Search

BackgroundTraditional analysis of High Resolution peripheral Quantitative Computed Tomography (HR-pQCT) images results in a multitude of cortical and trabecular parameters which would be potentially cumbersome to interpret for clinicians compared to user-friendly tools utilising clinical parameters. A computer vision approach (by which the entire scan is ‘read’ by a computer algorithm) to ascertain fracture risk, would be far simpler. We therefore investigated whether a computer vision and machine learning technique could improve upon selected clinical parameters in assessing fracture risk.MethodsParticipants of the Hertfordshire Cohort Study (HCS) attended research visits at which height and weight were measured; fracture history was determined via self-report and vertebral fracture assessment. Bone microarchitecture was assessed via HR-pQCT scans of the non-dominant distal tibia (Scanco XtremeCT), and bone mineral density measurement and lateral vertebral assessment were performed using dual-energy X-ray absorptiometry (DXA) (Lunar Prodigy Advanced). Images were cropped, pre-processed and texture analysis was performed using a three-dimensional local binary pattern method. These image data, together with age, sex, height, weight, BMI, dietary calcium and femoral neck BMD, were used in a random-forest classification algorithm. Receiver operating characteristic (ROC) analysis was used to compare fracture risk identification methods.ResultsOverall, 180 males and 165 females were included in this study with a mean age of approximately 76 years and 97 (28 %) participants had sustained a previous fracture. Using clinical risk factors alone resulted in an area under the curve (AUC) of 0.70 (95 % CI: 0.56–0.84), which improved to 0.71 (0.57–0.85) with the addition of DXA-measured BMD. The addition of HR-pQCT image data to the machine learning classifier with clinical risk factors and DXA-measured BMD as inputs led to an improved AUC of 0.90 (0.83–0.96) with a sensitivity of 0.83 and specificity of 0.74.ConclusionThese results suggest that using a three-dimensional computer vision method to HR-pQCT scanning may enhance the identification of those at risk of fracture beyond that afforded by clinical risk factors and DXA-measured BMD. This approach has the potential to make the information offered by HR-pQCT more accessible (and therefore) applicable to healthcare professionals in the clinic if the technology becomes more widely available.

Background: Radiofrequency Echographic Multi Spectrometry (REMS) is a non-ionizing technology for the densitometric assessment of osteoporosis. It has already been validated in Italian women with respect to the current clinical reference technology, Dual-energy X-ray Absorptiometry (DXA). Purpose: Aim of the current study was to assess the diagnostic accuracy of REMS technology with respect to DXA in a wider European clinical context. Methods: A total of 4307 female Caucasian patients aged between 30 and 90 years underwent DXA and REMS scans at femoral neck and/or lumbar spine (the site depending on the medical prescription). The acquired data underwent a rigorous quality check in order to exclude the erroneous DXA and REMS reports. The diagnostic agreement between the two technologies was assessed, also stratifying for patients' age groups. The ability to recognise previously fractured patients was also investigated. Results: Overall, 4245 lumbar spine scans and 4271 femoral neck scans were performed. The ability to discriminate patients with and without osteoporosis by femoral neck investigation resulted in sensitivity and specificity of 90.4% and 95.5%, respectively. For lumbar spine scans, a sensitivity of 90.9% and a specificity of 95.1% were obtained. The areas under the curve (AUCs) of the Receiver Operating Characteristic (ROC) curve evaluating the ability to discriminate groups of patients with previous osteoporotic fracture using DXA and REMS T-score values were 0.631 and 0.683 (p < 0.0001), respectively, for femoral neck scans, whereas 0.603 and 0.640 (p = 0.0002), respectively, for lumbar spine scans. Conclusion: The diagnostic effectiveness of REMS technology at reference anatomical sites for the assessment of osteoporosis has been confirmed in a large series of female patients, spanning from younger and pre-menopausal to elderly women up to 90 years, in a multicenter European clinical context.

Studies in childhood suggest that both body composition and early postnatal growth are associated with bone mineral density (BMD). However, little is known of the relationships between longitudinal changes in fat (FM) and lean mass (LM) and bone development in pre-pubertal children. We therefore investigated these associations in a population-based mother-offspring cohort, the Southampton Women's Survey.Total FM and LM were assessed at birth and 6-7 years of age by dual-energy x-ray absorptiometry (DXA). At 6-7 years, total cross-sectional area (CSA) and trabecular volumetric BMD (vBMD) at the 4% site (metaphysis) of the tibia was assessed using peripheral quantitative computed tomography [pQCT (Stratec XCT-2000)]. Total CSA, cortical CSA, cortical vBMD and strength-strain index (SSI) were measured at the 38% site (diaphysis). FM, LM and bone parameters were adjusted for age and sex and standardised to create within-cohort z-scores. Change in LM (ΔLM) or FM (ΔFM) was represented by change in z-score from birth to 7 years old and conditioned on the birth measurement. Linear regression was used to explore the associations between ΔLM or ΔFM and standardised pQCT outcomes, before and after mutual adjustment and for linear growth. The β-coefficient represents SD change in outcome per unit SD change in predictor.DXA at birth, in addition to both DXA and pQCT scans at 6-7 years, were available for 200 children (48.5% male). ΔLM adjusted for ΔFM was positively associated with tibial total CSA at both the 4% (β=0.57SD/SD, p0.001) and 38% sites (β=0.53SD/SD, p0.001), cortical CSA (β=0.48SD/SD, p0.001) and trabecular vBMD (β=0.30SD/SD, p0.001), but not with cortical vBMD. These relationships persisted after adjustment for linear growth. In contrast, ΔFM adjusted for ΔLM was only associated with 38% total and cortical CSA, which became non-significant after adjustment for linear growth.In this study, gain in childhood LM was positively associated with bone size and trabecular vBMD at 6-7 years of age. In contrast, no relationships between change in FM and bone were observed, suggesting that muscle growth, rather than accrual of fat mass, may be a more important determinant of childhood bone development.

The FRAX(tr) algorithm uses clinical risk factors (CRF) and bone mineral density (BMD) to predict fracture risk but does not include falls history in the calculation. Using results from the Hertfordshire Cohort Study, we examined the relative contributions of CRFs, BMD and falls history to fracture prediction. We studied 2299 participants at a baseline clinic that included completion of a health questionnaire and anthropometric data. A mean of 5.5years later (range 2.9-8.8years) subjects completed a postal questionnaire detailing fall and fracture history. In a subset of 368 men and 407 women, bone densitometry was performed using a Hologic QDR 4500 instrument. There was a significantly increased risk of fracture in men and women with a previous fracture. A one standard deviation drop in femoral neck BMD was associated with a hazards ratio (HR) of incident fracture (adjusted for CRFs) of 1.92 (1.04-3.54) and 1.77 (1.16-2.71) in men and women respectively. A history of any fall since the age of 45years resulted in an unadjusted HR of fracture of 7.31 (3.78-14.14) and 8.56 (4.85-15.13) in men and women respectively. In a ROC curve analysis, the predictive capacity progressively increased as BMD and previous falls were added into an initial model using CRFs alone. Falls history is a further independent risk factor for fracture. Falls risk should be taken into consideration when assessing whether or not to commence medication for osteoporosis and should also alert the physician to the opportunity to target falls risk directly.

Osteoporosis is a major healthcare problem which is conventionally assessed by dual energy x-ray absorptiometry (DXA). New technologies such as high resolution peripheral quantitative computed tomography (HRpQCT) also predict fracture risk. HRpQCT measures a number of bone characteristics that may inform specific patterns of bone deficits. We used cluster analysis to define different bone phenotypes and their relationships to fracture prevalence and areal bone mineral density (BMD). 177 men and 159 women, in whom fracture history was determined by self-report and vertebral fracture assessment, underwent HRpQCT of the distal radius and femoral neck DXA. Five clusters were derived with two clusters associated with elevated fracture risk. “Cluster 1” contained 26 women (50.0% fractured) and 30 men (50.0% fractured) with a lower mean cortical thickness and cortical volumetric BMD, and in men only, a mean total and trabecular area more than the sex-specific cohort mean. “Cluster 2” contained 20 women (50.0% fractured) and 14 men (35.7% fractured) with a lower mean trabecular density and trabecular number than the sex-specific cohort mean. Logistic regression showed fracture rates in these clusters to be significantly higher than the lowest fracture risk cluster (5) (p

Osteoporosis and sarcopenia are common in older age and associated with significant morbidity and mortality. Consequently, they are both attended by a considerable socioeconomic burden. Osteoporosis was defined by the World Health Organisation (WHO) in 1994 as a bone mineral density of less than 2.5 standard deviations below the sex-specific young adult mean and this characterisation has been adopted globally. Subsequently, a further step forward was taken when bone mineral density was incorporated into fracture risk prediction algorithms, such as the Fracture Risk Assessment Tool (FRAX®) also developed by the WHO. In contrast, for sarcopenia there have been several diagnostic criteria suggested, initially relating to low muscle mass alone and more recently low muscle mass and muscle function. However, none of these have been universally accepted. This has led to difficulties in accurately delineating the burden of disease, exploring geographic differences, and recruiting appropriate subjects to clinical trials. There is also uncertainty about how improvement in sarcopenia should be measured in pharmaceutical trials. Reasons for these difficulties include the number of facets of muscle health available, e.g. mass, strength, function, and performance, and the various clinical outcomes to which sarcopenia can be related such as falls, fracture, disability and premature mortality. It is imperative that a universal definition of sarcopenia is reached soon to facilitate greater progress in research into this debilitating condition. This article is part of a Special Issue entitled "Muscle Bone Interactions".

Peripheral quantitative computed tomography (pQCT) captures novel aspects of bone geometry that may contribute to fracture risk and offers the ability to measure both volumetric bone mineral density (vBMD) and a separation of trabecular and cortical compartments of bone, but longitudinal data relating measures obtained from this technique to incident fractures are lacking. Here we report an analysis from the Hertfordshire Cohort Study, where we were able to study associations between measures obtained from pQCT and DXA in 182 men and 202 women aged 60-75 years at baseline with incident fractures over 6 years later. Among women, radial cortical thickness (HR 1.72, 95% CI 1.16, 2.54, p=0.007) and cortical area (HR 1.91, 95% CI 1.27, 2.85, p=0.002) at the 66% slice were both associated with incident fractures; these results remained significant after adjustment for confounders (age, BMI, social class, cigarette smoking and alcohol consumption, physical activity, dietary calcium, HRT and years since menopause). Further adjustment for aBMD made a little difference to the results. At the tibia, cortical area (HR 1.58, 95% CI 1.10, 2.28, p=0.01), thickness (HR 1.49, 95% CI 1.08, 2.07, p=0.02) and density (HR 1.64, 95% CI 1.18, 2.26, p=0.003) at the 38% site were all associated with incident fractures with the cortical area and density relationships remaining robust to adjustment for the confounders listed above. Further adjustment for aBMD at this site did lead to attenuation of relationships. Among men, tibial stress-strain index (SSI) was predictive of incident fractures (HR 2.30, 95% CI 1.28, 4.13, p=0.005). Adjustment for confounding variables and aBMD did not render this association non-significant. In conclusion, we have demonstrated relationships between measures of bone size, density and strength obtained by pQCT and incident fracture. These relationships were attenuated but in some cases remained significant after adjustment for BMD measures obtained by DXA, suggesting that some additional information may be conferred by this assessment.

Hip fracture incidence is lower in Japan than in the West. Although differences have been found in peak bone mass and hip geometry between white and Japanese populations, these do not fully explain the difference in hip fracture rates. Variation in the rates of involutional bone loss may be an additional contributing factor. We address this issue in a prospective epidemiological study comparing bone loss rate among elderly people in Britain and Japan. Two population-based studies of bone loss rate in a British and a Japanese cohort were performed. Annual bone loss rates were obtained for 172 Hertfordshire men and 143 Hertfordshire women of mean age 66 years, and a questionnaire administered to obtain information on known confounding lifestyle factors. Eighty-six Japanese men and 90 Japanese women of mean age 69 years completed a similar study in Taiji, Japan. British men and women were heavier than Japanese men and women. Differences in lifestyle were also evident; the British men were less likely to smoke and the women more likely to consume alcohol than their Japanese counterparts. The British population also spent more time walking outdoors. Statistically significant differences between the two populations were apparent in baseline bone mineral density at lumbar spine (p < 0.05) and trochanter (p < 0.001) in men and women with Japanese subjects having lower values. There were also significant differences in bone density at the femoral neck (p < 0.001) between British and Japanese males. Men gained bone at the lumbar spine over the follow-up period in both populations. Bone loss rates were generally greater in the British female population than in Japanese women: the difference was statistically significant at the femoral neck (p < 0.05) and femoral trochanter (p < 0.001). These differences all remained significant after adjustment for differences in age between the two populations. Japanese subjects appear to have lower peak bone mass, but slower bone loss rates in later life than their European counterparts. These differences in bone loss rate help to explain the relatively low hip fracture rates found in Japan.

We are most grateful to Dr Kawada for their comments on our study. We were delighted that they felt the topic of importance and were so supportive of the methodology and results. They correctly point out that, in keeping with the FRAX model(1), we utilised height and weight as clinical risk factors in the assessment of fracture prediction(2). In our model, the hazard ratios (HR) for fracture associated with a history of falls were 6.96 and 2.64 in men and women respectively after adjustment for clinical risk factors and femoral neck bone mineral density (BMD)(3). When body mass index (BMI) was utilised in place of height and weight, the values changed little (6.68 and 2.75 respectively). We entirely agree that the limited number of fractures in men will affect the stability of the male model risk prediction. In this regard, we would point out the imprecision of our estimates of risk, as illustrated by the 95% confidence intervals for men. Of course, our research paves the way for similar studies using larger populations or longer follow up to increase fracture numbers and statistical power. The correspondent also correctly identifies the important issue of exercise as a predictor of fracture. This was not incorporated in our study models as it is not a constituent of the FRAX algorithm. However assessments have been made in our cohort of self-perceived walking speed and habitual physical activity. The latter was calculated as a standardised score ranging from 0–100 derived from frequency of gardening, housework, climbing stairs and carrying loads in a typical week. Higher scores indicated greater levels of activity(4). An association was demonstrated between faster walking speed and a lower rate of both falls and fractures in women. However, when included in our fracture prediction models, these variables did not materially affect the HR for fracture in either sex, and multivariate analysis did not reveal any association between habitual physical activity and subsequent fracture in either men or women. Overall our study confirms the role of clinical risk factors and BMD in fracture risk prediction and provides evidence that fall history may further augment this. These findings are in accord with the recommendations of the FRAX Clinical Task Force(5) that suggested falls be accounted for alongside FRAX. They are also commensurate with their suggestion that a 30% increased risk of fracture be applied for every additional fall in the preceding year.

During treatment of childhood acute lymphoblastic leukemia (ALL) fracture incidence is increased. Studies using DXA, which measures a composite of both trabecular and cortical BMD, have shown reduced BMD during treatment. We investigated changes in compartmental (cortical and trabecular) volumetric BMD (vBMD) and bone geometry using peripheral quantitative computed tomography. These outcomes were also analysed in relation to adiposity and treatment factors. Thirty nine patients with ALL (64% male, median age 7.2 years (4.1-16.9)) were compared to 34 healthy controls (50% male, median age 9.1 years (4.4-18.7)). DXA-derived age-specific standard deviation scores (SDS) of the lumbar spine (LS) and femoral neck (FN) were reduced in subjects with ALL compared to controls (p ≤ 0.01). This persisted following adjustment for body size using height-specific SDS (LS -0.72 ± 1.02 vs -0.18 ± 0.72, p=0.01; FN -1.53 ± 0.96 vs -0.74 ± 0.74, p=0.001) and bone mineral apparent density (BMAD) SDS (LS -0.76 ± 1.14 vs 0.04 ± 1.08, p=0.01; FN -1.63 ± 1.38 vs -0.16 ± 1.20, p

Infant growth is a determinant of adult bone mass, and poor childhood growth is a risk factor for adult hip fracture. Peripheral quantitative computed tomography (pQCT) allows non-invasive assessment of bone strength. We utilised this technology to examine relationships between growth in early life and bone strength. We studied 313 men and 318 women born in Hertfordshire between 1931 and 1939 who were still resident there in adult life, for whom detailed early life records were available. Lifestyle factors were evaluated by questionnaire, anthropometric measurements made, and peripheral QCT examination of the radius and tibia performed (Stratec 4500). Birthweight and conditional weight at 1 year were strongly related to radial and tibial length in both sexes (p

Evidence is accruing that environmental exposures during critical periods of early development induce persisting changes in skeletal growth, and alter fracture risk in later life. We have previously demonstrated that placental calcium transport, partly determined by maternal 25-(OH) vitamin D status, may underlie this phenomenon. However, the precise relationship between expression of calcium transport proteins in the human placenta, and neonatal bone mineral accrual in the offspring, remains unknown. Tissue samples from 70 human placentae were fast frozen in liquid nitrogen and stored at -70 degrees C. A quantitative real time reverse transcriptase polymerase chain reaction was used to measure the mRNA expression of PMCA isoforms 1-4, using beta-actin as a control gene. Neonatal whole body bone area, mineral content and areal density (BA, BMC, BMD) were measured within 2 weeks of birth using DXA. PMCA3 mRNA expression predicted BA (r=0.28, p=0.02), BMC (r=0.25, p=0.04), placental weight (r=0.26, p=0.04) and birth weight (r=0.33, p=0.006) of the neonate. In a multivariate model, the relationship between placental PMCA3 expression and neonatal BMC was independent of maternal height, pre-pregnant fat stores, parity, physical activity, smoking, and calcium intake (p

Evidence is accumulating that the risk of osteoporosis may be influenced by environmental factors during intrauterine and early postnatal life; such programming might be mediated through modification of the GH/IGF-1 axis during critical periods in its development. To address this issue, we explored the relationships among birth weight, circulating GH profile, bone density, and bone loss rate in a group of British women. The study population consisted of 38 women 60-75 years old resident in Hertfordshire for whom detailed birth records were available. Twenty-four-hour circulating GH profiles were obtained during an inpatient stay on a metabolic ward, after an overnight rest. The circulating profile of GH was characterised by estimating the peak, median, trough, and total concentrations from 72 samples measured sequentially over 24 h in each subject. Bone mineral density was assessed at the lumbar spine and femoral neck at baseline and at follow-up 4 years later. Lumbar spine bone mineral content (BMC) and density (BMD) were positively associated with all measures of GH concentration, although relationships were strongest for BMC with trough GH (r = 0.47, P < 0.01). Associations persisted after adjustment for age, body mass index, smoking, alcohol consumption, physical activity, and osteoarthritis score in multiple regression models. However, associations of GH concentration with femoral neck BMC were weak, and there was no association between any measure of GH concentration and bone loss at either site. Total (integrated) daily GH concentration tended to increase (P = 0.08) with rising birth weight, while IGF-1 concentration fell (P = 0.05) with rising birth weight, suggesting a role for the GH/IGF-1 axis in the programming of adult bone mass among women.

The 7Q31 locus, containing WNT16, is associated with bone mineral density, osteoporotic fracture and bone strength

Records from the General Practice Research Database were used to derive age- and gender-specific fracture incidence rates for England and Wales during the period 1988-1998. In total, 103,052 men and 119,317 women in the sample of 5 million adults sustained a fracture over 10.4 million and 11.2 million person-years (py) of follow-up. Among women, the most frequent fracture sites were the radius/ulna (30.2 cases per 10,000 py) and femur/hip (17.0 per 10,000 py). In men, the most common fracture was that of the carpal bones (26.2 per 10,000 py); the incidence of femur/hip fracture was 5.3 per 10,000 py. Varying patterns of fracture incidence were observed with increasing age; whereas some fractures became more common in later life (vertebral, distal forearm, hip, proximal humerus, rib, clavicle, pelvis), others were more frequent in childhood and young adulthood (tibia, fibula, carpus, foot, ankle). The lifetime risk of any fracture was 53.2% at age 50 years among women, and 20.7% at the same age among men. Whereas fractures of the proximal femur and vertebral body were associated with excess mortality over a 5 year period following fracture diagnosis among both men and women, fractures of the distal forearm were associated with only slight excess mortality in men. This study provides robust estimates of fracture incidence that will assist health-care planning and delivery.