2.1 Liver Pathology in Refractory Celiac Disease

Refractory celiac disease (RCeD) is a rare complication of CeD, defined as the persistence or recurrence of malabsorption and villous atrophy despite adherence to GFD for at least 12 months, once other causes of villous atrophy and malignancy have been excluded [52-55]. The estimated cumulative incidence of RCeD is about 1%, affecting around 0.01% of the population [56, 57].

There are two subtypes of RCeD, type I and type II, according to duodenal histological, molecular (clonal T-cell receptor-γ, TCRγ, rearrangements) and flow cytometry (detection of aberrant phenotype intraepithelial lymphocytes, aIELs) features [56, 58]. Particularly, RCeD type II is now considered a pre-T-Cell lymphoma with an “in situ” clonal expansion of aIELs and a 33% to 53% risk of developing enteropathy-associated T-cell lymphoma (EATL) at 5 years from the diagnosis [56, 59, 60]. RCeD type I has a better prognosis, typically characterised by a less severe intestinal insufficiency and lower risk of malignant complications [57]. No curative treatment is currently available for RCeD. To date, therapy is based on immunosuppressive (steroids, thiopurines, infliximab, anti-Janus Kinase) or ablative treatments (cladribine, high-dose chemotherapy followed by stem cell transplantation) and nutritional support [15].

The frequency of liver affection in RCeD is unknown because of the rarity of the condition. Most data on RCeD and its complications come from retrospective studies, case series and case reports. Similarly to responsive CeD [61, 62], in RCeD, there is a higher incidence of autoimmune diseases. For instance, Weber et al. [63] have observed in their cohort of 46 RCeD that 20 (43.5%) of them had another AI disease, including 1 patient with AIH (2.3%). A similar finding was presented in an Italian retrospective study [64] with 13 out of 37 RCeD patients (35%) with an autoimmune disease, of which 3 (8%) had autoimmune liver disease. However, the nature of these studies and their sample size pose major limitations, and further studies are required to confirm these findings. Another form of liver involvement in this group of patients, specifically in RCeD type II, could be related to liver enzymes' alterations due to aIELs migration to the liver [65]. Indeed, in RCeD type II, aIELs have been observed also in the duodenal lamina propria [66-69]. They are not only disseminated in the small intestine, but they can also have extra-intestinal localisations. Lymphocytes presenting the same aberrant phenotype of duodenal IELs have been detected in blood flow cytometry [66, 69], as well as at the level of the skin, lungs and liver [70]. Specifically, in a case report, the authors described aIELs parenchymal infiltrate with the same phenotype as those found in the duodenal mucosa of a patient with a recent diagnosis of RCeD type II [71]. Therefore, altered liver function tests in RCeD with suspected EATL should raise the suspicion of a possible hepatic disease localisation. This rare T-cell lymphoma is diagnosed half of the time in an emergency setting because of perforation, occlusion or haemorrhage and at a late stage, stage III/IV, with liver metastasis in 40%–60% of cases [72-74]. The primary site of EATL location is the small bowel in 90%–95% of cases, often with multifocal intestinal involvement. However, in a small portion of patients, EATL can present on extraintestinal sites, most frequently from RCeD type II skin lesions [75].

2.2 Management of Liver Function Test Alterations in Celiac Disease and Refractory Celiac Disease

The initial approach to patients with incidental elevation of transaminases (> 2–3 times the upper limit of normal, ULN) of unclear origin includes CeD-specific serology and the exclusion of the most frequent causes such as medications, hepatitis B and C (HBsAg and anti-HCV), hemochromatosis (serum ferritin and transferrin saturation), AIH (anti-nuclear antibodies, ANA, and anti-smooth muscle antibodies, ASMA) and PBC (antimitochondrial antibodies, AMA) [19]. Table 1 reports the etiologies of the liver disease in CD patients with hepatic involvement. An abdominal ultrasound to investigate the hepatic morphology and rule out fatty liver or changes highly suggestive of chronic liver disease cirrhosis such as hepatic surface nodularity or a coarsened echotexture is recommended [49]. The clinical assessment should evaluate other factors and conditions promoting liver disease such as alcohol intake and metabolic diseases including Wilson's disease (ceruloplasmin) and alpha-1 antitrypsin deficit [77].

The initial monitoring of CeD patients with hepatic involvement can be done similarly to patients with other comorbidities. This approach includes the measurement of liver transaminases (ALT and AST), alkaline phosphatase and gamma-glutamyl transpeptidase. In the case of elevation of transaminases up to 3 to 5 times ULN, without evidence of chronic liver disease, this could be considered as a pattern secondary to celiac hepatitis [20]. Once a strict GFD is initiated, liver enzymes should be repeated after 6 to 12 months to assess response to therapy [78, 79]. However, there is a subset of patients in whom liver transaminases will remain elevated after 1 year of GFD. Therefore, a thorough assessment to exclude other hepatic complications such as MASLD, alcoholic liver disease, chronic viral hepatitis, autoimmune liver disease, and drug or toxin-related injury (see Table 1). The management of associated liver disease in patients with celiac hepatitis is like that of patients without CeD and can be supported by the use of non-invasive tests for liver fibrosis, such as transient elastography and serum biomarkers. These methods reduce the need for liver biopsies, providing safer and faster evaluations. Transient elastography has been used in patients with CeD, showing alterations in a subset of patients with increased transaminases and an improvement in patients on a GFD [80]. The presence of portal hypertension, with a normal transient elastography (< 10 Kpa) is suggestive of PVSD; usually, PSVD is also associated with high spleen stiffness [81]. Serum biomarkers can be useful in the early detection of liver fibrosis, which is crucial for managing liver complications, although untested in CeD. In spite of the absence of clear evidence, monitoring with non-invasive tests could improve patient outcomes and treatment decisions in selected cases [82].

The approach to liver enzymes alterations in RCeD is similar to those without CeD. At the time of RCeD diagnosis, patients have already performed most of the first-line tests excluding the most frequent causes of liver disease [15, 83]. In the case of diagnosis of AIH in an RCeD patient, it should be underlined that a multidisciplinary approach and shared strategy are crucial as both conditions are treated with immunosuppressive drugs including corticosteroids, azathioprine and methotrexate [84, 85].

In RCeD patients presenting with severe malnutrition, in particular RCeD type II, the alteration of transaminases and/or cholestasis should raise suspicion for an underlying neoplastic condition with liver involvement [15, 52, 53, 83]. A liver biopsy is required to confirm the diagnosis with an evaluation of liver lymphocyte infiltrate phenotype using immunohistochemistry (or when available, in off-label settings, employing flow-cytometry) [86]. The work-up should also include blood tests with ferritin, triglycerides, lactate dehydrogenase, liver function tests and complete blood count to exclude haemophagocytic lymphohistiocytosis.

3.1 Non-Immune-Mediated Liver Disorders Associated With Crohn's Disease

MASLD is a growing cause of chronic liver disease and is considered the most common liver disease in patients with CD. According to a recent systematic review [90], the pooled prevalence of MASLD in CD patients is 34.4%, with a risk two times higher in IBD patients versus healthy controls. The prevalence of advanced liver fibrosis was reported at 14% [90]. CD patients with MASLD have fewer metabolic risk factors than patients with MASLD alone, suggesting that CD itself is a risk factor for developing MASLD [91]. Indeed, Mancina et al. [92] showed that PNPLA3 148 M allele carriers with IBD had a higher risk for hepatic steatosis. A cross-sectional study conducted by McHenry et al. [93] with a cohort of 311 CD patients compared with 1244 controls have found CD as an independent risk factor of MASLD. Specific risk factors such as disease duration and previous surgery were reported, whereas MASLD development seems not to be affected by the treatment [94].

Prevalence of cholelithiasis in CD ranges from 11% to 34% [88]. A population-based study showed a significantly higher prevalence of cholelithiasis in CD patients than in control subjects [95]. Ileitis and ileal resection in CD seem to be risk factors for cholelithiasis [88], because of bile salt malabsorption leading to alteration in bile composition.

Furthermore, patients with CD are at increased risk of venous thromboembolism and the portal vein is a common site of thrombosis [96]. Many risk factors have been described such as extent of disease, surgery, disease activity, use of steroids and smoking [6]. Thus, thromboprophylaxis with anticoagulants is recommended in hospitalised patients with active CD and after surgery [97].

Liver abscesses are a rare complication of CD. They can result from direct extension of an intra-abdominal abscess or be secondary to bacterial translocation [98]. Persistent trans-mural inflammation, fistulising phenotype, long-term steroid use and malnutrition are the main risk factors reported. Recurrent fever is the most common symptom associated with abdominal pain in about half of patients. Blood tests are usually not specific, and diagnosis is made with ultrasound or computed tomography (CT) scan [99]. The treatment is based on antibiotic therapy and drainage in case of a large abscess [100].

More recently, in a large cohort patients with PSVD, 5.6% presented a concomitant unspecified IBD suggestive a possible rare association [22].

3.2 Immune-Mediated Liver Disorders Associated With Crohn's Disease

Different immunomediated liver diseases are associated with CD. Primary sclerosing cholangitis (PSC) is characterised by chronic inflammation of intra- and/or extra-hepatic bile ducts leading to biliary strictures and may result in end-stage liver disease requiring a liver transplant [101]. PSC patients have a significantly higher risk of developing cholangiocarcinoma and gallbladder cancer [99, 101]. The prevalence of PSC in patients with CD ranges from 1.2% to 3.4% [88]. A milder course of IBD and involvement of the right colon and terminal ileum is often present [102]. The histological findings compatible with PSC are peri-ductal fibrosis, fibro-obliterative cholangitis, ductular reaction, peri-ductal inflammation and ductopenia [103]. Small duct PSC is a variant of PSC characterised by abnormal liver biochemistry with a cholestatic pattern and histological findings of PSC but normal cholangiography [101]. The long-term prognosis of small duct PSC is better with a reduced risk of progression to end-stage liver disease and of developing cholangiocarcinoma [101]. The prevalence of small duct PSC is higher in CD patients than in ‘classic’ PSC [104]. CD and PSC seem to have an independent course, and CD therapies do not influence the progress of PSC.

Only a few case series have described the association between CD and other immune-mediated liver diseases such as AIH and PBC, with some patients affected by overlap syndromes presenting with common features between two entities, AIH-PSC being the most common [104, 105].

3.3 Drug-Induced Liver Injury in Crohn's Disease

Different immunosuppressive drugs are used for CD and are associated with hepatotoxicity. Methotrexate is known to cause liver injury ranging from asymptomatic hypertransaminasemia to fibrosis and cirrhosis, especially in prolonged use and/or high doses [106]. Hepatotoxic effects are thought to be linked to folate depletion and oxidative stress, which disrupts cellular antioxidant defences. Histological changes in the liver due to methotrexate are graded using the Roenigk classification system, which ranges from fatty infiltration to cirrhosis. Folic acid supplementation may reduce the risk of elevated transaminases, although its efficacy in preventing severe hepatotoxicity remains inconclusive [107].

Azathioprine-induced hepatotoxicity is less common but can manifest as hepatitis, cholestasis or mixed patterns of liver injury [108, 109]. The mechanism involves mitochondrial dysfunction and depletion of intracellular glutathione, leading to necrosis and impaired ATP production. Elevated levels of its metabolite, 6-methylmercaptopurine nucleotide (6-MMPN), are associated with hepatotoxicity. Discontinuation of azathioprine typically leads to recovery of liver function. Co-treatment with antioxidants like N-acetyl-L-cysteine has shown protective effects against mitochondrial damage caused by azathioprine [110]. A recent study on a large cohort of patients with PSVD described the presence of an azathioprine treatment in 3.4% of the subjects; most of them had a concomitant unspecified IBD [22].

5-Aminosalicylic acids (5-ASAs) can cause rare (but serious) liver injury. Hepatotoxicity may present as acute hepatitis or cholestatic injury. The mechanism is not fully understood but may involve idiosyncratic reactions or direct toxicity to hepatocytes. Monitoring liver enzymes during treatment is recommended to detect early signs of liver damage [111]. Monoclonal antibody targeting TNF-alpha has been associated with mild elevations in liver enzymes and rare cases of severe hepatotoxicity such as autoimmune hepatitis-like syndromes. The exact mechanism remains unclear but may involve immune-mediated responses. Discontinuation of infliximab and corticosteroid therapy often results in resolution of liver injury [112, 113].

3.4 Management of Liver Function Test Alterations in Crohn's Disease

CD patients should be routinely screened for liver function test abnormalities [100]. If liver test alterations are detected, drug-induced liver injury because of CD treatment (e.g., thiopurines, methotrexate or anti-TNF drugs) and viral hepatitis, especially HBV and HCV related, should be ruled out. These conditions are beyond the scope of our reviews. Most patients with PSC are initially asymptomatic; thus, diagnosis is based on biochemical alterations and radiological features. PSC should be excluded in every CD patient with persistent or fluctuating alteration of liver biochemical tests with a cholestatic pattern [114]. Magnetic resonance cholangiopancreatography (MRCP) has high sensitivity and specificity (86% and 94% respectively) in identifying PSC [115]; therefore, it should be performed in CD patients with suspected PSC [114]. A liver biopsy is usually not required to assess a diagnosis of PSC unless there is a suspicion of small duct PSC [103]. Patients with evidence of small or large-duct PSC should be considered for liver transplant evaluation, not only in the presence of portal hypertensive complications such as ascites or oesophageal varices but also in the presence of two or more episodes of bacterial cholangitis [114, 116]. No specific therapy for PSC is currently available; ursodeoxycholic acid (UDCA) at the dose of 15–20 mg/Kg/daily has a beneficial effect on cholestasis but fails to demonstrate a reduction of mortality or an increase of liver transplant-free survival [117]. Some retrospective studies have evaluated the efficacy of biologic therapies in patients with concomitant IBD. Infliximab, a monoclonal antibody directed against tumour necrosis factor-alpha (TNFα), was effective in the treatment of IBD associated with PSC but has not shown any improvement on liver biochemistry or imaging, considered as a prognostic surrogate marker of PSC [118]. Adalimumab, another monoclonal anti-TNFα agent, has shown a significant decrease of alkaline phosphatase compared to Infliximab and Vedolizumab after 6–8 months of treatment; however, this improvement was not confirmed after 12–14 months [118]. A retrospective study has suggested a potential specific role of Adalimumab in the improvement of cholestasis in PSC [119]. However, for both biologics, there is a lack of data on clinical outcomes and long-term prognosis in PSC. Vedolizumab is a humanised monoclonal antibody directed against the human lymphocyte α4β7 integrin specific to gut-homing lymphocytes. This peculiar mechanism of action was encouraging but failed to show improvement in liver biochemistry. In the largest cohort published by Lynch et al. [120], the mean alkaline phosphatase level was slightly increased after treatment with vedolizumab and 21% of the patients had liver-related events, including cholangitis, cirrhosis decompensation or liver transplantation [120]. Some concerns were raised by the study of Caron et al. that showed the occurrence of digestive neoplasia in 10% of patients during a follow-up of 19 months [121]. However, the retrospective nature of the studies and the lack of a control group make these findings difficult to interpret.

Diagnosis of AIH is based on biochemical tests and histology. The treatment is based on immunosuppression and the response to treatment is not affected by the presence of CD [122].

Anti-mitochondrial antibodies are usually detected for PBC diagnosis. Moreover, the evidence of elevated immunoglobulin M is also helpful in identifying this group of patients. The treatment is based on the use of ursodeoxycholic acid with a high rate of response [123].