Postmenopausal women with established vertebral osteoporosis were studied for 2 years to determine the terminal elimination half-life and the duration of response to treatment with intravenous alendronate (30 mg) given over 4 days. The urinary excretion of alendronate followed a multiexponential decline. Approximately 50% of the total dose was excreted over the first 5 days, and a further 17% was excreted in the succeeding 6 months. Thereafter, there was a much slower elimination phase with an estimated mean terminal half-life of greater than 10 years (n = 11). Urinary excretion of hydroxyproline and calcium decreased significantly from pretreatment values by day 3, reaching a nadir by 1 week (40% and 67% decrease, respectively). Thereafter, hydroxyproline remained suppressed for the following 2 years. In contrast, urinary calcium excretion returned gradually toward pretreatment values over the first year and during the second year was comparable to pretreatment values. Serum activity of alkaline phosphatase activity decreased over 3 months (23% reduction), increased gradually thereafter, and returned to pretreatment values at month 24. Bone mineral density measured at the spine increased by approximately 5% during the first year and remained significantly higher than pretreatment values at 2 years. We conclude that a short course of high doses of intravenous alendronate is associated with a prolonged skeletal retention of the agent. This open study also suggests that this regimen has a sustained effect on bone turnover persisting for at least 1 year.
BISPHOSPHONATES are stable analogs of pyrophosphate which bind strongly to hydoxyapatite crystals and inhibit osteoclast-mediated bone resorption.1 Their P-C-P structure permits the targeting of the bisphosphonates to skeletal sites, whereas additional structural modifications affect their potency.2 Bisphosphonates have been used with success in the treatment of skeletal disorders characterized by increased bone turnover, including hypercalcaemia of malignancy,1 hyperparathyroidism,3 and Paget's disease of bone.4 They are also used widely for the treatment of osteoporosis.5,6 The pharmacokinetics of bisphosphonates have been partially characterized in man.7–9 Preclinical and experimental studies have shown that after absorption, the bisphosphonates are either deposited in bone or excreted rapidly by the kidneys with little or no accumulation in noncalcified tissues.10,11 They do not appear to be metabolized in vivo, and the fraction which adheres to skeletal surfaces may either be buried during new bone formation or re-enter the circulation following its physicochemical release from bone surfaces (desorption) or release by bone resorption10,11 and subsequently redistributed.
The terminal elimination of bisphosphonates has not been studied in man. Estimates of half-life for various bisphosphonates are reported to be on the order of 1–2 h,8,9,11,12 but these estimates primarily reflect the initial distribution between the skeleton and renal elimination. Experimental studies in rats, mice, and dogs suggest that the terminal elimination of bisphosphonates is prolonged with a half-life of several months.11,13,14 For this reason, it has been suggested that skeletal retention of bisphosphonates might contribute to their prolonged duration of effect when treatment is stopped, for example in the treatment of Paget's disease.4 The aim of the present study was to determine the terminal elimination half-life of alendronate in women with postmenopausal osteoporosis. In addition, we also wished to evaluate the duration of therapeutic response in osteoporosis following a short exposure to high doses of alendronate.
MATERIALS AND METHODS
We studied 21 postmenopausal women with vertebral osteoporosis with a mean age of 66 (range 56–75) years. All patients had a lumbar bone mineral density (BMD) measured by dual X-ray absorptiometry (DXA; Hologic QDR 1000, Hologic, Inc., Waltham, MA, U.S.A.) of more than 2.5 standard deviations below the mean for adult premenopausal women.15 In addition, 12 of the patients had established vertebral deformities as judged by vertebral morphometry on lateral spine radiographs.16 Patients with secondary causes of osteoporosis or who were taking medication likely to affect skeletal metabolism were excluded. The study was approved by the local ethics committee, and all patients gave written informed consent.
Patients were admitted for 1 week to the metabolic bone unit for baseline investigations and initiation of treatment. On each of the first 4 days, alendronate (7.5 mg) was diluted in 500 ml of saline and infused intravenously over 12 h (total dose 30 mg). Patients were examined as out-patients on days 9, 16, and 30 and thereafter on a monthly basis for the next 5 months, two monthly for the next 6 months, and every 3 months for the next year (total follow-up 2 years).
Twenty-four hour urine samples were collected on each of the 7 days following the administration of the first intravenous dose and on the day prior to subsequent out-patient visits to determine the urinary excretion of alendronate. Urine samples were collected in polypropylene, wide-mouth jars containing boric acid as a bacteriostat. All samples were weighed and acidified (5 ml of 6.0 N HCl/200 ml of urine) after collection to prevent the precipitation of alendronate. Aliquots of 50 ml were frozen and stored at −20°C for the subsequent determination of alendronate and creatinine concentration. The concentrations of urinary alendronate were determined by methods previously described with a lower limit of 1 ng/ml for reliable quantitation.17
Fasting blood and urine samples were collected before, during, and after treatment to measure indices of bone turnover and mineral homeostasis. Two-hour fasting urine samples were obtained by voluntary voiding into acidified bottles and analyzed for hydroxyproline,18 calcium, phosphate, and creatinine. Both fasting urinary calcium and hydroxyproline were expressed as a ratio of creatinine excretion measured on the same sample.
Serum was analyzed for calcium, phosphate, alkaline phosphatase (ALP) activity, creatinine, transaminases, urea, and electrolytes using a Technicon multiple channel analysis (SMAC) (Technicon Corp., Tarrytown, NJ, U.S.A.). Serum calcium was corrected for fluctuations in serum albumin concentration.19 Aliquots of serum on days 1, 5, 8, 17, and months 1 and 2 were frozen for the subsequent assay of osteocalcin (between batch CV of 13%),20 intact parathyroid hormone (Nichols Allegro intact parathyroid hormone [PTH] assay, San Juan Capristrano, CA, U.S.A.; between batch CV of 3.9% at 34 pg/ml and 12.4% at 251 pg/ml) and 1,25-dihydroxyvitamin D (1,25(OH)2D) assay.21 Serum 25-hydroxyvitamin D (25(OH)D) was assayed before and 2 months after treatment.21
Lumbar BMD measurements were performed by DXA and were repeated at 6, 12, and 24 months from the start of treatment (Hologic QDR 1000). Seven patients underwent a transiliac bone biopsy after double tetracycline labeling 15 months into the study using a Sheffield trephine (Bolton Surgical Instruments, Sheffield, U.K.).
Alendronate excreted over the first week was assessed directly from the urine collections in each patient. Thereafter, excretion was calculated from the area under the urine concentration-time curve by trapezoidal integration. The skeletal retention of alendronate was assumed to be equal to the total dose administered minus the urinary excretion.11
Terminal elimination half-life (t1/2z) was calculated in each patient by log linear regression of the percentage retained versus time curve between days 240 and 540 and substituting the slope of the regression line into the equation, t1/2 = −log 2/slope.
Intermediate half-lives (t1/2e) were calculated by stripping the residual retention (observed-asymtotic value) and replotting the values on a logarithmic scale.22 From this plot, a linear phase over the next time interval was identified and the half-life calculated as before. This procedure was repeated up to and including data on day 4 of the study.
A separate estimate of terminal half-life was undertaken for each patient from the sums of exponentials computed from the plot of urinary excretion rates with time to estimate the cumulative excretion of alendronate at 240 days. Half-life was then calculated by averaging the excretion rates thereafter (the regression method).
Several of the urine samples collected after the first 8 months of the study had alendronate concentrations which either fell below the validated limit of the assay (1 ng/ml) or were lower than the limit of detection. In order to estimate of the amount excreted, those samples in which alendronate was detected but values were below this limit a concentration was determined by extrapolation below the standard curve. Those samples below the detection limit were either assigned a value of zero or, where alendronate was detected in subsequent samples, estimated by taking the average of the preceding and subsequent urinary alendronate concentration.
For serum ALP, osteocalcin, PTH, and urinary hydroxyproline and calcium, data were log-transformed to normalize the distributions before statistical analysis. Sequential changes from pretreatment values were assessed using Student's paired t-test with Bonferroni adjustment for multiple comparisons. The results are expressed graphically as the mean (SEM) percentage change from pretreatment. The results in Tables 1 and 2 are antilogged values (i.e., geometric mean and 95% confidence intervals) for all available data.
Table Table 2. CUMULATIVE EXCRETION OF ALENDRONATE OVER THE FIRST 540 DAYS FOLLOWING INTRAVENOUS INFUSION OF 30 MG OF ALENDRONATE
Table Table 1. CHARACTERISTICS OF PATIENTS WITH POSTMENOPAUSAL OSTEOPOROSIS TREATED WITH INTRAVENOUS ALENDRONATE
Of the 21 patients enrolled, three patients were withdrawn from the study for reasons not attributed to treatment with alendronate. One patient experienced a deterioration of her established Parkinson's disease on the fourth day of the study, and a further two patients discontinued the study following elective surgery 4 and 16 months into the study. In assessing the change from pretreatment values, pharmacodynamic data from these latter two patients were used before their withdrawal. Seven of the 18 patients completing the study were excluded from the pharmacokinetic calculations because of incomplete urine collections, due to an administrative error during their in-patient stay.
Oral temperature increased in eight patients. The highest change in temperature seen was 2.9°C. On average temperature increased by 0.05°F from pretreatment values on day 4 (p < 0.05), and nine of the patients experienced one or more mild to moderate symptoms including pain, aches, chills, malaise, and neck stiffness. All patients recovered with no residual symptoms.
The 11 patients in whom a pharmacokinetic evaluation was undertaken did not differ significantly in their pretreatment characteristics from the other nine patients (Table 1).
The cumulative urinary excretion of alendronate is shown in Fig. 1. After the first infusion of alendronate, 44% of the dose given (95% CI = 38–51%) was excreted before the subsequent infusion. Two days following the last infusion 52% (48–56%) of the total dose given had been excreted. By 18 months the total excretion rose to 70% (65–75%). Thus, following the rapid uptake of alendronate after the first infusion, approximately 50% of the total drug administered was retained in the first week after the start of treatment. The total retained decreased thereafter by approximately 17% over the next 6 months. Over the next 12 months, whole body retention of alendronate decreased by 2.5% (Fig. 1) so that, at the end of 18 months, approximately 30% (25–35%) of the administered dose was unaccounted for by urinary excretion (Table 2).
The urinary excretion of alendronate approximated a monoexponential decline from the 240th day onward. For each of the patients over this period, there was a significant and close negative linear correlation between the logarithmic values of the residual retention vs time (0.981 < r < 0.998; Table 2). The terminal elimination half-life calculated from the mean residual whole body retention was 10.5 years (95% CI = 7.9–13.2 years). The estimation of terminal half-life by the regression method was similar (mean 10.9 years, range 5.4–19 years).
The urinary excretion profile of alendronate before day 240 followed a multiexponential pattern with three phases identifiable after stripping the terminal retention (Fig. 1). The half-life for the first three phases were calculated as 0.80 days (days 4–7), 6.6 days (days 9–16), and 35.6 days (days 30–180).
Five patients with a terminal half-life of less than 10 years were compared with the other six patients. There was no significant difference in their pretreatment characteristics nor evidence of any differences in the duration or extent of the pharmacodynamic response to alendronate.
Mean urinary hydroxyproline excretion decreased significantly from pretreatment values by day 3 (29–21 μmol/mmol creatinine, p < 0.05), and were maximally suppressed at 1 week (by 40%). Thereafter, following a transient rebound, the mean hydroxyproline excretion remained significantly below pretreatment values for 2 years after the start of treatment (18 μmol/mmol creatinine, p < 0.001; Fig. 2). A similar response to treatment was observed in the urinary calcium excretion in that a significant reduction from pretreatment values was noted by day 3 (0.34–0.13 mol/mol creatinine). The decrease in urinary calcium excretion was maximal on day 8 (0.34–0.11 mol/mol creatinine, p < 0.001) and gradually increased toward pretreatment values over the first year. Between 12 and 24 months, fasting urinary calcium excretion was comparable to pretreatment values (Fig. 2).
A significant reduction in the serum calcium was observed on day 3 (2.34–2.23 mmol/l, p < 0.001; Fig. 3) and the minimum values attained by day 5 (2.18 mmol/l, 7% decrease from pretreatment values). By month 4, serum calcium values had returned to pretreatment values and thereafter remained constant. Nine patients had asymptomatic hypocalcaemia (range 1.89–2.11 mmol/l) on one or more occasions between days 3 and 10 after the start of treatment. Hypocalcamia in one patient persisted up to month 3. A significant reduction in serum phosphate was observed on day 3 (1.13–0.96 mmol/l, p < 0.001; Fig. 3) maximal on day 4 (0.87 mmol/l, mean decrease 21% from pretreatment) with a return to pretreatment values by month 2 (Fig. 3).
Serum PTH increased significantly from pretreatment values, coincidently with the fall in serum calcium reaching a maximal value on day 5 (30.2–64.2 pg/ml; 113% increase) which persisted up to and including month 2 (47.6 pg/ml; Fig. 3). A significant mean increase in serum 1,25(OH)2D was noted on days 5 and 8, respectively (p < 0.05), but values returned to pretreatment values by day 17. Serum 25-(OH)D concentrations assayed at month 2 did not differ from pretreatment values.
In contrast to the changes in fasting urinary hydroxyproline excretion, a progressive increase in the serum ALP activity was observed from day 5 and was maximally increased on day 10 by 14 ± 4% (Fig. 2). Thereafter, serum activity fell, and at 3 months values were (23 ± 2%) lower than pretreatment values. Thereafter, activity rose toward pretreatment values. A similar pattern of response was observed in serum osteocalcin (Fig. 3). Mean values increased from pretreatment values on day 8, but this was not statistically significant (p = 0.06). Thereafter, a significant decrease from pretreatment values was observed.
BMD measured at the spine showed a nonlinear increase of 5.3 ± 1.1% (p = 0.05) during the first year of the study (3.3 ± 0.7%; p < 0.05) during the first 6 months and 2.0% during the second 6 months). BMD measured in seven patients after the second year was significantly higher than pretreatment values.
Bone biopsies taken 15 months after the infusions showed reduced cancellous bone volume and thinning of cortical bone consistent with the diagnosis of osteoporosis. Osteoid surfaces were sparse but present in all biopsies with active-looking (plump) osteoblasts. Osteoid seam width and the number of osteoid lamellae were normal. Double tetracycline labels were clearly visible and mineralization fronts were normal in appearance.
This is the first study to evaluate the terminal elimination kinetics of any bisphosphonate in man. Our study shows that the acute whole body retention of alendronate is high and that the apparent terminal elimination half-life of alendronate is on the order of 10 years in patients with postmenopausal osteoporosis. These data are consistent with experimental studies with alendronate which estimated the terminal half-life to be 200 days in rats and 1000 days in dogs11,23 when differences in the relative rates of bone turnover are taken into account.
The pharmacokinetics of bisphosphonates are complex. Bisphosphonates are rapidly cleared from the plasma8,11,12 related to the avid skeletal uptake of the drug, and this skeletal retention accounts for the fraction not appearing in the urine.9–11 Skeletal uptake of bisphosphonates accounts for their therapeutic action, but the skeletal distribution is not homogenous, and significantly higher concentrations are found in cancellous bone compared with cortical sites,23 most likely related to the greater blood flow and metabolic activity in cancellous bone. In addition, the amount of alendronate retained by bone is likely to be related to the prevailing rate of skeletal turnover as shown for other bisphosphonates13,14 where increased bone turnover is associated with higher skeletal retention and a correspondingly lower urinary excretion. This property of bisphosphonates forms the basis of bone scanning, and whole body bisphosphonate retention has been used for monitoring bone metabolism.24
Several studies suggest that bisphosphonates are taken up largely at sites of bone formation,25,26 but alendronate has been shown to be localized preferentially to sites of bone resorption.27 It seems likely that the mineral at resorption sites is bound with high affinity but readily saturated and that excess amounts are bound at additional sites of accessible mineral such as at the mineralizing surface but with less affinity.28 For this reason, low doses of bisphosphonates are likely to have a higher fractional retention in the skeleton. This phenomenom may account for the high skeletal retention that we observed in our patients (very nearly 50% 2 days after the last infusion and 30% at 18 months) compared with the much lower acute retention (approximately 25%) with the less potent bisphosphonates where much larger doses were used.14,26,29 The effects of disease activity and disease type on the disposition of alendronate have not yet been studied, but compared with less potent agents, retention of alendronate is more likely to be influenced by differences in resorption rates rather than by differences in rates of bone formation.
The turnover time of bone in man has been estimated to be approximately 11 years at cancellous sites and 40 years in cortical bone, although it varies from site to site and even within sites.4 Since the terminal elimination half-life of alendronate was approximately 10 years, this suggests that the half-life measured is largely dependent on bone remodeling at cancellous sites. This finding is consistent with the view that the selective distribution of bisphosphonates within the skeleton is dependent upon the prevailing rate of bone turnover.1,7,23 In reality the situation is likely to be more complex for several reasons. First, bisphosphonates are also taken up by cortical bone, and the terminal elimination at cortical sites is likely to be much longer than that measured in the present study. Indeed estimates of half-life vary according to the length of observation.9 A secure definition of a terminal half-life would require sample collection over three or four half-lives, which is not feasible in humans. Second, the turnover of bone is a surface-based phenomenon. Thus, the likelihood that surface bound bisphosphonate would be released by bone resorption is higher than that buried deep within bony trabeculae. Also, there may be significant desorption of alendronate from the resorption surface. Both phenomena may account for the intermediate half-lives that we observed (7 and 36 days). Finally, the doses of alendronate that we used clearly had pharmacological activity in that bone turnover was substantially decreased, so that the activity of the agent on bone is also likely to affect its kinetics.
The pharmacodynamic responses that we observed are generally consistent with a decrease in the turnover of bone. Indices of bone resorption (hydroxyproline and calcium excretion) decreased within days of administration, which was followed later by a decrease in indices of osteoblast function such as osteocalcin and ALP. This type of response, characteristic of all inhibitors of bone resorption,4 is thought to be due to the preservation of “coupling” of bone formation at sites of bone resorption. This transient imbalance between bone resorption and formation is thought to be responsible for the early increase in bone mass following treatment. Subsequently, a new lower steady state of bone turnover is attained. This may take up to 3 years since bone turnover is a slow process.30,31 It is of interest that serum activity of ALP rose in the few days following treatment at a time when hydroxyproline excretion decreased. This phenomenon has also been observed with pamidronate32 and may be due to increased matrix synthesis within resorption sites.
Several of the other metabolic responses that we observed are expected consequences of a primary effect of alendronate on bone resorption. These include the decrease in serum calcium, the rise in PTH and calcitriol, and the fall in serum phosphate likely due to secondary hyperparathyroidism. Such responses are well characterized with most bisphosphonates which do not additionally impair the mineralization of bone.33 The return of fasting urinary calcium excretion after 1 year to pretreatment values, but continued suppression of urinary hydroxyproline, is also an expected effect of the sustained inhibition of bone resorption when bone formation (and thus net skeletal calcium efflux) is also decreased.
It is of interest that both serum osteocalcin and activity of ALP increased over the first 10 days of observation. The initial rise in serum ALP activity has not been described previously in osteoporosis but may be due to the secondary hyperparathyroidism, or the increase in synthesis in calcitriol, both of which modulate the activity of osteoblasts in vitro.34 However, it is possible that the bisphosphonates might also directly affect osteoblast metabolism.35
Although the lack of a control group limits conclusions on the duration of the therapeutic response, the significant changes observed in the biochemical indices were greater than the long-term coefficient of variation for each variable.36 Exposure to a short course of alendronate in high doses thus appears to induce a prolonged effect on bone resorption, which remained significantly suppressed below pretreatment values for at least 1 year and probably 2 years following treatment. The effects on BMD would support this view. It is of interest that, assuming oral absorption of 0.7% for alendronate,37 the systemic dose that we gave would be equivalent to a treatment of more than 400 days with an oral dose of 10 mg daily. The effects of oral treatment on bone resorption are similar,38,39 suggesting that the pharmacodynamic response over the first year may be dependent on the total dose given rather than the duration of exposure.
It has been suggested that the relatively long skeletal half-life of bisphosphonates may account for the long duration of response in Paget's disease of bone and osteoporosis.4,40 This seems unlikely since in disorders where skeletal retention is increased, such as hypercalcaemia and metastatic bone disease, the rate of reversal of the therapeutic effect is relatively rapid after stopping treatment.41,42 For these reasons, it seems unlikely that alendronate buried in mineralized bone is of clinical importance. Unfortunately, we could not address this issue in the present study since the pharmacodynamic response persisted for the duration of study which was of shorter duration than the estimated skeletal half-life. The relationship between the pharmacokinetic and pharmacodynamic response following bisphosphonate treatment still remains to be fully characterized. Notwithstanding, the prolonged response that we and others have shown40 following short exposures with high doses of alendronate poses challenges for the optimization of treatment regimens with bisphosphonates in osteoporosis.
This work was supported by Merck Research Laboratories, Rahway, NJ, U.S.A.