Breast cancer imaging phenotype is diverse and may relate to molecular alterations driving cancer behavior. We systematically reviewed and meta-analyzed relations between breast cancer imaging features and human epidermal growth factor receptor type 2 (HER2) overexpression as a marker of breast cancer aggressiveness. MEDLINE and EMBASE were searched for mammography, breast ultrasound, magnetic resonance imaging (MRI), and/or [(18)F]fluorodeoxyglucose positron emission tomography studies through February 2013. Of 68 imaging features that could be pooled (85 articles, 23,255 cancers; random-effects meta-analysis), 11 significantly related to HER2 overexpression. Results based on five or more studies and robustness in subgroup analyses were as follows: the presence of microcalcifications on mammography [pooled odds ratio (pOR), 3.14; 95% confidence interval (CI), 2.46-4.00] or ultrasound (mass-associated pOR, 2.95; 95% CI, 2.34-3.71), branching or fine linear microcalcifications (pOR, 2.11; 95% CI, 1.07-4.14) or extremely dense breasts on mammography (pOR, 1.37; 95% CI, 1.07-1.76), and washout (pOR, 1.57; 95% CI, 1.11-2.21) or fast initial kinetics (pOR, 2.60; 95% CI, 1.43-4.73) on MRI all increased the chance of HER2 overexpression. Maximum [(18)F]fluorodeoxyglucose standardized uptake value (SUVmax) was higher upon HER2 overexpression (pooled mean difference, +0.76; 95% CI, 0.10-1.42). These results show that several imaging features relate to HER2 overexpression, lending credibility to the hypothesis that imaging phenotype reflects cancer behavior. This implies prognostic relevance, which is especially relevant as imaging is readily available during diagnostic work-up.
©2014 American Association for Cancer Research.