A combination of circulating tumor cells and CA199 improves the diagnosis of pancreatic cancer

Abstract Background Early diagnosis of pancreatic ductal adenocarcinoma (PDAC) is difficult due to the lack of effective screening tests. CA199, the standard biomarker for PDAC management, is not sufficiently reliable for early diagnosis. This prospective study aimed to evaluate whether circulating tumor cells (CTCs) could complement or perform better than CA199 in determining PDAC. Methods A total of 168 blood samples were collected from 80 patients with PDAC, 32 patients with acute pancreatitis, 22 patients with benign pancreatic masses, and 34 healthy donors. CTCs were detected by a novel system combining negative enrichment with immunostaining and fluorescence in situ hybridization (NE‐imFISH). Next, ROC curves and AUC analyses were conducted to assess diagnostic abilities of CA199, CTCs, and the combination of the two biomarkers in PDAC. Results CTCs were stained as CD45–/DAPI+/CEP8 ≥3. With 2 CTCs/3.2 ml as the cut‐off value, the sensitivity/specificity of the CTC number was 0.76/0.94, which was comparable to that of CA199 (0.78/0.83; Delong test p = 0.3360). Improved performance was achieved through a logistic regression model integrating CA199 and CTC number (AUCCTC+CA199 = 0.95, AUCCA199 = 0.80, AUCCTC number = 0.85; Delong test p vs. CA199 < 0.0001 and p vs. CTC number = 0.0002). CTC subtype was inferior to CTC number as a diagnostic marker (AUCCTC subtype = 0.73; Delong test p vs. CTC number < 0.0001). Conclusion The dual‐marker panel consisting of CA199 and CTC number can significantly improve upon the diagnostic performance of CA199 alone, highlighting the promising clinical utilization as an effective strategy for PDAC surveillance.

with similar signals to surrounding tissues. 6,7 Although Endoscopic ultrasound-guided fine-needle aspiration (EUS-FNA) can provide preoperative biopsy, it is an invasive method accompanied by various complications, including pancreatitis, hemorrhage, infection, perforation, and needle tract seeding. 8,9 Thus, a minimally invasive, early diagnostic test with high sensitivity and specificity is highly desirable. Currently, Carbohydrate antigen 199 (CA199) is the sole serum biomarker that has received approval from the US Food and Drug Administration (FDA) for routine management of PDAC. This standard marker has 60%-90% sensitivity and 68%-91% specificity for diagnosis, with false-negative results observed in Lewis antigen-negative patients and false-positive results caused by benign diseases, such as pancreatitis and cholangitis. 10 Because of its modest diagnostic performance, there is a vital necessity to identify effective biomarkers to either complement or perform better than CA199.
Circulating tumor cells (CTCs) were first discovered in the peripheral circulation of a metastatic breast cancer patient by Prof.
Ashworth in 1869. 11 As cancer cells that detach from primary or metastatic tumors and spread through the bloodstream, CTCs may lead to tumor metastasis and relapse. 4,12 Owing to the technical difficulties, it took almost 150 years to establish a method to isolate 1 CTC from 5 billion erythrocytes and 10 million leukocytes present in the 1 ml of whole blood. 3,4 CellSearch ® is the only assay approved by FDA in 2004 for CTC detection in metastatic breast, prostate, and colorectal cancer, but the detection rate is only 21%-45% in PDAC patients. 3,[13][14][15][16][17][18][19] The low efficiency of this platform is attributed to the dependence on epithelial cell adhesion molecule (EpCAM) and cytokeratin (CK) for CTC enumeration and identification, which fails to detect highly aggressive cells that downregulate the expression of CK and EpCAM during the process of epithelial-mesenchymal transition (EMT). 20 In addition, non-tumor epithelial cells derived from inflammation, trauma, and benign epithelial hyperplasia may result in false-positive results. 21 As the CellSearch ® platform cannot meet clinical requirements, an EpCAM-independent system integrating negative enrichment with immunostaining and fluorescence in situ hybridization (NE-imFISH) has been introduced as an alternative option. 13 The enrichment process of this system involves exclusion of leukocytes by anti-CD45 antibody and erythrocytes by hemolysis, followed by CTC identification by centromere probe 8 (CEP8) staining, as chromosome 8 polysome is the most common form of chromosome instability observed in PDAC. [22][23][24][25] In addition to existence in advanced and metastatic stages, CTCs are also widely present in PDAC patients at early stages, suggesting its potential value as a diagnostic tool. 26

| CTC detection and CA199 measurement
The NE-imFISH approach developed by Cyttel ® was employed to identify CTCs. The enrichment and identification process were based on the standard protocol reported previously. 13 In short, a 3.2 ml blood sample combined with CS1 buffer (Cyttel Biosciences INC.) was centrifuged at 650 g for 5 min to remove plasma. Then, erythrocytes were lysed with CS2 (Cyttel) for 8 min and removed by another round of centrifugation (650 g for 5 min). After discarding the supernatants and re-suspending the residual cells with CS1 buffer, CS3 buffer (Cyttel) and magnetic beads (Cyttel) were added to remove most of the leukocytes by magnetic separation and gradient centrifugation (300 g for 5 min). The sedimented cells were fixed on the slides and dried at room temperature for CTC identification.
Afterward, the slides were dehydrated in ethanol and hybridized with CEP8 (Cyttel). Subsequently, slides were incubated with anti-CD45 conjugated to Alexa Fluor 594 (Cyttel) and mounted with mounting media containing DAPI (Vector Laboratories). The results were read by fluorescence microscope (Nikon Ci). The whole detection procedure was conducted blindly by two trained authors to ensure the reliability of the detection result. Any disagreement was resolved through discussion with a senior technician. CTCs were characterized as CD45−/ DAPI+/CEP8 ≥3.
CA199 in blood samples was measured using a chemiluminescence assay kit (Roche). Levels <37 U/ml were considered as negative.

| Sample size calculation and statistical analysis
This study utilized a non-inferiority design. On the basis of reviewing the literature on CA199 and CTCs identified by NE-imFISH, we estimated the sensitivity/specificity of CA199 and CTCs were 75%/80% and 83%/86%, respectively. 21,23,32-34 As we only focused on the non-inferiority rather than the equivalence, the sample size calculation was based on a one-tailed alpha of 0.025 with a power of 0.8. two-sided p-value of <0.05 was considered statistically significant.
All graphs were plotted utilizing GraphPad Prism 8.4 (GraphPad Software) and the "ggalluvial" R package.

| Identification and classification of CTCs
Cells enriched from blood samples were identified and classified according to the leukocyte marker (CD45), cell nucleus marker (DAPI), and chromosome ploidy (CEP8). Negative cells were stained as CD45+/DAPI+/CEP ≥1 ( Figure 1A) and CD45-/DAPI+/CEP8 ≤2 ( Figure 1B,C), while CTCs were characterized by CD45-/DAPI+/ CEP8 ≥3. In our study, cells with more than five CEP8 signals were collectively referred to as multiploidy CTCs because of their rarity. Thus, detected cancer cells were categorized as triploid CTCs for three copies of chromosome 8 ( Figure 1D), tetraploid for 4 ( Figure 1E), and multiploid for five or more ( Figure 1F). As different kinds of CTCs may present in one sample, CTC subtype is used to indicate the number of karyotypes.

| Distributions of CA199 and CTCs
The distributions of CA199 and CTCs in different groups were first explored. In the control group, CTC subtype and CTC number were not significantly different among patients with acute pancreatitis, patients with benign pancreatic masses, and health controls ( Figure 2C,E). As for CA199, statistical significance was only found between the comparison between health controls and acute pancreatitis (p = 0.0002; Figure 2A). Similarly, CA199, CTC subtype, and CTC number could not distinguish between early-stage (stage Ⅰ-Ⅱ) and advanced-stage (stage Ⅲ-Ⅳ) patients ( Figure 2B,D,F). Given the within-group similarities, in the subsequent analyses, we only divided enrolled participants into PDAC group and control group.
As illustrated in Figure Figure 2H).

| Comparison of the diagnostic performance among CTCs, CA199, and the dual-marker panel
ROC curves and AUC analyses were conducted to assess diagnostic abilities in distinguishing PDAC cases from controls (Table 1;

| Subgroup analysis
To investigate why CTC number was complementary to CA199, we evaluated the performances of 3 individual markers in different subgroups.
(1) Subgroup analysis by CA199 ( Table 3). One of the main shortcomings of CA199 as the biomarker for PDAC is the limited sensitivity, such as false-negative results observed in 5%-10% of people with Lewis antigen-negative genotype. 35 Because of this, we divided our participants into CA199 low (CA199 < 37 U/ml) and high (CA199 ≥ 37 U/ml) groups. In the low expression group, 18 PDAC patients were falsely diagnosed as negative by CA199. CTC number could correctly identify 14 of them while maintaining high specificity (sensitivity = 0.78; specificity = 0.95; Figure 4D). Falsepositive results were observed in 15 controls, including 11 patients with acute pancreatitis, of which 14 could be correctly classified by CTC number (sensitivity = 0.76; specificity = 0.93; Figure 4A).
Therefore, CTC number could complement CA199 in determining the diagnosis of PDAC.
In addition, only five patients showed false-positive results of CTC number, of which 4 patients could be correctly differentiated by CA199 (sensitivity = 0.77; specificity = 0.80; Figure 4C). These findings demonstrated that CA199 could also complement CTCs in the diagnosis of PDAC.

| DISCUSS ION
PDAC is the deadliest cancer with a dismal prognosis, which can be improved through early diagnosis. To date, CA199 is the routinely used serum biomarker for PDAC management with several unfavorable drawbacks. 32,33 Thus, there is an urgent need to establish F I G U R E 2 Distributions of CA199 and CTCs in PDAC patients and controls. (A, C, E) The distributions of CA199, CTC number, and CTC subtype among acute pancreatitis (AP), benign pancreatic masses (BPM), and health control (HC). (B, D, F) The distributions of CA199, CTC number, and CTC subtype between early-stage (stageⅠ-Ⅱ) and late-stage (stage Ⅲ-Ⅳ) PDAC patients. (G) The karyotype distribution of PDAC patients was significantly different from that of controls. (H) A significant difference was observed in CTC subtype distribution between PDAC patients and controls a more accurate diagnostic system. As a non-invasive approach harboring the great potential to compensate for the shortcomings of CA199, CTC detection, also known as liquid biopsy, has remarkable value in diagnosis, prognosis, as well as evaluation of tumor metastasis and relapse. 20 CTCs are cancer cells shed into the peripheral bloodstream either from primary or from metastatic tumors. 12  As far as we know, this study is the first to combine CA199 and CTC identified by the NE-imFISH method for the diagnosis of PDAC.
The sensitivity and specificity of CA199 in our study reached 0.80 and 0.78. No significant difference was found in the comparison with AUCs between CTC number and CA199 (Delong test p = 0.3360), indicating that CTCs are comparable to CA199 as a diagnostic TA B L E 1 Diagnostic performances of CTC, CA199, and the combination of the two markers Han et al. pointed out that CTC subtype could be used as a diagnostic marker with a sensitivity of 54% and a specificity of 85% using 2 subtypes/3.2 ml as the cut-off value. 23 In this study, the optimal cut-off value of CTC subtype was the same, achieving 40% sensi- diseases. As our CTC detection approach was based on the aneuploidy of chromosome 8, the cells identified as CTCs in the control group were most probably cells with abnormal metabolism instead of real tumor cells. Nevertheless, the frequency of aneuploid cells was much lower in the control group.
In conclusion, the current data support CTC number identified by the NE-imFISH method is a feasible biomarker comparable to CA199 in detecting PDAC. The dual-marker panel consisting of CA199 and CTC number can significantly improve upon the diagnostic performance of CA199 alone, highlighting the promising clinical utilization as a convenient and effective strategy for PDAC surveillance. However, our study was conducted in a single center with limited sample size. Further clinical studies with more participants are warranted to verify our findings.

ACK N OWLED G M ENTS
We are thankful to the reviewers for their useful thoughts and suggestions.

CO N FLI C T O F I NTE R E S T
The authors have no conflicts of interests to declare.

AUTH O R CO NTR I B UTI O N S
Xiaodong Tan and Junliang Chen conceived and planned this experi- Junliang Chen took the lead in writing the manuscript. All authors provided vital feedback to shape the final version of the manuscript.

DATA AVA I L A B I L I T Y S TAT E M E N T
The data that support the findings of this study are available on request from the corresponding author, XD Tan.