Paraffin-embedded tissues were obtained from the Department of Pathology of the Chungbuk National University Hospital. Three hundred and two CRC patients (mean age, 63.6 ± 11.1 yr; range 25-86 yr; male-to-female ratio, 181:121), who underwent complete resection (R0) and were followed for more than 5 yr, were enrolled in this study. The patients did not receive any chemotherapy or radiation therapy before surgery. The CRCs were located in the right colon in 69 cases, in the left colon in 70 cases, and in the rectum in 160 cases (Table 1).

For each tumor, medical records and tissue slides were re-evaluated consistently by one pathologist. Tumor size, cell grade, depth of tumor invasion, lymph node or distant metastasis, and microsatellite status (characterized in 146 cases) were assessed. Microsatellite status was evaluated by demonstrating the expression of specific microsatellite repeats by polymerase chain reaction (PCR). Briefly, PCR was performed using DNA extracted from resected CRC tissues about two mononucleotide repeats (BAT25, BAT26) and three dinucleotide repeats (D5S346, D2S123, D17S250). MSI High (MSI-H) indicates that 2 or more microsatellites show instability, MSI low (MSI-L) indicates only one microsatellite shows instability and microsatellite stable (MSS) indicates the absence of unstable microsatellites (13). We categorized the CRCs as MSI-H versus MSI-L or MSS (14). The stages were defined according to the TNM staging system based on the American Joint Committee on Cancer (AJCC) 7th edition (15). We also investigated invasion into lymphatics or vessels and perineural invasion. Types of tumor growth (expanding and infiltrative) and the presence of tumor budding (defined as the presence of clusters of less than 5 tumor cells (16)) were evaluated. The number of buds was counted in one field of the invasive margin that demonstrated intense tumor budding in a high power field (× 200 magnification). A field with ≥ 1 bud was considered positive.

After the slide review, 3-mm tissue micro-arrays (TMA) were reconstructed. Regions near the tumor center and near the invasive margin of CRC, which demonstrated the lowest degree of differentiation and had an abundance of mitotic cells, were chosen from the original blocks. An experienced pathologist marked these representative areas on hematoxylin and eosin (H&E)-stained slides from the selected paraffin blocks. Sections of 4 µm thickness were cut and IHC was performed using the BenchMark XT automated IHC stainer (Ventana Medical Systems, Inc., Tucson, AZ) and the Ventana Ultraview DAB Kit (Ventana Medical Systems) was used for detection. For IHC, the sections were deparaffinized using the EZ Prep solution and antigen retrieval was done using the CC1 standard automated process (pH 8.4 buffer containing Tris/Borate/EDTA). The DAB inhibitor (3% H₂O₂ endogenous peroxidase) was blocked for 4 min at 37℃. The slides were incubated with the respective primary antibodies for 30 min at 37℃, followed by incubation with the secondary antibody (Universal HRP Multimer) for 8 min at 37℃. Subsequently, the slides were treated with the DAB + H₂O₂ substrate for 8 min followed by counterstaining with hematoxylin and the bluing reagent at 37℃. The reaction buffer (pH 7.6 Tris buffer) was used as a wash solution. The primary antibodies used included: anti-podoplanin (Dako Cytomation, Glostrup, Denmark, 1:50), anti-α-SMA (Dako Cytomation, Glostrup, Denmark, 1:400), and anti-S100A4 (Dako Cytomation, Glostrup, Denmark, 1:800). Dual staining of podoplanin (cytoplasm, Red) plus S100A4 (nucleus, DAB/brown) and α-SMA (cytoplasm, Red) plus S100A4 (nucleus, DAB/brown) were also performed. For the dual staining, the slides were incubated with primary antibodies (S100A4) for 32 min at 37℃, followed by the secondary antibody of Universal HRP Multimer for 8 min at 37℃. The slides were then treated with the DAB substrate for 8 min at 37℃ followed by denaturation at 90℃ for 4 min. The slides were incubated with primary antibodies (anti-podoplanin or α-SMA) for 32 min at 37℃ and the secondary antibody of Universal AP Multimer for 12 min at 37℃. Slides were sequential treated with UV Red enhancer for 4 min at 37℃, UV Fast Red A plus UV Red naphthol for 8 min at 37℃, and UV Fast Red B for 8 min at 37℃, and finally counterstained using hematoxylin for 4 min and bluing reagent for 4 min at 37℃.

For the identification of PDPN-, α-SMA-, and S100A4-expressing CAFs among α-SMA-positive smooth muscles of muscularis mucosa, muscularis propria and blood vessels, PDPN-positive lymphatic endothelial cells, and S100A4-expressing inflammatory cells, the distinct histological morphology was carefully examined. Muscularis mucosa and muscularis propria appeared as smooth muscle bundles and the blood vessels and lymphatic channels were identified by their vascular or lymphatic lumens. In addition, the CAFs appeared as spindle shaped stromal cells with long cytoplasmic extensions, which were different from the lymphocytes characterized by a small, round nucleus with scanty cytoplasm (Fig. 1).

The stained slides were semiquantitatively evaluated according to the method described previously (17); stromal staining was semiquantitatively assessed as 0 (absence of or weak immunostaining in < 1% of the tumor stroma), 1 (focal positivity in 1%-10% of stromal cells), 2 (positive immunostaining in 11%-50% of stromal cells, and 3 (51%-100% stromal staining). Assessment of the entire tumor was made at the tumor center and the invasive front, or areas of active tumor growth. For statistical analysis, PDPN⁺ CAFs were classified as negative (score 0) and positive (score 1-3), while α-SMA or S100A4 CAFs were classified as low expression (score 0 or 1) and high expression (score 2 or 3).

The patient outcomes were overall survival and disease-free survival as measured from the time of surgery and were estimated by the Kaplan-Meier method. The log-rank test and the Cox proportional hazard model were used to conduct univariate and multivariate analyses. The relationship between the CAF staining intensity (measured semiquantitatively) and the clinicopathological parameters were analyzed using the Fisher's exact test, the chi-square test, Student's t test, or the Mann-Whitney U test. All statistical tests were 2-sided, and statistical significance was accepted at P < 0.05. All analyses were performed using the SPSS version 12.0 (SPSS Inc., Chicago, IL, USA).

The study protocol was reviewed and approved by the institutional review board of the Chungbuk National University College of Medicine (IRB No. GR 2011-06-009). Written informed consent was obtained from all the study participants.

Representative pictures of PDPN, α-SMA, and S100A4 staining are shown in Fig. 1. All the study cases presented with CAFs expressing S100A4 or α-SMA in the stroma of both the tumor center and the invasive front, while 74% of PDPN⁺ CAFs were identified at the tumor center and 42% were at the invasive front among the cases examined. Dual-stained CAFs expressing PDPN plus S100A were less frequently identified (10%) than those expressing α-SMA plus S100A4 (40%) at the tumor invasive front. The immunoreactivity for PDPN was exclusively confined to the stromal fibroblasts of both the intra- and the peri-tumoral stroma, and was sparse in the stromal cells that surrounded the cancer cell budding in the tumor nests of the invasive front. On the other hand, CAFs expressing α-SMA or S100A4 demonstrated a similar pattern, both at the tumor center and at the invasive margin (Table 2).

PDPN⁺ CAFs at the tumor invasive front were inversely related to the pre-operative carcinoembryonic antigen (CEA) level (P = 0.020), tumor size (P = 0.039), T-stage (P = 0.002), and AJCC stage (P = 0.033). They were frequently observed in tumors that exhibited an expanding tumor border (P = 0.001), absence of tumor budding (P = 0.001), and high MSI (P = 0.010). The presence of PDPN^- CAFs with high α-SMA expression at the tumor invasive front was associated with the pre-operative CEA level (P = 0.032), tumor size (P = 0.011), T-stage (P = 0.001), and AJCC stage (P = 0.020). These tumors frequently presented with an infiltrative tumor border (P = 0.004), presence of tumor budding (P = 0.001), and low MSI or microsatellite stable tumors (P = 0.010). In addition, CAFs with PDPN^- and high S100A4 expression at the tumor invasive front were associated with lymphovascular invasion (P = 0.013) and the presence of tumor budding (P = 0.002) (Table 3).

During the follow up period of 58.6 months (mean), recurrence and death occurred in 58 (19.2%) patients and 59 (19.5%) patients, respectively. Univariate analysis revealed that PDPN⁺ CAFs at the tumor center were significantly associated with prolonged disease-free survival time (P = 0.042 by log rank test, Fig. 2A). Multivariate analysis revealed that the AJCC stage, perineural invasion, and PDPN^- CAFs at the tumor center were independent risk factors of disease-free survival (Table 4). Notably, the PDPN⁺ CAFs at the invasive front demonstrated a prolonged disease-free survival time, although without statistical significance (Fig. 2B). Patients that expressed PDPN^-/α-SMA^high CAFs presented with a shorter disease-free survival time but it was statistically insignificant (Fig. 2C). The presence of PDPN^-/S100A4^high CAFs did not show any relationship with disease-free survival (Fig. 2D). In this study, there was no association between the CAF subtypes observed and overall survival.