Dr. Berland is a Professor and the Vice-Chairman for Development in the Department of Radiology, University of Alabama at Birmingham, Birmingham, AL and is Vice-President of the Society of Computed Body Tomography and Magnetic Resonance (SCBT/MR).

Controversies about medical screening tests often center on ethics and economics. ¹ Knowledgeable opinions regarding these issues abound but beg to be put into perspective with the simple question of whether the screening test works. Unfortunately, scientific proof always comes at a high price and with long delay, if ever. Furthermore, whole-body screening (WBS) with computed tomography (CT) did not originate within the scientific community, which has received it coolly and has studied it little.

The purpose of this article is to assess the value of WBS by reviewing what little we know about it and to evaluate it by deconstructing the examination. That is, by analyzing the most common important diseases that are likely to be detected by WBS, we may learn whether the sum of such detection events are likely to lead to enough benefit to warrant the cost.

Defining WBS

Whole-body screening is unusual among screening tests because it is intended to detect multiple conditions. The press and radiology literature usually refer to WBS as a practice of performing CT examinations of the chest, abdomen, and pelvis. The scan protocol is usually at a low radiation dose and with no contrast administration, intended to provide CT scanning at relatively lower risk and cost compared with conventional CT. The subjects of these scans are usually self-referred, asymptomatic individuals.

Some centers perform scans with intravenous contrast material to increase the detection rate and improve the characterization of abnormalities. However, this adds time, cost, and risk. A recent article cited the example of a successful malpractice suit against a radiologist who performed a contrast-enhanced WBS on an asymptomatic person who then died from an anaphylactic reaction. ²

The purpose of the chest CT component of the scan is primarily for early detection of lung cancer, which represents a distinct controversy, and will not be discussed further. ^3,4 Therefore, this article will address WBS as a noncontrast examination of the abdomen and pelvis.

CT colonography (CTC) is not usually considered part of WBS because it requires bowel preparation, multidetector helical scanning, and expert radiologists and technologists with focused training and experience. CT colonography has received considerable attention, is under intense study, and already represents a promising screening technique for colon cancer and colon polyps. Ironically, while CTC is not included as a part of WBS, CTC actually represents a form of WBS because the entire abdomen and pelvis are scanned without contrast. Therefore, regardless of whether WBS becomes an accepted method for screening patients, it will be important to learn how to manage patients undergoing WBS because so many will have CTC.

Defining success

Conclusions regarding the value of WBS often hinge on differences about how to measure the value of a screening examination. Proposed measures of success may include: 1) the detection rate of cancers; 2) the detection rate of other medical conditions; 3) the value conferred by self-referring consumers, who may believe that they are exerting increased control over their health; and 4) the value the consumer experiences by increased contact with a physician, usually in reassuring them of their health. There is no disagreement that WBS detects diseases and thus can lead to improving the health of some subjects. What is in doubt is whether the value of the lives extended by detecting diseases with WBS is greater than the cost measured in both health outcomes and dollars.

Limitations of screening: False-positives, overdiagnosis, and incidental findings

Why isn't medical screening as beneficial as it intuitively seems? It would be, except for the disadvantages of false-
positives, overdiagnosis, and incidental findings, which will be defined and discussed more fully. This trio of obstacles to accuracy leads us to finding many more lesions than actually could harm the subject. Identifying such false lesions may actually lead to costs and harm, because of additional testing and procedures. These adverse effects can be measured in pain, anxiety, morbidity, loss of productivity, and even death.

False-positives represent the most well-recognized source of error on screening tests. However, the magnitude of the problem is often underappreciated. For example, if one considers testing a population of 100,000 people with a prevalence of disease of 1%, using a test with a high sensitivity and specificity of 95%, this would lead to 950 true-positive tests, but also 4950 false-positive tests, for a positive predictive value of only 16% (Personal communication, G. Scott Gazelle, MD, PhD, MPH, March 2003). This would lead to a large number of other tests to verify or exclude the diagnosis.

Overdiagnosis is a more insidious form of false-positive because it signifies a true detection of the target disease, but in a form that would not lead to mortality or morbidity. An example would be a small renal cell carcinoma (RCC) that grows so slowly that it would not reach a significant size or metastasize before the patient dies of other causes. Overdiagnosis is surprisingly common and varies with the condition considered.

It is well known by radiologists (and less known by other clinical physicians and even less so by healthcare consumers) that incidental lesions of no clinical consequence are extraordinarily common. These incidental findings can be described as benign lesions that do not lead to symptoms, morbidity, or mortality. These include such conditions as adrenal nodules and liver lesions that are too small to characterize. Unfortunately, radiologists have not come to a consensus on the rational management of such findings, and inconsistently recommend further studies.

How, then, can we decide whether to do a specific type of medical screening? The commonly accepted requirements for an effective screening test are: 1) the disease should be common and associated with clinically important mortality or morbidity; 2) the screening test should be available, acceptable, feasible, and sufficiently accurate to detect early disease;
3) earlier diagnosis and treatment should be associated with improved prognosis; and 4) the sum of benefits should outweigh the sum of potential harms and costs.

Lesions found with WBS

The conditions that CT can find in the abdomen and pelvis that might fit these criteria--especially ones that are accepted as being associated with clinically important mortality or morbidity--include: 1) liver lesions, 2) adrenal nodules, 3) abdominal aortic aneurysms, 4) renal cell carcinoma, and 5) ovarian carcinoma, each of which will be considered individually. Additional abdominal conditions may be detectable on abdominal scans, such as colon cancer, or lung cancer in the upper abdominal images. However, specialized CT screening tests have been developed for these conditions and are under extensive study, as noted above. Miscellaneous conditions may also occur, as outlined in a Web site of a screening organization:

[Our] Full Body Scan is the world's most complete health evaluation. The Full Body Scan can safely and accurately detect over 100 life-threatening diseases in the arteries, heart, lungs, liver and other major vital organs before its too late. ⁵

This Web site, and sites of other screening companies, list detectable conditions including lymphoma, pancreatic cancer, gallstones, kidney stones, osteoporosis and many others. However, while all of these conditions are reported on abdominal and pelvic CT, the probability of benefit of detecting asymptomatic gallstones, kidney stones, or even lymphoma by CT is doubtful. Such conditions as osteoporosis and carotid stenosis have their own more sensitive, more specific, and less expensive tests.

Liver

Two studies have evaluated the importance of small liver lesions (<1.0 to 1.5 cm). ^6,7 These studies have primarily evaluated cancer patients, finding a frequency of 13% and 17% of such lesions, with a malignancy rate of 12% and about 22%. However, of 262 patients without a primary malignancy among the 1454 studied, not one small liver nodule was malignant.

I performed an informal review of 100 consecutive outpatient abdominal CT examinations, excluding patients with cirrhosis, obvious liver metastases, or polycystic kidney and liver disease. Small liver lesions were present in 22% of these patients, many with multiple lesions. It is likely that the increased resolution with modern helical and multidetector CT has increased the visibility of small lesions. Therefore, the probability that a small lesion is clinically important in someone without a known malignancy (the vast majority of screening candidates) is much less than 1%, while the chance of finding at least one small lesion is >20%.

What about detecting liver lesions >1.5 cm? An example of the difficulty in managing such findings is highlighted in the case of an 18-year-old woman with renal calculi who had an incidentally discovered liver lesion approximately 2 cm in diameter (Figure 1). At a follow-up scan 6 months later, the lesion had appeared to grow. The patient and her parents expressed great concern and encouraged the physician to pursue the finding. A contrast-enhanced scan demonstrated that the lesion was the same size as on the original scan and was a benign hemangioma.

Most lesions >1.5 cm are benign hemangiomas, focal nodular hyperplasia, noncalcified granulomas, or hamartomas. This illustrated example highlights the futility of attempting to minimize the pursuit of some lesions, because most physicians would probably consider the apparent growth as an appropriate indication for further study. Structured criteria for decisions to pursue such lesions based on cystic appearance, number of lesions, age, and history of malignancy may help minimize unnecessary studies.

Adrenal

Adrenal nodules have been the subject of several recent studies and even of an NIH Adrenal Incidentaloma Consensus Conference in February, 2002. ⁸ However, an asymptomatic adrenal nodule in a person without a primary cancer diagnosis is rarely clinically important. The prevalence of adrenal nodules in otherwise healthy people is reported as 0.4%, and approximately 4% in patients with a prior history of cancer. However, in our informal review discussed above, we discovered small nodules in >10% of patients. One presumed objective of finding adrenal nodules is to prevent death from adrenal cortical carcinoma, which is usually fatal. However, this disease is rare, affecting only 4 to 12 people in 1,000,000 and only 1 in 4000 adrenal nodules. Also, evidence suggests that early detection is unlikely to lengthen life.

Detecting adrenal metastases from other cancers is unlikely to substantially affect outcome. There is little evidence that asymptomatic, benign, hormonally active adrenal nodules represent a significant public health problem or that detecting them would be a cost-effective justification for screening. The consensus conference re-port recommended that incidentally discovered adrenal nodules be worked up with a history, physical examination, biochemical evaluation, and possibly more imaging. ⁸ However, in its conclusions, the conference report stated that they: ...call into question the advisability of the current practice of intense clinical follow-up of this common condition. ⁸ Therefore, the types of lesions likely to be detected, the frequency with which nodules are detected (requiring further imaging or biopsy), and the low probability of affecting outcomes makes adrenal nodule detection a poor justification for CT screening.

Abdominal aortic aneurysms

Abdominal aortic aneurysms (AAA) account for approximately 10,000 deaths per year in the United States. This accounts for 1.2% of the mortality in men over the age of 65 years and is a small fraction of the total deaths from other cardiovascular diseases. However, studies have suggested there is value to ultrasound screening for AAA, with studies showing a mortality reduction from ruptured AAA of 21% to 55%. ^9,10 Therefore, while ultrasound screening may reduce deaths, the cost-effectiveness of screening with CT remains in doubt because of the relatively low prevalence in the population. Furthermore, if one chose to screen for AAA, ultrasound would be a better choice, because it is cheaper, yet still has a high sensitivity and specificity. Therefore, even if it is concluded that screening for AAA should be done, it still cannot be used to justify WBS.

Renal cell carcinoma

The rate of deaths per year in the United States from RCC is approximately 12,000--similar to that for AAA--although 31,000 new cases per year are detected. This represents approximately 2% of deaths from all forms of cancer. While radiologists are quite familiar with the common discovery of incidental RCC during CT scanning for other reasons, the question must be asked as to what benefit derives from such discoveries. Analysis of incidence and death statistics reveals that the rate of detection of RCC has increased three-fold more than the increase in mortality from RCC ¹¹ ; this increase is almost certainly related to the marked increase in the use of CT and other imaging studies. Sixty-one percent of all cases of RCC are now discovered incidentally, and most are cured. There is a marked discrepancy between death statistics of 3.5 per 100,000 population and detection rate of lesions in the few ultrasound screening studies that have been done of 1 to 3 in 1000. Also, RCC is found in 22% of autopsies.

All of these facts infer that a high percentage of cases of RCC represent disease that the patient would die with rather than from, which is termed pseudodisease and represents overdiagnosis. Therefore, the increased detection rate caused by screening is likely to lead to a substantial number of unnecessary radical or partial nephrectomies, with their consequent morbidity and small mortality. Therefore, the cost-effectiveness of screening for RCC does not seem highly plausible, based on the relatively low incidence of disease and the high rate of pseudodisease. Thus, it is difficult to use the detection of early RCC to justify WBS.

Ovarian carcinoma

There has been substantial interest in ultrasound screening for ovarian carcinoma because it is usually discovered in a late stage and has a poor cure rate. There are approximately 24,000 new cases of ovarian cancer yearly in the United States with approximately 14,000 deaths. Studies of screening have been disappointing, with one report showing the detection of 17 cancers among more than 57,000 women ¹² and another finding 17 stage I cancers in more than 183,000 subjects. ¹³ Unfortunately, with today's technology, one must conclude that screening is not cost-effective for ovarian carcinoma. Again, as with AAA, even if one were to believe that screening was a reasonable strategy, most would agree that ultrasound is less expensive and would have a much higher sensitivity and specificity than CT for early-stage cancer. Therefore, detecting ovarian cancer also cannot be used to justify WBS.

Conclusion

The practice of whole-body CT screening has led to a raucous debate over the wisdom and ethics of performing such screening for self-paying, asymptomatic volunteers. Subjects for such studies are often implicitly promised peace of mind or extended life from early detection of otherwise fatal diseases.

In this article, I have reviewed the five most common potentially fatal diseases that proponents of WBS might cite as justifying this test. In each case, the probability of benefit is doubtful or an alternative screening test is more reasonable. Therefore, it is also likely that the sum of additional instances of false-positive findings, pseudodisease, and incidental findings from WBS would likely lead to considerable additional cost, morbidity, and even mortality from subsequent tests and procedures. Even accurate early detection of disease may not translate to lengthening life. Does the benefit to helping one sick person balance the harm caused to one or several healthy ones? I believe that this analysis suggests that WBS is currently not a reasonable medical practice.