Electronic Health Record (EHR) and Genomics-Based Machine Learning (ML) to Predict Therapeutic Effectiveness among Patients with Hormone-Receptor Positive (hr)+/her2advanced Breast Cancer (Abc)

Deep K. Hathi,Steven Pisano,Hossein Honarvar,Ravi Bharat Parikh,Rahul K. Das

JOURNAL OF CLINICAL ONCOLOGY（2023）

Concert Pharmaceuticals (United States) | University of Pennsylvania

Cited 0|Views22

Abstract

e13574 Background: Patients (pts) with HR+/HER2- aBC may eventually become resistant to endocrine therapies and CDK4/6 inhibitors (CDK4/6i). ML algorithms on EHR linked with NGS data may enable more accurate predictions of therapeutic resistance and identify clinicogenomic (CG) risk factors. Methods: HR+/HER2- aBC pts treated between Jan 2015 – Dec 2022 with at least one NGS test report before line end were identified in ConcertAI’s Genome360 oncology database, consisting of molecular alterations abstracted from NGS reports linked with deeply curated US-based EHR. XGBoost (XGB) and Cox proportional hazards (CoxPH) models were trained to predict real-world progression-free survival (rwPFS) derived from curated tumor progression (TP), as assessed by the providers based on clinical observations, pathology results, and radiologic evidence. Data was split into 60:20:20 for training/validation/test. Pts in the 1st and 4th quartiles of predicted risk from XGB model were defined as low- and high-risk. Predictors were identified from Shapley Additive exPlanations (SHAP)-derived marginal hazard ratios (HR), derived in the 180 days prior to index date, and included demographics, labs, vitals, comorbidities, concomitant medications, procedures, tumor properties, and molecular alterations as features. Results: Among 624 pts (CDK4/6i treated = 519) with median age = 62.5 years and 10.5% black, 46.8% developed TP. The test cumulative dynamic ROC-AUC in the best-performing XGB model was 0.68. The precision at 180 days (%TP = 27) was 40%. The 180-day cumulative incidence of TP in low- and high-risk groups was 11.6% vs 37.7%. Higher line number, lower hemoglobin (HGB), higher alkaline phosphatase (ALP), presence of liver metastasis, and history of thoracic radiation were clinical risk factors. Genomic risk factors were FGF aberrations, alterations in TP53 and in genes within the MAPK, cell cycle, and ESR pathways, and co-alterations in the PIK3CA+TP53, and MAPK+ESR pathways. ALP-by-TP53 interaction was found. ALP level was a risk factor in TP53-wildtype (wt) but not in TP53-mutant (mut) pts. Conclusions: ML on EHR linked with NGS data enables identification of high-risk pts and multimodal predictors of CDK4/6i resistance among women with HR+/HER2- aBC, identifying both known and undescribed interactions between clinical and genomic risk factors as well as co-alteration risk factors. [Table: see text]

Translated text

求助PDF

上传PDF

Bibtex

AI Read Science

AI Summary

AI Summary is the key point extracted automatically understanding the full text of the paper, including the background, methods, results, conclusions, icons and other key content, so that you can get the outline of the paper at a glance.

Example

Background

Key content

Introduction

Methods

Results

Related work

Fund

Key content

Pretraining has recently greatly promoted the development of natural language processing (NLP)
We show that M6 outperforms the baselines in multimodal downstream tasks, and the large M6 with 10 parameters can reach a better performance
We propose a method called M6 that is able to process information of multiple modalities and perform both single-modal and cross-modal understanding and generation
The model is scaled to large model with 10 billion parameters with sophisticated deployment, and the 10 -parameter M6-large is the largest pretrained model in Chinese
Experimental results show that our proposed M6 outperforms the baseline in a number of downstream tasks concerning both single modality and multiple modalities We will continue the pretraining of extremely large models by increasing data to explore the limit of its performance

Upload PDF to Generate Summary

Must-Reading Tree

Example

Generate MRT to find the research sequence of this paper

Data Disclaimer

The page data are from open Internet sources, cooperative publishers and automatic analysis results through AI technology. We do not make any commitments and guarantees for the validity, accuracy, correctness, reliability, completeness and timeliness of the page data. If you have any questions, please contact us by email: report@aminer.cn

Chat Paper

【要点】：该研究使用电子健康记录（EHR）和基因组学数据，通过机器学习（ML）预测激素受体阳性（HR+/HER2-）晚期乳腺癌（abc）患者对治疗的反应效果，发现了新的临床和基因组风险因素及其相互作用。

【方法】：研究利用ConcertAI的Genome360肿瘤学数据库中的EHR和下一代测序（NGS）数据，采用XGBoost（XGB）和Cox比例风险（CoxPH）模型预测患者的无进展生存期（rwPFS）。

【实验】：数据被分为训练集、验证集和测试集（60:20:20）。通过Shapley Additive exPlanations（SHAP）分析确定风险预测因子。在624名患者中（其中519名接受CDK4/6抑制剂治疗），中位年龄为62.5岁，10.5%为黑人，46.8%的患者出现了肿瘤进展（TP）。最佳性能的XGB模型的测试累积动态ROC-AUC为0.68，180天的精确度为40%。低风险和高风险组的180天TP累积发生率分别为11.6%和37.7%。