A Deep Learning Approach Validates Genetic Risk Factors for Late Toxicity After Prostate Cancer Radiotherapy in a REQUITE Multi-National Cohort

Michela Carlotta Massi; Francesca Gasperoni; Francesca Ieva; Anna Maria Paganoni; Paolo Zunino; Andrea Manzoni; Nicola Rares Franco; Liv Veldeman; Piet Ost; Valerie Fonteyne; Christopher J. Talbot; Tim Rattay; Adam Webb; Paul R. Symonds; Kerstie Johnson; Maarten Lambrecht; Karin Haustermans; Gert De Meerleer; Dirk de Ruysscher; Ben Vanneste; Evert Van Limbergen; Ananya Choudhury; Rebecca M. Elliott; Elena Sperk; Carsten Herskind; Marlon R. Veldwijk; Barbara Avuzzi; Tommaso Giandini; Riccardo Valdagni; Alessandro Cicchetti; David Azria; Marie-Pierre Farcy Jacquet; Barry S. Rosenstein; Richard G. Stock; Kayla Collado; Ana Vega; Miguel Elias Aguado-Barrera; Patricia Calvo; Alison M. Dunning; Laura Fachal; Sarah L. Kerns; Debbie Payne; Jenny Chang-Claude; Petra Seibold; Catharine M. L. West; Tiziana Rancati; REQUITE consortium

doi:10.3389/fonc.2020.541281

A Deep Learning Approach Validates Genetic Risk Factors for Late Toxicity After Prostate Cancer Radiotherapy in a REQUITE Multi-National Cohort

Michela Carlotta Massi, Francesca Gasperoni, Francesca Ieva^*, Anna Maria Paganoni, Paolo Zunino, Andrea Manzoni, Nicola Rares Franco, Liv Veldeman, Piet Ost, Valerie Fonteyne, Christopher J. Talbot, Tim Rattay, Adam Webb, Paul R. Symonds, Kerstie Johnson, Maarten Lambrecht, Karin Haustermans, Gert De Meerleer, Dirk de Ruysscher, Ben VannesteEvert Van Limbergen, Ananya Choudhury, Rebecca M. Elliott, Elena Sperk, Carsten Herskind, Marlon R. Veldwijk, Barbara Avuzzi, Tommaso Giandini, Riccardo Valdagni, Alessandro Cicchetti, David Azria, Marie-Pierre Farcy Jacquet, Barry S. Rosenstein, Richard G. Stock, Kayla Collado, Ana Vega, Miguel Elias Aguado-Barrera, Patricia Calvo, Alison M. Dunning, Laura Fachal, Sarah L. Kerns, Debbie Payne, Jenny Chang-Claude, Petra Seibold, Catharine M. L. West, Tiziana Rancati, REQUITE consortium

^*Corresponding author for this work

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

5 Citations (Web of Science)

Abstract

Background: REQUITE (validating pREdictive models and biomarkers of radiotherapy toxicity to reduce side effects and improve QUalITy of lifE in cancer survivors) is an international prospective cohort study. The purpose of this project was to analyse a cohort of patients recruited into REQUITE using a deep learning algorithm to identify patient-specific features associated with the development of toxicity, and test the approach by attempting to validate previously published genetic risk factors.

Methods: The study involved REQUITE prostate cancer patients treated with external beam radiotherapy who had complete 2-year follow-up. We used five separate late toxicity endpoints: >= grade 1 late rectal bleeding, >= grade 2 urinary frequency, >= grade 1 haematuria, >= grade 2 nocturia, >= grade 1 decreased urinary stream. Forty-three single nucleotide polymorphisms (SNPs) already reported in the literature to be associated with the toxicity endpoints were included in the analysis. No SNP had been studied before in the REQUITE cohort. Deep Sparse AutoEncoders (DSAE) were trained to recognize features (SNPs) identifying patients with no toxicity and tested on a different independent mixed population including patients without and with toxicity.

Results: One thousand, four hundred and one patients were included, and toxicity rates were: rectal bleeding 11.7%, urinary frequency 4%, haematuria 5.5%, nocturia 7.8%, decreased urinary stream 17.1%. Twenty-four of the 43 SNPs that were associated with the toxicity endpoints were validated as identifying patients with toxicity. Twenty of the 24 SNPs were associated with the same toxicity endpoint as reported in the literature: 9 SNPs for urinary symptoms and 11 SNPs for overall toxicity. The other 4 SNPs were associated with a different endpoint.

Conclusion: Deep learning algorithms can validate SNPs associated with toxicity after radiotherapy for prostate cancer. The method should be studied further to identify polygenic SNP risk signatures for radiotherapy toxicity. The signatures could then be included in integrated normal tissue complication probability models and tested for their ability to personalize radiotherapy treatment planning.

Original language	English
Article number	541281
Number of pages	15
Journal	Frontiers in Oncology
Volume	10
DOIs	https://doi.org/10.3389/fonc.2020.541281
Publication status	Published - 15 Oct 2020

Keywords

prostate cancer
late toxicity
snps
deep learning
autoencoder
validation
GENOME-WIDE ASSOCIATION
QUALITY-OF-LIFE
RADIATION-THERAPY
RADIOGENOMICS
METAANALYSIS
CONSORTIUM
BIOMARKERS
SELECTION
VARIANTS
DESIGN

Access to Document

10.3389/fonc.2020.541281Licence: CC BY

Cite this

Massi, M. C., Gasperoni, F., Ieva, F., Paganoni, A. M., Zunino, P., Manzoni, A., Franco, N. R., Veldeman, L., Ost, P., Fonteyne, V., Talbot, C. J., Rattay, T., Webb, A., Symonds, P. R., Johnson, K., Lambrecht, M., Haustermans, K., De Meerleer, G., de Ruysscher, D., ... REQUITE consortium (2020). A Deep Learning Approach Validates Genetic Risk Factors for Late Toxicity After Prostate Cancer Radiotherapy in a REQUITE Multi-National Cohort. Frontiers in Oncology, 10, [541281]. https://doi.org/10.3389/fonc.2020.541281

@article{7402ba83fcd142b28317abe7143f6f0d,

title = "A Deep Learning Approach Validates Genetic Risk Factors for Late Toxicity After Prostate Cancer Radiotherapy in a REQUITE Multi-National Cohort",

abstract = "Background: REQUITE (validating pREdictive models and biomarkers of radiotherapy toxicity to reduce side effects and improve QUalITy of lifE in cancer survivors) is an international prospective cohort study. The purpose of this project was to analyse a cohort of patients recruited into REQUITE using a deep learning algorithm to identify patient-specific features associated with the development of toxicity, and test the approach by attempting to validate previously published genetic risk factors.Methods: The study involved REQUITE prostate cancer patients treated with external beam radiotherapy who had complete 2-year follow-up. We used five separate late toxicity endpoints: >= grade 1 late rectal bleeding, >= grade 2 urinary frequency, >= grade 1 haematuria, >= grade 2 nocturia, >= grade 1 decreased urinary stream. Forty-three single nucleotide polymorphisms (SNPs) already reported in the literature to be associated with the toxicity endpoints were included in the analysis. No SNP had been studied before in the REQUITE cohort. Deep Sparse AutoEncoders (DSAE) were trained to recognize features (SNPs) identifying patients with no toxicity and tested on a different independent mixed population including patients without and with toxicity.Results: One thousand, four hundred and one patients were included, and toxicity rates were: rectal bleeding 11.7%, urinary frequency 4%, haematuria 5.5%, nocturia 7.8%, decreased urinary stream 17.1%. Twenty-four of the 43 SNPs that were associated with the toxicity endpoints were validated as identifying patients with toxicity. Twenty of the 24 SNPs were associated with the same toxicity endpoint as reported in the literature: 9 SNPs for urinary symptoms and 11 SNPs for overall toxicity. The other 4 SNPs were associated with a different endpoint.Conclusion: Deep learning algorithms can validate SNPs associated with toxicity after radiotherapy for prostate cancer. The method should be studied further to identify polygenic SNP risk signatures for radiotherapy toxicity. The signatures could then be included in integrated normal tissue complication probability models and tested for their ability to personalize radiotherapy treatment planning.",

keywords = "prostate cancer, late toxicity, snps, deep learning, autoencoder, validation, GENOME-WIDE ASSOCIATION, QUALITY-OF-LIFE, RADIATION-THERAPY, RADIOGENOMICS, METAANALYSIS, CONSORTIUM, BIOMARKERS, SELECTION, VARIANTS, DESIGN",

author = "Massi, {Michela Carlotta} and Francesca Gasperoni and Francesca Ieva and Paganoni, {Anna Maria} and Paolo Zunino and Andrea Manzoni and Franco, {Nicola Rares} and Liv Veldeman and Piet Ost and Valerie Fonteyne and Talbot, {Christopher J.} and Tim Rattay and Adam Webb and Symonds, {Paul R.} and Kerstie Johnson and Maarten Lambrecht and Karin Haustermans and {De Meerleer}, Gert and {de Ruysscher}, Dirk and Ben Vanneste and {Van Limbergen}, Evert and Ananya Choudhury and Elliott, {Rebecca M.} and Elena Sperk and Carsten Herskind and Veldwijk, {Marlon R.} and Barbara Avuzzi and Tommaso Giandini and Riccardo Valdagni and Alessandro Cicchetti and David Azria and Jacquet, {Marie-Pierre Farcy} and Rosenstein, {Barry S.} and Stock, {Richard G.} and Kayla Collado and Ana Vega and Aguado-Barrera, {Miguel Elias} and Patricia Calvo and Dunning, {Alison M.} and Laura Fachal and Kerns, {Sarah L.} and Debbie Payne and Jenny Chang-Claude and Petra Seibold and West, {Catharine M. L.} and Tiziana Rancati and {REQUITE consortium}",

year = "2020",

month = oct,

day = "15",

doi = "10.3389/fonc.2020.541281",

language = "English",

volume = "10",

journal = "Frontiers in Oncology",

issn = "2234-943X",

publisher = "Frontiers Media S.A.",

}

Massi, MC, Gasperoni, F, Ieva, F, Paganoni, AM, Zunino, P, Manzoni, A, Franco, NR, Veldeman, L, Ost, P, Fonteyne, V, Talbot, CJ, Rattay, T, Webb, A, Symonds, PR, Johnson, K, Lambrecht, M, Haustermans, K, De Meerleer, G, de Ruysscher, D , Vanneste, B , Van Limbergen, E , Choudhury, A, Elliott, RM, Sperk, E, Herskind, C, Veldwijk, MR, Avuzzi, B, Giandini, T, Valdagni, R, Cicchetti, A, Azria, D, Jacquet, M-PF, Rosenstein, BS, Stock, RG, Collado, K, Vega, A, Aguado-Barrera, ME, Calvo, P, Dunning, AM, Fachal, L, Kerns, SL, Payne, D, Chang-Claude, J, Seibold, P, West, CML, Rancati, T & REQUITE consortium 2020, 'A Deep Learning Approach Validates Genetic Risk Factors for Late Toxicity After Prostate Cancer Radiotherapy in a REQUITE Multi-National Cohort', Frontiers in Oncology, vol. 10, 541281. https://doi.org/10.3389/fonc.2020.541281

TY - JOUR

T1 - A Deep Learning Approach Validates Genetic Risk Factors for Late Toxicity After Prostate Cancer Radiotherapy in a REQUITE Multi-National Cohort

AU - Massi, Michela Carlotta

AU - Gasperoni, Francesca

AU - Ieva, Francesca

AU - Paganoni, Anna Maria

AU - Zunino, Paolo

AU - Manzoni, Andrea

AU - Franco, Nicola Rares

AU - Veldeman, Liv

AU - Ost, Piet

AU - Fonteyne, Valerie

AU - Talbot, Christopher J.

AU - Rattay, Tim

AU - Webb, Adam

AU - Symonds, Paul R.

AU - Johnson, Kerstie

AU - Lambrecht, Maarten

AU - Haustermans, Karin

AU - De Meerleer, Gert

AU - de Ruysscher, Dirk

AU - Vanneste, Ben

AU - Van Limbergen, Evert

AU - Choudhury, Ananya

AU - Elliott, Rebecca M.

AU - Sperk, Elena

AU - Herskind, Carsten

AU - Veldwijk, Marlon R.

AU - Avuzzi, Barbara

AU - Giandini, Tommaso

AU - Valdagni, Riccardo

AU - Cicchetti, Alessandro

AU - Azria, David

AU - Jacquet, Marie-Pierre Farcy

AU - Rosenstein, Barry S.

AU - Stock, Richard G.

AU - Collado, Kayla

AU - Vega, Ana

AU - Aguado-Barrera, Miguel Elias

AU - Calvo, Patricia

AU - Dunning, Alison M.

AU - Fachal, Laura

AU - Kerns, Sarah L.

AU - Payne, Debbie

AU - Chang-Claude, Jenny

AU - Seibold, Petra

AU - West, Catharine M. L.

AU - Rancati, Tiziana

AU - REQUITE consortium

PY - 2020/10/15

Y1 - 2020/10/15

N2 - Background: REQUITE (validating pREdictive models and biomarkers of radiotherapy toxicity to reduce side effects and improve QUalITy of lifE in cancer survivors) is an international prospective cohort study. The purpose of this project was to analyse a cohort of patients recruited into REQUITE using a deep learning algorithm to identify patient-specific features associated with the development of toxicity, and test the approach by attempting to validate previously published genetic risk factors.Methods: The study involved REQUITE prostate cancer patients treated with external beam radiotherapy who had complete 2-year follow-up. We used five separate late toxicity endpoints: >= grade 1 late rectal bleeding, >= grade 2 urinary frequency, >= grade 1 haematuria, >= grade 2 nocturia, >= grade 1 decreased urinary stream. Forty-three single nucleotide polymorphisms (SNPs) already reported in the literature to be associated with the toxicity endpoints were included in the analysis. No SNP had been studied before in the REQUITE cohort. Deep Sparse AutoEncoders (DSAE) were trained to recognize features (SNPs) identifying patients with no toxicity and tested on a different independent mixed population including patients without and with toxicity.Results: One thousand, four hundred and one patients were included, and toxicity rates were: rectal bleeding 11.7%, urinary frequency 4%, haematuria 5.5%, nocturia 7.8%, decreased urinary stream 17.1%. Twenty-four of the 43 SNPs that were associated with the toxicity endpoints were validated as identifying patients with toxicity. Twenty of the 24 SNPs were associated with the same toxicity endpoint as reported in the literature: 9 SNPs for urinary symptoms and 11 SNPs for overall toxicity. The other 4 SNPs were associated with a different endpoint.Conclusion: Deep learning algorithms can validate SNPs associated with toxicity after radiotherapy for prostate cancer. The method should be studied further to identify polygenic SNP risk signatures for radiotherapy toxicity. The signatures could then be included in integrated normal tissue complication probability models and tested for their ability to personalize radiotherapy treatment planning.

AB - Background: REQUITE (validating pREdictive models and biomarkers of radiotherapy toxicity to reduce side effects and improve QUalITy of lifE in cancer survivors) is an international prospective cohort study. The purpose of this project was to analyse a cohort of patients recruited into REQUITE using a deep learning algorithm to identify patient-specific features associated with the development of toxicity, and test the approach by attempting to validate previously published genetic risk factors.Methods: The study involved REQUITE prostate cancer patients treated with external beam radiotherapy who had complete 2-year follow-up. We used five separate late toxicity endpoints: >= grade 1 late rectal bleeding, >= grade 2 urinary frequency, >= grade 1 haematuria, >= grade 2 nocturia, >= grade 1 decreased urinary stream. Forty-three single nucleotide polymorphisms (SNPs) already reported in the literature to be associated with the toxicity endpoints were included in the analysis. No SNP had been studied before in the REQUITE cohort. Deep Sparse AutoEncoders (DSAE) were trained to recognize features (SNPs) identifying patients with no toxicity and tested on a different independent mixed population including patients without and with toxicity.Results: One thousand, four hundred and one patients were included, and toxicity rates were: rectal bleeding 11.7%, urinary frequency 4%, haematuria 5.5%, nocturia 7.8%, decreased urinary stream 17.1%. Twenty-four of the 43 SNPs that were associated with the toxicity endpoints were validated as identifying patients with toxicity. Twenty of the 24 SNPs were associated with the same toxicity endpoint as reported in the literature: 9 SNPs for urinary symptoms and 11 SNPs for overall toxicity. The other 4 SNPs were associated with a different endpoint.Conclusion: Deep learning algorithms can validate SNPs associated with toxicity after radiotherapy for prostate cancer. The method should be studied further to identify polygenic SNP risk signatures for radiotherapy toxicity. The signatures could then be included in integrated normal tissue complication probability models and tested for their ability to personalize radiotherapy treatment planning.

KW - prostate cancer

KW - late toxicity

KW - snps

KW - deep learning

KW - autoencoder

KW - validation

KW - GENOME-WIDE ASSOCIATION

KW - QUALITY-OF-LIFE

KW - RADIATION-THERAPY

KW - RADIOGENOMICS

KW - METAANALYSIS

KW - CONSORTIUM

KW - BIOMARKERS

KW - SELECTION

KW - VARIANTS

KW - DESIGN

U2 - 10.3389/fonc.2020.541281

DO - 10.3389/fonc.2020.541281

M3 - Article

C2 - 33178576

VL - 10

JO - Frontiers in Oncology

JF - Frontiers in Oncology

SN - 2234-943X

M1 - 541281

ER -