The role of immunohistochemistry in the detection of vascular invasion in specimens of endoscopic submucosal dissection

First submission on 02/04/13; last submission on 16/06/13; accepted for publication on 17/06/13; published on 20/08/13 1. Master’s in Pathology by Universidade Federal de Minas Gerais (UFMG); pathologist at LAPAC. 2. Graduate student in Medicine at UFMG. 3. Master’s in Infectology and Tropical Medicine by UFMG; professor at Universidade Federal de Ouro Preto (UFOP). 4. Pathologist at Centro de Estudos em Anatomia e Patologia (CEAP). 5. Doctor in Health Sciences by UFMG; adjunct professor at UFMG. 6. Doctor in Pathology by UFMG; adjunct professor at the Anatomy Pathology and Legal Medicine Department-UFMG. ABStrACt


introDuCtion
The endoscopic resection of neoplasms of the gastrointestinal tract (GIT) has been accepted as an alternative to surgery in selected cases (4) .The technique of endoscopic submucosal dissection (ESD) was developed in Japan in order to enable en bloc resection of neoplastic lesions larger than 2 cm.A major advantage of DES is to produce an adequate sample for histological assessment (10,11) .Nevertheless, endoscopic resection, obviously, does not allow the removal or histopathological evaluation of locoregional lymph nodes.In general, in these cases, lymph node staging is performed with imaging methods such as computed tomography (5) .
Regarding the risk of lymph node metastasis, previous studies demonstrated that the presence of neoplastic vascular emboli, either in lymphatic (lymphatic vascular invasion [LVI]) or in blood vessels (blood vascular invasion [SVI]), is deemed to The role of immunohistochemistry in the detection of vascular invasion in specimens of endoscopic submucosal dissection be a positive predictive factor for the occurrence of lymph node metastasis (6) .Therefore, since lymphadenectomy is not performed during the endoscopic resection, the presence of LVI or BVI in ESD specimens should be carefully considered.

oBJECtivE
The aim of this study was to compare the detection of LVI and BVI by using CD34 and D2-40 IHC markers and the analysis of histological pattern by HE in DES specimens.

MEthoD
It is a cross-sectional study of a series of specimens obtained through DES from 30 patients with GIT neoplasias, who underwent this procedure at the Endoscopy Service of Instituto Alfa de Gastroenterologia, Universidade Federal de Minas Gerais-Clinical Hospital (UFMG-CH) during the period of 2008 to 2012.The study was approved by the Research and Ethics Committee of UFMG.By the end of DES, the specimen was carefully recovered with a foreign body extractor.Initially, it was kept in saline solution for several minutes.Subsequently, the piece was stretched and fixed on a cork board, photographed, immersed in a solution of 10% formalin and sent to the pathology department.The specimen was coated with India ink on the deep margin and 2 mm serial sections were performed.All the material was embedded in paraffin blocks, enabling the suitable histopathological analysis of the entire specimen.
The general features of the primary tumor analyzed by HE were the following: histological classification of the tumor according to World Health Organization (WHO), histological differentiation grade, type classification, lateral and deep margins, BVI and LVI and depth of the tumor.
For IHc assessment, we chose the paraffin blocks that best represented deep tumor invasion.From selected paraffin blocks, serial 4 mm histological sections were performed and, subsequently, placed on silanized slides.Monoclonal antibodies were applied: D2-40 (clone D2-40, Dako) and CD34 (clone QBAND/10, Biogenex).The IHC technique was amplified polymer according to the method recommended by the Immunohistochemistry Manual from the Brazilian Society of Pathology (SBP) and the slides were processed with diaminobenzidine (DAB).The background staining was performed with hematoxylin.Negative controls consisted in the replacement of primary antisera for rabbit or goat immunoglobulins of the same primary antibody class.
The presence of BVI or LVI by HE staining was considered positive only when a neoplastic cell embolus was clearly visualized within the vascular lumen lined by clear endothelium.The BVI was defined by the presence of erythrocytes and neoplastic emboli on the vascular lumen, with or without smooth muscle surrounding the endothelium positively stained for CD34.The BVI was identified when the immunohistochemistry was negative for D2-40 and positive for CD34, associated with the presence of erythrocytes in the vascular lumen.For the definition of LVI, neoplastic emboli should be surrounded by endothelium, positively stained for D2-40 and without erythrocytes or smooth muscle.
The investigation database and all statistical calculations were carried out with the following statistical program: Statistical Package for Social Science for Windows (SPSS) 17.0 (SPSS, Inc, Chicago, III).The p value < 0.05 was considered statistically significant.Cohen's Kappa coefficient was deployed to verify if the agreement between different methods of vascular invasion assessment (HE and IHC) was purely random variation.Chi-square test was applied to analyze the association between categorical variables such as the presence of BVI or LVI and depth of the tumor.

General characteristics of the sample
All patients (30) underwent endoscopic submucosal dissection.Table 1 shows the clinicopathological characteristics of the sample.

Assessment of LVI and BVI
The detection of LVI was considerably higher in comparison with BVI.The presence of LVI was detected in six out of 30 cases (20%) by HE and in nine out of 30 cases (30%) by IHC.The BVI was detected in five out of 30 cases (16.7%) by HE and in only one case by IHC (3.3%).

Comparison of the effectiveness of HE and IHC methods in identifying vascular invasion
The comparison between the detection of vascular invasion by HE staining methods and IHC presented some differences.
Concerning LVI, there was no statistically significant difference between the values (p = 0.232).The presence of BVI detected by IHC was lower in comparison with HE (p = 0.272).Kappa coefficient for the diagnostic agreement between both methods was weak for LVI (Kappa = 0.211) and poor for BVI (Kappa = 0.167), as shown in Table 2.

Correlation between clinicopathological variables and LVI and BVI detected by IHC
When we analyze LVI detected by immunohistochemical examination, we may identify a statistically significant relation with the depth of the tumor in layers m1 (p = 0.031) and m3 (p = 0.047), when compared with other levels of tumor depth.
There was no significant correlation between the presence of LVI and the following variables: gender, topography, histological type of tumor and degree of differentiation.As to BVI, all variables demonstrated no correlation with the presence of BVI.

DiSCuSSion
Endoscopic submucosal dissection (ESD) is increasingly deployed in the management of early neoplasias of the gastrointestinal tract.Furthermore, it produces a high-quality Three (50%) out of six cases with LVI detected by HE were not confirmed by IHC.In contrast, we observed LVI by IHC in six sample, allowing a proper pathological evaluation and reliable information on the neoplasia, such as size, histological grade, type of differentiation, depth of invasion, presence of neoplastic cell emboli in blood and lymphatic vessels and the status of surgical resection margins.
The evaluation of these factors allows a more precise diagnosis and T staging of the neoplastic lesion.Moreover, it permits to determine the prognosis and the need for additional procedures.
In the present study, we investigated if the addition of IHC with specific markers (D2-40 and CD34) for standard histopathological evaluation with HE staining affects the detection of LVI and BVI.The yielded results revealed that detection of LVI by HE staining was higher, although not statistically different (p = 0.232), in comparison with IHC.The presence of LVI was detected in six cases by HE and nine by IHC.Among six cases of LVI observed by HE staining, only three were confirmed by IHC, suggesting that there were false-positive results in three cases.Furthermore, IHC allowed the detection of LVI in six patients, which may have gone unnoticed by HE staining.These results generated a low coefficient of agreement (kappa = 0.211) between HE and IHC analysis.This difference was corroborated by other studies that underline the importance of IHC analysis of vascular invasion in advanced neoplasias (1,3,8,11) .
It is worth highlighting that histopathological analysis of slides stained with HE with the aim to detect LVI is prone to these inaccuracies.By using only this routine staining, pathologists may not be able to differentiate retraction artifacts around cell groups or neoplastic glands from real vascular invasion.Cases in which the neoplastic emboli totally fill the vascular lumen are also hindered, inasmuch as their identification without specific immunostaining of the lymphatic endothelium is precluded.Moreover, falsepositive results for LVI may occur when the presence of BVI is misinterpreted as LVI by HE staining.These discrepancies were also observed in other series of studies, mainly focusing on gastric carcinomas (1,11) .
As to the assessment of BVI in histological sections stained by routine HE, we obtained five positive cases.Nevertheless, IHC did not confirm the positivity observed by HE in any of them.Therefore, these five cases correspond to false-positive results for BVI by HE.In contrast, when subjected to IHC, we identified one positive case for BVI among 25 cases considered negative for BVI by conventional HE, thus demonstrating a false-negative case by HE.Despite the apparent discrepancy between HE and IHC findings, there was no statistically significant difference (p = 0.272) in six cases (20%) of the sample.This bias could be attributed to the fact that some among these cases may have been misinterpreted as BVI instead of LVI or they may be tissue retraction surrounding neoplastic cells in HE assessment, inasmuch as this staining does not allow the proper distinction between these factors in all cases (9) .The analysis of tumor invasion depth on the organ walls revealed that most early GIT neoplasias investigated in this study were intramucosal tumors (83.3%) and superficial submucosal (13.3%), hence demonstrating that almost all cases met the criteria established for the indication of SED.Only one case (3.4%) reached the middle third of the submucosal layer (sm2), which was evidenced only histologically.It is worth noting that there was a significant correlation between the depth of tumor invasion (m1 and m3) and the presence of LVI (p = 0.031 and p = 0.047) herein.
In general, it is believed that in cases of early carcinomas there is less likelihood of neoplastic invasion into blood and lymphatic vessels.However, studies on the density of lymphatic network on the wall of the GIT organs, particularly in the normal gastric wall, attest that the concentration of lymph vessels is considerably higher in the muscular mucosa layer, which may be infiltrated in the so-called early neoplasias (8) .These studies corroborate the statistically significant results obtained through the correlation of depth of tumor invasion and LVI in m3 layer.
A statistically significant correlation between depth of tumor invasion in layer m1 and LVI is due to the fact that there were no positive cases of LVI insofar as they are neoplasias restricted to the epithelium (in situ).

ConCLuSion
In conclusion, LVI and BVI are parameters to be systematically analyzed in histopathology due to its considerable prognostic value, determining the choice of complementary therapy after SED.Accordingly, they should be evaluated by more sensitive and specific methods and incorporated into the protocol of routine histopathological examination.
Our results demonstrate that IHC method using the combined markers (CD34 and D2-40) is superior in detecting LVI and BVI when compared with routine HE staining, which may be of great value in clinical practice in order to optimize the accuracy of pathological staging of early endoscopically resected neoplasias.
Unitermos: dissecção endoscópica da submucosa; invasão vascular sanguínea; invasão vascular linfática; CD34; D2-40.rEfErEnCES of them (25% of negative cases), which had not previously been detected by HE.Some examples of LVI not detected by conventional histopathological examination (HE) are shown in Figure as well as a case in which LVI detected by HE was not confirmed by immunohistochemical examination.

tABLE 1 -
Clinicopathological features of the sample WHO: World Health Organization.

tABLE 2 -
Comparison of diagnostic agreement in the detection of LVI and BVI by HE and IHC LVI: lymphatic vascular invasion; BVI: blood vascular invasion; HE: hematoxylin and eosin; IHC: immunohistochemistry.