INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Low bone mineral density (BMD) from osteoporosis or osteomalacia complicates the lives of many children and adults with cystic fibrosis (CF). Numerous reports from the United States and Europe have documented a high prevalence of low BMD in CF patients (1), and several studies have recently shown that fracture and kyphosis rates in these patients are significantly higher than normal (4, 5). The pathogenesis of bone disease in CF is not well understood (6). Many factors including, but not limited to, poor nutrition, delayed puberty, vitamin D insufficiency (secondary to pancreatic insufficiency and diminished exposure to sunlight), hypogonadism, and reduced activity are likely to play important roles in bone disease (7). Additional alterations in calcium homeostasis in CF include calcium malabsorption and mild secondary hyperparathyroidism (8). Associations between low BMD have been reported with serum vitamin D levels, poor nutrition (i.e., low BMI), lung function, and age (9). Reported correlation coefficients for these variables suggest that a significant portion of low BMD in CF occurs for reasons other than those just cited.

Bone remodeling is a complex process. Recently, the role of cytokines in bone metabolism as it pertains to bone loss has been intensely studied (10). In particular, tumor necrosis factor (TNF)- and interleukin (IL)-1, both of which are acute-phase cytokines, have been shown to be powerful promoters of bone resorption both in vitro and in vivo through osteoclast activation and osteoclastogenesis, the latter occuring by both direct mitogenic effects and the indirect stimulation of colony stimulating factors, IL-6, and IL-11 from stromal cells and osteoblasts. IL-1 and TNF- also inhibit bone formation (10, 11). IL-6, a pleiotropic inflammatory cytokine, although not directly involved in bone resorption, potentiates bone breakdown induced by other cytokines (12).

Lung infection in CF causes increases in serum TNF-, IL-1, and IL-6 levels. Conventional treatment leads to decreases in the levels of these cytokines, but even after treatment their levels remain elevated when compared with those of healthy controls (13, 14). For this reason, CF patients provide a unique opportunity to study the dynamic interactions between inflammatory cytokines and bone metabolic markers. We hypothesized that pulmonary infection in CF leads to increases in serum acute-phase cytokines (i.e., TNF-, IL-1, and IL-6) that play a role in altering bone remodeling. To test this hypothesis, we monitored both specific (serum TNF-, IL-1, and IL-6 levels) and general (serum C-reactive protein [CRP] and chondrex levels) inflammatory markers, and bone metabolic indicators (i.e., osteocalcin, a marker of bone formation, and urinary N-telopeptide of type I collagen and deoxypyridinolines, both of which are markers of bone breakdown) throughout the course and treatment of lung infection in CF patients.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Patients

The study group consisted of 17 patients with CF, aged 18 to 41 yr. CF had been previously diagnosed in these patients from increased sweat chloride concentrations and an appropriate clinical picture. The Committee on Human Research of the School of Medicine, University of North Carolina (UNC) approved the prospective study, and written consent was obtained from all participants. The main inclusion criterion was admission to UNC Hospitals for a pulmonary exacerbation of CF. All study subjects were recruited within 72 h of admission and were studied between June 1998 and August 1999. The exacerbation was diagnosed by the subjects' primary CF caregivers (none of whom was involved directly in the study) on the basis of increased sputum production, cough, fatigue, and acute reductions in pulmonary function. Pregnancy, corticosteroid use, or CF exacerbation within the preceding 2 mo, and renal failure (creatinine > 176.8 µmol/L or 2 mg/ dl) were exclusion criteria. The CF patients were treated with conventional therapy for lung infection, including intravenous antipseudomonal (and, as needed, antistaphylococcal) antibiotics chosen on the basis of recent sputum culture results (usually tobramycin and either ceftazidine or imipenem), chest physical therapy applied two or three times daily, nebulized albuterol as needed, oxygen in cases of oxyhemoglobin saturation < 90%, and aggressive nutritional intervention.

Protocol

Height, weight, body mass index (BMI), age, medication use, FVC, and FEV₁ were recorded. Blood and urine samples from each patient were collected and analyzed at three different time points: (1) within the first 72 h after the diagnosis of CF exacerbation; (2) at the completion of antibiotic therapy; and (3) 3 wk after the end of antibiotic therapy, when the patient returned to the ambulatory care clinic for routine follow up. The General Clinical Research Center of the UNC Hospitals provided nursing support for measurements, sample acquisition, centrifugation, labeling, and storage. Serum samples were obtained by collecting blood by venipuncture into two serum separator tubes, clotting the blood for 40 min at room temperature, and centrifuging at 10,000 rpm for 12 min. Serum was stored at 70° C. A second morning void specimen of urine was collected and a sample was sent to the Clinical Chemistry Laboratory of UNC Hospitals to obtain a creatinine measurement with an automatic analyzer (Hitachi 911; Boehringer Mannheim, Indianapolis, IL). The urine was shielded from light during transport and storage to avoid breakdown of bone markers. The remaining urine was stored at 70° C.

Assays

All longitudinal samples from a single individual were analyzed in duplicate in the same assay after thawing of the stored specimens. All normal-range data were provided by the assay manufacturer unless otherwise specified.

N-telopeptides of type I collagen were measured in urine with an enzyme-linked immunosorbent assay (ELISA) (Ostex International Inc., Seattle, WA). All N-telopeptide data were corrected on the basis of urinary creatinine levels and were expressed as nanomoles of bone collagen equivalents (BCE) per millimole of creatinine. The normal ranges for premenopausal women and men are 35.6 ± 13 (mean ± SD) and 27 ± 12 nmol BCE/mmol creatinine, respectively. Intraassay variability was 5 to 19%, and the lower limit of detection was 20 nm BCE. Free deoxypyridinoline was measured in urine with the Pyrilinks-D two-site immunoassay (Metra Biosystems, Mountain View, CA). Data generated were corrected on the basis of urinary creatinine levels. The intraassay precision of the immunoassay averages 6% and the interassay precision averages 4%. The normal range for premenopausal women is 5.5 ± 2.3 nmol/mmol creatinine. Serum osteocalcin was measured with an immunoradiometric assay (Elsa-Osteo; CIS US Inc., Bedford, MA) and a gamma scintillation counter (LS6500; Beckman Instruments, Fullerton, CA). This assay has a detection limit of 0.4 ng/ml of osteocalcin and an intraassay variability of 3.8 to 3.9%. The normal osteocalcin values for males and females in our subject group age are 22.7 ng/ml (range: 10.7 to 37 ng/ml) and 21.8 ng/ml (range: 8.8 to 39.4 ng/ml), respectively.

All cytokines (IL-1, IL-6, and TNF-) were assayed in serum with Quantikine High Sensitivity ELISAs (R&D Systems, Minneapolis, MN) specific for each particular cytokine. For the TNF- assay, the normal serum range is 0 to 3.62 pg/ml (mean: 1.25 pg/ml) and the intraassay variation is 14.3%. For the human IL-1 assay, the normal serum range is 0 to 1.996 pg/ml (mean: 0.536 pg/ml), with an intraassay variability of 6.9 to 10.2%. For the IL-6 assay, the normal serum range is 0.378 to 10.1 pg/ml (mean: 1.62 pg/ml), with an intraassay variability of 3.8 to 5.9%.

Serum levels of CRP and chondrex were measured as markers of acute inflammation. Serum CRP levels were measured with an enzyme immunoassay (EIA) (Hemagen Diagnostics, Waltham, MA). The assay range is 1.0 to 50.0 µg/ml and the normal serum range is 0.07 to 29.0 (median: 0.6 to 1.9) µg/ml. The coefficient of variation (CV) is 6 to 25%. Serum chondrex levels were measured with a two-site ELISA (Metra Biosystems). This assay measures YKL-40, a 40 kD glycoprotein secreted by cultured human macrophages, neutrophils, and other cells. It is thought to play a role in cartilage remodeling and is an inflammatory marker (15). Normal serum chondrex values are 25 to 93 ng/ml for females and 24 to 125 ng/ml for males, with a minimal detectable level of 20 ng/ml and an intraassay variability of 5.6 to 6.6%.

Statistical Analysis

Statistical analyses were performed with commercially available software (SAS version 6.12; SAS Institute, Cary, NC) (16). We tested the two distinct hypotheses that: (1) bone metabolic markers would improve with treatment of CF lung infection, and (2) bone metabolic markers would show sustained improvements at 3 wk after the cessation of treatment. Separate analyses were conducted to determine whether values obtained at Time 2 and Time 3 differed from those at baseline (Time 1). Pulmonary function endpoints, serum levels of cytokines, osteocalcin, chondrex, and CRP, and urine levels of N-telopeptides and free deoxypyridinoline were compared with Student's t tests for paired samples, using a two-tailed curve. The differences in the data for each of the tested variables were normally distributed. Nonetheless, nonparametric tests (i.e., Wilcoxon's signed ranks test) were performed to confirm the findings with the t tests. Only the t-test analyses are presented, since results were similar with both analyses. In all analyses, a value of p < 0.05 was considered significant with two-tailed tests. Using Bonferroni's correction for multiple analyses with seven different variables in the testing of each hypothesis would lead to the use of a much more conservative p value (i.e., 0.007). Univariate regression analysis was used to test associations between variables; our sample size was too small for multivariate analyses.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Clinical Endpoints

Subject characteristics are given in Table 1. At study entry, seven patients were taking inhaled corticosteroids, four had diabetes mellitus (three of whom required insulin), one had asymptomatic cholestasis (serum gamma glutamyl transferase and alkaline phosphatase more than three times the respective upper limits of normal for the assays), but none had abnormal liver transaminase levels, and five were wait-listed for lung transplantation. Patients were initiated into the study, on average, at 1.6 ± 0.8 (mean ± SD) hospital days. The length of intravenous antibiotic treatment was 17.5 ± 7.6 d (range: 13 to 28 d). Clinical improvement in response to treatment was assessed through symptoms, sputum production, and increases in pulmonary function test results (FEV₁: pretreatment: 1.28 ± 1.20 L, posttreatment: 1.45 ± 1.20 L, p = 0.020; FVC pretreatment 2.06 ± 1.37, posttreatment 2.38 ± 1.41 L, p = 0.012). For all patients, the determination of type, duration, and response to therapy was made by the attending physician on the pulmonary service, who was completely uninvolved in the conduct of the study. The three sample collections occurred at 1.1 ± 0.9 d, 18.2 ± 5.1 d, and 39.7 ± 12.1 d, respectively, after the beginning of treatment.

Cytokines, Inflammatory Markers and Bone Metabolites

Osteocalcin and N-telopeptide levels are shown in Figure 1. Osteocalcin levels were depressed as compared both with those of age-matched, ambulatory CF patients (28.4 ± 7.9 ng/ml) and age-matched, healthy controls (27.8 ± 6.7 ng/ml) (17) at baseline (14.5 ± 5.4 ng/ml), and rose significantly by the end of treatment (22.5 ± 8.7 ng/ml, p = 0.010) and 3 wk later (33.0 ± 21.7 ng/ml, p = 0.006). Urine N-telopeptide levels were increased at baseline (147.3 ± 77.5 BCE/mmol creatinine) and fell significantly by the end of treatment (95.5 ± 57.3 BCE/ mmol creatinine, p = 0.0014) and 3 wk later (76.4 ± 48.7 BCE/ mmol creatinine, p = 0.008), but remained higher (p < 0.003) than assay control levels at all time points. Urine deoxypyridinoline levels were increased at all time points as compared with those of assay controls (p = 0.003 to 0.06), trended downward (p = 0.08) by the end of antibiotic therapy, and then increased 3 wk later (8.4 ± 2.8, 6.8 ± 3.0, and 8.0 ± 4.3 nmol/ mmol creatinine, respectively). N-telopeptide levels were not significantly associated with either deoxypyridinoline or osteocalcin levels (p > 0.2 for both). Renal function, as measured by serum creatinine, remained unchanged throughout antibiotic therapy (baseline and end-of-treatment, both: 53.0 ± 8.8 µmol/L, or 0.6 ± 0.1 mg/dl, p > 0.2).

View larger version (18K): [in this window] [in a new window]   Figure 1.   Serum levels (mean ± SEM) of N-telopeptides (NTx; open bars) decreased and those of osteocalcin (OC; closed circles) increased significantly from the beginning to the end of treatment, and remained significantly different from pretreatment levels at Day 40 (p  0.013 for all comparisons).

TNF-, IL-1, and IL-6 levels were all highest at the beginning of the exacerbation and declined by the end of treatment, with IL1- and IL-6 showing statistically significant differences (p  0.015 for both) (Figure 2A). TNF- levels, although lower after treatment, were not significantly different from those at the start of treatment or 3 wk later (3.82 ± 2.03, 3.16 ± 1.39, and 3.60 ± 1.60 pg/ml, respectively, p = 0.11 for Day 1 versus Day 18). Levels of these cytokines rose from Day 18 to Day 40 to values intermediate between those before and after treatment (p > 0.2 for Day 1 versus Day 40). As general markers of inflammation, CRP (p = 0.04) and chondrex (p = 0.006) levels fell from the beginning to the end of treatment and remained persistently lower at Day 40 (p = 0.05 for both, versus Day 1) (Figure 2). IL-1 levels were associated with N-telopeptide levels (r = 0.41, p = 0.01). Osteocalcin levels were inversely associated with chondrex levels (r = 0.33, p = 0.02). TNF-, IL-1, and IL-6 levels were associated with each other (r = 0.33 to 0.40, p < 0.02 for each) and with levels of chondrex (r = 0.32 to 0.61, p < 0.02 for each), but not with levels of CRP.

View larger version (29K): [in this window] [in a new window]   Figure 2.   (A) Serum levels (mean ± SEM) of IL-1 (solid line) and IL-6 (dashed line) decreased significantly (p  0.015 for both comparisons) from the beginning to the end of treatment, but increased at Day 40 (p = NS for Day 1 versus Day 40). (B) Serum levels (mean ± SEM) of CRP (solid line) and chondrex (dashed line), both general markers of inflammation, decreased significantly from the beginning to the end of treatment (p  0.04 for both), and remained significantly different from pretreatment levels at Day 40 (p  0.05 for all comparisons). See text for specific p values.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Three important findings resulted from this study. First, bone metabolic markers were adversely altered by exacerbation of lung infection in CF patients, and responded favorably to treatment of infection. However, markers of bone resorption (N-telopeptides and deoxypyridinoline) remained significantly above the normal range despite treatment of lung infection. Second, cytokines (TNF-, IL-1, and IL-6) thought to be important in increasing bone resorption and inhibiting bone formation were increased at the beginning of treatment for lung infection and fell in response to therapy. Third, bone remodeling was uncoupled, with accelerated bone breakdown and inadequate bone formation throughout the course of the study in a manner similar to, but greater in magnitude, than that previously found in ambulatory CF patients (17, 18). Although a study of this kind cannot prove cause and effect, the temporal linkage between initial elevations and subsequent declines in serum TNF-, IL-1, and IL-6, and in urinary N-telopeptides, is evidence of a potentially important interaction of these substances with bone metabolism. The rebound of serum osteocalcin levels to normal in response to treatment provides further evidence for the deleterious impact of pulmonary inflammation on bone physiology. Moreover, the statistically significant associations between several inflammatory and bone metabolic markers strengthens the potential causal role for lung inflammation in bone pathology. Taken together, these results provide insights into the impact of inflammatory diseases on bone metabolism, suggest additional, plausible mechanisms for CF bone disease, and may influence the therapeutic options for improving bone health in CF patients and other patients with systemic inflammatory diseases.

Our results are in keeping with those of Norman and colleagues, Nixon and associates, and others who have reported increases in serum cytokine levels in CF patients at the start of treatment of a CF exacerbation and a subsequent decline in these levels following the completion of therapy (12, 13). The source of these inflammatory cytokines is presumed to be the lung, because sputum and bronchoalveolar lavage studies have shown local increases in these cytokines (19). Preliminary studies of markers of bone resorption in CF have produced conflicting data on the impact of pulmonary infection on bone metabolism. Cunningham and coworkers demonstrated that urinary pyridinoline and deoxypyridinoline levels in seven children with pulmonary infection from CF were lower than those of healthy controls and did not change over a 6-wk period of treatment for lung infection (24). Ionescu and colleagues confirmed that urine deoxypyridinoline levels did not change, but that serum TNF- and IL-6 levels declined in response to treatment of infection in CF (25). Their results also suggested an association between abnormalities in bone metabolism and a greater inflammatory status and persistent catabolic state as measured by serum TNF- and IL-6 levels. Haworth and associates were the first to show an association between an inflammatory marker (i.e., serum CRP levels) and bone density (Z-scores) in a large cohort of adult CF patients (r = 0.33, p < 0.01) (26).

Clinical reports of cases other than CF have drawn links between cytokines and alterations in bone remodeling. The best studied clinical area is that of postmenopausal osteoporosis, in which estrogen deficiency, whether natural or surgical, has been associated with increased peripheral blood monocyte production of IL-1, IL-6, TNF-, and colony-stimulating factors (27). The changes in these cytokines occur in a temporal sequence, suggesting that they have a causal role in the pathogenesis of ovariectomy-induced bone loss (28). Furthermore, the bone-resorbing activity of the supernatants of cultured peripheral blood monocytes obtained from both pre- and postmenopausal women has been correlated with supernatant IL-1, IL-6, and TNF- levels and with total pyridinoline excretion (29). Other conditions characterized by an increased production of IL-1 or TNF-, including rheumatoid arthritis, idiopathic hypercalciuria, Paget's disease, inflammatory bowel disease (IBD), and endometriosis, are associated with a higher prevalence of osteopenia and osteoporosis (30). Other evidence suggesting a causal role for inflammatory cytokines in inducing bone disease comes from reports that serum from Crohn's disease patients adversely affected bone formation in an in vitro rat calvarium tissue model (34), and that IBD patients with osteoporosis had higher serum IL-6 levels than IBD patients without osteoporosis (35).

Our study has several potential limitations. First, although provocative, the association between acute-phase cytokines and markers of bone breakdown and formation does not prove causality. The relatively low r values for these associations is probably due to the complexity of the underlying pathophysiology, in which multiple mediators might affect bone metabolism in a variety of ways. In addition, a number of other (unmeasured) factors, including, but not limited to, sex hormones, colony stimulating factors (e.g., macrophage colony-stimulating factor and granulocyte-macropahge colony-stimulating factor), IL-6-like cytokines (e.g., IL-11, leukemia inhibitory factor, oncostatin M, ciliary neurotrophic factor, cardiotrophin-1), and IL-17 may be related to the changes in bone metabolic markers (10, 28). Second, both the exacerbation of pulmonary infection and the response to therapy are heterogeneous processes in CF. Thus, a better, albeit more difficult, study design, assessing endpoints on a daily basis and studying a more clinically homogeneous patient population with regard to lung function, age, and other variables, may provide more meaningful data. Third, some controversy exists over the reliability of spot urine levels of deoxypyridinoline and N-telopeptides and serum levels of osteocalcin as indicators of bone resorption and formation, respectively. However, these molecules are generally regarded as the most sensitive and specific markers of bone resorption and formation yet discovered. Also, we studied moderately to severely ill CF patients, and our findings might have been different in patients who were in better health. Furthermore, the data support the possibility that lung-derived inflammatory cytokines promote bone resorption in CF, but do not demonstrate an actual loss in bone mass. Further studies will be needed to determine whether episodic bone hyperresorption and diminished bone formation occurring with infections are pathophysiologic changes that lead to progressive bone loss and osteoporosis.

In conclusion, this study provides an important link between pulmonary infection and inflammation in CF and unfavorable alterations in bone metabolism. We speculate that accelerated bone loss and, possibly, diminished bone formation, modulated at least in part by cytokines, may result from intermittent exacerbations of chronic pulmonary infection in CF. Similar mechanisms may operate in patients with other chronic inflammatory disorders and contribute to bone disease. The pathogenesis of bone disease in CF is undoubtedly multifactorial, with many potential variables affecting bone remodeling. Further study will be needed to determine which factors are most important in the bone disease that occurs in the setting of CF and other chronic inflammatory disorders. Future studies, designed to elucidate the relative importance of TNF-, IL-1, and IL-6 in bone hyperresorption, will use in vitro assays with cytokine antagonists. Longitudinal measurements of serum cytokines and BMD in CF patients may prove useful in determining the role of cytokines in bone loss.