Two-Year Clinical Trial of Oral Alendronate Versus Intravenous Pamidronate in Children With Osteogenesis Imperfecta


  • Linda A DiMeglio MD, MPH,

    Corresponding author
    1. Department of Pediatrics, Section of Pediatric Endocrinology and Diabetology, Indiana University School of Medicine, Indianapolis, Indiana, USA
    • Riley Hospital, Room 5960 702 Barnhill Drive Indianapolis, IN 46202–5225, USA
    Search for more papers by this author
  • Munro Peacock

    1. Department of Medicine, Indiana University School of Medicine, Indianapolis, Indiana, USA
    Search for more papers by this author

  • The authors have no conflict of interest


A 2-year prospective, partially randomized open-label trial comparing oral alendronate with intravenous pamidronate therapy in children with OI showed equivalence in increasing total body BMD, spine BMD, and linear growth, and decreasing bone turnover and fracture incidence. Children with mild OI had greater responses than severe OI in BMD and growth.

Introduction: Bisphosphonate therapies increase BMD and may reduce fractures in children with osteogenesis imperfecta (OI). A study directly comparing oral with intravenous bisphosphonate has not been published. This clinical trial compares oral alendronate with intravenous pamidronate in children with OI using an open-label, prospective, 2-year, randomized design.

Materials and Methods: Children over the age of 3 years were stratified by bone age, pubertal stage, and type of OI and then randomized to receive oral alendronate 1 mg/kg/day in tablet form or intravenous pamidronate, 3 mg/kg/4 months. One child was assigned to pamidronate. One child randomized to intravenous pamidronate changed to oral alendronate. Eighteen children completed 12 months of therapy: nine on oral alendronate and nine on intravenous pamidronate. Primary outcome efficacy was increase in BMD. Secondary outcomes included changes in bone turnover biomarkers, fracture incidence, and growth.

Results: Total body and lumbar spine BMD increased, turnover markers decreased, and linear growth increased equivalently with oral and intravenous therapy. Fracture incidence showed a trend to decrease in both groups, with a significant decrease in fracture rates when the oral and intravenous groups were pooled. There were greater responses in BMD and growth in children with milder OI (type I) than those with more severe disease (types III and IV), but there were no significant effects of age or pubertal stage.

Conclusions: Oral and intravenous bisphosphonate therapies are equally effective in children with OI and are particularly effective in milder forms. The oral route is highly acceptable in children and has practical advantages over the intravenous route.


ALTHOUGH OSTEOGENESIS IMPERFECTA (OI) is a genetic disease of collagen synthesis, it manifests clinically with increased bone fragility. (1) In most affected children, the recurrent fractures are associated with reduced BMD. Severity of OI is greatest in type II (infantile lethal form) and types III-V and mildest in type I, a form that may have only a few fractures over a lifetime. (2, 3)

Bisphosphonates are established as therapy to reduce the rate of osteoclastic bone resorption, thereby increasing BMD and decreasing osteoporotic fracture incidence in adults. (4, 5) In children with OI treated with intravenous(6–10) or oral(11–13) bisphosphonates, increases in BMD have been reported. Several studies in OI have shown decreases in fracture rates using either historical controls(8) or comparisons of fracture rates pretherapy and on therapy. (6, 14) A placebo-controlled trial that used olpadronate showed a reduction in long bone fractures in the treated group. (13)

We have reported the results of a feasibility study comparing oral to an intravenous bisphosphonate therapy in children with OI over 8–12 months. (15) In that preliminary study, we showed that oral bisphosphonate was well tolerated and seemed to increase BMD. We now report the results of the extension of that study over 2 years of treatment in 18 children assessed every 4 months for BMD, fracture incidence, bone turnover markers, and growth in relation to type of OI, age, and pubertal stage.



Children with OI were treated with either oral alendronate or intravenous pamidronate using an open-label, prospective, randomized clinical trial design approved by the Indiana University-Purdue University Indianapolis and Clarian Institutional Review Board. The diagnosis of OI was made clinically by a team consisting of a geneticist, orthopedic surgeon, and pediatric endocrinologist. No child had previously received any bisphosphonate therapy.

To minimize effects of probable confounding factors, children were stratified according to clinical severity of OI, pubertal stage, and radiological bone age into four groups: group 1, mild OI (type I or unclassified with ≤3 fractures/year), Tanner I puberty, with bone age ≤9 years for girls and ≤11 years for boys; group 2, severe OI (type III or IV or unclassified with >3 fractures/year) and Tanner I; group 3, mild OI, Tanner II or higher with bone age >9 (girls) or >11 (boys); group 4, severe OI and Tanner II or higher. The bone ages of 9 years for girls and 11 years for boys were chosen as representing −2 SD for the normal timing of the onset of puberty. (16)

A randomization schedule with a computer random number generator using a block-randomization scheme was developed. Random number sequences were concealed until the children were assigned to a group. A nurse determined from the randomization schedule whether the child would receive oral or intravenous treatment.


Study protocol

Informed consent was obtained from parents with assent from children over the age of 7. Children attended the General Clinical Research Center at Indiana University every 4 months. Fasting blood and second-morning urine voids were collected at each visit. Heights were measured using a Harpenden stadiometer and lengths using a recumbent length board. Heights/lengths and weights were converted to age- and sex-specific Z scores using reference software. (17) LAD graded pubertal stage. (18, 19) Dietary calcium intake was evaluated by the General Clinical Research Center Nutritionist through diet history recall at each visit and intake was maintained at >1000 mg/day through dietary advice. Diary records and parental recall assessed compliance with oral bisphosphonate therapy. Fracture histories were obtained by parental recall and by diary.

Children received alendronate (Fosamax; Merck) 1 mg/kg/day orally in the form of tablets or pamidronate disodium (Aredia; Novartis) 1 mg/kg/day intravenously for 3 consecutive days (i.e., 3 mg/kg) every 4 months. To provide equivalent oral and intravenous doses, it was assumed based on available in vitro and in vivo data that the oral alendronate dose had 0.5% intestinal absorption, (20) 50% skeletal bioavailability, (21) and five times greater potency than intravenous pamidronate. (22) Thus, daily oral alendronate over 4 months was considered to be equivalent to intravenous pamidronate given over 3 consecutive days, every 120 days. Alendronate was taken as either one or two 10-mg tablets, rounded to the nearest 10-mg dose for weight. Pamidronate was diluted in 250 or 500 ml of normal saline depending on body weight and slowly infused over 4 h with the maximum dose ≤180 mg over 3 days. Other medications were continued, and stool samples were collected for assessment for occult gastrointestinal bleeding.

Skeletal measurements

BMD, BMC, and area at the spine (L2-L4) and total body were measured every 4 months by DXA using a Lunar DPXL instrument (Lunar, Madison, WI, USA). All measures were performed and analyzed using pediatric software. BMD was transformed to an age-specific Z score using manufacturer reference data. (23) Because the available software for the Lunar densitometer only offers Z scores starting at the age of 5, for children under the age of 5, Z scores were generated by comparing children to the 5-year-old norms.

Bone ages were assessed from posteroanterior radiographs of the left hand and wrist according to the method of Greulich and Pyle. (24) These annual radiographs and interim radiographs taken for suspected fractures or other orthopedic assessments were examined to confirm fracture history, to examine for excess mineralization and sclerosis, and to examine for changes in the growth plates. Delayed fracture healing was assessed both by review of available radiographs for fracture lines at least partially visible 1 year after the initial fracture event(25) and by parental report of orthopedic surgeon assessments of fracture healing rates.


After voiding the overnight urine, blood and the second morning urine sample were collected in the fasting state. Serum samples were processed after clotting at room temperature for 30 minutes. They were aliquoted and stored at −80°C until analyzed for calcium by Arsenazo Dye method of Roche Cobas-Mira, total alkaline phosphatase (ALP) and bone-specific ALP (BALP) by ELISA assay (Quidel, Mountain View, CA, USA) osteocalcin (OC; assayed by IRMA), intact PTH(1-84) (IRMA assay; Nichols Institute, San Juan Capistrano, CA, USA), 25-hydroxyvitamin D (RIA; Dia Sorin, Stillwater, MI, USA), and 1,25-hydroxyvitamin D (using an RIA kit after extraction and purification; Dia Sorin). CVs for these assays have been described previously. (15)

Urine samples were aliquoted and stored at −40°C until analyzed. Urine calcium and creatinine were analyzed on the Roche Cobas Mira (Roche Diagnostics, Indianapolis, IN, USA). Urine calcium was analyzed using the Arsenazo III method. Urine creatinine was analyzed using the Jaffe Kinetic method. Urine cross-linked N-teleopeptide of type I collagen (NTx) was assayed after 1:4 dilution using an ELISA assay (Ostex, Seattle, WA, USA). Free deoxypyridinoline (FDPD) was assayed after 1:20 dilution using an ELISA assay (Quidel). CVs for assays have been described previously. (15) Biochemical variables are reported at 0, 4, and 24 months to show acute and chronic responses to therapy.

Statistical analysis

Analysis included children completing 2 years of therapy (Table 1). Reproducibility analyses for densitometry variables were done using intraclass correlation coefficients. (26) Changes over time were examined by repeated-measures ANOVA and, if the ANOVA was significant, as paired t-tests. Treatment effects on BMD were calculated both as absolute and Z score changes in BMD from baseline to 2 years. A positive change in Z score indicated that treatment maintained faster accrual than normal, whereas a negative value indicated a slower mineral accrual than normal. Treatment effects on BMC and area were calculated as absolute changes. Fracture incidence was calculated as the annual fracture rate before treatment versus the annual fracture rate on treatment, and differences were tested using a paired sample t-test. Differences between oral and intravenous groups, children with mild and severe OI, pre- and postpubertal children, and males and females were assessed by ANOVA followed by a least significant difference test. All tests were two-tailed, and p < 0.05 was considered significant. All statistical analyses were performed with SPSS software release 12 (SPSS, Chicago, IL, USA).

Table Table 1.. Patient Characteristics at Baseline
original image



Nine children were randomized to intravenous pamidronate, nine to oral alendronate, and one child was assigned to intravenous treatment because of concerns of a history of vague abdominal pain accompanied by heme-positive stools. One child randomized to intravenous pamidronate was changed to oral alendronate before any intravenous was given because of great difficulty obtaining intravenous access. One child on oral alendronate was lost to follow-up after 8 months and is therefore not included in this analysis. Eighteen children completed 2 years of study (Table 1).

There were no significant differences in group characteristics. In both groups, three had severe OI (type III/IV) and six had mild OI (type I). Six in the oral group and five in the intravenous group were prepubertal at the start of study. The mean age was 8.7 years (oral group, 9.0 years; range, 3.8-12.7 years; intravenous group 8.4 years; range, 3.0-13.7 years). The oral group had a mean of 1.7 fractures/year (range, 0.6-4.5 fractures/year), and the intravenous group had an average of 2.1 fractures/year (range, 0.4-4.8 fractures/year). Sixteen were small for age, with a mean SD score of −2.7 (range, 1.1 to -6.6) for height. In all children, bone age corresponded to chronological age (data not shown). Nine children (five on oral; four on intravenous) had orthopedic hardware (rods, plates) at the start of the study.

There were no significant differences between intravenous and oral groups in BMD, BMC, area (Table 2), or biochemical markers of bone turnover (Table 3). Total body BMD was low for chronological age, with a mean Z score of −1.5 (range, −3.8 to 0.9). BMD at L2-L4 was markedly low, with a mean Z score of −3.4 (range, −5.7 to −1.6). Baseline serum calcium, 25-hydroxyvitamin D, and 1,25-dihydroxyvitamin D were normal.

Table Table 2.. BMD (g/cm2 and as Z Score), BMC (g), and Area (cm2) for the Oral Group (n = 9), Intravenous Group(n = 9), and for Both Groups Combined (n = 18)
original image
Table Table 3.. Biochemistry for Oral and Intravenous Groups
original image


Dose of bisphosphonate

Children in the oral group received total doses ranging from 220 to 750 mg/kg of alendronate over 24 months, equivalent to an absorbed dose of 1.1-3.8 mg/kg. Children on intravenous therapy received a total of 18 mg/kg of pamidronate. These doses yielded an estimated skeletal retention of 3.5 μg/kg/day of alendronate and 12 μg/kg/day of pamidronate.

BMD, BMC, and area

Children had significant increases in total body and L2-L4 (Table 2) BMD beyond that expected with normal BMD accrual for age (Fig. 1). The group on oral therapy increased the Z score at L2-L4 from −3.2 ± 1.0 to −1.1 ± 1.2; the group on intravenous therapy increased from −3.2 ± 1.3 to −1.3 ± 2.0. These changes translated to a mean annualized change in total body BMD of 8.4% and in L2-L4 BMD of 34.7%. There were corresponding increases in both BMC and bone area. In children with OI, the reproducibility of repeated BMD measures assessed by intraclass correlation coefficients were 0.994 for total body and 0.977 for L2-L4 spine BMD.

Figure FIG. 1..

Shaded areas represent the 1 (dark hatch) and 2 (light hatch) SD range for age- and sex-matched children. The numbers in the box refer to patient number in Table 1 and oral and intravenous (IV) indicate treatment group.20


There were no significant differences in response between the two treatment groups at 0, 4, 12, or 24 months. Table 3 shows the data from the 0, 4, and 24 month time-points. Serum markers of bone turnover showed decreases in total ALP and BALP. Urine NTX/Cr and DPD/Cr decreased over time.

Longitudinal growth

Both the oral and intravenous group showed trends in improvement in height Z score, with the oral group increasing from −3.0 to −2.7 and the intravenous group increasing from −2.5 to −2.1. Because there were no significant differences in response in height between the two treatments, changes were analyzed in the two treatment groups combined. Z scores in height or length compared with normal children increased significantly (p = 0.02).

Orthopedic interventions/fracture incidence

During the 2 years on protocol, three children (two oral; one intravenous) had hardware added, and three different children (two oral; one intravenous) had hardware removed. The annualized fracture rates in both groups trended toward a decline, with a significant decrease in the combined group (Table 4; p < 0.05).

Table Table 4.. Annualized Fracture Rates for Oral Group, Intravenous Group, and Both Groups Combined
original image


Radiographs of patients on intravenous pamidronate showed multiple discrete bands of hypermineralization under growth plates corresponding to infusion courses. In contrast, children on oral alendronate showed a uniform broad band of subepiphyseal hypermineralization corresponding to the time on therapy. Treatment with pamidronate did not alter the rate of fracture healing clinically or the radiographic appearance of growth plates.

Comparison by OI type, gender, and pubertal status

Children with severe OI (type III/IV) were shorter than those with milder OI (type I; height SD for severe OI, −5.3 ± 1.3 versus −1.4 ± 1.8; p < 0.001). There was an increase of 0.4 in height Z score in the type I children (p < 0.01) and a change of 0.2 in the type III/IV group (p = 0.47).

Children with mild OI had greater mean total body area than those with severe disease. L2-L4 area in the type I children was not significantly different from that of type III/IV children, although it was higher at all times measured (Table 5). At all time-points measured, children with mild OI had a significantly greater BMD and BMC at both the total body and the lumbar spine. The total body changes in BMC, area, and BMD were all significantly greater in mild OI than in severe OI. The mean total body BMD Z score in the children with mild OI increased from −1.1 to 0.3 at 24 months (p < 0.001), whereas in those with mild OI, it did not change significantly (−2.2 to −2.1; p = 0.45; Fig. 2). The changes at the lumbar spine were only significantly different between groups for BMC at 12 months. In type I, the mean BMD Z score at L2-L4 spine increased from −2.9 to −0.5 at 24 months (p ≤ 0.001), whereas in type III/IV, the increase, although significant (p = 0.02), was less (−3.9 to −2.5).

Table Table 5.. BMD (g/cm2), BMC (g) and Area (cm2) for the Mild Group (n = 12) and Severe Group (n = 6)
original image
Figure FIG. 2..

Change in (A) total BMD and (B) lumbar spine BMD Z score at 12 and 24 months of treatment in mild and severe OI. *p < 0.05 between groups. Data shown are mean and SD.20

There were few significant differences by severity in response to treatment in turnover markers or calcium-regulating hormones. At baseline, children with milder OI had lower serum phosphorus levels (5.1 ± 0.5 versus 4.4 ± mg/dl; p < 0.02) and lower urinary NTx levels (568 ± 278 versus 921 ± 199 nMBCE/mM; p < 0.02); by 4 months, these levels were comparable and remained so throughout the treatment period. Osteocalcin tended to be higher in the mild OI children at all time-points, and significantly different from children with severe OI at 12 months (56 ± 39 versus 22 ± 7; p < 0.02).

Children with milder OI had fewer fractures per year both before and on therapy than those with severe OI. Before therapy, children with type I OI had 1.3 ± 1.0 fractures/year compared with 3.1 ± 1.4 fractures/year for those with type III/IV (p < 0.01). On therapy, fracture rates were 0.6 ± 0.7 for the mild group and 2.2 ± 1.5 for the severe group (p < 0.01 for comparison by group).

Gender did not affect the biochemical, growth, or densitometric measures of response. Nor did pubertal status, with the exception of lumbar spine area, which increased more in the pubertal group over time than the prepubertal children (p < 0.05, data not shown).


The only side effect of treatment was an acute phase reaction on the second day of the first infusion cycle. (27) This consisted of fevers, accompanied in some cases by myalgia and emesis. Symptoms were well controlled with acetaminophen. None of the children in the oral group had gastrointestinal problems (abdominal pain, bleeding) or positive occult blood.


Children with OI in this partially randomized study, whether treated with oral or intravenous bisphosphonates, showed increases in BMD over 2 years above that expected for normal skeletal growth. There were no significant differences in densitometric variables between those treated with oral and intravenous therapy. In other studies, intravenous pamidronate at similar doses to those used in our study showed annualized percent changes in spine BMD of 42%(6) and 48%. (7) The increases in BMD in our study are of the same magnitude. A study using oral alendronate in children with OI found a 2-year increase in spine BMD of 53% compared with an increase of 69% in our study. (11) However, that study used a lower dose of alendronate, with children receiving between 5 (if they weighed <40 kg) and 10 mg/day (if they were >40 kg). Thus, some of the apparent difference in responses among studies may be caused by differences in dose, but as discussed below, it may also be caused by the presence of a larger percentage of type I patients in our study.

Children, whether treated with intravenous or oral medication, had decreases in bone turnover. Thus, our results are similar to published studies with intravenous pamidronate that showed significant decreases in serum total ALP, (6, 28, 29) osteocalcin, (28) and urinary calcium, (30) NTx, (29, 30) and DPD. (28) In our study, bone markers were decreased by 4 months and thus were very useful as an early sign that bisphosphonate was effective therapy. The ranges of markers of bone turnover are not well established in healthy children. Nevertheless, it is clear that, in children with OI, all bone markers are in the high normal range. However, it is not clear from our study whether this is a feature of the genetic abnormality in collagen or a more general combined effect of immobilization and fractures on bone turnover. Irrespective of mechanism, the decrease in bone markers indicates that bisphosphonates are effective in OI, at least in part, through a decrease in bone turnover.

In contrast to a previous report(31) using intravenous pamidronate, we found better responses in BMD to treatment in children with milder OI (type I) treated with intravenous pamidronate or oral alendronate. The greater increase in BMD in type I OI may be related to an increase in trabecular number, (31) in apatite crystal size, (32) in the mineralization of the osteons, (33) or a combination of all three. In our study, despite the large differences in BMD, there were no significant differences in the response in bone markers between those with mild and severe disease. This may indicate that the differences in response in BMD are not due solely to decreases in bone turnover and filling of a resorption space but may relate to the genetic defect itself. Thus, it is possible that some of the difference resides in the ability of bisphosphonate treatment in type I OI, which is characterized by reduced collagen quantity but with normal fibril structure, to produce more or better apatite deposition.

Although we have not genotyped our patients, most children classified with type I OI have null mutations affecting the amount of type I collagen produced, whereas those with type III or IV OI have dominant negative mutations that disrupt the entire collagen matrix. (3) It is plausible that, in the latter, bisphosphonate is less effective. It is perhaps relevant that in our long-term follow-up of studies beyond 2 years reported here, three mild patients increased BMD above +1.5 SD for age and were discontinued from treatment as dictated by our institution's protocol, whereas none of the severe cases have done so over the same period of time. If this difference in BMD effect by severity of OI is confirmed in other studies, it implies that pediatric treatment regimens should be tailored by type of OI.

Children in both the oral and intravenous treatment groups showed improvements in linear growth. The explanation for such an effect is not clear. It may represent a direct stimulation on growth, but perhaps more likely it is related to decreased numbers of fractures and to improvement in leg deformities from remodeling. The improvements were primarily in the patients with mild OI, which suggests that the children with the best response in bone also had the best improvements in overall growth. Our results differ from the published experience that either shows no overall growth improvement(29) or improvement only in patients with type IV OI. (34) Perhaps the likeliest explanation is that we had a relatively high proportion of children with type I OI in our study.

As has been shown in other studies, (6, 13, 29) we found a reduction of fracture incidence with treatment when we pooled the oral and intravenous groups. Clinically, many of the fractures that were sustained during treatment were less severe than those before therapy. However, a true effect on fracture rate is difficult to document in the absence of a placebo trial because of the natural history toward a reduction in fractures with age.

There are a number of limitations in this study. Our DXA results have several limitations. Available Z score references are based on limited normative data from healthy, normal-sized children. Because they are based on areal densities, they may be artifactually low in OI children who are small for age. Several of our children also had hardware added or removed during the course of the study, which may have affected our total body BMD results. Also, assessing longitudinal BMD measurements using cross-sectional pediatric reference data is not accurate. However, the changes in absolute BMD are similar to those reported in previous studies of bisphosphonate treatment for children with this disease. (6) Our single-site trial enrolled 18 children because OI is a rare disease. This number does not provide power to detect small differences in responses, prove equivalence of treatment, or establish an optimal oral dose. However, overall, the DXA, laboratory, and clinical responses of patients on oral and intravenous bisphosphonates appear similar after 24 months of treatment.

Our data provide modest reassurance that oral and intravenous bisphosphonate therapies are equally safe for children with OI. However, oral treatment for children with OI has clear advantages. The frequent hospitalizations required for intravenous bisphosphonate therapy are expensive, time intensive, and require intravenous access (difficult in children with OI because of vascular fragility). In addition, children on oral therapy showed bones that were more uniformly labeled during growth. This perhaps may impart more uniform bone quality. However, it is unknown if this results in greater biomechanical strength. It is perhaps of note that radiologically fractures sustained on treatment have not been noted through areas of hypermineralization either in the intravenous or oral groups. In this study, we used oral bisphosphonate in the form of tablets. At the time of starting the study, a liquid oral preparation was not available (Fosamax liquid was released in the United States on February 23, 2004). It is likely that the latter will be equally acceptable and effective.

Long-term effects of bisphosphonates beyond 2 years on growing skeletons are needed. This is especially important because bisphosphonates are known to accumulate in bone for many years. Recent reports of hypermineralization of bone, (35) avascular necrosis, (36) and delayed osteotomy healing(25) with bisphosphonate therapies highlight the need for continued careful monitoring of treated children both on and after cessation treatment.


The authors acknowledge the assistance of the members of the nursing, biostatistics, nutrition, and core laboratory staff at the Indiana University General Clinical Research Center. In particular, we recognize the contributions of LeeAnn Ford, RN, as study coordinator, Cindy McClintock, CDT, with the bone densities, and Ronald McClintock in the laboratory. We also acknowledge the cooperation of Drs David Weaver and Wilfredo Torres of Medical Genetics and Dr Kosmas Kayes of Orthopedic Surgery. This work was supported by NIH Grants K23 RR15538-01 and MO1 RR 00750 and a Clarian Values Fund Award (VF-47).

This paper was presented in part at the 26th Annual Meeting of the American Society of Bone and Mineral Research, in Seattle, Washington, USA, October 1–5, 2004.