Strange Glow

by Timothy J Jorgensen

  Teresa started having regular mammograms when she turned 40 and has already had one false-positive result—a shadow on an image that turned out to be nothing. Nevertheless, the positive mammogram, the follow-up procedures, and the consult with the breast surgeon, had taken their toll on her emotionally. Intellectually, she knows that the majority of positive mammograms turn out not to be cancer (i.e., false-positive results), but knowing that doesn’t make going through the experience any easier. Despite it all, when she’s due, she dutifully musters her courage, submits to yet another mammogram, and holds her breath until the results come in.

  Teresa’s understanding of the mammography’s high false-positive rate is actually better than many doctors’. When 160 gynecologists were asked at a Continuing Medical Education meeting what the chances were of a 50-year-old woman actually having breast cancer if her mammogram came back positive, 60% of them said either 8 or 9 out of 10.3 The truth is that the odds the woman actually has cancer are only 1 in 10. Before you conclude that gynecologists don’t know much about mammograms, you should know that radiologists don’t do much better, consistently overestimating their ability to identify cancer on a mammogram. Some radiologists, practicing for decades, fail to notice that most of their patients with a positive mammogram do not have breast cancer.4

  To add to all her other concerns, Teresa now worries that the radiation from her mammograms has actually increased her chances of getting breast cancer, making it even more important that she keep up her screening regimen. As her age-related breast cancer risk increases with each birthday, so do the cumulative risks of her multiple mammograms. To beat it all, now she hears that mammograms haven’t lived up to their expectations in terms of saving lives. Ugh!

  Teresa’s situation is indeed more complicated than Matthew’s. But on the risk side of the equation, determining the risk of a breast x-ray is no more difficult than determining the risk of an arm x-ray. A quick Internet search tells us that the effective dose for a typical mammogram for a woman of average breast size is about 0.5 mSv.5 As before, we simply multiply this effective dose by the cancer risk per mSv to get the overall cancer risk associated with the procedure:

  0.5 mSv (effective dose) × 0.005% per mSv = 0.0025%

  (or odds of 1 in 40,000)

  Stated in terms of the NNH, for every 40,000 women having a mammogram, one would actually get cancer from the procedure. It follows that, if a woman has 10 annual mammograms in a row, then the NNH for the entire series of 10 would be 4,000 (i.e., the risk would increase 10-fold).

  The first thing we notice is that a mammogram has more cancer risk than an arm x-ray. This is because mammograms require much higher image resolution than bone x-rays. Increasing resolution means more radiation and higher doses, and therefore, higher risk. There is also a sensitivity correction made in calculating the effective dose for breast irradiations because breast tissue has somewhat higher radiation sensitivity than other tissues.6

  Another thing that’s different for mammography compared to an x-ray for a broken arm is that, for the breast, the nature of the benefit and the risk is the same. Whereas the price of a restored broken arm is the risk of cancer, the benefit and risk for mammography are both breast cancer—either finding one or causing one, respectively. Since the benefit and risk are identical, one would think that the risk-benefit analysis simplifies to just the numbers. If the chance of finding a cancer is greater than the risk of getting a cancer, then the benefit outweighs the risk, and the matter should be closed. Fortunately, in the case of mammography, there are numerous studies that quantify the benefits of mammography. Let’s just calculate the chances of benefiting from mammography and we should be done. Or so one would think.

  Determining the value of mammography has, unfortunately, been quite a challenge and highly controversial, largely because the benefit depends on age. The older the woman, the greater the likelihood that a cancer will be found and, therefore, the greater the benefit. The increased likelihood of finding breast cancer with age is due to a combination of the fact that breast cancers are more common in older women and that it’s easier to see cancerous growths on older women’s mammograms because the higher breast density of younger women makes tumors harder to spot. But we need not get into these details. These are the kinds of issues that expert panels consider when trying to formulate recommendations as to what age women should begin having mammograms. This is also why the recommended age for beginning mammograms keeps creeping up. But for our purposes here, we just want to know the overall success rate of mammograms saving lives in the general population. What is the ultimate societal benefit? To put it even more simply, we want to know how successful mammograms are in doing what we want them to do. Do they prevent breast cancer deaths in women?

  It is the answer to this question that has spawned the controversy, because the answer turns out to be, unfortunately, that mammograms have not met expectations. They don’t save nearly the number of lives that everyone had hoped for. Let’s review some recent numbers to illustrate this.

  If 10,000 women, 50 years of age, begin receiving annual mammograms for 10 consecutive years (i.e., a total of 10 mammograms each), then 10 curable breast cancers would be found because of the mammography. If we frame the question as to how many women need to be screened in this way to save one life—that is, the number needed to treat (NNT)7—we can see the answer is simply the 10,000 screened, divided by 10 lives saved, or an NNT of 1,000. It is this number (1,000) that has disappointed so many people. It was anticipated that the NNT would be much lower than 1,000 (i.e., you wouldn’t need to screen so many women to find a lethal breast cancer). In fact, given the reality that breast cancer is a leading killer of women, it seems counterintuitive that lethal cancers would be so hard to find when you are actively looking for them. After a 40-year history with mammography screening, and a lot of anecdotal testimonies from breast cancer survivors, it came as quite a shock that mammography wasn’t doing any better than that at saving lives.8

  It is important to note that this 1,000 number is not without controversy.9 In particular, some say that mammography has been getting better with time, so that any long-term study of screening in the past will always underestimate current benefits. This is the “moving target” argument, in that it postulates that you can never get an accurate assessment of the value of any contemporary technology because, by the time you analyze the data, the state of the art for that technology has moved on to a different place. But even if this moving target criticism is valid, it probably will not have a huge effect on the NNT number. In any event, more and more breast cancer advocacy groups are coming to grips with the idea that an NNT of 1,000, even if it isn’t exactly correct, isn’t too far from the truth either. But for now, let’s get back to our risk-benefit analysis of mammography radiation.

  Even though mammography screening seems to be an inefficient process because 1,000 women need to be screened annually for 10 years in order to save one, at least you are saving that one life. Furthermore, if you need to screen 4,000 women to cause one breast cancer from the radiation (NNH), but only 1,000 women to prevent one breast cancer death (NNT), it would mean that the chances of finding cancers are better than the chances of causing it. So mammography wins the day, right? Yes, that would all be true if radiation-induced breast cancer were the only harm caused by screening. Unfortunately, for every one life saved by a decade of annual mammography screening, 613 of the 1,000 screened women (61%) will experience at least one false-positive reading over those 10 years, just like Teresa did.10 To reiterate, a false positive is a mammogram that is read as positive for cancer when there actually is no cancer. Nancy Keating, a breast cancer physician at Harvard-affiliate Brigham and Women’s Hospital in Boston, says that mammogram false positives are so common that “if you choose to participate [in a screening program], you should assume this is going to happen to you.”11

  False-positive mammograms set into motion a cascade of follow-up medical procedures that e
ach takes its toll on the patients. In fact, out of those 10,000 women in the 10-year screening program, approximately 940 (nearly 10%), will end up having unnecessary biopsies and surgeries; some of them will also develop postsurgical complications, including infections and possibly death.

  The bottom line on the radiation risk of mammography is this: If it were just the radiation dose alone that were driving the risk-benefit analysis of mammography, screening would get the thumbs up, albeit by a much smaller margin than what we found for x-rays of broken arms. It is not, however, the radiation risk alone that is turning up the heat on the mammography screening debate these days. Rather, it is the complete spectrum of harm caused by false-positive findings (added to the modest radiation risks) that is causing the medical community to rethink routine mammography screening. So, the issues with mammography are legion, and only one small part of the downside is the risk of causing cancer from the radiation. In fact, the radiation risk is usually considered so small that few critics of mammography even feel the need to mention it.

  There is one caveat before you draw any premature conclusion about the usefulness of mammography. The above screening scenario does not apply if women have risk factors for breast cancer. Since women with risk factors—particularly a family history of breast cancer—are at higher risk going into the screening than the average woman, they would be expected to benefit more from mammography. And the greater their breast cancer risk factors, the more they should benefit.

  Lest you think that mammography is the only radiation screening procedure facing this false-positive challenge, consider lung cancer. A study of radiographic screening for lung cancer found many early-stage lung cancers among smokers before they experienced any symptoms. This early detection should translate into better treatment responses and better cure rates. Great news, right? Here’s the problem. The same numbers of early-stage lung cancers were also found among nonsmokers, a population with far fewer lung cancer deaths.12 This suggests that most of the supposed lung cancers being found by the radiographic screening would never have developed into clinical disease. It seems that some cancers just don’t progress while others do, and no one knows why. What we really need to do with cancer screening is to be able to distinguish the bad tumors from the nonthreatening ones (the majority). X-rays alone simply can’t do that. They just reveal lumps (abnormal tissue growths) but can’t predict which of those lumps will progress to clinical disease.

  SUPERSIZE ME

  If diagnostic radiography of breasts and lungs can find cancer in those organs, why not screen every organ in the body, and do it all at one time? That’s the idea behind using spiral whole-body computed tomography scanning (spiral whole-body CT) to perform a complete body screen for diseases in patients with no apparent symptoms. These scanners are called spiral CT because they take a helical path around the body in order to produce a computer-generated x-ray image of internal organs.

  This spiral whole-body CT is not the same as a traditional CT scan. A traditional CT scan (which may or may not use spiral technology) is used by doctors to look at cross-sectional “slices” of a person’s body where disease is suspected to be located, based on symptoms. For a coughing patient, it might be the lungs. For a patient with a headache it might be the brain. The point is that the image is restricted to the region of the body of clinical interest, and the dose is delivered to a limited number of narrow cross sections through a particular organ rather than every organ (i.e., a small amount of total body tissue is exposed). The CT x-rays produce cross-sectional images, similar to how a meat slicer cuts cross sections of bologna. The slices are comparable to the images, and only the sliced parts of the bologna that get a radiation dose, not the whole sausage. Granted, bologna is a crude comparison to a human body, but actually it’s not that far off the mark. Everyone has probably seen a cold-cut slice that has some curious imperfection in it. Is it an indication of a bad sausage, or just some normal variation in the texture of the meat? This is the challenge facing physicians reading CT scans.

  Since the slices of a standard CT scan are relatively thin, the amount of tissue receiving the dose is a relatively small portion of the body. As such, the effective doses of a standard CT scan are significantly smaller than the local tissue dose. Spiral whole-body CT scans, in contrast, are quite different because the imaging device spirals around the body and slices the entire body into hundreds of two-dimensional images that are then stacked and reconstructed by computer into a three-dimensional image of all the internal organs. Because the whole body is imaged, and not just part of the body, the effective dose is equal to the actual body dose, typically about 20 mSv13—nearly 40 times the effective dose of a mammogram and 20,000 times that of an arm x-ray. And this means, of course, that spiral CT scans entail 40 times and 20,000 times the cancer risks of mammography and arm x-rays, respectively.

  It is important to interrupt our story here, in order to point out that these whole-body diagnostic CT scanners used to screen for disease are completely unrelated to the full-body scanners used to screen travelers for weapons at airports. Although they both use x-rays, the doses from the airport scanners are negligible. In contrast to the dose of 20 mSv that a diagnostic whole-body CT scanner delivers to a patient, an airport scanner delivers only 0.000001 mSv (one millionth of a mSv) to the traveler. In fact, the x-ray dose received from the airport scanner is comparable to the dose from cosmic radiation that the traveler will receive from just 12 seconds of high-altitude flight, or just 1.8 minutes of background radiation while standing at ground level. Ironically, travelers accumulate a higher radiation dose while waiting in the long airport lines to be scanned than they do from the few seconds they spend in the scanner.14

  As you might have already guessed, spiral CT images are no better at finding life-threatening cancers in the rest of the body than mammographic images are at finding lethal tumors in the breast. They also entail all of the same false-positive problems that mammography does, with one big exception. Mammography, at least, has 40 years of research behind it measuring its benefits. Although the benefits are not as big as we had hoped, they are there nonetheless, and lives are saved. The benefits of a spiral whole-body CT scan as a screening tool for cancer are entirely hypothetical. There is currently no evidence that these screens are saving anyone’s lives.

  You may ask: But what about all those commercials where people claim their lives were saved because a cancer was found during spiral whole-body CT screening? Weren’t their lives actually saved? Maybe, maybe not. Perhaps their cancer was one that would never have progressed to the point that it gave them any clinical symptoms. Perhaps their cancer would have been cured even if it had been discovered later, when symptoms appeared. Or perhaps they aren’t really cured, and their tumor comes back some time later. The point is that no individual can definitively know whether or not her life was saved by cancer screening. In fact, no one can point out the individuals that were or were not saved by screening. All we can say is what happens to patients on average, and from that we infer that there are benefits to individuals. The truth of the benefit is entirely in the numbers. That’s why we need statistics. The truth about spiral whole-body CT scanning for cancer screening is that we currently lack the statistics required to demonstrate a benefit. At this point, the benefit is purely theoretical.

  FIGURE 13.1. WHOLE-BODY COAXIAL TOMOGRAPHY (CT) SCAN. X-ray scans of the whole body are sometimes used to find and diagnose sites of disease. Physicians can use this technique to visualize slices of the body at any desired body depth by producing a series of two-dimensional (2D) images in either frontal (left) or side (right) view orientation. Serial 2D CT scan slices, like the ones shown, can also be combined, colorized, and reconstructed by computers to produce 3D images that physicians use to explore the anatomy of inner organs at any depth and from any angle. An example of such a reconstructed 3D image can be seen in video format at the following website: https://www.youtube.com/watch?v=sEnr6FJZOJM. (Source: The 2D images shown
here are from the website http://radiologystudio.com/ and are used with permission generously provided by Dr. Xinhua Cao.)

  Therefore the benefit side of the risk-benefit equation is questionable. Now let’s take a quick look at the box score for spiral whole-body CT scan risk:

  20.0 mSv (effective dose) × 0.005% per mSv = 0.1%

  (or odds of 1 in 1,000)

  This means the NNH for each scan is 1,000. If a man has ten annual screening scans, his effective dose from the multiple procedures would, therefore be, 200 mSv (10 times 20 mSv), giving an NNH value of 100 (1,000 divided by 10). This dose and risk level is comparable to the doses received by some of the lower-dose victims in the atomic bomb survivor studies. In fact, most of the people in the Life Span Study received doses lower than 200 mSv.15

  The comparison with atomic bomb victims is not meant to frighten, but rather to emphasize an underappreciated point. Although we think of atomic bomb survivors as high-dose victims, and many were, the vast majority of the participants in the long-term atomic bomb survivor study—the ones that are actually the major drivers of the cancer risk estimates we now use—did not receive doses anywhere near what would be required to produce radiation sickness. In fact, they had no radiation symptoms at the time of the bombing, and their major health consequence was increased risk of cancer. They are normal people, just like everyone else, concerned about whether they will be among the unlucky few who develop cancer because of a radiation exposure they had many years ago.