Prognostic and predictive significance of the tumor microenvironment in hepatocellular carcinoma

Abstract

BACKGROUND:

Identification of molecular markers that reflect the characteristics of the tumor microenvironment (TME) may be beneficial to predict the prognosis of post-operative hepatocellular carcinoma (HCC) patients.

OBJECTIVE AND METHODS:

A total of 100 tissue samples from HCC patients were separately stained by immunohistochemistry to examine the expression levels of CD56, CD8 $\alpha$ , CD68, FoxP3, CD31 and pan-Keratin. The prognostic values were analyzed by Cox regression and the Kaplan-Meier method.

RESULTS:

Univariate and multivariate logistic analysis showed that FoxP3 was the independent factor associated with microvascular invasion (MVI), tumor size and envelop invasion; CD68 was associated with envelope invasion and AFP. Kaplan-Meier survival curves revealed that CD68 and FoxP3 expression were significantly associated with relapse free survival (RFS) of HCC patients ( $P<$ 0.05). The ROC curve indicated that the combination of tumor number, MVI present and CD68 expression yielded a ROC curve area of 82.3% (86.36% specificity, 68.75% sensitivity) to evaluate the prognosis of HCC patients, which was higher than the classifier established by the combination of tumor number and MVI (78.8% probability, 63.64% specificity and 85.42% sensitivity).

CONCLUSIONS:

Our study indicated that CD68 and FoxP3 are associated with prognosis of HCC patients, and CD68 can be considered as a potential prognostic and predictive biomarker.

Keywords

HCC microvascular invasion (MVI)relapse survival (RFS)CD68 FoxP3 CD31

1. Introduction

Hepatocellular carcinoma (HCC), the majority of primary liver cancers (accounting for 75%–85%), is the sixth most common tumor and the fourth leading cause of cancer-related deaths worldwide, with about 841,000 new cases and 782,000 deaths annually [1]. The main risk factors associated with HCC are chronic infections with hepatitis B virus or hepatitis C virus, nonalcoholic fatty liver disease, and alcoholic liver disease [2]. Surgical resection is still the major radical treatment for patients with HCC, however, most patients develop tumor recurrence even metastasis after surgery, which results in poor prognosis and affects long-term survival [3]. It has been reported that the main mechanisms of recurrence include residual micro-lesions, hepatitis, immunosuppression, and microvascular invasion [4]. Therefore, identification of potential molecular markers involved in tumor recurrence could be beneficial to HCC patients.

The tumor microenvironment (TME) plays a vital role in tumor initiation, progression and metastasis, which consists of cancer cells, immune cells, vascular endothelial cells, fibroblasts, secreted cytokines/chemokines and other components [5]. The tumor microenvironment could be divided into immune microenvironment and non-immune microenvironment. TME could be involved in anti-tumor and pro-tumor effect. Natural killer (NK) cells and CD8 ${}^{+}$ cytotoxic T lymphocytes (CTLs) contribute to cytotoxic effect among the cells involved in innate and adaptive immune system, respectively [6]. Nevertheless, the TME of HCC is characterized by immune tolerance [7], due to a variety of immune suppressor cells, such as tumor-associated macrophages (TAMs), regulatory T cell (Tregs) and myeloid-derived suppressor cells (MDSCs), which could lead to tumor immune evasion by interfering with immune surveillance and result in poor prognosis of HCC patients eventually [8]. Moreover, it has been reported that the microvascular invasion (MVI) is an important risk of early postoperative recurrence and closely related to TME in invasive breast cancer as well as non-small-cell lung cancer [9, 10]. However, the correlation between MVI and immune microenvironment in HCC has not been well understood.

A lot of markers have been found to evaluate different types of immune cells. Recent studies showed that the level of active CD56 ${}^{+}$ NK cells were notably decreased in HCC patients, showing the implication in the tumor progression [11]. In contrast, the high levels of CD8 ${}^{+}$ tumor infiltrating lymphocytes (TILs) in HCC had a better prognostic value on overall survival (OS) and disease/recurrence-free survival (DFS/RFS) [12]. CD68, as a pan-macrophage marker, has been increasingly used to identify TAMs in HCC. Ding et al. showed that the higher density of CD68 ${}^{+}$ TAMs in either intratumor or peritumor was correlated with a poorer OS [13]. Similarly, Tu et al. found that higher percentage of intratumoral FoxP3 ${}^{+}$ Treg cells were positively associated with higher grade, larger tumor size, and poorer prognosis of HCC patients [14], whereas Shi et al. showed that high expression of FoxP3 in tumor tissues was related to low level of serum AFP level, absence of vascular invasion and early TNM stage, which suggested a tumor-inhibiting effect of FoxP3 in HCC [15]. Although a lot of studies on the markers of immune cells have been done, the prognostic value of them in HCC is still controversial. In addition, the marker of endothelium CD31 could take part in vascular development [16], which is a significant prognostic indicator. It has been reported that CD31 expression notably increased in HCC compared to the adjacent liver parenchyma [17]. Zhang et al. found that the expression of CD31 could promote HCC invasion and metastasis [18]. Cytokeratin, especially cytokeratin-19 (CK-19), expressing in a subset of HCC with poor prognosis [19], plays an important role in distinguishing between dysplastic and malignant liver nodules [20]. These two markers could be used to predict the prognosis of HCC patients, but the relation between them and immune cells is still not clear.

In this study, we examined the expression of CD56, CD8 $\alpha$ , CD68, FoxP3, CD31, and pan-Keratin by immunohistochemistry in the tumor tissues of 100 HCC patients. And we also analyzed the association between the main clinicopathological factors and the six molecular markers and the prognostic and predictive values of them.

2. Methods and materials

2.1 Patients with HCC

From January 2017 to July 2019, a total of 100 HCC patients who underwent tumor resection at Cancer Hospital of Shandong Province were included in this study. HCC was diagnosed according to the NCCN (National Comprehensive Cancer Network) guidelines. Main clinicopathologic information of the patient was summarized in Supplementary Table 1, including MVI, tumor number, satellite nodule, tumor size, envelope invasion, and AFP. AFP levels were determined by immunoenzymatic chemiluminescence (Roche Diagnostics, Germany). Paraffin embedded tissue samples for immunohistochemistry were used in this study. All the clinical tissue specimens were identified by using H&E staining in Department of Pathology. All of the patients were followed-up until June 2020. One patient died of postoperative complications. The clinicopathologic information and tumor tissue of this patient were analyzed, but the patient was excluded in survival analysis.

This study was approved by the medical ethics committee of Cancer Hospital of Shandong Province, and written informed consent was obtained from the participants.

2.2 Immunohistochemistry and scoring

Immunohistochemistry was performed on paraffin embedded tissue samples. Tissues were cut into 5 $\mu$ m sections. After deparaffinization and hydration, antigen retrieval was performed and endogenous peroxidase was blocked by 3% hydrogen peroxide. Primary antibodies, including CD68 (D4B9C, 1:800, Cell Signaling Technology), CD8 $\alpha$ (C8/144B, 1:200, Cell Signaling Technology), FoxP3 (D2W8E, 1:100, Cell Signaling Technology), CD56 (123C3, 1:800, Cell Signaling Technology), CD31 (89C2, 1:1600, Cell Signaling Technology), pan-Keratin (C11, 1:500, Cell Signaling Technology), were incubated at 4 ${}^{\circ}$ C overnight. Sections were washed with phosphate-buffered saline then treated with a PV-9000 Polymer Detection System for Immuno-Histological Staining (GBI Labs cat# PV-9000), and finally stained with DAB (Zhongshan cat# ZLI-9018).

For CD68, FoxP3 and pan-Keratin, the staining intensity and proportion of positive cells were determined. Positive cells had light yellow or brown granules. Staining intensity was determined according to the staining characteristics of most cells (contrast to the background) and was scored as follows: no staining, 0 (None); light yellow, 1 (Weak); brown yellow, 2 (Moderate); brown, 3 (Strong). Five fields were randomly selected at a high magnification (200 $\times$ ), and 100 cells were counted in each field in which the proportion of positive cells was determined and scored as follows: 0 $=$ 0% to 10%; 1 $=$ 11%–25%; 2 $=$ 26%–50%; 3 $=$ 51–75%; 4 $=>$ 75%. The staining intensity and proportion of positive cells were determined in each field, and the total score was the sum of both scores: 0, negative (-), 1–2, weakly positive (+), 3–4, positive (++), 5–7, strong positive (+++). For CD8 $\alpha$ and FoxP3, the staining of positive cells in the section were determined as “+” and no positive cells were determined as “-”. For CD31, the number of the micro-vessel was counted in three random field of the section. All the sections were scored independently by two experienced pathologists.

2.3 Follow-up

Post-surgery follow-up was conducted through hospital-based follow-up visit and/or telephone. All the patients were followed-up until death or June 20, 2020. The postoperative surveillance program complied with a periodical routine follow-up at 3-month intervals for the first 2 years and at 6-month intervals thereafter. Each follow-up visit will include Alpha-fetoprotein (AFP), liver function, enhanced computed tomography (CT), and/or digital subtraction angiography (DSA). If the recurrence was suspicious by the above-mentioned examinations, enhanced magnetic resonance imaging (MRI) will be performed to confirm diagnosis. The recurrence of HCC was diagnosed based on typical imaging findings and/or continually increased serum AFP. Biopsies were undertaken to achieve histopathology or cytopathology evidence but were not necessarily for the assessment of recurrences.

The treatment strategy for recurrence mainly based on the comprehensive consideration of tumor characteristics, liver function and general condition. Local curative treatment consisted of transcatheter arterial chemoembolization (TACE) and radiofrequency ablation (RFA); systemic palliative treatment, such as molecular targeted therapy and chemotherapy was performed as alternative methods for recurrence treatment.

Relapse free survival (RFS) was defined from the date of surgery to the date of detected recurrence or last follow-up.

2.4 Survival analysis

Univariate Cox proportional hazards regression analysis were done to evaluate the association of each molecular markers or clinical parameters to overall patient survival. The $p$ values were calculated using the Wald test. Multivariate Cox proportional hazards regression analysis were done to evaluate the independent prognostic value of the six molecular markers or clinical parameters. The Kaplan-Meier estimator was used to evaluate the median survival time of the RFS that was based on the six molecular markers or clinical parameters. The $p$ value of the Kaplan-Meier analysis was calculated with the log-rank test.

Table 1
Comparisons of baseline immunology characteristics between the groups with different pathological factors

Characteristic	FoxP3	$P$	CD8 $\alpha$	$P$	CD31	$P$	CD68	$P$	Pan-Keratin	$P$	CD56	$P$
MVI
$+$	7/47	0.000	3 (2-5)	0.015	40.41 $\pm$ 10.21	0.005	4.5 (4-5.25)	0.912	6 (4-7)	0.918	13/41	0.105
$-$	21/25		4 (3-6)		35.24 $\pm$ 7.95		4 (4-6)		6 (4-7)		18/28
Tumor number
$+$	5/21	0.247	3 (2-5)	0.643	38.31 $\pm$ 10.36	0.864	4 (4-5)	0.462	6 (4.75-7)	0.278	7/19	0.601
$-$	23/51		3 (2-5)		37.93 $\pm$ 9.32		4.5 (4-6)		6 (4-7)		24/50
Satellite nodule
$+$	2/12	0.218	3 (2-5.25)	0.596	41.36 $\pm$ 11.27	0.161	4 (4-6)	0.755	6 (3.75-6.25)	0.526	4/10	0.551
$-$	26/60		4 (2-5)		37.49 $\pm$ 9.20		4 (4-6)		6 (4-7)		27/56
Tumor size ( $>$ 5 cm)
$+$	6/29	0.076	4 (2-5)	0.894	39.17 $\pm$ 10.02	0.383	5 (4-6)	0.422	4 (3-7)	0.059	16/19	0.020
$-$	22/43		3 (2-5)		37.42 $\pm$ 9.31		4 (4-5.5)		6 (4-7)		15/50
Envelope invasion
$+$	5/37	0.002	3 (2-5)	0.324	38.62 $\pm$ 10.85	0.616	5 (4-6)	0.002	6 (4-7)	0.826	11/31	0.376
$-$	23/35		4 (2-5.25)		37.60 $\pm$ 8.55		4 (3-5)		6 (4-7)		20/38
AFP
$+$	11/44	0.049	3 (2-6)	0.534	39.65 $\pm$ 10.09	0.060	5 (4-6)	0.028	6 (4-7)	0.321	20/35	0.200
$-$	17/28		3 (2-5)		36.04 $\pm$ 8.54		4 (3.5-5)		6 (4-7)		11/34

Variables are expressed as the mean $\pm$ SD or median with quartiles range or number, unless otherwise indicated.

2.5 Statistical analysis

SPSS26.0 software was used for the statistical analysis. Continuous data were expressed as mean $\pm$ standard deviation and compared using $t$ test (Student’s $t$ test). Ranked ordinal data were described as median with quartiles range and compared using Mann-Whitney U test. Categorical data was presented as number and compared using the Chi-square test or Fisher’s exact probability test. Logistic regression analysis was performed to analyze various combinations of clinical parameters and immunological marker. The $p$ value was bilaterally tested, and values less than 0.05 were regarded as statistically significant. The receiver operating characteristic (ROC) curve and the area under the curve (AUC) were used to determine the feasibility. As defined, the corresponding sensitivity and specificity was showed.

3. Results

3.1 The association between clinicopathological factors and the six molecular markers in 100 HCC patients

According to the standard of staging system of liver cancer reported by NCCN and Guidelines for diagnosis and treatment of primary liver cancer in China, we just analyzed the main clinical pathological factors, including MVI, tumor number, satellite nodule, tumor size ( $>$ or $\leqslant$ 5 cm), envelope invasion, and the level of AFP in the following analysis. We examined the expression of CD56, CD8 $\alpha$ , CD68, FoxP3, CD31, and pan-Keratin by immunohistochemistry in tumor tissue specimens of 100 HCC patients. Representative examples and the staining intensity were showed in Fig. 1.

Figure 1.

Representative immunohistochemical staining images of CD8 $\alpha$ , CD56, CD68, FoxP3, CD31 and pan-Keratin protein expression. Distribution diagrams of staining intensity in 100 cases were showed, respectively.

As shown in Table 1, the association between clinicopathological factors and the molecular markers were analyzed. We found that the expression of FoxP3, CD8 $\alpha$ , and CD31 was significantly correlated with MVI ( $P=$ 0.000, $P=$ 0.015, and $P=$ 0.005, respectively); Additionally, the expression of FoxP3 and CD68 were significantly correlated with envelope invasion ( $P=$ 0.002 and $P=$ 0.002, respectively) and the level of AFP ( $P=$ 0.049 and $P=$ 0.028, respectively). Furthermore, the expression of CD56 was associated with tumor size ( $>$ or $\leqslant$ 5 cm) ( $P=$ 0.02). No significant association was found between pan-Keratin expression and the clinicopathological factors that mentioned above.

Table 2

Univariate and multivariate logistic analysis of the six markers and pathology characters in 100 HCC patients

Variable	Univariate analysis		Multivariate analysis
	HR (95% CI)	$p$ value	HR (95% CI)	$p$ value
MVI ( $+$ / $-$ )
FoxP3	0.177 (0.066–0.474)	0.001	0.174 (0.061–0.496)	0.001
CD8	0.749 (0.584–0.961)	0.023	0.720 (0.545–0.952)	0.021
CD31	1.063 (1.016–1.114)	0.009	1.065 (1.012–1.121)	0.015
CD68	1.007 (0.757–1.339)	0.963
pan-Keratin	0.976 (0.769–1.239)	0.840
CD56	0.493 (0.209–1.166)	0.107
Multi-nodule ( $+$ / $-$ )
FoxP3	0.528 (0.177–1.574)	0.252
CD8	1.063 (0.813–1.391)	0.655
CD31	1.044 (0.958–1.052)	0.862
CD68	0.924 (0.669–1.275)	0.630
pan-Keratin	1.184 (0.890–1.575)	0.246
CD56	0.768 (0.284–2.074)	0.602
Satellite nodule ( $+$ / $-$ )
FoxP3	0.385 (0.080–1.841)	0.232
CD8	0.921 (0.647–1.309)	0.645
CD31	1.043 (0.983–1.108)	0.163
CD68	1.099 (0.724–1.668)	0.658
pan-Keratin	0.913 (0.653–1.277)	0.594
CD56	0.874 (0.251–3.038)	0.832
Tumor size ( $>$ 5 cm/ $\leqslant$ 5 cm)
FoxP3	0.404 (0.146–1.119)	0.081	0.277 (0.089–0.858)	0.026
CD8	0.990 (0.772–1.271)	0.939
CD31	1.020 (0.976–1.065)	0.380
CD68	1.106 (0.818–1.495)	0.514
pan-Keratin	0.771 (0.599–0.992)	0.043
CD56	2.807 (1.164–6.771)	0.022	3.503 (1.327–9.249)	0.011
Envelope invasion ( $+$ / $-$ )
FoxP3	0.206 (0.070–0.601)	0.004	0.211 (0.068–0.657)	0.007
CD8 $\alpha$	0.872 (0.682–1.115)	0.275
CD31	1.011 (0.970–1.055)	0.598
CD68	1.629 (1.168–2.272)	0.004	1.804 (1.243–2.619)	0.002
pan-Keratin	1.028 (0.808–1.308)	0.823
CD56	0.674 (0.281–1.618)	0.377
AFP ( $<$ 20 ng/ml/ $\geqslant$ 20 ng/ml)
FoxP3	0.412 (0.168–1.007)	0.052
CD8	1.065 (0.837–1.353)	0.609
CD31	1.042 (0.998–1.089)	0.063
CD68	1.424 (1.048–1.935)	0.024	1.424 (1.048–1.935)	0.024
pan-Keratin	0.901 (0.708–1.147)	0.399
CD56	1.766 (0.737–4.233)	0.202

HR, hazard ratio; CI, confidence interval.

Univariate and multivariate logistic analysis were performed to further understand the relationship between the molecular markers and clinicopathological factors as shown in Table 2. Univariate analysis showed FoxP3 (HR $=$ 0.177, 95% CI 0.066–0.474, $P=$ 0.001), CD8 $\alpha$ (HR $=$ 0.749, 95% CI 0.584–0.961, $P=$ 0.023) and CD31 (HR $=$ 1.063, 95% CI 1.016–1.114, $P=$ 0.009) were associated with the MVI presence. Pan-Keratin (HR $=$ 0.771, 95% CI 0.599–0.992, $P=$ 0.043) and CD56 (HR $=$ 2.807, 95% CI 1.164–6.771, $P=$ 0.022) were associated with tumor size. FoxP3 (HR $=$ 0.206, 95% CI 0.070–0.601, $P=$ 0.004) and CD68 (HR $=$ 1.629, 95% CI 1.168–2.272, $P=$ 0.004) were associated with the envelope invasion. Furthermore, CD68 (HR $=$ 1.424, 95% CI 1.048–1.935, $P=$ 0.024) was associated with the level of AFP. All the six molecular markers were not associated with multi-nodule or satellite nodule. Multivariate analysis showed that FoxP3 (HR $=$ 0.174, 95% CI 0.061–0.496, $P=$ 0.001), CD8 $\alpha$ (HR $=$ 0.720, 95% CI 0.545–0.952, $P=$ 0.021) and CD31 (HR $=$ 1.065, 95% CI 1.012–1.121, $P=$ 0.015) were the independent risk factor associated with MVI presence; so were FoxP3 (HR $=$ 0.277, 95% CI 0.089–0.858, $P=$ 0.026) and CD56 (HR $=$ 3.503, 95% CI 1.327–9.249, $P=$ 0.011) associated with tumor size ( $>$ 5 cm/ $\leqslant$ 5 cm), FoxP3 (HR $=$ 0.211, 95% CI 0.068–0.657, $P=$ 0.007) and CD68 (HR $=$ 1.804, 95% CI 1.243–2.619, $P=$ 0.002) associated with envelope invasion. Furthermore, CD68 (HR $=$ 1.424, 95% CI 1.048–1.935, $P=$ 0.024) was also an independent risk factor associated with AFP.

3.2 The prognostic values of the six molecular markers and clinicopathological factors in HCC patients

In this cohort, recurrence was observed in 55/99 patients (55.6%) in total. The cumulative 1-, 2-, 3-, 4-year RFS rates of the 99 patients were 50.7%, 25.3%, 25.3% and 8.4%, respectively. Among the 55 recurrent patients, 47 had recurrence in the first year. To explore the prognostic values of the six molecular markers and main clinicopathological factors, we used Kaplan-Meier survival curves for relapse free survival (RFS) analysis in this cohort As shown in Fig. 2, our result revealed that CD68 and FoxP3 showed significant relationships with RFS in HCC patients ( $P<$ 0.05). The patients in CD68 high expression group had shorter RFS than those in CD68 low expression group ( $P=$ 0.006), and patients in FoxP3 positive expression group had longer RFS than those in FoxP3 negative expression group ( $P=$ 0.014). However, CD8 $\alpha$ , CD56, CD31 and pan-Keratin expression levels did not affect HCC patients’ RFS ( $P>$ 0.05).

Figure 2.

Kaplan-Meier curves for RFS in patients with HCC according to the level of CD8 $\alpha$ (A), CD56 (B), CD68 (C), FoxP3 (D), CD31 (E), pan-Keratin (F) and the status of tumor number (G), MVI (H) and envelope invasion (I).

Table 3

Univariate and multivariate survival analysis of RFS in HCC patients

Variable	COX
	Univariate analysis		Multivariate analysis
	HR (95% CI)	$p$ value	HR (95% CI)	$p$ value
Pathological factors
Tumor size	1.073 (0.988–1.165)	0.094
Tumor number	2.745 (1.551–4.859)	0.001	3.020 (1.609–5.669)	0.001
Satellite nodules	2.449 (1.243–4.823)	0.010
MVI classification	3.244 (1.755–5.997)	0.000	2.506 (1.342–4.679)	0.004
Envelope invasion	1.791 (1.048–3.062)	0.033	2.007 (1.116–3.608)	0.020
HBV	2.401 (0.952–6.053)	0.063
Cirrhosis	1.285 (0.547–3.021)	0.565
AFP level	1.157 (1.064–1.258)	0.001
Postoperative TACE	1.466 (0.771–2.788)	0.243
Immunologic marker
FoxP3	0.486 (0.244–0.970)	0.041	0.486 (0.244–0.970)	0.041
CD8 $\alpha$	0.974 (0.831–1.142)	0.747
CD31	1.006 (0.977–1.036)	0.684
CD68	1.206 (0.993–1.465)	0.059
pan-Keratin	1.031 (0.873–1.217)	0.723
CD56	1.004 (0.558–1.806)	0.989
Pathological factors and immunologic marker
Tumor size	1.073 (0.988–1.165)	0.094
Tumor number	2.745 (1.551–4.859)	0.001	2.440 (1.360–4.377)	0.003
Satellite nodules	2.449 (1.243–4.823)	0.010
MVI classification	3.244 (1.755–5.997)	0.000	3.096 (1.655–5.789)	0.000
Envelope invasion	1.791 (1.048–3.062)	0.033
HBV	2.401 (0.952–6.053)	0.063
AFP level	1.157 (1.064–1.258)	0.001
FoxP3	0.486 (0.244–0.970)	0.041
CD68	1.206 (0.993–1.465)	0.059	1.296 (1.055–1.592)	0.013

HR, hazard ratio; CI, confidence interval.

Univariate and multivariate survival analysis were used to analyze the prognostic values of the six markers and clinicopathological factors in HCC patients. As shown in Table 3, the analysis was divided into three sections. Among the clinicopathological factors listed above, univariate analysis revealed that tumor number (HR $=$ 2.745, 95% CI 1.551–4.859, $P=$ 0.001), satellite nodules (HR $=$ 2.449, 95% CI 1.243–4.823, $P=$ 0.010), MVI classification (HR $=$ 3.244, 95% CI 1.755–5.997, $P=$ 0.000), envelope invasion (HR $=$ 1.791, 95% CI 1.048–3.062, $P=$ 0.033) and AFP level (HR $=$ 1.157, 95% CI 1.064–1.258, $P=$ 0.001) were associated with RFS. No statistical significance related to RFS was detected in other clinicopathological factors including tumor size, HBV, cirrhosis postoperative TACE. Multivariate analysis revealed that tumor numbers (HR $=$ 3.020, 95% CI 1.609–5.669, $P=$ 0.001), MVI classification (HR $=$ 2.506, 95% CI 1.342–4.679, $P=$ 0.004) and envelope invasion (HR $=$ 2.007, 95% CI 1.116–3.608, $P=$ 0.020) were independent prognostic factors of RFS in HCC patients. Kaplan-Meier survival curves of tumor number, MVI and envelope invasion were also showed in Fig. 2. About the six molecular markers, the results of univariate analysis showed that only the expression of FoxP3 (HR $=$ 0.486, 95% CI 0.224–0.970, $P=$ 0.041) was related to RFS in HCC patients. No significant association was found between other molecular markers and RFS. By multivariate analysis, the results showed that the expression of FoxP3 (HR $=$ 0.486, 95% CI 0.224–0.970, $P=$ 0.041) was an independent prognostic factor of RFS in HCC patients after hepatectomy.

Based on the analysis of the two sections above, we included the clinicopathological factors and molecular markers whose $P$ value was below 0.1 in the univariate analysis to further analyze. Univariate analysis of Section 3 showed that tumor number (HR $=$ 2.745, 95% CI 1.551–4.859, $P=$ 0.001), satellite nodules (HR $=$ 2.449, 95% CI 1.243–4.823, $P=$ 0.010), MVI classification (HR $=$ 3.244, 95% CI 1.755–5.997, $P=$ 0.000), envelope invasion (HR $=$ 1.791, 95% CI 1.048–3.062, $P=$ 0.033), AFP level (HR $=$ 1.157, 95% CI 1.064–1.258, $P=$ 0.001) and the expression of FoxP3 (HR $=$ 0.486, 95% CI 0.224–0.970, $P=$ 0.041) were correlated to RFS in HCC patients. No statistical significance was found in the association of tumor size, HBV and the expression of CD68 with RFS in HCC patients. Multivariate analysis showed that number of tumors (HR $=$ 2.440, 95% CI 1.360–4.377, $P=$ 0.003), MVI classification (HR $=$ 3.096, 95% CI 1.655–5.789, $P=$ 0.000) and the expression of CD68 (HR $=$ 1.296, 95% CI 1.055–1.592, $P=$ 0.013) were independent prognostic factors of RFS in this cohort.

Table 4

Univariate and multivariate survival analysis of RFS in 28 HCC patients without postoperative TACE

Variable	COX
	Univariate analysis		Multivariate analysis
	HR (95% CI)	$p$ value	HR (95% CI)	$p$ value
Treg	0.741 (0.156–3.511)	0.706
CD8 $\alpha$	1.040 (0.770–1.405)	0.798
CD31	1.105 (1.000–1.220)	0.049	1.105 (1.000–1.220)	0.049
CD68	1.569 (0.934–2.635)	0.089
pan-Keratin	0.968 (0.702–1.335)	0.843
CD56	0.509 (0.109–2.368)	0.389

HR, hazard ratio; CI, confidence interval.

Figure 3.

Receiver operating characteristic curve analysis for predicting the RFS of HCC patients. ROC curves for the combination of tumor number and MVI, and the combination of tumor number, MVI and CD68, respectively. The sensitivity, specificity and AUC were indicated below the ROC graph.

Additionally, to exclude the influence of postoperative adjuvant therapy on TME, we also did stratification analysis in 28 HCC patients with no postoperative treatment to further identify the prognostic values of the six markers at baseline. As shown in Table 4, the results of univariate analysis showed that only the expression of CD31 (HR $=$ 1.105, 95% CI 1.000–1.220, $P=$ 0.049) was associated with RFS in HCC patients without postoperative treatment. No significant association was found between other molecular markers and RFS in this cohort. The results of multivariate analysis showed that the expression of CD31 (HR $=$ 1.105, 95% CI 1.000–1.220, $P=$ 0.049) was an independent prognostic factor of RFS in HCC patients without postoperative treatment.

3.3 The predictive value of the six molecular markers and pathological factors in HCC patients

As early recurrence is significantly associated with survival, we wanted to predict early recurrence with those highrisk factors identified above. Multivariate analysis revealed that tumor number, MVI classification and the expression level of CD68 were the independent predictive factor associated with RFS. Then, discriminative power of these three factors in predicting early-recurrence was evaluated by using the ROC curve. As shown in Fig. 3, the ROC curve of the two clinical pathological factors (tumor number and MVI) showed an AUC of 0.788 (85.42% sensitivity and 63.64% specificity). However, the ROC curve of the combination of tumor number, MVI and CD68 showed an AUC of 0.823 (68.75% sensitivity and 86.36% specificity). These results indicated that the combination of clinical pathological factors and the molecular markers was more accurate as a predictive indicator to evaluate the prognosis of HCC patients.

4. Discussion

The tumor microenvironment, especially the non-cancer cells, significantly influences tumor growth and survival and is closely related to clinical outcomes. To identify the immunologic cellular characteristics of TME may provide new perspective on the treatment of HCC [21]. In the present study, we assessed the expression of CD56, CD8 $\alpha$ , CD68, FoxP3, CD31, and pan-Keratin by using immunohistochemistry in HCC tumor tissues. We found that the expression of FoxP3 and CD31 were the independent prognostic factor for HCC patients that received surgical resection (Tables 3 and 4).

Forkhead box (Fox) proteins constitute a huge family of transcriptional regulators, and the roles of Fox proteins in tumorigenesis and cancer progression have been intensively studied [22]. A variety of Fox family members can be used as prognostic markers, such as FoxQ1 in breast cancer [23], FoxL1 in osteosarcoma [24, 25] and FoxO1 in epithelial ovarian cancer [26]. FoxP3 has been reported to act as a tumor suppressor in HCC. Shi et al. reported that higher expression of FoxP3 was related to better survival and lower recurrence risk. Moreover, FoxP3 may suppress tumor progression via the TGF- $\beta$ /Smad2/3 pathway in HCC [15]. Recently, Gong et al. claimed that the overexpression of FoxP3 can inhibit the growth of HCC via suppressing c-Myc directly or indirectly through interacting with Smad2/3/4 [27]. Consistent with these data, we also found that increased FoxP3 expression in tumor tissue was correlated with better RFS in HCC patients, and FoxP3 expression was negatively associated with envelop invasion, which was also an evidence suggesting that FoxP3 could be a favorable factor in the TME of HCC. As is well known, FoxP3 is a specific marker of Treg cells. To date, most of the published studies have showed that FoxP3 ${}^{+}$ Treg cells play an immunosuppressive role in the TME. Higher density or number intratumoral infiltrating FoxP3 ${}^{+}$ Treg cells were positively associated with worse clinical outcome in HCC patients [14, 28], and vice versa. Wang et al. reported that the early stage HCC patients with lower expression of tumour-infiltrating FoxP3 ${}^{+}$ Tregs had remarkably longer OS and RFS [29]. According to the expression levels of CD45RA ${}^{+}$ and FoxP3, Treg cells can be classified as three subpopulations including CD45RA ${}^{+}$ FoxP3 ${}^{\text{low}}$ resting Treg cells (rTreg cells), CD45RA ${}^{-}$ FoxP3 ${}^{\text{high}}$ activated Treg cells (aTreg cells) and cytokine-secreting nonsuppressive CD45RA ${}^{-}$ FoxP3 ${}^{\text{low}}$ non-Treg cells [30]. The aTreg cells are terminally differentiated from the rTreg cells, which are highly immune suppressive and have stable function. In contrast, non-Treg cells are immune stimulatory by producing inflammatory cytokines like IFN- $\gamma$ and IL-17 [30]. Takuro Saito et al. found that high-density of FoxP3 ${}^{\text{low}}$ Treg infiltration showed significantly better prognosis in colorectal cancer patients than those with predominantly FoxP3 ${}^{\text{high}}$ Treg cell infiltration; and TME with abundant infiltration of FoxP3 ${}^{\text{low}}$ Treg cells may depend on the high secretion of inflammatory cytokines such as TGF- $\beta$ and IL-12 [31]. Due to the deficiency of immunohistochemistry, we could not distinguish the subtype of Treg cells. The controversy about the prognostic significance of FoxP3 ${}^{+}$ Treg cells may be relevant to the subset of Treg cell infiltration and the inflammatory cytokines secretion in the TME of HCC. The function of the dominant subtype of Treg cell in HCC are needed to be further studied.

CD31 is a reliable marker for endothelial cells, which could involve in vascular development and maintain vascular endothelial barrier function [32, 33]. In this study, we found that CD31 expression was associated with MVI presence and RFS in HCC patients without any postoperative adjuvant therapy. Previous study showed that CD31 expression was associated with a poor survival in HCC patients [18], which was consistent with our result. Furthermore, we found that CD8 $\alpha$ expression was negatively correlated with MVI presence. Hitherto, the relationship between CD8 $\alpha$ expression and MVI are rarely reported in HCC. Armani et al. showed that a high number of intratumoral CD8 ${}^{+}$ cells was related to increased vascular invasion in non-small cell lung cancer [10], which was contradictory with our result. So further studies are required to identify the correlation between CD8 $\alpha$ expression and MVI and the underlying mechanisms. MVI presence has been identified as a critical factor for early recurrence and poor outcome in HCC patients [34, 35]. Our previous work also found that the patients with MVI had a shorter RFS than those without MVI in postoperative HCC patients [36]. Here, we again demonstrated that MVI was an independent prognostic factor for the RFS in HCC patients after hepatectomy.

CD68 is widely accepted as a cell surface marker of macrophage. In this work, we found CD68 expression was an independent risk factor for AFP, and the higher AFP level was associated with shorter RFS in HCC patients. AFP is commonly considered as a prognostic factor of HCC [37]. When the clinicopathological factors and the 6 molecular biomarkers are included in the multivariate analysis, the results showed that CD68 expression was an independent prognostic factors of RFS in HCC patients, which was consistent with the study that reported previously [13]. Moreover, we found that CD56 expression was associated with tumor size. It has been reported that active CD56 ${}^{+}$ NK cells were associated with the transition from cirrhosis to HCC [11]. Additionally, Aloysius et al. found that CD56 could be a prognostic marker in periampullary cancer [38]. All these results provide hints that CD56 may play a role in promoting tumor progression.

The limitation of our work is that we just detected the expression of these markers in the resected tumor tissue but did not follow the subtype of these immune cells. As we known, different subtypes of the immune cells can perform different or even opposite functions. Further studies are required to determine the function of the dominant subtype of the immune cells in HCC.

In summary, our study found that CD68 and FoxP3 showed significant relationships with RFS of HCC patients. Tumor number, MVI classification and the expression level of CD68 were the independent predictive factors associated with RFS. The classifier established by the combination of tumor number, MVI and CD68 was higher than that of just tumor number and MVI. All these suggest that the characteristics of TME are associated with prognosis of HCC patients.

Author contributions

Conception: Mei Liu.

Interpretation or analysis of data: Liming Wang.

Preparation of the manuscript: Shuwen Kuang, Mei Liu, Liming Wang, Jibing Liu.

Revision for important intellectual content: Jibing Liu, Yiling Zheng, Liming Wang.

Supervision: Jibing Liu, Mei Liu.

Supplementary data

The supplementary files are available to download from http://dx.doi.org/10.3233/CBM-203003.

sj-docx-1-cbm-10.3233_CBM-203003.docx - Supplemental material

Supplemental material, sj-docx-1-cbm-10.3233_CBM-203003.docx

Footnotes

Acknowledgments

This work was supported by Beijing Municipal Science and Technology Project (Z181100001718191), National Natural Science Foundation (81641113).

Conflict of interest

There are no conflicts to disclose.

References

Bray

Ferlay

Soerjomataram

Siegel

R.L.

Torre

L.A.

and Jemal

, Global cancer statistics 2018: GLOBOCAN estimates of incidence and mortality worldwide for 36 cancers in 185 countries, CA Cancer J Clin 68 (2018), 394–424.

Yang

J.D.

Hainaut

Gores

G.J.

Amadou

Plymoth

and Roberts

L.R.

, A global view of hepatocellular carcinoma: trends, risk, prevention and management, Nat Rev Gastroenterol Hepatol 16 (2019), 589–604.

Krenzien

Schmelzle

Struecker

Raschzok

Benzing

Jara

Bahra

Ollinger

Sauer

I.M.

Pascher

Pratschke

and Andreou

, Liver transplantation and liver resection for cirrhotic patients with hepatocellular carcinoma: comparison of long-term survivals, J Gastrointest Surg 22 (2018), 840–848.

Bruix

Gores

G.J.

and Mazzaferro

, Hepatocellular carcinoma: clinical frontiers and perspectives, Gut 63 (2014), 844–855.

Novikova

M.V.

Khromova

N.V.

and Kopnin

P.B.

, Components of the hepatocellular carcinoma microenvironment and their role in tumor progression, Biochemistry (Mosc) 82 (2017), 861–873.

Cariani

and Missale

, Immune landscape of hepatocellular carcinoma microenvironment: Implications for prognosis and therapeutic applications, Liver Int 39 (2019), 1608–1621.

Liu

Zeng

and Shen

, From bench to bed: the tumor immune microenvironment and current immunotherapeutic strategies for hepatocellular carcinoma, J Exp Clin Cancer Res 38 (2019), 396.

Nishida

and Kudo

, Immunological microenvironment of hepatocellular carcinoma and its clinical implication, Oncology 92(Suppl 1) (2017), 40–49.

Tamma

Ruggieri

Annese

Simone

Mangia

Rega

Zito

F.A.

Nico

and Ribatti

, Bcl6/p53 expression, macrophages/mast cells infiltration and microvascular density in invasive breast carcinoma, Oncotarget 9 (2018), 22727–22740.

10.

Armani

Madeddu

Mazzaschi

Bocchialini

Sogni

Frati

Lorusso

Falco

Lagrasta

C.A.

Cavalli

Mangiaracina

Vilella

Becchi

Gnetti

Corradini

Quaini

Urbanek

Goldoni

Carbognani

Ampollini

and Quaini

, Blood and lymphatic vessels contribute to the impact of the immune microenvironment on clinical outcome in non-small-cell lung cancer, Eur J Cardiothorac Surg 53 (2018), 1205–1213.

11.

Zekri

A.N.

El Deeb

Bahnassy

A.A.

Badr

A.M.

Abdellateif

M.S.

Esmat

Salama

Mohanad

El-Dien

A.E.

Rabah

and Abd Elkader

, Role of relevant immune-modulators and cytokines in hepatocellular carcinoma and premalignant hepatic lesions, World J Gastroenterol 24 (2018), 1228–1238.

12.

Yao

J.C.

Yang

Wang

J.M.

Qian

Y.W.

Yang

and Ji

, The prognostic value of tumor-infiltrating lymphocytes in hepatocellular carcinoma: a systematic review and meta-analysis, Sci Rep 7 (2017), 7525.

13.

Ding

Tan

Qian

Xue

Wang

Jiang

and Xu

, Clinicopathologic and prognostic significance of tumor-associated macrophages in patients with hepatocellular carcinoma: a meta-analysis, PLoS One 14 (2019), e0223971.

14.

J.F.

Ding

Y.H.

Ying

X.H.

F.Z.

Zhou

X.M.

Zhang

D.K.

Zou

and Ji

J.S.

, Regulatory T cells, especially ICOS(+) FOXP3(+) regulatory T cells, are increased in the hepatocellular carcinoma microenvironment and predict reduced survival, Sci Rep 6 (2016), 35056.

15.

Shi

J.Y.

L.J.

Zhang

J.W.

Duan

Ding

Z.B.

Yang

L.X.

Cao

Zhou

Fan

Zhang

Zhao

Y.J.

Wang

X.Y.

and Gao

, FOXP3 Is a HCC suppressor gene and Acts through regulating the TGF-beta/Smad2/3 signaling pathway, BMC Cancer 17 (2017), 648.

16.

Park

DiMaio

T.A.

Scheef

E.A.

Sorenson

C.M.

and Sheibani

, PECAM-1 regulates proangiogenic properties of endothelial cells through modulation of cell-cell and cell-matrix interactions, Am J Physiol Cell Physiol 299 (2010), C1468–1484.

17.

Bosmuller

Pfefferle

Bittar

Scheble

Horger

Sipos

and Fend

, Microvessel density and angiogenesis in primary hepatic malignancies: differential expression of CD31 and VEGFR-2 in hepatocellular carcinoma and intrahepatic cholangiocarcinoma, Pathol Res Pract 214 (2018), 1136–1141.

18.

Zhang

Y.Y.

Kong

L.Q.

Zhu

X.D.

Cai

Wang

C.H.

Shi

W.K.

Cao

M.Q.

X.L.

K.S.

Zhang

S.Z.

Chai

Z.T.

J.Y.

B.G.

and Sun

H.C.

, CD31 regulates metastasis by inducing epithelial-mesenchymal transition in hepatocellular carcinoma via the ITGB1-FAK-Akt signaling pathway, Cancer Lett 429 (2018), 29–40.

19.

Govaere

Komuta

Berkers

Spee

Janssen

de Luca

Katoonizadeh

Wouters

van Kempen

L.C.

Durnez

Verslype

De Kock

Rogiers

van Grunsven

L.A.

Topal

Pirenne

Vankelecom

Nevens

van den Oord

Pinzani

and Roskams

, Keratin 19: a key role player in the invasion of human hepatocellular carcinomas, Gut 63 (2014), 674–685.

20.

Hassan

Mansy

S.S.

Tabak

S.A.

AbdelFattah

A.S.

Abdel-Aziz

A.M.

Hamam

Seleem

M.I.

and Abdelaal

, Immunohistochemical and electron microscopic morphometric image analysis of hepatocellular carcinoma in association of HCV infection, Ultrastruct Pathol 42 (2018), 97–107.

21.

Atanasov

Dino

Schierle

Dietel

Aust

Pratschke

Seehofer

Schmelzle

and Hau

H.M.

, Immunologic cellular characteristics of the tumour microenvironment of hepatocellular carcinoma drive patient outcomes, World J Surg Oncol 17 (2019), 97.

22.

Bach

D.H.

Long

N.P.

Luu

T.T.

Anh

N.H.

Kwon

S.W.

and Lee

S.K.

, The dominant role of forkhead box proteins in cancer, Int J Mol Sci 19 (2018).

23.

Elian

F.A.

Are

Ghosh

Nuin

Footz

McMullen

T.P.W.

Brindley

D.N.

and Walter

M.A.

, FOXQ1 is differentially expressed across breast cancer subtypes with low expression associated with poor overall survival, Breast Cancer (Dove Med Press) 13 (2021), 171–188.

24.

Chen

Deng

Zhou

Xiao

Zhou

Zhang

and Wu

, Inhibitory effects of forkhead box L1 gene on osteosarcoma growth through the induction of cell cycle arrest and apoptosis, Oncol Rep 34 (2015), 265–271.

25.

Zhang

Duan

Song

Zhang

and Chen

, The emerging roles of forkhead box (FOX) proteins in osteosarcoma, J Cancer 8 (2017), 1619–1628.

26.

Han

G.H.

Chay

D.B.

Nam

Cho

Chung

J.Y.

and Kim

J.H.

, Prognostic implications of forkhead box protein O1 (FOXO1) and paired box 3 (PAX3) in epithelial ovarian cancer, BMC Cancer 19 (2019), 1202.

27.

Gong

Jia

Liu

Ren

Yang

Liu

Lai

P.B.S.

and Chen

G.G.

, Nuclear FOXP3 inhibits tumor growth and induced apoptosis in hepatocellular carcinoma by targeting c-Myc, Oncogenesis 9 (2020), 97.

28.

Huang

Wang

Zhu

Gao

and Du

, Intrahepatic interleukin-17+ T cells and FoxP3+ regulatory T cells cooperate to promote development and affect the prognosis of hepatocellular carcinoma, Journal of gastroenterology and hepatology 29 (2014), 851–859.

29.

Wang

Jing

Wang

Yang

Zhu

Gao

Zhang

Yang

Wang

and Du

, Foxp3+ regulatory T cells are associated with the natural history of chronic hepatitis B and poor prognosis of hepatocellular carcinoma, Liver Int 32 (2012), 644–655.

30.

Miyara

Yoshioka

Kitoh

Shima

Wing

Niwa

Parizot

Taflin

Heike

Valeyre

Mathian

Nakahata

Yamaguchi

Nomura

Ono

Amoura

Gorochov

and Sakaguchi

, Functional delineation and differentiation dynamics of human CD4+ T cells expressing the FoxP3 transcription factor, Immunity 30 (2009), 899–911.

31.

Saito

Nishikawa

Wada

Nagano

Sugiyama

Atarashi

Maeda

Hamaguchi

Ohkura

Sato

Nagase

Nishimura

Yamamoto

Takiguchi

Tanoue

Suda

Morita

Hattori

Honda

Mori

Doki

and Sakaguchi

, Two FOXP3(+)CD4(+) T cell subpopulations distinctly control the prognosis of colorectal cancers, Nat Med 22 (2016), 679–684.

32.

Lertkiatmongkol

Liao

Mei

and Newman

P.J.

, Endothelial functions of platelet/endothelial cell adhesion molecule-1 (CD31), Curr Opin Hematol 23 (2016), 253–259.

33.

Fernandez-Martin

Marcos-Ramiro

Bigarella

C.L.

Graupera

Cain

R.J.

Reglero-Real

Jimenez

Cernuda-Morollon

Correas

Cox

Ridley

A.J.

and Millan

, Crosstalk between reticular adherens junctions and platelet endothelial cell adhesion molecule-1 regulates endothelial barrier function, Arterioscler Thromb Vasc Biol 32 (2012), e90–102.

34.

Lauwers

G.Y.

Terris

Balis

U.J.

Batts

K.P.

Regimbeau

J.M.

Chang

Graeme-Cook

Yamabe

Ikai

Cleary

K.R.

Fujita

Flejou

J.F.

Zukerberg

L.R.

Nagorney

D.M.

Belghiti

Yamaoka

Vauthey

J.N.

and C. International Cooperative Study Group on Hepatocellular, Prognostic histologic indicators of curatively resected hepatocellular carcinomas: a multi-institutional analysis of 425 patients with definition of a histologic prognostic index, Am J Surg Pathol 26 (2002), 25–34.

35.

Lim

K.C.

Chow

P.K.

Allen

J.C.

Chia

G.S.

Lim

Cheow

P.C.

Chung

A.Y.

Ooi

L.L.

and Tan

S.B.

, Microvascular invasion is a better predictor of tumor recurrence and overall survival following surgical resection for hepatocellular carcinoma compared to the Milan criteria, Ann Surg 254 (2011), 108–113.

36.

Liu

Wang

Zhu

Rong

Liang

and Wu

, A preoperative measurement of serum MicroRNA-125b may predict the presence of microvascular invasion in hepatocellular carcinomas patients, Transl Oncol 9 (2016), 167–172.

37.

Sauzay

Petit

Bourgeois

A.M.

Barbare

J.C.

Chauffert

Galmiche

and Houessinon

, Alpha-foetoprotein (AFP): a multi-purpose marker in hepatocellular carcinoma, Clin Chim Acta 463 (2016), 39–44.

38.

Aloysius

M.M.

Zaitoun

A.M.

Awad

Ilyas

Rowlands

B.J.

and Lobo

D.N.

, Mucins and CD56 as markers of tumour invasion and prognosis in periampullary cancer, Br J Surg 97 (2010), 1269–1278.

Prognostic and predictive significance of the tumor microenvironment in hepatocellular carcinoma

Abstract

BACKGROUND:

OBJECTIVE AND METHODS:

RESULTS:

CONCLUSIONS:

Keywords

1. Introduction

2. Methods and materials

2.1 Patients with HCC

2.2 Immunohistochemistry and scoring

2.3 Follow-up

2.4 Survival analysis

Table 1 Comparisons of baseline immunology characteristics between the groups with different pathological factors

3. Results

3.1 The association between clinicopathological factors and the six molecular markers in 100 HCC patients

4. Discussion

Author contributions

Supplementary data

sj-docx-1-cbm-10.3233_CBM-203003.docx - Supplemental material

Footnotes

Acknowledgments

Conflict of interest

References

Table 1
Comparisons of baseline immunology characteristics between the groups with different pathological factors