Sub-Optimal CD4 T-Lymphocyte Responses among HIV Infected Patients who Develop TB during the First Year of ART

Sub-Optimal CD4 T-Lymphocyte Responses among HIV Infected Patients who Develop TB during the First Year of ART Ingrid Eshun-Wilson1,2*, Jantjie J Taljaard3 and Jean B Nachega1,4 1Centre for Infectious Diseases, Stellenbosch University, Parow, 7505, Cape Town South Africa 2Department of Epidemiology and Biostatistics, University of San Francisco, California, 41097, USA 3Centre for Infectious Diseases and Division of Internal Medicine, Stellenbosch University, Parow, 7505, Cape Town South Africa 4John Hopkins School of Public Health, Johns Hopkins University, Baltimore, Maryland, USA

In South African cohorts HIV infection is complicated by high rates of TB co-infection; during the fi rst year of ART, TB incidence is reported to be 3.5 cases per 100 person-years. Th is decreases substantially with increased time on ART and at 5 years; incidence rates reduce to 1.01 cases per 100 person-years [9]. Studies evaluating immunological responses to ART have found poor responses to occur in patients who develop incident TB aft er initiating ART, these studies have however been limited by a small numbers of incident TB cases or a lack of corresponding viral load measurements [9,10]. Further studies which evaluated the combined eff ect of TB prior to ART, at baseline and aft er ART initiation have found no association with CD4 T-lymphocyte count or virological response [11,12].
Th e aim of this study was to determine the association between incident TB occurring during the fi rst year aft er ART initiation, and sub-optimal immunological responses to ART. It was hypothesized that HIV infected patients who were receiving ART and developed TB during the fi rst year of ART would have poorer CD4 responses as compared to those who did not develop incident TB.

Study design and study population
A retrospective cohort study was conducted of patients initiating ART through the national ART program at Tygerberg Academic Hospital Infectious Diseases Clinic, in Cape Town, South Africa between 1 January 2004 and 15 May 2008. Patients initiating ART received a backbone of stavudine or zidovudine with lamivudine and either efavirenz or nevirapine. Patients with TB were preferentially initiated or switched to an efavirenz containing regimen. Follow-up data was censored on 28 Feb 2009. TB treatment was Rifampicin based and routinely provided for 6-8 months according to WHO guidelines [13]. All patients initiating ART were screened for active TB through clinical evaluation and microbiological tests and chest-radiograph. Viral load and CD4 T-lymphocyte count measurements were performed routinely every 6 months aft er initiating ART. If clinical failure was suspected viral load measurements were performed outside of the 6 month window at the physicians' discretion. Baseline viral load measurements were only available between 2004 and 2005 due to changes in national policy aft er this period. In an attempt to decentralize care a large proportion of patients were transferred to peripheral community ART clinics aft er initiating ART diminishing the number of patients with follow-up beyond the fi rst 6 months of ART.

Defi nitions and laboratory measurements
Baseline measurements (CD4 T-lymphocyte count and viral load) were included if performed within six months prior to ART initiation and once on treatment follow up measurements were included if performed within 2 months (+/-) of the 6 or 12 month time points.
Th e diagnosis of active TB was made both at peripheral TB clinics and at the study site with the use of World Health Organisation (WHO) TB case defi nitions [14]. Th ese diagnoses were based on clinical, microbiological and radiological fi ndings. Data from peripheral clinic TB record cards were routinely captured in the clinic database if patients developed TB while on ART. 'Incident TB' was defi ned as active TB which developed aft er initiating ART and 'prevalent TB' as active TB for which treatment was initiated prior to ART.
A sub-optimal CD4 response at 6 months was defi ned as a failure of the CD4 T-lymphocyte count to rise from baseline by more than 50 cells/ µl [2,3,8]. A detectable viral load during the fi rst year was determined as any viral load measurement above 400 copies/ml between 6 and 12 months on ART.

Sample size estimation
All available cohort data was used for evaluation in this study.

Data sources and ethics approval
Data was abstracted from clinical records and clinic and laboratory databases. Th e Health Research Ethics Review Committee of the University of Stellenbosch approved the protocol.

Statistical analysis
Stata version 11 was used for data management and analysis. Logistic regression was used for all univariable and multivariable analyses. Th e results are presented as odds ratios (OR) with 95% confi dence intervals (CI). All signifi cance test results represent likelihood ratio tests and are two-sided (α =0.05). Multivariable models were constructed using a forward step-wise approach. Age and sex were included a-priori in models. Interactions between terms were evaluated for each model.

Results
Of 936 patients who had received ART for 6 months or longer, 691 were eligible for inclusion into the study ( Figure 1) with a total of 660 patient years of follow up.
Th e median age of the fi nal cohort was 35 years and 66% of the cohort was female (Table 1). Incident TB developed in 49 (7.1%) of patients within the fi rst 12 months of ART, with an incidence rate of 7.4 cases per 100 patient years of follow up. Th e median timing of incident TB during the fi rst year of ART was 4.24 months (IQR: 1.8-6.9 months). 179 patients were already receiving TB treatment (prevalent TB) for a median of 3.2 months (IQR: 2.03-4.83 months) at the time of ART initiation.
Sub-optimal CD4 T-cell responses at 6 months on ART were seen in 141 (20.4%) of patients in the cohort. A detectable viral load between 6 and 12 months aft er initiating ART occurred in 72 (10.42%) patients and among these the median detectable viral load was 5.21 copies/ml (IQR: 4.63-5.81 copies/ml).
Among patients who developed incident TB within the 1 st year of ART, 17 of 49 (34.7%) had sub-optimal CD4 T-cell responses as compared to 104 of 478 (21.8%) patients who had no evidence of TB.
Th e univariable and multivariable analyses of the association between incident TB and sub-optimal CD4 responses (  * Analysis restricted to the 527 patients with no TB at baseline ** No interactions detected between variables. † Detectable viral load between 6 and12 months on ART. found have a 2.2 times greater odds of a sub-optimal CD4 response at 6 months on ART (95% CI: 1.14-4.23; p-value: 0.021).

Discussion
In this cohort, patients who developed incident TB during the first year after initiating ART were more likely to have suboptimal CD4 responses at 6 months of treatment. It is however not possible to determine the causal direction of this relationship from such observational data. Active TB has been shown to cause CD4 lymphopaenia independently of HIV status this may have contributed to the poor immunological responses seen during active TB infection in this cohort [15].
In a similar Ugandan cohort of 5982 HIV infected patients initiated on ART, incident TB was found to occur in 5.6% of patients within 12 months and was associated with sub-optimal CD4 responses at up to 24 months [10]. This suggests that incident TB in HIV infected patients is associated with long lasting immune suppression, virological suppression and adherence data were however not available for analysis in this study. A South African study evaluating 27 incident TB cases occurring over 5 years found sub-optimal CD4 responses to be associated with incident TB independently of virological failure [9].
Strength of our study is the availability of viral load data after ART initiation and substantially more cases of incident TB as compared to those evaluated by Lawn et al. [9]. The ability to adjust for elevated viral load in the multivariable analysis compensates to some degree for lack of direct data on ART adherence in this study. The high viral load measurements seen in patients with a detectable viral load suggest poor ART adherence or treatment failure in these cases. Direct adherence data would however have contributed further to this analysis and reduced residual confounding.
The association between age and sub-optimal CD4 recovery has been attributed to the deteriorating thymic function which occurs with advancing age and lower naive/memory CD4 ratios among older patients [16,17]. The need to re-establish immune function more rapidly during early ART among severely immunocompromised patients may explain the better short term CD4 responses among patients with low baseline CD4 count [5]. Patients with low CD4 T-lymphocyte count at baseline have however been shown to have poorer long term CD4 responses and do not reach the same CD4 plateau over time as compared to those with higher CD4 T-lymphocyte counts at baseline [6,7].
The proportion of patients with either missing baseline or followup CD4 measurements limited the number of observations available for analysis at the 6 month time point and could have introduced bias into the study findings. Demographic characteristics and occurrence of TB of these excluded patients did however not differ from the remaining cohort. A further study limitation was the lack of information on exact ART treatment regimen at the time of TB treatment initiation. This would have helped in the evaluation of whether a time lag between starting TB treatment and switching from NVP to EFV impacted on virological responses. A longer duration of follow-up could have helped to ascertain whether early sub-optimal responses in patients with incident TB persist beyond the first year of ART. Other unmeasured confounders such as co-morbidities and other opportunistic infections may have impacted on CD4 responses in this cohort.
In this setting where ART is initiated at an advanced stage of immunosuppression, incident TB is associated with poor CD4 T-lymphocyte responses. Although the direction of causality cannot be determined from these data, these findings provide additional support for the initiation of ART at higher CD4 counts where the underlying risk of developing active TB is lower and where long term immune recovery is improved.