Endogenous metabolites as active drivers of acute pancreatitis: Mechanisms and advances in targeted therapy

Abstract

Acute pancreatitis (AP) is a common acute abdominal emergency initiated by pancreatic acinar cell injury. Approximately 20% of patients progress to severe disease with multiple organ dysfunction, and mortality rates remain high. However, current clinical management remains largely supportive fluid resuscitation, nutritional support, and complication management and no specific drug has yet been approved directly targeting its key pathogenic mechanisms. Advances in metabolomics have revealed disruptions in multiple pathways, including lipid mediators, bile acids, short-chain fatty acids, amino acids, and tryptophan metabolites. These disruptions are not merely passive epiphenomena of AP but rather active participants in injury initiation, inflammatory amplification, and immune dysregulation. Consequently, interventions targeting metabolic network imbalance have emerged as a novel therapeutic direction. This narrative review classifies metabolites into three categories based on their origin: Endogenous metabolites directly generated by host enzyme systems (lipids, purines, tryptophan, amino acids and their derivatives, polyamines, and oxidative stress-related metabolites); Microbiota-derived metabolites (short-chain fatty acids and indoles); and co-metabolites that depend on host-microbiota collaboration (bile acids). We then systematically review the pathological roles and targeted intervention strategies of eight metabolic pathways in AP. From a spatiotemporal perspective, we further propose two integrative conceptual frameworks the mitochondrion as a multi-hit integration node and the gut as a source of three messengers. We also establish a four-tier evidence stratification ranging from human therapeutic evidence to mechanistic extrapolation, and design a phase-based combination intervention strategy informed by the temporal features of the disease course. Our aim is to provide a comprehensive reference spanning from mechanistic networks to clinical translation for metabolism-targeted therapy in AP.

Key Words: Acute pancreatitis; Metabolic network; Metabolomics; Targeted therapy; Drug repurposing; Mitochondrial dysfunction; Gut microbiota

Core Tip: Acute pancreatitis still has no therapy directed at its core mechanisms, and treatment remains supportive. Metabolomic studies have revealed widespread disturbance of metabolic pathways during the disease, and accumulating evidence indicates that these changes are not merely markers but active contributors to acinar injury, inflammatory amplification, and immune dysregulation. This review classifies the relevant metabolites by origin host enzymes, gut microbes, or the cooperation of both and examines eight pathways together with the interventions being tested for each. It treats the mitochondrion as a point where several injurious pathways converge and the gut as a source of metabolites that act on the pancreas from a distance. As the contribution of each pathway shifts as the disease evolves, it also considers how interventions might be matched to the disease phase. As the evidence is still largely preclinical, the framework is meant to guide further study rather than current treatment.

INTRODUCTION

Acute pancreatitis (AP) is an abdominal emergency initiated by pancreatic acinar cell injury, with a global annual incidence of approximately 34 per 100000 that continues to rise. Approximately 20% of patients progress to severe AP (SAP), frequently complicated by multiple organ dysfunction syndrome, with mortality rates reaching 20%-40%[1]. Over the past several decades, basic research has progressively uncovered the interplay of multiple mechanisms in AP, including abnormal trypsinogen activation, calcium signaling dysregulation, mitochondrial damage, and inflammatory cascades[2,3]. However, clinical management remains largely supportive fluid resuscitation, nutritional support, and complication management and no drug specifically targeting these core pathogenic mechanisms has been approved. The widening gap between growing mechanistic insights and the persistent lack of effective therapies underscores the urgent need to reassess the pathological logic of AP and explore novel therapeutic dimensions.

The development of metabolomics has provided a new entry point in this regard. Using untargeted metabolomics, researchers have profiled the metabolic landscape of plasma, serum, urine, and pancreatic tissue from patients with AP, revealing disruptions across multiple classes of metabolites, including lipid mediators and bile acids, with several differential metabolites correlating with disease severity, organ function, and mortality risk[4]. Targeted metabolomics has further enabled more precise quantification of specific pathways: Targeted lipidomics has revealed the aberrant conversion of phosphatidylcholine to lysophosphatidylcholine (LPC) and an imbalance between pro-inflammatory and pro-resolving lipid mediators[5]; Targeted bile acid profiling has delineated altered ratios of primary to secondary bile acids[6]; and integrated metabolomic and microbiome analyses have demonstrated that certain metabolic alterations are directly linked to gut dysbiosis and to functional pathways such as short-chain fatty acid (SCFA) production[7]. Collectively, these findings indicate that disruptions in metabolic pathways involving lipid mediators, bile acids, SCFAs, and amino acids are not passive bystanders but rather active contributors that drive disease progression in AP[8,9]. Metabolic dysregulation has thus transitioned from a mere downstream marker to a potential pathogenic driver, and targeted interventions aiming to restore metabolic network balance have accordingly emerged as a new frontier in therapeutic exploration.

From the perspective of metabolic network integration, this review examines the cross-regulatory mechanisms among the various metabolic pathways in AP and their translational prospects. We classify metabolites into three groups based on their origin: Endogenous metabolites directly generated by host enzyme systems, microbiota-derived metabolites produced through fermentation or transformation by gut microbes, and host-microbiota co-metabolites (bile acids). Within this framework, we sequentially discuss the physiological and pathological roles of eight metabolic pathways and their corresponding targeted intervention strategies. We then propose two conceptual frameworks, the mitochondrion as a multi-hit integration node and the gut as a source of three messengers to characterize the convergence and synergy of multiple pathways at key nodes. We further establish a four-tier evidence stratification ranging from human therapeutic evidence to mechanistic extrapolation and design a phase-based combination intervention strategy informed by the temporal features of the disease course, with the aim of providing a comprehensive reference for metabolism-targeted therapy in AP that spans from the mechanistic network to clinical translation.

METHODOLOGY

Literature search strategy

This work adopted a narrative review design, with PubMed and Web of Science as the principal databases searched and the search window extending through 2026 for both. The search strategy used “acute pancreatitis” as the core MeSH term, combined separately with terms pertaining to each metabolic pathway (e.g., lipid mediator, bile acids). In addition, the reference lists of the key articles thus identified were screened to capture relevant evidence that the database searches might have missed, thereby improving the completeness of literature coverage.

Principles for evidence selection

When selecting evidence, we prioritized studies directly addressing AP. When direct AP evidence for a given metabolic pathway was lacking, studies from adjacent disease areas such as inflammatory bowel disease (IBD), sepsis, and cancer were cited with caution; such evidence is explicitly labeled as “extrapolated evidence” in both the text and the summary tables to signal that its applicability to AP remains uncertain.

Grading of evidence strength and standardization of wording

To prevent overinterpretation, we graded the causal strength of the included evidence and standardized the corresponding wording throughout the manuscript. Evidence based solely on cross-sectional or observational human studies was designated the association level, with wording confined to “is associated with”. Evidence from in vitro or animal models demonstrating that pathway intervention alters pathological markers was designated the contributory level, with terms such as “participates in” or “promotes”. Evidence in which genetic or pharmacological intervention was sufficient to initiate or fundamentally alter the disease course was designated the driving level, with terms such as “initiates” or “plays a core pathogenic role”. Only when driving-level evidence existed and pathway-targeted intervention improved clinically meaningful outcomes with translational feasibility was a pathway described as a “therapeutic target”.

LIPID METABOLISM

The arachidonic acid cascade and eicosanoids

Arachidonic acid (AA) is a major constituent of membrane phospholipids and is released from them by the action of phospholipase A2 (PLA2). Once released in large quantities, AA is metabolized through three main pathways: Cyclooxygenase (COX), lipoxygenase (LOX), and cytochrome P450[10].

In the COX pathway, COX-2 expression is markedly upregulated in pancreatic tissues during AP, catalyzing the conversion of AA to prostaglandin E2 (PGE2) and thromboxane A2 (TXA2). Of these, PGE2 exerts bidirectional regulatory effects via the EP receptor family: EP3 activation promotes intracellular calcium release and inflammatory cytokine secretion in acinar cells, whereas EP4 activation exhibits certain anti-inflammatory and tissue-protective effects[11,12]. TXA2, in turn, mediates platelet aggregation and microvascular constriction via the thromboxane receptor, thereby exacerbating pancreatic microcirculatory impairment[13].

In the LOX pathway, 5-LOX catalyzes the conversion of AA into leukotriene B4 (LTB4) and cysteinyl leukotrienes (CysLTs). The former mediates neutrophil infiltration into the pancreas via the block lipid transport-1 (BLT1) receptor[14], whereas the latter increases vascular permeability through the CysLT1 receptor, promoting edema formation in both pancreatic and pulmonary tissue[15].

Unlike the pro-inflammatory products described above, AA metabolism also generates the pro-resolving mediator lipoxin A4 (LXA4), which exerts anti-inflammatory and pro-resolving functions through the formyl peptide receptor 2/Lipoxin A4 receptor (FPR2/ALX) receptor[16]. In AP models, a decline in LXA4 levels frequently accompanies sustained inflammation and aggravated tissue injury, suggesting that an imbalance between pro-inflammatory and pro-resolving mediators may be one of the metabolic underpinnings of uncontrolled inflammation in AP[17].

Specialized pro-resolving mediators

Specialized pro-resolving mediators (SPMs) are derived chiefly from ω-3 polyunsaturated fatty acids and comprise three major families, resolvins, protectins, and maresins, serving as key signaling molecules in the active resolution of inflammation[18].

Several SPMs have demonstrated clear protective effects in animal models of AP. Resolvin D1 (RvD1) inhibits excessive neutrophil infiltration and reduces the release of inflammatory cytokines such as interleukin (IL)-6 through the GPR32 and FPR2/ALX receptors[19]. Resolvin E1, acting via chemokine-like receptor 1 and BLT1, antagonizes the pro-inflammatory signaling of LTB4 and promotes the resolution of inflammation. Protectin D1 attenuates pancreatic necrosis and pulmonary inflammatory injury in AP animals by inhibiting neutrophil infiltration and the formation of neutrophil extracellular traps[20]. Although maresin 1 (MaR1) has not yet been directly validated in AP, it has been shown in other models of sterile inflammation to enhance phagocytosis and efferocytosis via the LGR6 receptor and to promote tissue regeneration, suggesting that it may hold similar potential in AP[21].

Clinical observations show that, in patients with gallstone-induced AP, serum RvD1 and LXA4 levels are reduced while the pro-inflammatory mediator LTB4 is elevated, with RvD1 inversely correlated with LTB4. This suggests that an imbalance between pro-inflammatory lipid mediators and SPMs may be associated with persistent inflammation and disease exacerbation, although the causal relationship awaits further validation[22].

Lysophospholipid and sphingolipid metabolism

Beyond the classical eicosanoid pathways, lysophospholipid and sphingolipid metabolism also play important regulatory roles in AP. LPC is a particularly notable metabolite in this context. When the common channel of the bile duct and pancreatic duct becomes obstructed, biliary phosphatidylcholine is hydrolyzed to LPC by PLA2. High concentrations of LPC can directly damage acinar cells, inducing calcium overload and collapse of the mitochondrial membrane potential, while also activating innate immunity via Toll like receptor-dependent pathways[23].

In sphingolipid metabolism, ceramide production increases under various cellular stress conditions and induces acinar cell apoptosis and necrosis through activation of protein phosphatase 2A and promotion of mitochondrial outer membrane permeabilization[24]. Sphingosine-1-phosphate (S1P) plays a broad role in immune cell migration, the maintenance of vascular integrity, and the regulation of inflammation[25]. Clinical data have shown that plasma S1P levels are elevated in patients with mild AP, whereas they are significantly reduced in the early phase of SAP, suggesting that S1P not only participates in inflammatory processes but also correlates with disease severity[26]. Compared with the AA and SPM pathways, targeted intervention studies on lysophospholipid and sphingolipid metabolism in AP remain relatively sparse, with most strategies still at the conceptual stage; therefore, these pathways will not be further addressed in the intervention sections of this review.

Targeted intervention strategies for lipid mediators

SPM supplementation and optimization of delivery: Exogenous SPM supplementation has demonstrated therapeutic efficacy in multiple animal models of AP. In a mouse model of SAP, intraperitoneal injection of RvD1 (300 ng/mouse) attenuated pancreatic edema and inflammatory infiltration while reducing plasma cytokine levels, conferring protective effects on both the pancreas and lung[19]. MaR1 alleviated pancreatic injury and fibrosis in a cerulein-induced AP model by promoting macrophage polarization toward a reparative phenotype[27]. These gain-of-function studies indicate that the SPM pathway is not merely an epiphenomenon of inflammation but is directly involved in the regulation of tissue injury in AP, thereby supporting its potential as a therapeutic target.

However, targeted intervention of this pathway is currently constrained by the pharmacokinetic limitations of natural SPMs: They contain multiple unsaturated double bonds, have a half-life of only a few minutes in vivo, and lack target specificity upon systemic administration[28]. To overcome these limitations, ongoing engineering strategies include chemically stabilized analogs (e.g., benzo-RvD1)[29], nano-formulation delivery systems[30], and ω-3 polyunsaturated fatty acid precursor supplementation[31]. Most of these approaches remain at the preclinical stage and have been validated in only a single model of AP each, leaving a considerable gap before clinical application.

Targeting the LXA4 pathway and modulating pro-inflammatory mediators: In the LXA4 pathway, studies in mouse models of AP-associated acute lung injury have shown that LXA4 markedly attenuates pathological injury of pancreatic and pulmonary tissues and suppresses the generation of inflammatory cytokines such as tumor necrosis factor-α and IL-1β, as well as reactive oxygen species (ROS). Its receptor agonist BML-111 likewise activates the nuclear factor erythroid 2-related factor 2 (Nrf2)/heme oxygenase-1 (HO-1) antioxidant pathway, thereby reducing ROS accumulation and apoptosis in the intestinal epithelium and restoring E-cadherin expression. Consequently, it not only alleviates pancreatitis but also ameliorates the concomitant disruption of the intestinal barrier[32-34].

When pro-inflammatory lipid mediators are suppressed, whether protective pathways can be spared is central to evaluating the merit of a given strategy. Selective inhibitors of microsomal prostaglandin E synthase-1 reduce pro-inflammatory PGE2 without affecting the synthesis of prostacyclin (PGI2), which is essential for pancreatic microcirculation. Their selectivity is therefore superior to that of conventional non-steroidal anti-inflammatory drugs (NSAIDs), which inhibit both COX-1 and COX-2[35].

Beyond such selective inhibition, direct blockade of individual pro-inflammatory lipid-mediator receptors has also shown protective effects in AP models. The BLT1 inhibitor LY293111 suppresses aberrant neutrophil migration and attenuates both AP and AP-associated lung injury[36]. Likewise, the CysLT1 antagonist montelukast blocks CysLT-induced vascular permeability and neutrophil recruitment, demonstrating protective effects in multiple AP models[37]. Montelukast is already approved for the treatment of asthma, which lowers the barrier to its repurposing; however, clinical trials for an AP indication are still lacking.

Multi-target natural products (baicalein as an example): Baicalein, the principal flavonoid constituent of Scutellaria baicalensis, offers an alternative approach to lipid-mediator intervention by virtue of its multi-target properties. It potently inhibits 12-LOX activity and binds directly to the high mobility group box-1 protein[38]. In a hypertriglyceridemic AP (HTG-AP) model, baicalein downregulates NOD-like receptor thermal protein domain associated protein 3 (NLRP3) expression through the miR-192-5p/TXNIP axis, thereby suppressing inflammasome activation[39]. Recent studies have further shown that baicalein upregulates glutathione peroxidase 4 (GPX4) and activates the Nrf2/HO-1 pathway, suggesting that it may antagonize ferroptosis an effect that complements its anti-pyroptotic properties described above[40]. However, its multi-target nature complicates the structure–activity relationship, and its low oral bioavailability represents a significant barrier to clinical application.

In summary, the imbalance between pro-inflammatory and pro-resolving lipid mediators represents an important metabolic node underlying the amplification of inflammation and the failure of its resolution in AP. SPM supplementation, LXA4 receptor agonists, and selective pro-inflammatory receptor antagonists have shown protective potential in animal models, while multi-target natural products such as baicalein offer novel perspectives for synergistic modulation. Yet these strategies remain constrained by pharmacokinetic bottlenecks or insufficient clinical evidence; overall, lipid-mediator-targeted strategies are still largely at the preclinical stage.

PURINE METABOLISM

Adenosine triphosphate and adenosine signaling pathways

In AP, necrotic acinar cells release large amounts of adenosine triphosphate (ATP) into the extracellular space, making it a key damage-associated molecular pattern (DAMP). Upon binding to the P2X purinoceptor 7 (P2X7) receptor, extracellular ATP (eATP) activates the NLRP3 inflammasome, promoting the maturation and release of IL-1β and IL-18 and inducing pyroptosis; this constitutes one of the central steps in the amplification of sterile inflammation[41].

Under physiological conditions, eATP is sequentially hydrolyzed to adenosine by cluster of differentiation (CD) 39 and CD73, with adenosine exerting anti-inflammatory and tissue-repairing effects[42]. However, under pathological conditions such as AP, intense inflammation and oxidative stress can downregulate or functionally inhibit the CD39/CD73 ectonucleotidase system. This delays eATP clearance and shifts the pro-inflammatory anti-inflammatory balance toward a pro-inflammatory state. Consistent with this, clinical studies have found that circulating soluble CD73 activity declines with increasing AP severity and correlates with the risk of persistent organ failure[43]. In animal experiments, clearance of eATP by apyrase markedly attenuates the systemic inflammatory response in AP, providing functional preclinical evidence for interventions targeting this pathway[44].

Uric acid and xanthine oxidase

Xanthine oxidase (XO) catalyzes the oxidation of hypoxanthine and xanthine to uric acid while generating large amounts of ROS, making it a major source of oxidative stress in AP. The pancreatic ischemia-reperfusion associated with AP markedly enhances XO activity[45].

Clinical observations indicate that hyperuricemia is relatively common in patients with AP[46]. At physiological concentrations, uric acid is an important plasma antioxidant that scavenges free radicals such as peroxynitrite and exerts a protective effect on the vascular endothelium[47]. However, when massive cell necrosis causes local uric acid concentrations to rise sharply, uric acid instead acts as a DAMP. Through pathways including increased mitochondrial ROS production, it activates the NLRP3 inflammasome, forming a “uric acid-ROS-NLRP3” positive feedback loop that amplifies local and systemic inflammation[48].

Targeted intervention strategies for purine metabolism pathways

P2X7 receptor antagonism and NLRP3 inflammasome blockade: Genetic deletion of the P2X7 receptor and functional antagonists such as oxidized ATP and brilliant blue G have consistently reduced inflammation and tissue injury in animal models of AP and related inflammatory conditions, supporting P2X7 as a potential therapeutic target in AP[49,50]. However, translation from bench to bedside faces the challenge of species differences: The human and murine P2X7 receptors differ markedly in their pharmacological characteristics, and identifying compounds with high affinity for the human P2X7 receptor is key to clinical translation[51].

The NLRP3 inhibitor MCC950 markedly attenuates pancreatic and intestinal injury across several AP models[52,53]. Another NLRP3 inhibitor, dapansutrile, has demonstrated a favorable safety profile in a phase II clinical trial in gout[54]. Given the shared central role of the NLRP3 inflammasome in AP and gout, the potential application of this agent in AP warrants attention, although it has not yet been validated in AP models.

Strategies to enhance adenosine signaling: Currently, adenosine signaling enhancement strategies in AP remain at an early exploratory stage, with the various intervention pathways differing in their benefit-risk profiles. Direct activation of adenosine A2A receptors has shown anti-inflammatory and tissue-protective effects in AP models[55], yet carries the potential risk of exacerbating edema[56]. In contrast, ABT702 which indirectly elevates endogenous adenosine levels by inhibiting adenosine kinase has consistently attenuated inflammatory responses and acinar cell necroptosis in both in vitro and in vivo SAP models[57].

XO inhibition: A substantial body of preclinical evidence has accumulated for XO inhibition in AP. Selective XO inhibitors such as allopurinol and febuxostat attenuate pancreatic necrosis and systemic inflammation in most animal models of AP. Beyond scavenging ROS, their mechanisms also involve inhibition of the hypoxia inducible factor-1α/Lactate dehydrogenase A glycolytic pathway and NLRP3 inflammatory signaling[58]. Available evidence suggests that intervention yields the greatest benefit during the early rise in XO activity or the ischemia-reperfusion phase; however, owing to differences among models, the optimal timing and dosing remain to be defined[59]. Moreover, current clinical evidence is confined to a few randomized controlled trials and meta-analyses of allopurinol for the prevention of post-endoscopic retrograde cholangiopancreatography (ERCP) pancreatitis, and the overall results have been negative[60,61]. At present, no reliable clinical studies support the use of XO inhibitors for the treatment or prognostic improvement of AP in general.

Emodin (a natural P2X7 antagonist and modulator of multiple metabolic pathways): Emodin possesses both P2X7-antagonistic and antioxidant activities and can act simultaneously on the P2X7-NLRP3 and XO-ROS axes, offering an approach to integrated intervention in purine metabolism[62]. In vitro studies have shown that, acting as a natural P2X7 receptor antagonist, it suppresses ATP-induced secretion of inflammatory cytokines by macrophages and human pancreatic ductal epithelial cells[63]. In multiple AP models, it not only downregulates the P2X7-NLRP3-apoptosis-associated speck-like protein containing a CARD signaling axis but also attenuates pancreatic and pulmonary injury by inhibiting the cold-inducible RNA-binding protein-mediated nuclear factor kappa-B (NF-κB)/CXCL1 pathway[64]. However, its clinical translation faces core bottlenecks: Low oral bioavailability, reported hepatotoxicity and nephrotoxicity, and insufficient independent validation as most mechanistic studies have originated from a limited number of research groups[65]. Therefore, although its multi-target profile is of considerable research interest, clinical translation should be approached with caution. The role of emodin in the Nrf2-mediated anti-ferroptosis pathway and engineered delivery strategies are discussed further in “anti-ferroptotic metabolic intervention”.

Current evidence indicates that purine metabolism converges on the NLRP3 inflammasome through two axes “eATP-P2X7-NLRP3” and “uric acid/XO-ROS-NLRP3” representing one of the principal drivers of sterile inflammatory amplification in AP. Strategies targeting P2X7, NLRP3, adenosine signaling, and XO have shown protective potential in animal models, but face challenges including species differences, insufficient clinical evidence (such as the negative results of allopurinol for preventing post-ERCP pancreatitis), and safety concerns. Overall, the clinical translation of interventions targeting this pathway remains at an early stage.

TRYPTOPHAN METABOLISM

Although chemically an amino acid, tryptophan functions in AP primarily as a pathogenic driver via pathways such as the kynurenine pathway (KP), with corresponding interventions targeting key enzymes. This fundamentally distinguishes it from other amino acids, whose role is mainly nutritional support; therefore, tryptophan is discussed separately. The remaining amino acids and their derivatives are addressed in “amino acids and their derivatives”. Over the course of AP, dysregulation of tryptophan metabolism is closely associated with systemic inflammation, immune dysregulation, and remote organ injury. In vivo, tryptophan is metabolized along three main pathways: The KP, the 5-hydroxytryptamine (5-HT) pathway, and the indole pathway. The first two are driven by host enzyme systems and are the focus of this section, whereas the indole pathway depends on the gut microbiota; its products are classified as microbiota-derived metabolites and are discussed in “the indole-aryl hydrocarbon receptor (AhR)-IL-22 axis and barrier protection”.

Tryptophan metabolic pathways in AP (the KP and 5-HT pathways)

The rate-limiting step of the KP is mediated by enzymes such as indoleamine 2,3-dioxygenase (IDO), which can be upregulated by pro-inflammatory cytokines during systemic inflammation. In patients with AP, early plasma tryptophan depletion and elevated downstream KP metabolites have been observed, indicating activation of this pathway. Of clinical significance is the kynurenine-3-monooxygenase (KMO) branch: This enzyme catalyzes the formation of cytotoxic 3-hydroxykynurenine (3-HK), which exacerbates remote organ injury through oxidative stress[66]. A prospective cohort of patients with AP showed that plasma 3-HK levels correlated significantly with concurrent acute physiology and chronic health evaluation II scores and C-reactive protein (CRP), rising with increasing systemic inflammatory burden and disease severity. This represents one of the key pieces of human evidence supporting a role for the KP in AP-related organ dysfunction[67].

Beyond generating toxic metabolites, flux through the KP is also a core pathway for de novo nicotinamide adenine dinucleotide (NAD⁺) synthesis in vivo. However, in the AP state, persistent DNA damage and inflammatory stress can deplete NAD⁺ through two concurrent mechanisms: Overactivation of poly-adenosine diphosphate ribose polymerase (PARP)-1, which consumes NAD⁺ for PAR polymer synthesis, and upregulation of CD38, which accelerates NAD⁺ hydrolysis. Together, these pathways exacerbate the cellular energy crisis and contribute to the triggering of PARP-1-dependent cell death (parthanatos)[68,69].

The 5-HT pathway, in turn, drives peripheral 5-HT synthesis via tryptophan hydroxylase 1 in enterochromaffin cells. Gene-knockout studies suggest that peripheral 5-HT is not only required for the initiation of experimental pancreatitis but also participates in regulating acinar secretion and injury[70]. In AP, injury signals stimulate enterochromaffin cells to release 5-HT into the peripheral circulation, which activates vagal afferents via the 5-HT3 receptor and elicits nausea and vomiting[71].

Targeted intervention strategies for tryptophan metabolism

Inhibition of the KMO pathway: KMO inhibition is the best-supported AP target within the KP. Inhibiting KMO selectively blocks the toxic 3-HK branch while redirecting metabolic flux toward neuroprotective kynurenic acid[72]. In multiple AP models, both Kmo gene deletion and intravenously administered highly selective KMO inhibitors markedly attenuate injury to remote organs such as the lung, kidney, and liver, and in some models also improve the critical disease course and survival[66,73]. The primary endpoints of existing animal studies have focused on organ function and histological injury, whereas survival benefits have emerged only in a limited number of severe models and require further validation. Overall, this approach is considered the most translationally promising intervention within the KP. However, no clinical trial of KMO inhibition in AP patients has been conducted to date.

Modulation of the 5-HT pathway: The 5-HT3 receptor antagonist ondansetron is already widely used as an antiemetic in the clinical management of AP. In a caerulein-induced murine AP model, it lowers plasma IL-6 and attenuates pancreatic tissue injury and neutrophil infiltration; the underlying mechanism may involve suppression of the 5-HT3-mediated vagal-inflammatory reflex and of monocyte activation[74]. A retrospective study using an intensive care unit database of approximately 1000 patients with AP found that ondansetron use was associated with better in-hospital, 90-day, and overall survival, with adjusted hazard ratios of approximately 0.5 to 0.66[75]. However, as a single-center retrospective study reflecting only an adjusted association with all-cause mortality and subject to biases such as confounding by indication, this benefit currently constitutes only associative evidence. As a conventional drug with a well-established safety profile, ondansetron would be an ideal candidate for drug repurposing if its anti-inflammatory benefits are confirmed in randomized controlled trials. Furthermore, as a 5-HT precursor, orally administered 5-hydroxytryptophan increases small intestinal T helper cell 17 and IL-17A levels, upregulates antimicrobial peptides and barrier molecules, thereby reducing bacterial burden in the pancreas and lungs and consequently improving 72-hour survival and alleviating infectious complications in AP mice[76].

NAD⁺ supplementation: NAD⁺ depletion is a common downstream consequence of aberrant tryptophan-KP metabolism converging with inflammatory stress. Restoring NAD⁺ metabolism has therefore emerged as a promising intervention, with current approaches centering on two routes: Reducing consumption and increasing supply. The PARP-1 inhibitor olaparib, by suppressing PARP overactivation, restores NAD⁺ levels and thereby markedly attenuates pancreatic and renal injury and dampens inflammatory responses in animal models of AP. PAR accumulation has also been detected in human pancreatic specimens, providing preliminary clinical evidence for this strategy[77,78].

Supplementation with the NAD⁺ precursor nicotinamide riboside, a complementary strategy, has been shown in several clinical trials to be safe and to effectively raise NAD⁺ levels, providing a sound basis for clinical translation[79]; however, direct evidence in AP is lacking. Mechanistically, combining a PARP inhibitor with NAD⁺ precursor supplementation would reconstitute NAD⁺ homeostasis from both ends reducing consumption and increasing supply, and would thus carry a strong theoretical rationale for synergy. Nevertheless, this combination strategy remains at a conceptual stage and still lacks direct validation in AP models.

Taken together, dysregulation of tryptophan metabolism links systemic inflammation, immune dysregulation, and remote organ injury through the KP and the 5-HT pathway. With respect to intervention, KMO inhibition shows organ-protective potential in animal models but lacks clinical validation; ondansetron is associated with improved survival in retrospective studies and represents a promising candidate for drug repurposing; and NAD⁺ supplementation could theoretically restore metabolic homeostasis from both the consumption and supply ends, although combination regimens still lack direct validation in AP models. Overall, clinical evidence for tryptophan-metabolism-targeted interventions remains very limited.

AMINO ACIDS AND THEIR DERIVATIVES

Glutamine metabolism

Glutamine is the most abundant free amino acid in plasma and the principal energy substrate of small-intestinal epithelial cells. AP is a hypercatabolic stress state in which, although glutamine release from skeletal muscle increases, its output still fails to meet the enormous systemic demand, rendering it a “conditionally essential amino acid” under these conditions[80].

At the pathophysiological level, glutamine exerts protective effects through several routes. On the one hand, it directly promotes the proliferation and differentiation of intestinal epithelial cells and maintains the expression of tight-junction proteins, thereby stabilizing the intestinal barrier. On the other hand, it induces the expression of heat shock protein 70 via heat shock factor-1, enhancing cellular resistance to stress-induced injury. In addition, as a precursor for glutathione (GSH) synthesis, it participates in maintaining intracellular redox homeostasis, while as a key energy substrate it supports the immune functions of lymphocytes and macrophages[81,82].

Arginine-nitric oxide metabolic axis

Arginine is the common precursor for the synthesis of nitric oxide (NO) [via inducible NO synthase (iNOS) and endothelial NO synthase (eNOS)], polyamines, and urea, and its metabolic fate exerts divergent effects on the course of AP. Physiological levels of NO produced by constitutive eNOS help maintain pancreatic microcirculatory perfusion and endothelial function, exerting a protective role during the initiation phase of AP. Consistently, eNOS gene deletion or NOS inhibition exacerbates experimental AP[83]. Under inflammatory conditions, however, iNOS upregulation leads to excessive NO production. This excess NO readily reacts with superoxide anion to form the highly cytotoxic peroxynitrite, thereby intensifying oxidative and nitrosative stress in vivo[84]. Nevertheless, direct evidence for injury to the pancreas itself through this pathway is inconsistent, whereas its injurious role is more clearly established in secondary remote organ injury[85].

Multifaceted protective effects of taurine

Taurine exhibits multifaceted protective effects in AP. As a bile acid-conjugating moiety, it enhances their hydrophilicity while reducing their detergent properties and cytotoxicity, thereby attenuating tissue injury caused by the aberrant bile acid profile in AP[86]. Moreover, across several animal models of AP, taurine preconditioning significantly lowers serum amylase and lipase levels and improves histological injury scores, while reducing malondialdehyde (MDA) content and restoring GSH levels, demonstrating both antioxidant and anti-inflammatory effects[87,88]. However, the available evidence derives chiefly from preconditioning regimens in animal models, which differs from the clinical reality where intervention is typically initiated only after disease onset.

Targeted interventions for amino acids and their derivatives

Targeting glutamine and protection of the gut-pancreas axis: Glutamine currently has relatively substantial clinical evidence in AP, although the overall certainty remains low. Meta-analyses show that parenteral glutamine supplementation reduces the incidence of infectious complications in SAP; enteral supplementation, by contrast, improves the intestinal mucosal barrier but has not reliably translated into improvements in hard endpoints such as infected necrosis or mortality. In addition, the relevant clinical trials are generally limited by small sample sizes, high heterogeneity, and a substantial risk of bias[89,90]. Clinical application therefore requires a strict distinction between general regimens for critical illness and recommendations specific to AP. The ESPEN pancreatitis guidelines are comparatively conservative, recommending the addition of L-glutamine (0.20 g/kg/day) only when enteral nutrition is not feasible and parenteral nutrition is required, rather than as a first-line option[91]. The appropriate role of glutamine in AP is thus limited to a conditional option for patients receiving parenteral nutrition, and high-quality randomized controlled trials are still needed to establish its translational value.

Targeting the arginine-NO metabolic axis: Interventions targeting the arginine-NO axis fall into two categories. The first strategy is to inhibit overactivated iNOS. In several animal models of AP, administering selective or relatively selective iNOS inhibitors such as L-NIL and aminoguanidine after disease onset effectively lowers NO levels, attenuates pancreatic injury, improves hemodynamics, and alleviates secondary lesions in remote organs such as the lung[92,93]. These findings suggest that targeting the iNOS-NO pathway has potential therapeutic value. However, the benefits vary across models and the effects on remote organs are heterogeneous; clinical evidence in human populations is still lacking. Thus, the benefit-risk balance requires further elucidation. The second strategy is indirect modulation through citrulline supplementation. As a precursor for arginine synthesis, citrulline bypasses the substantial hepatic first-pass metabolism that depletes arginine and, when given orally, raises plasma arginine levels more effectively[94]; it has been shown to improve intestinal barrier function in sepsis models[95]. In clinical studies, low plasma citrulline has emerged as a marker reflecting intestinal failure and predicting poor prognosis in patients with AP (often in combination with intestinal fatty acid binding protein), suggesting that the translational value of citrulline in AP warrants further exploration[96].

The amino acids and derivatives discussed above primarily exert “conditional protection” in AP: Glutamine maintains intestinal barrier function and antioxidant reserves; taurine enhances bile acid hydrophilicity and exerts antioxidant effects; and the arginine-NO axis exhibits functional heterogeneity. Among these, glutamine has relatively clinical evidence but with low certainty, and serves only as a conditional supplement for patients on parenteral nutrition rather than a first-line strategy. iNOS inhibition and citrulline supplementation still lack evidence in AP populations and remain, overall, adjunctive directions for intervention.

POLYAMINE METABOLISM

Pathological alterations in polyamine metabolism

Polyamines-spermidine, spermine, and putrescine play fundamental roles in cell proliferation, the regulation of gene expression, and protein synthesis. Their biosynthesis is initiated by ornithine decarboxylase and proceeds through sequential catalysis by spermidine synthase and spermine synthase, forming the progressive pathway putrescine-spermidine-spermine; their catabolism is mediated mainly by spermidine/spermine N¹-acetyltransferase (SSAT) and polyamine oxidases (PAO). In the pancreas, beyond their role in proliferation and protein synthesis, polyamines also maintain the structural integrity and stress tolerance of acinar cells. In AP, aberrant activation of SSAT rapidly depletes spermidine and spermine while leading to the accumulation of putrescine and acetylated polyamines, accompanied by endoplasmic reticulum stress and autophagic dysfunction. This in turn triggers premature zymogen activation, resulting in acinar cell death and an inflammatory response[97].

Dysregulation of polyamine metabolism is a key link connecting metabolic abnormalities with acinar necrosis and is observed across several experimental pancreatitis models. Its causal status, however, varies by model: It is an upstream event triggering pancreatitis in gene-knockout models, whereas in the L-ornithine model it is secondary to tissue injury[98]. Although clinical and animal follow-up data show that erythrocyte polyamine levels correlate with the extent of pancreatic necrosis and with inflammatory markers (IL-6, CRP), at present they reflect only disease severity; whether they can serve as early markers remains to be confirmed by prospective studies[99].

Targeted intervention strategies for polyamine metabolism

Supplementation with higher polyamines or their stable analogues aims to block the acinar necrosis triggered by polyamine depletion. The strongest evidence comes from a methylated analogue resistant to SSAT/PAO degradation (1-methylspermidine): In an SSAT-overexpression model, prior replacement of native polyamines with 1-methylspermidine completely prevents the development of pancreatitis. This effect, however, is markedly model- and phase-dependent, with limited efficacy in conventional exogenous-injury models. For example, in the taurodeoxycholate model it only transiently delays necrosis within 24 hours without affecting trypsinogen activation, whereas in the L-ornithine model it is entirely ineffective[100]. Furthermore, the benefit of this analogue is confined to the pancreas and to the early phase of disease; it fails to ameliorate systemic multi-organ injury and, at high doses, tends to accumulate in the kidney, thereby increasing mortality. By contrast, although direct supplementation with exogenous native spermidine shows preliminary signs of reducing acinar necrosis in the cerulein model, most related studies are largely based on network pharmacology predictions and early-stage validation. They still lack the cross-model causal evidence available for the methylated analogue and, at present, can serve only as a mechanistic clue[101].

The second strategy is to inhibit the oxidative degradation of polyamines, reducing toxic by-products and remodeling the polyamine profile through irreversible inhibition of PAO/SMOX. However, when its representative agent MDL 72527 is used alone in AP induced by SSAT overactivation, it does not improve histological injury[102,103]. This suggests that acinar injury arises mainly from the acute depletion of higher polyamines rather than from PAO-mediated downstream oxidative stress. A converse experiment provides further support: After polyamine catabolism is activated by agents such as gossypol, profoundly depleting higher polyamines, transgenic mice spontaneously develop pancreatitis[104]. Therefore, merely blocking degradation by-products without correcting polyamine depletion is unlikely to effectively reverse pancreatic injury.

In summary, polyamine-targeted interventions remain highly model-dependent. While correcting higher polyamine depletion via methylated analogues holds preventive potential in specific genetic models, this approach demonstrates limited efficacy in exogenous-injury models and carries an inherent risk of nephrotoxicity. Furthermore, merely blocking the production of toxic by-products from polyamine degradation without correcting polyamine depletion is ineffective as a monotherapy, highlighting that a considerable gap persists between mechanistic insights into polyamine metabolism and the clinical translation of polyamine-targeted strategies.

OXIDATIVE STRESS-RELATED METABOLITES

Metabolic basis of oxidative stress

During AP, the neutrophil respiratory burst, activation of XO, and electron leak from the mitochondrial electron transport chain together drive the excessive generation of ROS and reactive nitrogen species, disrupting redox homeostasis[105]. As the most important intracellular small-molecule antioxidant, GSH is continuously consumed during free-radical scavenging, resulting in an elevated glutathione oxidized/GSH ratio. The lipid peroxidation products MDA and 4-hydroxynonenal serve both as markers of oxidative damage and as signaling molecules that modulate antioxidant pathways such as Nrf2/HO-1[106]. Once GSH depletion and GPX4 inactivation reach a critical threshold, iron-dependent lipid peroxidation becomes uncontrolled and triggers ferroptosis. As a pivotal node linking oxidative stress to cell death, ferroptosis has been shown to contribute to acinar cell injury in AP and other pancreatic diseases[107,108]. In addition, mitochondrial oxidative damage and XO-mediated ischemia-reperfusion injury increase local oxygen consumption and precipitate an energy crisis. Coupled with the hypoxia resulting from microcirculatory impairment, these factors collectively drive glycolytic reprogramming of pancreatic tissue, resulting in marked lactate accumulation.

Targeted intervention strategies for redox metabolism

Reconstitution of the GSH system: In multiple animal models of AP, N-acetylcysteine (NAC) replenishes cysteine and elevates pancreatic GSH levels, thereby attenuating acinar injury and oxidative stress, improving calcium ion (Ca²⁺) homeostasis, reducing the accumulation of exocrine enzymes, and alleviating pancreatic injury[109,110]. Clinically, a small randomized controlled trial showed that oral NAC reduces the incidence of post-ERCP pancreatitis without increasing adverse events; however, most meta-analyses have not confirmed a definitive preventive effect[111]. Furthermore, several reviews have concluded that the clinical efficacy of NAC is heterogeneous across studies, with dose, timing of administration, and disease stage being potential key determinants of its effect[112].

Although S-adenosylmethionine (SAMe) exhibits certain antioxidant effects in chronic liver disease[113], the transsulfuration pathway and GSH metabolism are frequently impaired in the setting of AP[114]. Available experimental evidence suggests that SAMe supplementation alone not only fails to effectively restore pancreatic GSH but may instead aggravate the obstruction of the transsulfuration pathway and the inflammatory response. Therefore, SAMe cannot be regarded as equivalent to NAC for reconstituting the GSH system in AP; its use warrants greater caution and requires validation in future studies.

Anti-ferroptotic metabolic intervention: Ferroptosis is regulated at multiple levels. In the canonical pathway, iron overload, GSH depletion, and GPX4 inactivation act in concert to drive the progressive accumulation of peroxidized membrane phospholipids and ultimately cell death; among these, the system cystine/glutamate antiporter (Xc^-)-GSH-GPX4 axis is the most thoroughly characterized anti-ferroptotic route to date[115]. In addition, the endogenous metabolites coenzyme Q10 (CoQ10) and tetrahydrobiopterin (BH4) counteract ferroptosis through GPX4-independent routes, providing new entry points for metabolically targeted intervention in AP[116]. As an element essential to the catalytic center of GPX4, selenium supplementation upregulates the expression and activity of GPX4/GPX1, reinforcing the GSH-GPX4 antioxidant system, and has shown potential for attenuating acinar cell injury in multiple AP models[117,118].

As described in “emodin (a natural P2X7 antagonist and modulator of multiple metabolic pathways)”, emodin activates the Nrf2/HO-1/GPX4 pathway and upregulates GSH and GPX4 levels, thereby inhibiting ferroptosis. Conversely, the Nrf2 inhibitor ML385 markedly attenuates this protective effect, further confirming the critical role of Nrf2 in its anti-ferroptotic mechanism[119]. Given the low oral bioavailability of emodin, it has been encapsulated in extracellular vesicles derived from human umbilical cord mesenchymal stem cells, which has enhanced its anti-inflammatory efficacy across cellular, organoid, and animal models of AP[120,121]. This delivery strategy offers a promising approach to overcoming the bioavailability bottleneck, yet the scalability of its manufacturing process and its in vivo pharmacokinetic profile remain to be evaluated.

Activation of the Nrf2-antioxidant response elements pathway: Itaconate and its derivatives alkylate cysteine residues of KEAP1, block KEAP1-mediated ubiquitination and degradation of Nrf2; this promotes the nuclear translocation of Nrf2 and the transcription of antioxidant and anti-inflammatory genes, thereby suppressing oxidative stress and inflammation[122,123]. Among these, the cell-permeable derivative 4-octyl itaconate (4-OI) efficiently alkylates KEAP1 and activates Nrf2 signaling. In mouse models of AP, 4-OI inhibits ferroptosis and attenuates oxidative stress and inflammation, thereby markedly alleviating pancreatic and pulmonary injury[124]. The available evidence remains confined to preclinical animal models and lacks clinical translation data in patients with AP, so its efficacy and safety have yet to be established.

Lactate metabolism and protein lactylation: Recent studies have revealed a novel role for lactate in promoting epigenetic reprogramming in AP, overturning the traditional view of it as merely a glycolytic waste product. In models of the recovery phase of AP, lactate remodels the macrophage phenotype through two routes: via monocarboxylate transporter 1-mediated lactylation, it upregulates reparative genes such as Lrg1, Vegfa, and Retnla; and via the GPR132 receptor, it inhibits Janus kinase 2-signal transducer and activator of transcription 1 (STAT1) signaling and downregulates pro-inflammatory genes including Tnf, Il6, and Nos2. Together, these pathways drive macrophage polarization toward a reparative phenotype[125]. Both genetic reduction of lactate production (Pfkfb3 deficiency) and pharmacological blockade of lactate transport (AZD3965) attenuate this effect and delay pancreatic repair. Nevertheless, the role of lactate in AP is bidirectional and dependent on disease phase and microenvironment, and should not be simply classified as a pro-resolving signal. In contrast to the reparative effect described above, lactate and H3K18 lactylation can also aggravate pancreatic injury and inflammation by upregulating NCOA4-mediated ferroptosis, whereas inhibiting lactate production attenuates these changes[126]. The current evidence derives mainly from animal and in vitro models, and the precise modification sites of lactylation, its dynamic regulatory network, and its relevance to humans remain to be clarified. Given its dual role in both repair and injury, the feasibility of targeting this pathway to promote repair in the later phase of AP hinges on whether its phase-specific effects can be elucidated.

Overall, oxidative stress represents a key metabolic node at which multiple injury pathways converge on cell death, with ferroptosis being particularly pivotal. Antioxidant and anti-ferroptotic strategies such as 4-OI, selenium supplementation, and CoQ10/BH4 provide complementary entry points, yet the evidence largely remains at the preclinical stage. Through lactylation, lactate can both promote a reparative macrophage phenotype and exacerbate ferroptotic injury, so any targeting strategy must take its phase specificity into account. By comparison, NAC acting through GSH replenishment has accumulated a degree of exploratory clinical evidence and represents one of the more clinically advanced candidates within the field of redox-targeted intervention. On the whole, however, intervention strategies in this field remain predominantly preclinical, and the systematic advancement of clinical translation still awaits support from additional high-quality evidence.

SCFAS AND INDOLES

SCFAs and indole compounds derived respectively from microbial fermentation of dietary fiber and microbial metabolism of tryptophan are key mediators through which the gut microbiota influences host immunity. In AP, both are depleted owing to microbial dysbiosis, thereby compromising intestinal barrier integrity. However, because patients in the acute phase are typically fasted, with enteral intake drastically reduced, the accurate assessment of circulating levels of both metabolites is subject to this shared confounding effect.

Microbiota-derived metabolites and intestinal barrier protection

SCFAs primarily acetate, propionate, and butyrate maintain a molar ratio of approximately 60:20:20 in the colonic lumen of healthy individuals. Most small-sample clinical studies and animal models have demonstrated that SCFA-producing bacteria and both circulating and fecal SCFA levels decline during AP, correlating negatively with disease severity[7]. However, a large prospective multicenter cohort yielded contrasting results, suggesting that the SCFA biosynthetic pathway is upregulated in patients with SAP; thus, the direction of this change remains controversial[127]. This discrepancy is tightly linked to key confounders, such as acute-phase fasting and a sudden reduction in dietary fiber intake. Based on available cross-sectional data alone, it remains challenging to discern whether these SCFA alterations drive microbial dysbiosis or occur secondary to fasting and intestinal injury. Furthermore, research on acetate and propionate in AP remains sparse, with existing literature focusing predominantly on butyrate; therefore, the independent contributions of the former two metabolites await further elucidation.

Multiple protective mechanisms of butyrate

As the preferred energy substrate of colonic epithelial cells, butyrate exerts multiple protective mechanisms in AP that have been directly verified in relevant models. Butyrate upregulates tight junction proteins to maintain intestinal barrier integrity and suppresses the systemic inflammation triggered by bacterial and lipopolysaccharide (LPS) translocation; as a natural histone deacetylase (HDAC) inhibitor, it also downregulates the expression of pro-inflammatory cytokines, induces the differentiation of regulatory T (Treg) cells, and promotes the polarization of macrophages from the pro-inflammatory M1 phenotype toward the anti-inflammatory M2 phenotype[128,129]. The regulation of NLRP3 by butyrate is tissue- and cell-specific: In the cerulein-induced AP model, its inhibitory effect in the colon is mediated mainly by GPR109A, whereas in pancreatic and peritoneal macrophages it is achieved through the HDAC1-STAT1/AP1 pathway[130]. This inhibition is not universal; in LPS-preactivated human macrophages, butyrate and propionate may instead activate NLRP3 and promote IL-1β release by inhibiting HDAC[131]. This indicates that their effects depend on cell type, activation state, and dose.

The indole-AhR-IL-22 axis and barrier protection

Tryptophan metabolism proceeds along three distinct pathways: The host-dominated kynurenine [IDO1/tryptophan 2,3-dioxygenase; see “tryptophan metabolic pathways in AP (the KP and 5-HT pathways)”] and 5-HT pathways, and the microbiota-dominated indole pathway. Upon microbial conversion of tryptophan into indole compounds, these metabolites activate the AhR within the intestinal epithelium, subsequently inducing innate lymphoid cells and T cells to secrete IL-22. In turn, IL-22 acts on the intestinal epithelium to upregulate tight junction proteins and antimicrobial peptides, thereby maintaining intestinal barrier integrity[132]. Population-based metabolomics has corroborated these findings: A fecal and plasma metabolomic study of 92 patients with AP revealed a marked decrease in metabolites from the indole and serotonin pathways, coupled with a significant increase in those from the KP. This shift likely reflects a competitive consumption of the tryptophan substrate driven by IDO1 activation during the disease course, compounded by the suppression of indole production due to microbial dysbiosis. Among these, the levels of metabolites such as indole-3-propionic acid and indole-3-acetic acid correlated strongly and negatively with disease progression, indicating that the depletion of the indole pathway is closely associated with AP deterioration[76].

Targeted intervention strategies for SCFAs and the indole pathway

Supplementation and delivery optimization of butyrate and its derivatives: A large body of preclinical research has provided in vivo and mechanistic evidence supporting the clinical translation of direct butyrate supplementation in AP. In various models of SAP, including the complex setting of SAP accompanied by intra-abdominal hypertension, intragastric administration of butyrate significantly reduced local pancreatic inflammatory infiltration and stabilized intestinal barrier repair[133,134]. This strategy has entered the stage of clinical validation: A phase II randomized, double-blind trial in patients with newly diagnosed AP (ClinicalTrials.gov identifier: NCT06147635) is under way, aiming to assess whether microencapsulated tributyrin can improve plasma toxin levels and thereby advance its translation from preclinical research to targeted therapy in humans[135]. With regard to butyrate-producing probiotics, Clostridium butyricum (particularly the MIYAIRI 588 strain) has been used in Japan and other countries to modulate the gut microbiota, and multiple animal models of AP have confirmed that it significantly attenuates pancreatic inflammation by maintaining intestinal homeostasis and barrier function[134,136]. Among prebiotics, dietary fibers such as resistant starch and inulin can significantly raise the luminal levels of SCFAs such as butyrate by remodeling the microbiota, and have been shown to ameliorate inflammation and barrier injury in models of obesity, metabolic disorder, and obesity-related SAP[137,138]. For GV-971, only a single animal study in SAP has so far suggested that it may act through the microbiota-SCFA-macrophage axis; however, its efficacy in Alzheimer’s disease remains controversial, clinical data in AP are lacking, and the reliability and reproducibility of the proposed mechanism still require independent validation[139].

Translational potential and limitations of indole-targeted therapies: In the sodium taurocholate-induced rat model of AP, the abundance of Akkermansia muciniphila declined, and its metabolite indole-3-lactic acid (ILA) was identified as an important protective molecule. Administration of ILA (50 mg/kg for 3 consecutive days) attenuated intestinal structural damage and reduced the levels of inflammatory cytokines[140]. However, the bioavailability of exogenous supplementation is limited and its effect is short-lived, making it difficult to maintain stable luminal ligand concentrations.

Indole-producing probiotics offer an alternative approach: By colonizing with specific strains, tryptophan is efficiently converted into AhR ligands in situ within the intestine, addressing both microbiota reconstitution and the sustained supply of metabolites, and this has been preliminarily validated in IBD models[141]. However, the relevant evidence derives mainly from idiopathic IBD, whose pathological features differ fundamentally from those of AP; in the field of AP, only a few studies have so far reported that supplementation with Bifidobacterium spp. can lower serum amylase and tissue inflammation scores[142], and whether this effect is mediated through the indole pathway remains unclear.

In summary, SCFAs (chiefly butyrate) and indole compounds are central mediators by which the microbiota protects the intestinal barrier and thereby remotely modulates inflammation in AP. Among them, butyrate supplementation is the strategy supported by the most direct evidence for this pathway, and a phase II randomized controlled trial of microencapsulated tributyrin in newly diagnosed AP is under way; however, acute-phase fasting constitutes a key confounder for the assessment of its circulating levels. By contrast, some of the efficacy evidence for indoles derives from IBD rather than AP, so caution is warranted when extrapolating these findings to AP.

BILE ACID METABOLISM: A CLASSIC HOST-MICROBIOTA CO-METABOLISM

The generation of bile acids relies on cooperation between the host and the gut microbiota. Primary bile acids are synthesized by the host liver from cholesterol, and following deconjugation by bile salt hydrolase (BSH) and conversion by 7α-dehydroxylase, secondary bile acids are produced by the gut microbiota. This dual dependence on host synthesis and microbial conversion distinguishes bile acids both from the endogenous metabolites generated directly by host enzyme systems (part one) and from purely microbial products (part two), establishing this class as the third core category of interest: Host-microbiota co-metabolites.

The dual role of bile acids

In AP, the net effect of bile acids depends primarily on two classes of properties: Their hydrophobicity, which determines their direct toxicity toward membranes and organelles, and their affinity for receptors such as the farnesoid X receptor (FXR) and Takeda G protein-coupled receptor 5 (TGR5), which determines the direction of downstream signaling. The bile acids regarded as harmful are mainly the more hydrophobic secondary bile acids, such as lithocholic acid, which at high concentrations disrupt membrane structure in a detergent-like manner, induce sustained Ca²⁺ overload and mitochondrial depolarization, and thereby trigger acinar cell necrosis[143,144]. However, the influence of hydrophobic bile acids on AP severity is not unidirectional; across different animal models, they may exert opposite effects depending on differences in the hydrophobicity of the bile acid pool and on the specific etiology of the model, indicating that interspecies differences in bile acid pool composition must be considered when extrapolating animal findings to the clinic[145]. By contrast, the hydrophilic ursodeoxycholic acid (UDCA) and its taurine conjugate tauroursodeoxycholic acid (TUDCA) have demonstrated protective effects in various biliary and experimental AP models by alleviating endoplasmic reticulum stress and reducing premature activation of trypsinogen[146]. The two core bile acid receptors operate in distinct modes. FXR is expressed in the ileum, liver, and pancreas, exerting anti-inflammatory effects by inhibiting NF-κB and maintaining intestinal barrier integrity while limiting bacterial overgrowth and translocation; pancreatic FXR overexpression suppresses necrosis and is directly relevant to the prevention and control of infectious complications in AP[147]. TGR5, in contrast, exhibits cell-type-dependent bidirectional regulation: In immune cells such as macrophages and natural killer T cells, it exerts anti-inflammatory effects by inhibiting NF-κB and the NLRP3 inflammasome, whereas in acinar cells its activation can aggravate trypsinogen activation and cell injury through Ca²⁺ dysregulation[148]. This cell-type-dependent directionality is one of the reasons why bile acid signaling exerts dual effects in AP. In AP, gut dysbiosis disturbs the primary-to-secondary bile acid ratio, and this imbalance in turn feeds back on the intestinal barrier and systemic immunity through FXR and TGR5, forming a vicious cycle of bile acids-microbiota-immunity[149,150].

Targeted intervention strategies for bile acid signaling

Targeting FXR and TGR5: Activation of FXR consistently reduces pancreatic pathological scores and pro-inflammatory cytokine levels across various AP models[151]. The agonist activity of obeticholic acid (OCA) at FXR is approximately 100-fold that of the natural ligand chenodeoxycholic acid (CDCA)[152]. A metabolomic study of 326 patients combined with mouse experiments demonstrated that both CDCA and its derivative OCA attenuated acinar injury and pancreatic necrosis, and that pancreas-specific FXR overexpression further enhanced this protection[145]. However, OCA can cause hepatotoxicity, atherogenic dyslipidemia, and severe pruritus, so its clinical use requires caution. The intestine-selective agonist fexaramine has minimal systemic exposure after oral administration and locally activates the FXR-fibroblast growth factor 15 (FGF15) axis in the ileum, thereby circumventing the adverse effects of hepatic FXR overactivation; this offers an alternative path that balances efficacy and safety, but still requires validation in AP models[153,154].

For TGR5, given its cell-type-dependent effects, the ideal strategy would be to achieve selective activation in immune cells. The receptor-selective agonist INT-777 does not yet possess cell-type specificity, but in AP models it is overall protective through inhibition of the ROS/NLRP3 pathway, suggesting that, on current evidence, the net benefit of TGR5 activation is positive[155].

Protective effects of hydrophilic bile acids: As a hydrophilic bile acid, UDCA reduces overall cytotoxicity by remodeling the bile acid pool composition, while also stabilizing mitochondrial membranes and attenuating oxidative stress. Its taurine conjugate TUDCA exhibits even stronger endoplasmic reticulum stress-inhibitory activity[146]. UDCA is already approved for cholelithiasis and primary biliary cholangitis, with ample safety data available[156]. A recent small prospective study suggests potential benefit in high-risk patients, but the sample size is limited and confirmation in large-scale trials is still awaited[157].

Targeting the gut microbiota-bile acid metabolic axis: The probiotic-like strain Blautia coccoides is reduced in patients with AP, and its intragastric administration attenuates pancreatic and intestinal injury in mice; the mechanism involves enhancing BSH activity, thereby activating FXR-FGF15 and inhibiting NF-κB and the NLRP3 inflammasome[158,159]. In recent years, engineered bacteria and postbiotics have provided more controllable alternative routes for this approach: Engineered bacteria can precisely express specific BSH subtypes to achieve targeted regulation, whereas postbiotics avoid the infection risk of live biotherapeutic products and are better suited to the immunocompromised state during the acute phase of AP[160,161]. Although microbiota-targeted interventions in AP (probiotics, fecal microbiota transplantation, engineered bacteria, and postbiotics) hold potential for regulating the “microbiota-bile acid axis”, repairing the intestinal barrier, and exerting anti-inflammatory effects, their clinical translation faces substantial safety barriers. The largest trial to date, PROPATRIA (296 patients with predicted SAP), showed that multispecies probiotic prophylaxis did not reduce infectious complications but instead significantly increased mortality[162]. Several authoritative guidelines accordingly advise against its routine use. Fecal microbiota transplantation carries a risk of transmitting pathogenic bacteria in immunocompromised populations; moreover, existing AP studies have shown no benefit, and animal experiments suggest that it may aggravate the disease. Engineered bacteria remain constrained by unresolved problems such as biocontainment and genetic stability, and although postbiotics circumvent the risks of live bacteria, almost all of the supporting evidence comes from preclinical studies[159]. Given that intestinal barrier injury and bacterial translocation are common during the acute phase of AP, such interventions should be evaluated cautiously in rigorously designed clinical trials, with safety as a prerequisite.

In summary, bile acids, as host-microbiota co-metabolites, exert a net effect that depends on their hydrophobicity (direct toxicity) and affinity for FXR/TGR5 (signaling direction), and is therefore bidirectional. FXR agonists (e.g., OCA) and replacement therapy with UDCA/TUDCA show protective effects in animal models, but OCA raises safety concerns and the clinical evidence for UDCA remains insufficient. Intervention strategies targeting the microbiota-bile acid axis face a major safety warning owing to the negative results of the PROPATRIA trial; although engineered bacteria and postbiotics can circumvent the risks of live bacteria, they remain at an early exploratory stage. Overall, the clinical translation of bile acid metabolism-targeted intervention requires a careful balance between efficacy and safety.

INTEGRATION OF METABOLIC NETWORKS AND THERAPEUTIC IMPLICATIONS

The pathological cascade of the AP metabolic network: From single pathways to network integration

The pathogenesis of AP is not driven by a single molecular event, but rather results from the temporal and spatial interplay of multiple metabolic pathways. This section organizes these metabolic events into a cascading framework of “injury initiation to inflammatory amplification to immune dysregulation to systemic crosstalk via the gut-pancreas axis” to elucidate their interdependencies from a network perspective, as summarized in Figure 1. This framework remains an integrative hypothesis: Although each individual pathway is supported by experimental evidence, the majority of this evidence derives from animal models and ex vivo cell lines, and human data remain relatively limited.

Figure 1 An integrated view of pathological cascades in the metabolic network of acute pancreatitis. This pathological network can be conceptualized into four sequential cascade phases: Injury initiation, inflammatory amplification, immune dysregulation, and gut-pancreatic axis dysregulation. Bile acids induce calcium overload and premature trypsinogen activation, concurrently triggering arachidonic acid release and mitochondrial injury. Functioning as a critical hub, the mitochondrion further amplifies inflammation via mitochondrial permeability transition pore opening, reactive oxygen species attack, and nicotinamide adenine dinucleotide depletion, which collectively drive NOD-like receptor thermal protein domain associated protein 3 inflammasome activation. At the immunoregulatory level, an imbalance emerges between pro-inflammatory mediators (leukotriene B4 and extracellular adenosine triphosphate) and pro-resolving factors (specialized pro-resolving mediators and butyrate). Multiple metabolic pathways cooperatively dictate the macrophage phenotype, establishing these cells as the central node of metabolic-immune crosstalk. Regarding the gut-pancreatic axis, gut dysbiosis leads to the simultaneous depletion of three distinct classes of microbial metabolites short-chain fatty acids, secondary bile acids, and indole derivatives subsequently triggering lipopolysaccharide translocation and fueling a vicious pathological cycle. TGR5: Takeda G protein-coupled receptor 5; PIP2: Phosphatidylinositol 4,5-bisphosphate; IP3: Inositol 1,4,5-trisphosphate; cPLA₂: Cytosolic phospholipase A2; COX: Cyclooxygenase; LOX: Lipoxygenase; Ca²⁺: Calcium ion; IP3R/RyR: Inositol 1,4,5-trisphosphate receptor/ryanodine receptor; LTB4: Leukotriene B4; PGE2: Prostaglandin E2; ATP: Adenosine triphosphate; XO: Xanthine oxidase; ROS: Reactive oxygen species; PARP-1: Poly-adenosine diphosphate ribose polymerase 1; nDNA: Nuclear DNA; NAD⁺: Nicotinamide adenine dinucleotide; mPTP: Mitochondrial permeability transition pore; GSH: Glutathione; GPX4: Glutathione peroxidase 4; HMGB1: High mobility group box 1; eATP: Extracellular adenosine triphosphate; SSAT: Spermidine/spermine N¹-acetyltransferase; H₂O₂: Hydrogen peroxide; NLRP3: NOD-like receptor thermal protein domain associated protein 3; IL: Interleukin; BLT1: Block lipid transport-1; NET: Neutrophil extracellular trap; P2X7: P2X purinoceptor 7; TLR4: Toll-like receptor 4; SPMs: Specialized pro-resolving mediators; HDAC: Histone deacetylase; LPS: Lipopolysaccharide; SCFA: Short-chain fatty acid; FXR: Farnesoid X receptor; AhR: Aryl hydrocarbon receptor.

Injury initiation: Sequential activation of bile acids, calcium signaling, and the lipid cascade. Hydrophobic bile acids (see “the dual role of bile acids”) enter acinar cells via specific receptors and transporters and act on endoplasmic reticulum Ca²⁺ channels such as inositol 1,4,5-trisphosphate receptor and ryanodine receptor, causing a sustained elevation of cytosolic Ca²⁺ that is closely linked to the aberrant early activation of intra-acinar trypsinogen; meanwhile, Ca²⁺-dependent PLA2 and the AA release it mediates participate in the inflammatory amplification of lipid mediators (“the arachidonic acid cascade and eicosanoids”). Bile acid-induced Ca²⁺ overload can also open the mitochondrial permeability transition pore, leading to mitochondrial depolarization and ATP depletion, which impairs Ca²⁺ clearance and the autophagy-lysosomal pathway. Undegraded activated pancreatic enzymes and damaged mitochondria amplify each other, further driving cell death and inflammation. Thus, early injury in AP results from the concerted action of bile acid toxicity, calcium signaling dysregulation, lipid mediator release, and energy metabolism impairment, rather than from any single factor.

Inflammatory amplification: Mitochondria as convergence nodes of multiple metabolic insults. Following injury initiation, the inflammatory amplification phase is marked by the convergence of multiple metabolic pathways at the mitochondrial level, subjecting mitochondria to superimposed insults: Bile acid-induced Ca²⁺ overload promotes opening of the mitochondrial permeability transition pore and collapse of the membrane potential (“the dual role of bile acids”); ROS generated by XO activation attack the mitochondrial membrane (“uric acid and xanthine oxidase”); GSH depletion and GPX4 inactivation contribute to uncontrolled lipid peroxidation (“metabolic basis of oxidative stress”); and PARP-1 overactivation and the upregulation of NAD⁺-consuming enzymes consume NAD⁺, aggravating the energy crisis [“tryptophan metabolic pathways in AP (the KP and 5-HT pathways)].

The downstream consequences are similarly multifaceted: Leakage of mitochondrial DNA (mtDNA) not only activates the NLRP3 inflammasome but also acts as a DAMP. Together with eATP (“adenosine triphosphate and adenosine signaling pathways”) and uric acid (“uric acid and xanthine oxidase”) released from necrotic cells, this mtDNA constitutes a potent set of DAMP signals that drive the maturation and release of IL-1β and IL-18. In addition, the vicious cycle between polyamine metabolic disruption and the autophagy-mitochondria axis (“pathological alterations in polyamine metabolism”) further amplifies oxidative stress and energy imbalance. As the mitochondrion lies at the intersection of these injury pathways, interventions centered on mitochondrial protection may simultaneously alleviate several downstream processes.

Immune regulation: Multi-axis imbalance between pro-inflammatory and pro-resolving mediators. This pro-inflammatory loop is counterbalanced by endogenous pro-resolving mediators, such as SPMs (see “specialized pro-resolving mediators”) and butyrate (see “multiple protective mechanisms of butyrate”). In addition, tryptophan metabolism exhibits a phase-dependent shift: In the early stage of AP, tryptophan is rapidly depleted with concomitant accumulation of 3-HK, which can amplify the innate immune response and contribute to immune dysregulation. As the disease progresses, the KP remains activated and its metabolites accumulate; under tryptophan-deficient conditions, their activation of AhR may become more effective, thereby promoting the differentiation of Tregs and the production of IL-10, driving subsequent immunosuppression[76,163]. Macrophage polarization represents the convergent integration of these metabolic signals at the cellular level: Butyrate promotes M2 polarization by inhibiting HDAC (see “multiple protective mechanisms of butyrate”), SPMs promote efferocytosis (see “specialized pro-resolving mediators”), and lactate participates in reparative macrophage polarization through histone lactylation-mediated reprogramming (see “lactate metabolism and protein lactylation”). Signals from different metabolic pathways jointly determine the macrophage phenotype, making it a central node of metabolic-immune crosstalk.

The gut-pancreas axis: Remote regulation mediated by microbial metabolites. The gut microbiota participates in the remote regulation of AP through three classes of metabolites: SCFAs (chiefly butyrate) reach the pancreas via the circulation and downregulate the NLRP3 inflammasome through GPR109A in the colon and through HDAC inhibition in the pancreas (see “multiple protective mechanisms of butyrate”); Microbiota-catalyzed conversion of primary to secondary bile acids determines the ligand repertoire and signaling strength of the FXR and TGR5 (“the dual role of bile acids”). Indole compounds produced by microbial metabolism of tryptophan maintain barrier integrity through the intestinal epithelial AhR-IL-22 axis (“the indole-AhR-IL-22 axis and barrier protection”). In AP, microbial dysbiosis simultaneously reduces all three classes of metabolites, while the translocation of LPS caused by intestinal barrier disruption in turn aggravates inflammation, forming a vicious cycle. It may therefore be inferred that restoring any single class of metabolite alone may be insufficient to interrupt this cycle, and that an integrated strategy to jointly restore microbial metabolic function may hold greater therapeutic potential, albeit with correspondingly greater implementation complexity (see “supplementation and delivery optimization of butyrate and its derivatives”, “translational potential and limitations of indole-targeted therapies”, and “targeting the gut microbiota-bile acid metabolic axis”).

Design rationale for phase-based combined intervention

The pathological evolution of AP is phase-dependent: The early stage is dominated by the initiation of acinar cell injury, the middle stage is centered on inflammatory amplification and organ damage, and the later stage enters a complex phase in which inflammatory resolution and immune repair coexist. Taking the preceding sections together, it is evident that the activity and pathogenic weight of each pathway differ markedly across disease stages: The same metabolic node may amplify injury in one phase yet promote repair in another. Therefore, combined intervention should follow the principle of “phase matching”, that is, targeting the corresponding metabolic pathway within an appropriate time window. This section sets out the design rationale and the principle of mechanistic complementarity for combined intervention in each phase; the specific strategies are presented in Table 1, and the evidence level and references for each single-agent strategy are given in Table 2.

Table 1 Phase-specific integrated intervention framework for the metabolic network in acute pancreatitis.

Disease phase	Intervention level	Representative strategies	Design rationale
Injury initiation phase (approximately 0-6 hours)	Upstream blockade (source)	UDCA/TUDCA (“protective effects of hydrophilic bile acids”) + taurine (“multifaceted protective effects of taurine”)	Hydrophilic bile acids dilute cytotoxic components while taurine stabilizes Ca2+, providing dual buffering at the source of injury
	Downstream priming (early protection)	NAC (“reconstitution of the GSH system”) + early SPMs (“SPM supplementation and optimization of delivery”)	Pre-stocks GSH antioxidant capacity and deploys pro-resolving signals before the pro-inflammatory storm
Inflammation amplification phase (approximately 6-48 hours)	Upstream: DAMP-signal interruption	Allopurinol/febuxostat (“XO inhibition”) + P2X7 antagonist (“P2X7 receptor antagonism and NLRP3 inflammasome blockade”)	XO inhibition lowers ROS and uric acid; P2X7 antagonism interrupts eATP-NLRP3 signaling
	Midstream: Inflammasome-antioxidant axis	MCC950 (“P2X7 receptor antagonism and NLRP3 inflammasome blockade”) + 4-OI (“activation of the Nrf2-antioxidant response elements pathway”)	Direct blockade of NLRP3 assembly plus Nrf2 activation to attenuate upstream oxidative injury
	Collateral: Gut-pancreas axis protection	Targeted butyrate delivery (“supplementation and delivery optimization of butyrate and its derivatives”) + FXR agonist (“targeting FXR and TGR5”)	Butyrate and FXR jointly repair the intestinal barrier, blocking the second hit of bacterial/LPS translocation
	Phase-sensitive strategy	e.g., short-course tryptophan-pathway inhibition [“tryptophan metabolic pathways in AP (the KP and 5-HT pathways)”]	Brief inhibition during the SIRS phase blocks tryptophan-depletion driven immune dysregulation; a defined stop point is required, as prolonged inhibition may impair later immune tolerance
Immune repair phase (approximately after 48 hours)	Macrophage-polarization reprogramming	Sustained SPMs supplementation (“SPM supplementation and optimization of delivery”) + butyrate/HDAC inhibition (“multiple protective mechanisms of butyrate”) + lactate-histone lactylation (see “lactate metabolism and protein lactylation”)	SPMs supplementation, butyrate-mediated HDAC inhibition, and lactate-driven lactylation jointly drive reparative macrophage polarization
	Restoration of the triple messengers	Combined butyrate- and indole-producing probiotics (“supplementation and delivery optimization of butyrate and its derivatives”/“translational potential and limitations of indole-targeted therapies”/“targeting the gut microbiota-bile acid metabolic axis”)	Synchronous restoration of the three messengers SCFAs, secondary bile acids, and indoles to break the vicious cycle; note that PROPATRIA showed multi-strain live probiotics increased mortality
	Reconstruction of energy homeostasis	NR supplementation (“NAD+ supplementation”) + olaparib/PARP inhibition (“NAD+ supplementation”)	Olaparib inhibits PARP to curb NAD+ consumption, while NR supplies precursors to increase NAD+ availability
	Phase switch	Discontinuation of tryptophan-pathway inhibition [“tryptophan metabolic pathways in AP (the KP and 5-HT pathways)”, “immune repair phase (approximately after 48 hours)”]	Withdrawal after the SIRS peak restores the kynurenine-AhR axis, promotes Treg/IL-10 generation, and establishes immune tolerance

The time windows (e.g., approximately 0-6 hours) are based primarily on the tempo of pathological progression in animal experiments and should be regarded clinically as approximate references rather than precise boundaries; they require individualized calibration according to the dynamic changes of indicators such as C-reactive protein, interleukin-6, bedside index for severity in acute pancreatitis/sequential organ failure assessment, and procalcitonin (see “design rationale for phase-based combined intervention” for details). DAMP: Damage-associated molecular pattern; UDCA: Ursodeoxycholic acid; TUDCA: Tauroursodeoxycholic acid; NAC: N-acetylcysteine; GSH: Glutathione; SPMs: Specialized pro-resolving mediators; XO: Xanthine oxidase; P2X7: P2X purinoceptor 7; NLRP3: NOD-like receptor thermal protein domain associated protein 3; 4-OI: 4-octyl itaconate; Nrf2: Nuclear factor erythroid 2-related factor 2; FXR: Farnesoid X receptor; TGR5: Takeda G protein-coupled receptor 5; AP: Acute pancreatitis; KP: Kynurenine pathway; 5-HT: 5-hydroxytryptamine; HDAC: Histone deacetylase; LPS: Lipopolysaccharide; NAD⁺: Nicotinamide adenine dinucleotide; PARP: Poly-adenosine diphosphate ribose polymerase; Ca²⁺: Calcium ion; ROS: Reactive oxygen species; eATP: Extracellular adenosine triphosphate; SIRS: Systemic inflammatory response syndrome; SCFA: Short-chain fatty acid; AhR: Aryl hydrocarbon receptor; Treg: Regulatory T cell; IL: Interleukin; NR: Nicotinamide riboside.

Table 2 Evidence stratification and translational priority of metabolite-targeted intervention strategies in acute pancreatitis.

Tier	Strategy	Target pathway	Evidence nature and context	Key limitations/caveats	Ref.
Tier 1: Human therapeutic evidence	Glutamine	Amino acid metabolism (“targeting glutamine and protection of the gut-pancreas axis”)	Human therapeutic (RCT/meta-analysis; adjunct to parenteral nutrition in severe AP) + animal therapeutic multiple models3	Hard endpoints (infected necrosis, mortality) not consistently improved; trials small, heterogeneous, high risk of bias	Yong et al[89]; Arutla et al[90]; Tao et al[91]
	Ondansetron	5-HT pathway (“modulation of the 5-HT pathway”)	Human therapeutic (retrospective cohort associative) + animal therapeutic multiple models; antiemetic use well established, safety known	Survival signal only from a single-center retrospective study with indication confounding; needs prospective RCT	Tsukamoto et al[74]; Wu et al[75]
Tier 2: Human data available, but not therapeutic evidence in established AP	Olaparib	NAD+ metabolic repair (“NAD+ supplementation”)	Animal therapeutic multiple models + safety from another indication (approved in oncology)1; PAR accumulation in human samples (mechanistic relevance)	No human efficacy data in AP (only PAR accumulation in human samples)	Ahmad et al[77]; Mota et al[78]
	Montelukast	Pro-inflammatory lipid-mediator inhibition (“targeting the LXA4 pathway and modulating pro-inflammatory mediators”)	Animal therapeutic multiple models + safety from another indication (approved for asthma)	No clinical trial for an AP indication	Ozkan et al[37]
	Clostridium butyricum	Microbiota-SCFAs axis (“supplementation and delivery optimization of butyrate and its derivatives”)	Animal therapeutic multiple models + safety from another indication (approved for microbiota modulation)	Poor acute-phase colonization; no clinical efficacy data; live-probiotic safety caution (PROPATRIA)4	Zhao et al[134]; Xiong et al[136]
	UDCA/TUDCA	Bile acid metabolism (“protective effects of hydrophilic bile acids”)	Animal therapeutic multiple models + human prophylaxis (small prospective study) + safety from other indications (approved for gallstone disease/PBC)	AP evidence limited to small prophylactic studies; no treatment evidence in established AP2; needs multicenter RCTs	Seyhun et al[146]; Hall et al[156]; Maringhini et al[157]
	NAC	GSH-system reconstruction (“reconstitution of the GSH system”)	Animal therapeutic multiple models + human prophylaxis (post-ERCP; one small positive RCT, but most RCTs negative)	Positive in one early small RCT only; pooled/meta-analysis data show no clear preventive effect	Minati et al[109]; Yagci et al[110]; Alavi Nejad et al[111]; Tenório et al[112]
	Allopurinol/febuxostat	XO-ROS-uric acid axis (“XO inhibition”)	Animal therapeutic multiple models + safety from another indication (approved for hyperuricemia); human prophylaxis (post-ERCP RCTs negative)	Post-ERCP prophylaxis RCTs/meta-analyses overall negative; no evidence for treating or improving prognosis of general AP	Rong et al[58]; Han et al[59]; Cao et al[60]; Pekgöz[61]
Tier 3-A: Animal therapeutic evidence multiple models	Butyrate/tributyrin	SCFAs (“supplementation and delivery optimization of butyrate and its derivatives”)	Animal therapeutic multiple models + ongoing trial	Rapid oral absorption, short half-life; phase II RCT (NCT06147635) ongoing, no results yet	Xia et al[133]; Zhao et al[134]; NIH[135]
	SPMs supplementation	Lipid mediators/SPMs (“SPM supplementation and optimization of delivery”)	Animal therapeutic multiple models	Native SPMs: Minute-scale half-life and poor targeting after systemic delivery (central bottleneck)	Liu et al[19]; Lu et al[27]
	ABT-702	Adenosine-signaling enhancement (“strategies to enhance adenosine signaling”)	Animal therapeutic multiple models (consistent in vitro and in vivo SAP)	No human AP data	Sun et al[57]
	MCC950	NLRP3 inhibition (“P2X7 receptor antagonism and NLRP3 inflammasome blockade”)	Animal therapeutic multiple models	No human AP data	Shen et al[52]; Gao et al[53]
	1-Methylspermidine	Polyamine metabolism (“targeted intervention strategies for polyamine metabolism”)	Animal therapeutic multiple models	Highly model-/phase-dependent; benefit confined to pancreas/early phase; renal accumulation and higher mortality at high doses	Hyvönen et al[100]
	FXR agonist	Bile acid signaling (“targeting FXR and TGR5”)	Animal therapeutic multiple models	OCA: Hepatotoxicity and other adverse effects; Gut-selective fexaramine safer but unvalidated in AP	Zhu et al[145]; Zheng et al[151]; Zhang et al[152]
	LXA4	LXA4 pathway (“targeting the LXA4 pathway and modulating pro-inflammatory mediators”)	Animal therapeutic multiple models	No human AP data	Yu et al[32]; Fu et al[33]; Shi et al[34]
	LY293111	LTB4/BLT1 axis (“targeting the LXA4 pathway and modulating pro-inflammatory mediators”)	Animal therapeutic multiple models	No human AP data	Li et al[36]
	KMO inhibitors	Downstream kynurenine pathway (“inhibition of the KMO pathway”)	Animal therapeutic multiple models (improved survival in some models)	No KMO-inhibition clinical trial in AP patients yet	Mole et al[66]; Hayes et al[73]
	Selenium supplementation	GPX4/GPX1 upregulation (“anti-ferroptotic metabolic intervention”)	Animal therapeutic multiple models	Clinically applicable dose and safety window undetermined	Han et al[117]; Swetha et al[118]
	iNOS inhibitors	Arginine-NO axis (“targeting the arginine-NO metabolic axis”)	Animal therapeutic multiple models	Benefits inconsistent across models; no population evidence	Shields et al[92]; Nishino et al[93]
	A2A receptor agonists	Adenosine pathway (“strategies to enhance adenosine signaling”)	Animal therapeutic multiple models	Potential risk of aggravating edema; no population evidence	Celiński et al[55]; Satoh et al[56]
	Emodin	P2X7/Nrf2/NF-κB multi-target [“emodin (a natural P2X7 antagonist and modulator of multiple metabolic pathways)”, “anti-ferroptotic metabolic intervention”]	Animal therapeutic multiple models	Low oral bioavailability; reported hepatic/renal toxicity; insufficient independent validation	Dong et al[62]; Hui et al[63]; Xu et al[64]; Shen et al[119]; Liu et al[120]; Dong et al[121]
	Baicalein	Multi-target [“multi-target natural products (baicalein as an example)”]	Animal therapeutic multiple models	Complex structure activity relationship; low oral bioavailability	Kuhnert et al[38]; Wang et al[39]; Dai et al[40]
	Lactate-metabolism modulation	Lactate/Lactylation (see “lactate metabolism and protein lactylation”)	Animal therapeutic multiple models (bidirectional effect)	Phase-/context-dependent bidirectional effect; modification sites and human relevance unclear	Jiang et al[125]; Han et al[126]
Tier 3-B: Animal therapeutic evidence single model	5-HTP	5-HT pathway (“modulation of the 5-HT pathway”)	Animal therapeutic single model	Single-model evidence; no human data	Wang et al[76]
	4-OI	Nrf2/ARE activation (“activation of the Nrf2-antioxidant response elements pathway”)	Animal therapeutic single model	No clinical translational data in AP	Mills et al[122]; Song et al[123]; Lu et al[124]
	ILA	Indole metabolites (“translational potential and limitations of indole-targeted therapies”)	Animal therapeutic single model	Limited bioavailability and transient effect; hard to maintain stable intraluminal ligand levels	Li et al[140]
	INT-777	TGR5 pathway (“targeting FXR and TGR5”)	Animal therapeutic single model	TGR5 may promote acinar injury; needs immune-cell selective activation/delivery	Li et al[155]
	P2X7 receptor antagonists	eATP-P2X7-NLRP3 axis (“P2X7 receptor antagonism and NLRP3 inflammasome blockade”)	Animal therapeutic single model	Human-murine P2X7 pharmacological differences; needs human-P2X7 selective compounds	Zhang et al[49]; Hoque et al[50]; Fujiwara et al[51]
	Selective mPGES-1 inhibitors	Selective PGE2 inhibition (“targeting the LXA4 pathway and modulating pro-inflammatory mediators”)	Animal therapeutic single model	Limited direct efficacy data in AP	Steinmetz-Späh and Jakobsson[35]
	Resistant starch/inulin (prebiotics)	Microbiota-SCFA (“supplementation and delivery optimization of butyrate and its derivatives”)	Animal therapeutic single model	No human AP data	Baxter et al[137]; Li et al[138]
	Blautia coccoides	Microbiota-bile acid axis (“targeting the gut microbiota-bile acid metabolic axis”)	Animal therapeutic single model	Single-model evidence; live-bacteria safety caution (PROPATRIA)4	Fu et al[158]; Nista et al[159]
	Spermidine (exogenous)	Polyamine metabolism (“targeted intervention strategies for polyamine metabolism”)	Animal therapeutic single model (plus network-pharmacology prediction)	Mostly network-pharmacology prediction/early validation; dose response and safety window undetermined	Shen et al[101]
	GV-971	Microbiota remodeling–enhanced butyrate production (“supplementation and delivery optimization of butyrate and its derivatives”)	Animal therapeutic single model	Single AP study; original indication controversial; reliability/reproducibility need independent validation	Chen et al[139]
Tier 4: Mechanistic extrapolation/no AP model	Nicotinamide riboside	NAD+ precursor supplementation (“NAD+ supplementation”)	Mechanistic extrapolation; no AP model	Combination with PARP inhibitor still conceptual; no direct validation in AP models	Conze et al[79]
	Citrulline (exogenous)	Arginine-NO axis (“targeting the arginine-NO metabolic axis”)	Mechanistic extrapolation (sepsis models) + AP biomarker; no AP model	Evidence from sepsis models and AP biomarker only; no AP efficacy model	Uyanga et al[94]; Chen et al[95]; Goswami et al[96]
	Dapansutrile	NLRP3 inhibition (“P2X7 receptor antagonism and NLRP3 inflammasome blockade”)	Safety from another indication (gout, phase II) + mechanistic extrapolation; no AP model	No direct efficacy data in AP	Li et al[54]
	Colonization with indole-producing probiotics	AhR-IL-22-intestinal barrier (“translational potential and limitations of indole-targeted therapies”)	Mechanistic extrapolation (evidence mainly from IBD models); no AP model	Validated only in IBD models; unconfirmed in AP models	Cui et al[141]; Li et al[142]
Appendix: Negative/cautionary evidence	SAMe	GSH system (“reconstitution of the GSH system”)	Animal negative/cautionary	Alone fails to restore pancreatic GSH, may worsen blockade and inflammation	Anstee and Day[113]; Rius-Pérez et al[114]
	MDL 72527	Inhibition of oxidative polyamine degradation (“targeted intervention strategies for polyamine metabolism”)	Animal negative	Monotherapy did not improve histological injury; blocking degradation by-products alone cannot reverse injury	Alhonen et al[102]; Mandal et al[103]; Räsänen et al[104]

¹Tiers represent an integrated ranking of translational readiness/directness of evidence rather than strength of efficacy in acute pancreatitis alone. Tier 2 ranks above Tier 3 because available human data (prophylaxis or safety from other indications) and an established safety profile bring it closer to clinical trials not because its efficacy evidence for treating established acute pancreatitis is stronger than that of Tier 3.

²The “evidence nature and context” column distinguishes the different evidence types: Human therapeutic (established acute pancreatitis); Human prophylaxis (e.g., post-endoscopic retrograde cholangiopancreatography, biliary); Animal therapeutic multiple models; Animal therapeutic single model; Safety extrapolation from other indications; and mechanistic extrapolation. Three boundaries must be strictly kept apart prophylactic evidence not equal to therapeutic evidence (e.g., N-acetylcysteine, ursodeoxycholic acid); efficacy in animals not equal to efficacy in humans; and approval/safety in other indications not equal to efficacy in acute pancreatitis.

³Tier definitions: Tier 1: Human therapeutic evidence for established acute pancreatitis (randomized controlled trial/meta-analysis/prospective or retrospective cohort); excludes prophylactic or other-indication data. Tier 2: Human data available but limited to prophylaxis, safety from other indications, or ongoing trials (not treatment of established acute pancreatitis), combined with animal therapeutic evidence; therapeutic evidence for established acute pancreatitis remains animal-only. Tier 3A: Animal therapeutic evidence only (post-onset administration), consistent across multiple models, with no human therapeutic data in acute pancreatitis. Tier 3B: Animal therapeutic evidence only (post-onset administration), a single model, with no human therapeutic data in acute pancreatitis. Tier 4: No acute pancreatitis model; evidence derives from models of other diseases (e.g., sepsis, inflammatory bowel disease), safety in other indications, or purely mechanistic extrapolation.

⁴Safety warning for microbiota-based interventions: The largest trial to date, PROPATRIA (296 patients with predicted severe acute pancreatitis), showed that multi-strain probiotic prophylaxis did not reduce infectious complications and significantly increased mortality; several guidelines accordingly do not recommend routine use.

UDCA: Ursodeoxycholic acid; TUDCA: Tauroursodeoxycholic acid; NAC: N-acetylcysteine; AP: Acute pancreatitis; SPMs: Specialized pro-resolving mediators; FXR: Farnesoid X receptor; LXA4: Lipoxin A4; KMO: Kynurenine-3-monooxygenase; iNOS: Inducible nitric oxide synthase; 5-HTP: 5-Hydroxytryptophan; 4-OI: 4-octyl itaconate; ILA: Indole-3-lactic acid; P2X7: P2X purinoceptor 7; mPGES: Membrane-associated prostaglandin E synthase; SAMe: S-adenosylmethionine; NAD⁺: Nicotinamide adenine dinucleotide; 5-HT: 5-hydroxytryptamine; SCFA: Short-chain fatty acid; GSH: Glutathione; GPX4: Glutathione peroxidase 4; XO: Xanthine oxidase; ROS: Reactive oxygen species; NLRP3: NOD-like receptor thermal protein domain associated protein 3; TGR5: Takeda G protein-coupled receptor 5; eATP: Extracellular adenosine triphosphate; PGE2: Prostaglandin E2; AhR: Aryl hydrocarbon receptor; IL: Interleukin; NO: Nitric oxide; PAR: Poly-adenosine diphosphate ribose; RCT: Randomized controlled trial; PBC: Primary biliary cirrhosis; ERCP: Endoscopic retrograde cholangiopancreatography; SAP: Severe acute pancreatitis; IBD: Inflammatory bowel disease; OCA: Obeticholic acid; PARP: Poly-adenosine diphosphate ribose polymerase.

It should be noted that the time annotations below (e.g., approximately 0-6 hours) are based mainly on the rhythm of pathological progression observed in animal experiments and should be regarded clinically as approximate references rather than precise boundaries. In actual clinical practice, the rate of inflammatory progression varies with etiology and severity, and the inflammatory amplification phase is often compressed or advanced in severe cases. Therefore, determining the disease stage solely on the basis of time since onset is unreliable and must be calibrated with measurable indicators. Such indicators serve two functions. The first is severity stratification, to predict whether a patient will rapidly enter the inflammatory amplification phase: The bedside index for severity in acute pancreatitis and sequential organ failure assessment (SOFA) scores can identify a tendency toward severe disease at an early stage, whereas elevated procalcitonin more specifically indicates the risk of infectious complications, helping to distinguish between the two scenarios predominantly sterile inflammation vs established infection that require different intervention priorities[164,165]. The second is phase localization: CRP usually peaks at 48-72 hours after onset, and its dynamic curve can approximately reflect whether the inflammatory amplification phase is still ongoing, whereas IL-6 rises earlier and is suitable for assessing early inflammatory intensity and providing early warning of organ damage[166]. In practice, the dynamic changes in biomarkers should be combined with score trends to make an individualized judgment of the patient's stage, rather than mechanically applying the time boundaries given in this section; for phase-sensitive strategies such as short-course KMO/IDO inhibition, the timing of initiation and discontinuation should be determined by the dynamic trends of indicators such as CRP/SOFA rather than by a fixed number of hours.

Injury initiation phase (approximately 0-6 hours): Source blockade and early protection. The core pathological event of this phase is the sequential activation of the bile acid-calcium signaling-lipid cascade (“injury initiation”), with the predominant link being the early transmission from calcium signaling to mitochondrial injury. Intervention accordingly proceeds along two complementary routes. The first targets the source of injury: Hydrophilic bile acids remodel the composition of the bile acid pool and dilute hydrophobic toxic components, while taurine enhances bile acid hydrophilicity and stabilizes intracellular Ca²⁺ homeostasis corresponding, respectively, to the two pathways of direct cytotoxicity and calcium signaling dysregulation described in “the dual role of bile acids”. The second addresses the oxidative stress that ensues, building up defensive capacity in advance: On the one hand by enhancing the antioxidant capacity of the pancreas, and on the other by administering pro-resolving signals early, so that the balance between pro-inflammatory and pro-resolving mediators is partially restored before the massive release of lipid mediators.

The clinical challenge of this phase lies in capturing the time window: Most patients present only several hours after onset and may already have partially entered the next phase. For high-risk populations (e.g., post-ERCP), prophylactic administration may be the most feasible application scenario for the above strategies.

Inflammatory amplification phase (approximately 6-48 hours): Multi-target anti-inflammatory intervention. The core feature of this phase is the convergence of multiple metabolic pathways on the coordinated activation of mitochondria and the NLRP3 inflammasome (“inflammatory amplification”). Mitochondria simultaneously sustain multiple insults such as Ca²⁺ overload, XO-ROS damage, GSH depletion, and NAD⁺ consumption while the NLRP3 inflammasome integrates various DAMP signals. Intervention accordingly follows a three-tiered, progressive logic. Upstream, interventions intercept the continuous input of DAMP signals, simultaneously reducing ROS and uric acid at the source and blocking eATP-driven inflammasome activation. At the intermediate level, interventions directly inhibit inflammasome assembly while upregulating the transcription of antioxidant genes. As a parallel measure, interventions protect the intestinal barrier to block the secondary insult of bacterial/LPS translocation.

This phase requires attention to the phase-limited nature of intervention. Taking tryptophan as an example, during the window dominated by the systemic inflammatory response syndrome (SIRS), moderate, short-term inhibition of this pathway aims to interrupt the cascade of immune dysregulation induced by tryptophan depletion, thereby protecting organ function and creating conditions for subsequent immune homeostasis. However, such inhibition requires a clearly defined withdrawal time point, as prolonged use may instead interfere with the later establishment of immune tolerance. Determining the timing of withdrawal is precisely what distinguishes the strategy of this phase from traditional continuous anti-inflammatory regimens.

Immune repair phase (approximately after 48 hours): Resolution promotion and homeostasis reconstruction. The goal of this phase shifts from suppressing inflammation to promoting resolution and tissue repair. As noted in “immune regulation”, macrophages serve as the central integrative node of metabolic-immune crosstalk. Mechanistically, HDAC inhibition by butyrate, efferocytosis promoted by SPMs, and lactate-mediated histone lactylation ultimately converge to dictate macrophage M1/M2 polarization. The core strategy during this phase is to drive macrophage phenotypic switching toward a reparative profile through the coordinated input of multilayered metabolic signals.

Concurrently, the three classes of gut-derived messengers described in “the gut-pancreas axis”, are significantly compromised during AP. This specific window where the intestinal barrier has initiated repair but gut dysbiosis remains unresolved represents the optimal opportunity for restoring intestinal metabolic function. Multiple microbiota-targeted strategies can be combined to synchronously replenish these three classes of messengers, thereby breaking the vicious cycle. However, a critical safety caveat must be underscored: Clinical evidence indicates that multispecies live-bacterial prophylaxis not only fails to reduce infectious complications but significantly increases mortality. Therefore, live-bacterial interventions must be evaluated with extreme caution in rigorously designed trials, with safety established as the paramount prerequisite.

Furthermore, two pivotal phase transitions must be achieved during this stage. The first involves reconstructing cellular energy homeostasis via a dual-pronged approach: Mitigating the excessive consumption of NAD⁺ while concurrently supplementing its precursors to bolster supply, thereby restoring energy metabolism. The second transition requires the timely withdrawal of tryptophan-catabolism inhibition once the peak of SIRS has subsided. This allows the kynurenine-AhR axis to regain physiological activation under low-grade inflammatory conditions, promoting the generation of appropriate pools of Treg cells and IL-10 to limit residual inflammation, facilitate tissue repair, and ultimately re-establish long-term immune homeostasis.

The impact of etiological heterogeneity on stratified intervention: Beyond the temporal dimension, etiological heterogeneity also constrains stratified intervention strategies for AP. Different etiological subtypes differ substantially in metabolic background, core injury mechanisms, and potential targets: Biliary AP is centered on bile acid reflux and Ca²⁺ overload, so intervention focuses on regulating bile acids and calcium signaling[167]; HTG-AP is dominated by lipotoxicity, in which the abnormal accumulation of free fatty acids, LPC, and pro-inflammatory lipid mediators is the main driving link, making intervention in lipid metabolism more critical[168]; Alcoholic AP is prominently characterized by mitochondrial dysfunction and oxidative stress, for which energy metabolism and antioxidant strategies are more targeted[169]; and post-ERCP pancreatitis, owing to its predictable time of onset, is better suited to a prophylactic intervention window, such as the administration of rectal NSAIDs or UDCA before or after the procedure[170]. Thus, an ideal translational approach should be a precise design stratified by both etiology and disease phase, rather than a uniform regimen applied to all patients with AP; when interpreting the therapeutic strategies for each metabolic pathway in this review, their applicability and priority should likewise be judged in the context of the underlying etiology.

LIMITATIONS AND FUTURE DIRECTIONS

This review must also frankly acknowledge its own limitations. First, as a narrative review, the selection of literature and the distillation of viewpoints inevitably carry a degree of subjectivity and may be subject to selection bias. Second, the integrated mechanistic evidence is highly heterogeneous at the species and model levels, with many conclusions derived from rodents and in vitro experiments, so caution is required when extrapolating to humans. Third, the frameworks proposed here, the “mitochondrial multi-hit integration node”, the “gut triple messengers” and “phase-based combination intervention”, are conceptual integrations that have not yet undergone prospective clinical validation, and the efficacy and safety of the corresponding combined-intervention regimens still await systematic preclinical and clinical evaluation. Fourth, metabolomic data themselves have inherent limitations in sample handling, inter-platform variation, and reproducibility, which also constrain the robustness of some conclusions. These limitations indicate that the metabolic network framework constructed here is better regarded as a working hypothesis to guide further research than as an established clinical conclusion. To genuinely advance this hypothesis toward mechanism-oriented clinical intervention, several key obstacles must still be overcome. The four most pressing issues are discussed in turn below.

The most fundamental of these is the establishment of causality. Metabolomics largely reveals correlations between metabolites and disease severity, but acute-phase fasting, organ failure, and stress can themselves reshape the metabolic profile; cross-sectional designs cannot distinguish pathological drivers from stress-related consequences, and the net causal effect awaits confirmation in time-resolved models such as acinar cell-specific knockouts. At the same time, the safety boundaries of intervention remain undefined: Supplementing NAD⁺ precursors during the peak of inflammation may aggravate oxidative stress, and administering high-dose butyrate during barrier disruption may promote pathogenic bacteria, so both dose and timing must be carefully considered.

Even when causality has been clarified, determining the timing of intervention remains a challenge. Phase-based intervention uses hours since onset as its coordinate, but the time of onset is often difficult to trace and the rate of progression varies widely between individuals, so guiding clinical practice by elapsed time alone is not feasible. Therefore, it is currently advisable to combine the dynamic changes in biomarkers with score trends to make an individualized judgment of the patient’s stage; in the future, metabolite ratios may replace the time axis for example, using the serum kynurenine/tryptophan ratio to demarcate the immunosuppressive phase and the plasma SPMs/LTB4 ratio to reflect the exhaustion of pro-resolving signals so that, after calibration in longitudinal cohorts, the stage could be determined from a single blood draw.

Delineation of the extent of involvement likewise remains to be clarified. AP often induces multiple organ dysfunction, and its metabolic disturbance is not confined to the gut-pancreas axis: There is as yet no direct evidence as to whether lipid peroxides from pancreatic necrosis involve the liver via the portal vein and create substrate competition between acute-phase protein synthesis and GSH synthesis. Spatially resolved metabolomics is well suited to addressing such questions, but existing studies have largely been confined to static imaging of the pancreas; future work should advance multi-organ, multi-time-point analyses to characterize the evolution of metabolites in the pancreas, liver, lung, and intestine and to determine whether metabolic collapse represents local spillover or systemic dissemination.

The safety of combination strategies requires careful evaluation. Such strategies are based on mechanistic complementarity, but AP is often accompanied by capillary leakage, hypoalbuminemia, and fluctuating renal function, with consequent changes in the volume of distribution and clearance of drugs; whether simultaneous inhibition of oxidative stress and the inflammasome weakens immune surveillance and raises the risk of infection, and whether modulation of coagulation metabolism aggravates hemorrhage in necrotizing AP, also remain to be assessed. Preclinical research should not stop at single-agent validation but should examine the pharmacodynamic synergy, pharmacokinetic interactions, and safety window of sequential combination use.

These four issues are interrelated from causal validation and the determination of intervention timing to the delineation of the extent of involvement and the evaluation of safety, and together constitute the evidence required for metabolically targeted therapy to enter the clinic. For AP, which still lacks a specific drug, the significance of metabolically targeted research lies not in offering new possibilities but in using measurable mechanistic evidence and detectable biomarkers to ensure that every intervention decision is well founded.

CONCLUSION

The core value of the metabolic network framework constructed in this review lies in providing a preliminary, integrated pathological logic for fragmented metabolomic findings and, through concepts such as the mitochondrial integration node and the gut triple messengers, in shifting the therapeutic paradigm for AP from single-target intervention toward phase-based combination therapy. It must be emphasized, however, that this framework remains a working hypothesis awaiting validation. Whether confirming causation from correlation, calibrating phase markers from time of onset, or tracing metabolic disturbances from the gut-pancreas axis to multiple organs, these are all key problems that must be resolved between theoretical conception and clinical practice. For AP, a disease that still lacks specific targeted drugs, the foremost task at this stage may not be to rush these strategies into clinical translation, but rather to use this framework to generate testable hypotheses and to systematically clarify these uncertainties through rigorous prospective studies, thereby accumulating sufficient evidence for rational intervention decisions in the future.