Skip Navigation

Circulating Autoantibodies to Erythropoietin Are Associated with Human Immunodeficiency Virus Type 1—Related Anemia

  1. Nikolaos V. Sipsas1,
  2. Styliani I. Kokori1,
  3. John P. A. Ioannidis3,4,
  4. Despina Kyriaki2,
  5. Athanasios G. Tzioufas1 and
  6. Theodore Kordossis1
  1. 1Department of Pathophysiology, Laikon General Hospital and School of Medicine, National and Kapodistrian University of Athens, Athens
  2. 2Department of Nuclear Medicine, Laikon General Hospital and School of Medicine, National and Kapodistrian University of Athens, Athens
  3. 3Department of Hygiene and Epidemiology, University of Ioannina School of Medicine, Ioannina, Greece
  4. 4Department of Medicine, Tufts University School of Medicine, Boston, Massachusetts
  1. Reprints or correspondence: Dr. Theodore Kordossis, Dept. of Pathophysiology, Athens University Medical School, 75 Mikras Asias St., Athens, GR-115 27, Greece (nsipsas{at}otenet.gr).
 
Next Section

Abstract

In a cohort of 204 unselected consecutive human immunodeficiency virus type 1 (HIV-1)—infected patients, the association of circulating autoantibodies to endogenous erythropoietin (EPO) with HIV-1—related anemia was studied. Circulating autoantibodies to EPO were present in 48 (23.5%) of the 204 patients studied. Circulating autoantibodies were an independent predictor of anemia (odds ratio [OR] = 5.0; 95% confidence interval [CI], 2.5–9.9), as strong as other known causes of anemia. The association of anti-EPO antibodies with anemia became stronger when the analysis was limited to the group of patients without any medical condition causing anemia (OR = 10.4; 95% CI, 3.2–33.9). Moreover, the effect on hemoglobin levels remained significant even after adjusting for other anemia parameters. Anti-EPO autoantibodies were associated with higher EPO levels (r = .25, P = .012) and with a more prominent EPO response to anemia. Our findings suggest that autoimmunity, among other factors, may contribute to the pathogenesis of HIV-1—related anemia.

Anemia is a common complication of human immunodeficiency virus type 1 (HIV-1) infection and is associated with an increased risk of death. The overall prevalence of HIV-1—related anemia increases as the disease progresses, ranging from 10% among asymptomatic persons to 80% in AIDS patients [1]. HIV-1—related anemia has the characteristics of the “anemia of chronic disease” [2] and is associated with an impaired erythropoietin (EPO) feedback [3], possibly caused by proinflammatory cytokines [4]. Although the pathophysiology of HIV-1—related anemia is not well understood, there are many contributing factors, including the direct effect of HIV-1 on bone marrow cells, adverse reactions to antiretroviral drugs, opportunistic infections and neoplasms infiltrating the bone marrow, vitamin B12 or iron deficiency, autoimmune hemolytic anemia, and other coexisting medical conditions [2].

There are no data on whether autoimmunity may contribute to HIV-1—related anemia. HIV-1 infection is characterized by a plethora of autoimmune phenomena caused by the extreme immune activation. A high prevalence of circulating autoantibodies against a constellation of antigens has been reported in HIV-1—infected patients [5–8], but their clinical significance is unclear [9]. In a recent report, 2 AIDS patients with anemia caused by pure red cell aplasia (PRCA) responded favorably to prednisone treatment, suggesting the presence of an autoimmune mechanism [10]. Two subsequent studies reported the presence of serum autoantibodies to EPO in a patient with PRCA [11] and in 15% of patients with systemic lupus erythematosus (SLE), respectively [12]. Both studies suggested that the presence of anti-EPO autoantibodies was associated with anemia. Autoantibodies against EPO have not been determined in HIV-1—infected patients. Their prevalence and clinical significance in HIV-1—infected patients not treated with EPO is unknown.

This study was undertaken to determine the prevalence of circulating anti-EPO autoantibodies in HIV-1—infected patients and investigate whether their presence is associated with HIV-1—related anemia.

Previous SectionNext Section

Patients and Methods

Study population

Sera from 204 unselected consecutive HIV-1—infected patients (180 men and 24 women; mean, age ± SD 35.24 ± 10.80 years; range, 17–80 years; median CD4 count, 268 cells/mL; interquartile range, 62–522) were used in the study. None of the patients was treated with EPO before the time of sampling. All patients were followed up with clinic visits at 4-month intervals. Blood samples were obtained at each patient visit for flow cytometric measurements of CD4+ T lymphocytes and other laboratory studies. Hematologic studies, performed by use of standard methods described elsewhere, included hematocrit and hemoglobin (Hb) levels, white blood cell count and differential, platelet numbers, review of peripheral blood smear, and determination of iron, ferritin, vitamin B12, and folic acid. EPO serum levels were measured by radioimmunoblot assay with a commercial kit (Incstar, Stillwater, MN). Anemia was defined as Hb level <12 g/dL. In the presence of anemia, further laboratory evaluation, consisting of bone marrow examination (aspiration, biopsy), hemoglobin electrophoresis, Coombs test, examination of a stool specimen for occult blood, bilirubin levels, and reticulocyte count, was performed as needed. Serum from each blood sample was separated, aliquoted, frozen, and stored at −80°C. The following information, corresponding to the sampling time point, was recorded for each patient: age, sex, risk group, CDC stage, CD4 count, antiretroviral treatment, concurrent opportunistic infection or neoplasm, and presence of fever of >1 month's duration or other anemia-associated medical condition.

Detection of anti-EPO autoantibodies

Recombinant human EPO (rHuEPO), purified by analytical gel filtration and characterized by amino acid composition and NH2 terminal analysis (Cilag AG, Switzerland), was used as antigen. Sera from patients with SLE, containing anti-EPO autoantibodies as defined elsewhere [12], were used as positive controls. Sera from 40 healthy blood donors were used as normal controls. Autoantibodies to EPO were detected by use of the ELISA technique described elsewhere [12]. Briefly, 96-well polystyrene plates were coated with 10 µg of r-HuEPO in PBS, pH 7.2. Optimum blocking conditions for nonspecific binding were achieved by adding to each well 100 µL of 5% bovine serum albumin (BSA)—Tris-NaCl, pH 7.2, and incubating at 4°C overnight. After washing 3 times with PBS, the samples were added in duplicate, at a 1 : 25 dilution, in PBS containing 2% BSA and 0.2% Tween 20. After 1 h of incubation at room temperature (RT), the plates were washed 5 times with PBS and subsequently incubated with goat anti-human IgG conjugated with alkaline phosphate, at a 1 : 2000 dilution for 1 h at RT. The substrate buffer (p-nitrophenyl phosphate, 2 mg/mL [Sigma, St. Louis]) was added, and, after 30 min of incubation at 37°C, the final reaction was read at 405 nm using a Dynatech (London, UK) UR 4000 ELISA reader. The cutoff point for positive samples was calculated as the mean optical density for the normal controls plus 3 SD. Only 5 patients were between 2.58 and 3 SD. Intra- and interassay coefficients of variation were <8% and <15%, respectively. The specificity of the method has been evaluated with homologous and cross inhibition assays in previous experiments in our laboratory [12].

Statistical methods

Logistic regressions used the presence of anemia as the dependent variable, while linear least squares regressions used the levels of hemoglobin and the log10 of EPO as the dependent variables. Multivariate regression models were built with backward elimination of variables, according to standard criteria of P> .10 for variable removal and P < .05 for variable entry. Backward elimination on the logistic models was based on the likelihood ratio. All P values are two-tailed. All analyses were performed in Advanced SPSS v. 6.1 (SPSS, Chicago).

Previous SectionNext Section

Results

Patient characteristics

Anemia was present in 73 of the 204 study patients. As expected, the prevalence of anemia was much higher among the 108 patients with AIDS than among the 96 patients without AIDS (59.2% and 9.4%, respectively). Of the 204 patients included in the study, 101 had at least 1 other medical condition causing anemia, and 57 of these had anemia. Such medical conditions included zidovudine treatment (n = 69; n = 38 with anemia), neoplastic disease (n =17; n = 13 with anemia), opportunistic infection (n = 26; n = 18 with anemia), fever of at least 1 month's duration (n = 27; n = 20 with anemia), and miscellaneous causes such as thalassemia, renal failure, or a ferritin level <10 ng/mL, suggestive of iron deficiency anemia (n = 14; n = 9 with anemia). No patient had significant hemolysis.

Relationship of anti-EPO autoantibodies with anemia in the total study population

Of 73 patients with anemia, 31 had anti-EPO autoantibodies, as compared with only 17 of 131 patients without anemia (odds ratio [OR] = 5.0; 95% confidence interval [CI], 2.5–9.9). In multivariate logistic regression analyses (table 1), the presence of anti-EPO autoantibodies was associated with an OR of 7.8 for the presence of anemia, while other independent predictors of anemia included the CD4 cell count, a clinical AIDS diagnosis, fever of at least 1 month duration, and miscellaneous causes of anemia (thalassemia, low ferritin, renal failure). Several other variables associated with anemia in univariate regressions were not significant as independent predictors in the multivariate regression.

Figure 1
View larger version:
Figure 1

Levels of erythropoietin (EPO) as a function of hemoglobin levels in patients with anti-EPO antibodies (■) and in patients without such antibodies (▽). Only the 103 patients without other potential causes of anemia are included in this graph. Also shown are regression lines for logarithm of EPO as a function of hemoglobin levels (thick line for group with anti-EPO antibodies, thin line for group without antibodies). Regression lines were similar to those shown when analyses were further limited to include only patients with anemia.

Table 1
View larger version:
Table 1

Logistic regression analyses for predictors of anemia (hemoglobin <12 g/dL).

Patients with anti-EPO autoantibodies had, on average, a hemoglobin level that was 1.15 g/dL lower than the hemoglobin level of patients without anti-EPO (P = .003). In multivariate linear regression, the anti-EPO effect remained significant (−0.62 g/dL; P = .015), while the hemoglobin was also higher in male patients and decreased in patients with lower CD4 cell counts, a clinical AIDS diagnosis, increasing age, and in the presence of neoplasia, fever of at least 1 month's duration, or miscellaneous causes of anemia. In the same multivariate regression, a 10-mU higher EPO was associated with a significant decrease of 0.08 g/dL in the hemoglobin level (P < .0001).

Relationship of anti-EPO autoantibodies with anemia in patients without other medical conditions resulting in anemia

When the analysis was limited to 103 patients who did not have any other known medical condition that could cause anemia, the association of anti-EPO autoantibodies with anemia became even stronger. Ten of 16 patients with anemia of unknown cause had anti-EPO autoantibodies, while only 12 of 87 patients without anemia had such autoantibodies. The OR was 10.4 (95% CI, 3.2–33.9). In multivariate regression (table 1), the only significant, independent predictors of anemia were the presence of anti-EPO autoantibodies (OR = 11.5; P = .0011) and a lower CD4 cell count (P = .001).

The hemoglobin level was, on average, 1.46 g/dL lower in patients with anti-EPO autoantibodies, compared with patients without them (P = .0006). The effect of anti-EPO autoantibodies remained significant (0.87 g/dL, P = .017) after adjusting for other significant predictors of hemoglobin levels (sex, age, clinical AIDS diagnosis, CD4 cell count, and EPO levels).

Anti-EPO autoantibodies and EPO levels at different levels of hemoglobin

The presence of anti-EPO antibodies was also correlated with higher levels of EPO (correlation coefficient, 0.25; P = .012). As shown in figure 1, patients with anti-EPO autoantibodies not only had lower hemoglobin levels, but their EPO response to the presence of anemia tended to be more prominent than it was among patients without anti-EPO autoantibodies. Regression analysis showed that among patients without anti-EPO autoantibodies, the levels of EPO increased 1.09-fold (SD, 1.04; P = .04) per each decrease of 1 g/dL in the hemoglobin level. Among patients with anti-EPO autoantibodies, the levels of EPO increased by 1.34-fold (SD, 1.07; P = .0002) per each decrease of 1 g/dL in hemoglobin. The 2 slopes were significantly different (figure 1). The regression parameters estimated a similar EPO level (∼21 mU/mL) in the 2 patient groups at a hemoglobin of 13.5 g/dL; on the contrary, at a hemoglobin level of 9 g/dL, the EPO level was expected to be 31 mU/mL versus 80 mU/ml in patients without versus with anti-EPO autoantibodies, respectively.

Previous SectionNext Section

Discussion

We have shown that serum autoantibodies to EPO were present in 48 (23.5%) of the 204 patients of the study population. We observed a significant association between the occurrence of anti-EPO autoantibodies and anemia. Circulating anti-EPO autoantibodies were an independent predictor of anemia, as strong as other known causes of anemia and associated with lower hemoglobin and higher EPO levels. The association of anti-EPO autoantibodies with anemia became stronger when the analysis was limited to the group of patients without any other medical condition explaining their anemia. The effect on hemoglobin levels was independent of other parameters associated with anemia.

Autoimmunity is a feature of HIV-1 infection, and many circulating autoantibodies have been detected in HIV-1—infected patients. The relatively high percentage of anti-EPO positive patients in our study was not surprising, taking into consideration that HIV-1—associated autoimmunity is characterized by a high prevalence of circulating autoantibodies, ranging from 4% for non—organ-specific autoantibodies [7] to 47% for histone H2B autoantibodies [8]. The strength of the clinical association of anti-EPO autoantibodies is in sharp contrast with the lack of association between various circulating autoantibodies and clinical signs of autoimmunity reported by other studies [9].

Our study does not provide direct evidence that autoantibodies against EPO have a detrimental effect on erythropoiesis in HIV patients. The retrospective design does not allow us to elucidate whether anti-EPO autoantibodies preceded the development of anemia in these patients or were simply a concomitant marker of the disease process. However, the strength of the association and its independence from other predictors suggests that an etiologic role is possible. A longitudinal cohort study is under way.

Anti-EPO autoantibodies were also associated with higher EPO levels. Decreased hemoglobin levels are the stimulus for increased renal EPO synthesis [13]. In several anemias of chronic disease, including HIV infection, EPO response is inadequate for the levels of anemia [13, 14]. There is limited evidence about the role of anti-EPO antibodies in non-HIV—related chronic anemias, with the exception of SLE, where anti-EPO antibodies have been frequently detected and associated with the presence of anemia [12]. In our study, EPO response to anemia was more prominent in the presence of anti-EPO autoantibodies. Although this response is typically blunted in HIV-1 infection, there have been reports of anemic HIV-1 patients with high levels of EPO [14]. It is unknown whether autoantibodies might be a response to increased EPO levels in an immunologically distorted milieu. In this case, autoantibodies might be an indirect marker rather than a pathogenetic factor of the anemia. Alternatively, a high EPO level with a failure to carry out its biologic action may be caused by a neutralizing effect of anti-EPO autoantibodies.

Previous SectionNext Section

Acknowledgments

We thank Dr. H. M. Moutsopoulos for his comments and suggestions; Cilag A.G., Switzerland, for the generous donation of purified human recombinant EPO; and Ms. Stefania Molangeli for excellent technical assistance.

  • Received April 25, 1999.
  • Revision received July 26, 1999.
Previous Section
 

References

  1. 1.↵
    1. Sullivan PS,
    2. Hanson DL,
    3. Chu SY,
    4. Jones JL,
    5. Ward JW
    ; the Adult/Adolescent Spectrum of Disease Group. Epidemiology of anemia in human immunodeficiency virus (HIV) -infected persons: results from a multistate adult and adolescent spectrum of HIV disease surveillance project. Blood 1998;91:301-8.
    Abstract/FREE Full Text
  2. 2.↵
    1. Kreuzer KA,
    2. Rockstroh JK
    . Pathogenesis and pathophysiology of anemia in HIV infection. Ann Hematol 1997;75:179-87.
    CrossRefMedlineWeb of Science
  3. 3.↵
    1. Spivac JL,
    2. Barnes DC,
    3. Fuchs E,
    4. Quinn TA
    . Serum immunoreactive eryth-ropoietin in HIV-1 infected patients. JAMA 1989;261:3104-7.
    Abstract/FREE Full Text
  4. 4.↵
    1. Kreuzer KA,
    2. Rockstroh JK,
    3. Jelkmann W,
    4. Theisen A,
    5. Spengler U,
    6. Sauerbruch T
    . Inadequate erythropoietin response to anemia in HIV patients: relationship to serum levels of tumor necrosis factor-alpha, interleukin 6 and their soluble receptors. Br J Haematol 1997;96:235-9.
    CrossRefMedlineWeb of Science
  5. 5.↵
    1. Dimitrakopoulos AN,
    2. Kordossis T,
    3. Hatzakis A,
    4. Moutsopoulos HM
    . Mixed cryoglobulinemia in HIV-1 infection: the role of HIV-1. Ann Intern Med 1999;130:226-30.
    Abstract/FREE Full Text
  6. 6.
    1. Montero A,
    2. Giovannoni AG,
    3. Fernandez MA,
    4. Pons-Estel B,
    5. Sen L
    . Autoantibodies in human immunodeficiency virus-infected patients with and without concurrent hepatitis C infection. Arthritis Rheum 1998;41:2077-9.
    CrossRefMedlineWeb of Science
  7. 7.↵
    1. Kordossis T,
    2. Piha L,
    3. Guialis A,
    4. Kosmopoulou O,
    5. Tzioufas AG,
    6. Moutsopoulos HM
    . Non-organ specific autoantibodies against cellular components in HIV-1 infected individuals. Clin Exp Rheumatol 1997;5:185-8.
  8. 8.↵
    1. Williams WM,
    2. Whalley AS,
    3. Comacchio RM,
    4. et al
    . Correlation between expression of antibodies to histone H2B and clinical activity in HIV-infected individuals. Clin Exp Immunol 1996;104:18-24.
    CrossRefMedlineWeb of Science
  9. 9.↵
    1. Massabki PS,
    2. Accetturi C,
    3. Nishie IA,
    4. da Silva NP,
    5. Sato EI,
    6. Andrade LEC
    . Clinical implications of autoantibodies in HIV infection. AIDS 1997;11:1845-50.
    MedlineWeb of Science
  10. 10.↵
    1. Majluf-Cruz A,
    2. Luna-Castanos G,
    3. Nieto-Cisneros L
    . AIDS-related pure red cell aplasia. Am J Hematol 1996;51:171.
    CrossRefMedlineWeb of Science
  11. 11.↵
    1. Casadevall N,
    2. Dupuy E,
    3. Molho-Sabatier P,
    4. Tobelem G,
    5. Varet B,
    6. Mayeux P
    . Autoantibodies against erythropoietin in a patient with pure red cell aplasia. N Engl J Med 1996;334:630-3.
    CrossRefMedlineWeb of Science
  12. 12.↵
    1. Tzioufas AG,
    2. Kokori SI,
    3. Petrovas CI,
    4. Moutsopoulos HM
    . Autoantibodies to human recombinant erythropoietin in patients with systemic lupus erythematosus. Arthritis Rheum 1997;40:2212-6.
    MedlineWeb of Science
  13. 13.↵
    1. Barosi G
    . Inadequate erythropoietin response to anemia: definition and clin-ical relevance. Ann Hematol 1994;68:215-23.
    CrossRefMedlineWeb of Science
  14. 14.↵
    1. Fischl M,
    2. Galpin JE,
    3. Levine JD,
    4. et al
    . Recombinant human erythropoietin for patients with AIDS treated with zidovudine. N Engl J Med 1990;322:1488-93.
    MedlineWeb of Science
| Table of Contents

This Article

  1. J Infect Dis. (1999) 180 (6): 2044-2047. doi: 10.1086/315156
  1. AbstractFree
  2. » Full Text (HTML)Free
  3. Full Text (PDF)Free

Classifications

Share

  1. Email this article

Navigate This Article

Current Issue

  1. March 1, 2012 205 (5)
  1. Journal of Infectious Diseases
  1. Alert me to new issues

Most