Identifying a Cardio‐Hepatic Phenotype by Integrating the Albumin‐Bilirubin Score Into Risk Assessment in Pulmonary Arterial Hypertension

ABSTRACT

Risk stratification is essential in pulmonary arterial hypertension (PAH), yet the prognostic utility of the albumin‐bilirubin (ALBI) score as a marker of hepatic dysfunction remains underexplored. In this single‐center study, we evaluated the clinical, neurohumoral, echocardiographic and hemodynamic correlates and prognostic value of the ALBI score, and to assess whether this score offers an incremental benefit over the Registry to EValuate Early And Long‐term pulmonary arterial hypertension disease management (REVEAL) 2.0 risk score in patients with PAH. This study evaluated the prognostic value of the ALBI score and its incremental benefit in a cohort of 380 PAH patients (mean age 47.6 ± 18.3 years, 74.3% female). Patients were stratified into low (<= −2.60) and high (> −2.60) ALBI groups. High ALBI scores were significantly associated with worse functional class, reduced exercise capacity, elevated Nt‐proBNP, right ventricular systolic dysfunction, and vena cava plethora (all p < 0.001). Hemodynamically, elevated ALBI correlated with increased right atrial pressure (RAP) (p = 0.005), reduced cardiac index (p = 0.002), and a narrowed trans‐systemic pressure gradient (p = 0.001). Multivariable Cox regression demonstrated that ALBI was an independent predictor of mortality [Adjusted HR: 1.80 (95% CI: 1.30–2.49), p < 0.001], outperforming MELD variants and the FIB‐4 index. Integrating ALBI into REVEAL 2.0 yielded higher risk discrimination (C‐index: 0.756) than adding direct echocardiographic indices of right ventricular function or coupling (C‐indices: 0.748–0.749). In sensitivity analyses adjusting for TAPSE/PASP, invasive cardiac index, mean RAP, and background therapy, ALBI maintained its independent prognostic value [Adjusted HR: 1.44 (95% CI: 1.01–2.05), p = 0.044], with the full model achieving a C‐index of 0.801. Within the REVEAL 2.0 intermediate‐risk subgroup, a high ALBI score identified patients with a three‐fold higher mortality risk [HR: 3.04 (95% CI: 1.25–7.38), p = 0.014], improving discriminative performance by 26.3%. The ALBI score independently predicts mortality beyond standard pulmonary hemodynamics and echocardiographic right ventricular parameters. Incorporating ALBI into REVEAL 2.0 enhances survival prediction and aids risk refinement in intermediate‐risk PAH.

1. Introduction

Pulmonary arterial hypertension (PAH) is chronic and relentlessly progressive disease spectrum characterized by obstructive remodeling of distal pulmonary arterial bed resulting in elevated pulmonary vascular resistance (PVR), pulmonary arterial systolic and mean pressures, and impaired right ventricular–pulmonary arterial (RV/PA) coupling eventually leading to right‐sided heart failure (HF) [1, 2, 3, 4, 5, 6, 7].

In patients with HF and pulmonary hypertension (PH), hepatic dysfunction primarily manifests through two distinct but often overlapping clinical scenarios; congestive hepatopathy due to backward failure and ischemic hepatitis caused by forward failure [1, 2, 3, 4, 5, 6, 8, 9, 10]. In first situation, elevated central venous pressure transmitting back through the hepatic veins leads to passive congestion, sinusoidal dilation, hepatocyte atrophy, and perisinusoidal edema, which impairs oxygen diffusion [1, 2, 3, 4, 5, 6, 8, 9, 10]. Moreover, long‐standing venous stasis can also lead to hepatic fibrosis by activation of stellate cells, and finally “cardiac cirrhosis”. Ischemic hepatitis or “shock liver” develops secondary to a sudden decrease in cardiac output and hepatic blood flow, resulting in centrilobular necrosis [1, 2, 3, 4, 5, 6, 8, 9, 10]. While the liver can normally adapt for reduced flow by increasing oxygen extraction, this compensatory mechanism fails in cases of acute cardiogenic shock or treatment‐resistant systemic arterial hypotension concomitant with elevated central venous pressures [8, 9, 10]. Therefore, hepatic dysfunction has been reported as a strong independent predictor of mortality in both HF and PH [1, 2, 3, 4, 5, 6, 8, 9, 10].

Several multiparametric risk scores (MRSs) have been developed and validated for risk stratification at both initial and follow‐up assessments in patients with PAH, and are also used in other groups of PH [1, 2, 6, 11, 12, 13, 14, 15, 16, 17, 18, 19]. In patients with PH, hepatic dysfunction scores, developed for originally cirrhosis, have also been used for assessing multi‐organ involvement due to right‐sided HF, and predicting the clinical course and long‐term survival [8, 9, 10, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35]. Among these, the Albumin‐Bilirubin (ALBI) score is a simple two‐component index that has shown independent prognostic value in HF, idiopathic PAH, and portopulmonary hypertension [27, 28, 29, 31]. However, its comparative performance against other liver‐based scores and its incremental value beyond established multiparametric risk models have not been systematically evaluated in PAH.

In this single‐center study, we aimed to evaluate the clinical, neurohumoral, echocardiographic and hemodynamic correlates and prognostic value of the ALBI score, and to assess whether this score offers an incremental benefit over the REVEAL 2.0 risk score in patients with PAH.

2. Methods

2.1. Study Population

Our study cohort c onsisted of a retrospectively evaluated 380 patients with PAH diagnosis (mean age 47.6 ± 18.3 years, 74.3% female). These patients were enrolled between January 2007 and December 2025 and were selected from a larger pool of overall 1210 patients with PH enrolled in our single‐center study, EvalUation of Pulmonary Hypertension Risk Factors AssociaTEd with Survival (EUPHRATES). The diagnostic algorithms, hemodynamic confirmation, clinical sub‐classification of PH, and incident and prevalent PAH definitions have been based on the recommendations of the European Society of Cardiology/European Respiratory Society 2022 PH Guidelines [1, 2, 5, 6, 7]. The pre‐capillary PH definition includes criteria of a pulmonary artery wedge pressure (PAWP) ≤ 15 mmHg and a PVR > 2 Wood Units (WU) [1, 2, 5]. Patients with portopulmonary hypertension and those with documented liver disease were excluded from the study. Documented liver disease was defined as a prior clinical diagnosis of cirrhosis, portal hypertension, chronic viral hepatitis, autoimmune liver disease, primary biliary cholangitis, hepatocellular carcinoma, or other chronic hepatic disorders recorded in the medical records. Patients with active malignancy at the time of enrollment were also excluded. The stepwise selection process from the initial EUPHRATES cohort (n = 1210) to the final study population (n = 380) is illustrated in Figure1.

Figure 1.

Flowchart showing the inclusion and exclusion criteria for the study cohort from the EUPHRATES registry. Out of 1,210 patients with pulmonary hypertension, 380 patients with Group 1 PAH were included in the final analysis after applying predefined exclusion criteria and removing missing data. Patients were stratified into low ALBI (<= −2.60, n = 226) and high ALBI (> −2.60, n = 154) groups. Abbreviations: ALBI, albumin‐bilirubin; PAH, pulmonary arterial hypertension; PH, pulmonary hypertension.

Baseline risk scores were calculated based on clinical etiologies and initial assessments including World Health Organization functional class (FC), 6‐min walk distance (6MWD), blood biochemistry and cell counts, N‐terminal pro‐B‐type natriuretic peptide (Nt‐proBNP) levels, as well as echocardiographic and invasive hemodynamic measures of pulmonary circulation and RV function [5, 6, 11, 12, 13, 14, 15, 16, 17, 18, 19]. For baseline risk stratification, established MRSs developed from the Swedish PAH Registry (SPAHR), the Comparative Prospective Registry of Newly Initiated Therapies for Pulmonary Hypertension (COMPERA), the 3‐ and 4‐component French Pulmonary Hypertension Network (FPHN) low‐risk models, and Registry to EValuate Early And Long‐term PAH disease management (REVEAL) including REVEAL 1.0, REVEAL 2.0, and the 6‐component REVEAL Lite 2 score were utilized [5, 6, 11, 12, 13, 14, 15, 16, 17, 18, 19].

The study protocol was reviewed and approved by the Institutional Ethics Committee (registry decision number: 2013.3/4). This study was conducted in accordance with the principles of the Declaration of Helsinki.

2.2. Echocardiography

Transthoracic echocardiography (TTE) was performed on all patients. Tricuspid annular plane systolic excursion (TAPSE) and tissue Doppler (TDI) measurements were obtained to assess RV function in TTE, and estimated pulmonary artery pressures (PASP) were calculated from the tricuspid regurgitation jet. All measurements and assessments were made in accordance with the American Society of Echocardiography and the European Association of Cardiovascular Imaging guidelines [36, 37, 38].

2.3. Cardiac Catheterization

A 7Fr balloon‐tipped Swan–Ganz catheter (Edwards Lifesciences, Irvine, CA, USA) or pigtail catheter was used through the right jugular or femoral vein for RHC. Indirect Fick method was used for cardiac output estimation. All pressure tracings were visually inspected for physiological accuracy, and end‐expiratory pressure values were recorded.

2.4. Laboratory Assessment and Calculation of Liver‐Based Risk Scores

Routine venous blood samples collected during follow‐up visits were reviewed, including complete blood count, creatinine, Nt‐proBNP, alanine transaminase (ALT), aspartate transaminase (AST), total bilirubin, and albumin levels. Serum albumin was measured using the colorimetric method, and total bilirubin was determined via the diazo method. Sodium and creatinine levels were measured using standard automated laboratory assays.

The ALBI score was calculated using the following formula [27, 28, 29, 31]:

• ALBI = (log10 bilirubin [µmol/L] × 0.66) + (albumin [g/L] × −0.085)

Patients were stratified into low (≤ −2.60) and high (> −2.60) ALBI groups based on the internationally validated threshold [27, 28]. To compare the prognostic performance of ALBI against conventional liver dysfunction markers, the following scores were calculated [20, 21, 22, 23, 24, 25, 26, 32, 33, 34, 35]:

• MELD Score: 3.78 × ln(bilirubin [mg/dL]) + 11.2 × ln(INR) + 9.57 × ln(creatinine [mg/dL]) + 6.43

• MELD‐Na: MELD ‐ Na ‐ (0.025 × MELD × (140 ‐ Na)) + 140 (where Na is limited between 125 and 140 mmol/L)

• MELD 3.0: This updated version incorporated female sex and updated coefficients for bilirubin, albumin, and creatinine to improve mortality prediction.

• FIB‐4 Index: (Age [years] × AST [U/L])/(Platelet count [10^9/L] × √ALT [U/L])

For all scoring systems, higher values represented more advanced degrees of hepatic or systemic dysfunction, with ALBI specifically evaluated for its continuous and binary prognostic impact.

2.5. Statistical Analysis

Continuous variables were expressed as mean ± standard deviation or median with interquartile range based on the Shapiro‐Wilk test for normality, while categorical variables were presented as frequencies and percentages. Baseline characteristics were compared between low and high ALBI groups using Independent Samples t‐tests, Mann‐Whitney U tests, or Chi‐square tests as appropriate. The primary endpoint was all‐cause mortality over a 5‐year follow‐up period. Survival distributions were estimated using the Kaplan‐Meier method and compared with the log‐rank test.

To evaluate the independent prognostic value of the ALBI score, we utilized Cox proportional hazards regression models. The potentially non‐linear relationship between continuous ALBI values and mortality risk was further explored using Restricted Cubic Splines (RCS). A major focus of our analysis was the incremental benefit of integrating ALBI into the existing REVEAL 2.0 risk framework. This was assessed and compared against other established liver dysfunction markers, specifically MELD, MELD‐Na, MELD 3.0, and FIB‐4. To confirm the independent prognostic value of the ALBI score beyond RV function, central hemodynamics, and treatment intensity, comparative and multivariable sensitivity Cox models were constructed adjusting for TAPSE, TAPSE/PASP, invasive cardiac index, RAP, and targeted PAH therapy regimens alongside REVEAL 2.0.

We performed Time‐dependent ROC analysis to evaluate the discriminative accuracy of each marker‐enhanced model throughout the entire follow‐up duration. Model fit and parsimony were compared using the Akaike Information Criterion (AIC) and Likelihood Ratio Tests (LRT), while changes in the Concordance index (C‐index) quantified the improvement in predictive power. Clinical utility was examined through Decision Curve Analysis, determining the net benefit of the ALBI‐augmented model across various threshold probabilities. Finally, a dedicated subgroup analysis was conducted on patients classified as Intermediate Risk by the REVEAL 2.0 score. Statistical significance was defined as a two‐sided p‐value < 0.05. All analyses were performed using R software version 4.3.2, utilizing the following packages: survival, survminer, rms, timeROC, dcurves, dplyr, and ggplot2.

3. Results

3.1. Baseline Clinical and Laboratory Characteristics

Mean age was 47.6 ± 18.3 years and 74.5% of patients were female. A total of 380 patients were stratified according to their ALBI scores into low (≤ 2.6, n = 226) and high (> 2.6, n = 154) groups. Baseline demographic and clinical characteristics, along with treatment patterns for both groups, are summarized in Table 1.

Table 1.

Baseline Characteristics of Patients According to ALBI Score.

Variable	Low ALBI Score (≤ −2.60) (n = 226)	High ALBI Score (> −2.60) (n = 154)	p‐value
Age, years	44.3 (18.2)	52.5 (17.4)	< 0.001
Gender, Female (n)	168 (74.3%)	115 (74.7%)	0.941
NYHA Class
1  2  3  4	4 (1.8%) 39 (17.3%) 136 (60.4%) 46 (20.4%)	0 (0) 2 (1.3%) 79 (51.6%) 72 (47.1%)	< 0.001
Subgroups, n (%)
IPAH  CHD‐PAH	91 (40.3%) 112 (49.6%)	79 (51.3%) 53 (34.4%)	0.034 0.003
6MWD, meters	300 (180–376)	180 (47.5–300)	< 0.001
Nt‐proBNP, pg/mL	357 (133–926)	825 (349–2196)	< 0.001
Baseline SpO2, %	95 (89–98)	93 (85.8–96)	0.019
REVEAL	8 (7–10)	10 (8.25–11)	< 0.001
REVEAL 2.0	7.5 (6–9)	10 (8–12)	< 0.001
REVEAL Lite	7 (6–9)	10 (8–11)	< 0.001
SPAHR
Low  Intermediate  High	97 (42.9) 121 (53.5) 8 (3.5)	31 (20.1) 100 (64.9) 23 (14.9)	< 0.001
COMPERA 3 strata
Low  Intermediate  High	83 (36.7%) 123 (54.4%) 20 (8.8%)	19 (12.3%) 96 (62.3%) 39 (25.3)	< 0.001
FPHN
0  1  2  3	91 (40.3%) 100 (44.2%) 25 (11.1%) 10 (4.4%)	96 (62.3%) 56 (36.4%) 2 (1.3%) 0 (0)	< 0.001
Hemoglobin, g/dL	13.8 (2.58)	13.0 (2.81)	0.008
Platelet count (x103/μL)	228 (85.4)	225 (90.3)	0.792
Albumin, g/dL	4.35 (0.38)	3.49 (0.45)	< 0.001
Bilirubin (total), mg/dL	0.73 (0.48)	1.09 (0.71)	< 0.001
AST, U/L	21.5 (18–26)	23.9 (19–32)	0.006
ALT, U/L	16 (12–22.8)	16.2 (11–24)	0.632
INR	1.17 (0.26)	1.26 (0.38)	0.006
Targeted PAH therapies (n)
Monotherapy  Double combination therapies  Triple combination therapies	75 (33.2) 74 (32.7) 77 (34.1)	48 (31.1) 55 (35.7) 31 (33.1)	0.742

Note: Patients were stratified into low (≤ −2.6) and high (> −2.6) ALBI score groups. Data are presented as mean (SD), median (IQR), or number (percentage), as appropriate. ALBI = Albumin‐Bilirubin; 6MWD = 6‐min Walk Distance; Nt‐proBNP = N‐terminal pro‐B‐type Natriuretic Peptide; NYHA = New York Heart Association; REVEAL = Registry to Evaluate Early and Long‐term PAH Disease Management; COMPERA = Comparative, Prospective Registry of Newly Initiated Therapies for Pulmonary Hypertension; FPHN = French Pulmonary Hypertension Network Risk Score.

Patients with a high ALBI score were significantly older (52.5 [17.4] vs. 44.3 [18.2] years, p < 0.001), while gender distribution was similar between groups (female: 74.7% vs. 74.3%, p = 0.941). Regarding etiology, CHD‐PAH was more frequent in the low ALBI group (49.6% vs. 34.4%, p = 0.003), whereas IPAH was more common in the high ALBI group (51.3% vs. 40.3%, p = 0.034).

Higher ALBI scores were associated with worse functional status (p < 0.001), with a greater proportion of patients in WHO class III and IV in the high ALBI group. The high ALBI was associated with significantly lower 6MWD, [median 180 (IQR 47.5–300 vs. 300 m (180–376), p < 00.001]and pulse oximetric SpO2 [median 95 (IQR 89–98) vs 93 (85.8–96), p = 0.019], and higher Nt‐proBNP level (pg/mL) [median 825 IQR (349–2196) vs. 357 (133–926), p < 0.001) as compared to those in low ALBI group. Laboratory parameters demonstrated consistently lower hemoglobin and albumin levels, and higher total bilirubin, AST and INR values in the high ALBI group reflecting impaired liver function (all p < 0.01).

Baseline risk status with MRSs was also significantly worse in the high ALBI group, and REVEAL, REVEAL 2.0, REVEAL Lite, SPAHR, COMPERA and FPHN scores showed a significant higher risk in association with elevated ALBI scores (p < 0.001).

No significant differences were observed between the low and high ALBI groups regarding the distribution of targeted PAH therapies, including mono, double, or triple combination regimens (p = 0.742).

3.2. Echocardiographic and Hemodynamic Assessments

Compared to the low ALBI cohort, patients in the high ALBI group demonstrated a greater prevalence of d‐shaped septum (83% vs. 73%, p = 0.023) and pericardial effusion (28.9% vs. 9.9%, p < 0.001). Echocardiographic markers of RV congestion and dysfunction were more pronounced in the high ALBI group, characterized by larger IVC diameter (2.06 ± 0.44 vs. 1.78 ± 0.46 cm, p < 0.001), shortened pulmonary artery acceleration time (PA ACT: 63.8 ± 16.2 vs. 71.8 ± 18.9 ms, p < 0.001), and reduced TAPSE (1.73 ± 0.5 vs. 1.95 ± 0.5 cm, p < 0.001). Additionally, RV S’ (11.1 ± 3.26 vs. 12.1 ± 3.13 cm/s, p = 0.009) and the TAPSE/PASP ratio (0.20 [IQR0.15–0.29] vs. 0.24 [IQR 0.17–0.34], p = 0.003) were significantly lower in the high ALBI group (Table 2).

Table 2.

Echocardiographic and Hemodynamic Parameters According to ALBI Score.

Variable	Low ALBI Score (≤ −2.60) (n = 226)	High ALBI Score (> −2.60) (n = 154)	p‐value
d‐shaped Septum(n)	162 (73%)	127 (83%)	0.023
Pericardial effusion (n)	22 (9.9%)	44 (28.9%)	< 0.001
TR Vmax, m/s	4.15 (0.73)	4.22 (0.71)	0.387
IVC, cm	1.78 (0.46)	2.06 (0.44)	< 0.001
PASP, mm Hg	77.9 (26.9)	83.0 (25.8)	0.085
PAMP, mm Hg	52.5 (15)	54.8 (14.9)	0.299
RA Area, cm2	23.7 (10.5)	24.5 (8.79)	0.224
MPA, cm	3.39 (0.88)	3.42 (0.80)	0.319
PA ACT, ms	71.8 (18.9)	63.8 (16.2)	< 0.001
TAPSE, cm	1.95 (0.5)	1.73 (0.5)	< 0.001
RV TDI, cm/s	12.1 (3.13)	11.1 (3.26)	0.009
TAPSE/PASP ratio	0.24 (0.17–0.34)	0.20 (0.15–0.29)	0.003
LMCA compression	20 (11.2%)	10 (8.4%)	0.427
PCWP, mm Hg	11.7 (4.28)	12 (3.98)	0.492
PASP, mm Hg	83 (29)	82.6 (29.6)	0.884
PAMP, mm Hg	51.6 (20)	50.7 (19.5)	0.681
PADP, mm Hg	32.4 (16.1)	32.2 (15.7)	0.866
RAP, mm Hg	8.32 (4.73)	10.2 (5.98)	0.005
CO, L/min	4.42 (1.05)	4.07 (1.30)	0.007
CI, L/min/m2	2.58 (0.63)	2.35 (0.66)	0.002
TPG, mm Hg	39.9 (20.5)	38.9 (20.0)	0.705
TSG, mm Hg	87.1 (17.5)	80.9 (17.7)	0.001
PVR, WU	8.76 (6.01)	10.1 (7.68)	0.112
SVR, WU	21 (7.19)	21.9 (7.15)	0.231
PVR/SVR	0.42 (0.26)	0.44 (0.25)	0.190

Note: Patients were stratified into low (≤ −2.6) and high (> −2.6) ALBI score groups. Data are presented as mean (SD) or number (percentage), as appropriate. ALBI = Albumin‐Bilirubin; TR Vmax = Tricuspid Regurgitation Maximum Velocity; IVC = Inferior Vena Cava; PASP = Pulmonary Artery Systolic Pressure; PAMP = Mean Pulmonary Artery Pressure; RA = Right Atrium; MPA = Main Pulmonary Artery; PA ACT = Pulmonary Artery Acceleration Time; TAPSE = Tricuspid Annular Plane Systolic Excursion; RV TDI = Right Ventricular Tissue Doppler Imaging Systolic Velocity; LMCA = Left Main Coronary Artery; PCWP = Pulmonary Capillary Wedge Pressure; PADP = Pulmonary Artery Diastolic Pressure; RAP = Right Atrial Pressure; CO = Cardiac Output; CI = Cardiac Index; TPG = Transpulmonary Gradient; TSG = Transseptal Gradient; PVR = Pulmonary Vascular Resistance; SVR = Systemic Vascular Resistance.

Hemodynamically, high versus low ALBI scores were associated with higher RAP (10.2 ± 5.98 vs. 8.32 ± 4.73 mmHg, p = 0.005), lower cardiac output (4.07 ± 1.30 vs. 4.42 ± 1.05 L/min, p = 0.007), and reduced cardiac index (2.35 ± 0.66 vs. 2.58 ± 0.63 L/min/m², p = 0.002) while other hemodynamic measures including PASP, PAMP, PVR, SVR and PVR/SVR ratio did not differ significantly between the groups.

3.3. Survival Analysis and Clinical Utility

The median follow‐up duration was 438 days (IQR: 78–1116 days). Kaplan‐Meier analysis demonstrated a profound and early divergence in mortality based on ALBI status (log‐rank p < 0.001). In the low and high ALBI cohorts, cumulative mortality rates at 6 months, 1 year, 2 years, 3 years, and 5 years were 10.9% versus 27.5%, 13.9% versus 41.7%, 17.9% versus 55.0%, 23.0% versus 61.7%, and 34.3% versus 67.2%, respectively (log‐rank p < 0.001) (Figure 2).

Figure 2.

Kaplan‐Meier Survival Analysis of 5‐year Overall Survival According to ALBI Group. The Kaplan‐Meier curves illustrate a significant divergence in 5‐year survival outcomes between patients in the low ALBI group (≤ −2.60, blue line) and the high ALBI group (> −2.60, red line). Cumulative mortality was substantially higher in the high ALBI cohort, reaching 41.7% at 1 year and 67.2% at 5 years, compared to 13.9% and 34.3% in the low ALBI cohort, respectively (log‐rank p < 0.001).

3.4. Prognostic Performance and Comparison With Other Markers

In univariate Cox regression analysis, both continuous ALBI [HR: 2.82 (2.14–3.72), p < 0.001] and binary ALBI [HR: 3.23 (2.27–4.58), p < 0.001] were strong predictors of mortality (Figure 3). Notably, ALBI exhibited the highest discriminative power among all liver‐related markers (C‐index: 0.710). After adjusting for the REVEAL 2.0 risk score, ALBI remained the strongest independent predictor of mortality, either with continuous [adjusted HR: 1.80 (1.30–2.49), p < 0.001, C‐index :0.756] or binary [1.96 (1.35–2.82) p < 0.001, C‐index: 0.754] definitions. The integration of ALBI continuous and ALBI binary definitions into the REVEAL 2.0 framework yielded the higher adjusted C‐index values outperforming prognostic performance of other liver dysfunction scores including MELD, MELD‐Na, MELD 3.0, and FIB‐4—ALBI (Table 3, Figure 4A).

Figure 3.

Adjusted Hazard Ratio Curves and Comparative Prognostic Performance of Hepatic Dysfunction Markers. The panels illustrate the relationship between liver‐related scores and the adjusted Hazard Ratio (HR) for 5‐year mortality, with all models corrected for the continuous REVEAL 2.0 risk score. (A) Adjusted HR curve for the ALBI score using Restricted Cubic Splines (RCS), demonstrating a significant non‐linear relationship with mortality risk (p for non‐linearity < 0.05). (B) Adjusted HR curve for the FIB‐4 index using RCS, which also exhibited significant non‐linearity (p for non‐linearity < 0.05). (C) Comparative HR plots for MELD‐derived scores (MELD, MELD‐Na, and MELD 3.0), where a predominantly linear relationship with mortality was observed. Across all comparisons, the ALBI‐augmented model achieved the highest discriminative performance. Abbreviations: ALBI, Albumin‐Bilirubin; FIB‐4, Fibrosis‐4 Index; MELD, Model for End‐Stage Liver Disease; RCS, Restricted Cubic Spline; HR, Hazard Ratio.

Table 3.

Univariate and Multivariable Cox Regression Analysis of Liver‐Based Markers.

Marker	Univariate HR (95% CI)	p value	C‐Index	Adjusted HR (95% CI)	p value	Adjusted C‐index
REVEAL 2.0	1.33 (1.25–1.41)	< 0.001	0.736	—	—	—
ALBI	2.82 (2.14–3.72)	< 0.001	0.710	1.80 (1.30–2.49)	< 0.001	0.756
ALBI Binary	3.23 (2.27–4.58)	< 0.001	0.651	1.96 (1.35–2.82)	< 0.001	0.754
MELD	1.13 (1.08–1.17)	< 0.001	0.633	1.07 (1.02–1.11)	0.002	0.750
MELD‐Na	1.11 (1.08–1.15)	< 0.001	0.639	1.05 (1.01–1.09)	0.006	0.747
MELD 3.0	1.12 (1.07–1.16)	< 0.001	0.641	1.06 (1.01–1.11)	0.012	0.746
FIB‐4	1.25 (1.14–1.38)	< 0.001	0.627	1.20 (1.08–1.32)	0.001	0.746

Note: Adjusted models were corrected for the continuous REVEAL 2.0 risk score.

Abbreviations: ALBI, albumin‐bilirubin; CI, confidence interval; C‐Index, concordance index; FIB‐4, fibrosis‐4 index; HR, hazard ratio; MELD, model for end‐stage liver disease.

Figure 4.

Incremental Predictive Performance and Model Comparison of ALBI‐Augmented Risk Stratification. This figure quantifies the statistical improvement gained by integrating hepatic markers into the baseline REVEAL 2.0 risk model. (A) Bar plot illustrating model parsimony and discriminative gain. (B) Time‐dependent ROC curves, demonstrating that the superior discriminative accuracy of the ALBI‐augmented model is consistently maintained over the 5‐year follow‐up period compared to all other liver‐based and baseline models. AIC, Akaike Information Criterion; ALBI, Albumin‐Bilirubin; C‐index, Concordance Index; FIB‐4, Fibrosis‐4 Index; MELD, Model for End‐Stage Liver Disease; ROC, Receiver Operating Characteristic.

Furthermore, time‐dependent ROC analysis confirmed the superior discriminative accuracy of the ALBI‐augmented model, maintaining consistently higher AUC values across the entire 5‐year follow‐up period compared to baseline and other liver‐based models (Figure 4B).

DCA confirmed the clinical utility of this stratification, as the ALBI‐augmented REVEAL 2.0 model provided superior net benefit across all relevant threshold probabilities compared to the baseline score alone (Supplementary Figure 1). Finally, comparative model analyses revealed that adding ALBI to REVEAL 2.0 provided higher risk discrimination (C‐index: 0.764) than adding direct echocardiographic indices of RV function or coupling, such as TAPSE or TAPSE/PASP (C‐indices: 0.749 and 0.748, respectively). In multivariable sensitivity analyses, the ALBI score maintained its independent prognostic value across all models, including the fully adjusted model incorporating TAPSE/PASP, invasive cardiac index, mean RAP, and background PAH therapy [adjusted HR: 1.44 (95% CI: 1.01–2.05), p = 0.044; model C‐index: 0.801] (Supplementary Table 1).

3.5. Subgroup Analysis: Intermediate‐Risk Cohort

The prognostic impact of ALBI status was most marked in the intermediate‐risk cohort, and high versus low ALBI score related to three‐fold increased risk of death [HR: 3.04 (95% CI: 1.25–7.38), p = 0.014] (Figure 5).

Figure 5.

Impact of ALBI Score on Risk Stratification within the REVEAL 2.0 Intermediate‐Risk Group. This figure focuses on the prognostic utility of the ALBI score in the “gray zone” of risk assessment. (A) Kaplan‐Meier survival curves showing effective re‐stratification of intermediate‐risk patients; those with high ALBI exhibited significantly higher mortality (35.3% vs. 14.8%, p = 0.014). (B) Cardio‐Hepatic Risk Radar; a circular plot illustrating the combined impact of ALBI status and REVEAL risk categories on patient survival. (C) Incremental predictive power; bar plots demonstrating that adding ALBI to the continuous REVEAL 2.0 score improved discriminative performance by 26.3% (C‐index increase from 0.558 to 0.704), significantly refining risk prediction in this subgroup. ALBI, Albumin‐Bilirubin; REVEAL, Registry to Evaluate Early and Long‐term PAH Disease Management.

Furthermore, adding ALBI to the continuous REVEAL 2.0 score resulted in a 26.3% relative improvement in discriminative performance (C‐index: 0.558 to 0.704; LRT p = 0.015), addressing the risk uncertainty in this clinical “gray zone.”

4. Discussion

Our main finding is that the ALBI score independently predicts long‐term mortality in PAH, providing better prognostic accuracy than traditional liver markers like MELD, MELD 3.0, MELD‐Na, and FIB‐4. Hemodynamically, higher ALBI levels are associated with more severe right heart failure, showing direct links to worse RV systolic function, higher RAP, inferior vena cava plethora, and lower cardiac index. The real clinical value of ALBI, however, came from adding it to REVEAL 2.0. This integration significantly improved risk prediction, especially in the intermediate‐risk ‘gray zone’ where prognostic decisions are often the hardest. In this subgroup, a high ALBI score singled out patients with a three‐fold higher mortality risk and boosted the model's C‐index by 26.3%.

In patients with PH, several liver‐related risk scores have been proposed to quantify cardio‐hepatic involvement and predict long‐term outcomes. Among these are the modified Child–Pugh score, MELD and its derivatives (MELD‐XI, MELD‐Na, and MELD 3.0), the LFPI, the FIB‐4 index, and more recently the ALBI score [11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35]. Among these, MELD and its variants capture different aspects of hepatic and systemic dysfunction. A MELD‐XI variation removes the INR component as a confounder in patients on anticoagulant treatments with vitamin K antagonists, and score ≥ 12 is significantly associated with increased mortality risk in critically ill patients with PH [21, 22, 23]. MELD‐Na score mainly targeted to hyponatremia as a marker of vicious cycle caused by activation of the renin‐angiotensin‐system in advanced HF and CTEPH [30, 33, 34]. Although MELD provides insights for multiorgan failure, its components are susceptible to confounders such as chronic liver disease and acute kidney injury, limiting its specificity in PAH populations [20, 21, 22, 23, 24, 25]. Moreover, the contribution of liver as an inflammatory mediator to the pulmonary endothelium was hypothesized in the pathogenesis of PAH [34].

The ALBI score is a simpler yet biologically meaningful hepatic function assessment tool that reflects the interactions between hepatic congestion, perfusion, and systemic disease severity in HF and PH [27, 28, 29, 31]. Hepatic congestion due to elevated central venous pressures and reduced hepatic perfusion pressures secondary to low cardiac output states eventually impair synthesis of albumin and excretion phase of conjugated bilirubin [8, 9, 10, 27, 28, 29, 31]. The ALBI has been reported to correlate with increased RAP, reduced cardiac index and lower mixed venous oxygen saturation levels reflecting the high‐pressure/low‐flow status of cardio‐hepatic axis in HF [27, 28, 29, 31]. Importantly, beyond hepatic congestion, the ALBI score may also reflect systemic vulnerability, including nutritional status and frailty, both of which may contribute to adverse outcomes in advanced cardiopulmonary disease. Therefore, this score may integrate several consequences of systemic arterial and venous hemodynamic disturbances into a single composite index, and can predict increased risk for clinical deteriorations, hospitalizations and mortality in patients with HF and PH [27, 28, 29, 31]. When added to NT‐proBNP, ALBI score can enhance the predictive potential of NT‐proBNP in HF patients, and was found to be an independent predictor for in‐hospital mortality [29, 31]. Moreover, ALBI score was reported to discriminate the patients with severe Porto‐PH among the patients with chronic liver disease [31]. However, MELD, MELD‐XI, MELD‐Na and ALBI scores are not PH‐specific markers [20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34], and ALBI score may also be influenced by malnutrition, chronic liver disease and systemic inflammation [26, 27, 28, 29, 30, 31]. Because of these confounders, these scores have not been considered to replace well‐established cardiac‐specific multiparametric risk score models in PH [6, 11, 12, 13, 14, 15, 16, 17, 18, 19]. Rather, as demonstrated in the present study, ALBI offers complementary prognostic information when integrated into established frameworks such as REVEAL 2.0.

While scores of hepatic dysfunction reflect the severity of right‐sided HF, direct comparisons of their prognostic power have remained underexplored. Our results confirmed that higher ALBI scores were consistently associated with a worsening hemodynamic profile represented by reduced RV systolic function, increased prevalence of pericardial effusion, increased RAP and inferior vena cava plethora, reduced cardiac index, and narrowing in the pressure difference between mean systemic arterial pressure and mean RAP as an essential, but ignored, driving force in maintaining the blood circulation throughout the arterial and venous systems. In mortality prediction, ALBI demonstrated the highest discriminatory performance among the evaluated liver‐based markers, including MELD, MELD 3.0, MELD‐Na and FIB‐4. This superior performance may relate to the fact that ALBI, by combining albumin and bilirubin, captures both the synthetic and excretory consequences of hepatic congestion in a single index, without the confounding effect of INR or renal function that complicates MELD‐based scores [20, 21, 22, 23, 24, 25, 27, 28, 29, 31]. Although the baseline discriminative performance of REVEAL 2.0 in our cohort was lower than originally reported, likely owing to single‐center mechanics, prevalent enrollment, local treatment practices, and a substantial subgroup of congenital heart disease‐associated PAH, integrating ALBI consistently enhanced risk prediction. Adding ALBI to REVEAL 2.0 provided greater prognostic discrimination than incorporating standard echocardiographic parameters of right ventricular function or coupling, such as TAPSE and TAPSE/PASP. This independent predictive value persisted after adjusting for TAPSE/PASP, invasive cardiac index, RAP, and background PAH therapy. These sensitivity analyses demonstrate that ALBI does not only track mechanical RV failure or central hemodynamics; rather, it captures incremental prognostic information driven by systemic visceral congestion and cardio‐hepatic uncoupling. Consequently, ALBI appears to reflect the cumulative end‐organ burden of chronic right heart failure rather than isolated pulmonary vascular load at a single time point, marking out a patient subgroup with a distinctly adverse clinical and hemodynamic trajectory. Whether this represents a discrete hepatic phenotype or a systemic manifestation of end‐stage RV failure remains to be clarified in prospective longitudinal cohorts. One of the most clinically relevant findings of the present study was the ability of the ALBI score to refine risk stratification within the REVEAL 2.0 intermediate‐risk category. Despite advances in multiparametric risk assessment, patients classified as intermediate risk remain a heterogeneous group in whom the decision to escalate therapy is often uncertain, and for whom current guidelines provide limited discriminatory tools. The present findings suggest that the ALBI score, derived from two routinely available laboratory parameters, may help resolve this ambiguity by identifying a subset of intermediate‐risk patients who carry a disproportionately high mortality burden. The primary clinical advantage of the ALBI score lies in its simplicity and accessibility. Derived solely from routine laboratory parameters, it avoids the costs and risks of repeated invasive hemodynamics or complex imaging, making it easily adaptable to everyday practice. Integrating ALBI into established risk frameworks offers a practical means to refine prognostic assessment, helping clinicians identify intermediate‐risk patients who carry an unrecognized mortality burden and may benefit from timely treatment escalation. Although patients with portopulmonary hypertension were excluded from the present analysis to avoid confounding from underlying liver disease, findings from the GoDeep Metaregistry further support the broader prognostic relevance of hepatic dysfunction across the pulmonary vascular disease spectrum. In that registry, the survival benefit associated with PAH‐targeted therapies persisted even in patients with severe liver disease, as defined by a MELD score ≥ 13, and was not primarily explained by baseline PVR or cardiac index [25]. Together with our findings, these observations suggest that hepatic dysfunction may capture clinically relevant dimensions of disease severity that are not fully reflected by conventional pulmonary hemodynamic parameters alone. In contrast, the strong association between ALBI and reduced CI, increased RAP, impaired RV longitudinal systolic function measures, enlarged IVC diameter, and increased prevalence of pericardial effusion seems to suggest that ALBI score, through its components of albumin and bilirubin, may reflect the combined hepatic consequences of back‐ward pressure and volume overload at hepatic veins, and low‐flow state at the hepatic arterial circulation level. Similar physiopathology seems to be observed as cardio‐pulmonary‐renal interactions in patients with pulmonary hypertension, and an integrative approach regarding cardio‐pulmonary‐renal interactions may provide a more comprehensive perspective for circulatory status and renal function in this setting [39]. More importantly, even small increases of serum creatinine levels within the normal range seems to be associated long‐term survival differences in these patients [39]. Moreover, using machine learning‐based clustering analysis, phenotypic clustering of pulmonary vascular and cardiometabolic co‐morbidity trajectories seems to provide new insights for longitudinal therapeutic responses and long‐term survival in patients with idiopathic pulmonary arterial hypertension [40].

4.1. Limitations

Despite its significant findings, this study has several limitations that warrant consideration. First, its retrospective and single‐center design may introduce selection bias and limit the generalizability of the results to broader, more diverse PAH populations. Second, while we excluded patients with documented liver disease — including cirrhosis, portal hypertension, chronic viral hepatitis, and related disorders — as well as those with portopulmonary hypertension, the potential for subclinical or undiagnosed hepatic pathologies to influence ALBI scores cannot be entirely ruled out. Third, our analysis was based on baseline assessments; thus, it does not account for longitudinal changes in ALBI scores over time. Investigating how serial changes in ALBI correlate with treatment escalations, NT‐proBNP fluctuations, and hemodynamic remodeling could provide deeper insights into its utility as a dynamic monitoring tool. Fourth, although RHC was performed, the study lacks a direct comparison with advanced congestion markers such as Venous Excess Ultrasound (VExUS) scores, which might offer a more granular grading of hepatic congestion in the context of RV failure [41, 42]. Fifth, only baseline hemodynamic measurements were available in the present study. Serial right heart catheterization data were not systematically obtained, precluding assessment of longitudinal changes in pulmonary vascular resistance and their relationship with hepatic dysfunction. Consequently, potential temporal dissociation between pulmonary vascular remodeling and cumulative hepatic injury could not be evaluated. Sixth, although the distribution of PAH‐targeted therapies did not differ significantly between ALBI groups (p = 0.742), residual confounding related to treatment intensity, treatment response, or indication bias cannot be completely excluded because of the retrospective nature of the study. Therefore, the observed dissociation between ALBI and PVR should be interpreted with caution and requires confirmation in prospective longitudinal cohorts. Finally, while the ALBI‐augmented REVEAL 2.0 model demonstrated superior discriminative power, external validation in independent prospective cohorts is necessary to confirm the robustness of the “cardio‐hepatic phenotype” identified in the intermediate‐risk “gray zone.” Future studies should also explore the potential of the ALBI score to guide early aggressive therapy in patients showing early or subclinical signs of cardio‐hepatic deterioration.

4.2. Conclusion

The ALBI score is a simple and reliable surrogate biomarker that captures the complex cardio‐hepatic interactions and independently predicts long‐term mortality in patients with PAH. The ALBI outperforms the conventional liver‐based scores including MELD, MELD 3.0, MELD Na and FIB‐4. Its integration into the REVEAL 2.0 framework significantly refines the risk stratification across the intermediate‐risk “gray zone,” and allows for the earlier identification of higher risk individuals who may benefit from more proactive therapeutic strategies. These findings support the routine use of the ALBI score as a complementary tool for individualization of risk‐based treatment algorithms in PAH.

Author Contributions

Hacer Ceren Tokgoz: Conceptualization, study design, clinical data curation, patient follow‐up, medical record review, verification of longitudinal hemodynamic parameters, interpretation of data, and drafting the initial manuscript. Cihangir Kaymaz: Senior clinical supervision, study oversight, conceptual guidance, interpretation of data, and critical revision of the manuscript for important intellectual content. Barkin Kultursay: Conceptualization, study design, methodology, advanced statistical analysis, data curation, interpretation of results, manuscript drafting, and critical revision of the final manuscript. Nihal Ozdemir: Study oversight, clinical validation of registry data, and critical revision of the final manuscript. Dicle Sırma; Critical revision of the final manuscript. Seda Tanyeri, Dicle Sırma, Cagdas Bulus, Tezel Kovanci, Mustafa Kara, Seyma Zeynep Atici, Seyma Nur Cicek, Azmican Kaya, Faruk Akalin, Dogan Sen, Leyla Aliyeva: Clinical data collection, patient follow‐up, medical record review, and verification of longitudinal hemodynamic parameters. Caglar Emre Cagliyan: Data validation, quality control, and providing critical feedback on the manuscript. All authors contributed to the work, reviewed the final version of the manuscript, approved it for submission, and agreed to be accountable for all aspects of the study.

Funding

The authors have nothing to report.

Ethics Statement

Ethical approval was obtained from the Kartal Kosuyolu High Specialization Training and Research Hospital Ethics Committee (Approval no: 2013.3/4, Date: 31.05.2013). The data that support the findings of this study are available on request from the corresponding author. The data are not publicly available due to the privacy of research participants. The author(s) affirm that the methods used in the data analyses are suitably applied to their data within their study design and context, and the statistical findings have been implemented and interpreted correctly. The author(s) agrees to take responsibility for ensuring that the choice of statistical approach is appropriate and is conducted and interpreted correctly as a condition to submit to the Journal. There is a researcher Dr. Barkin Kultursay who has statistical knowledge of the author team. The author(s) agrees to take responsibility for ensuring that the choice of statistical approach is appropriate and is conducted and interpreted correctly as a condition to submit to the Journal.

Conflicts of Interest

The authors declare no conflicts of interest.

Guarantor

Prof. Dr. Cihangir Kaymaz accepts full responsibility for the integrity of the data and the accuracy of the data analysis, and serves as the guarantor for this manuscript.

Supporting information

Supporting File 1

Supporting File 2

Data Availability Statement

The data that support the findings of this study are available on request from the corresponding author. The data are not publicly available due to privacy or ethical restrictions.