Correspondence
Erythropoietin for Zidovudine-Induced Anemia
N Engl J Med 1990; 323:1069-1070October 11, 1990
- Article
To the Editor:
Fischl and coauthors have reported a beneficial effect of recombinant human erythropoietin in the management of zidovudine-induced anemia in patients with the acquired immunodeficiency syndrome (AIDS) (May 24 issue).1 They found that erythropoietin given in a dosage of 300 IU per kilogram of body weight per week decreased transfusion requirements among patients with pretreatment levels of endogenous erythropoietin ≤500 IU per liter. Unfortunately, tolerance of zidovudine was not improved. In fact, in the subset of patients in which benefit was seen, the mean weekly dose of zidovudine in patients receiving erythropoietin had been reduced by the third month of treatment by 25 percent from base line, as compared with only 6 percent in patients receiving placebo. This result not only is disappointing but also casts doubt on the authors' conclusion. The difference between the treatment groups in the mean weekly dose of zidovudine was not found to be statistically significant, but the difference in the dose reduction was not addressed and may have accounted for much of the observed decrease in transfusion requirements. Because the mean daily dose of zidovudine at base line was considerably higher (750 mg) than those now recognized to be effective,2 , 3 many of the patients might have been treated successfully with dose reduction alone. Because the study design did not include a standardized approach to the dosage of zidovudine, the relative roles of erythropoietin therapy and zidovudine dose reduction are unclear.
Our own limited experience with recombinant erythropoietin suggests that the frequency of improved tolerance of zidovudine is low. We administered erythropoietin to five patients in whom severe anemia developed while they were receiving 300 to 500 mg of zidovudine daily and had levels of endogenous erythropoietin ≤500 IU per liter. The starting dose was 24,000 IU per week; it was increased to 36,000 and 48,000 IU per week at six-week intervals. Zidovudine was continued throughout the period of erythropoietin therapy. The highest dose level was reached in four of the five patients. No patient had a hemoglobin level above 10 ml per deciliter, reticulocytosis of ≥1 percent, or any reduction in transfusion requirements as compared with requirements in the three months before treatment. The daily dose of zidovudine could not be increased in any of them.
3 ReferencesDavid H. Shepp, M.D.
Bruce D. Agins, M.D.
Bruce F. Farber, M.D.
North Shore University Hospital, Manhasset, NY 110301
Fischl
Full Text | Web of Science | Medline2
Volberding
Full Text | Web of Science | Medline3
Progress in the use of zidovudine . FDA Drug Bull 1990; 20:6–7.
To the Editor:
We would like to take exception to the implication of Fischl et al. that the increased burden placed on the nation's blood supply by the population with AIDS receiving zidovudine therapy would be substantially reduced by the use of erythropoietin. If the authors' own figures are used, only 48 of their 63 patients (76 percent) would be expected to qualify for erythropoietin treatment on the basis of their proportion of "low" levels (≤500 IU per liter) and "high" levels (>500 IU per liter) of erythropoietin. Moreover, patients with "high" levels have the largest transfusion requirements and are clearly not benefited by erythropoietin treatment. Thus, approximately 30,400 patients with AIDS would meet the entry criteria (40,000 patients × 0.76). Besides, the end-point red-cell transfusion requirement in the group with low erythropoietin levels was still 0.84 unit per month per patient, representing an improvement of 0.47 unit per month per patient over the base-line need (1.31 units per month). This represents a saving of 171,456 units per year (30,400 patients × 0.47 unit per month × 12 months) of the projected 1 million units required.
The usefulness of erythropoietin could in fact be greater if the subgroup represented by the five patients in the group with low erythropoietin levels requiring all the blood transfusions at the end point could be identified beforehand and excluded as nonresponders. In this small but important group, which accounts for approximately one quarter of the patients in the low-erythropoietin category, the transfusion requirement determined according to the figures provided could have increased over the three-month study period, rising from an average of 1.80 units per patient at base line (1.31 units per patient × 22 patients/16 patients at risk) to 3.70 units per patient at the end point (0.84 unit per patient × 22 patients/5 patients at risk). Setting strict criteria for transfusion would eliminate the need for a placebo group in the future and allow patients to serve as their own controls.
Brian M. O'Connor, M.D.
Kaushik A. Shastri, M.D.
Veterans Affairs Medical Center, Buffalo, NY 14215The above letters were referred to the authors of the article in question, who offer the following reply:
To the Editor: The study by Fischl et al. was a preliminary study intended to demonstrate the activity of recombinant human erythropoietin when administered to anemic patients with AIDS treated with zidovudine. Although it is true that the dose of zidovudine decreased in both groups, the difference between them was neither statistically significant nor clinically important. At the end of the study, the dose of zidovudine in the erythropoietin-treated group remained 4336 mg per week — at or above the current recommended dose. In a series of analyses in this and other studies, which used a linear model not described in our article, the effect of transfusions and variations in the dose of zidovudine on the hematocrit was determined. In patients with serum levels of endogenous erythropoietin ≤500 IU per liter, the effect on the hematocrit of a change in zidovudine doses in the range observed in these studies was relatively minor and would not be expected to account for the hematologic benefit observed. We have no reason to believe that the effect reported in the article could have been duplicated by a simple dose reduction alone.
The experience of Shepp et al. with erythropoietin (as an investigational new drug) in a population with human immunodeficiency virus infection stands in distinction to the general experience. On the basis of a preliminary analysis, these patients had a decrease of 60 percent in transfusion requirements and a change in the hematocrit from 28 percent to 36 percent. We did not suggest in our paper that the use of recombinant erythropoietin alone will improve tolerance of zidovudine. Such tolerance goes well beyond hematocrit changes and includes concurrent changes in platelet count, white-cell count, and concomitant infection; erythropoietin therapy is directed at only one component of that effect.
Although not intending to argue with the calculations of O'Connor and Shastri, we believe that they did not fully account for the increase in the transfusion rate in the placebo group over time. At the end of study, the differential between the groups was 1.9 units per patient per month. In both this double-blind experience and subsequent open-label experience, the erythropoietin-treated patients with base-line serum erythropoietin levels ≤500 IU per liter had maintenance or improvement of their hematocrit and a further decrease in their transfusion requirements (with escalation of the dose of erythropoietin) for up to 16 months. Whether 170,000 units, 680,000 units, or 1 million units will be saved depends not on our calculations but on the actual effect on practice of the use of recombinant erythropoietin and the continued refinement of zidovudine administration. We look forward to seeing those results and to providing them for subsequent review.
Robert I. Abels, M.D.
Seth A. Rudnick, M.D.
Huei C. Tsai, Ph.D.
R.W. Johnson Pharmaceutical Research Institute, Raritan, NJ 08869
