Heaney originally proposed that estrogen effects were mediated by inducing resistance of the skeleton to the effects of parathyroid hormone (PTH) [15]. Given that PTH has been implicated in facilitating bone loss through stimulation of osteoclast recruitment [7, 15], resistance to its actions might decrease the rate of bone loss. In ovariectomized rats given exogenous PTH, the calcium content and cortical thickness of the femur were diminished to a greater extent in estrogen-depleted animals than in those with a normal endogenous estrogen supply, implying an increased sensitivity of the estrogen-deficient skeleton to PTH [16]. In-vitro data also support the concept that estrogens inhibit osteoclastic resorption [17, 18] and PTH-stimulated osteoclastic resorption [19]. One limited study of postmenopausal women showed a reduced response of hydroxyproline excretion to exogenously administered PTH after estrogen was administered [20]. Further, estrogens have been used successfully to treat primary hyperparathyroidism with decreases in urinary hydroxyproline [21, 22] and serum alkaline phosphatase [22], suggesting a reduction in bone turnover. Because PTH levels are not diminished by estrogen treatment, it is tempting to speculate that estrogens might exert their effects in hyperparathyroidism by decreasing the sensitivity of the skeleton to PTH. Little in-vivo evidence, however, supports the concept that estrogen makes the skeleton more resistant to PTH. Therefore, we sought to determine in a systematic fashion by infusing (1-34) human PTH and measuring biochemical indices of skeletal turnover in groups of untreated and estrogen-treated osteoporotic women, whether estrogen induced resistance to the skeletal remodeling effects of PTH.

Methods

Top Methods Results Discussion Author & Article Info References

All patients attending our bone metabolism clinic during a 1.5-year period who had primary postmenopausal osteoporosis and had not received treatment with calcitonin, fluoride, or diphosphonates were asked to participate in the protocol (n = 40). Of these, 32 agreed to participate. Untreated osteoporotic women (n = 15) had a history of atraumatic fractures or a bone mineral density more than 2 standard deviations less than that of mean young normal women (as determined in our laboratory using dual-energy photon or x-ray absorptiometry) [23]. Estrogen-treated osteoporotic women (n = 17) had, in addition, been treated with estrogen for 1 month to 28 years (mean, 3.9 years; median, 2.5 years). Estrogen treatment was in the form of either oral conjugated estrogens (n = 15) (Premarin, 0.625 mg/d; Wyeth-Ayerst, Philadelphia, Pennsylvania) for 25 to 31 days per month, or transdermal estradiol (n = 2) (Estraderm, 0.05 mg/d; Ciba Geigy, Summit, New Jersey). Women whose uterus had not been removed (n = 11) were also treated with a progestin (Provera, 5 to 10 mg/d; Upjohn, Kalamazoo, Michigan) cyclically for 10 to 15 days each month. All patients gave written informed consent; and the study was approved by the institutional review board of Helen Hayes Hospital.

Infusion Protocol

We used a previously described protocol [24] involving infusion of (1-34) human parathyroid hormone (Rhone-Poulenc Rorer; Collegeville, Pennsylvania), 0.55U/(kg x h) for 20 hours (a minor modification from the originally described protocol) after an 8-hour fast. Participants were maintained on a low-hydroxyproline diet for 72 hours before the investigation and resumed a regular hospital diet of moderately low hydroxyproline content (median, 280 mg/d) during the study. Three basal blood samples and a basal urine sample were obtained in a 30-minute period, followed by blood and urine sampling every 4 hours.

Biochemical Testing

Serum was analyzed for tartrate-resistant acid phosphatase in the presence of 20-mM tartaric acid in citrate buffer, pH 4.8, using a Sigma Diagnostic kit for phosphatase (Sigma Company, St. Louis, Missouri) [25]. Tartrate-resistant acid phosphatase assays were not done on samples that were not frozen immediately at –80°C. Serum was also analyzed for bone Gla protein using a commercial radioimmunoassay kit (Incstar Company, Stillwater, Minnesota) [26]; insulin-like growth factor-1 with a commercial immunoradiometric assay (Diagnostic Systems Laboratory, Webster, Texas) [27]; ionized calcium (NOVA 8 Ionized Calcium Analyzer; Nova Biomedical, Newton, Massachusetts); alkaline phosphatase using an Automated Discrete Chemistry Analyzer (Cobas, Mira-S, Roche Diagnostic System; Montclair, New Jersey). Urine was analyzed for deoxypyridinoline and pyridinoline using ion-paired, reversed-phase high-performance liquid chromatography by a modification of the technique of Black and colleagues [28, 29]; hydroxyproline was analyzed by the method of Kivirikko and colleagues [30]. To minimize assay variation at different time points, each participant's serum sample was measured twice. The average of these duplicate determinations was used as the data point. Interassay and intra-assay coefficients of variation for the assays are shown in Table 1. Intra-assay coefficients of variation were calculated from the means and standard deviations of a single basal sample measured 10 times in the same assay. Interassay coefficients of variation were calculated from means and standard deviations of duplicate determinations of a single basal sample assayed on six different occasions. Hydroxyproline and bone Gla protein were measured at all time points, whereas deoxypyridinoline, pyridinoline, tartrate-resistant acid phosphatase, and insulin-like growth factor-1 were measured only at baseline and 20 hours.

Statistical Analysis

Analysis of variance was used to assess differences in biochemical variables over time and between groups. Analysis of covariance was used to assess the importance of factors such as bone mineral density, age, and years from menopause. For variables in which data were available only for the beginning and end of the infusion, paired t-tests were used to evaluate differences from basal to postinfusion levels within groups. For variables that were not normally distributed, nonparametric tests were used to determine differences over time (Wilcoxon signed-rank test) and between groups (Wilcoxon rank-sum test). Linear regression equations were calculated using the method of least squares. Spearman correlations were determined for relationships between variables that were not normally distributed. The statistical software we used was SAS (SAS Institute; Cary, North Carolina).

Results

Top Methods Results Discussion Author & Article Info References

Characteristics of the women are shown in Table 2. Mean age and years from menopause were both approximately 8 years higher in the untreated osteoporotic women, although the difference was significant only for age (P < 0.02). Mean heights and weights were similar in the two groups. Mean bone mineral density measurements were higher at all sites in the estrogen-treated group, but none of the differences reached statistical significance.

Basal laboratory values are shown in Table 3. All resorption and formation indices were higher (mean, 41.7% higher for all variables; range, 8.5% to 102%) in the untreated women, but statistically significant differences were found only for urine deoxypyridinoline and pyridinoline (P < 0.02).

When the two groups were combined, basal bone Gla protein correlated positively with alkaline phosphatase (r = 0.58, P < 0.002) and negatively with insulin-like growth factor-1 (r = –0.60,P < 0.001). These variables did not correlate significantly with age; however, deoxypyridinoline, pyridinoline, and tartrate-resistant acid phosphatase did correlate with age (r = 0.37 to 0.55, P < 0.04). Deoxypyridinoline and pyridinoline correlated strongly with each other (r = 0.87, P < 0.001) but not with the other indices. A negative correlation with insulin-like growth factor-1 was observed for deoxypyridinoline and pyridinoline (r = –0.38,P < 0.04). Hydroxyproline related weakly to tartrate-resistant acid phosphatase (r = 0.41, P = 0.08) but not with the other indices.

Biochemical Response to Parathyroid Hormone Infusion

Resorption Indices

Changes in urinary hydroxyproline (micromole/micromole creatinine) during infusion are shown in Figure 1. In both groups, hydroxyproline increased from baseline by 4 hours in untreated women (P < 0.05) and by 8 hours in estrogen-treated women (P < 0.02) and remained significantly elevated throughout infusion in both groups. The response in estrogen-treated women reached a plateau at 4 hours, whereas levels in untreated women continued to increase slowly throughout the infusion. Group differences were seen by analysis of variance (P = 0.05) with a time trend for untreated (P < 0.005) but not for estrogen-treated women. For hydroxyproline data, when bone mineral density, age, duration of time from menopause, duration or mode of estrogen therapy, or presence of progestin was analyzed separately as covariates in the ANOVA, none of these factors was a statistically significant independent predictor.

View larger version (17K): [in this window] [in a new window]   Figure 1.

For both deoxypyridinoline and pyridinoline, values were greater at 20 hours in untreated than in estrogen-treated osteoporotic women (Table 3). In the estrogen-treated women, small increments were seen in both deoxypyridinoline and pyridinoline (P < 0.05 for deoxypyridinoline only), whereas larger increments in both deoxypyridinoline and pyridinoline levels were seen in untreated women (P < 0.05 for both variables). The difference in deoxypyridinoline increment between groups just missed significance (P = 0.06), whereas the increase in pyridinoline was higher at 20 hours in untreated compared with estrogen-treated women (P < 0.05). Tartrate-resistant acid phosphatase data were available for only 8 untreated and 10 estrogen-treated women (see Methods and Table 3. Levels rose significantly in untreated [P = 0.02] but not in estrogen-treated women. No relationship was found between duration or mode of estrogen therapy or addition of progestin treatment and change in deoxypyridinoline, pyridinoline, or tartrate-resistant acid phosphatase.

When data for both groups were combined, increments in pyridinoline and deoxypyridinoline correlated positively with age (r = 0.38 and 0.41, P < 0.03), a relationship not seen with other resorptive indices. None of the increments in resorption variables correlated with bone mineral density in the total patient sample. Increments in hydroxyproline at 20 hours correlated weakly with increments in tartrate-resistant acid phosphatase (r = 0.38, P = 0.11). Neither increments in hydroxyproline nor tartrate-resistant acid phosphatase correlated significantly with those of deoxypyridinoline or pyridinoline.

When the untreated and estrogen-treated groups were analyzed separately, increments in hydroxyproline, deoxypyridinoline, and pyridinoline did not show a significant relationship to age. No correlations were found between the increments in any variable and the duration of estrogen use. The 20-hour hydroxyproline increase correlated significantly with increases in tartrate-resistant acid phosphatase, deoxypyridinoline, and pyridinoline (r = 0.69 to 0.76, all P < 0.03) in estrogen-treated women, whereas none of these relationships was seen in untreated women.

Formation Indices

Bone Gla protein levels throughout the infusion showed parallel changes in untreated and estrogen-treated women (see Figure 1). Both groups showed early increments followed by return to basal levels within 16 hours of infusion. When the two groups were evaluated together, a time trend was seen [P < 0.003]. Group differences approached significance (P = 0.06). At no time point in either group did bone Gla protein diminish significantly below baseline.

Figure 1(bottom panel) shows alkaline phosphatase levels, which remained parallel and below basal values in the two groups throughout the infusion. Significant time trends were seen in both groups (P < 0.02), with no statistical difference between the groups. Decreases in alkaline phosphatase at both 4 and 20 hours did not correlate with increments in any of the other metabolic indices or with age or bone mineral density.

Insulin-like Growth Factor-1

Basal insulin-like growth factor-1 levels were 6.10 nmol/L in untreated women and 5.92 nmol/L in estrogen-treated women (see Table 3). Levels for both untreated and estrogen-treated osteoporotic women increased slightly above basal within 20 hours [P < 0.03], but incremental changes were identical. Changes in insulin-like growth factor-1 at 20 hours did not correlate with age, bone mineral density, or changes in any other variables in either group separately or combined.

Discussion

Top Methods Results Discussion Author & Article Info References

In the performance of its major homeostatic function—maintenance of serum calcium—parathyroid hormone liberates calcium from the skeleton in addition to increasing renal calcium reabsorption and, indirectly, gastrointestinal calcium absorption [31]. The skeletal action is accomplished at least in part through an increase in the frequency of remodeling activation. Increased osteoclast activation in response to PTH has been shown in neonatal rat osteoclasts on bone slices [32]. In addition, PTH induces a rapid increase in osteoclast number and osteoclast nuclei in intact rats [33]. A small net loss of bone occurs on average in each remodeling unit in adults, particularly in women after menopause [34]. Because PTH is known to increase activation frequency, a factor such as estrogen, which could decrease the sensitivity of the skeleton to PTH, would be expected to protect against bone loss.

By using continuous infusion of human (1-34) PTH as a skeletal challenge, we have shown an acute increase in circulating levels of biochemical indices of skeletal resorption in both estrogen-treated and untreated osteoporotic women. The increments in all resorption markers were much more modest, however, and reached statistical significance less frequently in the estrogen-treated group. We have no reason to believe that metabolism or excretion of these markers is changed by estrogen. Therefore, because previous work has shown that urine and serum levels of these markers correlate with skeletal turnover rates [35, 36], we postulate that PTH causes a quantitatively smaller increase in overall skeletal turnover in estrogenized bone, possibly due to a decrease in the sensitivity of the skeleton to the effects of PTH. This might be a mechanism by which estrogen prevents further bone loss in established osteoporosis.

Although all resorption and formation markers were somewhat lower in estrogen-treated women in the basal state, only deoxypyridinoline and pyridinoline levels were statistically different. This finding is probably due to small sample sizes and the variance with each measure. The group differences in other variables were unmasked by dynamic challenge similar to subtle abnormalities in other endocrine systems that are unmasked by stimulation or suppression tests [37].

Our data differ somewhat from those of Tsai and colleagues who, using a different protocol, did not show any statistical difference between hydroxyproline excretion among groups of normal premenopausal, normal postmenopausal, and untreated osteoporotic women during a 3-day (1-34) PTH infusion [38]. In fact, no statistically significant increment in hydroxyproline excretion (as a function of creatinine clearance) was found in any group before the third day. Our protocol used on average 50% more PTH (adjusted for body weight) and was administered for only 20 hours.

The responses of resorption and formation markers to PTH differed. Although resorption indices increased more briskly in untreated than estrogen-treated osteoporotic women, the response of bone formation indices was modest at most, and no statistical differences in response were found between the two groups. Parathyroid hormone inhibits or has no effect on alkaline phosphatase activity in osteoblastic cells in vitro [39, 40]. Our in-vivo results, showing a rapid small decrement in alkaline phosphatase, are consistent with the in-vitro studies but differ somewhat from a recent clinical study where a small significant alkaline phosphatase increase was seen in premenopausal (but not postmenopausal) women subjected to 24-hour subcutaneous PTH infusion [20]. The alkaline phosphatase we measured included that from hepatic and other organ synthesis as well as that from bone so the specific skeletal response might have been diluted. A decrease might reflect the nonskeletal enzyme fraction, although we have no documentation for this. A more bone-specific alkaline phosphatase assay would be required to elucidate this issue further.

Osteocalcin has never been shown in-vivo to increase in response to PTH, although at least one in-vitro study showed that PTH induced increased bone Gla protein mRNA [41]. In our study, although bone Gla protein levels fluctuated throughout the 20 hours, the early increase was followed by a return to basal levels. This same pattern was seen in those patients not receiving PTH (data not shown) and more likely reflects an intrinsic diurnal variation in bone Gla protein rather than a response to PTH. Other investigators have shown similar diurnal rhythms in both normal and osteoporotic women [42, 43]. The explanation for the observation of Joborn and colleagues [20], who found a decrease in bone Gla protein after subcutaneous PTH stimulation, remains unclear. In our study, bone Gla protein never diminished significantly from baseline in either group of patients.

Although estrogen is known to increase insulin-like growth factor-1 production from some osteoblastic cells in-vitro [44], we found no substantial difference in basal levels between untreated and estrogen-treated women, perhaps because in the basal state the hepatic fraction contributes most of the serum level [45]. The insulin-like growth factor-1 response to PTH, like that of the formation markers, was not statistically different in untreated compared with estrogen-treated osteoporotic women but did increase significantly in both groups. Because the insulin-like growth factor-1 we measured was a combination of the hepatic and osteoblastic products, it is impossible to know how much of the serum increase with PTH infusion was due to the intrinsic skeletal response. Previous work suggests, however, that PTH does not cause a marked hepatic production of insulin-like growth factor-1, although it definitely increases osteoblast production of insulin-like growth factor-1 [45, 46]. This finding implies that the increase might have been from osteoblastic stimulation.

Of the variables measured in the basal state, deoxypyridinoline, pyridinoline, and tartrate-resistant acid phosphatase correlated with age in our population, consistent with what others have seen with these and other turnover variables [47]. We did not find this age relationship with bone Gla protein, alkaline phosphatase, and hydroxyproline, at least in part due to the limited sample size. Basal resorption indices did not strongly correlate with each other possibly because different stages of resorption are reflected by these biochemical variables. Although increments in resorption variables did not relate well to each other in untreated osteoporotic women or in the two groups combined, they did correlate well in the estrogen-treated group alone. In part, this is due to the much smaller variance with regard to all the variables at almost all time points in the estrogen-treated women.

Although bone Gla protein and alkaline phosphatase were correlated with each other in basal conditions, neither correlated well with resorption indices, possibly because of poor coupling between resorption and formation in these osteoporotic patients. We found no relationships between changes in formation and resorption variables in response to PTH, as expected, because acute PTH infusion has disparate effects on these two processes.

A confounding variable in this study was that the untreated osteoporotic group was older than the estrogen-treated group. If anything, however, this difference might bias against the conclusion we have drawn from the data. Numerous densitometric studies have documented a slowing of skeletal loss with advanced age and longer periods of time past menopause [48], in contrast to the early years after menopause. Despite this, some biochemical data, largely cross-sectional, support an overall increase in skeletal turnover with age [47]. Increases in urinary hydroxyproline and serum alkaline phosphatase levels, however, were not seen in women between the mean ages of 55 and 65 in the study by Delmas and colleagues (similar to the mean ages of the participants in our study), and even bone Gla protein increased only 7.7% over this decade [49]. It is unlikely, therefore, that the difference in basal bone Gla protein that we found between our two groups (36.8%) or the basal differences in the other turnover variables would have resulted from an intrinsic age-related change in skeletal metabolism. Indeed, others have also found estrogen-mediated decreases in markers of skeletal remodeling [48].

Moreover, most endocrine systems respond less as age increases [50]. One example is decreased responsiveness of the 1--hydroxylase to PTH stimulation [51]. One might expect, therefore, that unless estrogen was the dominant factor in modifying the skeletal response to PTH, the older, untreated group of women would have less of a skeletal turnover response than the younger, treated group. Our data, however, showed the opposite finding.

We conclude that the estrogenized osteoporotic skeleton is less sensitive to the acute bone resorbing effects of PTH, while not causing differential effects on markers of bone formation. Whether this is caused by an estrogen-induced reduction in the population of active bone cells, reduction in the activity of a similar number of cells (as in the untreated osteoporotic skeleton), or decreased recruitment of new osteoclasts cannot be determined from our study. Resistance to parathyroid-hormone-induced bone resorption may be one of the mechanisms by which estrogen prevents further bone loss in osteoporosis. If that theory is correct, it provides additional rationale for the theoretic efficacy of parathyroid hormone in combination with estrogen administration in the therapy of osteoporosis, where estrogen might diminish the resorbing effects of parathyroid hormone and allow the anabolic effects to proceed unhindered.