Effect of Bisphosphonates on Function and Mobility Among Children With Osteogenesis Imperfecta: A Systematic Review

Osteogenesis imperfecta (OI) is a rare genetic connective tissue disorder that results in bone fragility and deformity. Management is multi‐disciplinary. Although pharmacologic intervention with bisphosphonates (BP) is a standard of care for individuals with severe OI, no consensus or reviews were found that focus on the effects of bisphosphonates on function and mobility. PubMed, CINAHL, Cochrane Library, Web of Science, and PEDro databases were searched for eligible articles for this review. Methodological quality was assessed using the Cochrane Collaboration's tool for risk of bias. Twenty‐six studies (801 children) were reviewed and five showed a low risk of bias. Included studies showed significant variability among clinical protocols for administering BP. Randomized controlled trials did not demonstrate a significant improvement in function and mobility with oral BP administration, while non‐randomized open‐label uncontrolled studies demonstrated that oral and intravenous BP administration objectively improved function and mobility. The most common outcome measure used by the studies included in this review was the Bleck score. Effect sizes (d = 0.28 ‐ 4.5) varied among studies. This systematic review also summarized the apparent confounding variables affecting results of previous studies and provided suggestions to improve the quality of future studies.


Introduction
O steogenesis imperfecta (OI) is a rare, clinically heterogenous genetic connective tissue disorder marked by low bone mass and increased bone fragility, resulting in increased susceptibility to fractures, deformities, and substantial growth abnormalities. (1) It has a reported incidence of 1 per 10,000 to 20,000 births. (1)(2)(3)(4)(5) Early genetic studies on OI documented that it is commonly caused by autosomal dominant heterozygous mutations in one of the two genes encoding type I collagen, COLIA1 and COLIA2, and recently, mutations in other genes were documented to be involved in its pathogenesis. (1)(2)(3)(4)(5) With variable clinical manifestations, OI was initially classified into four types based on severity of signs and symptoms using the Sillence classification system: types I and IV being mild and moderate, type II being lethal, and type III being severe and progressively deforming. (6,7) Recent studies have broadened the classification of OI into up to 19 types based on the genes involved. (5,(8)(9)(10) The classification system by Sillence is still being used in a modified fashion in current studies to stratify subject populations. (11)(12)(13)(14) As a consequence of the physical impairments brought about by OI, varying degrees of pain, (11,12) gait deviations, (13) and functional limitations (11)(12)(13)(14) have also been reported depending on the type and severity of OI. Lower mobility scores, (11)(12)(13) limited performance of activities of daily living, (12) and lower levels of participation in sports, exercise, or physical function (11)(12)(13) have been reported in individuals with OI. Despite limitations in functional activity, studies show individuals with OI are still able to participate and ambulate in the community but may show difficulty keeping up with typically developing peers. (12)(13)(14) Without a genetic cure for OI, management of the disease is aimed at symptom reduction through a multidisciplinary approach consisting of pharmacologic agents, orthopedic interventions, physiotherapy, and rehabilitation. (15,16) Among pharmacologic interventions, bisphosphonates (BP) have been considered standard of care for children with severe OI. (17)(18)(19)(20) BP can be administered orally or intravenously with varied efficacy, and there are two types, both acting on osteoclasts (cells that break down bone tissue) by disrupting their formation (nitrogenous type BP) or initiating their apoptosis (nonnitrogenous type BP). (19,21) Although BP therapy is widely used to treat OI, results on improvements to function and mobility outcomes have been variable. (22)(23)(24) Previously published systematic reviews and meta-analyses on BP focused on their effects on increasing bone mineral density and reducing fracture rate. The reports mention function and mobility outcomes, but these were considered secondary variables of interest. (16,(25)(26)(27) To the best of our knowledge, no consensus or systematic reviews have been published to quantitatively describe how BP therapy affects measures of function and mobility among individuals with OI. (12)(13)(14) Therefore, the purpose of the current work was to systematically review existing literature and describe the effects of BP therapy on improving measures of function and mobility.

Inclusion criteria
Included studies were limited to populations involving male and female children who have an established diagnosis of OI in which at least one of the outcomes was the effect of bisphosphonates on function and/or mobility using objective outcome measures. Studies that included quality-of-life (QOL) or well-being as outcomes were included if objective parameters of function and mobility were included in the measuring tools used in those studies. Randomized controlled trials (RCTs), non-randomized open-label uncontrolled studies (NROs), NROs with a historic control group, and retrospective studies were included. Non-randomized open-label uncontrolled studies are defined as studies that are not randomized, all subjects are given treatment (no control or placebo group), and both the researchers and subjects are aware of the treatment administered (no blinding).

Search strategy
We searched the following electronic databases: PubMed, CI-NAHL (Cumulative Index to Nursing and Allied Health Professions), Cochrane Library (Cochrane Database of Systematic Reviews, Cochrane Central Register of Controlled Trials), Web of Science, and PEDro (Physiotherapy Evidence Database). The initial search strategy was developed in PubMed, using a combination of Medical Subject Terms (MeSH) and key words. Once the initial search was determined, it was modified to fit the parameters of the other databases.

Selection of studies and data extraction
The two lead authors (CSC and JJK) screened the articles for eligibility independently. In the event of disagreement, a third reviewer (KMK) was included in the discussion until a consensus was reached. A standardized data extraction form was created during study selection and was used to compile relevant descriptive information and relevant data among the included studies.

Strategy for data synthesis
The reviewers provided a narrative synthesis of the findings from the included studies, structured around OI population type, route (oral or intravenous), dosage of BP administered, and outcomes used to measure function and mobility. Based on the anticipated heterogeneity of outcomes and data completeness, the reviewers provided a summary of intervention effects for each study ( Table 1). Effects of BP on measures of function and mobility were calculated with standardized mean differences on each outcome and effect size when applicable.

Risk of bias (methodological quality) assessment
The two primary reviewers (CSC and JJK) assessed the methodological quality (risk of bias) of the articles independently using the adapted version of the Cochrane Collaboration's tool. This tool reviews five domains, with 11 items. Each item was scored "1" if the item was present in the article, and "0" if it was not ( Table 2). Studies fulfilling six or more items were regarded as having a low risk of bias. (28,29) Disagreements were resolved by including three reviewers (KMK, JRR, and KR) in the discussion until a consensus was reached. The strength of interrater agreement was measured using Cohen's Ƙ coefficient (95% confidence interval), with Ƙ = 0.41 to 0.60 indicating moderate agreement, Ƙ = 0.61 to 0.80 indicating good agreement, and Ƙ ≥ 0.81 indicating very good agreement. (28,29) Analysis of subgroups or subsets Subgroup analyses were performed based on the route of administration (oral or intravenous) of BP, function and mobility outcome measure used, and population. Figure 1 illustrates the article selection process. The search strategy yielded 423 articles. Removing duplicates, the number was reduced to 210, and 173 were excluded based on title and abstract. Full texts of the 37 remaining articles were screened, and one was excluded because some participants were given growth hormone in addition to pamidronate, and the results obtained from these participants were not differentiated from those who received pamidronate only. (30) Two more were excluded because the main intervention was not bisphosphonate administration. (31,32) Six more were excluded because the outcomes used were not objective. (33)(34)(35)(36)(37)(38) Two more were excluded because they were found to be abstracts presented in proceedings. (39,40) A total of 26 full-text articles (801 male and female children) including four RCTs, (23,(41)(42)(43) 17 NROs, (44,45,47,48,(50)(51)(52)(53)(54)(55)57,58,(60)(61)(62)(63)(64) three NROs with a historic control group, (49,56,59) and two retrospective studies (46,65) met the inclusion criteria and were included. Population sizes (n) ranged from n = 4 (65) to n = 139. (42) Three NROs did not specify sex distribution. (48,51,63) OI types recruited by most of the studies were I, III, and IV. Other NROs included participants with types V, (64) VI, (57) VII, (65) and VIII. (53) Two NROs did not specify the OI type of their population. (50,60) Two studies included participants with unclassified types in addition to types I, III, and IV. (42,54) One of the retrospective studies recruited only females. (65) Bleck score (including its modified forms) (66,67) was the most common outcome tool used in 19 (73%) studies, followed by the pediatric disability inventory (PEDI), (68) used in six (23%) studies. and was 6 (range 2-9) after treatment (p = 0.001).

Results
(Continues) Bleck score During the study period, ambulation score improved in 4 patients with OI type V and remained unchanged in 1 patient. One patient was not assessed because of young age.
The other 5 patients were independent walkers (grade 4) before pamidronate treatment was started and remained so during the observation interval.
In the control group, 4 patients gained 1 to 3 grades, and in 2 patients no progress was noted.
The other 5 patients were independent walkers before and after 2 years of treatment.  Risk of bias (methodological quality) assessment Summary of scores of the adapted Cochrane Collaboration's tool is shown in Table 2. Interrater agreement was very good (Ƙ = 0.94). Five studies scored ≥6 (low risk of bias). (23,(41)(42)(43)59) Studies demonstrating a positive effect of bisphosphonates on function and mobility

Oral bisphosphonate administration
Two NROs administered oral alendronate on patients with OI types I, III, and IV. (52,62) Cho and colleagues (52) measured function and mobility in 16 patients using a scale the researchers devised themselves. This scale was a 9-point scale, with a highest possible score of 1 (able to sprint and participate in contact sports) and a lowest possible score of 9 (wheelchair-or bedridden; always requiring assistance from others, including self-care activities). Seven of their patients had surgical or implant-related problems that may have affected their ambulatory/mobility status. Of the remaining nine that did not have problems, five had improved scores on latest follow-up. Vyskocil and colleagues (62)
Comparison with a historic untreated population: In another study by Astrom and colleagues, (49) they compared Bleck and Wilson scores of patients who received intravenous pamidronate to scores of a historical untreated group. Better scores were found among patients who received treatment on both Bleck (3.0 ± 0.77 versus 0.36 ± 0.67 for controls; p < 0.001, d = 3.63) and Wilson (2.82 ± 2.09 versus 7.45 ± 1.51 for controls; p < 0.001, d = 2.55) scores. (49) In their 2006 study, Land and colleagues (56) compared historic data from 48 untreated patients and compared their scores to 48 age-matched patients who received treatment for 3 years. Their results showed significantly higher Bleck scores among the treatment group (2.1 ± 1.2) versus untreated (1.0 ± 1.2), p = 0.001. The mobility domain of PEDI was also significantly higher among the treatment group (76.3) versus untreated (58.3), p = 0.002. Munns and colleagues (59) showed increased Bleck scores among 58 treated patients compared with an age-matched historical untreated control group (2.3 ± 1.0 versus 0.8 ± 1.0; p < 0.001, d = 0.75).
Intravenous zoledronic acid administration: Garganta and colleagues (53) used both pamidronate (1.1 mg/kg/d every 3 months for patients 2 to 3 years old; 1.5 mg/kg/d every 4 months for patients aged >3 years old) and zoledronic acid (0.05 mg/kg/d every 6 months) to treat 22 children with OI types I, III, IV, and VIII and used the Pediatric Quality of Life Inventory 4.0 Generic Core Scales for Physical Functioning (PedsQL) (69) to measure physical function. The choice of BP was based on what regimen the patients were already taking at the start of the study. Not all patients in this study were able to complete study visits, and only 5 patients had data on all visits. These patients had significant changes in scores from the first visit to the postvisit with mean PedsQL scores of 49.48 ± 25.49 before infusion and 57.03 ± 25.29 after 4 weeks post-infusion (p = 0.007, d = 1.24). Physical function improved after the first infusion and diminished to pre-infusion levels during the time of the next infusion. Mean score was 46.88 ± 28.13 by the second infusion (p = 0.008, d = 0.51). There were no significant differences between patients treated with pamidronate and zoledronic acid with respect to age, sex, or OI type. Sanchez-Sanchez and colleagues (61) treated 14 children with OI types I, III, and IV with zoledronic acid 0.05 mg/kg administered over 1 to 2 hours every 6 months with varying treatment durations. The study used the modified Bleck score (9-point scale) and noted improved mean scores of 4 at pretreatment and 6 after treatment (p = 0.001).
Studies demonstrating no effect on function and mobility All RCTs used oral BP including olpadronate 10 mg/m 2 /d (23,41) and alendronate 5 mg/d (patients <40 kg) or 10 mg/d (patients >40 kg). (42,44) Kok and colleagues (41) used the ambulation domain of the Health-utility index-mark III scale (70) and athletic performance domain of the Harter self-perception profile for children (71) for participants >6 years old to measure function and mobility. Sakkers and colleagues (22) used both modified Bleck scores and PEDI. The remaining two RCTs used PEDI. (42,43) All RCTs reported no significant difference between treatment and placebo groups. In the 2006 study of Land and colleagues, (56) the self-care domain of PEDI (measure of function) was not significant between the two groups. Falk and colleagues (45) also noted inconclusive PEDI scores in their study.

Discussion
The purpose of the current work was to systematically review existing literature describing the effect of BP therapy on measures of function and mobility. The results of this current review show that children given intravenous BP have increased mobility, as measured by Bleck scores, after treatment. Results also showed that mobility scores were greater in patients with OI type I, III, and IV compared with other types. Moreover, based on the studies by Astrom and colleagues, (49) Land and colleagues, (56) and Munns and colleagues, (59) improvements are expected to be significantly greater than one would expect when compared with a historical untreated population.
All RCTs showed that oral BPs had no significant effect on function and mobility scores between treatment and placebo groups. In contrast, two NROs that used oral alendronate noted improvements in mobility. (52,62) Cho and colleagues (52) used an unpublished scale that the researchers devised themselves; hence their findings cannot be compared with those from other studies included in this review. Vyskocil and colleagues (62) used the Bleck score, which is the scale most of the studies in this review used, and it was also used by one of the RCTs (Sakkers and colleagues (23) ). Comparing these studies, population sizes were close (34 for Sakkers and colleagues; 30 for Vyskocil and colleagues), but these studies differed in terms of treatment time (Sakkers and colleagues administered for 2 years; Vyskocil and colleagues for 3 years), age group (Sakkers and colleagues = 10 ± 3.1 years old; Vyskocil and colleagues = 4 to 16 years old), and drug of choice (Sakkers and colleagues used olpadronate; Vyskocil and colleagues used alendronate). Between the two studies, Sakkers and colleagues has a lower risk of bias, scoring 11/11 in the Cochrane Collaboration's tool, whereas Vyskocil and colleagues only scored 4/11. All studies that administered BP intravenously were not randomized and uncontrolled, and if control groups were included, they were primarily limited to historical data. The paucity of RCTs in the available literature can be explained by the fact that intravenous BP administration became a standard of care for the management of severe OI before the RCTs were performed; thus, including a placebo or untreated control group in a study of severe OI is problematic. The three studies (49,56,59) that collected data from untreated groups obtained the data earlier than the data from the treatment groups, and intravenous BP may have not yet been used during that time. Relatively small sample sizes among the included studies reflect the rarity of OI. An improvement in mobility was consistent among all the NROs that administered intravenous BP and used Bleck scores (including its modified version) to measure mobility. (44,(47)(48)(49)(50)(51)(54)(55)(56)(57)(59)(60)(61)63,64) Adiyaman and colleagues (44) showed a very large effect size (d = 4.5), which can be explained by all his subjects starting with a score of 0 at baseline. In the study by Oztemur and colleagues, (46) the varying dosages and the small sample size reduced the power of the findings, but it is still important to note that all patients either improved or retained their baseline mobility scores and none had a decrease after treatment. Alharbi and colleagues (47) reported a moderate effect size (d = 0.78), which can be explained by a number of poor responders to treatment. The overall effect to mobility, however, was still an increase. The Wilson scale that Astrom and colleagues used in both their studies is unpublished and has not been verified or tested for reliability, but since both studies also used Bleck, findings from both studies also contribute to the overall improved mobility scores found in this review. (48,49) Moreover, it is consistent in all studies that treatment with i.v. BP does not decrease function or mobility scores. Another consistent finding in this review is that baseline and post-treatment function and mobility scores are higher among patients with types I, III, and IV compared with other types (types V, (64) VI, (57) VII (65) ) as measured by both Bleck (55,57,64) and PEDI (57,65) scores. Function and mobility scores measured by PEDI were inconsistent. Falk and colleagues (45) noted inconclusive results and Cheung and colleagues (65) noted non-significant increases in scores, while four other studies (56)(57)(58)(59) reported improvement after i.v. BP administration. In the study performed by Garganta and colleagues (53) using both i.v. pamidronate and zoledronic acid, increased function was noted with the use of the PedsQL scales, but because this is the only study that used this scale, no comparisons with other studies can be made. The retrospective review by Cheung and colleagues, (65) which had a sample size of 4 and only included females, was the only study to show non-significant increases in Bleck scores after i.v. BP administration.
Limitations of this review include a high number of studies (81%) with low methodologic quality. In addition, among the studies that administered BP intravenously, only one study had high methodologic quality. (59) This could be explained by the Cochrane collaboration tool, which was used to appraise the included studies regarding methodologic quality. As RCTs were included in this review, an appraisal tool specific to RCTs was employed. Because of this, any study that was not an RCT is discredited in five of the 11 categories in the assessment tool. Additionally, most of the studies included in this review scored low on the appraisal tool for not having comparable groups at baseline. This was not because they had different baseline groups, but because they had no control groups since these were single-group studies that tracked changes over time. These were longitudinal studies that lacked comparison.
Another point, which was expected because of the rarity of OI, has made the administration of BP, sample sizes, and followup periods very heterogeneous among the included studies. Also, because of the wide ranges of age groups and differences in demographic profiles and OI types, findings in the studies may not have enough power to draw strong correlations. Lastly, some studies included in this review may have been underpowered for outcome measures related to function and mobility. Most of the included studies were powered based on a primary outcome of bone mineral density. Therefore, we are uncertain if sufficient sample sizes were recruited in all studies to identify an effect of bisphosphonates on the chosen measures of function and mobility.
Hence, it is advised that our results be interpreted with caution because of these limitations. We suggest that future studies be conducted among comparable cohorts at baseline by ensuring participants with similar ages, OI types, and functional status at baseline to reduce confounding variables that may affect the outcome of these studies. Furthermore, these populations should receive consistent treatment regimens and durations of BP administration in order to draw stronger conclusions regarding the effectiveness of BP administration on improving function and mobility. Although intravenous BP is the standard of care and RCTs involving placebo controls are not possible, controlled studies that compare dosage regimens or combinations of BP treatment with other modalities like physical therapy are also suggested. It is suggested that better-powered studies on the effect of BP on function and mobility among children with OI be performed in the future.
This review, however, cannot answer which i.v. BP dosage best improved mobility because of the heterogeneous studies included, and future studies with more consistent dosages and treatment times with bigger and more homogeneous samples that measure mobility using the Bleck score are suggested. ◼ 14 CONSTANTINO ET AL.