Antiretroviral therapy (ART) must be started during tuberculosis (TB) treatment in HIV-TB–coinfected patients to reduce mortality.1 ART should be started as early as 2 weeks after TB treatment initiation in patients with low CD4+ cell counts to decrease mortality and AIDS complications.2–4 However, cotreatment of HIV and TB is complicated by drug–drug interactions, overlapping medication toxicities, programmatic challenges of implementing earlier ART, and an increased risk of paradoxical immune reconstitution inflammatory syndrome (IRIS). Paradoxical TB IRIS refers to recurrence or worsening clinical features of TB despite effective TB treatment, in contrast to unmasking IRIS, which is a new presentation of unrecognized TB that is “unmasked” after initiation of ART.5
Paradoxical TB IRIS incidence increases when ART is started in close proximity to TB treatment and in those with low CD4+ cell counts.6 Thus, increased rates of TB IRIS are anticipated as programs scale-up implementation of earlier ART in TB coinfection to reduce AIDS progression and death.7
To inform clinicians and programs providing ART in HIV-TB–coinfected patients, this planned secondary analysis was conducted to characterize the frequency, presentation, severity, and risk factors for TB IRIS within the AIDS Clinical Trials Group A5221 Strategy Study of Immediate Versus Deferred Initiation of Antiretroviral Therapy for AIDS Disease-Free Survival in HIV-Infected Persons Treated for Tuberculosis with CD4 < 250 Cells/mm3 (STRIDE). The STRIDE study was an open-label, randomized study comparing ART started earlier (within 2 weeks after TB treatment initiation) vs. later (8–12 weeks after TB treatment initiation) in HIV-infected participants with CD4+ cell counts <250 cells per cubic millimeter receiving treatment for suspected TB.2
The STRIDE study enrolled HIV-infected ART-naive participants with confirmed or probable TB, stratified by screening CD4+ cell count of <50 or ≥50 cells per cubic millimeter; eligibility criteria for the STRIDE study were described in detail elsewhere.2 Confirmed TB was defined as the detection of acid-fast bacilli (AFB) in sputum smear or lymph node specimen or a positive culture for Mycobacterium tuberculosis from sputum, lymph node, or another sterile site. Probable TB required clinician's assessment that signs and symptoms warranted empiric TB treatment. Participants were required to have received 1–14 days of rifamycin-based TB treatment at the time of study entry. Study-provided ART was 600 mg of efavirenz daily (Stocrin, donated by Merck, Whitehouse Station, NJ) and a fixed-dose combination of 200 mg emtricitabine daily and 300 mg tenofovir disoproxil fumarate daily (Truvada, donated by Gilead Sciences, Foster City, CA). Participants were followed for 48 weeks after study entry. The study protocol was approved by an institutional review board or ethics committee at each participating site. The National Institutes of Health funded this study and provided study oversight.
TB IRIS Definition and Severity Index
Paradoxical TB IRIS was defined as the presence of at least 1 major or 2 minor TB IRIS criteria, from the International Network for the Study of HIV-Associated IRIS case definition.8 Major TB IRIS criteria were new or worsening (1) lymph nodes, cold abscesses, or other focal tissue involvement, (2) radiologic features of TB, (3) central nervous system TB, or (4) serositis (pleural effusion, ascites, or pericardial effusion). Minor TB IRIS criteria were new or worsening (1) constitutional symptoms (fever, night sweats, or weight loss), (2) respiratory symptoms (cough, dypnea, or stridor), or (3) abdominal pain accompanied by peritonitis, hepatomegaly, splenomegaly, or abdominal adenopathy. The STRIDE TB IRIS case definition required neither concomitant ART administration nor demonstration of initial response to TB treatment. Site investigators were not blinded to treatment arm allocation when evaluating possible TB IRIS. All suspect TB IRIS cases were reported on a standardized case report form (see Supplemental Digital Content, http://links.lww.com/QAI/A478) and were confirmed by an independent reviewer blinded to treatment allocation (C.A.B.). TB IRIS was managed as per local standard of care and was not specified in the study protocol. A study-defined TB IRIS severity index was applied as follows: severe—IRIS complicated by hospitalization or death; moderate—IRIS necessitating corticosteroid use or invasive procedure for evaluation and/or symptom management (such as lymph node drainage); or mild—IRIS cases without hospitalization, steroid use, or invasive procedures.
The primary end point of this secondary analysis was time from the date of TB treatment initiation to the date of TB IRIS diagnosis. Participants who died or were lost to follow-up or completed follow-up without a TB IRIS diagnosis were censored at their last clinic visit. Failure-time plots were calculated using the Kaplan–Meier method.9 Cox proportional hazards models, stratified by screening CD4+, examined associations between baseline covariates (age, male sex, race/ethnicity, HIV RNA, body mass index, presence of AIDS-defining illness other than TB, randomized treatment strategy, confirmed vs. probable TB, South Africa vs. elsewhere, and South America vs. elsewhere) and time to TB IRIS.10 Covariates that were univariately significant (P < 0.1) were examined in a multivariate Cox proportional hazards analysis. Using the backward elimination method, the final stratified multivariate model contained only significant covariates. Because participants often died or withdrew from study before a TB IRIS diagnosis, a competing risks analysis was completed as a sensitivity analysis. Because a stratified competing risks analysis could not be done using the available software, the multivariate model as per the nonstratified Cox proportional hazards analysis was fit using proportional hazards regression modeling of subdistribution functions in competing risks.11 Between-group differences were assessed using Fisher exact, proportion, Wilcoxon rank sum, exact ordered Wilcoxon, and log-rank tests. Interactions were assessed using logistic regression. These tests were 2 sided with a 5% type 1 error rate. Incidence of TB IRIS was calculated using standard epidemiological methods.
From September 2006 through August 2009, 806 STRIDE participants were enrolled from 26 sites on 4 continents. TB IRIS occurred in 7.6% of participants (61/806) (Table 1), with an overall incidence of 9.0/100 person-years. The proportion with TB IRIS was twice as high in those randomized to earlier ART vs. later ART (10.4% vs. 4.7%) and in those with screening CD4+ cell count <50 vs. ≥50 cells per cubic millimeter (11.5% vs. 5.4%) (Table 2); there was a significant interaction between CD4+ cell count strata and ART arm (earlier vs. later, P = 0.014). Nearly half of the TB IRIS cases [44.3% (27/61)] occurred in participants with screening CD4+ <50 randomized to earlier ART. In those with screening CD4+ ≥50, there was no difference in TB IRIS incidence between earlier and later ART (5.7% vs. 5.0%, proportion test P = 0.85).
TB IRIS Manifestations and Timing
Of the total 61 TB IRIS cases, 42 (69%) occurred in the earlier ART arm and 19 (31%) in the later ART arm. The majority [93.4% (57/61)] of TB IRIS cases met one or more major TB IRIS criteria, whereas 6.6% (4/61) met only 2 or more minor criteria for TB IRIS (Table 3). The most common manifestations of TB IRIS were new or worsening lymphadenopathy (59.0%) and radiographic features of TB (41.0%), with infrequent serositis (8.2%) and central nervous system manifestations (6.6%). Minor TB IRIS criteria of constitutional (54.1%) and respiratory (34.4%) symptoms were also frequent. The majority of IRIS cases occurred in the first 4 months (Fig. 1) after TB treatment initiation [median 57 days (Q1, Q3 22, 84)] and presented significantly earlier in participants randomized to earlier ART vs. later (median 29 vs. 82 days, P < 0.001) (Table 4). Four TB IRIS cases occurred before ART initiation in the later ART arm; no IRIS cases occurred before ART in the earlier arm. Among those who initiated ART before experiencing TB IRIS, median time to TB IRIS after ART initiation was 16 days (Q1, Q3 10, 49), 18 days with earlier and 15 days with later ART initiation (P = 0.28). TB IRIS symptoms lasted a median of 87 days (Q1, Q3 44, 139) with a nonsignificant difference between earlier and later ART initiation (median 92 vs. 70 days, respectively, P = 0.79).
TB IRIS Management and Severity
More than half of the participants [54.1% (33/61)] with TB IRIS received corticosteroids to manage symptoms, with a median treatment duration of 15 days (Q1, Q3 7, 32). Twenty-one participants (34.4%) with TB IRIS required one or more invasive procedures to aid in diagnosis or to manage complications, such as abscess drainage. These procedures included fine-needle aspiration (10), outpatient abscess drainage (3), surgical abscess drainage (1), lymph node biopsy (5), lumbar puncture (2), thoracentesis (1), liver biopsy (1), and surgical pleural drainage (2). Hospitalization was required in 31.1% (19/61) of participants with TB IRIS; no deaths were attributed to TB IRIS. TB IRIS infrequently led to TB treatment interruption (>7 days) (3/61) or HIV treatment interruption (>3 days) (3/61); 2 participants interrupted both ART and TB treatment. Applying the study-defined TB IRIS severity index, severe TB IRIS requiring hospitalization occurred in 19/61 (31.1%), moderate TB IRIS (steroids and/or invasive procedure) in 25/61 (41.0%), and mild TB IRIS in 17/61 (27.9%) (see Table S1, Supplemental Digital Content,http://links.lww.com/QAI/A478). There was no difference in TB IRIS severity with earlier vs. later ART or with CD4+ <50 vs. ≥50 cells per cubic millimeter (exact ordered Wilcoxon P = 0.84 and 0.17, respectively).
Predictors of TB IRIS
Participants developing TB IRIS had significantly lower baseline CD4+ counts (45 vs. 82 cells/mm3, P < 0.001), higher baseline HIV RNA (5.7 vs. 5.4 log10 copies/mL, P < 0.001), and started ART earlier in relation to TB treatment (12 vs. 24 days, P = 0.028) (Table 1). TB IRIS occurred significantly more frequently in participants with confirmed TB compared with those with only probable TB [47/374 (12.6%) vs. 14/432 (3.2%), P < 0.001]. TB IRIS occurred in 14.9% (33/221) of the participants enrolled from South Africa compared with lower proportions among those from other African countries [1.8% (6/333)], South America [9.9% (16/161)], Asia [3.8% (2/52)], and North America [10.3% (4/39)]. Hispanic ethnicity was significantly associated with a 2-fold higher occurrence of TB IRIS compared with non-Hispanic ethnicity [14.0% (19/136) vs. 6.3% (42/670), P = 0.004].
The time to TB IRIS from TB treatment initiation was significantly shorter for those with baseline CD4+ <50 cells per cubic millimeter compared with other CD4+ categories (P = 0.004, Fig. 1) and for earlier ART compared with later (P = 0.002, see Figure S1, Supplemental Digital Content,http://links.lww.com/QAI/A478). In the stratified multivariate Cox proportional hazards model, earlier ART initiation was significantly associated with a shorter time to TB IRIS compared with later ART initiation. The hazard ratio (HR) of developing TB IRIS was 2.47 [90% confidence interval (CI): 1.56 to 3.91, P = 0.001] for earlier ART vs. later ART, after adjusting for male sex (HR: 1.80, 90% CI: 1.09 to 2.98, P = 0.054), enrollment from South Africa vs. elsewhere (HR: 2.78, 90% CI: 1.79 to 4.32, P < 0.001), baseline HIV RNA (HR: 1.83, 90% CI: 1.24 to 2.71, P = 0.011), baseline body mass index (HR: 1.07, 90% CI: 1.01 to 1.13, P = 0.066), and confirmed vs. probable TB at entry (HR: 3.16, 90% CI: 1.90 to 5.24, P < 0.001). We also considered the competing risks of death and lost to follow-up; the results of the competing risks analysis were not different from the Cox analysis.
In HIV-coinfected patients on TB treatment, TB IRIS occurred in 7.6%, with TB IRIS occurring twice as frequently in those with CD4+ counts <50 cells per cubic millimeter vs. ≥50 and in those starting earlier ART (within 2 weeks after TB treatment initiation) compared with those starting later ART. However, in those with CD4 ≥50, earlier initiation of ART did not increase rate of TB IRIS compared with those starting later, indicating that the increased risk of TB IRIS with earlier ART is concentrated in those with very low CD4+ cell counts of <50 cells per cubic millimeter. TB IRIS was not significantly more severe with earlier ART or in those with CD4+ <50 cells per cubic millimeter. Notably, there were no deaths associated with TB IRIS in this series.
As programs globally implement earlier ART in TB coinfection to reduce AIDS progression and death, increased rates of TB IRIS are an expected consequence. As ART implementation is increasingly task shifted from physicians to nurses and other nonphysician providers in resource-limited settings,12 appropriate training for health-care workers in the recognition and management of TB IRIS will be key, as well as access to experienced clinicians to aid in the management of severe and/or complex cases. Providers need the tools and training necessary to diagnose and manage TB IRIS, which includes access to diagnostic testing to exclude other AIDS-related diagnoses and drug-resistant TB, which can each present similarly to TB IRIS.13 Access to invasive procedures may also be necessary, as one-third of STRIDE participants with TB IRIS underwent procedures ranging from fine-needle aspiration to thoracotomy as part of IRIS diagnosis and/or management. The absolute number of invasive procedures was low (21/806); however, this number does not take into account the procedures performed on those in whom TB IRIS was suspected, but ultimately excluded. Once diagnosed, TB IRIS was addressed in the outpatient setting for the majority of patients (69%), and there were no deaths attributed to TB IRIS, indicating that TB IRIS was generally a manageable condition. However, considerable resources were used during TB IRIS management, with inpatient care used in 31.1% of participants and corticosteroids used in over half, with concomitant monitoring for steroid-related complications. TB IRIS episodes were of moderate (41.0%) or severe (31.1%) intensity in the majority of participants experiencing TB IRIS and lasted a median of nearly 3 months, leading to ongoing clinical resources to monitor and manage patients.
Finally, programs may require the resources and infrastructure to provide closer monitoring during the first 2–3 months of TB and HIV cotreatment, as this is a particularly vulnerable time for treatment-related complications. The risk for TB IRIS was greatest in the first month after ART initiation, with median times to TB IRIS presentation of 18 and 15 days in the earlier and later ART initiation arms, respectively. Further research is needed to define optimal monitoring and management strategies for TB IRIS in resource-constrained settings.
Two randomized trials examining when to initiate ART in HIV–TB coinfection had similar findings to the current study, but with several notable differences. Both the Cambodian-based Cambodian Early versus Late Introduction of Antivirals (CAMELIA)14,15 and South African–based Starting Antiretroviral Therapy at Three Points in Tuberculosis (SAPIT)16 studies found a significantly higher rate of TB IRIS with earlier ART (36% vs. 16%, P < 0.001, and 20.1% vs. 8.4%, P < 0.001, respectively). However, both studies had a higher overall TB IRIS rate than the current study (26.0% and 12.5%, respectively). This increased rate is attributable in part to the enrollment requirement for AFB smear-positive specimens, increasing the likelihood of culture-positive TB in the TB suspects; indeed, among STRIDE enrollees, the TB IRIS rate was higher in AFB smear or culture-positive participants (12.6%). CAMELIA's high TB IRIS rate of 25% was also derived from the lower median CD4+ cell count of participants (25 vs. 77 cells/mm3 in STRIDE). It is possible that there were additional factors, including racial differences in the study populations. Of note, STRIDE's TB IRIS rate in South African participants (14.9%) is quite similar to the 14.2% reported in the South African SAPIT study, suggesting that racial background may contribute to TB IRIS, in addition to local provider focus on recognition of TB IRIS and access to radiographic tools for IRIS evaluation. CAMELIA reported 154 TB IRIS cases with 7 (4.5%) deaths attributed to TB IRIS, 6 of which occurred in the earlier arm, and SAPIT reported 2 (2.5%) deaths out of 80 TB IRIS events, both in the earlier ART arm. The mortality benefit of earlier ART initiation in both these studies outweighed the mortality attributed to TB IRIS; however, TB patients with advanced AIDS need to be monitored closely in the first months of ART for complications of TB IRIS, which can be severe and, in rare cases, fatal.
TB IRIS is a well-characterized clinical syndrome8,17; however, diagnosis can be challenging as there is no confirmatory laboratory test, and other AIDS- and TB-related processes need to be excluded. Presence of TB IRIS was determined by study investigator evaluation during required study visits every 4 weeks and any unscheduled visits for toxicity evaluations. It is possible that participants with more advanced immunosuppression and those in the earlier ART arm may have been seen more frequently at unscheduled visits because of AIDS complications or medication toxicity, respectively, and thus assessed more often for TB IRIS creating a diagnostic bias. However, an increase in TB IRIS has consistently been reported in other series in association with earlier ART and low CD4+ cell counts.6,17
TB IRIS was more common in participants with earlier ART initiation and CD4+ cell counts <50 cells per cubic millimeter. Overall, TB IRIS occurrence was infrequent at 7.6%. When TB IRIS occurred, the majority (69%) of cases were mild to moderate in severity; however, 31% were hospitalized with TB IRIS and over half received corticosteroids. As ART is increasingly implemented within 2 weeks after TB treatment initiation to reduce AIDS-related complications and mortality, HIV–TB programs will need the diagnostic capabilities, clinical resources, and provider training necessary to diagnose and manage TB-related IRIS.
The authors thank the study participants, the site principal investigators, and staff for their exceptional efforts to conduct the study, coordinate efforts with the in-country TB control programs, and to help build the capacity of integrated HIV–TB services; the data managers, Carol Suckow, BSN, and Lynne Jones, BS; the clinical trials specialist Evelyn Hogg; the DAIDS protocol pharmacist, Ana Martinez, RPh; the field representative, Janet Nicotera, RN, BSN; the study's laboratory technologist, Patty Anthony, BS, CLS; the laboratory data coordinator, Travis Behm, BS; and the community representative, Martha Tholanah Mensah-King.
1. Abdool Karim SS, Naidoo K, Grobler A, et al.. Timing of initiation of antiretroviral drugs during tuberculosis therapy. N Engl J Med. 2010;362:697–706.
2. Havlir DV, Kendall MA, Ive P, et al.. Timing of antiretroviral therapy for HIV-1 infection and tuberculosis. N Engl J Med. 2011;365:1482–1491.
3. Blanc FX, Sok T, Laureillard D, et al.. Earlier versus later start of antiretroviral therapy in HIV-infected adults with tuberculosis. N Engl J Med. 2011;365:1471–1481.
4. Abdool Karim SS, Naidoo K, Grobler A, et al.. Integration of antiretroviral therapy with tuberculosis treatment. N Engl J Med. 2011;365:1492–1501.
5. Meintjes G, Lawn SD, Scano F, et al.. Tuberculosis-associated immune reconstitution inflammatory syndrome: case definitions for use in resource-limited settings. Lancet Infect Dis. 2008;8:516–523.
6. Lawn SD, Myer L, Bekker LG, et al.. Tuberculosis-associated immune reconstitution disease: incidence, risk factors and impact in an antiretroviral treatment service in South Africa. AIDS. 2007;21:335–341.
7. Haddow LJ, Moosa MY, Mosam A, et al.. Incidence, clinical spectrum, risk factors and impact of HIV-associated immune reconstitution inflammatory syndrome in South Africa. PLoS One. 2012;7:e40623.
8. Meintjes G, Lawn SD, Scano F, et al.. Tuberculosis-associated immune reconstitution inflammatory syndrome: case definitions for use in resource-limited settings. Lancet Infect Dis. 2008;8:516–523.
9. Kaplan EL, Meier P. Nonparametric estimation of incomplete observations. J Am Stat Assoc. 1958;53:457–481.
10. Cox DR. Regression models and life-tables. J R Stat Soc. 1972;34:187–220.
11. Fine JP, Gray RJ. A proportional hazards model for the subdistribution of a competing risk. J Am Stat Assoc. 1999;94:496–509.
12. Emdin CA, Millson P. A systematic review evaluating the impact of task shifting on access to antiretroviral therapy in sub-Saharan Africa. Afr Health Sci. 2012;12:318–324.
13. Meintjes G, Rangaka MX, Maartens G, et al.. Novel relationship between tuberculosis immune reconstitution inflammatory syndrome and antitubercular drug resistance. Clin Infect Dis. 2009;48:667–676.
14. Blanc FX, Sok T, Laureillard D, et al.. Earlier versus later start of antiretroviral therapy in HIV-infected adults with tuberculosis. N Engl J Med. 2011;365:1471–1481.
15. Laureillard D, Marcy O, Madec Y, et al.. Paradoxical tuberculosis-associated immune reconstitution inflammatory syndrome after early initiation of antiretroviral therapy in the camelia randomized. AIDS. 2013;27:2577–2586.
16. Naidoo K, Yende-Zuma N, Padayatchi N, et al.. The immune reconstitution inflammatory syndrome after antiretroviral therapy initiation in patients with tuberculosis: findings from the SAPiT trial. Ann Intern Med. 2012;157:313–324.
17. Meintjes G, Rabie H, Wilkinson RJ, et al.. Tuberculosis-associated immune reconstitution inflammatory syndrome and unmasking of tuberculosis by antiretroviral therapy. Clin Chest Med. 2009;30:797–810.