The presence of tumor associated macrophages in tumor stroma as a prognostic marker for breast cancer patients

The breast cancer cohort analyzed consists of 144 patients diagnosed with invasive breast cancer at Skåne University Hospital, Malmö, Sweden, between 2001 and 2002. The cohort and TMA construction have previously been described in detail [15‐17]. The majority, 109, of the patients had luminal A breast cancer (79%), while 15 patients (11%) had triple-negative/basal-like breast cancer. Immunohistochemical staining of ER, PR, and human epidermal growth factor receptor (HER) was performed as previously described [18] and used as surrogate markers for molecular subtypes [19]. Mean age was 65 years (range 34-97). During follow-up, 41 patients (28%) died and 29 patients (20%) had recurrence. Median follow-up was 6.55 years (range 0.33-7.55 years) for the full cohort and 6.74 years (range 4.88-7.55) for patients alive. Ninety-six patients (67%) had endocrine therapy: of whom 3 patients (2%) had been given aromatase inhibitors (AI) only, 67 (47%) tamoxifen only and 25 patients (17%) both AI and tamoxifen, and in 1 (1%) case unspecified. Of the 30 patients (21%) receiving chemotherapy, 26 (87%) had a combined treatment with fluorouracil, epirubicin and cyclophosphamide (FEC). Ethical approval for the use of breast cancer specimens for this study was obtained from the Ethics Committee at Lund University (ref no 447-07), whereby written consent was not required and patients were offered the option to opt out.

Four μm thick TMA sections were mounted onto glass slides and deparaffinised followed by antigen retrieval using the PT-link system (DAKO, Glostrup, Denmark) and then stained in an Autostainer Plus (DAKO) with the EnVisionFlex High pH-kit (DAKO). Primary antibodies included: anti-CD163 (10D6 dilution 1:250; Novocastra), anti-CD68 (dilution 1:1500; DAKO) and anti-DC-LAMP (also known as CD208; a marker for mature myeloid dendritic cells; clone 101E1.01 dilution 1:1000; Dendritics). Immunohistochemistry (IHC) had previously been performed on sections from the same TMA blocks with the anti-Granulin antibody HPA028747 (expressed in tumor-resident bone marrow derived cells responsible for instigation; 1:100; AtlasAntibodies) and the outcome of immunoreactivity was assessed by image analysis [15]. The estimated fraction of cells with nuclear Ki67 expression was denoted as low (0-10%), intermediate (11-25%) and high (>25%) according to previous studies [20]. ER-negativity and PR-negativity was defined as <10% positively staining nuclei, according to current clinical guidelines in Sweden. The CD163 and CD68 staining was scored as the infiltration density of CD163⁺ and CD68⁺ cells with a monocyte/macrophage morphology, ranging from 0 (absent) up to 3 (dense). For statistical analyses, these categories were dichotomized into absent/sparse (0-2) or dense (3) macrophage infiltration. The CD163⁺ and CD68⁺ macrophages were scored in the TS and TN separately. The amount of TS content was scored into the categories ≤50% or >50% of the tumor core as visualized by HE-staining and microscopic evaluation. The CD208/DC-LAMP staining was scored as the presence of CD208⁺ cells with a DC morphology ranging from 0 (absent) – 1 (present). The CD208⁺ cells were present in 25% of all cases (31 out of 121) and were located in the peri-tumoral T cell rich areas only.

The gene expression levels of CD163 and CD68 in different breast cancer patient groups were analyzed using microarray profile sets, [GenBank:GDS1329] [21] and [GenBank:GDS806] [22], from NCBI Gene Expression Omnibus profiles [23].

Spearman’s Rho and χ² tests were used for comparison of CD163 and CD68 expression and patient and tumor characteristics. Kaplan-Meier analysis and log rank tests were used to illustrate differences in recurrence free survival (RFS), breast cancer specific survival (BCSS) and overall survival (OS) according to CD163 and CD68 expression. Cox regression proportional hazards models were used for estimation of hazard ratios (HR) for death from breast cancer or overall causes according to CD163 and CD68 expression in both uni- and multivariable analysis. Covariates with a P value ≤ .05 in the univariable analysis were included in the multivariable analysis. All statistical tests were two sided and P ≤ .05 was considered significant. Calculations were performed with IBM SPSS Statistics version 19.0 (SPSS Inc).

Due to loss of material or low tumor cell content, 121 (84%) samples were annotated for CD163 in TS, 105 (73%) for CD163 in TN and 108 (75%) samples were scored for CD68 in both TS and TN. As shown in Figure 1A-D, CD163⁺ and CD68⁺ cells with a macrophage-morphology were present in both TS and TN of primary breast cancers. There was a strong correlation between CD163 and CD68 in both TS (P<.001) and TN (P<.001). There was no correlation between the infiltration density of macrophages in TS and the infiltration density of macrophages in the TN. However, CD163 in TN correlated to the presence of CD208⁺ cells in peri-tumoral T cell zones (P = .009). CD208 (DC-LAMP) is a marker for mature myeloid dendritic cells (MDC), which could be reasoned to express CD163 [24]. CD208⁺ cells were only found in the T cell rich peri-tumoral areas and not in TN (Figure 1E).

To analyze if the localization of CD163⁺ and CD68⁺ macrophages had any correlation to clinical characteristics, the distribution of CD163⁺ and CD68⁺ macrophages in TS and TN was evaluated separately (Table 1 and see Additional file 1). The staining categories initially scored from 0 to 3 in macrophage infiltration density were further dichotomized into two groups with absent/sparse (0-2) or dense (3) macrophage infiltration. Seventeen percent of the tumors were denoted as having dense infiltration of CD163⁺ macrophages in TS, while 9% had dense infiltration in TN. Nine percent had dense infiltration of CD68⁺ macrophages in TS, while 6% had dense infiltration in TN. Hence, the majority of tumors had absent or sparse macrophage infiltration.

CD163 ⁺ and CD68 ⁺ macrophages were more equally distributed among the patients with luminal A breast cancer (see Additional file 1). The majority had absent/sparse macrophage infiltration density in both the TS and TN. Dense infiltration of CD163⁺ and CD68⁺ macrophages in TS was observed in only 8% and 6% of the cases respectively. Eighty percent of the triple-negative/basal-like breast cancer patient had dense infiltration of CD163 ⁺ macrophages in TS, while 23% had dense infiltration of CD68 ⁺ macrophages in TS. This was not due to an increase in the amount of TS since we saw an inverse correlation between the amount of TS and the density of TS-associated CD163⁺ and CD68⁺ TAMs (see Additional file 2). We did not find any correlation between the amount of TS and breast cancer subtypes in our cohort (data not shown). The majority of the cases (92% for CD163⁺ and 100% for CD68⁺) had absent/sparse macrophage infiltration in TN.

Breast cancer tumors with dense infiltration of CD163⁺ macrophages in the TS were of higher grade (P<.001), larger size (P<.001) and had a higher proliferation index as indicated by Ki67 positivity (P = .007). Dense infiltration of CD163⁺ macrophages in the TS was further associated with estrogen receptor (ER) negativity (P = .001), progesterone receptor (PR) negativity (P<.001), triple-negative/basal-like breast cancer (P<.001), inversely correlated with luminal A breast cancer (P<.001) (Table 1) and correlated with the extent of granulin (GRN) expression (P = .01) (see Additional file 3). CD163⁺ macrophages in TN did not correlate with any clinicopathological features.

To further evaluate the association between CD163⁺ macrophages and different breast cancer subtypes, we analyzed the gene expression levels of CD163 in both basal-like and luminal breast cancer, using a publically available gene expression array dataset [GenBank:GDS1329] [21] from NCBI Gene Expression Omnibus profiles [23]. In line with the findings from the TMA-based analysis, basal-like breast cancer had significantly higher gene expression levels of CD163 (P<.001) compared to luminal breast cancer (Figure 1F), but also of CD68 (P<.05) (data not shown).

Dense infiltration of CD68⁺ macrophages in the TS positively correlated with large tumor size and high grade and inversely correlated with luminal A breast cancer. There was no significant association between CD68⁺ macrophages in TN and any clinicopathological features.

We found that dense infiltration of CD163⁺ and in particular CD68⁺ macrophages in TS correlated with poor OS (Figure 2A-D and Table 2) and poor BCSS (Table 2). CD68⁺ macrophages in TS further correlated to recurrence (Figure 3A and Table 2). Using a publically available gene expression array dataset [GenBank:GDS806] [22] from NCBI Gene Expression Omnibus profiles [23] we were able to identify that breast cancer patients receiving endocrine therapy, who had recurrence, had significantly higher gene expression levels of CD68 (Figure 3B).

There was no observed correlation between CD163⁺ or CD68⁺ macrophages in TN with OS, BCSS or RFS (Table 2).

Since CD163⁺ macrophages in the TS positively correlated with triple-negative/basal-like breast cancer and inversely correlated with luminal A breast cancer (Table 1 and 2), the prognostic value of CD163⁺ macrophages in TS was evaluated separately in the two different breast cancer subtypes (Figure 4A, B). Luminal A breast cancer patients with dense infiltration of CD163⁺ macrophages had a worse OS (Figure 4A). The majority (74%) of the luminal A breast cancer patients received endocrine therapy.

There was no difference in patient outcome according to CD163⁺ macrophage infiltration in TS in the triple-negative/basal-like breast cancer patient group (Figure 4B).

Multivariate analyses revealed that CD163 was not an independent risk factor for OS, BCSS or RFS (Table 3). Dense infiltration of CD68⁺ macrophages in TS was not an independent risk factor for OS and RFS but was an independent risk factor for BCSS (HR = 0.12; 95% CI, 0.02 to 0.72; P = .02), (Table 3).