Oops, you're using an old version of your browser so some of the features on this page may not be displaying properly.

MINIMAL Requirements: Google Chrome 24+Mozilla Firefox 20+Internet Explorer 11Opera 15–18Apple Safari 7SeaMonkey 2.15-2.23

ePoster Display

948P - Predicting the efficacy of lenvatinib plus anti-PD-1 antibodies in unresectable hepatocellular carcinoma (uHCC) using radiomics features of tumors extracted from baseline MRI

Date

16 Sep 2021

Session

ePoster Display

Topics

Staging and Imaging;  Targeted Therapy;  Immunotherapy

Tumour Site

Hepatobiliary Cancers

Presenters

Huichuan Sun

Citation

Annals of Oncology (2021) 32 (suppl_5): S818-S828. 10.1016/annonc/annonc677

Authors

H. Sun1, S. Rao2, B. Xu1, X. Zhu1, S. Dong2, X. Li1, C. Huang1, Y. Shen1, J. Zhu1, M. Li1, J. Liu1, M. Zeng2, J. Zhou1, J. Fan1

Author affiliations

  • 1 Department Of Liver Surgery And Transplantation, Zhongshan Hospital and Liver Cancer Institute, Fudan University, 200032 - Shanghai/CN
  • 2 Department Of Radiology, Zhongshan Hospital, Fudan University and Shanghai Institute of Medical Imaging, 200032 - Shanghai/CN

Resources

Login to get immediate access to this content.

If you do not have an ESMO account, please create one for free.

Abstract 948P

Background

Development of a method to predict the response of HCC to combination therapy with lenvatinib plus anti-PD-1 antibodies before treatment initiation would have great clinical benefit.

Methods

Consecutive patients with uHCC receiving first-line lenvatinib plus an anti-PD-1 antibody between Sep 2018 and Feb 2021, and who had at least one radiological response evaluation, were eligible for this study. Intrahepatic tumor response was assessed every 2 months (± 2 weeks) using modified RECIST; patients with a best intrahepatic tumor response of complete or partial response were defined as radiological responders and those with stable or progressive disease were defined as radiological non-responders. Radiomic features of intrahepatic tumors were extracted from the enhanced arterial and delayed phase of baseline MRI images. A Least Absolute Shrinkage and Selection Operator (LASSO) model was employed for feature selection. A Neural Network was used to develop the prediction model. The optimal cutoff value was determined using a receiver operating characteristic (ROC) curve by maximizing the Youden index.

Results

Of 96 eligible patients, 50% (n = 48) were radiological responders and 50% (n = 48) were non-responders. All patients were randomly divided into training (n = 72) and validation (n = 24) sets. A total of 2,420 radiomic features were extracted and normalized with min-max normalization. Features in the training set with intraclass correlation coefficients ≥ 0.80 were introduced into the LASSO model. Five features in the arterial phase and five in the delayed phase were identified as significant and used to build a neural network. The optimal cutoff value was 0.504. The area under the ROC curve, accuracy, specificity, and sensitivity for predicting objective response were 0.971 (P < 0.001), 97.2%, 97.2% and 97.2% in the training set, respectively; and were 0.778 (P = 0.010), 75.0%, 91.7% and 58.3% in the validation set, respectively.

Conclusions

Radiomics features from baseline MRI may serve as predictors for objective response to lenvatinib plus anti-PD-1 antibodies in uHCC patients before treatment initiation.

Clinical trial identification

NCT04639284.

Editorial acknowledgement

Legal entity responsible for the study

Zhongshan Hospital, Fudan University.

Funding

Clinical Research Special Fund of Zhongshan Hospital, Fudan University.

Disclosure

All authors have declared no conflicts of interest.

This site uses cookies. Some of these cookies are essential, while others help us improve your experience by providing insights into how the site is being used.

For more detailed information on the cookies we use, please check our Privacy Policy.

Customise settings
  • Necessary cookies enable core functionality. The website cannot function properly without these cookies, and you can only disable them by changing your browser preferences.