2020medRxivOpen access

An iterative supervised learning method identifying two subgroups of FOLFOX resistance patterns and predicting FOLFOX response in colorectal cancer patients

Sun Tian, Fulong Wang, Shi‐Xun Lu, Rujia Wu, Gong Chen

Open full text 0 citations

Abstract

Abstract Background FOLFOX is a combination of drugs that is widely used to treat colorectal cancer. The response rate of FOLFOX in colorectal cancer(CRC) is 30-50%. We develop a method that analyzes mechanisms of FOLFOX resistance and predicts whether a patient will benefit from FOLFOX. Methods Gene expression data of 83 stage IV CRC tumor samples (FOLFOX responder n=42, non-responder n=41) were used to develop a supervised learning method IML and analyze subgroups of FOLFOX resistance mechanism. Datasets of 32 FOLFOX treated stage IV CRC patients and 55 FOLFOX treated stage III CRC patients were used as independent validations. Results An iterative supervised learning (IML) method identified two distinct subgroups of CRC patients who resist FOLFOX. Each subgroup relies on different types of DNA damage repair proteins and they are mutually exclusive. Protein-protein networks showed the main mechanism might be the synergistic effect of resisting apoptosis and an altered cell cycle. IML method was validated in two independent validation sets, one FOLFOX treated stage IV CRC patients(HR=2.6, p-value=0.02, 3-years survival rate of the predicted responder group 61.9%, predicted nonresponder group 18.8%) and one FOLFOX treated stage III CRC patients (estimated HR=2.36, p-value=0.02). A subgroup of mesenchymal subtype patients shows the pattern as FOLFOX responders. Conclusions IML method reflects the underlying biology of FOLFOX resistance and predicts FOLFOX response.

Open-access reader

About this research paper

What this paper is about

Abstract Background FOLFOX is a combination of drugs that is widely used to treat colorectal cancer. The response rate of FOLFOX in colorectal cancer(CRC) is 30-50%. We develop a method that analyzes mechanisms of FOLFOX resistance and predicts whether a patient will benefit from FOLFOX. Methods Gene expression data of 83 stage IV CRC tumor samples (FOLFOX responder n=42, non-responder n=41) were used to develop a supervised learning method IML and analyze subgroups of FOLFOX resistance mechanism. Datasets of 32 FOLFOX treated stage IV CRC patients and 55 FOLFOX treated stage III CRC patients were used as independent validations. Results An iterative supervised learning (IML) method identified two distinct subgroups of CRC patients who resist FOLFOX. Each subgroup relies on different types of DNA damage repair proteins and they are mutually exclusive. Protein-protein networks showed the main mechanism might be the synergistic effect of resisting apoptosis and an altered cell cycle. IML method was validated in two independent validation sets, one FOLFOX treated stage IV CRC patients(HR=2.6, p-value=0.02, 3-years survival rate of the predicted responder group 61.9%, predicted nonresponder group 18.8%) and one FOLFOX treated stage III CRC patients (estimated HR=2.36, p-value=0.02). A subgroup of mesenchymal subtype patients shows the pattern as FOLFOX responders. Conclusions IML method reflects the underlying biology of FOLFOX resistance and predicts FOLFOX response.

Why it matters

A significance statement is not available in the OpenAlex record.

Key contribution

A contribution statement is not available in the OpenAlex record.

Method / approach

Method details are not available in the OpenAlex metadata.

Main findings

Findings are not separately available in the OpenAlex metadata.

Limitations

Limitations are not available in the OpenAlex metadata.

Applications

Application details are not available in the OpenAlex metadata.

Available abstract

Abstract Background FOLFOX is a combination of drugs that is widely used to treat colorectal cancer. The response rate of FOLFOX in colorectal cancer(CRC) is 30-50%. We develop a method that analyzes mechanisms of FOLFOX resistance and predicts whether a patient will benefit from FOLFOX. Methods Gene expression data of 83 stage IV CRC tumor samples (FOLFOX responder n=42, non-responder n=41) were used to develop a supervised learning method IML and analyze subgroups of FOLFOX resistance mechanism. Datasets of 32 FOLFOX treated stage IV CRC patients and 55 FOLFOX treated stage III CRC patients were used as independent validations. Results An iterative supervised learning (IML) method identified two distinct subgroups of CRC patients who resist FOLFOX. Each subgroup relies on different types of DNA damage repair proteins and they are mutually exclusive. Protein-protein networks showed the main mechanism might be the synergistic effect of resisting apoptosis and an altered cell cycle. IML method was validated in two independent validation sets, one FOLFOX treated stage IV CRC patients(HR=2.6, p-value=0.02, 3-years survival rate of the predicted responder group 61.9%, predicted nonresponder group 18.8%) and one FOLFOX treated stage III CRC patients (estimated HR=2.36, p-value=0.02). A subgroup of mesenchymal subtype patients shows the pattern as FOLFOX responders. Conclusions IML method reflects the underlying biology of FOLFOX resistance and predicts FOLFOX response.

Key concepts: FOLFOX, Colorectal cancer, Oncology, Internal medicine, Medicine, Cancer, Oxaliplatin

Related papers

Back to paper searchBrowse research topicsOriginal source
An iterative supervised learning method identifying two subgroups of FOLFOX resistance patterns and predicting FOLFOX response in colorectal cancer patients — Research Paper | ScholarLens