Background A recent pooled analysis of randomised trials found an increased risk of myocardial infarction with use of the antiosteoporotic drug strontium ranelate. We conducted a nationwide cohort study in Denmark, 2005–2011, to investigate the risk of acute coronary syndrome among postmenopausal women treated with strontium ranelate.
Methods The study involved two analytic setups. The first analysis included new users of either strontium ranelate (n=1798) or one of the two first-line bisphosphonates in Denmark (alendronate and risedronate; n=65 236). The second analysis included patients who had first been treated with a first-line bisphosphonate and subsequently switched to either strontium ranelate (n=1219) or ibandronate (n=2290). The primary outcome was acute coronary syndrome (unstable angina or myocardial infarction). The secondary outcome was any-cause mortality. Cox regression was used to estimate HRs, with adjustment using a disease risk score.
Results Compared with use of alendronate/risedronate, use of strontium was not associated with significantly increased risk of acute coronary syndrome (rate per 1000 person-years 5.7 for strontium vs 6.3 for alendronate/risedronate; adjusted HR 0.89, 95% CI 0.52 to 1.55) or any-cause mortality (adjusted HR 0.96, 95% CI 0.76 to 1.21). In the analysis of switchers, strontium was not associated with significantly increased risk of acute coronary syndrome (rate per 1000 person-years 9.9 for strontium vs 9.9 for ibandronate; adjusted HR 1.00, 95% CI 0.49 to 2.05) or of any-cause mortality (adjusted HR 1.45, 95% CI 0.95 to 2.21).
Interpretation These real-world data of antiosteoporotic drug users do not support a significant association between use of strontium ranelate and acute coronary syndrome.
- Cardiovascular Disease
Statistics from Altmetric.com
Strontium ranelate is one of few non-bisphosphonate alternatives for the prevention of osteoporotic fractures and has recently shown positive effects in the treatment of knee osteoarthritis.1 ,2 A recent routine European Medicines Agency assessment of the benefit–risk profile of strontium ranelate, however, found increased risk of serious cardiac events. This pooled analysis of randomised placebo-controlled trials included approximately 7500 women with osteoporosis and found a relative risk of 1.6 (95% CI 1.07 to 2.38) for myocardial infarction with use of strontium ranelate.3 The assessment also found an imbalance of serious cardiac events in two other studies: one in men with osteoporosis and another in patients with osteoarthritis.3
These data prompted the European Medicines Agency to modify recommendations for the use of strontium ranelate, including a contraindication in patients with cardiovascular disease.3 After subsequent in-depth review, the European Medicines Agency's Pharmacovigilance Risk Assessment Committee has recommended that strontium ranelate should no longer be used to treat osteoporosis.4 A decision by the European Medicine Agency’s Committee for Medicinal Products for Human Use on this issue will be made in February 2014.5
Notably, the trials included in the assessment that found an increased risk of myocardial infarction were designed to investigate efficacy, and not to analyse cardiovascular safety, which may raise issues with study definitions of cardiovascular outcomes. After the assessment by the European Medicines Agency was communicated, two observational studies have been published, neither of which found an increased risk of myocardial infarction with use of strontium ranelate.6 ,7
We conducted a population-based historical cohort study of postmenopausal women in Denmark to investigate whether use of strontium ranelate was associated with increased risk of acute coronary syndrome (unstable angina or myocardial infarction) in a real-world setting.
Using Danish registers, we studied a nationwide cohort of women aged 50–84 years in the period 1 January 2005 to 31 December 2011. Although strontium ranelate was marketed in Europe from 21 September 2004, we postponed the study start to 1 January 2005 since very early users of a newly marketed drug may be selected individuals who differ from later and more representative users. Eligible for inclusion were all women of target age residing in Denmark during the study period. We excluded women who had less than 2 years of registered residence in Denmark prior to cohort entry to assure adequate capture of covariates. In the first analysis, we included women who were new users of either strontium ranelate or one of the two first-line bisphosphonates in Denmark (alendronate and risedronate) and had no history of use of any antiosteoporotic drug (bisphosphonates, strontium, raloxifene, denosumab, teriparatide).
Because the characteristics of patients who start treatment with strontium ranelate, which is mainly viewed as a second-line drug, as their first osteoporotic drug may differ from those who start treatment with a first-line bisphosphonate,8 we conducted a second analysis among patients who had used a first-line bisphosphonate and subsequently switched to use of either strontium ranelate or the second-line bisphosphonate ibandronate. Furthermore, in the first analysis, we also included a third exposure group, which consisted of new users of the second-line bisphosphonate ibandronate; this exposure group may be seen as a negative control aimed to assess whether factors related to choice of a non-first-line drug as the first antiosteoporotic drug are associated with acute coronary syndrome.
The primary outcome was fatal and non-fatal acute coronary syndrome at 5 years follow-up. The secondary outcome was any-cause mortality. Additional analyses of the primary outcome were conducted with follow-up of 1 and 3 years, stratified according to history of cardiovascular disease, and with adjustment for concurrent use of calcium supplements with or without vitamin D (which might be associated with cardiac risk9).
The source population was identified from the Danish Civil Registration System.10 The National Prescription Registry was used to identify prescriptions for study drugs.11 Drug exposure was defined as the filling of a prescription, and patients were defined as drug users as long as prescriptions were refilled. The duration of each prescription was estimated from the number of tablets (alendronate 10 mg once daily or 70 mg once weekly; risedronate 5 mg once daily or 35 mg once weekly; ibandronate 50 mg once daily or 150 mg once monthly) or sachets (strontium 1 sachet of 2 g once daily) dispensed. Patients filling prescriptions for parenteral ibandronate were not included because this preparation is indicated for the treatment of tumour-induced hypercalcaemia and prevention of skeletal events among patients with skeletal metastases. We allowed for a gap of 2 weeks between two individual prescriptions in order to account for irregularities in drug intake habits. Information about drug exposure during hospitalisation was not available but was assumed to continue during any hospital stay.
Registers, outcome definitions and potential confounders are described in the online supplementary appendix.
Cox regression was used to estimate the HRs of acute coronary syndrome, comparing current users of the study drugs. Censoring criteria were death due to causes other than acute coronary syndrome, switch to another antiosteoporotic drug, loss to follow-up (disappearance, emigration), end of follow-up and end of study period (31 December 2011). Time since treatment start was the time metric in the models. Proportional hazards assumptions were checked through assessment of the interaction (Wald test) between treatment status and time. Our strategy to control for confounders included the use of a disease risk score in order to capture baseline cardiovascular risk (see online supplementary appendix: Methods). The disease risk score was included as an adjustment variable in the models, categorised according to deciles of the score's distribution (deciles categorised: 1–4, 5–7, 8, 9, 10). In a sensitivity analysis, use of calcium with or without vitamin D was adjusted for through inclusion in models as a time-updated covariate. Sensitivity analysis according to intention-to-treat principles, with the initial treatment carried forward throughout follow-up, was also conducted. SAS V.9.4 software was used (SAS Institute, Cary, North Carolina, USA).
Cohort of new antiosteoporotic drug users
From a source population of 1 195 284 individuals, 1798 users of strontium ranelate, 65 236 users of first-line bisphosphonates alendronate or risedronate (reference group) and 3022 users of ibandronate (negative control) were included in the cohort of women with no previous use of antiosteoporotic drugs (figure 1). The distribution of baseline characteristics according to drug exposure group is shown in table 1. Compared with users of first-line bisphosphonates, users of strontium ranelate were more likely to have started treatment early in the study period; had higher prevalence of peptic ulcer/oesophageal disorder, dementia/functional decline, osteoporotic fractures and lower prevalence of rheumatic disease; had more use of opiate analgesics, antidepressants/anxiolytics/antipsychotics and less use of oral corticosteroids; and had used a higher number of prescription drugs in the last year.
During the 5-year follow-up period, median time on treatment was 0.7 years (IQR 0.2–2.0) among strontium ranelate users, 1.4 years (IQR 0.4–2.9) among alendronate/risedronate users and 1.9 (IQR 0.5–3.4) among ibandronate users. A total of 794 cases of acute coronary syndrome occurred in the cohort. Assessment of the interaction between treatment status and the underlying time scale showed that the proportional hazards assumption was met for all models. The risk of acute coronary syndrome was not significantly increased among strontium ranelate users compared with alendronate/risedronate users (crude HR 0.89, 95% 0.51–1.54, and adjusted HR 0.89, 95% CI 0.52 to 1.55; figure 2). Estimates were similar with follow-up truncated at 1 and 3 years (figure 2). Use of strontium was not significantly associated with increased risk of acute coronary syndrome in the subgroups of women with and without history of cardiovascular disease, respectively (table 2).
In a sensitivity analysis, the estimate for acute coronary syndrome was unchanged when additionally adjusted for concomitant use of calcium supplements with or without vitamin D (adjusted HR 0.89, 95% CI 0.52 to 1.55). In a sensitivity analysis according to intention-to-treat principles, use of strontium was not associated with increased risk of acute coronary syndrome (adjusted HR 1.09, 95% CI 0.76 to 1.56).
Use of strontium use was not associated with significantly increased risk of the secondary outcome of any-cause mortality (73 deaths during 2287 person-years among strontium users (rate 31.9 per 1000 person-years) vs 3427 deaths during 118 472 person-years among alendronate/risedronate users (rate 28.9 per 1000 person-years); adjusted HR 0.96, 95% CI 0.76 to 1.21). In the comparison of ibandronate users (36 cases of acute coronary syndrome (rate 5.8 per 1000) and 174 any-cause deaths (rate 27.8 per 1000)) and alendronate/risedronate users, the adjusted HR was 1.04 (95% CI 0.74 to 1.45) for acute coronary syndrome and 1.14 (95% CI 0.98 to 1.33) for any-cause mortality.
Cohort of switchers from first-line antiosteoporotic drug
The cohort of new users of first-line bisphosphonates (alendronate or risedronate) was followed until patients switched to a second antiosteoporotic drug (figure 1). Among patients who switched, 1219 initiated strontium ranelate and 2290 initiated ibandronate. Patients using strontium and ibandronate were similar in baseline characteristics, although users of strontium were more likely to live in the region of Southern Denmark and less likely to live in Copenhagen; had more use of opiate analgesics; were more likely to have an outpatient hospital contact in the last 6 months; and had switched from first-line bisphosphonates earlier in treatment (table 1).
During the 5-year follow-up period, median follow-up on treatment was 0.3 years (IQR 0.1–1.1) among strontium ranelate users and 0.9 (IQR 0.3–2.4) among ibandronate users, and a total of 42 cases of acute coronary syndrome occurred. Assessment of the interaction between treatment status and the underlying time scale showed that the proportional hazards assumption was met for all models. The risk of acute coronary syndrome was not significantly increased among strontium ranelate users (crude HR 0.95, 95% CI 0.46 to 1.94, and adjusted HR 1.00, 95% CI 0.49 to 2.05; figure 3). Estimates were similar with follow-up truncated at 1 and 3 years (figure 3). Use of strontium was not associated with significantly increased risk of acute coronary syndrome in the subgroups of women with and without history of cardiovascular disease, respectively (table 2). In a sensitivity analysis, the estimate for acute coronary syndrome was unchanged when additionally adjusted for concomitant use of calcium supplements with or without vitamin D (adjusted HR 1.01, 95% CI 0.49 to 2.08). In sensitivity analysis of acute coronary syndrome according to intention-to-treat, the results were consistent with the main analysis (HR 0.89, 95% CI 0.53 to 1.48).
Use of strontium was not associated with significantly increased risk of the secondary outcome of any-cause mortality (32 deaths during 1022 person-years among strontium users (rate 31.3 per 1000 person-years) vs 74 deaths during 3246 person-years among ibandronate users (rate 22.8 per 1000 person-years); adjusted HR 1.45, 95% CI 0.95 to 2.21).
This Danish population-based study of users of strontium ranelate found no association with a significantly increased risk of acute coronary syndrome compared with use of bisphosphonates. Adding robustness to the analyses, results were consistent both in the cohort of new users of antiosteoporotic drugs, comparing strontium with a first-line bisphosphonate, and in the cohort of patients who had switched from a first-line antiosteoporotic bisphosphonate, comparing strontium with a second-line bisphosphonate.
Strengths and weaknesses
We implemented a design with active comparators and with an analytic setup that aimed to capture two scenarios. In the first scenario, a classical new user design was applied comparing strontium ranelate users with users of one of the two first-line antiosteoporotic drugs in Denmark, alendronate or risedronate. Although such a design, that is, where all previous users of any of the drugs in the class are excluded, is often preferred in pharmacoepidemiology, it may suffer shortcomings when a second-line agent is compared with a first-line agent. That is, an active choice has been made to prescribe strontium rather than one of the first-line drugs, and the measured or unmeasured characteristics of patients who start a second-line agent as their first drug within a class may differ from those who start a first-line agent. Therefore, we also did a second analysis in which we first followed users of first-line antiosteoporotic bisphosphonates until they switched to a second drug and then compared the risk of acute coronary syndrome between strontium users and users of the second-line bisphosphonate ibandronate. These two analyses may be seen as complimentary and probably capture somewhat different groups of patients, with switchers perhaps representing patients who experienced adverse events or poor effectiveness from the first-line drugs. In addition, our use of a negative control, the second-line bisphosphonate in the cohort of new antiosteoporotic drug users, should limit concerns over unmeasured confounding associated with not choosing the first-line drug.
We used a disease risk score to capture and adjust for baseline cardiovascular risk. Additionally, the use of active comparators in a head-to-head design versus other drugs with the same indication and with no known signals of increased risk of ischaemic heart disease14 ,15 likely reduced the potential of selection bias and may have reduced confounding by unmeasured characteristics.16 Therefore, while residual confounding will always remain a possibility in observational research, we believe appropriate measures have been taken to reduce this. Data on smoking and body mass index, which are important risk factors for acute coronary syndrome, were not available for inclusion in the disease risk score. However, these factors are unlikely to be associated with choice of either strontium ranelate or a bisphosphonate and are therefore unlikely to confound the association. The number of events among patients with no history of cardiovascular disease was low, yielding limited power for the subgroup analyses according to cardiovascular disease.
The average time on treatment was shorter among users of strontium compared with users of bisphosphonates. Post hoc, this was found to be attributable to earlier discontinuation of treatment, more frequent occurrence of gaps between treatment episodes and higher probability of treatment switch among users of strontium. The shorter time on treatment among users of strontium is thus a result of the on-treatment exposure definition, allowing the course of treatment to influence the duration of follow-up, as opposed to traditional intention-to-treat designs. On the other hand, by disregarding time after treatment has ended, on-treatment designs have the advantage of being able to incorporate the circumstance that discontinuation of treatment may be on the pathway of a deteriorating health status; this may particularly apply to prophylactic treatment in an elderly population, where prophylaxis is likely given low priority in patients with declining health or acute illness. Results from sensitivity analyses with an intention-to-treat exposure definition were consistent with the primary on-treatment analyses.
Comparison with other studies
Our study was prompted by a recent assessment of the benefit–risk profile of strontium ranelate conducted by the European Medicines Agency. This assessment included a pooled analysis of randomised trials that showed an increased risk of myocardial infarction among postmenopausal women using strontium ranelate as compared with placebo (relative risk 1.6 (95% CI 1.07 to 2.38)) as well as two studies that showed an imbalance in serious cardiac events in men with osteoporosis and patients with osteoarthritis.3 Given that the findings of the pooled analysis are not published in the peer-reviewed literature, detailed comparisons with our study are not possible, but it can probably be assumed that the trials included in the pooled analysis are the trials on strontium efficacy with respect to osteoporotic fractures that have been published.1 ,17–20 Hence, the trials included in the pooled analysis were designed to investigate efficacy and not to analyse the risk of myocardial infarction, which may raise issues with study definitions of this outcome. Indeed, cardiovascular outcomes are not mentioned in the published studies.1 ,17–20 In the published report of what we believe to be the study of men with osteoporosis mentioned by the European Medicines Agency,3 there was an imbalance in the incidence of coronary artery disorder (8.7% strontium and 4.6% placebo).21 However, the distribution of history of myocardial ischaemia at baseline was skewed, being more common in patients assigned to strontium (10.3%) than in those assigned to placebo (3.4%).21 Cardiac events were not reported in the published report22 of what we believe to be the study of patients with osteoarthritis mentioned by the European Medicines Agency.3
Since the initial safety assessment was communicated by European Medicines Agency, two observational studies addressing the risk of myocardial infarction with use of strontium ranelate have been published. In a British case–control study, no increased risk of myocardial infarction was found among users of strontium ranelate compared with users of other antiosteoporotic agents (OR 1.05, 95% CI 0.68 to 1.61).7 Similarly, a recent cohort study in Denmark found no increased risk of myocardial infarction in women treated with strontium ranelate compared with other antiosteoporotic drugs (HR 1.05, 95% CI 0.79 to 1.41).6 In neither of these studies, however, did the definition of antiosteoporotic drug exposure differentiate between new users and those who had switched treatment. Given that strontium ranelate is mainly indicated as second-line drug, the lack of such a distinction may result in susceptibility to channelling bias, as fragile patients with relatively severe disease may preferentially be switched to strontium ranelate.23 A major strength of the present study is the well-characterised population in which we were able to assess the risk of acute coronary syndrome in, respectively, users of strontium ranelate who were treatment-naïve to antiosteoporotic agents and those who had switched to strontium ranelate from other antiosteoporotic treatment. As is evident from the present study, these two treatment groups represent two distinct populations that are relevant to study independently. As such, we believe the present paper adds to and expands on previous data.
This Danish nationwide cohort study of women aged 50–84 years found no significant association between use of strontium ranelate and the risk of acute coronary syndrome. Results were consistent through two analytic setups, one of new antiosteoporotic drug users and one of switchers from a first-line bisphosphonate, and helps support clinical and regulatory decision making.
This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.
Files in this Data Supplement:
- Data supplement 1 - Online supplement
Handling editor Tore K Kvien
Contributors All authors contributed to conception and design of the study, the analysis and interpretation of the study results, and critically revised the manuscript. HS acquired the data and conducted the statistical analyses. HS and BP drafted the manuscript. AH supervised the study and is the guarantor. All authors approved the final version of the manuscript for submission.
Competing interests None.
Provenance and peer review Not commissioned; externally peer reviewed.
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.