Between 1986 and 1987, 9704 white women ages 65 and older were recruited for the Study of Osteoporotic Fractures at four metropolitan areas in the United States, including 2401 women from the Monongahela Valley near Pittsburgh, PA. Age-eligible participants were contacted by mail using community-based listings such as HMO membership and voter registration. In 1997, 662 African American women were added to the national study, including 177 at the Pittsburgh site. The current analysis was limited to the 258 surviving members of the Pittsburgh cohort (ages 79 to 96 years, mean 87.0 years) who attended a clinic visit in 2006 to 2008, a median of 20 years after their baseline examination (Fig. 1).
Measurements at the follow-up visit
Serum samples were collected in a fasting state and after processing were stored at −80°C until assay. Sclerostin assays were carried out at the Heinz Nutrition Laboratory at the Graduate School of Public Health, University of Pittsburgh. Sclerostin levels were measured by ELISA sandwich assay (Biomedica Medizinprodukte GmbH & Co. KG, Wien, Austria), which has a sensitivity ranging from 50 to 50,000 pg/mL and coefficients of variation (CV) between 4.1% and 9.8%. Serum sclerostin measured with this assay has been shown to correlate with levels in bone marrow. Serum-b-Crosslaps (CTX) was measured on Elecsys 2010 (Roche GmbH, Mannheim, Germany) by electrochemiluminescence immunoassay (ECLIA), a sandwich assay measuring ranges of 0.010 to 2.880 ng/mL with intra- and interassay variation of <4.7% and 4.8%, respectively. Serum procollagen type 1 N-terminal propeptide (PINP) was also measured on Elecsys 2010 (Roche GmbH) by ECLIA, measuring ranges of 5.0 to 437.2 ng/mL with intra- and interassay variations of <2.1% and 2.1%. Testosterone and estradiol were quantified in serum (0.5 ml) by previously described radio-immunoassay (RIA) methods.[25-27] Before the RIAs, steroids are extracted with hexane:ethyl acetate (3:2), and then testosterone and estradiol are separated from each other and their metabolites by Celite column partition chromatography. Appropriate tritiated internal standards are added to each serum sample before the extraction step to follow and correct for procedural losses. The assay sensitivities for the testosterone and estradiol RIAs are 1.5 ng/dL and 2 pg/mL, respectively, and the interassay CVs for these assays are 8%, 12%, and 12% at 13, 30, and 96 ng/dL, and 11%, 13%, and 12% at 15, 36, and 101 pg/mL, respectively.
Analysis of sex hormone–binding globulin (SHBG) is carried out by chemiluminescent immunometric assay on the Immulite analyzer (Siemens Healthcare Diagnostics, Deerfield, IL, USA). The assay utilizes monoclonal murine anti-SHBG attached to a bead and a polyclonal rabbit anti-SHBG conjugated to alkaline phosphatase. The assay sensitivity is 1 nmol/L, and the interassay CV is 9.1% at 69 nmol/L.
Free and bioavailable (non-SHBG-bound) testosterone and estradiol are calculated using the measured total testosterone and estradiol levels, respectively, and SHBG concentrations as well as an average assumed concentration for albumin.[28, 29] This method has been found to have high validity. Blind duplicates from 26 randomly selected women showed excellent correlations for all biomarkers (r = 0.87–0.99, p < 0.05). P1NP and CTX were measured at Synarc (Lyon, France).
Measurements obtained at the year 20 visit included anthropometry, vital signs, physical function, and hip and whole body areal (aBMD). BMI was calculated as weight in kilograms (measured on a balance beam scale) divided by height (measured by Harpenden stadiometer) squared. Vital signs were obtained using a digital ear thermometer (Omron, Lake Forest, IL, USA, MC-514) and automatic blood pressure monitor (Omron HEM-780). Physical function evaluation utilized the short physical performance battery (SPPB) developed by Guralnik and colleagues, which included grip strength (measured by Preston Grip or TEC dynamometer), gait speed, single and multiple chair stands, and tandem stands. BMD was measured by dual-energy X-ray absorptiometry (Hologic QDR 4500W, Bedford, MA, USA) for the hip and whole body.
Self-reported functional status was obtained by in-clinic interview. Participants were asked to bring into the clinic all their prescription and over-the-counter medications for review by study staff. Information on medical history, lifestyle, and tobacco use was collected by a self-administered questionnaire.
Peripheral quantitative computed tomography
The Stratec Three XCT-2000 pQCT (Stratec Medizintechnik, Pforzheim, Germany) was used to perform scans on different sites: 4% (T4%), 33% (T33%), and 66% (T66%) of the total length of the tibia. Phantom scans were performed on a daily basis as a quality-control measure. Trained technicians followed a standardized protocol for scanning, using anatomical landmarks to optimize patient positioning in the machine. Tibia length was measured from medial malleolus to medial epicondyle. Scans at 4% of tibial site represent predominantly trabecular bone, whereas scans at the 33% and 66% sites consisted mainly of cortical bone. A single axial slice of 2.5 mm thickness with a voxel size of 0.5 mm and a speed of 20 mm/second was taken. Image processing was performed by a single investigator using the Stratec software package (version 5.5E).
pQCT bone parameters
Of 36 available pQCT parameters, we analyzed three that reflect bone density and size as well as five strength parameters shown to be predictive of fracture risk. All scanning sites were measured for total volumetric bone mineral density (mg/cm3). At T33% and T66%, size parameters of cortical area (mm2) and thickness (mm) were measured, in addition to strength parameters such as cross-sectional moment of inertia (CSMI, mm4), polar moment of inertia (PMI, mm4), section modulus (SM, mm3), and polar (SSIp) and axial (SSIx) stress strain indices. Formulas for calculating CSMI, SM, and SSI have been described in a previous publication. CSMI and PMI reflect estimations of bone resistance to bending and torsion. SM derives from the CSMI and is an estimator of torsional strength. SSI estimates bending strength by accounting for material properties of bone and multiplying SM by the quotient of the measured cortical density and normal physiologic cortical density (1200 mg/cm3). SSIp accounts for torsional load, where SSIx does not, and has proven to be a precise indicator of architectural strength in bone-bending tests. In terms of breaking force, however, it is SSIx that correlates stronger than areal BMD, CSMI, or cortical vBMD.
We compared participant characteristics across quartiles of sclerostin. For normally distributed variables, a test of linear trend was performed by treating sclerostin quartile cutoffs as category integer valued steps. For non-normally distributed or skewed variables, the Jonckheere-Terpstra test for trend was performed. The Cochran-Armitage test for trend was used for dichotomous variables.
Analysis of covariance was used to test for differences across sclerostin quartiles in age-adjusted and multivariable-adjusted means of continuous variables (DXA BMD, pQCT tibia bone strength and geometry, and bone turnover markers, P1NP and CTX). The multivariable-adjusted model included age, race, weight, BMI, and diabetes status. To explore whether the results are modified by 25-hydroxyvitamin D (25(OH)D), sex hormone–binding globulin, or bioavailable estradiol and testosterone, we subsequently individually adjusted for each marker in the multivariable models. All analyses were conducted with Statistical Analysis System (SAS; version 9.3; SAS Institute, Cary, NC, USA).