The clinical and pathological data of 79 patients who were diagnosed with colorectal cancer at Dukes' stage B or C, and underwent radical surgery at Kunshan First People's Hospital (Kunshan, China) between January 2008 and December 2010 were reviewed. The mean age was 59.5 years, with 38 males and 41 females. None of the patients had received radiotherapy or chemotherapy prior to surgery. Thirty pairs of fresh-frozen colorectal tumors and matched normal tissues were also collected for total protein extraction and stored at −80°C. Written informed consent was obtained from all patients and the study was approved by the ethical approval of Kunshan First People's Hospital Ethics Committee. Follow-up data were available for all patients and the duration ranged between 3 and 60 months with a mean of 40.29±2.42 months.

In each case, three representative tumor regions were selected, from which tissue cylinders with diameters of 0.6 mm were arrayed into a recipient block using a tissue chip microarrayer (Beecher Instruments, Inc., Silver Spring, MD, USA). Subsequently, the recipient block was cut into 5-µm thick sections on slides pretreated with adhesion agent (Jiangsu Haimen Shitai Experimental Equipment Co. Ltd., Haimen, China) to support adhesion of the tissue samples.

Thirty pairs of fresh-frozen CRC specimens and corresponding adjacent normal tissues were used for western blotting. Total protein of each tissue was extracted using RIPA lysis buffer (Beyotime Institute of Biotechnology, Haimen, China) for 10–15 min at 0°C, and then the supernatant was collected, in which the concentration of protein was measured using a BCA protein assay kit (Pierce; Thermo Fisher Scientific, Inc., Waltham, MA, USA). Supernatants from each sample were all mixed with 5X SDS-PAGE Sample Loading Buffer (Beyotime Institute of Biotechnology, Haimen, China) and boiled for 8–10 min at 96°C. A total of 15 mg sample was resolved per lane of 6–12% SDS-PAGE, transferred to polyvinylidene difluoride membranes (EMD Millipore, Billerica, MA, USA), and then blocked in 5% non-fat dry milk with TBS-Tween 20 for 1–2 h at room temperature (depending on seasonal changes in room temperature; 1 h in summer, 2 h in winter). Subsequently, the primary antibodies, mouse anti-human SPARCL1 polyclonal antibody (1:500; cat. no. ab107533, Abcam, Cambridge, UK) and mouse β-actin monoclonal antibody (1:500; cat. no. AA128, Beyotime Institute of Biotechnology), were used to incubate the membranes in 4°C overnight. Accordingly, horseradish peroxidase (HRP)-conjugated goat anti-mouse IgG secondary antibody (1:1,000; cat. no. A0216, Beyotime Institute of Biotechnology) were used to incubate the membranes for 1 h at 37°C. Membranes were then detected using an Enhanced Chemiluminescence Detection system (Beyotime Institute of Biotechnology). The relative densities of proteins were quantified using ImageJ software (version 1.8.0; National Institutes of Health, Bethesda, MD, USA). The formula using to calculate the relative SPARCL1 expression was as follows: Gray value (SPARCL1)/gray value (β-actin).

A mouse anti-human SPARCL1 polyclonal antibody (cat. no. ab107533; 1:200; Abcam, UK) was used as the primary antibody. A Streptavidin-HRP kit (CW2069A; CWBio, Beijing, China) was used for IHC according to the manufacturer's protocol. The slides were deparaffinized, rehydrated and heated in a microwave for 10–15 min in 10 mmol/l citrate buffer (Sigma-Aldrich; Merck KGaA, Darmstadt Germany). After the slides were cooled to room temperature, the slides were treated with reagent1 and reagent2 to block non-specific staining, and then treated with the primary antibody for 12 h. Slides were then incubated with reagent3 and reagent4 (anti-rabbit/mouse IgG) for 5 min, then treated with DAB and hematoxylin for visualization.

A stained TMA slide was scanned using electron microscopy and analyzed with ImageScope software (version 11; Leica Microsystems GmbH, Wetzlar, Germany). Protein expression was assessed in a semi-quantitative manner by two pathologists (Dr Xiao-jiao Gao and Dr Fang Chen; Department of Pathology, Kunshan First People's Hospital Affiliated to Jiangsu University), who were unaware of any patient information. The percentage of tumor cells with staining of the cytoplasm was evaluated as follows: 0%, 0; 1–10%, 1; 11–50%, 2; 51–80%, 3; 81–100%, 4. The staining intensity was also evaluated as follows: No expression, 0; weak, 1; moderate, 2; or strong, 3. The values were multiplied, resulting in an immunoreactivity score (IRS) ranging between 0 and 12. Mean scores calculated from three TMAs were the final numeric values. Samples were divided into negative/under-expressed (IRS<3) and positive/over-expressed (IRS>2) levels according to SPARCL1 expression. Any disagreement was resolved by discussion.

Data are presented as the means ± standard deviation. The SPSS software (version 20.0; IBM Corp., Armonk, NY, USA) was used for statistical analyses and P<0.05 was considered to indicate a statistically significant difference. Paired Student's t-tests were used to compare SPARCL1 protein expression in tumors with normal tissues. Pearson's Chi squared and Fisher's tests were used to analyze the associations of SPARCL1 expression with clinicopathological characteristics. In addition, the Cox univariate and multivariate regression analyses, and Kaplan-Meier curves with log rank test were also used to analyze overall survival (OS). Hazard ratios (HRs) with 95% confidence intervals (CI) were used as the main indicate for OS.

SPARCL1 protein expression in thirty paired CRC and normal tissues was visualized by western blot analyses. The data of four representative samples derived from a total of thirty paired specimens are presented in Fig. 1A. The comparison of average relative SPARCL1 expression rates between total thirty tumors and normal tissues are summarized in Fig. 1B (the relative SPARCL1 expression of tumors vs. normal tissues, was 1.370±0.185 vs. 0.901±0.136; P<0.0001), which indicated a significantly lower expression pattern of SPARCL1 protein in colorectal tumors compared with corresponding normal tissues.

A total of 79 patients with radically resected Duke's stage B or C tumors, were included in the present study. Of the 79 patients, 48 cases (60.76%) were negative expression for SPARCL1 (Fig. 2A and B) and the other 31 (39.24%) were positive (Fig. 2C and D). Significant differences were identified in the SPARCL1 expression status in relation to differentiation and Duke's stages (P<0.05; Table I), but no significant differences were identified among other parameters. The associations between clinicopathological features of patients with colorectal cancer and SPARCL1 expression are presented in Table I.

Although no difference between positive and negative expression status was identified in overall survival analysis, and no significance in SPARCL1 expression was detected in the Cox regression analysis (P>0.05; Tables I and II; Fig. 4A), immunohistochemical staining scores of patients who survived were significantly higher compared with those who succumbed (Fig. 3A). In addition, scores of patients at Duke's stage B who survived were significantly higher compared with patients with stage C (Fig. 3). Furthermore, the following Kaplan Meier curves demonstrated that positive expression of SPARCL1 was a potentially good indicator of prognosis at Duke's stage B, but not at stage C (P=0.038 and P=0.872, respectively; Fig. 4). The HR of overall survival calculated by log rank test, was 0.5755 (95% CI, 0.3108–1.045; P=0.078). The HR of patients at stage B was 0.3850 (95% CI, 0.1603–0.9277; P=0.038). The HR of patients at stage C was 0.9337 (95% CI, 0.3928–2.198; P=0.872).