Breast cancer screening can sometimes identify cancers that would never have caused symptoms or threatened a woman’s health during her lifetime. This phenomenon is known as overdiagnosis and has long been considered one of the possible drawbacks of screening.
For decades, researchers have debated how often overdiagnosis actually occurs. Estimates from randomized trials have differed markedly, with some studies suggesting that 30% to 50% of breast cancers detected by screening could fall into this category. Those numbers have influenced international debates about the benefits and potential harms of population-based breast cancer screening.
“The goal of our study was to pool the evidence from all randomized controlled trials to get a clearer picture of the extent of overdiagnosis in breast cancer screening,” says Sisse Helle Njor, professor at the University of Southern Denmark and Lillebælt Hospital. She continues:
“Randomized trials have often been cited as evidence that overdiagnosis is a substantial problem. Our study shows that this interpretation is not as straightforward as it might seem.”
The researchers found that the additional cases of breast cancer detected in randomized trials closely resembled patterns seen in Denmark, where overdiagnosis associated with screening is estimated to be less than 5%.
A new look at mammography trials
To investigate the problem, the team combined and reanalyzed the results of all randomized trials on screening mammograms and compared them with data from Denmark.
Denmark offered a useful real-world reference because organized breast cancer screening began in some regions 17 years earlier than in others. That difference allowed researchers to track how breast cancer diagnoses changed immediately after screening tests were introduced and how those patterns evolved over longer periods.
“When screening is introduced, the number of breast cancer diagnoses initially increases because the cancers are detected earlier than they would have been detected without screening. Over time, this should be followed by a drop, as some of these cancers would otherwise have been diagnosed later. This pattern may also be affected if women in either group continue to undergo screening after finishing the trials, which was common. If researchers do not take these factors into account, the initial increase may be confused with an overdiagnosis,” says Elsebeth Lynge, professor emeritus of the Department. of Public Health, University of Copenhagen.
The researchers compared breast cancer incidence at coincident points in time in randomized trials and in routine screening programs in Denmark. This allowed them to assess whether the patterns were similar and what those similarities might reveal about the true scale of overdiagnosis.
“Taken together, we believe that some previous high estimates of overdiagnosis, which influenced screening guidelines and communication, were based on evidence before trial data had fully matured. When interpreted in their full temporal context, the randomized trial data are consistent with overdiagnosis of less than five percent, rather than estimates close to 50%,” says Matejka Rebolj, senior epidemiologist at Queen Mary University of London.
Facts: overdiagnosis
Overdiagnosis occurs when screening tests detect breast cancer that would never have been life-threatening or caused symptoms during a woman’s lifetime. Without screening, the woman would never have known the cancer was present.
The definition may also include women who die from another cause shortly after receiving a breast cancer diagnosis. In these cases, screening may have offered little benefit because poor health or limited life expectancy meant that finding and treating cancer was unlikely to improve health or prolong life.
Why time can change numbers
A key issue is that screening tests change when cancer is diagnosed.
When mammography is introduced, more cancers are detected earlier, leading to an initial increase in the number of diagnoses. Some of those cancers would have eventually been found even without screening, meaning diagnosis rates should decline later as those cases progress over time.
If a study ends before enough time has passed for that decline to become visible, researchers may incorrectly interpret some of the initial increase as overdiagnosis. Estimates may also be distorted when women in control groups are later screened.
The new analysis suggests that taking these time effects into account can substantially change estimates of how often mammography identifies cancers that would otherwise never have caused a problem.
What overdiagnosis means for women
Understanding both the benefits and potential drawbacks of screening is important for women to decide to participate in breast cancer screening programs.
“The majority of women will not develop breast cancer, but with this study we can now be sure that the benefits of detecting breast cancer early and preventing premature death will outweigh the small risk of unnecessary treatment,” says Sisse Helle Njor.
“With this in mind, we hope that this study will provide a framework for a more realistic interpretation of the evidence and help us better inform women when they are invited for screening.”
Facts: about the study
The researchers conducted a new analysis of existing research on mammography screening.
Their analysis included the eight randomized trials in this field: the New York Health Insurance Plan, Malmö, Two Counties, Edinburgh, the Canadian National Breast Screening Study, Stockholm, Gothenburg and Age UK.
Two regional screening programs in Denmark were used as references. The researchers examined both invasive breast cancer and ductal carcinoma in situ (DCIS).
In reevaluating previous trials, the team focused on three factors that may influence estimates of overdiagnosis:
- Whether women in the control group were subsequently screened, for example after the trial ended or after routine screening programs were introduced.
- The number of screening rounds offered to women in each group. Additional rounds of screening create more opportunities to detect cancers earlier.
- The duration of follow-up of women in the screening and control groups. Longer follow-up gives more time for cancers detected earlier by screening to eventually appear among women in the control group.
After accounting for these differences in screening exposure and follow-up, the researchers concluded that overdiagnosis may be substantially less common than previous estimates suggested.
Funds
Casper Urth Pedersen is supported by the Novo Nordisk Foundation (reference: NNF22OC0076184) and Matejka Rebolj is supported by Cancer Research UK (reference: C8162/A29083).