Prognostic Value of SPARC Expression in Unresectable NSCLC Treated with Concurrent Chemoradiotherapy

In human, SPARC, also termed as osteonectin or basal membrane protein 40 (BM-40), is a protein encoded by SPARC gene. Human SPARC gene is 26.5 kb long; contains 10 exons and 9 introns and is located at chromosome 5q31-33. SPARC is an acidic, cysteinrich glycoprotein with a molecular weight of 40 Kd (Kaufmann et al., 2004). It is a non-structural component of extracellular matrix. SPARC is released by osteoblasts during bone formation, initiation of mineralization and formation of mineral crystals. It shows affinity to collagen in addition to bone mineral calcium. SPARC is a specialized, glycosylated protein that regulates changes and remodeling in multifunctional intercellular matrix. SPARC mediates to interactions between cell and its surrounding environment by fusing with collagen, laminin, fibronectin and vitronectin (Podhajcer et al., 2008; Huang et al., 2010). SPARC regulates many


Introduction
In human, SPARC, also termed as osteonectin or basal membrane protein 40 (BM-40), is a protein encoded by SPARC gene. Human SPARC gene is 26.5 kb long; contains 10 exons and 9 introns and is located at chromosome 5q31-33. SPARC is an acidic, cysteinrich glycoprotein with a molecular weight of 40 Kd (Kaufmann et al., 2004). It is a non-structural component of extracellular matrix. SPARC is released by osteoblasts during bone formation, initiation of mineralization and formation of mineral crystals. It shows affinity to collagen in addition to bone mineral calcium. SPARC is a specialized, glycosylated protein that regulates changes and remodeling in multifunctional intercellular matrix. SPARC mediates to interactions between cell and its surrounding environment by fusing with collagen, laminin, fibronectin and vitronectin (Podhajcer et al., 2008;Huang et al., 2010). SPARC regulates many biological processes such as cell proliferation and survival, apoptosis, adhesion or migration (Tai et al., 2008;Chlenski et al., 2010). In normal tissues, the SPARC expression is limited to intestinal epithelium, osseous epithelium and tissues undergoing remodeling or repairing (Chin et al., 2005). Because of intense matrix formation in tumor development, various SPARC expression patterns can be detected in cancer-related stroma or malign cells, representing tumor development (Miyoshi et al., 2010). Different SPARC expressions have been demonstrated in several human cancers; however, differential effects of SPARC on tumor growth in distinct tissues haven't been fully elucidated yet.
In several studies, it has been shown that high SPARC expression is an independent prognostic factor for disease progression and poorer overall survival in breast (Gradishar et al., 2005;Watkins et al., 2005;Nagai et al., 2010), pancreas (Gradishar et al., 2006), gastric (Sato et al., 2013), bladder (Chlenski et al., 2006) and tongue cancers (Schiemann et al., 2003) as well as malign melanoma (Desai et al., 2009) head-neck cancers (Watkins et al., 2005) glioblastoma (Cheetham et al,. 2008), meningioma (Yiu et al., 2001) and osteosarcoma (Tai et al., 2005). In contrast, in some other studies, it was found that low SPARC expression is a prognostic factor for poorer survival in colorectal cancer (Taghizadeh et al., 2007), ovarian cancers (Said et al., 2007), pancreas cancer (Infante et al., 2007) and acute myeloid leukemia (DiMartino et al., 2006). Thus, the effect of SPARC expression on prognosis of human cancers is unclear and rather controversial. It is needed to evaluate its prognostic and predictive value in patients.
In contrast to other studies, the prognostic and predictive value of SPARC was evaluated in patients with locally advanced, stage III non-small cell lung cancer (NSCLC), who received same treatment modality in this study.

Materials and Methods
In this study, 84 patients who were treated with the diagnosis of locally advanced NSCLC between January, 2007 and December, 2011 were retrospectively evaluated. The pathology specimens of patients, who were treated with chemoradiotherapy and adjuvant chemotherapy after confirmation of the diagnosis of unresectable, locally advanced, stage III NSCLC by a multidisciplinary council, were evaluated. This study was approved by Local Ethics Committee (Approval #: 2012/192). The inclusion criteria were as follows: patients with ECOG performance score of 0 and 1; patients without pleural fluid requiring drainage, those with serum creatinine≤1.5mg/dl and creatinine clearance≥60ml/min, those with normal liver function tests and complete blood count, those without comorbid heart disease such as coronary artery disease or congestive cardiac failure; and those didn't previously receive radiotherapy/chemotherapy. Patients younger than 18 years of age, those who have medical contraindication for radiotherapy, those with severe comorbid disease (severe heart failure, respiratory failure, liver failure or renal failure etc.), those with history of severe infection within prior 2 weeks, patients with history of additional malignancy and those in which histopathological diagnosis was made at another facility; thus, had insufficient pathology specimen, were excluded. By X-ray simulator, a 2-dimensional treatment planning system was used for radiotherapy planning and radiotherapy was given by using linear accelerator device (6 and 18 MV photons) (Varian CDX 2300). Simultaneous RT was initiated on the day 1 of chemotherapy and given at a fractioned doses of 2 Gy (total dose of 66 Gy) in combination with simultaneous weekly docataxel (20mg/m 2 ) and cisplatin (20mg/m 2 ) infusions. Initially, planning target volume was estimated to include primary tumor and ipsilateral hilar and mediastinal lymph nodes with margins of 2cm. After the dose of 46 Gy, additional doses of 20 Gy (in 10 fractions) were given to gross tumor volume at both sides and involved lymph nodes with margins of 1.5cm.

Immunohistochemical staining
The best exemplifies the tumor slights were selected and that were re-evaluated under the light microscope. All tissue used in this study were fixed in 10% neutralbuffered formalin and embedded in paraffin. The tissue parts cut into 5µ thick slices and mounted on poly-L-lysine coated slides. A step of heat-induced antigen retrieval in pH: 6.0 sodium citrate buffers were included for 10 minutes cycles for SPARC (Bioss, bs-1133R, Rabbit Anti-SPARC Polyclonal Antibody, Unconjugated). Immunohistochemical staining was examined by using of avidin-biotin-peroxidase method. The renal tissue was used as positive controls for immunohistochemıcal staining of SPARC. The cytoplasmic staining was accepted as positive for more than 5% of tumor cells. The intensity of immunostaining for SPARC was reviewed and scored according to the location of cytoplasmic without knowledge of the clinicopathological parameters of the patients. The staining results of SPARC were evaluated by a score corresponding to the sum between: the percentage of cytoplasmic positive tumor cells : negative, 0% immunopositive cells; weakly positive(1+), ≤25% positive cells; moderately positive(2+), 26~74% positive cells; ≥75% positive cells, strongly positive(3+). However, tumor cells were stratified according to binary scale system as follows: negative and weak staining as low expression and moderate or strong staining as high expression.

Statistical analysis
Data were expressed as frequency, percentage, mean± standard deviation and median (min-max). Normal distribution of data was tested by Shapiro-Wilks's test and assessed by histogram and q-q plot. The patients were stratified into subgroups according to SPARC expression (low or high). Mann Whitney U and chi-square tests were used to compare differences between groups stratified according to age (<58 or ≥58 years), gender (male or female), T stage (T0, T1, T2, T3, T4), N stage (N0, N1, N2, N3), stage (IIIA and IIIB) and histopathology (adenocarcinoma, SCC, unclassified). Kaplan-Meier survival analysis was used to determine and compare overall survival, local control and disease-free survivals between groups. Statistical differences were calculated by log rank test. Univariate and multivariate Cox regression analysis was used to identify risk factors including age (<58 or ≥58 years), gender (male or female), T stage (T0, T1, T2, T3, T4), N stage (N0, N1, N2, N3), stage (IIIA and IIIB), histopathology (adenocarcinoma, SCC, unclassified), SPARC expression (low or high). p<0.05 was considered as statistically significant. SPSS version 20.0 (IBM; SPSS Inc., Chicago, IL) was used in statistical analysis.
According to SPARC expression, it was found that median overall survival was 16±2.73 (11.55-20.46) months for low expression vs 7±1.79 months (7.92-16.08) months for high expression. A significant difference was detected between groups regarding overall survival (p=0.032).
In addition, according to SPARC expression, it was found that median local control was 13±2.31 (8.48-17.52) months for low expression vs 6±0.85 (4.34-7.66) months for high expression. A significant difference was detected between groups regarding local control (p=0.045).
No significant difference was detected between SPARC expression and response to chemoradiotherapy (p=0.502).
In both univariate and multivariate analyses, high

Figure 1. Positive Staining of Tissue Section with SPARC in Cytoplasm in Non-small Cell Lung Cancer
Cells. The cytoplasmic staining was accepted as positive for more than 5% of tumor cells. Pozitif immunostaining for SPARC x10 (left), pozitif immunostaining for SPARC x20 (right)

Discussion
Only 15% of the patients with lung cancer, which is leading cause of cancer-related deaths, can able to survive 5 years or more after diagnosis. NSCLC is often diagnosed at an advanced stage (Oguz,et al., 2013) and tumor stage is the most important prognostic factor in NSCLC; followed by histopathological diagnosis. However, different prognostic factors are needed as patients at same stages shows different survival times. There are several biomarkers as prognostic and predictive markers for NSCLC. Best known of these biomarkers are Epidermal growth factor receptor (EGFR), excision repair crosscomplementation group l (ERCC1), anaplastic lymphoma kinase (ALK), KRAS oncogene, ribonucleotide reductase subunit M1 (RRM1). In various studies, effects of these biomarkers on survival and/or response to therapy are shown in NSCLC (Tsao et al., 2007;Wu et al., 2014;Xie et al., 2014;).
SPARC is an extracellular matrix protein and it can be detected in cancer-related stroma or malign cells in tumor development. Effect of SPARC expression on prognosis is unclear in lung cancers. Therefore, prognostic and predictive value of SPARC expression should have to be evaluated in patients with the disease at same stage who received same treatment modalities. In the present study, we evaluated prognostic and predictive value of SPARC in patient who received chemoradiotherapy with a diagnosis of stage III NSCLC.
In previous studies, it was failed to demonstrate SPARC expression in normal lung tissue other than focal and weak staining at vessel walls and alveolar septa (Siddiq et al., 2004). In contrast, majority of tissues with NSCLC (75%) showed cytoplasmic SPARC expression. Thus, it was reported that SPARC expression could be related to in vivo and in vitro malign cell transformation and might be a good marker for NSCLC.
In a study by Koukourakis et al. (2003). stromal and cytoplasmic SPARC expressions were evaluated in 102 patients with T1-2, N0-1 and M0 disease who underwent surgical resection. It was found that there was high SPARC expression in 37 (%36) and low SPARC expression in 65 (%64) of 102 patients. In survival analysis, a significant    DOI:http://dx.doi.org/10.7314/APJCP.2014.15.20.8911 Prognostic Value of SPARC Expression in NSCLC Treated with Concurrent Chemoradiotherapy correlation was shown between high stromal SPARC expression and poor prognosis (p=0.006). In our study, SPARC expression was found as low in 48 (57%) patients and high in 36 (43%) patients in IHC staining. Our expression rates were in agreement with the study by Koukourakis et al., as low expression rate being high. In addition, poor prognostic value of SPARC detected in multivariate survival analysis (p=0.007) in our study was in agreement with this study.
In another study, SPARC was evaluated pathology preparations by using IHC method in 105 patients (Huang et al., 2012). Authors detected positive SPARC staining in 57 patients (54.3%) and negative SPARC staining in 48 patients (45.7%). When compared to adenocarcinoma, epidermoid carcinoma showed both higher SPARC expression and longer disease-free survival (p=0.041).
In that study, overall survival was significantly higher in the group with high SPARC expression (HR: 0.32; 595% CI: 0.16-0.65; p=0.001); however, no such relationship was shown in disease-free survival (p=0.543). The positive effect of SPARC on prognosis is linked to affinity of SPARC to albumin, based on the study which demonstrated higher response rate to nab-paclitaxel in SPARC positive patients when compared to SPARC negative patients in NSCLC (Socinski et al., 2010). We failed to demonstrate longer disease-free survival in epidermoid carcinomas compared to adenocarcinomas, as shown in the study by Huang et al. This could be due to smaller number of cases with adenocarcinoma. Although Schneider et al. also found that low SPARC level is related to long median survival, in terms of estimated probability of survival it was not statistically significant (p=0,095) (Schneider et al., 2004).
In a study assessing 89 patients with stage I-IV NSCLC, relationship between SPARC expression and age, gender, histopathological type, tumor size, tumor differentiation, T stages and N stage was investigated in normal pulmonary and cancer tissue . Higher SPARC expression was found in cases with lymph node metastasis when compared to those without (81.3% vs 58.5%; p<0.05); however, no such relationship was detected for tumor size, age and gender. Authors found that there was significantly higher SPARC expression in patients with stage III-IV disease when compared to those with stage I-II disease (p=0.04). In our study, it was failed to demonstrate high SPARC expression in patients with lymph node metastasis However, the cytoplasmic assessment in that study is supportive for our methodology. In our study, lack of significant relationship between SPARC expression and age, gender or tumor size was in agreement with the study by Zhang et al. In addition, the finding of higher SPARC expression in stage III and IV disease than in stage I and II, supports our study which made in patients with stage III.
On the baseline of these findings, we can understand that SPARC is closely related to tumor formation and especially poor prognosis in advanced stages NSCLC independent from age, sex and tumor size.
Recently it is announced by Grant et al. that how poor prognostic impact of SPARC occurred in NSCLC. According to this study, overexpression of Snail causes the upregulation of SPARC and increased SPARC-dependent invasion, therefore in NSCLC progression, SPARC can play an important role. Transforming growth factor beta, Extracellular signal-regulated kinase1/2, and miR-29b are potential intermediaries in this process.
In conclusion, in this study, it was shown that SPARC is an important prognostic factor in cases with locally advanced, Stage III NSCLC. Our study differs from others regarding homogeneous patient group and treatment, thus this poor prognostic effect of SPARC was demonstrated more clearly.