Paradoxical TB-IRIS is a clinical entity frequently encountered when ART is initiated in HIV and TB co-infected patients. Understanding of this phenomenon is particularly relevant for HIV clinics in South East Asia and Africa where many patients with active TB enter ART programs [1, 7].
Retrospective and prospective studies have reported that 7- 43% of patients with HIV-TB co-infection develop IRIS . The frequency of TB-IRIS among patients starting ART during ATT in our series (12.6%) is within this range. Our cases developed IRIS a median of 15 days after starting ART (IQR 15–36) (Table 2), which is similar to that reported by other authors. In most patients, TB-IRIS is seen to develop within the first 2–6 weeks [3, 7–10], though symptoms can occur as early as five days after starting ART. Case reports describing development of TB-IRIS after the first three months of ART initiation are infrequent .
Baseline CD4 T cell counts or CD4 T cell recovery after six months of ART were not different between patients who developed TB-IRIS and those who did not. Earlier studies from South India  and Thailand  report similar findings. However, there are cohort studies from South Africa [7, 13] which report a definite association of development of TB-IRIS with low baseline CD4 counts. The reason for this difference is not clear. It may be a reflection of the overall low CD4 counts of the cohort of HIV-TB cases in this analysis.
In our study from North India, we report serious neurologic and pulmonary manifestations of IRIS, with poor outcomes in patients with extensive and disseminated disease. Neurological disease following initiation of ART due to inflammation associated with previously subclinical M. tuberculosis infection in the central nervous system (CNS) is of particular concern . In this study, we report 4 cases (31%) of meningitis TB-IRIS. They were receiving ATT for extensive TB disease, but had no evidence of CNS involvement at time ART initiation. Following development of meningitis, all four patients were hospitalized and received steroids, but three died. The mean CD4 cell count of three meningitis patients who died was higher as compared to the other two paradoxical TB IRIS patients who died (99 vs.34 cells/μl). The published literature on CNS TB IRIS is mostly restricted to a few case reports [14, 15]. A recent prospective case series from South Africa by Pepper et al, with a larger sample size (190 cases of TB IRIS) has reported neurologic TB-IRIS in 23 (12%) cases where also meningitis was the most common manifestation, followed by tuberculoma and myeloradiculopathy. In their study, Pepper et al also describe a low (70%) six month survival for neurologic TB-IRIS cases .
Working under program conditions, our study was limited by non-availability of plasma viral loads. Also the sample size was inadequate to define risk factors for TB IRIS or draw statistical inferences in general. However it is a precise retrospective review of data, especially outcome data, from a referral TB hospital which cares for TB patients referred from all over north India.
In conclusion, similar to other regions of the world, paradoxical TB-IRIS is a frequent problem during concomitant ATT and ART in HIV-TB co infected patients from northern India and mostly occurs during the initial weeks of ART. Meningitis is a potentially life threatening manifestation of TB-IRIS. Future studies in areas with high prevalence of M. tuberculosis infection should therefore address the issue of how patients at risk of developing paradoxical CNS TB-IRIS might be identified.