Machine learning approaches for the prediction of bone mineral density by using genomic and phenotypic data of 5130 older men.
Ontology highlight
ABSTRACT:
SUBMITTER: Wu Q
PROVIDER: S-EPMC7904941 | biostudies-literature | 2021 Feb
REPOSITORIES: biostudies-literature
Publications
Machine learning approaches for the prediction of bone mineral density by using genomic and phenotypic data of 5130 older men.
Wu Qing Q Nasoz Fatma F Jung Jongyun J Bhattarai Bibek B Han Mira V MV Greenes Robert A RA Saag Kenneth G KG
Scientific reports 20210224 1
The study aimed to utilize machine learning (ML) approaches and genomic data to develop a prediction model for bone mineral density (BMD) and identify the best modeling approach for BMD prediction. The genomic and phenotypic data of Osteoporotic Fractures in Men Study (n = 5130) was analyzed. Genetic risk score (GRS) was calculated from 1103 associated SNPs for each participant after a comprehensive genotype imputation. Data were normalized and divided into a training set (80%) and a validation ...[more]
Similar Datasets
Project description:Studying dietary patterns is often more informative than individual nutrients or foods. We found that a Prudent dietary pattern (rich in vegetables and fish) was associated with reduced loss of total hip BMD in older men. A Prudent dietary pattern may be a potential lifestyle strategy for minimizing bone loss. INTRODUCTION:This study aimed to identify baseline dietary patterns using factor analysis in a cohort of older men and to evaluate whether the dietary patterns were associated with bone mineral density change (%ΔBMD) at the total hip and femoral neck over time. METHODS:Participants (n = 4379; mean age 72.9 ± 5.5 years) were from the Osteoporotic Fractures in Men (MrOS) prospective cohort study and had dietary data collected at baseline (March 2000-April 2002) and BMD measured at baseline and Visit 2 (March 2005-May 2006). Dietary intake was assessed with a brief Block food frequency questionnaire (FFQ); factor analysis was used to derive dietary patterns. BMD was measured by dual-energy x-ray absorptiometry (DXA); %ΔBMD was calculated from baseline to Visit 2. We used generalized linear regression to estimate least square (LS) means of %ΔBMD in quartiles of the dietary pattern scores adjusted for potential confounding factors. RESULTS:Two major dietary patterns were derived: Prudent (abundant in vegetables, salad, and non-fried fish) and Western (rich in hamburger, fries, processed meats, cheese, and sweets/desserts). There was an inverse association between adherence to the Prudent pattern and total hip %ΔBMD (p-trend = 0.028 after adjusting for age and clinical site; p-trend = 0.033 after further adjustment for smoking, calcium supplement use, diabetes, hypertension, and total energy intake). No other consistent associations between dietary patterns and %ΔBMD were observed. CONCLUSIONS:Greater adherence to a Prudent dietary pattern may attenuate total hip BMD loss (%ΔBMD) in older men.
| S-EPMC6842326 | biostudies-literature
Project description:BackgroundVitamin K modulates cytokines involved in bone turnover, including interleukin-6 (IL-6) and osteoprotegerin in vitro.ObjectiveThe objective of this study was to assess 1) associations between measures of vitamin K status [plasma phylloquinone and serum percentage of undercarboxylated osteocalcin (%ucOC)] and IL-6, osteoprotegerin, and C-reactive protein (CRP) concentrations and 2) the effect of daily 500 mug phylloquinone supplementation for 3 y on cytokine concentrations.DesignConcentrations of IL-6, osteoprotegerin, and CRP and bone mineral density (BMD) were measured at baseline and after 3 y of follow-up in 379 healthy men and women (60-81 y; 58.5% women) participating in a randomized trial that studied the effect of vitamin K supplementation on bone loss.ResultsCross-sectionally, plasma phylloquinone was inversely associated with IL-6 and CRP, whereas serum %ucOC was inversely associated with IL-6. Osteoprotegerin was associated positively with plasma phylloquinone and inversely with %ucOC. No differences were observed in the 3-y change in IL-6, osteoprotegerin, and CRP concentrations between participants who received phylloquinone supplementation and those who did not. Overall, no association was observed between the 3-y changes in circulating cytokines and BMD.ConclusionsPoor vitamin K status was associated with high concentrations of cytokines involved in bone turnover, but vitamin K supplementation did not confer a decrease in cytokine concentrations. The healthy status of this cohort may explain a lack of effect of vitamin K supplementation on cytokine concentrations. This trial was registered with www.clinicaltrials.gov as NCT00183001.
| S-EPMC2532855 | biostudies-literature
Project description:The study aims were to develop fracture prediction models by using machine learning approaches and genomic data, as well as to identify the best modeling approach for fracture prediction. The genomic data of Osteoporotic Fractures in Men, cohort Study (n = 5130), were analyzed. After a comprehensive genotype imputation, genetic risk score (GRS) was calculated from 1103 associated Single Nucleotide Polymorphisms for each participant. Data were normalized and split into a training set (80%) and a validation set (20%) for analysis. Random forest, gradient boosting, neural network, and logistic regression were used to develop prediction models for major osteoporotic fractures separately, with GRS, bone density, and other risk factors as predictors. In model training, the synthetic minority oversampling technique was used to account for low fracture rate, and tenfold cross-validation was employed for hyperparameters optimization. In the testing, the area under curve (AUC) and accuracy were used to assess the model performance. The McNemar test was employed to examine the accuracy difference between models. The results showed that the prediction performance of gradient boosting was the best, with AUC of 0.71 and an accuracy of 0.88, and the GRS ranked as the 7th most important variable in the model. The performance of random forest and neural network were also significantly better than that of logistic regression. This study suggested that improving fracture prediction in older men can be achieved by incorporating genetic profiling and by utilizing the gradient boosting approach. This result should not be extrapolated to women or young individuals.
| S-EPMC7492432 | biostudies-literature
Project description:Osteoporosis is an important health issue for older adults, and has been relatively understudied in older men. This study aimed to examine ethnic differences in bone mineral density (BMD), and elucidate the role of bone turnover markers (BTMs), fat and fat biomarkers on these ethnic differences. BMD at the lumbar spine and femoral neck, marrow fat at femoral neck, visceral adipose tissue (VAT) and subcutaneous adipose tissue, bone and fat biomarkers were evaluated in 120 healthy men aged ? 60 years. Indians had higher BMD values compared to Chinese at the lumbar spine (? = 20.336, SE = 4.749, p < 0.001) and the femoral neck (e ? = 1.105, SE = 0.032, p < 0.001), after adjusting for BTMs, fat composition and lifestyle choices. Marrow fat, VAT and adiponectin were independent predictors of BMD. However, these factors did not explain the lower BMD observed in older Chinese men. Our findings suggest that older Chinese men are at significant risk of osteoporotic fractures due to lower BMD. Fat appears to be a key factor associated with lower BMD, and warrants further longitudinal studies to elucidate the complex interactions between adipose tissue and bone strength.
| S-EPMC5760609 | biostudies-literature
Project description:In contrast to conventional dual-energy X-ray absorptiometry, quantitative computed tomography separately measures trabecular and cortical volumetric bone mineral density (vBMD). Little is known about the genetic variants associated with trabecular and cortical vBMD in humans, although both may be important for determining bone strength and osteoporotic risk. In the current analysis, we tested the hypothesis that there are genetic variants associated with trabecular and cortical vBMD at the femoral neck by genotyping 4608 tagging and potentially functional single-nucleotide polymorphisms (SNPs) in 383 bone metabolism candidate genes in 822 Caucasian men aged 65 years or older from the Osteoporotic Fractures in Men Study (MrOS). Promising SNP associations then were tested for replication in an additional 1155 men from the same study. We identified SNPs in five genes (IFNAR2, NFATC1, SMAD1, HOXA, and KLF10) that were robustly associated with cortical vBMD and SNPs in nine genes (APC, ATF2, BMP3, BMP7, FGF18, FLT1, TGFB3, THRB, and RUNX1) that were robustly associated with trabecular vBMD. There was no overlap between genes associated with cortical vBMD and trabecular vBMD. These findings identify novel genetic variants for cortical and trabecular vBMD and raise the possibility that some genetic loci may be unique for each bone compartment.
| S-EPMC3153388 | biostudies-literature
Project description:Aiming to identify more genomic loci associated with bone mineral density (BMD), we conducted a joint association analysis of 2 genome-wide association study (GWAS) by the integrative association method multi-trait analysis of GWAS (MTAG). The first one is the single GWAS of estimated heel BMD (eBMD) in the UK biobank (UKB) cohort (N = 426,824), and the second one is the GWAS meta-analysis of total body BMD (TB-BMD) in 66,628 participants from 30 studies. Approximate conditional association analysis was performed in the identified novel loci to identify secondary association signal. Statistical fine-mapping was conducted to prioritize plausible credible risk variants (CRVs). Candidate genes were prioritized based on the analyses of cis- expression quantitative trait locus (cis-eQTL) and cis-protein QTL (cis-pQTL) information as well as the functional category of the SNP. By integrating the information carried in over 490,000 participants, this largest joint analysis of BMD GWAS identified 12 novel genomic loci at the genome-wide significance level (GWS, p = 5.0 × 10-8), nine of which were for eBMD and four were for TB-BMD, explaining an additional 0.11 and 0.23% heritability for the two traits, respectively. These loci include 1p33 (lead SNP rs10493130, peBMD = 3.19 × 10-8), 5q13.2 (rs4703589, peBMD = 4.78 × 10-8), 5q31.3 (rs9324887, pTB-BMD = 1.36 × 10-9), 6p21.32 (rs6905837, peBMD = 3.32 × 10-8), 6q14.1 (rs10806234, peBMD = 2.63 × 10-8), 7q21.11 (rs10806234, pTB-BMD = 3.37 × 10-8), 8q24.12 (rs11995866, peBMD = 6.72 × 10-9), 12p13.31 (rs1639122, peBMD = 4.43 × 10-8), 12p12.1 (rs58489179, peBMD = 4.74 × 10-8), 12q24.23 (rs75499226, peBMD = 1.44 × 10-8), 19q13.31 (rs7255083, pTB-BMD = 2.18 × 10-8) and 22q11.23 (rs13056137, pTB-BMD = 2.54 × 10-8). All lead SNPs in these 12 loci are nominally significant in both original studies as well as consistent in effect direction between them, providing solid evidence of replication. Approximate conditional analysis identified one secondary signal in 5q13.2 (rs11738874, pconditional = 5.06 × 10-9). Statistical fine-mapping analysis prioritized 269 CRVs. A total of 65 candidate genes were prioritized, including those with known biological function to bone development (such as FGF1, COL11A2 and DEPTOR). Our findings provide novel insights into a better understanding of the genetic mechanism underlying bone development as well as candidate genes for future functional investigation.
| S-EPMC7188784 | biostudies-literature
Project description:BackgroundSevere gonadal steroid deficiency induces bone loss in adult men; however, the specific roles of androgen and estrogen deficiency in hypogonadal bone loss are unclear. Additionally, the threshold levels of testosterone and estradiol that initiate bone loss are uncertain.MethodsOne hundred ninety-eight healthy men, ages 20-50, received goserelin acetate, which suppresses endogenous gonadal steroid production, and were randomized to treatment with 0, 1.25, 2.5, 5, or 10 grams of testosterone gel daily for 16 weeks. An additional cohort of 202 men was randomized to receive these treatments plus anastrozole, which suppresses conversion of androgens to estrogens. Thirty-seven men served as controls and received placebos for goserelin and testosterone. Changes in bone turnover markers, bone mineral density (BMD) by dual-energy x-ray absorptiometry (DXA), and BMD by quantitative computed tomography (QCT) were assessed in all men. Bone microarchitecture was assessed in 100 men.ResultsAs testosterone dosage decreased, the percent change in C-telopeptide increased. These increases were considerably greater when aromatization of testosterone to estradiol was also suppressed, suggesting effects of both testosterone and estradiol deficiency. Decreases in DXA BMD were observed when aromatization was suppressed but were modest in most groups. QCT spine BMD fell substantially in all testosterone-dose groups in which aromatization was also suppressed, and this decline was independent of testosterone dose. Estradiol deficiency disrupted cortical microarchitecture at peripheral sites. Estradiol levels above 10 pg/ml and testosterone levels above 200 ng/dl were generally sufficient to prevent increases in bone resorption and decreases in BMD in men.ConclusionsEstrogens primarily regulate bone homeostasis in adult men, and testosterone and estradiol levels must decline substantially to impact the skeleton.Trial registrationClinicalTrials.gov, NCT00114114.FundingAbbVie Inc., AstraZeneca Pharmaceuticals LP, NIH.
| S-EPMC4767351 | biostudies-literature
Project description:BackgroundDistal radius fractures can adversely affect wrist function; for men with this fracture, the role played by fracture severity, age and osteoporosis on fracture outcome has not been sufficiently studied.ObjectiveTo describe patient-reported outcome and the association with bone integrity, fracture severity and future fracture risk among young and older men with distal radius fracture.MethodsThis prospective study includes 133 men with acute distal radius fracture, mean age 54 (range 21-88), who were followed for 12 months. They were categorized as younger (< 65) and older (65+). Main outcome was DASH (Disability of the Arm, Shoulder and Hand) at 12 months; DASH > 15 was defined as poor outcome. Fractures were classified and radiographic displacement identified at initial presentation and follow-up. BMD was measured and FRAX 10-year probability of fracture calculated.ResultsDisability was higher in older men (DASHmedian 10 vs 2; p = 0.002); a clinically meaningful difference (ΔDASH = 10, p = 0.017) remained after adjustment for displacement, fracture classification and treatment method. Almost 50% of older men vs 14% in younger had poor outcome, p < 0.001. Bone mineral density did not independently predict outcome. Older men with a displaced fracture at initial presentation had greater disability (DASHmedian, IQR 45, 14;73) and risk of fracture (FRAXmajor osteoporotic 14, 8;21).ConclusionMen over the age of 65 with a distal radius fracture are more likely to have post-fracture disability regardless of radiographic appearance. Fracture displacement, indicating impaired bone strength, is also more common and associated with an increased risk of fracture within 10-years. Secondary fracture prevention should therefore be considered in men presenting with distal radius fracture.
| S-EPMC7722451 | biostudies-literature
Project description:Osteoporosis is a medical condition of global concern, with increasing incidence in both sexes. Bone mineral density (BMD), a highly heritable trait, has been proven a useful diagnostic factor in predicting fracture. Because medical information is lacking about male osteoporotic genetics, we conducted a genome-wide association study of BMD in Korean men. With 1,176 participants, we analyzed 4,414,664 single nucleotide polymorphisms (SNPs) after genomic imputation, and identified five SNPs and three loci correlated with bone density and strength. Multivariate linear regression models were applied to adjust for age and body mass index interference. Rs17124500 (p = 6.42 × 10(-7)), rs34594869 (p = 6.53 × 10(-7)) and rs17124504 (p = 6.53 × 10(-7)) in 14q31.3 and rs140155614 (p = 8.64 × 10(-7)) in 15q25.1 were significantly associated with lumbar spine BMD (LS-BMD), while rs111822233 (p = 6.35 × 10(-7)) was linked with the femur total BMD (FT-BMD). Additionally, we analyzed the relationship between BMD and five genes previously identified in Korean men. Rs61382873 (p = 0.0009) in LRP5, rs9567003 (p = 0.0033) in TNFSF11 and rs9935828 (p = 0.0248) in FOXL1 were observed for LS-BMD. Furthermore, rs33997547 (p = 0.0057) in ZBTB and rs1664496 (p = 0.0012) in MEF2C were found to influence FT-BMD and rs61769193 (p = 0.0114) in ZBTB to influence femur neck BMD. We identified five SNPs and three genomic regions, associated with BMD. The significance of our results lies in the discovery of new loci, while also affirming a previously significant locus, as potential osteoporotic factors in the Korean male population.
| S-EPMC4951402 | biostudies-literature
Project description:Background:Disrupted rest-activity circadian rhythm (RAR) patterns have been associated with poor health outcomes (i.e. diminished cognitive function, increased risk of dementia and falls). Circadian time cues in bone influence the differentiation of osteoblasts and osteoclasts, and bone turnover markers exhibit circadian variation; relationships between bone outcomes and RAR are emerging areas of research. We evaluated associations between RAR and areal bone mineral density (aBMD) at the total hip and femoral neck in older men from the Osteoporotic Fractures in Men (MrOS) cohort. We hypothesized that weaker RAR patterns would be associated with lower aBMD. Methods:MrOS is an ongoing prospective cohort study following ambulatory men ? 65 years (n = 5994) at 6 U.S. clinics (baseline enrollment 3/2000-4/2002); participants for this analysis are from an ancillary study, Outcomes of Sleep Disorders in Older Men (MrOS Sleep). We included data from men who had technically adequate measures of RAR and aBMD at Sleep Visit 1 (12/2003-3/2005), with repeat aBMD at core Visit 3 (3/2007-3/2009) (n = 2412; mean age at Sleep Visit 1: 75.7 ± 5.2 years). aBMD was measured by dual energy x-ray absorptiometry (DXA). Actigraphs worn on the non-dominant wrist were used to collect circadian activity data over 4.8 ± 0.8 consecutive 24-hour periods. An extension of the traditional cosine curve was used to fit RAR to the activity data [Ancoli-Israel et al., 2003; Marler et al., 2006]. Six RAR parameters were evaluated: acrophase (time of peak activity), amplitude (rhythm strength), mesor (mean of activity fitted curve), pseudo F-statistic (overall circadian rhythmicity of rest and activity), alpha statistic (daytime to nighttime activity ratio), and beta statistic (daytime activity). Associations between RAR and aBMD (Sleep Visit 1), and RAR and ?aBMD (Sleep Visit 1-Visit 3) were assessed with generalized linear models. Covariates included age, clinic site, physical activity, race, comorbidity, body mass index (BMI), smoking, alcohol, caffeine, beta blocker use, serum 25(OH) vitamin D and urinary melatonin and calcium. Results:Pseudo F-statistic was significantly associated with total hip aBMD (p-trend = 0.009), femoral neck aBMD (p-trend = 0.007) and total hip ?aBMD (p-trend = 0.017) in minimally adjusted models but not after multivariate (MV) adjustment. Alpha statistic was significantly associated with femoral neck aBMD (p-trend = 0.029) and femoral neck ?aBMD (p-trend = 0.019) in minimally adjusted models; significance was retained in the femoral neck ?aBMD model (p-trend = 0.034) after MV adjustment. There were no consistent, significant associations between the other RAR variables and aBMD or ?aBMD. Conclusions:The data demonstrate modest associations between overall circadian rhythmicity of rest and activity (measured by pseudo F-statistic), as well as daytime to nighttime activity ratio (measured by alpha statistic), aBMD and ?aBMD, but adjustment for covariates related to lifestyle, BMI and comorbidities attenuated most of these associations. These results suggest that RAR patterns are not independently associated with aBMD or four-year ?aBMD at the total hip or femoral neck in older men, but additional research is needed.
| S-EPMC5695538 | biostudies-literature