点击显示 收起
1 From the Departments of Infectious Diseases (AS, GEG, and DCM) and Chest Medicine (CFJR), St Georges Hospital Medical School, and the Department of Clinical Dietetics, St Georges Healthcare NHS Trust (LH), London.
2 Supported by the Wellcome Trust (International Fellowship to AS) Medical Research Council and Serono International SA (DCM). 3 Address reprint requests to A Schwenk, Department of Infectious Diseases, Royal Free Hospital, Hampstead, London NW3 2QG, United Kingdom. E-mail: a.schwenk{at}doctors.org.uk.
ABSTRACT
Background: Pulmonary tuberculosis is the classic cause of "consumption," but the pathogenesis of such wasting is largely unknown. Animal studies in other conditions suggest that leptin may be a mediator between proinflammatory cytokine activity and wasting.
Objective: We tested whether the leptin concentration, after control for body fat mass, is higher during active pulmonary tuberculosis than after recovery and whether it correlates with energy metabolism and proinflammatory cytokine activity.
Design: Nondiabetic adults with pulmonary tuberculosis (n = 32) were recruited into a prospective observational study. Patients found to be antibody positive for human immunodeficiency virus were excluded from the study. Dual-energy X-ray absorptiometry, indirect calorimetry, and food intake protocols were performed at baseline and after 1 and 6 mo of tuberculosis treatment. Fasting plasma leptin, tumor necrosis factor and its soluble receptor, and interleukin 6 were measured by enzyme-linked immunosorbent assay.
Results: Resting energy expenditure was close to Harris-Benedict predictions and did not change significantly during treatment, but energy intake increased. Leptin concentration was correlated in a log-linear fashion with percentage body fat but was independent of cytokines and energy intake. There was no significant difference in leptin, corrected for energy balance and fat mass, at baseline and after 1 and 6 mo of treatment.
Conclusions: These data are compatible with recovery from anorexia or starvation without discernible hyper- or hypometabolism. The close correlation of leptin with body fat mass is similar to observations in healthy subjects. No additional influence of disease state or proinflammatory cytokine activity was found. Leptin does not appear to be a component of the immune response to human pulmonary tuberculosis, and thus it cannot account for the weight loss and anorexia associated with tuberculosis.
Key Words: Basal metabolism body composition cytokines densitometry X-ray energy intake interleukin 6 leptin receptors tumor necrosis factor tuberculosis wasting syndrome
INTRODUCTION
Several million persons will experience wasting this year because they are among the 812 million persons per year worldwide who are newly diagnosed with tuberculosis (1). In an unselected US cohort of patients diagnosed with tuberculosis, 45% lost weight and 26% had anorexia (2). In resource-poor countries, weight loss is an almost invariable sign of tuberculosis (3). Co-infection with human immunodeficiency virus (HIV; 4) and delayed diagnosis of tuberculosis contribute to the burden of wasting in resource-poor settings. Although antimycobacterial treatment often induces weight gain, patients may remain underweight even 6 mo after the initiation of successful chemotherapy (5). Despite the scale of the problem, the pathogenesis of wasting in tuberculosis is largely unknown (6).
Proinflammatory cytokines are prime candidates as causative agents of the metabolic changes that eventually result in tuberculosis-associated wasting (7). However, evidence for such a link between the immune response and wasting is equivocal and incomplete, as recently reviewed (6). In the search for other mediators, leptin has emerged as a new candidate.
Leptin is best known as a key mediator of energy metabolism, and it reports the status of body energy stores to feeding centers in the hypothalamus (8). In addition, leptin is now recognized as both a recipient and an effector of immune stimuli, belonging to the same class of cytokines as interleukin 6 (IL6; 9, 10). It has been suggested that leptin mediates anorexia in chronic inflammatory states (11). In patients with pulmonary tuberculosis, a recent study found increased leptin concentrations and a correlation with increased concentrations of tumor necrosis factor (TNF-; 12).
In this report, we present leptin, body composition, and energy metabolism data from a prospective observational study of patients with pulmonary tuberculosis. The primary hypothesis was that, after control for fat mass, leptin would be higher in patients during active tuberculosis than after their recovery. Furthermore, we hypothesized that higher leptin concentrations would correlate with increased proinflammatory cytokine activity, increased resting energy expenditure (REE), and reduced food intake.
SUBJECTS AND METHODS
Study design and subjects
In this prospective longitudinal study, measures of energy metabolism, body composition, and immune response were assessed at 3 time points: within 3 d of the initiation of antimycobacterial treatment, after 1 mo of treatment, and after 6 mo of treatment. Patients (n = 32) with a clinical diagnosis of pulmonary tuberculosis were recruited from 2 London hospitals. Diagnosis was based on standard microbiologic or clinical criteria, ie, acid-fast bacilli in sputum microscopy, Mycobacterium tuberculosis in sputum culture, or a combination of typical signs on chest X-ray, typical clinical symptoms, and clinical response to empirical antimycobacterial treatment. Patients were required to be 18 y old and able to give written informed consent.
Patients were excluded if HIV co-infection was documented but were allowed to enter the study if the HIV status was still unknown. Patients who had undergone surgery, pregnancy, or childbirth < 2 mo before the study were excluded. Further exclusion criteria were severe renal, hepatic, or cardiac insufficiency; diabetes mellitus; or intake of corticosteroids at baseline. Patients with a final primary diagnosis other than tuberculosis and those later diagnosed with HIV co-infection were excluded from the data analysis. Healthy subjects recruited from staff and students of St Georges Hospital Medical School [7 women, 4 men; age: 23 ± 5 y; body mass index (in kg/m2): 22.5 ± 3.0] served as a control group for REE only, as described below and elsewhere (13).
Measurement methods
A heparinized blood sample was taken around 0900 after an overnight fast and centrifuged at 1215 x g for 30 min at 4 °C and then at 2380 x g for 15 min at 4 °C to obtain platelet-poor plasma, which was immediately frozen at -80 °C. Repeated freeze-thaw cycles were avoided. Leptin, IL-6, TNF-, and soluble TNF- receptor type 1 (sTNFR-1) were measured by enzyme-linked immunosorbent assay with the use of paired antibodies (R&D Systems, Abingdon, United Kingdom). All measurements were done in triplicate. REE was measured by indirect calorimetry on a Deltatrac calorimeter (Datex-Ohmeda, Helsinki). Gas calibration was performed before each measurement. To prevent cross-infection in patients with active pulmonary tuberculosis, the tube connecting the canopy to the calorimeter was replaced with a single-use disposable polyethylene tube attached to an airway filter (Ultipor BB50TE; Pall Corp, Newquay, United Kingdom). A pilot study in the 11 healthy controls found that, although flow was reduced by 40%, these changes in the Deltatrac setup did not introduce a systematic error. Validation of this method is described in a separate report (13). REE was compared between patients and controls and with values predicted by the Harris-Benedict equation (14). Food energy intake was assessed by 24-h recall protocols and a food-frequency questionnaire covering the previous wk. The higher estimate of energy intake from these 2 methods was used for further analysis. Professional interpreters were available during the assessment for any subjects who were not fluent in English.
Body weight was measured on a calibrated scale to the nearest 0.1 kg while the patient was wearing light clothing. The weight of such clothing, calculated from the average of 10 measurements of each standard clothing item, was subtracted. Height was measured on a calibrated wall-mounted scale to the nearest 0.1 cm. Body fat was determined by dual-energy X-ray absorptiometry (Lunar DPX; Aura Scientific, Milton Keynes, United Kingdom). The local research ethics committee approved the study, and the rules of the Helsinki Declaration (revision of 1983) were followed.
Statistical analysis
Because of skewed distribution, concentrations of leptin, IL-6, and TNF- were log transformed for further statistical analysis. Prior weight loss was categorized as 10%, < 10%, and unknown. The ratio between reported energy intake and measured REE was used as a variable of energy balance. Changes between time points (ie, baseline, 1 mo, and 6 mo) were tested by repeated-measures analysis of variance with sex, time, and sex-by-time interaction as independent variables. Data are given as mean (± SD) unless indicated otherwise. The correlation between fat mass and leptin was tested by analysis of variance, with controls for time and sex. Two methods were used to test whether associations between leptin and values of proinflammatory cytokine response or energy metabolism were partly independent from fat mass and energy balance. For continuous variables, this testing was done by partial correlation, calculated separately for each time point. For categorial variables (time points, sex, prior weight loss), this testing was done by calculating the residuals for leptin, ie, the difference between measured and predicted log10 (leptin), from linear regression against percentage body fat and energy balance. Residuals were then compared between time points or between subgroups at each time point by paired or unpaired t test. Significance testing was adjusted for multiple comparisons with the use of the Bonferroni method. Statistics were calculated with SPSS software, version 10.0 (SPSS Inc, Chicago; Internet: http://www.spss.com).
RESULTS
Baseline characteristics
Between June 1999 and November 2000, 102 patients with a clinical diagnosis of pulmonary tuberculosis were screened for eligibility. Diagnosis was confirmed in 80 of them, but 24 had concomitant diseases that excluded them from the study (5 diabetes, 11 HIV co-infection, and 8 other). Sixteen patients did not give informed consent. Therefore, 40 patients entered the study, 32 (80%) of whom completed all assessments. One patient died of a cause unrelated to tuberculosis, and 7 patients withdrew for personal reasons. The final analysis was therefore based on 32 patients, 23 (72%) of whom were male. The age of the subjects was 42.5 ± 21.1 y (range: 1884 y). The racial or ethnic background was South Asian, white, African, and Hispanic in 18, 9, 4, and 1 patients, respectively. All patients but one were considered to have made a satisfactory clinical response to anti-tuberculous chemotherapy at the end of 6 mo, and none were found to have multi-drugresistant mycobacteria. One patient developed new pulmonary infiltrates in the 4th mo of antimycobacterial treatment but recovered during an extended 9-mo course of treatment.
Nutritional status
The body mass index at baseline was 19.7 ± 2.6 (range: 15.224.5). A weight history could be obtained in 28 patients, 26 (92.9%) of whom reported a weight loss during the previous year of 11.6 ± 6.7% of their usual body weight (range: 1.326.5%). Weight loss had first been noted 151 ± 70 d before baseline (range: 16724 d).
Clinical recovery was accompanied by increases in weight and body fat, both in absolute values and in percentage of body weight (Table 1). These changes tended to be more pronounced in men than in women, although this sex difference was not statistically significant (Table 1). Longitudinal gain over 6 mo amounted to 9.4 ± 8.4% for weight (ranging from 4.7% loss to 33.3% gain) and 41.7 ± 3.1% for body fat (ranging from 27% loss to 238% gain). This weight gain represented recovery of most of the weight previously lost: patients reached 93.1 ± 6.9% of their pre-illness weight after 1 mo and 99.0 ± 8.8% after 6 mo.
View this table:
TABLE 1 . Body composition and leptin concentrations during treatment of tuberculosis, by group1
Energy metabolism
As expected, absolute REE was higher in men than in women (5.8 ± 0.7 and 4.9 ± 1.0, respectively, at baseline; P < 0.001). However, neither changes in REE during treatment nor REE results normalized for age, sex, height, and weight (Harris-Benedict prediction; 14) or for fat-free mass differed significantly between men and women. Therefore, all further data are presented for the sexes combined (Table 2). REE did not differ significantly from the value predicted by Harris-Benedict at baseline and at 1 mo, but it was significantly (P < 0.01) lower than predicted at 6 mo (Table 2). Because measured REE was also significantly lower than predicted in 11 healthy controls (92.9 ± 9.1%, P < 0.05), REE as a percentage of the predicted value did not differ significantly between healthy controls and patients at any time point. REE was highly reproducible for individual patients, differing +3.8 ± 2.8% from the individual baseline at 1 mo and -1.7 ± 0.3% at 6 mo. Likewise, REE did not change between time points when expressed per kilogram of fat-free mass (Table 2).
View this table:
TABLE 2 . Cytokines and energy metabolism during treatment of tuberculosis1
By contrast, energy intake increased significantly during treatment (Table 2). Intra-individual increases were 35 ± 8% from baseline to 1 mo and 44 ± 17% from baseline to 6 mo (P < 0.01).
Cytokines
Plasma IL-6, sTNFR1, and C-reactive protein were above the reference range at baseline, and they decreased by 1 mo and 6 mo, without a difference between the sexes (Table 2). By contrast, no change in plasma TNF- concentration was found.
Leptin
Fasting plasma leptin concentrations increased between baseline and 1 mo and 6 mo (Table 1). The median increase in leptin in individual patients amounted to 42.9% during the first month and to 84.7% over the entire 6-mo treatment. Leptin concentration and percentage fat mass (%FM) were strongly correlated, following a log-linear curve (Figure 1). With each increase in %FM, leptin was estimated to increase by a factor of 1.043 (95% CI: 1.037, 1.051), after we controlled for time and sex. This corresponds to a doubling in leptin concentration for each 16% increment in %FM. Women had higher leptin concentrations than did men (Table 1).
FIGURE 1. . Correlation between leptin concentration and percentage fat mass. Scatterplots of leptin concentration (logarithmic scale) against percentage fat mass at 3 time points (n = 32 each) are combined in this graph: baseline (), 1 mo (), and 6 mo (). Correlation coefficients were r2 = 0.55, r2 = 0.80, and r2 = 0.84 at baseline, 1 mo, and 6 mo, respectively. Regression lines did not differ significantly between time points.
Patients with weight loss > 10% were compared with those with lesser degrees of weight loss. Patients with such a degree of wasting had lower baseline leptin concentrations than did patients with lesser degrees of weight loss (6.4 ± 7.3 and 9.0 ± 9.9 x 10-6 g/L, respectively; P = 0.02); they also had significantly lower baseline %FM (P = 0.04) but similar concentrations of circulating cytokines and similar values for energy metabolism (data not shown).
Correlations between leptin, cytokine concentrations, and energy values were explored with and without control for %FM and energy balance and at baseline and 6 mo (Tables 3 and 4). No further significant correlations between these variables were found at 1 mo (data not shown). Log(leptin) was not significantly correlated to any of the cytokines. Neither REE nor reported food intake was correlated with leptin in this study. After control for %FM and energy balance by calculation of the residual leptin concentration, no significant difference between time points was found. Residual leptin at 1 mo was 1.27 times (95% CI: 0.94, 1.71) the baseline value, and at 6 mo, it was 1.23 times (95% CI: 0.91, 1.68) the baseline value. Residual leptin concentration also did not differ at any time point between men and women or between patients who did and patients who did not have weight loss > 10%. Changes in log(leptin) and changes in %FM did not correlate between baseline and 1 mo, but they were found to be strongly correlated between 1 mo and 6 mo (r 2 = 0.74, P < 0.001). Thus, %FM and energy balance were the only variables found to be associated with leptin, whereas disease status and cytokine response to tuberculosis were independent from leptin.
View this table:
TABLE 3 . Pearsons bivariate (top right triangular area) and partial (bottom left triangular area) correlation coefficients (r) between leptin, proinflammatory cytokines, and energy balance at baseline1
View this table:
TABLE 4 . Pearsons bivariate (top right triangular area) and partial (bottom left triangular area) correlation coefficients (r) between leptin, proinflammatory cytokines, and energy balance at 6 mo1
DISCUSSION
This study provides no support for the concept that leptin is the missing link between immune defense and wasting in pulmonary tuberculosis. Leptin concentrations closely reflected %FM and were not correlated with the proinflammatory cytokine response in active tuberculosis. No additional influence of disease activity or inflammatory cytokines was found.
In theory, leptin would have been an intriguing candidate for this role. Leptin concentrations closely reflect %FM and energy balance (15), but a growing number of additional roles in neuroendocrine regulation are being recognized for this hormone (16, 17). Leptin is both a recipient of and a stimulus for immunologic signals. In animal models, its production increases in response to bacterial lipopolysaccharide (10, 11) and turpentine (9). Its molecular structure is similar to that of the IL-6 cytokine family (18). Furthermore, it has a trophic effect on components of the immune system. The immunosuppression induced by starvation in mice is abolished if leptin is substituted (19), including stimulation of a Th1 pattern of proinflammatory cytokine production (20).
Evidence for such a role for leptin in humans is equivocal. Ex vivo stimulation of human peripheral blood mononuclear cells with leptin results in a Th1 pattern of cytokine release (21). Increased leptin concentrations were found in patients with sepsis (2224). It is not known to what extent such findings are explained by hyperglycemia and hyperinsulinemia, which are common findings in such patients and a major stimulus for leptin production (25). Leptin concentration increases during acute surgical stress but returns to normal within 24 h (2629). In patients with the AIDS wasting syndrome, leptin concentration was reduced, but this finding was explained by lower %FM rather than by the presence of HIV infection per se (30, 31). Similar findings were reported in patients with cancer-related cachexia (32). Finally, people with genetic leptin deficiency have largely normal immune function (33). Taken together, these data suggest that the leptin system reacts to acute metabolic stress but does not to contribute to chronic wasting in humans.
Leptin has been suggested as a component of a Th1 pattern of cytokines, the same pattern that is required for protective immunity against M. tuberculosis (34). The production of interferon and TNF- is crucial to the host defense against tuberculosis (3537), but it may also be associated with anorexia and fever (38). Stimulation of leptin production during active tuberculosis may therefore contribute to wasting, in concert with other proinflammatory cytokines. Indeed, one study found higher leptin concentrations in patients with active pulmonary tuberculosis than in healthy controls, and leptin increased further during treatment (12). A positive correlation between TNF- and leptin concentrations was found at baseline. However, both effects were significant only in 8 female patients, and not in 22 male patients. Body fat mass was estimated from BMI rather than being directly measured in that study (12). Such estimates are of very limited validity (39).
By using dual-energy X-ray absorptiometry as an accurate measurement of body fat, we found a close correlation between leptin and %FM, which followed a curve similar to that found in healthy subjects (40). Correlations between leptin concentration and values of energy metabolism gave inconsistent results. The weak negative correlation between leptin concentration and REE and the lack of an association with energy intake are at odds with findings of other studies in which leptin was strongly influenced by recent energy balance (41).
Because of the short-lived and paracrine effects of TNF- itself, concentrations of sTNFR-1 were used as a marker of TNF- activity (38, 42). The sTNFR-1 concentrations decreased with recovery from tuberculosis, as previously shown in patients co-infected with M. tuberculosis and HIV (43). Concentrations of sTNFR-1 and leptin were not correlated to each other in the present study. By contrast, a positive correlation between sTNFR-1 and leptin was found in obese healthy subjects, after control for body fat mass (44). This may reflect different metabolic roles of the TNF- system in obesity and in infection. Adipose tissue itself is a site of production of both TNF- and sTNFR-1, most probably for purposes other than defense against infection, and such production may become apparent in the peripheral blood of obese persons (45).
Some limitations of this study should be noted. Plasma concentrations may not always reflect the biologic activity of compounds such as leptin and cytokines, in which diurnal rhythms and pulsatile release may occur (46) and the interaction with circulating receptors may be important (42, 47). Food intake was measured with the use of a 24-h recall protocol and a food-frequency questionnaire, methods that are often biased by over- or underreporting (48). Such limitations are shared with other studies on this subject (12, 2224). However, this study has the benefit of systematic direct assessment of body composition by a validated technique. Our observations show that data on leptin concentrations in clinical human studies may be misleading if fat mass and energy metabolism are not measured concomitantly.
In conclusion, leptin does not appear to be part of the proinflammatory cytokine response in human pulmonary tuberculosis. Changes in leptin are entirely appropriate for the changes in body fat mass and energy balance. Altered leptin activity cannot, therefore, be held responsible for the weight loss and anorexia so often associated with tuberculosis infection.
ACKNOWLEDGMENTS
We are grateful to the nurses and physicians in the departments of Infectious Diseases and Chest Medicine, St Georges Hospital, and of Chest Medicine, St Helier Hospital (N Cooke), who were of great help in recruiting patients to this study. We are also grateful to all patients who contributed their time and effort to the study.
REFERENCES