Ten years too long: strontium ranelate, cardiac events, and the European Medicines AgencyBMJ 2016; 354 doi: https://doi.org/10.1136/bmj.i5109 (Published 30 September 2016) Cite this as: BMJ 2016;354:i5109
- Mark J Bolland, associate professor of medicine,
- Andrew Grey, associate professor of medicine
- Correspondence to: M Bolland
- Accepted 19 September 2016
Strontium ranelate was registered in Europe in 2004 for postmenopausal osteoporosis. Sales reached about €200m (£170m; $220m) a year by 2010,1 with 3.4 million patient years of use by February 2013.2 In April 2013 the European Medicines Agency (EMA) reported that strontium increased the risk of myocardial infarction.2 Surprisingly, this was driven by the results of TROPOS, a phase III trial completed in November 2003, almost 10 years earlier. No further large clinical trials of strontium had been conducted, and cardiac safety concerns were not mentioned in previous trial publications or publicly available regulatory documents (fig 1⇓).3
To understand how it could take 10 years for an important adverse effect to be reported, we requested the clinical safety reports (CSRs, box 1) of the two phase III trials, SOTI and TROPOS. Our request was declined by the EMA on the grounds of commercial sensitivity, but we successfully appealed. The CSRs, together with data derived from safety assessments carried out by the EMA in 2006-07, confirm that the agency identified cardiac safety concerns in 2007 but did not report them, raising concerns about the regulatory process.
Box 1: EMA definitions, documents, and committees
Adverse event—Any untoward medical occurrence in a patient or clinical trial participant given a medicinal product that does not necessarily have a causal relationship with treatment
Adverse reaction—A response to a medicinal product that is noxious and unintended. A causal relationship between the product and an adverse event is at least a reasonable possibility
Serious adverse reaction—An adverse reaction that results in death; is life threatening; requires hospitalisation or prolongation of existing hospitalisation; results in persistent or significant disability or incapacity; or is a congenital anomaly
Severity of adverse events—The intensity of a specific event (mild, moderate, or severe)
Clinical study report (CSR)—A standardised document submitted for regulatory review as part of a marketing or registration application that reports the results of one or more clinical trials and includes data on the efficacy and safety of a therapeutic agent
Summary of product characteristics—A regulatory document that describes the properties and approved conditions of use of a medicine
Periodic safety update report (PSUR)—A report prepared by the manufacturer describing the safety experience with a medicine at that time
Pharmacovigilance Risk Assessment Committee (PRAC)—A 40 member committee responsible for assessing and monitoring safety issues for medicines. Its recommendations are considered by the CHMP
Committee for Medicinal Products for Human Use (CHMP)—A 37 member committee responsible for preparing European Medicine Agency opinions on human medicines including initial assessment and postmarketing review. It considers PRAC recommendations and has final responsibility for regulatory decisions
Cardiac adverse events in TROPOS
TROPOS was a five year, randomised, placebo controlled trial in 5091 postmenopausal women conducted by Servier, the manufacturer of strontium. It finished in November 2003,4 5 6 and the final CSR is dated 30 May 2006.7 Strontium reduced the absolute risk of the primary endpoint, peripheral osteoporotic fracture, by 2.3%.7
The CSR did not contain a summary table of adverse events by randomised treatment group; instead adverse events were presented in a confusing series of categories relating to severity, seriousness, and treatment adherence (table 1⇓). A 2.0% increase in the absolute risk of serious adverse cardiac events with strontium was found, largely due to myocardial infarctions and cardiac failure. The absolute risk of myocardial infarction (emergent on treatment) was 0.9% higher with strontium. Several terms were used for cardiac failure: pooling these categories showed a statistically significant 2.0% increase in the absolute risk of severe heart failure in the strontium group. Treatment withdrawals for cardiac disorders were almost twice as common with strontium as with placebo. The imbalance in myocardial infarction between treatment groups was specifically mentioned at several points in the CSR,7 including the conclusions, but not in the CSR study synopsis or the primary trial publications.5 6
EMA regulatory response to TROPOS
Adverse effects of strontium were considered by the EMA in 2006-07.8 The only available data were from TROPOS, SOTI,9 10 11 and small phase II trials. Pooled analyses of SOTI and TROPOS showed an increased risk of myocardial infarction with strontium and an increase in heart failure cases categorised as congestive cardiac failure (table 2⇓).8 The safety assessment concluded that “there was a slightly higher incidence of cardiac disorders; and specifically coronary artery disorders and heart failure” and requested that Servier perform more detailed analyses on “ischemic heart disease and heart failure” in this pooled dataset.8
Servier conducted new analyses for heart failure and the broad endpoint of coronary artery ischaemic disease, exploratory subgroup analyses, and analyses adjusting for risk factors.12 Neither endpoint was explicitly defined. Servier concluded that strontium did not increase the risk of either outcome in the entire cohort; in those with hypertension, hypercholesterolaemia, or venous thromboembolism; or in those aged >80.12 The report did not mention the increased risk of myocardial infarction or report any additional analyses of that endpoint. Evidence that any of these extra analyses were independently verified by the EMA is lacking. Servier said that revised wording of the Summary of Product Characteristics was unnecessary.12 The EMA agreed: “There are no signals that raise further concerns. The conclusion by [Servier] is agreed. Issue resolved.”12 The EMA did not mandate any postregistration surveillance of cardiac events.
MALEO trial and EMA response
Concerns about cardiovascular safety might not have been made public if the MALEO trial had not been submitted for regulatory review in February 2011. MALEO was a two year randomised, placebo controlled trial in men with osteoporosis (173 strontium, 87 placebo).13 It showed a higher absolute risk of cardiac events in men randomised to strontium (8.7% versus 4.6% for placebo).13 Despite the concerns already raised by TROPOS, Servier and the MALEO investigators said that this was a chance finding, attributable to differences in baseline cardiac risk factors.13 14
The EMA asked for further monitoring of cardiac events with strontium.14 Servier proposed (and the EMA endorsed) a three year cohort study assessing tolerability of strontium in men and a retrospective case control study in men and women to assess ischaemic cardiac events.14
The 2013 Periodic Safety Update Report (PSUR) said that strontium increased the absolute risk of myocardial infarction by 0.6% in a pooled analysis of seven trials in postmenopausal women, with similar risks in osteoarthritis and male osteoporosis.2 Of the 104 women with myocardial infarction, 103 participated in SOTI and TROPOS, completed about 10 years earlier. Although they reported similar absolute risks for myocardial infarction, the safety assessments from 2006 and 2013 reported (as yes unexplained) different numbers of participants with myocardial infarction in the pooled analyses of SOTI and TROPOS (57 strontium, 36 placebo in 2006; 64 strontium, 39 placebo in 2013; table 2⇑). Heart failure was reported in the 2013 PSUR only in a table of serious adverse events from all clinical studies in postmenopausal osteoporosis,2 using several different terms (table 2⇑). Pooling these terms showed a 0.6% increase in the absolute risk of cardiac failure with strontium, which was not mentioned in the PSUR.
The EMA Pharmacovigilance Risk Assessment Committee (PRAC) reviewed the PSUR data in March 2013 and asked Servier for supplementary data on cardiac safety and the risk-benefit profile of strontium for its final report.2 Servier presented data from observational studies and exploratory post hoc subgroup analyses, without any evidence of independent verification. It reported no increased risk of myocardial infarction with strontium (but maintenance of fracture prevention) in a subgroup of patients with no history of ischaemic heart disease and without baseline diastolic blood pressure >90 mmHg or systolic blood pressure >160 mmHg.2
PRAC considered the results of seven trials in postmenopausal osteoporosis, one trial in male osteoporosis, and one trial in osteoarthritis, together with the results of a registry based prospective cohort study of postmenopausal women taking strontium, postmarketing surveillance data, and preliminary results from a case control study approved by the EMA and funded by Servier.2 In April 2013 it recommended, by majority decision, stringent prescribing restrictions for strontium and close additional regulatory monitoring. It concluded that the risk-benefit profile of strontium was favourable in patients without ischaemic heart disease, peripheral arterial disease, cerebrovascular disease, uncontrolled hypertension, or significant cardiovascular risk factors.2 But 10 members of PRAC thought that the risk-benefit profile of strontium favoured harm and that its registration should be suspended.2
More regulatory reviews
The EMA Committee for Medicinal Products for Human Use (CHMP), which has final responsibility for EMA regulatory decisions, initiated a further regulatory review of strontium by PRAC in April 2013, although there were no new clinical trial data on safety or efficacy.15 Servier presented more subgroup analyses. An ad hoc group of unnamed experts in osteoporosis, cardiology, epidemiology, and general practice provided advice to PRAC. This time, by majority decision, PRAC concluded that the benefits of strontium did not outweigh its risks, and in January 2014 recommended that its registration be suspended. Fifteen members disagreed.15
CHMP overruled PRAC,15 concluding in February 2014 that—in the restricted population identified in the post hoc exploratory subgroup analyses—the benefits outweighed the risks. By majority decision CHMP recommended that strontium remain registered for patients with no other treatment options, with the previous restrictive prescribing conditions in place. Ten members disagreed, saying “We are not convinced that the proposed risk minimisation measures are realistic, and therefore they cannot achieve what is intended.”15
The CHMP decision placed greater weight on the observational studies and subgroup analyses, all of which had substantial limitations. Two of the observational studies and the postmarketing surveillance data had no controls and very few events (≤33), and the case control study identified only 64 users of strontium (average duration 11 months) who had a myocardial infarction.16 In the case control study, baseline characteristics of strontium users differed substantially from users of other agents, indicating potential for confounding.16 These inherent weaknesses prevent reliable conclusions being made. The subgroup analyses included only 11-30 of 104 myocardial infarctions, indicating substantial underpowering, and strontium had no effect on non-vertebral fractures in these subgroups.15 Furthermore, the subgroups were chosen retrospectively, results were presented only for selected subgroups (the number of other subgroup analyses performed is unclear), and interaction tests were not reported. Basing decisions on these results contravenes recommended approaches to interpretation of subgroup analyses.17
The EMA reviews of virtually identical data led to markedly different responses in 2006-07 and in 2013-14. The main consequence was that failure to report the risk of myocardial infarction in 2006-07 distorted the risk-benefit profile of strontium in favour of its widespread use (box 2). By 2010, sales in Europe were approximately €200m a year,1 but 20-30% of people taking strontium in Denmark had conditions considered to be prescribing contraindications, and a further 35% had probable prescribing contraindications.18 By 2012, the drug was registered in 101 countries and marketed in 81.14 If the regulatory decisions made in 2012-14 had been applied in 2006-07, strontium use would have been less widespread, doctors and their patients could have included the cardiovascular risk estimates in their deliberations about strontium therapy, and patient harm might have been avoided.
Box 2: Consequences of regulatory failure
Lack of public notification of adverse cardiac events (myocardial infarction and heart failure) created a distorted risk-benefit profile that favoured the widespread use of strontium ranelate, including registration in other countries
Missed opportunity to adjudicate cardiovascular events from existing trials and closely monitor cardiovascular events in later trials or in postmarketing surveillance
Missed opportunity to investigate potential mechanisms and risk factors for increased cardiovascular risk in later trials and by independent investigators
Loss of confidence in regulatory processes
Several avoidable problems contributed to the regulatory failure (box 3). Additionally, heart failure events were recorded using multiple terms, which may have led to missed cases or lack of power to detect differences between groups. Pooling results from a broad range of terms may have included incorrect diagnoses or obscured specific adverse effects. The outcome would have been improved by using better formats for collation and reporting of adverse events in clinical trials and CSRs, by making CSRs publicly accessible, and by ensuring statistical analyses were verified by an independent party free of conflict of interests.
Box 3: Avoiding regulatory failure
CSRs should have a table of all adverse events by treatment allocation for all randomised participants
Differences in participant numbers or adverse events from the intention-to-treat analysis should be justified and explained
CSRs should be made available for public scrutiny. Academic investigators should report findings in primary trial publications or subsequent manuscripts
Subgroup analyses should be verified by an independent statistician and interpreted cautiously using recommended approaches
Final datasets should be submitted for verification by regulatory authorities, and anonymised datasets should be made accessible to independent researchers after registration
CSRs and publications should include clear statements about the role of academics in designing the study, interpreting data, and writing manuscripts, including whether analyses conducted by companies were independently verified, and whether academics had free access to all available data
Composition of expert advisory committees should be published together with declarations of conflict of interest
Safety concerns recognised in 2003
In June 2003 Servier applied for registration of strontium based on three year data from SOTI and TROPOS. Limited four year safety data were also considered. The 2013 PSUR shows that the relative risk of myocardial infarction in these trials was elevated (by 1.6%) at 12 months and remained similar subsequently (fig 2⇓). The absolute risk was elevated by 0.18% at 12 months, rising to 0.73% at 60 months. Four year and five year follow-up were completed by November 2002 and 2003, respectively, so a safety concern about myocardial infarction should have been flagged before registration in September 2004.
Strontium ranelate should not be registered
Ten years after strontium was introduced to clinical practice, its risks and benefits can be independently evaluated (fig 3⇓). Fifty five fewer non-vertebral fractures and five fewer hip fractures occurred in the strontium groups. But so did 25 more cases of myocardial infarction, 27 more cases of venous thromboembolism (including 11 more cases of pulmonary embolism), and 19 more cases of heart failure.
The available trial data show that any benefit of strontium in fracture prevention is outweighed by risk of cardiovascular events and venous thromboembolism. In more recent observational studies, strontium has been associated with higher all cause mortality in one report,18 but otherwise no association has been found with acute coronary syndrome,19 myocardial infarction,16 18 stroke,18 cardiovascular mortality,16 18 or all cause mortality.16 19 Inherent methodological issues—in particular confounding by indication, few events, and short duration of exposure—mean that these data do not diminish the concerns about adverse events that arose in the randomised trials. We think that strontium ranelate should not be registered for clinical use.
Regulatory documents from 2006-07 show that strontium ranelate increases the risk of myocardial infarction and congestive heart failure
The EMA did not act on these data at the time or make them publicly known until 2013, indicating regulatory failure
The risks of strontium ranelate outweigh the benefits, and it should not be prescribed for fracture prevention
Competing interests: All authors have completed the ICMJE uniform disclosure form at www.icmje.org/coi_disclosure.pdf and declare: the study was funded by the Health Research Council (HRC) of New Zealand. MB is the recipient of a Sir Charles Hercus Health Research Fellowship; AG receives research funding from Pharmac, the New Zealand Government’s Pharmaceutical Management Agency, and is a shareholder in Auckland Bone Density, an organisation that provides bone densitometry services.
Contributors and sources: MB and AG are endocrinologists with research interests in the management of osteoporosis. We carried out the research because we could not understand the long delay between the completion of the TROPOS trial and the public notification of increased myocardial infarction risk in TROPOS. MB and AG designed the research. MB applied to the EMA for the clinical study reports, extracted the data, and performed the analyses. AG checked the data. MB drafted the paper. AG critically reviewed and improved it. MB is the guarantor for the article. Both authors had access to all the data and take responsibility for the integrity of the data and the accuracy of the data analysis.
Provenance and peer review: Not commissioned; externally peer reviewed.