Possibility of Breast Cancer Overdiagnosis Should Inform Mammography Decisions

Although their findings of breast cancer overdiagnosis from annual or biennial mammography were not as high as previous studies, the authors of a new study published online today in Annals of Internal Medicine caution their findings also indicate need for increased collaborative decision-making on this screening method.

For their study, they defined overdiagnosis as, “mammographic detection of cancer that would not become clinically relevant in the woman’s remaining lifetime,” and they placed blame on variations in overdiagnosis definitions, study settings, and estimation methods for the 0% to 54% estimate in overdiagnosis range found.

“Knowledge about overdiagnosis is critical for supporting shared decision-making, as recommended by the US Preventive Services Task Force and the American Cancer Society,” they wrote. “However, the risk for breast cancer overdiagnosis in contemporary screening programs remains uncertain.”

Their data on 35,986 women aged 50 to 74 (12.1% Black, 19.0% Asian, 64.4% White) who underwent 1 or more screening mammograms from 2000 to 2018 at a Breast Cancer Surveillance Consortium (BCSC) facility showed 82,677 total mammograms (mean, 2.3 [range, 1-17]) and a 0.87% breast cancer diagnosis rate (n = 718 cancer diagnoses; 79.0% invasive; 21.0% in situ). For those cases determined to be preclinical cancers—those that have not produced symptoms or signs of cancer—4.5% (95% CI, 0.1%-14.8%) were nonprogressive.

The authors’ lead-time estimation method incorporated progressive and nonprogressive cancer probability, and their final model comprised age-dependent incidence of preclinical disease, sensitivity of screening examinations to detect a preclinical cancer, chance of a preclinical cancer to be nonprogressive, and time from onset of progressive preclinical disease to detection with clinical symptoms or signs. Study inclusion was prohibited based on mammography or breast cancer history before a first BCSC-based mammogram, and the final date of follow-up was first breast cancer, death, 2009 for 2 BCSC registries, 2018, and lack of attendance at a BCSC facility.

In addition, when considering contributions from biennial screening, an overall 15.4% (95% CI, 9.4%-26.5%) were characterized as being overdiagnosed. Of this group, 6.1% (95% CI, 0.2%-20.1%; 40% overall) were determined to be indolent, or slow-growing, preclinical cancers. The remaining 9.3% (95% CI, 5.5%-13.5%; 60% overall) were progressive preclinical cancers, or “cancer that would not have progressed to clinical cancer before death from a breast cancer unrelated cause,” wrote the authors.

According to their findings, the study authors posit that for women aged 50 to 74 years who undergo a screening mammography every other year, among those with cancer diagnosed, just over 14% of those cases constitute an overdiagnosis. The rate of overdiagnosis was also shown to almost double among the age range analyzed: 11.5% (95% CI, 3.8%-28.3%) to 23.6% (95% CI, 17.7%-31.9%) from aged 50 to 74 years, respectively.

Further, according to screening round (first or last mammogram, respectively), detected progressive and nonprogressive preclinical cancers influenced the total predicted overdiagnosis rate differently:

• Nonprogressive cancer: 8.4% (95% CI, 0.3%-26.4%) to 5.5% (95% CI, 0.2%-17.0%)
• Progressive cancer: 3.1% (95% CI, 1.6%-5.1%) to 18.1% (95% CI, 11.9%-24.5%)

The authors note that their findings on overdiagnosis are both higher than previous modeling studies, due to differences in screening practices, diagnostic practices, and modeling assumptions, and lower than excess-incident studies, because they are “prone to overestimation.”

An accompanying editorial noted the ongoing controversy surrounding breast cancer overdiagnosis and proposed several solutions:

• Prediction models need to be more accurate to influence treatment decisions
• Screening technologies need to improve to reduce the risks of overdiagnosis and missing breast cancer already missed by mammography
• Screening mammographies could be more effective if they are part of individualized (by risks and preferences) multipronged strategies for cancer risk reduction

Still, the authors of the present study belive their results are strong because of the several sensitivity analyses they conducted. Their Bayesian modeling approach showed lack of influence on their predictions from model structure and prior distribution variations and accounted for parameter uncertainty, respectively. Because of this, they hope their findings will contribute to a more-informed mammography screening process.

“We hope that our findings will bring the field closer to a consensus estimate and facilitate decision-making about mammography screening,” they concluded. “Our estimates of the frequency and the age dependence of overdiagnosis can be provided along with information about false-positive rates to balance estimates of mammography screening benefits as part of a process of shared and informed decision-making.”

Reference

Ryser MD, Lange J, Inoue LYT, et al. Estimation of breast cancer overdiagnosis in a U.S. breast screening cohort. Ann Intern Med. Published online February 28, 2022. doi:10.7326/M21-3577