Surveillance of BRCA1 and BRCA2 CarriersSurveillance of

To the Editor: The study by Dr Warner and colleagues¹ comparing magnetic resonance imaging (MRI), mammography, ultrasound, and clinical breast examination in screening BRCA1 and BRCA2 carriers for breast cancer has several potential limitations that could affect their conclusions.

First, the study compared prevalent cancer screening by MRI with incident cancer screening by mammography. This may bias against mammography since prevalent screening generally detects more cancer than incident screening. The sensitivity of MRI in prevalent screen 1 was better than its performance in subsequent incident screens, with sensitivity decreasing from 85% to 71% and 50% on annual screens 1, 2, and 3, respectively. This same pattern was observed by Kriege et al² with MRI sensitivity decreasing from 79% at screen 1 to 62% at screen 2.

Second, different interpretive threshold behaviors for mammography vs MRI may have biased the reported sensitivities. The positive predictive value for mammography was much higher than that for MRI, suggesting a relatively greater emphasis on specificity for mammography and on sensitivity for MRI. There was also a much higher recommended short-term follow-up rate for MRI compared with mammography (7.6% vs 0.4%). In addition, MRI had a high recall rate of 16.5%; mammography recall rate was not reported.

Third, it was stated that those individuals who performed ultrasound were blinded to the outcome of other modalities. It is important to know whether those individuals reading the MRIs were also blinded to mammography and ultrasound results because there were only 2 study readers.

Finally, a more informative receiver operating characteristic analysis would use data only for the incident screening results of the various methods.