Following up incidental findings may do more harm than good

A healthy 40-year-old, doctors sometimes joke, is one who hasn't been worked up yet. And therein lies a dilemma for radiology and for all medicine. Mirroring progress in so many fields today, the technologies that have advanced medical imaging to its current state have far outpaced the ethical and medicolegal decisions that are crucial for determining the most appropriate use of imaging findings.

If all of this sounds a bit esoteric, it's not. In a discussion at the 2002 International Symposium on Multidetector-Row CT in San Francisco, Dr. Lincoln Berland from the University of Alabama in Birmingham used research and real-life incidental findings to illustrate the need for a more conservative approach to screening, follow-up, and intervention. A key question is what to do with incidental radiologic findings, which have exploded in the past two decades thanks to higher-resolution scanners and the advent of CT screening for a variety of conditions.

"False positives, overdiagnosis, and incidental findings are really the primary flaws of screening," Berland said. "Screening would be great if you weren't concerned that you were finding lesions that even proved to be cancer -- but really weren't significant because they were indolent and growing so slowly. If you didn't have false-positive findings that you had to pursue, and find out they were benign, screening would be great. The problem is that people are not paying attention, and it's becoming a great problem."

Inappropriate screening, follow-up, and intervention raises patient risks as well as medical costs. And there is mounting evidence, Berland said, that the current approach of working up all but the most insignificant incidental findings leads to dramatically higher costs with little or no medical benefit.

Psychology is a key factor. An incidental finding can lead to apprehension and curiosity that results in more tests, and still more incidental findings, risks, and costs, Berland said, relating the story of a physician who underwent CT screening that showed liver and lung lesions.

"He had a biopsy of the liver; he ended up having surgery in his lung because the lesion they found was relatively inaccessible," Berland said. "He had a complication from surgery, was in the hospital for weeks, he almost died, and everything was benign. This is medical cascading syndrome, where you go from one thing to the next."

Whether or not each biopsy and contrast CT exam was medically indicated became secondary to the psychology of the case, amid family concerns and a powerful need to know, a phenomenon that drives inappropriate exams every day, he said.

"So the question about incidental findings is what to pursue, what to ignore, what to follow, and what to study right now to the point of diagnosis," Berland said. He defines an incidental finding as one not related to symptoms, or previously detected abnormal medical results of any kind, including:

  • Benign lesions such as small liver cysts and granulomas that do not cause symptoms, or lead to morbidity or mortality.
  • Benign neoplasms, such as adrenal adenomas.
  • Anatomic abnormalities such as retroaortic renal veins.
  • Findings that show normal and age-appropriate degenerative changes, such as aortic atherosclerosis or spinal degeneration.
  • Findings related to old trauma or prior surgery.

Berland focused his talk on the first two categories above as they relate to the liver, the adrenal glands, and the kidneys.


Thin-section scanning with spiral CT has contributed to a marked rise in the detection of small, indeterminate liver lesions, Berland said, referencing three studies.

A study of 1,454 outpatients who underwent abdominal CT screening found hepatic lesions 1.5 cm or smaller in 17% of patients (262/1,454), 22% of which were determined to be malignant. The incidence of malignant hepatic tumors remained very low, the group concluded, even in patients with a known primary extrahepatic tumor. And there was no case in which a patient without a primary extrahepatic tumor was found to have a malignant hepatic lesion (AJR, March 1992, Vol. 158:3, pp. 535-553).

A study of 100 patients at Berland's institution (mean age 55) found small liver lesions in 20% of patients, and small adrenal lesions in 10%.

And in a recent study that reviewed CT reports of 2,978 patients, hepatic lesions of 1 cm or less were reported in 12.7% (378) of patients. Interval growth in subsequent imaging exams (6-56 months later) was reported in 11.6% of these 378 patients, while 80.2% of patients demonstrated no interval growth. Among the three most common tumors (lymphoma and colorectal and breast cancers), small hepatic lesions were metastatic in 4%, 14%, and 22%, respectively, the authors wrote (Radiology, January 1999, Vol.210:1, pp. 71-74).

"Our experience in the liver is that the incidence of detecting these is even higher than what's been reported previously, because we're using thinner slices than the studies used," Berland said. If a patient has any nodule or lesion that looks like a cyst, and there's no primary malignancy, "you'd have to conclude that the probability of malignancy is probably a fraction of a percent."

Adrenal glands

The prevalence of adrenal masses seen in imaging ranges from about 0.4% in the screening population to 4% among patients previously diagnosed with cancer, Berland said. Nearly all lesions smaller than 4 cm are benign, a few of them grow, and a few develop biochemical findings. Once they're found, should they be followed up on?

Probably not, Berland said. The most serious finding is adrenal cortical carcinoma, which constitutes fewer than 1 in 4,000 adrenal nodules. And the prognosis is so poor, with survival less than 50% at two years, that finding it probably won't help the patient.

"The smallest adrenal carcinoma we've ever seen was under 4 cm ... and it looked malignant," he said. "And the patient had it all resected and developed metastases two years later anyway."

A study in Radiology evaluated 166 adrenal masses with unenhanced, contrast-enhanced, and delayed-enhanced CT. The protocol diagnosed adenoma in 127 masses and nonadenoma in 39. In all, 96% of the masses were correctly characterized, and the group concluded that CT alone is sufficient for the task (Radiology, March 2002, Vol. 222:3, pp. 629-633).

Earlier this year, a National Institutes of Health conference on the management of incidentalomas produced a draft consensus report recommending follow-up of such findings with a complete physical history, biochemical evaluation, and possible additional radiologic studies, Berland said.

Nevertheless, the report also stated that the low incidence of carcinomas and infrequent progression to hyperfunction call into question the advisability of intense clinical follow-up of this common condition, Berland said.

As a hypothesis for study, Berland suggested that the most reasonable approach might be to do nothing with incidentalomas 3 mm or smaller, except for hormonally active nodules.


There are about 30,000 cases and 12,000 deaths from renal cell carcinoma (RCC) each year in the U.S., and about half of the cases are discovered incidentally. However, population statistics show that RCC mortality is very low, only about 7 in 100,000 cases. Yet a few small studies have shown that prevalence in the screening population is much higher, Berland said.

So if one guesses that incidence is about one-fourth of prevalence, one might conclude that only about 20% of renal cell cancers are fatal.

"That is, we may be identifying a huge number of things that are actually cancers, and if you left them alone they would never hurt the patient," Berland said.

Much more common than RCC are indeterminate Bosniak II cystic renal masses, but the risk of malignancy is extremely small. The vast majority don't warrant follow-up.

Radiologists might consider not reporting simple renal cysts up to 5 in number that are smaller than 3 cm, and not reporting most Bosniak cysts, even those with septations or mild rim calcifications if they're small.

"I'm just suggesting this as a proposal for study that I think may be more cost-effective than the haphazard approach we currently have," he said.


Many challenges lie ahead, in terms of different standards for measuring lesions, interreader variability, and the difficulty of defining the endpoints of the lesion, Berland said. Virtual colonoscopy can produce measurements of adenomatous polyps that differ from those of traditional colonoscopy, for example.

"What should be done after a finding is found?" he asked. "How do you follow these results, what are the medical and legal obligations (of changing policies) ... and what are the appropriate examination intervals?" Guidelines also need to be developed for recommending additional examinations, and for performing invasive procedures following positive imaging results.

"I think this is a huge area that may be boring, but is ripe for further discussion and study," he said. "We must develop selective, reasoned approaches to dealing with these kinds of findings."

In conclusion, Berland said that incidental findings are quite common, and that pursuing them aggressively probably does more harm than good.

By Eric Barnes staff writer
September 5, 2002

Additional information regarding the University of Alabama's approach to screening can be found at, and by visiting, and clicking on "Other Links," and then "CT Screening." Symposium sponsor Stanford Radiology offers online CME at

Related Reading

Pennsylvania takes hard line on self-referral CT screening, May 31, 2002

Experts weigh benefits, drawbacks of whole-body CT, May 5, 2002

For the person who has everything, whole-body CT makes inroads, September 11, 2001

New ARRS president cautions against turning ‘healthy people into sick people’, April 30, 2001

Copyright © 2002

Page 1 of 654
Next Page