Risk Stratification for Overall Survival among Metastatic Prostate Cancer Patients Treated by ADT as the First Line Treatment Open Access

Ji,Yuhan (Summer 2020)

Permanent URL: https://etd.library.emory.edu/concern/etds/fx719n62n?locale=en
Published

Abstract

Background: Risk stratification is an important clinical process to divide patients into different groups based on their health signs and to carry right level of care, and there have already existed numerous standards to do risk stratification towards prostate cancer patients. In this study, we examine if nomogram, decision tree and regression coefficient-based scoring systems are useful as risk stratification tools towards prognostic assessment in metastatic prostate cancer patients.

Method: In this study, we implemented 3 alternative methods for risk stratification: nomogram, decision tree, three kinds of regression coefficient-based scoring system (Beta/Schneeweiss score, Beta/Sullivan score and HR/Charlson score). These 3 methods were conducted for an application on metastatic prostate cancer from NCDB dataset, where overall survival among patients were examined through Cox proportional hazard model. Prediction ability of 3 risk stratification methods were examined.

Conclusion: This study reveals that race, age, Charlson-Deyo Score, clinical stage, PSA, bone metastasis involvement, positive biopsy cores percentage are independent prognostic indicators for overall survival for metastatic prostate cancer patients. For risk stratification, nomogram was constructed with 1 and 2-year survival rate and had a C-index of 0.614, decision tree was not ideal with only age, Charlson-Deyo Score and positive biopsy cores percentage left in the model. Beta/Sullivan scoring system outperform other two regression coefficient-based scoring algorithms and had an average C-index of 0.595.

KEYWORD: Risk stratification, Nomogram, Decision tree, Regression coefficient-based scoring system

Table of Contents

1 Introduction

2 Method

2.1 Define study population

2.2 Nomogram

2.3 Decision tree

2.4 Regression coefficient-based scoring system

3 Result

3.1 Patient characteristics

3.2 Univariate and multivariate association

3.3 Risk stratification

3.3.1 Nomogram construction

3.3.2 Nomogram validation

3.3.3 Decision tree

3.3.4 REgression coefficient-based scoring system

4 Discussion

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
Department
Degree
Submission
Language
  • English
Research Field
Keyword
Committee Chair / Thesis Advisor
Committee Members
Last modified

Primary PDF

Supplemental Files