Introduction

Paget disease of bone (PDB) is a chronic, progressive metabolic skeletal disorder characterized by focal abnormal increases in bone resorption and formation leading to weakened bone strength and abnormal bone architecture [1–4]. Antiresorptive therapy with bisphosphonates is the first-choice treatment for PDB [1]. Of the six agents in clinical use, zoledronic acid has the highest affinity for bone mineral, resulting in a long half-life, and is the most potent inhibitor of farnesyl diphosphate synthase in vitro and bone resorption in vivo [3, 5–7]. The 6-month efficacy and safety of a single dose of 5 mg zoledronic acid administered intravenously over 15 min versus daily oral treatment with risedronate for 60 days were documented in the pooled analysis of two identical randomized, double-blind, controlled phase 3 trials [8]. At 6 months, 96.0 % of patients in the zoledronic acid arm had a therapeutic response, defined as normalization of serum alkaline phosphatase (ALP) levels or a reduction of ≥75 % in serum alkaline phosphatase (sALP) excess, compared to 74.3 % of patients in the risedronate arm (p < 0.001). ALP normalized in 88.6 vs. 57.9 % of patients in the respective arms (p < 0.001). Median time to first therapeutic response was 64 vs. 89 days, respectively (p < 0.001).

The effects of zoledronic acid may extend beyond 6 months. Patients who showed therapeutic response in the registration trials [8] were eligible to enter an open extension observational study. At 2 years, 98 % of patients on zoledronic acid were still in remission versus only 57 % for those on risedronate (p < 0.0001) [9]. A recent analysis compared a residual of 44 % of the original responders in the zoledronic acid arm and a residual of 18 % of those randomized to the risedronate arm up to 6.5 years following treatment. At 3 years, 88 % of patients treated with zoledronic acid had normalized ALP levels versus 43 % for those who received risedronate; at 6.5 years, these rates were 84 and 43 %, respectively [10].

Using data from the Belgian Paget’s Disease Registry, we report on the remission rates and remission duration observed over 3 years in 98 evaluable PDB patients treated with a single 5 mg infusion of zoledronic acid. We also report on determinants of and retreatment following relapse, as well as serum creatinine and calcium levels over the same time period.

Methods

Registry

The Belgian Paget’s Disease Registry is a multicenter disease and product registry of patients with PDB treated with a single IV infusion of 5 mg zoledronic acid. Patients were recruited into the registry at the time of their zoledronic acid treatment and followed for a minimum of 36 months on a schedule determined by their physician to monitor treatment outcomes and disease remission. There were no fixed time points for visits and data collection and no required data to be collected; only data available from routine clinical practice were recorded. The data model included patient demographics, clinical parameters of PDB, relevant laboratory and imaging tests, treatment and retreatment decisions, concomitant medications, and remission. Twenty-two Belgian sites participated in the registry, mainly rheumatology centers (n = 16) but also rehabilitation (n = 3) and internal medicine, geriatrics, and endocrinology centers (n = 1 each). Seven were academic medical centers.

The registry includes male and female adults (age ≥18 years) diagnosed with PDB and being treated de novo with a single IV infusion of 5 mg zoledronic acid over 15 min per their treating physician’s best clinical judgment and in accordance with treatment guidelines, a scientific leaflet, and Belgian reimbursement criteria. Both newly diagnosed patients and patients previously treated for PDB but for whom a new treatment was indicated were eligible for inclusion. Excluded were patients previously treated with zoledronic acid or with a treatment indication other than PDB. All patients (or their legal guardians) provided written informed consent to be included in the registry.

Time Points

The 1- and 2-year time points were calculated, respectively, as 12 ± 4 and 24 ± 4 months following infusion with zoledronic acid. The 3-year time point was calculated as any visit at least 32 months following infusion.

Effectiveness and Safety Outcomes

The primary outcome of interest was disease remission versus relapse as rated by the patients’ treating physician. Physicians were instructed to consider ALP normalization or a reduction of 75 % or more in excess phosphatase as indicators of therapeutic response, but this was not verified independently. The study protocol also specified that treatment should conform to the approved label and reimbursement criteria. In Belgium, zoledronic acid is reimbursed for PDB upon written certification by the treating physician that the patient (1) has either clinical active symptoms of Paget disease (e.g., pain, osteolytic lesions, fractures, bone deformities, nerve compressions) and that the ALP levels exceed the upper normal limit or (2) is clinically asymptomatic but has radiological signs of Paget disease and (2.1), for patients less than 50 years old, ALP levels exceed the upper normal limit or (2.2), for patients 50 years or older, ALP levels exceed more than twice the upper normal limit.

While we did not verify during the course of the study whether patients indeed showed the above-specified reduction of 75 % or more in excess phosphatase, we did so post hoc. We also applied post hoc the Russell et al. [11] historical definition of positive treatment response as a 30 % or greater fall in ALP compared to pretreatment values.

Secondary outcomes included duration of remission (in months) and, alternately, time to first relapse (in months), with early relapse possibly indicating non- or poor response to treatment. Time to first relapse was modeled as a function of comorbidities. Safety outcomes included adverse events (AEs) suspected to be related to zoledronic acid, as well as serum creatinine and calcemia.

Statistical Analysis

The analysis population consisted of the 98 patients for whom data were available at the 3-year time point. The safety population included the 142 patients who received an infusion of zoledronic acid.

Descriptive statistics included measures of frequency, central tendency, and dispersion under consideration of the level of measurement of the variables. Comparison with literature data was done using Fisher’s exact test for discrete data and the t test for independent samples for continuous variables. The latter test was also applied to compare the duration of remission between comorbid conditions. Proportional hazards regression analysis was performed to model the time between baseline and first visit in which the patient showed relapse, taking into account that the majority of patients remained in remission. Testing for trends over time in serum creatinine and calcemia employed mixed effects regression using time as a fixed-effects and patient as a random-effects variable. Creatinine was logarithmized to meet the assumption of normal distribution of the residuals.

Results

Patient Disposition

In total 142 patients were enrolled between February 10, 2006, and the census date for analysis of October 27, 2010. Of these, 98 (69.0 %) had follow-up data at 32 or more months compared to 44 (31.0 %) patients who were discontinued for various reasons and/or had no visit at 32+ months (Fig. 1). At the time of census patients had been followed for a mean of 37.3 ± 2.6 months and had been seen between two and nine times following treatment. Most patients (58.2 %) were followed by the physician who prescribed zoledronic acid versus external follow-up by general practitioners (18.4 %), rheumatologists (13.3 %), orthopedic surgeons or internists (3.1 % each), and endocrinologists (1.0 %). Three (3.1 %) patients were followed by other clinicians.

Fig. 1
figure 1

Patient disposition at census date

Full size image

Demographics and Clinical Status at Enrollment

Evaluable patients had a mean age of 74.2 years, and the majority (84.7 %) were age 65 or older (Table 1). Most (60.2 %) were male, and all were Caucasian. Median body mass index was 25.8 kg/m2. In 67.4 % of patients PDB had been diagnosed 5 or more years prior to zoledronic acid treatment (median = 11.5 years). Pagetic sites were most frequently located in the spine (82.6 % of patients), pelvis (63.3 %), femur (40.8 %), tibia (31.6 %), and skull (22.5 %). The most common complaints at enrollment were pain (82.7 % of patients), impaired mobility (51.0 %), bone deformities (34.7 %), joint disease (24.5 %), and pathological fractures (15.3 %). Some patients presented with comorbid osteoarthritis (36.7 %) or osteoporosis (16.3 %). About one-third (32.6 %) of patients had been treated previously with various other bisphosphonates and calcitonin and/or had received supportive therapy with calcium, vitamin D, and anti-inflammatory or analgesic agents.

Table 1 Patient demographics and clinical status at enrollment
Full size table

Compared to patients randomized to the zoledronic acid arm in the original trials [8] and in the extended observational study [9, 10], in our registry subjects were older (p = 0.006 and 0.002, respectively) and fewer of them had been treated previously with other bisphosphonates (p = 0.011 and 0.018, respectively). There were no significant differences in mean sALP at the time of infusion between the three samples. Compared to the original trial subjects [8], patients in our registry tended to weigh less (p = 0.031) but gender distributions between the two were statistically similar.

Bisphosphonate Exposure and Concomitant Treatment

All evaluable patients received one dose of 5 mg zoledronic acid administered intravenously over 15 min according to their treating physician’s decision. The most common reasons to treat with zoledronic acid were elevated ALP (82.7 % of patients), pain (78.6 %), and/or findings from imaging studies (52.0 %) (Table 2). In the first year following initial treatment with zoledronic acid, one patient was retreated with zoledronic acid and one with pamidronate. By the end of year 3, 11 patients had been retreated, including nine with zoledronic acid, one with pamidronate, and one with zoledronic acid complemented with pamidronate.

Table 2 Physicians’ reported reasons to treat patients with 5 mg zoledronic acid
Full size table

Nearly all patients were prescribed calcium (95.9 %) and vitamin D (96.9 %) for at least 10 days after zoledronic acid treatment. One patient was taking NSAIDs, and one was taking analgesics at inclusion and continued thereafter.

Remission

Per treating physicians’ reports, 93.3 % of patients were in remission at 1 year, 89.5 % at 2 years, and 91.6 % at 3 years, rates similar to those reported for the same time points in the extended observation study (all p values nonsignificant) [10]. Of the 87 patients in remission at 3 years, 20 were newly diagnosed patients, whereas 67 had been diagnosed at an earlier time point. One of the eight patients not in remission at 3 years was newly diagnosed. There was no association between the 3-year remission rate and being newly diagnosed with PDB at the time of infusion or with previous bisphosphonate treatment.

Relapse and Retreatment Outcomes

As shown in the time plot in Fig. 2 and the flowchart in Fig. 3, 21 patients (21.4 %) showed a relapse at some point during the follow-up period. Of these, 13 (1.26, 10.8, 14.1, 15.2, 16.1, 16.4, 25.5, 25.6, 3.12, 4.1, 5.12, 6.3, 9.1) regained their remission status by the 3-year time point, including five patients (10.8, 14.1, 16.1, 25.5, 25.6) who evidenced a slow treatment response. Eight of the 21 relapsing patients (1.15, 25.1, 3.13, 3.16, 3.18, 3.27, 3.8, 9.2) did not show treatment response at the end of the observation period. Of these, 5 patients showed a treatment response during the observation period but not at the 3-year time point (1.15, 25.1, 3.13, 3.18, 3.8), whereas three patients (3.16, 3.27, 9.2) showed (virtually) no response to treatment with zoledronic acid over the study period. Five of the eight patients who did not show a treatment response at study end had been treated with other bisphosphonates prior to the zoledronic acid infusion.

Fig. 2
figure 2

Management of patients with no treatment response and/or relapse. The x-axis represents month of follow-up after treatment with zoledronic acid. The time line for each of the 21 patients who showed relapse at some point during follow-up is presented horizontally. Circle visit on which patient was in remission, square visit on which patient was not in remission, asterick visit on which patient was retreated with zoledronic acid, greater than visit on which patient was retreated with pamidronate. Patients 3.16, 3.27, and 9.2 may be considered nonresponders and patients 10.8, 14.1, 16.1, 25.5, and 25.6 slow responders to treatment with zoledronic acid

Full size image
Fig. 3
figure 3

Flowchart of the 21 patients experiencing relapse

Full size image

We examined the reported sALP recordings for the three patients who did not respond to zoledronic acid treatment (all data as available). Patient 3.16 had sALP values of 105 U/L at enrollment, 123 U/L at 114 days, 123 U/L at 218 days, 108 U/L at 521 days, and 98 at 703 days. Patient 3.27 showed sALP values of 175 U/L at enrollment, 231 U/L at 338 days, 161 U/L at 422 days, 190 U/L at 800 days, and 219 U/L at 1,164 days. Patient 9.2 had sALP values of 112 U/L at enrollment, 80 U/L at 376 days, 73 U/L at 558 days, 76 U/L at 649 days, and 98 at 703 days. All three patients reported pain at one or more time points.

We also assessed the sALP results for the 18 patients with an initial response that was not sustained at the 3-year time point. Their sALP values were not statistically different at visits with relapse compared to visits in remission (p = 0.10).

Relapses occurred anywhere from 1.0 to 38.2 months postinfusion, with a median of 20.7 months (Table 3). The mean and median relapse times for patients with and without osteoarthritis, osteoporosis, or other comorbidities were not statistically significant (all p values nonsignificant). Proportional hazards regression modeling did not retain these conditions as independent predictors of time to first relapse.

Table 3 Time to first relapse (months) following treatment with 5 mg zoledronic acid
Full size table

Safety

A total of 180 AEs were reported in 79 of the 142 (55.6 %) patients treated with zoledronic acid, for an AE rate per patient year of observation of 0.06. In 23 (29.1 %) of these patients at least one AE was suspected to be related to zoledronic acid. Treatment with zoledronic acid was the suspected cause of 3 of the 38 musculoskeletal and connective tissue AEs, 0 of the 23 cardiovascular AEs, and 12 of the 19 flu-like symptoms and fever AEs. Eight patients experienced fractures, but only one case was assumed to be associated with treatment. Three cases of atrial fibrillation and two cases each of arrhythmia and palpitations were reported, none suspected of being related to zoledronic acid. No cases of osteonecrosis of the jaw were reported.

In total 38 serious AEs (SAEs), including 9 fatal, were reported in 31 patients, for an SAE rate per patient year of observation of 0.02. None of the fatal and only two of the nonfatal SAEs were suspected of being treatment-related. The first patient was an 85-year-old male diagnosed with PDB 2 years prior and admitted to the hospital the day following treatment with zoledronic acid. The most common AEs included general malaise, fever, fatigue, abdominal pain, and dyspnea. Laboratory assessments showed increased levels of creatinine kinase, C-reactive protein, and lactate dehydrogenase, as well as hypocalcemia and hypophosphatemia. Hypocalcemia was associated with subobstruction (abdominal pain), which resolved spontaneously. The second patient was a 65-year-old female with PDB diagnosed 12 years prior, known osteoporosis, a family history of PDB, and prior treatment with pamidronate. She was hospitalized about 3 months following treatment with zoledronic acid with a spontaneous right hip fracture. Though PDB was a possible etiology, the fracture was suspected to be related to treatment with zoledronic acid because of its nontraumatic character, its unusual subtrochanteric transversal (horizontal) shape, and, per the surgeon’s report, the unusually hard fractured bone. Both patients remained in remission.

Serum creatinine levels declined by 1 % per year (p = 0.02). Calcemia values remained stable over the observation period (p = 0.64).

Discussion

Prior to the advent of long-acting intravenous zoledronic acid, oral bisphosphonates taken daily for 2–6 months were the standard of care for PDB. However, gastrointestinal side effects, poor absorption, potentially impaired bioavailability, and the need to fast before and after intake have been associated with poor adherence and subsequent AEs among patients with osteoporosis [12–14], findings that, arguably, can be extended to patients with PDB. Further, pamidronate requires multiple serial infusions, each lasting several hours. Zoledronic acid represents a marked advance in therapeutic benefit and patient convenience owing to a convenient single 15-min IV administration every 12 months; the drug’s high affinity for bone mineral and its potent inhibition of the enzyme farnesyl diphosphate synthase [5, 6]; the qualitative difference in terms of stability of biochemical markers of bone turnover [10]; and, as now documented in two previous reports [9, 10] and our findings, long-term remission of PDB.

To our knowledge, apart from a small case series at one center extending 12–18 months postinfusion [15, 16], our registry offers the first adequately powered long-term multicenter data on remission and relapse in routine (“real-world”) clinical practice. The findings confirm the remission rates at 1, 2, and 3 years reported by Reid and colleagues [10]. Further, patients in our registry were on average 3.4 and 4 years older compared to the samples of, respectively, the original trials [8] and the extended study subsample [10]. Combined with weighing on average 4.2 kg less at the time of zoledronic acid treatment [8], our sample represents a frailer population than those studied to date.

Being an observational study, there were no mandated tests or other assessments and only data from routine clinical practice were used. From a scientific point of view, asking physicians for their clinical evaluation of PDB remission and relapse may have made for a less rigorous assessment method compared to prior longitudinal reports but is consistent with routine clinical practice [9, 10]. In Belgium zoledronic acid treatment for PDB is reimbursed only if the prescribing physician provides written certification as to active clinical symptoms or radiological evidence, along with sALP levels. Hence, physicians have to consider objective data in their clinical evaluation and reimbursement certification. Most likely, this information was the basis for their data entries on the registry. Further, despite the potential for opportunistic ratings of treatment success, such assessments better reflect routine clinical practice.

Our findings confirm the relapse rates reported by Reid et al. [10] yet offer some new insights as well. Unlike Reid et al., we found no association between relapse and prior bisphosphonate use or being newly versus previously diagnosed. Comorbid osteoarthritis or osteoporosis was not predictive of relapse. That 13 of the patients experiencing relapse regained remission by the end of the observational period underscores the importance of retreatment with zoledronic acid, including in patients showing either no early treatment response or a relatively early relapse (see Fig. 2). Of the eight patients failing to show remission over the course of the study, five had been previously treated with bisphosphonates. Possible treatment failure, treatment resistance, or slow treatment response should be evaluated in future studies as has already been shown for pamidronate [17].

On the safety side, the AE and SAE rates per patient year of observation in our sample were statistically similar to those reported in the extended observation study [10]. Flu-like symptoms and fever are known side effects of zoledronic acid treatment. The frequency of musculoskeletal and connective tissue AEs may have been confounded by comorbid osteoarthritis. None of the cardiovascular AEs were attributed to treatment with zoledronic acid and may instead be due to the advanced age of patients. Only one fracture was suspected of being related to treatment. No cases of osteonecrosis of the jaw were reported. There was no evidence of toxicity to a major organ system.

Our analyses are limited to 98 patients from one country. Future observational studies might benefit from a multicenter and multicountry design to increase sample size, mirror regional variation in the epidemiology of PDB, and evaluate regional differences in remission. This will also enable further assessment of possible age differentials in remission; any role of elapsed time between PDB diagnosis and treatment with zoledronic acid; variation in patients and treatments and associated variations in treatment outcomes; possible treatment resistance to bisphosphonates; the effectiveness of supportive therapy with nonsteroidal anti-inflammatory or analgesic agents [18]; and, per data availability, formal validation for the accuracy of physician assessments of remission and objective markers of response. Larger sample sizes will also increase the precision of AE and SAE rates and the adjudication of their relationship to zoledronic acid.

In conclusion, in this somewhat frailer sample of patients with PDB, effectiveness and safety data were similar to those observed in an open extended study of the original trial populations. These findings, which are the first on the long-term use of zoledronic acid for PDB in routine clinical practice, underscore the therapeutic benefits and relative safety of zoledronic acid in the management of PDB in “real-world” settings.