Journal of the American College of Cardiology
Do Bisphosphonates Slow the Progression of Aortic Stenosis?
Author + information
- Received August 10, 2011
- Revision received December 30, 2011
- Accepted January 2, 2012
- Published online April 17, 2012.
Author Information
- Olcay Aksoy, MD⁎,
- Akin Cam, MD†,
- Sachin S. Goel, MD⁎,
- Penny L. Houghtaling, MS‡,
- Sarah Williams, MS‡,
- Ernesto Ruiz-Rodriguez, MD†,
- Venu Menon, MD⁎,
- Samir R. Kapadia, MD⁎,
- E. Murat Tuzcu, MD⁎,
- Eugene H. Blackstone, MD‡ and
- Brian P. Griffin, MD⁎,⁎ (griffib{at}ccf.org)
- ↵⁎Reprint requests and correspondence:
Dr. Brian Griffin, Cleveland Clinic, 9500 Euclid Avenue/J1-5, Cleveland, Ohio 44195
Abstract
Objectives The aim of this study was to investigate the impact of bisphosphonates on the progression of aortic stenosis.
Background Valvular calcification is associated with the development and progression of aortic stenosis. Bisphosphonates have been suggested to slow this progression.
Methods Female patients older than the age of 60 years with an aortic valve area (AVA) between 1.0 and 2.0 cm2 were identified and studied retrospectively. Only those who had follow-up echocardiograms at least a year apart were included. Primary outcomes were the change in AVA and valvular gradients over time. Mortality and freedom from aortic valve replacement were also studied. A propensity-matching method was applied for the probability of the use of bisphosphonates.
Results The study included 801 female patients (mean age, 76 ± 7.6 years) with a mean follow-up of 5.1 ± 2.4 years. The mean duration of bisphosphonate use was 3.1 ± 2.6 years. At the time of the initial echocardiogram, 323 patients (38%) were taking bisphosphonates. The mean ejection fraction at baseline was 56.7 ± 9.6% with a mean AVA of 1.32 ± 0.25 cm2. Peak and mean gradients were 28.4 ± 11 mm Hg and 15.6 ± 6.8 mm Hg, respectively. Propensity matching was successfully performed for 438 patients. On follow-up, there were no differences in the rate of change in AVA or peak and mean gradients when patients were stratified based on the use of bisphosphonates. Bisphosphonates also had no impact on survival or freedom from aortic valve replacement.
Conclusions In this retrospective analysis of older female patients, bisphosphonates do not have a significant impact on the hemodynamic or clinical progression of aortic stenosis.
Aortic stenosis (AS) in the elderly is commonly associated with leaflet calcification and thickening leading to progressive obstruction of the aortic valve (1–6). As symptoms ensue, the outcomes of these patients are poor, with a mean survival of 2 years (7–10). The only intervention proven to alter disease course is aortic valve replacement (AVR), which normalizes the life expectancy to that of age- and sex-matched controls (11). Because the disease becomes more prevalent with the aging population, strategies to slow down and reverse the progressive stenosis of the valve become important, both from patient and population standpoints. Such efforts should be based on a thorough study of the histopathology and the mechanisms of progression of AS. Retrospective studies of the use of statin therapy to slow AS progression suggest a benefit in hyperlipidemic patients, but prospective, randomized studies have not demonstrated a significant favorable effect of statins on AS progression (12–17).
Calcification has been shown to have a central role in the progression of AS along with inflammation (1,3–5). Although several trials have evaluated the role of statins in altering the inflammatory process of the valve, calcification has not been clearly targeted. Bisphosphonates, drugs that are approved for use in patients with osteoporosis, have been shown to inhibit vascular calcification (18–21). Furthermore, several small observational retrospective studies have shown a possible link between the use of bisphosphonates and slowing of AS progression (22–24). More recently, a cross-sectional study of nitrogen-containing bisphosphonate therapy found an association with a lower prevalence of cardiovascular calcification in women older than the age of 65 years (25).
In our study, we sought to investigate the use of bisphosphonates in a large patient population to study the longitudinal effects of these agents on the progression of AS.
Methods
Patient population
Women older than the age of 60 years who were found to have mild to moderate AS (defined as an aortic valve area [AVA] between 1.0 and 2.0 cm2) were included in the study. Those who did not have follow-up echocardiograms at least a year apart were excluded. Study subjects were identified from the computerized echocardiography database at the Cleveland Clinic from January 2000 to January 2009. This study was performed after approval from the Institutional Review Board of the Cleveland Clinic (IRB #09-913).
Baseline characteristics including hypertension, hyperlipidemia, chronic kidney disease, diabetes mellitus, and coronary artery disease were obtained from electronic medical records. The medications used by the patients including bisphosphonates, angiotensin-converting enzyme inhibitors, angiotensin receptor blockers, beta-blockers, statins, vitamin D, and calcium supplements at the time of the index echocardiogram were extracted from the electronic medical records. Echocardiographic data included left ventricular ejection fraction, AVA, aortic peak and mean gradients, and the presence and severity of mitral regurgitation, tricuspid regurgitation, and right ventricular systolic pressure. The effective AVA was obtained on 2-dimensional echocardiography using the continuity equation. Peak and mean gradients across the aortic valve were obtained using pulse and continuous wave Doppler evaluation of the aortic valve. Left ventricular ejection fraction was calculated using a combination of the biplane Simpson method and 3-dimensional measurements incorporating the visual estimation as assessed by the staff specialized in reading these studies. Aortic valve calcification was determined via a calcification score depending on severity (0 to 3: none, mild, moderate, and severe, respectively) based on the semiquantitative assessment by the staff interpreting the imaging. Images and Doppler data were acquired by an experienced sonographer and subsequently interpreted by staff with expertise in echocardiographic imaging. Occurrence of AVR was adjudicated using the databases kept at our center. Mortality was assessed using Social Security Death Index.
Statistical analysis
The primary endpoint to be assessed was the change in AVA and in peak and mean gradients over time. Secondary endpoints evaluated were all-cause mortality and freedom from AVR.
Continuous variables are expressed as mean ± SD and categorical variables as proportions. Continuous variables were analyzed with a 2-sided Student t test, and categorical variables were compared with the chi-square test. The design of data collection ensured that the variables considered in the analysis were complete. Missing values were imputed via mean value replacement. If a variable had transformations, then that transformation was done on the imputed value. If a variable was missing >50% of the time, it was excluded from analysis.
After determining the use of bisphosphonates across the cohort, patients were stratified based on the bisphosphonate use. Two separate analyses were performed: 1) unadjusted analysis of the entire cohort after stratification by bisphosphonate use, and 2) adjusted analysis after propensity matching, in which patients not taking bisphosphonates were matched to those taking bisphosphonates after solving a saturated model for probability of bisphosphonate use.
AVA was analyzed longitudinally for change over time. A nonlinear mixed model was used to resolve a number of time phases on a cumulative odds domain to form a temporal decomposition model and to estimate the shaping parameters at each phase. Longitudinal cumulative logistic regression for repeated measurements (SAS PROC NLMIXED, SAS Institute, Cary, North Carolina) was used to implement the temporal decomposition model. In this analysis, all follow-up echocardiograms were factored into the analysis (ranging from 1 to 11 studies per patient). Initially, the longitudinal model was used to find a trend in the data during follow-up. Using this initially developed model, we accounted for the nonindependence of the observations for each person, creating a separate model for each patient. The final model incorporated the best possible algorithm that provided the best fit for all patients. The effect of bisphosphonates on the need for AVR and survival was studied with the Kaplan-Meier nonparametric method. Survival estimates were then estimated parametrically by a multiphase hazard model. The parametric model was used to resolve a number of phases of instantaneous risk of death (hazard function) and to estimate shaping parameters. To determine the difference between group risk factors for death, the indicator of bisphosphonate use was then added to the multiphase hazard function domain. During all computations of survival and freedom from AVR, the follow-up period was divided into 2 phases (early and late) to allow for the time-dependent effects of bisphosphonates.
Propensity matching
Using variables known at the time of study entry and multivariable logistic regression, factors associated with the use of bisphosphonates were determined. Subsequently, a propensity score was calculated for each patient by solving the saturated model for the probability of bisphosphonate use. Using only the propensity score, patients without a history of bisphosphonate use were matched to patients with a history of bisphosphonate use using a greedy matching strategy (26). With this strategy, a bisphosphonate propensity score was considered matched to the closest nonbisphosphonate propensity score within a difference of 0.1. A nonbisphosphonate patient who was matched was no longer eligible for consideration. This algorithm was repeated until all patients taking bisphosphonates were matched or all propensity scores deviated >0.1 between the groups. Patients with a history of bisphosphonate use whose propensity scores deviated >0.1 from the scores of those without a history of bisphosphonate use were considered unmatched. All analyses were performed using SAS statistical software versions 8.2 and 9.1 (SAS Institute).
Results
Baseline variables and analysis of the unmatched cohort
There were 801 female patients in the echocardiography database who had an AVA between 1.0 and 2.0 cm2 and older than 60 years of age who also had follow-up echocardiograms at least 1 year after the index study.
Of the total cohort of 801 patients, 313 (39%) were taking bisphosphonates and 488 (61%) were not at the time of the first echocardiogram. The mean duration of use of bisphosphonates was calculated from the difference between the first documentation of use and the last verified date of use in the medical records, giving a mean duration of 3.1 ± 2.6 years. The severity of calcification based on mean calcification score across both cohorts at the time of index echocardiogram was similar (p = 0.45). Analysis of the baseline variables showed that patients who were taking bisphosphonates were more likely to be older and have hyperlipidemia and osteoporosis. Furthermore, if patients were taking bisphosphonates, they were also more likely to be taking vitamin D, calcium supplements, calcitonin, and selective estrogen receptor modulators. The baseline echocardiographic parameters were similar (Table 1).
Baseline Characteristics of the Entire Cohort (N = 801)⁎
In the entire group, the change in AVA over time was analyzed with a median follow-up period of 1.6 years (range, 0 to 9.2 years) after the qualifying echocardiogram (first study after the index echocardiogram). The mean AVA at the time of the index echocardiogram was 1.32 ± 0.25 cm2. Findings were consistent with an overall rate of change that was nonlinear in nature. There appeared to be a higher rate of change for the first 3 years, which then stabilized at 0.05 cm2 per year after 3 years.
When patients who were taking bisphosphonates and those who were not were compared, there was no significant difference in the rate of progression of the AVA when change was analyzed over time (p = 0.87). The change in peak and mean gradients was also similar across strata (p = 0.75 and 0.43, respectively). Survival data were obtained with a mean follow-up of 3.1 ± 2.3 years and a total of 2,458 patient-years (with time 0 defined as the time of the qualifying echocardiogram). Unadjusted analysis comparing survival between the bisphosphonate groups showed no significant difference (p = 0.53 and 0.70 for early and late phases of follow-up, respectively). Occurrence of AVR was also compared across study groups with a mean follow-up of 1.5 ± 1.9 years (time 0 defined as the time of the qualifying echocardiogram). This analysis showed no difference across bisphosphonate groups with respect to freedom from AVR (p = 0.33 and 0.55 for early and late phases, respectively).
Analysis after propensity matching according to the use of bisphosphonates
From the cohort of 801 patients, a total of 438 patients were matched according to the use of bisphosphonates (219 patients in each group). After matching, all measured baseline comorbid conditions, medications, and echocardiographic parameters were similar (Table 2). The severity of calcification at the time of the index echocardiogram was similar in both groups (p = 0.32). The C-statistic of the propensity-matching model was 0.79.
Baseline Characteristics of the Matched Cohort of Patients (n = 438)
In the matched cohort, the change in AVA over time was analyzed and showed no significant difference in the rate of progression of AVA across groups with a median follow-up of 1.6 years (p = 0.34) (Fig. 1A). This finding was also true for the similarity in change in peak and mean gradients (p = 0.79 and 0.20, respectively) (Figs. 1B and 1C). Survival analysis did not show a significant difference between the bisphosphonate use groups for a mean follow-up of 3.1 ± 2.3 years (p = 0.94 and p = 0.11 for early and late phases during follow-up, respectively) (Fig. 2A). Analysis evaluating freedom from AVR showed no difference across bisphosphonate groups for a mean follow-up of 1.5 ± 1.9 years (p = 0.26 and p = 0.98 for early and late phases, respectively) (Fig. 2B).
Effect of Bisphosphonate Use on Aortic Stenosis Progression
Analysis of the matched cohort of patients stratified by their bisphosphonate use does not demonstrate any differences in change in aortic valve (AV) area (p = 0.34) (A), peak AV gradient (p = 0.79) (B), or mean AV gradient (p = 0.20) (C), as demonstrated by Kaplan-Meier plots. (The number of patients included at each time point is shown in the first row and number of echocardiograms [Echo] shown in the second row.)
Effect of Bisphosphonate Use on Survival and Freedom From AVR in the Propensity Matched Cohort
Analysis of the matched cohort of patients stratified by their bisphosphonate use does not demonstrate any differences in survival (p = 0.11) (A) or freedom from aortic valve replacement (AVR) (p = 0.98) (B). (The number of bisphosphonate users at each time point is shown in the first row and nonusers in the second row.) Echo = echocardiograms.
Analysis of patients who had AVR during follow-up
We studied the subset of patients who had AVR during follow-up to evaluate for the role of bisphosphonates in disease progression in this patient subset (n = 118). Of these patients, 77 were not taking bisphosphonates, with the remaining 41 taking these agents. In the entire cohort of 118 patients, the rate of progression of AS as measured by change in AVA was high in the first year (0.17 cm2), which then slowed in the subsequent years.
Analyzed with respect to the use of bisphosphonates (median follow-up of 1.9 years), both treatment cohorts had similar progression of their AS as measured by AVA over time (p = 0.07) (Fig. 3A) with their time course of progression to AVR also being similar (p = 0.66) (Fig. 3B).
Effect of Bisphosphonate Use on Progression of Aortic Stenosis and Survival in Patients Undergoing AVR
Analysis of cohort of patients who underwent aortic valve replacement (AVR) during follow-up does not show any differences in change in aortic valve area (p = 0.07) (A) or time course to AVR (p = 0.66) (B). (For A, the number of patients included at each time point is shown in the first row and the number of echocardiograms [Echo] is shown in the second row. For B, the number of bisphosphonate users at each time point is shown in the first row and nonusers in the second row.)
Discussion
In this study of older women (older than 60 years of age) with mild to moderate AS at baseline, we found that the use of bisphosphonates did not affect the hemodynamic progression of valve stenosis. Furthermore, the use of bisphosphonates did not affect the survival or the rate of AVR during follow-up. The study also illustrates that the rate of progression of AS in this population is not linear but tends to lessen over time.
Our understanding of AS, a disease once thought to result from a passive process of “wear-and-tear” of the valve, has evolved in recent years (1,3–5,10,22). Now understood to be the result of an active process of chronic inflammation involving the renin-angiotensin system, lipid accumulation, and calcium deposition, AS has been targeted by several drugs in an effort to slow its progression. Statins have been used to modify the chronic inflammation involved in the disease progression with disappointing results in nonhyperlipidemic patients. Although retrospective data analyses had suggested that the use of statins would be associated with slowing of disease progression, randomized, controlled trials have not demonstrated such clear benefit (12–17). Angiotensin-converting enzyme inhibitors have also been evaluated in retrospective studies with conflicting results (27,28). As yet, no therapeutic option is available to slow the rate of progression of stenosis in patients with mild to moderate AS.
Calcification has been shown to be an integral part of disease onset and progression in the histopathologic studies of AS. Epidemiological studies have associated the increased presence of calcium with adverse clinical outcomes as well (29–31). As such, retarding, inhibiting, or reversing the calcification of the valve might provide an avenue for treatment of patients with mild to moderate AS. As an active process, calcification of the aortic valve has been associated with pathways that use osteopontin, bone morphogenic proteins, and receptor activator of nuclear factor-κB ligand (32–34). Through complex interactions of these molecules with those of lipid pathways and fibroblasts, initial calcific nodules have been shown to form in regions of lipid accumulation (5). Bisphosphonates are active in 2 pathways, which potentially could affect the pathophysiology in AS. Nitrogen-containing bisphosphonates have been shown to inhibit farnesyl-pyrophosphate synthase, an enzyme in the cascade in which ultimately statins exert their effects (35). As a result, bisphosphonates have an effect similar to that of the statins, by affecting lipid metabolism and inflammation. Furthermore, bisphosphonates also prevent bone resorption and slow the release of calcium phosphate particles from the bone, which might play a role in retarding calcium deposition in vascular and valvular tissues (35,36). These effects of bisphosphonates have actually been evaluated in several small case series in which they were shown to slow the progression of AS (22–24). However, these case series have included a total of 234 patients with only 54 patients taking bisphosphonates, precluding the generalizability of these results to a broader population.
Given the increasing prevalence of AS, medical therapies that slow or reverse the progression of disease are urgently needed from patient and population standpoints. In our study, we evaluated the use of bisphosphonates in women older than the age of 60 years with mild to moderate AS and showed that bisphosphonates were not associated with any beneficial effects in this patient population. Specifically, it did not have any effect on the change in AVA over time, survival, and freedom from AVR. Furthermore, bisphosphonates did not delay surgery or slow the rate of narrowing of the valve in patients who eventually underwent AVR. This is in contrast to earlier case series, and although the discrepancy can be partially explained by different patient populations and the disease severity studied, the larger sample size presented in this study is likely more reflective of the role of bisphosphonates in the progression of AS.
This study is the largest retrospective analysis to date and argues against the use of bisphosphonates in older women to prevent the progression of AS. The potential reasons for the negative results are multiple: Complex mechanisms are involved in the regulation of calcium phosphate metabolism, and bisphosphonates, effects on bone tissue may not be replicated in the valve. Furthermore, the anti-inflammatory effects of biphosphonates may not play a significant role in the regulation of valvular inflammation. We also hypothesized that the effects of potential positive effects of bisphosphonates on AS progression might be offset by the concomitant use of agents such as supplemental calcium and vitamin D, which may accelerate the calcification process. However, the propensity analysis (Table 2) allowed us match for the use of these ancillary agents in both groups and despite adequate matching of these agents, no effect of bisphosphonates on progression was evident.
Many previous studies that have examined the progression of AS showed that the rate of change tends to be relatively constant over time; however, these studies evaluated the progression over a relatively short time frame and used linear analysis, often using the annualized change in valve area or pressure gradients as the mode of expression of progression (37,38). Our study is novel in that it incorporated multiple echocardiograms in individual patients over a longer period, suggesting that progression of AS, at least in this population of women with mild to moderate AS, does not have a true linear progression, but rather the progression tends to slow after an early period of acceleration. Future studies of longer term progression of AS should take into account the likelihood of a nonlinear change in both valve area and pressure gradients over time.
Study limitations
This study has several limitations. Its retrospective nature exposes the data to unidentified sources of bias, which might have led to findings of reduced efficacy of bisphosphonates. We tried to circumvent this problem with propensity matching. However, we cannot account for unmeasured variables by using this approach. We do not have data on patient compliance with taking bisphosphonates throughout the course of follow-up, nor did we match for changes in therapy during follow-up. Furthermore, bisphosphonates are a large group of drugs, each with its own specific characteristics and affinities to distinct pathways in the calcification and inflammatory process. As such, categorizing all these drugs as 1 group might have prevented us from identifying positive signals in the data. To further elucidate the role of these drugs, research that evaluates the role of distinct bisphosphonates on valvular calcification is necessary, together with prospective clinical studies to assess for their efficacy in preventing the progression of AS.
Conclusions
In summary, in this largest study to date of the effect of bisphosphonates on valvular disease progression, we show that bisphosphonates do not have a significant impact on the progression of AS. We believe that further studies are necessary to better understand the mechanisms of calcification and inflammation in AS to identify appropriate therapeutic targets and drugs to modify the natural course of disease.
Footnotes
This study was supported in part by the Dellovade Family Fund, Canonsburg, Pennsylvania, and the Mareb Foundation, Jupiter, Florida. All authors have reported that they have no relationships relevant to the contents of this paper to disclose.
- Abbreviations and Acronyms
- AS
- aortic stenosis
- AVA
- aortic valve area
- AVR
- aortic valve replacement
- Received August 10, 2011.
- Revision received December 30, 2011.
- Accepted January 2, 2012.
- American College of Cardiology Foundation
References
- ↵
- Stewart B.F.,
- Siscovick D.,
- Lind B.K.,
- et al.
- O'Brien K.D.
- Parolari A.,
- Loardi C.,
- Mussoni L.,
- et al.
- ↵
- Freeman R.V.,
- Otto C.M.
- ↵
- Turina J.,
- Hess O.,
- Sepulcri F.,
- Krayenbuehl H.P.
- Horstkotte D.,
- Loogen F.
- Bonow R.O.,
- Carabello B.A.,
- Chatterjee K.,
- et al.
- ↵
- Lindblom D.,
- Lindblom U.,
- Qvist J.,
- Lundstrom H.
- ↵
- Novaro G.M.,
- Tiong I.Y.,
- Pearce G.L.,
- Lauer M.S.,
- Sprecher D.L.,
- Griffin B.P.
- Moura L.M.,
- Ramos S.F.,
- Zamorano J.L.,
- et al.
- Chan K.L.,
- Teo K.,
- Dumesnil J.G.,
- Ni A.,
- Tam J.
- ↵
- Price P.A.,
- Faus S.A.,
- Williamson M.K.
- Kramsch D.M.,
- Chan C.T.
- ↵
- ↵
- Elmariah S.,
- Delaney J.A.,
- O'Brien K.D.,
- et al.
- ↵
- ↵
- Rosenhek R.,
- Rader F.,
- Loho N.,
- et al.
- ↵
- Messika-Zeitoun D.,
- Aubry M.C.,
- Detaint D.,
- et al.
- ↵
- Bevilacqua M.,
- Dominguez L.J.,
- Rosini S.,
- Barbagallo M.
- ↵
- ↵
- Otto C.M.,
- Burwash I.G.,
- Legget M.E.,
- et al.
- Otto C.M.,
- Pearlman A.S.,
- Gardner C.L.