In our study we found that the median time to immunological recovery after ART initiation was 271 days (9.3 months) in the ADC group compared to 208 days (6.9 months) in the non-ADC group. The ADC group status was associated with a longer median time to immunological recovery after ART initiation in comparison to the non-ADC group. Factors that were independently associated with time to immunological recovery were group status (presence or absence of ADC), baseline CD4 count, baseline total lymphocyte count and adherence to ART. Although, immunological responses to ART have been estimated to be at least an increase of 50–100 CD4 cells/μl/year after ART initiation, the median time to attain this CD4 cell threshold has not been documented [12, 16]. We took the worst case scenario and estimated a gain of at least 50 CD4 cells/μl/year from the baseline value to indicate good immunological recovery process.
The differences in median times to immunological recovery between the ADC and non-ADC group found in our study may be a result of profound pre-treatment immuno-suppression which is a common feature of ADCs. The ADCs cause a loss of CD4 memory cells that represent a component of the T-cell repertoire that is specific to most ADC infections. The defect in the CD4 memory cells as a result of ADCs accounts for the inability of patients with ADCs to respond to recall antigens and evoke immunological responses leading to a slower immunological recovery process compared to patients without ADCs . Another explanation could be that the presence of ADCs causes immune activation that leads to increased infection of CD4 cells by HIV virus. This leads into CD4 cell destruction and increased viral replication . Our findings seem to support these findings and are similar to other studies in a multi-centre clinical trial where patients with ADCs who were followed-up for 144 weeks who had a slower immunological recovery than patients without ADCs . The contrasting feature of this study from our study is that it did not determine the median time to immunological recovery. We found that the hazard of attaining immunological recovery when a patient had no ADC increased by 1.3 times compared to when a patient had an ADC.
Other determinants that favored a good immunological recovery process after ART initiation were a baseline CD4 count (> 200 cells/μl), baseline total lymphocytes count (≥ 1200 cells/mm3) and adherence to ART (≥ 95%). The hazard of attaining immunological recovery reduced when comparing a patient with ≤ 200 CD4 cells/μl to a patient with > 200 cells/μl by 50%. Our results are consistent with other studies where immunological recovery is largely dependent on baseline CD4 count and thus the timing of ART initiation is important in order to maximize the CD4+ T-cell response to therapy . In contrast, one study from South African community cohort that followed-up patients for 44 weeks, found that patients who had a baseline CD4 count below or equal to 50 cells/μl had equivalent or greater immunological recovery compared to patients with higher baseline CD4 count . We found out that there was no interaction between CD4 count and other variables (group status, viral load count, total lymphocyte count, age, sex), although it has been documented else where [20, 21]. The lack of interaction may be a result of inadequate power due to limited sample size among the various CD4 count strata.
We found a significant association between adherence level and time to immunological recovery. The hazard of attaining immunological recovery when a patient had adherence level of ≥ 95% compared to one of lower adherence level was 2.2. The adherence level to ART in the ADC group was much lower than the non-ADC group. Similarly, the proportion of patients who achieved virological suppression in the ADC group was lower than the non-ADC group. The non-adherence in the ADC group could have resulted from the pill burden as a result of the concomitant medications other than the drug-drug interactions from the rifampicin-based and ART regimens which has been reported else where . During the intensive phase of the anti-tuberculosis treatment, patients were started on efavirence instead of niverapine-based regimens, although in some patients ART was deferred until the intensive phase was completed. Another reason for the non-adherence was that the patients were too ill to adhere to ART medications due to the concurrent co-morbidities which is a common feature in the sub-Saharan Africa . Our adherence levels to attain immunological recovery are similar to what other studies have found. In a multi-centre study in Africa ≥ 95% to ART was necessary to achieve viral suppression and increase in CD4 counts . In another study in the USA, adherence of ≥ 95% to protease inhibitors was necessary to achieve a good immunological recovery . During ART initiation, the majority of our patients (89.7%) had been started on non-protease ART regimens.
A baseline total lymphocyte count ≥ 1200 cells/mm3 was found to be associated with a short median time to immunological recovery. The hazard of attaining immunological recovery increased by 1.8 times when comparing a patient with a total lymphocyte count of ≥ 1200 cells/mm3 to one of a lower total lymphocyte count. Some studies have confirmed the significant association between a total lymphocyte count of < 1200 cells/mm3 and subsequent poor immunological recovery, disease progression or mortality [25, 26].
In our study, old age was not associated with median time to immunological recovery, although a significant association at bivariate analysis was observed. Later, this association between old age and time to immunological recovery disappeared after adjusting for other variables. Our findings are similar to those in a cohort of older participants in the USA (≥ 55 years) where increasing age was associated with a slower and lower immunological recovery due to age-related decreases in thymopoiesis . The 2 cohorts differed in their mean age. The mean age of the patients in our cohort was 40 years whereas for the cohort in the USA was 55 years.
We also found no effect of sex on median time to immunological recovery. Our findings are similar to other findings else where that show that there are no gender differences between males and females regarding response to ART . On the contrary, one study of antiretroviral-naive patients found that women had a greater CD4 cell response than men, even after adjusting for baseline CD4 cell count and viral load. Pharmacokinetic differences may have resulted in women having higher antiretroviral drug levels, which led to more-profound virus suppression and resulted in a greater increase in CD4 cell count .
We recognize our study had the following limitations;
Our study design did not allow for the control of all possible confounders as it was retrospective cohort. We relied on measurements taken in the past and where a possible confounder was not measured we could not control for it. For example, we were not able to determine the effect of Body Mass Index (BMI) as height measurements were only available in about 35 patients. Although the baseline body weights were available, subsequent measurements during the follow-up visits were missing and only available in 40 patients. A baseline BMI of < 20.3 kg/m2 for men and < 18.5 kg/m2 for women is predictive of poor immunological recovery, increased morbidity and mortality [29, 30]. We were also not able to determine the effect of payment for ART services as this information was not also available, although we found from the chart reviews that some patients were paying for ART drugs.
The time taken to attain 50 CD4 cells/μl after ART initiation was estimated according nearest date of visit to JCRC outpatient clinic when the CD4 count was done. This estimation could have over or underestimated the true estimate. To minimize errors in estimation of time to attain 50 CD4 cells/μl both the principal investigator and data entrant independently calculated the time, and where differences arose an average was taken. Overall inter-observer kappa statistic was 0.68.
Another bias could have resulted from under diagnosis of ADCs because of lack of diagnostic equipments, a common finding in resource limited settings. It is surprising that some of the ADCs were not diagnosed during 2002–2006 when we reviewed records, whereas had been found to be prevalent by other study sites. For example in a nearby main referral hospital, 18% of hospitalized patients presented with non-typhoid salmonella, as an ADC in blood stream . We minimized under diagnosis of ADC by chart review to compare diagnosis of what was in data base whether conforms to what is in the charts. Another explanation could be that our facility was predominantly an out-patient setting.
Some of the immune responses after ART initiation were likely to have been evoked by immune reconstitution syndrome (IRS) and diagnosed as ADCs. This is an inflammatory response against infectious and non-infectious antigens and usually occurs within 12 weeks after ART initiation . However, only less than 3% of patients had ADCs within 12 weeks of ART initiation, thus ruling out the confounding effect of IRS.
We could also not explain a big loss to follow-up of patients who had enrolled for ART as it was inadequately documented. Of 609 who were eligible and enrolled for ART, 182 (29.9%) could not come for second or subsequent review visits. We could explain the loss follow-up could have resulted from deaths due to profound immuno-suppression at ART initiation. Our findings may have overestimated individuals achieving immunological recovery, as those who died from an AIDS-defining disease shortly after starting ART were excluded from the analysis. It is also possible that these patients sought treatment in other treatment centers elsewhere. Since 2003, Uganda has benefited from roll-out program for ART services where more centers have been accredited to offer HIV services and treatment has been either subsided or become free. The loss to follow-up of 29.9% could have resulted into affecting the generalizability of the study findings. This was minimized by taking precautions to prevent under or over diagnose ADCs by chart review, and correct estimation of time to end points from baseline value by taking independent measurements. Secondly, the baseline characteristics of the patients who were lost to follow-up were similar to the patients who had a subsequent follow-up after the baseline visit (CD4 counts 131 vs 115 cells/μl; p = 0.640). Lastly, findings of this study are strengthened by the relatively homogeneous study population receiving treatment at a single health facility using the national treatment protocols. Patients were all ART-naïve and received a standard triple-drug regimen.