Machine Learning Solutions for Osteoporosis—A Review

Osteoporosis and its clinical consequence, bone fracture, is a multifactorial disease that has been the object of extensive research. Recent advances in machine learning (ML) have enabled the field of artificial intelligence (AI) to make impressive breakthroughs in complex data environments where human capacity to identify high-dimensional relationships is limited. The field of osteoporosis is one such domain, notwithstanding technical and clinical concerns regarding the application of ML methods. This qualitative review is intended to outline some of these concerns and to inform stakeholders interested in applying AI for improved management of osteoporosis. A systemic search in PubMed and Web of Science resulted in 89 studies for inclusion in the review. These covered one or more of four main areas in osteoporosis management: bone properties assessment (n = 13), osteoporosis classification (n = 34), fracture detection (n = 32), and risk prediction (n = 14). Reporting and methodological quality was determined by means of a 12-point checklist. In general, the studies were of moderate quality with a wide range (mode score 6, range 2 to 11). Major limitations were identified in a significant number of studies. Incomplete reporting, especially over model selection, inadequate splitting of data, and the low proportion of studies with external validation were among the most frequent problems. However, the use of images for opportunistic osteoporosis diagnosis or fracture detection emerged as a promising approach and one of the main contributions that ML could bring to the osteoporosis field. Efforts to develop ML-based models for identifying novel fracture risk factors and improving fracture prediction are additional promising lines of research. Some studies also offered insights into the potential for model-based decision-making. Finally, to avoid some of the common pitfalls, the use of standardized checklists in developing and sharing the results of ML models should be encouraged. © 2021 American Society for Bone and Mineral Research (ASBMR).     

Dietary Inflammatory Index,Bone Mineral Density,and Risk of Fracture in Postmenopausal Women: Results From the Women's Health Initiative

Previous studies suggest that bone loss and fracture risk are associated with higher inflammatory milieu, potentially modifiable by diet. The primary objective of this analysis was to evaluate the association of the dietary inflammatory index (DII), a measure of the inflammatory potential of diet, with risk of hip, lower‐arm, and total fracture using longitudinal data from the Women's Health Initiative Observational Study and Clinical Trials. Secondarily, we evaluated changes in bone mineral density (BMD) and DII scores. DII scores were calculated from baseline food frequency questionnaires (FFQs) completed by 160,191 participants (mean age 63 years) without history of hip fracture at enrollment. Year 3 FFQs were used to calculate a DII change score. Fractures were reported at least annually; hip fractures were confirmed by medical records. Hazard ratios for fractures were computed using multivariable‐adjusted Cox proportional hazard models, further stratified by age and race/ethnicity. Pairwise comparisons of changes in hip BMD, measured by dual‐energy X‐ray absorptiometry from baseline, year 3, and year 6 were analyzed by quartile (Q1 = least inflammatory diet) of baseline DII scores in a subgroup of women (n = 10,290). Mean DII score improved significantly over 3 years (p < 0.01), but change was not associated with fracture risk. Baseline DII score was only associated with hip fracture risk in younger white women (HR Q4,1.48; 95% CI, 1.09 to 2.01; p = 0.01). There were no significant associations among white women older than 63 years or other races/ethnicities. Women with the least inflammatory DII scores had less loss of hip BMD (p = 0.01) by year 6, despite lower baseline hip BMD, versus women with the most inflammatory DII scores. In conclusion, a less inflammatory dietary pattern was associated with less BMD loss in postmenopausal women. A more inflammatory diet was associated with increased hip fracture risk only in white women younger than 63 years. © 2016 American Society for Bone and Mineral Research.     

Change in Trabecular Bone Score (TBS) With Antiresorptive Therapy Does Not Predict Fracture in Women: The Manitoba BMD Cohort

Bone mineral density (BMD) and trabecular bone score (TBS), along with additional clinical risk factors, can be used to identify individuals at high fracture risk. Whether change in TBS in untreated or treated women independently affects fracture risk is unclear. Using the Manitoba (Canada) DXA Registry containing all BMD results for the population we identified 9044 women age ≥40 years with two consecutive DXA scans and who were not receiving osteoporosis treatment at baseline (baseline mean age 62 ± 10 years). We examined BMD and TBS change, osteoporosis treatment, and incident major osteoporotic fractures (MOFs) for each individual. Over a mean of 7.7 years follow‐up, 770 women developed an incident MOF. During the interval between the two DXA scans (mean, 4.1 years), 5083 women initiated osteoporosis treatment (bisphosphonate use 80%) whereas 3961 women did not receive any osteoporosis treatment. Larger gains in both BMD and TBS were seen in women with greater adherence to osteoporosis medication (p for trend <0.001), and the magnitude of the increase was consistently greater for BMD than for TBS. Among treated women there was greater antifracture effect for each SD increase in total hip BMD change (fracture decrease 20%; 95% CI, 13% to 26%; p < 0.001), femoral neck BMD change (19%; 95% CI, 12% to 26%; p < 0.001), and lumbar spine BMD change (9%; 95% CI, 0% to 17%; p = 0.049). In contrast, change in TBS did not predict fractures in women who initiated osteoporosis treatment (p = 0.10). Among untreated women neither change in BMD or TBS predicted fractures. We conclude that, unlike antiresorptive treatment–related changes in BMD, change in lumbar spine TBS is not a useful indicator of fracture risk irrespective of osteoporosis treatment. © 2016 American Society for Bone and Mineral Research.     

Low Serum DHEAS Predicts Increased Fracture Risk in Older Men: The MrOS Sweden Study

The adrenal‐derived hormones dehydroepiandrosterone (DHEA) and its sulfate (DHEAS) are the most abundant circulating hormones and their levels decline substantially with age. DHEAS is considered an inactive precursor, which is converted into androgens and estrogens via local metabolism in peripheral target tissues. The predictive value of serum DHEAS for fracture risk is unknown. The aim of this study was, therefore, to assess the associations between baseline DHEAS levels and incident fractures in a large cohort of older men. Serum DHEAS levels were analyzed with mass spectrometry in the population‐based Osteoporotic Fractures in Men study in Sweden (n = 2568, aged 69 to 81 years). Incident X‐ray validated fractures (all, n = 594; non‐vertebral major osteoporotic, n = 255; hip, n = 175; clinical vertebral, n = 206) were ascertained during a median follow‐up of 10.6 years. DHEAS levels were inversely associated with the risk of any fracture (hazard ratio [HR] per SD decrease = 1.14, 95% confidence interval [CI] 1.05–1.24), non‐vertebral major osteoporotic fractures (HR = 1.31, 95% CI 1.16–1.48), and hip fractures (HR = 1.18, 95% CI 1.02–1.37) but not clinical vertebral fractures (HR = 1.09, 95% CI 0.95–1.26) in Cox regression models adjusted for age, body mass index (BMI) and prevalent fractures. Further adjustment for traditional risk factors for fracture, bone mineral density (BMD), and/or physical performance variables as well as serum sex steroid levels only slightly attenuated the associations between serum DHEAS and fracture risk. Similarly, the point estimates were only marginally reduced after adjustment for FRAX estimates with BMD. The inverse association between serum DHEAS and all fractures or major osteoporotic fractures was nonlinear, with a substantial increase in fracture risk (all fractures 22%, major osteoporotic fractures 33%) for those participants with serum DHEAS levels below the median (0.60 μg/mL). In conclusion, low serum DHEAS levels are a risk marker of mainly non‐vertebral fractures in older men, of whom those with DHEAS levels below 0.60 μg/mL are at highest risk. © The Authors. Journal of Bone and Mineral Research Published by Wiley Periodicals Inc.     

Assessing Depressive Symptoms in Multiple Sclerosis: Is It Necessary to Omit Items from the Original Beck Depression Inventory?

Overlap between depression scale item content and medical symptoms may exaggerate depression estimates for patients with multiple sclerosis (MS). We reconsider Mohr and co-workers' (1997) recommendation to omit Beck Depression Inventory (BDI) items assessing work ability (item 15), fatigue (17), and health concerns (20) for MS patients. Subjects were medical patients with either MS (n = 105) or a medical disorder for which the BDI is empirically supported [diabetes mellitus (DM), n = 71; chronic pain (CP), n = 80], psychiatric patients with depressive disorder (MDD; n = 37), and healthy controls (HC; n = 80). Relative scores for the eight somatic BDI items were analyzed by multivariate analysis of variance with demographic variables and BDI total as covariates. The only significant difference was MS > HC (item 15). On raw scores, MS patients exceeded HCs on items 15 and 21 (sexual disinterest), but this was attributable to the low HC item endorsement. There were no other differences on somatic items or item-total correlations. Scale consistency was good across groups, regardless of item omission. Somatic items were unassociated with major MS parameters. We thus encourage continued application of the full BDI for assessing depressive symptoms in patients with MS.     

Low Plasma Level of Leucine‐Rich Repeat‐Containing 17 (LRRc17) Is an Independent and Additive Risk Factor for Osteoporotic Fractures in Postmenopausal Women

A novel role of leucine‐rich repeat‐containing 17 (LRRc17), an LRR protein secreted by osteoblasts, as a negative regulator of receptor activator of NF‐κB ligand–induced osteoclast differentiation was found. However, the clinical association between LRRc17 and osteoporotic fracture (OF) has not yet been investigated. We hypothesized that low circulating plasma level of LRRc17 might serve as an independent and additive risk factor for OF, including vertebral fractures (VF) and non‐vertebral fractures (non‐VF). In this case‐control study, 102 OF cases and 102 age‐ and body mass index–matched controls (mean age, 63.2 years) were analyzed among 532 postmenopausal women. VF (n = 49) and non‐VF (n = 60) participants were identified using lateral thoracolumbar radiographs and an interviewer‐assisted questionnaire, respectively. Median LRRc17 levels were significantly lower in participants with any OF (117.5 versus 197.3 pg/mL, p < 0.001), VF (93.2 versus 172.4 pg/mL, p = 0.002), and non‐VF (124.5 versus 206.9 pg/mL, p = 0.008) compared with the respective controls without fractures. The prevalence of OF increased from the highest LRRc17 tertile (≥228.5 pg/mL, 33.8%) to the lowest (<95.6 pg/mL, 63.2%). Each log unit decrease of LRRc17 was associated with greater risk of OF (odds ratio [OR] = 1.46; 95% confidence interval [CI] 1.10–1.96; p = 0.009) and VF (OR = 2.42; 95% CI 1.39–4.23; p = 0.002). Plasma LRRc17 significantly improved discrimination of OF, particularly VF, when added to models including clinical risk factors and bone mineral density according to the area under receiver operating characteristics curves (AUC 0.71 to 0.81, p = 0.036), category‐free net reclassification improvement (0.79; 95% CI 0.37–1.21; p < 0.001), and integrated discrimination improvement (0.13; 95% CI 0.06–0.20; p < 0.001). Low plasma LRRc17 was an independent risk factor for OF, which improved risk stratification, particularly in the spines of postmenopausal women. © 2016 American Society for Bone and Mineral Research.     

Muscle Strength and Physical Performance Improve Fracture Risk Prediction Beyond Garvan and FRAX: The Osteoporotic Fractures in Men (MrOS) Study

Muscle strength and physical performance are associated with fracture risk in men. However, it is not known whether these measurements enhance fracture prediction beyond Garvan and FRAX tools. A total of 5665 community-dwelling men, aged ≥65 years, from the Osteoporotic Fractures in Men (MrOS) Study, who had data on muscle strength (grip strength) and physical performance (gait speed and chair stand tests), were followed from 2000 to 2019 for any fracture, major osteoporotic fracture (MOF), initial hip, and any hip fracture. The contributions to different fracture outcomes were assessed using Cox's proportional hazard models. Tool-specific analysis approaches and outcome definitions were used. The added predictive values of muscle strength and physical performance beyond Garvan and FRAX were assessed using categorical net reclassification improvement (NRI) and relative importance analyses. During a median follow-up of 13 (interquartile range 7–17) years, there were 1014 fractures, 536 MOFs, 215 initial hip, and 274 any hip fractures. Grip strength and chair stand improved prediction of any fracture (NRI for grip strength 3.9% and for chair stand 3.2%) and MOF (5.2% and 6.1%). Gait speed improved prediction of initial hip (5.7%) and any hip (7.0%) fracture. Combining grip strength and the relevant performance test further improved the models (5.7%, 8.9%, 9.4%, and 7.0% for any, MOF, initial, and any hip fractures, respectively). The improvements were predominantly driven by reclassification of those with fracture to higher risk categories. Apart from age and femoral neck bone mineral density, muscle strength and performance were ranked equal to or better than the other risk factors included in fracture models, including prior fractures, falls, smoking, alcohol, and glucocorticoid use. Muscle strength and performance measurements improved fracture risk prediction in men beyond Garvan and FRAX. They were as or more important than other established risk factors. These measures should be considered for inclusion in fracture risk assessment tools. © 2021 American Society for Bone and Mineral Research (ASBMR).     

Preventing the development and progression of diabetic kidney disease: Where do we stand?

Diabetic kidney disease (DKD) is a major factor associated with increased cardiovascular (CV) and all-cause mortality and morbidity in patients with diabetes. Current standard therapy includes intensive management of hyperglycemia and blood pressure control with renin-angiotensin-aldosterone system (RAAS) blockers. Despite the implementation of this strategy, DKD remains the leading cause of end-stage renal disease (ESRD), mainly because of the increasing burden of diabetes mellitus. The aim of this review is to evaluate the available evidence, focusing on the benefit of current treatment in the development and progression of DKD.     

Correction of Malnutrition Following Gastrectomy with Cyclic Enteral Nutrition

This study was designed to evaluate the efficacyand tolerance of cyclic enteral nutrition (CyEN) ingastrectomized patients and to compare the nutritionalresponse to refeeding of 28 gastrectomized patients and 38 nongastrectomized undernourishedpatients. Total (enteral + oral) energy intake (292% and284% of resting energy expenditure in gastrectomized andnongastrectomized patients, respectively) and the duration of CyEN (27 days) were similar in bothgroups. Tolerance was good and not different in the twogroups. In the gastrectomized patients, a globalnutritional deficiency score based on 10 biological and anthropometric parameters significantlyimproved (P < 0.0001) with six of the 10 nutritionalparameters studied being improved by renutrition.Comparison of the two groups revealed similar efficacy of refeeding; the global nutritional deficiencyscore improved by 42.7 ± 17.3% in group 1 and40.0 ± 17.4% in group 2. After one year, theprobability of being alive without relapse was 77.8% inthe gastrectomized and 72.2% in thenongastrectomized patients (P = 0.70). The efficacy ofrenutrition was similar regardless of the type ofgastrectomy (eg, partial or total). CyEN is a safe,effective and durable treatment for undernutrition ingastrectomized patients.