The aim of this study was to develop and validate a machine learning algorithm to predict response of individual HER2-amplified colorectal cancer liver metastases (lmCRC) undergoing dual HER2-targeted therapy. Twenty-four radiomics features were extracted following 3D manual segmentation of 141 lmCRC on pre-treatment portal CT scans of a cohort including 38 HER2-amplified patients; feature selection was then performed using genetic algorithms. lmCRC were classified as non-responders (R-), if their largest diameter increased more than 10% at a CT scan performed following 3 months of treatment, responders (R+) otherwise. Sensitivity, specificity, negative (NPV) and positive (PPV) predictive values in correctly classifying individual lesion and overall patient response were assessed on a training dataset and then validated on a second dataset using a Gaussian naïve bayesian classifier. Per-lesion sensitivity, specificity, NPV and PPV were 89%, 85%, 93%, 78% and 90%, 42%, 73%, 71% respectively in the testing and validation datasets. Per-patient sensitivity and specificity were 92% and 86%. Heterogeneous response was observed in 9 of 38 patients (24%). Five of the 9 patients were carriers of non-responder lesions correctly classified as such by our radiomics signature, including 4 of 7 harboring only one non-responder lesion. The developed method has been proven effective in predicting behavior of individual metastases to targeted treatment in a cohort of HER2 amplified patients. The model accurately detects responder lesions and identifies non-responder lesions in patients with heterogeneous response, potentially paving the way to multimodal treatment in selected patients. Further validation will be needed to confirm our findings. This article is protected by copyright. All rights reserved.

Radiomics predicts response of individual HER2-amplified colorectal cancer liver metastases in patients treated with HER2-targeted therapy / V. Giannini, S. Rosati, A. Defeudis, G. Balestra, L. Vassallo, G. Cappello, S. Mazzetti, C. De Mattia, F. Rizzetto, A. Torresin, A. Sartore-Bianchi, S. Siena, A. Vanzulli, F. Leone, V. Zagonel, S. Marsoni, D. Regge. - In: INTERNATIONAL JOURNAL OF CANCER. - ISSN 0020-7136. - (2020). [Epub ahead of print] [10.1002/ijc.33271]

Radiomics predicts response of individual HER2-amplified colorectal cancer liver metastases in patients treated with HER2-targeted therapy

V. Giannini
Primo
;
C. De Mattia;F. Rizzetto;A. Torresin;A. Sartore-Bianchi;S. Siena;A. Vanzulli;
2020

Abstract

The aim of this study was to develop and validate a machine learning algorithm to predict response of individual HER2-amplified colorectal cancer liver metastases (lmCRC) undergoing dual HER2-targeted therapy. Twenty-four radiomics features were extracted following 3D manual segmentation of 141 lmCRC on pre-treatment portal CT scans of a cohort including 38 HER2-amplified patients; feature selection was then performed using genetic algorithms. lmCRC were classified as non-responders (R-), if their largest diameter increased more than 10% at a CT scan performed following 3 months of treatment, responders (R+) otherwise. Sensitivity, specificity, negative (NPV) and positive (PPV) predictive values in correctly classifying individual lesion and overall patient response were assessed on a training dataset and then validated on a second dataset using a Gaussian naïve bayesian classifier. Per-lesion sensitivity, specificity, NPV and PPV were 89%, 85%, 93%, 78% and 90%, 42%, 73%, 71% respectively in the testing and validation datasets. Per-patient sensitivity and specificity were 92% and 86%. Heterogeneous response was observed in 9 of 38 patients (24%). Five of the 9 patients were carriers of non-responder lesions correctly classified as such by our radiomics signature, including 4 of 7 harboring only one non-responder lesion. The developed method has been proven effective in predicting behavior of individual metastases to targeted treatment in a cohort of HER2 amplified patients. The model accurately detects responder lesions and identifies non-responder lesions in patients with heterogeneous response, potentially paving the way to multimodal treatment in selected patients. Further validation will be needed to confirm our findings. This article is protected by copyright. All rights reserved.
CT liver metastases; genetic algorithms; machine learning; prediction of response to therapy; radiomics;
Settore MED/06 - Oncologia Medica
2020
1-set-2020
Article (author)
File in questo prodotto:
File Dimensione Formato  
2020 Giannini V et al, Int J Cancer.pdf

Open Access dal 02/09/2021

Tipologia: Post-print, accepted manuscript ecc. (versione accettata dall'editore)
Dimensione 2.4 MB
Formato Adobe PDF
2.4 MB Adobe PDF Visualizza/Apri
Pubblicazioni consigliate

I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.

Utilizza questo identificativo per citare o creare un link a questo documento: https://hdl.handle.net/2434/761836
Citazioni
  • ???jsp.display-item.citation.pmc??? 16
  • Scopus 24
  • ???jsp.display-item.citation.isi??? 20
social impact