Distinct patterns and prognostic values of tumor-infiltrating macrophages in hepatocellular carcinoma and gastric cancer

Archived, formalin-fixed, paraffin-embedded (FFPE) tissues from 188 HCC patients and 138 GC patients who had all undergone radical resection for tumors at the Sun Yat-Sen University Cancer Center between 2002 and 2012 were enrolled in this study. Patients who exhibited signs of distant metastasis and had received anti-cancer therapies before surgery, or experienced concurrent autoimmune disease, were excluded. The diagnosis of HCC and GC in each patient was confirmed histopathologically. The tumor stage was determined according to the tumor-node-metastasis (TNM) classification system of the International Union Against Cancer, 7th Edition. Data was censored at the last follow-up for surviving patients. Overall survival (OS) was defined as the interval between the time of surgery and either the last follow-up or death.

This study conformed strictly to the ethical guidelines of the Declaration of Helsinki and was approved by the Research Ethics Committee of Sun Yat-Sen University Cancer Center. Written informed consent was obtained from all patients before sample collection. All samples were coded and data was stored anonymously. The clinicopathological characteristics of the patients are summarized in Table 1.

IHC was performed using a two-step method (DakoCytomation, Glostrup, Denmark) using protocols described in our previous studies [29, 30]. Sections of FFPE tissues were cut using a microtome, and then sequentially dried, dewaxed, and re-hydrated with xylene and a decreasing ethanol series. Endogenous peroxidase activity was then blocked with 0.3% H₂O₂ for 10 min. For antigen retrieval, sections were steamed in 10 mM citrate buffer (pH 6.0) for 10 min. Glass slides were incubated overnight at 4 °C with anti-CD204 (Transgenic, Kumamoto, Japan), anti-CD169 (R&D Systems, Minneapolis, MN, USA), or anti-CD68 (DakoCytomation, Carpinteria, CA, USA) antibodies. Horseradish peroxidase-conjugated anti-rabbit and anti-mouse antibodies from Dako EnVision systems (DakoCytomation) were used as secondary detection reagents and the immunoreactivities were visualized using 3,3′-diaminobenzidine (DAB). All sections were lightly counterstained with Mayer’s Hematoxylin Solution (Sigma) and mounted using non-aqueous Permount™ mounting medium. Negative controls comprised slides for which the primary antibodies were replaced by the same concentration of an irrelevant, isotype-matched antibody.

Double immunofluorescent staining was carried out as previously described [30]. Briefly, re-hydrated FFPE sections were incubated at 4 °C overnight with mouse anti-human CD68, rabbit anti-human CD204, or sheep anti-human CD169 antibodies. The sections were then incubated for 30 min at 37 °C with a mixture of primary-antibody-matched fluorescently labeled secondary antibodies (Invitrogen; Carlsbad, CA, USA). Nuclei were counterstained using 4′,6-diamidino-2-phenylindole (DAPI).

To quantify CD204⁺ and CD169⁺ cell density, the Vectra-Inform image analysis system (Perkin-Elmer/Applied Biosystems, Foster City, CA, USA) was used, as described in a previous study [28, 30]. Target signals were quantified in selected tissues and cellular compartments of interest. The percentage of each immune cell subset was calculated by dividing the absolute number of each cell subset by area of the tissue surface.

Quantification methods for immunofluorescence were performed as previously described [30]. Immunofluorescence images were captured using a confocal microscope (Olympus, Essex, UK) and analyzed using FV10-ASW Viewer (Olympus, Essex, UK). The number of single-positive or double-positive cells in each of five representative fields at 400× magnification were counted. From these numbers, the proportions of CD204⁺ or CD169⁺ cells in CD68⁺ Mφs were calculated as: (number of CD204⁺CD68⁺ cells)/(number of CD68⁺ Mφs), or (number of CD169⁺CD68⁺ cells)/(number of CD68⁺ Mφs).

OS curves were obtained using the Kaplan–Meier method, and compared using the log-rank test for each prognostic variable. Variables with effects on survival in univariate analysis were included in a multivariate Cox proportional hazard regression model, which was used to estimate the adjusted hazard ratio (HR) and 95% confidence interval (CI), and to identify independent prognostic factors. Subgroups of each immunostaining parameter were divided by the median values. Associations between immunostaining parameters and clinicopathological features were evaluated using the χ ² test or Fisher’s exact test, as appropriate. A threshold of P < 0.05 denoted statistical significance. SPSS 20.0 (IBM) was used for the statistical analyses.

To evaluate the in situ distribution of different Mφ subpopulations, we used immunostaining to detect CD68⁺ Mφs, CD204⁺ Mφs, and CD169⁺ Mφs in the non-tumor (NT) and intra-tumor (IT) areas of HCC and GC. Clear and distinguishable staining was observed for all the phenotypic markers. In HCC, Mφs were evenly distributed in the parenchyma of both the NT and IT regions. In GC, Mφs were gathered in the stromal area surrounded the glandular tubes of gut tissue, but were scattered distributed in the tumor nest (Fig. 1a).

We compared the density of Mφs in the NT and IT regions of HCC and GC. Statistics showed that the numbers of CD68⁺ Mφs were high in the NT of HCC, but were relatively low in the NT of GC tissues, with mean (±SEM) densities of 859 ± 19, and 378 ± 28 in HCC and GC, respectively (P < 0.001; Fig. 1b). However, the density decreased in the IT of HCC (660 ± 28), while it remarkably increased in the IT of GC (604 ± 29). We also compared the distribution of different Mφ subpopulations. In HCC, the densities of both CD204⁺ Mφs (826 ± 77 and 526 ± 23 in NT and IT, respectively; P < 0.001; Fig. 1c) and CD169⁺ Mφs (760 ± 22 and 187 ± 16 in NT and IT, respectively; P < 0.001; Fig. 1d) also decreased in the IT compared with the NT region. In GC, only a few CD204⁺ Mφs were found in the NT, but they were significantly enriched in the IT (31 ± 5 and 411 ± 27 in the NT and IT, respectively; P < 0.001; Fig. 1c). Moreover, CD169⁺ Mφs could be detected in the NT region and were also increased in the IT of GC (242 ± 20 and 514 ± 37 in the NT and IT, respectively; P < 0.001; Fig. 1d). In addition, the ratios of CD204⁺/CD169⁺ Mφs were significantly increased in IT as compared with NT regions of HCC but not in GC (1.2 ± 0.05 and 14.1 ± 1.9 in the NT and IT, respectively; P < 0.01; Fig. 1e). Taken together, the distribution of Mφs in the NT and IT areas differed in HCC and GC.

CD68 is always used as a pan-Mφ marker, while CD204 and CD169 might represent different Mφ subpopulations with pro- or anti-tumor functions during tumor progression. Multiple immunofluorescence staining and confocal analyses confirmed that most CD204⁺ and CD169⁺ cells are CD68⁺ Mφs in both the NT and IT areas of HCC and GC (Fig. 2a; Additional file 1: Figures S1 and S2).

We then examined the proportion of each Mφ subpopulation within CD68⁺ Mφs in HCC and GC. As shown in Fig. 2b, most CD68⁺ Mφs are CD204⁺ (92.8 ± 1.0%) and CD169⁺ (88.4 ± 3.2%) cells in the NT area of HCC; however, the phenotype changed in the IT area, as shown by a significant decrease in the percentage of CD169⁺ Mφs (P = 0.004; Fig. 2b). Moreover, the CD169⁺ (46.3 ± 8.0%) proportion was significantly lower than that of the CD204⁺ (82.3 ± 4.1%; P < 0.001; Fig. 2b) subpopulation within total CD68⁺ Mφs in the IT region of HCC.

The composition of Mφs displayed a different pattern in GC, in that CD68⁺ Mφs comprised fewer CD204⁺ cell (15.7 ± 4.4%) than CD169⁺ cell (60.9 ± 4.3%) in the NT region (P < 0.001; Fig. 2c). In contrast, more CD204⁺ (44.3 ± 6.1%) but fewer CD169⁺ (30.4 ± 8.5%) cells were found in CD68⁺ Mφs in the IT region. Notably, the CD169⁺ subpopulation of Mφs was also significantly decreased in the IT compared with the NT region of GC (P = 0.008; Fig. 2c).

The relationship between the densities of CD204⁺/CD169⁺ Mφs and patient survival was further investigated. Patients were divided into two groups, based on the median value of CD204⁺ or CD169⁺ Mφ density in the IT regions of HCC (median density, 460 and 112 for CD204⁺ and CD169⁺ Mφ, respectively) and GC (median density, 352 and 452 for CD204⁺ and CD169⁺ Mφ, respectively). Kaplan–Meier survival analysis revealed a negative correlation between the density of CD204⁺ Mφs and the OS of HCC patients (P = 0.004; Fig. 3a); however, no significant association was found for GC patients (P = 0.899; Fig. 3a). In contrast, a high density of intra-tumor CD169⁺ Mφs predicted favorable survival in both HCC (P = 0.01) and GC (P = 0.027; Fig. 3b) patients (Additional files 2, 3).

To further assess whether CD204⁺ or CD169⁺ Mφ density could be used as an independent predictor of OS, we performed multivariate Cox proportional hazards analysis. As shown in Table 2, the CD169⁺ Mφ density was associated with a decreased risk of death in HCC (HR 0.561, 95% CI 0.358–0.878, P = 0.011) and GC (HR 0.569, 95% CI 0.343–0.943, P = 0.029) patients. By contrast, the CD204⁺ Mφ density was associated with an increased risk of death in HCC (HR 1.922, 95% CI 1.217–3.034, P = 0.005), but no significant association was found for GC patients (HR 1.033, 95% CI 0.625–1.709, P = 0.899). Clinicopathological variables that were shown to be significant in the univariate analysis were used as covariates in the multivariate analysis. We found that the CD169⁺ Mφ density could act as an independent prognostic factor for OS in both HCC (HR 0.436, 95% CI 0.270–0.703, P = 0.001) and GC (HR 0.587, 95% CI 0.354–0.974, P = 0.039) patients.

We also tested the association of CD204⁺/CD169⁺ Mφ density with CD8⁺ T cell infiltration. As shown in Fig. 4, the density of CD169⁺ Mφ was positively correlated with CD8⁺ T cells infiltration in both HCC and GC (P < 0.0001), indicating the anti-tumor functions of these Mφs in both tumors. However, no association was found for CD204⁺ Mφ in either HCC or GC (P > 0.05). The associations between the CD204⁺/CD169⁺ Mφ density and clinicopathological variables have also been analyzed. The density of CD204⁺ cells was significantly correlated with tumor number, tumor size, TNM stage and histological grade (P = 0.006, P = 0.004, P = 0.004, and P = 0.002, respectively; Table 3) in HCC patients. No significant correlations were found between CD169⁺ cells density and clinicopathological variables in either HCC or GC.

Our findings indicated that the density of CD204⁺ and CD169⁺ Mφs represented a valuable independent factor to predict the prognosis of HCC. Therefore, we analyzed whether the combination of intra-tumor CD204⁺ and CD169⁺ Mφs (namely, the Mφ index) could represent a more powerful criterion for predicting patient prognoses.

In HCC, patients in the CD204^low and CD169^high group exhibited the best OS (5-year OS rate: 90.3%) compared with those in the CD204^low and CD169^low group (5-year OS rate: 64.2%, P = 0.019), CD204^high and CD169^high group (5-year OS rate: 63.0%, P = 0.01) and CD204^high and CD169^low group (5-year OS rate: 38.9%, P < 0.0001; Fig. 3c). A similar trend was found in GC patients, but did not reach statistical significance. In addition, we also analyzed the prognostic value of CD204⁺/CD169⁺ Mφs ratio, and found high CD204⁺/CD169⁺ Mφs ratio was correlated with poor survival in HCC patients (P < 0.001 Fig. 3d). In the multivariate Cox analysis, the Mφ index in HCC was also associated with OS in HCC, but not in GC (Additional file 1: Table S1).