Clinicopathologic and prognostic significance of tumor-associated macrophages in patients with hepatocellular carcinoma: A meta-analysis

Summary

Purpose

Tumor-associated macrophages (TAMs) deserve more focus because of its clinicopathologic and prognostic roles in solid tumors. However, the prognostic value of TAMs in patients with hepatocellular carcinoma (HCC) is still controversial. We performed a meta-analysis to resolve the issue.

Methods

We selected relevant studies from the Cochrane Library, Embase and PubMed databases. The hazard ratios (HRs) and 95% confidence intervals (CIs) were calculated employing fixed-effect or random-effect models depending on the heterogeneity of the included trials. Moreover, we also performed subgroup analysis, cumulative meta-analysis, sensitivity analysis, and bias analysis (Egger’s test).

Results

A total of 20 observational studies with 4297 patients were enrolled. For TAMs subsets, high density of CD68+ TAMs in either intratumor (IT) (pooled HR = 1.417; 95% CI = 1.092–1.839; P = 0.009) or peritumor (PT) (pooled HR = 1.393; 95% CI = 1.022–1.899; P = 0.036) was associated with a poor OS. High density of CD68+ TAMs in IT was also associated with high AFP value, large tumor size, absent encapsulation, present vascular invasion, and later tumor-nodes-metastasis (TNM) stage. High density of CD163+ macrophages in serum was associated with a poor OS (pooled HR = 5.698; 95% CI = 3.062–10.603; P < 0.001). High density of CD204+ TAMs in IT was associated with a poor OS (pooled HR = 1.947; 95% CI = 1.387–2.733; P < 0.001. High density of CD206+ TAMs in IT was associated with a poor OS (pooled HR = 1.723; 95% CI = 1.308–2.270; P < 0.001) and DFS (pooled HR = 1.711; 95% CI = 1.214–2.412; P = 0.002). However, high density of CD169+ TAMs in IT was associated with a good OS (pooled HR = 0.471; 95% CI = 0.343–0.647; P = 0.037).

Conclusions

TAMs could serve as independent predictive indicators and therapeutic targets for HCC. Further trials are needed to elucidate the exact relationship and the underlying mechanism.

Keywords:

Cytokines – Cytotoxic T cells – Database searching – Macrophages – China – hepatocellular carcinoma – Prognosis

Introduction

Hepatocellular carcinoma (HCC) is the fifth most common malignant tumor in the world with approximately 850,000 new cases every year and represents the third leading cause of global cancer deaths and the incidence is raising [1, 2]. Surgical resection is the primary therapy for HCC. Following surgery, transcatheter arterial chemoembolization was taken as effective supplement measure to prevent or reduce the recurrence, especially for high risk liver cancer. Radiofrequency ablation, targeted medical therapy and Chinese medicinal therapy were also choices for advanced liver cancer [3]. However, high rates of recurrence and metastasis after liver resection result in poor prognosis. Early diagnosis and comprehensive treatment in time can increase the efficacy of HCC. Therefore, it is critical to explore and validate new markers of HCC with high sensitivity and specificity.

Macrophages are terminally differentiated cells that reside in all tissues including tumors [4]. They exist two main functional phenotypes, the type 1 (M1, classical) and type 2 (M2, alternative) phenotype [5]. Tumor-associated macrophages (TAMs) refer to the macrophages present in the tumor microenvironment as the most abundant immune cell populations that orchestrate various factors [6, 7]. Due to the important role of TAMs in tumor progression, the level of TAMs may be used as a prognostic factor in cancers [8]. However, the contradictory results were shown in prognostic studies. For instance, Li et al. demonstrated that patients with high expression levels of intratumoral CD68+ TAMs got a better survival while Ding et al. found that there was a negative correlation between intratumoral CD68+ TAMs and survival [9, 10]. In many studies, CD68 was used as an indicator for tissue macrophages, but this marker was not sufficiently specific. Immunohistochemistry to detect the expression of CD68, CD86 (M1), or CD163 and CD206 (M2) is frequently used to quantify and classify the TAMs [11]. In reality, the TAMs' phenotypes are diverse. CD169 and CD204 were also used as indicators for TAMs. Single usage of a biomarker to evaluate the density of TAMs couldn’t reflect its actual condition in the tumor microenvironment.

Although some reviews on TAMs for HCC have been conducted, there was no precise calculation of the hazard ratios. Most articles focused on intratumoral CD68+ TAMs while the prognostic values of peritumoral CD68+ TAMs or other subtypes (including CD163+, CD169+, CD204+ and CD206+ TAMs) were ignored. Moreover, the association between TAMs and clinicopathologic features in patients with HCC has not been analyzed systematically. Therefore, we performed a meta-analysis to make a more accurate estimation of the clinicopathologic and prognostic value of TAMs in patients with HCC.

Materials and methods

Literature search

PubMed, EMBASE and Cochrane databases were comprehensively searched for relevant articles published by two independent researchers from inception to Apr 30th, 2019 with the following keywords: (liver cancer or hepatocellular carcinoma), (prognosis or prognostic or survival or outcome) and (tumor-associated macrophages or TAMs or CD68 or CD86 or CD204 or CD206 or CD163 or CD169). Additionally, relevant references were also searched. Unpublished literatures and conference papers were not included. A third researcher made the final decision of the disagreement on candidate articles.

Criteria for inclusion and exclusion

Inclusion criteria were as follows: (1) reported the relationship between prognosis and the density of TAMs in HCC; (2) sufficient data to acquire hazard ratio (HR) and 95% confidence interval (CI); (3) selected the most valuable report, if more than one paper were reported by the same author or group.

Exclusion criteria were as follows: (1) duplicate articles, abstracts, letters, case reports and meetings reports were excluded; (2) didn’t provide enough information on overall survival (OS) or disease free survival (DFS); (3) metastatic or recurrent live cancer.

Data extraction

The required data were extracted by two researchers from all included studies independently. The following information was extracted: the first author's name, publication year, country, biomarkers of macrophage phenotypes, sample source, number of patients, cut-off value and the HRs and CIs for survival.

Statistical analysis

HRs and 95% CIs were used to quantify the association between TAMs and prognosis. If they could not be acquired directly, they were extracted from Kaplan-Meier curves using the method described by Parmar et al [12]. The survival data from Kaplan-Meier curves were extracted by Engauge Digitizer version 10.6 as described previously [13]. The pooled HR with its 95% CI was calculated by STATA version 14.0. The evaluation of heterogeneity among studies was performed by Cochran's Q test and Higgin's I² statistics. The heterogeneity among all included studies was suggested significant when I²>50% and/or P<0.05, then a random-effect model was used (DerSimoniane-Laird method); otherwise, a fixed-effect model (Mantel-Haenszel method) was used. The cumulative meta-analysis was performed according to publication time. The sensitivity analysis was performed by omitting one study at a time in turn to assess the stability and reliability of this review. Egger’s test was performed to identify the potential publication bias [14].

Results

Study selection and characteristics

The initial search in databases gathered a total of 1034 potentially eligible articles. After removing all duplicate articles and checking all titles and abstracts, 24 studies reporting the association between TAMs and the clinicopathologic characteristics or the prognosis of HCC remained. Since four researches among them lacked enough useful data, a total of 20 studies including 4297 patients from China, Germany and Japan were included in the present meta-analysis finally (Fig 1) [9, 10, 15–32].

The main characteristics of the included studies were summarized in Table 1. Among these 20 articles, 13 studied the TAMs in intratumor (IT), one studied the TAMs in the margin of tumor (MT) and 10 studied the TAMs in peritumor (PT), respectively. In addition, two articles studied the macrophages in serum. Totally, 4091 patients were evaluated for the association between TAMs density and survival in patients with HCC. Case numbers of included studies ranged from 73 to 368. The description of tumor-nodes-metastasis (TNM) stage was provided in 14 articles. Followed-up time was offered in 14 articles. All 20 articles were classified as high-quality studies according to the Newcastle-Ottawa quality assessment scale (score ≥ 6 points) (Table 1).

Subgroup analysis

Given that multiple biomarkers of macrophage phenotypes were used to identify the effects on liver cancer patients, a subgroup analysis was carried out on the subsets of TAMs and the distribution location (Table 2).

CD68+ TAM subset

A total of 16 articles, including 3550 cases, studied the association between the prognosis in patients with HCC and the density of CD68+ TAMs [9, 10, 15–19, 21–23, 25, 27–29, 31, 32].

There were 15 studies evaluating the correlation between the density of CD68+ TAMs and OS [9, 10, 15, 16, 18, 19, 21–23, 25, 27–29, 31, 32]. Among them, 11 articles [9, 10, 19, 21–23, 25, 27–29, 32] focused on the association between OS and the density of CD68+ TAMs in IT while eight articles [9, 15, 16, 18, 22, 28, 29, 31] focused on the density of CD68+ TAMs in PT. A random model was used because of a significant heterogeneity (P < 0.001, I² = 69.3%), and the result demonstrated that high density of intratumoral CD68+ TAMs in patients with HCC was associated with a poor OS (pooled HR = 1.417; 95% CI = 1.092–1.839; P = 0.009) and high density of peritumoral CD68+ TAMs was also associated with a poor OS (pooled HR = 1.393; 95% CI = 1.022–1.899; P = 0.036) (Fig 2).

There were nine studies evaluating the correlation between the density of CD68+ TAMs and DFS [9, 15–18, 22, 28, 29, 31]. Among them, eight articles [9, 15, 16, 18, 22, 28, 29, 31] focused on the association between the density of intratumoral CD68+ TAMs and DFS while eight articles [9, 15–18, 22, 28, 31] focused on the density of peritumoral CD68+ TAMs. A random model was used because of a significant heterogeneity (P < 0.001, I² = 71.4%), and the result demonstrated that no significant relation was observed between DFS and the density of CD68+ TAMs in IT (pooled HR = 1.095; 95% CI = 0.871–1.376; P = 0.436) or in PT (pooled HR = 1.223; 95% CI = 0.895–1.671; P = 0.207) (Fig 3).

CD163+ TAM subset

A total of six articles, including 1295 cases, studied the association between the prognosis in patients with HCC and the density of CD163+ TAMs [18, 20, 22, 24–26]. Among them, two articles [22, 25] focused on the association between survival and the density of CD163+ TAMs in IT, three articles [18, 22, 26] focused on the density of CD163+ TAMs in PT, and two articles [20, 24] focused on the density of CD163+ macrophages in serum. As a result, there were no significant relation between OS and the density of CD163+ TAMs in IT (pooled HR = 1.293; 95% CI = 0.537–3.111; P = 0.566) or in PT (pooled HR = 1.150; 95% CI = 0.740–1.787; P = 0.533). However, the result demonstrated that patients with high density of CD163+ macrophages in serum were associated with a poor OS (pooled HR = 5.698; 95% CI = 3.062–10.603; P < 0.001).

CD169+ TAM subset

A total of two articles, including 542 cases, focused on the association between the density of intratumoral CD169+ TAMs and OS [29, 30]. A fixed model was used because of no significant heterogeneity (P = 0.674, I² = 0), and the result demonstrated that high density of intratumoral CD169+ TAMs in patients with HCC was associated with a good OS (pooled HR = 0.471; 95% CI = 0.343–0.647; P = 0.037).

CD204+ TAM subset

A total of two articles, including 412 cases, focused on the association between the density of intratumoral CD204+ TAMs and OS [21, 30]. A fixed model was used because of no significant heterogeneity (P = 0.632, I² = 0), and the result demonstrated that high density of intratumoral CD204+ TAMs in patients with HCC were associated with a poor OS (pooled HR = 1.947; 95% CI = 1.387–2.733; P < 0.001).

CD206+ TAM subset

A total of three articles, including 601 cases, studied the association between the prognosis in patients with HCC and the density of CD206+ TAMs [23, 27, 31]. Among them, three articles [23, 27, 31] focused on the association between the density of intratumoral CD163+ TAMs and OS while two articles [27, 31] focused on the association between the density of intratumoral CD163+ TAMs and DFS. The result demonstrated that high density of intratumoral CD206+ TAMs was associated with a poor OS (pooled HR = 1.723; 95% CI = 1.308–2.270; P < 0.001) and a poor DFS (pooled HR = 1.711; 95% CI = 1.214–2.412; P = 0.002).

Subgroup analyses of the prognostic effect of CD68+ TAMs

Subgroup analysis showed that patients with high density of CD68+ TAMs in IT were associated with poor OS in the groups with large sample size (≥100; pooled HR = 1.485, 95% CI = 1.119–1.970; P = 0.006), median cutoff value (pooled HR = 1.480, 95% CI = 1.027–2.133; P = 0.035), and other cutoff value (pooled HR = 1.436, 95% CI = 1.092–1.889; P = 0.010). Moreover, patients with high density of CD68+ TAMs in PT were associated poor OS in the groups with median cutoff value (pooled HR = 1.431, 95% CI = 1.146–1.787; P = 0.002) (Table 3).

The cumulative meta-analysis revealed that the association between the density of CD68+ TAMs and OS got more and more stable and the confidence interval got narrowed. The association between the density of CD68+ TAMs in IT (Fig 4A) and OS became statistically significant since Zhang’s research in 2016. And the association between the density of CD68+ TAMs in PT (Fig 4B) and OS became statistically significant since Kuang’s research in 2009. It is convinced that high density of CD68+ TAMs was associated with poor prognostic for HCC.

Relationship between CD68+ TAMs and clinicopathologic characteristics

Eleven studies reported the association between CD68+ TAMs and clinicopathologic parameters [9, 10, 15, 18, 22, 23, 25, 27–29, 31]. The information including 15 clinicopathologic parameters and their correlation with CD68+ TAMs in IT is summarized in Table 4. The results of meta-analysis demonstrated that patients with high density of CD68+ TAMs in IT were associated with high AFP value (pooled OR = 1.31, 95% CI = 1.07–1.61; P = 0.01), large tumor size (pooled OR = 1.46, 95% CI = 1.19–1.79; P < 0.01), absent encapsulation (pooled OR = 0.77, 95% CI = 0.61–0.96; P = 0.02), present vascular invasion (pooled OR = 1.33, 95% CI = 1.03–1.71; P = 0.03), and later TNM stage (pooled OR = 1.51, 95% CI = 1.17–1.95; P < 0.01). Besides, the results of meta-analysis showed no correlation between CD68+ TAMs and age, gender, Child-Pugh score, cirrhosis, hepatitis B surface antigen, hepatitis C virus antibody, alanine transaminase, γ-glutamyl transpeptidase, tumor number or differentiation.

Sensitivity analyses and publication bias

Sensitivity analyses showed that none of the included studies were found to significantly affect the pooled HR (Fig 5). Additionally, Egger’s test was performed to assess the publication bias of the included studies in this meta-analysis. As a result, there was no statistical significance of publication bias about CD68+ TAMs in IT (P = 0.994) (Fig 6A) or in PT (P = 0.628) (Fig 6B). Therefore, it was believed to be reliable of this meta-analysis.

Discussion

TAMs are prominent immune cells that orchestrate various factors in the tumor microenvironment and play important roles in the progression of human cancers and angiogenesis [6, 33]. Macrophages are present in the inflammatory environment, especially in its chronic stage, which may trigger cancer initiation. Recently, the presence of TAM-derived inflammatory cytokines IL-12, IL-10, IL-23 and IL-17 has been shown to be closely associated with cancer initiation and progression [34, 35]. However, researches showed that TAMs may exhibit both promoting and inhibiting activities in tumor growth [8, 36]. Due to the crucial role of TAMs in tumor progression, the level of infiltrated TAMs was used as a prognostic factor in cancers. There were a number of meta-analyses available on the impact of TAMs in tumors [37–40]. The prognostic value of TAM for survival in patients with solid tumor remains controversial [37]. Mei J, et al. suggested that although the density of total CD68+ TAMs was not associated with OS, the localization and M1/M2 polarization of TAMs were potential prognostic predictors of non-small cell lung cancer [38]. Troiano G, et al. revealed that CD68+ TAMs had no prognostic utility in patients with squamous cell carcinoma of the head and neck, however CD163+ TAMs predicted poor prognosis [39]. So far, a group of original articles has studied the association between outcome and TAMs in HCC, and the presence of both negative and positive results addressing the significance of TAMs on survival made it necessary to conduct a quantitative aggregation of the conclusion. The presented meta-analysis of 20 articles including 4297 patients revealed that elevated CD68+, CD204+ and CD206+ TAMs infiltration predicted worse survival while elevated CD169+ TAMs infiltration predicted better survival.

CD68 is a heavily glycosylated type I transmembrane glycoprotein that is highly expressed in macrophages and other mononuclear phagocytes. Recently, there have been extensive studies on the relevance between TAMs and prognosis using biomarker CD68, the most extensively used macrophage marker, is expressed on all macrophages [41, 42]. CD68+ TAMs play a role in immunosuppression and inhibit cytotoxic activity of CD8+ T cells in tumors [43]. In the presented meta-analysis, high expression of CD68+ TAMs in the tumor stroma correlates with higher AFP value, larger tumor size, absent encapsulation, present vascular invasion and more advanced TNM stage. It was coincidence with prognostic results. High CD68+ TAMs expression level was associated with worse OS both in IT and PT. However, there were no correlation between CD68+ TAMs and DFS. Most articles believed that there was a positive correlation between CD68+ TAMs and survival while only four articles thought not [10, 17, 22, 31]. According to Li et al., macrophages could directly kill cancer cells as innate immune cell and activate antitumor response via presenting tumor-associated antigens [10]. They believe that the liver is a unique immunologically privileged organ, which the discrepancy between their results and other kinds of cancers could be reasonably explained by. According to Ding et al., although marginal CD68+ TAMs density was not associated with survival or recurrence, it was associated with Child-Pugh score, AFP value, tumor size, tumor number, vascular invasion, TNM stage and fibrous capsule [9]. Therefore, it is reasonable to infer that CD68+ TAMs in MT could help the tumor to spread into surrounding normal tissue and/or blood vessel. Although some conflicting findings existed, the result of the presented meta-analysis manifested a tendency to support the anti-tumor effect of CD68+ TAMs.

TAMs can be divided into two classes of type 1 macrophages (M1) and type 2 macrophages (M2) [5]. M1 macrophages, known as classically activated macrophages, are activated by interferon-γ and microbial products. They can ignite anti-tumor immune responses via expressing high levels of pro-inflammatory cytokines (tumor necrosis factor α, interleukin 1 [IL-1], IL-6, IL-12 or IL-23), major histocompatibility complex (MHC) molecules and inducible nitric oxide synthase. M2 macrophages, known as alternatively activated macrophages, are induced by IL-4, IL-10 and IL-13 in vitro. They can suppress tumor-specific immune responses via downregulating MHC class II and IL-12 expression and upregulating anti-inflammatory cytokine IL-10, scavenger receptor A, and arginase expression [44]. Immunohistochemical staining for M1 (CD86+) or M2 (CD163+, CD204+ and CD206+) is frequently used to quantify and classify TAMs [40, 45]. CD86+ TAMs was only researched in one article which demonstrated that high density of CD86+ TAMs in IT was associated with a better OS [23]. According to the presented meta-analysis, although there was no association between CD163+ TAMs and survival, high density of either CD204+ or CD206+ TAMs in IT was associated with worse OS. It meant the opposite functions of M1 and M2 in HCC prognosis. Therefore, M1/M2 rate may be a better predictor of survival for HCC. However, further study about the relationship between M1/M2 rate and HCC is needed.

CD169+ macrophages, as the founder member of the Siglec superfamily, constitute a minor macrophage population present in lymphoid organs [46, 47]. The combination of antigen and interferon-1, produced by CD169+ macrophages, can stimulate DCs to cross-present cell-associated tumor antigens and to induce T cell activation [47]. Several studies have reported a correlation between CD169+ macrophages and CD8+ T cell infiltration, improving overall survival rates in cancers, such as colorectal cancer, malignant melanoma, endometrial carcinoma, breast cancer and bladder cancer [48–52]. Similarly, according to the presented meta-analysis, high density of CD169+ TAMs was associated with better OS. Considering the significance of the protective and predictive effect of CD169+ TAMs, it may be an important subject whether the selective overexpression of CD169 might represent a novel therapeutic approach to reprogram the anti-tumor activities of macrophages.

Based on the previous research, the expression of TAMs in HCC is correlated with prognosis. Hence, significant attention has been drawn towards development of TAM-targeted therapy; either eliminating them present in the tumor and blocking their pro-tumoral functions, or restoring their immunostimulatory/tumoricidal properties [53]. Liposomes loaded with clodronate can induce apoptosis of macrophages after intracellular release from the liposomes [54]. Tocilizumab, an anti-IL-6 receptor antibody, can be used to block IL-6 signaling and inhibit TAM-stimulated activity of cancer stem cells in vitro and in vivo [55]. Sorafenib, an antiangiogenic oral multikinase inhibitor, can interfer with the polarization of TAMs and their cytokine production and polarize macrophage-induced epithelial-mesenchymal transition and migration of HCC cells [56]. In general, although the clinical application of TAM-targeted therapy is in the initial stage, many preclinical studies in HCC murine models have shown excellent results [7].

As we know, this is the first meta-analysis to evaluate the prognostic role and clinic correlation of TAMs in HCC. Several limitations should be taken into consideration in explaining the conclusion of this research. First, the use of different cut-off values in different studies could reduce the accuracy of TAMs in estimating prognosis. Second, some HRs were obtained by calculating the data extracted from the survival curves when they could not be acquired from the original article directly. Third, some HRs, acquired from univariate analysis, may overestimate the predictive effect compared to multivariate analysis. Fourth, there were few studies on some subsets including CD86+, CD163+, CD169+, CD204+ and CD206+ TAMs. Finally, no prospective trials or randomized controlled trials were reported.

In conclusion, the presented meta-analysis demonstrated that the density of TAMs was associated with survival of patients with HCC. High density of CD68+ TAMs was associated with worse OS, higher AFP value, larger tumor size, absent encapsulation, present vascular invasion, and later TNM stage. High density of M1 TAMs (CD86+ TAMs) predicted better survival in HCC patients, while high density of M2 TAMs (CD204+ and CD206+ TAMs) or CD169+ TAMs predicted worse survival. Nevertheless, further randomized controlled trials and multiple centers clinical trials are needed to elucidate the exact relationship and the underlying mechanism and more non-Asian studies were needed to be enrolled to decrease heterogeneity.

Supporting information

S1 File [doc]
PRISMA checklist.

S2 File [doc]
PRISMA flow diagram.