Low-density lipoprotein cholesterol and survival in pulmonary arterial hypertension

Low-density lipoprotein cholesterol(LDL-C) is a well established metabolic marker of cardiovascular risk, however, its role in pulmonary arterial hypertension (PAH) has not been determined. Therefore we assessed whether LDL-C levels are altered in PAH patients, if they are associated with survival in this group and whether pulmonary hypertension (PH) reversal can influence LDL-C levels. Consecutive 46 PAH males and 94 females were age matched with a representative sample of 1168 males and 1245 females, respectively. Cox regression models were used to assess the association between LDL-C and mortality. The effect of PH reversal on LDL-C levels was assessed in 34 patients with chronic thromboembolic pulmonary hypertension (CTEPH) undergoing invasive treatment. LDL-C was lower in both PAH (2.6 ± 0.8 mmol/l) and CTEPH (2.7 ± 0.7 mmol/l) patients when compared to controls (3.2 ± 1.1 mmol/l, p < 0.001). In PAH patients lower LDL-C significantly predicted death (HR:0.44/1 mmol/l, 95%CI:0.26–0.74, p = 0.002) after a median follow-up time of 33(21–36) months. In the CTEPH group, LDL-C increased (from 2.6[2.1–3.2] to 4.0[2.8–4.9]mmol/l, p = 0.01) in patients with PH reversal but remained unchanged in other patients (2.4[2.2–2.7] vs 2.3[2.1–2.5]mmol/l, p = 0.51). We concluded that LDL-C level is low in patients with PAH and is associated with an increased risk of death. Reversal of PH increases LDL-C levels.

Comparison with age and sex-matched controls. LDL-C was stable during 3 month follow-up: 2.57 ± 0.88 mmol/l at enrollment and 2.59 ± 0.98 mmol/l at follow-up; p = 0.67. As compared to controls it was lower in PAH women, men and patients who did not use statins ( Table 2). LDL-C levels were similar in the subgroups of idiopathic PAH (IPAH), PAH associated with connective tissue disease (CTD-PAH), and PAH associated with congenital heart disease (CHD-PAH) (2.4 [1.7-3.0] vs 2.8 [2.1-3.6] vs and 2.6 [2.1-3.0] mmol/l, respectively, p = 0.83). Sex-specific comparisons of other cardiovascular risk factors between PAH patients and age-matched controls are presented in Table 2.
Clinical characteristics and LDL-C levels. The associations of age, sex, aspartate aminotransferase (AST), alanine aminotransferase (ALT), creatinine, cardiovascular risk factors, markers of PAH severity and LDL-C levels are presented in Table 3. In univariate linear regression analysis triglycerides (TG), WHO-FC class and sex were associated with LDL-C level, however in a multiple regression model [R 2 = 0.26, F(2, 86) = 15.2, p < 0.001] including all variables given in Table 3 significant relationship with LDL-C remained only for TG (beta coefficient  Table 1. Characteristics of patients with pulmonary arterial hypertension. 6-MWD -6-minute walk distance, Ao -aorta, CHD-PAH -pulmonary arterial hypertension associated with congenital heart disease, CI -cardiac index, CO -cardiac output, CTD-PAH -pulmonary arterial hypertension associated with connective tissue disease, IPAH -idiopathic pulmonary arterial hypertension, mPAP -mean pulmonary arterial pressure, NT-proBNP -N-terminal pro-brain natriuretic peptide, PAH -pulmonary arterial hypertension, PAWP -pulmonary artery wedge pressure, PVR -pulmonary vascular resistance, RAPright atrial pressure, WHO-FC -World Health Organization functional class.  Table 3. Univariate linear regression associations between different variables and low-density cholesterol level in patients with pulmonary arterial hypertension (n = 140). 6-MWD -6-minute walk distance, ALTalanine aminotransferase, AST -aspartate aminotransferase, BMI -body mass index, CHD-PAH -pulmonary arterial hypertension associated with congenital heart disease, CI -cardiac index, CTD-PAH -pulmonary arterial hypertension associated with connective tissue disease, HDL-C -high density lipoprotein cholesterol, IPAH -idiopathic pulmonary arterial hypertension, mPAP -mean pulmonary arterial pressure, NT-proBNP -N-terminal pro-brain natriuretic peptide, PAH -pulmonary arterial hypertension, PVR -pulmonary vascular resistance, RAP -right atrial pressure, SE -standard error, WHO-FC -World Health Organization functional class. In a subgroup of 108 patients in whom high-sensitivity C-reactive protein (hsCRP) was available (patients recruited after January 2011) we found a significant correlation (r = − 0.24 and p = 0.01) between this marker and LDL-C level.

LDL-C for the prediction of death in PAH.
In the PAH sample, 32 patients died after a median follow-up time of 33 20-36 months. Spline analysis revealed the association between LDL-C levels and survival to be linear ( Fig. 1), allowing to model LDL-C in this way. Lower LDL-C levels were significantly associated with all-cause mortality in univariable analysis (HR: 0.44 per 1 mmol/l, 95% CI: 0.26-0.74, p = 0.002, Table 4) and after hierarchical adjustment for risk factors (HR in fully-adjusted model: 0.18 per 1 mmol/l, 95% CI: 0.07-0.47, p < 0.001, Table 4). Similar results were found when only patients who did not use statins were analysed (HR in fully-adjusted model: HR = 0.19 per mmol/l, 95% CI = 0.06-0.46, p < 0.001).

Discussion
In the present study we show that PAH patients have decreased LDL-C levels, and that low LDL-C levels are associated with increased risk of all-cause mortality. Furthermore, we demonstrate that in CTEPH patients significant improvement of hemodynamic parameters is associated with reversal of decreased LDL-C levels.
Metabolic changes as reflected by decreased levels of HDL-C and increased insulin resistance have been described previously in the PAH population [4][5][6][7] . Low HDL-C in PAH patients has been associated with worse clinical outcomes and higher mortality. This finding was independent from cardiovascular risk factors, insulin resistance and severity of the disease. This phenomenon may be explained by antiinflammatory properties of HDL-C, its ability to enhance prostacyclin half-life and its protective role in endothelial dysfunction.
The phenomenon of reduced LDL-C levels has already been described in other chronic diseases such as rheumatoid arthritis 12 , cancer 20 , end-stage renal failure 11 or chronic heart failure 21 however, to the best of our knowledge this is the first study reporting changes in LDL-C levels and their prognostic role in PAH patients. Potential explanations for these findings include a chronic inflammatory state 22 , malnutrition 18 , and altered liver metabolism 17 .
Inflammatory cytokines are able to activate the reticuloendothelial system leading to enhanced receptor-independent clearance of LDL-C 23 . Conversely, suppression of the reticuloendothelial system leads to inhibition of LDL-C clearance 24 . Studies in rheumatoid arthritis have shown that suppression of inflammation with disease-modifying antirheumatic drugs or biological agents which target specific proinflammatory cytokines was accompanied by an increase in total cholesterol and LDL-C levels 25,26 . What is more, this increase was able to differentiate responders from non-responders to antirheumatic therapies 27 . An expanding body of knowledge suggests that inflammation and altered immune processes underlie both development and progression of PAH 14,28 . In fact we found a significant correlation between hsCRP and LDL-C in subgroup analysis of our PAH patients. However further studies are needed to fully understand the relationship between inflammation and lipid profile in this group.
As the liver plays a central role in lipid metabolism controlling both production and clearance of serum lipoproteins it seems likely that its dysfunction could contribute to altered LDL-C levels in PAH. Right heart dysfunction can lead to congestive hepatopathy, which is related to right atrial pressure and the degree of tricuspid regurgitation 17 . In fact, in our study we demonstrated that significant improvement of pulmonary hemodynamics in CTEPH patients was associated with normalization of LDL-C plasma levels. However, the exact pathophysiological pathways involved still need to be identified, since LDL-C normalization might not only be a direct consequence of hemodynamic improvement but also may result from improvement of the general condition of the patient 29,30 .
Low LDL-C levels have also been found in malnourished patients with chronic heart failure 9 . Accordingly, malnutrition could also contribute to the observed reduction of LDL-C in PAH patients. However, in our cohort none of the available markers of malnutrition such as body mass index (BMI) were associated with LDL-C level.
Several concomitant disorders and lifestyle interventions can influence LDL-C level 31 . Dietary changes, reduction of excessive body weight and increase in habitual physical activity may lead to lowering LDL-C concentration 32 . However we do not think that these lifestyle factors had significant impact on LDL-C level in our population. First, no specific recommendations on dietary patterns or weight reduction has been given to patients due to lack of scientific evidence. Second, the ability to increase of physical activity is limited in PAH patients due to exertional dyspnea. Third, most of our patients were within reference values for BMI and we did not find any correlation between LDL-C and BMI. Renal or hepatic impairment as well as thyroid abnormalities can also influence LDL-C level. In our patients, however, we did not find any associations between markers of these disorders (ALT, AST, creatinine) and LDL-C level. Thyroid abnormalities when present were well controlled. An important modulator of LDL-C concentration is the use of statins. Therefore, when we compared LDL-C levels to controls we analyzed separately all PAH patients and PAH patients who did not use statins. In both comparisons LDL-C levels were significantly decreased in PAH group. We monitored the use of statins at follow-up and did not find significant change in the number of patients who used these drugs as well as doses of statins in individual patients. In multivariable analysis we considered the use of statins as a potential confounder.
Our study suggests that traditional interpretation of hypercholesterolemia as a risk factor for increased mortality may not apply in PAH population. Instead, our finding of the deleterious signal of low LDL-C levels in PAH corresponds with the concept of cholesterol paradox and reverse epidemiology whereby lower levels of traditional risk factors are associated with worse prognosis 22 . The evidence of a survival advantage associated with higher cholesterol levels has been provided for several populations with debilitating disorders such as heart failure 9 , rheumatoid arthritis 12 , acute myocardial infarction 33,34 and in the elderly 35,36 .
The mechanism of interaction between LDL-C and clinical outcome has not been investigated in PAH. In other populations the importance of lipoproteins in down-regulation of the inflammatory immune response 9 and prevention of malnutrition 18 has been claimed. Lipids and lipoproteins can modulate inflammatory markers 22,[37][38][39][40] . Cholesterol can bind and neutralize endotoxin and liposaccharide components of infectious agents, thereby down-regulating the inflammatory state, which has been linked to progression of PAH 14,28,41 . Other data from experimental studies suggest that the effects of LDL-cholesterol on the arterial wall may be different in the pulmonary and systemic circulation. In rabbits, it has been demonstrated that hypercholesterolemia can increase pulmonary artery relaxation in response to methacholine 42 . Another experimental study 43 in rats with monocrotaline induced PAH showed that high-fat diet and hypercholesterolemia was associated with better prognosis as compared with standard diet. We have recently shown that low LDL-C levels, together with BMI < 25 kg/m 2 and higher pulmonary artery pressure are independent determinants of increased pulmonary artery stiffness and together explain most of the variation in pulmonary artery pulse wave velocity 16 . The different role of LDL-C in pulmonary circulation disease compared to the systemic circulation could also have contributed to the negative results of studies with statins in PAH patients 44 .
Our data suggests a strong linear relationship between LDL-C levels and mortality with a 2-to 5-fold increase in risk of death for every 1 mmol/l decrease in LDL-C. We cannot exclude the possibility that low LDL-C do not have any direct pathophysiologic connection to mortality and is only a consequence of the severity of PAH or malnutrition, which are also associated with poor survival. However, LDL-C levels were associate with all-cause mortality independently from markers of PAH severity -WHO FC and malnutrition -low BMI. Therefore, our data suggests that the prognostic impact of LDL-C in PAH cannot be fully explained by PAH severity or malnutrition, thereby supporting its potential role as an independent risk marker in PAH. Nevertheless, whether low LDL-C levels represent a modifiable risk marker in PAH or just an epiphenomenon with no direct pathophysiological impact on prognosis still needs to be elaborated.
Our study has several limitations. First, our data cannot provide explanations on the mechanisms by which low LDL-C levels influence survival in PAH patients. However, some plausible hypotheses can be proposed from our data for further studies. Second, as this is an observational study, we cannot establish a cause-and-effect relationship between LDL-C and outcome. Third, we could not ascertain the cause of death and compare the impact of low LDL-C on cardiovascular versus non-cardiovascular mortality or cardiovascular morbidity. Additionally we did not assess specifically dietary patterns of our patients which could have influenced LDL-C level. We also have not given the patients any specific advice on nutrition. This was mainly due to the fact that currently there is no evidence on the proper diet in patients with PAH.
Our study has several important strengths. First, the findings of LDL-C level alterations in the PAH population and their prognostic importance are novel and have not been previously reported. Second, as a control group we used a large representative sample of adults. Third, by using interventional and surgical treatment of CTEPH as a model for reliable, rapid improvement of hemodynamics, we could elaborate the dynamic relationship between LDL-C levels and severity of pulmonary hypertension beyond simple associations at baseline.
In summary, LDL-C levels are low in patients with PAH and are associated with an increased risk of death. Reversal of pulmonary hypertension in CTEPH patients increases low LDL-C levels. Future studies are necessary to gain more insights into the underlying pathophysiology of these findings and their role in the management of PAH.
Scientific RepoRts | 7:41650 | DOI: 10.1038/srep41650 Methods Patient population. Consecutive patients were recruited from the PH Program at the John Paul II Hospital (Krakow, Poland) between February 2009 and January 2015. Adult patients diagnosed with PAH except portopulmonary PAH were eligible for the study. PH was defined as an increase in mPAP ≥ 25 mmHg at rest as assessed by right heart catheterization 3 . PAH was defined as precapillary PH (pulmonary artery wedge pressure ≤ 15 mmHg) with PVR > 3 Wood units in the absence of other causes of precapillary PH such as lung diseases, CTEPH or other rare diseases 3 . As a control group we chose a representative sample of 2413 Polish residents aged 18-79 years from the NATPOL 2011 Survey, a cross-sectional observational study aimed to assess the prevalence and control of cardiovascular disease risk factors in Poland 45 . All-cause mortality was ascertained by data collection from the Department of Nationals' and Foreigners' Affairs and through phone follow-up. The observation period started at the first assessment of the patient in our center and extended until January 2016. We also studied patients with CTEPH from our centre to see whether PH reversal can change LDL-C levels. For this purpose, we formed two subgroups to be compared: CTEPH patients with a significant hemodynamic improvement (rCTEPH) by interventional (BPA), surgical (PEA) or drug treatment. The second subgroup included CTEPH patients not meeting our definition of hemodynamic improvement (nrCTEPH). Significant hemodynamic improvement was defined as a reduction of mPAP by ≥ 10 mmHg and to < 30 mmHg. This analysis was performed only in patients without changes in statin therapy from 3 months before baseline until follow-up. CTEPH was defined as precapillary PH characterized by mismatched perfusion defects on lung scan and specific diagnostic signs for CTEPH seen by conventional pulmonary cineangiography. The diagnosis of CTEPH could have been established only after at least 3 months of effective anticoagulation 3  Clinical assessment and laboratory measurements. Clinical assessment, laboratory measurements and hemodynamic evaluation were made at the first evaluation of a patient in our center. Clinical assessment included demographic information, patient's history, calculation of BMI, measurement of blood pressure, and N-terminal pro-brain natriuretic peptide (NT-proBNP) level, six minute walk test and assessment of the WHO FC, as well as other tests routinely performed at first evaluation of a PAH patient in our center but not analyzed in the study such as echocardiography. Obesity was defined as BMI ≥ 30 kg/m 2 , and overweight as BMI ≥ 25 kg/m 2 and < 30 kg/m 2 46 . Hypertension was diagnosed in patients with elevated systolic or diastolic blood pressure (BP) ≥ 140 or 90 mmHg, respectively or patients having been prescribed a BP-lowering drugs for high BP. On the contrary patients were not classified as hypertensives when they were assuming drugs with BP-lowering effect but prescribed for other reasons than hypertension, e.g. for angina 47 . Diabetes was diagnosed based on the plasma glucose criteria of the American Diabetes Association, either the fasting plasma glucose or the 2-h plasma glucose value after a 75-g oral glucose tolerance test 48 . Active smoker was defined as having smoked at least for 1 month during the previous 12 months; all others were classified as nonsmokers 16 . Peripheral venous blood was drawn after overnight fast in all patients (PAH and CTEPH group) on the day of right heart catheterization. Plasma lipids were measured twice in every patient. In the PAH group measurements were performed at enrollment and three months after enrollment while in the CTEPH group before starting therapy and after at least three months of therapy. Blood chemistry was performed using the Cobas ® 6000 analyzer (Roche Diagnostics International Ltd). LDL-C, HDL-C, and TG plasma concentrations were measured directly by using colorimetric technique with PEG-esterase, PEG-oxidase and peroxidase for HDL-C and LDL-C, and with lipase and peroxidase for TG concentrations. AST and ALT assay was performed accordingly to International Federation of Clinical Chemistry and Laboratory Medicine recommendations without pyridoxal phosphate activation. hsCRP was assessed using high sensitive latex assay. Creatinine level was assessed using kinetic colorimetric assay. Hemodynamic evaluation was performed during right heart catheterization using a Swan-Ganz-Catheter. Cardiac output was assessed using the oxygen consumption method. Clinical assessment and laboratory measurements performed in the NATPOL survey control group were described in detail elsewhere 45 . Statistical analysis. Continuous variables were reported as mean ± standard deviation or as median (interquartile range) and categorical variables as counts and percentages. For the comparison of continuous variables between two groups we used Student's t-test or Mann-Whitney U-test as appropriate and for categorical variables the chi-squared test. The Wilcoxon rank sum test was used for paired comparisons of continuous parameters assessed at baseline and at follow-up. For the comparison of LDL-C levels in three different groups of patients with PAH we used the Kruskal-Wallis test.
To compare patients with controls, the groups were matched for age separately in males and females. Multiple regression analysis using forward stepwise selection was used to assess significant associations between patients' characteristics and LDL-C levels. To exclude possible multicollinearity, we used multicollinearity diagnostic statistics and examined the variance inflation factor.
To assess the relationship between LDL-C and the risk of all-cause mortality, we used Cox proportional hazards regression. To allow for flexible modeling of the association and account for potential non-linearity, we modeled LDL-C using restricted cubic splines.
Models were hierarchically adjusted for major cardiovascular risk factors and markers of severity of PAH: model 1 -unadjusted analysis; model 2: adjusted for age and sex; model 3: adjusted additionally for diabetes, BMI, statin use, hypertension, smoking; model 4 -adjusted additionally for HDL-C, TG; model 5 -adjusted additionally for WHO-FC, etiology of PAH, PAH-specific treatment at the end of observation, NT-proBNP, 6-minute walk distance, RAP, PVR, CI; model 6 -similar to model 5 but PVR was substituted with mPAP. Sample size was calculated for the comparison of LDL-C between the rCTEPH and nrCTEPH at follow-up. We expected that in the rCTEPH group the LDL-C level would increase to the values of control group and that in the nrCTEPH group LDL-C would not change. Consequently, we expected that the difference of mean LDL-C between nrCTEPH group and rCTEPH group at follow-up would be 0.74 mmol/l. For alpha level of 0.05 and beta level of 0.2 the minimal number of patients in each group was 10. When comparing several parameters in the CTEPH group before and after treatment adjustment for multiple comparisons has been done. The significance level was set at alpha level of 0.05. Statistical analysis was performed with Statistica PL software [StatSoft, Inc. (2010). STATISTICA (data analysis software system), version 9.1. Tulsa, USA www.statsoft.com] and MedCalc version 11.6.1.0 (MedCalc Software, Mariakerke, Belgium) and STATA. 14.1 (StataCorp).