Tuberculosis (TB) and HIV infection are independently associated with elevated serum concentrations of tumour necrosis factor receptor type 1 and β₂-microglobulin, respectively

Abstract

The aim of this study was to identify immune markers that are independently associated with HIV infection or TB in vivo. Using commercially available assays, we measured concentrations of five immune markers in sera from 175 out-patients attending medical clinics in Cote D'Ivoire and Ghana, West Africa. Patients were categorized into groups with TB only (TB⁺HIV⁻, n = 55), TB and HIV co-infection (TB⁺HIV⁺, n = 50), HIV infection only (TB⁻HIV⁺, n = 35), or neither infection (TB⁻HIV⁻, n = 35). TB⁺HIV⁺ and TB⁻HIV⁺ groups were matched for blood CD4⁺ lymphocyte count. Mean ±s.d. concentrations of β₂-microglobulin were similarly increased in both the TB⁻HIV⁺ (5·3 ± 2·1 μg/ml, P < 0·0001) and the TB⁺HIV⁺ (5·0 ± 1·5 μg/ml, P < 0·0001) groups compared with the TB⁻HIV⁻ group (2·2 ± 1·8 μg/ml), but were only slightly increased in the TB⁺HIV⁻ group (3·2 ± 1·8 μg/ml, P = 0·01). In contrast, mean serum concentrations of soluble tumour necrosis factor receptor type I (sTNF-RI) were similarly elevated in the TB⁺HIV⁻ (1873 ± 799 pg/ml, P < 0·0001) and TB⁺HIV⁺ (1797 ± 571 pg/ml, P < 0·0001) groups compared with uninfected subjects (906 ± 613 pg/ml), but there was only a small increase in sTNF-RI in the TB⁻HIV⁺ group (1231 ± 165 pg/ml, P = 0·03). Both TB and HIV infection were associated with substantial elevation of serum concentrations of soluble CD8, soluble CD54, and sTNF-R type II. Analysis of additional samples from groups of TB⁺HIV⁻ and TB⁺HIV⁺ patients receiving anti-TB treatment showed significant and equal reductions in mean serum sTNF-RI concentrations, but no significant change in mean β₂-microglobulin. Thus, serum β₂-microglobulin and sTNF-RI serve as relatively independent markers of HIV infection and TB, respectively, in studies of co-infected persons.

INTRODUCTION

World-wide, TB is one of the most common opportunistic infections in patients with HIV-1 infection [1]. The major burden of co-infection with these two diseases is in sub-Saharan Africa [2] where, as in other parts of the world, HIV-1 infection has had a profound effect on the incidence, clinicopathological features, and radiological presentation of TB [3–5]. Moreover, TB modulates HIV-1 infection reciprocally as a result of its effects on the immune system. Activation of mononuclear cell pools by Mycobacterium tuberculosis enhances HIV-1 replication in vitro and in vivo[6–10] and leads to increased plasma HIV-1 load in co-infected individuals [9]. By these means, TB may act as a cofactor that accelerates the decline in immune function and shortens survival in HIV-infected persons [11], although the evidence for this is not conclusive at present (reviewed in [12]).

In addition to TB, it has been hypothesized that chronic immune activation due to helminth infections, recurrent malaria, and water-borne pathogens may also accelerate the progression of HIV-1 infection to AIDS in persons in Africa [13,14]. Furthermore, it has been suggested that treatment of TB and other co-infections may reverse or diminish this adverse effect—a hypothesis that is supported in part by the finding of significant reductions in plasma viral load in HIV-infected persons following treatment of acute co-infections [15,16]. However, no significant reduction in mean plasma HIV-1 load was seen in two studies of patients treated for TB in sub-Saharan Africa [17,18], in contrast to a study of a small group of western patients treated for TB in whom a reduction was seen [9]. These contrasting findings may be due to differences in immunological activation and immunoregulation that exist between African and western subjects [19,20].

Further studies are clearly needed to understand more fully the immunological mechanisms underlying the copathogenesis of TB and HIV-1 in vivo and to evaluate the role of immune-modulating drugs used in conjunction with anti-TB treatment in co-infected patients. Moreover, there is a great need for studies to determine the rate of progression of HIV-1 infection and the impact of copathogens on progression of HIV-1 infection in persons in Africa. However, assessment of the level of immunological activation resulting from the presence of a copathogen cannot be reliably made by measuring soluble immune markers in serum of HIV-infected persons if concentrations of those markers are also markedly affected by HIV-1 infection itself. Conversely, the use of serum immune correlates of progression of HIV infection in field studies may be confounded if concentrations of those markers are transiently affected by opportunistic infections.

To our knowledge, immune markers in serum that are independently associated with either HIV infection or TB have not previously been clearly established. Thus, in this study we measured concentrations of five soluble immune receptors in serum samples obtained from West African subjects with TB only, HIV infection only, TB and HIV co-infection, or neither disease.

PATIENTS AND METHODS

Patients

In the initial cross-sectional part of the study, subjects were categorized into one of four groups: patients with TB only (TB⁺HIV⁻, n = 55), TB and HIV co-infection (TB⁺HIV⁺, n = 50), HIV infection alone (TB⁻HIV⁺, n = 35), or neither infection (TB⁻HIV⁻, n = 35). Samples from adult patients with TB were selected from among those collected in a cohort study of response to anti-TB treatment as reported previously [21]. In brief, patients with sputum smear-positive pulmonary TB or extrapulmonary TB (based on clinical grounds) were recruited at the time of diagnosis at two large out-patient treatment centres in Abidjan, Cote D'Iviore. Following consent, blood was obtained for HIV testing and lymphocyte subtyping (FACScan; Becton Dickinson, San Jose, CA). For this study, we randomly selected serum samples from HIV⁺ persons and then selected age- and sex-matched samples from among the HIV⁻ subjects.

Serum samples from HIV⁺ and HIV⁻ patients who did not have TB were selected from among patients attending a clinic for sexually transmitted diseases in Abidjan. Following consent, blood samples were obtained for HIV testing and lymphocyte subtyping. Anonymous unlinked serum aliquots were selected and matched with samples from the TB⁺HIV⁺ and TB⁺HIV⁻ groups based on the patient's HIV serologic status, age, sex and blood CD4⁺ lymphocyte count.

To analyse further the use of selected immune markers, additional serum samples were also obtained prospectively from patients with smear-positive pulmonary TB receiving treatment at the Komfo Anokye Teaching Hospital, Kumasi, Ghana. Four serial serum samples were obtained during the first 3 months of directly observed short-course treatment from HIV⁻ (n = 10) and HIV⁺ (n = 10) patients consecutively diagnosed as having smear-positive pulmonary TB. Treatment included 2 months of daily streptomycin, isoniazid, rifampicin and pyrazinamide, followed by 6 months of daily isoniazid together with either thioacetazone (TB⁺HIV⁻ patients) or ethambutol (TB⁺HIV⁺ patients). All patients were fully compliant with treatment, and all follow-up sputum smears were negative after 2 months of treatment.

Serum markers

Concentrations of immune markers in patient sera stored at −80°C were measured using commercially available assays. The markers included β₂-microglobulin (β₂-MG), soluble CD8 (sCD8), soluble CD54 (sCD54, soluble intercellular adhesion molecule-1), and soluble tumour necrosis factor receptor types I and II (sTNF-RI and sTNF-RII). Serum sCD8, β₂-MG, and sCD54 were measured using ELISAs (T Cell Diagnostics/Endogen Inc., Woburn, MA), and serum sTNF-RI and sTNF-RII were measured using Quantikine ELISAs (R&D Systems, Minneapolis, MN).

Data analysis

Prism version 2.0 software (GraphPad Software Inc., San Diego, CA) was used for the analysis. The proportions of extrapulmonary TB cases in the TB⁺HIV⁺ and TB⁺HIV⁻ groups were compared by χ² test, and levels of serum markers and CD4⁺ lymphocyte counts in the patient groups were compared by t-test. Pearson's correlation coefficients were calculated to assess the relationship between concentrations of serum markers and blood CD4⁺ lymphocyte counts. Statistical significance was defined as P ≤0·05.

RESULTS

Patient characteristics

Patient characteristics and blood CD4⁺ lymphocyte counts are summarized in Table 1. As a result of the matched sample selection, age and sex of patients were similar in all four groups and there was no significant difference between CD4⁺ lymphocyte counts in the two groups of HIV-infected patients. The proportions of cases of extrapulmonary TB in the TB⁺HIV⁺ and TB⁺HIV⁻ groups were not statistically different (P = 0·3; χ² test). Five serum samples in the TB⁺HIV⁺ group were not analysed because of poor specimen quality; therefore, the TB⁺HIV⁻ and TB⁺HIV⁺ groups were of unequal size.

Table 1.

Patient characteristics, site of TB, and blood CD4⁺ lymphocyte counts in patients with TB only (TB⁺HIV⁻), HIV infection only (TB⁻HIV⁺), dual infection (TB⁺HIV⁺) and in uninfected controls (TB⁻HIV⁻)

	TB+HIV+	TB+HIV−	TB−HIV+	TB−HIV−
No. subjects	50	55	35	35
No. (%) male	33 (66)	37 (67)	21 (60)	21 (60)
Mean (s.d.) age (years)	35·4 (9·4)	31·8 (11·6)	32·2 (11·7)	30·6 (15·7)
No. (%) extrapulmonary TB	12 (24·0)*	9 (16·4)*	–	–
Mean (s.d.) CD4 count × 106/l	376 (269)**	1044 (523)	391 (269)**	998 (335)

^*,**No statistical differences between groups for either parameter.

Elevated serum β₂-MG is predominantly associated with HIV infection

Although it was not possible to examine a more extensive list of immune markers in this study, we have previously demonstrated that TNF-α[17], soluble CD14 [22] and soluble CD25 (unpublished data) are not independently associated with HIV infection or TB. Mean concentrations of all five markers in this study were substantially increased in the dually infected (TB⁺HIV⁺) patients compared with the uninfected (TB⁻HIV⁻) group (Figs 1 and 2). However, differential elevation of these markers was observed in the groups with either TB or HIV infection alone.

Fig. 1.

Serum concentrations of (a) β₂-microglobulin (β₂-MG, μg/ml), (b) sCD8 (U/ml), (c) sCD54 (ng/ml) in patients with TB (TB⁺HIV⁻, n = 55), HIV infection (TB⁻HIV⁺, n = 35), or dual infection (TB⁺HIV⁺, n = 50), and in uninfected controls (TB⁻HIV⁻, n = 35). Box and whisker plots indicate the median (bar), 25th and 75th centiles (box), and 10th and 90th centiles (whiskers).

Fig. 2.

Serum concentrations of (a) sTNF-RI (pg/ml) and (b) sTNF-RII (pg/ml) in the four study groups. Patient groups and box and whisker plot format are as in Fig. 1.

Mean ±s.d. serum concentrations of β₂-MG were markedly higher in the TB⁻HIV⁺ (5·3 ± 2·1 μg/ml; P < 0·001) and TB⁺HIV⁺ (5·0 ± 1·5 μg/ml; P < 0·001) groups compared with the TB⁻HIV⁻ group (2·2 ± 1·8 μg/ml) (Fig. 1a). In contrast, the increase in the mean serum β₂-MG level in the TB⁺HIV⁻ group was small (3·2 ± 1·8 μg/ml; P = 0·02). It was notable that there was no significant difference in levels of this marker comparing the TB⁻HIV⁺ and TB⁺HIV⁺ groups (5·0 ± 1·5 versus 5·3 ± 2·1; P = 0·5). Moreover, the significant negative correlation between CD4⁺ lymphocyte counts and serum β₂-MG concentrations observed in the TB⁻HIV⁺ group (r = −0·33, P = 0·05) was also seen in the TB⁺HIV⁺ group (r = −0·29, P = 0·05). Together, these data suggest that the increased serum β₂-MG levels were predominantly associated with HIV infection rather than TB and reflect progression of HIV infection both in the presence and absence of active TB.

In comparison with the TB⁻HIV⁻ group, serum sCD8 concentrations were significantly increased in both HIV-infected groups (TB⁺HIV⁺, P < 0·0001, and TB⁻HIV⁺, P < 0·001), but were increased to a much smaller degree in those with TB only (TB⁺HIV⁻, P = 0·03) (Fig. 1b). However, mean ± s.d. serum levels of sCD8 were significantly higher in the TB⁺HIV⁺ group (853 ± 320 U/ml) compared with the TB⁻HIV⁺ group (655 ± 324 U/ml; P < 0·01), suggesting that both TB and HIV contributed to the greatly elevated levels of sCD8 in the co-infected group. This fact, together with the finding of no significant correlation between serum sCD8 concentrations and CD4⁺ lymphocyte counts in the TB⁻HIV⁺ group (r = −0·12, P > 0·2), suggested that sCD8 was a less independent marker of HIV infection compared with β₂-MG.

Significantly increased levels of sCD54 were seen in samples from patients with TB alone (P < 0·001), HIV infection alone (P < 0·001), and dual infection (P < 0·0001), with no apparent disease specificity (Fig. 1c). In the TB⁺HIV⁺ group, there was no significant correlation of serum sCD54 concentrations with CD4⁺ lymphocyte count (r = − 0·13, P > 0·2).

Elevated serum TNF-RI is associated with TB

Mean ±s.d. serum concentrations of sTNF-RI were substantially higher in the TB⁺HIV⁻ group (1873 ± 799 pg/ml; P < 0·001) and the TB⁺HIV⁺ group (1797 ± 571 pg/ml; P < 0·001) compared with uninfected TB⁻HIV⁻ subjects (906 ± 614 pg/ml) (Fig. 2a). In contrast, the increase in mean sTNF-RI concentration in the TB⁻HIV⁺ group was comparatively small (1232 ± 952 pg/ml; P = 0·03). It was notable that there was no significant difference in mean levels of serum TNF-RI in the TB⁺HIV⁺ group (1797 ± 571 pg/ml) compared with the TB⁺HIV⁻ group (1873 ± 799 pg/ml; P = 0·6), further suggesting that the increased levels of the marker predominantly reflected active TB. Moreover, there was no significant correlation between serum concentrations of TNF-RI and CD4⁺ lymphocyte count in the TB⁻HIV⁺ group (r = −0·16, P = 0·3). Together, these data suggest that the increased shedding of sTNF-RI is predominantly associated with TB rather than HIV infection.

Soluble TNF-RII has been shown to be a useful prognostic marker in asymptomatic HIV-infected persons [23]. However, in this study significantly increased levels of sTNF-RII were seen not only in persons with HIV infection alone but also in those with TB alone compared with the uninfected group (P < 0·001 for each); the highest levels were seen in those with dual infection (P < 0·001). Thus, TNF-RII showed no disease specificity. The correlation between TNF-RII and β₂-MG, both recognized markers of HIV infection, was very strong in the TB⁻HIV⁺ patients (r = 0·83, P = < 0·0001), but considerably less strong in the TB⁺HIV⁺ patients (r = 0·458, P < 0·001). This further supports our finding that TB co-infection in HIV-infected persons leads to differential effects on levels of these immune markers. In the TB⁺HIV⁺ group there was a significant negative correlation between CD4⁺ lymphocyte counts and levels of both TNF type I (r = −0·32, P < 0·05) and type II (r = −0·415, P < 0·01) receptors.

Reduction of serum TNF-RI but not β₂-MG during treatment of TB

Data from the cross-sectional study suggest that increased serum sTNF-RI concentrations were predominantly associated with TB rather than HIV infection. We therefore hypothesized that serum levels of TNF-RI should decline significantly in both TB⁺HIV⁺ and TB⁺HIV⁻ patients during anti-TB treatment. Furthermore, results suggested that β₂-MG was predominantly associated with HIV infection rather than TB and we therefore hypothesized that serum concentrations of β₂-MG would not be affected by anti-TB treatment in either group of patients. Indeed, in TB⁺HIV⁺ and TB⁺HIV⁻ subgroups of patients, similar significant reductions in mean sTNF-RI concentrations were observed during the first 3 months of treatment (Fig. 3). Furthermore, mean serum concentrations of β₂-MG did not significantly change in either patient group during treatment (Fig. 3), which is consistent with our previous finding that mean plasma HIV-1 load also did not change significantly in these TB⁺HIV⁺ patients during anti-TB treatment [17].

Fig. 3.

Mean ±s.e.m. serum concentrations of β₂-microglobulin (β₂-MG) and sTNF-RI in additional groups of patients with (a) TB only (TB⁺HIV⁻, n = 10) or (b) TB and HIV co-infection (TB⁺HIV⁺, n = 10). The graphs display the change in mean concentration of each immune marker as a percentage of the mean baseline concentration during the first 3 months of anti-TB treatment. There was no significant change in mean serum β₂-MG concentration at 3 months relative to pretreatment levels in either group. However, there were significant and equal reductions in mean serum TNF-R1 concentrations in the TB⁺HIV⁻ and TB⁺HIV⁺ groups (P = 0·01 for each).

DISCUSSION

In this study we measured the concentrations of five soluble receptors in serum samples obtained from clinically defined groups of patients in West Africa in order to identify immunological markers that show relatively independent associations with TB or HIV infection in dually infected persons. Elevated serum levels of β₂-MG were associated with HIV infection, with minor influence from TB co-infection. Conversely, increased serum concentrations of TNF-RI were associated with TB with minor influence from HIV co-infection. These data will serve to rationalize the choice of immune markers in future immunological studies of the copathogensis and treatment of TB and HIV infection in vivo, and further suggest that the use of β₂-MG as a simple marker of progression of HIV infection in field studies is not confounded by coincident TB.

In addition to CD4⁺ lymphopenia, HIV co-infection leads to systemic activation of the immune system in both western and African patients with TB [24,25]. This is confirmed by the results of this study, which found substantial elevation of serum levels of all five immune markers in dually infected subjects. Negative correlations between blood CD4⁺ lymphocyte counts and serum levels of each of these markers indicate that greater systemic immune activation is associated with more advanced HIV infection in patients co-infected with TB. This is probably attributable to effects of both organisms: low CD4⁺ lymphocyte counts are associated with both increased plasma HIV load and greater likelihood of systemic spread of mycobacteria [26]. By means of cellular activation, TB enhances HIV-1 replication in vivo[9] and, in view of the marked systemic immune activation observed, enhanced replication probably occurs both at the principal site of TB [27] and systemically.

β₂-MG forms the light chain of class I MHC molecules and is therefore present on nearly all cell types [28]. Circulating β₂-MG is generated during normal MHC I turnover [28] and thus is not specific to HIV-related cell death. Although increased concentrations of this molecule are predictive of progression of HIV infection to AIDS [29], the prognostic value in HIV⁺ patients who also have active TB is not clear [24,30–34]. [Differences in study design, including sample size, patient selection, and absence of either control groups or prospective data collection, probably account for these broadly contradictory data. Our findings that serum β₂-MG is markedly increased in TB/HIV-co-infected patients and is only minimally elevated in those with TB only concur with the data of Vanham et al. [24]. In addition, the fact that CD4-matched groups of TB⁺HIV⁺ and TB⁻HIV⁺ patients had similar levels of β₂-MG suggests that TB did not substantially contribute to the level of the marker in the TB⁺HIV⁺ group. This conclusion is further supported by the prospective data that showed that serum levels of β₂-MG did not decline during anti-TB treatment (Fig. 3). Instead, β₂-MG levels during TB treatment in the TB⁺HIV⁺ group paralleled the persistently elevated plasma HIV-1 load that was previously demonstrated [17]. Together, these data suggest that β₂-MG may be useful as an immunological marker of HIV infection in the presence of TB co-infection. This conclusion is supported by our finding of significant correlations between serum β₂-MG concentrations and CD4⁺ lymphocyte counts in both the TB⁺HIV⁺ and TB⁻HIV⁺ groups.

Increased levels of sCD8 have been found in the serum of HIV-infected individuals, and it has been suggested that this molecule might serve as a useful prognostic marker [35,36]. However, although we confirmed that serum levels of sCD8 were increased in the TB⁻HIV⁺ group (Fig. 1b), levels did not correlate significantly with CD4⁺ lymphocyte counts. Furthermore, serum levels of sCD8 were significantly higher in the TB⁺HIV⁺ group compared with the TB⁻HIV⁺ group, reflecting increased shedding of sCD8 by lymphocytes with greater cellular activation [35] due to the presence of TB. Thus, sCD8 levels were significantly influenced by TB co-infection and showed less specificity for HIV infection than did β₂-MG.

This study confirms previous findings that serum levels of the immune activation markers sCD54, sTNF-RI and sTNF-RII are all elevated in patients with active TB [37,38]. Soluble CD54 and sTNF-RII were also significantly increased in sera of persons with HIV infection only, indicating that they are not independently associated with either TB or HIV infection. In contrast, serum sTNF-RI concentrations were markedly elevated only in patients with TB; levels were affected to a relatively minor degree by HIV infection, confirming the findings of Zangerle et al. [39]. Furthermore, it has previously been shown that sTNF-RII, but not sTNF-RI, serves as a marker of host antiviral reactivity [40]. From our data, we concluded that sTNF-RI was relatively independently associated with TB in TB/HIV-infected patients, and this was strongly supported by the finding of similar reductions in mean serum sTNF-RI in both the TB⁺HIV⁻ and TB⁺HIV⁺ groups during anti-TB treatment (Fig. 3).

In conclusion, although TB and HIV co-infection lead to broad activation of the immune system, β₂-MG and sTNF-RI are relatively independent markers of the immunological effects of HIV infection and TB, respectively. As such, these immune markers may serve a useful role in future studies of the copathogenesis and treatment of TB and HIV co-infection and in field studies of the progression of HIV infection in developing countries.