Amphiregulin Is a Prognostic Factor in Colorectal Cancer

Materials and Methods

Human tissues. The study population consisted of 174 consecutive patients who underwent curative resection of colorectal cancer at the Department of Surgery, Kurume University Hospital from January 2002 to December 2004. Patients with a second cancer were not included. The patients' characteristics are described in Table I. The median follow-up duration was 80 months (ranging from 2 to 118 months). Patients had stage 0 to IIIC disease (TNM 7th). There were 18 cases of recurrence, seven cases of liver metastasis, seven cases of lung metastasis, and seven cases of local recurrence. Adjuvant chemotherapy was used in 61 cases (35%).

Immunohistochemistry. For amphiregulin immunostaining with goat polyclonal antibodies, tissue sections (3 μm) were deparaffinized in xylene and rehydrated in an ethanol series. The sections were then treated for 5 min with 3% hydrogen peroxide to block endogenous peroxidase activity. The sections were unmasked in Tris/EDTA buffer, pH 9 (DAKO Target Retrieval Solution) in an autoclave for 10 min, at 105°C, and were subsequently washed with phosphate-buffered saline (PBS). The sections were incubated with rabbit serum for 30 min at room temperature and were then incubated with the primary antibody [goat polyclonal antibody to amphiregulin (1/200): R&D: McKinley Place NE Minneapolis, MN, USA] for 30 min, at room temperature. The bound primary antibodies were detected by adding anti-goat secondary antibodies and an avidin/biotin/horseradish peroxidase complex (VECTOR: Ingold Road, Burlingame, CA, USA) for 30 min, at room temperature. The sections were visualized using solid diaminobenzine diluted in PBS, counterstained with Mayer's hematoxylin, and were finally mounted.

Evaluation of immunostaining. The slides were graded according to the staining intensity and the percentage of immunopositive cells, as previously described (5). Specific staining with postimmune serum was semiquantitated by assigning a score of 0 to 3 based on the color intensity of the brown diaminobenzidine precipitate, as 1, representing light brown staining; 2, a moderately brown color; or as 3, an intense brown color. The number of positive cells per slide was stratified into three groups based on the percentage of positive cells: group 1 with <33%; group 2 with 33 to 67%; and group 3 with >67%. Semiquantitative scores ranging from 1 to 9 for the specific staining of each specimen were obtained by multiplying the staining intensity by the number of the group that represented the percentage of positive cells within each specimen. A score of zero represents no specific staining (Figure 1).

Data analysis. The association between the results of the amphiregulin staining and each of the clinicopathological factors were tested using the chi-squared test or t-test. Cox proportional hazard model was employed to examine the effect of amphiregulin on overall survival (OS) and relapse-free survival (RFS). Before accessing the effect of amphiregulin on OS and RFS, we used the survival tree model (6) to stratify the study sample into a number of clinically meaningful and homogeneous subgroups based on the clinicopathological factors. These strata were included in the Cox model to control confounding. When the number of events is low with a large number of potential confounding variables (7), the tree-based method such as the survival tree model may be a practical method for controlling confounding. The R (version 2.14.1) package RPART (recursive partitioning and regression trees) by Terry Therneau and Beth Atkinson (http://www.mayo.edu/biostatistics) was used to conduct the survival tree modeling. Other analyses were carried out by JMP 9.0.0 (SAS: roppongi, minatoku, Tokyo, Japan).