Judy Dean, MD

This is the second of a two-part series. Last week, Dr. Dean 

At RSNA 2011 there were dozens of posters and presentations focused on breast density and the use of breast ultrasound to improve the early detection of cancer. A number of large studies, including ACRIN 6666, continue to bolster the scientific platform validating the use of breast ultrasound to find mammographically occult cancers. So, with published studies from some of the nation’s leading breast imaging experts, why hasn’t breast ultrasound screening been incorporated into routine clinical practice?

In their paper, “The CT Experiment: The Role of Ultrasound Screening in Dense Breasts,” Weigert and Steenberger conducted a retrospective chart review of five Connecticut radiology practices with more than 50,000 mammograms and more than 8,000 screening ultrasounds where approximately half received follow up breast ultrasound exams. While the results clearly demonstrated that breast ultrasound is effective in finding cancers missed by mammography, the researchers raised practical questions about the implementation of breast ultrasound screening, such as scheduling, technology factors, physician supervision, as well as protocols for determining true findings versus false positives. 

clinical benefits of implementing an automated breast ultrasound screening (ABUS) program

, namely the goal of reducing interval cancers being missed by using mammography alone in women with dense breast tissue. Here, I’ll examine real and perceived barriers to screening with ABUS, including clinical and technical workflow, training, and density notification.

Barrier: Breast Ultrasound Takes Too Much Time 

Few radiologists argue that breast ultrasound has clinical benefits, particularly when focused on women with dense breasts. Studies presented at RSNA demonstrated that ABUS increased the yield of invasive cancers found, and that the average size of the cancers found was smaller. Despite these results, many radiologists are concerned that breast ultrasound screening will take too much time.  

Chart source: Kelly, KM, Dean, J, Comulada WS, Sung-Jae L: "Breast Cancer Detection Using Automated Whole Breast Ultrasound and Mammography in Radiographically Dense Breasts," Eur Radiol (2010) 20: 734-742

Ultrasound is a proven breast imaging tool that has been used for diagnostic purposes for years. However, handheld ultrasound (HHUS) is operator-dependent, which can result in variable quality and inconsistent exams. Additionally, it is difficult to consider HHUS true screening as I believe the patient influences acquisition and the study quickly shifts from detection-based screening to a diagnostic exam. It is natural to be distracted when the patient is watching with intense interest as you interrogate specific areas. 

From a time perspective, physician-performed HHUS exams averaged 20 minutes in the ACRIN study. In routine practice, it is probably closer to 20 to 30 minutes. Using trained non-physician personnel that understand patient positioning and the anatomy of the breast as seen on ultrasound, ABUS has the potential to dramatically improve technical workflow.

In our practice, acquisition of a normal ABUS exam takes about 15 minutes and we book ABUS exams in 30-minute increments. This gives us ample time to prep the patient, set up the exam room, conduct the ABUS exam, and clean up to prepare the room for the next patient.

Barrier: I Don’t Know How to Interpret ABUS Exams

I have read more than 8,000 automated breast ultrasound exams since 2005 and would consider myself to be an experienced reader. However, when I recently served as a reader in the U-Systems ROC Reader Study, I participated in a training program designed to help accurately read ABUS exams faster.

Users start with a self study component to get oriented to ABUS images and software, which is followed by a 90-minute peer-to-peer web conference where trainees review cases with experienced ABUS experts. Finally, users attend an eight hour peer-to-peer workshop where they read 100 cases and get feedback from an expert. The results of this training program have been that users accurately read ABUS cases faster with repeatability, reproducibility and consistent outcomes.

In addition to improving the accuracy of the interpretations, clinical workflow can show tremendous improvement as a result of a training program. Before the peer-to-peer workshop, the average read time was about 15 minutes per case for radiologists in the U-Systems Reader Study. After the training, average review time decreased to about three minutes for a normal somo•v® ABUS case.

Additional breast ultrasound courses that include Sonocine and U-Systems’ somo•v® are available, and I highly recommend attendance at a formal course before starting an ABUS program.

Barrier: My Practice Can’t Handle ABUS Clinical Workflow 

Although three minutes is longer than the average one- to two-minute read of a normal handheld breast ultrasound, with the ABUS exam the radiologist is able to review the entire data set, not just the few images selected for review by the technologist. This provides the ability to interrogate suspicious areas or quickly rule out normal findings, resulting in greater diagnostic confidence. 

Additionally, since the entire data set is available, ABUS exams can be batch read without the need to hold the patient in case the radiologist needs additional views or images not acquired by the technologist. In HHUS, the technologist essentially performs a diagnostic process while conducting the study and selecting sample images for review. You, as a radiologist must decide if you’re comfortable with this process. If your technologist does not recognize and document the abnormality, you will never know it is there.

Now that we’ve established that ABUS improves technical workflow and provides consistent, reliable coverage of breast tissue for improved image quality and diagnostic confidence, how do you implement a breast ultrasound screening program without disrupting your workflow and productivity?

As with any new modality or system, you should expect a learning curve as you start offering automated breast ultrasound exams. Initially, you will question more findings than you would on HHUS and your recall rate will be higher as you learn to rule out incidental findings, such as edge artifacts or nipple shadows. You can accelerate your learning curve by correlating immediately with HHUS for the first few weeks. To facilitate this, review the ABUS exams before patients leave, and allow enough time for HHUS in case you want to verify your findings.    

As I mentioned in the first article, I don’t recommend a 

biopsy on any screening study without a work up

, whether it is a mammogram or ultrasound. If findings from screening studies are carefully worked up the PPV should be the same no matter how you initially identified the lesion. While recall rates for screening ABUS may always be higher than mammography, the increase is proportional to the additional cancers found.

Educating Patients and Referring Physicians 

You may experience cost resistance from patients at first as there is no formal reimbursement for ABUS exams. However, as you begin to identify cancers with ABUS that were missed on mammography, word will spread. Since 2005, we have identified more than 30 patients with mammographically occult cancers using ABUS, and we have found that women are willing to pay for care when they understand the value.

Physicians have been very accepting of ABUS studies, as I make a point to personally call referring physicians when mammographically occult cancers are detected in their patients using ultrasound. I also am open about presenting these cases at tumor boards, when appropriate and with the patients’ permission, and I list on pathology referral forms “ultrasound-detected mass” when appropriate, just as I would list “suspicious microcalcifications” when a stereotactic biopsy specimen is submitted.

Barrier: Referring Physicians Will Be Upset if We Talk to Women about Density

With the goal of empowering women with the information they need to better protect their health, we have notified patients about breast density since 2005. In nearly eight years we have not received a single complaint from referring physicians about informing patients about breast density. Despite the widely publicized political opposition to Density Inform legislation in California, we have experienced nothing but strong support from our referring physicians. In fact, we routinely receive referrals for ABUS exams, sometimes for women who have not yet had a mammogram or who do not have dense tissue. In these situations we find ourselves explaining why they 

need an ABUS study if their mammogram does not show dense breast tissue.

Patient results letters from my office include density information and any woman with a BIRADS density rating of 3 or 4 is informed that new research suggests they might benefit from an ABUS exam and to consult with their doctor. Women with benign findings and high density are flagged in physicians’ reports with the suggestion that an ABUS exam might provide clinical value.

All patient recall information is recorded in the practice database - for both mammography and ABUS. This ensures that the receptionist sees the ABUS component when scheduling the next year’s exam. This results in fewer “lost” exams and reduces the likelihood the women will need to return for a second appointment. 

Barrier: My Staff Does Not Want to Sell ABUS Exams 

Technologists are one of the best resources to help patients understand how and why an ABUS exam might be appropriate for them. Some techs are initially resistant to talk to patients as they “don’t want to be sales people.” However, they quickly come around when they realize that what they’re really doing is helping find early cancers with ABUS.

We have all experienced the heartbreak of a regular screening patient returning with a lump that turns out to be an interval cancer. My technologists understand that informing patients about breast density and the option of additional screening with ABUS can help stop that from happening. They are very comfortable showing patients the difference between fatty breasts and dense breasts, and explaining to them what their density is.

Healthcare is moving closer and closer to personalized medicine as opposed to a one-size-fits-all approach. Mammography remains an important breast cancer screening tool, effective for all women in the appropriate age range. However, for at least 40 percent of women who have dense breast tissue, it should be viewed as a gateway or entry point to the screening process. For these women, it is not appropriate to stop at mammography as dense breast tissue may be hiding a cancer that we can find on ABUS.



Judy Dean, MD, is a diagnostic radiologist with more than 20 years of experience in women’s imaging.  She completed an ultrasound and body imaging fellowship at Stanford University, and medical school and residency at Loma Linda University.

Articles

Why hasn’t breast ultrasound screening been incorporated into routine clinical practice? In this second part of a series, I examine real and perceived barriers to screening with ABUS, including clinical and technical workflow, training, and density notification.

Incorporating Automated Breast Ultrasound Screening into Routine Clinical Practice

This is the first of a two-part series. Next week, Dr. Dean explores incorporating ABUS into clinical practice.

Breast density has been well established as a risk factor for cancer, increasing a woman’s risk by as much as four to six times. Numerous studies have also demonstrated that increased breast density makes cancer more difficult to detect with mammography. By masking the conspicuity of tumors, as breast density goes up, the accuracy of mammography goes down.

Breast tissue density has also been shown to be a major determinant of interval cancer, which is defined as cancer that becomes apparent clinically, usually because of a lump, between routine mammographic screenings. Retrospective studies have shown that 15 percent to 30 percent of interval cancers appear on the original screening mammogram that was initially interpreted as negative. The other 70 percent to 85 percent are not seen mammographically, even in retrospect.

 showed that all interval cancers (true interval cancers and missed cancers) were diagnosed at a higher stage and grade than screen detected cancer. These findings continue to suggest a need for more sensitive modalities and revised screening protocols to improve the sensitivity and efficacy of breast cancer screening.

With demonstrated value as an adjunct to screening mammography, I believe automated breast ultrasound (ABUS) can be the modality that improves screening for women with dense breasts. I believe ABUS can be used to reduce interval cancers in women with dense breast tissue now being missed by relying on screening mammography only.

In my Santa Barbara, Calif. practice, we have been offering ABUS since 2005 and have performed more than 8,000 ABUS exams on more than 3,000 women. As part of my experience with ABUS, I have participated in a prospective trial with SonoCine and more recently served as a reader in a pivotal ROC Reader Study for U-Systems.

At RSNA, I presented updated results to the 2010 

Breast cancer detection using automated whole breast ultrasound and mammography in radiographically dense breasts

.” The update included 3,778 digital mammograms and automated breast ultrasound studies; all women had dense breast tissue (BIRADS 3 or 4 density score) or a high risk of breast cancer. Results showed that ABUS found 13 additional cancers, an increased yield of 76 percent over mammography alone. All the additional cancers found by automated breast ultrasound were invasive, and the average size of invasive cancers found was 9 mm with ABUS compared to 14 mm with digital mammography.

Since we have been routinely offering ABUS exams, the number of interval cancers among screened women has dropped dramatically. We are finding smaller cancers, potentially a year or more before they would be palpable or visible on mammography. Earlier detection means different treatment options and potentially better outcomes. Before we began offering ABUS, interval cancers were not unusual in my practice, occurring 10 to 15 times every year. In 2011 there was just one interval cancer in patients with prior ABUS within 16 months. These results, if extrapolated over the dense breast patient population, could have a tremendous impact on effective breast cancer screening and overall breast health.

The most common reason cited by physicians for not accepting ultrasound screening for breast cancer is the low Positive Predictive Value (PPV) of biopsy shown in the ACRIN 6666 study. But this trial involved a very high risk cancer population with a cancer rate of 12.6/1000. If a lesion was found in a woman in this very high risk group it would be difficult to justify short-term follow-up instead of biopsy.

To avoid low PPV using ABUS, the first step is to correlate the exam to the original mammogram. If a possible abnormality is identified the next step is to use focused breast ultrasound with color Doppler, to determine if there is a lesion, and to characterize the lesion using a systematic approach like the ACR BIRADS™ characterization method. If you are not familiar with breast ultrasound, there are a number of terrific courses available, and you should enroll in one before starting an ultrasound screening program.

In my practice we treat ABUS findings in the same way we treat screening mammography findings. The patient is recalled for the appropriate imaging to clarify whether there is cause for concern or not. That may mean magnification views, focused breast ultrasound, Color flow Doppler, or other imaging. Biopsy is never recommended on the basis of the initial ABUS finding alone. Therefore my PPV for lesions first identified by ultrasound and lesions first identified by mammography is similar, since the decision whether to biopsy is based on the work-up and not the initial screening. I was not a participant in the ACRIN 6666 trial but I think it may have blurred the line between screening and diagnostic ultrasound since these were hand-held, radiologist performed ultrasounds.

The radiologist also has to be able to confidently identify benign ultrasound findings like intramammary lymph nodes and fibroadenoma, or the biopsy rate will be unacceptably high.

The Balancing Act: Anxiety from False Positives or Missing the Opportunity to Prevent a Delayed Diagnosis of Breast Cancer 

The use of adjunctive screening tools in breast cancer is frequently criticized - not for lack of sensitivity, but for the potential anxiety created by false positives. This concern is one of the most frequently cited objections to legislative efforts seeking to standardize the communication of breast density to women. However, what this argument fails to consider is the far greater stress and anxiety created by delayed diagnosis. The personal pain, cost and grief generated by a late diagnosis of breast cancer cannot be overstated. ABUS gives us the tool we need to prevent a great many delayed diagnoses.

While imaging tests detect suspicious areas, only a biopsy can provide a definitive diagnosis. Biopsies are an integral step in the diagnostic process and I do not hesitate to recommend a biopsy if the imaging results warrant it. At the same time, I wouldn’t recommend a biopsy on any screening study without a work up, whether it is a mammogram or an ultrasound. In a true screening scenario, when you work up each case, the PPV should be the same no matter how you first identified the lesion.

A practice utilizing ABUS will have significantly higher recall rates at the outset until the readers gain experience differentiating normal and benign findings, and identifying artifacts. Patient anxiety can be greatly reduced during this learning curve by initially having patients wait while ABUS studies are reviewed, and allowing time on the schedule for hand-held ultrasound if the radiologist feels it is needed. While recall rates for screening ABUS may always be higher than mammography, the increase is proportional to the additional cancers found.

There is a great deal of discussion about the classification of breast density and what BIRADS density score warrants the use of ABUS. While it is intuitive that there are more cancers missed by mammography the higher the density, in fact we have found occult cancers in density categories 2, 3, and 4.

This is where breast imagers may have to think differently than we’re used to. Rather than deciding for or against an ABUS recommendation solely on the BIRADS density score I also include the patient’s risk profile in my decision-making. From a protocol perspective, if I have a high risk patient with BIRADS 2 density rating, I discuss the surveillance options and let her decide. However, I recommend an ABUS exam for all patients with BIRADS 3 or 4 ratings. This is obviously an area where additional research will be of benefit, but the fact that a radiologist’s judgment of the density score may vary is not a reason to withhold density information or ABUS studies from women. Whether or not an ABUS is warranted is a judgment call, and radiologists are the best qualified physicians I know to make that judgment.

When we’re talking about reducing or eliminating interval cancers, we’re really talking about down-staging the cancers we find. In our 3,778 case study, we found 13 additional cancers using ABUS - all invasive, all node negative, and 12 of 13 were T1 (under 20 mm). These are early stage cancers with excellent prognoses and potentially less treatment required. Besides dramatically reducing the medical and financial impact of the cancer, we can’t ignore the social and emotional aspects of an early versus late stage cancer on women and their families.

I expect that the adjunctive use of automated breast ultrasound will have a huge impact on overall breast health. I believe it is the right thing to do for women with dense breasts.

Left: Retro areolar irregular mass seen in the coronal and transverse views. Right: Spiculated mass seen in the coronal and transverse views. 

Multicyctic breast tissue seen in the coronal, transverse and sagital views. 

Judy Dean, MD, is a diagnostic radiologist in Santa Barbara, Calif., with more than 20 years of experience in women’s imaging. She completed an ultrasound and body imaging fellowship at Stanford University, and medical school and residency at Loma Linda University.

With demonstrated value as an adjunct to screening mammography, I believe ABUS can be the modality that improves screening for women with dense breasts. I believe ABUS can be used to reduce interval cancers in women with dense breast tissue now being missed by relying on screening mammography only.

Author | Judy Dean, MD

Articles

Incorporating Automated Breast Ultrasound Screening into Routine Clinical Practice

Using Automated Breast Ultrasound to Reduce or Eliminate Interval Cancers