Effect of Stopping Risedronate after Long-Term Treatment on Bone Turnover

Abstract

Context:

Determining how quickly bisphosphonate treatment effects begin to regress is crucial when considering termination of treatment.

Objective:

Our objective was to assess the effects of 1 yr discontinuation of risedronate use in postmenopausal women with osteoporosis who had previously received risedronate for 2 or 7 yr.

Design and Setting:

Before initiation of the current study, placebo/5-mg-risedronate patients had received placebo for 5 yr and risedronate for 2 yr, whereas 5-mg-risedronate patients had received risedronate for a total of 7 yr. Risedronate was then discontinued for 1 yr (yr 8).

Patients:

Postmenopausal women with osteoporosis who had previously completed the 3-yr Vertebral Efficacy with Risedronate Therapy MultiNational (VERT-MN) pivotal trial, plus a 2-yr extension comparing risedronate or placebo for a total of 5 yr, followed by 2 yr of open-label risedronate treatment were enrolled in these trial extensions.

Main Outcome Measures:

Evaluations included changes in type I collagen cross-linked N-telopeptide (NTX)/creatinine (Cr) and bone mineral density (BMD) values, fracture incidence, and adverse events.

Results:

After 1 yr of risedronate discontinuation, NTX/Cr levels increased toward baseline in both patient groups vs. the values at the end of yr 7. In both treatment groups, off-treatment total hip and femoral trochanter BMD values decreased, whereas lumbar spine and femoral neck BMD were maintained or slightly increased. The adverse event profiles were similar between the two treatment groups during yr 8.

Conclusions:

One year of discontinuation of risedronate treatment in patients who had received 2 or 7 yr of risedronate therapy led to increases in NTX/Cr levels toward baseline and decreases in femoral trochanter and total hip BMD.

Although bisphosphonate therapy has proven efficacious in the treatment of osteoporosis, the optimal duration of treatment is still unknown. Knowing how quickly bisphosphonate treatment effects begin to resolve is important when making decisions about changing or terminating treatment. Some evidence suggests that the length of treatment effect after discontinuation varies based upon the physical-chemical characteristics of the particular bisphosphonate.

The Fracture Intervention Trial Long-Term Extension (FLEX) has provided data on the effects of stopping alendronate on bone mineral density (BMD) and bone turnover markers (BTM) (1–3). Compared with women who discontinued alendronate after an average of 5 yr, those continuing treatment maintained a higher BMD and greater reduction of BTM after 3 yr (3) and 5 yr (1). Those discontinuing alendronate treatment demonstrated a moderate decline in hip and spine BMD and a gradual, albeit sustained, increase in BTM. When comparing fracture-risk outcomes after 5 yr, clinically recognized vertebral fractures were significantly more common among those discontinuing alendronate, but the cumulative risk for nonvertebral fractures and morphometric vertebral fractures were similar (1). However, this study should be interpreted with care because there was no placebo group. Zoledronic acid has been shown to sustain antiresorptive effects for 3 yr compared with the placebo group. In a 3-yr, double-blind, randomized, placebo-controlled trial with 50 postmenopausal women with osteopenia, participants received a single dose of zoledronic acid or placebo. After 3 yr, β-C-terminal telopeptide of type I collagen (β-CTX) and procollagen type-I N-terminal propeptide (PINP) levels were 40% lower in the zoledronic acid group. Whether such dosing reduces fracture risk is not known (4). In a very small study, 30 patients (including seven men) with osteoporotic vertebral fractures were treated with oral pamidronate for 5–9 yr and then followed for 12–24 months after discontinuation (5). During follow-up, no changes were observed in lumbar spine or femoral neck BMD. Serum alkaline phosphatase (ALP), which was depressed by about 10% at the end of treatment, rose to pretreatment values by 6 months after stopping treatment. The hydroxyproline/creatinine (Cr) ratio, which had been depressed by nearly 30%, rose to about 15% below the pretreatment level and then remained stable. In contrast, in a follow-up of women with postmenopausal osteoporosis treated with oral pamidronate, 150 mg/d, for 3 yr, there were significant losses in BMD in the total body and femoral trochanter 1 yr after discontinuing treatment (6). BMD also decreased, nonsignificantly, at the lumbar spine, femoral neck, and Ward's triangle.

Conversely, a 1-yr discontinuation study with risedronate showed loss of previous antiresorptive effects (7). Postmenopausal women with previous vertebral fracture received placebo or risedronate 5 mg daily for 3 yr, plus calcium and vitamin D (if indicated). They subsequently discontinued the study drug (except for calcium and vitamin D) during a fourth year of follow-up off-treatment with risedronate (7). Follow-up was completed by 309 former risedronate and 290 placebo patients. One year after treatment discontinuation, evaluations found that BMD decreased, although values remained higher than those for the placebo group. BTM, however, showed complete resolution to levels of the control group. Incident morphometric vertebral fractures remained statistically significantly lower in the group previously treated with risedronate (46% lower than the control group, P = 0.009). The number of nonvertebral fractures was similar for both groups regardless of cause and location. Therefore, partial or complete resolution of antiresorptive effects, as measured by BMD and BTM, did not correspond to loss of the protective effect of risedronate on vertebral fractures during the year after treatment discontinuation.

The aim of the present study was to assess the effect of risedronate discontinuation after short-term (2 yr) and long-term (7 yr) risedronate therapy for postmenopausal osteoporosis by evaluating bone resorption using urinary type I collagen cross-linked N-telopeptide (NTX) and BMD values at the lumbar spine and total hip.

Patients and Methods

This report describes an extension of the pivotal Vertebral Efficacy with Risedronate Therapy MultiNational (VERT-MN) study, designed to assess the effect of discontinuation of risedronate in postmenopausal women with osteoporosis. The extension trial was approved by local institutional review boards or independent ethics committees. Informed consent was secured from all study participants. The study was conducted according to the principles of the International Conference on Harmonisation of Technical Requirements for Registration of Pharmaceuticals for Human Use Harmonised Tripartite Guideline for Good Clinical Practice; the U.S. Code of Federal Regulations, Title 21, Parts 50, 56, and 312; the European Community Guideline Good Clinical Practice for Trials on Medicinal Products in the European Community; and the ethical principles of the Declaration of Helsinki (www.ClinicalTrials.gov identifier: NCT01249261).

Patients and treatments

A design summary of the initial 3-yr, placebo-controlled VERT-MN study and subsequent extensions, including the current study, is shown in Fig. 1. Inclusion criteria for the initial 3-yr study have been reported previously (8). Women enrolled in VERT-MN were at least 5 yr postmenopausal, no more than 85 yr of age, and had at least two vertebral fractures (T4–L4, inclusive) at baseline. Patients from centers completing the second extension (yr 7 of treatment) were offered entry into the study at yr 8, unless they were less than 60% compliant with study drug during the previous extension or had started or continued a bisphosphonate, glucocorticoid, anabolic steroid, calcitonin, fluoride, vitamin D, calcitriol, or hormone therapy between completion of the second extension and entry into yr 8 of the study.

Fig. 1.

Study design of the 8-yr trial.

Patients received elemental calcium (1 g daily) throughout the study. Patients who required vitamin D supplementation, according to the investigator's medical judgment, received vitamin D supplements (up to 500 IU/d) throughout the trial.

Before enrollment, the change in bone mass over the preceding 7-yr period was reviewed. If there was evidence of excessive bone loss as defined by a loss in femoral neck BMD of more than 12% and more than 0.080 g/cm² or a loss in lumbar spine BMD of more than 10% and more than 0.080 g/cm², the physician and patient discussed whether continuation in the trial was in the patient's best interests. These criteria were independent of the time on study drug because they were already used as a guideline (not a requirement) in the first extension (9) and the second extension (10). A loss of bone mass of more than 0.080 g/cm² was assumed as being a high risk for fractures.

Measurements

NTX levels were measured in second morning void urine samples collected at baseline and 3, 6, 36, 48, 60, 72, 84, 90, and 96 months using an automated electrochemiluminescent immunoassay (Ortho Clinical Diagnostics, Rochester, NY); the intraassay coefficient of variation (CV) was less than 6%. Urinary Cr was measured by dry-slide chemistry (Ortho Clinical Diagnostics); the intraassay CV was less than 3%. Urinary NTX levels were expressed as a ratio to Cr (NTX/Cr), and the results are expressed as nanomoles bone collagen equivalent per millimole Cr. Samples were stored at below −20 C until analysis and analyzed in four batches at the end of each stage of the study (Fig. 1). In each batch, the original baseline sample was remeasured, and the percentage change from baseline was calculated from the value of the baseline obtained in the same batch. Adverse events were recorded, and laboratory evaluations and physical examinations were conducted. BMD values were determined by dual-energy x-ray absorptiometry (DXA) scans of the lumbar spine and total hip. Within a specific study site, the same machine (Lunar or Hologic) was used for all DXA measurements. All reported fractures sustained during this extension were confirmed radiographically. Bone ALP was measured using the Tandem-R Ostase immunoradiometric assay (Hybritech, San Diego, CA); the intraassay CV was 3.7–6.7%.

For yr 8, baseline data included data collected during the previous two extension studies, the 3-yr core study, and from the month-24 visit in yr 6 and 7 or within 180 d of the month-24 visit. DXA of the spine and total hip and adverse events were assessed every 6 months. Serum chemistry, hematology, and physical examination were evaluated at baseline (i.e. month 0 of yr 8) and at month 12 of yr 8.

Statistical analysis

All patients enrolled in yr 8 of the study were included in the intent-to-treat cohort. Analyses were performed on this cohort. Eligibility was defined as completion of the study or the last assessment for which data were available for any given patient (last observation carried forward). Patient demographics and baseline data were analyzed using descriptive statistics. All demographic variables and baseline medical characteristics (excluding medical history) were established at baseline of the pivotal study (8).

Descriptive statistics were provided for NTX and BMD percent changes from the pivotal study baseline and from month 0 of yr 8.

Results

Patient characteristics and disposition

Of 136 patients who completed yr 6–7 (10), 61 entered yr 8, and 60 patients completed month 12 of yr 8.

Patient characteristics for patients enrolled into study yr 8 are shown in Table 1. Treatment groups are identified according to treatments received during the previous 7 study years. The placebo/5-mg-risedronate group received placebo for 5 yr and then risedronate for 2 yr, whereas the 5-mg-risedronate group received risedronate for 7 yr.

Table 1.

Demographics and baseline characteristics for the yr-8 study patients (intent-to-treat population)

Characteristic	Placebo/5 mg risedronate (n = 30)	5 mg risedronate (n = 31)
Age (yr), mean (sd)	66.9 (6.2)	67.3 (5.8)
Time since last menstrual period (yr), mean (sd)	20.0 (7.0)	20.3 (6.5)
Race [n (%)]
Caucasian	30 (100)	31 (100)
Weight (kg), mean (sd)	66.0 (11.04)	64.9 (10.03)
Height (cm), mean (sd)	157.2 (7.2)	155.1 (6.8)
Lumbar spine BMD (mg/cm2 standardized)
n	19	24
Mean (se)	862.6 (34.4)	833.0 (44.7)
Prevalent vertebral fractures at pivotal study baseline
n	28	29
Median (range)	4.0 (2–11)	3.0 (0–12)
Prevalent vertebral fractures at month 0 of yr 8
n	28	29
Median (range)	4.5 (2–11)	3.5 (0–12)

At yr 8, the two treatment groups were relatively balanced in terms of age and lumbar spine BMD. The median number of prevalent vertebral fractures per patient was higher in the placebo/5-mg-risedronate group compared with the 5-mg-risedronate group.

NTX/Cr levels

In both treatment groups, NTX/Cr levels increased during yr 8 when compared with values at the end of yr 7. There was no significant difference between NTX levels during yr 8 (90 and 96 months) compared with month 60; i.e. NTX levels returned to placebo levels. Mean percent changes from pivotal study baseline to the end of yr 8 for NTX/Cr values are shown in Fig. 2.

Fig. 2.

Mean percent change of NTX/Cr levels from pivotal study baseline to the end of yr 8. Bl, Baseline; RIS, risedronate.

BMD values

At the end of yr 8, total hip and femoral trochanter BMD decreased from the levels at the end of yr 7 in both the placebo/5-mg-risedronate group and the 5-mg-risedronate group (Fig. 3), whereas lumbar spine and femoral neck off-treatment BMD were maintained or increased in both treatment groups compared with levels at end of yr 7 (Table 2).

Fig. 3.

Mean percent change of the total hip (A) and femoral trochanter (B) BMD from pivotal study baseline to the end of yr 8. Bl, Baseline; RIS, risedronate.

Table 2.

Mean percent change from pivotal study baseline in BMD and bone ALP after discontinuation of risedronate

Skeletal site	Mean percent change from pivotal study baseline
	Placebo/5 mg risedronate (n = 30)				5 mg risedronate (n = 31)
	n	End of yr 7	n	End of yr 8	n	End of yr 7	n	End of yr 8
Lumbar spine BMD (se)	13	5.94a (2.53)	11	5.53a (1.66)	18	12.02a (1.91)	18	13.35a (2.10)
Femoral neck BMD (se)	21	−1.28 (1.66)	21	−1.11 (1.95)	22	1.23 (0.90)	22	3.22 (3.56)
Bone ALP (se)	22	−18.46a (9.10)	21	2.14 (9.37)	24	−31.96a (5.53)	24	−7.42a (8.13)

^aIndicates a statistically significant difference from baseline determined from within-treatment 95% confidence interval.

Adverse events

Approximately 60% of patients reported an adverse event during yr 8; however, there were no withdrawals due to adverse events or deaths. The adverse event profile was similar between treatment groups, but more events, serious adverse events, and upper gastrointestinal adverse events were reported in the placebo/5-mg-risedronate group than in the 5-mg-risedronate group (Table 3). No vertebral fractures were reported as adverse events, and one traumatic fracture of the right proximal humerus (due to a fall) was reported as an adverse event in one patient (3.2%) in the 5-mg-risedronate group.

Table 3.

Adverse events for the yr-8 study patients (intent-to-treat population)

	Placebo/5 mg risedronate (n = 30)	5 mg risedronate (n = 31)
No. of AE reported	50	34
No. (%) of patients with AEa	18 (60)	17 (55)
Mean no. of AE per patientb	3	2
No. of serious AE	3	0
No. (%) of patients with serious AEa	2 (7)	0
No. of upper GI AEc	4	0
No. (%) of patients with upper GI AEa,c	3 (10)	0
No. of moderate-to-severe upper GI AEd	1	0
No. (%) of patients with moderate-to-severe upper GI AEa	1 (3)	0

AE, Adverse events; GI, gastrointestinal.

^aPatients who experienced one or more AE are counted once.

^bFor patients who had AE.

^cUpper GI adverse events according to COSTART descriptions.

^dModerate-to-severe upper GI AE are those which were upper GI AEs as defined above and had moderate or severe clinical severity.

Discussion

Our study shows that 1 yr discontinuation of risedronate after long-term treatment induces changes in NTX/Cr and BMD values. NTX/Cr levels increased from levels on risedronate treatment, consistent with previous findings (7). It is notable that the offset of treatment effect on bone resorption is similar after 2 and 7 yr exposure to risedronate; this may mean that the cumulative effect of the drug is very small, and this could relate to the low affinity of risedronate for hydroxyapatite (11, 12). At the end of yr 8 (the risedronate-free year), total hip and femoral trochanter BMD levels decreased from those at the end of yr 7 in both treatment groups. Lumbar spine and femoral neck BMD remained stable or increased in both treatment groups. Reinitiation of risedronate therapy for 2 yr after this 1-yr discontinuation study led to decreases in NTX/Cr levels and increases in lumbar spine BMD, indicating that the bone is still responsive to risedronate therapy after long-term treatment (13).

Previously published studies have shown that improvements in BMD and BTM generally persisted after discontinuation of alendronate, with only modest returns toward control values (1, 3). The continued effect shown in this study has encouraged the use of drug holidays after 5 yr of alendronate therapy, although if the femoral neck BMD is −2.5 or less, then there is still reduction in nonvertebral fracture risk from continuing therapy (14). Antiresorptive effects with zoledronic acid were sustained for 3 yr after a single dose compared with placebo control subjects (4). Positive effects were found in a different study with zoledronic acid compared with risedronate in Paget's disease, which in this case might be desirable (15). In a previously published study with risedronate, resolution of antiresorptive effects of both BMD and BTM occurred 1 yr after discontinuing risedronate, but the risk of morphometric vertebral fracture remained reduced (7). The increase in bone turnover during offset of treatment is often associated with some bone loss; this is best seen in the clinical trials of denosumab in which the bone turnover markers are transiently higher than at baseline (16, 17). However, we do not yet know the relevance of this for fracture risk. The current study similarly showed loss of antiresorptive effects after discontinuation of risedronate after both short- and long-term risedronate therapy with a decrease in femoral trochanter BMD similar to that reported previously (7) but less loss from the spine and femoral neck; the reason for the latter finding is unclear, but the current study design differs from that of Watts et al. (7) in that there are few subjects and no control group. The changes in NTX/Cr were similar in the current study to that of Watts et al. (7); the change in NTX/Cr in response to placebo was about 30%, and risedronate 60%, and at offset about 40% below baseline. Although the findings from these various studies imply differences regarding the persistence of antiresorptive effect after discontinuation among various bisphosphonates, there have been no head-to-head studies investigating these effects, and therefore, no direct comparisons can be made at this point. It is imperative that head-to-head studies are conducted comparing the offset of treatment effect of currently licensed bisphosphonates; this would have practical clinical implications because it would help inform the likely time course of changes during drug holidays. Better still, such studies should be powered to allow study of fracture effects.

In the present study, the adverse event profile was similar between the two treatment groups; no new safety concerns were observed during yr 8 compared with the first 5 or 7 yr of study (9, 10). Reinitiation of risedronate therapy for 2 yr after this 1-yr discontinuation showed a similar adverse event profile to that observed in previous VERT studies, and no new safety concerns were detected (13).

Interpreting data from the current study is limited by the small sample sizes and lack of data on morphometric vertebral fractures or data on vitamin D status. Thus, we cannot comment on whether the resolution of the effect of risedronate on NTX/Cr and hip BMD is associated with any change in fracture risk. Although all patients who could be contacted were invited to participate, some investigators closed their sites and patients were lost to follow-up over this 8-yr period. Baseline characteristics of the extension cohort were generally similar to the original study, but average age was obviously higher in the extension. Furthermore, samples for NTX/Cr level analysis were not always collected after an overnight fast or at the same time of day, which would have resulted in greater individual variability.

One year of discontinuation of risedronate treatment, in patients who had received 2 or 7 yr of risedronate therapy, led to increases in NTX/Cr levels toward baseline and decreases in femoral trochanter and total hip BMD. After 7 yr of risedronate therapy, bones remain reactive after a year of treatment discontinuation.