1096 - OncotypeDX® predictive nomogram for recurrence score output: a machine learning system based on quantitative immunochemistry analysis - ADAPTED01

Background

OncotypeDX (ODX®) can enhance prediction of breast cancer recurrence, guiding adjuvant treatment options. However, the opportunity to access this test is not always possible. The aim of this study is to investigate the correlation between phenotypical tumor characteristics, quantitative classical immunohistochemistry (IHC) and recurrent score (RS) resulting from ODX®.

Methods

All breast cancer patients who underwent ODX® between 2014 and 2018 were retrospectively included in the study. The data selected for analysis were age, menopausal status, pathological and IHC features. IHC was performed with standardized quantitative methods. Dataset was split into two subsets (70% for training and 30% for internal validation). Logistic models were built with statistically significant features for predicting RS ≤ 25 or ≤ 20. An external validation set, provided by another center, was used to test reliability of prediction models.

Results

The internal dataset included 407 patients (Table) who underwent ODX®. Mean age was 53.7 (31-80) and 222 patients (54.55%) were > 50 years old. ODX® results showed: 67 patients (16.6%) between 0-10, 272 patients between 11-25 (66.8%) and 68 pts > 26 (16.6%). At the logistic regression analysis, RS score was significantly associated with ER (p = 0.004), PgR (p < 0.001), and Ki67% (p < 0.001) with the threshold equal to 25. Above patients with RS ≤ 25, the generalized linear regression resulted in a well calibrated model with an AUC of 92.2% (sensitivity 84.2%; specificity 80.1%). External validation set included 180 patients and confirmed the model performance with an AUC of 82.3% and good calibration. A nomogram predicting RS score ≤25 was generated.Table:

261P Tumor characteristics training set + internal test set

Conclusions

Quantitative IHC presents a good correlation with RS score in patients with RS ≤ 25, also in external validation set. A nomogram for physician that enhances a cost/effectiveness clinical approach practice has been developed. Prospective clinical application will be tested in further studies.

Clinical trial identification

Editorial acknowledgement

Legal entity responsible for the study

Fabio Marazzi.

Funding

Has not received any funding.

Disclosure

All authors have declared no conflicts of interest.