Prevalent vertebral fractures predict subsequent radiographic vertebral fractures in postmenopausal Korean women receiving antiresorptive agent
- First Online:
- Cite this article as:
- Kim, S.H., Choi, H.S., Rhee, Y. et al. Osteoporos Int (2011) 22: 781. doi:10.1007/s00198-010-1298-y
- 116 Views
The relationship between prevalent vertebral fractures and new vertebral fractures in Korean women has not been previously studied. We found that prevalent vertebral fracture is a strong risk factor for subsequent radiographic vertebral fracture, independent of age and spine bone mineral density (BMD) in postmenopausal Korean women receiving antiresorptive agents.
The presence of prevalent vertebral fractures increases the risk of subsequent vertebral fractures. This observational study examined whether prevalent vertebral fractures influenced future incident vertebral fracture in postmenopausal women receiving antiresorptive treatment.
Using the medical records, we retrospectively assessed 1,200 postmenopausal women who had visited the osteoporosis clinic. Among them, we selected patients whose BMD was measured at least once a year for 3 years. We excluded patients whose lateral spine X-ray film at baseline or follow-up period was missing, which is the key assessment. In the end, we analyzed 666 postmenopausal women. The mean duration of follow-up period was 3.2 years.
The majority of patients (83%) received bisphosphonates or raloxifene during the mean duration of 3.2 years. Fifty-five of the 666 women had new incident radiographic vertebral fractures during the follow-up period. Prevalent vertebral fractures were associated with a five-fold increased risk (odd ratio 5.7, 95% CI 3.09–10.74) of incident vertebral fractures after adjusting for age. These associations remained significant after the adjustment for multiple confounding factors (odd ratio 4.4, 95% CI 1.97–9.75). The risk of incident vertebral fracture increased with the number of prevalent vertebral fractures at baseline: 3.4 (95% CI 1.54–7.70) for those with a single fracture and 7.4 (95% CI 3.10–17.54) for those with two or more after adjustment for age and spine BMD.
We confirmed that prevalent vertebral fracture is a strong risk factor for new vertebral fractures, independent of age and spine BMD in postmenopausal Korean women receiving antiresorptive agents.
KeywordsAlendronateBone densityOsteoporosisRaloxifeneVertebral fracture
Vertebral fracture is the most common osteoporotic fracture, and it is a well-recognized and serious consequence of postmenopausal osteoporosis . All vertebral fractures, whether symptomatic or radiographically identified, are associated with increased mortality and morbidity . It is well established that a prior osteoporotic fracture increases the risk of future fractures [3–7]. Prevalent vertebral fractures have been shown to be strong predictors of incident vertebral fractures, independent of bone mineral density (BMD) [7–10]. Moreover, prevalent vertebral deformities increase the risk of subsequent hip fractures [8, 11]. Many published guidelines recommend that pharmacologic therapy, in order to reduce fracture risk, be considered for postmenopausal women with radiographic vertebral fracture, even in the absence of osteoporosis by BMD criteria [12, 13].
Fracture risk is strongly affected by ethnicity ; therefore, fracture risk factors should be evaluated according to each ethnic and population group. Moreover, the association of prevalent vertebral fractures and new vertebral fractures in patients receiving antiresorptive might be different when compared with those without osteoporosis medication in epidemiologic studies. Previously, the relationship between prevalent vertebral fractures and new vertebral fractures, especially in Korean women taking osteoporosis medication, has not been studied. We therefore retrospectively observed the effects of prevalent vertebral fractures on the risk of new vertebral fractures in postmenopausal Korean women receiving antiresorptive agents.
Materials and methods
Using the medical records of Severance Hospital, we retrospectively assessed postmenopausal women who had visited the osteoporosis clinic for the first time between January 2000 and December 2004. We further reviewed medical records of 1,200 patients to select patients whose BMD was measured at least three times annually until December 2006. We excluded patients whose lateral spine X-ray film taken upon onset of osteoporosis or upon the final year of the follow-up period was missing. Patients were also excluded if (1) they were taking any osteoporosis medication (bisphosphonate, raloxifene, or calcitonin) at baseline or had a history of osteoporosis medication, (2) they had other diseases such as rheumatoid arthritis, malignancy, renal dysfunction (serum creatinine >1.4 mg/dL), and Paget's disease, or (3) they were taking glucocorticoid. We also excluded patients who stopped osteoporosis medication during the follow-up period. In the end, we selected and analyzed 666 postmenopausal women. The mean duration of follow-up period was 3.2 years.
BMD of the lumbar spine and hip was assessed by a dual energy X-ray absorptiometry (DXA, Hologic QDR4500W, Bedford, MA) at baseline and annually thereafter. The coefficients of variation for the lumbar spine and hip were both less than 1%.
Confirmation of prevalent radiographic vertebral fractures and incident radiographic vertebral fractures
The vertebral fracture was assessed using the visual semiquantitative vertebral deformity (SQ) scoring system of Genant et al.  An SQ score of 0 was assigned to normal vertebra; mild grade 1 deformity was defined as a 20–25% reduction in anterior, middle, or posterior vertebral height; moderate grade 2 deformity was defined as a 25–40% reduction in any height; and severe grade 3 deformity was defined as a 40% reduction in any height. A prevalent vertebral fracture was defined as more than grade 1 deformity. A new incident fracture was defined as a vertebral fracture occurring in a vertebra that was not fractured at baseline. One radiologist read 50 randomly selected radiographs on two separate occasions to evaluate the intraobserver reproducibility of the SQ. The intraobserver agreement based on the dichotomous fracture/non-fracture decision was 94% (κ = 0.86; P value < 0.01).
Statistical analysis was carried out using SPSS for Windows (v 11.0, SPSS Inc., Chicago, IL, USA). Baseline characteristics of women with prevalent radiographic vertebral fractures were compared with those of women without vertebral fracture using t tests for continuous variables and χ2 statistics for categorical variables. Logistic regression analyses were performed to determine the association between prevalent radiographic vertebral fractures and incident vertebral fractures. The first model was adjusted for age only, and the second model for age and baseline spine BMD. Finally, we included covariates which were significant in univariate analyses (P < 0.1). Covariates included in the final regression model were age, baseline spine BMD, body mass index (BMI), and use of antiresorptive agents (bisphosphonate or raloxifene) during the follow-up period.
To examine whether the impact of prevalent vertebral fractures on the risk of incident radiographic vertebral fractures varies according to the level of BMD, we divided the patients into two groups defined by the level of spine BMD (T-score between −1.0 and −2.5, and T-score ≤ −2.5). For each group, logistic regression models were used to evaluate the risk of incident vertebral fractures attributable to baseline prevalent vertebral fractures.
Characteristics of subjects with/without vertebral fracture at baseline
Baseline characteristics of study subjects
Without prevalent vertebral fractures (n = 567)
With prevalent vertebral fractures (n = 99)
62.1 ± 7.0
68.1 ± 7.6
Years since menopause
11.0 ± 6.8
17.5 ± 9.1
155.0 ± 4.8
152.1 ± 5.5
Body mass index (kg/m2)
23.2 ± 2.6
23.6 ± 3.3
Serum calcium (mg/dL)
9.6 ± 0.4
9.6 ± 0.3
Inorganic phosphate (mg/dL)
3.8 ± 0.4
3.9 ± 0.4
34.7 ± 15.8
34.8 ± 18.8
16.4 ± 6.6
17.7 ± 8.0
Past or current
−2.0 ± 0.9
−2.8 ± 1.1
−1.7 ± 0.7
−2.2 ± 0.9
−1.1 ± 0.8
−1.8 ± 0.9
Prevalent vertebral fractures and the risk of incident radiographic vertebral fractures
During the follow-up period of 3.2 years, 370 (55.5%) received bisphosphonates (daily or weekly), and 142 (21%) received raloxifene. Forty-four subjects (6.6%) received bisphosphonate followed by raloxifene or raloxifene followed by bisphosphonate. Ninety-seven percent of patients with prevalent vertebral fractures and 80% of patients without fracture were treated with antiresorptive agents.
Relative risk (OR) of incident vertebral fracture in subjects with or without prevalent vertebral fracture
Age-adjusted OR (95% CI)
Age- and BMD-adjusted OR (95% CI)
Multivariate-adjusted ORa (95% CI)
I. In all patients (n = 666)
Any prevalent vertebral fracture
One vertebral fracture
Two or more vertebral fractures
II. In patients with OP medication (n = 556)
Any prevalent vertebral fracture
One vertebral fracture
Two or more vertebral fractures
Additionally, we examined the effect of prevalent vertebral fractures on incident vertebral fractures in 556 patients who were taking osteoporosis medication during the follow-up period. The risk of incident vertebral fractures appeared to be attenuated compared to that of all patients. The ratio increased with the number of prevalent vertebral fractures at baseline, which was 2.6 (95% CI 1.07–6.17) for those with a single fracture and 6.8 (95% CI 2.80–16.35) for those with two or more after the adjustment for age and spine BMD (Table 2).
Association between prevalent vertebral fractures and the risk of incident vertebral fractures according to the BMD level
Logistic regression analyses with new vertebral fractures as a dependent variable
Prevalent vertebral fracture
Use of antiresorptive agents
Relative risk of age and BMD adjusted incident radiographic vertebral fracture in those with prevalent vertebral fractures according to BMD level
Age and BMD-adjusted incident radiographic vertebral fracture [OR (95% CI)]
Prevalence of new incident radiographic vertebral fractures (number of new vertebral fractures)
Without prevalent vertebral fractures
With prevalent vertebral fractures
Osteopenia (n = 383) (T-score −1.0 to −2.5)
Osteoporosis (n = 283) (T-score ≤−2.5)
In this observational study of 666 postmenopausal Korean women, we found that subjects with prevalent vertebral fractures were at a much higher risk of new incident vertebral fractures compared to those without them. Women with prevalent vertebral fractures have approximately four to five times higher risk of further vertebral fractures after the adjustment for age and spine BMD. Furthermore, the risk of new incident vertebral fractures increased with the number of previous vertebral fractures.
Previous studies have found that the prevalent vertebral fracture is a risk factor for an incident fracture. Black and colleagues  showed a five-fold increased risk of new vertebral fractures in elderly women with prevalent vertebral fractures after a mean of 3.7 years. Linsay and colleagues reported that almost 20% of postmenopausal women experience another fracture within 1 year of an incident vertebral fracture in a large clinical trial data analysis . Moreover, a meta-analysis showed that women with preexisting vertebral fractures had an approximately a four times higher risk of subsequent vertebral fractures than those without prior fractures .
The relationship between prevalent vertebral fractures and new vertebral fractures in Korean women has not been previously studied. In the present study, we retrospectively observed the effect of prevalent vertebral fracture on the risk of new vertebral fractures in postmenopausal Korean women receiving antiresorptive agents. Many randomized clinical trials have shown that these antiresorptive agents reduced the incidence of vertebral fractures by up to 40–50% when compared with placebo in postmenopausal osteoporotic women [17–19]. In our study, 83% of patients took oral bisphosphonate or raloxifene during the mean of 3.2 years, and more patients with prevalent vertebral fractures received either bisphosphonate or raloxifene compared to those without the fractures (97% vs. 80%). Nonetheless, the risk of incident vertebral fractures was 4.4 times higher in patients with prevalent vertebral fractures than those without fractures, even after the adjustment for multiple confounding factors. When the effect of prevalent vertebral fractures in patients who were taking antiresorptive agents was assessed exclusively, the risk of new vertebral fractures seemed attenuated but remained significant (2.6 for those with a single fracture and 6.8 for those with two or more).
In the Fracture Intervention Trial (FIT) among subjects with vertebral fracture at baseline, 15% of those in the placebo group and 8% of those in the alendronate group had experienced a new morphometric vertebral fracture . In subjects without an initial vertebral fracture, new vertebral fractures had occurred in 4% in the placebo group and 2% in the alendronate group after 4.2 years . Although there were some differences in age and baseline characteristics, subjects with baseline vertebral fractures seemed to have a four-fold increased risk of new vertebral fractures compared to those without baseline fractures in both group (15% vs. 4% in the placebo group, 8% vs. 2% in the alendronate group, respectively). In the Multiple Outcomes Raloxifene Evaluation (MORE) study, the risk of incident vertebral fracture was higher in patients with baseline vertebral fractures compared with those without fractures (21.2% vs. 4.5% in the placebo group, 14.7% vs. 2.3% in the raloxifene group) . These results could be consistent with our study where women with prevalent vertebral fractures have a four times higher risk of further vertebral fractures after the adjustment for multiple factors. Moreover, these results could be consistent with a concept that the presence of previous vertebral fractures reflects the failure of bone quality regardless of BMD. Therefore, in addition to low BMD, a failure of bone quality may affect the occurrence of incidence vertebral fracture.
Recently, Lee et al.  identified clinical risk factors for osteoporotic fracture in Korean men and women. They reported that age, BMI, and history of previous fractures were independent risk factors of osteoporotic fracture in both sexes. Also, subjects with prior fractures had a higher risk of a clinical osteoporotic fracture (RR = 1.80 in women). The risk of subsequent fractures is not as great as that identified in our study. The discrepancy might be attributed to the fact that they included fracture events at several sites (hip, wrist, humerus, rib and pelvis, and vertebra). Vertebral fracture is commonly related to low BMD, while other sites of fractures are associated more with bone-related risk factors such as previous fracture history or parent's fracture history or fall-related risk factors than low BMD. Moreover, fractures at any particular site are most predictive of future fractures at that site. Other possible reasons for the discrepancy may be related to differences in the populations studied. In our study, the subjects were selected from the osteoporosis clinic at Severance Hospital which is a tertiary medical center. Subjects in our study were older (63.1 vs. 52.2 years) and more likely to have coexisting morbidity that might have increase the risk of fractures.
Low BMD is a well-known independent risk factor for new vertebral fractures. As expected, our study also showed that spine BMD as along with prevalent vertebral fractures was an independent risk factor for new vertebral fractures. In our study, prevalent vertebral fracture was associated with a greater risk of new vertebral fractures in patients with osteopenia compared to those with osteoporosis (6.5-fold vs. 3.7-fold). It was very hard for us to explain these results. Among patients without prevalent vertebral fracture, 3.7% of patients with osteopenia and 6.9% of osteoporosis patients developed new vertebral fractures. This result was consistent with the concept that low BMD was an important factor for new vertebral fracture. Meanwhile, among patients with prevalent vertebral fracture, the difference of new vertebral fracture incidence between the both groups was smaller than those without prevalent fracture (22.5% vs. 29.4%). These results suggest that prevalent vertebral fracture was another risk factor for new vertebral fractures regardless of BMD.
Our study has several limitations. First, as with all observational studies, systematic errors (e.g., selection bias) have been included in the observed results . Observational errors may arise from differences in antiresorptive agents used during the follow-up period in our study. Second, we did not examine the effect of prevalent fracture severity on the risk of new vertebral fractures. Previous studies have shown that prevalent fracture severity, in addition to the number of fracture, was a valuable predictor of new vertebral fractures . Third, we have no data on patients' compliance with osteoporosis treatment. However, we selected postmenopausal women whose BMD was measured at least three times. Thus, we consider the women in our study had a relatively good compliance.
In conclusion, within this observational study of women receiving antiresorptive agents, subjects with prevalent vertebral fractures had a much higher rate of new incident vertebral fractures than those without. We have reconfirmed that prevalent vertebral fractures increase the risk of further vertebral fractures in postmenopausal Korean women.
Conflicts of interest