Content uploaded by Chris I Flowers
Author content
All content in this area was uploaded by Chris I Flowers on Sep 07, 2015
Content may be subject to copyright.
1
doi:10.1038/nrclinonc.2012.126
Breast cancer screening: the questions answered
Philippe Autier, Laura Esserman, Chris I. Flowers and Nehmat Houssami
Abstract | Early detection of cancer has long been thought to be the first step towards eradicating the
mortality associated with the disease. National screening programmes for breast cancer have been
implemented in many countries. However, there is controversy regarding the efficacy and optimal methods of
screening, which is regularly discussed in articles, conferences and even conflicting guidelines. In this article,
Nature Reviews Clinical Oncology asks four experts their opinions on some of the pressing questions associated
with breast cancer screening.
Q. What is the optimal age to start screening?
Philippe Autier. Meta-analyses of randomized trials suggest that a 15% reduction in the risk of death from
breast cancer is associated with annual mammography screening of women 40 to 49 years of age.1 This risk
reduction implies that 1,800 to 2,000 women in this age bracket need to participate in screening to prevent one
breast cancer death. By contrast, 377 women 60 to 69 years of age need to participate every 2 years to
prevent one breast cancer death. In addition to the low efficacy, screening of women younger than 50 years
leads to more false-positive results, and the use of digital instead of film-based mammography also increases
the rate of false-positive screening results. Doses of radiation associated with mammography when screening
postmenopausal women would entail a very low carcinogenic effect. However, the risk of radiation-induced
breast cancer increases with decreasing age, and recent estimates suggest that the 15% reduction in breast
cancer death associated with mammography screening could be offset by breast cancer deaths due to the
carcinogenic effects of mammography X-rays.2 Hence, screening should not be offered to women younger than
50.
Laura J. Esserman. Age 50; this is the age chosen to start screening in almost all countries outside the US
that have instituted screening programmes. Screening should not be routinely offered to women under the age
of 50. Breast cancer risk is lower in women in their 40s, interval cancers (faster growing cancers that become
clinically apparent between routine screens) are also more prevalent,3 and the chance of a false positive is
higher than in older women. These are some of the reasons that routine screening for women in their 40s will
have a low benefit for that population as a whole. It has recently been shown that women in their 40s with at
least a 1.9-fold increase in the risk of developing breast cancer have equivalent benefits from screening as a
woman in her 50s.4 So, understanding risk factors will be important going forward in helping to individualize
screening recommendations for women in their 40s. In the future, risk-based screening is likely to make the
most sense for the whole population.
2
Chris I. Flowers. This is an interesting question, as optimal benefits for screening probably lie somewhere
between 40 and 50 years of age. The choice of lowering the age of screening in the UK to 47 is therefore likely
to be a good move.
http://www.controlled-trials.com/ISRCTN33292440/ or
http://www.publications.parliament.uk/pa/cm201011/cmhansrd/cm110118/text/110118w0003.
htm Parliamentary answers under ‘Breast Screening’.
Nothing magically changes at age 50, but that is the most common age to start screening in European
screening programmes, and concurs with the updated US Preventive Services Task Force guidelines of 2009.
Ref: Screening for Breast Cancer: U.S. Preventive Services Task Force Recommendation Statement
Ann Intern Med. 2009;151:716-726
However, a significant number of cancers arise in women between 40 and 50 with the distribution towards the
upper end of that age group, therefore screening should be offered to some women under 50
When it comes to a public health message, it is simple to say that a woman should have her first mammogram
aged 40 years and continue with annual mammograms, as supported by the ACS, ACR, ACOG and AMA. But,
this message does not recognize, or adapt to. the biology of the disease. A baseline exam could be performed
at age 40 when we can assess breast density (which may increase risk by up to four times), Ref: Boyd NF,
Martin LJ, Bronskill M, et al. Breast tissue composition and susceptibility to breast cancer. J Natl
Cancer Inst 2010; 102:1224–1237. and then potentially could stratify by risk. The downside to this
approach is that we do not currently have the data available to determine who benefits from mammography
and who does not, so saying that screening should depend on personal risk is problematic. There is evidence
that screening starting at 40 is effective in detecting cancer early, The Nelson paper showed a 19%
mortality reduction from the UK Age Trial which should still benefit the patient even if it does not
necessarily reduce mortality. In younger women, we also have to consider the number of productive years of
life saved.
Nehmat Houssami. A recent summary of the evidence from all randomized controlled trials of mammography
screening (women aged 40–74 years) estimated a 19% (intention-to-treat) or approximately 25% (adjusted for
participation) reduction in breast cancer deaths attributable to screening.5 Mammography screening has the
potential to confer benefit in women within that age range; however, the breast cancer mortality reduction in
40–49 year olds is of relatively lower magnitude, estimated as 15% in a meta-analysis.1 In addition, the
associated harms from screening are higher in these younger women relative to older age groups—hence, a
very fine balance exists between the screening benefits and harms in 40–49 year old women.
The offer of screening from age 50 is an appropriate breast screening strategy at a population level—screening
women aged 50–69 years has been shown to be effective in randomized controlled trials,5 in observational
studies of screening programs,6,7 and in modelling various screening strategies.8 However, this should not deny
40–49 year olds access to the same organized or dedicated breast screening services as their older
3
counterparts. In practice, 40–49 year olds frequently express an interest in breast screening, so it is essential
that these younger women are informed about screening benefits and harms (particularly the risk of false
positives) and be supported in making an informed decision about whether or not to commence screening.9
When fully informed through a decision tool a randomized controlled trial showed that younger women were
less inclined to commence screening; however, approximately half wanted to have screening.9
Q. How frequently should patients be offered screening? Does this
depend on risk or should this be a population-level message?
P. A. Shortening intervals between mammography screenings should increase sensitivity and lead to greater
reduction in the incidence of advanced-stage breast cancer, which would in turn lead to greater reductions in
breast cancer mortality. However, in randomized and population studies, the ability of screening to decrease
the incidence of advanced-stage cancer or the risk of breast cancer death is not associated with intervals
between screens.10,11 Higher screening frequency is associated with greater likelihood of a false-positive result
over the entire screening period. There is so far no evidence that in high-risk women, shorter screening
intervals decrease the risk of breast cancer death. Hence, screening every 2 or 3 years is adequate.
L. E. Today, we should offer screening on an every-other-year basis. The data that support a benefit from
screening came from the Swedish trials in which screening was conducted every 2–3 years. Modelling has
shown that screening more frequently has little if any value.8 The data from the Breast Cancer Screening
Consortium not only show that biennial screening leads to the discovery of essentially the same fraction of
stage 2 and higher cancers, but leads to a 50% decrease in the number of unnecessary biopsies performed on
the basis of abnormal mammographic findings.12 The recent analysis that suggested that women with at least
twofold increase in risk (compared to the average risk at age 40) may benefit from screening, but the
recommended frequency for these women is still biennial.4
Going forward, however, we should work towards developing a much more risk-based screening approach.
Today, for example, we recommend that BRCA mutation carriers have annual mammography and annual MRI
screening, staggered at 6-month intervals. However, such recommendations are extended to women with even
moderate risk, which is likely a lot of testing for very little additional value. Instead, we should use additional
biological information to identify lower and higher risk groups, based on breast density, emerging single
nucleotide polymorphisms (SNPs), and models to estimate risk of both genetic and sporadic breast cancer.
C. I. F. When screening programmes were started, it made sense to deal with the ‘single disease’ of breast
cancer by offering regular screenings at a fixed interval. We know much more about breast cancer now than we
did in the 1980s, and already have identified that women at high risk (>25% risk) should have annual
screening with mammograms and MRI. Ref: Kuhl CK, Schrading S, Leutner CC, et al. Mammography, breast
ultrasound, and magnetic resonance imaging for surveillance of women at high familial risk for breast cancer. J
4
Clin Oncol 2005; 23: 8469–76
Warner E, Messersmith H, Causer P, Eisen A, Shumak R, Plewes D. Systematic review: using magnetic
resonance imaging to screen women at high risk for breast cancer. Ann Intern Med 2008; 148: 671–79. We
know that there is a higher risk for women who have dense breasts, and in whom mammography has a lower
sensitivity. However, this risk does not reach 20%, Ref: Boyd NF, Martin LJ, Bronskill M, et al. Breast
tissue composition and susceptibility to breast cancer. J Natl Cancer Inst 2010; 102:1224–1237
and so the addition of expensive tests such as MRI is difficult to justify.
This knowledge leads to the question of how we deal with this complexity at a population level. It may be
optimal for the majority of women to have biennial mammography, but for those women with a more-
aggressive cancer, a shorter interval would work better and increase the chances of successful treatment at an
earlier stage of the disease. We could stratify screening according to risk if we had the data on which to base
the decision. In the meantime, if we are going to screen women younger than 50, then annual or 18-monthly
mammograms make much more sense based on the biology of faster growing tumours in that age group.
Equally, for women over 70, in whom cancer usually grows slower, 3-yearly mammography (as used in the UK)
Ref: Forrest P. Department of Health and Social Security. Breast cancer screening, Report to the
Health Ministers of England, Wales, Scotland and Northern Ireland. London, UK: Her Majesty’s
Stationery Office; 1986] would work.
N. H. A population-level message helps support policy and practice in organized screening programmes that are
aimed at ‘average’ (standard) risk women. Biennial breast screening is effective and has been the population
screening model adopted in Europe, Australia and the UK for most of the duration of these programmes.
Modelling of various screening strategies from Mandleblatt et al.8 provided further evidence that biennial
screening is an efficient and effective population screening strategy as it confers most (on average 81%) of the
benefit that would be achieved with annual screening, and reduces the harms of screening, substantially
reducing false positives and to a lesser extent reducing overdiagnosis.8 The challenge for clinicians is to identify
women who are at sufficiently increased breast cancer risk to warrant a recommendation for more-frequent
(annual) screening. Despite the increased harms, women at increased risk might accept small incremental
benefits (in terms of mortality reduction) from annual, relative to biennial screening, because the trade-off
between the benefit and harm will be viewed differently given their higher underlying risk of breast cancer.
Clearly, I am not talking here about women with breast cancer gene mutations (such as BRCA) who represent a
special group with particularly high risk and who require an entirely different screening approach or primary
prevention strategies.
Q. What information should be provided to women about the benefits,
limitations and harms associated with breast screening?
5
P. A. Information about screening is usually provided by institutions and health professionals involved in
screening. In most instances, that information does not really address uncertainties on health benefits and on
potential harms associated with screening. In addition, the advantages of screening are amplified by the media
and cancer prevention campaigns. As a result, women have expectations that go well beyond the health
benefits that screening can reasonably achieve, while they largely ignore the disadvantages. A recent editorial
summarized the problem: “the public is left totally confused, and usually at the mercy of screening providers.
Before becoming a patient, a healthy individual deserves fully informed consent, with information provided at
the individual and population level”.13 The key items on likely health benefits and potential harms associated
with screening, and their formulation that appear in information materials should be issued by a writing
committee whose members have no professional, academic or financial involvement in screening. Also, before
their first screen, women should be asked to sign a consent stipulating that they have read and understood the
information material issued by the writing committee.
L. E. Women should be aware that by undergoing screening they have an increased risk for callbacks and
biopsies, the majority of which are benign. They need to know that there is some risk that screening will
identify inconsequential as well as consequential cancers, and that there are molecular tests that help better
characterize the risks of disease progression and the benefits of therapy. Women should be told of the risks and
benefits of screening, the chance of a false positive, the importance of paying attention to the symptom of a
new mass or change in their breast shape even if they have had a recent normal mammogram, and be
evaluated for risk factors that would make screening more beneficial.4
C. I. F. Women should be fully educated and informed about the benefits, limitations and harms of screening
before they have their screening. This should also include information about current comorbidities minimizing
the benefits of attending for a screening exam. An understanding of potential false-positive exams, and that a
recall does not necessarily mean you have cancer should also be explained. The difficulties and complexities of
the screening detection of ductal carcinoma in situ (DCIS) should be explained. It is better to introduce the
issues in a way that can be discussed prior to having to deal with the consequence of a DCIS diagnosis at
screening, given the heterogeneous nature of the disease, and the controversies as to what if any treatment
should be offered. The ability of screening to catch cancer at an earlier stage, with potentially less serious
treatment needs to be emphasized as well as the reduced chances of dying of breast cancer.
N. H. It may seem to be stating the obvious, but women should be given a clear message that only
mammography screening has been shown to confer health benefit; this clarification would prevent any
confusion regarding screening using alternative methods, such as MRI and thermography. Age-appropriate
information is necessary given the age-related differences in both benefits and harms of mammography
screening. Information about the potential for screening to reduce breast cancer deaths is generally provided by
screening programmes, so here the focus should be on improving the framing of that information such that it is
both valid and clearly presented, and includes estimates of the effect of screening in absolute (rather than
relative) terms—some examples are available in decision aids that have been tested in practice.9,14 Information
6
on the harms of screening is an area that needs more attention, and should include accurate information on the
possibility of false-negative screens and false-positive screening, which (in my opinion) represents the largest
harm from screening and also includes the risk of biopsy or surgical intervention for non-malignant findings.
Overdiagnosis is emerging as a serious issue; we need to investigate how to inform women about it given the
uncertainty around the true magnitude of overdiagnosis from breast screening. Before incorporating
information on overdiagnosis, it will be critical to examine how women perceive and understand the divergent
estimates of overdiagnosis, and whether (and how) it affects their decision to have screening—this is a priority
area for screening research.
Q. Studies using different methods have obtained highly variable results
on effectiveness of screening in general populations. So, how should
effectiveness of breast screening be evaluated?
P. A. The goal of cancer screening is to reduce the risk of cancer death by detecting cancers when they are not
yet clinically apparent, at a stage when they are less life threatening and more curable. So, reduction in the
incidence of advanced-stage cancer is an early indicator of screening effectiveness. This indicator has the
immense advantage of not being influenced by subsequent treatments. Cancer registry data from areas where
breast screening is widespread for 7 years or more show no decline in the incidence of breast cancers.11 Routine
screening programmes can be evaluated most readily by time trends and differential mortality from the disease
for which screening is being performed.15 A study on temporal trends of breast cancer death that mimicked the
Nordic study on cervical cancer screening effectiveness showed no difference in mortality reduction between
countries with longstanding breast screening programmes and countries where screening was implemented 19
to 15 years later.16 Case-control studies should not be used to evaluate the effectiveness of breast screening
because they cannot disentangle the role of screening and of treatment in mortality changes.
L. E. Certainly not by looking at the number of early stage cancers as a fraction of all cancers diagnosed, as
that figure can be very misleading and significantly inflate the value of screening.17 The number of overall
cancers, the biology of those cancers and the trends over time are critical to report. If the goal is to stage shift,
you should not only see an increase in early stage cancers, but a concomitant drop in the later stage cancers,
which we have not seen in the population data.
Over time, in the randomized trials, there should be a mortality reduction, or certainly a breast cancer mortality
reduction. Of course, today, screening is not the only contributor to the reduction in mortality, which makes it
complicated, but modelling can be very helpful in understanding the overall and relative contributions of
screening and treatment.18
C. I. F. The effectiveness of a test should be judged according to a person’s response to that test. In other
7
words, measurements of screening should be based on the women who attended screening, and the number of
women who needed to be screened (NNS), rather than number of women who needed to be invited (NNI).19
Using this approach during modelling, significantly affects the perception of that test.
N. H. Population health decisions regarding adopting breast screening are based on evidence from randomized
controlled trials that mammography breast screening reduces breast cancer deaths.5 Non-randomized studies of
various designs can be used to evaluate the impact of screening in practice, but can be affected by design
limitations that may bias their results,6 and explains some of the variability in reported estimates of screening
effect. At present, the challenge in breast screening is to evaluate its impact in well-designed observational
studies6 that use an appropriate control group for the group receiving screening, and identify potential or actual
confounding variables and integrate methods that adjust for these confounding factors. Here, I refer to factors
that affect risk of breast cancer and its outcome (such as use of hormone-replacement therapy) and variables
known to have an effect on breast cancer mortality, in particular treatment-related factors. If these factors
significantly differ between the group being screened and the comparison group, this will bias or distort
estimates of screening effect. Because of the progressive improvement in breast cancer therapy, it is important
that screening programmes invest time and resources into high-quality observational studies for future
monitoring of the impact of breast screening at the population level.6
Q. What is the amount of overdiagnosis in populations in which breast
screening is widespread?
P. A. Estimations of overdiagnosis (that is, the diagnosis of a condition that would not have become clinically
significant had it not been detected by screening) vary from less than 5% to 50%. Studies finding low
overdiagnosis rates have often recourse to statistical adjustments (for example, on lead time) or complex
modelling, and usually omit in situ breast cancer. Before the screening era, in situ cancer represented about 2–
4% of all breast cancers. In areas with widespread screening, 10–15% of cancers are detected in situ, most of
which would not have become invasive in the absence of treatment. In Norway, where high-quality population-
based cancer registry data exists, where participation in biennial screening is high, and where data on
hormone-replacement therapy use is available, a study estimated that 15–25% of invasive breast cancers
would be overdiagnosed.20 If in situ cancers are also considered, then overdiagnosis may be in the region of
25–35%.
L. E. The likelihood of overdiagnosis is in the range of 20–30% of screen-detected cancers.21,22 Furthermore,
this figure does not include the in situ lesions. After taking 60,000 DCIS lesions out of the population for well
over 10 years (enough time for these lesions to have matured into invasive cancer), we have not seen the
concomitant decline in invasive cancers that we would have expected.23 Here, we have not shown that early
detection is necessarily of benefit.
8
Any time there is widespread screening, we will necessarily identify tumours that display a range of biological
behaviours. We have to recognize that finding indolent lesions—or IDLE tumours (InDolent Lesions of Epithelial
origin)—is going to be a part of screening,17 which will allow us to develop strategies for minimizing
overtreatment and reducing thresholds for biopsy.24
C. I. F. The answer to this question depends on what you mean by overdiagnosis. Some recent papers suggest
that this should be applied to finding any condition, including cancer, which does not eventually kill you. Ref: 1.
Elmore JG, Fletcher SW. Overdiagnosis in breast cancer screening: time to tackle an underappreciated harm.
Ann Intern Med. 2012;156(7):536-7 2. Kalager M, Adami HO, Bretthauer M, Tamimi RM. Overdiagnosis of
invasive breast cancer due to mammography screening: results from the Norwegian screening program. Ann
Intern Med. 2012;156(7):491-9.
This is a new definition, which many would not accept. Screening is there to catch early stage breast cancer,
and saying we should not screen for DCIS or low-grade cancer at this point is illogical. It is acknowledged that
there are slow-growing cancers that may never be lethal in the woman’s lifetime. Unfortunately, we cannot
identify which women will develop this type of low-grade cancer by imaging alone; a biopsy in which we can
measure tissue biomarkers is required. Similarly, with DCIS, research is helping us to determine how we might
differentiate between low or intermediate-grade and high-grade DCIS. Refs: 1. Jensen RA, Page DL. Ductal
carcinoma in situ of the breast: impact of pathology on therapeutic decisions. Am J Surg Pathol.
2003;27(6):828-31 2. Evans A, Clements K, Maxwell A, Bishop H, Hanby A, Lawrence G, Pinder S E
on behalf of the Sloane Project Steering Group. Lesion size is a major determinant of the
mammographic features of ductal carcinoma in situ: findings from the Sloane Project. Clinical
Radiology 2010; 65: 181-184 3. Sunil S. Badve et al. Correlation between the DCIS score and
traditional clinicopathologic features in the prospectively designed E5194 clinical validation study.
J Clin Oncol 30, 2012 (suppl; abstr 1005)]
High-grade DCIS is the nearest thing we have to a precursor lesion, and should be treated, whereas low-grade
DCIS might be amenable to active surveillance, similar to that used in prostate cancer. However, until this type
of conceptual approach is widely accepted, and until we can identify which women will go on to develop these
lesions, then screening should continue as is. Treatment of these lesions should be personalized according to
risk, which is the responsibility of surgeons and oncologists.
N. H. There is little doubt that overdiagnosis occurs in breast screening; however, currently available estimates
range between none to over 50%, and some of the credible estimates are in the range of 10–25% for invasive
breast cancer alone.20,25 One of the difficulties is the absence of a standard definition for overdiagnosis25,27
(hence differences in the denominator used and the proportion that is being measured in various studies), as
well as whether overdiagnosis refers to DCIS or invasive cancer or both. A study by de Gelder et al.26 showed
that the amount of overdiagnosis could vary by a factor of 3.5 depending on the denominator in the calculation.
Another problem relates to differences in methodology as highlighted by Biesheuvel et al.25 (for example,
9
whether the assessment is based on cumulative incidence or incidence rates) and the finding that estimates of
overdiagnosis may be affected by bias from differences in the underlying risk of breast cancer in screened and
unscreened populations (if not using data from randomized controlled trials), and by whether the study included
adjustment for lead-time.25 These factors could lead to the amount of overdiagnosis being over or under
estimated.
Q. How does the heterogeneity of breast cancer impact screening
benefit?
P. A. The density of breasts (that is the amount of X-rays retained by breast tissues) influences the ability of
mammography to detect anomalies that indicate the presence of a probably cancerous lesion. Breasts are
denser among premenopausal women and in women taking hormone-replacement therapy. Digital
mammography is more sensitive than film-based mammography for dense breast screening. However, one
large-scale study suggested that digital technology is probably not better for preventing advanced-stage
disease and thus for reducing the risk of breast cancer death.27 Ultrasonography is often used in dense breasts
but the real ability of this technology to detect potentially fatal cancers missed by mammography remains to be
established.
L. E. Heterogeneity impacts breast cancer screening tremendously. We just have not really considered it or
made it a mainstream issue. Breast cancer is not one disease. Some cancers are slow growing, and may never
progress (indolent), or progress slowly. Others grow rapidly and may appear as interval cancers.28 Surely, the
period prior to diagnosis for these tumours will also vary tremendously, so it cannot be that there is one way to
screen for and prevent breast cancer. Going forward, we should be thinking about other biological risk factors
that can help us to predict the risk of breast cancer by type (for example, molecular subtypes). Breast density,
for example, is the strongest breast cancer risk factor after age and genetic mutations. It is also a candidate
marker of disease associated with a poor outcome and early recurrence, and can potentially be used to
determine screening frequency.29,30 Another example is inherited genetic variation, as measured by common
SNPs that have been identified and/or validated through the Collaborative Oncology Gene-environment Study
(COGS) consortium, an international collaboration supported by European Commission, will be reported in the
next year, hopefully adding tools for risk refinement not just for breast cancer but for which type of breast
cancer.
We need to develop new models to help predict the type of cancer for which women are at risk, and that cancer
type is what should dictate screening frequency. For example, a woman with fatty breast tissue at risk for a
luminal A tumour, may be best screened every 3–4 years. On the other hand, women at high risk for fast-
growing tumours, for example BRCA1 mutation-driven or triple-negative tumours, will be best screened at
earlier ages on an annual or semiannual basis, using MRI alternating with mammography, particularly in the
10
setting of dense breast tissue. Women at risk for luminal B tumours may need screening every 2 years.
C. I. F. As we discover more, we may be able to identify which women will benefit from screening and which
ones will not. Different growth rates of tumours mean that the sampling (screening interval) may need to be
different for different women. Very-slow-growing tumours may not even need to be screened for. Research
needs to be directed towards the determination of who will benefit from the screening test, as well as who will
not, and how to identify the target group before the intervention.
N. H. If screening detected more of the breast cancers that were destined to progress it would enhance
screening benefit. However, if screening preferentially detected more of the biologically indolent cancers
(unlikely to progress or metastasize) then this bias would reduce the impact of screening, and might add to
overdiagnosis. Because ‘lead time’ will vary by cancer type, indolent cancers may be diagnosed by screening a
long time before they become symptomatic, and thus for some patients who die in the following few years from
other causes screening will have been of no benefit. At present, estimates of benefit from the mammography
screening trials included the full spectrum of breast cancers. Therefore, I would expect the impact of breast
cancer heterogeneity to be mostly on how we manage and treat these heterogeneous tumours and on
therapeutic response.
Q. Why do experts sharing the same published data come to different
conclusions on the effectiveness and harms associated with breast
screening?
P. A. Broadly speaking, one type of conclusion is that breast cancer mammography screening is effective and
causes little harm in terms of overdiagnosis and overtreatment. The other type of conclusion is that
mammography screening is not effective, or has limited effectiveness, and causes substantial overdiagnosis and
overtreatment. Those adhering to the first type of conclusion consider that randomized trials on breast
screening represent strong evidence for the ability of screening to reduce the risk of breast cancer death. Those
adhering to the second type of conclusion consider that results obtained by several mammography trials were
due to flaws in design and conduct. Recent arguments indicate that conflicts of interests have an enormous role
in the mammography controversy,31 and that these conflicts involve not only financial interest but also
academic or institutional positions and activities.32 This conflict is the reason why evaluation of, and
recommendations and information on screening need to be issued by people or institutions having no
involvement in screening.
L. E. Different perspectives and biases shape the perception of benefit versus harm. Some people are
motivated because of the fear of missing a benefit, some from the fear of harm. The nature of specialty medical
training has an impact on the conclusions drawn. Defending or attacking screening has taken on religious
11
fervour in some circles and the frequent articles about why only certain trials should be cited inflames the
debate. Everyone is taught that early detection is the answer to cancer mortality reduction. That can make it
more difficult to recognize that for some cancers, early detection will not improve outcome, either because the
tumour is not destined to become fatal, or because it has the capacity to metastasize early on and may grow at
such a pace that serial detection will have minimal impact.
Whether the benefit of screening is considered significant depends on how people weigh risks and benefits—this
assessment is often subjective when made on a patient-by-patient basis. However, screening is a common
good, conducted on large numbers of patients to benefit a few. We do not know the absolute benefit for a given
patient, although we may understand the benefit for the population. In making recommendations, we often fail
to take an epidemiological view of the data, and to put screening into the perspective of treatment and biology.
The ability to tailor screening to individual risk factors should help to ameliorate this situation.
Regardless of the absolute benefit of screening, it is less than we had hoped, reduced by modern treatment
strategies, and will not be sufficient to eliminate the risk of dying of breast cancer. It is time for us to turn our
energy and attention to how to improve screening and to tailor recommendations to risk for specific breast
cancer subtypes and to focus on prevention.
C. I. F. We all read journals mainly within our subject interest. When we write up a study, it is submitted to a
journal thought most likely to publish our findings. Therefore, there is self selection even within publication. We
are all also guilty of selectively quoting papers that support our cause, and criticizing those that do not. It is the
same as in a court of law. No attorney is going to produce evidence that goes against what they are fighting
for. The way things are going it is unlikely that we will ever reach a consensus. It is far more important to
understand breast cancer better and personalize imaging and screening for that patient., However, to do that
we first need the data with which we can develop appropriate tools.
N. H. There are various reasons for the different conclusions and some of these I have touched on earlier in
discussing differences in study methods and design. Even when using the same available data or the same
published trials, conclusions can differ because, broadly speaking: different methods may have been used to
analyse, adjust, or to pool data, and, in the case of evidence synthesis, different criteria may have been applied
to include or exclude studies; the specific metric used to define or measure effectiveness (or harm) might have
been different even if based on the same data sources, for example whether effect was measured directly by
estimating mortality reductions or by estimating life-years gained8 or measured indirectly by looking for
reductions in advanced-stage cancer rates;10,33 and, finally, when considering benefit and harm the
interpretation of the trade off is inherently shaped by personal values and experience.
The contributors*
Philippe Autier
Laura Esserman
12
Chris I. Flowers
Nehmat Houssami
REFERENCES
1. Nelson, H. D. et al. Screening for breast cancer: an update for the U.S. Preventive Services Task
Force. Ann. Intern. Med. 151, 727–737 (2009).
2. Berrington de González, A. & Reeves, G. Mammographic screening before age 50 years in the UK:
comparison of the radiation risks with the mortality benefits. Br. J. Cancer 93, 590–596 (2005).
3. Lin, C. et al. Locally advanced breast cancers are more likely to present as Interval Cancers: results
from the I-SPY 1 TRIAL (CALGB 150007/150012, ACRIN 6657, InterSPORE Trial). Breast Cancer Res.
Treat. 132, 871–879 (2012).
4. van Ravesteyn, N. T. et al. Tipping the balance of benefits and harms to favor screening
mammography starting at age 40 years: a comparative modeling study of risk. Ann. Intern. Med. 156,
609–617 (2012).
5. Glasziou, P. & Houssami, N. The evidence base for breast cancer screening. Prev. Med. 53, 100–102
(2011).
6 Harris, R., Yeatts, J. & Kinsinger, L. Breast cancer screening for women ages 50 to 69 years a
systematic review of observational evidence. Prev. Med. 53, 108–114 (2011).
7 Schopper, D. & de Wolf, C. How effective are breast cancer screening programmes by mammography?
Review of the current evidence. Eur. J. Cancer 45, 1916–1923 (2009).
8 Mandelblatt, J. S. et al. Effects of mammography screening under different screening schedules: model
estimates of potential benefits and harms. Ann. Intern. Med. 151, 738–747 (2009).
9 Mathieu, E. et al. Helping women make choices about mammography screening: an online randomized
trial of a decision aid for 40-year-old women. Patient Educ. Couns. 81, 63–72 (2010).
10 Autier, P., Héry, C., Haukka, J., Boniol, M. & Byrnes, G. Advanced breast cancer and breast cancer
mortality in randomized controlled trials on mammography screening. J. Clin. Oncol. 27, 5919–5923
(2009).
11. Autier, P. et al. Advanced breast cancer incidence following population-based mammographic
screening. Ann. Oncol. 22, 1726–1735 (2011)
12 Hubbard, R. A. et al. Cumulative probability of false-positive recall or biopsy recommendation after 10
years of screening mammography: a cohort study. Ann. Intern. Med. 155, 481–492 (2011).
13 [No authors listed] The trouble with screening. Lancet 373, 1223 (2009)
13
14 Mathieu, E. et al. Informed choice in mammography screening: a randomized trial of a decision aid for
70-year-old women. Arch. Intern. Med. 167, 2039–2046 (2007).
15. Vainio, H. & Bianchini, F. (Eds) IARC Handbooks of Cancer Prevention. Volume 7. Breast Cancer
Screening. (IARC Press, Lyon, 2002).
16. Autier, P., Boniol, M., Gavin, A. & Vatten, L. J. Breast cancer mortality in neighbouring European
countries with different levels of screening but similar access to treatment: trend analysis of WHO
mortality database. BMJ 343, d4411 (2011)
17. Esserman, L., Shieh, Y. & Thompson, I. Rethinking screening for breast cancer and prostate cancer.
JAMA 302, 1685–1692 (2009).
18. Berry, D. A. et al. Effect of screening and adjuvant therapy on mortality from breast cancer. N. Engl. J.
Med. 353, 1784–1792 (2005).
19. Hendrick, R. E. & Helvie, M. A. Mammography screening: a new estimate of number needed to screen
to prevent one breast cancer death. AJR Am. J. Roentgenol. 198, 723–728 (2012).
20. Kalager, M., Adami H. O., Bretthauer, M. & Tamimi, R. M. Overdiagnosis of invasive breast cancer due
to mammography screening: results from the Norwegian screening program. Ann. Intern. Med. 156,
491–499 (2012)
21. Welch, H. G. & Black, W. C. Overdiagnosis in cancer. J. Natl Cancer Inst. 102, 605–613 (2010).
22. Esserman, L. J. et al. Impact of mammographic screening on the detection of good and poor prognosis
breast cancers. Breast Cancer Res. Treat. 130, 725–734 (2011).
23. Ozanne, E. M. et al. Characterizing the impact of 25 years of DCIS treatment. Breast Cancer Res.
Treat. 129, 165–173 (2011).
24. Esserman, L. & Thompson, I. Solving the overdiagnosis dilemma. J. Natl Cancer Inst. 102, 582–583
(2010).
25 Biesheuvel, C., Barratt, A., Howard, K., Houssami, N. & Irwig, L. Effects of study methods and biases
on estimates of invasive breast cancer overdetection with mammography screening: a systematic
review. Lancet Oncol. 8, 1129–1138 (2007).
26 de Gelder, R. et al. Interpreting overdiagnosis estimates in population-based mammography
screening. Epidemiol. Rev. 33, 111–121 (2011).
27 Kerlikowske, K. et al. Comparative effectiveness of digital versus film-screen mammography in
community practice in the United States: a cohort study. Ann. Intern. Med. 155, 493–502 (2011)
28 Esserman, L. J. et al. Biologic markers determine both the risk and the timing of recurrence in breast
cancer. Breast Cancer Res. Treat. 129, 607–616 (2011).
14
29 Schousboe, J. T., Kerlikowske, K., Loh, A. & Cummings, S. R. Personalizing mammography by breast
density and other risk factors for breast cancer: analysis of health benefits and cost-effectiveness.
Ann. Intern. Med. 155, 10–20 (2011).
30 Kerlikowske, K. & Phipps, A. I. Breast density influences tumor subtypes and tumor aggressiveness. J.
Natl Cancer Inst. 103, 1143–1145 (2011).
31 Quanstrum, K. H. & Hayward, R. A. Lessons from the mammography wars. N. Engl. J. Med. 363,
1076–1077 (2010).
32. Bewley, S. Effect of population-based screening on breast cancer mortality. Lancet 379, 1296 (2012)
33 Ciatto, S. et al. Proportional incidence and radiological review of large (T2+) breast cancers as
surrogate indicators of screening programme performance. Eur. Radiol. 22, 1250–1254 (2012).
International Prevention Research Institute (iPRI), 95 Cours Lafayette, 69006 Lyon, France (P. Autier). UCSF
Carol Franc Buck Breast Care Center, 1600 Divisadero Box 1710, San Francisco, CA 94115, USA (L. Esserman).
H. Lee Moffitt Cancer Center & Research Institute, 12902 Magnolia Drive, Tampa, FL 33612, USA (C. I.
Flowers). Screening and Test Evaluation Program, School of Public Health (A27), Sydney Medical School;
University of Sydney, Sydney 2006, Australia (N. Houssami).
Correspondence to:
philippe.autier@i-pri.org, laura.esserman@ucsfmedctr.org, chrisflowers@mac.com and
nehmath@med.usyd.edu.au