Bone density, microarchitecture and estimated strength in stone formers: a cross-sectional HR-pQCT study Free

ABSTRACT

Background

Low areal bone mineral density (BMD), increased fracture risk and altered bone remodeling have been described among stone formers (SFs), but the magnitude of these findings differs by age, sex, menopausal status and urinary calcium (uCa). This study aimed to investigate volumetric BMD (vBMD), bone microarchitecture and biomechanical properties by high-resolution peripheral quantitative computed tomography (HR-pQCT) and finite element analysis (FEA) in young SFs, irrespective of calciuria, further distinguishing trabecular from cortical compartments.

Methods

HR-pQCT/FEA was performed at the distal tibia (DT) and distal radius (DR) in 106 SFs (57 males and 49 premenopausal females; median age 37 years) and compared with 106 non-SFs (NSFs) retrieved from an existing database, matched for age, sex and body mass index (BMI). Biochemical/hormonal serum and urinary parameters were obtained from SFs.

Results

SFs exhibited significantly lower trabecular number (TbN) and higher trabecular separation (TbSp) than NSFs at both anatomical sites and lower cortical porosity in the DR. In a subgroup analysis separated by sex, female SFs presented significantly lower TbvBMD, relative bone volume fraction (BV/TV) and TbN and higher TbSp than NSFs at both sites, while male SFs showed significantly lower stiffness and failure load. Multivariate analysis showed TbN to be independently associated with sex and BMI at both sites and with uCa at the DR.

Conclusions

The present findings suggest that bone disease represents an early event among SFs, associated at least in part with calcium excretion and mainly characterized by trabecular bone microarchitecture impairment, especially among women, but with reduced bone strength parameters in men.

INTRODUCTION

Nephrolithiasis is associated with low bone mineral density (BMD) and increased fracture risk worldwide [1–4]. Genetic [5], dietary [6] and hormonal factors [7] and increased bone expression of cytokines and receptor activator of nuclear factor κB ligand (RANKL) play important roles in nephrolithiasis-associated bone disease [7–11]. However, there are still conflicting data regarding the pathogenic effects of age, sex and menopausal status on bone loss in this population. Some studies have reported increased bone fracture risk in both sexes [2–4], while others detected it only in male stone formers (SFs) [12] but not in female SFs [13, 14]. Moreover, although several studies indicate the presence of hypercalciuria as a paramount risk factor [15–17], it is still a controverted matter [6, 18].

Previous histomorphometric analyses by our group and others [8, 10, 11, 19, 20] disclosed altered bone remodeling characterized by reduced bone formation, high bone resorption and a mineralization defect. Dual-energy X-ray absorptiometry (DXA) [18, 19] and quantitative computed tomography (QCT) [21] previously suggested trabecular bone to be predominantly affected.

The advent of high-resolution peripheral QCT (HR-pQCT) enabled not only trabecular and cortical bone compartment volumetric BMD (vBMD) measurements but also assessment of bone microarchitecture [22] and biomechanical properties estimation by finite element analysis (FEA) [23]. Both HR-pQCT and bone strength parameters by FEA have been strongly and better related to fracture risk [22, 23]. Interestingly, their measurement from peripheral sites correlated with central skeletal sites [24]. Secondary osteoporosis (e.g. rheumatoid arthritis, type 2 diabetes mellitus, CKD, lupus erythematosus, etc.) and metabolic disorders have been studied with HR-pQCT, but not nephrolithiasis [25]. The present study aimed to determine vBMD, bone microarchitecture and biomechanical properties by HR-pQCT/FEA in young male and exclusively premenopausal female SFs, focusing on possible differences in trabecular and cortical compartments and further investigating whether urinary calcium (uCa) is associated with any alteration.

MATERIALS AND METHODS

Participants

A total of 223 SFs were referred to the nephrolithiasis outpatient clinic at Universidade Federal de São Paulo from January 2016 to April 2019. The diagnosis of kidney stones was confirmed using noncontrast computed tomography (CT) scans. Only premenopausal women ages 18–49 years and men ages 18–60 years with an estimated glomerular filtration rate (eGFR) ≥90 mL/min/1.73 m² [calculated using the Chronic Kidney Disease Epidemiology Collaboration (CKD-EPI) equation], naïve to medical therapy, with evidence of radiopaque calculi on CT scout images consistent with calcium composition were considered eligible. The exclusion criteria were pregnancy, hypercalcemia, nephrocalcinosis, cystinuria, hyperoxaluria, diabetes, inflammatory intestinal diseases, neoplasia, diseases affecting bone metabolism such as primary hyperparathyroidism, renal tubular acidosis, rickets during childhood, osteomalacia or abnormal height as adults, as well as previous treatment with thiazides, corticosteroids and anticonvulsants. SFs were matched 1:1 with non-SFs (NSFs), retrieved from a local database of 244 individuals [26], using propensity score matching accounting for age, sex and body mass index (BMI). Smoking, use of oral contraceptives or use of other medications was derived from questionnaires. The study protocol was approved by the local Ethics Advisory Committee of the Universidade Federal de São Paulo, in accordance with the Helsinki Declaration of 1975 as revised in 2013. Written consent was obtained from all participants.

HR-pQCT

Trabecular and cortical vBMD (TbvBMD and CtvBMD) and microarchitecture were determined in all participants by HR-pQCT (XtremeCT, Scanco Medical, Brüttisellen, Switzerland) at the nondominant distal radius (DR) and distal tibia (DT), with a spatial resolution of 82 µm, as previously described [26]. Variables evaluated by HR-pQCT included total TTbvBMD and Ct vBMD; trabecular microarchitectural parameters [trabecular number (TbN), thickness (TbTh), separation (TbSp) and relative bone volume fraction (BV/TV)] and cortical microarchitectural parameters [cortical thickness (CtTh) and porosity (CtPo)].

FEA

Bone strength was estimated by FEA, a computer modeling technique that converts each voxel from HR‐pQCT images into a cubic finite element (Image Processing Language and FE Extension IPLFE software; Scanco Medical), resulting in a model that represents the complex bone structure and estimates biomechanical properties. A virtual mechanical test applies a compressive force to the bone [26, 27], determining the bone strength parameters stiffness and failure load.

Biochemistry, hormonal and nutritional data

SFs were instructed to collect two nonconsecutive 24-h urine samples under an unrestricted diet to determine calcium, sodium, citrate, oxalate, phosphorus and urea and averaged values were used for analyses. Urinary sodium was used to calculate sodium chloride (NaCl) intake and urinary urea to estimate protein intake through the protein equivalent of nitrogen appearance (PNA) formula. Urinary creatinine measurement adjusted for body weight was utilized to determine the accuracy of 24-h urine sample collection. Calcium intake was assessed through a 3-day dietary record (Dietpro 6.0; Agromídia Software, Minas Gerais, Brazil) delivered with the second urine sample. A fasting blood sample was drawn to measure creatinine, ionized calcium, phosphorus, parathyroid hormone (PTH), bone alkaline phosphatase (BAP), 25-hydroxyvitamin D, 1,25-dihydroxyvitamin D₃, fibroblast growth factor 23 (FGF23), sclerostin, C-terminal crosslinks (CTx) and N-pro-peptide of type I collagen (P1NP)_. Creatinine levels were determined according to a modified Jaffe's reaction, using an isotope dilution mass spectrometry (ID-MS)-traceable method. All biochemical parameters were measured using a Beckman Clinical Chemistry Analyzer (AU480; Olympus America, Center Valley, PA, USA). Vitamin D metabolites and PTH were determined by chemiluminescent immunoassays (Architect, Abbott Laboratories, Abbott Park, IL, USA). FGF23 was determined by enzyme-linked immunosorbant assay (ELISA) (Immutopics, San Clemente, CA, USA), as were BAP and sclerostin (Quidel, San Diego, CA, USA). CTx and P1NP were measured by electrochemiluminescence (Roche Diagnostics, Indianapolis, IN, USA).

Statistical analysis

Statistical analyses were performed using SPSS version 23.0 (IBM, Armonk, NY, USA). In all analyses, P < 0.05 was considered significant. Variable distribution was evaluated by Kolmogorov–Smirnov test. Categorical variables are presented as n (%), normally distributed variables as mean ± standard deviation (SD) and nonnormally distributed ones as median [interquartile range (IQR)]. Comparison of categorical variables was performed using a chi-squared test. Differences between groups were tested through the Student's t test or Mann–Whitney test according to variable characteristics. HR-pQCT/FEA from SFs and NSFs were compared using a propensity-matched analysis (1:1), creating propensity scores for each sex through a non-parsimonious logistic regression model with kidney stone presence as the dependent variable and age and BMI as independent variables and a greedy matching algorithm with a caliper width of 0.02. Possible determinants of TbN at the DT and DR in SFs were studied using univariable linear regression. Variables of known clinical importance for bone mass, such as age, sex, BMI, dietary calcium, sodium, protein, serum levels of PTH, sclerostin, P1NP, CTx, BAP, FGF-23, 25-hydroxyvitamin D and 1,25-dihydroxyvitamin D were included in the univariate analysis. Subsequently, all variables with a P-value <0.10 were included in a multivariable linear regression model to identify independent clinical determinants of TbN at the DT/DR. Moreover, in order to study potential associations between TbN with TbvBMD and other cortical parameters at the DT/DR in the entire population (SFs + NSFs), a multivariate linear regression analysis adjusted for age and BMI was performed. Residuals were checked for normality and variables were natural log-transformed when appropriate.

RESULTS

From the 223 SFs initially evaluated, 169 were eligible based on the inclusion/exclusion criteria, of which 35 refused to participate, 16 were excluded due to inaccurate 24-h urine collection (urinary creatinine <18 mg/kg/day in males and <15 mg/kg/day in females) and 12 were lost to follow-up, giving a total of 106 SFs (57 males and 49 females). Table 1 shows the mean biochemical and hormonal parameters. The study population was characteristically endowed with normal renal function (eGFR ≥ 90 mL/min/1.73 m²) and no alteration in calciotropic hormones. In our sample, 3 of 106 SFs presented hypophosphatemia, i.e. serum phosphorus <2.5 mg/dL (data not shown in the table).

Table 2 shows baseline characteristics and HR-pQCT and FEA parameters at the DT and DR of all SFs and NSFs. Matching by age, sex and BMI rendered both groups similar for statistical comparisons and the percentage of smokers and oral contraceptive users among women did not differ statistically between groups. vBMD analysis revealed a trend for lower TbvBMD at the DT in SFs versus NSFs (P < 0.06). The trabecular microarchitectural analysis revealed significantly lower TbN and higher TbSp in SFs compared with NSFs at both anatomical sites, whereas TbTh was not statistically different. At the cortical compartment, SFs presented a significantly lower CtPo at the DR. A trend for lower stiffness (P = 0.06) and failure load (P = 0.05) was detected by FEA at the DT in SFs versus NSFs but not at the DR. Subgroup analyses by sex are shown in Table 3. Among SF males, the CtTh and CtPo of SFs did not differ from NSFs at the DT but were significantly lower at the DR. However, stiffness and failure load were significantly lower in SFs at both the DT and DR. Among females, SFs exhibited significantly lower TbvBMD, BV/TV and TbN and significantly higher TbSp than NSFs at both sites.

Table 4 synthesises the significant differences between SFs and NSFs with respect to bone microarchitecture and strength parameters.

Table 5 shows the potential determinants of TbN at the DT and DR as dependent variables, given it was the parameter more uniformly distinguishable by statistical analysis between SFs and NSFs. At the DT, the univariate linear regression analysis revealed direct associations of TbN with BMI, PNA and serum sclerostin and inverse associations with uCa and female sex, but only BMI and sex remained independently associated in the multivariate analysis. At the DT, the univariate linear regression analysis revealed direct associations of TbN with BMI, PNA, serum PTH, sclerostin and 25-hydroxyvitamin D and inverse associations with uCa and female sex. The multivariate analysis showed sex, BMI and uCa as potential determinants of TbN at the DR. The mean daily intakes of NaCl and protein (as addressed by 24-h urinary collection and PNA) were 10.0 ± 4.0 g/day  and 1.1 ± 0.3 g/kg/day,  respectively. The mean daily intake of calcium was 511 ± 214 mg/day (Table 5).

Another multivariate linear regression analysis adjusted for age and BMI was applied to investigate potential associations between TbN with TbvBMD and other cortical parameters at the DT and DR in the entire population (SFs + NSFs) divided by sex, since previous analyses have suggested gender-specific alterations (Table 6). Interestingly, while there was a direct association between TbN and TbvBMD at both sites in both females and males, there was an inverse association between TbN and CtvBMD at both sites among females and only at the DR among males. Accordingly, TbN and CtPo demonstrated a consistent direct association in all analyses. There was a significant association between TbN and CtTh only at the DT among females.

DISCUSSION

To the best of our knowledge, the present study is the first to apply HR-pQCT and FEA in a relatively young population of SFs, including only premenopausal females. Trabecular bone microarchitectural impairment was observed in SFs, especially in women, and uCa was negatively associated with a TbN at the DR in the SF group as a whole. There was a trend for lower bone strength at the DT in the whole sample, but a significant reduction was evidenced only in male SFs.

The current findings in the trabecular compartment are in line with those of an earlier histomorphometric study from our group [11] that revealed a lower TbN and higher trabecular separation in hypercalciuric SFs compared with healthy subjects.

Although the present results are in accordance with DXA studies indirectly revealing trabecular sites as preferentially affected, experimental data on genetic hypercalciuric rats demonstrated reduced trabecular volume and thickness, with more brittle and fracture-prone bones coupled with more compromised mechanical properties in the cortical bone as well [28].

HR-pQCT, a noninvasive technique, is capable of measuring trabecular microarchitecture and vBMD in vivo, providing additional information to DXA and better distinguishing cortical from trabecular components at the DR and DT.

In the present series, trabecular alterations at both anatomical sites, namely TbSp, TbN and TbvBMD were disclosed in SFs and the significant associations of TbN with uCa by univariate analysis at the DR and DT seem to suggest the trabecular bone as the preferred source of calcium, given its higher metabolic activity compared with cortical bone. However, the multivariate analysis showed the TbN to be independently associated with uCa only at the distal radius. As BMI was shown to be a strong predictor of TbN at both sites, it can be speculated that the mechanical loading to the tibia bones might have counteracted the calciuric effect at this site. On the other hand, calciuria may not reflect a causal effect of bone loss in SFs but rather the consequence [29]. In the general population, age, gender and BMI are known determinants of TbN [25].

Vezzoli et al. [30] found a higher proportion of low TbvBMD in the general population belonging to the higher uCa tertiles, whereas elderly men in the Osteoporotic Fractures in Men study [31] did not exhibit an association of osteoporosis with hypercalciuria. Asplin et al. [16, 32] observed an inverse correlation between uCa and spine and femoral z-scores among SFs but not NSFs. Conversely, Sakhaee et al. [18] did not find correlations between uCa and BMD in a large group of calcium SFs of all ages, except for the estrogen-untreated postmenopausal group. Of note, the present series differs from all previous studies because of the exclusion of postmenopausal female SFs and men >60 years of age.

Subgroup analysis by sex indicated that male SFs presented no significant alterations in trabecular architecture, significantly reduced CtTh at DR and rather better mean values of CtPo than NSFs, whereas significantly reduced strength parameters by FEA where disclosed among them at both sites. The underlying mechanism for such dissociation could not be explained through the other parameters evaluated by HR-pQCT. Nevertheless, since FEA has the advantage of assessing bone mechanical properties as a whole, contrary to the individual analysis of each variable by HR-pQCT, it provides better insights into the final product of bone quality. However, although trabecular impairment in male SFs was not depicted here, it has been recognized that even small microarchitectural alterations may considerably affect load distribution, causing a great impact on bone mechanical properties [23, 33]. In addition, variables not assessed in the present study (e.g. geometry, accumulated damage and material properties) might also have accounted for the reduction of bone strength in male SFs. Corroborating these hypotheses, an epidemiological study from the Framingham cohort demonstrated a strong correlation between failure load and CtTh, TbvBMD, BV/TV, TbN and some geometric variables (cortical area and total area) [34].

Conversely, the female SF subgroup analysis revealed significantly reduced density and microarchitectural trabecular parameters, whereas strength parameters assessed by FEA remained preserved. Notwithstanding the unknown reasons for such sex-related differences, we speculate that cortical bone might have exerted a protective effect. A comparison between sexes was performed to test this hypothesis (data not shown in table), showing that female SFs presented better cortical parameters than male SFs at both anatomical sites: higher CtvBMD (DT: P < 0.001; DR: P < 0.001) and lower CtPo (DT: P < 0.001; DR: P < 0.001). This finding agrees well with a large Canadian cohort in the general population, focusing on sex-specific differences in the cortical and trabecular compartments, wherein young women presented with lower CtPo and higher CtvBMD than young men [35].

Moreover, in the current series, a compensatory increase in cortical parameters induced by load distribution changes caused by trabecular alterations might have contributed to preserving strength in the female population. In the multiple linear regression we observed a consistent association between TbN and CtPo in all analyses and a significant negative association between TbN and CtvBMD at both sites in the female population, demonstrating that loss of TbN was associated with enhanced cortical parameters, namely CtPo and CtvBMD. One possible explanation relies on the fact that the bone has a sophisticated mechanosensory system that responds to strain to adjust its mineral content and architecture to maintain strength. The Wnt/β-catenin pathway participates in the bone response to mechanical strain and requires estrogen receptor α to be effective [36]. In addition, estrogen has a particular influence on the cortical bone [37, 38]. In line with our findings, previous studies have also observed a net increase in CtvBMD associated with decreasing trabecular parameters in healthy premenopausal women [38, 39]. As mentioned, Sakhaee et al. [18] also demonstrated the relevance of estrogen for bone quality in female SFs. Therefore, the better adaptive response observed in female SFs compared with male SFs could be attributed to an estrogen-replete state.

Given that failure load can better predict bone strength and fracture risk than bone density [23, 40–42] or microarchitectural parameters from HR-pQCT [40], the reduction in bone strength observed in this young male SFs population might suggest an increased risk for future incident fractures [23, 40]. A recent study reinforced the hypothesis that bone disease in male SFs may start early in life, as demonstrated by the lower body BMD z-score in male adolescent SFs compared with controls, while no significant differences were observed among females [43]. Cross-sectional data from the Third National Health and Nutrition Examination Survey (NHANES III) showed an association between the history of nephrolithiasis and incident fractures in men, but not in women [12], while incident fractures in postmenopausal women with nephrolithiasis were not detected in the Women's Health Initiative study [13]. In contrast, Denburg et al. [3], in a huge population-based cohort study using The Health Improvement Network (THIN) database, showed a high fracture risk among women with a history of kidney stones at all ages and the highest risk among those ages 30–39 years. It should be noted that TbN, which was reduced in the female SFs, has been demonstrated as a significant predictor of incident fractures in a series of studies, including the Bone Microarchitecture International Consortium [40], which showed that TbN was also one of the bone measures most strongly associated with the risk of fractures. Summarizing the findings of the present study, failure load and TbN were significantly lower in both sites for male and female SFs versus NSFs.

A low calcium intake was observed in the present series of SFs, which might have contributed to the alterations of bone parameters disclosed. Although dietary intake of NSFs was unavailable for comparison, a significant difference among groups would not be expected, considering the usual low calcium consumption found in the Brazilian general population [44]. Although a high mean NaCl intake here depicted by SFs could have played a role in bone loss by inducing a negative calcium balance [45], it was not associated with the TbN and neither was PNA.

Our data suggest early evaluation of bone microarchitecture and strength could identify patients at risk for future osteoporotic fractures, which may have major practical implications since it is easier to prevent bone loss at an earlier stage than to restore it [46]. The absence of any biochemical surrogate markers associated with trabecular alterations, as demonstrated in the linear regression analysis, strengthens the need for a direct evaluation of bone parameters in SFs. HR-pQCT is an emerging and promising technique, as it can detect early changes in bone quality and assess vBMD and microarchitecture of both trabecular and cortical compartments. Moreover, it presents a lower effective radiation dose (3–5 μSv) compared with DXA (9 μSv) and may be able to discern patients with moderately low BMD in regard to the level of their microstructural deterioration. Nevertheless, the current clinical application of HR-pQCT still presents some obstacles, as it is not widely available and the cost is about twice that of DXA. However, a recent study has demonstrated that it is cost-effective to treat women with osteopenia and severe microstructural deterioration [47]. Anyway, normative datasets are still needed to guide the interpretation of the results of each parameter at an individual level and not only among distinct populations.

The present study has some limitations. First, the cross-sectional design could not determine a cause–effect relationship between nephrolithiasis and bone alterations. Information about potential confounding factors such as alcohol intake, race and physical activity were not obtained and sex hormone measurements were not performed. In the NSF database, nutritional and biochemistry data were not available. Calcium composition stones were an assumption based on radiopacity seen in all CT scout images, but some SFs of the cohort might have had noncalcic or mixed kidney stones, since stone analysis was not obtained from most of them. Renal phosphate leak has been described among calcium nephrolithiasis patients and/or bone loss [48–50]. In the present series, one of three SFs did present hypophosphatemia, low BMD with DXA and uCa and serum FGF23 levels of 235 mg/24 h and 65 pg/mL, respectively. Nevertheless, since we did not perform a systematic genetic panel nor functional studies in the current study, it cannot be ruled out that allelic variants of genes coding for NPT2a, FGF23, NHERF1 and CYP24A1, among others, might have been associated at least in part with either bone or stone phenotypes.

In conclusion, present findings suggest that bone disease in SFs is mostly characterized by early trabecular impairment, especially concerning microarchitecture, with an association to calcium excretion. Sex-related differences were disclosed, namely bone strength reduction only in male SFs. Further studies are warranted to confirm whether these sex-related differences are due to the beneficial effects of estrogen on cortical bone among premenopausal women with nephrolithiasis.

ACKNOWLEDGEMENTS

We would like to thank Liliam Takayama and Lysien I. Zambrano for technical assistance. Portions of this study were presented as a poster at the Annual Meeting of the American Society of Nephrology (7–10 November 2019; Washington, DC, USA).

FUNDING

This study was supported by Fundação de Amparo à Pesquisa do Estado de São Paulo (grant 2016/25359-0), Conselho Nacional de Desenvolvimento Científico e Tecnológico [grant 309045/2018-5 (to I.P.H.), grant 305556/2017-7 (to R.M.R.P)], Coordenação de Aperfeiçoamento de Pessoal de Nível Superior and Fundação Oswaldo Ramos—Hospital do Rim.

AUTHORS’ CONTRIBUTIONS

P.L.G.E. and I.P.H were the study leads and wrote the first draft of the manuscript. P.L.G.E., I.P.H and R.M.R.P. contributed to the study conception and design. Material preparation and data collection were performed by P.L.G.E, T.L.M, M.S.O., J.C.A., V.F.C. Analysis and interpretation of data were performed by P.L.G.E, F.G.R., C.M.C., A.B.C., R.M.R.P and I.P.H. All authors provided intellectual content of critical importance, revisited the article and provided the final approval of the version to be published.

CONFLICT OF INTEREST STATEMENT

The authors declare no conflicts of interest. The results presented in this article have not been published previously in whole or part, except in abstract format.

DATA AVAILABILITY STATEMENT

The data in this article are available from the corresponding author upon reasonable request.

REFERENCES