Activity Transcript

Linda Moy, MD: Hello, I'm Linda Moy, professor and director of Breast MRI at the Department of Radiology at the NYU Grossman School of Medicine. Welcome to this program titled "Ultrafast, Abbreviated, and Conventional Breast MRI: Who and Why?"

Today I'm going to spend some time talking about the advancements we've made using MRI for breast cancer screening.

Let me begin by stating that magnetic resonance imaging (MRI) is the most sensitive imaging test for breast cancer detection. It finds cancers that are occult, meaning not detected on clinical breast exam and not detected on mammogram. Here's an example of a woman with a large breast cancer and an enlarged metastatic lymph node. This patient underwent neoadjuvant chemotherapy with a good response. The tumor shrunk significantly and the lymph node is now normal in size.

MRI is a very sensitive exam because it visualizes tumor angiogenesis. Tumor cells produce vascular endothelial growth factors, or VEGF, leading to increased density of microvessels to feed the tumor. These are leaky blood vessels that have increased permeability, resulting in preferential uptake of gadolinium by breast cancers.

Breast MRI, in my opinion, is the most effective screening tool, and my definition of an effective screening exam is a test that has high cancer detection and a low interval cancer rate. Studies have shown that MRI outperforms the combination of mammography and ultrasound in high-risk women.

In the EVA trial, a prospective multi-center study of 687 asymptomatic high-risk women, sensitivity of MRI was 93%, much higher than ultrasound at 37% and mammography at 33%. The ACRIN 6666 trial had included 2809 women who underwent 3 rounds of screening ultrasound and mammography. At year 3, they were offered a single breast MRI, and this MRI found 14.7 cancers per 1000. The take-home point for me is MRI finds cancers in this highly screened population that underwent annual mammography and ultrasound.

This is a schema showing you how we perform the MRI. We obtain pre-contrast images -- it can be diffusion or T1 volumetric interpolated breath-hold examination (VIBE). In the United States, we tend to obtain the T2 pre-contrast; in Europe, it's usually done at the end. Then you inject contrast and you can run ultrafast images, along with several post-contrast scans, through both breasts.

In other words, you obtain your pre-contrast images, inject, and obtain your post-contrast scans. And what you want is for the enhancing lesions to pop out. We call that lesion conspicuity, as you can see with a small white oval mass in the breast.

Since we're obtaining serial post-contrast images, we can generate time-signal intensity curves. There are 3 types. We have type 1 in blue. This is the persistent curve. Contrast is consistently picked up on the breast tissue, and that is persistent benign. Type 2 is going to be what we call the plateau indeterminate curve in green. It picks up contrast and then it plateaus off. Type 3 in red is the washout. We have rapid wash-in and washout of contrast.

What about cancer biology? Well, this is a retrospective review of over 7500 high-risk women. And of these women, we found 222 cancers diagnosed in 219 women. Three-quarters of these were found by magnetic resonance (MR).

What are the types of cancers that we find? I want you to focus on that x-axis. On the left side we have the more indolent cancers. These are the low- and intermediate-grade ductal carcinoma in situ (DCIS). On the right are the invasive cancers, both intermediate and high grade. What you're seeing with the dark green columns is that these invasive cancers are more likely to be detected by MRI. In fact, 71% of these cancers were high grade compared to those seen in mammography. A mammogram detects those in the red column, mainly calcifications seen on mammography.

Now let's talk about the clinical indications for a breast MRI. These include diagnostic. This is when a woman presents with an issue. They can be newly diagnosed with breast cancer and we're performing a preoperative MRI for an extent of disease evaluation. We also perform it in neoadjuvant chemotherapy. Here we're giving chemo upfront to try to shrink the tumor prior to surgery and now we are assessing treatment response. Some women may already have surgery but have positive margins, and we perform the MR to see if patients are still a candidate for breast conservation or if they're better served with a mastectomy. Infrequently, women will present with a lump in their armpit, it turns out to be a metastatic axillary lymph node, and were doing the MRI to search for the unknown primary cancer. We also perform MR to see if there's evidence of rupture of a silicone implant.

Screening MRI makes up the bulk of what most of the breast imaging centers in the US perform. This is screening of asymptomatic women who are usually at above-average risk.

This is the latest recommendation on screening of high-risk women from the American College of Radiology (ACR) breast commission. This was published recently, and then there's an update to the guidelines we published in 2018. We concluded that for women at higher than average risk of breast cancer, the ACR recommends annual digital mammography beginning between ages 25 and 40, and supplemental screening, depending on the risk, with breast MRI as a supplemental method of choice. For those women who qualify but cannot undergo breast MRI, contrast-enhanced mammography or ultrasound should be considered.

In 2007, the American Cancer Society published their screening breast MRI guidelines and classified women into 3 groups: high risk, intermediate risk, and low risk. These guidelines have not been updated, and in a high-risk group, the American Cancer Society states annual screening MRI is recommended. In the intermediate-risk group, there's no recommendation for or against screening MRI, and in the low-risk group annual screening MR is not recommended. Let's take a closer look.

These are criteria for women that are considered to be high risk, the BRCA 1 and BRCA 2 gene mutation carriers, as well as their first-degree relatives, or women that have had usually childhood lymphoma, had mantle radiation to a chest wall between ages 10 and 30, or they may carry other genetic mutations that are considered to be high risk -- TP53 gene mutations or PTEN -- and again, if they're also first-degree relatives. The bottom line is these women have a lifetime risk of greater than 20%.

The next group is the intermediate group, sometimes classified as a higher than average risk group. These women, their lifetime risk is 15% to 20% and a much larger group of women than those higher risk group. These women can have a history of atypia, including lobular carcinoma in situ (LCIS), atypical lobular hyperplasia (ALH), and atypical ductal hyperplasia (ADH). We can also have dense breast tissue. Also, they will carry genetic mutations considered to be moderate risk. This includes neurofibromatosis type 1, the PALB, and a woman that has a personal history of breast cancer, including ductal carcinoma in situ (DCIS).

Finally, the American Cancer Society recommends against screening MRI in what they call the low-risk women. Here, lifetime risk less than or equal to 15%. And they say that the benefits are not outweighed by the potential limitations, and therefore MRI screening is not recommended. This includes women that are at average risk and have fatty breasts. Here they state neither routine screening with MRI or ultrasound is appropriate.

In the US, there is a large grassroots movement to perform supplemental screening in women with dense breast tissue and a normal mammogram. But what is the data on performing MRI in woman with dense breasts? And do these studies meet the Breast Imaging-Reporting and Data System (BI-RADS®) benchmark? The earliest data on a screening in dense breast comes from these 3 studies on an abbreviated MR. The authors found that both their abbreviated MR and the full MR have a very high sensitivity, and the specificities meet the BI-RADS® benchmarks of 85% to 90%.

This brings me to the DENSE trial. In this Dutch trial, 40,000 women aged 50 to 75 with extremely dense breasts and a recent negative screening mammogram were randomized to undergo or not undergo a supplemental breast MRI. The authors found that the MRI group had significantly fewer interval cancers than mammography alone during the 2-year screening period. Let's look at the data.

The women were assigned in a 1:4 ratio to MRI screening group or to the mammogram only. The red box shows that 32,000 women were assigned to the mammogram only, and at 2 years follow-up, the interval cancer rate was 5 per 1000. The purple box shows that 59% of the 8000 participants in the MRI group actually had MR. They found 79 cancers, giving a cancer detection rate of 16.5 per 1000, interval cancer rate of 0.8. In women who declined MRI, in the green box, they had an interval cancer rate of 4.9, pretty close to the mammogram-only group of 5 per 1000. This DENSE trial was powered to estimate the interval cancer rate, and that 0.8 interval cancer rate is similar to the interval cancer rate in woman with fatty breasts.

However, randomized clinical trials need to include all women who were assigned to the MRI group. That means the interval cancer rate was 2.5 in the MRI group and met the BI-RADS® benchmark, and it was 5 in the mammogram-only group. So the PPV3, which is the positive predictive value (PPV) for biopsies, was 26.3 -- again, meeting the BI-RADS benchmark.

What kind of cancers did they find? Well, they're more likely to be small and to be DCIS. So, although they detected more cancers, it came with the risk of more benign biopsies.

The authors then published their second screening round last year. To refresh your memory, the first or prevalence screening MRI round cancer detection rate of 16.5 per 1000 and a false-positive rate of 79.8 per 1000. And of the close to 4000 women in the MRI group, 72% underwent a second MRI. Now the cancer detection rate was 5.8 per 1000 and a false-positive rate was 26.3. So, the cancer detection rate decreased in a normal rate that we see in a general population. This means we're detecting cancers earlier but you don't detect more cancers. The recall rate was 3%, better than mammography. The specificity was great.

So, what is this news that you can use? First, cancer detection rate is lower than a prevalent round, 5.8 vs 16.5 per screening, but still substantial. There was a small decrease in the invasive cancer detection rate and a number of false-positive results sharply decreased from 79.8 to 26.3. The PPVs were similar in both rounds and met BI-RADS benchmark. The take-home point is that supplemental MRI screening improves cancer detection also in the incident rounds, and the cancers that they found were mainly invasive cancers that were small, less than 10 mm.

The results of the DENSE trial led EUSOBI, which is the European Society of Breast Imaging, to publish this guideline in woman with extremely dense breast tissue. They now recommend supplemental screening for women age 50 to 70 at least every 4 years and preferably every 2 to 3 years. If MRI is unavailable, ultrasound and mammography may be an alternative. And of note, EUSOBI recommends shared decision-making.

Well, what about MRI in average-risk woman? There's only one study, a prospective study by Christiane Kuhl, published in Radiology in 2017. Here are about 2100 average-risk woman aged 40 to 70, and this means average risk, lifetime risk less than 15%. Data went close to 3900 MRIs. Cancer detection rate was 15.5, which were mainly invasive cancers. Interval cancer rate, 0%. It's one of my take-home points. MRI is an effective exam. And also, it's not about density.

It turns out MRI benefits all women.

You see this with this table that looks at MRI screening at average risk. We have all 4 types of breast density from ACR A down to D, and we find that the cancers detected were, again, small, 8 mm, node negative, high grade. These are the early cancers we want to detect.

Forty-one point seven percent (41.7%) of cancers were detected at prevalence screening, and 46% at incident screening. The specificity of MRI screening was 97.1%, false-positive rate of 2.9%. So they concluded that MRI screening improves detection of biologically relevant breast cancers in women at average risk, independent of mammographic breast density.

Now, let's transition and talk about abbreviated MR.

Given the compelling data I showed you, why isn't breast MRI more widely utilized? Well, it's expensive and time-consuming to perform. This table shows the cost of screening mammography, ultrasound, and MRI in the US. MR is much more expensive than mammography and ultrasound. Also, there's a limited number of MR magnets in the US and worldwide.

Healthcare policy makers use QALY, which is the quality of adjusted life years, to decide whether Medicare should cover certain exams. A QALY threshold value in the US ranges from 50,000 to 100,000. Applying this range, annual screening MRI in the younger, highest risk BRCA1 patients would likely be cost-effective, with the upper end at 70,000 per QALY. It's interesting that based on a cancer detection rate, a similar QALY should apply for woman with dense breasts and woman at average risk. However, both are based on a limited prospective trial, the DENSE trial, and Dr Kuhl's study.

The issue is, can patients tolerate the MR? Claustrophobia is a big issue: 25% of the ACRIN 6666 participants and 40% of the woman in the DENSE trial declined MR despite being able to undergo it for free. There's also concern about gadolinium. For me the take-home point is, for MR to be effective, it needs to be quick exam and we also need to be quick with our evaluations. So, here is the million-dollar question. Can we create a faster, more efficient, and cheaper MRI while preserving accuracy?

In this seminal study, Christiane Kuhl investigated weather an abbreviated protocol consisting of one pre- and one post-contrast subtracted maximum intensity projection (MIP) image was suitable for MRI screening. Prospective observational study, 443 women who are at mild to moderately increased risk, who underwent 606 screening MRs.

What did she find? Eleven cancers, 7 invasive or diagnosed, detection rate of 18.2. Again, no interval cancers were detected, and overall there was equivalent sensitivity and specificity between the abbreviated MR and the conventional MRI protocol.

They were able to read these exams very quickly. After looking at the MIP, or the maximum intensity projection image, it took them 2 seconds, reading the full protocol --30 seconds. I have to tell you, most of us cannot replicate those numbers. We can read much faster, but not quite that fast.

Importantly, table time, 3 minutes for the abbreviated protocol compared to 21 minutes for the conventional MRI. And we're finding the cancers we want to detect early. Abbreviated MR finds these cancers, it tends to be high nuclear grade, and Dr Kuhl says this might be the ideal screening tool because there's no breast compression and no radiation.

We recently looked at whether abbreviated MR can provide value. Twenty-one studies done in 8 countries, 4500 women, and these included all different types of abbreviated MR protocols. Bottom line, we found that the diagnostic accuracy of abbreviated MR was similar to that full conventional MR protocol. I'm going to show you the systematic review and meta-analysis. Most people groan when they see this -- there's lots of numbers and statistics. However, it is these types of analyses that are used to see if a test truly works, whether it's no longer a research tool, and whether we can bill for it.

The prime objective was to compare the diagnostic accuracy of breast cancer detection of abbreviated MR or the full protocol MR in a screening setting. Only 7 studies met the qualifications to be in in this meta-analysis. Note the heterogeneous inclusion criteria: high-risk women, women with dense breast tissue, or women with a personal history of breast cancer. Look at the heterogeneous imaging protocols; and to date, there isn't a single standardized abbreviated MR protocol.

This is a forest plot showing that sensitivity and specificity for abbreviated MR. Now these values did not differ significantly between studies that had involved high-risk patients or those that did not. The high specificity is a big strength. However, there's a lack of precision with the sensitivity. You can see those horizontal lines are associated with each study that represents a wide confidence interval, and that's due to a small number of women and a small number of cancers in each of these studies.

There were 3 studies that had 1450 women who presented with follow-up data. Again, similar take-home point: abbreviated MR and a full protocol had a similar sensitivity and specificity. However, the authors concluded there was a very low level of certainty in that the abbreviated MR exam has a similar level of diagnostic accuracy to the conventional MR.

Why is that? Those issues of bias. The full MR protocol was interpreted directly after viewing the abbreviated MR protocol and not all the lesions were biopsied. Also, one of the goals was to compare the diagnostic accuracy of abbreviated MR with our routine mammography exams. Those include digital mammography, tomosynthesis, and contrast-enhanced mammography. There was insufficient data to look at that.

That leads me to this ECOG-ACRIN trial that was published recently. It's EA1141, which is a prospective study comparing invasive cancer detection by abbreviated MR compared with tomosynthesis in 1144 women. These women each underwent 2 abbreviated MR and 2 tomosynthesis studies, and these exams were performed in a randomized order. So, abbreviated MR and tomo 1 year and repeated again in year 2.

What they found was abbreviated MR had a significantly higher rate of invasive breast cancer detection: 11.8 vs 4.8 compared to tomosynthesis. And so these results broadly agree with that systematic review and meta-analysis I just showed you. In total, 22 cancers were detected, mostly by MRI. Abbreviated MR, though, did have a lower specificity and PPV compared with tomosynthesis. So with MR, we are working up a lot more enhancing lesions that turn out to be benign. So abbreviated MR, just like conventional MR, finds these biologic significant cancers, these include intermediate-grade and high-grade cancers.

Let's pause for a minute. With the pros of abbreviated MR -- certainly more comfortable for the patient, and this can be wider availability of breast MRI. If you have a short exam it could be cheaper, and as a radiologist I can read these studies faster. However, I've already showed you the drawbacks -- the lower specificity and the lower PPV means we're doing more biopsies, and right now we are waiting for the cost-effective analysis of this ECOG-ACRIN study.

What is Achilles heel? Well, with abbreviated MR, we're only getting morphologic information. Those time-signal intensity curves I showed you earlier, they're not available because only one time point is included.

So, what if we performed ultrafast imaging on the initial phase of enhancement in the first 2 minutes after contrast injection? Here we're going to scan quickly, that's what we mean by high temporal resolution, of both breasts during this initial period. We looked at this earlier and we found that during that early enhancement, those that have the fast or high initial enhancement ratio, IER, turns out to be those cancers that are higher grade and have a higher Ki-67. They also have a higher lesion conspicuity, meaning they pop out. So, the take-home point is the early enhancement is seen in bad prognosis cancers less often than seen in false-positive lesions.

This is a complex chart showing you this literature is still developing, so there are many ways of performing ultrafast protocol. What we do at New York University (NYU) is looking at the time of initial enhancement. That means, what is the time between when you see contrast in the aorta and when you see a mass that's enhancing in the breast? We found this to be easiest, and for us, the time to initial enhancement cutoff is during the first 15 seconds that usually separates cancers from the benign lesions. Let me walk you through this.

First, ultrafast sequences can be obtained without lengthening the protocol. Usually we're not scanning these patients as we're waiting for the contrast to reach the breast. Ultrafast allows us to maintain a high diagnostic accuracy. So, on your top left, A, you only see the pulmonary artery enhancing; in B, it's reached the aorta; and in C, the tumor now and the right breast are starting to enhance. And you see that clearly on subsequent images. By point E, that cancer is clearly enhancing. You also see that some of the adjacent skin is enhancing, and the skin was found involved at surgery.

So, time to enhancement -- again, what is that? That defines as time points between when the lesion started to enhance minus the time point where the aorta started to enhance. In this case, the cancer enhanced in the first frame. That's within 4 seconds. Remember I told you the cutoff is less than 15 seconds. In the far right you can see what the cancer looks like on the routine post-contrast images.

This is the NYU ultrafast protocol. We do pre-contrast, then we go ahead and run the ultrafast images, followed by the routine post-contrast images. There's another way of looking at the abbreviated protocol, and we include the ultrafast. Our total scan time is about 14 minutes, compared to the usual conventional MR, about 30 minutes.

Here are a couple of cases to show you. A 44-year-old woman with invasive lobular cancer, and going there from A through B I'm showing, on A, when you see the aorta is enhancing, you can see on the top row, at B, that cancer is enhancing at 4 seconds -- should be suspicious. In the bottom row here, you can see these other little smaller dots enhancing out to point D. Those are probably going to be benign. That was shown at surgery to be benign.

Another example: 45-year-old woman, invasive lobular cancer. On the top and the center, label B, the cancer is enhancing at 5 seconds. Contralateral breast, you're seeing enhancement at 15 seconds. And you see it get larger on the bottom row. That was benign fat necrosis.

This is just to show you ultrafast can work beyond screening MR. This is showing you neoadjuvant chemotherapy. And here, these authors used the wash-in slope -- it just means a steeper uptake of contrast -- and that was found to predict pathologic response in these patients.

This is a summary slide showing you, right now, what are indications for abbreviated MR, ultrafast, and conventional MR. Take-home point: conventional MR is clearly the standard of care, very limited data on ultrafast, and right now I would say that data is showing that abbreviated MR works just as well as conventional MR in women at intermediate risk and at average risk.

However, there's lots of challenges to clinical implantation of abbreviated MR. We certainly need more data, standardization protocol, and we need reimbursement for this study. However, in this era of value-based care, abbreviated MR is here to stay. It's performed in many different body parts.

I would conclude by saying that we can create a faster, more efficient, and cheaper MRI while preserving accuracy with the abbreviated MR. My take-home points are the following. Breast MRI is the most effective tool to detect breast cancer in a diagnostic and screening setting. The abbreviated MR screening setting has a similar diagnostic accuracy compared to conventional MR. Ultrafast can be added to conventional MR or abbreviated MR to increase the specificity of the MR.

Thank you for participating in this activity. Please continue on to answer the questions that follow and complete the evaluation.

This transcript has not been copyedited.