Controlled Trial of Pamidronate in Children With Types III and IV Osteogenesis Imperfecta Confirms Vertebral Gains but Not Short-Term Functional Improvement



Bisphosphonates have been widely administered to children with OI based on observational trials. A randomized controlled trial of q3m intravenous pamidronate in children with types III and IV OI yielded positive vertebral changes in DXA and geometry after 1 year of treatment, but no further significant improvement during extended treatment. The treated group did not experience significantly decreased pain or long bone fractures or have increased motor function or muscle strength.

Introduction: Bisphosphonates, antiresorptive drugs for osteoporosis, are widely administered to children with osteogenesis imperfecta (OI). Uncontrolled pamidronate trials in OI reported increased BMD, vertebral coronal area, and mobility, and decreased pain. We conducted a randomized controlled trial of pamidronate in children with types III and IV OI.

Materials and Methods: This randomized trial included 18 children (4-13 years of age) with types III and IV OI. The first study year was controlled; 9 children received pamidronate (10 mg/m2/day IV for 3 days every 3 months). Four children in each group also received recombinant growth hormone (rGH) injections (0.06 mg/kg/day for 6 days/week). Seven children in the treatment group received pamidronate for an additional 6-21 months. All patients had L1-L4 DXA, spine QCT, spine radiographs, and musculoskeletal and functional testing.

Results: In the controlled phase, treated patients experienced a significant increase in L1-L4 DXA z score (p < 0.001) and increased L1-L4 midvertebral height (p = 0.014) and total vertebral area (p = 0.003) compared with controls. During extended treatment, DXA z scores and vertebral heights and areas did not increase significantly beyond the 12-month values. Fracture rate decreased significantly in the upper extremities (p = 0.04) but not the lower extremities (p = 0.09) during the first year of treatment. Gross motor function, muscle strength, and pain did not change significantly during the controlled or extended treatment phases.

Conclusions: A controlled trial confirmed the spine benefits of short-term pamidronate treatment in children with types III and IV OI. Pamidronate increased L1-L4 vertebral DXA and decreased vertebral compressions and upper extremity fractures. Vertebral measures did not improve during the extended treatment phase. The treatment group did not experience decreased lower extremity long bone fractures, significant improvement in growth, ambulation, muscle strength, or pain. There was substantial variability in individual response to treatment.


Osteogenesis imperfecta (OI) is a genetic disorder of extracellular matrix.(1) Affected individuals have osteoporosis and fracture from minimal trauma. Patients may also have short stature, bone deformity, joint laxity, impaired mobility, muscle weakness, and chronic pain.(2) Defects in type I collagen, the major structural component of the extracellular matrix of bone, cause most cases of OI.(1,3) Type I (mild)(4) OI is caused by quantitative defects of type I collagen; affected individuals secrete structurally normal collagen in reduced amounts and osteoporosis results from matrix insufficiency. Most cases of types II, III, and IV OI, the clinically lethal, severe, and moderate forms, respectively, are caused by structural defects in one of the type I collagen chains. As much as 50-75% of matrix collagen may be composed of mutant helices that actively contribute to a weakened bone structure, although a portion of the mutant collagen may be degraded in the cell. Thus, osteoporosis in type II, III, or IV OI results from a combination of matrix structural abnormality and matrix insufficiency. Recent literature has described three new variants of OI, based primarily on histologic distinctions within the Sillence type IV phenotype. Patients with type V OI have a triad of hypertrophic callus, dense metaphyseal bands, and calcification of the interosseous membrane. Their bone histology has a unique “mesh-like” lamellar pattern under polarized light.(5) The bone histology of type VI OI patients is characterized by the absence of the normal birefringent pattern of lamellar bone under polarized light and decreased mineral apposition rate and prolonged mineralization time on histomorphometry.(6) Type VII OI is a rare recessive disorder. Patients have severe OI plus prominent rhizomelia. Histomorphometry reveals decreased cortical width and trabecular number, with increased bone turnover.(7) Types V, VI, and VII OI are not caused by defects in type I collagen.

In recent years, antiresorptive drugs, specifically the aminobisphosphonates alendronate, pamidronate, and risedronate, have received much attention for medical management of OI.(8–11) Bisphosphonates have been shown to increase BMD and decrease bone turnover in postmenopausal and juvenile osteoporosis, Paget's disease, malignant osteolytic lesions, fibrous dysplasia, and McCune-Albright syndrome.(12–16) Treatment of OI with bisphosphonates is not expected to decrease the synthesis of mutant collagen or its incorporation into bone. However, the presence of increased cortical or trabecular bone, even with an abnormal composition, could positively affect bone mechanics. The ultimate mechanical effect on bone depends on matrix composition, quantity, and distribution of cortical and trabecular bone and bone geometry.

Uncontrolled trials of pamidronate in children with OI reported increased vertebral BMD and/or coronal area, decreased fracture rate and bone pain, and increased functional level.(17–27) Plotkin et al.(28) reported similar effects of pamidronate in nine infants with types III and IV OI. Iliac crest bone histomorphometry of children with OI showed increased cortical width and trabecular number after pamidronate infusions.(29)

Because of the abnormal bone matrix in OI, the relationship of BMD and bone functional strength is not clear, especially for long bones in which prolonged treatment with antiresorptives might lead to hyperdensity. Sakkers et al.(30) published results of a 2-year placebo-controlled trial of the oral bisphosphonate olpadronate in children with OI. Vertebral BMD increased, but no significant functional change was shown. The NICHD OI program initiated the first randomized controlled study of pamidronate for children with types III and IV OI. The goals were to determine the effect of pamidronate on vertebral BMD, strength, and geometry, lower extremity long bone fractures, and participant pain and motor performance.


Patient population and study design

For this randomized, controlled, nonblinded study of pamidronate in OI, children 4-16 years of age with types III and IV OI, who had not received any prior bisphosphonates, were recruited from the NICHD longitudinal OI population under an Institutional Review Board-approved protocol. Informed consent was obtained from parents or legal guardians; patients signed an informed assent if age-appropriate. Patients were assigned to treatment (nine children) or control (nine children) groups using randomly generated numbers. Four patients per group were co-enrolled in a recombinant growth hormone (rGH) treatment protocol. At entry into the study, all patients had vertebral compressions on lateral spine radiographs, but no patients had spinal instrumentation. All patients were seen quarterly at the NIH Clinical Center. Patients in the treatment group received intravenous pamidronate (Aredia; Novartis) 10 mg/m2/day for 3 days (∼1 mg/kg/cycle) every 3 months. Serum ionized calcium was analyzed the morning of each infusion. Patients <10 years of age received 500 mg/day calcium supplementation; patients >10 years of age received 1000 mg/day. After 1 year of treatment, the treatment group was shown to have significantly greater rate of increase in L1-L4 DXA z scores than the control group, which was the criterion for termination of the controlled portion of the trial. Seven patients in the treatment group were given an additional 6-21 months of intravenous pamidronate.

Serum markers of bone formation were measured at each visit. Bone-specific alkaline phosphatase (BSALP) was measured by IRMA (Esoterix, Calabasas, CA, USA), osteocalcin (OC) by homologous equilibrium RIA(31) (Dr Caren M Gundberg, Yale University, New Haven, CT, USA), and procollagen peptide type I (PICP) by RIA (Quest Diagnostics, Baltimore, MD, USA). Complete blood count, metabolic panel, 1,25(OH)D2 and vitamin 25(OH)D were measured at each visit.

Eight children received rGH injections during the course of the study. The study was carefully structured to allow the children to continue receiving their rGH injections, participate in the pamidronate trial, and yield valid comparison data. rGH recipients were evenly distributed between the pamidronate and no pamidronate groups, with four rGH recipients in each group. These children received 0.06 mg/kg/day for 6 days per week of Humatrope (generous gift of Eli Lilly and Co), which is our standard GH protocol regimen. Six of the eight children receiving pamidronate were determined to be responders to rGH before enrollment in the pamidronate protocol. The remaining two children began receiving rGH at the visit at which they were randomized into the pamidronate protocol.

Growth parameters were measured at each visit using the same equipment for each patient. Growth parameters included length and sitting height to the nearest 0.01 cm (average of five supine measurements on each side on a length board and three measurements on a stadiometer with chain, respectively), upper and lower segment measurements, and arm span. All measurements were obtained by nurses on an endocrine/genetics unit who are well versed in obtaining precise measurements for clinical trials. Pubertal status was assessed through Tanner staging at each visit.

BMD and geometry

Antero-posterior (AP) and lateral radiographs of the spine and lower extremity long bones and DXA at vertebrae L1-L4 were obtained at baseline and every 6 months. DXA measurements were obtained using a Hologic QDR 4500 densitometer and low-density software package (Hologic, Bedford, MA, USA). Measurements have a precision of 0.011 SD. All DXA scans were analyzed by one of us (JCR). Raw measurements were converted to z scores for analysis using the manufacturer's reference standards for age and pubertal status.

QCT scans of the spine were performed at the NIH Clinical Center at 0 and 12 months. QCT measurements were done using a spiral volumetric (3D) protocol with 2.5 mm section thickness, reconstructed at 1-mm increments. Midvertebral location of the axial sections was identified on sagittal and coronal images for each data set. Axial section thickness used for analysis was typically 3 mm, although it was reduced to 2 mm in several cases where endplate sclerosis was severe. QCT analysis was done by one of us (CEC) using QCT Pro software; precision of the 3D spine measurement is 0.8 mg/cm3.(32) Z scores were calculated using pediatric reference data matched for age and pubertal status.(33)

Vertebral area and compressions were measured on lateral spine radiographs by one of us (JCM). During measurements, the investigator was blinded to patient identifiers, the order of the films, and whether they came from the treatment or control groups. Compressions of T12-L4 were calculated as the ratio of central to anterior vertebral height (c/a ratio). NIH Image was used to calculate vertebral area from scanned tracings of lateral spine radiographs.

Upper and lower extremity fracture rates were assessed by parent/patient report, confirmed by X-ray, and compared with the 12 months before the study. Rib and other small bone fractures were not included because they do not bear the functional brunt of weight bearing and ambulation.

Assessment of function and activity level

All patients received standardized rehabilitation assessments and individualized physical therapy interventions at each visit. The Brief Assessment of Motor Function (BAMF) scale, a quantitative method of assessing gross motor function that has shown reliability and concurrent validity in children with a variety of neurological and musculoskeletal disorders, including OI,(34) was completed at each visit. Manual muscle testing (MMT) was completed at each visit by the same physical therapist and physiatrist. MMT involved evaluation of abdominal muscles and lower extremity muscle strength, including evaluation of bilateral straight leg raises, hip abduction, extension and flexion, and knee extension (1-10 points each).(35) We compared the sum of the MMT measures (maximum score, 110/child/visit) between treatment and control groups. All children were evaluated for pain using the NIH Functional Assessment Pain Score, a four-point pain self-assessment at each visit: 4 = no pain, 3 = pain not interfering with functional activities, 2 = pain interfering with functional activities, 1 = intractable pain.

Statistical methods

Primary outcome variables of the study were vertebral DXA z score, height, and area. The functional measures were secondary outcomes. The power calculation done before the study was based on a t-test of the DXA z score between pamidronate and no pamidronate groups and found that nine patients per group gave 90% power to detect a difference of about 1.5 SD.

For analysis of protocol data at the 1- and 2-year time-points, a more sensitive repeated measures model was used to incorporate serial observations taken on the patients. Each primary outcome variable was analyzed separately to determine changes caused by pamidronate treatment. The spatial correlation structure was used to model dependence between observations taken on the same patient that decrease as the time difference increases. The model includes terms for baseline age and the presence or absence of pamidronate or rGH. The pamidronate × GH interaction was not significant. Each term is allowed to affect the slope against time. The effect of treatment is estimated as the difference in slope against time in the two groups. Correlations between changes in z score and vertebral height and area were estimated using the sample Pearson product-moment correlation. Two-tailed t-tests were used to analyze fracture rate, QCT z scores, BAMF scores, MMT, functional pain score, and serum markers of bone formation.


Study population and general response

Characteristics of the 18 children who participated in this randomized, controlled, nonblinded study are shown in Table 1. Average age at entry for the pamidronate group was 11.05 ± 2.4 years (range, 7-13 years) and for the control group was 9.97 ± 3.1 years (range, 4-13 years). All treated children experienced acute phase reactions with the first infusion cycle. No other complications were noted.

Table Table 1.. Characteristics of Participants, Growth, Fractures, and Motor Function
original image

Growth rates in the two groups were unchanged. Children in the treatment group had a growth rate of 4.36 ± 2.13 cm/year in the year before treatment, 4.44 ± 2.00 cm/year in the first study year (p = 0.94), and 3.32 ± 3.78 cm/year (p = 0.66, with respect to baseline) in the second treatment year. The control group children had growth rates of 3.37 ± 4.66 cm/year before the study year and 3.25 ± 2.46 cm/year in the study year (p = 0.94).

Serum parameters of bone synthesis were measured just before the next infusion and compared with the time 0 baseline. BSALP, OC, and PICP showed no significant change from baseline at any infusion time, indicating that each interval inhibition had returned to baseline.

Spine density and geometry, fractures

Z scores for DXA measurements at L1-L4 (Table 2) increased significantly in treated patients compared with the control group by repeated measures analysis (p < 0.001). DXA z scores for the treated group increased steadily with each infusion (Fig. 1; Tables 2 and 3) through 12 months but did not increase further during the extended treatment period. The individual z score change in the first study year ranged from 0.1-2.75 SD, with an average of 1.4 ± 0.7 SD. The control group had an average z score indistinguishable from the treatment group at baseline, but did not show an increase over baseline in z score during the study (Tables 2 and 3). It is worth noting that BMD of control patients was not artifactually increased by decreased vertebral area because of compressions. Decreased vertebral area (L1-L4) occurred in three patients, all in the control group. Each of these patients experienced a decrease in DXA L1-L4 z score during the same interval.

Table Table 2.. Effect of Pamidronate on BMD, Vertebrae, Vertebral Area, and Compressions
original image
Table Table 3.. Effect of Pamidronate on BMD, Vertebrae, Vertebral Area, and Compressions
original image
Figure FIG. 1..

Changes in L1-L4 BMD of all patients enrolled in the study. L1-L4 BMD raw values plotted versus patient age for (A) treatment and (B) control groups. Gray zone shows normal values for age. L1-L4 BMD z scores plotted versus time from entry into protocol for (C) treatment and (D) control groups. Within each panel, successive BMD determinations for the same participant are connected, using symbols shown in inset in each panel.

Spine QCT z scores (Tables 2 and 3) did not increase significantly in the treatment group. However, the z score interval change was significantly greater in the treatment than in the control group (treatment, 0.51 ± 0.49; control, −0.06 ± 0.59, p = 0.05).

All patients had baseline vertebral compressions. By repeated measures analysis, the treated group had a significantly greater rate of increase than the controls in summed L1-L4 midvertebral height (Table 4) and area. Within the treated group, the changes in summed L1-L4 midvertebral height and area were significant at 1 year of treatment (height: baseline, 2.03 ± 1.28 versus 12 months, 2.5 ± 1.03, p = 0.018; area: baseline, 1.35 ± 0.95 versus 12 months, 1.74 ± 0.99, p = 0.006); the improvements were maintained but did not increase significantly after the second year of treatment (height, 2.86 ± 0.75 at 24 months, p = 0.25; area, 1.76 ± 0.59, p = 0.11).

Table Table 4.. Comparison of Vertebrae of Treatment vs. Control Groups
original image

The vertebrae most likely to experience compressions from weight bearing, T12, L1, and L2, were examined individually. During the first year, the treatment group had a significantly greater rate of increase than the control group in L2 midvertebral height and area, but not L1 or T12 (Table 4). The L2 height and area did not increase further in the second treatment year (p = 0.60 and 0.34, respectively). The rate of increase in summed T12-L1-L2 height and area was also significantly greater in the treatment than in the control group during the first year (Table 4) but did not increase significantly during the extended treatment period (p = 0.52 and 0.25, respectively).

The incidence of long bone fractures of the lower extremities (Table 3) did not change significantly compared with baseline in either the first (p = 0.09) or second year (p = 0.29) of treatment. The incidence of upper extremity fractures decreased significantly in the first year of treatment (p = 0.04) but did not further decrease in the second year (p = 0.84). Time to first lower extremity fracture during the study was compared between the treatment and control groups. The average time to first fracture was 11.3 versus 9.3 months for the treatment and control groups, respectively (two-tailed t-test, p = 0.6).

Functional and activity levels

The effect of pamidronate infusions on lower extremity gross motor function, muscle strength, and pain was determined. Pamidronate did not significantly affect these functional assessments. The BAMF gross motor scale is a 10-point assessment that rates gross motor ability from preambulatory head control (1), through transfers (4), standing (5), walking with decreasing gait aids (6–8), walking with no gait aid (9), to running (10).(34) BAMF scores of treatment and control groups were not significantly changed during either the controlled or the extended treatment portions of the study (Table 5). On average, both groups remained in the functional category of ambulation with maximum gait aids.

Table Table 5.. Results of Assessment of Functional and Activity Levels
original image

Evaluation of the sum of lower extremity and abdominal muscle strength (Table 5) showed no difference between the treatment and control groups at entry into study (p = 0.49). Neither group changed significantly during the 2 years of the study.

Evaluation of pain status with a four-point pain scale yielded no change in self-evaluation in the treatment or control groups (Table 5). We did not observe the previously reported perceived decreases in skeletal discomfort and well being. More subtle pain relief might have increased activity level and risk-taking behavior, leading to a decrease in time to first fracture in the treated group. This was not seen. Several patients did report increased endurance and/or decreased back discomfort in the morning; most reported no perceptible changes.

Variability of response within the treatment group

Of interest was the variability in response to pamidronate treatment (Table 2; Fig. 1). Some treated patients had a robust response in all measurements. Others had increased BMD but not improved measurements of vertebral body strength. Most striking were increases in vertebral heights in patients 3, 4, 5, and 9, whereas other treated patients experienced decreases or no change in vertebral height (patients 2, 6, and 7). Patients 1, 2, 4, 5, and 9 more than doubled L1-L4 vertebral area; vertebral areas of patient 7 were essentially unchanged. Some patients had more robust DXA increases than others. Patient 5 had a >3 SD change in DXA z score, the greatest increase of any treated patient. Patients 1, 3, 4, and 8 had increases of >2 SD, whereas patients 2 and 7 had <1 SD changes. Two patients who had clear increases in BMD did not have increased vertebral height (patients 1 and 6). There was a moderately strong correlation between changes in DXA and changes in L1-L4 or T12-L2 summed height (r = 0.52 and 0.67, respectively) but low correlation of DXA changes with L1-L4 or T12-L2 summed area (r = 0.18 and 0.19, respectively). Variability of response did not correlate with patients receiving rGH. In the treatment group, patients 2, 3, 4, and 6 received rGH as well as pamidronate. This group included two of the best vertebral DXA and height responders (patients 3 and 4) and two of the worst (patients 2 and 6).


In the last 5 years, a series of reports has appeared on bisphosphonate treatment of children and infants with OI.(9,17–24,26,27) Most have reported uncontrolled observational studies or have involved historical controls. Vertebral BMD increased in treated children, as expected from the known action of pamidronate. The study by Glorieux et al.(20) reported an increase of 41.9%/year in L1-L4 BMD and a z score increase of 1.9 SD. A study of 18 Australian children with OI reported a 62% increase in L1-L4 BMD/year with an improvement in z score of 1.8 SD.(25) Small trials with 10,(9) 7,(36) or 6(26) children with OI reported similar increases in lumbar spine BMD and z scores. In infants <3 years of age, BMD was reported to increase 86-227%, and z scores increased 2.5 SD.(28) The one reported controlled study of bisphosphonate in OI children used olpadronate(30); the treated group experienced an increase in DXA z score of 1.67 SD over 2 years of treatment.

The study of pamidronate in types III and IV OI reported here was controlled for the first year. Seven children in the treatment group continued on pamidronate for an additional 6-21 months (the extended treatment phase). Patients in the treatment group received 30 mg/m2 intravenous pamidronate divided over 3 days every 3 months (∼4 mg/kg/year). All treated children had a first cycle febrile reaction. DXA z scores increased steadily in the treatment group from baseline through 12 months (p = 0.054), with an average increase of 1.48 SD. The change in average DXA z score obtained using 4 mg/kg/year was similar to the change in DXA z score obtained using ∼12 mg/kg/year.(20,26,27) The lower dose of pamidronate is sufficient to obtain significant DXA increases in children with OI. Higher pamidronate doses may have resulted in significant changes in variables showing a trend toward improvement, such as lower extremity (LE) fracture rate or further increases in vertebral height and area. However, given the nonlinear relationship of DXA improvement and dose, it is less likely that higher doses would have caused changes in pain, motor function, and muscle strength. At the dose of pamidronate used in this study, markers of bone formation were stable before each infusion cycle, but we are unable to comment on the effect of the drug on bone resorption markers.

DXA z scores were maintained but not further increased during the extended treatment phase. Volumetric BMD (QCT) also increased significantly in treated children in comparison with the control group (p = 0.05).

The increased DXA z score probably reflects retention of mineralized cartilage as well as augmented bone volume. Rauch et al.(29) reported increased mineralized cartilage in the iliac crest biopsies of treated children, and similar findings occur in the treated long bones of the Brtl mouse model for type IV OI (JC Marini, unpublished data, 2004). However, even for a normal matrix, BMD is not a direct measure of bone strength. Vertebral area and height measurements provide a functional assessment of vertebral resistance to the compressive forces of the axial skeleton. Comparison of L1-L4 and T12-L2 vertebrae in our treated and control groups by repeated measures revealed a significant increase in vertebral height (p = 0.014 and 0.018, respectively) and area (p = 0.005 and 0.05, respectively) in the treated children. The average L1-L4 area change in our treated children (≈13%) is less than the previously reported 33.9%(20) or 30.7%(28) increases in vertebral area. The controlled trial of olpadronate in children with OI did not detect a significant change in vertebral geometry.(30) The gains in summed vertebral height and area were maintained but not significantly increased during the extended treatment period. The children in our trial exhibited substantial variability in the response of DXA and vertebral measurements to pamidronate, perhaps related to the extent of matrix insufficiency caused by specific mutations, which may contribute to statistical limitations of area and height measurements.

The effect of pamidronate on long bone strength was reflected in fracture rate. The upper extremity fracture rate decreased significantly during the first year of treatment (p = 0.04) but did not further decrease during the second year (p = 0.84). The lower long bone fracture rate did not decrease significantly in the treated group during the controlled (p = 0.09) or extended treatment trial (p = 0.29). When upper and lower long bone fracture rates were combined, the treatment group had a trend toward significance (p = 0.06). The trend in the first year is similar to the decrease in fracture rate in small treatment trials with seven (p = 0.09)(36) and six children with OI.(26) The controlled trial of olpadronate reported a significant decrease in relative fracture risk for the treatment group, although not a significant decrease in fracture number between the two groups.(30) Significant decreases in fracture rates were seen with larger groups in uncontrolled trials.(19,20,22–25) The early decrease in fracture rate may reflect increased bone mass, especially increased trabecular number,(29) which may compensate for OI matrix deficiency. In this study, the decrease in upper extremity fracture rate was significant, whereas the decrease in lower extremity fracture rate was not. A larger study group or higher dose of bisphosphonate during the first year may have resulted in a significant decrease in lower extremity fracture rate as well. Long bone fracture resistance did not improve in the extended treatment trial. Cautions have been raised about long-term bisphosphonate treatment.(37) Roldan et al.(18) used QCT to show decreased moment of inertia in treated long bones. Alendronate treatment of the Brtl mouse model for OI worsened the brittleness of treated murine femurs.(38) These results suggest that risk of fracture in response to torque persists in treated OI bones and may be exacerbated by longer length of treatment.

Whereas a larger sample size would be needed to detect a small but significant interaction of pamidronate and GH, the presence of such an interaction would not change our conclusions about pamidronate. The reason for this is that a model with interaction between pamidronate and GH yields the following estimated slopes for vertebral DXA z scores: (1) no drug, 0.36; (2) rGH only, 0.66; (3) pamidronate only, 2.21; (4) pamidronate + rGH, 1.81. The difference in slopes for pamidronate only versus untreated is 1.85 (SE = 0.52, p = 0.0012). Thus, pamidronate has a more marked effect on bone than does GH, which allows us to clearly and accurately detect the effect of pamidronate in this distribution of drug regimens.

Concerns about growth rate in children treated with bisphosphonates have been raised. This is of particular relevance in OI, because affected children already have syndromic short stature. In this study, growth rates in both groups were unchanged compared with the 2 years before the study. The children treated with rGH were evenly distributed between the treatment and control groups. Before enrolling in the pamidronate study, six of eight children on rGH had a documented positive response to rGH, which would be reflected in the “prior year” growth rate. The addition of bisphosphonate did not result in a further increase in growth rate. There was no positive or negative interaction detected between pamidronate and GH, although we cannot exclude the possibility of such an interaction in this small sample. This agrees with the majority of small, uncontrolled studies.(17,19,22,23,25,36) The single study in which the height z scores of patients with type IV OI were reported to be significantly increased(39) after 4 years of infusions involved comparison with a group of untreated historical controls and is difficult to evaluate on this topic. Furthermore, treated OI children experienced a greater decrease in bone turnover with pamidronate than healthy children,(29) making a positive change in syndromic short stature unlikely.

We also used measures of function in our controlled trial, including ambulatory motor function, MMT, and self-assessments of pain. Uncontrolled trials have reported significant gains in ambulatory status, with some children progressing from wheelchair mobility to full or partial ambulation.(19,20,22,23,25) Many participants in uncontrolled trials had not been previously enrolled in longitudinal studies that emphasized intensive rehabilitation. For those children not receiving regular rehabilitation before enrollment in uncontrolled trials, some gross motor gains during treatment may have come from the systematic rehabilitation intervention they received from the multidisciplinary team. Trial participation, per se, may thus have increased compliance with a rehabilitation regimen or may have imparted the confidence to make functional advances. Children in our longitudinal population all received regular rehabilitation before enrollment in this study, using interventions designed to promote strength and endurance. We found no difference in functional mobility between groups at baseline, after 1 year of treatment, or after the extended treatment phase. Similar results were obtained in the controlled olpadronate study, in which no significant change in Pediatric Evaluation of Disability Inventory (PEDI) score was detected.(30)

Muscle strength, measured as grip force, has been reported to increase in uncontrolled trials,(40) but a similar result was not obtained in the controlled study of olpadronate.(30) The results from uncontrolled trials were attributed to decreased bone pain, although a direct effect on muscle was not ruled out. In this NICHD trial, MMT was used to follow the effect of the drug on muscles related to gross motor function and ambulation. There was no significant change in summed muscle test scores from baseline to 1 year of treatment or during the extended treatment phase.

Multiple uncontrolled studies reported improvement in bone pain in subject participants.(19,20,23,24,28) Aström and Soderhall(23) reported a significant decrease in days per month with pain in younger participants but a less pronounced effect in older children. Using a four-point self-assessment of pain, we did not detect an improvement in treated or control groups. Children did not report an increase in discomfort in the weeks just before an infusion or a decrease in pain in the weeks after an infusion.

Thus, our controlled trial of pamidronate in children with types III and IV OI confirmed the benefits of short-term use for lumbar vertebral DXA (L1-L4) and coronal area. In the extended treatment interval, gains from the first year of treatment were maintained but were not further improved. These changes may lead to increased resistance to scoliosis, which would result in improved long-term quality of life. Our controlled trial does not, however, support the improvements in motor function, muscle strength, or pain reported in uncontrolled trials. In a context of maximized physical rehabilitation, we did not see an additional functional effect from bisphosphonate treatment, in agreement with findings from the controlled study of olpadronate in OI.(30) Furthermore, the treated group in this controlled trial of pamidronate showed considerable variability in response to pamidronate.

These data also support our interpretation that children with OI obtain the greatest benefit from pamidronate during ∼2 years of infusions, after which the beneficial bone response tapers. There was no significant difference between the 1- and 2-year data for the vertebral DXA z score, L1-L4 summed vertebral height and area, L2 height and area, and T12-L2 summed vertebral height and area.

In the NICHD OI program, we have considered the benefits and risks to the growing skeleton associated with pamidronate treatment. A high cumulative dose of bisphosphonates causes increased risk of abnormal long bone modeling(41) and osteotomy nonunion.(42) Remnants of mineralized cartilage in the treated growing skeleton(43,44) create matrix discontinuities that could increase fracture risk. Alendronate treatment of the Brtl mouse model for type IV improves bone geometry and loading before fracture but decreases bone material quality and alters osteoblast morphology.(44) Based on the tapering of skeletal benefit seen in this study and the increased risk of side effects with higher cumulative doses, we have decided to limit the duration of pamidronate treatment in our pediatric OI patients to 2-3 years. After this treatment interval, the children are followed regularly for maintenance of vertebral parameters.


The authors thank the 9 West Inpatient Unit and 9 West Day Hospital nurses at the Clinical Center, NIH, for assistance with pamidronate infusions and gathering data. We also thank Eli Lilly and Co. for continued generosity in providing Humatrope for the OI GH protocol. The authors thank Scott M Paul, MD, of the Rehabilitation Department, Clinical Center, NIH, for helpful and productive discussions of patient care and assistance with data collection. We are grateful to the children and their families for their continued dedication to research on OI and for their participation in a controlled trial.