Regional Elevation of Liver T1 in Fontan Patients

Abstract

Background

Fontan-associated liver disease (FALD) is characterized by hepatic congestion and progressive hepatic fibrosis in patients with the Fontan operation. This condition is generally clinically silent until late, necessitating techniques for early detection. Liver T1 mapping has been used to screen for FALD, but without consideration of regional variations in T1 values.

Methods

Liver T1 measured with a liver-specific T1 mapping sequence (PROFIT1) in Fontan patients was compared with cohorts of patients with biventricular congenital heart disease (BiV-CHD) and controls with normal cardiac function and anatomy.

Results

Liver T1 was significantly elevated in the Fontan cohort (n = 20) compared with patients with BiV-CHD (n = 12) and controls (n = 9) (781, 678, and 675 milliseconds, respectively; P < 0.001), with a consistent pattern of significantly elevated T1 values in the peripheral compared with central liver regions (ΔT1 = 54, 2, and 11 milliseconds; P < 0.001). PROFIT1 also yielded simultaneous T2∗ maps and fat fraction values that were similar in all groups. Fontan liver T1 values were also significantly elevated as compared with BiV-CHD and controls as measured with the cardiac (modified Look-Locker inversion) acquisitions (728, 583, and 583 milliseconds, respectively; P < 0.001) and values correlated with PROFIT1 liver T1 (R = 0.87, P < 0.001).

Conclusions

Fontan patients have globally increased liver T1 values and consistent spatial variations, with higher values in the peripheral liver regions as compared with spatially uniform values in BiV-CHD and controls. The spatial patterns may provide insight into the progression of FALD. Liver T1 mapping studies should include uniform spatial coverage to avoid bias based on slice locations in this population.

Graphical abstract

The Fontan procedure is the final surgical stage for congenital heart disease (CHD) patients with single ventricles, routing systemic venous return directly to the lungs without a pump, while the single ventricle pumps blood to the body. Since the first Fontan procedure carried out in 1971, Fontan survival has significantly improved from 79% before the 1990s to the current 10-year survival of over 90%.^1,2 With an increasing number of Fontan survivors, a serious complication known as Fontan-associated liver disease (FALD) has emerged as a result of chronically elevated systemic venous pressures from the passive venous flow into the lungs.³

FALD is characterized by liver fibrosis, which can lead to cirrhosis and even hepatocellular carcinoma.⁴ Once this has occurred, the only treatment option is a combined heart-liver transplant. Screening for FALD has proved to be challenging. Although every Fontan patient has an element of subclinical liver disease, severity does not necessarily correlate with the time since Fontan.³ Liver enzymes remain normal or slightly increased until an advanced state and do not correlate with the degree of fibrosis.⁵ Many centres initially recommended screening liver biopsy, which is considered the gold standard for fibrosis, in asymptomatic patients no less than 10 years after the Fontan operation.⁶ Other studies have suggested that biopsies should be reserved for when imaging or laboratory studies suggest cirrhosis. Recent publications by the European Society of Cardiology and the American Heart Association indicate that the current consensus recommendation is for regular blood biomarkers and imaging, but the exact type of imaging is not specified.^7,8

Magnetic resonance imaging (MRI) T1 mapping offers a widely available test of tissue characteristics that is commonly used to detect fibrosis via increased T1 values.^9,10 Reports of increased liver T1 values in the Fontan population are emerging.11, 12, 13, 14 MRI is routinely used for the assessment of the Fontan circuit, ventricular function, and collateral burden, and views of the liver are consistently acquired within the short-axis cardiac T1 mapping view, which has enabled opportunistic liver T1 studies.11, 12, 13, 14, 15 However, the commonly used modified Look-Locker inversion (MOLLI) cardiac T1 mapping approach has well-documented systematic T1 errors, in particular from the presence of fat or iron,16, 17, 18 and the oblique partial coverage of the liver in many previous studies has precluded consideration of regional variations in T1. Proton density fat fraction with T2∗ imaging with water-specific T1 (PROFIT1)¹⁶ is a novel liver-specific sequence that aims to overcome some of these limitations. It incorporates proton density fat fraction (PDFF) imaging to separate and quantify water and fat content. The water and fat separation enables the measurement of water-specific T1 values, without contamination by fat, simultaneously with T2∗ maps for correction for iron. The aim of this study was to use PROFIT1 to evaluate and compare liver T1 values in Fontan patients with biventricular CHD patients and controls, with consideration of the spatial variations in T1 values. Secondary aims were to compare these findings with conventional functional cardiac parameters and the more commonly used cardiac-specific T1 mapping sequence (MOLLI).¹⁹

Materials and Methods

Patient population

This cross-sectional cohort study involved patients aged 0-18 who underwent cardiac MRI at the Mazankowski Heart Institute, University of Alberta. These were divided into 3 groups: Fontan patients, patients with biventricular CHD (BiV-CHD), and control subjects. The inclusion criterion for the Fontan group was any patient with a Fontan circulation. The inclusion criterion for BiV-CHD was any patient with a congenital heart lesion with 2 ventricles whether they had undergone previous repair or not, with or without residual lesions. The criteria for the control subjects were otherwise healthy patients who underwent screening MRI for electrocardiogram (ECG) abnormalities, unexplained cardiac symptoms, or a family history of inherited conditions with normal physical examination, no congenital heart lesions, and normal cardiac anatomy and function on MRI. Exclusion criteria included patients with iron overload.

Informed consent was obtained to acquire additional liver sequences during the time of their cardiac MRI. Parental consent was obtained for patients aged 0-14, and patient consent was obtained in the case of mature minors (>15 years of age). The study was approved by our institution’s research ethics board and adhered to institution research guidelines.

Clinical data

Demographic and clinical data collected included age, heart rate, blood pressure, height, weight, body surface area, cardiac anatomy, and indication for MRI. For Fontan patients, data also included age at Fontan, presence or absence of fenestration (including device closure), liver elastography and biopsy results (if available), and other Fontan complications. Study data were anonymized and entered using a secure, web-based software platform: REDCap (Research Electronic Data Capture) hosted at University of Alberta.^20,21

Magnetic resonance imaging

MRI examinations were performed on a 1.5 T Siemens Aera scanner (Siemens Medical Solutions, Erlangen, Germany). The cardiac protocol included balanced steady-state-free-precession contiguous short-axis oblique cine images with echo time (TE) = 1.17-1.43 milliseconds, repetition time (TR) = 34.56-52.35 milliseconds, 56° flip angle, 4-7 mm slice thickness, 0%-20% gap, matrix size = 256 × 126, field of view (FOV) = 273 × 350, and 30 reconstructed cardiac phases. MRI analysis quantified ventricular volumes, mass, and ejection fraction (EF) of the dominant ventricle in single ventricles and the left ventricle in controls using CVI 42 (Circle Cardiovascular Imaging, Calgary, Canada).

Liver images were acquired using the PROFIT1 method.¹⁵ TE = 1.60, 3.65, 5.70, and 7.75 milliseconds, TR = 12.0 milliseconds, 30° flip angle, 6 mm slice thickness, FOV = 420 × 260 mm with a 160 × 120 matrix, and GeneRalized Autocalibrating Partial Parallel Acquisition = 2. The images were acquired during free breathing in sedated patients (3 averages over 40 seconds) and breath-holding (13 seconds) in nonsedated patients if possible. Two or three transverse slices were acquired at the widest dimension of the liver (Fig. 1, A-C). Average T1, PDFF, and T2∗ values from all slices were calculated after automated removal of blood vessels. Regional liver T1 analysis included the placement of a circular region of interest in the central liver region to identify central and peripheral liver regions. All regional analyses were performed blinded to the group.

Figure 1.

Sample PROFIT1 and MOLLI images of a Fontan patient. PROFIT1 (1 of 3 slices shown) yields T1 maps (A), PDFF maps (B), and T2∗ maps (C). A MOLLI T1 map from the same subject (D), acquired using the cardiac short-axis slice orientation. The dashed lines outline the liver. MOLLI, modified Look-Locker inversion; PDFF, proton density fat fraction; PROFIT1, proton density fat fraction with T2∗ imaging with water-specific T1.

Myocardial T1 mapping used a 5(3)3 MOLLI sequence acquired in a single short-axis slice at the midventricular level as previously described.¹⁷ Two inversion recovery prepared ECG-synchronized Look-Locker acquisitions were performed with inversion times of 100 and 180 milliseconds, respectively, followed by 5 and 3 single-shot images after these inversion pulses with TE = 1.08 milliseconds, TR = 277.68 milliseconds, 35° flip angle, 7 mm slice thickness, and FOV = 300 × 400 with a 192 × 256 matrix. The number of heart beats between image acquisitions, 3 by default, was increased for higher heart rates to allow for adequate T1 recovery between inversion pulses. The images were acquired during free breathing in sedated patients and breath-holding in nonsedated patients if possible. An inline motion correction algorithm was used to register and align the 8 inversion recovery source images before the calculation of T1 maps. Myocardial T1 was measured in the septum and lateral wall.

Liver T1 values were also measured from the short-axis cardiac MOLLI acquisitions, which typically include a parasagittal region of the liver (Fig. 1D). Three regions of interest were drawn in the liver: (1) near the diaphragm, (2) near the hepatic hilum, and (3) caudal and peripheral portion of the liver, taking care to avoid major blood vessels, the gallbladder, and the surface of the liver. Results from all 3 regions were averaged. The oblique slice orientations used for short-axis cardiac imaging precluded analysis of central and peripheral liver regions with MOLLI.

Statistical analysis

Continuous variables are presented as median values (interquartile range). Cohort medians were compared with a nonparametric analysis of variance Kruskal-Wallis test, with pairwise comparisons calculated with Dunn’s post hoc test. The Kruskal-Wallis P value was used for significance with Dunn’s post hoc test used to determine where differences were, and P values were adjusted for multiple comparisons. Correlations were assessed using univariate regression analysis with nonparametric Spearman’s r value. Data analysis was performed with IBM SPSS Statistics for Windows, Version 25.0 (IBM Corp, Armonk, NY).

Results

Baseline clinical characteristics

Forty-two subjects met inclusion criteria, with 1 subject was excluded from the BiV-CHD cohort because of iron overload leaving a total of 41 subjects. There were 20 Fontan patients, 12 patients with BiV-CHD, and 9 controls. The cohorts did not differ significantly in age, height, or weight, and generally were either clinically asymptomatic or had mild exercise intolerance (Table 1). Fontan patients underwent the Fontan operation at a median age of 40 months (interquartile range: 33-44 months) with MRI scans at a median of 102 months (76-144 months) after surgery. Five Fontan patients had a fenestration at the time of the MRI, 9 patients had fenestrations that were previously closed, and 6 patients were nonfenestrated. The CHD diagnoses in Fontan patients were hypoplastic left heart (n = 12), tricuspid atresia (n = 2), isomerism (n = 3), complex transposition of the great arteries (n = 1), and unbalanced atrioventricular septal defect (n = 2). The types of CHD in the BiV-CHD group were unoperated atrial septal defect (n = 2), ventricular septal defect (n = 1), and pulmonary atresia with ventricular septal defect and collaterals (n = 1); and status post-repair or tetralogy of Fallot (n = 2), atrioventricular septal defect (n = 1), coarctation of the aorta (n = 1), interrupted aortic arch (n = 1), partial anomalous pulmonary venous return (n = 1), and d-transposition of the great arteries (n = 2). The indications for MRI in controls were family history or inherited cardiac condition (n = 4), palpitations (n = 1), syncope (n = 2), and ECG abnormalities (n = 2).

Table 1.

Demographics and clinical characteristics

	Fontan (n = 20)	BiV-CHD (n = 12)	Control (n = 9)	P value
Age (y)	11.5 (9.6-15.0)	13.1 (10.2-16.7)	10.6 (8.5-17.3)	0.728
BSA (m2)	1.20 (1.06-1.58)	1.54 (1.18-1.91)	1.19 (1.04-1.89)	0.323
Heart rate (bpm)	87 (74-103)	75 (62-92)	83 (52-101)	0.373
Clinical status	Mild exercise intolerance (n = 6), exercise intolerance (2), nocturnal hypoxemia (1), hepatomegaly (3), clubbing (3)	Mild exercise intolerance (4), fatigue (4), 1 midline sternotomy (5), 2 sternotomies (2), left thoracotomy (1), bypass required (6)	Mild fatigue (1)

BiV-CHD, biventricular congenital heart disease; BSA, body surface area.

Baseline MRI characteristics

Fontan patients had significantly increased ventricular volumes compared with controls but not as compared with patients with BiV-CHD (Table 2). Fontan systemic ventricular EF was significantly decreased compared with both BiV-CHD and controls (P < 0.001). There were no differences in indexed mass between the 3 groups. Myocardial septal and lateral MOLLI T1 values were significantly increased in Fontan patients compared with controls with no significant differences between patients with BiV-CHD and controls or Fontan patients (Table 2, Fig. 2A).

Table 2.

Cardiac and liver MRI characteristics and their significance in study cohorts

	Fontan (n = 20)	BiV-CHD (n = 12)	Control (n = 9)	P value (Kruskal-Wallis)
EDVi (mL/m2)	101.0 (83.0-125.5)∗	83.5 (72.5-90.3)	78.0 (74.5-86.5)∗	0.024
ESVi (mL/m2)	52.5 (35.5-83.0)∗	36.5 (27.8-41.0)	33.0 (29.0-39.5)∗	0.009
EF (%)	48.5 (37.0-54.8)∗,†	58.9 (53.5-61.8)†	58.0 (55.3-61.5)∗	<0.001
Mass index (g/m2)	49.5 (40.0-63.3)	51.0 (43.0-61.0)	48.0 (43.5-55.3)	NS
Septal MOLLI T1 (ms)	1051 (1024-1088)∗	1010 (988-1050)	994 (978-1019)∗	0.014
Lateral MOLLI T1 (ms)	1048 (1001-1073)∗	998 (946-1076)	980 (951-991)∗	0.032
Liver
T1 (average) (ms)	781 (745-836)∗,†	678 (622-705)†	675 (629-713)∗	<0.001
T1 (central) (ms)‡	714 (698-773)∗,†	676 (637-706)†	689 (624-698)∗	0.003
T1 (outer) (ms)‡	767 (741-842)∗,†	694 (632-720)†	677 (629-710)∗	<0.001
Outer-central (ms)‡	53.9 (41.6-67.3)∗,†	1.7 (-6.0-16.2)†	10.5 (1.5-17.7)∗	<0.001
Central/outer (%)	93.5 (91.3-94.5)∗,†	99.8 (97.8-101.0)†	98.5 (97.4-99.6)∗	<0.001
Liver PDFF (%)	1.6 (1.4-1.8)	1.8 (1.4-2.7)	1.9 (1.6-2.8)	NS
Liver T2∗ (ms)	29.2 (23.9-34.2)	25.6 (21.4-29.1)	23.7 (21.6-28.2)	0.053
Liver MOLLI T1 (ms)	728 (714-744)∗,†	583 (573-638)†	583 (555-617)∗	<0.001

BiV-CHD, biventricular congenital heart disease; EDVi, end diastolic volume indexed to BSA; EF, ejection fraction; ESVi, end systolic volume indexed to BSA; MOLLI, modified Look-Locker inversion; MRI, magnetic resonance imaging; NS, not significant; PDFF, proton density fat fraction.

^∗Indicates the significant difference between Fontan and control.

^†Indicates the significant difference between Fontan and BiV-CHD.

^‡From a single slice.

Figure 2.

T1 value comparison between study cohorts. (A) Myocardial T1 for the 3 cohorts. (B) Whole liver average PROFIT1 liver T1 values in the 3 cohorts. (C) Difference in PROFIT1 liver T1 values between the peripheral and central liver regions for the 3 cohorts. BiV-CHD, biventricular congenital heart disease; PROFIT1, proton density fat fraction with T2∗ imaging with water-specific T1.

Liver findings

PROFIT1 Fontan liver T1 values (whole liver average) were significantly higher than both BiV-CHD and controls (P < 0.001), with no differences between BiV-CHD and controls (Table 1, Fig. 2B). T2∗ values were similar in all groups (P = 0.053), and all 3 groups had similar fat fraction (PDFF) with median values of <2% in all groups (Table 2). The Fontan group had a consistent pattern of elevated T1 values in the peripheral liver regions relative to T1 values in the central region, as compared with spatially uniform liver T1 values in BiV-CHD and controls (Table 2, Fig. 3). The absolute difference in T1 values between central and peripheral regions was significantly larger in the Fontan group as compared with the other groups (Table 2, Fig. 2C), and the magnitude of the differences was also significantly associated with globally elevated liver T1 values (Fig. 4A). There was no difference in those with or without a fenestration at the time of MRI in PROFIT1 (P = 0.896) or liver T1 MOLLI (P = 0.961). Liver T1 values were inversely correlated with EF (r = −0.5; P = 0.001) and positively correlated with ventricular volumes (Table 3). Liver T1 values did not correlate with age (Table 3).

Figure 3.

Sample PROFIT1 liver T1 maps. (A) The control group, (B) the BiV-CHD group, and (C) the Fontan group. The lower panel is zoomed with a reduced range of T1 values to highlight the global and regional differences that also includes the blood vessel removal. BiV-CHD, biventricular congenital heart disease.

Figure 4.

Correlation plots between the PROFIT1 liver T1 values (whole liver average). (A) With the T1 differences between peripheral and central liver regions, (B) with myocardial T1 values (septum), and (C) with the MOLLI liver T1 values from the cardiac acquisition. BiV-CHD, biventricular congenital heart disease; MOLLI, modified Look-Locker inversion; PROFIT1, proton density fat fraction with T2∗ imaging with water-specific T1.

Table 3.

Correlations of PROFIT1 liver T1 with clinical and MRI characteristics of all patient cohorts combined and the Fontan cohort alone

Variable	All patients			Fontan patients
	Correlation coefficient	P value (2-tailed)	N	Correlation coefficient	P value (2-tailed)	N
Age (mo)	0.150	0.348	41	0.217	0.359	20
Time since Fontan				0.185	0.435	20
Age at Fontan				−0.103	0.667	20
EF of primary ventricle	−0.501	0.001	41	0.163	0.493	20
SVEDVi	0.355	0.023	41	−0.041	0.863	20
SVESVi	0.425	0.006	41	−0.053	0.826	20
Mass index	0.119	0.524	31	0.578	0.049	12
Septum T1	0.346	0.036	37	−0.148	0.559	18
Lateral wall T1	0.241	0.157	36	0.324	0.205	17
Liver T1 (MOLLI)	0.869	<0.001	37	0.515	0.029	18

EF, ejection fraction; MOLLI, modified Look-Locker inversion; MRI, magnetic resonance imaging; PROFIT1, proton density fat fraction with T2∗ imaging with water-specific T1; SVEDVi, systemic ventricle end diastolic volume indexed to BSA; SVESVi, systemic ventricle end systolic volume indexed to BSA.

Comparison between PROFIT1 and MOLLI

The liver T1 values measured from the MOLLI cardiac acquisitions were significantly correlated with the whole-liver average PROFIT1 liver T1 values (Fig. 4C, r = 0.87; P < 0.001) but underestimated T1 values by 50-100 milliseconds in all groups. There was a weak correlation between PROFIT1 liver T1 and MOLLI cardiac septal T1 (r = 0.346; P = 0.036) (Table 3, Fig. 4B).

Discussion

Using a liver-specific T1 mapping method (PROFIT1), we demonstrated globally elevated liver T1 values in Fontan patients with a nonuniform spatial distribution of increased T1 values in the periphery of the liver and a larger spatial variation in those with more elevated global liver T1 values. This pattern has not previously been reported to our knowledge and may provide insight into the distinct presentation of congestion and fibrosis in FALD, which remains an unresolved challenge. By contrast, both controls and BiV-CHD did not have significant differences between central and peripheral liver regions.

Quantitative T1 mapping with MRI has been shown to be strongly associated with diffuse myocardial fibrosis in patients with myocardial disease.^9,22 Quantification of liver T1 has been found to reflect histologically proven fibrosis in a mouse model of liver fibrosis,²³ although no biopsy-validated studies with comparison to T1 mapping have been performed in the Fontan population. Like previous studies in the Fontan population,11, 12, 13, 14, 15 opportunistic MOLLI liver T1 maps were obtained at the time of a cardiac T1 acquisition and similarly showed liver T1 elevation. Although this approach requires no additional scan time, it may miss regional variations in T1 as found in the current study, which incorporated more uniform spatial coverage of the liver. The observed regional variations in liver T1 is a new finding, and while unexpected, it is consistent with literature describing FALD as a heterogeneous disease.²⁴ Morphologic abnormalities of the liver in Fontan patients have shown nonuniform and patchy changes on MRI, which is not captured using the gold standard liver biopsy.²⁵ With very few reports in children, a comparable population is adults with chronic hepatitis C where the right lobe of the liver demonstrated the highest fibrotic grade, indicating regional variations in liver fibrosis. They postulated that the proximity of the lobe to the main portal vein may explain their finding of more advanced fibrosis in the right lobe of the liver.²⁶ Extrapolating this to our population, we can only hypothesize that the periphery of the liver may be more affected by chronic venous congestion, resulting in regional variations in liver T1 values.

The poorer agreement between PROFIT1 and MOLLI within the Fontan group may reflect the regional variations in liver T1 in this group, and thus, a dependence on slice locations. Future studies with matched slice locations with more uniform coverage of the liver will be necessary to compare the MOLLI and PROFIT1 methods. Previous MOLLI T1 studies in the Fontan liver have yielded similarly increased T1 values as the current study,11, 12, 13, 14, 15 using cardiac slice orientations, and thus, regional variations in T1 values comparable to the PROFIT1 component of the current study would not have been feasible.

The PROFIT1 approach, in addition to being more time efficient than the MOLLI (eg, 3 slices of coverage vs 1 slice in the same acquisition time), also provides accurate water-specific T1 values that are not confounded by the presence of fat, as well as providing intrinsically registered fat fraction and T2∗ maps. The ability to detect fatty liver in older subjects will add to interpretation of elevated T1 values, where both chronic congestion and high liver fat content can contribute to fibrosis. Similar to previous studies, the MOLLI technique was found to significantly underestimate T1 values.16, 17, 18^,27,28 All participants in the current study had relatively low PDFF values (<2%), indicating that fat-related liver T1 errors using MOLLI were likely small, and T2∗ values were also normal and similar in all groups. Despite the T1 differences between PROFIT1 and MOLLI, the T1 values were highly correlated, suggesting that either method could be appropriate for the identification of increased values; however, a systematic comparison with matched slice locations is necessary to better characterize potential method differences. T2∗ mapping to detect the accumulation of iron is valuable as recent research shows that even normal physiologic increases in iron decrease T1 values in a mostly linear manner,¹⁸ which can confound the interpretation of T1 values. A method for correction of liver T1 values for the effects of iron has been illustrated.²⁹ Finally, MOLLI T1 values are elevated in the presence of increased T2 values (ie, T1 underestimation errors are reduced with increasing T2), and thus, the T1 values potentially reflect multiple underlying mechanisms.

Noninvasive imaging methods to detect FALD using liver elastography (ultrasound or MR elastography) are based on measurements of liver stiffness. Paediatric studies applying ultrasound elastography have found that hepatic stiffness was markedly increased in Fontan patients compared with controls but also increased as soon as 2.5 and 7.5 days after the Fontan operation. This and other studies have suggested that an earlier process may be responsible for these changes.^30,31 Increasing Fontan liver stiffness was found to occur with increasing time from Fontan surgery using MR elastography in adult Fontan patients.^32,33 Of note is that age or time from Fontan did not correlate with liver T1, given that time from Fontan is the biggest risk factor for FALD. However, because our population was a younger age group, it is likely that it was too early to detect such a correlation. Elastography studies in the Fontan population have not yet been shown to be predictive of outcomes, and the technique is somewhat limited by the availability of specialized equipment.³⁴ PROFIT1 has the advantage that it can be incorporated into MRI protocols commonly used in children already undergoing a cardiac MRI.

In addition to controls, we chose a group of BiV-CHD patients with a range of lesions, some of whom had previously undergone cardiopulmonary bypass surgery. Although Fontan patients typically undergo 3 cardiac surgical procedures compared with a range of 0-2 surgeries in the BiV-CHD group, the finding of similar liver and myocardial T1 in the BiV-CHD group and controls suggests that residual shunt lesions and cardiopulmonary bypass are not major factors responsible for these differences.

As previously mentioned, there are no current treatments for FALD, and the only option is to optimize the Fontan circulation as much as possible, particularly anatomic features such as stenoses in any portion of the circuit. Within the Fontan group, the weak positive correlation between ventricular mass and liver T1 suggests that patients with more cardiac compensation may have more fibrosis or congestion. However, liver T1 in the Fontan group lacks correlation with other cardiac indicators such as myocardial T1 and EF. Although this may be partially explained by the limited number of patients in the study, it also suggests a more complex relationship between the heart and liver and indeed may argue for the use of a new measurement such as liver T1, given that the indicators do not always agree. Unfortunately, correlations with other common indicators of liver disease were not possible in our study due to lack of blood biomarkers close to the time of MRI.

Our study is limited by a small sample size, as well as difficulties in obtaining a large number of young controls. Our BiV-CHD group was also a heterogeneous group of patients with a range of shunt lesions and surgical histories. The different T1 values between MOLLI and PROFIT1 were influenced by differing slice positions, and future studies should match slice locations. Few Fontan patients had livery enzyme tests close to the time of their MRI, although these have not shown to correlate with severity of fibrosis.²⁵ Further, this is largely due to a paediatric population, but it is difficult to correlate fibrosis with clinical outcomes given that our patient population lacks significant complications like protein losing enteropathy or Fontan failure. In the Fontan group, 4 underwent ultrasound elastography within 1 year of the MRI showing stiffness values ranging from 5.64 to 7 kPa. Liver biopsy was performed in another 3 patients due to either abnormal liver enzymes or imaging findings. Patient 1 showed sinusoidal dilation with mild congestion and mild pericentral fibrosis; patient 2 showed focal subcapsular bridging fibrosis, sinusoidal dilatation, no hepatocyte injury, or cirrhosis—findings consistent with congestive hepatopathy; and patient 3 showed no sinusoidal dilatation or intercellular fibrosis. The limited number of biopsies collected in the current study were not sufficient to provide validation of the PROFIT1 values. Further studies performing concurrent MRIs near the time of a liver biopsy are necessary to validate our technique, particularly in the adult Fontan population. Being aware of the underlying mechanisms for increased liver T1 should enable the development of Fontan-specific liver metrics, and serial evaluations with correlative liver biopsies may help to determine at what level or what change over time increased liver T1 indicates diffuse fibrosis.

Conclusions

Using a novel liver T1 mapping technique, we found increased liver T1 in a young Fontan population compared with biventricular CHD and controls with a nonuniform spatial distribution of elevated T1 values in the peripheral liver regions. The observed spatial variations in liver T1 may provide new insight into the time course of events in FALD, in particular, the transition from congestion to fibrosis. In the future, incorporation of dedicated liver T1 mapping sequences in routine Fontan protocols will provide the serial information needed to characterize the progression of the FALD pathology, ideally in conjunction with tissue biopsy. Different T1 mapping methods can yield dramatically different T1 values with characteristic dependences on other factors, such as fat content and pulse sequence parameters, and thus, the use of consistent protocols and uniform spatial coverage are essential.