Dysregulated Arginine Metabolism, Hemolysis-Associated Pulmonary Hypertension, and Mortality in Sickle Cell Disease FREE

L-Arginine, the substrate for nitric oxide (NO) synthesis, is deficient in sickle cell disease (SCD).^1- 4 Increased NO consumption by cell free plasma hemoglobin⁵ and reactive oxygen species^6- 7 leads to decreased NO bioavailability^8- 9 that is exacerbated by decreased availability of the NO synthase substrate L-arginine. This state of resistance to NO is accompanied by a compensatory up-regulation of NO synthase and non–NO-dependent vasodilators.^10- 13 Under conditions of low arginine concentration, NO synthase is uncoupled, producing reactive oxygen species in lieu of NO,^14- 15 potentially further reducing NO bioavailability in SCD and enhancing oxidative stress. Recent reports of elevated arginase activity in SCD^16- 18 offer another avenue for decreased arginine bioavailability. Arginase, an enzyme that converts L-arginine to ornithine and urea, can limit NO bioavailability through increased consumption of the substrate for NO synthase.^19- 21 Arginase, which is found predominantly in the liver and kidneys, is also present in human red blood cells^22- 23 and can be induced in many cell types by a variety of cytokines and inflammatory stimuli.^20,24- 25 Furthermore, since arginine and ornithine compete for the same transport system for cellular uptake,^26- 27 a decrease in the ratio of arginine to ornithine resulting from increased arginase activity could further limit arginine bioavailability for NO synthesis.

We have previously reported high plasma arginase activity in 10 SCD patients with pulmonary hypertension.¹⁸ Death within a year of enrollment in that study occurred in 2 patients with the highest arginase activity. Pulmonary hypertension is common in both adults and children with SCD^28- 31 and is an important predictor of early mortality.²⁹ Pulmonary hypertension also develops in most other hereditary and chronic hemolytic anemias, including thalassemia,³² hereditary spherocytosis,³³ paroxysmal nocturnal hemoglobinuria,³⁴ and other hemolytic disorders,^35- 38 supporting the existence of a clinical syndrome of hemolysis-associated pulmonary hypertension.^29,39- 40

Since endothelial dysfunction may contribute to the pathogenesis of pulmonary hypertension through impaired production and bioavailability of and responsiveness to NO,^41- 44 we hypothesized that elevated arginase activity and dysregulated arginine metabolism may contribute to the endothelial dysfunction syndrome that occurs in SCD. Consistent with this hypothesis, arginase activity and alterations in arginine metabolic pathways have recently been implicated in the pathophysiology of primary pulmonary hypertension.⁴⁵ The goal of this study was to identify the source of increased plasma arginase activity in a large cohort of patients with SCD and to evaluate the contribution of dysregulated arginine metabolism to morbidity and mortality.

The patient population was sequentially enrolled between February 2001 and March 2005 and comprised 228 patients with SCD hemoglobinopathies for whom measurements of arginase (n = 140), arginine and ornithine (n = 209), or both (n = 121) were available. This study includes 188 of 195 patients who have been described in detail.²⁹ Written informed consent was obtained from each patient for an institutional review board–approved protocol to obtain clinical information, echocardiography, blood specimens, and prospective clinical follow-up data for research analysis. All laboratory assays were performed using the blood specimens that were collected prospectively at enrollment. In this population, right heart catheterization studies have confirmed that a tricuspid regurgitant jet velocity less than 2.5 m/s corresponds to normal pulmonary artery pressures, tricuspid regurgitant jet velocity of at least 2.5 m/s but less than 3.0 m/s corresponds to mild pulmonary hypertension, and tricuspid regurgitant jet velocity of at least 3.0 m/s corresponds to moderate to severe pulmonary hypertension.²⁹ Pulmonary hypertension was prospectively defined as a tricuspid regurgitant jet velocity of at least 2.5 m/s on Doppler echocardiography.

Thirty-six African American individuals recruited from a list of volunteers maintained at the National Institutes of Health were evaluated as controls for comparisons of laboratory and echocardiographic data. Volunteers similar in age and sex distribution to the sickle cell pulmonary hypertension screening cohort were selected. An additional 9 African American controls were enrolled specifically to obtain plasma and erythrocyte arginase measurements for comparison of these 2 measures of arginase activity.

Plasma amino acids were quantified via ion exchange chromatography (Beckman model 6300 amino acid analyzer, Fullerton, Calif) at the Mayo Clinic (Rochester, Minn) by methods recommended by the manufacturer. In addition to arginine and ornithine, citrulline, the endogenous precursor for de novo arginine synthesis, which occurs primarily in the kidney,^46- 47 and proline, which is synthesized from ornithine,⁴⁶ were also measured.

After 153 patients had been enrolled, it was decided to obtain measurements of plasma arginase activity in all patients for whom a sufficient quantity of stored frozen plasma was available (n = 140). Arginase activity was determined as the conversion of L-arginine that is carbon 14 (¹⁴C)–labeled on the guanidino carbon to ¹⁴C-labeled urea, which was converted to ¹⁴C-labeled carbon dioxide by urease and trapped as ¹⁴C-labeled sodium carbonate for scintillation counting, as previously described.⁴⁸ Briefly, aliquots of plasma or red blood cell lysate were spun down on collection and frozen at −80°C. Thawed samples were later incubated for 10 minutes at 55°C in complete assay mixture lacking arginine. The reaction was initiated by addition of labeled arginine and incubation was continued at 37°C for 2 hours. The reaction was terminated by heating at 100°C for 3 minutes. Samples were incubated with urease at 37°C for 45 minutes, and ¹⁴C-labeled sodium carbonate was trapped on sodium hydroxide–soaked filters following acidification of the samples with hydrochloric acid to volatilize the ¹⁴C-labeled carbon dioxide.

Plasma levels of endothelial and platelet-specific soluble (s) adhesion molecules (sE-selectin, sP-selectin, vascular cell adhesion molecule 1 [sVCAM-1], and intracellular adhesion molecule 1 [sICAM-1]) were measured using commercially available enzyme-linked immunosorbent assay (ELISA) kits (R&D Systems, Minneapolis, Minn).

Plasma levels of hemoglobin were measured by ELISA as previously described.⁴⁹

Myeloperoxidase levels were measured by ELISA (PrognostiX, Cleveland, Ohio).

Results are presented as mean and standard deviation, geometric mean and 95% confidence interval (CI), or percentage of participants with characteristic, as appropriate. Two-sided 2-sample t tests were used to compare continuous variables in 2 groups (for example, amino acid values in sickle cell patients and healthy controls). Linear regression on 3 categories of tricuspid regurgitant jet velocity (<2.5, 2.5-2.99, and ≥3.0 m/s; coded 0, 1, and 2, respectively) was used to evaluate relationships between amino acid values or arginase and level of pulmonary hypertension in patients with SCD. Since normal distributions provided poor approximations for many of the variables of interest, bivariate correlations were assessed using the Spearman rank correlation coefficient.

Multiple regression analysis of arginase activity used log₁₀-transformed values for arginase, as well as for laboratory correlates for which normal distributions fit logarithms better than untransformed values; use of logarithms also reduced the influence of extremely high values of independent variables. This modeling used a stepwise procedure to add independent variables, beginning with the variables most strongly associated with log₁₀ arginase and considering all potential covariates associated with log₁₀ arginase with P≤.15. Deletion of variables after initial inclusion in the model was allowed. The procedure continued until all independent variables in the final model had P≤.05, adjusted for other independent variables, and no additional variable had P≤.05. The final model was confirmed by fitting a similar model using ranks of all variables, including log₁₀ arginase.

Proportional hazards (Cox) regression was used to study relationships between mortality in patients with SCD and covariates of interest. The proportional hazards assumption was evaluated by assessing whether scaled Shoenfeld residuals⁵⁰ showed a trend with time.

In all regression modeling, observations with missing values for any variable included in a particular model were deleted; no imputation of missing values was performed.

P<.05 was considered statistically significant; no adjustment for multiple comparisons was made. Analysis was performed using NCSS 2004 software (Number Cruncher Statistical Systems, Kaysville, Utah).

The analyses described above were preplanned. Additional analyses evaluated arginase activity within erythrocytes and the association between arginase activity in plasma and red blood cell lysate. These analyses involved an unselected subgroup of 16 patients, as well as 45 controls (the original 36 and the additional 9 for whom plasma and erythrocyte arginase measurements were made).

General characteristics of the study population of patients with SCD and controls are shown in Table 1.

Plasma amino acid levels in SCD patients were compared with those in African American control participants without SCD (Table 2). An abnormal amino acid profile was observed in patients with SCD that is consistent with altered arginine metabolism. The observed dysregulation of the arginine-to-NO metabolism was greatest in SCD patients with pulmonary hypertension. Although mean (SD) plasma arginine concentrations were low in patients with SCD compared with healthy controls (41 [16] vs 67 [18] μmol/L; P<.001), these levels were similar in patients with and without pulmonary hypertension. However, plasma ornithine levels were elevated in patients with SCD and severe pulmonary hypertension (tricuspid regurgitant jet velocity ≥3.0 m/s), and the elevation was significant compared with patients with SCD without pulmonary hypertension (mean [SD], 81 [24] vs 63 [21] μmol/L; P<.001 by 2-sided t test) and is likely the result of elevated arginase activity. The ratio of arginine to ornithine, an indirect measure of arginase activity and relative arginine bioavailability, was low in patients with SCD compared with controls (mean [SD], 0.71 [0.39] vs 1.20 [0.49]; P<.001) (Table 2 and Figure 1A) and was particularly low in the group with severe pulmonary hypertension (mean, 0.49 [SD, 0.18]). Ratios were clearly lower even in patients with SCD without evidence of pulmonary hypertension than in controls (mean [SD], 0.74 [0.41] vs 1.20 [0.49]; P<.001 by 2-sided t test). The association between arginine-ornithine ratio and level of pulmonary hypertension was significant by linear regression of the ratio on category of tricuspid regurgitant jet velocity (P = .03) (Table 2), but ratios were low only in patients with severe pulmonary hypertension compared with patients with no evidence of pulmonary hypertension by 2-sided t test adjusted for unequal variances (mean [SD], 0.49 [0.18] vs 0.74 [0.41]; P<.001).

Plasma proline concentrations were also significantly increased in patients with SCD compared with controls (mean [SD], 210 [75] vs 154 [59] μmol/L; P<.001 (Table 2). Patients without pulmonary hypertension had higher proline levels than controls (mean [SD], 202 [70] vs 154 [59]; P<.001 by 2-sided t test), and even higher levels occurred in patients with severe and moderate pulmonary hypertension (mean [SD], 245 [88] vs 219 [80] vs 202 [70] μmol/L for tricuspid regurgitant jet velocity ≥3.0, 2.5-2.99, and <2.5 m/s, respectively; P = .01 by linear regression analysis; Table 2). This possibly indicates increased conversion of ornithine to proline in SCD that is amplified in patients with pulmonary hypertension (Table 2) and may well account for the fact that plasma ornithine levels did not increase in most patients with SCD, even as plasma arginase activity increased and plasma arginine levels declined. However, the possibility that reductions in proline catabolism also may contribute to the elevated proline concentrations cannot be ruled out. More precise explanations of the changes in amino acid levels in this patient population will require further studies using isotopic tracer methods.

Although citrulline levels tended to increase slightly with level of pulmonary hypertension in patients, mean levels were almost identical in patients and controls (Table 2). Citrulline was significantly correlated with creatinine level (ρ = 0.51; P<.001), which is consistent with impaired renal function. The ratio of arginine to ornithine plus citrulline, which takes into account the impact of renal dysfunction on global arginine bioavailability, showed very similar relationships among controls and patients to those for arginine-ornithine ratio (Table 2). In aggregate, these data indicate significant modulation of L-arginine metabolism in SCD that is associated with the development of pulmonary and renal vasculopathy.

To understand the mechanism responsible for dysregulation of L-arginine metabolism, plasma arginase activity was measured in patients with SCD and controls. Plasma arginase activity was significantly elevated in patients with SCD (n = 140) (mean [SD], 2.1 [2.1] μmol/mL per hour) compared with controls (n = 36) (mean [SD], 0.4 [0.2] μmol/mL per hour; P<.001 by t test on log₁₀ arginase) (Figure 1B). In patients with SCD, arginase activity tended to increase with level of pulmonary hypertension (mean [SD], 1.9 [1.8], 2.6 [2.8], and 2.8 [2.0] μmol/mL per hour for tricuspid regurgitant jet velocity <2.5, 2.5-2.99, and ≥3.0, respectively), although the association was not significant in linear regression of log₁₀ arginase on the 3 categories of tricuspid regurgitant jet velocity, coded 0, 1, or 2 (R² = 0.017; P = .13). However, even in patients without pulmonary hypertension, patients with SCD had significantly higher arginase activity compared with controls (mean [SD], 0.4 [0.2] vs 1.9 [1.8]; P<.001). Arginase activity was significantly correlated with arginine-ornithine ratio (ρ = −0.34; P<.001) (Figure 1C); however, it likely is only one of several factors affecting this ratio and arginine bioavailability in patients with SCD.

The relationship between arginase activity and clinical laboratory markers of disease severity was evaluated to identify mechanisms for increased enzymatic activity and associated effects on organ function (Table 3). Plasma arginase activity was significantly associated with several markers of increased hemolytic rate, including cell free plasma hemoglobin (ρ = 0.56; P<.001) (Figure 2), lactate dehydrogenase (ρ = 0.35; P<.001), aspartate aminotransferase (ρ = 0.34; P<.001), and hematocrit (ρ = −0.20; P = .02). The lack of correlation between arginase and reticulocyte count in this cohort likely reflects the suppressive effects of transfusions, renal impairment, and hydroxyurea therapy on reticulocytosis in the most severely affected patients. Other significant associations included oxygen saturation, white blood cell count, myeloperoxidase, alanine aminotransferase, endothelial and platelet-specific soluble adhesion molecules (sE-selectin, sP-selectin, sVCAM-1, and sICAM-1), triglycerides, and cholesterol (Table 3). No association of arginase activity with age (ρ = −0.09; P = .31) or sex (P = .63 by t test on log₁₀ arginase) was identified. There was no evidence of association between elevated arginase activity and markers of renal function (Table 3).

In multiple regression analysis of log₁₀ arginase activity, all variables associated with log₁₀ arginase with P≤.15 in Table 3 were considered in a stepwise model-fitting process. In the final model, log₁₀ arginase activity was related to log₁₀ cell free hemoglobin, log₁₀ sP-selectin, and log₁₀ triglycerides (n = 110; R² = 0.40; adjusted P<.001 for all independent variables). Although the strongest association was with hemolysis, these data also demonstrate a link between arginase activity and abnormal lipid metabolism as well as adhesion. The low-variance inflation factors (≤1.11 for all 3 covariates in the model) suggest that there was no multicollinearity problem in this model and that each of the 3 covariates was independently related to log₁₀ arginase. However, since the residuals were not well fit by a normal distribution, the analysis for the final model was repeated using ranks for all the variables. The results were similar (R² = 0.43; adjusted P<.001 for ranks of cell free hemoglobin and sP-selectin and P = .001 for rank of triglycerides).

No adjustment was made for multiple comparisons in these analyses. This seems appropriate, since the objective of this study was to detect potentially important associations involving arginase, not to control the overall type I error rate. Even if we made a conservative adjustment, however (for example, multiplying P values in Table 3 by the number of correlations shown) the major associations would still have P<.05. Furthermore, regression modeling enabled us to estimate the number of variables, among those considered, that are independently associated with arginase activity and with mortality.

These data indicate that increased plasma arginase activity in patients with SCD is associated with intravascular hemolysis, endothelial activation, and inflammation.

To further identify the source of increased plasma arginase activity, arginase activity was also determined for red blood cell lysates of 45 controls and a subset of 16 patients with SCD in whom both frozen plasma and red blood cell lysates were available for comparison (Figure 3A). Specific activities of arginase in red blood cell lysate of patients with SCD were significantly higher than those of controls (mean [SD], 37.7 [2.9] vs 23.5 [1.7] nmol/mg per min; P<.001 by t test on log₁₀ arginase). For purposes of comparison, “normal-range” boundaries for controls were set arbitrarily at approximately the 80th percentile for arginase activities of both red blood cell lysates and plasma. Two thirds of all control values fell within these boundaries, while, in striking contrast, 94% of all values for plasma and erythrocyte arginase activities of patients with SCD fell outside of these boundaries (Figure 3B).

Information on deaths was collected during a follow-up period of up to 49 months. Between enrollment in the study and March 2005, 18 patients with SCD had died, with a median survival time of 14 months (range, 2-41 months). Median follow-up was 33 months for the 210 patients who survived (range, 7-49 months). Nine patients had not responded to attempts to contact them and were considered lost to follow-up. Confirmation of all deaths with death certificates and the absence of available death certificates in the United States for patients lost to follow-up suggest that we have not missed any deaths and all patients lost to follow-up were alive at the time of data analysis.

Fourteen of the 18 patients who died had a tricuspid regurgitant jet velocity of at least 2.5 m/s; and by proportional hazards regression analysis, the presence of pulmonary hypertension by this definition was the most significant risk factor for death (risk ratio, 7.4; 95% CI, 2.4-22.4; P<.001) (Table 4). Plasma amino acid concentrations and plasma arginase activities were available for all 18 who died. Low ratios of plasma arginine to ornithine and arginine to ornithine plus citrulline were associated with mortality in proportional hazards regression (Table 4 and Figure 4). After adjustment for high tricuspid regurgitant jet velocity and log₁₀ creatinine level, the ratios of arginine to ornithine and arginine to ornithine plus citrulline remained significantly related to mortality; thus, either of these ratios may be an independent risk factor for death in patients with SCD. Age was not significantly related to mortality after adjustment for the above variables. Estimated adjusted risk ratios (for 25th percentile relative to 75th percentile) were 2.2 (P = .02) for ratio of arginine to ornithine and 2.9 (P = .007) for ratio of arginine to ornithine plus citrulline (Table 4). There was no evidence that risk ratios were different for patients with low (<2.5 m/s) and high (≥2.5 m/s) tricuspid regurgitant jet velocity.

The assumption of proportional hazards (ie, constant risk ratio) for ratios of arginine to ornithine and arginine to ornithine plus citrulline seemed reasonable, since scaled Shoenfeld residuals⁵⁰ for these variables did not show a strong trend with follow-up time.

Arginase activity was not directly associated with mortality; however, in shifting L-arginine metabolism away from NO production and toward ornithine-dependent pathways, increased arginase activity may contribute to events that put patients at risk of early death.

This large cohort study of patients with SCD not only confirms previous reports of elevated arginase activity in SCD^16- 18,23 but, more importantly, also demonstrates important associations among dysregulated L-arginine metabolism, pulmonary hypertension, and prospective mortality. Additionally, these data suggest that elevated plasma arginase activity in SCD is primarily the consequence of erythrocyte arginase release during intravascular hemolysis, with some possible contribution from endothelial cell activation associated with inflammation, and liver injury in some patients. Global arginine bioavailability is diminished further through impairment of de novo arginine synthesis in patients with renal dysfunction. These observations support a novel mechanism of disease that links oxidative stress,^6- 7 chronic organ damage,^13,51 and hemolytic rate^{5,29,39- 40,52} to endothelial dysfunction and pulmonary hypertension.

Arginase is an intracellular enzyme that appears in plasma only after cell damage or death. Thus, inflammation, chronic end-organ damage and hemolysis are all potential sources of elevated arginase activity in SCD. Arginase activity is higher in immature red blood cells and reticulocytes,²³ both of which are plentiful in SCD. Nearly 94% of the patients with SCD had arginase values outside the normal-range boundaries, with two thirds of the SCD population exhibiting elevated arginase activity in both plasma and red blood cell lysate (Figure 3). Elevated erythrocyte arginase activity has also been reported in patients with megaloblastic anemia²³ and thalassemia patients.^53- 54 In addition to NO scavenging by cell free plasma hemoglobin, erythrocyte arginase release during hemolysis may represent a mechanistic link between the pulmonary hypertension syndrome that develops in SCD and other hemolytic disorders, such as thalassemia, hereditary spherocytosis, and paroxysmal nocturnal hemoglobinuria,^29,39 and warrants further investigation. These combined mechanisms support a model whereby hemolysis produces a state of reduced endothelial NO bioavailability that might contribute to endothelial dysfunction, intimal and smooth muscle proliferation, oxidant stress, and thrombosis.⁵

Association of arginase activity with common markers of inflammation such as white blood cell count and myeloperoxidase is not unexpected because arginase is induced in monocytes in response to helper T-cell type 2 cytokines,²⁴ inflammatory mediators that are elevated in SCD.⁵⁵ Elevated arginase activity has also recently been discovered in asthma,^56- 59 another T-cell type 2 cytokine–related disorder^60- 61 that may be relevant to the pathophysiology of SCD since 30% to 70% of pediatric patients with SCD have reactive airways.^62- 63

An association of arginase activity with plasma soluble adhesion molecules is a novel observation, although also not unexpected since endothelial cells assume an inflammatory phenotype in SCD.^64- 65 In particular, endothelial sP-selectin is thought to contribute to the microcirculatory abnormalities in SCD.^66- 68 Significant correlations with sE-selectin, sVCAM-1, and sICAM-1 were also identified; however, only an independent relationship of arginase activity to sP-selectin was maintained in multiple regression analysis. This may also represent a link between arginase activity and platelet activation. It remains to be determined whether these findings represent a causal relationship between arginase and adhesion molecules or reflect a common response to endothelial injury or proinflammatory cytokine release.

Equally intriguing is the association of arginase activity with triglycerides and cholesterol, given growing support for triglyceride involvement in endothelial dysfunction.^69- 71 Although triglyceride levels are generally lower in SCD,⁷² they are increased by tumor necrosis factor α, interleukin 1, and other cytokines,⁷³ which are elevated in SCD.⁷⁴ Although the true significance of this finding remains to be determined, this relationship may be a reflection of a broader common pathway between arginase activity and abnormal lipid metabolism in endothelial dysfunction and warrants further investigation.

Plasma arginine concentration in SCD is approximately 40 to 50 μmol/L at baseline, well below the arginine concentration at which the rate of cellular uptake via the arginine transport system is half-maximal (K_m, approximately 100 μmol/L^26- 27). Thus, even modest reductions in plasma arginine concentration can significantly affect cellular arginine uptake and bioavailability. In the current study, we find that the ratio of arginine to ornithine is associated with arginase activity. Because arginine and ornithine compete for uptake via the same transport system,^26- 27 decreases in the arginine-ornithine ratio in patients with SCD also represent decreases in arginine bioavailability. Consistent with a metabolic marker of arginine bioavailability, a low arginine-ornithine ratio was associated with worsening severity of pulmonary hypertension and independently associated with mortality. Because conversion of citrulline to arginine occurs primarily within the kidney,^46- 47 the increased mortality risk ratio observed after citrulline was included in the Cox regression analysis probably reflects effects of renal dysfunction on arginine bioavailability. Indeed, citrulline levels trended higher in SCD patients with pulmonary hypertension and correlated with rising creatinine levels (Spearman ρ = 0.51; P<.001).

The alterations in arginine metabolism in SCD and their implications for clinical complications are summarized in Figure 5. As arginase and cell free hemoglobin correlate strongly with one another and are both released from erythrocytes as they undergo hemolysis, the independent contribution of each toward decreasing bioavailability of NO cannot be determined in this clinical study, and causality cannot be assumed. However, increased catabolism of arginine via arginase may not only compromise the ability to synthesize NO but also may contribute to the pulmonary vascular remodeling that occurs in pulmonary hypertension through increased production of ornithine, a precursor for synthesis of proline and polyamines⁴⁶ (Figure 5), which are required for the collagen synthesis and cell proliferation, respectively, that occur in vascular remodeling.^57,59 Analogous to its proposed roles in asthma,^57,59 elevated proline levels in SCD demonstrated in this and other studies^3,75 may contribute to pulmonary fibrosis and lung pathogenesis by promoting collagen synthesis.

These findings suggest that dysregulated arginine metabolism is associated with intravascular hemolysis, inflammation, and endothelial cell activation. Alterations in the normal balance of arginine and its catabolic byproducts, ornithine, citrulline, and proline, are associated with pulmonary hypertension and prospective risk of death. These easily measured ratios may provide clinicians with an objective index of disease severity that could identify patients at risk and allow for earlier and more aggressive therapeutic intervention.