# The Impact of Clinical and Demographic Factors on High-Risk Patient Classification Frequencies by the EndoPredict Test: A Review and Single-Site Study

## TL;DR

This study explores why different patient groups show varying high-risk classifications using the EndoPredict test for breast cancer, finding that tumor size and lymph node involvement are key factors.

## Contribution

The study identifies tumor size, lymph node involvement, and histological grade as primary factors influencing EndoPredict risk classification variability across patient cohorts.

## Key findings

- Tumor size and lymph node involvement are primary determinants of high-risk classification by EndoPredict.
- Higher histological grade and Ki-67 levels above 25% are significantly associated with high-risk status.
- Variability in risk distribution across studies is largely due to differences in tumor size, nodal involvement, and histological grade.

## Abstract

Gene expression tests such as EndoPredict are widely used to support treatment decisions in hormone-receptor-positive breast cancer. However, different studies often report varying proportions of patients classified as high- or low-risk, which may raise concerns when local results differ from published data. In this work, we combined a descriptive review of published studies with data from our own patient cohort to better understand the reasons behind these differences. We found that tumor size, lymph node involvement, histological grade, and tumor proliferation strongly influence risk classification, while several demographic and reproductive factors play a more limited role. Our results indicate that variability in risk distributions mainly reflects differences in patient populations and case mix across cohorts. These findings help contextualize the divergent risk frequencies observed in clinical practice and support confidence in the robustness of EndoPredict across diverse clinical settings.

Background/Objectives: EndoPredict is a second-generation prognostic assay for estrogen-receptor-positive, HER2-negative breast cancer that integrates molecular and clinical parameters for risk stratification. Multiple studies have reported its clinical utility, while differences in the proportion of patients classified as high- or low-risk have been observed across cohorts. This study aimed to characterize clinical, pathological, and demographic factors associated with these differences. Methods: We conducted a descriptive review of 17 published studies and analyzed a single-institution cohort of 140 patients. Associations between clinicopathological variables and high-risk classification were assessed, including tumor size, lymph node status, histological grade, Ki-67 expression, and reproductive and demographic factors. Differences in inclusion criteria and cohort characteristics were also examined. Results: Tumor size and lymph node involvement emerged as primary determinants of high-risk classification. A high histological grade and Ki-67 levels above 25% were significantly associated with high-risk status (p < 0.001). Conversely, age, age at menarche, menopausal status, Body Mass Index, progesterone receptor expression, molecular subtype, and histological type showed no significant association. A higher number of pregnancies correlated with a lower frequency of high-risk classification (p < 0.01). Heterogeneity in risk distribution across studies was largely attributable to differences in tumor size, nodal involvement, and histological grade. Additional variability was associated with inclusion criteria, sample selection, and regional demographic characteristics. Conclusions: Variability in EndoPredict risk classification reflects both tumor biological features and population-specific factors. These findings emphasize the importance of interpreting genomic risk scores within their clinical and demographic context and support the comparison of risk distributions across heterogeneous patient cohorts.