Introduction
Fontan-associated liver disease is a progressive disorder with cirrhosis and hepatocellular carcinoma in the severe stage, which significantly affects the prognosis of patients. Reference Dipchand, Honjo, Alonso-Gonzalez, McDonald and Roche1,Reference Lin and Hsu2 An ultimate treatment for the liver dysfunction is a heart transplantation, i.e., terminating the Fontan circulation. However, the presence of cirrhosis or hepatocellular carcinoma excludes patients as a recipient of a heart transplantation. Therefore, evaluating the progression of Fontan-associated liver disease is essential in managing these patients.
The Model of End-stage Liver Disease Excluding International Normalized Ratio score is the most sensitive haematological marker of liver fibrosis in Fontan-associated liver disease and predictor for post-heart transplant mortality in Fontan patients. Reference Amdani, Simpson and Thrush3 However, a recent study demonstrated that this score was not associated with adverse events when applied to unselected outpatients with Fontan circulation. Reference Aldweib, Wei and Lubert4 Although prognostic stratification based on multiple Fontan-associated liver disease-specific findings has also been attempted, the physiological findings, such as ascites and varices, are seen after progression to cirrhosis, Reference Schleiger, Kramer and Salzmann5,Reference Lewis, Reardon and Aboulhosn6 making this score unsuitable for early assessment. More quantitative clinical markers that can be used from the early stages of Fontan-associated liver disease need to be established.
Non-contrast computed tomography attenuation values of the parenchyma have been widely used to assess the progression of steatohepatitis, another diffuse hepatic disorder, because the technique is quantitative and reproducible. Reference Iwasaki, Takada and Hayashi7,Reference Goto, Kawamura and Watanabe8 We hypothesised that early liver histological changes caused by Fontan-associated liver disease would also be detected by changes in hepatic CT attenuation values, potentially serving as a quantitative clinical marker. This study sought to determine the usefulness of hepatic CT attenuation values for assessing the progression of Fontan-associated liver disease.
Methods
Ethical statement
This observational study was approved by the institutional ethics committee of Osaka University Hospital (date of approval: August 22, 2023; approval number: 23,135(T3)-3). Written consent for the study was obtained from all included patients.
Selection of patients
We reviewed all data of patients who underwent Fontan operation between 1980 and 2023 at our institution. Patients with follow-up data for more than 20 years postoperatively were included. Patients without any abdominal CT after Fontan and those who received any treatment for liver diseases other than Fontan-associated liver disease were excluded.
Cirrhosis
Diagnosis of cirrhosis was made by F4 grade fibrosis on liver biopsy using the METAVIR scoring system. Reference Manning and Afdhal9 The indication of liver biopsy for post-Fontan patients in this study period was the presence of ascites or irregularity of the liver surface on ultrasonography or CT. The representative histological findings of liver biopsy in a patient with cirrhosis are shown in Supplementary Fig. 1.
Hepatic CT attenuation values
The following CT instruments were used: Discovery CT750, Revolution, LightSpeed Ultra (GE Healthcare, Chicago, IL, USA); Aquilion, Aquilion ONE (CANON, Tokyo, Japan); and XVigor (Toshiba, Tokyo, Japan). All tube voltages were 120 kVp, and the slice thickness was 5 mm. A SYNAPSE system (FUJIFILM, Tokyo, Japan) was used to view all CT images. To extract hepatic CT attenuation values, a region of interest with a 20-mm-diameter circle was established. The minimum, maximum, mean, and standard deviation of the hepatic CT attenuation values were obtained from each region of interest. To correct batch effects between examinations, CT attenuation values of the descending aorta within the same slice were also measured using 10-mm-diameter circular regions of interest. Five regions of interest were set in the same slice, at a 10 mm margin from the liver surface, in the slice containing the umbilical portion of the axial image. The minimum, maximum, and mean values obtained from the five regions of interest were averaged and divided by the mean value of the regions of interest in the descending aorta, defined as the adjusted minimum, maximum, and mean values, which were used for the downstream analyses. The same adjustment was conducted using the CT attenuation values of the spleen. Supplementary Fig. 2A shows the schema to obtain hepatic CT attenuation values.
Study methods
Changes in the adjusted minimum, maximum, and mean values from postoperative abdominal CT scans were analysed as time-dependent covariates to examine their association with the development of cirrhosis, i.e., using the values obtained from the latest CT examination as of the time of censoring or the outcome. Additionally, the relationships between these values and the Model of End-Stage Liver Disease Excluding International Normalized Ratio score at each examination time point were evaluated. In a subgroup of patients with cirrhosis, the time-course changes in adjusted minimum, maximum, and mean values before the development of cirrhosis were analysed and compared with concomitant Model of End-Stage Liver Disease Excluding International Normalized Ratio scores. Each measured time point was classified into three categories: within 5, 10, and 30 years after the Fontan operation, defined as t5, t10, and t30, respectively. The scheme of the treating CT attenuation values as time-dependent covariates is presented in Supplementary Fig. 2B.
Statistical analysis
Measurements are reported as median (range or interquartile range [represented as 25–75%] as appropriate) or frequencies (%). Associated factors for the development of cirrhosis were determined using the Cox proportional hazard model. The adjusted minimum, maximum, and mean values and Model of End-Stage Liver Disease Excluding International Normalized Ratio scores were used as time-dependent covariates. Preoperative factors, including age at operation, type of Fontan operation, mean pulmonary artery pressure, and moderate or greater atrioventricular valve regurgitation, were treated as time-independent covariates. For all time points of CT examination, Pearson’s product-moment correlations among adjusted minimum, maximum, and mean values; standard deviation; and Model of End-Stage Liver Disease Excluding International Normalized Ratio scores were analysed. Time-dependent receiver operating characteristic curves were generated for the adjusted minimum and maximum values at 20 and 30 years postoperatively, with cut-off values determined using Youden’s index. Time-course changes in these values and scores were assessed based on changes between two consecutive time points, t5–t10 and t10–t30 that were examined in mixed effect models. Inter-individual variations were treated as random effects, and time was treated as a fixed effect. All statistical analyses were performed using R version 4.3.1 (R Foundation for Statistical Computing, Vienna, Austria).
Results
Patient characteristics
During the study period, 69 patients underwent the Fontan operation at our institution. Thirty-six patients were followed up for more than 20 years. All had a history of at least one blood transfusion before or at the Fontan operation, and no case was diagnosed as having any liver disease other than Fontan-associated liver disease. Six patients did not undergo postoperative abdominal CT scans. A total of 30 patients were included in this study (Figure 1 a). The median age at Fontan operation was 4.7 years. The type of Fontan operation was atriopulmonary connection in 7 patients (23%), Bjork in 4 (13%), and total cavopulmonary connection in 19 (63%). Sixteen patients had dominant left ventricle. Fenestration was placed in six patients. Moderate or greater atrioventricular valve regurgitation was identified in five patients (17%). Table 1 shows the patient characteristics.
Figure 1. (a) Flowchart of patient selection and timing of computed tomography (CT) examinations. (b) Survival curve for diagnosis of cirrhosis among all patients.
Table 1. Patients’ characteristics
APC = atriopulmonary connection; TCPC = total cavopulmonary connection; AVVR = atrioventricular valve regurgitation; SVCP = superior caval vein pressure.
Outcomes
The median follow-up period after the Fontan operation was 29.0 years (interquartile range: 23.7–33.5). There was no mortality during the follow-up period. Liver biopsy was performed in nine patients at a median 26.2 years after the Fontan operation, and cirrhosis was diagnosed in all patients. Event-free curves for all patients are shown in Figure 1(b).
Associated factors for cirrhosis
In univariates analysis of cox proportional hazards models, the factors associated with cirrhosis were age at Fontan operation (hazard ratio: 1.08 [1.02–1.15, p = 0.01]), adjusted minimum (hazard ratio: 0.01 [0.00–0.02, p < 0.01]), adjusted maximum (hazard ratio: 159 [4.34–5831, p < 0.01]), standard deviation (hazard ratio: 1.89 [1.29–2.76, p < 0.01]), and Model of End-Stage Liver Disease Excluding International Normalized Ratio score (hazard ratio: 1.2 [1.02–1.43, p = 0.03]) (Table 2). Supplemental Table shows the hazard ratios of the hepatic CT attenuation values adjusted by that of the spleen.
Table 2. Univariate analysis with Cox proportional hazard model
HR = hazard ratio; CI = confidence interval; LV = left ventricle; TCPC = total cavopulmonary connection; SVCP = superior caval vein pressure; AVVR = atrioventricular valve regurgitation; MELD-XI = Model of End-Stage Liver Disease Excluding International Normalized Ratio.
Correlation between CT attenuation values and Model of End-stage Liver Disease Excluding International Normalised Ratio Scores
The adjusted minimum value (r2 = −0.22, p = 0.04) and standard deviation (r2 = 0.23, p = 0.04) were significantly correlated with the concomitant Model of End-Stage Liver Disease Excluding International Normalized Ratio score among 84 time points, whereas the adjusted maximum and mean values did not show a significant correlation with the Model of End-Stage Liver Disease Excluding International Normalized Ratio scores (Figure 2 a). In the receiver operating characteristic curves at 30 years postoperatively, the area under the curve values for the adjusted minimum and maximum were 0.93 and 0.94, respectively, with cut-off values of 0.48 and 2.11 (Supplementary Fig. 4).
Figure 2. (a) Scatterplots depicting correlations between hepatic CT attenuation values and concomitant MELD-XI scores at each CT examinations. Correlation coefficients (r2) and p-values are also shown alongside. MELD-XI = Model for End-Stage Liver Disease Excluding International Normalized Ratio; aVMIN, aVMAX, and aVMEAN = adjusted minimum, maximum, and mean of the hepatic CT attenuation values, respectively. (b) Boxplots depicting longitudinal trend of hepatic CT attenuation values; aVMIN, aVMAX, aVMEAN, SD, and corresponding MELD-XI scores prior to diagnosis of cirrhosis. Significance denoted: ns, (p > 0.05); *, (p < 0.05); **, (p < 0.01); ***, (p < 0.001) above brackets. MELD-XI = Model for End-Stage Liver Disease Excluding International Normalized Ratio; aVMIN, aVMAX, aVMEAN = adjusted minimum, maximum, and mean of the hepatic CT attenuation values, respectively.
Time-course change of hepatic CT attenuation values and Model of End-stage Liver Disease Excluding International Normalised Ratio scores in patients with cirrhosis
In nine patients diagnosed with cirrhosis, 27 CT examinations were included for longitudinal assessment. At a median 26.8 years after the Fontan operation, adjusted minimum value significantly decreased (0.73 vs. 0.46, p < 0.05) and standard deviation significantly increased (9.03 vs. 12.0, p < 0.05) between t10 and t30, whereas Model of End-Stage Liver Disease Excluding International Normalized Ratio scores and adjusted maximum value were not. Boxplots and statistical significances of values are shown in Figure 2(b). Trends in CT attenuation values adjusted for the spleen are shown in supplementary fig. 3.
Discussion
In this single-centre retrospective study, the hepatic CT attenuation values adjusted by that of the descending aorta obtained at multiple time points were evaluated in post-Fontan patients. In the peripheral region of the liver, the minimum CT attenuation values decreased, and the maximum value was unaltered; subsequently, the standard deviation increased. The hepatic CT attenuation and Model of End-Stage Liver Disease Excluding International Normalized Ratio scores were significantly associated with cirrhosis in the Cox proportional hazard model. In patients with cirrhosis, the hepatic CT attenuation showed a significant time-course change, whereas the Model of End-Stage Liver Disease Excluding International Normalized Ratio scores did not. Imaging modalities such as ultrasonography, CT, and magnetic resonance imaging are employed in Fontan-associated liver disease evaluation. However, these findings were limited to highlighting the findings specific to cirrhosis, which is the terminal presentation of all liver diseases. Reference Sethasathien, Leemasawat, Silvilairat, Sittiwangkul, Chattipakorn and Chattipakorn10 In the present study, CT attenuation values were also examined in the period prior to the diagnosis of cirrhosis, highlighting the possibility of detecting the hepatic parenchymal alterations specific to Fontan-associated liver disease in early stages.
Hepatic CT attenuation value has been commonly used for the assessment of the progression of steatohepatitis. In steatohepatitis, as the examination equipment type and tube voltage affect the hepatic CT attenuation values, the values divided by those of the spleen within the same slice are used to correct variances between examinations. Reference Piekarski, Goldberg, Royal, Axel and Moss11 In patients with Fontan circulation, however, the effect of splenic congestion should be considered, as the splenic congestion could lead to unprecise correction. In this study, we selected the CT attenuation value of the descending aorta as the control instead of that of the spleen because the CT attenuation value of blood is not involved in the degree of venous congestion. Reference New and Aronow12 In the present study, CT attenuation values of the spleen in the same slice were also measured to use as a control, which showed that the values were altered similarly to those with the aortic control.
Hepatic venous congestion leads to the gradual progression of sinusoidal dilatation towards the peripheral region in patients with Fontan-associated liver disease. Reference Kendall, Stedman and Hacking13,Reference Schwartz, Sullivan and Glatz14 Additionally, hepatic venous endothelial injury subsequently leads to fibrosis of the space of Disse. Reference Wu, Jonas and Opotowsky15 Given that the liver parenchyma or hepatocytes are not simultaneously nor diffusely affected, the CT attenuation values may remain unchanged in areas where function and morphology are preserved. This suggests that the CT attenuation values within given regions of interest do not uniformly decrease, as demonstrated in this study.
A more accurate assessment of Fontan-associated liver disease progression by imaging findings could bring a clinical implication of multi-modality evaluation of the progression, i.e., a scoring system of Fontan-associated liver disease. Recently, various modalities such as imaging, histology, or even genetics can contribute to giving features for those scoring models in liver disease. Reference Kendall, Jimenez-Ramos and Turner16 A similar concept could be developed for Fontan-associated liver disease, which has been recognised as difficult to predict in the early stage.
This is a pilot study with several limitations, including being conducted in a single institution, being retrospective, and having a small number of participants. Therefore, further robust validation studies are needed to develop suitable methods and verify the results of this study.
Supplementary material
The supplementary material for this article can be found at https://doi.org/10.1017/S1047951125000393.
Data availability statement
The data used for this study are available from the corresponding author upon reasonable request.
Acknowledgements
We thank Editage (www.editage.jp) for their assistance with English language editing.
Author contributions
Yosuke Kugo: Conceptualisation, methodology, formal analysis, investigation, data curation, resources, and original draft writing. Takashi Kido: Total advisement of the study and review of the draft. Moyu Hasegawa: Review. Toshiaki Nagashima: Review. Takuji Watanabe: Review. Masaki Taira: Review. Shigeru Miyagawa: Review.
Financial support
The authors declare they have not received any particular funding for this study.
Competing interests
None.
Ethical standards
The authors assert that all procedures contributing to this work comply with the ethical standards of the relevant national guidelines on human experimentation (please name) and with the Helsinki Declaration of 1975, as revised in 2008, and has been approved by the institutional ethics committee of Osaka University Hospital (date of approval: August 22, 2023; approval number: 23,135(T3)-3).
Open access
