Roxanne Nelson, RN, BSN
October 07, 2019
Cancer mortality is declining, as has been evidenced by a number of recent publications.
However, this decline in mortality has not been accompanied by a decrease in cancer incidence.
But this doesn't necessarily mean that cancer occurrence is increasing, says the lead author of a recently published study.
"Rising incidence may not reflect rising true cancer occurrence, but may instead reflect increasing diagnostic intensity," explains H. Gilbert Welch, MD, MPH, from the Center for Surgery and Public Health, Brigham and Women's Hospital, Boston, Massachusetts.
"While most people might think that rising overall cancer incidence reflects dangers in our environment, it instead reflects a danger in our medical care system," Welch told Medscape Medical News. "The biggest force causing overall incidence to rise is overdiagnosis — an unfortunate side effect of our drive to find cancer early."
Overdiagnosis refers to the phenomenon by which sophisticated methods of imaging and other methods of detection reveal cancers that in many cases would never have become evident clinically. This is a subject that Welch has long emphasized — for example, in a 2016 NEJM article on overdiagnosis of breast cancer from mammography screening.
Welch explores this issue once again, together with two colleagues, in a new study published in the October 3 issue of the New England Journal of Medicine.
This time, however, the authors have used a new approach to illustrate the issue: they looked at "epidemiologic signatures" for various cancer types.
"Cancer incidence is not a reliable measurement of cancer burden," Welch said in a statement. "Rising incidence may not reflect rising true cancer occurrence but instead reflect overdiagnosis ... It's an issue that's been around for years but this paper highlights that we can't rely on cancer incidence alone."
Welch and colleagues emphasize that the primary goals of this article are to "enable general medical readers to interpret trends in the basic measures of population-based cancer burden and to provide insight into true cancer occurrence, overdiagnosis, and treatment advances."
Importantly, the authors encourage readers not to interpret cancer-incidence trends in isolation, but instead to view them in the context with trends in cancer mortality and trends in the incidence of metastatic disease. While there is little ambiguity when incidence is stable, evolving incidence needs more cautious interpretation.
Analyzing Epidemiologic Signatures
In order to analyze the epidemiologic signatures of various cancers, Welch and colleagues looked at both mortality and incidence patterns from 1975 through 2015.
The mortality data come from the National Vital Statistics System and incidence data from the original nine registries of SEER — the Surveillance, Epidemiology, and End Results Program. The incidence of metastatic disease was also examined, but only included cases when the patient first presents with metastases.
The team then categorized the 'epidemiologic signatures' as being desirable, undesirable, or as having mixed signals.
Desirable signatures comprised the cases where stable incidence signals true cancer occurrence and where the associated drop in mortality clearly demonstrates improvements in cancer treatment.
One example cited was Hodgkin's lymphoma, which has shown a gradual decline in mortality over time and a generally stable incidence. This reflects a steady improvement in therapy, the authors note.
Another is cigarette smoking and lung cancer. After smoking was positively linked to lung cancer and usage rates began to decrease, what was most striking about these signatures is that incidence and mortality move together, the authors point out. For both men and women, the rise and eventually the fall of smoking was followed by the same pattern for lung cancer.
"The biggest force, by far and away, for reducing cancer mortality is the decline in cigarette smoking," said Welch. "Next is better treatment."
Other examples of cancers showing 'desirable' epidemiologic signatures include chronic myeloid leukemia (CML), where there have been rapid treatment improvements, and stomach cancer, where there has been a decline (due to treatment) of a powerful risk factor (the bacterium Helicobacter pylori).
Cancers Where Overdiagnosis Plays a Role
Undesirable epidemiologic signatures are those that show discordance between occurrence and mortality, or suggest overdiagnosis.
For example, for thyroid and kidney cancer, and also for melanoma, the reported incidence is rising but mortality has remained stable. "Stable mortality should be viewed as a marker for stable true cancer occurrence," the authors point out. "Although it is possible that stable mortality could result from a combination of increasing true cancer occurrence and improvement in treatment, such a perfect annual counterbalancing of opposing forces would be a remarkable coincidence."
It is more likely that these signatures suggest overdiagnosis, in that cancers that were not "destined to cause death" were detected.
Mixed signals, the third category, is more complex and highlighted by changes in breast and prostate cancer incidence — both of these show rising incidence and declining mortality, but both are cancers where screening has played an intensive role.
The researchers point out that the introduction of widespread screening mammography led to a rapid increase in breast-cancer: the incidence increased rapidly and has apparently settled at a new, higher baseline. This could either be a true increase in occurrence, or overdiagnosis associated with widespread population based screening, they say.
Breast cancer associated mortality also began to drop in the 1990s, which could reflect either improved treatment or screening or a combination of both. "Other data suggest that improved treatment is the primary explanation," they write.
For prostate cancer, the incidence dramatically increased with the introduction of widespread prostate-specific antigen (PSA) screening, but has subsequently declined almost back to its 1975 baseline, they note. This "remarkable volatility" cannot be explained by changes in true cancer occurrence, the authors emphasize, and instead, "it highlights how sensitive prostate cancer is to diagnostic scrutiny."
The incidence of metastatic prostate cancer has also markedly declined after screening was introduced, which suggests that screening advances the time of diagnosis for prostate cancers that are destined to metastasize, the authors note. Similar to breast cancer, decreasing mortality that was first observed in the 1990s could reflect either improved treatment or screening or a combination of both.
The authors have disclosed no relevant financial relationships.