Introduction

Primary liver cancer is one of the most common malignant tumors in China1,2,3. The high postoperative recurrence rate remains the main reason for poor prognosis in patients with hepatocellular carcinoma (HCC), which seriously reduces treatment efficacy. The lung is the most common site of metastasis and extrahepatic recurrence of HCC after surgery. Reports have shown that postoperative lung recurrence after hepatectomy accounts for 39.5–53.8% of extrahepatic metastasis and seriously compromises prognosis4,5. Therefore, investigating potential predictive biomarkers involved in lung recurrence after hepatectomy is crucial for early detection of HCC recurrence and for development of therapeutic strategies to prevent it.

Lysyl oxidase (LOX) belongs to the lysyl oxidase family. The most important role of LOX-family enzymes is to covalently cross-link collagen and elastin in order to maintain the normal structure and function of the extracellular matrix6. In recent years, studies have shown that LOX is closely related to malignant transformation, tumor cell migration, adhesion, invasion, and the formation of the tumor microenvironment before metastasis. In primary tumors, Systemic secretion of LOX leads to its accumulation in the lung, where it may act on extracellular matrix proteins to establish a permissive niche for infiltrating cancer cells7. Whether LOX plays a role in HCC, particularly in metastasis or recurrence, is unclear.

In this study, we conducted cDNA microarray analysis to identify genes differentially expressed between HCC patients with and without lung recurrence following hepatectomy. LOX emerged as a gene of interest due to its strong association with postoperative lung recurrence. We subsequently investigated the relationship between LOX expression, circulating tumor cell (CTC) count in peripheral blood, and microvessel density (MVD) in tumor tissues. Finally, we evaluated the predictive potential of serum LOX levels for lung recurrence after hepatectomy in two independent patient cohorts from our institution.

Methods

Patient samples

This study enrolled 128 HCC patients who were treated with curative resection in the Department of Hepatobiliary Surgery at the Affiliated Cancer Hospital of Guangxi Medical University in China from September 2016 to September 2019. The Ethics Committee of the Affiliated Cancer Hospital of Guangxi Medical University has approved the study protocol (Grant No.: LW2020077). In accordance with the Declaration of Helsinki, all subjects provided written informed consent for their anonymized medical data to be analyzed and published for research purposes. The treatment guidelines and diagnostic criteria of primary liver cancer are based on the Chinese guidelines for the diagnosis and treatment of primary liver cancer8. For the training cohort, tumor tissues and paired adjacent non-tumor tissues from the 128 HCC patients were used for quantitative real time PCR (qRT-PCR) and immunohistochemical analysis. Their sera, along with sera from 30 healthy outpatient controls, were analyzed for LOX levels by enzyme-linked immunosorbent assay (ELISA). The healthy controls were age- and sex-matched to the HCC patient cohort. In addition, 14 patients with lung recurrence after hepatectomy and 23 patients without lung recurrence were selected from this cohort for analysis of differential gene expression.

Isolation of CTCs by CanPatrol system and tricolor RNA in situ hybridization (RNA-ISH) assay

Blood sample collection procedure and Isolation of CTCs with the CanPatrol system and Tri-color RNA-ISH assay was described in our previous study9. Briefly, red blood cell lysis buffer was used to remove erythrocytes, and the cells were resuspended in phosphate-buffered saline (PBS) with 4% formaldehyde (Sigma, St. Louis, MO, USA) for 5 min. RNA-ISH was used to detect sequences encoding CD45 (leukocyte biomarker), EpCAM and CK8/18/19 (epithelial biomarkers), and vimentin and twist (mesenchymal biomarkers). The assay was performed in a 24-well plate (Corning, NY, USA), and adherent cells were treated with a protease (Qiagen, Hilden, Germany) before hybridization to capture a specific RNA probe. Nuclei were stained with 4’,6-diamidino-2-phenylindole (DAPI), and the cells were analyzed using a fluorescence microscope.

RNA extraction and reverse transcription of HCC and adjacent tissues

Total RNA Kit (R6834-02, Omega Bio-tek, Norcross, USA) was used to extract total RNAs from liver cancer tissues and adjacent tissues. Total RNAs were stored at-80 °C. Reverse transcription was performed using a reverse transcription Kit (RR047a, Takara Bio, Japan). The cDNAs were stored at -20 °C.

Detection of LOX expression by RT-PCR

RT-PCR to detect LOX expression was performed using a commercial kit (RR820a, Takara Bio), and the reaction system was configured according to the manufacturer’s instructions. The relative expression of the target gene and an internal control gene was compared using the 2−ΔΔCT method. The following primers were used: β-Actin (internal reference) forward primer, CTGGAACGGTGAAGGTGACA; reverse primer, CGGCCACATTGTGAACTTTG; LOX forward primer, TTCTTACCCAGCCGACCAAGATA; and reverse primer, GTGTTGGCATCAAGCAGGTCA.

Detection of LOX and CD34 expression by immunohistochemistry

LOX and CD34 proteins were detected in 128 samples by immunohistochemistry kit (Pv-6000, Zhongshan Jinqiao, China). CD34 antibody was purchased from Zhongshan Jinqiao; LOX antibody (ab174316, abcam, UK) was used at a dilution of 1:300. Operate according to the instructions. Immunostained tissue was observed under an optical microscope. The percentage of LOX positive cells was divided into 4 grades (percentage scores): <5% (0); 5–25% (1); 26–50% (2); 51–75% (3); >75% (4). The intensity of staining was divided into 4 grades (intensity scores): no staining (0), weak staining (1), moderate staining (2) and strong staining (3). Therefore, the staining positivity was determined by the formula (overall score = percentage score × intensity score): overall score 0: negative (-); overall score: 1–3 (+); overall score: 4–8 (++); overall score: 9–12 (+++).

Cell lines

Hepatoma cell lines SK-Hep 1, Hep G2, LM3 and SMMC-7721 were purchased from the Cell Bank in the Chinese Academy of Medical Sciences (Shanghai, China) keeping it for our laboratory. They were Included into 25 cm2 culture flask, DMEM (Gibco, USA) high glucose medium containing 10% fetal bovine serum. Cells in logarithmic growth phase were used for subsequent experiments.

LOX knockdown in SK-Hep1 cells and LOX overexpression in SMMC-7721 cells

The expression of LOX in SK-Hep1, Hep G2, LM3, and SMMC-7721 were detected by western blot and RT-PCR. We screened a low LOX expression cell (SK-Hep1) and a high LOX expression cell (SMMC-7721). LOX overexpression plasmid and LOX lentiviral vector were constructed in Shanghai Jikai Gene Technology Co. According to the instructions, two lentiviruses were transfected into SK-Hep1 cells, the transfected cells were screened by puromycin, stable transfected cell lines were obtained. They were divided into sh-1 group, sh-2 group, mock control group. SK-Hep1 cells were set as non-treated control group (NC group). According to the instructions of Invitrogen Lipofectamine 3000 (Invitrogen USA), the plasmid was transfected into SMMC-7721 cells, pBABE group, mock control group, SMMC-7721 cells were set as non-treated control group (NC group). The transfection efficiency was detected by WB.

Cell proliferation analysis

We used cell counting kit-8 (CCK-8) (dojindo, Japan) to compare the proliferation of SK- Hep 1 cells. Sh-1 group, sh-2 group, mock control group and NC group cells were seeded into 96 well plates, and 1 × 104 cells were added into each well. 110 µl CCK-8 mixture (including 10 µl CCK-8 reagent and 100 µl DMEM medium) was added to each well at 0 h, 24 h, 48 h and 72 h, respectively. After incubation for 1.5 h, the optical density (OD) at 450 nm was measured by microplate. Repeat the experiment 3 times.

Cell migration and invasion assays

Before cell invasion test, the cells were coated with Matrigel in the following steps: 50 mg/L Matrigel was diluted in 1:8 and incubated at 37 °C for 1 h for coating. Sh-1 group, sh-2 group, mock control group, NC group, pbabe group, mock control group, NC group cells were collected. The cells in each group were washed with pure DMEM medium, and the cell concentration was adjusted. Then 1 × 105 cells/100 µl were inoculated in the upper chamber, and 600 µl of complete medium containing 10% serum were added into the lower chamber as chemotactic agent. After 48 h of culture at 37 °C in an atmosphere of 5% carbon dioxide, Take out the cotton swab and wipe off the cells without membrane, fixed with methanol for 30 min, stained with 0.1% crystal violet for 20 min, and washed with PBS 3 times. Under the microscope, 5 fields were selected to analyze the number of cell invasion.

Matrigel was not needed to coat the cells in cell migration test, the cells concentration was adjusted to 5 × 104/100 µl, and cultures were incubated for 18 h. The rest was the same as invasion test.

Detection of LOX antibodies in peripheral patients’ serum using ELISA

Fasting venous blood samples (2 mL) from patients were centrifuged, and frozen at -80 °C. A total of 270 individuals were included for serum LOX analysis, including 30 healthy controls. They were randomly assigned into a training cohort (n = 158) and a validation cohort (n = 112) using a computer-generated random number table, following an approximately 6:4 ratio to ensure robust model development and validation. The 112 patients in the validation cohort met the same inclusion criteria as those in the training cohort: pathologically confirmed HCC, curative-intent hepatectomy, available preoperative serum samples, and complete clinical follow-up. No significant differences were observed in key clinical characteristics of patients between the two cohorts. A detailed comparison of baseline variables is provided in Supplementary Table 1. Then use an ELISA kit against human LOX (catalog no. ml060424, Shanghai Enzyme Linked Biological, China). The OD values of each well were measured using enzyme-labelled methods. Drawing ROC curve, according to ROC analysis, the expression of LOX antibody in serum between training cohort and validation cohort was analyzed and compared.

Statistical analysis

SPSS 23.0 and graphpad prism 8.0 were used to analyze the data. Univariate and multivariate Cox regression was used to identify factors associated with recurrence of HCC. Rates of survival and recurrence were calculated using the Kaplan Meier method and compared between groups using the log rank test. When P < 0.05, the difference was considered statistically significant.

Results

Identification of genes related to lung recurrence after hepatectomy in HCC

To search for potential biomarkers of lung recurrence after hepatectomy of HCC, we compared the gene expression profiles 14 patients with lung recurrence and 23 without lung recurrence after curative resection, We identified a total of 860 differential genes (Fig. 1A). A subset of common differentially expressed genes was selected by initial filtering at P < 0.05, followed by filtering based on a difference of ≥ 2-fold in expression level. Using these selection criteria, we found 452 upregulated genes (Fold change ≥ 2) and 408 downregulated genes (Fold change ≤ 0.5) in HCC involving lung recurrence after hepatectomy (Fig. 1B). The top 40 differentially expressed genes (20 upregulated and 20 downregulated) for which expression levels were > 3-fold are shown. Among them, LOX was > 4-fold which ranked first (Fig. 1C). The LOX standardized fluorescence of differed significantly between patients with lung recurrence and patients without lung recurrence after hepatectomy (Fig. 1D).

The GO enrichment analysis suggests that LOX is mainly involved in 8 major biological pathways “cancer-related”, including wound healing, response to steroid hormone, response to decreased oxygen levels, vascular development, extracellular matrix organization, endothelial cell proliferation, lung development, and positive regulation of epithelial migration (Table 1).

Fig. 1
figure 1

Identification of biomarker candidate gene LOX in HCCs with or without lung recurrence after hepatectomy. (A) Gene expression profile microarray (cDNA microarray) was used to screen for genes differentially expressed between patients with or without lung recurrence after hepatectomy. The level of upregulation and downregulation are represented by the intensity of the red or blue colors, respectively. (B) The Cluster Profiler package was used to analyze the differentially expressed genes; 452 upregulated genes and 408 downregulated genes were found. (C) Forty differentially expressed genes (20 upregulated and 20 downregulated) for which expression levels were > 3-fold are shown; among them, LOX ranked the forth. (D) Standardized fluorescence of differed significantly (> 4-fold) between 37 HCC patients with or without lung recurrence after hepatectomy.

Full size image
Table 1 Biological processes enriched among the genes differentially expressed between HCC patients with or without lung recurrence after hepatectomy.
Full size table

LOX is upregulated in HCC involving lung recurrence after hepatectomy

LOX expression was examined by RT-PCR and immunohistochemistry analysis in 46 patients with lung recurrence and 82 patients without lung recurrence after hepatectomy. Most of the 128 HCC patients (101,78.9%) showed higher expression of LOX mRNA in the tumor tissues than in the adjacent normal tissues (Fig. 2A). Levels of LOX mRNA were significantly higher in HCC patients with lung recurrence after hepatectomy than in those without lung recurrence (Fig. 2B). Immunohistochemistry staining confirmed higher LOX levels (≥ 2-fold) in HCC with lung recurrence after hepatectomy (Fig. 2C).

Univariate Logistic analysis showed that: HBsAg, tumor size, Edmondson grade, node number, tumor capsule insufficiency, microvessel density (MVD), microvascular invasion (MVI), BCLC stage and high expression of LOX are 9 major factors that lead to postoperative HCC. Related risk factors for lung recurrence after hepatectomy (all P < 0.05). Incorporating the above factors into the multivariate logistic analysis: Only high expression of LOX is an independent risk factor for postoperative lung recurrence (Table 2). Univariate Cox analysis showed that: HBsAg, tumor size, Edmondson grade, node number, tumor capsule insufficiency, microvessel density (MVD), microvascular invasion (MVI), BCLC stage and high LOX expression are the top 10 risk factors of post-operative HCC recurrence. Incorporating the above factors into the multivariate Cox analysis showed that: age, MVI and high LOX expression are independent risk factors of post-operative HCC recurrence (Table 3).

Fig. 2
figure 2

Relationship between LOX expression and lung recurrence after hepatectomy in human HCC tissues. (A) RT-PCR was used to detect the expression of LOX in cancer and adjacent tissues. (B) Expression of LOX mRNA by RT-PCR in patients with or without lung recurrence after hepatectomy. (C) Representative immunohistochemistry staining showed stronger LOX expression in human HCC than in the adjacent normal liver tissue. Magnification: 40×. High expression is shown in red and low expression in blue.

Full size image
Table 2 Comparison of preoperative characteristics between HCC patients with or without postoperative lung recurrence after hepatectomy (n = 128).
Full size table

The degree of liver cirrhosis was classified into two categories based on imaging, intraoperative findings, and histopathological examination: none/mild and moderate-to-severe. Resection margin status was stratified using a 1 cm cutoff: < 1 cm was considered a narrow margin, and ≥ 1 cm a wide margin.

Table 3 Univariate and multivariate Cox regression to identify preoperative patient characteristics associated with risk of postoperative lung recurrence in HCC (n = 128).
Full size table

High expression of LOX was associated with greater metastatic potential and risk of post-operative recurrence

Using the CanPatrol™ CTC-enrichment technique, we detected abundant CTCs undergoing the Epithelial-mesenchymal transition (EMT) in patients’ blood samples. Analysis of CTCs by RNA-ISH revealed 3 subpopulations: epithelial CTCs (E-CTCs), epithelial/mesenchymal hybrid CTCs (E/M-CTCs), and mesenchymal CTCs (M-CTCs) (Fig. 3A). In total, 118 of 128 (90.2%) HCC patients displayed CTC-positivity, comprising 89.1% (73/82) of patients without lung recurrence after hepatectomy and 97.8% (45/46) with it (Fig. 3B). We found that the rate of CTC-positivity was significantly higher among patients with strong LOX immunostaining (97.3%) than among those with weak immunostaining (85.2%, P < 0.05) among patients with lung lung recurrence after hepatectomy, patients without lung recurrence after hepatectomy and healthy controls (Fig. 3C). The same was observed for median counts of total CTCs, M-CTCs, and E/M-CTCs (Fig. 3D).

In addition, we examined whether LOX expression was correlated with microvessel density (MVD) in our HCC patients. Specific staining of capillary-like vessels using anti-CD34 antibody was observed in all tumor specimens. Examples of high and low MVD images are shown in Fig. 3E. High MVD occurred significantly more often among patients with high LOX expression than among those with low LOX expression (P < 0.05; Fig. 3F).

Fig. 3
figure 3

Relationship of LOX expression with CTC counts and MVD. (A) Representative images of three types of CTCs isolated from patients with HCC based on RNA-ISH staining of epithelial markers (red fluorescence) and mesenchymal markers (green fluorescence). (B) Total CTC counts (top) and percentages of each CTC type (bottom) in patients with or without lung recurrence after hepatectomy. (C) Total CTC-positivity rates in HCC patients showing high or low LOX expression. (D) Total CTC count (1) and percentage of M-CTCs (2), E/M-CTCs (3), and E-CTCs (4) among patients with lung recurrence after hepatectomy, patients without lung recurrence after hepatectomy and healthy controls.**P < 0.05. (E) CD34-positivity rates in HCC tissues was significantly higher in patients with high MVD than in those with low MVD. Magnification: 40× (left panel), 100× (right panel). (F) The expression of MVD in patients with high LOX expression was significantly higher than that in patients with low LOX expression.

Full size image

Effects of LOX on proliferation, migration, and invasion of HCC cells

We next analyzed the effect of LOX on proliferation and invasive ability of the HCC cell lines SMMC7721 and SK-Hep1. Both shRNA-LOX knockdown clones sh-1 and sh-2 showed lower LOX expression than the NC and Mock control cells (Fig. 4A).

LOX knockdown by the two shRNAs inhibited cell proliferation in SK-Hep 1 cells (Fig. 4B), and it led to significantly lower migration and invasion than in NC and Mock control cells. In contrast, LOX over-expression in SMMC-7721 cells was associated with significantly increased migration and invasion compared with that in NC and Mock controls (Fig. 4C and D).

Fig. 4
figure 4

Effects of LOX overexpression or knockdown on cell proliferation, migration, and invasion. (A) Western blot analysis of LOX in each type of cells (after cropping). Original blots/gels are presented in Supplementary Fig. 1. (B) CCK8 assay to determine the effect of LOX knockdown on proliferation of SK-Hep1 cells. Magnification, 10×. (C, D) Transwell invasion assays to determine the effect of LOX overexpression and knockdown on invasion (C) and migration (D) of HCC cells. ** P < 0.05, *** P < 0.001.

Full size image

Serum LOX as a potential predictor of lung recurrence after hepatectomy

We next investigated whether LOX is a secretory protein that could be a serum marker for HCC. We also evaluated the potential of serum LOX levels for predicting post-surgery lung recurrence after hepatectomy. We included two cohorts of patients and measured their serum LOX values using ELISA. In the training cohort, the average serum LOX level was significantly higher in HCC patients with lung recurrence after hepatectomy than in those without it (Fig. 5A). The ROC curve for serum LOX levels showed an AUC of 0.736 for predicting lung recurrence after hepatectomy. The optimal cutoff value for serum LOX level based on the Youden index was 94.86 ng/mL, giving sensitivity of 65.2% and specificity of 75.6% (Fig. 5C).

Similarly, in the validation group, the average serum LOX level was significantly higher in HCC patients with lung recurrence after hepatectomy than in those without it (Fig. 5B), and the ROC curve had an AUC of 0.704. Using the same cutoff value of LOX as in the training cohort, LOX showed 75.6% sensitivity and 61.6% specificity for predicting lung recurrence after hepatectomy (Fig. 5D).

Fig. 5
figure 5

Ability of serum LOX levels to predict lung recurrence after hepatectomy in HCC patients after hepatectomy. (A, C) Results for the training cohort. (B, D) Results for the validation cohort. **P < 0.05.

Full size image

Discussion

Several studies have found that LOX expression is upregulated in various type of cancers, including breast cancer10 gastric cancer11 colorectal cancer12 esophageal squamous cell carcinoma13 as well as head and neck tumors14. These studies suggest that LOX may be a marker for tumor prognosis. Some reports found that high levels of LOX were detected in liver cancer tissues compared with adjacent normal liver tissues15,16. Some reports have shown an association between LOX and angiogenesis markers such as vascular endothelial growth factor (VEGF)15,17,18. Venning FA et al.‘s study confirmed that LOX overexpression is related to EMT or metastasis of hepatocellular carcinoma19. In addition, compared with normal tissues, the level of LOX-PP expressed in HCC tissue is lower20. The expression of LOX-PP by adenovirus in hepatocellular carcinoma cells can enhance cell apoptosis and inhibit proliferation, migration and invasion through mitogen-activated protein kinase (MAPK) pathway21. These results together indicate the carcinogenic effect of LOX in HCC. In a microarray study, we found that LOX is overexpressed in HCC patients compared with controls. We also found that LOX expression was higher in patients with lung recurrence after hepatectomy than in those without lung recurrence, and that serum LOX might be a predictive marker for postoperative lung recurrence.

Gene expression profile analysis of HCCs showed that LOX was significantly upregulated in patients with lung recurrence after hepatectomy and was involved in eight biological processes. Some of these biological processes are related to recurrence, such as vascular development, extracellular matrix organization, endothelial cell proliferation, lung development and positive regulation of epithelium migration. Thus, LOX expression may be involved in compensatory mechanisms in HCC, helping drive metastasis and thereby increasing risk of postoperative lung recurrence.

Studies have suggested that extrahepatic metastasis is facilitated by the flow of tumor blood to distant sites, while intrahepatic metastasis occurs due to systemic rehoming of CTCs to the remnant liver22,23. Umezaki et al. mainly focus on intrahepatic recurrence, they found that high LOX expression was associated with EMT markers and predicted early recurrence and poor survival14. This is consistent with our study, LOX may be the key molecule for HCC metastasis or recurrence. Unlike intrahepatic recurrence, which some of them likely arises through de novo primary HCC, all extrahepatic metastasis recurrence may represent ‘CTC-induced recurrence’. If so, the EMT of CTCs likely underlies HCC recurrence, so it may be possible to evaluate the risk of recurrence and metastasis by counting CTCs in blood. Indeed, a previous study from our group showed that high CTC count before resection was significantly associated with early recurrence and lung metastasis recurrence9,24. In the present study, our results revealed that LOX upregulation was associated with higher counts of total CTCs as well as M- or E/M-CTCs. In addition, our in vitro studies indicate that LOX promotes migration, invasion, and proliferation in HCC cells, which points to possible biological functions of LOX in metastasis.

CTCs need to adapt to the microenvironment of target organs in order to survive and proliferate into recurrent tumors. One study showed that tissue stiffness modulated by LOX regulates postnatal lung development through LRP5-Tie2 signaling25. Bioinformatic prediction in this study also suggested that LOX is involved in lung development. Given the biological characteristics of LOX, we speculate that over-expression of LOX in HCC not only regulates CTCs and MVD, but also promotes the survival and growth of cancer cells in the lung. Further research should investigate the molecular mechanisms of LOX-regulated lung colonization by CTCs.

Tumor recurrence and metastasis require angiogenesis, a complex, multi-stage process mediated by many cytokines. High angiogenesis is a prerequisite for HCC26. A recent study has shown that LOX can indirectly activate protein kinase B to increase the expression of vascular endothelial growth factor, thereby promoting angiogenesis27. In our study, expression of LOX was significantly higher among patients with high MVD than among those with low MVD group. This implies a role for LOX in tumor angiogenesis.

About 30% of cases of HCC recurrence after resection are extrahepatic, and most involve lung metastasis recurrence28,29. Numerous studies have shown that modeling can predict tumor metastasis17,30,31,32 unfortunately, there are still no effective methods to predict lung metastasis. Since LOX is secreted into the extracellular interstice via autocrine and paracrine mechanisms10 we examined whether LOX might be a serum marker of lung metastasis recurrence in HCC after hepatectomy. Indeed, we found that serum LOX levels predicted lung recurrence after hepatectomy with relatively high positive predictive values in both the training and validation cohorts. Serum LOX levels can be measured non-invasively and conveniently, strengthening its potential as a biomarker.

Our study presents several limitations. First, our sample was small and came from a single center. A large, multi-site study is needed to assess the general validity of the results. Second, we did not directly confirm the effects of LOX overexpression or knockdown on the metastatic potential of cancer cells in vivo. Future studies will involve orthotopic and metastatic animal models to validate the contribution of LOX to lung metastasis of HCC. Nevertheless, previous studies such as that by Tse et al.33 have indirectly shown that HBx remodels the extracellular matrix through the HIF-1α/LOX pathway to promote HCC metastasis in vivo, which supports our mechanistic hypothesis. Despite these limitations, our study provides evidence that high expression of LOX can promote recurrence by promoting the formation of CTCs and by providing a transfer channel for MVD. LOX can be considered a candidate biomarker and therapeutic target for HCC recurrence, and it may help promotes the proliferation, migration, and invasion of HCC cells. Serum LOX level may be a non-invasive predictor of lung recurrence after hepatectomy in HCC.