Pharmacogenomics-Driven Prediction of Antidepressant Treatment Outcomes: A Machine-Learning Approach With Multi-trial Replication

Arjun P. Athreya, Drew Neavin, Tania Carrillo-Roa, Michelle Skime, Joanna M Biernacka, Mark A Frye, A. John Rush, Liewei Wang, Elisabeth B. Binder, Ravishankar K. Iyer, Richard M Weinshilboum, William V Bobo

Research output: Contribution to journalArticle

4 Citations (Scopus)

Abstract

We set out to determine whether machine learning–based algorithms that included functionally validated pharmacogenomic biomarkers joined with clinical measures could predict selective serotonin reuptake inhibitor (SSRI) remission/response in patients with major depressive disorder (MDD). We studied 1,030 white outpatients with MDD treated with citalopram/escitalopram in the Mayo Clinic Pharmacogenomics Research Network Antidepressant Medication Pharmacogenomic Study (PGRN-AMPS; n = 398), Sequenced Treatment Alternatives to Relieve Depression (STAR*D; n = 467), and International SSRI Pharmacogenomics Consortium (ISPC; n = 165) trials. A genomewide association study for PGRN-AMPS plasma metabolites associated with SSRI response (serotonin) and baseline MDD severity (kynurenine) identified single nucleotide polymorphisms (SNPs) in DEFB1, ERICH3, AHR, and TSPAN5 that we tested as predictors. Supervised machine-learning methods trained using SNPs and total baseline depression scores predicted remission and response at 8 weeks with area under the receiver operating curve (AUC) > 0.7 (P < 0.04) in PGRN-AMPS patients, with comparable prediction accuracies > 69% (P ≤ 0.07) in STAR*D and ISPC. These results demonstrate that machine learning can achieve accurate and, importantly, replicable prediction of SSRI therapy response using total baseline depression severity combined with pharmacogenomic biomarkers.

Original languageEnglish (US)
JournalClinical pharmacology and therapeutics
DOIs
StatePublished - Jan 1 2019

Fingerprint

Pharmacogenetics
Serotonin Uptake Inhibitors
Antidepressive Agents
Major Depressive Disorder
Citalopram
Depression
Single Nucleotide Polymorphism
Biomarkers
Kynurenine
Area Under Curve
Serotonin
Outpatients
Machine Learning
Therapeutics
Research

ASJC Scopus subject areas

  • Pharmacology
  • Pharmacology (medical)

Cite this

Pharmacogenomics-Driven Prediction of Antidepressant Treatment Outcomes : A Machine-Learning Approach With Multi-trial Replication. / Athreya, Arjun P.; Neavin, Drew; Carrillo-Roa, Tania; Skime, Michelle; Biernacka, Joanna M; Frye, Mark A; Rush, A. John; Wang, Liewei; Binder, Elisabeth B.; Iyer, Ravishankar K.; Weinshilboum, Richard M; Bobo, William V.

In: Clinical pharmacology and therapeutics, 01.01.2019.

Research output: Contribution to journalArticle

Athreya, Arjun P. ; Neavin, Drew ; Carrillo-Roa, Tania ; Skime, Michelle ; Biernacka, Joanna M ; Frye, Mark A ; Rush, A. John ; Wang, Liewei ; Binder, Elisabeth B. ; Iyer, Ravishankar K. ; Weinshilboum, Richard M ; Bobo, William V. / Pharmacogenomics-Driven Prediction of Antidepressant Treatment Outcomes : A Machine-Learning Approach With Multi-trial Replication. In: Clinical pharmacology and therapeutics. 2019.
@article{ddc95bc0ddb24818bf3bf789a3d5c155,
title = "Pharmacogenomics-Driven Prediction of Antidepressant Treatment Outcomes: A Machine-Learning Approach With Multi-trial Replication",
abstract = "We set out to determine whether machine learning–based algorithms that included functionally validated pharmacogenomic biomarkers joined with clinical measures could predict selective serotonin reuptake inhibitor (SSRI) remission/response in patients with major depressive disorder (MDD). We studied 1,030 white outpatients with MDD treated with citalopram/escitalopram in the Mayo Clinic Pharmacogenomics Research Network Antidepressant Medication Pharmacogenomic Study (PGRN-AMPS; n = 398), Sequenced Treatment Alternatives to Relieve Depression (STAR*D; n = 467), and International SSRI Pharmacogenomics Consortium (ISPC; n = 165) trials. A genomewide association study for PGRN-AMPS plasma metabolites associated with SSRI response (serotonin) and baseline MDD severity (kynurenine) identified single nucleotide polymorphisms (SNPs) in DEFB1, ERICH3, AHR, and TSPAN5 that we tested as predictors. Supervised machine-learning methods trained using SNPs and total baseline depression scores predicted remission and response at 8 weeks with area under the receiver operating curve (AUC) > 0.7 (P < 0.04) in PGRN-AMPS patients, with comparable prediction accuracies > 69{\%} (P ≤ 0.07) in STAR*D and ISPC. These results demonstrate that machine learning can achieve accurate and, importantly, replicable prediction of SSRI therapy response using total baseline depression severity combined with pharmacogenomic biomarkers.",
author = "Athreya, {Arjun P.} and Drew Neavin and Tania Carrillo-Roa and Michelle Skime and Biernacka, {Joanna M} and Frye, {Mark A} and Rush, {A. John} and Liewei Wang and Binder, {Elisabeth B.} and Iyer, {Ravishankar K.} and Weinshilboum, {Richard M} and Bobo, {William V}",
year = "2019",
month = "1",
day = "1",
doi = "10.1002/cpt.1482",
language = "English (US)",
journal = "Clinical Pharmacology and Therapeutics",
issn = "0009-9236",
publisher = "Nature Publishing Group",

}

TY - JOUR

T1 - Pharmacogenomics-Driven Prediction of Antidepressant Treatment Outcomes

T2 - A Machine-Learning Approach With Multi-trial Replication

AU - Athreya, Arjun P.

AU - Neavin, Drew

AU - Carrillo-Roa, Tania

AU - Skime, Michelle

AU - Biernacka, Joanna M

AU - Frye, Mark A

AU - Rush, A. John

AU - Wang, Liewei

AU - Binder, Elisabeth B.

AU - Iyer, Ravishankar K.

AU - Weinshilboum, Richard M

AU - Bobo, William V

PY - 2019/1/1

Y1 - 2019/1/1

N2 - We set out to determine whether machine learning–based algorithms that included functionally validated pharmacogenomic biomarkers joined with clinical measures could predict selective serotonin reuptake inhibitor (SSRI) remission/response in patients with major depressive disorder (MDD). We studied 1,030 white outpatients with MDD treated with citalopram/escitalopram in the Mayo Clinic Pharmacogenomics Research Network Antidepressant Medication Pharmacogenomic Study (PGRN-AMPS; n = 398), Sequenced Treatment Alternatives to Relieve Depression (STAR*D; n = 467), and International SSRI Pharmacogenomics Consortium (ISPC; n = 165) trials. A genomewide association study for PGRN-AMPS plasma metabolites associated with SSRI response (serotonin) and baseline MDD severity (kynurenine) identified single nucleotide polymorphisms (SNPs) in DEFB1, ERICH3, AHR, and TSPAN5 that we tested as predictors. Supervised machine-learning methods trained using SNPs and total baseline depression scores predicted remission and response at 8 weeks with area under the receiver operating curve (AUC) > 0.7 (P < 0.04) in PGRN-AMPS patients, with comparable prediction accuracies > 69% (P ≤ 0.07) in STAR*D and ISPC. These results demonstrate that machine learning can achieve accurate and, importantly, replicable prediction of SSRI therapy response using total baseline depression severity combined with pharmacogenomic biomarkers.

AB - We set out to determine whether machine learning–based algorithms that included functionally validated pharmacogenomic biomarkers joined with clinical measures could predict selective serotonin reuptake inhibitor (SSRI) remission/response in patients with major depressive disorder (MDD). We studied 1,030 white outpatients with MDD treated with citalopram/escitalopram in the Mayo Clinic Pharmacogenomics Research Network Antidepressant Medication Pharmacogenomic Study (PGRN-AMPS; n = 398), Sequenced Treatment Alternatives to Relieve Depression (STAR*D; n = 467), and International SSRI Pharmacogenomics Consortium (ISPC; n = 165) trials. A genomewide association study for PGRN-AMPS plasma metabolites associated with SSRI response (serotonin) and baseline MDD severity (kynurenine) identified single nucleotide polymorphisms (SNPs) in DEFB1, ERICH3, AHR, and TSPAN5 that we tested as predictors. Supervised machine-learning methods trained using SNPs and total baseline depression scores predicted remission and response at 8 weeks with area under the receiver operating curve (AUC) > 0.7 (P < 0.04) in PGRN-AMPS patients, with comparable prediction accuracies > 69% (P ≤ 0.07) in STAR*D and ISPC. These results demonstrate that machine learning can achieve accurate and, importantly, replicable prediction of SSRI therapy response using total baseline depression severity combined with pharmacogenomic biomarkers.

UR - http://www.scopus.com/inward/record.url?scp=85068223528&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85068223528&partnerID=8YFLogxK

U2 - 10.1002/cpt.1482

DO - 10.1002/cpt.1482

M3 - Article

C2 - 31012492

AN - SCOPUS:85068223528

JO - Clinical Pharmacology and Therapeutics

JF - Clinical Pharmacology and Therapeutics

SN - 0009-9236

ER -