The breast cancer screening “industry” in the U.S. is sometimes pinned to a 6-8 billion dollar figure that is supposed to reflect high cost and low gain. That is, few lives saved for such an impressive price tag. And when compared to $6 billion, $2 million seems piddly. But it’s $2 million that has been awarded by the National Cancer Institute in a R-01 grant to my collaborators (“inventors”) and me to see if we can revolutionize how we screen for breast cancer.
First, some background. Here sits a breast MRI machine, less than 20 feet from my office. I have to consider, every day, that if we were able to screen the entire population with this device, very few women would ever die of breast cancer. So, while everyone waits for a “breakthrough” to radically alter the treatment of breast cancer, I’m looking at that breakthrough in the past tense. It’s already been done. Yet, many experts either ignore it or criticize it.
And for sure, it’s not the long term answer – the final answer will come when we are able to cure or control metastatic breast cancer 100% of the time. Or, alternatively, when we can immunize against cancer, so that the disease becomes an historical relic. Then, we won’t need to screen. Early diagnosis won’t be required. But that day is not yet on the horizon. For the next 50 years, at least, early diagnosis through screening will be important. And, we could do so much better than the status quo with the technology currently at our disposal. What a shame that it is nearly impossible to use. Strict guidelines and stricter insurance companies make it difficult to identify patients who qualify for MRI screening.
Currently, women qualify only when they are at very high risk for breast cancer. Sounds both obvious and justifiable, but in fact, it’s an inefficient approach piggy-backed onto inefficient baseline screening. First of all, it excludes the vast majority of women who are headed toward breast cancer, the 80% who do not have a family history. Then, under current guidelines, women qualify on the basis of “lifetime risk,” an unwieldy number that declines as you age, given that you are “passing through” your risk. So, young women with risks qualify more easily, but as time goes on, when individuals actually enter the danger zone where incidence peaks, their remaining lifetime risk may not allow them to be screened with MRI.
Risk-based screening is on the tip of everyone’s tongue, but for me, it leaves a bad taste even though my area of expertise is breast cancer risk assessment. Why? Because the difference in cancer yields between very high-risk women vs. normal-risk women is not great enough to warrant completely separate approaches. If you screen 100 women with MRI who have already had a normal mammogram, you will find 1 cancer in a population of “normal risk” women. If you screen 100 women at the very highest level of risk, e.g., BRCA-positive women, you will find 3 cancers. All that work to find 3 cancers instead of one.
My point is this: research should focus on how to identify women that have mammographically-occult breast cancer on the day of the negative mammogram, NOT using the surrogate of breast cancer risk spread out over a lifetime. For 20 years, my only idea to make this happen has been through a low cost screening blood test that would tell you to proceed with MRI if mammograms were negative. And, I am recently encouraged by new developments in this area, and will be writing about it more in the future.
But I’ve always been haunted by another fact, well-known to all breast radiologists, but seldom discussed. Often, when you diagnose breast cancer, you can look back one year earlier and see a subtle change in the density level in the area where cancer has recently been diagnosed. In fact, so many attorneys were taking advantage of this, and successfully swaying juries into rendering guilty verdicts against radiologists, that a group of experts wrote an article on the topic, admitting that 58% of the time, you can detect “something” happening in a zone where cancer will be diagnosed 1-2 years later. Yet, these changes are too minor to hold the radiologist accountable. If breast radiologists called back everyone with such subtle changes for diagnostic work-ups, then they would be calling back the majority of patients being screened.
The article was designed to address the absurd standard to which radiologists were being held by the courts, not simple perfection, but prescience beyond perfection. But when I first read the article, my “take home” message was entirely different. What if those women had undergone breast MRI? I suspect nearly 100% would have been positive for cancer. But again, you can’t do MRI on all those with such subtle changes. Or can you?
One year ago, a publication caught my eye in The Breast Journal, where an accomplished computer scientist (Bin Zheng, PhD) and his mentor (Hong Liu, PhD) had been working on image analysis through computers in the detection of subtle differences in the comparison of one breast to the other, and over time. Using their invention, they reported being able to find women at nearly 10-fold short-term risk for breast cancer based on mammographic density changes. This is not CAD (computer-assisted diagnosis), already in current use where specific lesions are identified by the computer. This is a “second line” computer analysis, an “ultra-CAD” if you must, identifying changes after routine CAD has “signed off” on normal mammograms.
It was a brilliant approach, and my mind raced back to the 58% who have subtle changes in the year(s) prior to diagnosis. This “ultraCAD” would serve the same purpose as a screening blood test, that is, in the efficient selection of patients for MRI, not based on future long-term risk, but based on the high probability of a current malignancy missed by mammography.
And to my surprise, these computer scientists were working out of an Advanced Cancer Imaging Lab located only 30 miles away at my alma mater, the University of Oklahoma, Norman campus, only a short walk to the basketball arena and a short jog to the football field. As always, basic scientists need clinical collaborators in order to lift their inventions from laboratory into actual practice. I contacted Dr. Zheng, and we ran a quick pilot study that included 30 patients with normal mammograms, but 5 of whom actually had cancer discovered on MRI. Dr. Zheng’s system identified 9 of 30 as “very high short-term risk,” and all 5 of the cancers were included in the 9. Had we used his system to select patients for MRI in the first place, we would have only performed 9 MRIs instead of 30 to find the cancers. This is efficiency. I won’t take up a lot of space here, describing what this means in terms of cancer yields on MRI, but in summary, if it works, 1) the cancer yields will dwarf anything ever accomplished through risk stratification, and 2) it will open up MRI screening to all women, not just the minority who have risk factors.
The National Cancer Institute seemed to agree that we may be onto to something. In July, 2015, the NCI awarded us over $2 million for a 5-year study that will involve approximately 10,000 breast images. In keeping with NIH policy, “Research reported in this publication was supported by the National Cancer Institute of the National Institutes of Health under Award Number R-01CA197150. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.”
Our grant title is: “Increasing Cancer Detection Yield Using Breast MRI Screening Modality.” Breast MRI can detect more than 90% of breast cancers at an early stage, twice the number detected by mammography in head-to-head comparisons. While molecular biologists whittle away at the secrets locked inside the cancer cell, going for the eventual cure, it’s well past time to take full advantage of the miracle of multi-modality imaging, something invented in the past, yet presently offered only to a select few.