Fontan-associated liver disease and hepatocellular carcinoma in adults

The Fontan operation creates a unique circulation, and is a palliative therapy for patients with single-ventricle congenital heart disease. Increased venous pressure and decreased cardiac output and hepatic venous drainage result in sinusoidal dilatation around the central veins. This causes congestion and hypoxia in the liver, leading to Fontan-associated liver disease (FALD). Non-invasive and invasive markers enable diagnosis and evaluation of the fibrosis status in chronic liver disease; however, these markers have not been validated in FALD. Additionally, regenerative nodules such as focal nodular hyperplasia (FNH) are frequently found. The severity of fibrosis correlates with the duration of the Fontan procedure and the central venous pressure. Cirrhosis is a risk factor for hepatocellular carcinoma (HCC), the annual risk of which is 1.5–5.0%. HCC is frequently difficult to diagnose and treat because of cardiac complications, coagulopathy, and congenital abnormalities. The mortality rate of FALD with liver cirrhosis and/or FALD-HCC was increased to ~ 29.4% (5/17 cases) in a nationwide survey. Although there is no consensus on the surveillance of patients with FALD, serial monitoring of the alpha fetoprotein level and imaging at 6-month intervals is required in patients with cirrhosis.


Description
Type I (systolic heart failure) Type II (diastolic heart failure) Type III (non-cardiac failure) Relevant to discussion of FALD  www.nature.com/scientificreports/ Fontan physiology. Central venous pressure (CVP) typically increases after Fontan surgery (Fig. 1b), as compared to normal (Fig. 1a). The hepatic venous pressure and pressure load on the central vein in the hepatic lobule cause congestion of the liver 15,16 . Mutual buffering between the portal blood flow and the hepatic artery blood flow is known as the hepatic arterial buffer response (HABR) 15 (Fig. 2). When the portal blood flow decreases, the hepatic artery is dilated to regulate the hepatic blood flow. Thus, the oxygen supply to the liver tissue is constantly maintained. Mechanical stimulation caused by dilation of the hepatic sinusoids, thrombus formation as a result of abnormal coagulation, and congestion and hepatocyte hypoxia are observed 17 . HABR increases the hepatic artery blood flow and hypernodular lesions may form in hypoxic areas, particularly in peripheral areas of the liver. Additionally, hypoxia and thrombosis within the sinusoids promote the activation of hepatic stellate cells (HSCs) and the production of fibronectin, leading to portal and sinusoidal fibrosis. Perisinusoidal edema and ischemic liver cause fibrosis progression without centrilobular inflammation. When fibrosis further progresses, cross-linking fibrosis mainly connecting the central zones is observed histologically, and a fibrous septum is formed, a finding of cirrhosis. Congestion causes the formation of an inverted image of the hepatic lobules ('reverse lobulation'; an image in which the portal area is located in the center and the hepatocyte population is surrounded by a congestion zone). Postsinusoidal hepatic outflow obstruction lead to accumulation of ascites presenting a high protein level. The ascites showed a protein level of > 2.5 g/dL and the serum ascites an albumin gradient of > 1.1 g/dL 18 . Ascites may be caused by increased sinusoidal pressure and/ or impaired lymphatic drainage 4 .
Prevalence and diagnosis of FALD. Baek 19 . In the blood test, transaminase levels were typically within the normal range or mildly elevated in FALD ( Table 2). The γ-glutamyltransferase (GGT) level was mildly elevated (median = 69 U/L) in 75% of patients 20 20,22 . Procollagen-III-peptide (P-III-P) is more susceptible to inflammation and it is said diagnosing the stage of fibrosis is inferior than hyaluronic acid or type IV collagen 22 . Shimizu et al. suggested hyaluronic acid and GGT as markers of the progression of liver fibrosis in Fontan patients 23 . While mac-2 binding protein glycan isomer (M2BPGi) is a useful marker of chronic hepatitis, Table 2. Non-invasive and invasive assessment of FALD. ALT alanine aminotransferase, ARFI acoustic radiation force impulse, AST aspartate aminotransferase, CT computed tomography, FALD Fontan-associated liver disease, Gd-EOB-MRI gadolinium ethoxybenzyl diethylenetriamine pentaacetic acid -enhanced magnetic resonance imaging, GGT γ-glutamyl transferase, HPVG hepatic venous pressure gradient, M2BPGi mac-2 binding protein glycosylation isomer, MELD model for end-stage liver disease, MRE magnetic resonance elastography, NPV negative predictive value, P-III-P procollagen-III-peptide, PPV positive predictive value, SWE shear wave elastography, T-BIL total birirubin, VAST score varices, ascites, splenomegaly, thrombocytopenia.  24 . However, it is unlikely to rise of M2BPGi in the cases of Fontan, because FALD is not accompanied by inflammation.
In the imaging, the liver may appear normal on radiological examination or slightly hypoechoic on ultrasound at the early stage of congestive hepatopathy. As fibrosis develops, a coarse heterogeneous hyperechoic parenchymal pattern and surface nodularity become evident 18 . The liver is often enlarged with caudate lobe hypertrophy, similar to Budd-Chiari syndrome. Irregular parenchymal fatty infiltration of a perivascular distribution can be seen on ultrasound. Contrast-enhanced ultrasound (CEUS) shows heterogeneous and decreased enhancement of the liver in the portal venous phase 25 . Hepatic vein waveforms assessed by doppler ultrasound change in accordance with liver fibrosis progression 26 .
In computed tomography (CT), hepatic fibrosis may be seen as reticular pattern on delayed-phase CT (i.e., peripheral diffuse patchy enhancement) 13 . Zonal enhancement (i.e., altered enhancement of the liver periphery) is correlated with lower hepatic vein pressure and a lower likelihood of cardiac cirrhosis 13 . On magnetic resonance imaging (MRI), there are areas of increased T2-weighted and diffusion-weighted signals with reduced T1-weighted signal intensity in the periphery of the liver, corresponding to areas of abnormal contrast enhancement 27 . Gadolinium ethoxybenzyl diethylenetriamine pentaacetic acid-enhanced MRI (Gd-EOB-MRI) revealed a characteristic reticular or mosaic pattern of diminished enhancement (i.e., 'frog spawn' appearance) 28 . The apparent diffusion coefficient (ADC) calculated by diffusion-weighted imaging (DWI) enables estimation of the degree of hepatic fibrosis 25 .
Assessment of liver stiffness by transient elastography (TE), acoustic radiation force impulse (ARFI) elastography, and magnetic resonance elastography (MRE) enables evaluation of hepatic fibrosis [29][30][31] . However, it was not correlated with the histopathologic findings at a single time point 32 . ARFI and TE might be useful for monitoring liver stiffness in patients with Fontan physiology. The mean shear wave propagation velocity in liver tissue by ARFI elastography in Fontan patients was 1.86 ± 0.5 m/s in 21 patients 33 . Of that, 76% of patients had a value over the cirrhosis threshold of 1.55 m/s. In contrast, the mean shear wave propagation velocity was significantly lower in patients who had undergone heart transplant 31 . However, TE/shear-wave elastography (SWE) cannot distinguish hepatic congestion from fibrosis 34 . While MRE reportedly enables evaluation of liver fibrosis 30 , its utility in FALD needs to be evaluated.
The gross appearance of the liver is termed 'nutmeg liver' in patients with FALD 18 . The pathological findings showed liver sinus fibrosis in 76%, centrilobular necrosis in 33%, pericentral fibrosis in 79%, and portal vein fibrosis in 52% of cases 35 . FALD showed dilatation and fibrosis of hepatic sinusoids, fibrosis of the portal area, and no inflammation 36,37 . Pathological evaluation of the liver via percutaneous or transvenous biopsy is the gold standard for assessing the degree of fibrosis; however, obtaining liver samples is difficult and there is a risk of bleeding. Also, the role and timing of the initial and follow-up liver biopsies in this population are unclear.

Diagnosis of liver cirrhosis by invasive and non-invasive biomarkers in FALD. Cardiac cirrhosis
results from prolonged passive liver venous congestion secondary to right-sided congestive heart failure and is defined as stage-4 fibrosis on liver biopsy 38 . At this stage, obtaining liver samples is hampered by the risk of bleeding and ascites accumulation. As another invasive examinatioin, transjugular measurement of the hepatic venous pressure gradient (HVPG, the gradient between the hepatic and portal veins) of > 5 mmHg is suggestive of sinusoidal portal hypertension 13 . The HVPG is useful for differential diagnosis of ascites of cardiac origin (normal HVPG) or hepatic origin (elevated HPVG) 39 .
Instead of these invasive examinations, typical imaging findings of the liver, formation of esophageal and gastric varices, ascites accumulation, and splenomegaly can facilitate diagnosis of liver cirrhosis. Moreover, several scores are used to assess patients with end-stage liver disease 40 . The model for end stage liver disease (MELD) XI score, which is based on the serum bilirubin and serum creatinine levels, may be predictive of the outcomes of Fontan patients 41 . A recent retrospective review revealed a positive correlation between the MELD-XI score and hepatic fibrosis scores on pathology (correlation coefficient = 0.4; p = 0.003) 42 . Although, a receiver operator characteristic analysis did not identify a score cutoff with adequate sensitivity and specificity 42 , patients with a MELD-XI score of ≥ 19 had a higher mortality rate 43 .
Proprietary tests such as FibroSURE, which includes assessment of multiple serum markers, have been validated only in patients with HCV and rely on inflammatory markers that are unlikely to be relevant in FALD. FibroSure for identifying evolving or established cirrhosis when compared to liver biopsy had a positive predictive value (PPV) of 33.3% and a negative predictive value (NPV) of 52.6% 20 . A hyaluronic acid level of > 46 ng/mL is indicative of liver cirrhosis and the PPV and NPV were 33.3% and 38.5%, respectively 20 .
Besides, the aspartate aminotransferase (AST)/alanine aminotransferase (ALT) ratio, AST-to-platelet ratio index (APRI) score, Forns index, and fibrosis-4 (FIB-4) score are reported as markers of fibrosis in FALD 19,39,44,45 . While the FIB-4 index includes age and has lower predictive utility among young adults with other liver disease 46 . In contrast, the VAST score is used to evaluate portal hypertension according to the liver-related complications 47 . A VAST score of ≥ 2 was reported to be significantly related to major adverse events (odds ratio = 9.8, 95% confidence interval [CI] = 2.9-32.7) 47 .
Treatment for FALD. Medical therapies specific for FALD are not available. Nevertheless, preventive, medical, surgical, and transplant strategies beneficial for similar disease processes may be applicable in FALD 14,16,48 . Prior to initiating a treatment strategy, it is important to improve cardiac output and/or raise the Fontan pressure (Fig. 1b).
Regarding treatments for liver cirrhosis, ursodeoxycholic acid (UDCA), lactulose, kanamycin, and heart-lung transplantation were used in a nationwide study 49 . Generally, UDCA treatment decreased the elevation of liver enzymes; however, the long-term benefit for FALD was uncertain. A branched-chain amino acid (BCAA),

Scientific Reports
| (2020) 10:21742 | https://doi.org/10.1038/s41598-020-78840-y www.nature.com/scientificreports/ an antibacterial agent (rifaximin), carnitine, and/or synthetic disaccharide may be used for FALD to prevent encephalopathy by reducing the ammonia level. For ascites, since the renin-angiotensin system is upregulated in liver cirrhosis, spironolactone is used prior to frosemide as diuretics 50 . Additionally, tolvaptan, a highly selective vasopressin-V2 receptor antagonist, is used to treat cardiac failur and refractory ascites 50 . In more severe cases of ascites, cell-free and concentrated ascites reinfusion therapy (CART) or ascites drainage therapy can be considered. Infusion volume load and fever should be care. Control of ascites is frequently hampered by postsinusoidal outflow obstruction.
In patients with more severe disease, cardiac transplantation was selected for cases with no evidence of liver cirrhosis. Overall, the 5-and 10-year survival rates were 72% and 69%, respectively, after cardiac transplantation 51 .
Combined heart-liver transplantation should be considered in severe cases 52 . It reportedly has a favorable outcome, with a 10-year survival rate of 83% 52 .
Liver nodules are missed on ultrasound in 30% of cases 31 . Contrast-enhanced CT and Gd-EOB-MRI enable detection of FALD-related HCC, however, 14% of FNH cases show portal/delayed washout, which is also present in patients with HCC 57 . The sensitivity of positron emission tomography (PET)-CT scan is only 55% for diagnosis of HCC 58 . Wells et al. demonstrated that mosaic architecture and an elevated alpha fetoprotein (AFP) level are associated with HCC, especially an AFP level of ≥ 400 ng/mL 59 . Some HCC cases are difficult to diagnose because of the lack of an increased AFP level. Again, a large proportion of the patients were treated with warfarin potassium and obtaining a tumor biopsy sample for diagnosis was problematic. Also, it affected the level of des-gamma-carboxy prothrombin (DCP), a marker of HCC, in turn hampering the diagnosis of HCC. Therefore, a new marker for FALD-HCC is needed.
There are no reports of the misdiagnosed rate of patients with FALD-HCC. Almost all investigations have been case reports, and large studies are limited. In our 124 cases, we detected 77 (62.1%) cases with hyperechoic lesions on ultrasound. Twelve patients were diagnosed with HCC 56 . An increase in AFP was observed in seven cases. Five cases were finally diagnosed with HCC by imaging and the clinical course. We experienced several HCC cases that were difficult to distinguish from FNH. Case 1: A 37-year-old female had a complicated hypervascular tumor periphery on CT (Fig. 3a). The nodules were increasing in size, ultrasound could not detect the nodule, and MRI could not be performed because of a pacemaker. Although preoperatively diagnosed as HCC by CT, the pathological findings of the surgically removed tumor indicated FNH (Fig. 3b). A non-cancerous liver specimen showed sinusoidal dilatation and mild fibrosis. Case 2: A 30-year-old male was detected with a hypervascular tumor on CT of the late arterial phase (Fig. 3c). The tumor was positive by PET-CT (Fig. 3e) and surgically removed, because TACE and PBT were ineffective. It was diagnosed as confluent-multinodular-type Table 3. Characteristics of liver nodules in FALD 53 . CT computed tomography, MRI magnetic resonance imaging. www.nature.com/scientificreports/ poorly differentiated-HCC based on liver cirrhosis (Fig. 3d). Two cases (2.6%) including case 1 were diagnosed as FNH by surgery. Therefore, the false-positive misdiagnosed rate was 2.6% (2 of 77 cases with nodules). It is difficult to take samples from the peripheral type of nodule. In one case, we performed a tumor biopsy just before transcatheter arterial chemoembolization (TACE) therapy to prevent bleeding. It is a method to diagnose HCC; however, it might increase the bleeding risk in patients in the margin. Therefore, new diagnostic method or marker for FALD-HCC is needed.
Treatment and outcomes of HCC. The treatment of HCC is dependent on liver function and the number and size of tumors, establishing the various guidelines 60,61 . In addition to volumetry of HCC and residual liver, MELD-XI score and indocyanine green retention rate at 15 min (ICG-R15) 62 may assist evaluation of liver function before surgery. In patients with FALD, cardiac function should be considered when selecting a treatment for HCC. There is limited evidence to suggest the optimal treatment strategy for FALD-HCC. Liver resection increases the CVP if the IVC clumping is applied, and so was regarded as unsuitable for patients with FALD. Nemoto et al. performed surgery in the reverse Trendelenburg position without IVC clamping 63 . This procedure reduced the CVP from 12 to 10 mmHg without decreasing the systemic blood pressure resulted in reduced blood loss. Moreover, laparoscopic hepatectomy was reported as safe proceasear in the FALD setting 64,65 . The laparoscopic hepatectomy was safely performed keeping the pneumoperitoneum pressure at less than 6-10 mmHg and adequate fluid infusion was given to maintain cardiac preload 64 . Central venous pressure was monitored (11-21 mmHg) and end-tidal carbon dioxide tension was shifted to 36-40 mmHg 65 . The Pringle maneuver was applied during liver resection.
In contrast, the utility of radiofrequency ablation (RFA) is limited in the patients with a pacemaker, accumulation of ascites, and coagulopathy or anticoagulant therapy.
In previous studies, several cases were treated with transcatheter arterial chemo-embolization (TACE) or hepatic arterial infusion chemotherapy (HAIC) 6,56,66 based on the greater hypervascularity of FALD-HCC ( Table 4). The efficacy of TACE is limited in some cases of abnormal vasculature. We reported proton beam therapy (PBT) as a treatment option for patients with FALD-HCC 56,67 . PBT is potentially more beneficial in sparing organs-at risk 68 . For liver tumors, the tolerance of surrounding normal liver, biliary tracts, and gastrointestinal structures is the main limiting factor for dose escalation. Therefore, PBT has a dosimetric advantage compared to X-ray therapy. We treated four patients with HCC; no serious adverse event was observed 68 .  www.nature.com/scientificreports/ In addition to intrahepatic metastasis, most extrahepatic metastases are to the lungs. We experienced three cases of lung metastases among 12 FALD-HCC (Table 4). We speculated that the increase in pulmonary vascular resistance may reduce the blood flow speed, facilitating adhesion of cancer cells. This might promote the metastasis of HCC to the lungs.

Risk factors for FALD and FALD-HCC.
The severity of fibrosis correlated with the duration of the Fontan procedure and the CVP 13,69 (Fig. 4). In one case series, 43% of patients had advanced fibrosis 30 years after Fontan operation 51 . Additionally, aging, underlying hepatitis B or C infection, alcohol intake, and hepatotoxic drug use were associated with FALD development (Fig. 4). Timing of diagnosis, type of Fontan, cardiac complications, comorbid systemic disease and obesity may influence the clinical picture in ways that are poorly understood 14 .
Notably, cirrhosis is a strong risk factor for FALD-HCC 70 . The annual risk of HCC in cirrhotic patients with FALD was estimated to be 1.5-5.0% 6 . CVP has been reported to be 16.4 ± 6.1 mmHg in patients with liver cirrhosis after the Fontan procedure and 11.3 ± 2 mmHg in non-cirrhotic cases 13 . Although cirrhosis is a risk factor of HCC, it does not predict the prognosis. Ohuchi et al. reported that a high CVP and low arterial oxygen saturation strongly predict clinical events in children (p < 0.001), whereas these prognostic factors were marginal in adults 71 . Instead of CVP, renal dysfunction and metabolic abnormalities predicted clinical events in adults (p < 0.05). Therefore, medication and fenestration that lowers right atrial pressure are effective for decreasing CVP and might inhibit the progression of FALD; however, it may be insufficient to prevent FALD-HCC. The liver stiffness values on ARFI elastography were significantly higher in patients with hepatic nodules 31

Surveillance of FALD and HCC.
A recent long-term follow-up study reported 10-, 20-, and 30-year survival rates of 74%, 61%, and 43%, respectively, among 1,052 patients after the Fontan procedure 72 . However, a recent systemic review of 65 FALD-HCC cases, which reported that 1-year survival is 50% 73 . Only four patients (6.2%) were under liver imaging surveillance for FALD-HCC, suggesting that HCC surveillance is necessary.   39 . In the presence of cirrhosis, serial monitoring by AFP and imaging every 6 months should be recommended similar to patients with HCV 74 . We think we could also follow the patients by this algorism to surveillance for FALD-HCC (Fig. 5). We experienced 12 cases of FALD-HCC, for an incidence of 0.8%, 2.9%, and 13.3% after 10, 20, and 30 years, respectively; these values are lower than those for HCV-related  www.nature.com/scientificreports/ HCC 75 . The mortality rate of liver-related death among FALD is 0.19% (5/2,700 cases) and increased by ~ 29.4% (5/17 cases) in only those with liver cirrhosis and/or FALD-HCC (Overall, 25 year-survival rates after Fontan procedure were 68.6% and 97.9% in FALD-HCC and non-FALD-HCC, respectively, p < 0.01.) 49 . HCC must be diagnosed as at an early stage as possible to facilitate timely treatment. The incidence increases approximately 20 years after the operation. Therefore, we recommend that HCC surveillance should begin 10 years after the Fontan procedure 53,56,76 .

Mental health care and transitional care of FALD in adults.
People with a Fontan circulation have a higher rate of lifetime psychiatric diagnosis (65%) than their healthy peers (22%), particularly for anxiety and behavioral disorders 77 . Therefore, there needs to be a low threshold for the provision of mental health care. Also, from 21 to 76% of patients experience a break in regular follow-up cardiology care after transfer from pediatric to adult care 78 . Transitioning from child to adult care, including clinical and social care, is necessary for patients with FALD. FALD must be followed-up continuously for the lifetime of the patient.

Conclusion and future perspective
The prevalence of FALD is increasing worldwide and the frequency of liver complications is rising because of improvement of cardiac survival. Evaluation of Fontan is different in children and adults, so further studies to identify non-invasive markers of fibrosis and FALD-HCC criteria are needed.
Ethical approval. All procedures performed in studies involving human participants were in accordance with the ethical standards of the institutional and/or national research committee and with the 1964 Helsinki declaration and its later amendments or comparable ethical standards.
Informed consent. Informed consent was obtained from the patient for the publication of our study.