1 INTRODUCTION

Liver abnormalities are common in persons with Cystic Fibrosis (PwCF) with up to 32% affected by 25 years of age¹ and is the fourth leading identifiable cause of death.² Multilobular cirrhosis, which affects 5%–10% of PwCF, is the most significant form of liver disease and presents at a median age of 10 years.³ This early and aggressive manifestation of CF demonstrates a need for improved understanding of disease identification, progression and treatment. As with many liver diseases, the damage may be silent until advanced disease has been established.

The gravity of CF liver disease (CFLD) is emphasized as it is an independent cause of mortality and is consistently a leading cause of death in PwCF.^{4, 5} A recent systematic review determined a threefold increased risk of death for CFLD over PwCF without liver disease and a retrospective study from the Netherlands concluded there is 10-year reduction in life expectancy with CF related cirrhosis.⁶ We do not yet understand which patients will progress to advanced disease and treatment options are limited. Ursodeoxycholic acid (UDCA) has commonly been used albeit with no evidence to support improved long-term outcomes.⁷ Ultimately, liver transplantation is the only treatment for advanced disease.

Advances in our understanding of the cystic fibrosis transmembrane conductance regulator (CFTR) protein have led to the development of highly effective modulators that improve CFTR function and many aspects of health for PwCF. With this, the predicted median survival of PwCF rose from 36.3 years in 2006 to 53.1 years in 2021.² Evidence of the impact of CFTR modulators on the liver is evolving. An increasing number of studies suggest a possible improvement in liver disease markers including liver stiffness, bile flow, and the incidence of cirrhosis.^8-10 However, other case reports show significant hepatotoxicity.^{11, 12} The approval of CFTR modulators for use in younger ages may help prevent disease progression, but mutations that are not eligible for modulator treatment remain. There is a pressing need to elucidate the pathogenesis to identify early markers of disease and possible targets for treatment intervention.

2 LIVER INVOLVEMENT IN CYSTIC FIBROSIS

Although CFLD often affects the entire liver, the CFTR protein is located in the apical membrane of cholangiocytes and is notably absent in hepatocytes.¹³ The effect of CF on the liver is apparent through a wide array of liver abnormalities including neonatal cholestasis, elevations of AST/ALT and GGT, steatosis, cholelithiasis, multilobular cirrhosis, and biliary cirrhosis. Intermittent elevations of alanine aminotransferase (ALT), aspartate aminotransferase (AST), and gamma glutamyl transferase (GGT) are reported in up to 85% of children diagnosed with CF.¹³ Radiographic changes such as steatosis of the liver are also common.^14-16 These changes may remain asymptomatic or evolve to clinically significant portal hypertension with esophageal varices, ascites, splenomegaly or encephalopathy.

Current recommendations include annual screening with physical exam (hepatosplenomegaly) and laboratory studies (AST, ALT, GGT). A longitudinal study determined that the finding of nodular liver disease by ultrasound identified individuals at highest risk for development of portal hypertension.¹⁷ Thus, some practitioners advocate for including ultrasound although whether clinically identified nodular changes will be as valid a predictor remains unclear.

A significant percentage of ALT elevations or sonographic evidence of hepatic steatosis will not result in prolonged or severe liver disease. Transient elevations of ALT may occur due to an infection or medication use and do not necessarily represent a substantial liver injury, nor are they a harbinger of more severe disease. The transient nature of these results raises the question of how to define what constitutes CFLD. Should everyone with ALT over normal for 6 months be considered to have CFLD or is true CFLD defined by portal hypertension with or without multilobular cirrhosis? Comparison between research papers is challenging due to the wide variation in presentation and lack of clear indicators of progression. Currently there is no single unifying definition accepted. Several definitions have been used with Table 1 showing that proposed by the North American Cystic Fibrosis Foundation.^{3, 18}

These recently published Guidelines by the Cystic Fibrosis Foundation provide clear recommendations on optimal strategies for Screening, Diagnosis, Monitoring and Management of persons with CF with possible liver involvment,¹⁹ covered in more detail in an accompanying chapter in this series (Early Detection, Elastography, and Biomarkers).

3 PATHOGENESIS CONSIDERATIONS

Given the multiple clinical manifestations of CFLD including steatosis, focal biliary fibrosis, multilobular cirrhosis, nodular regenerative hyperplasia (NRH), and biliary strictures, it is challenging to identify a single pathophysiologic process that could account for all variations. However, the lack of CFTR function on the cell membrane of the cholangiocytes,^{13, 16} the impact of dysregulated nutrition on the liver parenchyma leading to steatosis, and the abnormal gut function that leads to a pathogenic liver-gut interaction, can all contribute to the pathogenesis of liver manifestations in pwCF (Figure 1).

The presence of CFTR on the apical membrane of the cholangiocyte is critical for the maintenance of bile fluid control through secretion of chloride into the bile duct lumen when stimulated by secretin. This process also leads to the secretion of bicarbonate into the biliary tree.²² The loss of this protective alkaline “bicarbonate umbrella” within the biliary tree can lead to cholangiocyte epithelial damage by hydrophobic bile acids. The subsequent reactive inflammatory response can become chronic, resulting in the recruitment of immune and fibrotic cells and subsequent cholestasis and biliary fibrosis, a process shared by multiple cholangiopathies.²³ This pathogenic sequence is one reasonable explanation for the inspissated bile, focal biliary fibrosis, gallbladder dysfunction and biliary strictures found in a subset of patients with CFHBI.²⁴ In addition to reduction in bile acid alkalinization, dysfunction of CFTR in the biliary tree interestingly results in increased sensitivity to the biliary tree to bacterial endotoxin,²⁵ which can in turn significantly potentiates the inflammatory reaction of the cholangiocyte.²⁶

The wide variability of hepatobiliary manifestations of CF among patients with the same class mutations, and even with the same genetic mutation profile, remains unexplained. However, environmental factors likely contribute heavily to the risk of disease development and perhaps serve as modifiers leading to particular phenotypes. The bile of patients with CFHBI and biliary strictures has not been extensively studied, however, recent reports from patients with primary sclerosing cholangitis (PSC), demonstrating that bile colonization with bacteria results in significant deterioration of the biliary tree,²⁷ raise the question of whether bacterial products within the bile pool could serve as a such a disease modifier in CF. Additional candidate environmental triggers include obesity, malnutrition, drug toxicity, infections, and toxins such as alcohol.

One major limitation of current animal models of hepatobiliary disease in CF is the absence of a CFTR transgenic mouse which automatically develops liver abnormalities, although notable progress has been made in recent years using larger animal models.^{28, 29} Mouse models can, nonetheless, be very useful to study particular aspects of the interaction between genes and environment in the context of CF. One illustrative example is a study where the induction of biliary damage in the absence of CFTR required induction of experimental colitis.²⁵ Furthermore, the role of diet in hepatic manifestations of CF has been highlighted using such mouse models, where dietary alterations led to multiorgan manifestations in Cftr (−/−) mice, including biliary disease and steatosis.³⁰ A separate recent study found that a high medium-chain triglyceride diet resulted in cholangiopathy in Cftr (−/−) mice.³¹ These insights may lead to potential new avenues for research. The important role of diet in the gut-lung axis in CF was recently reported,³² and the many negative consequences to the liver following perturbations in the gut-liver axis through the portal vein is already well known.³³ Last, mouse models have proven to be useful to study the interaction between bile acids and the intestine in the pathophysiology of CF through the enterohepatic circulation.³⁴

4 PATHOGENESIS OF NON-CIRRHOTIC PORTAL HYPERTENSION AND OBLITERATIVE PORTAL VENOPATHY

Given the location of the CFTR protein in cholangiocytes, biliary sclerosis with subsequent cirrhosis has traditionally been recognized as a logical mechanism of injury for advanced liver disease. Hepatic fibrosis is progressive with stages one, two and three being incomplete. Cirrhosis (stage four or F4) is defined by complete bands of fibrosis surrounding liver lobules. However, there is increasing evidence that NCPH plays a role in this process. NCPH, recently renamed Porto-Sinusoidal Vascular Disease, presents with portal hypertension and may have mild degrees of fibrosis without actual cirrhosis.

NRH is one etiology of NCPH that can be difficult to distinguish serologically and radiographically from cirrhosis. Both may show thrombocytopenia, hepatosplenomegaly, liver nodularity, and have increased elastography scores. The preservation of liver synthetic function (normal albumin, INR and glucose) may suggest a diagnosis of NRH rather than cirrhosis. The diagnosis of NCPH is made histologically with liver biopsy or tissue from an explant at the time of liver transplant. Differentiation of NCPH from cirrhosis is important since shunt surgeries may alleviate the former without requiring transplantation. Furthermore, in these patients contrast-enhanced MRI imaging should be performed to distinguish between hepatic nodularity due to NRH vs enhancing lesions that could be concerning for malignancy.

In 2011, Peter Witters and colleagues published a Letter to the Editor describing histologic comparison of 12 patients of whom only five had histologic findings of cirrhosis.³⁵ Two of the patients had mildly elevated hepatic venous pressure gradients. These gradients were not elevated to a degree that would usually be considered clinically significant portal hypertension, yet they did have esophageal varices. This finding is consistent with pre-sinusoidal NCPH. On histologic review Witters and colleagues describe a portal venopathy with complete absence of portal veins within the portal triad. The authors theorized that the venopathy could be from inflammatory spillover from bile ducts, microthrombosis or primary endothelialitis.

Subsequently, in 2017, two case series of adult patients were published describing additional evidence of NCPH as the etiology of CFLD. In the first, Witters et al described eight patients with portal hypertension and preserved liver function.³⁶ Over 60% (5/8) of the patients had explant specimens without cirrhosis. Specimens from all patients revealed portal vein branch obliteration and dense portal vein fibrosis. Additionally, specimens from 3 patients had calcifications of the portal vein branches. Shortly afterwards, Hillaire et al described 10 patients with portal hypertension, nine of whom had combined lung-liver transplant and one who had isolated liver transplant. Biliary cirrhosis was identified in only 20% (2/10) of these patients. The other 80% (8/10) had an idiopathic NCPH with nodular hyperplasia and obliterative venopathy present.³⁷

A 2019 report from Texas Children's Hospital examined 17 explanted livers from pediatric patients with CFLD. The median patient age was 15 years with the youngest being 8 years. Only 35.3% of these specimens had focal biliary cirrhosis and none had signs of biliary obstruction. Conversely, NRH without cirrhosis was present in 94% of the specimens. A vascular etiology of disease was suggested by diminished portal vein diameters and smooth muscle surrounding intermediate sized portal veins.³⁸

These reports suggest that a biliary etiology is not the sole contributor to the development of CFLD. The question remains whether the portal vein is affected by its close proximity to the damaged and inflamed biliary system or from the inflow of inflammatory products and mediators from the intestines. A recent study from Ireland by Oneill et al has attempted to answer this question by comparing explanted livers from adults with CFLD and two other classic cholestatic diseases (primary biliary cirrhosis [PBC] and PSC).³⁹ The study found differences in rates of cirrhosis (0/9 patients with CFLD vs. 11/14 in PBC/PSC group) and NRH (8/9 in CFLD vs. 0/14 in PBC/PSC group). These findings suggest a venous etiology of damage in CFLD that is distinct from other cholestatic liver disease.

Several aspects of vascular disease and coagulation are affected in CF. Vasculitis is a rare, but severe complication in CF that may result from bacterial colonization, the presence of immune complexes or hypergammaglobulinemia.⁴⁰ Vascular damage may be exacerbated by platelet activation from systemic inflammation⁴¹ and hypercoagulability which is increased in PwCF.⁴² Finally, intestinal inflammation can weaken tight junctions between enterocytes and allow infiltration of gut bacteria into the portal system. Escherichia Coli can generate septic emboli that travel via the portal vein and result in obstruction of portal venules.

These proposed mechanisms of injury open possible areas of investigation for prevention of disease progression and treatment. The pathogenesis is likely multifactorial and there is a clear need for further research into these possibilities.

5 NEW APPROACHES TO STUDY CF LIVER DISEASE

The heterogeneity of clinical manifestations of CF liver involvement, the uncertainty as to the interplay between the CFTR mutations and the environment, and the limitations of existing animal models highlight the urgent need for innovative technological approaches (Figure 2).

One such approach is the incorporation of three-dimensional (3D) organoids to the study of CF liver disease. Organoids, defined as a collection of self-organizing cells originating from primary tissue or pluripotent stem cells, have increasingly served as a powerful tool for understanding the pathophysiology of a wide variety of disorders.⁴³ Advantages of organoids over conventional methods include the possibility of biobanking from individual patients and accessing precious primary cell types that are otherwise not readily accessible. Furthermore, induced pluripotent stem cells (iPSC) can be obtained from skin fibroblasts or peripheral blood cells and subsequently derived into a cell type of interest including hepatocytes and cholangiocytes. These iPSC derived cells can subsequently be used in two-dimensional (2D) monolayer or 3D cell culture investigations. The 3D structure of the organoid allows for a better assessment of cell-cell interactions and junctions within the organoid sphere, and assessment of the basolateral and apical membranes (using apical-out techniques)⁴⁴ compared to standard 2D cell techniques. Multicellular organoids can incorporate key cell types into a single sphere which allows for analysis of the relevant cellular interplay.⁴⁵ Last, organoid technology has been utilized to enable recreation of complex multiorgan systems to study organogenesis.⁴⁶

Organoids have been increasingly used for studies of hepatocellular and cholestatic liver disorders^{47, 48} and the cellular material can be derived from iPSC, from biopsy tissue, or from cells present in the bile.⁴⁹ Human cholangiocyte organoids express CFTR and the cholangiocytes perform typical relevant channel functions such as CFTR-dependent fluid secretion, which can be assessed through an assay where addition of forskolin increases CFTR function leading to fluid transport and swelling of the organoid, termed Forskolin-Induced Swelling (FIS).⁵⁰ This swelling, in turn, can be measured and the degree of swelling used as an assay readout. In addition, organoids can used to study the impact of CFTR modulators on chloride and bicarbonate transport.⁵¹

Building on these principles, an early study of biliary organoids derived from skin fibroblast iPSC of a patient with F508del homozygous mutation confirmed the absence of proper CFTR function and the negative impact of this on fluid secretion.⁵² Culture of these organoids with CFTR corrector VX-809 restored CFTR functionality and significantly increased the size of the organoid compared to untreated organoids. This was also demonstrated in a study of intrahepatic cholangiocyte organoids from a patient compound heterozygote for F508del/R1162X, where treatment with X-809/VX-770 significantly improved FIS.⁵³

Therefore, cholangiocyte organoids have a significant potential as a tool to understand the consequences of CFTR dysfunction in the biliary tree. Furthermore, although a 3D organoid does not fully replicate the multicellular microenvironment of the liver, advances in platforms such as bile duct-on-a-chip technology can make use of stored organoids to replicate organ-level functions by adding other relevant cells around the cholangiocyte in the chamber such as neutrophils, lymphocytes, macrophages, and hepatic stellate cells.⁵⁴

In addition to organoids, the ability to derive iPSC from patients with CF represents a broader technological advance by which CFLD can be studied a variety of settings. A study by Fiorotto et al.⁵⁵ demonstrated the utility of iPSC to explore novel approaches to understand the pathophysiology and potential treatment of CFLD. In this study, cholangiocytes were derived from iPSC from a healthy control and from an F508del homozygous PwCF. Using 2D culture techniques, the CF cholangiocytes recapitulated impairment of fluid secretion after protein kinase A/cAMP stimulation due to dysfunction of CFTR, as would be expected. However, interestingly these CF cholangiocytes derived from iPSC also showed higher phosphorylation of Src kinase and Toll-like receptor 4. This resulted in significantly more proinflammatory signaling by the CF cholangiocytes, including secretion of IL-8 and MCP-1 compared to the healthy control. Furthermore, the reactive and pathogenic response by the cholangiocytes could be abrogated with the administration of a Src inhibitor. This proof of principle shows that greater accessibility to cells derived from PwCF can allow for progress in our understanding of the complex mechanisms leading to hepatobiliary dysfunction in CF and enable novel therapeutic discoveries.

Another promising technological advance in the study of CF liver involvement is found in the field of artificial intelligence (AI). There has been great progress in the study and early implementation of AI in the diagnosis, management, and prognostication for patients with chronic liver diseases.⁵⁶ A major aspect of this advance revolves around histopathology and radiology assessment. Studies of patients with steatotic liver disease demonstrated a superior ability of machine learning analysis to diagnose and prognosticate liver outcomes.⁵⁷ In addition, in this study investigators found that a Deep Learning Treatment Assessment Liver Fibrosis score could detect an impact of antifibrotic therapy on liver histology that was not detected by liver pathologists using manual analysis. Although use of such AI techniques has not yet been reported for histologic assessment in CF liver disease, there is a potential to utilize these techniques to address unmet needs, such as unbiased evaluation of NRH, whose pathophysiology is poorly understood and centers on nonparenchymal abnormalities that are not well characterized through standard liver histopathology analysis.

Application of AI methods to radiologic imaging in CFHBI also holds promise and the analysis of digital images in clinical hepatology is advancing rapidly.⁵⁶ A prospective study of ultrasound in pediatric CF has provided rich data on the prognostic implication of abnormal findings such as heterogeneous parenchyma and nodularity.^{17, 58} However, a real limitation of these investigations is the significant interobserver variability among radiologists, particularly for more subtle findings in early CFHBI, which can be challenging even in the setting of a protocolized study let alone in routine clinical practice. One way to overcome this is through implementation of AI for analysis of routine ultrasound imaging in CF which could provide a way to improve standardization and decrease heterogeneity in the interpretation of results. Indeed, the application of AI for ultrasound images of the liver is under intense study at this time and was recently demonstrated to be excellent for detection of steatotic liver disease.⁵⁹

6 PERSONALIZED MEDICINE FOR CF LIVER DISEASE IN THE ERA OF CFTR MODULATORS

The treatment of many PwCF has been revolutionized over the past 5 years since the introduction of the current generation of modulators, specifically the Elexacaftor/Tezacaftor/Ivacaftor (ETI) regimen which has impacted clinical outcomes and quality of life.^{60, 61} Uncommonly, a small number of patients may develop drug-induced liver injury from this regimen, and liver enzymes should be monitored once the regimen is initiated.^{14, 15, 62} However, the majority of patients will not develop liver injury and it remains to be determined the degree to which liver involvement from CF can be reduced by long-term administration of ETI. In this respect, there is an urgent need for well-designed studies to be conducted using specific liver clinical endpoints to properly analyze the impact of ETI and related modulator therapies on hepatobiliary outcomes in CF.

Unfortunately, in addition to the above considerations, not all PwCF and liver disease will be eligible for ETI based on their mutation analysis. A recent study from Colombo et al.⁶³ reported that in a cohort including 171 patients with severe CFLD, 11% of such patients were ineligible for ETI based on their genetic profile. This was principally due to the presence of severe mutations resulting in truncated protein and complete loss of CFTR function. This study raises a valid concern that a significant minority of patients who develop advanced CFLD will not derive any benefit from ETI even if treated before the onset of CFLD, and as such will remain ineligible for the drug. It remains to be determined whether these genetic mutations, which are not treatable with ETI, themselves play a role in the pathophysiology of advanced CFLD. Therefore, there is an ongoing, urgent need both to study CFTR modulators in patients with CHFBI and identify additional avenues for therapy.

The above challenges also highlight the role for a personalized approach to care of PwCF to determine their risk of developing CHBHI, and their chances of responding to therapy if liver involvement is detected. Again, the potential for organoid technology to help predict disease on a personalized has been recently demonstrated. Muilwijk et al.⁶⁴ evaluated a biobank of intestinal organoids of 173 PwCF, determining in each individual organoid the degree of induced swelling by Forskolin, FIS, and the accompanying change in organoid size. They found that FIS strongly correlated with the individual risk of future lung decline (measured by FEV1). Strikingly, a higher FIS predicted a lower risk of CFLD (odds ratio: 0.18; 0.06–0.54; p = .002), as well as a lower risk of exocrine pancreatic insufficiency and CF-related diabetes. By contrast, there was no association between sweat chloride values from individual patients and these complications.

Another potential method to apply a personalized approach to the treatment of CFHBI, by building on an informed understanding of pathophysiology, is through regenerative medicine and gene editing. Geurts et al.⁶⁵ demonstrated the feasibility of correcting nonsense CFTR mutations in CF organoids using CRISPR technology. This could hold future promise for difficult-to-treat mutations. To this end, there is great interest in field of regenerative medicine to address chronic biliary disorders by using organoids as a means to repopulate the biliary tree through bioengineering methods, pioneered by Sampaziotis et al.^{66, 67}

As such, patients with advanced CFLD may have future therapeutic options to restore hepatobiliary function in addition to the use of modulator drug regimens. This concept will increasingly center on the personalized medicine paradigm of characterizing an individual's disease risk based on assays using patient biospecimens.^{68, 69} However, the ability to harness these methods to benefit patients with liver involvement will depend on continuous improvements in the understanding of key pathophysiologic mechanisms of CHBHI, to determine the optimal method and timing to intervene throughout natural history of the disease.

7 SUMMARY AND CONCLUSIONS

Multiple advances in the care of PwCF have significantly improved longevity and quality of life in recent decades, including the current era of highly effective modulators with ETI. However, our understanding of the pathophysiology of hepatic and biliary manifestations of CF remains incomplete and has lagged in an underdeveloped state relative to other fields within CF. In addition, there are a variety of clinical manifestations of CFHBI ranging from mild steatosis to multilobular cirrhosis to NRH. The terminology for these clinical types are only now becoming standardized, and their full characterization requires further research and consensus to promote clarity in clinical studies.

While there is, undoubtedly, a critical interaction between pathogenic CFTR mutations and a variety of environmental factors including microbiome dysbiosis, dietary factors, toxins, and an inflamed gut, how these interactions result in the various clinical manifestations of CFHBI remains to be determined. More research into the study of pathophysiology in CF in the hepatobiliary system is critical to overcome these challenges and should be prioritized by investigators and clinicians, advocacy organizations, funding partners, and patients alike. Furthermore, the limitations of animal models, the increased prioritization of scientific studies based on human specimens, and the scarcity of primary cells from the hepatobiliary system highlights the need to apply innovative technologies to study CFHBI, including iPSC and organoids in addition to bringing AI decisively into the investigative arsenal for CF liver research.

Finally, the paradigm of individual, personalized medicine is tailor-made for CF. Indeed, the CF community has truly stood out within medicine as a whole and is leading the way in developing this approach. Critically necessary among these considerations is a commitment to focus such energies on CFHBI itself, thereby creating new opportunities to predict complications and deliver novel therapies to individual patients at risk of liver failure and death. These efforts will ultimately be successful insofar as the pathophysiologic mechanisms and clinical phenotypes are addressed in a robust manner.

AUTHOR CONTRIBUTIONS

Vania L. Kasper: conceptualization; writing—original draft; writing—review & editing. David N. Assis: conceptualization; writing—original draft; writing—review & editing.

CONFLICT OF INTEREST STATEMENT

The authors declare no conflicts of interest.