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

451P - Utility of circulating free DNA 5’-end motif profile in the prediction of pathological response after neoadjuvant chemoradiotherapy in patients with locally advanced rectal cancer

Date

16 Sep 2021

Session

ePoster Display

Topics

Radiation Oncology

Tumour Site

Colon and Rectal Cancer

Presenters

Yaqi Wang

Citation

Annals of Oncology (2021) 32 (suppl_5): S530-S582. 10.1016/annonc/annonc698

Authors

Y. Wang1, X. Fan2, Y. Xu2, H. Bao2, F. Xia1, J. Wan1, L. Shen1, X. Wu2, Y. Shao2, X. Li3, Y. Xu3, S. Cai3, Z. Zhang1

Author affiliations

  • 1 Department Of Radiation Oncology, Fudan University Shanghai Cancer Center, 200032 - Shanghai/CN
  • 2 Department Of Research And Development, Nanjing Geneseeq Technology Inc., 210032 - Nanjing/CN
  • 3 Department Of Colorectal Surgery, Fudan University Shanghai Cancer Center, 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 451P

Background

For locally advanced rectal cancer (LARC) patients who achieved complete clinical response (cCR) after neoadjuvant chemoradiotherapy (nCRT), a “Watch & Wait” (W&W) approach is adopted to improve quality of life. However, W&W may increase the recurrence risk of those who are not pathological complete response (non-pCR). Cell-free DNA (cfDNA) end motifs have been used in cancer detection. But its utility in predicting pCR status has not been reported. In this study, we explored the potential of cfDNA end motif profile in the prediction of pathological response in LARC patients.

Methods

We recruited 119 LARC patients (cT3-4/N0-2, M0) who received nCRT plus total mesorectal excision (TME) from February, 2016 to October, 2017. Plasma samples at baseline, during and after nCRT were collected and applied to deep targeted-panel sequencing covering 422 genes. Frequencies of 4096 6-mer 5'-end motifs during and after nCRT were calculated and used as input to construct an elastic-net logistic regression model to predict pCR status. Its predictive performance was evaluated by internal 5-fold cross validation and 30-time repeats.

Results

Among 119 patients, 103 patients completed all sample collection and were included. The out-of-sample validation AUC of the motif model for predicting pCR status was 0.95±0.02 (mean ± standard deviation). With Youden index as cut off, the sensitivity and specificity for predicting non-pCR were 0.95±0.05 and 0.85±0.04, and accuracy was 0.93±0.03. The AUC of MRI-based tumor regression grade (mTRG) was 0.72 with a maximum accuracy of 0.74. Particularly, the end motif based classifier maintained good predictive ability for patients whose circulating tumor DNA (ctDNA) was undetectable during and after nCRT (n=74). The out-of-sample AUC of the model was 0.94±0.02 and sensitivity and specificity for predicting non-pCR were 0.89±0.07 and 0.88±0.06, and the accuracy was 0.89±0.03.

Conclusions

cfDNA 5'-end motif profile has a potential to predict pathological response after nCRT. The model based on the 5'-end motif profile shows superior predictive performance compared to MRI, therefore, may improve the selection of patients for W&W approach.

Clinical trial identification

Editorial acknowledgement

Legal entity responsible for the study

Fudan University Shanghai Cancer Center.

Funding

National Natural Science Foundation of China (No. 81773357).

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.