A stratified sample, according to the histological grade (WHO classification) of 70 cases of Oral Epidermoid Carcinoma of 48 (65.7%) males and 24 (34.3%) females patients, with 56/70 (80%) lesions distributed in the 5th, 6th, and 7th decades of life, were reviewed and reassessed with relation to the histological variables presence of keratin, polymorphism nuclear and cellular, mitoses, pattern of invasion, stage (depth) of invasion, and cellular response. Initially, the histological diagnosis of the sample was reviewed with basis in the register number of the material of biopsies, with the objective of stratification of the sample, according to the histological grade. To oral epidermoid carcinomas classified as well differentiated, moderately differentiated and poor differentiated, according to World Health Organization (WHO), were attributed the grades 1, 2, and 3, respectively, (Wahi et al., 1971). The conversion was realized because the histological grades 1, 2, and 3 of WHO classification were similar at to respective grades of the MGHS (tumor cell population). However, conserving the ordinal nature of this evaluation. The initial evaluation of each lesion was realized 4 times, for three observers, at different time (2 times in pilot study and two times in the present study). Afterward, these lesions were evaluated more two times (MGHS) for two observers. Therefore, in the present study, each lesion was evaluated 4 times - two times by the WHO classification and two by MGHS - at different time and without knowledge, at the moment of this evaluation, of the clinic stage (TNM), treatment, or any other clinic information reported in the files from histopathologic diagnosis. While one observer evaluated the material of routine (1 or 2 specimens of various cases and WHO classification), the other evaluated the semi-serial another sections by the MGHS. Afterward, these specimens were changed without any change of information until the final classification of the sample, when then, were listed and compared the gradation histological gives for the two examiners. Although had not been realized the statistic study of interpersonal reproducibility of the variables histological individuals, nor of statistic correlation between these variables and points (score) attributed by each observer; with relation to the histological grade, occurred concordance between them, after the histological classification to have been make individually. The paraffin blocks, contain the material of biopsy, were sectioning by the histologic technique of semi-serial cuts with 6 micrometers (6µm), having been one section processed and 10 cuts (60µm) rejected, until the exhaustion of all included material. After at the histological processing, the cuts were numbered in crescent order, and colouring by hematoxilin and eosin. The study of slides was realized at the Ortholux microscopic, Leitz, at ocular (x12.5) and objectives (x10) and (x20) promoting x125 and x250 augmentation. In the evaluation of the histological variables polymorphism nuclear and mitoses were used the latter augmentation; to others, the former augmentation. For all histological variables, the initial localization of the microscopic fields that would be used was gone with the objective x5, and to the observation of details, objective x20. In the histologic gradation of malignancy of the lesions was used a modification of the multi-factorial system of histological gradation developed by Anneroth et al. (1987). With objective of turning more accurate the evaluation of each histologic variable were done the following modifications: for each prepare histologic was done a casual analyze of a number minimum of 5 microscopic fields, at different levels of profundity; all the series had more 10 sections analyzed; the evaluation of the keratinization or cellular differentiation had as basis the criterion reported by Crissman et al., 1984, who introduced the evaluation of the structure of the keratin pearls in addition to analysis of the percentage of differentiated cells. The presence of well-structured keratin pearls was graduated with the grade 1. The identification of bad-structured keratin pearls received the grade 2; only the visualization of keratinization at individual cells was graded as grade 3; and the minimum presence or absence of keratin as grade 4; in addition to evaluation of the polymorphism nuclear and cellular was partially used the criterion of anaplasia reported by Robbins et al., 1989. The absence of giants multi-nucleate cells, or with unique, voluminous and hypercromatic nucleus received grade 1. The identification of a number minimum (grade 2), intermediary number of cells (grade 3) and many cells (grade 4); the counting of the number of mitoses per microscopic field was only based in the casual quantitative analyze (Lund et al., 1975). The evaluation of the variables pattern and stage (depth) of invasion was considered the more invasive region (profound parts of the specimens and in intermediaries cuts at each series); for the graduation of the stage of invasion was considered the identification of the lamina propria. Always that this structure and/or the superficial portion of the submucosa were identified, was attributed grade was 2, and when not, grade 3. The grades 1 and 4 were not used. The former, for the fact that the lesions with the diagnosis of Carcinoma in situ not to have been identified in the final classification with semi-serial cut, while the latter due to the material of this study to be resulting of incisional biopsy. Therefore, the scores attributed for this variable were 2 (minimum) and 3 (maximum). Were attributed scores 1, 2, 3, and 4 for all other 5 histological variable studied. Always that observed different histological grades, for one determined variable, in one same microscopic field, the greater grade was considered representative of this variable. The "histologic score" - defined as the arithmetic mean of scores attributed for each variable - was always used in the determination of histological grade. Considering that nor always the six variables received the same histologic score or grade, was used the following criterion: grade 1 (histologic score 1.17 or 1.2) - because of the minimum score to variable stage of invasion was 2); grade 2 (histologic scores from 1.33 to 2); grade 3 (histologic scores from 2.17 to 3); and, grade 4 (histologic scores from 3.17 to 4). Histological score between 1 and 2 was considered as of low histologic grade, while histologic scores between greater than 2,17 was considered as of high histologic grade (3 and 4). When one variable might not be evaluated it was excluded of the calculation of the histologic score. The descriptive terms well-differentiate carcinomas was used to the grade 1 (scores 1, 1.17 or 1.2); moderately differentiate, grade 2 (scores 1.4 to 2); poorly differentiate, grades 3 and 4 (scores 2.1 to 4), when referred to WHO classification. The data were assessed quantitatively in ordinal scale by the statistic test of Chi-squared (c2) to identification of contrast (Pimentel-Gomez, 1987, Bulman, 1989). Were established as null hypothesis: the histologic grade is not connected with the histologic technique and histologic grading system of the malignancy of Oral Epidermoid Carcinoma on 3df; there is not association between ulceration and histologic gradation of malignancy on 1df. The value of Chi-squared (c2) lower than 5% of probability (p<0.05) was considered statiscally significant for rejecting the null hypothesis, at a valor greater than 7.82 and 3.84, respectively. At a statistic level greater than 5% (p>0.05) or lower (7.82 and 3.84) than these values, the null hypotheses were confirmed. The results of these tests were confirmed by Excel (Chi-squared test) program, Microsoft.

Fig. 5: Routine and WHO classification:  Histological grade: 3  (histologic score: 2.25). Routine and MGHS:  Histologic grade: 3 (histologic score, 2.8).  

Fig. 6: Semi-serial sectioning and WHO classification:  Histological grade 3 (score histological, 3.0); Semi-serial and MGHS:  Histologic grade 4 (Histologial score: 3.5).

Results                                            Same case

Analyze comparative between histologic techniques Using the routine histological technique and the histological grade suggested by WHO 26 (37.68%) cases received grade 1; 26 (37.68%) cases grade 2; and 17 (24.63) cases. After the semi-serial cuts, using the multi-factorial system of gradation of malignancy, 18 (25.71%) cases received grade 1 (score 1 or 1.2), 34 (48.57%) grade 2 (Score from 1.4 to 2), 13 (18.57%) grade 3 (score from 2.1 to 3), and 5 (7.10%) cases grade 4 (score 3.1 to 4). So, 21 (30%) cases had their histologic grade, with basis in the WHO classification and routine cuts, changed; in 49 (70%) lesions the histological grade stay unalterable (table 4). The statistic test of the Chi-squared proved to be this difference (8.48) statically significant on 3df (r=4 and c=2) (p<0.05). Of these lesions, 1 (4.8%) case with the diagnostic of Carcinona in situ (WHO classification) received the histological grade 1; 10/21 (47.61%) grade histological 2; 5/21 (23.8%) cases grade 3; and 5/21 (23.8%) grade 4. Of the 21 lesions that had alteration in the histological grade, 5/21 (23.8%) lesions was localized in the floor of mouth; 4/21(19.04%) in the tongue, and 3/21 (14.28%) in the inferior gum. In the other anatomic regions of the mouth, in lower number and percentages, distributed themselves the remaining cases (tables 1 and 2). The greater number of cases with alteration in the histological grade occurred in inferior gum and alveolar edge 8/19 (42.1%) cases; floor of the mouth, 5/17 (30%) cases; and tongue, 4/15 (26.6%) cases. The gum with 5/7 (71.4%) cases was the anatomic region that suffers greater increase in the histological gradation. Association between the presence of ulceration and histological grade: The distribution of the lesions according to the histologic grade (tables 1, 2, 3) showed that 51/70 (72.85%) cases presented ulceration, and 19/70 (27.14%) not. Among the latter, 15/19 (78.94 %) lesions presented themselves with the grades 1 and 2. Nine/15 (60%) cases were described as leukoeritroplastic, 2/15 (13.33%) cases as leukoplastic, and 4/15 (26.6%) cases as eritroplastic. Four/19 (21 %) cases of lesions not ulcerated, which presented the histologic grades 3 and 4, possessed leukoplastic aspect at 1/4 (25%) case; eritroplastic at 1/4 (25%) case; and leukoeritroplastic at 2/4 (50%) cases. Among the 51 ulcerated lesions, 37/51 (72.54 %) presented 1 and 2. histological grades With objective of the statistic analysis, two lesions in that was not described the clinic aspect, were considered as ulcerated (table1, 3). Of these, 24/37 (64.86%) lesions were only ulcerated. Among the other lesions ulcerated presenting 1 and 2 histological grade, 3/37 (8.10%) lesions presented leukoplastic component, 5/37 (13.51%) lesions eritroplastic component, and 3/37 (8.10%) lesions leukoeritroplastic component (table 1, 3). Among the 14/51 (27.45 %) ulcerated lesions with histological grades 3 and 4, 4/14 (28.57%) lesions were only ulcerated, 2/14 (14.28%) lesions showed leukoplastic component, 2/14 (14.28%) eritroplastic component and 1/14 (7.1 %) leukoeritroplastic component. With objective of statistic analysis 5 lesions with the clinic aspect did not describe were consider ulcerated (Table 1, 2, and 3). The result of the Chi-squared test (0.38) on 1df (r=2, c=2) was considered statiscally not significant (p>0.05), confirming the null hypothesis, i.e., the presence of ulceration is not influentiated for the histological grade (table 3). Variable differentiation cellular The majority of the lesions 65/70 (92.85%) presented the grades 1, 2 or 3 of keratinization (in individual cells - fig. 2, 4 - or in the pearls form - fig 1,3), while only 5/70 cases (7.14%) showed quantity not significant (0-5%) (fig 5, 6, 7). In the floor of mouth, 53 % of the cases presented moderate keratinization (grade 2), while in the tongue, at 80% of the cases was observed moderated or poor keratinization (grades 3 and 4). The greatest percentages of keratinization (87.5% and 70%) were observed in the cheek and inferior alveolar edge, respectively. The recognition or not of the cells keratinization in individual cells was easy, particularly, at the grades 1 and 4 - grades extremes of the rank - (fig. 1, 3, 6,7), however, at the intermediary grades (2 and 3), the estimation of the percentage was difficult, because usually all the microscopic fields presented different percentages. However, in these cases, the analysis of the structure of the keratin pearls was particularly useful in the distinction between the grades 2 and 3 (fig. 2, 4). With the use of the semi-serial cuts was verified that the keratinization variable presented different histologic grades, not only in the different profundities of one same histological prepare, but also among the different histologic sections at in one same series, having been observed a decreased of the keratinization grade in the profound parts of the specimen (fig. 1, 2, 3, 4). In only one case was observed a decrease at the histologic grade in the intermediary sections with relation to initial sections (fig. 7, 8). Variable nuclear and cellular pleomorfism In 52/70 (74,28%) cases, was identified the grades 1 and 2 (minimum or moderate). In these cases, the alterations more frequently observed were the alteration in the cytoplasm/nuclear ratio and nuclear hypercormatism (fig. 1, 3). Eighteen 18 (25.7%) cases presented histological grade 3 or 4 (accentuate or extreme). These lesions were observed with greater frequency in the gum, tongue, and inferior alveolar edge at 50%, 33.3%, and 30%, respectively. In these histological grades, all the alterations of form, size and colouring were observed. However, the presence of numerous multi-nucleate giants cells or with the unique, voluminous or hypercromatic nucleus were considered to separate the lesions from a small histological grade (1 and 2) of them presenting a rise histological grade (3 and 4) because the former (fig. 1, 3) these cells were not observed, or when observed, showed themselves in small number (fig. 6, 7). Variable mitoses Of 70 cases studied, 18 (25.7%) cases presented from 0 to 1 mitoses per microscopic field (grade 1), at x250 augmentation, while 34 (48.6%) cases presented from 0 to 3 mitoses (grade 2); and 18 (25.7%) cases more than 3 mitoses (grades 3 and 4). In the extremes grades (1 and 4), presenting until 0 or 1 mitoses (grade 1), and presenting more than 5 mitoses (grade 4), this evaluation was easily gone, due to the fact that all microscopic presented similar counting. However, in the intermediary grades (2 and 3), some times, all the counted microscopic fields presented differing counting. In these cases the greatest score was considered representative of the histological grade of the variable. This evaluation was only done quantitatively, independently, if they were typical or atypical. However, the latter were observed with more frequency in the grades 3 and 4 than 1 and 2 (fig 1, 2, 3, 4, 5, 6, 7, 8). Variable pattern of invasion In 52 (74.3%) cases was observed the invasion of the connective tissue by epithelial cells arranged with well-defined border, nest, strip, and solid string (grades 1 and 2) (fig. 1, 3, 4, 5) while 18 (25.3%) was observed invasion kind thin string and/or at the pattern small groups or dissociated individual cells (fig. 2, 4, 6, 7, 8). Variable stage of invasion All the cases showed invasion of the connective tissue by squamous cells. The grades 1 (Carcinoma in situ) and 4 (invasion of the bone) were not observed. The former due to have not been included in this study, while the latter because the material studied was proceeding of incisional biopsies. Therefore, these lesions only received grades 2 and 3. In 52 cases (74.3%) was observed grade 2, i. e., the invasion was until lamina propria and superficial portion of the submucosa (fig. 1, 3, 4, 5). In 18 (25.7%) cases, involved the portion profound of the submucosa (not having been identified lamina propria), in some cases, reached gland, muscle and nerve (fig. 2, 6, 7, 8). Variable linphoplasmocitary infiltrate In the majority of the cases this infiltrate was predominantly linfoplasmocitary, at grade moderate or minimum (grade 2 and 3) (fig 1, 4, 5, 6). Only in few cases were observed grade accentuate or absence of this infiltrate (grade 1 and 4) (fig. 3, 7, 8). In addition these kinds of cells was also identified in smaller number macrophage, and seldom neutrophils and eosinophils, which usually distribute themselves near of areas of ulcerations.

                  Tables>>>>                  

Fig. 7: Routine and WHO classification:  Histological grade: 3  (histologic score: 3). Routine and MGHS: Histological grade: 4 (histological score, 3.8).   

Fig. 8: Semi-serial sectioning and WHO classification:  Histological grade 3 (score histological, 2.25); Semi-serial and MGHS: Histological grade: 4 (histological score, 3.4).

Discussion                                     Same case

Differentiation cellular is the process in that occur the normal transformation from an immature state to the mature, or specialized state. In it, the cells are functionally died and filled by keratin, losing the proliferative activity. Therefore, the percentage of mature cells would indicate the growth potential of the neoplasm. Well-differentiated neoplasm would present slow growth because their cells at the to produce keratin would lose the activity proliferative, while the poorly differentiated, which conserve this potential, would grow quickly. Consequently, the well-differentiated neoplasm would present a biologic behavior less aggressive and better prognostic than the poorly differentiated. However, was verified that the use of a histological system of grading of the malignancy, based only in the differentiation cellular, would determine a erroneous histological grade, usually inferior to the real; because might be observed high nuclear polymorphism and invasion of the connective tissue by individual neoplastic cells, thin strings or small groups connected with an high grade of keratinization (fig.1). In these cases the establishment of a histological grade based only in the cellular proliferation and differentiation would be incorrect. This observation was confirmed in our study, whose histological aspect in many cases presented the difference of 1 histological grade between WHO classification and MGHS (fig. 4, 7, 8). This result might explicate the potential of errors of the routine microscopic exam, based only in the proliferating and differentiation cellular and analysis of one or two sections. Other important aspect of the differentiation cellular is its relation with the presence of metastasis in regional lymph nodes, since the incidence of metastasis is significantly greater in patients with poorly differentiated carcinomas than in well-differentiated carcinomas (Mcgavran et al., 1961). Studies of electronic microscopy proved that at poorly differentiated carcinomas there is a decrease of the number and complexity of the desmosome, as well as that these cells, when in culture, present the tendency to suffer loses of the contact inhibition, presenting a greater probability to metastasize (Shingaki et al., 1988). However, this biologic comportment would be better evaluated for the stage (depth) and pattern of invasion than by its cellular differentiation (Crissman et al., 1984). Although this variable looks to be important to determine the cellular differentiation of the neoplasm, it presents little value in the determination of the survival and presence of metastasis (Crissman et al., 1984). This is particularly important to the lesions of gum and alveolar edge, floor of mouth and tongue, which in our study presented a greater percentage of lesions with alterations in the histologic grade (42.1%, 30%, and 26%, respectively). In spite of us only to have evaluated few cases of oral carcinoma in gum, 5/7 (71.42%) cases suffer increase in the histological grade with semi-serial sectioning. Possibly this occurs because this region is keratinized. So, another studies with a greater number of cases must be realized to confirm this result. The results of our study also proved that 92.4% (grades 1 and 2) presented rise or moderate presence of keratin, while only 7.6 showed minimum keratinization (3 and 4), confirming the results of studies of immunohistochemistry that reported the presence of keratin in all the neoplasm, independently of the histologic grade (Santos-Pinto, 1991). However, in the floor of mouth, 53 % of the cases presented moderate keratinization (grade 2), while in the tongue, at 80% of the cases was observed moderated or poor keratinization (grades 3 and 4). This result contrast with the results of the dissertation realized at the Faculty of Dentistry at the Federal University of Rio Grande do Sul (FO/UFRGS), using the WHO classification, was verified at patients with the clinic stages T1 or T2 a frequency of 76 % (32/41) tongue neoplasm presenting the histologic grades 1 and 2, while at only 9 (21%) cases was verified the histological grade 3. For these patients was noticed that neoplasm in the stages T1 presented more differentiated histopathologic behavior, while T4 showed a less differentiated microscopic aspect (Martins-Neto, 1996), as well as confirm the study of Lund et al., (1975), in which was used the multi-factorial histological system of grading of the malignancy (MGHS), verifying that in the stages T1 and T2 of the carcinomas of tongue, there was 14/28 (50%) cases with the histological grades 3 and 4 (score 2.6 - 4), particularly, because in our study was used semi-serial cuts, which increase the gradation histological of malignance around of 30%, while in the study of Lund et al., (1975) was use routine cuts. Anaplasia is the probable reversion of a cell to an embrionary form whose proliferative activity is elevated. The observation of a high histological grade of anaplasia indicates lose of the differentiation cellular. In this state, the cells are morphologically and functionally changed, however, their proliferative activity is increased. In it, is possible to observe all alterations of form, size and citplasmatic-nuclear coloring (fig. 6, 7, 8). These alterations together with the number of atypical mitoses are histological variables important in the establishment of the malignancy grade, because determine the anaplasia grade (Robbins et al., 1989). Nevertheless, this histological variable, when compared with others, such as the stage (depth) and pattern of invasion, presents lower prognostic valor, since the foresight about the occurrence of metastasis has statistic correlation with the invasion vascular, which also has direct correlation with the pattern and stage of invasion (Crissman et al., 1984). In our study, there was not difficulty in the recognition of nuclear and citoplasmatic alterations. However, its evaluation quantitative, in percentage, was problematic, because so as in the evaluation of the histological variable keratinization as in the evaluation of the anaplasia, the histological grades of these variables differed in the different levels of the specimen, being, in some cases, difficult to distinct between the grades 2 and 3. In addition, this evaluation presents some grade of subjectivity, considering that the increase of the nuclear/citoplasmatic ratio can promote the illusion of the increase of number of cells/microscopic field; as well as factors connected to histological preparation such as the coloration technique may interfere in the analysis of that variable (Anneroth, Hansen, 1987). Although the increase of the mitotic activity may indicate rise of the proliferative rate of a neoplasm, its growth occur due to disturb in the process of cellular differentiation, because the neoplastic cells are not eliminate in the same rate with that are generate, always resulting in increase of the cellular population of the neoplasm (Catanzaro-Guimarâes, 1982). Therefore, the evaluation of the mitotic activity as well as of the polymorphism nuclear would indicate the growth potential of the neoplasm. So far, the presence together of numerous mitoses and high nuclear polymorphism would sign the elevation of the proliferative rate of its cells, and, consequently, at its velocity of growth. Possibly, one neoplasm presenting a high histologic grade to these two histological variables would present an aggressive biologic behavior. In this study, in some cases, was not possible to obtain the same counting of mitoses in all microscopic fields, because, particularly, in the histological grades 2 and 3, all the five fields presented different counting. This possibly is caused by the variable distribution of the mitoses in the different levels of the neoplasm. However, factors such as the delay in the fixation of the biopsies material and thickness of the sectioning can also difficult this evaluation (willén, Erklund, 1985). In spite of the presence of numerous atypical mitoses to be an important histological variable to characterize the malignancy neoplastic (Robbins et al., 1987), and one of the principals histological variables of the anaplastic carcinomas, it presents little value in the distinction of the initial grades of anaplasia, because can be also observed in greater number at lesions of lesser malignancy grade (epithelial dysplasia) than oral epidermoid carcinoma of histological grade 1 (Araújo, 1986). In the present study was impracticable to establish a numeric valor to atypical mitoses capable of separating between these two grades (grade 2 an 3). Approximately 80% of patients with oral epidermoid carcinoma died at consequence of metastasis in regional nodes, in spite of the neoplasm to have bee correctly treated. In many cases this occurs for the presence of sub-clinical metastasis that occurred before or during the treatment. Consequently, the increase of the proliferative cellular activity would suggest an aggressive biologic behavior of the neoplasm, as well as the potential to metastasize, because the increase of the frequency of metastasis has been correlated with the invasion vascular by the neoplastic cells (Boyce et al. 1984, Lund et al. 1976, Poleksic, Kalwaic, 1978). Therefore, the analysis of the invasion vascular is an important histologic variable in the evaluation of the biologic behavior of the neoplasm. Thus, its recognition, when of the exam of the biopsy material, is a important factor to the evaluation about the prognosis and selection of the treatment (Poleksik and Kalwaic, 1978). As was verified, there is a statistic correlation between the histologic variables vascular invasion and pattern of invasion, because there is, a greater possibility of cells individually dissociated penetrate among the endothelial cells (fig. 2, 4, 6, 7) than the big nest of cells with the border well definite (fig. 5). Consequently, the evaluation of the histologic variable pattern of invasion might reflect the prognostic capacity of the vascular invasion variable. However, the prognosis valor of this latter variable has been questioned because was not always possible to observe a number significant of vessels at their histological prepares (Larsen et al., 1978). This same difficulty has been verified in immune-histochemical study, where although had occurred facility in the observation of the vessels, was reported vascular invasion in only 2/43 (4.6%) cases (Santos-Pinto Jr., 1991). In many cases, the difficulty of evaluation in "step" sections (Larsen et al., 1978) is promoted for the rupture of vascular structures, when of the histological sectioning, caused by bad fixation (Willén et al., 1983). In others, has as cause the decreased number of "step" sections observed (Larsen et al., 1978), considering that there is an increase in the frequency of vascular invasion, when are examined more than eight specimens (Boyce et al., 1984). The decreased number of microscopic fields analyzed in each specimen, as well as the decreased number of specimens might have also caused the difficulty reported in the studies of Larsen et al. (1978) and Santos-Pinto Jr. (1991). In spite of the sample studied for us not to have been fixed at a mixture of 10 % formalin with 0.17 mol of phosphate buffer at ph 7.0 (Willén et al., 1983), we observed, using the semi-serial sectioning that the embolization of neoplastic cells is another important histological variable in the establishment of the prognostic of a neoplasm to metastasize. Other histologic variable that is directly connected with the vascular invasion and, therefore, with the frequency of metastasis is the stage (depth) of invasion, because it has been too connected with the rise of the frequency of metastasis, which was of 8.3% in patients with neoplasm presenting the depth of 4 mm, increasing to 83.3% in lesions with more of 8 mm (Shingaki, 1986). However, the importance prognostic of this histological variable has been questioned, because in many cases is not possible to distinguish between the grades 3 and 4 at incisional biopsies (Larsen et al., 1978). This argumentation is only true, in part, because same being this gradation inferior to real, it would cause the reduction of only 0.16 points in the final histological score, which is not sufficient to alter the histological grade, and to remove them of the rank of oral epidermoid carcinomas, presenting high histological grade (score 2.5 to 4); therefore, conserving the prognosis valor of the histological score. Besides, the stage (depth) of invasion may be evaluated in 4 histological grades using the criterion reported by Anneroth and Hansen, 1984, which was, partially, modified fur us. In our study was observed a relation between the pattern and stage of invasion, because in 25.7% of the lesions, the neoplastic cells extended itself under lamina propria and superficial portion of the submucosa, being of the kind small groups, thin strings or at dissociates cells. In some cases, they occurred connected with the presence of keratin (fig. 3, 4, 8). These results suggest an aggressive biologic behavior to lesions presenting these histological variables with the grade 3 and 4. Other reason by which think that the evaluation of the variable stage (depth) of invasion must be realized, same in incisional biopsies, was the fact that the lesion with the diagnosis of carcinoma in situ, after the semi-serial sectioning, to have the diagnosis altered to invasive oral epidermoid carcinoma grade I, confirming that big part of lesions with the histological diagnosis of high epithelial dysplasia or carcinoma in situ are invasive carcinomas needing so only to increase the depth of the histologic sectioning to identify them (Araújo, 1986). The presence of a high inflammatory infiltrate lympho-plasmocytic around the neoplastic cels has correlation with the survival of 5 years or more (Willén and Nathanson, 1973). This infiltrate might indicate the immune response to the neoplasm and, therefore, to be a histologic variable suggestive of a probable involution of the cancer, because it would decrease the number of dysplatic clones and invasion cellular at the connective tissue, retarding the growth of the neoplasm (Noone et al., 1974, Berg, 1971). However, the immune response nor always reaches its objective of protection, and can until same to blockage the resistance anti-neoplastic. In many cases, the presence of these cells has been connected with the inflammatory response to ulceration or necrosis, which could indicate a bad prognosis (Greenough, 1971). The results of our study showed that the great part of the cases presented chronic inflammatory infiltrate constituted by lymphocytes, plasmatic cells and histiocytes. In lower number was identified the presence of neutrophils. This infiltrate might indicate the immune response, which in the majority of the cases does not destroy the neoplasm (fig. 1, 3, 4, 5). Considering that in the material that constitute the present study was identified the presence of ulceration, is possible that in some situations this infiltrate represent the inflammatory response to the ulceration. With the use of semi-serial sectioning and MGHS, Eighteen/70 (25.7) cases were graded as grade 1 (histological score 1 or 1.17); 31 (44.2%) cases, grade 2 (histological score 1.33 to 2); 13 (18.5%) cases grade 3 (histological score 2.17 to 3); and at 5 (7.1%) cases, grade 4 (3,17 to 4). This result contrast with the result of studies, which reported a histological score minimum of 1.8 point (Lund et al., 1975, Anneroth, Hansen, 1984), as well as with the results of immune-histochemistry study, which excluding the histologic stage of invasion parameter, reported, at only one case, the histological score of 1.3 points (H&E), and histological score minimum of 1.8 point in the immune-histochemistry (Pinto-Junior, 1991). According to the histological scores for us used, the histological grade minimum (1.3) reported by Pinto-Junior (1991), in our study, would have been graded as two. In other words, the sample studied would not have lesions graduated with the grade I. In spite of not to have been done statistic analysis of correlation between the histological variables, we observed that, in oral epidermoid carcinomas with histological grade 1 (semi-serial sectioning and MGHS), the histological variables keratinization and polymorphism nuclear, as well as mitoses presented negative correlation. However, the variables polymorphism nuclear and mitoses presented positive correlation, occurring similar increase or decrease of the histological grades in the majority of the cases. In these lesions, all the histological variables of the tumoral cellular population, when evaluated, showed grade 1. As only the histological variable of the evaluation of the relation tumor-normal tissue, which presented greater grade, was the stage of invasion, the attributed histological score was 1.17. If the variable stage of invasion, pattern of invasion or lympho-plasmocytic were excluded, the histological score would have been 1. Possibly these results may be explicated by fact that in sample used in our study there be a high percentage of lesions in anatomic regions where the carcinomas usually present themselves with greater cellular differentiation and a biologic behavior less aggressive (cheek, gum, and alveolar edge). Similar results were verified in study of the epidermoid carcinoma of the palate (Enneroth and Moberger, 1973). The presence of carcinomas in these anatomic localizations might explicate the difference in the histological minimum score verified between our study (1.17) and (1.8) obtained in the studies of Lund et al. (1975), Anneroth and Hansen (1984), whose sample was restrict to carcinomas of tongue and tongue/floor of mouth, which generally present themselves with increase histological score. With relation to the work of Pinto Jr. (1991), realized at the Faculty of Dentistry at the University of Sao Paulo (FO/USP), we do not know to explicate the difference in the histological score minimum between this study (1.8 points) and our results (1 or 1.7 points). Same considering that it was obtained with immune histochemistry coloration; only one case (1.3 points) and two cases (1.4 points) were reported presenting histological score inferior to 1.7 points in H&E. Since the parameter stage of invasion was not evaluated in the Santos-Pinto Jr.' s study (1991), is possible to conclude that this author did not identified neither carcinoma presenting keratinization (score 1), nuclear polymorphism (score 1), mitoses (score 1), pattern of invasion (score 1) and lymph-plasmocytic infiltrate (score 2); or score 2 only in the keratinization or polymorphism cellular and score 1 in the other four histological parameter; nor carcinomas showing keratinization (grade 1), nuclear polimorfism (score 1), mitoses (score 1), pattern of invasion (score 2) and lympho-plamocytic infiltrate (score 1), which present themselves with histological score = 1. In our study, these types of carcinoma, conjunctly with the carcinomas that presented score 1 in all histological parameters evaluated represented 25.70% of the studied sample. As in the study of Santos-Pinto Jr. and Cavalcanti de Araújo (1991), possibly, was not used serial or semi-serial sectioning, the histological scores reported in this study, like us increased with relation to the real histological score of these lesions in H&E, since the serial and semi-serial sectioning increases the score and grade histological around of 20-30% of the studied sample (Schütz, 1992, Rosolen, 1997). So, with the use semi-serial or serial cut, for the seven neoplasm with the score lower than 2 points (possibly grade 1), was expected the elevation of the histological grade, at least, two lesions, which would decrease the number of neoplasm with the grade 1 from 7 to 5 (5/43) 11.42% (carcinomas with histological grade 1). This percentage differs of the 25.70% verified in our study. Another interesting result presented for the professors Décio dos Santos Pinto Jr. and Vera Cavalcanti de Araújo (Master of Science at the Indiana University), at the University of Sao Paulo (FO/USP), was the histological score 1.3 (case 21), because it is not possible when only four or five histological parameters are evaluated. This score is only possible when 3 histological parameters are evaluated (2 parameters with score 1 and 1 parameter with score 2; 4/3=1.3). However, only with the evaluation of these three histological parameters, same not considering the subjectivity of the evaluation of the parameters possibly evaluated (keratinization, cellular polimorphism and mitoses) is equally difficult to understand the score of 2.6 obtained, in this same case (21), using immune-histochemistry coloration. For this, would have occurred the elevation of two grade or points (from 1 to 3) at two of the three histological parameters evaluated, increasing the histological score at 1.3 points. I confess that had difficulty to understand as this occurred, because the identification immune-histochemistry of a histological parameter, for example, of the keratin only turn more easy its identification, however, without alter for more the score of this parameter (from 1 to 3), but decreasing, considering that possibly both studies (H&E and immunehistochemistry) were realized in a similar depth. This technique only is valid for decreasing the histological grade of keratinization from 4 for 3, where is impossible determines it at H&E. In our study, using the modified histological criteria of Anneroth et al (1987) for the analyze of these variables, I do not remember of having verified the increase of more than 1 point or grade in the histological score or grade of each histological variable, except for the parameter keratinization, specially with relation to pearls of keratin (Crissmann et al., 1984), which in the grade 1 was well-structured, while absence in grade 3, at the contrary of the study of Santos-Pinto e Cavalcanti (1991). Usually, the increase of more than 1 grade in the histological grade was caused by the evaluation of parameters connected with the relation neoplasm-tissue of patient, which were not evaluated with the WHO classification. Using semi-serial sectioning, I do also not remember of having verified the diminution of the histological score in any lesion, such as was reported in the histochemistry study of Santos-Pinto Jr. at 9.3% cases (4/43). But these observations, certainly, do not decrease the academic merit of this thesis defended with force and enthusiasm for the professor at the FO/USP and rigorously evaluated at the University of Sao Paulo (FO/USP) with the grade 10 more laud and distinction. Besides, the study of Santos-Pinto Jr. has support on the study of Rosolen, at the Federal University of Rio Grande do Sul (FO/UFRGS), which reported similar result also using semi-serial sectioning in oral leukoplakia. In our study, the histological technique of semi-serial sectioning changed the histological grade of 21/70 (30%) cases of oral epidermoid carcinomas graded by routine cuts and WHO classification. In only one (1.4%) case (1/70) was observed the diminution of the histological grade in the initial and final sections with relation to intermediary sections decreasing the histological score (fig. 7 e 8). In other words, only in this case was verified an increase of cellular differentiation in the intermediary and final sections with relation to initial portion. A possible explication to this strange result might be the fact that the keratin pearls, being spherical structures, in the dependence of their number and of where were sectioned, to have occurred the sectioning in the intermediary portion of a great number of these pearls, at the final series, producing the impression of the increase of the grade of keratinization (figs. 7, 8). Considering the variable distribution of these pearls in the neoplasm, this would be little probable. Other possibility for explicate this our strange case (figs. 7 and 8), might have been the erroneous inclusion or the sectioning of the specimens of biopsy only in the direction anterior, posterior, and not medial, lateral. Because the specimens were block out and again block, anything similar to bad inclusion might have occurred in the study of Rosolen (1997) and Santos-Pinto, (1991). Therefore, in our study, by the fact of this percentage (1.4%) (1/70) case not to be statiscally significant, the histological grade was not changed, considering also the prognostic valor of other histological variables and the initial histological grade to have been greater. In the other cases, was used for final classification the histological grade identified in the intermediary and final sectioning (greater histological grade), which in these 21 cases was greater than in the initial cuts. So, our study contrasted in percentage valor (1.4%) with the studies (9.3%) realized for Santos-Pinto Jr. (1991) at the Faculty of Dentistry at the University of Sao Paulo (FO/USP), as well as, at the Faculty of Dentistry at the Federal University of Rio Grande do Sul (FO/UFRGS), Porto Alegre, Brazil, in leucoplakia, which at 4/17(23.5%) cases using semi-serial sectioning was verified rise of the cellular differentiation, since the diagnostic in the intermediary sections (epithelial dysplasia) changed to (hiperkeratose) in the final sections (Rosolen, 1997). The result of this latter study like me strange and contrast with the vision that many epithelial dysplasias have multi-centric origin. In summary, the results of our study proved that MHSG and semi-serial sectioning technique is a more confident method to establish the histological grade of malignancy of oral carcinoma than WHO classification and routine sectioning, because promotes a more accurate analysis of the histological variables pattern and stage of invasion, as well as of the vascular invasion by embolus of neoplastics cells, which are variables important to the evaluation of the tendency of a neoplasm to metastasize. In addition, these results may explicate the lack of correlation between the histological grades of the malignance of oral carcinoma, prognostic and metastatic dissemination, when the histological examination of routine is used (WHO and routine cuts).  

Fig. 9: Routine and WHO classification:  Histological grade: 1 (histologic score, 1.0). Routine and MGHS: Histologic grade: 2 (histologic score, 1.2).  

Fig. 10: Semi-serial and WHO classification:  Histologic grade  3 (histological score: 2.5). Semi-serial sectioning and MGHS: Histologic grade 3 (histological score 2.75)