This article is more than 5 years old. Information may no longer be current.
Gene expression assay successfully predicts benign disease in indeterminate thyroid nodules
Click Here to Manage Email Alerts
HOUSTON — A novel gene-expression classifier has been validated in reclassifying otherwise cytologically inconclusive results from thyroid biopsies as either benign or suspicious, according to data presented here at the Endocrine Society’s 94th Meeting and Expo. The results were simultaneously published today in the New England Journal of Medicine.
“Thyroid nodules are very common; this year alone, in the US, nearly half a million diagnostic fine-needle aspirations will occur in the work-up of these nodules. Though many of them are benign, there’s about a 5% to 15% risk that any thyroid nodule is cancerous. So evaluation is recommended, though the current diagnostic methods are both imprecise and, at times, inaccurate,” Erik Alexander, MD, study author and physician-researcher in the division of endocrinology, diabetes and hypertension at Brigham and Women’s Hospital, said during a press conference here.
Erik Alexander
A new option
The classifier, the Afirma Gene Expression Classifier (Veracyte), measures the expression of 167 genes, and has previously shown promise in improving preoperative risk assessment.
Over the course of 19 months, Alexander and colleagues prospectively collected 4,812 fine-needle aspirations for thyroid nodules ≥1 cm from 3,789 patients treated at 49 clinical sites in the US. Twelve percent (577) of samples were indeterminate, 413 of which had subsequent surgery allowing for histopathological review of the excised lesions.
A total of 265 independent, indeterminate nodules were used in the primary analysis to examine the efficacy of the novel gene-expression classifier. The reference standard was based on results of central, blinded histopathological review by two expert pathologists.
Alexander said the goal was to “create a real-world experience, and therefore allow high translatability of these results back to the clinic.”
According to this standard, 32% (85 lesions) were found to be malignant.
The novel gene expression classifier correctly identified 78 of these samples as “suspicious,” yielding a sensitivity of 92% (95% CI, 84-97). Additionally, 93 of 180 nonmalignant samples were correctly identified as benign, yielding a specificity of 52% (95% CI, 44-59).
The sensitivity and specificity for nodules classified as atypia of undetermined significance were 90% (95% CI, 68-90) and 53% (95% CI, 43-63), respectively. For those classified as follicular neoplasms or lesions suspicious for follicular neoplasm, the sensitivity was 90% (95% CI, 68-90) and the specificity was 49% (95% CI, 36-62). Lastly, for those nodules classified as lesions suspicious for malignancy, the sensitivity was 94% (95% CI, 80-99) and the specificity was 52% (95% CI, 30-74).
Promising NPVs
Among these three independent subtypes (atypia of undetermined significance, follicular neoplasms and suspicious for malignancy), the percentage of malignant lesions was 24%, 25% and 62%, yielding negative predictive values (NPVs) of 95%, 94% and 85%, respectively.
The authors also tested 47 independent cytologically benign nodules, as well as 55 cytologically malignant nodules. Three of the 47 cytologically benign samples proved malignant on histopathological review; the assay correctly identified all three of these samples as suspicious.Additionally, the classifier had 100% sensitivity when considering the 55 cytologically malignant samples, calling them all ‘suspicious’ for malignancy.
There were seven false-negative results, analysis of which demonstrated that six had a scarcity of thyroid follicular cells, which suggested a strong likelihood of insufficient sampling of the nodule, they said.
Future use, costs
“These data confirm the performance of the novel gene-expression classifier with high NPV designed to preoperatively classify the malignancy risk for cytologically indeterminate thyroid nodules,” Alexander concluded. “For all cytologically indeterminate nodules, the NPV was 93% in a population with a cancer prevalence of 32%. I would also just mention that NPV is strongly influenced by prevalence of cancer in a population, and this may or may not be fully representative of what exactly is seen in clinical practice; others have found lower cancer prevalence, which would raise NPV higher, if applied.”
During a question and answer session following the press conference, Alexander said that sonography is viewed as critically important to risk-stratify patients and nodules, and remains a mainstay diagnostic approach to anyone with a nodule. However, “a substantial concern is suboptimal inter-rater reliability; the ability for two separate radiologists to independently and accurately identify the same radiologic findings is imperfect.”
Richard T. Kloos
When asked about cost, study author Richard T. Kloos, MD, said the price for the assay is in the range of $3,200, and the test is already covered by Medicare and other health care providers. In addition, a patient assistance program is currently in place.
“We certainly are all very aware of cost issues in terms of total US health care expenditure,” Alexander added. He also said results of a cost analysis demonstrated “that this can be a cost savings, overall, to the system - separate from the reduced morbidity of avoiding surgery. – by Stacey L. Fisher
Reference:
Disclosures:
- Brigham and Women’s hospital has received a grant from Veracyte and support for travel to meetings for the study and other purposes. Dr. Alexander has received consulting fees from Asuragen, Inc. and Genzyme, Inc. He has also received grants from Veracyte, Inc. and Genzyme, Inc. He also receives fees from Boston Clinical Research Institute as part of the Adjudication Committee for Thyroid Disease for ongoing GLP-1 studies in diabetes. He is an unpaid member of the American Thyroid Association, and the Thyroid Cancer and Nodules Guideline Committee. He is on the editorial board for the Journal of Clinical Endocrinology & Metabolism.