Predicting invasive breast cancer versus DCIS in different age groups

Area of Science:

• Radiology
• Oncology
• Breast Imaging

Background:

• Growing concern over overtreatment of certain breast cancers.
• Need to understand age-related differences in predicting invasive breast cancer versus DCIS.

Purpose of the Study:

• To investigate if clinical and mammographic features predicting invasive breast cancer vs. DCIS differ by patient age.
• To compare the predictive performance of these features across different age groups.

Main Methods:

• Analysis of 1,475 malignant breast biopsies (1,063 invasive, 412 DCIS) from 35,871 mammograms.
• Development of logistic regression models for women ≥ 65, 50-64, and < 50 years.
• Evaluation of model performance using the area under the receiver operating characteristic curve (AUC).

Main Results:

• Predictive models performed significantly better for older (AUC=0.848) and middle-aged (AUC=0.851) groups compared to younger women (AUC=0.778).
• Palpability and principal mammographic finding were significant predictors across all ages.
• Specific predictors varied: family history, mass shape/margins for older women; mass margins for middle-aged; mass size for younger women.

Conclusions:

• Clinical and mammographic features are more effective in predicting invasive breast cancer vs. DCIS in older women.
• Age-specific predictive variables are crucial for accurate breast cancer diagnosis.