- Email: [email protected]
An epidemiological perspective on cancer screening
Clinical Biochemistry, Vol. 28, No. 1, pp. 41-48, 1995 Copyright © 1995 The Canadian Society of Clinical Chemists Printed in the USA. All rights reserved 0009-9120/95 $9.50 + .00
Pergamon 0009-9120(94)00069-7
An Epidemiological Perspective on Cancer Screening ANTHONY B. MILLER Department of Preventive Medicine and Biostatistics, University of Toronto, Toronto, Ontario, M5S 1A8, Canada Objective: To provide an epidemiological perspective on cancer screening.
Methods and Results: For screening to be applicable as public health policy, the disease has to be an important health problem, there has to be evidence that early detection results in improved outcome, that adequate facilities for diagnosis, therapy, and subsequent management of true and false positives are available, that screening is acceptable to the target groups, and that programs are cost effective in the population. Although it is relatively easy to demonstrate that screening results in earlier detection of cancer, survival is a biassed measure of its effectiveness. The only valid design to study the efficacy of screening is the randomized trial. Cendcal cancer screening was introduced before these requirements were recognized. There is, however, good evidence of its effectiveness; the challenge is to make programs cost effective. For breast cancer, studies show little or no evidence of effectiveness of mammography screening in women age 40-49. For women age 50-69, there is good evidence of effectiveness in trials comparing screening with no screening. These support the introduction of population-based programs for this age group. The challenge is to put the research results into practice to ensure cost-effective programs. For colo-rectal cancer, there is some evidence that both screening sigmoidoscopy and the fecal occult blood test will reduce mortality. It is not clear, however, whether programs using either or both these tests will be cost effective. For lung cancer, there is good evidence of no benefit for screening. For ovarian, prostate, mouth, and skin cancer, although early detection has been demonstrated, there is no evidence of reduction in mortality in the target groups; indeed, prostate screening could result in lowering the overall quality of life. Conclusion: Screening, which offers a fairly rapid return from appropriate investment, should remain part of our armamentarium for cancer control.
K E Y W O R D S : cancer; screening; cervical;breast; colorectal;lung. Introduction f the approaches to cancer control that can reduce mortality from cancer m prevention, treatO ment, and screening -- ~screening holds perhaps the greatest promise for a rapid major impact, but for a number of practical and organizational difficulties, Presented at the combined meeting of CSCC, SQBC, CAMB, Quebec City, June 1994. Manuscript received August 9, 1994; accepted September 21, 1994. CLINICAL B I O C H E M I S T R Y , V O L U M E 28, F E B R U A R Y 1995
its potential may not be achieved. Further, there are several scientific reasons why the early detection of cancer does not automatically guarantee reduced cancer mortality (1). I shall first review the scientific basis for screening and then consider the evidence on screening for a number of major cancer sites. General principles of screening There are both benefits and disadvantages to screening for cancer (2). The benefits include an improved prognosis for some patients whose disease is detected by screening, but not for all. Those who benefit are primarily those who in the absence of screening would have died of their disease. This is the major benefit sought in screening programs. A second benefit of screening is that less radical treatment may be needed to cure some cases of their disease. A third benefit is reassurance for those with negative test results. Indeed, many people participate in screening programs for just this reassurance. A fourth benefit is resource savings, in particular a lower treatment cost if less radical treatment is instituted, and lower costs for treating patients who otherwise would have died of their disease, as these costs for cancer can often be substantial. The list of disadvantages is somewhat longer. The first is a longer period of morbidity, due to the advancement of the time of diagnosis or lead time from screening, for patients whose prognosis is unaltered. A second disadvantage is overtreatment of borderline abnormalities. Many abnormalities brought to light by screening might never have been recognized without a positive screening test. A third disadvantage is false reassurance for those with falsenegative screening tests. In such cases, the subsequent development of symptoms may be ignored, with postponement of the diagnosis and a poorer prognosis than if the symptoms had led to prompt diagnosis. A fourth disadvantage is unnecessary morbidity for those with false-positive screening tests, which may initiate a complex series of diagnostic tests. A fifth disadvantage is a potential hazard of the test itself. Finally there are resource costs, 41
MILLER
particularly those arising from the costs of administering the test, as well as those from the diagnostic tests induced by false-positives and those arising from the overtreatment of borderline abnormalities. Because of these disadvantages, it is important to insist upon appropriate evaluation of the effectiveness of screening. This process begins with evaluation of the test itself, which should be demonstrated to be of adequate sensitivity (the proportion of those with the disease who test positive) and specificity (the proportion of those free of the disease who test negative); and to be acceptable to those who are invited to be screened. The process then continues with evaluation of the screening program. The ideal approach for such an evaluation is the randomized controlled trial, in which people are randomized to be screened and their outcome in terms of mortality from the cancer in question assessed in comparison with the randomized control group (2). Although expensive, it is increasingly being realized that such trials are the only way to avoid the biases that otherwise accompany attempts to use other methods to evaluate the effectiveness of screening, especially those that attempt to use numbers of cases detected, stage distribution or survival as the principal end points. These biases are lead time, the time by which diagnosis is advanced by screening; length bias, the tendency of screening to detect cases of disease with a more prolonged natural history and, thus, a better prognosis than normal; selection bias, reflecting the recruitment of volunteers to screening who tend to have different risks of disease or death from disease than the general population; and overdiagnosis bias, the tendency of screening to bring to light and label as disease lesions that might never have been diagnosed in the screenee's lifetime (1). If a precursor of cancer is identified by the screening test, treatment of this lesion may interrupt the natural history of the cancer and result in reduction in incidence of disease in the screened population. However, if the cancer itself, albeit at an early stage, is detected as a result of screening, the incidence of the disease will be increased (because of lead time and possibly overdiagnosis)and mortality from the disease in the population screened will be the only valid outcome measure. Designs other than the randomized trial that use mortality in the population screened as the end point are able to avoid lead time, length bias, and overdiagnosis bias, but cannot avoid selection bias and, thus, may over or underestimate the effectiveness of screening. This has recently been demonstrated to especially affect the case-control approach, in which the screening history of cases (deaths or cases with advanced disease) are compared with those of controls drawn from the same population (3). Even though screening has been judged to be effective, once programs have been introduced in the population they should be monitored to ensure that they do achieve the expected benefits and that changes are not required in their organization. Screening should not be introduced unless there are 42
adequate facilities for diagnosis and treatment of the detected abnormalities and effective treatment for the detected disease. Even if these are present, failure of the referral process for diagnosis or treatment will result in failure of screening. Screening for cervical cancer
Because screening for cervical cancer was introduced before there was recognition of the need to evaluate programs by randomized trials, other mechanisms have been used to evaluate the effectiveness of screening for this disease. They include geographical comparisons of screened and unscreened populations combined with studies of the trends of incidence and mortality that followed introduction of screening (4), studies that correlated the intensity of screening with reduction in mortality from the disease (5), and case-control studies (6). Because a precursor is detected by screening (carcinoma in situ or severe dysplasia), reduction in incidence of the disease as well as reduction in mortality are both valid end points. However, very few countries have noted a dramatic reduction of incidence or mortality from the disease that could be directly related to the introduction of screening. The exceptions to this are largely the Nordic countries, where organized programs were introduced, and where an impressive reduction in incidence and mortality followed within 6 to 10 years (5). In North America and much of Europe, where organized screening was not introduced, and where screening was largely opportunistic (utilizing approaches that depend on the physician or the woman herself), although there have been some reductions, they have not been dramatic, and it has often been difficult to determine the extent that screening has been responsible for the reduction observed. Hence, there is increasing interest in introducing organized programs in such countries, as indicated by a workshop report in Canada (7). The essential features of organized programs have been described as follows: a) the individual women in the target population are identifiable; b) measures are available to guarantee high coverage and attendance, such as a personal letter of invitation; c) adequate field facilities for taking the smears and adequate laboratory facilities to examine them; d) an organized quality control program on taking of the smears and on interpreting them; e) adequate facilities exist for diagnosis and for appropriate treatment of confirmed neoplastic lesions; f) a carefully designed and agreed referral system . . . for management of any abnormalities found and for providing information about normal screening tests; g) evaluation and monitoring of the total program is organized... (6). One of the most controversial issues on cervical cancer screening relates to the frequency of rescreening. Several years ago the American Cancer Society (ACS) (8), largely echoing an earlier report by a Canadian Task Force (9), recommended threeCLINICAL BIOCHEMISTRY, VOLUME 28, FEBRUARY 1995
CANCER SCREENING
yearly screening as a norm. This proved intensely controversial, especially' to a number of clinical interests, and later the recommendation was changed back to one that could be interpreted as approving annual smears, though the physician has the discretion to perform tests less frequently (10,11). This change was then reflected in the Early Detection Working Guidelines of the National Cancer Institute (12). Unfortunately, these changes in the US Guidelines failed to reflect the natural history of the disease, a n u m b e r of studies having shown that the time to progression to invasive cancer of preclinical disease was of the order of several years if not decades, for example, a study carried out on records of screened women in British Columbia (13). British Columbia and other data had, in fact, been used in a mathematical model, the findings of which largely underlay the earlier ACS recommendations (14). These data, together with data from other studies largely in Europe, were combined by a working group of the International Agency for Research on Cancer (15). The analysis documented the expected degree of benefit from a negative screening test. Protection was maximal in the first 3 years after a negative test and persisted to some degree for at least 5 years. On the basis of this analysis, it was possible to compute the expected degree of benefit from different schedules of rescreening. It was clear that almost as much benefit is obtained from tests conducted at three-year intervals as annual screening, at a considerable saving in terms of numbers of tests required in a lifetime and, thus, in use of resources. The analysis assumes that the test is of high sensitivity and that a high proportion of at-risk women attend for screening. Adequate sensitivity of the test can only be achieved by ihigh quality laboratory services and high compliance by an organized program, both are a prerequisite for effective screening programs (7). However, defects in either of these parameters cannot be overcome by increasing the frequency of rescreening; rather, actions have to be in place to ensure an effective, indeed, a cost-effective program (6). The difficulty with annual screening, even in a wealthy country, is that it places emphasis on rescreening of women already in programs, while the emphasis needs to shift to bring women who are poorly screened, or not screened at all into programs, if failures of screening policies are to be avoided (9,16). Two other features of cervical cancer screening have proven controversial." the age range over which screening should be performed, and the management of the preclinical lesions found as a result of screening. Initiating screening at age 25 is almost as effective as initiating screening at age 20, b u t postponing initiation of screening to age 35 is less effective, though not markedly so. Most jurisdictions in North America, however, recommend initiation of screening at around ages 18 to 20 for sexually active women, because of a high frequency of preclinical lesions in young women (7). One advantage of such a policy is that it can capitalize on the attendance of CLINICAL BIOCHEMISTRY, VOLUME 28, FEBRUARY 1995
young women for contraceptive advice or antenatal care. However, many successful screening programs in Europe do not initiate screening until age 25 (3,6). Further, concentrating all screening on special health services for women is not wise, as in most countries those most at risk for the disease are the older women in their fifties and sixties who have not been screened or who have lapsed from screening. It is one of the main purposes of organized programs of screening to ensure that such women are screened. In terms of age to stop screening, three-yearly screening has been determined to be cost effective in women over the age of 65 in the US (17). Although cessation of screening used to be recommended at age 60 in Canada, recently the upper age was extended to 69, though with identification that this was a research issue (7). However, for women older than 65 who have not had a number of negative smears, screening should continue until they have achieved such a record (at l e a s t two n e g a t i v e smears) (3,7). In terms of management of dysplasia, concern has been expressed over substantial observer disagreement in classifying less severe forms of dysplasia as well as their importance in indicating increased risk of invasive cancer. Further, the role of H u m a n Papilloma Virus (HPV)-associated lesions in management is also controversial. Attempts to reduce the confusion over classification by setting up a twostage system [the Bethesda system (18)] was not judged in Canada to be particularly helpful (7). Reinforced by data confirming a low probability of progression for mild dysplasia, the Canadian group recommended cytologic surveillance for women with such lesions every 6 months for up to 2 years, unless there was evidence of progression cytologically. For cytologic abnormalities compatible with moderate or severe dysplasia, however, immediate referral for colposcopy for diagnosis was recommended (7). Screening for breast cancer
Much is known about screening for breast cancer, largely because this is a site for which a number of controlled trials and other studies have been completed, or are in progress. The first trial of screening was that of the Health Insurance Plan (HIP) of greater New York. Approximately 62,000 women age 4 0 - 6 4 were randomized in the early 1960s to screening by annual m a m m o g r a p h y and physical examination of the breasts to a total of four screens or to a control group. Of those invited to be screened, 65% were screened at least once. The early (5 year) mortality results indicated a significant reduction in breast cancer mortality but restricted to those age 5 0 - 6 4 (19). Subsequent follow-up (eventually completed at 18 years) suggested a benefit also in the younger women, commencing much later than that in older women (20), but the number of study subjects in the different age groups were small, and the findings are controversial as to whether a true benefit occurred in the younger women (21). The HIP 43
MILLER
study was conducted in the early years of mammography, and, therefore, only 33% of cases of breast cancer were detected by mammography alone, only 25% in women age 4 0 - 4 9 . Physical examination (conducted by carefully t r a i n e d surgeons), thus, made a substantial contribution to case detection and some have estimated that as much as 70% of the benefit from the screen may have been derived from the physical examinations (21). The HIP study was followed by the Breast Cancer Detection Demonstration Project (BCDDP). In this project, the combination of annual mammography, thermography, and physical examination of the breasts was used; 280,000 women being recruited age 35 and older. However, there was no control group, so that it was difficult to evaluate its effectiveness (22). It was clear, however, that thermography was of too low sensitivity and specificity to be a viable screening test, while mammography was of higher sensitivity than in the HIP study, especially in women under the age of 50. Extensive data on case detection (23) and survival (24) have been published for this study, data that cannot be used to infer effectiveness of the screen because of the biases reviewed earlier in this paper. More recently, the breast cancer mortality up to 9 years in the study population has been compared with that expected from the general population (25). This indicates little benefit in younger women, but in those age 50 or more who were recorded in the study files as asymptomatic on entry, a 40% reduction of breast cancer mortality below expectation was computed. Two case-control studies evaluating the possible role of breast self examination (BSE) in reducing advanced breast cancer have been reported (26,27). Both were negative, though in one (26), there appeared to be some benefit in compliers with BSE. Three population breast screening programs in Europe, two in the Netherlands (28,29), and one in Italy (30) have also been evaluated with the casecontrol approach. All three show an effect of screening in reducing deaths in women over the age of 50, but in those that studied younger women (29,30), no effect. These latter projects used m a m m o g r a p h y alone, the other (28), mammography plus physical examination. Four randomized trials are ongoing in Sweden. The largest, the two-country trial, used single medio-lateral oblique view mammography, every 21 months in women age 4 0 - 4 9 , and every 33 months in women age 5 0 - 7 4 . A significant reduction in breast cancer mortality in women age 5 0 - 6 9 was seen, starting at about 4 years after initiation of screening, which persists through to 10 years, but no effect in the first 10 years in women age 4 0 - 4 9 (31,32). The other long-running trial (10 years) was in Malm5 and used double-view mammography in women age 4 5 - 6 9 (33). The effect was less, and restricted to women over the age of 55. In younger women, there was a nonsignificant excess breast cancer mortality in the mammography group in the first 5 years after initiation of screening. The third 44
Swedish trial in Stockholm started later; this shows at 7 years an apparent effect in older women b u t not in those age 4 0 - 4 9 (34). Mortality results have not yet been independently reported on the fourth Swedish trial (in Gothenberg). However, they have been incorporated in an overview of the four Swedish trials (35). For women age 4 0 - 4 9 the overview suggested a beneficial effect beginning about 10 years after initiation of screening, but it was not statistically significant. A statistically significant benefit was seen for women age 5 0 - 6 9 . In the United Kingdom, a "quasi-experimental" s t u d y is ongoing i n v o l v i n g w o m e n age 4 5 - 6 4 screened by biennial m a m m o g r a p h y and annual physical examinations, with a randomized component in one center (Edinburgh). In the main study, breast cancer mortality was reduced at years 6 and 7 in the two screening centers compared to the four control areas, but not in the two areas where only BSE teaching was offered (36). In the Edinburgh randomized trial, no benefit was seen (37). Data from the nonrandomized component of the U K study is now available by age at 10 years. For women age 4 5 - 4 9 , the relative risk was 0.74 (95% CI 0 . 5 4 1.01). So this finding was of borderline significance, but there is uncertainty about it as it was based on a geographic comparison. For older women, the results were compatible with a benefit (38). Breast Self-Examination (BSE) has the potential to improve the outlook for interval cancers, while its teaching has probably gone some w a y to diminish false reassurance in those studies where it has been advocated together with other screening tests. The World Health Organization has concluded that only BSE has the potential to provide early diagnosis of breast cancer in many parts of the world (39). A cohort study has been conducted of the M a m a BSE program introduced by Gastrin in Finland in 1973. The records of nearly 30,000 women who enrolled in the program from 1973-75 and who returned BSE calendars to Dr. Gastrin were identified, and passive follow-up conducted by record linkage to the Finnish Cancer Registry. These BSE compliers had elevated breast cancer incidence but reduced breast cancer mortality compared to that expected from the Finnish population (40). The other two studies are randomized trials. The first, being conducted with group randomization by factory in Moscow and polyclinic in St Petrograd has enrolled nearly 200,000 women, half of whom have been instructed in BSE; follow-up is expected to last through to 1998 at least (41). The o t h e r trial, p l a n n e d to enroll a b o u t 300,000, is underway in Shanghai. The National Breast Screening Study (NBSS) in Canada is an individually randomized trial of volunteers recruited through 15 screening centers in six provinces, who all signed an informed consent form and provided data on risk factors for breast cancer. Participants were well balanced with respect to breast cancer risk factors. There was an equal distribution of women with breast symptoms and a similar proportion were referred to review for palCLINICAL BIOCHEMISTRY, VOLUME 28, FEBRUARY 1995
CANCER SCREENING
pable abnormalities. There was good compliance with the study procedures. Over 90% enrolled in each group attended screening and/or returned follow-up questionnaires from year 2 through 5. For women age 40-49, screening with yearly mammography and physical examination detected considerably more node-negative and small breast cancers than usual care, but had no impact on mortality from breast cancer in the first 7 years from entry (42). In women age 50-59, the objective of the NBSS was to determine the additional contribution of routine annual mammographic screening to screening by physical examination alone. Again, screening with yearly mammography in addition to physical examination detected considerably more nodenegative and small breast cancers than screening with physical examination alone, but had no impact on mortality from breast cancer in the first 7 years from entry (43). NBSS mammography has been criticized, but comparison with other studies shows that by all usual measures, the NBSS achieved as good cancer detection (44). Thus, the absence of benefit from mammography cannot to attributable to poor mammograms. Current breast screening programs have been planned on the expectation of a 30% or greater breast cancer mortality reduction in the age group screened. The NBSS suggests, however, that the contribution of mammography to breast cancer mortality reduction may be less than has been assumed. Screening for colorectali cancer There is now some evidence to support both sigmoidoscopy and fecal occult blood tests as possibly appropriate for colorectal cancer screening. A trial of multiphasic screening incorporating sigmoidoscopy was difficult to interpret, but the conclusion was that sigmoidoscopy was not responsible for lower colorectal mortality in the screened group compared to the control (45). The recent evidence in favor of sigmoidoscopy has come from two case-control studies (46,47), and it is known that such studies cannot eliminate the effect of selection bias, so that benefit has probably been overestimated. The long-running Minnesota trial has produced evidence that annual but not biennial fecal occult blood tests reduce mortality from colorectal cancer after about a 10-year period (48). This is a greater delay than had been expected, and it was achieved at a substantial cost in terms of false-positive results. Thus, the acceptability and cost effectiveness of these screens have not yet been established, and it seems that primary prevention should have higher priority than screening for this disease.
graphs with or without sputum cytology did not improve the prognosis of lung cancer in the populations studied (2). In the 1970s, three randomized trials of lung cancer screening were initiated in the United States. They showed that although chest radiographs and sputum cytology are capable of detecting cases in high-risk male heavy smokers, sputum cytology at 4-month intervals adds a small proportion of cases to those discovered by annual chest radiographs and does not demonstrably reduce lung cancer m o r t a l i t y . F u r t h e r , chest r a d i o g r a p h s at 4-month intervals confer no mortality advantage over routine care that includes annual chest radiographs. The UICC group (2) recommended assessing the value of the annual chest radiograph and suggested that case-control studies may have promise in this regard. One such study was performed in the German Democratic Republic, evaluating the role of mass miniature radiographs obtained at 2-year intervals, and found no evidence that such radiologic screening reduced the mortality from lung cancer (49). Both the UICC project (2) and the ACS (8) concluded that screening with sputum cytology and/or chest radiographs could not be recommended. All organizations agree that the emphasis for control of lung cancer has to be on primary prevention. BLADDER CANCER
Urinary cytology is clearly capable of detecting urinary bladder cancers in persons in high-risk occupations, as well as those living in areas where urinary schistosomiasis is endemic (2). Improved survival of patients with screen-detected bladder cancer has been demonstrated, but it is not known if a reduction in mortality will follow. Screeningdetected cases in areas where urinary schistosomiasis is endemic are at an earlier stage and require less extensive therapy than those diagnosed by routine measures. However, if bladder cancer screening is to play a role in cancer control, further research is needed. The role of urinary cytology in reducing mortality from bladder cancer in high-risk occupational groups and in areas of urinary schistosomiasis must be evaluated, preferably by means of controlled trials. ORAL CANCER
Several early studies suggested that screening for lung cancer by means of six monthly chest radio-
Visual examination is capable of identifying presymptomatic oral cancers. Exfoliative cytology is less sensitive than visual examination (2). Neither test has yet been shown to reduce mortality from oral cancer. In technically advanced countries, it will be difficult to evaluate the place of routine dental examinations, as those tend to be undertaken on people at low risk, high-risk individuals often avoiding dental care. In developing countries where the disease incidence is high, programs have been proposed based on inspection of the mouth by allied
CLINICAL BIOCHEMISTRY, VOLUME 28, FEBRUARY 1995
45
Screening for other cancers LUNG CANCER
MILLER
health professionals (50). These have not yet been fully evaluated. The role of primary prevention in those with oral epithelial dysplasia could be combined with an oral cancer screening program. Oral examination as part of the periodic health examination is recommended by the National Cancer Institute (12), especially for those at high risk due to tobacco and alcohol addiction, though no guidance is offered on how to ensure that those most at risk for the disease can be persuaded to accept such examinations. OVARIAN CANCER
Ovarian cancer is common in Western countries and has a relatively high mortality, now in excess of cervical cancer in these countries, largely because of the inaccessibility of the ovaries and the late stage of diagnosis of m a n y o v a r i a n cancers. Several screening tests are u n d e r development (6), with tests for CA 125 and vaginal ultrasound already being evaluated (3). One of the difficulties already encountered is the low specificity of the tests in premenopausal women. High specificity of the tests is essential because diagnostic confirmation is invasive. In premenopausal women, disease incidence is also low. This had led to the restriction of studies to postmenopausal women, where the disease incidence is higher and the specificity is higher. It is still not clear, however, whether the tests are sufficiently sensitive to expect a major benefit. Although using two or more tests can increase sensitivity if a positive to either is acted upon, this will lead to a loss of specificity, and, thus, increase in costs and morbidity. Several issues, therefore, need to be addressed before ovarian cancer screening can be considered as public health policy. ENDOMETRIAL CANCER
Although the disease is relatively common in W e s t e r n countries, t h e p e r s o n - y e a r s saved by screening m a y be limited because it is largely a disease of elderly women. Available screening tests are based on endometrial samples subjected to cytologic analysis. These are not simple tests, requiring considerable care in their application (6). Much further research into the effectiveness of endometrial cancer screening is required before it can be determined whether screening is likely to play a role in control of the disease. STOMACH CANCER
Screening programs for stomach cancer were introduced in J a p a n over 20 y e a r s ago (51). The screening test used has gradually been standardized and now comprises a photofluorographic b a r i u m meal technique with six standard views. The UICC project concluded that considerable observational evidence had a c c u m u l a t e d t h a t the widespread application of screening in J a p a n had 46
contributed to a fall in mortality, though its contribution is probably small in relation to that resulting from falling incidence (3,51). In view of the uncertainty over its effectiveness, screening for gastric cancer in countries other t h a n J a p a n cannot, at present, be recommended as public health policy. LIVER CANCER
Screening by means of serum ~-fetoprotein levels is being evaluated in some parts of China where the risk of liver cancer is very high (51). There is some evidence that small liver cancers can be successfully treated. However, pending further research, screening for liver cancer cannot be recommended as public health policy (51). ESOPHAGEAL CANCER
A similar conclusion was reached by the UICC group on esophageal cancer screening (51). Although a possible screening test is available that involves cytologic e x a m i n a t i o n of cells obtained with the aid of a balloon, there are difficulties associated with the cytologic diagnosis of esophageal dysplasia (a possible precancerous lesion); another concern is the unavailability of suitable t r e a t m e n t for any abnormalities found. Research into chemoprevention might provide a solution to the latter problem. PROSTATE CANCER
Prostate cancer is the leading cancer in incidence in men in the United States and second (after lung) in Canada. However, it is a condition t h a t has a steep rise of incidence with increasing age relatively late in life; therefore, it has less importance as a cause of Potential Years of Life Lost. Nevertheless, like m a n y cancers, it can be an important cause of morbidity, and there is increasing interest in the possibility of screening for prostate cancer using the digital rectal examination (DRE) and/or the prostate specific antigen (PSA) and possibly transrectal ultrasound, though the latter m a y be of more value as a diagnostic test (52). Screening for prostate cancer using the DRE is recommended by the ACS and in the early detection guidelines of the National Cancer Institute. It is not clear, however, t h a t this is a valid screening test, in that its sensitivity, at least for early prostate cancer, appears to be low. There are m a n y obstacles in the way of an effective screening program for prostate cancer. An acceptable and valid screening test has to be available and there must also be an acceptable and effective t r e a t m e n t for the preclinical lesions found as a result of screening (3). This problem is particularly acute for prostate cancer because of the increasing frequency of latent prostatic carcinoma with increasing age and the not inappreciable morbidity and mortality of the radical procedures usually used to treat prostate cancer. CLINICAL BIOCHEMISTRY, VOLUME 28, FEBRUARY 1995
CANCER SCREENING
The crucial distinction between screening and normal medical diagnosis and care is t h a t the encounter is not originated by the individual who is the subject of the screening; rather, the provider of screening initiates the process (53). The screener believes t h a t as a result of screening, the health of the community will be better. It is not possible to be certain t h a t the h e a l t h of everyone offered screening will be better, but at the very least the screener has an obligation to reduce to the m i n i m u m the possibility t h a t individuals will receive harm. Clearly, these conditions cannot yet be met for screening for prostate cancer, and, tlherefore, screening for this condition should only be offered in the context of a properly designed randomized trial with validly constituted informed consent forms. Such trials will have to be completed before a recommendation on including screening in a program of cancer control could be developed (3). We must be at least a decade away from such a recommendation.
achieved before the year 2000. As primary prevention may require m a n y decades to achieve its full potential, screening, which offers a fairly rapid ret u r n from appropriate investment, should remain part of our a r m a m e n t a r i u m for cancer control. References
There are a number of obstacles to a major contribution of screening to cancer control (54). These include an unfavorable n a t u r a l history of m a n y cancers and possibly particularly lung cancer; poor organization of screening programs; poor compliance of those at risk, especially noted for cancer of the cervix, though poor compliance m a y sometimes be exacerbated by economic barriers. Continued research is also needed to reduce the cost of and morbidity from t r e a t m e n t consequent on false-positive tests, as for screening for colorectal cancer, for example. In c o n s i d e r i n g t h e p o t e n t i a l c o n t r i b u t i o n of screening to reduction in cancer mortality by the year 2000, only screening for cancer of the breast and cervix were judged likely to make an impact, and even the relatively small impact envisaged (a 3% reduction in overall cancer mortality) would require s u b s t a n t i a l increases in the proportion of women being screened (55). For the other sites, it seems u n l i k e l y t h a t a n y major i m p a c t will be
1. Miller AB. General principles of evaluation of screening. In: Miller AB, ed. Screening for cancer, Pp. 3-24. Orlando, FL: Academic Press, 1985. 2. Prorok PC, Chamberlain J, Day NE, Hakama M, Miller AB. UICC workshop on the evaluation of screening programmes for cancer. Int J Cancer 1984; 34: 1-4. 3. Miller AB, Chamberlain J, Day NE, Hakama M, Prorok PC. Report on a workshop of the UICC project on evaluation of screening for cancer. Int J Cancer 1990; 46: 761-9. 4. Hakama M. Trends in the incidence of cervical cancer in the Nordic countries. In: Magnus K, ed. Trends in cancer incidence. Pp. 279-92. Washington, DC: Hemisphere Publ, 1982. 5. Miller AB, Lindsay J, Hill GB. Mortality from cancer of the uterus in Canada and its relationship to screening for cancer of the cervix. Int J Cancer 1976; 17: 602-12. 6. Hakama M, Chamberlain J, Day NE, Miller AB, Prorok PC. Evaluation of screening programmes for gynaecological cancer. Br J Cancer 1985; 52: 669-73. 7. Miller AB, Anderson G, Brisson J, et al. Report of a National Workshop on screening for cancer of the cervix. Can Med Assoc J 1991; 145: 1301-25. 8. American Cancer Society. Guidelines for the cancerrelated check-up. Recommendations and rationale. CA 1980; 30: 193-240. 9. Task Force. Cervical cancer screening programs. The Walton report. Can Med Assoc J 1976; 114: 1003-33. 10. Fink DJ. Change in American Cancer Society checkup guidelines for early detection of cervical cancer. CA 1988; 38: 127-8. 11. Mettlin C, Dodd GD. The American Cancer Society guidelines for the cancer-related checkup: An update. CA 1991; 41: 279-82. 12. Early Detection Branch. Working guidelines for early cancer detection. Bethesda: Division of Cancer Prevention and Control, National Cancer Institute, 1987. 13. Boyes DA, Morrison B, Knox EG, Draper G, Miller AB. A cohort study of cervical cancer in British Columbia. Clin Invest Med 1982; 5: 1-29. 14. Eddy D. Screening for cancer: Theory, analysis, and design. Englewood Cliffs, NJ: Prentice Hall, 1980. 15. IARC Working Group on cervical cancer screening. Summary chapter. In: Hakama M, Miller AB, Day NE, eds. Screening for cancer of the uterine cervix. Pp. 133-42. IARC Scientific Publications No. 76. Lyon, International Agency for Research on Cancer, 1986. 16. Chamberlain J. Reasons that some screening programmes fail to control cervical cancer. In: Hakama M, Miller AB, Day NE, eds. Screening for cancer of the uterine cervix. Pp. 161-8. IARC Scientific Publications No. 76. Lyon, International Agency for Research on Cancer, 1986. 17. Muller C, Mandelblatt J, Schechter CB, et al. Costs and effectiveness of cervical cancer screening in elderly women -- background paper. US Congress, Office of Technology Assessment. OTA-BP-H-65. Washington, DC: US Government Printing Office, 1990. 18. National Cancer Institute Workshop. The 1988 Be-
CLINICALBIOCHEMISTRY.VOLUME 28, FEBRUARY 1995
47
OTHER CANCERS The UICC project has also considered screening for melanoma, neuroblastoma, and nasopharyngeal carcinoma (3). For none of these conditions is screening recommended as public health policy. For melanoma, the issue is the lack of documented effectiveness of screening, though self-examination and referral to special s k i n clinics as p a r t of public education campaigns have been advocated and are being assessed in some areas (3). The National Cancer Institute Working Guidelines also recommend testicular self-examination and routine palpation of the testicles as part of the periodic health examination (12). However, no data relevant to the evaluation of screening for t h i s site u n c o n t a m i n a t e d by screening biases are offered in support of this recommendation.
Future perspectives
MILLER thesda system for reporting cervical/vaginal cytological diagnoses. J A M A 1989; 262: 931-4. 19. Shapiro S, Strax P, Venet L. Periodic breast cancer screening in reducing mortality from breast cancer. J A M A 1971; 215: 1777-85. 20. Shapiro S, Venet W, Strax P, Venet L. Periodic screen-
37. 38.
ing for breast cancer. The health insurance plan project and it sequelae, 1963-1986. Baltimore: The Johns
Hopkins University Press, 1988. 21. Miller AB. Breast cancer screening. Who should be included? J Gen Intern Med 1990; 5: S19-22. 22. Beahrs OH, Shapiro S, Smart C. Report of the working group to review the National Cancer Institute, American Cancer Society Breast Cancer Detection Demonstration Projects. J Natl Cancer Inst 1979; 62: 640-709. 23. Baker LH. Breast cancer detection demonstration project: Five-year summary report. CA 1982; 32: 194225. 24. Seidman H, Gelb SK, Silverberg E, et al. Survival experience in the Breast Cancer Detection Demonstration Project. CA 1987; 37: 258-90. 25. Morrison AS, Brisson J, Khalid N. Breast cancer incidence and mortality in the breast cancer detection demonstration project. J Natl Cancer Inst 1988; 80: 1540-7. 26. Newcomb PA, Weiss NS, Storer BE, Scholes D, Young BE, Voigt LF. Breast self-examination in relation to occurrence of advanced breast cancer. J Natl Cancer Inst 1991; 83: 260-5. 27. Muscat JE, Huncharek MS. Breast self-examination and extent of disease: A population-based study. Cancer Detect Prevent 1991; 15: 155-9. 28. CoUette HJA, Day NE, Rombach JJ, et al. Evaluation of screening for breast cancer in a non-randomized study (the DOM project) by means of a case-control study. Lancet 1984; i: 1224-6. 29. Verbeek ALM, Hendricks JHCL, Holland R, et al. Mammographic screening and breast cancer mortality age specific effects in Nijmegen project, 19751982 [Letter]. Lancet 1985; i: 865. 30. Palli D, del Turco MR, Buiatti E, et al. A case-control study of the efficacy of a non-randomised breast cancer screening program in Florence (Italy). Int J Cancer 1986; 38: 501-4. 31. Tabar L, Fagerberg CJG, Gad A, et al. Reduction in mortality from breast cancer after mass screening with mammography: Randomized trial from the breast cancer screening working group of the Swedish National Board of Health and Welfare. Lancet 1985; i: 829-32. 32. Tabar L, Fagerberg G, Duffy SW, Day NE. The Swedish two county trial of mammographic screening for breast cancer: Recent results and calculation of benefit. J Epidemiol Community Health 1989; 43: 107-14. 33. Andersson I, Aspergren K, Janzon L, et al. Mammographic screening and mortality from breast cancer: The Malm5 mammographic screening trial. Br Med J 1988; 297: 943-8. 34. Frisell J, Eklund G, HellstrSm L, et al. Randomized study of mammographic screening - - Preliminary report on mortality in the Stockholm trial. Breast Cancer Res Treat 1991; 18: 49-56. 35. Nystr{im L, Rutqvist LE, Wall S, et al. Breast cancer screening with mammography: Overview of Swedish randomized trails. Lancet 1993; 341: 973-78. 36. UK Trial of Early Detection of Breast Cancer Group. First results on mortality reduction in the UK Trial of 48
39.
40.
41.
42.
43.
44. 45.
46.
47. 48. 49. 50.
51.
52.
53. 54. 55.
Early Detection of Breast Cancer. Lancet 1988; ii: 411-16. Roberts MM, Alexander FE, Anderson TJ, et al. Edinburgh trial of screening for breast cancer: Mortality at seven years. Lancet 1990; 335: 241-6. UK Trial of Early Detection of Breast Cancer Group. Breast cancer mortality after 10 years in the UK trial of early detection of breast cancer. Breast 1993; 2: 13-20. Miller AB, Chamberlain J, Tsechovski M. Selfexamination in the early detection of breast cancer. A review of the evidence, with recommendations for further research. J Chron Dis 1985; 38: 527-40. Gastrin G, Miller AB, To T, et al. Incidence and mortality from breast cancer in the Mama program for breast screening in Finland, 1973-1986. Cancer 1994; 73: 2168-74. Semiglazov VF, Sagaidack VN, Moiseyenko VM, Mikhailov EA. Study of the role of breast selfexamination in the reduction of mortality from breast cancer. Eur J Cancer 1994; 29A: 2039-46. Miller AB, Baines CJ, To T, Wall C. The Canadian national breast screening study: Breast cancer detection and mortality in women age 40-49 on entry. Can Med Assoc J 1992; 147: 1459-76. Miller AB, Baines CJ, To T, Wall C. The Canadian national breast screening study: Breast cancer detection and mortality in women age 50-59 on entry. Can Med Assoc J 1992; 147: 1477-88. Fletcher SW, Black W, Harris R, Rimer BK, Shapiro S. Report of the International Workshop on screening for breast cancer. J Natl Cancer Inst 1993; 85: 1644-56. Selby JV, Friedman GD, Collen MF. Sigmoidoscopy and mortality from colorectal cancer: The Kaiser Permanente multiphasic evaluation study. J Clin Epidemiol 1988; 41: 427-34. Selby J, Friedman GCD, Quesenberry CP Jr, et al. A case-control study of screening sigmoidoscopy and mortality from colorectal cancer. N Engl J Med 1992; 326: 653-7. Newcomb PA, Norfleet RG, Storer BE, et al. Screening sigmoidoscopy and colorectal cancer mortality. J Natl Cancer Inst 1992; 84: 1572-5. Mandel JS, Bond JH, Church TR, et al. Reducing mortality from colorectal cancer by screening for fecal occult blood. N Engl J Med 1993; 328: 1365-71. Ebeling K, Nischan P. Screening for lung cancer: Results from a case-control study. Int J Cancer 1987; 40: 141-4. Warnakulasuriya KAAS, Ekanayake ANI, Sivayoham S, et al. Utilization of primary health care workers for early detection of oral cancer and precancer cases in Sri Lanka. Bull World Health Organ 1984; 62: 243-50. Chamberlain J, Day NE, Hakama M, Miller AB, Prorok PC. UICC workshop of the project on evaluation of screening programmes for gastrointestinal cancer. Int J Cancer 1986; 37: 329-34. Miller AB. Issues in screening for prostate cancer. In: Miller AB, Chamberlain J, Day NE, et al., eds. Cancer screening. Pp. 289-93. Cambridge: Cambridge University Press, 1991. Miller AB. The ethics, the risks and the benefits of screening. Biomed Pharmacother 1988; 42: 439-42. Miller AB. Screening for cancer: Issues and future directions. J Chron Dis 1986; 39: 1067-77. Greenwald P, Sondik EJ, eds. Cancer control objectives for the nation: 1985-2000. Natl Cancer Inst Monogr 2: 1986.
CLINICAL BIOCHEMISTRY, VOLUME 28, FEBRUARY 1995
Pergamon 0009-9120(94)00069-7
An Epidemiological Perspective on Cancer Screening ANTHONY B. MILLER Department of Preventive Medicine and Biostatistics, University of Toronto, Toronto, Ontario, M5S 1A8, Canada Objective: To provide an epidemiological perspective on cancer screening.
Methods and Results: For screening to be applicable as public health policy, the disease has to be an important health problem, there has to be evidence that early detection results in improved outcome, that adequate facilities for diagnosis, therapy, and subsequent management of true and false positives are available, that screening is acceptable to the target groups, and that programs are cost effective in the population. Although it is relatively easy to demonstrate that screening results in earlier detection of cancer, survival is a biassed measure of its effectiveness. The only valid design to study the efficacy of screening is the randomized trial. Cendcal cancer screening was introduced before these requirements were recognized. There is, however, good evidence of its effectiveness; the challenge is to make programs cost effective. For breast cancer, studies show little or no evidence of effectiveness of mammography screening in women age 40-49. For women age 50-69, there is good evidence of effectiveness in trials comparing screening with no screening. These support the introduction of population-based programs for this age group. The challenge is to put the research results into practice to ensure cost-effective programs. For colo-rectal cancer, there is some evidence that both screening sigmoidoscopy and the fecal occult blood test will reduce mortality. It is not clear, however, whether programs using either or both these tests will be cost effective. For lung cancer, there is good evidence of no benefit for screening. For ovarian, prostate, mouth, and skin cancer, although early detection has been demonstrated, there is no evidence of reduction in mortality in the target groups; indeed, prostate screening could result in lowering the overall quality of life. Conclusion: Screening, which offers a fairly rapid return from appropriate investment, should remain part of our armamentarium for cancer control.
K E Y W O R D S : cancer; screening; cervical;breast; colorectal;lung. Introduction f the approaches to cancer control that can reduce mortality from cancer m prevention, treatO ment, and screening -- ~screening holds perhaps the greatest promise for a rapid major impact, but for a number of practical and organizational difficulties, Presented at the combined meeting of CSCC, SQBC, CAMB, Quebec City, June 1994. Manuscript received August 9, 1994; accepted September 21, 1994. CLINICAL B I O C H E M I S T R Y , V O L U M E 28, F E B R U A R Y 1995
its potential may not be achieved. Further, there are several scientific reasons why the early detection of cancer does not automatically guarantee reduced cancer mortality (1). I shall first review the scientific basis for screening and then consider the evidence on screening for a number of major cancer sites. General principles of screening There are both benefits and disadvantages to screening for cancer (2). The benefits include an improved prognosis for some patients whose disease is detected by screening, but not for all. Those who benefit are primarily those who in the absence of screening would have died of their disease. This is the major benefit sought in screening programs. A second benefit of screening is that less radical treatment may be needed to cure some cases of their disease. A third benefit is reassurance for those with negative test results. Indeed, many people participate in screening programs for just this reassurance. A fourth benefit is resource savings, in particular a lower treatment cost if less radical treatment is instituted, and lower costs for treating patients who otherwise would have died of their disease, as these costs for cancer can often be substantial. The list of disadvantages is somewhat longer. The first is a longer period of morbidity, due to the advancement of the time of diagnosis or lead time from screening, for patients whose prognosis is unaltered. A second disadvantage is overtreatment of borderline abnormalities. Many abnormalities brought to light by screening might never have been recognized without a positive screening test. A third disadvantage is false reassurance for those with falsenegative screening tests. In such cases, the subsequent development of symptoms may be ignored, with postponement of the diagnosis and a poorer prognosis than if the symptoms had led to prompt diagnosis. A fourth disadvantage is unnecessary morbidity for those with false-positive screening tests, which may initiate a complex series of diagnostic tests. A fifth disadvantage is a potential hazard of the test itself. Finally there are resource costs, 41
MILLER
particularly those arising from the costs of administering the test, as well as those from the diagnostic tests induced by false-positives and those arising from the overtreatment of borderline abnormalities. Because of these disadvantages, it is important to insist upon appropriate evaluation of the effectiveness of screening. This process begins with evaluation of the test itself, which should be demonstrated to be of adequate sensitivity (the proportion of those with the disease who test positive) and specificity (the proportion of those free of the disease who test negative); and to be acceptable to those who are invited to be screened. The process then continues with evaluation of the screening program. The ideal approach for such an evaluation is the randomized controlled trial, in which people are randomized to be screened and their outcome in terms of mortality from the cancer in question assessed in comparison with the randomized control group (2). Although expensive, it is increasingly being realized that such trials are the only way to avoid the biases that otherwise accompany attempts to use other methods to evaluate the effectiveness of screening, especially those that attempt to use numbers of cases detected, stage distribution or survival as the principal end points. These biases are lead time, the time by which diagnosis is advanced by screening; length bias, the tendency of screening to detect cases of disease with a more prolonged natural history and, thus, a better prognosis than normal; selection bias, reflecting the recruitment of volunteers to screening who tend to have different risks of disease or death from disease than the general population; and overdiagnosis bias, the tendency of screening to bring to light and label as disease lesions that might never have been diagnosed in the screenee's lifetime (1). If a precursor of cancer is identified by the screening test, treatment of this lesion may interrupt the natural history of the cancer and result in reduction in incidence of disease in the screened population. However, if the cancer itself, albeit at an early stage, is detected as a result of screening, the incidence of the disease will be increased (because of lead time and possibly overdiagnosis)and mortality from the disease in the population screened will be the only valid outcome measure. Designs other than the randomized trial that use mortality in the population screened as the end point are able to avoid lead time, length bias, and overdiagnosis bias, but cannot avoid selection bias and, thus, may over or underestimate the effectiveness of screening. This has recently been demonstrated to especially affect the case-control approach, in which the screening history of cases (deaths or cases with advanced disease) are compared with those of controls drawn from the same population (3). Even though screening has been judged to be effective, once programs have been introduced in the population they should be monitored to ensure that they do achieve the expected benefits and that changes are not required in their organization. Screening should not be introduced unless there are 42
adequate facilities for diagnosis and treatment of the detected abnormalities and effective treatment for the detected disease. Even if these are present, failure of the referral process for diagnosis or treatment will result in failure of screening. Screening for cervical cancer
Because screening for cervical cancer was introduced before there was recognition of the need to evaluate programs by randomized trials, other mechanisms have been used to evaluate the effectiveness of screening for this disease. They include geographical comparisons of screened and unscreened populations combined with studies of the trends of incidence and mortality that followed introduction of screening (4), studies that correlated the intensity of screening with reduction in mortality from the disease (5), and case-control studies (6). Because a precursor is detected by screening (carcinoma in situ or severe dysplasia), reduction in incidence of the disease as well as reduction in mortality are both valid end points. However, very few countries have noted a dramatic reduction of incidence or mortality from the disease that could be directly related to the introduction of screening. The exceptions to this are largely the Nordic countries, where organized programs were introduced, and where an impressive reduction in incidence and mortality followed within 6 to 10 years (5). In North America and much of Europe, where organized screening was not introduced, and where screening was largely opportunistic (utilizing approaches that depend on the physician or the woman herself), although there have been some reductions, they have not been dramatic, and it has often been difficult to determine the extent that screening has been responsible for the reduction observed. Hence, there is increasing interest in introducing organized programs in such countries, as indicated by a workshop report in Canada (7). The essential features of organized programs have been described as follows: a) the individual women in the target population are identifiable; b) measures are available to guarantee high coverage and attendance, such as a personal letter of invitation; c) adequate field facilities for taking the smears and adequate laboratory facilities to examine them; d) an organized quality control program on taking of the smears and on interpreting them; e) adequate facilities exist for diagnosis and for appropriate treatment of confirmed neoplastic lesions; f) a carefully designed and agreed referral system . . . for management of any abnormalities found and for providing information about normal screening tests; g) evaluation and monitoring of the total program is organized... (6). One of the most controversial issues on cervical cancer screening relates to the frequency of rescreening. Several years ago the American Cancer Society (ACS) (8), largely echoing an earlier report by a Canadian Task Force (9), recommended threeCLINICAL BIOCHEMISTRY, VOLUME 28, FEBRUARY 1995
CANCER SCREENING
yearly screening as a norm. This proved intensely controversial, especially' to a number of clinical interests, and later the recommendation was changed back to one that could be interpreted as approving annual smears, though the physician has the discretion to perform tests less frequently (10,11). This change was then reflected in the Early Detection Working Guidelines of the National Cancer Institute (12). Unfortunately, these changes in the US Guidelines failed to reflect the natural history of the disease, a n u m b e r of studies having shown that the time to progression to invasive cancer of preclinical disease was of the order of several years if not decades, for example, a study carried out on records of screened women in British Columbia (13). British Columbia and other data had, in fact, been used in a mathematical model, the findings of which largely underlay the earlier ACS recommendations (14). These data, together with data from other studies largely in Europe, were combined by a working group of the International Agency for Research on Cancer (15). The analysis documented the expected degree of benefit from a negative screening test. Protection was maximal in the first 3 years after a negative test and persisted to some degree for at least 5 years. On the basis of this analysis, it was possible to compute the expected degree of benefit from different schedules of rescreening. It was clear that almost as much benefit is obtained from tests conducted at three-year intervals as annual screening, at a considerable saving in terms of numbers of tests required in a lifetime and, thus, in use of resources. The analysis assumes that the test is of high sensitivity and that a high proportion of at-risk women attend for screening. Adequate sensitivity of the test can only be achieved by ihigh quality laboratory services and high compliance by an organized program, both are a prerequisite for effective screening programs (7). However, defects in either of these parameters cannot be overcome by increasing the frequency of rescreening; rather, actions have to be in place to ensure an effective, indeed, a cost-effective program (6). The difficulty with annual screening, even in a wealthy country, is that it places emphasis on rescreening of women already in programs, while the emphasis needs to shift to bring women who are poorly screened, or not screened at all into programs, if failures of screening policies are to be avoided (9,16). Two other features of cervical cancer screening have proven controversial." the age range over which screening should be performed, and the management of the preclinical lesions found as a result of screening. Initiating screening at age 25 is almost as effective as initiating screening at age 20, b u t postponing initiation of screening to age 35 is less effective, though not markedly so. Most jurisdictions in North America, however, recommend initiation of screening at around ages 18 to 20 for sexually active women, because of a high frequency of preclinical lesions in young women (7). One advantage of such a policy is that it can capitalize on the attendance of CLINICAL BIOCHEMISTRY, VOLUME 28, FEBRUARY 1995
young women for contraceptive advice or antenatal care. However, many successful screening programs in Europe do not initiate screening until age 25 (3,6). Further, concentrating all screening on special health services for women is not wise, as in most countries those most at risk for the disease are the older women in their fifties and sixties who have not been screened or who have lapsed from screening. It is one of the main purposes of organized programs of screening to ensure that such women are screened. In terms of age to stop screening, three-yearly screening has been determined to be cost effective in women over the age of 65 in the US (17). Although cessation of screening used to be recommended at age 60 in Canada, recently the upper age was extended to 69, though with identification that this was a research issue (7). However, for women older than 65 who have not had a number of negative smears, screening should continue until they have achieved such a record (at l e a s t two n e g a t i v e smears) (3,7). In terms of management of dysplasia, concern has been expressed over substantial observer disagreement in classifying less severe forms of dysplasia as well as their importance in indicating increased risk of invasive cancer. Further, the role of H u m a n Papilloma Virus (HPV)-associated lesions in management is also controversial. Attempts to reduce the confusion over classification by setting up a twostage system [the Bethesda system (18)] was not judged in Canada to be particularly helpful (7). Reinforced by data confirming a low probability of progression for mild dysplasia, the Canadian group recommended cytologic surveillance for women with such lesions every 6 months for up to 2 years, unless there was evidence of progression cytologically. For cytologic abnormalities compatible with moderate or severe dysplasia, however, immediate referral for colposcopy for diagnosis was recommended (7). Screening for breast cancer
Much is known about screening for breast cancer, largely because this is a site for which a number of controlled trials and other studies have been completed, or are in progress. The first trial of screening was that of the Health Insurance Plan (HIP) of greater New York. Approximately 62,000 women age 4 0 - 6 4 were randomized in the early 1960s to screening by annual m a m m o g r a p h y and physical examination of the breasts to a total of four screens or to a control group. Of those invited to be screened, 65% were screened at least once. The early (5 year) mortality results indicated a significant reduction in breast cancer mortality but restricted to those age 5 0 - 6 4 (19). Subsequent follow-up (eventually completed at 18 years) suggested a benefit also in the younger women, commencing much later than that in older women (20), but the number of study subjects in the different age groups were small, and the findings are controversial as to whether a true benefit occurred in the younger women (21). The HIP 43
MILLER
study was conducted in the early years of mammography, and, therefore, only 33% of cases of breast cancer were detected by mammography alone, only 25% in women age 4 0 - 4 9 . Physical examination (conducted by carefully t r a i n e d surgeons), thus, made a substantial contribution to case detection and some have estimated that as much as 70% of the benefit from the screen may have been derived from the physical examinations (21). The HIP study was followed by the Breast Cancer Detection Demonstration Project (BCDDP). In this project, the combination of annual mammography, thermography, and physical examination of the breasts was used; 280,000 women being recruited age 35 and older. However, there was no control group, so that it was difficult to evaluate its effectiveness (22). It was clear, however, that thermography was of too low sensitivity and specificity to be a viable screening test, while mammography was of higher sensitivity than in the HIP study, especially in women under the age of 50. Extensive data on case detection (23) and survival (24) have been published for this study, data that cannot be used to infer effectiveness of the screen because of the biases reviewed earlier in this paper. More recently, the breast cancer mortality up to 9 years in the study population has been compared with that expected from the general population (25). This indicates little benefit in younger women, but in those age 50 or more who were recorded in the study files as asymptomatic on entry, a 40% reduction of breast cancer mortality below expectation was computed. Two case-control studies evaluating the possible role of breast self examination (BSE) in reducing advanced breast cancer have been reported (26,27). Both were negative, though in one (26), there appeared to be some benefit in compliers with BSE. Three population breast screening programs in Europe, two in the Netherlands (28,29), and one in Italy (30) have also been evaluated with the casecontrol approach. All three show an effect of screening in reducing deaths in women over the age of 50, but in those that studied younger women (29,30), no effect. These latter projects used m a m m o g r a p h y alone, the other (28), mammography plus physical examination. Four randomized trials are ongoing in Sweden. The largest, the two-country trial, used single medio-lateral oblique view mammography, every 21 months in women age 4 0 - 4 9 , and every 33 months in women age 5 0 - 7 4 . A significant reduction in breast cancer mortality in women age 5 0 - 6 9 was seen, starting at about 4 years after initiation of screening, which persists through to 10 years, but no effect in the first 10 years in women age 4 0 - 4 9 (31,32). The other long-running trial (10 years) was in Malm5 and used double-view mammography in women age 4 5 - 6 9 (33). The effect was less, and restricted to women over the age of 55. In younger women, there was a nonsignificant excess breast cancer mortality in the mammography group in the first 5 years after initiation of screening. The third 44
Swedish trial in Stockholm started later; this shows at 7 years an apparent effect in older women b u t not in those age 4 0 - 4 9 (34). Mortality results have not yet been independently reported on the fourth Swedish trial (in Gothenberg). However, they have been incorporated in an overview of the four Swedish trials (35). For women age 4 0 - 4 9 the overview suggested a beneficial effect beginning about 10 years after initiation of screening, but it was not statistically significant. A statistically significant benefit was seen for women age 5 0 - 6 9 . In the United Kingdom, a "quasi-experimental" s t u d y is ongoing i n v o l v i n g w o m e n age 4 5 - 6 4 screened by biennial m a m m o g r a p h y and annual physical examinations, with a randomized component in one center (Edinburgh). In the main study, breast cancer mortality was reduced at years 6 and 7 in the two screening centers compared to the four control areas, but not in the two areas where only BSE teaching was offered (36). In the Edinburgh randomized trial, no benefit was seen (37). Data from the nonrandomized component of the U K study is now available by age at 10 years. For women age 4 5 - 4 9 , the relative risk was 0.74 (95% CI 0 . 5 4 1.01). So this finding was of borderline significance, but there is uncertainty about it as it was based on a geographic comparison. For older women, the results were compatible with a benefit (38). Breast Self-Examination (BSE) has the potential to improve the outlook for interval cancers, while its teaching has probably gone some w a y to diminish false reassurance in those studies where it has been advocated together with other screening tests. The World Health Organization has concluded that only BSE has the potential to provide early diagnosis of breast cancer in many parts of the world (39). A cohort study has been conducted of the M a m a BSE program introduced by Gastrin in Finland in 1973. The records of nearly 30,000 women who enrolled in the program from 1973-75 and who returned BSE calendars to Dr. Gastrin were identified, and passive follow-up conducted by record linkage to the Finnish Cancer Registry. These BSE compliers had elevated breast cancer incidence but reduced breast cancer mortality compared to that expected from the Finnish population (40). The other two studies are randomized trials. The first, being conducted with group randomization by factory in Moscow and polyclinic in St Petrograd has enrolled nearly 200,000 women, half of whom have been instructed in BSE; follow-up is expected to last through to 1998 at least (41). The o t h e r trial, p l a n n e d to enroll a b o u t 300,000, is underway in Shanghai. The National Breast Screening Study (NBSS) in Canada is an individually randomized trial of volunteers recruited through 15 screening centers in six provinces, who all signed an informed consent form and provided data on risk factors for breast cancer. Participants were well balanced with respect to breast cancer risk factors. There was an equal distribution of women with breast symptoms and a similar proportion were referred to review for palCLINICAL BIOCHEMISTRY, VOLUME 28, FEBRUARY 1995
CANCER SCREENING
pable abnormalities. There was good compliance with the study procedures. Over 90% enrolled in each group attended screening and/or returned follow-up questionnaires from year 2 through 5. For women age 40-49, screening with yearly mammography and physical examination detected considerably more node-negative and small breast cancers than usual care, but had no impact on mortality from breast cancer in the first 7 years from entry (42). In women age 50-59, the objective of the NBSS was to determine the additional contribution of routine annual mammographic screening to screening by physical examination alone. Again, screening with yearly mammography in addition to physical examination detected considerably more nodenegative and small breast cancers than screening with physical examination alone, but had no impact on mortality from breast cancer in the first 7 years from entry (43). NBSS mammography has been criticized, but comparison with other studies shows that by all usual measures, the NBSS achieved as good cancer detection (44). Thus, the absence of benefit from mammography cannot to attributable to poor mammograms. Current breast screening programs have been planned on the expectation of a 30% or greater breast cancer mortality reduction in the age group screened. The NBSS suggests, however, that the contribution of mammography to breast cancer mortality reduction may be less than has been assumed. Screening for colorectali cancer There is now some evidence to support both sigmoidoscopy and fecal occult blood tests as possibly appropriate for colorectal cancer screening. A trial of multiphasic screening incorporating sigmoidoscopy was difficult to interpret, but the conclusion was that sigmoidoscopy was not responsible for lower colorectal mortality in the screened group compared to the control (45). The recent evidence in favor of sigmoidoscopy has come from two case-control studies (46,47), and it is known that such studies cannot eliminate the effect of selection bias, so that benefit has probably been overestimated. The long-running Minnesota trial has produced evidence that annual but not biennial fecal occult blood tests reduce mortality from colorectal cancer after about a 10-year period (48). This is a greater delay than had been expected, and it was achieved at a substantial cost in terms of false-positive results. Thus, the acceptability and cost effectiveness of these screens have not yet been established, and it seems that primary prevention should have higher priority than screening for this disease.
graphs with or without sputum cytology did not improve the prognosis of lung cancer in the populations studied (2). In the 1970s, three randomized trials of lung cancer screening were initiated in the United States. They showed that although chest radiographs and sputum cytology are capable of detecting cases in high-risk male heavy smokers, sputum cytology at 4-month intervals adds a small proportion of cases to those discovered by annual chest radiographs and does not demonstrably reduce lung cancer m o r t a l i t y . F u r t h e r , chest r a d i o g r a p h s at 4-month intervals confer no mortality advantage over routine care that includes annual chest radiographs. The UICC group (2) recommended assessing the value of the annual chest radiograph and suggested that case-control studies may have promise in this regard. One such study was performed in the German Democratic Republic, evaluating the role of mass miniature radiographs obtained at 2-year intervals, and found no evidence that such radiologic screening reduced the mortality from lung cancer (49). Both the UICC project (2) and the ACS (8) concluded that screening with sputum cytology and/or chest radiographs could not be recommended. All organizations agree that the emphasis for control of lung cancer has to be on primary prevention. BLADDER CANCER
Urinary cytology is clearly capable of detecting urinary bladder cancers in persons in high-risk occupations, as well as those living in areas where urinary schistosomiasis is endemic (2). Improved survival of patients with screen-detected bladder cancer has been demonstrated, but it is not known if a reduction in mortality will follow. Screeningdetected cases in areas where urinary schistosomiasis is endemic are at an earlier stage and require less extensive therapy than those diagnosed by routine measures. However, if bladder cancer screening is to play a role in cancer control, further research is needed. The role of urinary cytology in reducing mortality from bladder cancer in high-risk occupational groups and in areas of urinary schistosomiasis must be evaluated, preferably by means of controlled trials. ORAL CANCER
Several early studies suggested that screening for lung cancer by means of six monthly chest radio-
Visual examination is capable of identifying presymptomatic oral cancers. Exfoliative cytology is less sensitive than visual examination (2). Neither test has yet been shown to reduce mortality from oral cancer. In technically advanced countries, it will be difficult to evaluate the place of routine dental examinations, as those tend to be undertaken on people at low risk, high-risk individuals often avoiding dental care. In developing countries where the disease incidence is high, programs have been proposed based on inspection of the mouth by allied
CLINICAL BIOCHEMISTRY, VOLUME 28, FEBRUARY 1995
45
Screening for other cancers LUNG CANCER
MILLER
health professionals (50). These have not yet been fully evaluated. The role of primary prevention in those with oral epithelial dysplasia could be combined with an oral cancer screening program. Oral examination as part of the periodic health examination is recommended by the National Cancer Institute (12), especially for those at high risk due to tobacco and alcohol addiction, though no guidance is offered on how to ensure that those most at risk for the disease can be persuaded to accept such examinations. OVARIAN CANCER
Ovarian cancer is common in Western countries and has a relatively high mortality, now in excess of cervical cancer in these countries, largely because of the inaccessibility of the ovaries and the late stage of diagnosis of m a n y o v a r i a n cancers. Several screening tests are u n d e r development (6), with tests for CA 125 and vaginal ultrasound already being evaluated (3). One of the difficulties already encountered is the low specificity of the tests in premenopausal women. High specificity of the tests is essential because diagnostic confirmation is invasive. In premenopausal women, disease incidence is also low. This had led to the restriction of studies to postmenopausal women, where the disease incidence is higher and the specificity is higher. It is still not clear, however, whether the tests are sufficiently sensitive to expect a major benefit. Although using two or more tests can increase sensitivity if a positive to either is acted upon, this will lead to a loss of specificity, and, thus, increase in costs and morbidity. Several issues, therefore, need to be addressed before ovarian cancer screening can be considered as public health policy. ENDOMETRIAL CANCER
Although the disease is relatively common in W e s t e r n countries, t h e p e r s o n - y e a r s saved by screening m a y be limited because it is largely a disease of elderly women. Available screening tests are based on endometrial samples subjected to cytologic analysis. These are not simple tests, requiring considerable care in their application (6). Much further research into the effectiveness of endometrial cancer screening is required before it can be determined whether screening is likely to play a role in control of the disease. STOMACH CANCER
Screening programs for stomach cancer were introduced in J a p a n over 20 y e a r s ago (51). The screening test used has gradually been standardized and now comprises a photofluorographic b a r i u m meal technique with six standard views. The UICC project concluded that considerable observational evidence had a c c u m u l a t e d t h a t the widespread application of screening in J a p a n had 46
contributed to a fall in mortality, though its contribution is probably small in relation to that resulting from falling incidence (3,51). In view of the uncertainty over its effectiveness, screening for gastric cancer in countries other t h a n J a p a n cannot, at present, be recommended as public health policy. LIVER CANCER
Screening by means of serum ~-fetoprotein levels is being evaluated in some parts of China where the risk of liver cancer is very high (51). There is some evidence that small liver cancers can be successfully treated. However, pending further research, screening for liver cancer cannot be recommended as public health policy (51). ESOPHAGEAL CANCER
A similar conclusion was reached by the UICC group on esophageal cancer screening (51). Although a possible screening test is available that involves cytologic e x a m i n a t i o n of cells obtained with the aid of a balloon, there are difficulties associated with the cytologic diagnosis of esophageal dysplasia (a possible precancerous lesion); another concern is the unavailability of suitable t r e a t m e n t for any abnormalities found. Research into chemoprevention might provide a solution to the latter problem. PROSTATE CANCER
Prostate cancer is the leading cancer in incidence in men in the United States and second (after lung) in Canada. However, it is a condition t h a t has a steep rise of incidence with increasing age relatively late in life; therefore, it has less importance as a cause of Potential Years of Life Lost. Nevertheless, like m a n y cancers, it can be an important cause of morbidity, and there is increasing interest in the possibility of screening for prostate cancer using the digital rectal examination (DRE) and/or the prostate specific antigen (PSA) and possibly transrectal ultrasound, though the latter m a y be of more value as a diagnostic test (52). Screening for prostate cancer using the DRE is recommended by the ACS and in the early detection guidelines of the National Cancer Institute. It is not clear, however, t h a t this is a valid screening test, in that its sensitivity, at least for early prostate cancer, appears to be low. There are m a n y obstacles in the way of an effective screening program for prostate cancer. An acceptable and valid screening test has to be available and there must also be an acceptable and effective t r e a t m e n t for the preclinical lesions found as a result of screening (3). This problem is particularly acute for prostate cancer because of the increasing frequency of latent prostatic carcinoma with increasing age and the not inappreciable morbidity and mortality of the radical procedures usually used to treat prostate cancer. CLINICAL BIOCHEMISTRY, VOLUME 28, FEBRUARY 1995
CANCER SCREENING
The crucial distinction between screening and normal medical diagnosis and care is t h a t the encounter is not originated by the individual who is the subject of the screening; rather, the provider of screening initiates the process (53). The screener believes t h a t as a result of screening, the health of the community will be better. It is not possible to be certain t h a t the h e a l t h of everyone offered screening will be better, but at the very least the screener has an obligation to reduce to the m i n i m u m the possibility t h a t individuals will receive harm. Clearly, these conditions cannot yet be met for screening for prostate cancer, and, tlherefore, screening for this condition should only be offered in the context of a properly designed randomized trial with validly constituted informed consent forms. Such trials will have to be completed before a recommendation on including screening in a program of cancer control could be developed (3). We must be at least a decade away from such a recommendation.
achieved before the year 2000. As primary prevention may require m a n y decades to achieve its full potential, screening, which offers a fairly rapid ret u r n from appropriate investment, should remain part of our a r m a m e n t a r i u m for cancer control. References
There are a number of obstacles to a major contribution of screening to cancer control (54). These include an unfavorable n a t u r a l history of m a n y cancers and possibly particularly lung cancer; poor organization of screening programs; poor compliance of those at risk, especially noted for cancer of the cervix, though poor compliance m a y sometimes be exacerbated by economic barriers. Continued research is also needed to reduce the cost of and morbidity from t r e a t m e n t consequent on false-positive tests, as for screening for colorectal cancer, for example. In c o n s i d e r i n g t h e p o t e n t i a l c o n t r i b u t i o n of screening to reduction in cancer mortality by the year 2000, only screening for cancer of the breast and cervix were judged likely to make an impact, and even the relatively small impact envisaged (a 3% reduction in overall cancer mortality) would require s u b s t a n t i a l increases in the proportion of women being screened (55). For the other sites, it seems u n l i k e l y t h a t a n y major i m p a c t will be
1. Miller AB. General principles of evaluation of screening. In: Miller AB, ed. Screening for cancer, Pp. 3-24. Orlando, FL: Academic Press, 1985. 2. Prorok PC, Chamberlain J, Day NE, Hakama M, Miller AB. UICC workshop on the evaluation of screening programmes for cancer. Int J Cancer 1984; 34: 1-4. 3. Miller AB, Chamberlain J, Day NE, Hakama M, Prorok PC. Report on a workshop of the UICC project on evaluation of screening for cancer. Int J Cancer 1990; 46: 761-9. 4. Hakama M. Trends in the incidence of cervical cancer in the Nordic countries. In: Magnus K, ed. Trends in cancer incidence. Pp. 279-92. Washington, DC: Hemisphere Publ, 1982. 5. Miller AB, Lindsay J, Hill GB. Mortality from cancer of the uterus in Canada and its relationship to screening for cancer of the cervix. Int J Cancer 1976; 17: 602-12. 6. Hakama M, Chamberlain J, Day NE, Miller AB, Prorok PC. Evaluation of screening programmes for gynaecological cancer. Br J Cancer 1985; 52: 669-73. 7. Miller AB, Anderson G, Brisson J, et al. Report of a National Workshop on screening for cancer of the cervix. Can Med Assoc J 1991; 145: 1301-25. 8. American Cancer Society. Guidelines for the cancerrelated check-up. Recommendations and rationale. CA 1980; 30: 193-240. 9. Task Force. Cervical cancer screening programs. The Walton report. Can Med Assoc J 1976; 114: 1003-33. 10. Fink DJ. Change in American Cancer Society checkup guidelines for early detection of cervical cancer. CA 1988; 38: 127-8. 11. Mettlin C, Dodd GD. The American Cancer Society guidelines for the cancer-related checkup: An update. CA 1991; 41: 279-82. 12. Early Detection Branch. Working guidelines for early cancer detection. Bethesda: Division of Cancer Prevention and Control, National Cancer Institute, 1987. 13. Boyes DA, Morrison B, Knox EG, Draper G, Miller AB. A cohort study of cervical cancer in British Columbia. Clin Invest Med 1982; 5: 1-29. 14. Eddy D. Screening for cancer: Theory, analysis, and design. Englewood Cliffs, NJ: Prentice Hall, 1980. 15. IARC Working Group on cervical cancer screening. Summary chapter. In: Hakama M, Miller AB, Day NE, eds. Screening for cancer of the uterine cervix. Pp. 133-42. IARC Scientific Publications No. 76. Lyon, International Agency for Research on Cancer, 1986. 16. Chamberlain J. Reasons that some screening programmes fail to control cervical cancer. In: Hakama M, Miller AB, Day NE, eds. Screening for cancer of the uterine cervix. Pp. 161-8. IARC Scientific Publications No. 76. Lyon, International Agency for Research on Cancer, 1986. 17. Muller C, Mandelblatt J, Schechter CB, et al. Costs and effectiveness of cervical cancer screening in elderly women -- background paper. US Congress, Office of Technology Assessment. OTA-BP-H-65. Washington, DC: US Government Printing Office, 1990. 18. National Cancer Institute Workshop. The 1988 Be-
CLINICALBIOCHEMISTRY.VOLUME 28, FEBRUARY 1995
47
OTHER CANCERS The UICC project has also considered screening for melanoma, neuroblastoma, and nasopharyngeal carcinoma (3). For none of these conditions is screening recommended as public health policy. For melanoma, the issue is the lack of documented effectiveness of screening, though self-examination and referral to special s k i n clinics as p a r t of public education campaigns have been advocated and are being assessed in some areas (3). The National Cancer Institute Working Guidelines also recommend testicular self-examination and routine palpation of the testicles as part of the periodic health examination (12). However, no data relevant to the evaluation of screening for t h i s site u n c o n t a m i n a t e d by screening biases are offered in support of this recommendation.
Future perspectives
MILLER thesda system for reporting cervical/vaginal cytological diagnoses. J A M A 1989; 262: 931-4. 19. Shapiro S, Strax P, Venet L. Periodic breast cancer screening in reducing mortality from breast cancer. J A M A 1971; 215: 1777-85. 20. Shapiro S, Venet W, Strax P, Venet L. Periodic screen-
37. 38.
ing for breast cancer. The health insurance plan project and it sequelae, 1963-1986. Baltimore: The Johns
Hopkins University Press, 1988. 21. Miller AB. Breast cancer screening. Who should be included? J Gen Intern Med 1990; 5: S19-22. 22. Beahrs OH, Shapiro S, Smart C. Report of the working group to review the National Cancer Institute, American Cancer Society Breast Cancer Detection Demonstration Projects. J Natl Cancer Inst 1979; 62: 640-709. 23. Baker LH. Breast cancer detection demonstration project: Five-year summary report. CA 1982; 32: 194225. 24. Seidman H, Gelb SK, Silverberg E, et al. Survival experience in the Breast Cancer Detection Demonstration Project. CA 1987; 37: 258-90. 25. Morrison AS, Brisson J, Khalid N. Breast cancer incidence and mortality in the breast cancer detection demonstration project. J Natl Cancer Inst 1988; 80: 1540-7. 26. Newcomb PA, Weiss NS, Storer BE, Scholes D, Young BE, Voigt LF. Breast self-examination in relation to occurrence of advanced breast cancer. J Natl Cancer Inst 1991; 83: 260-5. 27. Muscat JE, Huncharek MS. Breast self-examination and extent of disease: A population-based study. Cancer Detect Prevent 1991; 15: 155-9. 28. CoUette HJA, Day NE, Rombach JJ, et al. Evaluation of screening for breast cancer in a non-randomized study (the DOM project) by means of a case-control study. Lancet 1984; i: 1224-6. 29. Verbeek ALM, Hendricks JHCL, Holland R, et al. Mammographic screening and breast cancer mortality age specific effects in Nijmegen project, 19751982 [Letter]. Lancet 1985; i: 865. 30. Palli D, del Turco MR, Buiatti E, et al. A case-control study of the efficacy of a non-randomised breast cancer screening program in Florence (Italy). Int J Cancer 1986; 38: 501-4. 31. Tabar L, Fagerberg CJG, Gad A, et al. Reduction in mortality from breast cancer after mass screening with mammography: Randomized trial from the breast cancer screening working group of the Swedish National Board of Health and Welfare. Lancet 1985; i: 829-32. 32. Tabar L, Fagerberg G, Duffy SW, Day NE. The Swedish two county trial of mammographic screening for breast cancer: Recent results and calculation of benefit. J Epidemiol Community Health 1989; 43: 107-14. 33. Andersson I, Aspergren K, Janzon L, et al. Mammographic screening and mortality from breast cancer: The Malm5 mammographic screening trial. Br Med J 1988; 297: 943-8. 34. Frisell J, Eklund G, HellstrSm L, et al. Randomized study of mammographic screening - - Preliminary report on mortality in the Stockholm trial. Breast Cancer Res Treat 1991; 18: 49-56. 35. Nystr{im L, Rutqvist LE, Wall S, et al. Breast cancer screening with mammography: Overview of Swedish randomized trails. Lancet 1993; 341: 973-78. 36. UK Trial of Early Detection of Breast Cancer Group. First results on mortality reduction in the UK Trial of 48
39.
40.
41.
42.
43.
44. 45.
46.
47. 48. 49. 50.
51.
52.
53. 54. 55.
Early Detection of Breast Cancer. Lancet 1988; ii: 411-16. Roberts MM, Alexander FE, Anderson TJ, et al. Edinburgh trial of screening for breast cancer: Mortality at seven years. Lancet 1990; 335: 241-6. UK Trial of Early Detection of Breast Cancer Group. Breast cancer mortality after 10 years in the UK trial of early detection of breast cancer. Breast 1993; 2: 13-20. Miller AB, Chamberlain J, Tsechovski M. Selfexamination in the early detection of breast cancer. A review of the evidence, with recommendations for further research. J Chron Dis 1985; 38: 527-40. Gastrin G, Miller AB, To T, et al. Incidence and mortality from breast cancer in the Mama program for breast screening in Finland, 1973-1986. Cancer 1994; 73: 2168-74. Semiglazov VF, Sagaidack VN, Moiseyenko VM, Mikhailov EA. Study of the role of breast selfexamination in the reduction of mortality from breast cancer. Eur J Cancer 1994; 29A: 2039-46. Miller AB, Baines CJ, To T, Wall C. The Canadian national breast screening study: Breast cancer detection and mortality in women age 40-49 on entry. Can Med Assoc J 1992; 147: 1459-76. Miller AB, Baines CJ, To T, Wall C. The Canadian national breast screening study: Breast cancer detection and mortality in women age 50-59 on entry. Can Med Assoc J 1992; 147: 1477-88. Fletcher SW, Black W, Harris R, Rimer BK, Shapiro S. Report of the International Workshop on screening for breast cancer. J Natl Cancer Inst 1993; 85: 1644-56. Selby JV, Friedman GD, Collen MF. Sigmoidoscopy and mortality from colorectal cancer: The Kaiser Permanente multiphasic evaluation study. J Clin Epidemiol 1988; 41: 427-34. Selby J, Friedman GCD, Quesenberry CP Jr, et al. A case-control study of screening sigmoidoscopy and mortality from colorectal cancer. N Engl J Med 1992; 326: 653-7. Newcomb PA, Norfleet RG, Storer BE, et al. Screening sigmoidoscopy and colorectal cancer mortality. J Natl Cancer Inst 1992; 84: 1572-5. Mandel JS, Bond JH, Church TR, et al. Reducing mortality from colorectal cancer by screening for fecal occult blood. N Engl J Med 1993; 328: 1365-71. Ebeling K, Nischan P. Screening for lung cancer: Results from a case-control study. Int J Cancer 1987; 40: 141-4. Warnakulasuriya KAAS, Ekanayake ANI, Sivayoham S, et al. Utilization of primary health care workers for early detection of oral cancer and precancer cases in Sri Lanka. Bull World Health Organ 1984; 62: 243-50. Chamberlain J, Day NE, Hakama M, Miller AB, Prorok PC. UICC workshop of the project on evaluation of screening programmes for gastrointestinal cancer. Int J Cancer 1986; 37: 329-34. Miller AB. Issues in screening for prostate cancer. In: Miller AB, Chamberlain J, Day NE, et al., eds. Cancer screening. Pp. 289-93. Cambridge: Cambridge University Press, 1991. Miller AB. The ethics, the risks and the benefits of screening. Biomed Pharmacother 1988; 42: 439-42. Miller AB. Screening for cancer: Issues and future directions. J Chron Dis 1986; 39: 1067-77. Greenwald P, Sondik EJ, eds. Cancer control objectives for the nation: 1985-2000. Natl Cancer Inst Monogr 2: 1986.
CLINICAL BIOCHEMISTRY, VOLUME 28, FEBRUARY 1995