Treatment guidelines from the World Health Organization (WHO) for the use of antiretroviral therapy (ART) in resource-limited settings state that CD4 cell counts may be used to monitor clinical response to therapy in programs in which viral load testing is not available.1 The return of CD4 cell counts to pretherapy levels, declines in the absence of coexistent infections, or a >50% decrease in CD4 cell counts from peak levels without coexistent infections are proposed as definitions of treatment failure whereby a change in an ART regimen may be warranted.1
In industrialized countries, however, viral load has been the primary tool that clinicians and researchers have used to monitor patients on ART to indicate when therapy should be changed.2,3 Immunologic parameters in isolation from viral load measurements are not used to indicate treatment failure. The clinical performance of immunologic responses to ART, without concomitant virologic monitoring, in terms of predicting response to therapy has not been previously evaluated. Therefore, we designed a study to evaluate the clinical utility of changes in CD4 cell count parameters at 6 and 12 months after treatment initiation in terms of their ability to identify subjects who have not achieved virologic suppression on their primary ART regimen using data from the British Columbia HIV/AIDS Drug Treatment Program (BCDTP). Such subjects would be recommended to change their initial ART regimen, based on the current WHO guidelines for resource-limited settings.
The BCDTP has been previously described.4 Participants in the present study were, aged >18 years, were ART naive, and had CD4 cell counts ≤200 cells/μL or an AIDS-defining illness at baseline and started ART between August 1, 1996 and September 30, 2003. Subjects also needed to have CD4 cell counts and CD4 percentage (CD4%) measured 3 to 9 months after treatment initiation or CD4 cell counts at 9 to 15 months after treatment initiation. Included subjects must also have had at least 2 viral load measurements during follow-up. Plasma HIV-1 RNA levels were determined using the Roche Amplicor Monitor sensitive assay before the year 2000 and the ultra-sensitive assay thereafter (Roche Diagnostics, Laval, Quebec, Canada). CD4 cell counts were measured by flow cytometry, followed by fluorescent monoclonal antibody analysis (Beckman Coulter, Inc., Mississauga, Ontario, Canada).
Logistic regression analysis was used to define receiver-operator characteristic (ROC) curves for each of the following measures in terms of their performance in predicting failure to achieve 2 consecutive viral load measurements <500 copies/mL within the first year on the primary ART regimen: change in CD4 cell counts from baseline at 6 months and at 12 months; change in CD4 cell counts from 6 months to 12 months; absolute CD4 cell counts at 6 and 12 months; change in CD4 cell counts as a percentage of baseline CD4 cell counts; and absolute CD4% and change in CD4% at 6 months. The ROC curves were then used to identify the thresholds that maximized the diagnostic accuracy (true-positive results + true-negative results/all subjects) for each measure of immunologic response. In addition, these parameters were examined using current WHO criteria for immunologic monitoring. Data from a subgroup of patients who initiated therapy after January 1, 2000 were used to develop ROC curves for immunologic parameters using a viral load measurement of ≥50 copies/mL at 6 and 12 months after treatment initiation as evidence of treatment failure.
A total of 1125 subjects (325 with AIDS-defining illnesses at baseline and 800 with baseline CD4 counts <200 cells/μL) had available viral load measurements on 2 occasions within 1 year of treatment initiation. The median baseline CD4 cell count was 90 cells/μL (interquartile range [IQR]: 30-150), and the median baseline CD4% was 9% (IQR: 5-14). Viral load suppression (defined as the first episode of achieving a viral load <500 copies/mL on 2 consecutive measurements) occurred in 674 (60%) of subjects in a median of 2.4 months (IQR: 1.4-3.7).
Follow-up CD4 cell counts at 6 months were available for 905 subjects. The median CD4 cell count at 6 months was 180 cells/μL (IQR: 110-300), reflecting a median CD4 cell count increase of 100 cells (IQR: 30-170) from baseline. The CD4% at 6 months was available for 835 subjects. The median CD4% was 14% (IQR: 8-21) at 6 months after treatment initiation, with median increases of 5% (IQR: 2-8). Twelve-month CD4 cell counts were available for 855 subjects with median counts of 220 cells (IQR: 130-340), reflecting median increases of 130 cells (IQR: 50-220) from baseline.
ROC curves for 3 measures of immunologic response for predicting virologic suppression are shown in Figure 1. Of all the measures examined, CD4 cell count changes from 0 to 12 months had the largest area under the curve (0.76), and were thus the best predictor of virologic suppression. ROC curves were also developed for changes in CD4 cell counts at 6 months as a percentage of baseline CD4 counts and for all immunologic measures using 6- and 12-month viral load measurements <50 copies/mL as the outcome measure for 450 and 452 subjects, respectively. All these analyses produced ROCs with much lower accuracy than those presented (data not shown).
Table 1 shows the clinical performance of failure to achieve CD4 cell count increases at 6 and 12 months (WHO criteria) in terms of predicting virologic suppression and derived thresholds for each other measure of immunologic response that corresponds to the best clinical performance in terms of accuracy. Using no increase in CD4 cell counts at 6 months as a definition of treatment failure had an accuracy of 0.71, with a sensitivity of 0.34, specificity of 0.94, positive predictive value of 0.75, and negative predictive value of 0.71. Using no increase in CD4 cell counts at 12 months, the measures were 0.75, 0.35, 0.95, 0.79, and 0.73, respectively. Of the other measures of immunologic response examined, only raising the threshold of CD4 cell count changes at 12 months to ≥40 cells/μL had slightly greater accuracy (0.76) than using no change in CD4 cell counts at 12 months.
This analysis has shown that using immunologic criteria to predict which patients have not achieved virologic suppression results in significant misclassification of therapeutic responses. Health professionals with only CD4 cell count monitoring available to assess ART treatment failure should interpret these values and the WHO monitoring guidelines quite cautiously. The WHO guidelines state that subjects who experience a return of CD4 cell counts to pretherapy levels or declines in CD4 cell counts from baseline, in the absence of coexistent infections, should be considered for ART regimen changes to a second-line regimen. If failure to achieve any increase from baseline CD4 cell counts at 6 months or 12 months were used to define treatment failure in this population, approximately 21% to 25% of subjects would be improperly labeled as failing treatment. Clearly, this would have negative consequences for such patients because they would be prematurely switched off a regimen that was effectively controlling viral replication.
Furthermore, only 34% to 35% of true treatment failures were identified using these criteria. This poor sensitivity may be less of a concern, because subjects with CD4 cell count increases in the absence of virologic suppression may experience CD4 cell count declines later in their course of treatment that would allow them to be identified, albeit at a later date. The clinical consequences of the late identification of these treatment failures are unknown but would allow greater time for viruses in these patients to accumulate multiple drug resistance mutations.5 This may be less of an immediate concern for patients in African countries, where second-line regimens commonly use 3 completely new drugs to which the client has not previously been exposed. Such prolonged exposure to partially suppressive therapy may limit the effectiveness of third- and fourth-line regimens later on, however.
We cannot conclude that immunologic monitoring in resource-limited settings does not have value. We have shown that it poorly predicts those who have not achieved virologic suppression, and thus should not be used in isolation to identify those who are failing treatment. Virologic and immunologic responses are associated with disease progression on ART.6,7 Nevertheless, it is not clear that changing ART regimens in the face of a poor immunologic response is a safe or effective strategy to improve patient outcomes. Failure to achieve 6-month immunologic thresholds in the presence of virologic suppression is associated with poor adherence to therapy.8 Therefore, switching therapy for such patients without addressing possible adherence problems may lead to earlier failure on second- and third-line regimens.
Previous studies have shown that CD4 cell count measurements have significant intraindividual variability.9 Therefore, serial measurements may have more prognostic value than assessments based on single measurements. Alternatively, CD4 cell counts may be better used as a screening tool to define which patients require viral load testing.10 Such a scenario is predicated on viral load testing being accessible to all ART programs, however, which is a situation that currently does not exist. It is possible that additional clinical information such as weight gain, improved physical activity levels, or resolution of specific opportunistic infections could be used in addition to CD4 cell count monitoring improve the positive predictive value. Further research in such settings is needed to determine if immunologic monitoring can be enhanced through such measures.
The criteria used to define treatment failure (not achieving 2 consecutive viral load measurements <500 copies/mL) were relatively conservative, given that current ART guidelines in industrialized countries use failure to achieve viral load measurements of <50 copies mL as evidence of treatment failure.2,3 All immunologic measures assessed against this more stringent definition of failure performed much worse than using 500 copies/mL, however.
The generalizability of this study to resource-limited settings may be limited, because hematologic parameters in African countries can differ markedly from those in Europe or North America.11 In addition, the proportion of study subjects who failed to achieve virologic suppression was higher (40%) than that reported from several recent cohort studies in Africa,12,13 which may affect the clinical performance of these parameters. Clearly, more research is needed in African and other developing country settings to inform ART programs better about how best to monitor clients and what interventions can improve patient outcomes.
The authors thank Jennifer Adachi, Bonnie Devlin, Elizabeth Ferris, Nada Gataric, Kelly Hsu, Myrna Reginaldo, and Peter Vann for their research and administrative assistance.
1. World Health Organization. Scaling up Antiretroviral Therapy in Resource Limited Settings: Guidelines for a Public Health Approach
. Geneva: WHO; 2003.
2. Yeni PG, Hammer SM, Hirsch MS, et al. Treatment for adult HIV infection: 2004 recommendations of the International AIDS Society-USA Panel. JAMA
4. Hogg RS, Yip B, Chan KJ, et al. Rates of disease progression by baseline CD4 cell count and viral load after initiating triple-drug therapy. JAMA
5. Lafeuillade A, Hittinger G, Delbeke E, et al. Resistance selection in patients with stable low levels of HIV-1 viremia. Presented at: XV International AIDS Conference; 2004; Bangkok.
6. Moore DM, Hogg RS, Yip B, et al. Discordant immunologic and virologic responses to highly active antiretroviral therapy are associated with increased mortality and poor adherence to therapy. J Acquir Immune Defic Syndr
7. Chene G, Sterne JA, May M, et al. Prognostic importance of initial response in HIV-1 infected patients starting potent antiretroviral therapy: analysis of prospective studies. Lancet
8. Moore DM, Hogg RS, Chan K, et al. Disease progression in patients with virological suppression in response to HAART is associated with the degree of immunological response. AIDS
9. Raboud JM, Haley L, Montaner JS, et al. Quantification of the variation due to laboratory and physiologic sources in CD4 lymphocyte counts of clinically stable HIV-infected individuals. J Acquir Immune Defic Syndr Hum Retrovirol
. 1995;10(Suppl 2):S67-S73.
10. Bisson G, Gross R, Gaolathe T, et al. Diagnostic characteristics of CD4 cell count response in predicting virologic response in HIV-infected patients initiating HAART in Botswana. Presented at: 13th Conference on Retroviruses and Opportunistic Infections; 2006; Denver.
11. Lugada ES, Mermin J, Kaharuza F, et al. Population-based hematologic and immunologic reference values for a healthy Ugandan population. Clin Diagn Lab Immunol
12. Coetzee D, Hildebrand K, Boulle A, et al. Outcomes after two years of providing antiretroviral treatment in Khayelitsha, South Africa. AIDS
13. Laurent C, Diakhate N, Gueye NF, et al. The Senegalese government's highly active antiretroviral therapy initiative: an 18-month follow-up study. AIDS