McCormack · Cancer epidemiology, biomarkers & prevention : a publication of the American Association for Cancer Research, cosponsored by the American Society of Preventive Oncology 2006 · systematic review and meta-analysis · n=42 studies (>14,000 cases, 226,000 noncases)

Breast density and parenchymal patterns as markers of breast cancer risk: a meta-analysis.

Cited 2168 times in the scientific literature.

Level 1 - systematic review of randomized trials

Systematic review and meta-analysis of observational studies

PubMed 16775176 · doi:10.1158/1055-9965.EPI-06-0034 · record verified 2026-08-26

What was done

Systematic review and meta-analysis of 42 studies with aggregate data on over 14,000 breast cancer cases and 226,000 noncases. Random-effects models were used to pool study-specific relative risks of mammographic patterns and breast cancer risk. Analyses compared general versus symptomatic populations, percentage density versus Wolfe grade and BI-RADS classification, incident versus prevalent cancers, age and menopausal status, ethnicity, and prediagnostic timing.

What was found

Associations were consistent in general population studies but highly heterogeneous in symptomatic populations. Relative risks were 20% to 30% stronger for incident than prevalent cancer and were substantially stronger for percentage density than for Wolfe grade or BI-RADS. No differences were observed across age, menopausal status, or ethnicity. For percentage density on prediagnostic mammograms in the general population, pooled relative risks for incident breast cancer compared with density under 5% were 1.79 (95% CI, 1.48-2.16) for 5% to 24%, 2.11 (95% CI, 1.70-2.63) for 25% to 49%, 2.92 (95% CI, 2.49-3.42) for 50% to 74%, and 4.64 (95% CI, 3.64-5.91) for 75% or greater. The association remained strong after excluding cancers diagnosed within one year post-mammography.

Why it matters

This review establishes high mammographic percentage density as one of the strongest risk factors for breast cancer and demonstrates that the risk is not merely an artifact of masking bias or restricted to specific age groups.

Limits

The meta-analysis used aggregate data rather than individual participant data. High heterogeneity was observed among symptomatic populations, and the abstract does not report adjustment for potential confounders such as body mass index or reproductive factors across the included studies.

Cited by