Surgical whole mounts from a total of 100 human breast carcinoma specimens from females diagnosed between January 1987 and December 1988 from the pathology archives at Thomas Jefferson University (Philadelphia, USA) were used in the present study. The average age of patients at the time of surgery was 61.8±16.4 years. Cases with tissue containing ductal carcinoma in situ (DCIS) only, inflammatory breast cancer or other concurrent malignancies were excluded from the present study. Cases with no reactivity to vimentin staining were eliminated from the study. Only cases with fully annotated information regarding demographics, estrogen receptor (ER) expression, histology grade and overall survival (OS) were used in final analyses.

For quality control, cases were first immunohistochemically stained with anti-vimentin monoclonal antibody (cat. no. 550513; BD Biosciences, San Jose, CA, USA) at a 1:250 dilution overnight at 4°C. Those without vimentin reactivity were removed from the study. Double immunohistochemistry was performed using formalin-fixed paraffin-embedded tumor sections (4 µm thickness). Briefly, following deparaffinization and rehydration, antigen retrieval was performed using Envision Flex Target Retrieval Solution (pH 6.1; Dako; Santa Clara, CA, USA) in a pressure cooker for 20 min at 102°C. Endogenous peroxidase and nonspecific epitopes were blocked with 0.3% hydrogen peroxide in absolute methanol for 30 min at room temperature and 5% normal horse serum and 1% normal goat serum (Sigma-Aldrich; St. Louis, MO, USA) for 1 h at room temperature. Sections were incubated with mouse anti-E-selectin monoclonal antibody at 1:100 (cat. no. MO20039; Neuromics, Inc., Minneapolis, MN, USA) overnight at room temperature. Following washing with PBS and subsequent blocking with 5% normal horse serum and 1% normal goat serum for 5 min at room temperature, the slides were incubated with pre-dilute secondary horseradish peroxidase (HRP)-polymer conjugated anti-mouse IgG (cat. no. K4001; Dako) for 30 min at room temperature. HRP was detected using 3–3′-diaminobenzidine (DAB; Biocare Medical LLC, Paheco, CA, USA) substrate for 10 min at room temperature and enhanced using DAB Sparkle (Biocare Medical LLC) for 1 min at room temperature. Residual antibodies were eluted using Denaturing solution (Biocare Medical LLC) at a 1:3 dilution for 3 min at room temperature to ensure no cross reaction between the first and second staining. Slides were blocked with 5% normal horse serum and 1% normal goat serum for 5 min at room temperature and then incubated with mouse anti-CD68 monoclonal antibody at 1:25 (cat. no. M0876; Dako) overnight at 4°C. Following a brief wash with PBS, the slides were incubated with pre-dilute secondary alkaline phosphatase-polymer conjugated MACH2 anti-mouse IgG (cat. no. MALP521; Biocare Medical LLC) for 1 h at room temperature and then visualized using Fast-Red (Biocare Medical LLC) for 7 min, followed by counterstain with Mayer Hematoxylin (Dako) for 4 min both at room temperature. The slides were air-dried and mounted. As a negative control, breast carcinoma tissues were immunostained with the secondary IgG only.

All immunohistochemically stained slides were evaluated by a board certified surgical pathologist (Department of Pathology, School of Medicine, University of Oklahoma Health Sciences Center, Oklahoma City, OK, USA) for breast pathology. The tumors were classified and graded according to the protocol from of the College of American Pathologists (Protocol for the Examination of Specimens from Patients with Invasive Carcinoma of the Breast according to InvasiveBreast 3.3.0.0.) (35). Tumor inflammation was defined as positive or negative, as characterized by the presence of lymphocyte clusters. Immunohistochemical reaction to E-selectin was graded using intensity as a score of 0, 1+, 2+ or 3+ for no, weak, moderate or strong reaction in endothelial cells within the tumor, respectively. Immunohistochemical staining of CD68 was quantitatively categorized as a score of 0, 1+, 2+ or 3+ for no CD68⁺ cells, ≤10 CD68⁺ cells, 11–20 CD68⁺ cells or ≥21 CD68⁺ cells in the observing field at ×200 magnification within the tumor, respectively. All images were viewed under a light microscope (DM2500; Leica, Buffalo Grove, IL, USA) and images were captured using digital cooling color camera (DFC450; Leica).

The Fisher's exact test was used to analyze the association between age and tumor pathological parameters using CD68 and E-selectin expression levels. Spearman's rho was used to determine the correlation between CD68⁺ TAMs, and E-selectin expression level and tumor inflammation. The Kaplan-Meier method was used to estimate OS as a function of time, and differences were analyzed using the log-rank test. Cox proportional hazards regression analysis was used for multivariable analysis of prognostic factors in relation to OS. The statistical software SAS, version 9.4 (SAS Institute Inc., Cary, NC, USA) was used to perform statistical analyses. GraphPad Prism version 6 (Graphpad Software, Inc., La Jolla, CA, USA) was used to generate Kaplan-Meier curves. P<0.05 was considered to indicate a statistically significant difference.

Following exclusions, a total of 53 invasive breast cancer cases that had been first-time diagnosed by surgical resection in 1986–1988 at Thomas Jefferson University were used in the present study. Tumor characteristics of these patients are presented in Table I. Cases were categorized as either high (score 3+) or low (score 0–2+) CD68 expression level according to the abundance of CD68⁺ TAMs in the tumor core and high (score 3+), or low (score 0–2+) expression level of E-selectin on vessels in the tumor area. CD68⁺ TAMs were significantly associated with tumor inflammation (P=0.005) and ER status (P=0.037). E-selectin expression level was associated with age (P=0.016) and was significantly higher among females over the age of 60 years (Table I).

Representative images of high and low expression levels of CD68⁺ TAMs, and E-selectin⁺ inflamed vessels are presented in Fig. 1. Various expression levels of CD68⁺ TAMs were consistently present in the tumor stroma and core of all 53 cases (Fig. 1A-C). Considerable, multilayered CD68⁺ TAM deposition in the necrotic area of the tumors was observed. CD68⁺ TAMs were also abundant in the mammary fat adjacent to the tumor, as well as around the luminal surface of vessels (Fig. 1C). E-selectin was predominantly expressed on vessels within the tumor stroma (Fig. 1D and E). No positive signal for E-selectin was detected in other tumor components, including cancer cells, fibroblasts, immune infiltrates or the apical side of the vessels. The size of E-selectin-expressing vessels varied from small capillaries to large vessels in the stroma and fat adjacent to the invasive front of tumors. Vessel structure was well retained within the stroma but often compressed, crushed or even absent in the tumor core. Overall, 88.7% of the breast carcinoma cases exhibited E-selectin expression on their tumor vessels. Consistent with inflammation of the adipose tissue adjacent to the tumor, E-selectin expression level was also high in vessels of the neighboring peripheral adipose tissue. Of note, vascular E-selectin expression level was present in the stroma surrounding the mammary duct in invasive carcinomas that retained a ductal structure (Fig. 1F). A similar pattern of E-selectin expressing vessels in the stroma around the mammary duct in DCIS only cases was also revealed (data not shown), suggesting the presence of peritumoral inflammation at the pre-invasive stage.

Double immunohistochemistry for CD68 (pink) and E-selectin (brown) was performed to evaluate their association. CD68⁺ TAMs were abundant in close proximity to E-selectin-expressing vessels in the tumor stroma and peripheral tissue adjacent to the tumor (Fig. 2A and B). CD68⁺ TAMs were sparsely present in carcinoma cell rich areas; however, E-selectin expression was limited to the surrounding inflamed area and absent or weakly present in the tumor core (Fig. 2B). CD68⁺ TAMs were also highly abundant in necrotic areas, but E-selectin was absent within and adjacent to the necrotic core (Fig. 2C). TAMs and E-selectin were present at the location where the ductal structure was retained (Fig. 2D). Association between the abundance of CD68⁺ TAMs and E-selectin⁺ vessels was evaluated using a 4-level scoring scale (0, 1+, 2+, 3+) for the expression level of each marker. High abundance of CD68⁺ TAMs and E-selectin⁺ vessels (3+/3+) was demonstrated in 7.5% of the overall analyzed samples. CD68⁺ TAMs and E-selectin expression levels were positively correlated (r=0.30, P=0.030; Table II). Additionally, CD68⁺ TAMs were significantly correlated with tumor inflammation (r=0.54, P=0.001; Table II).

OS was determined and graphically presented using the Kaplan-Meier method, and a Cox proportional hazards regression model was used for multivariable analysis of the association between clinicopathological parameters and marker expression level with OS. Tumor inflammation was significantly associated with OS among patients with breast cancer [hazard ratio (HR), 2.00; confidence interval (CI) 95%, 1.03–4.06; P=0.044; Fig. 3A]. However, inflammation in the tumor periphery lacked association with OS (HR, 1.20; CI 95%, 0.560–2.61; P=0.629; data not shown). Compared with tissue samples with lower expression levels of CD68⁺ TAMs, higher expression levels of CD68⁺ TAMs in the tumor core were significantly associated with shorter OS at the 10-year follow-up (HR, 2.23; CI 95%, 1.12–5.54; P=0.017; Fig. 3B). Conversely, CD68⁺ TAMs in the tumor periphery were not significantly associated with OS (HR, 1.39; CI 95%, 0.59–3.50; P=0.420; data not shown). Although the abundance of E-selectin⁺ vessels and CD68⁺ TAMs were positively correlated, E-selectin expression level alone did not impact OS at the tumor (HR, 0.71; CI 95%, 0.32–1.57; P=0.395; Fig. 3C) or periphery. Kaplan-Meier analysis demonstrated that OS was unaffected by age for CD68, E-selectin or tumor inflammation statuses (data not shown). Finally, multivariable analysis of procedure-naïve breast tumor tissues revealed that the presence of abundant CD68⁺ TAMs was an independent predictor of OS (HR, 2.37; 95% CI, 1.02–5.36; P=0.045) following adjustment for ER status, tumor inflammation and E-selectin expression level (Table III).