Prediction of lithium response using genomic data

Abstract

Predicting lithium response prior to treatment could both expedite therapy and avoid exposure to side effects. Since lithium responsiveness may be heritable, its predictability based on genomic data is of interest. We thus evaluate the degree to which lithium response can be predicted with a machine learning (ML) approach using genomic data. Using the largest existing genomic dataset in the lithium response literature (n = 2210 across 14 international sites; 29% responders), we evaluated the degree to which lithium response could be predicted based on 47,465 genotyped single nucleotide polymorphisms using a supervised ML approach. Under appropriate cross-validation procedures, lithium response could be predicted to above-chance levels in two constituent sites (Halifax, Cohen's kappa 0.15, 95% confidence interval, CI [0.07, 0.24]; and Würzburg, kappa 0.2 [0.1, 0.3]). Variants with shared importance in these models showed over-representation of postsynaptic membrane related genes. Lithium response was not predictable in the pooled dataset (kappa 0.02 [- 0.01, 0.04]), although non-trivial performance was achieved within a restricted dataset including only those patients followed prospectively (kappa 0.09 [0.04, 0.14]). Genomic classification of lithium response remains a promising but difficult task. Classification performance could potentially be improved by further harmonization of data collection procedures.

Figure 1

Performance of the logistic regression (LR), and XGBoost (XGB) classifiers on the aggregate and site-level analyses. Faceting of plots along columns corresponds to different classification statistics. Faceting along rows corresponds to the aforementioned classification algorithms. Within each plot, the x-axis represents the statistic value, and the y-axis corresponds to the dataset under which the classification analysis was performed. Plot markers denote the mean performance over the respective number of cross-validation folds, with error bars denoting the empirical 95% confidence intervals. In the column for Kappa, where the p value from the criticism analysis fell below 0.01 in comparison to the null classifier we coloured the point yellow to represent exceedance of chance; all other points are coloured dark blue. Abbreviations: all sites (ALL; i.e. aggregate analysis), area under the receiver operating characteristic curve (AUC), positive predictive value (PPV), negative predictive value (NPV), F-1 score (F1).