Bone geometry, density, and microarchitecture in the distal radius and tibia in adults with osteogenesis imperfecta type I assessed by high-resolution pQCT

Authors

  • Lars Folkestad,

    Corresponding author
    1. Department of Endocrinology, Odense University Hospital, Odense, Denmark
    2. Department of Endocrinology, Hospital of Southwest Denmark, Esbjerg, Denmark
    3. Institute of Clinical Research, University of Southern Denmark, Odense, Denmark
    • Osteoporosis Clinic, Department of Endocrinology M, Odense University Hospital, Kloevervaenget 6, 5000 Odense C, Denmark.
    Search for more papers by this author
  • Jannie Dahl Hald,

    1. Department of Endocrinology and Internal Medicine, Aarhus University Hospital, Aarhus, Denmark
    Search for more papers by this author
  • Stinus Hansen,

    1. Department of Endocrinology, Odense University Hospital, Odense, Denmark
    2. Institute of Clinical Research, University of Southern Denmark, Odense, Denmark
    Search for more papers by this author
  • Jeppe Gram,

    1. Department of Endocrinology, Hospital of Southwest Denmark, Esbjerg, Denmark
    Search for more papers by this author
  • Bente Langdahl,

    1. Department of Endocrinology and Internal Medicine, Aarhus University Hospital, Aarhus, Denmark
    Search for more papers by this author
  • Bo Abrahamsen,

    1. Institute of Clinical Research, University of Southern Denmark, Odense, Denmark
    2. Department of Internal Medicine, Gentofte Hospital, Copenhagen, Denmark
    Search for more papers by this author
  • Kim Brixen

    1. Department of Endocrinology, Odense University Hospital, Odense, Denmark
    2. Institute of Clinical Research, University of Southern Denmark, Odense, Denmark
    Search for more papers by this author

Abstract

Osteogenesis imperfecta (OI) is a hereditary disorder characterized by decreased biosynthesis or impaired morphology of type I collagen that leads to decreased bone mass and increased bone fragility. We hypothesized that patients with OI have altered bone microstructure and bone geometry. In this cross-sectional study we compared patients with type I OI to age- and gender-matched healthy controls. A total of 39 (13 men and 26 women) patients with OI, aged 53 (range, 21–77) years, and 39 controls, aged 53 (range, 21–77) years, were included in the study. Twenty-seven of the patients had been treated with bisphosphonates. High-resolution peripheral quantitative computed tomography (HR-pQCT) at the distal radius and distal tibia and dual-energy X-ray absorptiometry of total hip, femoral neck, trochanteric region, and the lumbar spine (L1–L4) were performed. The patients were shorter than the controls (159 ± 10 cm versus 170 ± 9 cm, p < 0.001), but had similar body weight. In OI, areal bone mineral density (aBMD) was 8% lower at the hip (p < 0.05) and 13% lower at the spine (p < 0.001) compared with controls. The trabecular volumetric bone mineral density (vBMD) was 28% lower in radius (p < 0.001) and 38% lower in tibia (p < 0.001) in OI compared with controls. At radius, total bone area was 5% lower in OI than in controls (p < 0.05). In the tibia, cortical bone area was 18% lower in OI (p < 0.001). In both radius and tibia the number of trabeculae was lower in patients compared to the controls (35% and 38%, respectively, p < 0.001 at both sites). Furthermore, trabecular spacing was 55% higher in OI in both tibia and radius (p < 0.001 at both sites) when compared with controls. We conclude that patients with type I OI have lower aBMD, vBMD, bone area, and trabecular number when compared with healthy age- and gender-matched controls. © 2012 American Society for Bone and Mineral Research.

Introduction

Osteogenesis imperfecta (OI) is a hereditary disorder characterized by decreased biosynthesis or impaired structure of type I collagen that leads to decreased bone mass and increased bone fragility.1–3 The most common symptom of OI is fracture after minimal or no trauma. Patients with OI often have bone deformities and short stature, but the clinical severity varies. OI is divided into 11 different subtypes according to clinical features and genetics.4 Mutations in the COL1a1 and COL1a2 genes are identified in 80% to 90% of patients with type I OI.5 The phenotype of type I OI is heterogeneous and spans from patients with few fractures, normal body height, and normal bone mineral density (BMD) to patients with multiple fractures, low body height, and very low BMD.6 Some patients with OI have blue or bluish-gray sclerae.2 The population prevalence of OI in Denmark is estimated to be 10.6:100,000 with type I being the most prevalent (71% of all cases).7

Bone strength is determined by bone geometry, bone microstructure, bone mass, and the quality of the bone matrix.8 In children and adults with OI, BMD is lower than in healthy controls.9–11 In OI, bone stiffness is increased due to areas of hypermineralization.12 The bones are thus “brittle” and absorb less energy before fracturing.13 The bone microstructure evaluated by histomorphometry on iliac bone biopsies or pQCT is altered, with fewer and thinner trabeculae and increased cortical porosity.12, 14 The refinement of bone imaging technologies in recent years has improved the assessment of bone microarchitecture. Measures of bone microarchitecture, bone geometry, and volumetric BMD (vBMD) can now be obtained using high-resolution pQCT (HR-pQCT).15

Whereas bone mass in patients with OI is well-described,3, 4, 16, 17 other determinants of bone strength are less explored. Findings from studies on bone structure and geometry using traditional histomorphometric analysis of bone biopsies in adults with OI need to be corroborated by studies using more advanced stereological techniques such as HR-pQCT. Moreover, HR-pQCT allows multiple assessments in the same patient. The objective of this study was to investigate the differences in bone geometry, density, and microarchitecture in patients with OI compared to age- and gender-matched healthy controls with HR-pQCT. We hypothesize that patients with type I OI have altered microarchitecture and bone geometry.

Materials and Methods

A total of 159 adults with suspected OI were identified through our clinical database and the membership registry of the Danish Osteogenesis Imperfecta Society (DFOI) and were invited by mail to participate in the study. Patients were included if they had clinical type I OI and were above 18 years of age. The patients were only included in the study: (1) if patients either had a family history of OI and blue sclerae or dentinogenesis imperfecta; or (2) if patients had experienced at least one fracture and had either blue sclerae or dentinogenesis imperfecta. Exclusion criteria were other types of OI as assessed clinically, uncertain diagnosis, other metabolic bone disease, and residence in the Northern Region of Denmark (due to extended transport time to the investigation site). Of the 75 patients who replied by mail, 16 declined participation and a further 20 fulfilled one or more exclusion criteria (Fig. 1). The remaining 39 patients (13 males and 26 females) with clinically diagnosed type I OI entered the study.

Figure 1.

Flow diagram showing how the patients and controls were selected for inclusion in the study.

Age- and gender-matched controls were selected from an ongoing study on bone microarchitecture in normal healthy volunteers (Regional Ethics Committee S-20090069) (unpublished results). In this reference study, a random sample of adults drawn from the Civil Registration System in the municipality of Odense, Denmark, were invited by mail to participate in the study. Exclusion criteria were treatment for or diagnosis with conditions known to alter bone metabolism (such as bisphosphonate treatment, overt thyrotoxicosis). At the time of the current study, 237 women and 225 men, aged 18 to 80 years of age, had participated in the reference study. From this cohort, 39 gender- and age-matched controls were selected at random prior to any analysis of their data.

All participants gave informed consent, and the study was performed according to the Declaration of Helsinki. The Regional Ethics Committee approved the study (Regional Ethics Committee S-20100049).

Dual-Energy X-Ray Absorptiometry (DXA)

Areal BMD (aBMD) at the lumbar spine (aBMDspine), hip (trochanteric region (aBMDtroch), femoral neck (aBMDneck), and total hip (aBMDhip) were measured using dual-energy X-ray absorptiometry (DXA) (Hologic Discovery, Waltham, MA, USA). In our unit, the coefficient of variation (CV) for measurements of aBMDspine and aBMDhip is 1.5%.

HR-pQCT

We assessed bone geometry, volumetric BMD (vBMD), and microarchitecture of the nondominant distal radius and distal tibia (or in case of a previous fracture the nonfractured limb) using an HR-pQCT system (Xtreme CT; Scanco Medical AG, Brüttisellen, Switzerland). Image acquisition and analysis were performed as described.18 In short, we applied the manufacturer's default protocol for in vivo patient scanning with provision of a 3D-representation of 9.02 mm of the distal radius and distal tibia. A fixed offset of 9.5 mm and 22.5 mm from the endplate to the first slide were applied at radius and tibia, respectively. We measured total vBMD (vBMDtot), cortical vBMD (vBMDCt), and trabecular vBMD (vBMDTb). Trabecular bone volume fraction (bone volume/tissue volume = BV/TV) was derived from vBMD in the trabecular compartment. Cancellous architecture was assessed using a direct 3D measure of trabecular number (Tb.N) and its distribution (SD.1/Tb.N), whereas trabecular thickness (Tb.Th) and spacing (Tb.Sp) were derived from BV/TV and Tb.N. Measures of geometry (total, trabecular, and cortical) were derived from the respective volumes and length of the scan using an annular approach. We used a two-threshold input technique to extract the endosteal and periosteal edges of the cortex,19 which provided a 3D measure of cortical thickness (Ct.Th) and cortical porosity (Ct.Po) measured as void cortical volume relative to total cortical volume, as described.20, 21 Due to the voxel resolution of the HR-pQCT technology, pores with a diameter smaller than 82 µm cannot be identified using HR-pQCT.

In our unit, the intraindividual CVs for geometrical parameters were 0.2% to 1.8% for total bone area, cortical area, and trabecular area, and 0.4% to 0.9% for vBMDtot, vBMDCt, and vBMDTb. The intraindividual CVs for microarchitectural parameters (Ct.th, Ct.Po, Tb.N, Tb.Th, and Tb.Sp were 0.6% to 7.2%).22

Anthropometry and medical history

We measured body weight to the nearest 0.1 kg on a Seca Model 708 scale (Seca, Hamburg, Germany) with participants dressed in indoor clothing without shoes. Body height was measured to the nearest 0.1 cm using a wall-mounted Harpeden stadiometer (Holtain Ltd, Crymych, UK). Data on bisphosphonate use, menopausal status, and fractures were obtained through a self-administered questionnaire and a structured medical interview. Cumulative dose was estimated as years on bisphosphonates regardless of specific drug, administration (oral or intravenous [i.v.]) and regime (eg, weekly, quarterly, etc.).

Statistical analysis

Data are presented as mean ± SD, median [interquartile range] or median (range) as appropriate. We compared the matched groups using Wilcoxon matched-pairs signed-rank test, paired t test, or chi-squared test as appropriate. Values of p were two-sided, and the statistical significance level was set at 0.05. We assessed the distribution of each parameter via normality plots and Shapiro-Wilks W test. To identify the parameters that best distinguish patients with OI from the normal controls, we performed logistic regression analyses using a dummy variable (OI yes/no) as the dependent variable and the measured parameters as predictor variables. Pearson's goodness-of-fit and the Akaike information criterion (AIC) were then calculated. The model with the best fit and lowest AIC was selected as identifying the variable that best distinguished the OI and control groups. All analyses were done based on assessment of prestudy planed primary parameters, and using Stata Statistical Software Release 11.0 (StataCorp LP, College Station, TX, USA). Using the Bonferroni correction by dividing the alpha of 0.05 with the number of variables for bone mass, geometry, and microarchitecture (n = 44), we obtain a significance level of 0.0011.

Results

Anthropometrics

Patients with OI had a median age of 53 (21–77) years, similar to the controls (Table 1). Patients with OI were significantly shorter than the controls, but did not differ in body weight. There were no between-group differences in the menopausal state or the median age at menopause. The majority of the patients had received bisphosphonates. Thirty-three of the patients had a family history of OI. Thirteen of the patients had been diagnosed with dentinogenesis imperfecta. All had blue sclera and had experienced multiple fractures, including at least one long-bone fracture.

Table 1. Participant Characteristics
CharacteristicOI type I (n = 39)Healthy controls (n = 39)pa
  • BSA = body surface area; BMI = body mass index; i.v. = intravenous; IQR = interquartile range; PTH = parathyroid hormone; HRT = hormone replacement therapy; OI = osteogenesis imperfecta.

  • a

    Between-group differences were tested using the Wilcoxon matched-pairs signed-rank test (w), t test (t), or (x) chi-squared test.

Age, years (range)53 (21–77)54 (21–77)0.07w
Sex (male/female)13/2613/26
Height (cm)159.2 ± 10.3170.1 ± 9.4<0.001t
Weight (kg)73.1 ± 17.376.6 ± 15.50.25t
BSA (m2)1.79 ± 0.261.89 ± 3.88<0.05t
BMI (kg/m2)28.6 ± 5.726.5 ± 4.90.07t
Treatment with bisphosphonates, n (%)27 (70%)0 (0%)
Current oral bisphosphonate users, n (%)16 (41%)0 (0%)
Current i.v. bisphosphonate users, n (%)11 (28%)0 (0%)
Cumulative years on bisphosphonates, years [IQR]5 [0–7]0 [0–0]
Treatment with PTH, n (%)3 (8%)0 (0%)
Current PTH users, n (%)1(3%)0 (0%)
Treatment with HRT, n (%)3 (8%)0 (0%)
Number of previous fractures, median [IQR]15 [10–30]0 [0–1]<0.001w
Dentinogenesis imperfecta, n (%)13 (33%)0 (%)
Family history of OI, n (%)33 (85%)0 (%)
Previous or current smoker, n (%)19 (49%)17 (43%)0.8x
Menopausal status (pre/post) (n)13/1312/140.8x
Mean age at menopause, years (range)46 (40–54)50 (40–57)0.18w

aBMD and vBMD assessed by DXA and HRpqCT

aBMD in the spine (p < 0.001), trochanter (p < 0.05), femoral neck (p < 0.001), and total hip (p < 0.001) was lower in the patients with OI compared with the controls (Table 2). In the radius, patients with OI had lower vBMDTb (p < 0.001), but not vBMDCt (p = 0.24) or vBMDtot (p = 0.13) (Table 2). In the tibia, patients had lower vBMDTb (p < 0.001) and vBMDtot (p < 0.001), but not vBMDCt (p = 0.35).

Table 2. Areal and Volumetric BMD as Measured by DXA and HR-pQCT, Respectively
 OI type I (n = 39)Healthy controls (n = 39)paAICbGOFb
  • Data are presented as mean ± SD or median [interquartile range].

  • BMD = bone mineral density; HR-pQCT = high-resolution peripheral quantitative computed tomography; OI = osteogenesis imperfecta; AIC = Akaike information criterion; GOF = Pearsons goodness-of-fit; vBMD = volumetric bone mineral density; mgHA = milligram hydroxyapatide.

  • a

    Between-group differences were tested using the Wilcoxon matched-pairs signed-rank test (w) or t test (t) for normally distributed variables.

  • b

    AIC and GOF are outcomes of the regression analyses performed to identify the parameters that best distinguish OI from normal. The lower the AIC and the higher the GOF, the better the parameter distinguishes between OI and normal.

Areal BMD
 Total spine (g/cm2)0.845 [0.751–0.939]0.969 [0.847–1.045]<0.001w87.972
 Total hip (g/cm2)0.851 [0.746–0.915]0.920 [0.806–1.019]<0.05w102.976
 Trochanteric BMD (g/cm2)0.584 ± 0.1040.686 ± 0.131<0.001t97.475
 Femoral neck BMD (g/cm2)0.671 ± 0.1090.752 ± 0.110<0.001t100.475
vBMD
 Ultradistal radius
  Total vBMD (mgHA/cm3)285 ± 75316 ± 830.13t109.273
  Cortical vBMD (mgHA/cm3)899 [859–932]885 [833–928]0.24w110.576
  Trabecular vBMD (mgHA/cm3)103 ± 41144 ± 42<0.001t95.476
 Ultradistal tibia
  Total vBMD (mgHA/cm3)215 ± 62278 ± 63<0.001t94.085
  Cortical vBMD (mgHA/cm3)869 [794–897]864 [784–892]0.35w111.678
  Trabecular vBMD (mgHA/cm3)101 ± 35164 ± 38<0.001t67.372

Geometry as assessed by HR-pQCT

In the radius, total bone area (p < 0.05) was lower in patients with OI compared with controls, but there were no differences in cortical (p = 0.06) or trabecular bone area (p = 0.07) (Table 3). In contrast, cortical bone area (p < 0.001) was lower in the tibia of patients with OI, but there were no differences in the total bone area (p = 0.23) or trabecular bone area (p = 0.59). Representative HR-pQCT images are shown in Fig. 2.

Table 3. Microstructural Parameters at the Ultradistal Radius and Tibia as Measured by HR-pQCT
 OI type I (n = 39)Healthy controls (n = 39)paAICbGOFb
  • Data are presented as mean ± SD or median [interquartile range].

  • HR-pQCT = high-resolution peripheral quantitative computed tomography; OI = osteogenesis imperfecta; AIC = Akaike information criterion; GOF = Pearsons goodness-of-fit; SD.1/Tb.N = trabecular network distribution.

  • a

    Between-group differences were tested using the Wilcoxon matched-pairs signed-rank test (w) or t test (t) for normally distributed variables.

  • b

    AIC and GOF are outcomes of the regression analyses performed to identify the parameters that best distinguish OI from normal. The lower the AIC and the higher the GOF, the better the parameter distinguishes between OI and normal.

Ultradistal radius
 Bone area
  Total bone area (cm2)256 [216–314]271 [225–336]<0.05110.676
  Cortical bone area (cm2)51.3 [44.1–60.5]54.7 [47.5–68.5]0.06w109.370
  Trabecular bone area (cm2)197 [164–251]208 [161–268]0.07w111.276
 Cortical parameters
  Cortical thickness (mm)0.88 [0.75–0.98]0.92 [0.7–1.03]0.32w110.248
  Cortical porosity (%)2.15 [1.36–3.27]1.70 [1.23–2.45]0.10w104.666
 Trabecular parameters
  Bone volume/tissue volume (ratio)0.09 ± 0.030.120 ± 0.035<0.001t95.563
  Trabecular number (1/mm)1.29 [0.86–1.76]1.97 [1.71–2.19]<0.001w81.459
  Trabecular thickness (mm)0.06 [0.06–0.08]0.06 [0.05–0.07]<0.05w105.941
  Trabecular spacing (mm)0.69 [0.50–1.09]0.44 [0.40–0.53]<0.001w85.068
  SD.1/Tb.N0.42 [0.24–0.86]0.18 [0.16–0.24]<0.001w82.5
 Estimated bone strength
  Total polar moment of inertia (mm4)551 [428–816]731 [526–998]<0.001w104.875
Ultradistal tibia
 Bone area
  Total bone area (cm2)749 ± 198785 ± 1750.23t111.378
  Cortical bone area (cm2)97.0 ± 22.4118.3 ± 36.3<0.001t102.575
  Trabecular bone area (cm2)642 ± 192660 ± 1740.59t112.076
 Cortical parameters
  Cortical thickness (mm)0.93 [0.87–1.10]1.17 [0.91–1.31]<0.05w106.258
  Cortical porosity (%)5.8 [3.5–8.6]6.0 [4.0–7.9]0.75t109.376
 Trabecular parameters
  Bone volume/tissue volume (ratio)0.08 ± 0.030.14 ± 0.03<0.001t67.250
  Trabecular number (1/mm)1.30 [0.75–1.53]1.94 [1.74–2.13]<0.001w55.743
  Trabecular thickness (mm)0.07 [0.06–0.08]0.07 [0.06–0.08]0.54w110.134
  Trabecular spacing (mm)0.68 [0.57–1.19]0.44 [0.39–0.51]<0.001w75.5
  SD.1/Tb.N0.40 [0.31–1.11]0.20 [0.16–0.25]<0.001t59.7
 Estimated bone strength
  Total polar moment of inertia (mm4)3499 [2640–4294]4423 [3403–5927]<0.001w101.075
Figure 2.

HR-pQCT images of ultradistal radius and ultradistal tibia The bone images were selected from the two pairs of participants who had the largest and the smallest difference in trabecular number in the tibia, respectively.

Microarchitecture as assessed by HR-pQCT

In the radius, cortical thickness (p = 0.32) and porosity (p = 0.10) were similar in the two groups (Table 3). Patients with OI had lower trabecular BV/TV (p < 0.001), lower Tb.N (p < 0.001), and greater Tb.Sp (p < 0.001) compared with controls.

In the tibia, patients with OI had thinner cortical bone (p < 0.05), although cortical porosity (p = 0.75) was similar in the two groups (Table 3). Patients had lower trabecular BV/TV (p < 0.001), lower Tb.N (p < 0.001), and greater Tb.Sp (p < 0.001) compared with controls.

For both radius (p < 0.001) and tibia (p < 0.001), the trabecular network distribution (SD.1/Tb.N) was more inhomogeneous for patients with OI than the controls (Table 3).

Estimation of bone strength

The total polar moment of inertia was lower in both radius (p < 0.001) and tibia (p < 0.001) of patients with OI (Table 3).

The best between-group determinant

Results of the logistic regression modeling (Tables 2 and 3) showed that trabecular number was the parameter that most effectively distinguished between patients with OI and controls overall, and was also the best microarchitecture parameter in this regard. Trabecular vBMD was the best volumetric density parameter, while cortical area was the best geometry parameter.

HR-pQCT results are summarized in Fig. 3.

Figure 3.

Bone mass, geometry, and microstructure in adults with OI and healthy controls assessed by HR-pQCT. Dots mark the mean, and error bars show 1 SD. Box plots show median and range. The parameters presented here are those with the lowest Akaike information criterion. **Between-group significance, p < 0.001.

We found 23 parameters to be significantly different between the two groups, and 18 of these had a significance level below the Bonferroni correction alpha of 0.0011.

Discussion

To the best of our knowledge, this is the first study published using HR-pQCT to investigate bone structure in patients with OI. We found lower BMD, altered bone geometry, and altered bone microarchitecture in adults with type I OI compared with healthy controls.

Patients with type I OI had lower aBMD in all sites. This is in accordance with previous literature.12, 16, 17, 23 Thus, Wekre and colleagues17 found that adults with type I OI had a BMD whole-body Z-score of −0.29 ± 1.02, while Cepollaro and colleagues16 found significantly lower aBMDspine in a group of 21 adults with OI types I and IV compared to age- and gender-matched controls (aBMDspine 0.753 ± 0.01 g/cm2 versus 0.979 ± 0.07 g/cm2). Furthermore, we found that the patients with type I OI had a lower trabecular vBMD in both tibia and radius. In the tibia, total vBMD was also significantly lower in the patients. Using pQCT, Gatti and colleagues23 found decreased total and trabecular vBMD at the radius in patients with OI types I, III, and IV. The fact that we were only able to show a lower total vBMD in the tibia and not the radius may reflect that we only included patients with type I OI. In a pQCT study by Rauch and colleagues,24 children and adolescents with type I OI had an increased cortical vBMD in the diaphysis of the radius. We found no such significant difference between the two groups in our study, possibly because we scanned a more distal region of the forearm, and only included adults.

Patients with OI had lower total bone area in the radius. This is in accordance with Gatti and colleagues,23 who found a trend toward decreased total bone area in the radius measured by pQCT in adults with OI types I, III, and IV compared to healthy controls. Furthermore, we found a lower cortical bone area in the tibia. The different results for bone area in the tibia and radius may be due to the differences in mechanical loading in the two regions. This is supported by our finding of a near-significant decreased cortical (p = 0.06) and trabecular (p = 0.07) bone area in the radius of patients with OI compared to the healthy controls, whereas the between-group differences in both total (p = 0.23) and cortical (p = 0.59) bone area in the tibia were clearly nonsignificant. No reports of altered bone geometry in the lower extremities of patients with type I OI have been published.

Our patients with OI had a thinner cortex in the tibia compared to controls, but not in the radius. In contrast, Gatti and colleagues23 found lower cortical thickness in the ultradistal radius in patients with OI. Gatti and colleagues,23 however, included more severe phenotypes, which may explain the difference of our findings. We expected patients with type I OI to have increased cortical porosity compared with controls due to increased bone turnover,25, 26 but were unable to show a significant difference between the groups. This may be due to lack of statistical power or to the fact that HR-pQCT cannot capture cortical pores that are smaller than the image resolution of 82 µm. Therefore, a portion of pores is not detected and a possible difference in porosity caused by small-in-size pores is thus missed.21 A third possible explanation could be that many of the OI patients had received bisphosphonates and therefore have reduced bone turnover.

Patients with type I OI had lower trabecular number and increased trabecular spacing in both radius and tibia, with increased inhomogeneity of the trabecular network, compared to the controls. We found a tendency toward thicker trabeculae. This is possibly due to increased bone turnover that in turn increases trabecular perforations and thus leads to preferential loss of thin trabeculae—thus increasing the mean thickness of the remaining trabeculae.27 This is in line with histomorphometric data from iliac crest biopsies in adults with type I OI.12, 28

We found that parameters derived from measurements in the lower limb best distinguished between patients with OI and healthy controls. This is in line with the observation that patients with OI are often less affected in the upper extremities than in the lower extremities.29

Our patients with OI were recruited through our clinical database and the national patient society. They came from both rural and urban Denmark and belong to 29 different families. Patients from four of the five Danish regions were represented in the sample. We cannot conclusively rule out recruitment bias, as it is likely that nonsymptomatic OI patients are not diagnosed or followed in specialist clinics and may not be members of the patient society. The controls chosen for this study were invited from a random sample drawn from persons living in the city of Odense, on the island of Funen, Southern Denmark. The population of Funen comprises approximately 9% of the total Danish population and has previously been shown to be a representative segment of the Danish general population.30 Studies comparing bone mass in persons living in rural or urban parts of Norway and Sweden found that urban-dwelling women and men had lower bone mass than rural dwellers.31, 32 The OI patients live in both rural and urban parts of Denmark, and we cannot conclusively rule out the introduction of bias when comparing these patients with an urban population.

Weight is a well-known confounder of bone geometry and structure.33 Our patients with type I OI had altered anthropometry with higher body mass index and lower body surface area than the controls, but similar weight. Physical activity is also a confounder of bone geometry and structure.34 All of the study patients were ambulatory, leading normal lives with little or no help with activities of daily living. A limitation of our study is the lack of a validated measurement tool to record physical activity. Tobacco smoking is another potential confounder of bone structure, especially affecting the trabecular bone microarchitecture.35 In our study we found no significant difference in the number of smokers in the two groups. Menopause is also known to alter bone structure. Nishiyama and colleagues21 showed that postmenopausal women had higher cortical porosity and thinner cortices at the distal radius and tibia than premenopausal women. However, the OI patients and the normal controls were comparable with respect to menopausal status and age at menopause.

In our study, 27 patients were previously or currently treated with bisphosphonates. Both oral and intravenous bisphosphonates increase BMD, total bone volume, and trabecular and cortical thickness, and decrease cortical porosity in adult patients with OI.36–38 Two of our study patients had received bisphosphonates before the age of 20 years. Bisphosphonate treatment during growth would appear to alter the bone structure in patients with OI. Rauch and colleagues14 found significantly increased cortical width, bone volume per tissue volume, and trabecular number in children treated with bisphosphonates. Three of our patients were previously or currently treated with parathyroid hormone (PTH) or PTH analogues. This relatively large portion of patients being treated with anti-absorptive or bone-anabolic medications may have affected the results of the comparative analysis. We consider, however, that the between-group differences found were most likely related to the OI disorder, as antiresorptive treatment would tend to minimize the differences between OI patients and normal controls.

We have evaluated the differences between patients with type I OI and healthy controls on several parameters and cannot conclusively rule out a mass significance effect, but using the Bonferroni transformation most of the parameters remain significant, as shown by the p values in Tables 2 and 3, indicating that the differences between the groups are disease-specific.

A limitation of our study is that we did not measure bone quality or undertook a collagen analysis. Altered bone material properties due to defective collagen or decreased amounts of collagen are present in all patients with OI.

In conclusion, patients with type I OI had altered bone geometry (lower total bone area in the radius), altered bone microstructure (decreased trabecular number, increased trabecular spacing, and greater trabecular inhomogeneity), and lower bone mass (decreased areal and volumetric BMD) compared to healthy controls. Our results suggest that the increased risk of fractures in patients with type I OI is a combined result of altered bone matrix quality, low bone mass, and altered bone microstructure and geometry.

Disclosures

All authors state that they have no conflicts of interest.

Acknowledgements

This work was supported by grants from the University of Southern Denmark, Odense University Hospital, Region of Southern Denmark, Karola Jørgensen's Research Foundation, Guldsmed A.L. & D. Rasmussen's Memorial Foundation, and Edith and Vagn Hedegaard Jensen's Foundation. We thank Lotte Hørlyck and the other staff at the Osteoporosis Clinic, Odense University Hospital, for excellent administrative and technical support, and Claire Gudex for language editing the manuscript.

Authors' roles: Study design: KB, LF, BA, JG, BLL; Study conduct: LF, SH; Data Analysis: LF, KB; Drafting manuscript: LF; Revising manuscript: JDH, SH, JG, BLL, BA, KB; Approving final version of manuscript: LF, JDH, SH, JG, BLL, BA, KB. LF accepts responsibility for the integrity of the data analysis.

Ancillary