Virchows Arch DOI 10.1007/s00428-014-1552-3
ORIGINAL ARTICLE
Morphology predicts BRAFV600E mutation in papillary thyroid carcinoma: an interobserver reproducibility study Renu K. Virk & Constantine G. A. Theoharis & Avinash Prasad & David Chhieng & Manju L. Prasad
Received: 13 August 2013 / Revised: 9 January 2014 / Accepted: 2 February 2014 # Springer-Verlag Berlin Heidelberg 2014
Abstract Papillary thyroid carcinomas (PTC) with BRAFV600E mutation are morphologically distinctive. They are typically classic or tall cell variants, show infiltrative borders, and are associated with desmoplasia/fibrosis, psammoma bodies, and well-developed nuclear features of papillary carcinoma. We hypothesize that morphologic features of PTC can help in the prediction of BRAFV600E mutation, and we evaluate the accuracy and the interobserver reproducibility of such prediction. Hematoxylin and eosin-stained sections from 50 PTCs comprising of 26 mutation-positive and 24 mutation-negative tumors were examined. BRAFV600E mutation was predicted correctly in 42/50 tumors (accuracy, 84 %) with 96 % sensitivity, 71 % specificity, and 78 % positive and 94 % negative predictive values (NPV). Subtle nuclear features of PTC (n=10) had the highest (100 %) negative predictive value followed by wellcircumscribed non-infiltrative tumor borders (17/22 mutationnegative tumors, 95 % NPV). The positive predictive value of infiltrative tumor borders (21/28 [75 %] mutation-positive), desmoplasia/fibrosis (23/31 [74 %] mutation-positive), and psammoma bodies (13/20 [65 %] mutation-positive) increased to 100 % when all three features were present (n=8/8 mutationpositive). To assess interobserver reproducibility, two pathologists blinded to the mutational status evaluated 30 PTCs (15 mutation-positive and 15 mutation-negative) after self-training on 10 PTCs with known BRAFV600E mutational status (five mutation-positive and five mutation-negative). The prediction of the mutation was achieved with substantial agreement (κ value, 0.79) and accuracy (25/30, 83 %). This study R. K. Virk : C. G. A. Theoharis : D. Chhieng : M. L. Prasad (*) Department of Pathology, Yale School of Medicine, 310 Cedar Street, LH-108, New Haven, CT 06520, USA e-mail: [email protected] A. Prasad Department of Neurology, Hartford Hospital, 80 Seymour St, Hartford, CT 06102, USA
demonstrates that BRAFV600E mutation in papillary thyroid carcinoma can be predicted on morphology with accuracy and with substantial interobserver agreement. Keywords Papillary thyroid carcinoma . BRAFV600E . Interobserver reproducibility
Introduction Papillary thyroid carcinoma (PTC) is the most common malignancy of the endocrine system. In general, PTCs have excellent overall survival; however, a small subset shows aggressive behavior including recurrence and metastasis. Controversy exists about the role of BRAFV600E mutation in PTC. In cell culture studies of thyroid cancer, the mutation activates the mitogen-activated protein-kinase (MAPK) kinase (MEK)-extracellular signal-regulated kinase (ERK)MAP kinase pathway leading to uncontrolled cell proliferation and invasiveness in the cells [1]. The elevated kinase activity has a transforming effect in NIH3T3 cell lines transfected with BRAF mutants [2]. BRAFV600E mutation has been reported in 28–70 % of PTCs making it the most frequent genetic abnormality in PTC, and is increasingly being utilized to diagnose it in fine needle aspiration cytology [3–6]. Some investigators have reported an association between somatic BRAFV600E mutation and lymph node metastasis, extrathyroidal extension and higher tumor stage, recurrence, persistence, and resistance to radioactive iodine therapy in PTC while others have found no evidence towards it [4, 7–10]. Thus, the value of preoperative BRAFV600E mutational analysis to determine the extent of surgery, use and/or dosage of radioactive iodine therapy, and the intensity of medical surveillance is debatable [11–13]. It has been proposed that targeted therapy against BRAFV600E mutation may be an alternative in radioactive iodine-resistant PTC [14–16]. Given
Virchows Arch
the current state of knowledge, the determination of BRAFV600E mutational status in all PTCs is of dubious value and is expensive. PTCs with and without BRAFV600E mutation show distinctive morphologic features [5, 7, 10, 17, 18]. The mutated PTCs are more likely to have infiltrative tumor borders; tumor-associated stromal reaction including desmoplasia, fibrosis, and sclerosis; well-developed nuclear features of PTC; focal polygonal eosinophilic (plump pink) cells or tall cells; and psammoma bodies. Classic, tall cell, and subcapsular sclerosing variants of PTCs are more frequently associated with BRAFV600E mutation, whereas the follicular variant is not. These differences were also noted in microcarcinomas (tumors≤1 cm) [19]. In this study, we examine these distinctive morphologic features in routine hematoxylin and eosin (H&E)-stained sections, and evaluate their accuracy in predicting BRAFV600E mutation and the interobserver reproducibility of such prediction.
Materials and methods The study commenced following approval from the institutional review board. We defined a set of six morphologic features (Table 1; Figs. 1, 2, 3, 4, and 5) previously found to be significantly associated with the BRAFV600E mutation in PTCs [5, 19]. Classic, follicular, and tall cell variants of PTC were defined per World Health Organization criteria (2004) [20]. We defined the subcapsular sclerosing variant as a predominantly solid PTC with a peripheral subcapsular location, abutting and involving the thyroid pseudocapsule along at least 20 % of the tumor’s circumference, associated with a sclerosing pattern of infiltration, usually with a central stellate scar, and with no or few papillae (Fig. 4). Tumors that did not fall under any of the foregoing categories were classified as other for the purpose of this study. Follicular variants of PTC that were well-circumscribed or encapsulated (non-infiltrative tumor borders) and had subtle nuclear features (Fig. 2) were more likely to be negative for the mutation, whereas tumors that were associated with stromal reaction including desmoplasia and/or fibrosis (Figs. 4 and 5), infiltrative tumor borders, well-developed nuclear features of PTC, tall cells (Fig.3) or polygonal eosinophilic (plump pink) cells (Fig. 1b) and psammoma bodies were likely to test positive for the mutation (Fig. 1a). All H&E-stained sections of PTCs were prospectively evaluated during routine examination of total thyroidectomy or lobectomy specimens received in the Yale pathology laboratory (MLP and RKV). Criteria for inclusion were (1) tumors ≥5 mm as the morphologic features may not be adequately developed in smaller tumors and thus, may be unreliable in predicting the mutational status, and (2) adequacy of tumor tissue for DNA extraction for molecular analysis. Metastatic
or recurrent PTCs and all consultation cases where the primary tumors were not available for histologic evaluation or mutational analysis were excluded. The prediction regarding the presence or absence of BRAFV600E mutation was made prior to the availability of the results of the mutational analysis. Any discrepancy between the two pathologists was resolved by consensus on a double-headed microscope. To assess interobserver variability, a self-training set comprising of a single most representative H&E-stained slide from 10 PTCs was provided to two participating pathologists (DC and CT) along with a list of characteristic histologic features associated with BRAFV600E mutation with their definitions (Table 1) and the mutational status of these tumors. The selftraining set comprised of five PTCs with and five PTCs without BRAFV600E mutation, and included classic (n=2), follicular (n=4), tall cell (n=1), subcapsular sclerosing (n=1), and other (n=2) variants of PTC. After self-training, both pathologists predicted the presence or absence of the BRAFV600E mutation on a test set of 30 PTCs that comprised of one most representative H&E slide from 15 mutation-positive and 15 mutationnegative tumors without prior knowledge of their mutational status. Mutational analysis BRAFV600E mutation was tested by single-strand conformational polymorphism after PCR amplification of DNA extracted from tumor cells in formalin-fixed paraffin-embedded tissue sections as described previously [5, 19]. The tumors were identified and marked on the H&E-stained glass slides by an experienced endocrine pathologist. Five to ten unstained sections (2 of 5-μm thick) from the tumor were deparaffinized and macrodissected using the marked H&E-stained slide as guide. DNA was extracted using Qiagen tissue kit according to the manufacturer’s protocol (Qiagen, Chatsworth, CA, USA). Five to 20 ng of extracted DNA was amplified using 0.2-μM PCR primers flanking the region of T1799A mutation of BRAF (5′ flank primer: CTCTTCATAATGCTTGCTCT GATAGG and 3′ flank primer: TAGTAACTCAGCAGCA TCTCAGG) in a 50-μl PCR reaction solution containing 1× PCR buffer, 0.1 mM dNTP, 1.5 mM MgCl2, and 2.5 units of AmpliTaq Gold DNA polymerase. PCR started with initial denaturation at 95 °C for 8 min, followed by 35 cycles of denaturation at 94 °C for 1 min, annealing at 55 °C for 1 min and synthesis at 72 °C for 2 min, and finished by a final extension at 72 °C for 10 min (ABI Veriti Thermal Cycler, Applied Biosystem, Foster City, CA, USA). The 250 base pair amplification product was analyzed by single-strand conformation polymorphism using 4 μl of the PCR product on MDE non-denaturing gel. Electrophoresis was carried out on ice for 2 h and 45 min at 325 V. The gel was then stained with SYBR Gold (Molecular Probes) 1:10,000 in TE added for 20 min and imaged by BioRad GelDoc UV System (BioRad, Hercules,
Virchows Arch Table 1 Morphologic features of papillary thyroid carcinoma (PTC) associated with BRAFV600E mutation Morphologic features
Variables
Definition
Histologic variant
Classic variant (Fig. 1) Follicular variant (Fig. 2) Tall cell variant
Classic papillary architecture with true papillae (>5 %); may be cystic Exclusively follicular pattern of growth without or 50 % of tumor cells with moderate to abundant eosinophilic cytoplasm and at least twice as tall as wide; well-developed nuclear features, frequent pseudonucleoli; may have trabecular architecture with back to back papillae and slit-like spaces (Fig. 3 shows tall cell features) Peripheral subcapsular/capsular location; ≥20 % of the tumor circumference involving the thyroid capsule, central stellate fibrosis ±; infiltrating borders; classic nuclear features, no or few papillae Any tumor with features other than above At least 5 of 6 nuclear features (enlargement, overlapping, grooves, irregular nuclear membrane, chromatin clearing and pseudoinclusions) present Pseudoinclusions are absent, other nuclear features may be focal, subtle or less well developed Tumor with infiltrative interface with adjacent non-neoplastic thyroid parenchyma Fibroblasts in collagenous (not myxoid) stroma (Figs. 3 and 4) Proliferating fibroblasts in myxoid stroma (Fig. 5) Paucicellular, eosinophilic, dense bundles of collagen Polygonal tumor cells with moderate to abundant homogeneous, eosinophilic cytoplasm, well developed nuclear features of PTC; however, height of tumor cells is less than twice the width (not “tall” enough); may be focal, usually present at the periphery or advancing edge of the tumor (Fig. 1b)
Subcapsular sclerosing variant (Fig. 4)
Nuclear features
Other Well-developed (Fig. 1b) Subtle (Fig.2b)
Infiltrative tumor borders Tumor-associated stromal reaction Polygonal eosinophilic (plump pink) cells
Psammoma bodies
Present or absent Fibrosis Desmoplasia Sclerosis Present or absent
Present or absent
Concentric lamellated calcifications (Fig. 1a)
CA, USA). BRAFV600E mutation was determined by comparing the banding pattern of the test case with BRAFV600E mutation-positive (colonic adenocarcinoma and papillary thyroid carcinoma) and mutation-negative (normal tonsil and benign thyroid nodule) controls.
Morphologic features associated with BRAFV600E mutationpositive and mutation-negative PTCs were compared using the two-tailed Fisher’s exact test (GraphPad InStat 3.1,
GraphPad, San Diego, CA, USA) with α set at ≤0.05. In order to achieve appropriate statistical power, a similar number of mutation-positive and mutation-negative tumors were included. The sensitivity (true positive/true positive+false negative), specificity (true negative/true negative+false negative), accuracy (true positive+true negative/all positive+all negative), positive predictive value (true positive/true positive+false positive), and negative predictive value (true negative/true negative+false negative) for BRAFV600E mutation were calculated for the histologic features individually and collectively. Interobserver agreement for predicting BRAF V600E
Fig. 1 a Papillary thyroid carcinoma, classic variant with papillary architecture, and a psammoma body (arrow). b Higher magnification shows focal polygonal cells (left) with moderate amount of homogeneous eosinophilic cytoplasm (plump pink cells) alongside the usual tumor cells
(right) with well-developed nuclear features of papillary carcinoma including nuclear enlargement, overlapping, chromatin clearing, nuclear membrane irregularity, grooves, and pseudonucleoli (BRAFV600E mutation-positive)
Statistical analysis
Virchows Arch
Fig. 2 a Papillary thyroid carcinoma, follicular variant: wellcircumscribed, thinly encapsulated tumor with predominantly microfollicular architecture with inspissated colloid. b Higher magnification shows subtle nuclear features of papillary carcinoma including some
nuclear enlargement, chromatin clearing, nuclear membrane irregularity, grooves, some loss of polarity but no significant overlapping and no pseudonucleoli (BRAFV600E mutation-negative)
mutation was assessed by percentage agreement and kappa statistics. Kappa (κ) values of ≤0.2, 0.21–0.40, 0.41–0.60, 0.61–0.80, and >0.80 were considered slight, fair, moderate, substantial, and excellent agreement, respectively [21].
Prediction of BRAFV600E mutation in PTCs
The study included 90 PTCs—50 tumors for the prospective prediction of BRAFV600E mutation and 30 PTCs for assessing interobserver agreement after self-training with 10 PTCs.
Twenty-six mutation-positive and 24 mutation-negative tumors (n=50) were accrued for prospective prediction of the mutation. The patients included 42 women and 8 men (5:1) aged 21 to 83 years (mean, 49 years). The tumors ranged in size from 0.5 to 5 cm (mean, 1.6 cm). The distribution of histologic subtypes of PTCs was as follows: 16 classic variant (Fig. 1), 18 follicular variant (Fig. 2), 5 tall cell variants, 5 subcapsular sclerosing type (Fig. 4), and 6 other PTCs that did not meet the definitions of any of the preceding. The association of tumor morphology and BRAFV600E mutation is shown in Table 2. The mutation was not detected
Fig. 3 Papillary thyroid carcinoma with tall cell features showing backto-back arrangement of papillae, tumor cells with moderate amount of eosinophilic cytoplasm, at least a subset of cells are twice as tall as wide (BRAFV600E mutation-positive)
Fig. 4 Papillary thyroid carcinoma, subcapsular sclerosing variant with superficial location on the surface of the thyroid as denoted by black ink, abundant tumor-associated fibrosis and sclerosis (BRAFV600E mutationpositive)
Results
Virchows Arch
Fig. 5 Papillary thyroid carcinoma with tumor-associated desmoplasia (BRAFV600E mutation-positive)
in all follicular variants of PTC except for one (P=0.0001). Infiltrating tumor borders was significantly associated with the mutation while encapsulation well-circumscribed borders was not (P=0.0005). Although 5 of 22 (23 %) encapsulated or well-circumscribed PTCs with non-infiltrative tumor borders harbored the BRAFV600E mutation, these tumors had other histologic features characteristically associated with the mutation, e.g., four tumors were classic variants, had tumorassociated stromal reaction, and had focal plump pink cells (Fig. 1); two tumors had psammoma bodies; and one tumor was classified as follicular variant but had well-developed nuclear features of PTC. The mutation was present in 74 % (23/31) of the PTCs with tumor-associated stromal reaction including desmoplasia, fibrosis, and/or sclerosis in contrast to only 16 % (3/19) of PTCs without stromal reaction (P= 0.0001). Forty of 50 PTCs exhibited well-developed nuclear Table 2 Association of specific morphologic features with BRAFV600E mutation
features of PTC, whereas subtle nuclear features were exclusively seen in mutation-negative follicular variants of PTC (n= 10, 20 %) (P=0.0002). Although presence of tumor cells with moderate to abundant eosinophilic homogeneous cytoplasm (plump pink cells, Fig. 1a) or with tall cell features (Fig. 3), even focally, was significantly associated with BRAFV600E mutation (21/29, 72 %), psammoma bodies were not. Table 3 shows the predictive value of each histologic feature with regards to BRAFV600E mutation. Well-developed nuclear features of PTC were most sensitive (100 %) in predicting the mutation, whereas infiltrating tumor borders was the most specific (71 %); the latter also had the highest positive predictive value (PPV, 75 %). In contrast, subtle nuclear features had 100 % negative predictive value (NPV). Infiltrative tumor borders, tumor-associated stromal reaction, and histologic subtypes of tumors were >80 % accurate individually in predicting the BRAFV600E mutational status. We then asked if a combination of several histologic features could improve the prediction (Table 4). As expected, sensitivity and negative predictive value did not improve with a combination of features. However, the specificity and PPV increased to as high as 100 % in tumors that displayed a combination of infiltrating tumor borders, tumor-associated stromal reaction and psammoma bodies. The presence of other combinations of histologic features, e.g., focal plump pink cells with tumor-associated stromal reaction or with infiltrative tumor borders also improved specificity and PPV to >80 %. We found psammoma bodies to be least useful in predicting BRAFV600E mutation unless associated with infiltrative tumor borders, tumor-associated stromal reaction, and/ or focal plump pink cells. Overall, BRAFV600E mutation was correctly predicted in 42 of 50 PTCs (accuracy, 84 %) prospectively. Among the 26 mutation-positive tumors, correct prediction was made in 25 tumors (96 % sensitivity), whereas among the 24 mutationnegative tumors, correct prediction was made in 17 tumors only (71 % specificity), with a PPV and NPV of 78 and 94 %, respectively. The only mutation-positive tumor that was incorrectly predicted to be negative was a follicular variant of PTC with well-circumscribed, non-infiltrative tumor borders
Morphologic feature
BRAFV600E positive (n=26, 52 %)
BRAFV600E negative (n=24; 48 %)
P value
Histologic subtype (follicular variant, n=18) Encapsulated or well-circumscribed (non-infiltrative) tumor borders (n=22) Infiltrative tumor borders (n=28) Tumor-associated stromal reaction (n=31) Subtle nuclear features of PTC (n=10) Plump pink cells (n=29) Psammoma bodies (n=20)
1 (6 %) 5 (23 %)
17 (94 %) 17 (77 %)
0.0001 0.0005
21 (75 23 (74 0 21 (72 13 (65
7 (25 %) 8 (26 %) 10 (100 %) 8 (28 %) 7 (35 %)
0.0005 0.0001 0.0002 0.0013 0.1588
%) %) %) %)
Virchows Arch Table 3 Predictive value of morphologic features for BRAFV600E mutation
Infiltrative tumor borders Tumor-associated stromal reaction Plump pink cells Histologic variant other than follicular variant Well-developed nuclear features Psammoma body
Sensitivity (%)
Specificity (%)
PPV (%)
NPV (%)
Accuracy (%)
81 88 81 96 100 50
71 67 67 67 41 71
75 74 72 74 65 65
77 84 76 94 100a 57
82 82 74 82 72 60
a
Tumors with subtle nuclear features of papillary thyroid carcinoma (n=10) and with non-infiltrative borders (n=22) had negative predictive values of 100 and 95 %, respectively PPV positive predictive value, NPV negative predictive value
but had well-developed nuclear features of papillary carcinoma. Seven mutation-negative tumors were incorrectly predicted to be positive; all seven showed well-developed nuclear features of papillary carcinoma, six showed psammoma bodies, five each showed infiltrative tumor borders and tumorassociated stromal reaction, and four showed focal plump pink cells. Interobserver agreement Two pathologists predicted the presence or absence of BRAFV600E mutation retrospectively in 30 PTCs without knowledge of the actual mutational status. Both pathologists correctly predicted the mutational status in 25 of 30 PTCs (accuracy, 83 %) with substantial interobserver agreement (κ, 0.79). The correct prediction rate for the presence of mutation was better (n = 14/15, sensitivity 93 %) than for its absence (n=11/15; specificity, 73 %).
Discussion and conclusion Our findings show that BRAFV600E mutational status can be predicted with accuracy (>80 %) and with substantial interobserver agreement (κ, 0.79) in PTCs using a few key morphological features that are appreciable on routine H&E sections,
i.e., histologic subtype, nuclear features, infiltrative tumor borders, tumor-associated stromal reaction, and focal plump pink cells or focal tall cells. In the current study, all histologic features except psammoma bodies demonstrated high sensitivity (>80 %) but limited specificity (41–71 %) in predicting the mutation. The PPV ranged from 65 to 75 %. However, the NPV reached 100 % when subtle nuclear features were present in the follicular variant of PTC. One of the histologic features that predicted the mutation with good accuracy was infiltrative tumor border. Lupi et al. also found absence of tumor capsule to be significantly associated with the presence of BRAFV600E mutation on multivariate analysis [7]. Another histologic feature, tumor-associated stromal reaction, that has been reported to be significantly associated with mutated PTCs, also predicted the mutation with >80 % sensitivity and accuracy [5, 19]. Prevalence of BRAFV600E mutation is not uniform among different histologic subtypes of PTC. The tall cell, classic, and the Warthin-like variants of PTC are more frequently mutated than the follicular variant [5, 7, 17–19, 22]. Consistent with these reports, the current study also shows that the histologic subtyping of PTC is helpful in predicting the mutational status. However, subtyping alone is not sufficient as follicular variants of tumors can be mutation-positive while classic variants may be mutation-negative, albeit less frequently. In
Table 4 Predictive value of combination of morphologic features for BRAFV600E mutation Sensitivity (%)
Specificity (%)
PPV (%)
NPV (%)
Accuracy (%)
Infiltrative tumor borders and stromal reaction
69
75
75
69
72
Infiltrative tumor borders, stromal reaction and psammoma body Infiltrative tumor borders, stromal reaction, and plump pink cells Stromal reaction with plump pink cells Infiltrative tumor borders and plump pink cells Plump pink cells and psammoma bodies Infiltrative tumor borders and psammoma body Stromal reaction with psammoma body
33 54 73 65 53 38 46
100 83 83 83 83 75 75
100 78 83 81 69 62 67
57 63 74 69 54 53 56
64 68 78 74 58 56 60
Virchows Arch
our experience, follicular variants of PTC with welldeveloped nuclear features, infiltrating tumor borders, and tumor-associated stromal reaction tend to be positive for the mutation despite their follicular growth pattern. The association between tumor borders and BRAFV600E mutation has also been reported by others. Eloy et al. reported a higher prevalence of BRAFV600E mutation in poorly circumscribed papillary thyroid carcinomas compared to well-circumscribed tumors [23]. Segregating the PTCs as well-circumscribed noninfiltrative follicular pattern tumors vs. other tumors, we were able to achieve a NPVof 94 % that further increased to 100 % when combined with subtle nuclear features; the latter virtually excludes BRAFV600E mutation. Interestingly, the accuracy of predicting BRAFV600E mutation was similar when two pathologists predicted the mutations retrospectively in 30 tumors, and another two pathologists predicted it prospectively in an additional 50 PTCs (83 vs. 84 % respectively). This highlights the robustness of the histologic features outlined in the current study in predicting the mutation. We are cognizant of the fact that this is a singleinstitution study, and pathologists at the same institution tend to apply similar morphologic criteria in practice that may reflect in better interobserver agreement. The prediction of BRAFV600E mutation may also be affected by the accuracy of molecular techniques used for detecting the mutation. In our practice, the pathologist responsible for the final diagnosis selects the specific area of tumor on the H&E-stained slide for DNA extraction ensuring tumor representation. We cannot overemphasize this critical step, as poor tumor representation in the extracted DNA, e.g., selection of the wrong nodule by a less-trained person may lead to falsenegative results on mutational analysis. A mouse monoclonal antibody to the BRAF V600E-mutated protein, clone VE1 has recently become available for immunohistochemical detection of the mutation [24]. Although this antibody is reportedly highly specific and sensitive in detecting the mutation, the staining reaction may be weak and difficult to interpret (personal communication). There are other logistical issues, e.g., most immunohistochemistry laboratories do not offer this antibody routinely requiring a send-out to a specialized laboratory, and finally, it does not replace the molecular testing for BRAFV600E mutation. In conclusion, we propose a simple and well-defined set of morphologic criteria that are easily learned and applied to routinely stained H&E sections of PTC in order to predict their BRAFV600E mutational status with >80 % accuracy and with substantial interobserver reproducibility. While these morphologic criteria do not replace molecular analysis for the mutation, they may help in triaging PTCs for such testing, and thus limiting the cost. This could be especially helpful in multifocal PTCs when the most appropriate tumor needs to be selected for molecular testing.
Disclosure A part of this study was presented as an abstract (poster) at the annual meeting of the United States and Canadian Academy of Pathology, Vancouver, 2012. Conflict of interest The authors have no conflict of interest to report.
References 1. Baquero P, Sanchez-Hernandez I, Jimenez-Mora E, Orgaz JL, Jimenez B, Chiloeches A (2013) (V600E)BRAF promotes invasiveness of thyroid cancer cells by decreasing E-cadherin expression through a snail-dependent mechanism. Cancer Lett 335:232–241 2. Davies H, Bignell GR, Cox C, Stephens P, Edkins S, Clegg S, Teague J, Woffendin H, Garnett MJ, Bottomley W, Davis N, Dicks E, Ewing R, Floyd Y, Gray K, Hall S, Hawes R, Hughes J, Kosmidou V, Menzies A, Mould C, Parker A, Stevens C, Watt S, Hooper S, Wilson R, Jayatilake H, Gusterson BA, Cooper C, Shipley J, Hargrave D, Pritchard-Jones K, Maitland N, Chenevix-Trench G, Riggins GJ, Bigner DD, Palmieri G, Cossu A, Flanagan A, Nicholson A, Ho JW, Leung SY, Yuen ST, Weber BL, Seigler HF, Darrow TL, Paterson H, Marais R, Marshall CJ, Wooster R, Stratton MR, Futreal PA (2002) Mutations of the BRAF gene in human cancer. Nature 417:949–954 3. Namba H, Nakashima M, Hayashi T, Hayashida N, Maeda S, Rogounovitch TI, Ohtsuru A, Saenko VA, Kanematsu T, Yamashita S (2003) Clinical implication of hot spot BRAF mutation, V599E, in papillary thyroid cancers. J Clin Endocrinol Metab 88:4393–4397 4. Rivera M, Ricarte-Filho J, Tuttle RM, Ganly I, Shaha A, Knauf J, Fagin J, Ghossein R (2010) Molecular, morphologic, and outcome analysis of thyroid carcinomas according to degree of extrathyroid extension. Thyroid 20:1085–1093 5. Finkelstein A, Levy GH, Hui P, Prasad A, Virk R, Chhieng DC, Carling T, Roman SA, Sosa JA, Udelsman R, Theoharis CG, Prasad ML (2012) Papillary thyroid carcinomas with and without BRAF V600E mutations are morphologically distinct. Histopathology 60: 1052–1059 6. Soares P, Trovisco V, Rocha AS, Lima J, Castro P, Preto A, Maximo V, Botelho T, Seruca R, Sobrinho-Simoes M (2003) BRAF mutations and RET/PTC rearrangements are alternative events in the etiopathogenesis of PTC. Oncogene 22:4578–4580 7. Lupi C, Giannini R, Ugolini C, Proietti A, Berti P, Minuto M, Materazzi G, Elisei R, Santoro M, Miccoli P, Basolo F (2007) Association of BRAF V600E mutation with poor clinicopathological outcomes in 500 consecutive cases of papillary thyroid carcinoma. J Clin Endocrinol Metab 92:4085–4090 8. Durante C, Puxeddu E, Ferretti E, Morisi R, Moretti S, Bruno R, Barbi F, Avenia N, Scipioni A, Verrienti A, Tosi E, Cavaliere A, Gulino A, Filetti S, Russo D (2007) BRAF mutations in papillary thyroid carcinomas inhibit genes involved in iodine metabolism. J Clin Endocrinol Metab 92:2840–2843 9. Kim TH, Park YJ, Lim JA, Ahn HY, Lee EK, Lee YJ, Kim KW, Hahn SK, Youn YK, Kim KH, Cho BY, do Park J (2012) The association of the BRAF(V600E) mutation with prognostic factors and poor clinical outcome in papillary thyroid cancer: a metaanalysis. Cancer 118:1764–1773 10. Eloy C, Santos J, Soares P, Sobrinho-Simoes M (2011) The preeminence of growth pattern and invasiveness and the limited influence of BRAF and RAS mutations in the occurrence of papillary thyroid carcinoma lymph node metastases. Virchows Arch 459:265–276 11. Yip L, Nikiforova MN, Carty SE, Yim JH, Stang MT, Tublin MJ, Lebeau SO, Hodak SP, Ogilvie JB, Nikiforov YE (2009) Optimizing surgical treatment of papillary thyroid carcinoma associated with BRAF mutation. Surgery 146:1215–1223
Virchows Arch 12. Xing M, Clark D, Guan H, Ji M, Dackiw A, Carson KA, Kim M, Tufaro A, Ladenson P, Zeiger M, Tufano R (2009) BRAF mutation testing of thyroid fine-needle aspiration biopsy specimens for preoperative risk stratification in papillary thyroid cancer. J Clin Oncol 27: 2977–2982 13. Sarne DH (2012) A piece of the puzzle: what does BRAF status mean in the management of patients with papillary thyroid carcinoma? J Clin Endocrinol Metab 97:3094–3096 14. Salvatore G, De Falco V, Salerno P, Nappi TC, Pepe S, Troncone G, Carlomagno F, Melillo RM, Wilhelm SM, Santoro M (2006) BRAF is a therapeutic target in aggressive thyroid carcinoma. Clin Cancer Res 12:1623–1629 15. Kloos RT, Ringel MD, Knopp MV, Hall NC, King M, Stevens R, Liang J, Wakely PE Jr, Vasko VV, Saji M, Rittenberry J, Wei L, Arbogast D, Collamore M, Wright JJ, Grever M, Shah MH (2009) Phase II trial of sorafenib in metastatic thyroid cancer. J Clin Oncol 27:1675–1684 16. Gupta-Abramson V, Troxel AB, Nellore A, Puttaswamy K, Redlinger M, Ransone K, Mandel SJ, Flaherty KT, Loevner LA, O’Dwyer PJ, Brose MS (2008) Phase II trial of sorafenib in advanced thyroid cancer. J Clin Oncol 26:4714–4719 17. Trovisco V, Vieira de Castro I, Soares P, Maximo V, Silva P, Magalhaes J, Abrosimov A, Guiu XM, Sobrinho-Simoes M (2004) BRAF mutations are associated with some histological types of papillary thyroid carcinoma. J Pathol 202:247–251 18. Adeniran AJ, Zhu Z, Gandhi M, Steward DL, Fidler JP, Giordano TJ, Biddinger PW, Nikiforov YE (2006) Correlation between genetic
19.
20.
21. 22.
23.
24.
alterations and microscopic features, clinical manifestations, and prognostic characteristics of thyroid papillary carcinomas. Am J Surg Pathol 30:216–222 Virk RK, Van Dyke AL, Finkelstein A, Prasad A, Gibson J, Hui P, Theoharis CG, Carling T, Roman SA, Sosa JA, Udelsman R, Prasad ML (2013) BRAFV600E mutation in papillary thyroid microcarcinoma: a genotype-phenotype correlation. Mod Pathol 26: 62–70 DeLellis RA, Lloyd R, Heitz PU, Eng C (eds) (2004) World Health Organization classification of tumours. Pathology and genetics of tumours of endocrine organs. IARC Press, Lyon Carpenter CR (2005) Kappa statistic. CMAJ 173:15–16, author reply 17 Basolo F, Torregrossa L, Giannini R, Miccoli M, Lupi C, Sensi E, Berti P, Elisei R, Vitti P, Baggiani A, Miccoli P (2010) Correlation between the BRAF V600E mutation and tumor invasiveness in papillary thyroid carcinomas smaller than 20 millimeters: analysis of 1,060 cases. J Clin Endocrinol Metab 95:4197–4205 Eloy C, Santos J, Cameselle-Teijeiro J, Soares P, Sobrinho-Simoes M (2012) TGF-beta/SMAD pathway and BRAF mutation play different roles in circumscribed and infiltrative papillary thyroid carcinoma. Virchows Arch 460:587–600 Capper D, Preusser M, Habel A, Sahm F, Ackermann U, Schindler G, Pusch S, Mechtersheimer G, Zentgraf H, von Deimling A (2011) Assessment of BRAF V600E mutation status by immunohistochemistry with a mutation-specific monoclonal antibody. Acta Neuropathol 122:11–19
ORIGINAL ARTICLE
Morphology predicts BRAFV600E mutation in papillary thyroid carcinoma: an interobserver reproducibility study Renu K. Virk & Constantine G. A. Theoharis & Avinash Prasad & David Chhieng & Manju L. Prasad
Received: 13 August 2013 / Revised: 9 January 2014 / Accepted: 2 February 2014 # Springer-Verlag Berlin Heidelberg 2014
Abstract Papillary thyroid carcinomas (PTC) with BRAFV600E mutation are morphologically distinctive. They are typically classic or tall cell variants, show infiltrative borders, and are associated with desmoplasia/fibrosis, psammoma bodies, and well-developed nuclear features of papillary carcinoma. We hypothesize that morphologic features of PTC can help in the prediction of BRAFV600E mutation, and we evaluate the accuracy and the interobserver reproducibility of such prediction. Hematoxylin and eosin-stained sections from 50 PTCs comprising of 26 mutation-positive and 24 mutation-negative tumors were examined. BRAFV600E mutation was predicted correctly in 42/50 tumors (accuracy, 84 %) with 96 % sensitivity, 71 % specificity, and 78 % positive and 94 % negative predictive values (NPV). Subtle nuclear features of PTC (n=10) had the highest (100 %) negative predictive value followed by wellcircumscribed non-infiltrative tumor borders (17/22 mutationnegative tumors, 95 % NPV). The positive predictive value of infiltrative tumor borders (21/28 [75 %] mutation-positive), desmoplasia/fibrosis (23/31 [74 %] mutation-positive), and psammoma bodies (13/20 [65 %] mutation-positive) increased to 100 % when all three features were present (n=8/8 mutationpositive). To assess interobserver reproducibility, two pathologists blinded to the mutational status evaluated 30 PTCs (15 mutation-positive and 15 mutation-negative) after self-training on 10 PTCs with known BRAFV600E mutational status (five mutation-positive and five mutation-negative). The prediction of the mutation was achieved with substantial agreement (κ value, 0.79) and accuracy (25/30, 83 %). This study R. K. Virk : C. G. A. Theoharis : D. Chhieng : M. L. Prasad (*) Department of Pathology, Yale School of Medicine, 310 Cedar Street, LH-108, New Haven, CT 06520, USA e-mail: [email protected] A. Prasad Department of Neurology, Hartford Hospital, 80 Seymour St, Hartford, CT 06102, USA
demonstrates that BRAFV600E mutation in papillary thyroid carcinoma can be predicted on morphology with accuracy and with substantial interobserver agreement. Keywords Papillary thyroid carcinoma . BRAFV600E . Interobserver reproducibility
Introduction Papillary thyroid carcinoma (PTC) is the most common malignancy of the endocrine system. In general, PTCs have excellent overall survival; however, a small subset shows aggressive behavior including recurrence and metastasis. Controversy exists about the role of BRAFV600E mutation in PTC. In cell culture studies of thyroid cancer, the mutation activates the mitogen-activated protein-kinase (MAPK) kinase (MEK)-extracellular signal-regulated kinase (ERK)MAP kinase pathway leading to uncontrolled cell proliferation and invasiveness in the cells [1]. The elevated kinase activity has a transforming effect in NIH3T3 cell lines transfected with BRAF mutants [2]. BRAFV600E mutation has been reported in 28–70 % of PTCs making it the most frequent genetic abnormality in PTC, and is increasingly being utilized to diagnose it in fine needle aspiration cytology [3–6]. Some investigators have reported an association between somatic BRAFV600E mutation and lymph node metastasis, extrathyroidal extension and higher tumor stage, recurrence, persistence, and resistance to radioactive iodine therapy in PTC while others have found no evidence towards it [4, 7–10]. Thus, the value of preoperative BRAFV600E mutational analysis to determine the extent of surgery, use and/or dosage of radioactive iodine therapy, and the intensity of medical surveillance is debatable [11–13]. It has been proposed that targeted therapy against BRAFV600E mutation may be an alternative in radioactive iodine-resistant PTC [14–16]. Given
Virchows Arch
the current state of knowledge, the determination of BRAFV600E mutational status in all PTCs is of dubious value and is expensive. PTCs with and without BRAFV600E mutation show distinctive morphologic features [5, 7, 10, 17, 18]. The mutated PTCs are more likely to have infiltrative tumor borders; tumor-associated stromal reaction including desmoplasia, fibrosis, and sclerosis; well-developed nuclear features of PTC; focal polygonal eosinophilic (plump pink) cells or tall cells; and psammoma bodies. Classic, tall cell, and subcapsular sclerosing variants of PTCs are more frequently associated with BRAFV600E mutation, whereas the follicular variant is not. These differences were also noted in microcarcinomas (tumors≤1 cm) [19]. In this study, we examine these distinctive morphologic features in routine hematoxylin and eosin (H&E)-stained sections, and evaluate their accuracy in predicting BRAFV600E mutation and the interobserver reproducibility of such prediction.
Materials and methods The study commenced following approval from the institutional review board. We defined a set of six morphologic features (Table 1; Figs. 1, 2, 3, 4, and 5) previously found to be significantly associated with the BRAFV600E mutation in PTCs [5, 19]. Classic, follicular, and tall cell variants of PTC were defined per World Health Organization criteria (2004) [20]. We defined the subcapsular sclerosing variant as a predominantly solid PTC with a peripheral subcapsular location, abutting and involving the thyroid pseudocapsule along at least 20 % of the tumor’s circumference, associated with a sclerosing pattern of infiltration, usually with a central stellate scar, and with no or few papillae (Fig. 4). Tumors that did not fall under any of the foregoing categories were classified as other for the purpose of this study. Follicular variants of PTC that were well-circumscribed or encapsulated (non-infiltrative tumor borders) and had subtle nuclear features (Fig. 2) were more likely to be negative for the mutation, whereas tumors that were associated with stromal reaction including desmoplasia and/or fibrosis (Figs. 4 and 5), infiltrative tumor borders, well-developed nuclear features of PTC, tall cells (Fig.3) or polygonal eosinophilic (plump pink) cells (Fig. 1b) and psammoma bodies were likely to test positive for the mutation (Fig. 1a). All H&E-stained sections of PTCs were prospectively evaluated during routine examination of total thyroidectomy or lobectomy specimens received in the Yale pathology laboratory (MLP and RKV). Criteria for inclusion were (1) tumors ≥5 mm as the morphologic features may not be adequately developed in smaller tumors and thus, may be unreliable in predicting the mutational status, and (2) adequacy of tumor tissue for DNA extraction for molecular analysis. Metastatic
or recurrent PTCs and all consultation cases where the primary tumors were not available for histologic evaluation or mutational analysis were excluded. The prediction regarding the presence or absence of BRAFV600E mutation was made prior to the availability of the results of the mutational analysis. Any discrepancy between the two pathologists was resolved by consensus on a double-headed microscope. To assess interobserver variability, a self-training set comprising of a single most representative H&E-stained slide from 10 PTCs was provided to two participating pathologists (DC and CT) along with a list of characteristic histologic features associated with BRAFV600E mutation with their definitions (Table 1) and the mutational status of these tumors. The selftraining set comprised of five PTCs with and five PTCs without BRAFV600E mutation, and included classic (n=2), follicular (n=4), tall cell (n=1), subcapsular sclerosing (n=1), and other (n=2) variants of PTC. After self-training, both pathologists predicted the presence or absence of the BRAFV600E mutation on a test set of 30 PTCs that comprised of one most representative H&E slide from 15 mutation-positive and 15 mutationnegative tumors without prior knowledge of their mutational status. Mutational analysis BRAFV600E mutation was tested by single-strand conformational polymorphism after PCR amplification of DNA extracted from tumor cells in formalin-fixed paraffin-embedded tissue sections as described previously [5, 19]. The tumors were identified and marked on the H&E-stained glass slides by an experienced endocrine pathologist. Five to ten unstained sections (2 of 5-μm thick) from the tumor were deparaffinized and macrodissected using the marked H&E-stained slide as guide. DNA was extracted using Qiagen tissue kit according to the manufacturer’s protocol (Qiagen, Chatsworth, CA, USA). Five to 20 ng of extracted DNA was amplified using 0.2-μM PCR primers flanking the region of T1799A mutation of BRAF (5′ flank primer: CTCTTCATAATGCTTGCTCT GATAGG and 3′ flank primer: TAGTAACTCAGCAGCA TCTCAGG) in a 50-μl PCR reaction solution containing 1× PCR buffer, 0.1 mM dNTP, 1.5 mM MgCl2, and 2.5 units of AmpliTaq Gold DNA polymerase. PCR started with initial denaturation at 95 °C for 8 min, followed by 35 cycles of denaturation at 94 °C for 1 min, annealing at 55 °C for 1 min and synthesis at 72 °C for 2 min, and finished by a final extension at 72 °C for 10 min (ABI Veriti Thermal Cycler, Applied Biosystem, Foster City, CA, USA). The 250 base pair amplification product was analyzed by single-strand conformation polymorphism using 4 μl of the PCR product on MDE non-denaturing gel. Electrophoresis was carried out on ice for 2 h and 45 min at 325 V. The gel was then stained with SYBR Gold (Molecular Probes) 1:10,000 in TE added for 20 min and imaged by BioRad GelDoc UV System (BioRad, Hercules,
Virchows Arch Table 1 Morphologic features of papillary thyroid carcinoma (PTC) associated with BRAFV600E mutation Morphologic features
Variables
Definition
Histologic variant
Classic variant (Fig. 1) Follicular variant (Fig. 2) Tall cell variant
Classic papillary architecture with true papillae (>5 %); may be cystic Exclusively follicular pattern of growth without or 50 % of tumor cells with moderate to abundant eosinophilic cytoplasm and at least twice as tall as wide; well-developed nuclear features, frequent pseudonucleoli; may have trabecular architecture with back to back papillae and slit-like spaces (Fig. 3 shows tall cell features) Peripheral subcapsular/capsular location; ≥20 % of the tumor circumference involving the thyroid capsule, central stellate fibrosis ±; infiltrating borders; classic nuclear features, no or few papillae Any tumor with features other than above At least 5 of 6 nuclear features (enlargement, overlapping, grooves, irregular nuclear membrane, chromatin clearing and pseudoinclusions) present Pseudoinclusions are absent, other nuclear features may be focal, subtle or less well developed Tumor with infiltrative interface with adjacent non-neoplastic thyroid parenchyma Fibroblasts in collagenous (not myxoid) stroma (Figs. 3 and 4) Proliferating fibroblasts in myxoid stroma (Fig. 5) Paucicellular, eosinophilic, dense bundles of collagen Polygonal tumor cells with moderate to abundant homogeneous, eosinophilic cytoplasm, well developed nuclear features of PTC; however, height of tumor cells is less than twice the width (not “tall” enough); may be focal, usually present at the periphery or advancing edge of the tumor (Fig. 1b)
Subcapsular sclerosing variant (Fig. 4)
Nuclear features
Other Well-developed (Fig. 1b) Subtle (Fig.2b)
Infiltrative tumor borders Tumor-associated stromal reaction Polygonal eosinophilic (plump pink) cells
Psammoma bodies
Present or absent Fibrosis Desmoplasia Sclerosis Present or absent
Present or absent
Concentric lamellated calcifications (Fig. 1a)
CA, USA). BRAFV600E mutation was determined by comparing the banding pattern of the test case with BRAFV600E mutation-positive (colonic adenocarcinoma and papillary thyroid carcinoma) and mutation-negative (normal tonsil and benign thyroid nodule) controls.
Morphologic features associated with BRAFV600E mutationpositive and mutation-negative PTCs were compared using the two-tailed Fisher’s exact test (GraphPad InStat 3.1,
GraphPad, San Diego, CA, USA) with α set at ≤0.05. In order to achieve appropriate statistical power, a similar number of mutation-positive and mutation-negative tumors were included. The sensitivity (true positive/true positive+false negative), specificity (true negative/true negative+false negative), accuracy (true positive+true negative/all positive+all negative), positive predictive value (true positive/true positive+false positive), and negative predictive value (true negative/true negative+false negative) for BRAFV600E mutation were calculated for the histologic features individually and collectively. Interobserver agreement for predicting BRAF V600E
Fig. 1 a Papillary thyroid carcinoma, classic variant with papillary architecture, and a psammoma body (arrow). b Higher magnification shows focal polygonal cells (left) with moderate amount of homogeneous eosinophilic cytoplasm (plump pink cells) alongside the usual tumor cells
(right) with well-developed nuclear features of papillary carcinoma including nuclear enlargement, overlapping, chromatin clearing, nuclear membrane irregularity, grooves, and pseudonucleoli (BRAFV600E mutation-positive)
Statistical analysis
Virchows Arch
Fig. 2 a Papillary thyroid carcinoma, follicular variant: wellcircumscribed, thinly encapsulated tumor with predominantly microfollicular architecture with inspissated colloid. b Higher magnification shows subtle nuclear features of papillary carcinoma including some
nuclear enlargement, chromatin clearing, nuclear membrane irregularity, grooves, some loss of polarity but no significant overlapping and no pseudonucleoli (BRAFV600E mutation-negative)
mutation was assessed by percentage agreement and kappa statistics. Kappa (κ) values of ≤0.2, 0.21–0.40, 0.41–0.60, 0.61–0.80, and >0.80 were considered slight, fair, moderate, substantial, and excellent agreement, respectively [21].
Prediction of BRAFV600E mutation in PTCs
The study included 90 PTCs—50 tumors for the prospective prediction of BRAFV600E mutation and 30 PTCs for assessing interobserver agreement after self-training with 10 PTCs.
Twenty-six mutation-positive and 24 mutation-negative tumors (n=50) were accrued for prospective prediction of the mutation. The patients included 42 women and 8 men (5:1) aged 21 to 83 years (mean, 49 years). The tumors ranged in size from 0.5 to 5 cm (mean, 1.6 cm). The distribution of histologic subtypes of PTCs was as follows: 16 classic variant (Fig. 1), 18 follicular variant (Fig. 2), 5 tall cell variants, 5 subcapsular sclerosing type (Fig. 4), and 6 other PTCs that did not meet the definitions of any of the preceding. The association of tumor morphology and BRAFV600E mutation is shown in Table 2. The mutation was not detected
Fig. 3 Papillary thyroid carcinoma with tall cell features showing backto-back arrangement of papillae, tumor cells with moderate amount of eosinophilic cytoplasm, at least a subset of cells are twice as tall as wide (BRAFV600E mutation-positive)
Fig. 4 Papillary thyroid carcinoma, subcapsular sclerosing variant with superficial location on the surface of the thyroid as denoted by black ink, abundant tumor-associated fibrosis and sclerosis (BRAFV600E mutationpositive)
Results
Virchows Arch
Fig. 5 Papillary thyroid carcinoma with tumor-associated desmoplasia (BRAFV600E mutation-positive)
in all follicular variants of PTC except for one (P=0.0001). Infiltrating tumor borders was significantly associated with the mutation while encapsulation well-circumscribed borders was not (P=0.0005). Although 5 of 22 (23 %) encapsulated or well-circumscribed PTCs with non-infiltrative tumor borders harbored the BRAFV600E mutation, these tumors had other histologic features characteristically associated with the mutation, e.g., four tumors were classic variants, had tumorassociated stromal reaction, and had focal plump pink cells (Fig. 1); two tumors had psammoma bodies; and one tumor was classified as follicular variant but had well-developed nuclear features of PTC. The mutation was present in 74 % (23/31) of the PTCs with tumor-associated stromal reaction including desmoplasia, fibrosis, and/or sclerosis in contrast to only 16 % (3/19) of PTCs without stromal reaction (P= 0.0001). Forty of 50 PTCs exhibited well-developed nuclear Table 2 Association of specific morphologic features with BRAFV600E mutation
features of PTC, whereas subtle nuclear features were exclusively seen in mutation-negative follicular variants of PTC (n= 10, 20 %) (P=0.0002). Although presence of tumor cells with moderate to abundant eosinophilic homogeneous cytoplasm (plump pink cells, Fig. 1a) or with tall cell features (Fig. 3), even focally, was significantly associated with BRAFV600E mutation (21/29, 72 %), psammoma bodies were not. Table 3 shows the predictive value of each histologic feature with regards to BRAFV600E mutation. Well-developed nuclear features of PTC were most sensitive (100 %) in predicting the mutation, whereas infiltrating tumor borders was the most specific (71 %); the latter also had the highest positive predictive value (PPV, 75 %). In contrast, subtle nuclear features had 100 % negative predictive value (NPV). Infiltrative tumor borders, tumor-associated stromal reaction, and histologic subtypes of tumors were >80 % accurate individually in predicting the BRAFV600E mutational status. We then asked if a combination of several histologic features could improve the prediction (Table 4). As expected, sensitivity and negative predictive value did not improve with a combination of features. However, the specificity and PPV increased to as high as 100 % in tumors that displayed a combination of infiltrating tumor borders, tumor-associated stromal reaction and psammoma bodies. The presence of other combinations of histologic features, e.g., focal plump pink cells with tumor-associated stromal reaction or with infiltrative tumor borders also improved specificity and PPV to >80 %. We found psammoma bodies to be least useful in predicting BRAFV600E mutation unless associated with infiltrative tumor borders, tumor-associated stromal reaction, and/ or focal plump pink cells. Overall, BRAFV600E mutation was correctly predicted in 42 of 50 PTCs (accuracy, 84 %) prospectively. Among the 26 mutation-positive tumors, correct prediction was made in 25 tumors (96 % sensitivity), whereas among the 24 mutationnegative tumors, correct prediction was made in 17 tumors only (71 % specificity), with a PPV and NPV of 78 and 94 %, respectively. The only mutation-positive tumor that was incorrectly predicted to be negative was a follicular variant of PTC with well-circumscribed, non-infiltrative tumor borders
Morphologic feature
BRAFV600E positive (n=26, 52 %)
BRAFV600E negative (n=24; 48 %)
P value
Histologic subtype (follicular variant, n=18) Encapsulated or well-circumscribed (non-infiltrative) tumor borders (n=22) Infiltrative tumor borders (n=28) Tumor-associated stromal reaction (n=31) Subtle nuclear features of PTC (n=10) Plump pink cells (n=29) Psammoma bodies (n=20)
1 (6 %) 5 (23 %)
17 (94 %) 17 (77 %)
0.0001 0.0005
21 (75 23 (74 0 21 (72 13 (65
7 (25 %) 8 (26 %) 10 (100 %) 8 (28 %) 7 (35 %)
0.0005 0.0001 0.0002 0.0013 0.1588
%) %) %) %)
Virchows Arch Table 3 Predictive value of morphologic features for BRAFV600E mutation
Infiltrative tumor borders Tumor-associated stromal reaction Plump pink cells Histologic variant other than follicular variant Well-developed nuclear features Psammoma body
Sensitivity (%)
Specificity (%)
PPV (%)
NPV (%)
Accuracy (%)
81 88 81 96 100 50
71 67 67 67 41 71
75 74 72 74 65 65
77 84 76 94 100a 57
82 82 74 82 72 60
a
Tumors with subtle nuclear features of papillary thyroid carcinoma (n=10) and with non-infiltrative borders (n=22) had negative predictive values of 100 and 95 %, respectively PPV positive predictive value, NPV negative predictive value
but had well-developed nuclear features of papillary carcinoma. Seven mutation-negative tumors were incorrectly predicted to be positive; all seven showed well-developed nuclear features of papillary carcinoma, six showed psammoma bodies, five each showed infiltrative tumor borders and tumorassociated stromal reaction, and four showed focal plump pink cells. Interobserver agreement Two pathologists predicted the presence or absence of BRAFV600E mutation retrospectively in 30 PTCs without knowledge of the actual mutational status. Both pathologists correctly predicted the mutational status in 25 of 30 PTCs (accuracy, 83 %) with substantial interobserver agreement (κ, 0.79). The correct prediction rate for the presence of mutation was better (n = 14/15, sensitivity 93 %) than for its absence (n=11/15; specificity, 73 %).
Discussion and conclusion Our findings show that BRAFV600E mutational status can be predicted with accuracy (>80 %) and with substantial interobserver agreement (κ, 0.79) in PTCs using a few key morphological features that are appreciable on routine H&E sections,
i.e., histologic subtype, nuclear features, infiltrative tumor borders, tumor-associated stromal reaction, and focal plump pink cells or focal tall cells. In the current study, all histologic features except psammoma bodies demonstrated high sensitivity (>80 %) but limited specificity (41–71 %) in predicting the mutation. The PPV ranged from 65 to 75 %. However, the NPV reached 100 % when subtle nuclear features were present in the follicular variant of PTC. One of the histologic features that predicted the mutation with good accuracy was infiltrative tumor border. Lupi et al. also found absence of tumor capsule to be significantly associated with the presence of BRAFV600E mutation on multivariate analysis [7]. Another histologic feature, tumor-associated stromal reaction, that has been reported to be significantly associated with mutated PTCs, also predicted the mutation with >80 % sensitivity and accuracy [5, 19]. Prevalence of BRAFV600E mutation is not uniform among different histologic subtypes of PTC. The tall cell, classic, and the Warthin-like variants of PTC are more frequently mutated than the follicular variant [5, 7, 17–19, 22]. Consistent with these reports, the current study also shows that the histologic subtyping of PTC is helpful in predicting the mutational status. However, subtyping alone is not sufficient as follicular variants of tumors can be mutation-positive while classic variants may be mutation-negative, albeit less frequently. In
Table 4 Predictive value of combination of morphologic features for BRAFV600E mutation Sensitivity (%)
Specificity (%)
PPV (%)
NPV (%)
Accuracy (%)
Infiltrative tumor borders and stromal reaction
69
75
75
69
72
Infiltrative tumor borders, stromal reaction and psammoma body Infiltrative tumor borders, stromal reaction, and plump pink cells Stromal reaction with plump pink cells Infiltrative tumor borders and plump pink cells Plump pink cells and psammoma bodies Infiltrative tumor borders and psammoma body Stromal reaction with psammoma body
33 54 73 65 53 38 46
100 83 83 83 83 75 75
100 78 83 81 69 62 67
57 63 74 69 54 53 56
64 68 78 74 58 56 60
Virchows Arch
our experience, follicular variants of PTC with welldeveloped nuclear features, infiltrating tumor borders, and tumor-associated stromal reaction tend to be positive for the mutation despite their follicular growth pattern. The association between tumor borders and BRAFV600E mutation has also been reported by others. Eloy et al. reported a higher prevalence of BRAFV600E mutation in poorly circumscribed papillary thyroid carcinomas compared to well-circumscribed tumors [23]. Segregating the PTCs as well-circumscribed noninfiltrative follicular pattern tumors vs. other tumors, we were able to achieve a NPVof 94 % that further increased to 100 % when combined with subtle nuclear features; the latter virtually excludes BRAFV600E mutation. Interestingly, the accuracy of predicting BRAFV600E mutation was similar when two pathologists predicted the mutations retrospectively in 30 tumors, and another two pathologists predicted it prospectively in an additional 50 PTCs (83 vs. 84 % respectively). This highlights the robustness of the histologic features outlined in the current study in predicting the mutation. We are cognizant of the fact that this is a singleinstitution study, and pathologists at the same institution tend to apply similar morphologic criteria in practice that may reflect in better interobserver agreement. The prediction of BRAFV600E mutation may also be affected by the accuracy of molecular techniques used for detecting the mutation. In our practice, the pathologist responsible for the final diagnosis selects the specific area of tumor on the H&E-stained slide for DNA extraction ensuring tumor representation. We cannot overemphasize this critical step, as poor tumor representation in the extracted DNA, e.g., selection of the wrong nodule by a less-trained person may lead to falsenegative results on mutational analysis. A mouse monoclonal antibody to the BRAF V600E-mutated protein, clone VE1 has recently become available for immunohistochemical detection of the mutation [24]. Although this antibody is reportedly highly specific and sensitive in detecting the mutation, the staining reaction may be weak and difficult to interpret (personal communication). There are other logistical issues, e.g., most immunohistochemistry laboratories do not offer this antibody routinely requiring a send-out to a specialized laboratory, and finally, it does not replace the molecular testing for BRAFV600E mutation. In conclusion, we propose a simple and well-defined set of morphologic criteria that are easily learned and applied to routinely stained H&E sections of PTC in order to predict their BRAFV600E mutational status with >80 % accuracy and with substantial interobserver reproducibility. While these morphologic criteria do not replace molecular analysis for the mutation, they may help in triaging PTCs for such testing, and thus limiting the cost. This could be especially helpful in multifocal PTCs when the most appropriate tumor needs to be selected for molecular testing.
Disclosure A part of this study was presented as an abstract (poster) at the annual meeting of the United States and Canadian Academy of Pathology, Vancouver, 2012. Conflict of interest The authors have no conflict of interest to report.
References 1. Baquero P, Sanchez-Hernandez I, Jimenez-Mora E, Orgaz JL, Jimenez B, Chiloeches A (2013) (V600E)BRAF promotes invasiveness of thyroid cancer cells by decreasing E-cadherin expression through a snail-dependent mechanism. Cancer Lett 335:232–241 2. Davies H, Bignell GR, Cox C, Stephens P, Edkins S, Clegg S, Teague J, Woffendin H, Garnett MJ, Bottomley W, Davis N, Dicks E, Ewing R, Floyd Y, Gray K, Hall S, Hawes R, Hughes J, Kosmidou V, Menzies A, Mould C, Parker A, Stevens C, Watt S, Hooper S, Wilson R, Jayatilake H, Gusterson BA, Cooper C, Shipley J, Hargrave D, Pritchard-Jones K, Maitland N, Chenevix-Trench G, Riggins GJ, Bigner DD, Palmieri G, Cossu A, Flanagan A, Nicholson A, Ho JW, Leung SY, Yuen ST, Weber BL, Seigler HF, Darrow TL, Paterson H, Marais R, Marshall CJ, Wooster R, Stratton MR, Futreal PA (2002) Mutations of the BRAF gene in human cancer. Nature 417:949–954 3. Namba H, Nakashima M, Hayashi T, Hayashida N, Maeda S, Rogounovitch TI, Ohtsuru A, Saenko VA, Kanematsu T, Yamashita S (2003) Clinical implication of hot spot BRAF mutation, V599E, in papillary thyroid cancers. J Clin Endocrinol Metab 88:4393–4397 4. Rivera M, Ricarte-Filho J, Tuttle RM, Ganly I, Shaha A, Knauf J, Fagin J, Ghossein R (2010) Molecular, morphologic, and outcome analysis of thyroid carcinomas according to degree of extrathyroid extension. Thyroid 20:1085–1093 5. Finkelstein A, Levy GH, Hui P, Prasad A, Virk R, Chhieng DC, Carling T, Roman SA, Sosa JA, Udelsman R, Theoharis CG, Prasad ML (2012) Papillary thyroid carcinomas with and without BRAF V600E mutations are morphologically distinct. Histopathology 60: 1052–1059 6. Soares P, Trovisco V, Rocha AS, Lima J, Castro P, Preto A, Maximo V, Botelho T, Seruca R, Sobrinho-Simoes M (2003) BRAF mutations and RET/PTC rearrangements are alternative events in the etiopathogenesis of PTC. Oncogene 22:4578–4580 7. Lupi C, Giannini R, Ugolini C, Proietti A, Berti P, Minuto M, Materazzi G, Elisei R, Santoro M, Miccoli P, Basolo F (2007) Association of BRAF V600E mutation with poor clinicopathological outcomes in 500 consecutive cases of papillary thyroid carcinoma. J Clin Endocrinol Metab 92:4085–4090 8. Durante C, Puxeddu E, Ferretti E, Morisi R, Moretti S, Bruno R, Barbi F, Avenia N, Scipioni A, Verrienti A, Tosi E, Cavaliere A, Gulino A, Filetti S, Russo D (2007) BRAF mutations in papillary thyroid carcinomas inhibit genes involved in iodine metabolism. J Clin Endocrinol Metab 92:2840–2843 9. Kim TH, Park YJ, Lim JA, Ahn HY, Lee EK, Lee YJ, Kim KW, Hahn SK, Youn YK, Kim KH, Cho BY, do Park J (2012) The association of the BRAF(V600E) mutation with prognostic factors and poor clinical outcome in papillary thyroid cancer: a metaanalysis. Cancer 118:1764–1773 10. Eloy C, Santos J, Soares P, Sobrinho-Simoes M (2011) The preeminence of growth pattern and invasiveness and the limited influence of BRAF and RAS mutations in the occurrence of papillary thyroid carcinoma lymph node metastases. Virchows Arch 459:265–276 11. Yip L, Nikiforova MN, Carty SE, Yim JH, Stang MT, Tublin MJ, Lebeau SO, Hodak SP, Ogilvie JB, Nikiforov YE (2009) Optimizing surgical treatment of papillary thyroid carcinoma associated with BRAF mutation. Surgery 146:1215–1223
Virchows Arch 12. Xing M, Clark D, Guan H, Ji M, Dackiw A, Carson KA, Kim M, Tufaro A, Ladenson P, Zeiger M, Tufano R (2009) BRAF mutation testing of thyroid fine-needle aspiration biopsy specimens for preoperative risk stratification in papillary thyroid cancer. J Clin Oncol 27: 2977–2982 13. Sarne DH (2012) A piece of the puzzle: what does BRAF status mean in the management of patients with papillary thyroid carcinoma? J Clin Endocrinol Metab 97:3094–3096 14. Salvatore G, De Falco V, Salerno P, Nappi TC, Pepe S, Troncone G, Carlomagno F, Melillo RM, Wilhelm SM, Santoro M (2006) BRAF is a therapeutic target in aggressive thyroid carcinoma. Clin Cancer Res 12:1623–1629 15. Kloos RT, Ringel MD, Knopp MV, Hall NC, King M, Stevens R, Liang J, Wakely PE Jr, Vasko VV, Saji M, Rittenberry J, Wei L, Arbogast D, Collamore M, Wright JJ, Grever M, Shah MH (2009) Phase II trial of sorafenib in metastatic thyroid cancer. J Clin Oncol 27:1675–1684 16. Gupta-Abramson V, Troxel AB, Nellore A, Puttaswamy K, Redlinger M, Ransone K, Mandel SJ, Flaherty KT, Loevner LA, O’Dwyer PJ, Brose MS (2008) Phase II trial of sorafenib in advanced thyroid cancer. J Clin Oncol 26:4714–4719 17. Trovisco V, Vieira de Castro I, Soares P, Maximo V, Silva P, Magalhaes J, Abrosimov A, Guiu XM, Sobrinho-Simoes M (2004) BRAF mutations are associated with some histological types of papillary thyroid carcinoma. J Pathol 202:247–251 18. Adeniran AJ, Zhu Z, Gandhi M, Steward DL, Fidler JP, Giordano TJ, Biddinger PW, Nikiforov YE (2006) Correlation between genetic
19.
20.
21. 22.
23.
24.
alterations and microscopic features, clinical manifestations, and prognostic characteristics of thyroid papillary carcinomas. Am J Surg Pathol 30:216–222 Virk RK, Van Dyke AL, Finkelstein A, Prasad A, Gibson J, Hui P, Theoharis CG, Carling T, Roman SA, Sosa JA, Udelsman R, Prasad ML (2013) BRAFV600E mutation in papillary thyroid microcarcinoma: a genotype-phenotype correlation. Mod Pathol 26: 62–70 DeLellis RA, Lloyd R, Heitz PU, Eng C (eds) (2004) World Health Organization classification of tumours. Pathology and genetics of tumours of endocrine organs. IARC Press, Lyon Carpenter CR (2005) Kappa statistic. CMAJ 173:15–16, author reply 17 Basolo F, Torregrossa L, Giannini R, Miccoli M, Lupi C, Sensi E, Berti P, Elisei R, Vitti P, Baggiani A, Miccoli P (2010) Correlation between the BRAF V600E mutation and tumor invasiveness in papillary thyroid carcinomas smaller than 20 millimeters: analysis of 1,060 cases. J Clin Endocrinol Metab 95:4197–4205 Eloy C, Santos J, Cameselle-Teijeiro J, Soares P, Sobrinho-Simoes M (2012) TGF-beta/SMAD pathway and BRAF mutation play different roles in circumscribed and infiltrative papillary thyroid carcinoma. Virchows Arch 460:587–600 Capper D, Preusser M, Habel A, Sahm F, Ackermann U, Schindler G, Pusch S, Mechtersheimer G, Zentgraf H, von Deimling A (2011) Assessment of BRAF V600E mutation status by immunohistochemistry with a mutation-specific monoclonal antibody. Acta Neuropathol 122:11–19
