High-Resolution Peripheral Quantitative CT Imaging: Cortical Porosity, Poor Trabecular Bone Microarchitecture, and Low Bone Strength in Lung Transplant Recipients

Introduction

Organ transplant recipients often exhibit low bone mineral density (BMD) (1) and a high risk for fragility fractures (2). Transplantation osteoporosis is a complex skeletal condition because pre-existing secondary bone disease (3–8) is complicated by adverse effects of immunosuppressive therapy (1). Posttransplantation fracture risk has been shown to vary according to organ type, and recipients of lung and liver grafts exhibit the lowest BMD and the highest risk for fragility fractures (2).

Although BMD is a key parameter for the prediction of fracture risk at dual-energy x-ray absorptiometry or quantitative computed tomographic (CT) imaging (9), bone strength is only partially predictable by BMD (10–13). Along with clinical risk factors (eg, age, sex, and body mass index [BMI]) and serum markers of bone turnover, bone material properties and bone microarchitecture are recognized as pivotal components for the quantitative prediction of bone strength. Previously, quantification of bone microarchitecture required invasive tissue sampling with bone biopsy. Recent high-resolution bone imaging techniques enabled noninvasive assessment of bone microarchitecture (14). High-resolution (HR) peripheral quantitative CT, also known as HR-pQCT, and magnetic resonance imaging were used to investigate the microstructural basis of metabolic bone diseases including postmenopausal and age-related osteoporosis (15–17), hyperparathyroidism (18), diabetic bone disease (19), and transplantation osteoporosis (20). Of particular interest, HR peripheral quantitative CT imaging provides the possibility to quantify cortical porosity, a structural feature linked with increased bone fragility (17,19,21,22).

To date, we know of no studies that analyze bone microarchitecture in lung transplant (LT) recipients or examine the role that microarchitectural features might play in increased bone fragility within these patients. We hypothesized that LT recipients would exhibit poorer bone microarchitecture, higher cortical porosity, and worse bone strength than healthy control subjects. The aims of our study were to characterize bone microarchitecture and quantify bone strength in LT recipients by using HR peripheral quantitative CT in the ultradistal radius.

Materials and Methods

Participants

This prospective study was approved by our institutional ethics committee. All participants gave written informed consent. We enrolled 58 patients who underwent LT between April 2010 and May 2012 and compared them to 60 healthy control subjects who were enrolled between December 2008 and May 2013. In the control participants, demographic data were self-reported. In the LT group, demographic and clinical data were obtained from a standardized patient documentation and monitoring system. To be included, participants needed to be at least 18 years old. LT had to have taken place at our institution within the last 6 months. After transplantation, the standardized immunosuppressive triple regimen was composed of a calcineurin inhibitor (tacrolimus, 0.05–0.1 [mg · kg⁻¹]/day; or cyclosporine A, 5–10 [mg · kg⁻¹]/day), mycophenolate mofetil (2–3 g/day) and corticosteroids (methylprednisolone, 125 mg every 8 hours for the first 24 hours followed by prednisone, 1 [mg · kg⁻¹]/day). Prednisone was then tapered to 5 mg/day.

All patients received prophylaxis against postoperative bacterial and fungal infections, infections with cytomegalovirus, and Pneumocystis jirovecii. We administered 1000–1500 mg calcium, 400–800 IU oral vitamin D, and oral or intravenous bisphosphonates as a standardized antiosteoporotic treatment. Before transplantation (ie, before study enrollment), all patients received oral low-dose corticosteroids (5–10 mg daily). Other than that, individual pretransplantation drug regimens were tailored to underlying lung diseases and clinical patient needs. Control subjects were without self-reported history of chronic diseases. The foreseeable inability to sit motionless for 3 minutes and bilateral distal radius fractures were exclusion criteria.

HR Peripheral Quantitative CT Imaging

The ultradistal radius of all patients was scanned by using an HR peripheral quantitative CT imager (XtremeCT; Scanco Medical, Brüttisellen, Switzerland). Images were acquired from the nondominant side unless there was a local fracture history, in which case the dominant side was scanned. To reduce motion artifacts, a carbon fiber cast with attached pneumatic pad (Perltec AG, Schlieren, Switzerland) was used to immobilize the forearms of patients during acquisitions. The image (110 sections, 9.02 mm) was placed 9.5 mm proximal to the distal radius endplate. The tube settings defined by the manufacturer standard in vivo protocol were 60 kVp, 900 μA, and 100-msec integration time. Total scan time for a single examination was 3 minutes with an effective radiation dose less than 4 μSv. An isotropic voxel size of 82 × 82 × 82 μm was achieved by reconstructing the 126-mm field of view across a 1536 × 1536 matrix. The full-width–half maximum resolution of the section sensitivity profile of the CT imager was 215 μm (standard protocol). If patient motion was noticed during the measurement, scans were repeated once.

Each image was reviewed by a radiologist (J.P., with 5 years of experience in HR peripheral quantitative CT imaging) section by section and graded for motion artifacts according to Pialat et al (23) on a scale from 1 to 5. Only images with motion grades 1 (no motion) to 3 (moderate motion with minimal effect on quantitative measures) were included in the subsequent image analysis.

Image Analysis

Densitometry and trabecular bone morphometry.—Densitometric and morphometric evaluations were performed according to the manufacturer (24). From the binary trabecular ridge image, trabecular number was measured by using the direct three-dimensional distance transform approach (25,26). Trabecular thickness and separation were derived by using plate model assumptions (L.F.) (27).

Feature clustering and texture analysis of trabecular bone.—Cortical and trabecular bone was segmented by using a threshold-independent segmentation tool (28). Bone quality maps of trabecular microarchitecture clusters (TMACs) (29) were then generated for each volume (L.F., M.D.D., and A.V.). TMACs are computed by using three-dimensional gray-level co-occurrence features and structure tensors on the segmented trabecular volume. By using a Gaussian mixture model, each voxel was assigned to one of three TMACs. These clusters represent the following main features: TMAC 1 is rich in thick trabeculae with low intertrabecular spacing, TMAC 2 is characterized by trabeculae of intermediate morphometric properties, and TMAC 3 contains thin inhomogeneous trabeculae and demonstrates high intertrabecular spacing. Per patient and cluster, relative cluster volumes were computed and visualized by using bone quality maps (29).

Cortical bone analysis.—Apparent cortical thickness was computed as the mean three-dimensional distance between the periosteal and endosteal boundaries (21). The apparent cortical BMD was derived as average for all voxels within the cortical compartment. Percentage of cortical porosity was quantified according to Burghardt et al (21) (L.F.).

Microstructural finite element analyses.—Mechanical competence was assessed by linear microstructural finite element analyses with previous ex vivo validation (30). Voxel conversion into isotropic eight-node hexahedral finite elements was performed, and each element was assigned an isotropic linear elastic material behavior with a Young modulus of 10 GPa and a Poisson ratio of 0.3. Boundary conditions represented an experimental compression test with high friction at the contact areas (31). All nodes on the most distal plane were fixed in the transversal plane while a prescribed displacement in axial direction was applied corresponding to 1% apparent strain. All nodes on the most proximal plane were fully constrained. Linear analyses were performed on a parallel operating system (Linux Operating System; Linux, Ogdensburg, NY) with the following parameters: 2 × 6 Xeon ×5680 processor and 144 GB of RAM in the ParFE solver (32). Apparent stiffness was evaluated by computing the ratio of the resultant force and the imposed displacement. Ultimate load was computed according to Pistoia et al (33). Strength was defined as the quotient of ultimate load and the mean total bone area (T.G.).

Statistical Analyses

Analyses were performed by using statistical software (PASW Statistics 18.0 Statistical Database; IBM, Armonk, NY). Nominal data are presented by using absolute numbers and percentages. Mean values and standard errors of the mean were assessed for metric data. T test for independent samples was used to verify differences in age and BMI. A two-way factorial analysis of covariance (factors were LT and sex) with age-correction was used to determine differences in HR peripheral quantitative CT–derived parameters between LT recipients and control subjects. P values of .05 or less indicated statistical significance. P values less than .1 indicated statistical significance for interactions. To avoid increasing errors of the second type and because of the small sample size, we refrained from multiplicity corrections (M.W.).

Results

In this study, 118 participants (consisting of 58 LT recipients [mean age, 46.8 years ± 1.9; 30 women: mean age, 48.0 years ± 2.4; 28 men: mean age, 45.4 years ± 3.2] and 60 control participants [mean age, 39.9 years ± 1.9; 41 women: mean age, 40.6 years ± 2.2; 19 men: mean age, 38.5 years ± 3.5]) underwent HR peripheral quantitative CT imaging of the ultradistal radius. All patients were white. LT patients were older (LT patients were a mean 7 years older than control participants; P = .013) and had a lower BMI (LT patients, 20.3 kg/m² ± 0.5; healthy control subjects, 23.5 kg/m² ± 0.5; difference of −13.9%; P < .001) than control subjects. The median time interval between surgery and HR peripheral quantitative CT imaging was 54 days (range, 11–296 days).

The majority of LT patients underwent transplantation for end-stage chronic obstructive pulmonary disease (44.8%; 26 of 58 patients), while 22.4% (13 of 58 patients) underwent transplantation for other parenchymal diseases, including pulmonary fibrosis (n = 6), lymphangioleiomyomatosis (n = 2), α1-antitrypsin deficiency (n = 2), bronchiectasis (n = 1), graft-versus-host disease (n = 1), and acute respiratory distress syndrome (n = 1). A further 19.0% (11 of 58) had underlying vascular diseases, which included idiopathic pulmonary hemosiderosis (n = 4), chronic thromboembolic pulmonary hypertension (n = 3), primary pulmonary hypertension (n = 2), and pulmonary veno-occlusive disease (n = 2). The final 13.8% (eight of 58) of LT recipients who were enrolled in this study had cystic fibrosis. Patient characteristics are reported in Table 1.

Table 1 Participant Characteristics

Note.—Data are numerator and denominator except where indicated; data in parentheses are percentages. All participants were white. COPD = chronic obstructive pulmonary disease, NA = not applicable.

*Data are ± standard error of the mean.

^†P = .013; t test for independent samples

^‡P < .001; t test for independent samples.

Motion grading (23) was performed for all 118 scans, which yielded motion grade 1 in 50 scans (42%), motion grade 2 in 54 scans (46%), and motion grade 3 in 14 scans (12%). No images were excluded. HR peripheral quantitative CT-derived parameters are shown in Table 2.

Table 2 Bone Mineral Density, Trabecular and Cortical Bone Microarchitecture, and Mechanical Competence of the Ultradistal Radius in Healthy Control Participants and Patients with Recent LT

Note.—P values were adjusted for age and sex.

*Data are mean ± standard error of the mean.

^†P values of .05 or less were indicative of statistical significance.

^‡A measure of trabecular network heterogeneity.

Image analysis time was about 30 minutes per case and included operator time and computational time.

BMD and Bone Microarchitecture

Total BMD (−13.4%; P = .001) and trabecular BMD (−16.4%; P = .001) of the ultradistal radius were significantly lower in LT recipients than in healthy control subjects.

Cortical BMD differences between LT recipients and control subjects were not statistically significant (−0.2%; P = .850).

LT recipients had lower trabecular number (trabecular number, −9.7%; P = .004) and lower trabecular thickness (−8.1%; P = .025). Trabecular separation (+24.3%; P = .007) and trabecular network heterogeneity (+63.9%; P = .012) were higher in LT recipients than in healthy control subjects.

LT recipients showed significantly higher cortical porosity (+31.3%, P = .035) and lower cortical thickness (−10.2%, P = .005) at the ultradistal radius than healthy control subjects. Example visualizations of cortical porosity are given in Figure 1. Transplantation-related differences in cortical thickness were more pronounced in men than in women (women: LT vs healthy control subjects, −4.9%; men: LT vs healthy control subjects, −14.5%; interaction, P = .093).

Figure 1:

Download as PowerPointOpen in Image Viewer

Trabecular Texture

Alterations in trabecular microarchitecture were also reflected by three-dimensional texture analysis. The mean relative volume of TMAC 3 (ie, poor quality cluster containing thin, inhomogeneous trabeculae; Fig 2) was significantly larger in LT recipients than in healthy control subjects (+55.7%; P = .001). LT recipients exhibited lower TMAC 2 (ie, trabeculae of intermediate morphometric properties; Fig 2) than did healthy control subjects (−18.9%; P < .001).

Figure 2a:

Download as PowerPointOpen in Image Viewer

Figure 2b:

Download as PowerPointOpen in Image Viewer

The relative volume of TMAC 1 did not differ between LT recipients and healthy control subjects. However, visual examination of bone quality maps of individual patients revealed marked irregularity and cluster discontinuity of TMAC 1 in LT patients (Fig 2).

Mechanical Competence

LT patients exhibited significantly lower stiffness (stiffness, −15.0%; P < .001), failure force (−14.8%; P < .001), and bone strength (bone strength, −14.6%, P < .001). Transplantation-related deficits in stiffness (stiffness, women: LT vs healthy control subjects, −7.5%; stiffness, men: LT vs healthy control subjects, −19.5%; interaction, P = .013), failure force (women: LT vs healthy control subjects, −7.3%; men: LT vs healthy control subjects, −19.6%; P = .009) and bone strength (bone strength, women: LT vs healthy control subjects, −7.2%; bone strength, men: LT vs healthy control subjects, −20.7%; interaction, P = .040) were more pronounced in men than in women. Statistically significant group-sex interactions are shown in Table 3.

Table 3 Significant Interactions Including Relative Differences between LT Status and Sex

Note.—Interactions between group and sex were determined by two-way analysis of covariance. P values were adjusted for age and sex.

*Data are mean ± standard error of the mean.

^†P values less than .1 were indicative of statistical significance for interactions.

Discussion

The additive effect of pre-existing secondary bone disease and posttransplantation bone loss predispose LT recipients to fragility fractures (1,2). Low BMD is a frequent finding in LT recipients; nevertheless, little is known about the underlying microarchitectural changes. To address this issue, we performed high-resolution quantitative CT imaging (HR peripheral quantitative CT imaging) of the ultradistal radius in LT recipients and compared them with healthy control subjects. Specifically, we used HR peripheral quantitative CT to determine quantitative measures of bone microarchitecture, which included cortical porosity, and to calculate mechanical competence at a mean postsurgical interval of 3 months.

In line with the literature (1,2), we found lower overall BMD in LT patients than in control subjects, and trabecular BMD was lower. Cortical porosity was higher in LT recipients, which constitutes one of our main findings. The clinical relevance of cortical porosity is in altered cortical integrity and impaired strength, which contributes to severe deficits in bone biomechanics (34–37). Cortical porosity can be modulated by pharmacologic intervention (eg, administration of denosumab, alendronate [38], or ibandronate [39]), thereby lending clinical relevance to its radiologic detection, quantification, and monitoring by high-resolution imaging (38,39). Somewhat surprisingly, we found no significant differences in cortical BMD between transplant recipients and control subjects.

Deficiencies in trabecular bone microarchitecture were also seen in LT recipients. In particular, both male and female LT recipients had lower trabecular number in comparison to healthy control subjects, implying loss of trabecular structures and reduced mechanical competence as predicted in computer simulations (40). Additionally, low trabecular number in male and female LT patients reflects a differential microarchitectural deficiency to that seen in age-related osteoporosis in men or corticosteroid-induced osteoporosis (41). Those diseases are primarily characterized by low trabecular thickness, which indicates trabecular thinning (42) rather than loss of trabeculae. In general, our findings indicate that microarchitectural features of bone disease after LT were independent of sex, which further highlights the differentiation between transplantation osteoporosis and primary osteoporosis.

Mechanical competence of the ultradistal radius was estimated by using a computer simulation of axial loading, which models a fall on the outstretched hand. Specifically, we quantified stiffness, failure load, and resultant overall bone strength by a standardized technique (30) and found severe deficits of all three parameters in LT recipients. We have not investigated associations between bone microarchitecture, bone strength, and clinical presentation, but literature suggests that low bone strength can be caused by a multitude of factors including systemic inflammation (43), use of corticosteroids and immunosuppressive agents, limited physical activity, sarcopenia, sex hormone deficiency, history of smoking, and a low BMI (44–48). We recognize that low BMI is typical in LT recipients (44), and we refrained from statistical adjustment for this parameter, which therefore further avoided the issue of multicolinearity.

There were some limitations to our study. Because of the lack of pretransplantation HR peripheral quantitative CT imaging, it was not possible to demonstrate whether the observed deficits in bone density and bone quality occurred before or after transplantation or both. Heterogeneity of underlying chronic lung diseases and possible skeletal adverse effects of drugs taken before and/or after transplantation require cautious interpretation of HR peripheral quantitative CT data, in particular with respect to the reasons and timing for the observed differences. LT recipients received calcium, vitamin D supplements, and bisphosphonates; therefore, their bones were not treatment-naive for antiosteoporotic drugs. Furthermore, HR peripheral quantitative CT imaging was restricted to peripheral skeletal sites. Based on correlations between biomechanical estimates for the proximal femur and the peripheral skeleton, we only scanned a single location, the ultradistal radius, which minimized exposure to radiation (49). Weight-bearing bones were not scanned, and we did not account for potential differences in forearm length (eg, between men and women). In cases where the dominant radius was scanned because of previous fracture incidents of the nondominant side, corresponding control subjects were not proactively matched by dominant hand. Cortical porosity has been shown to be underestimated by HR peripheral quantitative CT because of its spatial resolution (50). Moreover, the definition of the cortico-trabecular transition zone (which affects the quantification of bone microarchitecture) is the subject of ongoing research (21,28,51).

Patients and control participants were not explicitly age matched, and the mean age of control subjects was younger than that of the LT recipients. We have thus adjusted our statistical analyses for age and sex. For a broader perspective on the differences between data from LT recipients and control subjects, we have additionally compared and validated group-related differences with normative HR peripheral quantitative CT data published by other groups (17,52–54). Further research is warranted to address interactions between bone microarchitecture, age, sex, transplant indication, and clinical parameters, including rejection episodes in longitudinal datasets.

Dual-energy x-ray absorptiometry scans were only available in a subset of patients and were thus not reported. Because our approach was cross-sectional, future work should apply the same methodologic analysis to investigate longitudinal changes in bone quality and to quantify their relationship with serum markers of bone turnover, fracture incidence, and clinical parameters, such as graft rejection.

We conclude that LT recipients exhibit low bone mineral density; however, our data indicate heretofore unrecognized microarchitectural skeletal deficiencies, including elevated cortical porosity, low cortical thickness, diminished trabecular microarchitectural quality, and lower estimated bone strength. Trabecular defects and, in particular, elevated cortical porosity could provide a potential explanation for bone fragility after LT.

Author Contributions

Author contributions: Guarantors of integrity of entire study, C.S.W., D.K., H.R., W.K., J.M.P.; study concepts/study design or data acquisition or data analysis/interpretation, all authors; manuscript drafting or manuscript revision for important intellectual content, all authors; approval of final version of submitted manuscript, all authors; literature research, L.F., A.V., M.D.D., D.K., H.R., B.Z., P.P., G.L., J.M.P.; clinical studies, L.F., A.V., D.K., H.R., P.J., W.K., B.Z., F.K.; experimental studies, L.F., A.V., D.K., H.R., T.G.; statistical analysis, L.F., A.V., M.D.D., D.K., H.R., M.W., P.P., G.L.; and manuscript editing, L.F., A.V., M.D.D., C.S.W., D.K., H.R., T.G., M.W., B.Z., P.P., G.L., J.M.P.