Liver fibrosis detected by diffusion-weighted magnetic resonance imaging and its functional correlates in Fontan patients

Abstract

OBJECTIVES

The evaluation of Fontan-associated liver disease is often challenging. Diffusion-weighted magnetic resonance imaging can detect hepatic fibrosis from capillary perfusion and diffusion abnormalities from extracellular matrix accumulation. This study investigated its role in the evaluation of liver disease in Fontan patients and explored possible diagnostic methods for early detection of advanced liver fibrosis.

METHODS

Stable adult Fontan patients who could safely be examined with magnetic resonance imaging were enrolled, and blood biomarkers, transient elastography were also examined.

RESULTS

Forty-six patients received diffusion-weighted imaging; and 58.7% were diagnosed with advanced liver fibrosis (severe liver fibrosis, 37.0%, and cirrhosis 21.7%). Two parameters of hepatic dysfunction, platelet counts (Spearman’s ρ: –0.456, P = 0.001) and cholesterol levels (Spearman’s ρ: –0.383, P = 0.009), decreased with increasing severity of fibrosis. Using transient elastography, a cut-off value of 14.2 kPa predicted the presence of advanced liver fibrosis, but with a low positive predictive value. When we included platelet count, cholesterol, post-Fontan years and transient elastography values as a composite, the capability of predicting advanced liver fibrosis was the most satisfactory (C statistic 0.817 ± 0.071, P < 0.001). A cut-off value of 5.0 revealed a sensitivity of 78% and a specificity of 82%.

CONCLUSIONS

In Fontan patients, diffusion-weighted imaging was helpful in detecting liver fibrosis that was correlated with hepatic dysfunction. A simple score was proposed for long-term surveillance and early detection of advanced liver disease in adult Fontan patients. For adult Fontan patients with a calculated score > 5.0, we may consider timely diffusion-weight imaging and early management for liver complications.

Open in new tabDownload slide

INTRODUCTION

After the introduction of the Fontan operation in 1968 and the subsequent series of modifications, the life expectancy of patients with single-ventricle physiology has been much improved in recent decades [1]. Fontan circulation allows for passive caval blood flow to the pulmonary arteries in the absence of a subpulmonary pump. The unique circulation maintains nearly normal systemic oxygenation, but chronic haemodynamic changes including diminished cardiac output and elevated central venous pressure persist from the time of Fontan completion [2]. Therefore, Fontan circulation is associated with late hepatic complications such as liver fibrosis, cirrhosis or even hepatocellular carcinoma, which are increasingly recognized as Fontan-associated liver disease (FALD) [3]. This disease is usually clinically silent, and its early detection has been difficult. Consensus recommendations for the surveillance of hepatic fibrosis after Fontan operation are still lacking [3]. Liver biopsy is often considered as a reference standard for the diagnosis and staging of liver fibrosis. However, it has several limitations in Fontan patients, including the procedural risks, sampling error and interpretation bias [4]. It is also impossible to monitor the progression of the disease over time using repeat biopsies. Adoption of non-invasive diagnostic methods is currently recommended for routine disease surveillance for liver fibrosis [5]. Serum biomarkers have been developed for the diagnosis of liver fibrosis and cirrhosis across a range of chronic liver diseases [6], but the correlations between these biomarkers and the severity of FALD are still uncertain due to the unique pathophysiology. The measurement of liver stiffness, either by transient elastography (TE) [7] or magnetic resonance elastography [8], may be helpful to assess hepatic fibrosis. But, an elevation of central venous pressure and liver congestion in Fontan patients also increase the measured liver stiffness by elastography and compound its interpretation for assessing the severity of liver fibrosis [9].

A newly developed type of intravoxel incoherent motion (IVIM) diffusion-weighted magnetic resonance imaging (DW-MRI), which detects the capillary perfusion and diffusion abnormalities from extracellular matrix accumulation, has been applied to the assessment of liver fibrosis in Fontan patients without interference from elevated central venous pressure [10]. Because of the excellent correlation between IVIM and hepatic histopathology, IVIM can be a good diagnostic tool in detecting and staging liver fibrosis and serve as a noninvasive substitute for liver biopsy [11, 12]. We therefore hypothesized that the severity of hepatic fibrosis could be effectively assessed by DW-MRI and that the assessment, with less interference from the elevated central venous pressure, might be closely associated with the degree of hepatic functional derangement in Fontan patients. We investigated the hepatic DW-MRI and its functional correlates in stable Fontan patients. Our goal was to establish a noninvasive strategy based on a multivariable regression model to assist in serial monitoring for early detection of advanced liver fibrosis in Fontan patients.

METHODS

Study population

This was a cross-sectional study and was approved by the clinical research ethics board of the National Taiwan University Hospital. The procedures followed were in accordance with the Declaration of Helsinki and the ethical standards committee of the clinical research ethics board of the National Taiwan University Hospital. Adult patients (age ≥ 18 years) with an anatomical and/or functional univentricular heart who had undergone the Fontan procedure and were New York Heart Association functional class I or II without symptoms of Fontan failure (oedema, ascites, protein-losing enteropathy or plastic bronchitis) during regular follow-up between August 2017 and July 2019 were enrolled. Patients who could not safely be examined with DW-MRI, including patients with severe renal dysfunction, those with a pacemaker or implantable cardioverter defibrillator and those with known liver disease other than congestive liver disease, were excluded. After providing written informed consent, all patients underwent peripheral blood sampling for liver function testing and measurement of the levels of circulating biomarkers, TE and DW-MRI. The medical records including history of prior cardiac interventions, haemodynamic data from cardiac catheterization and clinical cardiac performance were reviewed.

Biochemical evaluation

Routine liver biochemical analyses including aspartate transaminase (AST) and alanine transaminase, bilirubin, gamma-glutamyl transferase (gamma-GT) and albumin were examined within 3 months prior to the DW-MRI study. Direct liver fibrosis markers [hyaluronic acid, procollagen type III N-terminal peptide (P3NP) and tissue inhibitor of metalloproteinases-1 (TIMP-1)], viral hepatitis markers for hepatitis B and C, platelet counts and cholesterol levels were tested after at least 6 h of fasting. Thrombocytopaenia was defined as a blood platelet count <150 × 10³/uL.

By using these data, known fibrosis indexes were also calculated, including (1) the AST-to-platelet ratio index (APRI) (2) Forn’s index (3) the Fibrosis-4 (FIB-4) score (4) the Enhanced Liver Fibrosis Score (ELFS) and (5) the Model for End-stage Liver Disease excluding the international normalized ratio (MELD-XI) score. (See Supplementary Materials for detailed description.)

Transient elastography

Transient elastography was performed using FibroScan (Echosens, Paris, France). The detailed method is described in the Supplementary Materials. The results of TE were also used to calculate the Fontan hepatic index: hepatic elastography measurement (kPa) + MELD-XI scores + (years post-Fontan surgery)^1/2 [13].

Intravoxel incoherent motion diffusion-weighted magnetic resonance imaging

Magnetic resonance imaging was performed with a 3.0-Tesla unit (Magnetom Verio, Siemens Medical Solutions, Erlangen, Germany). All patients fasted for at least 4 h before the examination. Echo-planar diffusion-weighted imaging was performed in all patients under free breathing conditions with 8 b values: 0, 50, 100, 200, 300, 400, 500 and 1000 s/mm [11]. Parallel imaging with an integrated parallel acquisition technique factor of 2 was used to reduce susceptibility artifacts. Eight slices were sampled with a 5-mm thickness. The scanning time was less than 7 min for acquiring diffusion-weighted imaging in all patients. All regression algorithms in this study were analysed using the open source software MitkDiffusion (https://www.mitk.org/wiki/MitkDiffusion), which yielded values for D, D* and f, following the functional descriptions provided by Lagarias et al. [14]: f is the diffusion fraction linked to microcirculation and is also called perfusion fraction; D is the diffusion parameter that represents the true molecular diffusion (slow component); and D* is the diffusion parameter that represents the IVIM (fast component). The influence of D* on signal decay can be neglected by using b values greater than 200 s/mm², which allowed the extraction of the parametric maps representing the f, D and D* parameters, which were fitted on a pixel-by-pixel basis. The liver signal intensity was recorded as the mean of all values generated by placing 4 separate square 1-cm regions of interest over the right hepatic lobe on the D, D* and f mapping images. The left lobe was not used because of cardiac motion artifacts that could potentially alter the diffusion measurements. Subsequently, the average value of the 4 region-of-interest values obtained was carefully measured by avoiding large vessels or bile ducts. The severity of liver fibrosis (no fibrosis, mild, moderate, or severe fibrosis and liver cirrhosis) was graded by the IVIM (fast component, D*) of DW-MRI according to our previous report [11] (Fig. 1).

Statistical analyses

The statistical software SPSS version 22 (IBM-SPSS, Inc., Armonk, NY, USA) was used for all analyses. Continuous variables are expressed as the median and 25th to 75th percentiles. The distribution of baseline characteristics and risk factors among patients with different degrees of liver fibrosis was examined using Fisher’s exact test (for categorical variables) or the Kruskal–Wallis test (for continuous variables). The correlations between the severity of the cases of liver fibrosis as assessed by MRI and liver elastography and the hepatic fibrosis markers were analysed using Spearman’s correlation coefficient. We developed a logistic regression model to identify potential factors to predict advanced liver fibrosis by MRI. Variables with a theoretical contribution in the univariable analyses were included in the multivariable logistic regression model. We generated a new clinical score by using the results of the multivariable logistic regression model. We used the Hosmer–Lemeshow goodness-of-fit test to evaluate the calibration of this logistic regression model to predict the advanced liver fibrosis. A model with a smaller χ² value and a higher P-value indicated a good calibration of the multivariable model.

In addition, the diagnostic accuracies of parameters for diagnosing advanced hepatic fibrosis were compared using receiver operating characteristic (ROC) curve analysis. We computed the C statistic, which is equivalent to the area under the ROC curve. A model with a C statistic > 0.75 is generally considered to have a meaningful discriminatory ability, and a C statistic < 0.5 denotes a poor discriminative capacity.

RESULTS

Patient characteristics

We enrolled 50 adult Fontan patients (age ≥ 18 years) who were in New York Heart Association functional class I or II without symptoms of Fontan failure. Four of them were excluded due to an implanted pacemaker and the inability to undergo MRI safely. The median age of these 46 patients was 22.6 years (25th–75th percentiles: 19.6–29.6 years); 47.8% were male (n = 22). The median duration from Fontan completion to the time of the study was 15.4 years (25th–75th percentiles: 8.3–19.2 years).

Liver fibrosis evaluation by diffusion-weighted magnetic resonance imaging

Among the 46 stable Fontan patients evaluated by DW-MRI of the liver, 93.5% had variable degrees of liver fibrosis, including 7 (15.2%) mild, 9 (19.6%) moderate and 27 (58.7%) advanced [17 (37.0%) severe and 10 (21.7%) cirrhosis] (Table 1). Only 3 (6.5%) did not have liver fibrosis. Four patients (8.7%) had hepatic nodular hyperplasia, but none of our patients had hepatocellular carcinoma (Table 1). The distributions of age, gender, years after the Fontan operation and the presence of heterotaxy or Fontan fenestration showed no differences among patients with variable degrees of liver fibrosis (Supplementary Material, Table S1).

Analysis of the correlation of liver function/fibrosis biomarkers, transient elastography and contemporary composite liver fibrosis indexes with various severities of fibrosis

Two major parameters of hepatic dysfunction, platelet counts (Spearman’s ρ: 0.456, P = 0.001) and cholesterol levels (Spearman’s ρ: –0.383, P = 0.009), decreased significantly as the severity of the liver fibrosis increased (Table 2, Fig. 1), but they both had weak correlations (Spearman’s ρ < 0.5). Thrombocytopaenia was not observed in patients without liver fibrosis or in those with mild liver fibrosis, but it was present in most patients with liver cirrhosis. The AST, alanine transaminase, bilirubin, albumin and gamma-GT levels showed no correlation with the degree of liver fibrosis (Table 2).

The direct serum markers of liver fibrosis, including hyaluronic acid, P3NP and TIMP-1, were not correlated with the severity of liver fibrosis as determined by MRI (Table 2).

The results of TE correlated with the severity of liver fibrosis (Spearman’s ρ: 0.335, P = 0.026). The median (25th–75th percentiles) TE value was 11.6 kPa in those without liver fibrosis and was 13.8 (11.6–17.4), 18.7 (10.0–21.5), 20.9 (12.8–25.0) and 19.0 (14.8–23.6) kPa in those with mild, moderate and severe fibrosis, and those with cirrhosis, respectively (Table 2, Fig. 2). The contemporary composite liver fibrosis indexes, including the APRI (Spearman’s ρ: 0.351, P = 0.017), Forn’s index (Spearman’s ρ: 0.467, P = 0.001), the Fib-4 score (Spearman’s ρ: 0.344, P = 0.019) and the Fontan hepatic index (Spearman’s ρ: 0.314, P = 0.038) also showed positive correlation with the severity of liver fibrosis as determined by MRI, but the MELD-XI score and ELFS did not (Table 2, Fig. 2).

Composite scores to predict severe liver fibrosis/cirrhosis by diffusion-weighted magnetic resonance imaging

When we developed univariable and multivariable logistic regression models for potential factors to predict severe liver fibrosis and liver cirrhosis by DW-MRI, we found that the years after Fontan operation, platelet count, cholesterol level and TE measurement helped to predict advanced FALD (severe fibrosis and cirrhosis) (Supplementary Material, Table S2). The calibration of this regression model evaluated with the Hosmer–Lemeshow goodness-of-fit test was satisfactory, with a χ² value of 10.34 and a P-value of 0.24. (Supplementary Material, Fig. S3) Therefore, by using the multivariate regression model, we generated a new liver fibrosis score for FALD named YACHT (Years after Fontan, plAtelet count, CHolesterol, Transient elastography) score, which is calculated as follows: 12.8 + 0.9 × (years after Fontan) −0.07 × platelet count (10³/uL) −0.2 × cholesterol (mg/dl) + 1.14 × FibroScan (kPa). The YACHT score showed the best correlation with the severity of liver fibrosis as assessed by DW-MRI (Spearman’s ρ: 0.552, P < 0.001) (Table 2, Fig. 2). We further used ROC curves to demonstrate the prediction capability of the YACHT score for advanced liver fibrosis, which revealed the largest area under the curve (C statistic: 0.817 ± 0.071, P < 0.001) (Fig. 3, Table 3). When we used the cut-off value of 5.0 for the YACHT score, the sensitivity was 78%, and the specificity was 82% for predicting severe liver fibrosis/cirrhosis. The positive predictive value (88%) and negative predictive value (70%) were the most satisfactory (Table 4). Using TE only, a cut-off value by TE such as 14.2 kPa was suggested to determine advanced liver fibrosis, but with a low positive predictive value (Table 4).

Among the other contemporary composite scores for liver fibrosis, although APRI, Forn’s index, the FIB-4 score and the Fontan hepatic index showed positive correlation with the severity of liver fibrosis as determined by MRI, only Forn’s index (C statistic: 0.725 ± 0.080, P = 0.013) showed good predictive ability for advanced liver fibrosis (Fig. 3, Table 3). Adopting a cut-off value of 3.2 for Forn’s index also resulted in a sensitivity of 78% and a good specificity of 82%, but the positive (75%) and negative (67%) predicting values were decreased (Table 4).

DISCUSSION

The Fontan population is increasing in size and medical complexity, including FALD that exhibit a unique pathophysiology. This prospective study demonstrates that liver fibrosis could be detected early by DW-MRI and that the changes correlated well with 2 parameters (platelet count and cholesterol level) of hepatic functional deterioration. Surrogate scoring to predict such advanced liver fibrosis is effective when using a composite score that includes years after Fontan operation, the platelet count, cholesterol level and TE measurement. Thereby, surveillance of FALD can be developed to incorporate simple parameters for timely DW-MRI and interventions.

Because of the safety and potential efficacy to detect early liver fibrosis independent of elevated central venous pressure, we applied DW-MRI to evaluate the severity of liver fibrosis in stable Fontan patients. The severity of liver fibrosis (no fibrosis, mild, moderate or severe fibrosis and liver cirrhosis) was graded by the IVIM, which might reflect the distinct hepatic changes in Fontan circulation. Using DW-MRI, only 6.5% of our patients were free from liver fibrosis. Severe liver fibrosis/cirrhosis was detected in 57.8% of these patients. In other studies in a population of a similar age using liver biopsy as a diagnostic standard, the proportion of patients with advanced liver fibrosis (stage 3–4 by gross architectural distortion score or stage 3 sinusoidal liver fibrosis) ranged between 41 and 68% [15, 16]. In the current study, the severity of liver fibrosis determined by DW-MRI correlated well with the changes in platelet counts and serum cholesterol levels, suggesting that DW-MRI can effectively detect FALD with global hepatic functional deterioration. Those Fontan patients are free of clinical symptoms, but the structural changes in liver fibrosis in these Fontan patients might already reduce the circulating number of platelets and jeopardize the lipid production of the liver. Meanwhile, the liver enzyme, bilirubin and gamma-GT levels in these patients did not increase, despite the evident liver fibrosis, that had been similarly described previously [17].

The causes of the decreased platelet count in chronic liver disease may be multifactorial [18]. Thrombocytopaenia (platelet count < 150 × 10³/uL) had been reported in 57.9% of Fontan patients with features of portal hypertension (VAST score $≥$ 2), possibly due to platelet sequestration and destruction in a congested spleen [19]. Furthermore, thrombopoietin, which is predominantly produced by the liver, is reduced in advanced liver fibrosis, resulting in decreased thrombopoiesis in the bone marrow [20]. A recent study had demonstrated that liver fibrosis scores that incorporated platelet counts, such as APRI or the FIB-4 score, correlated better with the severities of bridge fibrosis by liver biopsy in Fontan patients [21]. Thrombocytopaenia may further increase the bleeding risk in Fontan patients, especially in those with coagulopathy or those who are receiving anticoagulation therapy. Hypocholesterolaemia, which might be related to hepatic dysfunction, has been noted in Fontan patients [22]. Hypocholesterolaemia represents an independent predictor of survival in patients with liver cirrhosis [23]. In this study, we were probably the first to find a close negative correlation between the serum cholesterol level and the severity of liver fibrosis. This change, along with the decreased platelet count, may be used to detect hepatic functional deterioration in Fontan patients.

The interpretation of TE may be confounded by the increased liver stiffness due to elevated central venous pressure and liver congestion in Fontan patients. The optimal cut-off value of TE to detect severe liver fibrosis/liver cirrhosis depends upon the aetiology of the underlying liver disease and upon the prevalence of the condition under study in the target population. For example, the American Gastroenterological Association Institute Guideline recommended a cut-off value of 9.5 kPa to rule out advanced liver fibrosis, a cut-off value of 12.5 kPa for liver cirrhosis in patients with hepatitis C virus infection and a cut-off value of 11.0 kPa to detect liver cirrhosis in patients with hepatitis B [24]. The cut-off value suggested by TE to determine advanced liver fibrosis in our study was 14.2 kPa, slightly higher than the cut-off value recommended by the American Gastroenterological Association for patients with hepatitis C or hepatitis B. This recommendation may reflect the influence of elevated central venous pressure in Fontan patients.

The ELFS comprises 3 direct biomarkers of liver fibrosis, including hyaluronic acid, P3NP and TIMP-1, which had recently been described as a poor predictor of liver fibrosis in Fontan patients [25]. Though higher than the threshold, the ELFS was not different between patients with mild and severe liver fibrosis [15]. In this study, these 3 direct biomarkers all revealed poor correlations with the severity of liver fibrosis. Collectively, in this study, we developed a YACHT score, which consists of the years after Fontan, the platelet count, the cholesterol level and the TE measurement. The severity of FALD is strongly correlated to the time after Fontan [8, 26]. This surrogate revealed good performance in identifying severe liver fibrosis or cirrhosis detected by DW-MRI in stable adult Fontan patients. Compared with other contemporary liver fibrosis indexes, using a cut-off value of 5.0 for the YACHT score, revealed the highest positive and negative predictive values for predicting severe fibrosis or cirrhosis. Both the sensitivity and specificity were approximately 80% in stable adult Fontan patients.

Liver biopsy remains an important reference standard for the diagnosis of FALD. However, in addition to the non-neglectable procedural risks, even using the trans-jugular approach for liver biopsy [27], liver biopsy has potential limitations in FALD. Firstly, a sampling bias of liver biopsy can occur due to the patchy nature and heterogeneous distribution of liver fibrosis observed in FALD [28]. The result of a liver biopsy may not accurately reflect the overall status of the liver. Secondly, traditional pathological staging systems of liver fibrosis, such as the METAVIR (Meta-analysis of Histological Data in Viral Hepatitis) or Ishak scores, have been developed for chronic hepatitis and may ignore the common centrilobular and sinusoidal fibrosis observed in early FALD [3, 29]. Being combined with other scoring systems such as percentage collagen deposition [26] or congestive hepatic fibrosis score [29] may be a more precise approach to describe the pathological severities of FALD. Thirdly, several studies also indicated that the pretransplant liver biopsy cannot accurately predict post-heart transplant hepatic outcomes [4, 30]. As shown in the current study, the DW-MRI may not be regarded as the gold standard for the diagnosis of FALD, but it is effective for detecting the hepatic fibrosis that is associated with hepatic synthetic function deterioration (such as decreased platelet counts or cholesterol level) and shall be a potentially powerful way to reduce the need for a liver biopsy for the management of FALD in Fontan patients.

Introducing the YACHT score in clinical practice when caring for patients after the Fontan operation offers several benefits. Firstly, the YACHT score facilitates early detection of liver damage in adult patients with a Fontan circulation, a condition almost universally associated with liver damage. Secondly, it helps clinicians differentiate between mild liver damage and severe liver disease, enabling appropriate risk stratification for patient management. Thirdly, the score aids in monitoring disease progression over time, crucial for effective management and intervention strategies. The YACHT score comprises 4 parameters that are easily obtained in clinical settings, enhancing its feasibility and practicality for routine use. Additionally, by assessing the 4 parameters included in the YACHT score, clinicians can evaluate not only the risk factors associated with FALD after the Fontan operation but also the extent of congestion and parenchymal changes using TE and impaired hepatic synthesis function through decreased platelet counts and cholesterol levels. Overall, the introduction of the YACHT score empowers clinicians to more effectively diagnose, stratify risk and monitor patients with FALD, ultimately improving patient care and outcomes. Finally, the simplicity of the YACHT score calculation, which does not involve complex logarithmic functions like those in Forn's index or the MELD-XI score, enhances its ease of use in clinical practice. When we notice patients with a YACHT score above 5 in the clinical setting, we may consider detailed imaging surveys for advanced liver fibrosis or hepatocellular carcinoma, initiating investigations and management of specific complications (such as oesophageal varices and bleeding tendency) and detailed evaluation of the patient’s Fontan circulation for the possible interventions to slow down the progression of FALD.

Study limitations

Our study has several limitations. Firstly, the sample size was small, and the age at which the final Fontan operation was completed in this study cohort was relatively old. The first Fontan operation in Taiwan was performed in 1982, and the total cavopulmonary connection was performed since 1989. This operation was performed more than a decade later than those reported in the early Western literature. Many of the patients in this study received Fontan completion during the early era of this operation in Taiwan. The Fontan procedure was unavailable when these patients were young. In addition, some patients were referred to or visited our centre for further treatment when they were older; therefore, they could not have had the Fontan operation at a younger age. Longer exposure to cyanosis and ventricular volume overload before the Fontan completion may have occurred in our patient cohort. The liver disease may have developed prior to the final Fontan procedure rather than be related to the haemodynamics of the Fontan circulation. This observation may potentially limit the applicability of our findings to the broader Fontan population. Secondly, because this study only enrolled adult Fontan patients (age ≥ 18 years), the period after Fontan completion was longer than that in the paediatric patient population. The application of indirect biomarkers such as the YACHT score, Forn’s index and TE measurements in paediatric patients after the Fontan procedure remained undefined. Future studies for external validation of the performances of the YACHT score are mandatory. Finally, there is a paucity of histopathological data in our study due to the difficulties of performing a liver biopsy in stable Fontan patients. Nevertheless, the correlation of liver fibrosis identified by DW-MRI with liver histopathological analysis was validated by our previous study of other chronic liver diseases [11].

CONCLUSIONS

In this prospective study on stable adult Fontan patients, DW-MRI effectively detected the existence of liver fibrosis that was associated with decreased platelet counts and cholesterol levels, suggesting the appearance of hepatic functional deterioration. The YACHT score, which incorporates Years after the Fontan, the plAtelet count, the CHolesterol level and the Transient elastography measurement, can be applied as a surrogate for surveillance of FALD, timely DW-MRI and subsequent interventions. This observation is particularly important when DW-MRI cannot be performed safely (such as in patients with implanted pacemakers) or is not available due to limited resources.

SUPPLEMENTARY MATERIAL

Supplementary material is available at EJCTS online.

ACKNOWLEDGEMENTS

The authors would like to express our thanks to the staff of the National Taiwan University Hospital-Statistical Consulting Unit for statistical consultation and analyses.

FUNDING

This work was supported by the grants from the National Taiwan University Hospital (NTUH105-S3157) and the Ministry of Science and Technology (107–2314-B-002-172-MY2).

Conflict of interest: None of the authors have a conflict of interest to disclose. The authors take responsibility for all aspects of the reliability and freedom from bias of the data presented and their discussed interpretation.

DATA AVAILABILITY

The data underlying this article cannot be shared publicly to protect the privacy of the individuals who participated in the study. The data will be shared on reasonable request to the corresponding author.

ETHICAL STATEMENT

The present study was approved by the clinical research ethics board of the National Taiwan University Hospital. Reference number: 201504004RIND. The procedures followed were in accordance with the Declaration of Helsinki and the ethical standards of the committee of the clinical research ethics board of the National Taiwan University Hospital.

Author contributions

Chun-Wei Lu: Conceptualization; Data curation; Formal analysis; Funding acquisition; Investigation; Methodology; Writing—original draft. Chih-Horng Wu: Conceptualization; Data curation; Investigation. Miao-Tzu Huang: Data curation; Investigation; Methodology. Chee-Seng Lee: Data curation; Investigation. Huey-Ling Chen: Investigation; Methodology; Resources. Ming-Tai Lin: Data curation; Investigation. Shuenn-Nan Chiu: Investigations. Wei-Chieh Tseng: Investigation. Chun-An Chen: Investigation. Jou-Kou Wang: Investigation; Resources. Mei-Hwan Wu: Conceptualization; Investigation; Supervision; Writing—review and editing.

Reviewer information

The European Journal of Cardio-Thoracic Surgery thanks Victor O. Morell and the other anonymous reviewers for their contribution to the peer review process of this article.

REFERENCES

ABBREVIATIONS

ABBREVIATIONS
 
• APRI

  AST-to-platelet ratio index

 
• AST

  aspartate transaminase

 
• DW-MRI

  diffusion-weighted magnetic resonance imaging

 
• ELFS

  Enhanced Liver Fibrosis Score

 
• FALD

  Fontan-associated liver disease

 
• FIB-4 score

  Fibrosis-4 score

 
• γ-GT

  gamma-glutamyl transferase

 
• IVIM

  intravoxel incoherent motion

 
• MELD-XI

  Model for End-stage Liver Disease excluding the international normalized ratio

 
• MRI

  magnetic resonance imaging

 
• P3NP

  procollagen type III N-terminal peptide

 
• ROC

  receiver operating characteristic

 
• TE

  transient elastography

 
• YACHT

  Years after Fontan, plAtelet count, CHolesterol, Transient elastography