Prognostic impact of CD68+ tumor-associated macrophages in hepatocellular carcinoma: A meta-analysis

Abstract

Background:

Evidence from clinical research suggests that the tumor-associated macrophages (TAMs) were associated with prognosis in hepatocellular carcinoma (HCC). The aim of the present meta-analysis was to conduct a qualitative analysis to explore the prognostic value of CD68 + TAMs in HCC.

Methods:

This study conducted a systematic search in Pubmed, Embase, the Cochrane Library and China National Knowledge Internet from inception of the databases to November 2023. The hazard ratio (HR) and 95% confidence interval (CI) were calculated employing fixed-effect or random-effect models depending on the heterogeneity of the included trials. The Newcastle-Ottawa Scale was used to evaluate the risk of prejudice.

Results:

We analyzed 4362 HCC patients. The present research indicated that the expression levels Of CD68 + TAMs were significantly associated with overall survival (OS) (HR = 1.55, 95% CI: 1.30–1.84) and disease-free survival (DFS) (HR = 1.44, 95% CI: 1.17–1.78). Subgroup analysis based on cutoff values showed that the “Median” subgroup showed a pooled HR of 1.66 with a 95% CI ranging from 1.32 to 2.08, which was slightly higher than the “Others” subgroup that exhibited a pooled HR of 1.40 and a 95% CI of 1.07 to 1.84. The “PT” subgroup had the highest pooled HR of 1.68 (95% CI: 1.19–2.37), indicating a worse OS compared to the “IT” (pooled HR: 1.50, 95% CI: 1.13–2.01) and “Mix” (pooled HR: 1.52, 95% CI: 1.03–2.26) subgroups. Moreover, in the sample size-based analysis, studies with more than 100 samples (>100) exhibited a higher pooled HR of 1.57 (95% CI: 1.28 to 1.93) compared to studies with fewer than 100 samples (<100), which had a pooled HR of 1.45 (95% CI: 1.00–2.10).

Conclusions:

The analysis suggests that CD68 + TAMs were significantly associated with unfavorable OS and DFS in HCC patients, and may be served as a promising prognostic biomarker in HCC. However, more large-scale trials are needed to study the clinical value of TAMs in HCC.

1. Introduction

Hepatocellular carcinoma (HCC) ranks as the seventh most common and third most lethal cancer globally, presenting significant challenges in treatment and prognosis.^[1] Within its complex tumor microenvironment (TME), the interplay between tumor-associated macrophages (TAMs) and T lymphocytes is pivotal. T lymphocytes face metabolic competition, immunosuppressive influences, and challenging physicochemical conditions that undermine their functionality and viability, crucial for combating tumor growth.^[2] Concurrently, cancer-associated fibroblasts actively participate in HCC advancement through extracellular matrix remodeling, growth factor and cytokine secretion, and interactions with other stromal components, presenting resistance to standard therapies.^[3] TAMs in HCC are characterized by specific markers such as CD68 + and CD141, alongside HLA-DR1, CD163+, CD206+, and elevated arginase activity, highlighting their multifaceted role in HCC progression.^[4]

The prognostic significance of CD68 + tumor-associated macrophages (TAMs) in cancer remains a subject of debate, as studies report varying impacts on patient outcomes. For instance, Ding et al reported that a high density of CD68 + TAMs within tumors correlates with decreased overall survival (OS) and is ties to more aggressive cancer traits such as elevated AFP levels, larger tumors, and vascular invasion.^[5] Furthermore, Minami et al highlighted that the impact of TAMs can vary based on their location within the tumor, emphasizing their role in the peritumoral environment and their association with poor OS.^[6] Contrarily, other studies have identified scenarios where elevated intratumoral CD68 + TAMs are linked to improved patient survival, while some found no significant correlation between CD68 + TAM densities in tumor islets and stroma with OS, sometimes even associating high densities in tumor islets with better OS.^[7,8] Research into TAMs as therapeutic targets, proposing methods to modulate their activity, unveils potential for enhancing treatment efficacy.^[9,10] TAMs’ multifaceted roles in tumor angiogenesis, immunosuppression, and their interactions with cancer stem cells underscore their potential as targets for therapy.^[11,12]

This meta-analysis aims to synthesize these diverse findings, providing a more comprehensive understanding of the prognostic significance of CD68 + TAMs in HCC. By integrating data from multiple studies, this analysis seeks to clarify the role of TAMs in HCC and potentially guide future therapeutic strategies.

2. Methods

2.1. Systematic literature search

This study was conducted using the guidance from Preferred Reporting Items for Systematic reviews and Meta-Analysis protocols (PRISMA flowchart). The study conducted a systematic search in Pubmed, Embase, the Cochrane Library and China National Knowledge Internet from inception of the databases to November, 2023. The reference list of articles was also checked for further reference. The literature search used the following combination of terms: (“hepatocellular carcinoma” or “HCC” or “liver cancer” or “hepatic cancer” or “liver neoplasms”) and (“tumor associated macrophages” or “TAMs” or “macrophage” or “CD68+”) and (“overall survival” or “OS” or “survival” or “Prognostic”) (Table S1, http://links.lww.com/MD/M170). The 2 authors carried out systems electronic search and data extraction.

2.2. Ethical review

As this meta-analysis constitutes a secondary analysis based on previously published data, no ethical approval or informed consent is necessary.

2.3. Inclusion and exclusion criteria

Inclusion criteria: Studies investigating the prognostic values of CD68 + TAMs in HCC; Research with sample size more than 20; sufficient information for computing pooled hazard ratios (HR) and 95% confidence intervals (CI).

Exclusion criteria: Duplicated articles; Study on insufficient data integrity or lack of prognostic results; case reports, correspondences, letters, non-human research, review articles and other studies without original data.

2.4. Data extraction

Two independent researchers will meticulously assess titles, abstracts, and full texts to identify potentially eligible studies. After reviewing the eligible articles, 2 authors extracted the data independently. For each article, the following information was recorded: last name of first author, publication year, country, study period, number of patients, age, sex, location, tumor stage, hepatitis B history, cutoff value, follow-up time, and results (overall survival, disease-free survival [DFS]). Discordances will be resolved through consensus or corresponding author.

2.5. Quality assessment

Quality assessment will be conducted independently by 2 researchers according to standardized criteria, with disagreements resolved through consensus or third-party mediation. The Newcastle-Ottawa Scale will be employed for this purpose. The evaluation items have a 9-point total value. Except for Comparability of Cases and Controls on the Basis of the Design or Analysis, which receives a 2-point rating, all other items received a score of 1, while those that were partially or totally noncompliant received a score of 0. Higher cumulative scores are thought to have a lower chance of bias in the research.

2.6. Statistical analysis

In this meta-analysis, we employed a random effects model to calculate the pooled HR and 95% confidence intervals (CIs), chosen for their ability to accommodate between study variability. Heterogeneity among studies was quantitatively assessed using the I² statistic, Tau², and the Q statistic, with I² values over 50% indicating substantial heterogeneity. Publication bias was evaluated through Begg and Egger tests, which assess the correlation between study sizes and effect sizes. Visual representations included forest plots for individual and pooled HRs, and funnel plots with 95% CI dashed lines for assessing publication bias symmetry. All analyses were conducted using Python software.

3. Results

3.1. Study selection

645 of the 1184 papers found in the initial literature search (Fig. 1) were duplicates. Automation tools marked 466 papers as ineligible after scanning the titles and abstracts. 52 studies were subsequently excluded for reasons such as insufficient data, absence of overall survival data, and studies with fewer than 20 patients. 21 papers that met the inclusion criteria from these publications after further evaluation and selection were included in our quantitative synthesis.

Figure 1.

Flow diagram of study selection.

3.2. Characteristics of the enrolled studies

Table 1 summarizes the main characteristics of the enrolled 21 eligible articles.^[7,13–32] All participants were from Asia (China and Japan). Among the publications, 18 studies reported the number of patients with stage I to II or III to IV tumors, and 15 studies reported the patient hepatitis B history. All studies reported the distribution location of CD68 + TAMs, with 8 studies distributed in intratumor (IT), 6 studies in peritumor (PT), and the remaining studies in IT, PT and margin of tumor. In addition, the number of patients included in 17 studies was all above 100, with only 4 studies having a patient count below 100.

Table 1.

Characteristics of included studies.

Study	Country	Study period	No. of patients	Age	Male, n (%)	Hepatitis B history (yes/no)	Location	Tumor stage n (%)		Follow-up (mo)	Cutoff value
								I–II	III–IV
Zhu, 2008	China	1999–2006	105	NR	96 (91.4)	90/15	IT/PT	86	19	22.1 (12–96)	Median
Li, 2009	China	1997–1999	302	> 60 yr, 70 (23%)	260 (86.1)	256/46	IT	237	65	58 (2–121)	Median
Kuang, 2009	China	1999–2004	262	NR	NR	NR	PT	NR	NR	60	Median
Ding, 2009	China	1999–2004	137	Median, 48 (17–78)	117 (85.4)	14/123	IT/MT/PT	98	39	30 (2–95)	Median
Ju, 2009	China	2002–2005	130	> 52 yr, 63 (48.5)	112 (86.2)	111/19	PT	74	56	31.8 ± 1.7	20%
Jia, 2010	China	2008–2009	105	Median, 51 (18–75)	96 (91.4)	NR	PT	86	19	31.7	Median
Kong, 2013	China	NR	295	Median, 52 (22–80)	247 (83.7)	NR	PT	202	93	60	75%
Lin, 2013	China	2004–2011	132	Median, 51 (20–71)	122 (92.4)	11/121	IT	82	50	NR	Minimum P value
Ohno, 2014	Japan	1999–2002	225	Mean, 65.5 yr	168 (74.7)	109/116	IT	103	122	43 (1–166)	Average
Yeung, 2015	China	2004–2008	95	18–83	74 (78)	72/22	IT/PT	NR	NR	48 < 60	ROC curve
Shu, 2016	China	2005–2009	80	Median, 57 (30–79)	57 (71.25)	60/20	IT	48	32	31 (1–54)	Median
Dong, 2016	China	2009–2011	253	> 51 yr, 122 (48%)	220 (87)	246/12	IT	176	77	16	Median
Hu, 2016	China	2008–2013	368	> 50 yr, 225 (61%)	303 (83.3)	315/53	IT	228	140	84	Median
Zhou, 2016	China	2006	390	NR	NR	NR	IT	NR	NR	84	Median
Zhang Y, 2016	China	2002–2012	354	Median, 48 (20–78)	316 (89.3)	322/53	IT/PT	253	91	120	Median
Zhang Q, 2016	China	NR	149	Mean, 67.5 yr	82 (55.0)	NR	IT/PT	115	34	NR	Scores ≥ 8
Kono, 2016	Japan	2000–2008	77	Median, 67 (44–85)	61 (79.22)	NR	PT	73	4	96	ROC curve
Ren, 2017	China	2011–2016	268	< 54 yr, 133 (49%)	229 (85.4)	208/58	IT/PT	203	65	44 (1–54)	ROC curve
Xie, 2018	China	2009–2014	316	< 60 yr, 222 (70%)	258 (81.6)	213/103	IT	167	122	NR	Minimum P value
Chen, 2019	China	2008–2017	94	> 47 yr, 47 (50%)	82 (87.2)	85/9	PT	64	30	NR	Median
Yusa, 2022	Japan	2004–2013	225	Mean, 67.6 ± 8.4	171 (76.0)	141/84	IT/PT	134	91	53	Median

IT = intratumor, MT = margin of tumor, NOS = Newcastle-Ottawa scale, NR = not report, PT = peritumor, ROC = receiver operating characteristic.

3.3. Quality assessment

The Newcastle-Ottawa Scale was used to evaluate the bias risk of every included article, and every article received an average score of over 6 points. All 21 papers completely complied with the standards for research of fair quality, according to the findings of the risk of bias assessment (Table S2, http://links.lww.com/MD/M171).

3.4. Meta-analysis results

Our meta-analysis included 4362 patients to analyze the impact of CD68 + TAMs on survival rate. A random effects model was conducted to calculate the pooled effect size because significant heterogeneity existed among the enrolled studies (I² = 79.59%). Our results revealed that the increased CD68 + TAMs was significantly related to the unfavorable OS (HR = 1.55, 95% CI: 1.30–1.84, P < .01) (Fig. 2). In addition, our research also found that the increase of CD68 + TAMs was associated with a poor DFS (pooled HR = 1.44, 95% CI = 1.17–1.78, P < .01, I² = 70.76%) (Fig. 3).

Figure 2.

Forest plot of enrolled studies for the association between CD68 + TAMs with overall survival. TAMs = tumor-associated macrophages.

Figure 3.

Forest plot of enrolled studies for the association between CD68 + TAMs with disease-free survival. TAMs = tumor-associated macrophages.

3.5. Subgroup analyses

The results of subgroup analysis are shown in Table 2. In the meta-analysis, subgroup analysis based on cutoff values revealed distinct trends (Figure S1, http://links.lww.com/MD/M172). The “Median” subgroup showed a pooled HR of 1.66 with a 95% CI ranging from 1.32 to 2.08, which was slightly higher than the “Others” subgroup which exhibited a pooled HR of 1.40 and a 95% CI of 1.07 to 1.84. The location-based analysis differentiated 3 subgroups: “IT,” “Mix,” and “PT.” The “PT” subgroup had the highest pooled HR of 1.68 (95% CI: 1.19–2.37), indicating a greater effect size compared to the “IT” (pooled HR: 1.50, 95% CI: 1.13–2.01) and “Mix” (pooled HR: 1.52, 95% CI: 1.03–2.26) subgroups (Figure S2, http://links.lww.com/MD/M173). Lastly, in the sample size-based analysis, studies with more than 100 samples (> 100) exhibited a higher pooled HR of 1.57 (95% CI: 1.28–1.93) compared to studies with fewer than 100 samples (< 100), which had a pooled HR of 1.45 (95% CI: 1.00–2.10) (Figure S3, http://links.lww.com/MD/M174).

Table 2.

Subgroup analyses results.

Subgroup	Number of study	Sample size	HR (95%CI)	Cochran Q	df	I 2	P value
Location
PT	6	963	1.68 (1.19–2.37)	21.88	5	77.15%	.0006
IT	8	2066	1.50 (1.13–2.01)	30.62	7	77.14%	.0001
Mix	7	1333	1.52 (1.02–2.26)	24.71	6	75.72%	.0004
Sample size
> 100	17	4016	1.57 (1.28–1.93)	61.01	16	73.77%	.0000
< 100	4	346	1.45 (1.00–2.10)	7.48	3	59.89%	.0581
cutoff value
Median	12	2675	1.66 (1.32–2.08)	38.59	11	71.50%	.0001
Others	9	1687	1.40 (1.07–1.84)	31.33	8	74.76%	.0001

CI = confidence interval, HR = hazard ratio.

3.6. Publication bias

Publication bias was assessed visually using funnel plots. The I² statistic was found to be 79.60%, indicating a high level of heterogeneity among the included studies. The Q statistic P value was exceedingly low, confirming significant heterogeneity. Therefore, we conducted Begg and Egger. Begg test, which examines the correlation between the effect sizes and their variances, yielded a Kendall Tau of 0.40 with a P value of .0111. Similarly, Egger test, which performs a linear regression of the effect size against its standard error, indicated a significant slope of 2.26 (P value = .0111). These findings from both Begg and Egger tests are consistent and point toward a significant publication bias within the included studies. These findings emphasize the need for a cautious and nuanced interpretation of the pooled HR and its implications (Fig. 4).

Figure 4.

Funnel plot for assessment of publication bias.

4. Discussion

The findings of our meta-analysis highlight the complex role of CD68 + TAMs in HCC. The variability in CD68 + TAMs’ prognostic impact underscores the intricate dynamics within the tumor microenvironment.

Previous studies reveal a comprehensive analysis of the prognostic significance of TAMs across various cancer types, particularly focusing on head and neck squamous cell carcinoma, esophageal cancer, and ovarian cancer. Bisheshar et al 2022 and Troiano et al 2019 both focus on head and neck squamous cell carcinoma. Bisheshar et al emphasize the importance of TAMs, which are associated with worse OS, DFS, and progression-free survival (PFS).^[33] Troiano et al found similar results, noting that high stromal expression of CD163 + TAMs correlates with poor OS and PFS. Moreover, the role of TAMs was examined in esophageal cancer, finding that CD68 + TAM density is not associated with esophageal cancer progression.^[34] Another study underscores the varying roles of different TAM subtypes in cancer prognosis.^[35] Besides, Yuan et al 2017 conducted a meta-analysis on ovarian cancer, highlighting that a higher M1/M2 TAM ratio in tumor tissues is associated with favorable OS and PFS.^[36] They also find that high density of CD163 + TAMs and a higher CD163+/CD68 + TAMs ratio is correlated with poor PFS and advanced tumor-nodes-metastasis stages, suggesting the pivotal role of TAMs in the progression and prognosis of ovarian cancer. Overall, these studies collectively highlight the complex and significant role of TAMs in cancer prognosis. While CD163 + M2-like TAMs often correlate with poorer outcomes across different cancer types, the prognostic significance of CD68 + TAMs appears to be more variable. These findings underscore the importance of TAMs in cancer biology and their potential as targets for therapeutic intervention. However, the decision to recommend adjuvant therapy should not be based solely on CD68 status. It should also take into account the overall clinical context, including tumor type, stage, patient health status, and the specific molecular and histopathological characteristics of the tumor. In contrast, CD68-negative patients might have a different TME profile, potentially less infiltrated by these immunosuppressive macrophages, which could suggest a different prognosis or response to therapy. However, again, the therapeutic approach should be personalized and based on a comprehensive evaluation of all clinical and pathological data.^[37,38]

Studies consistently indicate that TAMs, play a significant role in the progression and prognosis of HCC. High density of TAMs in intratumor regions is associated with poor OS and DFS in HCC patients.^[39–42] The polarization of TAMs into pro-inflammatory M1 and immunosuppressive M2 macrophages is crucial. High CD11c-positive M1 macrophage density is associated with better OS, whereas high CD206-positive M2 macrophage density correlates with worse OS.^[43–47] Moreover, Studies reveal that increased expression level of peritumoral infiltrated CD68 + macrophages had a poor prognostic value on OS, DFS. High density of CD68 + TAMs in either IT was associated with a poor OS, high AFP value, large tumor size, absent encapsulation, present vascular invasion, and later tumor-nodes-metastasis stage.^[4,48–50] This was consistent with our research findings. The infiltration and polarization status of TAMs, particularly the CD11c-positive and CD206-positive macrophages, emerge as independent prognostic factors for HCC, useful for risk stratification in clinical settings. Therefore, understanding the role of TAMs in HCC can inform treatment strategies. For example, targeting M2 macrophages might be a viable therapeutic approach to hinder tumor progression. Integrating the study of TAMs with established HCC staging systems could refine prognostic assessments and guide treatment decisions. Besides, HCC underlying etiology, whether viral hepatitis or conditions like nonalcoholic steatohepatitis (NASH) or alcoholic liver disease (ALD), also influences its clinical course and outcomes. Hepatitis B or hepatitis C virus (HBV, HCV)’s impact on HCC outcomes can indeed vary when compared to HCC arising from other etiologies like ALD or NASH. NASH-related HCC patients presented with more preserved liver function and higher rates of curative-intent therapy compared to those with other etiologies. Despite similar tumor stages at diagnosis, NASH was associated with worse overall survival compared to ALD but showed similar survival rates compared to HBV and HCV-related HCC.^[51,52]

In our study, the subgroup analysis provided nuanced insights into the variability of effect sizes across different parameters. The marginally higher pooled HR observed in the “Median” subgroup compared to the “Others” subgroup could be indicative of methodological differences or inherent characteristics within these groups impacting the outcomes. The notable difference in the pooled HRs among the location-based subgroups, particularly the higher HR in the “PT” subgroup, might reflect related variations in the study conditions. The larger effect size in studies with sample sizes >100 suggests a potential influence of sample size on the reliability and variability of results. In our study, the total sample size of subgroups with patients >100 was much larger than that of subgroups with patients <100 (4016 vs 346). An excessive number of participants can introduce numerous interfering factors, leading to larger errors. These findings underscore the importance of considering subgroup characteristics in interpreting meta-analytic results. They highlight how factors such as cutoff criteria, location, and sample size not only influence the pooled HR and its confidence interval but also potentially reflect underlying differences in study populations, methodologies, or context-specific variables.

This meta-analysis encountered several limitations, notably significant heterogeneity across most subgroups, as evidenced by high I² values, suggesting considerable variability in study outcomes. Particularly, smaller studies exhibited lower heterogeneity, hinting at a potential small study effect, whereas larger studies might disproportionately influence the pooled estimates. Secondly, the presence of publication bias, where studies with significant findings are more likely to be published, could have led to an overestimation of effects. The varying quality of the included studies further complicates the reliability of the findings, as methodological differences and biases within individual studies could skew the overall results. Additionally, the generalizability of the findings might be limited, given that the included studies may not represent broader populations or diverse settings, thus necessitating cautious interpretation and application of the results. Lastly, our study did not report cases with worse survival outcomes and inability to undergo surgery, which may have a certain impact on actual clinical outcomes. Well-designed studies and multi-ethnic clinical research with larger sample sizes should be carried out in the future.

5. Conclusions

In conclusion, the present analysis implicated that CD68 + TAMs are strongly associated with OS and DFS with HCC patients, and may be served as a promising prognostic biomarker in HCC. In the further, more large-scale trials and more non-Asian populations are needed to study the clinical value of TAMs in HCC.

Author contributions

Conceptualization: Danwen Jin, Jinliang Dong.

Data curation: Liyong Qian, Jiayao Chen.

Formal analysis: Danwen Jin.

Funding acquisition: Jinliang Dong.

Investigation: Danwen Jin.

Methodology: Jinliang Dong.

Project administration: Jiayao Chen, Ze Yu.

Resources: Jinliang Dong.

Software: Ze Yu.

Supervision: Jinliang Dong.

Validation: Liyong Qian.

Visualization: Ze Yu.

Writing – original draft: Danwen Jin.

Writing – review & editing: Liyong Qian, Jiayao Chen, Ze Yu, Jinliang Dong.

Supplementary Material