Abstract 147P
Background
Correctly classifying early invasive breast cancer (eiBC) cases into high-risk and low-risk is considered a key issue in breast cancer treatment. This classification is notably crucial in determining the treatment regimen: chemo-endocrine therapy versus, endocrine therapy alone. Last year, we presented RACE, an AI-based tool for assessing the risk of distant relapse at 5 years of ER+HER2- eiBC patients from HES (hematoxylin-eosin-safran) whole slide images (WSI). In the present study, we performed a one-shot blind validation of RACE on an independent cohort of 676 HES WSI.
Methods
HES WSI of 676 ER+/HER2- eiBC diagnosed at Gustave Roussy from 2012 to 2017 included in the CANTO cohort, constituted the validation dataset (19 patients relapsed at 5 years). We compared RACE performance to the two most relevant clinical scores: Predict Breast and CTS0. To assess performance, we proceeded as follows. The scores were first compared in terms of both cumulative sensitivity and dynamic specificity at 5 years to assess the accuracy of the scores to identify relapses. For this purpose, each score has been binarized (low risk/high risk) with respect to a threshold that has been set beforehand.
Results
The cumulative sensitivity (resp. dynamic specificity) at 5 years is 64% (resp. 78%) for Race, 61% (resp. 77%) for CTS0 and 43% (resp. 80%) for Predict Breast. Further analyses showed that among the low risk population treated with endocrine therapy alone, the distant relapse rate was 0.3% (1 out of 324).
Conclusions
We performed a first fully blind validation of Race, an AI-based tool for assessing the risk of distant relapse. First, the obtained results showed the ability of RACE to generalize on independent data and thus endorse the soundness of the method. Furthermore, additional analysis brings to light the clinical value of Race and that it could be used for therapeutic de-escalation purposes. This validation will be extended to multi-site and multi-scanner eiBC WSI from the CANTO cohort under completion.
Clinical trial identification
Editorial acknowledgement
Legal entity responsible for the study
Gustave Roussy.
Funding
Owkin.
Disclosure
V. Gaury, V. Aubert, C. Saillard, K. Elgui: Financial Interests, Personal, Stocks/Shares: Owkin. F. André: Financial Interests, Personal, Advisory Role: Owkin. All other authors have declared no conflicts of interest.