Clinical case definition and manifestations of paradoxical tuberculosis-associated immune reconstitution inflammatory syndrome.

Background: The International Network for the Study of HIV-associated IRIS (INSHI) recently published criteria for tuberculosis-associated immune reconstitution inflammatory syndrome (TB-IRIS) diagnosis. The performance of this definition and clinical manifestations of TB-IRIS were studied.

Methods: Antiretroviral therapy-naive HIV/TB Thai patients receiving antituberculous therapy were enrolled during 2006-2007 and prospectively followed through 24 weeks of antiretroviral therapy. Patients were defined as having paradoxical TB-IRIS if they fulfilled the 'study definition' by French 2004 and were confirmed by an external reviewer. All were later compared by the classification according to 'INSHI-2008'.

Results: For the 126 patients, median baseline CD4 cell count was 43 cells/microl and HIV-1 RNA was 5.9 log(10) Y copies/ml. Seventy-three (58%) had extrapulmonary/disseminated TB. Twenty-two (18%) and 21 (17%) fulfilled TB-IRIS criteria according to the study definition and INSHI-2008 definition, respectively. Two (2%) were diagnosed by study definition only and one (1%) by INSHI-2008 definition only. Twenty (16%) were concordantly diagnosed by both definitions and 103 (82%) were consistently negative. Eighteen (82%) had worsening of a preexisting site, whereas four (18%) had TB-IRIS in a new location. Lymph node enlargement (73%) and fever (59%) were common in TB-IRIS. Sensitivity and specificity of INSHI-2008 was 91% (95% confidence interval, 72-98%) and 99% (95% confidence interval, 95-99.8%), respectively. Positive predictive value was 95% and negative predictive value was 98%. By multivariate analysis, factors predicting TB-IRIS were extrapulmonary TB (odds ratio, 8.63) and disseminated TB (odds ratio, 4.17).

Conclusion: There was high concordance between the INSHI-2008 and French 2004 definition for TB-IRIS diagnosis in HIV/TB patients with relatively high rate of paradoxical TB-IRIS. This suggests that lack of HIV-1 RNA and CD4 cell count monitoring does not impede the ability to diagnose TB-IRIS.