Effects of Cotrimoxazole on Hematologic Parameters in HIV-Infected Adults in a Community-Based Clinic in Entebbe, Uganda

Abstract

To the Editor: We and others have previously reported reduced mortality and malaria incidence with cotrimoxazole prophylaxis for HIV-1-infected adults in Uganda and elsewhere.1-4 This evidence has led to widespread implementation of cotrimoxazole prophylaxis for HIV-1-infected persons in sub-Saharan Africa. The World Health Organization (WHO) recommends that implementation at the country level should be carefully monitored and that this monitoring should evaluate rates of toxicity, including hematologic toxicity, through the measurement of blood cell counts every 6 months.5 Cotrimoxazole interferes with folate metabolism, and there have been several reports of hematologic abnormalities associated with its use.6 These abnormalities could potentially worsen if cotrimoxazole is used in conjunction with myelosuppressive or antifolate drugs such as zidovudine.7,8 There is scanty evidence of these abnormalities associated with cotrimoxazole use in Uganda, however.2,8 Cotrimoxazole prophylaxis was introduced in August 2000 in an open cohort of HIV-1-infected participants in Entebbe, Uganda. Here, we report hematologic changes in participants before and after the introduction of cotrimoxazole. Since 1995, HIV-1-infected adults in Entebbe have been evaluated every 6 months using a standard protocol for clinical, hematologic, and immunologic parameters, as previously described.9 Hemoglobin level, total white cell blood counts (WCCs), platelets, and CD4 cell counts were measured in the same way throughout the study at scheduled visits. The percentages of lymphocytes (including monocytes) and granulocytes were measured using a simple counter initially; however, from 2002 onward, a Coulter, differential analyzer (Beckman Coulter Inc., Fullerton, CA) was utilized. Definitions of abnormalities were taken from the AIDS Clinical Trials Group (ACTG) studies. The reference time point was taken as the time of enrollment on cotrimoxazole prophylaxis. Participants were excluded from the analysis if they did not have a measurement within 3 months of enrollment and if they did not have at least 1 measurement before and after the enrollment measurement. All measurement data taken between 24 months before and after enrollment were included in the analysis. The number and proportion of participants who experienced hematologic abnormalities were calculated in the 2 years before and after enrollment. Separate analyses were performed stratifying participants into 3 categories according to the CD4 count at enrollment (<200, 200 to 499, and ≥500 cells/μL). To assess the change in each hematologic parameter over time, generalized estimating equations (GEEs) were used with a linear spline at the time of enrollment and were adjusted for age, gender, and CD4 cell counts at enrollment. From the model, the difference in the rate of change and the 95% confidence interval of the difference were calculated for each hematologic parameter. Separate models were then fitted for the 3 CD4 cell count strata, using the mean CD4 cell count for each of the periods before and after enrollment. Between August 2000 and April 2003, 936 (95%) of 982 participants of the Entebbe cohort were enrolled and received cotrimoxazole prophylaxis. Analysis is restricted to 786 participants with hematologic data across all time points, 548 (70%) of whom had data on all hematologic parameters. The median age of the 786 participants was just less than 34 years (range: 18 to 63 years), and 611 (78%) were female. Before and after enrollment in the cotrimoxazole study, there was a drop in most hematologic parameters over time (Table 1). Women had higher platelet counts than men but lower WCCs, granulocytes, and hemoglobin (data not shown); however, the declines in the mean values over time were similar in both genders.TABLE 1: Annual Change in Selected Hematologic Parameters for 786 Subjects Enrolled on Cotrimoxazole Prophylaxis Between August 2, 2000 and April 25, 2003, and With Hematologic Data Before and After EnrollmentBefore enrollment, few participants were categorized as having grade 3 or 4 leukopenia, neutropenia, or thrombocytopenia. There were more participants with all 3 abnormalities after enrollment than before, although the numbers were still small. Comparing the periods before and after the start of cotrimoxazole prophylaxis, there was an increase in the percentage of participants with grade 3 or 4 abnormalities: 2.1% with leukopenia, 2.0% with neutropenia, 2.3% with thrombocytopenia, and 5.4% with anemia. Most of these abnormalities occurred in participants with a CD4 count <200 cells/μL. In the period before starting cotrimoxazole, a CD4 count <200 cells/μL was found in 75% of those with leukopenia, 82% with neutropenia, 57% with thrombocytopenia, and 59% with anemia compared with 81%, 66%, 73%, and 82%, respectively, after starting cotrimoxazole. There was a significant annual decline in all hematologic parameters, except platelets, in the period before introduction of cotrimoxazole. After the introduction of cotrimoxazole, all hematologic parameters continued to show a decline, and the mean rate of decline was significantly greater after the introduction of cotrimoxazole than before for granulocytes (−390 vs. −160 cells/μL; P = 0.002) and platelets (−12,500 vs. −1800 cells/μL; P < 0.001). For granulocytes, the increase in the rate of annual decline was larger in male participants (−480 cells/μL per year) compared with female participants (−160 cells/μL per year), as was the rate of decline of platelets (−7540 cells/μL per year in female participants, −18,940 cells/μL per year in male participants), although the difference between genders was not significant for either parameter. We stratified these results by mean CD4 cell count before and after enrollment (see Table 1). In all 3 strata, there was an annual decline in the hematologic parameters before and after enrollment into cotrimoxazole prophylaxis, and differences were observed between the strata. There was a significantly greater annual decline in WCCs (370 cells/μL; P = 0.001), granulocytes (390 cells/μL; P < 0.001), and platelets (13,600 cells/μL; P = 0.003) in participants with a CD4 count of 500 cells/μL or more after the start of cotrimoxazole. In addition, there was a greater annual decline in CD4 cell counts in this stratum. Granulocytes and platelets in HIV-1-infected adults in Uganda decline over time, but this natural decline is significantly greater after enrollment into cotrimoxazole prophylaxis. Although this is of concern, our data show that few participants are seen with severe or moderate leukopenia or neutropenia. This concern should be weighed against the benefits of cotrimoxazole prophylaxis.1-4 From an open cohort in which everyone was offered cotrimoxazole, we used historical data, where each participant was his or her own control, to compare hematologic changes in the periods before and after cotrimoxazole prophylaxis treatment. We used GEEs to assess the change in hematologic parameters within each participant and to estimate accurately the overall rates of decline before and after the introduction of cotrimoxazole prophylaxis. We included all participants of this open cohort who had hematologic measurements before and after their enrollment in the study. Greater declines in hematologic parameters were seen in the participants with higher CD4 cell counts at enrollment, suggesting that this is not attributable to the effect of terminal illness. Another study in Tororo, Uganda in patients with the same range of CD4 cell counts at enrollment found that the annual rate of decline in CD4 cell counts was less during prophylaxis than before.2 The different results could be partly explained by differences in the 2 study populations. In our study, more than a third of participants had been part of the cohort for more than 2 years before starting cotrimoxazole, whereas in Tororo, only a few months of observation were possible before the start of the intervention. Some selection bias in the populations could also explain the different results, with our study comprising sick people and long-term HIV-positive slow progressors, whereas in Tororo, patients were recruited after attending an HIV clinic, with a median CD4 count of 124 cells/μL, and perhaps had larger declines in CD4 cell counts before receiving cotrimoxazole. Finally, there were differences in the adjustment for CD4 cell counts in the analysis, because we used enrollment CD4 cell counts to define groups at the time the participants started cotrimoxazole, whereas in Tororo, the CD4 cell count groups were defined at the beginning of the period before cotrimoxazole. Although methods for measuring the WCC remained the same, the laboratory method used to assess the differential counts of granulocytes and lymphocytes changed in 2002, and this may have introduced some differences in the calculation of the absolute counts between measurements before and after this date. The results presented here show that cotrimoxazole has affected hematologic parameters in these participants and may consequently lead to higher rates of hematologic problems, given the cumulative nature of these toxicities. Guidelines are needed on how to monitor the clinical signs suggestive of hematologic problems in patients taking long-term cotrimoxazole. ACKNOWLEDGMENTS The cotrimoxazole prophylaxis study was financially supported by the Medical Research Council (MRC) Programme on AIDS in Uganda and other institutions within the Uganda Virus Research Institute (UVRI) Collaboration. The authors thank all the patients for participating in the study and all The AIDS Support Organization (TASO) and MRC/UVRI program staff who worked on this project. Christine Watera, MD Jim Todd, MSc Gertrude Mutonyi, BSc George Miiro, MD Juliet Mpendo, MD Peter Hughes, MSc Jessica Nakiyingi-Miiro, MSc Jimmy Whitworth, MD, PhD Heiner Grosskurth, MD, PhD Medical Research Council Unit on AIDS Uganda Virus Research Institute Entebbe, Uganda