0021-972X/91/7202-0367$03.00/0 Journal of Clinical Endocrinology and Metabolism Copyright© 1991 by The Endocrine Society
Vol. 72, No. 2 Printed in U.S.A.
Effect of Menopause and Hormone Replacement Therapy on the Urinary Excretion of Pyridinium Cross-Links DANIEL UEBELHART, ANNETTE SCHLEMMER, JULIA S. JOHANSEN, EVELYNE GINEYTS, CLAUS CHRISTIANSEN, AND PIERRE D. DELMAS INSERM Unit 234 and Service de Rhumatologie et de Pathologie Osseuse, Hopital E. Herriot, Lyon, France; and the Department of Clinical Chemistry, University of Copenhagen, Glostrup Hospital (A.S., J.S.J., C.C.), Glostrup, Denmark
ABSTRACT. Pyridinoline (Pyr) and deoxypyridinoline (DPyr) are two cross-links of collagen molecules that are present in the extracellular matrix and released during its degradation. In contrast to the wide distribution of collagen, Pyr is present in bone and cartilage, but not in significant amounts in other connective tissues, and D-Pyr appears to be specific for bone tissue. Therefore, the urinary excretion of Pyr and D-Pyr might be a sensitive marker of bone matrix degradation. Using a specific high pressure liquid chromatography assay we have measured Pyr and D-Pyr cross-links in a 24-h and a fasting urine sample in 60 early postmenopausal women and 19 premenopausal women matched for age. Menopause induced a 62% increase in Fu Pyr (49.8 ± 18.7 vs. 30.8 ± 8.0 pmol//imol creatinine; P < 0.001) and an 82% increase in Fu D-Pyr (8.2 ± 3.4 vs. 4.5 ± 1.4 pmol/Vmol creatinine; P < 0.001). In 20 postmenopausal women on hormone replacement therapy, urinary Pyr and D-Pyr returned to premenopausal levels within 6 months, contrasting with unchanged levels during placebo treatment. The 24-h excretion of Pyr and D-Pyr was significantly lower than the fasting excretion, but was similarly decreased after hormone replacement therapy. Pyr and D-Pyr excretion measured in the same urinary sample were highly correlated (r = 0.85 for fasting and 0.83 for 24-h sampling), but correlations
between fasting and 24-h values were weak (D-Pyr, r = 0.30; Pyr, r= 0.29; P < 0.05 for both). Correlations between urinary cross-links and other markers of bone turnover (Fu hydroxyproline/creatinine and plasma osteocalcin) were significant but low (Pyr vs. osteocalcin, r = 0.29, P < 0.05; Pyr vs. hydroxyproline, r = 0/.34; P < 0.01; D-Pyr vs. osteocalcin, r = 0.39; P < 0.01), except for D-Pyr us. hydroxyproline (r = 0.24; P = 0,07), suggesting that these markers reflect different events of bone metabolism. Finally, a single measurement of the fasting excretion, but not of the 24-h excretion, of cross-links was significantly correlated (Pyr, r = 0.34; P < 0.05; D-Pyr, r = -0.46; P < 0.01), with the subsequent spontaneous rate of bone loss assessed by repeated measurements of the radial bone mineral content in 37 postmenopausal women. The correlation with the rate of bone loss was improved when the simultaneous measurement of plasma osteocalcin and urinary hydroxyproline/creatinine was combined (Pyr, r = 0.75; P < 0.001; D-Pyr, r = 0.77; P < 0.001). We conclude that the fasting excretion of Pyr and D-Pyr reflects bone turnover changes at the menopause and the effects of hormone replacement therapy. The combination of cross-links, hydroxyproline, and osteocalcin measurement seems promising to evaluate the rate of bone loss in postmenopausal women. (J Clin Endocrinol Metab 72: 367-373, 1991)
A
resorption, respectively, but their sensitivity and specificity are limited (3). Other markers of bone formation include the bone-specific alkaline phosphatase, serum osteocalcin, also called bone gla protein (4, 5), and the carboxy-terminal propeptide of type 1 procollagen. All of these markers are increased at the time of menopause, and we have recently shown that serum osteocalcin is significantly correlated with the spontaneous rate of bone loss in recently menopausal women (6). Similarly, serum osteocalcin was reported to be the best single predictor of the rate of bone loss in a large cohort of periand postmenopausal women followed prospectively during 4 yr (7). Because an increase in bone resorption is the primary event following estrogen deficiency, a sensitive marker of bone resorption would be valuable to analyze the effects of menopause on bone turnover. As stated above, urinary hydroxyproline is not specific for bone collagen
N EFFECTIVE prevention of osteoporosis can be achieved at the time of menopause with long term hormone replacement therapy. Because of the constraints, the cost, and the potential side-effects of such treatment, it is generally accepted that intervention should be targeted at women who present a high risk of osteoporosis. The two main factors responsible for that risk are represented by a low peak bone mass and a high rate of bone loss at the time of hormone replacement therapy deprivation (1, 2). The postmenopausal rate of bone loss is related to an overall increase in bone turnover, which can be assessed by several biochemical markers. Serum alkaline phosphatase and urinary hydroxyproline are commonly used to assess bone formation and Received April 15,1990. Address all correspondence and requests for reprints to: Dr. P. D. Delmas, Hopital E. Herriot, Pav. F, 69437 Lyon Cedex 03, France. 367
The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 15 July 2016. at 19:39 For personal use only. No other uses without permission. . All rights reserved.
UEBELHART ET AL.
368
resorption, as it is also derived from collagen synthesis during the breakdown of the procollagen N-terminal extension peptides and of neosynthesized collagen molecules (8, 9). Plasma tartrate-resistant acid phosphatase and urinary galactosyl-hydroxylypsyl glycoside, two recently described markers of bone resorption, are increased in osteoporotic women (10, 11) and warrant further investigation to establish their sensitivity and specificity. Recently, the urinary excretion of the pyridinium cross-links of collagen have been suggested to be a sensitive and specific marker of bone resorption (12). The extracellular matrix is stabilized by the formation of covalent cross-links between adjacent collagen chains, and it is thought that the reducible aldimine cross-links initially formed are converted to mature nonreducible compounds (13). Two mature cross-linking amino acids, pyridinoline (Pyr) and deoxypyridinoline (D-Pyr), also called hydroxylysylpyridinoline (HP) and lysylpyridinoline (LP), have been so far identified by their natural fluorescent properties in many connective tissues, such as bone, cartilage, and, to a lesser extent, other connective tissues, except skin (14). Pyr is the major mature component of these tissues, whereas D-Pyr is significantly present only in bone tissue. They can be measured precisely in urine by fluorescence detection after high pressure liquid chromatography (HPLC) (15-17), but other techniques have also been reported. Pyr and D-Pyr levels have been reported in normal subjects (18, 19), in patients with osteoarthritis (18, 20) and rheumatoid arthritis (19, 21, 22), and more recently in patients with various metabolic bone diseases and in perimenopausal women (12). Urinary Pyr was markedly increased in patients with primary hyperpathyroidism and Paget's disease of bone. Intravenous treatment with a diphosphonate induced a rapid and dramatic decrease in Pyr and D-Pyr, demonstrating that their excretion reflects resorption and not formation. Finally, urinary Pyr and D-Pyr were significantly higher in postmenopausal normal women than in young women in their thirties (12). In this study we have applied the same sensitive and specific assay for pyridinium cross-links to a well defined population of normal women screened according to their menopausal status in order 1) to assess the increased bone resorption that follows the decrease in hormone replacement therapy production, 2) to measure the effects of a prophylactic estrogen plus gestagen hormone replacement therapy on the urinary excretion of both Pyr and D-Pyr, and 3) to test the ability of urinary Pyr and D-Pyr to predict bone loss in the early postmenopausal period. Materials and Methods The study population comprised 1) 19 healthy premenopausal women, aged 45-53 yr, who had a history of regular vaginal
JCE & M • 1991 Vol72«No2
bleeding and were not taking any medication, including oral contraceptives, known to influence calcium metabolism; and 2) 60 healthy postmenopausal women, aged 45-53 yr, who had all passed a natural menopause '3 months to 3 yr before the study; these women were taking part in a larger clinical trial. Both the premenopausal and postmenopausal women were recruited by a questionnaire about bleeding patterns, medications, and former and present diseases, which was sent to 9836 women, aged 45-55 yr, living in the catchment area of Glostrup Hospital. After medical and aiochemical examinations those women who were free of pas1; and present disease and medications known to influence calcium metabolism were included in the study. The FSH level (mean ± 1 SD) was 19.3 ± 10.2 IU/L in the premenopausal women and 134.3 ± 37.0 in the postmenopausal women. Further details about the selection procedure were given previously (23). The blood and urine samples used for the present study were randomly collected from 40 women receiving 2 yr of placebo tr3atment and 20 women receiving treatment with a cyclic combination of 2 mg estradiol valerate (days 1-21) and 1 mg cyproterone acetate (days 11-21). This study includes the values of Pyr and D-Pyr initially and after 6 and 9 months of hormonu and placebo treatment. Urinary samples were taken on a 24-h based recovery for Pyr and DPyr measurements. Furthermore, blood samples were taken and urine collected in the morning after an overnight fast and tobacco abstinence in orde:: to measure the Pyr and D-Pyr cross-links, total hydroxyprcline, creatinine, and plasma osteocalcin. All blood and urine samples were taken during days 18 and 21. All samples were stored at -20 C until analysis. In accordance with the Helsinki II Declaration, all participants gave their consent after receiving thorough information, and research protocols were approved by the Ethical Committee of Copenhagen County. Pyridinoline assay Pyridinoline was measured in urine, using a slight modification of our previously described method (12), and corrected for creatinine excretion. Urine extraction. A urine sample of 3 mL was hydrolyzed in 6 M HC1 for 3 h in glass tubes with a screw cap and a Teflon liner in a pressure cooker, a procedure which has been shown to give the same results as tie classical 24-h hydrolysis at 105108 C in glass sealed tubes. The cold hydrolysates were first centrifugated at 1000 X g for 10 min, then the supernatant was carefully removed and applied to a CF1-cellulose column for fractionation according to the method originally described for the purification of the elastin cross-links (24). According to the expected relative content o:i" mature cross-linking amino acids in 24-h urine, 0.5-1 mL urine hydrolyzate were mixed with acetic acid, water, and n-butanol (1:1:4, vol/vol/vol) and applied to the CFl-cellulose column (Whatman Co. Ltd., Maidstone, England). After a 3-column vol (15 mL) washing with the acetic acid-water-n-butanol mixture, the cross-linking amino acids were eluted from the colurr.n with 5 mL distilled water into a glass tube and evaporated to dryness on a Savant Speed-Vac (Hicksville, NY) overnight, and the dry residue was stored at 4 C before analysis. For the HPLC analysis, the dry residues were redissolved in 100 nL of a 1% sequanal grade rc-heptafluo-
The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 15 July 2016. at 19:39 For personal use only. No other uses without permission. . All rights reserved.
URINARY EXCRETION OF PYR AND D-PYR DURING MENOPAUSE robutyric acid solution (Merck, Darmstadt, Germany). Pyridinoline standard. A bone hydrolysate preparation was used regularly as an external standard before the application of urine samples to the chromatographic procedure. The Pyr and D-Pyr standards were prepared from powdered human cortical bone according to a method previously described (12). HPLC assay. Pyr and D-Pyr were assayed in the urine extract by HPLC with a modification of the method described by Eyre et al. (16) on a Kontron HPLC System 400 (Kontron Instruments Co., Zurich, Switzerland) equipped with a serial dual piston two-pump system and an automatic injector (Autosampler 460) controlled by a central multitask computerized unit (Data System 450). The reverse phase column used was an Altex Ultrasphere ODS (5 pm; 25 cm x 4.6 mm). Eluant A was 0.01 Mrc-heptafluorobutyricacid in 15% (vol/vol) acetonitrile. Eluant B was 100% acetonitrile. The column was equilibrated in 100% A, and the samples were eluted with the same isocratic gradient run at a flow rate of 1 mL/min during 20 min. The column was then stripped with 100% B for 5 min and reequilibrated at 100% A for 5 min before the next injection. For each urine specimen, two samples of 25 and 50 nh were injected. The fluorescence of the eluted peaks was monitored with a Kontron spectrofluorimeter SFM-25, with excitation at 297 nm and emission at 395 nm. The results of Pyr and D-Pyr were given according to a comparison with an external standard injected at two different amounts and expressed as picomoles per ^mol creatinine. All measurements were performed in a blind fashion, and the code was broken after completion of the study. Other methods RIA for plasma osteocalcin. Plasma osteocalcin was determined by RIA (25). Calf osteocalcin was used for tracer, for standard, and to raise antibody. The tracer was prepared by the Iodogen method (26). Antibody-bound and free 125I-labeled osteocalcin were separated by use of swine antirabbit serum and polyethylene glycol. The sensitivity of the assay is 0.8 ng/mL. The intra- and interassay variations are less than 7% and less than 12%, respectively. Urinary hydroxproline. Fasting urinary hydroxyproline was determined by a spectrophotometric method (27) and corrected by the urinary excretion of creatinine, measured on an SMA 6/ 60 analyzer. Bone densitometry. The bone mineral content (BMC) of the forearm was measured by single photon absorptiometry with an 125I source (100 mCi) and calculated as the mean of 12 scans, 6 on each distal forearm. The system software was developed in Glostrup Hospital. The long term in vivo precision in early postmenopausal women is 1% (28). BMC was measured at the start of treatment and every 3 months for 2 yr. Statistical analysis Comparison between premenopausal and postmenopausal groups was assessed with unpaired Student's t test. The comparison between 2-h and 24-h urinary values within a patient
369
group was made with paired Student's t test. The changes in cross-link excretion with time under hormone replacement therapy and placebo were evaluated by analysis of variance, using as dependent variables the factors time and treatment. The individual changes in BMC were calculated as the slope of nine measurements («BMC) by linear regression analysis and are expressed as the percent change over 2 yr. The correlation coefficients between the biochemical markers and the rate of bone loss were assessed by single and multiple linear regression analysis. Results Effects of menopause and hormone replacement therapy on urinary excretion of Pyr cross-links As shown in Table 1 and Fig. 1, there was a significant 62% increase in Pyr and an 82% increase in D-Pyr in postmenopausal women compared to age-matched premenopausal women. The height and weight of the two populations were virtually the same (not shown). Table 2 shows the time course of fasting urinary Pyr and DPyr during treatment with hormone replacement therapy and placebo. Both cross-links returned to premenopausal levels within 6 months of hormone replacement therapy (Fig. 1) and did not change during placebo treatment. In untreated postmenopausal women, the 24-h excretion was significantly lower than the 2-h excretion for both Pyr (34.3 ± 9.8 vs. 49.8 ± 18.7 pmolAimol creatinine; n = 56; P < 0.001) and D-Pyr (5.4 ± 1.7 us. 8.2 ± 3.4 pmol/Vmol creatinine; n = 56; P < 0.001). However, hormone replacement therapy induced a similar decrease in the 24-h excretion of Pyr and D-Pyr, without any significant change in the placebo group (Fig. 2). Correlations within cross-link excretion and with other biochemical markers Correlations between the 24-h and the fasting excretion of Pyr and D-Pyr were determined in 60 untreated postmenopausal women. In the same urine sample (fasting or 24-h), there was a correlation between the 2 crosslinks (r = 0.85; P < 0.0001; n = 57 for fasting and r = 0.83; P < 0.001; n = 59 for 24-h urine). In contrast, the urinary excretion of Pyr and D-Pyr in the fasting urine TABLE 1. Pyr and D-Pyr in pre- and postmenopausal women Groups Premenopausal (n = 19) Postmenopausal (n = 57)
Age (yr)
D-Pyr (pmol/^mol creatinine)
50.3 ± 2.5
30.8 ± 8.0
4.5 ± 1.4
50.2 ± 2.8
49.8 ± 18.7°
8.2 ± 3.4°
+62
+82
% Change 0
Pyr (pmol//imol creatinine)
P < 0.001 us. premenopausal.
The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 15 July 2016. at 19:39 For personal use only. No other uses without permission. . All rights reserved.
370
UEBELHART ET AL.
and urinary hydroxyproline (r = 0.34; P < 0.01), except for D-Pyr vs. urinary hydrc xyproline (r = 0.24; P = 0.07). Correlations between the 24-h urinary excretion of crosslinks and either fasting urinary hydroxyproline or osteocalcin were not significant, except for a marginal correlation between hydroxyproline and Pyr (r = 0.26; P < 0.05).
FU Pyr 100 -
8o
80 -
60 I
40 -
oo oo • o o ooo* o oooo o•o• *
•
#
0~0
«
Biochemical prediction of the spontaneous rate of bone loss
«
*• • o °
u
oo o o o • oo 20 "
0
i
untreated
hormonal replacement therapy
FU D-Pyr (pmoi/umol c r ) 18 16 " 14 12 10 8 6 4 O
oo ooo
o°o" o °
••!o*o# •oo O°OM TToo^
The spontaneous rate of bone loss was evaluated in 37 untreated early postmenopausal women by repeated measurements of BMC. BMC was significantly correlated with the fasting urinary level of D-Pyr and Pyr measured once at the beginning of the study (Table 3). The rate of bone loss was not predicted by the 24-h urinary excretion of Pyr and D-Pyr. When the mean value of the measurement obtained at 0, 6, and 9 months was used, the prediction o:" the spontaneous rate of bone loss was improved, i.e. the correlation coefficient between the rate of bone loss and the cross-links increased (fasting Pyr, r = 0.49; P < 0.0!.; fasting D-Pyr, r = 0.48; P < 0.01; 24-h Pyr, r = 0.56; P < 0.01; 24-h D-Pyr, r = 0.68; P< 0.001). In the same women the rate of bone loss was also correlated with the initial urinary hydroxyproline (r = -0.45; P < 0.01) and plasma osteocalcin (r = -0.59; P < 0.001) level. The best prediction of the rate of bone loss was obtained when fasting D-Pyr, fasting urinary hydroxyproline, and plasma osteocalcin were combined by multiple regression analysis (Table 3 and Fig. 4).
• •
,oo, o 2 o2 o8
—
JCE&M»1991 Vol 72 • No 2
_•_*
0 I
I
untreated
hormonal replacement therapy
FIG. 1. Pyr (upper panel) and D-Pyr (lower panel) in picomoles per ^mol creatinine. Individual values of fasting D-Pyr in 57 untreated postmenopausal women are shown. • , Values for 16 women before (left) and after (right) 6 months of hormone replacement therapy. The horizontal lines represent the range of fasting D-Pyr of age-matched premenopausal women (mean value given as a solid line and ±2 SD as dashed lines). FU, Fasting urinary sample.
sample could not be reliably predicted by the 24-h urine collection (r = 0.29; P < 0.05; n = 56 for Pyr and r = 0.30; P < 0.05; n = 56 for D-Pyr; Fig. 3). In early postmenopausal women, the fasting urinary Pyr and D-Pyr were correlated with plasma osteocalcin (r = 0.29; P < 0.05, and r = 0.40; P < 0.01, respectively)
Discussion Estrogen deficiency results in a postmenopausal increase in bone resorption, whereas increased bone formation is secondarily elevated as a result of the coupling phenomenon (29). The rate of the postmenopausal bone loss is due to an imbalance between the two turnover processes, and a sensitive marker of bone resorption would be expected to reflect these changes in bone turnover. As expected, menopause results in a significant increase in fasting pyridinium cross-link excretion. The relative increase (62% fcr Pyr and 82% for D-Pyr) is lower than the increase that we previously reported (128% for 24-h Pyr and 200% for 24-h D-Pyr) (12). In the previous study, however, premenopausal women were younger (31 ± 6 yr old), while in the present study preand postmenopausal women were age matched in order to exclude the effect of age. Some minor fluctuation in estrogen production by the ovary of a 50-yr-old woman are likely to occur and account for a mild increase in bone turnover.
The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 15 July 2016. at 19:39 For personal use only. No other uses without permission. . All rights reserved.
URINARY EXCRETION OF PYR AND D-PYR DURING MENOPAUSE
371
TABLE 2. Time course of fasting urinary Pyr and D-Pyr before and during treatment Treatment Groups
Before
Pyr (pmol/Vmol creatinine) Estrogen (n = 20) Controls (n = 37) D-Pyr (pinol/nmol creatinine) Estrogen (n = 20) Controls (n = 37)
3 months
6 months
9 months
48.6 ± 15.4 50.4 ± 20.4
39.7 ± 14.9°
NM
29.8±6.5 a ' 6 54.2 ± 28.3
36.1 ± 13.5"'6 47.4 ± 27.6
8.2 ± 3.3 8.2 ± 3.4
6.0 ± 2.3° NM
4.3 ± 1.2°'6 9.1 ± 4.1
5.5±2.4 a ' 6 7.4 ± 4.3
C
0
P < 0.001 vs. before. * P < 0.05 us. controls (by ANOVA). c Not measured. FU Pyr 90-i
70-
503030r-o.85* n-57
10-
11 Before
9 months
8-
13
15
FU D-Pyr 19
17
24h D-Pyr 12 -[
10 -
8 6 -
24hr
4 2
J
12 Before
9 months
FiG. 2. Effects of hormone replacement therapy (O and • ) and placebo ( • and • ) on the urinary excretion of cross-links measured on a fasting sample (O and • ) and on a 24-h collection (D and • ) . The upper panel depicts the Pyr levels, and the lower panel the D-Pyr levels. Data are expressed as the mean ± SE in picomoles per /*mol creatinine.
The effect of hormone replacement therapy on postmenopausal bone loss is generally accepted to be caused by a reduction in bone resorption and a secondary decrease in bone formation, reflecting a decrease in the activation frequency of new bone-remodeling units (3032). The fact that urinary Pyr and D-Pyr, which have been demonstrated to be specific markers of bone resorption, decreased significantly 3 months after the onset of hormone replacement therapy, is consistant with this
\ FU D-Pyr 18
16
FIG. 3. Correlation between the fasting excretion of D-Pyr and either the 24-h urinary D-Pyr (lower panel) or the fasting urinary (FU) Pyr {upper panel) in 56 untreated postmenopausal women. TABLE 3. Correlations (r) between the rate of bone loss measured over 2 years and the initial levels of pyridinium crosslinks alone or combined with other biochemical markers of bone turnover. Parameters correlated with the rate of bone loss Cross-link Cross-link + hydroxyproline Cross-link + hydroxyproline + osteocalcin
D-Pyr
Pyr r
P
r
P
-0.34 -0.53 -0.75
Vol. 72, No. 2 Printed in U.S.A.
Effect of Menopause and Hormone Replacement Therapy on the Urinary Excretion of Pyridinium Cross-Links DANIEL UEBELHART, ANNETTE SCHLEMMER, JULIA S. JOHANSEN, EVELYNE GINEYTS, CLAUS CHRISTIANSEN, AND PIERRE D. DELMAS INSERM Unit 234 and Service de Rhumatologie et de Pathologie Osseuse, Hopital E. Herriot, Lyon, France; and the Department of Clinical Chemistry, University of Copenhagen, Glostrup Hospital (A.S., J.S.J., C.C.), Glostrup, Denmark
ABSTRACT. Pyridinoline (Pyr) and deoxypyridinoline (DPyr) are two cross-links of collagen molecules that are present in the extracellular matrix and released during its degradation. In contrast to the wide distribution of collagen, Pyr is present in bone and cartilage, but not in significant amounts in other connective tissues, and D-Pyr appears to be specific for bone tissue. Therefore, the urinary excretion of Pyr and D-Pyr might be a sensitive marker of bone matrix degradation. Using a specific high pressure liquid chromatography assay we have measured Pyr and D-Pyr cross-links in a 24-h and a fasting urine sample in 60 early postmenopausal women and 19 premenopausal women matched for age. Menopause induced a 62% increase in Fu Pyr (49.8 ± 18.7 vs. 30.8 ± 8.0 pmol//imol creatinine; P < 0.001) and an 82% increase in Fu D-Pyr (8.2 ± 3.4 vs. 4.5 ± 1.4 pmol/Vmol creatinine; P < 0.001). In 20 postmenopausal women on hormone replacement therapy, urinary Pyr and D-Pyr returned to premenopausal levels within 6 months, contrasting with unchanged levels during placebo treatment. The 24-h excretion of Pyr and D-Pyr was significantly lower than the fasting excretion, but was similarly decreased after hormone replacement therapy. Pyr and D-Pyr excretion measured in the same urinary sample were highly correlated (r = 0.85 for fasting and 0.83 for 24-h sampling), but correlations
between fasting and 24-h values were weak (D-Pyr, r = 0.30; Pyr, r= 0.29; P < 0.05 for both). Correlations between urinary cross-links and other markers of bone turnover (Fu hydroxyproline/creatinine and plasma osteocalcin) were significant but low (Pyr vs. osteocalcin, r = 0.29, P < 0.05; Pyr vs. hydroxyproline, r = 0/.34; P < 0.01; D-Pyr vs. osteocalcin, r = 0.39; P < 0.01), except for D-Pyr us. hydroxyproline (r = 0.24; P = 0,07), suggesting that these markers reflect different events of bone metabolism. Finally, a single measurement of the fasting excretion, but not of the 24-h excretion, of cross-links was significantly correlated (Pyr, r = 0.34; P < 0.05; D-Pyr, r = -0.46; P < 0.01), with the subsequent spontaneous rate of bone loss assessed by repeated measurements of the radial bone mineral content in 37 postmenopausal women. The correlation with the rate of bone loss was improved when the simultaneous measurement of plasma osteocalcin and urinary hydroxyproline/creatinine was combined (Pyr, r = 0.75; P < 0.001; D-Pyr, r = 0.77; P < 0.001). We conclude that the fasting excretion of Pyr and D-Pyr reflects bone turnover changes at the menopause and the effects of hormone replacement therapy. The combination of cross-links, hydroxyproline, and osteocalcin measurement seems promising to evaluate the rate of bone loss in postmenopausal women. (J Clin Endocrinol Metab 72: 367-373, 1991)
A
resorption, respectively, but their sensitivity and specificity are limited (3). Other markers of bone formation include the bone-specific alkaline phosphatase, serum osteocalcin, also called bone gla protein (4, 5), and the carboxy-terminal propeptide of type 1 procollagen. All of these markers are increased at the time of menopause, and we have recently shown that serum osteocalcin is significantly correlated with the spontaneous rate of bone loss in recently menopausal women (6). Similarly, serum osteocalcin was reported to be the best single predictor of the rate of bone loss in a large cohort of periand postmenopausal women followed prospectively during 4 yr (7). Because an increase in bone resorption is the primary event following estrogen deficiency, a sensitive marker of bone resorption would be valuable to analyze the effects of menopause on bone turnover. As stated above, urinary hydroxyproline is not specific for bone collagen
N EFFECTIVE prevention of osteoporosis can be achieved at the time of menopause with long term hormone replacement therapy. Because of the constraints, the cost, and the potential side-effects of such treatment, it is generally accepted that intervention should be targeted at women who present a high risk of osteoporosis. The two main factors responsible for that risk are represented by a low peak bone mass and a high rate of bone loss at the time of hormone replacement therapy deprivation (1, 2). The postmenopausal rate of bone loss is related to an overall increase in bone turnover, which can be assessed by several biochemical markers. Serum alkaline phosphatase and urinary hydroxyproline are commonly used to assess bone formation and Received April 15,1990. Address all correspondence and requests for reprints to: Dr. P. D. Delmas, Hopital E. Herriot, Pav. F, 69437 Lyon Cedex 03, France. 367
The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 15 July 2016. at 19:39 For personal use only. No other uses without permission. . All rights reserved.
UEBELHART ET AL.
368
resorption, as it is also derived from collagen synthesis during the breakdown of the procollagen N-terminal extension peptides and of neosynthesized collagen molecules (8, 9). Plasma tartrate-resistant acid phosphatase and urinary galactosyl-hydroxylypsyl glycoside, two recently described markers of bone resorption, are increased in osteoporotic women (10, 11) and warrant further investigation to establish their sensitivity and specificity. Recently, the urinary excretion of the pyridinium cross-links of collagen have been suggested to be a sensitive and specific marker of bone resorption (12). The extracellular matrix is stabilized by the formation of covalent cross-links between adjacent collagen chains, and it is thought that the reducible aldimine cross-links initially formed are converted to mature nonreducible compounds (13). Two mature cross-linking amino acids, pyridinoline (Pyr) and deoxypyridinoline (D-Pyr), also called hydroxylysylpyridinoline (HP) and lysylpyridinoline (LP), have been so far identified by their natural fluorescent properties in many connective tissues, such as bone, cartilage, and, to a lesser extent, other connective tissues, except skin (14). Pyr is the major mature component of these tissues, whereas D-Pyr is significantly present only in bone tissue. They can be measured precisely in urine by fluorescence detection after high pressure liquid chromatography (HPLC) (15-17), but other techniques have also been reported. Pyr and D-Pyr levels have been reported in normal subjects (18, 19), in patients with osteoarthritis (18, 20) and rheumatoid arthritis (19, 21, 22), and more recently in patients with various metabolic bone diseases and in perimenopausal women (12). Urinary Pyr was markedly increased in patients with primary hyperpathyroidism and Paget's disease of bone. Intravenous treatment with a diphosphonate induced a rapid and dramatic decrease in Pyr and D-Pyr, demonstrating that their excretion reflects resorption and not formation. Finally, urinary Pyr and D-Pyr were significantly higher in postmenopausal normal women than in young women in their thirties (12). In this study we have applied the same sensitive and specific assay for pyridinium cross-links to a well defined population of normal women screened according to their menopausal status in order 1) to assess the increased bone resorption that follows the decrease in hormone replacement therapy production, 2) to measure the effects of a prophylactic estrogen plus gestagen hormone replacement therapy on the urinary excretion of both Pyr and D-Pyr, and 3) to test the ability of urinary Pyr and D-Pyr to predict bone loss in the early postmenopausal period. Materials and Methods The study population comprised 1) 19 healthy premenopausal women, aged 45-53 yr, who had a history of regular vaginal
JCE & M • 1991 Vol72«No2
bleeding and were not taking any medication, including oral contraceptives, known to influence calcium metabolism; and 2) 60 healthy postmenopausal women, aged 45-53 yr, who had all passed a natural menopause '3 months to 3 yr before the study; these women were taking part in a larger clinical trial. Both the premenopausal and postmenopausal women were recruited by a questionnaire about bleeding patterns, medications, and former and present diseases, which was sent to 9836 women, aged 45-55 yr, living in the catchment area of Glostrup Hospital. After medical and aiochemical examinations those women who were free of pas1; and present disease and medications known to influence calcium metabolism were included in the study. The FSH level (mean ± 1 SD) was 19.3 ± 10.2 IU/L in the premenopausal women and 134.3 ± 37.0 in the postmenopausal women. Further details about the selection procedure were given previously (23). The blood and urine samples used for the present study were randomly collected from 40 women receiving 2 yr of placebo tr3atment and 20 women receiving treatment with a cyclic combination of 2 mg estradiol valerate (days 1-21) and 1 mg cyproterone acetate (days 11-21). This study includes the values of Pyr and D-Pyr initially and after 6 and 9 months of hormonu and placebo treatment. Urinary samples were taken on a 24-h based recovery for Pyr and DPyr measurements. Furthermore, blood samples were taken and urine collected in the morning after an overnight fast and tobacco abstinence in orde:: to measure the Pyr and D-Pyr cross-links, total hydroxyprcline, creatinine, and plasma osteocalcin. All blood and urine samples were taken during days 18 and 21. All samples were stored at -20 C until analysis. In accordance with the Helsinki II Declaration, all participants gave their consent after receiving thorough information, and research protocols were approved by the Ethical Committee of Copenhagen County. Pyridinoline assay Pyridinoline was measured in urine, using a slight modification of our previously described method (12), and corrected for creatinine excretion. Urine extraction. A urine sample of 3 mL was hydrolyzed in 6 M HC1 for 3 h in glass tubes with a screw cap and a Teflon liner in a pressure cooker, a procedure which has been shown to give the same results as tie classical 24-h hydrolysis at 105108 C in glass sealed tubes. The cold hydrolysates were first centrifugated at 1000 X g for 10 min, then the supernatant was carefully removed and applied to a CF1-cellulose column for fractionation according to the method originally described for the purification of the elastin cross-links (24). According to the expected relative content o:i" mature cross-linking amino acids in 24-h urine, 0.5-1 mL urine hydrolyzate were mixed with acetic acid, water, and n-butanol (1:1:4, vol/vol/vol) and applied to the CFl-cellulose column (Whatman Co. Ltd., Maidstone, England). After a 3-column vol (15 mL) washing with the acetic acid-water-n-butanol mixture, the cross-linking amino acids were eluted from the colurr.n with 5 mL distilled water into a glass tube and evaporated to dryness on a Savant Speed-Vac (Hicksville, NY) overnight, and the dry residue was stored at 4 C before analysis. For the HPLC analysis, the dry residues were redissolved in 100 nL of a 1% sequanal grade rc-heptafluo-
The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 15 July 2016. at 19:39 For personal use only. No other uses without permission. . All rights reserved.
URINARY EXCRETION OF PYR AND D-PYR DURING MENOPAUSE robutyric acid solution (Merck, Darmstadt, Germany). Pyridinoline standard. A bone hydrolysate preparation was used regularly as an external standard before the application of urine samples to the chromatographic procedure. The Pyr and D-Pyr standards were prepared from powdered human cortical bone according to a method previously described (12). HPLC assay. Pyr and D-Pyr were assayed in the urine extract by HPLC with a modification of the method described by Eyre et al. (16) on a Kontron HPLC System 400 (Kontron Instruments Co., Zurich, Switzerland) equipped with a serial dual piston two-pump system and an automatic injector (Autosampler 460) controlled by a central multitask computerized unit (Data System 450). The reverse phase column used was an Altex Ultrasphere ODS (5 pm; 25 cm x 4.6 mm). Eluant A was 0.01 Mrc-heptafluorobutyricacid in 15% (vol/vol) acetonitrile. Eluant B was 100% acetonitrile. The column was equilibrated in 100% A, and the samples were eluted with the same isocratic gradient run at a flow rate of 1 mL/min during 20 min. The column was then stripped with 100% B for 5 min and reequilibrated at 100% A for 5 min before the next injection. For each urine specimen, two samples of 25 and 50 nh were injected. The fluorescence of the eluted peaks was monitored with a Kontron spectrofluorimeter SFM-25, with excitation at 297 nm and emission at 395 nm. The results of Pyr and D-Pyr were given according to a comparison with an external standard injected at two different amounts and expressed as picomoles per ^mol creatinine. All measurements were performed in a blind fashion, and the code was broken after completion of the study. Other methods RIA for plasma osteocalcin. Plasma osteocalcin was determined by RIA (25). Calf osteocalcin was used for tracer, for standard, and to raise antibody. The tracer was prepared by the Iodogen method (26). Antibody-bound and free 125I-labeled osteocalcin were separated by use of swine antirabbit serum and polyethylene glycol. The sensitivity of the assay is 0.8 ng/mL. The intra- and interassay variations are less than 7% and less than 12%, respectively. Urinary hydroxproline. Fasting urinary hydroxyproline was determined by a spectrophotometric method (27) and corrected by the urinary excretion of creatinine, measured on an SMA 6/ 60 analyzer. Bone densitometry. The bone mineral content (BMC) of the forearm was measured by single photon absorptiometry with an 125I source (100 mCi) and calculated as the mean of 12 scans, 6 on each distal forearm. The system software was developed in Glostrup Hospital. The long term in vivo precision in early postmenopausal women is 1% (28). BMC was measured at the start of treatment and every 3 months for 2 yr. Statistical analysis Comparison between premenopausal and postmenopausal groups was assessed with unpaired Student's t test. The comparison between 2-h and 24-h urinary values within a patient
369
group was made with paired Student's t test. The changes in cross-link excretion with time under hormone replacement therapy and placebo were evaluated by analysis of variance, using as dependent variables the factors time and treatment. The individual changes in BMC were calculated as the slope of nine measurements («BMC) by linear regression analysis and are expressed as the percent change over 2 yr. The correlation coefficients between the biochemical markers and the rate of bone loss were assessed by single and multiple linear regression analysis. Results Effects of menopause and hormone replacement therapy on urinary excretion of Pyr cross-links As shown in Table 1 and Fig. 1, there was a significant 62% increase in Pyr and an 82% increase in D-Pyr in postmenopausal women compared to age-matched premenopausal women. The height and weight of the two populations were virtually the same (not shown). Table 2 shows the time course of fasting urinary Pyr and DPyr during treatment with hormone replacement therapy and placebo. Both cross-links returned to premenopausal levels within 6 months of hormone replacement therapy (Fig. 1) and did not change during placebo treatment. In untreated postmenopausal women, the 24-h excretion was significantly lower than the 2-h excretion for both Pyr (34.3 ± 9.8 vs. 49.8 ± 18.7 pmolAimol creatinine; n = 56; P < 0.001) and D-Pyr (5.4 ± 1.7 us. 8.2 ± 3.4 pmol/Vmol creatinine; n = 56; P < 0.001). However, hormone replacement therapy induced a similar decrease in the 24-h excretion of Pyr and D-Pyr, without any significant change in the placebo group (Fig. 2). Correlations within cross-link excretion and with other biochemical markers Correlations between the 24-h and the fasting excretion of Pyr and D-Pyr were determined in 60 untreated postmenopausal women. In the same urine sample (fasting or 24-h), there was a correlation between the 2 crosslinks (r = 0.85; P < 0.0001; n = 57 for fasting and r = 0.83; P < 0.001; n = 59 for 24-h urine). In contrast, the urinary excretion of Pyr and D-Pyr in the fasting urine TABLE 1. Pyr and D-Pyr in pre- and postmenopausal women Groups Premenopausal (n = 19) Postmenopausal (n = 57)
Age (yr)
D-Pyr (pmol/^mol creatinine)
50.3 ± 2.5
30.8 ± 8.0
4.5 ± 1.4
50.2 ± 2.8
49.8 ± 18.7°
8.2 ± 3.4°
+62
+82
% Change 0
Pyr (pmol//imol creatinine)
P < 0.001 us. premenopausal.
The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 15 July 2016. at 19:39 For personal use only. No other uses without permission. . All rights reserved.
370
UEBELHART ET AL.
and urinary hydroxyproline (r = 0.34; P < 0.01), except for D-Pyr vs. urinary hydrc xyproline (r = 0.24; P = 0.07). Correlations between the 24-h urinary excretion of crosslinks and either fasting urinary hydroxyproline or osteocalcin were not significant, except for a marginal correlation between hydroxyproline and Pyr (r = 0.26; P < 0.05).
FU Pyr 100 -
8o
80 -
60 I
40 -
oo oo • o o ooo* o oooo o•o• *
•
#
0~0
«
Biochemical prediction of the spontaneous rate of bone loss
«
*• • o °
u
oo o o o • oo 20 "
0
i
untreated
hormonal replacement therapy
FU D-Pyr (pmoi/umol c r ) 18 16 " 14 12 10 8 6 4 O
oo ooo
o°o" o °
••!o*o# •oo O°OM TToo^
The spontaneous rate of bone loss was evaluated in 37 untreated early postmenopausal women by repeated measurements of BMC. BMC was significantly correlated with the fasting urinary level of D-Pyr and Pyr measured once at the beginning of the study (Table 3). The rate of bone loss was not predicted by the 24-h urinary excretion of Pyr and D-Pyr. When the mean value of the measurement obtained at 0, 6, and 9 months was used, the prediction o:" the spontaneous rate of bone loss was improved, i.e. the correlation coefficient between the rate of bone loss and the cross-links increased (fasting Pyr, r = 0.49; P < 0.0!.; fasting D-Pyr, r = 0.48; P < 0.01; 24-h Pyr, r = 0.56; P < 0.01; 24-h D-Pyr, r = 0.68; P< 0.001). In the same women the rate of bone loss was also correlated with the initial urinary hydroxyproline (r = -0.45; P < 0.01) and plasma osteocalcin (r = -0.59; P < 0.001) level. The best prediction of the rate of bone loss was obtained when fasting D-Pyr, fasting urinary hydroxyproline, and plasma osteocalcin were combined by multiple regression analysis (Table 3 and Fig. 4).
• •
,oo, o 2 o2 o8
—
JCE&M»1991 Vol 72 • No 2
_•_*
0 I
I
untreated
hormonal replacement therapy
FIG. 1. Pyr (upper panel) and D-Pyr (lower panel) in picomoles per ^mol creatinine. Individual values of fasting D-Pyr in 57 untreated postmenopausal women are shown. • , Values for 16 women before (left) and after (right) 6 months of hormone replacement therapy. The horizontal lines represent the range of fasting D-Pyr of age-matched premenopausal women (mean value given as a solid line and ±2 SD as dashed lines). FU, Fasting urinary sample.
sample could not be reliably predicted by the 24-h urine collection (r = 0.29; P < 0.05; n = 56 for Pyr and r = 0.30; P < 0.05; n = 56 for D-Pyr; Fig. 3). In early postmenopausal women, the fasting urinary Pyr and D-Pyr were correlated with plasma osteocalcin (r = 0.29; P < 0.05, and r = 0.40; P < 0.01, respectively)
Discussion Estrogen deficiency results in a postmenopausal increase in bone resorption, whereas increased bone formation is secondarily elevated as a result of the coupling phenomenon (29). The rate of the postmenopausal bone loss is due to an imbalance between the two turnover processes, and a sensitive marker of bone resorption would be expected to reflect these changes in bone turnover. As expected, menopause results in a significant increase in fasting pyridinium cross-link excretion. The relative increase (62% fcr Pyr and 82% for D-Pyr) is lower than the increase that we previously reported (128% for 24-h Pyr and 200% for 24-h D-Pyr) (12). In the previous study, however, premenopausal women were younger (31 ± 6 yr old), while in the present study preand postmenopausal women were age matched in order to exclude the effect of age. Some minor fluctuation in estrogen production by the ovary of a 50-yr-old woman are likely to occur and account for a mild increase in bone turnover.
The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 15 July 2016. at 19:39 For personal use only. No other uses without permission. . All rights reserved.
URINARY EXCRETION OF PYR AND D-PYR DURING MENOPAUSE
371
TABLE 2. Time course of fasting urinary Pyr and D-Pyr before and during treatment Treatment Groups
Before
Pyr (pmol/Vmol creatinine) Estrogen (n = 20) Controls (n = 37) D-Pyr (pinol/nmol creatinine) Estrogen (n = 20) Controls (n = 37)
3 months
6 months
9 months
48.6 ± 15.4 50.4 ± 20.4
39.7 ± 14.9°
NM
29.8±6.5 a ' 6 54.2 ± 28.3
36.1 ± 13.5"'6 47.4 ± 27.6
8.2 ± 3.3 8.2 ± 3.4
6.0 ± 2.3° NM
4.3 ± 1.2°'6 9.1 ± 4.1
5.5±2.4 a ' 6 7.4 ± 4.3
C
0
P < 0.001 vs. before. * P < 0.05 us. controls (by ANOVA). c Not measured. FU Pyr 90-i
70-
503030r-o.85* n-57
10-
11 Before
9 months
8-
13
15
FU D-Pyr 19
17
24h D-Pyr 12 -[
10 -
8 6 -
24hr
4 2
J
12 Before
9 months
FiG. 2. Effects of hormone replacement therapy (O and • ) and placebo ( • and • ) on the urinary excretion of cross-links measured on a fasting sample (O and • ) and on a 24-h collection (D and • ) . The upper panel depicts the Pyr levels, and the lower panel the D-Pyr levels. Data are expressed as the mean ± SE in picomoles per /*mol creatinine.
The effect of hormone replacement therapy on postmenopausal bone loss is generally accepted to be caused by a reduction in bone resorption and a secondary decrease in bone formation, reflecting a decrease in the activation frequency of new bone-remodeling units (3032). The fact that urinary Pyr and D-Pyr, which have been demonstrated to be specific markers of bone resorption, decreased significantly 3 months after the onset of hormone replacement therapy, is consistant with this
\ FU D-Pyr 18
16
FIG. 3. Correlation between the fasting excretion of D-Pyr and either the 24-h urinary D-Pyr (lower panel) or the fasting urinary (FU) Pyr {upper panel) in 56 untreated postmenopausal women. TABLE 3. Correlations (r) between the rate of bone loss measured over 2 years and the initial levels of pyridinium crosslinks alone or combined with other biochemical markers of bone turnover. Parameters correlated with the rate of bone loss Cross-link Cross-link + hydroxyproline Cross-link + hydroxyproline + osteocalcin
D-Pyr
Pyr r
P
r
P
-0.34 -0.53 -0.75
