Abstract 5054: Breast cancer survival analysis with the size of the infiltrative cancer area

Background: Analyzing breast cancer prognosis factors is important for survival analysis. In general, the approximate size of the tumor is considered in pathologic staging, but it is not known whether the exact size of the infiltrative cancer area is related to the patients' survival. To analyz...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Veröffentlicht in:Cancer research (Chicago, Ill.) Ill.), 2022-06, Vol.82 (12_Supplement), p.5054-5054
Hauptverfasser: Choi, JunYoung, Kwak, Tae-Yeong, Kim, Sun Woo, Chang, Hyeyoon
Format: Artikel
Sprache:eng
Online-Zugang:Volltext
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
Beschreibung
Zusammenfassung:Background: Analyzing breast cancer prognosis factors is important for survival analysis. In general, the approximate size of the tumor is considered in pathologic staging, but it is not known whether the exact size of the infiltrative cancer area is related to the patients' survival. To analyze the effectiveness of the size of the infiltrative cancer area for survival analysis of patients, we developed a machine learning algorithm that predicts death risk by analyzing clinical data and the actual size of the infiltrative cancer area from hematoxylin and eosin stained whole slide images (WSIs) of breast cancer. Design: To obtain an area of infiltrative cancer, an experienced pathologist conducted annotations using WSIs of breast cancer resection tissue. The data analyzed in this study was from The Cancer Genome Atlas’s breast cancer data set (TCGA-BRCA). There were 871 WSIs labeled with survival events and periods, and several clinical features according to patients including the age at diagnosis, ER, PR, HER2 status, and pathologic TNM stages. We extract the size of the infiltrative cancer area from annotations and incorporate them with clinical features. There existed 53 uncensored data, and we splitted the whole dataset into 8:2 for training/validation. We performed 5-fold cross validation with random sampling. We trained a Cox Regression model with features and compared C-index according to features. Result: The C-index with the infiltrative cancer area for the test set was 0.806. This figure is an improvement of 0.2 from 0.781 without the infiltrative cancer area. The hazard ratio of the infiltrative cancer area was 1.84. The p-value of the infiltrative cancer area was 0.019 for the log-rank test. Conclusion: This result shows that the infiltrative cancer area is a useful feature even in the presence of the TNM stage. Citation Format: JunYoung Choi, Tae-Yeong Kwak, Sun Woo Kim, Hyeyoon Chang. Breast cancer survival analysis with the size of the infiltrative cancer area [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2022; 2022 Apr 8-13. Philadelphia (PA): AACR; Cancer Res 2022;82(12_Suppl):Abstract nr 5054.
ISSN:1538-7445
1538-7445
DOI:10.1158/1538-7445.AM2022-5054