Prognostic impact of renal function in precapillary pulmonary hypertension

Authors


Abstract

Background

Impairment of renal function is associated with adverse outcome in various diseases. Patients with pulmonary hypertension (PH) show diminished cardiac function and organ perfusion. The aim of this study was to investigate the associations between renal function and both haemodynamic parameters and long-term survival in patients with PH.

Methods

Blood was collected from 64 patients with PH (Dana Point class 1, 3 and 4) during right heart catheterization, and plasma was prepared. Creatinine, blood urea nitrogen (BUN), cystatin C, neutrophil-gelatinase-associated lipocalin (NGAL), fibroblast growth factor 23 (FGF-23) and α-Klotho levels were determined, and glomerular filtration rate (GFR) was estimated (eGFR). Parameters were evaluated using c-statistics and dichotomized for survival analysis based on receiver operating characteristic curves.

Results

The median follow-up time was 9.92 years with all-cause mortality as the primary end-point. Elevated BUN, cystatin C and creatinine levels were associated with decreased survival, with hazard ratios (HRs) of 3.237, 4.514 and 2.006, respectively, and equivalent performance according to c-statistics. Estimating GFR by CKD-EPI, MDRD and Cockcroft–Gault formulas resulted in HRs of 2.942, 2.694 and 3.306, respectively. Amongst these formulas, eGFR (Cockcroft–Gault) had the highest c-statistics of 0.674. There was a correlation between BUN and both cardiac index (τ = −0.39) and pulmonary vascular resistance index (τ = 0.249), whereas eGFR (CKD-EPI) was correlated with cardiac index (τ = 0.225). No correlations between either BUN or eGFR and right atrial pressure (RAP) were observed. NGAL, FGF-23 and α-Klotho had no prognostic impact or association with haemodynamic parameters.

Conclusion

Comparison of markers of renal function for prognosis in PH demonstrated superiority of creatinine, cystatin C and BUN over NGAL, FGF-23 and α-Klotho. Minor decreases in eGFR influence long-term prognosis, and measurement of cystatin C levels might be useful to detect renal impairment in patients with a normal serum concentration of creatinine. Renal function in patients with PH is linked to cardiac index rather than RAP.

Introduction

Pulmonary hypertension (PH) is a heterogeneous clinical disorder characterized by a marked increase in pulmonary vascular resistance and pulmonary artery pressure caused by rarefaction, obliteration and vasoconstriction of the pulmonary capillary bed. Its clinical implications include profound exercise intolerance and reduced life expectancy [1]. Prognosis is tightly linked to right ventricular function as defined by right atrial pressure and cardiac output [2, 3].

As perfusion is critical for glomerular filtration, decreased cardiac output and perfusion pressure affect renal function independently of the underlying disease. Renal function has been associated with poor prognosis in a wide variety of disorders including acute and chronic cardiovascular [4, 5] and lung diseases [6, 7]. Recent studies examined the occurrence of PH in patients with end-stage renal disease and undergoing haemodialysis. A prevalence of precapillary PH of 13% was observed in the PEPPER (PrEvalence of Precapillary Pulmonary hypertension in End-stage Renal disease) study, whilst the majority of patients had postcapillary PH [8]. These patients primarily suffered from kidney failure; however, less is known about the effects of chronic renal insufficiency secondary to precapillary PH.

Shah and colleagues demonstrated an association between elevated serum creatinine in patients with pulmonary arterial hypertension (PAH) and compromised haemodynamics, particularly with elevated right atrial pressure [9]. Associations with haemodynamics or prognosis might be of clinical value, as measurements of creatinine levels, as well as blood urea nitrogen (BUN) and cystatin C, are well standardized and broadly available in clinical laboratories. The value of the measurement of neutrophil-gelatinase-associated lipocalin (NGAL; also known as lipocalin-2) or fibroblast growth factor 23 (FGF-23), and its co-factor α-Klotho is under investigation in an increasing number of studies. The use of NGAL was investigated in cardiorenal syndrome [10] and chronic heart failure [11, 12], but data on right-sided heart failure and prognosis are limited. Koca and co-workers found no differences in NGAL levels in either serum or urine in patients with right heart failure attributed to cardiomyopathy or PAH and preserved renal function [13].

The phosphaturic hormone FGF-23 was found to be associated with left ventricular hypertrophy (LVH) and dysfunction not only in patients with chronic kidney disease (CKD) but also in individuals with intact renal function [14]; this finding was independent of the predictive value of elevated levels of FGF-23 for progression of CKD and all-cause and cardiovascular mortality [15]. A direct cardiac effect of FGF-23 was demonstrated in vivo and vitro by Faul et al. These experimental findings were supplemented by observational data from the chronic renal insufficiency cohort (CRIC) study, correlating LVH with increased FGF-23 levels [16].

Until recently, the co-factor α-Klotho had only been implicated in the physiological functions of FGF-23 in the parathyroid gland and kidney [17, 18]. New experimental and observational data, however, suggest an FGF-23-independent role of α-Klotho, whereby decreased levels may be a risk marker of mortality and a very early indicator of CKD [19-21]. To date, there are no data on the relationship between FGF-23 and α-Klotho as markers of calcium phosphate metabolism and PH.

There were three main objectives of the present study: first, we aimed to investigate the influence of compromised renal function on long-term prognosis in patients with precapillary PH. Secondly, we compared the value of standard parameters of renal function (creatinine, BUN and cystatin C) and different methods for glomerular filtration rate (GFR) estimation (eGFR), as well as the additional variables NGAL, FGF-23 and α-Klotho. Thirdly, we evaluated a possible association between haemodynamic parameters and renal function. This analysis was undertaken in a cohort of patients with precapillary PH; all patients had stable disease on therapy and lacked signs of acute right heart failure.

Methods

The study was conducted in accordance with the Declaration of Helsinki and was approved by the local ethics committee. Written informed consent was obtained from all patients before enrolment.

Study design

A total of 64 patients were included retrospectively from the local database of biosamples from patients with PH. The patients had incident and prevalent diagnosis of precapillary PH including PAH (= 32), PH associated with lung disease (= 11) and chronic thromboembolic PH (= 21). The patient cohort was registered in the database between 2000 and 2003, at the Centre for Pulmonary Vascular Disease in the Department of Pulmonology at Saarland University.

All patients had a mean pulmonary artery pressure above 25 mmHg, with pulmonary artery occlusion pressure below 15 mmHg. Thermodilution method was used to determine cardiac output. Patients with postcapillary PH, significant left heart disease and malignancies were excluded from the study. Furthermore, patients diagnosed during the follow-up period with left-sided heart disease, primary kidney disease or diabetes were excluded from the analysis. None of the patients was in an acutely decompensated haemodynamic state at enrolment.

At completion of the study, the vital status was confirmed by review of medical records and by telephone and in-person patient interviews. The primary end-point was all-cause mortality.

Measurements

Blood samples were collected during routine right heart catheterization using commercially available standard EDTA tubes (Sarstedt, Nümbrecht, Germany). The samples were supplemented with Trasylol for protease inactivation and processed immediately by centrifugation at 2700 g at 4 °C for 20 min. Plasma was separated and stored in aliquots at −80 °C until analysis. Levels of serum creatinine and BUN, as well as additional standard biochemistry parameters, were determined at the local clinical chemistry laboratory.

Cystatin C levels were measured by an enzyme-linked immunosorbant assay (ELISA) according to the manufacturer's instructions (human Cystatin C ELISA, DRG Diagnostics, Marburg, Germany). Intra-assay variability was determined to be 2.5%. There was a high recovery rate of 97% cystatin C in EDTA compared with serum samples (coefficient of determination, R= 0.97).

Neutrophil-gelatinase-associated lipocalin concentration was determined in the same samples using a commercial ELISA kit, with an intra-assay variability of 8.06%, according to the manufacturer's instructions (Boster Biological Technology Co. Ltd., Fremont, CA, USA). Different commercially available ELISA kits were also used for measurement of c-terminal FGF-23 (USCN Life Sciences, Wuhan, China; intra-assay variability of 6.96%) and soluble α-Klotho (Cusabio, Wuhan, China; intra-assay variability of 5.62%), according to the manufacturers' instructions.

Optical density was measured using a Tecan Spectra III reader set to 450 nm, with a reference wavelength of 620 nm. Mid-regional pro-atrial natriuretic peptide (MR-pro-ANP) concentration was determined fluorometrically with a commercially available system using TRACE™ technology (BRAHMS, Hennigsdorf, Germany); the interassay variability was calculated to be 1.00% for EDTA plasma.

Statistical analysis

Estimated glomerular filtration rate was calculated using three previously published methods: four-variable Modification of Diet in Renal Disease (MDRD) [22], Cockcroft–Gault [23] and Chronic Kidney Disease Epidemiology Collaboration (CKD-EPI) equations [24] (see Data S1).

Continuous variables were tested for normality and variance homogeneity using the Shapiro–Wilk test and Hartley's Fmax test, respectively. As most variables were non-normally distributed, results are expressed as median (interquartile range) unless indicated otherwise. The frequencies of categorical variables are expressed as percentage (number). Data points were defined as outliers if outside a three-SD interval from the mean of the whole data set. Differences between groups were detected by Mann–Whitney U-test. Comparison of categorical variables was made by chi-squared test.

Associations between continuous variables were investigated using Kendall's rank order correlation analysis. Cut-off values were optimized in a stepwise manner by minimization of Matthews coefficient. Variables were then dichotomized, and their prognostic significance tested using Kaplan–Meier analysis for all-cause mortality. Survival differences between groups were compared by log-rank test. A P-value of less than 0.05 (two-tailed) was considered statistically significant. Data were stored and processed using aabel (version 3.0.3, Gigawiz, Tulsa, OK, US) and R-Project (version 2.15.0, open source) for survival analyses (Efron model). For time-dependent comparison of parameters, c-statistics were calculated as described previously [25].

Results

PH was indicated in all patients in the study, based on the Dana Point classification (categories 1, 3 and 4). Patient characteristics are shown in Table 1. Patients with PAH were slightly younger than those in the other two groups. Furthermore, there were differences in therapy regimens as specific therapeutics are predominantly used in PAH. The majority of patients were in New York Heart Association functional class 3 or 4, with no statistically significant differences between subgroups.

Table 1. Patient characteristics. Data expressed as median [interquartile range]
 Overall, = 64PAH, = 32LD, = 11CTEPH, = 21
  1. PAH, pulmonary arterial hypertension; LD, PH associated with lung disease; CTEPH, chronic thromboembolic pulmonary hypertension; Age [years]; BSA, body surface area [sqm]; BMI, body mass index [kg m−2]; survival time [months]; 6MWD, 6-min walking distance [m]; NYHA-FC, modified functional class in percentage of group (absolute number); PDEI, phosphodiesterase inhibitor; ERA, endothelin receptor antagonist; PC, prostacyclin; Combination: therapy with ≥2 PH-specific drugs.

  2. *= 0.05 vs. PAH; = 0.05 vs. LD.

Age56.4 [48.0–66.1]51.8 [38.2–60.4]69.5* [62.9–74.3]59.4*,† [50.4–67.4]
Gender (f/m)39/2519/136/514/7
Survival27 (42.2%)14 (43.8%)1 (9.1%)12 (57.1%)
BSA1.81 [1.66–1.98]1.80 [1.62–1.95]1.82 [1.57–2.01]1.86 [1.73–2.02]
BMI24.93 [22.65–28.34]24.22 [21.97–28.48]25.20 [23.74–30.05]25.65 [22.09–26.94]
Survival time38.6 [11.8–103.9]56.8 [11.4–104.2]20.0 [14.6–42.8]49.8 [20.0–115.9]
6MWD303.5 [175–384]361.5 [240–400]180 [167–330]245 [120–393]
NYHA-FC III/IV66.6% (40)69.2% (18)90% (9)68.4% (13)
PDEI18.8% (12)15.6% (5)36.4% (4)14.3% (3)
ERA29.7% (19)34.4% (11)18.2% (2)23.8% (5)
PC32.8% (21)56.3% (18)0.0% (0)*14.3% (3)*
Combination20.3% (13)25.0% (8)9.1% (1)*19.0% (4)*

Haemodynamic parameters are shown in Table 2 and echocardiographic parameters in Supplementary Table 1. Patients with PH associated with lung disease had slightly lower pulmonary artery pressures, compared with those with PAH. Consequently, pulmonary vascular resistance index (PVRI) and the pulmonary/systemic vascular resistance index ratio were lower in this group.

Table 2. Hemodynamic parameters. Data expressed as median [interquartile range]
 Overall, = 64PAH, = 32LD, = 11CTEPH, = 21
  1. SAP, systemic artery pressure [mmHg]; PAP, pulmonary artery pressure [mmHg]; PAOP, pulmonary artery occlusion pressure [mmHg]; RAP, right atrial pressure [mmHg]; SVRI, systemic vascular resistance index [dyn s cm−5 sqm]; PVRI, pulmonary vascular resistance index [dyn s cm−5 sqm].

  2. *= 0.05 vs. PAH.

SAP mean92.0 [81.0–100.0]93.0 [80.0–99.0]95.0 [89.0–98.0]90.0 [80.5–105.5]
PAP mean46.5 [37.0–57.0]51.0 [41.0–66.5]39.0* [28.0–42.0]48.0 [32.0–55.0]
PAOP9.0 [8.0–11.0]8.0 [8.0–10.0]9.5 [8.0–12.0]10.0 [8.0–11.5]
RAP8.0 [5.0–13.0]8.0 [5.0–12.0]5.0 [4.0–6.0]10.0 [6.0–13.0]
CI2.25 [1.88–2.72]2.14 [1.81–2.80]2.50 [2.33–2.79]2.15 [1.76–2.58]
SVRI2592.5 [2070.1–3471.7]2434.2 [2056.7–3567.3]2401.6 [2083.6–2861.8]2955.6 [2327.3–3650.4]
PVRI1217.4 [639.7–1911.1]1285.0 [873.6–1767.5]852.7 [542.6–1616.6]1199.1 [622.0–2086.8]
PVRI/SVRI0.471 [0.312–0.692]0.532 [0.367–0.875]0.333* [0.253–0.405]0.500 [0.303–0.639]

With regard to biochemical measurements (Table 3), lower creatinine and BUN levels were observed in patients with PH associated with lung disease compared with those with PAH. MR-pro-ANP concentration was also lower in this group although the difference did not reach statistical significance. These biochemical results mirror the haemodynamic differences between groups. No differences in levels of cystatin C, NGAL, FGF-23 and α-Klotho were found in subgroup analyses. Similarly, there were no differences in eGFR and electrolyte levels amongst groups.

Table 3. Biochemistry parameters. Data expressed as median [interquartile range]
 Overall, = 64PAH, = 32LD, = 11CTEPH, = 21
  1. NGAL, neutrophil-gelatinase-associated lipocalin; BUN, blood urea nitrogen; eGFR, calculated glomerular filtration rate; CKD-EPI, chronic kidney disease epidemiology collaboration; MDRD, modified diet in renal disease; Cockcroft, Cockcroft–Gault; CysC, cystatin C; Crea, creatinine; MR-pro-ANP, mid-regional pro-atrial natriuretic peptide.

  2. *= 0.05 vs. PAH; = 0.05 vs. LD.

Creatinine (mg dL−1)0.90 [0.80–1.10]1.00 [0.80–1.10]0.80* [0.60–0.80]0.90 [0.80–1.00]
Cystatin C (mg mL−1)1.23 [1.05–1.43]1.30 [0.97–1.43]1.12 [1.05–1.21]1.21 [0.95–1.41]
NGAL (ng mL−1)49.33 [29.82–69.80]53.68 [29.68–71.17]41.10 [25.28–70.84]52.75 [33.21–65.25]
FGF-23 (pg mL−1)27.50 [11.75–43.25]25.50 [5.75–43.25]27.00 [19.00–33.00]33.00 [12.00–50.00]
α-Klotho (pg mL−1)93.99 [74.85–109.60]92.74 [74.14–108.40]94.59 [84.53–107.40]101.00 [76.18–120.00]
BUN (mg dL−1)33 [27–44]36 [30–47]24* [21-30]33 [31.5–47]
eGFR (CKD-EPI)79.63 [64.16–97.59]76.97 [62.94–101.50]90.77 [75.26–101.10]79.26 [62.91–94.91]
eGFR (MDRD)78.6 [61.7–99.8]76.0 [59.8–95.1]103.8 [74.8–121.1]77.6 [61.0–87.8]
eGFR (Cockcroft)79.0 [62.8–115.2]77.7 [61.7–101.3]82.8 [61.7–119.7]77.9 [64.8–103.9]
MR-pro-ANP (pmol L−1)193.20 [84.16–340.10]194.85 [86.36–350.40]84.24 [49.15–256.70]280.40 [109.02–348.45]

For survival analyses, biochemical parameters were dichotomized using optimized cut-off values. The results of the survival analyses are summarized in Table 4. The end-point for all analyses was all-cause mortality. The median follow-up time was 9.92 [9.14–11.04] years. Elevations of creatinine and cystatin C were associated with a hazard ratio (HR) of 2.006 (95% confidence interval (CI) 1.035–3.887, = 0.0391; Fig. 1a) and 4.514 (95% CI 1.380–14.760, = 0.0127; Fig. 1b), respectively. A higher BUN concentration was associated with an HR of 3.237 (95% CI 1.645–6.367, = 0.000669; Fig. 1c). The performance of the three parameters was comparable according to c-statistic values of 0.610, 0.603 and 0.595 for creatinine, BUN and cystatin C, respectively. To compare the diagnostic value of cystatin C and creatinine, patients were categorized using standardized cut-off values (creatinine, ≤0.9 mg dL−1 for women and ≤1.2 mg dL−1 for men; cystatin C, ≤1.09 mg dL−1). Amongst patients with normal creatinine levels, 59% had to be reclassified as having compromised renal function, according to cystatin C levels. By contrast, only one patient amongst all those with elevated cystatin C levels had elevated creatinine concentration too (Table 5, chi-squared 6.23, = 0.013). No impact of NGAL, FGF-23 and α-Klotho on all-cause mortality was observed in this study (see Figs S1–S3).

Table 4. Results of log-rank tests for over-all survival with cut-off values listed by parameter
 Cut-offHR [CI] P
Creatinine (mg dL−1)>1.0 mg dL−12.006 [1.035–3.887]0.0391
Cystatin C (mg ml−1)>1.0 mg dL−14.514 [1.380–14.760]0.0127
NGAL (ng mL−1)>45 ng mL−11.760 [0.877–3.535]0.112
FGF-23 (pg mL−1)>93 pg mL−11.357 [0.669–2.673]0.378
α-Klotho (pg mL−1)>74 pg mL−11.406 [0.618–3.203]0.417
BUN (mg dL−1)>38 mg dL−13.237 [1.645–6.367]0.000669
eGFR (CKD-EPI)<79 mL min−12.942 [1.467–5.900]0.00236
eGFR (MDRD)<80 mL min−12.694 [1.317–5.507]0.00662
eGFR (Cockcroft)<80 mL min−13.306 [1.520–7.193]0.00257
Table 5. C-statistics of single parameters ordered by value
 C-statisticSELower CIUpper CI
MR-pro-ANP0.6790.0580.5650.794
Cockcroft–Gault (eGFR)0.6740.0520.5730.775
CKDEPI (eGFR)0.6370.0470.5460.729
Creatinine0.6100.0660.4810.740
MDRD (eGFR)0.6080.0740.4620.754
BUN0.6030.0600.4860.720
Cystatin C0.5950.0570.4830.707
NGAL0.5250.0630.4010.648
a-Klotho0.5240.0700.3880.661
FGF230.5150.0670.3830.647
Figure 1.

Survival analysis according to serum levels of creatinine (Crea; a), cystatin C (b) and blood urea nitrogen (BUN; c).

Various methods for calculation of eGFR were compared (see Table 4). The best results according to c-statistic values were obtained using the CKD-EPI and Cockcroft–Gault formulas with HR values of 2.942 (95% CI 1.467–5.900, = 0.00236) and 3.306 (95% CI 1.520–7.193, = 0.00257), respectively (Fig. 2). c-Statistic values for survival are summarized in Table 6.

Table 6. Contingency table of cystatin C versus creatinine
  Cystatin C ≤ 1.09 mg dL−1Cystatin C > 1.09 mg dL−1 
  1. Cut-offs chosen by standardized normal values. Of 37 patients with normal creatinine levels, 59% (22) were reclassified according to elevated cystatin C. (χ2 6.23, P = 0.013).

Male, Crea ≤ 1.2 mg dL−1Creatinin normal152237
Female, Crea ≤ 0.9 mg dL−1
Male, Crea > 1.2 mg dL−1Creatinine increased11516
Female, Crea > 0.9 mg dL−1
  163763
Figure 2.

Survival analysis according to glomerular filtration rate estimated (eGFR) using the Cockcroft–Gault (Cockcroft; a), four-variable Modification of Diet in Renal Disease (MDRD; b) and Chronic Kidney Disease Epidemiology Collaboration (CKD-EPI; c) equations.

Renal parameters were examined for associations with invasive cardiac haemodynamic parameters. Compared with lower concentrations, BUN above 38 mg dL−1 was associated with low cardiac index (1.88 vs. 2.47 L min−1 m−2, < 0.001) and increased PVRI (1372.11 vs. 852.6511 dyn s cm−5 m2, = 0.03) as well as higher right atrial pressures (12 vs. 6 mmHg, = 0.015). Consequently, higher levels of MR-pro-ANP (323.3 vs. 111.9 pmol L−1, = 0.007) were observed. Although median eGFR was lower in the group with BUN >38 mg L−1 (eGFR CKD-EPI 71.68 vs. 87.04 mL min−1, < 0.001), there was only a weak correlation between BUN and eGFR (CKD-EPI; τ = −0.316, < 0.001). However, a moderate negative correlation with cardiac index (τ = −0.39, < 0.001) and a weak correlation with PVRI (τ = 0.249, = 0.015) were observed. There were no correlations between BUN and RAP, mean systemic arterial pressure (mSAP) or systemic perfusion pressure (mSAP-RAP).

When patients were stratified by eGFR (CKD-EPI) at a cut-off of 79 mL min−1, a lower median cardiac index (2.055 vs. 2.570 L min−1 m−2, = 0.018), higher right atrial pressure (10.5 vs. 6.0 mmHg, = 0.032) and higher circulating level of MR-pro-ANP (309.60 vs. 86.36 pmol L−1, = 0.008) were observed in those with reduced renal function. No significant difference in PVRI was found between groups of patients with eGFR below and above 79 mL min−1 (1314.9 vs. 953.1 dyn s cm−5 m2, respectively, n.s.). eGFR (CKD-EPI) was weakly correlated with cardiac index (τ = 0.225, = 0.028), but not with PVRI, right atrial pressure, mSAP or systemic perfusion pressure.

There were no associations between therapy with phosphodiesterase inhibitors, endothelin receptor antagonists and prostacyclins and differences in the measured or calculated parameters. Parameters of renal function or haemodynamics did not differ in incident and prevalent cases of PH.

Discussion

In this study, impairment of renal function was associated with worse outcome in patients with PH, even when the degree of impairment was minor. Renal function is an important parameter in various cardiovascular diseases. It reflects both haemodynamic state and general microcirculatory disorders, known as cardio-renal syndrome. In PH patients, progression of pulmonary microvascular dysfunction increases pulmonary vascular resistance and right ventricular afterload and subsequently leads to impaired right ventricular function and right heart failure. The consequence is increased central venous pressure, which reduces the effective perfusion pressure of organs, whilst, at the same time, decreased left ventricular filling due to forward failure of the right ventricle leads to decreased cardiac output and lower arterial pressure. These profound circulatory changes affect exercise capacity and result in chronic organ hypoperfusion. The autoregulatory mechanisms of the kidneys maintain perfusion and glomerular filtration by activation of the renin-angiotensin-aldosterone system (RAAS). Dysregulation of the RAAS has been shown to support the development of experimental PAH in the monocrotaline model [26]. Retrospective analysis of renin, angiotensin II and aldosterone as well as angiotensin II receptor type 1 in pulmonary arteries confirmed the activation of the RAAS in patients with idiopathic PAH. Interventional experiments in a monocrotaline model with losartan resulted in reduced progression and reduction in afterload of the right ventricle [27]. In the present study, no effects of medication on either parameters of kidney function or outcome were observed. Indeed, neither the type of specific medication for PH nor supplemental therapy with angiotensin-converting enzyme inhibitors, angiotensin receptor blockers or aldosterone antagonists showed any association with outcome or kidney function. However, the number of patients receiving supplemental medication was small, and analysis of parameters of the RAAS was prohibited by sample preparation in this cohort.

In contrast to growing evidence on the incretory function of the kidney, data on excretory kidney function in PH, particularly regarding its relation to outcome, are limited. Shah et al. described associations between increased serum creatinine levels and both increased right atrial pressure and lower cardiac index in the largest cohort of PAH patients to date. The authors found that prognosis was worse with decreasing renal function in groups stratified according to creatinine cut-off levels of 1.0 and 1.4 mg dL−1 [9]. In another study, several variables were examined during follow-up of patients with PAH. The goal of this study was to determine whether changes in these variables were predictive of outcome. Both creatinine at baseline and change in BUN at follow-up were predictors of outcome in univariate models, but were not incorporated into multivariate models. Whilst both studies provided excellent data on PAH, no patients with other types of PH (non-PAH) were included. Therefore, it remains unclear whether the results apply to these non-PAH patient groups. Furthermore, cystatin C, which is increasingly used in clinical practice, and other novel markers of renal function were not evaluated.

In the present study, we examined patients with different subtypes of PH and recorded long-term all-cause mortality under standard therapy (as recommended by guidelines) over more than 10 years. Standard parameters of renal function – serum creatinine, eGFR (Cockcroft–Gault and CKD-EPI) and BUN – were measured and demonstrated a highly significant impact on survival even with minor renal impairment. We were surprised to note that, according to c-statistics, the Cockcroft–Gault formula was superior in terms of performance to the CKD-EPI and MDRD formulas.

Blood urea nitrogen, which is generally considered to be a short-term marker of renal function, had the highest HR and statistical significance, but was only of average performance according to c-statistics. Increased concentrations of BUN are associated with hypovolaemia and acute hypoperfusion of the kidney. The patients in the present study were selected during a period of stable disease, and therefore, acute haemodynamic deterioration could be excluded. We found no correlations between BUN and either right atrial pressure or perfusion pressure, indicating that loading conditions of the right ventricle were not responsible for any increases in the level of BUN. Instead, BUN was correlated moderately with cardiac index and weakly with PVRI. Thus, central haemodynamic parameters appear to be crucial in this respect. However, no correlation between BUN levels and eGFR was found in this study, and thus, the relation of this variable to renal function in this specific context must be questioned. BUN is degraded in the liver and might increase with decreased liver perfusion, although no sufficient data on liver function were available in this study.

Serum creatinine as a marker of renal function is susceptible to confounding variables such as muscle mass. New biomarkers of kidney function have been successfully integrated into clinical practice to avoid this disadvantage. Cystatin C demonstrated a consistent performance over a wide range of anthropometric criteria. In the present study, we were able to provide for the first time long-term follow-up data in patients with PH based on cystatin C. In a direct comparison of cystatin C and creatinine as markers of renal function, the majority of patients with a normal serum creatinine concentration had compromised renal function according to their cystatin C levels. Serum levels of cystatin C above the normal range were associated with worse prognosis, but did not add prognostic value in conjunction with creatinine. The performance of cystatin C according to c-statistics was slightly lower than creatinine-based indices. The most common known confounders for cystatin C are thyroid disorders, inflammation and malignancy, which were excluded in patients in this study.

Amongst the tested models of GFR estimation, the CKD-EPI equation was only marginally exceeded in performance by the Cockcroft–Gault formula and clearly outperformed the widely used MDRD formula according to c-statistics for all-cause mortality. The CKD-EPI formula has previously shown advantages for risk assessment in cardiovascular disorders [28]. Most equations for GFR estimation have been tested in patients with compromised renal function. With regard to studies using the MDRD formula, patients were below 70 years of age and the majority had diabetic nephropathy [29]. The performance of the CKD-EPI formula in patients with PH in this study may have been a result of the fact that it was intended for use in all patients, not only those younger than 70 years and with diabetic nephropathy. In this study, serum markers of renal function were measured and used for statistical evaluation. On the other hand, eGFR was derived from these measures and used for risk assessment in statistical models. These calculations were developed to estimate GFR in kidney disease, and use of different formulas resulted in different estimates. The true GFR for the index substance (creatinine, BUN, cystatin C) was not determined in this study due to the absence of urine sampling or scintigraphy. Therefore, we were not able to evaluate the accuracy of the estimation itself. Nevertheless, measured GFR does not necessarily outperform eGFR with regard to prognosis [30]. Although cut-off values differed substantially between different models, there may be a prognostic value beyond GFR. This possibility has previously been suggested [31] and warrants further investigation in studies evaluating measured GFR in PH, as well as evaluation of the accuracy of GFR estimation in this context.

Mielniczuk et al. [32] provided data on the deterioration of kidney function in 32 patients with right heart failure and its association with prognosis. Approximately two-thirds had moderately to severely compromised kidney function. The authors reported a reduction in short-term survival in a group with an acute increase in creatinine of more than 0.3 mg dL−1 in the first 48 h after admission. Whilst these data reveal the importance of decreased renal function, acute deterioration of right ventricular function was evaluated and only short-term prognosis is reported. Markers other than creatinine were not investigated.

Maintenance of volume, fluid and electrolyte balance is a central role of renal function. This function is impaired in right heart failure as seen in advanced PH, and the aim of adjunctive therapy with diuretics is to decrease volume overload but at the risk of electrolyte disorders. Forfia and colleagues examined serum sodium concentration in 40 patients with PAH and found that hyponatraemia was associated with increased right atrial pressure and decreased cardiac output. GFR was estimated using the MDRD formula, and patients with sodium levels below 136 mmol L−1 had significantly higher concentrations of serum creatinine and BUN and lower eGFR. In addition, hospitalization rates were higher in these patients, exercise tolerance was lower, and they had a greater number of symptoms and a worse prognosis [33]. The data were analysed with respect to sodium levels, but revealed no further association with renal function or different markers of renal function. Determining whether this reflects adjunctive therapy for progressive right heart failure or an intrinsic excretory dysfunction of the kidney was beyond the scope of the study. By contrast, the data presented here referred to patients with decreased excretory renal function without electrolyte disorders, and we were therefore able to demonstrate the prognostic value of renal function in presumably stable PH.

Furthermore, several novel serum biomarkers of renal function were examined in this study for the first time in the context of PH. Amongst these, NGAL represents an established marker of acute kidney injury. Tubular cells of the loop of Henle release NGAL upon ischaemic insults as seen after vascular procedures or contrast media-induced kidney damage, resulting in a rapid increase in plasma levels [34]. We hypothesized that chronic hypoperfusion of the kidney and low arterial oxygen tension as seen in patients with PH might lead to an equivalent release of NGAL. Survival analyses showed no prognostic impact of NGAL, and no difference in NGAL levels was detected between survivors and nonsurvivors. Furthermore, no associations were found between NGAL levels and either haemodynamic variables or central venous oxygen saturation. Therefore, NGAL might serve as a prognostic marker in acute right heart failure [10-12], but not in stable PH as demonstrated in this study.

The concentration of the phosphaturic protein FGF-23 is elevated in patients with impaired renal function [35] and is a marker of cardiovascular events in patients with moderately impaired renal function and patients with chronic heart failure [36, 37]. Levels of α-Klotho, the co-factor of FGF-23, decrease with impaired renal function [20]. However, data suggesting a prognostic impact of circulating α-Klotho are scarce [19, 21]. Our results showed no associations between FGF-23 or α-Klotho and survival or haemodynamic parameters.

Therefore, NGAL as a measure of acute – mostly ischaemic – tubular kidney damage and both FGF-23 and α-Klotho as measures of calcium phosphate metabolism and renal function in cardiovascular disease were not found to be of prognostic importance in this study. By contrast, cystatin C caused significant reclassification of renal function and was able to identify a significantly higher proportion of PH patients with renal impairment. Whilst the prognostic value was comparable to creatinine levels and eGFR, accurate estimation of renal function is necessary to avoid potentially harmful overdosing of vasoactive drugs or in the face of novel oral coagulants.

In conclusion, the findings of this study provide insight into the use of a wide range of markers of renal function in PH. Amongst these biomarkers, GFR estimated by either the Cockcroft–Gault or CKD-EPI formulas was strongly associated with long-term all-cause mortality and negatively correlated with cardiac index. Furthermore, the novel biomarkers cystatin C, NGAL, FGF-23 and α-Klotho were evaluated for the first time in patients with PH. Of these, only cystatin C was found to be of prognostic value in this study. Of note, analysis of cystatin C demonstrated the underdiagnosis of renal impairment by serum creatinine in patients with PH; this highlights the importance of using the most appropriate diagnostic tools.

Conflict of interest statement

The authors have the following perceived conflict of interest: R.K.: Honoraria for consultancy for GSK, travel grants from Actelion, Bayer and GSK; S.S.: none; M.H.: Honoraria for lectures and/or consultancy for Actelion, Bayer, Boehringer, GSK, Lilly, Novartis, Pfizer and Servier; R.B.: The University received financial support for educational symposia; H.W.: Honoraria for lectures and/or consultancy for Actelion, Bayer, GSK, Lilly and Pfizer

Ancillary