Imagers seek clarity on pool of supplemental screening patients

February 6, 2008

Imagine hundreds of thousands of women, informed by a list of potential risk factors for breast cancer, knocking on radiologists’ doors, demanding another imaging study to supplement mammography. Will radiology practices know how to respond to this scenario?

Imagine hundreds of thousands of women, informed by a list of potential risk factors for breast cancer, knocking on radiologists’ doors, demanding another imaging study to supplement mammography. Will radiology practices know how to respond to this scenario?

Recent developments have generated much uncertainty in breast imaging when it comes to managing not only women at high risk but also the large number of women deemed to be at intermediate risk. Last year, the American Cancer Society recommended that high-risk women undergo breast MRI along with mammography in annual screening. The definition of high risk includes a lifetime likelihood of breast cancer of 20% or greater (see table), but some factors cited are hard to pin down.

In an ideal world, family history would be available and clear for all women. But, in reality, pedigrees are sometimes incomplete and cancers can be easily overlooked, said Dr. Robert Smith, director of screening at the ACS.

“Determining risk based on family history can be very vexing for radiology practices,” Smith said during a presentation on high-risk screening at the tail-end of the 2007 RSNA meeting.

Concerned and curious about the potential impact of the ACS guidelines, Dr. Rand Stack performed a study measuring demand and supply. About 1.8 million women are at high risk according to the ACS definitions, making them eligible for breast MRI as an adjunct to annual screening, said Stack, who presented results at the RSNA meeting.

Most of these women-1.7 million-qualify on the basis of a strong family history, meaning they have multiple relatives who have had breast or ovarian cancer, said Stack, section chief of the Breast Imaging Center at White Plains Hospital in New York.

In these cases, a family pattern suggests inheritance but a gene has not yet been isolated, Stack said in an interview with Diagnostic Imaging. Of the 1.8 million women at risk for all reasons, only an estimated 87,000 have the BRCA1 and BRCA2 mutations.

As for supply, Stack's study shows that if the 14,000 physicians currently performing mammography all offered breast MRI, workforce capacity would be sufficient to meet demand. But it's unclear whether such an equal distribution of work is achievable in practice.

FALSE-POSITIVE HIT

In another key 2007 development, principal investigator Dr. Wendie Berg released eagerly awaited results from the American College of Radiology Imaging Network 6666 high-risk ultrasound screening trial at the RSNA meeting. Previous studies had not been able to clarify the value of ultrasound due to limitations in trial design and incomplete reporting of results. In the new, well-designed ACRIN study, funded by the National Cancer Institute and the Avon Foundation, ultrasound increased cancer detection by 30% but at the cost of far too many false positives.

The literature indicates that for high-risk women, MRI and mammography together are most appropriate and that adding ultrasound will not pick up many more additional cancers, said Dr. Edward Sickles, a professor of radiology at the University of California, San Francisco, during a presentation at the RSNA meeting.

“There is a big false-positive hit with ultrasound, without very much gain. Why think about ultrasound at all? It's more widely available, less expensive, and better tolerated. How will we use it ultimately? We don't know yet,” Sickles said.

If MRI is not available, for example, ultrasound makes sense for high-risk screening, he said. Though MRI's sensitivity is significantly higher than mammography, specificity is significantly lower.

“MRI has twice the number of recalls and more than twice the number of biopsies [as mammography]. You get something for that price-higher cancer detection, lower rates of false negatives, and lower rates of interval cancers,” Sickles said.

It's possible that rates of interval cancers could be reduced even further if providers stagger annual screening exams in high-risk women, performing MRI and mammography six months apart, he said.

Assuming average charges of $1038 for initial MRI and 75% compliance with ACS screening guidelines, direct costs for adding MRI as a supplemental screening tool would reach $1.4 billion a year, according to Stack's analysis (see chart). Use of breast MRI would also create indirect expenses related to follow-up studies and biopsies at the annual rate of $429 million, again assuming 75% compliance.

The literature shows that short-interval follow-up is recommended much more frequently with MR than mammography and that relatively few of these patients have cancer. The proportion of mammographic findings deemed “probably benign” is about 7% to 8% in Europe, whereas it can be as high as 20% to 25% in the U.S., possibly due to defensive medicine practices.

“Radiologists [in the U.S.] are more cautious about borderline lesions. Hopefully, the percentage of probably benign lesions will come down to below 10% as people get more experienced with breast MRI screening,” Stack said.

A new study shows that acceptable rates of probably benign findings were achieved in the ACRIN 6667 screening trial of high-risk women, according to Dr. Susan Weinstein, who presented results at the RSNA meeting.

There are multiple advantages to scoring a lesion as “probably benign,” said Weinstein, an assistant professor of radiology at the University of Pennsylvania.

“If the probably benign category is appropriately applied, the malignancy rate is low. It decreases the number of unnecessary biopsies, decreases expenditures, and helps relieve patient anxiety,” she said.

Of 969 women, 10% (106) had findings deemed probably benign. Extensive follow-up showed that only one woman was subsequently diagnosed with cancer. In this case, the woman had ductal carcinoma in situ detected on elective mastectomy. The malignancy rate for probably benign lesions is in line with mammography, Weinstein said.

PATIENT HISTORY

When managing patients, radiology practices should first ask if any first-degree or second-degree relatives have a history of breast or ovarian cancer and remind patients to include the paternal as well as the maternal side, Smith said. The number of relatives affected and their ages of onset will affect the patient's overall lifetime risk.

Most patients will not have a family history of breast cancer. Furthermore, most of those with a positive history will not develop cancer, and if they do, risk of death is low.

“This pattern is reassuring for patients. They are not certain to be diagnosed even if they have two first-degree relatives with cancer,” Smith said.

Free downloadable software tools can help estimate risk in patients with a positive family history.

“These packages are simple to use and can be very useful in your practice to identify those who will benefit from MRI screening,” Smith said.

Commonly used risk estimation models include the following:

  • BRCAPRO: astor.som.jhmi.edu/BayesMendel/brcapro.html.

A recent article, “Management of an inherited predisposition to breast cancer,” by Robson and Offit in The New England Journal of Medicine provides a concise review of genes linked to breast cancer and the strengths and limitations of programs listed above (2007;357:154-162). The preferred models for estimating risk based on family history are Tyrer-Cuzick and BRCAPRO, whereas the Gail model is a bit simplistic, according to Smith.

Sickles advises that radiologists are not particularly well trained in deciding which women are eligible for high-risk screening and suggests they seek out experts in evaluating risk.

“Information is available but not easy to use. Until it becomes easier to use, it's best to partner with someone who knows how to use the programs and does so on a regular basis,” Sickles said.

BREAST DENSITY

Evidence is insufficient to recommend MRI for those with a personal history of cancer or dense breasts as a sole risk factor, according to the ACS guidelines.

Radiologists are accustomed to thinking of density in terms of its effects on the quality of a mammogram. But density is actually an important, underappreciated, and very common risk factor for breast cancer, said Dr. Jennifer Harvey, head of breast imaging at the University of Virginia Health System.

Compared with a woman with fatty breasts, a woman with extremely dense breasts has a fivefold increased risk for breast cancer. Contrary to common perception, the increased risk is not due to later diagnosis of cancers in dense tissue. Some studies, controlling for epidemiological factors such as age and menopausal status, have shown that density is an independent risk factor.

Half of women aged 40 to 49 have breasts that are 50% or more dense, as do nearly one-third of women aged 70 to 79, said Harvey during a presentation at the RSNA meeting. She cited two studies that have estimated that breast density is a risk factor in 28% to 30% of all cancers.

In the late 1970s and early 1980s, when research on breast density was beginning, the dense part of the mammogram was reported to represent predominantly fibrous stroma, Harvey said. Because the dense tissue did not contain a large number of epithelial cells, it was thought to be not very important.

However, more recent studies have demonstrated the importance of interactions between stroma and epithelial cells, particularly paracrine effects of growth factors and aromatase, effects that are produced by macrophages and fibroblasts in the stroma. Growth factors could stimulate the replication of normal breast tissue as well as cancer cells. A new study by Harvey and others shows that dense tissue is composed of stroma and greater numbers of lobules, but not ducts, compared with fatty breast tissue. (Menopause, published online June 7, 2007).

All of the available software models underpredict the risk of breast cancer, but if breast density is added into the equation, the accuracy of the models increases for all age groups, according to Harvey.

Though mammography is proven to be less sensitive in dense tissue, it is nevertheless unclear whether MRI or ultrasound should be used in supplemental screening.

“The data are just not there for a woman whose sole risk factor is breast density. We don't know if MRI is effective or what the cancer detection rate, sensitivity, or specificity might be,” she said.

Sickles agreed that density alone is not enough to warrant an MRI but said that ultrasound might be appropriate in some cases. Ultrasound offers a substantial increase in detection, and the malignant findings tend to be small, invasive, and node-negative. But the positive predictive value is much lower than with mammography.

“The bottom line is we don't know the answers. I would not recommend that we all do ultrasound screening on women with dense breasts on the basis of the data reported. Right now, there is insufficient evidence to do any more than mammography,” he said.

At Stack's hospital, however, primary-care doctors often order a follow-up ultrasound if the mammography report indicates that a patient's breasts are extremely dense.

“I predict that in the future, there will be enough data to demonstrate that everyone with dense breasts should have screening ultrasound,” Stack said.

If a patient's mammogram is unclear due to density, it is reasonable to explain the option of ultrasound, as long as the downside of false positives is thoroughly explained, Smith said. Then it is up to the woman to determine if she wants to accept the risks.

Emily Hayes is feature editor of Diagnostic Imaging.