Prediction of Response to Lenvatinib Monotherapy for Unresectable Hepatocellular Carcinoma by Machine Learning...

Area of Science:

• Radiomics
• Machine Learning
• Oncology

Background:

• Hepatocellular carcinoma (HCC) is a leading cause of cancer death.
• Lenvatinib monotherapy is a treatment option for unresectable HCC.
• Predicting treatment response is crucial for optimizing patient outcomes.

Purpose of the Study:

• To develop and validate machine learning (ML) radiomics models for predicting response to lenvatinib monotherapy in unresectable HCC.
• To identify radiomics-based subtypes associated with treatment outcomes.

Main Methods:

• Retrospective analysis of 109 patients with unresectable HCC treated with lenvatinib monotherapy.
• Extraction of radiomics features from contrast-enhanced CT images.
• Construction and validation of 10 ML radiomics models, including AutoGluon, using K-means clustering for subtype identification.

Main Results:

• Two radiomics-based subtypes were identified, with subtype 1 showing higher overall response rates (ORR) and longer progression-free survival (PFS).
• The AutoGluon ML model achieved high predictive performance (AUC = 0.97 internally, 0.93 externally).
• Responders demonstrated significantly better overall survival and PFS compared to non-responders.

Conclusions:

• Machine learning radiomics models show promise in predicting lenvatinib response for unresectable HCC.
• These models can aid in stratifying patients and guiding treatment decisions.
• Further validation in larger cohorts is warranted.