL13+ T cells, and CXCR6+ NKT cells). Furthermore, research has identified that CD4⁺ FOXP3⁺ T cells, a core Treg subset, play a pro-tumor role by suppressing the immune microenvironment in HCC progression (Table 1): these cells are specifically enriched in the carcinomatous areas of the liver, where they impair anti-tumor immune responses through their classic immunosuppressive functions. They interact with PD-L1-expressing LAMP3⁺ dendritic cells and are selectively retained by the tumor microenvironment, collectively maintaining an immunosuppressive environment favorable for tumor growth (22).

During the progression of CLDs, immune cell dysfunction is closely linked to the driving mechanisms of liver injury and fibrosis. Macrophages, as central immunoregulatory cells, exhibit a dual role through abnormal polarization states: on one hand, under hepatocyte activation conditions, macrophages can polarize towards the pro-inflammatory M1 phenotype, directly exacerbating liver inflammation and promoting fibrosis via upregulation of signaling pathways like YAP/TAZ/CYR61 (Table 1); on the other hand, during the transition from metabolic dysfunction-associated steatotic liver disease (MASLD) to cirrhosis, macrophages can polarize towards the M2 phenotype, accelerating liver fibrosis and cirrhosis by secreting prostaglandin E2 (PGE2), which binds to the EP4 receptor, further activating HSC autophagy and proliferation (16) (Table 1). In summary, this dynamic imbalance in M1/M2 polarization not only reflects the contradictory role of macrophages in liver disease but also reveals their complex regulation of the liver microenvironment across different pathological stages.

The in-depth application of scRNA-seq has further refined our understanding of macrophage heterogeneity. Recent studies using scRNA-seq in CLDs have revealed the critical roles of several specific gene-positive macrophage subpopulations, providing new perspectives on the complexity of the liver immune microenvironment. Firstly, in metabolic dysfunction-associated steatohepatitis (MASH), CX3CR1⁺CD63⁺ liver portal-associated macrophages (LPAMs) have garnered attention for their unique anti-inflammatory and neuroprotective functions (Table 1). These macrophages delay MASH inflammatory progression by inhibiting neutrophil infiltration; simultaneously, they closely interact with hepatic sympathetic nerve endings, maintaining the structural integrity of sympathetic nerves and providing neural support for liver immune homeostasis. Their loss leads to increased inflammatory granulocyte infiltration and exacerbated sympathetic neurodegeneration (23). Secondly, in the fibrotic stage, CD206⁺ macrophages display functional duality: they maintain local immune homeostasis by phagocytosing apoptotic cells and debris, yet they are significantly enriched in fibrotic areas, express the typical M2 macrophage marker CD206, and drive HSC activation and collagen deposition by secreting factors like Thbs1, which binds to integrin receptors on HSCs, highlighting their potential as interventional targets (24) (Table 1). Furthermore, scar-associated macrophages (SAMs) are a monocyte-derived population expressing TREM2/CD9 that specifically expands in fibrotic regions (Table 1). They directly activate HSCs by secreting factors like SPP1 and PDGFB, promote collagen deposition, and form interaction networks with other cells within the scar, synergistically driving fibrosis progression. This conserved cell population offers a new target for anti-fibrotic therapy (7). Collectively, these studies indicate that macrophage subsets with different genetic markers have distinct functional division at various stages of CLD (e.g., MASH, fibrosis, cirrhosis). They may delay disease progression through anti-inflammatory or neuroprotective actions, or exacerbate pathological damage due to pro-fibrotic effects or dysregulated immunomodulation, providing important rationale for developing targeted macrophage therapies.

Meanwhile, abnormal activation of other immune cells should not be overlooked. For instance, in alcoholic hepatitis, neutrophil infiltration and their ability to form neutrophil extracellular traps (NETs) have been clearly associated with the degree of liver injury (Table 1). Further analysis shows this heterogeneity is manifested as high-density and low-density neutrophils possessing highly activated or exhausted transcriptomic signatures, respectively, further highlighting the dysregulated crosstalk among multiple immune cell types within the CLD immune microenvironment (25). Therefore, whether it’s the skewed polarization of macrophages or the aberrant activation of neutrophil functions, the dysregulation of these immune cells collectively contributes to the persistent worsening of liver fibrosis and drives cirrhosis towards irreversible stages.

While scRNA-seq reveals cellular heterogeneity, understanding the spatial distribution and interactions of these cells within their native tissue context is equally crucial. Spatial transcriptomics has emerged to directly address the lack of spatial context in single-cell technologies by capturing transcriptome-wide data within preserved tissue architecture (26). By integrating whole transcriptome data with spatial location information, spatial transcriptomics overcomes the spatial context limitation of traditional sequencing methods, enabling in situ analysis of the liver immune microenvironment in two or even three dimensions. This approach has proven particularly powerful in CLDs, where it can delineate disease-specific transcriptional landscapes within anatomical structures such as fibrotic septa (27). Consequently, this technology can not only precisely map the distribution patterns of immune cells in structural regions like liver lobules, portal tracts, and fibrous septa but also reveal region-specific immune-stromal interaction networks. For example, by analyzing the spatial proximity and ligand-receptor pair expression between macrophages and activated HSCs at the fibrotic front, it can elucidate the local transmission mechanisms of pro-fibrotic signals. Indeed, integrative single-cell and spatial analyses have been indispensable for defining the cellular niche that drives fibrosis in human cirrhosis (7). Simultaneously, the technology can identify the coupling between cell states and spatial locations within micro-regions like germinal centers or inflammatory foci, providing an indispensable technical perspective for understanding the regionalized regulation and structural remodeling of immune responses in CLDs as demonstrated in the spatial mapping of immune-stromal interfaces in primary sclerosing cholangitis (28). In recent years, the advent of high-resolution spatial transcriptomics platforms like Visium HD and Xenium has achieved subcellular resolution, allowing clear distinction of transcriptomic features from LSECs, hepatocytes, and infiltrating immune cells. Furthermore, multi-omics integration technologies like DBiT-seq can simultaneously obtain transcriptome and proteome information from the same tissue section, providing a powerful tool for spatially resolved molecular phenotyping of liver immune cells (29). Notably, the rapidly evolving multiplexed spatial platforms—including MERFISH, seqFISH+, and CosMx SMl—have further pushed the boundaries of spatial resolution and gene detection throughput beyond conventional spatial transcriptomics. These platforms enable quantitative mapping of thousands of target genes at subcellular resolution, allowing precise delineation of cell-cell communication networks within heterogeneous liver microenvironments that were previously undetectable. Complementing these transcriptomic advances, spatial epigenomics technologies such as CUT & Tag-spatial and ATAC-spatial integrate spatial localization with epigenetic signatures, uncovering region-specific chromatin accessibility and histone modification patterns in key cell types. This integration bridges the gap between spatial gene expression and its regulatory mechanisms, offering deeper insights into how epigenetic remodeling shapes the regionalized immune response in CLDs like MASH and HCC. While scRNA-seq has greatly advanced the dissection of immune cell heterogeneity in CLDs, this technology relies heavily on tissue dissociation. Notably, fibrotic/cirrhotic liver samples are characterized by tissue stiffness and extensive extracellular matrix(ECM) deposition (30), which may introduce significant methodological biases. Additionally, immune cell dropout and insufficient capture efficiency of rare cell subsets could obscure the true features of the immune microenvironment, and these limitations require careful consideration in result interpretation.

While single-cell and spatial transcriptomics have greatly deepened our understanding of the in vivo immune microenvironment, causal validation of mechanisms and facilitating clinical translation still require in vitro models that highly mimic the complex structure and function of the human liver. Organoid technology demonstrates its irreplaceable value in meeting this need. Organoid technology involves culturing human liver cells in vitro to simulate tissue structure and function. Derived from patient-specific stem cells or primary cells, organoids retain key liver structures, such as biliary networks and lobular polarity, as well as critical functions like the expression of liver-specific metabolic enzymes and bile secretion, thereby enabling more authentic modeling of the disease microenvironment. Furthermore, this technology avoids the species-specific functional differences between traditional mouse models and the human liver, facilitating direct study of human-specific immune mechanisms. Additionally, organoids can be cultured at scale, making them suitable for large-scale drug library screening and personalized drug testing, for instance, to evaluate the effects of different adjuvants on immune responses.

Tissue dissociation is an indispensable prerequisite for scRNA-seq but introduces inherent, disease-specific biases—particularly in fibrotic or cirrhotic liver tissues, the core research objects of CLD. Excessive ECM deposition and tissue stiffening in these samples severely hinder the penetration of dissociation enzymes such as collagenase and dispase, leading to two critical artifacts that distort the true immune microenvironment. Selective cell loss is a major issue: activated HSCs, KCs tightly adherent to sinusoidal endothelial cells, and dendritic cells (DCs) embedded in fibrotic septa are prone to incomplete dissociation, resulting in underrepresentation of these key immunoregulatory cells in scRNA-seq datasets and overestimation of easily dissociated cells like circulating T cells (45). Additionally, prolonged dissociation time induces stress responses in immune cells, upregulating genes related to inflammation and apoptosis, leading to misclassification of “stress-induced activated cells” as “disease-specific activated cells” and confounding interpretations of in vivo immune cell function. Comparative studies show that dissociation protocols optimized for normal liver tissue yield a lower recovery rate of tissue-resident macrophages in cirrhotic samples compared to fibrosis-tailored protocols, highlighting how the lack of disease-specific standards exacerbates result heterogeneity. Notably, neutrophils—critical for hepatic inflammation in alcoholic hepatitis and MASH—exhibit distinct recovery biases: their fragile cytoplasmic structure and rapid activation upon tissue manipulation cause significant lysis during dissociation, with recovery rates lower than in normal liver tissue (46). This uneven recovery not only underestimates their proportion in the intrahepatic immune microenvironment but also obscures functional heterogeneity, further distorting understanding of their role in CLD pathogenesis.

Immune cell dropout—where low-abundance transcripts or fragile immune cells fail to be captured by sequencing platforms—poses a major challenge for scRNA-seq-based CLD research, especially in identifying rare but functionally critical subsets. LAMs and DNT cells account for only 1–3% in the peripheral blood and lymphoid organs, and their detection relies heavily on high RNA capture efficiency (47). Studies using different scRNA-seq platforms (e.g., 10x Genomics vs. Smart-seq2) report conflicting frequencies of these subsets, with Smart-seq2 showing a higher DNT cell detection rate due to improved transcript coverage, suggesting platform-specific limitations may lead to underestimation or omission of key rare populations (48). Another issue is the misclassification of circulating vs. tissue-resident cells: peripheral T cells and other circulating immune cells are more prone to detachment during tissue processing, leading to their overrepresentation in scRNA-seq datasets. This bias is amplified in cirrhotic livers with increased vascular permeability, making it difficult to distinguish truly tissue-resident immune cells from transient circulating cells and misleading interpretations of “liver-specific immune responses.

Spatial transcriptomics addresses the lack of spatial context in scRNA-seq but faces inherent technical constraints and disease-related tissue damage in CLD applications. The spatial resolution bottleneck remains unresolved: even high-resolution platforms (e.g., Visium HD, Xenium) have spot sizes encompassing 2–5 cells in dense regions like portal tracts, making it impossible to assign transcripts to individual cells or verify direct cell-cell contacts. For example, predicted interactions between macrophages and HSCs at the fibrotic front may merely reflect co-localization within the same spot rather than true functional crosstalk. RNA degradation is an underappreciated issue exacerbating biases: fibrotic and cirrhotic tissues require prolonged fixation and decrosslinking due to dense ECM, accelerating the degradation of short-lived, immune-regulatory transcripts. RNA degradation is an underappreciated issue exacerbating biases: fibrotic and cirrhotic tissues require prolonged fixation and decrosslinking due to dense ECM, accelerating the degradation of short-lived, immune-regulatory transcripts. This degradation pattern introduces unavoidable bias when quantifying immune cell activation states, which rely on accurate detection of short-lived cytokines and chemokines (spatial transcriptomics) (27). Tissue integrity compromise further limits utility: severe architectural distortion and ECM cross-linking in fibrotic/cirrhotic tissues make intact sectioning difficult, resulting in fragmented samples and lost spatial context in fibrotic septa, which directly impairs the accuracy of gene expression localization. Additionally, spatial transcriptomics cannot distinguish between membrane-bound and soluble ligands, leading to ambiguous interpretation of paracrine vs. autocrine signaling—for instance, predicted PD-L1/PD-1 interactions in HCC may reflect soluble PD-L1 secretion rather than direct cell-cell binding.

L-R interaction prediction is a core tool for inferring intercellular communication but is frequently over-interpreted in CLD research, leading to misleading conclusions. Most studies rely on computational tools such as CellPhoneDB and NATMI, which use gene co-expression as the sole criterion without validating protein expression, binding affinity, or functional relevance. For example, the predicted MDK-LRP1 interaction between fibroblasts and macrophages in fibrotic niches has not been confirmed by co-immunoprecipitation or neutralization experiments, leaving open the possibility of false-positive predictions. Conflicting results across studies further highlight the need for critical reflection: regarding Tregs in PBC, some studies report Treg dysfunction, while others suggest Tregs retain immunosuppressive capacity but are sequestered by stromal cells—a discrepancy likely driven by differences in tissue dissociation protocols and L-R prediction thresholds (49, 50). Similarly, the pro-fibrotic role of SAMs is well-established in human cirrhosis (51), but murine models show inconsistent results (52), with some reporting anti-fibrotic SAM subsets, highlighting species-specific limitations and the need for cross-validation. Notably, most predicted interactions lack functional validation via targeted approaches such as neutralizing antibody blocking, CRISPR-mediated gene knockout, or in vitro co-culture assays. This gap between computational inference and experimental confirmation risks misidentifying non-physiological interactions as therapeutic targets, hindering clinical translation.

Current liver organoid systems, despite their value in disease modeling, fail to replicate key physiological features of the human liver, severely restricting their translational relevance. Vascularization is absent in most conventional organoids, precluding the simulation of immune cell recruitment via the circulatory system, ischemia-reperfusion injury, and vascular-immune crosstalk that drives fibrosis progression. Bile-flow physiology is also not recapitulated, as organoids lack intact biliary excretion pathways—limiting the modeling of cholestatic liver diseases where bile stasis-induced immune activation is a core pathogenesis. Immune-cell longevity is compromised: primary immune cells co-cultured with organoids survive shortly, making it impossible to study long-term immune responses such as chronic exhaustion or immune memory formation. Additionally, metabolic zonation—the regional specialization of hepatocytes—is absent, failing to replicate the spatial heterogeneity of metabolic-immune crosstalk that regulates liver inflammation and fibrosis. These physiological limitations mean organoid-based findings may not accurately reflect in vivo immune microenvironment dynamics.

Multi-omics integration is hailed as a solution to unravel complex immune regulatory networks but introduces significant interpretational gaps in CLD research. Discrepancies between omics layers are common: scRNA-seq may detect upregulation of pro-inflammatory genes in macrophages (53), but epigenomic data may show reduced chromatin accessibility at these loci—indicating transcriptional priming rather than active expression—yet most studies fail to reconcile such discrepancies, leading to oversimplified conclusions about cell function. Integrating multi-omics datasets is also computationally intensive due to the high dimensionality of single-cell data and the need for complex algorithms for batch correction, cell matching, and interaction inference, which requires advanced computational resources and specialized expertise not accessible to most research teams. The lack of standardized computational tools further exacerbates the issue: different studies use distinct algorithms for batch correction and integration, leading to variable cell clustering outcomes and hindering cross-study comparisons.

Core computational tools for CLD multi-omics analysis include functional modules tailored to different data types: single-cell analysis frameworks for scRNA-seq preprocessing, dimensionality reduction, and cell-type annotation, with Harmony and scVI critical for batch correction; spatial analysis suites for spatial clustering, cell-type deconvolution, and immune-stromal interaction mapping; L-R inference tools for constructing communication networks; and trajectory/lineage tools for modeling cell state transitions. However, key challenges persist: batch correction and donor variability confound datasets, sparse sampling from scarred tissue compromises trajectory inference and L-R analysis, inconsistent cell-type annotation stems from the lack of a unified liver-specific immune reference atlas, and bridging in vivo and in vitro data remains hindered by experimental system differences.

Building on the critical reflection of these limitations, targeted future directions are needed to advance CLD immunomicroenvironment research toward precision and reproducibility. Technological innovations should focus on developing high-sensitivity, low-damage single-cell capture technologies to address small biopsy sizes and challenges in isolating intrahepatic immune cells. For spatial transcriptomics, optimizing tissue processing protocols for fibrotic/cirrhotic samples—including improved fixation and RNA protection methods—will enhance gene localization accuracy. Organoid model optimization should prioritize functional enhancement and standardization of co-culture conditions to improve reproducibility.

Multi-technology integration strategies will be key: combining scRNA-seq and spatial transcriptomics to construct 3D maps of “cell type - spatial location - functional state,” and integrating metabolomics with immunology to reveal how metabolic reprogramming regulates the immune microenvironment. Computationally, developing donor-aware integration tools that account for clinical covariates, CLD-tailored sparse data imputation methods, and a unified liver-specific immune reference atlas will standardize analyses and enable cross-study comparisons. Finally, strengthening the link between computational predictions and experimental validation—via functional assays of key L-R interactions and cell trajectories—will ensure that discoveries translate reliably to clinical applications, advancing personalized immunotherapy for CLD.