Abstract
Psoriatic arthritis (PsA) can have peculiar effects on bone, including mechanisms of bone loss such as erosions, but also of bone formation, such as ankylosis or periostitis. The aim of the present study was to describe the prevalence of fractures in patients with PsA as compared to healthy controls and to investigate determinants of fractures among cases. For both cases and controls, radiographs were read to identify vertebral fractures (VF), and the presence of femoral neck or other nonvertebral fractures was obtained from patients’ medical history. The prevalence of fragility fractures on radiographic readings did not differ between cases and controls. The number of subjects showing a VF was 33 (36%) among PsA patients and 36 (36%) among controls, with a prevalence of severe VF of 8% among cases and 4% among controls. Controlling for covariates in a logistic model, the only variables showing a significant correlation with the presence of nonvertebral fractures (NVF) were disease duration (p = 0.02), age (p = 0.03), and bone mineral density at femoral neck (inverse correlation, p = 0.04). Fractures should be carefully considered when evaluating the global picture of the patient with PsA for their contribution to the “fragility” profile.
Psoriatic arthritis (PsA) can have peculiar effects on bone, including mechanisms of bone loss such as erosions, but also of bone formation, such as ankylosis or periostitis. This is probably one of the reasons why the prevalence of osteoporosis (OP) in PsA is still debated1,2,3,4,5,6,7. However, an issue of growing importance today is a major complication of OP — fragility fractures. Fragility fractures are defined as fractures that occur spontaneously or on low trauma impact and are therefore considered the consequence of “reduced bone strength,” which is basically the current definition of OP8.
Fragility fractures are a very relevant clinical event, not only because of the severe and frequent disability that results, but also because fractures cause an increased mortality risk. This holds true for both sexes and not only for hip fracture, but also for spine fractures, which trigger a complex chain of events. They represent an occurrence that cannot be overlooked in the management of a patient with spine and joint complaints. However, surprisingly, few studies have been performed — and only recently — on the prevalence of fractures in patients with PsA9,10,11.
We sought to contribute to this area with a retrospective study using as cases subjects from our PsA outpatient clinic, whose data are collected following a standardized protocol. A peculiarity of this study is the use of control subjects selected from a population-based epidemiological study that we performed in the female population of the Naples, Italy, area and whose data analysis continues.
The aim of the present study was to describe the prevalence of fractures in patients with PsA as compared to healthy controls and to investigate determinants of fractures among cases. This is the first study, to our knowledge, of fragility fracture prevalence in patients with PsA as primary aim and fracture identification based on radiographic findings.
MATERIALS AND METHODS
Our continuing population-based epidemiological study is aimed at defining prevalence and risk factors for OP in a healthy, active, noninstitutionalized female population from the Naples area12. Sample size was representative of the general population and was adequately powered. Recruitment for the epidemiological study has reached a very relevant completion rate (85.3%). According to the initial design of the study, in the global sample, 100 subjects were randomly selected by age group to undergo spinal radiography. They represent the control subjects of the present study.
Patients were 100 postmenopausal women with PsA attending the PsA outpatient clinic of the Rheumatology Unit of the University Federico II over the last year. They were consecutively recruited to complete matching with the control subjects by 5-year age classes. Exclusion criteria were early menopause, early PsA (< 2 yrs of diagnosis), secondary causes of OP (drugs and diseases), fractures due to major trauma from impact, and bisphosphonate use for > 1 year for subjects with no fractures. After recruitment completion, 8 cases withdrew their consent for inclusion into the study and their data were not used. The age-group distribution of these 8 subjects did not significantly affect case-control matching. The final number of cases analyzed was 92. For cases, the data used were those relative to the most recent visit with a spinal radiograph.
Radiographic readings aimed at identifying vertebral fractures (VF) were performed in both cases and controls by one of the first 4 authors of the present contribution, using the Genant semiquantitative method13. Findings on the presence of femoral neck fractures or other nonvertebral fractures (NVF) were obtained from the patients’ medical history. For each subject the following were collected: age, years since menopause, maternal history of hip fracture, weight, height, smoking habits (and mean number of cigarettes per day), hip fractures (total number), major peripheral non-hip fractures (humerus, wrist, tibia, pelvis, VF, severe VF), t score (lumbar dual-energy x-ray absorptiometry, femoral neck, femoral total), and 25-hydroxy Vitamin D. Among cases only: disease duration (yrs since first diagnosis), drugs and duration of administration (disease-modifying antirheumatic drugs, biologic therapies, bisphosphonates), steroids (cumulative dose), arthritic subset, and Health Assessment Questionnaire. Correlation data were obtained using multiple logistic regression analysis to account for the different confounding factors involved.
RESULTS
Cases and controls, all postmenopausal women, did not differ in their major demographic, anthropometric, and menopause duration data. Mean age (yrs ± SD) was 62.2 ± 7.3 in cases and 58.5 ± 9.4 in controls. The number of years since menopause was 12.3 ± 7.2 in cases and 13.9 ± 9.7 in controls. Body mass index (BMI) was 24.1 ± 3.7 in cases and 22.7 ± 3.8 in controls. Among cases, disease duration was 7.9 ± 4.5 years. Bone mineral density (BMD) t score at femoral neck did not differ significantly between cases (−1.1 SD) and controls (−1.6 SD).
The prevalence of fragility fractures in the radiographic readings did not differ between cases and controls. The number of subjects showing a VF was 33 (36%) among patients with PsA and 36 (36%) among controls, with a prevalence of severe VF of 8% among cases and 4% among controls. When considering the prevalence of any fracture we found 38 cases (41%) among patients with PsA and 43 among controls. Femoral neck fractures were 2 each in cases and controls. Other major fractures (pelvis, humerus, wrist, tibia) were found in 8 cases (9%) and 9 controls. When densitometric data were considered, the prevalence of osteopenic subjects was not statistically different between cases (41%) and controls (52%), while prevalence of osteoporotic subjects was greater among controls (20%) as compared to patients with PsA (9%), even if the data did not reach statistical significance (p = 0.054).
We then analyzed the fracture data for the entire study population using multiple logistic regression to control results for potential confounding factors. Using as dependent variable the presence of NVF, the only variable showing a significant (direct) correlation was age (p = 0.03), while no correlation was found with being case or control, and with other covariates (femoral neck BMD, smoking habit, BMI). Similar results were obtained using the presence of VF as dependent variable in multiple logistic regression. Only age showed a direct significant correlation (p < 0.01), while no correlation was found with being case or control or with other covariates.
We decided then to analyze among cases the correlation between the presence of fragility fractures and the other study variables to evaluate determinants of fractures in patients with PsA. Using the presence of NVF as dependent variable in multiple logistic regression, we analyzed the correlation of several variables: age, sex, subset type, disease duration, femoral BMD, cumulative steroid dose, steroid use (yes/no), biological therapies use (yes/no), and smoking habits. Controlling for the covariates in the logistic model, the only variables showing a significant correlation with the presence of NVF were disease duration (p = 0.02), age (p = 0.03), and BMD at the femoral neck (inverse correlation, p = 0.04). Similarly, we evaluated the correlation of VF with the study variables in the logistic model. Again, age showed a strong positive correlation (p < 0.01), as did disease duration (p = 0.03) and the presence of a peripheral subset (p = 0.01), while no other covariate significantly correlated with the presence of VF among subjects with PsA.
DISCUSSION
Vertebral and nonvertebral fragility fractures represent the most serious complication of OP. Fractures cause disability, and significant clinical, social, and economic burden for patients and their families. With regard to disability, fewer than 20% of women who fracture a hip will recover to their prefracture competence in activities, and more than 40% of them will never walk again without assistance. Another area of concern about fractures relates to mortality. Not only hip fracture but also VF are associated with a significant increase in mortality in both sexes. Therefore, fragility fractures represent an occurrence that cannot be overlooked in the management of chronic patients with spine and joint complaints as PsA patients.
However, as we pointed out in another report8, data on fracture in patients with PsA were completely lacking, with the exception of only 1 article reporting, as an ancillary finding, data on prevalence of low-impact fractures in postmenopausal women with PsA9. Data refer to a sample of 45 patients with PsA, 52 subjects with psoriasis, and 98 healthy controls. The authors describe a 33.3% prevalence of fragility fractures among patients with PsA, significantly greater as compared to the finding (28.8%) among psoriatic patients. Both these fracture rates are significantly greater versus controls, but the prevalence of fractures in controls is not described. Interestingly, in this study BMD data of the lumbar spine and proximal femur are not significantly different in the 3 groups.
Two other reports have studied the question10,11. Both investigated fracture prevalence in PsA outpatients excluding subjects with axial involvement, and only 1 had a control group11. Fracture ascertainment was based on medical history and not on radiographic findings. Therefore, most of the events recorded were wrist fractures, with likely underestimation of VF. The reported prevalence of clinical fragility fractures was, respectively, 13% and 14%. The latter result was significantly higher as compared to the prevalence in the control group11.
It is surprising how little attention this topic has received in the literature. The present study wishes to help reverse this trend. This may contribute to better knowledge of such a complex condition as PsA, which involves not only skin and joints but is increasingly recognized as psoriatic disease, affecting multiple domains14,15,16.
Our present study reports data from a group of cases and controls using radiographs read for VF ascertainment. The control group was recruited from a representative sample. Data obtained need to be considered in the general context. A preliminary consideration is the lack of data standardized for age and sex. This makes thorough comparison between different results difficult. However, it is worth emphasizing that VF prevalence among patients with PsA found in the present work (36%) is higher as compared to 2 other studies on clinical fractures10,11, while it is comparable with prevalence reported in the only other available study9 (33%) using radiographic readings, even if the latter did not have as primary aim the evaluation of fracture prevalence. In addition, the prevalence among PsA subjects seems to be comparable with the VF prevalence observed in the available studies among patients with rheumatoid arthritis, ranging up to 36%17,18,19.
On the other hand, when considering the data among healthy subjects, it should be noted that in the present study the VF prevalence among controls was 36%, which seems to be considerably higher when compared to literature data. In fact, in the general population aged 50 years and over, the most robust studies show a VF prevalence ranging from 6 to 25%20. This higher prevalence seems to be consistent with the finding of a high prevalence of osteoporotic subjects among the general population in the Naples area12. Further, this population shows a high prevalence of hypovitaminosis D, which is present in over 25% of subjects aged 50 years and over21. Hypovitaminosis D is a relevant risk factor for demineralization and fracture. This relationship has also been described in the general population of Naples. In a 2-year longitudinal study, using stepwise analysis, femoral neck BMD was shown to directly correlate with baseline 25(OH)D serum levels, which turned out to be the only variable to show statistical correlation in a large set of study variables21. Therefore, we believe that controls in the present study have a peculiar profile. They may be considered representative of a population at particularly high fracture risk and this risk, mainly due to the large prevalence of hypovitaminosis D, may be substantially shared with the subjects with PsA living in the same area, as part of the same general population. This may help to explain the lack of difference between cases and controls, given that control subjects show a prevalence of fracture that is higher versus other populations and otherwise similar to that observed among patients with PsA and other forms of arthritis.
Data suggest the need for further studies aimed at evaluating fracture prevalence and its relationship with Vitamin D concentration among patients with PsA in different populations.
Our present study is the first attempt, to our knowledge, to evaluate, as a primary aim, fragility fracture prevalence among patients with PsA as compared to a group of healthy controls representative of the Naples general population, using radiographic evaluation of fractures. No differences were found in VF or non-VF prevalence between cases and controls in our population. Age was shown to be the only determinant of fracture presence.
It is important to consider that for patients with PsA, age and disease duration were the only variables that correlated with the presence of VF and non-VF, while non-VF was also inversely correlated with femoral neck bone densitometry, and VF correlated with the presence of a peripheral subset of joint involvement.
One in 3 subjects with PsA presents a vertebral fragility fracture, and 41% of these patients have a major fracture. These data indicate a finding with a significant clinical relevance. Fractures should be carefully considered when evaluating the global picture of the patient with PsA for their contribution to the “fragility” profile, emphasizing, even in this domain, the relevance of the concept of psoriatic disease.