| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Perspectives |
1 From the Department of Radiological Sciences, University of California, Irvine Medical Center, Orange. Received October 15, 2001; accepted October 17. Address correspondence to the author, 18961 Castlegate Ln, Santa Ana, CA 92705 (e-mail: rmfriede@uci.edu).
Index terms: Cancer screening • Cost-effectiveness • Perspectives • Radiology and radiologists, socioeconomic issues
This is the age of screening examinations—cervical, breast, lung, and now cardiac. This is also the age of managed health care, which is based on the concept of cost-effectiveness; currently, the only accepted screening procedures are for mammography and cervical carcinoma. In 1968, Wilson and Junger established the classical criteria for screening: (a) The burden of disease must be sufficient, (b) the disease must be detectable in the preclinical phase, (c) there must be an effective test to help detect the disease, and (d) there must be an effective treatment (1). An ideal screening test should be noninvasive, inexpensive, widely available, and cost-effective. It should have high specificity and show improved outcomes over the control group. It has been estimated that the cost of the annual Papanicolaou test for women, with testing starting at age 20 years, is $99,000 per life saved, and the cost of annual mammography for women aged 55–64 years is $132,000 per life saved (2). Both of these numbers were calculated by statisticians and are presented in 1998 dollars. The appropriate amount to be spent on screening examinations will vary with the importance of the disease and the effectiveness of the screening examination. In the aforementioned illustrations, the cost of screening was acceptable up to $132,000 per life saved.
When these numbers are presented, they are impersonal, and individuals do not relate to them as affecting their own family situation. Since we do not expect to be caught in the web of malignancy, we accept whatever the experts tell us is reasonable. However, if malignancy becomes personal and occurs in their family, most individuals resent cost-effectiveness concepts; they believe that the life of their family member is priceless, and they resent controls. That is the beauty of the use of population-based cost-effectiveness: It is acceptable because it is impersonal.
Screening for carcinoma of the cervix and breast are proven techniques. Variations may occur in the future to reduce costs but maintain a reasonable cost-effectiveness ratio. It may be determined, for example, that if Papanicolau smears are normal for 5 years in patients without a family history, then repeat studies may be necessary only every 4 or 5 years after that.
I will briefly discuss other possible screening examinations that are on our horizon and comment on their benefits and the problems associated with screening.
Colorectal carcinoma is the second leading cause of cancer mortality in the United States, with approximately 56,000 deaths in 1999 (2). Multiple types of screening examinations have been proposed for this malignancy, including double-contrast barium enema examination, fecal occult blood testing (FOBT), colonoscopy, sigmoidoscopy, and combinations of the above. Frazier and associates (2) evaluated 22 screening combinations in the attempt to find the one that is most cost-effective. The cost-effectiveness ratio for the different combinations varied from $1,200 per life saved (one sigmoidoscopy at age 55 years) to $92,900 per life saved (FOBT annually and sigmoidoscopy every 5 years). They thought that the costs of these procedures were within acceptable limits. They also noted that compliance with screening is low in the United States, which keeps costs down. Only 20% of respondents reported having undergone FOBT in the past year, and 30% reported having undergone sigmoidoscopy in the past 5 years. Frazier and associates believe that the most effective starting point for a national policy would be a one-time screening colonoscopy at age 55 years, which they calculated would reduce mortality by 31% and would have a cost-effectiveness ratio of $22,400 (in 1998 dollars) per life saved.
The suggestion that FOBT results be used to select candidates for colonoscopy does not appear to be practical. In a series with 3,196 patients, Lieberman and Weiss (3) found that FOBT was positive in only 7% of subjects with small tubular adenomas and in 35.6% of subjects with cancer or adenoma with high-grade dysplasia. They also noted that if sigmoidoscopy was relied on as a screening procedure, 31.9% of cases of advanced neoplasia that occurred in the proximal colon would be missed.
McMahon and associates (4) analyzed the cost-effectiveness of double-contrast barium enema examination versus colonoscopy in average-risk populations. They emphasized that the fundamental goal of a cost-effectiveness analysis, which is to compare the relative values of alternative interventions, is often complicated by variations in the reporting of results. In their analysis, a double-contrast barium enema examination every 5 years coupled with annual FOBT was safer and more cost-effective than colonoscopy, with an incremental cost of less than $55,600 per life saved.
Colorectal cancer screening is today considered a reasonable strategy, although, to my knowledge, the most cost-effective approach has not been determined. Obviously, a colonoscopy every 5 years with a repeat colonoscopy in 2 years if adenomatous polyps are present would be efficient but probably not cost-effective as a national policy. Sigmoidoscopy and FOBT, although cost efficient, are not cost-effective by themselves. Computed tomographic (CT) colonography is now being assessed. The current established choices in procedure are either colonoscopy or double-contrast barium enema examination. Perhaps an initial colonoscopy at age 50–55 years and, if negative, a single follow-up colonoscopy at age 65 would be a reasonable approach. Glick et al (5) calculated that colonoscopy every 10 years results in an increased life expectancy of 21.64 days per person, while colonoscopy every 5 years only increases this to 23.73 days per person.
Screening for cancer of the lung is not yet an approved procedure, since studies have not determined if use of the screening tests will result in lower mortality rates. Lung cancer is the leading cause of cancer deaths worldwide and accounts for 28% of cancer deaths in the United States (6). Current 5-year survival rates range from 8% to 14% (6,7).
The initial randomized trials, conducted in the 1970s, were performed at the Mayo Clinic, the Johns Hopkins University, and Memorial Sloan-Kettering Cancer Center and also included a clinical trial in the former Czechoslovakia. The trials consisted of variations of chest radiography and sputum cytologic tests (6–8). For screening to be effective, the cases found at screening must show a better outcome than the cases detected after the occurrence of symptoms. Superficially, this seemed to be the case, with 5-year survival doubling in those cases discovered at screening in the 1970s. However, many authors (6,9,10) have shown that lead-time bias, where the screening test uncovers the lesion earlier, prior to symptom occurrence, may increase the 5-year survival but not necessarily the final outcome. Also, screening is likely to allow detection of slow-growing indolent tumors (length-time bias), which may never produce symptoms in the elderly patient. This explains why more cancers are detected in a screened population than in a control group. Currently, no scientific advisory committee recommends lung cancer screening. In the previous studies (6–10), there was no proof that the overall lung cancer mortality rate was lower in the screened group. Ellis and Gleeson (8) stated that mortality must be the outcome measured, since other measures such as survival are affected by screening biases.
However, there is potential for lung cancer screening in the future. Boucot, as early as 1959, demonstrated that the 5-year survival rate for localized cancers detected at screening was 37% (11). Today, stage I carcinomas discovered and surgically removed are associated with a 5-year survival rate of over 60% (10). How lead-time bias may affect this is not known. Jain and Auroliga (9) stated that lung cancer is a suitable target for screening tests because (a) it is a common health problem, with an incidence of 172,000 cases in 1998 and with 160,000 deaths; (b) lung cancer has a preclinical phase, in that radiologic abnormalities almost always precede clinical symptoms; and (c) early diagnosis appears to improve patient outcome. The first two points are indisputable. The claim that early diagnosis leads to improved outcome is not yet proven.
Henschke et al (7) reported in 1999 on results from the Early Lung Cancer Action Project (ELCAP), which was initiated in 1992. They enrolled 1,000 symptom-free volunteers aged 60 years and older with a cigarette smoking history of at least 10 pack-years. Noncalcified nodules were detected by using low-dose CT in 233 (23.3%) volunteers. Among these, the nodules in 27 (2.7% of the total) were malignant, and 26 were resectable. Of the 233 individuals with nodules, 159 had one nodule, 43 had two, and 31 had three or more. The frequency of malignancy was 12% for individuals with one nodule and varied between 10% and 13% for individuals with multiple nodules. Of major interest is the fact that only 28 biopsies were performed for the 233 nodules, and 27 of these were malignant. The ELCAP protocol called for thin-section CT of all discovered nodules. If CT demonstrated benign calcifications in a nodule with smooth edges and a size smaller than 20 mm, the nodule was classified as benign. If no calcifications were detected, nodules larger than 5 mm were subjected to biopsy and nodules smaller than 5 mm were followed to detect growth. Of the resected nodules, over 80% were stage I, which theoretically should lead to a 5-year survival rate of over 70%. Again, how much the eventual outcome will be affected by lead-time and length-time bias is not known.
In Japan, under the Health and Welfare Law for the Aged, physicians must screen for gastric, cervical, breast, colorectal, and lung cancer. Screening for lung cancer is with sputum cytologic tests and chest radiography (8). It is claimed there is a 46% reduction in mortality following the introduction of mandatory screening products for all of these lesions, with an estimated amount of $93,000 per life saved.
For lung cancer screening to be most effective, one must screen separately for peripheral and central lesions. Screening with low-dose CT is effective for peripheral lesions, and sputum cytologic testing is useful for central lesions (9). New technologies are being developed, such as sputum immunocytologic and monoclonal antibodies analyses, both of which may decrease the need for biopsy by helping identify malignant disease (10).
In a recent study by Patz et al (11) with over 500 patients, no correlation was detected between the size of the tumor and survival. This raises the question as to whether detection of early small lesions on CT scans will decrease mortality. How early in the course of lung cancer do metastases occur? How variable are tumor biologic features and individual immune responses in affecting mortality?
Lung cancer screening is still in limbo. However, evidence is accumulating (eg, in the ELCAP) to show that we have the equipment to achieve early detection, and, even if there is no logarithmic relationship between size and cure, certainly early disease will be more curable. New trials will develop that will include new techniques such as immunocytologic and monoclonal antibodies analyses, and the results will be evaluated on the bases of efficiency and cost-effectiveness, particularly in individuals who are at risk. With the development of low-dose spiral CT screening for lung cancer, the concept of lung cancer screening for high-risk individuals is being reevaluated, and if over the next several years outcomes prove to be beneficial in the screened population, screening of the high-risk population may be attempted. With low-dose CT, the entire thorax can be screened in 10–15 seconds with a radiation exposure equal to that of about 10 chest radiographs (9).
Obviously, the addition of each new screening procedure in the general population reduces the finite amount of funds available for health care. This may be particularly troublesome with lung cancer screening. One of the unanswered problems in lung cancer screening is the approach to the small lesion. Thin-section CT can delineate 2- or 3-mm lesions, which, if we accept the ELCAP results, may be present in 25% of adults over 60 years of age. The ELCAP investigators recommend biopsy of noncalcified lesions larger than 5 mm, while smaller lesions are followed to detect growth. This would provide an enormous group of patients requiring costly procedures, needle biopsies and follow-up CT examinations. Jain and Auroliga (9) believe that the predictive value for low-dose spiral lung CT detection of malignancy in any given nodule is less than 10%. In the ELCAP series, CT demonstrated suggestive findings in 233 of 1,000 patients, but only 27 (12%) of these proved to be cancer. In a study by Sone et al (12) with 3,967 persons, 219 had suggestive findings, but only 19 (9%) of the 219 had lung carcinoma. Even if we assume that the majority of the malignant lesions are stage I and that the 5-year survival will be over 60%, the total cost may be prohibitive. In addition, if we use the measurable outcome of mortality, we do not know how lead-time bias and length-time bias will affect the results. We still do not know if the screening leads to a reduction in mortality.
The latest screening concept that has generated much interest is coronary artery screening. Radio and television commercials are constantly harping on the need to evaluate coronary artery status to determine whether a heart attack is impending. Since this examination is not covered by any existing managed health care plan, individuals must pay directly for the examination. The potential of high profits from these studies has stimulated considerable interest. The test has been performed with electron-beam CT to measure the amount of coronary artery calcification and has been scored with the scale developed by Agatston et al (13). The underlying concept is that coronary artery calcification has a prognostic relevance in that high calcium burdens predict future cardiac events.
Haberl and associates (14) correlated electron-beam CT results with conventional angiography results in a total of 1,764 symptomatic patients. Although a higher calcium score did, in general, relate to the degree of stenosis, there was no close correlation between the amount of calcium and arterial luminal narrowing. Coronary arterial stenosis greater than 50% was present in 56% of men and 47% of women; stenosis greater than 75% was present in 37% of men and 30% of women. In 128 (23%) of the 549 men and 116 (41%) of the 284 women without significant coronary arterial disease, no calcium was found, while in patients with significant coronary arterial disease (stenosis > 75%), fewer than 1% had no calcium. The authors thought that patients with typical angina should directly undergo angiography, but patients with atypical angina or poorly defined symptoms could be screened with electron-beam CT. Patients with low calcium scores would probably not require angiography.
Hoff et al (15) reported on 35,246 asymptomatic subjects aged 30–90 years who were self-referred for calcium screening. Calcium scores increased with age. The calcium scores in men were approximately equal to those in women 15 years older. Most men younger than 40 years and women younger than 50 years did not have detectable calcium. Hoff et al suggested a threshold value of 400 (on the Agatston scale) as the cutoff for high-risk coronary arterial disease. Rumberger et al (16) suggested a calcium threshold of 371, which they believe approximates a 70% luminal stenosis.
Raggi et al (17) reported on 676 asymptomatic patients who were followed up for 32 months ± 7 (SD). The mean calcium score in the 646 patients who did not experience a coronary event was 87, while the mean score in the 30 patients who did experience a coronary event was 388. Raggi et al thought that the overall calcium score was the best predictor of events and was proportioned to the overall plaque burden in younger patients but that the predictive value of a high calcium score was low in patients after the age of 55–60 years. Raggi et al believe that evidence of severe coronary calcification (calcium score of 
200) justifies aggressive medical intervention even in the absence of symptoms, while the absence of calcium is indicative of patients at very low risk.
It would appear that coronary calcium correlates with the degree of soft fatty plaque formation. There does not appear to be a close correlation between the amount of calcium and arterial luminal narrowing, however, and it is not known whether there is any correlation between the degree of calcification and plaque rupture (14). We do not know what percentage of patients with a high calcium score will have occlusion of their coronary arteries. There is evidence to suggest that the absence of coronary calcification is associated with a low risk of coronary arterial disease. Newman and associates (18) attempted to determine the extent of coronary arterial caclification in older adults. The mean age of their 106 participants was 78 years. Many elderly adults have considerable calcification that, in many, may represent quiescent lesions of long duration that may not pose an increased risk of a serious event. Therefore, measurements of coronary arterial calcification may be more important in younger individuals. Feuerstein et al (19) suggest that screening should start in men at age 45 and in women at age 50.
More recently, investigators have claimed that the more common fast CT scanners found in most hospitals can be synchronized with an electrocardiogram to record an image during the resting phase of the cardiac cycle. In an article in The New York Times (20), information from physicians in Miami Beach, Fla, Framingham, Mass, the University of Iowa (Iowa City), and Wake Forest University (Winston-Salem, NC) was correlated with the statement that the ability to detect calcium in the coronary arteries by using the more common CT scanners was "virtually identical" to the same ability with electron-beam CT. This would allow most of the hospitals in the country to utilize the former technique.
I do not believe that we can commit resources to screen for the presence or degree of coronary calcification at this time. There are too many questions regarding the importance of calcification. Certainly patients with typical symptoms will undergo catheter placement, but I do not believe we can afford to screen all patients with atypical or questionable symptoms. Although angioplasty has become the standard for initial treatment of coronary luminal narrowing, it has still not passed the outcome test. It has been estimated that we spend $6 billion each year on performing over 600,000 angioplasties in the United States (21). Many cardiologists question whether this adds to survival time. Steve Nisson, vice chairman of the Department of Cardiology at the Cleveland Clinic, Ohio, says that heart disease kills 50% of Americans, and we still do not know the optimal way to treat it (21). There are no studies to prove that angioplasty prevents heart attacks, but all agree that it is effective in providing symptom relief.
The evaluation of coronary arterial calcification may be the latest rage in medicine today, but it is not ready for screening. Many of my acquaintances have undergone coronary screening. One with angina showed considerable calcification and underwent angiography, which revealed a 70% stenosis; a stent was implanted. The test was unnecessary; he was symptomatic. Two other asymptomatic acquaintances underwent coronary screening; one had considerable calcification, the other almost none. Both have modified their diets—one avoiding fat and cholesterol, the other increasing his cholesterol intake. Entrepreneurs will profit from coronary artery screening, but I do not see it as a cost-effective measure for the immediate future.
Managed health care accepts screening for breast and cervical carcinoma. In my opinion, screening for colorectal carcinoma is warranted and might be accomplished with two colonoscopies, one at age 50 and one at age 65 years. Patients with adenomatous lesions or a family history would be screened more frequently. I believe that before screening for lung carcinoma can be established, we need evidence that it will affect outcome—namely, a major decrease in mortality. If this occurs, screening for lung carcinoma, particularly in individuals with a family history or who are at high risk, might be performed with low-dose CT at age 50 and 65 years, following a similar pattern to that described in the ELCAP report (7). I do not believe that screening for coronary arterial calcification is warranted at the present time.
Obviously, new screening techniques will be developed and evaluated as to their cost-effectiveness. They will have to pass the basic test, which requires that the burden of disease evaluated at screening must be sufficient, that it can be detectable with an effective test in the preclinical phase, and that there is an effective treatment. If new screening procedures are accepted, even with a cost of under $100,000 per life saved, and if compliance increases into the 70%–80% range, we may find that even with the reduction in mortality from these procedures, we are consuming too much of the health care dollar. We may not like the concept that health must be constrained by cost, but this is a truism. Only so much of the available resources can be committed to health care. As each potential screening test is evaluated, it must be balanced against what must be sacrificed to allow us to perform this test.
REFERENCES
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| RADIOLOGY | RADIOGRAPHICS | RSNA JOURNALS ONLINE |
