Relationship of erythropoietin, fetal hemoglobin, and hydroxyurea treatment to tricuspid regurgitation velocity in children with sickle cell disease

Studies in both adults¹  and children^2,3  with sickle cell disease have found correlations between hemolysis and pulmonary hypertension, a complication associated with increased mortality.¹  Intravascular hemolysis may contribute to a hemolytic vasculopathy in part by scavenging nitric oxide, a key modulator of microvascular function⁴  and by limiting availability of arginine, the substrate for nitric oxide synthase.⁵  Hemolysis, however, does not fully explain the finding of pulmonary hypertension in this setting. Pulmonary hypertension develops in patients with hemoglobin SC disease or Sβ⁺-thalassemia, conditions with a substantially lower hemolytic rate than that of homozygous hemoglobin SS disease.¹  Furthermore, once hemoglobin SC disease patients develop pulmonary hypertension, their prognosis is as poor as in hemoglobin SS patients with this complication.⁶  Hydroxyurea decreases hemolysis^7,8  and induces nitric oxide in endothelial cells,⁹  but the largest prospective studies of patients with sickle cell disease have not found less pulmonary hypertension among those receiving hydroxyurea.^1,3,10  It is also not clear whether high hemoglobin F levels reduce pulmonary hypertension risk in sickle cell patients. Some studies have found an association of high hemoglobin F with lower pulmonary hypertension risk,^11-14  but several others have detected no such association.^1,2,15-18 

We recently reported a prospective, multicenter study of 310 children and adolescents with sickle cell disease at steady state in which tricuspid regurgitation velocity of 2.6 m/s or higher occurred in 11% of participants and had independent associations with hemolysis and hemoglobin oxygen desaturation.³  In addition, we have found that an elevated screening tricuspid regurgitation velocity in children and adolescents with sickle cell disease predicts functional impairment over 2 years of follow-up (V.R.G., unpublished observations, April 2009). The present report involves an investigation in these and additional participants of potential relationships among pulmonary hypertension and serum erythropoietin concentration, hemoglobin F levels, and hydroxyurea use. We also attempt to explain why, despite the well-documented beneficial effects of high hemoglobin F levels and hydroxyurea treatment, neither spontaneous nor hydroxyurea-induced elevations of hemoglobin F have been convincingly demonstrated to lower pulmonary hypertension risk in sickle cell disease patients.

This report includes 399 children and adolescents with sickle cell disease from 3 to 20 years who were evaluated at steady state as previously described.³  Of the children in the present report, 307 were also included in our previous publication³  and 92 were newly enrolled. The patients had hemoglobin SS, SC, Sβ-thalassemia, or other major sickling phenotypes as confirmed by hemoglobin electrophoresis or high-performance liquid chromatography. One hundred fifty patients (38%) were receiving hydroxyurea therapy. Doppler echocardiography was used to estimate systolic pulmonary artery pressure through measurement of the tricuspid regurgitation velocity. Transthoracic echocardiography was performed using the Philips Sono 5500/7500 or iE33, Acuson Sequoia, or General Electric VIVID 7 or VIVID I instruments. Cardiac images were obtained, measurements performed, and studies interpreted centrally according to guidelines of the American Society of Echocardiography. A nonencouraged 6-minute walk test was performed. Participants were recruited at 3 centers: Howard University, Children's National Medical Center, and the University of Michigan. The institutional review boards of all participating institutions approved the study protocol, and all subjects provided written informed consent to participate in accordance with the Declaration of Helsinki.

Serum concentrations of erythropoietin were measured with a commercially available enzyme-linked immunosorbent assay kit (R&D Systems) following the manufacturer's recommendations. Other measurements were performed as previously described.³  Hemoglobin F was determined by high-performance liquid chromatography or hemoglobin electrophoresis by the laboratories of each institution. In some cases the hemoglobin electrophoresis results did not report a value for hemoglobin F; these cases were considered to have missing hemoglobin F data in this paper rather than assigning them 0 or an arbitrarily low hemoglobin F value.

For continuous variables that did not follow a normal distribution, the best transformation to a normal distribution was made for statistical analyses. To overcome colinearity of related markers and to point to underlying mechanisms, principal component analysis of 4 markers of hemolysis (reticulocyte count, and serum concentrations of aspartate aminotransferase, lactate dehydrogenase, and total bilirubin) was performed.³  Principal component analysis produces several components equal to the number of variables in the analysis; each component represents a normalized standard distribution with a mean value of 0. In this analysis, the first component had an Eigen value of 2.56 (explaining 64% of variability) and was termed a hemolytic component. Continuous variables were compared between patients according to hydroxyurea treatment at the time of the study with analysis of variance models that adjusted for severe (ie, Hbs SS, Sβ⁰-thalassemia, and SD^LA) versus mild (Hbs SC and Sβ⁺-thalassemia) sickling phenotype and other important covariates. Categoric variables were compared with the χ² test. The associations of erythropoietin and tricuspid regurgitation velocity with other variables were assessed by Pearson correlation or multiple linear regression. In these analyses, up to 5 outlier values were excluded. P values less than .05 were considered statistically significant. Analyses were performed with STATA 10.0 (StataCorp).

Table 1 summarizes the general clinical characteristics of this cohort. Seventy-six percent had the severe sickling phenotypes of hemoglobin SS, Sβ⁰ thalassemia, or SD^LA, and 38% were receiving hydroxyurea treatment at the time of the study. Twenty-two percent had tricuspid regurgitation velocity of 2.5 m/s or higher, and 11% had velocity of 2.6 m/s or higher. There was only 1 participant who had a tricuspid regurgitation velocity higher than 2.9 m/s.

Table 2 shows that sickle cell disease patients taking hydroxyurea were older and more likely to have severe sickling phenotypes (hemoglobin SS, Sβ⁰ thalassemia, or SD^LA) than those not taking the medication. They had higher values for hemoglobin and mean corpuscular volume, and lower values for white blood cell and reticulocyte counts and the hemolytic component, indicating compliance with the medication and an effect of the drug on the body's hematologic status. In addition, hydroxyurea-treated patients had significantly higher values for hemoglobin oxygen saturation, hemoglobin F, and erythropoietin. The tricuspid regurgitation velocity did not differ according to treatment with hydroxyurea among all phenotypes as shown in Table 2 or when restricted to patients with hemoglobin SS (data not shown). For these analyses, we statistically adjusted for patients who were on a chronic transfusion program. Essentially the same results were found if these patients were excluded from the analyses.

With the exception of erythropoietin-expressing tumors and rare conditions of altered hypoxia sensing, erythropoietin expression sensitively reflects tissue oxygenation status.^19,20  In fact, hypoxia-inducible factor-α, the master regulator of the body's response to hypoxia, was discovered by studying the regulation of the erythropoietin gene.¹⁹  In our data, bivariate analyses revealed significant relationships between lower hemoglobin concentration and log erythropoietin (n = 356, r = −0.66, P < .001), between lower hemoglobin oxygen saturation and log erythropoietin (n = 354, r = −0.28, P < .001), and between higher hemoglobin F percentage and log erythropoietin (n = 189, r = 0.21, P = .003). Multiple linear regression confirmed that lower hemoglobin concentration (P < .001) and higher log hemoglobin F percentage (P < .001) each correlated independently with higher log erythropoietin concentration among 179 patients with sickle cell disease (Table 3, “All patients”). In subanalyses of the patients being treated with hydroxyurea (Table 3, “Patients on hydroxyurea”) and those not receiving hydroxyurea (Table 3, “Patients not on hydroxyurea”), the inverse relationship between hemoglobin concentration and erythropoietin persisted in both subgroups (P ≤ .005). An independent positive association of hemoglobin F with erythropoietin was found in the patients receiving hydroxyurea (P < .001) but this relationship was not statistically significant in the analysis of patients not on hydroxyurea (P = .09).

Erythropoietin, hemoglobin F, and hemolytic component correlated positively with tricuspid regurgitation velocity, whereas hemoglobin concentration and hemoglobin oxygen saturation correlated negatively in analyses adjusted for age, sex, chronic transfusion program, and research site (Table 4). When the same analyses were restricted to patients who were being treated with hydroxyurea, the hemolytic component (P < .001), erythropoietin level (P < .001), and hemoglobin F category (P = .003) correlated positively with tricuspid regurgitation velocity, whereas hemoglobin concentration (P < .001) and hemoglobin oxygen saturation (P = .028) correlated negatively. The positive correlation of hemoglobin F percentage (P = .2) with regurgitation velocity did not achieve statistical significance. When restricted to patients who were not taking hydroxyurea, the hemolytic component (P < .001) and erythropoietin concentration (P < .001) correlated positively with regurgitation velocity, whereas hemoglobin concentration (P = .001) correlated negatively. The positive correlations of hemoglobin F percentage (P = .2) and hemoglobin F category (P = .1) and the negative correlation of oxygen saturation (P = .2) with regurgitation velocity were not statistically significant (data not shown).

Both the degree of hemolysis as reflected in the hemolytic component and the erythropoietin concentration were significantly and independently associated with tricuspid regurgitation velocity among 294 patients (Table 5, “Analysis based on all participants with erythropoietin available”). Hemoglobin F percentage was available in a subset of 142 participants. Multivariate analysis in these patients showed that hemoglobin F was also an independent positive predictor of higher tricuspid regurgitation velocity in addition to the degree of hemolysis (Table 5, “Analysis based on participants with value for hemoglobin F available”).

This study indicates that serum erythropoietin and hemoglobin F levels, in addition to or in concert with hemolysis, are associated with higher tricuspid regurgitation velocities in children and adolescents with sickle cell disease. In our patient population and also in some published studies, hydroxyurea treatment failed to predict lower tricuspid regurgitation velocity, despite its association with lower hemolysis. Our findings suggest that this might be explained by the induction of higher erythropoietin and hemoglobin F levels.

Increased erythropoietin has been reported to protect from the development of pulmonary hypertension in some studies²¹  and to be associated with the development of pulmonary hypertension in others.^22,23  In a study of 124 adults with sickle cell disease, no relationship between erythropoietin and pulmonary hypertension was detected.²⁴  We, however, observed that serum erythropoietin concentration is associated with higher tricuspid regurgitation velocity in sickle cell disease patients even after adjustment for the degree of hemolysis and other significant covariates, providing evidence for the concept that erythropoietin may be related to the development of pulmonary hypertension. Circulating erythropoietin concentrations reflect the degree of tissue hypoxia, and are known to increase with lower hemoglobin concentrations and hemoglobin oxygen saturations.²⁰  Therefore, the observed association of erythropoietin with higher tricuspid regurgitation velocity could be reflective of an association of hypoxia with elevated velocity rather than a primary relationship.

Hemoglobin F has high affinity for oxygen²⁵  due to its low affinity for 2,3-diphosphoglycerate.²⁶  Hemoglobin F also inhibits hemoglobin S polymerization,²⁷  which would be expected to reverse in part the low oxygen affinity of hemoglobin S that results from its polymerization.²⁸  Therefore, hemoglobin F could conceivably contribute to a relative tissue hypoxia, despite its well-documented effect in ameliorating the course of sickle cell disease and increasing patient survival.²⁹  In a sense, hemoglobin F's left shifting of the oxygen saturation curve is similar to what can be seen in sickle cell patients after red cell exchanges: both hemoglobin levels and blood oxygen affinity increase modestly after exchange and are associated with increased exercise capacity.³⁰  It is interesting also that thalassemia intermedia patients with high hemoglobin F levels have significantly higher erythropoietin concentrations than those with low hemoglobin F levels despite their similar degree of anemia.³¹ 

In the present study, lower hemoglobin concentration and higher hemoglobin F each correlated independently and strongly with higher erythropoietin concentration. Furthermore, in multiple linear regression analyses, both erythropoietin and hemoglobin F independently were associated with higher regurgitation velocities in an interchangeable manner. From this perspective, the associations of greater levels of erythropoietin and hemoglobin F with higher regurgitation velocities may serve to reflect the known association of hypoxia with the development of pulmonary hypertension in other conditions. On the other hand, erythropoietin has functions other than the stimulation of erythropoiesis, such as regulation of the development of endothelial progenitor cells,³²  and the inducement of such processes might contribute to vascular remodeling and the risk of pulmonary hypertension. Our finding of a positive association of hemoglobin F with tricuspid regurgitation velocity in children and adolescents is in contrast to studies in adults with sickle cell disease that reported no such association¹⁷  or an association with lower regurgitation velocities.¹² 

Hydroxyurea treatment lowers hemolysis^7,8  and decreases morbidity and mortality in patients with sickle cell disease.^33,34  Hydroxyurea may also have an impact on nitric oxide signaling by evoking nitric oxide synthase and decreasing arginine levels^9,35 ; the agent promotes the synthesis of nitric oxide by endothelial cells.⁹  These factors may serve to protect from pulmonary hypertension. However, although some investigators have reported that hydroxyurea therapy provides a protective effect from pulmonary hypertension,¹²  most reports including the largest, prospective investigations of pulmonary hypertension in sickle cell disease have not found such a protective effect.^1,10,17  The present study provides some possible insights into this observation. Compared with children not receiving hydroxyurea at the time of study, those receiving hydroxyurea had higher hemoglobin levels, mean corpuscular volumes, and hemoglobin F concentrations, and lower leukocyte counts, indicating their compliance with the regimen for a sufficient time to experience its effects on hematopoiesis. At the same time, our patients on hydroxyurea had higher erythropoietin concentrations, as has been previously reported in sickle cell disease patients on this drug,^36,37  and higher hemoglobin F percentages, features that are associated with higher tricuspid regurgitation velocities. Interestingly, a nitric oxide signal for fetal hemoglobin induction has been described.⁹  Tricuspid regurgitation velocities did not differ according to whether the children were receiving hydroxyurea, suggesting that factors associated with higher tricuspid regurgitation velocities may have been balanced by those associated with lower velocities. Alternatively, the fact that the patients included in this study had received hydroxyurea in a nonrandomized manner represents a potentially important confounder. We cannot rule out the possibility that patients receiving hydroxyurea were at higher risk for pulmonary hypertension before starting therapy than those not treated with hydroxyurea, and that they might have had higher tricuspid regurgitation velocities if they were not on hydroxyurea. Consistent with this possibility, a recent publication reported reduction in mean pulmonary artery pressures with hydroxyurea therapy in 5 patients.³⁸  On the other hand, prospective administration of hydroxyurea in the Multicenter Study of Hydroxyurea in Sickle Cell Anemia did not influence concentrations of N-terminal pro–brain natriuretic peptide, an index of pulmonary hypertension.¹⁰ 

There are several additional limitations to our study. First, we did not collect information on how long the children had been receiving hydroxyurea or what their doses were. Second, although, as shown in Table 1, there were significant differences in hemoglobin concentration, hemoglobin F percentage, and mean corpuscular volume according to hydroxyurea in the present study, these differences were not as great as those reported in the HUG-KIDS study.³⁹  Thus, the hydroxyurea group may not have been receiving optimal amounts of hydroxyurea and this may have contributed to a lack of association with lower tricuspid regurgitation velocity. Third, we have not controlled for low arginine bioavailability that may be independently associated with high tricuspid regurgitation velocity in adult studies.^1,5  Fourth, the reliability of a single echocardiographic measurement of tricuspid regurgitation velocity has not been established in children with sickle cell disease. Fifth, hemoglobin F results were not available for approximately one-half of the group studied.

Our findings have implications for future studies examining the causes and treatment of pulmonary hypertension in patients with sickle cell disease. It is likely that the etiology of pulmonary hypertension in this setting is multifactorial. Furthermore, children may be different from adults,⁴⁰  and the clinical implications of an elevated tricuspid regurgitation velocity are largely unknown in the pediatric population. The independent association of erythropoietin with higher tricuspid regurgitation velocity suggests that the safety of high doses of human recombinant erythropoietin in patients with sickle cell disease should be studied further, specifically, whether erythropoietin therapy may increase tricuspid regurgitation velocity even as it increases hemoglobin concentration. Prospective studies of the effect of hydroxyurea therapy on pulmonary artery pressure in patients with sickle cell disease should also be carried out. These trials should indicate whether the degree of hemolysis reduction by hydroxyurea compensates for the drug's effect in increasing hemoglobin F and erythropoietin levels.

We thank the patients and their families who participated in the study. We thank the research coordinators and nurses from all 3 facilities who contributed to this project and Mr Bak Kim for diligence and work throughout the course of this project.

This study was supported in part by grant nos. 2 R25 HL003679-08 and 1 R01 HL079912-02 from the National Heart, Lung, and Blood Institute (NHLBI); by Howard University General Clinical Research Center (GCRC) grant no. 2MOI RR10284-10 from the National Center for Research Resources (NCRR), National Institutes of Health (NIH); and by the intramural research program of the NIH.

Contribution: V.R.G. and O.L.C. participated in study design, data analysis, and writing the paper; A.C., S.R., C.P.M., C.S., D.D., N.D., O.O., G.J.K., and M.T.G. participated in study design, data collection, and writing the paper; M.N. participated in data analysis and writing the paper; and X.N., T.A., and S.N. participated in study design, collecting laboratory data, and writing the paper.

Conflict-of-interest disclosure: V.R.G. has received research grants from Biomarin and TRF Pharma and has received consulting fees from Ikaria. The remaining authors declare no competing financial interests.

Correspondence: Victor R. Gordeuk, Center for Sickle Cell Disease, Howard University, 2041 Georgia Ave NW, Washington, DC 20060; e-mail: vgordeuk@howard.edu.