HIV-associated immune reconstitution inflammatory syndrome (IRIS) occurs in 10–25% of persons initiating antiretroviral therapy (ART) [1–4], and represents a significant challenge to ART programs in resource-limited settings [5,6]. IRIS comprises a diverse spectrum of clinical presentations in response to a range of pathogens , but most existing diagnostic criteria are generic rather than disease-specific [8,9]. Mycobacterium tuberculosis infection accounts for up to two thirds of IRIS in developing countries [2,3,10], and clear diagnostic criteria for tuberculosis-associated IRIS (TB-IRIS) would represent a valuable tool in clinical and research settings.
We, therefore, welcome the recent publication of a case definition for TB-IRIS by the International Network for the Study of HIV-associated IRIS (INSHI) . Compared with previous definitions [8–10], the main modifications proposed are: inclusion of two categories of clinical presentation (paradoxical IRIS and ART-associated TB – of which a subset is unmasking IRIS); TB-associated rather than generic clinical criteria; restriction to the first 3 months of ART; and no CD4 or viral load criteria. An important next step is to validate this definition in various settings. We evaluated the performance of the INSHI definition of TB-IRIS using positive, negative, and chance-corrected agreement with consensus expert opinion in a prospective cohort study of the epidemiology of IRIS in South Africa.
Study population and data collection
Four hundred and ninety eight HIV-1-infected patients initiating ART for the first time were enrolled at two public hospital-based clinics in Durban, South Africa, between December 2006 and October 2007. Patients were initiated on ART (stavudine, lamivudine, and efavirenz or nevirapine) according to South African national guidelines, which require a pre-ART CD4 cell count of less than 200 cells/μl or WHO stage 4 disease . All patients underwent pre-ART clinical and laboratory assessment, and symptomatic patients were screened for TB by chest radiography and/or sputum examination. Patients with TB started antituberculous therapy (ATT) prior to ART initiation. Patients were reviewed every 2–4 weeks until 24 weeks, with additional visits if indicated. Serology for hepatitis viruses was performed on all patients at baseline and at follow-up in all individuals with liver enzyme elevation. The cohort study was approved by UKZN Biomedical Research Ethics Committee (E06/024).
In this analysis, any clinical event in which TB was considered to play a role was reviewed by two investigators (L.H. and M.Y.M.), both experienced HIV clinicians. The diagnosis and underlying cause of the event were based on consensus expert opinion after consideration of all relevant clinical data and investigations. Routinely available investigations included plain radiography, ultrasonography, cerebrospinal fluid analysis, mycobacterial culture and sensitivity, bacterial culture, viral hepatitis serology, and cryptococcal antigen detection and culture. The third investigator (P.E.) arbitrated when there was a lack of consensus between the other investigators.
Classification of clinical events by expert opinion
Competing explanations for events were assessed for their relative likelihood and events were classified as IRIS, if this was the most likely cause for the deterioration, possible IRIS, if IRIS was plausible but competing explanations were equal in likelihood, or non-IRIS, if the event was more consistent with an alternative cause or there was little evidence to support IRIS. The likelihood of IRIS according to expert opinion was thus weighted by both positive evidence in its favor, and the lack of evidence for other causes. Events were further categorized as either suspected paradoxical TB-IRIS (events arising in patients on ATT when ART was initiated) or suspected unmasking TB-IRIS (events in patients not on ATT when ART was initiated).
Classification of clinical events using the International Network for the Study of HIV-associated IRIS case definition
A diagnosis of paradoxical IRIS based on the INSHI case definition required both a diagnosis of TB  and an initial response to ATT, together with at least one of four major or two of three minor clinical criteria, and exclusion of alternative explanations . Consistent with the detailed INSHI criteria, events in which drug-resistant TB, other opportunistic infections, and neoplasms could not be fully excluded were initially regarded as probable paradoxical TB-IRIS. In a retrospective evaluation, we were then able to confirm spontaneous resolution and reclassify some cases as paradoxical TB-IRIS.
The INSHI definition of ART-associated TB required three criteria: not receiving TB treatment at ART initiation; diagnosis of active TB after ART initiation; and fulfilling WHO diagnostic criteria for TB. Additional criteria required for unmasking TB-IRIS are presentation within 3 months of ART initiation and either ‘heightened intensity of clinical manifestations’ or development of a paradoxical reaction once on TB treatment.
After classifying clinical events using both expert opinion and the INSHI case definitions, positive, negative, and chance-corrected agreement were calculated using expert opinion as the reference. Possible IRIS cases (under expert opinion) and ART-associated TB (under the INSHI definition) were considered to be non-IRIS events. Events with discordant conclusions between the two approaches were reviewed to determine reasons for disagreement.
Of 498 patients initiating ART, 375 (75%) were women, median age was 35 years, median pre-ART CD4 cell count was 106 cells/μl, 63 (13%) were WHO stage 4, and 102 (20%) patients were receiving ATT for a median of 12 weeks (interquartile range 8–22 weeks). There were 25 deaths and 21 patients were lost to follow-up. There were 620 clinical events in total, including 333 in which an association with TB was considered (74 suspected paradoxical IRIS events and 259 suspected unmasking or post-ART TB events).
Overall, there was good agreement between expert opinion and the INSHI case definition (Tables 1 and 2). Positive agreement for suspected paradoxical TB-IRIS was 13/18 (72.2%), negative agreement 52/56 (92.9%), and chance-corrected agreement (kappa) 0.66. In the potential unmasking cases, there was 100% negative agreement, but a lower positive agreement (12/19, 63.2%), and a kappa value of 0.76.
Paradoxical tuberculosis-associated immune reconstitution inflammatory syndrome
Based on expert opinion, there were 18 paradoxical TB-IRIS events, of which 13 (72.2%) were also IRIS under the INSHI case definition. The remaining five cases were non-IRIS under the INSHI definition, including two that only met one minor clinical criterion (constitutional symptoms) and three with another disease process (Kaposi's sarcoma or cryptococcosis) as well as TB, so that other opportunistic infections could not be excluded.
Expert opinion classified 41 cases as non-IRIS and 15 as possible paradoxical IRIS (56 negatives in total). Expert opinion was unable to classify 12 of the 15 possible cases as IRIS because of uncertainty regarding either whether the clinical course was consistent with the expected progress of disease (n = 2) or whether elevated liver enzymes were caused by drug hepatotoxicity or TB-IRIS with hepatic involvement, after exclusion of viral hepatitis (n = 10). Fifty-two of the 56 non-IRIS cases based on expert opinion were also negative under the INSHI definition, whereas two were considered definite and two probable paradoxical TB-IRIS. The two definite cases were patients who developed respiratory and constitutional symptoms within a month of completing ATT, neither of whom had microbiological evidence of a TB relapse. The investigators' consensus opinion of non-IRIS in both instances arose because the clinical history was most consistent with another respiratory tract infection, and the recent completion of a full course of ATT made TB-IRIS unlikely. The two ‘probable’ TB-IRIS cases (INSHI definition) included one sudden death at home and one patient who was lost to follow-up; both were classified as only possible IRIS under expert opinion because of equal likelihood with competing diagnoses.
Antiretroviral therapy-associated tuberculosis and unmasking tuberculosis-associated immune reconstitution inflammatory syndrome
There were 19 unmasking TB-IRIS events based on expert opinion, 12 (63.2%) of which were considered TB-IRIS under the INSHI definition. Six of seven unmasking events did not qualify as IRIS by INSHI, either because symptoms started more than 3 months after ART initiation (n = 2) or they did not display ‘heightened intensity of clinical manifestations’ (n = 4). Clinical features of the latter four cases included very early onset within 2 weeks of ART initiation (n = 3), sweats or fever as well as respiratory symptoms (n = 4), conversion to sputum smear positivity (n = 3), new opacity on chest radiography (n = 4), and hilar or peripheral lymphadenopathy (n = 3). In all four patients, consensus expert opinion was that immune reconstitution played an important role in the clinical presentation of their disease. One additional unmasking TB-IRIS event (under expert opinion) was neither IRIS nor ART-associated TB according to the INSHI definition because the diagnosis of TB did not meet WHO criteria prior to death. Of 240 non-IRIS and possible IRIS events in the potential unmasking TB-IRIS group, there was 100% agreement by the INSHI definition.
We have shown strong agreement between the INSHI case definition for TB-IRIS and consensus expert opinion (the best approximation of a gold standard, in the absence of a specific diagnostic test for IRIS), for both paradoxical and unmasking events in a research cohort in South Africa. These data support the adoption of this definition in clinical practice, and for defining clinical events in future studies of TB-IRIS. We identified minor caveats in the application of this definition, which may assist other clinicians and researchers in the field.
Disagreements in which events were classified as IRIS by expert opinion but non-IRIS by the INSHI case definition (12 cases) arose for a number of reasons. First, in cases of potential paradoxical IRIS, there were three individuals with dual disease (TB and another AIDS-defining illness) in whom either or both may have been responsible for IRIS. In this situation, the INSHI definition ruled out TB-IRIS because ‘another opportunistic infection or neoplasm’ could not be excluded. Second, four cases of unmasking TB-IRIS (by expert opinion) were excluded by the INSHI definition because they did not exhibit ‘heightened intensity of clinical manifestations’. This is a very subjective criterion, likely to be problematic for many clinicians, especially if one considers the normal wide variation in character and intensity of TB symptoms and signs. Third, we found one individual who was diagnosed with unmasking TB-IRIS on expert opinion, but was excluded by the INSHI definition because of incomplete data to fulfill WHO TB diagnostic criteria . It is likely that this situation will occur more frequently in routine practice in resource-limited settings than in the research cohort evaluated here. Fourth, two cases arose after 3 months of ART, and finally, two only met one minor clinical criterion, but IRIS was still held to be the most likely diagnosis by investigators.
In contrast, there were four events classified as non-IRIS by expert opinion but IRIS using the INSHI definition. These comprised two individuals who were lost to follow-up resulting in insufficient information to exclude other causes, and two others in whom the INSHI definition led to a diagnosis of paradoxical TB-IRIS after completion of ATT.
The INSHI case definition does not offer guidance on the diagnosis of TB-IRIS presenting with elevated liver enzymes (unless there is hepatomegaly), a common clinical scenario represented by 10 possible IRIS cases (under expert opinion). Even after exclusion of viral hepatitis and other pathogens, differentiating TB-IRIS with hepatic involvement from drug-related hepatotoxicity is challenging [14,15].
The authors of the INSHI case definition highlighted the limited literature on unmasking TB-IRIS and the limitations of their criteria for differentiating unmasking TB-IRIS from ART-associated TB. We anticipate that ongoing studies on the immunopathogenesis of TB-IRIS will contribute toward resolving this issue. We endorse the recommendation that researchers and clinicians report the clinical histories of patients with ART-associated TB to help with the continuing refinement of this case definition. INSHI is currently developing IRIS case definitions specific to other important opportunistic infections such as cryptococcosis and Kaposi's sarcoma.
We would like to acknowledge the valuable contributions of the following individuals for patient care, data collection and assessment of clinical events: Anisa Mosam, Nonhlanhla Khanyile, Ntombizethu Sithole, Queen Zulu, Jaysingh Brijkumar, Ganasen Manikam Govender, and Devilliers Zitha.
Two authors (L.H. and P.E.) participated in the INSHI meeting in Uganda in 2006, in which the case definition under evaluation was conceived, but neither are co-authors of the publication of the proposed case definition for TB-IRIS.
The Medical Research Council, UK (G0700530), British Society for Antimicrobial Chemotherapy (GA745), and Doris Duke Charitable Foundation (2006029) supported the study.
Contributors: All authors contributed towards the concept, coordination, clinical management, data collection, assessment of clinical events, and analysis of this study. L. Haddow drafted the manuscript, which all authors subsequently reviewed, edited, and approved.
There are no financial conflicts of interest.
1. Ratnam I, Chiu C, Kandala NB, Easterbrook PJ. Incidence and risk factors for immune reconstitution inflammatory syndrome
in an ethnically diverse HIV type 1-infected cohort. Clin Infect Dis 2006; 42:418–427.
2. Murdoch DM, Venter WDF, Feldmann C, Van Rie A. Incidence and risk factors for the immune reconstitution inflammatory syndrome
in HIV patients in South Africa
: a prospective study. AIDS 2008; 22:601–610.
3. Huruy K, Mulu A, Mengistu G, Shewa-Amare A, Akalu A, Kassu A, et al
. Immune reconstitution inflammatory syndrome
among HIV/AIDS patients during highly active antiretroviral therapy in Addis Ababa, Ethiopia. Jpn J Infect Dis 2008; 61:205–209.
4. French MA, Lenzo N, John M, Mallal SA, McKinnon EJ, James IR, et al
. Immune restoration disease after the treatment of immunodeficient HIV-infected patients with highly active antiretroviral therapy. HIV Med 2000; 1:107–115.
5. Colebunders R, John L, Huyst V, Kambugu A, Scano F, Lynen L. Tuberculosis immune reconstitution inflammatory syndrome
in countries with limited resources. Int J Tuberc Lung Dis 2006; 10:946–953.
6. Lawn SD, French MA. Immune reconstitution disease: recent developments and implications for antiretroviral treatment in resource-limited settings
. Curr Opin HIV AIDS 2007; 2:339–345.
7. Price P, Lim A, French MA. Immune reconstitution syndrome is not a single entity. AIDS 2008; 22:911.
8. French MA, Price P, Stone SF. Immune restoration disease after antiretroviral therapy. AIDS 2004; 18:1615–1627.
9. Robertson J, Meier M, Wall J, Ying J, Fichtenbaum CJ. Immune reconstitution syndrome in HIV: validating a case definition and identifying clinical predictors in persons initiating antiretroviral therapy. Clin Infect Dis 2006; 42:1639–1646.
10. Lawn SD, Bekker L-G, Miller RF. Immune reconstitution disease associated with mycobacterial infections in HIV-infected individuals receiving antiretrovirals. Lancet Infect Dis 2005; 5:361–373.
11. Meintjes G, Lawn SD, Scano F, Maartens G, French MA, Worodria W, et al
-associated immune reconstitution inflammatory syndrome
: case definitions for use in resource-limited settings
. Lancet Infect Dis 2008; 8:516–523.
12. South African National Guidelines for Antiretroviral Treatment. Pretoria: South African Department of Health; 2003.
13. WHO. Improving the diagnosis
and treatment of smear-negative pulmonary and extrapulmonary tuberculosis
among adults and adolescents: recommendations for HIV-prevalent and resource-constrained settings. Geneva: Stop TB Department, Department of HIV/AIDS; 2007.
14. Hoffmann CJ, Charalambous S, Thio CL, Martin DJ, Pemba L, Fielding KL, et al
. Hepatotoxicity in an African antiretroviral therapy cohort: the effect of tuberculosis
and hepatitis B. AIDS 2007; 21:1301–1308.
15. Lawn SD, Wood R. Hepatic involvement with tuberculosis
-associated immune reconstitution disease. AIDS 2007; 21:2362–2363.