Evaluation of a Bead-based Enzyme Immunoassay for the Rapid Detection of Osteocalcin in Human Serum

Clinical Chemistry 46:2 252–257 (2000) Endocrinology and Metabolism Evaluation of a Bead-based Enzyme Immunoassay for the Rapid Detection of Osteocalcin in Human Serum Alexandra M. Crăciun,1 Cees Vermeer,1 Hans-Georg Eisenwiener,2 Norbert Drees,2 and Marjo H.J. Knapen1* Background: Circulating osteocalcin is a well-known marker for bone formation, but none of the commercial kits currently available can be used in automated systems. Here we present the first semiautomated assay for human serum osteocalcin. Methods: Polystyrene beads were coated with antibodies against the COOH terminus of osteocalcin and used in the COBAS® EIA System. Osteocalcin was detected with peroxidase-conjugated antibodies against the osteocalcin NH2 terminus. Results: The time required to analyze an unknown sample was 60 min, with a lower detection limit of 4.5 mg/L and a linear dose–response curve between 4.5 and 100 mg/L. The intraassay imprecision (CV) was 5– 8% (n 5 21); the interassay variation was 6 –9% (n 5 14). In samples from human volunteers and patients, data generated with the newly developed assay were comparable to those obtained with standard microtiter plate-based assays. Conclusions: The coated beads assay may be implemented on fully automated analyzers, which not only may further reduce imprecision but may also substantially increase the applicability of osteocalcin as a marker for bone metabolism in the routine clinical setting. posttranslational formation of its three g-carboxyglutamate (Gla) residues. Although its function on a molecular level has remained unclear to date, increased bone formation, including higher bone mass and improved bone strength, was observed in OC-deficient (knock-out) mice (5 ). These experiments have demonstrated that OC has an important role in the regulation of bone growth and in the correct deposition of the mineral matrix in bone. Because 20 –30% of the de novo synthesized OC is not accumulated in the bone tissue, but is secreted into the blood stream, circulating OC is widely used as a biomarker for bone formation (6 – 8 ). During episodes of vitamin K deficiency or during treatment with vitamin K antagonists, Gla-containing proteins are synthesized in an undercarboxylated (Gladeficient) form. Fully carboxylated and undercarboxylated OC may be quantified separately on the basis of their different affinities for hydroxyapatite (9 ). It has been reported by various groups that in the general population, a substantial fraction of circulating OC occurs in its undercarboxylated form (9 –11 ), and that the concentration of undercarboxylated OC may have an independent diagnostic value for the assessment of bone mass and bone fracture risk (12–14 ). During recent years, an increasing number of test kits for serum OC have become commercially available, all of which are either radioimmunoassays or enzyme-based immunoassays for microtiter plates. Some of the drawbacks of such tests are that they are laborious and prone to errors by the persons performing the tests, and that they generally require long incubation steps (either all day or overnight) before the data become available. In addition, the various OC kits © 2000 American Association for Clinical Chemistry Osteocalcin (OC),3 also known as bone Gla protein, is the most abundant noncollagenous protein in mature bone (1, 2 ). Osteoblasts are the exclusive site of OC biosynthesis, which is regulated at the transcription stage by vitamin D (3, 4 ), whereas vitamin K is required for the 1 Department of Biochemistry and Cardiovascular Research Institute, Maastricht University, P.O. Box 616, 6200 MD Maastricht, The Netherlands. 2 Hoffmann-La Roche Diagnostics, Basel 4070, Switzerland. *Address corresponding to this author at: Department of Biochemistry, University of Maastricht, P.O. Box 616, 6200 MD Maastricht, The Netherlands. Fax 31-43-367-0992; e-mail m.knapen@bioch.unimaas.nl. Received September 29, 1999; accepted November 30, 1999. 3 Nonstandard abbreviations: OC, osteocalcin; Gla, g-carboxyglutamate; BAP, bone-specific alkaline phosphatase; mAb, monoclonal antibody; PTH, parathyroid hormone; OHPro, hydroxyproline; DPD, deoxypyridinoline; and CTX, type I collagen C-terminal telopeptide. 252 253 Clinical Chemistry 46, No. 2, 2000 may give widely different values when the same serum sample is tested (6, 15 ). This may be related to different specificities of the antibodies used for fully carboxylated and undercarboxylated OC and whether OC degradation products are recognized. To improve the accuracy and reproducibility of OC quantification, we used commercial antibodies that were coated onto polystyrene beads and used in the Roche COBAS® EIA System. We compared the performance of this semiautomated enzyme immunoassay with commercial radiometric and microtiter plate-based assays for various markers of bone metabolism. Materials and Methods subjects The reference values for the concentrations of OC and bone-specific alkaline phosphatase (BAP) were established in serum samples obtained from 151 apparently healthy subjects (60 men and 91 women) between 19 and 86 years of age who were recruited via a local newspaper. The within-day and day-to-day variations of OC and several other markers were examined in 12 apparently healthy volunteers (6 men and 6 women) between 19 and 27 years of age recruited from the students of the Maastricht University. During the first 24 h of the experiment, urine and blood samples were collected at 0900 (start) and at 1100, 1400, 1700, 2100, 2300, 0300, and 0700. Subsequent samples were collected at 0900 on days 2, 8, 15, 22, 29, and 57. All samples taken at 0900 were obtained after an overnight fast (only water allowed after 1900 of the preceding night). Changes in bone markers during the follow-up of treatment were recorded in serum and urine samples from 30 osteoporotic women (.65 years of age), before and after treatment with either bisphosphonate (Alendronate, 5–10 mg/day for 15 months; Merck), estrogen (Livial, 2.5 mg/day for 6 months; Organon), or calcitonin (Miacalcic nasal application, 100 IU/day for 12 months; Sandoz). All studies were approved by the University Hospital Medical Ethics Committee. sample collection and storage Blood (10 mL) was taken by venipuncture to prepare serum, 0.5-mL aliquots of which were frozen at 280 °C within 2 h after sample collection until use. All urine samples collected at the 0900 time points were obtained after an overnight fast (14 h) and represent the 2-h second morning void. Other urine samples collected during the first day were nonfasting samples. All urine samples were stored in 0.5-mL aliquots at 230 °C until use. procedure for the semiautomated oc immunoassay The assay (hereafter called the Coated Beads Assay) is based on commercial monoclonal antibodies (mAbs) against synthetic peptides homologous to OC residues 1–19 and 20 – 43 (Osteometer), and will be designated here as mAb1–19 and mAb20 – 43. Polystyrene beads (2 mm Fig. 1. Calibration curve for OC by Coated Beads Assay. Calibrators used were from the commercial N-mid Osteocalcin assay (Osteometer). Bars, SD. diameter) were coated with mAb20 – 43 by Osteometer, other components for the test were those in the microtiterbased N-midTM Osteocalcin ELISA (Osteometer), and are described by Rosenquist et al. (16 ). The test procedure, for which we used the COBAS EIA System from Hoffmann-La Roche (Basel, Switzerland), was as follows: 25 mL of sample (calibrator, control, or serum) was pipetted into polycarbonate tubes, together with 225 mL of buffer A (0.14 mol/L NaCl, 0.01 mol/L sodium phosphate, pH 7.4) and 25 mL of peroxidase-conjugated mAb1–19. Subsequently, one coated bead was added to each tube and incubated for 30 min at 37 °C while shaking in the COBAS EIA Incubator. The beads were then washed with distilled water in the COBAS EIA Washer, and 250 mL of substrate solution (tetramethylbenzidine) was added, after which the tubes were incubated for another 15 min at 37 °C with constant shaking. The reaction was stopped by the addition of 1 mL of 0.1 mol/L H2SO4; within 1 h after Table 1. Recovery of OC in pathological serum samples after serial dilution.a Recovery of OC after dilution, % Dilution factor Sample A Sample Bc Sample Cd 1.00 1.25 1.67 2.00 2.50 5.00 100.0 6 1.2 102.7 6 1.8 105.0 6 4.7 99.7 6 4.7 96.6 6 2.4 80.0 6 5.1 100.0 6 2.3 91.9 6 3.4 88.4 6 3.9 90.0 6 8.9 84.0 6 6.2 77.5 6 3.5 100.0 6 2.9 93.6 6 4.2 88.7 6 7.4 85.1 6 5.4 79.9 6 4.5 69.5 6 6.0 b a The samples were diluted with calibrator A, and the observed OC concentrations were multiplied by the dilution factor. All samples were tested in triplicate, and the experiment was repeated on 3 different days. Mean values are given 6 SD. b Undiluted serum: OC concentration, 117.4 mg/L. c Prediluted twofold: OC concentration, 114.3 mg/L. d Prediluted fivefold: OC concentration, 75.3 mg/L. 254 Crăciun et al.: Coated Beads Assay for Human Serum Osteocalcin Table 2. Bone formation markers in different age groups of men and women.a OC, mg/L Age range, years Men 19–40 59–86 Women 19–40 41–58 59–86 Mean age, years n Beads IRMA BAP, mg/L 27.5 6 6.3 74.4 6 6.5 30 30 21.7 6 10.1 16.8 6 11.9 25.2 6 10.4 16.6 6 9.6b 16.7 6 5.3 13.9 6 8.1 30.5 6 6.1 51.1 6 4.9 75.1 6 6.3 31 28 32 18.8 6 6.0 18.4 6 6.6 22.0 6 10.6 16.6 6 6.1 16.5 6 6.8 17.5 6 9.4 11.5 6 4.7 9.5 6 5.0 15.1 6 6.4c a All samples were tested in triplicate, and mean values are given 6 SD. Statistics were performed with the unpaired Student t-test. Significantly different from men in the 19 – 40 age group. c Significantly different from women in the 41–58 age group. b termination of the reaction, the absorbance at 450 nm was recorded with a 25-channel COBAS EIA Photometer. Serum OC concentrations were calculated using a fourparameter logistic curve fit based on the calibrators of the assay (0, 6.25, 12.5, 25, 50, and 100 mg/L). other tests used The data obtained with the experimental OC assay were compared with commercial kits for OC (ELSA-Osteo; CIS Bio-international) and BAP (Tandem-R Ostase; Hybritech). Both tests are two-site IRMAs. Intact parathyroid hormone (PTH) was quantified in serum with the N-tact PTH radioimmunoassay from Incstar. Markers tested in urine were hydroxyproline (OHPro; hypronosticon; Organon Teknika), deoxypyridinoline (DPD; Pyrilinks-D; Metra Biosystems), and type I collagen C-terminal telopeptide (CTX; CrossLaps; Osteometer BioTech). Creatinine was assessed in urine by standard enzymatic techniques (Boehringer Mannheim) on a Beckman Synchron CX7-2 automated analyzer. Urinary calcium was determined by atomic absorption spectrophotometry (PerkinElmer). Urinary markers are expressed as the ratio between these markers and creatinine throughout this report. data analysis Statistical analysis was performed with the software package SPSSWin, Ver. 7.5 (SPSS). All results are given as the mean value 6 SD. Differences between the groups were investigated with the unpaired Student t-test. The Wilcoxon test was used for the evaluation of differences within groups. Differences were considered significant at P ,0.05. Correlations between the data obtained with different test procedures were evaluated using linear regression. Results calibration curve and test characteristics Calibration curves were constructed on 12 different days using six calibrator solutions with OC concentrations in the range of 0 –100 mg/L. Each calibrator was measured in triplicate, and the mean absorbance at 450 nm (6 SD) was expressed as a function of the OC concentration (Fig. 1). The lower limit to detection, defined as the mean absorbance 1 3 SD for calibrator A (0 mg/L), was 4.45 mg/L [absorbance 5 0.05 1 3 3 0.02 5 0.11]. The intra- and interassay variation of the test was determined using three serum pools of known OC concentrations. The intraassay variation was calculated by expressing the SD as a percentage of the mean concentration as calculated Table 3. Day-to-day variation of bone markers during 2 months. All (n 5 12) Test In serum OC, mg/L (coated beads) OC, mg/L (ELSA-Osteo) BAP, mg/L PTH, ng/L In urine DPD/creat,a mmol/mol OHPro/creat, g/mol CTX/creat, mg/mol Calcium/creat, mol/mol a creat, creatinine. Mean 6 SD 23.7 6 6.0 26.0 6 6.1 15.8 6 7.7 31.1 6 5.6 6.9 6 2.8 2.2 6 0.9 160.8 6 104.2 0.25 6 0.09 Men (n 5 6) Women (n 5 6) Day-to-day variation, % Mean 6 SD Day-to-day variation, % 10.8 9.1 11.7 29 23.1 6 6.5 25.0 6 6.4 11.9 6 3.2 28.9 6 6.1 8.3 9.6 13.3 27.5 24.1 42.3 52.4 36.3 5.2 6 2.2 1.7 6 0.5 109.3 6 56.6 0.2 6 0.06 20.7 44.5 58.7 37 Mean 6 SD 24.3 6 2.7 27.2 6 6.1 19.6 6 9.2 33.2 6 4.5 8.6 6 2.5 2.7 6 0.9 212.3 6 119.7 0.3 6 0.1 Day-to-day variation, % 15.5 8.4 9.9 30.5 27.6 40 46.1 35.5 Clinical Chemistry 46, No. 2, 2000 from 21 replicates of each serum pool, which was repeated on 3 different days. Mean values obtained were 7.6%, 6.0%, and 4.8% for serum pools containing 10, 22, and 83 mg/L of OC, respectively. The same serum pools were measured in duplicate on 14 consecutive days, and interassay variations of the means of duplicates were calculated by expressing the SDs as percentages of the means: 6.0%, 9.1%, and 5.9%, respectively. To determine the linearity on dilution, we used three human serum samples containing pathologically high OC concentrations (from patients with renal failure). Samples with OC concentrations well above the highest calibrator (calibrator F; 100 mg/L) were prediluted with calibrator A (0 mg/L) before use to give the following OC concentrations: sample A (undiluted), 97.4 6 2.2 mg/L; sample B (prediluted twofold), 94.3 6 2.6 mg/L; and sample C (prediluted fivefold), 75.3 6 0.7 mg/L. On 3 different days, each sample was diluted serially and tested in duplicate. Recoveries were calculated and expressed as percentages of the starting values (Table 1), and it was apparent that sample A could be diluted twofold without significant loss of recovery. As was the case with all (n 5 6) of the microtiter plate-based kits and radioimmunoassays we have checked to date (data not shown), the recovery declined at higher dilutions. Both prediluted samples (B and C) showed a strong decrease of recovery after further dilutions, which indicates that more than twofold dilution of serum samples may lead to substantial underestimation of the OC concentration in these samples if the dilutions are prepared with the calibrator A (0 mg/L), which does not have a matrix sufficiently similar to serum. To investigate whether the antibodies used in our assay discriminate between OCs containing different numbers of Gla residues, full-length synthetic OCs (17 ) containing either 0 or 3 Gla residues were dissolved in calibrator A and tested in various dilutions. Both synthetic peptides were recognized well, but the antibodies did not differentiate between fully carboxylated and noncarboxylated OC (data not shown). 255 day-to-day and within-day variations Twelve subjects (6 men and 6 women) were enrolled in an experiment in which blood and urine samples were collected at various time points during the first 24 h and at weekly intervals during the first month, with a final sample collection after 2 months. We measured serum concentrations of OC (by two assays), BAP, and PTH, and urine concentrations of DPD, CTX, OHPro, total Ca21, and creatinine. For each subject and each variable sepa- variations related to age and gender In this experiment, we assembled serum samples from apparently healthy men (n 5 60) and women (n 5 91) of different ages and compared the new assay with commercially available test kits for OC and for BAP. The results obtained with the two OC assays were very similar, with all three bone formation markers slightly increased in men 19 – 40 years of age, possibly because in this group the peak bone mass was not yet reached (Table 2). Only in the case of the commercial OC kit was this difference statistically significant, however. In addition, in women 59 – 86 years of age, the bone formation markers had a tendency to increase, which may be related to the increased bone turnover frequently seen during postmenopausal bone loss. Only in the case of BAP was this increase statistically significant, however. Fig. 2. Within-day variation of bone formation markers. (Top), individual plots of OC determined with the Coated Beads Assay during consecutive time points within 24 h; (middle), individual plots of OC determined with the Elsa-Osteo assay; (bottom), individual plots of BAP. All curves are presented as the concentration of each marker (mg/L). The mean values per time point are given by the thick line; bars, SD for each time point. 256 Crăciun et al.: Coated Beads Assay for Human Serum Osteocalcin Table 4. Bone markers in postmenopausal women before and after therapy.a,b OC, mg/L OC Treatment Biphosphonates Before After Hormone replacement Before After Calcitonin Before After CTX/creat, mg/mol 0.61 6 0.40 0.63 6 1.14 29.8 6 54.0 17.6 6 26.2 532 6 1698 173 6 332b 6.6 6 2.3 5.6 6 2.9b 0.49 6 0.33 0.31 6 0.29b 19.8 6 26.4 10.1 6 4.8 578 6 307 210 6 144b 14.1 6 2.7 11.9 6 1.7b 0.4 6 0.3 0.4 6 0.3 13.0 6 6.6 11.7 6 5.2 801 6 459 609 6 338b ELSA-Osteo BAP, mg/L 28.2 6 13.1 10.3 6 3.07a 29.8 6 12.5 8.9 6 3.91a 12.9 6 4.0 7.0 6 2.06a 23.6 6 7.7 17.6 6 4.2b 28.0 6 8.7 18.5 6 5.0a 21.1 6 8.0 16.0 6 5.1b 23.8 6 8.7 19.1 6 9.7b a,b c Ca/creat,c mol/mol DPD/creat, mol/mol Beads Changes within each treatment group were analyzed with the Wilcoxon test: creat, creatinine. rately, the individual mean values of the seven fasting morning samples (taken at 0900) were calculated and used to calculate the group mean values (Table 3, columns 2, 4, and 6). Individual day-to-day CVs were expressed as percentages of the corresponding individual mean values and were used to calculate the mean intraindividual variation in the group (Table 3, columns 3, 5, and 7). It turned out that the serum markers (except PTH) were relatively stable with time and that both the inter- and intraindividual variations of the Coated Bead Assay were comparable with commercial kits for bone markers. The urinary markers, notably OHPro and CTX, had large intra- and interindividual CVs. The within-day variation of the various markers was assessed using the samples obtained during the first 24 h of the experiment (Fig. 2). The individual variation of the bone formation markers OC (determined with both assays) and BAP are given in the different plots. No distinct diurnal variation was found for any of the three assays. A more pronounced diurnal pattern was observed for the bone resorption markers, but because of the large interindividual variation, the difference between zenith and nadir did not reach statistical significance (data not shown). follow-up during osteoporosis treatment To test the ability of the Coated Beads Assay to detect changes in serum OC concentration during therapy, three groups of 10 postmenopausal women were followed during their treatment with bisphosphonates, estrogen, or calcitonin (Table 4). Changes during therapy were more pronounced for OC than for BAP, with similar relative a P ,0.005; b P ,0.05. changes for the two OC assays. Therapy-induced changes were also large in the bone resorption markers, but because of the large standard deviation of urinary markers, the changes were not statistically significant in all cases. correlation between bone formation markers To investigate the correlation between the various tests for bone formation, the data from the former experiments were pooled. In total, the data of 181 subjects (60 men, 91 women, and the baseline measurements of 30 postmenopausal women) were used to perform linear regression. The correlation between both OC assays was r 5 0.879 (P ,0.0001), and the standard deviation of the residuals (Syux) was 4.66 mg/L. Regression analysis between BAP and both OC assays produced rather poor regression coefficients and higher Syux values (Table 5). Discussion OC is the most abundant noncollagenous protein in bone. It is synthesized by the osteoblasts, and after its cellular secretion, ;80% is bound to the hydroxyapatite matrix. The remainder is released into the blood stream, where it is available for detection (2 ). Circulating OC is used as a marker for bone formation, and high concentrations have been observed in children (2 ) and in patients with high bone turnover, as is found in Paget disease and postmenopausal osteoporosis (18, 19 ). Commercial test kits for OC detection are based on one of two principles: radioimmunoassays or enzyme-linked immunoassays. The use of OC Table 5. Linear regression analysisa of bone formation markers. Dependent variable OC (Elsa-Osteo) BAP BAP Independent variable Slope (SD) Intercept (SD) R P Syzx, mg/L OC (coated beads) OC (coated beads) OC (Elsa-Osteo) 0.89 (0.036) 0.32 (0.059) 0.22 (0.061) 1.86 (0.81) 7.21 (1.32) 9.41 (1.35) 0.879 0.378 0.259 ,0.0001 ,0.0001 ,0.0001 4.66 7.58 7.91 a Linear regression analysis was performed according to the equation: y 5 a0(SD) 1 a1(SD) 3 x, in which y is the dependent variable and x is the independent variable; a0 is the intercept, and a1 is the slope of the equation, both with corresponding SD. For each equation, the regression coefficient (R) and significance (P) are given. Syux is the standard deviation of the residuals. Clinical Chemistry 46, No. 2, 2000 as a marker for bone metabolism in routine clinical chemistry is mainly restricted because of (a) the large differences between the various kits and (b) the fact that no automated test is available to date. The former problem may be related to the fact that serum OC occurs in at least two conformations: fully carboxylated OC, which contains three residues of the unusual amino acid Gla, and undercarboxylated OC, which contains 0 –2 Gla residues (12, 20 ). The antibodies (either polyclonal or monoclonal) used in different kits differ from each other in their relative affinity for fully carboxylated and undercarboxylated OC (15 ), and thus in the amounts that are detected in the same serum sample. An additional problem is that some kits detect only intact OC, whereas others recognize both intact OC and OC fragments. Hence, standardization of reference samples and antibodies used is an absolute requirement before data obtained with various kits may be compared. From a technical point of view, a serious drawback of existing kits is that they are test tube- or microtiter plate-based assays that must be pipetted by hand. This is laborious and forms a potential source of imprecision and mistakes. Therefore, we have used and evaluated a semiautomated method for OC detection, using the antibodies and calibrators from a commercial enzyme-linked immunosorbent assay. The test is based on the sandwich principle, with antibodies directed against epitopes at the NH2 and COOH termini of OC outside the Gla domain, thus ensuring that only intact OC (both fully carboxylated and undercarboxylated) is detected. It turned out that the newly developed assay (provisionally designated as the Coated Beads Assay) has a high precision and compares well with the kit from CIS Biointernational in both withinday and day-to-day variation, as well as in patient follow-up studies. Like other test for bone formation, the standard deviation of the Coated Beads Assay was much lower than that of bone resorption markers, and a good correlation was observed with OC values determined with the CIS kit. In conclusion, we have demonstrated that with the semiautomated assay, OC may be determined within 1 h, and that the accuracy of the assay is at least comparable to existing kits. The new test still requires a few pipetting steps, but the coated beads may be used equally well in fully automated analyzers. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. This study was supported by Hoffmann-La Roche Diagnostics (Basel, Switzerland). We thank Dr. P. Proost (University of Leuven, Belgium) for kindly supplying us with synthetic OC. 19. References 1. Price PA. Role of vitamin K-dependent proteins in bone metabolism [Review]. Annu Rev Nutr 1988;8:565– 83. 2. Hauschka PV, Lian JB, Cole DEC, Gundberg CM. Osteocalcin and 20. 257 matrix Gla protein: vitamin K dependent proteins in bone [Review]. Physiol Rev 1989;69:990 –1047. Kesterson RA, Stanley L, DeMayo F, Finegold M, Pike JW. The human osteocalcin promotor directs bone-specific vitamin Dregulatable gene expression in transgenic mice. Mol Endocrinol 1993;7:462–7. Ozono K, Liao J, Kerner SA, Scott RA, Pike JW. The vitamin D-responsive element in the human osteocalcin gene. Association with a nuclear proto-oncogene enhancer. J Biol Chem 1990;265: 21881– 8. Luo G, Ducy P, McKee MD, Pinero GJ, Loyer E, Behringer RR, Karsenty G. Spontaneous calcification of arteries and cartilage in mice lacking matrix GLA protein. Nature 1997;386:78 – 81. Power MJ, Fottrell PF. Osteocalcin: diagnostic methods and clinical applications [Review]. Crit Rev Clin Lab Sci 1991;28:287–335. Garnero P, Grimaux M, Demiaux B, Preaudat C, Seguin P, Delmas PD. Measurement of serum osteocalcin with a human-specific twosite immunoradiometric assay. J Bone Miner Res 1992;7:1389 –98. Eastell R, Robins SP, Colwell T, Assiri AMA, Riggs BL, Russell RGG. Evaluation of bone turnover in type I osteoporosis using biochemical markers specific for both bone formation and bone resorption. Osteoporos Int 1993;3:255– 60. Knapen MHJ, Hamulyák K, Vermeer C. The effect of vitamin K supplementation on circulating osteocalcin (bone Gla-protein) and urinary calcium excretion. Ann Intern Med 1989;111:1001–5. Knapen MHJ, Jie K-SG, Hamulyák K, Vermeer C. Vitamin K-induced changes in markers for osteoblast activity and urinary calcium loss. Calcif Tissue Int 1993;53:81–5. Plantalech L, Guillaumont M, Vergnaud P, Leclercq M, Delmas PD. Impairment of gamma carboxylation of circulating osteocalcin (bone Gla protein) in elderly women. J Bone Miner Res 1991;6:1211– 6. Szulc P, Chapuy M-C, Meunier PJ, Delmas PD. Serum undercarboxylated osteocalcin is a marker of the risk of hip fracture in elderly women. J Clin Investig 1993;91:1769 –74. Szulc P, Arlot M, Chapuy M-C, Duboeuf F, Meunier PJ, Delmas PD. Serum undercarboxylated osteocalcin correlates with hip bone mineral density in elderly women. J Bone Miner Res 1994;9: 1591–5. Knapen MHJ, Nieuwenhuijzen Kruseman AC, Wouters RSME, Vermeer C. Correlation of serum osteocalcin with bone mineral density in women during the first 10 years after menopause. J Bone Miner Res 1998;63:375–9. Knapen MHJ, Eisenwiener H-G, Vermeer C. Osteocalcin detection in aging serum and whole blood: stability of different osteocalcin fractions. Clin Chim Acta 1996;256:151– 64. Rosenquist C, Qvist P, Bjarnason N, Christiansen C. Measurement of a more stable region of osteocalcin in serum by ELISA with two monoclonal antibodies. Clin Chem 1995;41:1439 – 45. Houben RJTJ, Jin D, Stafford DW, Proost P, Ebberink RHM, Vermeer C, Soute BAM. Osteocalcin binds tightly to the g-glutamyl carboxylase at a site distinct from that of the other known vitamin K-dependent proteins. Biochem J 1999;341:265–9. Gundberg CM, Lian JB, Gallop PM, Steinberg JJ. Urinary gammacarboxyglutamic acid and serum osteocalcin as bone markers: studies in osteoporosis and Paget’s disease. J Clin Endocrinol Metab 1983;57:1221–5. Kaddam IMS, Iqbal SJ, Holland S, Wong M, Manning D. Comparison of serum osteocalcin with total and bone specific alkaline phosphatase and urinary hydroxyproline:creatinine ratio in patients with Paget’s disease of bone. Ann Clin Biochem 1994;31: 327–30. Cairns JR, Price PA. Direct demonstration that the vitamin Kdependent bone Gla protein is incompletely g-carboxylated in humans. J Bone Miner Res 1994;9:1989 –97.