ALF develops from rapid liver function loss triggered by various acute injuries, including hepatotoxic drugs, immune-mediated attacks or viral infection-induced rapid hepatocellular necrosis (24,25). The condition manifests when hepatocellular death exceeds the regenerative capacity of the liver, primarily through necrosis, apoptosis and necrotizing apoptosis (26). Apoptotic hepatocytes interact with Kupffer cells (KCs), newly infiltrated bone marrow-derived macrophages and hepatic stellate cells (HSCs). However, extensive hepatocyte death overwhelms the clearance capacity, preventing liver function restoration (27). The inflammatory response distinguishes necrotic from apoptotic cell death. Necrotic cell rupture triggers inflammation through the release of intracellular contents, notably the NLR family pyrin domain containing 3 (NLRP3) inflammasome. This well-established human inflammasome, when aberrantly activated, associates with various ALF types and induces different programmed cell death forms (Fig. 4) (28).

The predominant mechanism of ALF cell death induced by viral infection involves the activation of death receptors. Blood samples from patients with ALF showed significantly elevated levels of death receptors and ligands, including CD95L, TNF-α and TNF receptor (29-32). Studies using in vitro-constructed TNF-α and anti-CD95 mouse models have demonstrated toxicity patterns similar to those observed in patients with ALF (33,34). Additionally, HCV-infected human livers demonstrated increased TRAIL receptor expression and enhanced susceptibility to TRAIL-induced apoptosis (31).

Unlike viral infection, drug-induced liver injury primarily involves signaling pathways originating from mitochondria (35). The mechanism of acetaminophen (APAP)-induced ALF cell death primarily involves metabolic and adaptive mechanisms regulating toxicity in the cytochrome P450 system (36). Mitochondrial autophagy, a selective form of autophagy, eliminates damaged mitochondria through the interaction between mitochondrial and autophagic mechanisms. This process occurs through ubiquitin-dependent or ubiquitin-independent pathways, characterized primarily by PINK/Parkin-mediated and light chain 3 interacting region-domain receptor-mediated pathways, respectively (37).

Mitochondrial autophagy serves to attenuate local inflammatory responses and tissue necrosis by eliminating damaged or excess mitochondria, maintaining mitochondrial homeostasis, and inhibiting pro-inflammatory factor secretion triggered by inflammatory vesicles such as NLRP3 (38,39). During ALF onset, injured hepatocyte mitochondria release damage-associated molecular patterns and secrete inflammatory factors IL-1β and TNF-α. This leads to increased reactive oxygen species (ROS) production and ATP depletion, followed by mTOR activity inhibition and enhanced autophagic flux. Autophagic vesicles accumulate around necrotic foci, limiting necrosis expansion into normal hepatic areas. Mitochondrial autophagy proteins, activated by ectopic or ubiquitinated mitochondrial proteins, facilitate damaged mitochondria removal, ROS reduction, inflammation decrease and hepatocyte repair (40). However, insufficient activation of mitochondrial autophagy during disease onset impairs timely removal of damaged mitochondria and necrotic hepatocyte repair, resulting in increased local and systemic inflammation and ultimately irreparable massive liver parenchyma necrosis (41). In APAP-induced ALF, Parkin protein inhibition/knockdown significantly reduces mitochondrial autophagy, leading to NLRP3 inflammatory vesicle accumulation and increased APAP-induced hepatotoxicity (42). By contrast, mitochondrial autophagy activation removes ROS-damaged mitochondria, inhibits IL-1β release and alleviates APAP-induced ALF symptoms (43). This evidence demonstrates the crucial role of mitochondrial autophagy in liver protection and homeostasis maintenance, while its dysfunction leads to damaged mitochondria accumulation, disrupted mitochondrial homeostasis and ALF progression (Fig. 5) (41).

Previous evidence has suggested that extensive extracellular matrix (ECM) deposition and scarring from persistent HSC activation impairs tissue regeneration by hepatocytes (44). However, another study showed that preventing fibrosis through activated HSC depletion in ALF mouse models results in more severe liver injury and reduced survival (45). Fibrosis is an intrinsic injury response that maintains organ integrity during extensive necrosis or apoptosis, suggesting its beneficial role during acute tissue injury (46). Previous studies have proposed that ALF-related fibrosis may possess a protective function. The acute production of collagen and fibrosis onset in ALF may represent an intentional physiological process (47-49). Further investigation is needed to determine whether fibrosis formation could serve as an intervention target in ALF animal models.

From a clinical perspective, the primary cause of mortality in ALF is systemic inflammatory response syndrome, which results from excessive inflammatory responses, leading to concentrative renal anemia and multi-organ failure (50). Disruptions in the innate immune system constitute the principal mechanism leading to ALF (51). Notably, monocyte-macrophages, serving as crucial effector cells in both natural and intrinsic immunity, generate numerous proinflammatory and anti-inflammatory factors, playing a central role in ALF initiation and progression. As hepatic macrophages, both KCs and monocyte-derived macrophages regulate the hepatic inflammatory response. During ALF onset, hepatic KCs become activated by injury-associated molecular patterns released from damaged hepatocyte mitochondria, subsequently activating the NF-κB signaling pathway to produce inflammatory cytokines and reactive oxygen clusters (52). This leads to a substantial increase in hepatic KCs, partially through KC proliferation and division, and partially through circulating monocytes that migrate to injured hepatic areas via monocyte chemoattractant protein-1 receptor signaling. These monocytes differentiate into KCs, secreting additional inflammatory mediators and promoting neutrophil infiltration, thereby intensifying local inflammation and tissue necrosis (53). The resulting local and systemic inflammatory responses rapidly amplify, ultimately causing irreversible inflammation and tissue death. This evidence indicated that KCs represent a potential therapeutic target for ALF, where timely inhibition of KC activation may reduce liver injury and enhance liver regeneration. This area warrants further investigation, given its significance in ALF progression and outcomes (54,55).

The most commonly used model organisms currently include non-mammalian models (brewer's yeast, C. elegans, Drosophila melanogaster and Danio rerio), and mammalian models (Mus musculus). Patient-derived xenograft models and cell lines are also used in cancer research. In addition, in situ tissues and stem cell-derived organoids provide a new platform for basic research, each with unique advantages and limitations.

C. elegans has served as a prominent research model since the mid-1970s (62), and its highly deterministic developmental process makes it an ideal model for cellular lineage analysis, through which numerous genes essential to the apoptotic program were initially identified (63).

The selection of zebrafish (Danio rerio) was primarily due to the near-complete transparency of its early embryo and robust reproductive capacity with manageable costs, enabling researchers to elucidate fundamental principles of early embryonic development (64,65). Large-scale genetic screening in zebrafish commenced in the 1990s with two concurrent extensive screens for mutations affecting various developmental processes in Tübingen and Boston (66-68).

However, these non-mammalian species exhibit notable differences from humans in their growth and development processes, including their physiological characteristics, genomes and disease pathogenesis.

Among mammalian models, mice represent closer analogues to human genetics, development and disease compared with Drosophila or worms. The mouse serves as the preferred animal model for biomedical research, enabling researchers to identify interventions for neurological disorders (69,70), validate initial stem cell origin hypotheses (71,72) and advance emerging fields such as 'fertility preservation' (73) and reversal of epigenetic changes induced by assisted reproduction techniques (74). Researchers selected the mouse as a model organism for three primary reasons: i) The mouse genome is more thoroughly characterized than other animal models, with sophisticated gene editing tools and germline totipotent embryonic stem cells, facilitating genetic modification through evolving tools for genetics, genotyping and phenotyping; ii) mouse populations can be readily expanded, enabling large-scale experiments to derive statistically significant conclusions regarding developmental and physiological mechanisms and systemic properties; and iii) feeding management and environmental conditions can be precisely controlled, allowing systematic multi-omics studies at molecular, cellular and organ levels (75-100). Although studies using identical mouse models may yield different findings, such variations often result from differences in technical analysis methods, a reproducibility challenge also present in human studies. This necessitates enhanced optimization of husbandry environments, stringent verification of mouse strain identity and standardization of basic experimental conditions, which, with appropriate mouse models, can provide accurate insights into human conditions, facilitate biological principle discovery, prioritize human studies and validate related hypotheses.

Prior to the emergence of organoid technology, various approaches to simulate human organ biology have been explored, including 2D human stem cell differentiation with or without 3D matrices, bio-3D printing of human cells and cell culture in microfluidic devices ('organ-on-a-chip') (101). Organoid generation from human patient-derived cells or human-induced pluripotent stem cells has demonstrated notable value in studying mechanisms related to differentiation, morphogenesis, pathogenesis and drug action, while also facilitating biomarker discovery for diseases at cellular and tissue levels (102).

However, organoids, being closed structures, lack tissue-tissue interfaces, vascular flow, circulating immune cells and physiologically relevant mechanical cues, limiting their capacity to fully capture organ-level responses or study drug effects under pharmacologically relevant conditions dependent on dynamic drug pharmacokinetic exposure profiles. The incorporation of self-organizing capillary networks in 3D organoid cultures (103) and immune cell addition to surrounding ECM gel may address some limitations (104). Nevertheless, substantial constraints persist, including the inability to control vascular structure and flow dynamics, and challenges in understanding chemical, inflammatory molecule, drug and immune cell movement within living vascularized organs. Due to their closed structure surrounded by dense ECM, organoids present difficulties in measuring nutrient, chemical or drug transport and uptake in epithelial tissues, sampling inner lumen contents, maintaining microbiome co-cultures and integrating sensors for on-line functional measurements (105).

Previous research indicates that organoids demonstrate notable potential in complementing existing model systems and advancing basic biology, medical research and drug discovery within physiologically relevant human environments (101). However, organoid technology remains in its early developmental stages compared with established cell lines and animal models, with numerous challenges yet to be addressed.

The utilization of distinct animal models, 2D human cell lines and 3D organoids throughout the past three decades has substantially enhanced the understanding of disease pathogenesis, while simultaneously revealing the limitations of these systems in replicating human pathophysiology. The mouse model maintains its position as the predominant model organism for disease research, with particular validation coming from disease-related genes and pathological features initially discovered in mouse studies being subsequently confirmed in humans (106,107), establishing the mouse model as the preferred animal model for biomedical research.

Drug-induced liver injury serves as a key contributor to ALF (Fig. 6); consequently, researchers frequently establish drug-induced liver injury models for treatment or prevention studies in the early stages of the disease (108). Pharmacologic liver injury presents one of the most notable challenges for hepatologists due to several factors, including: i) The extensive range of drugs used in clinical practice; ii) the availability of potentially hepatotoxic herbs and dietary supplements; iii) the diverse clinical and pathologic manifestations of the disease; and iv) the absence of specific biomarkers, which renders diagnosis inherently complex. Current pharmacological models of ALI/ALF primarily utilize intraperitoneal injection of excessive APAP, a medication widely recognized as safe and effective for analgesic and antipyretic purposes at therapeutic doses. However, overdose results in liver injury and potentially progresses to ALF. Research has demonstrated that APAP notably contributes to the development of drug-induced ALI and ALF (31).

In 1973, Mitchell et al (109) discovered that rats exhibited resistance to APAP-induced hepatic injury following APAP injection, while C57BL/6J mice provided accurate indicators of hepatic injury status after APAP overdose. The C57BL/6J mouse remains the predominant animal model for APAP hepatotoxicity research, although other susceptible mouse strains, including ICR mice (110), C3Heb/FeJ mice (111) and B6C3F1 mice, have also been employed in ALI/ALF research (112).

For ALI modeling in C57BL/6J mice, APAP doses typically range from 300 to 600 mg/kg, with 600-750 mg/kg considered the lethal dose for ALF modeling. However, a standardized dose for ALI/ALF modeling remains undefined, with researchers selecting different APAP doses for various mechanistic studies. APAP overdose triggers excessive production of reactive-free radicals and n-acetyl-p-benzoquinone imine (NAPQI) (113). NAPQI rapidly depletes cellular glutathione and forms complexes with biologically active molecules, inducing oxidative stress, causing mitochondrial damage and ultimately resulting in hepatocellular injury (114).

Herbal extracts demonstrate particular promise as therapeutic agents for ALI/liver failure, functioning through antioxidants. Platanoside (PTS), isolated from American sycamore (Platanus occidentalis), represents a potential novel tetramolecular phytopharmaceutical antibiotic effective against drug-resistant infectious diseases and exhibits antioxidant properties. A previous study demonstrated reduced inducible nitric oxide synthase expression and c-Jun-N terminal kinase (JNK) activation when APAP-overdosed mice (300 mg/kg) received PTS treatment (115). Atractylenolide I (AO-I), a phytochemical derived from Atractylodes macrocephala Koidz., demonstrates established antioxidant properties. Li et al (116) administered AO-I to C57BL/6 mice following 500 mg/kg APAP-induced hepatotoxicity, revealing that AO-I provided protection against APAP-induced hepatotoxicity through the TLR4/MAPKs/NF-κB pathway (117). Research utilizing an ALF mouse model constructed with high-dose APAP (700 mg/kg) demonstrated that mannose-binding lectin regulates CYP2E1 expression through the ROS-dependent JNK/SP1 pathway and mitigates APAP-induced hepatotoxicity (116).

Mitochondrial damage plays a crucial role in APAP-induced hepatocyte necrosis and liver dysfunction. Research demonstrates that APAP activates mitophagy, evidenced by increased Parkin translocation to mitochondria and ubiquitination of mitochondrial proteins, along with concurrent sequestration of damaged mitochondria in autophagosomes and degradation of mitochondrial proteins in primary hepatocytes from mouse livers (118). In studies examining the relationship between mitochondrial autophagy and ALF occurrence, researchers predominantly established ALI models through intraperitoneal injection of 500 mg/kg APAP for 24 h (119-124). The initial implementation of the APAP-ALI model was conducted by Ni et al (125), who utilized the 500 mg/kg APAP C57BL/6J mouse model in 2012 and demonstrated that treatment with the autophagy inducer rapamycin inhibited APAP-induced hepatotoxicity. Subsequently, in 2015, the authors identified distinct responses to mitochondrial autophagy and hepatic injury in mice induced by chronic deletion of Parkin and acute knockdown of APAP using the same model (126). In 2019, their research revealed that PINK1 and dual deletion of Parkin compromised hepatic mitochondrial autophagy and intensified APAP-induced hepatic injury in mice (127). Studies investigating autophagy through amelioration of inflammatory response and apoptosis-related research typically employ a lower dose of 300 mg/kg APAP (42,128). Chen et al (129) administered 400 mg/kg APAP coenzyme Q10 to C57BL/6J mice to activate mitochondrial autophagy and prevent APAP-induced hepatic injury. APAP doses >700 mg/kg are considered lethal, with C57BL/6J mice survival rates <60 h, paralleling ALF disease progression (130).

The ER facilitates a specialized intracellular environment for protein processing, folding and sorting for intracellular and extracellular transport. ER stress (ERS) occurs when the ER structure sustains damage or when protein synthesis exceeds the functional capacity of the ER (131). Transient ERS serves as a protective mechanism for normal cellular function under stress. However, prolonged ERS may trigger apoptosis, potentially compromising organ function (132). Sustained ERS can enhance ROS production, intensifying oxidative stress (133), or initiate an inflammatory response, further aggravating cellular damage (134). In 2019, Torres et al (135) first established that APAP induces ALI through ERS, utilizing C57BL/6J 300 mg/kg APAP for ALI modeling. By contrast, studies examining rhodiola rosea glycosides for APAP-induced ALI prevention employed C57BL/6J 500 mg/kg APAP (136). This dosage also revealed the transcription factor CHOP as a key regulator of APAP-induced hepatotoxicity (137). The investigation of hepatotoxicity alleviation through ERS remains in its early stages, leaving uncertainty about future optimal APAP model development.