![]() |
||||||
|
||||||
What is the most appropriate
breast-cancer screening interval for women aged 45 to 49 years in New
Zealand?
Simon Baker, Madeleine Wall, Ashley Bloomfield
Between
December 1998 and June 2004, the New Zealand breast-cancer screening
programme—BreastScreen Aotearoa (BSA)—offered publicly funded,
two-yearly, two-view mammography to all New Zealand women without symptoms of
breast disease aged 50 to 64 years, with the aim of reducing mortality from
breast cancer in this population.
In February 2004, the Minister of Health announced an
extension to the eligible age-range of BSA. In addition to women aged 50 to 64
years already being screened, from 1 July 2004 women aged 45 to 49 years and 65
to 69 years became eligible for publicly-funded screening mammography. This age
extension significantly increases the number of women eligible for
publicly-funded mammography in New Zealand and will substantially increase
demand for breast-screening services.
Women aged 50 to 69 years are screened two-yearly in BSA.
Many international breast-screening experts recommend annual screening for women
under 50, and there is a strong perception that this is best practice for this
age group. One of the key decisions for the age extension was deciding on the
most appropriate screening interval for women aged under 50.
A multidisciplinary Expert Advisory Group was convened to
consider the evidence and to make a recommendation to the Minister. An earlier
version of this review provided the evidence base.
MethodsA systematic search and review of the medical
literature was carried out in May 2004, using the search terms
breast and
interval and
(screening or
cancer) in MEDLINE. A web-based
literature search and retrieval service (PubCrawler) was also used to access
articles, and this was continued up to March 2005. The search terms
screening,
screen,
mammogram, and
mammography were used for searches in
the title field. All titles were reviewed for relevance, and articles retrieved
and reviewed when relevant. Further articles were retrieved from reference
lists.
Any review of breast cancer screening will want to
examine the highest quality evidence available. The randomised controlled trial
(RCT), with mortality as the outcome, is the only type of study designed
specifically to eliminate the effects of biases that are common in studies of
screening programmes—namely lead time bias, length-biased sampling,
overdiagnosis, and selection bias.1,2 However, there is a limited amount of
trial evidence on the most appropriate screening interval for breast cancer.
Consequently, less robust forms of evidence were also examined.
ResultsScreening
interval trials—There have only been two randomised controlled
trials (RCTs) comparing breast screening intervals:
Sojourn
time—The choice of screening interval in a breast-screening
programme depends directly on the epidemiology and natural history of breast
cancer in the age group of interest and, in particular, on the ‘sojourn
time’. The sojourn time is the duration of the period during which a
cancer is symptom free, but potentially detectable by screening. The shorter the
sojourn time of a cancer, the less likely it is that a screening programme will
improve mortality from that disease—as the cancer is more likely to have
spread before it is detected.
Estimates suggest that the sojourn time among women aged 40
to 49 is between 1¼ and 2½ years,5–17 compared to between 3 and
6½ years in women aged 50 to 74.5–11,16,17 It is generally
recommended that the screening interval should not exceed the sojourn time.18 As
the sojourn time is shorter among younger women, a shorter screening interval is
usually recommended for younger women.1,12,14,17,19–22
Doubling
time—The ‘doubling time’ is the time taken for a tumour
to double in volume. In a study from the Netherlands, Peer et al estimated that
the doubling time for breast cancer in women aged 40 to 49 years was 80 days
(95%CI=44–147 days), compared to 157 days (95%CI=121–204 days) in
women aged 50 to 70 years.23
The authors concluded by suggesting that the screening
interval should be shorter in younger women, although they also pointed out that
more frequent screening will not necessarily lead to a clear mortality reduction
in this age group. Tabar et al calculated a doubling time for breast cancer of
178 days among women aged 40 to 49 years, compared to 255 days for women aged 50
to 74 years (p value not given).9
Breast
cancer screening trials—In the RCTs of breast cancer
screening—designed to examine the efficacy of breast screening rather than
the screening interval—women were invited to be screened at a mixture of
12 month, 18 month, 24 month and 33 month intervals—the majority being
screened at two-year intervals.1 Trials using a shorter screening interval for
women aged 45 to 49 have not produced a greater reduction in breast cancer
mortality than those using a longer interval.24,25 Preliminary results from the
UK ‘Age’ Trial suggest a 10 or 11%reduction in predicted deaths at
10 years in 40 to 49 year old women invited for annual screening26,27—less
than in recent meta-analyses of screening in this age group, which included
trials in which women were screened at 24 and 33 month intervals.1,24 An ideal
screening interval has not been determined from these trials.28
Tabar et al analysed the
Swedish Two-County Trial. Among women over 50 years of age at entry to the
trial, relatively few interval cancers were seen in the first two years after a
screening test; in the third year, the rate rose to 45% of the comparable rate
in the control group.14 By contrast, among women aged 40 to 49 years at entry,
the proportion of interval cancers (even in the first post-screening year) was
38% of that in controls, and in the second year, it was 68% (p value<0.001
for difference between age groups), suggesting a shorter screening interval may
be more appropriate in younger women.
Modelling—Some
authors have used mathematical modelling to assess the effect of different
screening intervals. Pelikan and Moskowitz used modelling based on the Health
Insurance Plan (HIP) trial to predict that the mortality reduction for annual
screening of women aged 40 to 49 should be the same, or somewhat less, than that
for two yearly screening of women aged 50 and over.29
Tabar et al used modelling
based on tumour size and nodal status to predict deaths from breast cancer among
women under 50 years,9 and predicted a 12% (RR=0.88, 95%CI=0.54–1.41)
reduction in mortality from breast cancer from two-yearly screening, compared to
a 19% (RR=0.81, 95%CI not given) reduction in mortality from annual
screening.
Feig used data from the
Swedish Two-County Trial, and calculated that if women aged 40 to 49 years were
screened every year, their ratio of interval to control incidence cancers at one
year would have been similar to the same ratio for older women two to three
years after screening, suggesting that annual screening among 40 to 49 year old
women would produce a mortality reduction close to that of women aged 50 to 74
years who are screened every two or three years. Mathematic modelling suggested
a mortality reduction of 35% versus 23% (assuming a 100% compliance rate, p
value not given) for annual versus biennial screening of women aged 40 to
49.30,31
The results of Markov-chain
modelling of breast tumour progression to determine the optimal screening
interval using data from the Swedish trials suggested that the screening
interval is critical for women aged 40 to 49 years, but less so for older women.
Among women aged 40 to 49 years, a three-year screening interval was predicted
to result in only a 4% reduction in breast cancer mortality. A two-year
screening interval was predicted to result in an 18% reduction in breast cancer
mortality, while annual screening should result in a 36% reduction (p value for
differences not given).8,21 Jansen and Zoetelief produced similar results when
modelling data from the Swedish Two County study.32
Chen et al used Markov chain
modelling to predict deaths from breast cancer among women under 50 years.
Results suggested that two-yearly screening would yield a mortality reduction of
10–20%, while annual screening would yield a mortality reduction of
20–30%.33
Boer et al used modelling to
estimate the benefits and harms of different screening intervals for women aged
over 50 years. The authors postulated that longer screening intervals would tend
to pick up slower growing cancers—cancers that may benefit less from early
detection. The authors demonstrated that the ratio of benefits to harms
increases with shorter screening intervals, but that cost-effectiveness
decreases markedly.34
Service
screening—The sensitivity of a screening test is commonly reported.
The sensitivity is the proportion of people with the disease who are detected as
having it by the test. A test with a low sensitivity will miss a lot of cancers.
A test with a sensitivity of 100% will detect all cancers present. Brekelmans et
al analysed data from the DOM project—a breast screening programme
established in Utrecht (the Netherlands) in 1974. Among women aged 40 to 49
years, the age-specific sensitivity at six months was 89%
(95%CI=77–101%)—close to the age-specific sensitivity among women
aged 50 to 64 after 12 months (88%, 95%CI=81–95%). Similarly, the
age-specific sensitivity at one year among women aged 40 to 49 (76%,
95%CI=61-91%), was close to the sensitivity at two years among women aged 50 to
64 (72%, 95%CI=63-81%).35
Feig reviewed an analysis of
the Uppsala programme in Sweden, and concluded that women aged 40 to 49 need to
be screened every 12 months to attain roughly the same mortality reduction as
older women screened every 24–30 months.36
Peer et al examined the
interval cancer rate among women aged 40 to 49 years in the Nijmegen breast
screening programme in the Netherlands. In the second year after screening, the
interval cancer rate for women aged 40 to 49 years approached the expected
breast cancer incidence rate in the absence of screening, leading the authors to
suggest that a two-year interval for women under 50 appears to be too long.
However, this result was heavily influenced by the very poor performance of
mammography in the 40 to 44 age-group (interval cancers making up 94% of the
expected breast cancer incidence rate in the absence of screening). The
performance of screening was better in the 45 to 49 age-group (71%). For
comparison, the rate for women aged 50 to 54 years was 55% (p values for
differences not given).11
Using data from the
University of California San Francisco screening programme, Kerlikowske et al
demonstrated that the sensitivity of screening for women aged 40 to 49 at a
one-year screening interval is virtually the same (84%) as that of screening
women age 50 and older at a two-year interval (86%).37 Breast density did not
influence the sensitivity of mammography in women under 50 years—thus
supporting the view that rapid tumour growth, rather than breast density, is the
main reason for the reduced sensitivity of mammography in younger women.
Proportional incidence is
also commonly measured in screening programmes. The proportional incidence is
the number of interval breast cancers, expressed as a proportion of the number
of breast cancers expected in the absence of screening.38 The higher the
proportional incidence, the less likely it is that the programme will reduce
mortality from breast cancer. A review of the two-yearly BreastScreen Victoria
(Australia) programme revealed a proportional incidence among women aged 40 to
49 years of 59% (95%CI=39–85%) in the first year after screening, and 93%
(95%CI=59–140%) in the second year.38
Hunt et al reported on the
outcome of screening annually versus two-yearly in a screening programme for
American women aged 40 to 79. Although women in their dataset were not assigned
to annual versus two-yearly screening at the outset, Hunt et al were able to
extract estimates of the effect of various screening intervals by sorting women
into those who returned to screening at either 10–14 month intervals
(‘annual’), or those who returned at 22–26-month intervals
(‘two-yearly’). The authors combined this data with information from
the cancer registry on women found to have interval tumours (defined as tumours
not detected at screening that were noted within either one year [annual], or
two years [two-yearly] of a negative screen). They found that the size of the
tumours was smaller in the annual group (p=0.04), suggesting an enhanced value
of annual screening.39 40 Differences between annual and two-yearly screening in
tumour size did not reach statistical significance when the age group 40-49
years was analysed separately.
Broeders et al examined the
Nijmegen breast screening programme in The Netherlands, and calculated a
non-significant, 44% reduction in breast cancer mortality among women aged 45 to
49 years who were screened two-yearly. The authors suggested that, even in a
programme with a two-year screening interval, there may be a benefit of starting
screening around age 45.41
Buist et al examined outcomes among women aged 40 and over,
who developed invasive breast cancer, enrolled in a health maintenance
organisation in Washington State.42 Within 12 months of a negative screen,
interval cancers had occurred in 28% of women aged 40 to 49, and 14% of women
aged 50 plus (OR=2.36, 95%CI=1.14–4.77). Within 24 months of a negative
screen, interval cancers had occurred in 52% of women aged 40 to 49, and 25% of
women aged 50 plus (OR=3.58, 95%CI=2.15–5.97). Greater breast density
explained 68% of the decreased mammographic performance in younger women at 12
months. At 24 months, greater breast density explained 38%, and rapid tumour
growth explained 31% of the decreased mammographic performance in younger women.
The authors concluded that mammography screening of younger women at 12 versus
24 month intervals may remove the adverse effect of faster growing tumours on
mammographic sensitivity, but will not remove the adverse effect of breast
density on mammographic sensitivity.
Taylor et al analysed the New
South Wales (Australia) screening programme, and calculated a proportional
incidence in the 40 to 49 year age group of 56% (95%CI =50–62%) for the
first year, and 86% (95%CI =82–90%) for the second year.43
White et al investigated
whether American women diagnosed with breast cancer after having screening
mammograms separated by a two-year interval were more likely to be diagnosed
with late-stage disease (positive lymph nodes or metastases) than women
diagnosed with breast cancer after having screening mammograms separated by a
one-year interval.44 Women aged 40 to 49 with a two-year interval were more
likely to have late-stage disease than those with a one-year interval (28% vs
21%, OR=1.35, 95%CI=1.01–1.81). No differences were observed in older
women, and there was no indication that women with dense breasts would benefit
more from a one-year rather than a two-year screening interval.
Wai et al compared the
long-term impact of one- and two-year screening intervals among 50 to
74-year-old Canadian women, using prognostic modelling. The shorter screening
interval was associated with a higher proportion of screen-detected cancers,
compared to interval cancers—but this translated to only a 1.2% increase
in the estimated 10-year breast cancer-specific survival among women screened
every year, compared to those screened every second year.45
Meta-analyses—Kerlikowske
et al performed a meta-analysis of the association between screening interval
and mortality across the eight randomised controlled trials. Mortality
reductions for screening every 18–33 months were similar to reductions for
annual screening for women aged 40 to 49.25
The United States Preventive
Services Taskforce (USPSTF) concluded (in their meta-analysis) that annual
screening may be more important among women aged 40 to 49 years—but they
found no direct proof for this hypothesis in the controlled trials that had been
completed at the time the meta-analysis was undertaken.24 However, the authors
noted that some experts recommended annual mammography for women aged 40 to 49
years—based on the lower sensitivity of the test, and on evidence that
tumours grow more rapidly among this younger age-group.46
Taylor et al carried out a
meta-analysis of proportional incidence rates for women aged 40–49 years43
(both for randomised trials of screening, and for service screening), and they
included some of the studies already documented in this
report.12,14,35,37,38,47,48 The meta-analysed proportional incidence rate for
women aged 40 to 49 for randomised trials was 42% (95%CI=21–62%) for the
first year, and 63% (95%CI=55–71%) for the second year—and for
service screening it was 44% (95%CI=31–58%) for the first year, and 72%
(95%CI=51–92%) for the second year.
Overseas
programmes—In Sweden, mammography is recommended every 18 months
for women aged 40 to 49 years;39,49,50 Australia, Iceland, and the Netherlands
(Nijmegen) recommend two-yearly mammography for women aged 40 to 49
years;39,51,50 Canada offers one-yearly or two-yearly screening (depending on
the province); and in the United States, no single group establishes national
policy as different institutions recommend either one-yearly or two-yearly
screening for women aged 40 to 49.50
Other countries (including Portugal, Spain [Navarre], and
Greece) recommend two-yearly mammography for women aged 40 to 49 years.1,52
However, screening interval recommendations do not automatically result in that
screening interval in practice—as capacity is not always present, and
women do not always attend as recommended.
Capacity-limited
studies—Fett et al created a computer model to examine the
mortality benefits of adjusting the balance between population coverage of
screening and screening interval among 50 to 64 year old women. That is, does
screening twice the population half as often produce the same reduction in
mortality? Analysis suggested that (for a given number of screens) the higher
the coverage and the longer the screening interval, the greater the reduction in
breast cancer mortality. For example, screening 30% of the population annually
would produce a 20% reduction in breast cancer mortality, whereas screening 60%
of the population every two years would produce a 29% reduction. The authors
concluded that the results supported policies that seek to deploy resources
available for mammographic screening most evenly across the target
population.6
Prospects
for future research—An RCT to evaluate the relative effect of
different screening intervals (with breast cancer death as the end-point) would
require groups large enough to yield 350 breast cancer deaths in the absence of
screening, and a very long follow-up.4,14 Such a trial would need to be three
times the size of the Swedish trials of 135,000 women, and would need to run for
considerably longer.14 It is unlikely that such a trial will ever be carried
out.
DiscussionLike other countries, New Zealand is experiencing
difficulties meeting demand for breast screening services, due largely to a
shortage of both breast-screening medical radiation technologists (MRTs) to take
mammograms and breast radiologists to read mammograms and carry out assessments.
Screening women (aged under 50 years) every year would place
great demands on limited resources, to the extent that some women would
inevitably need to wait to be screened. However, if women aged 45 to 49 years
were screened every two years in New
Zealand, instead of every one year,
more women could be screened sooner. It is clearly important, however, to make
sure that when using a two-yearly screening interval for women (aged under 50),
the benefit of screening this age group is not diminished.
No robust trial evidence
exists on which to base a decision on the screening interval for women aged 45
to 49 years, and it is unlikely that definitive trial evidence will ever emerge.
Evidence from less robust studies supports a screening interval of one year or
18 months—although in the randomised controlled trials, shorter screening
intervals did not produce greater benefits among women aged 45 to 49
years.
However, it is important to
note that the eight breast screening trials varied on more than just screening
interval. Trials had one or two mammography readers, one or two mammography
views, variable quality of mammography equipment, widely varying recall rates,
and radiologists with differing mammography expertise. All of these factors
independently influence screening performance over and above the screening
interval, thus complicating the interpretation of the RCT data and the
subsequent modelling of the effect of screening interval. Lisby carried out an
evidence-based review, and concluded that there was ‘not enough evidence
to recommend optimal intervals for screening with mammography in women in their
forties’.28
Overseas screening programmes either recommend intervals of
12, 18, or 24 months in women aged 45 to 49 years. There is some evidence that,
in a situation of limited capacity, screening 1000 women every two years
produces a greater mortality benefit than screening 500 women every year. In the
absence of definitive evidence, those charged with determining the screening
interval for women aged 45 to 49 years in a breast-screening programme have to
weigh up the available evidence and consider it alongside other relevant
factors.
The Expert Advisory Group, convened to recommend
implementation policies for BSA Age Extension, considered the evidence
summarised above and the likely ability of the programme to increase capacity to
meet the demand. They decided, after much discussion, to recommend a two-yearly
screening interval for women aged 45 to 49 at this time—taking into
account the available evidence and the ability of the health sector to cope with
the increased workload. The Group also recommended that emerging evidence on the
screening interval be regularly monitored. The Minister of Health accepted these
recommendations.
Disclaimer: The
authors are employees of the New Zealand Ministry of Health. The views expressed
in this paper are the authors’ own and do not represent the views or
policies of the Ministry of Health. The paper was submitted for publication with
the permission of the Director General of Health.
Author information:
Simon Baker, Public Health Physician, National Screening Unit; Madeleine Wall,
BreastScreen Aotearoa Clinical Leader, National Screening Unit; Ashley
Bloomfield, Chief Advisor, Public Health; Ministry of Health, Wellington
Acknowledgement: We
thank the members of the BSA Age Extension Expert Advisory Group for their hard
work in reaching a difficult decision. We also thank Karen Mitchell of the
National Screening Unit, who commissioned and provided feedback on this
work.
Correspondence: Dr
Simon Baker, Public Health Physician, National Screening Unit, Private Bag
92522, Wellesley Street, Auckland. Fax: (09) 580 9195; email:
Simon_Baker@moh.govt.nz
References:
|
||||||
| Current
issue | Search journal |
Archived issues | Classifieds
| Hotline (free ads) Subscribe | Contribute | Advertise | Contact Us | Copyright | Other Journals |