Vitamin D, chronic pain, and depression: linear and non-linear Mendelian randomization analyses.

Vitamin D deficiency has been linked to various chronic pain conditions. However, randomized trials of vitamin D supplementation have had mixed results. In contrast, systematic reviews of randomized trials indicate a protective effect of vitamin D supplementation on depression. We undertake a Mendelian randomization investigation in UK Biobank, a study of UK residents aged 40-65 at recruitment. We perform linear and non-linear Mendelian randomization analyses for four outcomes: fibromyalgia, clinical fatigue, chronic widespread pain, and probable lifetime major depression. We use genetic variants from four gene regions with known links to vitamin D biology as instruments. In linear analyses, genetically-predicted levels of 25-hydroxyvitamin D [25(OH)D], a clinical marker of vitamin D status, were not associated with fibromyalgia (odds ratio [OR] per 10 nmol/L higher 25(OH)D 1.02, 95% confidence interval [CI] 0.93, 1.12), clinical fatigue (OR 0.99, 95% CI 0.94, 1.05), chronic widespread pain (OR 0.95, 95% CI 0.89, 1.02), or probable lifetime major depression (OR 0.97, 95% CI 0.93, 1.01). In non-linear analyses, an association was observed between genetically-predicted 25(OH)D levels and depression in the quintile of the population with the lowest 25(OH)D levels (OR 0.75, 95% CI 0.59, 0.94); associations were null in other strata. Our findings suggest that population-wide vitamin D supplementation will not substantially reduce pain or depression; however, targeted supplementation of deficient individuals may reduce risk of depression.

Associations between vitamin D and autoimmune diseases: Mendelian randomization analysis.

OBJECTIVE: The VITAL trial of vitamin D supplementation suggested a possible protective effect for autoimmune diseases but uncertainties remain. We investigated potential causal effects of vitamin D on composite and individual autoimmune diseases using Mendelian randomization. METHODS: We used data from 332,984 participants of the UK Biobank of whom 23,089 had at least one autoimmune disease defined using ICD code and/or self-report. Diseases were further considered in mechanistic subgroups driven by "autoimmunity" (n = 12,774) or "autoinflammation" (n = 11,164), then individually. We selected variants within gene regions implicated in vitamin D biology to generate a weighted genetic score. We performed population-wide analysis using the ratio method, then examined non-linear effects across five quantiles based on 25-hydroxycholecalciferol levels. RESULTS: Genetically-predicted vitamin D was associated with lower risk of diseases in the autoinflammation group (OR 0.95 per 10 ng/ml increase in 25-hydroxycholecalciferol; 95%CI 0.91-0.99; p = 0.03) but not the autoimmunity group (OR 0.99; 95%CI 0.95-1.03; p = 0.64) or combined. When considering individual diseases, genetically-predicted vitamin D was associated with lower risk of psoriasis (OR 0.91; 95%CI 0.85-0.97; p = 0.005), the most common disease in the autoinflammation group, and suggestively with systemic lupus erythematosus (OR 0.84; 95%CI 0.69-1.02; p = 0.08); results were replicated using data from independent studies. We found no evidence for a plausible non-linear relationship between vitamin D and any outcome. CONCLUSIONS: We found genetic evidence to support a causal link between 25-hydroxycholecalciferol concentrations and psoriasis and systemic lupus erythematosus. These results have implications for potential disease prevention strategies, and the interpretation and design of vitamin D supplementation trials.

Predictive Utility of a Machine Learning Algorithm in Estimating Mortality Risk in Cardiac Surgery.

BACKGROUND: This study evaluated the predictive utility of a machine learning algorithm in estimating operative mortality risk in cardiac surgery. METHODS: Index adult cardiac operations performed between 2011 and 2017 at a single institution were included. The primary outcome was operative mortality. Extreme gradient boosting (XGBoost) models were developed and evaluated using 10-fold cross-validation with 1000-replication bootstrapping. Model performance was assessed using multiple measures including precision, recall, calibration plots, area under the receiver-operating characteristic curve (C-index), accuracy, and F1 score. RESULTS: A total of 11,190 patients were included (7048 isolated coronary artery bypass grafting [CABG], 2507 isolated valves, and 1635 CABG plus valves). The Society of Thoracic Surgeons Predicted Risk of Mortality (STS PROM) was 3.2% ± 5.0%. Actual operative mortality was 2.8%. There was moderate correlation (r = 0.652) in predicted risk between XGBoost and STS PROM for the overall cohort and weak correlation (r = 0.473) in predicted risk between the models specifically in patients with operative mortality. XGBoost demonstrated improvements in all measures of model performance when compared with the STS PROM in the validation cohorts: mean average precision (0.221 XGBoost vs 0.180 STS PROM), C-index (0.808 XGBoost vs 0.795 STS PROM), calibration (mean observed-to-expected mortality: XGBoost 0.993 vs 0.956 STS PROM), accuracy (1%-3% improvement across discriminatory thresholds of 3%-10% risk), and F1 score (0.281 XGBoost vs 0.230 STS PROM). CONCLUSIONS: Machine learning algorithms such as XGBoost have promise in predictive analytics in cardiac surgery. The modest improvements in model performance demonstrated in the current study warrant further validation in larger cohorts of patients.