Plasma B-type natriuretic peptide is independently associated with cardiovascular events and mortality in patients with chronic kidney disease

The association between B-type natriuretic peptide (BNP) and cardiovascular (CV) events and mortality has not been well characterized in patients with chronic kidney disease (CKD). We prospectively investigated whether BNP was associated with CV events or mortality beyond cardiac alterations in 1078 patients with CKD. Participants were divided into the following 3 groups according to circulating BNP concentration: < 40 pg/mL, low; 40–100 pg/mL, middle; and > 100 pg/mL, high. Primary outcome was fatal or nonfatal CV events, and alternative outcome was a composite of fatal or nonfatal CV events, or non-CV deaths. During a median follow-up of 2.6 years, CV and composite events occurred in 158 and 248 participants, respectively. Cox analyses after adjustment for covariates, including cardiac parameters, showed that the hazard ratios (HRs) (95% confidence intervals [CIs]) for CV events of middle and high groups were 1.00 (0.63, 1.58) and 1.72 (1.06, 2.79), respectively, compared with low group. Additionally, similar results were obtained for composite events; the HRs (95% CIs) of middle and high groups were 1.10 (0.77, 1.57) and 1.54 (1.04, 2.27), respectively, compared with low group. Thus, in CKD, high BNP concentrations were independently associated with CV events and mortality, independent of cardiac alterations.


Subgroup analyses regarding the associations of log BNP levels with CV and composite events
The adjusted HRs for CV events and the composite events per 1-log unit increment in log BNP in subgroups stratified by demographic and clinical characteristics are shown in Tables 4 and 5, respectively.No significant interactions for those outcomes were identified (P for interactions, 0.06-0.86).www.nature.com/scientificreports/

Associations of BNP with CV and composite events in participants with LVEF ≥ 45% (sensitivity analyses)
We excluded 27 participants with LVEF < 45% who therefore may have had CHF, according to a previous study 25 .
In the remaining 1051 participants, multivariable Cox analyses found that participants with high BNP levels had increased risk of both CV and composite events (Supplementary Table 2).

Results of receiver operating characteristic (ROC) curve analyses for CV and composite events
The ROC curves for CV events (Supplementary Fig. 1) and composite events (Supplementary Fig. 2) were modelled according to stages 1-3 and 4-5 CKD.The area under the curve (AUC) for BNP with respect to CV events in participants with stages 1-3 and 4-5 CKD were 0.709 and 0.637, respectively (Supplementary Fig. 1A,B).AUC for BNP with respect to composite events in participants with stages 1-3 and 4-5 CKD were 0.733 and 0.640, respectively (Supplementary Fig. 2A,B).The cutoff values of BNP for CV events in participants with stages 1-3

Baseline characteristics and HRs for CV and composite events of BNP levels after inverse probability of treatment weighting (IPTW)
Baseline characteristics after IPTW are shown in Supplementary Table 3.Additionally, the c-statistics between two groups based on propensity score were calculated as follows: low BNP vs. middle BNP, 0.904; low BNP vs. high BNP, 0.756; and middle BNP vs. high BNP, 0.747.Supplementary Table 4 shows the HRs for CV and composite events by BNP level after IPTW.Participants with high BNP levels had higher CV and composite

Discussion
In the present study, we investigated whether high BNP concentrations are associated with CV events and mortality, independent of cardiac structure and function, in patients with CKD.Independent associations of log BNP with cardiac parameters (LAD, LVMI, and LVEF) were identified using multiple linear regression analysis, and even if these parameters were included in multivariable Cox analyses, high BNP concentration was identified as an independent risk factor for adverse outcomes (fatal or nonfatal CV events, and the composite event of fatal or nonfatal CV events, or non-CV death).Furthermore, subgroup analyses showed that baseline covariates such as prior CVDs, eGFR, LVEF, LAD, and LVMI did not significantly affect the associations between BNP and the two outcomes.Alternatively, we examined the effect of high BNP levels in combination with the prior presence of CVDs on poor outcomes.It was confirmed that the combination had strongest predictive power for adverse events.Furthermore, the associations of higher BNP levels with adverse outcomes in multivariable analyses after IPTW were similar to the findings obtained before IPTW.
In the heart-kidney hemodynamic model, the kidneys control the extracellular fluid volume by regulating sodium excretion and reabsorption, and the heart controls systemic hemodynamics.When one of these organs fails, a vicious circle develops in which the RAAS, the nitric oxide-reactive oxygen species balance, the sympathetic nervous system, and inflammation interact and synergize, thereby leading to an increase in morbidity and mortality 26,27 .Cardiorenal syndrome (CRS) is defined as a pathophysiologic disorder of the heart and kidneys, whereby acute or chronic dysfunction in one organ induces acute or chronic dysfunction in the other.Five subtypes of this syndrome have also been defined: type 1, acute cardio-renal; type 2, chronic cardio-renal; type 3, acute reno-cardiac; type 4, chronic reno-cardiac; and type 5, secondary CRS.In CRS type 2, chronic cardiac abnormalities result in kidney injury or dysfunction.In contrast, in CRS type 4, CKD contributes to a decrease in cardiac function, cardiac hypertrophy, and/or an increase in the risk of adverse CV events 28,29 .To date, several studies have demonstrated that poor kidney function or a decline in function are associated with higher risks of CVD and mortality [30][31][32] .Cardiac biomarkers, such as BNP and NT-proBNP, have shown prognostic value in patients at various stages of kidney dysfunction, and therefore have potential applications in patients with CRS types 2 or 4 28 .
Several previous studies have documented associations of BNP/NT-proBNP with kidney function or cardiac alterations.A previous study demonstrated that BNP concentrations are less affected than those of NT-proBNP by kidney dysfunction 33 .Vickery et al. reported that LVMI is independently associated with both BNP and NT-proBNP, whereas eGFR had an independent effect on the BNP, and especially on the NT-proBNP, concentrations in patients with CKD 24 .BNP concentration has also been reported to be independently associated with left ventricular overload, but not with kidney function 14 .Furthermore, Tagore et al. reported that there is an independent inverse correlation between eGFR and NT-proBNP, but not BNP, in patients with stages 3 or 4 CKD 23 .In the present study, simple linear regression analysis revealed a significant inverse association between log BNP and eGFR, but multiple linear regression analysis did not.In addition, in multiple analysis, independent associations between log BNP and cardiac parameters were identified, as in previous studies 24 .These findings suggest that in patients with CKD, BNP concentration may be influenced by abnormalities in cardiac structure and function, rather than kidney dysfunction.However, the previous studies that characterized the associations of BNP with CV outcomes and mortality in patients with CKD did not include cardiac parameters as covariates [13][14][15] ; therefore, the fact that significant associations of BNP concentrations with adverse outcomes identified in the present study were independent of indices of cardiac structure and function provides valuable additional information.
Volume overload is not uncommon in patients with CKD and is associated with both traditional and nontraditional CV risk factors 34 .A previous study demonstrated that volume overload, assessed using bioelectric impedance spectroscopy, is an independent risk factor for CV morbidity and all-cause mortality in patients with advanced CKD stages 35 .Furthermore, a previous systematic review demonstrated that volume overload was associated with CHF events 36 .In the present study, the volume status of the participants was not assessed using a body composition monitor, but their LADs were assessed using echocardiography, and log BNP was found to be independently and positively associated with LAD.It has previously been reported that LAD has a significant positive correlation with estimated plasma volume 37 , and BNP has been reported to be a biomarker of volume status in patients with CKD 38 .Therefore, in the present cohort, participants with high BNP concentrations may have had more severe volume overload.Furthermore, the present study found a significant association of BNP with CHF events, consistent with other several studies [10][11][12]14 . Givn these findings, the implied association between high BNP and volume overload might explain the link between high BNP concentration and adverse outcomes, especially CHF events.
BNP concentration may increase in response to myocardial ischemia, even if this is not accompanied by heart failure 39,40 .Therefore, it may have value for the identification of subclinical cardiac disease in patients who do not Vol:.(1234567890 41 .Subclinical myocardial injury has also been reported to be associated with higher risks of CVD and all-cause mortality in individuals who do not have CVD 42 .In a previous study, BNP was also reported to be associated with nonfatal myocardial infarction in patients with stable coronary heart disease 43 .Therefore, high BNP-related subclinical myocardial injury may be explained by the association between BNP and IHD events.However, like previous large cohort studies [10][11][12] , the present study showed no significant association between BNP and IHD events.Furthermore, findings on the association of BNP and stroke are also conflicting. A previous large cohort study 12 reported a significant association between BNP and stroke; conversely, two large cohort studies 10,11 and the present study did not.Given these findings, it remains unclear whether higher BNP levels may contribute to subsequent CV events such as IHD or stroke, and further investigations are warranted to determine the association of BNP with these CV events.The present study had several limitations.First, all the participants were recruited at a single regional hospital; therefore, the sample was fairly homogeneous and subject to selection bias.Second, we recruited consecutive patients who were admitted to the hospital; they were relatively old, were all Japanese, and the number of male participants was approximately 1.8-times higher than that of female participants.Third, we did not measure the concentrations of cardiac troponin I (cTnI) which is a sensitive and specific marker of myocardial injury 44 .Higher cTnI was also reported to be associated with a large spectrum of incident CV events in the community-based cohort 45 .Furthermore, it was demonstrated that combined measurements of cTnI and BNP were more reliable predictors of increased CV events in patients with hypertrophic cardiomyopathy 46 .Therefore, simultaneous measurements of these cardiac biomarkers could provide more novel information regarding the risk stratification for subsequent CV events.Fourth, we did not evaluate the New York Heart Association function class for each participant.We performed sensitivity analyses in participants with LVEF ≥ 45%; however, it remained unclear whether these participants did or did not have CHF.Fifth, the numbers of CV events by category (i.e., CHF, IHD, stroke, and other CV events) were relatively small.Therefore, statistical power may have been low when performing multivariable Cox analyses.Sixth, although the maximum absolute standardized difference (ASD) of each covariate after IPTW was lower compared with that calculated before applying IPTW, maximum ASDs in all covariates after IPTW were not less than 0.1 (as shown in Supplementary Table 5).An ASD of < 0.1 has been taken to indicate a negligible difference in the mean or prevalence of a covariate between treatment groups 47 .Therefore, the patient population after IPTW may not be well-balanced across all covariates.In clinical epidemiology, a very high c-statistic value can indicate considerable non-overlap in propensity score distribution between exposed and unexposed subjects 48 .Furthermore, it is possible that IPTW may perform poorly when the treatment groups are initially very different and when some patients have extreme propensity scores near 1 or 0 49 .In the present study, the c-statistics were also calculated between each two-group comparison.In particular, the c-statistic between the low and middle BNP groups was > 0.9, suggesting very high non-overlap between the two groups.Accordingly, this may indicate unstable estimation via IPTW.Finally, we only used a single BNP measurement, which may not be a highly accurate predictor of adverse outcomes.In optimally treated patients with chronic heart failure, the use of a BNP-guided strategy has been reported to be associated with a lower risk of chronic heart failure-related death or hospitalization because of chronic heart failure than the use of a conventional strategy based on clinical expertise 50 .In addition, a recent study demonstrated that BNP monitoring is associated with a lower risk of a requirement for kidney replacement therapy in patients with CKD who are not on dialysis 51 .Therefore, further studies are warranted to determine whether longitudinal BNP monitoring lower the risks of CV events and mortality in patients with CKD.
In conclusion, in patients with CKD, high BNP concentrations are associated with adverse outcomes, such as CV events and mortality, independent of confounding variables such as cardiac parameters.These findings suggest that BNP is a useful biomarker of the risk associated with adverse outcomes in patients with CKD.

Study design
Between June 2009 and November 2022, 1279 consecutive Japanese patients who were admitted to the NHO Kyushu Medical Center for the evaluation of, and education regarding, CKD were selected.Of these, we excluded 61 who showed acute-on-chronic kidney injury, 7 who lacked BNP data, and 5 who did not undergo echocardiography.The remaining 1206 patients were discharged from hospital without initiating kidney replacement therapy and were subsequently followed up at the same hospital.Of these, 60 who were lost to follow-up within 6 months of discharge and 68 who initiated kidney replacement therapy within the same period were also excluded.Therefore, data for 1078 patients that were collected up to June 2023 were prospectively analyzed.
The study was approved by the Ethics Committee of the NHO Kyushu Medical Center (approval number: 09-09), registered with the University Hospital Medical Information Network (UMIN000017519), and performed in accordance with the guidelines of the Declaration of Helsinki.Written informed consent was obtained from all the participants.

Outcome definitions
The primary endpoints were fatal or nonfatal CV events in the absence of kidney replacement therapy, which were defined as follows: IHD (those requiring percutaneous intervention or coronary artery bypass grafting, acute myocardial infarction, and myocardial ischemia identified using myocardial scintigraphy), hospitalization for the treatment of CHF, stroke (brain infarction, brain hemorrhage, subarachnoid hemorrhage, and non-traumatic acute subdural hematoma), requirement for interventions to treat peripheral artery disease (percutaneous transcatheter angioplasty, lower-limb amputation, and bypass surgery), dissecting aneurysm of the thoracic and/or abdominal aorta, rupture of a thoracic and/or abdominal aortic aneurysm, requirement for bypass or stent placement in the thoracic or abdominal aortic aneurysm, or iliac artery aneurysm, requirement for stent www.nature.com/scientificreports/placement for the treatment of internal carotid or vertebral artery stenosis, cardiac valvular diseases (sudden onset of severe aortic regurgitation, valve replacement surgery, or transcatheter aortic valve implantation for the treatment of aortic stenosis), and sudden death.The alternative outcome was a composite of fatal or nonfatal CV events, or non-CV deaths, without the initiation of kidney replacement therapy.Non-CV death was defined as death that occurred in the absence of a CV event.The follow-up period was defined as the period between baseline and a first event in participants who experienced events, or as the time to the completion of the study or loss to follow-up in participants who were censored.

Data collection
Blood samples (serum creatinine, serum phosphorus, hemoglobin, serum albumin, CRP, and BNP) were obtained from participants early in the morning following an overnight fast on the second day of admission.Daily proteinuria was also measured.eGFR (mL/min/1.73m 2 ) was calculated using the following new Japanese equation: eGFR = 194 × SCr −1.094 × age −0.287 (× 0.739 if female), where SCr is the serum creatinine concentration of the participant 52 .All the participants were interviewed regarding their medical history, including hypertension, diabetes mellitus, prior CVDs, and malignancy.The demographic information (age and sex), medication history, and history of smoking at presentation were recorded for each participant.Prior CVDs were defined as a history of IHD, CHF, stroke, peripheral artery disease, thoracic and/or abdominal aortic aneurysm, and/or aortic dissection.Peripheral artery disease was defined as having a low ankle-brachial blood pressure index (< 0.9) or having undergone treatment for lower limb ischemia.Participants were also categorized according to their cigarette smoking status as current or past smokers.BMI was calculated as body mass in kg divided by height in m, squared.Blood pressure was measured on three separate occasions on day 2 of hospitalization, with the participants in a sitting position, and the mean of the three values obtained was recorded.

Assessment of cardiac structure and function
Left ventricular mass (LVM) was calculated using M-mode data obtained from parasternal long-axis images, according to the following formula 53 : LVM = 1.04 ([IVSd + LVPWd + LVDd] 3 − LVDd 3 ) − 13.6, where IVSd and LVPWd are the thicknesses of the interventricular septum and the posterior wall of the LV during diastole, respectively, and LVDd is the diameter of the LV during diastole.LVMI was expressed as LVM per square meter of body surface area, calculated using the Du Bois formula 54 as body mass 0.425 × height 0.725 × 0.007184.

Statistical analyses
Continuous data are expressed as the median (interquartile range) because Shapiro-Wilk testing indicated that none of the continuous variables were normally distributed.The participants were divided into the following three categories based on their BNP concentration, according to the JCS 2017/JHFS 2017 Guidelines on Diagnosis and Treatment of Acute and Chronic Heart Failure 55 : BNP levels < 40 pg/mL, low group; 40-100 pg/mL, middle group; and > 100 pg/mL, high group.The BNP concentration was non-normally distributed; therefore, it was log-transformed to achieve an approximately normal distribution prior to statistical analysis.Linear regression analysis was performed to identify variables associated with log BNP.Multiple linear regression analyses were adjusted for all variables used in simple linear regression analyses to determine independent variables for log BNP.Survival curves were constructed using the Kaplan-Meier method and evaluated using the log-rank test.We selected traditional CV risk factors (age, sex, smoking, diabetes mellitus, systolic blood pressure, dyslipidemia, BMI, and prior CVDs); non-traditional CV risk factors (hemoglobin, CRP, serum albumin, and eGFR); the presence of malignancy, which affects mortality; and cardiac parameters (LAD, LVEF, and LVMI, which are closely related to BNP concentration) as covariates.Common variables associated with both univariable linear regression for log BNP and Cox analyses for outcomes were considered confounding factors, and the above covariates-with the exceptions of dyslipidemia, malignancy, daily proteinuria, and BMI-were confirmed to be confounding factors.Cox proportional hazards models were used to determine whether BNP was associated with CV and the composite events, and HRs and 95% confidence intervals were calculated for each variable.Subgroup analyses were performed according to sex, the presence or absence of categorical variables, and the status of continuous data (values below or above the median value).The effects of interactions between log BNP and other variables on the outcomes were evaluated by adding interaction terms for the associations between log BNP and other variables to the relevant model.ROC curve analyses were used to evaluate the diagnostic performance of BNP for the prediction of CV and composite events, and the sensitivity and specificity of BNP to predict CV and composite events were also determined, according to stages 1-3 and 4-5 CKD.The propensity scores were evaluated using multivariable logistic regression models that included the following variables: age, sex, smoking, diabetes mellitus, dyslipidemia, prior CVDs, malignancy, systolic blood pressure, BMI, CRP, hemoglobin, serum albumin, eGFR, LVEF, LAD, and LVMI.Inverse probability of treatment weighting (IPTW) was calculated to balance the covariates among three groups, namely: low, middle, and high BNP groups 56 .We used Stata's streg command, which fits proportional hazard models with a Weibull distribution, to evaluate the association of BNP levels with CV and composite events 57 .Statistical analyses were performed using STATA version 15 (Stata Corp., College Station, TX, USA), and P < 0.05 was taken to indicate statistical significance.

Figure 1 .
Figure 1.Kaplan-Meier curves for the absence of CV events (A) and the composite events (B) in participants stratified according to their BNP concentration, compared using the log-rank test.CV cardiovascular, BNP B-type natriuretic peptide. https://doi.org/10.1038/s41598-024-67529-1

Table 1 .
Baseline clinical characteristics of participants according to BNP levels.Values are expressed as number (percent) or median (interquartile range).BNP B-type natriuretic peptide, CVD cardiovascular disease, IHD ischemic heart disease, CHF congestive heart failure, PAD peripheral artery disease, RAAS reninangiotensin-aldosterone system, BMI body mass index, CRP C-reactive protein, CRP C-reactive protein, eGFR estimated glomerular filtration rate, LVEF left ventricular ejection fraction, LAD left atrial diameter, LVMI left ventricular mass index.

Table 3 .
Hazard ratios for CV and composite events of BNP levels. No.

Table 5 .
Adjusted hazard ratios for composite events per 1-log unit increment in log BNP among subgroups stratified by clinical parameters.CV cardiovascular, BNP B-type natriuretic peptide, HR hazard ratio, CI confidence interval, CVD cardiovascular disease, SBP systolic blood pressure, BMI body mass index, CRP C-reactive protein, eGFR estimated glomerular filtration rate, LVEF left ventricular ejection fraction, LAD left atrial diameter, LVMI left ventricular mass index.events compared with those with low BNP levels.Additionally, log BNP levels were independently associated with both outcomes.