In Connecticut you’ve been doing some extremely interesting things because you’ve particularly been looking at women with dense breasts. Can you tell me what it is that you did?
A little bit of background to this is very important. In 2009 Connecticut became the first state in our country to sign landmark legislation mandating that we tell patients what their breast density is, whether it’s low density or high density, but we had to include in our plain language report a sentence that was mandated by this legislation saying that you have breast density of x, if you have density over 50% you may be at greater risk of developing cancer or that your mammogram may not find the cancer and that adjunct imaging such as ultrasound may be available to you, talk to your doctor.
So what have you been doing about that then?
As soon as that happened, obviously it was the law, all of us in the state had to add that piece of paper, add that information to our plain language reports, those plain language reports that have been mandated by the MQSA for many years, giving the patients if they have a normal or abnormal mammogram. But this adds this little bit of information to them. For the most part in the beginning we basically told patients, “Go talk to your doctors. See if you think you would like the adjunct imaging and if you do we will schedule that for you.”
Now you’ve conducted what’s called the Connecticut Experiment, you’ve done four years of screening now with ultrasound, what did you find?
What we found, but I do want to do just a little bit of background on this, that in the very beginning when this all came out I, like most radiologists, was against this legislation. I felt that we were going to be at risk of finding what I like to call too many ditzels, too many things that we don’t need to find. We already know that mammography is an imperfect test but we do find cancers. In the screening population we find between 3-4 new cancers per year per thousand patients. The question is would we find more cancers, statistically significant, if we added this additional test to this patient population. At the time when this legislation came out there was only one study, the ACRIN 666 study that had shown that indeed in patients who had high risk, either family history or other risk factors, they could find additional cancers. But, that being said, we didn’t know if just a normal patient population with no risk factors, other than dense breasts, we would find additional cancers. So I thought this was fertile ground to do this research and I was able to engage a medical student from our university; I do not have medical students at my disposal, I’m a clinical radiologist, but we do have a local medical school right around the corner from where I practise and she was very excited about working with me. We gathered data from around the state, twelve sites, over 8,000 screening ultrasounds in the first year and we found 3.2 per thousand additional cancers. That is statistically significant.
So you found cancers which would have been missed?
Yes, these were all negative mammograms. However, the reverse side was that our positive predictive value was only about 6% which means we were biopsying an enormous number of negative findings. That’s disturbing but, that being said, it was only the first year. We knew that there was going to be a learning curve to adding this procedure and that is what has been so interesting about following it for four years. I went back, the second year of data we found similar numbers. I then went to my own practice and just gleaned the data just from my five sites out of those twelve sites for years 1 and 2 and then looked at years 3 and 4. Because my staff was very good at keeping records for me and helping me with that, I was able to do that fairly simply. We found that years 3 and 4 we continued to find 3.2 per thousand additional cancers but by year 4 our positive predictive value had increased to almost 18%. 20-30% is the positive predictive value from mammography, all mammography, low dense, high dense. So this was clearly the jump we needed to make.
It’s looking very interesting, then, for bilateral ultrasound. What are the clinical implications, do you think, coming out of this with the four years of data?
Clinically it indicates that this is a viable test to do. It is simple, it is low cost, it has no radiation and now there are developing tools to actually make it easier for us. Right now our ultrasonographers do it, it’s very labour intensive for them. We can only do six or seven per ultrasonographer a day because it’s tiring, both physically and intellectually because they’re the first line to look to see if there’s something abnormal. Then they call me or one of my breast colleagues in to review the study. If we have automated breast ultrasound that level will be taken away from them and then we are the people that will review, in some sort of a cineloop I think, all the breast ultrasound images and decide if there’s anything there.
Does this mean that you could potentially use ultrasound instead of mammography?
No. We must do the mammogram first because the mammogram, number one, it’s our road map. We look for changes from year to year, subtle changes, whether there’s a mass, whether there’s an asymmetry, whether there are calcifications. It would be inappropriate to do the ultrasound in a vacuum separate from that.
What sort of evidence do you have so far, it obviously is early days, about the clinical benefits of detecting these lesions?
What we found was that over the four year period the size of the lesions in patients, particularly those patients that are coming back now year after year, because they are coming back, that the cancers that we’re finding that were not there the year before are very small, less than 1cm in size. Regardless of the type of cancer, whether it’s a lobular cancer, an invasive ductal cancer, grade 2, grade 3, they are node negative. Clearly that is the type of cancer we want to find because those are the ones that will be easier to treat and presumably if we already know that screening mammography has reduced mortality of cancer by finding cancers over time, one has to intellectually presume that these additional three cancers per thousand of this type of cancer, small, node negative, will add to that mortality reduction but it’s going to take more time.
Is this helping you, do you think, to personalise detection and therefore personalise treatment?
I think that’s absolutely a very good question and, yes, I believe that what we’re coming up with is a new paradigm of risk assessment and that different patients should have different types of screening depending on what their risk is. Low risk patients who do not have dense breasts, no significant family history, can they still develop a breast cancer? Absolutely, it should not be stopped. I believe that women over the age of 40 should have routine mammography every year. We find cancers that were not there the year before. On patients that do not have dense breasts that’s really easy – the mammogram. We see something new and different, we work it up. Patients who are very high risk, BRCA patients, greater than what the American Cancer Society had, the 20% increase in breast risk, cancer risk, should have a mammogram and then perhaps an MRI. But there’s this huge portion of patients that fall in the middle. 40% of women have dense breasts. Of those 40% most of them don’t have any other risk factors but should we be getting additional imaging for those patients? Yes, and I think the ultrasound is the easiest, most cost effective approach.
That’s a fantastic summary. Could you distill it into a few seconds, a few sentences, what the take home messages for doctors ought to be right now?
We’re on the cusp of a new paradigm of screening for breast cancer and that this paradigm will be based on risk assessment and this risk assessment will then decide how we routinely treat these patients with what technology we have, and we have so much technology we have to use the right technology for the right patient.