[Note: If you haven’t already, you should read PART 1 of this two-part series. It defines several terms that I will be using in this post, and I don’t plan on explaining them again, given that they were explained in detail in Part 1. Of course, if you’re a medical professional and already know what lead time bias, length bias, and stage migration are, then you should still read Part 1 for its scintillating writing.]
I hadn’t expected that it would take me this long to get around to part 2 of my post from Monday, but, alas, other matters intervened. Better late than never, and besides, we wouldn’t want to make our friendly neighborhood creationist neurosurgeon Dr. Egnor wait an extra day for his much needed dose of Respectful Insolenceâ¢, would we?
Of course not.
When last I left this topic, I had discussed why detecting cancer at ever-earlier stages and ever-smaller sizes is not necessarily an unalloyed good. At that time, I discussed in detail a landmark commentary in the New England Journal of Medicine entitled, Advances in Diagnostic Imaging and Overestimations of Disease Prevalence and the Benefits of Therapy. The article, although nearly 14 years old, rang just as true today in its cautioning doctors about whether ever-increasing diagnostic sensitivity that imaging technology and new blood tests were (and are) providing was actually helping patients as much as we thought it was. Moreover, the discussion is timely, given recent high profile cases of cancer patients whose tumors had recurred, coupled with a at least two sets of confusing new recommendations and and one study about breast cancer screening that have come out all within the last week or so.
Before we dive in, let’s see what Drs. Black and Welch had to say about screening tests for breast cancer 14 years ago, as way of background and linking my last post and this one:
Before the widespread use of mammography, most breast cancers were discovered on physical examination, as palpable lumps. In one of the few studies to assess directly the accuracy of physical examination in screening for breast cancer, only 27 percent of tumors more than 1.0 cm in diameter and 10 percent of those less than 1.0 cm in diameter were detected by physical examination. However, the mean size of breast cancers detected by state-of-the-art screening mammography is about 1.0 cm, and many of the cancers detected as microcalcifications are only a few millimeters in size.
Again, prevalence depends on the degree of scrutiny. According to the Connecticut Tumor Registry, clinically apparent breast cancer afflicts about 1 percent of all women between the ages of 40 and 50 years. In a recent medicolegal autopsy study, however, small foci of breast cancer were found in 39 percent of women in this age group. Most cancers were in the form of ductal carcinoma in situ. Furthermore, over 45 percent of the women with cancer had two or more lesions, and over 40 percent had bilateral lesions. Although it has been argued that such small in situ lesions are not detected by and are therefore irrelevant to screening mammography, about half the lesions in that study were detected, usually as microcalcifications, on postmortem plain-film radiography of the resected breasts. Because of continual technical improvements and increasingly broad criteria for the interpretation of mammograms, the detection threshold for breast cancer has fallen considerably since the time of the Breast Cancer Screening Project of the Health Insurance Plan of Greater New York (1963 to 1975). This can explain the increased prevalence of cancer on mammographic screening, from 2.717 to 7.614 per 1000 examinations (with the incidence increasing from 1.517 to 3.214 per 1000 examinations). The lower detection threshold can also explain the increase in the percentage of carcinomas in situ (stage 0) among all mammographically detected cancers — from 12.7 percent to over 30 percent. The principal indication for biopsy has changed from suspicious mass to suspicious microcalcifications. This can explain why the reported incidence of breast cancer has increased and why most of the increase is in smaller lesions, particularly ductal carcinoma in situ.
The serendipity of this is almost frightening. Were I into the paranormal, I might wonder if there was some higher purpose that pointed this hoary paper to me less than a week before three major papers about breast cancer imaging were released. The first of these is a major set of metanalyses about the value of screening mammography in just this population (women between 40-49 years old) were published in the Annals of Internal Medicine as clinical guidelines, thus putting me in exactly the right frame of mind to look them over critically. Remember, as mentioned for prostate cancer and thyroid cancer in my previous post, a large percentage of apparently healthy women, if you look at their breast tissue closely enough, will have evidence of invasive cancer or ductal carcinoma in situ (DCIS), the latter of which is, depending on how you look at things, either a premalignant lesion or cancer that has not invaded outside of the breast ducts yet. Either way, few women die of DCIS, and the vast majority of women from 40-49 with microscopic evidence of breast cancer or DCIS will never die of cancer, because either does not progress or progresses so slowly that, even in the 30-40 year remaining life expectancy for women in this age range, it does not manifest itself as clinical cancer. This situation is a classic example of how lead time bias and length bias lead to an increase in the number of diagnoses of cancer and a shift towards less rapidly progressive disease being diagnosed, as I described in part 1. The problem, of course, is that we have no way of identifying which of these lesions will turn into clinically apparent cancer and thus have little choice but to treat all detected cancer, even tiny microscopic lesions, as though they will.
These were the issues complicating the question are now, as they were then, that breast cancer is not nearly as common in this age group as it is in women over 50, for whom study after study demonstrate the value of yearly mammography for decreasing mortality from breast cancer. Indeed, the value of yearly screening mammography in women aged 40-49 has been somewhat controversial, and a recent set of guidelines published in the Annals of Internal Medicine, includes a metanalysis of studies on screening mammography in this age range, accompanied by a set of consensus clinical guidelines from the American College of Physicians. The introduction sets the stage:
In the United States, breast cancer is one of the most common causes of death for women in their 40s. In 2002, almost 5000 women between 40 and 49 years of age died of breast cancer, compared with the 6800 women who died of heart disease or 1500 women who died of HIV (5). However, despite the relative importance of breast cancer in this age group, the burden of breast cancer among women in their 40s is low for a population-based screening program. More than 98% of women will not develop breast cancer between 40 and 50 years of age, but they will be subject to the risks of population-based screening. Of the 44 000 women who die of breast cancer each year, fewer than one fifth received their diagnoses between the ages of 40 and 49 years.
That is indeed the problem with all mass screening programs. The disease being screened for must be common and treatable, such that the treatment impacts survival, for example, and there must be a screening test that has an acceptable sensitivity and does not turn up too many false positives. In addition, if it is being applied to mass populations, it must be inexpensive and safe. For women over 50, these conditions certainly apply with respect to breast cancer and mammography, and, indeed, various studies show a 20-25% reduction in breast cancer mortality in this age range in populations that are screened compared to populations that are not. In contrast, for women from 40-49 years of age, the benefit is more modest. What is cited is approximately a 15% decrease in mortality after 14 years; however the error bars are very wide. Indeed, the low end of the range is a mere 1% benefit in decreased mortality. These are the reasons why some European countries do not recommend routine screening mammography for women of average or low risk for breast cancer until age 50, unless there are risk factors, such as a strong family history or mutations in breast cancer susceptibility genes, such as BRCA1 and BRCA2.
Among the other findings of the metanalysis was that the risk from the radiation is quite small. However, the most problematic finding was that the risk of false positive results is high. After 10 mammograms, there is a 20-56% risk of a false positive. All of these will require, at the minimum, further imaging and workup, often with a referral to a breast surgeon for evaluation. Many will end up getting biopsies, the majority of which (75-80%) will be benign. This does not even count the mental and emotional distress that abnormal mammograms cause between the time the woman is informed of the result and the time that either additional imaging or a biopsy. Given this information, unfortunately, the guidelines suggested by the ACP are so vague as to be almost useless:
- In women 40 to 49 years of age, clinicians should periodically perform individualized assessment of risk for breast cancer to help guide decisions about screening mammography.
- Clinicians should inform women 40 to 49 years of age about the potential benefits and harms of screening mammography.
- For women 40 to 49 years of age, clinicians should base screening mammography decisions on benefits and harms of screening, as well as on a woman’s preferences and breast cancer risk profile.
- We recommend further research on the net benefits and harms of breast cancer screening modalities for women 40 to 49 years of age.
Those of you who are gynecologists, internists, or family practice docs (you know, the docs who generally take care of routine screening tests), tell me: Do those guidelines help you at all to decide whether you should screen women between 40-49? The problem is, the above guidelines do not tell me as a clinician how I should balance risk and benefit and how I should counsel patients to balance risks and benefits, and, quite frankly, the discussion after each guideline in the original paper do not help much. In addition, the elephant in the room that no one wants to acknowledge is the medical-legal climate. One of the most common causes of lawsuits, if not the most common, is failure to diagnose breast cancer. What primary care doctor has the cojones not to screen women from 40-49, when the majority of radiology and medical organizations still recommend mammography every 1-2 years between ages 40 and 50, even if the evidence upon which they base these recommendations is not as compelling as that for women over 50?
Not me if I were doing primary care. And not me doing breast surgery. Besides, I think that even the more modest benefit of screening women from 40-50 years old is probably worth it.
Ironically, there is now a newer imaging modality that really brings home the points in the Black and Welch commentary from 14 years ago. I am referring, of course, to MRI. MRI stands for magnetic resonance imaging, and, over the last decade, MRI of the breast has been developed and fine-tuned. MRI can produce stunning images and is highly sensitive. There’s little doubt that it can pick up lesions that mammography can miss, although it is not as good for picking up DCIS as it is for breast cancer. Given that we have probably come close to the limits of resolution and sensitivity in mammography with the newer digital mammography machines, not surprisingly, the question of what role MRI should play in the screening of women for the early detection of breast cancer is now coming to the forefront. As I mentioned very early in the history of this blog, there is good evidence that MRI adds benefit in the screening young women at a high risk of breast cancer, such as women carrying BRCA1 or BRCA2 mutations. But what about the general population, whose risk of breast cancer is much lower than these very high risk women?
Glad you asked, because last week, the American Cancer Society reviewed the literature on the topic and tried to come up with some recommendations for when using MRI as an adjunct to screening mammography is appropriate. The last time it updated its guidelines in 2003, the ACS stated that there was insufficient evidence to recommend MRI as a screening test, except perhaps for women at high risk of breast cancer. I’ll boil down the most recent recommendations, which come in a large paper published in CA: A Cancer Journal for Clinicians by listing them as the appear in Table 1:
blockquote>Recommend Annual MRI Screening (Based on Evidence*)
First-degree relative of BRCA carrier, but untested
Lifetime risk 20-25% or greater, as defined by BRCAPRO or other models that
are largely dependent on family history
Recommend Annual MRI Screening (Based on Expert Consensus Opinion)
Radiation to chest between age 10 and 30 years
Li-Fraumeni syndrome and first-degree relatives
Cowden and Bannayan-Riley-Ruvalcaba syndromes and first-degree relatives
Insufficient Evidence to Recommend for or Against MRI Screening
Lifetime risk 15-20%, as defined by BRCAPRO or other models that are largely
dependent on family history
Lobular carcinoma in situ (LCIS) or atypical lobular hyperplasia (ALH)
Atypical ductal hyperplasia (ADH)
Heterogeneously or extremely dense breast on mammography
Women with a personal history of breast cancer, including DCIS
Recommend Against MRI Screening (Based on Expert Consensus Opinion)
Women at <15% lifetime risk
Of course, it’s hard not to point out that recommending MRI screening for women with BRCA mutations is nothing new; such women are at extraordinarily high risk for developing cancer, as high as 80% over the course of their lifetimes. Indeed, these are the women who often undergo prophylactic bilateral mastectomies or take estrogen-blocking drugs in order to try to decrease their risk. What I am quite unclear about is how they came up with the recommendation that women who have a lifetime risk of developing breast cancer of 20% or greater should undergo MRI screening. The rationale for choosing that number as the cutoff was not at all clear to me from reading the paper; certainly no clear risk-benefit analysis was presented that suggests that, above a 20% risk MRI is worthwhile and below a 15% risk it is not.
Once again, consistent with the Black and Welch paper, I will point out that the increased sensitivity of MRI comes at a price. If a woman gets an MRI, she is twice as likely to get a biopsy as compared to getting just a mammograph and, if needed, an ultrasound. Remember, MRI is more sensitive than but not as specific as mammography. It picks up all sorts of lesions that mammography misses, but most of them are not cancer. Remember, also, the possible confounding effects of lead time bias and length bias. It is not at all clear if detecting breast cancer by MRI at an even earlier point in its development than it would be detected by mammography and/or physical examination will truly improve breast cancer survival independent of lead time bias, nor is it clear what percentage of these smaller cancers would ever grow to cause any problems at all. Remember the autopsy study cited earlier. Among even 40-50 year old women, 40% will have microscopic evidence of breast cancer if you look hard enough. Also, if you look at the Gail Model, it’s pretty easy to come up with a common scenario that pushes a patient’s lifetime risk of breast cancer to 20% or more. For instance, I modeled a 40 year old woman with a single breast biopsy (which was benign), age at first menstruation of 11, and her age at first childbirth over 30, and the lifetime risk is already approaching 18%. True, the Gail model has its problems, but it is still widely used. It doesn’t take much to reach this level of risk.
MRI has other problems as well. For one thing, it’s expensive, easily costing up to $1,000 for the test. In addition, it requires the intravenous injection of a contrast agent, gadolinium. It’s a pretty safe agent, but if we start doing millions of screening MRIs in which it is used as a contrast agent, we’re going to start seeing a lot more adverse and allergic reactions to it, even though they are uncommon now. Finally, there is a big access problem. Although the large urban areas are almost covered now, MRI is still not available everywhere, particularly in rural areas (heck, in some rural areas, there aren’t enough mammography machines to go around). There are also not enough radiologists trained in reading breast MRI to permit any sort of mass screening of even high risk women, nor are there enough genetic counselors to identify all the women who might be at high enough risk to qualify for MRI screening. Moreover, there are not enough facilities with the capability of doing MRI-guided biopsies of lesions that aren’t visible on any other imaging study. Finally, these recommendations were made on the basis of zero data regarding whether such screening results in a decrease in mortality due to breast cancer in the population undergoing MRI. Once again, remember lead time and length bias.
Finally, one last paper in this trifecta of breast cancer imaging papers landed on my desk late last week, right in the New England Journal of Medicine and simultaneously all over the news media. Before I discuss it, remember two terms from last time: stage migration and the Will Rogers effect. (If you don’t know what they are, go back and read part 1.) This particular paper involved the use of MRI in patients diagnosed with breast cancer. As a background, I will mention that careful evalution of the opposite breast in newly diagnosed breast cancer patients will reveal occult tumors up to 10% of the time. Here’s a summary of the findings:
MRI detected clinically and mammographically occult breast cancer in the contralateral breast in 30 of 969 women who were enrolled in the study (3.1%). The sensitivity of MRI in the contralateral breast was 91%, and the specificity was 88%. The negative predictive value of MRI was 99%. A biopsy was performed on the basis of a positive MRI finding in 121 of the 969 women (12.5%), 30 of whom had specimens that were positive for cancer (24.8%); 18 of the 30 specimens were positive for invasive cancer. The mean diameter of the invasive tumors detected was 10.9 mm. The additional number of cancers detected was not influenced by breast density, menopausal status, or the histologic features of the primary tumor.
This study, as the others did, has gotten wide play in the media. Indeed, my partner, whose practice is nearly 100% breast surgery, quipped during tumor board this week while presenting a patient that his patient had come in after the results of this study had been announced; so she was getting a bilateral breast MRI. This study tells us nothing to help us decide whether detecting these tumors at the time of diagnosis of the main breast cancer, as opposed to later, when clinically relevant tumors would certainly show up on mammography or ultrasound, improves patient survival at all. All it says is that they can be detected. Not surprisingly, some of the media accounts featured women in which MRI found small cancers in the opposite breast telling how MRI “saved my life.” For some, that may be true, but for many probably not. Remember again the concepts of lead time and length biases. Indeed, for really small tumors in the other breast (particularly DCIS) that will now be detected by MRI, it is quite possible that the adjuvant chemotherapy given after surgery could well eradicate a significant percentage of them such that they never become a problem or that small foci of DCIS might never grow to cause the patient problems. We just don’t know. Finally, also remember that only 3% of breast cancer patients in this study had a tumor found in the other breast that was not apparent on mammography. That’s a pretty small number. In addition, only 18 of these (approximately 1.8%) were invasive cancer; the rest were DCIS, a precancerous lesion whose course we cannot predict and some percentage of which may never turn into cancer. So, if you boil it all down, you have to do MRI on close to 10,000 breast cancer patients to identify 18 cancers in the opposite breast. From this data, I consider it quite possible that MRIs could, as seen by Black and Welch stated 14 years ago, lead to additional surgery that ultimately does not benefit the patient.
Another issue to consider is an unintended consequence that is likely to come about if MRIs are done for all patients with breast cancer. For this discussion, remember the term “stage migration.” Most doctors are not going to order an MRI on just the opposite breast in a patient with newly diagnosed cancer. Indeed, prior to the publication of this paper, there was a growing tendency to order MRI for breast cancer to evaluate the extent of disease in the involved breast. That means that these patients are probably all going to get bilateral breast MRI, and this is where the problem of the extreme sensitivity of MRI comes in. There is little doubt that MRI will be more likely to find additional suspicious lesions in the breast with the known cancer an/or to indicate that the cancer goes further than it appears on mammography. As I asked a few days ago (rhetorically, of course), how can this not be good thing? Trust me, there’s a way. One unintended consequence of using MRI willy-nilly to evaluate extent of disease in breast cancer could well be an increase in the percentage of breast cancer patients undergoing mastectomies as opposed to lumpectomies. Dr. Monica Morrow, Chairman of the Department of Surgical Oncology at the Fox Chase Cancer Center put it very well in an editorial in an editorial in JAMA in 2004:
Is it possible to reconcile the fact that MRI detects additional carcinoma [in the same breast] in as many as a third of patients, yet clinical outcomes for patients undergoing breast-conserving therapy on the basis of clinical and mammographic selection are excellent? At this point, some clinicians may be experiencing a sense of dÃ©jÃ vu. In the 1970s when the prospective randomized trials of breast conserving therapy were being initiated, pathological studies demonstrating multicentric foci of carcinoma separate from the primary tumor site in 38% to 54% of patients were used to argue that even early stage breast carcinoma was a disease of the entire breast, and treatment with less than complete mastectomy was inappropriate. Extensive clinical experience has shown that the majority of these tumor foci are controlled with breast irradiation and these deposits of microscopic cancer have largely been forgotten for the past 20 years. Now, MRI technology is capable of identifying some of these occult tumor foci. Is this an appropriate reason to start doing more mastectomies? The answer is no. Prior to adopting MRI as a routine part of the evaluation of the patient with breast cancer, there should be evidence of benefit to the patient in the form of a reduced rate of ipsilateral breast tumor recurrence. This reduction in breast recurrences will need to be weighed against the number of extra biopsies resulting from the routine use of MRI, the cost, and the delay in surgical treatment.
In the absence of trials with a clinical end point, breast cancer patients undergoing MRI should be advised that this step forward in technology may take them right back to the 1970s and result in a mastectomy for disease that can be controlled with radiation.
And, in reference to this most recent study, Dr. Morrow said:
“I don’t think you can conclude from this study that MRI is justified,” said Monica Morrow, the chair of surgery at Fox Chase Cancer Center. “Look at how good the outcomes are without MRI. Maybe MRI just leads to more aggressive and unnecessary surgery.” Some research suggests that small undetected tumors can be eradicated simply with standard hormonal drugs such as tamoxifen, which cuts off the supply of estrogen that fuels tumors. These drugs are particularly effective against early, pre-invasive lesions that are confined to milk ducts – those that accounted for almost half of the MRI-detected cancers in the new study.
Yes, everything old is new again. Alas, the older I get, the more I realize how often this is the case in medicine, with the very same issues popping up decade after decade in new guises. Black and Welch nailed it 14 years ago when they wrote:
Misperceptions of disease prevalence and therapeutic effectiveness can promote a cycle of increasing medical intervention, despite the best intentions of all parties. The cycle usually begins with some form of increased testing that lowers the threshold for detecting disease, such as technical improvement in imaging tests, more frequent testing, or closer scrutiny of the images. This immediately leads to a higher diagnostic yield of the disease and a spectrum of milder cases. These effects are almost always interpreted as indicating progress and provide immediate reinforcement for the increased testing, despite the caveat that earlier detection is a double-edged sword…Tests that are more sensitive…are accepted as better, even though they detect a broader spectrum of disease that includes a subgroup whose natural history and response to intervention are unknown. Consequently, the assessment legitimizes the use of the more sensitive imaging test and becomes a distraction from the fundamental question: How should patients with this newly detectable subclinical disease be treated?
Don’t get me wrong. I’m no Luddite; I love new gadgets and tests as much as the next doctor. Nor am I saying that MRI doesn’t have a role in breast screening or the workup of breast cancer, and I don’t entirely buy into the first paper’s recommendations, which suggest that screening for breast cancer with mammography in women ages 40-49 may do more harm than good, at least for some women. There is little doubt that a major cause of the decline in breast cancer mortality since the 1980’s has been the detection of breast cancer at earlier stages with mammography. It is quite possible that detecting tumors at still smaller and earlier stages could result in a further decrease in mortality from breast cancer, even taking into account the confounding factor of lead time bias. Moreover, there is an argument that taking care of the cancer in the opposite breast at the same time saves the patient a second set of operations and treatments later. It is nonetheless premature to conclude from the last two studies that I mentioned that MRI should be used in screening for any but the highest risk women or that MRI should be part of the routine battery of tests that women newly diagnosed with breast cancer undergo. It’s an expensive test (radiologists like it because they get paid a lot more for it than mammography); it’s operator-dependent and has a steep learning curve; there is no evidence that using it decreases mortality in women screened by it; and there is no evidence that its wanton use improves survival or decreases recurrence in patients with breast cancer thought to be treatable with breast conserving therapy, such as lumpectomy. Moreover, it has the real potential to result in more aggressive and disfiguring surgery for an unclear and possibly even nonexistent benefit to patients. Until more studies are done, it’s hard to justify the widespread use of MRI suggested by these studies.
Unfortunately, in the wake of these new recommendations and studies, doing the high quality followup studies that answer the question of whether the use of MRI truly improves breast cancer care will be very difficult, and it will be very difficult for a clinician to justify not doing bilateral breast MRIs in, for example, patients with newly diagnosed cancer. Because of the medical-legal climate with respect to malpractice in this country coupled with media coverage that stokes patient demand, I reluctantly predict that it will not be long before most women (at least those with insurance, anyway) will undergo MRI screening (and a lot more biopsies as a result) and every breast cancer patient will have bilateral breast MRI upon diagnosis (and a lot more mastectomies as a result). It will take several years, if not a decade, to sort out whether this change in the standard of care for screening and for breast cancer patients actually results in improved survival and/or decreased rates of recurrence, or whether even the best technology and intentions will be thwarted again by lead time bias, length bias, and stage migration.
Back to the future we go!