Prevention of bone loss and vertebral fractures in patients with chronic epilepsy—Antiepileptic drug and osteoporosis prevention trial

Authors

  • Antonio A. Lazzari,

    Corresponding author
    1. Primary Care Service, Boston VA Healthcare System, Boston, Massachusetts, U.S.A
    2. Osteoporosis Clinic, Boston VA Healthcare System, Boston, Massachusetts, U.S.A
    3. Rheumatology Section, Boston VA Healthcare System, Boston, Massachusetts, U.S.A
    4. Department of Medicine, Boston VA Healthcare System, Boston, Massachusetts, U.S.A
    5. Department of Medicine, Boston University School of Medicine, Boston, Massachusetts, U.S.A
    6. Department of Medicine, Harvard Medical School, Boston, Massachusetts, U.S.A
    • Address correspondence to Antonio A. Lazzari, Boston VA Healthcare System, 120 South Huntington Avenue, Suite F2-29, Boston, Massachusetts 02130, U.S.A. E-mail: antonio.lazzari@va.gov

    Search for more papers by this author
  • Philip M. Dussault,

    1. Pharmacy Services, Boston VA Healthcare System Osteoporosis Clinic, Boston, Massachusetts, U.S.A
    Search for more papers by this author
  • Manisha Thakore-James,

    1. Neurology Department, Boston VA Healthcare System, Boston, Massachusetts, U.S.A
    Search for more papers by this author
  • David Gagnon,

    1. MAVERIC, Boston VA Healthcare System, Boston, Massachusetts, U.S.A
    2. Department of Biostatistics, Boston University School of Public Health, Boston, Massachusetts, U.S.A
    Search for more papers by this author
  • Errol Baker,

    1. Research and Development Department, Boston VA Healthcare System, Boston, Massachusetts, U.S.A
    Search for more papers by this author
  • Samuel A. Davis,

    1. Research and Development Department, Boston VA Healthcare System Osteoporosis Clinic, Boston, Massachusetts, U.S.A
    Search for more papers by this author
  • Antoun M. Houranieh

    1. Pharmacy Services, Boston VA Healthcare System, Boston, Massachusetts, U.S.A
    Search for more papers by this author

Summary

Purpose

To evaluate whether use of a bisphosphonate (risedronate) in addition to calcium and vitamin D in male veterans with epilepsy who were taking antiepileptic drugs (AEDs) long term can prevent the loss of bone mass (BMD, bone mineral density) associated with AED use compared to patients who were treated with a placebo plus calcium and vitamin D. As a secondary end point we studied the incidence of new morphometric vertebral and nonvertebral fractures.

Methods

Antiepileptic drug and osteoporosis prevention trial (ADOPT) was designed as a prospective 2-year double-blind, randomized placebo controlled study involving 80 male veterans with epilepsy who were being treated with AEDs such as phenytoin, phenobarbital, sodium valproate, or carbamazepine for a minimum of 2 years. All enrolled participants received calcium and vitamin D supplementation, and were randomized to risedronate or matching placebo. Total body, bilateral proximal femora, and anteroposterior (AP) lumbar spine BMDs in addition to morphometric lateral vertebral assessments (LVAs) were evaluated by a dual energy x-ray absorptiometry (DXA) instrument. Comparisons of BMDs were made between baseline, 1 year, and after 2 years of enrollment in the study. The incidence of new vertebral and nonvertebral fractures was secondary end point.

Key Findings

Of the 80 patients initially enrolled in the study, 53 patients completed the study. Baseline characteristics of the two groups were similar. At the end of the study, in the placebo plus calcium and vitamin D group, we observed a significant improvement in BMD at any of the evaluated sites when compared to their baseline scans in 69% (18/26) of the participants. In the risedronate plus calcium and vitamin D group, we observed significant improvement of BMDs in 70% (19/27) of the participants. At the end of the study, the risedronate group experienced a significant increase of BMD at the lumbar spine L1-4 (1.267–1.332 g/cm2), which was significantly larger than that seen in the placebo group) (1.229 g/cm2 vs. 1.245 g/cm2; p = 0.0066).There were nonsignificant differences between the two groups regarding changes of total body BMD or at the proximal bilateral femora. Five new vertebral fractures and one nonvertebral fracture were observed only in the placebo group.

Significance

Calcium and vitamin D supplementation or calcium and vitamin D supplementation in addition to risedronate improved BMD in more than 69% of male veterans with epilepsy who were taking AEDs. In the group receiving risedronate plus calcium and vitamin D there was a significant improvement of BMD at the lumbar spine as compared to the placebo group, which also received calcium and vitamin D. The use of risedronate plus calcium and vitamin D prevented the incidence of new vertebral fractures and one nonvertebral fracture in this cohort.

It is well recognized that chronic use of antiepileptic drugs (AEDs), both cytochrome P450 (CYP) enzyme-inducing such as phenytoin, phenobarbital, carbamazepine, and primidone or non–enzyme-inducing AEDs such as valproate, are known to be associated with accelerated rate of bone loss and development of secondary osteoporosis and consequently osteoporotic fractures (Kruse, 1968; Dent et al., 1970; O'Hare et al., 1980; Chung & Ahn, 1994; Cummings et al., 1995; Sheth et al., 1995; Aboukasm & Smith, 1997; Vestergaard et al., 1999; Sato et al., 2001; Lotfizadeh & Montouris, 2004; Carbone et al., 2010). Although some studies have suggested that the increased prevalence of fractures in this population was related to the seizure activity (Aboukasm & Smith, 1997; Persson et al., 2002), other studies have suggested that chronic use of AEDs was an independent risk factor for fractures (Nilsson et al., 1986; Cummings et al., 1995; Vestergaard et al., 2004; Vestergaard, 2005). Clinical studies have confirmed that >50% of adults on AEDs have decreased bone density of either the hip or the spine (Farhat et al., 2002; Lotfizadeh & Montouris, 2004); however, another cross-sectional study suggested that the increased prevalence of vertebral and nonvertebral fractures could not be directly correlated with changes in bone mineral density (BMD) over time but was associated with the number and type of AEDs used (Carbone et al., 2010).

Other unique risk factors, including the use of AEDs with sedative effect, have been described as playing important roles in increasing the risk of fractures in the epileptic population (Gidal & Sheth, 2004; Mattson & Gidal, 2004). Lack of sun exposure, excessive alcohol and tobacco use, and poor dietary habits, are also considered to be responsible for the increased prevalence of osteoporosis in both the male and female epileptic population (Ensrud et al., 2004, 2008).

In a retrospective study of 750 patients with epilepsy who sustained a fracture over a period of 7 years, the authors concluded that age and gender played a role in the incidence of pathologic or traumatic fractures in that population (Sheth et al., 2006).

Although evidence-based clinical guidelines do not exist, several authors recommend supplementation with calcium and vitamin D in this population. In a randomized study, Mikati et al. (2006) studied two different doses of vitamin D and concluded that high-dose vitamin D (4,000 units) resulted in increased bone density at several skeletal sites in adults.

However, in a retrospective study, calcium and vitamin D supplementation did not appear to influence the prevalence of fractures in 3,303 veterans who were receiving AEDs (Espinosa et al., 2011).

To the best of our knowledge no randomized clinical trials have been published on the use of a bisphosphonate in the prevention of osteoporosis and osteoporotic fractures in male veterans with epilepsy who were taking anticonvulsants (Ali et al., 2004; Gidal & Sheth, 2004; Pack & Morrell, 2004; Lee et al., 2010).

We hypothesized that treatment with a bisphosphonate (risedronate) would prevent accelerated bone loss in male patients with epilepsy who underwent long-term AED treatment.

Methods

Study design

Anti-Epileptic Drug and Osteoporosis Prevention Trial (ADOPT) was designed as a 2-year, longitudinal, randomized, double-blind placebo-controlled phase IV trial (Clinicaltrials.gov number registration # NCT00869322) involving male veterans with epilepsy who were taking AEDs for more than 2 years prior to enrollment. This study was conducted at the Veteran Affairs Boston Healthcare System (VABHS) Outpatient Osteoporosis Prevention and Treatment Clinic at the Jamaica Plain, Boston campus, and approved and overseen by and monitored by the safety subcommittee of the local research and development and institutional review board committees.

Study drugs

Patients were assigned randomly to receive either a risedronate 35 mg tablet or matching placebo tablet to take weekly according to manufacturer's package insert. All enrolled patients received calcium supplementation (1,000–1,500 mg) and vitamin D (500–750 IU) daily. For patients who had 25-hydroxy vitamin D levels <20 ng/ml, a loading dose of ergocalciferol 50,000 IU weekly for 12 weeks was prescribed before those patients being enrolled and randomized into the study.

Participants and inclusion and exclusion criteria

Participants were included if they were male; had a diagnosis of epilepsy; were actively taking phenytoin, carbamazepine, phenobarbital, or sodium divalproex for 2 years prior to enrollment, and had normal calcium levels (8.5–10.5 mg/dl) and vitamin D levels (>20 ng/ml). Subjects were excluded if they were female; were treated with glucocorticoids; were organ transplant recipients; had an estimated glomerular filtration rate (EGFR) of <30 ml/min; did not have a diagnosis of epilepsy or seizure disorder; were treated previously with miacalcin, bisphosphonate, teriparatide, or testosterone; had severe swallowing disorder or esophagitis; were osteoporotic at baseline (BMD T-score <−2.5 at AP spine or femoral neck or total proximal femur or forearm), or were unable to follow instructions for taking medications.

Women were excluded from this study owing to the low number of female veterans followed in our epilepsy/seizure clinic.

Based on power calculations from previous studies (Stephen et al., 1999; Andress et al., 2002; Farhat et al., 2002), our own clinical data (A. A. Lazzari, P. M. Dussault, and S. A. Davis, unpublished data), and assuming a drop-out rate of 30–35%, we recruited a total of 80 participants with 40 patients assigned by a randomization without replacement procedure into each treatment arm. The final cohort who completed the study was 27 patients in the risedronate group (67.5%) and 26 patients in the placebo group (65%).

Randomization and blinding

For randomization we utilized QM for Windows Version 1.41 to run a simulation to randomize 80 patients in a 1:1 ratio. Assignment of active versus placebo was placed in an alphabetical placeholder prior to run (Weiss, 1996).

Study design and assessments of efficacy

The primary goal of this study was to determine the effects of treatment with bisphosphonates in preventing further bone loss and possibly improving BMD as measured by the same DXA instrument in patients who were taking long-term AEDs.

We also determined the rate of new verterbral fractures that were identifed in the cohort by serial lateral vertebral assessment (LVA) scans and changes in 25-hydroxy vitamin D levels during the study period.

Patients were monitored for up to 2 years, with clinic visits scheduled at 0, 3, 6, 12, 18, and 24 months. At the initial visit, patients were assessed for existence of modifiable risk factors such as smoking, alcohol consumption, sedentary lifestyle, poor nutrition, and the presence of hypogonadism. Education and counseling were provided to all patients with respect to nutrition, weight-bearing exercises, smoking cessation, and alcohol intake.

BMD evaluations were performed at baseline, and at 12 and 24 months. All studies were performed on the same GE Lunar Prodigy DXA instrument, performed by the same research radiology technician, and analyzed with the same enCore software version 13.6 (GE Medical Systems Lunar, Madison, WI, U.S.A.). Positioning of patients for imaging and analysis was undertaken according to recommended instrument protocol (GE Medical Systems Lunar). BMD assessment was performed at the bilateral proximal femora, AP lumbar spine (AP), lateral spine including morphometric evaluation (LVA), and forearm. All densitometric studies were analyzed by an International Society of Clinical Densitometry certified clinical densitometrist (AAL). Standard dual-energy LVA imaging was performed in the lateral decubitus position as described previously (Binkley et al., 2005). Assessment of vertebral fractures using DXA morphometry is known to provide an excellent specificity, with moderate sensitivity (Duboeuf et al., 2005). All morphometric vertebral fractures were confirmed by two independent nonradiologist clinicians who were blind to study participants; both are certified and trained densitometrists (AAL, PD).

All measurements of each participant BMD were compared to their baseline measurements. The least significant change (LSC) calculated for the GE Lunar densitometer instrument utilized in this study and for the technician (SAD) is 0.012 g/cm2 for the proximal femur and 0.010 g/cm2 for lumbar spine and total body. LSC is calculated using a 95% confidence interval. LSC is the least amount of BMD change measured by the DXA instrument that can be considered statistically significant (Hind et al., 2010). Comparisons of BMDs were performed at lumbar spine, total mean bilateral proximal femora, and total body owing to better precision at those sites.

Statistical analysis

Statistical tests were conducted on demographic data to determine the existence of any subtle selection bias. Quantitative measures (e.g., age) were tested by unpaired t-test; qualitative measures (e.g., ethnicity) were tested by the chi-square test. All data for the primary outcome was analyzed on an intent-to-treat basis. Each patient's BMD at each specific site was compared to his own baseline BMD at year 1 and year 2 after enrollment. Increases or losses of BMD above the LSC for the instrument and our technician were considered significant. Means within groups and differences in BMD were calculated from baseline to year 1 and then baseline to year 2. The differences of BMD between the individuals that completed the study were compared using a paired t-test. A p-value of <0.05 was considered statistically significant. Differences in BMD between risedronate and placebo over time were tested using a mixture model (SAS v. 9.2, PROC MIXED, SAS Institute, Inc., Cary, NC, U.S.A.), where an exchangeable matrix was used to model within-subject correlations over time and where ethnicity was entered into the model as a confounder. Least squared means were calculated, giving ethnicity adjusted mean estimates of bone density for each treatment group at each of the three observation times. To test for the effect of risedronate on fractures compared to placebo, Fisher's exact test was used to test for significant differences between the two groups.

Results

Baseline characteristics of the participants

Eighty patients provided informed consent to participate in the study (Table 1). Forty patients each were randomized to risedronate 35 mg per week or to a matching placebo tablet. All participants received supplementation with calcium and vitamin D. Baseline characteristics including alcohol and smoking history were similar between groups, with the exception of a greater number of vertebral fractures identified at baseline by LVA in the risedronate group compared to placebo (Table 1; 37% vs. 22%). Mean baseline vitamin D level in the risedronate group was 29.10 ng/ml and in the placebo group was 29.31 ng/ml (p = 0.57). The two groups had similar baseline characteristics in terms of prevalence of controlled or refractory seizures or in the number of patients receiving one or more AEDs or on the length of AEDs treatment (p = 0.158; Table 1). In the placebo group, 58% were using phenytoin, 38% were taking carbamazepine, and 8% were taking divalproex, whereas in the risedronate group 78% were taking phenytoin; 7% carbamazepine, 11% divalproex, and 4% phenobarbital. At the end of the study period (2 years after enrollment), 65% of patients in the placebo group and 65% patients in the risedronate group completed the protocol, having at least one scan after baseline.

Table 1. Baseline demographics and characteristics of enrolled participants
 Risedronate (n = 40)Placebo (n = 40)
  1. Values are given as means (±1 SD). Values of BMD are expressed in g/cm2 (±1 SD). Values for 25-hydroxy vitamin D are given as ng/ml. Values for urinary N telopeptide of collagen type I are expressed in nmol bone collagen equivalents/nmol creatinine.

Demographics  
Age63 (±13)58 (±13)
Race or ethnic group  
White (%)9592.5
Black (%)57.5
Body mass index29 (±3.9)29 (±5.4)
Current smoker (%)72.570
Consumption of alcohol drinks >1 per day (%)3040
Weight – mean (kg)85.0 (±11.8)84.7 (±19)
Seizure type and AED use  
Grand mal (%)47.555
Other types (%)52.545
Refractory epilepsy (%)2522.5
Controlled epilepsy (%)7577.5
Antiepileptic drug use (%)  
One AED (%)9090
Two or more AEDs (%)1010
Length of use of AEDs (years)28 (±16)22 (±12)
Baseline BMD  
AP lumbar spine BMD – L1–L41.246 (±0.189)1.244 (±0.162)
Bilateral total proximal femur BMD0.990 (±0.144)1.018 (±0.144)
Lowest femoral neck BMD0.882 (±0.121)0.924 (±0.139)
Total body BMD1.204 (±0.092)1.229 (±0.107)
Vertebral fracture at baseline (%)15 (37.5%)9 (22.5%)
One vertebral fracture7/157/9
Two or more vertebral fractures8/152/9
Baseline – 25-hydroxy vitamin D29.10 (±18.95)29.31 (±15.65)
Baseline – urinary N telopeptide39.24 (±26.30)37.18 (±16.79)

At baseline it was observed that subjects who had compression fractures had lower BMD at the femoral neck and bilateral proximal femora as compared to those who did not have compression fractures (p = 0.02 and 0.04, respectively) (Table 2).

Table 2. Mean baseline BMDs in patients with vertebral compression at baseline compared to patients with no vertebral fractures
Compression fracture (n)Lowest femoral neck BMDBilateral total proximal femora BMDAP lumbar spine BMD
  1. Values are given as means g/cm2 (±1 SD). p-Values were calculated using a Student's t-test – two-sample assuming unequal variance.

No (56)0.925 (±0.129)1.024 (±0.130)1.240 (±0.168)
Yes (24)0.827 (±0.152)0.897 (±0.195)1.309 (±0.103)
p-Value0.020.040.7

Changes in BMD

In the first follow-up DXA study after 1 year of enrollment for the patients who completed the study (Table 3), the mean total bilateral proximal femora BMD decreased nonsignificantly in the placebo group (1.009–0.989 g/cm2; p = 0.44), whereas in the risedronate group, the mean BMD increased significantly by 3% (1.007–1.042 g/cm2; p = 0.02). The mean BMD at the AP lumbar spine as measured at L1 to L4 did not change significantly in the placebo group (1.229–1.231 g/cm2; p = 0.8), whereas the BMD significantly increased in the risedronate group (Table 4) (1.267–1.334 g/cm2; p ≤ 0.005). Total body BMD in the placebo group decreased nonsignificantly (1.216–1.211 g/cm2; p = 0.10), whereas in the risedronate group total body bone density increased significantly (1.204–1.230 g/cm2; p = 0.001).

Table 3. BMD placebo group
SiteBaselineYear 1Year 2
  1. Values are given as means g/cm2 (±1 SD).

  2. a

    p > 0.05 using the paired T-test as compared to baseline values.

  3. b

    p < 0.05 using the paired T-test f as compared to baseline values.

Bilateral proximal femora1.009 (±0.150)0.989 (±0.161)a0.999 (±0.174)a
L1–L4 AP spine1.229 (±0.153)1.231 (±0.143)a1.245 (±0.154)a
Total body1.216 (±0.108)1.211 (±0.101)a1.192 (±0.127)b
Table 4. Mean BMD risedronate group
SiteBaselineYear 1Year 2
  1. Values are given as means g/cm2 (±1 SD).

  2. a

    p > 0.05 using the paired T-test as compared to baseline values.

  3. b

    p < 0.05 using the paired T-test f as compared to baseline values.

Bilateral proximal femora1.007 (±0.118)1.042 (±0.109)b1.025 (±0.111)a
L1–L4 AP spine1.267 (±0.204)1.334 (±0.242)b1.332 (±0.221)b
Total body1.204 (±0.092)1.230 (±0.088)b1.205 (±0.096)a

At the end of the study (Table 3), 2 years after enrollment, there was a 1% nonsignificant increase in the mean BMD at AP lumbar spine in the placebo group (1.229–1.245 g/cm2; p = 0.19), whereas the risedronate group (Table 4) experienced a significant increase of BMD (1.267–1.332 g/cm2; p = 0.004). In comparison, at the AP lumbar spine as measured at L1 to L4, the increase in the risedronate group (1.267 g/cm2 vs. 1.332 g/cm2) was significantly larger than that seen in the placebo group (1.229 g/cm2 vs. 1.245 g/cm2; p = 0.0066). Over the course of the study, the ethnicity-adjusted mean BMD of the femur dropped slightly (1.043 g/cm2 vs. 1.041 g/cm2) in the placebo group, whereas there was a small increase (1.049 g/cm2 vs. 1.053 g/cm2) in the risedronate group. There was no statistically significant difference in these changes between the treatment groups (p = 0.2806). At the end of the study, there was a significant decrease in total body BMD in the placebo group (1.216 g/cm2 vs. 1.192 g/cm2; p = 0.0066), whereas in the risedronate group there was a nonsignificant change (1.204 g/cm2 vs. 1.205 g/cm2; p = 0.454). At the end of the study, there was no statistically significant difference in these measurements between the treatment groups (p = 0.8477). In contrast, at the AP lumbar spine as measured at L1 to L4, the increase in the risedronate group (1.332 g/cm2 vs. 1.267 g/cm2) was confirmed to be significantly larger than that seen in the placebo group (1.245 g/cm2 vs. 1.229 g/cm2; p = 0.0066).

At the end of the study period, of the 26 patients in the placebo group who completed the protocol, 30% had a significant improvement of the BMD at the bilateral proximal femora as determined by increases above LSC for the site and instrument; 26% had no significant changes and 44% had a decrease of BMD (lower than the LSC). At the same site, in the 27 patients in the risedronate group who completed the protocol, 38% had significant gains in BMD, 27% had a significant decrease of BMD, and 35% had no significant changes. In the placebo group, total body BMD decreased in 69% of the patients, increased in 4%, and there was no change in 27%. In the risedronate group 48% had a significant decrease of total body BMD, 19% gained, and 33% had no change. At the lumbar spine L1 to L4, for the entire study period, in the placebo group, 65% gained BMD, whereas 34% had a significant decrease of BMD. In the risedronate group, from L1 to L4, 88% gained BMD, whereas 11% had a significant decrease in BMD.

Vitamin D levels

Mean vitamin D levels at baseline in the placebo group were 30 ng/ml and at the final visit were 38 ng/ml. Mean vitamin D levels in the risedronate group at baseline were 28 ng/ml and at the final visit were 29 ng/ml. These changes in vitamin D levels did not differ significantly (p > 0.05).

Fractures

Using semiquantitative methods as previously described (Binkley et al., 2005), we found throughout the duration of the study five new vertebral fractures in the placebo group and none in the risedronate group (p = 0.0229). In addition, one of the patients in the placebo group was found to have had an upper arm fracture 1 year into the study. This patient was withdrawn from the study as he was found to have significant loss of BMD (T-score <2.5) at his proximal femur. His results were not included in the final analysis.

Conclusions

To our knowledge, this study is the first longitudinal prospective randomized placebo controlled trial of prevention and treatment of bone loss in male veterans with epilepsy who underwent long-term AEDs treatment and had normal or low bone mass.

We conceived and designed the ADOPT study to determine if treatment with an oral bisphosphonate (risedronate) compared to placebo resulted in an improvement in BMD in this population. We observed that more than 65% of the patients in the placebo group receiving supplementation of calcium and vitamin D had a significant improvement of their BMD at any studied site, this supports a beneficial effect of this supplementation on the overall bone health in this patient population. In the risedronate group, we observed that a greater number of subjects sustained a significant increase in BMD at the lumbar spine and at the bilateral proximal femora when compared to the placebo group and to their own baseline BMDs. This significant improvement in bone density at the lumbar spine in the risedronate group may explain why no new compression fractures were observed in this group during the study period. We observed five new vertebral fractures and one nonvertebral fracture only in the placebo group, even though both groups were well matched for their baseline characteristics (Table 1) including the number of AEDs used, and length of therapy and other cofactors such as age, body mass index (BMI), and tobacco and alcohol use. The small sample size in both groups did not allow subanalyses of confounding factors, in particular; however, the statistical analysis was performed as each subject was evaluated as their own control. The overall baseline prevalence of vertebral fractures in the entire ADOPT cohort was 30%, (37.5% were in the treatment group and 22.5% were in the placebo group) (Table 2). This prevalence of vertebral compression fractures is remarkable, since the ADOPT cohort was selected based on T-scores above the threshold for defining the diagnosis of osteoporosis. In this study cohort there was a prevalence of tobacco use (71%) and alcohol use (35%), which may have contributed to the remarkable prevalence of baseline vertebral fractures of the participants, since those habits are considered to be independent risk factors for osteoporosis and osteoporotic fractures (Kanis et al., 2009; Buns, 2013).

The identification of subclinical vertebral fractures is important because a previous fragility vertebral fracture is considered to be a strong predictor for future vertebral and nonvertebral fractures (Binkley et al., 2005; Holmberg et al., 2006). The benefit of identification of subclinical vertebral fractures and other risk factors for osteoporosis in this population would be the initiation of treatment and establishment of preventive measures, which may justify the cost and the possible risk of the use of bisphosphonates as demonstrated in other populations at risk (Little & Eccles, 2010; Majumdar et al., 2013; National Osteoporosis Foundation, 2013). We also observed that BMD at the bilateral total proximal femur at baseline in this cohort was significantly lower in the group of subjects who had baseline compression fractures. We did not, however, when comparing patients at baseline who had a previous compression vertebral fracture to patients with no vertebral deformities, observe significant differences of BMD at the AP spine (Table 2). This may because vertebral compression fractures falsely increase BMD of the spine, as reported previously (Krege et al., 2006). The remarkable frequent finding of baseline vertebral fractures in this population supports the importance of screening epileptic patients for subclinical vertebral fractures either by routine traditional vertebral radiographs or by performing LVAs on a DXA scanner. We believe that by performing an LVA at the same exam as a central DXA scan is performed is time and cost efficient and can disclose asymptomatic and not previously diagnosed compression fractures (Binkley et al., 2005). We also believe that the identification of asymptomatic compression fractures in a patient with epilepsy may improve management and treatment of future bone loss in patients at risk for future fractures. Furthermore, LVA scans expose patients to a considerable lower level of radiation than conventional spine x-rays (2–50 microsievert vs. 600–800 microsievert, respectively, per exposure) (Damilakis et al., 2010; Hind et al., 2010; Robinson et al., 2013).

Our results are in agreement with previously reported risk reduction of vertebral and nonvertebral fractures in other male populations at risk for osteoporotic fractures, supporting the paradigm that the addition of a bisphosphonate in combination with calcium and vitamin D supplementation for 2 years significantly reduced the incidence of new vertebral and possibly nonvertebral fractures (Orwoll et al., 2000; Boonen et al., 2009; Ringe et al., 2009).

There were a number of limitations to our study. Although enrolled patients were provided travel support, there was a large percentage of dropouts owing to travel difficulties, although the dropout rate (40%) was similar for both groups.

Although weight-bearing exercises have been shown to improve BMD, we did not quantify the amount of exercises in this ambulatory population; however, both groups were instructed in all visits to improve their level of physical activity and instructed on the importance of weight-bearing exercises during the initial and all the follow-up visits. Even though the duration of AED use did not differ significantly in this cohort, owing to the small numbers of patients we are unable to determine whether one AED resulted in significant bone density changes when compared to the others. Our cohort was limited to a population of male veterans with epilepsy with a T-score at any of the traditional sites above −2.5; therefore, its application to a wider epileptic population is to be determined.

In attempting to develop a treatment and preventive paradigm for patients with epilepsy regarding their skeletal health, it is important to identify other primary risk factors of bone loss and future fracture risk such as family history of osteoporosis or hip fractures, personal history of fractures and other conditions such as rheumatoid arthritis, and chronic use of glucocorticoids. Well-established fracture risk factors should be evaluated and calculated using the “WHO Fracture Risk Assessment Tool” (FRAX (Kanis et al., 2009; National Osteoporosis Foundation, 2013). Pharmacologic intervention to prevent osteoporotic fractures in patients with epilepsy is indicated and justified with an increased risk of fractures such as a previous fragility fracture, for instance, a vertebral or a hip fracture, or if patients are determined to have a 10-year increased risk of hip fractures above 3% or of a major osteoporotic fracture above 20% according to the FRAX calculator (Kanis et al., 2009). Modifiable risk factors in this cohort included the following: low vitamin D levels, tobacco use, alcohol use, sedentary lifestyle, polypharmacy, and monitoring of patients for central nervous system side effects from AEDs (sedation, ataxia), which may increase fall risk.

The results of this present study suggest that treatment of low bone mass and prevention of fractures in the epileptic population not only can improve bone mass and possibly reverse the progression of bone loss, but also may prevent the development of new vertebral and possibly nonvertebral fractures. Therefore, we believe that in order to address the increased risk for the development of osteoporosis and osteoporotic fractures, clinicians providing care to patients with epilepsy should focus on the following: (1) identification of patients at an increased risk of fracture; (2) identification and management of additional modifiable risk factors for osteoporosis and falls; (3) provision of adequate seizures prevention with as few AEDs or psychotropic medications as possible and with minimal sedative effects; (4) use of nonpharmacologic means of preventing falls and fractures, such as improving physical activity, including but not limited to multifactorial interventions such as muscle strengthening and balance exercises such as tai chi; (5) home safety evaluation should be considered to address environmental factors that may contribute to a fall such as area carpets, lighting, grab rails in bathroom; (6) physical therapy evaluation for gait, balance, and strength; (7) periodic evaluation for BMD loss through DXA and LVA studies to be repeated at 1–2 year intervals; (8) ensure adequate calcium and vitamin D supplementation; and (9) consider initiation of antiosteoporotic drugs (antiresorptive agents such as bisphosphonates, receptor activator of nuclear factor kappa-B ligand (RANK-L) inhibitors, selective estrogen receptor modulators (SERMS), or drugs that improve bone formation such as the parathyroid hormone (PTH) analogues in patients at high risk for future fracture (National Osteoporosis Foundation, 2013). Recently, concerns about the long-term effects of bisphosphonates have been published (McClung et al., 2013). In accordance with current recommendations, we advise that all patients be asked about active dental issues and potential invasive dental procedures prior to initiating therapy. If possible all extensive invasive dental work should be performed before the patient starts taking a bisphosphonate. In addition, therapy should be limited to 5 years if possible, to possibly reduce the rare side effect of atypical femoral fractures and other complications. Future studies of long-term use of bisphosphonates in the epileptic patient population should be encouraged.

Acknowledgments

We thank the following who have contributed to the design and statistical analysis or manuscript revision: Rachael Burns, RA; Roanna Bamford, RNP; Kent Creamer, MD; and Jack Bukowski, MD. We also thank the Boston VA Research Institute for their assistance and support and the VA Boston Healthcare System.

Disclosure

This work was supported by a grant from the Alliance for Better Bone Health, Boston VA Research Institute, and VA Boston Healthcare System. The authors have no conflicts to disclose or affiliation with any commercial entity. We confirm that we have read the Journal's position on issues involved in ethical publication and affirm that this report is consistent with those guidelines.

Ancillary