MRI promises to contribute greatly to cancer management

July 1, 2005

The benchmark of any breast imaging tool is its ability to match mammography in cancer detection. Breast MRI not only holds its own in population subsets, it also promises to contribute more to breast cancer management than mammography can.

The benchmark of any breast imaging tool is its ability to match mammography in cancer detection. Breast MRI not only holds its own in population subsets, it also promises to contribute more to breast cancer management than mammography can.

That's not a knock against the undisputed front-line tool for general breast cancer screening. It is, instead, a sign that radiologists are using MR to extend the boundaries of imaging's traditional role in detecting and characterizing breast cancer.

New techniques to heighten MR's specificity are having a ripple effect on patient care. In a study presented at the 2004 RSNA meeting, MR changed clinical management of patients with newly diagnosed cancer in 29% of cases, and it offered better assessment of tumor size than mammography or ultrasound.

"That's a high number," said author Dr. Gillian Newstead, director of breast imaging at the University of Chicago. "We have patients every day who are sent in for evaluation, and we find cancer in another quadrant that would have been missed."

At a community hospital in Florida, a kinetic and morphologic analysis used to interpret contrast-enhanced MRI is finding significantly more cancerous lesions than mammography in patients whose primary breast lesions were malignant (AJR 2005;184:878-886).

"We're probably understaging breast cancer at least 15% of the time when we do a standard mammogram supplemented by targeted ultrasound," said Dr. Jonathan Wiener, lead investigator and director of MRI at Boca Raton Community Hospital.

MRI is also being used to chart the success or failure of neoadjuvant chemotherapy. And in the not-too-distant future, breast MR may be pressed into service for image-guided therapy.

"We're only at the beginning of looking at the promise of breast MR," Newstead said. "In the future, we'll be able to get more functional data. We should be able to analyze the molecular level when we do spectroscopy and analyze breast cancers. Ten years from now, we will look back and be amazed at what we have found out about breast MR and cancer."

RISKY BUSINESS

There is little dispute that MRI is valuable in screening patients at high risk for breast cancer. But there is less certainty about what constitutes high risk.

"It's clear that in high-risk women, MR is able to pick up cancers that are not seen on other modalities, particularly mammography. But you can define high risk in many different ways," said Dr. Elizabeth Morris, director of breast MRI at Memorial Sloan-Kettering Cancer Center.

Obvious members of the high-risk group include women with a genetic predisposition for breast cancer who have tested positive for BRCA1 or BRCA2 mutations. In a large, U.K.-based prospective study published in May, annual screening of gene-positive women with a combined regimen of breast MR and mammography was most successful at detecting cancers. MRI was nearly twice as sensitive as mammography, but the pairing of the two yielded a 94% detection rate in the group of 650 women (Lancet 2005; 365:e-published).

The U.K. study confirms findings from two other prospective studies conducted in the Netherlands and Toronto over the past several years. Based on the strength of this and other research, the National Comprehensive Cancer Network, a consortium of 19 U.S. cancer centers, now advocates mammography and/or MR beginning at age 25 for BRCA1/2 women. The American Cancer Society may make a similar recommendation later this year.

Breast MR is undeterred by dense breasts and fibrocystic disease, conditions that undermine mammography's accuracy. That attribute is important in screening gene-positive patients, who are typically younger women, often with dense breasts.

At many sites around the U.S., a broader definition of high risk is applied for MR screening. The screening population at Sloan-Kettering includes patients with a personal history of cancer, a first-degree relative with breast cancer, and atypia and lobular carcinoma in situ, as well as gene-positive women. That policy evolved after a 2003 study (AJR 2003;181: 619-626) documented increased cancer detection by MR in the group, Morris said.

Similarly, Newstead sees value in breast MR screening for women with a family history of breast cancer and those with a personal history of breast cancer who are 35 or older. Post-cancer therapy patients whose initial cancer was mammographically occult are also screened with MR.

"We are not as strict with our definition of high risk. The only reason not to screen these patients, in my mind, is cost. It's not a medical reason," she said.

In time, MR breast screening may be customized to individual risk profiles, including women with increased risk of breast cancer and those who have already been diagnosed.

"Eventually, we will be performing breast MR differently," Newstead said. "We'll be tailoring it. There will probably be a screening set of techniques that we'll use to evaluate both breasts in a general way. Then we will have other techniques that we'll use to answer specific questions on lesions that we find."

NEW TECHNIQUES

The most common diagnostic application for breast MRI in the general population is evaluating patients with an abnormal mammogram and ultrasound examination. It's also helpful in assessing postsurgical patients to differentiate between scarring and cancer recurrence. With optimal techniques and top-of-the-line scanners, MR can achieve 100% sensitivity in detecting invasive breast cancers.

But specificity can range from 50% to 95%, due to wide variations in technique. As a result, MR has far to go before it can replace biopsy to determine the seriousness of suspicious lesions. In a study conducted by the International Breast MRI Consortium, researchers found that contrast-enhanced 3D dynamic MRI, the optimal breast MR technique to evaluate women with an abnormal mammogram, is still considered controversial. Moreover, breast MR's negative predictive value is too high to consider it as a biopsy alternative (JAMA 2004; 292[22]: 2735-2742).

Several new techniques in development are aimed at increasing specificity and, potentially, identifying which patients need biopsy and which do not.

One of these techniques, high temporal and spatial resolution (HiTS) MR, shows particular promise. Newstead presented research at the 2004 RSNA meeting showing that HiTS offers improved lesion visualization in patients with suspected breast cancer compared with standard breast MR.

Differentiating benign from malignant enhancement is one of the most difficult challenges in MR imaging, and HiTS may help out there, too. In a follow-up study presented at the American Roentgen Ray Society meeting in May, Newstead used HiTS to assess suspicious microcalcifications. The technique differentiated lesions from nonspecific parenchymal enhancement, provided clinically useful kinetic information, and clearly depicted morphology of rapidly enhancing lesions.

MR's sensitivity and specificity in small ductal carcinoma in situ lesions were high in this pilot study compared with published studies, she said. The findings suggest that HiTS MR could significantly improve detection of DCIS, long an area of weakness.

HiTS MR evaluates only the area of concern in the breast, not unlike a focused diagnostic mammogram performed when x-ray screening raises questions. The technique acquires pre- and postcontrast images in eight-second intervals. The experimental work has yielded valuable information about morphologic changes in DCIS lesions and how they enhance, Newstead said.

"Early literature suggested that DCIS is not well diagnosed by MR, but we do not believe that to be true," she said. "Those early results were primarily due to technical factors. Now we understand more about how DCIS enhances and how its morphology is different. And as a result, we are better able to recognize it."

The key is to image early, because with time DCIS blends in with normal breast tissue and is more difficult to find, she said.

Another experimental technique, high spectral and spatial resolution (HiSS) MR, is being applied to masses as well as calcifications. Like HiTS, the aim is to increase MR sensitivity and specificity for detecting very small early lesions, said Gregory Karczmar, Ph.D., an associate professor of radiology at the University of Chicago.

With HiSS MR, radiologists can access an entire dimension of spectral data by creating images based on water resonance. While the technique is 75% engineering, its potential is definitively clinical.

"The resulting images are very clean and crisp," Karczmar said. "Ultimately, I see HiSS being used to obtain a clear morphologic, anatomic image of the breast to show suspicious areas. Then the HiTS can be applied, obtaining higher temporal and spatial resolution in the regions of concern."

Compared with HiSS and HiTS, parallel imaging techniques are relatively mainstream, even though they have only recently been applied to the breast. Most are vendor-specific, but all involve a reduction of phase-encoding steps, which speeds image acquisition. Moreover, parallel techniques allow image acquisition in the sagittal plane, a key advantage, according to Newstead.

One such iteration of parallel imaging, sensitivity encoding (SENSE), resulted in images with higher temporal and spatial resolution compared with standard MR in a study conducted at Saint Barnabas Medical Center in Livingston, NJ (AJR 2005;184: 448-451). That led to better contrast uptake and lesion characterization. The shorter scanning times made possible with SENSE also improved patient comfort, generated increased throughput, and contributed to lower costs.

Image acquisition techniques are not the only factor in breast MR. Image processing also plays a role. In the Boca Raton study, morphologic and kinetic analysis was performed using dedicated postprocessing and display software. Breast MR demands 3D visualization in order to fully appreciate all of the collected data, Wiener said.

"Without such software, you might spend 45 minutes to an hour agonizing over one case and still possibly miss the most important information there," he said. "Software allows you to quickly appreciate kinetic information."

CANCER EVALUATION

With breast MR's ability to evaluate lesion size, metabolism, and kinetics, as well as tissue vascularity, it's no surprise that appraising response to neoadjuvant chemotherapy is another winning application.

Patients who undergo neoadjuvant chemotherapy are more likely to be treated with breast conservation surgery. The wide range of agents available for breast cancer treatment makes it critical to know as soon as possible if one regimen is not working and a swift shift to another is called for.

Published studies have shown correlations between MR and clinical findings, including changes in lesion size, neoplastic phenotype, dynamic contrast enhancement, and extraction flow product (Radiology 2004;233: 424-431). The clinical gold standard for assessing initial tumor size and response to therapy is physical exam.

In a recent study, researchers found that MRI not only provided more information about therapeutic response than mammography and ultrasound, but it also correlated better than physical exam with pathology (AJR 2005;184:868-877).

"Clinicians want a way to assess how patients are treated and whether treatments are working," said Dr. Eren Yeh, a radiologist at Massachusetts General Hospital and an instructor at Harvard Medical School. "We found that MRI is much better than physical exam in determining treatment response in this subset of patients."

MR may soon go one step further, providing image-guided therapy of early breast cancer, Wiener said.

"The potential exists for a whole new area of development with MR-guided therapy. Various types of devices and facilities are doing trials. The focus now is on gaining experience by destroying benign tumors such as fibroadenomas," he said.

STANDARDIZATION

As breast MR embraces new clinical horizons, standardization of routine imaging is woefully lacking. In late 2004, the American College of Radiology released standards for breast MR, and users also now have the Breast Imaging Reporting and Data System to guide them. But that is not enough.

"If you get a group of radiologists in a room and ask if they think breast MR is useful, they are all going to say yes," Wiener said. "If you then ask, 'When is it useful?', you are going to begin seeing some disagreement. If the next question is 'How do you perform it, how do you interpret it, and which tools do you use to interpret it?', you are going to see a huge disparity."

Compared with mammography, which has standards for everything from how images are acquired to how technologists are trained, MRI practice is still maturing, Yeh said.

"There are attempts to make breast MR more uniform, but it is a challenging technology. We are still trying to combine anatomic and morphologic information and enhancement curves," she said. "People are scanning differently in the U.S. from in Europe, from imaging one breast to both and acquiring data axially and sagittally. There is no uniformity-but it's coming."

Biopsy of MR-suspicious lesions is one aspect of breast MR operations that desperately needs a standard. Like most experienced imagers, Morris and Wiener believe that breast MR should not be performed unless the facility is equipped for MR-guided biopsies.

Boca Raton Community Hospital was one of the first sites in the U.S. to perform the procedure. The team uses MR-guided vacuum-assisted breast biopsy devices with a large-bore needle that permits substantial excision of tissue from the area of suspected abnormality. Wiener is compiling data on their biopsy experience to determine false-positive and false-negative rates with breast MR. Such data tracking is critical for a breast MR program, he said.

"If you are going to do breast MR, it's important to think about how you are going to do it and how effective you are going to be," he said. "Are you really going to help patients or just create a lot of unnecessary expense and anxiety?"

Ms. Dakins is feature editor of Diagnostic Imaging.