Evaluation of Bcl-2 , Bcl-x and Cleaved Caspase-3 in Malignant Peripheral Nerve Sheath Tumors and Neurofibromas

AIMS: To study the expression of Bcl-2, Bcl-x, as well the presence of cleaved caspase-3 in neurofibromas and malignant peripheral nerve sheath tumors. The expression of Bcl-2 and Bcl-x and the presence of cleaved caspase 3 were compared to clinicopathological features of malignant peripheral nerve sheath tumors and their impact on survival rates were also investigated. MATERIALS AND METHODS: The evaluation of Bcl-2, Bcl-x and cleaved caspase-3 was performed by immunohistochemistry using tissue microarrays in 28 malignant peripheral nerve sheath tumors and 38 neurofibromas. Immunoquantification was performed by computerized digital image analysis. CONCLUSIONS: Apoptosis is altered in neurofibromas and mainly in malignant peripheral nerve sheath tumors. High levels of cleaved caspase-3 are more common in tumors with more aggressive histological features and it is associated with lower disease free survival of patients with malignant peripheral nerve sheath tumors.

In individuals with Neurofibromatosis type 1 (NF1), they occur in 4.6% of all cases, being this syndrome the major risk factor for the development of MPNSTs (Ducatman et al. 1986).
MPNSTs may appear de novo or may develop from the transformation of a benign neural neoplasm, mainly from a plexiform neurofibroma, which occurs almost exclusively in NF1 (Scheithauer et al. 1999).
Nevertheless, the molecular events involved in the development of neurofibromas and MPNSTs are still not completely understood.Abnormal programmed cell death is a hallmark of human tumors and apoptosis resistance is one of the most marking features of the majority of malignant neoplasms (Bray et al. 2009).Among the many known regulators and effectors of apoptosis, the family of caspases plays an important role in the execution phases of this type of cell death.Caspase-3 is considered to be the central protein in the execution of apoptosis and the detection of its activated form (cleaved caspase-3; CC3) is a valuable tool to detect apoptotic cells, even before the morphological features of apoptosis are present (Arai et al. 2005).Many proteins participate in the regulation of apoptosis; between them, there are the Bcl-2 family proteins, which include proapoptotic, e.g.Bax protein, and anti-apoptotic proteins, e.g.Bcl-2 and Bcl-x L .
The altered expression of Bcl-2 proteins family members and CC3 has been analysed in various benign and malignant neoplasms and their prognostic value has been demonstrated (Arai et al. 2005, Friess et al. 1998, Satoh et al. 2000, Biroccio et al. 2001).Nevertheless, few studies were performed with neurofibromas and MPNSTs and, to our knowledge, there is no study that investigated the presence of Bcl-2, Bcl-x and CC3 in these tumors and their the correlation to the clinicopathological features of MPNSTs and their impact on survival rates.
Studies that investigated the Bcl-2 expression in MPNSTs present different results, showing that there is still a lack of understanding of the involvement of this protein in their pathogenesis (Suster et al. 1998, Hirakawa et al. 1996, Miettinen et al. 1998, Dan'ura et al. 2002, Thanakit et al. 2006, Köhler et al. 2002).These divergent results may be a consequence of the small number of cases included in many investigations and also of the fact that most studies investigated MPNSTs together with other sarcomas, probably reflecting the difficulty of conducting research with larger number of cases of MPNSTs due to its rarity.Soft tissue sarcomas represent a heterogeneous group of tumors with different clinical behaviors.Therefore, a question raised about the studies evaluating prognostic factors in various histological types of sarcomas taken together, is that although the findings may be applicable to many sarcomas, certainly they not apply to all types sarcomas.Another important factor is that all previous studies used conventional pathologist-based manual scoring to quantify Bcl-2-staining, which increases inter-observer and intraobserver variability.
The aim of this research was to study the expression of Bcl-2, Bcl-x, as well the presence of CC3 in MPNSTs and neurofibromas.The expre ssion of Bcl-2 and Bcl-x and the presence of CC3 were compared to clinicopathological features of MPNSTs and their impact on survival rates were also investigated.

MATERIALS AND METHODS
Ethical Committee of National Institute of Cancer (INCA), RJ, Brazil, approved this study.

SAMPLE
The study sample comprises 16 cases of plexiform neurofibromas, 22 cases of non-plexiform neurofibromas (from NF1-patients and from non-NF1 patients), and 28 cases of MPNSTs, which were obtained from the pathology files of INCA and Hospital Universitário Antônio Pedro of the Universidade Federal Fluminense.All the cases were used in previous studies (Cunha et al. 2003(Cunha et al. , 2008(Cunha et al. , 2009)).The samples of the tumors were included in two tissue microarray (TMA) paraffin blocks.For detailed information about the construction method for TMA block, please refer to the previous published paper (Cunha et al. 2012) All the cases were reviewed by two pathologists.For neurofibromas, only cases with heterogeneous and diffuse immunostaining for S-100 protein were included.The samples of MPNSTs included in this study were obtained from the paraffin blocks of the resected primary tumor (in cases submitted to radiotherapy and/or chemotherapy prior to primary tumor resection, the biopsy material was used, since it had the same histological grade of the resection material).MPNSTs with one of the following features were included: arose within a peripheral nerve; arose in transition from a benign neural tumor; developed in a NF1 patient and exhibited the same histological features of most MPNSTs originating from a nerve; and developed in a non-NF1 patient, exhibited the same histological features as most MPNSTs and expressed S-100 protein (Dako Corp., Carpinteria, CA, USA, 1:4,000) and/or CD57 (clone TB01; Dako, 1:50).All samples were immunoreactive for anti-vimentin (clone V9; Dako, 1:800) and negative for anti-cytokeratin (clone AE1/ AE3; Dako, 1:400), anti-melanosome, (clone HMB-45, 1:200), anti-actin (smooth muscle; clone 1A4; Dako, 1:250), antiactin/muscle (clone HHF35; Dako, 1:1,000), and anti-desmin (clone D33; Dako, 1:100) antibodies, except the malignant Triton tumors, which exhibited anti-actin/muscle and desmin immunopositive areas.

CLINICAL AND PATHOLOGICAL FEATURES OF MPNSTS INCLUDED IN THE STUDY
Tables I and II show the clinical and pathological features of MPNSTs included in this research, respectively.MPNSTs were classified in low and high-grade (Scheithauer et al. 1999).Tumors were classified as low-grade if they presented mild to moderate cellularity, mild nuclear atypia, absence of necrosis and no or few mitoses (mitotic index <4).Tumors were classified as high-grade if they had marked cellularity, variable cellular pleomorphism and easily identified mitoses (mitotic index ≥ 4).Mitotic index was calculated by the sum of mitotic figures in 10 high-power fields.In each case, at least 30 fields were analyzed and the mean value of the sum of mitotic figures in each 10 high-power fields was obtained.
Epithelioid MPNSTs included in this study were composed predominantly of epithelioid cells and presented spindle cells identical to that of conventional MPNSTs.Malignant triton tumors had areas of cells with rabdomyoblast morphology, which expressed desmin and/or actin/muscle.

IMMUNOHISTOCHEMICAL STAINING
Sections with 3 µm from the two TMA paraffin blocks were cut and collected on silane-coated slides.After dewaxing, proteins were demonstrated by IHC, using the protocol described in a previous study (Cunha et al. 2009).Sections were incubated overnight at 4°C with a 1:200 dilution of the primary antibody anti-Bcl-2 (clone 124; Dako Corporation, USA), with a 1:25 dilution of the primary antibody anti-Bcl-x (clone A35-10; Dako Corporation, USA), and with a 1:600 dilution   -, information not available (the tumor resection was performed in other institution); sd, standard deviation *, in each case, at least 30 fields were analyzed and the mean value of the sum of mitotic figures in each 10 high-power fields is showed in the table.
of the primary antibody anti-CC3 (clone Asp175; Cell Signaling Technology, USA).Palatine tonsil was used as a positive control.The negative control was established by omission of primary antibody.
The glass slides were reviewed independently by two pathologists and were scored as negative or positive.The location (nucleus, cytoplasm and/ or cytoplasmic membrane) of immunopositivity for Bcl-2, Bcl-x and CC3 was also evaluated.Following the independent evaluation of the immunohistochemical staining, if there were any divergence between the pathologists, the glass slides were discussed at a separate evaluation in order to reach a consensus on the evaluation.For the cases classified as positive, quantification of Bcl-2, Bcl-x and CC3 immuno staining was performed by computerized digital image analysis (Image-Pro Plus software v4.5;Media Cybernetics), as previously described (Cunha et al. 2009), and expressed as positivity index (PI), defined as positive area divided by total tissue area.
To avoid searching the optimal cut-off point for the immunostaining to perform the statistical analyzes (overall and disease-free survivals), which could be misleading and associated with difficulty in its interpretation, we arranged the cases of MPNSTs in ascending order according to PI values and divided them into two groups: one with tumors with high PI (cases in the third third of the list) and another with tumors with low PI (cases in first two-thirds of the list).Immunostaining for proteins Bcl-2 and Bcl-x was found in the cytoplasm of tumor cells of neurofibromas and MPNSTs (Fig. 1).Moreover, expression of Bcl-2 and Bcl-x was also observed in the endothelial cells of blood vessels, as well as in lymphocytes, when present in the tumor.
Regarding the presence of CC3, neurofibromas showed only nuclear immunostaining (Fig. 2A).In contrast, half of MPNSTs showed only nuclear immunostaining and the other 50% of the cases showed cytoplasmic and nuclear staining (Fig. 2B, 2C and 2D).

NEUROFIBROMAS
There was no statistically significant difference between plexiform neurofibromas and nonplexiform neurofibromas regarding the percentage of tumors that expressed Bcl-2 (Fisher's exact test, p=0.166) and Bcl-x (Chi-square test; p=0.539) proteins.With respect to CC3, all neurofibromas were immunopositive, expressing the protein only in the nucleus.
Considering the immunopositive cases for each antibody, there was no statistically significant difference of PIs between the two groups of

COMPARISON OF IMMUNOSTAINING FOR BCL-2, BCL-x AND CC3 WITH CLINICOPATHOLOGICAL FEATURES OF CASES OF MALIGNANT PERIPHERAL NERVE SHEATH TUMORS
We investigated the correlation of the immunostaining of Bcl-2, Bcl-x and CC3 with the presence of NF1, recurrence and metastasis, histological grade, mitotic index and necrosis.In all tests performed, presence of NF1 (Fisher's exact test, p=0.011), high histological grade (Fisher's exact test, p=0.003) and high mitotic index (Fisher's The overall 5-year survival rate of patients with MPNST was 46%.Considering the patients who received curative treatment, the 5-year diseasefree survival rate was 39%.All these data were showed in the previous study in which we used the same sample (Cunha et al. 2012).In that study, we evaluated the influence of clinical and pathological features on overall and disease free survival in patients with MPNSTs: sex, presence of NF1, tumor location, treatment performed, status of the surgical margins, local recurrence, metastasis, histological grade, mitotic index, necrosis, epithelioid variant, Triton tumor and chondrosarcoma differentiation.The variables, which were significant in univariate log-rank analysis for survival rates, are shown in Table IV, as well as the results of the evaluation of Bcl-2, Bcl-x and CC3.Presence of NF1 (p=0.044),high histological grade (p=0.045),presence of necrosis (p= 0.003) and high CC3 PI (p=0.024) were significant in univariate log-rank analysis for shorter overall survival and were included in Cox regression model, which showed that the presence of necrosis was an independent prognosis factor for lower overall survival (p=0.007).(Table V) Epithelioid variant (p=0.036) and high CC3 PI (p<0.001) were significant in univariate log-rank analysis for disease-free survival (Table II).Cox regression model showed that high CC3 PI could be considered an independent prognosis factor for disease-free survival (p=0.002)(Table V).

DISCUSSION
To our knowledge, the present study was the first to use computerized image analysis to calculate the Bcl-2 expression in neurofibromas and MPNSTs.The results showed a significantly higher percentage of positive MPNSTs for Bcl-2 than neurofibromas (p<0.001).Other studies also observed increased expression of Bcl-2 in malignant neoplasms, such as cutaneous carcinomas and melanomas, when compared to their benign counterparts, suggesting that the expression of Bcl-2 protein could be an oncogenic factor (Leiter et al. 2000, Hussein et al. 2004).Corroborating to our findings, other studies also observed that most MPNSTs express Bcl-2 (Miettinen et al. 1998, Watanabe et al. 2001, Sabah et al. 2007).Contrary to our results, Suster et al. (1998) found that 87.5% of eight neurofibromas expressed Bcl-2, while only 39.3% of 25 MPNSTs were immunoreactive.In another survey with 38 sarcomas, none of the four studied MPNSTs expressed the Bcl-2 protein.In two studies that included MPNSTs, which arose from neurofibromas (one study with eight and the other with 20 cases), there was no difference in Bcl-2 expression between the benign and malignant counterparts (Thanakit et al. 2006, Watanabe et al. 2001).
Bcl-2 expression has been correlated with histological grade and clinical behavior of several malignant neoplasms (Hirakawa et al. 1996, Sabah et al. 2007, Silvestrini et al. 1994, Ofner et al. 1995, Groeger et al. 2004).In certain cancers, like breast carcinomas, the expression of this protein is more commonly observed in lowgrade neoplasms, whereas in other tumors, such as bladder carcinoma and some sarcomas, its expression has been more common in high-grade tumors (Sabah et al. 2007, Yang et al. 2000, Eissa and Seada 1998).It has been demonstrated that the immunoreactivity to the Bcl-2 is associated with increased survival in patients with colorectal, breast, lung and ovary carcinomas (Silvestrini et SV, survival; PI, positive index; *, the value could not be calculated because all cases were censored.

TABLE IV Results of the influence of variables on overall and disease-free survival
in patients with malignant peripheral nerve sheath tumors.al. 1994, Ofner et al. 1995, Yoo et al. 2007, Nadler et al. 2008, Yang et al. 2000).Conversely, in other tumors, such as non-small-cell lung carcinomas, the expression of this protein has been correlated with poor survival (Groeger et al. 2004).Our results showed no correlation between the expression of Bcl-2 and histological grade, as well as no correlation between the expression of Bcl-2 and survival of patients with MPNSTs, similarly to other studies with different tumors.(Hirakawa et al. 1996, Paik et al. 2006, Tsutsui et al. 2006).Bcl-x is another member of the Bcl-2 proteins family that has been extensively studied in many tumors.Two isoforms, Bcl-x S and Bcl-x L , can be generated by alternative splicing of Bcl-x transcripts, having an inductor and inhibition function on apoptosis, respectively (Schoelch et al. 1999).In some malignant neoplasms, higher expression of anti-apoptotic Bcl-x L protein was observed, when compared to corresponding normal tissues (Leiter et al. 2000, Guo et al. 2002).In a study with an established NF1-associated MPNST cell line and primary tissue cultured MPNST cells derived from a NF1 patient, high expression of Bcl-x L was observed (Lee et al. 2012).In another recent study with cell lines, it was found that Bcl-x L is upregulated in Schwann cells derived from NF1-associated MPNST compared to neurofibromas from NF1 patients and was suggested that this upregulation may be caused by NF1 deficiency-mediated elevation in Ras signaling.(Park et al. 2013) Our results showed that there was a higher number of positive MPNSTs for Bcl-x than neurofibromas (p <0.001).Moreover, positive MPNSTs had higher PIs than neurofibromas (p=0.002).Interestingly, in a study with many normal tissues, the expression of Bcl-x L and Bcl-x S was not observed in Schwann cells, neither in other cells from peripheral nerve sheath (Krajewski et al. 1994).However, in the already mentioned study of Park et al. (2013) with cell lines, the expression of Bcl-x L in normal Schwann cells could be detected by Western blot, but they presented lower expression when compared to Schwann cells derived from NF1-associated MPSNTs.Because the antibody we used does not distinguish the two isoforms of Bcl-x protein, we cannot know if the expression of Bcl-x observed in the tumors of our study is referred to the anti or pro-apoptotic isoform.Nevertheless, it probably corresponds to Bcl-x L isoform.
Investigations with other malignant neoplasms demonstrated a correlation between high expression of Bcl-x L protein with poor prognosis (Friess et al. 1998, Thamboo et al. 2006).Köhler et al. (2002) evaluated the expression of messenger RNA (mRNA) of Bcl-x L , Bcl-2 and other genes involved in apoptosis in 82 soft tissue sarcomas, including 15 MPNSTs.They concluded that high levels of mRNA of Bcl-x L were related to a poor prognosis.Unfortunately, this study, as many others, investigated a variety of histological types of soft tissue sarcomas together, and there is no report about the expression of these genes in each type of sarcoma separately.Our research did not show correlation of Bcl-x expression to the prognosis.
In our study, we performed the identification of apoptotic cells through the investigation of the presence of CC3 using IHC.The activation of effectors caspases, like caspase-3, is an event also known as "point of no return" of apoptosis, beyond which the cells certainly will entry to apoptosis (Asselin et al. 2001).The presence of CC3 has been demonstrated in many neoplasms, such as breast cancer, gliomas, neuroblastomas and sarcomas (Konstantinidou et al. 2007, Kobayashi et al. 2007, Matsubara et al. 2008, Nassar et al. 2008).
Double inactivation of NF1 gene in Schwann cells seems to be the common event in neurofibromas and MPNSTs.(Laycock- van Spyk et al. 2011) This leads to high levels of Ras-GTP, since neurofibromin, which is an important negative regulator of Ras pathway (similar to GTPase-activating protein -GAP) is deficient as a consequence of the NF1 gene mutation.It is known that Ras activation elicits antiapoptotic effects in many cells and that there is a positive correlation between neurofibromin deficiency and resistance to apoptosis (Dasgupta and Gutmann 2005, Guerrero et al. 2000, Shapira et al. 2007).Therefore, the observation that all neurofibromas and MPNSTs were immunopositive for CC3 in our study was unexpected.Interestingly, a study with cultured Schwann cells from a high-grade MPNST showed high levels of expression of caspase-3 gene (Lee et al. 2004).
The observation of the presence of CC3 in the nucleus and cytoplasm of MPNSTs cells in our study is in accordance with the results of a previous study (Kamada et al. 2005), which demonstrated that this protein is actively translocated, by one of more unknown proteins, from the cytoplasm to the nucleus.The reason why nuclear location of CC3 was only observed in MPNST and not in neurofibromas needs to be further investigated.
In the study with pancreatic neoplasms, using IHC technique and mRNA analysis, the pattern of cellular location of caspase-3 reflected the biological behavior of benign and malignant neoplasms (Satoh et al. 2000).Nuclear location of caspase-3 had statistically significance with the benign characteristic of pancreatic neoplasms, whereas the cytoplasmic location of this protein was correlated to the invasivity of malignant tumors.In some tumors, i.e. meningiomas (Konstantinidou et al. 2007), and Burkitt's and Burkitt-like lymphomas (Nomura et al. 2008), the location of caspase-3 protein has been shown to be a prognostic marker.In a study with 41 sarcomas, but none MPNSTs, the overall and disease free survival were lower in cases of patients with higher levels of CC3 (Matsubara et al. 2008).
Our results showed that high level of CC3 was associated to lower disease free survival.We found that the presence of cytoplasmic staining and high level of CC3 were more frequently associated to high-grade tumors and high mitotic index.Moreover, cytoplasmic location of CC3 was more common in MPNSTs from NF1 patients and high CC3 PI was associated to necrosis.These data show that higher quantities of CC3 are more common in tumors with more aggressive histological features and it is associated with a poor prognosis.
Because CC3 is associated to induction of apoptosis, it would be expected that tumors with high quantity of this protein would have better prognosis or more indolent features.Nevertheless, we should keep in mind that tumor growth results from lack of balance between proliferation and cell death by apoptosis and our results show that, in more aggressive tumors, although there was high number of apoptotic cells, there was also high number of proliferating cells (high mitotic index).
Finally, our study shows that apoptosis is altered in neurofibromas and mainly in MPNSTs and that the accumulation of Bcl-2, Bcl-x and CC3 may be correlated with the biology of these tumors.Further studies, including other members of Bcl-2 proteins family, such as Bax, Bak, Bad, are required for a better understanding of the alterations in apoptotic mechanisms involved in the biology of neurofibromas and MPNSTs.

TABLE I (continuation) Patient Age at diagnosis Sex Race NF1 Site Free surgical margins Treatment Local recurrence Metastasis Site of metastasis Death Overall survival (months) Disease-free survival (months)
-, information not avaiable; F, female; M, male; B, black; W, white; Rxt, radiotherapy; Ct, chemotherapy; * these patients presented metastasis at the moment of the tumor diagnosis.

TABLE II Pathological data of the malignant peripheral nerve sheath tumors.
in neurofibromas and MPNSTs are described in TableIII.