Clinicopathological and prognostic values of fibronectin and integrin αvβ3 expression in primary osteosarcoma

We carried out a retrospective study of 60 patients with primary osteosarcoma who had undergone complete surgical resection at the First Affiliated Hospital of Fujian Medical University between 2009 and 2014. Histopathological diagnosis of all specimens was confirmed by a senior doctor of pathology. All patients underwent standardized neoadjuvant and postoperative chemotherapy with ifosfamide, cisplatin, and doxorubicin. Relevant clinical data were retrieved from medical records, including sex, age at diagnosis, tumor size, tumor location, histologic subtype, Enneking staging, and response to chemotherapy. Formalin-fixed and paraffin-embedded surgical tumor specimens for immunohistochemical staining were obtained from the archives of the Department of Pathology.

Follow-up of osteosarcoma patients was terminated on 31 August 2017 either by phone call or outpatient visit. To assess the development of local recurrence and distant metastasis, all patients with osteosarcoma were monitored by X-ray or lung computed tomography (CT) scans after surgical excision every 3 months during the first 3 years and every 6 months thereafter. DFS time was calculated from the date of diagnosis until the date of first tumor progression. OS time was calculated from the date of diagnosis until the date of death. Patients were censored at the date of the last follow-up if tumor progression or death had not occurred.

Written informed consent was provided by all participants involved in this study. The study protocol conformed to the ethical guidelines of the Declaration of Helsinki and was approved by the Ethics Committee of the First Affiliated Hospital of Fujian Medical University.

Immunohistochemical expression of FN and αvβ3 in archival osteosarcoma specimens were examined by applying a PV-9000 Polymer Detection System (Zhongshan Goldenbridge Inc., Beijing, China), with 30 corresponding osteochondroma tissues which were resected at the First Affiliated Hospital of Fujian Medical University between 2009 and 2014 used as controls. Paraffin-embedded specimens were serially sectioned (4 μm) and incubated for 1 h at 60 °C. The tissue sections were then deparaffinized, hydrated, and incubated with 3% hydrogen peroxide for 10 min at room temperature to block endogenous peroxidase activity. During the antigen retrieval process, the sections were placed in citrate buffer (pH 6.0) in an electromagnetic oven for 2 min and then allowed to cool to room temperature. The sections were incubated overnight at 4 °C with antibodies against FN (mouse monoclonal 2755-8; 1:50; Santa Cruz, USA) and αvβ3 (rabbit polyclonal orb10927; 1:50; Biorbyt, UK). Next, the sections were incubated with a polymer helper reagent for 20 min at room temperature and then poly-peroxidase-anti-mouse/rabbit IgG for 30 min at room temperature according to the manufacturer’s instructions. After staining with diaminobenzidine (Zhongshan Goldenbridge Inc.), the sections were counterstained with hematoxylin, dehydrated, and mounted. Negative (PBS (0.01 M, pH 7.2) rather than primary antibodies) and known positive (human esophagus tissue for FN and human lung cancer tissue for αvβ3) controls were stained in parallel with each set of sections studied.

The staining results were evaluated by two independent observers (CYP and ZZZ). Cytoplasmic staining for FN or αvβ3 in tumor cells was interpreted as a positive result. The average labelling index of FN from five random high-power fields was semi-quantitatively recorded as follows: 0, no positive staining; ±, only a few scattered positive cells accounting for less than 20% of tumor cells; +, cluster(s) of positively stained cells accounting for 20–30% of tumor cells; ++, cluster(s) of positively stained cells accounting for greater than 30% of tumor cells [15]. The average labelling index of αvβ3 from five random high-power fields was recorded as follows: 0, absent; ±, weak expression, accounting for greater than 20% of tumor cells; +, moderate expression, accounting for greater than 20% of tumor cells; ++, strong expression, cells accounting for greater than 20% of tumor cells [16]. Specimens showing immunostaining of ++ were defined as high expression of FN or αvβ3; expression levels of ± or + were defined as low expression, and 0 as negative expression.

Tissues (100 mg) of 60 osteosarcoma and 30 corresponding osteochondroma cases were ground into powder in liquid nitrogen and lysed in lysis buffer (cat. no. G2002; Servicebio Technology, Wuhan, China). Protein concentrations in the lysates were then quantitated using a Bicinchoninic Acid Protein Assay kit (cat. no. G2026; Servicebio Technology) and preserved at − 80 °C. Proteins (40 μg) were separated by 10% SDS-PAGE and transferred to PVDF membranes. The membranes were incubated with primary anti-FN (1:1000; Santa Cruz) and anti-αvβ3 (1:1000; Biorbyt) antibodies overnight at 4 °C, followed by incubation with a horseradish peroxidase-conjugated secondary antibody (cat. no. GB23404; 1:3000; Servicebio Technology) for 1 h at 37 °C. The signal was visualized using an Enhanced Chemiluminescence detection system (Amersham Biosciences, UK) and Image Lab software version 3.0 (Bio-Rad Laboratories, Inc., USA). β-actin was simultaneously detected using mouse anti-β-actin antibody (1:5000; Servicebio Technology) as a loading control.

The chi-square test, Fisher’s exact test, or Student’s t test (independent-sample) was used to compare FN and αvβ3 expression between osteosarcoma and osteochondroma and to determine whether their expression was correlated with the clinicopathological data of the osteosarcoma patients. Spearman’s rank coefficient was applied to determine the correlation between FN and αvβ3 expression. Kaplan-Meier survival plots were employed for univariate analysis and the log-rank test was utilized to compare differences in survival distributions. The Cox proportional hazards model was used to perform multivariate analysis for all parameters significant in the univariate analysis. The Harrell concordance index (C-index) was calculated to measure the performance of the model. All statistical analyses were performed using SPSS software version 19.0 (SPSS Inc., Chicago, USA). P < 0.05 was considered statistically significant.

FN and αvβ3 protein distribution were primarily observed in the cytoplasm of tumor cells (Fig. 1). The statistical results of immunohistochemistry are summarized in Table 1. FN and integrin αvβ3 were highly expressed in 19 (31.7%) and 16 (26.7%) of 60 osteosarcoma cases, which were not observed in the 30 osteochondroma cases. FN (P = 0.002) and αvβ3 (P < 0.001) showed higher rates of expression in osteosarcoma than in osteochondroma. The expressional levels of the two proteins were further verified by western blotting analysis. Similarly, the expression of FN (P < 0.001) and αvβ3 (P = 0.003) was also found to be upregulated in osteosarcoma tissues compared with osteochondroma tissues (Fig. 1). These results demonstrated that the expressional levels of FN and αvβ3 proteins were markedly increased in osteosarcoma tissues compared with the corresponding osteochondroma tissues. Furthermore, high expression (FN⁺/αvβ3⁺; Table 1; Fig. 2) and low/negative expression (FN⁻/αvβ3⁻; Table 1) of both FN and αvβ3 were identified in 10 and 50 osteosarcoma cases, respectively. The co-expression rates of the two proteins were significantly different in osteosarcoma and osteochondroma specimens (P < 0.001). A significant correlation for the level of expression was observed between FN and αvβ3 in osteosarcoma (r = 0.379, P = 0.003, Table 2).

The osteosarcoma patients’ clinicopathological characteristics are summarized in the Additional file 1: Table S1. Association of FN and αvβ3 expression individually (Table 3) and together (Table 4) with clinicopathological parameters, including sex, age, tumor size, tumor location, histologic subtype, Enneking staging, and response to chemotherapy, were analyzed. Expression of FN and integrin αvβ3 was stratified according to conventional (osteoblastic, chondroblastic, and fibroblastic types) and special (small cell and telangiectatic types) osteosarcoma, rather than each histological subtype. In osteosarcoma, high FN expression was significantly associated with a poor response to chemotherapy (P = 0.001; Table 3), whereas high expression of αvβ3 was significantly associated with advanced Enneking staging (P = 0.028; Table 3) and a poor response to chemotherapy (P = 0.002; Table 3). Tumors were more likely to develop to an advanced surgical stage (P = 0.013; Table 4) and exhibit a worse response to chemotherapy (P = 0.002; Table 4) as the combined expression levels progressed from FN⁻/αvβ3⁻ to other groups (FN⁺/αvβ3⁻ plus FN⁻/αvβ3⁺) and then an FN⁺/αvβ3⁺ status. In addition, a sex difference (P = 0.011) was found among the three expression levels of FN and αvβ3.

The mean patient follow-up time was 45.2 months (range 8 to 86 months). By the end of the follow-up period, 28 (46.6%) patients survived with no evidence of disease, 13 (21.7%) remained alive with disease, and 19 (31.7%) succumbed to osteosarcoma (Additional file 1: Table S1).

Univariate analysis of patient survival in relation to FN and αvβ3 expression levels is presented in Table 5, and survival curves are shown in Fig. 3. Mean DFS and OS times decreased with increasing expression of either FN (DFS, P < 0.001, Fig. 3a; OS, P < 0.001, Fig. 3b) or αvβ3 (DFS, P < 0.001, Fig. 3c; OS, P < 0.001, Fig. 3d). High expression of both FN and αvβ3 (FN⁺/αvβ3⁺) was associated with a shorter DFS time (P < 0.001, Fig. 3e) and OS time (P < 0.001, Fig. 3f) compared with the other groups (FN⁺/αvβ3⁻ plus FN⁻/αvβ3⁺ plus FN⁻/αvβ3⁻). Furthermore, the mean survival time decreased based on the extent of FN⁺ or αvβ3⁺ expression, which was longest for the FN⁻/αvβ3⁻ group (DFS, 67.14 months; OS, 78.43 months) followed by the single high-expression groups (FN⁺/αvβ3⁻ plus FN⁻/αvβ3⁺; DFS, 27.60 months; OS, 58.57 months) and then the FN⁺/αvβ3⁺ group (DFS, 15.10 months; OS, 24.10 months). The DFS time difference between the FN⁺/αvβ3⁺ group and FN⁻/αvβ3⁻ group (P < 0.001, Fig. 3g) as well as between the single high-expression groups (FN⁺/αvβ3⁻ plus FN⁻/αvβ3⁺) and the FN⁻/αvβ3⁻ group (P < 0.001) was statistically significant. A significant difference in OS time was also noted between the FN⁺/αvβ3⁺ group and the FN⁻/αvβ3⁻ group (P < 0.001, Fig. 3h), the single high-expression groups (FN⁺/αvβ3⁻ plus FN⁻/αvβ3⁺) and the FN⁻/αvβ3⁻ group (P = 0.022), and between the FN⁺/αvβ3⁺ group and the single high-expression groups (P = 0.005, Fig. 3h).

To explore the independent prognostic ability of FN and αvβ3 co-expression, three clinicopathological factors (i.e., tumor size, Enneking staging, and response to chemotherapy) that were significant predictors of survival time in univariate analysis (Table 6, Fig. 4) were evaluated by multivariate analysis. As shown in Table 7, according to multivariate analysis, FN⁺/αvβ3⁺ independently predicted worse DFS (hazard ratio (HR) = 2.66, P = 0.025, C-index = 0.75) and OS (HR = 3.75, P = 0.011, C-index = 0.84) of patients with osteosarcoma compared with other groups (FN⁺/αvβ3⁻ plus FN⁻/αvβ3⁺ plus FN⁻/αvβ3⁻). Compared with FN⁻/αvβ3⁻, FN⁺/αvβ3⁺ exhibited a statistically more significant predictive value for shorter DFS (HR = 5.62, P = 0.002, C-index = 0.79) and OS (HR = 6.35, P = 0.010, C-index = 0.85). Similarly, high expression of FN and αvβ3 individually (FN⁺/αvβ3⁻ plus FN⁻/αvβ3⁺) was significantly associated with worse DFS compared with FN⁻/αvβ3⁻. Therefore, co-expression of FN and αvβ3 was significantly correlated with poor DFS and OS in osteosarcoma patients.