Circulating natural antibodies against 3’-sialyllactose complement the diagnostic performance of CA19-9 for the early detection of pancreatic ductal adenocarcinoma

Abstract

BACKGROUND:

Pancreatic ductal adenocarcinoma is a devastating malignancy with an extremely poor prognosis. Although the most widely used biomarker for pancreatic cancer is carbohydrate antigen CA19-9, it is elevated mainly in the late stage of pancreatic cancer. Some serum natural antibodies against carbohydrates have been shown to be possible diagnostic markers for cancer.

OBJECTIVE:

This study was conducted to determine whether the level of natural antibodies against carbohydrates fluctuates in pancreatic ductal adenocarcinoma.

METHODS:

Serum from pancreatic cancer subjects ( $n=$ 55) and 43 subjects free of malignant disease were studied. The contents of natural antibodies against sialyl glycans and CA19-9 in serum were determined by enzyme-linked immunosorbent assay.

RESULTS:

The level of serum anti-3’-sialyllactose antibodies in pancreatic cancer subjects was significantly lower than that in healthy controls. In contrast, the amounts of serum antibodies against other sialyl glycans were comparable between the two groups. Concentration of serum anti-3’-sialyllactose IgG provided excellent AUC of 0.86, with sensitivity 82%, specificity 81%, and accuracy 82%. The combination of serum anti-3’-sialyllactose IgG with CA19-9 improved the sensitivity of pancreatic cancer detection at an early stage.

CONCLUSIONS:

Natural antibodies against 3’-sialyllactose constitute a promising biomarker for pancreatic cancer detection. The measurement of serum anti-3’-sialyllactose antibodies could play a supportive role in diagnostics and complement the performance of CA19-9 for the early detection of pancreatic ductal adenocarcinoma.

1. Introduction

Pancreatic ductal adenocarcinoma, which accounts for more than 90% of all pancreatic tumors, is a devastating malignancy with an extremely poor prognosis. Because of the asymptomatic nature of its early stage, and coupled with inadequate methods for early detection, most of the patients present with locally advanced and inoperable disease at the time of diagnosis [1]. Therefore, there is an urgent need to develop accurate markers of pre-invasive pancreatic neoplasms to facilitate prediction of cancer risk and to aid in diagnosis of pancreatic cancer at an early stage [2]. The most widely used serum biomarker for pancreatic cancer is carbohydrate antigen 19-9 (CA19-9), sialyl Lewis ${}^a$ , found on the surface of proteins including mucins. In addition, elevated expressions of sialyl Lewis ${}^c$ (DUPAN-2), sialyl Lewis ${}^x$ (SLX), and sialyl Tn antigen (STN) have been shown in pancreatic cancer [3, 4]. We previously developed a highly sensitive and efficient method capable of accurate identification of $O$ -glycans on mucins [5, 6], and revealed that the expression profiles of sialylated polylactosaminyl $O$ -glycans are significantly different between moderately and poorly differentiated pancreatic cancer cell lines [7]. Moreover, we demonstrated that the expression of NeuAc $\alpha$ 2-3Gal $\beta$ 1-4Glc (3’-sialyllactose) residue is observed in human pancreatic cancer cells and its level is augmented in metastatic cells [8]. These results suggest that the aberrant glycosylation occurs also in pancreatic cancer, and many oligosaccharide epitopes, particularly sialyl glycans, constitute tumor-associated antigens.

Specific antibodies are commonly monitored for the diagnosis of diseases such as HIV, malaria, autoimmune diseases, and cancers. Immune system success is largely dependent on the presence of several components and well-equipped repertoire of antibodies that are the first line of defense against pathogens and eliminate dysfunctional or malignant cells [9]. Natural antibodies are present in the serum of individuals in the absence of known antigenic stimulation. Human natural antibodies against Gal $\alpha$ 1-3Gal $\beta$ 1-4GlcNAc ( $\alpha$ Gal) are the most abundant (about 1%) in immunoglobulins throughout the life of all individuals, and mediate the rejection of pig xenograft organs by binding $\alpha$ Gal epitopes on the pig cells, inducing complement- and antibody-dependent cell destruction [10].

In contrast, some natural antibodies are supposed to play a role in surveillance on appearance of aberrant cancer antigens, including carbohydrates [11]. A hexa-saccharide, Globo H, is highly expressed on a various type of cancers, especially breast, prostate, and lung cancers. Wang et al. presented that the amounts of serum antibodies against Globo H from breast cancer patients were significantly higher than those of healthy subjects [12]. CA125 is a high molecular weight, heavily glycosylated transmembrane mucin 16, one of the major carriers of CA19-9 and SLX, and is overexpressed in 80% of ovarian cancer. A previous study showed that the serum CA125 level is increased in ovarian cancer; however, the contents of some anti-sialyl glycan antibodies in serum are significantly decreased in ovarian cancer compared with healthy controls [13].

Since the repertoire of glycan expressed on cell surface undergoes structural alterations with the onset of cancer, the corresponding natural antibody levels can be changed [14]. Although previous studies identified some natural antibodies as diagnostic markers for cancer such as breast, ovarian, and prostate cancers, the involvement of natural antibodies in the detection of pancreatic cancer remains unclear. The current study was conducted to determine whether the level of natural antibodies against pancreatic cancer-associated sialyl oligosaccharides fluctuates. In addition, we examined whether serum anti-3’-sialyllactose antibodies as a biomarker could increase the diagnostic capacity of CA19-9 to detect the early stage of pancreatic cancer.

2. Materials and methods

2.1 Serum

Human sera from healthy ( $n=$ 43) and pancreatic ductal adenocarcinoma ( $n=$ 55) individuals were obtained from PromedDx (Norton, MA, USA) as shown in Table 1. Sera from cancer subjects were collected at time of diagnosis, before initiation of treatment. Sera from male animals, 7-yr monkey (Macaca), and 8-wk hamster (Syrian), rabbit (NZW), guinea pig (Hartley), rat (F344), and mouse (Balb/c) were purchased from Japan SLC (Shizuoka, Japan).

Table 1

Demographic data of the subjects

	Number	Male/female	Mean age (range)
Healthy	43	20/23	56.2 (40–75)
Pancreatic ductal adenocarcinoma	55	21/34	69.7 (25–96)
Stage I	15	6/9	62.5 (25–88)
Stage II	15	4/11	72.7 (57–96)
Stage III	11	3/8	71.8 (38–91)
Stage IV	14	8/6	72.6 (53–84)

2.2 Human anti-3’-sialyllactose IgG

Human serum IgG from healthy subjects was purchased from Wako Pure Chemical Industries (Osaka, Japan). Biotinylated 3’-sialyllactose conjugated with polyacrylamide (PAA) was obtained from GlycoTech (Gaithersburg, MD, USA). To prepare human anti-3’-sialyllactose IgG, biotinylated 3’-sialyllactose-PAA and streptavidin-tagged agarose (Thermo Scientific, Waltham, MA, USA) in PBS (-) were incubated for 12 h at 4 ${}^{\circ }$ C, and the resin was packed into a column. Next, human IgG in PBS (-) containing 0.1% Tween 20 was applied to the column, incubated for 12 h at 4 ${}^{\circ }$ C, and then the flow throw as an unbound fraction was obtained. After washing with PBS (-), a bound fraction was prepared using 3 mM of 3’-sialyllactose (Tokyo Chemical Industry, Tokyo, Japan). The eluent was concentrated using Amicon Ultra (Merck, Darmstadt, Germany) and protein concentration was determined.

2.3 Enzyme-linked immunosorbent assay (ELISA)

The wells of 96-well plates (MaxiSorp, Nunc, Denmark) were coated with 10 $\mu$ g/ml glycan polymer (GlycoTech) for 16 h at room temperature. After blocking with 5% BSA and 1% Triton X-100, the diluted sera were added to the wells and incubated for 1 h at room temperature. The wells were washed five times and incubated with HRP-conjugated goat anti-human IgG (SeraCare Life Sciences, Milford, MA, USA) for 1 h at room temperature. After washing, peroxidase substrate and stop solution (SeraCare Life Sciences) were added to the wells, and then the optical density (OD) at 450 nm was measured. The final OD ${}_{450}$ was determined by subtracting the negative control OD ${}_{450}$ from subject sample reading. CA19-9 ELISA kit was purchased from Abnova (Taipei, Taiwan), and used in accordance with the supplier’s protocol.

2.4 Glycan array

Glycan polymer (1 $\mu$ g) was spotted on an epoxy-coated glass slide (Schott AG, Mainz, Germany) and blocked with Tris-HCl buffer (25 mM Tris-HCl [pH 7.4], 0.8% NaCl, 1% Triton-X100, and 4% BSA). Sample (80 $\mu$ l) in the probing buffer (25 mM Tris-HCl [pH 7.4], 0.8% NaCl, 1% Triton-X100, 1 mM MnCl ${}_2$ , and 1 mM CaCl ${}_2$ ) followed by Cy3-conjugated appropriate secondary antibody was applied to the array, and then fluorescent images were acquired using an Ettan DIGE imager (GE Healthcare, Buckinghamshire, England).

2.5 Statistical analysis

All statistical calculations were carried out using Prism and InStat software (GraphPad software, La Jolla, CA, USA). The diagnostic ability of serum marker to distinguish between healthy control and pancreatic cancer was determined through cutoffs, derived through analysis of the Area Under the Curve (AUC) using Receiver Operating Characteristic (ROC) curve analysis. Optimal cutoff values for biomarkers were identified using Youden’s J statistic. Linear discriminant analysis was applied to the data using R software (Ver. 3.5.2) with packages MASS. Cross-validation was estimated using a limited number of data sample by the leave-one-out-cross-validation (LOOCV) method. Values are expressed as mean $\pm$ S.D. Student’s $t$ -test was used to evaluate the statistical significance of differences between groups. A $p$ -value of less than 0.05 was considered significant.

Figure 1.

Diagnostic performance of serum anti-3’-sialyllactose IgG for pancreatic cancer. (A) Both the unbound and bound fractions prepared using a 3’-sialyllactose affinity column were applied to two arrays of $\alpha$ Gal and 3’-sialyllactose polymers, incubated with Cy3-conjugated anti-human IgG antibody, and the bindings were detected using a fluorescence-assisted scanner (left). The standard curve of anti-3’-sialyllactose IgG was analyzed by ELISA (right). (B) The concentrations of serum anti-3’-sialyllactose IgG from 43 healthy and 55 pancreatic cancer subjects was determined by ELISA. (C) Distribution of serum anti-3’-sialyllactose IgG from subjects with different stages of pancreatic cancer. Scatter plots of the level of anti-3’-sialyllactose IgG from healthy ( $n=$ 43), stage I ( $n=$ 15), stage II ( $n=$ 15), stage III ( $n=$ 11), and stage IV ( $n=$ 14) pancreatic cancer subjects are shown. (D) Serum level of anti-3’-sialyllactose IgG was subjected to analysis by ROC curve analysis for pancreatic cancer detection. The AUC of serum anti-3’-sialyllactose IgG was 0.86 (95% confidence interval 0.78–0.94, $p<$ 0.0001) for differentiation between subjects with pancreatic cancer and healthy control.

Figure 2.

Combined diagnosis of pancreatic cancer at an early stage by serum anti-3’-sialyllactose IgG and CA19-9. (A) The concentration of serum CA19-9 from 43 healthy and 55 pancreatic cancer subjects was determined by commercially available ELISA. (B) Distribution of serum CA19-9 from subjects with different stages of pancreatic cancer. Scatter plots of the level of CA19-9 from healthy ( $n=$ 43), stage I ( $n=$ 15), stage II ( $n=$ 15), stage III ( $n=$ 11), and stage IV ( $n=$ 14) pancreatic cancer subjects are shown. (C) Serum level of CA19-9 was subjected to ROC curve analysis for pancreatic cancer detection. The AUC of serum CA19-9 was 0.92 (95% confidence interval 0.86–0.99, $p<$ 0.0001) for differentiation between subjects with pancreatic cancer and healthy control. (D) The composite biomarker to diagnose having pancreatic cancer had a sensitivity of 95%, a specificity of 95%, and accuracy of 95% with an AUC of 0.97 (95% confidence interval 0.93–1.0, $p<$ 0.0001).

3. Results

3.1 Natural antibodies against pancreatic cancer-associated sialyl glycans

Glycan arrays have emerged as powerful tools for evaluating the interactions between carbohydrates and lectin, whole cells, viruses, or natural antibodies. Using glycan microarrays, previous studies have shown that numerous natural antibodies against oligosaccharides are included in human serum and IgG [15, 16]. To investigate whether the level of natural antibodies against pancreatic cancer-associated sialyl glycans fluctuates, semi-quantitative analysis of anti-sialyl glycan IgG in serum from healthy and pancreatic cancer subjects was performed by ELISA (Supplementary Fig. 1). The concentrations of HRP-conjugated goat anti-human IgG as a detecting antibody were determined necessary to achieve greater linearity from 1:800 to 1:200 dilution of serum (data not shown). Semi-quantitative analysis showed that the content of serum anti-3’-sialyllactose IgG in pancreatic cancer subjects was significantly lower than that in healthy controls, whereas the levels of antibodies against CA19-9, DUPAN-2, SLX, and STN were comparable between the healthy and pancreatic cancer groups(Supplementary Fig. 2).

3.2 Diagnostic potential of natural antibodies against 3’-sialyllactose

Previous studies have suggested that human antibodies that bind to 3’-sialyllactose were included in purified IgG [16]. To quantify natural antibodies against 3’-sialyllactose, we first isolated human anti-3’-sialyllactose IgG as a standard using 3’-sialyllactose affinity columns. After purification by affinity, both the unbound and bound fractions were applied to two arrays of $\alpha$ Gal and 3’-sialyllactose polymers, incubated with Cy3-conjugated anti-human IgG antibody, and the bindings were detected using a fluorescence-assisted scanner. As expected, the bound fraction did not totally include anti- $\alpha$ Gal IgG, the most abundant natural antibody, suggesting that anti-3’-sialyllactose IgG was successfully purified (Fig. 1A, left) and was estimated as 0.1% of total IgG (data not shown). The standard curve using the obtained anti-3’-sialyllactose IgG exhibited high linearity in the range of 15.6–250 ng/ml in ELISA (Fig. 1A, right). Under the same experimental conditions, the concentration of serum anti-3’-sialyllactose IgG from 43 healthy and 55 pancreatic cancer subjects was determined. As shown in Fig. 1B, the level of serum anti-3’-sialyllactose IgG in pancreatic cancer subjects was remarkably lower than that in healthy controls. Stage-dependent decreases in the content of anti-3’-sialyllactose IgG were also observed (Fig. 1C). In contrast, the level of anti- $\alpha$ Gal IgG was comparable between both groups (data not shown). The discriminatory capacity of serum levels of anti-3’-sialyllactose IgG to detect pancreatic cancer was profiled by ROC curve analysis. As shown in Fig. 1D, the concentration of serum anti-3’-sialyllactose IgG provided an excellent AUC of 0.86 (95% confidence interval 0.78–0.94, $p<$ 0.0001). At a serum level cutoff of 53.35 $\mu$ g/ml, anti-3’-sialyllactose IgG had a sensitivity of 82%, a specificity of 81%, and accuracy of 82%. In addition, the AUCs for Stage I and Stage IV pancreatic cancer were 0.75 and 0.88, respectively(Supplementary Fig. 3).

3.3 Combined diagnosis for pancreatic cancer at an early stage

The most widely used biomarker for pancreatic cancer is CA19-9; however, CA19-9 is elevated mainly in the late-stage of pancreatic cancer [17]. The results described above led us to raise the possibility that an integrated approach (a combination of serum anti-3’-sialyllactose IgG and CA19-9 tests) would improve the detection of early-stage pancreatic cancer. Serum CA19-9 values from sera of the same individuals as for the anti-3’-sialyllactose IgG test were calculated using the commercially available ELISA kit. Consistent with previous studies, the concentration of serum CA19-9 in pancreatic cancer subjects was significantly elevated compared with healthy controls (Fig. 2A). In addition, the stage-dependent increases in the CA19-9 content were observed (Fig. 2B). The CA19-9 value exhibited an AUC of 0.92 (95% confidence interval 0.86–0.99, $p<$ 0.0001, Fig. 2C). At a serum level cutoff of 21.23 U/ml, CA19-9 had a sensitivity of 89%, a specificity of 95%, and accuracy of 92%. In addition, the AUCs for Stage I and Stage IV pancreatic cancer were 0.86 and 0.95, respectively(Supplementary Fig. 3).

We next explored the possibility that combining anti-3’-sialyllactose IgG and CA19-9 tests would improve diagnostic performance for the early stage of pancreatic cancer. When all data from healthy and pancreatic cancer subjects were plotted against each other, the values were not totally correlated(Supplementary Fig. 4). Similarly, data from pancreatic cancer subjects at stage I were plotted against anti-3’-sialyllactose IgG and CA19-9, and each cutoff value was applied. As shown in Supplementary Figs 5 and 6 out of 15 subjects were diagnosed as having cancer by both tests (solid circle), and 3 or 5 were determined as the presence of cancer by only anti-3’-sialyllactose IgG or CA19-9 test, respectively (open circle). However, 1 out of 15 pancreatic cancer subjects at stage I was diagnosed as a non-cancerous subject by both tests (closed square). In addition, linear discriminant analysis produced the following predictive equations for pancreatic cancer: Composite biomarker $=-$ 18.054 $\times$ (1/Conc ${}_{\text{anti-3’-sialyllactose IgG}}$ ) $-$ 0.014 $\times$ Conc ${}_{\text{CA19-9}}+$ 1.724; where Conc ${}_{\text{anti-3’-sialyllactose IgG}}$ and Conc ${}_{\text{CA19-9}}$ represent the serum concentrations of anti-3’-sialyllactose IgG ( $\mu$ g/ml) and CA19-9 (U/ml), respectively. The cross-validated classification showed that overall 87% of the subjects were correctly classified. When the cutoff value of 0.9 was applied to this equation, the composite biomarker to diagnose having pancreatic cancer had a sensitivity of 95%, a specificity of 95%, and accuracy of 95% with an excellent AUC of 0.97 (95% confidence interval 0.93–1.0, $p<$ 0.0001) (Fig. 2D).

4. Discussion

Pancreatic cancer is one of the most aggressive and lethal malignancies worldwide, with a very poor prognosis and a five-year survival rate of less than 8%. This dismal outcome is largely due to delayed diagnosis, early distant metastasis, and resistance to conventional chemotherapies [18]. CA19-9 is a serum marker measured in the clinical management of pancreatic cancer; however, CA19-9 is not overexpressed in the early stage of pancreatic cancer [17]. In addition, CA19-9 expression is not elevated in patients who lack the Lewis antigen, even if they have a large pancreatic tumor [19]. Accordingly, the guidelines of the American Society of Clinical Oncology do not recommend the measurement of CA19-9 as a screening test for gastrointestinal cancers, particularly pancreatic cancer [20]. On the other hand, aberrant expression of $O$ -linked glycans, including DUPAN-2, SLX, STN, and 3’-sialyllactose, is also associated with progression of pancreatic cancer [4, 5].

In the present study, the content of natural antibodies against the sialyl glycans were measured and found that the level of serum anti-3’-sialyllactose IgG in pancreatic cancer subjects was markedly reduced compared with healthy controls (Fig. 1B). In addition, we demonstrated that the concentration of serum anti-3’-sialyllactose IgG serves as a novel diagnostic marker for pancreatic cancer (Fig. 1D). The CA19-9 test exhibited high sensitivity; however, the cancer of 4 out of 15 pancreatic cancer subjects at stage I failed to be detected in this study. To more sensitively detect the early stage of pancreatic cancer, we examined a combination assay based on the determination of anti-3’-sialyllactose IgG and CA19-9. As expected, the combination assay enabled detection of stage I pancreatic cancer with high sensitivity (93.3%), and the anti-3’-sialyllactose IgG test detected 3 out of 4 subjects with clinical stage I pancreatic cancer in which no increase in CA19-9 was observed (Fig. 2D). In addition, 1 out of 15 pancreatic cancer subjects at stage II was predicated to be non-cancerous in the CA19-9 test, but properly diagnosed as a pancreatic cancer subject in the 3’-sialyllactose IgG test (data not shown). Thus, these results indicated that the combination of anti-3’-sialyllactose IgG with CA19-9 test increases the sensitivity of pancreatic cancer detection at an early stage.

Serum carbohydrates have been studied extensively as potential cancer biomarkers; however, many are restricted to the tumor site and levels that are often proportional to the size of tumor, which is not ideal for early detection. In contrast, a small tumor can potentially give rise to large changes in antibody levels, and these antibodies can be easily detected in biofluids such as serum [14]. We have previously proposed novel methods that facilitate the discovery of carbohydrate-binding molecules as disease markers using glycan arrays [21]. Since previous studies have shown that numerous natural antibodies against oligosaccharides are included in human serum, we determined the levels of serum natural antibodies against 50 species of glycans in addition to sialyl glycans used in this study using glycan arrays, and concluded that the concentration of serum anti-3’-sialyllactose antibodies can robustly distinguish pancreatic cancer subjects from healthy subjects (data not shown). In addition, the diagnostic capacity of anti-3’-sialyllactose IgG to detect pancreatic cancer was superior to that of IgM class (data not shown). Since serum CA19-9 is also elevated in benign diseases of the pancreas and in other malignancies of gastrointestinal tract, we preliminarily evaluated the specificity of anti-3’-sialyllactose antibodies using sera from several cancers and chronic pancreatitis, and obtained prospective results on the clinical availability as a pancreatic cancer marker (data not shown). Collectively, natural antibodies as a serum biomarker for pancreatic cancer are presented here for the first time. In addition, the measurement of serum anti-3’-sialyllactose IgG could play a supportive role in diagnostics, complementing the defects in CA19-9 determination. To apply current evidence to the diagnosis for pancreatic cancer, the findings from large-scale clinical trials of interventions would be the most relevant.

Natural antibodies are supposed to play a role in surveillance on appearance of aberrant cancer antigen [14]. $N$ -glycolylneuraminic acid (NeuGc) ganglioside is not naturally expressed, because the enzymes involved in the biosynthesis of NeuGc are deficient in humans. However, NeuGc-containing foods such as animal meat and milk are taken in and incorporated into gangliosides as NeuGc gangliosides in human body. Previous studies showed that NeuGc gangliosides are frequently overexpressed in human tumors, including non-small cell lung cancer [22, 23]. Quantitative analyses revealed that the level of serum anti-NeuGc ganglioside antibodies from non-small cell lung cancer patients was significantly lowered compared with healthy donors [24]. In addition, anti-NeuGc ganglioside antibodies present in healthy human sera recognize and kill NeuGc ganglioside-expressing tumor cells in complement-dependent and-independent manners [24]. These results raise the possibility of anti-tumor activity by human anti-3’-sialyllactose antibodies. The concentration of anti-3’-sialyllactose IgG in serum from pancreatic cancer subjects was remarkably reduced compared with healthy subjects, whereas other natural antibodies against pancreatic cancer-associated sialyl glycans were notchanged(Supplementary Fig. 2). Since tumor cells express the transmembrane and secreted forms of mucins that are associated with CA19-9, DUPAN-2, SLX, and STN [4], a specific decrease in serum natural antibodies against 3’-sialyllactose could be explained in part by the adhesion of anti-3’-sialyllactose antibodies infiltrated from the circulation to their antigens on the plasma membrane but not the secreted molecules of cancer cells. We next examined the species difference in the amount of serum anti-3’-sialyllactose IgG, human, monkey, hamster, rabbit, guinea pig, rat, and mouse sera were subjected to glycan arrays. Interestingly, the results suggested that serum anti-3’-sialyllactose antibodies are inherent in human beings(Supplementary Fig. 6). Further studies to clarify the physiological significance of serum anti-3’-sialyllactose antibodies in the human innate immune system are now under investigation.

In conclusion, we determined the amounts of natural antibodies that exist against numerous oligosaccharides and demonstrated that serum anti-3’-sialyllactose IgG is a promising biomarker for pancreatic cancer detection. In addition, we observed that the level of serum anti-3’-sialyllactose IgG could improve the diagnostic performance of CA19-9 for the early stage of pancreatic cancer. Future clinical trials will validate the possibility of serum anti-3’-sialyllactose IgG as a promising biomarker for the early detection of pancreatic cancer.