The clinical and biochemical characteristics of the study population are shown in Table 1. The average age at the time of assessment was 74 years but participants had been first diagnosed with PDB approximately 10 years previously on average. There were a higher proportion of males than females and 8.9% had a family history of PDB. Many individuals had complications of PDB, including bone deformity (30.4%), limb shortening (11.9%), previous pathological fractures (7.1%), and deafness with skull involvement (2.9%). Musculoskeletal pain was present in 122/168 (72.6%) of individuals and osteoarthritis was present in 111/168 (66.1%). In 46 participants, the OA was at a site neighbouring affected bone and in 65 osteoarthritis was at a site distant from Pagetic bone. The most common sites of osteoarthritis were the lumbar spine (26.1%), hands (25.0%), feet (25.0%), hips (20.2%), cervical spine (13.6%), knees (13%), shoulders (12.5%), and thoracic spine (7.7%). One individual (0.6%) had a history of osteosarcoma.

The pattern of skeletal involvement was typical for PDB. In total, 107/168 (63.6%) had monostotic disease and the median number of bones involved was 1, with a range of 1-10. The commonest involved sites were the pelvis (56%), the lumbar spine (20.8%, the femur (20.2%) the skull (12.5%), the tibia (11.3%), the thoracic spine (11.3%), the humerus (5.4%), and the scapula (3.6%). Other sites included the ribs, sternum, clavicle, radius, mandible, maxilla, ulna, patella, and sacrum (13.7%). Just over half of the individuals had previously received bisphosphonates for PDB.

The average estimated creatinine clearance was 76.4 mL/min at the time of enrolment. Serum total ALP was increased above the reference range in 40/163 (24.5%) of individuals, PINP was increased in 40/164 (23.8%), and BAP was increased in 25/164 (15.2%). In contrast CTX was increased in only 4/164 (2.4%). Circulating concentrations of IL-6 were above the reference range in 8/165 (4.8%) of patients. In contrast, the mean circulating concentration of M-CSF was increased above the reference range in 38/165 (23.0%) of individuals. There was no significant difference in circulating concentrations of IL-6 in people who had previously been treated with bisphosphonates and those who had not. Mean ± SD values for IL-6 were 3.36 ± 12.2 pg/ml vs. 3.05 ± 5.6 pg/ml, p=0.50) and the same was true for M-CSF concentrations (421.1 ± 264.2 pg/ml vs. 435.6 ± 315.8 pg/ml, p=0.45)

The causes of pain in individual participants as assessed clinically are shown in Fig. 1. The most common cause was osteoarthritis at a site distant from affected bone occurring in 54/122 (44.1%), followed by metabolically active PDB in 18/122 (14.7%); bone deformity associated with PDB in 14 (11.4%); and osteoarthritis of joints neighbouring an affected bone in 11 (9.0%). Other causes were neuropathic pain in 10 (8.2%), fibromyalgia in 3 (2.4%), and a wide variety of other causes in the remainder 34 (27.8%), including rotator cuff syndrome, plantar fasciitis, myositis, tendonitis, sciatica or nerve root pain, fractures, recent injuries, and pain following orthopaedic surgery. In 4 individuals, the cause of pain was unknown. In 83/122 individuals, (68.0%) a single cause of pain was identified; in 37 (30.3%) two causes were identified and in 2 individuals, 3 causes were identified.

Causes of pain in the study population. The percentages of individuals with different causes of pain are shown. The percentages add up to more than 100% as several individuals had more than one cause of pain. See text for a breakdown of the individual causes of pain in the “other causes” group

Ninety-two individuals had previously been treated with bisphosphonate therapy but data on response to treatment were available for only 90 of these individuals. The patient-reported response of pain to previous bisphosphonate therapy is summarised in Fig. 2. The most frequently used bisphosphonate was intravenous zoledronic acid 5 mg by infusion (n=82) followed by intravenous pamidronate 60 mg by infusion on between one and three occasions (n=9) and oral risedronate 30 mg orally for 2 months (n=7). Eighty-one individuals had received a single bisphosphonate; 5 had received two different bisphosphonates and 3 had received three bisphosphates. Of the 7 treated with risedronate, 1 (14%) reported that the pain had improved a lot; 3 (43%) that the pain had improved a little; and 3 (43%) reported that it had not changed. For pamidronate, 1 individual (11%) reported the pain had disappeared, 2 (22%) reported it had improved a lot, 2 (22%) reported that it had improved a little, 4 (44%) that it had not changed, and 1 (11%) that it had worsened. For zoledronic acid, 14 individuals (17%) reported the pain had disappeared, 25 (30%) reported it had improved a lot, 15 (18%) reported that it had improved a little, and 21 (26%) that it had not changed. The remaining 7 participants treated with zoledronic acid had not experienced pain before receiving treatment. When data from all bisphosphonates were combined, pain disappeared in 15 (16%), improved a lot in 28 (31%), improved a little in 20 (22%), did not change in 28 (31%), and worsened in 1 (1%). There was no difference in the magnitude of pain response to previous bisphosphonate therapy in the groups of individuals with and without pain overall. However, in an exploratory analysis, we found that 7/16 individuals (44%) who reported that the pain has previously disappeared in response to bisphosphonate therapy were in the no pain group compared with 7/44 (16%) in the current pain group (p=0.038, Fisher’s exact test).

Response of pain to previous bisphosphonate therapy. The values are the proportion of patients who reported that, following bisphosphonate treatment, their pain disappeared, improved a lot, improved a little, did not change, or worsened. The number of individuals in each group and shown on the x-axis

The demographics, clinical characteristics, and biomarkers in those with and without pain are shown in Table 2. Factors significantly associated with the presence of pain at the level of p<0.05 or below were comparing groups of individuals with and without pain were as follows: female sex, increased age, bone deformity, body mass index, the presence of osteoarthritis, and serum M-CSF concentrations. There was no significant association between any of the biochemical markers of bone turnover and the presence and absence of musculoskeletal pain or between IL-6 concentrations and presence or absence of musculoskeletal pain in the study group as a whole. Since many participants had been previously treated with bisphosphonates, we conducted an exploratory subgroup analysis to determine if biochemical markers were associated with pain in participants who had not previously been treated with bisphosphonates. As expected, the circulating concentrations of all markers were higher in this subgroup than in the whole study group, but there was no significant difference in any of the markers in those with and without pain. These results are summarised in supplementary Table 1.

Logistic regression analysis was carried out to determine which variables were independent predictors of the presence of pain and of these, only osteoarthritis remained an independent predictor of pain in the study population (p=0.019; beta = 0.986, S.E. 0.420, Wald statistic 5.51)

When we subdivided participants into groups who had or had not reported pain at the time of study, all subdomains of SF36 were significantly lower in those with pain as compared with those who did not have pain (Table 3).