Computer-aided-detection
Reading screening mammograms is difficult. Just two to six cases of breast cancer are typically detected per thousand mammograms. Radiologists compare the task to “looking for a needle in a haystack.”
Radiologists would rather not make mistakes, but if they do, it seems they prefer making false positives to false negatives; they would rather erroneously decide a woman might have breast cancer than miss a breast cancer. I can think of two possible reasons for this.
One, they’re more likely to be sued for false negatives than false positives. Missed or late diagnoses of breast cancer are the leading cause of radiology malpractice lawsuits in the U.S.
Two, there seems to be a general bias in medicine to avoid misses. For example, the common wisdom amongst surgeons seems to be that if they haven’t operated a certain number times for a non-existent appendicitis, they’re not diagnosing appendicitis often enough--- a reasonable strategy since missing an appendicitis can be immediately catastrophic. Patients could die, not in months or years, but in the next few days if not sooner. Plus, the patients are sick, not well. They are usually in lots of pain. Although avoiding misses is a good strategy for imminent catastrophe and ill patients, it may not be in other cases.
Radiologists have strategies for avoiding mistakes. In clinics and departments that have a number of radiologists, the more accurate mammography readers will be the ones who screen mammograms. If previous mammograms are available, radiologists will use them to check for changes. And, in some places in the U.S., mammograms are read independently by two radiologists. If their readings differ, no decision is made until they agree. Of all the strategies, double-reading is the most accurate.
Computer-aided detection (CAD) was developed about eight years ago to improve the accuracy of a single radiologist. And, it does. When using CAD the accuracy of a lone radiologist almost equals that of double-reading. Its boost in accuracy however, is achieved primarily by reducing the number of false negatives. It was designed to “avoid false negatives without unduly increasing the number of false positives.”
CAD searches for suspicious- looking regions on mammograms, marking possible masses and micro-calcifications. It generates an average of one false mark per mammogram. That’s a lot considering that between 994 and 998 out of 1000 mammograms are likely to be normal. The radiologist considers each mark and accepts or rejects it. Given the bias against missing something, it’s more likely false CAD marks will be accepted, adding to the number of false positives.
And, there’s a specific protocol. First radiologists interpret a mammogram without CAD. They then review the same mammogram after it’s been marked by CAD. At this point they may change their interpretation---but not if it’s a false positive. If a radiologist initially suspected an abnormality and decided to recall a patient for further tests, the patient is recalled even if CAD didn’t mark anything. The protocol makes an increase in false positives almost impossible to avoid.
Finally, CAD is better at detecting micro-calcifications than masses. It finds very small abnormalities that may never lead to breast cancer or if biopsied, contain so few cells pathologists are unable to analyze them appropriately.
Recent research reported CAD significantly increased the number of false positives. Its greater ability to avoid missing a breast cancer does not benefit well women.
If you decide to be screened regularly, you can reduce your risk of being a false negative or a false positive by looking for radiology departments and clinics where the better mammography readers review screening mammograms or, even better, where screening mammograms are double-read. If neither is available you can ask for a second opinion---preferably from a radiologist at another facility. It might also be a good idea to get second opinions for any biopsies---again preferably from a pathologist at another facility.
When radiologists are unsure they will call women back in three or six months for repeat mammograms. There’s no reason you can’t request that yourself. If it appears that you may have DCIS or LCIS or a miniscule micro-calcification, you might suggest waiting to see if any changes occur before agreeing to a biopsy. And, if possible, be sure to be screened at the same facility every year. Your risk of being a false positive is cut in half when radiologists have past mammograms for comparison.
Ionizing Radiation
Screening presents a risk in addition to the risks due to radiologists’ errors. It’s ionizing radiation, a known carcinogen.
I would like you to imagine being in the following experiment. Your ears are plugged so that you can’t hear; your eyes are covered so that you can’t see. And, at a random time during the day you’re going to run across a highway. The highway has two lanes and little traffic. You avoid getting hit and try again the next day. Again you’re not hit and think you might be able to run blind and deaf across the highway every day for a month without being hit. If imagining this is making you begin to feel uncomfortable and sense the danger is getting too great, you’re right. The risk is thirty times as large.
The same logic applies to the risk of getting breast cancer from ionizing radiation. The risk of developing breast cancer from one mammogram is small, but the risk of developing breast cancer from 25 mammograms (the number a woman would have if she had an annual mammogram between the ages of 50 and 74) is 25 times as large. It’s not that the ionizing radiation is building up in your breast, but that each time you have a mammogram there’s a small risk cells will be damaged and become malignant. Since each time you have a mammogram that risk is repeated, being screened every other year (biennially) rather than annually will cut your risk in half. But what is the risk? Maybe it’s so small, you won’t care.
The dose of ionizing radiation from one mammogram may be as low as 0.30 or as high as 0.42 cGy (centigray). If a young woman has 10 mammograms and if the dose is 0.30 cGy, her total dose would equal about 3 cGy (10 x 0.3) and could increase her risk of getting breast cancer by 1.2% (Cornell University’s Program on Breast Cancer and Environmental Risks Fact Sheet #52, 2005). Keeping in mind that the older a woman gets the less susceptible her breast tissue is to ionizing radiation, if she has 25 mammograms, her risk increases by about 3%. That means three additional women out of a hundred over their lifetime will get breast cancer if they’re screened annually. But, if she’s screened biennially her risk would increase by 1.5%. Between one or two additional women out of a hundred over their lifetime will get breast cancer if they’re screened every other year.
Depending upon how you feel about taking those increased risks, you may decide to be screened annually, biennially, or not at all. In 2002, the U.S. Preventive Services Task Force found an increased risk of 3% acceptable; it no longer was in 2009. But, an increased risk of 1.5% was. However, sometimes an increased risk as small as 0.1% is said to be unacceptable.
One out of a thousand women (0.1%) will get endometrial cancer during their lifetime. If they take tamoxifen for a year, their risk is increased by 0.1% - two out of a thousand (0.2%) will get endometrial cancer. Tamoxifen and aromatase inhibitors (AIs) are two different types of anti-estrogen drugs. They’re often compared. AIs don’t increase one’s risk of developing endometrial cancer, but they have other side effects, one of which is significant bone loss. One study recently reported a 6.1% decline in bone mineral density in women taking one of the AIs vs.1.8% in women taking a placebo. Women are often warned tamoxifen doubles their risk of developing endometrial cancer, but they’re not told how small the risk is. If they knew, some might choose the 0.1% increased risk of getting endometrial cancer over the bone loss, especially if they had an idea of how painful and crippling osteoporosis can be.
Be wary. Don’t be satisfied with simply being told a dose is small or a risk is some number of times as great. You have to know the size of the risk and how often you’re going to be exposed to it to make an informed decision.
Next post: Dissecting the Task Force’s recommendations
