Assessing the association of epigenetic age acceleration with osteoarthritis in the Multicenter Osteoarthritis Study (MOST).

PURPOSE: Advancing age is one of the strongest risk factors for osteoarthritis (OA). DNA methylation-based measures of epigenetic age acceleration may provide insights into mechanisms underlying OA. METHODS: We analyzed data from the Multicenter Osteoarthritis Study in a subset of 671 participants ages 45-69 years with no or mild radiographic knee OA. DNA methylation was assessed with the Illumina Infinium MethylationEPIC 850K array. We calculated predicted epigenetic age according to Hannum, Horvath, PhenoAge, and GrimAge epigenetic clocks, then regressed epigenetic age on chronological age to obtain the residuals. Associations between the residuals and knee, hand, and multi-joint OA were assessed using logistic regression, adjusted for chronological age, sex, clinical site, smoking status, and race. RESULTS: Twenty-three percent met criteria for radiographic hand OA, 25% met criteria for radiographic knee OA, and 8% met criteria for multi-joint OA. Mean chronological age (SD) was 58.4 (6.7) years. Mean predicted epigenetic age (SD) according to Horvath, Hannum, PhenoAge, and GrimAge epigenetic clocks was 64.9 (6.4), 68.6 (5.9), 50.5 (7.7), and 67.0 (6.2), respectively. Horvath epigenetic age acceleration was not associated with an increased odds of hand OA, odds ratio (95% confidence intervals) = 1.03 (0.99-1.08), with similar findings for knee and multi-joint OA. We found similar magnitudes of associations for Hannum epigenetic age, PhenoAge, and GrimAge acceleration compared to Horvath epigenetic age acceleration. CONCLUSIONS: Epigenetic age acceleration as measured by various well-validated epigenetic clocks based on DNA methylation was not associated with increased risk of knee, hand, or multi-joint OA independent of chronological age.

A two-cohort study on the association between the gut microbiota and bone density, microarchitecture, and strength.

The gut microbiome affects the inflammatory environment through effects on T-cells, which influence the production of immune mediators and inflammatory cytokines that stimulate osteoclastogenesis and bone loss in mice. However, there are few large human studies of the gut microbiome and skeletal health. We investigated the association between the human gut microbiome and high resolution peripheral quantitative computed tomography (HR-pQCT) scans of the radius and tibia in two large cohorts; Framingham Heart Study (FHS [n=1227, age range: 32 - 89]), and the Osteoporosis in Men Study (MrOS [n=836, age range: 78 - 98]). Stool samples from study participants underwent amplification and sequencing of the V4 hypervariable region of the 16S rRNA gene. The resulting 16S rRNA sequencing data were processed separately for each cohort, with the DADA2 pipeline incorporated in the16S bioBakery workflow. Resulting amplicon sequence variants were assigned taxonomies using the SILVA reference database. Controlling for multiple covariates, we tested for associations between microbial taxa abundances and HR-pQCT measures using general linear models as implemented in microbiome multivariable association with linear model (MaAslin2). Abundance of 37 microbial genera in FHS, and 4 genera in MrOS, were associated with various skeletal measures (false discovery rate [FDR] ≤ 0.1) including the association of DTU089 with bone measures, which was independently replicated in the two cohorts. A meta-analysis of the taxa-bone associations further revealed (FDR ≤ 0.25) that greater abundances of the genera; Akkermansia and DTU089, were associated with lower radius total vBMD, and tibia cortical vBMD respectively. Conversely, higher abundances of the genera; Lachnospiraceae NK4A136 group, and Faecalibacterium were associated with greater tibia cortical vBMD. We also investigated functional capabilities of microbial taxa by testing for associations between predicted (based on 16S rRNA amplicon sequence data) metabolic pathways abundance and bone phenotypes in each cohort. While there were no concordant functional associations observed in both cohorts, a meta-analysis revealed 8 pathways including the super-pathway of histidine, purine, and pyrimidine biosynthesis, associated with bone measures of the tibia cortical compartment. In conclusion, our findings suggest that there is a link between the gut microbiome and skeletal metabolism.

Transcriptome prediction performance across machine learning models and diverse ancestries.

Transcriptome prediction methods such as PrediXcan and FUSION have become popular in complex trait mapping. Most transcriptome prediction models have been trained in European populations using methods that make parametric linear assumptions like the elastic net (EN). To potentially further optimize imputation performance of gene expression across global populations, we built transcriptome prediction models using both linear and non-linear machine learning (ML) algorithms and evaluated their performance in comparison to EN. We trained models using genotype and blood monocyte transcriptome data from the Multi-Ethnic Study of Atherosclerosis (MESA) comprising individuals of African, Hispanic, and European ancestries and tested them using genotype and whole-blood transcriptome data from the Modeling the Epidemiology Transition Study (METS) comprising individuals of African ancestries. We show that the prediction performance is highest when the training and the testing population share similar ancestries regardless of the prediction algorithm used. While EN generally outperformed random forest (RF), support vector regression (SVR), and K nearest neighbor (KNN), we found that RF outperformed EN for some genes, particularly between disparate ancestries, suggesting potential robustness and reduced variability of RF imputation performance across global populations. When applied to a high-density lipoprotein (HDL) phenotype, we show including RF prediction models in PrediXcan revealed potential gene associations missed by EN models. Therefore, by integrating other ML modeling into PrediXcan and diversifying our training populations to include more global ancestries, we may uncover new genes associated with complex traits.