INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Tuberculosis, a reemerging disease, is recognized to be associated with a degree of compromise of the immune system, particularly failing Th1 cytokine responses and overproduction of IL-10 and TGF- (1). One mechanism by which this might be achieved is via the exposure of immune cells to abnormal relative local concentrations of glucocorticoids (5). To evaluate this possibility we investigated the hypothesis that within the tissues infected by M. tuberculosis there is a local imbalance in the ratio of cortisol to biologically inactive cortisone.

There are complex relationships between glucocorticoids and defense against tuberculous infection. Administration of synthetic glucocorticoids induces reactivation of tuberculosis in humans (8, 9) and in animals (10). Conditions of war, stress, and deprivation lead to spontaneous reactivation of tuberculosis (11) and it may be that elevated concentrations of the principal endogenous glucocorticoid, cortisol, are responsible. Adrenal function in human tuberculosis has previously been investigated with variable and inconclusive results (12). In a previous study of 24-h urinary cortisol metabolites in patients with active pulmonary tuberculosis (13), we noted a marked elevation in the ratio of metabolites of cortisol to those of its inactive derivative cortisone.

The interconversion of cortisol and cortisone is catalyzed by 11-hydroxysteroid dehydrogenases (11-HSDs), which exist as at least two distinct enzymes. 11-HSD type 2 is a widely expressed high-affinity enzyme. It acts as an exclusive 11-dehydrogenase, notably in the distal nephron, where it converts cortisol to cortisone and thereby protects renal mineralocorticoid receptors from inappropriate activation by cortisol (14- 17). 11-HSD2 is also present in human lung (18) and in lymph nodes (19). In contrast, 11-HSD type 1 (20) is a low-affinity enzyme expressed in multiple tissues, including liver and lung (21). In most tissues, 11-HSD1 acts a reductase, converting cortisone to cortisol. In this study we have examined both central control of glucocorticosteroids and peripheral metabolism by assessing basal cortisol secretion, responses to stimulation and suppression of the hypothalamo-pituitary adrenal axis, net whole-body metabolism of cortisol by 11-HSDs, and specific metabolism of cortisol in the lung. These investigations were performed in patients with active tuberculosis, previous cured tuberculosis, acute nontuberculous pneumonia, and matched normal control subjects.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Patient Groups

The study groups included 30 patients with active pulmonary tuberculosis confirmed by auramine staining of sputum or bronchoalveolar lavage fluid, and by subsequent culture; 14 patients with microbiologically proven previous pulmonary tuberculosis who completed treatment between 6 and 23 mo before the study; 23 healthy volunteers with no clinical evidence of tuberculosis and receiving no medication; and six patients with a diagnosis of acute pneumonia, based upon clinical presentation, chest radiograph revealing consolidation, and/or sputum culture. Demographic details for all groups are recorded in Table 1.

All participants gave written informed consent for the studies, which were approved by the local ethical committees of St. Mary's Hospital and University College Hospital, London. For all participants, exclusion criteria included: corticosteroid or androgen therapy by any route during the preceding 12 mo; abnormal thyroid or liver function tests; hypertension requiring medical therapy; diabetes mellitus; clinical depression during the preceding 12 mo; obesity (body mass index > 30 kg/m²); previous organ-specific autoimmune disease; positive HIV serology (all patients were tested for HIV infection). Not all participants underwent all investigations; numbers are given with results.

Protocol

Strict adherence to protocol was maintained for all subjects to prevent effects caused by stress responses. The following measurements were made before any treatment: (1) 24-h urine collection for cortisol metabolites; (2) intravenous cannulation at 8:30 A.M. after overnight fast and withdrawal of blood after 30 min supine at 9:00 A.M. for serum cortisol, renin activity, aldosterone, and electrolytes; and (3) 4 hourly saliva collection (using Salivette; Sarstedt, Leicester, UK) for cortisol.

All healthy volunteers and patients with active or cured tuberculosis were then treated with pyrazinamide (< 50 kg body weight 1.5 g, > 50 kg body weight 2 g), pyridoxine (10 mg), ethambutol (25 mg/kg body weight) and isoniazid (300 mg). Administration was once daily by mouth for all drugs. Rifampicin, as an inducer of hepatic drug metabolizing enzymes that reduces the bioavailability of glucocorticoids, was omitted for the duration of the investigations. After 48 h, the following measurements were made. Day 1: 24-h urine collection for cortisol metabolites and four hourly saliva collection for cortisol. Day 2: corticotrophin-releasing hormone (CRH) test in which 100 µg ovine CRH (Ferring Pharmaceuticals Ltd. Feltham Middlesex UK) was injected at 9:00 A.M. and blood was withdrawn for 60 min from an intravenous cannula for cortisol and ACTH. Day 3: overnight threshold dose dexamethasone suppression test in which 250 µg dexamethasone was taken orally at 11:00 P.M., and subjects attended at 9:00 A.M. for cannulation and blood withdrawal for measurement of serum cortisol and dexamethasone. This was followed by oral administration of cortisone acetate (25 mg) and withdrawal of blood at intervals of 150 min for serum cortisol. Day 4: a second low dose suppression test performed as on Day 3 but with 2 mg dexamethasone. This was followed by sequential stimulation with boluses of ACTH_1-24 (Synacthen; Ciba, Horsham, UK) 60 ng, 150 ng, and 250 µg at 90-min intervals and withdrawal of blood at intervals for serum cortisol over 240 min.

Control Patients with Pneumonia

Six patients with nontuberculous pneumonia were studied during antibiotic treatment. Samples were collected for baseline urine measurements as above. These patients were not given antituberculosis therapy, and did not participate in dynamic tests.

Bronchoalveolar Lavage

Eleven patients with tuberculosis and 13 healthy volunteers provided specimens for bronchoalveolar lavage. The procedure was performed before any antituberculosis treatment was initiated. In both patients and healthy volunteers lavage was performed during an infusion of cortisol 4 µg/kg/min in order to ensure high substrate concentrations for 11-HSD activities and counteract any interindividual effects of stress. The infusion followed a 3-mg intravenous bolus of cortisol and was run for 240 min before the procedure in order to obtain a steady-state condition. The infusion continued during the lavage. Blood was sampled before the infusion was commenced, and immediately after the procedure. Both patients and healthy volunteers were sedated before a flexible bronchoscope was passed into either the diseased segment or the right middle lobe. Then 250 ml of normal saline were instilled in 50-ml aliquots with the end of the bronchoscope wedged in a subsegmental bronchus and fluid aspirated after minimal dwell time (25).

Assays

Urine was analyzed for conjugated and unconjugated metabolites of cortisol by gas chromatography and mass spectrometry (GC/MS), as previously described (13). Serum and salivary cortisol and serum ACTH were measured by radioimmunoassays, as were plasma renin activity and aldosterone. Serum dexamethasone was measured by gas chromatography and mass spectrometry. Cortisol binding globulin was measured by radioimmunoassay (Medgenix radioimmunoassay kit; Lifescreen, Watford, UK). Cortisol and cortisone were measured in bronchoalveolar lavage fluid by radioimmunoassay after extraction on a C18 SepPak (Waters Millipore, Watford, UK), elution with methanol, and separation by HPLC.

Data Interpretation

Activity of the hypothalamo-pituitary adrenal axis was reflected in several indices. (1) Cortisol production rate, estimated from the sum of the daily excretion of the principal urinary metabolites of cortisol and cortisone (5-tetrahydrocortisol [THF], 5-tetrahydrocortisol [allo-THF], tetrahydrocortisone [THE], - and -cortols, and - and -cortolones); (2) diurnal variation of salivary cortisol; (3) stimulation of both ACTH and cortisol by CRH; (4) stimulation of cortisol by threshold doses of exogenous ACTH; and (5) suppression of plasma cortisol at two different circulating concentrations of dexamethasone.

Metabolism of glucocorticoids was estimated by: (1) ratios of urinary metabolites of cortisol, from which 11-HSD activities are reflected in (THF + allo-THF)/THE ratio, and the balance of 5- and 5- reductase activities are reflected in THF/allo-THF ratio; (2) the accumulation of cortisol in peripheral serum after oral administration of cortisone, which reflects predominant 11-reductase activity (26); and (3) the ratio of cortisol to cortisone in bronchoalveolar lavage fluid and plasma.

Statistical Analysis

Variables were compared between groups, and across time-courses, by analysis of variance followed by least squares difference tests when appropriate, or by the Mann-Whitney U test. Relationships between continuous variables were compared by multiple regression analysis, in which qualitative variables were assigned values of 0 and 1 (27).

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The Hypothalamo-pituitary Adrenal Axis

Baseline and dynamic tests found no differences in the hypothalamic pituitary adrenal access between patients with acute tuberculosis and the control groups.

Twenty-four-hour urine. In patients with active pulmonary tuberculosis, compared with other groups, there was no change in total cortisol metabolite excretion rate (Table 2). There was no significant change in cortisol production rate after 3 d of treatment.

The diurnal rhythm of cortisol concentrations. Salivary cortisol concentrations were higher at 8:00 P.M. in samples from the patients with active tuberculosis than in those from other groups (Figure 1a and 1b). This abnormality was found in the samples taken before initiation of antituberculosis therapy, but was not significant after 3 d of treatment. This observation might have been linked to stress of hospitalization since only the patients with active tuberculosis were admitted.

View larger version (18K): [in this window] [in a new window]   Figure 1.   Diurnal variation of salivary cortisol, mean ± SEM. (a) Before antituberculosis therapy: patients with tuberculosis (n = 14) (closed circles); patients cured of tuberculosis (n = 13) (open triangles); healthy volunteers (n = 11) (open circles). (b) During antituberculosis therapy: patients with tuberculosis (n = 12) (closed circles); patients cured of tuberculosis (n = 11) (open triangles); healthy volunteers (n = 10) (open circles). Asterisk indicates significantly elevated value in patients with tuberculosis. ANOVA before therapy between groups p < 0.05, significant between acute tuberculosis and both other groups to p < 0.05. ANOVA during therapy not significant.

Responsiveness of the adrenal to ACTH. After suppression of endogenous cortisol with a 2-mg dose of dexamethasone, the adrenals of patients with acute tuberculosis responded normally to the physiological challenges (60 ng and 150 ng) and also to the supraphysiological challenge (250 µg) with ACTH (Figure 2).

View larger version (20K): [in this window] [in a new window]   Figure 2.   Serum cortisol responses, mean ± SEM, to sequential ACTH stimulation tests after dexamethasone 2 mg. Patients with tuberculosis (n = 11) (closed circles); patients cured of tuberculosis (n = 11) (open triangles); healthy volunteers (n = 11) (open circles). No significant differences between groups.

Response of the pituitary to CRH. Plasma cortisol was not suppressed by dexamethasone before this test was performed. There were no significant differences between the groups in the increases in cortisol or ACTH that were observed after CRH (Figures 3a and 3b).

View larger version (13K): [in this window] [in a new window]   Figure 3.   (a) Cortisol and (b) ACTH responses to CRH administered at time 0, mean ± SEM. Patients with tuberculosis (n = 12) (closed circles); patients cured of tuberculosis, (n = 11) (open triangles); healthy volunteers (n = 12) (open circles). No significant differences between groups.

Suppression by dexamethasone. Suppression of 9:00 A.M. plasma cortisol levels after 250 µg dexamethasone at 11:00 P.M. the previous night was similar in all groups (Table 2). Similarly, suppression of 9.00 A.M. cortisol levels to less than 50 nmol/L was achieved in all subjects after 2 mg dexamethasone (Table 2). The concentrations of plasma dexamethasone achieved at each dose did not differ between groups (Table 2).

Cortisol Metabolism

Both the 24-h urine analysis and study of conversion of an oral cortisone load identified a deviation of peripheral metabolism of glucocorticoids in favor of the active metabolite cortisol.

Twenty-four-hour urine collections. In patients with active pulmonary tuberculosis the ratio of urinary metabolites of cortisol relative to metabolites of cortisone (Figure 4) was increased. Thus, the urinary cortisol: cortisone metabolite ratio in patients with active pulmonary tuberculosis (n = 16) of 1.19 (± 0.1) was greater than that in healthy volunteers (n = 11) of 0.78 ± 0.04 (p < 0.005) or in patients with previously cured pulmonary tuberculosis (n = 13) who had values of 0.89 ± 0.05 (p < 0.01). This remained significant during 3 d of treatment in hospital. Thus, the results of measurements after 48 h of therapy (excluding rifampicin) revealed in those with acute pulmonary tuberculosis (n = 12), a urinary cortisol: cortisone metabolite ratio of 1.37 ± 0.18, which was greater than in the healthy volunteers whose ratio was 0.85 ± 0.07 (n = 10) p < 0.005, those with cured tuberculosis with a ratio of 1.03 ± 0.18 (n = 10) p < 0.01, and acute pneumonia with a ratio of 0.9 ± 0.16 (n = 6) p < 0.05. On treatment ANOVA between groups was p < 0.05. Those patients with acute nontuberculous pneumonia, who were receiving various antibiotic regimens at the time of sampling, showed a range in the ratio of cortisol to cortisone metabolites, but with no differences from other control groups. In multiple regression analyses, sex, ethnic origin, body mass index, and age did not account for differences in cortisol/cortisone metabolite ratios between groups.

View larger version (27K): [in this window] [in a new window]   Figure 4.   Ratio of urinary metabolites of cortisol (THF + alloTHF) to metabolites of cortisone (THE) in groups before (top panel ) and during (bottom panel ) treatment, mean ± SEM. "During treatment" refers to antituberculosis chemotherapy for 48 to 72 h in all groups apart from those with acute pneumonia, who were treated with various antibiotics and for varying durations. Patients with tuberculosis (TB) had significantly elevated ratios compared with healthy volunteers, cured patients, or patients with pneumonia. ANOVA between groups before treatment p < 0.005, significant between acute tuberculosis and both cured TB and healthy volunteers to p < 0.01. ANOVA between groups during treatment p < 0.02. The differences between individual ratios during treatment were significant to the following p values: Acute TB:HV, p < 0.005; Acute TB:Cured TB, p < 0.01; Acute TB:pneumonia, p < 0.05. Discrepancies in sample numbers compared with those in Table 1 relate to incomplete 24-h urine collections.

There was also an increase in the ratio of 5- to 5-tetrahydrocortisol in patients with active pulmonary tuberculosis (Table 2). Thus, the urinary 5- to 5-tetrahydrocortisol ratio in patients with active pulmonary tuberculosis (n = 16) of 2.94 ± 0.4 was greater than that in healthy volunteers (n = 11), 1.54 ± 0.22 (p < 0.05). Patients with previously cured pulmonary tuberculosis (n = 13) had values of 2.28 ± 0.3 (p = NS). This remained significant during 3 d of treatment in hospital. Thus, the results of measurements after 48 h of therapy (excluding rifampicin) revealed in those with acute pulmonary tuberculosis (n = 12), a urinary 5- to 5-tetrahydrocortisol of 2.82 ± 0.5, which was greater than in the healthy volunteers whose ratio was 1.67 ± 0.23 (n = 10) p < 0.01, those with cured tuberculosis with a ratio of 1.63 ± 0.23 (n = 10) p < 0.01, and those with acute pneumonia with a ratio of 1.93 ± 0.25 (n = 6) p = 0.07.

Conversion of cortisone to cortisol after an oral cortisone load. After suppression of plasma cortisol with 250 µg dexamethasone, and administration of 25 mg oral cortisone, peak plasma cortisol concentrations were higher in patients with active tuberculosis relative to other groups (Figure 5). Patients with tuberculosis (n = 14) achieved a significantly higher peak of plasma cortisol (1,157 ± 55 nmol/L) than healthy volunteers (n = 10) whose peak was 882 ± 73 nmol/L, p < 0.005, or patients cured of tuberculosis (n = 10) who had a maximum of 862 ± 50 nmol/L, p < 0.001. ANOVA between groups was p < 0.005. There were no differences in corticosteroid binding globulin (Table 2) or baseline plasma cortisol (Figure 5) between groups.

View larger version (16K): [in this window] [in a new window]   Figure 5.   Accumulation of cortisol in serum after oral administration of cortisone acetate 25 mg, mean ± SEM. Subjects had received dexamethasone 250 µg at 11:00 P.M. the previous night. The cortisone tablet was administered at 9:00 A.M. Patients with tuberculosis (n = 14) (closed circles); patients cured of tuberculosis (n = 10) (open triangles); healthy volunteers (n = 10) (open circles). Patients with tuberculosis achieved a significantly higher peak of plasma cortisol (1,157 ± 55 nmol/L) than healthy volunteers (882 ± 73 nmol/L, p < 0.005) or patients cured of tuberculosis (862 ± 50 nmol/L, p < 0.001). ANOVA between groups p < 0.005.

Pulmonary Metabolism of Glucocorticoids

Cortisol/cortisone ratios in bronchoalveolar lavage fluid were higher in patients with tuberculosis when compared with other groups (Table 3). Patients with acute tuberculosis (n = 13) achieved a higher cortisol/cortisone ratio of 7.73 ± 1.48 compared with healthy volunteers (n = 11) whose ratio was 4.05 ± 0.38 (p < 0.05). By contrast plasma cortisol and cortisone concentrations were not different in patients with tuberculosis compared with those in healthy control subjects at baseline or during cortisol infusion. Cortisol/cortisone ratios in plasma were rather variable, but not different between groups. Cortisol/cortisone ratios in lavage fluid did not correlate with ratios or absolute levels of cortisol in plasma (BAL ratio versus plasma ratio R = 0.01, p = 0.95; BAL ratio versus plasma cortisol concentration R = 0.125, p = 0.57).

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

This detailed and comprehensive investigation of cortisol metabolism examined patients with active pulmonary tuberculosis, patients with cured tuberculosis, patients with nontuberculous lung infection, and healthy control subjects. We found central regulation of the hypothalamo-pituitary-adrenal axis to be remarkably normal in active tuberculosis, but we confirmed our previous observation (13) that peripheral metabolism of cortisol is perturbed. We have further demonstrated that this disturbance occurs during active tuberculous infection only. The normal response in patients with previously cured tuberculosis argues against the possibility that the abnormalities were intrinsic to subjects predisposed to tuberculosis. The study found evidence that the change in peripheral steroid metabolism was associated with an increased ratio of active glucocorticoid metabolite cortisol in the lung.

Previous studies of patients with tuberculosis have included only limited analyses of regulation of cortisol secretion. These studies showed variable loss of diurnal rhythm of plasma cortisol associated with a highly variable impaired response to a pharmacologic dose of exogenous ACTH. The confounding effects of hospitalization and drug therapy were not always accounted for. In the present study, we were unable, for ethical reasons, to complete all of our investigations before starting antituberculosis therapy, but we did collect baseline saliva, blood, and urine specimens and repeated these tests 48 h after starting therapy. The baseline tests showed that total cortisol metabolite excretion, an index of cortisol production rate, did not differ in patients with active tuberculosis. The higher salivary cortisol in the evening in these patients may be attributable to the effect of acute hospitalization, particularly since it was less striking 3 d later, when the patients had started therapy and acclimatized to the hospital environment. It was not possible for subjects in the control groups to be admitted to hospital.

After these assessments, we administered the same drugs (which excluded the liver enzyme inducer rifampicin) to patients and to the two control groups, a healthy group and a group who had previously been cured of tuberculosis. Urinary cortisol metabolite excretion did not change after drug therapy in any group, indicating that further dynamic tests of cortisol secretion and metabolism can be interpreted with reference to the untreated state. Dynamic tests of the hypothalamo-pituitary-adrenal axis performed during antituberculosis treatment showed no difference between patients with active tuberculosis and any of the control groups in cortisol responses to suppression with dexamethasone or stimulation with CRH or ACTH. Importantly, both the dexamethasone and ACTH responses were assessed using very low threshold doses as well as pharmacologic doses. The lack of difference in response to ACTH is in contrast to previous reports (12). It is possible that the different results after ACTH stimulation reflect differences in the prevalence of adrenocortical tuberculous infection in different groups of patients.

Although hypothalamic pituitary adrenal axis function appeared normal, the study did detect changes in peripheral metabolism of corticosteroids and suggests altered 11-HSD function resulting in a relative increase in cortisol. This study does not address whether the shift in the equilibrium point of the enzymes that convert cortisol and cortisone occurs in all tissues where it is expressed, or only in some sites. Although the generation of cortisol in the peripheral circulation after oral cortisone administration is thought to reflect predominately first-pass metabolism in the liver (28), it may also be influenced by the cortisol/cortisone equilibrium in other sites. Given its potential relevance to pulmonary tuberculous infection, we addressed specifically whether the altered cortisol/ cortisone equilibrium affected intrapulmonary glucocorticoid concentrations. There is substantial expression of both 11-HSD1 and 11-HSD2 in the lung (18, 22, 24, 29, 30), and the enzymes are known to be active with cortisol and cortisone having previously been quantified in BAL by gas chromatography negative ion chemical ionization mass spectrometry (22). In this single previous study by Hubbard and colleagues (22) BAL cortisone was measured in healthy volunteers as 1.05 nmol/L and cortisol as 0.42 nmol/L. In the present study we have measured cortisol and cortisone in bronchoalveolar lavage fluid using HPLC and RIA. We attribute the discrepancy in the absolute concentrations obtained in normal subjects between the present study (Table 3) and that of Hubbard and colleagues (22) to the supraphysiologic dosage of cortisol, which we infused for the reasons outlined above in all subjects. Crucially, despite there being no difference in circulating plasma cortisol concentrations or plasma cortisol/cortisone ratios between patients with acute tuberculosis and healthy volunteers, the patients with active pulmonary tuberculosis had higher cortisol/cortisone ratios in bronchoalveolar lavage fluid. With the methodologies available it was not possible to distinguish the influence of pulmonary 11-HSD2 and 11-HSD1 activities on the cortisol: cortisone ratio in the lung.

Although the data from this study do not address the mechanism for altered cortisol metabolism in tuberculosis, possibilities include the recognized regulation of 11-HSD type 1 by the cytokines IL-1 and TNF- (29). One or both of these cytokines is present in increased concentrations in bronchoalveolar lavage fluid and pleural effusions in human (31) and in murine (32) tuberculosis. Another physiologically plausible explanation would be decreased expression of 11-HSD type 2 (33).

This study has thus provided at least a theoretical mechanism to link the release of proinflammatory cytokines IL-1 and TNF in tuberculosis via modulation of 11 HSD type 1 and a relative increase in tissue cortisol with a local impairment of the immune response to tuberculosis. Increased local conversion of cortisone to cortisol could explain the increased production of TGF- (7) and IL-10 (5, 6) in human tuberculosis (1), and impaired macrophage function (34).

Whether or not the shift in cortisol metabolism in favor of the active glucocorticoid is a primary or secondary phenomenon in reactivated pulmonary tuberculosis, this observation may provide a link to the recognized increase in susceptibility to tuberculosis reactivation during stress. Therapeutic opportunities may also arise, since manipulation of corticosteroids such as corticosterone and androstenediol have been shown to affect survival curves in a mouse model of tuberculosis (35). The interactions between the immune system and the factors controlling glucocorticoid metabolism therefore offer a novel therapeutic approach in tuberculosis.