Data-Driven Prediction of Drug Effects and Interactions

By Nicholas P. Tatonetti, Patrick Ye, Roxana Daneshjou, Russ B. Altman. 2012

898 citations recorded by the source corpus.

Two new databases—one of drug effects and a second of drug-drug interaction side effects—permit identification of drug targets, prediction of drug indications, and discovery of drug class interactions. Avoiding Adversity For some disease-therapy pairs, Francis Bacon was right: “The remedy is worse than the disease.” And when several drugs collide in an individual, the troubling effects can multiply. One goal of the developing field of pharmacogenomics is to make use of clinical data to predict adverse drug events so that future patients can be protected from the sometimes serious consequences. Now, Tatonetti et al. describe two new databases—one of drug effects and a second of drug-drug interaction side effects—that permit the identification of drug targets, prediction of drug indications, and discovery of drug-class interactions. Adverse drug events aren’t merely a nuisance; these toxic interactions can cause debilitating illness and death. The nature of clinical trials doesn’t allow the detection of all serious side effects and drug interactions before approval of the therapy, by regulatory agencies, for use in patients. But these agencies along with pharmaceutical companies, hospitals, and other institutions collect adverse event reports after the drugs are in use in the clinic. When delineated in databases and coupled with the impressive computing power now available, these reports have the potential to permit characterization of drug effects at the population level. However, even with the recent move toward electronic health records, adverse event data often lack crucial information about co-prescribed medications, patient demographics and medical histories, and the reasons that a given drug was prescribed in the first place. One can easily see how that lack of such information thwarts the ability to obtain meaningful analyses of drug side effects and interactions. To address this problem of omission and improve the ability to analyze drug effects, Tatonetti et al. use an adaptive data-driven approach to correct for the lack of such information—the so-called unknown “covariates.” Using this information, the authors developed two comprehensive databases—one of drug effects (Offsides) and another of drug-drug interaction side effects (Twosides)—and then used their new databases to pinpoint drug targets and discover drug-class interactions. Finally, the authors validated 47 of the drug-class interactions in an independent analysis of patient electronic health records. When prescribed together, widely used antidepressant drugs (selective serotonin reuptake inhibitors) and thiazide diuretics were associated with an increase in the incidence of prolonged QT, which indicates a delayed repolarization of the heart after a heartbeat. Prolonged QT can increase a patient’s risk of palpitations, fainting, and even sudden death resulting from ventricular fibrillation. Better than tarot cards or crystal balls, the authors show that intricate analyses of observational clinical data can improve physicians’ ability to predict the future—at least with respect to as yet uncharacterized adverse drug effects and interactions. Adverse drug events remain a leading cause of morbidity and mortality around the world. Many adverse events are not detected during clinical trials before a drug receives approval for use in the clinic. Fortunately, as part of postmarketing surveillance, regulatory agencies and other institutions maintain large collections of adverse event reports, and these databases present an opportunity to study drug effects from patient population data. However, confounding factors such as concomitant medications, patient demographics, patient medical histories, and reasons for prescribing a drug often are uncharacterized in spontaneous reporting systems, and these omissions can limit the use of quantitative signal detection methods used in the analysis of such data. Here, we present an adaptive data-driven approach for correcting these factors in cases for which the covariates are unknown or unmeasured and combine this approach with existing methods to improve analyses of drug effects using three test data sets. We also present a comprehensive database of drug effects (Offsides) and a database of drug-drug interaction side effects (Twosides). To demonstrate the biological use of these new resources, we used them to identify drug targets, predict drug indications, and discover drug class interactions. We then corroborated 47 (P < 0.0001) of the drug class interactions using an independent analysis of electronic medical records. Our analysis suggests that combined treatment with selective serotonin reuptake inhibitors and thiazides is associated with significantly increased incidence of prolonged QT intervals. We conclude that confounding effects from covariates in observational clinical data can be controlled in data analyses and thus improve the detection and prediction of adverse drug effects and interactions.

Subfields: Computational Theory and Mathematics, Molecular Biology, Toxicology

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Computational Theory and Mathematics

In Molecular Biology

In Toxicology

Citations in the corpus

Modeling polypharmacy side effects with graph convolutional networks (2018, methodology): “The idea is that Decagon learns how to transform and propagate information, captured by node feature vectors, across the graph.”
Deep Patient: An Unsupervised Representation to Predict the Future of Patients from the Electronic Health Records (2016, background): “Recent studies have shown that secondary use of EHRs has enabled data-driven prediction of drug effects and interactions 2 , identification of type 2 diabetes subgroups 3 , discovery of comorbidity clusters in autism spectrum disorders 4 , and improvements in recruiting patients for clinical trials…”
Embedding Logical Queries on Knowledge Graphs (2018, methodology): “We integrated it with the OFFSIDES database, which details off-label associations between drugs and side effects [41].”
Computational Drug Repositioning: From Data to Therapeutics (2013, background): “58 These efforts relating to mining of observed clinical side effects also offer a great opportunity to reposition drugs.”
Pharmacological inhibition of cystine–glutamate exchange induces endoplasmic reticulum stress and ferroptosis (2014, methodology): “First, we identified those reports of patients with exposure to sorafenib, and a set of reports that could serve as controls; for this, we used a high dimensional propensity-score model previously validated for this use in the FAERS that has been shown to mitigate confounding bias and to improve the”
Label Propagation Prediction of Drug-Drug Interactions Based on Clinical Side Effects (2015, methodology): “Of all the drug pairs, 1,892 DDIs are found only in DrugBank, but not found in TWOSIDES.”
Analysis of chemical and biological features yields mechanistic insights into drug side effects. (2013, background): “Indeed, the best event predictors are trained mostly from observational clinical reports that relate drug intakes to SEs (Cami et al., 2011; Tatonetti et al., 2012), with little attention paid to either biological or chemical details.”
A framework for data-driven analysis of materials under uncertainty: Countering the curse of dimensionality (2017, background): “…presence of microbiota in the human gut [75] and termites [76], protein analysis and genetics [77]; biotechnology with the design of new drugs [78,79]; artificial intelligence with robots adapting like animals [80] and programs mastering the GO game [81]; psychology by profiling…”
Network-based approach to prediction and population-based validation of in silico drug repurposing (2018, background): “Focusing on just single targets results in failure to anticipate off-target toxicity, unintended beneficial effects, or multiple target interactions leading to suboptimal efficacy (6-10).”
Identification of Adverse Drug Events from Free Text Electronic Patient Records and Information in a Large Mental Health Case Register (2015, background): “The ability to mine for, and eventually predict, occurrences of ADRs could have significant patient and cost benefits in the future [2].”
The Emerging Paradigm of Network Medicine in the Study of Human Disease (2012, background): “100 Alternatively, network approaches offer substantial insight into the elucidation of the unanticipated “off-target” effects of otherwise presumed direct targeted therapies.”
Graph convolutional networks for computational drug development and discovery (2019, background): “The Off-label Side Effect (OFFSIDES) [98] is a database of 438 801 off-label side effects for 1332 drugs and 10 097 adverse events.”