Diagnostic and Prognostic Value of Aminoterminal Prohormone of Brain Natriuretic Peptide in Heart Failure with Mildly Reduced Ejection Fraction Stratified by the Degree of Renal Dysfunction

Limited data concerning the diagnostic and prognostic value of blood-derived biomarkers in heart failure with mildly reduced ejection fraction (HFmrEF) is available. This study investigates the diagnostic and prognostic value of aminoterminal prohormone of brain natriuretic peptide (NT-proBNP) in patients with HFmrEF, stratified by the estimated glomerular filtration rate (eGFR). Consecutive patients with HFmrEF were retrospectively included at one institution from 2016 to 2022. First, the diagnostic value of NT-proBNP for acute decompensated heart failure (ADHF) was tested. Thereafter, the prognostic value of NT-proBNP levels was tested for 30-months all-cause mortality in patients with ADHF. From a total of 755 patients hospitalized with HFmrEF, the rate of ADHF was 42%. Patients with ADHF revealed higher NT-proBNP levels compared to patients without (median 5394 pg/mL vs. 1655 pg/mL; p = 0.001). NT-proBNP was able to discriminate ADHF with an area under the curve (AUC) of 0.777 (p = 0.001), with the highest AUC in patients with eGFR ≥ 60 mL/min (AUC = 0.800; p = 0.001), and no diagnostic value was seen in eGFR < 30 mL/min (AUC = 0.576; p = 0.210). Patients with NT-proBNP levels > 3946 pg/mL were associated with higher rates of all-cause mortality at 30 months (57.7% vs. 34.4%; HR = 2.036; 95% CI 1.423–2.912; p = 0.001), even after multivariable adjustment (HR = 1.712; 95% CI 1.166–2.512; p = 0.006). In conclusion, increasing NT-proBNP levels predicted the risk of ADHF and all-cause mortality in patients with HFmrEF and preserved renal function; however, NT-proBNP levels were not predictive in patients with HFmrEF and eGFR < 30 mL/min.


Introduction
Although improvements in the management of coronary artery disease (CAD) and heart failure (HF) stabilized the incidence of HF over the past years, HF still affects about 64 million people worldwide, with a corresponding prevalence of 4% in the general population [1][2][3].Related to an increasing number of individuals with cardiac and non-cardiac comorbidities [4,5], related to the higher supply of invasive cardiac devices, higher rates of coronary revascularization, and cardiac pharmacotherapies, risk stratification for HF patients has even become more difficult and complex [6][7][8][9].Even multi-morbid patients have the highest risk of acute decompensated heart failure (ADHF), which, by now, represents one of the leading causes of hospitalization in the Western world [10].ADHF is characterized by an increased risk of cardiovascular mortality in patients with HF with reduced (i.e., HFrEF) and preserved (i.e., HfpEF) left ventricular ejection fraction (LVEF) [10].Recently, our study group demonstrated adverse long-term prognosis in patients with ADHF and HF with mildly reduced ejection fraction (HFmrEF) compared to patients without ADHF.The rate of ADHF was 22% in patients with HFmrEF [11].
By now, many biomarkers have been evaluated to identify individuals with ADHF; whereas, specifically, the measurement of brain natriuretic peptide (BNP) and aminoterminal prohormone of brain natriuretic peptide (NT-proBNP) was embedded into daily clinical practice [12][13][14][15][16][17].Although the predictive value of NT-proBNP levels may differ across the spectrum of HF stratified by LVEF, Savarese et al. recently demonstrated the discriminative capacity of NT-proBNP in patients with HFmrEF [18].A decline in NT-proBNP levels was associated with improved mortality rates [19].Contrarily, other studies concluded a limited prognostic impact of NT-proBNP values in patients with HFmrEF [20].In patients with HF, levels of NT-proBNP may further reflect patients' comorbidities; as such, a strong inverse correlation between renal function and NT-proBNP levels was demonstrated [20][21][22][23].Concerning the discriminatory capacity of NT-proBNP, the prognostic value of NT-proBNP levels was more pronounced in patients with preserved renal function among patients undergoing cardiac surgery; whereas, the prognostic value of NT-proBNP levels was poor in patients with impaired renal function [24].However, the diagnostic and prognostic value of NT-proBNP in patients with HFmrEF, stratified by renal function has never been investigated.
The present study sought to investigate (1) the diagnostic value of NT-proBNP levels to discriminate ADHF in patients with HFmrEF, as well as (2) the prognostic value of NT-proBNP levels in patients with ADHF, stratified by the presence and severity of concomitant renal dysfunction.

Study Patients, Design, and Data Collection
The aim of the present study was to evaluate the diagnostic capacity of NT-proBNP for ADHF.Additionally it was aimed to investigate the prognostic value of NT-proBNP for long-term all-cause mortality.All consecutive patients hospitalized with HFmrEF at one university medical centre were included from January 2016 to December 2022, as recently published [11].Using the electronic hospital information system, all relevant clinical data related to the index event were documented, such as baseline characteristics; vital signs on admission; prior medical history; prior medical treatment; length of index hospital and intensive care unit (ICU) stay; laboratory values; data derived from all non-invasive or invasive cardiac diagnostics and device therapies, such as echocardiographic data; coronary angiography and data being derived from prior or newly implanted cardiac devices.Every re-visit at the outpatient clinic or rehospitalization related to HF or adverse cardiac events was documented until the end of the year 2022.
The present study is derived from the "Heart Failure With Mildly Reduced Ejection Fraction Registry" (HARMER), representing a retrospective single-center registry including consecutive patients with HFmrEF hospitalized at the University Medical Centre Mannheim (UMM), Germany (clinicaltrials.govidentifier: NCT05603390).The registry was carried out according to the principles of the Declaration of Helsinki and was approved by the Medical Ethics Committee II of the Medical Faculty Mannheim, University of Heidelberg, Germany (ethical approval code: 2022-818).No written informed consent was deemed necessary for the present study.

Inclusion and Exclusion Criteria
All consecutive patients ≥18 years of age hospitalized with HFmrEF at one institution were included, irrespective of the department of hospital admission.The diagnosis of HFmrEF was determined according to the "2021 ESC Guidelines for the diagnosis and treatment of acute and chronic heart failure" [25].Accordingly, all patients with a LVEF of 41-49% and symptoms and/or signs of HF were included.The presence of elevated NT-proBNP levels and the evidence of structural heart disease were considered to make the diagnosis more likely but were not mandatory for diagnosis of HFmrEF.Transthoracic echocardiography was performed by cardiologists during routine clinical practice in accordance with current European guidelines [26,27].Echocardiographic operators were blinded to the final study analyses.Patients without measurements of NT-proBNP levels and/or estimated glomerular filtration rate (eGFR) during index hospitalization were excluded.No further exclusion criteria were applied.

Measurement of Creatinine, eGFR, and NT-proBNP Levels
Measurements of creatinine were carried out predominantly using lithium heparinate plasma.This assay is a modification of the Jaffé method with blank correction and axis segment adjustment.This blank correction is used to minimize interferences with bilirubin.The assay was carried out on a clinical chemistry analyzer (Atellica CH 930, Siemens Healthineers, Erlangen, Germany).A linear measurement range in plasma of 0.15 mg/dL (13 µmol/L) to 30.00 mg/dL (2652 µmol/L) was expandable to 60 mg/dL (5304 µmol/L) by automated dilution.The manufacturer specifies a limit of detection (LoD) of ≤0.10 mg/dL (≤9 µmol/L) and a limit of quantitation (LoQ) of ≤0.30 mg/dL.The reference range for healthy is given as 0.55-1.02mg/dL (49-90 µmol/L) for women and 0.70-1.30mg/dL (62-115 µmol/L) for men.Additionally, eGFR was estimated by using the CKD-EPI formula.This is more accurate compared to the MDRD formula in estimating the eGFR in the threshold region of beginning renal insufficiency [28,29].
Measurements of NT-proBNP were performed as a direct chemiluminescence sandwich immunoassay on the Atellica Solution IM (Siemens Healthineers, Erlangen Germany).The linear quantification range of the assay for serum and plasma is 35-35,000 pg/mL (4.13-4130 pmol/L).The clinical decision threshold for the NT-proBNP assay to separate healthy from sick patients is 125 pg/mL for patients aged < 75 years and 450 pg/mL for patients aged ≥75 years.

Study Endpoints
First, the diagnostic value of NT-proBNP for the diagnosis of ADHF was tested within the entire study cohort.ADHF was defined according to current European guidelines [25] based on congestion, characterized by the apparent worsening of clinical signs and/or symptoms of HF requiring intravenous diuretic therapy.
The prognostic impact of NT-proBNP levels was tested for long-term all-cause mortality at 30 months in patients with ADHF at index hospitalization.All-cause mortality was documented using the electronic hospital information system and by directly contacting state resident registration offices ('Bureau of Mortality Statistics').

Statistical Methods
Quantitative data was presented as the mean ± standard error of the mean (SEM), median and interquartile range (IQR), and ranges depending on the distribution of the data.They were compared using the Student's t-test for normally distributed data or the Mann-Whitney U test for non-parametric data.Deviations from a Gaussian distribution were tested by the Kolmogorov-Smirnov test.Qualitative data was presented as absolute and relative frequencies and were compared using the Chi-square test or the Fisher's exact test, as appropriate.
C-statistics were applied by calculating the receiver operating characteristic (ROC) curves and investigating the corresponding areas under the curves (AUCs) to assess (1) the diagnostic performance of NT-proBNP levels with regard to the diagnosis of ADHF during index hospitalization, as well as (2) the prognostic performance of NT-proBNP with regard to 30-month all-cause mortality in patients with ADHF.ROC analyses were performed within the entire study cohort, as well as stratification by eGFR, including patients with eGFR ≥ 60 mL/min, ≥30-<60 mL/min, and <30 mL/min.Optimal cut-offs were determined in accordance with the maximum Youden index.AUCs for the diagnostic and prognostic performance stratified by eGFR were compared by the method of Hanley et al. [30].Thereafter, Kaplan-Meier analyses were performed according to NT-proBNP levels based on the optimal cut-off and univariable hazard ratios (HRs) were given together with 95% confidence intervals.The prognostic impact of NT-proBNP levels was finally investigated within multivariable Cox regression models.
The results of all statistical tests were considered significant at p ≤ 0.05.SPSS (Version 28, IBM, Armonk, NY, USA) was used for statistics.

Study Population
A total of 2228 consecutive patients were hospitalized with HFmrEF from 2016 to 2022.In addition, 44 patients with incomplete follow-up, 1400 patients without measurements of NT-proBNP, and 29 patients without measurement of eGFR during index hospitalization were excluded.The final study cohort comprised 755 patients with HFmrEF (Figure 1; Flow chart).
test, as appropriate.
C-statistics were applied by calculating the receiver operating characteristic (ROC) curves and investigating the corresponding areas under the curves (AUCs) to assess (1) the diagnostic performance of NT-proBNP levels with regard to the diagnosis of ADHF during index hospitalization, as well as (2) the prognostic performance of NT-proBNP with regard to 30-month all-cause mortality in patients with ADHF.ROC analyses were performed within the entire study cohort, as well as stratification by eGFR, including patients with eGFR ≥ 60 mL/min, ≥30-<60 mL/min, and <30 mL/min.Optimal cut-offs were determined in accordance with the maximum Youden index.AUCs for the diagnostic and prognostic performance stratified by eGFR were compared by the method of Hanley et al. [30].Thereafter, Kaplan-Meier analyses were performed according to NT-proBNP levels based on the optimal cut-off and univariable hazard ratios (HRs) were given together with 95% confidence intervals.The prognostic impact of NT-proBNP levels was finally investigated within multivariable Cox regression models.
The results of all statistical tests were considered significant at p ≤ 0.05.SPSS (Version 28, IBM, Armonk, NY, USA) was used for statistics.

Correlations of NT-proBNP with Clinical, Echocardiographic, and Laboratory Data
In patients hospitalized with HFmrEF, NT-proBNP levels on admission correlated with age and body mass index, as well as with echocardiographic data, such as LVEF and TAPSE (Table 3).Among other laboratory values, inverse correlations with eGFR (r = −0.476;p = 0.001) and hemoglobin (r = −0.466;p = 0.001) were observed.
Figure 5. Kaplan-Meier analyses investigating the prognostic value of NT-proBNP levels regarding the risk of all-cause mortality at 30 months in all patients with ADHF, as well as stratified by patients with eGFR ≥ 60 mL/min, eGFR ≥ 30-<60, and eGFR.

Discussion
The present study aimed to investigate the diagnostic and prognostic value of NT-proBNP levels in a large retrospective cohort of consecutive patients hospitalized with HFmrEF.The main findings of this study can be summarized as follows: -NT-proBNP levels were higher in patients with ADHF as compared to patients without within the entire study cohort, as well as in patients with eGFR ≥ 30 mL/min.NT-proBNP levels did not differ in patients with ADHF vs. without ADHF and eGFR < 30 mL/min; -In line with this, NT-proBNP levels discriminated the presence of ADHF within the entire study cohort (AUC = 0.777); whereas, the diagnostic value of NT-proBNP was lower in patients with impaired renal function; -Furthermore, NT-proBNP levels predicted the risk of 30-months all-cause mortality in patients with HFmrEF and ADHF, especially in patients with preserved renal function and eGFR ≥ 60 mL/min.The prognostic impact of NT-proBNP was confirmed, even after multivariable adjustment.
In addition to clinical signs of congestion and patients' symptoms, the measurement of blood-derived biomarkers, especially the measurement of natriuretic peptides, such as BNP and NT-proBNP, has been embedded into daily clinical practice for the diagnostic decision making and treatment of HF.From a pathophysiological point of view, BNP causes diuresis and natriuresis and further leads to smooth muscle relaxation; whereas, NT-proBNP is physiologically inactive [31].By this point, the high diagnostic value of NT-proBNP with regard to the presence of ADHF was yet demonstrated both in patients with HFrEF and HFpEF, related to increased production and elevated plasma concentrations in patients with HF.For instance, Ibrahim et al. suggested NT-proBNP was useful in identifying patients with ADHF with an AUC of 0.926 in Asia and 0.866 in the Western world, including 1106 patients admitted to an emergency department with breathlessness [32].Of note, the diagnosis of ADHF may be improved when incorporating baseline characteristics, clinical characteristics, and the measurement of eGFR and hemoglobin in addition to the measurement of NT-proBNP-this approach, the so-called CoDE-HF decision support tool, was recently introduced, including 10,369 patients with suspected ADHF, and revealed an AUC of 0.846 in patients with previous HF and of 0.925 in patients without prior HF, respectively [14].The present study confirms the high diagnostic accuracy of NT-proBNP measurement for the diagnosis of ADHF in patients with HFmrEF, which was specifically observed in patients with preserved renal function.Other than the diagnostic value of impaired prognosis patients with HF [12,17,20,32,33] and atrial fibrillation [34,35]; whereas, heterogeneous findings concerning their prognostic impact in septic and cardiogenic shock was demonstrated [36][37][38].In line with this, Kang et al. demonstrated that patients with higher NT-proBNP levels had an increased risk of all-cause mortality and rehospitalization for worsening HF after 1 year, irrespective of the presence of HFrEF and HFpEF, including 1670 patients enrolled in the Korean Heart Failure registry [39].These findings were confirmed by Salah et al., suggesting a comparable prediction of all-cause mortality in HFrEF and HfpEF; whereas, specifically, a higher burden of comorbidities contributed to the prognosis of patients with HFpEF than HFrEF [33].This is of major importance since the number of comorbidities in individuals with HF is steadily increasing.
From this perspective, especially, the number of patients with HF and concomitant arterial hypertension, atrial fibrillation, and chronic kidney disease was shown to increase from 2001 to 2016 [5].In line with this, even 35% of patients with HFmrEF included in the present study suffered from concomitant chronic kidney disease.Thus, the diagnostic and prognostic value of NT-proBNP measurement was shown to be limited in patients with impaired renal function, especially in patients with eGFR < 30 mL/min.The current literature is characterized by heterogeneous findings concerning the prognostic value of NT-proBNP levels in patients with impaired renal function.For instance, Horii et al. suggested NT-proBNP was associated with reliable discrimination of all-cause mortality irrespective of renal function; whereas, NT-proBNP was associated with better discrimination of allcause mortality compared to BNP in patients with chronic kidney disease stages 4-5 [40].In line, NT-proBNP levels predicted the risk of mortality among 341 patients with congestive HF, irrespective of the presence or absence of chronic kidney disease [41].In contrast, lower prognostic accuracy with regard to all-cause mortality was observed in patients with chronic kidney disease stage 3b (AUC = 0.616) as compared to patients with stage 3a (AUC = 0.697), including 168 patients of at least 80 years of age [42].Poor prediction of all-cause mortality in patients with advanced stages of chronic kidney disease may be attributed to the very high risk of all-cause mortality related to chronic kidney disease itself (i.e., at least 50% within the present study in patients with eGFR < 30 mL/min at 30 months).Thus, the presence of chronic kidney disease was recently shown to increase the risk of both ventricular tachyarrhythmias and sudden cardiac death [43,44].Further studies are, therefore, necessary to investigate the prognostic role of NT-proBNP in patients with advanced stages of chronic kidney disease with regard to the risk of HF-related mortality and to separate its impact on cardiovascular death.
Within the present study, NT-proBNP levels were associated with prognosis despite a high rate of patients with an optimal pharmacological treatment, including beta-blockers and inhibitors of the renin-angiotensin-aldosterone system.However, within the present study, specifically, the proportion of patients treated with a sodium-glucose-linked transporter 2 (SGLT2) inhibitor was rather low and only 6.7% of patients without ADHF and 4.7% with ADHF were treated with a SGLT2 inhibitor.The low prescription rates of SGLT2 inhibitors are in line with a previous study from the Swedish HF registry; whereas, only 5.5% of patients with concomitant diabetes mellitus were treated with a SGLT2 inhibitor from 2016 to 2018 [45].This may be in accordance with the upgrade within the updated ESC HF guidelines; whereas, specifically, treatment with SGLT2 inhibitors gained more importance in 2023 [46].From this perspective, the prognostic impact of SGLT2 inhibitors may be superior in patients with increased NT-proBNP levels and may decrease NT-proBNP levels in HF patients [47].However, further studies are needed in patients with HFmrEF receiving an optimal HF pharmacotherapy, including treatment with SGLT2 inhibitors after being embedded into daily clinical practice.
The findings of the present study, suggesting the superior diagnostic and prognostic value of NT-proBNP in patients with HFmrEF and preserved renal function, are of the utmost importance given ongoing demographic changes and the aging of the population.In line with this, the burden of cardiovascular and non-cardiovascular comorbidities was demonstrated to increase, leading to a higher of patients with chronic kidney disease [5,48].Specifically in patients with advanced stages of chronic kidney disease, both the diagnostic and prognostic accuracy of NT-proBNP were limited within the present all-comer study, including patients hospitalized with HFmrEF.Furthermore, the limited diagnostic capacity of NT-proBNP was not yet demonstrated in patients with obesity and atrial fibrillation [49].Given these findings, further studies are warranted to improve risk stratification for patients with HF and chronic kidney disease.From this perspective, the combined assessment of biomarkers may improve risk stratification.The measurement of soluble (s)ST2 was recently shown to improve the prognostic accuracy when combined with NT-proBNP measurement [50].Especially in patients with chronic kidney disease, sST2 was shown to be associated with the risk of developing HF, including 3314 patients [51].Therefore, further studies are needed to evaluate the diagnostic and prognostic accuracy of sST2 in patients with HF and chronic kidney disease.

Study Limitations
This study has several limitations.Due to the retrospective and single-center study design, results may be influenced by measured and unmeasured confounding variables.NT-proBNP was not measured in a large proportion of patients, which may bias the conclusions diagnostically, especially for the non-ADHF group.In line with this, serial measurements of NT-proBNP levels were only available in a minor part of the study population and, therefore, not included.No information on eGFR beyond index hospitalization was available.The proportion of patients with eGFR < 30 mL/min was rather low within the present study.In addition, changes in LVEF during the course of follow-up were available in a minor part of the study population and, therefore, were beyond the scope of the present study.Patients with ambulatory visits only were not included in the present study.This may further impact the findings of the present study by presumably including a higher proportion of sicker patients.Furthermore, no sub-analyses were performed further stratifying by the patients' age, sex, body mass index, or atrial fibrillation, which may further affect NT-proBNP levels.Finally, causes of death beyond index hospitalization were not available for the present study.

Conclusions
NT-proBNP was associated with reliable diagnostic and prognostic value and independently predicted the risk of 30-months all-cause mortality, especially in patients with preserved renal function.In contrast, NT-proBNP revealed neither diagnostic nor prognostic value in patients with eGFR < 30 mL/min.Informed Consent Statement: Not applicable.

Data Availability Statement:
The datasets used and/or analyzed during the current study are available from the corresponding author upon reasonable request.

Figure 1 .
Figure 1.Flow chart of the study population.

Figure 1 .
Figure 1.Flow chart of the study population.

Figure 2 .
Figure 2. Box plots demonstrating the distribution of NT-pro BNP levels among patients with HFmrEF stratified by patients with ADHF vs. without ADHF.Data are presented as the median with interquartile ranges (boxes) and 5-95% percentiles (whiskers).

Figure 2 .
Figure 2. Box plots demonstrating the distribution of NT-pro BNP levels among patients with HFmrEF stratified by patients with ADHF vs. without ADHF.Data are presented as the median with interquartile ranges (boxes) and 5-95% percentiles (whiskers).

Figure 3 .
Figure 3. Receiver operator characteristic (ROC) curve analyses for the discrimination of ADHF from non-ADHF within the entire study cohort, as well as stratification by patients with eGFR ≥ 60 mL/min, eGFR ≥ 30-<60, and eGFR.

Figure 3 .
Figure 3. Receiver operator characteristic (ROC) curve analyses for the discrimination of ADHF from non-ADHF within the entire study cohort, as well as stratification by patients with eGFR ≥ 60 mL/min, eGFR ≥ 30-<60, and eGFR.

Figure 4 .
Figure 4. Receiver operator characteristic (ROC) curve analyses for the discrimination of 30-months all-cause mortality in patients with ADHF, as well as stratified by patients with eGFR ≥ 60 mL/min, eGFR ≥ 30-<60, and eGFR.

Figure 4 .
Figure 4. Receiver operator characteristic (ROC) curve analyses for the discrimination of 30-months all-cause mortality in patients with ADHF, as well as stratified by patients with eGFR ≥ 60 mL/min, eGFR ≥ 30-<60, and eGFR.

Figure 5 .
Figure 5. Kaplan-Meier analyses investigating the prognostic value of NT-proBNP levels regarding the risk of all-cause mortality at 30 months in all patients with ADHF, as well as stratified by patients with eGFR ≥ 60 mL/min, eGFR ≥ 30-<60, and eGFR.

Table 2 .
Heart-failure-related and procedural data.

Table 4 .
Multivariable Cox analyses with regard to 30-month all-cause mortality in patients with ADHF and HFmrEF., confidence interval; eGFR, estimated glomerular filtration rate; HR, hazard ratio; NT-pro BNP, aminoterminal pro-B-type natriuretic peptide.Level of significance p < 0.05.Bold type indicates statistical significance.* Model 2: multivariable Cox regression analyses were additionally performed and stratified by eGFR and HR; additionally, 95% CI and p-values were provided for NT-proBNP > 3946 pg/mL. CI