Imaging Biomarkers as Predictors for Breast Cancer Death.

Wendy Yi-Ying Wu,Hsiu-Hsi Chen,Tibor Tot,Ching-Yuan Fann,Chen-Yang Hsu,Kerri R Beckmann,Stephen W Duffy,May Mei-Sheng Ku,Laszlo Tabar,Sherry Yueh-Hsia Chiu,Amy Ming-Fang Yen,Robert A Smith,Sam Li-Sheng Chen

doi:10.1155/2019/2087983

Abstract

Background To differentiate the risk of breast cancer death in a longitudinal cohort using imaging biomarkers of tumor extent and biology, specifically, the mammographic appearance, basal phenotype, histologic tumor distribution, and conventional tumor attributes. Methods Using a prospective cohort study design, 498 invasive breast cancer patients diagnosed between 1996 and 1998 were used as the test cohort to assess the independent effects of the imaging biomarkers and other predictors on the risk of breast cancer death. External validation was performed with a cohort of 848 patients diagnosed between 2006 and 2010. Results Mammographic tumor appearance was an independent predictor of risk of breast cancer death (P=0.0003) when conventional tumor attributes and treatment modalities were controlled. The casting type calcifications and architectural distortion were associated with 3.13-fold and 3.19-fold risks of breast cancer death, respectively. The basal phenotype independently conferred a 2.68-fold risk compared with nonbasal phenotype. The observed deaths did not differ significantly from expected deaths in the validation cohort. The application of imaging biomarkers together with other predictors classified twelve categories of risk for breast cancer death. Conclusion Combining imaging biomarkers such as the mammographic appearance of the tumor with the histopathologic distribution and basal phenotype, accurately predicted long-term risk of breast cancer death. The information may be relevant for determining the need for molecular testing, planning treatment, and determining the most appropriate clinical surveillance schedule for breast cancer patients.

Highlights

The first generation of breast cancer prognostic indicators, tumor size, histologic malignancy grade, and spread to lymph nodes have long influenced treatment regimens [1,2,3,4]
The aim of the present study is to explore whether imaging biomarkers such as the mammographic appearance of the tumor, the tumor phenotype with particular emphasis on basal phenotype, and the histologic tumor distribution can be combined to differentiate risk groups for breast cancer death in a longitudinal cohort of women diagnosed with breast cancer that was derived from a population-based organized mammography screening program
The survival curves by mammographic appearance of the tumor, basal phenotype, and histologic tumor distribution were plotted using the lifetable method

Summary

Introduction

The first generation of breast cancer prognostic indicators, tumor size, histologic malignancy grade, and spread to lymph nodes have long influenced treatment regimens [1,2,3,4] These conventional tumor attributes may not be sufficient for predicting the prognosis of breast cancer after the era of early detection via the wide spread use of mammography and molecular tumor subtypes. To differentiate the risk of breast cancer death in a longitudinal cohort using imaging biomarkers of tumor extent and biology, the mammographic appearance, basal phenotype, histologic tumor distribution, and conventional tumor attributes. Combining imaging biomarkers such as the mammographic appearance of the tumor with the histopathologic distribution and basal phenotype, accurately predicted long-term risk of breast cancer death. The information may be relevant for determining the need for molecular testing, planning treatment, and determining the most appropriate clinical surveillance schedule for breast cancer patients

Objectives

Methods

Results

Discussion

Conclusion