INTRODUCTION
Fragility fractures of the spine and hip are the most frequently associated with increased morbidity and mortality.[1,2] Although back pain is the primary clinical manifestation of vertebral fractures (VF), the clinical presentation is highly variable from disabling pain that seriously disrupts quality of life to asymptomatic fractures that go undiagnosed, resulting in delays in treatment and secondary prevention of new fragility fractures.[3]
There are numerous skeletal and non-skeletal factors that determine the development of fragility VF (FVF) and its severity.[4] In a previous study, we identified that older age, low body mass index (BMI), low bone mineral density (BMD), previous fragility fractures, and smoking habit are fracture risk factors associated with more severe FVF.[5]
Fragility fractures increase the risk of a subsequent fracture for at least 10 years, with the first two years being the period of highest or imminent risk.[6-8] In 42% of individuals with a sentinel VF, the new fracture occurs within the first year with a relative risk of 6.6 times higher than that of an age-matched cohort.[9] Older age, low BMD, and the severity and number of baseline VF are independent risk factors for future FVF.[8,10] Low back pain or its exacerbation and height loss are common clinical symptoms and signs that alert of the probability of a new FVF in postmenopausal women with osteoporosis and VF.[11]
The objective of this study is to describe the radiological evolution of FVF in patients followed in the clinical setting of a fracture liaison service (FLS). Additionally, the study aims to analyze the factors associated with this evolution, including the clinical and radiological characteristics of prevalent FVF, as well as the presence of fragility fracture risk factors.
METHODS
This study was approved by the Ethics Committee for Research at Hospital Universitari de Bellvitge (Approval no. UFO-2015-01). All patients signed a written informed consent upon enrollment.
1. Patient identification
All patients diagnosed with FVF that attended the FLS at Hospital Universitari de Bellvitge from 2015 and 2018 were included. Patients with FVF are referred to FLS from different clinical settings, including the emergency department, primary health care centers, or other medical specialties. Additionally, to identify the presence of FVF in patients from the emergency department, a systematic weekly revision of the discharge reports of patients who received care in the trauma section is made.
The initial in-person visit at the FLS includes a comprehensive assessment of fracture risk factors. The fracture mechanism is systematically reevaluated and patients whose fractures are attributed to high-energy trauma are excluded from the FLS pathway and, consequently, from this study. Clinical follow-up visits are very frequent after the diagnosis of a VF, occurring every 2 to 4 weeks depending on pain levels, to adjust analgesic treatment if necessary, and become more spaced out over time. Radiological follow-up imaging is ordered by the FLS, primary care physicians, or emergency physicians based on clinical progression. Generally, follow-up imaging due to worsening or lack of pain improvement typically involves a simple radiograph of the spine, while in cases with neurological complications, a computed tomography (CT) or magnetic resonance imaging (MRI) is performed.
All data were entered prospectively in a specifically designed database.
2. Study design
This was an observational longitudinal study. Included participants had FVF confirmed by imaging and patients with poor-quality images or incomplete data in the FLS database were excluded. A VF was defined as any reduction in anterior, middle, or posterior vertebral height greater than 20% compared to the same vertebra or the immediately adjacent vertebra (above or below) on spinal imaging. The vast majority of fractures were identified using conventional spine X-rays, although some were through CT or MRI. While some VF were detected incidentally (morphometric VF), most cases were clinical VF. We identified the baseline radiological characteristics of the VF upon admission to the FLS. All images analyzed for this study were stored in the picture archiving communication system of the “x” for later evaluation.
The primary outcome was the presence or absence of radiological worsening (RW), classified as follows: progression of the prevalent VF (PVF), new VF (NVF), or both (PNVF). The progression of VF was defined as an increase in the grade of any of the baseline FVF as described by Delmas et al. [7] (explained below), in consecutive images taken after admission of the patients to the FLS. A NVF was defined as the occurrence of a VF in a vertebra that was normal in a previous imaging study. Any case of RW—whether PVF or NVF—that was clearly related to a traumatic event or a fall from a height greater than one’s own, or walking at normal speed was not considered RW.
We use five categories of secondary variables: (1) demographic data: sex, age and BMI; (2) fracture risk factors: age ≥65 years, BMI ≤20 kg/m2, previous vertebral and non-vertebral fragility fracture (as evidenced in available imaging and clinical records), parental hip fracture, current smoking, rheumatoid arthritis, alcohol intake of 3 or more units/day, use of oral glucocorticoids for more than 3 months, secondary osteoporosis (disorders strongly associated with osteoporosis, including type I diabetes, osteogenesis imperfecta in adults, untreated long-standing hyperthyroidism, hypogonadism or premature menopause, chronic malnutrition, malabsorption, and/or chronic liver disease), number of falls in the previous year, daily intake of dietary calcium less than 500 mg, serum calcidiol level, lumbar spine and proximal femur BMD, World Health Organization (WHO) diagnostic category (normal, osteopenia, or osteoporosis), and trabecular bone score; (3) characteristics of pain associated to the baseline VF: history of chronic back pain, acute or gradual onset, localized or irradiated distribution, analgesia requirements, according to the WHO analgesia ladder (first, second, and third steps), as an indirect assessment of pain intensity; (4) pharmacological treatment of osteoporosis: calcium and vitamin D supplementation, antiresorptive drugs (bisphosphonates or denosumab) and anabolic agents (teriparatide); (5) follow-up time (months).
Acute onset pain was defined as pain that appeared suddenly and was clearly attributable to a specific moment or event, while gradual onset pain referred to pain that developed progressively over time without a clear triggering event.
Patients received osteoporosis treatment for at least five years. Those treated with teriparatide for two years were subsequently prescribed alendronate or denosumab for at least three additional years, based on baseline and follow-up BMD. Treatment adherence was monitored through telephone follow-ups at 3 and 12 months. Calcium and vitamin D supplementation was adjusted individually according to dietary calcium intake and serum calcidiol levels.
3. Image assessment
Available vertebrae were assessed at baseline and in every follow-up radiological examination. Follow-up imaging was requested at the physician’s discretion and did not follow a standardized protocol. For every image (X-ray, CT, or MR), the following data were recorded:
- date
- location(s) of the FVF
- number of FVF (single or multiple FVF)
- kyphotic angle, measured in grades, between the upper-end plate of the upper vertebra and the lower-end plate of the lower vertebra of the dorsal spine
- spine deformity index (SDI), calculated by adding the deformity grade of each fracture
- grade of FVF according to the Genant’s semiquantitative method.[12] Briefly, we measured the anterior, middle, and posterior heights of the vertebral body at each fractured vertebra and of the adjacent vertebrae. Height loss (%) was calculated as a percentage of the mean of the height of the superior and inferior vertebrae. Severity was assigned as grade 1 (20%-25% height reduction), 2 (26%-40% height reduction), or 3 (>40% height reduction).
- For each patient and scan, we also recorded the highest VF severity grade observed in the available image.
4. Statistical analysis
Data was expressed as means±standard deviation. Absolute and relative frequencies were used for dichotomous variables. The Kolmogorov-Smirnov test was performed to evaluate data distribution. Differences between groups of patients were examined using the χ2 test or Fishers exact test for categorical variables, and analysis of variance or Wilcoxon’s signed rank test for quantitative variables.
The factors associated with RW were analyzed using the multivariate Cox proportional hazards regression model. Kaplan-Meier curves were drawn to express the cumulative incidence of RW over time.
A P-value of less than 0.05 was considered statistically significant. Data were collated in an Access 2010 database and analyzed using SPSS software version 18 (SPSS Inc., Chicago, IL, USA).
RESULTS
We identified 452 patients and excluded 30 (6.6%). A total of 422 patients presented a FVF and were evaluated at the FLS. Of these patients, 223 completed at least one follow-up evaluation (53%) (Fig. 1). The other 199 patients visited the FLS in the same period but had no radiological follow-up. The mean radiological follow-up period was 11.74± 11.7 months (median, 7 months; range, 1-48 months), with a total observation period of 217 patient-years.
The demographic data, fracture risk factors, baseline clinical and radiological characteristics of index FVF, and pharmacological treatment of osteoporosis in patients with or without radiological follow-up are presented in Table 1. The patients with follow-up had a better fracture risk factor profile (they were younger, had higher BMD, and lower prevalence of vitamin D deficiency), but they required more potent analgesia at the moment of admission to FLS than those with no follow-up.
Table 2 shows the same data for patients with and without RW. One hundred and sixty-one patients had RW, representing 69% of the patients with follow-up. We performed a sensitivity analysis considering two extreme scenarios: (1) assuming that none of the patients without follow-up had RW (best-case scenario), the overall RW incidence would be 36.5%; and (2) assuming that all patients without follow-up had RW (worst-case scenario), the overall RW incidence would rise to 83.6%. The median time to identified RW was 8 months (95% confidence interval [CI], 5.7-10.3) (Fig. 2). A hundred and forty (87%) of these patients worsened before 2 years of follow-up.
In the univariate analysis (Table 2), the variables collected in the evaluation of baseline FVF that associated with a RW were: (1) fracture risk factors: BMI ≤20 kg/m2 and calcidiol ≤25 nmol/L; (2) patients’ clinical characteristics: gradual low back pain at onset; and (3) baseline FVF radiological characteristics: lumbar location, and higher worst grade, SDI and kyphosis angle.
In the multivariate analysis, we included all variables that were statistically significant or approached significance. RW was significantly associated with multiple FVF at baseline (odds ratio [OR], 1.7; 95% CI, 1.05-2.62); baseline FVF grade 1 (OR, 3.4; 95% CI, 1.56-7.21); and grade 2 (OR, 1.78; 95% CI, 1.07-2.95).
Of the 161 patients with RW, 61% had PVF, 27% had NVF, and 12% had PNVF.
There were 73 NVF; 36% in the thoracic and 64% in the lumbar spine. Sixteen NVF were in grade 2 and 57 grade 3. All NVF grade 2 occurred in patients whose worst grade baseline FVF was grade 1. Of the NVF grade 3, 15 occurred in patients whose worst grade baseline FVF was grade 1 and 42 in grade 2 (P=0.000). In 63 (43%) patients with NVF, their location was adjacent to baseline FVF.
There were 131 FVF with PVF; 39% in the thoracic and 61% in the lumbar spine. Of the grade 1 PVF, 27 progressed to grade 2 and 22 to grade 3. Eighty-three grade 2 FVF progressed to grade 3.
Factors significantly associated with PVF were baseline FVF grade 1 (OR, 3.22; 95% CI, 1.47-7.02) and grade 2 (OR, 1.97; 95% CI, 1.05-3.68) and for a NVF was grade 3 (OR, 3.19; 95% CI, 1.39-7.33) (Table 3).
DISCUSSION
In this study, we analyzed the incidence of RW in a series of patients with recent clinical FVF. We studied the relationship between RW and patients’ risk factor profile, including the clinical and radiological characteristics of the prevalent FVF.
We observed an incidence rate of RW of 73.8/patient-year. Nearly three-quarters of patients with radiological follow-up exhibited RW, primarily due to PVF. In the best case-scenario, assuming that none of the patients without radiological follow-up had RW, at least 36.5% of patients admitted to the FLS have RW. In the worst-case scenario, assuming that all patients without follow-up had RW, the overall RW incidence would rise to 83.6%. While this highlights the potential impact of missing radiological follow-up, we consider the worst-case assumption unlikely, given that follow-up was mainly performed in patients with greater pain severity. A higher number of FVF and a higher grade of severity at baseline are associated to RW.
The increase in risk of NVF in patients with prevalent FVF is well-known. In a prospective population-based cohort study of 3,469 women and men (mean age of 68.8 and 66.3 years, respectively), 253 of them with prevalent FVF, the incidence of NVF was 19%, 26.2% in women and 8.7% in men, after a follow-up of 6.3 years.[13] Similar to our study, a higher age was correlated with a higher incidence of NVF.
In a randomized, double-blind, and three-year trial of 7,705 postmenopausal women with osteoporosis, the mean age of whom was 66 years, the authors analyzed the relationship between prevalent FVF grade and the risk of new vertebral and non-VF. In the placebo group (N=2,565), the incidence of NVF was 20% in patients with prevalent FVF at baseline (N=938). The incidence of NVF was 38.1% higher in women with severe VF (defined as grade 3 VF) than in those with mild or moderate prevalent FVF.[7]
A prospective population-based study in postmenopausal women, with an average age of 74 years, followed for a period of four years, evaluated the impact of mild VF on the risk of new vertebral and non-VF. In patients with prevalent FVF, the incidence of NVF was 49%. The incidence was higher (39.8%) in patients with at least one grade ≥2 FVF at baseline compared to patients with a grade 1 FVF (27.4%).[14]
Several studies have reported the incidence of NVF ranging from 24.5% to 52% in elderly patients over a follow-up period of 3 to 6 years.[15-17] In a study of 1,533 Chinese women aged 75 years with a 4-year follow-up, the incidence of RW was 21% among patients with FVF (N=193) at baseline, either due to PVF or NVF.[18] In this series, in all patients who had RW (N=41) it was by PVF and 33 of them had also NVF. Patients with prevalent grade 3 FVF (N=24) had more RW than the ones with grade 1 or 2 FVF.
All the studies described have a prospective design and systematic radiological follow-up. Different protocols are used to identify an incidental FVF: serial spine films, VF assessment (VFA), self-reported incident clinical fractures, or scheduled interviews. Additionally, these studies differ in the imaging methods used to identify FVF, such as X-ray images [13,14,16,18] or VFA,[15,17] as well as the method to score the FVF, the Genant method in most studies [7,16-18] or McCloskey-Kanis [13,15] in others.
In terms of identifying the main variable in the RW, only one of studies described assessed both,[18] PVF and the presence of NVF. The majority of the data reported are limited to the incidence of NVF.
The most significant distinction between our study and the aforementioned studies is the timing of patient enrollment. This difference may be an important factor contributing to the elevated incidence of RW observed in our study, ranging from 36.5% to 72%. Our patients were enrolled in the study immediately following a prevalent clinical FVF, which is defined as the period of imminent risk of fracture. It is well established that the initial two-year period following a sentinel FVF is characterized by the highest risk.[6-8] Our findings are in that line, as the majority of our patients had RW before the two-year follow-up. However, we acknowledge that some less symptomatic patients without follow-up imaging might have had undiagnosed RW, which could have influenced the reported RW incidence and its associated factors.
In all the previous studies described the inclusion criteria for the patients were not related to the identification of an FVF. Additionally, the date when the FVF occurred was not consistently known, which may have resulted in a percentage of the FVF being classified as either morphometrical, not clinical, or even of traumatic origin. A comparison of our findings with the studies described reveals a degree of concordance with regard to the risk factors for RW. These include advanced age,[13] the presence of pre-existing FVF,[13,17] the number of pre-existing FVF,[14,19] and the grade of severity of pre-existing FVF.[7]
In relation to the grade of prevalent FVF, a meta-analysis of four phase-three trials assessed their prognostic significance in women.[20] The authors identified FVF using the Genant method and found that mild FVF at baseline increased the risk of developing NVF by approximately twofold to threefold. Moderate or severe FVF at baseline, increased the risk fourfold. Consistent with this, we found a significant association between the degree of severity of prevalent FVF and RW, either by PVF or NVF.
Regarding the distribution of RW within the spine, PVF involved primarily L1 and T12 and NVF, L2, and L4. These findings are consistent with other studies that have demonstrated the thoracolumbar junction to be the most affected area, likely because it is the site of the highest vertebral compression load.[13,17,21,22]
Regarding the clinical and radiological characteristics of FVF at baseline, although patients with RW due to NVF had the highest SDI and kyphosis angle at admission to the FLS, multivariate analysis showed that the variables significantly associated with RW were the presence of multiple FVF at baseline and a more severe grade of the prevalent FVF. These findings are not contradictory, as both variables are representative of a severe form of presentation of FVF.
The results of the multivariate analysis indicated that there was no statistically significant association between fragility fracture risk factors and RW. The variables that correlated with RW in the univariate analysis, namely low BMI, low calcidiol, gradual low back pain at onset, and lumbar location of the baseline FVF, may require further investigation in new prospective studies, as some of them could be considered as expected associations in studies of this nature. A low BMI is a significant risk factor for fragility fractures, and the gradual onset of pain may be indicative of an evolving fracture cascade.
When we compared patients’ characteristics when admitted to the FLS, those who had at least one radiological evaluation on follow-up were younger, had higher calcidiol level and higher femoral neck and total hip BMD, with respect to those with no follow-up. It can be hypothesized that patients with follow-up had a lower risk for RW. We also found significant differences in the level of analgesia administered to the patients, higher in the follow-up group. These findings are consistent with those of the European Prospective Osteoporosis Study, in which patients with a second radiological image were younger and had fewer previous fragility fractures and fewer diagnoses of osteoporosis.[23] It can be concluded that when radiological follow-up is not protocolized, the decision to perform a radiological evaluation on follow-up is not based on the risk of RW but on the poor evolution of pain.
We found no association between osteoporosis treatment or calcium/vitamin D supplementation and RW. However, this study was not designed to compare the anti-fracture efficacy of different osteoporosis drugs, as the sample size is insufficient and patients were not randomized to specific therapies. While no differences were observed between drugs, this does not imply that such differences do not exist.
The limitations of this study include the absence of a pre-defined protocol for patient follow-up and the absence of predetermined intervals for X-ray examinations. The decision to perform imaging tests was based on the clinical evolution of the patients, as observed during each medical check-up, whether outpatient, emergency department care, or regular telephone follow-up. Despite that fact, there are no significant differences between patients with and without follow-up in our series, with respect to the risk factors of fractures. It is important to acknowledge the inherent limitations of plain radiography in the early diagnosis of VF. In clinical practice, some patients may present with symptoms strongly suggestive of a VF but have normal spinal radiographs at symptom onset and even in the days that follow. In such cases, repeat imaging performed later may reveal a fracture that retrospectively explains the initial symptoms. While this phenomenon is uncommon, it cannot be entirely excluded in our cohort and reflects a well-described limitation of standard radiological evaluation in real-world clinical settings. As another limitation, a single reviewer assessed all images for all participants
Our study design has several strengths. Fragility fracture risk factors were prospectively collected, based on a predefined protocol, which was implemented for all the patients admitted to the FLS. All this information was incorporated into a database that was specifically designed for patients with fragility fractures. The series is large, homogeneous, and can be considered representative of patients with FVF in our community. The length of follow-up was 4 years and therefore, an appropriate time can include recent FVF and those produced during the period of imminent risk of fracture.
While our study may not provide an accurate incidence of RW due to the lack of follow-up imaging for all patients, it highlights a critical point: in patients with VF who undergo clinical driven repeat imaging-often due to increased pain, emergency visits, or clinician judgment-the likelihood of identifying RW compared to previous radiological assessments is very high.
Not all clinicians implement systematic radiological follow-up at regular intervals for all patients with FVF. In many cases, identifying RW does not necessarily imply a change in treatment for secondary fracture prevention. For FLS or bone metabolism units that perform clinical follow-up without routine radiological assessments, the information provided in our study may offer valuable information for real-world clinical practice.
PDF Links
PubReader
ePub Link
XML Download
Full text via DOI
Download Citation
Print
