Alendronate Treatment of the Brtl Osteogenesis Imperfecta Mouse Improves Femoral Geometry and Load Response Before Fracture but Decreases Predicted Material Properties and Has Detrimental Effects on Osteoblasts and Bone Formation


  • Thomas E Uveges,

    1. Bone and Extracellular Matrix Branch, The Eunice Kennedy Shriver National Institute of Child Health and Human Development (NICHD), NIH, Bethesda, Maryland, USA
    2. Current address: Trevigen, Gaithersburg, Maryland, USA
    3. These authors contributed equally to this work and are co-first authors
    Search for more papers by this author
  • Kenneth M Kozloff,

    1. These authors contributed equally to this work and are co-first authors
    2. Department of Orthopaedic Surgery, University of Michigan, Ann Arbor, Michigan, USA
    Search for more papers by this author
  • Jennifer M Ty,

    1. Bone and Extracellular Matrix Branch, The Eunice Kennedy Shriver National Institute of Child Health and Human Development (NICHD), NIH, Bethesda, Maryland, USA
    2. Current address: Department of Orthopaedic Surgery, duPont Hospital for Children, Wilmington, Delaware, USA
    Search for more papers by this author
  • Felicia Ledgard,

    1. Department of Surgery, University of Connecticut Health Center, Farmington, Connecticut, USA
    Search for more papers by this author
  • Cathleen L Raggio,

    1. Center for Skeletal Dysplasias, Hospital for Special Surgery, New York, New York, USA
    Search for more papers by this author
  • Gloria Gronowicz,

    1. Department of Surgery, University of Connecticut Health Center, Farmington, Connecticut, USA
    Search for more papers by this author
  • Steven A Goldstein,

    1. Department of Orthopaedic Surgery, University of Michigan, Ann Arbor, Michigan, USA
    Search for more papers by this author
  • Joan C Marini

    Corresponding author
    1. Bone and Extracellular Matrix Branch, The Eunice Kennedy Shriver National Institute of Child Health and Human Development (NICHD), NIH, Bethesda, Maryland, USA
    • Address correspondence to: Joan C. Marini, MD, PhD, Chief, Bone and Extracellular Matrix Branch, NICHD, NIH Building 10, Room 10N260, 9000 Rockville Pike, Bethesda, MD 20892, USA
    Search for more papers by this author

  • Dr. Ty owns stock in Merck. All other authors state that they have no conflicts of interest.


Long courses of bisphosphonates are widely administered to children with osteogenesis imperfecta (OI), although bisphosphonates do not block mutant collagen secretion and may affect bone matrix composition or structure. The Brtl mouse has a glycine substitution in col1a1 and is ideal for modeling the effects of bisphosphonate in classical OI. We treated Brtl and wildtype mice with alendronate (Aln; 0.219 mg/kg/wk, SC) for 6 or 12 wk and compared treated and untreated femora of both genotypes. Mutant and wildtype bone had similar responses to Aln treatment. Femoral areal BMD and cortical volumetric BMD increased significantly after 12 wk, but femoral length and growth curves were unaltered. Aln improved Brtl diaphyseal cortical thickness and trabecular number after 6 wk and cross-sectional shape after 12 wk. Mechanically, Aln significantly increased stiffness in wildtype femora and load to fracture in both genotypes after 12 wk. However, predicted material strength and elastic modulus were negatively impacted by 12 wk of Aln in both genotypes, and metaphyseal remnants of mineralized cartilage also increased. Brtl femoral brittleness was unimproved. Brtl osteoclast and osteoblast surface were unchanged by treatment. However, decreased mineral apposition rate and bone formation rate/bone surface and the flattened morphology of Brtl osteoblasts suggested that Aln impaired osteoblast function and matrix synthesis. We conclude that Aln treatment improves Brtl femoral geometry and load to fracture but decreases bone matrix synthesis and predicted material modulus and strength, with striking retention of mineralized cartilage. Beneficial and detrimental changes appear concomitantly. Limiting cumulative bisphosphonate exposure of OI bone will minimize detrimental effects.


Bisphosphonates are antiresorptive drugs used to treat postmenopausal osteoporosis and Paget's disease and as an adjuvant treatment for bone metastases.(1) More recently, they have been used off-label in children, including administration for osteogenesis imperfecta (OI). While the goal of pediatric use of bisphosphonates is to restore their bones to normal functional strength and ductility, the short-term outcomes in the growing skeleton may differ from those in adults. Furthermore, the prolonged half-life and recirculation of pamidronate in pediatric patients up to 8 yr after treatment cessation(2) may pose pediatric-specific skeletal and reproductive risks.

In the last decade, many children with OI have received 5 or more yr of cyclic pamidronate. The bone fragility of Sillence types II, III, and IV OI (lethal, progressive deforming, and moderately severe OI) is caused by structurally defective type I collagen, which forms an abnormal bone matrix that cannot accumulate a properly organized mineral phase.(3,4) Administration of bisphosphonates does not selectively affect synthesis of defective type I collagen. However, inhibition of bone resorption by osteoclasts increases bone mass, albeit of OI quality, and was postulated to improve clinical symptoms. Observational or historically controlled studies reported increased lumbar BMD by DXA, improved vertebral shape, and decreased fracture rates.(5–13) In paired iliac crest biopsies from OI children who received 2.5 yr of cyclic pamidronate administration, cortical width and bone volume increased significantly.(14) However, the baseline hardness, elastic modulus, and degree of mineralization of OI bone were unchanged by pamidronate treatment, leading the investigators to suggest that long-term administration might not increase the brittleness and fragility of OI bone matrix.(15)

Controlled trials of bisphosphonates in OI children administered several different compounds but provided unifying insights.(16–19) Vertebral height and area were consistently improved, indicating enhanced resistance to vertical compression. However, no controlled trial reported a decreased incidence of long bone fractures, although two groups(17,18) obtained decreased relative risk of nonvertebral fractures in treated children. The double-blind alendronate (Aln) trial involving 139 children did not obtain significant improvement in fractures or bone pain in treated children,(16) resulting in a labeling change specifying that Aln was not indicated for treatment of OI in children.(20)

Accumulating reports indicate that prolonged administration of commonly used doses of bisphosphonates has detrimental skeletal effects. Bisphosphonate-induced osteopetrosis and defective bone modeling were described in a child who received prolonged high doses of pamidronate; less dramatic modeling effects were seen with lower cumulative doses.(21) Delayed osteotomy healing was increased at conventional doses. Munns et al.(22) speculated that use of an oscillating osteotomy saw cauterized the site, implying that the blood supply needed for bone healing may be reduced after bisphosphonate treatment. Anecdotal surgical reports describe the inferior, rock-hard, shatter-prone quality of OI bone after prolonged treatment.

Studies in animal models are essential to understanding the effects of bisphosphonate treatment on long bone strength and quality. Previous studies provided important controlled data but involved the recessive oim mouse model, in which bone matrix is composed of homotrimeric α1(I)3 collagen rather than the heterotrimer α1(I)2α2(I) collagen found in normal or OI bone.(23) We present here coordinated growth, areal and volumetric bone density, μQCT, four-point bending to failure, and static and dynamic histomorphometry obtained from Aln-treated Brtl mice. Brtl is a knock-in murine model for dominantly inherited moderately severe OI, which synthesizes standard heterotrimeric type I collagen and has a classical OI-causing glycine substitution (Gly349Cys) in one col1a1 allele.(24) Brtl shows phenotypic features of type IV OI, including small size, decreased matrix synthesis,(25) low BMD, and spontaneous fractures during the first few weeks of life. Brtl femora have weak geometry and brittle failure properties.(26) We treated Brtl mice and wildtype littermates for 6 or 12 wk to determine the effects of bisphosphonate treatment and the time course of their appearance.



Male Brtl(24) mice, maintained on a mixed background of Sv129/CD-1/C57BL/6S, were mated to female CD-1 mice (Charles River). Pups were genotyped by PCR using DNA from tail clippings of 21-day-old mice.(25) Only male offspring of the Brtl × wildtype CD-1 matings were used for these studies. Regular rodent chow and water were available to the mice ad libitum. All experiments were performed with approval from the NICHD ACUC committee.

Aln treatment

Male Brtl mice and their wildtype littermates were treated with Aln and compared with untreated littermates. Treated mice were injected subcutaneously with 0.219 mg/kg/wk of pharmacological grade Aln (generously provided by Merck) in PBS from 2 to 14 wk (skeletal maturity) or 2 to 8 wk (puberty) of age. The Aln dose was chosen to match the treatment of oim mice(27) so that the effects of Aln in different murine OI models could be directly compared. The oim dose was based on the lowest dose that increased bone mineralization without impeding resorption in a pilot study.(28) Control mice were initially injected subcutaneously each week with PBS as placebo controls or received no injections. Mice were killed by lethal injection at 8 or 14 wk of age, respectively. The mice treated for 12 wk underwent all the analyses conducted for this study. The mice treated for 6 wk were used for analyses of BMD and biomechanical studies to determine progression of bisphosphonate effects over the course of the 12-wk treatment.

Growth analysis and areal BMD

Treated and untreated Brtl and wildtype mice were weighed weekly for growth curves during the 12-wk treatment regimen. Areal BMD of Brtl and wildtype lumbar spines and femora were measured using a GE Lunar PIXImus2 (GE Healthcare) on the whole mouse specimen. An internal calibration standard provided by GE was run before each measurement group. After euthanasia, femora of mice in the 12-wk treatment group were excised and cleaned of soft tissue, leaving the growth plates intact. Femora were imaged from the proximal head to the distal end of the medial and lateral condyles by μCT (see below).

Evaluation of femoral geometry by μCT

Left femora were analyzed by μQCT (GE Healthcare) for both structural and mineral parameters. Scanned specimens were reconstructed at 18-μm voxel size. Regions of interest were isolated for cortical and trabecular parameters. A central cortical region was located at the midpoint of each femur, extending distally for 2.5 mm, and assessed for cross-sectional area (CSA), cortical thickness, bending moment of inertia across the medial-lateral axis, and shape factor. Shape factor (SF) refers to the ratio of bending moments in the medial-lateral to anterior-posterior directions and provides an evaluation of how round or flattened a structure is. An SF of 1 would indicate a circular cross-section, whereas an SF >1 indicates ellipsoid cross-section with a long axis in the medial-lateral direction. Alterations in SF value reflect changes in bone shape toward a more circular (approaching SF = 1) or more flattened (SF > 1) cross-section because of changes in bone mass distribution in the anterior-posterior or medial-lateral direction. Total femoral length was established, and a trabecular region consisting of a 1-mm-diameter cylinder was located just proximal to the distal femoral growth plate and extended 1/10th total femoral length in the proximal direction. Trabecular number and thickness and bone volume fraction were measured after application of a threshold algorithm.(29) Tissue-specific volumetric BMD (vBMD) was assessed for both cortical and trabecular regions. Regions of interest described above were assessed for BMC using calibrated tissue phantoms and normalized by total bone volume of each region of interest. Local differences in tissue vBMD are represented graphically by applying a pseudo-color map to grayscale voxel values of mineralization in Hounsfield units. Backscatter scanning electron microscopy (BSEM) was performed on a subset of specimens after μQCT to verify local variations in mineral density within distal femoral trabecular bone.

Whole bone mechanical properties

After μQCT, femora were loaded to failure in four-point bending at 0.5 mm/s in the anterior-posterior direction using a servohydraulic testing machine (810 Material Test System; Eden Prairie). Regions loaded in four-point bending corresponded to those measured by μCT. Displacement of loading points was measured using an external linear variable differential transducer (LVDT; Lucas Schavitts), while load data were collected simultaneously with a 50-lb load cell (Sensotec). Pre-yield effects of genotype and treatment were measured by yield load, yield displacement, and stiffness. Ultimate load and post-yield displacement were considered post-yield effects.

Predicted matrix properties

Four-point bending results reflect both structural and material properties of bone. To determine which effects were caused by material differences, results from four-point bending were normalized by structural parameters through beam theory.(30) These equations normalize four-point bending results by the bending moment of inertia of the central diaphysis of the bone, measured by μCT, to estimate inherent material properties of the extracellular matrix, independent of bone size and shape, assuming a bending mode of failure.

Bone histomorphometry

Femora from 2- to 14-wk treated Brtl (n = 20) and wildtype (n = 20) and 2- to 14-wk untreated Brtl (n = 19) and wildtype (n = 21) mice were analyzed. The right femur was dissected and fixed in 70% ethanol for 3–5 days. It was dehydrated with increasing concentrations of ethanol, cleared in xylene, embedded undecalcified in methyl methacrylate, and used for both static and dynamic histomorphometry. Five-micrometer-thick longitudinal serial sections were cut on a Reichert-Jung Polycut S microtome (Reichert-Jung) with a D profile knife (Delaware Diamond Knives). For static histomorphometry, sections taken from the middle of the femur, where the central vein is located, were stained with modified Masson trichrome stain. Measurements were made by a blinded observer using the OsteoMeasure image analysis system (OsteoMetrics), interfaced with an Optiphot Nikon microscope (Nikon). The recommendations of the ASBMR(31) Histomorphometry Nomenclature Committee were followed. All measurements were confined to the secondary spongiosa and restricted to an area between 400 and 2000 μm proximal to the growth plate-metaphyseal junction of the distal femur. Cortical measurements were made 4000 μm proximal to the same growth plate. For dynamic histomorphometry, mice received intraperitoneal injections of calcein (Sigma), 10 mg/kg in 2% NaHCO3, and xylenol orange (Sigma), 90 mg/kg, 12 and 2 days before death, respectively.

Statistical analysis

μCT and four-point bending data were statistically analyzed using multivariate general linear models for each group, with significance attributed to p < 0.05. Least significant difference posthoc tests were used to assess genotype differences when overall model significance was ≤0.05. Areal BMD and bone histomorphometry data were analyzed using Student's t-test; significance was p < 0.05. Data are presented as mean ± SD.


Aln treatment does not alter Brtl growth curve or femoral length

All four sets of mice in the 12-wk treatment study were weighed weekly. The resulting growth curves (Fig. 1) showed that neither wildtype nor Brtl genotypes experienced a change in their growth with Aln treatment. Specifically, Aln treatment did not relieve the smaller size of Brtl mice with respect to wildtype littermates. Femur lengths measured after death after 12 wk of treatment are shown (Fig. 1, inset). Neither genotype experienced a significant change in femoral length with Aln (wildtype: untreated, 32.07 ± 0.92 mm versus treated, 31.88 ± 0.74 mm, p = 0.48; Brtl: untreated, 30.35 ± 1.19 mm versus treated, 30.09 ± 1.00 mm, p = 0.49), showing that Aln had no beneficial effect on growth of OI long bone.

Figure Figure 1.

Growth curves of treated and untreated Brtl and wildtype littermates from 4 to 14 wk of age. Aln treatment did not change the body weight or femoral length of either genotype of mice. Brtl remains significantly smaller than wildtype. Weights plotted as mean ± SD *p < 0.05. (Inset) Femur length is unaffected by 12 wk of Aln treatment. #p < 0.05 between untreated wildtype and treated Brtl; +p < 0.05 between untreated wildtype and Brtl.

Aln affects areal BMD and vBMD of Brtl and wildtype femur and spine

Vertebral (L1–L4) areal BMD is frequently reported as a measure of response to bisphosphonate in children with OI.(16–19) In our murine study groups, we determined areal femoral and lumbar vertebral BMD (aBMD) using a small mammal densitometer (PIXImus) and vBMD of the femoral cortex and trabeculae using μQCT (Table 1). Untreated Brtl spine and femur have significantly lower aBMD than those of wildtype littermates by 7–16% and 16–17%, respectively. aBMD of Brtl spines and femora was unchanged after 6 wk of treatment, remaining significantly lower than untreated wildtype samples but increased 30% (spine) and 26% (femur) after 12 wk of Aln. The wildtype control groups increased aBMD with both treatment durations (22% and 20%, and 27% and 20%, respectively, for spine and femur for the 6- and 12-wk treatment sets, respectively). After 12 wk of treatment, Brtl spine and femoral aBMD were significantly greater than those of untreated wildtype mice.

Table Table 1.. BMD
original image

Untreated Brtl had femoral cortical vBMD 3–4% greater than untreated wildtype, consistent with the hypermineralization of pediatric OI cases noted on nanoindentation measurements.(15) The cortical vBMD of both genotypes was unchanged after 6 wk of treatment but was increased 8% after 12 wk of Aln. Treated Brtl cortical vBMD was 5% and 11% greater than in untreated wildtype after 6 and 12 wk of Aln, respectively.

Trabecular vBMD did not differ significantly between genotypes in untreated mice. Treated Brtl trabecular vBMD was decreased 17% and 38% in the 6- and 12-wk treatment sets compared with untreated Brtl. Similarly, trabecular vBMD of treated wildtype mice decreased progressively 12% and 25% compared with untreated wildtype mice in the 6- and 12-wk treatment sets, respectively. Treated Brtl mice had trabecular vBMD 13% (6-wk treatment set) and 24% (12-wk treatment set) lower than those of untreated wildtype.

Aln increases both cortical and trabecular bone parameters in Brtl and wildtype mice

After both 6- and 12-wk treatment durations, the cortical thickness of treated Brtl and wildtype femora were each increased compared with untreated mice of their genotype (Figs. 2 and 3; Table 2). Most of the >20% increase in cortical thickness of treated Brtl mice accomplished in 12 wk occurred during the first 6 wk of Aln treatment, as did essentially all of the >50% increase in cortical thickness of 12-wk treated wildtype. Treated Brtl mice attained the cortical thickness of control wildtype mice. The CSA of the Brtl diaphysis was also significantly increased after 12 wk of treatment. In part, this reflects a change in the shape of the diaphysis (Fig. 2). Untreated Brtl femoral diaphyses have a more elliptical profile than the rounded wildtype shape (1.80 ± 0.22, wildtype; 2.04 ± 0.22, Brtl; p < 0.05). The wildtype shape factor does not change with Aln treatment, but Aln restores a more normal shape to the Brtl shaft after 12 wk (1.81 ± 0.25, p < 0.05 versus untreated). The shape change is achieved primarily by adding periosteal bone to enlarge the outer diameter. Based on the composite geometric changes, there is a significant increase in the bending moment of inertia, a geometric measure of the resistance of bone to bending, after 12 wk of Aln.

Table Table 2.. Mechanics
original image
Figure Figure 2.

Representative μCT images of femora from the 12-wk treatment group scanned at 18-μm voxel resolution to obtain average geometric parameters for cortical (A) and trabecular (C) regions. (B) Representative femoral cross-sectional reconstruction of the 12-wk Aln treatment group shows increased CSA and cortical thickness in both wildtype and Brtl. The shape factor (SF), a ratio of anterior-posterior to medial-lateral bending moments, was also significantly changed in the Brtl mouse after the 12-wk Aln treatment from a more flattened geometry to a more normal, rounded geometry. (D) Representative images of the dramatic increase in femoral trabecular bone volume per total volume in both Brtl and wildtype treated mice from the 12-wk Aln treatment group.

Figure Figure 3.

Geometric parameters of femoral cortical and trabecular regions after 6 and 12 wk of Aln treatment, scanned by μCT (A-D). Cortical thickness (A) and trabecular bone volume/total volume (B) are significantly increased by treatment in both wildtype and Brtl femora. The increased trabecular bone volume was caused by a dramatic increase in trabecular number (C), whereas trabecular thickness (D) is unchanged. Biomechanical parameters from femora loaded in four-point bending to failure (E and F). Aln treatment for 12 wk increases stiffness of wildtype femora (E), but Brtl stiffness was unchanged. The ultimate load, or load at which the femur fractured, was significantly increased in both wildtype and Brtl mice after 12 wk of Aln treatment (F). Data are represented as mean ± SD. *p < 0.05 between untreated and treated; #p < 0.05 between untreated wildtype and treated Brtl; +p < 0.05 between untreated wildtype and Brtl.

As expected, distal femoral bone volume fraction was sharply increased by Aln (Fig. 3; Table 2). Wildtype BV/TV doubled and Brtl BV/TV tripled after 6 wk of treatment; no further increases were found in the 12-wk treatment group. As expected, bone volume increases were caused by dramatically increased trabecular number, whereas trabecular thickness was unchanged (Figs. 2 and 3; Table 2). With both treatment durations, treated Brtl trabecular number was double untreated wildtype. Interestingly, the TbN of treated Brtl and treated wildtype mice did not differ significantly after either treatment duration, although untreated Brtl TbN is significantly lower than untreated wildtype (Table 2), suggesting that TbN had reached a “biological maximum” in both genotypes.

Aln treatment increases femoral stiffness and load to fracture

Biomechanical testing of femora in four-point loading provided insight into the changes effected by Aln treatment along the loading curve (Fig. 3; Table 2). Untreated Brtl femoral stiffness was ∼20–25% lower than in wildtype and did not increase significantly with Aln (p = 0.23) or attain wildtype untreated levels. The stiffness of wildtype femora was increased 21% compared with untreated wildtype (p = 0.003) after 12 wk of Aln.

There were no significant changes in yield load, the load at which the femur could no longer recover from deformation, in either genotype. The ultimate load, or the force at which the femur fractured, was 23–36% lower in Brtl at baseline. Twelve weeks of Aln treatment were required for significant change in Brtl (14% increase) and wildtype (13% increase) ultimate load, although treated Brtl remained significantly lower than untreated wildtype load to fracture.

Aln does not improve femoral brittleness and decreases predicted material strength

In OI patients and in the Brtl mouse, OI bone is brittle, that is, it breaks after it is bent a shorter distance beyond the yield point than does normal bone. The brittleness (post-yield displacement) of the Brtl mouse femur is not improved after either 6 or 12 wk of Aln administration (Fig. 4; Table 2). Although, on average, wildtype bone becomes more brittle after treatment, the difference is not statistically significant.

Figure Figure 4.

Material properties of Aln-treated femora. (A) Untreated Brtl femora have a significantly lower post-yield displacement, a measure of bone brittleness, than wildtype femora, but do not improve with Aln treatment. Aln treatment reduced WT post-yield displacement. (B) Predicted material strength calculations combine four-point bending results with bone geometry. Predicted material strength of both wildtype and Brtl femora was significantly reduced after 12 wk of Aln treatment. Data are represented as mean ± SD. *p < 0.05 between untreated and treated; #p < 0.05 between untreated wildtype and treated Brtl; +p < 0.05 between untreated wildtype and Brtl.

The predicted strength and elastic modulus of the bone material itself can be calculated from the combined geometry and breaking load of the femora. Although Brtl bone is geometrically smaller than wildtype bone, the predicted strength of untreated Brtl and wildtype bone material is equivalent (Fig. 4; Table 2). However, after 12 wk of Aln, Brtl and wildtype predicted strengths are reduced 13% and 21%, respectively, compared with untreated mice of the same genotype. Similar reductions occur in predicted elastic modulus (Table 2).

Osteoblast surface and matrix production decreased by Aln treatment

Static histomorphometry of femora treated 12 wk confirms that Aln treatment increases bone volume because of increased trabecular number in both Brtl and wildtype mice, as delineated by μQCT (Table 3). Interestingly, given the well-documented inhibition of osteoclast function by bisphosphonates, the osteoclast surface (OcS/BS) is not altered by treatment in either genotype (Fig. 5). Instead, we found an effect on osteoblast surface and matrix production. Osteoblast surface (ObS/BS) is significantly decreased in treated wildtype mice compared with untreated wildtype, although the decrease of ObS/BS in Brtl is not significant. Furthermore, treatment significantly diminishes the mineral apposition rate (MAR), a surrogate for osteoblast matrix production, by >50% and dramatically depresses the bone formation rate >85% in both genotypes (Table 3). We noted that after 12 wk of Aln treatment, the treated osteoblasts have altered morphology (Fig. 5A), with decreased height compared with those of untreated mice and fewer plump, contiguous osteoblasts lining the bone surface compared with those of untreated mice. These qualitative observations correlated with the decrease in MAR.

Table Table 3.. Histomorphometry
original image
Figure Figure 5.

Histomorphometry of cellular elements from the 12-wk treatment group. (A) Representative images of trabecular bone from Masson trichrome-stained treated and untreated Brtl and wildtype femoral sections. Osteoblast morphology is altered in both genotypes after 12 wk of Aln treatment. Osteoblasts lining trabeculae in untreated wildtype mice have tall, plump cuboidal morphology, whereas those in treated wildtype mice are more rounded and rectangular in shape. Osteoblasts lining trabeculae in untreated Brtl mice are irregular in shape and are often noncontiguous; osteoblasts in treated Brtl mice have a flattened morphology, like lining cells. Bar scale is 25 μm. (B) Bar graph of osteoblast (ObS/BS) surface. Aln treatment for 12 wk had no effect on Brtl ObS/BS but significantly decreased wildtype ObS/BS. (C) Bar graph of osteoclast (OcS/BS) surface. Twelve weeks of Aln had no effect on OcS/BS in wildtype or Brtl mice. Data are represented as mean ± SD. *p < 0.05 between untreated and treated; #p < 0.05 between untreated wildtype and treated Brtl; +p < 0.05 between untreated wildtype and Brtl.

Bisphosphonate-treated Brtl and wildtype femora have retained mineralized cartilage

Both Brtl and wildtype treated femora have retention of mineralized cartilage (Fig. 6A). Chondrocytes are embedded in the light green-stained material at the growth plate. Also, the trabeculae adjacent to the growth plate contain a cartilaginous core; this finding is more pronounced in Aln-treated bones. In the metaphysis and diaphysis, there is persistence of trabecular cores stained pale green, which do not contain osteocytes or chondrocyte lacunae and are surrounded by darkly stained mineralized bone with osteocytes. Calibrated μQCT scans of the distal femoral trabecular bone regions show focal areas of intense mineralization after Aln treatment, in regions consistent with retained calcified cartilage, as seen histologically (Fig. 6B). BSEM images of distal femoral trabecular bone show trabecular bone structures are dominated by hypermineralized retained cartilage in treated specimens compared with smaller fractions of areas containing retained cartilage in untreated specimens (Fig. 6C).

Figure Figure 6.

Retention of mineralized cartilage in Aln-treated femora. (A) Representative images of Masson trichrome-stained treated and untreated wildtype and Brtl femora from the 12-wk treatment group. Boxes indicate portions of images for which enlargements are shown in B. Scale bar is 500 μm. (B) Enlargements of femoral sections show that both Brtl and wildtype treated femora retained mineralized cartilage, seen as pale green cores within the trabecular bone in the insets. Retention of mineralized cartilage is greater in the Aln-treated Brtl than wildtype bones. Scale bar is 225 μm. (C) Representative μQCT cross-sections of distal femoral trabeculae show local regions of highly mineralized material after treatment with Aln. Grayscale voxel values are pseudo-colored to represent calibrated CT attenuation values (Hounsfield units: purple, low mineralization; red, high mineralization) (D) BSEM micrographs of trabecular bone show trabecular bone in Aln-treated animals is dominated by hypermineralized retained cartilage compared with untreated controls.


Bisphosphonate treatment was embraced by the OI patient community before animal testing and controlled trials were completed because there was a lack of alternative treatment options. Many children with OI have now received >5 yr of cyclic pamidronate, some beginning treatment in infancy. This controlled study of Aln in the Brtl mouse model for classical OI focuses on the effects on long bone, which is inaccessible for direct studies in children. Mice were treated from 2 (before weaning) to 8 or 14 wk (full puberty or young adulthood, respectively). Both Brtl and wildtype long bone have similar responses to treatment.

Observational(5,8,9) or controlled pediatric trials(16–19) with bisphosphonates have reported no adverse effects on OI long bone growth, as well as persistence of OI syndromic short stature. We found no change in Brtl or wildtype femoral length or weight-based growth curves. Similarly, oim mice did not experience decreased femoral length until they were administered about double the dose used in this study.(27,32,33) At a 10-fold higher Aln dose, the humerii and ulna of oim were shortened and also the height of the growth plate hypertrophic zone was increased, suggesting failure of vascular invasion at the chondro-osseus junction.(33)

aBMD of lumbar vertebrae is a standard endpoint for pediatric bisphosphonate studies.(16–19) Aln increased Brtl aBMD after 12 wk of treatment and wildtype femoral and spine BMD after 6 wk. In oim, continuous Aln treatment did not increase cortical BMD.(32) In both Brtl and wildtype, femoral cortical vBMD increased, whereas trabecular vBMD decreased. However, this technique yields an average measurement, which does not accurately reflect the islands of highly mineralized cartilage retained in trabeculae after treatment and visualized on BSEM. OI children treated with pamidronate for 1–1.5 yr have unchanged spine qCT despite increased vertebral DXA Z-scores.(19)

Aln causes rapid changes in cortical thickness and trabecular bone volume of Brtl and wildtype femora, similar to the increased cortical width and bone volume seen in paired iliac crest biopsies in children with OI.(14) The oim Aln trial yielded increased bone volume, but not cortical thickness, after 12 wk of treatment.(27) Changes in mechanical properties of Brtl and wildtype femora, including load to fracture in both genotypes and stiffness in wildtype, required 12 wk of Aln treatment. Comparable whole femoral studies in oim yielded no increase in failure load or breaking strength after Aln.(27)

Aln does not improve the intrinsic brittleness of Brtl bone, just as brittleness was unchanged in Aln-treated oim.(34) In fact, in the wildtype mice in both Brtl and oim studies, Aln treatment yielded a trend toward (p = 0.083) or significantly increased (28%) femoral brittleness,(34) respectively. Furthermore, the predicted material properties of Brtl and wildtype bone decreased significantly, implying that Aln treatment lowers bone quality. Potential contributors to the decline in material properties include changes in bone architecture, such as disorganized woven bone or accumulated microdamage. Microdamage caused reduced vertebral toughness and energy absorption capacity in beagles treated with bisphosphonates for 1 yr.(35) However, it is not clear that mice remodel bone or accumulate microdamage.

The increased cortical vBMD detected in both genotypes after Aln treatment suggests that subtle but functionally important changes in cortex mineralization may also contribute to lower bone quality despite the additional mineralized tissue. For example, changes in the spatial distribution of mineral after treatment may be especially detrimental in a setting of accumulated microdamage, because it may tend to facilitate the propagation of microcracks.

Prior studies explored the effect of bisphosphonate on bone mineralization. Retention of mineralized cartilage was detected in growing mice(27,32–34) and children(14,21,36) treated with bisphosphonate. Pediatric studies found unchanged bone hardness or elastic indentation modulus of OI bone after 2.5 yr (range, 1.2–3.4 yr) of pamidronate treatment.(15) However, lower bone quality is apparently a risk of longer treatment in both children(21,37) and mice, because we found these changes after 12-wk treatment (human equivalent, treatment from toddler to young adult) but not after 6-wk treatment (human equivalent, treatment from toddler to prepubertal child). Mineral changes in oim were genotype dependent.(34,38) The oim/oim femora have atypical type I collagen homotrimers [α1(I)3]; they showed no changes in cortical strength, stiffness, or material properties after Aln, along with unchanged cortical mineral:matrix and crystallinity. In contrast, their wildtype controls had increased strength and brittleness, with reduced heterogeneity of cortical mineralization. The vertebrae of normal beagles treated with Aln for 1 yr(39) shift to higher BMD by density fractionation but did not have differences in crystal maturity. This study showed increased cortical vBMD after Aln in Brtl and wildtype femora, which both contain type I collagen heterotrimers and suggests that altered mineralization lowers bone quality in OI and normal bone. Further study is needed to determine the functional significance of increased cortical vBMD.

In this study, histomorphometry was done after 12 wk of Aln treatment. At that time point, the dramatic increase in trabecular number has already occurred and suppression of bone formation was more notable than decreased bone resorption. Although some decrease in osteoblast function is expected from coupling to bisphosphonate-suppressed osteoclasts, the substantial decreases detected in MAR and BFR in both genotypes suggest an additional direct effect on osteoblasts. These data are consistent with pediatric and animal studies. Two years of pediatric pamidronate treatment decreased OcS by >40% and increased the number of large Oc 4-fold; an overall greater decrease in bone formation than resorption was indicated by decreased osteoblast surface and a decline in BFR to one third of pretreatment levels.(14,36) Histomorphometry of Aln-treated beagles showed OcS/BS decreased by one half, as well as BFR/BS decreased by >90% of untreated values.(40) In rats, Aln significantly suppressed osteoblast activity on both periosteal and endocortical surfaces and a direct effect of bisphosphonates on osteoblast activity was postulated. The rats treated with Aln had significant reduction in periosteal MAR, a surface that only undergoes bone formation, supporting an interpretation of independent suppression of bone formation by a direct effect on individual osteoblasts.(41) In our study, we found that ObS/BS was decreased significantly in treated wildtype mice but not in Brtl. BFR was dramatically decreased in both genotypes. OcS/BS was not affected, although retention of mineralized cartilage indicated inhibited osteoclast function. We also detected a change in osteoblast morphology with Aln treatment. Wildtype osteoblasts were more rounded and rectangular after Aln treatment compared with the standard tall plump cuboidal cells seen before treatment. Untreated Brtl osteoblasts have an irregular appearance compared with the wildtype mice and seem to be flattened by Aln treatment. Flattened osteoblast morphology was previously reported in risedronate-treated osteoprotegerin (OPG)-deficient mice to a shape associated with quiescent or lining cells.(42)

We examined the progression of changes caused by bisphosphonate treatment to determine whether detrimental changes occur after beneficial changes and could be avoided simply by shorter treatment duration. Although changes in Brtl femoral cortical thickness, trabecular number and bone volume seem to have attained nearly their full increment over baseline in 6 wk, the changes that contribute to increased load to fracture and moment of inertia require longer duration treatment. Unfortunately, by the time there are positive mechanical changes, the predicted material strength of the bone has significantly declined, as has BFR, concomitant with altered osteoblast morphology. Not surprisingly, Aln affects wildtype bone similarly. After 12 wk, cortical vBMD and load to fracture were increased, without significant change in femoral shape or moment of inertia. However, the increase in load to fracture came at the cost of a trend toward increased bone brittleness and a greater decline in bone material predicted strength than seen in the Brtl femora.

Aln treatment of the Brtl mouse allowed direct examination of its effect on long bone (femora) with classical OI caused by a collagen structural defect. Because bisphosphonates do not selectively affect the secretion of mutant collagen, the logic of the treatment is that the inhibitory effect of bisphosphonates on bone resorption will be beneficial per se at the whole bone level. The controlled pediatric OI bisphosphonate trials have yielded equivocal effects on long bone fractures, and there are anecdotal accounts of increased fracture rates in children treated for many years. In Brtl, an increase in femoral load to fracture after Aln treatment is concomitant with a decline in bone material properties, decreased strength and elastic modulus, and decreased matrix production (MAR) by osteoblasts. The mouse data thus indicates that the new bone is of inferior quality and also raises concerns about the ability of treated growing bone to recover normal cellular functions after long-term treatment. Whereas further studies seek the cause of the decreased quality of treated bone material, it is prudent to limit the cumulative dose of bisphosphonates given to OI children with collagen structural defects to minimize detrimental effects on bone. This can be done by increasing cycle interval; we are currently completing a comparison of 3- versus 6-mo pamidronate cycle intervals in children with types III and IV OI. Cumulative dose can also be limited by capping treatment after 2–3 yr when aBMD plateaus.(19)


The authors thank Aileen Barnes, MS (BEMB, NICHD), for assistance with the microscopy and photography of bone sections for Fig. 5. This work was supported by NICHD Intramural Funding (J.C.M.), NIH AR46024 (S.G.), and The Center for Histology and Histomorphometry at the University of Connecticut (G.A.G.).