Selected Pool of Peptides from ESAT-6 and CFP-10 Proteins for Detection of Mycobacterium tuberculosis Infection

Abstract

We have validated a new test for detecting Mycobacterium tuberculosis infection. A pool of synthetic peptides derived from ESAT-6 and CFP-10 proteins was used to detect the number of specific gamma interferon-producing T cells by means of an enzyme-linked immunospot assay. Sixty-eight individuals positive for M. tuberculosis infection, either human immunodeficiency virus-seropositive or -seronegative, were studied. The test results were highly specific (87.5%) and sensitive (93.1%), more so than a classical lymphoproliferative assay (specificity and sensitivity of 77.27%), opening new possibilities for diagnosis and screening of tuberculosis. Moreover, the test allowed us to distinguish individuals infected with M. tuberculosis from those vaccinated with BCG.

The World Health Organization declared tuberculosis (TB) a global emergency since one-third of the world's population is infected, and ca. 5% of those infected develop active disease during the first years after exposure (35). Moreover, this number has increased in the last years due to the large diffusion of human immunodeficiency virus type 1 (HIV-1) infection, the strongest risk factor for the development of TB (27). Thus, new strategies for specific diagnosis and prevention of transmission are urgently needed.

The “gold standard” for the diagnosis for TB is a combination of clinical and radiologic examination and microbiological tests, both microscopic and cultural (the last one can take up to 8 weeks, and in 10 to 20% of cases Mycobacterium tuberculosis is not successfully cultured) (1). The tuberculin skin test (TST), using purified protein derivative (PPD), is largely utilized for both diagnosis and screening. However, the greatest drawback of PPD is its broad cross-reactivity with antigens (Ags) derived from several mycobacterial species, e.g., attenuated M. bovis bacillus Calmette-Guérin (BCG) used for vaccination, greatly decreasing the specificity of the TST (10). Moreover, 10 to 25% of TB patients have a negative TST result, and this percentage increases up to 50% in patients with advanced disease or with immunodeficiency due to HIV coinfection (12, 20, 21).

M. tuberculosis infection evokes a strong cell-mediated immune response, and detection of specific T cells might be a means to detect infection (17, 39). However, the problem of the assays currently used is the lack of well-defined specific Ags.

Recently, two secretory, low-molecular-mass proteins, early secreted antigenic target 6 (ESAT-6) and culture filtrate protein 10 (CFP-10) have been identified, and the corresponding genes have been coded (7, 38). These genes belong to the RD-1 genomic region of the M. tuberculosis complex, being absent in all vaccine strains of BCG and in environmental isolates with the exception of M. kansasii, M. marinum, M. szulgai, M. flavescens, and M. gastrii (6, 19, 31, 38). An ESAT-6 homologue has also been identified in the genome of M. leprae (18). Both proteins elicit a strong T-cell response in subjects with TB (3, 28, 34, 41, 42) and were used for detection either of early and active TB or of subclinical infection (14, 40). Synthetic overlapping peptides corresponding to the full length of each protein were recently synthesized and used as single peptides, as well as a pool. They were as efficient as use of the entire proteins in detecting T-cell responses in TB patients (4, 23, 34, 40) and permitted discrimination of BCG-vaccinated individuals from subjects with TB (22, 37). Moreover, of the different techniques used to evaluate T-cell responses, either in vivo (TST) or in vitro (lymphoproliferation assay [LPA]), the enzyme-linked immunospot (ELISPOT) assay for detecting single-cell gamma interferon (IFN-γ) secretion or the enzyme-linked immunosorbent assay (ELISA) method was the most sensitive (4, 11, 26, 30). However, the high number of peptides needed might make these assays too complex, limiting their application.

In order to obviate these difficulties, we selected five highly immunogenic and largely HLA-DR-restricted peptides, three derived from ESAT-6 and two from the CFP-10 protein (4, 41) and then tested them either in an ELISPOT assay to evaluate the number of specific IFN-γ-secreting T cells or in an LPA. Sixty-eight subjects, either HIV seropositive (HIV⁺) or HIV seronegative (HIV⁻), classified as having TB were enrolled in the study. We correlated the responses to the pool of M. tuberculosis peptides (MTPs) to those of BCG and H37Rv strains and of PPD. The ELISPOT assay, using the pool of MTPs, proved to be highly specific and sensitive for TB infection, not very expensive, and easy to apply, opening new possibilities for TB diagnosis and screening.

MATERIALS AND METHODS

Study population.

A total of 68 subjects (40 male, 28 female) were prospectively recruited at the Infectious Diseases Clinic of the San Raffaele Scientific Institute from March 2001 to May 2003 (Table 1). All subjects were classified for TB according to directions of the American Thoracic Society (2). (This official statement of the American Thoracic Society and the Centers for Disease Control and Prevention was adopted by the ATS Board of Directors, July 1999. The statement was endorsed by the Council of the Infectious Disease Society of America in September 1999.) A total of 32 subjects (17 HIV-infected) belonged to class 0 (no TB exposure, not infected, negative reaction to TST), 6 belonged to class 2 (positive reaction to TST and no clinical, bacteriological, or radiographic evidence of active TB), 23 (5 HIV infected) belonged to class 3 (active TB culture-confirmed), 6 (4 HIV infected) belonged to class 4 (history of episodes of TB, no clinically active TB), and 1 (HIV infected) belonged to class 5 (diagnosis pending). All but three HIV-infected individuals had a CD4⁺-T-cell count of <200 cells μl⁻¹ (median, 64; range, 7 to 858) and a log median HIV RNA of 9.8 copies ml⁻¹ (range, 4.4 to 14.6), indicating that the large majority were severely immunocompromised. Of the 32 class 0 subjects, 3 HIV-seropositive individuals were found to be infected by M. avium complex (MAC), as confirmed by blood culture, and 1 HIV-seronegative individual, with diabetes and chronic cardiac disease, was determined to have mycobacterial pneumonia, as confirmed by repeated sputum cultures positive for MAC.

TABLE 1.

Characteristics of individuals enrolled in the study

HIV status and TB class	No. of subjects	Male/female ratio	Mean age in yr (range)	TST (no. positive/total no.)
Seropositive	27	17/10	41.6 (26-72)	1/21
0	17	8/9	40.3 (27-53)	0/14
2	0
3	5	4/1	38.2 (26-72)	1/3
4	4	4/0	52.7 (40-60)	0/3
5	1	1/0	47	0/1
Seronegative	41	23/18	41.6 (15-75)	16/31
0	15	4/11	40.8 (26-75)	0/15
2	6	5/1	38.3 (33-42)	6/6
3	18	13/5	46.3 (31-33)	9/9
4	2	1/1	32 (31-33)	1/1
5	0

TST testing.

Of 68 (76.5%) enrolled individuals, 52 underwent TST administered by the Mantoux method with 0.1 ml (5 tuberculin units) of Biocinetest-PPD tubercolin (Chiron Italia srl, Milano, Italy) injected intradermally in the volar surface of the forearm. Induration at 48 to 72 h was measured and recorded. Subjects with an induration of ≥10 mm (≥5 mm in HIV-infected subjects) were classified as positive.

Ags.

Candida albicans Ags (Sanofi Diagnostic Pasteur, Marnes la Coquette, France) was tested at 25 μg/ml as recall Ag; PPD was prepared from M. tuberculosis (Statens Serum Institut, Copenhagen, Denmark) and tested at 1 μg/ml; attenuated BCG (Organon Teknika, Boxtel, The Netherlands) and virulent M. tuberculosis H37Rv strain (ATCC 22294) were grown in Middlebrook 7H9 broth (Difco, Detroit, Mich.) supplemented with oleic acid, albumin, dextrose, and catalase (OADC Enrichment; Difco) on an orbital shaker at 37°C for 7 days. The cultures were then both heat inactivated and sonicated and subsequently used at a final concentrations of 1:20 (vol/vol).

Synthetic peptides.

For the detection of a specific response to M. tuberculosis, a pool of five synthetic peptides (Primm srl, Milano, Italy) of 20 amino acids (aa) in length, >70% purified, derived from the sequences of ESAT-6 and CFP-10 secretory proteins from M. tuberculosis complex (Table 2) were used at a final concentration of 2 μg/ml per peptide (10 μg/ml in total). A synthetic peptide 10 aa in length from the sequence of heat shock protein 60 (HSP-60) of M. leprae was used at a concentration of 10 μg/ml as a control stimulus. The MTPs were chosen to be highly immunogenic and to represent the larger HLA-DR restriction as reported elsewhere (4).

TABLE 2.

Amino acid sequences of synthetic peptides used in this study

Protein	Position	Sequence
ESAT-6	1-20	MTEQQWNFAGIEAAASAIQG
ESAT-6	31-50	EGKQSLTKLAAAWGGSGSEA
ESAT-6	61-80	TATELNNALQNLARTISEAG
CFP-10	51-70	AQAAVVRFQEAANKQKQELD
CFP-10	71-90	EISTNIRQAGVQYSRADEEQ

Ex vivo ELISPOT assay for single-cell IFN-γ release.

Peripheral blood mononuclear cells (PBMC) from HIV⁺ or HIV⁻ individuals were obtained by blood centrifugation over Ficoll-Hypaque density gradient and suspended in RPMI 1640 medium (BioWhittaker Europe, Verviers, Belgium) supplemented with 2 mM [scap]l-glutamine, 10 U of penicillin-streptomycin/ml, and 10% heat-inactivated fetal calf serum (FCS; BioWhittaker). For IFN-γ-ELISPOT assay, 96-well polyvinylidene difluoride-bottom plates (MAIPS4510; Millipore, Bedford, Mass.) were precoated with anti-IFN-γ capture monoclonal antibody (MAb B-B1; Diaclone, Besançon, France) and kept at 4°C overnight. PBMC were seeded in triplicate at 5 × 10⁴ or 2 × 10⁵ cells/well (3 × 10⁵ cells/well for HIV⁺ individuals with circulating CD4⁺-T-cell counts of <200 cells μl⁻¹) and stimulated with the different Ags or with the pool of MTPs. Negative controls were represented by PBMC in medium alone or with added peptide from HSP-60. C. albicans Ags and phytohemagglutinin at 5 μg/ml (Sigma, St. Louis, Mo.) were used as the recall Ag and positive control, respectively. After 36 h of incubation at 37°C in air plus 5% CO₂, biotinylated anti-IFN-γ detection MAb (B-G1; Diaclone) was added. After 4 h of incubation at 37°C, streptavidin-alkaline phosphatase conjugate (Amersham Pharmacia Biotech Europe GmbH, Freiburg, Germany) was added for 1 h. Then, after a washing step, a chromogenic substrate (nitroblue tetrazolium-BCIP [5-bromo-4-chloro-3-indolylphosphate]) was added for 10 to 15 min, and the plates were washed with tap water. After the plates were dried, single spots were counted with an automated image analysis system ELISPOT reader (AID-GmbH, Strassberg, Germany).

The responses were empirically scored as positive if the test wells contained a mean number of spot-forming cells (SFC) higher than the mean value plus two standard deviations in negative control wells, and when the number of SFC per million PBMC in stimulated wells (subtracted of the values of negative control wells) was >25. Of note, in our experience with PBMC of 128 healthy subjects and plating 2 × 10⁵ to 3 × 10⁵ cells/well, we have never observed >25 spots in negative control wells (mean spots, 4.88; standard deviation, 7.82; range, 0 to 25).

In vitro lymphoproliferative assay.

Freshly isolated PBMC were seeded in 96-well round-bottom culture plates (Corning Costar, Concorezzo, Italy) at 10⁵ cells/well in RPMI 1640 medium plus 10% FCS and then stimulated for 6 days with the same Ags and MTPs and at the same concentrations as reported for the ELISPOT assay. [³H]thymidine (Amersham Italia S.r.l., Milano, Italy) was added at 1 μCi/well during the last 18 h of culture. Cultures were harvested on glass fiber filters, and the incorporated radioactivity was measured in a β-counter (Top Count; Packard Instrument Co, Meriden, Conn.). The results were expressed as counts per minute (cpm) and scored as positive if the stimulation index (SI) (i.e., cpm in test wells/cpm in control wells) was >3.

Quantitation of HIV viremia in plasma.

Plasma HIV RNA was quantitated by using the NASBA system (Organon Teknika, Boxtel, The Netherlands), which has a lower limit of detection of 80 RNA copies ml⁻¹.

CD4⁺-T-cell count.

Absolute CD4⁺-T-cell counts and percentages were calculated based on whole-blood samples. Flow cytometry analysis, with fluorescein-isothiocyanate-conjugated anti-CD4 monoclonal antibody (Coulter, Hialeah, Fla.) was performed on an Epics XL-MCL flow cytometer (Beckman Coulter Instrumentation Laboratory, Milano, Italy).

Statistical analysis.

Sensitivity and specificity, and their corresponding 95% confidence intervals (CIs), were calculated. Agreement between different immunological tests and TB classification was assessed by estimating the Cohen's kappa (κ) statistic (and their corresponding 95% CIs). According to Fleiss (16), κ values of >0.75 suggest excellent agreement beyond that determined by chance; values of >0.4 and <0.75 indicate fair to good agreement, and values of <0.4 indicate poor agreement. Differences between individuals classified as class 0 or 3 to 4 for TB infection or between HIV⁻ and HIV⁺ subjects for peptide-specific SFC mean values were compared by using the two-sample Wilcoxon rank sum test for nonparametric data. The probability level 0.05 (P = 0.05) was used to test the significance of the previous tests. Statistical analysis was carried out by using the SAS program (release 8.02).

RESULTS

Ex vivo ELISPOT assay for IFN-γ T-cell responses to the pool of MTPs.

The specificities and sensitivities of the pool of MTPs for the diagnosis of TB infections were analyzed by comparing the IFN-γ T cells responses of 32 individuals classified as class 0 for TB infection with those of 29 individuals classified as class 3 (23 subjects) or class 4 (6 subjects) in an IFN-γ-ELISPOT assay. The specificity was 87.5 (95% CI = 0.71 to 0.96), and the sensitivity was 93.1 (95% CI = 0.77 to 0.99) (Table 3A). Of 32, 4 (12.5%) class 0 individuals tested responded to the MTP. One was a medical doctor who had worked for a long time in an infectious disease department, and the three remaining subjects were highly immunocompromised HIV-infected individuals (CD4⁺-T-cell count of ≤90 cells μl⁻¹) with a negative TST result. Conversely, 2 of 29 (6.9%) class 3 or 4 individuals did not respond to the pool of MTPs. They were both HIV infected: one subject was highly immunocompromised (circulating CD4⁺-T-cell count of 32 cells μl⁻¹) with a negative TST result and unresponsive to all Ags tested (data not shown), whereas the other had an immune reconstitution syndrome characterized by a very high spontaneous activation of his lymphocytes able to mask any specific response (e.g., the mean number of SFC of 10⁶ lymphocytes in the three negative wells was 1,070 in the patient versus <25 in healthy subjects).

TABLE 3.

Response frequencies to MTP relative to TB classification: ELISPOT assay versus LPA^a

Assay and TB class	Response frequency (%) to MTP:		Total no. tested
	−	+
IFN-γ-ELISPOT assay
0	28 (87.5)	4 (12.5)	32
3-4	2 (6.9)	27 (93.1)	29
Total	30	31	61
LPA
0	17 (77.27)	5 (22.73)	22
3-4	5 (22.73)	17 (77.27)	22
Tot	22	22	44

^a−, Negative result, +, positive result. Values in bold face indicate specificity and sensitivity.

The IFN-γ-ELISPOT assay, as well as ELISA for detecting IFN-γ production, require short in vitro stimulation of T cells, whereas LPA requires longer stimulation (1 to 2 days versus 6 days, respectively). Besides, Pathan et al. (30) reported that T cells specific for certain ESAT-6 peptides were not detected by LPA. To verify which test was more reliable, for 44 of the 61 individuals tested for IFN-γ-ELISPOT assay, an LPA was also performed with the same panel of Ags as for the ELISPOT assay. In light of the responses to MTP, the specificity and sensitivity were both equal to 77.27 (95% CI = 0.55 to 0.92) for LPA (Table 3). IFN-γ T-cell responses versus MTP and classification of TB infection were highly concordant (κ = 0.80 and 95% CI = 0.65 to 0.95), whereas concordance was lower with LPA (κ = 0.55, and 95% CI = 0.30 to 0.79). Similar differences were also observed with the other Ags (data not shown), suggesting that they reflected differences in assay sensitivity per se.

Thus, the pool of MTPs was very specific and sensitive for the diagnosis of TB infection, and the IFN-γ-ELISPOT assay was determined to be more reliable than the LPA.

High correlation between in vitro T-cell responses to MTP and BCG or H37Rv Ags.

If MTPs were really capable of detecting TB infection, we can expect that all individuals that responded to peptides responded also to BCG and H37Rv Ags. To verify this possibility, we compared T-cell responses to MTP versus those to BCG (or to H37Rv [data not shown]) in the IFN-γ-ELISPOT assay. All but one individual tested responded to both stimuli (Table 4). The only one that responded to MTP but not to BCG nevertheless responded to H37Rv. Of the 33 individuals that did not respond to MTP, 21 responded to both BCG and H37Rv. These were BCG-vaccinated individuals or had been previously infected with MAC (one subject was HIV seronegative with a pulmonary infection, and three subjects were HIV seropositive with disseminated infection).

TABLE 4.

IFN-γ-ELISpot assay: comparison of T-cell response frequencies to MTP versus BCG^a

MTP response	T-cell response frequency (%) to BCG:		Total no. tested
	−	+
−	12 (36.36)	21 (63.64)	33
+	1 (2.86)	34 (97.14)	35
Total	13	55	68

^a−, Negative result; +, positive result. Values in bold face indicate specificity and sensitivity.

Discordance between TST and in vitro IFN-γ T-cell production after stimulation with PPD.

The TST is widely used to diagnose TB infections. However, up to 50% of patients with advanced disease or with immunodeficiency due to HIV infection have a negative TST (12, 20, 21). To verify whether there was correspondence between in vivo (TST) and in vitro (ELISPOT) responses to the same Ag (PPD), 52 of 68 (76.5%) subjects enrolled were tested by the TST (35 tests were TST negative, and 17 were TST positive) and by IFN-γ-ELISPOT assay with PPD as the Ag (Table 5). T lymphocytes of 18 of 35 TST-negative subjects (9 HIV⁻ and 9 HIV⁺) responded to in vitro stimulation with PPD, whereas only 2 of 17 TST-positive subjects (1 HIV⁻ and 1 HIV⁺) did not respond to PPD (the HIV⁺ individual was the mentioned above with the immune reconstitution syndrome). Of note, four HIV⁺ patients with active TB that were TST negative responded to in vitro stimulation with PPD.

TABLE 5.

Comparison of T-cell response frequencies to PPD in the IFN-γ-ELISpot assay versus TST

TST response	T-cell response frequency (%) to E-PPDa		Total no. tested
	−	+
−	17 (48.57)	18 (51.43)	35
+	2 (11.76)	15 (88.24)	17
Total	19	33	52

^aIFN-γ-ELISPOT assay with PPD as antigen. Concordance κ = 0.30; 95% CI = 0.09 to 0.50. −, Negative result; +, positive result. Values in boldface indicate specificity and sensitivity.

Thus, these results confirm the greater capability of the IFN-γ-ELISPOT assay versus the TST to diagnose TB and the discordance between in vivo and in vitro tests.

Higher numbers of SFC in response to MTP in HIV⁻ than in HIV⁺ individuals with TB.

The ELISPOT assay allowed us to determine not only whether there was a specific T-cell response but also the number of responding cells. As expected, the mean number of SFC per million PBMC in response to MTP was significantly higher in TB class 3 or 4 subjects than in class 0 subjects (Fig. 1). However, the mean value was significantly lower in HIV⁺ (median, 115 SFC) than in HIV⁻ (median, 595 SFC) subjects (P = 0.022). No significant differences were observed in the mean numbers of SFC between HIV⁺ and HIV⁻ TB class 0 subjects (medians, 10 and 5 SFC, respectively).

FIG. 1.

Values of IFN-γ-producing T cells (SFC per million PBMC) responding to the pool of MT peptides in HIV⁻ and HIV⁺ individuals classified as class 0 or 3 to 4 for TB infection. Each circle represents an individual subject. The horizontal bars represent the median values of peptide specific SFC in each group of individuals. The cutoff for a positive response was ≥25 SFC per million PBMC.

DISCUSSION

In this study we validated a new test for the detection of TB infection. Using a pool of five synthetic peptides, derived from ESAT-6 and CFP-10 secretory proteins in association with an ELISPOT assay for detecting IFN-γ-producing T cells, we tested 68 individuals that were either HIV⁺ or HIV⁻ and classified as having a TB infection. The test results were highly specific and sensitive, more reliable than a classical LPA, and able to distinguish BCG-vaccinated subjects from individuals with TB.

Due to the increasing numbers of persons with TB in the last years (35), the worldwide dissemination of HIV-1 infection, the strongest risk factor for TB development (27), and the continuous migration of peoples from areas with a high incidence of TB to highly industrialized countries, TB is considered a global emergency. To counteract and control this epidemic, early and accurate identification of person with active disease, as well as individuals who are latently infected, is absolutely mandatory (24, 25). However, diagnoses of TB based on clinical symptoms, chest radiography, and sputum microscopy are sensitive but not specific, especially in immunocompromised, HIV-infected individuals (9). Culture of bacteria is time-consuming, and in most cases the bacillus is not cultured (1), whereas nucleic acid-based methods such as PCR are not consistently accurate enough for the diagnosis of smear-negative pulmonary TB (36). The TST is widely used for the diagnosis of M. tuberculosis infection but has several drawbacks such as cross-reactivity, poor specificity, limited sensitivity, and the need for a followup visit limits its reliability (10, 12, 20, 21). In the past few years, in vitro tests of cell-mediated immunity, in particular LPA and IFN-γ detection in the supernatant of stimulated cultures, have been used for diagnosing TB (12, 13, 32, 33). However, these assays exhibited a low specificity due to the cross-reactivity of the Ags used (whole protein extracts or the 30-kDa Ag of M. tuberculosis, BCG, and PPD).

Recently identified proteins of low molecular mass, ESAT-6 and CFP-10 in particular, secreted by M. tuberculosis and virulent M. bovis strains but not by all BCG vaccine strains or by the majority of environmental isolates (6, 7, 18, 19, 31, 38) seem quite suitable for this purpose. These proteins can induce a strong T-cell response, as detected by LPA or IFN-γ release, in subjects with early and active TB or contacts with TB patients (3-5, 8, 14, 22, 28, 34, 37, 40-42). Synthetic peptides, of either a 15-mer or a 20-mer, derived from both proteins have been produced and were as effective as full-length proteins in eliciting in vitro specific immune T-cell responses (4, 23, 30, 34, 40). More interesting, they permitted rapid detection not only of active but also of latent TB infection (LTBI) in either HIV⁻ or HIV⁺ individuals (11, 15, 26, 30). However, the high number of peptides used might complicate and limit their application. For these reasons we decided to reduce the number of peptides to use, choosing those that are highly immunogenic and that should be presented by the largest number of HLA-DR molecules. We selected and pooled three peptides derived from ESAT-6 protein and two peptides derived from CFP-10 protein, all of which are 20 aa in length (4, 29, 34, 40, 41) (Table 2). These peptides were recognized by CD4⁺ T cells and hold multiple epitopes (4, 29, 34). In particular, the 1- to 20-aa N-terminal peptide from ESAT-6 protein was highly immunogenic (41). As a test for detecting T-cell responses we chose and correlated the LPA with the IFN-γ-ELISPOT assay, which is considered a more accurate and highly sensitive assay (15, 23, 25, 30). An ELISA method has also been proposed to detect IFN-γ production by PBMC in response ESAT-6 or CFP-10 proteins or their derived peptides (4, 5, 8, 22, 36). ELISA and ELISPOT assays are equally specific and sensitive; however, the ELISPOT method allow us to detect single cells producing IFN-γ and to generate quantitative results.

The association of IFN-γ-ELISPOT assay and MTPs was highly specific and sensitive (more than LPA) for detecting M. tuberculosis infection. Of note, four individuals classified as class 0 for TB infection, a medical doctor working in an infectious disease department and three HIV-infected individuals with very low levels of circulating CD4⁺ T lymphocytes, responded to MTP. Since there is no gold standard test for LTBI, we cannot know for sure whether these persons were infected or not. However, the test should be able to detect LTBI, particularly in severely immunocompromised subjects that were TST negative (unpublished data). Only two HIV⁺ patients with active TB did not respond to MTP both in the ELISPOT assay and in the LPA. The first one was severely immunocompromised and unresponsive to all Ags tested, whereas the other had an immune reconstitution syndrome characterized by an elevated spontaneous activation of his lymphocytes that masked the response to all Ags tested.

Once we established that the ELISPOT assay is more accurate and effective than the LPA, we further validated the test by correlating the responses to MTP with the responses to the BCG and H37Rv Ags. The results showed a high concordance among the responses to these Ags (Table 4).

It is well known that poor specificity is the major drawback for TST (1). Moreover, 10 to 25% of TB patients do not react to PPD, decreasing its sensitivity to <50% in patients with advanced disease or HIV infection (12, 20, 21). Our data showed that in vitro testing of T-cell responses to PPD by an IFN-γ-ELISPOT assay can overcome this problem, as reported by others using different methods for IFN-γ detection (5, 8, 12, 22, 33). In fact, we observed that 18 of 35 subjects with a negative TST result, 9 of whom were HIV-infected, responded to in vitro T-cell stimulation with PPD, whereas only 2 of 17 subjects with a positive TST result did not respond to PPD, thus confirming the high sensitivity and accuracy of the ELISPOT assay. Moreover, these results suggest that a discordance between in vivo and in vitro immunological responses may exist and should be taken into account, especially when immunocompromised patients are evaluated.

Finally, a quantitative analysis of single cell IFN-γ T-cell responses to MTP proved not only that individuals with TB have an increased number of specific T lymphocytes compared to class 0 individuals, as expected, but also that among the former, HIV-infected individuals have a significantly reduced number of SFC compared to uninfected individuals. In a recent study, Chapman et al. (11), using peptides from ESAT-6 and CFP-10 proteins and the IFN-γ-ELISPOT assay, did not observe any significant differences in the frequencies of SFC between HIV-positive and HIV-negative individuals with TB. However, some differences with our study exist with regard to the length and number of peptides used and the populations considered.