Epithelial-mesenchymal transition phenotypes of circulating tumor cells correlate with the clinical stages and cancer metastasis in hepatocellular carcinoma patients

Abstract

BACKGROUND:

Epithelial-mesenchymal transition (EMT) plays a crucial role in circulating tumor cells (CTCs) dissemination and cancer metastasis.

OBJECTIVE:

To investigate the EMT phenotypes of CTCs in hepatocellular carcinoma (HCC) patients and the clinical utility in the early diagnosis of HCC metastasis and progression.

METHODS:

We retrospectively analyzed the count and EMT classification of CTCs detected by the CanPatrol ${}^{\@setsize{\tiny}{6pt}{\vpt}{\@vpt}\textregistered}$ platform in 195 HCC patients. The clinical relevance with other pathological features was statistically evaluated.

RESULTS:

CTCs were detected in 95% of the 195 HCC patients with a range of 0–86 CTCs. Total CTCs numbers were correlated with BCLC stages, metastasis and serum AFP levels. The AUC of the ROC curve was 0.861 (95% CI: 0.782–0.940) in discriminating metastatic HCC patients with non-metastatic patients. Epithelial, hybrid and mesenchymal CTCs were found in about 53%, 83% and 57% patients, respectively. The proportion of hybrid and mesenchymal CTCs was associated with ages, BCLC stages, metastasis and AFP levels. Besides, recurrent HCC patients presented higher CTCs count and increased hybrid and mesenchymal CTCs.

CONCLUSIONS:

CTCs count and EMT classification are correlated with clinical stages and metastasis of HCC, suggesting that they may be potential markers for the early diagnosis of HCC metastasis and progression.

Keywords

Circulating tumor cells hepatocellular carcinoma epithelial-mesenchymal transition cancer metastasis

1. Introduction

Cancer-related mortality of hepatocellular carcinoma (HCC) ranks second in the world [1] and third in China [2]. Metastasis and recurrence account for most of the lethality, thus posing big challenges for HCC therapies such as hepatectomy, interventional therapy and liver transplantation. Therefore, early detection of metastasis and prevention of recurrence are important for improving the clinical efficacy and patient prognosis. The diagnosis of cancer metastasis mainly relies on imaging examinations including X-ray, Ultrasound, Computed Tomography and Magnetic Resonance Imaging. However, they are not satisfactory in discovering small metastases and micrometastases due to the technological limitations. And there is no reliable metastasis or relapse-specific serum biomarker up to now. Thus it is urgent to find effectual methods for the early diagnosis of HCC metastasis and monitoring of recurrence.

The detection of circulating tumor cells (CTCs) opens new horizons to address this problem. CTCs are tumor cells that released into the peripheral blood from the primary tumors or metastases. Since hematogenous spread is the dominant route of HCC dissemination [3], CTCs could be an active origin of metastasis and recurrence in HCC patients. Previous studies have shown that the presence of CTCs may indicate metastases or early recurrence of HCC in certain conditions [4, 5]. And a higher CTCs number is demonstrated associated with poor prognosis, higher recurrence risks, lower disease-free survival and overall survival after surgery [6, 7, 8, 9]. In addition, as a new pattern of liquid biopsy, CTCs test exhibits outstanding performance compared with traditional tissue biopsy, including easy sample collection, minimal invasiveness and achievable multi-monitoring. These advantages equip CTCs test to be a promising biomarker for HCC diagnosis and prognosis.

Lots of methods have been developed for CTCs isolation and detection [10], basing on physical properties [11, 12] and immunological principles [13, 14]. However, most immunological based methods target the epithelial markers to capture CTCs, such as epithelial cell adhesion molecule (EpCAM) or cytokeratins (CKs). This may result in the bias and inaccuracy of CTCs detection because of epithelial-mesenchymal transition (EMT) [15]. EMT is a vital process of metastasis because it enables epithelial cells to obtain enhanced migrating and invasive potential. CTCs may go through a dynamic process of EMT and MET (mesenchymal-epithelial transition) during their intravasation, dissemination and extravasation. Thus some CTCs may lose epithelial characteristics with decreased or deleted expression of epithelial markers and acquire a mesenchymal phenotype. Expressions of mesenchymal markers have been proved in CTCs from patients with lung, breast and prostate cancer [16, 17]. Therefore, combining the epithelial and mesenchymal markers can improve the sensitivity and accuracy of CTCs detection.

Canpatrol ${}^{\@setsize{\scriptsize}{8pt}{\viipt}{\@viipt}\textregistered}$ CTCs analysis system [18, 19] is a newly developed method for CTCs isolation and simultaneous classification by targeting both CTCs subpopulations that express epithelial markers and/or mesenchymal markers. This system combines microporous filtration with multiple mRNA in situ hybridization and branched DNA signal amplification technologies to isolate CTCs and classify them into epithelial type (E type), mesenchymal type (M type) and hybrid type (H type). Various fluorescence-labeled mRNA probes are used to target the epithelial markers (such as EpCAM, CK8/18/19, E-cadherin) and mesenchymal markers (such as Vimentin, Twist, AKT2 and Snail). Thus the numbers of total CTCs and each type CTCs are counted. The tumor cell recovery efficiency of this method is over 80% with high detection efficacies in common solid cancers [18]. The clinical significance of this method has been reported in gastric cancer [20] and colorectal cancer [19].

Here we explored the application of the Canpatrol ${}^{\@setsize{\scriptsize}{8pt}{\viipt}{\@viipt}\textregistered}$ CTCs analysis system in the metastasis diagnosis and monitoring of HCC patients. The results of CTCs count and EMT phenotypes detected with this method were retrospectively analyzed, and the correlations between CTCs detection and the clinical pathological characteristics of HCC patients were evaluated.

2. Subjects and methods

2.1 Patient information

A total of 195 HCC patients from Nanfang Hospital, Southern Medical University (Guangzhou, China) during December 1, 2014 to December 31, 2016 were investigated in this retrospective study. Patients were diagnosed with HCC by histopathological examination, including 99 newly diagnosed patients and 96 post-treatment patients.

2.2 Ethical approval

All procedures performed in studies involving human participants were in accordance with the ethical standards of the institutional and/or national research committee and with the 1964 Declaration of Helsinki and its later amendments or comparable ethical standards. All patients provided written informed consent and the research protocol was approved by the Ethics Committee of Nanfang Hospital (Guangzhou, China).

2.3 CTCs test and clinical data

We analyzed the data of CTCs tests and other clinical pathological characteristics of the 195 HCC patients from the Laboratory Information System (LIS) and Jiahe Electronic Medical Record System. CTCs test was conducted by a newly imported Canpatrol ${}^{\@setsize{\scriptsize}{8pt}{\viipt}{\@viipt}\textregistered}$ CTCs analysis platform (SurExam, China). The protocols and mRNA probes used for CTCs classification were performed as previously described [18]. The CTCs count and EMT classification were detected in 5 mL peripheral blood by microporous filtration and multiple mRNA in situ hybridization technology. CTCs were counted in total and by 3 different EMT phenotypes. Other clinical variables include gender, age, cirrhosis, Hepatitis B Surface Antigen (HbsAg), tumor size, BCLC stage and metastasis. The detailed patient information was confirmed and recorded by their clinicians. Besides, the serum levels Alpha-fetoprotein (AFP) and Carcinoembryonic antigen (CEA) were also simultaneously tested by the ADVIA Centaur ${}^{\@setsize{\scriptsize}{8pt}{\viipt}{\@viipt}\textregistered}$ XP analyzer (SIEMENS, Germany). Then we statistically evaluated the significance of CTCs test and the correlation between CTCs test and clinical characteristics.

2.4 Statistical analysis

Statistical analyses were performed using SPSS 13.0 (SPSS Inc., Chicago, IL, USA) and GraphPad Prism 5.0 (GraphPad Software, San Diego, CA, USA). Data were presented as median (Range) or median (InterQuartile Range, IQR) for discontinuous variables, and categorical variables were expressed as numbers and percentages. Correlations between CTCs test and clinical characteristics were evaluated using Chi-square test. Mann – Whitney U test was used to compare the differences between the groups. Receiver operating characteristic (ROC) curves and the area under the ROC curves (AUC) were applied to analyze the efficacy of assisted diagnosis of metastasis. All tests were two tailed, and we considered a $P$ value $<$ 0.05 was statistically significant.

3. Results

3.1 Patient characteristics and CTCs test

The clinical characteristics of the 195 HCC patients were shown in Table 1. We investigated 163 male and 32 female patients with median age of 50 years (Range 25–85 years). Among them, 123 patients had cirrhosis and 146 patients presented positive results of HbsAg. The distributions and percentages of other pathological parameters (tumor size, BCLC stage and metastasis) and serum tumor markers (AFP and CEA) were also described in Table 1.

Table 1
Clinical characteristics and CTCs test of 195 HCC patients

Characteristics	Median	Range	N	Percentage
				(%)
Gender	NA ${}^{\text{a}}$	NA
Male/Female			163/32	84/16
Age	50	25–85
$\leqslant$ 60/ $>$ 60 years			146/49	75/25
Cirrhosis	NA	NA
Yes/No			123/72	63/37
HbsAg	NA	NA
$+$ / $-$			146/49	75/25
Tumor size	5.7	1.6–14.3
$\leqslant$ 5/ $>$ 5 cm			94/101	48/52
BCLC stage	NA	NA
A/B $+$ C			89/106	46/54
Metastasis	NA	NA
Yes/No			79/116	41/59
Serum AFP	55	0.3–97397.0
$\leqslant$ 20/ $>$ 20 ng/mL			68/127	35/65
Serum CEA	2.1	0.12–64.6
$\leqslant$ 5/ $>$ 5 $\mu$ g/L			132/63	68/32
CTCs count	6	0–86
$<$ 3/ $\geqslant$ 3/5 mL			49/146	25/75
E type CTCs	1	0–33
$<$ 1/ $\geqslant$ 1/5 mL			92/103	47/53
H type CTCs	3	0–49
$<$ 1/ $\geqslant$ 1/5 mL			33/162	17/83
M type CTCs	1	0–21
$<$ 1/ $\geqslant$ 1/5 mL			84/111	43/57

${}^{\text{a}}$ : Not applicable.

Table 2

Correlations between CTCs detection and clinical pathological characteristics in pre-treated HCC patients

Pathological parameters	N	CTCs count ${}^{\text{a}}$		$P$	N	CTCs phenotypes ${}^{\text{b}}$		$P$
		$+$	$-$			H $+$ M $\leqslant$ E	H $+$ M $>$ E
Total cases	99	79	20		79	22	57
Gender				0.748				0.365
Male	87	69	18		69	18	51
Female	12	10	2		10	4	6
Age				0.675				0.010 ${}^{*}$
$\leqslant$ 60 years	73	59	14		59	12	47
$>$ 60 years	26	20	6		20	10	10
Cirrhosis				0.631				0.787
Yes	64	52	12		52	15	37
No	35	27	8		27	7	20
HbsAg				0.166				0.341
$+$	76	63	13		63	16	47
$-$	23	16	7		16	6	10
Tumor size				0.885				0.064
$\leqslant$ 5 cm	53	42	11		42	8	34
$>$ 5 cm	46	37	9		37	14	23
BCLC stage				0.029 ${}^{*}$				0.001 ${}^{*}$
A	43	30	13		30	15	15
B $+$ C	56	49	7		49	7	42
Metastasis				0.023 ${}^{*}$				0.021 ${}^{*}$
Yes	42	38	4		38	6	32
No	57	41	16		41	16	25
Serum AFP				0.047 ${}^{*}$				0.002 ${}^{*}$
$\leqslant$ 20 ng/mL	27	18	9		18	10	8
$>$ 20 ng/mL	72	61	11		61	12	49
Serum CEA				0.887				0.075
$\leqslant$ 5 $\mu$ g/L	73	58	15		58	13	45
$>$ 5 $\mu$ g/L	26	21	5		21	9	12

${}^{*}$ Correlation is significant at the 0.05 level (two-tailed). ${}^{\text{a}}$ : CTCs count analysis targeted all of the pre-treated patients ( $n=$ 99). ${}^{\text{b}}$ : CTCs phenotypes analysis targeted the patients with CTCs+ results ( $n=$ 79).

Figure 1.

CTCs EMT types were classified by fluorescence-labeled probes targeting epithelial and mesenchymal markers. (a) Epithelial type CTCs (E type). (b) Hybrid type CTCs (H type). (c) Mesenchymal type CTCs (M type). (d) Leukocyte as negative control.

Figure 2.

Correlations between CTCs count and clinical pathological characteristics in pre-treated HCC patients. (A) Correlation between CTCs count and BCLC stage. (B) Correlation between CTCs count and metastatic status. (C) Correlation between CTCs count and serum levels of AFP. (D) ROC curve for CTCs count to discriminate metastatic HCC patients from non-metastatic patients.

CTCs were detected ( $\geqslant$ 1/5 mL) in 186 out of 195 HCC patients (95%), and the median number was 6 CTCs (ranging from 0–86 CTCs) in 5 mL blood sample. Charging CTCs positive by CTCs count $\geqslant$ 3/5 mL peripheral blood, the positive rate of CTCs was 75% (146/195 cases) in these patients. With the Canpatrol ${}^{\@setsize{\scriptsize}{8pt}{\viipt}{\@viipt}\textregistered}$ CTCs analysis platform, all the isolated CTCs were classified into 3 EMT types (E type, H type and M type) using different labeled mRNA probes. Figure 1 showed the fluorescence images of 3 CTCs types. E type, H type and M type CTCs were found in about 53%, 83% and 57% HCC patients, respectively. And the median number of each type CTCs was 1 (Range 0–33), 3 (Range 0–49) and 1 (Range 0–21) in 5 mL peripheral blood (Table 1).

3.2 Canpatrol CTCs count helps discriminate metastatic HCC patients from non-metastatic patients

Among the 195 HCC patients, 99 patients were newly diagnosed and blood samples for CTCs test were collected before any surgical or radiochemical therapy. We evaluated the CTCs test results in these pre-treated patients, and CTCs were positive ( $\geqslant$ 3 CTCs/5 mL) in 79 out of the 99 patients (79.8%). Correlations between CTCs numbers and clinical pathological characteristics in the pre-treated HCC patients were shown in Table 2 and Fig. 2.

CTCs count was closely associated with BCLC stage ( $P=$ 0.029), metastasis ( $P=$ 0.023) and serum AFP levels ( $P=$ 0.047) in HCC patients (Table 2). The positive rate of CTCs was 69.8% (30/43 cases) in patients with BCLC-A stage HCC and 87.5% (49/56 cases) in patients with BCLC-B/C stage HCC. In non-metastatic patients, the CTCs positive rate was 71.9% (41/57 cases); while in metastatic patients the rate was 90.5% (38/42 cases). Furthermore, the CTCs positive rate was 66.7% (18/27 cases) and 84.7% (61/72 cases) in patients with serum AFP levels $\leqslant$ 20 ng/mL and $>$ 20 ng/mL, respectively. There was no significant correlation between CTCs numbers and other clinical characteristics, including gender, age, cirrhosis, HbsAg, tumor size and serum CEA levels (Table 2).

The median CTCs number in patients with BCLC-B/C stage HCC was 10 cells/5 mL (IQR 6–22.75), higher than that in patients with BCLC-A stage HCC (5 cells/5 mL, IQR 2–10) ( $P=$ 0.002, Fig. 2A). In non-metastatic patients, the CTCs median was 5 cells/5 mL (IQR 2–7); while in metastatic patients the median was 20 cells/5 mL (IQR 10–29.25) ( $P<$ 0.001, Fig. 2B). The CTCs median was 4 cells/5 mL (IQR 1–10) and 9 cells/5 mL (IQR 4.75–22) in patients with serum AFP levels $\leqslant$ 20 ng/mL and $>$ 20 ng/mL, respectively ( $P=$ 0.011, Fig. 2C). Additionally, we simulated the ROC curves to assess the role of CTCs count in discriminating metastatic HCC patients with non-metastatic patients (Fig. 2D). The AUC of the ROC curve was 0.861 (95% CI: 0.782–0.940), and the sensitivity and specificity were 85.7% and 80.7%. These results demonstrated total CTCs number could be an assistant diagnostic marker for cancer metastasis in HCC patients.

3.3 CTCs EMT phenotypes correlate with BCLC stages and metastasis of pre-treated HCC patients

Figure 3.

Correlations between CTCs EMT phenotypes and clinical pathological characteristics in CTCs positive pre-treated HCC patients. (A) Correlation between CTCs classification and age. (B) Correlation between CTCs classification and BCLC stage. (C) Correlation between CTCs classification and metastatic status. (D) Correlation between CTCs classification and serum AFP levels.

Figure 4.

CTCs counts and EMT classification results in 16 HCC patients experienced recurrence. (A) CTCs numbers of different EMT types. (B) Percentages of different EMT types of CTCs. (C) Proportions of CTCs expressing mesenchymal markers (H $+$ M type) were over 50% in all of the recurrent HCC patients.

Subsequently, we analyzed the EMT phenotypes of isolated CTCs in the 79 HCC patients with positive CTCs results ( $\geqslant$ 3/5 mL). CTCs expressing mesenchymal markers (H type and M type CTCs) were thought to have stronger invasive and metastatic ability than E type CTCs. Therefore, the CTCs EMT classification results were organized as H $+$ M $\leqslant$ E type and H $+$ M $>$ E type basing on the proportion of each type CTCs. We analyzed the distributions of CTCs EMT phenotypes and evaluated the correlations between CTCs phenotypes and clinical pathological characteristics (Table 2 and Fig. 3).

CTCs EMT phenotypes were closely associated with age ( $P=$ 0.010), BCLC stage ( $P=$ 0.001), metastasis ( $P=$ 0.021) and serum AFP levels ( $P=$ 0.002) in HCC patients (Table 2). The rate of patients presenting an H $+$ M $>$ E type was 79.7% (47/59 cases) in $\leqslant$ 60 years old patients and 50.0% (10/20 cases) in $>$ 60 years old patients. The rate of H $+$ M $>$ E type in patients with BCLC-A stage HCC was 50.0% (15/30 cases), and in patients with BCLC-B/C stage HCC the rate was 85.7% (42/49 cases). In non-metastatic patients, the rate of H $+$ M $>$ E type was 61.0% (25/41 cases); whereas in metastatic patients the rate was 84.2% (32/38 cases). Moreover, the rate of H $+$ M $>$ E type was 44.4% (8/18 cases) and 80.3% (49/61 cases) in patients with serum AFP levels $\leqslant$ 20 ng/mL and $>$ 20 ng/mL, respectively. There was no significant correlation between CTCs phenotypes and other clinical characteristics, including gender, cirrhosis, HbsAg, tumor size and serum CEA levels (Table 2).

In addition, we compared the proportions of H $+$ M type CTCs/Total CTCs between different groups defined by the levels of clinical parameters. The median proportion of H $+$ M type CTCs/Total CTCs was 0.60 (IQR 0.56–0.80) in $\leqslant$ 60 years old patients, higher than that in $>$ 60 years old patients (0.53, IQR 0.18–0.69) ( $P=$ 0.047, Fig. 3A). Patients with BCLC-B/C stage HCC presented a higher proportion (0.67, IQR 0.50–0.75) of H $+$ M type CTCs/Total CTCs than that in patients with BCLC-A stage HCC (0.40, IQR 0.18–0.57) ( $P=$ 0.002, Fig. 3B). In non-metastatic patients, the median proportion of H $+$ M type CTCs/Total CTCs was 0.60 (IQR 0.37–0.79); whereas in metastatic patients the proportion was 0.70 (IQR 0.58–0.83) ( $P=$ 0.029, Fig. 3C). The median proportion was 0.48 (IQR 0.19–0.71) and 0.70 (IQR 0.58–0.82) in patients with serum AFP levels $\leqslant$ 20 ng/mL and $>$ 20 ng/mL, respectively ( $P=$ 0.006, Fig. 3D). These results showed that a higher proportion of H and M type CTCs was correlated with tumor stage and metastasis in HCC patients, indicating the distribution of CTCs EMT phenotypes could provide information of HCC development.

3.4 CTCs count and EMT phenotypes in 16 recurrent HCC patients

In the 16 patients with recurrent HCC, CTCs numbers were $\geqslant$ 3/5 mL in all of them (100% positive) (Table 3). The median CTCs number was 11.5 cells (IQR 4.75–15.75) in 5 mL blood, ranging from 3 to 86 cells/5 mL (Fig. 4A). The median of E type, H type and M type CTCs was 1 cells (IQR 0–1.75), 6.5 cells (IQR 4.25–11.75) and 2 cells (IQR 0–5.5), respectively. The average rates of 3 types CTCs was 7.3% (IQR 0–14.3%) for E type, 61.2% (IQR 35.0%–90.2%) for H type and 17.4% (IQR 0–61.9%) for M type (Fig. 4B). Moreover, the percentage of patients with H $+$ M $>$ E type was 100%, which meant the proportions of H $+$ M CTCs/Total CTCs were all over 50% (Median 92.8%, IQR 85.7%–100%) in recurrent HCC patients, ranging from 58% to 100% (Fig. 4C and Table 3). These results further supported that the CTCs count and EMT phenotypes were associated with metastasis and progression of HCC.

4. Discussion

EMT is dynamic and reversible, thus the CTCs may consist of heterogeneous subgroups expressing epithelial markers, mesenchymal markers, or simultaneously expressing two types of markers. Studies have proved the hybrid and mesenchymal phenotype of CTCs by the subsequent identifications. Among CTCs isolated by cell size-based methods, the proportion of Vimentin $+$ CTCs could be as high as 80%–100% [21]. Armstrong et al. [17] reported that in metastatic prostate cancer patients, expressions of epithelial markers (such as EpCAM, E-cadherin and CKs) along with mesenchymal markers (such as Vimentin, O-cadherin and N-cadherin) were detected in more than 80% of the CTCs. And metastatic breast cancer patients also had more than 75% CTCs population with co-expression of two types of markers. Lecharpentier et al. [16] also found that almost all the CTCs isolated from lung cancer patients expressed EMT markers and some patients had only mesenchymal CTCs without epithelial CTCs. Accordingly, simultaneous detection of the CTCs subpopulations with a hybrid phenotype and mesenchymal phenotype can make up for the shortcomings of existing E markers-based methods, thus improving the capture rate and sensitivity of CTCs detection.

Table 3

CTCs detection results in 16 hepatocellular carcinoma patients experienced recurrence

Patient	Total CTCs/	CTCs phenotypes/5 mL			H $+$ M type/
number	5 mL	E type	H type	M type	Total CTCs
1	4	0	4	0	100%
2	11	0	11	0	100%
3	14	2	12	0	86%
4	24	2	22	0	92%
5	7	1	5	1	86%
6	3	1	0	2	67%
7	3	0	1	2	100%
8	3	0	1	2	100%
9	12	5	5	2	58%
10	11	1	8	2	91%
11	12	1	7	4	92%
12	25	1	16	8	96%
13	86	31	49	6	64%
14	15	0	6	9	100%
15	16	1	5	10	94%
16	7	0	7	0	100%

The CanPatrol ${}^{\@setsize{\scriptsize}{8pt}{\viipt}{\@viipt}\textregistered}$ CTCs analysis platform involved in this research is an integrated CTCs isolation and classification system. With the use of different combinations of fluorescence-labeled mRNA probes, the isolated CTCs were classified into heterogenous EMT phenotypes, including E type, H type and M type. CTCs were detected in 95% of the observed 195 HCC patients with a median number of 6 CTCs in 5 mL blood sample. The CTCs positive rate was 75% in these patients and 79.8% in the pre-treated HCC patients charging by CTCs count $\geqslant$ 3/5 mL. While the sensitivity of EpCAM-targeted CTCs detection was reported as 30.5%–42.6% [7, 22], this method demonstrated better detecting capability. And E type, H type and M type CTCs were found in about 53%, 83% and 57% HCC patients, respectively. Besides, among a total of 551 patients with common tumors (such as lung cancer, nasopharyngeal carcinoma and colorectal cancer), the positive rate of CTCs detected by this system was 61% (lung cancer) to 80% (osteosarcoma), with the CTCs count median ranges from 5 to 10 CTCs/5 mL (Table S1). The expressions of EMT markers were observed in CTCs from each kind of cancer patient, and the overall positive rate was 53% of E type, 89% of H type and 62% of M type (Fig. S1). Therefore, H type and M type CTCs comprised a large portion of total CTCs, suggesting that the combination of epithelial (EpCAM, CK8/18/19) and mesenchymal (Vimentin, Twist, AKT2, Snail) markers on CTCs test is crucial to improve the detecting performance.

CTCs are regarded as potential seeds of the metastatic dissemination of malignant tumors [23]. Some studies have demonstrated the clinical relevance of CTCs in patients with HCC and other cancers [24, 25, 26]. CTCs count could be a biomarker in evaluating the response to different therapies and predicting recurrence of HCC [5, 27, 28]. In line with the previous reports, we found that CTCs count was closely associated with BCLC stage, metastasis and serum AFP levels in pre-treated HCC patients. The higher positive rates and increased CTCs numbers were correlated with advanced BCLC stages, higher metastatic tendency and elevated serum levels of AFP. There was no significant correlation between CTCs numbers and other clinical characteristics, including gender, age, cirrhosis, HbsAg, tumor size and serum CEA levels. When using the CTCs count to discriminate metastatic HCC patients with non-metastatic patients, the AUC of the ROC curves was 0.861 (95% CI: 0.782–0.940), and the sensitivity and specificity were 85.7% and 80.7%. The metastasis might be more likely to have occurred if the CTCs count was over 7.5 cells/5 mL. These results demonstrated total CTCs number could be an assistant diagnostic marker for cancer metastasis in HCC patients, thus making up for the shortcomings of the image examinations.

CTCs expressing mesenchymal markers were thought to have stronger invasive and metastatic ability than epithelial CTCs. Li et al.’s study proved that CTCs co-expressed Twist and Vimentin was highly correlated with portal vein tumor thrombus, and the detection of both Twist and Vimentin in CTCs could predict HCC metastasis more accurately [29]. Min et al. [30] also reported that the expression of Snail in CTCs was associated with extra-hepatic metastasis of HCC. Accordingly, we calculated the index on proportions of H $+$ M type CTCs/Total CTCs to analyze the distribution of CTCs EMT phenotype. We found that the EMT phenotypes of CTCs were closely associated with age, BCLC stage, metastasis and serum AFP levels in HCC patients, but not significantly associated with other clinical characteristics. The higher proportion of H and M type CTCs was correlated with younger ages, advanced BCLC stages, higher metastatic tendency and elevated serum levels of AFP. These results showed a higher proportion of H and M type CTCs was correlated with tumor stage and metastasis in HCC patients, indicating the distribution of CTCs EMT phenotypes could provide information of disease development.

The classification of BCLC stage is based on the evolutionary course of tumor progression and liver disease, including tumor size, number, and assessment of liver function scores [31]. It is commonly used in clinical practice because it helps to predict life expectancy and choose specific treatment mode of HCC patients. The co-expression of Twist and Vimentin in CTCs was proved highly correlated with the portal vein tumor thrombus, TNM classification and tumor size of HCC [29]. In this study, both the CTCs count and EMT phenotypes were closely associated with advanced stages of HCC. Remarkably, we noticed that some of the patients in the early stage of BCLC (stage A) had relatively higher proportions of H and M type CTCs. Large tumor size was reported to be correlated with prognosis and recurrent metastasis of HCC. Studies proved that giant tumor size over 10 cm was an independent risk factor of elevated AFP level [32], ascites after hepatectomy [33] and 5-year survival [34]. In our study, relatively higher CTCs count along with higher H and M type CTCs proportions in some patients with tumor size smaller than 5 cm was also observed. These results suggest that there may be a tendency for HCC to metastasize at early stage when the tumor volume is small. Therefore, the combination of CTCs count and EMT phenotypes may complement the analysis of patients with metastatic status and BCLC staging.

Serum AFP is the most widely used serological marker for the diagnosis and management of HCC currently. Although there is heterogeneity in the levels of AFP in HCC patients, the clinicians consider the primary tumor size and serum AFP levels as important risk factors of recurrence. Wong et al. [35]reported that in conjunction with the serum AFP test, sequential quantification of serum AFP mRNA could predict clinical metastasis/recurrence in 56% of HCC patients during a 4-year follow-up. Schulze and the colleagues showed evidence for a significant correlation between the detection of epithelial CTCs and AFP values in HCC patients [7]. Li’s group isolated the asialoglycoprotein receptors (ASGPRs) – positive CTCs from the blood of HCC patients, but the immunofluorescence staining presented that the expression of Vimentin or Twist in CTCs had no relationship with serum AFP levels [29]. Our results indicated that both the higher CTCs counts and increased H $+$ M CTCs proportions were closely associated with elevated serum AFP levels. The reasons for these controversy results may be due to the difference between CTCs isolation methods and the inherent heterogeneity of AFP in HCC patients.

Age is a noticeable characteristic of HCC in clinics, because patients with HCC are known to consist of heterogeneous populations [36]. Compared with middle-aged group (50 to 79 years) and elderly group ( $\geqslant$ 80 years), the younger HCC patients ( $<$ 50 years) tended to have higher rates of Hepatitis B virus infection, cirrhosis occurrence, AFP elevation, vascular invasion, and moderate to poor differentiation [37, 38]. Consistent with previous reports [29, 39], our results showed that the CTCs numbers were not significantly associated with age of HCC patients. However, when it was in terms of CTCs EMT phenotypes, the younger ages were correlated with higher proportions of H and M type CTCs. It indicated that the characterization of CTCs EMT phenotypes might act as a vital supplement to CTCs count in determining the disease. In addition, CTCs numbers were found $\geqslant$ 3/5 mL in all of 16 recurrent HCC patients, with a higher CTCs count median of 11.5 cells (IQR 4.75–15.75) and increased proportion of H and M type CTCs (Median 92.8%, IQR 85.7%–100%). These results further supported that the CTCs count and EMT characteristics were associated with recurrence and progression of HCC.

In conclusion, we retrospectively analyzed the results of CTCs test detected by the CanPatrol ${}^{\@setsize{\scriptsize}{8pt}{\viipt}{\@viipt}\textregistered}$ platform in HCC patients, and statistically evaluated the clinical relevance of CTCs count and EMT classification with clinicopathological features. The results supported evidence for the EMT phenomenon in CTCs from HCC patients. Simultaneous detection of the CTCs subpopulations with a hybrid phenotype and mesenchymal phenotype could improve the sensitivity of CTCs detection. Besides, analyzing the EMT characteristics of the isolated CTCs may also provide information on the judgment and management of the disease. Both the CTCs counts and EMT phenotype distributions were demonstrated highly associated with the tumor stage and metastatic status of HCC, suggesting CTCs count and EMT classification to be potential assistant markers for the early diagnosis of metastasis and evaluation of disease progression in HCC patients. Further investigations on the molecular mechanisms of EMT dynamics in CTCs are needed to illustrate specific functions of heterogeneous CTCs types, ultimately benefiting the targeted therapy of metastasis and improving the prognosis.

Conflict of interest

The authors declare that they have no conflict of interest.

Footnotes

Acknowledgments

This work was supported by the Natural Science Foundation of Guangdong Province, China (Grant No. 2015A030313247) and the Science and Technology Planning Project of Guangdong Province, China (Grant No. 2016A010105006).

Supplementary materials

Table S1

CTCs count and EMT classification in 551 patients with common cancers

Cancer type	CTCs+/Total		Median (range)/5 mL				CTCs+ patients ( $n=$ 380)
	(rates)		Total patients		CTCs+ patients		E type CTCs+		H type CTCs+		M type CTCs+
Hepatocellular cancer	146/195	(75%)	6	(0–86)	10	(3–86)	42/146	(29%)	128/146	(88%)	69/146	(47%)
Nasopharyngeal cancer	24/38	(63%)	4	(0–34)	7	(3–34)	15/24	(63%)	20/24	(83%)	20/24	(83%)
Lung cancer	70/114	(61%)	7.5	(0–67)	7.5	(3–67)	55/70	(79%)	60/70	(86%)	46/70	(66%)
Colorectal cancer	50/74	(68%)	4	(0–58)	6	(3–58)	28/50	(56%)	46/50	(92%)	27/50	(54%)
Osteosarcoma	28/35	(80%)	6	(0–35)	10	(3–35)	20/28	(71%)	27/28	(96%)	25/28	(89%)
Renal cancer	13/18	(72%)	4	(0–54)	7	(3–54)	7/13	(54%)	11/13	(85%)	13/13	(100%)
Other cancers	49/77	(64%)	4	(0–51)	5	(3–51)	35/49	(71%)	46/49	(94%)	35/49	(71%)
Total	380/551	(69%)	–		–		202/380	(53%)	338/380	(89%)	235/380	(62%)

Figure S1.

CTCs count and EMT classification in 551 patients with common cancers. (A) The distribution of patients with different cancers. (B) CTCs positive rates in different cancers. (C) EMT phenotypes of CTCs in different cancers.

References

World cancer report 2014. World health organization. http://publications.iarc.fr/Non-Series-Publications/World-Cancer-Reports/World-Cancer-Report-2014.

Chen

Zheng

Baade

P.D.

Zhang

Zeng

Bray

Jemal

X.Q.

and He

, Cancer statistics in China, 2015, CA: A Cancer Journal for Clinicians 66 (2016), 115–132.

Imamura

Matsuyama

Tanaka

Ohkubo

Hasegawa

Miyagawa

Sugawara

Minagawa

Takayama

Kawasaki

and Makuuchi

, Risk factors contributing to early and late phase intrahepatic recurrence of hepatocellular carcinoma after hepatectomy, Journal of Hepatology 38 (2003), 200–207.

Colombo

Baldan

Mazzucchelli

Martin-Padura

Marighetti

Cattaneo

Foglieni

Spreafico

Guerneri

Baccarin

Bertolini

Rossi

Mazzaferro

Cadamuro

Maggioni

Agnelli

Rebulla

Prati

and Porretti

, Evidence of distinct tumour-propagating cell populations with different properties in primary human hepatocellular carcinoma, PloS One 6 (2011), e21369.

Fan

S.T.

Yang

Z.F.

D.W.

M.N.

W.C.

and Wong

, Prediction of posthepatectomy recurrence of hepatocellular carcinoma by circulating cancer stem cells: a prospective study, Annals of Surgery 254 (2011), 569–576.

Zhang

Cao

and Yin

, Circulating tumor cells in hepatocellular carcinoma: detection techniques, clinical implications, and future perspectives, Seminars in Oncology 39 (2012), 449–460.

Schulze

Gasch

Staufer

Nashan

Lohse

A.W.

Pantel

Riethdorf

and Wege

, Presence of EpCAM-positive circulating tumor cells as biomarker for systemic disease strongly correlates to survival in patients with hepatocellular carcinoma, International Journal of Cancer 133 (2013), 2165–2171.

Sun

Y.F.

Yang

X.R.

Guo

Zhang

Qiu

S.J.

Shi

R.Y.

Zhou

and Fan

, Circulating stem cell-like epithelial cell adhesion molecule-positive tumor cells indicate poor prognosis of hepatocellular carcinoma after curative resection, Hepatology (Baltimore, Md) 57 (2013), 1458–1468.

Yamashita

Honda

Nakamoto

Baba

Nio

Hara

Zeng

S.S.

Hayashi

Kondo

Takatori

Yamashita

Mizukoshi

Ikeda

Zen

Takamura

Wang

X.W.

and Kaneko

, Discrete nature of EpCAM+ and CD90+ cancer stem cells in human hepatocellular carcinoma, Hepatology (Baltimore, Md) 57 (2013), 1484–1497.

10.

Esmaeilsabzali

Beischlag

T.V.

Cox

M.E.

Parameswaran

A.M.

and Park

E.J.

, Detection and isolation of circulating tumor cells: principles and methods, Biotechnology Advances 31 (2013), 1063–1084.

11.

Vona

Sabile

Louha

Sitruk

Romana

Schutze

Capron

Franco

Pazzagli

Vekemans

Lacour

Brechot

and Paterlini-Brechot

, Isolation by size of epithelial tumor cells: a new method for the immunomorphological and molecular characterization of circulatingtumor cells, The American Journal of Pathology 156 (2000), 57–63.

12.

Fan

Zhao

Zeng

Zaslavsky

Chen

J.J.

Frohman

M.A.

Golightly

M.G.

Madajewicz

and Chen

W.T.

, Isolation of circulating epithelial and tumor progenitor cells with an invasive phenotype from breast cancer patients, International Journal of Cancer 126 (2010), 669–683.

13.

Cristofanilli

, Circulating tumor cells, disease progression, and survival in metastatic breast cancer, Seminars in Oncology 33 (2006), S9-14.

14.

Andreopoulou

Yang

L.Y.

Rangel

K.M.

Reuben

J.M.

Hsu

Krishnamurthy

Valero

Fritsche

H.A.

and Cristofanilli

, Comparison of assay methods for detection of circulating tumor cells in metastatic breast cancer: Adna Gen Adna Test Breast Cancer Select/Detect versus Veridex Cell Search system, International Journal of Cancer 130 (2012), 1590–1597.

15.

Gorges

T.M.

Tinhofer

Drosch

Rose

Zollner

T.M.

Krahn

and von Ahsen

, Circulating tumour cells escape from EpCAM-based detection due to epithelial-to-mesenchymal transition, BMC Cancer 12 (2012), 178.

16.

Lecharpentier

Vielh

Perez-Moreno

Planchard

Soria

J.C.

and Farace

, Detection of circulating tumour cells with a hybrid (epithelial/mesenchymal) phenotype in patients with metastatic non-small cell lung cancer, British Journal of Cancer 105 (2011), 1338–1341.

17.

Armstrong

A.J.

Marengo

M.S.

Oltean

Kemeny

Bitting

R.L.

Turnbull

J.D.

Herold

C.I.

Marcom

P.K.

George

D.J.

and Garcia-Blanco

M.A.

, Circulating tumor cells from patients with advanced prostate and breast cancer display both epithelial and mesenchymal markers, Molecular Cancer Research: MCR 9 (2011), 997–1007.

18.

Liu

Huang

Yang

Deng

Yang

and Xu

, Classification of circulating tumor cells by epithelial-mesenchymal transition markers, PloS One 10 (2015), e0123976.

19.

Zhao

Cai

Cheng

Gao

Liu

Dong

Zheng

Zhang

and Yao

, Expression and clinical relevance of epithelial and mesenchymal markers in circulating tumor cells from colorectal cancer, Oncotarget 8 (2017), 9293–9302.

20.

T.T.

Liu

F.P.

Y.F.

Mou

T.Y.

Lin

Zheng

and Li

G.X.

, Evaluation of epithelial-mesenchymal transitioned circulating tumor cells in patients with resectable gastric cancer: Relevance to therapy response, World Journal of Gastroenterology 21 (2015), 13259–13267.

21.

Harouaka

R.A.

Zhou

M.D.

Yeh

Y.T.

Khan

W.J.

Das

Liu

Christ

C.C.

Dicker

D.T.

Baney

T.S.

Kaifi

J.T.

Belani

C.P.

Truica

C.I.

El-Deiry

W.S.

Allerton

J.P.

and Zheng

S.Y.

, Flexible micro spring array device for high-throughput enrichment of viable circulating tumor cells, Clinical Chemistry 60 (2014), 323–333.

22.

Guo

Yang

X.R.

Sun

Y.F.

Shen

M.N.

X.L.

Zhang

C.Y.

Zhou

Zhang

Zhou

and Fan

, Clinical significance of EpCAM mRNA-positive circulating tumor cells in hepatocellular carcinoma by an optimized negative enrichment and qRT-PCR-based platform, Clinical Cancer Research: An Official Journal of The American Association for Cancer Research 20 (2014), 4794–4805.

23.

Pantel

and Brakenhoff

R.H.

, Dissecting the metastatic cascade, Nature Reviews Cancer 4 (2004), 448–456.

24.

Liao

Wang

S.Y.

Meng

X.Y.

Yang

Shi

M.J.

Liu

H.L.

Chen

F.F.

and Xiong

, Circulating tumor cells in breast cancer and its association with tumor clinicopathological characteristics: a meta-analysis, Medical Oncology (Northwood, London, England) 31 (2014), 343.

25.

Zhou

Dong

Cui

and Tang

, The role of circulating tumor cells in evaluation of prognosis and treatment response in advanced non-small-cell lung cancer, Cancer Chemotherapy and Pharmacology 79 (2017), 825–833.

26.

Wang

Cui

Xue

Tong

and Li

, Prognostic value of circulating tumor cells in patients with squamous cell carcinoma of the head and neck: a systematic review and meta-analysis, Medical Oncology (Northwood, London, England) 32 (2015), 164.

27.

Choi

G.H.

Kim

G.I.

Yoo

J.E.

D.C.

Han

D.H.

Roh

Y.H.

Park

Y.N.

and Choi

J.S.

, Increased Expression of Circulating Cancer Stem Cell Markers During the Perioperative Period Predicts Early Recurrence After Curative Resection of Hepatocellular Carcinoma, Annals of Surgical Oncology 22 Suppl 3 (2015), S1444–1452.

28.

Zhou

Wang

Zhang

Zhou

Yang

Zhou

Guo

and Fan

, Association of preoperative EpCAM Circulating Tumor Cells and peripheral Treg cell levels with early recurrence of hepatocellular carcinoma following radical hepatic resection, BMC Cancer 16 (2016), 506.

29.

Y.M.

S.C.

Han

K.Q.

H.F.

Zheng

Zuo

G.H.

Huang

X.B.

H.Y.

Zhao

H.Z.

Z.P.

Zhou

and Liang

, Epithelial-mesenchymal transition markers expressed in circulating tumor cells in hepatocellular carcinoma patients with different stages of disease, Cell Death & Disease 4 (2013), e831.

30.

Min

A.L.

Choi

J.Y.

Woo

H.Y.

Kim

J.D.

Kwon

J.H.

Bae

S.H.

Yoon

S.K.

Shin

S.H.

Chung

Y.J.

and Jung

C.K.

, High expression of Snail mRNA in blood from hepatocellular carcinoma patients with extra-hepatic metastasis, Clinical & Experimental Metastasis 26 (2009), 759–767.

31.

Rasool

Rashid

Arooj

Ansari

S.A.

Khan

K.M.

Malik

Naseer

M.I.

Zahid

Manan

Asif

Razzaq

Ashraf

Qazi

M.H.

Iqbal

Gan

S.H.

Kamal

M.A.

and Sheikh

I.A.

, New possibilities in hepatocellular carcinoma treatment, Anticancer Research 34 (2014), 1563–1571.

32.

J.B.

F.Z.

Chen

G.F.

Qin

L.X.

and Zhang

J.H.

, Value of alpha-fetoprotein and clinical characteristics in patients with liver neoplasm, Neoplasma 61 (2014), 218–224.

33.

Chen

L.P.

Wang

Wen

T.F.

Yan

L.N.

and Li

, Risk factors of ascites after hepatectomy for patients with hepatocellular carcinoma and hepatitis B virus-associated cirrhosis, Hepatogastroenterology 59 (2012), 292–295.

34.

Capussotti

Muratore

Amisano

Massucco

Polastri

and Bouzari

, Liver resection for large-size hepatocellular carcinomas in 47 non-cirrhotic patients – no mortality and long-term survival, Hepatogastroenterology 53 (2006), 768–772.

35.

Wong

I.H.

Yeo

Leung

Lau

W.Y.

and Johnson

P.J.

, Circulating tumor cell mRNAs in peripheral blood from hepatocellular carcinoma patients under radiotherapy, surgical resection or chemotherapy: a quantitative evaluation, Cancer letters 167 (2001), 183–191.

36.

El-Serag

H.B.

, Hepatocellular carcinoma, The New England Journal of Medicine 365 (2011), 1118–1127.

37.

Wei

X.L.

Luo

H.Y.

C.F.

Jin

Zeng

Z.L.

H.Q.

Q.N.

Wang

Mao

M.J.

Liu

W.L.

Jia

W.H.

Zhang

H.Z.

Y.H.

Wang

and Xu

R.H.

, Hepatitis B virus infection is associated with younger median age at diagnosis and death in cancers, International Journal of Cancer (2017).

38.

Katsuta

Tanaka

Mogushi

Matsumura

Ban

Ochiai

Irie

Kudo

Nakamura

Tanaka

Tanabe

and Arii

, Age-related clinicopathologic and molecular features of patients receiving curative hepatectomy for hepatocellular carcinoma, American Journal of Surgery 208 (2014), 450–456.

39.

Liu

Y.K.

B.S.

Z.L.

and Lu

L.G.

, An improved strategy to detect the epithelial-mesenchymal transition process in circulating tumor cells in hepatocellular carcinoma patients, Hepatology International 10 (2016), 640–646.

Epithelial-mesenchymal transition phenotypes of circulating tumor cells correlate with the clinical stages and cancer metastasis in hepatocellular carcinoma patients

Abstract

BACKGROUND:

OBJECTIVE:

METHODS:

RESULTS:

CONCLUSIONS:

Keywords

1. Introduction

2. Subjects and methods

2.1 Patient information

2.2 Ethical approval

2.3 CTCs test and clinical data

2.4 Statistical analysis

3. Results

3.1 Patient characteristics and CTCs test

Table 1 Clinical characteristics and CTCs test of 195 HCC patients

3.3 CTCs EMT phenotypes correlate with BCLC stages and metastasis of pre-treated HCC patients

4. Discussion

Conflict of interest

Footnotes

Acknowledgments

Supplementary materials

References

Table 1
Clinical characteristics and CTCs test of 195 HCC patients