Kevin P Nguyen1, Cherise Chin Fatt2, Alex Treacher1, Cooper Mellema1, Crystal Cooper3, Manish K Jha2, Benji Kurian2, Maurizio Fava4, Patrick J McGrath5, Myrna Weissman5, Mary L Phillips6, Madhukar H Trivedi7, Albert A Montillo8. 1. Lyda Hill Department of Bioinformatics, University of Texas Southwestern Medical Center, Dallas, Texas. 2. Department of Psychiatry, University of Texas Southwestern Medical Center, Dallas, Texas. 3. Department of Psychiatry, University of Texas Southwestern Medical Center, Dallas, Texas; Jane and John Justin Neuroscience Center, Cook Children's Health Care System, Fort Worth, Texas. 4. Department of Psychiatry, Massachusetts General Hospital, Boston, Massachusetts. 5. New York State Psychiatric Institute and Department of Psychiatry, College of Physicians and Surgeons of Columbia University, New York, New York. 6. Department of Psychiatry, University of Pittsburgh School of Medicine, Pittsburgh, Pennsylvania. 7. Department of Psychiatry, University of Texas Southwestern Medical Center, Dallas, Texas. Electronic address: madhukar.trivedi@utsouthwestern.edu. 8. Lyda Hill Department of Bioinformatics, University of Texas Southwestern Medical Center, Dallas, Texas; Department of Radiology, University of Texas Southwestern Medical Center, Dallas, Texas; Advanced Imaging Research Center, University of Texas Southwestern Medical Center, Dallas, Texas. Electronic address: Albert.Montillo@UTSouthwestern.edu.
Abstract
BACKGROUND: The lack of biomarkers to inform antidepressant selection is a key challenge in personalized depression treatment. This work identifies candidate biomarkers by building deep learning predictors of individual treatment outcomes using reward processing measures from functional magnetic resonance imaging, clinical assessments, and demographics. METHODS: Participants in the EMBARC (Establishing Moderators and Biosignatures of Antidepressant Response in Clinical Care) study (n = 222) underwent reward processing task-based functional magnetic resonance imaging at baseline and were randomized to 8 weeks of sertraline (n = 106) or placebo (n = 116). Subsequently, sertraline nonresponders (n = 37) switched to 8 weeks of bupropion. The change in Hamilton Depression Rating Scale was measured after treatment. Reward processing, clinical measurements, and demographics were used to train treatment-specific deep learning models. RESULTS: The predictive model for sertraline achieved R2 of 48% (95% CI, 33%-61%; p < 10-3) in predicting the change in Hamilton Depression Rating Scale and number-needed-to-treat (NNT) of 4.86 participants in predicting response. The placebo model achieved R2 of 28% (95% CI, 15%-42%; p < 10-3) and NNT of 2.95 in predicting response. The bupropion model achieved R2 of 34% (95% CI, 10%-59%, p < 10-3) and NNT of 1.68 in predicting response. Brain regions where reward processing activity was predictive included the prefrontal cortex and cerebellar crus 1 for sertraline and the cingulate cortex, caudate, orbitofrontal cortex, and crus 1 for bupropion. CONCLUSIONS: These findings demonstrate the utility of reward processing measurements and deep learning to predict antidepressant outcomes and to form multimodal treatment biomarkers.
BACKGROUND: The lack of biomarkers to inform antidepressant selection is a key challenge in personalized depression treatment. This work identifies candidate biomarkers by building deep learning predictors of individual treatment outcomes using reward processing measures from functional magnetic resonance imaging, clinical assessments, and demographics. METHODS: Participants in the EMBARC (Establishing Moderators and Biosignatures of Antidepressant Response in Clinical Care) study (n = 222) underwent reward processing task-based functional magnetic resonance imaging at baseline and were randomized to 8 weeks of sertraline (n = 106) or placebo (n = 116). Subsequently, sertraline nonresponders (n = 37) switched to 8 weeks of bupropion. The change in Hamilton Depression Rating Scale was measured after treatment. Reward processing, clinical measurements, and demographics were used to train treatment-specific deep learning models. RESULTS: The predictive model for sertraline achieved R2 of 48% (95% CI, 33%-61%; p < 10-3) in predicting the change in Hamilton Depression Rating Scale and number-needed-to-treat (NNT) of 4.86 participants in predicting response. The placebo model achieved R2 of 28% (95% CI, 15%-42%; p < 10-3) and NNT of 2.95 in predicting response. The bupropion model achieved R2 of 34% (95% CI, 10%-59%, p < 10-3) and NNT of 1.68 in predicting response. Brain regions where reward processing activity was predictive included the prefrontal cortex and cerebellar crus 1 for sertraline and the cingulate cortex, caudate, orbitofrontal cortex, and crus 1 for bupropion. CONCLUSIONS: These findings demonstrate the utility of reward processing measurements and deep learning to predict antidepressant outcomes and to form multimodal treatment biomarkers.
Authors: Jamie M Dupuy; Michael J Ostacher; Jeffrey Huffman; Roy H Perlis; Andrew A Nierenberg Journal: Int J Neuropsychopharmacol Date: 2011-02-24 Impact factor: 5.176
Authors: A John Rush; Madhukar H Trivedi; Stephen R Wisniewski; Andrew A Nierenberg; Jonathan W Stewart; Diane Warden; George Niederehe; Michael E Thase; Philip W Lavori; Barry D Lebowitz; Patrick J McGrath; Jerrold F Rosenbaum; Harold A Sackeim; David J Kupfer; James Luther; Maurizio Fava Journal: Am J Psychiatry Date: 2006-11 Impact factor: 18.112
Authors: Erika E Forbes; J Christopher May; Greg J Siegle; Cecile D Ladouceur; Neal D Ryan; Cameron S Carter; Boris Birmaher; David A Axelson; Ronald E Dahl Journal: J Child Psychol Psychiatry Date: 2006-10 Impact factor: 8.982
Authors: Diego A Pizzagalli; Christian A Webb; Daniel G Dillon; Craig E Tenke; Jürgen Kayser; Franziska Goer; Maurizio Fava; Patrick McGrath; Myrna Weissman; Ramin Parsey; Phil Adams; Joseph Trombello; Crystal Cooper; Patricia Deldin; Maria A Oquendo; Melvin G McInnis; Thomas Carmody; Gerard Bruder; Madhukar H Trivedi Journal: JAMA Psychiatry Date: 2018-06-01 Impact factor: 21.596
Authors: Chi-Hua Chen; Khanum Ridler; John Suckling; Steve Williams; Cynthia H Y Fu; Emilio Merlo-Pich; Ed Bullmore Journal: Biol Psychiatry Date: 2007-01-09 Impact factor: 13.382