Associations Between Characteristics of Individuals With Fontan Circulation With Blood and Urine Biomarkers of Kidney Injury and Dysfunction

Abstract

Background

Fontan circulation is associated with kidney injury and dysfunction, often unappreciated until Fontan circulatory failure. We hypothesized that cystatin C‐estimated glomerular filtration rate (eGFR) would identify chronic kidney disease more frequently and that urine kidney injury biomarkers would be higher with declining Fontan physiological features.

Methods and Results

We enrolled 100 ambulatory individuals. Blood and urinary laboratory measurements were compared with demographics and clinically obtained data. Different eGFR equations were used for individuals aged ≥19 years and <19 years. Chronic kidney disease was defined as eGFR <90 mL/min per 1.73 m². Median (25th–75th percentile) age was 19 (14–26) years, and 43% were female patients. Cystatin C eGFR detected chronic kidney disease (37%) in more patients than creatinine eGFR (11%). Cystatin C eGFR was positively associated, and skeletal muscle mass was negatively associated, with creatinine eGFR in both univariate (cystatin C eGFR β=0.44±0.12, P=0.0006; skeletal muscle mass β=−0.72±0.32, P=0.03) and multivariable analysis (cystatin C eGFR β=0.43±0.12, P=0.0005; skeletal muscle mass β=−0.69±0.29, P=0.02). Urine neutrophil gelatinase‐associated lipocalin concentration correlated with Fontan pressure (r=0.28; P=0.04), ventricular end‐diastolic pressure (r=0.28; P=0.04), and body fat mass (r=0.26; P=0.03).

Conclusions

Cystatin C eGFR identified more kidney dysfunction, likely attributable to creatinine eGFR being confounded by skeletal muscle mass. Elevated urine neutrophil gelatinase‐associated lipocalin was associated with worse Fontan hemodynamics and higher percentage body fat, suggesting that higher venous pressure and higher adiposity are associated with ongoing kidney injury.

Nonstandard Abbreviations and Acronyms

BIA

bioimpedance analysis

KIM‐1

kidney injury molecule‐1

NAG

N‐acetyl glucosaminidase

NGAL

neutrophil gelatinase‐associated lipocalin

Q_P:Q_S

ratio of pulmonary blood flow/systemic blood flow

SMM

skeletal muscle mass

Clinical Perspective.

What Is New?

• Our article shows that cystatin C–estimated glomerular filtration rate identifies chronic kidney disease in patients with a Fontan circulation better than creatinine estimated glomerular filtration rate secondary to the reliance of creatinine–estimated glomerular filtration rate on skeletal muscle mass (demonstrated with bioelectrical impedance analysis showing less skeletal muscle mass in patients with a Fontan circulation).

• Our article also shows that elevated urine neutrophil gelatinase‐associated lipocalin was associated with worse Fontan hemodynamics and higher percentage body fat, suggesting that higher venous pressure, lower cardiac index, and higher adiposity are associated with ongoing kidney injury.

What Are the Clinical Implications?

• Patients with a Fontan circulation should have the kidney function monitored with serial cystatin C measurements, as opposed to creatinine, as it may identify kidney dysfunction earlier (creatinine–estimated glomerular filtration rate is falsely elevated secondary to their low skeletal muscle mass).

• Urine neutrophil gelatinase‐associated lipocalin may be a biomarker that could be added to surveillance testing, as its elevation could signify that the patient's Fontan hemodynamics are elevated.

• Patients with a Fontan circulation should do their best to avoid excess adiposity, as it has an association with kidney injury (elevated urine neutrophil gelatinase‐associated lipocalin).

Individuals with single‐ventricle congenital heart disease require interventions in childhood that culminate with a Fontan procedure, which is an anastomosis between the inferior vena cava to the pulmonary arteries, usually with an extracardiac conduit. Combined with a bidirectional Glenn procedure, in which the superior vena cava is anastomosed to the pulmonary arteries, these individuals have passive systemic venous return. Although having a Fontan circulation allows these individuals to survive into adulthood without a subpulmonary ventricle, they often develop multiorgan sequalae, including kidney disease. ¹ For a variety of reasons, Fontan‐associated kidney disease is often identified late in the clinical course, after the development of Fontan circulatory failure. ² One reason may be that kidney function is most often estimated by measurement of the blood filtration marker creatinine, and this may overestimate the glomerular filtration rate (GFR) in this population with known relative sarcopenia. Cystatin C, another available filtration marker, is not dependent on muscle mass. In individuals with Fontan circulation, cystatin C–estimated GFR (eGFR) is commonly reduced and lower cystatin C eGFR is more strongly associated with increased risk for adverse outcomes, as compared with creatinine eGFR. ³

Circulating biomarkers have shown promise as a less invasive way to monitor the status of the Fontan circulation, in particular lower cardiac index. ⁴ Urine biomarkers of kidney injury have been shown to identify subclinical kidney injury in those with a reduced ejection fraction and increased risk of adverse outcomes in patients with a Fontan circulation. ³ , ⁵ The objective of this study was to assess whether urine biomarkers of kidney injury and kidney function markers correlate with clinical assessments of Fontan circulatory physiological features. Defining this relationship may lead to opportunities for early identification of kidney dysfunction and, thus, early interventions to preserve both the Fontan circulation of individuals and their kidneys. We hypothesized that cystatin C eGFR, compared with creatinine eGFR, would better reflect kidney function in individuals with Fontan circulation, and that urine markers of kidney injury would be elevated in individuals with declining Fontan circulatory physiological features. Furthermore, we hypothesized that any systematic difference in eGFR assessment by creatinine and cystatin C would be partly explained by measured skeletal muscle mass.

Methods

The data analyzed for this study are available to other researchers on request from the corresponding author. We enrolled 100 outpatients with a Fontan circulation between 2020 and 2021 seen at Cincinnati Children's Hospital Medical Center, who were at least 1 year after completion of the Fontan procedure, or Fontan revision. The current study was approved by the Cincinnati Children's Hospital Medical Center Institutional Review Board. Written informed consent was obtained from each participant aged ≥18 years or their parental caregiver if aged <18 years. Written assent was obtained from participants aged 11 to 18 years.

Demographic and clinical data, obtained within 2 years of biospecimen collection, were obtained via a detailed review of all available clinical notes and other data, as well as Cincinnati Children's Hospital Medical Center's Fontan Clinical Database. Blood and urine for this research were collected around the time of enrollment, typically on the same day.

GFR was estimated using equations appropriate for adults (aged ≥19 years) and children (aged <19 years). For adults, GFR was estimated using the Chronic Kidney Disease–Epidemiology Collaboration (CKD‐EPI) equations using creatinine:

A, B, and C are constants based on sex, and S_Cr. ⁶ Using cystatin C:

$$\begin{array}{c}\begin{array}{c}\mathrm{CKD‐}{\mathrm{EPI}}_{\text{CysC}}=133\mathrm{x}{\left({\mathrm{S}}_{\text{CysC}}/0.8\right)}^A\mathrm{x}{0.996}^{\mathrm{age}}\mathrm{x}\mathrm{B}\hfill \end{array}\hfill \end{array}$$

A and B are constants based on sex and S_CysC. ⁷ For children, creatinine‐based eGFR was calculated using the bedside Schwartz equation ⁸ :

$$\text{eGFR}=0.413\mathrm{x}\left(\text{height in}\mathrm{cm}/{\mathrm{S}}_{\mathrm{Cr}}\right);$$

Cystatin C–based eGFR was calculated using the Larsson equation ⁹ :

$$\text{eGFR}=77.24\mathrm{x}{{\mathrm{S}}_{\text{CysC}}}^{-1.2623}.$$

Demographic variables analyzed included the following: age at the time of biospecimen collection, sex, race, cardiac diagnosis, date of Fontan operation, type of Fontan pathway, presence of fenestration at the time of most recent Fontan procedure, the morphologic features of the dominant ventricle, saturation at the time of research laboratory tests, time since Fontan procedure, and history of Fontan revision. Clinical history included New York Heart Association functional class and any history of protein‐losing enteropathy, pacemaker, plastic bronchitis, thrombosis, catheter intervention, and arrhythmia. Laboratory data included both clinical and research assays: alkaline phosphatase, total bilirubin, albumin, aspartate aminotransferase, alanine aminotransferase, creatinine, cystatin C, platelet count, B‐type natriuretic peptide, NT‐proBNP (N‐terminal pro‐B‐type natriuretic peptide), urine kidney injury molecule‐1 (KIM‐1), urine N‐acetyl glucosaminidase (NAG), urine neutrophil gelatinase‐associated lipocalin (NGAL), urine creatinine, and albumin/creatinine ratio. Cardiac imaging assessments included either echocardiogram or cardiac magnetic resonance study, as previously described. ¹⁰ Data from catheterizations, cardiopulmonary exercise tests, and liver magnetic resonance elastography were also extracted from the medical record. A subset of individuals (n=75) underwent bioimpedance analysis (BIA) of body composition (InBody570; InBody USA, Cerritos, CA) as part of clinical care. We excluded clinical data obtained >2 years before the collection of the research biospecimen.

Urine was collected via clean catch voids and stored at −80 °C. For analysis, urine aliquots were thawed, vortexed, centrifuged at 2200g for 20 minutes at 20 °C, and then pipetted for assay, per manufacturer instructions. NAG and NGAL were measured via particle‐enhanced turbidimetric immunoassays on a Roche Cobas c 311 clinical chemistry analyzer. NAG reagent was supplied by Diazyme Laboratories; NGAL reagent was supplied by BioPorto Diagnostics. KIM‐1 was measured via Quantikine Human Urinary ELISAs (R&D Systems).

Statistical Analysis

Continuous variables are presented as mean±SD for normally distributed variables and as median (25th–75th percentile) for nonnormally distributed variables. We defined CKD as eGFR <90 mL/min per 1.73 m², as it represents at least stage 2 CKD per the Kidney Disease: Improving Global Outcomes definition. ¹¹ The cutoff for elevated urine NGAL being 50 ng/mL was based on the lowest detectable limit for urinary NGAL Test (BioPorto Diagnostics, Inc), which is commonly used for identifying low risk of acute kidney injury. Existing literature was used to determine the cutoff for elevated NGAL (>50 ng/mL), KIM‐1 (>1300 pg/mL), and NAG (>18 U/L). ¹² , ¹³ , ¹⁴ The McNemar test was used to compare the percentages of CKD defined by creatinine eGFR and cystatin eGFR. The Student t‐test was used to compare normally distributed continuous variables between 2 groups; for nonnormally distributed continuous variables, the Wilcoxon rank sum test was used. The Fisher exact test was used to compare categorical variables between groups. Univariate associations between nonnormally distributed continuous variables were determined using the Spearman correlation coefficient. Multivariable linear regression analysis was used to evaluate associations between variables after adjusting for skeletal muscle mass (SMM). A 2‐tailed P≤0.05 was considered statistically significant. Statistical analyses were performed using JMP version 14 (SAS Institute Inc, Cary, NC) and SAS version 9.4 (SAS Institute Inc).

Results

Demographics and Clinical Characteristics

A total of 100 individuals with Fontan circulation were enrolled. The median age at the time of laboratory collection was 19 (14–26) years (51% were aged <19 years), 43% were female patients, and 9% were Black race (with most of the patients being White race). The median time since the most recent Fontan procedure was 15 (10–22) years. The most common Fontan type and dominant ventricular morphologic feature were extracardiac conduit Fontan circulation (60%) and left ventricle (56%), respectively. A fenestration was created at the time of the Fontan procedure in most patients (77%). Table 1 depicts the demographics and clinical characteristics of the individuals in our cohort. There were no associations between clinical characteristics (New York Heart Association functional class, protein‐losing enteropathy, plastic bronchitis, pacemaker, prior clinical thrombosis, catheter intervention, and arrhythmia) and those with CKD determined by cystatin C eGFR versus creatinine eGFR.

Table 1.

Demographic and Clinical Characteristics of the Overall Cohort, and for Subsets With CKD Defined by Either Creatinine or Cystatin C eGFR

Variable	All patients	Creatinine‐based CKD	Cystatin C–based CKD	P value
Total No.	100	11	37
Age at laboratory testing, y	19 (14–26)	18 (16–32)	17 (14–23)	0.24
Aged <19 y, % (n)	51 (51)	63.6 (7)	67.6 (25)	1.0
Time since Fontan, y	15 (10–22)	15 (11–28)	14 (10–19)	0.33
Sex, % (n) female	43 (43)	36 (4)	41 (15)	1.0
Race, % (n) Black	9 (9)	18 (2)	18 (3)	0.32
Cardiac diagnosis, n (%)				0.34
Tricuspid atresia	24 (26)	9 (1)	27.5 (11)
HLHS	29 (31)	27 (3)	27.5 (11)
Double‐inlet left ventricle	15 (16)	45 (5)	17.5 (7)
Pulmonary atresia with IVS	5 (5)	18 (2)	7.5 (3)
Double‐outlet right ventricle	11 (12)	0 (0)	7.5 (3)
Unbalanced AVSD	9 (10)	0 (0)	2.5 (1)
Other	7 (7)	0 (0)	10 (4)
Fontan type, % (n)				0.80
Atriopulmonary	1 (1)	0 (0)	3 (1)
Lateral tunnel	39 (39)	55 (6)	46 (17)
Extracardiac	60 (60)	45 (5)	51 (19)
Medical history, % (n)
Dominant, left ventricle	56 (56)	73 (8)	62 (23)	0.72
NYHA class ≥2	22 (22)	27 (3)	24 (9)	1.00
Protein‐losing enteropathy	6 (6)	9 (10)	8 (3)	1.00
Plastic bronchitis	1 (1)	0 (0)	3 (1)	1.00
Pacemaker	11 (11)	18 (2)	14 (5)	0.65
Thromboembolism	24 (24)	36 (4)	27 (10)	0.71
Catheter intervention	65 (65)	45 (5)	62 (23)	0.49
Arrhythmia	35 (35)	45 (5)	33 (12)	0.49

Continuous variables are presented as median (25th–75th percentile), and categorical variables are presented as percentage (number). CKD is defined as eGFR <90 mL/min per 1.73 m², by either creatinine or cystatin C. Multiple diagnoses were present in a subset of patients (2 = double‐outlet right ventricle and HLHS; 1 = tricuspid atresia and pulmonary atresia with IVS; 3 = double‐outlet right ventricle and unbalanced AVSD; 1 = double‐outlet right ventricle and other [criss‐cross heart]). P value represents P values of Wilcoxon rank sum test used for comparison of continuous variables (eg, age), and Fisher exact test was used for comparison of categorical variables (eg, age <19 years), between creatinine‐based CKD and cystatin C–based CKD. AVSD indicates atrioventricular septal defect; CKD, chronic kidney disease; eGFR, estimated glomerular filtration rate; HLHS, hypoplastic left heart syndrome; IVS, intact ventricular septum; and NYHA, New York Heart Association.

Estimated GFR

Median creatinine eGFR was 121 (106–131) mL/min per 1.73 m², whereas median cystatin C eGFR was 98 (86–114) mL/min per 1.73 m². There was a positive correlation between cystatin C eGFR and creatinine eGFR (Spearman ρ=0.42; P=0.0006), but a greater proportion of individuals were defined as having CKD using cystatin C eGFR compared with creatinine eGFR (37% versus 11%; Figure 1). Of the 37 individuals with CKD using cystatin C eGFR, 12 were adults and 25 were children, whereas of the 11 individuals with CKD using creatinine eGFR, 4 were adults and 7 were children. There were no individuals with either a creatinine eGFR or cystatin C eGFR, which was <60 mL/min per 1.73 m².

Figure 1. Cystatin C– vs creatinine–estimated glomerular filtration rate (eGFR).

Scatterplot of cystatin C eGFR vs creatinine eGFR for 100 patients with a Fontan circulation. The red line represents the best fit, or trend, line identified using linear regression. The blue dashed lines represent eGFR <90mL/min per 1.73 m², a commonly used clinical cutoff that indicates chronic kidney disease (CKD). There was a positive correlation between cystatin C eGFR and creatinine eGFR (Spearman ρ=0.42; P=0.0006). Of the 100 patients, 37 had cystatin C eGFR <90 mL/min per 1.73 m², whereas only 11 patients had a creatinine eGFR <90 mL/min per 1.73 m². There were 26 patients with CKD by cystatin C, who would be classified as having normal eGFR by creatinine (bottom right quadrant, highlighted in yellow).

BIA was performed in 75 of 100 individuals; the median time between BIA and biospecimen collection was 182 (11–366) days. Table 2 displays the relationship between creatinine eGFR, cystatin C eGFR, and SMM. As expected, there was a strong positive association between creatinine eGFR and cystatin C eGFR (β=0.44±0.12; P=0.0006). There was a negative association between SMM and creatinine eGFR (β=−0.72±0.32; P=0.03); conversely, there was no significant association between SMM and cystatin C eGFR (β=−0.06±0.29; P=0.84). In multivariable analysis, adjusting for cystatin C eGFR, there remained a negative association between SMM and creatinine eGFR (β=−0.69±0.29; P=0.02). Conversely, SMM provided no additional information about cystatin C eGFR after accounting for creatinine eGFR.

Table 2.

Relationship Between Creatinine eGFR, Cystatin C eGFR, and Measured SMM

Row	Dependent variable	Predictor variable	β±SE	P value	r 2
1	Creatinine eGFR	SMM	−0.72±0.32	0.03*	0.07
2	Cystatin C eGFR	SMM	−0.06±0.29	0.84	0.0005
3	Creatinine eGFR	Cystatin C eGFR	0.44±0.12	0.0006*	0.15
4	Creatinine eGFR	Cystatin C eGFR	0.43±0.12	0.0005*	0.21
		SMM	−0.69±0.29	0.02*
5	Cystatin C eGFR	Creatinine eGFR	0.36±0.10	0.0005*	0.16
		SMM	0.20±0.28	0.47

The first 3 rows represent univariate analyses between the dependent and predictor variables listed. The last 2 rows represent multivariable analyses between the dependent and predictor variables listed. Row 1 demonstrates the presence of an association between SMM and creatinine eGFR, whereas row 2 shows a lack of such an association between SMM and cystatin C eGFR. Row 3 shows the association between creatinine and cystatin C eGFR. Rows 4 and 5 show that adding SMM as a covariate provides additional information to predicting creatinine eGFR, whereas the same is not true for the converse (ie, predicting cystatin C eGFR). After adjusting for cystatin C eGFR, subjects with 10 kg more SMM will have 6.9 mL/min per 1.73 m² lower creatinine eGFR. P value represents P values for each β coefficient. This analysis includes the 75 patients who had bioimpedance body composition analysis performed as part of clinical care. β indicates β coefficient; eGFR, estimated glomerular filtration rate; r ², Pearson correlation coefficient for the full model; and SMM, skeletal muscle mass.

^* P<0.05.

Associations With Urine and Serum Biomarkers

There were no statistically significant associations between CKD (via either cystatin C or creatinine eGFR) and any laboratory tests, including urinary biomarkers. All of the urine kidney injury biomarkers correlated with one another (Table 3). There was a negative correlation between cystatin C eGFR and urine KIM‐1 (Spearman ρ=−0.20; P=0.04) and NAG (Spearman ρ=−0.28; P=0.004; Table 3). In contrast, no urine biomarker measured was correlated with creatinine eGFR. Alkaline phosphatase had a negative correlation with NGAL (r=−0.25; P=0.01), but otherwise, there were no other statistically significant associations between clinical laboratory tests and any of the urine markers of kidney injury (Table 3). There were no significant differences in urine biomarkers of kidney injury based on whether the individual was an adult or a child (Figure 2).

Table 3.

Associations of Urine Biomarkers With Clinical Laboratory Tests

Variable	KIM‐1, pg/mL	NAG, U/L	NGAL, ng/mL
Cystatin C eGFR, mL/min per 1.73 m2	r=−0.20	r=−0.28	r=−0.02
	P=0.04*	P=0.004*	P=0.82
Creatinine eGFR, mL/min per 1.73 m2	r=0.01	r=−0.15	r=−0.07
	P=0.89	P=0.14	P=0.48
Platelet count, ×109/L (n=99)	r=0.05	r=0.05	r=−0.13
	P=0.62	P=0.65	P=0.21
Alkaline phosphatase, U/L	r=0.15	r=0.02	r=−0.25
	P=0.15	P=0.87	P=0.01*
ALT, U/L	r=−0.03	r=0.04	r=−0.05
	P=0.76	P=0.73	P=0.61
AST, U/L	r=0.14	r=0.11	r=−0.09
	P=0.16	P=0.28	P=0.36
Total bilirubin, mg/dL	r=0.04	r=0.00	r=0.02
	P=0.70	P=0.96	P=0.86
Albumin, g/dL	r=−0.04	r=−0.12	r=−0.02
	P=0.66	P=0.24	P=0.83
BNP, pg/mL (n=78)	r=0.06	r=0.11	r=0.08
	P=0.59	P=0.36	P=0.48
NT‐proBNP, pg/mL (n=64)	r=0.01	r=0.13	r=0.03
	P=0.90	P=0.31	P=0.79
hs‐CRP, mg/dL (n=82)	r=0.02	r=0.13	r=0.24
	P=0.88	P=0.26	P=0.03
KIM‐1, pg/mL	…	r=0.54	r=0.40
		P<0.01*	P<0.01*
NAG, U/L	…	…	r=0.37
			P=0.0002*
ACR, mg/g	r=0.09	r=0.19	r=0.03
	P=0.38	P=0.06	P=0.78

Spearman correlation between urine biomarker concentrations and clinical laboratory measurements. N=100 unless otherwise noted.

ACR indicates albumin/creatinine ratio; ALT, alanine aminotransferase; AST, aspartate aminotransferase; BNP, B‐type natriuretic peptide; eGFR, estimated glomerular filtration rate; hs‐CRP, high‐sensitivity C‐reactive protein; KIM‐1, kidney injury molecule‐1; NAG, N‐acetyl glucosaminidase; NGAL, neutrophil gelatinase‐associated lipocalin; NT‐proBNP, N‐terminal pro‐B‐type natriuretic peptide; and r, Spearman correlation coefficient.

^* P<0.05.

Figure 2. Comparison of urine biomarkers of kidney injury between age groups among 100 individuals with a Fontan circulation.

Median values of kidney injury molecule‐1 (KIM‐1; pg/mL), N‐acetyl glucosaminidase (NAG; U/L), and neutrophil gelatinase‐associated lipocalin (NGAL; ng/mL) are presented, along with error bars representing the 25th and 75th percentiles. Comparisons are between those aged <19 years (<19; children) and those aged ≥19 years (≥19; adults) for each urine biomarker of kidney injury using the Wilcoxon rank sum test (P values for each comparison are presented above each set of columns).

Associations Between Biomarkers and Catheterization Data

Catheterization data were available for 55 individuals (30 female patients and 25 male patients), and the median time between baseline laboratory measurements and catheterization was 241 (64–478) days. There were no statistically significant associations between the presence of CKD, determined by either cystatin C or creatinine eGFR, and hemodynamic variables. In terms of urine markers of kidney injury, there were positive correlations between NAG and Fontan pressure (r=0.29; P=0.03), as well as pulmonary artery wedge pressure (r=0.29; P=0.03). NGAL also correlated with Fontan pressure (r=0.28; P=0.04) and ventricular end‐diastolic pressure (r=0.28; P=0.04) (Table 4). Those with elevated NGAL (>50 ng/mL) had higher Fontan pressure (13 [11–16] versus 19 [14–21] mm Hg; P=0.02), lower cardiac index (2.88 [2.48–3.34] versus 2.45 [2.33–2.62] L/min per m²; P=0.05), and higher ratio of pulmonary blood flow/systemic blood flow (Q_P:Q_S) (0.87 [0.80–1.00] versus 1.00 [1.00–1.01]; P<0.01) (Table 5). Receiver operating characteristic curves demonstrated that Fontan pressure (area under the curve=0.78 [0.70–0.86]), CI (area under the curve=0.75 [0.67–0.84]), and Q_P:Q_S (area under the curve=0.88 [0.81–0.94]) were all fair to good predictors of NGAL >50 ng/mL (Table 5).

Table 4.

Associations Between Urine Biomarkers and Hemodynamics Among Patients Who Underwent Clinical Cardiac Catheterization Within 2 Years of Urine Biomarker Assessment

Variable	KIM‐1, pg/mL	NAG, U/L	NGAL, ng/mL
Fontan pressure, mm Hg	r=0.24	r=0.29	r=0.28
	P=0.07	P=0.03*	P=0.04*
PAWP, mm Hg	r=0.21	r=0.29	r=0.24
	P=0.12	P=0.03*	P=0.08
Ventricular EDP, mm Hg	r=0.21	r=0.16	r=0.28
	P=0.12	P=0.25	P=0.04*
TPG, mm Hg	r=0.03	r=0.10	r=0.09
	P=0.82	P=0.47	P=0.90
QPi, L/min per m2	r=0.07	r=0.14	r=0.17
	P=0.59	P=0.30	P=0.22
QSi, L/min per m2	r=−0.12	r=0.06	r=−0.22
	P=0.39	P=0.69	P=0.10
QP:QS	r=0.19	r=0.06	r=0.39
	P=0.16	P=0.66	P=0.004*
PVRi, WU×m 2	r=−0.03	r=−0.04	r=−0.06
	P=0.82	P=0.76	P=0.68
Oxygen saturation, %	r=0.11	r=0.03	r=−0.08
	P=0.28	P=0.74	P=0.42

N=55 patients. Spearman correlation between urine biomarkers and hemodynamic data. Oxygen saturation is from the time of urine biomarker collection.

EDP indicates end‐diastolic pressure; KIM‐1, kidney injury molecule‐1; NAG, N‐acetyl glucosaminidase; NGAL, neutrophil gelatinase‐associated lipocalin; PAWP, pulmonary artery wedge pressure; PVRi, indexed pulmonary vascular resistance; Q_P:Q_S, ratio of pulmonary blood flow/systemic blood flow; Q_Pi, right–sided (pulmonary) cardiac index; Q_si, cardiac index; r, Spearman correlation coefficient; TPG, transpulmonary gradient; and WU, Wood units.

^* P<0.05.

Table 5.

Relationship Between Urine NGAL and Hemodynamics Among Patients Who Underwent Clinical Cardiac Catheterization Within 2 Years of Urine Biomarker Assessment

Variable	NGAL <50 ng/mL (n=49)	NGAL ≥50 ng/mL (n=6)	P value	AUC for NGAL ≥50 ng/mL
Fontan pressure, mm Hg	13 (11–16)	19 (14–21)	0.02*	0.78 (0.70–0.86)
PAWP, mm Hg	9 (8–11)	12 (8–15)	0.21	0.66 (0.57–0.75)
Ventricular EDP, mm Hg	10 (8–12)	10 (9–15)	0.56	0.57 (0.48–0.67)
TPG, mm Hg	3 (2–5)	4 (3–8)	0.14	0.69 (0.59–0.78)
QPi, L/min per m2	2.49 (2.18–2.86)	2.5 (2.33–2.62)	0.89	0.48 (0.39–0.58)
QSi, L/min per m2	2.88 (2.48–3.34)	2.45 (2.33–2.62)	0.046*	0.75 (0.67–0.84)
QP:QS,	0.87 (0.8–1)	1 (1–1.01)	0.002*	0.88 (0.81–0.94)
PVRi, WU×m2	1.29 (0.82–1.89)	1.25 (1.11–2.93)	0.50	0.59 (0.49–0.68)
Oxygen saturation, %	95 (93–97)	95 (93–97)	0.92	0.51 (0.41–0.61)

Wilcoxon rank sum test comparing the hemodynamic data of those with NGAL <50 ng/mL and those with NGAL ≥50 ng/mL. Oxygen saturation is from the time of urine biomarker collection. AUC for NGAL ≥50 ng/mL was calculated using receiver operating characteristic and is presented with Wald 95% CIs.

AUC indicates area under the curve; EDP, end‐diastolic pressure; NGAL, neutrophil gelatinase‐associated lipocalin; PAWP, pulmonary artery wedge pressure; PVRi, indexed pulmonary vascular resistance; Q_P:Q_S, ratio of pulmonary blood flow/systemic blood flow; Q_Pi, right‐sided (pulmonary) cardiac index; Q_si, cardiac index; TPG, transpulmonary gradient; and WU, Wood units.

^* P<0.05.

Associations Between Biomarkers and Cardiopulmonary Exercise Test and BIA Data

Seventy‐five individuals underwent both cardiopulmonary exercise tests and BIA (35 female patients and 40 male patients), with a median time from laboratory testing of 190 (14–367) and 182 (11–366) days, respectively. There were no associations between CKD, determined by either cystatin C or creatinine eGFR, and any BIA or cardiopulmonary exercise test variable analyzed. In Table 6, body fat mass and percentage body fat were associated with higher NGAL (r=0.26; P=0.03 and r=0.29; P=0.01, respectively).

Table 6.

Associations of Urine Biomarkers With Cardiopulmonary Exercise Testing and Bioimpedance Body Composition Data

Variable	KIM‐1, pg/mL	NAG, U/L	NGAL, ng/mL
Peak HR, % maximum for age	r=−0.12	r=−0.15	r=−0.22
	P=0.32	P=0.19	P=0.054
Peak VO2, % predicted	r=−0.01	r=−0.17	r=0.01
	P=0.92	P=0.15	P=0.95
Peak VO2, mL/min per kg	r=0.04	r=−0.11	r=−0.22
	P=0.76	P=0.37	P=0.06
BMI, kg/m2	r=−0.09	r=0.08	r=0.21
	P=0.45	P=0.48	P=0.07
Weight, kg	r=−0.09	r=−0.02	r=0.16
	P=0.44	P=0.88	P=0.18
Skeletal muscle mass, kg	r=−0.06	r=−0.11	r=−0.05
	P=0.60	P=0.37	P=0.7
Body fat mass, kg	r=−0.10	r=0.09	r=0.26
	P=0.40	P=0.46	P=0.03*
Body fat, %	r=−0.10	r=0.13	r=0.29
	P=0.41	P=0.25	P=0.01*

N=75 patients. Relationships between continuous variables using Spearman correlation between urine biomarkers and cardiopulmonary exercise testing data, as well as bioimpedance body composition data.

BMI indicates body mass index; HR, heart rate; KIM‐1, kidney injury molecule‐1; NAG, N‐acetyl glucosaminidase; NGAL, neutrophil gelatinase‐associated lipocalin; r, Spearman correlation coefficient; and VO₂, oxygen consumption.

^* P<0.05.

Associations Between Biomarkers and Cardiac and Liver Imaging

Echocardiographic or cardiac magnetic resonance study data were available for all individuals, with a median time from laboratory assessment of 0 (0–20) days. Cardiac imaging variables were not associated with any of the kidney function or kidney injury markers described above. Liver magnetic resonance elastography was performed in n=48, with a median time from laboratory testing of 70 (0–362) days. Average liver stiffness was 4.08±0.97 kPa. There were no associations between liver stiffness and any of the kidney function or kidney injury markers described above.

Discussion

Results of this prospectively enrolled cross‐sectional study suggest that in individuals with Fontan circulation (1) creatinine overestimates GFR and, consequently, is poorly sensitive for CKD; (2) this discrepancy is partially explained by the dependence of creatinine eGFR accuracy on SMM; (3) higher concentrations of urine kidney injury biomarkers are associated with unfavorable Fontan hemodynamics; and (4) also with higher total body adiposity (eg, body fat mass).

Creatinine has previously been shown to overestimate creatinine eGFR, relative to cystatin C eGFR, in children and adults with Fontan circulation. ¹⁵ , ¹⁶ , ¹⁷ In our study, the prevalence of CKD was higher when determined by cystatin C eGFR, compared with creatinine eGFR. The most likely reason for this is that creatinine is derived from muscle cells and overestimates GFR in individuals who have lower SMM than expected for age, sex, and body size. ¹⁸ Individuals with a Fontan circulation are known to have less SMM compared with the general population. ¹⁹ This logical argument is further buttressed by our findings in the subset of individuals with measured SMM, which suggests that cystatin C may be preferable to estimate GFR for individuals with Fontan circulation. Although there was a negative correlation between cystatin C eGFR and NAG, as well as KIM‐1, neither urine biomarker of kidney injury was associated with an eGFR <90 mL/min per 1.73 m² (determined by creatinine or cystatin C).

Renal perfusion pressure, the difference between mean arterial pressure and central venous pressure, is associated with decreased GFR when Fontan pressure (central venous pressure) is elevated. ²⁰ Ongoing kidney injury, however, is not necessarily captured by eGFR but may be better identified using urinary biomarkers of tubular injury. In our study, there were a few notable findings about the relationship between urine biomarkers of kidney injury and hemodynamic data. Elevated pressure in the Fontan circulation (Fontan conduit, pulmonary artery wedge pressure, or ventricular end‐diastolic pressure) was associated with elevated urinary NGAL. This was also the case for NAG, but only one individual had a clinically elevated NAG (>18 U/L); so, despite the associations between NAG and catheterization data, it remains unclear whether this is clinically relevant. There were no associations between hemodynamics and KIM‐1, nor with the presence of dichotomously elevated KIM‐1 (>1300 pg/mL). Elevated urinary NGAL (>50 ng/mL), however, was associated with: higher Fontan pressure, lower cardiac index, and higher Q_P:Q_S. Also, using receiver operatic characteristics, these same hemodynamic characteristics (Fontan pressure, cardiac index, and Q_P:Q_S) were fair to good predictors of an individual having a urine NGAL >50 ng/mL. Our study demonstrated urine markers of kidney injury to be associated with higher central venous pressure (Fontan pressure) and lower cardiac index, findings suggestive of reduced renal perfusion.

The association with Q_P:Q_S, the ratio of pulmonary blood flow to systemic blood flow, also may have clinical implications. The presence of a Fontan fenestration or the presence of systemic‐to‐pulmonary venous collaterals may decompress pressure within the Fontan circuit or augment cardiac output at equivalent Fontan pressure. This may explain why having a lower Q_P:Q_S, suggesting a small degree of right‐to‐left shunting, was associated with urine NGAL <50 ng/mL.

The association between urine NGAL and body fat mass was unexpected. However, urine NGAL has previously been shown to correlate with the percentage body fat in prepubertal children (average age of 9 years). ²¹ This was reproduced by our study, as both body fat mass and percentage body fat correlated positively with urine NGAL. It is thought that adipose tissue may, via endocrine and inflammatory effects, cause kidney injury/disease in a similar way to how adipose tissue causes metabolic syndrome (increased oxidative stress, activation of the renin‐angiotensin‐aldosterone system, insulin resistance, and increased inflammation). ²² This may be particularly relevant in individuals with Fontan circulation, as they have more percentage body fat compared with healthy controls. ¹⁹

Limitations

These findings must be interpreted in the context of the underlying study design. Although participants were prospectively enrolled, several of the analyses depended on the availability of clinically ordered tests. Some of these data were missing, and there was often a time delay between the measurement of clinical variables and research laboratory tests. Our analysis assumes a constant renal function over a 1‐ to 2‐year period. Our study only had a single estimation of GFR, whereas reduced kidney function should be documented on multiple measurements over at least 3 months for a clinical diagnosis of CKD. ²³ In the absence of correction for multiple testing, caution should be taken in interpretation, particularly in the context of borderline level of statistical significance. The sample size of the cohort was relatively small, which limits statistical power. GFR was not measured directly, so we can only infer that cystatin C eGFR identifies more individuals with CKD compared with creatinine eGFR. We used the age of 19 years for deciding whether to estimate GFR using adult or pediatric equations, which can lead to an arbitrary variation of eGFR for individuals aged 18 to 19 years.

Conclusions

These data suggest that among those with Fontan circulation, eGFR may be overestimated by creatinine, leading to underdiagnosis of chronic kidney dysfunction. This is partially secondary to the reliance on creatinine eGFR on SMM, which is, on average, lower in individuals with Fontan circulation. Cystatin C is a preferable filtration in this population because it is not dependent on SMM. Finally, urine biomarkers of kidney injury (eg, NGAL) are associated with both unfavorable hemodynamics and higher body fat mass. Further research is warranted on whether such urinary biomarkers may be useful in clinical risk assessment or as indicators of clinical response in research studies.

Sources of Funding

This work was conducted with support from the Heart Institute Research Core at Cincinnati Children's Hospital.

Disclosures

Dr Opotowsky has previously received funding from Roche Diagnostics (Indianapolis, IN) for research on cystatin C; he has also consulted for and has served on an Independent Data Monitoring Committee for Janssen Pharmaceuticals, unrelated to this article. The remaining authors have no disclosures to report.