Endocrine Surgery Review: In Depth Review of Landmark Articles

The Clinical Utility of Molecular Testing in the Management of Thyroid Follicular Neoplasms (Bethesda IV Nodules)

Carty SE, Ohori NP, Hilko DA, McCoy KL, French EK, Manroa P, Morariu E, Sridharan S, Seethala RR, Yip L
Annals of Surgery. 2020; 272: 621-627
Reviewer: Catherine McManus, MD, Columbia University Medical Center

In Brief

Approximately 10-40% of thyroid nodules that undergo fine needle aspiration biopsy (FNAB) have indeterminate cytology.  The 2015 American Thyroid Association (ATA) Guidelines recommend consideration of molecular testing (MT) for indeterminate nodules, specifically those with Bethesda III and IV cytology, in order to preoperatively identify nodules that are more likely to be cancerous and avoid unnecessary surgery on indolent lesions (1).  The use of MT for indeterminate thyroid nodules has evolved significantly since the initial seven-gene panel developed in 2011 (2).  While studies have demonstrated the performance characteristics and external validity of commercially available molecular tests, the questions of whether MT appropriately guides clinical management and how MT can be incorporated when accounting for other clinical and patient related factors still remained largely unaddressed (3-5). 

In this clinical utility study, Carty and colleagues described their experience with MT of Bethesda IV thyroid nodules, suspicious for follicular neoplasms (FN) at a high-volume, multidisciplinary thyroid center (6).  Following current ATA guidelines, the authors aimed to determine whether performing MT for nodules with FN cytology would lead to surgical management for histologic thyroid cancer and would appropriately assign nodules that were likely benign to surveillance.

The authors retrospectively examined 389 patients with 405 thyroid nodules with FN cytology between 2014 and 2019.  Of note, oncocytic (Hurthle cell) neoplasms were excluded.  When histology was available, nodules were classified as thyroid cancer, noninvasive follicular thyroid neoplasm with papillary-like nuclear features (NIFTP, diagnosed after 1/2017), or benign.  Molecular testing, specifically Thyroseq version 2 (or version 3 after 11/2017), was routinely considered for all FN nodules and was positive if the identified mutation(s) indicated a >10% risk of thyroid cancer.  Patients were offered surgery if they had positive MT, symptoms related to the nodule, nodules >4cm, had hyperthyroidism, or concurrent hyperparathyroidism.

Of the 405 FNs, 359 (89%) underwent MT.  A total of 139 (39%) nodules were positive and 215 (53%) nodules were negative.  The majority (81%) of positive nodules contained a RAS-like mutation while only 10 (7.1%) contained a BRAF-like mutation.  Surgery was significantly more common among patients with positive MT (n=126, 91%) compared to nodules with negative MT (n=58, 27%), p<0.001.  Among MT positive nodules, 88 (70%) had histologic thyroid cancer (TC) whereas 9 (16%) MT negative nodules had TC (p <0.001).  Of note, all 9 MT negative nodules were tested with Thyroseq v2 (rather than v3) and all were histologically low risk thyroid cancer.

A total of 218 (54%) of the 405 nodules with FN underwent surgery.  Histology revealed 14 nodules with NIFTP and 109 thyroid cancers (85 papillary, 9 follicular, 6 Hurthle cell, 3 medullary and 6 poorly differentiated).  Importantly, MT was positive in 8 of 9 high-risk nodules and 13 of 16 intermediate risk nodules.  Additionally, the 3 medullary thyroid cancers and 6 Hurthle cell carcinomas were identified preoperatively with positive MT.  A total of 3 patients who had positive MT ultimately had benign histology upon total thyroidectomy.  Among the 14 nodules with NIFTP histology, 12 had positive MT and 2 had negative MT.   Of the 215 nodules with negative MT, 113 had follow up data and 92 (82%) were stable over a mean of 24.6 years.  However, 61 nodules (35%) of patients with negative MT assigned to surveillance were lost to follow up.

Critique and Future Directions

This study provides a long awaited illustration of how MT is implemented in a clinical setting.  When interpreting the results and outcomes of this study, it is important to recognize that the aim was to evaluate whether the use of MT appropriately led to surgery for nodules with thyroid cancer and surveillance for nodules that were more likely benign.  In this type of study, variability and different trends in the management for patients with thyroid nodules is expected based on updates in current guidelines, improvements in commercially available MT, and adaptations of the guidelines among clinicians at different rates.

There were several important clinical shifts that took place during the period of this study including the update of the 2015 ATA guidelines that opted for a less aggressive surgical approach to small thyroid cancers, the reclassification of NIFTP to a benign lesion, and the improvement in the sensitivity of Thyroseq v3.  The authors note that after 11/2017 when Thyroseq v3 was implemented, the percentage of nodules that underwent MT increased from 86% to 96% (p=0.0002) but surgery was less common (44% compared to 59% with v2, p=0.007), a finding that likely reflects the increased comfort of clinicians to observe MT negative nodules.

The authors conclude that the use of MT for FN nodules identified aggressive malignancies and appropriately recommended surveillance for likely benign nodules.  This experience can be used to counsel patients with FN who are deciding about MT and/or surgical management.  However, clinicians must keep in mind that this information does not apply to patients with Hurthle cell lesions or nodules with a Bethesda III or V cytology result.  Furthermore, although it appears that the majority of nodules for which surveillance was recommended remained stable over time, up to 1/3 of the patients were lost to follow up.  While we consider stability on ultrasound to be more characteristic of a benign nodule, we are cannot say with certainty that these nodules are true negatives without histology.   Additionally, the authors recognize the possibility that patients and referring physicians may have incorrectly assumed no further monitoring was necessary, which raises the importance of clear communication about the surveillance recommendations with the patient and referring physician.  

The authors recognize that although they found a four-fold increase in the yield of malignancy with surgical management using MT, this study was not powered to determine whether the use of MT in FN nodules altered the extent of the initial operation.  Consequently, surgeons cannot apply this data when considering whether to proceed with a diagnostic lobectomy versus a total thyroidectomy upfront.  Further studies are needed to evaluate whether MT can adequately and appropriately guide initial surgical management.  Finally, there remains a need for clinical utility studies of MT for other indeterminate cytologies including atypia of undetermined significance, follicular lesion of undermined significance (Bethesda III) and Hurthle cell lesions.


  1. Haugen BR, Alexander EK, Bible KC, et al. 2015 American Thyroid Association Management Guidelines for Adult Patients with Thyroid Nodules and Differentiated Thyroid Cancer. Thyroid. 2016; 26: 1-133.
  2. Nikiforov YE, Ohori P, Hodak S, et al. Impact of Mutational Testing on the Diagnosis and Management of Patients with Cytologically Indeterminate Thyroid Nodules: A Prospective Analysis of 1056 FNA Samples. J Clin Endocrinol Metab 2011; 96, 3390-7.
  3. Patel K, Angell TE, Babiarz J, et al.  Performance of a Genomic Sequencing Classifier for the Preoperative Diagnosis of Cytologically Indeterminate Thyroid Nodules. JAMA Surgery. 2018; 53, 817-24.
  4. Nikiforova MN, Mercurio S, Wald AI, et al. Analytical Performance of the ThyroSeq v3 Genomic Classifier for Cancer Diagnosis in Thyroid Nodules. Cancer2018; 1682-1690.
  5. Steward DL, Carty SE, Sippel RS, et al. Performance of a Multigene Genomic Classifier in Thyroid Nodules With Indeterminate Cytology: A Prospective Blinded Multicenter Study. JAMA Oncol. 2019; 5, 204-12.
  6. Carty SE, Ohri P, Hilko DA, et al.  The Clinical Utility of Molecular Testing in the Management of Thyroid Follicular Neoplasms (Bethesda IV Nodules). Annals of Surgery. 2020; 272(4): 621-627.


MiRNA let-7 from TPO (+) Extracellular Vesicles is a Potential Marker for a Differential Diagnosis of Follicular Thyroid Nodules

Lidia Zabegina, Sergey Titov, Anastasia Malek
N.N. Petrove National Medical Research Center of Oncology, St. Petersburg, Russia
Cells, August 2020
Reviewer: Muthusamy Kunnimalaiyaan, PhD, Department of Endocrine Neoplasia and Hormonal Disorders, MD Anderson Cancer Center, Houston, TX

In Brief:

The incidences of thyroid cancers (TC) are steadily increasing worldwide.  Among the TC, follicular thyroid cancer (FTC) is the second most common type with high mortality rates. The main clinical concern is to differentiate between FTC and benign follicular adenomas (FA) for treatment strategies at the time of diagnosis. Recently the development of noninvasive diagnostic tools for cancer analysis is steadily increasing, and this is the focal point of this paper for thyroid cancer.  In addition to the BRAF mutation as a marker for thyroid cancer, there are reports about other genetic mutations and miRNA expression profiles in tissue biopsy.  Fortunately, these types can be analyzed in the circulating plasma components (liquid biopsy), and extensive research has recently been focused on these aspects.  This group, earlier, reported exosome miRNA’s diagnostic potency (miR-31, miR-21, and miR-181a) for discrimination between benign FA, FTC, and papillary thyroid cancer (PTC).  In this study, they want to provide proof of principle 1.  Thyroid peroxidase (TPO) (+) extracellular vesicles (EV) will be reduced in FA patients after thyroidectomy, and 2. Development of follicular nodules in the thyroid gland is associated with enlargement of the gland’s size, which may increase/accommodate the number of TPO (+) EVs in circulating plasma.

In this study, Zabegina et al. reported the correlation between the differential amount of extracellular vesicles (EV) in FTC and FA. They have included in this study 5 healthy donors, 30 FA, and 30 FTC patients. They have purified from the thyroid-specific EV by immunoprecipitation using thyroid peroxidase (TPO) antibody, and this TPO (+) EVs were subjected to 85 cancer-associated miRNAs, identified let-7 family miRNA as a potential marker, and the results were statistically analyzed.  Overall, they have systematically carried out the experiments, as shown in Figure1.


Although this is a well-planned study, there are a couple of things that are not explained well; 1. Limitation of only 85 cancer-associated genes, including let-7 was analyzed, 2. Pooling 10 different patients from each group (FTC and FA) miRNA is not explained clearly, 3. In the experimental protocol they mentioned isolation of proteins from these EVs but no results were shown, 4. It is essential to add a combination of at least one additional thyroid-specific marker during the purification of thyroid-specific EVs.

Future directions

This is well-executed study and provides the feasibility of utilizing thyroid-specific EVs with their content that could be developed in the future as a potential diagnostic biomarker and/or therapeutic marker for the follicular thyroid nodules as well as other types.


Enhancer Signatures Stratify and Predict Outcomes of Non-Functional Pancreatic Neuroendocrine Tumors

Paloma Cejas, Yotam Drier, Koen M. A. Dreijerink, Lodewijk A. A. Brosens, Vikram Deshpande, Charles B. Epstein, Elfi B. Conemans, Folkert H. M. Morsink, Mindy K. Graham, Gerlof D. Valk, Menno R. Vriens, Carlos Fernandez-del Castillo, Cristina R. Ferrone, Tomer Adar, Michaela Bowden, Holly J. Whitton, Annacarolina Da Silva, Alba Font-Tello, Henry W. Long, Elizabeth Gaskell, Noam Shoresh, Christopher M. Heaphy, Ewa Sicinska, Matthew H. Kulke, Daniel C. Chung, Bradley E. Bernstein and Ramesh A. Shivdasani
Nature Medicine. August 2019; 25:1260-1265.
Reviewer: Sarina Bains, MD, Weill Cornell Medicine

In Brief

Nonfunctioning pancreatic neuroendocrine tumors (NF-PNETs) are rare pancreatic tumors of neuroendocrine origin that do not produce an excess of hormones and oftentimes are detected incidentally.  NF-PNETs have a wide spectrum of clinical behaviors ranging from indolent and localized to aggressive with distant metastases, most commonly to the liver. Tumor size has served as a strong prognostic factor in predicting malignancy, where tumors <2cm have excellent prognosis with survival rates up to 100% at 15 years.1,2 Thus, current guidelines recommend observation for asymptomatic patients with tumors <1cm and either surgery versus observation for tumors between 1 to 2 cm; tumors >2cm are recommended for resection.The World Health Organization (WHO) classification distinguishes PNETs as well-differentiated or poorly-differentiated tumors histopathologically. Well-differentiated tumors are further divided into Grade 1, 2 or 3 based on Ki-67 proliferation indices of <3%, 3 to 20%, and >20%, respectively; all poorly-differentiated PNETs are considered Grade 3.These factors are most reliably determined after surgery and are strong predictors of overall survival and disease-free survival.5  Recent genomics studies have revealed that PNETs are commonly driven by mutations in tumor suppressor genes (MEN1, ATRX, DAXX), with the latter two being associated with alternative lengthening of telomeres (ALT) and recurrence.  However, there is currently a paucity of clinically-available definitive biomarkers to help guide peri-operative management of these tumors.

The primary purpose of this study published by Cejas and colleagues was to identify novel epigenetic markers of NF-PNET cellular origin, and correlate these findings with clinical outcomes.  They used enhancer mapping to identify transcription factors (TFs) at active chromatin signatures to elucidate lineage-specific gene regulatory programs, specifically searching for markers of alpha or beta islet cell origin.  The authors found that most PNETs fall into two major epigenomic and transcriptomic subtypes based on H3K27ac and H3K4me2 chromatin signatures. Specifically, chromatin immunoprecipitation sequencing (ChIP-seq)-derived profiling of H3K27ac-associated enhancers was able to identify three distinct subtypes of PNETs. Across 8 non-functional PNETs, Type A was characterized by a two-fold higher H3K27ac signal in 288 enhancers, whereas Type B was characterized by increased H3K27ac signal at 104 unique regions.  Analyzing these data from a cell-lineage perspective revealed that Type A tumors expressed strong H3K27ac signals at a-cell-specific loci (ARX and IRX2) and Type B had strong signals at b-cell TFs (PDX1); type C had variable signal and mRNA expression at these loci.  ChIP-seq profiling for H3K4me2 in this cohort yielded similar results.  RNA-sequencing of these PNETs also confirmed differential ARX and PDX1 expression between A- and B- subtypes.

As shown in prior mouse models, ARX is necessary to drive a-cell fate while PDX1 is a maker of b-cell differentiation. This was confirmed in this study where the authors compared ChIP-Seq and RNA-seq profiles between PNET subtypes and normal pancreatic islets – this analysis showed that Type A and Type B tumors were highly enriched for a-cell-specific and b-cell-specific markers, respectively. These results were validated in an additional cohort of 13 PNETs.  Interestingly, this differential transcriptome profiling was more accurate in distinguishing PNET subtypes than glucagon and insulin mRNA levels.

An additional Dutch cohort was then evaluated by tissue microarray immunohistochemistry (IHC) consisting of 77 PNETs, of which 61 had a germline MEN1 mutation and 13 were insulinomas.  There were 34 (31%) Type A / ARX+, 31 (37%) Type B / PDX1+, 6 (8%) double negative, and 6 (8%) double positive tumors.  These findings were notable for variable expression of ARX in Type A tumors, whereas PDX1 was strong in all Type B tumors and insulinomas.  Selected tumors underwent ChIP-Seq and RNA-seq, which revealed concordant results as compared to IHC.  These data suggest that the authors have identified a unique marker that can differentiate tumor subtypes on conventional IHC.

From a translational perspective, ARX and PDX1 status was correlated to clinical outcomes in 61 NF-PNETs.  In this cohort, there were no pathologic differences between ARX+ and PDX1+ tumors, such as tumor size or grade.  However, at a median of 24 months follow-up, relapses (all occurring in the liver) were almost exclusively associated with ARX+ or double negative groups.  Notably, no PDX1+ or double positive tumors recurred.  These findings were confirmed in an additional cohort of NF-PNETs from Massachusetts (n=56).  With a longer median follow-up of 66 months, most relapses again occurred in the ARX+ or double negative tumors; instead, PDX1+ and double positive tumors were more indolent, accounting for three total relapses. 

While these data suggest that PDX1+ expression is correlated with a more favorable prognosis, the authors sought to analyze the prognostic impact of ALT status given its association with ATRX and DAXX mutations and known risk of recurrence.  Telomere-specific FISH analysis revealed that ALT positivity was more commonly found in ARX+ and double negative tumors (N=13/27, 48%) as compared to PDX1+ tumors (N=5/35, 14%).  As expected, ALT was associated with disease recurrence, however, these results were more striking when combined with PNET subtypes where relapses occurred in all ARX+/ALT+ tumors, only 9% of ARX+/ALT- tumors, and one PDX1+/ALT- tumor.  A final multi-variable analysis of a combined cohort including 83 NF-PNETs with 15 relapses demonstrated that only ALT positivity and absence of PDX1 expression both independently predicted relapse, while controlling for tumor size and WHO grade. 

In summary, this study concluded that enhancer profiles of non-functional PNETs subtypes uniquely resemble islet a and b-cell profiles. Identification of several of these markers, namely ARX and PDX1, can be performed readily on IHC and subsequently used as prognostic markers for disease recurrence.  Specifically, ARX+ tumors are at risk for relapse especially in the setting of ALT positivity, while PDX1 tumors appear more indolent.  In this study, these markers were better predictors than tumor size and WHO grade, thus, adaptation into clinical practice could help guide management of NF-PNET, particularly for tumors <2-3cm.

Critique and Future Directions

ARX and PDX1 expression identified in NF-PNETs represent a novel clinically-significant prognostic marker that can be potentially tested preoperatively to guide surveillance versus surgery decision-making, as well as perhaps alter surveillance strategies post-operatively. Given the significant difference in clinical course based on ARX and PDX1 expression, the reliability of signal detection is important – ARX strength on IHC was shown to be variable, thus there may be overlap between a staining failure and true ARX+ tumors with low expression. Further investigation of the clinical implications between potential staining failures in double negative samples versus weak signal strength will be important moving forward. Additionally, the double positive (ARX+ PDX1+) tumors need to be further investigated regarding the possibility of a dual cell lineage, and the resultant oncogenic implications. Evaluating how varying signal strengths correlate with prognosis may shed light on a tumor’s state of differentiation and ability to metastasize. In the Dutch cohort, the double positive group correlated with a good prognosis, a finding that was not clearly confirmed in the multivariable analysis of the combined cohort.  Mechanistic studies are warranted to determine how exactly ARX positivity is driving metastatic potential independent of ALT positivity, and why PDX1 positivity appears to protect against it.

A clinical limitation of this study is the relatively small number of relapses observed, which limits a more comprehensive multivariable analysis.  Future studies with larger cohorts of relapsed patients can also investigate associations between ARX and PDX1 positivity with response to adjuvant therapy (everolimus, lutetium Lu 177 dotatate, etc.)  Lastly, from a pre-operative management perspective, this study’s tumor analyses were performed on resected PNET specimens. It was not described whether this stratification can be reliably performed on preoperative FNA cytology. While one follow up study with a small sample size correlated ARX and PDX1 expression between cytologic and pathologic samples, an investigation in a larger cohort is needed to more accurately determine the reliability in expression from samples obtained by endoscopic FNA specimen.6   Overall, this research identifies novel preoperative prognostic markers to predict aggressive behavior in NF-PNETs with the potential benefit of guiding treatment algorithms.


  1. Falconi M. et al. ENETS Consensus Guidelines Update for the Management of Patients with Functional Pancreatic Neuroendocrine Tumors and Non-Functional Pancreatic Neuroendocrine Tumors. Neuroendocrinology 2016;103(2):153-171.
  2. Bettini R. et al. Tumor size correlates with malignancy in non-functioning pancreatic endocrine tumor. Surgery 2011;150:75-82.
  3. Howe, J., et al. The North American Neuroendocrine Tumor Society Consensus Paper on the Surgical Management of Pancreatic Neuroendocrine Tumors. Pancreas 2020;49:1-33.
  4. Lloyd RV, Osamura RY, Klöppel G, Rosai J. WHO Classification of Tumours of Endocrine Organs. 4th ed. Lyon: International Agency for Research on Cancer; 2017:355.
  5. Bu, J. et. Al. Prognostic factors of non-functioning pancreatic neuroendocrine tumor revisited: The value of WHO 2010 classification. Ann Hepatobiliary Pancreat Surg 2018:22:66-74
  6. Hackeng W, et al. Assessment of ARX expression, a novel biomarker for metastatic risk in pancreatic neuroendocrine tumors, in endoscopic ultrasound fine-needle aspiration. Diagnostic Cytopathology 2019;1-8