Journal Of Medical Screening

165

COMMENTARY

Computed tomography screening: safe and effective?

.................................................................................................. J Med Screen 2007;14:165–168

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 recommending the widespread use of a new technology. The development of multi-slice CT, enabling fast, lowdose 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 radiationinduced 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

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 investigation 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 colonography 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 calcification.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

Gonzalez

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 postmenopausal 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 overdiagnosis 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

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 appropriate 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 Singleorgan 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 abnormalities 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 radiationinduced 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 screening 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 screening 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 highrisk 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; [email protected]

REFERENCES 1 2

3 4 5 6 7 8

9

10 11 12

13

14 15 16

Whitmore WF Jr. Natural history of low-stage prostatic cancer and the impact of early detection. Urol Clin North Am 1990;17:689–97 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) 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 Bach PB, Jett JR, Pastorino U, et al. Computed tomography screening and lung cancer outcomes. JAMA 2006;297:953–61 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 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 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 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 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 Rockey DC, Paulson E, Niedzwiecki D, et al. Analysis of air contrast barium enema, computed tomographic colonography, and colonoscopy: prospective comparison. Lancet 2005;365:305–11 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 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 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 Wald NJ, Hackshaw AK, Frost CD. When can a risk factor be used as a worthwhile screening test? BMJ 1999;319:1562–5 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 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 18 19 20 21 22

Gonzalez 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 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 Ahlquist DA. Fecal occult blood testing for colorectal cancer. Can we afford to do this? Gastroenterol Clin North Am 1997;26:41–55 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 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 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

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