Paget's Disease of Bone in The Netherlands: A Population-Based Radiological and Biochemical Survey—The Rotterdam Study

Authors

  • Marelise EMW Eekhoff,

    1. Department of Endocrinology and Metabolic Diseases, Leiden University Medical Center, Leiden, The Netherlands
    2. Department of Internal Medicine, Erasmus University, Rotterdam, The Netherlands
    3. Department of Endocrinology, VU University Medical Center, Amsterdam, The Netherlands
    Search for more papers by this author
  • Marjolein van der Klift,

    1. Department of Internal Medicine, Erasmus University, Rotterdam, The Netherlands
    2. Department of Epidemiology, Erasmus University, Rotterdam, The Netherlands
    Search for more papers by this author
  • Herman M Kroon,

    1. Department of Radiology, Leiden University Medical Center, Leiden, The Netherlands
    Search for more papers by this author
  • Cyrus Cooper,

    1. MRC Environmental Epidemiology Unit, Southampton General Hospital, Southampton, United Kingdom
    Search for more papers by this author
  • Albert Hofman,

    1. Department of Epidemiology, Erasmus University, Rotterdam, The Netherlands
    Search for more papers by this author
  • Huibert AP Pols,

    1. Department of Internal Medicine, Erasmus University, Rotterdam, The Netherlands
    Search for more papers by this author
  • Socrates E Papapoulos

    Corresponding author
    1. Department of Endocrinology and Metabolic Diseases, Leiden University Medical Center, Leiden, The Netherlands
    • Department of Endocrinology and Metabolic Diseases, Leiden University Medical Center, PO Box 9600, 2300 RC Leiden, The Netherlands
    Search for more papers by this author

  • The authors have no conflict of interest.

Abstract

Serum ALP may be a good indicator of Paget's disease in epidemiologic studies. Subjects with raised and normal ALP from a population cohort were matched (1 in 6, total 548), and radiographs were taken. ALP was an excellent marker of the disease (RR, 10.9), but the majority of those affected had normal ALP.

Introduction: Evidence from radiographic surveys of limited skeletal sites has shown that Paget's disease of bone (PDB) is common in the elderly and has a distinct geographic variation. There is no information, however, about the relation of serum alkaline phosphatase (ALP) activity, a marker of the disease, and its prevalence in the population.

Materials and Methods: We analyzed data from a well-defined Dutch population cohort (the Rotterdam study) with the following specific aims: (1) to assess the relationship between serum ALP activity and prevalence of radiographically diagnosed PDB, (2) to estimate the overall prevalence of the disease in the Netherlands, and (3) to assess the appearance of the disease with time. Using a nested case-control design, subjects with an increased serum ALP and normal serum liver enzymes were matched for gender and age (1 to 6) with subjects with normal serum ALP activity. Radiographs of the thoracic and lumbar spine, pelvis, proximal femurs, knees, wrists, and hands were taken.

Results and Conclusions: PDB was diagnosed in 20.5% of subjects with elevated serum ALP activity and in 2.3% in those with normal serum ALP activity, increasing with age in both groups. The relative risk (RR) for PDB in the presence of raised serum ALP activity was 10.9 (95% CI, 4.8, 24.9). The estimated prevalence of PDB in the population was 3.6%, and the large majority (about 86%) had normal serum ALP activity, contrasting findings in bone clinics where the opposite is the case. Finally, in subjects with normal and raised serum ALP activity but no PDB at baseline, radiographs taken 6–9 years later showed no evidence of the disease. This study demonstrated that serum ALP activity is a sensitive marker of PDB in men and women >55 years of age, but the majority of those affected have normal serum ALP activity.

INTRODUCTION

PAGET's DISEASE OF bone (PDB) has a distinct geographical variation, its prevalence increases with age, and it affects 1-5% of those above 50 years of age. Estimates of the prevalence of the disease have been mainly derived from autopsies or radiographic surveys of subjects investigated in hospitals.(1-31) Only two population-based estimates of the prevalence of the disease have been reported. In the first, spine radiographs of French women >75 years of age were evaluated and showed a prevalence of 1.1-1.8% of PDB in this population.(32) The second was conducted in the United States and included a population-based cohort. The prevalence of PDB, estimated from pelvic radiographs, was 1-2% and increased with age.(33)

Although all these studies are very useful for estimating the prevalence of the disease, they have certain limitations. They can be biased (hospital-based), restricted to very specific populations (French survey), or may underestimate the true prevalence of the disease because of the limited skeletal sites assessed (U.S. survey). Skeletal radiographs are the gold standard for diagnosing PDB, but markers of disease activity, such as serum alkaline phosphatase (ALP) activity can also be very useful when considering that the majority of patients are asymptomatic.(1) To address these issues, we analyzed data from a well-defined Dutch population cohort with the following specific aims: (1) to assess the relationship between serum ALP activity and prevalence of PDB in men and women older than 55 years, (2) to estimate the overall prevalence of PDB in the Netherlands, and (3) to assess possible radiographic changes longitudinally.

MATERIALS AND METHODS

Study population

The Rotterdam Study is a prospective population-based cohort study of men and women ≥55 years of age and has the objective to investigate the incidence of, and risk factors for, chronic disabling diseases. Both the rationale and the study design have been described previously.(34) The focus of the Rotterdam Study is on neurologic, cardiovascular, ophthalmologic, and locomotor diseases. All 10,275 inhabitants of Ommoord, a district in Rotterdam, the Netherlands, were invited to participate. Of these, 7983 (4878 women) participated in the study (resulting in a response rate of 78%). The Medical Ethics Committee of the Erasmus University Medical School has approved the Rotterdam Study.

Between March 1990 and July 1993, an extensive baseline home interview on medical history, risk factors for chronic diseases, and medication use was performed by trained interviewers. After the home interview, the participants were invited to the research center for clinical examination, laboratory assessments, and skeletal radiographs.

Nonfasting blood samples were taken for the determination of serum total ALP activity, serum aminotransferase aspartate (ASAT), and aminotransferase alanine (ALAT).

Study design

A nested case-control design within the Rotterdam Study was used. Cases were subjects with a serum total ALP activity higher than 2 SD above the mean value of the whole cohort at baseline (>131.5 U/liter) and who had normal liver function, demonstrated by normal serum ALAT (<30.3 U/liter) and ASAT (< 30.6 U/liter) values. Controls were all subjects with normal serum ALP (<131.5 U/liter), ALAT, and ASAT values. Per case, six age- and gender-matched controls were randomly selected from the total study population per 5-year intervals up to 85 years. Serum total ALP activity was measured by a fotometric assay using 4-nitrophenylphosphate and H2O. Radiographs of hands, thoracic, and lumbar spine (anterior, posterior, and lateral), pelvis, and knees were taken as described below. Body mass index (BMI) was calculated from weight and length, and BMD of the femoral neck was measured by DXA (DPX-L; Lunar, Madison, WI, USA).

Complete data on serum ALP activity, ALAT, and ASAT were available in 4837 subjects. Of those, 108 subjects (2.23%) had an elevated level of serum ALP activity, and 105 also had normal serum ASAT and ALAT values and were therefore included as cases. Of the remaining subjects, 4301 had normal serum ALP activity, ALAT, and ASAT, and of those, 625 were selected as controls. Of the 730 included subjects (105 cases and 625 controls), a full set of radiographs was available in 548 subjects (73 cases and 475 controls). Radiographs were not available mainly in subjects above the age of 75 years, and the number of missing radiographs was similar in cases and controls (30% and 34%, respectively). Therefore, the final study population consisted of 73 cases and 475 controls with complete data on serum ALP activity, ALAT, ASAT, and radiographs. The characteristics of both cases and controls are shown in Table 1.

Table Table 1.. Characteristics of the Total Population Cohort and Studied Population
original image

Radiographic assessment of PDB

At baseline, seven radiographs per subject were taken by two trained research technicians following a standard protocol, with a distance between source and plate of 120 cm, using a Solarize FV (General Electric CGR, Utrecht, the Netherlands). In total, three lateral radiographs of the thoracolumbar spine (Th4-S1), one antero-posterior of the pelvis and proximal femora with both feet in 10° endorotation, one weight-bearing antero-posterior of the knees (distal femora and proximal tibias), and one postero-anterior of the hands and wrists were obtained. No skull X-rays were available because PDB was not a primary clinical outcome and were not included in the original design of the Rotterdam study. The criteria for radiographic diagnosis of PDB were those used in the epidemiologic studies in the UK studies.(4, 6, 12) These criteria were as follows.

• increased bone density, revealed as areas of decreased radiolucency in radiographs

• increased size of bone caused by rapid bone formation

• disorganization in the architecture of newly formed bone, resulting in bone deformation

• marked cortical thickening

• enhancement of the trabecular pattern

• additional features in the pelvis, included thickening of the iliopectineal line (“brim sign”) and protrusio acetabuli

To ensure comparability and consistency of radiologic diagnosis between our study and the UK study, the first observer (ME) was trained by one of the British observers who participated in the epidemiologic study in the UK.(4) The radiographs were classified by the first observer into three groups: positive (unequivocal evidence of PDB), doubtful, and negative. A second expert observer (SP) examined all positive and doubtful films and an equal number (1 in 10 sample) of negative films at random. Both observers subsequently re-examined all films when in disagreement, leading to a consensus score. The final diagnosis was made by an expert skeletal radiologist (HK), who examined all positive and doubtful films. A formal repeatability study was conducted to assess the within- and in-between observer variation using the first 42 radiographs. κ values (including positive, doubtful, and negative judgment) for the first two observers (0.68) as well as between the second observer and the expert radiologist (0.65) were good. All observers were blinded to the corresponding serum ALP activity values.

Statistical analysis

Differences in baseline characteristics were assessed by Student's t-test for the continuous variables and χ2 for categorical variables. Prevalence rates were calculated with 95% CIs. Logistic regression was used to calculate the age-adjusted relative risk (RR) for developing PDB according to the level of serum ALP activity. Relative risks were computed univariately and multivariately. We used the receiver operating characteristic (ROC) analysis of the levels of serum ALP activity for PDB. The ROC curve is a common test to quantitate the diagnostic accuracy of medical tests. The strategy plots the true positive fraction or sensitivity against the false positive fraction (1 − specificity) by varying the threshold value for the test. The threshold is varied with increasingly stringent criteria for positivity. The ROC curve thus indicates the probability of true positive result as a function of the probability of a false positive result for all possible threshold values. The area under the ROC curve assesses the relative accuracy of two diagnostic tests. An area of 0.5 indicates that the test results are no better than those obtained by chance, whereas an area of 1.0 indicates a perfectly sensitive and specific test. SPSS 9.0 for Windows was used.

RESULTS

PDB was diagnosed in 26 of the 548 subjects in whom a complete set of radiographs was available. In 14 subjects, only one bone was affected, whereas in the remaining, more than one bone was affected. The most commonly affected site was the pelvis (54% of the subjects), but other sites were also frequently affected, including the lumbar spine and the proximal and/or distal femora (Table 2). It is of interest to note that in 6 of the 26 patients (23%), the diagnosis was made on bones that are not evaluated in radiographic population or patients surveys. These bones included the hand, ulna, radius, distal femur, and proximal tibia.

Table Table 2.. Radiographic Distribution of Lesions of Paget's Disease in 26 Patients
original image

Relation of PDB to serum ALP activity

Eleven of the 26 subjects in whom PDB was diagnosed radiographically had normal serum ALP activity (range, 50-115 U/liter; median, 96 U/liter). This was 2.3% of the group of subjects with normal ALP activity. In those subjects, the disease was diagnosed mostly in one bone; only 2 of the 11 subjects had more than one localization. None of these subjects had received bisphosphonates in the past. In contrast, the prevalence of PDB was much higher among subjects with raised serum ALP activity. In these, the disease was diagnosed in 15 of the 73 who were evaluated, accounting for a prevalence of 20.5% among subjects with raised serum ALP activity. In this group, more than one bone was affected in 67% of the subjects. It is important to note that the radiographic changes of the disease were more marked in patients with raised serum ALP activity compared with those with normal serum ALP activity.

Subjects with an elevated serum ALP activity had a significantly higher risk for radiographically diagnosed PDB (age- and gender-adjusted: RR, 10.9; 95% CI, 4.8, 24.9). Other factors examined by multivariate analysis (BMI, serum creatinine, BMD) did not alter this risk. The ROC curve for serum ALP activity for cases versus controls is shown in Fig. 1, which underscores the sensitivity of this marker (area under the ROC curve, 0.821).

Figure FIG. 1..

ROC curves for the levels of serum ALP activity in relation to the radiographic diagnosis of PDB in the study population. The area under the ROC curve is 0.821 (serum ALP activity).

The prevalence of PDB increased with age in both groups of subjects (Fig. 2). For example, the prevalence in subjects with raised serum ALP activity was 12.7% between age 55 and 70 years and rose to 25.5% in those older than 70 years. The corresponding prevalence in subjects with normal serum ALP activity was 1.3% and 3.3%, respectively.

Figure FIG. 2..

(A) Prevalence of PDB in subjects with elevated serum ALP activity (EAP) per 5-year age strata. (B) Prevalence of PDB in subjects with normal serum ALP activity (NAP) per 5-year age strata.

Overall prevalence of PDB

The group with normal serum ALP activity was selected from the total population cohort to match the group of subjects with raised serum ALP activity. As a result of the study design, the final studied groups differed from the corresponding total population of the cohort in some age categories. To estimate the overall prevalence of PDB, the observed differences in age categories were accounted for, and results were adjusted for 5-year intervals to the frequency of representation of the various age categories for the whole cohort. This resulted in an overall prevalence of the disease of 3.6% in subjects >55 years of age. The estimated prevalence of PDB in subjects with normal serum ALP activity was 2.7% and rose to 26.7% in those with raised serum ALP activity. Thus, the prevalence was very similar to that observed in the studied groups.

Follow-up

To assess radiographic changes longitudinally, follow-up radiographs were obtained in 55% of the subjects without PDB at baseline (254 controls with normal serum ALP activity and 32 cases with raised serum ALP activity) after a median period of 7 years (range, 6-9 years). PDB was not diagnosed in any of these subjects.

DISCUSSION

The prevalence of PDB varies between 1% and 5% in people >50 years of age, with a marked geographical variation, which is highest in some regions of the UK, but the vast majority of these are asymptomatic.(1) There is, however, no information about the relation of serum ALP activity and the prevalence of PDB in the population. The unique design of the Rotterdam study provided an opportunity to accurately assess the radiographic prevalence of PDB in relation to serum ALP in a well-defined population cohort in the Netherlands. Moreover, the selection of subjects according to the level of serum ALP activity generated information that cannot be derived from other epidemiological studies.

We found that the prevalence of PDB increases with age in both subjects with normal and elevated serum ALP, being much higher in the latter group. One in 5 subjects with raised serum ALP activity had PDB, whereas the disease was diagnosed in 1 in 43 subjects with normal serum ALP activity. This accounts for an 11-fold higher risk for the disease in subjects with high serum ALP activity. The diagnostic accuracy of serum ALP activity as a marker for PDB was thus high (shown by the ROC curve).

Our study was primarily designed to assess the relationship between serum ALP activity and PDB in the population. The data obtained, however, also allowed the estimation of the overall prevalence of PDB in The Netherlands. The only available data for this country come from a small, hospital-based study of abdominal radiographs performed in 1982, which suggested a prevalence of 0.6%.(2) Our results revealed an overall estimated prevalence of PDB in The Netherlands of 3.6% in subjects >55 years of age that increased with age; this agrees with other studies. The prevalence was still 2.4% even after removing the 23% of the diagnoses based on localizations that were not investigated in any of the prior radiographic studies, indicating that the prevalence of the disease may have been underestimated until now. The distribution of the localizations showed a similar pattern to that of previous clinical studies, with the pelvis being most frequently affected; in 54% of the subjects with PDB, only one bone was affected.

Although serum ALP activity is clearly a sensitive marker of the disease, it should be appreciated that only 2.23% of the total cohort had an elevated level with normal liver enzymes, and we estimated that 26.7% of those had PDB (0.5% of the total cohort). Thus, of the total number of cases of PDB in the population, about 86% have normal serum ALP activity and only about 14% have raised serum ALP activity. These estimates contrast the findings in patients followed in bone clinics, 90% of whom have increased serum ALP activity.(1) It is clear, therefore, that only a small proportion of patients with PDB comes to clinical attention, most likely those with the more severe disease. This is underscored in our study, which showed more extensive radiographic changes in the subjects with raised serum ALP activity. These data further agree with estimates that only a small proportion of subjects (about 10%) with PDB are symptomatic.(1)

Although the precise etiology of PDB is not known, it is generally considered that the disease develops early in life and has a slow clinical course. These conclusions are derived from cross-sectional data. Our longitudinal radiographic follow-up, which showed no evidence of the disease 6-9 years after the first radiographs in subjects with normal and raised serum ALP activity, fully supports these ideas.

Comparison of our findings with other studies is difficult mainly because of the different methodologies used. However, the assessment of the prevalence of the disease in relation to a sensitive marker of its activity provided information that could not be derived from other epidemiologic surveys. Our study has also limitations. Older subjects from both groups could not travel to the X-ray center, and therefore, relatively less subjects in the oldest category were examined, and no skull X-rays were taken. Given the relations between age, PDB, and serum ALP activity, as well as between skull localizations of the disease and serum ALP activity,(35) we may have underestimated the prevalence of the disease. The population examined comes from one region of the country, and although unlikely when considering the size of the country, it is not known whether there may be regional differences in the prevalence of PDB in the Netherlands, as has been reported in the UK and the United States.

In conclusion, the combined use of biochemical and radiographic data in a well-defined cohort of women and men >55 years of age demonstrated the sensitivity of serum ALP activity as a marker of PDB and suggested that previously reported prevalences of the disease in The Netherlands may have been underestimated. Furthermore, there was no radiographic evidence of PDB after 7-year follow-up in either subjects with normal or raised serum ALP activity and normal baseline radiographs.

Ancillary