Differential Effects of Teriparatide on BMD After Treatment With Raloxifene or Alendronate


  • Interim data presented previously in abstract form at the 24th Annual Meeting of the American Soceity for Bone and Mineral Research, San Antonio, Texas, USA, September 20-24, 2002, and the 25th Annual Meeting of the American Society for Bone and Mineral Research, Minneapolis, Minnesota, USA, September 19-23, 2003. Dr Ettinger has received grants from Merck and has received grants and served as a consultant for Berlex, Eli Lilly and Company, and Procter & Gamble. Drs San Martin, Crans, and Pavo are employees of Eli Lilly and Company.


We investigated the effects of 18 months of treatment with teriparatide in patients previously treated with long-term antiresorptive therapy using bone turnover markers and bone densitometry. Previous raloxifene treatment allowed for teriparatide-induced early bone marker and BMD increases comparable with previously published results for treatment-naïve patients. Conversely, previous alendronate treatment reduced the bone marker and BMD response.

Introduction: Teriparatide [rhPTH(1–34)] has been shown to increase BMD and reduce the risk of fracture in postmenopausal women with osteoporosis. Our objective was to investigate the skeletal effects of 18 months of treatment with teriparatide in women whose osteoporosis was previously treated with either alendronate or raloxifene.

Materials and Methods: Daily subcutaneous injections of 20 μg teriparatide were administered for 18 months to 59 postmenopausal women, 60–87 years of age, with BMD T-scores ≤ −2.0 who had previously received either alendronate (ALN) or raloxifene (RLX) therapy for 18–36 months. All patients received daily calcium (1000 mg) and vitamin D (400 IU) supplementation. The primary study outcome was change in lumbar spine BMD measured by DXA. Secondary outcomes included changes in bone turnover markers, total hip BMD, and safety.

Results: Median baseline bone turnover marker levels in prior ALN patients were about one-half those of prior RLX patients. During teriparatide treatment, bone markers in prior ALN patients increased later and peaked at about one-third lower levels compared with prior RLX patients. During the first 6 months, there were statistically significant (p < 0.05) group differences in BMD change at the hip (prior ALN −1.8% versus prior RLX +0.5%) and at the spine (prior ALN +0.5% versus prior RLX +5.2%). The positive slopes in hip and lumbar spine BMD were similar in both groups between 6 and 18 months. After 18 months, mean lumbar spine BMD increased 10.2% in prior RLX compared with 4.1% in prior ALN (p < 0.05) patients. Furthermore, at 18 months, mean total hip BMD had significantly increased (1.8%, p < 0.05) in prior RLX patients but was not different from baseline in prior ALN patients.

Conclusions: Teriparatide treatment stimulates bone turnover in patients pretreated with both RLX and ALN. Prior treatment with RLX allows for the expected teriparatide-induced BMD increases comparable with those previously reported for treatment-naïve patients. In contrast, prior treatment with ALN prevents increases in BMD, particularly in the first 6 months.


Subcutaneous daily injection of parathyroid hormone (PTH) has been shown to increase bone formation, resulting in rapid increases in bone mass(1–4) and reducing fracture risk in women with osteoporosis.(1,2) PTH therapy yields sizable increases in markers of bone turnover and measures of BMD when added to ongoing estrogen therapy.(2,3,5) However, in contrast to preliminary short-term findings,(6) when PTH was administered for 1 year concomitantly with alendronate (ALN) in treatment-naïve patients(7) or for 2 years after 6-month ALN pretreatment,(8) the expected bone formation marker response was absent.

Antiresorptive therapies have long been the mainstay of osteoporosis treatment, and in 2002, 34 million prescriptions were written for the bisphosphonate raloxifene (RLX) and calcitonin in the United States.(9) Some patients are being considered for teriparatide [rhPTH(1-34)] treatment either because they remain severely osteoporotic or because they have failed antiresorptive therapy. It is not known whether teriparatide will yield skeletal benefits after long-term use of antiresorptive therapy, particularly ALN, a potent bisphosphonate. ALN markedly suppresses bone activation frequency,(10) has long retention time in the bone,(11) and suppresses bone turnover for many months after discontinuation.(12) The overall impact of these effects raises questions when switching patients from ALN to PTH therapy. Therefore, we sought to determine the effects of PTH therapy on bone turnover and BMD in women with osteoporosis who were previously treated with either ALN or RLX therapy for 18-36 months. We hypothesized that prior exposure to ALN would retard the skeletal response to PTH, whereas prior exposure to RLX would not.


Design and participants

This study was a prospective, open label, nonrandomized, single-armed 12-month trial, with a 6-month extension phase, examining the effects of rhPTH(1-34) treatment in patients previously treated with either ALN or RLX. The study was conducted at the Division of Research, Kaiser Permanente Medical Care Program, Oakland, CA. The trial was funded by Eli Lilly and Company (Indianapolis, IN, USA). Participants were women 60-87 years of age with a prior lumbar spine or total hip BMD measurement of ≤ −2.5 T-score, who regularly (using >70% of the pills prescribed) had used either ALN 10 mg/day or RLX 60 mg/day for 18-36 months before study entry. All participants were required to have a lumbar spine or total hip BMD T-score of ≤ −2.0 on entry. Administrative databases maintained by Kaiser Permanente were used to select participants who fit inclusion and exclusion criteria. Women were excluded if they had a history of metabolic bone disease, cancer (except non-melanoma skin cancer), coronary disease, stroke or transient ischemic attack, venous thromboembolism, uncontrolled hypertension, uncontrolled thyroid disease, liver disease, fasting triglycerides >300 mg/dl, fasting glucose >180 mg/dl, or had taken estrogen or progestin within 3 months of randomization. The Institutional Review Board approved the protocol, and informed consent was obtained from all participants.

Intervention and blinding

All women received daily self-administered subcutaneous injections of teriparatide (TPTD) 20 μg in addition to daily calcium (1000 mg) and vitamin D (400 IU). Teriparatide was provided in automatic injection pen devices that held a 4-week supply of the drug. Bone turnover markers and bone densitometry measurements were made by technicians blinded to the type of previous antiresorptive therapy.

Baseline and follow-up measures

At baseline, we obtained information on demographics, health history, and medication use. Fasting blood tests included hematology, routine chemistry, and serum markers of bone turnover; the latter consisted of osteocalcin (ELISA; CIS Biointernational, Gifsur Yvetter, France), bone-specific alkaline phosphatase (Bone ALP; immunoradiometric assay, Tandem-R Ostase; Hybritech, San Diego, CA, USA), N-propeptide of type I procollagen (PINP; Orion Diagnostica, Espoo, Finland), and N-telopeptide of collagen (NTX; Osteomark; Ostex International, Seattle, WA, USA).

BMD was measured using DXA of the L2-L4 lumbar spine and total hip. A single Hologic (Watham, MA, USA) model 2000-plus densitometer was used throughout the study. Daily scanning of a phantom showed absence of machine drift during the study. In vivo precision data based on repeated scans of 20 volunteers done 1-6 weeks apart and expressed as CVs were 1.4% for posterior-anterior spine and 2.2% for total hip.

Follow-up visits occurred 1, 3, 6, 12, and 18 months after the start of teriparatide therapy. Serum biochemical tests were obtained at all visits, whereas BMD was measured at all but the 1-month visit. Compliance was assessed by measuring the amount of teriparatide remaining in used and unused pen devices.


The primary study outcome was the change from baseline in lumbar spine BMD after 12 months of follow-up. An increase of ∼3% in each pretreatment group was the expected response to teriparatide treatment. Secondary outcomes included 12-month changes from baseline in total hip BMD in each of the four serum markers of bone turnover and comparisons of all these outcome variables between the two study groups. In addition, secondary outcomes included all these comparisons during the 6-month extension phase.

Statistical analyses

The primary comparison in this study was the mean difference in lumbar spine BMD from baseline to 12 months, calculated separately for each antiresorptive pretreatment group. The study was designed to enroll 60 patients, with ∼30 patients from each pretreatment group. This study design provides, using a one-sided paired t-test with a significance level of 0.05, at least 80% power to detect a mean within-group difference of 0.0225 g/cm2 (∼3%), assuming an SD of 0.043 g/cm2(1) and a 20% dropout rate.

Baseline characteristics of the two pretreatment groups were compared using Pearson's χ2 test for categorical variables and Student's t-test for continuous variables. For each pretreatment group, the statistical significance of changes and percentage changes from baseline was determined using the paired t-test for BMD and the Wilcoxon signed-rank test for markers of bone turnover markers. Differences in BMD response were calculated using analysis of covariance (ANCOVA), with pretreatment drug as the factor and the corresponding baseline BMD measurement as the covariate. Similar analyses were performed for bone markers, with additional ranked analyses performed to ensure the robustness of results. Additional exploratory analyses of bone turnover marker data were performed using a derived area under the curve (AUC) variable, representing the net gain above baseline level over time (see Appendix 1). Between-group comparisons were performed for the pretreatment groups, as well as an investigation of the BMD and AUC relationships using correlation analyses. The primary analysis was by intention-to-treat, without regard to adherence. We computed outcomes for women with missing measurements by carrying forward the most recent postenrollment measurement (last observation carried forward). We also performed analyses based on those completing 6-, 12-, and 18-month measurements. All serious adverse events and all episodes of hypercalcemia (>11.0 mg/dl) are reported.


During January-June 2001, 73 women visited the research center and were screened for eligibility; 10 women in the prior ALN group and 4 women in the prior RLX group were found to be ineligible (Fig. 1). Thus, a total of 59 eligible women were assigned to treatment with teriparatide.

Figure FIG. 1.

Recruitment and follow-up.

The mean ± SD age of participants was 70.1 ± 6.8 years, and mean BMD T-score at the lumbar spine and total hip was −2.4 and −2.2, respectively. With the exception of bone turnover markers, there were no statistically significant differences in clinical variables between treatment groups at study entry (Table 1). At baseline, the median levels of bone turnover markers in the prior ALN group were about one-half those observed in the prior RLX group (Fig. 2).

Figure FIG. 2.

Median values of serum bone turnover markers, by previous treatment with ALN or RLX: (A) osteocalcin, (B) N-propeptide of type I pro-collagen, (C) bone-specific alkaline phosphatase, (D) N-telopeptide of collagen. Prior RLX (•); prior ALN (▾). Error bars indicate 25-75% interquartile range. *p < 0.05 change from baseline; †p < 0.05 difference between groups.

Table Table 1.. Baseline Characteristics of 59 Women Enrolled During Study Stratified by Prior Antiresorptive Therapy
original image

Of 59 enrolled subjects, 58 (98.3%) completed 12 months of treatment, and 50 continued in the 6-month extension phase. Of the 50 subjects who elected to continue past 12 months, 48 (96%) completed the additional 6 months of teriparatide treatment (Fig. 1). Among women who enrolled in the study, pen measurements revealed that 93% of the women used at least 75% of the expected amount of teriparatide.

After 1 month of teriparatide treatment, both prior RLX and prior ALN groups showed statistically significant increases in median osteocalcin, PINP, and bone ALP (Fig. 2). There was a consistent trend among all bone turnover markers in prior RLX patients to show greater early increases and to remain about one-third higher during the entire 18 months than prior ALN patients, but the only statistically significant differences between prior treatment groups were at the 1-month observations for bone ALP, osteocalcin, and PINP. Median change in bone turnover markers was similar at 3, 6, 12, and 18 months. Median bone turnover markers reached a plateau at 6-12 months for both the prior ALN and prior RLX groups. Although the AUC was greater for the prior RLX group across all markers, only osteocalcin showed a statistically significant difference between groups (p = 0.01).

At 3 and 6 months of follow-up, lumbar spine BMD increased 2.1% and 5.2% relative to baseline in the prior RLX group, respectively (Fig. 3). However, the lumbar spine BMD in the prior ALN group remained close to baseline values at these time points. From 6 months onward, increases in BMD were similar between groups. That is, if we consider the 6-month time point as the beginning of the observation period, there were no statistically significant between-group differences in lumbar spine or total hip BMD at the 12- and 18-month time points. At 18 months, a 10.2% increase in lumbar spine BMD was observed for the prior RLX group, whereas the prior ALN group showed only a 4.1% increase (between-group difference, 6.1%; p < 0.001). Early changes in total hip BMD also markedly differed between groups. The mean change of total hip BMD at 6 months remained close to baseline in the prior RLX group (0.5%), whereas it decreased significantly in the prior ALN group (−1.8%, p = 0.002). Between 6 and 18 months, both groups showed similar mean hip BMD increases (∼1.5%). Within-group absolute change at 12 months in spine BMD was 0.06 g/cm2 (p < 0.0001) in the prior RLX group and 0.02 g/cm2 (p = 0.006) in the prior ALN group.

Figure FIG. 3.

Mean percent change in BMD from baseline for (A) lumbar spine and (B) total hip, by previous treatment with RLX or ALN. Prior RLX (•); prior ALN (▾). Error bars indicate ±SD. *p < 0.05 change from baseline; †p < 0.05 difference between groups.

We investigated the relationship between 12-month spine BMD percentage change and levels of bone turnover markers separately for each prior treatment group. For the prior RLX group, there were statistically significant correlations between 12-month spinal BMD percentage change and both baseline and AUC bone turnover markers; baseline osteocalcin, PINP, and bone ALP showed correlation coefficients of 0.43, 0.44, and 0.40, respectively, and AUC osteocalcin, AUC PINP, and AUC NTX showed coefficients of 0.59, 0.58, and 0.46, respectively. Conversely, in the prior ALN group, no statistically significant correlations with spinal BMD change were observed between baseline and AUC bone turnover markers.

Serious adverse events were reported by nine (15.3%) women: six (18.1%) in the prior ALN group and three (9.1%) in the prior RLX group. Hypercalcemia developed in three (9.1%) women in the prior RLX group, but all normalized without altering treatment. In the prior ALN group, four (12.1%) women developed hypercalcemia, but all normalized without altering treatment. No adverse treatment effects were observed in the other hematology or chemistry tests. There were no differences in the group comparisons for bone density at the lumbar spine or hip when the results were repeated with data restricted to per-protocol completers.


Our results show that patients previously treated with ALN or RLX respond to teriparatide. Statistically significant within-group increases were observed in both pretreatment groups for all bone markers at all time points during the study. With the exception of the 1-month values, there were no between-group differences in bone marker changes from baseline. The 200-300% increases in markers of bone formation and resorption we observed are comparable with previously published results for treatment-naïve patients.(13) These findings are in agreement with the evidence from animal studies that stimulation of bone turnover markers by PTH(1-34) was unaffected by pretreatment with ALN or RLX.(14) However, bone turnover markers did not increase when PTH and ALN were given concomitantly to postmenopausal women(7) and to men with osteoporosis.(8)

Within 1 month of teriparatide treatment, the prior RLX group experienced a marked increase in bone formation markers; a pattern observed in treatment-naïve patients.(13) While some increase in bone resorption could be expected after withdrawal of the antiresorptive effects of RLX, we found that NTX, the marker of bone resorption, showed much smaller increases during this time than the markers of formation. Furthermore, with the ∼30% average reduction in bone turnover achieved with RLX,(15) one would expect only a relatively small increase in bone turnover markers resulting from its withdrawal. The magnitude of the initial increase of bone formation markers was significantly less in prior ALN patients. Previous studies have shown that after discontinuation of long-term ALN therapy, bone turnover remains suppressed for at least 1 year.(12) We suspect the difference in early bone marker responses between prior ALN and prior RLX treatment is related to availability of target cells for the bone formation effect of teriparatide. After 2-3 years of ALN treatment, histomorphometric analyses of bone biopsy specimens show that bone activation frequency and mineralizing surfaces are reduced ∼90% from baseline.(10) Very low bone turnover reduces availability of osteoblasts, whose life span is prolonged by PTH.(16) Furthermore, low bone turnover could reduce availability of marrow lining cells(17) and preosteoblasts(18) that could be converted to active osteoblasts. Another possibility is that ALN reduces the PTH-induced increase in bone resorption, thereby preventing the osteoclast-dependent release of bone-formation stimulating cytokines. In contrast, RLX treatment, by causing less suppression of bone turnover,(19) may allow target cells to be available for the initial anabolic response to PTH.

The linearity of spine and hip BMD increases are similar in prior RLX patients to those observed in the Fracture Prevention Trial when teriparatide was given to treatment-naïve patients. In that study, after a median of 19 months of treatment, changes from baseline were 9.7% and 2.8% at the spine and hip, respectively.(1) In marked contrast, we found that prior ALN treatment inhibited the early increase in spinal BMD and was associated with an early decrease in BMD at the hip, with subsequent reversal of this trend. Similar biphasic patterns of hip BMD change have been reported when PTH(1-34) treatment was given to premenopausal women with endometriosis receiving nafarelin acetate,(20) to men with low bone density,(8) and when PTH(1-84) was given to women with postmenopausal osteoporosis.(21) The hip BMD results of the prior ALN group are different from those reported when PTH(1-84) and ALN were administered concomitantly. In the PaTH study, combination therapy with PTH(1-84) and ALN produced increases in hip BMD equal or greater than PTH(1-84) alone.(7)

The reason for the hip biphasic BMD response in ALN patients in the current study is not readily apparent, but 2 hypotheses can be suggested. First, the early anabolic effect of teriparatide could be diminished. Second, alendronate pretreatment is likely to have resulted in bone with little remodeling space and a high degree of mineralization.(10,22,23) Activation of new bone metabolic units by teriparatide reopens the closed remodeling space and causes bone with a high degree of mineralization to be removed and replaced with new, less mineralized bone. This process could explain the conflicting findings of decreasing hip BMD as bone turnover markers are increasing. This latter hypothesis is supported by the finding that, after PTH pretreatment, 1 year of alendronate treatment results in a substantial increase in BMD,(24) probably because of closure of the remodeling space and extensive secondary mineralization of bone that forms under the influence of parathyroid hormone. We suspect that measurement of BMD by DXA will underestimate the amount of new bone made by teriparatide until secondary mineralization concludes.(25,26)

In prior RLX patients, the AUC of bone formation markers correlated significantly with the increase in spinal BMD, but no such correlations were found among prior ALN patients. This lack of relation between bone activation and DXA-measured BMD in the prior ALN group supports the concept that change in remodeling space and mineralization degree are responsible for obscuring the true anabolic effects of teriparatide. Based on our findings, when assessing the teriparatide response in patients previously treated with ALN, DXA measurements performed within the first year may not provide useful clinical information. Instead measurements of bone turnover markers may be appropriate in assessing skeletal response. QCT may be a more accurate method for determining increase in volumetric bone density under PTH treatment. However, there are no clinical data to support a correlation between treatment-induced increases in BMD measured by QCT and fracture risk reduction.

Several strengths of our study deserve mention. We enrolled women of comparable age and within a narrow duration window of prior antiresorptive therapy. We measured outcome variables frequently, using one bone densitometer and a standardized reference laboratory. We obtained high compliance with teriparatide injections and follow-up measures. The study was adequately powered to show BMD changes from baseline.

Limitations of our study should be noted. We did not include a placebo or control group. Selection to antiresorptive therapies before our study was not random. Indeed, consistent with general prescribing patterns, women in our study at the time they were initially prescribed ALN tended to be a few years older and had somewhat lower T-scores than those begun on RLX. However, recent data suggest that skeletal response to teriparatide is independent of age, BMD, and presence of vertebral fractures in postmenopausal women with osteoporosis.(27) Finally, a longer treatment period with teriparatide could reveal important differences. In the 24-month treatment study reported by Finkelstein et al.,(8) most of the gain in hip BMD occurred between 12 and 24 months.

In conclusion, teriparatide treatment in postmenopausal women with osteoporosis previously treated with ALN produces less than expected bone marker and BMD responses, especially in the first months of treatment. Women previously treated with RLX show increases in bone turnover markers and BMD similar to those observed in teriparatide treatment-naïve patients from other trials. ALN pretreatment alters the usual time course and relationships between teriparatide stimulation of bone formation markers and changes in BMD.


We thank Kelli Pacuch and Karin Goldman, RN, for assistance in study implementation and site management. In addition, we thank Emmett Glass, PhD, for editorial assistance in the preparation of the manuscript and Mary Ellen Perron, BS, in the preparation of the figures. Funding was provided by Eli Lilly and Company.


The AUC for a given subject, which represents the net bone marker area gain from baseline, was defined as follows:

equation image


equation image

ti = ith time point, yi = bone marker value at the ith time point, and i = 0, 1, 2, 3, 4 corresponds to baseline and 1, 3, 6, and 12 months, respectively.