How Frequently should Women Undergo Mammography Screening?

“…mammography screening effectiveness in reducing BC mortality is achieved by earlier detection and therefore enables an increased cure rate of slow-growing disease.”

At the end of the 1980s, scientific evidence was available from randomized trials demonstrating that active invitation to mammography screening could lead to a substantial reduction in breast cancer (BC) mortality as compared with noninvitation [1]. Such evidence, and the relevance of BC as a frequent and lethal disease, justified the recommendation of mammography screening as a current preventive procedure [2], and free-of-charge population-based screening was implemented all over Western Europe by single nations' health services. Based on the existing evidence that demonstated a significant reduction of mortality for those aged 50–69 years [1], this was adopted in most European programs as the reference screening age, whereas no populationbased screening was offered to women aged 40–49 years, for whom the available evidence suggested a small, but not significant, benefit [3].

The decision regarding the recommended screening interval represents a major problem. Screening has to be repeated at regular intervals in order to be effective. At the first screening, only a small proportion of existing BCs have reached a sufficient size to be identified at mammography, while most are too small to be detected. In order to achieve early detection of these BCs, a second screening round must follow, which should ideally occur before these BCs have grown too much, in order to achieve early detection and effective treatment. In addition, considering that new cancers develop over time and enter this scenario, the need for repeat mammography at regular intervals continues until the upper screening-age limit is reached.

On the one hand, the ideal screening interval depends on the average growth speed of BC and, on the other hand, on the magnitude of diagnostic anticipation, which may be achieved by the screening test. The faster the cancer growth, the shorter the interval should be. BC has a relatively slow growth rate; BC cells duplicate every 100–150 days, and it takes approximately 8 years for the first cancer cell to multiply up to the size of a sphere of 1 cm in diameter. To give an example, let us imagine one BC with a diameter of 3 mm (a sphere of approximately 14 mm³), probably not detected at mammography owing to its small size. Assuming, for this BC, an average duplication time of 120 days, it will grow to a diameter of 6 mm in 1 year's time (3 cell duplications = 112 mm³) and to a diameter of 12 mm in 2 years (6 cell duplications = 896 mm³), being still at a limited stage and associated with a favorable prognosis. Indeed, there exist fast-growing BCs, for which even a screening interval of 6 months would not be sufficient to achieve early diagnosis in the asymptomatic status; fortunately, these BCs are a very small minority.

“…there exist fast-growing BCs, for which even a screening interval of 6 months would not be sufficient to achieve early diagnosis in the asymptomatic status…”

When the classic randomized screening trials were in the planning stage, different rescreening intervals had previously been adopted, from between 1 and 3 years, no controlled comparison of the different intervals was available and significant mortality reduction was achieved with all such intervals [1]. Based on the results of these trials, it was also calculated that in those aged 50–69 years, mammography allows an average diagnostic anticipation with respect to BC symptomatic onset (lead time) of 2–3 years, and this was the interval recommended for the implementation of population-based screening programs [2].

Population-based screening is a complex and costly procedure that involves planning to continuously invite and screen millions of women, employing trained and expert staff and sophisticated technical means, thus requiring the investment of significant resources. This is a crucial point to be dealt with by each nation's health service, and the choice of the rescreening interval is a major determinant of screening cost. While adopting a shorter interval (e.g., 1 year instead of 2) will increase diagnostic anticipation and, possibly, screening benefits, this might not be compatible with the existing resources, and the choice of a longer interval might be the only solution to obtain full coverage of the target population by regular screening. To give another example, let us assume that annual screening allows for a 5% additional reduction in mortality (35% instead of 30%) as compared with biennial screening, but, owing to limited resources, only for a 50% coverage of the target population; on the other hand, available resources could cover the reduced cost (actually half) of biennial screening, thus allowing for 100% coverage. It is evident that the choice will favor a biennial interval (which is actually a potentially less effective policy) since this will allow for a higher benefit, with 30% mortality reduction in the whole population, as compared with 35% in half the population; that is, only a 17% reduction, in total.

This is not just theory, since the economical aspect is having a major impact on screening in the European Community. Although the European Community recommendation of implementing screening has been adopted in most European countries, total coverage has been achieved in only a few countries, and where screening has not yet been implemented nationwide, this has not been due to political will, but is simply due to insufficient resources. The UK, one of the first large nations to achieve full national coverage of population-based screening in Europe, was able to do so only with a triennial, rather than a biennial screening interval, and this, apart from the existing evidence of triennial screening efficacy, was mainly motivated by the limitation of resources.

Indeed, it was the strong criticism against the adoption of such a long triennial interval that stimulated the UK NHS to design and perform a randomized trial comparing the adopted triennial interval with the supposedly more effective annual interval. The results of the trial quite clearly demonstrated that the impact of such a sharp reduction of the interval on screening efficacy was effectively null [4]. This finding, while confirming the triennial interval as a valid choice, also called into question the common adoption of a biennial interval throughout Europe, and confirmed the hypothesis that mammography screening effectiveness in reducing BC mortality is achieved by earlier detection and therefore enables an increased cure rate of slow-growing disease.

The problem of the screening interval is crucial in deciding the screening policy in a population-based scenario, where the implementation of a free-of-charge program must be fully paid for by the each nation's health service. This is not the case with other countries, such as the USA, where screening has been running for a long time on a spontaneous self-referral basis, and screening costs are paid for by the patient through personal insurance. Adopting an equally effective but more costly annual interval will contribute to increasing the cost of insurance, at least for those who can afford it.

The increased costs of shortening the interval are not only economical. Screening has several drawbacks, which are usually accepted in view of the screening benefits, but reference standards to limit such drawbacks (e.g., excess recall rate to assessment or excess biopsy rate) are provided [2], and not only for economical reasons. Anxiety while waiting for a screening test or diagnostic assessment results, as well as unnecessary invasive biopsies and unnecessary surgery generated by false-positive results from screening, will double with annual compared with biennial screening. Such psychological pressure, particularly if combined with a more aggressive diagnostic approach, might discourage regular attendance for screening, which might ultimately reduce screening efficacy. Screening is a process requiring multiple screening episodes over a long period of time, and each screening episode, the first as well as the last, might be the one that will save a woman's life.

“Anxiety while waiting for screening test or diagnostic assessment results, as well as unnecessary invasive biopsies and unnecessary surgery generated by false-positive results from screening, will double with annual as compared with biennial screening.”

Screening before the 50 years of age is worth a special mention with regard to the screening interval. Classic randomized trials demonstrated limited screening efficacy that did not justify the adoption of a population-based screening policy [3]. Even a recent randomized trial involving the adoption of annual frequency and utilization of modern mammography did not demonstrate better results as compared with the classic trials that were run 30 years ago [5]. This lack of efficacy has been explained by a lower sensitivity of mammography owing to the masking effect of a radiologically dense breast, a common feature in premenopausal women. Screening from 40 to 49 years of age is recommended on a voluntary basis, after properly informing the patient regarding the pros and cons of the procedure [6], particularly to avoid a gross overestimation of the expected benefit [7]. However, in order to maximize sensitivity, an annual rather than a biennial interval is commonly recommended at this age. Considering that radiologically dense breasts hamper mammography accuracy at any age, adopting a shorter (annual) interval for 50–69 year-old women with dense breasts, and a longer (triennial) interval for those with a fibro-fatty breast has been suggested in the attempt to personalize the screening approach, concentrating the efforts (and the costs) in the population for whom screening is less effective (interval BCs, missed at screening, are more likely to occur in association with dense breasts) [8]. Such an option has some associated problems, from the difficulties of combining two different screening frequencies to the limited interobserver reproducibility of determining breast density; however, this option might be taken into consideration in the future.

In summary, there is no reason to question the presently recommended biennial screening interval for women aged 50–69 years. Personalization of screening frequency in this age-group might be obtained by reducing the interval (e.g., to an annual interval) in women with radiologically dense breasts, and increasing it (e.g., to a triennial interval) in women with fatty breasts, since breast density is inversely associated to mammography sensitivity; however, there is no evidence that such an option, while optimizing sensitivity, will lead to greater efficacy. Screening those aged 40–49 years is not very effective, and might be offered on a spontaneous self-referral basis, with proper information provided regarding the pros and cons. Owing to the low sensitivity of screening related to dense breasts, annual screening intervals are commonly recommended.