Introduction
Pulmonary arterial hypertension (PAH) is defined as mean pulmonary arterial pressure (mPAP) > 20 mmHg at rest as assessed by right heart catheterization, pulmonary arterial wedge pressure ≤ 15 mmHg, and pulmonary vascular resistance > 2 wood units, according to the classification of 2022 European Society of Cardiology (ESC)/European Respiratory Society (ERS) guidelines [1]. Pathologic progressions of vascular remodeling leads to pulmonary hypertension, right-sided heart failure, and death, once compensatory mechanisms have failed [2–4].
Most of the ESC/ERS recommended multiparameters for risk assessment and outcome prediction are invasive hemodynamic measurements or effort-dependent exercise tests except serum natriuretic peptide, B-type natriuretic peptide (BNP), or N-terminal prohormone of B-type natriuretic peptide (NT-proBNP), which is only one type of effort-free biomarker that has been adopted for risk assessment [5]. However, any kind of effort-dependent exercise tests are limited by a patient’s physical restriction or exercise disability, an alternative or additional biomarker could provide more information of outcomes if invasive or effort-dependent tests are not accessible or available.
There are many circulating biomarkers involved in functional pathways associated with the pathobiology of pulmonary hypertension. Homocysteine is one of these circulating biomarkers, which involved several pathological functional pathways of PAH, and was considered to be correlated with diagnosis and prognosis of PAH [6, 7]. Elevated homocysteine levels have been toxic to the vascular endothelium and is an attribute for coronary disease, cerebrovascular disease, and peripheral vascular disease [8, 9]. However, the relationship between homocysteine and pulmonary hypertension remains unclear. This study aimed to investigate the application of homocysteine as an alternative or additional non-invasive and effort-free measurement in addition to serum natriuretic peptide for risk assessment of patients with pulmonary hypertension.
Methods
Samples in this study were obtained from the Kaohsiung Veterans General Hospital Biobank with approval from the respective ethics committees of Kaohsiung Veterans General Hospital. Deidentified data of patients diagnosed with pulmonary hypertension were analyzed to establish the association between circulating biomarkers and the risk levels of pulmonary hypertension. Pulmonary hypertension was defined as a mPAP ≥ 20 mmHg at rest, as assessed by right heart catheterization according to 2022 ESC/ERS guidelines [1].
Animal model
A monocrotaline (MCT)-induced PAH rat model was used in this study, and the Institutional Animal Care and Use Committee of Kaohsiung Veterans General Hospital approved the experimental protocols. Six-week-old male Sprague-Dawley rats in 220–280 g, were purchased from BioLASCO (Ilan, Taiwan) and handled according to the IACUC guidelines. To establish the MCT-induced PAH model, Sprague-Dawley rats were injected intraperitoneally with 60 mg/kg MCT (Sigma-Aldrich, St. Louis, MO, USA) as previously described [10, 11]. At the 1st, 2nd, 3rd, and 4th weeks, rat venous blood was drawn for analysis. On day 28, the animals were sacrificed, and PAH pathology was assessed as described previously [12]. All experimental protocols were performed in accordance with the European ethical regulation (Directive 2010/63/EU) and approved by the Institutional Animal Care and Use Committee, Kaohsiung Veterans General Hospital, Taiwan (Ref. 2019-2021-A054).
Serum NT-proBNP and homocysteine level of MCT rats
Rats were treated with phosphate-buffered saline or MCT (60 mg/kg) for 7, 14, 21, and 28 days. Blood samples were collected from the tail vein of the rats. Serum NT-proBNP and homocysteine concentrations were measured using ELISA kits (MBS2881463, MyBioSource, Inc., San Diego, CA, USA for NT-proBNP; MBS703069, MyBioSource, Inc., San Diego, CA, USA for homocysteine) according to the manufacturer’s instructions.
Information about hemodynamic measurements of MCT rats, histology and immunohistochemical analysis of pulmonary arteries, blood tests assay for human, multiplex immunoassay of human blood, hemodynamics and cardiopulmonary function tests of human, and risk level assessment are presented in the Supplementary Appendix.
Ethics statement
The Institutional Review Board (IRB) of Kaohsiung Veterans General Hospital approved this study (No. KSVGH21-CT9-04). Written informed consent was not required for this study as the Biobank research database consisted of de-identified secondary data for research purposes. The IRB of Kaohsiung Veterans General Hospital issued a formal written waiver of the requirement for informed consent.
Statistical analyses
SPSS version 22 (IBM Corp., Armonk, NY, USA) was used for data analysis. Percentile values were used to express categorical data and were analyzed using the chi-square test. Mean (μ) and standard deviation (SD) values were used for continuous variables using the Student unpaired test. Multiplex immunoassay biomarkers were analyzed by one-way analysis of variance (ANOVA) with Bonferroni correction, and statistical significance was defined as p < 0.05 after verifying the equality of variances.
Univariate and multivariate forward stepwise logistic regression analyses were performed to assess predictors for the high-risk group, and the odds ratios (OR) and the associated 95% confidence intervals (CI) for significant variables were calculated, and statistically significant predictor was set at p < 0.05. Correlation analysis was performed to assess the correlation between the biomarkers and NT-proBNP levels. To compare NT-proBNP and homocysteine levels with increasing severity of pulmonary hypertension by weeks following MCT infusion, ANOVA with post-hoc Fisher’s least significant difference test was adopted after verifying the equality of variances. In addition, statistical significance was set at p < 0.05.
To find the most appropriate cut-off value for selective biomarker to determine the risk level for pulmonary hypertension, a receiver operating characteristic (ROC) analysis was performed. Moreover, different biomarkers combinations and the comparison between respective predictive value of each model were illustrated. The areas under the curves (AUC) were calculated.
Results
The basic characteristics of patients with pulmonary hypertension based on the ESC/ERS guideline-recommended risk assessment are reported in Table 1 [13]. There were 3 patients in low-risk group, 24 intermediate-risk patient, and 23 patients in high-risk group. There were no disparities in sex and age between the low/intermediate-risk and high-risk groups. Biochemistry panel demonstrated worse renal function blood urea nitrogen = 15.0 ± 4.5 vs. 24.0 ± 13.7 mg/dL, p = 0.006; serum creatinine = 0.9 ± 0.2 vs. 1.2 ± 0.5 mg/dL, p = 0.030) in high-risk group. With regard to circulating biomarkers, higher homocysteine (10.6 ± 4.0 vs. 17.0 ± 7.0 µmol/L, p = 0.005, Fig. 1A), uric acid (UA; 6.0 ± 1.7 vs. 7.7 ± 2.5 mg/dL, p = 0.006, Fig. 1B), D-dimer (744.8 ± 579.1 vs. 1,525.5 ± 1,559.7 ng/mL, p = 0.040, Fig. 1C), and C-reactive protein (CRP; 0.7 ± 0.7 vs. 2.6 ± 2.7 mg/dL, p = 0.007, Fig. 1D) were observed in the high-risk group. Despite no significant differences of multiplex immunoassay circulating biomarkers, including angiopoietin-2, bone morphogenetic protein (BMP)-2, BMP-4, cluster of differentiation 40 (CD40), endoglin, interlukin-6, myeloperoxidase, osteopontin, and vascular endothelial growth factor (VEGF), there was an increased trend by disease severities. Furthermore, Bonferroni correction was applied for analysis of multiplex immunoassay biomarkers (Suppl. Table S1), and the insignificance could be attributed to the small sample size.
Hemodynamics and cardiopulmonary function tests for pulmonary hypertension risk assessment based on the ESC/ERS guidelines are also listed in Table 1. Compared to reports in low/intermediate-risk group, the high-risk group was reported to have worse World Health Organization (WHO) functional (Fc III = 11.1% vs. 73.9%, p < 0.001), worse exercise and cardiopulmonary exercise capacity (six-minute walking distance [6MWD] =367.7 ± 102.6 vs. 251.4 ± 143.0 m, p < 0.001; VE/VCO2 = 32.8 ± 7.4 vs. 41.3 ± 14.9, p = 0.049), higher NT-proBNP value (NT-proBNP = 794.5 ± 918.5 vs. 4390.6 ± 4843.6 pg/mL, p = 0.002). Regarding hemodynamic parameters, patients in the high-risk group had worse cardiac function (cardiac output = 5.4 ± 0.6 vs. 4.1 ± 1.5 L/min, p = 0.028; cardiac index = 3.6 ± 0.5 vs. 2.4 ± 0.9 L/min/m2, p = 0.001), worse vascular saturation (pulmonary artery saturation = 73.3 ± 4.8 vs. 50.7 ± 15.1%, p = 0.007; superior vena cava saturation = 71.3 ± 6.6 vs. 57.5 ± 12.0%, p < 0.001; inferior vena cava saturation = 74.5 ± 8.7 vs. 54.9 ± 14.1%, p =0.005), and higher pulmonary vascular resistance (6.0 ± 3.4 vs. 10.9 ± 8.6 woods, p =0.034) compared to the reports of patients in the low/intermediate-risk group. With regard to pulmonary function tests, forced expiratory volume in the first second (FEV1) and FVC (FEV1 = 1.9 ± 0.8 vs. 1.4 ± 0.5% predicted, p =0.024; FVC = 2.3 ± 1.2 vs. 1.7 ± 0.6 L, p = 0.020) were lower in the high-risk group than in the low/intermediate-risk group.
pulmonary hypertension based on risk levels.
|
Variables |
Low/intermediate risk |
High risk |
P |
|
Female |
22.0 (81.5%) |
20.0 (87.0%) |
0.711 |
|
Age [years] |
56.4 ± 14.7 |
63.7 ± 15.8 |
0.141 |
|
Body weight [kg] |
60.4 ± 12.5 |
68.6 ± 25.0 |
0.166 |
|
Body height [cm] |
156.5 ± 8.6 |
151.0 ± 22.5 |
0.279 |
|
Body surface area [m2] |
1.6 ± 0.2 |
1.7 ± 0.2 |
0.521 |
|
Hematology tests: |
|||
|
White blood cells [K/µL] |
6.8 ± 2.2 |
6.8 ± 2.2 |
0.992 |
|
Red blood cells [M/µL] |
4.5 ± 0.5 |
4.6 ± 0.9 |
0.630 |
|
Hemoglobin [g/dL] |
13.3 ± 1.7 |
13.3 ± 1.8 |
0.952 |
|
Hematocrit [%] |
40.1 ± 4.3 |
41.1 ± 6.0 |
0.510 |
|
Red blood cell volume distribution [%] |
14.8 ± 4.8 |
15.9 ± 4.3 |
0.424 |
|
Platelet [K/µL] |
250.5 ± 96.2 |
204.8 ± 73.2 |
0.069 |
|
Neutrophil [%] |
62.1 ± 13.6 |
65.4 ± 10.9 |
0.349 |
|
Lymphocyte [%] |
28.2 ± 11.9 |
23.0 ± 10.1 |
0.108 |
|
Neutrophil/Lymphocyte ratio |
3.2 ± 3.1 |
3.6 ± 2.2 |
0.625 |
|
Prothrombin time [s] |
11.1 ± 1.3 |
18.1 ± 22.8 |
0.158 |
|
International normalized ratio |
1.0 ± 0.1 |
1.2 ± 0.6 |
0.120 |
|
Partial thromboplastin time [s] |
31.0 ± 4.6 |
30.5 ± 6.9 |
0.779 |
|
Biochemistry panel: |
|||
|
Sodium [mmol/L] |
141.2 ± 3.1 |
139.2 ± 3.9 |
0.043 |
|
Blood urea nitrogen [mg/dL] |
15.0 ± 4.5 |
24.0 ± 13.7 |
0.006 |
|
Serum creatinine [mg/dL] |
0.9 ± 0.2 |
1.2 ± 0.5 |
0.030 |
|
Estimated GFR [mL/min/1.73 m2] |
73.6 ± 14.3 |
62.2 ± 26.1 |
0.070 |
|
Fasting plasma glucose level [mg/dL] |
100.2 ± 13.9 |
98.1 ± 32.7 |
0.796 |
|
Aspartate aminotransferase [U/L |
30.7 ± 20.4 |
27.8 ± 12.3 |
0.542 |
|
Alanine aminotransferase [U/L] |
25.9 ± 15.2 |
25.2 ± 18.2 |
0.874 |
|
Alkaline phosphatase [U/L] |
62.0 ± 32.4 |
79.3 ± 32.6 |
0.105 |
|
Total bilirubin [mg/dL] |
0.7 ± 0.5 |
0.9 ± 0.7 |
0.234 |
|
Albumin [g/dL] |
4.1 ± 0.6 |
3.8 ± 0.6 |
0.070 |
|
Lactate dehydrogenase [U/L] |
214.7 ± 89.2 |
212.6 ± 34.2 |
0.925 |
|
Lipid profile: |
|||
|
Total cholesterol [mg/dL] |
178.2 ± 40.9 |
162.3 ± 32.1 |
0.146 |
|
High-density lipoprotein [mg/dL] |
50.5 ± 17.5 |
46.5 ± 16.2 |
0.413 |
|
Low-density lipoprotein [mg/dL] |
96.0 ± 25.7 |
98.6 ± 30.6 |
0.751 |
|
Triglyceride [mg/dL] |
120.3 ± 65.7 |
96.4 ± 37.3 |
0.154 |
|
Multiplex immunoassay circulating biomarkers: |
|||
|
Angiopoietin-2 [pg/mL] |
6237.3 ± 4790.3 |
5871.0 ± 5029.6 |
0.793 |
|
BMP-2 [pg/mL] |
14.6 ± 0.0 |
12.4 ± 2.1 |
0.074 |
|
BMP-4 [pg/mL] |
4.5 ± 0.6 |
5.1 ± 1.2 |
0.084 |
|
CD40 [pg/mL] |
1689.6 ± 923.9 |
1666.3 ± 817.7 |
0.926 |
|
Endoglin [pg/mL] |
1319.1 ± 541.6 |
1411.1 ± 433.6 |
0.522 |
|
Interlukin-6 [pg/mL] |
2.5 ± 1.6 |
3.6 ± 6.5 |
0.452 |
|
Myeloperoxidase [pg/mL] |
6476.3 ± 1897.4 |
6345.4 ± 1753.7 |
0.802 |
|
Osteopontin [pg/mL] |
28901.9 ± 13600.6 |
35805.0 ± 34042.2 |
0.369 |
|
VEGF [pg/mL] |
34.7 ± 23.8 |
26.2 ± 13.8 |
0.141 |
|
von Willebrand factor [%] |
161.9 ± 53.5 |
164.8 ± 60.5 |
0.890 |
|
World Health Organization functional class III |
3.0 (11.1%) |
17.0 (73.9%) |
< 0.001 |
|
Six-minute walking distance [m] |
367.7 ± 102.6 |
251.4 ± 143.0 |
< 0.001 |
|
Cardiopulmonary exercise testing: |
|||
|
Peak oxygen consumption [mL/min/kg] |
74.4 ± 28.0 |
65.6 ± 25.4 |
0.315 |
|
VE/VCO2 |
32.8 ± 7.4 |
41.3 ± 14.9 |
0.049 |
|
NT-proBNP [pg/mL] |
794.5 ± 918.5 |
4390.6 ± 4843.6 |
0.002 |
|
Hemodynamics: |
|||
|
Heart rate [bpm] |
83.5 ± 16.5 |
85.9 ± 13.0 |
0.583 |
|
Right atrial pressure [mmHg] |
11.7 ± 3.8 |
13.9 ± 6.3 |
0.235 |
|
Cardiac output [L/min, Thermodilution method] |
5.4 ± 0.6 |
4.1 ± 1.5 |
0.028 |
|
Cardiac index [L/min/m2, Thermodilution method] |
3.6 ± 0.5 |
2.4 ± 0.9 |
0.001 |
|
Cardiac output [L/min, Fick formula] |
4.3 ± 1.3 |
3.5 ± 1.4 |
0.162 |
|
Cardiac index [L/min/m2, Fick formula] |
2.7 ± 0.8 |
2.1 ± 0.8 |
0.082 |
|
Pulmonary artery saturation [%] |
73.3 ± 4.8 |
50.7 ± 15.1 |
0.007 |
|
Superior vena cava saturation [%] |
71.3 ± 6.6 |
57.5 ± 12.0 |
< 0.001 |
|
Inferior vena cava saturation [%] |
74.5 ± 8.7 |
54.9 ± 14.1 |
0.005 |
|
Mean arterial pressure [mmHg] |
97.7 ± 12.1 |
98.6 ± 12.3 |
0.802 |
|
Mean pulmonary arterial pressure [mmHg] |
36.5 ± 14.1 |
43.6 ± 11.7 |
0.082 |
|
Pulmonary arterial wedge pressure [mmHg] |
8.7 ± 6.5 |
11.0 ± 7.0 |
0.238 |
|
Pulmonary vascular resistance [woods] |
6.0 ± 3.4 |
10.9 ± 8.6 |
0.034 |
|
Left ventricular ejection fraction [%] |
59.8 ± 4.0 |
58.9 ± 2.7 |
0.365 |
|
Peak tricuspid regurgitation peak gradient [mmHg] |
51.5 ± 16.0 |
59.6 ± 23.3 |
0.154 |
|
Pulmonary function tests: |
|||
|
Total lung capacity [L] |
4.6 ± 1.3 |
4.3 ± 0.9 |
0.587 |
|
FEV1 [s] |
1.9 ± 0.8 |
1.4 ± 0.5 |
0.024 |
|
FEV1/FVC (% predicted) |
82.0 ± 7.6 |
83.4 ± 13.0 |
0.667 |
|
Diffusing capacity for carbon monoxide (% predicted) |
58.3 ± 25.8 |
50.8 ± 25.3 |
0.385 |
Univariate (Table 2) and multivariate (Table 3) logistic regression analyses were performed to assess the predictors in the high-risk group. Multivariate logistic regression analysis demonstrated that homocysteine (OR: 1.256; 95% CI: 1.002–1.574, Table 3) was an independent predictor of high-risk levels. Furthermore, correlation analysis was performed to assess potential biomarkers that correlate with NT-proBNP levels (Table 4). Homocysteine (β = 0.75, p < 0.001) and UA (β= 0.44, p = 0.002) levels showed a good linear correlation with NT-proBNP levels. The linear correlation between NT-proBNP/homocysteine (Fig. 1E) and NT-proBNP/UA (Fig. 1F) was shown in Figure 1.
|
Variables |
B |
Standard |
Odds |
95% confidence interval |
P |
|
Female |
0.46 |
0.79 |
1.587 |
0.335–7.530 |
0.561 |
|
Age [years] |
0.04 |
0.02 |
1.039 |
0.993–1.086 |
0.100 |
|
Body surface area [m2] |
0.91 |
1.37 |
2.477 |
0.170–36.118 |
0.507 |
|
Height [cm] |
0.03 |
0.03 |
0.972 |
0.920–1.027 |
0.306 |
|
Weight [kg] |
0.02 |
0.02 |
1.025 |
0.990–1.060 |
0.161 |
|
Heart rate [bpm] |
0.01 |
0.02 |
1.009 |
0.971–1.048 |
0.659 |
|
Mean arterial pressure [mmHg] |
0.00 |
0.02 |
1.002 |
0.956–1.051 |
0.919 |
|
Hematology tests: |
|||||
|
White blood cells [K/µL] |
0.00 |
0.13 |
1.001 |
0.775–1.294 |
0.993 |
|
Red blood cells [M/µL] |
0.15 |
0.40 |
1.156 |
0.532–2.512 |
0.714 |
|
Hemoglobin [g/dL] |
0.01 |
0.17 |
1.005 |
0.728–1.388 |
0.976 |
|
Red blood cell volume distribution [%] |
0.05 |
0.07 |
1.047 |
0.919–1.192 |
0.493 |
|
Platelet [K/µL] |
0.01 |
0.00 |
0.993 |
0.986–1.001 |
0.071 |
|
Neutrophil [%] |
0.03 |
0.03 |
1.035 |
0.983–1.089 |
0.192 |
|
Lymphocyte [%] |
0.06 |
0.03 |
0.947 |
0.893–1.003 |
0.063 |
|
Neutrophil/Lymphocyte ratio |
0.26 |
0.17 |
1.290 |
0.922–1.805 |
0.137 |
|
Prothrombin time [s] |
0.32 |
0.19 |
1.370 |
0.941–1.995 |
0.100 |
|
Partial thromboplastin time [s] |
0.01 |
0.05 |
0.986 |
0.893–1.088 |
0.775 |
|
Biochemistry panel: |
|||||
|
Na [mmol/L] |
0.20 |
0.11 |
0.822 |
0.667–1.011 |
0.064 |
|
Estimated GFR [mL/min/1.73 m2] |
0.03 |
0.02 |
0.973 |
0.945–1.003 |
0.074 |
|
Aspartate aminotransferase [U/L] |
0.01 |
0.02 |
0.989 |
0.955–1.024 |
0.541 |
|
Alanine aminotransferase [U/L] |
0.00 |
0.02 |
0.996 |
0.963–1.031 |
0.835 |
|
Alkaline phosphatase [U/L] |
0.02 |
0.01 |
1.019 |
0.995–1.043 |
0.127 |
|
Total bilirubin [mg/dL] |
0.62 |
0.53 |
1.852 |
0.658–5.214 |
0.243 |
|
Albumin [g/dL] |
0.94 |
0.54 |
0.392 |
0.135–1.138 |
0.085 |
|
Lactate dehydrogenase [U/L] |
0.00 |
0.01 |
1.000 |
0.989–1.010 |
0.925 |
|
Lipid profile: |
|||||
|
High-density lipoprotein [mg/dL] |
0.01 |
0.02 |
0.986 |
0.952–1.022 |
0.441 |
|
Low-density lipoprotein [mg/dL] |
0.00 |
0.01 |
1.003 |
0.983–1.025 |
0.744 |
|
Total cholesterol [mg/dL] |
0.01 |
0.01 |
0.988 |
0.971–1.005 |
0.162 |
|
Triglyceride [mg/dL] |
0.01 |
0.01 |
0.991 |
0.979–1.004 |
0.190 |
|
Circulating biomarkers: |
|||||
|
Angiopoietin-2 |
0.00 |
0.00 |
1.000 |
1.000–1.000 |
0.765 |
|
BMP-2 |
6.15 |
6396.08 |
0.002 |
0.000–0.000 |
0.999 |
|
BMP-4 |
0.90 |
0.49 |
2.467 |
0.945–6.439 |
0.065 |
|
CD40 |
0.00 |
0.00 |
1.000 |
0.999–1.001 |
0.797 |
|
Endoglin |
1.00 |
1.00 |
1.000 |
0.999–1.002 |
0.449 |
|
Interlukin-6 |
0.07 |
0.09 |
1.067 |
0.900–1.266 |
0.453 |
|
Myeloperoxidase |
0.00 |
0.00 |
1.000 |
1.000–1.000 |
0.908 |
|
Osteopontin |
0.00 |
0.00 |
1.000 |
1.000–1.000 |
0.360 |
|
VEGF |
0.00 |
0.00 |
1.000 |
1.000–1.000 |
0.360 |
|
Homocysteine [µmol/L] |
0.26 |
0.10 |
1.293 |
1.054–1.586 |
0.014 |
|
von Willebrand factor [%] |
0.00 |
0.01 |
1.001 |
0.988–1.014 |
0.886 |
|
Uric acid [mg/dL] |
0.41 |
0.17 |
1.509 |
1.088–2.094 |
0.014 |
|
D-dimer [ng/mL] |
0.00 |
0.00 |
1.001 |
1.000–1.002 |
0.058 |
|
LVEF [%] |
–0.07 |
0.09 |
0.935 |
0.790–1.106 |
0.431 |
|
Peak tricuspid regurgitation peak gradient [mmHg] |
0.02 |
0.02 |
1.022 |
0.992–1.053 |
0.152 |
|
Pulmonary function tests: |
|||||
|
Total lung capacity [L] |
–0.24 |
0.42 |
0.787 |
0.348–1.783 |
0.566 |
|
FEV1 |
–1.12 |
0.50 |
0.328 |
0.122–0.880 |
0.027 |
|
FEV1/FVC (% predicted) |
0.01 |
0.03 |
1.013 |
0.957–1.073 |
0.654 |
|
Diffusing capacity for carbon monoxide (% predicted) |
–0.02 |
0.02 |
0.981 |
0.953–1.011 |
0.218 |
|
Variables |
B |
SE |
OR |
95% CI |
P value |
|
Homocysteine [µmol/L] |
0.20 |
0.10 |
1.256 |
1.002–1.574 |
0.048 |
|
Uric acid [mg/dL] |
0.30 |
0.20 |
1.338 |
0.834–2.147 |
0.227 |
|
FEV1 (L) |
–1.00 |
0.60 |
0.378 |
0.120–1.193 |
0.097 |
|
Variables |
Unstandardized coefficient |
P value |
||
|
B |
Standard error |
β |
||
|
Angiopoietin-2 [pg/mL] |
–0.04 |
0.11 |
–0.05 |
0.759 |
|
BMP-2 [pg/mL] |
–287.03 |
442.25 |
–0.28 |
0.545 |
|
BMP-4 [pg/mL] |
921.35 |
591.51 |
0.26 |
0.129 |
|
CD40 [pg/mL] |
0.18 |
0.65 |
0.04 |
0.783 |
|
Endoglin [pg/mL] |
0.02 |
1.17 |
0.00 |
0.988 |
|
Interlukin-6 [pg/mL] |
–51.42 |
120.84 |
–0.06 |
0.672 |
|
Myeloperoxidase [pg/mL] |
–0.04 |
0.35 |
–0.02 |
0.905 |
|
Osteopontin [pg/mL] |
0.00 |
0.02 |
0.00 |
0.978 |
|
VEGF [pg/mL] |
–18.76 |
36.81 |
–0.07 |
0.613 |
|
Homocysteine [µmol/L] |
489.53 |
77.85 |
0.75 |
< 0.001 |
|
Uric acid [mg/dL] |
750.24 |
233.61 |
0.44 |
0.002 |
To find the most appropriate cut-off value for homocysteine for determining the risk level for pulmonary hypertension, a ROC analysis was performed. The best cut-off value was homocysteine = 12 µmol/L, the area under the ROC curve was 0.82, with a 95% CI between 0.67 to 0.97. Hyperhomocysteinemia (homocysteine > 12 µmol/L) could discriminate high-risk levels from low/intermediate-risk levels in pulmonary hypertension, with more high-risk patients (≤ 12: 18.8%; > 12: 70.6%, p = 0.003, Fig. 1G) in patients with hyperhomocysteinemia. Patients with homocysteine > 12 µmol/L also had higher NT-proBNP (٨٠٣.٠ ± 1,165.4 vs. 4,057.7 ± ٥,٢٣٠.٩ pg/mL, p = 0.021, Fig. 1H) and lower diffusing capacity for carbon monoxide (DLCO) (64.6 ± 24.6 vs. 44.2 ± 25.4% predicted, p = 0.045, Fig. 1I).
The MCT-rat model was obtained successfully and reflected by the elevated right ventricular systolic pressure (21.4 ± 3.0 vs. 44.8 ± ٩.٠ mmHg, p = 0.001, Fig. 2B) and right ventricular hypertrophy (Fultons’s index: 25.2 ± 2.8 vs. 49.1 ± 12.5%, p = 0.003, Fig. 2C) indicated by a significantly increased Fultons’s index. MCT rats demonstrated the elevation of NT-proBNP (Fig. 2H) and homocysteine (Fig. 2I) levels with progressed severity of pulmonary hypertension by weeks. Comparative association of NT-proBNP and homocysteine level between MCT rats and humans by disease severity were illustrated in Figure 3A–D. In addition, different biomarker combinations and the comparison between respective predictive value of each model were illustrated in Figure 3E. NT-proBNP + homocysteine + UA had strongest predictive value (AUC = 0.898), following by NT-proBNP + homocysteine (AUC = 0.890), NT-proBNP + UA (AUC =0.871), NT-proBNP (AUC = 0.867), homocysteine (AUC =0.835), and then UA (AUC = 0.698).
Discussion
This study aimed to identify potential biomarkers correlated and comparable to the current guidelines recommending NT-proBNP. A higher homocysteine level was an independent predictor for high-risk levels, and it showed a linear correlation with NT-proBNP. Further analysis indicated that the most appropriate cut-off value of homocysteine for risk level discrimination of pulmonary hypertension was homocysteine = 12 µmol/L.
The rationale for exploring biomarkers compatible and comparable with NT-proBNP
There was no single attribution of regulators or signaling molecules has adequate capacity to estimate the risk [6, 7, 14]. Currently, both the U.S. REVEAL risk score and the ESC/ERS guidelines are the most widely used multidimensional tools for risk assessment [13]. Among these, right heart catheterization is the only test to obtain the precise hemodynamic parameters for diagnosis and therapies [5].
Surprisingly, a previous study reported that BNP or NT-proBNP had a 98% sensitivity for excluding high right atrial pressure (≥ 8 mmHg) and low cardiac index (< 2.5 L/min/m2), and in circumstances of extreme low BNP (< 50 pg/mL) or NT-proBNP (< 300 pg/mL) level, hemodynamic measurements no longer had independent prognostic predictive values [14]. Moreover, COMPERA and the SPAHR registries demonstrated that the ability of mortality prediction is excellent even when only about a third of patients are followed up under the assessment of right heart catheterization [15, 18]. Nevertheless, due to the complexity of pulmonary hypertension, any single biomarker is insufficient for the broad assessment of patients with different etiologies of pulmonary hypertension. This study aimed to explore potential biomarkers compatible and comparable with NT-proBNP for disease follow-up.
The investigations of novel biomarkers and application of homocysteine for PAH risk assessment
The investigations of novel biomarkers, such as angiopoietin-2, BMP-2, BMP-4, CD40, endoglin, interlukin-6, myeloperoxidase, and osteopontin are currently in progress [7, 19, 20]. Angiopoietin-2 is produced by vascular smooth muscle cells and is involved in vascular damage/remodeling, and expression of angiopoietin-2 was up-regulated in plexiform lesions PAH lung tissues [21]. BMP-2 and BMP-4 exert opposing roles in the hypoxic pulmonary vasculature mediated by increasing endothelial nitric oxide synthase expression and activity, and BMP-2 has suggested protective effect [22, 23]. CD40 is a type I transmembrane receptor and one of the members of the tumor necrosis factor superfamily, which is expressed on epithelial cells, fibroblasts, endothelia cells, vascular smooth muscle cells, and platelets. The expression of CD40 promotes pro-thrombotic and pro-inflammatory effects, and is associated with systemic sclerosis and PAH [24, 25]. Endoglin and VEGF are angiogenic modulatory factors [26, 27]. Interlukin-6 is associated with vascular remodeling and development of PAH, which is able to predict poor adverse outcomes within the following year [28, 29]. Myeloperoxidase is able to reduce the bioavailability of nitric oxide, which is an important anti-inflammatory and vasodilating molecule. It also predicts outcomes in patients with PAH [30]. Osteopontin is involved in tissue remodeling, inflammation, and metastasis, which is recognized in cardiomyocytes and fibroblasts. Previous studies supported its correlation with mPAP, NT-proBNP, 6MWD and function class [31–33]. Bonferroni correction was applied for analysis of multiplex immunoassay biomarkers (Suppl. Table S1); despite having no significant statistical difference between low-, intermediate- and high-risk groups, the increased trend by disease severity was demonstrated. The insignificance could be attributed to the small sample size.
This study demonstrated that patients in high-risk group for pulmonary hypertension had higher homocysteine, UA, D-dimer, and CRP base on univariate analysis (Table 2). However, multivariate logistic regression analysis demonstrated that homocysteine (OR: 1.256; 95% CI: 1.002–1.574, Table 3) was the only independent predictor for high-risk levels. In addition, studies in animals and in cell cultures also demonstrated that homocysteine has a variety of toxic effects on the vasculature, endothelial dysfunction, medial remodeling and adventitial inflammation [34–41] which supports the result of serum homocysteine level of MCT rats in the present study.
In comparison with angiopoietin-2, BMP-2, BMP-4, CD40, endoglin, interlukin-6, myeloperoxidase, and osteopontin, which need to be acquired by multiplex immunoassay of human blood and were not feasible in clinical tests, homocysteine is available in daily clinical care. Furthermore, homocysteine impairs endothelium-dependent vasodilatation and is an endogenous inhibitor of nitric oxide synthase. Moreover, increased homocysteine level in PAH was also reported in a previous study [42–44]. In addition, comparison between each model illustrated in Figure 3E reported higher predictive value of homocysteine (AUC = 0.835) compared to uric acid (AUC = 0.698). Therefore, homocysteine rather than other biomarkers was selected for final advanced analysis under the consideration of multivariate analysis and clinical feasibility compared to other biomarkers.
Correlation between homocysteine and NT-proBNP, and application of homocysteine for follow-up of pulmonary hypertension
Homocysteine interferes with the expression of endothelial nitric oxide synthetase, with which its multifactorial attributions increase vascular thickness and activate elastin fragmentation, which eventually leads to PAH [8, 45]. Pulmonary hypertension can develop rapidly under hypoxic situations, and hyperhomocysteinemia was reported in cyanotic PAH patients compared to non-cyanotic patients [42–44]. A low DLCO could be seen in patients with primary pulmonary hypertension and other pulmonary vascular diseases with or without the restriction of lung volumes [46]. Moreover, lower DLCO (< 45%) was demonstrated in PAH patients with lower arterial oxygen tension [47]. Low DLCO was also an index of worse prognosis, a strong and independent risk factor for survival in patients with pulmonary hypertension [48–50]. Hyperhomocysteinemia is an index for hypoxia and low DLCO [51]. This study reported that higher homocysteine group had more high-risk level patients (≤ 12 µmol/L: 18.8%; > 12 µmol/L: ٧٠.٦٪, p = 0.003, Fig. 1G), and higher NT-proBNP (803.0 ± 1,165.4 vs. 4,057.7 ± ٥,٢٣٠.٩ pg/mL, p = 0.021, Fig. 1H). This result supported the possibility of using homocysteine for disease follow-up.
A previous study demonstrated that higher homocysteine levels were correlated with higher concentrations of NT-proBNP when the differences were assessed in comparison with the upper quartile (≥ 18 μmol/L) with the lower quartile (≤ 12 μmol/L) [52]. Hyperhomocysteinemia predicted high NT-proBNP values via a link with impaired mitochondrial fatty oxidation [52]. Furthermore, homocysteine was one of the determinants of natriuretic peptide which was analyzed by univariate analyses [52]. Association between the log of plasma concentration of homocysteine and BNP was demonstrated with a correlation coefficient of +0.297 (95% CI: +0.097–+0.474, p = 0.004) [52]. In addition, homocysteine was reported to stimulate myocardial BNP and induce adverse left ventricular remodeling [53]. However, studies describing the correlation between homocysteine and NT-proBNP through a link with pulmonary hypertension are rare. This study showed that homocysteine had a linear correlation with NT-proBNP levels (β = 0.75, p < 0.001, Fig. 1E). The 1-year mortality was < 5%, 5–20%, and > 20% if NT-proBNP values are < 200, 300–1100, > 1100 pg/mL illustrated in 2022 ESC guideline [1]. As long as the biomarker identified had a good correlation with NT-proBNP, it was able to represent the estimated 1-year mortality as well as NT-proBNP does.
With regard to the use of homocysteine for pulmonary hypertension follow-up or severity evaluation, the current study demonstrated that hyperhomocysteinemia was present in pulmonary hypertension associated with the congenital heart disease group compared to the non-pulmonary hypertension group [42]. In addition, elevated total plasma homocysteine was reported in primary pulmonary hypertension patients compared to the control group (14.7 ± 7.2 vs. 10.2 ± 5.1, p =0.027), with the cut-off value of 15 µmol/L [54]. Hyperhomocysteinemia is a crucial factor in the pathogenesis of primary pulmonary hypertension as well as poor renal function [54]. These results support the present study, that a higher homocysteine value was reported in the high-risk group compared to the low/intermediate-risk group, and the most appropriate cut-off value based on ROC analysis was homocysteine = 12 µmol/L. In light of the small sample size of this study and ethical consideration, the present study used MCT induced PAH rats to evaluate the accessibility and reliability of using homocysteine to predict PAH risk level. The result proved a comparative association between disease severity and homocysteine level, which were demonstrated both in rats and humans (Fig. 3A–D).
Limitations of the study
Individual differences of metabolism and increased lipid profiles may interfere with homocysteine values. In addition, the acquisition of blood samples depends on the interval between application and permission. Blood samples of the present study were stored for an average of between 2 weeks and 1 month. The half-life of each biomarker and accuracy might affect the results of measurement. However, this was restricted by experimental accessibility and was a limitation of the present study. In addition, the small sample size of this study limits the reliability of the application of homocysteine as an index of risk assessment. Further investigation is needed to validate this study’s result.
Study strength
Previous studies have illustrated the correlation between hyperhomocysteinemia and high NT-proBNP value via a link with impaired mitochondrial fatty oxidation. However, the correlation between homocysteine and NT-proBNP through a link with pulmonary hypertension has been obscured. Based on previous evidence, hyperhomocysteinemia were related with hypoxia-induced pulmonary vascular constriction and pulmonary hypertension. This study demonstrated that homocysteine had a linear correlation with NT-proBNP. These results posed a potential new circulating biomarker to achieve more accurate risk assessment of pulmonary hypertension.
Conclusions
This study demonstrated that patients with higher homocysteine levels had higher risk levels, higher NT-proBNP levels, and lower DLCO. This study also delineated a linear correlation between homocysteine and NT-proBNP levels. In summary, homocysteine can help discriminate between low/intermediate and high-risk groups. It is a potential biomarker that could be compatible and comparable with NT-proBNP as a non-invasive and effort-free measurement for risk assessment and disease follow-up in pulmonary hypertension.
Acknowledgments
The authors appreciated the assistance of the Biobank, Department of Medical Education and Research, Kaohsiung Veterans General Hospital, for processing of clinical specimens.
Funding
This study was supported by grants from the Kaohsiung Veterans General Hospital, Kaohsiung, Taiwan, i.e., Grant Nos. KSVGH111-007, KSVGH110-077, VGHKS109-132, VAC110-001-4 and the Ministry of Science and Technology, i.e., grant numbers: MOST107-2314-B-075B-008-MY2 and MOST108-2314-B-075B-007-MY2.
