Mycobacterium tuberculosis-Specific T-Cell Responses Are Impaired During Late Pregnancy With Elevated Biomarkers of Tuberculosis Risk Postpartum

Abstract

Background

Pregnancy is a risk factor for progression from latent tuberculosis infection to symptomatic tuberculosis. However, how pregnancy influences T-cell responses to Mycobacterium tuberculosis is unknown.

Methods

We measured M. tuberculosis-specific cytokines, T-cell memory markers, and overall CD4⁺ and CD8⁺ T-cell activation by flow cytometry from 49 women (18 with and 31 without HIV) who became pregnant while enrolled in a randomized controlled trial of preexposure prophylaxis for HIV. We analyzed data using COMPASS, an established statistical method for evaluating overall antigen-specific T-cell responses.

Results

Pregnant women with latent tuberculosis infection demonstrated significantly diminished M. tuberculosis-specific CD4⁺ cytokine responses in the third trimester (COMPASS polyfunctional score [PFS], 0.07) compared before (PFS, 0.15), during (PFS, 0.13 and 0.16), and after pregnancy (PFS, 0.14; P = .0084, Kruskal-Wallis test). Paradoxically, M. tuberculosis-specific CD8⁺ cytokines and nonspecifically activated T-cells increased during late pregnancy. Nonspecific T-cell activation, a validated biomarker for progression from latent tuberculosis infection to tuberculosis disease, increased in latent tuberculosis infection-positive women postpartum, compared with latent tuberculosis infection-negative women.

Conclusions

Pregnancy-related functional T-cell changes were most pronounced during late pregnancy. Both M. tuberculosis-specific T-cell changes during pregnancy and increases in immune activation postpartum may contribute to increased risk for tuberculosis progression.

Clinical Trials Registration

NCT0557245.

Pregnancy is a risk factor for tuberculosis disease. The influence of pregnancy on Mycobacterium tuberculosis-specific T-cell responses is unknown. M. tuberculosis-specific CD4⁺T-cell responses were selectively diminished and simplified in the third trimester of pregnancy.

Tuberculosis in pregnancy and postpartum is associated with poor maternal and infant outcomes [1, 2]. Recent large cohort studies demonstrate that pregnancy is associated with increased rates of tuberculosis disease up to 6 months postpartum [3, 4]. However, the immune mechanisms underpinning this observation are not clear. Successful pregnancy requires alteration of the local immune environment [5–7]. These adaptations to fetal development are accompanied by cell-specific systemic immune alterations [8]. Pathogen-specific responses may be influenced by these immune changes, which can alter disease risk [9–11]. Understanding pathogen-specific changes to the immune response during pregnancy may improve efforts to prevent and control disease.

T-cell responses are a critical aspect of Mycobacterium tuberculosis host defense. Multiple lines of evidence, including studies in mouse and nonhuman primate studies of CD4⁺ T-cell knockout or depletion, human genetic studies, and observational studies, demonstrate that CD4⁺ T cells are essential for M. tuberculosis control [12–15]. Interferon-γ (IFN-γ) responses in CD4⁺ T cells are also essential for M. tuberculosis protection [16]. The frequency of CD4⁺ T cells simultaneously producing interleukin 2 (IL-2), tumor necrosis factor (TNF), and IFN-γ (eg, that are polyfunctional) is correlated with M. tuberculosis protection in mice [17]. Other T-cell phenotypes, such as the frequency of antigen-specific central memory (Tcm) and CD8⁺ T cells, contribute to M. tuberculosis host defense [18–21]. Conversely, nonspecifically activated HLA-DR⁺CD4⁺ T cells are a correlate of tuberculosis progression [22]. Because studies of pregnancy commonly begin enrollment during the first or second trimester of pregnancy, they are often unable to capture immune variation prior to and earlier in pregnancy. We leveraged a unique biorepository of samples and data collected longitudinally before, during, and after pregnancy from women in sub-Saharan Africa enrolled in a randomized controlled trial of preexposure prophylaxis for human immunodeficiency virus (HIV) prevention to evaluate the effects of various stages of pregnancy on M. tuberculosis-specific T-cell responses.

METHODS

Study Population

The protocol for this study was approved by the University of Washington Human Subjects Review Committee and ethics review committee at each study site. All participants provided written informed consent. The Partners PrEP Study (NCT0557245) was a randomized clinical trial of antiretroviral preexposure prophylaxis (PrEP) in HIV serodiscordant couples [23]. In total, 4758 participants were enrolled from 9 sites in Kenya and Uganda between 2007 and 2012. Partners with HIV and CD4 cell count > 250, without history of AIDS-defining diagnosis or current use of antiretroviral therapy (ART) were eligible. Participants without HIV underwent monthly HIV testing and had urine pregnancy testing at enrollment and monthly thereafter. Women living with HIV (WLWH) not already on ART who became pregnant were referred for prevention of maternal to child transmission services. Participants were eligible for this analysis if they had at least 2 samples available within 6 months prior to and during incident pregnancy. If available, additional samples were tested from within 6 months postpartum (Figure 1A). For participants with samples available from only before and during pregnancy, paired testing was done on participants with pregnancy samples from > 20 weeks’ gestation. WLWH were provided ART for prevention of maternal to child transmission of HIV.

Figure 1.

Baseline immune characteristics are stable across pregnancy. A, Flowchart of stored sample selection of participants with incidental pregnancies in the Partners PrEP study. B, Fold change of the proportion of CD4⁺ T cells expressing IFN-γ after incubation with ESAT-6 and CFP-10 pooled peptides, compared to vehicle control. C, COMPASS functional score and polyfunctional score stratified by latent tuberculosis infection status. D, Proportion of CD4⁺ T cells and CD8⁺ T cells, stratified by pregnancy trimester and HIV status. Prepregnancy sample, n = 49; 1st trimester, n = 9; 2nd trimester, n = 14; 3rd trimester, n = 24; postpartum, n = 22. Circles indicate HIV-negative study participants, while triangles indicate WLWH. Bars indicate median values. ∗Among participants with pre-pregnancy and pregnancy samples only. ∗∗Includes one participant with a sample collected during pregnancy < 20 weeks of gestation without an evaluable postpartum sample. ∗∗∗Includes two participants with two samples collected during pregnancy. Abbreviations: HIV, human immunodeficiency virus; IFN-γ, interferon-γ; LTBI, latent tuberculosis infection; PBMC, peripheral blood mononuclear cell; PrEP, preexposure prophylaxis; WLWH, women living with HIV.

Flow Cytometry

Cryopreserved peripheral blood mononuclear cells were thawed, washed, and rested overnight. Cells were stimulated and captured on a BD LSRFortessa flow cytometer (Supplementary Table 1) using established protocols (Supplementary Materials).

Identification of Latent Tuberculosis Infection-Positive Participants

Because real-time latent tuberculosis diagnostic tests were not available to us at the time of the parent study, we used a flow cytometry-based assay that demonstrated strong concordance with interferon-γ release assay (IGRA) results in pilot studies and in the published literature. We chose this approach due to its capacity to identify individuals with latent tuberculosis and to identify M. tuberculosis-specific immune response that are missed by traditional IGRA, including polyfunctionality, CD8⁺ T-cell responses, and IFN-γ–independent T-cell responses [24, 25]. We identified latent tuberculosis infection-positive individuals if the frequency of IFN-γ⁺CD4⁺ T cells doubled after ESAT-6 and CFP-10 pooled peptide stimulation compared to dimethyl sulfoxide with a frequency above 0.002% [24, 25].

Statistical Analysis

We used the R package COMPASS to analyze flow cytometry responses. COMPASS uses a Bayesian hierarchical framework to model all observed cell subsets and select those most likely to have antigen-specific responses [26, 27]. COMPASS provides a functional score (FS), which is the proportion of antigen specific subsets detected among all possible ones, and polyfunctional score (PFS), which weighs subsets to favor those producing multiple cytokines. Our primary outcome of interest was the COMPASS PFS from CD4⁺ and CD8⁺ T cells of latent tuberculosis infection-positive women, restimulated with ESAT-6 and CFP10 pooled peptides. Differences in median values were compared among time points of interest using Kruskal-Wallis tests followed by Dunn test to compare individual groups. Comparisons combined women with and without HIV as 1 group except where indicated, to optimize sample size. For studies of T-cell memory and activation markers in M. tuberculosis-specific, cytokine-producing cells, a minimum of 100 cytokine-producing cells were required for inclusion.

RESULTS

Cohort Characteristics and Timing of Sample Collection

We identified 70 participants with incident pregnancy. Of these, 49 of these met criteria for this study, with paired samples collected during pregnancy and another collected less than 6 months prior to the pregnancy; 17 participants had samples collected from 3 time points (pre-, during, and postpregnancy); and 32 participants had samples collected before and during pregnancy (Figure 1A). A total of 117 samples with evaluable results were analyzed: 48 prepregnancy (median 18.7 weeks; interquartile range [IQR], 12.4–22.2), 47 during pregnancy (median 26.9 weeks’ gestation; IQR, 12.7–30.9); and 22 postpartum (median 23.1 weeks; IQR, 18.4–39.6). The median age of study participants was 26.4 years (IQR, 23.2–31.1 years); 36.7% (18/49) were WLWH with median CD4 766 cells/mm³ (IQR, 507–1070 cells/mm³) (Table 1).

Table 1.

Baseline Characteristics of Participants and Timing of Samples With Evaluable Results

Characteristic	Overall	Latent Tuberculosis Infection Positivea	Latent Tuberculosis Infection Negative	RR (95% CI)	P
	Baseline
	n = 49b	n = 19	n = 30
Age, y, median (IQR)	26.4 (23.2–31.3)	26.8 (23.2–31.1)	26.2 (22.7–31.5)	1.02 (.94–1.10)	.624
WLWH	18 (36.7)	6 (31.6)	12 (40.0)	0.79 (.36–1.74)	.565
CD4, cells/mm3, median (IQR)	765.5 (507–1070)	868.5 (588–1272)	730.5 (476.5–1042)	1.00 (1.00–1.00)	.258
	Samples
	n = 117	n = 48	n = 69
Pregnancy stage				1.10 (.83–1.47)	.496
Prepregnancyd	48 (41.0)	18 (37.5)	30 (43.5)
Pregnancy	47 (40.2)	20 (41.7)	27 (39.1)	0.77 (.52–1.14)	.184
1st trimester (0 to < 14 wk)	9 (19.2)	5 (25.0)	4 (14.8)
2nd trimester (14 to < 28 wk)	14 (29.8)	7 (35.0)	7 (25.9)
3rd trimester (≥28 wk)	24 (51.1)	8 (40.0)	16 (59.3)
Postpregnancyc	22 (18.8)	10 (20.8)	12 (17.4)
Timing of sample collection, wk
Prepregnancy, median (IQR)	−18.7 (−22.2 to −12.4)	−18.3 (−22.3 to −11.7)	−18.8 (−22.1 to −13.1)	1.00 (.96–1.03)	.836
During pregnancy, median (IQR)	26.9 (12.7–30.9)	18.9 (11.7–28.1)	28.7 (22.4–32.0)	0.97 (.94–1.00)	.077
Postpregnancy, median (IQR)	23.1 (18.4–39.6)	19.5 (17.3–45.6)	30.5 (21.4–38.7)	1.00 (.97–1.03)	.980

Data are No. (%) except where indicated.

Abbreviations: CI, confidence interval; IFN-γ, interferon-γ; IQR, interquartile range; RR, relative risk; WLWH, women living with HIV.

Latent tuberculosis infection diagnosis determined by 2 × increase of frequency of IFN-γ⁺CD4⁺ cells after stimulation with Mycobacterium tuberculosis antigens over control.

Includes 1 participant whose prepregnancy sample was unevaluable.

Up to 6 months after delivery.

Samples collected up to 6 months before date of conception.

Identification of Latent Tuberculosis Infection-Positive Participants

We defined individuals as latent tuberculosis infection positive by flow cytometry if the frequency of ESAT-6 and CFP-10–specific IFN-γ⁺CD4⁺ T cells was double baseline, and above 0.002% frequency. This measurement correlates well with commercial IGRA tests [24, 25, 28, 29]. Twenty-two of 49 (45%) individuals were latent tuberculosis infection positive (Figure 1B). Increasing the threshold for positivity to 3-fold above background reduced the proportion of latent tuberculosis infection-positive individuals slightly, from 22 (44%) to 19 (41%). These observations are similar in frequency to other studies of women of childbearing age in regions with high M. tuberculosis burden [1]. We used the more inclusive cutoff to ensure we identified all individuals with preexisting immune responses to M. tuberculosis. We identified 3 individuals who were latent tuberculosis infection negative at baseline but latent tuberculosis infection positive by their pregnancy visit. In these situations, we included these individuals as latent tuberculosis infection positive from their first positive result onward. COMPASS FS and PFS of CD4⁺ T cells from all samples strongly correlated with our latent tuberculosis infection measurements (Figure 1C).

CD4⁺ and CD8⁺ T-Cell Frequency Is Consistent Before, During, and After Pregnancy

We identified the frequency of CD4⁺ and CD8⁺ T cells from samples collected before, during, and after pregnancy (gating shown in Supplementary Figure 1A). The frequency of CD4⁺ and CD8⁺ T cells was similar across this study (Figure 1D). We observed diminished frequency of CD4+ T cells in WLWH, irrespective of pregnancy status (Supplementary Figure 1B), and decreased CD8+ T-cell frequency (Supplementary Figure 1C).

M. tuberculosis-Specific CD4⁺ T Cells Polyfunctional Responses Decreased During the Third Trimester of Pregnancy

M. tuberculosis induces a polyfunctional T-cell response that may act independently of IFN-γ, and these cells are critical for M. tuberculosis control [30]. In M. tuberculosis-specific CD4⁺ T cells, PFS was significantly different before, during, and after pregnancy (Figure 2A, gating in Supplementary Figure 2; P = .0084, Kruskal-Wallis test). PFS was significantly decreased in the third trimester compared with prepregnancy (P = .0005), second trimester (P = .0034), and postpregnancy (P < .017), but not during the first trimester (P = .099; Figure 2A). We also evaluated the changes of T-cell responses during pregnancy over time. Using gestational age as a continuous variable, we performed simple linear regression of PFS as a function of gestational time from the second to the third trimester. We found that PFS decreased during this time (P = .0036, R² = 0.22, regression coefficient = −0.0037). FS results were similarly distributed (Figure 2B). We detected fewer M. tuberculosis-specific cytokine-producing subsets during third trimester, and very few subsets producing multiple cytokines simultaneously (Supplementary Figure 3A). We noted diminished single cytokine responses during the third trimester in IL-2⁺CD4⁺ cells compared with the second trimester (P = .026; Supplementary Figure 3B) which trended lower compared with prepregnancy (P = .13), first trimester (P = .12), and postpregnancy (P = .09). CD4⁺TNF⁺ or CD4⁺IFN-γ⁺ responses were similar across pregnancy (Supplementary Figure 3C and D). Therefore, using COMPASS, we were able to detect changes in M. tuberculosis-specific T-cell responses that may be clinically important but not immediately detectable by analyzing single cytokines separately.

Figure 2.

Mycobacterium tuberculosis-specific CD4+ T-cell responses are diminished during the third trimester of pregnancy. A-B, PFS (A) and FS (B) generated from COMPASS analysis of CD4+ T cell responses, stratified by pregnancy trimester. C-D, PFS (C) and FS (D) generated from COMPASS analysis of CD8+ T cell responses, stratified by pregnancy trimester. Circles indicate an HIV-negative study participant, while triangles indicate a WLWH. Bars indicate median values, Kruskal-Wallis test. Column-to-column comparisons are made with Dunn test. Abbreviations: FS, functional score; HIV, human immunodeficiency virus; PFS, polyfunctional score; WLWH, women living with HIV.

M. tuberculosis-Specific CD8⁺ T Cells Increase in Activity During the Second and Third Trimester of Pregnancy

CD8⁺ T cells contribute to M. tuberculosis host defense and are measured in latent tuberculosis infection diagnostic tests [31, 32], but little is known about how pregnancy impacts M. tuberculosis-specific CD8⁺ T-cell responses. We found low but detectable M. tuberculosis-specific CD8⁺ T-cell cytokine responses in most samples from latent tuberculosis infection-positive participants (gating strategy in Supplementary Figure 2). The PFS and FS in CD8⁺ T cells changed significantly over pregnancy (P = .0033 and P = .038, respectively, Kruskal-Wallis test; Figure 2C and 2D). Median PFS and FS in CD8⁺ T cells was nearly undetectable before pregnancy and significantly increased in the second trimester from prepregnancy (P = .015), the first trimester (P = .023), and postpregnancy (P = .034). The complexity of the CD8⁺ T-cell response increased during the second and third trimester, and we detected subsets producing multiple cytokines simultaneously (Supplementary Figure 4A). We did not detect changes in IL-2, TNF, and IFN-γ⁺ CD8⁺ T-cell responses when evaluated singly, but we found a trend toward increased responses during the second and third trimester (Supplementary Figure 4B–4D). Taken together, M. tuberculosis-specific CD8⁺ T-cell responses increased during the second and third trimester but M. tuberculosis-specific CD4⁺ T-cell responses diminished.

CD4⁺ T-Cell Responsiveness to a Mitogen Is Altered During the Third Trimester

Clinical laboratory tests for latent tuberculosis infection include mitogen controls to assess overall T-cell reactivity; impacts to mitogen-induced CD4⁺ and CD8⁺ T-cell responses during pregnancy directly alter latent tuberculosis infection diagnostic test performance [33]. We compared T-cell responses after stimulation with phorbol 12-myristate 13-acetate and ionomycin (PMA/ionomycin) over pregnancy in all study participants. We evaluated these T-cell responses using traditional flow cytometry analysis, not COMPASS, as COMPASS is designed to detect antigen-specific responses observed at low frequency. The frequency of IL-2⁺CD4⁺ T cells responding to PMA/ionomycin were significantly changed over the duration of pregnancy (P < .0001, Kruskal-Wallis test; Figure 3A). IL-2⁺CD4⁺ responses in the third trimester were increased compared to prepregnancy (P = .022), first trimester (P = .0006), and postpregnancy (P < .0001). TNF⁺CD4⁺ responses were also significantly increased during the third trimester compared to first trimester (P = .0073) and postpregnancy (P = .0002; P = .0011 overall; Figure 3A). IFN-γ⁺CD4⁺ cells decreased across pregnancy (P = .0036). The third trimester response was significantly diminished compared to prepregnancy (P = .016), first trimester (P = .026), second trimester (P = .049), and postpregnancy (P = .0001; Figure 3A). IL-2⁺, TNF⁺, and IFN-γ⁺ CD4⁺ T-cell frequencies were each significantly different before and after pregnancy (P = .006, P = .008, and P = .04, respectively). These data demonstrate distinct patterns of CD4⁺ T-cell activation across pregnancy as compared to M. tuberculosis-specific T cells.

Figure 3.

PMA/ionomycin-stimulated CD4⁺ T-cell cytokine responses are dynamic across pregnancy. A, Proportion of IL-2⁺CD4⁺, TNF⁺CD4⁺, or IFN-γ⁺CD4⁺ T cells stratified by pregnancy trimester. B, Proportion of IL-2⁺CD8⁺, TNF⁺CD8⁺, and IFN-γ⁺CD8⁺ T cells stratified by pregnancy trimester. Black dots indicate an HIV-negative study participant, while gold triangles indicate an HIV-positive study participant. Bars indicate median values. P values calculated by Kruskal-Wallis test. Column-to-column comparisons are made with Dunn test. Abbreviations: HIV, human immunodeficiency virus; IFN-γ, interferon-γ; IL-2, interleukin 2; TNF, tumor necrosis factor.

Mitogen-Induced TNF⁺CD8⁺ T-Cell Responses Increase During Pregnancy

We measured PMA/ionomycin-stimulated CD8⁺ T-cell responses over the course of pregnancy in all participants. The frequency of IL-2⁺CD8⁺ cells was unchanged over pregnancy (P = .14, Kruskal-Wallis test; Figure 3B). TNF⁺CD8⁺ T-cell frequency changed over the course of pregnancy (P = .0084; Figure 3B), increasing in the third trimester compared with the first trimester (P = .015) and postpregnancy (P = .0018). IFN-γ⁺CD8⁺ T-cell frequencies were similar (Figure 3B). These responses were distinct from the changes observed in M. tuberculosis-specific CD8⁺ T cells, where multiple subsets increased in frequency during the second and third trimester.

M. tuberculosis-Specific Effector Memory T-Cell Populations Diminish During the Third Trimester

T cells display cell surface markers CD45RA and CCR7, which define the T-cell memory phenotype [34]. Naive (CCR7⁺CD45RA⁺), central memory (Tcm; CCR7⁺CD45RA⁻), and effector memory (Tem; CCR7⁻CD45RA⁻) T cells display distinct functional profiles and tissue homing sites, and Tcm are important for M. tuberculosis control [18, 35, 36]. The frequency of total CD4⁺ and CD8⁺ T cells expressing CD45RA and CCR7 did not change over pregnancy (Supplementary Figure 5A–5C). We evaluated CD45RA and CCR7 expression in M. tuberculosis-specific CD4⁺ T cells, which were ESAT-6 and CFP-10 peptide-incubated, cytokine-producing, CD4⁺ T cells from latent tuberculosis infection-positive individuals. The proportion of M. tuberculosis-specific naive and Tcm CD4⁺ T cells remained stable throughout pregnancy (data not shown), but the frequency of M. tuberculosis-specific Tem decreased during the third trimester (P = .037, Kruskal-Wallis test; Figure 4A) compared to prepregnancy (P = .049), first trimester (P = .005), second trimester (P = .014), and postpartum (P = .031). Viewed proportionally, naive CD4⁺ cells increased while Tem decreased during the third trimester of pregnancy compared to the first trimester, second trimester, and postpartum, without substantial changes to Tcm frequency (Figure 4B). We did not detect adequate M. tuberculosis-specific CD8⁺ T cells to evaluate T-cell phenotypes in this population. In summary, Tem and IFN-γ⁺CD4⁺ cells were reduced during the third trimester, but without increases in the Tcm population, which is important for M. tuberculosis host defense and may predict worsened M. tuberculosis control.

Figure 4.

The proportion of Mycobacterium tuberculosis-specific naive effector memory T cells decreases during the third trimester of pregnancy. A, Overall proportion of Tem (CCR7⁻CD45RA⁻) among all cytokine-producing CD4⁺ T cells after restimulation with ESAT-6 and CFP-10 antigens, stratified by pregnancy status. Black dots indicate an HIV-negative study participant, while gold triangles indicate a sample from WLWH study participant. P values calculated by Kruskal-Wallis test. Column-to-column comparisons are made with Dunn test. B, Pie chart of Boolean-gated T helper memory phenotype frequencies, comparing the third trimester with other time points, connected by overhead bar. Arc: blue, CCR7⁺ cells; red, CD45RA⁺ cells. Pie slice: red, naive CD4+ T cells; blue, Tem CD4+ T cells; gray, Tcm; white, TEMRA cell. Signifcance determined by permutation test of 10 000 iterations. Abbreviations: HIV, human immunodeficiency virus; Tcm, central memory T cell; Tem, effector memory T cell; WLWH, women living with HIV.

Nonspecific T-Cell Activation Increases During Pregnancy

Increased nonspecific CD4⁺ T-cell activation is a correlate of risk for developing M. tuberculosis disease [22, 37]. We measured the frequency of HLA-DR⁺CD38⁺CD4⁺ and CD8⁺ T cells across pregnancy, using fluorescence minus 1 controls (Supplementary Figure 6). The frequency of HLA-DR⁺CD38⁺CD4⁺ T cells significantly changed over pregnancy (P = .0001, Kruskal-Wallis test; Figure 5A), highest in the third trimester compared to prepregnancy (P = .008), first trimester (P = .001), second trimester (P = .02), or postpartum (P = .01). HLA-DR⁺CD38⁺CD8⁺ T-cell frequency also changed similarly to CD4⁺ T cells (P = .002, Kruskal-Wallis test; Figure 5B).

Figure 5.

Nonspecific T-cell activation increases during pregnancy and latent tuberculosis infection-positive women demonstrate increased T-cell activation postpartum. A and B, Proportion of (A) HLA-DR⁺CD38⁺CD4⁺ T cells and (B) HLA-DR⁺CD38⁺CD8⁺ T cells stratified by pregnancy status. C and D, Proportion of (C) HLA-DR⁺CD38⁺CD4⁺ T cells from samples collected prepregnancy and (D) proportion of HLA-DR⁺CD38⁺CD4⁺ T cells from samples collected postpartum stratified by latent tuberculosis infection status of the participant. Circles indicate an HIV-negative study participant, while triangles indicate sample from WLWH. Bars indicate median values. P values calculated by Kruskal-Wallis test. Column-to-column comparisons are made with Dunn test in multiple group comparisons or Mann-Whitney U test for tests between 2 groups. Abbreviations: HIV, human immunodeficiency virus; LTBI, latent tuberculosis infection; WLWH, women living with HIV.

We leveraged the observation that CD4⁺ T-cell activation is a biomarker for tuberculosis risk to evaluate if pregnancy might increase the risk for tuberculosis progression from asymptomatic to symptomatic disease. We hypothesized that selective increases in nonspecific T-cell activation would be present in latent tuberculosis infection-positive women after pregnancy due to their increased risk for postpartum M. tuberculosis progression but not for latent tuberculosis infection-negative women. Prepregnancy, the proportion of HLADR⁺CD38⁺CD4⁺ T cells was equivalent between latent tuberculosis infection-negative and latent tuberculosis infection-positive individuals (Figure 5C). Postpartum, latent tuberculosis infection-positive individuals demonstrated significantly higher HLA-DR⁺CD38⁺CD4⁺ T cells as compared to latent tuberculosis infection-negative individuals (P = .008, Mann-Whitney test; Figure 5D). Immune changes of pregnancy in latent tuberculosis infection-positive individuals may substantially contribute to overall T-cell activation in the postpartum period and predict increased risk for tuberculosis progression.

DISCUSSION

In this study, we performed a comprehensive analysis of pregnancy stage on M. tuberculosis-specific T-cell phenotypes in sub-Saharan African women, including WLWH. This is, to our knowledge, the first longitudinal assessment of systemic T-cell responses to M. tuberculosis in pregnancy that includes samples collected prior to pregnancy. Late pregnancy is associated with multiple changes to M. tuberculosis-specific T-cell function: (1) overall M. tuberculosis-specific CD4⁺ T-cell responses diminished and simplified; (2) M. tuberculosis-specific CD4⁺ Tem, cells that strongly produce IFN-γ, decreased [34]; (3) M. tuberculosis-specific CD8⁺ T-cell responses increased; and (4) nonspecific T-cell activation increased. Nonspecific T-cell activation increased in latent tuberculosis infection-positive individuals compared to latent tuberculosis infection-negative individuals postpartum, despite being equivalent prepregnancy. Increased nonspecific T-cell activation, a correlate of risk for M. tuberculosis disease in latent tuberculosis infection-positive individuals, supports the observation that tuberculosis risk increases in the early postpartum period [22]. Together, late pregnancy induces complex changes in the known T-cell response to M. tuberculosis associated with tuberculosis progression, and latent tuberculosis infection-positive women develop increased correlates of risk for tuberculosis in the immediate postpartum period.

We observed distinct changes between M. tuberculosis-specific and more generalized mitogen (PMA/ionomycin)-induced T-cell cytokine responses (Table 2). These results have direct implications for the performance of latent tuberculosis infection diagnostics in pregnant women. M. tuberculosis-specific CD4⁺ T-cell responses decreased during the third trimester but increased in the second trimester, and M. tuberculosis-specific CD8⁺ T-cell responses were low but increased during the second and third trimester. By contrast, overall mitogen responses in CD4⁺ T cells, especially IL-2 responses, increased during late pregnancy. In other cohorts, discordance between the 2 commonest latent tuberculosis infection tests, tuberculin skin test (TST) and IGRA, has been noted, especially among women in the late second and early third trimester, which may be partly due to the changes in the immune responses of pregnancy [33, 38]. Part of the difference may be due to the absence of mitogen testing in most TSTs. Understanding this discordance in pregnancy may have implications for tuberculosis progression. Among those with positive IGRA but negative TST, M. tuberculosis-specific IL-2 responses associated with postpartum tuberculosis progression from asymptomatic infection to symptomatic disease [39]. In combination, discordance between M. tuberculosis-specific and nonspecific immune responses may contribute to increased false-negative latent tuberculosis infection testing during the third trimester and may contribute to M. tuberculosis progression.

Table 2.

Model Summary

T Cells	Prepregnancy	1st Trimester	2nd Trimester	3rd Trimester	Postpartem
Mycobacterium tuberculosis-specific T cells
CD4+	↔	↔	↓	↓↓↓	↔
CD8+	↓↓↓	↓↓	↑↑	↑	↓↓
Mitogen-stimulate T cells
CD4+	↔	↔	↔	↑↓	↑↓
CD8+	↔	↔	↑	↔	↔
Nonspecific T-cell activation
HLA-DR+CD38+CD4+	↔	↓	↑↑	↑↑↑	↑

We observed distinct changes in Mycobacterium tuberculosis-specific T-cell responses as compared to mitogen-stimulated T cells. Arrows indicate the overall frequency of the responses measured. A downward arrow followed by an upward arrow indicates diminished IL-2/TNF responses with simultaneous increases in IFN-γ⁺ cell frequency, and an upward arrow followed by a downward arrow indicates increases in IL-2/TNF⁺ T-cell frequency with decreases in IFN-γ⁺ responses.

Abbreviations: IFN-γ, interferon-γ; IL-2, interleukin 2; TNF, tumor necrosis factor.

M. tuberculosis-specific CD8⁺ T-cell responses increased during late pregnancy, when CD4⁺ T-cell responses were diminished. The role of CD8⁺ T cells in tuberculosis pathogenesis is uncertain. In children, M. tuberculosis-specific CD8⁺ T-cell frequencies are correlated with M. tuberculosis disease [21]. In animals, CD8⁺ T-cell responses correlate strongly with M. tuberculosis bacterial burden in the lung [40, 41]. CD8⁺ T cells recognize antigens presented on major histocompatibility complex I, which presents intracellular antigens. CD8⁺ T cells preferentially recognize cells with increased bacterial burden [42]. During active intracellular infection, these responses may increase proportionally. It is unclear if some of these responses reflect emerging tuberculosis disease or if there are general postpartum factors that enhance CD8⁺ responses. CD8⁺ T-cell responses may increase to compensate for diminished CD4⁺ T-cell responses but may be unable to fully reproduce CD4⁺ T-cell immune control.

The strengths of this study include its longitudinal nature, including samples obtained before, during, and after pregnancy. Most studies of the effects of pregnancy enroll women during pregnancy and often focus on the second and third trimester, which do not capture prepregnancy and first trimester. Several T-cell phenotypes change between prepregnancy and the first trimester, and other phenotypes differed between prepregnancy and postpartum. Local inflammation is critical for embryonic implantation, suggesting that immune changes begin immediately upon fertilization [43, 44]. Another strength was the use of COMPASS. COMPASS analysis revealed 2 findings that were not observable using standard data analysis techniques. This sensitive analysis permitted us to identify diminished complexity of the M. tuberculosis-specific CD4⁺ T-cell response and increases in overall CD8⁺ activity. Taken together, no single cytokine response accounted for the changes to the M. tuberculosis-specific T-cell response, suggesting broad changes in the functional capacity of T cells during the third trimester.

Weaknesses of this study include the lack of tuberculosis disease outcomes data. This study used repository samples from a trial not originally designed to evaluate tuberculosis. However, we performed detailed immune characterization of the M. tuberculosis-specific immune response across pregnancy. Importantly, latent tuberculosis infection-positive women had increased T-cell activation postpartum, which is a validated biomarker of tuberculosis disease progression [22]. IGRA or TST was not performed in the parent study, which prevents correlation with clinically validated latent tuberculosis infection testing. We identified latent tuberculosis infection-positive participants using a well-accepted alternate approach that correlates with latent tuberculosis infection diagnosis [28, 29]. Generally, flow-based testing is more sensitive to detect prior exposure to M. tuberculosis antigens than IGRA, so this approach permits us to evaluate a broader range of individual immune responses. Errors introduced by including individuals without preexisting M. tuberculosis exposures may reduce the study sensitivity, but should not detract from the observations made here. M. tuberculosis-specific T-cell responses during pregnancy were generally independent of HIV status, although direct comparisons were limited due to small sample size. The Partners studies featured women with regular exposure to HIV [23]. This regular exposure may influence nonspecific T-cell activation, as HIV exposed, uninfected infants develop altered T-cell responses [45]. However, its role in systemic T-cell responses and its role in tuberculosis susceptibility in adults are unknown. We hypothesize that our findings reflect changes due to pregnancy, not HIV exposure, because they predominate in late pregnancy with return to pre- and early pregnancy levels postpartum.

Tuberculosis is the third leading cause of death among women of child-bearing age in high-burden areas and is associated with mortality in pregnant women and their children [1, 38]. Systemic changes in the T-cell response during the third trimester may contribute to increased risk for tuberculosis. T-cell cytokine responses are reduced and simplified, T-cell memory phenotypes skewed toward naive phenotypes, and nonspecific activation is enhanced, which all are independently associated with tuberculosis risk. Data suggest that isoniazid preventative therapy is associated with adverse pregnancy outcomes, with a signal that this may occur more often if initiated early in pregnancy [46]. Late pregnancy may be advantageous to target for shorter course regimens. Understanding host factors that place pregnant mothers at risk for tuberculosis may provide insight into optimal public health strategies for tuberculosis control in this high-risk population.

Supplementary Data

Supplementary materials are available at The Journal of Infectious Diseases online. Supplementary materials consist of data provided by the author that are published to benefit the reader. The posted materials are not copyedited. The contents of all supplementary data are the sole responsibility of the authors. Questions or messages regarding errors should be addressed to the author.

Notes

Acknowledgments. The authors thank the study participants and their families. We acknowledge the support of Harald Haugen on database management of clinical samples; and the clinical teams who obtained informed consent and collected samples for this biorepository. We also thank Chetan Seshadri for helpful technical conversations.

Author contributions. A. S. contributed to the design, data acquisition, analysis, and manuscript preparation. J. E. contributed data analysis, manuscript preparation, sample collection, and transport. T. L. contributed statistical analysis and software. B. R. contributed data analysis, statistical support, and manuscript editing. S. H. and H. H. contributed sample transport, sample storage, database management. N. M. and A. M. contributed sample collection/storage, participant enrollment. C. C. and J. B. contributed cohort enrollment, funding, sample storage, manuscript editing. J. L. contributed cohort enrollment, sample storage, funding, manuscript editing. G. J. S. contributed experimental design, data analysis, funding, manuscript editing. S. L. and J. A. S. contributed experimental design, data analysis, grant funding, manuscript preparation and editing.

Financial support. This work was supported by National Institute of Child Health and Human Development (grant number R21-HD098746); National Institute of Allergy and Infectious Diseases (grant numbers K23-AI120793 and R01-AI136921); and Bill and Melinda Gates Foundation (grant numbers 26469, 41185, and 47674).

Potential conflicts of interest. All authors: No reported conflicts of interest. All authors have submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest. Conflicts that the editors consider relevant to the content of the manuscript have been disclosed.

Presented in part: 51st Union World Conference on Lung Health, virtual, 19 October 2020.