Method for diagnosing pancreatic ductal adenocarcinoma

Abstract

A method of diagnosing pancreatic ductal adenocarcinoma (PDAC) comprising detecting an autoantibody of GAGE1 (G antigen 1, GAGE4) and an autoantibody of ACVR2B (activin A receptor type 2B). Also claimed is a method of diagnosing PDAC comprising detecting an antibody which binds to LEMD1, MAGEB1, PAGE1, TSGA10, BAGE4, AURKA, MAGEA10, PLEKHA5, PLK4, XAGE3Av1. The autoantibodies may be IgA and / or IgG antibodies. The method may comprise contacting a sample with GAGE1 and ACVR2B antigens and detecting the binding of antibodies to the antigens. The levels of antibodies may be quantified and compared to healthy subjects or those with chronic pancreatitis. The antigens may be bound to a support, such as a bead, membrane, slide, plate, well, filter, or dipstick. The method may comprise detecting GAGE1, ACVR2B and other additional antigens. Also claimed is an antigen complex comprising a biotin carboxyl carrier protein (BCCP)-tagged antigen bound to a support, wherein the antigen is GAGE1, and wherein the support is coated with streptavidin, for use in the method. A kit comprising GAGE1 and another antigen. Also claimed is a computer implemented method of diagnosing PDAC comprising comparing data comprising the number of antibodies detected to reference data and displaying a diagnosis.

Description

FIELD OF THE INVENTION This invention relates to biomarkers, in particular IgG and IgA autoantibodies, for use in diagnosing pancreatic ductal adenocarcinoma. BACKGROUND TO THE INVENTION Pancreatic ductal adenocarcinoma (PDAC) is a heterogeneous cancer which is both difficult to diagnose and difficult to treat, and the overall survival rate of pancreatic cancer is reported to be less than 5%. A lack of druggable targets has hindered the development of effective drugs for treating the disease and early detection is therefore vital. However, due to the absence of early symptoms or the symptoms being associated with other abdominal diseases such as dyspepsia and pancreatitis, most cases of pancreatic cancer are only diagnosed at a late state when curative surgery is almost impossible. Traditionally, the diagnosis of PDAC has been significantly based on advanced radiological methodologies, such as computed tomography (CT), magnetic resonance imaging (MRI), endoscopic ultrasound (EUS) and 18-Fluorodeoxyglucose positron emission tomography (FDG-PET). However, such imaging methods are expensive, require instrumentation that is not readily available in developing countries, are low throughput, and fundamentally therefore require some other indication of risk of PDAC to justify their use. Percutaneous needle biopsies have also emerged as an adjunct diagnostic approach, representing a minimally-invasive method to obtain tissue which can then be used to confirm the presence of PDAC through histology. However, inherent sampling bias means that this is not a realistic early detection method on its own. It thus remains clear that despite these advancements in imaging and sampling technologies, the challenge of early stage PDAC diagnosis remains formidable, especially in resource limiting settings, yet has paramount significance in the context of combatting this malignancy. SUMMARY OF THE INVENTION According to a first embodiment of the invention, there is provided a method of diagnosing or screening for pancreatic ductal adenocarcinoma (PDAC) in a patient, the method comprising detecting an antibody which binds to GAGE1 and an antibody or antibodies which bind to at least one other antigen selected from the group consisting of ACVR2B, LEMD1, MAGEB1, PAGE1, TSGA10, BAGE4, AURKA, MAGEA10, PLEKHA5, PLK4 and XAGE3Av1 in a biological sample from a subject. The antibodies may be IgG and / or IgA antibodies. The method may comprise detecting antibodies which bind to GAGE1 and at least two of ACVR2B, LEMD1, MAGEB1, PAGE1, TSGA10, BAGE4, AURKA, MAGEA10, PLEKHA5, PLK4 and XAGE3Av1. The method may comprise detecting antibodies which bind to GAGE1 and at least three of ACVR2B, LEMD1, MAGEB1, PAGE1, TSGA10, BAGE4, AURKA, MAGEA10, PLEKHA5, PLK4 and XAGE3Av1. The method may comprise detecting antibodies which bind to GAGE1 and at least four of ACVR2B, LEMD1, MAGEB1, PAGE1, TSGA10, BAGE4, AURKA, MAGEA10, PLEKHA5, PLK4 and XAGE3Av1. The method may comprise detecting antibodies which bind to GAGE1 and at least five of ACVR2B, LEMD1, MAGEB1, PAGE1, TSGA10, BAGE4, AURKA, MAGEA10, PLEKHA5, PLK4 and XAGE3Av1. The method may comprise detecting antibodies which bind to GAGE1 and at least six, at least seven, at least eight, at least nine, at least ten, at least eleven or at least twelve of ACVR2B, LEMD1, MAGEB1, PAGE1, TSGA10, BAGE4, AURKA, MAGEA10, PLEKHA5, PLK4 and XAGE3Av1. The method may comprise detecting IgG antibodies which bind to GAGE1 and at least one of ACVR2B, LEMD1, MAGEB1, PAGE1, TSG10 and BAGE4. The method may comprise detecting IgG antibodies which bind to GAGE1 and at least two of ACVR2B, LEMD1, MAGEB1, PAGE1, TSG10 and BAGE4. The method may comprise detecting IgG antibodies which bind to GAGE1 and at least three ofACVR2B, LEMD1, MAGEB1, PAGE1, TSGIOand BAGE4. The method may comprise detecting IgG antibodies which bind to GAGE1 and at least four of ACVR2B, LEMD1, MAGEB1, PAGE1, TSG10 and BAGE4. The method may comprise detecting IgG antibodies which bind to GAGE1 and at least five of ACVR2B, LEMD1, MAGEB1, PAGE1, TSG10 and BAGE4. The method may comprise detecting IgG antibodies which bind to GAGE1 and at least six of ACVR2B, LEMD1, MAGEB1, PAGE1, TSG10 and BAGE4. The method may comprise detecting IgG antibodies which bind to: i) GAGE1, ACVR2B, LEMD1, MAGEB1 and PAGE1; ii) GAGE1, LEMD1, MAGEB1 and PAGE1; iii) GAGE1, LEMD1, MAGEB1 andTSGAIO; iv) ACVR2B, BAGE4, GAGE1, LEMD1, MAGEB1 and TSGA10; or v) ACVR2B, GAGE1, LEMD1, MAGEB1 andTSGAIO. The method may comprise detecting IgA antibodies which bind to GAGE1 or MAGEA10 and at least one other antigen selected from AURKA, MAGEB1, BAGE1, PLEKHA5, PLK4, XAGE3Av1, GAGE1 and MAGEA10. The method may comprise detecting IgA antibodies which bindtoGAGEI or MAGEA10 and at least two of AURKA, MAGEB1, BAGE1, PLEKHA5, PLK4, XAGE3Av1, GAGE1 and MAGEA10. The method may comprise detecting IgA antibodies which bind to GAGE1 or MAGEA10 and at least three of AURKA, MAGEB1, BAGE1, PLEKHA5, PLK4, XAGE3Av1, GAGE1 and MAGEA10. The method may comprise detecting IgA antibodies which bind to GAGE1 or MAGEA10 and at least four of AURKA, MAGEB1, BAGE1, PLEKHA5, PLK4, XAGE3Av1, GAGE1 and MAGEA10. The method may comprise detecting IgA antibodies which bind to GAGE1 or MAGEA10 and at least five of AURKA, MAGEB1, BAGE1, PLEKHA5, PLK4, XAGE3Av1, GAGE1 and MAGEA10. The method may comprise detecting IgA antibodies which bind to GAGE1 or MAGEA10 and at least six of AURKA, MAGEB1, BAGE1, PLEKHA5, PLK4, XAGE3Av1, GAGE1 and MAGEA10. The method may comprise detecting IgA antibodies which bind to: vi) GAGE1, AURKA, MAGEA10 and MAGEB1; vii) AURKA, MAGEA10, MAGEB1 and PAGE1; viii) AURKA, MAGEA10, MAGEB1 and PLEKHA5; ix) GAGE1, AURKA, MAGEA10, MAGEB1 and PLK4; or x) GAGE1, AURKA, MAGEA10, PLEKHA5 and XAGE3Av1. The method may comprise detecting any one of the panels of IgG antibodies and any one of the panels of IgA antibodies described above. The biological sample may be a liquid sample, such as a blood sample. The blood may be whole blood, serum or plasma. The blood may be fresh or dried blood. The method may comprise quantifying the levels of each antibody and comparing these to levels associated with healthy subjects or to levels associated with chronic pancreatitis. The method may comprise the steps of: contacting the biological sample with GAGE1 and at least one other antigen selected from the group consisting of ACVR2B, LEMD1, MAGEB1, PAGE1, TSGA10, BAGE4, AURKA, MAGEA10, PLEKHA5, PLK4 and XAGE3Av1; and detecting binding of autoantibodies to the antigens. The antigens may be immobilised on a support, such as a bead, a membrane, a slide, a plate, a well, a filter or a dipstick. Binding may be detected with labeled anti-human IgG antibodies and / or labeled anti-human IgA antibodies. The labels may be fluorescent labels. The IgG and IgA antibodies may be labeled differently. The method may further comprise the step of determining the sialylation, fucosylation and / or galactosylation of the autoantibodies. According to a second embodiment of the invention, there is provided an antigen complex comprising a biotin carboxyl carrier protein (BCCP)-tagged antigen bound to a support, wherein the antigen is selected from the group consisting of GAGE1, ACVR2B, LEMD1, MAGEB1, PAGE1, TSGA10, BAGE4, AURKA, MAGEA10, PLEKHA5, PLK4 and XAGE3Av1; and wherein the support is a bead, a membrane, a slide, a plate, a well, a filter or a dipstick and is coated with streptavidin, for use in the method According to a third embodiment of the invention, there is provided a kit including: - GAGE1; and - one or more other antigens selected from the group consisting of ACVR2B, LEMD1, MAGEB1, PAGE1, AURKA, MAGEA10, PLEKHA5 and XAGE3Av1. The antigens may be bound to a support. The kit may further include: - labeled anti-human IgG antibodies; and / or - labeled anti-human IgA antibodies. The kit may further include instructions, in written or computer-coded form, for performing the method described above. According to a further embodiment of the invention, there is provided a use of the immobilised antigen(s) or a kit as described above in a method of diagnosing PDAC. According to a further embodiment of the invention, there is provided the use of one or more antigens selected from the group consisting of GAGE1, ACVR2B, LEMD1, MAGEB1, PAGE1, TSGA10, BAGE4, AURKA, MAGEA10, PLEKHA5, PLK4 and XAGE3Av1 in the manufacture of a kit, composition, antigen complex or solid support for use in a method of diagnosing PDAC. According to a further aspect of the invention, there is provided a computer implemented method of diagnosing PDAC, the computer performing steps including: receiving inputted subject data comprising the number of antibodies detected in a biological sample from a patient, wherein the antibodies are IgG and / or IgA autoantibodies bound to one or more antigens selected from the group consisting of GAGE1, ACVR2B, LEMD1, MAGEB1, PAGE1, TSGA10, BAGE4, AURKA, MAGEA10, PLEKHA5, PLK4 and XAGE3Av1; comparing the data obtained from the sample to reference data for the same autoantibodies associated with a subject without PDAC or associated with a patient having PDAC and thereby determining whether the subject has, or possibly has, PDAC; and displaying a diagnosis. According to a further embodiment of the invention, there is provided a method of diagnosing and treating pancreatic ductal adenocarcinoma (PDAC) in a patient, the method comprising: detecting an antibody which binds to GAGE1 and an antibody or antibodies which bind to at least one other antigen selected from the group consisting of ACVR2B, LEMD1, MAGEB1, PAGE1, TSGA10, BAGE4, AURKA, MAGEA10, PLEKHA5, PLK4 and XAGE3Av1 in a biological sample from a subject; and administering a treatment for PDAC to the patient or performing surgery on the pancreas of the patient. It is contemplated that any method or composition described herein can be implemented with respect to any other method or composition described herein. Other objects, features and advantages of the present disclosure will become apparent from the following detailed description. It should be understood, however, that the detailed description and the specific examples, while indicating specific embodiments of the disclosure, are given by way of illustration only, since various changes and modifications within the spirit and scope of the disclosure will become apparent to those skilled in the art from this detailed description. BRIEF DESCRIPTION OF THE FIGURES Figure 1: Autoantibody response against tumour antigens in PDAC serum and tissue samples. Tissue analysis for (A) IgG and (B) IgA were used as surrogate indicators of antibody-isotype abundance between the three tissue types. (C) The proportion of antibodypositive antigens was also compared for IgG and IgA across the three tissue types. Forest plots showing the antibody subclass abundance for IgG and IgA antibody-isotypes in tissue. (D) Antibody subclass abundance determined by calculating the mean difference between antigen ratios (mean isotype antigen ratio - mean subclass antigen ratio), and the 95% and 5% confidence interval (Cl) and represented on a forest plot. The four IgG subclasses show lgG4 as the predominant antibody response in PDAC tissue, lgG1 and lgG3 have similar abundance. (E) lgA2 is the predominant IgA subclass, and lgA1 is comparably elevated in PDAC tumour tissue. Serum analysis showing bar graphs for differences in signal intensity values between cohorts with the pooled DYS+CP cohort segregated into two groups of Dyspepsia (DYS) and chronic pancreatitis (CP) for IgG and IgA. (F) IgG response shows that both confounding groups (CP and DYS) have no significant difference to the control group. (G) The CP group has a significantly different response to the control group for IgA, while the DYS group has a similar response to the control group. (H) Forest plot comparing abundant isotype distribution of antibody-positive antigens, showing that more antigens are IgA-positive. (I) Column graph showing the distribution of IgG subclasses based on signal RFU in the pooled DYS+CP cohort. (J) IgA subclass reactivity for the pooled DYS+CP group, indicating abundant lgA1 compared to lgA2. Signal intensity values for normal-adjacent, tumour, and chronic pancreatitis (CP) tissue, and serum samples for PDAC, CP,DYS, and LTBI were set at a threshold >mean+2SD of the normal-adjacent antibody-positive antigen response and the scatter plots showing the relative fluorescent units (RFU) of the mean ±SEM for Statistical analysis by one-way ANOVA, pairwise comparison of each group against the control group, graph represents mean values ± SEM (*P <0.05, **P <0.01, ***P <0.001, ****P <0.0001). Figure 2: Comparing glycosylation features of serum autoantibodies against CTAs in PDAC. (A) The total glycosylation distribution for the PDAC, pooled DYS+CP, and control cohort for galactosylation, fucosyaltion, and sialylation. Column graphs showing the mean ± SEM relative fluorescent units (RFU) of antigen-bound autoantibodies (threshold set >mean+2SD of the control RFU) that have glycans recognised by (A) RCA (P1,4-Gal), (B) SNA (a2,6-Sal), (C) LCA(a1,6-Fuc), and (D) ECL(P1,4-Gal) as a surrogate marker for assessing the detection of glycosylation moieties in serum autoantibodies using the Sengenics CT262 microarray platform. (E) Glycosylation features assessed in lectin-positive antigens observed in the PDAC cohort compared to the pooled DYS+CP group, showing the variable presence of different carbohydrate moieties. (F) A heatmap and dendogram of Iog2 transformed median normalised microarray data for each antigen-positive serum autoantibody feature and the associated detection of cancer testis antigens. Hierarchical clustering has grouped serum features and positive antigens by similarity. Figure 3: Scheme of a proposed method for diagnosing PDAC based on a panel of autoantibody biomarkers described herein. DETAILED DESCRIPTION OF THE INVENTION As discussed above, pancreatic ductal adenocarcinoma (PDAC) is a disease with minimal response to therapeutic intervention if diagnosed at a late stage. Early stage detection of the disease is crucial for the patient’s survival. In this study, an immunoproteomic approach was applied to investigate autoantibody responses against cancer-testis and tumour-associated antigens in PDAC using a high-throughput multiplexed protein microarray platform, comparing humoral immune responses in serum and at the site of disease. Serum or tissue IgG and IgA antibody isotypes and subclasses in a cohort of PDAC, disease control and healthy patients were simultaneously quantified. Subclass utilization in tumor tissue samples was observed to be predominantly immune suppressive lgG4 and inflammatory lgA2, contrasting with predominant lgG3 and lgA1 subclass utilization in matched sera and implying local autoantibody production at the site of disease in an immune-tolerant environment. By comparison, serum autoantibody subclass profiling for the disease controls identified lgG4, lgG1, and lgA1 as the abundant subclasses. Combinatorial analysis of serum autoantibody responses identified panels of candidate biomarkers. The top IgG panel included ACVR2B, GAGE1, LEMD1, MAGEB1 and PAGE1 (sensitivity, specificity and AUC values of 0.933, 0.767 and 0.906). Conversely, the top IgA panel included AURKA, GAGE1, MAGEA10, PLEKHA5 and XAGE3aV1 (sensitivity, specificity, and AUC values of 1.000, 0.800, and 0.954). Assessment of antigen-specific serum autoantibody glycoforms revealed abundant sialylation on IgA in PDAC, consistent with an immune suppressive IgA response to disease. “Sandwich” assays of (i) GAGE1 and ACVR2B, LEMD1, MAGEB1 and PAGE1 and (ii) GAGE 1 and AURKA, MAGEA10, PLEKHA5 and XAGE3aV1 were developed and employed as highly specific, sensitive, and stable diagnostic tests for PDAC. To the best of the inventors’ knowledge, this is the first study to demonstrate the diagnosis of PDAC based on the detection of autoantibodies in a sample from a patient. These and other aspects of the disclosure are described in detail below. Similar to autoimmune disorders, cancer produces autoantibodies. An autoantibody is an antibody produced by the immune system that is directed against one or more of the individual's own proteins. Typically, these autoantibodies are raised early in disease against mutated or aberrantly expressed / modified proteins. Numerous studies have identified the presence of autoantibodies in cancer, but the immunological importance of those antibodies in terms of whether they contribute to, or hamper cancer progression remains a matter of ongoing debate. Autoantibody-based tumor biomarkers have been studied as potential prognostic, diagnostic, and monitoring agents of therapeutic response in breast, prostate, and lung cancers amongst others. The common challenge with these biomarker studies is that the identified targets individually typically lack specificity and sensitivity, and some are only applicable to a specific tumor subset. As a result, the clinical application of antibodies in cancer to date has largely focused on their use as targeted immunotherapies. Of note, there are five human antibody isotypes, yet current therapeutics are based on the IgG isotype, particularly IgG 1, with other antibody isotypes having not been explored or developed for mAb therapies. Similarly, the aforementioned studies on autoantibody-based biomarkers were based solely on detection of IgG antibodies. Tumour antigens can be defined as tumour associated (TA) antigens, which are antigens similar to proteins found in normal cells but are modified or aberrantly expressed. Amongst the subset of proteins that are more likely to elicit an autoantibody response in cancers, the cancer-testis (CT) antigens are a family of ca. 500 tumor-specific antigens with highly restricted expression in normal adult somatic tissues and aberrant expression in various cancers as a result of disrupted gene regulation. As the testis is an immune-privileged site, aberrant expression of these antigens in cancer typically triggers a spontaneous cellular (T cell) and humoral (B cell) immune response to the relevant CT antigen. The latter includes the maturation of B cells against specific antigens to produce cognate antibodies which are detectable in the circulation. An unpublished assay of >3000 healthy individuals conducted by one of the inventors showed no detectable anti-CT antigen autoantibody titre. Antibody production is compartmentalised, with bone marrow-derived B-cells and tissue resident B-cells having distinct lineages and producing different antibody isotypes, potentially against different target antigens, which can confuse interpretation of the physiological significance of autoantibody production in cancers. In this study, the inventors therefore used a high-throughput multiplex microarray platform to profile serum samples for autoantibody responses, quantifying CT and TA antigen-specific IgG and IgA isotypes and their respective subclasses in PDAC and comparing them to autoantibodies extracted from diseased tissue. Moreover, because different subclasses of each antibody isotype have varying effector functions, which are also dependent on the glycan composition of the Fc region, their characterization would better reflect the antibody effector roles that are responsible for immune regulation in PDAC carcinogenesis. Thus, this study also aimed to identify the glycan moieties associated with the antigen-specific autoantibodies identified in serum. In order to provide more detailed characterization of the humoral response at the site of disease in PDAC patients, quantitative autoantibody profiling was carried out against 262 cancer-testis and tumour-associated antigens, utilizing a multiplexed, reproducible high through-put microarray platform to determine the isotype, subclass, and sialylation of antigenspecific autoantibodies in serum and matched tumour tissue from PDAC patients and controls. Amongst others, the data from this study revealed significant differences in anti-CT / TA antigen IgG and IgA autoantibody titres between PDAC patients and controls that are measurable in serum and which may provide the basis for early detection of PDAC. The data from this study also revealed significant differences in anti-CT / TA antigen isotype and subclass utilization in tissue biopsies at the site of disease (predominantly lgA2 and lgG4) compared to that found in matched sera (predominantly lgG3 and lgA1) or in matched adjacent normal biopsies, which argues against simple infiltration of antibodies from blood into the diseased tissue and instead argues for local autoantibody production at the site of disease. The identified anti-CT / TA antigen autoantibody responses were shown to predominantly reflect a tolerogenic antibody response at the site of disease. The predominant anti-CT / TA antigen IgG subclass was found to be the non-activating lgG4, whereas lgG1 and lgG3 -which can induce antibody-dependent cellular cytotoxicity (ADCC) and complement dependent cytotoxicity (CDC) through their effector functions - were found to be the least abundant IgG subclasses in tissue, implying that these functions are not effectively activated to drive tumour clearance; and the predominant anti-CT / TA antigen IgA subclass was found to be the pro-inflammatory lgA2. Antibodies are produced in the adaptive phase of an immune response and increased affinity for antigens is achieved through somatic hypermutations and isotype switching. Previously, studies have largely focused on IgG profiling in cancer and have demonstrated elevated expression levels of IgG in cancer cells. By simultaneously measuring IgG and IgA responses in PDAC, this study showed that whilst signal intensities were higher for IgG compared to IgA in serum, IgA has a broader selectivity for CT / TA antigens compared to IgG in both serum and tissue. Isotype switching increases the functional diversity of antibodies as the immune response proceeds. Indeed, the type of antibody activated during an immune response is important in determining downstream effector signaling and interaction with other immune cells. Previous studies have demonstrated that antibody isotype and subclass abundance fluctuate through the course of infectious diseases and it is plausible therefore that the same phenomenon occurs in cancers. Whilst the cohort studied here was cross-sectional, not longitudinal, in design, it is nonetheless interesting that in the serum analysis of this study, patient IgG subclass evaluation revealed that lgG1 and lgG2 reactivity was absent in some patients, which may be an indication of temporal changes that occur during cancer progression being reflected in the systemic immune response. Aberrant glycosylation has been a key feature in the acquisition and sustenance of hallmark characteristics that have been implicated in cancer development. Furthermore the glycobiology of antibodies and their respective receptors have been suggested to be an important factor in mediated effector responses and therapeutic development and activity. Additionally, research has shown that 2,6 sialic acid expression is associated with chemoresistance in PDAC, albeit that data was based on altered sialylation of tumour antigens, not of tumour-associated autoantibodies as observed here. The presence of specific glycan moieties on cancer antigen-specific serum autoantibodies was determined herein using a microarray platform, focused on three possible glycan sites by using a panel of four lectins: SNA, LCA, RCA, and ECL, in order to respectively investigate sialylation (a2,6-Sal), fucosylation (a1,6-Fuc), and galactosylation (pi,4-Gal) of antigen-bound serum autoantibodies. It is envisaged that by using the autoantibodies described herein for cancer diagnosis, increased autoantibody levels could be detected at early stages of the disease (e.g. stage 1A, 1B or 2A), making it possible to diagnose PDAC several months or years before any clinical signs of tumor development are presented. An immunoassay can be used to perform the diagnosis. In one embodiment, the immunoassay is a multiplexed “indirect” or “antigen-down” assay in which a target panel of antigens which can bind autoantibodies of interest are immobilised on a support, such as a bead, a membrane, a slide, a plate, a well, a filter or a dipstick. When a liquid sample from a patient is brought into contact with the antigens, any autoantibodies in the sample which recognise the antigens (“primary antibodies”) will bind to the antigens and become immobilised on the support. After a washing step to remove unbound and non-specifically bound antibodies, differentially labeled species-specific antibodies (i.e. anti-human IgA and / or anti-human IgG) (“secondary antibodies”) which bind to the primary antibodies are then added. The label is typically a fluorophore but could also be any other type of detectable label, such as an enzyme. Using any of the methods and devices known in the art (such as a microarray scanner or a Bioplex® or xMAP® system (Luminex)), the presence of antigen-specific autoantibodies in a sample are detected and quantified, with pg / ml limits of detection. Based on this data, a diagnosis can be made. An algorithm can be provided to compare the sample data to an autoantibody signature, to previously calculated reciprocal titres for the autoantibodies and / or to antigen ratios of median antibody titres in PDAC and non-cancer patients or patients with a confounding disease of the pancreas such as chronic pancreatitis (CP), and to discriminate between PDAC patients and other confounding diseases. A diagnosis of the patient having PDAC or possibly having PDAC can be made when IgG and / or IgA autoantibodies directed to the antigens in the panel are detected in the sample, or when the levels of the detected autoantibodies are higher than a typical level of the same autoantibodies in subjects without PDAC. Cut-off or threshold values can be determined based on levels of the same autoantibodies which are typically found in patients without PDAC. The levels referred to herein may be the concentration of the antibody in the sample. The test may be an initial diagnostic test, i.e. in the case of a positive diagnosis, the patient will be sent for further tests to confirm the diagnosis. Alternatively, or in addition, a patient with a positive result can be administered a suitable treatment for PDAC. For example, they may be treated with chemotherapy, radiation, surgery or a combination of these. When the cancer is advanced, palliative treatment may be administered, e.g. treatment to relieve pain or nausea. Given the paucity of clinical symptoms in early disease that might trigger a request for a specific test, a regular screening program amongst at-risk populations can be provided, as is done for a number of other cancers. In one embodiment, the biological sample is blood, which may be whole blood, serum or plasma. The blood may be fresh or dried blood (e.g. a dried blood spot). For simple, low-cost sample collection, and clinical implementation for patient access from resource limiting sites to a centralised laboratory, the use of dried blood spots collected on blood cards can be easily implemented. As autoantibody signals can be quantified after storage on blood cards for up to 3 months at room temperature, the blood cards can be posted or couriered at room temperature and at low cost, from remote sites to a central laboratory. Alternatively, the blood sample could be another liquid sample, such as urine. Other examples of suitable biological samples are a tissue biopsy from the site of disease, a tissue biopsy from an adjacent lymph node, exosomes prepared from blood, or stool. The target panel of antigens can be biotin carboxyl carrier protein (BCCP)-tagged. This can be done by, for example, expressing the antigens as fusions to a BCCP tag in vivo in insect cells, e.g. using a baculoviral system as described in Beeton-Kempen et al. (Beeton-Kempen, N., et al., Development of a novel, quantitative protein microarray platform for the multiplexed serological analysis of autoantibodies to cancer-testis antigens. Int J Cancer, 2014. 135(8): p. 1842-51). A bead-based test suitable for implementation in a clinical laboratory setting can be used, such as a Bioplex 200® test (Luminex). In this test, the support is a bead. The support can be coated with streptavidin and optionally also encoded with a distinguishable dye. The method of diagnosing PDAC described herein comprises detecting an antibody which binds to GAGE1 and an antibody or antibodies which bind to at least one other antigen selected from the group consisting of ACVR2B, LEMD1, MAGEB1, PAGE1, TSGA10, BAGE4, ALIRKA, MAGEA10, PLEKHA5, PLK4 and XAGE3Av1 in a biological sample from a subject. The antibodies are typically IgG and / or IgA autoantibodies. Antibodies which bind to GAGE1 and at least two, at least three, at least four, at least five, at least six, at least seven, at least eight, at least nine or at least ten of ACVR2B, LEMD1, MAGEB1, PAGE1, TSGA10, BAGE4, AURKA, MAGEA10, PLEKHA5, PLK4 and XAGE3Av1. In one embodiment, IgG antibodies which bind to GAGE1 and at least two, at least three, at least four, or at least five of ACVR2B, LEMD1, MAGEB1, PAGE1, TSG10 and BAGE4 are detected. For example, the method can comprise detecting IgG antibodies to any one of the following panels of antigens: i) GAGE1, ACVR2B, LEMD1, MAGEB1 and PAGE1; ii) GAGE1, LEMD1, MAGEB1 and PAGE1; iii) GAGE1, LEMD1, MAGEB1 andTSGAIO; iv) ACVR2B, BAGE4, GAGE1, LEMD1, MAGEB1 and TSGA10; or v) ACVR2B, GAGE1, LEMD1, MAGEB1 and TSGA10. In another embodiment, IgA antibodies which bind to GAGE1 or MAGEA10 and at least two, at least three or at least four of AURKA, MAGEB1, BAGE1, PLEKHA5, PLK4 and XAGE3Av1 are detected. For example, the method can comprise detecting IgA antibodies to any one of the following panels of antigens: vi) GAGE1, AURKA, MAGEA10 and MAGEB1; vii) AURKA, MAGEA10, MAGEB1 and PAGE1; viii) AURKA, MAGEA10, MAGEB1 and PLEKHA5; ix) GAGE1, AURKA, MAGEA10, MAGEB1 and PLK4; or x) GAGE1, AURKA, MAGEA10, PLEKHA5 and XAGE3Av1. In another embodiment, a panel of IgG antibodies and a panel of IgA antibodies are detected. The IgG and IgA antibodies may be directed to any of the antigens described herein. The method optionally further comprises a step of determining the sialylation, fucosylation and / or galactosylation of the autoantibodies. The pattern of antibody Fc glycosylation has a significant impact on the interaction of antigen-specific antibodies with other components of the innate immune system and modulates the associated effector functions, which are a key component of adaptive immune response control of disease. In this study, it was observed that the patterns of antigen-specific antibody glycoslation differs in PDAC patients compared to confounding disease controls. Thus, determining the sialylation, fucosylation and / or galactosylation of the antigen-specific autoantibodies serves to further increase the discriminatory power of the antigen-specific autoantibody diagnostic panels. A support having one or more of the immobilised antigens can be provided. A kit for use in the method can be provided which includes any one or two of the panels of antigens described above, wherein the antigens are optionally immobilized on a support. The kit can also include labeled anti-human IgG antibodies, labeled anti-human IgA antibodies, and / or instructions, in written or computer-implementable form, for performing the method described above. The antigens or the panel of antigens described above can be used in the manufacture of a kit, composition, complex, medicament or solid support for use in a method of diagnosing PDAC. The invention will now be described in more detail by way of the following non-limiting examples. Example 1: Identification of autoantibody biomarkers and biomarker panels for PDAC Materials and Methods Sample collection for study cohort This study was approved (HREC 654 / 2017& HREC 802 / 2020) by the Human Research Ethics Committee of the Faculty of Health Sciences, University of Cape Town. Blood samples were obtained from patients with early-stage (stage 1A, 1B or 2A) pancreatic ductal adenocarcinoma (n=30) who were diagnosed and underwent tumour resection surgery (pancreaticoduodenectomy) at Groote Schuur Hospital (GSH), Cape Town, South Africa. Informed consent was obtained from all patients involved in the study. Additionally, serum from patients with confounding diseases of the pancreas, chronic pancreatitis (n=16), non-ulcer dyspepsia (n=13) were used as the disease controls, and from patients characterised as having a latent tuberculosis infection (LTBI) (n=30) but who were otherwise healthy, were used as ‘healthy’ controls. Notably, in a South African context, public health data suggests that ca. 80% of adults carry a latent tuberculosis infection, thus making the LTBI group an appropriate control in the present study. Tissue matched from a subset of PDAC patients in the serum cohort was collected for assessment of local autoantibody production. Here, paired tissue sections of ~3mm2 for PDAC tumour (n=8), normal adjacent (n=8), and chronic pancreatitis (n=8) were lysed for antibody extraction and antibody presence in tissue lysates was confirmed by an immunoglobulin affinity purification method using magnetic Protein A and Protein G microbeads (MagReSyn®), as per to the manufacturer’s protocol. Serum and tissue samples were stored at -80°C until assays were performed. Serum and tissue antibody assays Antibody assays were performed as described in Smith et al. (Smith, M. et al., Age, Disease Severity and Ethnicity Influence Humoral Responses in a Multi-Ethnic COVID-19 Cohort. Viruses, 2021. 13(5)), with the following modifications: A dual colour format was adapted for antibody detection with simultaneous incubation of AF647-labelled anti-human IgG, and AF555-labelled anti-human IgA detection antibody (both at 1Opg / ml) for 30 min. Similarly, dualcolour microarray assays were performed for the respective antibody subclasses. The serum 5 autoantibody and lectin microarray assays were carried out using commercial cancer-testis antigen arrays comprising 213 CT plus 49 TA antigens (Sengenics), whereas microarray assays on tissue extracts were performed on a custom CT100+ platform as previously described in Beeton-Kempen et al. Both arrays contained triplicate spots of each antigen, and individual arrays were isolated using ProPlate 4-plex multi-well chambers. Detection antibody 10 subclasses (anti-human lgG1, lgG2, lgG3, lgG4, lgA1, and lgA2) and anti-human IgA were derivatised in-house with either Alexa Fluor (AF)647 or AF555 (Table 1). The assays were performed with a serum or tissue lysate dilution of 1:200 in phosphate-buffered saline with Tween20 (PBST), with minimal light, at room temperature, and all incubations were performed on a shaker at 100 rpm unless stated otherwise. 15 Table 1: Reagents used for antibody and lectin derivatisation, and for microarray assays. Reagent ^XXXXXXXXXXXXXXXXXXXVMXXXXXXXXXXXXXXXXXWKKKKKKKXXXXXX a-hlgG ! a-hlgA Stock : concentration x«Xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx 2 mg / ml i 2.4mg / ml Supplier xxxxxxxxxx^>>>>>>>>»xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx ThermoFisher j ThermoFisher i Catalogue number XXXX^XXXXXXX>>>>>>>>>XKKKKKKKKKKKKKKKKKSX^X\XXXXXXXXXXXXXXX : A21145 i 31140 i AF647-NHS ester : 10mg / ml : ThermoFisher AF647-NHS ester ! AF555-NHS ester : 10mg / ml ThermoFisher : AF555-NHS ester i a-hlgAi : 0.5mg / ml : Southern Biotech : 9130-01 j a-hlqA2 0.5mg / ml j Southern Biotech i 9140-01 j a-hlgGi................................ : 0.5mg / ml_______________ j Southern Biotech 9054-01 ? a-hlgG2 : 0.5mg / ml i Southern Biotech I 9060-01 § a-hlgG3 : 0.5mg / ml : Southern Biotech ! 9210-01 : a-hlqG4 ; 0.5mg / ml : Southern Biotech i 9200-01 § RCA : 10 mg / ml i Vector laboratories L-1080-10 SNA : 5 mg / ml Vector laboratories : L-1300-5 ! LCA : 10 mg / ml i Vector laboratories i L-1040-10 ! ECL j 10 mg / ml i Vector laboratories i L-11140-10 ! Biotin : 50 mM : Sigma® B4501-1G ! CaCI2 : 500 mM : Glentham : GK3739 ! Glycerol j 50% : Sigma® : G7757-5L i HEPES ; 2.5 M : Glentham i GM5581 KCI 1 M : Sigma® : P9333-1KG Milk powder : Powder : Sigma® I 70166-500G : MgCI2 : 1000 mM : Glentham i GK5046 ! MnCI2 : 100 mM : Glentham : GK2508 ! PBS : 10x : Gibco i 70011-036 ! Tris (pH 8.0) : 500 mM : Glentham I GP7166 ! Triton X-100 i 100% : Sigma® : 93443-100ML ! Tween 20 j 100 % : Glentham GK2245 Sodium bicarbonate 0.1M : Glentham__________________________ ; GE8414_________________________________ Biomarker antigens which had been biotinylated and BCCP-tagged such that oriented immobilisation and in situ purification of each antigen was obtained, were bound to a solid support, in this case glass slides with a hydrogel layer coated with covalently bound streptavidin tetramers. The microarray slides were stored at -20°C. Microarray slides were removed from -20°C storage and blocked with ice-cold blocking buffer for one hour at room temperature, and then then washed 2x in PBST and 2X PBS for five minutes at room temperature (100rpm). Thereafter, the slides were washed and dried by centrifugation at 1200 Xg for 2 minutes. Subsequently, slides were assembled in 4-plex gaskets and incubated with patient sera (1:200) for one hour at room temperature (100rpm). Thereafter, the slides were briefly rinsed with PBST, removed from the gaskets, and washed 3x for five minutes with PBST Following this, fluorescently labelled anti-human IgG and IgA detection antibody (10pg / ml) were added and incubated for 30min at room temperature on a shaker (1 OOrpm). Thereafter, the slides were washed 2x with PBST, and 2X with PBS for 5min at 10Orpm at room temperature. Finally, the slides were dried at 1200xg for 3 minutes at 23°C. Serological lectin assays Microarray slides were washed with PBST and incubated with gentle agitation for 3x 5 min, then washed with 2x 5min with PBS and dried by centrifugation at 1200X g for 2min. Individual arrays were incubated with the patient serum for 1hr with gentle agitation. Thereafter, the slides were briefly rinsed 3x with Tris-buffer saline (TBS) and incubated with 3% deglycosylated BSA for 1 hr. Subsequently, each lectin was diluted to 1ug / ml in lectin binding buffer (20 mM Tris (pH 8.0), 0.1 mM CaCI2, 0.1 mM MgCI2, 0.1m M MnCI2, 0.2% Tween 20), added to the slide, and incubated with gentle agitation for 30 min. Finally, the slides were washed 2x for 5 min each with TBST and TBS. Bioinformatic analysis The statistical estimation of power and sample size were calculated using power calculations and performed using G*Power version 3.1.9.4. Post hoc power calculations of matched tumour and normal-adjacent tissue was performed using the R package “ssize.fdr”. Microarray image analysis and raw data extraction Microarray slides were scanned at a fixed gain setting using an InnoScan 710 (Innopsys, Carbonne, France) fluorescence microarray scanner, generating a 16-bit TIFF file. A visual quality control check was conducted and any arrays showing artifacts were re-assayed. A GAL (GenePix Array List) file containing information regarding the location and identity of all antigen spots was used for image analysis. Automatic extraction and quantification of each spot were performed using Mapix software (Innopsys) to obtain the median foreground and local background pixel intensities for each spot. Data pre-processing and statistical analysis The mean net fluorescence intensity of each spot was calculated as the difference between the raw mean pixel intensity and its local background using in-house developed software (Protein Microarray Analyser; Da Gama Duarte, J., et al., PMA: Protein Microarray Analyser, a user-friendly tool for data processing and normalization. BMC Res Notes, 2018. 11(1): p. 156). The output files contained the relative fluorescent unit (RFU) and coefficient of variation for all antigens and controls spotted on the array. The R studio and R packages were used to perform clustering analysis, and the OptimalCutPoints package and receiver operating curves (ROC) were used to determine antigen specificity and sensitivity. Combinatorial ROC analysis (http: / / combiroc.eu / ) was used to determine the potential antigen biomarker panels. Other statistical analyses and graphical representations were generated using GraphPad Prism (v 9.5.1; GraphPad Software, San Diego, CA, USA). Results Quantifying autoantibody responses against cancer testis antigens To determine the specific autoantibody reactivity against the cancer antigens on the microarray platforms, a signal intensity threshold was set to distinguish true antibody-antigen binding from non-specific binding. After data normalization, the threshold for each protein was set as the mean plus 2SD for the normal-adjacent tissue samples (for tissue extracts) or the control group for serum; proteins with signals above this threshold were considered true autoantibody binding. It is well understood from ligand binding theory that the relative fluorescent units (RFU) measured for autoantibodies bound to individual autoantigens on a protein microarray depends on the density of the immobilised autoantigen, the concentration of the autoantibody in solution and the affinity of interaction between the autoantigen and autoantibody. Furthermore, the fluorophore labelling efficiency will vary between different isotype-specific detecting antibodies. Thus, whilst it is meaningful to compare RFU values for a given isotype bound to the same autoantigen across different samples, it is generally not considered meaningful to directly compare RFU values between the same isotype bound to different autoantigens on a microarray, or between different isotypes bound to the same autoantigen. However, RFU values for each autoantigen-bound autoantibody are nonetheless linearly related to antibody concentrations (titres). Given moreover that, when comparing autoantibody profiles in PDAC vs chronic pancreatitis (CP) and other controls, in serum and in tumour or normal-adjacent samples, different autoantigens were identified in the different disease and sample types, all antigen-specific RFU values for each autoantibody isotype (IgG; or IgA) in each sample type (tumour-; CP-; or normal-adjacent tissue) were plotted in order to provide a measure of relative autoantibody isotype abundance (Figures 1A &B). An increased number of autoantibody-positive antigens for a specific isotype (Figure 1C) should result in an increased mean autoantibody RFU value for that isotype. Thus, Figures 1A-C should be read together since they provide two different dimensions of relative autoantibody isotype abundance: number of autoantigens per isotype and mean titres of antigen-specific autoantibodies per isotype. For the tissue samples, post-hoc power calculations indicate that a single comparison of the paired PDAC tumour (n=8) and normal-adjacent (n=8) provides >99% power to detect a 5-fold change in IgG / lgA ratios either direction (FDR 0.05). Thus, the microarray results show that whilst there was little difference observed for the IgG levels between CP tissue and normal-adjacent tissue, significant differences were found in IgG levels between normal-adjacent and tumour tissue (p<0.0001) (Figurel A). The autoantibody levels for IgA also showed a significant difference for both tumour and CP tissue compared to the normal-adjacent samples (Figure 1B). Furthermore, IgA had a higher RFU compared to IgG in the tissue samples, with a 1:5 IgG / lgA ratio. These results thus provide an estimation of the abundant autoantibody isotype in pancreatic tissue and indicate an IgA-dominant cancer tissue microenvironment. In accordance with autoantibody abundance, the proportion of autoantibody-positive antigens showed significantly higher (p<0.0001) levels of IgA-positive antigens in tumour samples compared to IgG-positive antigens, but no statistically significant difference was observed for the normal-adjacent and the CP tissue samples (FigurelC). Together, these results indicate that in tissue, IgA-based responses against cancer antigens predominate in PDAC, but isotype-based specificity is not established in CP. Furthermore, the abundance of the respective subclasses for each antibody isotype was investigated, by calculating the antigen ratio (number of antibody-positive antigens 4- number of antibody-negative antigens) for each subclass and corresponding isotype in each sample and then representing the mean difference between the two for each sample type (mean isotype antigen ratio - mean subclass antigen ratio), together with the 95% and 5% confidence interval (Cl), for each subclass in a forest plot, in order to enable comparisons to be made independent of differences in isotype abundance between patients. The smaller and likely more negative the resultant mean difference, the greater the abundance of the subclass within the respective isotype. In tissue, the majority of IgG antigen-bound autoantibodies were found to be of the lgG4 subclass (Figure 1D). Moreover, the complement activating subclasses, lgG1 and lgG3 were observed to be the least abundant in the tumour environment of PDAC. The IgG subclass data thus indicated a tolerogenic immune response because the dominant antibody, lgG4, is considered a non-activating antibody. IgA subclass autoantibodies targeting tumour antigens in PDAC had a similar abundance, but with higher lgA2 than lgA1, as expected (Figure 1E). Although lgA1 and lgA2 mediate their effector functions by binding to the FcaRI as monomers, they are considered to be regulatory and pro-inflammatory respectively, with lgA2 typically dominant in mucosal tissue. Moreover, in the tissue extracts, it is likely that dimeric slgA is being detected, which is produced by mucosal plasma cells, rather than monomeric IgA (produced by bone marrow-derived B-cells). Together, the antibody subclass data from tissue samples indicates an antibody response that is pro-inflammatory, and not optimal for the activation of effector functions, thus promoting a tolerogenic tumour environment in PDAC. By contrast, when the mean signal intensities of IgG and IgA in serum were quantified, the overall mean intensity values were ~5-fold higher for IgG (Figure 1F) than for IgA (Figure 1G), as expected. Thereafter, the abundance of autoantibody isotypes between IgG and IgA were assessed by calculating the antigen ratio for each isotype (number of antibody-positive antigens 4- number of antibody-negative antigens), and the 95% and 5% confidence interval (Cl) and represented these ratios as a forest plot. The results showed that IgA had a high abundance of positive anti-CT / TA antigen autoantibody signals compared to IgG (Figure 1H), albeit IgG had a higher RFU signal overall. Furthermore, serum antibody subclass profiles - obtained by enumerating the number of antigens that had a positive response for a specific subclass in each patient - indicated that the most abundant isotypes in serum were lgG3 and lgG4, followed by lgG1, and the least abundant isotype was lgG2. Generally, in serum the overall pattern for all the patients indicated a dominant IgA subclass reactivity against the cancer antigens, with variation on the preferred subclass, indicating the significance of an IgA directed anti-CT / TA antigen response in PDAC. A reactivity pattern was observed that showed that patients (n=4 / 30; 13%) who had a dominant IgA subclass reactivity that was strong or very strong for both subclasses, had no lgG1 reactivity. The serum autoantibody subclass profiling of each patient is summarised in Table 2. By contrast, the serum autoantibody subclass profiling for the pooled DYS+CP group indicated lgG1, lgG4 and lgA1 as the abundant subclasses (Figure 11 and Figure 1J). Table 2: A summary of individual autoantibody subclass reactivity in serum outlining the presence and frequency of antibody-positive antigen frequency per patient. Patient (Px) ID Antibody subclass Pxi igGi + lgG2 igG3 ++ igG4 igAi +++ lgA2 + Px2 - + ++ ++ ++ ++ Px3 - + ++ ++ +++ +++ Px4 - + ++ ++ +++ ++ Px5................... + + ++ +++ ++ + Px6 + - ++ ++ +++ +++ Px7 - ++ ++ ++++ + Px8 + ++ ++ PSA ++ Px9 + +++ ++ ++ +++ Px10 + ++ ++ ++ + Px11 + - ++ ++ +++ + P12 + + ++ ++ PSA ++ P13 + +++ +++ ++ PSA + Px14 + + ++ ++ ++ + Px15 + + ++ ++ PSA + Px16 - + ++ ++ ++++ +++ Px17 + + ++ +++ + ++ pxis + + +++ ++ ++++ + Pxi 9 - + + ++ ++++ ++ Px20 + +++ ++ ++ ...........PSA + Px21 + ++ ++++ ++ ++++ ++ Px22 + + ++ ++ ++ + Px23 - + ++ ++ PSA ++ Px24 ............ + +++ ++ ++ ++++ + Px25 - + ++ +++ +++ PSA Px26............... + +++ ++ ++ +++ + Px27 + +++ PSA ++ +++ + Px28 - ++ ++ ++ ...........----- ----- Px29 + + + ++ ++ + Px30 ++ - ++ ++ +++ ++ Symbol key: PSA (Polyspecific antibody “sticky phenotype” i.e., antigen frequency >n=80) ++++ 5 (very strong; n= 64-80) +++(strong; n=48-63) ++ (moderate; n=30-47) + (weak; n=1-29) - (absent; n=0) Identification of candidate autoantibody-based serum biomarkers The minimal invasive nature of acquiring liquid biopsies makes them attractive candidates for cancer diagnosis, monitoring, and characterization. Thus, the most significant cancer antigenspecific autoantibodies from this study were identified and a subset of those autoantibodies that could be used as a panel of candidate serum biomarkers was determined. Here, 15% of IgG-positive significant antigens were obtained, compared to 61% for IgA-positive antigens, and 24% of the identified significant antigens were shared between the two antibody isotypes. Thereafter, a false discovery rate (FDR) of 1% was applied using the Benjamini-Hochberg method (Tong, T. and H. Zhao, Practical guidelines for assessing power and false discovery rate fora fixed sample size in microarray experiments. Stat Med, 2008. 27(11): p. 1960-72), after which no IgG-reactive antigens were retained individually as candidates. However, there were 12 IgA-reactive antigens retained as candidates after applying a 1% FDR. Subsequently, multiplex analysis of the data was performed by combinatorial ROC analysis on all antigens that were identified as being significantly different, with comparisons between the PDAC group and the pooled DYS+CP group. This allowed the determination of a subset of antigen combinations with the best specificity and sensitivity as potential PDAC biomarkers, thus creating a panel of antigens instead of single biomarkers. The top five antigen combinations for IgG and IgA are summarised in Table 3. The top IgG combinations had AUC, sensitivity, and specificity values of 0.906, 0.933, and 0.767, respectively. The top IgA combination showed AUC, sensitivity, and specificity values of 0.968, 1.00, and 0.833, respectively. Table 3: Combinatorial ROC analysis of top 5 antigen combinations for serum IgG and IgA PDAC classifiers showing the area under the curve (AUC), Sensitivity, and specificity values for each antigen combination. Autoantibody Combination symbol Antigens AUC sensitivity specificity igG I GAGE1-LEMD1-MAGEB1 - PAGE1 0.824 0.800 0.767 III GAGE1-LEMD1-MAGEB1 - TSGA10 0.833 0.833 0.800 XXIX ACVR2B-BAGE4-GAGE1 - LEMD1-MAGEB1-TSGA10 0.899 0.800 0.800 VII ACVR2B-GAGE1- LEMD1- MAGEB1-TSGA10 0.901 0.900 0.733 VI ACVR2B-GAGE1- LEMD1- MAGEB1-PAGE1 0.906 0.933 0.767 IgA X AURKA-GAGE1- MAGEA10-MAGEB1 0.956 1.000 0.833 XXIX AURKA -MAGEA10- MAGEB1-PAGE1 0.940 0.900 0.867 XXX AURKA -MAGEA10- MAGEB1-PLEKHA5 0.968 1.000 0.833 LXXV AURKA -GAGE1- MAGEA10-MAGEB1-PLK4 0.963 0.967 0.933 LXXVII AURKA -GAGE1- MAGEA10-PLEKHA5- XAGE3Av1 0.954 1.000 0.800 Antigen-specific autoantibody glycosylation patterns differ between PDAC and confounding cohort Having confirmed autoantibody reactivity against specific cancer-testis antigens on the microarray platform, whether there were differential glycan moieties on the antigen-specific autoantibodies in PDAC and controls was determined. This was achieved by adapting a fluorescently-labelled lectin-based protocol from published bead-based assays for detecting antibody glycosylation patterns (Li, C., et al., A multiplexed bead assay for profiling glycosylation patterns on serum protein biomarkers of pancreatic cancer. Electrophoresis, 2011. 32(15): p. 2028-35), profiling the glycoforms present on antigen-bound autoantibodies, detecting with 1pg / ml of RCA (P1,4-Gal), ECL(P1,4-Gal), SNA (a2,6-SA), or LCA(a1,6-Fuc), for the presence of each respective glycoform. As above, a threshold was set and a true positive signal was defined as any relative fluorescent unit (RFU) that was above the mean+2SD of the control RFU. Thereafter, the mean ± standard error of the mean (SEM) for all the positive antigens was used as a measure of the total galactosylation, sialylation, and fucosylation levels in each cohort. The investigated glycoforms had uniform low signal intensities across all the lectin-specific glycans for the control group (Figures 2A-D), further validating the specificity of the assay on a cancer antigen microarray platform. Moreover, the data shows relatively high sialylation (Figure 2B), and fucosylation (Figure 2C) for the PDAC cohort compared to the pooled DYS+CP cohort. These results suggest that although the antigen-bound autoantibodies in the pooled DYS+CP cohort may be sialylated and fucosylated, the carbohydrate content varies between the two disease conditions, favoring increased a2,6-Sal and a1,6-Fuc in PDAC. However, the galactosylation levels show contrasting outcomes respectively, showing high (Figure 2A) and low (Figure 2D) galactosylation for PDAC compared to the pooled DYS+CP group. Although both RCA and ECL have selectivity for pi ,4-Gal, their respective binding affinity is impacted by the presence of other carbohydrate groups, particularly sialylation for ECL, suggesting that the glycosylation features between PDAC and pooled DYS+CP diseases of the pancreas are notably distinct. Furthermore, the relationship between the observed glycan motifs between PDAC and the pooled DYS+CP group was assessed. Significance testing was followed by pairwise testing, adjusting the antigen significance level based on probability value (p-value) rank order by employing the stringent Bonferroni correction to limit the risk of a type I error from multiple pairwise tests performed on the data. Subsequently, each disease condition was plotted against the corrected significant antigen count for each lectin (Figure 2E). The results show a marginal increase for RCA binding (p-value =0.553), and a significant increase for SNA (p-value <0.0001) and LCA binding (p-va / ue<0.0001) for PDAC. In agreement with the above galactosylation data, a significant (p-value <0.0001) decrease in the ECL (P1,4-Gal) motif in the PDAC cohort was observed. Therefore, there are differences in the carbohydrate composition and motifs of serum autoantibodies identified for PDAC and pooled DYS+CP diseases of the pancreas, and these may be indicative of differential Fc region glycosylation that exists between serum autoantibodies present between the two different disease states. Thereafter, based on the lectin assay data of glycan-positive autoantibodies, and the data generated from autoantibody subclass assays, unsupervised hierarchical clustering was performed to determine whether the serum autoantibody signatures of the PDAC cohort could be associated with the observed glycosylation features (Figure 2F). To ensure comparability and avoid clustering driven by technical variation between signals for the same autoantigen in different samples, the Iog2 transformed data was scaled to the standard deviation. Antibody and lectin features form distinct clusters, with IgA subclasses showing shared characteristic antigen binding with the SNA (a2,6-Sal) lectin, indicating that both interactions are similar, and represent the serum autoantibody profile of the PDAC cohort. This data thus suggests that increased sialylation in both lgA1 and lgA2 is an immune feature of PDAC. Example 2: Bead-based diagnostic method directed to IgG and IgA serum biomarker panels A Luminex bead-based assay, in which an individual patient sample is added to a mixture of streptavidin-coated, dye-encoded xMap bead types, could be used to diagnose PDAC according to the method described herein. Each unique bead type is encoded by a unique dye combination as known in the art and is derivatised with a unique BCCP-tagged CT antigen drawn from the IgG and / or IgA diagnostic panel described above. Each antigen in the diagnostic panel is thus encoded by the xMAP bead type, as will be understood by a person skilled in the art. During incubation, autoantibodies in the patient sample bind to the cognate analytes of interest on the xMAP beads. Following a wash step, anti-human IgG and antihuman IgA detection antibodies - each labelled with a different fluorophore as known in the art - are added and allowed to bind to the patient autoantibodies captured on the xMAP beads. The beads are thereafter read, for example on a dual-laser Luminex Bioplex 200® for signal quantification. Sample preparation: A patient blood sample can be collected fresh, thawed from storage at -80°C or from blood cards. Sample concentrations can be prepared to at least 1:200 in wash buffer (PBST). Sample concentration can be varied but should be within dynamic range of the assay. Bead Preparation: Each xMAP bead type is individually coated with streptavidin and is non-covalently bound to a unique enzymatically biotinylated, BCCP-tagged CT antigen from the diagnostic panel via high affinity streptavidin-biotin interactions. Post binding of the BCCP-tagged CT antigen, the bead surface is blocked with blocking buffer containing free biotin for ~30 min, with gentle shaking at RT. The beads are washed in PBST prior to performing the capture assay. Hybridisation / capture assay: Technical duplicates of each patient sample are added to the mixture of antigen-derivatised xMAP beads and incubated at room temperature to allow PDAC-specific autoantibodies to form an antigen-antibody complex with their cognate bead-bound antigen. Thereafter, awash step is performed to remove unbound- and non-specifically bound antibodies. Differentially fluorophore-labelled anti-human IgG and anti-human IgA detection antibodies are added (at pg / ml concentrations) to the bead-bound antigen-antibody complexes, incubated for ca. 1 hr at room temperature, then washed with PBST and with PBS. The beads are then read on, for example, a Bioplex 200™ or Bioplex or xMAP system (Luminex) according to manufacturer’s protocols. The signal readings are exported to a spreadsheet for data analysis. Definitions used herein: The use of the word “a” or “an” when used in conjunction with the term “comprising” in the claims and / or the specification may mean “one,” but it is also consistent with the meaning of “one or more,” “at least one,” and “one or more than one.” The word “about” means plus or minus 5% of the stated number. Unless the context requires otherwise, the word “comprise” or variations such as “comprises” or “comprising” will be understood to imply the inclusion of a stated integer or group of integers but not the exclusion of any other integer or group of integers. Antigens: GAGE1 = G antigen 1 ACVR2B = Activin receptor type 11B LEMD1 = LEM domain containing 1 MAGEB1 = Melanoma-associated antigen family, member B1 PAGE1 = Prostate-associated antigen family, member 1 TSGA10 = Testis-specific gene antigen 10 BAGE4 = B melanoma antigen family, member 4 AURKA = Aurora kinase A MAGEA10 =Mage family, member A10 PLEKHA5 = Pleckstrin homology homain containing A5 PLK4 = Polo-like kinase 4 XAGE3Av1 = X antigen family, member 3 variant 1

Claims

1. A method of diagnosing pancreatic ductal adenocarcinoma (PDAC) in a patient, the method comprising detecting an antibody which binds to GAGE1 and an antibody or antibodies which bind to at least one other antigen selected from the group consisting of ACVR2B, LEMD1, MAGEB1, PAGE1, TSGA10, BAGE4, AURKA, MAGEA10, PLEKHA5, PLK4 and XAGE3Av1 in a biological sample from the patient.

2. The method of claim 1, wherein the antibodies are IgG and / or IgA antibodies.

3. The method of either of claims 1 or 2, wherein the biological sample is a blood sample.

4. The method of any one of claims 1 to 3, which comprises the steps of:contacting the biological sample with GAGE1 and at least one other antigenselected from the group consisting of ACVR2B, LEMD1, MAGEB1, PAGE1, TSGA10, BAGE4, AURKA, MAGEA10, PLEKHA5, PLK4 and XAGE3Av1; and detecting binding of antibodies to the antigens.

5. The method of claim 4, wherein binding of the antibodies to the antigens is detected using labeled anti-human IgG antibodies and / or anti-human IgA antibodies.

6. The method of either of claims 4 or 5, which further comprises the step of quantifying the levels of each antibody in the sample and comparing these to levels associated with healthy subjects or to levels associated with chronic pancreatitis.

7. The method of any one of claims 4 to 6, wherein the antigens are bound to a support.

8. The method of claim 7, wherein the support is a bead, a membrane, a slide, a plate, awell, a filter or a dipstick.

9. The method of any one of claims 5 to 7, wherein the anti-human IgG antibodies and antihuman IgA antibodies are fluorescently labeled.

10. The method of any one of claims 1 to 7, which comprises detecting antibodies which bind to GAGE1 and at least two, at least three, at least four, at least five, at least six, at least seven, at least eight, at least nine, at least ten, at least eleven, or twelve ofACVR2B, LEMD1, MAGEB1, PAGE1, TSGA10, BAGE4, AURKA, MAGEA10, PLEKHA5, PLK4 and XAGE3Av1.

11. The method of any one of claims 1 to 8, which comprises detecting IgG antibodies which bind to GAGE1 and at least one other antigen selected from the group consisting of ACVR2B, LEMD1, MAGEB1, PAGE1, TSG10 and BAGE4.

12. The method of claim 11, which comprises detecting IgG antibodies which bind toGAGEI and at least two, at least three, at least four, at least five, or at least six other antigens selected from the group consisting ofACVR2B, LEMD1, MAGEB1, PAGE1, TSG10 and BAGE4.

13. The method of any one of claims 1 to 12, which comprises detecting IgG antibodies which bind to:i) GAGE1, ACVR2B, LEMD1, MAGEB1 and PAGE1;ii) GAGE1, LEMD1, MAGEB1 and PAGE1;iii) GAGE1, LEMD1, MAGEB1 andTSGAIO;iv) ACVR2B, BAGE4, GAGE1, LEMD1, MAGEB1 and TSGA10; orv) ACVR2B, GAGE1, LEMD1, MAGEB1 and TSGA10.

14. The method of claim 13, which comprises detecting IgG antibodies which bind to ACVR2B, GAGE1, LEMD1, MAGEB1 and TSGA10.

15. The method of any one of claims 1 to 14, which comprises detecting IgA antibodies which bind to GAGE1 or MAGEA10 and at least one other antigen selected from the group consisting of AURKA, MAGEB1, BAGE1, PLEKHA5, PLK4, XAGE3Av1, GAGE1 or MAGEA10.

16. The method of claim 15, which comprises detecting IgA antibodies which bind to GAGE1 or MAGEA10 and at least two, at least three, at least four, at least five, or at least six other antigens selected from the group consisting of AURKA, MAGEB1, BAGE1, PLEKHA5, PLK4, XAGE3Av1, GAGE1 or MAGEA10.

17. The method of any one of claims 1 to 16, which comprises detecting IgA antibodies which bind to:vi) GAGE1, AURKA, MAGEA10 and MAGEB1;vii) AURKA, MAGEA10, MAGEB1 and PAGE1;viii) AURKA, MAGEA10, MAGEB1 and PLEKHA5;ix) GAGE1, AURKA, MAGEA10, MAGEB1 and PLK4; orx) GAGE1, AURKA, MAGEA10, PLEKHA5 and XAGE3Av1.

18. The method of claim 17, which comprises detecting IgA antibodies which bind to GAGE1, AURKA, MAGEA10, PLEKHA5 and XAGE3Av1.

19. The method of any one of claims 1 to 9, which comprises detecting any one of the combinations of IgG antibodies listed in any one of claims 11 to 14 and any one of the combinations of IgA antibodies listed in any one of claims 15 to 18.

20. An antigen complex comprising:a biotin carboxyl carrier protein (BCCP)-tagged antigen bound to a support,wherein the antigen is selected from the group consisting of GAGE1, ACVR2B, LEMD1, MAGEB1, PAGE1, TSGA10, BAGE4, AURKA, MAGEA10, PLEKHA5, PLK4 and XAGE3Av1; andwherein the support is a bead, a membrane, a slide, a plate, a well, a filter or a dipstick and is coated with streptavidin,for use in the method of any one of claims 1 to 19.

21. A kit comprising:GAGE1;andone or more other antigens selected from the group consisting of ACVR2B, LEMD1, MAGEB1, PAGE1, AURKA, MAGEA10, PLEKHA5 and XAGE3Av1.

21. The kit of claim 21, wherein the antigens are bound to a support, such as a bead, a membrane, a slide, a plate, a well, a filter or a dipstick.

23. The kit of claim 22, which further comprises:labeled anti-human IgG antibodies; and / orlabeled anti-human IgA antibodies;and which optionally further includes instructions, in written or computer-implementable form, for performing the method of any one of claims 1 to 19.

24. A computer implemented method of diagnosing PDAC, the computer performing steps including:receiving inputted subject data comprising the number of antibodies detected in a biological sample from a patient, wherein the antibodies are IgG and / or IgA 5 autoantibodies bound to one or more antigens selected from the group consistingof GAGE1, ACVR2B, LEMD1, MAGEB1, PAGE1, TSGA10, BAGE4, AURKA, MAGEA10, PLEKHA5, PLK4 and XAGE3Av1;comparing the data obtained from the sample to reference data for the same autoantibodies associated with a subject without PDAC or associated with a patient10 having PDAC and thereby determining whether the subject has, or possibly has,PDAC; anddisplaying a diagnosis.Application No: GB2401743.6Claims searched: 1-14, 19-24 in partExaminer: Vanessa LuuDate of search: 16 July 2024Patents Act 1977: Search Report under Section 17Documents considered to be relevant:Category Relevant to claims Identity of document and passage or figure of particular relevance X Y X: 20, 24; Y: 1-12, 21-23 PANCREATOLOGY, vol. 2, no. 2, 2002, Bert et al., "Expression spectrum and methylation-dependent regulation of melanoma antigenencoding gene family members in pancreatic cancer cells", page 146. [online] Available from: httDs: / / www. sciencedirect. com / science / article / nii / S 14243 9030280014X? via%3Dihub (accessed 09 / 07 / 24) See abstract, "Cell Lines and Tissue Samples" on page 147, and Figure 1. Y 1-12,21-23 WO 2023 / 230562 A2 (BOARD OF REGENTS OF THE UNIV OF NEBRASKA) See page 26 and page 30, lines 18-25. Y 10, 12 WO 2020 / 056162 Al (UNIV OREGON HEALTH &SCIENCE) See page 32. A - Frontiers in oncology, vol. 14, 2024, Maimela et al., "Humoral immunoprofiling identifies novel biomarkers and an immune suppressive autoantibody phenotype at the site of disease in pancreatic ductal adenocarcinoma", article no 1330419. [online] Available from: httos: / / www.frontiersin.org / iournals / oncologv / articles / 10.3389 / fonc.202 4,1330419 / full (accessed 09 / 07 / 24) See whole document.Categories:X Document indicating lack of novelty or inventive step A Document indicating technological background and / or state of the art. Y Document indicating lack of inventive step if combined with one or more other documents of same category'. P Document published on or after the declared priority date but before the filing date of this invention. & Member of the same patent family E Patent document published on or after, but with priority date earlier than, the filing date of this application.Field of Search:International Classification:Subclass Subgroup Valid From GOIN 0033 / 533 01 / 01 / 2006 C07K 0014 / 47 01 / 01 / 2006 GOIN 0033 / 543 01 / 01 / 2006 GOIN 0033 / 574 01 / 01 / 2006 GOIN 0033 / 58 01 / 01 / 2006

Images