• Open Access

Expression of TREM-1 in hepatic stellate cells and prognostic value in hepatitis B-related hepatocellular carcinoma

Authors


To whom correspondence should be addressed.

E-mail: qiu.shuangjian@zs-hospital.sh.cn; jiafan99@yahoo.com

Abstract

Hepatocellular carcinoma (HCC) is a typical inflammation-related malignancy characterized by high postoperative recurrence and metastasis. Although several inflammatory cells and inflammatory signatures have been linked to poor prognosis, the inflammation-associated molecular mechanisms of HCC development and progression are largely unknown. Here we show that triggering receptor expressed in myeloid cells (TREM)-1, a transmembrane receptor expressing in myeloid cells, was also expressed in tumor-activated hepatic stellate cells (HSCs) and associated with the aggressive behavior of HCC cells. Enzyme-linked immunosorbent assay was used to measure the expression levels of soluble TREM-1 (sTREM-1) in activated hepatic stellate cells supernatant and 92 preoperative and postoperative plasmas of patients with malignancy and/or benign liver tumor/disease, respectively. Expression levels of TREM-1 were assessed by immunohistochemistry in tissue microarray from 240 patients with HCC. As a result, increased secretion of sTREM-1 from activated HSCs was observed after co-culture with HCC cell lines (P < 0.001), and conditioned medium collected from activated HSCs/cancer associated myofibroblasts (CAMFs) with or without agonist/inhibitor of TREM-1 significantly changed the migratory ability of HCC cells. The levels of sTREM-1 were significantly higher in patients with HCC than those with benign liver tumors (P < 0.005). Peritumoral density of TREM-1 was shown to be an independent prognosis predictor according to univariate (P < 0.001 for both overall survival and time to recurrence) and multivariate analysis (P = 0.008 for overall survival; P = 0.005 for time to recurrence). Thus, these observations suggest that TREM-1 is related to the aggressive tumor behavior and has potential value as a prognostic factor for HCC. (Cancer Sci 2012; 103: 984–992)

Cancer is characterized by complex tissues composed of multiple distinct cell types contributing to progressive development, metastatic dissemination, and high postoperative recurrence. Its progression is not just determined by cancer cells themselves, but also depends on various inflammatory cells recruited to tumor and stroma.[1, 2] There is increasing evidence that various inflammatory cells are responsible for the tumor-promoting activities after seeding and establishment of metastatic cells.[3, 4] Hepatocellular carcinoma (HCC) is a typical inflammation-related malignancy mainly triggered by exposure to hepatitis viral infection.[5] Therefore, inflammatory status has contributed to the extremely poor prognosis of HCC. Our previous results[6-9] supported these connections by providing various inflammatory cells (such as activated hepatic stellate cells, macrophages, mast cells, and neutrophils) to discriminate populations of HCC with different survival outcomes. A unique peritumoral inflammatory/immune response signature of the liver microenvironment contributing to cancer metastatic propensity has also been linked to recurrence and metastasis of HCC,[10] suggesting the interplay between inflammation and cancer. Furthermore, some proinflammatory cytokines levels, like interleukin (IL)-2, -6, and -15, were useful for stratifying HCC patients according to different prognosis after curative resection,[11, 12] strongly indicating a tumor-promoting direction shift from pro-inflammatory response. Notably, hepatic stellate cells (HSCs) are pivotal members of the stromal cells accelerating this transition. To reveal the mechanisms involved in cancer progression, various interpretations were raised. First, in response to chronic hepatic inflammation, HSCs can undergo activation and transdifferentiation into highly proliferative myofibroblast-like cells, considered as a crucial event promoting liver fibrosis[13] and subsequent pro-tumor effects.[14, 15] Second, for both tumorigenesis and hepatic fibrogenesis, interventions with certain active signals in HSCs such as transforming growth factor-β[16], hepatocyte growth factor,[17] and platelet-derived growth factor[18] execute key roles in these pathological events. Finally, upon activation directly induced by tumor cells, HSCs are conferred potency as a promoter in angiogenesis,[19] invasiveness,[17] and metastasis[20] through the molecular framework of tumor–stromal interaction. Despite the mechanisms of HSCs implicated in tumor progression, very little is known about the nature and modulation of activated HSCs facilitating metastasis and recurrence in HCC.

Triggering receptor expressed on myeloid cells (TREM)-1 was initially recognized as a transmembrane receptor expressed together with DNAX activation protein of 12 kDa (DAP12) in neutrophils and monocytes/macrophages.[21] Significantly, during infection stimulated by LPS, engagement of TREM-1 per se can lead to myeloid cell activation producing large numbers of inflammatory cytokines such as tumor necrosis factor (TNF)-α, IL-1, and IL-6, indicating its ability to amplify inflammatory response.[22, 23] In addition to its proinflammatory effect, TREM-1 carries out other important biological functions, such as: (i) suppressive function against the induction of key immunomodulatory factors by LPS;[24] (ii) regulatory effect on the adaptive response, as well as the innate response, by recruiting effector lymphocytes through augmenting monocyte production of chemokines;[25] (iii) programming mature dendritic cell functions through its inducibility by hypoxia within pathologic tissues;[26] and (iv) tumor promoting effect on regulating the malignant behavior of cancer cells in lung cancer.[27] Of note, activated HSCs can express LPS-recognizing receptors such as CD14, TLR4, and MD2.[28] In turn, LPS signaling is clearly capable of enhancing liver damage by HSCs developing a strong proinflammatory phenotype and exacerbating cytokine production.[29, 30] Increased inflammatory cytokines (e.g. TNF-α, IL-6, and IL-10) from HSCs also participate in the modulation of TREM-1 signaling.[13, 31] Furthermore, after stimulating inflammatory cells in the tumor microenvironment, LPS can directly affect tumor cell proliferation, angiogenesis, invasion, and metastasis.[32, 33] Thus, it would be intriguing to examine whether the TREM-1 pathway, in HCC milieus, is related with the modulation of LPS signaling in activated HSCs. However, no related studies about the role of TREM-1 in HCC have been carried out.

We speculated that TREM-1 might directly perform a regulatory function and is a critical functional molecule in activated HSCs during liver inflammatory injury and cancer. In this study, our results provide the first evidence that TREM-1 was expressed in HSCs and promoted the migration of HCC cells. We investigated the prognostic value of TREM-1 and the association between TREM-1 and HSCs in HCC, and found peritumoral TREM-1 was related with poor prognosis of HCC after resection.

Materials and Methods

Patients and specimens

This study was specifically approved by the Ethics Review Committee of Zhongshan Hospital (Fudan University, Shanghai, China) and written informed consent from all participants involved in our study was obtained. As previously described,[6] all archival specimens were obtained from 130 consecutive patients with pathologically confirmed HCC who received surgical resection from 2002 to 2005 (Table S1). Another independent cohort of 110 patients was used to validate the prognostic value of TREM-1 (Table S1).

A total of 92 preoperative and postoperative (at 5 days) plasmas of malignancy and/or benign liver tumor/disease patients without acute infection were prospectively collected at our hospital from August to December 2010.

Tissue microarray design and evaluation of immunohistochemical variables

Paraffin blocks were selected (approximately 4 mm thick) and tissue microarrays were constructed as described previously.[6] Details regarding the work plan for the tissue microarrays and immunohistochemistry are presented in Data S1.

Soluble TREM-1, IL-6, and TNF-α assays in plasmas

Soluble TREM-1 (sTREM-1) in body fluids is a useful marker in establishing or excluding the diagnosis of sepsis or cancer, therefore, plasmas of patients with HCC were used to detect sTREM-1 levels. Both IL-6 and TNF-α assays were also carried out.

Isolation and culture of cells

Considering the predictive value of peritumoral activated HSCs,[6] we isolated HSCs from surgical peritumoral specimens and cancer associated myofibroblasts (CAMF) from matched cancer tissues from six patients with HCC or normal tissues of four hepatic hemangiomas, as described previously with minor modification (Fig. S1).[34] All selected patients were already infected with hepatitis B. Cells were cultured in DMEM supplemented with 10% FCS and 1% penicillin–streptomycin in 95% air and 5% CO2 at 37°C (see Data S1 for details).

Double-label immunofluorescent staining

Isolated HSCs/CAMFs were stained by double-label immunofluorescence in order to identify co-localization of TREM-1 with some phenotype markers of HSCs/CAMFs (described in detail in Data S1).

In vitro co-culture of activated HSCs/CAMFs and HCC cells lines or normal hepatocyte lines

In some experiments, for the assessment of the effects of LPS or hepatocarcinoma cells on HSCs/CAMFs and their modulatory effects on TREM-1 expression, HSCs/CAMFs were plated in 24-well plates with different concentration of LPS (10 μg/mL, 100 ng/mL, or 1 ng/mL) or HCC lines (Hep G2 or Hep 3B) or a normal hepatocyte line (LO2) for the indicated times (see Data S1 for details).

Peptide synthesis and collection of conditioned medium

For inhibition, TREM-1 peptide or control peptide was chemically synthesized as described previously.[35] Conditioned medium (CM) of HSCs/CAMFs was then collected as described elsewhere[14] with modification (see Data S1 for details).

Transwell migration assay

To investigate the effect of TREM-1 in HSCs/CAMFs on HCC cells, we seeded HSCs/CAMFs with HCC in Transwell filter inserts and carried out migration assays (detail in Data S1).

Statistic analysis

Statistical analysis was completed with spss 16.0 (SPSS, Chicago, IL, USA) and a two-tailed P-value <0.05 was considered significant (detail in Data S1).

Results

Hepatic stellate cells/myofibroblasts, hepatocytes, and cancer cells expressed TREM-1

Positive expression of TREM-1 was detected in both intratumoral areas and peritumoral tissues in most samples (Table 1). Expression of CD14+ monocytes was much weaker or negative in cancerous tissues of HCC patients (data not shown). Interestingly, we found cancer cells and hepatocytes both showed positive cytoplasmic staining (Fig. S2). Representative TREM-1 images and statistics of immunohistochemical variables are shown in Fig. 1(a–d) and Table S1. Strong expression of TREM-1 was found in CAMFs and activated HSCs (Fig. 1e). Freshly isolated from peritumoral tissues after 2 days, HSCs showed an early activated phenotype (α-SMA positive). LX-2 was also positive staining of TREM-1 by immunocytochemistry (Fig. S3).

Figure 1.

Expression levels of triggering receptor expressed on myeloid cells (TREM)-1 by immunohistochemistry and immunofluorescent analysis. High (a,c) and low (b,d) density of TREM-1 staining cells in intratumoral (a,b) and peritumoral areas (c,d). (e) Results of immunofluorescent analysis of primary hepatic stellate cells (HSCs) and cancer associated myofibroblasts (CAMFs) at 2, 7, and 14 days after isolation from human peritumoral and cancer tissues using fluorescence-labeled TREM-1 (red), Desmin (green), α-SMA (green), and DAPI (blue) stained cell nuclei.

Table 1. Descriptive statistics of peritumoral triggering receptor expressed on myeloid cells (TREM-1) immunohistochemical positive staining in testing and validation cohorts
VariableaMean ± SEMedianRangeP (paired t-test)b
  1. a

    Number of positive staining cells per field (magnification, ×400).

  2. b

    Peritumoral versus intratumoral.

TREM-1 of testing cohort (n = 130)
Peritumoral97.66 ± 4.281090–187 
Intratumoral82.86 ± 4.16800–169P = 0.011
TREM-1 of validation cohort (n = 110)
Peritumoral70.36 ± 3.04660–177 
Intratumoral59.33 ± 3.21520–156P = 0.014

Release of sTREM-1 from HSCs/myofibroblasts in LPS-induced inflammation was time-dependent, not dose-dependent

The release of sTREM-1 from activated HSCs/CMAFs by LPS was induced in a time-dependent, not dose-dependent, manner. The levels of sTREM-1 from activated HSCs/myofibroblasts were all detected to elevate at 30 min using LPS (10 μg/mL–1 ng/mL). A low concentration of LPS (1 ng/mL) began to enhance the secretion of sTREM-1 for 5 min (in vitro activated HSC, 48.34 ± 0.51 pg/mL; in vivo activated HSC, 48.66 ± 0.84 pg/mL; CAMFs, 51.73 ± 0.76 pg/mL, respectively), and the sTREM-1 expression level peaked at 1–2 h in activated HSCs/CAMFs after LPS stimulation. Increased dose of LPS stimulation did not significantly affect the secretion amount of activated HSCs/CAMFs. Although stimulated by increased time and dose of LPS, sTREM-1 levels secreted by quiescent phenotype HSCs showed no significant difference (Fig. 2).

Figure 2.

Release of triggering receptor expressed on myeloid cells (TREM)-1 by lipopolysaccharide (LPS)-induced inflammation. (a–c) Effects of various doses of LPS stimulation (10 μg–1 ng/mL). Data shown are the mean ± SD (n = 3). aHSC, activated hepatic stellate cell; CAMFs, cancer associated myofibroblasts; qHSC, quiescent phenotype hepatic stellate cell.

Expression levels of sTREM-1, IL-6, and TNF-α were increased in supernatants of activated HSCs/CAMFs co-cultured with HCC cells or normal hepatocytes

Tumor cells and activated HSCs/CAMFs produced an inflammatory cytokine milieu as well as provide cell–cell contact engagement that facilitates the generation of sTREM-1. Compared with quiescent HSCs, in vitro and in vivo activated HSCs/CAMFs secreted high levels of IL-6 (97.52 ± 10.56 pg/mL, 116.34 ± 24.78 pg/mL, and 130.92 ± 22.53 pg/mL, respectively) and TNF-α (17.10 ± 1.16 pg/mL, 17.22 ± 1.14 pg/mL, and 28.06 ± 2.67 pg/mL, respectively; P < 0.05). In vivo activated HSCs and CAMFs also increased the release of sTREM-1 (31.54 ± 1.58 pg/mL and 37.73 ± 2.38 pg/mL, respectively). The levels of sTREM-1, IL-6, and TNF-α were upregulated in the culture medium supernatants of activated HSCs/CAMFs after co-culture with HCC cell lines (Fig. 3a, P < 0.05). In addition, considering the close structural and functional relationship between HSCs and adjacent cells such as hepatocytes, normal hepatocytes lines LO2 were used to culture with activated HSCs/CAMFs. Activated HSCs/CAMFs co-cultured with hepatocytes also promoted the expression levels of sTREM-1, IL-6, and TNF-α (Fig. 3a, P < 0.05).

Figure 3.

Enzyme-linked immunosorbent assay for triggering receptor expressed on myeloid cells (TREM)-1, interleukin (IL)-6, and tumor necrosis factor (TNF)-α. (a) Activated hepatic stellate cells (HSCs) and cancer associated myofibroblasts (CAMFs) released high levels of IL-6 and TNF-α, and increased the release of TREM-1, IL-6, and TNF-α after co-culture with cancer cells or normal hepatocytes (P < 0.05). Cancer cells and normal hepatocytes had no significant effect on secretion of TREM-1, IL-6, or TNF-α by non-activated HSCs. (b) Expression levels of soluble TREM-1 (sTREM-1), IL-6, and TNF-α were upregulated in plasmas of hepatocellular carcinoma (HCC) patients compared with patients with benign disease or postoperative HCC at 5 days. *P < 0.005, vs other groups; #P < 0.005, vs other groups; ▴P < 0.005, vs other groups; **P < 0.05, vs other groups; ##P < 0.05, vs other groups; ▴▴P < 0.05, vs other groups. aHSC, activated HSC; qHSC, quiescent phenotype HSC.

Expression levels of sTREM-1, IL-6, and TNF-α upregulated in plasmas of patients with HCC

To investigate the expression levels of sTREM-1, ELISA was used to analyze the differences between patients with malignant and benign liver tumors. These patients were divided into five groups: HCC; hemangiomas; benign liver disease (cyst); focal nodular hyperplasia; and postoperative HCC. The levels of sTREM-1 (117.61 ± 12.05 pg/mL), IL-6 (8.36 ± 0.35 pg/mL), and TNF-α (59.48 ± 5.71 pg/mL) were significantly higher in patients with HCC than in those with benign liver tumor/disease and postoperative HCC at 5 days (P < 0.005). There was no difference between the groups of patients with benign liver tumors (Fig. 3b, P > 0.05).

Increased migration of HCC cells by TREM-1

In control medium, only a few cancer cells could migrate through the chambers. In contrast, activated HSCs/CAMFs-CM present in the lower compartment was able to increase the migration of cancer cells (P < 0.05). Furthermore, migration was greatly increased in both types of cancer cells after incubation with activated HSCs/CAMFs-CM and anti-TREM-1 agonist mAb (P < 0.05). Following inhibition of TREM-1 by LP17, migration was decreased in both type of cancer cells compared with those cultured in activated HSCs/CAMFs-CM (P < 0.01). Non-activated HSCs had no effect on migration of HCC cell lines (Fig. 4, P > 0.05).

Figure 4.

Migration assay of cancer cells. (A) Representative pictures of Hep3B cells' in vitro migration assay using conditioned medium (CM) of activated hepatic stellate cells (aHSCs) and/or inhibitor (peptide) of triggering receptor expressed on myeloid cells (TREM)-1 (LP17), or control peptide (LP17C), or anti-TREM-1 agonist mAb. (b,c) Number of migrated Hep3B (B) and HepG2 (C) cells (×200) both showed that CM of activated HSCs/CAMFs increase migration of cancer cells compared with control (P < 0.05). Furthermore, migration was greatly increased after incubation with CM and mAb (CM + mAb, P < 0.05). Migration was decreased after adding inhibitor of TREM-1 LP17 into CM (CM+LP17) compared with those cultured in CM of activated HSCs/cancer associated myofibroblasts (CAMFs) (P < 0.01). Non-activated HSCs had no effect on migration of HCC cell lines. *P < 0.05, vs CM or CM + LP17 or control; **P < 0.05, vs CM or control; ***P < 0.05, compared with control.

Association of TREM-1 density with clinicopathologic variables and survival analyses

In the testing cohort, intratumoral and peritumoral TREM-1 were investigated to compare the association with clinicopathologic variables. Peritumoral TREM-1 was found to be associated with several other variables, including vascular invasion (P < 0.001), tumor size (P = 0.001), and high TNM stage (P < 0.001) (Table S1). As shown in Table 2, on univariate analysis of our data, AFP, tumor multiplicity, tumor size, vascular invasion, tumor encapsulation, and TNM stage showed prognostic significance for both overall survival (OS) and time to recurrence (TTR). High density of peritumoral α-SMA and TREM-1 were associated with poor OS (both < 0.001) and TTR (both P < 0.001; Table 2, Fig. 5). In contrast, neither intratumoral α-SMA nor intratumoral TREM-1 were associated with death or recurrence. Significant factors were used for further multivariate analysis. Increased peritumoral α-SMA and TREM-1 were associated with death (P = 0.009 and P = 0.008, respectively) and elevated risks of recurrence (P = 0.015 and P = 0.005, respectively; Table 2, Fig. 5). Vascular invasion was shown to be related with both OS and TTR, and tumor differentiation was an independent predictor for TTR. According to TTR, recurrence was divided into early recurrence (metastasis after surgery ≤24 months) and late recurrence (new primary lesion after surgery >24 months).[36] Univariate and multivariate analysis both indicated that patients with high peritumoral TREM-1 levels were more likely to suffer from early tumor recurrences (univariate analysis, < 0.001; multivariate analysis, P = 0.006) rather than late recurrences (Table S2). Similarly, high density of peritumoral TREM-1 significantly correlated with poor OS (P < 0.001) and TTR (P = 0.002) in a validation cohort (Table S3). Moreover, there was positive relation between α-SMA and TREM-1 by correlation analysis (r = 0.199, P = 0.023). In those patients with a high density of α-SMA and TREM-1, 5-year OS rates were only 38% and 47%, and the TTR rates were 28% and 37%.

Figure 5.

Prognostic roles of triggering receptor expressed on myeloid cells (TREM)-1 by Kaplan–Meier analysis. (a,b) Kaplan–Meier estimates of overall survival according to high or low expression levels of TREM-1 positive cells in testing (a) and validation cohorts (b). (c,d) Kaplan–Meier estimates of time to recurrence of peritumoral TREM-1 positive staining cells in testing (c) and validation cohorts (d).

Table 2. Univariate and multivariate analyses of factors associated with survival and recurrence in the hepatocellular carcinoma group of testing cohort (n = 130)
FactorsOSTTR
UnivariateMultivariateUnivariateMultivariate
PHR (95% CI)PPHR (95% CI)P
  1. Univariate analysis, Kaplan–Meier method; multivariate analysis, Cox proportional hazards regression model. α-SMA, α-smooth muscle actin; AFP, alpha fetoprotein; ALT, alanine aminotransferase; HBeAg, hepatitis B e antigen; HR, hazard ratio; NA, not adopted; ND, no data; NS, not significant; OS, overall survival; TREM-1, triggering receptor expressed on myeloid cells; TTR, time to recurrence.

Age, years (≤52 vs >52)0.277NDNA0.058NDNA
Gender (male vs female)0.554NDNA0.268NDNA
ALT (U/L) (≤75 vs >75)0.776NDNA0.710NDNA
AFP (ng/mL) (≤20 vs >20)0.004NDNS0.003NDNS
Hepatitis history (yes vs no)0.344NDNA0.815NDNA
HBeAg (Positive vs negative)0.076NDNA0.160NDNA
Liver cirrhosis (yes vs no)0.067NDNA0.185NDNA
Tumor differentiation0.0361.706 (1.082–2.690)0.0220.289ND 
(I–II vs III–IV)     NA
Tumor number (single vs multiple)0.029NDNS0.046NDNS
Vascular invasion (yes vs no)<0.0013.030 (1.821–5.041)<0.001<0.0012.648 (1.661–4.220)<0.001
Tumor encapsulation (yes vs no)0.006NDNS0.003NDNS
Tumor size (≤5.0 vs >5.0)<0.001NDNS<0.001NDNS
TNM stage (I–II vs III–IV)<0.001NDNS<0.001NDNS
Intratumoral density (low vs high)      
TREM-10.448NDNA0.165NDNA
α-SMA0.008NDNS0.059NDNA
Peritumoral density (low vs high)      
TREM-1<0.0012.225 (1.235–4.009)0.008<0.0012.089 (1.256–3.477)0.005
α-SMA<0.0012.481 (1.260–4.885)0.009<0.0012.066 (1.150–3.712)0.015

Discussion

Substantial evidence shows that activated inflammatory cells in tumor micromilieu engage in a fine-tuned collaborative action with cancer cells and many molecules in these cells mediate this cross-talk.[37] In this study, we observed that TREM-1 was expressed highly in HSC cell lines LX-2, primary activated HSCs, and myofibroblasts, which secreted high concentrations of sTREM-1 to the extracellular compartment after stimulation by LPS and cancer cells. Those results revealed that expression of TREM-1 was not restricted to myeloid cells. Because an inflammatory environment seems to play a critical role in the transdifferentiation of monocytes or macrophages into myofibroblasts or smooth muscle cell-like cells,[38] HSCs and myofibroblasts expressing TREM-1 may phenotypically and functionally resemble monocytes or macrophage-derived smooth muscle cell-like cells. To our knowledge, this is the first study into the expression of TREM-1 locating on HSCs. Although the precise role of TREM-1 in HCC is largely uncertain, our in vitro observations found that TREM-1 showed a protumoral phenotype and accordingly enhanced the migratory ability of HCC cells. In addition to HSCs, TREM-1 was expressed in some other cell types (such as cancer cells, hepatic macrophages, and endothelial cells[39]), so it cannot be excluded that TREM-1 signaling in these cells may play an important role in affecting the tumor process. However, in our study, TREM-1 expressing in activated HSCs might, at least in part, exert its biological functions by mediating cancer cells' aggressive behavior. Further investigation is needed to unravel the mechanisms.

Hepatic stellate cells are versatile mesenchymal cells and vital to HCC progression, but the tumor-promoting mechanisms remain unclear.[6] We found that TREM-1 expression in HSCs was responsible for the aggressiveness of cancer cells in HCC based on the following findings. First, TREM-1 can also be expressed in HCC cells which may accordingly provide a cell surface receptor signal affecting intracellular functions. Second, exposure of LPS and cancer cells favored release of sTREM-1 from activated HSCs, suggesting cancer cells per se were probably triggers of inflammatory response, together with activated HSCs producing an inflammatory cytokine milieu and subsequently activation of TREM-1. Third, TREM-1 had a synergistic reaction on the malignant effects in both HepG2 and Hep3B cell lines, revealing that the effect of TREM-1 is not limited to a specific HCC cell line. Increased secretion of sTREM-1 after activated HSCs were co-cultured with normal hepatocytes may be implicated in the proliferation of hepatocytes by interaction with HSCs through cell adhesion molecules or cytokines.[40] This phenomenon possibly suggests the early phase of carcinogensis, in that changes have already occurred to hepatocytes because of the effects of activated HSCs. Fourth, we directly observed enhanced migration in HCC cells in the presence of activated HSCs/CAMFs-CM or agonist of TREM-1, and an inhibited effect by blockade of TREM-1 expression, strongly indicating a potential mechanism in mediating cancer cell invasive properties by TREM-1. A recent report also supported this role of TREM-1 to promote cancer cells' invasiveness.[27] Thus, it can not be excluded that the intracellular activity of TREM-1 acts as a functional gene in activated HSCs. In fact, there was positive relation between α-SMA (a biomarker of activated HSCs) and TREM-1 by correlation analysis of immunohistochemistry. In further analyses, our clinical data indicated that peritumoral TREM-1 showed powerful prediction ability in high-risk patients with unfavorable prognosis in HCC after resection. Compared with the low subgroup, the patients with high peritumoral TREM-1 expression had higher recurrence rate and reduced survival times in HCC after hepatectomy. Our study also showed that increased peritumoral TREM-1 expression was closely related to adverse HCC characteristics such as vascular invasion, tumor size, and stages of HCC. Therefore, our findings highlighted that close monitoring might be of clinical importance for patients with high peritumoral TREM-1 levels. Generally, the time and types of postoperative HCC recurrence can determine the clinical outcome of most individuals. If 24 months is set as the time limit, there are two distinct types of HCC recurrence, early (≤24 months) and late recurrence (>24 months). The former is identified as a true metastasis caused by dissemination of cancer cells and the latter represents a new primary lesion due to de novo hepatocarcinogenesis.[36] In this study, patients with high peritumoral TREM-1 were more likely to suffer from tumor early recurrences (P < 0.001 for univariate and P = 0.006 for multivariate) rather than late recurrences (P = 0.094 for univariate), revealing TREM-1 possibly promoting dissemination of cancer cells through inflammation-associated mechanisms. In this regard, TREM-1 seems to be a useful predictive molecular for triaging at-risk patients with early recurrence and metastasis of HCC following resection.

Soluble TREM-1 is considered as a soluble counterpart of TREM-1 and has the same extracellular domain as TREM-1.[31, 41] Although the origin and function of sTREM is controversial, it is thought to negatively regulate TREM receptor signaling.[23] To demonstrate direct involvement of sTREM-1 in HCC patients in cancer procession, we measured sTREM-1 in plasma of patients with HCC, which was significantly increased compared with those with benign disease. Decreased postoperative levels of sTREM-1 further indicated that it was a cancer-associated inflammatory mediator. This finding may reveal the complex systemic immune status of individuals with HCC. On one side, some factors (e.g. TREM-1) related to the immune response are thought of as an amplifier of the immune response; on the other side, pro-inflammatory effects can be downregulated by some anti-inflammatory mediators (at least sTREM-1) through competing with ligand(s) to block signaling pathways.[31] A dysregulation of the host immune status may be important in the progression of HCC resulting from continuous inflammation accumulation. In addition to the local upregulation of sTREM-1, release from blood is probably a potential route for its augmentation in tumor. Although sTREM-1 was also detected in lung cancer,[27] its functional properties remain to be defined. Our in vitro observations showed that, in addition to monocytes, activated HSCs were one of the sources of sTREM-1, providing novel evidence that sTREM-1 can perform its functional role in different inflammatory cell populations. Moreover, in line with previous reports,[42, 43] we found that release of IL-6 and TNF-α was significantly enhanced following engagement of sTREM-1, which was confirmed as an anti-inflammatory cytokine.[41] In the tumor milieu, most tumor cell migration in metastasis is dynamic, and such sequential cytokine production can influence tumor cells' ability to dissemine or invade.[44, 45] Furthermore, synergistic interactions of cytokine networks composed of anti-inflammatory (e.g. IL-10 and sTREM-1) and proinflammatory cytokines (e.g. IL-6 and TNF-α) could determine tumor progression of the host with cancer,[46] thus, these kinetics of cytokine production may reflect their potential immunoregulatory properties directing tumor growth to some extent. We thought, as potent tumor-promoting cytokines,[47] IL-6 and TNF-α may serve as critical homeostatic factors in an inflammatory “context” to correlate with anti-inflammatory mediators for keeping balance in cancer milieus to interfere with the spectrum of tumor development. However, given the complexity of cancerization, further mechanisms need to be elucidated.

We found TREM-1 positive cells accumulated in peritumoral liver tissue more than the intratumoral area (Table 1). This difference may be involved in its role in hepatocarcinogenesis. Furthermore, TREM-1 in the peritumoral liver tissue, but not in tumor tissue, can predict the outcome of patients with HCC, which probably depends on distinct compositions and functional properties in peritumoral and intratumoral microenvironments in the progression of HCC. Alhough TREM-1-induced protumoral mechanisms have not been clearly shown, we surmized that TREM-1, infiltrated abundantly into the peritumoral liver tissue, possibly acted as a stimulator in HSC by providing a fertilized soil for accelerating the formation of hepatic metastasis.

In this study, most tissue samples were obtained from patients infected with hepatitis B. Accordingly, we conjecture that the hepatitis B virus might also take a role in facilitating TREM-1 expression of HSCs in some way. However, the behavior of TREM-1 from other etiologies, such as hepatitis C virus or alcohol-related HCC, remains to be determined. In addition, it is unclear whether sTREM-1 released by HSCs has an effect on the chemoattraction of hepatocytes or cancer cells. Extensive studies are needed to confirm these mechanisms.

In summary, we have shown for the first time that peritumoral TREM-1 in HCC was a significant independent prognosticator for worse clinical outcome after resection. Expression of TREM-1 in HSCs promoted aggressive behavior of HCC cells. TREM-1 provides a rational biomarker and novel anti-inflammatory therapy target to make a judgment on prognosis and therapeutic effect after hepatectomy in HCC.

Acknowledgments

This work was supported by grants from the National Key Sci-Tech Special Project of China (Grant Nos. 2008ZX10002-018 and 2008ZX10002-019), the National Natural Science Foundation of China (Grant Nos. 81071707 and 81071995; key program no. 81030038), the Doctoral Fund of the Ministry of Education of China (Grant No. 200802460019), and the State Key Laboratory Open Project of Oncogenes and Related Genes (Grant No. 90-09-03).

Disclosure Statement

The authors have no conflicts of interest.

Ancillary