Pathological diagnosis of thyroid nodules based on core needle biopsies: comparative study between core needle biopsies and resected specimens in 578 cases

Thyroid nodule is a common disease of the endocrine system. The prevalence is 20 to 76% in the Chinese population as identified by high-resolution ultrasound, and 5 to 15% of nodules are malignant [1]. It is very important to screen these malignant cases for further treatment.

At present, the biopsy techniques for thyroid nodules used in practice include fine needle biopsy (FNB) and core needle biopsy (CNB). FNB aspirates cells for cytological examination with 22-27G needles, while CNB punctures tissues for histological examination with 18-21G needles.

FNA was applied to the preoperative diagnosis of thyroid nodules for the first time in 1960s and has been the standard diagnostic tool since the 1980s [2, 3]. CNB was first used in preoperative diagnosis of thyroid nodules in the 1990s. CNB has not been widely used for many years because of the pain, tolerability and complications associated with the operation [4]. However, with advances in CNB devices (i.e., spring-activated core needles) and the development of high-resolution ultrasound, these disadvantages have been significantly reduced. Recently, several large single-center studies have shown no significant differences between FNB and CNB in terms of pain, tolerability, or complications [5, 6]. In this case, the advantage of CNB for obtaining a large amount of tissue and providing more information on histological structures is highlighted. The National Cancer Institute, American Association of Clinical Endocrinologists/American College of Endocrinology/Associazione Medici Endocrinologi (AACE/ACE/AME), and the Korean Society of Thyroid Radiology (KSThR) have proposed CNB for thyroid nodules with previous nondiagnostic FNA results [7‐9]. Some publications even indicate that CNB may be a feasible and effective diagnostic tool for initially detected thyroid nodules [10‐12].

The application of CNB for preoperative diagnosis of thyroid nodules brings new challenges for pathologists. The lack of widely accepted pathological classification has led to inefficient communication between pathologists and clinicians and even misinterpretation of pathological results. Therefore, an effective pathological classification for thyroid nodules on CNB is urgently required by pathologists and clinicians. Of the very limited publications focusing on this issue, the majority are from Korea, the most important of which was the proposal of the Korean Endocrine Pathology Thyroid Core Needle Biopsy Study Group in 2015 [13]. The terminology and criteria for thyroid nodules on CNB specimen were defined for the first time in the proposal. This is a significant advancement, but the objectivity, accuracy and clinical value of this proposal should be tested in more centers and more case studies around the world to further improve and optimize the criteria based on worldwide data.

We retrieved 578 cases of thyroid nodules from Peking University First Hospital and classified them according to the Korean proposal. We tested the reliability of the Korean proposal and evaluated its clinical value by comparing classifications of a CNB sample to a resected specimen from the same nodule.

Pathological diagnosis of tumors with CNB specimens is unique. The reason is that, on the one hand, it is very different from the cytology observed with FNB and, therefore, either the diagnostic criteria or terminology should be different, and on the other hand, due to sampling limitations, it is difficult to apply the histological criteria established for resected specimens to CNB. In practice, the lack of uniform criteria and terminology often leads to deficiencies in the rigor and standardization of reports, consequently reducing the clinical value of CNB. Therefore, pathologists and clinicians urgently need a pathological classification that is not only based on the histological characteristics of CNB specimens but also objective, operable and highly valuable. This claim was first realized in the pathological diagnosis of lung cancer. The 2015 edition of WHO classification of tumours of the lung established the terminology and criteria for non-small cell carcinoma on CNB independently of resected specimens for the first time [16]. This system has been widely welcomed by pathologists and clinicians in the last 3 years. The successful experience has prompted the international pathology community to understand the specificity of the diagnosis for CNB specimens. After lung cancer, various studies have been performed in other subspecialties.

Unlike the preoperative diagnosis of lung nodules, the preoperative diagnosis of thyroid nodules around the world is still mainly dependent on FNB, although many publications have confirmed the safety of CNB for thyroid nodules [6]. In China, CNB is applied in the diagnosis of thyroid nodules in only a few medical centers. Therefore, research on the pathological diagnosis of thyroid nodules by CNB is relatively insufficient. There are very limited publications and the majority are from Korea, the most important of which is the proposal of the Korean Endocrine Pathology Thyroid Core Needle Biopsy Study Group 2015 [13]. The terminology and criteria for CNB thyroid nodule specimen were standardized for the first time in this proposal. This is a significant study, and the objectivity, accuracy and clinical value of the findings should be tested in more centers and more case studies around the world to improve and optimize the criteria based on a larger amount of data.

In our study, of 578 CNB samples, only 1 (0.17%) was classified as unsatisfactory, and 31 (5.36%) were classified as indeterminate. Compared to the unsatisfactory rate (5–17%) and indeterminate rate (10–30%) of FNB published in papers [17‐19], the rate of CNB is much lower. There were 28 cases classified as indeterminate based on CNB that were classified as malignant in resected specimens. In all of these cases, the histological pattern of the CNB specimens was entirely follicle, while that of resected specimens was a mixture of follicle, tubule and papillae. This demonstrates that sampling limitation is the main reason why the cases classified as malignant on resected specimens were classified as indeterminate based on CNB. Nevertheless, the atypia of the histological pattern and cell morphology of the CNB specimens highly suggested that the cases were malignant. There were 26 cases classified as suspicious for malignancy based on CNB that were classified as malignant in the resected specimens. Of these cases, a very few fragments of papillae were found in CNB specimens in 10 cases, and several microfollicles composed of cells with nuclear features of papillary thyroid carcinoma were present in 16 CNB specimens. Although, the components with the histological characteristics of papillary thyroid carcinoma did exist in CNB in these cases, they were classified as suspicious for malignancy only because the quantity was insufficient to make a definitive diagnosis. Therefore, sampling limitation is also the main reason that cases classified as malignant on resected specimens were classified as suspicious for malignancy on CNB.

By two-classification method A, the consistency between CNB classification and resected specimen classification is moderate (κ = 0.546) and CNB shows 90.39% sensitivity, 100.00% specificity, 91.00% accuracy, 100.00% PPV and 41.57% NPV for the preoperative evaluation of the malignancy of thyroid nodules. Therefore, we recommend that cases classified as suspicious for malignancy and malignant based on CNB be treated as malignant. By two-classification method B, the consistency between classification on CNB and resected specimens is substantial (κ = 0.737), and CNB shows 95.93% sensitivity, 97.30% specificity, 62.07% accuracy, 99.81% PPV and 62.07% NPV for the preoperative evaluation of the malignancy of thyroid nodules. Compared to method A, although 0.19% of the patients were at risk of over treatment, method B reduces the rate of missed diagnosis of malignancy by 20.50% and increases the accuracy of overall evaluation of malignancy by 5.02%. Thus, we think it may be the most beneficial strategy for cases classified as indeterminate, suspicious for malignancy and malignant based on CNB to be treated as malignant. Of the 40 cases classified as benign on CNB, 19 were classified as benign, 20 were classified as malignant and 1 was classified as suspicious for malignancy on resected specimens. A study of the morphology of the inconsistent cases shows that although the malignant components are obvious in resected specimens, they are absent in CNB. Failure to obtain tissue from the lesion is the only reason for the discrepancy. Therefore, cases classified as benign by CNB but considered malignant or suspicious for malignancy by ultrasound should be followed up closely and require rebiopsy.

FNB is incapable of differentiating follicular neoplasm from nodular hyperplasia, while CNB is very useful in making this distinction. Of 17 cases classified as follicular neoplasm with the resected specimens, 16 were classified as follicular neoplasm and 1 was classified as indeterminate by CNB. Therefore, cases classified as follicular neoplasm by CNB should be treated with surgical operation. The capsular and vascular infiltration should be detected microscopically to further clarify whether the lesion is benign or malignant.

CNB is a safe and effective method for the preoperative diagnosis of thyroid nodules. The proposal of the Korean Endocrine Pathology Thyroid Core Needle Biopsy Study Group suggestion 2015 is objective, operable and highly valuable for pathological classification of thyroid nodules on CNB. Under this framework, the inaccuracy and limitations of biopsy is the main cause of inconsistency between diagnosis based on CNB and resected specimens. The most beneficial strategy for cases classified as indeterminate, suspicious for malignancy and malignant on CNB may be to treat them as malignant. Cases classified as benign by CNB but considered malignant or suspicious for malignancy by ultrasound should be followed up closely and require rebiopsy. Cases classified as follicular neoplasm by CNB should be treated with surgical operation, and the capsular and vascular infiltration should be detected microscopically to further clarify whether the lesion is benign or malignant.