*Article* **Gastric Xanthelasma, Microsatellite Instability and Methylation of Tumor Suppressor Genes in the Gastric Mucosa: Correlation and Comparison as a Predictive Marker for the Development of Synchronous/Metachronous Gastric Cancer**

**Masashi Fukushima, Hirokazu Fukui \*, Jiro Watari, Chiyomi Ito, Ken Hara, Hirotsugu Eda, Toshihiko Tomita, Tadayuki Oshima and Hiroto Miwa**

> Division of Gastroenterology and Hepatology, Department of Internal Medicine, Hyogo College of Medicine, Nishinomiya 663-8501, Japan; ma-fukushima@hyo-med.ac.jp (M.F.); watarij@kinentou.or.jp (J.W.); s.aurantiaca66@gmail.com (C.I.); k-hara@hyo-med.ac.jp (K.H.); eda@hyo-med.ac.jp (H.E.); tomita@hyo-med.ac.jp (T.T.); t-oshima@hyo-med.ac.jp (T.O.); miwahgi@hyo-med.ac.jp (H.M.)

**\*** Correspondence: hfukui@hyo-med.ac.jp; Tel.: +81-798-45-6662

**Abstract:** A predictive marker for the development of synchronous/metachronous gastric cancer (GC) would be highly desirable in order to establish an effective strategy for endoscopic surveillance. Herein, we examine the significance of gastric xanthelasma (GX) and molecular abnormalities for the prediction of synchronous/metachronous GC. Patients (*n* = 115) were followed up (range, 12–122; median, 55 months) in whom the presence of GX and molecular alterations, including microsatellite instability (MSI) and methylation of *human mutL homolog 1* (*hMLH1*), *cyclin-dependent kinase inhibitor 2A* (*CDKN2A*) and *adenomatous polyposis coli* (*APC*) genes, had been confirmed in non-neoplastic gastric mucosa when undergoing endoscopic submucosal dissection (ESD) for early GC. At the start of surveillance, the numbers of positive subjects were as follows: GX, 59 (51.3%); MSI, 48 (41.7%); *hMLH1*, 37 (32.2%); *CDKN2A*, 7 (6.1%); *APC*, 18 (15.7%). After ESD treatment, synchronous/metachronous GCs occurred in patients with the following positive factors: GX, 16 (27.1%); MSI, 7 (14.6%); *hMLH1*, 6 (16.2%); *CDKN2A*, 3 (42.9%); *APC*, 3 (16.7%). The presence of GX had no significant relationship to positivity for MSI or methylation of *hMLH1*, *CDKN2A* or *APC*. GX was significantly (*p* = 0.0059) and independently (hazard ratio, 3.275; 95% confidence interval, 1.134–9.346) predictive for the development of synchronous/metachronous GC, whereas those genetic alterations were not predictive. GX is a simple and powerful marker for predicting the development of synchronous or metachronous GC.

**Keywords:** gastric xanthelasma; synchronous/metachronous gastric cancer; endoscopic submucosal dissection; genetic alteration; predicting marker

#### **1. Introduction**

Gastric cancer (GC) is still a leading cause of cancer-related mortality, especially in eastern countries [1,2]. With advances in endoscopy and diagnostic strategies, a considerable number of GCs can now be detected at an early stage and be treated curatively by endoscopic submucosal dissection (ESD) [3,4]. However, after endoscopic treatment of GC, the development of synchronous and/or metachronous lesions is a concern during follow-up [5,6]. For this reason, although endoscopic surveillance is recommended, no specific strategies for assisting the frequency or risk of such lesions have yet been established. Therefore, to improve the efficiency and effectiveness of endoscopic surveillance, a predictive marker for the development of synchronous/metachronous GC would be highly desirable.

It has been accepted that irreversible accumulation of molecular abnormality occurs in precancerous conditions, i.e., atrophic gastritis and intestinal metaplasia with chronic

**Citation:** Fukushima, M.; Fukui, H.; Watari, J.; Ito, C.; Hara, K.; Eda, H.; Tomita, T.; Oshima, T.; Miwa, H. Gastric Xanthelasma, Microsatellite Instability and Methylation of Tumor Suppressor Genes in the Gastric Mucosa: Correlation and Comparison as a Predictive Marker for the Development of Synchronous/ Metachronous Gastric Cancer. *J. Clin. Med.* **2022**, *11*, 9. https://doi.org/ 10.3390/jcm11010009

Academic Editor: Ugo Grossi

Received: 24 November 2021 Accepted: 17 December 2021 Published: 21 December 2021

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

*Helicobacter pylori* (*H. pylori*) infection [7]. Thus, the whole of the gastric mucosa with *H. pylori* infection has a high potential for development of GC, which is consistent with the frequent occurrence of synchronous/metachronous GC in this situation. Interestingly, it has been shown that microsatellite instability (MSI) or methylation of tumor-suppressor genes frequently occurs in the non-neoplastic gastric mucosa of patients with GC [8–10] and that, moreover, these molecular abnormalities do not completely normalize, even after successful eradication of *H. pylori* [7,8,11]. These findings offer a good explanation of the relatively high frequency of GC development, even after successful eradication of *H. pylori* infection. In these contexts, molecular abnormalities in the gastric mucosa may be candidate markers for prediction of the development of GC [10,12,13]. We recently reported that the incidence of molecular events related to carcinogenesis was mostly observed in IM, with very few in atrophic mucosa without intestinal metaplasia [14]. On the other hand, we recently reported that gastric xanthelasma (GX), characterized by accumulation of lipid in histiocytic foam cells [15], is a useful marker for prediction of the development of GC [16,17]. In the present study, therefore, we analyzed the correlation between GX and molecular abnormalities in the gastric mucosa, especially in intestinal metaplasia, of patients with early GC and investigated its significance for prediction of synchronous/metachronous GC.

#### **2. Materials and Methods**

#### *2.1. Patients*

This was a cohort study following our previous investigation in molecular alterations in the non-neoplastic gastric mucosa of patients with early GC [8]. Written informed consent had been obtained from all patients involved in the previous study (Ethics Nos. 136 and 154), and the opt-out for this observational study (Ethics No. 0404) was announced on the website of Hyogo College of Medicine. All studies were approved by the Ethics Committee of Hyogo College of Medicine.

A total of 115 patients were investigated in this study. All patients satisfied the following criteria: 1, subjects who had undergone ESD for GCs between August 2010 and December 2013; 2, subjects who had been enrolled in the previous study [8] and examined for molecular alterations in non-neoplastic gastric mucosa when receiving ESD treatment; 3, subjects who had been followed-up for ≥12 months by endoscopy to examine whether synchronous or metachronous GC had occurred after ESD treatment.

In the present study, we used the criteria of the Japanese Research Society for Gastric Cancer as the histological criteria for gastric cancer. The criteria of Kimura and Takemoto, reported previously [18,19], were also adopted for the severity of gastric atrophy. Endoscopists (M.F., K.H. and H.E.), who were blinded to the data of molecular alterations and patients' clinical course, confirmed the presence of xanthomas in endoscopic examination before ESD treatment, retrospectively. At the time of ESD, the status of *H. pylori* infection was determined by Giemsa staining of gastric biopsy samples and the obtained serum level of anti-*H. pylori* antibody and then defined as positive if at least one test gave a positive result. If *H. pylori* had been eradicated after ESD, the status of infection was examined by urease breath test.

#### *2.2. Analyses of MSI and Gene Methylation*

Molecular alterations in the intestinal metaplasia were analyzed as described previously [8]. In brief, biopsy specimens of non-neoplastic gastric mucosa at the greater curvatures of the antrum and corpus and the lesser curvature of the angulus were embedded in paraffin blocks. Seven-micrometer-thick tissue sections were cut, samples of epithelial cells were isolated by laser microdissection, and DNA was extracted only from the goblet intestinal metaplasia glands (incomplete type) using a QIAamp DNA Micro Kit (Qiagen, Hilden, Germany).

We examined five microsatellite loci on chromosomes for MSI based on the revised Bethesda panel [20], as follows: 2p (BAT26), 4q (BAT25), 2p (D2S123), 5q (D5S346) and 17p (D17S250). The MSI status was judged as previously reported [8]. To analyze the genetic methylation status, extracted DNA was modified using sodium bisulfite with the EpiTect Plus DNA Bisulphite Kit (Qiagen, Hilden, Germany). The modified DNA was amplified using specific primer pairs for the methylated or unmethylated sequences of *human mutL homolog 1* (*hMLH1*), *cyclin-dependent kinase inhibitor 2A* (*CDKN2A*) and *adenomatous polyposis coli* (*APC*) [8]. Thereafter, the methylation status of those genes was examined by methylation-sensitive high-resolution melting analysis, as previously described [8,21]. A methylation standard curve was prepared using a set of methylated (100%) and unmethylated (0%) DNA (EpiTect PCR Control DNA Set; Qiagen, Hilden, Germany). The methylation status of each target gene was scored as low (<10%), moderate (≥10% to <50%) or high (≥50%). Samples with a moderate or high methylation level were considered to be methylated.

#### *2.3. Statistical Analysis*

The Statview 5.0J statistical software package (Abacus Concepts Inc., Berkeley, CA, USA) was used for all analyses in the present study. Data for age and BMI were expressed as the mean ± SD, and categorical data were presented as frequencies with proportion. Differences in age and BMI between two groups were analyzed by unpaired two-tailed t test or by Mann-Whitney *U*-test when the data were not parametric. Fisher's exact test was performed to investigate the relationships between groups and clinical/genetic features. Cumulative incidence of synchronous/metachronous GC development after ESD treatment was evaluated by the Kaplan-Meier method and analyzed by log-rank test. Differences at *p* < 0.05 were considered to be statistically significant.

#### **3. Results**

#### *3.1. Relationship of GX to Clinical/Endoscopic Features in Patients with Early GC Treated by ESD*

Table 1 summarizes the clinical and endoscopic features of patients with early GC treated by ESD (*n* = 115). Most of the patients (*n* = 108, 93.9%) had open-type gastric atrophy. Eighty-nine patients (77.4%) were positive for *H. pylori* infection, and 11 were negative after *H. pylori* eradication. Fifteen were negative for *H. pylori* infection without eradication, and 13 of them had atrophy, suggesting previous *H. pylori* infection.


**Table 1.** Comparison of clinical features between patients with and without gastric xanthelasma.

GX, gastric xanthelasma; BMI, body mass index.

GX was detected in 59 (51.3%) of the 115 patients investigated. None of the parameters age, sex, BMI, severity of gastric atrophy or *H. pylori* infection status—showed a significant relationship to the prevalence of GX.

#### *3.2. Relationship of MSI or Methylation of hMLH1, CDKN2A or APC to Clinical/Endoscopic Features in Patients with Early GC Treated by ESD*

Among 115 patients with early GC who underwent ESD, 48 (41.7%) were positive for MSI (Table 2). None of the examined parameters—age, sex, BMI, severity of gastric atrophy or *H. pylori* infection status—showed a significant relationship with MSI positivity.

Methylation of the *hMLH1* gene was detected in 37 (32.2%) of the patients with early GC who underwent ESD. Positivity for *hMLH1* methylation showed no relationship with any of the above clinical/endoscopic features either.

Methylation of the *CDKN2A* and *APC* gene was detected in 7 (6.1%) and 18 (15.7%) of patients with early GC who underwent ESD, respectively. Positivity for *CDKN2A* or *APC* methylation showed no relationship with any of the above clinical/endoscopic features either.


**Table 2.**

Comparison

 of clinical features between patients with and without MSI or methylation

 of tumor suppressor

 genes.

#### *3.3. Relationship between GX and MSI or Methylation of hMLH1, CDKN2A or APC in Patients with Early GC Treated by ESD*

We next investigated the relationship between the prevalence of GX and molecular alterations in the gastric mucosa of patients with early GC (Table 3). Contrary to expectation, we found no significant correlation between the prevalence of GX and molecular alterations of MSI or methylation of *hMLH1*, *CDKN2A* or *APC*.

**Table 3.** Relationship between gastric xanthelasma and genetic alterations in early gastric cancer patients.


GX, gastric xanthelasma; MSI, microsatellite instability; *hMLH1*, *human mutL homolog 1*; *CDKN2A*, *cyclin-dependent kinase inhibitor 2A*; *APC*, *adenomatous polyposis coli*.

*3.4. Significance of GX, MSI and Methylation of Tumor Suppressor Genes as a Predictive Marker for the Development of Synchronous/Metachronous GC*

During the follow-up period, synchronous/metachronous GC was found in 21 (18.3%; 5 synchronous and 16 metachronous, respectively) of the 115 patients (Table 4). When investigating according to the prevalence of GX, 16 (27.1%) of the 59 patients with GX developed synchronous/metachronous GC after ESD treatment. On the other hand, 5 (8.9%) of 56 patients without GX had such lesions. As for the prevalence of MSI, 7 (14.6%) of 48 patients with MSI had synchronous/metachronous GC, and 14 (20.9%) of 67 patients without MSI had such lesions. In addition, 6 (16.2%) of 37 patients with *hMLH1* methylation had synchronous/metachronous GCs and 15 (19.2%) of 78 patients without *hMLH1* methylation had such lesions. Three (42.9%) of seven patients with *CDKN2A* methylation had synchronous/metachronous GCs, and 3 (16.7%) of 18 patients with *APC* methylation had such lesions.

Furthermore, we compared the cumulative incidence of synchronous/metachronous GC between GX-positive and -negative cases (Figure 1). Kaplan–Meir curves show that significantly more patients with GX developed synchronous/metachronous GC than those without GX (Figure 1). In terms of the status of MSI and methylation of *hMLH1*, *CDKN2A*, or *APC*, the Kaplan–Meir curves show no significant differences between the groups positive and negative for those genetic alterations (Figure 1).

**Figure 1.** Cumulative incidence of synchronous/metachronous gastric cancer during endoscopic follow-up (median, 55 months; range, 12–122 months) in patients after ESD treatment. GX, gastric xanthelasma; MSI, microsatellite instability; *hMLH1*, *human mutL homolog 1*; *CDKN2A*, *cyclin-dependent kinase inhibitor 2A*; *APC*, *adenomatous polyposis coli*.

We next examined whether the presence of GX is an independent factor predictive of synchronous/metachronous GC development. Univariate analysis showed that GX was significantly related to the development of synchronous/metachronous GC (Table 4). Moreover, multivariate analysis clarified that the presence of GX was independently related to the development of synchronous/metachronous GC (Table 4).


**Table 4.** Univariate and multivariate analyses of the cumulative incidence of synchronous or metachronous gastric cancer during endoscopic follow-up in patients after ESD treatment.

GX, gastric xanthelasma; MSI, microsatellite instability; *hMLH1*, *human mutL homolog 1*; *CDKN2A*, *cyclin-dependent kinase inhibitor 2A*; *APC*, *adenomatous polyposis coli*.

#### **4. Discussion**

On the basis of clinical/endoscopic features, the identification of a predictive marker for the development of synchronous/metachronous GC has long been desirable. Accumulating evidence has revealed that male sex and severe atrophy are independent risk factors for not only initial but also synchronous/metachronous GC [5,6]. In addition, we previously reported that GX is a powerful marker for prediction of the development of GC [16,17]. Moreover, in the present study, we have clarified that GX is a possible marker for prediction of the development of synchronous/metachronous GC, which is consistent with a report by Shibukawa et al. [22]. GX is characterized by accumulation of foamy histiocytes in the inflamed gastric mucosa and is thought to be the result of an inflammatory response to mucosal damage or aging [15,23]. In this regard, one might argue that gastric xanthelasma merely reflects the severity and long duration of gastric atrophy, which is a crucial risk factor for GC development. However, our previous multivariate analysis clearly indicates that GX is a factor independent of gastric atrophy for prediction of the development of GC [17]. Moreover, the present study similarly clarifies its significance as an independent predictor for synchronous/metachronous GC. It has been reported that increased release of oxygen free radicals, which cause DNA damage and play a role in the pathophysiology of various malignancies [24,25], is involved in the formation of GX [15]. Thus, it is tempting to speculate that the presence of GX may reflect the activation of oxygen free radicals and the associated promotion of genetic alterations in the gastric mucosa. In

this context, we therefore investigated the relationship between the presence of GX and molecular alterations in the gastric mucosa of patients with GC.

MSI is a form of genetic instability characterized by alterations in the length of the tandem repeat sequence (termed "microsatellite") [26], owing to inactivation of mismatch repair genes, such as *hMSH2* and *hMLH1* [27], and it is evident that MSI and/or methylation of *hMLH1* is frequent in various malignancies [28]. In addition, the methylation of tumor suppressor *CDKN2A* and *APC* is widely involved in gastrointestinal carcinogenesis by affecting cell cycle or proliferation [29–32]. In these contexts, we and others have shown that MSI and/or methylation of tumor-suppressor genes, including *hMLH1*, frequently occurs in the non-neoplastic gastric mucosa of patients with early GC [8,10,33] and that these molecular alterations can be potential markers for prediction of the development of GC [10,34]. In the present study, methylation of *CDKN2A* and *APC* was not very frequent in the non-neoplastic gastric mucosa, especially in intestinal metaplasia of patients with early GC and not predictive of the development of synchronous/metachronous GC, suggesting that those gene alterations may not be very critical in gastric carcinogenesis. On the other hand, it is noteworthy that *H. pylori* eradication is unable to normalize any molecular abnormality for MSI and *hMLH1* in patients with early GC who undergo ESD [8], which supports the contention that *H. pylori* eradication cannot necessarily prevent the development of metachronous GC [35]. In this context, it is interesting that GX persists even after *H. pylori* eradication [36] and that its presence is a predictive marker for the development of synchronous/metachronous GC. We then investigated the relationship between GX and the status of MSI or *hMLH1* methylation, but contrary to expectation, no significant correlations were evident. Besides these molecular alterations, considerable patterns of genetic abnormality are involved in the development of GC [37–39]. Therefore, it may be an interesting theme to identify the molecular alteration responsible for the occurrence of GX in the gastric mucosa.

We next investigated whether MSI or *hMLH1* methylation in the intestinal metaplasia is predictive for the development of synchronous/metachronous GC, as such molecular alterations may be applicable to prediction of the initial development of GC [10]. However, the results suggest that neither MSI nor *hMLH1* methylation is likely to predict the development of synchronous/metachronous GC in patients after ESD treatment. These results may be reasonable, as several studies have shown that MSI and/or *hMLH1* methylation is not useful for prediction of the development of GC [9,40]. On the other hand, genetic researchers have continuously investigated and identified some candidate molecular markers (methylation of *microRNA-34b/c* and *-124a3* or somatic mutation of *ARID1A* and *MAGI1*) for prediction of the development of metachronous GC [41–43]. However, since molecular alterations in GC patients are very complex and diverse [37–39], it might be difficult to select a specific genetic marker that can predict the development of synchronous/metachronous GCs.

In summary, although the molecular alteration responsible for the occurrence of GX in the gastric mucosa remains unclear, GX is a powerful marker for prediction of the development of synchronous/metachronous GC, at least compared with molecular alterations of MSI or methylation of *hMLH1*, *CDKN2A* or *APC* in patients with early GC. GX is very easy to detect in routine endoscopic examinations, whereas detection of molecular abnormality needs advanced equipment and technology. Thus, in clinical practice, GX may be a very useful marker for identification of patients, during follow-up surveillance, who are at high risk for development of synchronous/metachronous GC. The possibility that a powerful molecular marker might become available in the future for prediction of the development of synchronous/metachronous GC cannot be excluded. However, we believe that GX is a simple yet very effective marker in patients undergoing endoscopic surveillance for development of synchronous/metachronous GC and that the usefulness of GX should be validated in a large-scale, prospective, multi-center study.

**Author Contributions:** Conceptualization, M.F. and H.F.; methodology, H.F., J.W., C.I., K.H. and H.E.; validation, H.F., J.W. and C.I.; formal analysis, H.F., J.W., C.I., K.H. and H.E.; investigation, M.F., H.F., J.W., C.I., K.H. and H.E.; resources, M.F., J.W., K.H. and H.E.; data curation, M.F., H.F., J.W., C.I., K.H. and H.E.; writing—original draft preparation, M.F. and H.F.; writing—review and editing, M.F., H.F., J.W., C.I., K.H., H.E., T.T., T.O. and H.M.; supervision, T.T., T.O. and H.M.; project administration, H.F.; funding acquisition, H.F. All authors have read and agreed to the published version of the manuscript.

**Funding:** This work was supported in part by Grants-in-Aid for Scientific Research 21K08016 from the Ministry of Education, Culture, Sports, Science and Technology, Japan.

**Institutional Review Board Statement:** The study was conducted according to the guidelines of the Declaration of Helsinki, and the collection and analyses of biomaterials and clinical records were approved by the Institutional Ethics Committee of Hyogo College of Medicine (Ethics Nos. 136, 154 and 0404).

**Informed Consent Statement:** Informed consent was obtained from all subjects involved in the study. The opt-out for this observational study was announced on the website of Hyogo College of Medicine.

**Data Availability Statement:** Any data referred to in this work will be available on request.

**Acknowledgments:** We thank Mayumi Yamada and Kayo Tsubota (Hyogo College of Medicine) for their technical assistance.

**Conflicts of Interest:** The authors declare no conflict of interest. The funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript, or in the decision to publish the results.

#### **References**


## *Article* **Vimentin-Positive Circulating Tumor Cells as Diagnostic and Prognostic Biomarkers in Patients with Biliary Tract Cancer**

**Sung Yong Han 1,†, Sung Hee Park 1,†, Hyun Suk Ko 1, Aelee Jang 2, Hyung Il Seo 3, So Jeong Lee 4, Gwang Ha Kim <sup>1</sup> and Dong Uk Kim 1,\***


**Abstract:** Biliary tract cancer (BTC) has poor prognosis; thus, early diagnosis is important to decrease mortality. Although vimentin-positive circulating tumor cells (V-CTCs) are a good candidate for diagnostic and prognostic biomarkers, studies on the topic are limited. We aimed to evaluate the diagnostic efficacy of V-CTCs between BTC and benign biliary disease (BBD) and determine the prognostic value of V-CTCs in BTC patients. We recruited 69 participants who had BTCs and BBDs from a single tertiary referral center. We analyzed CTCs and V-CTCs in peripheral blood using the CD-PRIMETM system. Seven patients were excluded due to a technical failure of CTC detection. CTCs were detected in all 62 patients. CTC count > 40/mL blood (55.8% vs. 20%, *p* = 0.039), V-CTC count > 15/mL blood (57.7% vs. 10%, *p* = 0.005), and V-CTC/CTC ratio > 40% (48.1% vs. 10%, *p* = 0.025) were significantly different between BTCs and BBDs. Two or more of these three parameters (61.5% vs. 10%, *p* = 0.002) increased the accuracy. A combination of CTC markers with CA19-9 and biopsy increased the accuracy (90.4% vs. 10%, *p* = 0.000). V-CTC > 50/mL blood was a significant factor affecting survival (140 (66.6–213.3) vs. 253 (163.9–342.1) days, *p* = 0.008). V-CTC could be a potential biomarker for early diagnosis and predicting prognosis in patients with BTC.

**Keywords:** biliary tract cancer; circulating tumor cell; vimentin; diagnosis; prognosis

#### **1. Introduction**

Biliary tract cancer (BTC) is a rare type of cancer that occurs in 2–3 per 100,000 persons. The incidence is more than two times higher in northeast Asia than in other countries. However, the incidence is increasing worldwide, particularly in western countries. Furthermore, the mortality rate is relatively high compared to those of other gastrointestinal malignancies, despite the development of therapeutic agents [1–5]. The poor prognosis of BTC is largely due to delayed diagnosis from late examination because of non-specific symptoms such as dyspepsia, weight loss, and abdominal discomfort in the early disease stage. Additionally, BTC tissues are paucicellular with abundant fibrous stroma, leading to false negatives in pathology and resulting in late diagnosis and poor prognosis. Therefore, an exact early diagnostic method is needed for the improvement of prognosis of BTC patients.

Circulating tumor cells (CTCs) are good candidates for diagnostic or prognostic biomarkers because they enable frequent, non-invasive analysis and provide real-time dynamics of BTC. Efficient technologies for CTC analysis have been developed since the

**Citation:** Han, S.Y.; Park, S.H.; Ko, H.S.; Jang, A.; Seo, H.I.; Lee, S.J.; Kim, G.H.; Kim, D.U. Vimentin-Positive Circulating Tumor Cells as Diagnostic and Prognostic Biomarkers in Patients with Biliary Tract Cancer. *J. Clin. Med.* **2021**, *10*, 4435. https:// doi.org/10.3390/jcm10194435

Academic Editor: Matteo Neri

Received: 24 July 2021 Accepted: 23 September 2021 Published: 27 September 2021

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

U.S. Food and Drug Administration approved the CellSearch system for clinical use to detect CTCs in peripheral blood in January 2004 [6–8]; however, CTC isolation and characterization remains challenging due to their rarity and heterogeneity. The use of CTCs to predict clinical outcomes is far from being applied in the real world, but these applications are being actively researched since efficient CTC enrichment is possible with recent technological advances. A centrifugal microfluidic device with fluid-assisted separation technology (FAST disc) enables label-free CTC isolation from whole blood in a size-selective manner. This system uses tangential flow filtration (TFF), which allows clog-free, ultrafast (>3 mL/min) CTC enrichment with gentle reductions in pressure (~1 kPa) for collecting a large amount of tumor cells with high viability.

CTCs are detected even in precancerous lesions by circulating along blood vessels through the epithelial to mesenchymal process [9,10]. Thus, tumor detection could be possible by detecting CTCs, especially vimentin-positive CTCs developed during the epithelial-mesenchymal transition (EMT) process in the early disease stage. Furthermore, vimentin expression in CTCs is possibly highly correlated with cancer progression rather than CTCs [11,12].

There are limited studies on using CTCs for early tumor detection and prognosis of BTC. Further, the cut-off for a positive CTC value has not yet been defined. The aim of this study was to evaluate the diagnostic efficacy of vimentin-positive circulating tumor cells (V-CTCs) in BTCs and benign biliary diseases (BBDs). Additionally, we aimed to determine the prognostic value of V-CTCs in BTC patients.

#### **2. Materials and Methods**

#### *2.1. Patient Characteristics*

We recruited 69 participants from a single tertiary referral center in South Korea between June 2018 and February 2021. The inclusion criteria for BTCs were (1) age ≥ 18 years; (2) BTC diagnosis based on ultrasound (US), computed tomography (CT), and magnetic resonance imaging (MRI); and (3) histological confirmation as adenocarcinoma. The inclusion criteria for BBDs were (1) age ≥ 18 years; (2) benign biliary diseases such as cholelithiasis and benign biliary stricture based on US, CT, and MRI; and (3) no history of other malignancies within 5 years. Blood samples were collected at the initial visit. Seven patients were excluded due to a technical failure of CTC detection. Finally, 62 patients were enrolled for the assessment of CTC number (Figure 1).

Patients were followed clinically using medical records to determine treatment regimens and responses, including surgery, disease progression, and time of death. This prospective trial was conducted at a single tertiary medical center with institutional review board approval (H-H-1801-020-062), and all patients provided written informed consent. The Clinical Research Information Service (CRIS) approved this study (KCT0003511).

#### *2.2. CTC Enumeration and Characterization*

For all cases, peripheral blood samples were maintained at room temperature and pretreatment was performed within 2 h of collection. We used a CD-PRIMETM system (Clinomics, Ulsan, Korea) which is a commercialized version of the FAST disc. The system contains two parts, the CD-CTCTM Duo (disc) and a CD-OPR-1000TM (disc operating machine); we collected intact CTCs from the white buffy coat resuspended with phosphate buffered saline (PBS) in the same amount as the original blood of each BTC patient.

Immunostaining was performed to identify the isolated cells on the membrane in the filtration chamber of the FAST disc. The isolated cells were stained with fluorescenceconjugated antibodies, including FITC conjugated anti-EpCAM antibody (1:417, 9C4; BioLegend, San Diego, CA, USA), Alexa488 conjugated anti-pan-cytokeratin antibody (1:100, AE1/AE3; Invitrogen, Carlsbad, CA, USA), FITC conjugated anti-cytokeratin antibody (1:500, CAM5.2; BD Biosciences, San Diego, CA, USA), Alexa555 conjugated anti-vimentin antibody (1:125, D21H3; Cell signaling, MA, USA), and Cy5 conjugated anti-CD45 antibody (1:50, F10-89-4; Southern biotech, Birmingham, AL, USA) in PBS with 0.01% tween 20 and mounted with 4,6-diamidino-2-phenylindole (DAPI, Abcam, Cambridge, CB2, UK).

The cells were fixed in 4% paraformaldehyde for 20 min and stained with surface antibodies (CD45, EpCAM) in the dark for 20 min. Then, the cells were permeabilized with 0.01% Triton-X for 10 min and stained with intracellular antibodies (cytokeratin, pancytokeratin, and vimentin) in the dark for 20 min. Finally, the cells were stained with DAPI and examined under a fluorescence microscope. All staining processes were performed at room temperature, and the cells were washed with PBS in each step.

Cells were counted as CTCs if they had intact morphology (large cell with an intact nucleus; cut-off size of CTCs is 8 μm), stained positive for EpCAM, pan-cytokeratin, cytokeratin, and DAPI, and stained negative for CD45 by researchers blinded to the patient clinical status. In addition, V-CTCs were referred to positive staining for vimentin.

To validate the expression of CTCs, we spiked 100 cells of BTC cell lines such as SNU-1079, SNU-308, and SNU-1196 in 3 mL blood of healthy subjects. After the enrichment of cancer cells from spike-in blood, the cancer cells on the membrane were stained with fluorescence-conjugated antibodies and confirmed the CTCs expression marker.

#### *2.3. Outcome Assessment*

The primary study endpoint was to reveal the relationship between baseline CTC counts, V-CTC counts, V-CTC proportion, and the pathologic BTC diagnosis. The secondary endpoints were to find correlations between baseline V-CTC counts, progression free survival (PFS), and overall survival (OS).

The patients were followed for disease progression by imaging and laboratory testing. PFS was defined as the relapsed time from the time of pathologic diagnosis, and was assessed by peripheral blood sample collection, CT, MRI, and positron emission tomography (CA19-9) imaging. OS was defined as the time elapsed from the time of pathologic diagnosis until death.

#### *2.4. Statistical Analysis*

Statistical analysis was performed using IBM SPSS statistical software, version 21.0 (IBM Corp, Armonk, NY, USA). Descriptive statistics are presented as frequencies and percentages for categorical variables and as means ± standard deviations for continuous variables. Two or three-sample comparisons were performed using the Student's t-tests and ANOVA test for normally distributed variables. Wilcoxon rank sum tests and Kruskal– Wallis tests were used for non-parametric comparisons. A two-sided *p*-value of <0.05 was used to indicate statistical significance in all analyses. Differences in OS were plotted using Kaplan–Meier survival plots and tested using log-rank tests. The optimal cut-off value for CTC counts was determined using receiver operating characteristic (ROC) curves and the area under the curve (AUC) values were calculated. To evaluate the factors affecting the prognosis, COX regression analysis was performed, with factors known as prognostic markers and CTC markers as variables.

#### **3. Results**

#### *3.1. Patient Characteristics*

A total of 62 patients were enrolled for the assessment of CTC markers. Of them, 10 were diagnosed with BBDs and 52 were diagnosed with BTC (8 with gallbladder cancer (GB), 12 with intrahepatic cholangiocarcinoma (IHCC), 21 with extrahepatic cholangiocarcinoma (EHCC), and 11 with perihilar cholangiocarcinoma (PHCC)). Table 1 shows the characteristics of patients with benign, resectable, and unresectable BTC. All epidemiologic factors except smoking status were the same between patients. The mean age of patients with BBDs (50% male) was 66.1 years and that of patients with BTCs (61.5% male) was 69.2 years. The alanine transaminase ALT (28.6 vs. 119.9%, *p* = 0.026), alkaline phosphatase (ALP) (139.9 vs. 326.7, *p* = 0.035), and total bilirubin (0.81 vs. 5.76, *p* = 0.048) levels were significantly different between the BBD and BTC groups, respectively. These factors are markers of biliary obstruction. The CEA (3.0 vs. 3.9 vs. 11.1, *p* = 0.021) and CA19-9 (16.0 vs. 434.3 vs. 1165.1, *p* = 0.040) levels were also significantly different between BBD, resectable BTC, and unresectable BTC groups.


**Table 1.** Patient characteristics and CTC counts/v-CTC proportion.


**Table 1.** *Cont.*

IHD (intrahepatic duct), CBD (common bile duct), GB (gallbladder), IHCC (intrahepatic cholangiocarcinoma), EHCC (extrahepatic cholangiocarcinoma), HBV (hepatitis B virus), HCV (hepatitis C virus), LC (liver cirrhosis), NLR (neutrophil/lymphocyte ratio), PNI (prognostic nutrition index) \*: *p*-value < 0.05.

#### *3.2. CTC Counts in BTC and BBD*

Figure 2 shows the results of CTC and V-CTC analysis in patients with unresectable and resectable BTC and patients with BBD. Though the CTC and V-CTC counts differed between the BTC and BBD groups, this difference was not significant, whereas the V-CTC/total CTC count ratio (VCR) showed a statistically significant difference between the groups (35.7% vs. 23.8%, respectively, *p* = 0.048). There were no statistically significant differences in CTC count, V-CTC count, and VCR between patients with resectable and unresectable BTC (Figure 3). The CTC count, V-CTC count, and VCR cut-off values, determined via ROC curve analysis, were 40/mL blood, 15/mL blood, and 40%, respectively (Supplementary Figure S1.) When CTCs were analyzed using these cut-off values, significant difference across all three parameters were found between the BTC and BBD groups (CTC > 40: 55.8% vs. 20%, *p* = 0.039; V-CTC > 15: 57.7% vs. 10%, *p* = 0.005; VCR > 40%: 48.1% vs. 10%, *p* = 0.025, respectively). Analyzing any two of the three parameters in combination precipitated a more statistically significant difference between the BTC and BBD groups (61.5% vs. 10%, *p* = 0.002) than using any one parameter alone (*p* = 0.002). Notably, when patients showed two of three parameters plus biopsy results or elevated CA19-9 levels, the sensitivity and specificity of discrimination between BBD and BTC increased (90.4% vs. 10%, *p* < 0.001). (Table 2).



**Table 2.** *Cont.*

BTC, biliary tract cancer; V-CTC, Vimentin + CTC; VCR, vimentin/CTC ratio; Three parameter (CTC count > 40, V-CTC > 15, VCR > 40%), Biopsy (+): malignancy was proven by biopsy, UNL: upper normal limit, AUC: area under curve, \*: *p*-value < 0.05.

**Figure 2.** The yellow arrows indicate CTCs (PanCK+/CKCAM+/EpCAM+/CD45-) and white arrows indicate V-CTCs (PanCK+/CKCAM+/EpCAM+/CD45-, vimentin+) in BTC and BBD patients.

**Figure 3.** CTC counts and V-CTC counts in patients with biliary disease.

#### *3.3. Subgroup Analysis: Benign vs. Resectable Biliary Tract Cancer*

Table 3 shows the CTC counts of the BBD and resectable BTC groups. The indicators used to distinguish the BTC and BBD groups were applied between resectable BTC and BBDs. CTC count > 40/mL blood (54.5% vs. 20%, *p* = 0.002), V-CTC count > 15/mL blood (51.5% vs. 10%, *p* < 0.001), and VCR > 40% (45.5% vs. 10%, *p* = 0.000) were used to differentiate the two groups. Using two of the three parameters in combination (57.6% vs. 10%, *p* = 0.007) also yielded statistically significant results. Further, using a combination of these parameters, patient biopsy results, and elevated CA19-9 level data also increased the sensitivity and specificity of discriminating between BBDs and resectable BTC (Table 3).

**Table 3.** CTC count and sensitivity and specificity of each parameter between benign and resectable biliary tract cancer.



**Table 3.** *Cont.*

BTC, biliary tract cancer; V-CTC, Vimentin + CTC; VCR, vimentin/CTC ratio; Three parameter (CTC count > 40, V-CTC > 15, VCR > 40%) AUC: area under curve, \*: *p*-value < 0.05.

#### *3.4. Association of the CTC Count with Prognosis*

In the prognostic analysis of patients with BTC using their neutrophil/lymphocyte ratio (NLR), CA19-9 level, CTC count, V-CTC count, and VCR data, V-CTC counts > 50/mL blood was found to be the most significant (Table 4.) This cut-off value of V-CTC count was determined by ROC curve analysis, under or over 250 days (mean OS = 257 ± 184 days, AUC = 0.615, sensitivity = 32.1%, specificity = 87.5%). There was no significant difference in the baseline characteristics between the groups according to V-CTC counts of 50/mL blood (Table 5). Other non-significantly different prognostic markers included the NLR. Figure 4 shows the Kaplan–Meier survival analysis. Patients BTC with V-CTC count > 50/mL blood showed a poorer prognosis than other patients with BTC (median survival: 140 (66.6–213.3) vs. 253 (163.9–342.1) days, *p* = 0.008). In patients with resectable BTC, the prognosis was significantly different between patients with V-CTC count >50/mL blood and V-CTC count < 50/mL blood (median survival: 167 (97.7–236.3) vs. 311 (254.8–367.2) days, *p* = 0.004). The median survival of the V-CTC count > 50 and count < 50 groups in subgroup analysis according to the location of the cancer was 170 (0–345.3) vs. 95 (0–224.6) days for IHCC (*p* = 0.076), 307 (267.8–346.2) vs. 218 (117.1–318.9) days for EHCC (*p* = 0.072), and 324 (6.8–443.1) vs. 138 days for GB cancer (*p* = 0.353), which was similar to the result obtained with the total number of patients, though the number of patients in each subgroup was too low to obtain meaningful results. The PHCC group showed similar median survival between the >50 and <50 V-CTC count groups (293 (141.1–444.9) vs. 245 days, *p* = 0.835) (Supplementary Table S1.) However, PFS was not significantly different between the groups in accordance with any CTC marker, except CA19-9 level (CA19-9 < 40 vs. >40: 284 (168.5–399.5) vs. 163 (152.5–217.4) days, *p* = 0.011, respectively) (Supplementary Table S2).

**Table 4.** Prognostic factor analysis via Cox regression analysis.


HR, Hazard ratio; CI, confidence interval; CTC, circulating tumor cells; V-CTC, Vimentin + CTC; VCR, vimentin/CTC rate; UNL, upper normal limit; NLR, neutrophil-lymphocyte ratio, \*: *p*-value < 0.05.


**Table 5.** Baseline characteristics according to V-CTC level.

GB (gallbladder), IHCC (intrahepatic cholangiocarcinoma), EHCC (extrahepatic cholangiocarcinoma), PHCC (perihilar cholangiocarcinoma), HBV (hepatitis B virus), HCV (hepatitis C virus), LC (liver cirrhosis), NLR (neutrophil/lymphocyte ratio), PNI (prognostic nutrition index), \*: *p*-value < 0.05.

#### *3.5. Technical Failure of CTC Detection in Patients with Biliary Tract Cancers*

CTC detection failed in 7 of the 69 patients with BTC enrolled in this study, all of whom were in the advanced disease stage. In all seven patients, large amounts of amorphous necrotic matrices made it impossible to count the CTCs accurately. Further, all seven patients showed more frequent metastasis (71.4% vs. 23.1%, *p* = 0.007), significantly lower platelet counts (187k vs. 270k, *p* = 0.017), higher NLR (18.7 vs. 4.1, *p* < 0.001), and higher CA19-9 levels (4475 vs. 701, *p* < 0.001) than patients with detectable CTCs (Supplementary Table S3).

#### **4. Discussion**

The estimation of V-CTCs is a potential diagnostic approach for BTC, in addition to evaluating CA 19-9 levels, radiologic imaging, and core or forceps biopsy. Since the CTC markers have low diagnostic accuracy when used independently, we used the CTC markers in combination. Though evaluation using a combination of CTC markers improves accuracy, diagnosing BTC based on CTC markers alone is difficult. Thus, combining the results of this estimation with those obtained by traditional method, such as biopsy and CA19-9 level assessment, can facilitate accurate BTC diagnosis. Furthermore, the V-CTC count was related to OS, especially that of patients with resectable BTC. In multivariate analysis including CA19-9 levels, which is a well-known prognostic factor of BTC, the only significant prognostic factor was a V-CTC count > 50. However, additional studies are needed to support this result.

Efforts towards early diagnosis and prognosis prediction are constantly being made in cancer research. Recently, with the advent of precision medicine, interest in the use of target markers to provide personalized treatment, based on the systemic biology of cancer, has increased. However, there are no minimally invasive methods currently available to accurately diagnose early-stage cancer or predict cancer progression. Recently, studies have been conducted to analyze CTCs, circulating tumor DNA (ctDNA), and extracellular vesicles derived from tumors. Beyond aiding in early cancer diagnosis and prognosis determination, these circulating tumor markers form the basis of many key aspects of precision medicine, including determining actionable targets, monitoring treatment response and resistance, and selecting therapeutics.

There are two important steps in the assessment of CTCs. First, cell enrichment is performed using biological and physical properties. Then, protein-based techniques are used for positive CTC selection. This selection relies on the detection of specific markers by antibodies. However, the expression of epithelial markers such as EpCAM and pan-cytokeratin can be reduced during EMT, which can result in false negatives. Thus, mesenchymal markers, such as N-cadherin and vimentin, should be used [13]. The proportion of true mesenchymal phenotype of CTCs would be very limited because the EMT is a dynamic process when entering the circulation [14,15]. Another way to enrich CTCs is to distinguish CTCs based on their physical properties.

There are many challenges in the assessment of CTCs. The reproducibility of these assessments is difficult since the detected CTC subpopulations may vary across experiments. Additionally, CTCs are large and can be trapped in peripheral blood vessels. CTCs also undergo apoptosis 1–2 h after entering the bloodstream, which may result in low levels of CTCs being detected. Another challenge is the discrimination of CTCs from normal circulating cells. In a study involving patients with benign colonic disease, 11–19% of the patients had epithelial cells that were considered CTCs [16]. For this reason, CTC detection methods usually use epithelial cell adhesion molecules (EpCAMs), which may lead to an underestimation of CTC counts [17].

In this study, we used a platform comprising a centrifugal microfluidic device with a fluid-assisted separation filter membrane (FAST disc) to collect CTCs. The FAST disc enabled label-free CTC isolation from whole blood in a size-selective manner via tangential flow filtration (TFF). This system allowed a clog-free, ultrafast (>3 mL/min) CTC enrichment with gentle pressure drops (~1 kPa) for high viability. Since only gentle pressure was used, cells of various sizes were captured on the membrane, thus facilitating the counting of intact CTCs and allowing for the collection of a large number of tumor cells with high viability. Using vimentin to identify cells in the EMT process increased CTC counts [18,19]. In our study, high counts of CTCs, especially V-CTCs, were found even in patients with BBDs, indicating that the EMT process may also occur in BBDs.

A novel platform to diagnose BTC and predict its prognosis is required for several reasons. Patients with BTC usually present with non-specific symptoms, such as dyspepsia, weight loss, and abdominal discomfort in the early disease stage. A positive BTC diagnosis is usually only made in the later stages of the disease when overt symptoms, such as

jaundice, are present. Imaging by US, CT, and MRI is effective for detecting masses in the biliary tract. However, due to the low incidence of the disease, these methods are not cost-effective for BTC diagnosis. The pathologic diagnosis of BTC is difficult due to various anatomical factors, such as the deep location of the liver, the superficial spread of the bile duct, and the complex blood vessel distribution around the tumor. Although liver core biopsy, forceps biopsy through endoscopic retrograde cholangiopancreatography (ERCP), and brush cytology through ERCP are currently available techniques, they are both invasive and unsuitable for obtaining a sufficiently large cell for pathologic diagnosis because BTC tissues are paucicellular within abundant stroma. However, it is not always possible to obtain tissue samples from primary or metastatic sites. Even if tissue samples are obtained at the time of initial diagnosis, it is not certain that they can be obtained at recurrence or during tumor progression. Therefore, we aimed to develop a method for detecting high-risk groups by screening images during early BTC diagnosis.

In early BTC research, CTCs were detected using CEA-nested RT-PCR in the nucleated cell fraction. The detection rate of CEA-mRNA was 47.8–52.5% (21 of 40 patients with biliary-pancreatic cancers), which was relatively lower than that reported in a recent study [20,21].

Through analysis using the CellSearch system, low counts of CTCs were found in patients with BTC. The detection rate of CTCs in 3 of 13 BTCs is 23.1% [22]. The 12-month survival rates of the patients in the CTC-positive and CTC-negative groups were 25% and 50%, respectively. In another study [23], 88 patients (17%) were positive for CTCs with more than two, which was an independent predictor of survival. Although CTC detection is rare, assessing CTC counts may be useful for predicting the mortality risk of BTC. However, in a recent study evaluating the therapeutic efficacy of cediranib, no relationship between CTC count detection and survival was found. Furthermore, the benefits of cediranib treatment could not be predicted by the combined analysis of baseline and cycle 3 CTC count [24].

A new marker was evaluated for the detection of more CTCs in patients with BTC. In nonconventional CTCs (ncCTCs) lacking epithelial and leukocyte markers, the positive identification of CTCs increased from 19% to 83% [25]. ncCTCs are also correlated with disease-specific survival. Using a novel glycosaminoglycan, SCH45, CTCs were detected in 65 patients with advanced BTC. Furthermore, SCH45-based CTC counts were correlated with the prognosis of patients with BTC receiving chemotherapy [26]. Ninety percent of patients with pancreatic biliary cancers expressed pan-cytokeratin or V-CTCs, which increased the diagnostic accuracy of pancreatic biliary cancers [27].

We excluded seven patients in whom CTCs were not detected from the analysis. In all seven patients, only large amounts of amorphous necrotic matrices were found. These patients had a high ratio of metastasis, high levels of CA19-9, and a higher NLR compared to the other patients. Since CTCs were mostly not detected in patients with advanced cancers, the non-detection of CTCs with extensive necrotic materials may indicate an advanced cancer stage.

There are several limitations to this study. First, this study was conducted with a small number of patients with BTC and BBDs. Second, the patients with BBDs presented with active inflammation. Although blood was drawn immediately after the infection was controlled, the active inflammation may have affected the detection of epithelial cells in circulation. Third, we did not obtain follow-up blood samples to assess the dynamics of CTCs during therapy.

Many researchers have worked to identify the best biomarkers for diagnosing earlystage BTC. However, this is made difficult by the anatomical and histological characteristics of BTC. The combined assessment of circulating tumor markers and CA 19-9 levels, radiological imaging, and core or forceps biopsy may be helpful in discriminating CTCs between early-stage BTC and BBDs, and in determining future prognosis in patients with resectable BTCs. Although there are still limitations to early BTC diagnosis using V-CTCs, further studies may provide a framework for realizing precision medicine by conducting liquid biopsies using CTCs in a complementary manner.

**Supplementary Materials:** The following are available online at https://www.mdpi.com/article/10 .3390/jcm10194435/s1, Table S1: prognostic value according to V-CTC level in subgroup analysis, Table S2: comparison of PFS according to CTC marker and CA19-9 in 35 patients who underwent curative surgery or palliative chemotherapy, Table S3: Baseline characteristics between BTC patients with CTC detected and those without CTC detection, Figure S1: ROC curves of CTC, V-CTC, and VCR.

**Author Contributions:** S.Y.H. and S.H.P. contributed equally to the work. Study concept and design; D.U.K.; Data acquisition; S.Y.H., S.H.P., A.J., S.J.L.; Data analysis and interpretation; S.Y.H., S.H.P., H.S.K.; Drafting of the manuscript; S.Y.H., S.H.P., H.I.S., D.U.K.; Critical revision of the manuscript; G.H.K., D.U.K. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research was supported by a grant of the Korea Health Technology R&D Project through the Korea Health Industry Development Institute (KHIDI), funded by the Ministry of Health & Welfare, Republic of Korea (grant number: HI19C0521).

**Institutional Review Board Statement:** Approved by Institutional Review Board of Pusan National University Hospital (H-H-1801-020-062). The clinical research information service (CRIS) approved the study (KCT0003511).

**Informed Consent Statement:** Informed consent was obtained from all subjects involved in the study.

**Data Availability Statement:** All relevant data contained within the article and supplementary materials.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**

