Cancer incidence not always the best measure of cancer trends
Click Here to Manage Email Alerts
Cancer incidence rates represent one key measure of progress in the fight against cancer, but they are not the only consideration. According to a special report published in The New England Journal of Medicine, concordance between incidence and mortality rates provides a more complete picture of cancer burden.
H. Gilbert Welch, MD, MPH, senior investigator in the Center for Surgery and Public Health at Brigham and Women’s Hospital, said in an interview with HemOnc Today. “One of the things I’ve always wondered is whether to focus on absolute rates or look at the relative change. It occurred to me that I didn’t have to choose. I could look at each cancer in two ways.”
Welch and colleagues — including Barnett S. Kramer, MD, MPH, of the division of cancer control and population sciences at NCI, and William C. Black, MD, professor of radiology, community and family medicine at Dartmouth Institute at Geisel School of Medicine at Dartmouth — evaluated patterns of incidence and mortality for several cancers over a 40-year period, from 1975 through 2015. The researchers characterized each cancer’s “signature” based on concordance or discordance of the two measures.
“Cancer incidence isn’t necessarily about the true occurrence of disease; it can also be influenced by the degree to which physicians or the system are looking for the disease,” Welch said. “A major point to our paper is that reported incidence is not necessarily a good reflection of true disease occurrence.”
Welch spoke with HemOnc Today about signatures that are considered “desirable” and those that may be more indicative of overdiagnosis.
Question: What is an example of a “desirable signature” and why is it classified as such?
Answer: Incidence and mortality are the two measures we analyze in the report. We look at absolute rates and relative rates. The relative rate starts in 1975, which was the base year. What we see for Hodgkin lymphoma is a constant incidence over the last 40 years, reflecting a stable true occurrence of disease. Yet, mortality dropped steadily over the same period; it’s now one-quarter of what it was in 1975. This is a picture that should prompt oncologists and hematologists to pat themselves on the back. This is what a steady improvement in the treatment across time looks like. It’s pretty powerful.
In chronic myeloid leukemia, incidence is again constant over 40 years. Mortality also is constant for about 20 years and then drops precipitously. The question is, “What happened to mortality at the turn of the century?” It declined sharply, and oncologists know this reflects the introduction of imatinib, an incredibly effective drug in CML. These two signatures are desirable as they show improvements in treatment. On the one hand, we see a steady improvement in treatment, and on the other hand, it’s dramatic.
Q: What did the trends show in terms of lung cancer?
A: Lung cancer is, bar none, the most important cancer. It kills more Americans than the next three cancers combined. What’s interesting about lung cancer are the relative changes in incidence and mortality. They go up and down and track each other perfectly; they’re essentially superimposed. That’s because incidence again reflects the true occurrence of lung cancer which, in turn, reflects the rise and fall of cigarette smoking. The fact that mortality perfectly tracks incidence is indicative that our ability to treat the disease remains fundamentally unchanged. What drives mortality is how many cases there are. It’s that simple.
Incidence and mortality are now going down. The rates among men have been going down since the mid-1990s. Women peaked later because they started smoking later.
Q: You point out that signatures with declining incidence and mortality also are “desirable.” Which other cancers have this signature? How do they relate to screening practices?
A: Stomach, cervical and colorectal cancer — which showed declines in incidence and mortality at about the same rate, consistent with declining true cancer occurrence — have highly variable screening practices. We don’t really screen for stomach cancer, whereas we’ve screened for cervical cancer for many years. Then there is colorectal cancer, which is somewhere in between. In 1975, we weren’t screening for it. We started screening in the mid- to late-1990s, but even now, we’re not screening very completely for it. What this shows is that declining true cancer occurrence can have different explanations. In stomach cancer, it has nothing to do with medical care; it’s something about environment, something about Helicobacter pylori. With cervical cancer, we typically say, “this is about screening.” But, it also may be about general improvement in health and better ability to shed HPV. For colorectal cancer, we’re saying this can’t be all about screening, because mortality was declining long before screening was started.
Q: Can you discuss the cancer trends that were indicative of overdiagnosis?
A: In thyroid cancer, kidney cancer and melanoma, we have stable mortality and stable metastatic incidence, yet dramatic increases in overall incidence. This is not an increase in true cancer occurrence; this is an increase in our tendency to look for the cancer (eg, how often we test for it, in how many people), and the increased ability of our tests to find small cancer. In short, when we look harder, we find more.
The thyroid cancer data reflects a combination of physicians doing more physical exams, finding more nodules in the thyroid and then ordering an ultrasound. Some of it is incidental detection. A patient may get a chest CT because they’ve been in a car accident and a thyroid nodule was found, or they get an ultrasound of their carotid artery and a thyroid nodule was found. Most Americans diagnosed with thyroid cancer have their whole thyroid removed and need to go on thyroid replacement, and so forth.
Then there is melanoma, which is up about sixfold since 1975. That is all about skin exams, Melanoma Mondays, and most likely changing pathologic thresholds of what is classified as melanoma. This is an unhappy picture. It reflects a lot more people being treated, a lot more people spending money, missing time from work and having anxiety. Frequently, the money they spend is out of pocket.
Q: What do you think can be done to stop or prevent this overdiagnosis?
A: It takes a major education campaign, and dermatologists must change course. Dermatologists have made a massive effort targeting pigmented lesions with no evidence that this would affect melanoma mortality rate. They’re always going to be able to find more melanoma, but they’re not finding the melanomas that matter. In fact, some of the worst melanomas don’t happen in sun-exposed areas that would be visible. They can happen in the gums, in the eye, near the anus.
We also need to delve further and better characterize how the cancer came to be found. Was this a patient who presented to us with a complaint or a symptom? We’d call that clinically detected. Or was this a case where the patient was fine (had no symptoms of cancer) but we decided to do a certain test to check for early cancer? That would be screening detected. Or, is this a cancer we just stumbled upon when we were doing something else? Was the patient in a car accident, and we did a CT scan and found a thyroid cancer? I think the use of three categories would be a very important addition to our cancer incidence data. – by Jennifer Byrne
For more information:
H. Gilbert Welch, MD, MPH, can be reached at 75 Francis St., Boston, MA 02115; email: drgilwelch@gmail.com.
Disclosure: The authors report no relevant financial disclosures.