From the Journals

Cancer burden: Multiple metrics needed to clarify the big picture



A new analysis of 40 years of U.S. cancer data underscores the importance of looking at multiple metrics to discern the complex interplay of factors influencing cancer burden in the population. Findings showed that the epidemiologic signature – a composite of two or three key metrics – differed across cancer types and was favorable in some cases and unfavorable in others.

“Epidemiologic signatures that illustrate trends in population-based data on cancer burden provide insight into true cancer occurrence, overdiagnosis, and treatment advances,” explain the analysts, led by H. Gilbert Welch, MD, MPH, Center for Surgery and Public Health, Brigham and Women’s Hospital, Boston. “They are important indicators of the potential contribution of environmental exposures, primary preventive interventions, new treatments, and changing diagnostic and screening practices.”

Dr. Welch and colleagues analyzed data for the years 1975 through 2015, assessing juxtaposed trends in incidence, mortality, and, when available, metastatic incidence (cancer already metastatic at diagnosis) for 11 cancers individually and for all cancers combined. Incidence data combining invasive and in situ cancers were obtained from the original nine Surveillance, Epidemiology, and End Results (SEER) registries, and mortality data were obtained from the National Vital Statistics System.

The analysts then explored implications of the epidemiologic signatures as they pertain to true cancer occurrence (the underlying incidence of clinically meaningful cancer), overdiagnosis (detection of cancers that will not cause symptoms or death), and treatment advances.

Individual cancers

Findings of the analysis, published in a special report in the New England Journal of Medicine, revealed three broad categories of epidemiologic signatures having different implications for the public health and oncology fields.

Desirable signatures showed, for example, declining mortality against a backdrop of stable incidence over the 40-year period, signaling improved treatment, as seen for chronic myeloid leukemia following introduction of imatinib (Gleevec), according to the analysts. Lung cancer incidence and mortality rose and fell in tandem, reflecting an increase in smoking followed by a decrease in response to prevention efforts. Stomach, cervical, and colorectal cancers had both falling incidence – likely reflecting a true decline in occurrence related to prevention and/or screening detection and subsequent treatment of precancerous lesions – and falling mortality.

Undesirable signatures showed a rising incidence juxtaposed with stable mortality and stable or rising metastatic incidence, signaling likely overdiagnosis, Dr. Welch and colleagues proposed. Three cancers—thyroid cancer, kidney cancer, and melanoma—fell into this category; for thyroid cancer and melanoma, fairly recent upticks in metastatic incidence may reflect upstaging.

Finally, some signatures showed mixed signals, with rising incidence and falling mortality. Breast cancer incidence rose and stabilized, coinciding with introduction of screening mammography, and possibly reflecting an increase in true cancer occurrence or overdiagnosis (with stable metastatic incidence favoring the latter), the analysts speculate. Declining mortality since the 1990s may be due to improved treatment or screening, or both. Prostate cancer incidence rose sharply with introduction of prostate-specific antigen screening but then fell to initial levels, suggesting sensitivity of this cancer to diagnostic scrutiny. Falling metastatic incidence indicates screening leads to earlier diagnosis in some cases, while declining mortality starting in the 1990s may again reflect improved treatment or screening, or both.

Next Article: