Patients with HIV infection and a positive Mantoux test (tuberculosis skin test, TST) have an increased risk of progression from latent to active tuberculosis (TB) [1,2]. However, the TST as a method to determine latent TB infection is known to have major restraints. False positive results occur because of antigenic cross-reactivity of purified protein derivative with non-Mycobacterium tuberculosis infections, including Bacille Calmette–Guerin (BCG) vaccination [3,4]. False negative, that is, anergic, test reactions are frequently observed in HIV-coinfected patients . Further limitations include the difficulty in correct application and the requirement for a return visit.
Recently, two blood tests for the diagnosis of latent TB infection have become available. Both are based on the specific interferon-γ release of activated T-cells that were incubated ex-vivo with the M. tuberculosis antigens Early Secretory Antigen Target-6 (ESAT-6) and Culture Filtrate Protein 10 (CFP10). Whereas QuantiFERON-TB Gold (QFG) determines the level of soluble interferon-γ produced in 1 ml whole blood by enzyme-linked immunosorbent assay (ELISA), the T-SPOT.TB (ELISPOT) assay detects the number of interferon-γ-producing cells represented as spot forming units.
Published data on the comparative performance of the new T-cell interferon-γ-releasing assays (QFG, ELISPOT) in asymptomatic HIV-positive patients are limited. One study from San Francisco, USA compared QFG and TST in 294 HIV-infected individuals. The overall concordance was high (89.3%), but regarding 29 patients with any positive test result, only eight were positive by both modalities . Another study from Italy compared TST with both QFG and T-SPOT.TB in a TB reference clinic, but only a minority of patients were immunosuppressed for different reasons (e.g. cancer, chemotherapy, organ transplantation and HIV in 149 out of 393 patients) and the HIV-infected subgroup was not further characterized by CD4 cell counts .
The purpose of this study is to describe the prevalence and the predictors of a positive result for each of the three tests and to assess the agreement between tests in HIV-positive patients without clinical signs of active disease in the HAART era.
Patients and methods
Participants were recruited on a voluntary basis from our HIV outpatient department between 1 January 2006 and 31 July 2007. HIV-positive adults who were willing to undergo diagnostic procedures were enrolled in the study. All patients provided written and informed consent. Exclusion criteria were a previous severe reaction to TST, currently suspected TB, active AIDS-defining or other compromising disease. The University ethics committee approved the study.
A standardized patient questionnaire on the patients' TB history was completed with the assistance of a study nurse when necessary. In addition, prespecified risk factors such as recent close contact to someone with active pulmonary TB, social welfare status, nicotine and alcohol use, previous BCG vaccination as well as origin from a high TB prevalence country were collected. We assumed high TB prevalence in all countries outside the USA/Canada, Western Europe, Scandinavia, Germany, Austria, Switzerland, Japan and the Western Pacific Region.
Patient characteristics collected by chart review included sex, age, first available positive HIV test, antiretroviral treatment history, HIV category according to CDC (Centers for Disease Control and Prevention). The CD4 cell count and HIV RNA load were collected at time of study inclusion. In case of missing value for CD4 cell count and HIV RNA, most recent available results were accepted if they were collected within 12 weeks from the date of study inclusion.
Peripheral blood samples for both tests were obtained simultaneously and processed by laboratory staff unaware of the corresponding blood and TST result and the TB history (Table 1) . Blood tests were performed with commercially available versions of T-SPOT.TB and QFG in two separate laboratories, according to the manufacturer's instructions. Each test included a negative control well (no mitogen or antigens), a positive control well (phytohaemagglutinin) and two antigen wells: ESAT-6 and CFP10.
Categorical data were compared using Pearson's χ 2 test or Fisher's exact test when appropriate. The Mann–Whitney test was performed to determine whether the distribution of continuous variables differed between independent groups. Proportions of indeterminate and positive results for both blood tests were compared using the Wilcoxon test. Concordance between TST, QFG and T-SPOT.TB was assessed by kappa (κ) statistics. Strength of agreement was considered ‘poor’ for κ ≤ 0.20, ‘fair’ for 0.20 < κ ≤ 0.40, ‘moderate’ for 0.40 < κ ≤ 0.60, ‘substantial’ for 0.60 < κ ≤ 0.80 and ‘optimal’ for 0.80 < κ ≤ 1.00 . Logistic regression analysis was used to identify factors associated with positive results for TST, QFG and T-SPOT.TB. For correlation calculations, Spearman's rank correlation coefficient rho (ρ) was used. All statistical analyses were two-sided and considered significant in case of P value less than 0.05 . Statistical analysis was done with SPSS v 13.0 (SPSS Inc, Chicago, Illinois, USA).
In total, 286 participants were enrolled. The median age was 44 years (range 22–75), and 231 participants (81%) were men. The majority of patients were of white heritage (85%) and 62 (22%) originated from a country with a high TB prevalence. There were 29 participants (10.1%) with a history of active TB disease and 80 participants (30.7%) with a history of a prior AIDS-defining event.
The median CD4 cell count of all participants was 408/μl (range 7–1510); 48 patients (16.8%) were naive to any antiretroviral treatment. Among treated patients, 56 (19.6%) had detectable viral loads or were in a therapy interruption; 182 patients (63.6%) had undetectable HIV-1 RNA (<50 copies/ml). Table 2 shows the baseline characteristics of all patients and the participants with positive results in the three corresponding tests. During the patient enrolment period, 831 patients followed at the same department were not included in the study, mainly because they rejected a return visit for TST. They did not differ significantly from study patients with respect to median age (43.7 years; P = 0.2) and CD4 cell counts (407/μl, range 1–1681; P = 0.4).
Tuberculin skin test
A TST result was available in 275 out of 286 patients; TST was read within 48–72 h ideally but in some cases up to 168 h from application. Patients presenting before 48 h or after more than 168 h were excluded from this analysis as recommended . Of all participants with a valid TST result, 12% (n = 33) showed an induration of at least 5 mm. Among these, 20 had an induration of at least 10 mm. A statistically significant association was found between a positive TST result and the following risk factors: origin from a high prevalence TB country, history of AIDS events and prior active TB. In logistic regression analysis, patients with a history of prior active TB were at six-fold increased probability for a positive TST result [odds ratio (OR) 5.9, 95% confidence interval (CI): 2.1–16.2].
In only 19 patients, a prior BCG vaccination had been conducted with certainty. Because of this low case number (6.64%), this variable was not considered for further analysis.
In 250 patients with determinate results in all three tests, concerning patient groups with a positive or negative TST result, there was no significant difference in terms of median CD4 cell counts (TST+ 396/μl vs. TST− 409/μl; P = 0.817) as well as for CD4 nadir count (TST+ 97/μl vs. TST− 137/μl, P = 0.150).
A result for QuantiFERON-TB Gold (QFG) test was available in 275 out of 286 patients. Out of these, 52 (18.9%) were positive and 222 (80.7%) negative; only one (0.4%) had an indeterminate result. A positive QFG result was not statistically associated with any of the risk factors studied. However, a significant difference (P = 0.044) was found for median CD4 cell count between patients with a positive (457 cells/μl) and those with a negative QFG result (405 cells/μl), suggesting a dependence of positive QFG results on the CD4 cell count. This finding was confirmed by a significant correlation between CD4 cell count and the concentration of specific, soluble interferon-γ in cell culture supernatant as measured by the QFG assay (Spearman's correlation coefficient ρ = 0.199; P = 0.002). No significant association was detectable between CD4 nadir cell counts and QFG results.
For T-SPOT.TB test, a valid result was available in 275 out of 286 patients; among these, 66 (24%) patients had a positive, 201 (73.1%) a negative and eight (2.9%) an indeterminate result. Both positive and indeterminate test results were analyzed for an association with any of the risk factors studied. The only significant association was found between prior active TB and a positive T-SPOT.TB (P = 0.01). Logistic regression revealed a three-fold increased probability for a positive T-SPOT.TB result (OR = 3.1, 95% CI: 1.2–7.6, P = 0.015). Median CD4 cell count did not differ between patients with a positive (398 cells/μl) or negative T-SPOT.TB result (411 cells/μl; P = 0.343). No association was found between median CD4 nadir count and T-SPOT.TB (138 cells/μl for positive results, 120 cells/μl for negative results; P = 0.522), as well as between CD4 cell count and the number of spot-forming units, reactive to ESAT-6 (ρ = 0.062; P = 0.31) and CFP10 (ρ = −0.023; P = 0.71), respectively.
Agreement between three tests
Compared with TST, significantly more positive results were observed for both QFG (P = 0.008) and T-SPOT.TB (P < 0.001). There was no statistically significant difference between both blood tests (P = 0.133). Figure 1 illustrates the distribution of TST results stratified according to the results of the interferon-γ assay. The majority of patients had a negative TST, irrespective of the results in either of the blood assays. The agreement between TST and QFG (κ = 0.335), as well as between TST and T-SPOT.TB (κ = 0.201) was fair.
Figure 2 shows the study patient distribution subject to the result of T-SPOT.TB, according to the QFG result and vice versa. For each test result stratum, the majority of patients were negative for the other test, reflecting a poor agreement between the two interferon-γ-releasing assays (κ = 0.146).
Of all patients with a valid result in both blood tests (n = 268), the number of indeterminate results was one (0.4%) for QFG, but significantly more for T-SPOT.TB (n = 8, 3.0%; P < 0.01). Seven of the indeterminates in T-SPOT.TB tested negative in QFG, the remaining patient was indeterminate in QFG (Fig. 2). These eight could not be characterized in terms of origin from a TB-high prevalence country, history of active TB, intravenous drug abuse, CD4 cell count, sex, age, HIV-1 RNA replication status (<50 copies/ml), dependence on social welfare, CDC-category C or alcohol/nicotine abuse.
Patients with a history of active tuberculosis
Twenty-nine patients with a history of prior active TB did not differ significantly from those without such history in regard to age (median 40 years, range 22–61) or sex distribution (men, n = 22; 76%). Table 3 illustrates the test results for these patients with respect to site of previous TB infection. For all patients with available positive TST result (n = 11 out of 28; 39.3%), the median skin induration was 12 mm (range 8–18).
Among 26 patients with all three test results available, a positive result was found in 12 for T-SPOT.TB compared with 10 for TST and seven for ELISA, respectively. A substantial subset (n = 11 or 42.3%) was negative to all three tests. Agreement between TST and each interferon-γ-releasing assay was moderate (κ = 0.570, P = 0.003 between QFT and TST; κ = 0.530, P = 0.006 between T-SPOT.TB and TST); the concordance between QFG and T-SPOT.TB in this patient subset was poor (κ = 0.123, P = 0.50).
This cross-sectional analysis compares TST with the two commercially available interferon-γ-releasing assays for the detection of latent TB infection in representative 286 HIV-positive individuals from Germany, that is, a low TB prevalence country. For 250 patients with determinate results, both assays showed significantly more positive results (QFG 20.0% and T-SPOT.TB 25.2%) than TST alone (12.8%), reflecting a better sensitivity of interferon-γ-releasing assays.
Both interferon-γ-releasing assays have been demonstrated to be highly specific to detect latent TB infection in immunocompetent individuals. Based on case–control studies, the specificity of T-SPOT.TB was nearly 100% [12–15] and similarly high (96–98%) for QFG [16,17], whereas the estimated sensitivity was reported to be about 78% for QFG and 90% for T-SPOT.TB [18,19]. Considering 98 patients with at least one positive result in either of interferon-γ assays, the prevalence of latent TB infection should be at least 36.6%. This suggests a low test sensitivity of both interferon-γ assays in HIV patients: for QFG less than 53% (52/98) and for T-SPOT.TB less than 67% (66/98).
Studies comparing TST and QFG in asymptomatic, predominantly HIV-negative adults, revealed an excellent (κ = 0.866; 94% ) or good agreement (κ = 0.61; 81.4% ). However, one study of HIV-infected individuals found a moderate test concordance between TST and QFG (κ = 0.370) , which was similar to our result (κ = 0.335).
An Italian study  with predominantly immunocompetent patients (62%) with suspected TB revealed a high concordance between QFG and T-SPOT.TB (κ = 0.699). This finding is in contrast to the results of our cross-sectional analysis, as we observed a poor concordance between the interferon-γ-based assays (κ = 0.146 for all, κ = 0.123 for patients with a history of active TB). However, our study was performed on asymptomatic HIV patients and those with active TB were excluded. Lower sensitivity in HIV patients is probably an important explanation for the poor concordance. The interpretation of all these results with regard to latent TB infection is limited by the absence of a diagnostic gold standard. As all available tests focus on immunological response, this is a limitation for both interferon-γ assays in HIV-infected individuals.
When studying only cases of active TB, T-SPOT.TB has been found providing more positive results than TST in HIV-infected children (73 vs. 36%) , as well as in predominantly HIV-negative (95%) adults (89 vs. 79%) . On the contrary, in a multicenter US trial, among 69 adults, both HIV-positive and HIV-negative patients with culture-proven TB, the number of positive results for TST and QFG was comparable (73.9 vs. 69.6%, not significant) . For 87 patients with active disease, T-SPOT.TB was found more sensitive than QFG (96.6 vs. 70.1%, P < 0.05) . In our patients, the only risk factor associated with a positive blood test result was ‘history of active TB’ for T-SPOT.TB but not for QFG.
We found significantly more indeterminate results with T-SPOT.TB than QFG (eight vs. one patient, or 3.0 vs. 0.4%). Ferrara et al.  observed a similar proportion of indeterminate results with T-SPOT.TB (3%), but remarkably more with QFG (11%), and indeterminate results with QFG were associated with immunosuppression due to cancer chemotherapy and to age of 5 years or less. We enrolled only adult patients; the majority was on antiretroviral therapy (63.6% with undetectable HIV RNA), and the proportion with CD4 cell counts less than 100 cells/μl did not exceed 4%. All these factors may contribute to a different proportion of indeterminate QFG results.
Which test is best to use in HIV-positive individuals? Generally, the sensitivity of all assays to detect latent TB infection should not be overestimated: even in patients with proven historical TB disease, a substantial part (42%) remained nonreactive for all three tests. The CDC recommends the use of QFG in all circumstances in which TST has formerly been used . For HIV-infected individuals, combining QFG and T-SPOT.TB in order to improve sensitivity could be a possible solution. However, the cost-efficacy of such an approach remains to be assessed, especially in antiretroviral-treated patients followed in developed countries, for whom the risk of active TB is relatively low. TST is not only inferior to both interferon-γ-releasing assays in sensitivity but also less specific. In contrast to T-SPOT.TB, QFG is more dependent on CD4 cell count. Therefore, if a single test system has to be chosen, T-SPOT.TB should be preferred for HIV patients, particularly in case of more advanced immunosuppression.
The dependence of QFG on CD4 cell count may be due to the fix amount of 1 ml whole blood used for this analysis. Patients with low CD4 cell counts might fail to induce enough interferon-γ-secreting cells. This limitation of QFG in use for HIV-infected individuals could be overcome by CD4-adapted, higher amounts of collected blood. Concerning T-SPOT.TB, we could confirm the previously suggested independence from CD4 cell counts in AIDS patients with active TB  or asymptomatic HIV patients .
At present, we have not observed any case of active TB in our test population. However, not all patients with a positive result in any of the tests have been thoroughly screened to date, mainly due to a short observation period. A clinical follow-up analysis after a longer observation period could provide interesting results.
In conclusion, both QFG and T-SPOT.TB showed more positive results than TST, suggesting that TST alone underestimates the prevalence of TB infection. Agreement was fair between TST and QFG or T-SPOT.TB, but poor between both interferon-γ-based assays, probably due to a low sensitivity of both interferon-γ-based assays in HIV patients. The blood tests provided low proportions of indeterminate results (0.4% for QFG and 3.0% for T-SPOT.TB, difference P < 0.01). QFG results depend on CD4 cell count, that is, a limitation in an HIV-infected population. History of prior active TB was statistically associated with positive results for TST and SPOT but not with positive results for QFG.
We outstandingly thank Geetha Sarrach and the staff of the Medical HIVCENTER at the Hospital of the Johann Wolfgang Goethe-University in Frankfurt for recruitment and performance of this study and Hildegund Sauer-Eppel for laboratory assistance.
C.S., T.W. and U.G. contributed equally to this article. C.S., T.W., U.G., O.B., R.G., S.S. and S.S. participated in study design. C.S., T.W., G.N., G.O., Z.R. and S.S. contributed to data collection. C.S., T.W., U.G., G.O., Z.R. and H.R.B. participated in data analysis, all in data interpretation, and C.S., U.G. and T.W. in writing of the report.
The study was investigator-initiated and sponsored by the Medical Department II at the Hospital of the Johann Wolfgang Goethe-University. T-SPOT.TB kits were provided by Oxford Immunotec, Abingdon, UK; the QuantiFERON-TB Gold kits were provided by Cellestis, Carnegie, Australia. The Mantoux test agent was provided by the Frankfurt Public Health Office. The blood test manufacturer companies had no role in study design, data collection, data analysis, data interpretation, or writing of the report. The corresponding author had full access to all the data in the study and had final responsibility for the decision to submit for publication. T.W. has received money from Oxford Immunotec (manufacturer of T-SPOT.TB) for scientific lectures; all other authors declare that they have no conflict of interest.
1. Elzi L, Schlegel M, Weber R, Hirschel B, Cavassini M, Schmid P, et al
, for Swiss HIV Cohort Study. Reducing tuberculosis incidence by tuberculin skin testing, preventive treatment, and antiretroviral therapy in an area of low tuberculosis transmission. Clin Infect Dis 2007; 44:94–102.
2. 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 2005; 19:2109–2116.
3. Judson FN, Feldman RA. Mycobacterial skin tests in humans 12 years after infection with Mycobacterium marinum. Am Rev Respir Dis 1974; 109:544–547.
4. Snider DE Jr. Bacille Calmette–Guérin vaccinations and tuberculin skin tests. JAMA 1985; 253:3438–3439.
5. Selwyn PA, Sckell BM, Alcabes P, Friedland GH, Klein RS, Schoenbaum EE. High risk of active tuberculosis in HIV-infected drug users with cutaneous anergy. JAMA 1992; 268:504–509.
6. Luetkemeyer AF, Charlebois ED, Flores LL, Bangsberg DR, Deeks SG, Martin JN, et al
. Comparison of an interferon-gamma release assay with tuberculin skin testing in HIV-infected individuals. Am J Respir Crit Care Med 2007; 175:737–742.
7. Ferrara G, Losi M, D'Amico R, Roversi P, Piro R, Meacci M, et al
. Use in routine clinical practice of two commercial blood tests for diagnosis of infection with Mycobacterium tuberculosis: a prospective study. Lancet 2006; 367:1328–1334.
8. American Thoracic Society. Targeted tuberculin testing and treatment of latent tuberculosis infection
. Am J Respir Crit Care Med
9. Landis JR, Koch GG. The measurement of observer agreement for categorical data. Biometrics 1977; 33:159–174.
10. Altman DG. Practical statistics for medical research. London, UK: Chapman & Hall; 1991.
11. Singh D, Sutton C, Woodcock A. Tuberculin test measurement – variability due to the time of reading. Chest 2002; 122:1299–1301.
12. Pathan AA, Wilkinson KA, Klenerman P, McShane H, Davidson RN, Pasvol G, et al
. Direct ex vivo analysis of antigen-specific IFN-gamma-secreting CD4 T cells in Mycobacterium tuberculosisinfected individuals: associations with clinical disease state and effect of treatment. J Immunol 2001; 167:5217–5225.
13. Liebeschuetz S, Bamber S, Ewer K, Deeks J, Pathan AA, Lalvani A. Diagnosis of tuberculosis in South African children with a T-cell-based assay: a prospective cohort study. Lancet 2004; 364:2196–2203.
14. Chapman AL, Munkanta M, Wilkinson KA, Pathan AA, Ewer K, Ayles H, et al
. Rapid detection of active and latent tuberculosis infection in HIV-positive individuals by enumeration of Mycobacterium tuberculosis-specific T cells. AIDS 2002; 16:2285–2293.
15. Lalvani A, Pathan AA, McShane H, Wilkinson RJ, Latif M, Conlon CP, et al
. Rapid detection of Mycobacterium tuberculosis infection by enumeration of antigen-specific T cells. Am J Respir Crit Care Med 2001; 163:824–828.
16. Mori T, Sakatani M, Yamagishi F, Takashima T, Kawabe Y, Nagao K, et al
. Specific detection of tuberculosis infection: an interferon-gamma-based assay using new antigens. Am J Respir Crit Care Med 2004; 170:59–64.
17. Kang YA, Lee HW, Yoon HI, Cho B, Han SK, Shim YS, et al
. Discrepancy between the tuberculin skin test and the whole-blood interferon gamma assay for the diagnosis of latent tuberculosis infection in an intermediate tuberculosis-burden country. JAMA 2005; 293:2756–2761.
18. Menzies D, Pai M, Comstock G. Meta-analysis: new tests for the diagnosis of latent tuberculosis infection – areas of uncertainty and recommendations for research. Ann Intern Med 2007; 146:340–354.
19. 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; 149:177–184.
20. Brock I, Weldingh K, Lillebaek T, Follmann F, Andersen P. Comparison of tuberculin skin test and new specific blood test in tuberculosis contacts. Am J Respir Crit Care Med 2004; 170:65–69.
21. Pai M, Gokhale K, Joshi R, Dogra S, Kalantri S, Mendiratta DK, et al
. Mycobacterium tuberculosis infection in healthcare workers in rural India: comparison of a whole-blood interferon gamma assay with tuberculin skin testing. JAMA 2005; 293:2746–2755.
22. Dosanjh DP, Hinks TS, Innes JA, Deeks JJ, Pasvol G, Hackforth S, et al
. Improved diagnostic evaluation of suspected tuberculosis. Ann Intern Med 2008; 148:325–336.
23. Mazurek GH, Weis SE, Moonan PK, Daley CL, Bernardo J, Lardizabal AA, et al
. Prospective comparison of the tuberculin skin test and 2 whole-blood interferon-gamma release assays in persons with suspected tuberculosis. Clin Infect Dis 2007; 45:837–845.
24. Lee JY, Choi HJ, Park IN, Hong SB, Oh YM, Lim CM, et al
. Comparison of two commercial interferon-gamma assays for diagnosing Mycobacterium tuberculosis infection. Eur Respir J 2006; 28:24–30.
25. Mazurek GH, Jereb J, Lobue P, Iademarco MF, Metchock B, Vernon A. Guidelines for using the QuantiFERON-TB Gold test for detecting Mycobacterium tuberculosis infection, United States. MMWR Recomm Rep 2005; 54(RR–15):49–55.
26. Lawn SD, Bangani N, Vogt M, Bekker LG, Badri M, Ntobongwana M, et al
. Utility of interferon-gamma ELISPOT assay responses in highly tuberculosis-exposed patients with advanced HIV infection in South Africa. BMC Infect Dis 2007; 7:99.