Biomarker Profile in Stable Fontan Patients

Abstract

Background:

As the population of adults with congenital heart disease (CHD) grows, cardiologists continue to encounter patients with complex anatomies that challenge the standard treatment of care. Single ventricle Fontan palliated patients are the most complex within CHD, with a high morbidity and mortality burden. Factors driving this early demise are largely unknown.

Methods and Results:

We analyzed biomarker expression in 44 stable Fontan outpatients (29.2 ± 10.7 years, 68.2% female) seen in the outpatient Emory Adult Congenital Heart Center and compared them to 32 age, gender and race matched controls. In comparison to controls, Fontan patients had elevated levels of multiple cytokines within the inflammatory pathway including Tumor Necrosis Factor-α (TNF-α) (p<0.001), Interleukin-6 (IL-6) (p<0.011), Growth Derived Factor-15 (GDF-15) (p<0.0001), β2-macroglobulin, (p=0.0006), stem cell mobilization: SDF-1α (p=0.006), extracellular matrix turnover: Collagen IV (p<0.0001), neurohormonal activation: Renin (p<0.0001), renal dysfunction: Cystatin C (p<0.0001) and Urokinase receptor (uPAR) (p=0.022), cardiac injury: Troponin-I (p<0.0004) and metabolism: Adiponectin (p=0.0037). Within our baseline -stable- Fontan patients, 50% had hospitalizations, arrhythmias and worsening hepatic function within 1 year. GDF-15 was significantly increased in Fontan patients with clinical events (p<0.0001). In addition, GDF-15 moderately correlated with longer duration of Fontan (r=0.55, p=0.01) and was elevated in atriopulmonary (AP) Fontan circulation. Finally, in a multivariate model, VEGF-D and Collagen IV levels were found to be associated with a change in MELDXI, a marker of liver function.

Conclusion:

Multiple clinical and molecular biomarkers are upregulated in Fontan patients, suggesting a state of chronic systemic dysregulation.

Introduction:

Over the last 50 years, the Fontan operation has been transformative in improving the survival of children born with severe ventricular hypoplasia¹. However, Fontan patients have a higher burden of mortality and morbidity within 20 years of their Fontan operation². Fontan failure, a diverse range of clinical presentations encompassing heart failure and/or portal hypertension, results in a need for Fontan revision or transplantation, without which the patient can progress to premature death. Fontan failure is progressive, similar to heart failure resulting from acquired heart disease in adults. However, while acquired heart disease is often secondary to hypertension or atherosclerosis, Fontan failure is a consequence of long-term hemodynamic effects of a passive non-pulsatile pulmonary blood flow. Therefore, Fontan patients provide unique insights into effects of hemodynamic perturbations on cardiovascular disease in isolation from other comorbidities that affect the aging heart failure population. In addition, Fontan circulation models right heart failure seen in acquired heart failure. Clinically, Fontan patients have a high incidence of supraventricular arrhythmias, reduced aerobic exercise capacity, cardiac fibrosis, and diastolic and systolic heart failure that are driven mostly by their cardiovascular anatomy.

Identifying systemic biological factors that drive Fontan sequelae is integral to preventing clinical events that contribute to early patient demise. Standard markers of cardiac dysfunction such brain natriuretic peptide (BNP) are not reliable indicators of volume overload or worsening cardiac dysfunction and correlate instead with Fontan subtypes³. Understanding Fontan failure requires identifying clinical as well as molecular biomarkers that provide insight into expected adaptive responses of the human body with single ventricle physiology and differentiating them from pathologic markers of progressive Fontan failure. Extensive research has been conducted in biomarker profiles in acquired heart failure in adults. Braundwald⁴ and others^5–8 have classified these biomarkers into physiologic mechanisms of action and associated their dysregulation with outcomes. However, although studies have evaluated biomarkers in isolation, comprehensive studies related to biomarker profiles in patients with Fontan circulation have not been conducted.

In this study, we prospectively enrolled 44 stable Fontan patients at the Emory Adult Congenital Heart Center (ACHC), in an outpatient setting, between 2014 and 2016 and (a) measured serum biomarkers reflecting pathways associated with inflammation, hepatic extracellular matrix turnover, renal and cardiac dysfunction and angiogenesis, and compared these biomarker levels with 32 age-, gender-, and race-matched controls; and (b) evaluated association of serum biomarker levels with clinical events. We hypothesized that multiple biomarkers would be dysregulated in Fontan patients, given that Fontan circulation has systemic effects.

METHODS:

Study Population

The study was approved by the Institutional Review Board (IRB) at Emory University. We prospectively enrolled Fontan patients >18 years of age who were seen in the ambulatory setting at the Emory ACHC with univentricular physiology from December 2014–2016. Exclusion criteria included incomplete Fontan repair (including aortopulmonary shunts and Glenn circulation alone), patients with known genetic syndromes, those requiring continuous inotrope infusion, and patients unable to consent for themselves. “Stable” Fontan patients were defined as patients free from hospitalizations for >48 hrs within 3 months from enrollment, patients with anticipated Fontan revision, or listing for heart transplantation. Patients that died within 1 year of enrollment in the study were excluded. Fontan patients were matched to control subjects from the community for age (within ± 2.5 years in age), gender and race, and were self-reported to be healthy, were not taking medications and were non-smokers.

Clinical Assessment:

Fontan patient information including demographic data, anatomic diagnosis, New York Heart Association (NYHA) functional class, medication use, comorbidities, laboratory data, ejection fraction, valvular dysfunction, and clinical events including history of hospitalizations, history of arrhythmias and development of hepatocellular carcinoma (HCC) was obtained from review of medical records. “Cardiac” hospitalization were associated with arrhythmias, acute decompensated heart failure, PLE exacerbation or endocarditis. Clinical decompensation from non-cardiac causes such as dehydration or bacteremia can be exacerbated due to Fontan circulation, hence we also included non-cardiac causes of hospitalization as significant events. Hemodynamically significant arrhythmias, requiring either a change in anti-arrhythmic medication or procedural intervention such as cardioversion, ablation, or placement of a device were included in the study. HCC is a long-term sequalae that is associated with Fontan circulation and is fatal if not diagnosed early⁹. Diagnosis of HCC was based on characteristic contrast magnetic resonance imaging (MRI) abdominal imaging considered diagnostic of HCC. Patients had previous imaging that showed the absence of lesions suspicious of HCC. Patients considered to have developed HCC during the study had imaging changes considered diagnostic of HCC based on clinical practice guidelines for HCC¹⁰, which was confirmed by abdominal imaging experts and hepatologists (n=1). Patients with suspected metastatic HCC underwent biopsy of the metastatic lesion to confirm HCC (n=1). Clinical laboratory data and imaging data were within 6 months of baseline and 1-year mark.

Serum Biomarker Evaluation:

Blood was collected after venipuncture (BD Biosciences) and processed immediately after collection. Samples were centrifuged to remove cellular components and aliquots were distributed in 200 μl samples into Eppendorf tubes stored at −80°C. Biomarkers were evaluated using custom made Human Luminex Screening Multi-Plex assays (Single and Screening Multi-Plex assays), (R&D Systems). Biomarker panel included inflammatory biomarkers: Interleukin IL-1α, IL-1β, IL-6, IL-18 BPa, GDF-15, high sensitivity C reactive protein (hs-CRP) and β2-microglobulin; Angiogenesis: Vascular Endothelial Growth Factor–D (VEGF-D), and Stromal Derived Growth Factor-1α, (SDF-1α), Hepatic function: Collagen IV, Cardiac Injury: cardiac Troponin-I, (cTNI), Metabolic Function: Adiponectin, Neurohormonal activation: Renin, and Renal Dysfunction: Cystatin and uPAR. Renin levels were correlated to use of angiotensin converting enzyme inhibitors (ACEi) or Angiotensin Receptor Blocker (ARB). A subset of 23 random patients were further evaluated for Tumor Necrosis Factor-α (TNF-α) and compared similarly to their age, gender and race matched controls. The inter- and intra-assay coefficient of variation (CV%) were also calculated. In addition, standard laboratory values at baseline and 1 year (+/− 6 months) were recorded for direct bilirubin, total bilirubin, aspartate transaminase (AST), alanine transaminase (ALT), serum creatinine, alkaline phosphatase, albumin and platelet count. Model of End-Stage Liver Disease eXcluding INR (MELD XI) score was calculated using clinical laboratory data using the formula MELD-XI = 5.11 × ln (serum bilirubin in mg/dL) ± 11.76 × ln (serum creatinine in mg/dL) ± 9.44¹¹.

Statistical Analysis:

Initially, Spearman correlation was used to evaluate the correlations among the selected biomarkers. Biomarker levels were compared between Fontan patients and matched controls using Wilcoxon rank sum test. Moreover, biomarker levels were compared among Fontan patients with and without clinical events and control subjects using Kruskal-Wallis test, followed by post-hoc analysis using Dunn’s test. We set the false discovery rate (FDR) = 0.05 using the Benjamini–Hochberg procedure to control for type I error rate due to multiple testing¹². To identify biomarkers associated with an increase in MELD-XI score, we first used a simple linear regression of the change in MELD-XI score (baseline vs. 1 year) on each biomarker as a screening step to reduce the number of candidate biomarkers. Those biomarkers with P<0.15 in the screening step were entered into a subsequent multiple linear regression model with the change in MELD-XI score as the dependent variable. A p-value of <0.05 was considered to be significant for all analyses. To evaluate the statistical difference between baseline and 1 year follow up, values for standard laboratory markers as well as calculated scores such as the MELD XI score paired two-tailed t tests were used. Statistical analyses were conducted using SAS v.9.4 (Cary, NC) and graphs were prepared using Graph Pad Prism v. 8.0.1 (Graph Pad Software, San Diego, CA)

RESULTS

Distribution of Clinical Profiles and Characteristics

A total of 44 patients (30 female and 14 male) with Fontan operation and 32 age, gender and race matched controls were considered in the analysis. The average age of our Fontan population was 29.2 ± 10.7 years. Characteristics of these patients are reported in Table 1. A majority of our patients had a systemic left ventricle (n=28, 63.6%). In addition, 38.6% (n=17) had atriopulmonary (AP) connection, 47.7% (n=21) had a lateral tunnel (LT) connection and 13.6% (n=6) had an extra-cardiac conduit (ECC). Among the patients, the incidence of previous Fontan conversion was 13.6% and mean Fontan duration was 23.8 ± 5.1 years.

Table 1:

Characteristics of Fontan patients and controls within our EACH database that had biomarker profiles assessed. (^* Diuretic-Torsemide/Lasix, ^** Antiarrhythmic: Sotalol, Dronedarone, Propafenone, Amiodarone, ^*** Arterial vasodilators: Tadalafil, Sildenafil)

Characteristics	Fontan Patients
	N=44	%
Female	30	68.2%
Mean Age (yr)	29.2 ± 10.7	-
Mean BMI (kg/m2)	25.8 ± 4.1	-
Mean Fontan Years	23.8 ± 5.1	-
Age at Fontan (yr)	7.5 ± 6.3	-
Mean Oxygen Saturation	90.1 ± 4.9	-
Race
White	35	79.5%
Black/African American	8	18.2%
Asian	1	2.3%
Medications*
Aspirin	23	52.3%
Anti-platelet therapy (not aspirin)	1	2.3%
Warfarin	21	47.7%
DOAC	2	4.6%
Beta-blockers	20	45.5%
ACE/ARB	20	45.5%
Diuretic**	17	38.6%
Aldactone antagonist	12	27.3%
Antiarrhythmics***	8	18.2%
Pulmonary Vasodilators****	6	13.6%
History of Arrythmias
Supraventricular tachycardia	16	36.4%
Brady-arrhythmias including heart block	12	27.3%
Ventricular arrhythmias	2	4.6%
No history of arrhythmias	16	36.4%
Devices
Pacemaker	14	31.8%
ICD	1	2.3%
History of Hospitalizations
Previous hospitalization(s) lasting >48 hrs	9	20.5%
Anatomic Diagnosis
Double Outlet Right Ventricle	7	15.9%
Double Inlet Left Ventricle	10	22.7%
Hypoplastic Left Heart	6	13.6%
Pulmonary Atresia Intact Ventricular Septum	2	4.5%
Tricuspid Atresia	12	27.3%
Other	3	6.8%
Heterotaxy
Unbalanced AV Canal	2	4.6%
Double Outlet Right Ventricle	1	2.3%
Other	1	2.3%
Systemic Ventricle
Right	16	36.4%
Left	28	63.6%
Fontan Type
Atrio-pulmonary	17	38.6%
Lateral Tunnel#	21	47.7%
Extra-cardiac##	6	13.6%
Fontan Revisions
Fontan Revisions	6	13.6%
APC to LT	1	2.3%
APC to EC	5	11.4%
NYHA Class
I	10	22.7%
II	27	61.4%
III	7	15.9%
IV	0	0%

(Abbreviations used: DOAC: Direct oral Anticoagulant, ACE=Angiotensin converting enzyme, ARB= Angiotensin receptor blocker, ICD= Implantable Cardioverter Defibrillator

^*Listed medications are not mutually exclusive; patient can be on multiple medications

^**Diuretics include Furosemide, Torsemide and Bumetanide

^***Antiarrhythmic include Sotalol, Dronedarone, Propafenone, Amiodarone,

^***Pulmonary vasodilators include Tadalafil and Sildenafil

^#5 Lateral tunnel Fontans had fenestrations

^##4 Extra-cardiac Fontans had fenestrations

Clinical Events and Diagnostic Tests at Baseline and 1 Year:

After enrolling Fontan patients in the study between 2014–2016, we monitored clinical events up to 1 year after enrollment. Echocardiography and laboratory data were monitored at average time periods of 393.7±127.7 days and 428.4 ± 97.0 days respectively. Average time to clinical event was 155 ±122 days. Within the enrolled population, 50.0% (n=22) patients had clinical events as noted in Table 2a. Within the Fontan patient group, 18.2% (n=8) had hospitalizations. Cardiac and non-cardiac causes of hospital admissions are noted in Supplemental Table 3 and have n=1 each. 29.5% (n=13) had an increase in MELDXI score by at least 1 point, and 20.5% had arrhythmias (n=9) (Supplemental Table 4). 66.0% of arrhythmias (n=6) required cardioversion and 1 patient (2.3%) required ablation. 50% of Fontan patients had no events by these parameters. However, amongst those with events, 45.5% (n=10) had 1 event, 31.8% (n=7) had 2 events, 22.7% (n=5) had 3 or more events within the 1 year follow up period. Additionally, we monitored cardiac function via echocardiography and standard serum laboratory tests at baseline and at 1 year, and noted that there were no clinical differences at these two time points (Table 2b), although AST showed statistical difference. Similarly, we measured MELD XI scores in Fontan patients and noted that 13 patients had a significant increase in their MELD XI scores (Table 2c), in which case there was a significant increase in both creatinine and bilirubin. However, in those patients where MELD-XI score decreased or remained unchanged, there was only a statistical change in bilirubin levels. Change in MELD scores were not associated with hospitalizations or change in diuretics.

Table 2: Baseline and 1-year data on Fontan patients after enrollment.

2a: Clinical events observed in stable Fontan patients within 1 year of serum extraction for biomarkers analysis.

2b: Laboratory and echocardiography profile of stable SV Fontan patients at baseline and 1 year follow up after drawing serum for biomarker levels.

2c: Baseline and 1-year MELD-XI scores in Fontan patients. Corresponding creatinine and bilirubin levels in patients with an increase and stable/unchanged MELD-XI scores are also noted.

Average Time for Clinic Event	155 ± 122 days
Clinical Events	n	% (out of 44)
Hospitalizations	8	18.2%
Cardiac Causes	3	6.8%
Non-cardiac causes	5	11.4%
Increase in MELDXI* score by ≥ 1	13	29.5%
Diagnosis of Hepatocellular Carcinoma	2	4.5%
Arrhythmias requiring treatment	9	20.5%
Defibrillation	0	0.0%
Cardioversion	6	13.6%
Ablation	1	2.3%
Event rate#
No Events	22	50.0%
Any Events	22	50.0%
1 2 >2	10 7 5	45.5% 31.8% 22.7%

Echocardiography Parameters
Parameter	# Patients	Change in # Patients
Systemic Ventricular Function, qualitative (n=35)
Normal	24 (68.6%)	+2
Mildly Depressed	3 (8.6%)	+1
Moderate or severely depressed	8 (22.9%)	0
Systemic Valve Regurgitation (n=35)
None-Trace	7 (20.0%)	−1
Mild	15 (42.9%)	+2
Moderate	12 (34.3%)	0
Severe	1 (2.9%)	+1
Serum Chemistry Laboratory Values
Parameter	Average at 1 yr (%)	Change from Baseline	P value
Creatinine	0.95±0.66	+0.014±0.14	0.50
Total Bilirubin	1.1±0.6	0.0±0.5	0.76
Aspartate Aminotransferase (AST)	24.6±6.5	−3.5±8.4	0.03
Alanine Aminotransferase (ALT)	23.3±11.2	−4.3±11.9	0.27
Alkaline Phosphatase	74.2±23.0	4.5±13.9	0.31
Albumin	4.5±0.6	0.1±0.3	0.27
Platelets	174.02±65.05	−8.0±49.51	0.76

	n	Baseline MELD XI	MELD XI at 1 yr***	Change in Bilirubin	Change in Creatinine
Patients with increase in MELD XI Score	13	8.5 ± 1.2	10.0 ± 1.4*	0.48 ± 0.36#	0.10 ± 0.10$
Patients with decrease or unchanged MELD XI Score	31	9.8 ± 2.1	9.1 ± 2.1**	−0.24 ± 0.38##	−0.03 ± 0.14$ $

^*MELD score is calculated using the equation MELD-XI = 5.11 × ln (serum bilirubin in mg/dL) ± 11.76 × ln (serum creatinine in mg/dL) ± 9.44.

^#Events include cardiac hospitalization, non-cardiac hospitalization, presence off arrhythmia requiring intervention, increase in MELDXI score of ≥1, diagnosis of hepatocellular carcinoma.

^*p=0.0002,

^**p=0.003,

^#p=0.03,

^##p=0.04,

^$p=0.0006,

^{$ $}p=0.21,

^***MELD XI was remeasured at 428.4 ± 97.0 days.

Biomarker comparison between Fontan patients and Controls

We evaluated multiple biomarkers in Fontan patients as well as controls (Table 3). In comparison to age, gender and race matched controls, Fontan patients had significantly elevated levels of inflammatory markers IL-6 (p=0.011), GDF-15 (p<0.0001) and β2-microgloblulin (p<0.0006); hepatic extra-cellular matrix turnover marker, Collagen IV (p<0.0001); markers of cardiac injury, cTNI (p<0.0004), cell mobilization, SDF-1α (p<0.006), neurohormonal activation, renin (p<0.0001), renal dysfunction, uPAR (p=0.022) and cystatin (p<0.001) and fatty acid and glucose metabolism, adiponectin (p=0,0037). In a smaller population of 23 Fontan patients whose plasma was analyzed similarly, TNF-α was also upregulated, (p<0.001). The inter and intra assay coefficients of variation (CV%) are shown in Supplemental Table S1. Markers of thrombosis including vWF-A2 and platelets were statistically insignificant as compared to control patients. Use of angiotensin converting enzyme inhibitors (ACEi) or Angiotensin Receptor Blocker (ARB) did not correlate with renin levels in our Fontan patients. Furthermore, biomarkers showed a cluster pattern of correlation amongst each other as shown in Supplemental Figure 1.

Table 3:

Dysregulated Biomarkers in “stable” SV Fontan patients as compared to age, gender and race matched controls. Biomarkers were significantly different in various organ systems as noted above.

	Case (n=44)		Control (n=32)		Fold Changes	P value Wilcoxon ranked sum
	Median (pg/mL)	IQR (25th-75th percentile)	Median (pg/mL)	IQR (25th-75th percentile)
Inflammatory
IL-6	1.36	0.86,2.1	0.81	0.61, 1.41	1.69	0.011
TNF-α*	2.93	1.76, 2.85	0.57	0.17, 0.97	1.66	<0.001
GDF-15	908.47	675.71, 1383.68	545.77	450.885, 688.540	1.66	<0.0001
β2-microglobulin	2.41	1.551, 3.406	1.419	1.086, 1.702	1.70	0.0006
Cell mobilization
SDF1α	178.45	0.01, 556.925	0.25	0.01,253.82	713.78	0.006
Renal Function
uPAR	258.105	201.77, 308.47	203.075	120.11, 312.61	11.17	0.022
Cystatin c	438261	404053, 481752	370323	322646, 426023	1.18	<0.0001
Neurohormonal Activation
Renin	286.94	178.39, 721.42	148.5	88.36, 195.38	1.93	<0.0001
Cardiac Injury
Cardiac Troponin-I	15.375	7.99, 24.4	3.6075	0.01, 16.47	4.26	0.0004
Extracellular Matrix Turnover
Collagen-IVa	1633.31	864.65, 4138.67	497.53	362.555,805.335	3.34	<0.0001
Metabolism Regulator
Adiponectin	15.583	8.09, 39.013	8.831	5.676, 14.327	1.76	0.0037

^*TNF-α was tested in a random subset of 23 patients.

GDF-15 is associated with Clinical Events, Fontan Duration and APC Fontan:

GDF-15 correlated with multiple clinical markers in our Fontan patients (Figure 1). The average time from Fontan operation was 23.8 ± 5.1 years. After excluding patients who underwent Fontan revision, GDF-15 was found to be elevated in patients with AP Fontan (Figure 1a) and showed a moderately positive correlation (r=0.55) in AP Fontan patients with Fontan duration (Figure 1b), but not with LT or ECC Fontan patients. Both event free and Fontan patients with clinical events had GDF-15 levels that were significantly different from the levels in the control cohort (p<0.05 and p<0.0001 respectively) (Figure 1c).

Figure 1:

Correlation between clinical parameters and GDF-15. (A) GDF-15 was significantly elevated in Patients with APC Fontan as compared to LT or ECC Fontan. (B) Duration of APC Fontan moderately correlated positively with GDF-15 levels. (C) Biomarker levels were compared between control cohort, Fontan patients without events and Fontan patients with events. In Fontan patients with events, and without events, GDF-15 levels were increased as compared to control. *p≤0.05, ****p≤0.0001

MELDXI score and biomarker levels:

MELDXI scores were calculated at baseline and at 1 year follow up. An increase in MELDXI of at least 1 point was considered to be clinically significant. Serum VEGF-D (β coefficient −0.13; 95% CI −0.25 to −0.002; p<0.05) and Collagen IV (B coefficient −0.36; 95% CI −0.72 to −0.007; p<0.05) levels were found to be statistically associated with a change in MELDXI, controlling for serum vWF-A2, GDF-15, renin, and cystatin C levels, as shown in Supplemental Table 2.

Discussion

This study demonstrates chronic systemic biomolecular dysregulation in Fontan patients who are considered to be stable by clinical assessment. Adults with the Fontan operation may maintain his or her steady state of health until the circulation is no longer compensated; the “failing Fontan”, recognized by clinicians as a heterogeneous clinical syndrome of exercise intolerance and progressive decompensation eventually leading to transplant and/or death. However, Fontan failure is a progressive systemic decompensation and not a watershed event. It is important to recognize that even within a “stable” Fontan, a univentricular heart in combination with non-pulsatile pulmonary flow has chronic systemic consequences that are not reflected through standard laboratory and diagnostic tests, as shown in this study (Table 2(b)).

Given the lack of insight provided by standard diagnostic tests in predicting Fontan failure, there has been great interest in identifying biomarkers that can reflect physiological changes at baseline as well as portend clinical worsening in Fontan patients. Recent studies have evaluated biomarkers such as hs-CRP¹³ and galectin-3¹⁴ in isolation in Fontan patients and associated their dysregulation with poor outcomes. While these correlations provide valuable information, they do not provide insight into the physiological causes of dysregulation. With that in mind we evaluated a panel of biomarkers associated with various organ systems to understand the physiology of Fontan dysregulation. Given our relatively small population of Fontan patients, our goal was to evaluate baseline differences in biomarkers in stable Fontan patients as compared to control patients with normal cardiac physiology. To that end, we examined serum biomarkers using custom made multiplex and singleplex assays using the Luminex platform which has been validated and used extensively for biomarker quantification by other investigators^{15, 16}. We evaluated multiple biomarkers associated with inflammation, angiogenesis, renal function, thrombosis, neurohormonal activation, and glucose and fatty acid metabolism. Our analysis showed that multiple pro-inflammatory markers are upregulated as indicated in Table 3. Previous studies in adults with acquired heart failure have shown that various inflammatory cytokines are similarly upregulated, including, more commonly, TNF-α ^17–19 and IL-6 ^20–22 and less frequently IL-2, IL-6 and IFN-γ^{5, 23}. Furthermore, inflammatory cytokines continue to increase with clinical progression of heart failure^{22, 24}. It was novel and intriguing that some of these biomarkers were upregulated in patients that were deemed clinically stable, hence indicating that Fontan patients may have chronic subclinical inflammation throughout their life. Hence, it will be of clinical interest to evaluate these biomarkers longitudinally in Fontan patients and correlate their levels with clinical outcomes. Similarly, cystatin C^{25, 26} and uPAR²⁷, markers associated with renal dysfunction are elevated in our Fontan patients as compared to controls, whereas creatinine remains in the normal range.

Within our stable Fontan cohort, 50% had clinical events with 1 year of enrollment and blood draw (Table 2a). Within the Fontan population in general, incidence of arrhythmia has been shown to be higher in patients with APC Fontan as compared to LT and ECC Fontan^{1, 2, 28–34}. GDF-15 has been associated with atrial fibrosis and arrythmias³⁵ and was the only biomarker that was upregulated in our panel in the APC Fontan population as compared to LT and ECC Fontan (Figure 1). There is also a moderate positive correlation of GDF-15 with increasing duration of APC Fontan, which was not observed in LT and ECC Fontan. Given these findings, our study suggests that APC Fontans and those with clinical events have a chronic elevation of GDF-15 that plays a significant role over the lifespan of our Fontan patients. GDF-15 is a stress responsive cytokine from the TGF-β family and is expressed by cardiomyocytes and endothelial cells within the heart as well as inflammatory cells such as macrophages³⁶. GDF-15 has been associated with acquired cardiovascular diseases such as atherosclerosis, endothelial dysfunction, chronic kidney disease and cardiac hypertrophy. In addition, serum GDF-15 has been correlated with increasing fibrosis in acquired heart failure and found to decrease with cardiac assist devices in heart failure patients³⁷. While the specific physiologic consequence of GDF-15 upregulation in our patients is unknown, it can be hypothesized that progressive fibrosis within organs of Fontan patients may be contributing to their gradual physiological decline. The hypothesis is also supported by data from Opotowsky et al. which described galectin-3, a marker of fibrosis, as also upregulated in Fontan patients with poor outcomes¹⁴.

Fontan associated liver disease (FALD), is a common long term sequalae associated with Fontan physiology and has been reported at an annual rate of 1.5–5.0% of patients with cirrhosis. Fontan patients undergo a life-long predisposition to hepatic injury that starts immediately after the Fontan operation and has been attributed to venous congestion from the high pressure, non-pulsatile Fontan flow into the pulmonary vasculature³⁸. Hence, we investigated biomarkers that would signal worsening hepatic function in our Fontan patients as a function of MELDXI score, which has previously been shown to correlate with post-Fontan hepatic fibrosis^38–40 (Supplemental Table 2). In a multivariate regression model, we found that VEGF-D, a marker of angiogenesis and lymphangiogenesis^41–43 and Collagen IV, a marker of hepatic extracellular matrix turnover known be upregulated in chronic diseases associated with hepatic fibrosis^{44, 45} correlates with an increase MELD XI score. Studies have shown that VEGF levels progressively increase through successive steps of low-grade dysplasia, high grade dysplasia and early stage HCC⁴⁶. Studies have also shown that VEGF-D is involved in growth and spread of hepatocellular carcinoma⁴⁷, although only 2 of the patients in our cohort were diagnosed with HCC during the year of evaluation. Furthermore, Collagen IV has similarly been found to be significantly elevated in patients with HCC with and without cirrhosis as compared to controls⁴⁸. Given the long-term insult to the liver by the Fontan physiology from hypoxia and passive congestion, and the high incidence of liver cirrhosis, liver nodules and HCC, it would be interesting to study these biomarkers longitudinally as early markers of HCC.

Limitations:

Our study is a single center study with a small number of patients as compared to studies involving adults with acquired heart disease. Given the relatively low prevalence of Fontan patients in the general population, the small study size cannot be overcome without multi-center collaborative studies. However, initial single center studies are important in generating hypothesis and establishing methodologies. Given this is a single center study, there are significant population biases associated with geography and with accessing care at a tertiary care center. While the population of Fontan patients in the study were defined as “stable”, a significant portion had previous comorbidities, which reflects the general clinical course of these patients. It would be beneficial to evaluate biomarkers in adolescents without these comorbidities. Two patients developed HCC during the period of the study and were diagnosed based on characteristic MRI findings, with metastatic disease confirmed by biopsy in 1 patient. While current guidelines do not recommend histological confirmation of HCC in the presence of characteristic imaging findings^{10, 49}, Wells et al⁵⁰ have shown that imaging may result in false positive diagnosis in Fontan patients, which confirms the importance of portal phase washout on imaging, with review by hepatologists and imaging experts experienced in HCC diagnosis. Stable Fontan patients excluded those with end-outcomes of Fontan revision, death or transplant. Hence, surrogate end-points such as clinical events as well as increase in MELD-XI score were used, only the latter of which have been associated with death and transplant in the Fontan population³⁹. Furthermore, biomarkers were only measured at one time during the study period.

Conclusion and Relevance:

These data establish a baseline profile of systemic dysregulation in our stable Fontan patients that has previously not been described. Given this insight, additional longitudinal studies with larger biomarker groups are necessary to understand pathological changes that lead to Fontan failure.

Highlights:

• Fontan patients are the most complex within a fast-growing population of adults with CHD.

• Traditional diagnostic tests are not predictive of short-term outcomes in Fontan patients.

• Our study shows that multiple systemic biomarkers are dysregulated in stable Fontan patients.

• GDF-15 correlates with increased clinical events within 1 year in stable Fontan patients.

• VEGF-D and Collagen IV levels are elevated in Fontan patients with worsening hepatic dysfunction.

Glossary

AS

conceptualization, data curation, funding acquisition, investigation, methodology, project administration, supervision, writing of all versions of the draft

CD

data acquisition

IE

clinical data acquisition

AH

clinical data acquisition

YK

data validation, formal analysis

SJ

patient recruitment

JHK

control subject samples, reviewing and editing of manuscript

FHR

conceptualization, resources, manuscript review and editing

APK

conceptualization, methodology, manuscript review and editing

AQ

conceptualization, methodology, resources

WB

conceptualization, investigation, funding acquisition, methodology, project administration, resources, supervision, reviewing and editing of manuscript.