Predictive value of soluble haemoglobin scavenger receptor CD163 serum levels for survival in verified tuberculosis patients

Authors


Corresponding author and reprint requests: T. B. Knudsen, Department of Infectious Diseases, Copenhagen University Hospitals Hvidovre, Kettegaard alle 30, Copenhagen 2650 Denmark
E-mail: tbygum@gmail.com

Abstract

Pre-treatment serum levels of sCD163 were measured in a cohort of 236 suspected tuberculosis (TB) cases from Guinea-Bissau, with a median follow-up period of 3.3 years (range 0–6.4 years). In 113 cases, the diagnosis of TB was verified by positive sputum microscopy and/or culture. Among the verified TB cases, a decreased survival rate was found in 27 patients with sCD163 levels above the upper reference limit (3.95 µg/mL). The difference in survival was significant during TB treatment (log rank, p < 0.02) and after long-term follow-up (log rank, p < 0.001). The decrease in survival rate during TB treatment remained significant in a multivariate Cox model controlling for human immunodeficiency virus (HIV) status, age and gender, with a mortality increase of 1.19 (95% CI, 1.04–1.36) per µg of sCD163, and a hazard ratio (HR) for sCD163 levels above the upper reference limit of 4.18 (95% CI, 1.06–16.4). The difference was not significant after excluding patients with concomitant HIV-1 and HIV-2 infection in Kaplan–Meier analyses (log rank, p 0.11). In contrast, the difference in survival remained significant in Kaplan–Meier analyses after long-term follow-up, even after excluding patients with concomitant HIV-1 and HIV-2 infection (log rank, p 0.002). In the Cox model, the mortality increase per µg of sCD163 was 1.27 (95% CI, 1.14–1.40), with an HR for elevated sCD163 levels of 2.85 (95% CI, 1.44–5.63). The HRs for concomitant HIV-1 and HIV-2 infection were 6.92 (95% CI, 3.28–14.58) and 2.48 (95% CI, 1.09–5.67), respectively. Thus, sCD163 levels appeared to be an independent predictor of survival in verified TB patients.

Introduction

Tuberculosis (TB) remains a major cause of morbidity and mortality in resource-poor settings, despite the existence of effective treatment regimens and global efforts to control disease spread. Approximately one-third of the global population is infected with Mycobacterium tuberculosis, and there are > 8 million new cases of active TB and almost 2 million deaths annually [1]. The impact of TB has been augmented by the advent of human immunodeficiency virus (HIV) and the emergence of drug-resistant strains of M. tuberculosis. The countries most affected clearly require simple and inexpensive markers of disease progression and/or treatment efficacy.

CD163 belongs to the scavenger receptor cysteine-rich domain family. It is expressed exclusively on monocytes and macrophages, and is the endocytic receptor for haptoglobin–haemoglobin complexes [2]. Soluble CD163 (sCD163) is shed from macrophages by a protease-mediated release mechanism in response to inflammatory stimuli [3], and may modulate the inflammatory response [4,5]. The clinical implications of sCD163 in diseases involving the monocyte–macrophage system are being investigated currently [6]. At present, it seems clear that both proliferation [7–9] and inflammatory activation of macrophages [10,11] result in increased sCD163 levels. Elevated levels have been found in patients with sepsis [8,9], and sCD163 levels predict survival in patients with pneumococcal bacteraemia (H. J. Møller, personal communication). In addition, patients infected with the macrophage-tropic protozoan Leishmania have increased sCD163 levels [9]. As macrophages and monocytes play a pivotal role in the progression of TB, the aim of the present study was to investigate whether sCD163 serum levels are associated with mortality in verified TB patients.

Materials and methods

Patients

Patients from four suburban areas in Bissau, the capital of Guinea-Bissau in West Africa, were investigated and have been followed demographically for several years. In 1996, a TB surveillance system was established in Bissau in collaboration with the national referral hospital for lung diseases, Hospital Raoul Follereau [12]. Patients presenting either at the two local health centres, or directly at the referral hospital, with symptoms compatible with active TB were referred by the hospital for possible inclusion in the study. Inclusion criteria were age ≥ 15 years and the presence of at least one of the following, otherwise unexplained, symptoms or signs: persistent cough (> 1 month) without improvement following antibiotic treatment; constant or periodic fever (> 1 month); weight loss; dyspnoea; haemoptysis; night sweats; or lymphadenopathy. Patients were investigated with a chest X-ray and with direct microscopy and culture of sputum. Blood was screened for HIV using the Capillus HIV-1/HIV-2 kit (Cambridge Diagnostics, Galway, Ireland) and the Enzygnost Anti-HIV 1+2 Plus kit (Behring Diagnostics, Marburg, Germany). Pre-treatment serum samples were taken at the time of inclusion in the study.

Based on the findings, patients were divided into three diagnostic groups (Table 1): (1) verified TB, defined as patients who were positive for acid-fast bacilli following direct microscopy of sputum, and/or positive for M. tuberculosis in sputum culture; (2) presumed TB, defined as patients who were negative for acid-fast bacilli following direct microscopy and culture, but who had signs, symptoms and X-ray changes compatible with active intrathoracic TB—these patients were treated initially with antibiotics and were then re-evaluated clinically and with chest X-ray, followed by a presumption of TB and treatment if there was no improvement; and (3) non-TB disease, defined as patients who were diagnosed as having disease other than active intrathoracic TB. Verified and presumed TB patients were treated according to a regimen established by the national TB programme (see below), and all patients included in the study were followed with respect to mortality until January 2003. In total, 236 patients were enrolled in the study (128 males and 108 females). The TB surveillance study was approved by the Ministry of Public Health in Guinea-Bissau and the Central Ethical Committee of Denmark.

Table 1.  Cohort baseline characteristics used for survival analysis, with the data divided into two groups according to sCD163 levels
All casessCD163 < 3.95 µg/mL (n = 189; 76 deaths)sCD163 > 3.95 µg/mL (n = 47; 27 deaths)No. of cases (total n = 236)No. of deaths (total n = 103)
  • a

    Includes patients with a dual infection.

Age, years (median)39.035.0  
Male104 (42 deaths)24 (15 deaths)12857
Female85 (34 deaths)23 (12 deaths)10846
TB status
 Verified TB86 (24 deaths)27 (16 deaths)11340
 Presumed TB51 (31 deaths)12 (6 deaths)6337
 Non-TB disease52 (21 deaths)8 (5 deaths)6026
HIV status
 HIV-1a25 (23 deaths)15 (12 deaths)4035
 HIV-248 (26 deaths)11 (8 deaths)5934
 HIV-negative116 (27 deaths)21 (7 deaths)13734

TB treatment

The TB treatment regimen consisted of directly observed daily treatment with ethambutol, isoniazid, rifampicin and pyrazinamide for a period of 4 months, followed by self-administered treatment for a further 4 months with isoniazid and ethambutol, collected twice-monthly at the local health centres. HIV-positive and HIV-negative patients received identical treatment. All patients received daily doses of multivitamins and vitamin B complex. Adherence was verified by counting tablets and an isoniazid urine test after 2, 5 and 8 months. More than 85% of the patients completed treatment for a 6-month period. Patients were investigated clinically and with sputum smear microscopy following treatment for 8 months. No patient yielded a positive sputum result at this time, and all survivors were thus considered to be cured of TB.

Soluble CD163 serum levels

Adequate amounts of serum were available from all 236 patients included in the study. Duplicate measurements of sCD163 were performed retrospectively on frozen pre-treatment serum samples by ELISA as described previously [13]. Control samples and standards of purified CD163 were co-analysed in each run.

Statistics

Comparison between groups was performed using the Mann–Whitney U-test. Comparison of survival rates in groups according to sCD163 levels was performed by Kaplan–Meier analysis following stratification according to diagnostic group (verified TB cases, presumed TB, and non-TB disease). Survival analyses for patients with verified TB were controlled for HIV status. Differences between Kaplan–Meier curves were evaluated by the log-rank test. The association between sCD163 levels and mortality in verified TB cases was assessed by a Cox regression analysis, with adjustment for other known risk-factors (age, gender and HIV status) [12,14]. Analysis of survival during treatment was undertaken from the time of inclusion to the end of the treatment period, or the first of the following events: death, disruption of treatment following the outbreak of war in Guinea-Bissau (6 June 1998) [15], or abandonment of treatment or lost to follow-up. Multivariate long-term survival analysis was performed from the time of inclusion until January 2003 or the first of the following events: death or lost to follow-up. To evaluate survival rates in verified TB cases without the potential confounder of a concomitant HIV infection, Kaplan–Meier analyses excluding all patients with HIV-1, HIV-2 or dual infection were performed. In the Cox model, patients with dual infection were included as HIV-1-positive, as they have been shown to have a similar risk and severity of TB disease [14]. In the absence of an appropriate age-, sex- and race-matched control group, an upper reference limit of 3.95 µg/mL, identified in a previous study of sCD163 [8], was chosen as the dividing parameter for the statistical analysis. All statistical analyses were performed using SPSS v. 10 software (SPSS Inc., Chicago, IL, USA).

Results

Influence of TB and HIV status on sCD163 levels

A trend (p 0.06) towards elevated sCD163 levels was observed in verified TB cases (n = 113; median 2.50 µg/mL) compared with individuals with non-TB disease (n = 60; median 1.76 µg/mL). The difference was significant (p 0.02) in patients positive for acid-fast bacilli following direct microscopy of sputum (n = 82; median 2.70 µg/mL). No significant differences in sCD163 levels among presumed TB patients (n = 63; median 2.34 µg/mL) or TB patients negative following direct microscopy who were culture-positive (n = 33; median 2.09 µg/mL) were observed in comparison with TB-negative individuals. The highest levels were found in verified TB cases with concomitant HIV-1 infection (n = 19; median 3.11 µg/mL); these patients had significantly (p 0.02) higher levels than HIV-negative verified TB patients (n = 68; median 2.34 µg/mL). HIV-1- and HIV-2-positive patients were distributed uniformly among the groups; most notably, 16.8% of verified TB cases and 15.6% of patients with non-TB disease were HIV-1-positive.

Predictive value of sCD163 for mortality during treatment

Mortality during treatment was evaluated in both a univariate and a multivariate Cox regression model, since the latter may be of limited value with few events. The median follow-up period was 7.3 months (range 0.2–10.5 months). Of 113 verified TB patients, nine died during treatment, 29 were censored at the time of civil war, and 18 were censored at the time when treatment was abandoned or they were lost to follow-up. Five (56%) of the nine patients who died during treatment, and 23 (74%) of the 31 patients who died after completion of treatment, were HIV-1- or HIV-2-positive. In univariate analysis, an increased hazard ratio (HR) was found per µg of sCD163 of 1.18 (95% CI, 1.05–1.32), with an HR of 4.32 (95% CI, 1.2–16.1) for sCD163 levels above the upper reference limit. In the multivariate model controlling for HIV status, age and gender, only sCD163 remained significant (p 0.01), with an increased HR per µg of sCD163 of 1.19 (95% CI, 1.04–1.36) and an HR of 4.18 (95% CI, 1.06–16.4) for sCD163 levels above the upper reference limit. Thus, elevated sCD163 levels were associated with increased mortality during the 8-month treatment period in verified TB cases.

Value of sCD163 as an independent predictor of long-term mortality

Long-term mortality, defined as mortality between the time of inclusion and January 2003, was evaluated using the multivariate Cox model, controlling for HIV status, age and gender. The median follow-up period was 3.5 years (range 0–6.4 years). Of 113 verified TB patients, 40 died and 25 were lost to follow-up. The long-term mortality increase per µg of sCD163 was 1.27 (95% CI, 1.14–1.40), with an HR of 2.85 (95% CI, 1.44–5.63) for sCD163 levels above the upper reference limit. HIV status remained a significant risk-factor in the model, with HRs of concomitant infection of 6.92 for HIV-1 (95% CI, 3.28–14.58) and 2.48 for HIV-2 (95% CI, 1.09–5.67) for verified TB cases. There was no significant relationship with age or gender in this model.

Association between high sCD163 levels and increased mortality

Kaplan–Meier survival analysis showed that verified TB cases with sCD163 levels > 3.95 µg/mL were associated with increased mortality during treatment, as well as after long-term follow-up, compared with verified TB cases with sCD163 levels < 3.95 µg/mL (Fig. 1a,bn = 113; log rank, p 0.017 and p < 0.001, respectively). When HIV status was controlled for, the difference in long-term survival remained significant in both HIV-negative (Fig. 1c; n = 68; log rank, p 0.002) and HIV-1- and/or HIV-2-positive individuals (Fig. 1d; n = 45; log rank, p 0.02); however, the difference was not significant during treatment (log rank, p 0.11 and p 0.13, respectively). There was no significant effect of sCD163 levels on survival for patients with presumed TB (n = 63; log rank, p 0.28) or patients with non-TB disease (n = 60; log rank, p 0.20).

Figure 1.

Kaplan–Meier plots showing survival during treatment and long-term survival of verified cases of tuberculosis. Vertical lines indicate censoring. The upper bold line indicates patients with sCD163 levels below the upper reference limit of 3.95 µg/mL, and the lower line indicates levels above the upper reference limit of 3.95 µg/mL. (a) Overall survival during treatment in verified TB cases (n = 113; 27 cases > 3.95 µg/mL; p 0.017). (b) Overall long-term survival in verified TB cases (n = 113; 27 cases > 3.95 µg/mL; p < 0.001). (c) Long-term survival in HIV-negative verified TB cases (n = 68, 14 cases > 3.95 µg/mL; p 0.002). (d) Long-term survival in verified TB cases with concomitant HIV-1 and/or HIV-2 infection (n = 45; 13 cases > 3.95 µg/mL; p 0.02).

Discussion

The results of this study indicated that levels of sCD163 predict mortality during treatment, as well as long-term mortality, in verified cases of TB. There was a significant association between elevated sCD163 levels and mortality during treatment in both Kaplan–Meier (Fig. 1a) and univariate Cox regression analyses. The decrease in survival rate during TB treatment remained significant when controlling for HIV status in a multivariate Cox regression model, but not when completely excluding patients with concomitant HIV-1 and HIV-2 infection in Kaplan–Meier analysis. However, because of the relatively few events occurring during the treatment period, these results should be interpreted with caution. The association between sCD163 levels and long-term mortality remained significant for both TB patients and TB patients co-infected with HIV (Fig. 1b,d). Among verified TB cases, significantly higher sCD163 levels were found in HIV-1-positive than in HIV-negative individuals. Thus, both TB and HIV status influenced sCD163 levels, although their relative contributions to the observed increase remain unclear. However, the inherent limitations of the study design and the lack of data for standard markers of disease progression (e.g., CD4 cell counts and HIV virus loads) do not allow conclusions to be reached concerning possible associations between sCD163 levels and mortality in HIV disease.

The finding that patients with verified TB and HIV-1 infection had the highest sCD163 levels is in agreement with previous observations [14] that the severity of TB disease is augmented by concomitant HIV-1 disease (Fig. 1d). Most notably, the observation that sCD163 levels predict mortality, even when HIV-1- and HIV-2-infected patients are excluded, suggests that sCD163 may be a sensitive marker for the severity of TB disease (Fig. 1c).

Considering the natural course of TB, it might be expected that most deaths related directly to TB disease would occur during the first year of follow-up. In accordance with this hypothesis, elevated sCD163 levels were found to be associated with increased mortality during the 8-month treatment period. Surprisingly, the long-term survival analyses showed that substantial mortality occurred after the first year (Fig. 1b). Whether the predictive value of sCD163 for long-term survival reflects treatment failure, reactivation of latent TB, development of resistant bacteria, or a general susceptibility to disease, remains to be clarified. The present data suggest that sCD163 may be a sensitive marker for TB disease; however, it is possible that elevated levels may also reflect other conditions that carry an increased risk of mortality. A biological understanding of these observations requires further investigation, but it could be speculated that elevated levels of sCD163 may parallel an increased mobilisation and inflammatory activation of macrophages and monocytes that reflects the severity of TB disease. The putative function of sCD163 as a modulator of the inflammatory response is in agreement with our findings.

In conclusion, the results of this study demonstrate that both TB and HIV status influence sCD163 levels, and that high levels of sCD163 are associated with increased mortality in verified cases of TB. The findings suggest that measurement of sCD163 levels may be a useful prognostic tool in TB disease. However, further prospective studies are needed to confirm the results and to evaluate the effect of anti-tuberculosis drugs on sCD163 levels in order to determine whether such measurements can be used to monitor treatment efficacy.

Ancillary