Multicentre validation of CT grey-level co-occurrence matrix features for overall survival in primary oesophageal adenocarcinoma

Robert O’Shea; Samuel J. Withey; Kasia Owczarczyk; Christopher Rookyard; James Gossage; Edmund Godfrey; Craig Jobling; Simon L. Parsons; Richard J.E. Skipworth; Vicky Goh; Tom D.L. Crosby; Freddie Bartlett; Russell D. Petty; Helen Coleman; Damian McManus; Richard Turkington; Anna Grabowska; Krishna Moorthy; Christopher J. Peters; George B. Hanna; Sharmila Sothi; Michael Scott; Rehan Haidry; Laurence Lovat; John Saunders; Philip Kaye; Irshad Soomro; Loveena Sreedharan; Bhaskar Kumar; Ed Cheong; David Chan; Grant Sanders; Francesca D. Ciccarelli; Ula Mahadeva; Fuju Chang; Andrew Davies; Jesper Lagergren; Ben L. Grace; Timothy J. Underwood; Gianmarco Contino; Sonia Puig; Philippe Taniere; Andrew Beggs; Olga Tucker; J. Robert O’Neill; Vicki Save; Izhar Bagwan; Shaun R. Preston; Andrew Sharrocks; Yeng Ang; on behalf of the OCCAMS Consortium

doi:10.1007/s00330-024-10666-y

Multicentre validation of CT grey-level co-occurrence matrix features for overall survival in primary oesophageal adenocarcinoma

Robert O’Shea, Samuel J. Withey, Kasia Owczarczyk, Christopher Rookyard, James Gossage, Edmund Godfrey, Craig Jobling, Simon L. Parsons, Richard J.E. Skipworth, Vicky Goh^*, Tom D.L. Crosby, Freddie Bartlett, Russell D. Petty, Helen Coleman, Damian McManus, Richard Turkington, Anna Grabowska, Krishna Moorthy, Christopher J. Peters, George B. HannaSharmila Sothi, Michael Scott, Rehan Haidry, Laurence Lovat, John Saunders, Philip Kaye, Irshad Soomro, Loveena Sreedharan, Bhaskar Kumar, Ed Cheong, David Chan, Grant Sanders, Francesca D. Ciccarelli, Ula Mahadeva, Fuju Chang, Andrew Davies, Jesper Lagergren, Ben L. Grace, Timothy J. Underwood, Gianmarco Contino, Sonia Puig, Philippe Taniere, Andrew Beggs, Olga Tucker, J. Robert O’Neill, Vicki Save, Izhar Bagwan, Shaun R. Preston, Andrew Sharrocks, Yeng Ang, on behalf of the OCCAMS Consortium

^*Corresponding author for this work

School of Medicine, Dentistry and Biomedical Sciences

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

5 Downloads (Pure)

Abstract

Background: Personalising management of primary oesophageal adenocarcinoma requires better risk stratification. Lack of independent validation of proposed imaging biomarkers has hampered clinical translation. We aimed to prospectively validate previously identified prognostic grey-level co-occurrence matrix (GLCM) CT features for 3-year overall survival.

Methods: Following ethical approval, clinical and contrast-enhanced CT data were acquired from participants from five institutions. Data from three institutions were used for training and two for testing. Survival classifiers were modelled on prespecified variables (‘Clinical’ model: age, clinical T-stage, clinical N-stage; ‘ClinVol’ model: clinical features + CT tumour volume; ‘ClinRad’ model: ClinVol features + GLCM_Correlation and GLCM_Contrast). To reflect current clinical practice, baseline stage was also modelled as a univariate predictor (‘Stage’). Discrimination was assessed by area under the receiver operating curve (AUC) analysis; calibration by Brier scores; and clinical relevance by thresholding risk scores to achieve 90% sensitivity for 3-year mortality.

Results: A total of 162 participants were included (144 male; median 67 years [IQR 59, 72]; training, 95 participants; testing, 67 participants). Median survival was 998 days [IQR 486, 1594]. The ClinRad model yielded the greatest test discrimination (AUC, 0.68 [95% CI 0.54, 0.81]) that outperformed Stage (ΔAUC, 0.12 [95% CI 0.01, 0.23]; p =.04). The Clinical and ClinVol models yielded comparable test discrimination (AUC, 0.66 [95% CI 0.51, 0.80] vs. 0.65 [95% CI 0.50, 0.79]; p >.05). Test sensitivity of 90% was achieved by ClinRad and Stage models only.

Conclusions: Compared to Stage, multivariable models of prespecified clinical and radiomic variables yielded improved prediction of 3-year overall survival. Clinical relevance statement: Previously identified radiomic features are prognostic but may not substantially improve risk stratification on their own.

Key Points: • Better risk stratification is needed in primary oesophageal cancer to personalise management. • Previously identified CT features—GLCM_Correlation and GLCM_Contrast—contain incremental prognostic information to age and clinical stage. • Compared to staging, multivariable clinicoradiomic models improve discrimination of 3-year overall survival.

Original language	English
Pages (from-to)	6919-6928
Number of pages	10
Journal	European Radiology
Volume	34
Issue number	10
Early online date	25 Mar 2024
DOIs	https://doi.org/10.1007/s00330-024-10666-y
Publication status	Published - Oct 2024

Keywords

Adenocarcinoma
Oesophageal neoplasms
Precision medicine
Prognosis
Radiomics

ASJC Scopus subject areas

Radiology Nuclear Medicine and imaging

UN SDGs

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

Access to Document

10.1007/s00330-024-10666-yLicence: CC BY

Multicentre validation of CT grey-level co-occurrence matrix features for overall survival in primary oesophageal adenocarcinoma
Copyright 2024 the authors. This is an open access article published under a Creative Commons Attribution License (https://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution and reproduction in any medium, provided the author and source are cited.
Final published version, 2.73 MBLicence: CC BY

Cite this

O’Shea, R., Withey, S. J., Owczarczyk, K., Rookyard, C., Gossage, J., Godfrey, E., Jobling, C., Parsons, S. L., Skipworth, R. J. E., Goh, V., Crosby, T. D. L., Bartlett, F., Petty, R. D., Coleman, H., McManus, D., Turkington, R., Grabowska, A., Moorthy, K., Peters, C. J., ... on behalf of the OCCAMS Consortium (2024). Multicentre validation of CT grey-level co-occurrence matrix features for overall survival in primary oesophageal adenocarcinoma. European Radiology, 34(10), 6919-6928. https://doi.org/10.1007/s00330-024-10666-y

@article{87f26abb5b6c4db8b396d2b8ce2b1368,

title = "Multicentre validation of CT grey-level co-occurrence matrix features for overall survival in primary oesophageal adenocarcinoma",

abstract = "Background: Personalising management of primary oesophageal adenocarcinoma requires better risk stratification. Lack of independent validation of proposed imaging biomarkers has hampered clinical translation. We aimed to prospectively validate previously identified prognostic grey-level co-occurrence matrix (GLCM) CT features for 3-year overall survival. Methods: Following ethical approval, clinical and contrast-enhanced CT data were acquired from participants from five institutions. Data from three institutions were used for training and two for testing. Survival classifiers were modelled on prespecified variables ({\textquoteleft}Clinical{\textquoteright} model: age, clinical T-stage, clinical N-stage; {\textquoteleft}ClinVol{\textquoteright} model: clinical features + CT tumour volume; {\textquoteleft}ClinRad{\textquoteright} model: ClinVol features + GLCM_Correlation and GLCM_Contrast). To reflect current clinical practice, baseline stage was also modelled as a univariate predictor ({\textquoteleft}Stage{\textquoteright}). Discrimination was assessed by area under the receiver operating curve (AUC) analysis; calibration by Brier scores; and clinical relevance by thresholding risk scores to achieve 90% sensitivity for 3-year mortality. Results: A total of 162 participants were included (144 male; median 67 years [IQR 59, 72]; training, 95 participants; testing, 67 participants). Median survival was 998 days [IQR 486, 1594]. The ClinRad model yielded the greatest test discrimination (AUC, 0.68 [95% CI 0.54, 0.81]) that outperformed Stage (ΔAUC, 0.12 [95% CI 0.01, 0.23]; p =.04). The Clinical and ClinVol models yielded comparable test discrimination (AUC, 0.66 [95% CI 0.51, 0.80] vs. 0.65 [95% CI 0.50, 0.79]; p >.05). Test sensitivity of 90% was achieved by ClinRad and Stage models only. Conclusions: Compared to Stage, multivariable models of prespecified clinical and radiomic variables yielded improved prediction of 3-year overall survival. Clinical relevance statement: Previously identified radiomic features are prognostic but may not substantially improve risk stratification on their own. Key Points: • Better risk stratification is needed in primary oesophageal cancer to personalise management. • Previously identified CT features—GLCM_Correlation and GLCM_Contrast—contain incremental prognostic information to age and clinical stage. • Compared to staging, multivariable clinicoradiomic models improve discrimination of 3-year overall survival.",

keywords = "Adenocarcinoma, Oesophageal neoplasms, Precision medicine, Prognosis, Radiomics",

author = "Robert O{\textquoteright}Shea and Withey, {Samuel J.} and Kasia Owczarczyk and Christopher Rookyard and James Gossage and Edmund Godfrey and Craig Jobling and Parsons, {Simon L.} and Skipworth, {Richard J.E.} and Vicky Goh and Crosby, {Tom D.L.} and Freddie Bartlett and Petty, {Russell D.} and Helen Coleman and Damian McManus and Richard Turkington and Anna Grabowska and Krishna Moorthy and Peters, {Christopher J.} and Hanna, {George B.} and Sharmila Sothi and Michael Scott and Rehan Haidry and Laurence Lovat and John Saunders and Philip Kaye and Irshad Soomro and Loveena Sreedharan and Bhaskar Kumar and Ed Cheong and David Chan and Grant Sanders and Ciccarelli, {Francesca D.} and Ula Mahadeva and Fuju Chang and Andrew Davies and Jesper Lagergren and Grace, {Ben L.} and Underwood, {Timothy J.} and Gianmarco Contino and Sonia Puig and Philippe Taniere and Andrew Beggs and Olga Tucker and O{\textquoteright}Neill, {J. Robert} and Vicki Save and Izhar Bagwan and Preston, {Shaun R.} and Andrew Sharrocks and Yeng Ang and {on behalf of the OCCAMS Consortium}",

year = "2024",

month = oct,

doi = "10.1007/s00330-024-10666-y",

language = "English",

volume = "34",

pages = "6919--6928",

journal = "European Radiology",

issn = "0938-7994",

publisher = "Springer Verlag",

number = "10",

}

O’Shea, R, Withey, SJ, Owczarczyk, K, Rookyard, C, Gossage, J, Godfrey, E, Jobling, C, Parsons, SL, Skipworth, RJE, Goh, V, Crosby, TDL, Bartlett, F, Petty, RD, Coleman, H , McManus, D , Turkington, R, Grabowska, A, Moorthy, K, Peters, CJ, Hanna, GB, Sothi, S, Scott, M, Haidry, R, Lovat, L, Saunders, J, Kaye, P, Soomro, I, Sreedharan, L, Kumar, B, Cheong, E, Chan, D, Sanders, G, Ciccarelli, FD, Mahadeva, U, Chang, F, Davies, A, Lagergren, J, Grace, BL, Underwood, TJ, Contino, G, Puig, S, Taniere, P, Beggs, A, Tucker, O, O’Neill, JR, Save, V, Bagwan, I, Preston, SR, Sharrocks, A, Ang, Y & on behalf of the OCCAMS Consortium 2024, 'Multicentre validation of CT grey-level co-occurrence matrix features for overall survival in primary oesophageal adenocarcinoma', European Radiology, vol. 34, no. 10, pp. 6919-6928. https://doi.org/10.1007/s00330-024-10666-y

TY - JOUR

T1 - Multicentre validation of CT grey-level co-occurrence matrix features for overall survival in primary oesophageal adenocarcinoma

AU - O’Shea, Robert

AU - Withey, Samuel J.

AU - Owczarczyk, Kasia

AU - Rookyard, Christopher

AU - Gossage, James

AU - Godfrey, Edmund

AU - Jobling, Craig

AU - Parsons, Simon L.

AU - Skipworth, Richard J.E.

AU - Goh, Vicky

AU - Crosby, Tom D.L.

AU - Bartlett, Freddie

AU - Petty, Russell D.

AU - Coleman, Helen

AU - McManus, Damian

AU - Turkington, Richard

AU - Grabowska, Anna

AU - Moorthy, Krishna

AU - Peters, Christopher J.

AU - Hanna, George B.

AU - Sothi, Sharmila

AU - Scott, Michael

AU - Haidry, Rehan

AU - Lovat, Laurence

AU - Saunders, John

AU - Kaye, Philip

AU - Soomro, Irshad

AU - Sreedharan, Loveena

AU - Kumar, Bhaskar

AU - Cheong, Ed

AU - Chan, David

AU - Sanders, Grant

AU - Ciccarelli, Francesca D.

AU - Mahadeva, Ula

AU - Chang, Fuju

AU - Davies, Andrew

AU - Lagergren, Jesper

AU - Grace, Ben L.

AU - Underwood, Timothy J.

AU - Contino, Gianmarco

AU - Puig, Sonia

AU - Taniere, Philippe

AU - Beggs, Andrew

AU - Tucker, Olga

AU - O’Neill, J. Robert

AU - Save, Vicki

AU - Bagwan, Izhar

AU - Preston, Shaun R.

AU - Sharrocks, Andrew

AU - Ang, Yeng

AU - on behalf of the OCCAMS Consortium

PY - 2024/10

Y1 - 2024/10

N2 - Background: Personalising management of primary oesophageal adenocarcinoma requires better risk stratification. Lack of independent validation of proposed imaging biomarkers has hampered clinical translation. We aimed to prospectively validate previously identified prognostic grey-level co-occurrence matrix (GLCM) CT features for 3-year overall survival. Methods: Following ethical approval, clinical and contrast-enhanced CT data were acquired from participants from five institutions. Data from three institutions were used for training and two for testing. Survival classifiers were modelled on prespecified variables (‘Clinical’ model: age, clinical T-stage, clinical N-stage; ‘ClinVol’ model: clinical features + CT tumour volume; ‘ClinRad’ model: ClinVol features + GLCM_Correlation and GLCM_Contrast). To reflect current clinical practice, baseline stage was also modelled as a univariate predictor (‘Stage’). Discrimination was assessed by area under the receiver operating curve (AUC) analysis; calibration by Brier scores; and clinical relevance by thresholding risk scores to achieve 90% sensitivity for 3-year mortality. Results: A total of 162 participants were included (144 male; median 67 years [IQR 59, 72]; training, 95 participants; testing, 67 participants). Median survival was 998 days [IQR 486, 1594]. The ClinRad model yielded the greatest test discrimination (AUC, 0.68 [95% CI 0.54, 0.81]) that outperformed Stage (ΔAUC, 0.12 [95% CI 0.01, 0.23]; p =.04). The Clinical and ClinVol models yielded comparable test discrimination (AUC, 0.66 [95% CI 0.51, 0.80] vs. 0.65 [95% CI 0.50, 0.79]; p >.05). Test sensitivity of 90% was achieved by ClinRad and Stage models only. Conclusions: Compared to Stage, multivariable models of prespecified clinical and radiomic variables yielded improved prediction of 3-year overall survival. Clinical relevance statement: Previously identified radiomic features are prognostic but may not substantially improve risk stratification on their own. Key Points: • Better risk stratification is needed in primary oesophageal cancer to personalise management. • Previously identified CT features—GLCM_Correlation and GLCM_Contrast—contain incremental prognostic information to age and clinical stage. • Compared to staging, multivariable clinicoradiomic models improve discrimination of 3-year overall survival.

AB - Background: Personalising management of primary oesophageal adenocarcinoma requires better risk stratification. Lack of independent validation of proposed imaging biomarkers has hampered clinical translation. We aimed to prospectively validate previously identified prognostic grey-level co-occurrence matrix (GLCM) CT features for 3-year overall survival. Methods: Following ethical approval, clinical and contrast-enhanced CT data were acquired from participants from five institutions. Data from three institutions were used for training and two for testing. Survival classifiers were modelled on prespecified variables (‘Clinical’ model: age, clinical T-stage, clinical N-stage; ‘ClinVol’ model: clinical features + CT tumour volume; ‘ClinRad’ model: ClinVol features + GLCM_Correlation and GLCM_Contrast). To reflect current clinical practice, baseline stage was also modelled as a univariate predictor (‘Stage’). Discrimination was assessed by area under the receiver operating curve (AUC) analysis; calibration by Brier scores; and clinical relevance by thresholding risk scores to achieve 90% sensitivity for 3-year mortality. Results: A total of 162 participants were included (144 male; median 67 years [IQR 59, 72]; training, 95 participants; testing, 67 participants). Median survival was 998 days [IQR 486, 1594]. The ClinRad model yielded the greatest test discrimination (AUC, 0.68 [95% CI 0.54, 0.81]) that outperformed Stage (ΔAUC, 0.12 [95% CI 0.01, 0.23]; p =.04). The Clinical and ClinVol models yielded comparable test discrimination (AUC, 0.66 [95% CI 0.51, 0.80] vs. 0.65 [95% CI 0.50, 0.79]; p >.05). Test sensitivity of 90% was achieved by ClinRad and Stage models only. Conclusions: Compared to Stage, multivariable models of prespecified clinical and radiomic variables yielded improved prediction of 3-year overall survival. Clinical relevance statement: Previously identified radiomic features are prognostic but may not substantially improve risk stratification on their own. Key Points: • Better risk stratification is needed in primary oesophageal cancer to personalise management. • Previously identified CT features—GLCM_Correlation and GLCM_Contrast—contain incremental prognostic information to age and clinical stage. • Compared to staging, multivariable clinicoradiomic models improve discrimination of 3-year overall survival.

KW - Adenocarcinoma

KW - Oesophageal neoplasms

KW - Precision medicine

KW - Prognosis

KW - Radiomics

U2 - 10.1007/s00330-024-10666-y

DO - 10.1007/s00330-024-10666-y

M3 - Article

C2 - 38526750

AN - SCOPUS:85188602084

SN - 0938-7994

VL - 34

SP - 6919

EP - 6928

JO - European Radiology

JF - European Radiology

IS - 10

ER -

Multicentre validation of CT grey-level co-occurrence matrix features for overall survival in primary oesophageal adenocarcinoma

Abstract

Keywords

ASJC Scopus subject areas

UN SDGs

Access to Document

Fingerprint

Cite this