Using polygenic scores and clinical data for bipolar disorder patient stratification and lithium response prediction: machine learning approach

Micah Cearns; Azmeraw T. Amare; Klaus Oliver Schubert; Anbupalam Thalamuthu; Joseph Frank; Fabian Streit; Mazda Adli; Nirmala Akula; Kazufumi Akiyama; Raffaella Ardau; Bárbara Arias; Jean Michel Aubry; Lena Backlund; Abesh Kumar Bhattacharjee; Frank Bellivier; Antonio Benabarre; Susanne Bengesser; Joanna M. Biernacka; Armin Birner; Clara Brichant-Petitjean; Pablo Cervantes; Hsi Chung Chen; Caterina Chillotti; Sven Cichon; Cristiana Cruceanu; Piotr M. Czerski; Nina Dalkner; Alexandre Dayer; Franziska Degenhardt; Maria Del Zompo; J. Raymond Depaulo; Bruno Étain; Peter Falkai; Andreas J. Forstner; Louise Frisen; Mark A. Frye; Janice M. Fullerton; Sébastien Gard; Julie S. Garnham; Fernando S. Goes; Maria Grigoroiu-Serbanescu; Paul Grof; Ryota Hashimoto; Joanna Hauser; Urs Heilbronner; Stefan Herms; Per Hoffmann; Andrea Hofmann; Liping Hou; Yi Hsiang Hsu; Stephane Jamain; Esther Jiménez; Jean Pierre Kahn; Layla Kassem; Po Hsiu Kuo; Tadafumi Kato; John Kelsoe; Sarah Kittel-Schneider; Sebastian Kliwicki; Barbara König; Ichiro Kusumi; Gonzalo Laje; Mikael Landén; Catharina Lavebratt; Marion Leboyer; Susan G. Leckband; Mario Maj; Mirko Manchia; Lina Martinsson; Michael J. McCarthy; Susan McElroy; Francesc Colom; Marina Mitjans; Francis M. Mondimore; Palmiero Monteleone; Caroline M. Nievergelt; Markus M. Nöthen; Tomas Novák; Claire O'Donovan; Norio Ozaki; Vincent Millischer; Sergi Papiol; Andrea Pfennig; Claudia Pisanu; James B. Potash; Andreas Reif; Eva Reininghaus; Guy A. Rouleau; Janusz K. Rybakowski; Martin Schalling; Peter R. Schofield; Barbara W. Schweizer; Giovanni Severino; Tatyana Shekhtman; Paul D. Shilling; Katzutaka Shimoda; Christian Simhandl; Claire M. Slaney; Alessio Squassina; Thomas Stamm; Pavla Stopkova; Fasil Tekola-Ayele; Alfonso Tortorella; Gustavo Turecki; Julia Veeh; Eduard Vieta; Stephanie H. Witt; Gloria Roberts; Peter P. Zandi; Martin Alda; Michael Bauer; Francis J. McMahon; Philip B. Mitchell; Thomas G. Schulze; Marcella Rietschel; Scott R. Clark; Bernhard T. Baune

doi:10.1192/bjp.2022.28

Using polygenic scores and clinical data for bipolar disorder patient stratification and lithium response prediction: machine learning approach

Micah Cearns, Azmeraw T. Amare, Klaus Oliver Schubert, Anbupalam Thalamuthu, Joseph Frank, Fabian Streit, Mazda Adli, Nirmala Akula, Kazufumi Akiyama, Raffaella Ardau, Bárbara Arias, Jean Michel Aubry, Lena Backlund, Abesh Kumar Bhattacharjee, Frank Bellivier, Antonio Benabarre, Susanne Bengesser, Joanna M. Biernacka, Armin Birner, Clara Brichant-PetitjeanPablo Cervantes, Hsi Chung Chen, Caterina Chillotti, Sven Cichon, Cristiana Cruceanu, Piotr M. Czerski, Nina Dalkner, Alexandre Dayer, Franziska Degenhardt, Maria Del Zompo, J. Raymond Depaulo, Bruno Étain, Peter Falkai, Andreas J. Forstner, Louise Frisen, Mark A. Frye, Janice M. Fullerton, Sébastien Gard, Julie S. Garnham, Fernando S. Goes, Maria Grigoroiu-Serbanescu, Paul Grof, Ryota Hashimoto, Joanna Hauser, Urs Heilbronner, Stefan Herms, Per Hoffmann, Andrea Hofmann, Liping Hou, Yi Hsiang Hsu, Stephane Jamain, Esther Jiménez, Jean Pierre Kahn, Layla Kassem, Po Hsiu Kuo, Tadafumi Kato, John Kelsoe, Sarah Kittel-Schneider, Sebastian Kliwicki, Barbara König, Ichiro Kusumi, Gonzalo Laje, Mikael Landén, Catharina Lavebratt, Marion Leboyer, Susan G. Leckband, Mario Maj, Mirko Manchia, Lina Martinsson, Michael J. McCarthy, Susan McElroy, Francesc Colom, Marina Mitjans, Francis M. Mondimore, Palmiero Monteleone, Caroline M. Nievergelt, Markus M. Nöthen, Tomas Novák, Claire O'Donovan, Norio Ozaki, Vincent Millischer, Sergi Papiol, Andrea Pfennig, Claudia Pisanu, James B. Potash, Andreas Reif, Eva Reininghaus, Guy A. Rouleau, Janusz K. Rybakowski, Martin Schalling, Peter R. Schofield, Barbara W. Schweizer, Giovanni Severino, Tatyana Shekhtman, Paul D. Shilling, Katzutaka Shimoda, Christian Simhandl, Claire M. Slaney, Alessio Squassina, Thomas Stamm, Pavla Stopkova, Fasil Tekola-Ayele, Alfonso Tortorella, Gustavo Turecki, Julia Veeh, Eduard Vieta, Stephanie H. Witt, Gloria Roberts, Peter P. Zandi, Martin Alda, Michael Bauer, Francis J. McMahon, Philip B. Mitchell, Thomas G. Schulze, Marcella Rietschel, Scott R. Clark, Bernhard T. Baune

Research output: Contribution to journal › Article › peer-review

Abstract

Background Response to lithium in patients with bipolar disorder is associated with clinical and transdiagnostic genetic factors. The predictive combination of these variables might help clinicians better predict which patients will respond to lithium treatment. Aims To use a combination of transdiagnostic genetic and clinical factors to predict lithium response in patients with bipolar disorder. Method This study utilised genetic and clinical data (n = 1034) collected as part of the International Consortium on Lithium Genetics (ConLi+Gen) project. Polygenic risk scores (PRS) were computed for schizophrenia and major depressive disorder, and then combined with clinical variables using a cross-validated machine-learning regression approach. Unimodal, multimodal and genetically stratified models were trained and validated using ridge, elastic net and random forest regression on 692 patients with bipolar disorder from ten study sites using leave-site-out cross-validation. All models were then tested on an independent test set of 342 patients. The best performing models were then tested in a classification framework. Results The best performing linear model explained 5.1% (P = 0.0001) of variance in lithium response and was composed of clinical variables, PRS variables and interaction terms between them. The best performing non-linear model used only clinical variables and explained 8.1% (P = 0.0001) of variance in lithium response. A priori genomic stratification improved non-linear model performance to 13.7% (P = 0.0001) and improved the binary classification of lithium response. This model stratified patients based on their meta-polygenic loadings for major depressive disorder and schizophrenia and was then trained using clinical data. Conclusions Using PRS to first stratify patients genetically and then train machine-learning models with clinical predictors led to large improvements in lithium response prediction. When used with other PRS and biological markers in the future this approach may help inform which patients are most likely to respond to lithium treatment.

Original language	English (US)
Pages (from-to)	219-228
Number of pages	10
Journal	British Journal of Psychiatry
Volume	220
Issue number	4
DOIs	https://doi.org/10.1192/bjp.2022.28
State	Published - Apr 9 2022

Keywords

Mood stabilisers
bipolar affective disorders
depressive disorders
genetics
outcome studies

ASJC Scopus subject areas

Psychiatry and Mental health

Access to Document

10.1192/bjp.2022.28

Cite this

Cearns, M., Amare, A. T., Schubert, K. O., Thalamuthu, A., Frank, J., Streit, F., Adli, M., Akula, N., Akiyama, K., Ardau, R., Arias, B., Aubry, J. M., Backlund, L., Bhattacharjee, A. K., Bellivier, F., Benabarre, A., Bengesser, S., Biernacka, J. M., Birner, A., ... Baune, B. T. (2022). Using polygenic scores and clinical data for bipolar disorder patient stratification and lithium response prediction: machine learning approach. British Journal of Psychiatry, 220(4), 219-228. https://doi.org/10.1192/bjp.2022.28

Cearns, M, Amare, AT, Schubert, KO, Thalamuthu, A, Frank, J, Streit, F, Adli, M, Akula, N, Akiyama, K, Ardau, R, Arias, B, Aubry, JM, Backlund, L, Bhattacharjee, AK, Bellivier, F, Benabarre, A, Bengesser, S, Biernacka, JM, Birner, A, Brichant-Petitjean, C, Cervantes, P, Chen, HC, Chillotti, C, Cichon, S, Cruceanu, C, Czerski, PM, Dalkner, N, Dayer, A, Degenhardt, F, Zompo, MD, Depaulo, JR, Étain, B, Falkai, P, Forstner, AJ, Frisen, L, Frye, MA, Fullerton, JM, Gard, S, Garnham, JS, Goes, FS, Grigoroiu-Serbanescu, M, Grof, P, Hashimoto, R, Hauser, J, Heilbronner, U, Herms, S, Hoffmann, P, Hofmann, A, Hou, L, Hsu, YH, Jamain, S, Jiménez, E, Kahn, JP, Kassem, L, Kuo, PH, Kato, T, Kelsoe, J, Kittel-Schneider, S, Kliwicki, S, König, B, Kusumi, I, Laje, G, Landén, M, Lavebratt, C, Leboyer, M, Leckband, SG, Maj, M, Manchia, M, Martinsson, L, McCarthy, MJ, McElroy, S, Colom, F, Mitjans, M, Mondimore, FM, Monteleone, P, Nievergelt, CM, Nöthen, MM, Novák, T, O'Donovan, C, Ozaki, N, Millischer, V, Papiol, S, Pfennig, A, Pisanu, C, Potash, JB, Reif, A, Reininghaus, E, Rouleau, GA, Rybakowski, JK, Schalling, M, Schofield, PR, Schweizer, BW, Severino, G, Shekhtman, T, Shilling, PD, Shimoda, K, Simhandl, C, Slaney, CM, Squassina, A, Stamm, T, Stopkova, P, Tekola-Ayele, F, Tortorella, A, Turecki, G, Veeh, J, Vieta, E, Witt, SH, Roberts, G, Zandi, PP, Alda, M, Bauer, M, McMahon, FJ, Mitchell, PB, Schulze, TG, Rietschel, M, Clark, SR & Baune, BT 2022, 'Using polygenic scores and clinical data for bipolar disorder patient stratification and lithium response prediction: machine learning approach', British Journal of Psychiatry, vol. 220, no. 4, pp. 219-228. https://doi.org/10.1192/bjp.2022.28

@article{38905e968f404bdb97a312b57d5a3143,

title = "Using polygenic scores and clinical data for bipolar disorder patient stratification and lithium response prediction: machine learning approach",

abstract = "Background Response to lithium in patients with bipolar disorder is associated with clinical and transdiagnostic genetic factors. The predictive combination of these variables might help clinicians better predict which patients will respond to lithium treatment. Aims To use a combination of transdiagnostic genetic and clinical factors to predict lithium response in patients with bipolar disorder. Method This study utilised genetic and clinical data (n = 1034) collected as part of the International Consortium on Lithium Genetics (ConLi+Gen) project. Polygenic risk scores (PRS) were computed for schizophrenia and major depressive disorder, and then combined with clinical variables using a cross-validated machine-learning regression approach. Unimodal, multimodal and genetically stratified models were trained and validated using ridge, elastic net and random forest regression on 692 patients with bipolar disorder from ten study sites using leave-site-out cross-validation. All models were then tested on an independent test set of 342 patients. The best performing models were then tested in a classification framework. Results The best performing linear model explained 5.1% (P = 0.0001) of variance in lithium response and was composed of clinical variables, PRS variables and interaction terms between them. The best performing non-linear model used only clinical variables and explained 8.1% (P = 0.0001) of variance in lithium response. A priori genomic stratification improved non-linear model performance to 13.7% (P = 0.0001) and improved the binary classification of lithium response. This model stratified patients based on their meta-polygenic loadings for major depressive disorder and schizophrenia and was then trained using clinical data. Conclusions Using PRS to first stratify patients genetically and then train machine-learning models with clinical predictors led to large improvements in lithium response prediction. When used with other PRS and biological markers in the future this approach may help inform which patients are most likely to respond to lithium treatment.",

keywords = "Mood stabilisers, bipolar affective disorders, depressive disorders, genetics, outcome studies",

author = "Micah Cearns and Amare, {Azmeraw T.} and Schubert, {Klaus Oliver} and Anbupalam Thalamuthu and Joseph Frank and Fabian Streit and Mazda Adli and Nirmala Akula and Kazufumi Akiyama and Raffaella Ardau and B{\'a}rbara Arias and Aubry, {Jean Michel} and Lena Backlund and Bhattacharjee, {Abesh Kumar} and Frank Bellivier and Antonio Benabarre and Susanne Bengesser and Biernacka, {Joanna M.} and Armin Birner and Clara Brichant-Petitjean and Pablo Cervantes and Chen, {Hsi Chung} and Caterina Chillotti and Sven Cichon and Cristiana Cruceanu and Czerski, {Piotr M.} and Nina Dalkner and Alexandre Dayer and Franziska Degenhardt and Zompo, {Maria Del} and Depaulo, {J. Raymond} and Bruno {\'E}tain and Peter Falkai and Forstner, {Andreas J.} and Louise Frisen and Frye, {Mark A.} and Fullerton, {Janice M.} and S{\'e}bastien Gard and Garnham, {Julie S.} and Goes, {Fernando S.} and Maria Grigoroiu-Serbanescu and Paul Grof and Ryota Hashimoto and Joanna Hauser and Urs Heilbronner and Stefan Herms and Per Hoffmann and Andrea Hofmann and Liping Hou and Hsu, {Yi Hsiang} and Stephane Jamain and Esther Jim{\'e}nez and Kahn, {Jean Pierre} and Layla Kassem and Kuo, {Po Hsiu} and Tadafumi Kato and John Kelsoe and Sarah Kittel-Schneider and Sebastian Kliwicki and Barbara K{\"o}nig and Ichiro Kusumi and Gonzalo Laje and Mikael Land{\'e}n and Catharina Lavebratt and Marion Leboyer and Leckband, {Susan G.} and Mario Maj and Mirko Manchia and Lina Martinsson and McCarthy, {Michael J.} and Susan McElroy and Francesc Colom and Marina Mitjans and Mondimore, {Francis M.} and Palmiero Monteleone and Nievergelt, {Caroline M.} and N{\"o}then, {Markus M.} and Tomas Nov{\'a}k and Claire O'Donovan and Norio Ozaki and Vincent Millischer and Sergi Papiol and Andrea Pfennig and Claudia Pisanu and Potash, {James B.} and Andreas Reif and Eva Reininghaus and Rouleau, {Guy A.} and Rybakowski, {Janusz K.} and Martin Schalling and Schofield, {Peter R.} and Schweizer, {Barbara W.} and Giovanni Severino and Tatyana Shekhtman and Shilling, {Paul D.} and Katzutaka Shimoda and Christian Simhandl and Slaney, {Claire M.} and Alessio Squassina and Thomas Stamm and Pavla Stopkova and Fasil Tekola-Ayele and Alfonso Tortorella and Gustavo Turecki and Julia Veeh and Eduard Vieta and Witt, {Stephanie H.} and Gloria Roberts and Zandi, {Peter P.} and Martin Alda and Michael Bauer and McMahon, {Francis J.} and Mitchell, {Philip B.} and Schulze, {Thomas G.} and Marcella Rietschel and Clark, {Scott R.} and Baune, {Bernhard T.}",

note = "Publisher Copyright: Copyright {\textcopyright} The Author(s), 2022. Published by Cambridge University Press on behalf of the Royal College of Psychiatrists.",

year = "2022",

month = apr,

day = "9",

doi = "10.1192/bjp.2022.28",

language = "English (US)",

volume = "220",

pages = "219--228",

journal = "British Journal of Psychiatry",

issn = "0007-1250",

publisher = "Royal College of Psychiatrists",

number = "4",

}

TY - JOUR

T1 - Using polygenic scores and clinical data for bipolar disorder patient stratification and lithium response prediction

T2 - machine learning approach

AU - Cearns, Micah

AU - Amare, Azmeraw T.

AU - Schubert, Klaus Oliver

AU - Thalamuthu, Anbupalam

AU - Frank, Joseph

AU - Streit, Fabian

AU - Adli, Mazda

AU - Akula, Nirmala

AU - Akiyama, Kazufumi

AU - Ardau, Raffaella

AU - Arias, Bárbara

AU - Aubry, Jean Michel

AU - Backlund, Lena

AU - Bhattacharjee, Abesh Kumar

AU - Bellivier, Frank

AU - Benabarre, Antonio

AU - Bengesser, Susanne

AU - Biernacka, Joanna M.

AU - Birner, Armin

AU - Brichant-Petitjean, Clara

AU - Cervantes, Pablo

AU - Chen, Hsi Chung

AU - Chillotti, Caterina

AU - Cichon, Sven

AU - Cruceanu, Cristiana

AU - Czerski, Piotr M.

AU - Dalkner, Nina

AU - Dayer, Alexandre

AU - Degenhardt, Franziska

AU - Zompo, Maria Del

AU - Depaulo, J. Raymond

AU - Étain, Bruno

AU - Falkai, Peter

AU - Forstner, Andreas J.

AU - Frisen, Louise

AU - Frye, Mark A.

AU - Fullerton, Janice M.

AU - Gard, Sébastien

AU - Garnham, Julie S.

AU - Goes, Fernando S.

AU - Grigoroiu-Serbanescu, Maria

AU - Grof, Paul

AU - Hashimoto, Ryota

AU - Hauser, Joanna

AU - Heilbronner, Urs

AU - Herms, Stefan

AU - Hoffmann, Per

AU - Hofmann, Andrea

AU - Hou, Liping

AU - Hsu, Yi Hsiang

AU - Jamain, Stephane

AU - Jiménez, Esther

AU - Kahn, Jean Pierre

AU - Kassem, Layla

AU - Kuo, Po Hsiu

AU - Kato, Tadafumi

AU - Kelsoe, John

AU - Kittel-Schneider, Sarah

AU - Kliwicki, Sebastian

AU - König, Barbara

AU - Kusumi, Ichiro

AU - Laje, Gonzalo

AU - Landén, Mikael

AU - Lavebratt, Catharina

AU - Leboyer, Marion

AU - Leckband, Susan G.

AU - Maj, Mario

AU - Manchia, Mirko

AU - Martinsson, Lina

AU - McCarthy, Michael J.

AU - McElroy, Susan

AU - Colom, Francesc

AU - Mitjans, Marina

AU - Mondimore, Francis M.

AU - Monteleone, Palmiero

AU - Nievergelt, Caroline M.

AU - Nöthen, Markus M.

AU - Novák, Tomas

AU - O'Donovan, Claire

AU - Ozaki, Norio

AU - Millischer, Vincent

AU - Papiol, Sergi

AU - Pfennig, Andrea

AU - Pisanu, Claudia

AU - Potash, James B.

AU - Reif, Andreas

AU - Reininghaus, Eva

AU - Rouleau, Guy A.

AU - Rybakowski, Janusz K.

AU - Schalling, Martin

AU - Schofield, Peter R.

AU - Schweizer, Barbara W.

AU - Severino, Giovanni

AU - Shekhtman, Tatyana

AU - Shilling, Paul D.

AU - Shimoda, Katzutaka

AU - Simhandl, Christian

AU - Slaney, Claire M.

AU - Squassina, Alessio

AU - Stamm, Thomas

AU - Stopkova, Pavla

AU - Tekola-Ayele, Fasil

AU - Tortorella, Alfonso

AU - Turecki, Gustavo

AU - Veeh, Julia

AU - Vieta, Eduard

AU - Witt, Stephanie H.

AU - Roberts, Gloria

AU - Zandi, Peter P.

AU - Alda, Martin

AU - Bauer, Michael

AU - McMahon, Francis J.

AU - Mitchell, Philip B.

AU - Schulze, Thomas G.

AU - Rietschel, Marcella

AU - Clark, Scott R.

AU - Baune, Bernhard T.

PY - 2022/4/9

Y1 - 2022/4/9

N2 - Background Response to lithium in patients with bipolar disorder is associated with clinical and transdiagnostic genetic factors. The predictive combination of these variables might help clinicians better predict which patients will respond to lithium treatment. Aims To use a combination of transdiagnostic genetic and clinical factors to predict lithium response in patients with bipolar disorder. Method This study utilised genetic and clinical data (n = 1034) collected as part of the International Consortium on Lithium Genetics (ConLi+Gen) project. Polygenic risk scores (PRS) were computed for schizophrenia and major depressive disorder, and then combined with clinical variables using a cross-validated machine-learning regression approach. Unimodal, multimodal and genetically stratified models were trained and validated using ridge, elastic net and random forest regression on 692 patients with bipolar disorder from ten study sites using leave-site-out cross-validation. All models were then tested on an independent test set of 342 patients. The best performing models were then tested in a classification framework. Results The best performing linear model explained 5.1% (P = 0.0001) of variance in lithium response and was composed of clinical variables, PRS variables and interaction terms between them. The best performing non-linear model used only clinical variables and explained 8.1% (P = 0.0001) of variance in lithium response. A priori genomic stratification improved non-linear model performance to 13.7% (P = 0.0001) and improved the binary classification of lithium response. This model stratified patients based on their meta-polygenic loadings for major depressive disorder and schizophrenia and was then trained using clinical data. Conclusions Using PRS to first stratify patients genetically and then train machine-learning models with clinical predictors led to large improvements in lithium response prediction. When used with other PRS and biological markers in the future this approach may help inform which patients are most likely to respond to lithium treatment.

AB - Background Response to lithium in patients with bipolar disorder is associated with clinical and transdiagnostic genetic factors. The predictive combination of these variables might help clinicians better predict which patients will respond to lithium treatment. Aims To use a combination of transdiagnostic genetic and clinical factors to predict lithium response in patients with bipolar disorder. Method This study utilised genetic and clinical data (n = 1034) collected as part of the International Consortium on Lithium Genetics (ConLi+Gen) project. Polygenic risk scores (PRS) were computed for schizophrenia and major depressive disorder, and then combined with clinical variables using a cross-validated machine-learning regression approach. Unimodal, multimodal and genetically stratified models were trained and validated using ridge, elastic net and random forest regression on 692 patients with bipolar disorder from ten study sites using leave-site-out cross-validation. All models were then tested on an independent test set of 342 patients. The best performing models were then tested in a classification framework. Results The best performing linear model explained 5.1% (P = 0.0001) of variance in lithium response and was composed of clinical variables, PRS variables and interaction terms between them. The best performing non-linear model used only clinical variables and explained 8.1% (P = 0.0001) of variance in lithium response. A priori genomic stratification improved non-linear model performance to 13.7% (P = 0.0001) and improved the binary classification of lithium response. This model stratified patients based on their meta-polygenic loadings for major depressive disorder and schizophrenia and was then trained using clinical data. Conclusions Using PRS to first stratify patients genetically and then train machine-learning models with clinical predictors led to large improvements in lithium response prediction. When used with other PRS and biological markers in the future this approach may help inform which patients are most likely to respond to lithium treatment.

KW - Mood stabilisers

KW - bipolar affective disorders

KW - depressive disorders

KW - genetics

KW - outcome studies

UR - http://www.scopus.com/inward/record.url?scp=85126276148&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85126276148&partnerID=8YFLogxK

U2 - 10.1192/bjp.2022.28

DO - 10.1192/bjp.2022.28

M3 - Article

C2 - 35225756

AN - SCOPUS:85126276148

SN - 0007-1250

VL - 220

SP - 219

EP - 228

JO - British Journal of Psychiatry

JF - British Journal of Psychiatry

IS - 4

ER -

Using polygenic scores and clinical data for bipolar disorder patient stratification and lithium response prediction: machine learning approach

Abstract

Keywords

ASJC Scopus subject areas

Access to Document

Other files and links

Fingerprint

Cite this