Introduction

Coronary artery disease (CAD) is the leading cause of morbidity and mortality worldwide, accounting for 17.9 million deaths, annually.1 Coronary artery calcium (CAC) is a highly specific marker of established atherosclerotic plaques2 and is attained from axial non-contrast computed tomography (CT) slices. It is reported as the modified Agatston score, which is expressed as Agatston units (AUs).3 CAC scoring has been shown to predict the risk of future cardiovascular events in asymptomatic patients.4

Bisphosphonates and NF-κB ligand (RANKL) inhibitors are medications typically indicated for the management of osteoporosis. Evidence suggests that there may be a role for their use in reducing the progression of CAD, via their effects on plaque formation.5 6 Bisphosphonates have been shown to inhibit the crucial regulatory enzyme, farnesyl pyrophosphate synthase in the mevalonic acid pathway, which is implicated in both bone and lipid metabolism, and thus alter the progression of ectopic calcification.5 6 RANKL inhibitors, including denosumab, interfere with the glycoprotein, osteoprotegerin and other signalling pathways, again involved in lipid metabolism. Furthermore, calcified plaques shown on unenhanced CT images are much alike to bone itself. Together, this information suggests that the fundamental underlying biochemical pathways involved in bone formation and vascular calcification are likely shared. The implication of such could mean an additional therapeutic target in managing CAD, which is relevant in those for whom traditional cardiovascular therapies are no longer sufficient to control disease progression.

Etidronate is one bisphosphonate whose effects upon CAD have been studied. Three studies have shown that etidronate may delay the progression of CAD, which has been measured through the surrogate endpoints of aortic calcification scores, CAC scores and carotid artery intima–media thickness, respectively.7–9 The effects of nitrogen-containing bisphosphonates (NC-BPs), however, on vascular calcification are contradictory in the literature. Some randomised controlled trials (RCTs) have suggested that alendronate, the most studied of the NC-BPs, is protective against CAD progression, again through a reduction in carotid intima–media thickness,10 11 and total volume of vascular calcification.12 13 These trials all contained less than 75 patients, with effects on vascular calcification largely observed in patients with chronic kidney disease (CKD) or in those receiving haemodialysis, only. Meanwhile, one small pilot study by Hill et al, showed that there was no significant difference in CAC progression between those receiving alendronate and placebo.14

There is very limited evidence assessing the role of the RANKL inhibitor, denosumab, in the progression of vascular calcification. A recent RCT revealed that after 12-month follow-up there was no significant difference in CAC and carotid artery intima–media thickness between those on denosumab versus control.15 Conversely, another study demonstrated that denosumab may indeed suppress the progression of CAC,16 although this was restricted to patients with secondary hyperparathyroidism. These conflicting data highlight the need for consolidation of the literature by means of a systematic review.

Over the last decade, two systematic reviews have been performed on similar topics. The first investigated the effects of bisphosphonates on multiple vessels, including the carotids, coronaries and aorta in patients undergoing haemodialysis.17 However, the review published 10 years ago included only two papers which investigated the effects of etidronate, on CAC specifically, in a highly selected population group, limiting the generalisability of the findings. The second, more recent study was also limited by the inclusion of small sample size studies and a short duration of follow-up. Additionally, no systematic review to our knowledge has explored the impact of denosumab on vascular calcification.

The primary aim of the systematic review was to evaluate the relationship between the use of bisphosphonates and the RANKL inhibitor, denosumab, with CAC. We hypothesised that there would be an inverse relationship between bisphosphonate and denosumab use and CAC. Furthermore, this review aimed to assess the relationship between these medications and aortic and carotid calcification through its secondary outcomes. If a true association between bisphosphonate or RANKL inhibitor use and CAC can be established in a large, diverse cohort of patients, it may warrant their use in those with elevated CAC. This could prove vital in both the primary and secondary prevention of cardiovascular events in those who are at high risk of severe complications.

Methods

The methods of this review have been published as a protocol18 and are outlined in brief (please see online supplemental file 1).

Eligibility criteria

Definitions as per Patient-Intervention-Comparator-Outcome-Duration (PICO-D) were adapted for the purpose of this review. An article was included in the study if it met the PICO-D criteria as outlined in table 1.

Table 1

PICO-D criteria for inclusion of studies in the review

Search strategy

A structured search of MEDLINE (inception – present), Embase (inception – present) and the Cochrane Central Register of Controlled Trials (CENTRAL) was performed. Citation lists of any relevant papers found were handsearched to identify further pertinent articles. The search strategy was developed by a medical librarian (online supplemental file 2), with search syntax altered as appropriate according to each database’s subject headings and thesaurus. Search keywords included CAC, bisphosphonates, RANKL inhibitors and denosumab.

Study selection

The articles yielded by the search were screened by title and abstract against our inclusion and exclusion criteria. Following initial title/abstract screening, the full text of potentially eligible papers was then appraised for final inclusion in the systematic review. A third reviewer adjudicated if there was a discrepancy in the inclusion status of any study. This process was documented in a Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) flow diagram.

Data collection

Two independent reviewers (NSM and KG) extracted data items from included reviews as per a standardised data extraction form (online supplemental file 3). Extracted data included study design, country and setting, aims and objectives, study population, intervention, control, outcomes, risk of bias and demographical data.

Outcomes

The primary outcome extracted was the difference in CAC from baseline to follow-up in patients who used either a bisphosphonate or denosumab compared with those who did not. The secondary outcomes of the review were carotid artery intima–media thickness and aortic calcification, both measured in cubic millimetres, in those using either a bisphosphonate or denosumab compared with placebo.

Risk of bias

Risk of bias was assessed using the ROBINS-I tool19 for non-randomised studies and the RoB2 tool20 for RCTs. The studies were graded as low, moderate or high for risk of bias per criterion and for overall bias.

Data synthesis and analysis

Studies were included if they fulfilled the eligibility criteria. Data were presented narratively and complemented with tables and figures as appropriate. The outcomes of each study were extracted, with their effect size measured by OR or relative risk and their corresponding 95% CIs, where available. The statistical significance reported by their p values was also collected where available, as stated by the study authors. Lastly, the main conclusions drawn by the authors were extracted. Meta-analysis was not performed due to the heterogeneity of included studies, which was assessed per the definition reported in the Cochrane Handbook for Systematic Reviews of Interventions.21 Heterogeneity in the included studies was demonstrated in their clinical variation that is, diverse populations studied with variations in treatment type and dose, and methodological variation as evidenced by study design (prospective observational vs RCT) and variance in reporting of the outcome measure (CAC reported in cubic millimetres vs AU), the latter of which most markedly precluded meta-analysis.

Patient and public involvement

None.

ResultsStudy inclusion

112 potentially relevant articles were identified from a systematic search of the literature, of which five met the review inclusion criteria (see figure 1); four observational studies and one RCT were included with a total of 377 patients.7 8 14 16 22 Individual study characteristics are documented in table 2. Three studies reported outcomes pertaining to bisphosphonate use, one on denosumab use and one study investigated the use of both.

Figure 1

PRISMA diagram demonstrating how studies were included in the review. PRISMA, Preferred Reporting Items for Systematic Reviews and Meta- Analyses; RANKL, NF-κB ligand.

Table 2

Study characteristics

Risk of bias in included studies

A summary of the risk of bias in each study can be found in tables 3 and 4 for observational studies and RCTs, respectively. Low overall risk inferred a low risk in each criterion. Moderate overall risk was determined by moderate risk in at least one criterion, without high risk in any criterion. High overall risk was determined by high risk in at least one criterion.

Table 3

Risk of bias in included studies for observational studies and randomised controlled trials

Table 4

Risk of bias in included studies for randomised controlled trials

Demographical data and population

Demographical outcomes included age, sex, presence of hypertension, diabetes, dyslipidaemia, CKD and smoking status (see table 5). There was a large proportion of patients who had end-stage renal disease on dialysis (three studies, n=112) and osteoporosis (one study, n=115). One study did not report any demographical baseline characteristics of participants except for age. One study reported only on age and sex. The remaining studies reported on each of the aforementioned characteristics. No studies reported on family history of CAD, and consequently, these data are not included in table 5.

Table 5

Baseline characteristics

Primary outcomesBisphosphonate use

Nitta et al (n=56) reported the action of the bisphosphonate, etidronate on CAC progression and demonstrated a statistically significant reduction in CAC (−372 mm3 over 12 months) compared with patients not taking etidronate (+159 mm3 over 12 months) (p<0.01). Ariyoshi et al (n=14) who also reported the effect of etidronate use on CAC progression demonstrated a difference between intervention (+880 mm3 over 12 months) vs control (+2220 mm3 over 12 months), however, no p value directly comparing the two groups was reported.

A third study by Hill et al (n=115) found no statistically significant difference between intervention with the bisphosphonate, alendronate (+2.4 modified AU per month) and control (+3.1 modified AU per month), p=0.46. Pawade et al (alendronate use, n=51; control n=50) similarly found no statistically significant difference between alendronate use (+326 (138–813) modified AU over 24 months) and control (+354 (76–675) modified AU over 24 months), p=0.49.

Denosumab use

Chen et al (n=42) investigated denosumab use on CAC progression and found a statistically significant reduction at 6-month follow-up between intervention (−133±124 modified AU) and control (+188±72 modified AU), p=0.03. Pawade et al (denosumab use, n=49; control, n=50) found no statistically significant difference between denosumab use (+343 (198–804) modified AU) vs control (+354 (76–675)) over a 24-month period, p=0.41. Primary outcome data can be seen in table 6. Meta-analysis was not performed due to limited, heterogeneous studies as previously discussed.

Secondary outcomesAortic calcification

Ariyoshi et al (n=14) reported the progression of aortic calcification in those receiving etidronic acid versus those who were not.7 Aortic calcification was found to improve in the intervention group (–64.1% (–6.5 to –50.1)) versus in the control group (+130% (2.1–414%)) over a 12-month period (p=0.006).

Carotid artery calcification

None of the included studies reported on carotid artery intima–media thickness in those using either a bisphosphonate or denosumab compared with a placebo.

Discussion

Optimisation of risk factors remains challenging for many patients with moderate to severe CAD, and subsequently, the search for additional therapeutic targets is needed. Our review collected data from observational studies and a single RCT and is the first to summarise the evidence exploring the role of bisphosphonates and RANKL inhibitors, respectively, in the progression of CAD as objectively measured via CAC.

Bisphosphonates and CAC

Four studies investigated the action of bisphosphonates on CAC progression. Nitta et al’s study was the only study to report a statistically significant reduction in CAC progression in patients receiving the first-generation bisphosphonate, etidronic acid, which was given in doses of 200 mg/day for 14 days, every 90 days for a total of three cycles, compared with control in patients with end-stage CKD.8 Furthermore, the findings reported by Ariyoshi et al,7 also in a cohort of haemodialysis patients receiving etidronate but at a higher dose of 400 mg/day for 24 weeks, are worth noting; while the authors did not report any statistical hypothesis result, a large difference in CAC of 1340 mm3 between the groups was observed at 1-year follow-up. This would support the notion that etidronic acid is capable of reducing CAC in patients with CKD. The remaining studies in this category investigated the NC-BP, alendronate, in a broader range of individuals which could explain the differences in outcomes seen. This also correlates to the difference in molecular structure between first-generation bisphosphonates and NC-BPs, with etidronate theorised to have a greater capacity for inhibition of crystallisation and thus calcification of soft tissues.23 Furthermore, a literature review showed that inhibition of soft tissue calcification by bisphosphonates is likely restricted to etidronate.24 However, the review warns that etidronate poses a risk of causing osteomalacia in patients taking it at the high doses recommended to suppress vascular calcification. Conversely, other reviews25 26 suggest that etidronate is safe if given cyclically (as was performed in the Nitta et al, protocol) up to a dose of 400 mg daily. Osteomalacia was not reported in the 2–3 years follow-up period.26 Continuous therapy at these doses, however, is not recommended due to intolerable gastrointestinal side effects and subsequent poor adherence. Given that etidronate displayed benefit at 200 mg daily in the Nitta et al, paper, we posit that this could be an optimal starting dose that also allows for up-titration in the management of vascular calcification to balance the risks and benefits of treatment. This could be an area of future research.

RANKL inhibitors and CAC

Evidence regarding the action of denosumab on CAC progression also remains conflicting, with one study reporting a statistically significant hypothesis result between intervention and control, and one study demonstrating no statistically significant difference in this review. RANKL inhibitors are comparably novel to bisphosphonates, and subsequently, research into their use is not as ubiquitous. Denosumab’s ability to inhibit vascular calcification has been proven both in vitro27 and in vivo in mice models.28 In the latter study, osteoporosis was induced by prednisone in human RANKL knock-in mice, who became responsive to denosumab, unlike their wild-type counterparts. Subsequent denosumab therapy resulted in attenuation of aortic calcium deposition up to 50% compared with the control group. In humans the data remains conflicting. An RCT conducted in haemodialysis patients by Iseri et al, did not demonstrate a statistically significant difference in CAC progression from baseline to follow-up at 12 months. Notably, only 18 patients reached follow-up time. Additionally, this study did not include a control group and hence did not meet the eligibility criteria for inclusion in the present review. In contrast, a larger, more recent study published by Suzuki et al 29 conducted over a 30-month period demonstrated the capabilities of denosumab in reducing aortic arch calcification in haemodialysis patients. Given the narrow patient selection criteria, the results may not be generalisable to a wider cohort. However, these findings corroborate those published by Chen et al,16 suggesting that denosumab may have selective utility to improve vascular calcification in those with end-stage CKD.

CAC progression versus prevalence

Notably, the review included only papers which investigated the impact of antiosteoporotic medications on CAC progression over time. Hence, studies evaluating the point prevalence of CAC in bisphosphonate or denosumab users were excluded. The largest study to date investigating bisphosphonate use on the prevalence of atherosclerosis is the Multi-Ethnic Study of Atherosclerosis study.30 The study included over 3000 women from diverse ethnic backgrounds and found that bisphosphonate use was associated with a reduced prevalence of overall cardiovascular calcification in women over the age of 65. However, this was not statistically significant for CAC, which is supported by our review. While men made up the bulk of participants in our review, the mean age of participants was alike, which may explain the similarity to the review’s findings.

Antiosteoporotics and calcification in other vasculature

To the authors’ knowledge, the RCT conducted by Pawade et al is the first of its kind to apply CAC scoring to measure aortic valve stenosis (AS) in the context of bisphosphonate and denosumab use.22 The study did not find a statistically significant result between intervention and control in either case. Other papers have reported the effect of bisphosphonates on the progression of AS but measured via aortic valve area and mean and peak gradients. Using these parameters, the literature remains conflicting, with some evidence to suggest the slowing of AS by bisphosphonates,31 32 and other evidence contradicting such an effect.33

One study reported on the review’s secondary outcome of aortic calcification.7 A statistically significant reduction in aortic calcification in the etidronic acid group compared with control was demonstrated; whilst not statistically significant in the coronary arteries, a reduction was still noted. Whilst the underlying mechanism of calcification remains analogous between the two arteries, the differences in intimal and medial histological elastic and smooth muscle fibre composition34 as well as differences in shearing forces between the vessels could theoretically result in delayed improvement in calcification measurements over the given timeframe.35

Strengths and limitations

This review is likely to have captured most, if not all papers reporting on the topic of CAC progression in antiresorptive medication use, through its use of (1) inclusion of multiple study designs, (2) inclusion of wide study populations, that is, no limitation according to sex or presence of comorbidities and (3) a methodical search of three large databases and citation lists of relevant papers.

Meta-analysis was unable to be performed in view of the small number of heterogeneous papers eligible for inclusion in the review. Furthermore, three of the four studies were shown to have a moderate risk of bias which may limit the interpretation of the data. Two studies were undertaken in patients with end-stage renal failure on dialysis, specifically, and one in patients with concomitant secondary hyperparathyroidism with normocalcaemia. Therefore, these studies may not be generalisable to the broader population, particularly in those with alternate pathological mechanisms to cause their osteoporotic bony disease and simultaneous vascular calcification.

Future directions

Further research on etidronate in renal failure patients appears warranted, given previous promising results in this demographic. Larger scale studies which aim to balance therapeutic dosing with adverse events appear worthwhile. Additionally, data pertaining to the long-term use of both first generation and nitrogen-containing bisphosphonates, and RANKL inhibitors in patients with and without renal disease would be useful to establish the temporal relationship between antiosteoporotic medications and CAC. Large-scale RCTs would be most beneficial to draw valid conclusions.