Localized disruption of bone architecture leads to an increased risk of pathological fractures in patients with Paget's disease, but the impact of the disease on overall fracture risk is unknown. We addressed this issue among 236 Olmsted County, Minnesota residents (107 women and 129 men) first diagnosed with Paget's disease from 1950 through 1994. These subjects (mean ± SD age at diagnosis, 69.6 ± 12.2 years) were followed subsequently for 2798 person-years. During this period of observation, 33 pathological fractures were attributed to Paget's disease (1 skull, 11 vertebra, 1 shaft/distal humerus, 1 pelvis, 6 proximal femur, 2 shaft/distal femur, and 11 tibia/fibula). Excluding the fractures through pagetic bone, there was no increase in overall fracture risk in this cohort (standardized incidence ratio [SIR], 1.2; 95% CI, 0.9-1.4). However, there was a statistically significant increase in the risk of subsequent vertebra (SIR, 3.0; 95% CI, 2.2-4.1) and rib fractures (SIR, 1.7; 95% CI, 1.1-2.4) but not fractures of the proximal femur (SIR, 0.6; 95% CI, 0.3-1.1) or distal forearm (SIR, 1.4; 95% CI, 0.7-2.5). Thus, unselected patients with Paget's disease in the community, who mostly have mild disease, have a significantly increased risk of vertebral fractures, although this may relate partly to increased surveillance. Additional work is needed to clarify the relationship between Paget's disease and vertebral fractures and to identify individuals at increased risk for more aggressive therapy.
PATHOLOGICAL FRACTURES are a well-recognized complication of Paget's disease of bone.(1) It is clear on the basis of case reports and series of patients from tertiary referral centers that partial and complete pathological fractures occur with increased frequency in weight-bearing bones of the lower extremities.(2) The effect of Paget's disease on fracture risk at uninvolved skeletal sites is unknown, although there is evidence of increased remodeling in uninvolved bone(3) that has been attributed to a factor capable of stimulating osteoclast formation and activity both within the lesion and at distant sites. Reddy and colleagues(4) have shown that the factor stimulating osteoclast formation in marrow culture is interleukin-6 (IL-6), that osteoclast-like cells produce large amounts of IL-6 messenger RNA (mRNA), and that IL-6 levels are markedly increased in peripheral blood and marrow from patients with Paget's disease. Moreover, using in situ hybridization, Hoyland and coworkers(5) found that IL-6, IL-6 nuclear factor, and IL-6 receptor mRNA were increased markedly in osteoclasts from patients with Paget's disease compared with normal controls. The fact that IL-6 is capable of stimulating normal osteoclast formation(6,7) and preliminary data showing that normal peripheral blood monocytes infected with measles virus express high levels of IL-6(4) further support the concept that IL-6 may play an important role in pathogenesis by increasing osteoclast formation, with a resultant increase in bone remodeling(8) that would be expected to increase the risk of subsequent fractures.(9) Empirical evidence for an increase in fracture risk is lacking, however. Fracture rates of 1.5-6.6 per 100 person-years have been reported from referral centers,(10) but there are no data on the risk of fractures generally among unselected Paget's patients from the community. We sought to fill that gap with a population-based study comparing the risk of fracture among Olmsted County, Minnesota residents with Paget's disease with that expected for the general population.
MATERIALS AND METHODS
Population-based epidemiological research can be conducted in Olmsted County, Minnesota because medical records for the entire population are available from almost all providers of care. Most of the endocrinologic and orthopedic care is provided by the Mayo Clinic, which has maintained a common medical record with its two large affiliated hospitals (St. Marys and Rochester Methodist) for over 90 years.(11) Therefore, this dossier-type record contains both inpatient and outpatient data. The diagnoses and surgical procedures recorded in these records are entered into a computerized index. Medical records of other providers who serve the local population, most notably the Olmsted Medical Group and its affiliated Olmsted Community Hospital (Olmsted Medical Center), are indexed into the same system and also are available for review.(12) After approval by Mayo's Institutional Review Board, this unique database (the Rochester Epidemiology Project) was used to identify the 236 Olmsted County residents who initially presented with Paget's disease of bone during the period from 1950 through 1994, as described in detail elsewhere.(13)
These subjects were then followed forward in time through their linked medical records in the community (retrospective cohort study) until death or the most recent clinical contact. For each subject, all inpatient and outpatient records at any local provider of medical care were searched for the occurrence of fractures. Mayo Clinic records, for example, contain the details of every inpatient hospitalization, every outpatient office or clinic visit, and all emergency room and nursing home care, as well as all laboratory results, all radiographic and pathology reports, including autopsies, and all correspondence with each patient.(11) The records contained the clinical history and the radiologist's report of each fracture, but the original radiographs were not available for review. Thus, the diagnosis of vertebral fracture was accepted on the basis of a radiologist's report of compression or collapse of one or more thoracic or lumbar vertebrae. “Pagetic fractures” were those that involved affected skeletal sites, whereas fractures through normal bone in the patients with Paget's disease were considered “nonpagetic.” Fractures were classified further according to the circumstances of the injury: by convention, falls from standing height or less were considered moderate trauma, whereas motor vehicle accidents and falls from greater heights were deemed severe trauma. Ascertainment of fractures is believed to be complete except for vertebral fractures, some of which are never diagnosed.(14)
The influence of Paget's disease on subsequent fracture risk was evaluated using three basic methods of analysis. In the primary analysis, we calculated standardized incidence ratios (SIRs), comparing the number of fractures that were observed at each skeletal site (based on the first fracture of a given type per person) with the number expected in this cohort during their follow-up in the community. One patient, who had no follow-up after age 35 years, was deleted from this analysis. Expected numbers of fractures were derived by applying age- and sex-specific incidence rates from the local population for these fractures(14–19) to the age- and sex-specific person-years of follow-up in the Paget's cohort. Ninety-five percent CIs for the SIRs were calculated assuming that the expected rates are fixed and the observed fractures follow a Poisson distribution.(20)
In the second method of analysis, the cumulative incidence of new fractures (1 − survival free of fracture) was projected for up to 20 years after the initial diagnosis of Paget's disease, using product-limit life table methods.(21) Cumulative incidence curves were compared using the log-rank test statistic.(22)
Finally, Cox proportional hazards models,(23) which do not incorporate the population expected rates, were used to assess the impact of various covariates on subsequent fracture risk. Stepwise methods with forward selection and backward elimination were used to choose independent variables for the final models, and interactions among the significant main effects were assessed. The dependent variable was time until the first new fracture, and the independent variables were age, sex, and clinical characteristics. For the final multiple models, as well as for the univariate models, the assumption of proportional hazards was examined and was not violated for the variables considered.
Two hundred thirty-six Olmsted County, Minnesota residents were first diagnosed with Paget's disease during the period of 1950–1994. All but two were white, reflecting the racial composition of the community (98% white in 1980). Their mean (±SD) age at diagnosis was 69.6 ± 12.2 years (median, 70.6 years; range, 23–95 years); 129 (55%) were men compared with 107 women. These subjects were subsequently followed for 2798 person-years (median, 10.5 years per subject; mean, 11.9 ± 8.4 years). During this period of observation, 101 subjects experienced 240 different fractures. Forty-three of the fractures (18%) were caused by a specific pathological process, 33 of which were attributed to Paget's disease (1 skull, 11 vertebrae, 1 shaft/distal humerus, 1 pelvis, 6 proximal femur, 2 shaft/distal femur, and 11 tibia/fibula) with the remainder mostly caused by metastatic malignancies. Of interest, 79% of the tibia/fibula fractures and 62% of all fractures of the femur or tibia shaft involved pagetic bone. Some of these represented multiple fractures in the same individual (1 patient had 6 separate fractures of the tibia or fibula and another had 5 separate fractures). Sixteen (7%) of the remaining fractures were caused by severe trauma (4 from motor vehicle accidents and 12 because of falling from a height) and 11 (5%) resulted from miscellaneous other causes, but the majority of fractures (150, 63%) were attributed to moderate or minimal trauma. Most limb fractures were caused by falling from a standing height or less, while most of the vertebral fractures occurred “spontaneously” during the activities of daily living. An additional 20 (8%) fractures could not be associated with a specific episode of trauma. The immediate cause of the 33 pagetic fractures was almost always minimal or moderate trauma: 27% resulted from a fall and 58% occurred spontaneously.
The overall incidence of fractures after the diagnosis of Paget's disease was 8.6 per 100 person-years, while the incidence of subsequent fractures through pagetic bone was 1.2 per 100 person-years. The overall incidence was similar in women and men (9.7 vs. 7.6 per 100 person-years, respectively), as was the incidence of subsequent pagetic fractures (1.5 vs. 0.9 per 100 person-years, respectively). Fracture risk was relatively constant over time, both for pagetic fractures and for fractures generally, as shown in Fig. 1. The cumulative incidence of any fracture at 20 years was an estimated 61% and the cumulative incidence of any subsequent pagetic fracture was 19% at 20 years.
The number of nonpagetic fractures observed at each skeletal site is compared with the number expected and the corresponding SIRs in Table 1. The totals differ somewhat from those noted previously because the fractures through pagetic bone were excluded from this analysis and because subjects, who had had the particular fracture of interest previously, were excluded from analyses at specific skeletal sites. In general, there was a slight increase in overall fracture risk in men (SIR, 1.5; 95% CI, 1.1-1.9), but not in women (SIR, 0.9; 95% CI, 0.7-1.3), or both sexes combined (SIR, 1.2; 95% CI, 0.9-1.4). However, there were statistically significant increases in the risk of vertebral and rib fractures (Table 1). Even after exclusion of the pagetic vertebral fractures, the risk of a subsequent vertebral fracture was significantly increased in men (SIR, 4.3; 95% CI, 2.7-6.5), women (SIR, 2.4; 95% CI, 1.5-3.6), and both sexes combined (SIR, 3.0; 95% CI, 2.2-4.1). This is illustrated in Fig. 2. Likewise, because no rib fractures were attributed to Paget's disease, pathological fractures could not be the explanation for the increase in rib fractures either (SIR, 1.7; 95% CI, 1.1-2.4). There also was a significant increase in the risk of distal femur fractures in men (Table 1). However, if all long bone fractures of the lower limb (femur shaft/tibia/fibula) were combined, there was no statistically significant increase in risk either in men (SIR, 2.0; 95% CI, 0.7-4.4), women (SIR, 1.0; 95% CI, 0.3-2.3), or both sexes combined (SIR, 1.4; 95% CI, 0.7-2.4). At other skeletal sites that have been associated closely with osteoporosis, the observed number of fractures was not significantly different from the number of fractures expected in this cohort given the distribution by age, gender, and duration of follow-up; the SIR for fractures of the proximal femur was 0.6 (95% CI, 0.3-1.1) or for distal forearm fractures was 1.4 (95% CI, 0.7-2.5).
Table Table 1.. Nonpagetic Fractures Observed After the Initial Diagnosis of Paget's Disease of Bone in 1950–1994 Compared with the Numbers Expected and SIRs Among Olmsted County, Minnesota Residents
Excluding the pagetic fractures, fracture risk increased with age (hazard ratio [HR], 1.67 per 10-year increase; 95% CI, 1.35-2.06) for fractures generally and also for vertebral fractures specifically (HR per 10 years, 2.02; 95% CI, 1.51-2.69). Other clinical characteristics at diagnosis were not predictive of overall fracture risk in univariate analyses (Table 2). In particular, the baseline serum alkaline phosphate level was not a predictor of subsequent fracture risk, nor was the number of sites of pagetic involvement (67% of patients appeared to have monostotic disease). Eight patients received calcitonin treatment at some point in the course of their disease (beginning a mean of 7.7 years after diagnosis; range of calendar years, 1970–1985), while 23 were treated with bisphosphonates (mean, 11.9 years after diagnosis; range of calendar years, 1978–1993). Most patients received etidronate (96%); alendronate was used in only 1 patient beginning in 1992. Any treatment (either yes/no or as a time-dependent variable) was associated with a reduction in the risk of fractures in general and vertebral fractures specifically, but these differences were not statistically significant either in the univariate (Table 2) or multivariate analyses. After adjusting for age (HR per 10 years, 1.78; 95% CI, 1.33-2.40) in a multivariate model, later year of diagnosis was still an independent predictor of vertebral fracture risk (HR, 1.06; 95% CI, 1.02-1.09), but none of the other variables contributed. The only independent predictor of any fracture was age (HR per 10 years, 1.71; 95% CI, 1.37-2.13). Results were the same when pagetic fractures were included in the analysis.
Table Table 2.. Univariate HRsa with 95% CIs for the Development of Nonpagetic Fractures at Different Skeletal Sites Among Olmsted County, Minnesota Residents After the Initial Diagnosis of Paget's Disease of Bone in 1950–1994
This first population-based study revealed no significant increase in the overall risk of fractures among unselected patients from the general population after an initial diagnosis of Paget's disease. This is not necessarily the experience at tertiary medical centers where more severe involvement and/or complications of Paget's disease are indications for referral. By contrast, only one-third of the community patients in our study were symptomatic at diagnosis, and they typically had mild disease.(13) Although radiological confirmation was obtained in all cases, 67% apparently had involvement at a single skeletal site and only 37% had documentation of an elevated serum alkaline phosphatase level. The lack of an overall increase in fracture risk notwithstanding, there clearly was an excess of pathological fractures among patients with Paget's disease. Fractures through pagetic bone accounted for 14% of all subsequent fractures in this cohort, whereas pathological fractures of all types account for only 2% of fractures in the Olmsted County population generally and even those typically result from metastatic malignancy.(19) This re-emphasizes both the fact that patients with Paget's disease may present with a fracture and the fact that pagetic fractures tend to recur.(24) Indeed, a classic hallmark of the disease is pagetic fracture of the tibia or femur.(25) However, there was no overall increase in diaphyseal fractures of the lower limb either in men or women. The location of the other pagetic fractures, mostly of the spine and proximal femur, is consistent with previous reports.(10)
Although the incidence of subsequent fractures observed in this cohort (8.6 per 100 person-years) was actually greater than previously reported (rates of 1.5-6.6 per 100 in series of more selected patients(10)), the higher rate probably reflects more complete ascertainment of fractures in this study. However, compared with expected fracture rates in the elderly population generally, fractures were not more common among the Paget's patients in the community. Indeed, the only independent predictor of overall fracture risk in this cohort was advanced age, which is a well-known determinant of fracture risk in the general population.(26) Conversely, there was no significant decrease in overall fracture risk associated with specific treatments, but only a minority of these mildly affected patients were ever treated. Moreover, most such treatment was initiated long after the original diagnosis of Paget's disease was made and involved use of calcitonin or etidronate. Bisphosphonates that are more potent were only introduced near the end of the study period(1) so few patients benefited from them during this period of observation. For these reasons, the present analysis does not speak to the effectiveness of current recommendations for the treatment of patients with more severe Paget's disease.(27)
However, even after the pagetic fractures were excluded there was an increase in subsequent fractures of the vertebrae and ribs possibly related to a generalized increase in bone turnover as described in the Introduction. Predisposition for fractures at sites containing primarily cancellous bone, such as the vertebrae, is consistent with this theory, but there was no suggestion of any increase in the risk of the other fractures that have been most closely linked to loss of cancellous bone, fractures of the proximal femur and distal forearm.(26) Alternatively, an increased risk restricted to fractures of the spine and ribs might be explained by detection bias, that is, if Paget's patients are more likely relative to community residents in general to have X-rays for back or chest pain, there would be increased detection of otherwise undiagnosed vertebral and rib fractures. This is a potential problem because the expected number of vertebral fractures in this analysis was based on the incidence of clinically recognized fractures.(14) A more complete assessment of vertebral fracture occurrence can be obtained from incidence rates that are derived from prevalence data.(28) However, such rates overestimate true vertebral fracture incidence in the general population because some vertebral deformities do not represent actual clinical fractures. Nonetheless, if the latter rates are substituted in the analysis, overall vertebral fracture risk is still elevated (SIR, 1.7; 95% CI, 1.3-2.3).
Strengths of the present investigation include the use of a population-based inception cohort that included both institutionalized and community-dwelling individuals registered at the time that their Paget's disease was first recognized. Because of the unique medical records linkage system in Olmsted County, which provides access to the inpatient and outpatient medical records of the entire community,(12) there should be nearly complete ascertainment of Paget's disease to the extent that the condition came to clinical attention.(13) Likewise, with the exception of vertebral fractures and possibly some rib fractures, ascertainment of the fracture outcomes should be complete whether they were attended on an inpatient or outpatient basis. This is evidenced by the fact that fracture incidence rates in this community are three times higher(19) than those reported from the only comparable study.(29) Specifically, the vertebral fracture incidence rates in this community(14) are much greater than those from other populations.(30–34) Also, there was considerable follow-up and a large number of subsequent fractures, which provided adequate statistical power. A limitation of the study is the generalizability of these data from a small Midwestern community that is predominantly white and better educated than the white population of the United States.(12) However, there are no other population-based studies that estimate the risk of fractures after an initial diagnosis of Paget's disease.
In conclusion, these unselected patients in the community, who mostly had mild disease, experienced no overall increase in fracture risk, although there was an excess of pathological fractures attributed to Paget's disease. However, they did have a significantly increased risk of vertebral fractures despite the fact that some individuals received antiresorptive therapy in the form of calcitonin and etidronate during the latter part of the observation period. Although the increased vertebral fracture rate may have been related in part to closer surveillance, there appears to be an impact of Paget's disease on fracture risk that may be related directly to the effects of the disease process on uninvolved skeletal sites or indirectly related to the impact of the disease on the patient's general health. Additional studies will be needed to identify the individuals who are at increased risk for fracture in order to target these individuals for more aggressive preventive therapy.
The authors thank Mrs. Margaret L. Bitzer-Abramowitz for help with data collection, Ms. Sara J. Achenbach for assistance with data analysis, and Mrs. Mary G. Roberts for help in preparing the manuscript. This work was supported by a contract from Sanofi-Synthelabo, Inc. and by grants AG 04875 and AR 30582 from the National Institutes of Health, United States Public Health Service.