How Many Papillae in Conventional Papillary Carcinoma? A Clinical Evidence-Based Pathology Study of 235 Unifocal Encapsulated Papillary Thyroid Carcinomas,with Emphasis on the Diagnosis of Noninvasive Follicular Thyroid Neoplasm with Papillary-Like Nuclear Features

Introduction

Papillary thyroid carcinoma (PTC) is the most common thyroid cancer and it has shown a dramatic increase of incidence for the past several decades (1,2). For the past 40 years, the pathologic diagnosis of PTC has been based purely on its nuclear features: (i) enlarged elongated shape; (ii) chromatin clearing; and (iii) nuclear membrane irregularity (1,3,4). On the other hand, the subtyping of PTC factors depends on several parameters, including architectural pattern (e.g., follicular variant with follicular growth pattern and conventional [classic] PTC with well-formed papillae), cytomorphology (e.g., oncocytic variant, columnar variant, hobnail variant, and tall cell variant), encapsulation (e.g., encapsulated variant surrounded completely by a tumor capsule), and size (papillary microcarcinoma that measures 1 cm or less) (1,4). In particular, and pertinent to this study, follicular variants of PTC (FVPTC) are those PTCs with “exclusively or almost exclusively follicular growth pattern” (1,3) that have been traditionally translated into a cutoff of 1% papillae to separate FVPTC (<1% papillae) and conventional PTCs with a follicular predominant growth pattern (≥1% papillae) (4).

In 2014, a comprehensive genomic analysis by The Cancer Genome Atlas (TCGA) has shown that the follicular variant and conventional PTC have distinct molecular signatures. The FVPTC is associated with RAS mutations, an RAS-like molecular signature, and is highly differentiated with a high thyroid differentiation score (TDS). The classical PTC (CPTC) has a high frequency of BRAF^V600E mutations, a BRAF-like molecular signature, and a low TDS (5). Clinically, FVPTC, especially the encapsulated and noninvasive form, tends to follow a highly indolent clinical course with negligible risk of recurrence and lymph node metastasis (3,6,7). Based on these compelling clinical and molecular evidence, a subset of noninvasive encapsulated FVPTC with (near) exclusive follicular growth pattern, <30% solid growth, and ≤1% papillae was renamed as noninvasive follicular thyroid neoplasm with papillary-like nuclear features (NIFTP) in 2016, to highlight the indolent nature of these lesions by eliminating the word “carcinoma” from the diagnosis, with the aim of avoiding overtreatment (3,8).

However, since the publication of the NIFTP consensus paper, three studies have reported that encapsulated follicular lesions with less than 1% of true papillae and PTC nuclei have a small risk (2% to 5.5%) of lymph node metastasis (9 –11). As a result, commentaries were published subsequently on behalf of the NIFTP consensus group to revise the diagnostic criteria from less than 1% of true papillae to no true papillae allowed (8,12). This revision practically means that the finding of a single papilla in an otherwise noninvasive encapsulated follicular-patterned lesion with papillary-like nuclear features will result in a diagnosis of cancer.

In this retrospective cohort of 235 patients with unifocal encapsulated PTC (U-EPTC), we aimed at studying the correlation between architectural patterns, in particular the percentage of true papillae, with the risk of nodal metastasis and underlying BRAF/NRAS alterations. Our focus was to establish a clinical evidence-based cutoff for the percentage of papillae that can be used to define NIFTP and subclassify PTC. This, in turn, would lead to a better patient stratification, potentially sparing unnecessary therapy to individuals at a negligible clinical risk.

Materials and Methods

Pathologic and clinical review

The architectural patterns and their respective percentage were collected for each tumor. The papillary pattern was defined by the presence of projections lined by lesional follicular epithelium and a well-defined fibrovascular core (“true” papillae) (Fig. 1A–D). Obliquely cut septa with fibrovascular cores and abortive papillae lacking a fibrovascular core were not included (Fig. 1E, F). It is our experience that in the day-to-day practice it may be difficult to differentiate “true” papillae from obliquely-cut septa with fibrovascular cores and abortive papillae. Further, it is not uncommon to find foci of papillary hyperplasia in follicular-patterned lesions lacking the PTC nuclear features. For the purpose of this study as stated earlier, only “true” papillae lined by lesional follicular cells and a fibrovascular core were considered (Fig. 1). For cases with 0.1% to 10% papillae, the presence and percentage of papillae were determined at consensus review sessions between BX and RG. Follicular pattern was characterized by follicles of any size (macrofollicles or microfollicles) with central colloid surrounded by follicular cells. A solid pattern consisted of areas of solid, trabecular, or insular growth.

OPEN IN VIEWER

FIG. 1.

Papillae in U-EPTC. (A, B) True papillae with fibrovascular cores (arrows) (A, H&E) in a classical PTC almost completely composed of papillae BRAF^V600E-positive by immunohistochemistry (B). (C, D) An encapsulated PTC with follicular predominant growth pattern containing occasional (<1%) papillary structures (arrows). Inset shows typical PTC nuclei (C, H&E); neoplastic cells are positive for NRAS^Q61R by immunohistochemistry (D). (E) Pseudopapillae not fulfilling the definition of true papillae since they lack fibrovascular core (arrows) (F) Pseudopapilla not fulfilling the definition of true papillae since it lacks fibrovascular core and appears to represent an artefactually ruptured septa. H&E, hematoxylin and eosin; PTC, papillary thyroid carcinoma; U-EPTC, unifocal encapsulated PTC.

Tumor size was measured as the maximum diameter of the resected tumor specimen. Capsular invasion was defined as complete penetration of the capsule by tumor, and the number of these foci was recorded. Capsular invasion was recorded as “focal” when there were <4 foci of complete penetration of the tumor capsule and was recorded as “extensive” when the foci were ≥4. Vascular invasion (blood vessel invasion) was defined according to the criteria outlined by the Armed Forces Institute of Pathology (AFIP) fascicle (13) and the WHO classification (1). In brief, only when the invasive focus protruded into the lumen of the vessel in a polypoid manner covered by endothelial cells, or when it was attached to the vessel wall or associated with thrombus formation within an intra-capsular or extra-capsular vasculature, it was considered true vascular invasion. Vascular invasion was recorded as “focal” when there were <4 foci and was recorded as “extensive” when there were ≥4 foci.

The patient's charts were reviewed for age at diagnosis, sex, type of surgery, radioactive iodine therapy, and outcome. The primary outcome of the study was the presence (AJCC pN1) or absence (AJCC pNx or pN0) of lymph node metastases at the time of initial resection. The secondary outcomes were disease-free survival and disease-specific survival.

Results

Results are summarized in Tables 1 –4, and they are illustrated in Figures 1 and 2.

OPEN IN VIEWER

FIG. 2.

Percentage of papillae within the tumor, nodal metastasis, and BRAF/NRAS mutation. (A) Papillae within the tumor and risk of nodal metastasis. (B) Papillae within the tumor and BRAF^V600E/NRAS^Q61R status subdivided to BRAF^V600E positive, NRAS^Q61R positive and negative for both immunostains; on the y axis, the tumor groups are subdivided based on the percentage of papillae, and the x axis shows the number of tumors. (C) Diagram showing how an increasing percentage of papillae within the tumor is associated with a higher rate of N1 disease, BRAF^V600E, and a lower frequency of NRAS^Q61R. The BRAF and NRAS mutational status was determined by immunohistochemistry using monoclonal antibodies for BRAF^V600E and NRAS^Q61R in a subset of 100 randomly selected cases.

Correlation between papillary architecture and risk of lymph node metastasis at the time of initial surgery

Twenty-seven patients (12%) in our cohort had lymph node metastasis (AJCC pN1) at the time of initial resection. Compared with the patients without nodal metastases (AJCC pNx or pN0), those with nodal disease were associated with younger age (p < 0.001), had a higher prevalence of capsular invasion (66% in N1 group and 26% in N0/Nx group, p < 0.001), and of microscopic extrathyroidal extension into perithyroidal fibroadipose tissue (3 out of 77, 11% in N1 group and 1 out of 208, 0.5% in N0/Nx group, p < 0.001, Table 1). More importantly, the N1 and N0 groups were associated with different architectural patterns: 89% of the tumors in the N1 group had a predominance (≥50% of total tumor volume) of papillary architecture, whereas 75% of Nx/N0 tumors showed (near) exclusive follicular growth with 0% to <1% papillae (p < 0.001). A small percentage of tumors contained a significant amount of solid architecture: 5 cases (2%) with 25–49% and 8 (3%) with ≥50% of solid growth pattern. None of these 13 cases with significant solid growth had pN1 disease. The presence of any amount of solid pattern did not correlate with nodal status (N0/Nx vs. N1, p = 0.409).

All tumors were further classified into several categories according to the percentage of papillae within the tumor: 127 (54%) with no (0%) papillae, 31 (13%) with 0.1–0.9% of papillae, 14 (6%) had 1–9% papillae, 6 (3%) had 10–24% papillae, 5 (2%) had 25–49% papillae, and 52 (22%) had at least 50% papillae. The risk of lymph node metastasis according to the percentage of papillae within the tumor is illustrated in Table 3 and Figure 2A. None of the 158 patients with less than 1% papillae developed lymph node metastases, while 24 out of 52 tumors (46%) with at least 50% papillae had pN1 disease. The frequency of nodal metastasis in tumors with 1–9%, 10–24%, and 25–50% papillae was 7% (1 out of 13), 33% (2 out of 6), and 0% (0 out of 5), respectively.

When subdividing the tumors into those with invasion (capsular and/or vascular, n = 79) and without invasion (n = 156), there was a significant association between papillary architectural pattern and nodal metastasis regardless of the invasion status (p < 0.001, Table 2). Encapsulated PTC with ≥50% papillae had a 50% risk (11 out of 22 cases) of nodal metastasis even without invasion, whereas noninvasive encapsulated tumors with less than 10% of papillae (129 cases) did not metastasize to lymph nodes (Table 3).

The case of U-EPTC with nodal metastasis and the lowest percentage of papillae (Fig. 3 and the case marked with superscript “C” in Table 3) had 5% papillae, had extensive capsular invasion, and measured 2.0 cm.

OPEN IN VIEWER

FIG. 3.

Microscopic pictures of a patient with a 2-cm encapsulated PTC harboring 5% papillae with capsular invasion and nodal metastases. (A) Low-power view of primary tumor with area of papillae formation (blue arrows). (B) High-power view of papillary area in A showing papillae with fibrovascular cores (blue arrow). (C) Primary tumor displaying mushrooming capsular invasion (black arrows). (D) Lymph node with metastatic deposits (red arrows). Scale bar: 500 μm in (A, C, and D), 50 μm in (B).

The cases of U-EPTC without invasion with the lowest percentage of papillae had 0.1% papillae: They included 6 tumors and measured from 2.5 to 5.0 cm.

Within each PTC group based on the percentage of papillae, the overall invasion status (capsular and/or vascular) did not have a significant impact on the risk of nodal metastasis (p ranged from 0.286 to 1.000 for each individual category, data not shown).

Discussion

FVPTC are those PTCs with an almost exclusive follicular growth pattern (1,3) that, when encapsulated, need to be distinguished from encapsulated forms of conventional PTC, which may also have a predominantly follicular growth pattern with a few papillae. Traditionally, a cutoff of 1% has been used to separate FVPTC (<1% papillae) and conventional PTCs with a follicular predominant growth pattern (≥1% papillae). The distinction is important since FVPTC has an RAS-like molecular signature and a tendency to metastasize to distant sites rather than to regional lymph nodes, whereas conventional PTC, even when encapsulated, is BRAF-like and typically metastasizes to lymph nodes first and only later to distant sites (4). The existence of full encapsulation is relatively uncommon but well documented in classic PTC, and it has been associated with a favorable clinical outcome compared with the far more common infiltrative forms of conventional PTC (16 –18).

Surprisingly, however, less than a handful of studies included a systematic evaluation of the percentage of papillae formation in their study design. Liu et al. defined FVPTC as a tumor with <1% papillae, and reported an overall nodal metastasis rate of 17%, including a 5% rate of nodal metastasis in the encapsulated FVPTC, and 0% nodal metastasis in the noninvasive encapsulated forms (19). Zidan et al. (20) allowed up to 20% papillae to define FVPTC and reported a nodal metastasis rate of 22% in FVPTC.

How many papillae are there in encapsulated PTCs that behave like classic PTC that tend to develop lymph node metastasis? The ability to reliably answer this question has gained a new relevance since the introduction of the NIFTP terminology with the clinical implication of a negligible to absent risk for the patient carrying such a tumor to develop lymph node or distant metastases.

In the attempt to address the issue, Cho et al. have reported a nodal metastasis risk of 3% in noninvasive encapsulated FVPTC with rare (<1%) papillae, and 2% in those without any papillae (10). Similarly, Kim et al. (11) showed a 2% risk of lymph node metastasis in noninvasive encapsulated FVPTC with 0% papillae. However, these two studies included tumors with synchronous papillary microcarcinomas. As it is well-established that papillary microcarcinoma may give rise to nodal metastasis (21 –24), the reported nodal metastasis risk in noninvasive encapsulated FVPTC may be exaggerated.

Parente et al. (9) reported a 5% risk of nodal metastasis and 1% risk of distant metastasis in unifocal noninvasive encapsulated FVPTC without any papillae. However, molecular analyses were not done in the positive cases on the primary tumor and the metastases to confirm that the metastatic deposit emanated from the identified primary tumor. Since metastatic nodal and even distant disease has been reported in total thyroidectomy specimens negative for carcinoma and entirely submitted for microscopic examination (25,26), one cannot exclude without molecular analysis that the metastatic deposits may have originated from separate infiltrative small carcinomas embedded in the paraffin block or which had undergone regression. Thus, in our opinion, these findings remain to be validated in other studies.

In contrast to the reports just cited, multiple studies, including the NIFTP consensus cohort, showed 0% risk of nodal metastasis in NIFTP, originally defined as a noninvasive encapsulated follicular-patterned lesion with <1% papillae and papillary-like nuclear features (3,6,15,27 –32). Nevertheless, essentially based on the data from Cho et al. (10) and Parente et al. (9), a proposal has been made to revise the diagnostic criteria for NIFTP from <1% papillae to no true papillae (8,12).

All this recent work highlights the relevance of our study. We provide—to our knowledge for the first time—a detailed, clinical evidence-based analysis dissecting the relationship between papillary percentage, tumor behavior, and BRAF/NRAS status in encapsulated PTC.

In this study, we demonstrate a strong correlation between the percentage of papillae, risk of nodal metastasis, and underlying BRAF/RAS mutation status in encapsulated PTC. Tumors with ≥50% of papillae were associated with 46% risk of nodal metastasis and frequent (82%) BRAF^V600E mutation, whereas tumors with (near) exclusive follicular growth pattern and very few (<1%) true papillae had 0% risk of nodal metastasis and were enriched with NRAS^Q61R (37.5%) mutations.

There may be a certain degree of subjectivity and ambiguity in defining one single true papilla and, consequently, diagnosing papillary carcinoma, and thus establishing a “cancer” diagnosis. We show that in any case—even when accounting for the occasional equivocal interpretation of histologic findings of papillae—tumors with rare (<1%) papillae are not uncommon (13% of the entire cohort) and are highly indolent with 0% risk of recurrence or lymph node metastasis. Therefore, the <1% papillae criterion as a cutoff for NIFTP appears sound for diagnostic purposes. The discrepant nodal metastatic rates between this study and the earlier mentioned three publications (9 –11) could be due to two factors. First, we selected only unifocal carcinoma, while the study of Cho et al. and Kim et al. included tumors with synchronous papillary microcarcinomas. Further, a difference in the interpretation of the percentage of papillae could be an additional reason for these variable results.

We also demonstrate that there is an incremental risk of nodal metastasis according to the percentage of papillae, being 7% in those with 1–9% papillae and 46% in those with ≥50% papillae. In our cohort, invasion by itself did not predict the risk of nodal disease. Although capsular invasion is significantly more frequent in tumors with N1 disease, such an association is no longer significant when cases are substratified by the percentage of papillae within the tumor. Taken together, the percentage of papillae is an important histologic feature, and could be more relevant than invasive status in predicting the behavior of encapsulated PTC. As tumors with at least 1% papillae in an encapsulated follicular-predominant lesion are at risk of developing nodal metastasis, they are best classified as CPTC, with a follicular predominant growth pattern.

The link between papillae and BRAF/RAS signature has been previously implied by multiple studies (3,5,19,33,34). For example, the TCGA PTC study showed that tumors enriched with true papillae (e.g., classic and tall cell variants) have a high frequency (89% for tall cell and 67% for classic variant) of BRAF^V600E mutation, whereas PTCs with (near) exclusive follicular growth pattern (i.e., FVPTC) are enriched with RAS mutations (38%) (5,35). The reported frequency of BRAF^V600E mutations in a tumor fulfilling the original NIFTP diagnostic criteria (i.e., <1% papillae) ranges from 0% (15,30,36,37) to 8% (10,11). The 8% rate was reported by two Korean studies. The 0% rates were studies from North America and/or Europe. It is plausible to assume that variation in ethnic background may have an impact on the discrepancy. This study is the first to demonstrate a strong association between a measured percentage of papillae and BRAF/RAS alterations in encapsulated PTC.

Ten percent papillae in an encapsulated PTC appears to be associated with a genotype switch from RAS mutations to BRAF^V600E mutations. Tumors with ≥10% papillae are predominantly BRAF^V600E -related (73%) and lack NRAS mutations (0%), whereas PTC with <10% papillae are enriched in RAS mutations (37%). Two tumors with <1% papillae were positive for BRAF^V600E, one of which contained 10% of cells showing tall cell cytomorphology, which would exclude this tumor from a diagnosis of NIFTP. As 89% to 95% of tall cell variant PTC harbor BRAF^V600E mutations (5,38), the BRAF^V600E immunopositivity in this tumor is explained by the presence of tall cell cytomorphology. In our cohort, among the 119 tumors meeting the original morphologic diagnostic criteria of NIFTP (i.e., <1% of papillae), one (0.8%) was positive for BRAF^V600E. This tumor was a 0.2-cm well-circumscribed tumor completely devoid of papillae and tall cell cytomorphology on multiple hematoxylin and eosin levels examined. It is possible that papillae had not yet developed, given the very small size of the tumor. Whatever the reason, we fully agree with the statement by some of the authors of the NIFTP consensus group that BRAF^V600E immunopositivity should lead to an exhaustive search for papillae and invasion (12). However, as stated by the same authors, it cannot be used solely to exclude NIFTP, which is a morphologic diagnosis (12).

BRAF^V600E and NRAS^Q61R IHC have been shown to be highly sensitive and specific in detecting underlying mutations in thyroid tumors (39,40). An interesting finding of our study is that NRAS^Q61R IHC cross-reacts with HRAS^Q61R and KRAS^Q61R mutant proteins in thyroid carcinoma. A similar phenomenon has been previously described in medullary thyroid carcinoma (41) and melanoma (42). Regardless, we demonstrate a strong correlation between BRAF/NRAS IHC status and the route and risk of metastasis. While 14 of 24 (58%) of BRAF^V600E-positive encapsulated PTC developed nodal metastasis, none of the 26 NRAS^Q61R IHC-positive tumors had nodal metastasis. Rather, one of the tumors with 100% follicular growth and distant metastasis at presentation was positive for NRAS^Q61R by IHC and genetic analysis.

The link between BRAF^V600E -mutated tumors and nodal metastasis and between RAS-mutated carcinoma and distant metastasis have been previously reported by Fakhruddin et al. (43) in PTC and by Landa et al. (44) in poorly differentiated thyroid carcinoma. Given these results, BRAF^V600E and NRAS^Q61R IHC may be utilized clinically as an additional tool to predict the route of metastasis and assist clinical decision regarding appropriate follow-up and management. In addition, these IHC stains may help in the diagnosis of tumors that are morphologically bordering on NIFTP. BRAF^V600E immunopositivity, for example, should prompt extensive histopathologic examination for true papillae.

Our cohort also contains 13 tumors (6%) with a significant amount (≥25%) of solid architecture. Although none of these 13 tumors developed nodal metastases, the number is too small to draw any definite conclusion in regard to the impact of solid growth on nodal metastasis in encapsulated PTC.

The study has some limitation. First, it is a single institution study from a large tertiary cancer center and may not reflect the population at large. Second, the number of encapsulated PTC with 1% to 49% of papillae is relatively small as most of the tumors clustered at the two ends of the spectrum (<1% or ≥50% papillae).

In conclusion, in this study, we show that an increased percentage of papillae within an encapsulated PTC is associated with a high risk of nodal metastasis and BRAF^V600E mutation, regardless of invasion status. U-EPTCs have an overall favorable behavior. The only patients who developed distant metastases harbored them at presentation, had 100% follicular growth, were positive for NRAS^Q61R, and had capsular/vascular invasion. Tumors with less than 1% papillae do not develop lymph node metastases irrespective of their invasive status. Nodal disease or recurrence are also absent in encapsulated PTC without invasion with <10% papillary growth. These findings indicate that the original criterion of <1% papillae within the tumor is still sound for a diagnosis of NIFTP. Reinstituting this criterion will spare a carcinoma diagnosis and unnecessary therapy with its side effects on patients who have a really negligible clinical risk.