A Comprehensive Multistate Model Analyzing Associations of Various Risk Factors With the Course of Breast Cancer in a Population-Based Cohort of Breast Cancer Cases

TY - JOUR

T1 - A Comprehensive Multistate Model Analyzing Associations of Various Risk Factors With the Course of Breast Cancer in a Population-Based Cohort of Breast Cancer Cases

AU - Eulenburg, Christine

AU - Schroeder, Jennifer

AU - Obi, Nadia

AU - Heinz, Judith

AU - Seibold, Petra

AU - Rudolph, Anja

AU - Chang-Claude, Jenny

AU - Flesch-Janys, Dieter

N1 - © The Author 2016. Published by Oxford University Press on behalf of the Johns Hopkins Bloomberg School of Public Health. All rights reserved. For permissions, please e-mail: journals.permissions@oup.com.

PY - 2016/1/27

Y1 - 2016/1/27

N2 - We employed a semi-Markov multistate model for the simultaneous analysis of various endpoints describing the course of breast cancer. Results were compared with those from standard analyses using a Cox proportional hazards model. We included 3,012 patients with invasive breast cancer newly diagnosed between 2001 and 2005 who were recruited in Germany for a population-based study, the Mamma Carcinoma Risk Factor Investigation (MARIE Study), and prospectively followed up until the end of 2009. Locoregional recurrence and distant metastasis were included as intermediate states, and deaths from breast cancer, secondary cancer, and other causes were included as competing absorbing states. Tumor characteristics were significantly associated with all breast cancer-related endpoints. Nodal involvement was significantly related to local recurrence but more strongly related to distant metastases. Smoking was significantly associated with mortality from second cancers and other causes, whereas menopausal hormone use was significantly associated with reduced distant metastasis and death from causes other than cancer. The presence of cardiovascular disease at diagnosis was solely associated with mortality from other causes. Compared with separate Cox models, multistate models allow for dissection of prognostic factors and intermediate events in the analysis of cause-specific mortality and can yield new insights into disease progression and associated pathways.POM-Newsletter

AB - We employed a semi-Markov multistate model for the simultaneous analysis of various endpoints describing the course of breast cancer. Results were compared with those from standard analyses using a Cox proportional hazards model. We included 3,012 patients with invasive breast cancer newly diagnosed between 2001 and 2005 who were recruited in Germany for a population-based study, the Mamma Carcinoma Risk Factor Investigation (MARIE Study), and prospectively followed up until the end of 2009. Locoregional recurrence and distant metastasis were included as intermediate states, and deaths from breast cancer, secondary cancer, and other causes were included as competing absorbing states. Tumor characteristics were significantly associated with all breast cancer-related endpoints. Nodal involvement was significantly related to local recurrence but more strongly related to distant metastases. Smoking was significantly associated with mortality from second cancers and other causes, whereas menopausal hormone use was significantly associated with reduced distant metastasis and death from causes other than cancer. The presence of cardiovascular disease at diagnosis was solely associated with mortality from other causes. Compared with separate Cox models, multistate models allow for dissection of prognostic factors and intermediate events in the analysis of cause-specific mortality and can yield new insights into disease progression and associated pathways.POM-Newsletter

U2 - 10.1093/aje/kwv163

DO - 10.1093/aje/kwv163

M3 - SCORING: Journal article

C2 - 26823437

VL - 183

SP - 325

EP - 334

JO - AM J EPIDEMIOL

JF - AM J EPIDEMIOL

SN - 0002-9262

IS - 4

ER -