Introduction
Heart failure (HF) is an important medical and economic problem associated with high morbidity and mortality rates worldwide. Accurate risk assessment in patients with HF allows for more personalized follow-up and prediction of a patient’s possible response to advanced therapeutic interventions, such as heart transplant (HT) or mechanical circulatory support (MCS). In clinical practice, accurate risk estimation in HF patients is challenging due to a difficulty in assessing the relative importance of each prognostic parameter, and also because of physicians’ personal beliefs or their previous experiences.1-3 Considering heterogeneity of the HF population, combining clinical and laboratory data with novel biomarkers appears to be the best approach to developing new predictive tools.4-6
Although numerous prognostic scores have been developed, only the Heart Failure Survival Score (HFSS) and the Seattle Heart Failure Model (SHFM) are commonly used in clinical practice, and are recommended for stratifying the risk of death in patients with end-stage HF who are eligible for HT.7 Notably, HFSS and SHFM were derived from patient data beginning in the late 1990s and early 2000s, and these stratification models do not adequately account for changes in pharmacologic and device-based HF therapy over time. Moreover, they do not consider the role of novel biomarkers of HF.8-10 Therefore, new models derived from data of contemporary HF patients may be needed to improve prognostic assessment.4
The novel Barcelona Bio-Heart Failure (BCN Bio-HF) risk calculator consists of 11 clinical variables and 3 serum biomarkers (N-terminal pro–B-type natriuretic peptide [NT-proBNP], high-sensitivity soluble suppression of tumorigenicity 2 [sST2], and high-sensitivity cardiac troponin T [cTnT]).6 Both cTnT and natriuretic peptides are well-established biomarkers of HF,10 but in recent years, numerous studies have highlighted the importance of sST2 in the assessment of patients with HF.5,10 sST2 is a member of the interleukin-1 receptor family, and is released in response to vascular congestion as well as inflammatory and profibrotic stimuli. This biomarker reflects myocardial stress, ventricular remodeling, and fibrosis, and is a strong, independent predictor of poor outcomes and hospitalization in HF patients.5,10,11
Usefulness of the BCN Bio-HF risk calculator in end-stage HF has not been well established. The present study evaluated the predictive performance of this new risk score, as compared with the traditional scores (HFSS and SHFM), in patients with end-stage HF during 1-year follow-up. We also aimed to identify other potential risk factors associated with worse prognosis in the analyzed population.
Patients and methods
We prospectively analyzed 314 patients with end-stage HF awaiting HT who were hospitalized in our cardiology department between 2018 and 2021. The patients who underwent HT during the follow-up period (n = 35) were excluded from the analysis. The study end point was death from any cause within 1 year of follow-up. All patients received standard medical treatment, including angiotensin-converting enzyme inhibitors (ACEIs) / angiotensin II receptor blockers (ARBs) / angiotensin receptor–neprilysin inhibitors (ARNIs), mineralocorticoid receptor antagonists (MRAs), loop diuretics, and β-blockers at the maximum tolerated doses, according to the European Society of Cardiology guidelines for the diagnosis and treatment of acute and chronic HF.12,13 Fatal events were identified based on medical documentation and by means of telephone calls with the patients’ families.
This study followed the principles outlined in the Declaration of Helsinki, and its protocol was approved by the Local Bioethics Committee at the Medical University of Silesia in Katowice (PCN/0022/KB/159/20, PCN/0022/KB1/20/I/2, and PCN/0022/KB1/69/I/19). Written informed consent was obtained from all patients upon enrollment.
Scores
The HFSS was calculated according to a formula described by Aaronson et al8: HFSS = ([0.0216 × resting heart rhythm] + [−0.0255 × mean arterial blood pressure] + [−0.0464 × left ventricular ejection fraction (LVEF)] + [−0.047 × serum sodium] + [−0.0546 × peak oxygen uptake] + [0.6083 × presence (1) or absence (0) of interventricular conduction defect (QRS duration ≥0.12 s due to any cause)] + [0.6931 × presence (1) or absence (0) of ischemic cardiomyopathy]).
The SHFM score for each patient was determined as described by Levy et al.9 SHFM includes 10 continuous variables (age, LVEF, New York Heart Association [NYHA] class, systolic blood pressure, diuretic dose adjusted for weight, lymphocyte count, hemoglobin, serum sodium, total cholesterol, and uric acid) and 10 categorical variables (sex, ischemic cardiomyopathy, device-based therapy [implantable cardioverter-defibrillator, cardiac resynchronization therapy device], and treatment with β-blockers, ACEIs, ARBs, potassium-sparing diuretics, statins, or allopurinol) in an equation that yields a continuous risk score for each patient.
The BCN Bio-HF calculator incorporates 11 clinical variables (age, sex, NYHA class, LVEF, serum sodium, estimated glomerular filtration rate, hemoglobin, loop diuretic dose, β-blocker use, ACEI / ARB use, and statin treatment) and 3 serum biomarkers (NT-proBNP, high-sensitivity cTnT, and high-sensitivity sST2).6 According to Lupón et al,6 the score may be calculated with the availability of none, 1, 2, or 3 biomarkers. In our study, we used a formula that included the clinical variables and 2 biomarkers, NT-proBNP and high-sensitivity sST2. The BCN Bio-HF score was calculated as described previously.6
Laboratory testing
Peripheral venous blood samples were collected after 12 hours of fasting at the time of inclusion in the study. Hematologic parameters were measured with automatic blood cell counters (Sysmex XS1000i and XE2100; Sysmex Corporation, Kobe, Japan). Total bilirubin, creatinine, uric acid, alanine aminotransferase, aspartate aminotransferase, and cholesterol concentrations were determined with a Cobas Integra 800 analyzer (Roche Instrument Center AG, Rotkreuz, Switzerland). Serum fibrinogen levels were measured using an STA Compact analyzer (Roche Instrument Center AG, Rotkreuz, Switzerland). Plasma high-sensitivity C-reactive protein (hs-CRP) level was measured with a Cobas Integra 70 analyzer (Roche Diagnostics, Ltd., Mannheim, Germany). The detection limit for hs-CRP was 0.0175 mg/dl. Plasma level of NT-proBNP was measured with a commercially available kit from Roche Diagnostics on an Elecsys 2010 analyzer with analytical sensitivity below 5 pg/ml.
Human sST2 was measured using a sandwich enzyme-linked immunosorbent assay (ELISA) with a commercially available kit (Human ST-2 ELISA, SunRedBio Technology Co., Ltd., Shanghai, China). Sensitivity of the assay was 0.436 ng/ml and the assay range was 0.5–150 ng/ml. The ELISA was performed using a BioTek Elx50 reader (BioTek Instruments, Inc., Tecan Group, Männedorf, Switzerland).
Statistical analysis
Statistical analysis was performed with SAS software, version 9.4 (SAS Institute Inc., Cary, North Carolina, United States). Based on the Shapiro–Wilk test, continuous variables did not follow a normal distribution, and are expressed as median (interquartile range [IQR]). Differences between these variables were analyzed using the Mann‒Whitney test. Categorical variables are expressed as number (percentage) and were compared using the χ2 test. Receiver operating characteristic (ROC) curves were plotted, and the Youden index was used to determine the cutoff points for the analyzed parameters. The prognostic utility of the various scales and the NT-proBNP concentration was evaluated by the area under the ROC curve (AUC), as well as by their sensitivity, specificity, negative predictive value, positive predictive value, negative likelihood ratio, positive likelihood ratio, and accuracy. The AUC was calculated for death as a binary outcome, using a logistic regression model. We also performed the Hosmer–Lemeshow goodness of fit test. The DeLong test was used to quantitatively compare the ROC curves, while the Hanley and McNeil methods were used to compare differences between AUC values. An AUC greater than 0.7 was considered clinically relevant.
The Cox univariable proportional analysis was used to select predictors of death for inclusion in the multivariable analysis. The examined factors included 3 scores (HFSS, SHFM, and BCN Bio-HF), laboratory parameters (NT-proBNP, creatinine, bilirubin, uric acid, and sodium), body mass index, ischemic etiology of HF, and the presence of interventricular conduction defects. Correlations between explanatory variables were calculated, and multicollinearity was evaluated by calculating the tolerance and variance inflation factor. Schoenfeld residuals were used to check the proportional hazards assumption. As the proportional hazards assumption was not met, the weighted Cox regression analysis was conducted using the SAS macro.14,15 Univariable predictors of death with a P value below 0.2 were entered into the multivariable weighted Cox proportional hazards model. The Harrell concordance statistic was calculated for the final Cox regression model. The results are presented as hazard ratios (HRs) with 95% CIs. A P value below 0.05 was considered significant.
Results
The final study population consisted of 279 patients with end-stage HF awaiting HT. All patients had NYHA functional classes III and IV (88.9% and 11.1%, respectively), and profiles 4 to 7 according to the Interagency Registry for Mechanically Assisted Circulatory Support Classification (INTERMACS). A total of 95 patients (34.1%) died during 1-year follow-up. In the ACEI/ARB/ARNI group, the percentage of patients who received ACEIs was 70.8%, ARBs 23.4%, and ARNIs 5.8%. As for β-blockers, 42.6% of the patients received bisoprolol, 49.8% received metoprolol succinate, and 7.6% received carvedilol. Sodium-glucose cotransporter 2 (SGLT2) inhibitors were used in 17.2% of the patients. A summary of patient clinical characteristics and pharmacologic treatment is presented in Table 1. The ROC curves for the BCN Bio-HF calculator, HFSS, SHFM, and NT-proBNP levels are shown in Figure 1A–1D. The BCN Bio-HF calculator proved excellent at predicting 1-year mortality (AUC = 0.95; P <0.001), with high sensitivity (91%) and specificity (91%). It had significantly higher prognostic power than the NT-proBNP level, HFSS, and SHFM. The Hosmer–Lemeshow test demonstrated adequate calibration of the BCN Bio-HF calculator (P = 0.12), SHFM (P = 0.29), and NT-proBNP level (P = 0.15), and poor calibration of HFSS (P = 0.02).
![](https://pamw.pl/sites/default/files/json_zip_files/uncompressed/16757/IMAGES/KP_WEB__TAB_TABLE01.png)
Parameter | Whole population (n = 279) | Survivors (n = 184) | Nonsurvivors (n = 95) | P value |
---|---|---|---|---|
Baseline data | ||||
Age, y | 56 (50–60) | 56 (50–60) | 55 (50–60) | 0.8 |
Male sex, n (%) | 243 (87.1) | 164 (89.1) | 79 (83.2) | 0.16 |
BMI, kg/m2 | 26.93 (23.81–30.15) | 27.41 (24.58–30.78) | 26.13 (22.84–28.41) | 0.004 |
SBP, mm Hg | 102 (92–116) | 107 (99.5–120) | 96 (90–105) | <0.001 |
DBP, mm Hg | 64 (60–71) | 66 (60–74.5) | 60 (55–65) | <0.001 |
HR, bpm | 71 (66–78) | 70 (66–76) | 73 (66–80) | 0.16 |
Ischemic etiology of HF, n (%) | 159 (57) | 83 (45.1) | 76 (80) | <0.001 |
Hypertension, n (%) | 131 (47) | 83 (45.1) | 48 (50.5) | 0.39 |
Type 2 diabetes, n (%) | 143 (51.3) | 90 (48.9) | 53 (55.8) | 0.28 |
Hypercholesterolemia, n (%) | 181 (64.9) | 118 (64.1) | 63 (66.3) | 0.72 |
Atrial fibrillation, n (%) | 130 (46.6) | 84 (45.7) | 46 (48.4) | 0.66 |
Laboratory parameters | ||||
WBC, × 109/l | 7.28 (6.06–8.5) | 7.32 (6.06–8.56) | 7.26 (6.08–8.46) | 0.96 |
Hemoglobin, mmol/l | 8.8 (8.2–9.7) | 8.8 (8.2–9.7) | 8.9 (8.2–9.9) | 0.94 |
Creatinine, µmol/l | 108 (93–127) | 105 (89–124) | 117 (104–135) | <0.001 |
Platelet count, × 109/l | 196 (171–234) | 195 (170–238.5) | 201 (172–221) | 0.87 |
Total bilirubin, µmol/l | 18.3 (11.9–23.8) | 17.35 (11.35–23) | 19.7 (13.1–29) | 0.01 |
Uric acid, µmol/l | 441 (371–521) | 407 (345–479) | 512 (435–594) | <0.001 |
Urea, mmol/l | 8.3 (5.9–12.9) | 2.07 (1.59–2.83) | 2.05 (1.61–2.85) | 0.99 |
hs-CRP, mg/l | 3.84 (1.71–9.4) | 3.34 (1.6–10.38) | 4.52 (2.23–8.08) | 0.35 |
Albumin, g/l | 44 (41–46) | 44 (40.5–46) | 43 (41–46) | 0.6 |
Fibrinogen, g/dl | 384 (316–451) | 378 (312.5–442.5) | 399.5 (340–496) | 0.06 |
AST, U/l | 26 (20–31) | 25.5 (20–31) | 26 (20–34) | 0.91 |
ALT, U/l | 21 (16–31) | 21.5 (15.5–31) | 20 (16–30) | 0.9 |
Glucose, mmol/l | 5.8 (5.3–6.4) | 5.75 (5.25–6.3) | 5.9 (5.3–6.7) | 0.28 |
NT-proBNP, pg/ml | 4609 (1978–7215) | 3964.5 (1948–7396) | 5230 (2500–7102) | 0.13 |
ST2, ng/ml | 43.75 (33.87–80.46) | 37.12 (29.64–43.97) | 95.99 (75.17–109.16) | <0.001 |
Sodium, mmol/l | 139 (136–140) | 139 (138–141) | 136 (135–139) | <0.001 |
Echocardiographic parameters | ||||
LVEDd, mm | 73 (68–80) | 73 (68–80) | 73 (68–81) | 0.66 |
LVEF, % | 18 (15–20) | 18 (15–21) | 17 (15–20) | 0.12 |
TAPSE, mm | 15 (14–17) | 15 (14–17) | 15 (13–17) | 0.08 |
RVEDd, mm | 34 (31–40) | 34 (31–41.5) | 34 (30–39) | 0.52 |
Hemodynamic parameters | ||||
mPAP, mm Hg | 22 (17–30) | 22 (16–29.5) | 23 (18–31) | 0.11 |
CI, l/min/m2 | 17 (11–21) | 17 (10–22) | 17 (12–21) | 0.69 |
TPG, mm Hg | 8 (6–11) | 8 (6–10) | 9 (6–11) | 0.25 |
PVR, Wood units | 1.96 (1.48–2.35) | 1.92 (1.48–2.35) | 2.01 (1.5–2.48) | 0.32 |
Treatment | ||||
β-Blockers, n (%) | 263 (94.3) | 174 (94.6) | 89 (93.7) | 0.76 |
Bisoprolol, mg/d | 5 (2.5–5) | 5 (2.5–5) | 5 (2.5–7.5) | 0.25 |
Carvedilol, mg/d | 25 (12.5–25) | 25 (12.5–25) | 25 (12.5–25) | 0.81 |
Metoprolol succinate, mg/d | 71.25 (23.75–95) | 71.25 (23.75–95) | 71.25 (23.75–95) | 0.62 |
ACEI/ARB/ARNI, n (%) | 257 (92.1) | 168 (91.3) | 89 (93.7) | 0.48 |
Ramipril, mg/d | 5 (5–10) | 5 (5–10) | 7.5 (5–10) | 0.9 |
Perindopril, mg/d | 5 (2.5–5) | 5 (2.5–5) | 5 (2.5–5) | 0.95 |
Valsartan, mg/d | 160 (80–160) | 160 (160–160) | 160 (80–160) | 0.003b |
Sacubitril / valsartan; sacubitril, mg/d | 97.2 (48.6–97.2) | 72.9 (48.6–97.2) | 97.2 (72.9–97.2) | 0.47 |
Sacubitril / valsartan; valsartan, mg/d | 102.8 (51.4–102.8) | 77.1 (51.4–102.8) | 102.8 (51.4–102.8) | 0.7 |
MRA, n (%) | 260 (93.2) | 173 (94) | 87 (91.6) | 0.44 |
Spironolactone / epleronone, mg/d | 25 (25–50) | 25 (25–50) | 25 (25–50) | 0.38 |
Loop diuretics, n (%) | 279 (100) | 184 (100) | 95 (100) | 1 |
Statins, n (%) | 206 (73.8) | 131 (71.2) | 75 (78.9) | 0.16 |
Ivabradine, n (%) | 50 (18) | 35 (19.1) | 15 (15.8) | 0.49 |
ICD/CRT-D, n (%) | 279 (100) | 184 (100) | 95 (100) | 1 |
Inotropic support during follow-up, n (%) | 38 (13.6) | 23 (12.5) | 15 (15.8) | 0.45 |
Risk scores | ||||
HFSS | 7.6 (7.16–8.17) | 7.95 (7.39–8.33) | 7.24 (6.91–7.5) | <0.001 |
SHFM | 0.43 (0.01–0.92) | 0.27 (–0.13 to 0.76) | 0.63 (0.39–1.22) | <0.001 |
BCN Bio-HF | –21.03 (–60.1 to –21.03) | –16.40 (–21.45 to –12.09) | –84.39 (–107.6 to –52.54) | <0.001 |
Other parameters | ||||
VO2 max, ml/kg/min | 10.8 (9.8–11.5) | 10.8 (9.85–11.5) | 10.8 (9.8–11.4) | 0.68 |
IVCD (QRS >0.12 s), n (%) | 122 (43.7) | 69 (37.5) | 53 (55.8) | 0.004 |
Data are presented as median (interquartile range) unless indicated otherwise. P value <0.05 was considered significant. SI conversion factors: to convert hemoglobin to g/l, multiply by 1.611; hs-CRP to nmol/l, by 9.524; ALT and AST to μkat/l, by 0.0167; NT-proBNP to ng/l, by 1. Abbreviations: ACEI, angiotensin-converting enzyme inhibitor; ALT, alanine aminotransferase; ARB, angiotensin II receptor blocker; ARNI, angiotensin receptor–neprilysin inhibitor; AST, aspartate aminotransferase; BCN Bio-HF, Barcelona Bio-Heart Failure risk calculator; BMI, body mass index; CI, cardiac index; CRT-D, cardiac resynchronization therapy–defibrillator; DBP, diastolic blood pressure; HF, heart failure; HFSS, Heart Failure Survival Score; HR, heart rate; hs-CRP, high-sensitivity C-reactive protein; ICD, implantable cardioverter-defibrillator; IVCD, interventricular conduction defect; LVEDd, left ventricular end-diastolic dimension; LVEF, left ventricular ejection fraction; mPAP, mean pulmonary arterial pressure; MRA, mineralocorticoid receptor antagonist; NT-proBNP, N-terminal pro–B-type natriuretic peptide; PVR, pulmonary vascular resistance; RVEDd, right ventricular end-diastolic dimension; SBP, systolic blood pressure; SHFM, Seattle Heart Failure Model; ST2, soluble suppression of tumorigenicity; TAPSE, tricuspid annular plane systolic excursion; TPG, transpulmonary gradient; VO2 max, maximal oxygen uptake; WBC, white blood cell count |
![](https://pamw.pl/sites/default/files/json_zip_files/uncompressed/16757/IMAGES/KP_WEB__FIG_01.png)
Figure 1. Receiver operating characteristic curves for the Barcelona Bio-Heart Failure Risk Calculator (A), Heart Failure Survival Score (B), Seattle Heart Failure Model (C) and N-terminal pro–B-type natriuretic peptide level (D)
Abbreviations: AUC, area under the curve
Differences between the calculated AUCs for BCN Bio-HF vs NT-proBNP level, BCN Bio-HF vs HFSS, and BCN Bio-HF vs SHFM amounted to 0.391 (95% CI, 0.316–0.466; P <0.001), 0.137 (95% CI, 0.081–0.194; P <0.001, and 0.25 (95% CI, 0.18–0.32; P <0.001), respectively. A summary of the ROC analysis for the analyzed risk scores and the NT-proBNP level is presented in Table 2.
![](https://pamw.pl/sites/default/files/json_zip_files/uncompressed/16757/IMAGES/KP_WEB__TAB_TABLE02.png)
Parameter | AUC (95% CI) | Cutoff | Sensitivity (95% CI) | Specificity (95% CI) | PPV (95% CI) | NPV (95% CI) | LR+ (95% CI) | LR– (95% CI) | Accuracy |
---|---|---|---|---|---|---|---|---|---|
BCN Bio-HF | 0.95 (0.92–0.97) | <–42.12 | 0.91 (0.83–0.96) | 0.91 (0.86–0.95) | 0.84 (0.76–0.91) | 0.95 (0.91–0.98) | 10.41 (5.46–15.37) | 0.1 (0.04–0.17) | 0.91 (0.87–0.94) |
NT-proBNP | 0.56 (0.49–0.62) | ≥2223 | 0.8 (0.71–0.88) | 0.34 (0.27–0.41) | 0.38 (0.32–0.46) | 0.77 (0.66–0.85) | 1.21 (1.03–1.38) | 0.59 (0.32–0.86) | 0.49 (0.43–0.55) |
HFSS | 0.81 (0.76–0.86) | <7.741 | 0.93 (0.85–0.97) | 0.6 (0.53–0.67) | 0.55 (0.47–0.63) | 0.94 (0.88–0.98) | 2.33 (1.9–2.77) | 0.12 (0.03–0.21) | 0.71 (0.66–0.77) |
SHFM | 0.7 (0.63–0.76) | ≥0.364 | 0.78 (0.69–0.86) | 0.56 (0.48–0.63) | 0.48 (0.4–0.56) | 0.83 (0.75–0.89) | 1.77 (1.42–2.12) | 0.39 (0.24–0.55) | 0.63 (0.57–0.69) |
Abbreviations: LR, negative likelihood ratio; LR+, positive likelihood ratio; NPV, negative predictive value; PPV, positive predictive value; others, see Table 1 and Figure 1 |
According to the multivariable Cox proportional hazard analysis, lower BCN Bio-HF and HFSS scores, as well as a higher serum bilirubin concentration, were associated with a higher risk of mortality during 1-year follow-up. The Harrell concordance index for the final Cox regression model was 0.871. Factors associated with death in univariable and multivariable analyses are presented in Table 3.
![](https://pamw.pl/sites/default/files/json_zip_files/uncompressed/16757/IMAGES/KP_WEB__TAB_TABLE03.png)
Parameter | Univariable analysis | Multivariable analysis | ||
---|---|---|---|---|
HR (95% CI) | P value | HR (95% CI) | P value | |
Ischemic etiology of HFa | 2.851 (1.708–4.761) | <0.001 | – | – |
Presence of IVCDa | 1.939 (1.288–2.919) | 0.001 | – | – |
BCN Bio-HF scoreb | 1.02 (1.018–1.025) | <0.001 | 1.015 (1.012–1.019) | <0.001 |
HFSSb | 4.63 (3.067–6.993) | <0.001 | 2.801 (1.848–4.237) | <0.001 |
SHFMa | 1.459 (1.142–1.865) | 0.003 | – | – |
NT-proBNPc | 1.004 (0.999–1.008) | 0.14 | – | – |
BMIb | 1.086 (1.033–1.142) | 0.001 | – | – |
Creatininea | 1.014 (1.005–1.022) | 0.001 | – | – |
Bilirubina | 1.017 (1.004–1.03) | 0.008 | 1.015 (1.002–1.028) | 0.02 |
Uric acida | 1.004 (1.002–1.005) | <0.001 | – | – |
Sodiumb | 1.14 (1.083–1.2) | <0.001 | – | – |
a Per 1 unit increase b Per 1 unit decrease c Per 100 unit increase Abbreviations: HR, hazard ratio; others, see Table 1 |
Discussion
The present study evaluated the predictive performance of the contemporary BCN Bio-HF risk score in patients with end-stage HF awaiting HT during 1-year follow-up. This prospective, single-center study is the first to show that the BCN Bio-HF score is an independent predictor of 1-year mortality, with high power, sensitivity, and specificity, suggesting it can be used for prediction of death at 1 year in HF patients placed on an HT waiting list. Furthermore, this study is the first to report that the BCN Bio-HF calculator has significantly higher prognostic power than the traditional scores, HFSS and SHFM. The BCN Bio-HF risk calculator was developed in 2014 by Lupón et al6 using data from a cohort of 864 contemporarily treated patients with HF, and its discriminatory power ranged from 0.771 to 0.793, depending on the model. The BCN Bio-HF calculator was externally validated using data of patients included in the PROTECT study (Use of Amino-Terminal Pro-B Type Natriuretic Peptide to Guide Outpatient Therapy of Patients With Chronic Left Ventricular Systolic Dysfunction), which analyzed 151 individuals with left ventricular systolic dysfunction.16 In 2017, the BCN Bio-HF risk calculator was updated based on the PARADIGM-HF (Prospective Comparison of ARNI with ACEI to Determine Impact on Global Mortality and Morbidity in Heart Failure) cohort, accounting for the use of ARNIs and cardiac implantable electronic devices.17 Then, in 2023, a new version of the BCN Bio-HF calculator was created, which used the updated β coefficients to calculate the risk of death and / or hospitalization in patients with HF, and included the use of SGLT2 inhibitors.18 In studies evaluating the new versions of the BCN Bio-HF calculator, its prognostic power (AUC) ranged from 0.825 to 0.86 for all-cause mortality at 1 year.17,18 In our study, we noted a significantly higher prognostic power of the BCN Bio-HF score than reported for the initial and updated versions of the calculator. Notably, our study differed from the analysis by Lupón et al6 in some aspects. For example, we are the first to analyze HT candidates, whereas the population in the cited study comprised older patients, at a mean (SD) age of 68.2 (12) years, mainly with NYHA class I or II. So far, 2 studies have evaluated the predictive performance of the BCN Bio-HF score in various populations of patients with HF.19,20 Pérez-Sanz et al19 reported the utility of the BCN Bio-HF calculator in the assessment of short- and long-term mortality in HF patients hospitalized in an emergency department. Codina et al20 demonstrated good discriminatory power of the BCN Bio-HF risk score for 1-year mortality in patients with LVEF below 40%. In that study, the BCN Bio-HF risk score was shown to have the best discriminatory and overall performance among all the analyzed prognostic tools (SHFM, Meta-Analysis Global Group in Chronic Heart Failure risk score, and PARADIGM Risk of Events and Death in the Contemporary Treatment of Heart Failure score).20 In comparison with the prognostic tools developed earlier, the BCN Bio-HF calculator combines simple and conventional clinical variables and commonly used laboratory parameters with novel HF biomarkers based on data from cohorts of patients treated according to new standards. In their other work, Codina et al21 analyzed the prognostic value of the BCN Bio-HF risk score in a population of patients with advanced HF from the LEVO-D (Intermittent Inotropic Support with Levosimendan in Advanced Heart Failure as Destination Therapy) registry. In that study, the BCN Bio-HF score showed worse discrimination and calibration than in our analysis, with underestimation of the risk in the analyzed group of patients. Notably, the study design and population analyzed by Codina et al21 were different from those used in our analysis—the LEVO-D registry was a retrospective study, the patients received levosimendan as the destination therapy, and they were ineligible for advanced therapies, such as HT or an MCS device implantation. On the contrary, our study population was analyzed prospectively and the patients were HT candidates.
Another interesting finding of our study is the independent association between a lower HFSS score and worse 1-year prognosis in patients with end-stage HF. HFSS had good prognostic power and sensitivity but low specificity in the analyzed group of patients. Over the years, HFSS has been widely used as a predictive model facilitating risk stratification in patients with advanced HF. This score was developed in the 1990s based on data from outpatients referred for HT assessment,8 and has been validated in external populations of patients with HF, with a prognostic power ranging between 0.56 and 0.81.22-24 Our analysis of HFSS confirmed its prognostic value for patients with end-stage HF receiving optimal contemporary therapy. This finding is consistent with our previous studies assessing the importance of HFSS in ambulatory patients with end-stage HF. As we have previously shown, combining the original HFSS with new prognostic markers (apelin, modified Model for End-Stage Liver Disease [MELD] score, MELD-XI score, or procalcitonin) significantly improved the prognostic capability of these markers in patients with end-stage HF.1,4,25-28 Lund et al29 reported that HFSS may be useful in serial assessment of the mortality risk in ambulatory patients with end-stage HF referred for HT evaluation. Goda et al30 demonstrated acceptable prognostic power of HFSS for risk stratification of patients with end-stage HF undergoing HT evaluation. These authors showed that combining HFSS with SHFM improved the discriminatory power over either one alone.30
We also investigated the predictive performance of SHFM for 1-year mortality in our study group. SHFM was not an independent predictor of death at 1 year, and its prognostic power was limited in the analyzed cohort. Our results are consistent with previous studies demonstrating risk underestimation by SHFM in patients referred for HT evaluation.24,31 Gorodeski et al31 showed that SHFM had only modest predictive accuracy for the primary combined outcome, or the outcome of death alone, in consecutive ambulatory patients evaluated for HT and / or MCS eligibility. Alba et al24 also reported modest discrimination and questionable calibration of SHFM for the assessment of outcomes in patients with HF. Considering the above data, the risk predicted by SHFM in patients considered for HT must be interpreted with caution. One of the reasons for suboptimal predictive performance of SHFM may be that the management of patients with HF has changed significantly over the years. Although SHFM remains the most widely validated scale, it was originally derived from and validated in a historical population of HF patients, who were not treated according to the current HF therapy guidelines. In the derivation cohort, only 3% of the patients were taking MRAs, and none were prescribed β-blockers.9 In addition, in the validation cohort, data on uric acid levels and lymphocyte counts were missing in 65% and up to 100% of patients, respectively (in our study, all data were complete), which must have influenced the prognostic value of this score.9 Furthermore, SHFM does not include novel biomarkers that are strongly associated with prognosis of HF patients. In addition, factors previously associated with mortality in patients with HF may have a lower prognostic value in contemporary cohorts of HF patients. This may partially explain the poorer prognostic utility of SHFM in more recent validation cohorts of individuals with HF.24,31
In addition, we showed that bilirubin was a factor independently associated with 1-year mortality, and that it seems to be of value for risk stratification in HF patients. Several studies have demonstrated that a higher serum bilirubin concentration is associated with worse prognosis in end-stage HF individuals.32-34 Congestive tract hepatopathy is a common manifestation of HF. Both congestion and low perfusion of the liver are considered to reflect impaired hemodynamics related to HF. Congestive hepatopathy arises from chronic passive venous congestion caused by elevated central venous pressure secondary to HF, which is transmitted to the central veins of the liver. This results in presinusoidal dilation, decreased hepatic arterial blood flow, and decreased arterial oxygen saturation, which may gradually lead to irreversible liver damage.32,35,36 Another mechanism responsible for liver injury in HF patients is low hepatic blood flow due to low cardiac output, which leads to further hepatocellular hypoxia.33 Although congestive hepatopathy is usually clinically asymptomatic, it can be identified based on abnormalities in routine laboratory evaluations. One of the most common laboratory abnormalities in congestive hepatopathy secondary to HF is increased serum bilirubin concentration. Bilirubin is elevated in approximately 70% of HF cases.35 Other abnormalities associated with hepatopathy are increased serum alkaline phosphatase levels and hypoalbuminemia.34,35,37 Allen et al32 demonstrated that elevated total bilirubin concentration was one of the most important factors predicting worse prognosis, even after accounting for many demographic, clinical, and biochemical variables. Furthermore, changes in bilirubin concentration can a be useful marker for assessing left ventricular reverse remodeling in patients with advanced HF. This finding was corroborated by Hosoda et al,33 who also demonstrated that changes in serum bilirubin concentration are a predictor of outcomes in HF patients after cardiac resynchronization therapy.33 Other studies reported that high total bilirubin concentration was associated with HF severity and strongly correlated with worse prognosis in patients with HF.38,39 Simpson et al38 demonstrated that total bilirubin concentration was an important factor in a new prognostic model based on a contemporary cohort of HF patients. Bilirubin is also an important component of the MELD score and its modifications. Prognostic significance of the MELD score has been confirmed in studies involving patients with advanced HF awaiting HT.25,27,40 Several metabolic processes involving bilirubin are impaired secondary to hepatocellular hypoxia in HF. The uptake of indirect bilirubin from the blood, conjugation in hepatocytes, and secretion of direct bilirubin into bile in HF patients are impaired, which leads to an increase in serum total bilirubin concentration.34,36,37 Another cause of the increase in bilirubin is hemolysis, which occurs mainly secondary to lung congestion in HF patients.41 Notably, bilirubin is a widely available, inexpensive, and noninvasive marker that is typically evaluated during routine ambulatory and hospital care. Routine bilirubin assessment may improve mortality risk stratification in patients with end-stage HF.
Limitations
Several limitations of this work should be noted. First, it was a single-center, observational study with a relatively small sample size. Second, it included relatively stable chronic HF patients, so the findings cannot be readily extrapolated to patients with INTERMACS profiles 1 to 3 or NYHA functional classes I to II. Third, there was no independent validation cohort to confirm our results. It is likely that if an independent validation cohort had been used, the AUC for the BCN Bio-HF calculator and other scores would have been lower. In addition, we used a simplified version of the BCN Bio-HF calculator, without taking troponins into account. Future research should include the BCN Bio-HF risk score with all dedicated markers (cTnT, sST2, NT-proBNP, and SGLT2 inhibitors). Furthermore, the study was conducted during a period when guidelines for HF treatment were changing. A majority of our patients were enrolled during the pre-SGLT2 inhibitors era—they were included in the study between 2018 and 2021. Due to the low percentage of SGLT2 inhibitor use at the time of inclusion in the study, we did not analyze the latest version of the BCN Bio-HF calculator. Besides, we did not have data on whether SGLT2 inhibitors had been added to the patients’ medical regimens during follow-up. Thus, we could not adequately address the effects of these medications on our results. Finally, the only outcome analyzed was all-cause mortality, and the causes of death were not determined.
Conclusions
In summary, BCN Bio-HF and HFSS scores and serum bilirubin concentration can reliably predict the overall risk of mortality in patients with advanced HF awaiting HT. The excellent prognostic strength of the BCN Bio-HF risk calculator and the acceptable prognostic power of HFSS allowed for the prediction of 1-year mortality in the analyzed population. The prognostic performance of SHFM for 1-year mortality was limited in the analyzed cohort.
Bożena Szyguła-Jurkiewicz, MD, PhD, Third Department of Cardiology, Medical University of Silesia, Silesian Center for Heart Diseases, ul. Sklodowskiej-Curie 9, 41-800 Zabrze, Poland, phone: +48 32 373 38 60, email: bjurkiewicz@sum.edu.pl
February 15, 2024.
May 20, 2024.
May 24, 2024.
None.
This work was supported by an internal grant from the Medical University of Silesia in Katowice (BNW-1-047/K/3/K; to BS-J).
Conceptualization: BS-J and WS-W. Data curation: WS-W, MS, MJ, and AK. Formal analysis: BS-J, WS-W, AK, MJ, and MS. Funding acquisition: BS-J and MG. Investigation: WS-W, BS-J, AK, MS, MJ, and MG. Methodology: BS-J, WS-W, and MS. Resources: BS-J and MG. Software: MS. Visualization: WS-W and MS. Writing – original draft preparation: WS-W and BS-J. Writing – review and editing: WS-W, BS-J, MJ, MS, AK, and MG. All authors have read and accepted the published version of the manuscript.
None declared.
Szczurek-Wasilewicz W, Skrzypek M, Karmański M, et al. The Barcelona Bio-Heart Failure risk calculator may predict 1-year mortality in patients with advanced heart failure. Pol Arch Intern Med. 2024; 134: 16757. doi:10.20452/pamw.16757
- 1.
- Szczurek-Wasilewicz W, Gąsior M, Skrzypek M, et al. Predictors of 1-year mortality in ambulatory patients with advanced heart failure awaiting heart transplant. Pol Arch Intern Med. 2022; 132: 16151.Crossref
- 2.
- Leon-Justel A, Morgado Garcia-Polavieja JI, Alvarez-Rios AI, et al. Biomarkers-based personalized follow-up in chronic heart failure improves patient’s outcomes and reduces care associate cost. Health Qual Life Outcomes. 2021; 19: 142.Crossref
- 3.
- Hauptman PJ, Swindle J, Hussain Z, et al. Physician attitudes toward end-stage heart failure: a national survey. Am J Med. 2008; 121: 127-135.Crossref
- 4.
- Kurkiewicz K, Gąsior M, Szyguła-Jurkiewicz BE. Markers of malnutrition, inflammation, and tissue remodeling are associated with 1-year outcomes in patients with advanced heart failure. Pol Arch Intern Med. 2023; 133: 16411.Crossref
- 5.
- Pascual-Figal DA, Manzano-Fernández S, Boronat M, et al. Soluble ST2, high-sensitivity troponin T- and N-terminal pro-B-type natriuretic peptide: complementary role for risk stratification in acutely decompensated heart failure. Eur J Heart Fail. 2011; 13: 718-725.Crossref
- 6.
- Lupón J, de Antonio M, Vila J, et al. Development of a novel heart failure risk tool: the Barcelona Bio-Heart Failure risk calculator (BCN Bio-HF calculator). PLoS One. 2014; 9: e85466.Crossref
- 7.
- Mehra MR, Canter CE, Hannan MM, et al. The 2016 International Society for Heart Lung Transplantation listing criteria for heart transplantation: a 10-year update. J Heart Lung Transplant. 2016; 35: 1-23.Crossref
- 8.
- Aaronson KD, Schwartz JS, Chen TM, et al. Development and prospective validation of a clinical index to predict survival in ambulatory patients referred for cardiac transplant evaluation. Circulation. 1997; 95: 2660-2667.Crossref
- 9.
- Levy WC, Mozaffarian D, Linker DT, et al. The Seattle Heart Failure Model: prediction of survival in heart failure. Circulation. 2006; 113: 1424-1433.Crossref
- 10.
- Meijers WC, Bayes-Genis A, Mebazaa A, et al. Circulating heart failure biomarkers beyond natriuretic peptides: review from the Biomarker Study Group of the Heart Failure Association (HFA), European Society of Cardiology (ESC). Eur J Heart Fail. 2021; 23: 1610-1632.Crossref
- 11.
- Song Y, Li F, Xu Y, et al. Prognostic value of sST2 in patients with heart failure with reduced, mid-range and preserved ejection fraction. Int J Cardiol. 2020; 304: 95-100.Crossref
- 12.
- Ponikowski P, Voors AA, Anker SD, et al. ESC Scientific Document Group. 2016 ESC Guidelines for the diagnosis and treatment of acute and chronic heart failure: the Task Force for the diagnosis and treatment of acute and chronic heart failure of the European Society of Cardiology (ESC) developed with the special contribution of the Heart Failure Association (HFA) of the ESC. Eur Heart J. 2016; 37: 2129-2200.Crossref
- 13.
- McDonagh TA, Metra M, Adamo M, et al. 2021 ESC Guidelines for the diagnosis and treatment of acute and chronic heart failure: developed by the Task Force for the diagnosis and treatment of acute and chronic heart failure of the European Society of Cardiology (ESC). With the special contribution of the Heart Failure Association (HFA) of the ESC. Eur J Heart Fail. 2022; 24: 4-131.Crossref
- 14.
- Schemper M, Wakounig S, Heinze G. The estimation of average hazard ratios by weighted Cox regression. Stat Med. 2009; 28: 2473-2489.Crossref
- 15.
- Schemper M, Wakounig S, Heinze G. The estimation of average hazard ratios by weighted Cox regression. Stat Med. 2009; 28: 2473-2489.Crossref
- 16.
- Lupón J, Januzzi JL, de Antonio M, et al. Validation of the Barcelona Bio-Heart Failure risk calculator in a cohort from Boston. Rev Esp Cardiol (Engl Ed). 2015; 68: 80-81.Crossref
- 17.
- Lupón J, Simpson J, McMurray JJV, et al. Barcelona Bio-HF Calculator Version 2.0: incorporation of angiotensin II receptor blocker neprilysin inhibitor (ARNI) and risk for heart failure hospitalization. Eur J Heart Fail. 2018; 20: 938-940.Crossref
- 18.
- Codina P, Lupón J, Subirana I, et al. Barcelona Bio-HF calculator version 3.0: recalibration and incorporation of sodium-glucose cotransporter 2 inhibitor treatment. Eur J Heart Fail. 2023; 25: 131-132.Crossref
- 19.
- Pérez-Sanz TM, Gómez-Salvador I, Codina P, et al. Soluble ST2, BCN-Bio-HF calculator and MAGGIC-HF score in long-term risk prediction after an urgent visit for heart failure. Heart Vessels. 2024; 39: 216-225.Crossref
- 20.
- Codina P, Lupón J, Borrellas A, et al. Head-to-head comparison of contemporary heart failure risk scores. Eur J Heart Fail. 2021; 23: 2035-2044.Crossref
- 21.
- Codina P, Dobarro D, de Juan-Bagudá J, et al. Spanish LEVO‐D registry Collaborators. Heart failure risk scores in advanced heart failure patients: insights from the LEVO-D registry. ESC Heart Fail. 2023; 10: 2875-2881.Crossref
- 22.
- Nguyen LS, Coutance G, Ouldamar S, et al. Performance of existing risk scores around heart transplantation: validation study in a 4-year cohort. Transpl Int. 2018; 31: 520-530.Crossref
- 23.
- Aaronson KD, Cowger J. Heart failure prognostic models: why bother? Circ Heart Fail. 2012; 5: 6-9.Crossref
- 24.
- Alba AC, Agoritsas T, Jankowski M, et al. Risk prediction models for mortality in ambulatory patients with heart failure: a systematic review. Circ Heart Fail. 2013; 6: 881-889.Crossref
- 25.
- Szczurek W, Szyguła-Jurkiewicz B, Zakliczyński MW, et al. Prognostic value of selected risk scales in patients with end-stage heart failure. Kardiol Pol. 2018; 76: 1320-1326.Crossref
- 26.
- Szczurek W, Gąsior M, Skrzypek M, Szyguła-Jurkiewicz B. Apelin improves prognostic value of HFSS (Heart Failure Survival Score) and MAGGIC (Meta-Analysis Global Group in Chronic Heart Failure) scales in ambulatory patients with end-stage heart failure. J Clin Med. 2020; 9: 2300-2309.Crossref
- 27.
- Szczurek W, Gąsior M, Romuk E, et al. Usefulness of combining prognostic scores to predict survival in patients with advanced heart failure. J Heart Lung Transplant. 2019; 38: 1224-1227.Crossref
- 28.
- Szczurek-Wasilewicz W, Skrzypek M, Romuk E, et al. The utility of pentraxin and modified prognostic scales in predicting outcomes of patients with end-stage heart failure. J Clin Med. 2022; 11: 2567-2580.Crossref
- 29.
- Lund LH, Aaronson KD, Mancini DM. Validation of peak exercise oxygen consumption and the Heart Failure Survival Score for serial risk stratification in advanced heart failure. Am J Cardiol. 2005; 95: 734-741.Crossref
- 30.
- Goda A, Williams P, Mancini D, et al. Selecting patients for heart transplantation: comparison of the Heart Failure Survival Score (HFSS) and the Seattle Heart Failure Model (SHFM). J Heart Lung Transplant. 2011; 30: 1236-1243.Crossref
- 31.
- Gorodeski EZ, Chu EC, Chow CH, et al. Application of the Seattle Heart Failure Model in ambulatory patients presented to an advanced heart failure therapeutics committee. Circ Heart Fail. 2010; 3: 706-714.Crossref
- 32.
- Allen LA, Felker GM, Pocock S, et al; CHARM Investigators. Liver function abnormalities and outcome in patients with chronic heart failure: data from the Candesartan in Heart Failure: Assessment of Reduction in Mortality and Morbidity (CHARM) program. Eur J Heart Fail. 2009; 11: 170-177.Crossref
- 33.
- Hosoda J, Ishikawa T, Matsumoto K, et al. Significance of change in serum bilirubin in predicting left ventricular reverse remodeling and outcomes in heart failure patients with cardiac resynchronization therapy. J Cardiol. 2017; 70: 416-419.Crossref
- 34.
- Wu AH, Levy WC, Welch KB, et al. Association between bilirubin and mode of death in severe systolic heart failure. Am J Cardiol. 2013; 111: 1192-1197.Crossref
- 35.
- Gorter TM, van Veldhuisen DJ, Bauersachs J, et al. Right heart dysfunction and failure in heart failure with preserved ejection fraction: mechanisms and management. Position statement on behalf of the Heart Failure Association of the European Society of Cardiology. Eur J Heart Fail. 2018; 20: 16-37.Crossref
- 36.
- Lemmer A, VanWagner L, Ganger D. Congestive hepatopathy: differentiating congestion from fibrosis. Clin Liver Dis (Hoboken). 2018; 10: 139-143.Crossref
- 37.
- Shinagawa H, Inomata T, Koitabashi T, et al. Prognostic significance of increased serum bilirubin levels coincident with cardiac decompensation in chronic heart failure. Circ J. 2008; 72: 364-369.Crossref
- 38.
- Simpson J, Jhund PS, Lund LH, et al. Prognostic models derived in PARADIGM-HF and validated in ATMOSPHERE and the Swedish heart failure registry to predict mortality and morbidity in chronic heart failure. JAMA Cardiol. 2020; 5: 432-441.Crossref
- 39.
- Suzuki K, Claggett B, Minamisawa M, et al. Liver function and prognosis, and influence of sacubitril / valsartan in patients with heart failure with reduced ejection fraction. Eur J Heart Fail. 2020; 22: 1662-1671.Crossref
- 40.
- Kim MS, Kato TS, Farr M, et al. Hepatic dysfunction in ambulatory patients with heart failure: application of the MELD scoring system for outcome prediction. J Am Coll Cardiol. 2013; 61: 2253-2261.
- 41.
- Reese AJ. The effect of hypoxia on liver secretion studied by intravital fluorescence microscopy. Br J Exp Pathol. 1960; 41: 527-535.