An increased haemoglobin A2 percentage is an important diagnostic criterion for the diagnosis of beta thalassemia trait. Other causes of an increased percentage have, until recently, been uncommon, but in 1993 Routy and colleagues  described an increased A2 percentage in HIV-infected patients, some of whom were being treated with zidovudine. They found that 24 out of 78 patients on zidovudine had an A2 above 3.5%, whereas this was seen in none of the 68 patients not on zidovudine . No comparison of HIV-infected and normal individuals was performed. The increased A2 percentage in HIV-infected patients on treatment was confirmed by Howard and colleagues  in 2005. They reported two patients, both on zidovudine, with A2 percentages of 4.3 and 3.9, respectively (normal range 1.5–3.5%), in one of whom beta thalassemia trait was excluded by DNA analysis. In both these reports the increased A2 percentage was attributed to the therapy. Our own observations during routine thalassemia screening suggested that HIV-positive patients who were not on antiretroviral therapy also sometimes had an increased A2 percentage. We therefore carried out a systematic study to determine the frequency of an increase in haemoglobin A2 and to establish to what extent this was related to the HIV infection itself and to what extent to the treatment.
We performed both cross-sectional and cohort studies. In the cross-sectional study we performed a full blood count and measured the haemoglobin A2 percentage in 60 HIV-positive patients, 30 on HAART and 30 not on therapy. Three other HIV-positive patients were excluded because they were found to have sickle cell trait, which increases the measured haemoglobin A2 percentage by the technique we used (two patients) or thyrotoxicosis, which also raises the haemoglobin A2 percentage (one patient). The A2 percentages were compared with those of healthy haematologically normal volunteers using an unpaired t-test, and the proportions of elevated levels were compared using a χ2 test. The A2 percentage was related to haematological variables. A χ2 test was used to study the relationship between elevated haemoglobin A2 and individual drugs.
In the cohort study the same tests were performed on 23 HIV-positive patients immediately before starting on HAART and 4 months or more after starting treatment. The haemoglobin A2 percentages were compared using a paired t-test.
Full blood counts were performed on a Coulter LH700 automated blood cell analyser (Beckman–Coulter, Florida, USA). Haemoglobin A2 was determined by high performance liquid chromatography on a Bio-Rad variant II instrument (Bio-Rad, Hercules, California, USA) using the Beta Thal Short Programme. A reference range for haemoglobin A2 was derived from 40 healthy northern European volunteers with normal red cell indices.
The study was approved by our institutional research ethics committee. All patients were given detailed information as to the purpose of the study and gave written informed consent. Healthy volunteers were fully informed and gave verbal consent.
The normal volunteers were found to have a mean haemoglobin A2 of 2.89% (SD 0.179). HIV-positive patients not on treatment had a mean haemoglobin A2 of 3.09% (SD 0.255). This was significantly different from the values in the normal volunteers on an unpaired t-test (t = −3.711, P < 0.0001). HIV-positive patients on HAART had a mean haemoglobin A2 of 3.36% (SD 0.50; Fig. 1). The haemoglobin A2 percentage was significantly higher than values in normal volunteers (t = −5.449, P < 0.0001) and also values in untreated HIV-positive patients (t = −2.664, P < 0.01). Half of the treated patients were taking zidovudine and half were not. A χ2 test showed a significant association between this drug and an elevated haemoglobin A2 (elevated in 13/15 on zidovudine versus 3/15 not on zidovudine, P < 0.0001). No significant relationship to any other individual antiretroviral agent was found.
In the prospective cohort study, haemoglobin A2 percentages were significantly different from those of normal controls both before and after starting HAART (P < 0.001 pretreatment and P < 0.02 post-treatment). Two patients (two of the three who were taking zidovudine) had an appreciable rise in haemoglobin A2 after commencing HAART, both rising from high in the normal range to clearly above the normal range (3.9 and 4.4%, respectively). This occurred simultaneously with a marked increase in the mean cell volume (MCV), from 83 to 116 fl and from 89 to 120 fl; in the third patient with only a minor change in the haemoglobin A2 percentage (3 to 3.2), the rise in the MCV was less (91 to 101 fl). Two patients had an appreciable fall in haemoglobin A2; otherwise changes were minor. Overall, there was no significant change in the haemoglobin A2 percentage with therapy of 4 months or more duration (mean haemoglobin A2 pretherapy 3.0957, mean haemoglobin A2 post-therapy 3.0870, t = 0.117, P > 0.05).
We have confirmed that HIV infection is associated with a significant increase in the haemoglobin A2 percentage. This difference is seen in both treated and untreated patients. In a cross-sectional study, patients on HAART had a significantly higher haemoglobin A2 percentage than those not on treatment. The higher percentage of haemoglobin A2 in treated patients correlates with zidovudine therapy.
The rise in the haemoglobin A2 percentage in HIV-infected patients is clinically significant. If haematologists are not aware of this effect of HIV infection and its treatment, a misdiagnosis of beta thalassemia trait may occur, the absence of typical red cell indices being attributed to antiviral agents that raise the MCV. Diagnostic confusion is even more likely if the haemoglobinopathy laboratory is not aware of the patient's HIV status. Increasing numbers of HIV-infected women are becoming pregnant, and zidovudine remains an integral part of regimens to prevent mother-to-child transmission. A lack of awareness of the effect of HIV per se and of therapy with zidovudine on haemoglobin A2 may lead to unnecessary investigation and anxiety.
1. Routy JP, Monte M, Beaulieu R, Toma E, St Pierre L, Dumont M. Increase of hemoglobin A2
in human immunodeficiency virus-1-infected patients treated with zidovudine. Am J Hematol 1993; 43:86–90.
2. Howard J, Henthorn JS, Murphy S, Davies SC. Implications of increased haemoglobin A2
levels in HIV-positive women in the antenatal clinic. J Clin Pathol 2005; 58:556–558.