Introduction

Paget’s disease of bone (PDB) is very common in elderly caucasian populations in Europe, the United States and Australia; however, its rate of incidence in Asia (including Japan) is very low [1]. Increased morbidities pertaining to skeletal and extraskeletal clinical manifestations in PDB patients have been observed not only in western countries but also in Japan [1, 2].

Bisphosphonates are potent therapies for PBD and comparative trials have shown the relative efficacy of bisphosphonates [3–5]. Trials have demonstrated that risedronate is superior to etidronate [6] and zoledronate appears to be superior to risedronate [5]. Oral risedronate treatment (17.5 mg daily for 8 weeks) was efficacious in Japanese PDB patients [7] and this regimen was approved for the treatment of PDB patients in Japan. Long-term efficacy of the regimen was also demonstrated by reduction of serum alkaline phosphatase (ALP) levels [8].

Although the optimal levels of reduction for metabolic bone markers are not fully established in PDB patients, improvement in biochemical markers provides the most rapid indication of treatment effect, and normalization of metabolic bone marker levels has been proposed as the therapeutic goal in PDB treatment [9]. Indeed, measurement of metabolic bone markers (e.g., serum ALP level) is integral to diagnosis and management of PDB [10]. Typically, treatment effects are evaluated based on the duration of serum ALP level reduction.

Serum ALP levels showed the greatest decline compared to those observed for other metabolic bone markers in the treatment of PDB using bisphosphonates [10–12]; ALP is the most reliable marker for the evaluation of treatment efficacy [13, 14]. Other osteoblastic markers such as bone-specific ALP (BAP) and type I procollagen N-terminal pro-peptide (PINP) levels are as suitable as serum ALP levels are for monitoring the efficacy of bisphosphonate therapy for PDB [10]. BAP levels [15] or levels of bone resorption markers such as N-treminal crosslinking telopeptide of type I collagen (NTX) [16, 17] and C-treminal crosslinking telopeptide of type I collagen (CTX) [18, 19] were suggested to be superior markers as compared to serum ALP levels for PBD treatment.

We attempt to determine in this study the efficacy of oral risedronate treatment of 17.5 mg daily for 8 weeks in clinical settings and to evaluate the performance of currently available metabolic bone markers in assessing PDB therapy response.

Materials and methods

Patients

PDB was diagnosed according to the standard clinical criteria of the Japanese Osteoporosis Society [20]. Of the 21 PDB patients who visited our hospital for the first medical examination between 1994 and 2011, those whose ALP levels exceeded the upper limit of the normal range were treated with oral risedronate treatment of 17.5 mg daily for 8 weeks, aiming for normalization of the ALP levels. Retrospective analysis was carried out in 11 patients (6 males and 5 females) who could be traced for observation over 12 months or longer after the initial treatment, provided that risedronate or other bisphosphonates were not administered within 12 months after the initial oral risedronate treatments. None of the patients had liver disease, bone cancer metastasis, fractures or other conditions that would influence ALP levels. The affected areas of PDB were confirmed by bone scan and X-ray and classified as monostotic and polyostotic types. Of the 11 patients, 1 subject was treated at Hokkaido University Hospital. Two were enrolled in a Phase III study for oral risedronate treatment of 17.5 mg daily for 8 weeks [7] and their data were accumulated and analyzed upon completion of the clinical trial.

Treatment

The responder group included the patients whose ALP levels were normalized within 12 months after initiation of oral risedronate treatment and remained within the normal range for 12 months after initiation. The non-responder group included patients whose ALP levels were not normalized within 12 months after treatment initiation. Re-treatment was performed if bone pain recurred or if serum ALP levels increased at least 25% above the nadir, according to a published guideline [9]. Three of five non-responders were re-treated with oral risedronate but their serum ALP levels were not normalized within 12 months after initiation. Two of these three re-treated patients were treated one more time but their serum ALP levels were still not normalized within 12 months. So, five re-treatments were performed on three non-responder patients.

Serum specimens from three responders, four non-responders and three re-treated patients (five re-treatments) were collected and stored. Stored serum was used to measure metabolic bone markers. This study was approved by the Institutional Ethics Committee of Osaka City University Graduate School of Medicine and the Hokkaido University School of Medicine. Written informed consent was submitted by the subjects, in compliance with the Helsinki Declaration.

Statistics

Values are presented as mean ±SD. Time-course ALP changes were analyzed by a one-way repeated measures analysis of variance (ANOVA) in 11 subjects treated for the first time with oral risedronate at 17.5 mg daily for 8 weeks. A Mann–Whitney U test and Fisher’s exact test were used in comparing the responder and non-responder groups. Time-course changes of ALP in the responder and non-responder groups were compared by two-way repeated measures ANOVA. Comparison with baseline values was performed via a Dunnett’s test. The level of significance was set at 0.05. Statistical analysis was conducted by JMP software version 8.0.2.

Results

Oral risedronate treatment of 17.5 mg daily for 8 weeks resulted in serum ALP levels significantly decreasing over time (F value: 2.807; P value: 0.017) (Table 1). Further, an increasing number of subjects experienced ALP level normalization with the passage of time. At 12 months six of 11 subjects showed ALP normalization (responder group), while five did not (non-responder group).

Table 1 Serum ALP levels in 11 patients receiving oral risedronate treatment (17.5 mg daily for 8 weeks)
Full size table

No statistically significant differences existed in age or the male/female ratio between the responder and non-responder groups (Table 2). While the ratio of subjects with monostotic lesions was high in the responder group, there was no significant difference (P = 0.175, by Fisher’s exact test). While many non-responders received antecedent treatment with bisphosphonate, the number was not significantly different (P = 0.575, by Fisher’s exact test). A significant number of non-responders had lesions in the skull (P = 0.045, by Fisher’s exact test). Baseline ALP levels were high in the non-responder group, though it was not significant (P = 0.359, by Mann–Whitney U test). Serum calcium levels of the non-responder group were significantly high (P = 0.014, by Mann–Whitney U test).

Table 2 Clinical characteristics of PDB patients treated with oral risedronate (17.5 mg daily for 8 weeks)
Full size table

In comparing the ALP time-course between the responder and non-responder groups, intergroup factor (F value = 7.304, P = 0.024) and time factor (F value = 10.637, P < 0.001) were significant, while interactions between the intergroup factor and time factor were not significant (F value = 0.320, P = 0.923) (Fig. 1). ALP levels of the responder group significantly decreased after 1 month of treatment (by Dunnett’s test) when compared to pre-treatment levels, while ALP levels of the non-responder group tended to fall, though not significantly.

Fig. 1
figure 1

Comparison of time-course changes in ALP levels between responder and non-responder groups. The intergroup factor (F value = 5.887, P = 0.041) and time factor (F value = 8.695, P < 0.001) were significant, while interactions between the intergroup and time factors were not significant (F value = 0.230, P = 0.965) (by 2-way repeated measures ANOVA). *P < 0.05 vs 0 month by Dunnett’s test

Full size image

The effect of antecedent treatments of bisphosphonates on the time course changes in ALP levels was also examined. The time factor was concluded to be significant (F value = 13.287, P < 0.001), while the intergroup factor (F value = 1.918, P = 0.200) and interactions between intergroup and time factors were not significant (F value = 0.580, P = 0.744) (Fig. 2). ALP levels of the non-antecedent treatment group decreased significantly after 1 month of treatment (by Dunnett’s test) as compared to pre-treatment levels, while ALP levels of the antecedent treatment group tended to fall, though not significant statistically.

Fig. 2
figure 2

Comparison of time course changes in ALP levels resulting from antecedent treatments of bisphosphonates. The time factor (F value = 13.287, P < 0.001) was significant, while the intergroup factor (F value = 1.918, P = 0.200) and interactions between intergroup and time factors were not (F value = 0.580, P = 0.744) (by 2-way repeated measures ANOVA). *P < 0.05 vs 0 month by Dunnett’s test

Full size image

Correlations between baseline metabolic bone markers were examined for the subjects whose serum specimens could be stored. ALP levels showed significant positive correlations with P1NP, BAP and BGP, but no correlation with CTX (Table 3). The correlation coefficient to P1NP was the highest (R = 0.860).

Table 3 Correlations between metabolic bone markers at baseline
Full size table

We examined the treatment-associated evolution of metabolic bone markers using the stored serum. The responder group exhibited a significant decrease of P1NP after 1 month, ALP after 2 months, BAP after 6 months and BGP after 9 months (Fig. 3). While CTX significantly decreased at 1 and 2 months after treatment, significant differences disappeared thereafter. None of the metabolic bone markers showed significant changes in the non-responder and re-treatment groups.

Fig. 3
figure 3

Changes in metabolic bone markers during the treatment. P1NP, BAP, BGP and CTX were measured in the stored serum obtained from 3 responders, 4 non-responders and re-treated subjects (5 treatments). *P < 0.05 vs 0 month by Dunnett’s test

Full size image

Correlations between ALP levels and other metabolic bone markers were examined for the changes from baseline to 1, 2, 3 and 6 months after treatment: significant correlations were found between ∆-ALP and ∆-P1NP/∆-BAP during the 1–6 month period, though correlation with ∆-CTX was found only at 1 and 2 months, and no correlation to ∆-BGP was found at all (Table 4).

Table 4 Correlations of change markers to ∆-ALP resulting from risedronate treatment
Full size table

As for other serum data, no significant changes were observed except transient drops of calcium and phosphate levels (Table 5). Elevation of w-PTH was not observed, suggesting that development of the secondary hyperparathyroidism as a result of treatment did not occur. No subjects discontinued the treatment as a result of gastrointestinal symptoms (data not shown).

Table 5 Changes in serum parameters from oral risedronate treatment
Full size table

Discussion

In this study the efficacy of the first oral risedronate treatment of 17.5 mg daily for 8 weeks was observed in 6 of 11 PDB patients (55%) with ALP levels higher than the upper limit of the normal range. A Phase III study was conducted in patients whose ALP levels exceeded twice the upper limit of the normal range (serum ALP excess 1093.9 ±588.0 U/L), and ALP levels were normalized by treatment in 9 of 11 subjects (82%) [7]. In a post-marketing study of 17 patients (ALP: 799 ±877 IU/L) in the central part of Japan, 10 (59%) did not require re-treatment with bisphosphonate during the post-treatment for an average of 27 months [8]. Considering these data, the number of non-responders in the present study seemed higher than those in the previous Japanese studies [7, 8].

Pre-treatment ALP concentrations were reported to be useful in predicting the effect of bisphosphonate treatment [15, 21]. In the present study ALP levels in the non-responder group were high, though there was no statistically significant difference when compared to the responder group. Antecedent bisphosphonate treatment was known to lower ALP’s responsiveness to bisphosphonates [22, 23]. Although no difference existed in the frequency of antecedent bisphosphonate treatment between the responder and non-responder groups, a significant reduction of ALP levels was observed only in the non-antecedent bisphosphonate treatment group, suggesting antecedent bisphosphonate treatment is one of the factors for the resistance to the oral risedronate treatment.

Meanwhile, the non-responder group had a significantly higher prevalence of skull lesions and higher baseline serum calcium concentrations as compared to the responder group. Patients with skull lesions were known to have higher baseline ALP levels [24–26] and lower responsiveness to bisphosphonate treatment [27]. Hypercalcemia may result from a combination of increased bone turnover and immobilization [13]. As no subjects of the present study were immobilized, hypercalcemia and resistance to bisphosphonate treatment may have been the result of increased bone turnover. The factors that influence the therapeutic effect of oral risedronate treatment of 17.5 mg daily for 8 weeks should be further studied in a larger group of patients.

Examination of the stored serum revealed that baseline ALP levels best correlated to P1NP, followed by BAP and BGP. After oral risedronate treatment of 17.5 mg daily for 8 weeks the P1NP levels of the responder group was decreased at 1 month and a decrese in ALP and BAP levels followed. BGP was lowered quite late in the study. The post-treatment amount of change in various metabolic bone markers was examined in terms of their correlations to the amount of change in ALP. The degree of P1NP change was best correlated to the amount of ALP change at any time during 1–6 months after treatment. During the proliferation period of osteoblast differentiation, type I collagen is biosynthesized in the extracellular bone matrix, and P1NP is produced during the maturing process of type I collagen [28]. ALP levels increase during the maturation phase of the matrix after the end of the proliferation period. BGP is secreted during the next mineralization period and calcium is deposited onto the bone matrix, initiating calcification. Therefore, P1NP is an excellent marker for the early stage of osteoblast differentiation. Meanwhile, when osteoporosis is treated with teriparatide the amount of increase of P1NP at 1 month is correlated to the increase rate of bone density of the lumbar spine at 12 months [29]. In the present study P1NP dropped most rapidly among various metabolic bone markers for assessing the therapeutic effect on PDB, suggesting that P1NP measurement may predict the therapeutic effect of treatment.

PDB is primarily caused by dysregulation of osteoclast differentiation and function [30]. Bone resorption markers are considered useful for assessing therapeutic effects as a result of bisphosphonate suppressing osteoclast activity. The usefulness of bone resorption markers has been demonstrated in treating PDB, such as urinary NTX for pamidronate treatment [17] and CTX in serum for risedronate treatment [19]. However, urinary pyridinoline (PYD) and deoxypyridinoline (DPD) have been reported to be inferior to ALP in ibandronate treatment [12]. CTX dropped early in our study, though this downward trend did not continue and, therefore, the usefulness of CTX was inferior to other bone formation markers.

While the bone formation markers (P1NP, BAP, and BGP) showed a downward trend in both the non-responder and re-treatment groups, no significant differences were observed. CTX showed a transient decline, but it was not significant. In the patients requiring re-treatment the efficacy of risedronate was anticipated to be inferior as compared to the responder group. Further, re-treatment was itself a factor for suppressing responsiveness to treatment [22, 23] and, therefore, inhibited the responsiveness of various bone formation markers.

Our study demonstrated the usefulness of oral risedronate treatment (17.5 mg daily for 8 weeks) for more than half of the Japanese PDB patients when serum ALP was used as a marker. Skull lesions, high values of serum calcium concentration and antecedent treatments of bisphosphonates were the factors for resistance to treatment. The present study suggests that P1NP would be useful in assessing the therapeutic effect on PDB.