165
COMMENTARY
Computed tomography screening: safe and effective?
..................................................................................................
INTRODUCTION
‘Is cure necessary in those in whom it may be possible?
Is cure possible in those in whom it may be necessary’1
The urologist Willet Whitmore was referring to the early
detection and treatment of prostate cancer, but this question
is equally applicable to numerous diseases which can now
be detected with powerful screening tools such as multi-slice
computed tomography (CT). It is not easy to communicate
the potential risks from screening and it is rarely a popular
public health message. However, Whitmore’s quotation
elegantly summarizes the dilemma and reminds us that,
despite the clarion call for increased screening, we need to
assess the risks and the benefits carefully before recom-
mending the widespread use of a new technology.
The development of multi-slice CT, enabling fast, low-
dose CT scans to be produced, opened up the possibility of
using it for screening as well as for diagnosis. CT screens of
the lungs, colon, heart and also of the whole body are
available in many countries for asymptomatic individuals
who are willing to pay for them, despite the lack of direct
evidence to date that such screening is efficacious. In the
UK, none of these CT screening exams are recommended by
or paid for by the National Health Service, but it is relatively
easy to obtain them at private facilities. The advertising for
these private screening services rarely highlights the lack of
demonstrated benefit or the potential risks.
There are several potential risks from CT screening,
including the harms associated with over-diagnosis and
false-positive examinations, and also the risk of radiation-
induced cancer. These risks are not different from those
associated with screening with conventional X-rays. Their
magnitude may be greater, however, because the ability to
see more detail may result in the detection of more
abnormalities of uncertain significance as well as more
over-diagnosis, and the higher radiation exposure could
result in more radiation-induced cancers. Therefore, in
addition to recognizing the potential risks, it is necessary to
estimate their magnitude and to make comparisons with
estimates of the absolute benefits (if demonstrated).
The need to consider absolute as well as relative benefits is
an important issue that is often not appreciated. It is
common to say that breast cancer screening reduces breast
cancer mortality by about 30% while the maximum
absolute number of breast cancer deaths that could be
prevented by lifelong mammographic screening is about one
per 100 women screened, because about 4% of women
die of breast cancer.2
EFFICACY
Currently, CT is being evaluated as a potential screening test
for lung and colon cancer, as well as for coronary artery
calcification, a risk factor for heart disease. While it has been
established that CT is a relatively sensitive method for
detecting these diseases, this is not sufficient to recommend
its widespread use for screening. To date, there is insufficient
www.jmedscreen.com
J Med Screen 2007;14:165–168
evidence that these CT screening techniques will reduce
disease-specific mortality or incidence, although several
trials (National Lung Screening Trial, the Nederlands
Leuvens Longkanker Screenings Onderzoek (NELSON) trial
and the National CT colonography trial) are underway.
The US National Lung Screening Trial is designed to
compare the efficacy of annual low-dose lung CT with
conventional chest X-ray at reducing lung cancer mortality
in heavy smokers aged 55 years and over. Recruitment and
screening is complete and follow-up is underway, with the
first results for lung cancer mortality expected around 2009.
The European NELSON trial is designed to compare the
efficacy of low-dose lung CT scans with usual care; screening
started in 2004 and is due to continue through 2010. The
widespread claims about the efficacy of lung CT screening
are not based on results from randomized trials, but on
findings from observational studies such as the International
Early Lung Cancer Action Project.3 These observational
studies suggest that lung CT screening increases survival and
causes a stage-shift. However, a recent pooled analysis of
three observational studies, which used a lung cancer risk
prediction model to generate a theoretical comparison
group, questioned the reduction in mortality and suggested
that the high survival rate seen in previous studies could be
explained by over-diagnosis and lead-time.4 It is necessary
to wait for the results from the ongoing randomized trials to
clarify whether the benefit from lung CT screening is real.
Screening using faecal occult blood testing has already been
shown to be efficacious at reducing colorectal cancer
mortality,5 and trials of flexible sigmoidoscopy are in
progress.6–9 While optical colonoscopy may be the most
sensitive screening tool overall for colorectal cancer,10 there
are concerns about the rate of complications and its
acceptability as a general screening tool. CT colonography
has been proposed as a quicker and less invasive alternative to
colonoscopy, although those with positive test results still
have to undergo an optical colonoscopy for further investiga-
tion and polyp removal. One large multicentre found that CT
colonography sensitivity was similar to optical colonoscopy
(94% for polyps X10 mm),11 but two other large multicentre
studies found significantly lower sensitivity even for large
polyps (o60% for polyps X10 mm).10,12 There is no simple
explanation for this variation although possibilities include
the methods of stool tagging and three-dimensional rather
than two-dimensional primary reading that were used in the
study by Pickhardt et al.11 The first randomized trial
comparing polyp and cancer detection rates in CT colono-
graphy and colonosopy, the US National CT colonography
trial, is due to report its primary endpoint results in 2008.
Although CT colonography is a promising screening tool, data
on its performance in asymptomatic average-risk individuals
are still somewhat limited and the variability in accuracy
suggests that it is not yet appropriate for widespread use.
Multi-detector CT and electron beam CT have both been
investigated as tools for detecting coronary artery calcifica-
tion.13 It has been hypothesized that this screening test
could reduce coronary heart disease events either by
Journal of Medical Screening 2007 Volume 14 Number 4
166
detecting persons at high risk who could benefit from risk
factor modification or by detecting persons with existing
coronary artery stenosis whose lives might be prolonged by
coronary artery bypass grafting surgery or stent placement.
A recent Health Technology Assessment systematic review
did not identify any trials that had assessed whether
addition of coronary artery calcification scores to standard
risk factor assessment would reduce cardiac events.13 In the
observational studies of asymptomatic people with coronary
artery calcification versus those without, the relative risk of
a cardiac event was about four,13 which suggests that it is
unlikely to be a worthwhile screening test.14
If the results from the ongoing trials do show efficacy,
estimates of the absolute disease or mortality reduction are
necessary for risk–benefit analyses. The absolute benefit is
likely to vary according to the disease level in the target
population. For example, because mammography is less
efficacious in pre-menopausal women and breast cancer
incidence rates are lower, the absolute reduction in breast
cancer mortality is considerably lower in pre- than in post-
menopausal women. In the Swedish randomized screening
trials for women aged 40–49 years, the reduction in breast
cancer deaths was 0.6 per 1000 women screened, where as
for women aged 60–69 years it was 2.7 deaths per 1000
women screened after an average of approximately seven
years screening and 16 years follow-up.15 Even if a direct
estimate of the absolute benefit is not available for a
particular population of interest, it can be estimated using
relative risk reductions and incidence-based mortality rates
(for an example, see Berrington de Gonzalez and Reeves16).
RISKS
Over-diagnosis
Over-diagnosis, screen-detected disease that would never
normally have been detected during a patient’s lifetime, is a
well-established risk associated with screening. The harmful
impacts of over-diagnosis are not only the psychological
effect of living with the knowledge of the disease but also
the potential adverse effects from unnecessary treatment,
such as the mortality risks associated with surgery. The high
sensitivity of CT screening compared with standard X-rays
could result in higher levels of over-diagnosis and hence
increased levels of the associated risks.
In their pooled analysis of three lung CT screening studies,
Bach et al4 estimated that approximately three times more
cancers were diagnosed by low-dose lung CT screening than
expected, and that this resulted in 10 times more lung
resections; the absolute annual lung resection rate was about
10 per 1000 screened compared to a predicted rate in an
unscreened population of about 1 per 1000. Previous studies
have found that the postoperative mortality rate following
resection of lung cancer in the United States of America
averages 5%,17 which would translate into an excess annual
mortality rate of approximately 0.5 deaths per 1000 screened.
The natural histories of lung nodules, colon polyps and
coronary artery calcification are not yet well characterized.
The long-term follow-up of the ongoing randomized trials
will provide more information on the levels of over-
diagnosis from CT screening. More accurate estimates of
the associated risks are also needed.
False-positive results
False-positive findings may also be more frequent with CT
than with conventional X-ray screening if the higher
Journal of Medical Screening 2007 Volume 14 Number 4
Gonzalez
sensitivity has been obtained at the expense of lower
specificity. Apart from the anxiety that false-positive results
cause, which is difficult to quantify, there may also be
serious side-effects and complications from unnecessary
biopsies and other diagnostic procedures.
Discrete pulmonary nodules are proving to be a very
common finding on low-dose lung CT screens.18 Even
though a number of features can assist in differentiating
between benign and malignant lung lesions, such as size and
lack of calcification, the proportion of people with abnormal
screen findings has been found to vary widely, from 5% to
51%, in the prevalent screening round.18 In general, the
screen-positive rates were higher in the studies conducted in
the USA (21–51%), lower in the studies conducted in Japan
(5–12%) and varied widely in the European studies
(6–43%).18 Smaller nodules may be followed up with
imaging but larger nodules are typically biopsied, and
possible risks from biopsy include pneumothorax and
respiratory failure.18
Complications from colonoscopy following a positive CT
cololnography screening result include the risk of colonic
perforation, haemorrhage following polypectomy and, more
rarely, cardiovascular events from sedation.19 In a recent
study, 7.9% of average-risk individuals who were screened
with CT colonography underwent a follow-up colonoscopy
with polypectomy and 5.1% received ongoing imaging
surveillance.20 The per screening complication rate will be
lower than for primary screening with colonoscopy, but it is
not yet known what the cumulative rate of colonoscopy will
be after multiple CT screens and therefore what the
potential complication rate is compared to other colorectal
cancer screening tests.
More research is needed to further refine the best cut-off for
a positive screen in order to maximize the probability of a
malignancy being detected while minimizing the risks
associated with follow-up procedures. The variation in
screen-positive rates observed to date for lung CT screening
highlights the current level of uncertainty about the appro-
priate cut-off. Better characterization of the false-positive rates
and the associated risks can be obtained from screening trials
or from observational studies. While relatively small studies
may be sufficient to better characterize false-positive rates,
much larger studies are needed to accurately quantify the
risks associated with follow-up procedures.
Well-standardized protocols could help to improve the
risk–benefit profile from CT screening, similar to those
developed for mammography screening. While reduction of
radiation exposure is an important goal, if doses are reduced
to the extent that image quality is impacted, it is possible to
reach a point where there is only risk from screening and no
potential benefit. Low-quality images can also result in
higher rates of false-positive recall and the associated risks
discussed above. Efforts need to be made to ensure that
optimal protocols are developed and that they are used in
practice both for screening and for follow-up examinations.
The diagnosis of a large number of incidentalomas
‘findings of uncertain significance’ are one of the key
arguments against the use of whole-body CT for screening.
In a recent retrospective study of 1192 mainly self-referred,
asymptomatic individuals who underwent whole-body CT
screening, 86% had at least one abnormal finding, and 37%
received a recommendation for further evaluation.21 Single-
organ screening can also generate a number of incidental
findings in other organs at the periphery of the screen. In
their study of CT colonography, Kim et al.20 reported that
7.7% of the asymptomatic individuals who were screened
had extra-colonic screening findings that resulted in further
www.jmedscreen.com
CT screening: safe and effective?
imaging tests or surgical interventions. Similarly, Swensen
et al.22 reported that 14% of the smokers they screened with
low-dose lung CT had incidental non-pulmonary abnorm-
alities that required further evaluation, and approximately
75% of these findings were of unknown significance. While
these incidental findings are viewed by some as an
additional benefit from CT screening, the levels of further
imaging and interventions need to be monitored, along with
the added anxiety that could be caused.
RADIATION RISKS
Computed tomography involves higher levels of radiation
exposure than most conventional X-rays. This combined
with their widespread use means that they are estimated to
be the largest single cause of radiation-induced cancer from
diagnostic radiation exposures.23 Efforts are being made to
keep the radiation doses as low as possible without
compromising image quality. However, because radiation
exposure increases the risk of cancer for the remainder of
the person’s lifetime, even a small relative risk can cumulate
into a non-negligible lifetime risk of radiation-induced
cancer.24 While it is not practical to quantify these risks
directly, primarily because of the length of follow-up that
would be required,25 they can be estimated indirectly using
risk models developed from existing long-term studies such
as the studies of women who received multiple fluoroscopy
examinations to monitor tuberculosis.24 The results from
these studies show that risk varies according to age at
exposure, with younger age at exposure generally resulting
in higher risks, and also that some organs such as the
thyroid and female breast are more radiosensitive.
These risk projection methods have been used to estimate
the radiation-induced cancer risk for different types of CT
screening. The calculations suggest that the radiation-
induced cancer risk from a CT colonography at the age of
50 years is about 1.4 in 1000, but that this decreases to about
0.7 per 1000 for screening at the age of 70 years.26 The risk
of radiation-induced lung cancer mortality for a female
current smoker from a low-dose lung CT screen at the age 60
years is about 0.5 per 1000 screened.27 To be effective,
screening usually has to be repeated and so, for example, the
cumulative radiation-induced lung cancer risk from three
annual lung CT screens starting at age 60 years would
be about 1.5 per 1000 women screened.
The risks from screening are frequently described as small,
but how small are they compared with the potential
absolute benefits? We can calculate a crude estimate of the
absolute number of lung cancer deaths that could be
prevented by lung CT screening if it is shown to be effective
using the observed death rate in the study by Bach et al.4
The average annual lung cancer mortality rate after screen-
ing was 3.5 per 1000 over the six years follow-up. If
screening had reduced this by 20%, then this suggests that
without screening the average annual mortality rate would
have been 4.4 lung cancer deaths per 1000 screened, and
that after six years follow-up the cumulative absolute
reduction in lung cancer deaths was approximately 5.4 per
1000 people screened. So for this example of a 60-year-old
current smoker who undergoes three annual lung CT
screens, these crude calculation suggest that the net benefit
will be more than halved (5.4
1.5
1.4 ¼ 2.5) after the
potential deaths from radiation and surgery are taken in to
account. This illustration is just for short period of screening
to enable a direct comparison; for longer periods of screen-
ing both the absolute benefits and the absolute risks will be
higher.
www.jmedscreen.com
167
CONCLUSIONS
The efficacy of CT screening has not yet been established.
If the ongoing randomized trials demonstrate that CT
screening is effective, the benefits need to be considered
along with the risks. The example from lung CT screening
highlights the fact that although the risks from screening
might be ‘small’, the potential absolute benefit may also be
relatively ‘small’. Because the balance of the risks and
benefits from CT screening can vary according to the
age and disease level of the screened population, it cannot
be assumed that if screening has been demonstrated to
be safe and effective for one group that it can automatically
be extended to others. Conversely, there may be some high-
risk groups for whom the absolute benefits will be greater
and therefore the risk–benefit profile may be favourable
even if CT screening is not recommended for the general
population. The evidence for, and estimates of, each
associated risk and benefit need to be presented in a
transparent manner, and efforts need to be made to
communicate these risks and benefits effectively to both
physicians and people considering such screening to ensure
that informed decisions are made.
Amy Berrington de Gonzalez
Department of Epidemiology, Johns Hopkins Bloomberg School of
Public Health, 615 N Wolfe St, Baltimore, MD 21205, USA;
aberring@jhsph.edu
REFERENCES
1 Whitmore WF Jr. Natural history of low-stage prostatic cancer and the
impact of early detection. Urol Clin North Am 1990;17:689–97
2 Office for National Statistics. Death registrations, selected data tables,
England and Wales (2006). Available at http://www.statistics.gov.uk/
downloads/theme_population/Table_2_Death_Registrations_Cause.xls
(accessed 26 October 2007)
3 Henschke CL, Yankelevitz DF, Libby DM, et al. International Early Lung
Cancer Action Program Investigators. Survival of patients with stage I lung
cancer detected on CT screening. N Engl J Med 2006;355:1763–71
4 Bach PB, Jett JR, Pastorino U, et al. Computed tomography screening and
lung cancer outcomes. JAMA 2006;297:953–61
5 Hewitson P, Glasziou P, Irwig L, Towler B, Watson E. Screening for
colorectal cancer using the faecal occult blood test, Hemoccult. Cochran
Database Syst Rev 2007;1:CD001216
6 Segnan N, Senore C, Andreoni B, et al. Baseline findings of the Italian
multicenter randomized controlled trial of ‘once-only sigmoidoscopy’ –
SCORE. J Natl Cancer Inst 2002;94:1763–72
7 UK Flexible Sigmoidoscopy Screening Trial Investigators. Single flexible
sigmoidiscopy screening to prevent colorectal cancer: baseline findings of
a UK multicentric randomized trial. Lancet 2002;359:1291–3000
8 Gondal G, Grotmol T, Hofstad B, Bretthauer M, Eide TJ, Hoff G. The
Norwegian Colorectal Cancer Prevention (NORCCAP) screening study:
baseline findings and implementations for clinical work-up in age groups
50–64 years. Scand J Gastroenterol 2003;38:635–42
9 Weissfeld JL, Schoen RE, Pinsky PF, et al. PLCO project team. Flexible
sigmoidoscopy in the PLCO cancer screening trial: results from the baseline
screening examination of a randomized trial. J Natl Can Inst 2005;
97:989–97
10 Rockey DC, Paulson E, Niedzwiecki D, et al. Analysis of air contrast barium
enema, computed tomographic colonography, and colonoscopy: prospec-
tive comparison. Lancet 2005;365:305–11
11 Pickhardt PJ, Choi JR, Hwang I, et al. Computed tomography virtual
colonoscopy to screen for colorectal neoplasia in asymptomatic adults. N
Engl J Med 2003;349:2191–200
12 Cotton PB, Durkalski VL, Pineau BC, et al. Computed tomographic
colonography (virtual colonoscopy): a multicenter comparison with
standard colonoscopy for detection of colorectal neoplasia. JAMA 2004;
291:1713–19
13 Waugh N, Black C, Walker S, McIntyre L, Cummins E, Hillis G. The
effectiveness and cost-effectiveness of computed tomography screening for
coronary artery disease: systematic review. Health Technol Assess 2006;
10:1–41
14 Wald NJ, Hackshaw AK, Frost CD. When can a risk factor be used as a
worthwhile screening test? BMJ 1999;319:1562–5
15 Nystrom L, Andersson I, Bjurstam N, Frisell J, Nordenskjold B, Rutqvist LE.
Long-term effects of mammography screening: updated overview of the
Swedish randomized trials. Lancet 2002;359:909–19
16 Berrington de Gonzalez A, Reeves GK. Mammographic screening before
age 50 years in the UK: comparison of the radiation risks with the mortality
benefits. Br J Cancer 2005;93:590–6
Journal of Medical Screening 2007 Volume 14 Number 4
168
17 Bach PB, Cramer LD, Schrag D, et al. The influence of hospital volume on
survival after resection for lung cancer. N Engl J Med 2001;345:181–8
18 Black C, deVerteuil R, Walker S, et al. Population screening for lung cancer
using computed tomography, is there evidence of clinical effectiveness?
A systematic review of the literature. Thorax 2007;62:131–8
19 Ahlquist DA. Fecal occult blood testing for colorectal cancer. Can we afford
to do this? Gastroenterol Clin North Am 1997;26:41–55
20 Kim DH, Pickhardt PJ, Taylor AJ, et al. CT colonography versus colonoscopy
for the detection of advanced neoplasia. N Engl J Med 2007;357:
1403–12
21 Furtado CD, Aquirre DA, Sirlin CB, et al. Whole-body CT screening:
spectrum of findings and recommendations in 1192 patients. Radiology
2005;237:385–94
22 Swensen SJ, Jett JR, Sloan JA, et al. Screening for lung cancer with low-dose
spiral computed tomography. Am J Resp Crit Care Med 2002;165:508–13
Journal of Medical Screening 2007 Volume 14 Number 4
Gonzalez
23 Berrington de Gonzalez A, Darby SC. Risk of cancer from diagnostic
X-rays: estimates for the UK and 14 other countries. Lancet
2004;363:345–51
24 Committee to assess health risks from exposure to low levels of ionizing
radiation, National Research Council. Health Risks from Exposure to Low
Levels of Ionizing Radiation: BEIR VII. Washington, DC: National Academy
of Sciences, 2005
25 Land CE. Estimating cancer risks from low doses of ionizing radiation.
Science 1980;46:868–73
26 Brenner DJ, Georgsson MA. Mass screening with CT colonography: should
the radiation exposure be of concern? Gastroenterology 2005;129:
328–37
27 Brenner DJ. Radiation risks potentially associated with low-dose CT
screening of adult smokers for lung cancer. Radiology 2004;231:
440–5
www.jmedscreen.com