Ursodeoxycholic acid: cholangiocyte protection and clinical evidence

Ursodeoxycholic acid (UDCA, ursodiol) is a naturally occurring bile acid and a medicine with an established role in primary biliary cholangitis (PBC). Its protective effects involve several interacting processes: enrichment of the enterohepatic bile acid pool with relatively hydrophilic species, altered biliary secretion, and responses to cellular stress. The “bicarbonate umbrella” helps explain how the duct epithelium limits contact with membrane-permeant protonated bile acids. These mechanisms do not make UDCA a general remedy for every form of liver injury. PBC, primary sclerosing cholangitis (PSC), pregnancy-associated cholestasis and metabolic steatohepatitis have different evidence bases and clinical outcomes.

The historical connection between ursodiol and bear bile explains its name; modern pharmaceutical production does not require using bear bile. The relevant scientific question is how a change at one carbon atom affects bile acid behavior, and how that behavior interacts with transporters and epithelial physiology. UDCA and its taurine conjugate TUDCA also need to be distinguished: results obtained with TUDCA in isolated cells or rodents cannot automatically be assigned to oral UDCA in people.

Molecular configuration: a C7 epimer, not a membrane-inert substance

UDCA is 3α,7β-dihydroxy-5β-cholan-24-oic acid. Chenodeoxycholic acid (CDCA) has the same molecular formula and the corresponding 3α,7α configuration. They are epimers at C7: the configuration of one stereogenic center differs, while the rest of the specified steroid configuration is retained. UDCA has the formula $\mathrm{C}_{24}\mathrm{H}_{40}\mathrm{O}_{4}$ and molar mass approximately $392.57\ \mathrm{g\,mol^{-1}}$.

The α/β notation describes orientation relative to the steroid framework. It is not synonymous with an axial/equatorial assignment, which depends on ring conformation. The common 5β scaffold is bent, but assigning a universal 90° angle to all physiological bile acids is unnecessary and inaccurate. Some vertebrate bile alcohols and bile acids also have different configurations; the chemistry of human UDCA should not be generalized to every vertebrate bile molecule.

The C7β hydroxyl changes the distribution of polar groups across the steroid surface. Relative to CDCA, UDCA generally interacts less strongly with hydrophobic environments and is less injurious in many experimental membrane systems. Nevertheless, it remains an amphiphilic molecule. Its effects depend on concentration, protonation, conjugation, exposure time, phospholipids, cholesterol and the type of membrane. “Hydrophilic” does not mean that insertion into a membrane or toxicity is physically impossible.

What the Heuman index actually measures

Heuman’s 1989 study derived an empirical monomeric hydrophobicity index from reversed-phase HPLC retention. The scale was normalized to taurocholate at 0 and taurolithocholate at 1. Under the reported conditions, tauroursodeoxycholate had an index of −0.47, taurochenodeoxycholate +0.46 and taurodeoxycholate +0.59. These values refer to specified taurine-conjugated salts. They should not be relabeled as universal constants for the free acids or combined with an unsupported separate TUDCA value of −0.78.

The method describes partition-related behavior within a defined chromatographic system; it is not simply a clinical octanol–water partition measurement. The study also showed that pH particularly affects unconjugated and glycine-conjugated species. Conjugation, ionic strength and protonation must therefore be specified when numerical indices are compared (Heuman, 1989).

A pool index can be represented as a composition-weighted descriptor:

$$HI_{\mathrm{pool}}=\sum_i x_iHI_i,\qquad \sum_i x_i=1.$$

Here $x_i$ is the molar fraction of bile salt species $i$ and $HI_i$ its index measured or assigned on the same scale under compatible conditions. The expression explains why changing composition changes the average descriptor. It does not calculate an individual’s risk of cholangiocyte injury, establish a therapeutic target, or encode the protective contribution of phospholipids and the glycocalyx.

This is a physicochemical comparison, not a ranking of universally safe and unsafe molecules. Lithocholic acid (LCA), for example, has limited aqueous solubility and extensive detoxification pathways; its consequences depend on exposure and metabolism, not just a single index.

Microbial epimerization and enterohepatic pharmacokinetics

In humans, UDCA is a minor bile acid rather than a major product of de novo hepatic bile acid synthesis. Bacterial hydroxysteroid dehydrogenases can interconvert the C7 alcohol and ketone, allowing an epimerization pathway through 7-oxo-lithocholic acid. The nomenclature of this intermediate does not mean that its formation has already removed the C7 oxygen: 7-oxo-LCA retains the C7 ketone.

$$\mathrm{CDCA}\ \xrightleftharpoons[\mathrm{reduction}]{7\alpha\text{-HSDH}}\ 7\text{-oxo-LCA}\ \xrightleftharpoons[\mathrm{oxidation}]{7\beta\text{-HSDH}}\ \mathrm{UDCA}.$$

The reactions require nicotinamide cofactors, with cofactor preferences depending on the enzyme and bacterial strain. A biochemical investigation of Ruminococcus gnavus N53 demonstrated a 7β-HSDH favoring reduction of 7-oxo-LCA to UDCA. This supports a strain-specific route, not a claim that all members of a bacterial genus produce UDCA or that increasing that organism improves a patient’s cholestatic disease (Lee et al., 2013).

Oral UDCA is incompletely absorbed. Passive absorption of the unconjugated molecule and transporter-mediated reclamation of conjugated bile salts contribute at different stages of circulation. The terminal ileal transporter ASBT/SLC10A2 is particularly important for enterohepatic recovery of conjugated bile salts. Blood leaving the intestine reaches the liver through the portal vein; hepatic uptake involves transport systems including NTCP/SLC10A1 and organic anion transporters, with substrate preferences affected by conjugation.

Within hepatocytes, activation to a bile acid–CoA thioester precedes BAAT-mediated conjugation with glycine or taurine. The products are glycoursodeoxycholic acid (GUDCA) and tauroursodeoxycholic acid (TUDCA). Their relative abundance varies with species, nutrition and disease. Human bile commonly contains substantial glycine conjugation, whereas many rodent experiments emphasize taurine conjugates. A fixed 80:20 ratio should not be used as a universal pharmacokinetic constant.

Conjugated species enter bile through canalicular export and circulate repeatedly between liver and intestine. Gut bacteria can deconjugate and further transform them, including conversion into LCA. Sulfation and other metabolic routes reduce the reabsorption and persistence of some potentially injurious species. This is one reason that the administered dose, the biliary UDCA fraction, the serum concentration and the concentration within an experimental cell culture are different quantities. A pool turnover time of several days is not interchangeable with a single plasma elimination half-life.

The formulation, food, gastrointestinal anatomy, bile flow and adherence influence exposure. Exact absorption percentages and first-pass extraction rates from one experimental setting should not be applied to all patients. Product labeling describes the medicine’s pharmacology and its indication-specific use rather than a universal percentage of systemic bioavailability.

Hydrophilic enrichment, mixed aggregates and incomplete protection

During treatment, UDCA conjugates can become a substantial part of biliary bile salts. This lowers the relative contribution of more hydrophobic species and changes interactions among bile salts, phosphatidylcholine, cholesterol and membranes. The fraction achieved is variable. It is not appropriate to promise 50–70% enrichment after precisely two or three weeks in everyone receiving the standard PBC dose.

Enrichment is a dynamic consequence of absorption, conjugation, secretion, intestinal conversion and recirculation. Competition at transporters may occur in particular systems, but the whole process is more than UDCA physically displacing every other bile acid from ASBT, NTCP and BSEP. Changes in endogenous synthesis and clearance, species composition and retention during cholestasis also matter.

Bile contains a mixture of monomers, micelles, mixed micelles and vesicular lipid structures. Critical micellar concentration (CMC) is strongly condition-dependent: conjugation, temperature, ionic strength and accompanying lipids affect self-association. A CMC obtained in a simple aqueous solution is not a threshold above which physiological bile suddenly becomes cytotoxic. Membrane contact and monomer activity can be important below and above that concentration.

Phosphatidylcholine exported by MDR3/ABCB4 is a major protective constituent of bile. Incorporation of bile salts into phospholipid-rich assemblies changes their availability to interact with epithelial membranes. Vesicles and mixed micelles are distinct structures that coexist or interconvert according to composition; neither should be described as a permanently sealed container that makes all residual hydrophobic molecules harmless.

Hydrophilic enrichment can reduce injury in experimental settings and contributes to the rationale for treatment. It does not eliminate mechanical obstruction, neutralize every retained bile acid, or guarantee prevention of bile infarcts. Complete biliary obstruction is a contraindication in ursodiol tablet labeling. A protective molecular effect cannot replace restoration of drainage when an obstructed system requires intervention.

The biliary epithelium and its bicarbonate umbrella

Cholangiocytes line the biliary ducts and modify the fluid delivered from hepatocyte canaliculi. Their apical membrane faces the duct lumen; their basolateral surface communicates with surrounding tissue and blood. This polarity is essential for understanding the direction of ion movement. A canaliculus between hepatocytes and a duct lined by cholangiocytes are related parts of the biliary tree, not the same cellular compartment.

A protective bicarbonate-rich environment near the apical surface can decrease the fraction of bile acids in an uncharged protonated form. The concept proposed by Beuers and colleagues in 2010 was subsequently tested in human cholangiocyte models. Hohenester and colleagues linked bile acid injury to extracellular pH, AE2-dependent secretion and the apical glycocalyx (Hohenester et al., 2012).

The glycocalyx is a hydrated layer of glycoproteins and associated carbohydrates. It helps maintain the epithelial surface environment and limits damaging interactions. Protection also involves tight junctions, membrane composition, fluid secretion and intracellular defenses. The “umbrella” is therefore a physiological model of cooperating barriers, not a literal impermeable sheet.

Secretin signaling in responsive cholangiocytes raises cAMP and activates protein kinase A. This supports CFTR function and the coordinated secretory apparatus. CFTR allows chloride movement across the apical membrane; in the secretory configuration shown here, chloride moves from cytosol to duct lumen. AE2/SLC4A2 exchanges luminal chloride for intracellular bicarbonate, producing chloride influx and bicarbonate efflux. CFTR also has bicarbonate permeability and regulatory interactions in some epithelial settings, so it should not be described as a channel with only one possible physiological role.

$$\mathrm{Cl}^{-}_{\mathrm{lumen}}+\mathrm{HCO}_{3,\mathrm{cytosol}}^{-} \ \xrightleftharpoons{\mathrm{AE2}}\ \mathrm{Cl}^{-}_{\mathrm{cytosol}}+\mathrm{HCO}_{3,\mathrm{lumen}}^{-}.$$

The exchange is electroneutral and does not consume ATP as an ATPase pump. Its net direction depends on ion gradients and cellular conditions. The diagram depicts the direction associated with ductal bicarbonate secretion. CFTR regulation requires ATP-dependent gating, but that does not turn AE2 into an ATP-driven bicarbonate pump.

Functional representation of the apical secretory arrangement. CFTR chloride efflux is distinct from AE2 chloride uptake and bicarbonate secretion. The protein shapes are symbols, not atomic coordinates. Attenuation of the BAH entry arrow means reduced entry, not absolute exclusion of every neutral molecule; the protective effect is incomplete and depends on the surrounding physiology.

Protonation: why a small uncharged fraction matters

For a simplified weak acid equilibrium, the Henderson–Hasselbalch relation gives:

$$\mathrm{pH}=pK_a+\log_{10}\!\left(\frac{[\mathrm{BA}^{-}]}{[\mathrm{BAH}]}\right), \qquad f_{\mathrm{BAH}}=\frac{1}{1+10^{\mathrm{pH}-pK_a}}.$$

BAH denotes the uncharged protonated form, BA⁻ the anion, and $f_{\mathrm{BAH}}$ the protonated fraction. Conjugation changes acidity: taurine conjugates have a much lower apparent pKa than glycine conjugates, while unconjugated bile acids are weaker acids. Apparent pKa values also depend on aggregation and the chemical environment; a value measured in bulk water is not a complete description of the membrane interface.

As an illustrative calculation, a hypothetical bile acid with $pK_a=5.0$ has a protonated fraction of about 0.10% at pH 8.0 and about 0.99% at pH 7.0. This is roughly a tenfold change in the uncharged fraction, even though the overwhelming majority remains ionized at both pH values. It is not a measured UDCA concentration or a clinical cutoff. A glycine conjugate with a lower pKa has a smaller uncharged fraction under the same conditions.

Uncharged bile acids can cross lipid membranes more readily by passive mechanisms than their hydrated anions. Charge reduces passive permeability; it does not abolish every route of cellular entry because transporters and other membrane processes exist. At high total bile salt concentrations, a small permeant fraction can still matter. Secretion, mucus-like surface organization, diffusion and protonation therefore need to be considered together.

After entry, dissociation can affect intracellular acid–base balance. Injury is not explained by acidification alone. Bile acids can alter mitochondrial function, calcium signaling, membrane organization and apoptotic sensitivity. No single surface pH of exactly 7.8–8.2 has been established as a universal value for every human bile duct in health or PBC.

AE2 structure, carbonic anhydrase and secretory regulation

Human AE2 cryo-EM work by Zhang and colleagues resolved inward-facing, intermediate and outward-facing conformations. Each protomer contains a cytoplasmic regulatory region and a membrane domain with 14 transmembrane helices arranged into core and gate regions of a UraA-related fold. The structures showed an N-terminal/transmembrane interlock and inhibitory occupation of the intracellular vestibule by a C-terminal loop. These findings provide a mechanistic basis for pH-sensitive exchange, rather than a simple continuously open pore (Zhang et al., 2023).

The study identified a role for Arg1060 in anion binding and conformational transitions. Residue numbering belongs to the human sequence and construct analyzed; it should not be transferred casually between isoforms or related SLC4 proteins. Structural work on purified or heterologously expressed AE2 establishes transport architecture, but it does not by itself demonstrate direct binding of UDCA or restoration of bile duct function in treated patients.

Carbonic anhydrases accelerate the reversible conversion of carbon dioxide and water to bicarbonate and a proton:

$$\mathrm{CO}_{2}+\mathrm{H}_{2}\mathrm{O} \ \xrightleftharpoons{\mathrm{carbonic\ anhydrase}}\ \mathrm{HCO}_{3}^{-}+\mathrm{H}^{+}.$$

Carbonic anhydrase activity and bicarbonate uptake can supply the intracellular substrate for secretion. Proton handling must accompany this process. The broader idea of a transport “metabolon” describes functional cooperation between bicarbonate transporters and carbonic anhydrases, but enzyme proximity, a physical association and demonstrated substrate channeling are different claims. An obligatory CAII–AE2–CAIV complex that hands each bicarbonate ion directly into AE2 has not been established as the universal UDCA target in human cholangiocytes.

Secretory control includes cAMP/PKA, calcium-dependent pathways, ATP release and purinergic signaling. ATP binds P2 receptors; calcium is a second messenger rather than a ligand that activates P2Y receptors. In different models these pathways influence membrane trafficking, channels and secretion. They should not be compressed into one proven sequence in which every UDCA conjugate induces a fixed three- to fivefold increase in apical AE2 in all patients.

A human liver-cell study found that the combination of UDCA and dexamethasone, rather than either alone, increased alternative AE2b1/b2 expression and exchange activity. Promoter assays implicated interactions among glucocorticoid receptor, HNF1 and p300 (Arenas et al., 2008). This is a defined experimental finding. It is not evidence that routine UDCA monotherapy reproduces the entire transcriptional effect or an instruction to add glucocorticoids to every PBC treatment.

miR-506 and a second consequence of AE2 deficiency

Banales and colleagues found elevated miR-506 in PBC biliary epithelium and demonstrated regulation of AE2 through its mRNA 3′ untranslated region. Overexpression in human cholangiocyte models reduced protein expression and exchange activity; inhibition of miR-506 improved activity in cultured PBC cholangiocytes (Banales et al., 2012). These observations support a connection between post-transcriptional regulation and secretory failure. They do not prove that miR-506 is the sole cause of PBC, or that UDCA directly accelerates miR-506 degradation in patients.

PBC includes immune, genetic, environmental and epithelial components. AE2 deficiency can contribute to vulnerability without explaining all duct loss. The same caution applies to later work linking miR-506 with inflammatory signaling and epithelial changes: an experimental pathway may be relevant to disease while remaining an unvalidated therapeutic target.

AE2 also contributes to intracellular bicarbonate and pH regulation. Chang and colleagues showed that AE2 depletion in H69 human cholangiocytes increased intracellular bicarbonate and soluble adenylyl cyclase (sAC) activity, sensitizing the cells to apoptotic stimuli. Inhibiting or silencing sAC reduced bile salt-induced apoptosis. The experiments implicated intracellular calcium stores, Bax phosphorylation and mitochondrial translocation, and cytochrome c release (Chang et al., 2016).

This creates an important distinction: luminal bicarbonate supports the surface barrier, whereas intracellular bicarbonate can participate in a different signaling compartment. More bicarbonate everywhere is not equivalent to more protection. cAMP generated by soluble cyclase also need not have the same effect as a secretin-associated cAMP signal near the apical secretory apparatus. Intracellular location and downstream effectors matter.

Mitochondrial apoptosis: separate the outer membrane from permeability transition

Hydrophobic bile acids can activate stress pathways and promote mitochondrial injury in experimental hepatocytes. Bax and Bak regulate mitochondrial outer-membrane permeabilization (MOMP), allowing release of proteins such as cytochrome c. Cytochrome c contributes to Apaf-1-dependent apoptosome formation and caspase-9 activation, followed by downstream executioner caspases. The availability of these pathways depends on cell state and the stimulus.

The mitochondrial permeability transition pore (mPTP) involves an increase in inner-membrane permeability. It can dissipate the membrane potential, disturb energy production and produce swelling, but it is not identical to a Bax/Bak pore in the outer membrane. Cyclophilin D is an important regulator; describing ANT plus cyclophilin D as an exhaustively established pore structure oversimplifies continuing structural and functional questions. The distinction is developed in the article on mitochondrial permeability transition and cyclophilin D.

Rodrigues and colleagues reported that UDCA reduced mitochondrial depolarization, cytochrome c release and apoptotic features in primary rat hepatocytes exposed to selected injurious stimuli. The effects were substantial but incomplete, and the model did not demonstrate universal preservation of every hepatocyte under cholestasis (Rodrigues et al., 1999).

The Bax solution structure shows an inactive arrangement in which its C-terminal α9 helix occupies a hydrophobic groove (Suzuki et al., 2000). That structure explains a regulatory feature of Bax. It does not contain TUDCA and cannot establish a TUDCA binding affinity or a structurally resolved drug–Bax complex. Reduced Bax mitochondrial accumulation after treatment is a functional observation, not proof of a specific small-molecule docking site.

Survival kinase pathways provide another level of regulation. PI3K generates phosphoinositide signals that recruit and help activate Akt; Akt can phosphorylate proapoptotic Bad, permitting its sequestration by 14-3-3 and altering interactions with antiapoptotic Bcl-2 family proteins. This is a well-established signaling framework, but demonstration of individual steps depends on the experimental model.

In cultured rat hepatocytes, Schoemaker and colleagues found TUDCA protection against glycochenodeoxycholate-induced apoptosis involving PI3K, p38 and ERK pathways. In that model GUDCA did not reproduce TUDCA’s protection, and the effect was independent of several proposed alternatives, including NF-κB activation and competition at the cell membrane (Schoemaker et al., 2004). The result illustrates why UDCA, GUDCA and TUDCA should not be treated as interchangeable reagents.

TUDCA, ER stress, p53 and autophagy

The unfolded protein response (UPR) helps the endoplasmic reticulum adapt to increased folding demand through IRE1, PERK and ATF6 pathways. BiP/GRP78 is an adaptive chaperone, whereas persistent stress can activate pathways involving CHOP/DDIT3 and cell death. A lower stress-marker level after an intervention may indicate reduced ER stress, but it is not equivalent to direct binding and stabilization of each marker protein.

Özcan and colleagues studied chemical chaperones, including TUDCA, in cells and obese diabetic mice and observed improved ER-stress and metabolic measures (Özcan et al., 2006). This work supports a mechanistic research program. It does not establish TUDCA as a universally effective human diabetes or MASH therapy, nor show that oral UDCA at a PBC dose necessarily produces the same exposure and response.

Calling TUDCA a chemical chaperone describes its ability to improve folding-related homeostasis in experimental systems. It need not imply a resolved direct interaction with every misfolded protein or with BiP, PERK and CHOP. The concentrations used in a culture dish or the doses used in mice must be kept separate from prescribed human doses.

UDCA/TUDCA have also been studied in relation to p53-dependent apoptosis and autophagic flux. p53 can activate BH3-only mediators such as PUMA and NOXA, but these interactions depend on cell type and injury. Changes in p53 abundance or activity in a model do not establish a universal UDCA–MDM2 mechanism. Similarly, TFEB localization, LAMP proteins and mitochondrial quality-control pathways are useful experimental readouts, but they do not establish a complete TUDCA→TFEB→LAMP2→PINK1/Parkin pathway in human cholangiocytes.

Autophagic flux means the progression from cargo sequestration through lysosomal degradation. A higher level of LC3 or more autophagosomes can mean increased production or blocked disposal. Claims of restored flux require experiments that distinguish these alternatives, such as appropriate lysosomal inhibition or dynamic reporters. This qualification preserves the relevance of autophagy without turning individual marker changes into proof of complete organelle clearance.

Membranes, tight junctions and the limits of general cytoprotection

Membrane composition and epithelial junctions contribute to duct integrity. Hydrophobic bile acids can disturb lipid organization, cytoskeletal structures and tight-junction function. UDCA-related changes may mitigate some of these effects in experimental settings. However, a precise obligatory stoichiometric insertion of UDCA with cholesterol and sphingomyelin, or universal blockade of vinculin and occludin phosphorylation, requires direct evidence for that system and should not be inferred from a general protective effect.

The epithelium also responds to inflammation, pressure, infection and loss of transporter function. Hydrophilic enrichment cannot fully compensate for all of these insults. It is particularly misleading to describe treated cholangiocytes as permanently resistant to millimolar detergent exposure: total concentration, freely available monomers, conjugation and the local surface environment differ between bile and a cell-culture medium.

Hepatocyte secretion: BSEP, MDR3 and MRP2 have different jobs

BSEP/ABCB11 exports bile salts across the canalicular membrane. MDR3/ABCB4 transfers phosphatidylcholine to the outer leaflet, enabling its release into bile. MRP2/ABCC2 exports organic anions including bilirubin glucuronides and other conjugates. Their functions cooperate but are not interchangeable; “more bile export” does not describe the substrate specificity of each transporter.

Experimental studies of TUDCA have implicated calcium signaling, protein kinases and regulated insertion of transporter-containing vesicles into the canalicular membrane. These are especially relevant when cholestatic stimuli disturb trafficking. Transporter expression, membrane localization and net transport activity are distinct measurements, and a fixed increase within 15–30 minutes should not be stated as a general human response.

Integrin-associated signaling provides one investigated mechanism. Work on α5β1 integrin and related UDCA derivatives has linked conformational activation to kinase pathways and Bsep localization in experimental hepatocyte/liver systems. The studied derivative, the cellular compartment and the model must be named; molecular dynamics or inhibitor experiments are not equivalent to direct demonstration of every downstream branch in treated human liver (Gohlke et al., 2013).

Distinct hepatocyte export functions feed the extracellular biliary space, while cholangiocyte AE2 contributes to bicarbonate secretion at the duct surface. The shapes are functional symbols, not atomic structures. Colored groups in the lumen represent mixed bile contents: phosphatidylcholine and bilirubin are not bile acids and are excluded from their molar fractions and Heuman index. The map combines complementary physiological processes and does not claim that UDCA directly binds each transporter or guarantees a clinical response.

UDCA is not a potent endogenous FXR agonist in the same sense as CDCA. Changes in the bile acid pool can nevertheless affect receptor signaling indirectly. FXR, intestinal FGF19, hepatic synthesis and transport form a regulatory network; the absence of strong direct FXR agonism does not mean that every UDCA effect is necessarily non-genomic.

Hypercholeresis and the distinction from norUDCA

Bile salt export contributes to osmotic water movement, and bicarbonate secretion modifies ductal fluid. Cholehepatic shunting describes reabsorption of a biliary molecule through duct epithelium and return to hepatocytes without first completing a full intestinal circuit. It is particularly relevant to bile acid derivatives that remain relatively unconjugated and can repeatedly stimulate ductal secretion.

The side-chain-shortened compound 24-norUDCA is resistant to amidation relative to UDCA and has distinctive cholehepatic cycling and bicarbonate-rich hypercholeresis in experimental studies. Work in Mdr2-deficient mice showed that norUDCA and its modified derivatives did not share identical actions (Fickert et al., 2006; Halilbasic et al., 2009). These findings should not be transferred wholesale to ordinary UDCA, which undergoes extensive conjugation.

Increased bile flow may help dilute luminal contents, but “hydraulic flushing” is a model-level description. It does not prove that treatment removes obstructing stones, strictures, bacterial infection or protein plugs, or restores patency of an anatomically obliterated duct. Fluid secretion can contribute to protection while drainage remains impaired.

Immune and inflammatory effects: neither invisibility nor cure

PBC involves immune responses to mitochondrial antigens, particularly the E2 component of the pyruvate dehydrogenase complex (PDC-E2). This refers to immune recognition of a protein; it is different from the regulation of oxidative carbohydrate metabolism discussed in the article on PDH and metabolic flexibility.

A small biopsy comparison by Calmus and colleagues found lower aberrant hepatocyte HLA class I expression in UDCA-treated PBC patients. Aberrant biliary HLA class II expression was not changed (Calmus et al., 1990). This does not support a claim that all HLA-I/II molecules disappear from bile duct cells or that treated epithelia become invisible to T lymphocytes.

Inflammation, bile acid injury and immunity can influence one another. Changes in cytokines, immunoglobulins or cellular stress may accompany treatment, but a lower IgM concentration is not proof that the autoimmune process has been eradicated. NF-κB can be protective or injurious depending on cell type, stimulus and timing. One experimental demonstration of cytokine modulation does not establish uniform suppression of TNF, IL-1, IL-6 and interferon-γ in every patient.

PSC is also an immune-associated disease, frequently linked to inflammatory bowel disease, but it is not simply PBC affecting larger ducts. Differences in anatomy, fibrosis, infection, cancer risk and disease biology prevent direct extrapolation of the clinical benefit of UDCA in PBC to PSC.

PBC: first-line treatment and what survival evidence shows

In the 1991 two-year placebo-controlled trial, 146 patients were randomized equally to UDCA or placebo. Treatment failure occurred in 6 versus 13; the definition included a specified bilirubin increase, severe complications or adverse reactions. Histological improvement did not include demonstrated improvement in fibrosis. This concrete trial supports biochemical and selected progression effects without making its composite identical to mortality (Poupon et al., 1991). The historical PBC guideline framework should be read together with current labeling and the later regulatory changes discussed below (EASL, 2017).

For adults with PBC, 13–15 mg/kg/day is the established indication-specific UDCA dose in guidelines and product labeling. This is a description of standard therapy, not an individualized prescription or a dose to transfer to other populations. Formulation-specific schedules and clinical circumstances determine how a prescribed total is divided.

Early randomized trials demonstrated improvements in liver biochemistry and aspects of disease progression. Evaluating mortality or transplantation in a slowly evolving disease requires long follow-up and adequate event counts; many historical trials were limited for these outcomes. The survival evidence should therefore be explained using both the trials and later observational cohorts rather than attributing every conclusion to definitive placebo-controlled mortality trials.

Harms and colleagues analyzed 3,902 patients in an international PBC cohort and found an association between UDCA treatment and improved transplant-free survival using inverse-probability weighting. The association persisted among patients with incomplete biochemical response (Harms et al., 2019). The design strengthens the clinical evidence but cannot eliminate all residual confounding. It also argues against treating biochemical nonresponse as proof that UDCA has no value and should automatically be stopped.

Biochemical response is useful for risk stratification. It does not mean that the disease is cured, that fibrosis cannot progress, or that every responder has the same risk as the general population. Disease stage, bilirubin, albumin, platelet count, age and other clinical features remain relevant.

Paris criteria, GLOBE and UK-PBC: prediction without promised life-years

Paris I and Paris II classify biochemical findings after a year of UDCA treatment. Paris II was developed for early-stage PBC. The definitions have different populations and endpoints, and the ALP threshold used in a modern drug trial is not interchangeable with either definition (Corpechot et al., 2008; Corpechot et al., 2011).

ULN denotes the laboratory’s upper limit of normal; ALP is alkaline phosphatase, AST aspartate aminotransferase. A bilirubin value of 1 mg/dL corresponds to approximately 17.1 μmol/L, but “normal bilirubin” must use the relevant laboratory reference range. Different thresholds should not be mixed into a newly invented response criterion.

GLOBE and UK-PBC are derived and validated prediction systems (Lammers et al., 2015; Carbone et al., 2016). Their equations require specific units, normalization and timing of measurements. They do not prove that achieving ALP below 1.67×ULN adds precisely 10–12 years of life. A prognostic association is not a causal dose–response equation; risk estimates also have uncertainty and depend on applicability to the person being assessed.

Incomplete PBC response and second-line medicines

Second-line therapy is considered when response to appropriate UDCA treatment is inadequate or the medicine cannot be tolerated. Choice depends on the approved indication, liver disease stage, symptoms, interactions and monitoring. Experimental receptor biology explains complementarity but does not identify one universally safest combination.

BEZURSO studied 100 patients with inadequate UDCA response, adding bezafibrate or placebo for 24 months. Complete biochemical response occurred in 31% versus 0%, and ALP normalization in 67% versus 2%. The composite included bilirubin, ALP, aminotransferases, albumin and prothrombin index; it was not a trial proving a mortality benefit. Changes in creatinine and myalgia also matter when evaluating tolerability (Corpechot et al., 2018). Bezafibrate is a pan-PPAR agonist rather than simply equivalent to the more PPARα-selective fenofibrate; availability and off-label use vary by jurisdiction.

Seladelpar is a PPARδ agonist. In RESPONSE, the primary biochemical response occurred in 61.7% of treated patients versus 20.0% with placebo, with ALP normalization in 25.0% versus 0.0%; a prespecified itch endpoint also improved in the relevant symptomatic subgroup (Hirschfield et al., 2024). These findings concern the trial’s population, treatment duration and endpoints.

Elafibranor is a dual PPARα/δ agonist, not an α/γ agonist. ELATIVE found a biochemical response in 51% versus 4% and ALP normalization in 15% versus 0%; the reported itch comparison was not statistically significant (Kowdley et al., 2024). These trials do not support a head-to-head ranking or a claim that lowering ALP has already proved fewer deaths or transplants for every new medicine.

Obeticholic acid, an FXR agonist, is historically important because POISE demonstrated biochemical effects, but its current status differs from its earlier approval. As checked on 8 October 2026, EMA records show revoked EU authorization, and the FDA’s official withdrawal record documents withdrawal of US approval in 2025. It should not be presented as a current routine approved second-line option in those jurisdictions (EMA, 2024; FDA, 2025). Regulatory status elsewhere and individual treatment decisions require the applicable current local information.

PSC: laboratory improvement is not a survival endpoint

PSC provides a clear counterexample to extrapolation from protective mechanisms. In a randomized trial of 150 adults, high-dose UDCA at 28–30 mg/kg/day improved some liver tests but was associated with worse clinical outcomes. Prespecified endpoints occurred in 39% of the UDCA group versus 26% of the placebo group; the adjusted risk of the primary endpoint was 2.3 times higher. Serious adverse events were also more frequent (Lindor et al., 2009).

The composite included cirrhosis, varices, cholangiocarcinoma, transplantation and death. It should not be rewritten as a trial proving that LCA accumulation caused colorectal cancer. Altered bile acid metabolism is a proposed explanation, and colorectal neoplasia has been studied separately. A plausible mechanism does not establish the cause of the harm signal.

Guidelines allow nuanced discussion of selected moderate-dose use. AASLD guidance permits consideration of 13–23 mg/kg/day in selected patients with persistent ALP/GGT elevation, particularly when they are not participating in a trial, with continuation conditional on tolerability and meaningful biochemical or symptomatic improvement within 12 months. EASL guidance states that 15–20 mg/kg/day can be given because it may improve liver tests and surrogate markers (Bowlus et al., 2023; EASL, 2022). These conditional positions are not a unanimous proven “safe and effective corridor” or evidence of improved transplant-free survival. High-dose treatment should not be inferred from poor biochemical response.

UDCA is not established chemoprevention for cholangiocarcinoma or colorectal cancer in PSC. Management of strictures, bacterial cholangitis, inflammatory bowel disease and cancer surveillance remains distinct from changing the bile acid pool.

Pregnancy-associated cholestasis: maternal symptoms and fetal outcomes differ

The threshold distinction is supported by the primary aggregate and individual-participant analysis of pregnancy outcomes; its risk categories concern peak bile acids and the relevant pregnancy population. It does not establish a universal drug-response cutoff or show that lowering a laboratory result necessarily prevents stillbirth (Ovadia et al., 2019).

Intrahepatic cholestasis of pregnancy (ICP) is associated with maternal itching and elevated bile acids. The relationship between bile acid concentration and fetal risk depends on the degree of elevation, gestational age and other features. The highest bile acid categories, particularly ≥100 μmol/L, have important prognostic implications. A single value of 40 μmol/L should not be described as producing the same stillbirth risk as 100 μmol/L.

UDCA is used for maternal symptoms in relevant obstetric guidance, but symptom treatment and prevention of adverse perinatal outcomes are different questions. The PITCHES randomized trial enrolled 605 women. The primary composite of perinatal death, preterm birth or neonatal unit admission occurred in 23% of infants with UDCA versus 27% with placebo; the adjusted risk ratio was 0.85 with a 95% confidence interval of 0.62–1.15. This did not demonstrate a reduction in the primary outcome (Chappell et al., 2019).

No treatment-related serious adverse events were identified in that trial, which is reassuring within the studied setting. It is not proof of absolute embryonic safety at every gestational age or a guarantee that fetal risk has been normalized. Rare events and subgroup effects require appropriately powered evidence. Placental or fetal cardiac experiments can help investigate mechanisms but cannot substitute for clinical outcome data.

Treatment decisions, repeat measurements and timing of birth belong to obstetric care. A fall in enzymes or an improvement in itching does not remove the need for that assessment, and the PBC dose or follow-up schedule should not be transferred automatically to pregnancy.

Gallstones, rapid weight loss and bile reflux

UDCA can lower biliary cholesterol saturation and facilitate dissolution of selected cholesterol gallstones. A simple saturation descriptor is:

$$CSI=\frac{\text{measured biliary cholesterol concentration}} {\text{equilibrium cholesterol solubility at the same bile composition}}.$$

CSI is dimensionless; values above 1 indicate supersaturation relative to the specified equilibrium system. The denominator depends on bile salts, phospholipids and conditions, so this is an explanatory relation rather than a bedside calculation using a routine serum cholesterol result.

Selected noncalcified, radiolucent gallbladder stones can be candidates for dissolution, particularly when surgery is inappropriate or declined. Capsule labeling specifies stones smaller than 20 mm, but response varies, may take months and can be followed by recurrence. Radiolucency does not guarantee pure cholesterol composition. A universal CT attenuation cutoff or gallbladder ejection fraction should not replace a clinical assessment of suitability.

Acute cholecystitis, cholangitis, biliary obstruction or gallstone pancreatitis are different conditions and are not situations in which dissolution therapy can be assumed to replace needed intervention. “All doses must be taken strictly at bedtime” is also not a general rule across products and indications; capsule and tablet labels have different schedules.

After bariatric surgery, prevention of new stones is distinct from dissolution of existing stones and from prevention of symptomatic disease. UPGRADE tested 900 mg/day for six months. In the overall trial population, symptomatic gallstone disease occurred in 6.5% versus 9.7%, with a confidence interval crossing no effect; prespecified analyses supported benefit in patients without pre-existing stones undergoing Roux-en-Y gastric bypass (Haal et al., 2021). These findings do not justify promising the same reduction after every operation, particularly when sleeve-gastrectomy numbers were limited.

Bile reflux into the stomach has a different anatomical and symptom context. UDCA has been investigated in selected settings, including after gastrectomy, but a universal regimen of 250–500 mg at night for every patient with reflux gastritis is not established by the PBC evidence. Endoscopic findings, symptoms and the cause of reflux need separate evaluation.

MASLD/MASH, drug injury and pediatric cholestasis

Metabolic dysfunction-associated steatotic liver disease (MASLD) and steatohepatitis (MASH) are not established indications for UDCA as a disease-modifying treatment. A randomized trial of 166 patients with biopsy-proven NASH found no significant histological advantage over placebo after two years at 13–15 mg/kg/day (Lindor et al., 2004). Changes in ALT alone are not proof of fibrosis reversal. Preclinical TUDCA ER-stress findings cannot override this population-specific clinical evidence, and combinations with metformin or vitamin E should not be asserted to have established universal synergy.

Drug-induced liver injury (DILI) likewise includes different mechanisms and phenotypes. UDCA may be used by specialists in selected cholestatic situations, but robust evidence of improved hard outcomes is limited. It is not proven to universally accelerate elimination of the causative drug or restore a blocked BSEP pathway. Identifying and withdrawing the implicated agent, assessing severity and considering alternative diagnoses remain separate clinical tasks.

Cystic fibrosis alters ductal secretion through CFTR dysfunction, but restoring bile flow experimentally does not establish prevention of advanced CF liver disease. The Cystic Fibrosis Foundation’s consensus recommendations advise against routine UDCA use to prevent advanced liver disease in all people with CF (Sellers et al., 2024). Biochemical improvement, portal hypertension and liver-related outcomes must be separated; CF-associated liver disease can also include vascular pathology beyond a simple mucus-plug mechanism.

In biliary atresia, UDCA is sometimes part of specialist postoperative care, but it cannot reopen an obliterated extrahepatic biliary tree. Evidence for an independent transplant-free survival benefit is limited and should not be attributed to a combined postoperative regimen containing surgery, antibiotics and other medicines without isolating the UDCA effect.

Progressive familial intrahepatic cholestasis (PFIC) includes genetically distinct disorders. ATP8B1, ABCB11 and ABCB4 defects affect different functions. Residual transporter activity and variant severity help explain variable response; a medicine that changes bile acid composition cannot restore an absent transporter. ABCB4-associated disease may respond in some patients with residual function, but UDCA should not be described as making phospholipid-free bile harmless or guaranteeing avoidance of liver failure. Genotype-specific therapy, surgical options and transplantation are separate aspects of care.

Beyond the liver: a research field, not a universal tissue shield

TUDCA has been investigated in retinal, neurologic, cardiac and renal injury models. Findings involving oxidative stress, mitochondrial function, protein aggregation or ER stress are biologically interesting. They require the same distinction between compound, exposure, species and clinical endpoint used throughout hepatology.

An animal finding of tissue penetration is not proof that oral TUDCA reliably reaches therapeutic concentrations throughout the human brain by OATP1A2 transport. Likewise, changes in Nrf2-regulated genes do not demonstrate that TUDCA directly oxidizes specified Keap1 cysteines or activates every element of an antioxidant transcriptional program. Nrf2 and the nuclear respiratory factors termed NRF1/NRF2 are distinct regulatory systems despite confusingly similar abbreviations.

PGC-1α/TFAM, antioxidant enzymes and PINK1/Parkin are meaningful pathways to study, but an experimentally altered marker does not prove mitochondrial biogenesis or effective mitophagy unless function and flux are measured. The same applies to amyloid, tau, α-synuclein and other proteinopathy models: protection against an experimental insult does not establish approved treatment of Alzheimer’s disease, Parkinson’s disease or motor neuron disease. Human randomized outcomes must determine whether the molecular rationale produces useful treatment effects.

Tolerability, interactions and indication-specific interpretation

UDCA is generally well tolerated when appropriately prescribed, but diarrhea and other gastrointestinal effects can occur. Frequencies depend on population, dose and formulation. Hypersensitivity and complete biliary obstruction are relevant contraindications in labeling, and adverse effects cannot always be handled by simple dose titration without considering their cause.

Bile acid sequestrants can reduce absorption, and aluminum-containing antacids may interfere with exposure. Product-specific separation recommendations should be followed rather than describing all simultaneous use as absolutely contraindicated or all adsorption as irreversible. Medicines that change bile composition, and other interactions requiring monitoring, should be evaluated within the actual indication.

The useful scientific model is a set of interacting mechanisms constrained by indication-specific evidence: a relatively hydrophilic bile acid pool, distinct hepatocyte export functions, a polarized cholangiocyte barrier and context-dependent stress responses. Each can contribute to benefit without guaranteeing the clinical outcome of a different disease.

Evidence scope and limitations

This scientific educational account was checked on 8 October 2026 using targeted primary publication records and indexed abstracts, available publisher structural results, official prescribing information, relevant guideline passages and regulator records. Full text was not accessible for every cited clinical or mechanistic study. Cell and rodent experiments are identified as such, observational survival associations are distinguished from randomized endpoints, and biochemical surrogates are distinguished from patient-important outcomes. This is not a systematic review, a formal evidence grade or an individualized treatment protocol.

Particular care is required with bile acid conjugation, transporter direction, structural identifiers, the distinction between MOMP and mPTP, and the different evidence for PBC and PSC. The equations explain composition and acid–base principles; they are not validated calculators for prescribing UDCA or predicting an individual outcome.