Comparison of Intracellular Cytokine Flow Cytometry and an Enzyme Immunoassay for Evaluation of Cellular Immune Response to Active Tuberculosis

Study population.This prospective cross-sectional blinded study was conducted from December 2006 to February 2007 and had 28 patients (21 male and 7 female patients; mean age, 44 years) who were human immunodeficiency virus (HIV) negative, had suspected active TB, and attended TB and chest clinics which provide free public medical service to all TB patients in Hong Kong. Blood samples were taken for lymphocyte immunophenotyping and whole-blood assays for the detection of IFN-γ, and sputum was taken for bacteriological culture for M. tuberculosis.

Twenty-eight consecutive patients were finally categorized as belonging to either the active TB group or the nonactive TB group. Of these 28 patients, 17 (11 male and 6 female patients; mean age, 43 years; age range, 18 to 79 years) were subsequently classified as having active TB by either culture (10/17 patients were culture positive), histological confirmation, and/or radiological confirmation and were placed in the TB group. The other 11 patients were categorized in the nonactive TB group and included patients with inactive M. tuberculosis infection, nontuberculous mycobacterial (NTM) infection, posttreatment TB, and other related respiratory conditions. Inactive TB cases had chest radiographic evidence suggestive of TB, cultures of their sputa for acid-fast bacilli were negative, and their chest radiographic findings did not change significantly or remained static with anti-TB treatment on follow-up or when no anti-TB treatment was started. Patients with NTM infections were culture positive for NTM, including both colonization and NTM lung disease, and did not respond to standard anti-TB treatment. Posttreatment TB cases had received anti-TB treatment for more than 3 months. Other related respiratory conditions were those in which a diagnosis of TB could not be established by bacteriological culture or other investigations, including histological analysis or testing by PCR, or an alternate diagnosis was made.

A healthy control group included 12 healthy HIV-negative volunteers (3 male and 9 female volunteers; mean age, 43 years; age range, 24 to 55 years) with no known history of M. tuberculosis infection, as confirmed by all tests conducted (mycobacterial culture, lymphocyte immunophenotyping, and whole-blood assays for IFN-γ). Informed consent was obtained from all patients and controls prior to the study.

ELISA of whole blood for IFN-γ.The QuantiFERON-TB Gold (QFT-G) in-tube assay (Cellestis Ltd., Victoria, Australia) was performed according to the manufacturer's recommendations. In brief, 1 ml of blood was collected and placed into three blood collection tubes with the following stimulants: no stimulant (antigen-free control), TB antigen (which contained the ESAT-6, CFP-10, and TB7.7 peptide cocktail), and mitogen (which contained phytohemagglutinin). Following incubation for 16 to 24 h at 37°C, 50 μl plasma was removed from each tube and the concentration of IFN-γ was determined by the ELISA and by use of the QFT-G (in-tube) analysis software. The cutoff value for a positive response was taken to be 0.35 IU/ml. Indeterminate results were reported if the IFN-γ level in the tube with mitogen minus that in the tube with no antigen was less than 0.5 IU/ml.

Lymphocyte immunophenotyping.Venous heparinized blood was collected and processed for flow cytometry within 6 h of venipuncture. A standardized method with a recommended four-color panel (6) of combinations of fluorescein isothiocyanate (FITC), phycoerythrin (PE), peridinin chlorophyll protein (PerCP), and allophycocyanin (APC) conjugated to monoclonal antibody in TruCOUNT tubes obtained from a single manufacturer (Becton Dickinson, San Jose, CA) was used to detect the expression of the surface antigens. After the red blood cells were stained in the dark at room temperature for 15 min, they were lysed with 450 μl of 1× fluorescent-activated cells sorter analysis lysing solution for another 15 min. Samples were acquired within 1 h on a FACSCalibur cytometer (Becton Dickinson) and analyzed with MultiSET software (Becton Dickinson). The daily instrument setup and instrument control were performed with CaliBRITE beads 3, CaliBRITE APC beads, and FACSCOMP software. The data were analyzed and were within the recommended analytic reliability checks (6).

ICC technique.The ICC technique is based on the stimulation of T cells to produce cytokine in the presence of a pharmacologic inhibitor of secretion, followed by cell fixation and permeabilization and then intracytoplasmic staining of the accumulated cytokines (12, 18, 22, 29, 32, 34). The use of the drug brefeldin A disrupts the intracellular Golgi apparatus-mediated transport system and allows the cytokines to accumulate and then be detected by flow cytometry.

In vitro lymphocyte activation.Venous heparinized blood was collected and processed within 6 h of venipuncture: three sterilized Falcon tubes (12 by 75 mm) were labeled as the unstimulated control, TB antigen, and mitogen control tubes, respectively. To each tube, 0.5 ml heparinized whole blood with 5 μg brefeldin A was mixed with 0.5 ml supplemented RPMI 1640 medium containing 2 mmol/liter l-glutamine, 10% (vol/vol) heat-inactivated fetal bovine serum, and 100 IU penicillin-streptomycin. Activation was done in the presence of brefeldin A, which inhibited intracellular transport and allowed the antigens and the cytokines produced during the activation to be retained inside the cell. To the TB antigen tube, 5 μg each recombinant ESAT-6 and CFP-10 (ImmunoDiagnostic Inc.) was added; to the mitogen control tube, 25 ng phorbol 12-myristate 13-acetate (PMA) and 1 μg ionomycin were added. Costimulatory antibodies CD28 and CD49d were added to all tubes. The cells were then stained with monoclonal antibodies after 16 to 24 h of incubation at 37°C in the presence of 7% CO₂.

Monoclonal antibodies and staining.FITC-labeled anti-IFN-γ, PE-labeled anti-IL-4, APC-labeled anti-CD69, PE-Cy7-labeled anti-CD8, peridine-chlorophyll-Cy5.5-labeled anti-CD3, and APC-Cy7-labeled anti-CD4 were obtained from the manufacturer (Becton Dickinson). Surface staining was done on ice in the dark, followed by the addition of 0.5 ml of BD 1× permeabilizing solution. Subsequent intracellular staining was performed, and the cells were fixed in 2% (vol/vol) paraformaldehyde in phosphate-buffered saline for flow cytometric analysis. Isotype controls were used as the corresponding clones, in order to determine the level of background staining.

Flow cytometry data acquisition and analysis.Six-color flow cytometry was performed on a FACSCanto flow cytometer (Becton Dickinson) with FACSDiva (version 4.0) software (Becton Dickinson). The instrument was controlled daily by monitoring the optical alignment with Sphero rainbow fluorescent particles of 3.0 to 3.4 μm in diameter (Spherotech Inc.), and the stability of the optical alignment was verified when the optical alignment was within 2 standard deviations (SDs) of the preestablished values. The lymphocyte gating was set by forward and side light scatter (Fig. 1a) and was acquired with 30,000 to 100,000 CD3⁺ CD4⁺ lymphocytes events (Fig. 1b).

• Open in new tab
• Download powerpoint

FIG. 1.

Lymphocyte gating. (a) The lymphocyte gate was set by forward light scatter (FSC-H) and side light scatter (SSC-A). (b) Acquisition was performed with 30,000 to 100,000 CD3⁺ CD4⁺ lymphocyte events.

For data analysis, a sequential gating strategy was employed to detect cytokine-producing cells among the CD3⁺ CD4⁺ and CD3⁺ CD8⁺ lymphocytes. Briefly, a gate was set on viable lymphocytes (forward versus side scatter), CD3⁺ lymphocytes were then gated in a CD3-versus-side scatter histogram. CD3⁺ CD4⁺ and CD3⁺ CD8⁺ cells were then displayed and further gated to determine the expression of CD69⁺ and IFN-γ-producing (IFN-γ⁺) cells and CD69⁺ IL-4-producing (IL-4⁺) cells.

Assay controls.The following controls were included in the assay. (i) Fluorescein-conjugated isotype control antibodies (isotype control) were included in all samples with the isotype-matched antibodies in order to determine the amount of nonspecific staining present. (ii) An unstimulated control served as a means of determination of the background level of in vivo residual cytokine production for each sample. (ii) Treatment with mitogen, a potent stimulus, served as a positive control to ensure the immune status of the patient cells and also as an intracellular staining control with sufficient activation when the proportion of CD69⁺ T cells was >90%.

Calculation of specific response to TB antigen.The CD69 antigen is activated early, and its expression is induced during in vitro antigen stimulation. Upon activation, the level of CD69 antigen expression increases on lymphocytes (31, 35). The CD69 antibody is used to allow the better clustering of cytokine-positive cells and to ensure that cells defined as antigen responsive have been stimulated to express this activation marker. The specific response of the cells to the stimulus was obtained by subtracting the percentage of positive events for the unstimulated control cells (isotype-corrected response) from the percentage of positive events for the TB antigen-activated cells (isotype-corrected response). The antigen-reactive and cytokine-producing cells are expressed as CD69⁺ IFN-γ⁺ CD4⁺ T cells or CD69⁺ IL-4⁺ CD4⁺ T cells. The cutoff was determined by computing the mean and 2 SDs as the reference range for the healthy control group.

Bacteriological cultures.Sputum samples were processed according to standard operating procedures within the TB Reference Laboratory, Public Health Laboratory Centre, Hong Kong, as published previously (16). Briefly, sputum was first decontaminated with an equal volume of 3% NaOH-N-acetyl-l-cysteine for 20 min, neutralized with 0.067 M phosphate buffer (pH 6.8), and centrifuged at 3,000 × g for 15 min. After inoculation of the deposit onto two Lowenstein-Jensen slopes, the slopes were incubated at 37°C. When growth was observed, all isolates were identified by standard identification procedures (16, 38).

Statistical analysis.The performance characteristics of each test for the diagnosis of active M. tuberculosis infection were compared and are expressed statistically as sensitivity and specificity. Statistical analysis was performed with SPSS (version 15.0) software (SPSS Inc.). The results are presented as the mean values ± SDs of the means. Comparisons between groups with normal distributions were performed by Student's t test. Correlation analysis was performed with Pearson's correlation coefficient between two variables. Two-sided P values of less than 0.05 were considered significant.

T-cell subset characterization.Lymphocyte immunophenotyping was performed, and the results of the determinations for the main T-cell populations in the peripheral blood of the 17 TB patients and the 12 healthy controls are depicted in Table 1. The mean levels for all members of the healthy group fell within the reference ranges for our healthy population determined previously (14), and no significant differences were observed. For the TB group, generally slightly elevated levels of CD4⁺ T cells, CD8⁺ T cells, and B cells were detected, but they were not significantly higher, while a significant reduction in the numbers of peripheral natural killer (NK) cells (20% reduction; P < 0.05) was demonstrated.

Specific response by using ex vivo stimulation with PMA plus ionomycin.PMA plus ionomycin is a potent external stimulus for the activation of lymphocytes to produce cytokines. Dot plots (Fig. 2d) showed the activation results for the control assay, which utilized the surface expression of CD69 to assess whether activation had been achieved. Only activated cell populations expressed cytokines, while resting lymphocytes (T, B, or NK cells) from healthy individuals did not constitutively express cytokines (23, 28). If the expected level of CD69 expression of >90% was not seen, the whole set of cytokine data was invalid and the data were excluded from analysis.

• Open in new tab
• Download powerpoint

FIG. 2.

Cytokine activation assay results. (a) IFN-γ production for TB case by isotype control, 0.002% (Q2-2); (b) IFN-γ production for TB case by CD3⁺ CD4⁺ CD69⁺ cells, 0.096% (Q2-2); (c) IL-4 production for TB case by CD3⁺ CD4⁺ CD69⁺ cells, 0.002% (Q2-4); (d) CD69 production for TB case with PMA-ionomycin stimulus control by CD3⁺ CD4⁺ cells, >90% (Q1-2, 77.77%, Q2-2, 18.86%; CD3⁺ CD4⁺ CD69⁺, 96.63%). IgG1, immunoglobulin G1.

The general immune profiles of cytokine expression between the groups are shown in Table 2. It can be seen that the trend was for the activation of CD8⁺ IFN-γ⁺ T cells to be about three times higher than that of CD4⁺ IFN-γ⁺ T cells in all groups. Very low to undetectable levels of IL-4-producing cells were seen.

Specific response of mitogen control in QFT-G assay.The concentration of IFN-γ production by the mitogen control in the QFT-G assay was also determined (Table 2) to validate the activation process for all patients studied. A significant difference in the mean values for the TB group compared with those for the healthy control group was observed (16.61 IU/ml and 10.73 IU/ml, respectively; P < 0.05).

Specific response to ex vivo stimulation of M. tuberculosis antigens ESAT-6 and CFP-10.In 12 healthy individuals with no history of exposure to TB infection, the mean proportion of CD4⁺ CD69⁺ IFN-γ⁺ T cells was 0.0035 ± 0.0033% (Table 3). We then established a responder frequency of the mean ± 2 SDs, by which >0.01% activated CD4⁺ IFN-γ⁺ T cells was considered a positive response to M. tuberculosis infection. The pattern of the specific immune response by a healthy subject is illustrated in Fig. 3.

• Open in new tab
• Download powerpoint

FIG. 3.

IFN-γ production by CD3⁺ CD4⁺ CD69⁺ cells in a healthy control (0.002%; Q2-2).

A significant increase (P < 0.05) in the mean proportion of activated CD4⁺ IFN-γ⁺ T cells of 0.054% was seen in patients with active TB, whereas in the healthy controls the mean proportion was 0.0035% (Fig. 2a to d). There was also some contribution of CD69⁺ CD8⁺ IFN-γ⁺ T cells, the mean level of activation of which was 0.044%, although this level was not significantly greater than that for the healthy controls (0.017%). The distribution of activated CD4⁺ and CD8⁺ IFN-γ⁺ T cells for each group is illustrated in the box plots in Fig. 4. A very low to undetectable level of IL-4 expression could be found. From this, a clear pattern of type 1 cytokine production with a Th1-cell-mediated response in the TB group was demonstrated.

• Open in new tab
• Download powerpoint

FIG. 4.

Distribution of IFN-γ-producing cells among healthy controls and active TB group. The median, interquartile range, and 95% confidence interval are shown on the box plot. h_ifn, CD3⁺ CD4⁺ IFN-γ production; c_ifn, CD3⁺ CD8⁺ IFN-γ production.

Comparative performances of two diagnostic tests.Of the 17 confirmed cases of active TB, 15 (88.2%) were found to be positive by the QFT-G assay and 16 (94.1%) were found to be positive by the ICC assay (Table 4), whereas 10 (58.8%) were found to be positive for M. tuberculosis by culture. Of the 11 cases with nonactive TB, false-positive results were found for 9 (82%) cases by the QFT-G assay and 7 (64%) cases by the ICC assay. The sensitivities of the QFT-G and ICC assays were found to be 88.2% and 94.1%, respectively, and the specificities were 18% and 36.4%, respectively. Three cases of culture-positive non-TB infections occurred: two (one M. kansasii infection, one M. chelonei infection) showed false-positive responses by both the QFT-G and the ICC assays, while one (a Runyon group II mycobacterium infection) did not respond by either test. Both the QFT-G and the ICC assays showed negative results for the healthy control group.

Correlation of IFN-γ release detected by QFT-G and ICC assays with T-cell subsets.The concentration of IFN-γ released, as detected by the QFT-G assay, was significantly higher for the TB group (P < 0.05) than for the healthy controls (9.68 IU/ml and 0.11 IU/ml, respectively). There was also a good correlation between the quantity of IFN-γ produced, as detected by the QFT-G assay, and the CD4⁺ T-cell count in the TB group (r = 0.57, P < 0.05) (Table 5), while no significant association was found for the other lymphocyte subsets enumerated by flow cytometry.

When the ICC assay was used as the detection test, there was no significant correlation between the frequency of CD4⁺ CD69⁺ IFN-γ⁺ T-cell expression and the cell count.