Studying the Association between Overall Survival and Metastatic Sites by Breast Cancer Subtypes Based on National Cancer Database Público

Abstract

Background: Metastatic breast cancer is the main cause of breast cancer-associated deaths and receptor status has a large impact on prognosis. Other prognostic factors such as adjuvant systemic treatment also contribute to the heterogeneity in breast carcinomas. Within this scenario, how breast cancer subtype and metastatic site are associated with breast cancer overall survival remains unclear. 

Method: A total of 5211 female patients with stage IV breast cancer from the National Cancer Database (NCDB) (2010-2013) were examined. All patients received surgery and systemic treatment. The distribution of metastatic sites among breast cancer subtypes was evaluated through a χ^2test. Univariate and multivariate analyses using two semi-parametric approaches, including the Cox proportional hazard (PH) analyses and the censored quantile regression analyses, were conducted to assess the associations between metastatic sites and overall survival. 

Results: HR+/HER2- breast cancer was most likely to metastasize to bone, TNBC was most likely to metastasize to brain or lung and HR-/HER2+ was most likely to metastasize to liver or multiple organs. Overall, patients with bone metastasis appeared to have the best prognosis while patients with multiple metastasis had the worst prognosis. In univariate quantile regression analyses, the survival differences between bone metastasis versus multiple metastasis were varied over quantiles, except for HR-/HER2+ breast cancer. In multivariate analysis, age showed negative prognostic effect among patients with all subtypes and was varied in HR+/HER2- subtype. In particular, TNBC patients with bone metastasis versus multiple metastases had varying quantile effects above the median.  

Conclusion: This study showed different breast cancer subtypes had different metastatic patterns and survivals. Compared with the Cox model, the censored quantile regression model revealed a more comprehensive prognostic patterns in metastatic breast cancer. Adjusting clinical surveillance and treatment strategies were suggested based on the variation of prognostic effects in different metastatic sites over time. 

Table of Contents

LIST OF TABLES ---------------------------------------------------------------------------------------------------------I

LIST OF FIGURES -------------------------------------------------------------------------------------------------------II

CHAPTER I: INTRODUCTION -------------------------------------------------------------------------------------------1

1.1 Background ----------------------------------------------------------------------------------------------------------1

1.2 National Cancer Database -------------------------------------------------------------------------------------------3

1.3 Two Semiparametric Regression Models ----------------------------------------------------------------------------4

1.3.1 Cox Proportional Hazards Regression Model ----------------------------------------------------------------------4

1.3.2 Censored Quantile Regression Model ------------------------------------------------------------------------------4

1.3.3 Motivations for Considering Both Models -------------------------------------------------------------------------5

1.4 Outlines of Thesis ---------------------------------------------------------------------------------------------------6

CHAPTER II: METHOD --------------------------------------------------------------------------------------------------7

2.1 Data Set Information ------------------------------------------------------------------------------------------------7

2.1.1 Patient Selection --------------------------------------------------------------------------------------------------7

2.1.2 Data Cleaning -----------------------------------------------------------------------------------------------------7

2.1.3 Variable Description ----------------------------------------------------------------------------------------------8

2.2 Statistical Analyses -------------------------------------------------------------------------------------------------9

2.2.1 Descriptive Analysis ----------------------------------------------------------------------------------------------9

2.2.2 Analyses Based on Cox Proportional Hazard Model ---------------------------------------------------------------9

2.2.3 Analyses Based on Censored Quantile Regression Model --------------------------------------------------------12

CHAPTER III: RESULTS ------------------------------------------------------------------------------------------------15

3.1 Descriptive Summary ----------------------------------------------------------------------------------------------15

3.1.1 Demographic and Clinicopathological Characteristics -----------------------------------------------------------15

3.1.2 Metastatic Patterns in Breast Cancer Subtypes -------------------------------------------------------------------17

3.2 Cox Proportional Hazard Regression -------------------------------------------------------------------------------19

3.2.1 Univariate Analysis -----------------------------------------------------------------------------------------------19

3.2.2 Multivariate Analysis ---------------------------------------------------------------------------------------------22

3.3 Censored Quantile Regression --------------------------------------------------------------------------------------26

3.3.1 Univariate Analysis -----------------------------------------------------------------------------------------------26

3.3.2 Multivariate Analysis ---------------------------------------------------------------------------------------------35

CHAPTER IV: DISCUSSIONS -------------------------------------------------------------------------------------------44

4.1 Assumptions and Limitations ---------------------------------------------------------------------------------------45

4.1.1 Data Source and Patient Selection --------------------------------------------------------------------------------45

4.1.2 Regression Models ------------------------------------------------------------------------------------------------46

4.2 Future Research -----------------------------------------------------------------------------------------------------48

Appendix A: Regression Quantiles in Univariate Analysis --------------------------------------------------------------49

A.1 HR+/HER2- Subtype ------------------------------------------------------------------------------------------------49

A.2 HR+/HER2+ Subtype ------------------------------------------------------------------------------------------------53

A.3 HR-/HER2+ Subtype ------------------------------------------------------------------------------------------------58

A.4 TNBC Subtype ------------------------------------------------------------------------------------------------------63

Appendix B: Regression Quantiles in Multivariate Analysis ------------------------------------------------------------68

B.1 HR+/HER2- Subtype ------------------------------------------------------------------------------------------------68

B.2 HR+/HER2+ Subtype ------------------------------------------------------------------------------------------------73

B.3 HR-/HER2+ Subtype ------------------------------------------------------------------------------------------------76

B.4 TNBC Subtype ------------------------------------------------------------------------------------------------------82

REFERENCES -----------------------------------------------------------------------------------------------------------87

About this Master's Thesis

Rights statement

• Permission granted by the author to include this thesis or dissertation in this repository. All rights reserved by the author. Please contact the author for information regarding the reproduction and use of this thesis or dissertation.

School	Rollins School of Public Health
Department	Biostatistics
Subfield / Discipline	Biostatistics - MPH & MSPH
Degree	M.S.P.H.
Submission	Master's Thesis
Language	English
Research Field	Health Sciences, Pathology Health Sciences, Public Health Biology, Biostatistics
Palabra Clave	semiparametric model overall survival breast cancer subtype metastatic breast cancer censored quantile regression model
Committee Chair / Thesis Advisor	Limin Peng, Emory University
Committee Members	Xiaoxian Li, Emory University

Última modificación

Primary PDF

Thumbnail	Title	Date Uploaded	Actions
	Studying the Association between Overall Survival and Metastatic Sites by Breast Cancer Subtypes Based on National Cancer Database ()	2019-04-14 01:46:05 -0400	Press to Select an action Download

Supplemental Files

Thumbnail	Title	Date Uploaded	Actions