Association of Urine Galectin-3 with Cardiorenal Outcomes in Patients with Heart Failure

Abstract

Background and Objectives:

Approaches to distinguish pathologic cardiorenal dysfunction in heart failure (HF) from functional/hemodynamically mediated changes in serum creatinine are needed. We investigated urine galectin-3 as a candidate biomarker of renal fibrosis and prognostic indicator of cardiorenal dysfunction phenotypes.

Design, setting, participants, and measurements:

We measured urine galectin-3 in two contemporary HF cohorts: Yale Transitional Care Clinic (YTCC) cohort (n=132) and the Treatment of Preserved Cardiac Function Heart Failure with an Aldosterone Antagonist (TOPCAT) trial (n=434). We assessed the association of urine galectin-3 with all-cause mortality in both cohorts and the association with an established marker of renal tissue fibrosis, urinary amino-terminal propeptide of type III procollagen (PIIINP) in TOPCAT.

Results:

In the YTCC cohort, there was significant effect modification between higher urine galectin-3 and lower eGFR (P _interaction =0.046) such that low eGFR had minimal prognostic importance if urine galectin-3 was low but was high risk if urine galectin-3 was high. Similar observations were noted in the TOPCAT study (P _interaction=0.002). In TOPCAT, urine galectin-3 also positively correlated with urine PIIINP at both baseline (r=0.43; p< 0.001) and at 12-months (r=0.42; p< 0.001).

Conclusions:

Urine galectin-3 correlated with an established biomarker of renal fibrosis and in two cohorts was able to differentiate high vs. low-risk phenotypes of chronic kidney disease in HF. These proof-of-concept results indicate that additional biomarker research to differentiate cardiorenal phenotypes is warranted.

Graphical abstract

Lay Summary

New laboratory tests are needed to distinguish insignificant from bad changes in kidney function in patients with heart failure. We found that urine galectin-3 was higher in patients who died despite having the same values on other kidney tests. Urine galectin-3 requires further studies to determine if it can consistently provide new information on the meaning of kidney dysfunction in patients with heart failure.

Background

Kidney dysfunction plays a significant role in the pathophysiology of heart failure (HF).^1,2 Approximately one-half of all patients with HF have chronic kidney disease (CKD), and patients with comorbid CKD and HF are at a 2-fold higher risk of all-cause mortality.^3–5 This spectrum of heart-kidney disorders, termed cardio-renal syndrome (CRS), arises from variable etiologies and is precipitated by factors including neurohormonal and hemodynamic factors.^6–8 Yet, serum creatinine, the primary laboratory measure to identify patients with CRS, is unable to distinguish the mechanistic causes behind changes in glomerular filtration rate. Similar increases in serum creatinine have variable prognostic associations if caused by beneficial medical therapies or decongestion as opposed to pathologic CRS progression.^1,9,10 Biomarkers specific to pathologic CRS are an unmet need both for research and clinical care of patients.

Galectin-3 is a well-established mediator and marker of tissue fibrosis in HF and CKD.^11,12 Plasma galectin-3 is inversely associated with GFR, directly associated with albuminuria, and associated with risk of CKD progression.^13–16 However, in multiple HF trials, plasma galectin-3 was not independently associated with cardiac structural or functional abnormalities and did not provide additional prognostic value to natriuretic peptides or eGFR.^16–19 Plasma galectin-3’s lack of value in HF may stem from the absence of specificity for individual organ-level fibrosis, reflecting an amalgamation of fibrosis across the body, and confounding by kidney dysfunction, as plasma galectin-3 undergoes renal clearance.²⁰ Patients with HF have impaired plasma galectin-3 renal handling such that urine galectin-3 is not simply due to renal elimination of elevated plasma galectin-3 concentrations.^20,21 Given that kidney fibrosis is a final common pathway of CKD, urine galectin-3 may be more specific to renal fibrosis and could provide unique differentiation of CRS pathophysiology and prognosis in patients with HF.²²

Methods

We assessed urine galectin-3 in two contemporary HF cohorts. First, we assessed the association of urine galectin-3 with cardiorenal outcomes in the Yale Transitional Care Clinic (YTCC) cohort. Based on these preliminary results, we assessed urine galectin-3 among patients in the Treatment of Preserved Cardiac Function Heart Failure with an Aldosterone Antagonist (TOPCAT) trial.

YTCC Cohort

We prospectively enrolled consecutive patients (n=132 from 2013 – 2017) with HF receiving diuretic therapy at the Yale Transitional Care Clinic for outpatient intensified diuretic treatment of a worsening HF episode with hypervolemia. The primary outcome analyzed in this cohort was the association of urine galectin-3 with all-cause mortality and cardiorenal parameters over a follow up of 2.9 years. Mortality was assessed via electronic medical record review and social security death index query. All patients provided written informed consent and the study was approved by the Yale University School of Medicine Institutional Review Board.

TOPCAT Cohort

The TOPCAT trial was a multicenter, international, 1:1 randomized, double-blinded, placebo-controlled clinical trial that compared spironolactone vs. placebo on the rate of cardiac death, aborted cardiac arrest or HF hospitalization.²³ The trial methods and results have been previously reported.²³ Briefly, TOPCAT enrolled 3,445 patients with at least one sign and symptom of HF, age ≥ 50 years, left ventricular ejection fraction ≥ 45%, and either a history of HF hospitalization in the past year or elevated natriuretic peptide concentration. Key exclusion criteria included an estimated GFR < 30 ml/min/1.73m² or serum creatinine ≥ 2.5mg/dl. Treatment with spironolactone over a mean follow-up of 3.3 years did not reduce the primary outcome relative to placebo respectively (18.6% vs 20.4%; p=0.14).

We studied participants in the TOPCAT trial that participated in the biorepository sub-study with urine available at baseline (n=434 participants). We included patients in all geographical sites as there was no difference between the Americas (n=236 [54%]) versus Russia/Georgia (n=199 [46%]) sites in association of urine galectin-3 and mortality (p interaction =0.31). As in the YTCC cohort, we assessed the association with urine galectin-3 with all-cause mortality. Additionally, we assessed the correlation of urine galectin-3 with an established marker of renal tissue fibrosis, urinary amino-terminal propeptide of type III procollagen (PIIINP) at baseline and 12 months.^24–26 Changes in biomarker levels with placebo vs. spironolactone are not reported given the previously described issues with sites outside of the Americas^27,28 and a significant treatment interaction with spironolactone on urine galectin-3 concentrations based on region (region p interaction = 0.023).

Assays

To quantitate galectin-3 across wide range of concentration, we chose to develop and validate a urinary galectin-3 assay using a prototype multiplex assay on the Mesoscale platform (Meso Scale diagnostics, Gaithersburg, Maryland, USA), which employs proprietary electrochemiluminescence detection methods combined with patterned arrays. We employed the galectin-3 duoset containing the antibody pairs and standards from R and D systems (Catalog No: DY1154) to develop the custom assay. The raw data was analyzed using the Discovery workbench Software. Curve fitting was done using the same software and 4-PL curve was used to determine the concentration. Validation of the assay involved diluent standardization, calibration curve linearity, spike and recovery and dynamic range testing. The calibration curve obtained via the custom assay demonstrated a four-log dynamic range. All calibrators were run in duplicate, and the average intra-assay coefficient of variation was less than 10%. The lower limit of detection was determined from the respective standard curve and was defined as the calculated concentration of the signal that is 2.5 standard deviations over the blank. The average lower limit of detection obtained from multiple runs was 3.4 pg/ml and upper limit of detection was 4000 pg/ml. Urine galectin-3 levels were measured in duplicate using two different lots of galectin-3 antibodies and calibrators and the correlation between the two measurements was r2=0.91 indicating good reproducibility of data across batches of the duosets. Urine creatinine was measured by Jaffe’s method using the reagents and calibrators supplied by Randox laboratories on the automated analyzer RxImola (Randox Laboratories). For all analyses and results, urine levels of galectin-3 were indexed to urine creatinine levels to correct for differences in urinary concentration. Estimated glomerular filtration rate (eGFR) was calculated using the Chronic Kidney Disease Epidemiology Collaboration (CKD-EPI) formula and renal dysfunction was defined as an eGFR <60 ml/min/1.73m². Diuretic efficiency was defined as total sodium excreted per doubling of the loop diuretic dose as previously described.²⁹ The amino terminal propeptide of type III procollagen (PIIINP) was measured using the commercial P3NP ELISA kit from Cisbio (Perkin Elmer, USA) as per manufacturer’s instructions. The measuring range of the assay is from 0.036 – 33 μg/L and the detection limit is 0.036 μg/L. Plasma galectin-3 was measured using the enzyme-linked immuno-sorbent assay (ELISA) developed by BG Medicine (BG Medicine, Inc., Waltham, USA).

Statistical Analyses

Continuous data are reported as mean ± standard deviation for variables with normal distribution, and median (quartile 1 – quartile 3) for variables with skewed distribution. Categorical data are reported as frequency (percentage). Correlations between continuous variables are reported as Spearman’s rho. Pearson’s χ² test was used to compare categorical variables between groups. Urine galectin-3 was divided into “high” and “low” groups based upon the median galectin-3 in each cohort. This categorization was used to compare baseline characteristics in high vs low groups (Table and Supplementary Table). To compare continuous variables between groups, either Student’s t-test or the Wilcoxon Rank Sum test was used according to the observed distribution. Cox proportional hazards modeling was used to evaluate associations with all-cause mortality. Log-log survival curves and scaled Schoenfeld residuals were used to assess the proportionality hazards assumptions for each variable. For the TOPCAPT cohort, the model was adjusted for within-country correlation. Galectin-3 was log-transformed (natural logarithm e) to approximate normal distribution. Multivariable adjustment for outcome association included the following covariables: age, sex, race, eGFR, NT-proBNP, hemoglobin, chloride, sodium, treatment with ACEi or ARB, beta blocker, systolic blood pressure, heart rate, diabetes, and the interaction term of galectin-3*eGFR. This interaction term included both galectin-3 and eGFR as continuous variables. To assess for nonlinearity between galectin-3 and mortality, galectin-3 was modeled with a restricted cubic spline function with 3 knots determined by the percentiles recommended by Harrell (knots were 10, 50, and 90). In TOPCAT, NT-proBNP was missing in 37% of the patients, thus multiple imputation was used for to populate these values. The number of imputed datasets was 10. Statistical analysis was performed with IBM SPSS Statistics version 26 (IBM Corp, Armonk, NY) and Stata SE version 17.0 (StataCorp, College Station, TX). Statistical significance was defined as 2-tailed P<0.05 for all analyses.

Table:

Baseline Characteristics of TOPCAT Biorepository Cohort

Characteristic	Total Cohort (n=434)	Low Urine Gal-3/Ucr* (n=216)	High Urine Gal-3/Ucr* (n=218)	P-Value
Age (y)	69 ± 10	67 ± 9	71 ± 10	<0.001
Male	234 (54%)	135 (63%)	99 (45%)	<0.001
Caucasian	399 (92%)	202 (94%)	197 (90%)	0.23
Laboratory Values
Serum Sodium (mEq/L)	139 ± 4	140 ± 4	139 ± 4	0.02
Serum Chloride (mEq/L)	102 ± 5	102 ± 5	102 ± 5	0.98
BUN (mg/dl)	23 ± 11	21 ± 9	26 ± 13	<0.001
Serum Creatinine (mg/dl)	1.1 ± 0.3	1.1 ± 0.3	1.2 ± 0.4	0.57
eGFR (ml/min/1.73m2)	67 ± 20	69 ± 19	65 ± 20	0.01
Serum Albumin (g/dl)	4.0 ± 0.7	4.1 ± 0.6	4.0 ± 0.7	0.14
Hemoglobin (g/dl)	13.5 ± 1.7	13.8 ± 1.8	13.2 ± 1.6	< 0.001
NT-proBNP (pg/mL)	525 (207, 1049)	470 (218, 991)	555 (204, 1233)	0.25
Urine PIIINP/Ucr (mcg/mg)	2.9 (1.7 – 4.8)	2.2 (1.3 – 3.5)	3.6 (2.4 – 5.9)	<0.001
Comorbidities, n (%)
Hypertension	412 (95%)	206 (95%)	206 (94%)	0.68
Hyperlipidemia	304 (70%)	144 (67%)	160 (73%)	0.13
Atrial fibrillation	184 (42%)	76 (35%)	108 (50%)	0.002
Diabetes mellitus	140 (32%)	61 (28%)	79 (36%)	0.80
Vital Signs
Systolic blood pressure	126 ± 13	125 ± 12	127 ± 13	0.09
Heart rate	68 ± 10	68 ± 9	68 ± 11	0.58
Heart Failure Medications, n (%)
ACEI or ARB	329 (76%)	165 (76%)	164 (75%)	0.78
Beta Blocker	348 (80%)	177 (82%)	171 (78%)	0.36

Data presented as number (percentage), mean ± SD, or median (interquartile range) as appropriate.

^*Urine galectin-3 normalized to urine creatinine divided into low and high groups by the median value. ACEI = angiotensin converting enzyme inhibitor; ARB = angiotensin receptor blocker; BUN = blood urea nitrogen; eGFR = estimated glomerular filtration rate; NT-proBNP = N-terminal pro b-type natriuretic peptide; PIIINP = amino terminal propeptide of type III procollagen; Ucr= urine creatinine

Results

YTCC Cohort

The baseline characteristics of the 132 patients in the YTCC cohort are summarized in Supplementary Table. More patients had HF with preserved or mid-range left ventricular ejection fraction (LVEF) (56.1%) than a reduced LVEF, and the mean eGFR was 54±28mL/min/1.73m². The median urine galectin-3 concentration was 23.19 (IQR 13.39 – 34.09) pg/mg of urinary creatinine. Higher urine galectin-3 concentrations were more common in advanced age, higher BUN, and higher creatinine, (P< 0.05 for all); however, LVEF, comorbidities, and medical therapies were not different between higher and lower urine galectin-3 concentration groups. Urine galectin-3 did not correlate with baseline diuretic dose (ρ=−0.12, P=0.17) nor diuretic efficiency (ρ=0.02, P=0.86). Urine galectin-3 demonstrated correlation with plasma NT-proBNP (ρ=0.23, P=0.01), urine NGAL/creatinine (ρ=0.34, P=0.005), eGFR (ρ=−0.41, P<0.001), cystatin C (ρ=0.47, P<0.001), and microalbuminuria (ρ=0.24, P=0.006).

A total of 56 deaths occurred over 2.9 years. A higher urine galectin-3 concentration was associated with an increased risk of death [HR 1.37 per log increase, 95% CI (1.01–1.84), P=0.04]. This association did not appear to be nonlinear (P for non-linearity=0.40). Conversely, plasma galectin-3 was not associated with death [HR 1.13, 95% CI (0.67–1.92); P=0.64]. Importantly, the association between urine galectin-3 and risk of death appeared to be modified by eGFR as the continuous-by-continuous interaction was significant (P _interaction =0.046). Figure 1 shows a heat map where the increased mortality risk associated with lower eGFR was present only at higher urine galectin −3 concentrations.

Figure 1: Relationship between Urine Galectin −3 and Glomerular Filtration Rate with Survival.

The continuous relationship between eGFR and urine galectin-3 is displayed as a heat map for the YTCC (left) and TOPCAT (right) cohorts where blue is the lowest hazard ratio and red is the highest hazard ratio for survival. Heat maps included urine galectin 3 and eGFR as continuous variables in a full factorial model adjusted for age, sex, race, NT-proBNP, hemoglobin, chloride, sodium, treatment with ACEi or ARB, beta blocker, systolic blood pressure, and heart rate. At lower eGFR, increasing urine galectin-3 identified patients at higher risk of death. The P value for the interaction of urine galectin*eGFR was P=0.046 in YTCC and P=0.002 in TOPCAT in multivariable models.

TOPCAT Cohort

Because of the findings in YTCC, we explored urine galectin-3 in the TOPCAT study. Urine samples were available in 434 patients (98.6%) at baseline and 342 patients (77.7%) at 12-months. At baseline, the median urine galectin-3 was 139 pg/mg with an interquartile range (IQR) of (83 – 211 pg/mg). The baseline characteristics of the population are presented in Table. Similar to YTCC, higher urine galectin-3 values were associated with advanced age, lower eGFR, and higher BUN (P<0.05 for all). At 12-months, the median urine galectin-3 was 152 pg/mg (81 – 230 pg/mg) with a median change from baseline to 12-months of 0 (−47 to 58) pg/mg (P=0.59).

At baseline, higher urine galectin-3 had a higher median urine PIIINP, an established marker of renal tissue fibrosis, than lower urine galectin-3 (p<0.001). (Table) Urine galectin-3 also demonstrated significant positive correlations with urine PIIINP at both baseline (ρ=0.43; P< 0.001) and at 12-months (ρ=0.42; P< 0.001). (Figure 2) Urine galectin-3 correlated weakly with microalbuminuria (ρ=0.18; P= 0.02).

Figure 2: Association between urine galectin-3 and propeptide of type III procollagen (PIIINP).

Urine propeptide of type III procollagen (PIIINP) normalized to urine creatine (UCr) is plotted on the x-axis in deciles. Urine galectin-3 normalized to Ucr is plotted on the y-axis as mean (standard error of the mean). Urine PIIINP and urine galectin-3 displayed significant correlation at baseline (p trend<0.001) and 12-months (p trend<0.001)

All-cause mortality occurred in 66 patients, of which 44 (67%) were due to cardiovascular death. Urine galectin-3 was not associated with all-cause mortality (P=0.86; P for non-linearity=0.73). Similar to the YTCC cohort, effect modification was observed between eGFR and urine galectin-3 with an increased risk of all-cause mortality with high urine galectin 3 only at lower eGFR (P _interaction=0.002 in multivariable model; Figure 1).

Discussion

We investigated the associations of urine galectin-3 with mortality and an established renal tissue fibrosis biomarker (urine PIIINP) in contemporary HF cohorts. The key findings were: 1) urine galectin-3 differentiated patients with kidney dysfunction into high and low-risk subgroups, 2) urine galectin-3 demonstrated significant correlation with the renal fibrosis marker urine PIIINP. These observations with urinary galectin 3 provide proof-of-concept that differentiation of different cardio-renal phenotypes may be possible using biomarkers.

Kidney fibrosis is considered a final common pathway of most pathologies for CKD. However, functional changes in kidney function from stimuli such as initiation of renin-angiotensin-aldosterone system antagonists, sodium glucose co-transporter 2 inhibitors, hemodynamic fluctuations in blood pressure from medications such as vasodilators or beta blockers, and short-term changes in kidney from diuretic therapies are common in HF.³⁰ Importantly, it has been repeatedly demonstrated that stimuli such as the above commonly worsen kidney function but to not adversely affect prognosis.^9,10,31 It can be assumed that these stimuli also produce limited or no pro-fibrotic response, and thus renal fibrosis could be a pathway to differentiate pathologic vs. non-pathologic CKD in HF patients. Although accessible and specific approaches to identify renal fibrosis have been elusive, galectin-3 is a well-established fibrotic factor and thus the candidate marker we selected for this study.^11,12

A biomarker that can distinguish pathologic changes in glomerular filtration from functional changes due to decongestion or guideline-directed medical therapies that confer clinical benefits is greatly needed to improve the clinical care of patients. Despite significant effort, numerous novel plasma and urine biomarkers currently lack prognostic and clinical utility.^9,32,33 Based on the signal of urine galectin-3’s potential utility from the YTCC cohort, we investigated urine galectin-3’s association with clinical outcomes in the randomized, international TOPCAT clinical trial. In both cohorts we found that a low eGFR had the worst prognosis in patients with elevated urine galectin-3. Urine galectin-3 provided prognostic information on long-term mortality. The utility of urine galectin-3 to distinguish changes in serum creatinine during in-hospital treatment of acute HF is beyond the scope of these analyses and likely limited, as urine galectin-3 is hypothesized to measure renal fibrosis which may not develop during acute serum creatinine changes in acute HF. Although urine galectin-3 did not demonstrate a precise enough signal to have utility in individual patient care decisions, the observation that, in two independent and diverse cohorts, a biomarker could distinguish between high and low risk patients with a depressed eGFR is encouraging that additional research could identify a urinary biomarker accurate enough to provide clinical utility.

The TOPCAT study population has significant differences in the baseline characteristics, rates of the primary outcome, response to spironolactone, and markers of spironolactone adherence including serum potassium and serum drug metabolite concentrations in the Americas study sites.^27,28 Given these previously reported heterogenicities, we analyzed the associations with urine galectin-3 by regional study sites. As expected based on previously reported baseline differences, the Americas cohort had a higher baseline urine galectin-3 concentration. Despite differences at baseline, we found similar associations between urine galectin-3 and outcomes as well as changes over time in the Americas cohort and Russia/Georgia cohort and therefore reported these results without distinction by geographic region. Plasma galectin-3, plasma PIIINP, and other biomarkers associated with myocardial fibrosis also did not differ by treatment region in TOPCAT (abstract data).^34,35 However, we did find a significant treatment interaction (region p interaction = 0.023) where the change in urine galectin-3 in patients randomized to spironolactone was different in the Americas vs. non Americas study sites. These observations provide further evidence that there were likely differences in study conduct between regions and prohibit conclusions about the effects of spironolactone on urine galectin-3.^27,28

Several limitations warrant consideration. Although we analyzed urine galectin-3 in the international TOPCAT trial, only 434 patients participated in the biorepository sub-study of which 66 had a mortality event. Therefore, we may not have adequate power to detect an association between urine galectin-3 and mortality. Most patients studied had a LVEF > 40%, and further research is required in a reduced LVEF population. While the biomarker cut points identified in the YTCC cohort did not yield similar results in the TOPCAT cohort, qualitatively similar results were identified in both cohorts. Although this does indicate that urine galectin 3 is not likely to provide clinically actionable information for individual patient level decisions, it still provides proof of concept that different cardio-renal phenotypes may be identifiable using biomarkers. Lastly, significant regional heterogenicity existed in the TOPCAT trial. Although we found similar results in the Americas cohort, the small number of patients further limited the ability to detect associations or analyze the impact of spironolactone treatment.

In conclusion, urine galectin-3 may further distinguish the risk of adverse cardiorenal outcomes in patients with cardio-renal dysfunction. Further research is required to investigate this potential association and determine the clinical utility of urine galectin-3 and other biomarkers that may differentiate cardio-renal phenotypes.

Highlights:

• Biomarkers to distinguish pathologic cardiorenal dysfunction are needed in HF

• Urine galectin-3 may be more specific to renal fibrosis than plasma galectin-3

• Urine galectin-3 distinguished all-cause mortality risk at equal low eGFR values

• Urine galectin-3 correlated with P3NP, indicating specificity for renal fibrosis

• Urine galectin-3 may distinguish cardiorenal risk profiles

Biography