Latent tuberculosis infection (LTBI) is an asymptomatic condition after inhalation of aerosolized Mycobacterium tuberculosis organisms.1 Although a proportion of individuals progress immediately to active disease, a partially effective host immune response results in containment of the bacilli within pulmonary granulomata.2 In this state, viable bacilli can remain dormant for decades before reactivation. It is commonly quoted that acquisition of latent TB is associated with a 10% lifetime risk of active disease, with half of the subsequent infections occurring in the first 2 years after exposure.3 However, immunosuppressive conditions, particularly HIV infection, markedly increase the risk of progression, whereas effective anti-TB therapy reduces its likelihood substantially.4
Because of the increased risk of active infection, routine testing and treatment for LTBI are recommended in HIV-infected individuals, a practice which has been shown to reduce subsequent active TB and mortality in high TB prevalence settings.5–7 Testing for LTBI may include the use of tuberculin skin tests (TST) or whole-blood interferon-gamma release assays (IGRAs). Although both tests are in active clinical usage internationally, the predictive power of each is limited by the absence of a gold standard diagnostic test and limited comparative assessment in the setting of HIV infection.8,9 Limitations of the TST, which measures response to tuberculin purified protein derivative, include cross-reactivity with nontuberculous mycobacteria and bacille Calmette–Guerin,10 poor sensitivity among HIV-infected patients,11,12 and requirements for skilled reading of skin induration on subsequent patient visits by health-care workers.13 In some high-income countries, convenience, control for impaired T-cell response, and potentially improved detection of latent TB have led to IGRA being used for LTBI testing.14–16 International guidelines recommend anti-TB therapy for any HIV-infected patient with a positive screening IGRA or TST.17 However, existing LTBI testing recommendations and evidence for LTBI treatment are based predominantly on TST results and meta-analyses establishing clinical benefits (including reduced incident TB) when treating HIV-infected individuals with a reactive TST.18 Furthermore, there is limited evidence from high-income, low TB prevalence countries on the clinical utility of IGRA use in HIV infection.6,17 In Victoria, Australia, the incidence of active TB was 7 cases per 100,000 person-years (PY) in 2008 with >80% of cases occurring among overseas-born individuals.19,20
This study aimed to describe the frequency and predictors of latent TB assessed by the Quantiferon-TB Gold (QFT-G) (Cellestis, Melbourne, Australia) and how its use influenced clinical management among HIV-infected individuals in a high-income, low TB prevalence country. It further aimed to describe the risk and predictors of active TB development in a well-controlled Australian HIV-infected cohort.
Study Design and Participants
A retrospective cohort study was conducted among HIV-infected patients undergoing latent TB screening at a large urban sexual health service in Melbourne, Australia. All HIV-infected patients who attended the Melbourne Sexual Health Centre between March 2003 and March 2011, and had at least one QFT-G performed, were included in the cohort. They were followed up to their most recent HIV clinic visit where HIV monitoring blood tests were performed. The QFT-G is 1 of 2 commercially available IGRAs and was introduced into this health service in 2003. Since 2008, HIV-infected patients routinely had one QFT-G test performed free of charge close to their first visit at the HIV service.
All patients had demographic and risk factor information, including age, gender, country of birth, and indigenous status extracted from their electronic registration records. For patients with positive QFT-G results, medical records were reviewed for clinical information regarding potential TB exposures and past TB treatment, current symptoms, chest radiograph results, and subsequent anti-TB therapy. Active TB cases were defined by microbiological evidence (culture and/or polymerase chain reaction) of M. tuberculosis and/or health department case notification. Diagnosis and management information of active TB cases were obtained from clinic records and from reference laboratory surveillance.
Laboratory results for the cohort were collected from Victorian Infectious Diseases Reference Laboratory records, including QFT-G test–specific data (ESAT-10 or CFP-10 antigens, mitogen, and negative control value) and HIV confirmation testing and monitoring including HIV viral load and CD4 lymphocyte cell count. The same laboratory was used for all HIV monitoring and TB testing during the investigation period. Victorian Infectious Diseases Reference Laboratory also acts as the statewide reference laboratory for TB diagnosis, receiving all new TB isolates for drug susceptibility testing and mycobacterial interspersed repetitive unit/variable number of tandem repeats genetic typing, which should ensure complete detection of microbiologically confirmed active TB cases within this cohort.
A positive QFT-G assay was defined according to the manufacturer's instructions with either ESAT-10 or CFP-10 antigens (≥0.35 IU/mL) in the presence of a mitogen response (≥0.50 IU/mL) (corrected for negative control). Indeterminate QFT-G results were classified as either low mitogen responses (<0.50 IU/mL) or high negative controls (≥2.0 IU/mL). Results with high normal ESAT-10 or CFP-10 antigens (0.30–0.35 IU/mL) were retested.17 Where the first QFT-G result was indeterminate or patients had multiple testing, the first subsequent definitive QFT-G test result was used in the analysis instead. Approval was obtained from our institution's human research ethics committee.
Baseline demographic variables (age, gender, and country of birth) and clinical variables (CD4 count and HIV viral load) were tested for potential associations with positive QFT-G result using logistic regression. A second analysis added QFT-G to these predictors to examine active TB development over time using Cox proportional hazard models. The active TB analysis removed individuals who had previously received a TB diagnosis or received any TB (active or latent) treatment. Univariable and multivariable models for active TB estimated crude and adjusted hazard ratios (HR) with 95% confidence intervals (CIs) using these prespecified predictors in the final multivariable model. Proportionality assumptions were checked graphically and statistically using Schoenfeld residual estimation. All statistical tests were undertaken using STATA (Version 11.2; StataCorp, College Station, Tex).
There were 917 HIV-infected patients seen at least once with a QFT-G test performed at our center over 8 years. The cohort had a mean age of 41 years (range 18–84), with largely well-controlled HIV with median CD4 T-cell count of 490 cells per microliter [interquartile range (IQR) 348–676] and 64% had HIV RNA completely suppressed at the time of their QFT-G test. The majority of the cohort was Australian or New Zealand born (63%) and known of nonindigenous background (97.8%) (Table 1). The median length of time between the first QFT-G test and last follow-up was 26.4 (IQR 15.3–30.7) months, generating a total of 1796 PY of follow-up for active TB cases. There were no differences in CD4 count or viral load between those who received a QFT-G test and those who did not (Table 1). However, the proportion born overseas was high among those undertaking QFT-G compared with no QFT-G (37% vs. 28%, difference: 9.0%, 95% CI: 3.8 to 14.2, P < 0.001).
Twenty-nine (3.2%) patients had positive QFT-G results, whereas 12 (1.3%) had an initially indeterminate QFT-G test because of high negative controls (n = 8) or low mitogen controls (n = 4). Eight (0.9%) patients were retested given high normal TB antigens, of which 7 were definitively QFT-G negative and 1 QFT-G positive. The mean CD4 count in those with initially indeterminate QFT-G test was 496 cells per microliter (IQR 363–793) with no difference from those with definitively positive or negative results (Wilcoxon rank sum, P = 0.6). Ultimately, only 4 (0.4%) QFT-G results remained indeterminate—3 because of high negative controls and 1 low mitogen control—and were excluded from later logistic regression analyses. There was no difference in the proportion of patients testing QFT-G positive between CD4 cell counts greater than or less than 200 cells per microliter: 3.2% vs. 2.8%, respectively (risk difference: −0.3%, 95% CI: −4.4% to 3.7%, P = 0.8).
Participants born overseas were more likely to have a positive QFT-G result compared with those born in Australia or New Zealand [5.3% vs. 2.0%, odds ratio (OR): 2.6, 95% CI: 1.2 to 5.6, P = 0.017] (Table 2). Being born in Africa was the strongest predictor of being QFT-G positive (12.7% vs. 2.0%, OR: 7.1, 95% CI: 2.90 to 17.3, P < 0.001). There was no significant difference in the QFT-G result by age, gender, CD4 cell count, HIV viremia, or indigenous status; so, no adjustment was made for these prespecified factors.
After medical record review of the management of all 29 positive QFT-G tests, it was found that 6 patients had known previous TB exposure (4/6 born overseas) and 5 had previously been treated for LTBI (3/5 born overseas) (Table 3). In retrospect, 1 patient born overseas had respiratory symptoms of active TB at the time of QFT-G testing. Of the 24 patients without previous TB or LTBI treatment, 21 underwent chest radiograph and 19 later commenced anti-TB treatment (13/19 born overseas) with median age of 38 years (range 24–52). Four patients, aged 40–62 years of whom 1 was born overseas and the other 2 had no epidemiology of TB exposure, did not receive isoniazid therapy, and toxicity concerns were recorded as the main reason. One patient had been lost to follow-up after QFT-G testing.
There were 2 cases of active TB diagnosed after the QFT-G test, at an incidence rate of 111.4 (95% CI: 27.8 to 445) per 100,000 PY in this context. Both patients were diagnosed within 2 years of QFT-G screening, were born in sub-Saharan Africa, and had immigrated to Australia 3 months and 2 years before, respectively, and neither had received isoniazid therapy. Extended mycobacterial interspersed repetitive unit/variable number of tandem repeats genotyping to 21 loci showed that neither strain of TB had been previously diagnosed in Victoria. This suggests that both TB exposures occurred overseas in keeping with their known epidemiological links and did represent TB reactivation rather than a new locally acquired infection.21 QFT-G positivity was the strongest predictor of active TB in both univariable and multivariable analyses adjusted for CD4 cell count (adjusted HR: 42.4, 95% CI: 2.2 to 827, P = 0.013). There was a trend toward lower CD4 cell count increasing risk of active TB, most notable when CD4 cell count <200 cells per microliter (adjusted HR: 25.4, 95% CI: 0.70 to 918, P = 0.08). HIV-infected patients born in Africa were more likely to develop active TB than those born in Australia or New Zealand (incidence rate: 1202 vs. 0 per 100,000 PY; incidence rate ratio indeterminate, 95% CI: 1.87 to infinite; Exact test P = 0.009). Age and ethnicity were not shown to be predictive of active TB development based on these limited number of cases (Table 4). As a screening test in our low TB prevalence population, the positive predictive value for active TB during the 2-year median follow-up period was 3.4% (95% CI: 0.1% to 17.8%). The negative predictive value over the same period was 99.9% (95% CI: 99.4% to 99.9%). The number needed to screen to detect 1 case of active TB was 913.22
In a sensitivity analysis, variables used in the multivariable analyses were removed sequentially to examine for colinear effects on the data and no other new predictor variables achieved significance at the P <0.05 level. There was no difference in the model whether independent variables (age, CD4 count, and HIV viral load) were used as binary or continuous variables; however, the binary predictors are presented for ease of interpretation.
Our study is one of the largest cohort studies performed on QFT-G testing in HIV-infected patients in high-income countries and the only published report on QFT-G performance in this context from Australia. The results demonstrate that country of birth is a strong predictor of QFT-G positivity and that QFT-G positivity is a very strong relative predictor of active TB development. Furthermore, the QFT-G assay can be performed with very low indeterminate results even in the setting of HIV and immunodeficiency. Despite the overall good immunological and virological control of HIV across our cohort, we also confirmed an increased rate of TB among people with HIV infection in this setting: 111 per 100,000 PY (95% CI: 13.2 to 395 per 100,000 PY) in our cohort compared with the population rate of 7 per 100,000 PY.19 However, the positive predictive value of the QFT-G assay for development of active TB is limited by the low prevalence—by global standards—of TB in Australia.
Whereas other studies from high-income countries have examined the role of QFT-G in testing HIV-infected patients for symptomatic TB23,24 or used an alternative IGRA,25 only one article has similarly used QFT-G to screen asymptomatic patients for latent or active TB. The study by Aichelburg et al13 from Austria calculated the predictive value of QFT-G in HIV infection and included 830 patients followed for a median of 19 months. They found an increased odds of positive QFT-G results among patients born in Africa compared with Austria (OR: 6.6, 95% CI: 2.9 to 14.25, P < 0.001) or born in countries of high vs. low TB burden (>99 cases vs. <25 cases per 100,000 population, OR: 5.9, 95% CI: 2.4% to 13.4%).13 Only 3 cases of active TB developed during follow-up, all among those who tested QFT-G positive compared with QFT-G negative (8% vs. 0%, risk difference: 8%, 95% CI: −0.7% to 17%, P < 0.001). In our study and the study by Aichelburg et al,13 country of birth is probably a proxy for true exposure to TB, supported by the observation in both cohorts that LTBI risk increases across countries of increasing TB prevalence.
Although the QFT-G was widely used and accepted in this clinical setting as a screening test because of its convenience and reproducibility, we observed that the QFT-G result did not always change clinical decision making. That is, in a minority of cases, a clinical decision was made that positive QFT-G results would not be followed by anti-TB chemotherapy. Despite the relative odds of active TB among QFT-G positive patients, routine use of QFT-G in every HIV-infected patient in our context leads to tests being performed on some patients with low likelihood of true TB exposure. A targeted testing process, first involving a clinical assessment of likely TB exposure–based epidemiological risk factors, followed by a QFT-G test for those patients born abroad or with other potential TB exposures, could improve the clinical usefulness of the QFT-G test. Improving the pretest probability would increase the positive predictive value and may assist clinicians in diagnosing LTBI and then better weigh the benefits and toxicities of chemoprophylaxis. A targeted testing strategy would also reduce the number needed to screen to predict one future active TB case.
One recent cost-effectiveness study of TB control in the United States showed that IGRA screening costs more than not screening but was cost saving over the TST among HIV-infected patients, with an incremental cost-effectiveness ratio of IGRA compared with TST of <US $100,000 per quality adjusted life year gained.26 Given the public health importance of TB in the setting of HIV infection, we suggest that a detailed cost-effectiveness evaluation comparing universal and targeted LTBI screening strategies is now warranted.
These results also demonstrated a very low proportion (0.4%) of unresolved indeterminate QFT-G test results among HIV-infected patients, compared with previously published reports. One meta-analysis synthesized indeterminate rates from 7 studies from high-income settings using QFT-G to detect LTBI and calculated a pooled proportion of indeterminate QFT-G results at 4% (95% CI: 3% to 6%) among HIV-infected patients.12 Another recent meta-analysis included 3 studies from high-income countries using QFT-G to diagnose, rather than predict, active TB and found a pooled proportion of indeterminate results of 8.4% (95% CI: 6.8% to 10.2%).27 We also did not observe a decline in the proportion of patients testing QFT-G positive at lower CD4 T-cell counts, which had been previously observed in the HIV context.28 Two explanations might account for our findings. First, our HIV-infected cohort was generally well controlled on effective treatment. Only 8% of patients had CD4 <200 cells per microliter and 24% had CD4 <350 cells per microliter, and only 1 of the 4 tests that remained indeterminate after repeating was because of a low mitogen response. However, the small sample with low CD4 T-cell counts limits the ability to make observations about test performance in advanced immunodeficiency. Second, the QFT-G assays were all performed at the same reference laboratory soon after collection in a well-controlled setting, reducing processing delays, handling errors, and enhancing reproducibility. Given the shortcomings of the conventional TST among HIV-infected patients, these findings provide support for the use of QFT-G among HIV-infected patients in well-resourced settings.
This study had several limitations. Although a large cohort was screened for LTBI using the QFT-G, active TB cases were rare, limiting our study power. We are not able to make robust estimates of test parameters like sensitivity or the number of tests required to detect one active TB case. We used a retrospective cohort study design to deliberately capture all patients who had been routinely QFT-G tested and provide a practical lengthy follow-up period. But the low TB incidence would require either a substantially larger cohort or a longer period of observation to find more TB cases in Australia.
Although selection biases are possible in any retrospective study, any censoring by physicians in this cohort seems to favor selecting patients of perceived higher TB risk, based on the slightly higher proportion of non–Australian-born individuals receiving QFT-G testing. Importantly, CD4 count and viral load suppression were comparable between QFT-G participants and nonparticipants, and there were no cases of active TB development subsequently among the non-QFT-G testing group. Any clinician censoring therefore is more likely to lower further our estimates of QFT-G positivity and active TB risk.
There was a potential for active TB cases to have been underestimated if patients were diagnosed at a different health service after their initial QFT-G test. For this reason, we captured active TB case report information from the statewide Mycobacterium Reference Laboratory. Although any interstate or overseas TB diagnosis could still occur and underestimate the rate of active TB, in our experience, it would be uncommon. Our TB incidence is thus reported cautiously. However, because of privacy considerations limiting data linkage by statutory health protection agencies in our jurisdiction, this is the first local estimate of incident TB among HIV-infected patients.
The retrospective nature of the study also limits our ability to collect epidemiological information on LTBI risk factors or clinical management decisions given our reliance on contemporaneous medical records. Our health clinic database helps overcome this by consistently collected demographic data with complete country of birth data in 96% of the cohort. However, a prospective design would be needed to obtain more detailed survey information on travel, occupational, or other exposure risks.
This study demonstrated that the QFT-G assay is a strong predictor of active TB development in HIV-infected patients and in these practical conditions produced few indeterminate results. However, active TB is rare in the setting of well-controlled HIV and low population TB prevalence making its negative predictive value most helpful. This raises questions about the clinical utility of routine QFT-G testing in all HIV-infected patients born in low TB prevalence countries. A targeted, rather than routine, testing approach based on epidemiological risk factors for TB exposure may better assist clinicians to predict active TB development.
The authors wish to thank the database managers at Melbourne Sexual Health Centre (Afrizal Afrizal and Jun Kit Sze) and laboratory staff at the Victorian Infectious Diseases Reference Laboratory for their assistance in data collection. J.S.D. receives a National Health and Medical Research Council research fellowship and acknowledges the contribution of the Victorian Operational Infrastructure Support Program to the Burnet Institute.
1. Jasmer RM, Nahid P, Hopewell PC. Clinical practice. Latent tuberculosis infection. N Engl J Med. 2002;347:1860–1866.
2. Ulrichs T, Kaufmann SH. New insights into the function of granulomas in human tuberculosis. J Pathol. 2006;208:261–269.
3. Comstock GW, Livesay VT, Woolpert SF. The prognosis of a positive tuberculin reaction in childhood and adolescence. Am J Epidemiol. 1974;99:131–138.
4. Corbett EL, Watt CJ, Walker N, et al.. The growing burden of tuberculosis: global trends and interactions with the HIV epidemic. Arch Intern Med. 2003;163:1009–1021.
5. Gordin FM, Matts JP, Miller C, et al.. A controlled trial of isoniazid in persons with anergy and human immunodeficiency virus infection who are at high risk for tuberculosis. N Engl J Med. 1997;337:315–320.
6. Pozniak AL, Coyne KM, Miller RF, et al.. British HIV Association guidelines for the treatment of TB/HIV coinfection 2011. HIV Med. 2011;12:517–524.
7. Martinson NA, Barnes GL, Moulton LH, et al.. New regimens to prevent tuberculosis in adults with HIV infection. N Engl J Med. 2011;365:11–20.
8. Pai M, Zwerling A, Menzies D. Systematic review: T-cell–based assays for the diagnosis of latent tuberculosis infection: an update. Ann Intern Med. 2008;149:177.
9. Mandalakas A, Hesseling A, Chegou N, et al.. High level of discordant IGRA results in HIV-infected adults and children. Int J Tuberc Lung Dis. 2008;12:417–423.
10. Farhat M, Greenaway C, Pai M, et al.. False-positive tuberculin skin tests: what is the absolute effect of BCG and non-tuberculous mycobacteria? [Review Article]. Int J Tuberc Lung Dis. 2006;10:1192–1204.
11. Huebner RE, Schein MF, Hall CA, et al.. Delayed-type Hypersensitivity anergy in human immunodeficiency virus-infected persons screened for infection with Mycobacterium tuberculosis
. Clin Infect Dis. 1994;19:26–32.
12. Cattamanchi A, Smith R, Steingart KR, et al.. Interferon-gamma release assays for the diagnosis of latent tuberculosis infection in HIV-infected individuals: a systematic review and meta-analysis. J Acquir Immune Defic Syndr. 2011;56:230–238.
13. Aichelburg MC, Rieger A, Breitenecker F, et al.. Detection and prediction of active tuberculosis disease by a whole-blood interferon-gamma release assay in HIV-1-infected individuals. Clin Infect Dis. 2009;48:954–962.
14. Stephan C, Wolf T, Goetsch U, et al.. Comparing QuantiFERON-tuberculosis gold, T-SPOT tuberculosis and tuberculin skin test in HIV-infected individuals from a low prevalence tuberculosis country. AIDS. 2008;22:2471–2479.
15. Richeldi L, Losi M, D'Amico R, et al.. Performance of tests for latent tuberculosis in different groups of immunocompromised patients. Chest. 2009;136:198–204.
16. Dewan PK, Grinsdale J, Liska S, et al.. Feasibility, acceptability, and cost of tuberculosis testing by whole-blood interferon-gamma assay. BMC Infect Dis. 2006;6:47.
17. Mazurek GH, Jereb J, Vernon A, et al.. Updated guidelines for using interferon gamma release assays to detect Mycobacterium tuberculosis
infection—United States, 2010. MMWR Recomm Rep. 2010;59:1–25.
18. Akolo C, Adetifa I, Shepperd S, et al.. Treatment of latent tuberculosis infection in HIV infected persons. Cochrane Database Syst Rev. 2010:CD000171.
19. Communicable Disease Prevention and Control Unit. Mycobacterial infections. In: Rowe S, Melbourne HN. Surveillance of Notifiable Infectious Diseases in Victoria. Victoria, Australia: Department of Health; 2008. pp. 31–35.
20. Brown L. Mycobacterial infections. Victorian Infect Dis Bull. 2011;14:64–65.
21. Ashworth M, Horan KL, Freeman R, et al.. Use of PCR-based Mycobacterium tuberculosis genotyping to prioritize tuberculosis outbreak control activities. J Clin Microbiol. 2008;46:856–862.
22. Rembold CM. Number needed to screen: development of a statistic for disease screening. Br Med J. 1998;317:307–312.
23. Garcia-Gasalla M, Fernandez-Baca V, Mir-Viladrich I, et al.. Quantiferon-TB Gold In-Tube test in the diagnosis of pulmonary and extra-pulmonary tuberculosis[in Spanish]. Enferm Infecc Microbiol Clin. 2010;28:685–689.
24. Sauzullo I, Mengoni F, Scrivo R, et al.. Evaluation of QuantiFERON-TB Gold In-Tube in human immunodeficiency virus infection and in patient candidates for anti-tumour necrosis factor-alpha treatment. Int J Tuberc Lung Dis. 2010;14:834–840.
25. Clark SA, Martin SL, Pozniak A, et al.. Tuberculosis antigen-specific immune responses can be detected using enzyme-linked immunospot technology in human immunodeficiency virus (HIV)-1 patients with advanced disease. Clin Exp Immunol. 2007;150:238–244.
26. Linas BP, Wong AY, Freedberg KA, et al.. Priorities for screening and treatment of latent tuberculosis infection in the United States. Am J Respir Crit Care Med. 2011.
27. Chen J, Zhang R, Wang J, et al.. Interferon-gamma release assays for the diagnosis of active tuberculosis in HIV-infected patients: a systematic review and meta-analysis. PLoS One. 2011;6:e26827.
28. Aabye MG, Ravn P, PrayGod G, et al.. The Impact of HIV infection and CD4 cell count on the performance of an interferon gamma release assay in patients with pulmonary tuberculosis. PLoS One. 2009;4:e4220.