Background: The pyridinium cross-links pyridinoline (PYD) and deoxypyridinoline (DPD) are established markers of bone resorption. We evaluated the analytical and clinical performance of a commercially available PYD HPLC assay and established reference intervals in children and adults.
Methods: We used a commercially available reagent set (Chromsystems Instruments & Chemicals) to measure PYD and DPD in 319 healthy controls (156 premenopausal women, 80 healthy men, and 83 healthy children age 1 month to 14 years) and 397 patients with metabolic bone diseases (postmenopausal osteoporosis, n = 175; male osteoporosis, n = 176; hyperparathyroidism, n = 17; hyperthyroidism, n = 19; Paget disease, n = 10).
Results: The mean intraassay and interassay CVs were <6% and <8% for both PYD and DPD, respectively. The reference interval was constant for premenopausal women in the age group 20–49 years. In men, cross-link values peaked at 20–29 years and decreased thereafter. Women with postmenopausal osteoporosis had significantly higher PYD (51%) and DPD (58%) values compared to premenopausal women. Similar results were found in osteoporotic men. In children the highest values were found in the first weeks and months after birth, followed by a decrease of 50%–60% at age 11–14 years. In metabolic bone diseases cross-link concentrations were significantly increased. The DPD:PYD ratio (mean value approximately 0.2) was remarkably constant in all populations evaluated.
Conclusions: The automated HPLC assay is a precise and convenient method for PYD and DPD measurement. We established reference intervals for adult women and men and for children up to 14 years old. The cross-link concentrations we determined by use of this HPLC method confirm its clinical value in enabling identification of increased bone resorption in patients with metabolic bone diseases.
Considerable progress has been made in the isolation and characterization of cellular and extracellular components of the skeletal matrix. This information has facilitated the identification of biochemical markers that specifically reflect either bone formation or bone resorption (1)(2), greatly increasing the number and types of specific analytes used in the assessment of skeletal pathologies. Biochemical markers of bone turnover are being investigated for their clinical utility in identifying osteoporotic fracture risk and determining and monitoring effective treatment for established disease. Also under investigation is the use of these markers as diagnostic tools for bone diseases other than osteoporosis, such as Paget disease and genetic disorders of collagen metabolism [e.g., the kyphoscoliotic form of Ehlers-Danlos syndrome (EDS)1 (OMIM 225400) and its allelic entity Nevo syndrome (OMIM 601451)] (3)(4)(5)(6).
Most metabolic bone diseases are characterized by increased bone resorption; therefore biochemical markers for bone resorption are of special interest. The pyridinium cross-links pyridinoline (PYD) and deoxypyridinoline (DPD) are well-characterized markers for bone resorption and can be quantified by HPLC (7)(8)(9)(10)(11)(12). A problem in the interpretation of HPLC results, however, is the lack of a functional internal standard and of standardization of HPLC procedures (11)(13)(14)(15). Results from different laboratories are thus not necessarily comparable, and reference intervals must be established for each method, particularly because commercial reagent sets have become available. Manual HPLC assays are considered the reference methods, but they are cumbersome and labor intensive. Several modifications of the original HPLC assay have been reported to feature simplified sample cleanup procedures that use automated extraction units (9)(16). We sought to evaluate on an automated system the analytical and clinical performance of a commercially available PYD HPLC reagent set and to establish reference intervals for different populations (children, women, and men). In addition, we used the assay to evaluate DPD:PYD ratios in healthy children and adults and in patients with various metabolic bone diseases.
Patients and Methods
The study population included 319 healthy controls (156 premenopausal women, 80 healthy men, and 83 healthy children age 1 month to 14 years) and 397 patients with metabolic bone diseases (postmenopausal osteoporosis, n = 175; male osteoporosis, n = 176; hyperparathyroidism, n = 17; hyperthyroidism, n = 19; Paget disease, n = 10). The controls were healthy and did not take drugs affecting bone metabolism. The diagnosis of osteoporosis was documented by a t-score ≤−2.5 in the lumbar spine or proximal femur, or by the presence of a low-trauma fracture. Patients with secondary causes of osteoporosis were excluded. Informed consent was obtained from all study participants, in accordance with the rules of the local ethics committee.
To avoid the influences of circadian rhythm and food intake, we collected fasting 2-h morning urine samples (after the first void was discarded) from all adult study participants. In children, urine collection was performed at random times during the day; in small children a urine collection bag was used.
We measured PYD and DPD by HPLC using a commercially available assay from Chromsystems Instruments & Chemicals (Munich, Germany). Results are expressed in relation to urinary creatinine. Briefly, the urine samples were hydrolyzed with 6 mol/L hydrochloric acid (final concentration) for 16 h at 100 °C on a heating block. A partition chromatography step (CF1 cellulose) was performed, then the PYD cross-links were separated by reversed-phase ion-paired HPLC. Eluted fluorescence peaks were quantified by use of PYDs as external standard. An internal standard resistant to hydrolysis was used to correct for between-sample recovery variations. Additional external standards were added in the middle and at the end of the run to control the stability of the analysis.
The chromatographic system comprised an isocratic soft-start pump model 1350 (Bio-Rad) and a fluorescence detector (Jasco FP-1520). All extraction steps were performed automatically by use of an automated solid-phase extraction system (ASPEC XL apparatus; Gilson Medical) controlled by Gilson 721 sampler software. Column effluents were monitored by fluorescence detection with an excitation wavelength of 295 nm and an emission wavelength of 400 nm.
Intraassay imprecision of the cross-link measurements was determined by measuring 15 replicates of 5 urine controls (2 controls provided by the manufacturer and 3 in-house controls) in a single run. The interassay imprecision was assessed on 5 control urine samples in 30 runs performed on different days for a period of 6 months. Dilution linearity was assessed by serially diluting 4 samples with water (proportions of 1:2, 1:4, 1:8, 1:16, and 1:32). The results were plotted as least-square linear regression of the expected concentration vs observed concentration. The undiluted sample values, as determined in the assay, were used to establish the expected values for subsequent dilutions. Recovery was tested by adding 4 different amounts of cross-link calibrator to clinical urine samples. The percentage recovery was calculated as the observed value divided by the expected value.
To avoid multiple thawing, all urine samples were separated into aliquots with no added preservatives. Sample aliquots were stored at −20 °C. All data obtained were corrected according to the urine creatinine concentration measured by use of a standard colorimetric method on an automatic analyzer (Hitachi 911, Roche Diagnostics).
bone mineral density
Bone mineral density values at the lumbar spine and hip were measured by dual x-ray absorptiometry performed with a Lunar Expert densitometer (17). Lumbar vertebrae with prevalent or incident fractures at L1 to L4 were not included in the bone mineral density measurements.
data handling and statistics
All chromatographic data were collected by use of Borwin version 1.21.6 integration software (IBMS Developments). All data were expressed as mean (SD). To compare study variables in different age groups, one-way ANOVA was used. The reference intervals with 95% CIs were calculated by nonparametric methods because cross-link concentrations were not normally distributed. Differences between various groups were tested with the Mann–Whitney U-test for independent samples. Significance was defined as P < 0.05. Data were analyzed by use of Statistica for Windows (version 7.0; StatSoft).
A typical chromatogram of PYD and DPD with internal standard is shown in Supplemental Fig. 1, which accompanies the online version of this article at http://www.clinchem.org/content/vol54/issue9.
The analytes PYD and DPD were stable for several days at room temperature with daylight exposure, and bacterial growth in the urine had no effect on the PYD and DPD concentrations (data not shown).
The intra- and interassay CVs are summarized in Table 1⇓ and Supplemental Table 1 in the online Data Supplement. The mean intraassay CVs were <6% for PYD and DPD, and the interassay CVs were <8% for both cross-links. Dilution linearity was found across the range of 40–2400 μmol/L for PYD (r = 0.99) and 10–400 μmol/L for DPD (r = 0.99). The recovery was found to be 96.2% (5.3%) (range 86%–104%) for PYD and 93.8% (6.4%) (range 86%–106%) for DPD.
Anthropometric characteristics of the adult study participants are summarized in Table 2⇓ , and the cross-links and bone mineral density (BMD) results for the whole study population are listed in Table 3⇓ .
The mean (SD) spine and hip BMD values in healthy premenopausal women were 1.17 (0.13) g/cm2 and 0.98 (0.12) g/cm2, respectively. In agreement with the inclusion criteria, these BMD values correspond to normal t-scores of 0.14 (0.97) and 0.40 (0.95) for the spine and hip, respectively. In premenopausal women in this study no significant differences of BMD at the spine or hip were detected in the women in the age groups from 20–29 years through 40–49 years. The mean (SD) PYD for premenopausal women was 44.6 (11.9) nmol/mmol creatinine, and the nonparametric 95% CIs were 42.8–46.5 nmol/mmol creatinine (Table 3⇑ and Fig. 1⇓ ). The mean (SD) DPD was 9.3 (2.4) nmol/mmol creatinine, with 95% CIs of 8.9–9.7. The DPD:PYD ratio was 0.206 (0.017). There were no significant differences between the various age-groups.
postmenopausal osteoporotic women
BMD was significantly decreased in osteoporotic postmenopausal women, with no significant differences between those with or without fracture (Table 3⇑ and Fig. 1⇑ ). Creatinine-corrected PYD and DPD mean values were significantly increased compared to those in premenopausal women, PYD by 51% and DPD by 58%; the DPD:PYD ratio was approximately 0.2 for both groups.
The mean (SD) spine and hip BMD values in healthy men were 1.23 (0.14) g/cm2 and 1.09 (0.16) g/cm2, respectively, corresponding to t-scores of −0.04 (1.16) and 0.27 (1.35), respectively (Table 3⇑ ).
There was a significant decrease in BMD of the spine as well as for the hip with advancing age. In the age group of 50–60 years spinal BMDs were modestly increased, probably owing to an increase in degenerative changes with age. The mean (SD) values for PYD and DPD in healthy men were 45.2 (17.1) nmol/mmol creatinine and 9.5 (3.6) nmol/mmol creatinine, respectively (Table 3⇑ , Fig. 2⇓ ). The 95% CIs for PYD and DPD were 40.8–48.7 and 8.6–10.3, respectively. Both cross-links decreased significantly with age, with a nadir after the age of 40 years, a finding that has been reported previously (18). The DPD:PYD ratio was approximately 0.2.
BMDs at the spine and hip were significantly decreased in osteoporotic men, with no differences between men with or without fractures. At the same time resorption markers were significantly increased by a mean of 28% for both PYD and DPD (Table 3⇑ , Fig. 2⇑ ). The DPD:PYD ratio did not differ from that of the healthy men.
In the youngest age group (4 weeks up to 12 months), cross-link concentrations were very high, with a rapid and substantial decrease of 50%–60% occurring after the age of 1 year (Table 3⇑ , Fig. 3⇓ ). Thereafter cross-links appeared to reach a plateau until puberty. The DPD:PYD ratio was very similar to the ratio found in adults.
metabolic bone diseases
As we expected, cross-link concentrations were significantly increased in patients with diseases affecting bone turnover, such as primary hyperparathyroidism, Paget disease, or hyperthyroidism (Table 3⇑ , Fig. 1⇑ ). In patients with primary hyperparathyroidism the increase in cross-link concentrations was only modest, reflecting the mild activity of the disease in this group of patients. The mean (SD) hypercalcemia was 2.71 (0.18) mmol/L. Despite a substantial increase in cross-links the DPD:PYD ratio in these groups did not differ from those of sex- and age-matched control groups.
PYD and DPD are products of enzyme-mediated oxidation at specific hydroxylysine and lysine residues within the collagen molecule and are found only in mature collagens, not in immature or newly synthesized collagens (19)(20). These cross-links covalently link collagen molecules between 2 telopeptides and a triple-helical sequence at 2 intermolecular sites and therefore stabilize collagen fibrils. Of the 2 forms, PYD is the major cross-link in all connective tissues, whereas DPD is found in high amounts in mineralized tissue. In human bone, for example, the DPD:PYD ratio is about 1:4, whereas in human cartilage the ratio is 1:10. Significant amounts of DPD occur in bone and dentine, and very small amounts in the aorta and ligaments (21). DPD is therefore considered to be specific for bone collagen degradation.
There are 2 principal methods for the measurement of urine PYD cross-link concentrations: HPLC analyses with or without hydrolysis of the urine, and immunoassay detection. Until recently the interpretation of HPLC results has been a problem owing to the lack of an internal standard resistant to hydrolysis. Although the recovery of PYD and DPD from hydrolysis to identification is usually 80%–90% or higher, the addition of an internal standard is required to improve the precision of the overall method, particularly to correct for variation in recovery between samples (22). Internal standards reported to be successfully used include pyridoxine, isodesmosine (derived from elastin), and acetyl-PYD, a semisynthetic derivative (9)(10)(23)(24). Pyridoxine shows chromatographic behavior that differs from that of pyridinium cross-links. Whereas isodesmosine occurs endogenously in urine, a characteristic that affects its use in urine sample, acetylated PYD does not have this limitation (10)(11). However, acetylated PYD hydrolyzes quickly under acidic conditions and therefore can be added to samples only after hydrolysis (11). With use of the internal standard isodesmosine, mean recoveries of PYD and DPD were 97% and 93%, respectively, and interassay variation was decreased from 15.1% to 5.3% for PYD and from 20.8% to 4.6% for DPD (9)(22). We do not know the exact nature of the internal standard used in the Chromsystems kit (proprietary information). We obtained good analytical precision with the use of an internal standard; the CVs for the determination of PYD were 3.7% to 7.1% and for DPD were 3.4% to 7.8%. Dilution linearity and supplementation recovery were good and were comparable to those reported for other methods.
The HPLC assays are not yet standardized, and results from different laboratories are not necessarily comparable (11)(13)(14)(15). The HPLC assays use not only different techniques (e.g., manual or automated methods, isocratic or gradient elution) but also different calibrators from several sources. Laboratories using the same calibrator usually provide values with a correlation of r > 0.95, but values may differ in absolute terms. It is thus important to establish reference intervals for the different methods used to measure PYD and DPD, particularly because results of several prospective studies suggest that in menopausal and elderly women and also in men there is an association between osteoporotic fractures and indices of bone turnover, independent of BMD (24)(25)(26)(27)(28)(29).
In our evaluation the cross-links PYD and DPD were measured in a large population including children, adult women, and men to establish age- and sex-dependent reference intervals.
Biochemical marker concentrations in healthy premenopausal women have been recognized as useful baseline measurements for evaluating changes in bone turnover in patients with metabolic bone disease and for evaluating fracture risk. The reference intervals observed in our study for premenopausal women were constant for the age groups 20–49 years. In contrast, the highest values in men were observed in the age group 20–29 years, thereafter we observed a decrease in PYD and DPD in men, with a nadir in the age group 40–49 years. This finding is in agreement with previously published data (18). In women with osteoporosis we observed significantly higher PYD and DPD values, but these did not differ in women with fracture vs those without fractures. Similar results were found in men with osteoporosis.
In children the highest PYD and DPD concentrations were found in the first weeks and months after birth. Thereafter concentrations decreased, reaching a plateau at the age of 11–14 years, with some increase at puberty. Although we did not know the pubertal stage of the children in this age group, we expected a larger increase. As in previously reported studies, we did not detect a clearly discernible pubertal peak, although there is no doubt that bone turnover increases during puberty (30)(31)(32). During growth, urine creatinine is augmented owing to increases in lean mass and in glomerular filtration. This augmentation of urine creatinine falsely attenuates the age-related increase in bone marker excretion (33).
Another aspect of the present study was the evaluation of the DPD:PYD ratio as a diagnostic tool for disorders such as EDS. At birth, patients with the kyphoscoliotic type of EDS have severe muscular hypotonia, kyphoscoliosis (which is progressive), severe joint hypermobility and luxations, and marked skin hyperelasticity. Other findings include fragility of the skin with abnormal scarring and osteopenia and the frequent occurrence of microcornea and occasional rupture of the arteries and the eye globe. This autosomal recessive disorder is caused by a deficiency of the enzyme collagen lysyl hydroxylase (EC 184.108.40.206; procollagen-lysine,2-oxoglutarate 5-dioxygenase) [for review see (34)], which normally hydroxylates lysyl residues in -Xaa-Lys-Gly- to -Xaa-Hyl-Gly- sequences of the helical region of the collagen α-chains. In bone and other tissue collagens, 2 Hyl residues, 1 in each of 2 amino telopeptides or 2 carboxyl telopeptides, together with 1 Hyl or 1 Lys residue of the triple helix form the trivalent pyridinium cross-links hydroxylysyl pyridinoline (PYD) and lysyl pyridinoline (DPD), respectively. As a consequence, lysyl hydroxylase deficiency results in underhydroxylation of lysyl residues, an abnormal cross-link formation with consequent mechanical instability of the affected tissues, and an abnormal pattern of cross-links in urine, with a markedly increased DPD:PYD ratio 5.97 (0.99), range 4.3–8.1; n = 17) (5)(6). The diagnosis of this type of EDS can be confirmed by measurement of the activity of the enzyme in cultured skin fibroblasts and/or directly by mutation analysis of the procollagen-lysine 1, 2-oxoglutarate 5-dioxygenase 1 (PLOD1) gene, which encodes the lysyl hydroxylase.
Because of the importance of the DPD:PYD ratio in EDS, we sought to find out if other conditions can alter the DPD:PYD ratio. In all of the individuals in our study, both children and adults, this ratio was very constant. Even in diseases characterized by high bone turnover, such as hyperthyroidism, Paget disease, and hyperparathyroidism, the DPD:PYD ratio remained unchanged. Thus the altered ratio appears to be specific and diagnostic for the kyphoscoliotic type of EDS, the allelic condition called Nevo syndrome, and for a newly recognized disorder that we have termed the spondylo-cheiro dysplastic form of EDS (35).
In conclusion, our results demonstrate that the automated HPLC assay is a precise and convenient method for measuring PYD and DPD and enabled us to established reference intervals for adult women and men and for children up to 14 years old. The measurement of DPD and PYD cross-link concentrations has clinical value in enabling identification of increased bone resorption in metabolic bone diseases.
Grant/Funding Support: This study was supported by the Swiss National Science Foundation (NF grant 3200B0-109370/1 to B. Steinmann).
Financial Disclosures: None declared.
Acknowledgments: We thank Christine Plüss, Udo Redweik, and Angelika Schwarze for their technical assistance in PYD determination and sample collection.
1 Results are mean (SD).
2 Osteoporosis defined as a t-score ≤−2.5 at the spine or proximal femur or presence of low-trauma fracture.
1 Results are mean (SD).
2 t-Score for postmenopausal women or men aged >50 years, z-score for premenopausal women or men aged <50 years.
3 P < 0.01 compared to healthy premenopausal women.
4 P < 0.01 compared to age group 20–29 years.
5 P < 0.001 compared to healthy men.
↵1 Nonstandard abbreviations: EDS, Ehlers-Danlos syndrome; PYD, pyridinoline or hydroxylysyl pyridinoline; DPD, deoxypyridinoline or lysyl pyridinoline; BMD, bone mineral density.
- © 2008 The American Association for Clinical Chemistry