Mammograms and overdiagnosis – a hard truth to hear #breastcancerrealitycheck

Friday while flipping through my RSS blog feed reader, I came across an blog post about an article on overdiagnosis. Unfortunately, I cannot remember where it was or I’d link back to it. The blog post was talking about recent research study:

Welch, H. G., Prorok, P. C., O’Malley, A. J., & Kramer, B. S. (2016). Breast-Cancer Tumor Size, Overdiagnosis, and Mammography Screening Effectiveness. N Engl J Med, 375(15), 1438-1447. doi:10.1056/NEJMoa1600249

You can get the abstract from:

If you’d like to see the full document, feel free to leave a comment and I can send along my annotated version.The article itself is very well written and the language is not so filled with scientific mumble jumble such that it does not require a PhD in cancer science to understand.

The study compares population data (so not just those who have breast cancer but the population as a whole) for 1975-59 (pre-mammography era) and 2000-2002 (post-mammography era). This is the most recent one can get because they are using 10-year survival as a measure – so those who live 10 years post diagnosis are considered to have survived the disease. Also the study applies to the general population at “average risk”, so it does not apply to anyone who is high risk such as those with family history or generic mutations such as BRCA.

What I took from the article is that mammography has led to finding tumors while they are smaller. Those tumors which might have grown to be large are found sooner. However, and this is a big one, mammography also finds a LOT of tumors that will never grow to be clinically significant. That means, it might find a tumor that won’t kill you and won’t spread. This is very similar to the controversy around PSA screen for prostate cancer. The idea is that you may test positive but the cancer won’t be what kills you. You’ll die of other age related illness first.

“although trial data may provide an assessment of some negative consequences of screening, such as false positive results and associated diagnostic procedures, such assessments may underestimate what actually occurs when screening is implemented in the general community” (p. 1439)

It is easy for me to read this report, because it would not have changed anything for me. I had heard that it might be an issue for younger women (under 50), so I had chosen to not do the mammogram that my family doctor ordered back in March of 2014. I do not regret that decision. In June I found the lump. My cancer WAS clinically significant. The first primary tumor was growing rather rapidly. The other two tumors were growing at a slower pace. For me, I can look back and know that I made the right decision.

“Although the biologic characteristics of a tumor are now recognized to be more relevant to breast cancer prognosis than the size of the tumor,  tumor size is more relevant to the assessment of the proximate effect of screening” (p.1439)

The challenge with reading the report is more with people for whom overdiagnosis may very well be the case. These are some of the women that I see in support group meetings. They go through horrible body deforming surgeries and long term systemic treatments like hormone therapy for cancers that very well may never have become clinically significant. They do not want to hear about overdiagnosis. They are ristant to the whole idea. They need to believe that they went through this horrible treatment in order to save their life. It is the only way one can live – knowing that you made the best decision you could with the information you were given. No one who has experienced early stage breast cancer directly wants to hear that they did not need the treatment they underwent.

“given the stable incidence of metastatic breast cancer for more than three decades, despite spanning the era of increasing prevalence of screening-mediated breast cancer and changing patterns of hormone therapy” (p.1439)

Let that quote sink in a little. The stable incidence of metastatic breast cancer for more than three decades! What this means is that although more women are being identified and treated as having breast cancer, the same number of women are dying from it. Again think about that. It means that some of the women that are being treated would never have died from it even it if had not been treated.

“Assuming a stable underlying incidence of disease burden and no overdiagnosis of tumors, the additional detection of small tumors should be accompanied by a corresponding decrease in large tumors over time” (p.1440)

Now for the numbers …

“The incidence of large tumors decreased by 30 cases of cancer per 100,000 women” (p.1441)

This is the argument for screening and early detection.

“The incidence of small tumors increased by 162 cases of cancer per 100,000)” (1441)

“Assuming that the underlying burden of clinically meaningful breast cancer was unchanged, these data suggest that 30 cases of cancer per 100,000 women were destined to become large but were detected earlier, [meaning caught early or “early detection”], and the remaining 132 cases of cancer per 100,000 women were overdiagnosis (i.e. subtracted from 162)” (p.1441)

So for 30 people the mammogram meant they found the cancer sooner which may translate to less deforming surgery (I say MAY because treatment options are constantly changing/improving), but for 132 people new cancers were found that otherwise would not have been found or become clinically significant. Note that statistics don’t translate directly to the individual, but this is one way to represent what the data means.

“For large tumors, the declining case fatality rate predominantly reflected improved treatment” (p.1442)

One other challenge with overdiagnosis is that it causes a skew in the long term survivorship numbers. If more people are said to have cancer, and they survive, then it looks like our treatments and survival statistics are improving. However, it is an artificial improvement because a portion of the people said to have cancer never would have been diagnosed with it in the first place.

“Screening can advance the time of diagnosis of a tumor, thereby detecting the tumor when it is still small, without changing its prognosis, a phenomenon termed ‘biologic predeterminism'” (p. 1444)

In other words, you spend more of your life knowing you have breast cancer. Some women argue for mammography because it does matter for a portion of the population. However, what if you are the person for whom mammography found a tumor, you underwent life-altering deforming surgeries and chemotherapy that has horrible side effects only to learn that your cancer was not the kind that would kill you or even grow to a point that you ever noticed it? Would you not have preferred to not have known about it in the first place?

“Because tumors with favorable molecular features grow more slowly, they are disproportionately available to be detected by screening (so-called length-bias sampling). Thus, the expectation is that some tumors that are detected by screening when they are small would have favorable biologic characteristics and could have been treated equally effectively at clinical presentation. For this subset of tumors, earlier detection at a smaller size would not translate into a mortality reduction.” (p.1444)

This is largely what caused me to decide not to that have initial mammogram. I had heard that it didn’t make a difference in survival. I was happily naive. If the cancer is going to kill you anyways, would you not have preferred to have the two, three, five years of living your normal life cancer free before finding out you were going to die from the disease?

The challenge with the rhetoric around mammography saving lives is that it is not true. What this does say is that we should be focusing more of the cancer research dollars on figuring out what biological characteristics actually translate into cancers that do become clinically significant. We don’t need more complex 3D mammography. We could use more methods of detecting the biological characteristics, not the physical size. We do not need to detect cancers when they are so small that they may never actually be an issue. Rather, we need to be looking at what causes metastasis – what makes it spread – and how can we better predict which tumors will need aggressive treatment and which will never turn out to be a problem – an incidental finding.

  • Becky


  • Wanted to leave a comment on this because I had a great conversation about it with @bccww on twitter. During the conversation what occurred to me is that most of the people that express strong opinions on this topic have what I call ‘diagnosis bias’. When you have been diagnosed with breast cancer, then you are not in a good position to consider what a naive you would have thought. You are biased because you already know the outcome. But someone who is at average risk and has not been diagnosed would think very differently about whether or not they wanted to know. They do not see the world through the bias of having been diagnosed.

  • It bothers me that I don’t see more studies out there that focus on finding the characteristics that allow a stage 1a tumor to metastasize vs. a stage IIIb tumor that didn’t. It seems like the ability should be out there. We have a researcher here in Utah named Alana Welm that created a better mouse model for studying metastatic breast cancer. They found that if your tumor took to a mouse more easily, there was a greater chance that you would become metastatic (I believe they compared records later – I was one of the many people that submitted my tissue to be used to build this model). I sometimes want to ask if they can open up the records and find a way to determine if my tissue created a mouse model that is now in use. If you Google her name and “mouse model” you can find info on her. Now I am geeking out, but I find the science so interesting and want to see more and more research that could hopefully save lives someday.

Leave a Reply

%d bloggers like this: