Tumor-associated macrophages (TAMs) are candidate histological factors in invasive ductal carcinoma (IDC) of the pancreas. Tumor-associated macrophages can be affected by cancer-related inflammation and pancreatitis and interact with important invasive behavior in a recurrent manner in pancreatic IDC. These features may help elucidate the aggressiveness of pancreatic IDC. The aim of this study was to characterize TAMs in pancreatic IDC in comparison with chronic pancreatitis (CP) and to reveal TAM-related factors and the clinical impact of TAMs. CD68 (a pan-macrophage marker) and CD204 (an M2 macrophage marker) immunohistochemistry was carried out in pancreas head specimens from 107 IDC cases and 11 CP cases. Immunopositive cell areas were calculated at the periphery and center of the tumor. The distributions of macrophages in IDC and CP and the relationship between TAMs and histological tumor factors, survival, and recurrence were evaluated. Macrophages were more frequently observed in the lesion periphery than the center in IDC and CP. The density of macrophages was elevated in IDC compared to CP. Dense M2 macrophages at the tumor periphery were frequently seen in large tumors and showed an independent impact on overall survival and disease-free time. Early recurrence in the liver or the local manipulated area was associated with high accumulation of peripheral M2 macrophages. More M2 macrophages were seen in IDC than in CP in both the periphery and the center. High numbers of peripheral M2 macrophages were associated with large tumor size, early recurrence in the liver, local recurrence, and shortened survival time in patients with pancreatic IDC. (Cancer Sci, doi: 10.1111/j.1349-7006.2012.02411.x, 2012)
The prognosis of patients undergoing resection for pancreatic invasive ductal carcinoma (IDC) remains poor.[1-5] Histological studies have been carried out to elucidate the aggressiveness of pancreatic IDC and have revealed prognostic factors including tumor size, lymph node involvement, nerve plexus invasion, positive resected margin, and low tumor grade.[1-8] Tumor-associated macrophages (TAMs) have recently been reported as a candidate factor in poor prognosis.
Macrophages are the most abundant cancer stromal cells involved in the host immune system, and TAMs have been found to play important roles in tumorigenesis, angiogenesis, matrix remodeling, and metastasis.[11-13] Tumor-associated macrophages have a prognostic impact in prostate, breast, and lung cancers, as well as pancreatic IDC.[9, 14-16] The heterogeneity of macrophages has been discussed with regard to their different responses to various microenvironmental stimuli. Macrophages are classically activated towards the M1 phenotype by lipopolysaccharide and interferon-γ. M1 macrophages are characterized by high expression of pro-inflammatory cytokines, such as interleukin (IL)-1, IL-6, IL-12, and tumor necrosis factor. Alternatively, macrophages are activated towards the M2 phenotype by IL-4, IL-13, and IL-10. M2 macrophages are characterized by high expression of IL-4 and IL-10 and low expression of IL-12. Recent studies have revealed high CD204 expression in M2 macrophages and have shown that TAMs are polarized to the M2 phenotype.[12, 17, 18]
The distribution of TAMs was recently evaluated as a prognostic index in various cancers. A high number of TAMs in the peripheral area of the tumor is correlated with poor prognosis in gastric cancer, hepatocellular carcinoma, and non-small-cell lung cancer, although an increased number of TAMs in the invasive front of colon cancer is associated with favorable prognosis. Increased numbers of TAMs in many cancers are linked to reduced patient survival. In pancreatic IDC, high accumulation of TAMs in the periphery of the tumor is correlated with extrapancreatic invasion, lymph vessel invasion, lymph node involvement, and shortened survival time. Tumor-associated macrophages may be a key to elucidating the aggressiveness of pancreatic IDC. Detailed clinicopathological studies should be carried out to estimate the role of TAMs. First, the distribution of macrophages should be compared between mass-forming chronic pancreatitis (CP) and pancreatic IDC. Macrophages accumulate at the inflammatory site and play crucial roles in the diverse phase.[23, 24] Pancreatitis is prevalent in pancreatic IDC and CP due to obstruction of the main pancreatic duct. Tumor-associated macrophages in pancreatic IDC can be affected by both pancreatitis and inflammatory mediators from tumor cells; macrophages in CP are affected by pancreatitis only. The comparison of macrophages between pancreatic IDC and CP may provide evidence that tumor cells mainly lead to TAM accumulation in pancreatic IDC. Second, TAM-related tumor factors should be examined in detail. Tumor-associated macrophages are attracted to and retained in avascular and necrotic areas where they are exposed to tumor hypoxia.[26, 27] Our previous clinicopathological study showed that tumor necrosis is frequent in large tumors. Tumor size may be associated with TAM accumulation. Identification of the precise TAM-related tumor factors is useful for estimating microenvironmental interactions between TAMs and pancreatic IDC. Third, the impact of TAMs on tumor relapse should be evaluated. The prognostic value of TAMs may indicate that TAMs are predictive markers of recurrence. The impact of TAMs on recurrence will reinforce the clinical significance of TAMs. Finally, multivariate analysis should be carried out to confirm the impact of TAMs on prognosis. The prognostic value of TAMs has only been tested with univariate analysis. Establishment of the prognostic importance needs to show independence among various tumor factors with multivariate analysis.
The aim of this study was to characterize TAMs in pancreatic IDC in comparison with CP and to reveal TAM-related factors and the clinical impact of TAMs on tumor relapse and prognosis.
Materials and Methods
Between September 1992 and December 2007, 116 patients with a pathological diagnosis of pancreatic IDC who underwent a pancreaticoduodenectomy with curative intent at our institution were investigated, because pancreatitis due to obstruction of the main pancreatic duct is evident in the pancreatic head lesions of IDC and CP cases. Three in-hospital deaths, two patients with incomplete follow-up data, two patients who died of non-cancerous causes within 5 years of the pancreaticoduodenectomy (one due to liver cirrhosis and one due to brain infarction), and two patients whose surgical specimens were of poor quality were excluded from the study. The remaining 107 patients were investigated. For the CP cases in this study, 11 patients who underwent pancreaticoduodenectomy during the same period and were pathologically diagnosed with CP were assessed. Chronic pancreatitis was diagnosed according to The Revised Japanese Clinical Diagnostic Criteria for Chronic Pancreatitis. All CP cases showed fibrosis that was distributed primarily in the interlobular spaces, showing a nodular pattern of lobules called cirrhosis due to the disruption of dense interlobular fibrosis or the loss of exocrine parenchyma with irregular fibrosis. All patients signed an institutional review board-approved informed consent form. The median age of the IDC patients was 64.0 years (range, 37–82 years), and 44 were women (41.1%). The median age of the CP patients was 52.0 years (range, 38–72 years), and 1 (9.1%) was a woman (Table 1). None of the 107 IDC patients received neoadjuvant chemotherapy or radiotherapy; 30 received intraoperative radiotherapy, and 10 received adjuvant chemotherapy. Extended lymphadenectomy including regional and peripancreatic lymph node dissection was carried out with pancreaticoduodenectomy, according to the Japanese Classification of Pancreatic Cancer. Combined resection of the portal vein, inferior vena cava, colon, and para-aortic lymph node was carried out for macroscopically curative resection.
Table 1. Characteristics of patients who underwent pancreaticoduodenectomy with curative intent for a pancreatic head tumor
Invasive ductal carcinoma
S-1, an oral anti-cancer drug that combines tegafur, a prodrug of fluorouracil, with 5-chloro-2,4-dihydropyrimidine and potassium oxonate in a molar ratio of 1.0:0.4:1.0 (Taiho Pharmaceutical, Tokyo, Japan). CA19-9, carbohydrate antigen 19-9; CEA, carcinoembryonic antigen; GEM, gemcitabine.
To assess initial recurrence of the tumor, follow-up contrast computed tomography was done every 3 months after surgery or earlier if clinically indicated by examination, symptoms, or a rise in tumor markers, such as serum carcinoembryonic antigen and serum carbohydrate antigen 19-9, which were checked every month. If necessary, further examination such as cytology was carried out to diagnose peritoneal dissemination.
Evaluation of clinicopathological features
Clinical characteristics and pathological examination results were retrieved from the clinical records. Lymphatic (ly), venous (v), and intrapancreatic nerve invasion (ne) were classified into four groups according to the definition of the Japan Pancreas Society and were based on the most extensively involved area observed under low-power magnification (×100): 0, no invasion of cancer cells; (i) invasion of a few cancer cells (1–3 points); (ii) moderate invasion of cancer cells (4–8 points); and (iii) marked invasion of multiple cancer cells (>8 points).
The following clinicopathological factors were investigated retrospectively to assess their impact on survival: age (≤64 years vs >64 years); sex; serum carcinoembryonic antigen (≤3.5 ng/mL vs >3.5 ng/mL), serum carbohydrate antigen 19-9 (≤109 U/mL vs >109 U/mL); grade of tumor differentiation (well vs moderate or poor); tumor size (≤3 cm vs >3 cm); serosal invasion (absent vs present); retropancreatic tissue invasion (absent vs present); portal vein invasion (absent vs present); lymphatic invasion (ly0, 1 vs ly2, 3); venous invasion (v0, 1 vs v2, 3); intrapancreatic nerve invasion (ne0, 1 vs ne2, 3); extrapancreatic nerve plexus invasion (absent vs present); and lymph node involvement (absent vs present).
Antibodies and immunohistochemistry
Paraffin-embedded blocks of tumor at the maximum diameter were cut into 3-μm serial sections. The sections were deparaffinized in xylene, dehydrated in a graded ethanol series, and immersed in 0.3% hydrogen peroxide in methanol for 15 min to inhibit endogenous peroxidase activity. For antigen retrieval, the slides were heated at 95°C for 15 min in a microwave oven (H2800 Microwave Processor; Energy Beam Sciences, East Granby, CT, USA) in 0.1 M citric acid buffer then allowed to cool for 1 h at room temperature. After washing the slides three times in PBS, non-specific binding was blocked by pre-incubating in 2% normal swine serum in PBS (blocking buffer) for 30 min at room temperature. Individual slides were then incubated overnight at 4°C in mouse anti-human CD68 antibody (1:400 in blocking buffer; Dako, Glostrup, Denmark) or mouse anti-human CD204 antibody (Scavenger Receptor class A-E5, 1:400 in blocking buffer; Transgenic, Kumamoto, Japan). The slides were again washed three times with PBS and incubated with EnVision (Dako) for 1 h at room temperature. After extensive washing with PBS, the color reaction was developed with 2% 3, 3′-diaminobenzidine in 50 mM Tris-buffer (pH 7.6) containing 0.3% hydrogen peroxide. The sections were then counterstained with Mayer's hematoxylin, dehydrated, and mounted.
Definition of center of lesion and peripheral site
To identify the center of the lesion, H&E stained sections were scanned at a magnification of ×40, and the margin of the tumor was marked on each slide. The intersection of the major and minor axes was defined as the center of the lesion, and four fields including the center at a magnification of ×100 were defined as the center of the lesion. Peripheral sites were defined as fields that included cancer cells and adjacent non-cancerous cells at a magnification of ×100. In the pancreatitis cases, the same procedure was used to identify the center and the margin of the dense fibrosing area.
Evaluation of immunohistochemistry (IHC)
The IHC-positive cells were quantified by determining the percentage of IHC-positive cells in an area (IHC%) and the IHC-positive cell count, which was generally used to evaluate immunohistochemical staining.
Tumor-associated macrophages identified as CD68- or CD204-positive cells were defined as cells with oval to round nuclei that showed strong membranous/cytoplasmic staining but no nuclear staining. After scanning the immunostained slide at a magnification of ×100, the three areas with the greatest number of macrophages in both the center of the lesion and the peripheral site were selected as hot spots. The Automeasure function in Axio Vision 4.7.1 software (Carl Zeiss, Oberkochen, Germany) was used to distinguish the immunopositive area and to objectively calculate the summed areas of the immunopositive cells in each hot spot at a magnification of ×400. The IHC% (summed area of CD68- or CD204-positive cells/measured area ×100) was then calculated for each site (Fig. 1).
The number of macrophages was counted in three hot spots at ×400 magnification using a micrometer. The mean number of infiltrating macrophages was then calculated.
Correlations between IHC% and macrophage count for CD68 and CD204 in the center of the lesion and the peripheral sites were evaluated using Spearman's rank correlation coefficients. Differences in macrophage infiltration between the two groups were evaluated using the Mann–Whitney U-test. Overall survival time was calculated from the date of pancreaticoduodenectomy to August 24, 2010. Parameters that were significantly associated with disease-free survival (DFS) or overall survival rates evaluated in univariate analyses using log–rank tests were further analyzed with multivariate analysis using the Cox proportional hazard regression model. Crude overall survival curves were plotted using the Kaplan–Meier method. All P-values were two-sided, and the significance level was set at P < 0.05. All statistical analyses were carried out using the Statistical Package for the Social Sciences 11.5 J for Windows software (SPSS Inc., Chicago, IL, USA).
Comparison of the area ratio of IHC-positive cells and IHC-positive macrophage count
To validate auto-measurement of IHC-positive cell areas, the correlation between IHC-positive cell numbers and IHC% was examined. The median CD68 count was 21.0 (range, 1.7–64.0) at the center and 42.0 (range, 13.3–94.3) at the periphery of the lesion. The median CD204 count was 14.0 (range, 0.3–48.3) at the center and 24.7 (range, 4.0–75.3) at the periphery. The CD68% and CD204% strongly correlated with the number of CD68- and CD204-positive cells at the center and the periphery of the tumor in pancreatic IDCs (P < 0.001, R [correlation coefficient] >0.4). To ensure objectivity, auto-measurement of the IHC% was used to quantify immunoreactivity in this study (Fig. S1).
Distribution of CD68- and CD204-positive cells in pancreatic IDC and CP
A series of 107 IDC specimens of the pancreas and 11 specimens of CP were examined for CD68 and CD204 expression in the center and periphery of the lesion. In the IDC series, the median CD68% was 3.65% (range, 0.05–18.6%) at the center of the lesion and 9.92% (range, 0.37–25.1%) at the periphery, whereas the median CD68% of the CP series was 1.62% (range, 0.55–6.20%) at the center of the lesion and 2.29% (range, 1.13–19.5%) at the periphery (P = 0.031 at the center, P = 0.002 at the periphery). The median CD204% was 1.64% (range, 0.06–18.1%) at the center of the lesion and 3.38% (range, 0.27–14.0%) at the periphery in the IDC series, whereas the median CD204% in the CP series was 0.60% (range, 0.26–3.78%) at the center of the lesion and 1.59% (range, 0.32–3.54%) at the periphery (P = 0.018 at the center, P = 0.008 at the periphery). In each series, CD68- and CD204-positive cells were more frequently observed in the periphery than at the center of the lesions (Fig. 2). The CD204/CD68 ratios at the center and periphery were compared between IDC and CP cases to evaluate the population of cells with the M2 phenotype. In IDC cases, the median CD204/CD68 ratio was 67.6% (range, 3.6–185.4%) at the center of the lesion and 59.9 (range, 2.1–158.5%) at the peripheral sites, whereas the median CD204/CD68 ratio was 47.3% (range, 12.2–96.9%) at the center and 57.6% (range, 18.1–81.7%) at the periphery in CP cases. These differences were not significant (P = 0.238 at the center, P = 0.753 at the periphery; Fig. 2).
Distribution of CD68% and CD204% according to clinicopathological features
The relationship between clinicopathological features and macrophage infiltration was evaluated using Mann–Whitney U-tests (Tables 2, 3). The IDCs with lymph node involvement showed elevated expression of peripheral CD68 (P = 0.045), central CD204 (P = 0.018), and peripheral CD204 (P = 0.003). Cases with tumors >3.0 cm were significantly correlated with high peripheral CD204 expression (P = 0.031), and those with portal vein invasion were significantly correlated with high central CD204 expression (P = 0.012).
Table 2. Distribution of the percentage of the CD68-positive cell area at the center and periphery of lesions in pancreatic tumors according to clinicopathological features
Central CD68%, median (range)
Peripheral CD68%, median (range)
P < 0.05. Differences between the two groups were evaluated using the Mann–Whitney U-test. CA19-9, carbohydrate antigen 19-9; CEA, carcinoembryonic antigen; ly, lymphatic; ne, intrapancreatic nerve; v, venous.
Table 3. Distribution of central and peripheral CD204-positive cell area ratios in pancreatic tumors according to clinicopathological features
Central CD204%, median (range)
Peripheral CD204%, median (range)
P < 0.05. Differences between the two groups were evaluated using the Mann–Whitney U-test. CA19-9, carbohydrate antigen 19-9; CEA, carcinoembryonic antigen; ly, lymphatic; ne, intrapancreatic nerve; v, venous.
Univariate and multivariate analyses of parameters significantly associated with overall survival and DFS
The median IHC% of infiltrating macrophages was used to divide the cases into two groups, high (above the median value) and low (equal to or below the median value). Univariate analyses using log–rank tests were carried out to compare survival according to IHC% (Table 4), and overall survival curves were obtained with the Kaplan–Meier method (Fig. 3). Univariate analysis (Table 4) produced the following candidates for predicting prognosis: tumor size > 3.0 cm (P = 0.0001); lymph node involvement (P = 0.0106); lymphatic invasion (P = 0.0171); extrapancreatic nerve plexus invasion (P = 0.0025); and high central and peripheral CD204 expression (CD204high) (P = 0.0248 at the center, P < 0.0001 at the periphery). Multivariate analysis (Table 5) revealed the following independent prognostic factors: tumor size > 3.0 cm (hazard ratio [HR], 2.017; P = 0.002); extrapancreatic nerve plexus invasion (HR, 1.992; P = 0.002); and peripheral CD204high (HR, 2.781; P < 0.001).
Table 4. Univariate analyses of overall survival (OS) and disease-free survival (DFS) in patients with invasive ductal carcinoma of the pancreas
OS, median (range)
DFS, median (range)
P < 0.05. Univariate analysis (uni) was carried out using the log–rank test. CA19-9, carbohydrate antigen 19-9; CEA, carcinoembryonic antigen; CD68%, summed area of CD68-positive cells/measured area ×100; CD204%, summed area of CD204-positive cells/measured area ×100.
Table 5. Multivariate analyses of independent significant factors associated with overall survival and disease-free survival in patients with invasive ductal carcinoma of the pancreas
P < 0.05. Multivariate analyses (multi) were carried out using the Cox regression hazard model. Central CD204high, percentage of CD204-positive cells area over 1.64%; CI, confidence interval; HR, hazard ratio; Peripheral CD204high, percentage of CD204-positive cells area over 3.39%.
Univariate analysis (Table 4) showed that tumor size > 3.0 cm (P = 0.0058), serosal invasion (P = 0.0427), extrapancreatic nerve plexus invasion (P = 0.0057), and peripheral CD204high (P =0.0010) were correlated with shorter DFS. Multivariate analysis (Table 5) revealed that extrapancreatic nerve plexus invasion (HR, 1.882; P = 0.008) and peripheral CD204high (HR, 1.864; P = 0.010) were independent risk factors for DFS. Initial recurrent sites of IDC were considered to be liver metastasis (n = 38), local recurrence (n = 38), or peritoneal dissemination (n = 20). The DFS curves for these groups were plotted using the Kaplan–Meier method to determine any significant impact of high CD204 expression at the peripheral site. The peripheral CD204high group had a significantly shorter DFS period than the peripheral CD204low group when stratified by initial liver metastasis and local recurrence (Fig. 4).
This was the first study to evaluate the distributions of M2 macrophages (CD204-positive cells) in pancreatic IDC and CP. M2 macrophages preferentially accumulated in peripheral rather than central sites in pancreatic IDC and CP. This finding may indicate that non-cancerous cells play an important role in the recruitment of macrophages and the polarization toward M2 macrophages in pancreatic IDC and CP. In CP, macrophages are recruited using chemoattractants produced by myofibroblasts. Myofibroblasts are considered to be the activated state of pancreatic stellate cells (PSCs), and PSCs are activated by pancreatitis and pancreatic cancer cells. Macrophages in pancreatic IDC may have infiltrated because of chemoattractants produced by myofibroblasts derived from PSCs. The polarization toward M2 macrophages may be responsible for the cells producing IL-4 and IL-10 in both IDC and CP tumors. We considered mast cells and PSCs as candidates. Mast cells accumulate in peripheral areas of IDC and intestinal areas of CP and can produce IL-10. Activated PSCs are abundant in IDC and CP tumors and lead to IL-4 production by T cells. Mast cells and PSCs may play important roles in M2 accumulation in IDC and CP. In this study, most peripheral M2 macrophages in pancreatic IDC were dense along the stroma but not along tumor cells, a finding that may reinforce the above speculation.
Accumulated M2 macrophages in pancreatic IDC were more numerous than in CP. In pancreatic IDC, a large tumor was significantly correlated with dense peripheral M2 macrophages. These results indicate that the tumor volume affects accumulation of M2 macrophages. Recent studies have shown that monocyte recruitment is driven by several chemoattractants such as MIP-2, CCL3, and hypoxia-inducible factor-2α, which are secreted by malignant cells and stromal cells and induced by tumor hypoxia.[26, 27, 37] Tumor-associated macrophages are recruited to tumors by multiple growth factors and chemokines that are often produced by tumor cells themselves.[38, 39] Tumor necrosis is increased in large tumors, and TAMs are attracted to and retained in avascular and necrotic areas where they are exposed to tumor hypoxia.[26, 27] Large tumors may increase expression of inflammatory mediators from tumor cells, stroma cells, and tumor hypoxia. Thus, increased tumor volume may promote accumulation of M2 macrophages.
The independent impact of M2 macrophages on survival and time to relapse was first revealed with multivariate analysis in pancreatic IDC. Dense accumulation of peripheral M2 macrophages was established as a good predictive marker of survival and recurrence. According to the type of initial recurrence, dense peripheral M2 macrophages were associated with early relapse in liver and the manipulated area of the pancreaticoduodenectomy. This suggests that M2 macrophages may accelerate liver metastasis and local recurrence. Tumor-associated macrophages are important producers of proteases, including MMPs, and of a wide variety of growth factors, such as fibroblast growth factor and epidermal growth factor (EGF) receptor family ligands that can stimulate the growth and motility of tumor cells. Tumor-associated macrophages have been reported to be the most significant source of EGF in tumors, and they are associated with EGF receptor expression and poor outcome in breast cancer. Pollard et al. showed that tumor cells respond to macrophage-produced EGF ligands in vivo by chemotaxis and invasion, and that macrophages are often associated with vessels.[38, 42] Thus, M2 macrophages may provide chemotactic signals that recruit tumor cells to blood vessels and enhance their egress into vasculature, leading to tumor hematogenous metastasis and further local invasion. These effects of M2 macrophages may shorten DFS and overall survival.
Lymph node involvement was significantly correlated with high CD204 expression in peripheral sites of the lesion. Tumor-associated macrophages within the invasive tumor front have a profound influence on the regulation of tumor angiogenesis and lymphangiogenesis by production of vascular endothelial growth factor-C and -D.[9, 37, 41, 43] Elevated lymphangiogenesis by TAMs may promote lymph node metastasis.
The independent prognostic values of large tumor size and extrapancreatic nerve plexus invasion were reported in our previous study and reconfirmed by this study. Time to recurrence was associated with the presence of extrapancreatic nerve plexus invasion. Large tumor size did not show an impact on DFS, because high accumulation of peripheral M2 macrophages correlated with large tumor size.
In conclusion, dense M2 macrophages in peripheral sites were significantly correlated with large tumor size, lymph node involvement, and poor prognosis due to accelerated liver metastasis and local recurrence. The number of accumulated M2 macrophages was associated with tumor volume, but the distribution of M2 macrophages in CP was similar to that in IDC.
Supported by Grants-In-Aid for Cancer Research and by a Third Term Comprehensive 10-year Strategy for Cancer Control grant from the Ministry of Health, Labor and Welfare of Japan.