HIV-1 infection in infants severely impairs the immune response induced by Bacille Calmette-Guérin vaccine

Abstract

Background: Worldwide, most infants born to mothers infected with human immunodeficiency virus (HIV) receive bacille Calmette-Guérin (BCG) vaccine. Tuberculosis is a major cause of death among infants infected with HIV in sub-Saharan Africa, and it should be prevented. However, BCG may itself cause disease (known as "BCGosis") in these infants. Information regarding the immunogenicity of BCG is imperative for the risk/benefit assessment of BCG vaccination in HIV-infected infants; however, no such data exist.

Methods: We compared BCG-induced CD4 and CD8 T cell responses, as assessed by flow cytometry, in HIV-infected (n=20), HIV-exposed but uninfected (n=25), and HIV-unexposed (n=23) infants, during their first year of life.

Results: BCG vaccination of the 2 HIV-uninfected groups induced a robust response, which was characterized by CD4 T cells expressing interferon (IFN)-gamma, tumor necrosis factor (TNF)-alpha, and/or interleukin (IL)-2. In contrast, HIV-infected infants demonstrated a markedly lower response throughout the first year of life. These infants also had significantly reduced numbers of polyfunctional CD4 T cells coexpressing IFN-gamma, TNF-alpha, and IL-2, a finding that is thought to indicate T cell quality.

Conclusions: Infection with HIV severely impairs the BCG-specific T cell response during the first year of life. BCG may therefore provide little, if any, vaccine-induced benefit in HIV-infected infants. Considering the significant risk of BCGosis, these data strongly support not giving BCG to HIV-infected infants.

Figure 1

CD4 T cell cytokine responses induced by incubation of whole blood with BCG, as measured by flow cytometry. (A-D) Absolute (abs.) cytokine-producing CD4 T cell numbers (no.) in the three infant groups. Background values (unstimulated blood) were subtracted for each measurement. (A) Total cytokine response at each time point post-vaccination. (B-D) Total IFN-γ-producing CD4 T cells (B), total IL-2-producing CD4 T cells (C) and total TNF-α-producing CD4 T cells (D), in each infant, at each time point. The numbers of HIV-infected, Exposed HIV− and Unexposed HIV− infants per group for this analysis were: 15, 25 and 22 at 3 months; 9, 17 and 21 at 6 months; 8, 15 and 21 at 9 months; and 5, 12 and 17 at 12 months, respectively. For each boxplot, the median is represented by the horizontal line, the interquartile range by the box and the range by the whiskers. Comparisons (p values) were calculated from mixed effects maximum likelihood regression models on log-transformed responses that also included time as a continuous effect (Table 2 and Suppl. Fig. 3).

Figure 2

“Qualitative” differences of BCG-specific CD4 T cell responses in the three infant groups, based on co-expression of type I cytokines. (A) Absolute polyfunctional (IFN-γ⁺, IL-2⁺ and TNF-α⁺) CD4 T cell numbers at each time point. The numbers of HIV-infected, Exposed HIV− and Unexposed HIV− infants per group for this analysis were: 15, 25 and 22 at 3 months; 9, 17 and 21 at 6 months; 8, 15 and 21 at 9 months; and 5, 12 and 17 at 12 months, respectively. For each boxplot, the median is represented by the horizontal line, the interquartile range by the box and the range by the whiskers. Comparisons (overall effect p values) were calculated from mixed effects maximum likelihood regression models of log-transformed responses that also included time as a continuous effect (Table 2 and Suppl. Fig. 3). (B) Pie charts represent the median proportions of polyfunctional (cells producing 3 cytokines, red), bi-functional (cells producing 2 cytokines, blue) and mono-functional (cells producing 1 cytokine, green) out of the total cytokine CD4 T cell response for the three infant groups, at 3 and at 9 months post-vaccination.

Figure 3

BCG-specific IFN-γ CD8 T cell responses. Background values were subtracted and observed responses are shown as absolute counts (A) and as frequencies (B). For (A), comparisons (overall effect p values tabulated) were calculated from mixed effects maximum likelihood regression models of log-transformed responses that also included time as a continuous effect (Table 2). The numbers of HIV-infected, Exposed HIV− and Unexposed HIV− infants per group for this analysis were: 15, 25 and 22 at 3 months; 9, 17 and 21 at 6 months; 8, 15 and 21 at 9 months; and 5, 12 and 17 at 12 months, respectively. For (B), differences between groups were assessed by the Kruskal-Wallis test (overall effect) and the Mann Whitney U test (p-values are tabulated below each plot). The numbers of HIV-infected, Exposed HIV− and Unexposed HIV− infants per group for this analysis were: 19, 25 and 23 at 3 months; 12, 17 and 22 at 6 months; 8, 15 and 22 at 9 months; and 5, 12 and 17 at 12 months, respectively. Please refer to Materials and Methods for different statistical methods used for absolute number data and frequency data. IL-2 and TNF-α-expressing CD8 T cells are not shown because the detected frequencies were very low.

Figure 4

Levels of soluble cytokines in plasma from whole blood stimulated with BCG for 7 hours. IFN-γ (A), IL-2 (B) and TNF-α (C) were quantified by ELISA and background values (unstimulated blood) subtracted. For A, the numbers of HIV-infected, Exposed HIV− and Unexposed HIV− infants per group were: 14, 21 and 20 at 3 months; 11, 18 and 20 at 6 months; 8, 17 and 19 at 9 months; and 5, 14 and 16 at 12 months, respectively. For B, the numbers were: 14, 20 and 21 at 3 months; 10, 17 and 20 at 6 months; 8, 16 and 20 at 9 months; and 5, 13 and 16 at 12 months, respectively. For C, the numbers were: 14, 21 and 20 at 3 months; 11, 18 and 20 at 6 months; 8, 17 and 20 at 9 months; and 5, 14 and 17 at 12 months, respectively. Differences between groups were assessed by the Kruskal-Wallis test (overall effect) and the Mann Whitney U test (p-values are tabulated to the right of each plot).