Statistics from Altmetric.com
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.
Stakkestad et al reported a clinical trial where intravenous (IV) ibandronate injections, given every 3 months during 1 year, produced a dose dependent gain in mean (SD) lumbar spine bone mineral density (BMD) compared with placebo in prevention of bone loss in postmenopausal women.1 The treatment was then proposed as an alternative to oral bisphosphonates and hormonal therapy in preventing postmenopausal osteoporosis.
The primary outcome was the relative change from baseline in lumbar BMD after 2 years of treatment tested by analysis of variance, with treatment group and stratum as independent variables. We found several limitations.
Firstly, the initial study was planned for 2 years, but was stopped at 12 months because of the interesting results of the IV ibandronate pivotal phase III fracture study.2 Was it possible to infer that results obtained from a trial on fracture occurrence would apply to the study of BMD?
Secondly, given the limited changes in BMD, it could be suggested that they might be due to random variability in the procedure. The least significant change (LSC) in BMD is the percentage of change that is unlikely to be due to the precision error of the procedure. The LSC is 2.8-fold the precision error of the procedure on a specific device, site of measurement, and number of measurements.3 Precision error was not evaluated in the study. The sample size was calculated with an expectation of a significant change in lumbar spine BMD of 3% compared with placebo after 2 years of treatment. Would a 3% change be clinically relevant according to the variability of the measure?
Thirdly, analysis of variance was used for a dose-response model. Multiple comparisons were performed without using post hoc tests which would have adjusted the nominal threshold. The type I error would have been lower than 0.05.
Keeping these limitations in mind, on a clinical ground, IV ibandronate still needs to prove its efficacy on BMD gain.
On behalf of my fellow authors, I welcome the opportunity to respond to the points raised in the letter by Milka Maravic and Paul Landais. Their letter proposed several limitations of our recently published study, which investigated the efficacy and safety of intermittent intravenous (IV) ibandronate (0.5 mg, 1 mg, and 2 mg) injections in the prevention of postmenopausal osteoporosis.1
In their first comment, Maravic and Landais question why the suboptimal findings from the IV ibandronate pivotal phase III fracture prevention trial led to our study being stopped prematurely after 1 year. In that trial, 0.5 mg and 1 mg IV ibandronate injections every 3 months failed to reduce vertebral fracture risk significantly, relative to placebo.2 The decision to stop our study prematurely was taken by the sponsors, Hoffmann-La Roche Ltd, because of ethical concerns; it was decided at the time that patients should not continue to receive treatment potentially associated with suboptimal efficacy.
However, upon analysis of the results from the fracture prevention study, it became evident that the suboptimal antifracture efficacy was due to under dosing and that higher doses would probably provide the required gains in bone mineral density (BMD) and decreases in biochemical markers of bone turnover. The proposition that higher doses of ibandronate would produce optimal efficacy is supported by the results from a study that evaluated the dose-response relationship associated with intermittent IV ibandronate injections in postmenopausal women with osteoporosis.3 In that study, a 2 mg dose of ibandronate provided significantly greater BMD increases and suppression of bone resorption markers than the 1 mg dose investigated in the fracture prevention study.
The available results from our study were analysed to enable further evaluation of the dose-response relationship of intermittent IV ibandronate. As reported, although our study was terminated prematurely, sufficient results were collected to allow analysis of data for the first year of the study. These results support those seen previously: intermittent IV ibandronate produced a dose dependent effect on BMD and biochemical markers of bone resorption, with the highest dose producing the greatest efficacy. Owing to the findings from our study, together with those from prior studies, a large non-inferiority trial is continuing to determine whether a higher dose (3 mg every 3 months) or a shorter dosing interval (2 mg every 2 months) of IV ibandronate provides equivalent therapeutic efficacy relative to an oral daily ibandronate regimen, which has proved antifracture efficacy (the Dosing IntraVenous Administration study).
In their second observation on our study, the authors imply that the observed changes in BMD might have been due to the random variability associated with the procedure. However, all the BMD measurements in our study were centrally assessed by Synarc (Ballerup, Denmark), where they were also quality assured. Synarc is a well respected company that employs radiologists who are leading experts in clinical trial radiology. Thus, any variability in the assessment of spinal BMD in our study would have been kept to a minimum and the precision error (PE) associated with the measurements would be unlikely to have exceeded 1%. Hence, as defined by several authors, including Gluer et al,4 the percentage of BMD change that is unlikely to be due to the variability (PE) of these measurements (the least significant change) is expected to be about 2.8% (1% multiplied by 2.8). As a result, the sample size calculations used in our study, which were based on an expectation of a significant change in lumbar spine BMD of 3% versus placebo after 2 years of treatment, were justified. Because the 3% change exceeds the 2.8% variability of the measurement procedure, it is considered clinically relevant.
In their third comment, the authors correctly state that analysis of variance was used to investigate the efficacy of the three ibandronate regimens. They go on to advise that because multiple comparisons were performed, adjustments should have been made to ensure the maintenance of the overall type I error. However, as only three tests were performed for the primary end point (the relative effect of treatment on lumbar spine BMD), all of which were highly significant, it was not deemed necessary to apply adjustments.
Additional analyses were conducted to determine the influence of baseline spinal BMD and/or time since menopause on the primary end point. In retrospect, as a number of comparisons were performed when conducting these secondary analyses, we agree with Maravic and Landais that relevant adjustments would be optimal. Thus, we have retrospectively applied a Bonferroni correction, a well recognised test to adjust for multiple comparisons. Despite the highly conservative nature of such statistical adjustments when the number of repeated tests is >5, only two results changed (table 1).
Therefore, reassuringly, the findings observed after applying the Bonferroni correction were similar to those reported in our publication.
In summary, although our randomised, double blind, placebo controlled study was prematurely discontinued, sufficient 1 year data from almost 600 patients were available to allow an adequate assessment of the efficacy of IV ibandronate injections in postmenopausal women without osteoporosis. We found that ibandronate dose dependently prevents postmenopausal bone loss, even when considering the variability associated with measuring BMD, and even when adjusting for multiple, repeated statistical assessments. Of note, the highest dose of ibandronate produced gains in BMD that were comparable to those observed with other therapeutic agents in this setting. Thus, as stated in the conclusion of our publication, intermittent IV ibandronate injections offer the promise of an effective and convenient alternative to current treatments for the prevention of postmenopausal bone loss.