In the year 2003, the Ministry of Health in Burkina Faso introduced a program for the prevention of mother-to-child transmission of the human immunodeficiency virus (HIV) using voluntary HIV-counseling and testing and single-dose nevirapine in a remote rural area in the north-west of the country. The introduction and subsequent extension of that service was scientifically evaluated in collaboration between the Centre de la Recherche en Santé de Nouna (CRSN) and the University of Heidelberg, Germany. This led to determination of the HIV prevalence in the region to be 3.6% and identified HIV-1 circulating recombinant forms CRF02_AG and CRF06_cpx as predominant HIV strains [1]. Normal values for CD4⁺ T-cells in healthy adults in Nouna were assessed [2] and a simplified dual platform flow cytometric method for determination of lymphocyte subsets and T-cell maturation phenotypes was evaluated [3]. In that context it was observed that HIV-1-uninfected healthy adult controls living in Nouna had a skewed distribution pattern of CD4⁺ T-cell maturation phenotypes compared to published data from European adults [4].

When highly active antiretroviral therapy (HAART) was introduced in Burkina Faso with the help of the Global Fund, Nouna was chosen as the first rural site where this treatment was made available. We therefore conducted an observational cohort study of all patients in Nouna starting HAART in 2009 to clarify the following questions: (a) Do HIV-1-infected patients with higher baseline proportions of naive CD4⁺ T-cells and/or lower levels of T-cell activation show a more rapid or more extensive recovery of CD4⁺ T-cells under HAART and can these parameters be used as predictors of successful immune recovery? (b) Is the suppression of viral replication by HAART predicted or accompanied by reduced T-cell activation at baseline or during therapy and can these parameters be used as predictors or indicators of successful virologic treatment response?

The study was approved by the national ethics committee in Ouagadougou, Burkina Faso and the institutional ethics committee of the University of Heidelberg, Germany and was carried out in compliance with the Helsinki Declaration. Sixty-one treatment-naïve adult patients were recruited at the CRSN from January 2009 and followed until September 2010. All patients signed an informed consent form prior to study entry. Patients were admitted in the study if their CD4⁺ T-cell counts and/or their clinical status were indicative of the necessity to start HAART [5]. Patients were then seen at 2 weeks and 1, 3, 6, 9 and 12 months after starting HAART. Clinical, immunological, and virological assessments were done at each time point (only CD4⁺ T-cell counting at week 2 was omitted) using three to five milliliters of K₃ -EDTA-anticoagulated peripheral venous blood collected from each patient. Blood samples were transported to the fully climatized laboratory facilities at the CRSN. One specimen was used for immediate CD4⁺ and CD8⁺ T-cell counting performed on a FACSCount flow cytometer (BD Biosciences, San Jose, CA). Plasma was obtained from the second specimen by short centrifugation at 3000 rpm, frozen at -40°C and subsequently used for the quantification of viral RNA using the m2000rt RealTime HIV-1 test (the m2000rt test; Abbott, Chicago, IL) at the Centre Hospitalier Universitaire Yalgado Ouedraogo (CHUYO), Ouagadougou, Burkina Faso according to manufacturers’ protocols. The third specimen was used for multiparametric flow cytometry (FCM) on a three-color FACScan flow cytometer (BD Biosciences) using the CellQuest-Pro software (version 5.2). The instrument setting was regularly done using CaliBrite beads (BD Biosciences). Instrument settings were controlled longitudinally and remained constant during the study. A minimum of 10,000 events were analysed in the predefined gate for each analysis.

The distribution of T-cell maturation phenotypes and the expression of activation markers was assessed in peripheral blood T-cells from HIV-1-infected patients and control subjects (n=26) from healthy laboratory and health service personnel in Nouna. Monoclonal antibodies (BD Biosciences) specific for each assay were aliquoted in 3 ml BD FACS tubes and 100 μl anticoagulated venous blood were added. The percentages of CD4⁺ and CD8⁺ naïve and memory cell subsets were counted after incubating fresh blood with anti-CD4- or anti-CD8-PerCP, anti-CD45RA-FITC and anti-CCR7–PE-conjugated monoclonal antibodies. Naive CD4⁺ or CD8⁺ lymphocytes were identified as CD45RA⁺CCR7⁺ cells [3]. T-cell activation was determined with anti-CD95-PE, anti-CD3-PerCP and anti-CD4-FITC (expression of CD95 on CD4⁺ and CD4^neg T-cells) and anti-CD38-PE, anti-CD3-PerCP and anti-CD8-FITC (expression of CD38 on CD8⁺ and CD8^neg T-cells) essentially as described [6, 7] with minor modifications. Both the percentage of marker-positive cells (compared to unstained lymphocytes of each patient) and the median fluorescence intensity (MFI) obtained from FL-2 fluorescence intensity histogram curves of defined subpopulations of cells were assessed. CD38 MFI was measured on CD38⁺ (instead of total) T-cells in order to focus our analysis on the CD38^bright CD8⁺ T-cell population [8]. The two subpopulations of CD95^dim and CD95^bright CD4⁺ T-cells were separately analyzed since we had previously shown that CD95^dim cells essentially correspond to the resting/naïve subpopulation of CD4⁺ T-cells, whereas CD95^bright cells are of an activated/memory phenotype [6]. The absolute counts of naive and activated T-cells were calculated in HIV-1-infected patients by multiplying the percent values obtained on the FACScan flow cytometer with absolute CD4⁺ and CD8⁺ T-cell counts obtained on the FACSCount flow cytometer.

Unless otherwise indicated, data are given as median values and 10^th to 90^th percentiles. For mean differences between groups, the 95% confidence interval (CI) is shown. Graphical representations (box-and-whisker as well as dot plots) were used to describe the development of immunological parameters over time. Subjects with an increase of at least 100 CD4⁺ T-cells/µl above baseline after 12 months of therapy were classified as good immunological responders. Patients with a HIV-1 plasma viral load below detection limit (< 40 copies/ml) were defined as good virological responders [9, 10]. The statistical significance of differences between time points as well as between HIV-1-infected patients and healthy controls or between patient groups stratified according to immunological or virological response to treatment were analyzed using the non-parametric Kruskal-Wallis test with Dunn's correction for multiple comparisons or the Mann-Whitney U-test. Correlations between variables were explored by linear regression analysis using Pearson's correlation coefficients (r) and the coefficients of determination (R²) in patients with complete data sets at baseline, 6 and 12 month of HAART. All calculations were performed using the GraphPad Prism software, version 5.00 (Graph Pad Inc., San Diego, California, USA).

In our study we show that even under the environmental conditions in rural Burkina Faso CD4⁺ T-cell reconstitution following HAART is a biphasic phenomenon. It is characterized by a first phase mainly constituted of memory cells and a second phase during which naive T-cells progressively appear in the peripheral blood [11, 12]. CD4⁺ T-cell reconstitution after 12 months of therapy was similar to the observations from the CIPRA-SA cohort in South Africa, for which an increase from a mean (± standard error of the mean) of 186±8 cells/µl at baseline to 334±18 cells/µl at month 12 was reported [13]. The absolute rise in CD4⁺ T-cell counts after 12 months of HAART was mainly determined by the baseline CD4⁺ T-cell counts in our patients. More immunological benefit can be expected when therapy is started at higher CD4⁺ T-cell counts than actually practiced in Nouna [14]. Interestingly, not a single patient showed signs or symptoms of HAART-induced immune reconstitution inflammatory syndrome, a potential severe side-effect of successful T-cell regeneration in patients with advanced HIV-disease [15].

Over the entire year of treatment, a slight but continuous increase in the percentage of naive CD8⁺ T-cells was seen until month 6 of therapy. At the same time, the percentage of naive CD4⁺ T-cells reached pre-treatment levels again, and expression of the CD95 receptor was significantly reduced as compared to baseline. Regeneration of both naive CD4⁺ and CD8⁺ T-cells indicates that repetitive exposure to tropical infectious and non-infectious agents does not disturb the process of T-cell reconstitution under HAART in West African patients. After 12 months of HAART, however, HIV-1-infected patients in Nouna still had a significantly lower percentage of circulating naive CD8⁺ T-cells than healthy adult controls. A comparison with the reference values for naive T-cells in our study area [3] revealed lower absolute numbers of naive CD4⁺ T-cells after 12 months of HAART than in healthy controls (mean±standard deviation: 228±146 cells per µl; 5^th to 95^th percentile: 35 to 496 cells per µl). Absolute numbers of naive CD8⁺ T-cells after 12 months of HAART were also below the values observed in healthy adults (mean±standard deviation: 173±103 cells per µl; 5^th to 95^th percentile: 36 to 363 cells per µl).

Early immunologic treatment response was accompanied by a significant reduction in the percentage of activated T-cells in the peripheral blood, while absolute counts of these cells increased. A similar rapid decrease in the activation levels of peripheral blood T-cells has been described in European cohorts [18-20] and in studies on HIV-1-infected Africans receiving antiretroviral therapy [7, 13, 21]. This decrease may thus be explained by a causal association between high levels of HIV-1 replication leading to T-cell activation, followed by an acceleration in the proliferation rate of naive T-cells. The slow increase in naive T-cells during HAART may depend on thymopoiesis and on homeostatic autoproliferation of postthymic T-cells in the periphery when filling empty spaces in a lymphopenic host. Without additional measurements (e.g., quantification of the fraction of Ki-67⁺ proliferating naive T-cells or of T-cell receptor rearrangement excision cycles) it is not possible to discern the relative contribution of these regeneration pathways [16]. According to Di Mascio et al. [17], HAART-induced changes in naive T-cell dynamics cannot be explained by changes in thymic output but rather depend on a treatment-induced reduction in proliferation rates of naive T-cells. These cells may be lost at a lower rate into the activated/memory compartment once the HIV-1-induced chronic hyperactivation is reduced by effective HAART. Naive T-cells may then participate in homeostatic autoproliferation leading to the late increase in resting naive T-cells observed in the Nouna patient cohort.

Analysis of the development of expression patterns of CD95 on peripheral blood T-cells supports this interpretation. We [6, 22] and others [23] had shown in the past that the cell populations constituting the CD95^dim and the CD95^bright peak of CD4⁺ T-cells correspond to resting naïve and activated memory cells, respectively. In HIV-infected individuals, the slow and continuous change in the density of CD95 receptors on these cells seems to be more closely connected to the slow increase in the percentage of CD45RA⁺CCR7⁺ cells than to the rapid decrease in expression of the CD38 receptor. After 12 months of HAART, however, HIV-1-infected patients in Nouna still had a slightly higher expression of CD95 on CD4^neg (i.e., CD8⁺) T-cells than healthy adult controls. The percentage of naive CD8⁺ T-cells was reduced in HIV-1-infected patients, and the expression of CD38 on CD8⁺ T-cells was not significantly different from the controls. Skewed maturation and increased activation of peripheral blood T-cells in HIV-1-infected patients after 12 months of HAART compared to healthy controls may therefore be explained by several factors: (1) Previous HIV-induced T-cell loss may not be fully compensated by thymic reconstitution and autoproliferation of naive T-cells after 12 months of HAART. This may lead to homeostatic expansion of these cells in the periphery causing increased expression of CD95 or CD38 on circulating naive T-cells as a sign of cellular cycling. (2) In response to environmental stimuli circulating naive T-cells may become constitutively activated. We had observed high prevalences of, e.g., CMV and HBV in the study area [24]. T-cell responses to viral infection would be expected to increase CD38 expression in the CD8⁺ T-cell compartment [25]. However, after 12 months of HAART, HIV-1-infected patients in Nouna showed a similar expression of CD38 on CD8⁺ T-cells as healthy controls. Differences in the exposure to other viral infections than HIV is thus an unlikely cause of the deviations observed in the CD4⁺ and CD8⁺ T-cell compartments in these individuals. (3) The residual viral replication - which was seen in 17 patients without good virological response (NVR) - may also act as a continuous T-cell stimulus. However, continuing HIV-replication would also be expected to increase primarily the expression of CD38 on CD8⁺ T-cells [7].

These results point to a common problem in flow cytometric analysis of T-cell activation markers in HIV-infected individuals. In the individual patient with rising CD4⁺ T-cell counts under HAART, an increase in CD38 expression on CD8⁺ T-cells most probably indicates a rise in viral load (see supplementary material). In the lymphopenic patient with advanced disease and continuously low CD4⁺ T-cell counts (<50 cells µl^-1) despite HAART, the effects of cell cycling due to homeostatic autoproliferation may be the main determinants of the expression levels of molecular markers of T-cell activation. It is therefore not surprising that we were not able to predict a good virological treatment response by measuring T-cell activation - despite the statistically significant correlations between these parameters. While the predictive value of CD38 expression on CD8⁺ peripheral blood T-cells for virological treatment failure has been demonstrated in patient cohorts in South Africa [13] and Thailand [26], findings similar to ours were reported by Ondoa and colleagues in patients from Abidjan, Côte d'Ivoire [7]. Despite appearing to be a sensitive detector of HIV-1 plasma viral load, the percentage of CD38⁺ CD8⁺ T-cells lacks specificity if used as a screening test for the sensitive detection of early virological failure. Even the use of CD38 median fluorescence intensity did not improve the capacity of the test to discriminate treatment success and treatment failure (see also Supplementary Fig. 1). Assessment of CD38 expression on CD8⁺ T-cells may be more relevant if it were restricted to the memory (CD45RO⁺) subpopulation of CD8⁺ T cells [27] or if the percentage of CD8⁺ CD38^bright T cells (defined as having >8,500 CD38 binding sites per cell) were measured [8]. However, the observation that assessment of longitudinal changes of these parameters in the individual patient under HAART may indicate changes of plasma viral load levels [13] is corroborated by similar observations in single patients of our West African patient cohort (see Supplementary Table 4).

Repeated assessment of CD4⁺ T-cell counts remains the most important measure of immunological follow-up of HIV-infected patients before and during HAART. The expression of the T-cell activation markers CD95 and CD38 may be an indicator of the overall immune recovery in patients with advanced disease receiving HAART. T-cell activation markers should be used with caution as a substitute for repetitive measurements of HIV-1 plasma viral load in order to assess the virologic response to treatment.