In this population with good rates of viral suppression as was previously reported , the frequency of SO-CD4, using the CD4 increase criteria, was 21%, 45% and 54% at 6,12 and 24 months respectively. Our findings are comparable to results from other developing countries (Africa, Latin America and Asia) where 19% of patients had SO-CD4 using a similar criteria of a CD4 increase of < 50 cells/μl after 6 months of ART . Similarly, the frequency of SO-CD4 at 6 and 12 months is comparable to what has been reported in industrialized countries that used similar criteria [4, 11, 13]. Overall, our results show similar profiles of CD4 reconstitution in both the developing and industrialized countries despite the challenges with infrastructure for care delivery in sub-Saharan Africa.
We found that patients with baseline CD4 counts of 50–199 cells/μl were about 3 times more likely to have SO-CD4 than those with baseline CD4 counts of 0–49 cells/μl. Our results are similar to reports from South Africa where patients in the lower CD4 stratum had a higher gradient of CD4 increase . This is contrary to previous reports that advanced pre-treatment immunodeficiency is associated with diminished capacity to restore quantitative and functional CD4 T cell responses during antiretroviral therapy [14, 15]. We attribute our results to the peripheral expansion and/or redistribution of CD4 T cells that is described in the initial phase of CD4 reconstitution on ART . Our results imply that the CD4 increase criteria of SO-CD4 is not enough in a setting where patients present to hospitals and HIV care units with untreated advanced HIV disease [1, 17]. In addition, we used a threshold of 200 cells/μl below which patients were classified as SO-CD4 since this gives an indication of the general susceptibility to opportunistic infections. Using the CD4 threshold criteria, we found that 43%, 42% and 19% had SO-CD4 at 6, 12 and 24 months respectively; thereby remaining at risk of opportunistic infections.
Patients that initiated therapy with a zidovudine-containing regimen were 3.6 times more likely to develop SO-CD4 than patients on a d4T-containing regimen and we attribute this to the myelosuppressive effects of zidovudine . We interpret these results cautiously because only 26% of our patients initiated a zidovudine-containing regimen and they were not randomized. However, evidence in the US shows that use of a protease inhibitor (PI)-based regimen is protective against poor immune reconstitution [6, 19] because PIs modulate activation of peripheral blood CD4 T cells and decrease their susceptibility to apoptosis . Since the long term prognosis of patients exhibiting discordant responses remains unknown , we need to explore the use of the newer and less toxic first line regimens  for patients at risk of SO-CD4.
Age was not a significant predictor of SO-CD 4 in our study and this is consistent with what was reported in a US cohort . However, some previous studies showed that age above 30 years was associated with SO-CD4 [5, 11] because it correlated with thymic involution yet preserved thymic function is necessary for adequate CD4 T cell recovery [11, 23]. Similarly, hepatitis B co-infection did not predict SO-CD4 as was recently reported that hepatitis B co-infection had no impact on the response to ART regarding viral suppression and immune recovery.
Majority of the patients with SO-CD4 after 6 months, using either of the criteria, still had SO-CD4 at 12 months despite sustained HIV-RNA viral suppression. Since patients with SO-CD4 at 6 months are likely to maintain the phenomenon, they may need evaluation of the recovery of CD4 cell function, more so in Africa where there is an increased background risk of opportunistic infections. It is possible that the CD4 cells do not recover both in absolute numbers and function because of the high levels of T-cell activation in Africans due to frequent infections by the various pathogens endemic in the region [10, 25, 26].
It is also likely that these patients may require extended periods of prophylaxis against opportunistic infections. Our analysis was however limited to recovery of peripheral CD4 T cell counts and not CD4 T cell function. We recommend studies to examine other markers of recovery of immunological function among patients with SO-CD4.
We found that about a third of the opportunistic infections occurred among patients with SO-CD4 reconstitution as defined by either the CD4 increase or the threshold criteria. Similar to what has been reported in other cohorts, most of the AIDS-related events occurred in the first 6 months [27–29] and the spectrum of opportunistic infections was similar to what was found among patients at Mulago hospital where most patients with advanced HIV disease were hospitalized with severe bacterial pneumonias and tuberculosis . More AIDS-related events were recorded among patients without SO-CD4 and we postulate that immune reconstitution inflammatory syndrome (IRIS) contributed to this difference . However, Kaplan-Meier analysis showed no statistically significant differences in the rates of AIDS-related clinical events among patients with and without SO-CD4 in the setting of HIV-RNA viral suppression. On the contrary, in industrialized countries, patients with SO-CD4 (using similar criteria) have previously been reported to have a higher risk of developing an AIDS-related clinical events [4, 30]. In the Swiss cohort, suboptimal responders had a 1.5 fold higher incidence of opportunistic infections than the complete CD4 responders . However, we are cautious to compare our results with the latter cohort because the authors used a CD4 threshold below 500 cells/μl after 5 years of ART to define SO-CD4 at a frequency of 35.8%. We need to consider SO-CD4 after longer periods of follow up like has been done in the industrialized countries. Our results add to the emphasis that viral load testing is required for monitoring patients on ART in resource limited settings  especially those patients that present with unsatisfactory CD4 reconstitution in order to guide treatment decisions for this subgroup of patients.
The findings in this study are strengthened by the relatively homogenous study population of ART-naive individuals receiving ART at a single facility using standardized clinical protocols. Our patients used NNRTI-based ART regimen that are used in most HIV care facilities in Africa so our results can be generalized to most patients in Africa however they are limited to patients with sustained HIV-RNA viral suppression which, among others, is the ultimate goal of ART. We need to design studies of interventions for patients on ART with poor immune reconstitution and minimize the time spent with CD4 counts below the 200 cells/μl critical threshold. It is important to note that adherence to ART and previous exposure to ART were not considered to contribute to SO-CD4 in our study since all patients were naïve to ART and patients were included in the analysis only if they had HIV-RNA viral load < 400 copies/ml which we used as a proxy for good adherence.