Background: Laboratory-based monitoring of antiretroviral therapy is essential but adds a significant cost to HIV care. The World Health Organization 2006 guidelines support the use of CD4 lymphocyte count (CD4) to define treatment failure in resource-limited settings.
Methods: We compared CD4 obtained on replicate samples from 497 HIV-positive Ugandans (before and during ART) followed for 18 months by 2 manual bead-based assays, Dynabeads (Dynal Biotech), and Cytospheres (Beckman Coulter) with those generated by flow cytometry at the Infectious Diseases Institute in Kampala, Uganda.
Results: We tested 1671 samples (123 before ART) with Dynabeads and 1444 samples (91 before ART) with Cytospheres. Mean CD4 was 231 cells/mm (SD, 139) and 239 cells/mm (SD, 140) by Dynabeads and flow cytometry, respectively. Mean CD4 was 186 cells/mm (SD, 101) and 242 cells/mm (SD, 136) by Cytospheres and flow cytometry, respectively. The mean difference in CD4 count by flow cytometry versus Dynabeads were 8.8 cells/mm (SD, 76.0) and versus Cytospheres were 56.8 cells/mm (SD, 85.8). The limits of agreement were -140.9 to 158.4 cells/mm for Dynabeads and -112.2 to 225.8 cells/mm for Cytospheres. Linear regression analysis showed higher correlation between flow cytometry and Dynabeads (r=0.85, r=0.73, slope=0.85, intercept=28) compared with the correlation between flow cytometry and Cytospheres (r=0.78, r=0.60, slope=0.58, intercept=45). Area under the receiver operating characteristics curve to predict CD4<200 cells/mm was 0.928 for Dynabeads and 0.886 for Cytospheres.
Conclusion: Although Dynabeads and Cytospheres both underestimated CD4 lymphocyte count compared with flow cytometry, in resource-limited settings with low daily throughput, manual bead-based assays may provide a less expensive alternative to flow cytometry.