Whole-body imaging needs more tailored approach to be effective

August 1, 2007

Clinics offering whole-body imaging can be found readily by searching the Internet. Some now even offer gift certificates. But is buying a top-to-toe examination for a friend or relative money well spent? Or could the gift cause more harm than good? Mathematics may provide the answer, according to Prof. Myriam Hunink, a professor of radiology and clinical epidemiology at the Erasmus MC-University Medical Center Rotterdam in the Netherlands.

Clinics offering whole-body imaging can be found readily by searching the Internet. Some now even offer gift certificates. But is buying a top-to-toe examination for a friend or relative money well spent? Or could the gift cause more harm than good? Mathematics may provide the answer, according to Prof. Myriam Hunink, a professor of radiology and clinical epidemiology at the Erasmus MC-University Medical Center Rotterdam in the Netherlands.

The probability of disease is a key variable when weighing the pros and cons of whole-body imaging, Hunink told delegates at this year's European Congress of Radiology in Vienna.

"The bottom line is that we need an individualized, tailored approach," she said.

Hunink outlined a screening scenario: Probability of disease was 1% for a patient cohort of 100,000, and the sensitivity and specificity of the imaging examination were 90% and 80%, respectively. In a 100,000-patient cohort, imaging would find 900 of the 1000 people who actually had the disease (true positives) and miss 100 (false negatives). The 80% specific test would correctly identify 79,200 of the 99,000 remaining individuals as disease-free (true negatives) but classify 19,000 healthy subjects as having the target disease (false positives).

"The ratio of false positives to true positives is 22. That implies that for every true positive that you identify, you are going to label 22 healthy people as being ill," Hunink said. "Apart from the anxiety and risks, this is going to take up a lot of healthcare resources."

Screening for two diseases at the same time would only make matters worse. She illustrated this by adding a search for a second pathology with a 2% probability of disease. The sensitivity of finding one disease rises to 99%, but the sensitivity of finding both is just 81%, she said. Specificity drops to 64%, resulting in 55,000 false-positive findings.

"That's just for two diseases," she said. "If you go on and look for eight diseases, you can imagine how many false positives there will be. So the only normal patient is the one who has not undergone a complete diagnostic workup."

The benefit of a screening test also depends on the impact of early diagnosis. Advertisements for whole-body imaging checkups generally suggest that seeing pathology sooner has a positive impact on outcome. This may be true for certain pathologies, but not all. Genuine pathology detected on whole-body imaging may also be irrelevant if it is advancing slowly.

Randomized clinical trials would be the best way to measure the efficacy of whole-body imaging-led screening, Hunink said. Obtaining clinical trial data of sufficient quality can take many years, however.

An alternative option, available today, is to use a mathematical simulation tool known as decision modeling. This approach requires researchers to enter all possible strategies, together with their consequences and outcomes, into the computer model. The cost for each quality-adjusted life year can then be calculated. Researchers can also track how varying parameters will alter the extra cost required to achieve an increase in efficacy.

Similar calculations can be performed to examine the utility of whole-body imaging in symptomatic patients, Hunink said. Top-to-toe CT of multitrauma patients, for example, could have a real benefit given the high chance of seeing a suspicious finding on imaging and the possible harm that would be caused if an underlying injury went undiagnosed.

"Here, the picture is totally different. This could be a useful thing to do if you select your patients well. Again, that's the key issue, and that has everything to do with the probability of disease," she said.

Technologically, it is possible to scan the entire arterial system with CT and MRI, but clinically it has not been proven to be of great benefit in whole populations, according to Prof. Geoffrey Rubin, chief of cardiovascular imaging at Stanford University in Palo Alto, California.

Preliminary studies looking at cross-bed correlation show that certain vascular beds correlate well with some other vascular beds, while others do not. Many outcome studies need to be conducted to determine the path of preclinical lesions. Until then, clinicians run the risk of putting in stents or performing surgery when it is not really needed, he said.

Rubin favors using whole-body imaging to investigate atherosclerosis as a disease, to better understand the diversity of its manifestations. He does not advocate screening asymptomatic vascular territories with the intent to treat them upon finding lesions.

"There is very little evidence to support doing so," he said.

A few preliminary reports have been done for whole-body MRI, but no outcome studies compare treatment based upon 64-slice CT with the reference standard of catheter angiography. This is a recognized deficiency, and some outcome studies are in the works, particularly looking at the acute chest pain patient, Rubin said.

"Physicians, radiologists, entrepreneurs, and a large portion of the population were, and still are, attracted to the idea of whole-body imaging. Yes, there is a place for whole-body imaging, but only for very clear indications. In this setting, it is feasible," said ECR session moderator Prof. Gabriel P. Krestin, head of radiology at Erasmus University Medical Center in Rotterdam.