Introduction
According to the latest global cancer data released by the American Cancer Society, thyroid cancer has one of the highest incidences of cancer worldwide, ranking seventh among women [
1]. Ultrasound is the most commonly used imaging method to assess malignant thyroid nodules based on the Thyroid Imaging Reporting and Data System (TI-RADS) guidelines [
2‐
4]. For thyroid nodules that are difficult to identify only by ultrasound, ultrasound-guided fine-needle aspiration biopsy (FNAB) is usually recommended for routine diagnosis. Currently, FNAB results are divided into six diagnostic categories according to The Bethesda System for Reporting Thyroid Cytopathology (TBSRTC): (I) nondiagnostic or unsatisfactory, (II) benign, (III) atypia of undetermined significance or follicular lesion of undetermined significance (AUS/FLUS), (IV) follicular neoplasm or suspicious for a follicular neoplasm (FN/SFN), (V) suspicious for malignancy (SUSP), and (VI) malignant [
5,
6].
Approximately 20% of FNAB of thyroid nodules are classified as Bethesda III/IV/V, which are defined as cytologically indeterminate thyroid nodules (ITNs) [
6‐
8]. ITNs should be treated by surgery or follow-up, depending on the clinical risk factors, ultrasound patterns, and patient preferences [
9]. Currently, histopathological examination is the gold standard for diagnosing nodules, and the risk of malignancy in ITNs varies from 5 to 30% [
10‐
14]. In other words, many patients may have undergone unnecessary surgeries. Therefore, the challenge lies in distinguishing between benign and malignant ITNs before surgery [
15].
Some research indicates that a single test can help in the management of ITNs. Blood tests for carcinoembryonic antigen (CEA) and calcitonin are helpful for distinguishing medullary thyroid cancer (MTC) from other types of thyroid cancer in a sensitive manner [
16,
17]. In addition, the usefulness of CEA and calcitonin was also demonstrated by immunohistochemistry in histopathology [
16,
18]. Considering the risk of malignancy of ITNs, the side effects on patients with advanced or metastatic MTC, and the financial burdens of additional imaging evaluation, the use of serum indicators is economical and necessary [
19‐
22]. For ITNs, recent studies have shown that as an alternative to surveillance and diagnostic surgery, molecular testing may be used to supplement the risk assessment of malignant tumors. Two kinds of genetic tests, the Afirma Gene Expression Classifier and ThyroSeq genomic classifier, are the most common products in developed countries [
23‐
25]. Although these products can detect approximately 90% of benign thyroid nodules and can help half of patients with indeterminate cytology avoid diagnostic surgery, they are not yet available in most countries and are expensive. More importantly, these assays are invasive and require at least two additional needle insertions for the analysis of genomic expression. As an alternative, 5 of the most common thyroid cancer-associated somatic mutations (
BRAF,
HRAS,
NRAS,
KRAS, and
TERT) have been applied in clinical practice. The detection of these five DNA mutations does not require additional aspiration, and the cytological samples obtained from the first FNAB can be used. The five most common somatic mutations have been widely validated to identify a large proportion of thyroid cancers [
26‐
28]. In addition, it has been reported that repeat aspiration can help to confirm the benign or malignant of ITNs [
29‐
32].
However, combinations of current auxiliary diagnostic methods remain lacking. The aim of our research is to provide a risk assessment tool by integrating serum CEA and calcitonin levels, molecular alterations, and repeat cytology results that might be useful in surgical decision-making. To minimize damage and cost, the risk assessment starts with blood testing, followed by supplementary DNA mutation detection on the first FNAB, and finally repeat FNAB. We describe the results of a retrospective, single-center study validating this evaluation tool in patients with ITNs.
Methods
Study population
Our 34-month retrospective study included 254 patients and 265 ITNs at Beijing Hospital between June 2019 and April 2022 (Additional Table
1). Patients were selected on the basis of the following criteria: (a) thyroid nodules are classified as TI-RADS 4 or 5; (b) thyroid nodules are larger than or equal to 1 cm, or less than 1 cm but highly suspected of malignancy based on the doctors’ experience; (c) diagnosis as Bethesda III or Bethesda IV or Bethesda V based on TBSRTC; (d) older than 18 years old and younger than 85 years old. Patients who had history of other cancers, who had previously received radiotherapy in the head and neck region, or had a family history of thyroid cancer were excluded from the study. When available, patient demographics, TI-RADS and Bethesda classifications, DNA molecular alterations, repeat FNAB, and histopathological diagnosis were collected.
This study was approved by the Research Ethics Board of the Beijing Hospital, National Health Commission and performed in accordance with the Declaration of Helsinki (IRB number in Ethical approval: 2021BJYYEC-044-03). All participants provided written informed consent before participation.
Serum CEA and serum calcitonin measurement
We collected data for serum CEA and serum calcitonin levels from the medical record system of Beijing Hospital. The normal values of CEA and calcitonin were defined as less than 5.0 ng/ml and 19 pg/ml, respectively.
Fine needle aspiration biopsy
Fine needle aspiration was performed under the guidance of ultrasound by experienced doctors at Beijing Hospital based on Chinese guidelines on the diagnosis and treatment of thyroid nodules and differentiated thyroid carcinomas. According to TBSRTC, the FNAB results were classified into six diagnostic categories and confirmed by two experienced pathologists. If the specimen meets the criteria, mutation testing was recommended. Gene mutations were detected on the first fine-needle aspiration samples by amplification refractory mutation system, including BRAF exon 15, HRAS exon 3, NRAS exon 2/3/4, KRAS exon 2/3/4, TERT promoter mutations (C228T and C250T).
Surgical procedure and histological examination
Total thyroidectomy or lobectomy was performed by one expert surgeon in Beijing Hospital depending on the clinical risk factors, ultrasound patterns, FNAB, and patients’ preferences. With the help of an ultrasonic diagram, the surgeon located the nodule and oriented the resected specimen for pathological diagnosis. The results of a blinded histopathological examination were used as the reference standard.
Statistical analysis
Statistical analysis was conducted using SPSS software, version 26. Student’s
t tests and Mann–Whitney
U-rank sum tests were used to analyze continuous variables. Chi-square test was used to analyze categorical variables. Based on established methods, sensitivity, specificity, and positive and negative predictive values were calculated [
33].
P values less than 0.05 were considered statistically significant.
Discussion
We provided a risk assessment method for patients with indeterminate cytology diagnoses based on the integration of blood test results, molecular signatures, and repeat cytological findings of thyroid nodules. It is critical to diagnose MTC at an early stage due to the relatively high incidence of metastasis [
20,
22]. The levels of serum CEA and calcitonin have been proven to be risk indicators for medullary thyroid cancer and are essential to guide appropriate surgical treatment [
34,
35]. Nowadays, screening for CEA and calcitonin has already become a routine test for high-risk thyroid nodules in tertiary care hospitals in China, and the cost is covered by the National Health Insurance. In our study, blood tests identified all 3 medullary thyroid cancers of 87 ITNs, which is consistent with previous research by others. Some studies have shown that a single method, such as molecular testing or repeat FNAB, can help determine the prevalence of malignancy in ITNs [
7,
36‐
46]. In recent years, two kinds of gene tests have appeared on the market, the Afirma Gene Expression Classifier and ThyroSeq genomic classifier. A randomized clinical trial including 346 patients with indeterminate nodules indicated that both products have high specificity and can help 49% of patients with ITNs avoid diagnostic surgery [
23]. In this case series, as these molecular tests mentioned above are not yet available in China and are expensive, the department of pathology detected the most common cancer-associated somatic mutations (
BRAF,
HRAS,
NRAS,
KRAS, and
TERT) in 115 of 265 specimens from ITNs. These 5 molecular alterations are frequently represented in malignant thyroid nodules, especially
BRAF. Twenty-two of 24 ITNs carrying the
BRAFV600E mutation were diagnosed as papillary thyroid cancer. As reported in the literature,
BRAF mutations have high specificity, and
RAS mutations have more limits for relatively low specificity [
47‐
51]. Our findings also showed that almost all detected mutations were
BRAF mutations, but future studies will be needed to verify the presence of these mutations in a larger sample size. Our data showed that molecular detection might be beneficial to thyroid nodules because evaluations of the genetic alterations of Bethesda III and Bethesda IV samples improved the specificity of a cancer diagnosis to 90.0%. Considering the possibility of false negatives in molecular tests, 13 patients with no molecular alterations underwent repeat FNAB during follow-up according to the suspected ultrasound characteristics and patients’ wishes, and the second biopsies were performed at least 1 month after the first ones (Additional Table
7). Our research also indicated that repeat aspiration is related to improved specificity in the diagnosis of ITNs with high-risk features.
A possible strength of our study is that we integrated the results of three of the most frequently used tests for ITNs into a surgical decision-oriented tool. Considering the Afirma Gene Expression Classifier and ThyroSeq genomic classifier are not yet available in China and are expensive, 5 of the most common cancer-associated somatic mutations (
BRAF, HRAS, NRAS, KRAS and
TERT) were offered in our hospital. Because the results of molecular testing could be false negatives, some patients underwent repeated biopsies during follow-up. We developed the risk assessment process based on the principle of minimizing costs and minimizing damage. The evaluation tool starts with blood test results (CEA and calcitonin), followed by molecular tests, and finally repeat FNAB results. Our study finally retrospectively collected 87 ITNs and the assessment method was validated on more than 10 kinds of histopathological types. All the ITNs were highly suspected of malignancy on ultrasound scans and cytologically diagnosed as Bethesda III or Bethesda IV or Bethesda V based on TBSRTC. By using the risk evaluation tool, we could accurately identify the patients with high-risk ITNs that might benefit from surgical treatment. The results showed that the specificity was 90.0% for Bethesda III lesions and 100% for Bethesda IV lesions, indicating that malignant ITNs with these cytologic findings can be correctly distinguished from benign ITNs. In addition, 73 out of 87 histological malignant ITNs were correctly distinguished, showing that the integrated assessment tool can help to improve the accuracy of the preoperative diagnosis of ITNs. To be more specific, 58 (84.1%) malignant nodules will be missed without using the risk tool, and 15 (83.3%) benign nodules will be submitted to surgery [
52,
53]. We compared the diagnostic performance of our risk assessment with different tests (Additional Table
10). It showed that this integrated tool was not inferior to expensive multigene products. The risk assessment is a beneficial and cost-effective method for clarifying the malignant or benign nature of ITNs, taking into account the side effect of extensive surgery in patients with metastatic thyroid cancer and the cost of active surveillance [
52‐
57].
However, the lack of some equivalent risk consideration is a weakness of this integrated assessment. Given that germline mutation plays an important role in the evaluation of thyroid nodules and is detected on peripheral blood samples, it could be added in the subsequent evaluation, especially for patients with a family history [
58,
59]. Besides, NPV and PPV depend on the prevalence rate of disease in the population, and our research showed that the prevalence rate of cancer in ITNs was as high as 70.1% in our 87 samples. For ITNs classified as Bethesda III or IV or V, the risk of malignancy was 78.1%, 51.5%, and 93.8%, respectively. One factor for the high prevalence rate of malignancy is that patients who chose surgery had malignant signs of malignancy and surgeons recommended surgery based on their experience. Of the 265 nodules initially collected, 127 did not choose immediate surgery. Twenty out of 104 nodules that had histological results carried at least one of the most common cancer-associated molecular alterations and 21 patients undergone repeat FNAB. Another explanation is that the rate of malignancy of ITNs in China may be underestimated. A study that recruited 140 samples in China showed that the malignant rate was 74.1% for resected ITNs [
60]. Another multicenter study in Israel included 810 patients and indicated that the malignancy rates were higher than those reported earlier. The authors suggested that doctors should use validated data for their own country in addition to published values [
61].
It should be observed that the NPV is not very high, and it is the most meaningful value to avoid surgery in patients with negative results. Although the risk-assessment tool could prevent 15 out of 18 unnecessary surgeries, it also incorrectly stopped 11 out of 69 surgeries of malignant nodules. Given that 7 false-negative nodules in this study were histologically diagnosed as papillary thyroid microcarcinoma and 3 were noninvasive follicular thyroid neoplasm with papillary-like nuclear features, we recommend that active surveillance can be adopted as a treatment option for the low-risk ITNs as there are characteristic features of indolence of these subtypes, and either tumor enlargement or the novel appearance of nodal or distant metastasis is the indication for surgery. Besides, multiple studies have shown that delayed surgery is safe [
62‐
64]. It is important to note that NPV changes accordantly with the risk of malignancy in different Bethesda classifications, which indicates that the risk assessment is more critical for ITNs with lower risk of malignancy. Moreover, the three nodules with false-positive results cannot be ignored. As one was a Bethesda III nodule with a
BRAFV600E mutation detected, the two others were Bethesda V nodules, and all of the nodules were less than 0.5 cm, we believe that it is possible that postoperative histological pathology missed the neoplastic foci. It is worth mentioning that the small sample size and single-center nature of this research limit the generalizability of the results; therefore, more prospective studies are needed to verify the efficacy of this risk assessment method.
It is also worth noting that one limitation of this study is the small sample size. In our study, 265 ITNs were included in the beginning, but the risk assessment was done on only 87 samples, of which only 47 (54%) underwent genetic testing. This finding suggests that this risk assessment tool needs to be evaluated prospectively in larger sample populations in future studies.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.