Pettit, April C MD, MPH*; Jenkins, Cathy A MS†; Stinnette, Samuel E MS*; Rebeiro, Peter F BA*; Blackwell, Robert B RN, BSN‡; Raffanti, Stephen P MD, MPH*‡; Shepherd, Bryan E PhD†; Sterling, Timothy R MD*
From the *Division of Infectious Diseases, Department of Medicine, Vanderbilt University School of Medicine, Nashville, TN; †Department of Biostatistics, Vanderbilt University School of Medicine, Nashville, TN; and ‡Comprehensive Care Center, Nashville, TN.
Received for publication December 2, 2010; accepted March 1, 2011.
Supported by the National Institutes of Health: K24 A1065298 (T.R.S.) and National Institutes of Health 2 T32 AI07474-13 (A.C.P.) and the Vanderbilt-Meharry Center for AIDS Research (P30 AI54999; C.A.J., S.E.S., P.F.R., B.E.S., T.R.S.).
Presented in part at the 17th Annual International AIDS Conference, August 3-8, 2008, Mexico City, Mexico. Abstract WEPE0153; and at the 17th Annual Conference on Retroviruses and Opportunistic Infections, February 16-19, 2010, San Francisco, CA. Abstract #P-121.
The authors A.C.P. and T.R.S. had full access to all of the data and take responsibility for the integrity of the data and the accuracy of the data analysis. Concept and design: A.C.P., B.E.S., T.R.S. Acquisition of data: A.C.P., S.E.S., P.F.R., R.B.B. Analysis and interpretation of data: A.C.P., C.A.J., B.E.S. Drafting of the article: A.C.P. and T.R.S. Critical revision of the article: C.A.J., B.E.S., S.E.S., P.F.R., and S.P.R. All authors reviewed and approved the final version of the authors.
The authors have no conflict of interest to disclose.
Correspondence to: April C. Pettit, MD, MPH, Vanderbilt University Medical Center, 1161 21st Avenue South, A2209 Medical Center North, Nashville, TN 37232-2582 (e-mail: firstname.lastname@example.org).
In the United States, during 2009, there were an estimated 13,000 new cases of tuberculosis (TB), and 1300 (10%) were estimated to be among HIV-infected persons.1 TB risk is higher among HIV-infected persons with lower CD4+ lymphocyte counts.2-5 To prevent TB and other HIV-related illnesses and death, current US guidelines recommend initiation of highly active antiretroviral therapy (HAART) when the CD4+ lymphocyte count falls below 500 cells per cubic millimeter.6 HAART decreases TB risk up to 70%-90% in HIV-infected populations7-10 by restoring CD4+ lymphocytes and deficits in cell-mediated immunity. However, TB risk is high in the 3-6 months after HAART initiation.11-16
This high risk may be partially due to incomplete immune restoration and “unmasking” of previously undiagnosed TB in the setting of immune reconstitution.17,18 TB-associated immune reconstitution inflammatory syndrome (IRIS) is an important early complication of antiretroviral therapy resulting from a rapid restoration of TB-specific immune responses. New TB infection and previously undiagnosed TB may also contribute to the high TB risk during this period. Alternatively, the high TB risk during this period may reflect the low CD4+ lymphocyte counts at which HAART is initiated.
Routine screening and treatment for latent TB infection (LTBI) has been shown to reduce the risk of active TB in HIV-infected patients.19 In the United States, it is recommended that all HIV-infected persons are tested for LTBI at entry into care with a tuberculin skin test (TST) or interferon-gamma release assay regardless of their TB risk.20 However, it is known that adherence to these screening guidelines is poor.21,22 HIV-infected persons should receive isoniazid-preventive therapy if they have a positive diagnostic test for LTBI, have a negative diagnostic test for LTBI but were a close contact to an infectious pulmonary TB case, or have a history of untreated or inadequately treated healed TB regardless of LTBI diagnostic testing.20
It is important to improve our understanding of the epidemiology of TB, specifically soon after starting HAART. The decision to initiate or re-initiate HAART is guided by CD4+ lymphocyte counts. HAART affects CD4+ lymphocyte counts, which in turn affects TB risk. Therefore, CD4+ lymphocyte count is a confounder that lies on the causal pathway between HAART exposure and TB risk (Fig. 1). We are aware of only 2 studies that have assessed TB risk in the same population before and after HAART initiation and have appropriately adjusted for the most recent CD4+ lymphocyte count as a time-updated covariate using a marginal structural model.23,24 However, neither of these studies was performed in a cohort of adults from a low incidence TB setting. Additionally, neither of these studies incorporated the duration of HAART exposure into the marginal structural model, therefore, they did not assess both the short-term and long-term effect of HAART on TB disease.
We conducted an observational cohort study among patients receiving HIV care at the Comprehensive Care Center (CCC), a multidisciplinary clinic in Nashville, TN. Patients were included in this study if they were HIV-1 seropositive and had ≥2 provider visits at the clinic between January 1, 1998, and December 31, 2008. Screening for LTBI in this cohort was done using tuberculin skin testing; interferon-gamma release assay testing was not in use at the CCC during the study period. This study was approved by the Vanderbilt Institutional Review Board.
Patients were considered TB cases if they had been diagnosed with active TB after their first visit to the CCC. TB diagnoses were classified as either culture-confirmed or culture-negative. A diagnosis of culture-negative (ie, clinical) TB was based on the provider's clinical judgment or the combination of clinical judgment plus a pathological specimen with caseating or necrotizing granulomas and/or acid fast bacilli that was not culture positive for Mycobacterium tuberculosis, in addition to a clinical response to anti-TB therapy. Extrapulmonary TB cases were defined as any extrapulmonary disease, even if pulmonary disease was also present. Disseminated TB cases were defined as a positive blood culture for M. tuberculosis or miliary disease on chest radiography. An initial review of all such cases was conducted using the electronic medical record (EMR) database and the diagnoses “TB” or “PPD.” Hard copy charts were reviewed to confirm all TB diagnoses. Date of TB diagnosis was defined as the date of anti-TB treatment initiation. TB diagnosed before first CCC visit and recurrent TB was excluded.
Unmasking TB-associated IRIS was defined as an active TB case presenting within 3 months of starting HAART and 1 of the following: (1) increased intensity of the clinical presentation, specifically any inflammatory signs or symptoms or (2) a paradoxical reaction after initiation of anti-TB therapy.25 For all cases of TB diagnosed within 3 months of HAART initiation, re-initiation, or regimen change, hard copy charts were reviewed to determine if the criteria for unmasking TB-associated IRIS were met.
Demographic and laboratory data were obtained via the EMR. Hard copy charts were reviewed to validate laboratory data obtained from the EMR and to identify any data missing from the EMR. Baseline CD4+ lymphocyte and HIV-1 RNA values were defined as the first available values within 120 days before or up to 365 days after the first study visit. CD4+ lymphocyte values and HIV-1 RNA values at HAART initiation and TB diagnosis were defined as the first available values within 120 days before the respective event.
Utilization of HAART was determined and validated by chart review for all patients in the study period. HAART was defined as a regimen that contained 2 nucleoside reverse transcriptase inhibitors (NRTIs) plus a protease inhibitor (PI), a nonnucleoside reverse transcriptase inhibitor (NNRTI), or a third NRTI; 1 NRTI, 1 PI, plus 1 NNRTI; 2 PIs plus 1 NRTI or 1 NNRTI; or any regimen containing enfuvirtide plus at least 1 other agent. Antiretroviral therapy data (including regimen start and stop dates) were entered into an EMR by medical providers at the time of the patient encounter and validated by systematic chart review. If the chart noted that the patient had not taken the prescribed antiretroviral therapy, the patient was coded as not taking antiretroviral therapy during that period.
Length of HAART exposure was defined as the time on HAART because most recent regimen initiation. A patient was determined to be on HAART if they received ≥7 consecutive days of HAART and off HAART if they had not received HAART for >30 days.
HIV-1 RNA in plasma was quantified by reverse transcriptase polymerase chain reaction. CD4+ lymphocytes were quantified by flow cytometry. The median frequency of CD4+ lymphocyte count measurement was every 3.0 months [interquartile range (IQR): 1.9-4.0] over the entire study period, every 3.2 months (IQR: 2.2-5.5) when not on HAART, every 2.0 months (IQR: 1.1-3.7) in the first 180 days after HAART initiation, and every 3.0 months (IQR: 2.3-4.1) thereafter.
Fisher exact tests were used to compare categorical variables between categories. Wilcoxon Rank Sum and Kruskal-Wallis tests compared continuous variables between 2 or more than 2 categories, respectively. A marginal structural model was constructed for improved adjustment of CD4+ lymphocyte count, a time-dependent confounder affected by HAART exposure.26-28 The marginal structural model had 2 components as follows: (1) a model for predicting the probability of being on HAART, used to create inverse probability weights incorporated in the primary model; and (2) the primary model from which odds ratios for the risk of TB were computed. The model for predicting the probability of being on HAART for a given month included time since study entry (in months), whether the patient was on HAART during the preceding month, most recent CD4+ lymphocyte count, a variable indicating whether the patient had had a prior CD4+ lymphocyte count measurement, sex, race (black, non-black, unknown), age at study entry, foreign-born status (foreign born, not foreign born, unknown), and use of injection drugs. The primary model included time since study entry (in months) and current HAART status (not on HAART, ≤180 days of HAART, or >180 days of HAART). Results are reported with weights truncated at the first and 99th percentiles. All statistical tests were 2-sided. The P values <0.05 were considered statistically significant. Statistical analyses were performed using R version 2.11.1 (www.r-project.org). Analysis scripts are posted at http://biostat.mc.vanderbilt.edu/ArchivedAnalyses.
There were 4534 HIV-infected patients in care; median age at first visit was 38 years, 77% were male, 36% were black, 5% were foreign born, and 12% had injection drug use as an HIV risk factor. The median baseline CD4+ lymphocyte count was 320 cells per cubic millimeter (IQR: 154-513 cells/mm3), median baseline CD4 percentage 21% (IQR: 12%-30%), and median baseline HIV-1 RNA 15,982 copies per milliliter (IQR: 929-80,348 copies/mL). There were 3492 (77%) persons who initiated HAART during the study period, and the median CD4+ lymphocyte count before HAART initiation was 242 (IQR: 102-405). There were 1056 (23%) patients who received at least 1 screening TST; 63 (1.4%) patients received at least 6 months of isoniazid preventive therapy during the study period.
Thirty-four patients developed TB during the study period (165 per 100,000 person-years; 20,581 person-years of follow-up): 18 pulmonary cases, 12 extrapulmonary cases, and 4 cases of disseminated disease. Patients with TB were more likely to be male, black, foreign-born, and to report heterosexual contact as an HIV risk factor. They were also more likely to have a lower baseline CD4+ lymphocyte count/percentage and a higher baseline HIV-1 viral load. The frequency of CD4+ lymphocyte count measurements was similar between persons who developed and did not develop TB (Table 1).
Seventeen TB cases occurred among persons not on HAART or >30 days after HAART discontinuation (212 per 100,000 person-years; 8019 person-years of follow-up). Seventeen TB cases occurred among persons on HAART (135 per 100,000 person-years; 12,562 person-years of follow-up). Of these 17 TB cases, 10 occurred in the first 6 months of HAART (402 per 100,000 person-years; 2489 person-years of follow-up) and 7 occurred after more than 6 months of HAART (69 per 100,000 person-years; 10,073 person-years of follow-up) (Table 2).
Patients who developed TB while not on HAART, while on HAART ≤6 months, and while on HAART >6 months were similar with respect to age, race, HIV risk factor, baseline CD4+ lymphocyte count/percent, baseline HIV-1 viral load, and proportion with culture positivity (Table 2). Using a marginal structural model and adjusting for time-updated CD4+ lymphocyte count, age, sex, race, foreign born status, and injection drug use, the risk of TB after at least 180 days of HAART was 0.10 compared with persons not on HAART [95% confidence interval (CI): 0.03 to 0.31, P < 0.0001]. The risk of TB in the first 180 days after HAART initiation was 0.68 compared with persons not on HAART (95% CI: 0.14 to 3.22, P = 0.63). The risk of TB was higher in the first 180 days of HAART than after 180 days: odds ratio = 7.01 (95% CI: 1.35 to 36.31, P = 0.02) (Table 3). Results were similar when only adjusting for time-updated CD4+ lymphocyte counts (not shown).
The median time to TB diagnosis from time of HAART initiation, re-initiation, or regimen change was 144 days (IQR: 41-399 days). There were 7 cases of TB diagnosed within 3 months of HAART initiation, re-initiation, or regimen change. Five of these cases met criteria for unmasking TB-associated IRIS. The first case presented with cervical and supraclavicular lymphadenopathy and multiple hypodense ring-enhancing lesions in the chest and abdomen on computed tomography. The second case presented with bilateral supraclavicular lymphadenopathy and the third with multiple splenic lesions. The fourth case presented with splenomegaly and replacement of the spleen with soft tissue masses and right supraclavicular lymphadenopathy. The final case presented with head and neck lymphadenopathy.
Among these 5 provisional unmasking TB-associated IRIS cases, the median change in CD4+ lymphocyte count from the time of HAART initiation or re-initiation to TB diagnosis was +38 cells per cubic millimeter (IQR: 0 to +44), the median change in CD4+ lymphocyte percentage was +2% (IQR: 0% to +8%), the median change in HIV-1 viral load was −75,920 copies per milliliter (IQR: −268,123 to 0) (Table 4).
Of 34 TB cases, 9 (26%) had TST performed as routine screening before TB diagnosis. Of the 9 patients screened, 4 had a positive test, 3 of whom completed treatment for LTBI. Six (18%) of these patients had TST performed as routine screening before initiation of any HAART. None of the patients with unmasking TB-associated IRIS underwent screening for TB with TST before HAART initiation.
The TB rate during the first 180 days after HAART initiation (402 per 100,000 person-years) was higher than the TB rate in those persons not on HAART (212 per 100,000 person-years) and those on HAART more than 180 days (69 per 100,000 person-years). However, after incorporating HAART use as a time-updated exposure and adjusting for CD4+ lymphocyte count as a time-updated covariate, the TB risk was not significantly increased. In fact, the adjusted point estimate of TB risk suggests that HAART may even be protective in the first 180 days after initiation. Two previous studies utilizing the marginal structural model methodology assessed TB risk before and after HAART initiation but did not evaluate TB risk by duration of HAART exposure. Additionally, these cohorts were both in high TB incidence settings, and one was a cohort comprised only of children.23,24
Our results suggest that low CD4+ lymphocyte counts influence the high TB risk during the first 180 days after HAART initiation. This finding supports the earlier initiation of HAART at higher CD4+ lymphocyte counts to prevent TB disease.
Other causes of elevated TB risk during the period shortly after HAART initiation, such as new TB infection or previously undiagnosed TB are unlikely in this low incidence resource-rich setting of HIV patients in care. With this analysis, however, we cannot rule out unmasking TB-associated IRIS as a cause of elevated TB risk shortly after HAART initiation.
Data regarding the incidence of unmasking TB-associated IRIS in low TB incidence settings is unknown and exists mainly as case reports.17,29-31 We found that 5 (71%) of 7 TB cases diagnosed in the first 3 months after HAART initiation (or 15% of all TB cases in this cohort) met the definition for unmasking TB-associated IRIS. In this cohort, only 6 (18%) of 34 underwent screening for TB with a TST before initiation of any HAART. None of the 5 patients with unmasking TB-associated IRIS underwent screening for TB with TST before HAART initiation. It is possible that these cases could have been detected before HAART initiation had screening for TB been performed.
TB risk does decline with greater time on HAART, and in our study, the rate dropped to 69 per 100,000 person-years after 180 days on HAART. This corresponded to a 90% reduction in the risk of TB compared with not being on HAART after adjusting for CD4+ lymphocyte count. This is consistent with the risk reduction reported in previous studies,7-10 including one other low incidence cohort from the United States, which reported an 80% reduction in TB risk.9
The main limitation of our study is the low number of TB cases. This was a retrospective study and, therefore, some TB cases may have been missed. However, this is unlikely because TB is a reportable disease and persons were in care. This analysis should be repeated in a larger low-TB incidence cohort.
Additional limitations of our study are due to its retrospective observational nature. First, although we adjusted for CD4+ lymphocyte count, sex, race, age, foreign-born status, and intravenous drug use, the association between HAART use and TB could be confounded by other factors not included in our analysis.
Second, the frequency of CD4+ lymphocyte measurements could impact results. Specifically, when CD4+ lymphocyte counts are increasing/decreasing, the most recent value will tend to underestimate/overestimate its true value. This could lead to underestimation of the adjusted TB risk in the first 180 days after HAART initiation, presumably the time of most rapid change in CD4+ lymphocyte counts. However, CD4+ lymphocyte count measurements were frequent (median once every 3.0 months), and especially frequent during the first 180 days of HAART exposure (median once every 2.0 months), so we do not expect this potential bias to be great. It is also worth noting that the frequency of CD4+ lymphocyte count measurements was similar between those who developed and did not develop TB.
A final limitation lies in the completeness of documentation of HAART interruptions. If a patient was misclassified as taking HAART when they were not, this could lead to an overestimation of TB risk among those persons taking HAART. This type of bias should be limited by the extensive chart review for identification and documentation of interruptions.
In conclusion, crude TB risk in the first 6 months of HAART was higher than the risk in those persons not on HAART. However, this risk did not remain elevated after adjusting for the most recent CD4+ lymphocyte count and HAART exposure. This suggests that the elevated risk during this period is related at least in part to low recent CD4+ lymphocyte counts. These results emphasize the importance of screening for TB before the receipt of any HAART, and they underscore the need to maintain a clinical vigilance for TB, specifically shortly after HAART initiation. This is particularly true for HIV-infected persons at the highest risk for TB in a low incidence setting including the foreign born and those with low CD4+ lymphocyte counts. These results also support the earlier initiation of HAART at higher CD4+ lymphocyte counts to prevent TB disease.
2. Crowe SM, Carlin JB, Stewart KI, et al. Predictive value of CD4 lymphocyte numbers for the development of opportunistic infections and malignancies in HIV-infected persons. J Acquir Immune Defic Syndr
3. Guelar A, Gatell JM, Verdejo J, et al. A prospective study of the risk of tuberculosis among HIV-infected patients. AIDS
4. Hanson DL, Chu SY, Farizo KM, et al. Distribution of CD4+ T lymphocytes at diagnosis of acquired immunodeficiency syndrome-defining and other human immunodeficiency virus-related illnesses. The Adult and Adolescent Spectrum of HIV Disease Project Group. Arch Intern Med
5. Antonucci G, Girardi E, Raviglione MC, et al. Risk factors for tuberculosis in HIV-infected persons. A prospective cohort study. The Gruppo Italiano di Studio Tubercolosi e AIDS (GISTA). JAMA
6. Panel on Antiretroviral Guidelines for Adults and Adolescents. Guidelines for the use of antiretrovial agents in HIV-1-infected adults and adolescents. Department of Health and Human Services. January 10, 2011; 1-166. Available at: http://www.aidsinfo.nih.gov/ContentFiles/AdultandAdolescentGL.pdf
. Accessed January 26, 2011.
7. Badri M, Wilson D, Wood R. Effect of highly active antiretroviral therapy on incidence of tuberculosis in South Africa: a cohort study. Lancet
8. Santoro-Lopes G, de Pinho AM, Harrison LH, et al. Reduced risk of tuberculosis among Brazilian patients with advanced human immunodeficiency virus infection treated with highly active antiretroviral therapy. Clin Infect Dis
9. Jones JL, Hanson DL, Dworkin MS, et al. HIV-associated tuberculosis in the era of highly active antiretroviral therapy. The Adult/Adolescent Spectrum of HIV Disease Group. Int J Tuberc Lung Dis
10. Golub JE, Saraceni V, Cavalcante SC, et al. The impact of antiretroviral therapy and isoniazid preventive therapy on tuberculosis incidence in HIV-infected patients in Rio de Janeiro, Brazil. AIDS
11. Lawn SD, Myer L, Bekker LG, et al. Burden of tuberculosis in an antiretroviral treatment programme in sub-Saharan Africa: impact on treatment outcomes and implications for tuberculosis control. AIDS
12. Bonnet MM, Pinoges LL, Varaine FF, et al. Tuberculosis after HAART initiation in HIV-positive patients from five countries with a high tuberculosis burden. AIDS
13. Lawn SD, Badri M, Wood R. Tuberculosis among HIV-infected patients receiving HAART: long term incidence and risk factors in a South African cohort. AIDS
14. Seyler C, Toure S, Messou E, et al. Risk factors for active tuberculosis after antiretroviral treatment initiation in Abidjan. Am J Respir Crit Care Med
15. Moore D, Liechty C, Ekwaru P, et al. Prevalence, incidence and mortality associated with tuberculosis in HIV-infected patients initiating antiretroviral therapy in rural Uganda. AIDS
16. Girardi E, Sabin CA, d'Arminio MA, et al. Incidence of Tuberculosis among HIV-infected patients receiving highly active antiretroviral therapy in Europe and North America. Clin Infect Dis
17. Breen RA, Smith CJ, Cropley I, et al. Does immune reconstitution syndrome promote active tuberculosis in patients receiving highly active antiretroviral therapy? AIDS
18. Lawn SD, Wilkinson RJ, Lipman MC, et al. Immune reconstitution and “unmasking” of tuberculosis during antiretroviral therapy. Am J Respir Crit Care Med
19. Akolo C, Adetifa I, Shepperd S, et al. Treatment of latent tuberculosis infection in HIV infected persons. Cochrane Database Syst Rev
. 2010; CD000171.
20. Kaplan JE, Benson C, Holmes KH, et al. Guidelines for prevention and treatment of opportunistic infections in HIV-infected adults and adolescents: recommendations from CDC, the National Institutes of Health, and the HIV Medicine Association of the Infectious Diseases Society of America. MMWR Recomm Rep
21. Pettit AC, Barkanic G, Stinnette S, et al. Potentially preventable tuberculosis among HIV-infected persons in the era of highly active antiretroviral treatment. Int J Tuberc Lung Dis
22. Lee LM, Lobato MN, Buskin SE, et al. Low adherence to guidelines for preventing TB among persons with newly diagnosed HIV infection, United States. Int J Tuberc Lung Dis
23. Edmonds A, Lusiama J, Napravnik S, et al. Anti-retroviral therapy reduces incident tuberculosis in HIV-infected children. Int J Epidemiol
24. Fairall LR, Bachmann MO, Louwagie GM, et al. Effectiveness of antiretroviral treatment in a South African program: a cohort study. Arch Intern Med
25. 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
26. Hernan MA, Brumback B, Robins JM. Marginal structural models to estimate the joint causal effect of nonrandomized treatments. J Am Stat Assoc
27. Robins JM, Hernan MA, Brumback B. Marginal structural models and causal inference in epidemiology. Epidemiology
28. Sterne JA, Hernan MA, Ledergerber B, et al. Long-term effectiveness of potent antiretroviral therapy in preventing AIDS and death: a prospective cohort study. Lancet
29. Girardi E, Antonucci G, Vanacore P, et al. Tuberculosis in HIV-infected persons in the context of wide availability of highly active antiretroviral therapy. Eur Respir J
30. Girardi E, Antonucci G, Vanacore P, et al. Impact of combination antiretroviral therapy on the risk of tuberculosis among persons with HIV infection. AIDS
31. Goldsack NR, Allen S, Lipman MC. Adult respiratory distress syndrome as a severe immune reconstitution disease following the commencement of highly active antiretroviral therapy. Sex Transm Infect
© 2011 Lippincott Williams & Wilkins, Inc.