By F. Perry Wilson, MD, MSCE

You’ve seen the headlines. Cancer rates, across a variety of cancers, are on the rise among young people. The findings have led to more than just hand-wringing.

The National Cancer Institute and Cancer Research UK have launched a multimillion-dollar program to find causes, the US National Cancer Plan focuses on the issue, and recommendations from the US Preventive Services Task Force (USPSTF) — recommendations that are required to be followed by insurance companies without a copay — have recently changed to reduce the age of colon cancer screening from 50 to 45, and for breast cancer screening from 45 to 40. It’s all hands on deck to determine what is going on with young people and cancer.

But, according to a new special investigation, this rising incidence in early-onset cancer might be… fake?

Yes, it’s getting a bit spicy in the cancer epidemiology world these days, with takes and countertakes on underlying data that everyone basically agrees on. Which is this: The rate of early-onset cancer (cancer diagnosed before age 50) is rising. For some cancers, such as thyroid, kidney, endometrial, and colorectal cancer, these rates are rising quite dramatically — more than 1% per year from 1992 to 2022.

That much is not really up for debate.

The question being debated is whether these new cases of cancer are meaningful. Are they the result of a truly rising incidence of disease, or are we just getting better at detecting disease that, in prior years, we would have missed and wouldn’t have mattered anyway?

If you follow this space, you may have seen my article from a couple months ago, debating whether I, a 45-year-old man, should follow the USPSTF guidelines and schedule my colonoscopy. (I did, by the way. Good times.) In that article, I looked at the rates of colon cancer diagnoses in young people and suggested that the rates have gone up both because of increased diagnostic accuracy and a true increase in the rate of the underlying disease. This was based on data which suggested that much of the increase seen in young people was due to detection of early-stage cancers, but the incidence of metastatic colon cancer in young people was also going up.

Basically, I just want to get it out there early that the dichotomy between “there is a dramatic increase in cancer in young people” and “the rise is… more apparent than real,” as the article suggests, is somewhat artificial.

To figure this out, we’ll need to examine the data. But before we do that, we need to talk about the concept of “overdiagnosis.”

Overdiagnosis refers to the detection of a condition (in this case, cancer), that would never become clinically relevant were we not to have detected it. This is not misdiagnosis. These are real cancers; they are just cancers that never would have mattered. The idea actually started with autopsy studies. For example, studies show that 60% of men above age 79 who died from other causes had prostate cancer identified on autopsy. These are cancers that didn’t cause an issue. Had they been diagnosed before death, they would have been overdiagnosis.

But there are two problems with overdiagnosis as a concept. First, it’s impossible to know if any individual cancer we diagnose is an overdiagnosis. Data can suggest that overdiagnosis exists in a population, but we don’t know if your cancer is an overdiagnosis because we have no way to know if this is one of the many cancers that will harm you at some point or if you’ll die from some other cause before that.

The other problem with overdiagnosis in early-onset cancer is that overdiagnosis has often been associated with older age. The older you are, the more likely you are to die from something other than the cancer. Are we to believe that a 45-year-old, recently diagnosed with, say, endometrial cancer, would be fine for the next 40-plus years of her life had we never diagnosed it? That feels uncomfortable to me a priori.

But let’s give the paper its due. The authors cite a few lines of evidence to argue that the increase in the eight cancers most rapidly on the rise in young people, shown here, represents overdiagnosis, rather than a meaningful increase in disease.

The central argument is simple. It goes like this: The observed mortality rate from a cancer is driven by the true rate of the disease times some mortality rate of that particular cancer. The observed incidence of new diagnoses of cancer is driven by the true rate of the disease times some detection rate. They note that incidence has gone up dramatically, but the mortality rates from these cancers have not.

That’s true.

Let’s compare that graph of the increasing incidence rate of those eight cancers with this one, showing the mortality rate from those cancers.

Pretty darn flat.

If the true rate of cancer were rising, the authors argue, mortality would go up. If we’re simply ramping up the detection rate, then mortality would remain flat. Mortality did remain flat; therefore, we are catching cancers that would never be a problem.

I see the logic. But a couple of thoughts immediately come to mind.

First, this assumes that our treatments for cancer haven’t improved over the past three decades. They clearly have. The advent of immunotherapy alone has dramatically changed the landscape of cancer care. Maybe more people are getting cancer but we’re getting better at treating it — a standoff, so to speak.

The authors acknowledge this idea but dismiss it, saying, “For these opposing forces to align so precisely… seems implausible.” That is very much a judgment call. It’s also worth noting that it’s not like mortality rates are perfectly stable; there is always quite a bit of wiggle, so it’s not like these forces need to balance to some absurd degree. It is quite possible that, as cancer incidence has truly increased, we have kept pace with improvements in therapy.

It is also possible that there has been a true increase in disease, but the increase is in less-deadly cancers, driving down the overall mortality rate.

But “less deadly” doesn’t mean unimportant. Overdiagnosis occurs when the condition you diagnosed would never become clinically relevant. And there are plenty of “clinically relevant” consequences of cancer in a young person that fall short of mortality. In other words, there may be real value in identifying and treating cancers in young people that, while they may not lead to outright death, would lead to significant morbidity.

To be fair, cancer treatment is no walk in the park. And believe me, if I saw a 40-year-old with a new diagnosis of breast cancer and could tell her, with perfect certainty, that the cancer would never lead to any clinical issues for the rest of her life, I would be delighted to do that. But of course that is not possible because overdiagnosis is only useful when we talk about a population, not an individual. And I can’t imagine a doctor sitting on that diagnosis simply because it was detected incidentally.

We should also consider the possibility that early diagnosis leads to better treatment. This is not too much of a stretch. In fact, the very reason for population screening programs for cancer is the idea that catching cancers early can reduce the mortality rate of that cancer.

The authors acknowledge this possibility too but state that if we are catching important cancers earlier, why don’t we see a balanced reduction in later-stage cancers? Shouldn’t an increase in early diagnosis lead to a reduction in, for example, metastatic disease? We don’t see such a reduction. But if the underlying cancer rate is increasing, the metastatic cancer rate might be increasing as well, offset by the increased early diagnosis. In other words, this is complicated, but I’m not ready to write off all the epidemiology here as spurious.

The authors lean heavily on two cancers where there is quite a lot of evidentiary support for overdiagnosis: thyroid and kidney cancer. Rates of both of these have skyrocketed due to increased screening for the former and incidental detection in the setting of increased abdominal imaging in the latter. Meanwhile, mortality is flat.

These poster children for overdiagnosis are a bit special, however. These are notoriously slow-spreading cancers, for one, and tend to invade locally long before they metastasize. That is not the case for other cancers on the study’s list of the eight fastest-increasing cancers.

To be fair, when it comes to colorectal cancer, the authors came to the same conclusion I did a couple months ago. Here there actually has been an uptick in mortality; and so they buy that, while some of the increase is overdiagnosis, some reflects a true increase in the incidence of the disease. So I guess I still have to get my colonoscopy.

In short, I think this paper overstates its conclusions. And I worry about statements like this: “Searching for biological causes for rising incidence in cancers without evidence of a rise in clinically meaningful cancer is bound to be unproductive. Chasing potential exposures is not just a waste of time, but also diverts funding, talent, and attention from addressing more important issues affecting young people in the US.”

That is a very strong position to take and again presents a false dichotomy. We can acknowledge issues with too much screening and incidental findings while also acknowledging that there are likely exposures that are truly increasing cancer rates in younger people. And studying what those exposures might be (obesity, pollution, stress, and so on) can be done right alongside studying all the other important issues young people face, such as loneliness, drug and alcohol abuse, and not being able to afford a home.

In the end, I continue to feel that the likelihood of cancer being overdiagnosed rises with age, since the competing risk of death from other causes rises with age. Because of that, I would be very hesitant to write off any observed increase in cancer among younger people as unimportant.

F. Perry Wilson, MD, MSCE, is an associate professor of medicine and public health and director of Yale’s Clinical and Translational Research Accelerator.