Identification and validation of a machine learning model of complete response to radiation in rectal cancer reveals immune infiltrate and TGFβ as key predictors

Enric Domingo; Sanjay Rathee; Andrew Blake; Leslie Samuel; Graeme Murray; David Sebag-Montefiore; Simon Gollins; Nicholas West; Rubina Begum; Susan Richman; Phil Quirke; Keara Redmond; Aikaterini Chatzipli; Alessandro Barberis; Sylvana Hassanieh; Umair Mahmood; Michael Youdell; Ultan McDermott; Viktor Koelzer; Simon Leedham; Ian Tomlinson; Philip Dunne; Francesca M Buffa; Timothy S Maughan; S:CORT consortium

doi:10.1016/j.ebiom.2024.105228

Identification and validation of a machine learning model of complete response to radiation in rectal cancer reveals immune infiltrate and TGFβ as key predictors

Enric Domingo, Sanjay Rathee, Andrew Blake, Leslie Samuel, Graeme Murray, David Sebag-Montefiore, Simon Gollins, Nicholas West, Rubina Begum, Susan Richman, Phil Quirke, Keara Redmond, Aikaterini Chatzipli, Alessandro Barberis, Sylvana Hassanieh, Umair Mahmood, Michael Youdell, Ultan McDermott, Viktor Koelzer, Simon LeedhamIan Tomlinson, Philip Dunne, Francesca M Buffa^* (Corresponding Author), Timothy S Maughan^* (Corresponding Author), S:CORT consortium

^*Corresponding author for this work

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

2 Citations (Scopus)

4 Downloads (Pure)

Abstract

BACKGROUND: It is uncertain which biological features underpin the response of rectal cancer (RC) to radiotherapy. No biomarker is currently in clinical use to select patients for treatment modifications.

METHODS: We identified two cohorts of patients (total N = 249) with RC treated with neoadjuvant radiotherapy (45Gy/25) plus fluoropyrimidine. This discovery set included 57 cases with pathological complete response (pCR) to chemoradiotherapy (23%). Pre-treatment cancer biopsies were assessed using transcriptome-wide mRNA expression and targeted DNA sequencing for copy number and driver mutations. Biological candidate and machine learning (ML) approaches were used to identify predictors of pCR to radiotherapy independent of tumour stage. Findings were assessed in 107 cases from an independent validation set (GSE87211).

FINDINGS: Three gene expression sets showed significant independent associations with pCR: Fibroblast-TGFβ Response Signature (F-TBRS) with radioresistance; and cytotoxic lymphocyte (CL) expression signature and consensus molecular subtype CMS1 with radiosensitivity. These associations were replicated in the validation cohort. In parallel, a gradient boosting machine model comprising the expression of 33 genes generated in the discovery cohort showed high performance in GSE87211 with 90% sensitivity, 86% specificity. Biological and ML signatures indicated similar mechanisms underlying radiation response, and showed better AUC and p-values than published transcriptomic signatures of radiation response in RC.

INTERPRETATION: RCs responding completely to chemoradiotherapy (CRT) have biological characteristics of immune response and absence of immune inhibitory TGFβ signalling. These tumours may be identified with a potential biomarker based on a 33 gene expression signature. This could help select patients likely to respond to treatment with a primary radiotherapy approach as for anal cancer. Conversely, those with predicted radioresistance may be candidates for clinical trials evaluating addition of immune-oncology agents and stromal TGFβ signalling inhibition.

FUNDING: The Stratification in Colorectal Cancer Consortium (S:CORT) was funded by the Medical Research Council and Cancer Research UK (MR/M016587/1).

Original language	English
Article number	105228
Number of pages	13
Journal	EBioMedicine
Volume	106
Early online date	16 Jul 2024
DOIs	https://doi.org/10.1016/j.ebiom.2024.105228
Publication status	Published - 1 Aug 2024

Bibliographical note

Acknowledgements
F.M.B., A.B. and S.R. received funding from CRUK grant 23969 and ERC Consolidator Grant 772970 to F.M.B. The ARISTOTLE trial was funded by Cancer Research UK (CRUK/08/032). V.H.K. gratefully acknowledges funding by the Swiss National Science Foundation (P2SKP3_168322/1 and P2SKP3_168322/2) and the Promedica Foundation (F-87701-41-01). N.P.W acknowledges payment to institution from Yorkshire Cancer Research and Cancer Research UK (CRUK). P.D. acknowledges funding by CRUKearly detection project grant (grant no. A29834). I.T and TSM acknowledge funding from CRUK and MRC. This research was funded in whole, or in part, by the UKRI [MR/M016587/1].

Patients and/or the public were involved in the design and conduct of this work through the S:CORT consortium.

Funding

F.M.B., A.B. and S.R. received funding from CRUK grant 23969 and ERC Consolidator Grant 772970 to F.M.B. The ARISTOTLE trial was funded by Cancer Research UK (CRUK/08/032). V.H.K. gratefully acknowledges funding by the Swiss National Science Foundation (P2SKP3_168322/1 and P2SKP3_168322/2) and the Promedica Foundation (F-87701-41-01). N.P.W acknowledges payment to institution from Yorkshire Cancer Research and Cancer Research UK (CRUK). P.D. acknowledges funding by CRUKearly detection project grant (grant no. A29834). I.T and TSM acknowledge funding from CRUK and MRC. This research was funded in whole, or in part, by the UKRI [MR/M016587/1].

Funders	Funder number
Cancer Research UK	23969, CRUK/08/032, A29834
European Research Council	772970
Swiss National Science Foundation	P2SKP3_168322/1 , P2SKP3_168322/2)
UK Research and Innovation	MR/M016587/1

Keywords

Rectal neoplasms
Radiotherapy
Precision medicine
Prediction
TGFβ
Immune response
Genes

UN SDGs

This output contributes to the following UN Sustainable Development Goals (SDGs)

Access to Document

10.1016/j.ebiom.2024.105228Licence: CC BY

Domingo_etal_EBM_Identification_And_Validation_VoR
© 2024 The Author(s). Published by Elsevier B.V. This is an open access article under the CC BY license. (http://creativecommons.org/licenses/by/4.0/).
Final published version, 3.45 MBLicence: CC BY

Data from Identification and validation of a machine learning model of complete response to radiation in rectal cancer reveals immune infiltrate and TGFβ as key predictors
Domingo, E. (Creator), Rathee, S. (Creator), Blake, A. (Creator), Samuel, L. (Creator), Murray, G. (Creator), Sebag-Montefiore, D. (Creator), Gollins, S. (Creator), West, N. P. (Creator), Begum, R. (Creator), Richman, S. (Creator), Quirke, P. (Creator), Redmond, K. L. (Creator), Chatzipli, A. (Creator), Barberis, A. (Creator), Hassanieh, S. (Creator), Mahmood, U. (Creator), Youdell, M. (Creator), McDermott, U. (Creator), Koelzer, V. H. (Creator), Leedham, S. (Creator) & Maughan, T. S. (Creator), University of Aberdeen, Jul 2024
DOI: 10.1016/j.ebiom.2024.105228, https://www.s-cort.org/sites/default/files/exports/scort_ws3_grampian_export_84m9fndk/ws3_grampian_expression_raw.zip and one more link, https://www.s-cort.org/sites/default/files/exports/sample_sheets/SCORT_samples_Domingo_EBioMedicine.csv (show fewer)
Dataset

Cite this

Domingo, E., Rathee, S., Blake, A., Samuel, L., Murray, G., Sebag-Montefiore, D., Gollins, S., West, N., Begum, R., Richman, S., Quirke, P., Redmond, K., Chatzipli, A., Barberis, A., Hassanieh, S., Mahmood, U., Youdell, M., McDermott, U., Koelzer, V., ... S:CORT consortium (2024). Identification and validation of a machine learning model of complete response to radiation in rectal cancer reveals immune infiltrate and TGFβ as key predictors. EBioMedicine, 106, Article 105228. https://doi.org/10.1016/j.ebiom.2024.105228

Domingo, E, Rathee, S, Blake, A, Samuel, L, Murray, G, Sebag-Montefiore, D, Gollins, S, West, N, Begum, R, Richman, S, Quirke, P, Redmond, K, Chatzipli, A, Barberis, A, Hassanieh, S, Mahmood, U, Youdell, M, McDermott, U, Koelzer, V, Leedham, S, Tomlinson, I, Dunne, P, Buffa, FM, Maughan, TS & S:CORT consortium 2024, 'Identification and validation of a machine learning model of complete response to radiation in rectal cancer reveals immune infiltrate and TGFβ as key predictors', EBioMedicine, vol. 106, 105228. https://doi.org/10.1016/j.ebiom.2024.105228

@article{11d07152ce244cf180fbd7950444250f,

title = "Identification and validation of a machine learning model of complete response to radiation in rectal cancer reveals immune infiltrate and TGFβ as key predictors",

abstract = "BACKGROUND: It is uncertain which biological features underpin the response of rectal cancer (RC) to radiotherapy. No biomarker is currently in clinical use to select patients for treatment modifications.METHODS: We identified two cohorts of patients (total N = 249) with RC treated with neoadjuvant radiotherapy (45Gy/25) plus fluoropyrimidine. This discovery set included 57 cases with pathological complete response (pCR) to chemoradiotherapy (23%). Pre-treatment cancer biopsies were assessed using transcriptome-wide mRNA expression and targeted DNA sequencing for copy number and driver mutations. Biological candidate and machine learning (ML) approaches were used to identify predictors of pCR to radiotherapy independent of tumour stage. Findings were assessed in 107 cases from an independent validation set (GSE87211).FINDINGS: Three gene expression sets showed significant independent associations with pCR: Fibroblast-TGFβ Response Signature (F-TBRS) with radioresistance; and cytotoxic lymphocyte (CL) expression signature and consensus molecular subtype CMS1 with radiosensitivity. These associations were replicated in the validation cohort. In parallel, a gradient boosting machine model comprising the expression of 33 genes generated in the discovery cohort showed high performance in GSE87211 with 90% sensitivity, 86% specificity. Biological and ML signatures indicated similar mechanisms underlying radiation response, and showed better AUC and p-values than published transcriptomic signatures of radiation response in RC.INTERPRETATION: RCs responding completely to chemoradiotherapy (CRT) have biological characteristics of immune response and absence of immune inhibitory TGFβ signalling. These tumours may be identified with a potential biomarker based on a 33 gene expression signature. This could help select patients likely to respond to treatment with a primary radiotherapy approach as for anal cancer. Conversely, those with predicted radioresistance may be candidates for clinical trials evaluating addition of immune-oncology agents and stromal TGFβ signalling inhibition.FUNDING: The Stratification in Colorectal Cancer Consortium (S:CORT) was funded by the Medical Research Council and Cancer Research UK (MR/M016587/1).",

keywords = "Rectal neoplasms, Radiotherapy, Precision medicine, Prediction, TGFβ, Immune response, Genes",

author = "Enric Domingo and Sanjay Rathee and Andrew Blake and Leslie Samuel and Graeme Murray and David Sebag-Montefiore and Simon Gollins and Nicholas West and Rubina Begum and Susan Richman and Phil Quirke and Keara Redmond and Aikaterini Chatzipli and Alessandro Barberis and Sylvana Hassanieh and Umair Mahmood and Michael Youdell and Ultan McDermott and Viktor Koelzer and Simon Leedham and Ian Tomlinson and Philip Dunne and Buffa, {Francesca M} and Maughan, {Timothy S} and {S:CORT consortium}",

note = "Acknowledgements F.M.B., A.B. and S.R. received funding from CRUK grant 23969 and ERC Consolidator Grant 772970 to F.M.B. The ARISTOTLE trial was funded by Cancer Research UK (CRUK/08/032). V.H.K. gratefully acknowledges funding by the Swiss National Science Foundation (P2SKP3_168322/1 and P2SKP3_168322/2) and the Promedica Foundation (F-87701-41-01). N.P.W acknowledges payment to institution from Yorkshire Cancer Research and Cancer Research UK (CRUK). P.D. acknowledges funding by CRUKearly detection project grant (grant no. A29834). I.T and TSM acknowledge funding from CRUK and MRC. This research was funded in whole, or in part, by the UKRI [MR/M016587/1]. Patients and/or the public were involved in the design and conduct of this work through the S:CORT consortium.",

year = "2024",

month = aug,

day = "1",

doi = "10.1016/j.ebiom.2024.105228",

language = "English",

volume = "106",

journal = "EBioMedicine",

issn = "2352-3964",

publisher = "Elsevier BV",

}

TY - JOUR

T1 - Identification and validation of a machine learning model of complete response to radiation in rectal cancer reveals immune infiltrate and TGFβ as key predictors

AU - Domingo, Enric

AU - Rathee, Sanjay

AU - Blake, Andrew

AU - Samuel, Leslie

AU - Murray, Graeme

AU - Sebag-Montefiore, David

AU - Gollins, Simon

AU - West, Nicholas

AU - Begum, Rubina

AU - Richman, Susan

AU - Quirke, Phil

AU - Redmond, Keara

AU - Chatzipli, Aikaterini

AU - Barberis, Alessandro

AU - Hassanieh, Sylvana

AU - Mahmood, Umair

AU - Youdell, Michael

AU - McDermott, Ultan

AU - Koelzer, Viktor

AU - Leedham, Simon

AU - Tomlinson, Ian

AU - Dunne, Philip

AU - Buffa, Francesca M

AU - Maughan, Timothy S

AU - S:CORT consortium

N1 - Acknowledgements F.M.B., A.B. and S.R. received funding from CRUK grant 23969 and ERC Consolidator Grant 772970 to F.M.B. The ARISTOTLE trial was funded by Cancer Research UK (CRUK/08/032). V.H.K. gratefully acknowledges funding by the Swiss National Science Foundation (P2SKP3_168322/1 and P2SKP3_168322/2) and the Promedica Foundation (F-87701-41-01). N.P.W acknowledges payment to institution from Yorkshire Cancer Research and Cancer Research UK (CRUK). P.D. acknowledges funding by CRUKearly detection project grant (grant no. A29834). I.T and TSM acknowledge funding from CRUK and MRC. This research was funded in whole, or in part, by the UKRI [MR/M016587/1]. Patients and/or the public were involved in the design and conduct of this work through the S:CORT consortium.

PY - 2024/8/1

Y1 - 2024/8/1

N2 - BACKGROUND: It is uncertain which biological features underpin the response of rectal cancer (RC) to radiotherapy. No biomarker is currently in clinical use to select patients for treatment modifications.METHODS: We identified two cohorts of patients (total N = 249) with RC treated with neoadjuvant radiotherapy (45Gy/25) plus fluoropyrimidine. This discovery set included 57 cases with pathological complete response (pCR) to chemoradiotherapy (23%). Pre-treatment cancer biopsies were assessed using transcriptome-wide mRNA expression and targeted DNA sequencing for copy number and driver mutations. Biological candidate and machine learning (ML) approaches were used to identify predictors of pCR to radiotherapy independent of tumour stage. Findings were assessed in 107 cases from an independent validation set (GSE87211).FINDINGS: Three gene expression sets showed significant independent associations with pCR: Fibroblast-TGFβ Response Signature (F-TBRS) with radioresistance; and cytotoxic lymphocyte (CL) expression signature and consensus molecular subtype CMS1 with radiosensitivity. These associations were replicated in the validation cohort. In parallel, a gradient boosting machine model comprising the expression of 33 genes generated in the discovery cohort showed high performance in GSE87211 with 90% sensitivity, 86% specificity. Biological and ML signatures indicated similar mechanisms underlying radiation response, and showed better AUC and p-values than published transcriptomic signatures of radiation response in RC.INTERPRETATION: RCs responding completely to chemoradiotherapy (CRT) have biological characteristics of immune response and absence of immune inhibitory TGFβ signalling. These tumours may be identified with a potential biomarker based on a 33 gene expression signature. This could help select patients likely to respond to treatment with a primary radiotherapy approach as for anal cancer. Conversely, those with predicted radioresistance may be candidates for clinical trials evaluating addition of immune-oncology agents and stromal TGFβ signalling inhibition.FUNDING: The Stratification in Colorectal Cancer Consortium (S:CORT) was funded by the Medical Research Council and Cancer Research UK (MR/M016587/1).

AB - BACKGROUND: It is uncertain which biological features underpin the response of rectal cancer (RC) to radiotherapy. No biomarker is currently in clinical use to select patients for treatment modifications.METHODS: We identified two cohorts of patients (total N = 249) with RC treated with neoadjuvant radiotherapy (45Gy/25) plus fluoropyrimidine. This discovery set included 57 cases with pathological complete response (pCR) to chemoradiotherapy (23%). Pre-treatment cancer biopsies were assessed using transcriptome-wide mRNA expression and targeted DNA sequencing for copy number and driver mutations. Biological candidate and machine learning (ML) approaches were used to identify predictors of pCR to radiotherapy independent of tumour stage. Findings were assessed in 107 cases from an independent validation set (GSE87211).FINDINGS: Three gene expression sets showed significant independent associations with pCR: Fibroblast-TGFβ Response Signature (F-TBRS) with radioresistance; and cytotoxic lymphocyte (CL) expression signature and consensus molecular subtype CMS1 with radiosensitivity. These associations were replicated in the validation cohort. In parallel, a gradient boosting machine model comprising the expression of 33 genes generated in the discovery cohort showed high performance in GSE87211 with 90% sensitivity, 86% specificity. Biological and ML signatures indicated similar mechanisms underlying radiation response, and showed better AUC and p-values than published transcriptomic signatures of radiation response in RC.INTERPRETATION: RCs responding completely to chemoradiotherapy (CRT) have biological characteristics of immune response and absence of immune inhibitory TGFβ signalling. These tumours may be identified with a potential biomarker based on a 33 gene expression signature. This could help select patients likely to respond to treatment with a primary radiotherapy approach as for anal cancer. Conversely, those with predicted radioresistance may be candidates for clinical trials evaluating addition of immune-oncology agents and stromal TGFβ signalling inhibition.FUNDING: The Stratification in Colorectal Cancer Consortium (S:CORT) was funded by the Medical Research Council and Cancer Research UK (MR/M016587/1).

KW - Rectal neoplasms

KW - Radiotherapy

KW - Precision medicine

KW - Prediction

KW - TGFβ

KW - Immune response

KW - Genes

U2 - 10.1016/j.ebiom.2024.105228

DO - 10.1016/j.ebiom.2024.105228

M3 - Article

C2 - 39013324

SN - 2352-3964

VL - 106

JO - EBioMedicine

JF - EBioMedicine

M1 - 105228

ER -

Identification and validation of a machine learning model of complete response to radiation in rectal cancer reveals immune infiltrate and TGFβ as key predictors

Abstract

Bibliographical note

Funding

Keywords

UN SDGs

Access to Document

Fingerprint

Datasets

Data from Identification and validation of a machine learning model of complete response to radiation in rectal cancer reveals immune infiltrate and TGFβ as key predictors

Cite this