An article and editorial in today’s issue of the Journal of the National Cancer Institute speaks once again to what is becoming a recurring and repetitive theme, namely the overdiagnosis of cancer, its implications, and what to do about it.
My overriding concern is how an academic discussion is going to be interpreted and responded to by a profession and a public that is not particularly familiar with all of the nuances of that discussion and whether in fact we may end up drawing premature or incompletely informed conclusions about a very important topic.
What I find interesting about the entire concept of overdiagnosis is the thought that this is something new or something we haven’t thought about in the past. That's simply not the case.
A bit of history:
When I was a freshman medical student lo those many years ago (in the late 1960’s), one of the first things we learned was that there were cancers found in the bodies of people who were autopsied that were never discovered during life and would likely not have been a problem during life. We aren’t talking about large tumors that were simply not found because of medical mismanagement or a missed diagnosis. We are talking about fairly small cancers that were present in various organs of the body, not causing damage to nearby structures and not having spread elsewhere in the body.
The primary example of these cancers being found on postmortem was prostate cancer, but similar findings were noted for breast and thyroid cancers. We also learned about mysterious, unexplained cancer “regressions,” with kidney cancer being the primary example of that category of events. So the concept of “not all cancers cause a person problems during their lifetimes” is not a new one and certainly not one that wasn’t taught in medical schools years ago.
Fast forward to my early years as a medical oncologist, and I quickly learned that some cancers were very aggressive and some less so.
There was breast cancer in a woman that occurred in a few months after a normal screening mammogram and careful clinical examinations by three competent doctors who followed the same patient. There was a patient with a colon cancer—he happened to be in his 30’s--where a small cancer was found on colonoscopy one week, then evaluated for spread of disease and none was found. Two weeks after a negative workup with a small primary lesion, he went to surgery and the cancer was widespread in the liver. He died shortly afterwards.
And then there were the patients I treated for cancer that had spread to the lungs. There were several I was able to identify where the progress of their cancers was so slow—despite the fact that they had spread—that I talked to them at length about not taking chemotherapy. I followed those patients closely, and some of them went for a considerable period of time before their cancers started to progress and treatment was started. And there were other patients who had a particular type of lymph node cancer where they did well for years with no treatment because their disease didn’t progress.
What I am trying to say here is that we have known for decades that not all cancers behave the same way. We know some grow rapidly, some grow slowly, and some grow hardly at all. That’s old information, and it is not news.
Which brings me to today’s article, which basically says that we are “overdiagnosing” a lot of people with certain types of cancers. They define “overdiagnosis” as a cancer that “never progresses (or, in fact, regresses) or … the cancer progresses slowly enough that the patient dies of other causes before the cancer becomes symptomatic. Note that this second explanation incorporates the interaction of three variables: the cancer size at detection, its growth rate, and that patient’s competing risks for mortality.”
For example, the authors say that for breast cancer, the number is 16% by one analysis and 24% for another. What really caught my eye was their claim that 51% of lung cancers were overdiagnosed using the “old fashioned” methods of chest x-ray and sputum samples. For prostate cancer, the number is 60%--a number which many experts may agree with as being in the ball park.
The authors do conclude—in light of declining death rates—that screening may have a benefit for breast cancer and prostate cancer. And, because rates of diagnosis of cervical and colorectal cancer are declining, there is likely little overdiagnosis in those two diseases. But not so fast: they also indicate that this may be due to “more overdiagnosis of the precursor lesions, for example, cervical dysplasia or adenomatous polyps.”
They also acknowledge the “conundrum” (their word) that faces clinicians in determining whether or not a patient has an aggressive or indolent cancer:
“Overdiagnosis can only be identified in an individual if that individual 1) is never treated and 2) goes on to die from some other cause. Because clinicians do not know which patients have been overdiagnosed at the time of diagnosis, we tend to treat all of them. Thus, overdiagnosis contributes to the problem of escalating health-care costs. But even where there no money involved (sic), overdiagnosis would be a major concern: Although such patients cannot benefit from unnecessary treatment, they can be harmed.”
If that sentence leaving you scratching your head, then perhaps you can understand why this is such a difficult issue. Let me see if I can clarify the comment:
If a doctor finds a cancer, the doctor can’t tell whether or not the patient is overdiagnosed. The doctor frequently is compelled to suggest treatment to the patient, but doesn’t know whether that treatment is helpful because the only real way to know if the treatment didn’t make a difference would be to leave the cancer alone and see whether the patient dies from cancer or another cause. But if the doctor treats the patient, that costs money. Spending that money may be unnecessary, but the more important issue is that the treatment itself may lead to difficulties for the patient long term (I certainly agree with that last observation).
The researchers lay much of the “blame” for overdiagnosis to increased use of diagnostic imaging, such as CT scans and ultrasounds. CT scans find things that otherwise would not be found but frequently get diagnosed and treated. The authors point out the increased number of kidney cancers and thyroid cancers as examples of cancers found not infrequently by these imaging techniques. With the rapidly increasing use of CT scans of various parts of the body—namely abdominal and chest CT scans—the suggestion is that is responsible for lots of “overdiagnosis.” (There are also lots more brain CT and MRI scans being performed, but I am not aware that there is a contention that has led to a lot of overdiagnosis of brain tumors in this country).
The authors go on to say: “Often, the decision about whether or not to pursue early cancer detection involves a delicate balance between benefits and harms—different individuals, even in the same situation, might reasonably make different choices.” They admit that the issue is complex, and that “admittedly, quantifying overdiagnosis is challenging.” They further suggest that we somehow must raise the threshold of labeling a test abnormal, and perhaps ignoring smaller abnormalities, or examining the growth of a lesion over time. Their final recommendation is to incorporate the concept of overdiagnosis into the medical curriculum.
The editorial which accompanied the article makes many of the same arguments, again from very competent and qualified clinical researchers—but researchers who are “on board” with this thesis, as opposed to someone who might have a different point of view and who might provide the other side of the argument.
They do make a point that almost all of us familiar with this discussion agree on: “We must advocate for and demand innovation in diagnosis and management, fueled by science, harnessing modeling, molecular, and immunology tools to address this problem.”
“Is it too risky to not biopsy and to potentially miss a cancer? Perhaps it is just the opposite. It is too risky to continue on the path where we are compelled to know what every lesion is, and then invoking the …reflexive need to treat anything that resembles cancer. We need to curb the urge to intervene with more thought about what is truly valuable. We can ill afford to spend resources for diagnosis and treatment if we do not make a material contribution to a person’s well being.
“Perhaps most importantly, we have an obligation to educate patients and clinicians to explain more and do less when appropriate. We need to make sure that patients understand that not all cancers have the potential to kill and use language that engenders less fear… The challenge for the scientific and medical community is to work alongside our patients to make care more appropriate, more tailored, less resource intensive and less morbid.”
It all goes back to square one: how the heck do you tell whether or not the cancer you find is a bad cancer in the first place? And are you willing to move back the clock of time, find fewer cancers, and risk that some bad actors may get out of the barn but fewer patients will be diagnosed and treated?
That, my friends, is the crux of the issue.
Parallel to this discussion is the reality that we have developed better and better ways to diagnose cancer earlier. We have better mammograms, we have colonoscopes which find polyps before they become cancer, we have CT scans which can slice and dice the lung into such small sections that previously undetected lesions can now be seen.
It isn’t difficult to see how the pace of our technology has taken us from a time where we simply couldn’t find a cancer in the body with an old fashioned x-ray to a point where we find very small lesions with modern CT scanners and MRI machines. There will come a time in the future where our ability to detect cancer will far exceed what we are able to do today, probably through some form of genetic test, or blood sample or even a breath sample. I wouldn’t be surprised that if some day in the future we will be able to tell when even only a few cancer cells in our body will result in the detection of “cancer.”
In reality, our technology has become a blessing and curse. As we have moved further down the spectrum of finding cancers at smaller and smaller sizes, we move closer and closer to the old autopsy studies that I mentioned above. The not-so-surprising impact of all of this is that we find cancers today that will never cause harm.
In the past, we had to look at dead bodies to find these cancers. Today, it just takes a click of the switch on a CT scan, and presto! A lesion shows up that wasn’t expected in a thyroid gland, the lung, the kidney or one of many, many other places in the body. Or a mammogram finds a non-invasive cancer of very small size that was never destined to become larger.
The unresolved question is what does all this mean? What do we do about it? More importantly, what can we do about it?
We simply do not have reliable tools today that help us accurately determine which cancers are “bad” cancers, will act aggressively, and can lead to death. And we don’t have tools today that are sufficiently defined to tell us with high specificity whether or not a particular cancer needs treatment, and equally important, whether it will respond to treatment.
We are—in a very real sense—the victims of our own success. We all agree that we treat many cancers that would never have caused harm. I don’t know many doctors who want to put a patient through unnecessary cancer treatment. To suggest otherwise is ridiculous.
So we are left on the horns of a dilemma: We can find cancer early, and realize that we are treating people where the treatment may not impact their lives, and yet we don’t have real options that help us decide how to differentiate bad cancers from indolent cancers, or precisely tell us whether or not we made a difference in someone’s life.
There are some other “real world” issues, like medical malpractice that can take away everything you own (literally—I am not overemphasizing this) if you decide to counsel a patient that they are being overdiagnosed by not following up on a suspicious skin lesion or a breast mass that is consistent with cancer. What happens when the “wait and see” approach shows that your hunch that this was a “slow growing” cancer turns out to be wrong? Are these discussions about overdiagnosis going to lead us down a path where the doctor is darned if they treat a cancer and darned if they don’t? Are we going to see doctors sued because someone survives cancer and a claim is made that they were put through unnecessary treatment due to overdiagnosis? Stranger things have happened.
No one wants to be diagnosed with cancer. No one wants to be treated unnecessarily for cancer. Yes there are cancers we diagnose that will never cause harm. But to suggest that in some way we can accurately tell the difference isn’t where we are today. Our tests aren’t perfect, and our tests are fallible. But they are all we have right now—and I don’t see us moving back the clock anytime soon.
I suspect that many of us don’t want to go back to the future when autopsies were the way we found indolent cancers. We need to look forward to the time when our investment in research will provide us with the tools that will allow us to accurately predict which cancers need treatment, which don’t, which ones will harm a person, and which ones can be left alone.
Until we get to that moment in time, I believe that honesty compels us to acknowledge that what sounds theoretically intriguing is simply not practical. Hopefully we have the wisdom to understand the difference.