HIV Articles  
CD4 Lymphoctye Percentage versus Absolute CD4 Lymphocyte Count in Predicting HIV Disease Progression: An Old Debate Revisited EDITORIAL  
  The Journal of Infectious Diseases Sept 15 2005;192:945-947
Miguel Goicoechea and Richard Haubrich
Department of Medicine, Division of Infectious Diseases, University of California, San Diego, San Diego
HIV Antiretroviral Treatment: Early Versus Later
(JAIDS Aug 15 2005) Early therapy is more cost-effective when the impact of HAART on well-being is smaller. Conclusions: Initiation of HAART at a CD4+ T-cell count greater than ... ....Starting HAART when CD4s are >350 ......increases years of life by 1.21 years, increases...
Even before the first effective antiretroviral therapy (ART) became available nearly 20 years ago, various staging systems were used to predict HIV disease progression, and, subsequently, these systems have been used to guide initiation of therapy. They are based on measurements of CD4 lymphocytes, including the absolute CD4 lymphocyte count, the percentage of lymphocytes that are CD4 positive (CD4 lymphocyte percentage), and the CD4 : CD8 lymphocyte ratio [1]. In this issue of the Journal of Infectious Diseases, Hulgan et al. [2] present data that suggest that the baseline CD4 lymphocyte percentage may be an additional predictor of disease progression in a subset of individuals who have absolute CD4 lymphocyte counts >350 cells/mm3 but have low CD4 lymphocyte percentages. Although absolute CD4 lymphocyte counts and CD4 lymphocyte percentages provide similar information and are highly correlated, these correlations are not perfect; in one study, the correlation coefficient between the 2 markers was 0.5 [1]. Differences between absolute CD4 lymphocyte count and CD4 lymphocyte percentage may represent a type of immune discordance. A relatively high absolute CD4 lymphocyte count and a low CD4 lymphocyte percentage may occur in 8%–10% of untreated HIV-infected patients [3, 4]. This has potentially important clinical implications. Current guidelines for initiation of therapy for asymptomatic individuals with HIV RNA levels <100,000 copies/mL are based on the absolute CD4 lymphocyte count and recommend initiating ART at a count ⩽350 cells/mm3 [5, 6]. For minimally symptomatic patients, a CD4 lymphocyte count <200 cells/mm3 is used as a guide in resource-limited settings [7]. Given that response to therapy is dependent on disease stage, if a significant number of HIV-infected patients present with this type of discordance, it may partially explain why some individuals who initiate therapy with moderate disease have suboptimal virologic and immunologic responses [8–10].
A common method for obtaining the absolute CD4 lymphocyte count has been to use a calculated product of the CD4 lymphocyte percentage, obtained by flow-cytometric analysis, and the complete white blood cell count and lymphocyte differential, measured by use of an automated hematologic instrument. However, much of the variability of the calculated absolute CD4 lymphocyte counts results from the hematologic instrument and not from the flow-cytometric analysis [11]. Therefore, some investigators have argued that, since the CD4 lymphocyte percentage is directly measured and is less variable over time, it may be a better marker for monitoring patients [12]. Indeed, the study by Hulgan et al. [2] confirms the findings in earlier reports that, in some patients, the CD4 lymphocyte percentage may be a more accurate predictor of HIV disease progression than the absolute CD4 lymphocyte count [1, 13]. In contrast, Gebo et al. used a direct cytometry method to measure absolute CD4 lymphocyte counts, and CD4 lymphocyte percentage was not shown to be a better marker of symptomatic HIV disease progression [3]. Hulgan et al. suggest that the difference in findings between their study and the one by Gebo et al. may be due to dissimilar patient populations; Gebo et al. studied patients with advanced disease, whereas Hulgan et al. studied patients with CD4 lymphocyte counts >350 cells/mm3. An alternative explanation is that an increase in the precision of CD4 lymphocyte measurement (by use of a single-platform assay) reduced the variability of the absolute CD4 lymphocyte count and improved its reliability as a surrogate marker, compared with those of conventional methods.
The central finding in the study by Hulgan et al. was based on individuals with discordance between a high absolute CD4 lymphocyte count and an unexpectedly low CD4 lymphocyte percentage. Only patients whose absolute CD4 lymphocyte counts were >350 cells/mm3 and whose CD4 lymphocyte percentages were <17% had a greater risk of disease progression (hazard ratio [HR], 3.57; P = .045), compared with subjects with higher CD4 lymphocyte percentages and similar absolute CD4 lymphocyte counts. The authors chose a cutoff point of 17%, since it was the median CD4 lymphocyte percentage of all subjects enrolled in the study. Although biologically significant definitions of absolute CD4 lymphocyte count–CD4 lymphocyte percentage discordance are unknown, previous investigations have defined discordance as an absolute CD4 lymphocyte count ⩾200 cells/mm3 with a CD4 lymphocyte percentage <14% [3, 4]. The study by Hulgan et al. was not designed to evaluate discordance rates, nor could it define an appropriate discordance cutoff point for CD4 lymphocyte percentage. The study noted that 16% of the 223 subjects with an absolute CD4 lymphocyte count >350 cells/mm3 had a CD4 lymphocyte percentage <20%.
Given the controversies about when to start therapy, additional surrogate markers that could inform this decision would be welcome. The present study suggests that the CD4 lymphocyte percentage might be such a marker. However, some limitations of the current study need to be addressed before both absolute CD4 lymphocyte count and CD4 lymphocyte percentage can be included in evidence-based guidelines about timing of initiation of ART. The interpretation of the current findings is limited by the borderline significance of the finding; the P value for the analysis of progression to AIDS in the subgroup of patients with absolute CD4 lymphocyte counts >350 cells/mm3 was .045, a finding largely driven by a small number of events in the group of patients with absolute CD4 lymphocyte counts >350 cells/mm3 and CD4 lymphocyte percentages <17%. Careful evaluation of figure 2B suggests that the statistical difference was due to late-occurring events; a shift in 1 or 2 events could have rendered the results nonsignificant. The numbers of patients in the absolute CD4 lymphocyte count–CD4 lymphocyte percentage discordance group (i.e., CD4 lymphocyte percentages <17%) were small (<36). Larger studies with more patients in this discordant subgroup of interest are needed. Further, given the small numbers in the discordant group, analyses to define the CD4 lymphocyte percentage cutoff point that optimally separates patients with high risk of disease progression from those with absolute CD4 lymphocyte counts >350 cells/mm3 cannot be done on this data set.
Other factors may limit the ability to generalize these results. Since the response to therapy may be dependent on the immunological status of the patient at the time of initiation of therapy, future work should account for treatment-induced responses to therapy. The present study only required the patient to receive treatment for 30 days. Other potential confounding factors included differences in the duration of follow-up for the different CD4 lymphocyte count strata. Future work should also include full multivariate models that account for both absolute CD4 lymphocyte counts and CD4 lymphocyte percentages in the same model. Analytical challenges of such modeling (i.e., regression colinearity) may be one explanation for the omission of such models in the current study. Additionally, even though HIV RNA level was not a significant predictor of progression in the groups with higher CD4 lymphocyte counts in this study, given the proven predictive value of HIV RNA level in prior studies, future research should include this variable in multivariate models, to determine whether the HRs change with inclusion of HIV RNA (even given nonsignificant P values for that covariate in the particular model).
The current analyses considered a relatively small subgroup of patients with discordant CD4 lymphocyte counts and percentages (<16% of the population had a CD4 lymphocyte count >350 cells/mm3). Further research into how often this occurs, what is the optimal CD4 lymphocyte percentage cutoff point for defining risk of progression, and other predictive clinical factors (sex, age, prior conditions) is needed to define patient populations with absolute CD4 lymphocyte counts >350 cells/mm3 who would benefit from earlier initiation of therapy. The work by Hulgan et al. serves as a positive stimulus to further evaluate the role that CD4 lymphocyte percentage plays in defining the timing of initiation of treatment for ART-naive individuals. The decreased variability of the CD4 lymphocyte percentage intuitively makes this marker a more appealing predictor than absolute CD4 lymphocyte count, but current data are insufficient to recommend its routine use. Further investigation should also focus on the use of single-platform methods and whether such methods reduce variability of absolute CD4 lymphocyte count measurements.
  icon paper stack View Older Articles   Back to Top