Limits...
An Integrative Data Science Pipeline to Identify Novel Drug Interactions that Prolong the QT Interval.

Lorberbaum T, Sampson KJ, Woosley RL, Kass RS, Tatonetti NP - Drug Saf (2016)

Bottom Line: In the EHR at Columbia University Medical Center, we compared the QTc intervals of patients prescribed the flagged drug pairs with patients prescribed either drug individually.We created an AE fingerprint consisting of 13 latently detected side effects.Of 889 pairs flagged in FAERS, eight novel QT-DDIs were significantly associated with prolonged QTc intervals in the EHR and were not due to co-prescribed medications.

View Article: PubMed Central - PubMed

Affiliation: Department of Physiology and Cellular Biophysics, Columbia University, New York, NY, USA.

ABSTRACT

Introduction: Drug-induced prolongation of the QT interval on the electrocardiogram (long QT syndrome, LQTS) can lead to a potentially fatal ventricular arrhythmia known as torsades de pointes (TdP). Over 40 drugs with both cardiac and non-cardiac indications are associated with increased risk of TdP, but drug-drug interactions contributing to LQTS (QT-DDIs) remain poorly characterized. Traditional methods for mining observational healthcare data are poorly equipped to detect QT-DDI signals due to low reporting numbers and lack of direct evidence for LQTS.

Objective: We hypothesized that LQTS could be identified latently using an adverse event (AE) fingerprint of more commonly reported AEs. We aimed to generate an integrated data science pipeline that addresses current limitations by identifying latent signals for QT-DDIs in the US FDA's Adverse Event Reporting System (FAERS) and retrospectively validating these predictions using electrocardiogram data in electronic health records (EHRs).

Methods: We trained a model to identify an AE fingerprint for risk of TdP for single drugs and applied this model to drug pair data to predict novel DDIs. In the EHR at Columbia University Medical Center, we compared the QTc intervals of patients prescribed the flagged drug pairs with patients prescribed either drug individually.

Results: We created an AE fingerprint consisting of 13 latently detected side effects. This model significantly outperformed a direct evidence control model in the detection of established interactions (p = 1.62E-3) and significantly enriched for validated QT-DDIs in the EHR (p = 0.01). Of 889 pairs flagged in FAERS, eight novel QT-DDIs were significantly associated with prolonged QTc intervals in the EHR and were not due to co-prescribed medications.

Conclusions: Latent signal detection in FAERS validated using the EHR presents an automated and data-driven approach for systematically identifying novel QT-DDIs. The high-confidence hypotheses flagged using this method warrant further investigation.

No MeSH data available.


Related in: MedlinePlus

Overview of DIPULSE pipeline, which combines mining of FAERS and EHRs to flag novel QT-prolonging DDIs. FAERS: We generate an AE reporting frequency table (dimensions, N drugs by M AEs) for single drugs in FAERS. The value at a row and column represents the fraction of reports for drug i containing AE k (Fik). We label a drug as a positive example (shown in red) if it has a known risk of TdP (obtained from http://www.CredibleMeds.org). All drugs not found in CredibleMeds were labeled as negative examples (shown in green). We use machine learning to generate an AE fingerprint model that identified the most predictive subset of features (AE reporting frequencies, Fik) as latent evidence for predicting whether a drug does or does not prolong the QT interval (gray boxes). We then apply this fingerprint model to an independent test data set consisting of a matrix (with AE reporting frequencies Fijk) for drug pairs. We send pairs receiving high classifier probabilities (but where neither individual drug is known to prolong the QT interval) for EHR validation (in this case pairs (DN−1, DN−2) [purple-blue] and (DN−1, DN) [purple-orange]). EHR: We validate putative interactions using electrocardiogram laboratory results in the EHRs by determining whether patients prescribed a predicted interacting drug pair had increased QTc intervals compared with patients taking either drug alone. In this example, patients prescribed the drug pair (DN−1, DN−2) have a significantly increased QT interval compared with patients on either drug alone. This is not observed for drug pair (DN−1, DN) so it is filtered out. Finally, we performed a confounder analysis to confirm that the significant increase observed in QTc interval is not due to other co-prescribed medications. DIPULSE Drug Interaction Prediction Using Latent Signals and EHRs, EHRs electronic health records, FAERS FDA Adverse Event Reporting System, DDIs drug–drug interactions, AE adverse event, TdP torsades de pointes, QTc heart rate-corrected QT interval
© Copyright Policy - OpenAccess
Related In: Results  -  Collection


getmorefigures.php?uid=PMC4835515&req=5

Fig1: Overview of DIPULSE pipeline, which combines mining of FAERS and EHRs to flag novel QT-prolonging DDIs. FAERS: We generate an AE reporting frequency table (dimensions, N drugs by M AEs) for single drugs in FAERS. The value at a row and column represents the fraction of reports for drug i containing AE k (Fik). We label a drug as a positive example (shown in red) if it has a known risk of TdP (obtained from http://www.CredibleMeds.org). All drugs not found in CredibleMeds were labeled as negative examples (shown in green). We use machine learning to generate an AE fingerprint model that identified the most predictive subset of features (AE reporting frequencies, Fik) as latent evidence for predicting whether a drug does or does not prolong the QT interval (gray boxes). We then apply this fingerprint model to an independent test data set consisting of a matrix (with AE reporting frequencies Fijk) for drug pairs. We send pairs receiving high classifier probabilities (but where neither individual drug is known to prolong the QT interval) for EHR validation (in this case pairs (DN−1, DN−2) [purple-blue] and (DN−1, DN) [purple-orange]). EHR: We validate putative interactions using electrocardiogram laboratory results in the EHRs by determining whether patients prescribed a predicted interacting drug pair had increased QTc intervals compared with patients taking either drug alone. In this example, patients prescribed the drug pair (DN−1, DN−2) have a significantly increased QT interval compared with patients on either drug alone. This is not observed for drug pair (DN−1, DN) so it is filtered out. Finally, we performed a confounder analysis to confirm that the significant increase observed in QTc interval is not due to other co-prescribed medications. DIPULSE Drug Interaction Prediction Using Latent Signals and EHRs, EHRs electronic health records, FAERS FDA Adverse Event Reporting System, DDIs drug–drug interactions, AE adverse event, TdP torsades de pointes, QTc heart rate-corrected QT interval

Mentions: A graphical overview of DIPULSE can be found in Fig. 1. The individual steps of the pipeline corresponding to each panel of the figure are described in detail below. Briefly, we used AE reporting frequencies for individual drugs to identify an AE fingerprint for increased risk of TdP. We then apply this model to a test data set of AE reporting frequencies for drug pairs. We filtered for high-confidence predictions and proceeded to validate these putative QT-DDIs in the EHR by comparing the QTc (heart rate-corrected QT) intervals of patients prescribed the flagged drug pair with patients prescribed either drug alone. Finally, we perform a confounder analysis to remove any associations that can be explained by co-prescribed medications, and generated a final candidate list of novel QT-DDIs.Fig. 1


An Integrative Data Science Pipeline to Identify Novel Drug Interactions that Prolong the QT Interval.

Lorberbaum T, Sampson KJ, Woosley RL, Kass RS, Tatonetti NP - Drug Saf (2016)

Overview of DIPULSE pipeline, which combines mining of FAERS and EHRs to flag novel QT-prolonging DDIs. FAERS: We generate an AE reporting frequency table (dimensions, N drugs by M AEs) for single drugs in FAERS. The value at a row and column represents the fraction of reports for drug i containing AE k (Fik). We label a drug as a positive example (shown in red) if it has a known risk of TdP (obtained from http://www.CredibleMeds.org). All drugs not found in CredibleMeds were labeled as negative examples (shown in green). We use machine learning to generate an AE fingerprint model that identified the most predictive subset of features (AE reporting frequencies, Fik) as latent evidence for predicting whether a drug does or does not prolong the QT interval (gray boxes). We then apply this fingerprint model to an independent test data set consisting of a matrix (with AE reporting frequencies Fijk) for drug pairs. We send pairs receiving high classifier probabilities (but where neither individual drug is known to prolong the QT interval) for EHR validation (in this case pairs (DN−1, DN−2) [purple-blue] and (DN−1, DN) [purple-orange]). EHR: We validate putative interactions using electrocardiogram laboratory results in the EHRs by determining whether patients prescribed a predicted interacting drug pair had increased QTc intervals compared with patients taking either drug alone. In this example, patients prescribed the drug pair (DN−1, DN−2) have a significantly increased QT interval compared with patients on either drug alone. This is not observed for drug pair (DN−1, DN) so it is filtered out. Finally, we performed a confounder analysis to confirm that the significant increase observed in QTc interval is not due to other co-prescribed medications. DIPULSE Drug Interaction Prediction Using Latent Signals and EHRs, EHRs electronic health records, FAERS FDA Adverse Event Reporting System, DDIs drug–drug interactions, AE adverse event, TdP torsades de pointes, QTc heart rate-corrected QT interval
© Copyright Policy - OpenAccess
Related In: Results  -  Collection

Show All Figures
getmorefigures.php?uid=PMC4835515&req=5

Fig1: Overview of DIPULSE pipeline, which combines mining of FAERS and EHRs to flag novel QT-prolonging DDIs. FAERS: We generate an AE reporting frequency table (dimensions, N drugs by M AEs) for single drugs in FAERS. The value at a row and column represents the fraction of reports for drug i containing AE k (Fik). We label a drug as a positive example (shown in red) if it has a known risk of TdP (obtained from http://www.CredibleMeds.org). All drugs not found in CredibleMeds were labeled as negative examples (shown in green). We use machine learning to generate an AE fingerprint model that identified the most predictive subset of features (AE reporting frequencies, Fik) as latent evidence for predicting whether a drug does or does not prolong the QT interval (gray boxes). We then apply this fingerprint model to an independent test data set consisting of a matrix (with AE reporting frequencies Fijk) for drug pairs. We send pairs receiving high classifier probabilities (but where neither individual drug is known to prolong the QT interval) for EHR validation (in this case pairs (DN−1, DN−2) [purple-blue] and (DN−1, DN) [purple-orange]). EHR: We validate putative interactions using electrocardiogram laboratory results in the EHRs by determining whether patients prescribed a predicted interacting drug pair had increased QTc intervals compared with patients taking either drug alone. In this example, patients prescribed the drug pair (DN−1, DN−2) have a significantly increased QT interval compared with patients on either drug alone. This is not observed for drug pair (DN−1, DN) so it is filtered out. Finally, we performed a confounder analysis to confirm that the significant increase observed in QTc interval is not due to other co-prescribed medications. DIPULSE Drug Interaction Prediction Using Latent Signals and EHRs, EHRs electronic health records, FAERS FDA Adverse Event Reporting System, DDIs drug–drug interactions, AE adverse event, TdP torsades de pointes, QTc heart rate-corrected QT interval
Mentions: A graphical overview of DIPULSE can be found in Fig. 1. The individual steps of the pipeline corresponding to each panel of the figure are described in detail below. Briefly, we used AE reporting frequencies for individual drugs to identify an AE fingerprint for increased risk of TdP. We then apply this model to a test data set of AE reporting frequencies for drug pairs. We filtered for high-confidence predictions and proceeded to validate these putative QT-DDIs in the EHR by comparing the QTc (heart rate-corrected QT) intervals of patients prescribed the flagged drug pair with patients prescribed either drug alone. Finally, we perform a confounder analysis to remove any associations that can be explained by co-prescribed medications, and generated a final candidate list of novel QT-DDIs.Fig. 1

Bottom Line: In the EHR at Columbia University Medical Center, we compared the QTc intervals of patients prescribed the flagged drug pairs with patients prescribed either drug individually.We created an AE fingerprint consisting of 13 latently detected side effects.Of 889 pairs flagged in FAERS, eight novel QT-DDIs were significantly associated with prolonged QTc intervals in the EHR and were not due to co-prescribed medications.

View Article: PubMed Central - PubMed

Affiliation: Department of Physiology and Cellular Biophysics, Columbia University, New York, NY, USA.

ABSTRACT

Introduction: Drug-induced prolongation of the QT interval on the electrocardiogram (long QT syndrome, LQTS) can lead to a potentially fatal ventricular arrhythmia known as torsades de pointes (TdP). Over 40 drugs with both cardiac and non-cardiac indications are associated with increased risk of TdP, but drug-drug interactions contributing to LQTS (QT-DDIs) remain poorly characterized. Traditional methods for mining observational healthcare data are poorly equipped to detect QT-DDI signals due to low reporting numbers and lack of direct evidence for LQTS.

Objective: We hypothesized that LQTS could be identified latently using an adverse event (AE) fingerprint of more commonly reported AEs. We aimed to generate an integrated data science pipeline that addresses current limitations by identifying latent signals for QT-DDIs in the US FDA's Adverse Event Reporting System (FAERS) and retrospectively validating these predictions using electrocardiogram data in electronic health records (EHRs).

Methods: We trained a model to identify an AE fingerprint for risk of TdP for single drugs and applied this model to drug pair data to predict novel DDIs. In the EHR at Columbia University Medical Center, we compared the QTc intervals of patients prescribed the flagged drug pairs with patients prescribed either drug individually.

Results: We created an AE fingerprint consisting of 13 latently detected side effects. This model significantly outperformed a direct evidence control model in the detection of established interactions (p = 1.62E-3) and significantly enriched for validated QT-DDIs in the EHR (p = 0.01). Of 889 pairs flagged in FAERS, eight novel QT-DDIs were significantly associated with prolonged QTc intervals in the EHR and were not due to co-prescribed medications.

Conclusions: Latent signal detection in FAERS validated using the EHR presents an automated and data-driven approach for systematically identifying novel QT-DDIs. The high-confidence hypotheses flagged using this method warrant further investigation.

No MeSH data available.


Related in: MedlinePlus