Breast ultrasound adjusts to get past false positives

The massive American College of Radiology Imaging Network 6666 trial shows that adding ultrasound to the initial breast screening protocol for high-risk women could help detect 30% more cancers, but at the questionable cost of many more needless biopsies of benign lesions.

Results from the trial, funded by the National Cancer Institute and the Avon Foundation, were discussed at the RSNA meeting by principal investigator Dr. Wendie Berg. Despite the downside, the results suggest that ultrasound could still play a role in screening women at higher risk for breast cancer who are not eligible for MRI.

"Ultrasound is attractive for supplemental screening. It is widely available, well tolerated by patients, and relatively inexpensive, and it involves no radiation," Berg said.

This year, the American Cancer Society recommended breast MRI in addition to mammography for screening women at very high risk of breast cancer, such as those with the BRCA1 or BRCA2 mutation, those with a lifetime risk of breast cancer 20% or higher, and those who were exposed to chest radiation between the ages of 10 and 30 years. The guidelines do not apply to the large number of women considered to be at intermediate risk, including those whose only risk factor is breast density or prior history of cancer.

Research suggests that if MRI and mammography are performed, there is no need to perform screening ultrasound, Berg said. But the new data may support offering ultrasound to women at intermediate risk, when there is uncertainty about the use of MRI.

The ACRIN 6666 trial established that high-quality ultrasound can be performed across many sites. The study involved 2637 women at higher risk of breast cancer, most commonly due to a personal history of breast cancer. BRCA1 and BRCA2 mutations were uncommon.

Of 40 cancers, ultrasound alone detected 12 malignancies, most of which were invasive node-negative malignancies. Use of ultrasound would increase the detection yield by 4.2 per 1000 women screened, and the modality's value applied across the board, regardless of breast density.

The higher sensitivity made possible by the supplemental ultrasound scans came at a questionable price. Based on mammography, 69 women (2.6%) would be referred for biopsy, and 29% of these were positive. By comparison, the addition of ultrasound resulted in biopsies for 202 women (7.7%), and only 15% were positive, Berg said.

"There was a significant increase in the number of biopsies, by and large of benign lesions. If we offer screening ultrasound, we need to make women aware of the substantial risks of false positives. The risk exceeds standards for MRI and mammography. For some women, the risks outweigh the potential benefits," she said.

Berg expects that the false-positive rate will decline on subsequent screens and also said that new techniques such as ultrasound elastography could help cut down on unnecessary biopsies of benign lesions.

Elastography refers to the measurement of elastic properties of tissues, based on the well-established principle that malignant tissue is harder than benign tissue. In another study at the RSNA meeting, Dr. Richard Barr presented multisite trial results with real-time freehand ultrasound elastography on the Siemens Antares unit, as a follow-up to a positive single-center study presented in 2006.

The new trial involved 635 patients scheduled for ultrasound-guided biopsy at six sites. It took one to two additional minutes to obtain the elastography image and measure the lesion size (malignancies appear to be larger on elastography). The images were interpreted before pathology results were known.

In five of six sites, radiologists predicted all of the cancers based on ultrasound elastography. In one site, radiologists missed three malignancies, including one case of ductal carcinoma in situ, according to Barr, a professor of radiology at Northeastern Ohio Universities College of Medicine.

Across the six sites, sensitivity averaged 98.6%, whereas specificity ranged from 67% to 95%, with an average of 87%. Given this range, further research into interobserver variability is needed, if false positives are to be reduced across the board.

"Hopefully, we can decrease the number of biopsies," Barr said.