Urinary Dickkopf‐3 as a Potential Marker for Estimated Glomerular Filtration Rate Decline in Patients With Heart Failure Authors: Dennis Pieper, Anja Sandek, Ann‐Kathrin Schäfer, Hassan Dihazi, Gry Helene Dihazi, Andreas Leha, Michael Zeisberg, Stephan Lüders, Michael Koziolek, Manuel Wallbach Published: 2024 ABSTRACT BackgroundPatients with chronic heart failure (HF) show an increased risk for the occurrence of chronic kidney disease and cardiorenal syndrome. Urinary Dickkopf‐3 (uDKK3), a stress‐induced, tubular profibrotic glycoprotein, may be elevated in HF as early as in New York Heart Association class I HF and may indicate subsequent decline in estimated glomerular filtration rate (eGFR).Methods and ResultsuDKK3 levels in patients with HF and controls were measured by enzyme‐linked immunosorbent assay. eGFR was determined up to 5 years in HF. Change in eGFR was assessed with respect to baseline uDKK3 using (mixed‐effect) linear and logistic regression models. A total of 488 patients with chronic HF and 45 control patients were included. Patients with HF showed higher median uDKK3 levels than controls (259.6 pg/mg creatinine [interquartile range (IQR), 119.2–509.4 pg/mg creatinine] versus 107.5 pg/mg creatinine [IQR, 60.5–181.2 pg/mg creatinine], P<0.001). Regression models demonstrated a significant association between log uDKK3 and the decline in eGFR during a median of 13 months (IQR, 12–59 months) (estimated higher eGFR loss, 0.8039 mL/min per 1.73 m2/year [95% CI, 0.002–1.606 mL/min per 1.73 m2/year], P=0.049; odds ratio, 1.345 [95% CI, 1.049–1.741], P=0.021). uDKK3 levels ≥354 pg/mg creatinine were associated with a significantly higher risk for eGFR decline at 1‐year follow‐up (estimated higher eGFR loss, 4.538 mL/min per 1.73 m2 [95% CI, 1.482–9.593 mL/min per 1.73 m2]), P=0.004). Even patients with HF without chronic kidney disease (n=334) had higher uDKK3 levels compared with controls (233.4 [IQR, 109.0–436.9 pg/mg creatinine] versus 107.5 [IQR, 60.5–181.2 pg/mg creatinine], P<0.001).ConclusionsThe present findings indicate that uDKK3 is a promising prognostic biomarker for subsequent eGFR decline in patients with HF, irrespective of the presence of chronic kidney disease and even in the early stages of HF. This potential allows for early intervention to mitigate the deterioration of kidney function. Further investigation is warranted to validate its clinical utility. Nonstandard Abbreviations and Acronyms CKD‐Epidemiology Collaboration equation Dickkopf‐3 German Centre for Cardiovascular Research heart failure with mildly reduced ejection fraction heart failure with preserved ejection fraction heart failure with reduced ejection fraction ischemic cardiomyopathy Meta‐Analysis Global Group in Chronic Heart Failure nonischemic cardiomyopathy urinary albumin creatinine ratio urinary Dickkopf‐3 Clinical PerspectiveWhat Is New? Patients with chronic heart failure (HF), even in the early stages (New York Heart Association class I), exhibit elevated levels of urinary Dickkopf‐3 (uDKK3), a stress‐induced tubular profibrotic glycoprotein.Higher uDKK3 levels are associated with subsequent decline in estimated glomerular filtration rate over time, indicating a potential predictive value for renal function deterioration in patients with HF. What Are the Clinical Implications? uDKK3 could serve as a promising prognostic biomarker for monitoring renal function decline in patients with HF, regardless of the presence of chronic kidney disease.Early identification of elevated uDKK3 levels may allow for timely interventions to mitigate kidney function deterioration in patients with HF, potentially improving clinical outcomes.Further validation studies are warranted to confirm the clinical utility of uDKK3 as a prognostic marker and to explore its potential role in guiding personalized therapeutic strategies for patients with HF. Clinical Perspective What Is New? Patients with chronic heart failure (HF), even in the early stages (New York Heart Association class I), exhibit elevated levels of urinary Dickkopf‐3 (uDKK3), a stress‐induced tubular profibrotic glycoprotein.Higher uDKK3 levels are associated with subsequent decline in estimated glomerular filtration rate over time, indicating a potential predictive value for renal function deterioration in patients with HF. Patients with chronic heart failure (HF), even in the early stages (New York Heart Association class I), exhibit elevated levels of urinary Dickkopf‐3 (uDKK3), a stress‐induced tubular profibrotic glycoprotein. Patients with chronic heart failure (HF), even in the early stages (New York Heart Association class I), exhibit elevated levels of urinary Dickkopf‐3 (uDKK3), a stress‐induced tubular profibrotic glycoprotein. Higher uDKK3 levels are associated with subsequent decline in estimated glomerular filtration rate over time, indicating a potential predictive value for renal function deterioration in patients with HF. Higher uDKK3 levels are associated with subsequent decline in estimated glomerular filtration rate over time, indicating a potential predictive value for renal function deterioration in patients with HF. What Are the Clinical Implications? uDKK3 could serve as a promising prognostic biomarker for monitoring renal function decline in patients with HF, regardless of the presence of chronic kidney disease.Early identification of elevated uDKK3 levels may allow for timely interventions to mitigate kidney function deterioration in patients with HF, potentially improving clinical outcomes.Further validation studies are warranted to confirm the clinical utility of uDKK3 as a prognostic marker and to explore its potential role in guiding personalized therapeutic strategies for patients with HF. uDKK3 could serve as a promising prognostic biomarker for monitoring renal function decline in patients with HF, regardless of the presence of chronic kidney disease. uDKK3 could serve as a promising prognostic biomarker for monitoring renal function decline in patients with HF, regardless of the presence of chronic kidney disease. Early identification of elevated uDKK3 levels may allow for timely interventions to mitigate kidney function deterioration in patients with HF, potentially improving clinical outcomes. Early identification of elevated uDKK3 levels may allow for timely interventions to mitigate kidney function deterioration in patients with HF, potentially improving clinical outcomes. Further validation studies are warranted to confirm the clinical utility of uDKK3 as a prognostic marker and to explore its potential role in guiding personalized therapeutic strategies for patients with HF. Further validation studies are warranted to confirm the clinical utility of uDKK3 as a prognostic marker and to explore its potential role in guiding personalized therapeutic strategies for patients with HF. Impaired renal function is common in patients with heart failure (HF). In fact, up to 50% of patients with HF exhibit chronic kidney disease (CKD), which not only complicates the management of guideline‐directed pharmacological treatment for these patients but also worsens their prognosis. 1 Therefore, it is crucial to early identify patients who are at risk for a decline in their renal function by improved diagnostic methods. Recently, common pathophysiological pathways for the cardiorenal syndrome could be identified, suggesting a unifying pathogenesis in which several disease‐driving factors, such as inflammation, oxidative stress, endothelial dysfunction, hypertension, diabetes, and obesity lead to fibrosis, which is suggested to be a common feature in HF and CKD. 2 , 3 , 4 Thus, biomarkers that detect fibrotic processes and subclinical changes constitute a promising approach towards personalized intervention in cardiorenal syndrome. 5 , 6 To predict the loss of estimated glomerular filtration rate (eGFR) in patients with CKD, it was proposed to examine urinary Dickkopf‐3 (uDKK3), a stress‐induced tubular epithelium‐derived glycoprotein associated with tubulointerstitial fibrosis. 7 , 8 Whereas uDKK3 levels are exclusively elevated in patients with CKD, uDKK3 is generally absent in the general population without CKD. 7 However, it has been suggested that plasma Dickkopf‐3 (DKK3) may be involved in the pathogenesis and prognosis of HF. 9 , 10 While higher plasma concentrations of DKK3 have already been shown to be associated with HF, data on uDKK3, which may hold particular relevance in patients with HF with and without CKD, and its correlation with future eGFR course, are currently lacking. 9 Thus, uDKK3 might represent a promising biomarker that may provide insights into ongoing renal fibrosis and predict further renal function decline in cardiorenal patients. However, to date, the role of uDKK3 in HF as a predictor of renal function is unclear. We have previously demonstrated that uDKK3 is a valuable prognostic marker for renal function decline in patients with resistant hypertension. 11 Given the common occurrence of HF as a comorbidity, often complicated by cardiorenal syndrome, we aimed to assess whether uDKK3 retains prognostic value in early clinical stages (New York Heart Association [NYHA] class I) and to establish reference values for mild HF of various causes in patients with NYHA class I HF. METHODS Data Availability The data set analyzed in this study is available from the corresponding author upon reasonable request. Access requires prior approval in compliance with data protection regulations, including German Centre for Cardiovascular Research (DZHK) approval and ethics committee review. The data are not publicly available because of patient consent restrictions on public sharing. Study Design and Participants Baseline urine samples and clinical data of patients with NYHA class I HF were collected from cohort studies by the DZHK Heart Bank. The DZHK Heart Bank represents a research platform, which includes various DZHK‐funded studies, in which data and specimens were obtained using standardized procedures in all study centers of this national multicenter structure. The individual follow‐up data for the included patients differed depending on the underlying cohort study protocol. Patients in the control group were free of clinical or laboratory signs of heart or kidney failure and were included from our outpatient clinic. We excluded patients from the control group if they had a history of HF, clinical signs or symptoms of HF, prior echocardiography with a diagnosis of HF, elevated N‐terminal pro‐B‐type natriuretic peptide, eGFR <60 mL/min per 1.73 m2, albuminuria ≥30 mg/g creatinine or significant abnormalities in urine sediment. For subgroup analysis, HF with reduced ejection fraction (HFrEF), HF with mildly reduced ejection fraction (HFmrEF), and HF with preserved ejection fraction (HFpEF) were defined according to current HF guidelines. 12 The local ethics committee (#38/5/21) approved the study protocol. All participants gave written informed consent. From the control group, informed consent was obtained in our department, whereas data and urine from the HF group were from the DZHK Biobank (local Göttingen ethical vote number 12/6/19; ethical vote number Greifswald BB 135/17). The study complies with the principles of the Declaration of Helsinki. Laboratory Measurements Urinary chemistry measurements were performed using routine measurements. Determination of uDKK3 was performed using a commercial enzyme‐linked immunosorbent assay kit, which is validated for urine. 7 , 13 eGFR was calculated using the Chronic Kidney Disease Epidemiology Collaboration (CKD‐EPI) equation. Statistical Analysis Data are presented as mean±SD, median and interquartile range (IQR), or number (percentage). Statistical differences between groups for continuous or categorical variables were analyzed using Fisher exact tests, t tests, or Brunner‐Munzel tests, as appropriate. In analysis with >2 groups, Kruskal‐Wallis analysis with Dunn test for pairwise post hoc comparisons was performed. uDKK3 levels at baseline were compared on a log scale between groups using linear models controlling for sex, age, and body mass index (BMI) at baseline. Estimated marginal means per group were estimated with 95% CIs and compared between groups in contrast tests with Tukey adjustment of P values. Correlation analysis was performed using Spearman and rank‐based biserial correlation. Because of the screening character of these correlation analyses, no adjustment for multiple testing was conducted. To estimate the annual eGFR decline, a linear mixed‐effect model was fitted to the data using time as a predictor. Linear and logistic (mixed) regression models were used to model change in eGFR from baseline to follow‐up or the dichotomized outcome of >0.75 mL/min per 1.73 m2 decrease in eGFR from baseline to follow‐up, which corresponds to the average annual decline in kidney function in healthy individuals. 14 These models include uDKK3 (as a continuous log‐transformed variable or a variable categorized at >354 pg/mg creatinine) in interaction with time point as predictors and additionally controlled for eGFR at baseline, cause of HF, sex, age at baseline, BMI, urinary albumin creatinine ratio (UACR), and diabetes. Resulting model coefficients (for logistic models as odds ratios [ORs]) are reported with 95% CIs. For linear regression models, marginal means per group and time point were estimated with 95% CIs and compared per time point between groups using contrast tests. For all models, model diagnostics were visually inspected. We used receiver operating characteristic analysis of DKK3/creatinine as a predictor for eGFR decrease at 1 year >0.75 mL/min per 1.73 m2. The resulting area under the curve is given with 95% CIs and associated P value testing the null hypothesis of no predictive value. Using this ROC analysis, we graphically selected an uDKK3 cutoff within the clinically relevant range with a specificity for future eGFR decline of >80% and used it for the subsequent analysis. The significance level was set to an α of 5% for all statistical tests. All analyses were performed with the statistic software R version 4.2.3 (R Foundation for Statistical Computing: https://www.R‐project.org/) using the R packages lme4 (version 1.1.32) 15 for the mixed‐effect linear and logistic regression models, lmerTest (version 3.1.) 16 for the estimation of the Satterthwaite denominator degrees of freedom, and emmeans (version 1.8.5) (Estimated Marginal Means, aka Least‐Squares Means: https://CRAN.R‐project.org/package=emmeans) for the estimation of the marginal means and the contrast tests. RESULTS Study Population A total of 488 patients with HF (NYHA class I) and a control group of 45 patients without evidence of heart and kidney failure were included in the present study. Figure 1 shows a flow diagram of the study. Baseline creatinine values were available in 482 patients with HF. Patients with HF had a mean age of 57±15 years and 101 (20.7%) were women. Mean BMI was 27.9±4.9 kg/m2. UACR ≥30 mg/g creatinine was present in 116 patients (23.8%). Diabetes was present in 64 patients (13.1%). A total of 103 patients (21.1%) had ischemic cardiomyopathy (ICM), 301 patients (61.7%) had non‐ICM (NICM; including hereditary and inflammatory dilated cardiomyopathy, left ventricular noncompaction cardiomyopathy, hypertrophic cardiomyopathy, arrhythmogenic right ventricular cardiomyopathy, myocarditis, and amyloidosis), and 65 patients (13.3%) had a combination of ICM (coronary artery disease) and NICM (Figure 1). On average, patients with HF took 2.4±1.1 different classes of HF medications. The baseline characteristics of participants are shown in Table 1. Flow diagram of the study. eGFR indicates estimated glomerular filtration rate; HF, heart failure; ICM, ischemic cardiomyopathy; NICM, nonischemic cardiomyopathy; and uDKK3, urinary Dickkopf‐3. Table 1Baseline Characteristics in Patients With HF, Controls, and a Subgroup of Patients With HFHF cohort n=488† Controls n=45 P valueICM n=103NICM n=301 P value*Women, n (%)101 (21)25 (56)<0.00113 (13)72 (24)0.017Age, y56.7±14.548.2±13.3<0.00167.2±12.051.4±13.4<0.001BMI, kg/m2 27.9±4.926.4±4.10.02528.3±4.328.0±5.20.593eGFR, mL/min per 1.73 m2 85.3±19.7‡‡ 95.6±14.5<0.00176.3±18.6||| 89.6±19.1## <0.001UACR, mg/g creatinine12.6 (7.0–27.5)8.9 (6.1–13.1)‡‡‡‡ <0.00119.4 (9.4–72.1)11.6 (6.4–22.1)<0.001UACR >30 mg/g creatinine116 (23.8%)0<0.001SBP, mm Hg126±17§§ 139±17‡‡‡‡ <0.001137±17122±16§§§§ <0.001DBP, mm Hg77±11§§ 84±13‡‡‡‡ 0.00180±1076±10§§§§ 0.002Classes of HF and LVEFn=408n=62n=278HFrEF, n (%)234 (57)53 (85)142 (51)<0.001HFmrEF, n (%)82 (20)9 (15)59 (21)HFpEF, n (%)92 (23)0 (0)77 (28)LVEF, %41.1±10.736.6±5.342.2±11.3<0.001Medical history, n (%)Atrial fibrillation132 (27)‡ 0 (0)<0.00130 (29)71 (24)§§§ 0.270Arterial hypertension222 (46)‡ 16 (36)0.21375 (73%)102 (34)§§§ <0.001Hyperlipoproteinemia157 (34)§ 6 (13)0.00451 (54)‡‡‡ 66 (23)††† <0.001Diabetes64 (13)‡ 0 (0)0.00426 (25)26 (9)§§§ <0.001Apoplexy36 (7)0 (0)0.0619 (9)23 (8)0.678COPD40 (8)|| 0 (0)0.0398 (8)***25 (8)|||| 1.000History of smoking272 (56)22 (49)0.43475 (73)149 (50)<0.001History of ADHF73 (16)# 0 (0)0.0010 (0)***65 (24)### <0.001uDKK3, pg/mg creatinine259.6 (119.2–509.4)107.5 (60.5–181.2)<0.001442.9 (240.9–787.7)223.2 (99.8–430.4)<0.001Log uDKK3, pg/mg creatinine5.53±1.124.47±1.13<0.0016.1±0.95.4±1.1<0.001NT‐proBNP, ng/L304 (130–663)**54 (33–76)†† <0.001459 (249–924)†††† 230 (96–538)**** <0.100Cardiovascular medication, n (%)n=103n=294ACEI279 (58%)§§§§ 5 (11%)<0.00162 (60%)169 (56%)#### 0.728ARB124 (26%)§§§§ 6 (13%)0.06833 (32%)66 (22%)#### 0.064MRA255 (53%)||||| 1 (2%)<0.00164 (62%)146 (49%)0.030β‐Blocker420 (88%)||||| 5 (11%)<0.00198 (95%)249 (83%)0.005CCB40 (8%)§§§§ 5 (11%)0.51012 (12%)20 (7%)#### 0.141Loop diuretic82 (17%)||||| 0 (0%)0.0030 (0%)70 (23%)<0.001Thiazide71 (15%)§§§§ 2 (4%)0.05816 (16%)36 (12%)#### 0.400Values are expressed as mean±SD, number (%), or median (IQR). P value comparison of HF cohort and controls. P value*: comparison between the ICM and NICM subgroup, †The entire HF cohort (N=488) consisting of patients with ICM (n=103), NICM (n=301), ICM + NICM (n=65), and not classified (n=19). For better clarity and evaluation, the subgroup comparison in the table was limited to ICM and NICM. Data available for ‡ n=487, § n=459, || n=481, # n=37, ** n=434, †† n=44, ‡‡ n=482, §§ n=479, ||| n=102, ## n=297, *** n=102, ††† n=283, ‡‡‡ n=95, §§§ n=300, |||| n=295, ### n=268, **** n=260, †††† n=97, ‡‡‡‡ n=44, §§§§ n=479, ||||| n=480, and #### n=293. ACEI indicates angiotensin‐converting enzyme inhibitor; ADHF, acute decompensated heart failure; ARB, angiotensin receptor blocker; BMI, body mass index; CCB, calcium channel blocker; COPD, chronic obstructive pulmonary disease; DBP, diastolic blood pressure; eGFR, estimated glomerular filtration rate; HF, heart failure; HFmrEF, heart failure with mildly reduced ejection fraction; HFpEF, heart failure with preserved ejection fraction; HFrEF, heart failure with reduced ejection fraction; ICM, ischemic cardiomyopathy; IQR, interquartile range; LVEF, left ventricular ejection fraction; MRA, mineralocorticoid receptor antagonist; NICM, nonischemic cardiomyopathy; NT‐proBNP, N‐terminal pro‐B‐type natriuretic peptide; SBP, systolic blood pressure; UACR, urinary albumin creatinine ratio; and uDKK3, urinary Dickkopf‐3. Baseline Characteristics in Patients With HF, Controls, and a Subgroup of Patients With HF Values are expressed as mean±SD, number (%), or median (IQR). P value comparison of HF cohort and controls. P value*: comparison between the ICM and NICM subgroup, †The entire HF cohort (N=488) consisting of patients with ICM (n=103), NICM (n=301), ICM + NICM (n=65), and not classified (n=19). For better clarity and evaluation, the subgroup comparison in the table was limited to ICM and NICM. Data available for ‡ n=487, § n=459, || n=481, # n=37, ** n=434, †† n=44, ‡‡ n=482, §§ n=479, ||| n=102, ## n=297, *** n=102, ††† n=283, ‡‡‡ n=95, §§§ n=300, |||| n=295, ### n=268, **** n=260, †††† n=97, ‡‡‡‡ n=44, §§§§ n=479, ||||| n=480, and #### n=293. ACEI indicates angiotensin‐converting enzyme inhibitor; ADHF, acute decompensated heart failure; ARB, angiotensin receptor blocker; BMI, body mass index; CCB, calcium channel blocker; COPD, chronic obstructive pulmonary disease; DBP, diastolic blood pressure; eGFR, estimated glomerular filtration rate; HF, heart failure; HFmrEF, heart failure with mildly reduced ejection fraction; HFpEF, heart failure with preserved ejection fraction; HFrEF, heart failure with reduced ejection fraction; ICM, ischemic cardiomyopathy; IQR, interquartile range; LVEF, left ventricular ejection fraction; MRA, mineralocorticoid receptor antagonist; NICM, nonischemic cardiomyopathy; NT‐proBNP, N‐terminal pro‐B‐type natriuretic peptide; SBP, systolic blood pressure; UACR, urinary albumin creatinine ratio; and uDKK3, urinary Dickkopf‐3. CKD Progression in Patients With HF Figure 2 shows the eGFR slope within the HF cohort during follow‐up. The baseline eGFR was 85.3±19.7 mL/min per 1.73 m2, with higher baseline eGFR values in women (93.5±18.0 versus 83.2±19.6 mL/min per 1.73m2, P<0.001). Compared with baseline, eGFR reductions were −2.4 mL/min per 1.73 m2 (95% CI, −4.1 to −0.73 mL/min per 1.73 m2) at 1‐year follow‐up (n=160, P=0.005), −5.3 mL/min per 1.73 m2 (95% CI, −7.7 to −2.99 mL/min per 1.73 m2) at 4‐year follow‐up (n=70, P<0.001), and −12 mL/min per 1.73 m2 (95% CI, −14.0 to −9.53 mL/min per 1.73 m2) at 5‐year follow‐up (n=87, P<0.001). The mean annual eGFR decline was −1.94 mL/min per 1.73 m2 (95% CI, −2.3 to −1.6 mL/min per 1.73 m2). Patients with HF without CKD also showed clinically meaningful eGFR reductions during follow‐up from baseline: −2.5 (95% CI, −4.5 to −0.45 mL/min per 1.73 m2) at 1 year (n=115, P=0.017), −5.8 (95% CI, −8.2 to −2.6 mL/min per 1.73 m2) at 4 years (n=52, P=0.001), and −11.0 (95% CI, 14.0 to −8.3 mL/min per 1.73 m2) at 5 years (n=55, P<0.001). eGFR slope during follow‐up in patients with HF. The gray dots are the measurements with their distribution per time point additionally visualized as overlayed violin plots. The black bars show the 95% CIs around the expected marginal means estimated from a linear mixed‐effect regression model using time point as fixed effects and patient as random effect. The annotated P values comparing the follow‐up time points to baseline are also results from that model. eGFR indicates estimated glomerular filtration rate; and HF, heart failure. uDKK3 in Patients With HF Patients in the HF cohort showed a median uDKK3 level of 259.6 pg/mg creatinine (IQR, 119.2–509.4 pg/mg creatinine), which was higher compared with levels in the control group (107.5 pg/mg creatinine [IQR, 60.5–181.2 pg/mg creatinine], P<0.001) (Table 1). As shown in Table 1, patients with ICM had higher uDKK3 levels (median, 442.9 pg/mg creatinine [IQR, 240.9–787.7 pg/mg creatinine]) compared with patients with NICM (median, 223.2 pg/mg creatinine [IQR, 99.8–430.4 pg/mg creatinine], P<0.001). Table 2 shows uDKK3 levels separately according to the corresponding CKD stage within the HF group with available baseline eGFR values (n=482). A subgroup of 334 patients with HF without CKD (69%) (UACR <30 mg/g creatinine and eGFR ≥60 mL/min per 1.73 m2) showed higher median uDKK3 levels compared with controls (uDKK3 233.4 pg/mg creatinine [IQR, 109.0–436.9 pg/mg creatinine] versus 107.5 pg/mg creatinine [IQR, 60.5–181.2 pg/mg creatinine], P<0.001) (log uDKK3 data in Figure S1). Within the entire HF cohort, there were no correlations between uDKK3 and sex, baseline eGFR, left ventricular ejection fraction, or presence of arterial hypertension. uDKK3 was correlated with BMI, presence of diabetes, age, UACR, level of N‐terminal pro‐B‐type natriuretic peptide, intake of angiotensin‐converting enzyme inhibitors, and presence of edema (Table S1). Using a linear mixed model with adjustment for age, sex, and BMI, differences in uDKK3 levels between the HF cohort and controls (log uDKK3: 5.5±1.1 pg/mg creatinine versus 4.5±1.1 pg/mg creatinine, P<0.001) (Table 1 and Figure S2), as well as between patients with ICM and NICM (log uDKK3: 6.1±0.9 pg/mg creatinine versus 5.4±1.1 pg/mg creatinine, P<0.001) (Table 1 and Figure 3), remained statistically significant. Figure 3 shows log uDKK3 (pg/mg creatinine) levels within the HF cohort separated according to the different HF causes, as well as for the control group. Echocardiographic data were available for 408 patients (83.6%). Based on left ventricular ejection fraction data, 234 patients (57.4%) could be classified as having HFrEF (median uDKK3, 264.0 pg/mg creatinine [IQR, 121.2–501.2 pg/mg creatinine]), 82 patients (20.1%) as having HFmrEF (median uDKK3, 229.6 pg/mg creatinine [IQR, 78.6–483.3 pg/mg creatinine]), and 92 patients (22.5%) as having HFpEF (median uDKK3, 215.5 pg/mg creatinine [IQR, 110.2–424.9 pg/mg creatinine]). There were no differences in uDKK3 between the HF classes (HFrEF versus HFmrEF: P=0.209; HFmrEF versus HFpEF: P=0.846; HFrEF versus HFpEF: P=0.251) (Table S2). Table 2Distribution of Patients With HF According to CKD Stage and the Corresponding Median uDKK3 Level in pg/mg Creatinine for Each CategoryCKD stageUACRA1A2A3<30 mg/g30–300 mg/g>300 mg/geGFR (mL/min per 1.73 m2)G1≥90 n=334 (69%) 233.38 pg/mg n=34 (7%) 396.37 pg/mg n=5 (1%) 732.92 pg/mg G260–89 n=42 (9%) 427.33 pg/mg n=8 (2%) 995.03 pg/mg G3a45–59 n=27 (6%) 198.01 pg/mg n=16 (3%) 399.26 pg/mg n=1 (0.2%) 498.23 pg/mg G3b30–44 n=7 (1%) 173.88 pg/mg n=5 (1%) 1451.72 pg/mg n=1 (0.2%) 2227.15 pg/mg G415–29 n=1 (0.2%) 5.11 pg/mg n=1 (0.2%) 4894.89 pg/mg n=0 (0%)G5<15n=0 (0%)n=0 (0%)n=0 (0%)In 6 patients with HF, baseline serum creatinine levels were missing (n=482). Patients with eGFR ≥60 mL/min per 1.73 m2 and A1 albuminuria were summarized into a single category. CKD indicates chronic kidney disease; eGFR indicates estimated glomerular filtration rate; HF, heart failure; UACR, urinary albumin creatinine ratio; and uDKK3, urinary Dickkopf‐3. Distribution of Patients With HF According to CKD Stage and the Corresponding Median uDKK3 Level in pg/mg Creatinine for Each Category n=334 (69%) 233.38 pg/mg n=34 (7%) 396.37 pg/mg n=5 (1%) 732.92 pg/mg n=42 (9%) 427.33 pg/mg n=8 (2%) 995.03 pg/mg n=27 (6%) 198.01 pg/mg n=16 (3%) 399.26 pg/mg n=1 (0.2%) 498.23 pg/mg n=7 (1%) 173.88 pg/mg n=5 (1%) 1451.72 pg/mg n=1 (0.2%) 2227.15 pg/mg n=1 (0.2%) 5.11 pg/mg n=1 (0.2%) 4894.89 pg/mg In 6 patients with HF, baseline serum creatinine levels were missing (n=482). Patients with eGFR ≥60 mL/min per 1.73 m2 and A1 albuminuria were summarized into a single category. CKD indicates chronic kidney disease; eGFR indicates estimated glomerular filtration rate; HF, heart failure; UACR, urinary albumin creatinine ratio; and uDKK3, urinary Dickkopf‐3. Log uDKK3 in patients with different causes of HF and the control group. P values were adjusted for age, sex, and body mass index. HF indicates heart failure; ICM, ischemic cardiomyopathy; NICM, nonischemic cardiomyopathy; and uDKK3, urinary Dickkopf‐3. uDKK3 Predicts eGFR Decline in Patients With HF Median time to first available eGFR follow‐up was 13 months (IQR, 12–59 months). In a linear mixed‐effect regression model incorporating log uDKK3 (pg/mg creatinine), baseline eGFR, age, sex, BMI, presence of diabetes, log UACR, and cause of HF as predictors, log uDKK3 was a significant predictor for eGFR decline from baseline to first follow‐up (estimated higher drop in eGFR of 0.8039 mL/min per 1.73 m2/year [95% CI, 0.002–1.606 mL/min per 1.73 m2/year] per log unit increase in uDKK3, P=0.049) (Figure 4). Within this model, there was an association between subsequent higher eGFR loss and female sex (estimated higher eGFR loss of 4.7525 mL/min per 1.73 m2/year (95% CI, 2.384–7.066 mL/min per 1.73 m2/year; P<0.001) as well between subsequent higher eGFR loss and higher baseline eGFR (estimated higher eGFR loss 0.07971 mL/min per 1.73 m2/year [95% CI, 0.0014–0.0145 mL/min per 1.73 m2/year] per unit increase in baseline eGFR, P=0.017) with subsequent higher eGFR loss. No associations between the eGFR slope and age, BMI, cause of HF, presence of diabetes, and log UACR were detected (Table S3). uDKK3 showed an association with a decrease in eGFR >0.75 mL/min per 1.73 m2 from baseline to the earliest follow‐up as outcome within a logistic regression model using eGFR at baseline, age at baseline, sex, BMI at baseline, diabetes, log (UACR), reason for HF, and log uDKK3 (pg/mg creatinine) as predictors, with an OR of 1.345 (95% CI, 1.049–1.741; P=0.021) (Table S4). uDKK3 predicts eGFR decrease in patients with HF. Linear trend of the association between ΔeGFR in mL/min per 1.73 m2 to the earliest follow‐up and uDKK3 in patients with HF resulting from a linear model for earliest eGFR follow‐up. The blue line indicates the estimated change in eGFR with the respective 95% CI (gray area). Analyses and plots were adjusted for age, sex, body mass index, cause of HF, diabetes, log UACR, and baseline eGFR. ΔeGFR indicates change in estimated glomerular filtration rate; eGFR, estimated glomerular filtration rate; HF, heart failure; UACR, urinary albumin creatinine ratio; and uDKK3, urinary Dickkopf‐3. uDKK3 for the Prediction of 1‐Year eGFR Loss in Patients With HF Considering only patients with 1‐year follow‐up, uDKK3 showed a trend for an association in the linear regression model for change in eGFR (estimated higher eGFR loss of 1.091 mL/min per 1.73 m2 [95% CI, −0.098 to −2.280] per log unit increase of uDKK3, P=0.072) and reached statistical significance within a logistic regression model for eGFR decrease of >0.75 mL/min per 1.73 m3 (OR, 1.49 [95% CI, 1.096–2.079], P=0.014) (Tables S5 and S6). With regard to classes of HF, there was a significant correlation between log uDKK3 and change of eGFR after 1 year in HFrEF (Spearman ρ=−0.304 [95% CI, −0.516 to −0.057], P=0.017; n=61), whereas no significant correlation was observed in HFpEF: (Spearman ρ=−0.138 [95% CI, −0.389 to 0.132), P=0.316; n=55) and HFmrEF (Spearman ρ=0.106 [95% CI, −0.213 to 0.404], P=0.516; n=40) (Figure S3A). Correlation of uDKK3 With Change in eGFR After 1, 4, and 5 Years uDKK3 is correlated with change in eGFR after 1 year but not after 4 and 5 years. A visual approach using a scatter plot with log uDKK3 (pg/mg creatinine) and change in eGFR showed a fitting curve at 1 year (Spearman ρ=−0.225 [95% CI, −0.37 to −0.07], P=0.005; n=159) but not at 4 years (ρ=−0.023 [95% CI, −0.26 to 0.21), P=0.850; n=69) or 5 years (ρ=0.046 [95% CI, −0.17 to 0.26], P=0.675; n=86) (Figure S3B). Influence of Categorized uDKK3 on eGFR Decline ROC analysis of uDKK3 as a predictor for eGFR decrease at 1 year >0.75 mL/min per 1.73 m2 showed an area under the curve of 0.622 (95% CI, 0.534–0.709), indicating a moderate to high predictive power of the test. A uDKK3 cutoff of 354 pg/mg creatinine showed a specificity of 80.8% for eGFR decrease of >0.75 mL/min per 1.7 3m2 whereas the cutoff of 75 pg/mg creatinine showed a sensitivity of 84.7% (Figure S4). There was a difference in the 1‐year change in eGFR between the 3 groups (uDKK3 [A] 0–74 pg/mg creatinine, [B] 75–354 pg/mg creatinine, [C] ≥354 pg/mg creatinine, P=0.0018), with higher 1‐year eGFR drop in the group with uDKK3 levels ≥354 pg/mg creatinine compared with the group with uDKK3 levels 0 to 74 pg/mg creatinine (P=0.001) and the group with uDKK3 levels 75 to 354 pg/mg creatinine (P=0.05) (Figure 5). In patients with uDKK3 ≥354 pg/mg creatinine, a decrease in eGFR after 1 year was significantly higher than in patients with uDKK3 levels below this threshold (−5.57 mL/min per 1.73 m2 [95% CI, −1.18 to −9.96], n=50; versus −1.98 mL/min per 1.73 m2 [95% CI, 2.05 to −6.00], n=109; P=0.045). There was no difference in the change of eGFR after 4 and 5 years between both groups (P=0.850 and P=0.495, respectively) (Table S7). Categorization of uDKK3 at ≥354 pg/mg creatinine showed an association with the change in eGFR at the earliest follow‐up and at 1‐year follow‐up as part of linear regression models (estimated higher eGFR loss at earliest follow‐up: 2.76 mL/min per 1.73 m2 [95% CI, 0.868–4.652], P=0.004; at 1‐year follow‐up: 4.538 mL/min per 1.73 m2 [95% CI, 1.482–7.593], P=0.004) and within a logistic regression model for a decrease in eGFR >0.75 mL/min per 1.73 m2 (earliest follow‐up: OR, 2.493 [95% CI, 1.365–4.650], P=0.003; 1‐year follow‐up: OR, 4.167 [95% CI, 1.834–10.08], P=0.001) (Tables S8 through S11). Short‐term eGFR loss according to uDKK3 category in patients with HF. Change in eGFR from baseline to 1‐year follow‐up in mL/min per 1.73 m2 according to uDKK3/mg creatinine in patients with HF. Testing for significance was adjusted for age, sex, and body mass index. eGFR indicates estimated glomerular filtration rate; HF, heart failure; and uDKK3, urinary Dickkopf‐3. DISCUSSION For the first time, to our knowledge, we demonstrated uDKK3 to be a potential predictive marker for subsequent decline in kidney function in patients with NYHA class I HF. The finding that these patients without HF symptoms and in part even without evidence for CKD already show higher uDKK3 levels compared with controls suggests that uDKK3 might detect subclinical renal changes or ongoing renal fibrosis very early, even before CKD becomes clinically evident. The elevated uDKK3 levels in patients with HF without CKD might be caused by early HF‐induced hemodynamic changes and systemic inflammation leading to renal tubular stress. Of note, those patients with HF without CKD showed an eGFR decline in a similar manner to the overall HF cohort, indicating that diagnostic and therapeutic efforts with regard to CKD care in HF should also address patients without evident CKD. The present data demonstrate an interaction between uDKK3 levels and greater eGFR decline during a median of 13 months. The adjusted level of uDKK3 is higher in patients with HF compared with controls, suggesting that uDKK3 levels might be influenced by pathological processes occurring in HF. Given that HF is a complex condition involving inflammation, oxidative stress, tissue remodeling, and hemodynamic changes, it is possible that these factors influence the expression and secretion of DKK3. Higher uDKK3 levels in patients with ICM may indicate a specific risk in these patients as underlined by a recent report that emphasizes the higher rate of comorbidities and worse prognosis in patients with ICM as compared with patients with NICM. 17 The correlation of uDKK3 with various clinically relevant parameters underscores its potential relevance as a biomarker in HF. The main finding of the study is the predictive value of uDKK3 for eGFR decline in patients with HF. A categorical model revealed a moderate to high predictive power of the uDKK3 test. uDKK3 at a cutoff value of <75 pg/mg creatinine is suitable to rule out relevant future eGFR decline with high probability, whereas the uDKK3 cutoff point of 354 pg/mg creatinine is suitable to predict further eGFR decline after 1 year with high probability. In particular, patients with uDKK3 levels ≥354 pg/mg creatinine showed a distinct higher estimated eGFR loss of 4.54 mL/min per 1.73 m2 after 1 year of follow‐up compared with patients with lower uDKK3 levels. These data support the predictive value of uDKK3 for identifying patients with HF at higher risk of kidney function decline. While uDKK3 was predictive for an eGFR change after 1 year, it did not predict eGFR decline at 4 and 5 years of follow‐up. Since various external variables influence the course of patients with HF during a longer period, it seems plausible that uDKK3 represents preferably a short‐term prognostic parameter. In the general population, renal function decreases at a rate of ≈0.5 to 1.0 mL/min per 1.73 m2/year, which equates to an eGFR decrease in patients with HF. 18 , 19 Nonetheless, the present study found a significant annual decline in eGFR of −1.94 mL/min per 1.73 m2 in patients with HF. Even though the decline in renal function is generally modest in chronic HF, it is associated with increased cardiovascular event rates and all‐cause mortality. 19 For each 5‐mL/min per 1.73 m2 decrease in eGFR, there is a 26% higher risk of cardiovascular death, emphasizing the importance of monitoring even slight declines in kidney function in patients with HF, which directly increases the vulnerability to cardiovascular events. 18 , 20 , 21 Thereby, it should be taken into account that the clinical situation in which deterioration of kidney function occurs in HF is probably crucial. Worsening renal function during a persistent congestive state was related to poor outcome, while worsening renal function during acute recompensation was not. 22 In the context of chronic HF, as observed in the current study involving patients categorized as having NYHA class I HF, the deterioration of renal function is indicated to be linked with an unfavorable prognosis. 23 This study underscores both the possibility and the need for improved early diagnostic approaches that may pave the way for early interventions to mitigate the risk of renal function decline over time and its associated cardiorenal complications in this patient population as early as possible. Given the correlation between elevated levels of uDKK3 and heightened tubulointerstitial fibrosis, it is plausible to consider that individuals with HF exhibiting elevated uDKK3 levels may experience potential advantages from intensified antifibrotic and nephroprotective strategies. In a meta‐analysis, predictors for the occurrence of worsening renal function in HF could be identified: presence of CKD (OR, 2.17), arterial hypertension (OR, 1.36), diabetes (OR, 1.23), age (per 10 years) (OR, 1.38), and diuretic use (OR, 1.52). 24 In comparison, the present data demonstrated an OR of 4.17 for eGFR decline >0.75 mL/min per 1.73 m2 in the first year in patients with a uDKK3 level >354 pg/mg creatinine. The steeper eGFR decline observed in women in the present study aligns with previous reports, likely attributable to women being less frequently treated with guideline‐recommended HF medication. 23 , 25 , 26 As women had higher baseline eGFR values compared with men in the present study and higher baseline eGFR was associated with a steeper eGFR decline, the difference in baseline eGFR might also explain, at least in part, the association between female sex and higher eGFR loss in the present study, as higher eGFR loss in women might result from the effect of the regression to the mean. Since there are no other data on the effect of sex on eGFR decline in NYHA class I HF, further investigation into the impact of sex on kidney function decline in HF is warranted. Thereby, analyses stratifying female patients by hormonal status (eg, premenopausal versus postmenopausal) or by comorbid conditions more prevalent in women (eg, autoimmune diseases) might serve as interesting approaches to better understand sex‐specific influences on eGFR decline in HF. Notably, uDKK3 was not associated with sex. Although albuminuria is associated with the incidence and progression of HF and cardiovascular outcome, 27 , 28 , 29 , 30 , 31 , 32 there are no specific data available analyzing the prognostic value of albuminuria on the course of renal function in these patients. Though established for progression estimation in general CKD, in the present HF cohort, no association of albuminuria with subsequent eGFR decline was found, highlighting the prominent role of uDKK3 for predicting future renal function decline in HF.28 The limited association of albuminuria with eGFR course is in line with previous data on diabetic nephropathy, in which patients, characteristically women, progress to advanced CKD with normalbuminuria. 33 , 34 , 35 To what extent CKD in HF is driven by different potential risk factors is unknown, but the present data suggest independence of albuminuria. However, uDKK3 was correlated with a 1‐year eGFR decline in HFrEF, but this was not observed in HFpEF and HFmrEF, potentially driven by a higher prevalence of ICM in HFrEF displaying the highest risk of progression to CKD. CKD is closely linked to cardiovascular diseases, including ICM. The observation of higher uDKK3 levels in ICM compared with NICM could reflect greater kidney stress or injury associated with the more extensive vascular disease and higher mortality observed in ICM. 36 , 37 This distinction underscores the need for further research to elucidate the underlying mechanisms driving differential uDKK3 expression between these HF subtypes. In addition, previous bioinformatic analyses have identified several genes linking CKD and ICM, highlighting the role of fibrotic processes. 38 These findings suggest that elevated uDKK3 levels in ICM might be associated with increased fibrosis, potentially indicating a pathophysiological process in the interplay between CKD and ICM. Understanding these mechanisms could enhance the utility of uDKK3 as a biomarker for predicting kidney function decline and tailoring interventions based on HF cause. The observation that uDKK3 correlates with 1‐year eGFR decline in HFrEF, but not in HFpEF and HFmrEF, is notable in the context of cardiorenal interaction. CKD generally worsens outcomes among all HF types. 39 However, in HFpEF, CKD seems more of a comorbidity with less impact on mortality compared with HFrEF. 1 In HFrEF, kidney dysfunction may reflect disease progression due to reduced cardiac output and renal hypoperfusion, potentially promoting renal fibrosis and higher tubular DKK3 expression. 39 , 40 In support of this, a prior study suggested that eGFR is a stronger predictor of all‐cause mortality in HFrEF than in HFpEF, emphasizing the critical relevance of the cardiorenal interaction in these patients. 40 It is likely that a reduction in eGFR is a marker for reduced cardiac output, a more crucial prognostic factor in HFrEF than in HFpEF. 40 This might explain why DKK3 correlates with the deterioration of renal function after 1 year in HFrEF, in contrast to HFpEF and HFmrEF. The disease course in HFpEF is more likely influenced by coexisting comorbidities. 41 However, literature on mortality in HF shows conflicting results with respect to the impact of CKD on patients with different HF classes. 24 , 42 An individual patient data meta‐analysis (MAGGIC [Meta‐Analysis Global Group in Chronic Heart Failure]) found reduced eGFR to be a stronger predictor of outcomes in patients with HFrEF versus HFpEF. 42 This result was confirmed in the Swedish Heart Failure registry, where the association between CKD and mortality risk was less pronounced in patients with HFpEF. 1 HFrEF often involves ischemic conditions, leading to distinct renal impairment and fibrosis, which might increase DKK3 expression. 43 Since DKK3 is a stress‐induced profibrotic protein secreted by renal tubular epithelial cells, it is conceivable that HFrEF is more closely associated with elevated uDKK3 levels than HFpEF and HFmrEF. In contrast, kidney disease progression in HFpEF and HFmrEF may result from distinct pathophysiological mechanisms influenced more by comorbidities and involving different pathways apart from DKK3. Further evidence is needed to clarify these underlying processes. The present study has several limitations. This is a retrospective analysis of data that were derived from different multicenter HF cohort studies designed to identify biomarkers that may predict comorbidities and the clinical course of patients with HF with varying and partly incomplete follow‐up. Since there were differences between the HF cohort and the control group with respect to age, BMI, and sex distribution, appropriate adjustment was made for these variables within the statistical analyses. The present retrospective analyses yield intriguing findings; however, it is essential to validate these results to confirm their reliability and applicability in larger cohorts and to elucidate underlying mechanisms. The present analysis accounted for key clinical factors, including comorbidities such as diabetes and BMI, ensuring robust control of potential confounders in the relationship between uDKK3 and eGFR decline, although lifestyle factors (eg, diet and exercise) were not directly adjusted for. Despite the control group not having follow‐up data, their differences in baseline characteristics, such as a higher percentage of women, higher blood pressure, and higher baseline eGFR, could influence comparisons with the HF group. However, there were no significant correlations between uDKK3 levels and arterial hypertension, sex, or baseline eGFR (Table S1), suggesting that these imbalances did not influence uDKK3 levels. The lack of a correlation between uDKK3 and sex, baseline eGFR, left ventricular ejection fraction, or arterial hypertension suggests that it is not directly linked to baseline kidney or cardiac function. However, its association with BMI, diabetes, age, UACR, N‐terminal pro‐B‐type natriuretic peptide, angiotensin‐converting enzyme inhibitor use, and edema indicates that uDKK3 reflects metabolic dysfunction, renal stress, and HF severity. These findings suggest that uDKK3 may be more sensitive to dynamic, disease‐related factors and could serve as a marker for identifying patients at higher risk of adverse outcomes due to metabolic and kidney involvement, rather than baseline kidney or heart function alone. This supports the reliability of uDKK3 as a biomarker for kidney function decline in patients with HF, likely unaffected by sex, blood pressure, or initial eGFR, making it a sensitive marker for early kidney stress. CONCLUSIONS The present findings suggest that elevated uDKK3 levels are associated with an increased risk of eGFR decline in patients with HF in NYHA class I, particularly in the short term. Since uDKK3 levels were found to be associated with eGFR decline in patients with HF, uDKK3 might represent a link between heart and kidney dysfunction in cardiorenal syndrome. Therefore, uDKK3 may serve as a potential biomarker for predicting kidney function deterioration in patients with HF. Source of Funding This work was supported by the German Centre for Cardiovascular Research (DZHK). We acknowledge support by the Open Access Publication Funds/transformative agreements of the Göttingen University. Disclosures None. Supporting information Tables S1–S11 Figures S1–S4