MRI Scans as Overtreatment

Yesterday, while scouring KevinMD, I stumbled across a post from “Respectful Insolence,” a blog authored by an academic surgeon/scientist who dubs himself “Orac.” In the post, Orac  reports that this Wednesday at the American Society of Clinical Oncology (ASCO) 2008 Annual Meeting,researchers from the Mayo Clinic will be reporting on a disturbing correlation between the use of breast magnetic resonance imaging (MRI) and a rise in the number of women having mastectomies. In this context, Orac offers a cogent, compelling perspective on why too much cancer screening can harm patients.

Orac’s worries specifically relate to using MRI scans to detect breast cancer. Advocates of the procedure rightly claim that MRI scans can detect more growths than other techniques, including mammography [i.e. an x-ray] and a clinical examination. The MRI technology detects so much that, as the New York Times put it last year, the scans reveal “all sorts of suspicious growths in the breast, leading to many repeat scans and biopsies for things that turn out to be benign.”

In other words, breast MRI scans are so sensitive that if you have breast cancer, there’s an almost 100 percent chance that they’ll detect it; but the technology produces many  false positives because it’s not as good at distinguishing between malignant and benign growths. As Orac puts it: “…MRI [scans] now routinely "section" people into "slices" much thinner than 1 cm, making our imaging sensitivity considerably higher than it was 14 years ago.

The problem is that while many people undergo malignant changes in various organs as they grow older than most will never actually develop “clinically apparent cancer.” In fact, some studies have shown that MRI scans accurately detect breast cancer just 30 percent of the time. Though most studies place this rate at a higher level, they also show that mammographies (using x-rays to examine the breast) lead to fewer false positives than MRI scans.

The problem with high sensitivity but many false positives (what clinicians call “low specificity”) is obvious. Orac
again: “[MRI] picks up all sorts of lesions that mammography misses,
but most of them are not cancer.” You might guess that this dynamic
would result in more patients freaking out about having breast cancer
and taking relatively drastic measures to deal with it—say, having full
mastectomies. You would be right.

This brings us to the Mayo Clinic report
that will be presented Wednesday at ASCO’s annual meeting. The
researchers at Mayo looked at correlations between MRI scans and
mastectomies among 5464 women who had surgery for early-stage breast
cancer at the Mayo Clinic between 1997 and 2006. They found that rates
of mastectomies fell from 45% in 1997 to 30% in 2003, but then rose to
43% in 2006. This recent uptick was coupled with an increase in the use
of breast MRI scans in the clinic: the number of women who underwent
these scans doubled from 11% in 2003 to 22% in 2006. In that year, more
than half the women (52%) who underwent an MRI scan at Mayo went on to
have a mastectomy. In contrast, only 38% of women who did not have an
MRI had a mastectomy in 2006.

According to Dr. Rajini Katipamula, lead researcher of the study, these
data “demonstrate that a significant number of women with early-stage
breast cancer are undergoing mastectomy, and it appears to be partially
related to the introduction of preoperative MRI.” The scans detect more
tissue growth, which makes patients more worried that they have
cancer—and one would guess, sparks concerns that the cancer is
widespread, depending on how many growths are detected. So women are
increasingly undergoing mastectomies—total removal of the
breast—instead of getting a mammography, a lumpectomy (tumor removal)
or radiation therapy (zapping cancer cells with high-energy particles).

Orac points out that a big problem here: survival rates for patients
undergoing mastectomies are the same as for those undergoing other,
less severe procedures such as lumpectomies or radiation therapy. And
there is also no evidence that using an MRI scan to detect cancer
reduces mortality rates. This is an important consideration. Indeed, we
should always ask ourselves the following question posed by Orac: “Does
finding more disease before surgery that wouldn’t have been found
before [surgery] improve outcomes and survival rates?” 

So are we 100 percent sure that more MRI scans is causing more
mastectomies? No—correlation is not causation, so theirs is no
definitive cause-and-effect relationship between more MRI and more
mastectomies. But it sure seems likely, particularly when compared to
other explanations.

For example, Matthew Goetz, a co-author of this week’s Mayo study, suggests
that an increase in mastectomies might have to do with advances in
reconstructive surgery that make the prospect of losing both breasts
less frightening than it once was. But this isn’t a terribly compelling
explanation. Breast reconstruction is no walk in the park. Mastectomy
patients can get breast implants which will probably need to be
replaced over the years and can cause pain or infections. Another
option is tissue flap procedures, which take tissue from one part of
the body (usually the stomach or upper back) to reconstruct the breast.
According to the American Cancer Society,
this isn’t a picnic either: “There can…be complications at the donor
sites, such as abdominal hernias and muscle damage or weakness. There
can also be differences in the size and shape of the 2 breasts. Because
blood vessels are involved, these procedures usually cannot be offered
to women with diabetes, connective tissue or vascular disease, or to

The big problem with all this, of course, isn’t just that women are
undergoing more mastectomies—but that, because MRI scans have a
relatively high rate of false positives, that women are undergoing more
unnecessary mastectomies. This is especially worrisome  given
that the sensitivity of MRI scans are so high that, once cancer has
been found in one breast, they are increasingly being used to detect
early-stage cancer in the other breast. This means we may well be
seeing an increase in double mastectomies over time. Indeed we already have  seen a rise–although MRI scans are not the only reason for this trend.

As Dr. Seema Khan, director of the program for early detection and
prevention of breast cancer at Northwestern Memorial Hospital, is
quoted as saying on Orac’s blog, "there’s a huge question of whether
we’re being led down a path of overtreatment by routinely using MRI.”
As is so often the case in American medicine, the emergence of a new
technology has briskly outpaced our knowledge of how to best utilize
it—and of its associated risks.

Indeed, Orac notes that the results of “mammography and ultrasound,
along with clinical assessment,” have been “validated over three
decades with large clinical trials.” When it comes to MRI scans,
however, “there is no evidence yet…[that they]in any way leads to
improved survival in breast cancer.” What we do know is that the
all-seeing eye of the MRI “can…influence a clinician’s and woman’s
decision” about how to respond to breast cancer—or to something that
might or might not  be breast cancer.

The rush to MRI creates other problems. A breast MRI scan costs at least $1,000, and generally anywhere from 10-15
times as much as mammography. Because MRI machines are so expensive,
they also tend to be concentrated in high-income metropolitan
hospitals, raising issues of access. And though breast MRI can be
covered by Medicare, local Medicare contractors have wide discretion to determine the circumstances under which a MRI scan is covered.

But really, this isn’t about money or health coverage. It’s about the
fact that—as Orac succinctly puts it—“more is not always better.”

17 thoughts on “MRI Scans as Overtreatment

  1. There will always, unfortunately , be more surgical procedures than are deemed medically necessary after the fact. But tell that to my patient who had a mastectomy after an MRI found a large breast cancer that 2 mammograms and 2 ultrasounds did not.
    This patient went on her own for an MRI, not from my RX, and it saved her life.
    Have I ordered more MRI’s due to this? No. Each test ordered is based upon each individual patient, not on statistics of healthcare costs and savings.
    Physicians jobs are to treat and prevent disease, not to bow down to govt cost savings.

  2. Excellent Niko- Dr. Nortin Hadler writes extensively on this topic in at least two books (“The Last Well Person” and “Worried Sick”) and several excellent published papers.
    Dr. Rick Lippin
    Southampton, Pa

  3. Thinking of DrSH’s first comment that some surgeries will prove unneeded, there used to be a rule of thumb that if less than 10-15 percent of the appendices removed by a give surgeon were normal, the surgeon was dangerous in that he or she was waiting too long for a firm diagnosis, and risking peritonitis. Anyone know if that informal guideline has changed, since a laparoscopic appendectomy is even less risky than the classic method?
    If it has, I suspect that diagnostic imaging has played a role, with ultrasound being helpful in many cases, and abdominal CT being the “gold standard”.
    On the other paw (I’m being assisted by Rhonda, one of my feline staff), there’s not enough discussion, it seems of VOMIT. No, not gastrointestinal distress, but “Victim Of Modern Imaging Technology”, I hear this most from trauma surgeons who see people operating on pathology only on images — or don’t operate even though clinical judgment says it’s warranted, but nothing is visible in imaging.
    There’s not only a cost-benefit analysis, but a cost-risk analysis. We used to think MRI was benign, except for patients with embedded metal — and, thankfully since I have a pacemaker, that is no longer an absolute contraindication. Unfortunately, while the MRI itself may be benign, the gadolinium contrast agent used for many MRIs turns out to be potentially damaging to the kidneys.
    There’s a fairly solid base of evidence that CTs (and SPECT), which deliver more radiation than regular X-rays, slightly increase the risk of cancer. Nevertheless, I’ve had both MRIs and CTs that I considered quite appropriate — the imaging I’ve felt inappropriate were several X-rays and one ultrasound.
    We don’t yet know enough to be certain when and when not to order a particular imaging study, and what action to take as a result. Radiologists do not tend to be definitive in their reports; there’s a story about one who called 911 to report that he had symptoms “consistent with being mugged and shot.”
    With the significantly improved chemotherapeutic regimes for breast cancer, at least in postmenopausal women, the mastectomy vs. lumpectomy vs. chemotherapy alone decision may be revisited. Again complicating things, there’s the difficult choice of prophylactic bilateral mastectomy in women who carry the BRCA2 gene and have a family history of breast cancer.

  4. Doesn’t positron emission tomography (PEM) have equal sensitivity and greater specificity than MR? Why hasn’t it caught on?

  5. The different imaging methods used may depend upon the cost or personal physician preference. Computed tomography (CT) typically costs $500 to $700; Magnetic resonance imaging (MRI) typically costs $700 to $900; PET and other forms of nuclear medicine, in which a patient is given a radioactive material, typically costs $2,000 and up.
    In life or death situations, sometimes one must make judgements based upon preponderance of available evidence as opposed to proof beyond reasonable doubt.

  6. What is being said about MRI scans as overtreatment was also being said about Mammograms, according to Lancet. Mammograms not only pick up slow-growing tumors, but also identify cell changes that under a microscope look like cancer, but are biologically benign. As a result, doctors may aggressively treat something that may have gone unnoticed during the woman’s lifetime.
    While it is often asserted that early detection spares patients more aggressive treatments, screening results in over-diagnosis has led to a 20% increase in mastectomies and a 30% increase in the removal of tumor. Many women who come out suspiciously on a mammogram have their biopsy show that they did not have breast cancer but a benign condition.
    However, since PET images biochemical activity, it can accurately predict whether a tumor is benign or malignant, thereby avoiding surgical biopsy when the PET scan is negative. Conversely, since a PET scan images the entire body, confirmation of local or advanced metastases allows the physician and patient to more accurately decide on how to proceed.
    A non-local metastasis can alter treatment plans from surgical intervention to chemotherapy. A confirmation of local tumor without any metastasis can justify surgical intervention only. And of course, PET is the most accurate diagnostic procedure to differentiate tumor recurrences from radiation necrosis or post-surgical changes.
    Research indicates that PET can play a growing role in tracking the effectiveness of chemotherapy and evaluating early response to a selected drug regimen. Possible indications could be provided by PET imaging, but the only problem with this method is you give the patient potentially toxic and ineffective drugs and wait to make tumor measurements, then give more potentially toxic and ineffective drugs and wait to repeat measurements.
    You still have the patient getting potentially toxic and ineffective treatment and then you still have to wait until you could try Plan B. You measure the drug effects on tumors in the patient, one treatment regimen at a time, rather than in the laboratory where as many as twenty to thirty treatments can be done to see which one works best.

  7. For JR mostly, but some general observations. PET is not extremely precise in determining physical relationships; it’s often combined with CT to get an accurate anatomical reference.
    If it’s not an emergent situation where the much greater speed of CT makes the decision over MR, it’s not that one is better than another — they image different things. For the brain and spine, MR is often preferred, while CT would be preferred for abdominal structures.
    Perhaps you are thinking of the closer comparison between PET and _functional MRI_, rather than regular MRI. PET senses metabolism while fMRI senses blood flow; they can give complementary views, but neither have the anatomical accuracy of CT or regular MR — they are imaging moving things while, in principle, the other methods are looking at a still or reasonably still object.
    Things get even more complex when the procedure is invasive, variously to inject contrast media, or some specialized techniques such as ultrasound from the inside of a blood vessel, or just down the throat under local anesthesia.

  8. Alllan–
    This is actually Niko’s post, not mine, but let me respond to one comment.
    You quote Niko saying:
    “This is especially worrisome given that the sensitivity of MRI scans are so high that, once cancer has been found in one breast, they are increasingly being used to detect early-stage cancer in the other breast. This means we may well be seeing an increase in double mastectomies over time.”
    Allan, you write: Maybe that is a good thing (early detection and treatment of cancer of the other breast).”
    I think it’s easier to say that if you are a man. Some women who have been terrorized by the well-intended by someimtes over-the-top campaign against breast cancer might say, I’d be happy to lose both breasts even if there is only a very,very small
    chance that I will die of breast cancer.
    But the fact is that women who have mastectomies are no safer than women who have the much smaller procedure– a lumpectomy.
    Mortalities Are No Higher for Women Who Choose the Lumpectomy.
    And in many cases (breast cancer in situ) the tiny tumor detected by the MRI does not need treatment. It will never develop into full blown cancer.
    This has been a women’a health issue for a long time. Some male doctors recommend full mastectomies in an almost cavalier fashion–without
    seeming to understand the emotional and psychological consequences for many women.
    It has been suggested that the same male doctors would be far less likely to recommend removal of a testicle under circumstances where mortalities were no higher if the patient chose a less aggressive strategy.
    I’ll be writing about informed decision-making and breast cancer soon.

  9. Sometimes the frequentist approach is so narrowly focused and rigorous in its requirements that it limits innovation and learning.

  10. Despite the increased ability to offer breast-conservation techniques to patients with breast cancer, there exist certain groups who may be better served by traditional mastectomy procedures including:
    * women who have already had radiation therapy to the affected breast
    * women with 2 or more areas of cancer in the same breast that are too far apart to be removed through 1 surgical incision, while keeping the appearance of the breast satisfactory
    * women whose initial lumpectomy along with (one or more) re-excisions has not completely removed the cancer
    * women with certain serious connective tissue diseases such as scleroderma, which make them especially sensitive to the side effects of radiation therapy
    * pregnant women who would require radiation while still pregnant (risking harm to the fetus)
    * women with a tumor larger than 5 cm (2 inches) that doesn’t shrink very much with neoadjuvant chemotherapy
    * women with a cancer that is large relative to her breast size
    * Women who have tested positive for a deleterious mutation on the BRCA1 or BRCA2 gene and opt for prophylactic removal of the breasts
    * male breast cancer patients.

  11. Health is not an industry, people should understand. And when I hear, what about plastic surgery? Well, breasts implants are not about health, they’re about looks and appearance.