Prognostic Models for Breast Cancer Using Clinical and Gene Expression Data
Author Information
Author(s): Fan Cheng, Prat Aleix, Parker Joel S, Liu Yufeng, Carey Lisa A, Troester Melissa A, Perou Charles M
Primary Institution: University of North Carolina, Chapel Hill, USA
Hypothesis
Combining clinical variables with gene expression signatures can improve the prediction of outcomes in breast cancer patients.
Conclusion
Integrating gene expression signatures with clinical-pathological factors enhances prognostic accuracy for breast cancer patients.
Supporting Evidence
- Models that included both clinical and genomic parameters improved prognostication compared to models with either type alone.
- Statistically significant prognostic models for relapse-free survival were identified for the entire population and specific subgroups.
- Combined models for predicting pathological complete response were also statistically significant.
Takeaway
This study shows that using both clinical information and gene data together helps doctors better predict how breast cancer patients will do.
Methodology
Multivariate Cox proportional hazards models were built using clinical-pathological variables and 323 gene expression modules from a dataset of 550 breast cancer patients.
Potential Biases
Potential bias due to the dataset being primarily composed of tissue bank samples.
Limitations
The study's sample size limited the power for some patient subsets, and the implementation of many predictors may have resulted in loss of gene data.
Participant Demographics
The majority of patients had ER-positive (71.8%) and HER2-negative (80%) tumors.
Statistical Information
P-Value
p<0.05
Statistical Significance
p<0.05
Digital Object Identifier (DOI)
Want to read the original?
Access the complete publication on the publisher's website