Identification of Hürthle cell cancers: solving a clinical challenge with genomic sequencing and a trio of machine learning algorithms

Diagnostic performance of molecular testing in indeterminate (Bethesda III and IV) thyroid nodules with Hürthle cell cytology

BACKGROUND

Indeterminate thyroid nodules with Hürthle cell cytology remain a diagnostic challenge. The low benign call rate and positive predictive value of first-generation molecular tests precluded their use to rule out malignancy. We examined the diagnostic performance of current tests.

METHOD

This subset analysis of our prospective randomized trial compared the benign call rate and positive predictive value of Afirma Gene Sequencing Classifier and Thyroseq v3 in Bethesda III and IV nodules with Hürthle cell cytology. Molecular test samples were obtained at initial fine-needle aspiration (8/2017-7/2022) and reflexively sent for processing.

RESULTS

Molecular testing was performed on 140 Hürthle cell nodules. Of 79 nodules tested with the Afirma Gene Sequencing Classifier, the benign call rate was 84% (66/79). Nine of 66 nodules with benign results were resected, with no malignancies. Twelve of 13 nodules with suspicious results were resected, revealing 3 malignancies-2 papillary thyroid carcinomas and one Hürthle cell carcinoma (positive predictive value 25%). Of 61 nodules tested with Thyroseq v3, the benign call rate was 56% (34/61; (P < .01 versus Afirma Gene Sequencing Classifier). Five of 34 nodules with negative results were resected, with no malignancies. Nineteen of 27 nodules with positive results were resected, revealing 3 malignancies-2 papillary thyroid carcinomas and 1 Hürthle cell carcinoma (positive predictive value 16%).

CONCLUSION

The high benign call rate of current molecular tests in Hürthle cell nodules strengthens their value in enabling patients to avoid surgery.

Molecular testing for indeterminate thyroid nodules: past, present, and future

PURPOSE OF REVIEW

To examine the origin, current progress, and future directions of molecular testing in indeterminate Bethesda III and Bethesda IV thyroid nodules.

RECENT FINDINGS

The diagnostic performance of current genomic tests shows improved benign call rates, specificity and positive-predictive values over prior test versions. The choice of test platform for clinical use should consider test performance, institutional rate of malignancy, nodule cytology and the potential for prognostication to help guide decision-making. Current challenges include test reliability, defining the optimal duration of surveillance, and improving test performance in challenging cytology, such as oncocytic nodules and NIFTP. Opportunities also remain to optimize cost-effectiveness across multiple clinical and practice settings and to refine the use of molecular testing for dynamic risk stratification, such as with BRAF V600E mutation testing.

SUMMARY

Molecular testing of indeterminate thyroid nodules has helped to reduce the burden of diagnostic surgery, associated healthcare costs, and potential complications. Current-generation tests have demonstrated improvement in diagnostic performance, but challenges remain in improving test performance and refining the scope of testing in care. Decision-making for the management of indeterminate thyroid nodules should consider cytology, clinical and sonographic features, patient values and preferences and molecular testing results, whenever available.

Update on Molecular Diagnostics in Thyroid Pathology: A Review

Thyroid nodules are quite common, and the determination of a nodule of concern is complex, involving serum testing, radiology and, in some cases, pathological evaluation. For those nodules that raise clinical concern of neoplasia, fine needle aspiration biopsy is the gold standard for evaluation; however, in up to 30% of cases, results are indeterminate for malignancy, and further testing is needed. Advances in molecular testing have shown it to be of benefit for both diagnostic and prognostic purposes, and its use has become an integral part of thyroid cancer management in the United States and in several global nations. After The Cancer Genome Atlas (TCGA) consortium published its molecular landscape of papillary thyroid carcinoma (PTC) and reduced the "black matter" in PTC from 25% to 3.5%, further work ensued to clarify the remaining fraction not neatly attributed to the BRAF^V600E-like or RAS-like phenotypes of the TCGA. Over the past decade, commercial molecular platforms have been refined as data accrues, and they increasingly cover most genetic variants of thyroid carcinomas. Molecular reporting focuses on the nodule tested, including related clinical information for that nodule (size of nodule, Bethesda category, etc.). This results in a comprehensive report to physicians that may also include patient-directed, clear language that facilitates conversations about nodule management. In cases of advanced or recurrent disease, molecular testing may become essential for devising an individual therapeutic plan. In this review, we focus on the evolution of integrated molecular testing in thyroid nodules, and how our understanding of tumor genetics, combined with histopathology, is driving the next generation of rational patient management, particularly in the context of emerging small, targetable therapeutics.

Molecular diagnostics in the evaluation of thyroid nodules: Current use and prospective opportunities

Thyroid cancer is the most common endocrine malignancy with an estimated 43,800 new cases to be diagnosed in 2022 and representing the 7th most common cancer in women. While thyroid nodules are very common, being identified in over 60% of randomly selected adults, only 5-15% of thyroid nodules harbor thyroid malignancy. Therefore, it is incumbent upon physicians to detect and treat thyroid malignancies as is clinically appropriate and avoid unnecessary invasive procedures in patients with benign asymptomatic lesions. Over the last 15-20 years, rapid advances have been made in cytomolecular testing to aid in thyroid nodule management. Initially, indeterminate thyroid nodules, those with Bethesda III or IV cytology and approximately a 10-40% risk of malignancy, were studied to assess benignity or malignancy. More recently, next generation sequencing and micro-RNA technology platforms have refined the diagnostic capacity of thyroid nodule molecular testing and have introduced opportunities to glean prognostic information from both cytologically indeterminate and malignant thyroid nodules. Therefore, clinicians can move beyond determination of malignancy, and utilize contemporary molecular information to aid in decisions such as extent of surgery and post-therapy monitoring plans. Future opportunities include molecularly derived information about tumor behavior, neo-adjuvant treatment opportunities and response to thyroid cancer therapies.

Maximizing Small Biopsy Patient Samples: Unified RNA-Seq Platform Assessment of over 120,000 Patient Biopsies

Despite its wide-ranging benefits, whole-transcriptome or RNA exome profiling is challenging to implement in a clinical diagnostic setting. The Unified Assay is a comprehensive workflow wherein exome-enriched RNA-sequencing (RNA-Seq) assays are performed on clinical samples and analyzed by a series of advanced machine learning-based classifiers. Gene expression signatures and rare and/or novel genomic events, including fusions, mitochondrial variants, and loss of heterozygosity were assessed using RNA-Seq data generated from 120,313 clinical samples across three clinical indications (thyroid cancer, lung cancer, and interstitial lung disease). Since its implementation, the data derived from the Unified Assay have allowed significantly more patients to avoid unnecessary diagnostic surgery and have played an important role in guiding follow-up decisions regarding treatment. Collectively, data from the Unified Assay show the utility of RNA-Seq and RNA expression signatures in the clinical laboratory, and their importance to the future of precision medicine.

A primer on machine learning techniques for genomic applications

High throughput sequencing technologies have enabled the study of complex biological aspects at single nucleotide resolution, opening the big data era. The analysis of large volumes of heterogeneous "omic" data, however, requires novel and efficient computational algorithms based on the paradigm of Artificial Intelligence. In the present review, we introduce and describe the most common machine learning methodologies, and lately deep learning, applied to a variety of genomics tasks, trying to emphasize capabilities, strengths and limitations through a simple and intuitive language. We highlight the power of the machine learning approach in handling big data by means of a real life example, and underline how described methods could be relevant in all cases in which large amounts of multimodal genomic data are available.

Combined molecular and mathematical analysis of long noncoding RNAs expression in fine needle aspiration biopsies as novel tool for early diagnosis of thyroid cancer

PURPOSE

In presence of indeterminate lesions by fine needle aspiration (FNA), thyroid cancer cannot always be easily diagnosed by conventional cytology. As a consequence, unnecessary removal of thyroid gland is performed in patients without cancer based on the lack of optimized diagnostic criteria. Aim of this study is identifying a molecular profile based on long noncoding RNAs (lncRNAs) expression capable to discriminate between benign and malignant nodules.

METHODS

Patients were subjected to surgery (n = 19) for cytologic suspicious thyroid nodules or to FNA biopsy (n = 135) for thyroid nodules suspicious at ultrasound. Three thyroid-specific genes (TG, TPO, and NIS), six cancer-associated lncRNAs (MALAT1, NEAT1, HOTAIR, H19, PVT1, MEG3), and two housekeeping genes (GAPDH and P0) were analyzed using Droplet Digital PCR (ddPCR).

RESULTS

Based on higher co-expression in malignant (n = 11) but not in benign (n = 8) nodules after surgery, MALAT1, PVT1 and HOTAIR were selected as putative cancer biomarkers to analyze 135 FNA samples. Cytological and histopathological data from a subset of FNA patients (n = 34) were used to define a predictive algorithm based on a Naïve Bayes classifier using co-expression of MALAT1, PVT1, HOTAIR, and cytological class. This classifier exhibited a significant separation capability between malignant and benign nodules (P < 0.0001) as well as both rule in and rule out test potential with an accuracy of 94.12% and a negative predictive value (NPV) of 100% and a positive predictive value (PPV) of 91.67%.

CONCLUSIONS

ddPCR analysis of selected lncRNAs in FNA biopsies appears a suitable molecular tool with the potential of improving diagnostic accuracy.

Molecular Tests for Risk-Stratifying Cytologically Indeterminate Thyroid Nodules: An Overview of Commercially Available Testing Platforms in the United States

The past decade has witnessed significant advances in the application of molecular diagnostics for the pre-operative risk-stratification of cytologically indeterminate thyroid nodules. The tests that are currently marketed in the United States for this purpose combine aspects of tumor genotyping with gene and/or microRNA expression profiling. This review compares the general methodology and clinical validation studies for the three tests currently offered in the United States: ThyroSeq v3, Afirma GSC and Xpression Atlas, and ThyGeNEXT/ThyraMIR.

Molecular analysis of fine-needle aspiration cytology in thyroid disease: where are we?

PURPOSE OF REVIEW

The prevalence of thyroid nodules in the general population is high but only about 5% are malignant lesions. Cytology is usually appropriate to rule out malignancy in sonographically suspicious nodules but in many cases, reports are indeterminate. Molecular testing is a more recent approach to rule out malignancy and guide subsequent management.

RECENT FINDINGS

Although several different molecular testing approaches have proven useful in reducing unnecessary surgery, there are still several remaining issues, such as the possible occurrence of RAS mutations (which are difficult to interpret in clinical management) and the role of molecular analysis in specific histotypes, such as Hürthle cell carcinomas. Furthermore, conclusive evidence is lacking regarding the cost-effectiveness and appropriateness of surgical options following molecular tests.

SUMMARY

To be useful in clinical practice, molecular tests should be applied to appropriate candidates. In truly uncertain thyroid nodules in which diagnostic surgery may be considered, molecular testing may change the clinical approach and 'save' a number of thyroids.

Genetics, Diagnosis, and Management of Hürthle Cell Thyroid Neoplasms

Hürthle cell lesions have been a diagnostic conundrum in pathology since they were first recognized over a century ago. Controversy as to the name of the cell, the origin of the cell, and even which cells in particular may be designated as such still challenge pathologists and confound those treating patients with a diagnosis of "Hürthle cell" anything within the diagnosis, especially if that anything is a sizable mass lesion. The diagnosis of Hürthle cell adenoma (HCA) or Hürthle cell carcinoma (HCC) has typically relied on a judgement call by pathologists as to the presence or absence of capsular and/or vascular invasion of the adjacent thyroid parenchyma, easy to note in widely invasive disease and a somewhat subjective diagnosis for minimally invasive or borderline invasive disease. Diagnostic specificity, which has incorporated a sharp increase in molecular genetic studies of thyroid tumor subtypes and the integration of molecular testing into preoperative management protocols, continues to be challenged by Hürthle cell neoplasia. Here, we provide the improving yet still murky state of what is known about Hürthle cell tumor genetics, clinical management, and based upon what we are learning about the genetics of other thyroid tumors, how to manage expectations, by pathologists, clinicians, and patients, for more actionable, precise classifications of Hürthle cell tumors of the thyroid.

Hürthle cell neoplasms of the thyroid: Pathologic outcomes and ultrasonographic analysis

BACKGROUND

Fine-needle aspiration (FNA) of thyroid nodules suspicious for Hürthle cell neoplasms (HCN) have uncertain rate of malignancy. We aim to characterize rate and predictors of malignancy at our institution and compare these findings with established literature to help guide management.

METHODS

Single tertiary-referral center, retrospective study of 166 adults who underwent hemithyroidectomy or total thyroidectomy following FNA suspicious for HCN from 1998-2018. Demographic information and surgical histopathologic results were collected. Preoperative ultrasonography was independently scored on the American College of Radiology Thyroid Imaging Reporting and Data System (TI-RADS) by a board-certified head and neck radiologist.

RESULTS

There were 39 males and 127 females. Overall, 25 (15.1%) patients had carcinoma, with 15 (9%) being Hürthle cell carcinoma (HCC). Men had higher incidence of carcinoma (25.6% v. 11.8%, P = .035), and especially older males. Contralateral carcinoma was seen in 3 of 13 (23.1%) patients that underwent completion thyroidectomy. Patients with carcinoma had larger nodules (average diameter 3.3 cm versus 2.5 cm, respectively, P = .01), but no association with TI-RADS.

CONCLUSION

Adults with nodules suspicious for HCN have significant risk of malignancy consistent with prior studies. Older males and larger nodule diameter are associated with malignancy in this cohort, but TI-RADS grade is not. These findings provide a framework for management and counseling for lesions suspicious for HCN.

Features of Cytologically Indeterminate Molecularly Benign Nodules Treated With Surgery

BACKGROUND

Most cytologically indeterminate thyroid nodules (ITNs) with benign molecular testing are not surgically removed. The data on clinical outcomes of these nodules are limited.

METHODS

We retrospectively analyzed all ITNs where molecular testing was performed either with the Afirma gene expression classifier or Afirma gene sequencing classifier between 2011 and 2018 at a single institution.

RESULTS

Thirty-eight out of 289 molecularly benign ITNs were ultimately resected. The most common reason for surgery was compressive symptoms (39%). In multivariable modeling, patients aged <40 years, nodules ≥3 cm, presence of an Afirma suspicious nodule other than the index nodule, and compressive symptoms were associated with higher surgery rates with hazard ratios for surgery of 3.5 (P < 0.001), 3.2 (P < 0.001), 16.8 (P < 0.001), and 7.31 (P < 0.001), respectively. Of resected nodules, 5 were malignant. False-negative rate (FNR) was 1.7%, presuming all unresected nodules were truly benign and 13.2% restricting analysis to resected cases. The FNR was significantly higher in nodules with a high-risk sonographic appearance for cancer (American Thyroid Association high-risk classification and American College of Radiology Thyroid Imaging Reporting and Data Systems score of 5) compared with nodules with all other sonographic categories (11.8% vs 1.1%; P = 0.03 and 11.1% vs 1.1%; P = 0.02, respectively).

CONCLUSIONS

Younger age, larger nodule size, presence of an Afirma suspicious nodule other than the index nodule, and compressive symptoms were associated with a higher rate of surgery. The FNR of benign Afirma was significantly higher in nodules with high-risk sonographic features.

Use of artificial intelligence and machine learning for estimating malignancy risk of thyroid nodules

PURPOSE OF REVIEW

Current methods for thyroid nodule risk stratification are subjective, and artificial intelligence algorithms have been used to overcome this shortcoming. In this review, we summarize recent developments in the application of artificial intelligence algorithms for estimating the risks of malignancy in a thyroid nodule.

RECENT FINDINGS

Artificial intelligence have been used to predict malignancy in thyroid nodules using ultrasound images, cytopathology images, and molecular markers. Recent clinical trials have shown that artificial intelligence model's performance matched that of experienced radiologists and pathologists. Explainable artificial intelligence models are being developed to avoid the black box problem. Risk stratification algorithms using artificial intelligence for thyroid nodules are now commercially available in many countries.

SUMMARY

Artificial intelligence models could become a useful tool in a thyroidolgist's armamentarium as a decision support tool. Increased adoption of this emerging technology will depend upon increased awareness of the potential benefits and pitfalls in using artificial intelligence.

The Molecular Landscape of Hürthle Cell Thyroid Cancer Is Associated with Altered Mitochondrial Function-A Comprehensive Review

Hürthle cell thyroid carcinoma (HTC) accounts for 3-5% of all thyroid malignancies. Widely invasive HTC is characterized by poor prognosis and limited responsiveness to standard therapy with radioiodine. The molecular landscape of HTC is significantly different from the genetic signature seen in other forms of thyroid cancer. We performed a comprehensive literature review on the association between the molecular features of HTC and cancer metabolism. We searched the Pubmed, Embase, and Medline databases for clinical and translational studies published between 1980 and 2020 in English, coupling "HTC" with the following keywords: "genomic analysis", "mutations", "exome sequencing", "molecular", "mitochondria", "metabolism", "oxidative phosphorylation", "glycolysis", "oxidative stress", "reactive oxygen species", and "oncogenes". HTC is characterized by frequent complex I mitochondrial DNA mutations as early clonal events. This genetic signature is associated with the abundance of malfunctioning mitochondria in cancer cells. HTC relies predominantly on aerobic glycolysis as a source of energy production, as oxidative phosphorylation-related genes are downregulated. The enhanced glucose utilization by HTC is used for diagnostic purposes in the clinical setting for the detection of metastases by fluorodeoxyglucose positron emission tomography (FGD-PET/CT) imaging. A comprehensive metabolomic profiling of HTC in association with its molecular landscape might be necessary for the implementation of tumor-specific therapeutic approaches.

The Role of Molecular Testing for the Indeterminate Thyroid FNA

Thyroid nodules are common in the adult population where a majority are benign and only 4.0% to 6.5% are malignant. Fine needle aspiration (FNA) is a key method used in the early stages to evaluate and triage patients with thyroid nodules. While a definitive cytological diagnosis is provided in more than 70–75% of all thyroid FNA cases, the group of indeterminate lesions offers a challenge in terms of interpretation and clinical management. Molecular testing platforms have been developed, are recognized as an option by the 2015 American Thyroid Association Guidelines, and are frequently used in conjunction with FNA as an integral part of the cytologic evaluation. In this review, the utility of molecular testing options for nodules assigned to the group of indeterminate thyroid FNAs is described.

Afirma Gene Sequencing Classifier Compared with Gene Expression Classifier in Indeterminate Thyroid Nodules

Background: The Afirma Gene Expression Classifier (GEC) has been used to further characterize cytologically indeterminate (cyto-I) thyroid nodules into either benign or suspicious categories. However, its relatively low positive predictive value (PPV) limited its use as a classifier for patients with suspicious results. The Afirma Gene Sequencing Classifier (GSC) was developed to improve PPV while maintaining a high negative predictive value (NPV), yet real-world assessment of its performance is lacking. Methods: We analyzed all patients who had cyto-I nodules and molecular testing with either GEC or GSC between 2011 and 2018 at a single academic medical center. Clinical information was obtained for 343 GEC-tested nodules and 164 GSC-tested nodules. Results: The GSC had a statistically significant higher benign call rate (76.2% vs. 48.1%, p < 0.001), PPV (60.0% vs. 33.3%, p = 0.01), and specificity (94.3% vs. 61.4%, p < 0.001) than the GEC. Improvement was statistically significant in both Bethesda III and Bethesda IV nodules. In particular, the benign call rate of GSC was significantly higher in nodules with Hürthle cell changes (88.8% vs. 25.7%, p < 0.01). The rate of surgical intervention in the indeterminate nodule cohort has decreased by 66.4% since switching to the GSC; 52.5% of indeterminate nodules went to surgery while using the GEC compared with 17.6% with the GSC (p < 0.001). This reduction was statistically significant in nodules with Bethesda III diagnoses, demonstrating a 70.9% decrease (GEC 51.3% vs. GSC 14.9%, p < 0.001), and in nodules with Bethesda IV cytology, a 39.2% decrease was noted (GEC 54.8% vs. GSC 33.3%, p = 0.003). Conclusions: Data from a single academic tertiary center show an improved specificity and PPV while maintaining high sensitivity and NPV for GSC compared with GEC. A statistically significant increase in benign call rates was observed in GSC compared with GEC, likely indicating fewer false positive results. After implementation of GSC, surgical interventions have been reduced by 68%.

Extending expressed RNA genomics from surgical decision making for cytologically indeterminate thyroid nodules to targeting therapies for metastatic thyroid cancer

The Afirma Genomic Sequencing Classifier (GSC) is a rule‐out test for malignancy/noninvasive follicular thyroid neoplasms with papillary‐like nuclear features among patients with Bethesda category III/IV nodules, whereas the complimentary Xpression Atlas provides genomic insights from a curated panel of 511 genes among GSC suspicious and Bethesda category V/VI nodules. Together, they facilitate personalized treatment decisions based on genomic insights derived from the transcriptome of the biopsied target and extend the diagnostic and therapeutic reach of cytopathologists and fine‐needle aspiration biopsy sample collection.

Analytical Verification Performance of Afirma Genomic Sequencing Classifier in the Diagnosis of Cytologically Indeterminate Thyroid Nodules

Background: Fine needle aspiration (FNA) cytology, a diagnostic test central to thyroid nodule management, may yield indeterminate results in up to 30% of cases. The Afirma® Genomic Sequencing Classifier (GSC) was developed and clinically validated to utilize genomic material obtained during the FNA to accurately identify benign nodules among those deemed cytologically indeterminate so that diagnostic surgery can be avoided. A key question for diagnostic tests is their robustness under different perturbations that may occur in the lab. Herein, we describe the analytical performance of the Afirma GSC. Results: We examined the analytical sensitivity of the Afirma GSC to varied input RNA amounts and the limit of detection of malignant signals with heterogenous samples mixed with adjacent normal or benign tissues. We also evaluated the analytical specificity from potential interfering substances such as blood and genomic DNA. Further, the inter-laboratory, intra-run, and inter-run reproducibility of the assay were examined. Analytical sensitivity analysis showed that Afirma GSC calls are tolerant to variation in RNA input amount (5-30 ng), and up to 75% dilution of malignant FNA material. Analytical specificity studies demonstrated Afirma GSC remains accurate in presence of up to 75% blood or 30% genomic DNA. The Afirma GSC results are highly reproducible across different operators, runs, reagent lots, and laboratories. Conclusion: The analytical robustness and reproducibility of the Afirma GSC test support its routine clinical use among thyroid nodules with indeterminant FNA cytology.