Compromised Bone Microarchitecture and Estimated Bone Strength in Young Adults With Cystic Fibrosis

Abstract

Context:

Young adults with cystic fibrosis (CF) are at risk for low bone density and fractures, but the underlying alterations in bone microarchitecture that may contribute to their increased fracture risk are currently unknown.

Objective:

The main goal of this study was to use high-resolution peripheral quantitative computed tomography (HR-pQCT) to characterize the bone microarchitecture, volumetric bone mineral density (vBMD), and estimated strength of the radius and tibia in young adults with CF compared with healthy volunteers.

Design and Setting:

This was a cross-sectional study at an outpatient clinical research center within a tertiary academic medical center.

Participants:

Thirty young adults with CF, 18 to 40 years of age, were evaluated and compared with 60 healthy volunteers matched by age (±2 years), gender, and race.

Main Outcome Measures:

The primary outcomes were HR-pQCT–derived cortical and trabecular vBMD, bone microarchitecture, and estimates of bone strength.

Results:

At the radius and tibia, young adults with CF had smaller bone cross-sectional area and lower vBMD. Cortical and trabecular microarchitecture were compromised at both sites, most notably involving the trabecular bone of the tibia. These differences translated into lower estimated bone strength both at the radius and tibia. After accounting for body mass index differences, young adults with CF had lower bone area and estimated bone strength at the radius and had compromised trabecular microarchitecture and lower total and trabecular vBMD and estimated bone strength at the tibia. Alterations in trabecular bone density and microarchitecture and estimated strength measures of the tibia were also greater than expected based on dual-energy x-ray absorptiometry-derived areal BMD differences.

Conclusions:

Young adults with CF have compromised bone microarchitecture and lower estimated bone strength at both the radius and tibia, even after accounting for their smaller body size. These skeletal deficits likely explain the higher fracture risk observed in young adults with CF.

Over the past several decades, life expectancy for patients with cystic fibrosis (CF) has increased significantly. As patients live longer, other comorbidities related to this disease have emerged, including CF-related bone disease (1). The pathogenesis of the low bone mineral density (BMD) and increased fracture risk in patients with CF is currently not well understood. Obtaining new information about the underlying skeletal composition of these patients may be an important step in the monitoring, prevention, and potential treatment of this complication.

Fracture rates in patients with CF are higher than in healthy individuals, particularly rib and vertebral fractures (2–6). These fractures can have severe repercussions, leading to pain, kyphosis, chest wall deformities, reduced lung volumes, ineffective cough and airway clearance, and ultimately compromised lung function. In addition, severe bone disease may obviate lung transplantation. Due to this high morbidity, CF-related bone disease can significantly affect the health, well-being, and longevity of these patients.

Areal BMD, as measured by dual-energy x-ray absorptiometry (DXA), is low in patients with CF (7–10). DXA has important limitations, however, particularly in patients with CF. First, alteration in bone size introduces an artifact in DXA measurements such that values are reduced in smaller bones. Thus, in patients with CF, small bone size could account for their lower DXA values. Second, DXA cannot distinguish between trabecular and cortical bone (11, 12). Third, DXA may not predict fracture risk accurately in patients with CF, such that their risk of fracture is increased even with normal DXA areal BMD measurements (5).

High-resolution peripheral quantitative computed tomography (HR-pQCT) is a new imaging modality that has a resolution (voxel size) of 82 μm and allows for the noninvasive assessment of volumetric BMD (vBMD) and microstructure at peripheral skeletal sites (13–15). Furthermore, unlike DXA, HR-pQCT may be less affected by projection artifacts caused by bone size and adipose tissue. Finally, bone geometry and material characteristics can be integrated to provide an estimation of bone strength by applying microfinite element analyses (μFEA) to HR-pQCT images (16, 17).

The goals of this study were to use HR-pQCT to 1) characterize the bone microarchitecture, vBMD, and estimated strength of the radius and tibia in young adults with CF compared with healthy volunteers, 2) determine whether these HR-pQCT–derived parameters provide information independent of areal BMD that may help to explain the increased fracture risk in patients with CF, and 3) evaluate clinical characteristics of young adults with CF that may be associated with worsening bone structure and strength.

Subjects and Methods

Subjects and eligibility criteria

Young adults aged 18 to 40 years with CF were recruited from the Massachusetts General Hospital and Boston Children's Hospital Cystic Fibrosis Centers. Exclusion criteria included history of solid organ transplantation, current pregnancy, and Burkholderia dolosa infection (due to infection control issues). Data from healthy volunteers were obtained from studies enrolling subjects from the community to establish a normal database for HR-pQCT measures. Two healthy volunteers were matched to each subject with CF by age (within 2 years), race, and gender. Exclusion criteria included current pregnancy or a history of disorders known to affect bone metabolism. The protocol was approved by the Partners Healthcare Human Research Committee with ceded review by the Boston Children's Hospital Committee on Clinical Investigation. Written informed consent was obtained from all participants.

Clinical assessments

Subjects with CF were surveyed regarding medical history, medication use, history of fractures, pubertal and reproductive history, and lifetime glucocorticoid exposure. A registered bionutritionist assessed calcium and vitamin D intake using the validated Nutrition Data System for Research (18) and physical activity using the Modified Activity Questionnaire (19). Medical records were reviewed to obtain the most recent pulmonary function tests and CF genotype. When available, reported fractures were confirmed with x-ray reports. In all subjects, height was measured on a wall-mounted stadiometer and weight on an electronic scale. Race was self-reported.

Laboratory assessment

After fasting overnight, morning serum and urine samples were obtained in subjects with CF. Serum creatinine, calcium, and phosphorus levels were performed by Labcorp. The remaining labs were processed through the Massachusetts General Hospital Clinical Laboratory Research Core. Serum 25-hydroxyvitamin D (25(OH)D) levels were measured by chemiluminescent microparticle immunoassay (Abbott Diagnostics). Bone-specific alkaline phosphatase (BSAP) levels were obtained by enzyme immunoassay (Quidel Corporation). PTH and C-telopeptide (CTX) levels were measured by electrochemiluminescence immunoassay (Roche Diagnostics). Urine N-telopeptide (NTX) levels were measured by competitive-inhibition ELISA (Alere Scarborough Inc) and amino-terminal propeptide of type I collagen (P1NP) by quantitative RIA (Orion Diagnostica). References ranges for each assay were obtained from the manufacturer.

Assessment of areal BMD

Areal BMD of the lumbar spine in the posterior-anterior (PA) projection, total hip, femoral neck, and whole body (excluding the head) along with body composition were measured using DXA (Discovery A; Hologic Inc). A standard quality control program was employed that included daily measurement of a Hologic DXA anthropomorphic spine phantom and visual review of all images by an experienced investigator.

Assessment of vBMD, bone microarchitecture, and bone strength

The vBMD and microarchitecture of the distal radius and tibia were assessed by HR-pQCT (Xtreme CT; Scanco Medical AG) at the standard regions of interest using previously described methods (13–15). Quality control was maintained with daily scanning of the manufacturer's phantom. All HR-pQCT scans were reviewed for motion artifact and repeated if significant motion artifact was noted.

HR-pQCT scans were analyzed using Scanco software version 6.0 to provide total and trabecular vBMD (milligrams hydroxyapatite [HA]/cubic centimeter), trabecular thickness (millimeters), trabecular number (per millimeter), trabecular separation (millimeters), and trabecular distribution (millimeters). Cortical microarchitecture was characterized by processing HR-pQCT images with a semiautomated technique implemented in Scanco software (20–22). After image segmentation of cortical bone, the following measures were obtained: cortical vBMD (mg HA/cm³), cortical thickness (millimeters), cortical and trabecular area (square millimeters), and cortical porosity (percentage). μFEA was used to estimate radius and tibia biomechanical properties under uniaxial compression as previously described (23, 24), providing estimated stiffness (newtons per millimeter) and failure load (newtons).

In our laboratory, reproducibility for HR-pQCT measurements at the radius and tibia range from 0.2% to 1.4% for vBMD parameters, 0.3% to 8.6% for trabecular microarchitecture parameters, 0.6% to 2.4% for cortical microarchitecture parameters, 7.3% to 20.2% for cortical porosity measurements, and 2.1% to 3.0% for μFEA measures.

Statistical analysis

Statistical analysis was performed using SAS version 9.3 software (SAS Institute Inc). Clinical characteristics of subjects with CF and healthy volunteers were compared using independent-samples two-sided t tests for normally distributed data or Wilcoxon rank sum for nonnormally distributed data. Primary outcomes were HR-pQCT–derived microarchitectural and vBMD measures and μFEA results at the radius and tibia. The study was designed to match 2 healthy volunteers to each subject with CF to minimize effects of age, race, and gender on bone outcomes. To account for this matching, generalized linear models adjusting for each matched group were used to compare differences in means of areal BMD and HR-pQCT parameters between CF subjects and healthy volunteers. Multivariable linear regression was then performed adjusting for BMI, followed by an additional analysis adjusting for total hip areal BMD in addition to BMI. BMI was chosen as a covariate of interest based on clinical relevance and correlation with outcome variables (Spearman r = 0.05–0.52). Total hip areal BMD was chosen because this site represents a balance of both cortical and trabecular bone.

To evaluate associations between clinical characteristics and bone microarchitecture and strength among CF subjects, correlations were performed using Pearson correlation coefficient, or Spearman correlation coefficient in the case of nonnormally distributed data, between clinical covariates and bone outcomes (total vBMD and estimated failure load) among subjects with CF. Clinical covariates included 25(OH)D level, PTH level, bone formation and resorption markers, physical activity level, previous oral glucocorticoid use, and most recent pulmonary function as measured by percent predicted forced expiratory volume in 1 second (FEV1). Linear regression was used to determine whether the presence of F508del mutation was a significant predictor of bone strength in subjects with CF as well as to determine whether FEV1 effects on bone outcomes were independent of physical activity level.

Data are reported as mean ± SD unless otherwise noted, and P values < .05 are considered statistically significant.

Results

Cohort characteristics

Clinical characteristics of the healthy volunteers and the subjects with CF are presented in Table 1. The mean age of all subjects was 25 years, and 60% were female. Because a majority of patients with CF are Caucasian, all subjects with CF enrolled in this study were of Caucasian race. Young adults with CF were shorter, weighed less, and had a lower BMI than healthy volunteers. All women enrolled in this study were premenopausal.

Table 1.

Clinical and Demographic Characteristics of All Subjects

Clinical or Demographic Factor	Mean (SD) or n (%)		P Value
	Subjects With CF (n = 30)	Healthy Volunteers (n = 60)
Age, y	25.0 (5.0)	25.3 (4.3)	.80
Female	18 (60%)	36 (60%)
Caucasian race	30 (100%)	60 (100%)
Height, cm	165 (8)	170 (9)	.01
Weight, kg	57.0 (10.6)	71.6 (14.8)	<.01
BMI, kg/m2	20.8 (3.2)	24.5 (3.7)	<.01

Additional characteristics specific to subjects with CF are presented in Table 2. A broad range of cystic fibrosis transmembrane conductance regulator (CFTR) genotypes were represented in this study, although most subjects had at least 1 copy of the most common mutation, F508del. A majority of patients had pancreatic insufficiency requiring pancreatic enzyme replacement, and 17% had CF-related diabetes. Subjects reported excellent dietary and supplemental calcium and vitamin D intake. FEV1 was on average 75% predicted, indicating mild-to-moderate pulmonary impairment. Twenty-three percent of subjects had experienced at least 1 nondigital, nonfacial fracture, involving rib (3 patients), wrist (2 patients), or ankle (1 patient). One patient experienced vertebral compression fractures at T3 and T5 at age 23 years.

Table 2.

Clinical characteristics of subjects with CF.

Clinical Parameter	Mean (SD), Median (Range), or n (%)	Reference Range
CFTR Genotypes, n (%)
F508del/F508del	11 (37)
F508del/Other	12 (40)
Other	7 (23)
Most recent FEV1, % predicted, mean (SD)	73.2 (28.2)
Pancreatic insufficiency, n (%)	27 (90)
CF-related diabetes, n (%)	5 (17)
Calcium intake, mg/d, mean (SD)	2240 (1750)
Vitamin D intake, IU/d, mean (SD)	2380 (2490)
History of fracture, n (%)	7 (23)
Medication use in past year, n (%)
Oral glucocorticoids	9 (30)
Inhaled glucocorticoids	20 (67)
Proton pump inhibitors	21 (70)
Lifetime glucocorticoid exposure, mo, median (range)	7 (0–36)
Laboratory results
PTH, pg/mL	36 (14)	10–60
25(OH)D, ng/mL	28 (12–90)	>20
CTX, ng/mL	0.51 (0.22)	0.02–0.58
Urine NTX, nM BCE/mM creatinine	76 (19–236)	3–65
P1NP, μg/L	50 (19)	19–87
BSAP, U/L	29.1 (10.8)	11.6–41.3

Laboratory results in patients with CF are also presented in Table 2. Serum creatinine, calcium, and phosphorus levels were normal (data not shown). Six patients had a 25(OH)D level less than 20 ng/mL, and no patients had a level below 10 ng/mL. Median urine NTX/Cr level was above the normal range. Serum PTH, CTX, P1NP, and BSAP were within the expected reference range.

Areal BMD

Areal BMD measurements of the total hip, femoral neck, PA spine, and whole body were significantly lower in young adults with CF than age-, race-, and gender-matched healthy volunteers (Table 3). Seventy-three percent of young adults with CF had a Z-score of less than −1.0 at 1 or more sites, and 23% had a Z-score of less than −2.0. Although somewhat attenuated, differences remained significant after adjustment for BMI at all measured sites with the exception of areal BMD at the PA spine.

Table 3.

HR-pQCT Measurements of the Radius and Tibia in Subjects With CF and Healthy Volunteers^a

	CF	Healthy Volunteers	Mean Percent Differenceb	Unadjusted P Value	BMI-Adjusted P Value	BMI- and aBMD-Adjusted P Value
DXA aBMD, g/cm2
Total hip	0.870 ± 0.024	1.034 ± 0.017	−15.9	<.001	.007
Femoral neck	0.770 ± 0.024	0.936 ± 0.017	−17.7	<.001	.004
PA spine	0.973 ± 0.022	1.069 ± 0.016	−9.0	.001	.31
Whole body	0.970 ± 0.018	1.071 ± 0.014	−9.4	<.001	.042
Percent fat mass	25.9 ± 0.9	28.8 ± 0.7	−10.3	.011
Radius
Density, mg HA/cm3
Total vBMD	303.2 ± 8.4	332 ± 5.9	−8.7	.007	.27	.98
Cort vBMD	941.3 ± 5.0	952.6 ± 3.5	−1.2	.068	.68	.86
Trab vBMD	165.7 ± 5.5	186.7 ± 3.9	−11.3	.003	.08	.66
Area, mm2
Total area	263.9 ± 9.1	300.4 ± 6.4	−12.1	.002	.021	.21
Cort area	52.0 ± 1.7	61.6 ± 1.2	−15.6	<.001	.025	.43
Trab area	214.6 ± 8.9	241.7 ± 6.3	−11.2	0.015	.046	.25
Cort structure
Cort thickness, mm	0.823 ± 0.027	0.903 ± 0.019	−8.8	.021	.50	.81
Cort porosity, %	1.07 ± 0.13	1.31 ± 0.09	−18.0	.14	.06	.22
Trab structure
Trab thickness, mm	0.068 ± 0.001	0.076 ± 0.001	−9.8	.001	.011	.15
Trab number, mm−1	2.00 ± 0.04	2.06 ± 0.03	−2.7	.30	.83	.39
Trab separation, mm	0.44 ± 0.01	0.42 ± 0.01	+5.2	.14	.81	.31
SD Trab distribution, μm	0.18 ± 0.01	0.17 ± 0.01	+5.6	.33	.87	.31
Tibia
Density, mg HA/cm3
Total vBMD	286.3 ± 7.6	334.1 ± 5.3	−14.3	<.001	.002	.022
Cort vBMD	934.9 ± 6.9	958.6 ± 4.9	−2.5	.007	.13	.16
Trab vBMD	158.7 ± 5.0	202.9 ± 3.6	−21.8	<.001	<.001	<.001
Area, mm2
Total area	650.2 ± 23.8	739.3 ± 16.9	−12.1	.003	.08	.55
Cort area	109.3 ± 3.7	126.5 ± 2.6	−13.6	<.001	.15	.65
Trab area	544.7 ± 22.8	616.6 ± 17.2	−11.7	.013	.10	.58
Cort structure
Cort thickness, mm	1.19 ± 0.04	1.26 ± 0.03	−5.9	.10	.88	.62
Cort porosity, %	3.61 ± 0.33	2.96 ± 0.23	+21.9	.11	.22	.24
Trab structure
Trab thickness, mm	0.076 ± 0.002	0.0825 ± 0.002	−7.3	.025	.017	.07
Trab number, mm−1	1.71 ± 0.05	2.06 ± 0.034	−16.7	<.001	.001	.024
Trab separation, mm	0.52 ± 0.01	0.42 ± 0.01	+24.4	<.001	<.001	.010
SD Trab distribution, μm	0.23 ± 0.01	0.18 ± 0.01	+25.9	<.001	.040	.35

Abbreviations: aBMD, areal BMD; Cort, cortical; Trab, trabecular.

^aData are expressed as mean ± SE. Unadjusted group means are presented along with P values for the unadjusted analyses, BMI-adjusted analyses, and BMI- and total hip aBMD-adjusted analyses.

^bHealthy volunteers as reference.

HR-pQCT findings

Radius

In unadjusted analyses, subjects with CF had smaller bones, as indicated by the lower trabecular, cortical, and total bone cross-sectional areas (Table 3). Trabecular and total vBMD were also lower in patients with CF than healthy controls. Both cortical and trabecular thickness were lower in subjects with CF, but the remainder of cortical and trabecular microarchitecture, including cortical porosity and trabecular separation and number, were similar between the 2 groups. Subjects with CF also had lower stiffness and failure load than their healthy peers.

After adjustment for BMI, CF patients continued to have lower total, cortical, and trabecular area, although volumetric BMD differences were no longer significant. Trabecular thickness, estimated stiffness, and failure load also remained significantly lower in subjects with CF than healthy controls after BMI adjustment. The addition of total hip areal BMD to the multivariable model eliminated all differences between cohorts. Figure 1 illustrates μFEA results before and after multivariable adjustment.

Figure 1.

HR-pQCT scans of the tibia in a representative young woman with CF (A) and female healthy volunteer (B).

Tibia

Compromises in microarchitecture and vBMD were most notable at the tibia, and trabecular bone was predominately affected (Figure 2 and Table 3). In unadjusted analyses, subjects with CF had smaller bones and lower total, cortical, and trabecular vBMD. The lower trabecular vBMD was associated with fewer and thinner trabeculae along with wider separation and greater heterogeneity. Cortical microarchitecture was similar between groups. As in the radius, CF patients had significantly lower strength estimates of the tibia as measured by μFEA than their healthy peers (Figure 1).

Figure 2.

μFEA results. A and B, Stiffness (A) and failure load (B) of the radius and tibia in young adults with CF (black bars) and age-, gender-, and race-matched healthy volunteers (gray bars). Data are expressed as mean ± SE and are presented for the unadjusted analysis, the BMI-adjusted analysis, and the BMI- and total hip areal BMD-adjusted analysis. *, P < .05.

Adjustment for BMI attenuated some but not all of these differences. Bone size and cortical vBMD were no longer significantly different between groups; however, young adults with CF continued to demonstrate inferior trabecular bone characteristics, including lower vBMD, trabecular thickness, and trabecular number associated with greater trabecular separation and heterogeneity. Estimated tibial stiffness and failure load also remained lower in subjects with CF after adjusting for BMI differences.

Unlike the radius, deficits in trabecular bone of the tibia remained significant after accounting for total hip areal BMD, and patients with CF had lower estimated bone strength at this weight-bearing site independent of BMI and areal BMD (Figure 1).

Clinical predictors of vBMD and estimated bone strength

Within the CF cohort, total vBMD and estimated failure load of both the radius and tibia were moderately correlated with pulmonary status as measured by percent predicted FEV1 (Figure 3). Physical activity levels also were directly correlated with total vBMD of the radius (r = 0.39, P = .04) and tibia (r = 0.38, P = .04) and with failure load of the radius (r = 0.39, P = .04) and tibia (r = 0.41, P = .03). Percent predicted FEV1 remained a significant predictor of total vBMD and estimates of bone strength at the tibia even after adjustment for physical activity level (data not shown). Estimated failure load at the radius was inversely associated with PTH levels (r = −0.44, P = .01) with a similar but nonsignificant trend noted at the tibia (r = −0.33, P = .06). Failure load was positively associated with BSAP at the radius (r = 0.43, P = .01) and tibia (r = 0.40, P = .03). Vitamin D levels, previous oral glucocorticoid use, other bone turnover markers, and the presence of the F508del CFTR mutation were not significantly associated with these outcomes (data not shown). Due to the small number of patients with fractures (n = 7), this study was underpowered to detect differences in DXA or HR-pQCT outcomes in fracture vs nonfracture patients with CF.

Figure 3.

Correlations with FEV1. A–D, Scatter plots correlating percent predicted FEV1 with total vBMD of the radius (A) and tibia (B) and failure load of the radius (C) and tibia (D).

Discussion

Bone microarchitecture and vBMD are compromised in young adults with CF and are associated with lower estimated bone strength at both the radius and tibia. These skeletal alterations are greater than expected from the differences in BMI between patients with CF and their healthy peers, and differences in trabecular vBMD and trabecular microarchitecture at the tibia persisted even after adjustment for DXA-derived areal BMD of the hip, suggesting that DXA does not capture all of these bone structural and strength deficits in patients with CF. These findings may explain the increased fracture risk observed in patients with CF.

To our knowledge, this is the first study to characterize bone microarchitecture and predicted strength using HR-pQCT in young adults with CF. Histomorphometric data from iliac crest bone biopsies are limited but have suggested that patients with CF have decreased cortical and trabecular bone mass, with evidence of lower bone formation rates and increased osteoclastic activity (25–27). In one study assessing vBMD measured by standard pQCT (28), adolescents and young adults with CF had lower cortical thickness of the radius, consistent with our findings; however, unlike our study, trabecular and total vBMD were normal. This discrepancy may be explained by different study designs because comparisons were made to a reference database rather than the matched healthy controls that were used in our study. In addition, the lower resolution of pQCT (0.4 mm) is not ideal for assessing microarchitectural differences, particularly the detailed trabecular bone characteristics described in our study.

HR-pQCT studies involving other patient populations with secondary osteoporosis, such as acromegaly (29), have demonstrated that this technology is capable of providing important information about alterations in bone structure and strength. Deficits in HR-pQCT–derived cortical and trabecular parameters and in μFEA-derived strength measures described in the patients with CF in this study were of similar magnitude as those noted in patients with osteoporosis, chronic kidney disease, and incident fracture, suggesting that these differences may likely be of clinical significance (24, 30–32).

Young adults with CF in this study were shorter, weighed less, and had a lower BMI than matched healthy volunteers, as is commonly observed in this patient population. This size difference was reflected in the lower bone area at both the radius and tibia in these subjects. Lower weight applied to weight-bearing bones may also contribute to suboptimal bone structure and density, particularly if these patients are less physically active than their peers. The prominent deficits in trabecular and cortical bone structure and vBMD at the tibia observed in this study supports the hypothesis that weight bearing plays a role in these bone alterations.

To explore the confounding impact of body size on measures of bone density, size, and microarchitecture in patients with CF, we adjusted DXA and HR-pQCT values for BMI. Not only do adults with CF have smaller bones, but they also have inferior microarchitecture out of proportion to their lower BMI, particularly at the tibia, so that their estimated bone strength is lower than expected for their different body size. These factors may explain the increased fragility observed in this patient population.

Previous studies have demonstrated lower areal BMD in children and adults with CF, but these findings are difficult to interpret in the setting of their smaller body size. To determine whether HR-pQCT provides information independent of DXA, multivariable adjustment for areal BMD in addition to BMI was performed. In this analysis, differences in bone structure and strength at the radius were eliminated; however, subjects with CF continued to have compromised trabecular bone characteristics and lower estimated bone strength at the tibia even after normalizing for areal BMD differences. This finding suggests that alterations in the trabecular bone and estimated bone strength in the weight bearing bones of young adults with CF are greater than reflected in areal BMD findings, which may help explain the propensity to fracture trabecular-rich bones such as vertebrae. Further prospective studies will be required to determine whether HR-pQCT can potentially improve fracture prediction in this population.

There are many possible explanations for the bone alterations observed in patients with CF, including vitamin D deficiency, malabsorption, physical inactivity, malnutrition, glucocorticoid use, hormonal deficiencies, delayed puberty, inflammation, CF-related diabetes, and possibly CFTR dysfunction itself (1, 33, 34). The differences in vBMD and bone area observed in the patients with CF in our study, despite their young age, may provide support to the theory that they may fail to attain peak bone mass, predisposing to more fragile bones later in life. In this study, physical activity levels were positively associated with bone strength estimates at the radius and tibia, suggesting that a reduction in physical activity may play a role in CF-related bone disease. In addition, failure load at the radius and tibia were directly associated with measures of pulmonary function in patients with CF, a finding that has also been noted in other studies evaluating areal BMD and pulmonary function (2, 35, 36). Interestingly, in our study, FEV1 remained a predictor of tibial bone strength independent of physical activity level, implying that other factors associated with declining lung function, such as inflammation, poor nutrition, or chronic respiratory acidosis, may be contributing. Lastly, bone-specific CFTR genotype-phenotype correlations have not yet been fully elucidated, and in this cohort, F508del mutation was not a predictor of estimated bone strength, although the power to detect such differences was limited.

In this cohort of patients with CF, vitamin D deficiency did not appear to be a significant issue. Most patients had a serum 25(OH)D level greater than 20 ng/mL. Serum 25(OH)D levels were not significantly associated with vBMD or bone strength measures, possibly because reported vitamin D intake was high enough to produce adequate levels in this ambulatory relatively healthy cohort. Vitamin D deficiency and calcium malabsorption may be more common and have a greater influence on bone health in more severely ill patients. Interestingly, PTH levels were inversely associated with failure load at the radius, with a similar although nonsignificant trend at the tibia. Higher PTH values may reflect inadequate calcium intake or calcium malabsorption that is independent of vitamin D. Lower BSAP levels were also directly associated with lower failure load in subjects with CF. Urine NTX levels were elevated, suggesting increased bone resorption in these subjects, which is consistent with previous reports (37, 38). Further studies investigating the underlying cause of these bone turnover abnormalities and associations with bone structure will be required to understand the significance of these alterations.

The strengths of this study include the relatively homogeneous cohort of subjects with CF, limited to an ambulatory group of young adults evaluated before lung transplantation. In addition, by assigning each subject with CF to 2 healthy volunteers matched by age, race, and gender, the effects of these covariates on the outcomes were minimized. Our study also has limitations. The sample size was relatively small, and it is possible that additional bone abnormalities would be uncovered if a larger number of patients were evaluated; however, the significant findings that were noted suggest that there was adequate power to identify several important differences. In addition, only peripheral skeletal sites were evaluated in this study, and further studies are needed to determine whether findings are also representative of other skeletal regions. Finally, the cross-sectional design of this study cannot capture the dynamic changes that occur in the skeleton across childhood and the young adult years nor account for the effects of clinical variables occurring in the years before this evaluation.

In conclusion, several measures of bone microarchitecture are compromised in young adults with CF, and these alterations appear to contribute to lower estimated bone strength compared with age-, race-, and gender-matched healthy volunteers. These findings may explain the increased fracture risk observed in this patient population. Although some of these differences are eliminated by adjustment for BMI and areal BMD, many persist after these adjustments, suggesting that HR-pQCT provides additional unique information on bone integrity in patients with CF. Further studies are needed to understand the underlying pathophysiology and clinical implications of these bone alterations.