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

1134P - Personalized medicine in advanced breast cancer: AI-driven genomic test for CDK4/6 treatment response prediction

Date

16 Sep 2021

Session

ePoster Display

Topics

Cytotoxic Therapy;  Targeted Therapy;  Survivorship;  Supportive and Palliative Care

Tumour Site

Breast Cancer

Presenters

Paweł Zawadzki

Citation

Annals of Oncology (2021) 32 (suppl_5): S921-S930. 10.1016/annonc/annonc707

Authors

P. Zawadzki1, A. Woźna1, P. Sztromwasser2, W. Majer-Burman2, R. Stępień2, K. Pacewicz2, D. Sielski2, M. Gniot2, M. Wojtaszewska2, M. Dąbrowski2, M. Piernik3

Author affiliations

  • 1 Physics/biology Department, Adam Mickiewicz University - Faculty of Physics, 61-614 - Poznań/PL
  • 2 R&d, MNM Diagnostics, 61-614 - Poznań/PL
  • 3 Faculty Of Computing And Telecommunication, Poznań University of Technology, 60-965 - Poznań/PL

Resources

Login to get immediate access to this content.

If you do not have an ESMO account, please create one for free.

Abstract 1134P

Background

Breast cancer (BC) is the world's most commonly diagnosed cancer type among women. Recent clinical evidence suggests that CDK4/6 inhibitors (CDK4/6i) improve the outcomes across different subtypes of breast cancer. Based on these findings, the U.S. Food and Drug Administration (FDA) in February 2015 granted accelerated approval for palbociclib, ribociclib, and eventually abemaciclib for treatment of estrogen receptor-positive (ER+) human epidermal growth factor receptor-2-negative (HER2-) BC. Despite many studies supporting the use of CDK4/6i in ER+/HER2- BC, administration of these drugs is still limited, mainly due to relatively high occurrence of side effects and the current price of the therapy in comparison to endocrine monotherapy treatment. Currently, despite ongoing research, the field of cancer research still lacks reliable clinical biomarkers of response (i.e., predictive biomarkers) for CDK4/6-targeted therapeutics.

Methods

In our study we have applied approaches, such as machine learning algorithms bundled with molecular biology and genomics, enabling patient stratification and prediction of therapy responses for CDK4/6i treatment options. We have performed a retrospective analysis of CDK4/6i response, using our proprietary genomic database, consisting of over 1,000 genomes derived from BC patients, 171 BC patients of which were treated with CDK4/6i (i.e., palbociclib and ribociclib).

Results

The Random Forest-based AI algorithm achieved good performance on nine genomic features that were selected as the most biologically relevant. Hold-out validation, a performance metric of the CDK4/6iDx classifier, provided a discriminative power of 83.88% by area under receiver operating characteristic (AUROC) curve analysis. Consistently, the low rate of misclassifications (high precision and recall of 90% and 88%, respectively) demonstrates the high discriminating power of the final classifier algorithm.

Conclusions

These results show that the CDK4/6i genomic signature, consisting of 9 biomarkers, improves the predictive stratification of ER+/HER2- metastatic BC patients, increasing the chance of identifying patients that will experience clinical benefit from CDK4/6i therapy.

Clinical trial identification

Editorial acknowledgement

Legal entity responsible for the study

The authors.

Funding

MNM Diagnostics.

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.