Study raises questions about value of breast cancer biopsies

Breast biopsies are good for accurately diagnosing invasive cancerous cells, but are less accurate when it comes to finding other abnormalities, according to a new study. This means many women may receive unnecessarily aggressive treatment. Hari Sreenivasan learns more about the findings from lead author Dr. Joann Elmore of the University of Washington.

Read the Full Transcript


    More than 1.5 million women in the U.S. get a breast biopsy each year. But a new study raises doubts about their accuracy.

    Hari Sreenivasan explores the findings and the implications for treatment.


    The study found that, when it comes to invasive cancers, pathologists' diagnosis were generally correct. But an expert panel said pathologists had more trouble making the right diagnosis in about 25 percent of all cases.

    That was primarily true when it involved a case of ductal carcinoma in situ, or DCIS, cells that are abnormal, but not invasive. Doctors also had trouble diagnosing in cases of atypia, when abnormal non-cancerous cells are identified.

    Dr. Joann Elmore of the University of Washington School of Medicine is the study's lead author, and joins us now.

    So, Dr. Elmore, how big of a population are we talking about when there are error rates of possibly 20 or 25 percent?

    JOANN ELMORE, University of Washington School of Medicine: Well, I think you started correctly by saying that, every year, 1.6 million — 1.6 million biopsies are done in the U.S.

    Of those biopsies, some are interpreted as normal, others as cancer. And it's these in-between diagnoses that are the most problematic. They're very hard for us to give a diagnosis. There are probably 200,000 women each year that have a breast biopsy and get a report that says they're abnormal cells. It's not normal and it's not invasive cancer. It's somewhere in between.


    When you get a sentence like that out of your doctor's mouth, I'm sure that's one of the last things you want to hear, but what is the consequence of that? Does that mean that people are treated too aggressively or not treated aggressively enough?


    Well, let's start with your first comment, which is very correct.

    When you hear that sentence, especially a word like ductal carcinoma in situ — it has the word carcinoma — it can be very scary for women and their family. And even though it has the word carcinoma, it's not the same thing as invasive cancer.

    So, I guess, first, I would recommend that women stop, take a deep breath and realize that this is not a diagnosis of invasive breast cancer, and they have time to pause and reflect and gather information.

    We found that, with the diagnosis of DCIS, while four out of five agreed on a diagnosis, this also meant that one out of five disagreed. For the diagnosis of atypia, we found that — about 50 percent agreement. So this is similar to the agreement of flipping a coin, guessing heads or tails.

    When women are diagnosed with DCIS, they are told that they're at increased risk of breast cancer. We can't identify which woman is going to go on to be diagnosed with breast cancer. And so, understandably, a lot of women want to have what some would consider pretty aggressive treatment.

    Women are having mastectomies and lumpectomy with radiation therapy. Women with DCIS are having about the same kind of treatments as women with early stage invasive cancer.


    The idea that this is pathologists interpreting what the slide shows them, in this day and age, I almost imagine that a computer programmer, an algorithm of some sort could at least give us a lead. But this seems almost like an interpretation, that it is up to humans to interpret whether or not this cancer — or that these cells are exhibiting a cancer.


    In this day and age, I think many are surprised that much of medicine is an art.

    What we have found with breast biopsy interpretation, it's also a similar finding with radiologists interrupting mammograms, with cardiologists listening to heart murmurs using their stethoscope. And so this variability among physicians when we give a diagnosis has been noted in all specialties.

    So it's not unique to pathology. For hundreds of years, we have been diagnosing cancer by getting the tissue, putting it on a glass slide, looking at it under a microscope, and deciding, using our vision, whether it is normal vs. cancer. You would open, in this day and age, with modern technology, we would have other tools, other molecular markers or genetic tests, but, currently, our diagnosis is provided by the pathologist.

    You did ask about computers, though, and that is an interesting question. We now can take these slides and digitize them and put the image up on a screen. And so pathologists now can actually look at the images on a computer screen. It's not FDA-approved. But we can start having computers evaluate these images to see whether we can come up with computer-aided detection programs.

    They have developed this in many other areas of clinical medicine. So it's something that we need to look into.


    All right, Dr. Joann Elmore from the University of Washington, thanks so much.


    My pleasure.