July 14, 2008

Taking the Measure of Thyroid Cancer
By Beth W. Orenstein
Radiology Today
Vol. 9 No. 14 P. 20

Up to 60% of adults have thyroid nodules that can be seen on ultrasound, and nearly one half the population will have thyroid nodules at autopsy.

“If you have a circle of friends, half of them will have a thyroid nodule,” says Won-Jin Moon, MD, an associate professor in the department of radiology at Konkuk University Hospital in Seoul, South Korea.

Thyroid nodules are more common in older adults, women, and those living in iodine-deficient regions or who have had a history of radiation exposure. “As you may know, after the nuclear disaster at the Chernobyl power plant in 1986, the incidence of thyroid carcinoma was markedly increased in Russia,” Moon says. Genetics also may play a role.

People with thyroid nodules do not always have symptoms, although a few complain of neck, jaw, or ear pain. In rare instances, if a thyroid nodule rubs against the voice box, it can cause hoarseness.

Fortunately, the vast majority of thyroid nodules are benign, and thyroid cancer accounts for only 1% of cancers diagnosed in the United States. However, because some thyroid nodules may be malignant and, if they are, can often be successfully treated, it is important to investigate when they are found. “Although there are subtypes of thyroid carcinoma with relatively poor prognosis [anaplastic carcinoma, medullary carcinoma, papillary thyroid carcinoma] most thyroid carcinomas have a very good prognosis with a 10-year survival of 93% and a 20-year survival of more than 95%,” Moon says.

Some thyroid nodules are found by physical exam (palpitation) and some with sonography. Many thyroid nodules are found incidentally when imaging the head and neck area for other reasons.

If a nodule is found, a thyroid-stimulating hormone (TSH) blood test and a radionuclide thyroid scan are likely to be performed. However, TSH levels and radionuclide thyroid scans only indicate whether the nodule is functional or nonfunctional, not whether it is malignant or benign. Although almost all malignant nodules are nonfunctional, a functional nodule doesn’t mean it’s benign, since 5% to 8% of functioning nodules are found to be malignant, Moon says.

The most common method for evaluating thyroid nodules for malignancy is fine-needle aspiration (FNA), a procedure in which a very fine needle is inserted into the thyroid and sample cells extracted and examined under a microscope.

Until recently, FNA was considered the only method to determine whether a nodule was malignant or benign, Moon says. Now, she and her colleagues from the Korean Society of Neuro- and Head and Neck Radiology have published a retrospective study in the June issue of Radiology in which they show that sonography can be a good imaging technique for differentiating benign and malignant thyroid nodules.

“In our study,” Moon says, “we achieved, with reasonable inter-reviewer reliability, criteria on ultrasound that can be used to determine whether thyroid nodules are malignant.” The researchers found that five criteria showed a significant association with thyroid carcinomas. “In our study, the presence of at least one malignant finding on ultrasound had a sensitivity of 83.3% and a specificity of 74% and a diagnostic accuracy of 78%,” she says.

She adds that the implications of this study on patient care are exciting. “If you consider that about 60% of the general population has one or more thyroid nodules, it is impractical to differentiate benign from malignant nodules via FNA biopsy in all patients who have thyroid nodules,” she says. At the same time, ultrasound provides high-resolution anatomical images, does not require any contrast agents, and is readily available. “Therefore, ultrasound can be an alternative to FNA biopsy if the proper ultrasound criteria exists,” she says.

In 2003, the researchers examined sonograms from 831 patients (716 women and 115 men) with a mean age of 49.5 years. (The median age for a thyroid cancer diagnosis is 48.) The patients had 849 nodules that were diagnosed at surgery or biopsy. Of the nodules, 360 were malignant and 489 benign. The size of the nodules ranged from 2.7 to 64 millimeters.

Three radiologists retrospectively evaluated the characteristics of the nodules on the sonograms. They looked at nodule size, the presence of spongiform appearance, shape, margin, echotexture, echogenicity, and the presence of microcalcification, macrocalcification, or rim calcification. An X2 test and multiple regression analysis were performed, and sensitivity, specificity, and positive and negative predictive values were calculated.

They concluded that nodules are more likely to be malignant if they appear taller on ultrasound rather than wide, have spiculated margins, and show marked hypoechogenicity. Also, microcalcifications and macrocalcifications are likely to be present.

Moon says that the findings support previous study results that suggest a taller-than-wide shape is specific for differentiating malignant thyroid nodules from benign ones. “This result conveys the fact that malignant nodules [taller than wide] grow across normal tissue planes while benign nodules grow parallel to normal tissue planes,” she says.

The ultrasound findings for benign nodules were isoechogenicity and a spongiform appearance, Moon adds.

The researchers evaluated the diagnostic accuracy of the ultrasound features separately for nodules that were smaller or larger than 10 millimeters. “The ultrasound features suggestive of malignancy showed a higher specificity and a lower sensitivity for larger nodules than for smaller nodules,” Moon says.

After all their analyses were complete, the researchers concluded that the size of the thyroid nodules may affect the diagnostic accuracy of the sonograms. Some nodules can be falsely diagnosed as malignant even if they are benign and would require a biopsy follow-up, she says. “We feel that different strategies for ultrasound diagnosis are needed for nodules with different sizes,” she notes.

Moon and her coauthors began looking for the ultimate ultrasound criteria for differentiating between malignant and benign thyroid nodules even though it seemed impossible, she says. “Many ultrasound-based studies have their own weaknesses,” she says. Part of the problem is a lack of consensus about ultrasound terminology and vague definitions of ultrasound findings. Also, ultrasound is an operator-dependent modality. The sonographer’s inexperience and the lack of resolution on older equipment also can be problematic, she says.

To overcome these issues, the researchers first reached a consensus about the ultrasound terminology for thyroid nodules and then reviewed ultrasound data from high-end ultrasound equipment operated by highly experienced radiologists, she says. Having a standardized protocol for thyroid ultrasound and standardized training for ultrasound specialists allowed the researchers to control the variability due to operator dependency, Moon says. “In our study, we did have relatively good interobserver agreements in interpretation of the ultrasound findings,” she says.

The researchers’ experience with performing and evaluating thyroid sonograms ranged from six to nine years. They held two training sessions to further establish a baseline consensus. Each was asked to review the studies independently in a single session.

Some physicians believe that if they overdiagnose, they have done no harm, Moon says. “They say, ‘If you see a nodule and it looks like it is malignant, then aspirate it.’ But in reality,” she says, “there are small but not trivial amounts of false negatives. Indeed, if a nodule is smaller than 1 centimeter, the rate of false negative is even higher. What if you are a patient with a malignant-appearing nodule with just 0.3 centimeters? Do you want to have aspiration biopsy? Do you want to get another biopsy if the biopsy is also inconclusive?”

Interestingly, Moon points out that in the book Complications: A Surgeon’s Notes on an Imperfect Science,  Atul Gawande, MD, presents a similar situation with breast nodules and microcalcifications. “And thyroid carcinoma has a much better prognosis than does breast carcinoma,” she says.

Moon says the next step is a prospective study to support and corroborate the researchers results. “Also,” she says, “we have to look for whether there is any specific finding for other subtypes of thyroid carcinomas [follicular and medullary carcinomas). Our ultrasound findings for malignancy are in fact ultrasound findings for papillary thyroid carcinoma.”

Moon says research is also needed on which ultrasound findings predict poorer prognosis and higher recurrence rates for malignancy. “I believe there must be two or more different subsets of papillary thyroid carcinomas,” she says. According to the American Cancer Society, papillary thyroid carcinoma is the most common type of thyroid cancer, representing 60% to 80% of cases.

Moon believes that researchers will eventually develop a new method to identify different subsets of common papillary thyroid carcinomas with different prognoses without surgery. “It might be molecular imaging,” she says, “or it might be novel serum markers.”

In the meantime, Moon says, the application of ultrasound criteria for the differentiation of malignant from benign nodules may help in the accurate diagnosis of thyroid nodules and reduce the number of unnecessary FNA biopsies that are performed.

— Beth W. Orenstein is a freelance medical writer who lives in Northampton, Pa., and frequently contributes to Radiology Today.