Mycobacterial T Cell Responses in HIV-Infected Patients with Advanced Immunosuppression

Tuberculosis (TB) and AIDS respectively result in 2 and 3 million deaths every year worldwide [1]. In developing countries, the most common cause of death among HIV-infected individuals is TB [2]. Patients coinfected with HIV are more susceptible to endogenous reactivation of latent Mycobacterium tuberculosis infection and to exogenous reinfection [1]. Although protective immunity for both TB and HIV is associated with production of interferon (IFN)—γ, tumor necrosis factor (TNF)—α, and interleukin (IL)—2 after antigen stimulation, patients with advanced HIV infection experience a reduction in cell-mediated immune responses [1, 3].

The sensitivity of the traditional tuberculin skin test for detecting latent M. tuberculosis infection in individuals coinfected with HIV is reported to be low [4]. Additionally, the purified protein derivative (PPD) antigens used are not specific to M. tuberculosis, and the test may be confounded by prior bacille Calmette-Guérin (BCG) vaccination and exposure to environmental mycobacteria commonly encountered in the tropics [1, 4, 5].

Recently, the M. tuberculosis—specific antigens early secreted antigenic target (ESAT)—6 and culture filtrate protein (CFP)—10 have been proven to have improved specificity over PPD tuberculin (PPD-T) and are increasingly being used in IFN-γ release assays to identify individuals who have been exposed to M. tuberculosis [5–8]. In the ex vivo IFN-γ enzyme-linked immunospot (ELISpot) assay, the CD4 T cell is thought to be the main cell type responsible for M. tuberculosis—specific responses [6]. Despite this, M. tuberculosis—specific CD4 cell responses in severely immunocompromised individuals with HIV coinfection may be reduced when CD4 cell numbers are low [3], and it is also possible that specific CD8 cell responses can be induced by these antigens [9]. However, the performance of ESAT-6 and CFP-10 in HIV-infected subjects has not been evaluated in our setting, where TB is endemic. We therefore investigated, using the ELISpot assay, antimycobacterial T cell reactivity in relation to HIV disease progression and determined which T cell phenotypes were involved.

Methods. Venous blood was obtained from 145 HIVpositive patients recruited from a previously described [10] clinical HIV cohort in The Gambia, where the incidence of newly diagnosed sputum smear—positive TB is 80 cases/100,000 individuals/ year [6]. Of the 145 patients, 107 were infected with HIV-1, 29 were infected with HIV-2, and 7 were dually infected. These patients were grouped into 3 categories on the basis of their CD4 cell counts (71, 33, and 39 patients in categories of <200, 200–500, and >500 cells/μL, respectively). Seventeen patients with a low CD4 cell count had been treated for TB before the study, as had 1 and 4 in the 2 other CD4 cell count categories, respectively. Two patients in the low and 2 in the medium CD4 cell count categories developed TB disease afterward. Also recruited were 77 control subjects from the community without a history of TB disease or anti-TB treatment. All except 23 of these control subjects had a history of BCG vaccination and a visible scar. At recruitment, all of the HIV-infected subjects were antiretroviral therapy (ART) naive. Now, all eligible patients in the cohort are receiving ART. Informed consent was obtained from all study participants, and the project was approved by the Gambian Government/Medical Research Council Ethics Committee.

Patients were screened for HIV by use of various immunoassays, and their CD4 cell counts were enumerated as described elsewhere [10]. Frequencies of IFN-γ—producing T cells were determined using the ELISpot technique as described elsewhere [7]. Peripheral blood mononuclear cells (PBMCs) were plated at 2.2 × 10⁵ cells/well and were either left unstimulated (medium alone) or were stimulated with 2.5 μg/mL ESAT-6 or CFP-10 peptide pool (Advanced Biotech Centre), 5 μg/mL ESAT-6/CFP-10 fusion protein (LUMC) [6, 7], 20 μg/mL PPD-T (RT49; Statens Serum Institut), 250 IU/mL PPD avium (PPD-A; Serumwerk Mensen), 250 μg/mL streptokinase-streptodornase (Wyeth Laboratories), and 5 μg/mL phytohemagglutinin (Sigma-Aldrich) as a positive control. Cell depletions for 10 patient samples in each CD4 cell count category were obtained by use of 50 μL/mL human CD4 and CD8 RosetteSep antibody cocktail mix (Stem Cell Technologies), in accordance with the manufacturer's instructions. Flow cytometry analysis showed >95% purity of cell phenotype enrichment. ELISpot plates were developed, and spot-forming units were quantified as described elsewhere [6, 7]. HIV-1 Gag peptides (20mer overlapping by 10 aa) were obtained from the National Institute for Biological Standards and Control and used in the ELISpot assay at a final concentration of 2 μg/mL.

The number of spot-forming units per 10⁶ CD4 cells was calculated according to CD4 cell percentage. We excluded data from 2 patients because their ELISpot results did not meet the preset criteria [7]. By use of Stata software (version 8; Stata-Corp), the ratios were log transformed before analysis of variance was performed. Where a statistically significant difference was found, t tests were performed, and the P values were corrected for multiple testing by use of the Bonferroni correction. P < .05 was considered to indicate statistical significance. To compare antigen responses across CD4 cell count categories, the ratio of the median level of spot-forming units to CD4 cell count percentage was calculated for each group: the median levels of spot-forming units for the <200 cells/μL and the 200–500 cells/μL categories were divided by that for the >500 cells/μL category to give an n-fold change relative to responses in subjects with CD4 cell counts >500 cells/μL. Pearson's correlation coefficients were calculated for the associations between absolute CD4 cell count and responses to each mycobacterial antigen.

Results. Table 1 shows the proportion of patients in each CD4 cell count category with responses to TB antigens compared with healthy, HIV-uninfected control subjects. The proportion of people who responded to PPD-T fell as CD4 cell counts decreased (P = .04). On the other hand, the proportion of individuals responding to the other antigens was not different between the CD4 cell count categories. Despite the high proportion of responders to PPD-T among control subjects, a lower proportion responded to CFP-10 compared with those infected with HIV. When expressed as spot-forming units per 10⁶ PBMCs, responses to the antigens did not differ according to CD4 cell category or HIV-1 or HIV-2 status (data not shown). Neither the proportion of responders nor the magnitude of responses among the patients with a history of TB differed from the proportion or magnitude among those who did not. Three of the 4 patients who developed TB after this study had commenced had strong responses to all or most of the antigens.

There was a total reduction in IFN-γ—producing cells when CD4 cells were depleted in 14 patients and a marked reduction in another 6 (data not shown). Depletion of CD8 cells resulted in increased numbers of spot-forming units, presumably because this increased the relative proportion of responding CD4 cells. The majority of IFN-γ responses were thus mediated by CD4 cells.

We then expressed IFN-γ responses in relation to CD4 cell count. Surprisingly, patients with low levels of CD4 cells had significantly higher levels of IFN-γ responses (figure 1). However, when the number of CD4 cells producing IFN-γ spotforming units per microliter of blood was calculated, similar values were obtained for each of the antigens for each of the CD4 cell count categories; the exception was responses to PPD antigens, which were lower in those with CD4 cell counts <200 cells/μL (figure 2). In subsidiary experiments, we measured IFN-γ by ELISpot assay after stimulation of PBMCs with pooled HIV-1 Gag peptides. Responses were high in those with a normal CD4 cell count (median, 270 sfu/10⁶ CD4 cells; n = 5) and were low in those with a CD4 cell count >200 cells/μL (median, 84 sfu/10⁶ CD4 cells; n = 8).

View larger version:

• In this window
• In a new window

• Download as PowerPoint Slide

Figure 1

Magnitude of interferon (IFN)—γ responses to mycobacterial antigens in responder HIV-infected patients. Peripheral blood mononuclear cells (PBMCs) obtained from HIV-infected patients were stimulated with individual antigens in ex vivo enzyme-linked immunospot assays. The no. of spot-forming units per 10⁶ PBMCs was then related to the CD4 cell percentage in each category. Data were log transformed, t tests were performed, and the P values were corrected for multiple testing by use of the Bonferroni correction. Horizontal lines indicate medians. CFP, culture filtrate protein; ESAT, early secreted antigenic target; PPD, purified protein derivative.

View larger version:

• In this window
• In a new window

• Download as PowerPoint Slide

Figure 2

Absolute no. of CD4 cells per microliter that responded to the different mycobacterial antigens, by CD4 cell count category.

When the proportion of CD4 cells producing IFN-γ was related to CD4 cell count category, we observed a higher relative number of IFN-γ—producing CD4 cells in individuals with lower CD4 cell counts, compared with those in individuals with normal CD4 cell counts (table 1). Thus, the relative proportion of CD4 cells responding to M. tuberculosis antigens was consistently higher in subjects with CD4 counts <200 cells/μL by a factor ranging from 15.5 for CFP-10 to 6.0 for PPD-T, compared with the proportion found in those with CD4 cell counts >500 cells/μL. The relative proportion in those with intermediate CD4 cell counts (200–500 cells/μL) also increased by a factor ranging from 4.6 for CFP-10 to 1.7 for PPD-A. Consequently, there was a negative association between absolute CD4 cell count and IFN-γ ELISpot responses to ESAT-6, CFP-10, fusion protein, PPD-T, and PPD-A, which respectively were −0.81, −0.83, −0.80, −0.69, and −0.75 (Pearson's correlation coefficients), with each association being highly significant (P < .0001).

View larger version:

• In this window
• In a new window

• Download as PowerPoint Slide

Table 1

Proportion of HIV-infected patients and relative proportion of interferon (IFN)—γ—producing CD4 cells responding to mycobacterial antigens in each CD4 cell count category.

Discussion. In the present study, we observed similar proportions of responders to the various mycobacterial antigens among HIV-infected individuals with different CD4 cell counts. An exception was observed for the proportion of responders to PPD-T, which decreased with lower CD4 cell counts. We also determined that the main source of IFN-γ response was CD4 cells, which is in line with our earlier findings [7]. When normalized for CD4 cell count, the relative frequencies of ESAT-6—specific and CFP-10—specific IFN-γ—secreting T cells were relatively increased in those with low CD4 cell counts.

Although infection with M. tuberculosis is implied by the data, infection due to environmental mycobacteria is also indirectly implied by twice the proportion of HIV-infected patients in the lower category (<200 cells/μL) responding to PPD-A (table 1), compared with ESAT-6 responses and, to a lesser extent, with PPD-T responses. A number of studies have demonstrated that M. avium complex disease is opportunistic in patients with severe immunodeficiency [11], and this may be caused by latent M. avium infection or disease.

It was surprising that there were relatively increased responses in individuals with low CD4 cell counts and a strong negative association between absolute CD4 cell counts and responses (P < .0001), given that this category of HIV patients was expected to be anergic because of severe immunodepression. Indeed, their responses to HIV-specific Gag peptides were lower than those with normal CD4 cell counts. After M. tuberculosis infection, the magnitude of IFN-γ responses to ESAT-6 and CFP-10 have been shown to reflect bacterial load [12, 13]. The relative increase in mycobacteria-specific CD4 cell responses in our cohort may reflect early signs of increased mycobacterial burden [14], due to endogenous reactivation of TB and/or exogenous reinfection with other M. tuberculosis strains or other mycobacteria [3]. However, this is speculative, because few of the subjects developed TB disease after testing. These IFN-γ responses are unlikely to contain M. tuberculosis infection and may signal heightened risk for TB disease in this category of HIV-infected individuals. Prospective studies are aimed at defining the breadth of response in relation to TNF-α, IL-2, and other cytokines thought to be important in M. tuberculosis immunity.

The peptides we used were derived from relatively small proteins with the likelihood of stimulating both CD4 and CD8 cell responses [7]. However, only 1 subject demonstrated ESAT-6/CFP-10—specific CD4 and CD8 cell responses, and a lower frequency of IFN-γ—producing CD8 cell responses was seen in the ELISpot assay. Few studies have shown IFN-γ—producing CD8 cell responses during TB/HIV coinfection [15], whereas others have reported predominance of a CD4 cell—mediated IFN-γ response, consistent with our results [5, 6, 8]. These variable responses may perhaps be influenced by the mycobacterial species to which the person is exposed as well as host differences, such as HLA type and/or other genetic factors [3].

Despite a progressive loss of T cell numbers and function due to HIV infection [1], there is preservation and even increase in the proportion of CD4 cells that can effectively recognize mycobacterial antigens. Whether such immune recognition plays a role in containment of mycobacterial infection or indicates progressive TB disease needs further investigation. Our study is consistent with the work of others [15] in that it indicates that the M. tuberculosis-specific ELISpot assay is likely to have a niche in detecting and monitoring M. tuberculosis infection in patients with AIDS despite their low CD4 cell counts and further supports the growing view that PPD-T is insufficient for TB diagnosis in these individuals.

• Received February 5, 2007.
• Accepted July 23, 2007.

• © 2008 by the Infectious Diseases Society of America