Effect of Denosumab on Peripheral Compartmental Bone Density, Microarchitecture and Estimated Bone Strength in De Novo Kidney Transplant Recipients Open Access

Kidney transplant recipients (KTR) are at high risk to loose bone mass, especially in the first year after transplantation, due to the immunosuppressive medication (glucocorticoids in particular) and other factors such as persistence of elevated PTH and FGF-23 levels and low vitamin D status [1]. Thus, KTR have a heightened risk to develop or to worsen osteoporosis and to suffer from fractures [2,3]. Outcomes for KTR sustaining a fracture are significantly worse compared with the general population, with a 60% increased risk in mortality [4]. Preventing bone loss and reducing the risk of subsequent fractures among KTR is therefore of great clinical relevance [5,6].

Therapeutic recommendations to prevent bone loss in first-year KTR include glucocorticoid minimization or avoidance, calcium and vitamin D supplementation, and bisphosphonates [7,8,9,10]. Recently we demonstrated in a randomized controlled clinical trial among 90 first-year KTR that denosumab effectively increased bone mineral density (BMD), with a therapeutic effect that appeared to surpass previously described treatments [11]. In the present bone quality ancillary study we examined whether treatment with denosumab also improved bone microarchitecture and estimated bone strength. Specifically, we assessed volumetric BMD, cortical thickness, cortical porosity, stiffness and failure load at the distal tibia and radius by HRpQCT analysis in a subgroup of 24 first-year KTR.

Details of the “Prolia for Osteoporosis of Transplant Operated Patients” (POSTOP) study subjects have been reported previously [11]. In brief, we recruited a total of 90 KTR that had been transplanted less than 4 weeks ago to examine 1-year changes in aBMD at the lumbar spine and the hip by dual energy X-ray absorptiometry (DXA). A subset of 24 consecutive patients consented to have in addition two HRpQCT examinations of the distal tibia and radius at baseline and after 12 months.

Male (n=16) and female (n=8) transplant recipients were randomly assigned (1:1) to open-label treatment with denosumab 60 mg subcutaneously at baseline and after 6 months (n=10), or no treatment (n=14). All patients were prescribed daily calcium (1000 mg) and vitamin D (800 IU). All participants gave their written informed consent. The study was designed and conducted to conform to the principles of the Declaration of Helsinki, was approved by the local Ethics Committee (IRB approval number 2011-0032), and was registered on ClinicalTrials.gov, number NCT01377467.

Patients were seen at defined study visits for assessment of transplant function (creatinine-based eGFR determination by CKD-EPI formula) and to measure biomarkers of bone metabolism (parathyroid hormone [PTH], 25-hydroxyvitamin D, and 1,25-dihydroxyvitamin D) and turnover (procollagen type I N-terminal propeptide [P1NP], bone-specific alkaline phosphatase [BSAP], β-form of C-terminal telopeptide of type I collagen [β-CTx] and urine deoxypyridinoline to creatinine ratio). All study patients underwent DXA (Hologic Discovery®, Marlborough, MA, USA) and HRpQCT (XtremeCT®, Scano Medical AG, Brüttisellen, Switzerland) measurements at baseline and after 12 months.

At baseline and after 12 months, HRpQCT measurements were performed at the dominant radius and tibia according to the standard protocol which was provided by the manufacturer. The dominant side was chosen as most patients had a dialysis shunt at their non-dominant forearm which interfered with the positioning holder. With an isotropic voxel size of 82 microns, each scan included 110 slices representing a 9.02 mm region of interest. The final volume of interest for the standard analysis included the overlapping region from baseline and 12 months follow-up scans which was based on the manufacturer's 2D registration algorithm. Scan quality was graded according to Pialat et al. [12], with a five grade scheme ranging from grade 1 (no motion artifacts) to grade 5 (severe motion artifacts). Scans with insufficient quality (i.e. grade 4 or 5) were repeated. This was needed in 17 (35%) of the 48 radial scans and 6 (8%) of the 48 tibial scans.

The scans were evaluated using the standard patient evaluation protocol provided by the manufacturer [13]. First, cortical and trabecular regions were separated using a semi-automated contouring based on a threshold-based algorithm [14]. Voxels representing mineralized bone were then extracted using a Laplace-Hamming filter, followed by normalization and global thresholding to create segmented images. From the segmented images, the following parameters were calculated: volumetric bone mineral density (mg HA/cm3) for the total (Tot.vBMD), cortical (Ct.vBMD) and trabecular regions (Tb.vBMD), micro-architectural parameters of the trabecular structure (trabecular thickness [Tb.Th, mm], trabecular separation [Tb.Sp, mm], trabecular number [Tb.N, mm^-1]) [15], and cortical thickness (Ct.Th, mm). Additionally, advanced cortical analysis was performed to calculate cortical porosity (Ct.Po, %) [16].

Micro-finite element analysis (µFEA) models were created from segmented images of all 110 slices of baseline and follow-up scans. Each voxel representing bone tissue was converted into a brick element of the same size with linear-elastic and isotropic material behavior, a Young's modulus of 10 GPa and a Poisson's ratio of 0.3 [17]. The µFEA model simulated a uniaxial “high friction” compression test where bone stiffness (kN/mm) and failure load (kN) were estimated. Failure load was defined based on the Pistoia criterion assuming that failure would occur when more than 2% of bone volume is loaded beyond a critical strain limit of 0.7%.

A complete case analysis was performed for all endpoints. The percentage change between baseline and 12 months was computed for lumbar spine aBMD and distal tibia and radius HRpQCT parameters and compared between the treatment groups with an exact Wilcoxon rank sum test. For lumbar spine T-score, the absolute change between baseline and 12 months was computed and compared between the treatment groups with an exact Wilcoxon rank sum test. One-year changes are described by the median and 95% confidence intervals (95% CI) based on quantiles of the binomial distribution. Hodges-Lehmann median differences with 95% CI were calculated to determine the differences between the treatment groups. Statistical analyses were done using IBM SPSS Statistics Version 20. The 95% CI were computed in R, Version 3.1.0.

The mean age of the 24 transplant recipients was 53.4 ± 10.9 years; two thirds were male and all were of white ethnicity; the baseline eGFR amounted to 52.5 ± 14.4 ml/min/1.73 m². All patients were treated with triple immunosuppression, including a calcineurin antagonist (tacrolimus 92%, cyclosporine 8%), mycophenolate mofetil (100%) and glucocorticosteroids (100%). Of all patients, 10 (42%) had osteopenia and another 3 (13%) had osteoporosis. Table 1 shows the baseline clinical characteristics of the 2 treatment groups. Patients in the denosumab group were older and more often male, with a higher total lumbar spine and total hip aBMD. The baseline biochemical parameters of bone metabolism were balanced between the treatment groups. Thus patients had similar levels of PTH, 25-hydroxy- and 1,25-dihydroxyvitamin D, and biomarkers of bone formation (P1NP and BSAP) and bone resorption (β-CTx, urine deoxypyridinoline). Patients in the denosumab and control groups received similar amounts of supplementary daily calcium (average dose 929 ± 226 vs 867 ± 309 mg) and vitamin D (average dose 1597 ± 580 vs 1476 ± 1114 IU).

Figure 1 depicts the course of the eGFR, and the changes of PTH, 25-hydroxyvitamin D, and 1,25-dihydroxyvitamin D, as well as P1NP, BSAP, β-CTx and urine deoxypyridinoline to creatinine ratios from baseline to months 12 in the denosumab and control groups. Consistent with the full trial findings, the eGFR remained stable in both groups. PTH (mean ± SD) decreased similarly in both groups from baseline to 12 months (-21 ± 47 vs -42 ± 83 ng/l), although the decrease appeared to be slightly delayed in the denosumab group. 25-hydroxyvitamin D and 1,25-dihydroxyvitamin D (mean ± SD) increased similarly in both groups (10.2 ± 9.6 vs. 13.5 ± 11.7 µg/l; and 29.7 ± 34.5 vs 14.0 ± 29.3 ng/l). The markers of bone formation (P1NP and BSAP) and bone resorption (β-CTx and urine deoxypyridinoline to creatinine ratios) decreased in the denosumab group and tended to increase in the control group.

DXA of the lumbar spine was performed in all study subjects at baseline and after 12 months to determine the changes in aBMD over the one year study period. Total lumbar spine aBMD of patients in the control group (n=14) showed a median percentage decrease of -1.6% (95% CI -4.7% - 0.8%), whereas patients in the denosumab group (n=10) showed a median increase of 2.4% (95% CI 2.2% - 4.3%) (p < 0.001 for median difference of 4.7%; 95% CI 2.6% - 7.8%). The corresponding T-score decreased by a median of -0.10 (95% CI -0.30 - 0.10) in the control group and increased by a median of 0.30 (95% CI 0.20 - 0.40) in the denosumab group (p < 0.001 for a median T-score difference of 0.45; 95% CI 0.30 - 0.70).

HRpQCT of the distal tibia and distal radius was performed in all study subjects at baseline and after 12 months. The baseline HRpQCT parameters are shown in Table 2, and the median percentage changes from baseline to 12 months in vBMD and cortical thickness are shown in Figure 2. At the tibia, Tot.vBMD increased in the denosumab group by a median of 2.2% (95% CI 0.7% - 3.2%), whereas it decreased in the control group by 0.3% (-3.3% - 2.7%; p=0.019 for between-group difference). At the radius, Tot.vBMD increased in the denosumab group by 1.3% (-1.9% - 2.2%), whereas it decreased by -1.6% (-5.2% - 0.7%) in the control group (p=0.031). Denosumab significantly increased the Ct.vBMD at the tibia (0.1%; -0.5% - 1.4% vs. -0.5%; -2.0% - 0.1%; p = 0.042), but the difference was not significant at the radius (-0.1%; -1.8% - 1.3% vs. -0.9%; -2.7% - 0.4%; p=0.212). Denosumab did also not significantly change Tb.vBMD at the tibia (1.8%; 1.3% - 4.2% vs. 1.1%; -2.5% - 4.1%; p = 0.371) and the radius (2.4%; 1.1% - 4.1% vs. 0.2%; -9.4% - 3.1%; p=0.108). The Ct.Th increased significantly in the denosumab group at the tibia (2.6%; 0.9% - 8.5% vs. -1.1%; -5.9% - 2.2%; p=0.006) and also tended to increase at the radius (0.5%; -6.4% - 1.6% vs. -3.5%; -8.3% - -1.2%; p=0.058).

Using extended cortical analysis, changes in Ct.Po were then assessed in the two treatment groups. Furthermore, micro-finite element analysis (µFEA) was performed to assess changes in estimated stiffness and failure load. Data are shown in Figure 3.

At the tibia and the radius the Ct.Po was characterized by a large variability but did not appear to significantly change over the 12 months study period. Thus, in the denosumab vs the control group the median percentage change in Ct.Po amounted to -0.5% (-5.1% - 5.8%) vs. 2.5% (-0.2% - 10.8%) (p=0.371) at the tibia, and to -6.9% (-16.8% - 0.8%) vs.-5.2% (-23.7% - 28.6%) (p=0.977) at the radius. Likewise, the cortical pore diameter did not significantly change from baseline to 12 months at both sites (data not shown).

At the tibia, µFEA estimates of bone stiffness significantly increased in the denosumab group by a median of 3.9% (-0.8% - 6.6%) but decreased by a median of -2.1% (-3.9% - 2.7%) in the control group (p=0.002). At the radius however, changes in bone stiffness were not significantly different in the denosumab and control groups (-0.6%; -4.1% - 5.3% vs. -3.1%; -8.2% - 4.0%; p=0.369). Hodges-Lehmann estimates for group differences were 5.6% (1.8% - 9.2%; p=0.002) at the tibia and 2.9% (-3.7% - 9.1%; p=0.369) at the radius.

A similar pattern was seen for estimates of failure load which significantly increased at the tibia in the denosumab group by a median of 3.8% (-0.0% - 5.9%) but decreased by a median of -1.7% (-3.6% - 1.7%) in the control group (p=0.002). At the radius, changes in failure load were not significantly different in the denosumab and control groups (0.5%; -3.8% - 4.3% vs. -2.2%; -6.7% - 4.3%; p=0.336). Hodges-Lehmann estimates for group differences were 5.1% (2.1% - 8.1%; p=0.002) at the tibia and 2.4% (-3.2% - 8.5%; p=0.336) at the radius.

Changes in Tb.Th, Tb.Sp, Tb.N at tibia and radius did not significantly differ between the treatment groups (Table 3).

In this bone microarchitecture ancillary study of the POSTOP trial, we show among 24 patients assessed with the additional measures of bone microarchitecture and estimated bone strength that denosumab had significant benefits on tibial bone microarchitecture and bone strength. A similar pattern - albeit to a lesser extent and not significant - was documented at the distal radius. In parallel with the increase of aBMD at the lumbar spine and a decrease of classical markers of bone turnover, denosumab significantly increased vBMD at the tibia and radius, and improved Ct.Th, stiffness and failure load at the tibia. Denosumab did not appear to change trabecular BMD and structure at both sites. Taken together and consistent with the full trial findings, denosumab appeared to have significant beneficial effects on bone health in de novo recipients of a renal transplant.

HRpQCT examination represents a valuable tool to examine various parameters of bone mineralization and structure at a three-dimensional level [18,19,20]. Due to the high resolution of the technique it is possible to obtain volumetric measurements of bone density (reflecting in large part the mineralization of the bone) at various skeletal sites and compartments (cortical vs trabecular). Furthermore, differential structural information on cortical and trabecular bone can be obtained. Using complex computer analyses (extended cortical analysis and so-called micro-finite element analysis; µFEA) it is furthermore possible to study changes in cortical porosity and biomechanical properties of the bone, including stiffness and failure load [20].

HRpQCT analyses have previously been performed in patients with chronic renal failure, including patients on hemo- and peritoneal dialysis [21,22,23,24]. These cross-sectional studies provided useful baseline data for various HRpQCT parameters in patients with increasing levels of renal failure. KTR have also been examined by HRpQCT, and longitudinal analyses over the first 12 months after transplantation demonstrated that these patients have a decline of the radial and tibial cortical area, density and thickness, and an increase in cortical porosity, stiffness and failure load [25,26]. At variance with these studies the more subtle changes which we documented in our control group transplant recipients may be explained by the consistent supplementation with vitamin D and calcium in our patients.

In different patient populations (mostly postmenopausal women) HRpQCT has been used to study the effect of several drugs on bone, including bisphosphonates, teriparatide and denosumab [27,28,29]. Our study is the first of its kind to use HRpQCT to examine the therapeutic effect of a drug in KTR, demonstrating beneficial effects of denosumab beyond those already shown by standard DXA in this patient population [11]. Our results are in large agreement with the results of the DATA-HRpQCT study, which also found stronger effects on bone mineralization and strength for denosumab in the tibia compared with the radius, and a lack of an effect on cortical porosity [29].

We conclude that denosumab had important short-term beneficial effects on bone mineralization, structure, microarchitecture and strength in first-year KTR. Whether these favorable changes will translate into long-term benefits such as a reduced incidence of fractures remains to be determined in adequately designed and powered clinical trials.

All authors declare that they have no conflicts of interest. There was no commercial funding for this study.

We thank the study patients for their participation. This work was supported by the University Hospital Zürich and the University Zürich.