Here’s an article in the New York Times that exemplifies so much of what is wrong with the current debate about cancer screening.
The article describes a Dutch study on the effects of breast screening for women with so-called “breast cancer genes” as well as a family history of cancer. The author writes:
“For women with a high risk of breast cancer because of genetic mutations or family history, yearly M.R.I. scans in addition to mammograms and breast exams may save lives.”
A bit of background is in order here. Our bodies have evolved multiple safeguards to protect against cancer, and for most of us, they work very well, at least until senescence sets in. The BRCA1 and BRCA2 genes are often referred to in the media as “breast cancer genes,” but they are more accurately referred to as tumor-suppressor genes. The products of the genes make up part of a complex which repairs double-stranded breaks in DNA. In individuals who are born with one defective copy of either of these genes, a mutation in a cell in the good copy can set that cell on the road to becoming a cancer cell. (Several more mutations in that same cell are necessary to complete the process.) For this reason, these individuals are at increased risk not just for breast cancer, but for other kinds of cancer as well, including cancers of the ovaries, fallopian tubes, and prostate. In addition, a large proportion of the extremely rare cases of breast cancer in men are associated with the presence of a defective copy of the BRCA2 gene.
It's not a foregone conclusion that screening can help these folks. Not all cancers are created equal. Some grow so slowly that they never bother a person until that person dies of something unrelated. Others grow so rapidly that, even if detected early by any of the new screening technologies, end up killing the patient anyway. The whole idea of screening for cancer was based on the hope – that’s all it ever was, a hope – that there were some cancers which grow so rapidly that they are lethal if one waits until symptoms appear, but, which if detected before that, are curable. There is no convincing evidence that such cancers even exist -- or, of they do exist, that they are sufficiently common that screening can make a difference in the only statistic that matters, the death rate from all causes.
A meta-analysis published in 2008 by the respected Cochrane Collaboration found that screening women for breast cancer reduced the risk of dying of breast cancer by one in two thousand. Now, if you think such a tiny reduction in risk can even be measured reliably, well, bully for you, but the only statistic that means anything to you as an individual is the OVERALL death rate. What difference does it make if you die of breast cancer at 65, or you die of something else at the exact same time? We all die of something, be it breast cancer or a heart attack or stroke or being run over by a bus on your way to the mammography clinic. And the same meta-analysis found compared women who underwent screening mammography and those who did not and found NO DIFFERENCE in the overall death rate.
Last September, a Norwegian study published in the New England Journal of Medicine which controlled for advances in breast cancer treatment over the years found an even punier reduction in the risk of dying of breast cancer – one in two thousand five hundred, which was found not to be statistically significant. The NEJM study did not look at all-cause mortality.
In short, there is no convincing evidence that screening women for breast cancer saves lives. But, perhaps, in women with defective BRCA1 or BRCA2 genes, and/or a family history of early breast cancer, screening might conceivably result in a significant reduction in their risk of dying prematurely, since their overall risk is so much greater to begin with. So let’s take a look at what the researchers found.
The study enrolled 2,157 Dutch women judged to be at high risk for developing breast cancer. 594 of these women had defective BRCA1or BRCA2 genes, 5 had a defective version of another tumor-suppressor gene, and the remaining 1,558 were judged to be at risk because of family history of breast cancer. Women who had symptoms or a previous history of breast cancer were excluded from the study.
All women received biannual clinical breast examinations as well as annual mammograms and MRI scans. These women were followed for anywhere from 1.7 to 8.4 years, with a mean of 5.0 years.
Of the women judged to be at risk due to a family history of cancer, 36 cases of invasive breast cancer were recorded. None of these women died from breast cancer during the course of the study.
Of the women with defective versions of either the BRCA1 or BRCA2 genes, 42 cases of invasive breast cancer were recorded. Of these 42, four died from their cancer (two of these had their tumors detected while still at a “favorable” size of less than 1 cm in diameter). The overall five-year survival rate for these women was calculated to be 93%. Flipping that around, that means the death rate from breast cancer was 7%.
(Please note that means 7% of all women enrolled in this study with defective BRCA1 or BRCA2 genes AND who were also diagnosed with invasive breast cancer, not 7% of all women with defective BRCA1 or BRCA2 genes enrolled in the study. For the latter group, the death rate due to breast cancer was less than one percent.)
There was no control group that did not undergo screening, so the authors compared them to 26 cohorts of patients with defective BRCA1 or BRCA2 genes from previously published studies, which showed an average five-year survival rate of 74.5%.
So was the increase in the five-year survival rate due to screening, or to other factors such as improved treatment? And does it translate to a decrease in the only statistic that matters, which is all-cause mortality? Who knows?
It may be that a properly controlled study would show that screening these high-risk women does produce a significant reduction in their risk of dying prematurely. Or maybe not. Perhaps there is no benefit to early detection, and the increase in the five-year survival rate was due solely to advances in treatment. Or perhaps the five-year survival rate went up because all this screening is detecting a lot of cancers that never would have bothered the women until they died of something unrelated. That NEJM paper I linked to above showed that the incidence of “invasive” breast cancer nearly doubled in Norway in the years after screening mammography was made routine. Unless you believe that some mysterious factor caused the incidence of invasive breast cancer to spike at the exact same screening mammography was introduced, then you must believe that even a lot of cases of so-called “invasive” breast cancer never bother a woman until she dies of something unrelated.
Or maybe there is some benefit to early detection, but perhaps it is outweighed by the harmful effects of repeatedly blasting the breasts with radiation in individuals who already have one defective copy of a crucial tumor-suppressor gene; or for that matter maybe it is simply swamped by mortality due to other causes (remember, these women are at increased risk for many other kinds of cancer besides breast cancer). I just don’t know. The point is, the authors of this study don’t know either.
And in fairness, they don’t say otherwise. In the discussion section of the paper, they state:
“In view of the absence of randomization or correction for lead time or for potential differences in treatment between studies, definite conclusions on survival effects of specific screening strategies cannot be made.”
The problem is, by the time this message filters out to the mainstream media, it morphs into the coquettish “MAY save lives,” which I suspect in the minds of at least some readers means that it DOES save lives.
So the lesson for today is caveat emptor, whether we are talking about medical interventions or news stories about medical interventions.
All photos via Wikimedia Commons