https://orcid.org/0000-0003-2032-8997
Key Points
Patients with hemophilia have abnormal bone microarchitecture and decreased biomechanical bone strength compared with normal healthy controls.
Decreased BMD and bone strength are associated with severity of hemophilic arthropathy.
Abstract
Low bone density is a growing concern in aging men with hemophilia and may result in high-morbidity fragility fractures. Using high-resolution peripheral quantitative computed tomography (HR-pQCT), we demonstrate low trabecular and cortical bone density contributing to lower volumetric bone mineral density (BMD) at both distal radius and tibia in patients with hemophilia compared with age- and sex-matched controls. The low trabecular bone density found in hemophilia is attributed to significantly decreased trabecular number and increased separation; the lower cortical bone density results from thinner cortices, whereas cortical porosity is maintained. Microfinite element analysis from three-dimensional HR-pQCT images demonstrates that these microarchitectural deficits seen in patients with hemophilia translate into significantly lower estimated failure load (biomechanical bone strength) at the distal tibia and radius when compared with controls. In addition, an inverse association of joint score with BMD and failure load suggests the negative role of hemophilic arthropathy in bone density loss.
Introduction
With the increasing availability and affordability of safe factor concentrate replacement therapy, factor prophylaxis, and comprehensive care, persons with hemophilia (PWH) today should have a normal life expectancy, and more are living beyond age 65 years.1,2 However, comorbid complications associated with aging in this population are largely unknown, and background data on the clinical implications of these issues are lacking. Low bone mineral density (BMD) and associated osteoporosis is one of the emerging concerns that has been described, with an estimated prevalence of up to 70% in PWH.3-5 Although low BMD is well described in PWH, data remains limited on fracture risk, with case series reporting fracture prevalence of 12% to 18%.6-8
Osteoporosis is a pathologic bone disorder characterized by low BMD and microarchitectural bone disruption that results in increased risk for fracture.9 The majority of studies investigating BMD in males with hemophilia have used areal BMD measurement by dual X-ray absorptiometry, which is unable to account for geometric differences affecting true volumetric BMD.10 Furthermore, dual X-ray absorptiometry provides no information on the microarchitecture of the cortical and trabecular compartments that ultimately underpin bone strength.11
In this pilot study, we used high-resolution peripheral quantitative computed tomography (HR-pQCT) to measure true BMD and detect changes in cortical and trabecular bone microarchitecture (to an isotropic voxel size of 82 μm) and used novel microfinite element analysis (μFEA) to assess the effect of skeletal alterations on bone strength compared with matched controls. Because inactivity and cytokine-induced bone mineral loss resulting from chronic hemophilic arthropathy are proposed mechanisms for BMD loss in PWH,12,13 Gilbert joint scores for severity of hemophilia arthropathy were measured for correlation analysis.14
Study design
Subjects
After obtaining University of Calgary Research Ethics Board approval, 18 patients with hemophilia A and hemophilia B (≥18 years old) with severe (factors VIII or IX, <1 U/dL) or moderate (factors VIII/IX, 1-5 U/dL) disease were recruited through the Southern Alberta Rare Blood and Bleeding Disorders Comprehensive Care Program. Subjects receiving corticosteroids for more than 3 months were excluded. HR-pQCT data on age- and sex-matched controls (at a 2:1 ratio to PWH) were obtained from the Canadian Multicentre Osteoporosis Study on healthy subjects.15
Protocol
All procedures were conducted at the Southern Alberta Rare Blood and Bleeding Disorders Program and the University of Calgary Bone Imaging Laboratory. For each subject, anthropomorphic measurements and Gilbert joint score were collected. Gilbert joint score, a measure of severity of joint arthropathy, reflects the combined score of 6 joints (elbows, knees, ankles), where a higher score indicates worse arthropathy, and was determined by a single, expert-trained examiner (A.L.) for all subjects.16 Bone microarchitecture and strength of the nondominant distal radius and tibia were assessed by HR-pQCT (XtremeCT; Scanco Medical, Brüttisellen, Switzerland) and µFEA (Faim v6.0; Numerics88 Solution Ltd, Calgary, AB, Canada).
HR-pQCT imaging
Distal radius and tibia HR-pQCT measurements and analysis of PWH were performed identically to the Canadian Multicentre Osteoporosis Study.15 Standard morphologic analysis includes total volumetric BMD (mg HA/cm3), ratio of trabecular bone volume to total bone volume (BV/TV, reflecting trabecular BMD), trabecular number (Tb.N, 1/mm), trabecular thickness (Tb.Th, mm), and trabecular separation (Tb.Sp, mm), as well as cross-sectional cortical analysis, including cortical BMD (Ct.BMD, mg HA/cm3), cortical thickness (Ct.Th, mm), and cortical porosity (Ct.Po, %).17
μFEA
Linear homogenous finite element models from three-dimensional HR-pQCT image data were generated as per the Canadian Multicentre Osteoporosis Study.15 Finite element models were solved using µFEA software to calculate estimated biomechanical bone strength or failure load in Newtons, which is the amount of axial force needed to be applied to the bone for it to fail.
Statistical analysis
Analysis of variance model adjusted for age was used for comparison of bone parameters between PWH and controls. Significance is defined as P < .05. Spearman rank correlation analysis was used to investigate the relationships among BMD, failure load, and Gilbert joint score.
Results and discussion
As indicated in Tables 1 and 2, 18 patients with moderate or severe hemophilia were included in this pilot study, including 15 (83%) with hemophilia A. Among them, 15 (83%) have severe disease, 3 (17%) are HIV-positive, and 4 (22%) have documented liver cirrhosis. Only 5 (28%) had received primary prophylaxis. The control and hemophilia groups had similar average age (33.5 vs 33.7 years), whereas the controls had insignificantly higher height and weight and lower BMI. Vitamin D levels were available on 17 patients; 41% had vitamin D deficiency (<50 nmol/L). Compared with matched controls, PWH demonstrated significantly lower nondominant distal tibia and radius total BMD (288.9 and 313 mg HA/cm3; P = .003, P = .012), trabecular BMD (BV/TV: 0.15 and 0.14; P = .004, P = .014), and cortical BMD (880 and 916 mg HA/cm3; P = .032, P = .003). In addition, PWH had significantly lower Tb.N in the distal tibia, but not radius, and increased Tb.Sp at both tibia and radius. Ct.Th was significantly lower at both sites, but not Ct.Po.
The three-dimensional reconstructed image of the distal radius of a patient with severe hemophilia A illustrates fewer trabeculae that are less dense, as well as a thinner cortex compared with the matched healthy control (Figure 1A).
![Figure 1. Bone microarchitecture, bone strength, and correlation with joint score. (A) Representative HR-pQCT images at a nominal isotropic resolution of 82 μm of the distal radius in a 43-year-old male patient with severe hemophilia A (Tb.N, 1.82; Tb.Th, 0.061; Ct.Th, 0.94; Ct.Po, 2.23) and an age- and sex-matched control subject (Tb.N, 2.02; Tb.Th, 0.07; Ct.Th, 1.12; Ct.Po, 1.18). (B) Failure load (higher failure load = better mechanical bone strength) at the distal tibia in PWH and controls. The upper and lower boundaries of the box plot represent the interquartile range, the diamond represents the mean, the line dividing the box plot represents the median, and the whiskers indicate the maximum and minimum values. (C) Failure load at distal tibia vs total Gilbert joint score (higher joint score = more severe arthropathy) in 17 PWH (the remaining patient did not have Gilbert joint score available). Open square data points (□) indicate subjects who received primary prophylaxis (only 4 of 5 receiving primary prophylaxis shown, as 1 did not have joint score available) and as a whole appears to have lower Gilbert joint scores and higher failure load. The outlier (X) with the lowest tibial failure load has liver cirrhosis secondary to hepatitis C virus infection and HIV, which likely explains the lower tibial failure load and BMD. When the outlier is excluded from the correlation analysis, the statistically significant inverse correlation of Gilbert joint score and tibial BMD and failure load still stands (tibial failure load: R = −0.601 [P = .0139]; tibial BMD: R = −0.66 [P = .005]; R = Spearman ρ).](https://ash.silverchair-cdn.com/ash/content_public/journal/blood/125/13/10.1182_blood-2014-10-602060/4/m_2160f1.jpeg?Expires=1735530771&Signature=ynqCx8P47k7H6dJJmTeRnjJyuHkRrwZkblZ7wRGvtgvLNwEwNYTkKSIKXuDJwZ2KYJr6DSPQIj0BtW~J2BUgNXeUJkpthhRy7IbkpdBq-Rezh3i71UVfaC1kzTGKVoKx0A-f71Ws9aAj3u7Y8WC8-EpRpWzO~KZGFpr77a41OHdqTWpN18oenni14HGLtFuoaceLULnklSgyH9eTJCYeWT-v~luky0vLYOMjoPYe1cICS4NwbU76wPEbLbW3me9ls~hRaq36kYEY2JRGl45AHgQMnGyF2-dF2ux58h0b3ymrCKQbd~LcXuy374uXSlGeXVf3D6Fyo3vMaSy18~IuVA__&Key-Pair-Id=APKAIE5G5CRDK6RD3PGA)
Bone microarchitecture, bone strength, and correlation with joint score. (A) Representative HR-pQCT images at a nominal isotropic resolution of 82 μm of the distal radius in a 43-year-old male patient with severe hemophilia A (Tb.N, 1.82; Tb.Th, 0.061; Ct.Th, 0.94; Ct.Po, 2.23) and an age- and sex-matched control subject (Tb.N, 2.02; Tb.Th, 0.07; Ct.Th, 1.12; Ct.Po, 1.18). (B) Failure load (higher failure load = better mechanical bone strength) at the distal tibia in PWH and controls. The upper and lower boundaries of the box plot represent the interquartile range, the diamond represents the mean, the line dividing the box plot represents the median, and the whiskers indicate the maximum and minimum values. (C) Failure load at distal tibia vs total Gilbert joint score (higher joint score = more severe arthropathy) in 17 PWH (the remaining patient did not have Gilbert joint score available). Open square data points (□) indicate subjects who received primary prophylaxis (only 4 of 5 receiving primary prophylaxis shown, as 1 did not have joint score available) and as a whole appears to have lower Gilbert joint scores and higher failure load. The outlier (X) with the lowest tibial failure load has liver cirrhosis secondary to hepatitis C virus infection and HIV, which likely explains the lower tibial failure load and BMD. When the outlier is excluded from the correlation analysis, the statistically significant inverse correlation of Gilbert joint score and tibial BMD and failure load still stands (tibial failure load: R = −0.601 [P = .0139]; tibial BMD: R = −0.66 [P = .005]; R = Spearman ρ).
Bone microarchitecture, bone strength, and correlation with joint score. (A) Representative HR-pQCT images at a nominal isotropic resolution of 82 μm of the distal radius in a 43-year-old male patient with severe hemophilia A (Tb.N, 1.82; Tb.Th, 0.061; Ct.Th, 0.94; Ct.Po, 2.23) and an age- and sex-matched control subject (Tb.N, 2.02; Tb.Th, 0.07; Ct.Th, 1.12; Ct.Po, 1.18). (B) Failure load (higher failure load = better mechanical bone strength) at the distal tibia in PWH and controls. The upper and lower boundaries of the box plot represent the interquartile range, the diamond represents the mean, the line dividing the box plot represents the median, and the whiskers indicate the maximum and minimum values. (C) Failure load at distal tibia vs total Gilbert joint score (higher joint score = more severe arthropathy) in 17 PWH (the remaining patient did not have Gilbert joint score available). Open square data points (□) indicate subjects who received primary prophylaxis (only 4 of 5 receiving primary prophylaxis shown, as 1 did not have joint score available) and as a whole appears to have lower Gilbert joint scores and higher failure load. The outlier (X) with the lowest tibial failure load has liver cirrhosis secondary to hepatitis C virus infection and HIV, which likely explains the lower tibial failure load and BMD. When the outlier is excluded from the correlation analysis, the statistically significant inverse correlation of Gilbert joint score and tibial BMD and failure load still stands (tibial failure load: R = −0.601 [P = .0139]; tibial BMD: R = −0.66 [P = .005]; R = Spearman ρ).
μFEA assessment of the reconstructed HR-pQCT images indicates that the lower BMD and microarchitectural deficits present in PWH translated into significantly lower failure load at the distal tibia (6357 vs 7573 N; P = .009) and radius (2371 vs 3013 N; P = .002) compared with matched controls (Table 2; Figure 1B).
There is a significant inverse correlation of Gilbert joint score to tibia BMD (R = −0.697; P = .002), tibia failure load (R = −0.635; P = .006), and radius failure load (R = −0.552; P = .022) (Figure 1C). An inverse relationship with Gilbert score is seen for radius BMD (R = −0.372; P = .142), although it is not statistically significant. Individuals who received primary prophylaxis appear to have higher failure load.
These compartmental alterations in microarchitecture in PWH are similar to those reported in postmenopausal women with osteoporosis and fragility fractures; specifically, lower trabecular BMD and fewer Tb.N. Decreased CtPo, described in postmenopausal women, was not seen in PWH.18,19 Cortical thinning and trabecular bone loss are known to be important contributors to bone fragility, whereas low cortical porosity appears to be the main contributor to low bone strength in patients with monoclonal gammopathy of unknown significance.20 Studies in fractured osteopenic women demonstrate that lower trabecular BMD (BV/TV) resulting from loss of Tb.N (as opposed to decreased Tb.Th) and increased Tb.Sp are important contributors to low bone strength (2-5 times by finite element method).21,22 Population studies using HR-pQCT and μFEA demonstrate progressive cortical thinning and porosity, and loss of Tb.N leads to decreased bone strength with aging.23,24 These findings in our hemophilia study subjects are concerning, considering their average age was only 33.5 years. Low trabecular BMD specifically at the distal radius has also been described in boys with hemophilia (aged 6.6-19.8 years), using lower-resolution peripheral quantitative computed tomography (pQCT).25 Taken together, these findings suggest microarchitectural deficits start at an early age in hemophilia, and poor bone strength can be expected to worsen over time.
The mechanism for low BMD and osteoporosis in hemophilia is multifactorial. One of the hypothesized contributors to this process is severity of hemophilic arthropathy, which may result in decreased axial loading caused by inactivity, and/or inflammatory cytokine-induced bone resorption.9 This study suggests that hemophilic arthropathy results in greater BMD loss, as shown by the inverse association of joint score with BMD and failure load at both the tibia and radius.
Limitations of this study are the sample size and inability to perform subgroup analysis. In addition, the HR-pQCT data from this study are unable to describe the dynamic processes underlying the bone pathophysiology and identify whether there is excessive osteoclast-mediated resorption, decreased osteoblastic new bone formation, or both. Future inclusion of bone remodeling indices may help clarify the mechanisms and be pertinent to proposed interventions. Our data, however, are the first to our knowledge to describe abnormal microarchitectural changes in both cortical and trabecular compartments and significantly reduced bone strength associated with severity of hemophilic arthropathy in adult PWH. Further studies are needed to determine whether estimated bone strength using HR-pQCT is predictive of future fracture risk as these patients age.
The publication costs of this article were defrayed in part by page charge payment. Therefore, and solely to indicate this fact, this article is hereby marked “advertisement” in accordance with 18 USC section 1734.
Acknowledgments
We thank Michelle Kan, Duncan Raymond, and Anne Cooke from the University of Calgary Bone Imaging Laboratory for performing HR-pQCT scanning and μFEA calculations. A.L. is a recipient of the Bayer International Hemophilia Fellowship Award.
The study was supported in part by the University of Calgary Division of Hematology and Hematologic Malignancies Research fund.
Authorship
Contribution: A.L., S.K.B., G.K., and M.-C.P. designed the study; S.K.B. supervised the HR-pQCT scans and μFEA calculations performed at the University of Calgary Bone Imaging Laboratory and provided control data from the Canadian Multicentre Osteoporosis Study; A.L. performed the statistical analysis and wrote the manuscript for this article; and all authors participated in its editing and revision.
Conflict-of-interest disclosure: The authors declare no competing financial interests.
Correspondence: Man-Chiu Poon, Foothills Medical Centre, 1403 29th St NW, Calgary, AB, Canada T2N 2T9; e-mail: mcpoon@ucalgary.ca
![Figure 1. Bone microarchitecture, bone strength, and correlation with joint score. (A) Representative HR-pQCT images at a nominal isotropic resolution of 82 μm of the distal radius in a 43-year-old male patient with severe hemophilia A (Tb.N, 1.82; Tb.Th, 0.061; Ct.Th, 0.94; Ct.Po, 2.23) and an age- and sex-matched control subject (Tb.N, 2.02; Tb.Th, 0.07; Ct.Th, 1.12; Ct.Po, 1.18). (B) Failure load (higher failure load = better mechanical bone strength) at the distal tibia in PWH and controls. The upper and lower boundaries of the box plot represent the interquartile range, the diamond represents the mean, the line dividing the box plot represents the median, and the whiskers indicate the maximum and minimum values. (C) Failure load at distal tibia vs total Gilbert joint score (higher joint score = more severe arthropathy) in 17 PWH (the remaining patient did not have Gilbert joint score available). Open square data points (□) indicate subjects who received primary prophylaxis (only 4 of 5 receiving primary prophylaxis shown, as 1 did not have joint score available) and as a whole appears to have lower Gilbert joint scores and higher failure load. The outlier (X) with the lowest tibial failure load has liver cirrhosis secondary to hepatitis C virus infection and HIV, which likely explains the lower tibial failure load and BMD. When the outlier is excluded from the correlation analysis, the statistically significant inverse correlation of Gilbert joint score and tibial BMD and failure load still stands (tibial failure load: R = −0.601 [P = .0139]; tibial BMD: R = −0.66 [P = .005]; R = Spearman ρ).](https://ash.silverchair-cdn.com/ash/content_public/journal/blood/125/13/10.1182_blood-2014-10-602060/4/m_2160f1.jpeg?Expires=1735530772&Signature=bD0bq8vxCoj6F4tjgH7jQX5oFX5wKESnVC3i1exM2045dsYBUKjTnqxgagAAjh8Eo-n9~hWKoWkH0PAJmbRGUDdLsZW8VCK78-Hy9TcdJCD76vmec4g74NW3KOfnhTfO54SRzPh9BHF7QP5fJfxeRXDVZkd45yKYnTvvM~S4dJpKhxcaRA0q9VtTgafIVXiDj9uCXm9SN6doRQRFCUDVbM7SlbFYjovsr0SbURT1aK7GLFeND0Pp9BVI11hMkbxLrCeSDrph~dnCOGNhjwsiFwt3xsIMP0BLYPksl5bwPHx2gAk53LMeyY6dzXZVMVbGPXcG9FL3m-aQ~wRIvSNBug__&Key-Pair-Id=APKAIE5G5CRDK6RD3PGA)
