Novel Serum Markers of Bone Resorption: Clinical Assessment and Comparison with Established Urinary Indices

Abstract

Although urinary measurements of collagen degradation provide valid estimates of bone resorption, their clinical application is hampered by pronounced analytical and biological variability. Therefore, immunoassays for the determination of such parameters in serum have been developed. In this study, we assessed the performance of three new serum markers of bone turnover, i.e., C-terminal and N-terminal telopeptides of type I collagen (S-CTX and S-NTX) and bone sialoprotein. Results were compared with urinary total pyridinoline, total deoxypyridinoline, and urinary C-terminal telopeptides of type I collagen (U-CTX) and urinary N-terminal telopeptides of type I collagen (U-NTX). The study population included healthy men (n = 27), premenopausal (n = 30) and postmenopausal (n = 31) women, patients with hepatic dysfunction (HF, n = 24), renal failure (RF, n = 30), breast cancer without (BC–, n = 24) and with (BC+, n = 30) bone metastases, primary vertebral osteoporosis (OPO, n = 27), primary hyperparathyroidism (PHPT, n = 16), active Paget's disease of bone (n = 18), multiple myeloma (MM, n = 18), and patients with hypercalcemia of malignancy before and after treatment with pamidronate (HOM, n = 28). Changes in urinary and serum markers were similar in most metabolic bone diseases. However, differentiation between healthy controls and OPO, or PHPT, was improved by the serum markers. In MM, all serum and urinary markers were elevated (p < 0.05 vs. controls). In BC+, skeletal involvement was reflected by significant increments in all indices (p < 0.01 vs. BC–), except U-CTX and S-CTX. In HOM, pamidronate-induced changes in biomarkers were most pronounced for U-CTX and S-CTX and S-NTX. HF and RF were associated with elevated levels of all serum markers (p < 0.05 vs. controls). In conclusion, measurements in serum reflect bone resorption to the same extent as the urinary indices. Since serum markers circumvent some of the limitations of urinary measurements, their use potentially improves the assessment of skeletal disorders.

INTRODUCTION

Over the last decade, the assessment of bone diseases has been greatly improved by the development of specific and sensitive markers of bone metabolism.⁽¹⁾ Thus, the quantitation of collagen degradation products in urine provides a valid estimate of the rate of bone resorption.⁽²⁾ Presently, the hydroxypyridinium cross-links and the telopeptide-related epitopes of type I collagen in urine are considered the most sensitive measures of collagen breakdown.^(2-4) However, in contrast to serum markers, urinary measurements either require 24-h collections, or, in the case of untimed urine samples, need to be corrected for urinary creatinine. The mere addition of a secondary assay increases the overall variability of bone marker measurements in urine. This and other factors lead to significantly higher analytical and biological variability of the urinary indices.⁽⁵⁾

Recently, assays have been developed for the detection of collageneous and noncollageneous breakdown products of the organic bone matrix in serum.^(6-10) These developments concern the C-terminal and N-terminal telopeptides of type I collagen (CTX and NTX, respectively) and represent logical extensions of the established and well documented urinary assays.^{(8, 9)} Furthermore, a radioimmunoassay for immunoreactive bone sialoprotein (BSP) in serum has been described.⁽¹⁰⁾ BSP is synthesized by osteoblasts and is an important component of the noncollageneous organic matrix of bone. So far results indicate that serum BSP levels reflect processes preferentially related to bone resorption.^(10-12)

In reducing or eliminating some of the technical limitations specific to urine markers, these new assays may help to improve the assessment of skeletal disorders.^{(8-9, 11)} In the present study, therefore, the new immunoassays for serum CTX (S-CTX) and NTX (S-NTX), and for serum BSP, were evaluated in healthy adults, in patients with metabolic and malignant bone diseases, and in patients with nonskeletal disorders. A direct comparison was made between serum and urinary markers of bone resorption.

MATERIALS AND METHODS

Study population

A total of 303 individuals were included in the present investigation. Spot urine and venous blood samples were collected between 8:00 a.m. and 11:00 a.m. without dietary restrictions. Blood samples were centrifuged within 2 h after phlebotomy (1500g for 10 minutes) and stored as aliquots at −80°C until assayed. Urine specimens were protected from light exposure and stored within 2 h of collection at −30°C until analysis.

Healthy controls

Eighty-eight healthy individuals aged 25–82 years were recruited as an ambulatory control population. The group was comprised of 27 men and 30 premenopausal and 31 postmenopausal women (median duration of menopause 7.0 years; range 1.0–30 years). All subjects had a physical examination and routine laboratory measurements. In all individuals over the age of 50 years, radiologic and osteodensitometric evaluations were performed in order to exclude significant degenerative or other bone disease. None of the healthy controls had signs of present or a history of past metabolic bone disease, nor had they taken any medication known to affect bone metabolism, including osteotropic vitamins, calcium supplementation, or hormone replacement therapy.

Nonskeletal diseases

Seventy-eight hospitalized patients with chronic hepatic dysfunction, chronic renal failure (RF), or breast cancer without evidence of bone metastases (BC−) were included as a reference population.

Patients with chronic hepatic dysfunction (HF, n = 24, M/F 10:14) had clinical and biochemical evidence for liver disease. The diagnosis was established on the basis of liver biopsies. Fifteen patients had alcohol-induced liver cirrhosis, 4 had posthepatic cirrhosis, 2 had primary biliary cirrhosis, and 3 had liver cirrhosis of unknown etiology.

Patients with chronic RF (n = 30, M/F 15:15) were included on the basis of impaired glomerular function, as determined by an endogenous creatinine clearance below 50 ml/minute. Renal biopsies were performed in all 30 patients, of whom 13 were diagnosed with glomerulonephritis, 5 with glomerulosclerosis, 6 with renal malfunction of hereditary origin, and 6 with autoimmune disease with renal involvement (lupus erythematodus, Wegener granulomatosis). No signs of secondary hyperparathyroidism (parathyroid hormone [PTH] < 65 μg/l) or calcium imbalance (median serum calcium 2.32 mM/l; range 2.12–2.50 mM/l) were found in any of these patients.

Twenty-four female patients with histologically proven primary BC– were also included. Routine laboratory evaluation was essentially normal in all cases, and metastatic bone involvement was ruled out by radioisotopic bone scan and radiographic evaluation of painful areas.

None of the patients of the reference population had clinical, biochemical, or radiological evidence of metabolic bone disease, and none was terminal ill. Subjects taking medications known to interfere with bone turnover (including glucocorticoids) were excluded from the study.

Metabolic bone diseases

This group comprised 61 patients with primary vertebral osteoporosis (OPO), primary hyperparathyroidism (PHPT), and active Paget's disease of bone (PD).

OPO (n = 27, M/F 9:18) was diagnosed based upon clinical findings and a lumbar bone mineral density of < −2.5 SD below the young adult mean value.⁽¹³⁾ Bone mineral density was measured by dual-energy X-ray absorptiometry (QDR 1000; Hologic, Inc., Waltham, MA, U.S.A.). All patients had plain radiographs of the lumbar spine. Twelve patients had one or more vertebral fractures. None of the patients had evidence of secondary causes for osteoporosis. At the time of sampling, none of the patients received any major antiosteoporotic treatment such as hormone replacement therapy, bisphosphonates, or fluorides.

Asymptomatic PHPT (n = 16, M/F 5:11) was diagnosed by chronic hypercalcemia (see Table 2) in the presence of elevated serum intact PTH levels (>65 μg/l). None of the patients had radiologic signs of PHPT, nephrolithiasis, or gastrointestinal manifestations, or received treatment at the time of sample collection. All individuals had normal renal and hepatic function.

Table Table 2.. Urine and Serum Markers of Bone Resorption in Metabolic Bone Diseases

PD (n = 18, M/F 10:8) was diagnosed upon the basis of characteristic radiographic, scintigraphic, and laboratory findings. In all patients, serum total alkaline phosphatase (S-TAP) was above 250 U/l, indicating active disease. None of the patients had received treatment for at least 6 months before sample collection. Except for S-TAP, all other routine laboratory measurements were within normal range.

Malignant bone disease

A group of 76 patients presented with bone involvement in the presence of underlying malignant disease, including patients with metastatic breast cancer (BC+), multiple myeloma (MM), and hypercalcemia of malignancy (HOM) due to a variety of metastatic tumors.

A total number of 30 female individuals with histologically proven BC+ was included. Metastatic bone disease with one or more bone metastases was diagnosed by radioisotope bone imaging and confirmed by subsequent X-ray skeletal survey. All of the patients were normocalcemic (see Table 3), and all had serum creatinine values within the normal range. None of the BC+ subjects had received antineoplastic or antiresorptive treatment for at least 6 months before sample collection.

Table Table 3.. Urine and Serum Markers of Bone Resorption in Metastatic Bone Diseases

According to the criteria of Durie and Salmon, the diagnosis of MM (n = 18, M/F 9:9) was based on the plasma cell content of the bone marrow (>30% of nucleated cells), on the radiologic evidence of lytic bone lesions and on hematological impairment.⁽¹⁴⁾ Patients were studied at the time of diagnosis. All of them were normocalcemic (see Table 3) and none had received treatment prior to the investigation.

Patients with HOM (n = 28, M/F 17:11) suffered from histologically proven squamous cell carcinomas (n = 5), pancreatic carcinomas (n = 5), renal cell cancer (n = 6), breast cancer (n = 4), MM (n = 5), and metastatic tumors of unknown origin (n = 3). Neoplastic bone involvement was diagnosed by radioisotope bone imaging and subsequent radiographic survey. All patients were studied at the first occurrence of HOM. None of the patients was pretreated with a bisphosphonate, calcitonin or mithramycin. Patients were treated with intravenous 30–60 mg of pamidronate on day 0 and followed for 7 days. During the follow-up, no chemotherapy or other treatment known to interfere with bone turnover (e.g., glucocorticoids, mithramycin, calcitonin) was applied.

The study was approved by the local ethics committees and performed in accordance with the Declaration of Helsinki. Written informed consent was obtained from all individuals prior to the study.

Biochemical measurements

Serum calcium, creatinine, γ-glutamyltransferase, glutamic-oxalacetic, and glutamic-pyruvic transaminase, and albumin were all measured by standard automated techniques. Serum activity of TAP was determined by an automated colorimetric assay using a BM/Hitachi System 704 analyzer (Boehringer Mannheim, Mannheim, Germany) and p-nitrophenyl phosphate as substrate.⁽¹⁵⁾ A commercially available two-site immunoradiometric assay was applied to measure serum intact PTH (Nichols Institute, San Juan Capistrano, CA, U.S.A.).

Urine markers of bone resorption

Urinary total pyridinoline (U-PYD) and total deoxypyridinoline (U-DPD) were measured by automated reverse-phase ion-paired high-performance liquid chromatography using 0-acetylpyridinoline as internal standard, and after acid hydrolysis of the urine.⁽¹⁶⁾ The overall reproducibility of this assay was 8–12%.

Urinary C-terminal telopeptide of type I collagen (U-CTX) was measured using an enzyme-linked immunoassay (Enzymun-Test® β-Crosslaps; Boehringer Mannheim GmbH, Tutzing, Germany) on the automated ES 700 system (Boehringer Mannheim GmbH, Mannheim, Germany).⁽⁴⁾ Intra-assay variability ranged between 2.9% and 5.7%, and interassay variation ranged between 4.7% and 9.4%.

Urinary concentrations of the N-terminal telopeptide of type I collagen (U-NTX) were measured by a competitive enzyme immunoassay (ELItest® NTx; Brahms Diagnostica GmbH, Berlin, Germany) as described elsewhere.⁽³⁾ Mean coefficients of variation (CVs) were 5.8% for intra- and 7.6% for interassay variability.

All urinary results were expressed relatively to the creatinine excretion in urine (nM/mM creatinine) measured by a modified kinetic assay according to the method of Jaffé.⁽¹⁷⁾

Serum markers of bone resorption

Serum immunoreactive bone sialoprotein (S-BSP) was measured by radioimmunoassay as described previously.⁽¹⁰⁾ In brief, 100 μl of¹²⁵I-labeled BSP and 100 μl of chicken anti-human BSP antibody were added to an equal volume of serum or control samples, or standards. After incubation for 24 h at 4°C, 100 μl of a second antibody (donkey anti-chicken immunoglobulin Y) were added. A second incubation followed for 2 h at 4°C and a washing step including centrifugation (2×) for 10 minutes at 2000g, the radioactivity of the pellet was measured in a Berthold γ-counter for 1 minute. Results were calculated by interpolation of the unknown samples with a point-to-point curve fitting equation of the standards. Mean CVs were 7.0% for intra- and 9.1% for interassay variability.

S-CTX were determined using a solid phase enzyme-linked immunosorbent assay.⁽⁸⁾ Fifty microliters of standards, controls, and serum samples were pipetted into streptavidin-coated microtiter wells. One hundred and fifty microliters of a biotinylated monoclonal antibody (MAb) and a peroxidase-conjugated MAb specific for the CTX were added. After incubation for 2 h at room temperature and a washing step, 100 μl of tetramethylbenzidine as the chromogenic substrate were added, and the color reaction was stopped after 15 minutes with 100 μl of sulfuric acid. The absorbance was measured at 450 nm with 650 nm as reference, and calculation of results was performed by interpolation of the unknown samples with a lin-log standard curve. The intra-assay variability ranged between 4.7% and 4.9%, the interassay variability ranged between 5.4% and 8.1%.

Concentrations of N-terminal telopeptide of type I collagen in serum (S-NTX) were assessed by a newly developed competitive inhibition enzyme-linked immunoassay, in modification of the method described by Clemens et al.⁽⁹⁾ In brief, test specimens, controls and standards were diluted 1:5 in specimen diluent, and 100 μl of the diluted samples were added to the NTX antigen-coated microwell plate. One hundred microliters of horseradish peroxidase-labeled MAb directed against NTX were then added to the plate and incubated at room temperature for 90 minutes. At the end of the incubation, the microwells were washed five times, and 200 μl of chromogen reagent/buffered substrate (tetramethylbenzidine and hydrogen peroxide) was added. Following an incubation of 30 minutes at room temperature, the color reaction was stopped with sulfuric acid. Sample measurements were read from a calibration curve with use of a four-parameter logistic curve-fitting program, and results are reported as nanomole bone collagen equivalents (nM BCE). Mean CVs were 2.6% for intra-assay and 6.9% for interassay variability.

Statistical analyses

The Statistical Analysis System (SAS) software package was used for data analysis (SAS Institute, Cary, NC, U.S.A.). Descriptive values are presented as median (range) unless stated otherwise. Distribution of values were tested using the Chi-square statistic. To test for group differences, Student's t-test for parametric and Wilcoxon's rank-sum test for nonparametric variables were performed. To correct for multiple testing, the p value was adjusted according to the Bonferroni correction (= significance level/number of tests). Linear regression analyses were performed to determine the strength of association between two parameters. Z scores are expressed as the SD of the mean for premenopausal females aged 30–45 years and healthy males according to: z = (x − mean)/SD. Receiver-operating characteristic (ROC) analysis was performed to evaluate the diagnostic validity of the assays. All statistical analyses were two-tailed, and a significance level of p < 0.05 was considered statistically significant.

RESULTS

Healthy adults

Normal values of urinary and serum markers of bone resorption were analyzed according to gender and, where applicable, to menopausal status. Results are shown in Table 1. All measurements were normally distributed. Males had higher levels of U-PYD (p < 0.05), and S-NTX (p < 0.01) than premenopausal females. U-CTX (p < 0.05) and S-NTX (p < 0.001) were significantly higher in postmenopausal than in premenopausal females.

Table Table 1.. Urine and Serum Markers of Bone Resorption in Healthy Adults and in Patients with Nonskeletal Diseases (Median [Range])

In healthy adults, simple Pearson correlation coefficients between urine marker measurements ranged between r = 0.39 (U-PYD vs. U-NTX, p < 0.01), and r = 0.92 (U-PYD vs. U-DPD, p < 0.001). Simple Pearson correlations between serum markers were weaker, with the highest correlation coefficient for S-BSP versus S-NTX (r = 0.48, p < 0.001). Correlation coefficients were moderate between urine and serum markers in the healthy subgroup. Highest values were found for U-CTX versus S-NTX (r = 0.67, p < 0.001). Correlations between comparable markers were r = 0.26 (p < 0.05) for U-CTX versus S-CTX, and r = 0.50 (p < 0.01) for U-NTX versus S-NTX.

Nonskeletal disease

Compared with healthy adults, patients with HF had significantly (p < 0.01) elevated levels of all urinary and serum markers, except U-CTX (Table 1). No correlations were found between parameters of liver function (γ-glutamyltransferase, glutamic-oxalacetic + glutamic-pyruvic transaminase, albumin) and markers of bone turnover.

Compared with healthy controls, RF was associated with significantly higher mean levels of all urine and serum markers of bone resorption (Table 1). All serum markers correlated inversively with the endogeneous creatinine clearance, and values ranged between r = −0.49 (p < 0.001) for S-BSP, and r = −0.63 (p < 0.001) for S-NTX. In contrast, no such relation was observed for the urinary indices. To elucidate further the effect of RF on serum markers, the data were stratified into three subgroups according to the endogeneous creatinine clearance. All patients with a creatinine clearance of <20 ml/minute (n = 5) had elevated levels of S-BSP (median 38.9 ng/ml), of S-CTX (median 9.6 nM/l), and of S-NTX (median 93.2 nM BCE). In patients with a creatinine clearance of 20–39 ml/minute (n = 14), S-BSP was elevated in 86% (median 31.0 ng/ml), S-CTX in 90% (median 8.6 nM/l), and S-NTX in 93% (median 53.6 nM BCE). A creatinine clearance between 40 ml/minute and 50 ml/minute (n = 11) led to elevated BSP values in 36% (median 17.8 ng/ml), elevated S-CTX values in 56% (median 6.9 nM/l), and elevated S-NTX values in 73% of the patients (median 21.4 nM BCE).

In female patients with BC−, mean levels of U-Pyr (p < 0.01), U-Dpyr (p < 0.05), S-CTX (p < 0.05), and S-NTX (p < 0.05) were slightly but significantly higher than the respective normal values. Otherwise, no significant group differences were found.

Metabolic bone disease

When compared with healthy controls, urinary levels of U-PYD (p < 0.001), and of U-DPD (p < 0.001) were elevated in 19–27% of patients with OPO (Table 2 and Fig. 1). Similarly, S-NTX levels were elevated in 31% (p < 0.01), and S-BSP levels in 27% (p < 0.001) of osteoporotic patients (Table 2 and Fig. 1). For U-CTX, U-NTX, and S-CTX, no statistically significant changes were found. The differentiation between healthy and primarily osteoporotic subjects, as determined by ROC analysis, differed between the individual biochemical markers (Fig. 2). Thus, for S-BSP and S-NTX, the areas under the curves (AUCs) (i.e., the mean sensitivity across the range of possible specificities) were 0.849 and 0.777, respectively. For all of the urinary indices of bone resorption, lower values were calculated, ranging between 0.481 (U-CTX and U-NTX) and 0.742 (U-DPD).

Figure FIG. 1.. Urine and serum markers of bone resorption in metabolic and malignant bone disease. Values are expressed as Z scores (see Materials and Methods). The full lines represent the mean, and the dotted lines represent ±2 SD around the mean of healthy controls. *p < 0.05, **p < 0.01, ***p < 0.001 versus healthy controls. U-DPD, urinary total deoxypyridinoline; U-CTX, urinary total C-terminal cross-linked telopeptide of type I collagen; U-NTX, urinary total N-terminal cross-linked telopeptide of type I collagen; S-BSP, serum bone sialoprotein; S-CTX, serum C-terminal total cross-linked telopeptide of type I collagen; S-NTX, serum N-terminal total cross-linked telopeptide of type I collagen.

Figure FIG. 2.. Diagnostic validity of biochemical markers of bone resorption in patients with metabolic bone disease. Results of ROC analyses for urine (left panels) and serum (right panels) markers of bone resorption are given as the respective (area under the curve [AUC]) (i.e., the mean sensitivity across the range of possible specificities). The differentiation between healthy subjects and (i) OPO (upper panels), (ii) PHPT (mid-panels), and (iii) PD (lower panels) is shown for each individual biochemical marker. For abbreviations see legend of.

In patients with PHPT (Table 2 and Fig. 1), mean levels of all serum markers were significantly higher than in healthy males and premenopausal females. In contrast, urinary markers were only slightly increased in PHPT, and this difference was statistically significance for U-PYD and U-DPD only (p < 0.05, respectively). Results of ROC analyses were similar for all serum markers. Thus, the (AUCs) ranged from 0.852 (S-BSP) to 0.898 (S-NTX). ROC analyses of the urine markers revealed somewhat less pronounced results, with (AUCs) ranging from 0.748 (U-CTX) to 0.851 (U-DPD) (Fig. 2).

Patients with PD had elevated levels of all urine and serum markers (Table 2, Fig. 1). ROC analyses of healthy subjects and patients with PD revealed the highest diagnostic validity for U-NTX (AUC 0.994) and U-DPD (AUC 0.940) as urinary indices, and for S-NTX (AUC 0.991) and S-BSP (AUC 0.905) as serum markers of bone resorption (Fig. 2).

In the group of patients with metabolic bone disease, overall correlations between urine and serum markers of bone resorption ranged between r = 0.37 (U-CTX vs. S-BSP, p < 0.05) and r = 0.80 (U-NTX vs. S-NTX, p < 0.001). The correlation coefficient of U-CTX versus S-CTX was r = 0.63 (p < 0.01). In PHPT, no significant correlations were found between biochemical markers of bone resorption and S-TAP, or serum calcium. In contrast, all biochemical markers of bone metabolism showed significant correlations with intact PTH, ranging from r = 0.36 (U-CTX vs. S-PTH, p < 0.05), to r = 0.74 (S-BSP vs. S-PTH, p < 0.001). In PD, only U-PYD (r = 0.66), and U-DPD (r = 0.64) were significantly associated with S-TAP (p < 0.001).

Malignant bone disease

Compared with healthy subjects, mean levels of almost all serum and urinary markers were significantly higher in patients with malignant bone disease (Table 3 and Fig. 1). In patients with breast cancer, skeletal involvement (BC+) was reflected by significantly higher mean levels of all bone markers compared with patients without bone metastases (BC−) (p < 0.01), except for U-CTX and S-CTX (Fig. 1). In patients with MM, serum markers were elevated in over 60% of cases, whereas the urine markers were elevated in 33% (U-NTX and U-CTX) to 53% (U-PYD) (Table 3). No significant correlations were found between any of the metabolic bone markers and serum calcium levels, serum creatinine, or β₂-microglobulin.

A subgroup of 28 patients suffering form HOM was treated with intravenous pamidronate as described in Materials and Methods. Bisphosphonate-induced changes in urine and serum levels of biochemical markers of bone turnover were most pronounced for U-CTX, S-CTX, and S-NTX. After 24 h, urine levels of U-CTX, and serum levels of S-NTX and S-CTX were significantly reduced compared with baseline levels (−34.6%, −26.7%, and −43.9%, p > 0.05, respectively). All urine markers reached the nadir 4 days after pamidronate application, and showed a slight increase on day 7 (Fig. 3). Similar changes were found for the serum markers. However, an increase on day 7 was only found for S-BSP and S-NTX. S-CTX reached low values on day 2 (−66.0%, p < 0.001), but the nadir was found on day 7 (−66.4%, p < 0.01, vs. baseline levels).

Figure FIG. 3.. Percent change of urine (upper panel) and serum (lower panel) markers of bone resorption after intravenous treatment with 30–60 mg pamidronate at day 0 in patients with HOM. Results are presented as mean ± SEM. *p < 0.05 versus day 0, **p < 0.01 versus day 0, ***p < 0.001 versus day 0. For abbreviations see legend of.

DISCUSSION

It has been shown that the quantitation of collagen degradation products in urine provides a valid estimate of bone resorption.^(1-4) However, since urinary measurements imply both greater analytical and greater biological variability,⁽⁵⁾ attempts have been made to develop assays for the detection of bone resorption markers in serum.^(6-10) In the present study, we have found that serum measurements of collageneous and noncollageneous degradation products of the organic bone matrix provide similar, if not improved, information on bone resorption than the established urinary markers.

In primary osteoporosis, it has been shown that a single measurement of bone turnover is insufficient to diagnose the disease or to discriminate between healthy and diseased individuals.^{(18, 19)} However, several recent studies have demonstrated that patients with untreated osteoporosis often show accelerated bone turnover^(20-22) and that markers of bone resorption predict future bone loss⁽²³⁾ and fracture risk.⁽²⁴⁾ In the present study, serum levels of immunoreactive BSP and of NTX were significantly higher in patients with OPO than in healthy controls. Similar changes were found for the pyridinium cross-links of collagen in urine. As shown by ROC analyses, the disease-related changes in urinary markers were less pronounced than the ones seen in serum markers. However, for the diagnosis of osteoporosis, neither type of marker has sufficient diagnostic power as a single measurement. Biochemical markers of bone turnover therefore should always be combined with clinical and osteodensitometric measurements.

Conventional markers of bone turnover, such as S-TAP or urinary hydroxyproline, are usually normal in patients with asymptomatic PHPT. In contrast, it has been shown that the urinary excretion of hydroxypyridinium cross-links is increased even in the state of asymptomatic or mild disease.⁽²⁵⁾ These results are confirmed by the present study. Moreover, it could be demonstrated by ROC analyses that the group differences in serum levels of BSP, CTX, and NTX are clearly more pronounced than the corresponding alterations in urinary markers.

Although PD affects both bone formation and bone resorption, the gold standard in detecting disease activity is the measurement of S-TAP as an osteoblastic marker⁽²⁶⁾ which was elevated in all 18 patients with PD in the present study (= 100%). In contrast, markers of collagen breakdown were elevated in 94% at best (U-NTX). Serum NTX levels were elevated in 88% of patients, suggesting that it provides similar information than the corresponding urine marker. In summary, none of the applied urine or serum markers appears to be superior over S-TAP measurements in the diagnosis of PD.

In patients with breast cancer, overt bone metastases were associated with significantly elevated levels of all urine and serum markers, except S-CTX. Changes related to malignant bone involvement were most pronounced for the urinary hydroxypyridinium cross-links, and for the U-NTX and S-NTX. The reason for the more or less unchanged levels of U-CTX and S-CTX remains unclear. The development of metastatic bone disease is a very complex process, and it is conceivable that bone resorption markers of different origin represent different stages of skeletal involvement in metastatic tumors. However, the similar changes in urinary and the corresponding serum indices in the differentiation of skeletal involvement in patients with breast cancer again favor the idea that serum markers provides similar information of bone resorption.

We have recently shown that the measurement of urinary hydroxypyridinium cross-links is a reliable method to evaluate the degree of bone resorption in patients with untreated MM.⁽²⁷⁾ Moreover, serum BSP levels have been shown to be elevated in normocalcemic patients with newly diagnosed MM in a similar fashion than urinary cross-links.⁽¹¹⁾ From the results of the present study with a higher percentage of elevated values for all serum markers, it seems likely that the detection of collagen degradation products or BSP in serum would further improve the ability to identify patients with increased osteoclast activity. However, no further analyses concerning disease stages in MM have been performed due to the relatively small number of patients.

Bisphosphonates act by decreasing both bone resorption and bone formation in which the effect on bone formation occurs somewhat later in the course of the drug's action.⁽²⁸⁾ In the present study, the intravenous application of 30–60 mg of pamidronate resulted in a rapid and significant reduction of the urinary collagen breakdown products and of all circulating levels of bone turnover markers in serum. These results are concordant with previous studies showing that the administration of bisphosphonates induces a rapid decline in bone resorption markers.^{(11, 29 30)} The parallel changes of urine and serum parameters due to bisphosphonate treatment are a strong hint for a similar clinical usefulness of these markers in monitoring therapeutic intervention in metastatic bone disease.

Seen from an analytical perspective, measurements of circulating degradation products of the collageneous and noncollageneous bone matrix show equal, if not improved, performance than the respective measurements in urine. Thus, for all serum markers of bone resorption, intra- and interassay CVs were <10%, a technical performance which is comparable to that of the urinary markers. However, since serum measurements need no further correction by an additional assay, we expect the overall variability to be significantly lower than that of the urinary indices.⁽⁵⁾ Further studies are presently under way to define precisely the biological and overall variability of the novel serum markers of bone resorption.

A number of caveats have to be considered when using serum markers. Thus, in patients with RF, the severity of disease was associated with an increase in all serum marker levels. Moreover, levels of all serum indices of bone turnover were inversively correlated with renal function, as assessed by the endogenous creatinine clearance. In contrast, no such correlations were seen for any of the urinary markers. Very much like the results reported for osteocalcin,⁽³¹⁾ our present observations indicate that circulating degradation products of the collageneous and noncollageneous bone matrix are affected by kidney function. Robins et al. have previously shown that the urinary excretion of the pyridinium cross-links is unaffected by RF down to a creatinine clearance of ∼20 ml/minute.^{(18, 32)} Therefore, both serum and urinary markers of bone metabolism should be omitted in patients with renal insufficiency and a creatinine clearance of <20 ml/minute. In patients with a creatinine clearance between 20 ml/minute and 50 ml/min, measurements of serum markers of bone resorption need to be interpreted with caution. In these patients, serum measurements should be replaced by urinary indices.

Like renal insufficiency, HF was also associated with elevated mean levels of all applied parameters of bone turnover, except for U-CTX. The effect of liver disease on the excretion of urinary indices of collagen degradation has been reported earlier.⁽³³⁾ However, no direct correlation between parameters of disease stage and bone turnover was found in the present study. At least in the absence of liver disease, bone resorption markers seem to provide reliable and valid information for the detection of metabolic and metastatic bone disease.

In conclusion, serum levels of collagen degradation products and of BSP reflect bone resorption comparable or superior to the conventional urinary markers. Since serum measurements circumvent some of the limitations of the urinary indices, their use is likely to improve the clinical assessment of skeletal disorders.

Acknowledgements

We are indebted to Mrs. Beatrice Auler, Mrs. Heike Zimmermann, and Dr. Christian Kissling for excellent technical assistance, to Boehringer Mannheim GmbH for providing the U-CTX kits, and to Ostex International for providing S-NTX.