Determinants of Immune Reconstitution Inflammatory Syndrome in HIV Type 1-Infected Patients with Tuberculosis after Initiation of Antiretroviral Therapy

The occurrence of immune reconstitution inflammatory syndrome (IRIS) with paradoxical worsening of the manifestations of tuberculosis has been reported in HIV-1-infected patients after the initiation of antiretroviral therapy [1–5]. IRIS is a pleiomorphic syndrome that is sometimes associated with severe clinical manifestations that increase the complexity of therapeutic strategies [6, 7]. Apart from initiation of antiretroviral therapy, the determinants of IRIS occurrence in patients with tuberculosis remain unknown. Identification of these determinants could aid the diagnosis of IRIS and lead to preventive strategies. Thus, we studied patients who were coinfected with HIV-1 and Mycobacterium tuberculosis who initiated antiretroviral therapy after the initiation of antituberculosis therapy.

Methods. We retrospectively studied antiretroviral-naive HIV-1-infected patients who were treated subsequently for tuberculosis and HIV-1 infection at our unit during the period of March 1996 through December 2001. Patients were selected using the French national hospital health care information system, which classifies patients on the basis of their suspected or proven diagnosis at hospital discharge. Patients considered for this study were antiretroviral naive, initiated antiretroviral therapy within the 6 months after the initiation of antituberculosis therapy, and had a duration of follow-up of >2 months thereafter. Tuberculosis was defined by identification of M. tuberculosis by culture or by typical clinical features and favorable response to antituberculosis therapy after exclusion of other diagnoses. Tuberculosis was classified either as localized or as disseminated (if it involved ⩾2 organs).

The following data were studied by reviewing medical charts: age, sex, country of birth, risk factors for HIV-1 infection, Centers for Disease Control and Prevention stage of disease, and tuberculosis localization. HIV-1 RNA levels and CD4 and CD8 cell data (i.e., cell count, cell percentage, and ratio of CD4 cells to CD8 cells) were studied at the initiation of antituberculosis therapy and after 1 month of antiretroviral therapy. Among 590 patients classified as having HIV-1 infection, 73 were classified as having tuberculosis. Thirty-six subjects were excluded for the following reasons: identification of nontuberculosis mycobacteria by culture (n = 9), lack of confirmation of tuberculosis diagnosis (n = 9), receipt of previous antiretroviral treatment (n = 9), follow-up duration of <2 months (n = 4), incompletely filled medical chart (n = 3), lack of initiation of antiretroviral therapy (n = 2), and nonadherence to treatment (n = 1). Thirty-seven patients were finally included.

IRIS was defined using the criteria described elsewhere [1, 5–7]. It was classified as “possible” if there was a reappearance or worsening of previous tuberculosis manifestations or if there was an appearance of new manifestations, despite receipt of effective antituberculosis therapy, after exclusion of other diagnoses. Antituberculosis therapy was considered to be effective if bacteriological failure or relapse could not be documented, if the strain of M. tuberculosis was susceptible to antituberculosis drugs, and if the patient adhered adequately to antituberculosis therapy. IRIS was classified as “probable” if a case of possible IRIS met ⩾1 of the following criteria: (1) there was disappearance of symptoms after interruption of antiretroviral therapy and recurrence of symptoms after reinitiation of antiretroviral therapy; (2) there were positive results of direct bacteriological examination, showing the presence of acid-fast bacilli in an organ suspected of exhibiting IRIS, without further positive culture results; and (3) histological examination revealed granulomatous lesions in an organ suspected of exhibiting IRIS.

In the univariate analysis, categorical variables were compared using Fisher's exact test, and continuous variables were compared using the nonparametric Wilcoxon test. Associations that were statistically significant at the 10% level were entered into the multivariate analysis. Multivariate logistic regression analysis was performed. A manual stepwise selection procedure was performed. All tests were 2-tailed. All statistical analyses were performed with SAS software, version 8.2 (SAS Institute).

Results. The median age of patients was 35 years (range, 26–62 years), and 81% of patients were men. The leading risk factor for HIV-1 infection was heterosexual exposure (68%). Sixty-two percent of patients were migrants from intertropical areas, and 41% were migrants from sub-Saharan Africa. At the time of tuberculosis diagnosis, 73% of patients were unaware that they were HIV-1 infected, and 14% had had a prior AIDS-defining event. Tuberculosis was disseminated in 29 (78%) of 37 patients and involved a median number of 3 organs (range, 1–9). Tuberculosis was confirmed by culture in 33 of 37 patients (sputum culture, 30 patients; urine culture, 6 patients; and culture of a sample from another site, 6 patients). At the time of initiation of antituberculosis therapy, the median CD4 cell count was 100 cells/mm³ (range, 4–488 cells/mm³), and the median HIV-1 RNA level was 5.36 log₁₀ copies/mL (range, 4.13–7.2 log₁₀ copies/mL). Antiretroviral therapy was initiated after a median interval of 30 days (range, 0–105 days) after the initiation of antituberculosis therapy. Antiretroviral therapy consisted of a protease inhibitor-containing regimen for 23 patients (62%), a nonnucleoside reverse-transcriptase inhibitor-containing regimen for 8 patients (22%), a triple-nucleoside reverse-transcriptase inhibitor (NRTI) regimen for 1 patient, and a dual-NRTI regimen for 5 patients (14%). The median duration of antituberculosis therapy was 12 months (range, 3–60 months). The median duration of follow-up was 36 months (range, 2–65 months).

Sixteen patients (43%) had an IRIS, which was classified as probable in 10 patients and as possible in 6 patients. IRIS occurred at a median interval of 48 days (range, 23–172 days) after initiation of antituberculous therapy and at a median interval of 12 days (range, 2–114 days) after initiation of antiretroviral therapy. Manifestations associated with IRIS included reappearance of fever (14 patients), appearance or worsening of lymphadenopathy (11 patients), spleen abscess (3 patients), arthritis or osteomyelitis (3 patients), gastrointestinal involvement (3 patients), worsening of pulmonary infiltrates (2 patients), appearance of skin lesions (1 patient), and parotitis (1 patient). Severe manifestations (spleen rupture, compressive lymphadenopadenitis responsible for dyspnea, ureteral compression, and thromboembolic events) were observed in 5 patients. IRIS outcome was favorable without any treatment modification (6 patients) or required specific therapy (steroid treatment, 6 patients; nonsteroid treatment, 2 patients; antiretroviral therapy interruption, 7 patients; and surgery, 1 patient).

The characteristics of 16 patients with possible or probable IRIS (the IRIS group) were compared to those of 21 patients without IRIS (the non-IRIS group) (table 1). In univariate analysis of baseline characteristics, patients in the IRIS group more frequently had disseminated and extrapulmonary tuberculosis (P = .01) and tended to more frequently be migrants from sub-Saharan Africa (P = .09). A higher proportion of patients with HIV-1 RNA levels of >10⁵ log₁₀ copies/mL was observed in the IRIS group (P = .03).

View larger version:

• In this window
• In a new window

• Download as PowerPoint Slide

Table 1

Univariate analysis of factors associated with the occurrence of immune restoration inflammatory syndrome (IRIS) in 37 HIV-1-infected patients with tuberculosis (TB) who initiated antiretroviral therapy after antituberculosis therapy.

After 1 month of antiretroviral therapy, patients in the IRIS group had a higher CD4 cell percentage (P = .04), a higher ratio of CD4 cells to CD8 cells (P = .05), and a higher increase in both the CD4 cell percentage (P < .001) and the ratio of CD4 cells to CD8 cells (P < .001), compared with the non-IRIS group. The increase in CD4 cell count and the decrease in HIV-1 RNA level tended to be higher in the IRIS group (P = .07). The CD8 cell count and percentage at baseline and their variations after receipt of antiretroviral therapy were similar in the 2 groups (data not shown). In multivariate analysis, after adjustment for baseline CD4 cell count, the increase in CD4 cell percentage (OR, 1.34; 95% CI, 1.08–1.66) and the increase in the ratio of CD4 cells to CD8 cells (OR, 1.55; 95% CI, 1.06–2.26) after receipt of antiretroviral therapy were the 2 variables that remained independently associated with IRIS occurrence. None of the baseline characteristics remained significant.

Discussion. In antiretroviral-naive HIV-1-infected patients, IRIS is a frequent and sometimes severe event that happens during tuberculosis therapy, occurring in almost one-half of patients after initiation of antiretroviral therapy. Although one study, which evaluated antiretroviral-naive and -experienced patients together, found a lower frequency of IRIS (11%) [3], our results are concordant with the high frequency of IRIS observed in previous studies (range, 30%–45%) [1, 2, 4, 5].

Among the possible determinants of IRIS, dissemination of tuberculosis was associated with the occurrence of IRIS. One previous study found a higher frequency of IRIS among patients who presented with both extrapulmonary and pulmonary tuberculosis [3], and a recent review of IRIS in non-HIV-infected patients noted that IRIS occurs mostly (83%) in patients with extrapulmonary tuberculosis [8]. We observed, as have others, that patients who experienced IRIS tended to have a lower CD4 cell count and a higher HIV-1 RNA level at baseline [1, 5], which are associated with tuberculosis dissemination [9]. The trend towards higher frequency of IRIS among migrants from sub-Saharan Africa may be multifactorial, including genetic and environmental factors, and it may be related to the extent of tuberculosis or to the delay of therapy frequent in this population [10].

Although a delay in antiretroviral therapy in HIV-1-M. tuberculosis-coinfected patients could decrease side effects and drug interactions, it should be avoided for patients with a CD4 cell count of <100 cells/mm³ to prevent the occurrence of AIDS-related events [11]. This strategy of delay could be associated with a decrease in the rate of IRIS occurrence, which was not observed in the present study. Prospective studies will better address this question.

The parameters independently associated with IRIS occurrence in our study were greater increases in CD4 cell percentage and in the ratio of CD4 cells to CD8 cells. These results have implications for a better understanding of IRIS pathogenesis. In HIV-1-infected patients, antiretroviral therapy produces a partial restoration of immune response that is observed with common antigens and opportunistic agents and has been linked to increases in the CD4 cell count [12, 13]. In the present study, the fact that CD4 cell percentage and ratio of CD4 cells to CD8 cells, rather than CD4 cell count, were the only factors independently associated with IRIS raises the issue of the possible unbalanced T cell response in the pathogenesis of IRIS. Whether tuberculosis-associated IRIS shares common factors with the other pathogen-associated IRIS remains to be elucidated [14–16]. Nevertheless, the recognition of these markers can improve the accuracy of the diagnosis of IRIS, which is, at the present time, based on the association of multiple and heterogeneous criteria. Identification of additional factors associated with IRIS may contribute to the design of therapeutic and preventive strategies. Such knowledge is needed to face a potential IRIS epidemic, which could occur once antiretroviral therapy becomes available in developing countries where tuberculosis and HIV-1 infection are highly endemic.

• Received April 21, 2004.
• Accepted July 29, 2004.

• © 2004 by the Infectious Diseases Society of America