The contribution of molecular pathology to the classification of thyroid tumors

Abstract

Significant molecular advances have been undertaken for the past two decades in the field of thyroid follicular neoplasms, including a detailed genomic profile of papillary thyroid carcinoma (PTC) by The Cancer Genome Atlas (TCGA) project. These molecular discoveries led to a better understanding of the pathogenesis of thyroid neoplasms and resulted in reclassification of certain types of thyroid tumors. This review discusses how, 1) the molecular profiles of follicular-patterned lesions led to the reclassification of the follicular variant of PTC into non-invasive follicular thyroid neoplasm with papillary like nuclei, 2) the genotyping of Hürthle cell neoplasm provided the rationale to classify these tumors independently from follicular adenomas and carcinomas, and 3) BRAF and RAS molecular signatures have the potential of subclassifying PTC and poorly differentiated thyroid carcinoma into clinically relevant molecular subtypes.

Follicular patterned lesions

In the 1950s, carcinomas were classified purely based on the predominant architecture: carcinomas making papillae were labeled as papillary thyroid carcinoma while those with follicles were diagnosed as follicular carcinoma (FTC) (Figure 1) ¹. Lindsay was the first to describe the nuclear features of papillary thyroid carcinoma in follicular-pattered lesions and coin the term follicular variant of papillary thyroid carcinoma (FVPTC) ². Chen and Rosai (1977) defined FVPTC as tumors that “resembled papillary carcinoma in its biologic behavior and all morphologic features with the exception that papillae were not present”, thereby proving that it belonged to the papillary, as opposed to the follicular carcinoma family ³. During the 1980s and 1990s the diagnosis of papillary carcinoma began to be more common than follicular carcinomas as nuclear features became the defining criterion for the diagnosis of PTC, regardless of architectural pattern and invasive status ⁴. Based on that axiom and although FVPTCs first described by Lindsay and then by Chen and Rosai were infiltrative tumors with nodal metastasis, a new group of tumors began to appear, the noninvasive encapsulated follicular variant of PTC(NI-EFVPTC) defined as malignant solely on the basis of its nuclear characteristics ( e.g. nuclear enlargement, nuclear membrane irregularity, and chromatin clearing) (Figure 2). As the threshold for nuclear alteration required to diagnose a tumor as PTC became lower ^{5, 6}, this new entity, NI-EFVPTC, started to become a frequent diagnosis, accounting for nearly 15% of all PTCs diagnosed in Europe and United states ^{7, 8}. It is estimated that 45,000 patients worldwide are diagnosed with NI-EFVPTC every year ⁹. As most of the previous studies did not separate encapsulated from infiltrative FVPTC and some even included tumors with small proportion of papillae (e.g. >1%) as FVPTC, the reported rate of nodal metastasis in FVPTC overall was as high as 22% ^{10, 11}. It was thus assumed for more than 30 years that the encapsulated FVPTC behaved and spread like its classical counterpart. The crisis in the over-diagnosis of FVPTC led to the proposal formulated in Europe in 2000 of the well differentiated tumor of uncertain malignant potential(WDT-UMP): An encapsulated tumor composed of well differentiated follicular cells with questionable papillary carcinoma type nuclei, blood vessel invasion and/or capsular invasion that is either absent or questionable ¹². This diagnostic term gained acceptance in several European Institutions but was not used in the vast majority of Hospitals in the US ¹³.

Figure 1. Timeline depicting the evolution of histologic classification and the identification of common molecular events in well differentiated thyroid carcinoma.

The references pertaining to the above time points are as follows: Histology in chronological order: Lindsay S ², Chen and Rosai ³, WDT-UMP ¹², behavior of non-invasive PTC-EFV similar to follicular adenoma ²⁷, and NIFTP ⁹; molecular studies: RAS in FTC ¹⁴, RET-PTC fusion in PTC ¹⁸, PPARγ-PAX8 in FTC/FA ^{16, 17}, BRAF^V600E in PTC ^{19, 20}, RAS in FVPTC, pattern of chromosomal (chr.) gains/losses in FVPTC different than classical PTC ^{21, 22}, the Cancer Genome Atlas Research Network molecular analysis of PTC ²⁶. PTC, papillary thyroid carcinoma; FTC, follicular carcinoma; WDT-UMP, well-differentiated tumor of uncertain malignant potential; FVPTC, follicular variant of papillary thyroid carcinoma; NI-EFVPTC, noninvasive encapsulated follicular variant of papillary thyroid carcinoma; TCGA, the cancer genome atlas research network; NIFTP, noninvasive follicular thyroid neoplasms with papillary-like nuclear features.

Figure 2. Histologic features and molecular profile of common differentiated thyroid neoplasms.

Black arrows indicate vascular invasion (VI). NIFTP: noninvasive follicular thyroid neoplasm with papillary-like nuclear features, EFVPTC: encapsulated follicular variant papillary thyroid carcinoma, IFVPTC: infiltrative follicular variant of papillary thyroid carcinoma, FA: follicular adenoma, EFTC: encapsulated follicular thyroid carcinoma, CV: classical variant, TCV: tall cell variant, T: tumor. Arrows: vascular invasion.

Meanwhile, significant advances have taken place in the molecular understanding of thyroid neoplasms since the late 1980s (Fig 1). RAS point mutations were first discovered in follicular carcinoma in 1988 (Figures 1 and 2) ¹⁴, and soon after in follicular adenoma (FA) ¹⁵. Subsequent studies have also reported PAX8-PPARγ fusion in a subset of FTC/FA ^{16, 17}. On the other hand, a significant proportion of PTCs, in particular the classical variant, has been shown to harbor RET-PTC fusion and BRAF^V600E mutation ^18–20. It was not until 2003 that molecular studies began to indicate that a significant proportion (approximately 43%) of FVPTC harbor RAS mutation and a pattern of chromosome gain/loss akin to FA/FTC and in contrast to classical variant of PTC ^21–23. Furthermore, the expression profile of the follicular variant of PTC was shown to be different from that of classical and tall cell variant ²³. Additional studies showed that, as currently diagnosed, the encapsulated/well demarcated FVPTC has molecular alterations similar to follicular adenoma and carcinoma, characterized by the lack of BRAF mutations, a high prevalence of RAS mutations, and in some instances the presence of PAX8-PPARγ rearrangement ^{24, 25}. In 2014, the analysis of approximately 500 papillary carcinomas by The Cancer Genome Atlas (TCGA) research network confirmed all previous molecular studies demonstrating that the FVPTC group of tumors, invasive and non-invasive, have a high frequency of RAS mutation and a RAS-like molecular signature, unlike the classical variant and tall cell variant of papillary carcinoma that features BRAF mutations and a BRAF^V600E-like profile ²⁶.

Inspired by these molecular findings, Liu et al. in 2006 re-examined the clinical behaviors of EFVPTC. In that study, EFVPTC behaved like the FA/FTC group of tumors, and no recurrence or nodal metastasis developed in their NI-EFVPTC cohort treated by surgery alone with a median follow up of 11-years ²⁷. This indolent behavior has subsequently been confirmed by multiple studies ^{8, 28–34}. In 2016, the problems generated by the over-diagnosis and treatment of the NI-EFVPTC brought a working group of the Endocrine Pathology Society to critically re-examine this entity ⁹. As a result of this endeavor, Nikiforov et al. advocated a revision of diagnostic terminology to noninvasive follicular thyroid neoplasm with papillary-like nuclear features (NIFTP) with the fundamental goal of avoiding the term carcinoma, and the consequent risk of overtreatment for non-invasive tumors that are clinically benign, with a recurrence rate < 1% (Table 1) ⁹. Molecular profiling was performed in 27 tumors from the consensus NIFTP cohort, which again demonstrated a high frequency of RAS mutation (8/27, 30%) and PAX8-PPARγ fusion (6/27, 22%), as well as the absence of BRAF ^V600E in this group of tumor ⁹. In conclusion, the analysis of the molecular profile of follicular-patterned lesion was the key step that prompted a better histopathological appraisal and clinical correlation of different forms of FVPTC, which eventually led to proper re-classification of a subgroup of these lesions as NIFTP. The fourth edition of the World Health Organization (WHO) classification for endocrine tumors (2017) ³⁵ has adopted these new concepts and classified follicular patterned lesion based on both invasive status and nuclear characteristics (Figure 3). Tumors with invasion are classified as carcinoma, while encapsulated lesions without evidence of capsular or vascular invasion are follicular adenomas or NIFTPs. In addition, a third category of tumors, named as “tumors of uncertain malignant potential”, was added to address encapsulated lesions with questionable capsular or vascular invasion. Therefore, whenever we encounter a follicular patterned nodule, regardless of the presence and extent of the alterations of nuclear morphology in neoplastic cells, pathologists have to carefully analyze and report whether the tumor is invasive or not. If possible, the entire capsule of the nodule and the nodule/adjacent tissue interface should be submitted and examined histologically.

Table 1.

Consensus diagnostic criteria of noninvasive follicular thyroid neoplasms with papillary-like nuclear features (NIFTP, Nikiforov et al. 2016⁹).

1. Encapsulated or clear demarcation

2. Follicular growth pattern with

< 1% papillae

No psammoma bodies

< 30% solid/trabecular/insular growth pattern

3. Nuclear score 2–3

4. No vascular or capsular invasion

5. No tumor necrosis

6. No high mitotic activity (3 or more mitoses per 10 high power fields, 400x)

Figure 3. Evolution of classification of follicular patterned lesions.

The old concept adopted by WHO third edition (2004) was based predominantly on nuclear features, while the new concept endorsed by WHO fourth edition (2017) evaluates a tumor using a combination of invasion and nuclear features. FA: follicular adenoma, FTC: follicular thyroid carcinoma, FVPTC: follicular variant of papillary thyroid carcinoma, NIFTP: noninvasive follicular thyroid neoplasm with papillary-like nuclear features, FTUMP: follicular tumor of uncertain malignant potential, WDTUMP: well differentiated tumor of uncertain malignant potential, WDC: well-differentiated carcinoma, EFVPTC: encapsulated follicular variant of papillary thyroid carcinoma, CI: capsular invasion, VI: vascular invasion,

Practice points:

1. A subset of noninvasive encapsulated follicular variant of papillary thyroid carcinoma (NIEFVPTC) has been recently renamed as noninvasive follicular thyroid neoplasm with papillary-like nuclear features (NIFTP). This reclassification has been prompted by the molecular profile of FVPTC-which is close to FA/FTC and different from classical PTC.

2. Classification of follicular-patterned lesion is based on a combination of invasive status and nuclear features in the new WHO classification.

3. Extensive (preferably total) sampling of the tumor capsule is crucial for evaluating capsular and vascular invasion in an encapsulated follicular-patterned lesion especially in the era of NIFTP diagnosis.

Hürthle cell neoplasm: an independent entity rather than a subtype of follicular neoplasm

Hürthle cells, characterized by their abundant eosinophilic granular cytoplasm, hyperchromatic/vesicular nuclei, and prominent round central nucleoli, have been first described in the thyroid gland by Askanazy in 1898 ³⁶. Hürthle cells neoplasms, encompassing Hürthle cell adenomas (HCA) and Hürthle cell carcinomas (HCC), are composed predominantly or exclusively of Hürthle cells ³⁵. Similar to follicular-patterned lesions as discussed above, the presence of invasion can reliably predict the behaviors and outcome of Hürthle cell neoplasms ³⁷. HCA, which lacks capsular or vascular invasion, follow a benign clinical course, with no or negligible risk of relapse or disease-specific mortality ^{37, 38}. On the other hand, HCC demonstrates indisputable evidence of invasion and have been traditionally divided into minimally invasive (encapsulated), and widely invasive categories ³⁹. Encapsulated HCC can display capsular invasion and/or vascular invasion. The extent of vascular invasion predicts outcome with tumors displaying 4 or more vessels having a worse prognosis than those with lesser degree of angioinvasion ⁴⁰. Widely invasive carcinoma shows obvious gross invasion and prominent capsular penetration, often with extrathyroidal extension ³⁵. Encapsulated HCC with minimal or no vascular invasion are usually indolent while widely invasive HCCs are much more aggressive with 62% recurrence rate in one study ^40–42. It is interesting to note that that Hürthle cell carcinomas with significant vascular invasion (≥ 4 foci and/or extra-thyroid vascular invasion) have a different expression profile than those with lesser foci of angioinvasion (Figure 4) ⁴³.

Figure 4: Principal component analysis of expression data derived from Hurthle cell tumors.

The subclasses Hurthle cell adenoma (HCA), Hurthle cell carcinoma (HCC) with focal vascular invasion (VI), and HCC with significant/extensive VI are noted in the color legend (blue, red and green respectively). Heatmap shows differential gene expression between HCA, HCC with focal VI, and HCC with significant/extensive VI. HCC with focal VI (<4 foci) shows a similar profile to HA, whereas HCC with significant/extensive VI (>=4 foci and/or extra-thyroid VI) shows a more divergent expression profile. Modified and reproduced with permission from Ganly et al. ⁴³

In the third edition of WHO classification (2004) ⁴¹, Hürthle cell neoplasms were considered as histopathological variants of follicular adenoma and follicular carcinoma. Several recent studies have shown that Hürthle cell tumors have a higher frequency of mitochondrial DNA mutations, leading to defective multimeric complex of the inner mitochondrial membrane, impaired oxidative phosphorylation, and compensatory accumulation of mitochondria ^{44, 45}. An integrated analysis of mutational and copy number variation in HCC has shown that these tumors are distinctly different from follicular and papillary carcinomas and lack HRAS, KRAS, and BRAF mutations ⁴³. In regard to the patterns of chromosomal aberrations, HCC are also quite different from PTC and FTC (Figure 5). The most striking differentiating feature in HCC was large regions of gain on chromosomes 5, 7, 12, and 17 ^{43, 46–48}. Widely invasive HCC is further characterized by additional alterations in PIK3CA-AKT-mTOR and Wnt/β-catenin pathway. Compared with follicular carcinomas, Hürthle cell carcinomas are often associated with a decreased recurrence free survival ⁴⁹ and resistant to radioactive iodine therapy ⁵⁰. Therefore, in the fourth edition of WHO classification (2017), HCC has been separated from FTC, becoming an independent entity ³⁵.

Figure 5. Summary of chromosomal regions of gain and loss in HCC compared with PTC and FTC.

Regions of gain are shown in red, and regions of loss are shown in green. HCC have a different pattern of chromosome aberrations than the one reported in FTC and PTC. * Singh et al. ⁴⁶ and Wreesmann et al. ⁴⁷ ** Hemmer et al ⁴⁸. Reprinted with permission from Ganly et al. ⁴³

Practice points

• Hürthle cell carcinomas are now being recognized as independent tumor entities and are separated from follicular carcinomas in view of their particular molecular profile, RAI avidity and propensity for recurrence.

• The prognosis of Hürthle cell carcinomas is related to the extent of angioinvasion. The presence of 4 or more foci of vascular invasion denotes an aggressive behavior. Tumors with significant angioinvasion (4 or more foci and/or extrathyroidal vascular invasion) have an expression profile different from those carcinomas with a lesser degree of vascular invasion.

Future directions: molecular classification of papillary thyroid carcinoma and poorly differentiated thyroid carcinoma

The TCGA study on papillary thyroid carcinoma has introduced the concepts of BRAF^V600E-RAS score and thyroid differentiation score (TDS) ²⁶. The BRAF^V600E-RAS score was developed by comparing the RNA sequencing data of BRAF^V600E -mutated and RAS-mutated PTCs. A seventy one gene signature was derived from this comparison leading to a continuous measure (−1 to +1) with BRAF^V600E –like PTC being negative and RAS-like PTC being positive. The TDS is a single metric produced from the expression levels of 16 thyroid metabolism and function genes highly correlated in the TCGA cohort. PTCs BRAF^V600E –mutation and BRAF^V600E –like molecular signatures are often classical or tall cell variants by morphology, less differentiated with a low thyroid differentiation score, and are enriched with RAI-refractory PTCs. On the other hand, PTCs with RAS-mutation and RAS-like signatures are often follicular variants, highly differentiated with high thyroid differentiation score, occur in young patient, with low risk of recurrence, and do not convey RAI-resistance ²⁶. Applying unsupervised clustering methods using four genomic data sets (mRNA, microRNA, DNA methylation and protein expression), RAS like PTC were quite homogenous associated with a single cluster in all data sets except for DNA methylation. In contrast, BRAF^V600E –like were a heterogeneous group composed of different clusters in each data set with no overlap between the clusters. The most particular cluster identified within the BRAF^V600E –like PTC was mRNA 5 termed tall-cell like. This subgroup showed the strongest BRAF-like phenotype, least differentiation, more advanced stage, and higher risk of recurrence ²⁶. The heterogeneity within the BRAF tumors may explain the uncertainty of the prognosis and predictive power of this mutation. The results from the TCGA study suggest that the genetic differences between and within BRAF^V600E –like and RAS-like PTC may be used to create molecular subtypes that would lead to more refined surgical and medical therapy in the era of precision medicine.

In 1984, poorly differentiated thyroid carcinoma was first described by Carcangiu et al. ⁵¹ as a carcinoma with insular growth pattern and a prognosis in between the indolent well differentiated thyroid carcinoma and the extremely aggressive and often fatal anaplastic carcinomas. It was not until 2007 that a uniform set of diagnostic criteria for PDTC, namely the Turin proposal, had been put forward by a group of international thyroid experts. These are: 1) insular/solid/trabecular growth pattern; 2) the absence of nuclear features of papillary thyroid carcinoma; and 3) one or more of the following three features: convoluted nuclei, mitotic index of 3 or more per 10 high power fields, and tumor necrosis ⁵². Earlier, an alternative set of criteria for PDTC, the Memorial Sloan-Kettering Cancer Center (MSKCC) criteria, was proposed based solely on a mitotic index of ≥ 5/10 HPFs and/or tumor necrosis regardless of tumor growth patterns and nuclear features ⁵³. Both approaches were able to identify tumors with an intermediate prognosis with a reported mortality rate of 38 – 57% ^{52, 53}. In view of its intermediate prognosis, patients with PDTC have to be counseled, followed and treated differently than those with garden variety PTC.

In 2016, Landa et al. ⁵⁴ published the genomic profile of PDTC using ultra-deep targeted exome next generation sequencing platform in a large cohort of 84 PDTCs. Similar to the well-differentiated PTC, BRAF^V600E and RAS mutations remain the main driver mutations in PDTC, affecting 33% and 28% of cases respectively ⁵⁴. The BRAF/RAS mutations are also closely correlated with the histologic diagnostic criteria of PDTC: PDTCs fulfilling both Turin proposal and MSKCC criteria harbor a high frequency (42%) of RAS mutation, whereas 78% of PDTCs diagnosed solely by MSKCC criteria are associated with BRAF^V600E mutation ⁵⁴. Interestingly, Landa et al. have also found a significant association of BRAF/RAS status and metastatic routes: BRAF-mutated PDTCs often metastasize to regional lymph nodes (63% compared with 14% in RAS-mutated PDTCs), while RAS-mutated PDTCs have an increased risk to develop distant metastasis (67% compared with 14% in BRAF-mutated PDTCs). In conclusion, recent molecular advances seems to suggest that PDTCs can be further subdivided based on their driver mutations. RAS-driven and BRAF^V600E-driven PDTCs are morphologically different and carry distinct patterns of metastasis.

Practice points

• BRAF^V600E like PTC are heterogenous tumor while RAS like PTC are rather homogeneous. These genetic differences between and within BRAF^V600E like and RAS like tumors may lead to a molecular classification with potential clinical impact

• PDTCs can be further subdivided based on their driver mutations. RAS-driven and BRAF^V600Edriven PDTCs are morphologically different and carry distinct patterns of metastasis.

Conclusion:

Considerable molecular knowledge has been acquired in follicular cell derived thyroid tumors in the last 30 years. In addition to the development of new molecular based diagnostic and prognostic markers, these advances have led to a more clinically relevant histopathologic classification of certain neoplasms.

The prime example was the renaming of the noninvasive encapsulated FVPTC into NIFTP. This change in nomenclature will help thousands of patients throughout the world avoid unnecessary therapy and the psychosocial impact of a cancer diagnosis. This significant achievement would not have been realized without the intersection of the molecular, histopathologic and clinical outcome data. This clearly demonstrates that all aspects of a tumor should be considered when deciding about its nomenclature and therefore its therapy.