Francesco Sardanelli MD, Presenter: Consultant, Bracco Group

Franca Podo DSc, Abstract Co-Author: Nothing to Disclose

Flippo Santoro DSc, Abstract Co-Author: Nothing to Disclose

Luca Alessandro Carbonaro MD, Abstract Co-Author: Nothing to Disclose

Giovanni Di Leo, Abstract Co-Author: Nothing to Disclose

Alessandro Del Maschio MD, Abstract Co-Author: Nothing to Disclose

Daniele Vergnaghi MD, Abstract Co-Author: Nothing to Disclose

00030490-DMT et al, Abstract Co-Author: Nothing to Disclose

To evaluate the evolution of PPV of contrast-enhanced MRI during a multicenter multimodality surveillance of women at genetic-familial high-risk of breast cancer.

IRB approval and informed written consent were obtained. From June 2000 to January 2007, we enrolled 502 women in 18 Centers: 195 BRCA1 mutation carriers or their 1st degree relatives; 148 BRCA2 mutation carriers or their 1st degree relatives; and 159 women enrolled on the only basis of a strong family history of breast or ovarian cancer (three or more events in 1st-2nd degree maternal or paternal relatives, i.e. breast cancers in women younger than 60 years and ovarian cancers or male breast cancers at any age), including women with previous personal breast cancer history. Annual mammography, ultrasonography (US), and MRI were planned. Reference standard was histopathology or at least 1-year negative follow-up. False positive were those examinations which prompted a needle or surgical biopsy with negative pathology. ROC analysis and chi-square for trend were used.

A total of 502 women were enrolled with an entry age of 46.0±11.8 years and 1,596 screening events (mean 3.2 per woman) were performed. A total of 52 breast cancers were found (48 screen-detected and 3 interval cancers): 44 invasive and 8 ductal carcinoma in situ. Patient-based sensitivity was 0.913 (42/46) for MRI, 0.500 (25/50) for mammography and 0.52 (26/50) for US; specificity 0.974 (976/1002), 0.991 (1040/1049), and 0.992 (1011/1019), respectively. Positive likelihood ratio was 35.2, 58,3, and 66.2; negative likelihood ratio 0.09, 0.50, and 0.48, respectively. At ROC analysis, the area under the curve was in favor of MRI (0.97) versus mammography (0.83), US (0.82), and mammography plus US (0.87)(P<.005). Overall PPV was 0.618 (42/68), 0.735 (25/34), and 0.765 (26/34), respectively (P=n.s.). PPV of MRI was 0.571 (20/35) at the first round, 0.647 (11/17) at the second round, 0.667 (8/12) at the third round, 0.750 (3/4) at the fourth round (P=n.s.).

A trend for a learning curve with increasing PPV of MRI was observed in the screening setting. The relatively small number of positive cases did not allow to get a statistical significance.

As known from screening mammography, also for MRI the readers experience tends to reduce the frequency of false positive findings. An MRI PPV similar to that of mammography or US can be reached.

Sardanelli, F, Podo, F, Santoro, F, Carbonaro, L, Di Leo, G, Del Maschio, A, Vergnaghi, D, et al, 0, Evolution of Positive Predictive Value (PPV) of MRI during a Multimodality Multicenter Surveillance of High Risk Women.  Radiological Society of North America 2009 Scientific Assembly and Annual Meeting, November 29 - December 4, 2009 ,Chicago IL. http://archive.rsna.org/2009/8007215.html