A Machine Learning Platform to Optimize the Translation of Personalized Network Models to the Clinic

Manuela Salvucci, Arman Rahman, Alexa J Resler, Girish M Udupi, Deborah A McNamara, Elaine W Kay, Pierre Laurent-Puig, Daniel B Longley, Patrick G Johnston, Mark Lawler, Richard Wilson, Manuel Salto-Tellez, Sandra Van Schaeybroeck, Mairin Rafferty, William M Gallagher, Markus Rehm, Jochen H M Prehn

Research output: Contribution to journalArticle

23 Downloads (Pure)

Abstract

PURPOSE: Dynamic network models predict clinical prognosis and inform therapeutic intervention by elucidating disease-driven aberrations at the systems level. However, the personalization of model predictions requires the profiling of multiple model inputs, which hampers clinical translation.

PATIENTS AND METHODS: We applied APOPTO-CELL, a prognostic model of apoptosis signaling, to showcase the establishment of computational platforms that require a reduced set of inputs. We designed two distinct and complementary pipelines: a probabilistic approach to exploit a consistent subpanel of inputs across the whole cohort (Ensemble) and a machine learning approach to identify a reduced protein set tailored for individual patients (Tree). Development was performed on a virtual cohort of 3,200,000 patients, with inputs estimated from clinically relevant protein profiles. Validation was carried out in an in-house stage III colorectal cancer cohort, with inputs profiled in surgical resections by reverse phase protein array (n = 120) and/or immunohistochemistry (n = 117).

RESULTS: Ensemble and Tree reproduced APOPTO-CELL predictions in the virtual patient cohort with 92% and 99% accuracy while decreasing the number of inputs to a consistent subset of three proteins (40% reduction) or a personalized subset of 2.7 proteins on average (46% reduction), respectively. Ensemble and Tree retained prognostic utility in the in-house colorectal cancer cohort. The association between the Ensemble accuracy and prognostic value (Spearman ρ = 0.43; P = .02) provided a rationale to optimize the input composition for specific clinical settings. Comparison between profiling by reverse phase protein array (gold standard) and immunohistochemistry (clinical routine) revealed that the latter is a suitable technology to quantify model inputs.

CONCLUSION: This study provides a generalizable framework to optimize the development of network-based prognostic assays and, ultimately, to facilitate their integration in the routine clinical workflow.

Original languageEnglish
Number of pages17
JournalJCO Clinical Cancer Informatics
DOIs
Publication statusPublished - 17 Apr 2019

Fingerprint

Protein Array Analysis
Colorectal Neoplasms
Proteins
Immunohistochemistry
Workflow
Apoptosis
Technology
Machine Learning
Therapeutics

Cite this

Salvucci, M., Rahman, A., Resler, A. J., Udupi, G. M., McNamara, D. A., Kay, E. W., ... Prehn, J. H. M. (2019). A Machine Learning Platform to Optimize the Translation of Personalized Network Models to the Clinic. JCO Clinical Cancer Informatics. https://doi.org/10.1200/CCI.18.00056
Salvucci, Manuela ; Rahman, Arman ; Resler, Alexa J ; Udupi, Girish M ; McNamara, Deborah A ; Kay, Elaine W ; Laurent-Puig, Pierre ; Longley, Daniel B ; Johnston, Patrick G ; Lawler, Mark ; Wilson, Richard ; Salto-Tellez, Manuel ; Van Schaeybroeck, Sandra ; Rafferty, Mairin ; Gallagher, William M ; Rehm, Markus ; Prehn, Jochen H M. / A Machine Learning Platform to Optimize the Translation of Personalized Network Models to the Clinic. In: JCO Clinical Cancer Informatics. 2019.
@article{a475aebb4df24e028d8647ca65e5b962,
title = "A Machine Learning Platform to Optimize the Translation of Personalized Network Models to the Clinic",
abstract = "PURPOSE: Dynamic network models predict clinical prognosis and inform therapeutic intervention by elucidating disease-driven aberrations at the systems level. However, the personalization of model predictions requires the profiling of multiple model inputs, which hampers clinical translation.PATIENTS AND METHODS: We applied APOPTO-CELL, a prognostic model of apoptosis signaling, to showcase the establishment of computational platforms that require a reduced set of inputs. We designed two distinct and complementary pipelines: a probabilistic approach to exploit a consistent subpanel of inputs across the whole cohort (Ensemble) and a machine learning approach to identify a reduced protein set tailored for individual patients (Tree). Development was performed on a virtual cohort of 3,200,000 patients, with inputs estimated from clinically relevant protein profiles. Validation was carried out in an in-house stage III colorectal cancer cohort, with inputs profiled in surgical resections by reverse phase protein array (n = 120) and/or immunohistochemistry (n = 117).RESULTS: Ensemble and Tree reproduced APOPTO-CELL predictions in the virtual patient cohort with 92{\%} and 99{\%} accuracy while decreasing the number of inputs to a consistent subset of three proteins (40{\%} reduction) or a personalized subset of 2.7 proteins on average (46{\%} reduction), respectively. Ensemble and Tree retained prognostic utility in the in-house colorectal cancer cohort. The association between the Ensemble accuracy and prognostic value (Spearman ρ = 0.43; P = .02) provided a rationale to optimize the input composition for specific clinical settings. Comparison between profiling by reverse phase protein array (gold standard) and immunohistochemistry (clinical routine) revealed that the latter is a suitable technology to quantify model inputs.CONCLUSION: This study provides a generalizable framework to optimize the development of network-based prognostic assays and, ultimately, to facilitate their integration in the routine clinical workflow.",
author = "Manuela Salvucci and Arman Rahman and Resler, {Alexa J} and Udupi, {Girish M} and McNamara, {Deborah A} and Kay, {Elaine W} and Pierre Laurent-Puig and Longley, {Daniel B} and Johnston, {Patrick G} and Mark Lawler and Richard Wilson and Manuel Salto-Tellez and {Van Schaeybroeck}, Sandra and Mairin Rafferty and Gallagher, {William M} and Markus Rehm and Prehn, {Jochen H M}",
year = "2019",
month = "4",
day = "17",
doi = "10.1200/CCI.18.00056",
language = "English",
journal = "JCO Clinical Cancer Informatics",
issn = "2473-4276",
publisher = "American Society of Clinical Oncology",

}

Salvucci, M, Rahman, A, Resler, AJ, Udupi, GM, McNamara, DA, Kay, EW, Laurent-Puig, P, Longley, DB, Johnston, PG, Lawler, M, Wilson, R, Salto-Tellez, M, Van Schaeybroeck, S, Rafferty, M, Gallagher, WM, Rehm, M & Prehn, JHM 2019, 'A Machine Learning Platform to Optimize the Translation of Personalized Network Models to the Clinic', JCO Clinical Cancer Informatics. https://doi.org/10.1200/CCI.18.00056

A Machine Learning Platform to Optimize the Translation of Personalized Network Models to the Clinic. / Salvucci, Manuela; Rahman, Arman; Resler, Alexa J; Udupi, Girish M; McNamara, Deborah A; Kay, Elaine W; Laurent-Puig, Pierre; Longley, Daniel B; Johnston, Patrick G; Lawler, Mark; Wilson, Richard; Salto-Tellez, Manuel; Van Schaeybroeck, Sandra; Rafferty, Mairin; Gallagher, William M; Rehm, Markus; Prehn, Jochen H M.

In: JCO Clinical Cancer Informatics, 17.04.2019.

Research output: Contribution to journalArticle

TY - JOUR

T1 - A Machine Learning Platform to Optimize the Translation of Personalized Network Models to the Clinic

AU - Salvucci, Manuela

AU - Rahman, Arman

AU - Resler, Alexa J

AU - Udupi, Girish M

AU - McNamara, Deborah A

AU - Kay, Elaine W

AU - Laurent-Puig, Pierre

AU - Longley, Daniel B

AU - Johnston, Patrick G

AU - Lawler, Mark

AU - Wilson, Richard

AU - Salto-Tellez, Manuel

AU - Van Schaeybroeck, Sandra

AU - Rafferty, Mairin

AU - Gallagher, William M

AU - Rehm, Markus

AU - Prehn, Jochen H M

PY - 2019/4/17

Y1 - 2019/4/17

N2 - PURPOSE: Dynamic network models predict clinical prognosis and inform therapeutic intervention by elucidating disease-driven aberrations at the systems level. However, the personalization of model predictions requires the profiling of multiple model inputs, which hampers clinical translation.PATIENTS AND METHODS: We applied APOPTO-CELL, a prognostic model of apoptosis signaling, to showcase the establishment of computational platforms that require a reduced set of inputs. We designed two distinct and complementary pipelines: a probabilistic approach to exploit a consistent subpanel of inputs across the whole cohort (Ensemble) and a machine learning approach to identify a reduced protein set tailored for individual patients (Tree). Development was performed on a virtual cohort of 3,200,000 patients, with inputs estimated from clinically relevant protein profiles. Validation was carried out in an in-house stage III colorectal cancer cohort, with inputs profiled in surgical resections by reverse phase protein array (n = 120) and/or immunohistochemistry (n = 117).RESULTS: Ensemble and Tree reproduced APOPTO-CELL predictions in the virtual patient cohort with 92% and 99% accuracy while decreasing the number of inputs to a consistent subset of three proteins (40% reduction) or a personalized subset of 2.7 proteins on average (46% reduction), respectively. Ensemble and Tree retained prognostic utility in the in-house colorectal cancer cohort. The association between the Ensemble accuracy and prognostic value (Spearman ρ = 0.43; P = .02) provided a rationale to optimize the input composition for specific clinical settings. Comparison between profiling by reverse phase protein array (gold standard) and immunohistochemistry (clinical routine) revealed that the latter is a suitable technology to quantify model inputs.CONCLUSION: This study provides a generalizable framework to optimize the development of network-based prognostic assays and, ultimately, to facilitate their integration in the routine clinical workflow.

AB - PURPOSE: Dynamic network models predict clinical prognosis and inform therapeutic intervention by elucidating disease-driven aberrations at the systems level. However, the personalization of model predictions requires the profiling of multiple model inputs, which hampers clinical translation.PATIENTS AND METHODS: We applied APOPTO-CELL, a prognostic model of apoptosis signaling, to showcase the establishment of computational platforms that require a reduced set of inputs. We designed two distinct and complementary pipelines: a probabilistic approach to exploit a consistent subpanel of inputs across the whole cohort (Ensemble) and a machine learning approach to identify a reduced protein set tailored for individual patients (Tree). Development was performed on a virtual cohort of 3,200,000 patients, with inputs estimated from clinically relevant protein profiles. Validation was carried out in an in-house stage III colorectal cancer cohort, with inputs profiled in surgical resections by reverse phase protein array (n = 120) and/or immunohistochemistry (n = 117).RESULTS: Ensemble and Tree reproduced APOPTO-CELL predictions in the virtual patient cohort with 92% and 99% accuracy while decreasing the number of inputs to a consistent subset of three proteins (40% reduction) or a personalized subset of 2.7 proteins on average (46% reduction), respectively. Ensemble and Tree retained prognostic utility in the in-house colorectal cancer cohort. The association between the Ensemble accuracy and prognostic value (Spearman ρ = 0.43; P = .02) provided a rationale to optimize the input composition for specific clinical settings. Comparison between profiling by reverse phase protein array (gold standard) and immunohistochemistry (clinical routine) revealed that the latter is a suitable technology to quantify model inputs.CONCLUSION: This study provides a generalizable framework to optimize the development of network-based prognostic assays and, ultimately, to facilitate their integration in the routine clinical workflow.

U2 - 10.1200/CCI.18.00056

DO - 10.1200/CCI.18.00056

M3 - Article

C2 - 30995124

JO - JCO Clinical Cancer Informatics

JF - JCO Clinical Cancer Informatics

SN - 2473-4276

ER -