This article is more than 5 years old. Information may no longer be current.
Calculator uses clinical, ultrasound variables to better predict thyroid cancer risk
A thyroid cancer risk calculator that takes into account clinical and ultrasonographic variables, including age, sex, nodule size and the presence of cystic fluid, can reliably predict risk for malignancy in most patients, according to findings from a large database analysis.
“Although some sonographic features, such as microcalcifications, hypoechogenicity and irregular margins, associate with malignant disease, the interrater variability on interpreting these findings remains high,” Trevor E. Angell, MD, endocrinologist at Brigham and Women’s Hospital, and colleagues wrote in the study background. “In contrast, nodule size and presence of cystic fluid have much higher consistency when subjected to blinded review but have only variably been associated with increased [risk for malignancy] in selected populations. Importantly, no investigation of an unselected, consecutive cohort has yet placed all of these commonly obtained and reproducible clinical variables (age, sex, nodule size, cystic component and multinodularity) into a single model which can be applied to clinical care. The ability to do so would enable individualized integration of the risks posed by these variables and improve the approach to diagnostic fine needle aspiration biopsy.”
Angell and colleagues analyzed data from 9,967 consecutive adult patients evaluated with ultrasound-guided fine-needle aspiration for a thyroid nodule at least 1 cm in size between 1995 and 2017 at Brigham and Women’s Hospital (84% women; median age, 53 years; median nodule size, 1.7 cm). Researchers used mixed-effect logistic regression analysis to estimate the risk for thyroid nodule malignancy using patient and sonographic variables.
Within the cohort, thyroid cancer was confirmed in 1,974 of 20,001 thyroid nodules (9.9%), according to researchers. Predictors of malignancy included age younger than 52 years (OR = 1.82; 95% CI, 1.63-2.05) and male sex (OR = 1.68; 95% CI, 1.45-1.93). Risk for malignancy also increased with increasing nodule size. Compared with patients with thyroid nodules between 1 cm and 1.9 cm, the OR for malignancy was 1.3 (95% CI, 1.14-1.49) for a nodule between 2 cm and 2.9 cm, 1.59 (95% CI, 1.34-1.88) for a nodule between 3 cm and 3.9 cm, and 1.71 (95% CI, 1.43-2.04) for a nodule at least 4 cm.
In contrast, the presence of additional nodules at least 1 cm in size was associated with reduced odds for malignancy when part of a multinodular gland, according to researchers, with ORs of 0.69 (95% CI, 0.6-0.79) for patients with two nodules, 0.41 (95% CI, 0.34-0.49) for patients with three nodules and 0.19 (95% CI, 0.16-0.22) for patients with at least four nodules.
Researchers than used regression coefficients from the multivariable model to generate a prediction equation for the probability of malignancy for any given thyroid nodule based on the cumulative impact of clinical and sonographic variables. The thyroid nodule cancer risk calculator provides an estimated risk for malignancy for a given individual once patient age and sex, nodule size and degree of cystic component, and the number of nodules at least 1 cm in size are entered, the researchers wrote.
“For example, a solitary 1.5 cm thyroid nodule that is partially cystic (25-75%) in a 55-year-old woman has an estimated risk of 5.23%, whereas a 3.2 cm solid nodule in a 30-year-old man who has one other nodule 1 cm has a 29.9% risk of being malignant,” the researchers wrote.
The researchers noted that, for indeterminate nodules, final categorization is less certain and analyses categorized these nodules based only on histopathologic confirmation.
“These data demonstrate use of easily obtainable and commonly reproducible variables to generate precise risk of malignancy within any nodules to improve personalized discussion of care,” the researchers wrote.
The thyroid cancer risk calculator is available at http://thyroidcancerrisk.brighamandwomens.org/. – by Regina Schaffer
Disclosures: The authors report no relevant financial disclosures.