Abstract

Background

Neoadjuvant chemotherapy (NAC) is gaining attention as a treatment for advanced colorectal cancer owing to its potential to improve surgical outcomes and prognosis. However, reliable biomarkers to predict the response to NAC are lacking. We aimed to investigate the predictive value of cell-free DNA (cfDNA) integrity index for NAC response, using machine learning to compensate for the small cohort size.

Methods

This retrospective study included 31 locally advanced colorectal cancer patients who underwent NAC and surgery at the Nippon Medical School Hospital between 2016 and 2020. Blood samples were collected pre-post-NAC to assess cfDNA levels using quantitative polymerase chain reaction. The cfDNA integrity index was calculated based on the ratio of long to short fragments in the long-interspersed element-1 repeat sequence. Statistical analyses, including random forest modeling, were performed to evaluate the predictive value of the cfDNA integrity index for treatment response.

Results

Of the 31 patients, 19 (61.3%) were classified as responders and 12 (38.7%) as non-responders. The post-NAC cfDNA integrity index was significantly different between the groups (P = 0.002, odds ratio = 16.0). Random forest analysis identified changes in the cfDNA integrity index as the most important predictor of NAC response (%IncMSE: 15.79; IncNodePurity: 2.21), while sex, age, tumor site, and pre-NAC cfDNA levels were not significant predictors.

Conclusions

Variability in the cfDNA integrity index shows promise as a biomarker for predicting NAC efficacy in colorectal cancer.

Details

Title
Evaluating cell-free DNA integrity index as a non-invasive biomarker for neoadjuvant chemotherapy in colorectal cancer patients
Author
Iwai, Takuma; Yamada, Takeshi; Uehara, Kay; Shinji, Seiichi; Matsuda, Akihisa; Yokoyama, Yasuyuki; Takahashi, Goro; Miyasaka, Toshimitsu; Kanaka, Shintaro; Koizumi, Michihiro; Yoshida, Hiroshi
Pages
1-9
Section
Research
Publication year
2025
Publication date
2025
Publisher
Springer Nature B.V.
e-ISSN
14712407
Source type
Scholarly Journal
Language of publication
English
ProQuest document ID
3236996997
Copyright
© 2025. This work is licensed under http://creativecommons.org/licenses/by-nc-nd/4.0/ (the “License”). Notwithstanding the ProQuest Terms and Conditions, you may use this content in accordance with the terms of the License.