This multi-modal deep radiomics model, using PET/CT, clinical and histopathological data, was able to identify patients with bevacizumab-sensitive unresectable colorectal cancer liver metastases, providing a favorable approach for precise patient treatment.

Accurately predicting tumor response to targeted therapies is essential for guiding personalized conversion therapy in patients with unresectable colorectal cancer liver metastases (CRLM). Currently, tumor response evaluation criteria are based on assessments made after at least 2-months treatment. Consequently, there is a compelling need to develop baseline tools that can be used to guide therapy selection. Herein, the investigators proposed a deep radiomics-based fusion model which demonstrates high accuracy in predicting the efficacy of bevacizumab in CRLM patients. Further, the investigators observed a significant and positive association between the predicted-responders and longer progression-free survival as well as longer overall survival in CRLM patients treated with bevacizumab. Moreover, the model exhibits high negative prediction value, indicating its potential to accurately identify individuals who are unresponsive to bevacizumab. Thus, our model provides a valuable baseline method for specifically identifying bevacizumab-sensitive CRLM patients, which is offering a clinically convenient approach to guide precise patient treatment.

Inclusion Criteria:

1. Age ≥ 18 years and ≤75 years;
2. Patients were histologically confirmed for colorectal adenocarcinoma with unresectable liver-limited or liver-dominant metastases
3. PET/CT at baseline were available
4. First line treated with FOLFOX+ bevacizumab.

Exclusion Criteria:

1. Resectable liver metastases;
2. Wide-type KRAS/NRAS;
3. No measurable liver metastasis;
4. No efficacy assessment;
5. No follow-up information.

In this multicenter cohort study, the investigators collected 307 patients with colorectal cancer liver metastases. The training cohort and negative validation cohort were derived from the BECOME study (NCT01972490), for whom baseline PET/CT images were available. The internal validation cohort was derived from consecutive metastastic colorectal cancer patients of the multi-disciplinary team (MDT) at Zhongshan Hospital (ZSH), share the same MDT, surgical team, and PET/CT imaging equipment with training cohort, from 01 January 2018 to 31 December 2018. The external validation cohort came from the MDT of Zhongshan Hospital - Xiamen and the First Hospital of Wenzhou Medical University, from 01 January 2020 to 31 December 2020