Recommending Mammography Screening beyond 80 Years of Age: A Time for Caution

Summary of methods & results

The study of Badgwell et al. is based on a Surveillance, Epidemiology and End Results (SEER)–Medicare-linked large series of elderly breast cancer patients with detailed diagnostic, staging and survival information [1]. Results suggest that increasing exposure to mammography is associated with:

The latter finding suggests a typical ‘healthy screening effect’, with women undergoing regular mammography being different from nonusers with regards to lifestyle, healthcare and life expectancy. Based on their findings, the authors suggest that mammography screening may reduce breast cancer-specific mortality, and that mammography surveillance beyond the age of 80 years should be considered, informing women on possible screening benefits. Nevertheless, there are several aspects that have not been sufficiently dealt with by the authors, which may suggest much more cautious recommendations, as screening beyond the age of 80 years might turn out to be more detrimental than beneficial. These aspects are worth careful discussion.

Stage shift

Stage shift towards earlier stages among screen-detected as compared with nonscreen-detected cancers is necessary for screening efficacy. Screening may reduce mortality only by early detection at a more curable stage. A favorable stage shift was observed in this study with increasing mammography exposure. Dealing with a breast cancer series, stage shift could be measured only according to the proportion of different stages among total cancers, and not as the rate of cancers with a given stage among screened and unscreened populations. The former method (stage proportion) may be strongly biased in the presence of overdiagnosis, specifically, the detection of slow growing, indolent and not life-threatening cancers that are not likely to surface clinically during a lifetime. The most convincing overdiagnosis estimates in the screening age (55–69 years) are in the range of 5–10% [2]. Overdiagnosis depends on life expectancy and on competitive mortality causes, and its magnitude increases with age [3]; overdiagnosis over the age of 80 years is probably more than 10%, as estimated for women aged 50–69 years. Overdiagnosed cancers are most probably at an early (0–I) stage; thus, increasing mammography exposure will detect early-stage, overdiagnosed cancers that will have a diluting effect and will reduce the proportion of advanced stages among total cancers. We do not deny that use of mammography may detect an earlier-stage, clinically significant cancer, which is likely to surface at an advanced stage in the absence of screening, but the favorable stage shift that is claimed to be a screening benefit in the present study, on the basis of a reduced proportion of advanced stages with increasing mammography exposure, has probably been overestimated as the analysis did not take into account that mammography exposure, particularly in the considered age range, is probably associated with a substantial amount of overdiagnosis.

Healthy screening effect

The finding that increasing mammography exposure was also associated to overall and non-breast cancer-specific survival accounts for a typical healthy screening effect. Women seeking mammography surveillance, as compared with mammography nonusers, are more breast cancer aware and are more prone to adopting a healthy behavior and lifestyle (e.g., no smoking, healthy diet, breast self-examination and prompt referral to medical care when diseased). It may also be expected that these women, in case they develop breast cancer, seek medical advice sooner and are more likely to comply with proposed treatment and follow-up regimens and, thus, are exposed to a lower risk of dying from breast cancer, independent of its early or late detection. Therefore, the more favorable breast cancer-specific survival observed for mammography-exposed women as compared with nonusers might be explained, at least to some extent, with a healthy screening effect rather than with a beneficial effect of early detection. This aspect has been accounted for by the authors who, nevertheless, do not sufficiently stress that such a bias might totally flaw their conclusions as to the benefits of screening.

Lead time & survival

Early detection through screening implies diagnostic anticipation. The average early-detection lead time by mammography screening between 50 and 69 years of age has been estimated to be in the range of 2–3 years. Detection lead time increases with age; this may be owing to a different age-specific growth rate of cancer, but is definitely dependent on the sensitivity of mammography, which is known to increase with age, as shown by proportional interval cancer incidence, which clearly decreases with age [4]. Mammography sensitivity increases mostly with age since the masking effect of radiologically dense breast parenchyma subsides with menopause, and diagnostic anticipation increases as the breast becomes fatty and radiolucent. Such fatty-fibrous involution is almost the rule over the age of 80 years, and detection lead time is probably in the range of 3–5 years. Cancer survival is measured from the date of diagnosis and, in case of early detection, survival will always increase as the date of diagnosis is anticipated, even when early treatment will not prolong life, as compared with late diagnosis. Thus, when comparing subjects without early detection (e.g., mammography nonusers) with subjects with early detection (e.g., subjects with increasing mammography exposure), survival curves should be corrected according to the detection lead time. For example, if we assume that the detection lead time at the age of 80 years or above is approximately 3 years, then the 5-year survival in mammography nonusers should be compared with the 8-year survival in mammography users. Such a correction was not taken into account in the present study while, if adopted, would substantially – if not totally – reduce the observed survival difference.

Life expectancy

Although life expectancy over the age of 80 years may be as much as 7–10 years, it must be taken into account that the reduction in mortality by screening begins to be evident from mortality curves 6–7 years after the first screening [5]. Such a delay in the occurrence of visible mortality differences between screened and unscreened subjects is obvious; given a subject in whom clinical cancer would have surfaced in 1998 and killed the patient in 2004 (average clinical breast cancer survival = 6 years), and supposing that screening would anticipate diagnosis in 1996 (lead time = 2 years) and prolong life until 2008, no life prolongation could be claimed until 2004 (expected date of death in absence of screening), that is, 8 years after the date of detection at screening. Considering that lead time over 80 years of age is likely to be in the 3 to 5 years range, the assumption that subjects in whom cancer is screen-detected at this age will have a sufficient life expectancy to allow for screening benefit to become real is not that obvious, and should be carefully demonstrated, considering that previous modeling studies do not favor such a hypothesis [6].

Overall, the study of Badgwell et al. apparently provides evidence of earlier detection and of a favorable impact on survival with mammography surveillance over the age of 80 years, but the study findings have not been properly analyzed, since major possible systematic biases have not been taken into account, which might substantially flaw the study conclusions. Beyond the age of 80 years, mammography surveillance has a considerable chance of detrimental effects (e.g., overdiagnosis and overtreatment), whereas benefits are, as yet, unknown; until additional scientific evidence of screening benefits becomes available, allowing for a correct cost-effective evaluation, recommending screening beyond the age of 80 years appears to be highly questionable.

Future perspective

Although it is unlikely that new controlled trials of breast cancer screening in elderly people will be carried out in the near future, proper analysis of existing clinical material, taking into account all possible biases, might provide new convincing evidence of the screening benefit beyond 80 years of age. Moreover, life expectancy has been constantly increasing in recent decades and if such a trend continues, there will be increasing ‘space’ for the benefit of screening of elderly women to take place. Continuous improvements in the identification of aggressive or indolent cancer, offering new opportunities to identify overdiagnosed breast cancers that could be managed just by active surveillance, as well as less invasive treatment options might reduce the burden of overtreatment, which is presently one of the main drawbacks of cancer screening of elderly people. The combination of all these possible events might make routine screening beyond the age of 80 years a reasonable future option.