Combination detection of IgG- and IgA-related autoantibodies for the early diagnosis of gastric cancer.

Introduction
Gastric cancer (GC) is the fifth most common cancer worldwide and the third leading cause of cancer-related deaths.
1
The prognosis for advanced-stage GC remains poor, with a 5-year survival rate of less than 15%. In contrast, early-stage diagnosis significantly improves outcomes, with a 5-year survival rate of up to 75%. However, early detection is challenging due to the lack of specific symptoms in the initial stages and the absence of reliable, non-invasive screening methods. Among the major risk factors for GC, chronic inflammation plays a critical role, with Helicobacter pylori (H. pylori) infection as the leading cause. H. pylori is classified as a Group 1 carcinogen, contributing to GC development through persistent immune activation and epithelial damage.
2
Infection induces a strong inflammatory response, driven by cytokines such as IL-6, IL-8, and TNF-α, which promote immune cell infiltration and tissue remodeling. Strains carrying the CagA virulence factor further exacerbate oncogenesis by disrupting host signaling pathways and increasing genomic instability.
3
Despite its well-established role in GC, current diagnostic approaches remain reliant on invasive methods such as endoscopic examination and histological analysis, which are impractical for large-scale screening. This highlights an urgent need for novel, non-invasive biomarkers that can be detected in body fluids. Ideally, these biomarkers should offer high sensitivity and specificity, enabling the detection of early-stage GC before the onset of symptoms and disease spread, thereby improving patient outcomes and facilitating broader screening efforts.
The early detection and diagnosis of GC remain a significant challenge due to the limitations of current biomarkers. Serum antigen biomarkers, such as carcinoembryonic antigen (CEA), carbohydrate antibody 19-9 (CA19-9), carbohydrate antibody 72-4 (CA72-4), and carbohydrate antibody 50 (CA50), are widely used due to their accessibility and high titers, even in early-stage cancers.
4
However, these biomarkers often lack sufficient sensitivity and specificity, limiting their clinical application. To address this issue, an alternative approach focuses on detecting autoantibodies, which are produced in response to tumor-associated antigens (TAAs) and offer greater stability in circulation. Notably, IgG autoantibodies against specific TAAs are detectable in the blood approximately five years before the clinical manifestation of cancer, highlighting their potential to predict early-stage disease.
5
By targeting these immune responses, autoantibody-based diagnostics could overcome the limitations of conventional biomarkers and improve early detection strategies for GC.
Among the different immunoglobulin classes of autoantibodies, IgG and IgA hold particular promise due to their distinct yet complementary roles in systemic and mucosal immunity. IgA, primarily involved in mucosal defense, is produced at epithelial surfaces such as the gastric lining, making it a key marker for localized immune responses to TAAs.6,7 In contrast, IgG reflects systemic immune activation and immune memory.
8
Prior studies have demonstrated the diagnostic value of these immunoglobulins, including in breast cancer, where distinct IgG and IgA repertoires served as biomarkers,
9
and in oral cancer, where elevated levels correlated with tumor-specific immune responses.
10
Similarly, IgG and IgA autoantibody panels have shown promise for early detection of colon
11
and lung cancer,
12
reinforcing their superior diagnostic potential over IgM. Notably, Yu et al.
12
demonstrated that IgA and IgG autoantibodies against TIF1γ provided a stronger humoral immune response and better diagnostic performance in lung cancer than IgM, with IgA emerging as the most reliable biomarker. This further supports the rationale for focusing on IgA and IgG autoantibodies, as their combined detection enhances sensitivity and specificity in cancer diagnosis. By assessing this ratio, this study captures both localized mucosal immunity (IgA) and systemic immune changes (IgG), improving diagnostic accuracy. IgM autoantibodies were excluded due to their transient nature and lower stability in circulation, making them less reliable for biomarker detection. Given the influence of H. pylori-induced inflammation on these immune responses, this study also incorporates H. pylori-associated antigens (HpCagA and Hp0305) alongside key inflammatory mediators to evaluate their diagnostic relevance in gastric cancer.
In this study, we evaluated the diagnostic performance of serum IgA and IgG autoantibodies against 27 antigens using protein microarrays in a cohort of 265 healthy controls (HCs), 296 gastric cancer (GC) patients, and 195 gastritis patients. The 27 antigens selected in this study play essential roles in inflammation, immune regulation, and tumor progression. Cytokines and chemokines (IL-6, IL-8, IL-11, IL-18, IL-1β, TNF-α, MCP1, MCP3, MCP4, RANTES, MIP1β, MIP3α) were included due to their involvement in chronic inflammation, immune cell recruitment, and tumor microenvironment modulation. For example, IL-6 and TNF-α drive tumor initiation via immune evasion and angiogenesis,
2
while IL-8 and MCP1 promote immune infiltration and correlate with poor prognosis. Proteins involved in extracellular matrix remodeling, including MMP1, MMP7, MMP9, TIMP-1, ADAM10, and ADAM17, contribute to tumor invasion and metastasis.
3
Other antigens were included for their critical roles in immune activation, tumor progression, and metabolic regulation, reflecting their broader relevance in gastric cancer pathogenesis. Finally, HpCagA and Hp0305, included for their relevance to H. pylori-driven gastric cancer, contribute to chronic inflammation and oncogenic signaling.
In this study, we aimed to evaluate the diagnostic potential of IgG/IgA autoantibody ratios by developing predictive models for distinguishing GC from HCs, GC from gastritis, and subtypes of GC from subtypes of gastritis. Using supervised machine learning approaches, we identified optimal autoantibody panels based on IgG/IgA ratios and validated their performance in datasets. Additionally, we explored the correlations between the IgG/IgA ratios of Hp0305 and HpCagA with other tumor-associated antigens to assess their potential role in differentiating GC from gastritis.

Material and methods

Serum sample collection
The cohort was comprised of 756 serum samples from 296 gastric cancer (GC) patients, 265 healthy controls (HCs), and 195 patients with gastritis who had given informed consent and met the eligibility criteria (Table 1). All the GC patients and gastritis patients were collected from CHANGHAI hospital. Gastric cancer diagnosis was confirmed through endoscopic examination and histopathological biopsy, which remain the gold standard in clinical practice. Healthy control samples were collected from multiple centers, including five hospitals: CHANGHAI hospital, General hospital of Guangzhou military region, Sun Yat-Sen Memorial Hospital of Sun Yat-Sen University, Third Affiliated Hospital of Sun Yat-Sen University-Lingnan Hospital and Third Affiliated Hospital of Southern Medical University. All the healthy volunteers had no evidence of cancer history and gastric diseases. The serum samples were prepared according to the standard protocol. Briefly, 5 ml of whole blood was collected from each individual and placed in a Vacutainer (BD Biosciences) without anti-coagulant. The whole blood was then left undistributed at room temperature for 30 min. After centrifugation at 2,000x g for 10 min in a refrigerated centrifuge, the sera were then transferred as a 0.5-ml aliquot immediately to clean Eppendorf tubes and stored at −80°C until further measurements.

Recombinant antigen cloning and purification
The DNA fragments encoding 25 tumor-associated antigens and 2 Helicobacter pylori (H.pylori)-related antigens (Total 27 antigens were listed in Table 2) were cloned into the pFN19A (HaloTag®7) T7 SP6 Flexi bacterial expression vector (Promega, USA). First, a His-tag coding sequence was introduced into the 5′ region of Halo-tag coding region and then the selected antigen coding sequences were inserted downstream of the Halo-tag coding sequence. Recombinant His-Halo-tag fusion proteins were expressed in BL21-AI E. coli strain, and total cell lysate was applied to Protino® Ni-TED 150 nickel columns (Macherey-Nagel, Germany) to bind His-tagged proteins. columns were washed and bound proteins were eluted. The purity of the protein was assessed by PAGE and Coomassie blue staining; the specificity and solubility were verified by Western blot using rabbit anti-Halo-tag antibody (G9281, Promega, USA). Some of the purified antigens are shown in Supplementary Figure 1A. Antigens were diluted to concentrations of 0.03-0.1 mg/ml in PBS and supplemented with 2% of glycerol to ensure proper spot morphology. Proteins were aliquoted and stored at −80°C until use; throughout experiments, repeated thaw-freeze cycles were avoided.

Generation and processing of recombinant antigen microarrays
The 27 purified fusion proteins were diluted to concentration of 0.2 mg/ml in PBS and supplemented with 0.1% of BSA. Proteins were spotted onto glass slides (Corning, NY, USA) at a volume of 350 pl per spot at a pitch of 500 mm using a microarray printer (Scienion, Berlin, Germany). Each protein was printed duplicate. Control proteins, including biotinylated BSA, human IgG, and human IgA were also spotted in duplicate. Each glass slide contained 16 identical subarrays separated by a 16-well gasketed hybridization chamber to prevent sample cross-contamination. Individual targets of each microarray are shown and listed in Supplementary Figure 1B-C.

Detection of serum antibodies
All samples were analyzed for both IgA and IgG antibodies on the parallel recombinant antigen microarray slides using the same dilutions according to the manufacturer's instructions. Briefly, serum samples were diluted 1:200 in dilution buffer. After a 60 min incubation with blocking buffer, 100uL of 200-fold diluted serum samples were added to each well. After overnight incubation at 4°C and extensive washing, the biotin-conjugated anti-human IgG or biotin-conjugated anti-human IgA (1: 50,000) (Jackson ImmunoResearch, USA) was added for 1 h at room temperature and then washed away. Alexa Fluor 555-conjugated streptavidin was then added and incubated for 1 h at room temperature and protected from light. The signals (532 nm excitation, 635 nm emission) were scanned and extracted using an InnoScan 300 scanner (Innopsys, Carbonne, France). Raw data from the array scanner were provided as images (tif files) and spot intensities (tab-delimited.txt file) through Mapix 7.3.1 Software.

Microarray data processing and statistical analysis
Individual array spots were background-subtracted locally and normalized through positive controls. All the signal values were transformed using a log2 transformation to facilitate data analysis. To identify autoantibodies with significant changes in serum levels (based on log2-transformed values), we calculated False-Discovery Rates (FDR) for every protein through non-parametric p-value analysis using the p.adjust function in R studio. Differentially expressed autoantibodies (DEAs) were defined as those with an FDR < 0.05 and an absolute log2 fold-change >0.263. These were represented as a volcano pot using the ggplot function in R studio. ROC analysis was performed using the “pROC” package in R (R Foundation for Statistical Computing, Vienna, Austria. URL https://www.R-project.org/).
To identify a GC-specific autoantibody signature, we employed a supervised Support Vector Machine (SVM) model using the R package “caret” for classification. Model training and validation were conducted using an internal 3:1 data split, where 75% of the dataset (222 GC vs. 199 normal controls) was randomly assigned to training, while the remaining 25% (74 GC vs. 66 normal controls) was used for internal validation. To enhance model robustness and reduce overfitting, we implemented a 4-fold cross-validation scheme during training, ensuring that different subsets of data were iteratively used for training and validation. Diagnostic performance was assessed using Receiver Operating Characteristic (ROC) curve analysis, with area under the curve (AUC) values serving as the primary metric for classification accuracy. Model performance was further evaluated using sensitivity, specificity, and Youden's index to determine the optimal autoantibody panels. It is important to note that this validation was conducted within the study cohort, and no independent external dataset was used. All statistical analyses and visualizations, including heatmaps and ROC curves, were generated using RStudio, and figures were finalized using Adobe Illustrator CC5 (Adobe, San Diego, CA)

Results

Classification analysis of IgA and IgG autoantibodies between GC patients and HCs
Serum samples (265 HCs, 296 GC, and 195 gastritis) were collected at clinical centers, and the relative levels of 27 autoantibodies were examined by protein microarray. Three types of analyses were performed: individual autoantibody levels, ROC curve analysis of the prediction model, and correlations between the IgG/IgA ratio of HpCagA/Hp0305 and TAAs (Figure 1(a)). Patient demographic and clinical characteristics are described in Table 1 and a list of individual targets of the protein microarray is provided in Supplementary Figure 1.
Hierarchical clustering analyses were implemented to cluster subjects and autoantibodies. A heat map (Figure 1(b)) illustrates the analysis of autoantibody levels, revealing 24 differentially expressed autoantibodies (DEAs) to IgA after unsupervised clustering. This map separates samples into GC (red, left) and HCs (blue, right), as well as categorizes autoantibodies as higher in GC (red) and higher in normal controls (blue). The top 6 AUCs of IgA related TAAs (MIP1 beta, 0.797; MCP4, 0.776; MIP3 alpha, 0.776; IL6, 0.767; MCP1, 0.742; MMP7, 0.729) are shown in Figure 1(c). To develop and validate a GC diagnostic model, a supervised SVM model was conducted to classify GC patients with controls using the R package “caret”. The SVM model distinguished GC patients from HCs with an AUC of 0.934, a sensitivity of 83%, and a specificity of 89% in the training test (Figure 1(d) and (e)) and an AUC of 0.867, sensitivity of 82%, specificity of 86% in the validation test (Figure 1(f) and (g)).
Heat map representation of the autoantibody level analysis shows the 17 DEAs to IgG after unsupervised clustering between GC patients and HCs (Figure 2(a)). The top 6 AUCs of IgG related TAAs (AEG1, 0.829; IL-8, 0.823; TIMP.1, 0.821; MMP7, 0.791; IL-18, 0.789; IL-6, 0.768) are shown in Figure 2(b). The SVM model distinguished GC patients from HCs with an AUC of 0.916, a sensitivity of 87%, and a specificity of 79% in the training test (Figure 2(c) and (d)) and an AUC of 0.806, sensitivity of 84%, specificity of 76% in the validation test (Figure 2(e) and (f)).

Classification analysis of IgA and IgG autoantibodies between GC patients and gastritis patients
Heat map representation of the autoantibody level analysis shows the 20 DEAs to IgA after unsupervised clustering between GC patients and gastritis patients (Figure 3(a)). The top 6 AUCs of IgA related TAAs (MIP1 beta, 0.771; MIP3 alpha, 0.753; MCP4, 0.75; MCP1, 0.745; IGFBP2, 0.736; RANTES, 0.731) are shown in Figure 3(b). The SVM model distinguished GC patients from HCs with an AUC of 0.866, a sensitivity of 91%, and a specificity of 65% in the training test (Figure 3(c) and (d)) and an AUC of 0.792, sensitivity of 85%, specificity of 68% in the validation test (Figure 3(e) and (f)).
Heat map representation of the autoantibody level analysis shows the 23 DEAs to IgG after unsupervised clustering between GC patients and HCs (Figure 4(a)). The top 6 AUCs of IgG related TAAs (MMP7, 0.778; AEG1, 0.775; ADAM10, 0.773; ProApoA1, 0.756; TNF alpha, 0.753; IL-18, 0.752) are shown in Figure 4(b). The SVM model distinguished GC patients from HCs with an AUC of 0.917, a sensitivity of 100%, and a specificity of 49.6% in the training test (Figure 4(c) and (d)) and an AUC of 0.765, sensitivity of 100%, specificity of 54.1% in the validation test (Figure 4(e) and (f)).

Diagnostic performance of IgG/IgA ratio for single autoantibodies
To evaluate the potential serum autoantibodies as noninvasive biomarkers for the diagnosis of GC and gastritis, the ratio of IgG/IgA for each autoantibody was calculated and the diagnostic performance of each single autoantibody's ratio was examined.
For atrophic gastritis and GC-all, the IgG/IgA-ratio of MIP1 beta showed the highest diagnostic performance with an AUC of 0.81, a sensitivity of 87.9% and a specificity of 70.7%. For atrophic gastritis and diffuse-GC, the IgG/IgA-ratio of MIP1 beta antibody showed the highest diagnostic performance with an AUC of 0.801, a sensitivity of 86% and a specificity of 70.7%. For atrophic gastritis and intestinal-GC, the IgG/IgA-ratio of MIP1 beta antibody showed the highest diagnostic performance with an AUC of 0.808, a sensitivity of 88.1% and a specificity of 70.7% (Figure 5(a)).
For chronic gastritis and GC-all, the IgG/IgA-ratio of MMP7 antibody showed the highest diagnostic performance with an AUC of 0.713, a sensitivity of 50.7%, and a specificity of 84.4%. For chronic gastritis and diffuse-GC, the IgG/IgA-ratio of Hp0305 antibody showed the highest diagnostic performance with an AUC of 0.697, a sensitivity of 64%, and a specificity of 67.5%. For chronic gastritis and intestinal-GC, the IgG/IgA-ratio of MMP7 antibody showed the highest diagnostic performance with an AUC of 0.718, a sensitivity of 56.4%, and a specificity of 84.4% (Figure 5(b)).
For chronic gastritis and atrophic gastritis, the IgG/IgA-ratio of IL-11 antibody showed the highest diagnostic performance with an AUC of 0.597, a sensitivity of 64.9%, and a specificity of 61% (Supplementary Figure 2).

Diagnostic performance of autoantibody combinations based on IgG/IgA ratio
To evaluate the diagnostic performance of IgG/IgA ratio of different autoantibody combinations, we generated a series of marker combinations from 2 autoantibodies to 18 autoantibodies. The AUC, sensitivity, specificity and Youden's index for individual combinations under 3 different models (SVM, gbm, and rda) were conducted using the R package “caret”.
For the diagnosis of GC-all patients and HCs, the gbm prediction model with 14 autoantibodies presented the highest Youden's index of 1, and an AUC of 1 with a sensitivity of 100% and a specificity of 100% in the training test. It also had an AUC of 0.9018 with sensitivity of 83.72% and specificity of 85.5% in the validation test (Figure 6(a)). The 14-marker combination includes: MIP3 alpha, MIP1 beta, MCP1, MMP9, ADAM17, IGFBP2, IL-11, MCP4, IL-8, MCP3, RANTES, MMP7, GMCSF and ADAM10.
For the diagnosis of GC-all patients and gastritis-all patients, the gbm prediction model with 13 autoantibodies presented the highest Youden's index of 0.9979, and an AUC of 0.9994 with a sensitivity of 98.5% and a specificity of 99.2% in the training test and an AUC of 0.7369 with a sensitivity of 76.2% and a specificity of 85.7% in the validation test (Figure 6(b)). The 13-marker combination includes: MIP1 beta, Hp0305, MCP1, MCP4, MIP3 alpha, TNF alpha, IGFBP2, IL-11, ADAM17, MMP9, MCP3, GMCSF and RANTES.
For the diagnosis of GC (diffuse + intestinal) patients and gastritis (chronic + atrophic) patients, the svm prediction model with 17 autoantibodies presented the highest Youden's index of 0.7285, an AUC of 0.9188 with a sensitivity of 94.4% and a specificity of 78.4% in the training test and an AUC of 0.6814 with a sensitivity of 73.3% and a specificity of 70% in the validation test (Figure 6(c)). The 17-marker combination includes: MIP1 beta, MMP7, MCP1, IGFBP2, IL-11, MIP3 alpha, GMCSF, MCP4, MMP9, ADAM17, MCP3, RANTES, AEG1, TNF alpha, ADAM10, IL-18 and IL-8.

Correlation between ratio of individual autoantibodies to Hp0305 and HpCagA
As shown in Tables 2 and 3, IgG/IgA ratios of 19 TAAs in both chronic gastritis and atrophic gastritis were significantly correlated with H. pylori-related autoantibodies (HpCagA and Hp0305). The ratio of HpCagA autoantibody showed the highest correlation with the ratio of MMP7 in both atrophic gastritis (r = 0.902, p = 7.6X10−16) and chronic gastritis (r = 0.837, p = 2.6X10−21) groups. The ratio of Hp0305 autoantibody showed the highest correlation with the ratio of MMP7 in chronic gastritis (r = 0.756, p = 2.2 X10-15), while it showed the highest correlation with the ratio of ADAM17 in atrophic gastritis (r = 0.869, p = 1.82X10-13).
The IgG/IgA ratio of 16 TAAs was significantly correlated with the ratio of HpCagA, and 11 ratios of TAAs were significantly correlated with the ratio of Hp0305 in the diffuse-GC group (Table 4). The ratio of HpCagA autoantibody showed the highest correlation with the ratio of AEG1 in the diffuse-GC group (r = 0.598, p = 1.2X10−9). The ratio of Hp0305 autoantibody showed the highest correlation with the ratio of GMCSF in the diffuse-GC group (r = 0.404, p = 1.1X10−4).
The IgG/IgA ratios of 19 TAAs were significantly correlated with the ratio of HpCagA, and 18 ratios of TAAs were significantly correlated with the ratio of Hp0305 in the intestinal-GC group (Table 5). The ratio of H. pylori related autoantibodies showed the highest correlation with the ratio of MMP7 in the intestinal-GC group (HpCagA, r = 0.666, p = 2.8X10−14; Hp0305, r = 0.57, p = 4.9X10−10).

Discussion
During early tumorigenesis, B cell clones differentiate into plasma cells that can produce either IgG or IgA antibodies targeting TAAs.
13
These autoantibodies against TAAs present with high stability and persistence in sera and are easy to detect. Such excellent characteristics make it possible for TAA autoantibodies to serve as promising serological markers for the diagnosis of early GC.
14
However, the majority of previous studies have focused on IgG autoantibodies, whereas the role of IgA autoantibodies remains underexplored despite their predominant presence in gastric mucosa. Given that gastric mucosal linings produce more IgA than any other type of antibodies,
15
this study aimed to fill this gap by investigating both IgG and IgA autoantibody responses against a panel of 27 TAAs. Previous studies have also demonstrated that IgA autoantibodies, such as those against dihydrolipoamide dehydrogenase, serve as biomarkers for other cancers, including endometrial cancer.
16

Key findings and diagnostic potential of IgA
and IgG autoantibodies
This study identified distinct IgA and IgG autoantibody profiles in GC, suggesting their diagnostic value. We found that IgA and IgG autoantibody responses were significantly different (Figures 1 and 2). For IgA autoantibodies, MIP1 beta demonstrated the highest diagnostic performance in distinguishing GC patients from both HCs (AUC = 0.797; Figure 1(c)) and gastritis patients (AUC = 0.771; Figure 3(b)). For IgG autoantibodies, AEG1 showed the strongest separation of GC from HCs (AUC = 0.829; Figure 2(b)), while MMP7 performed best in differentiating GC from gastritis (AUC = 0.778; Figure 4(b)). Notably, when evaluating the IgG/IgA ratio, MIP1 beta had the highest diagnostic accuracy in differentiating atrophic gastritis from GC (AUC = 0.81), as well as diffuse (AUC = 0.801) and intestinal GC (AUC = 0.808).

Biological and clinical relevance of MIP1 beta, AEG1, and MMP7
MIP1 beta (CCL4), a CC chemokine, is known to promote tumor progression in various cancers.
17
Its high expression in diffuse-type GC9,18 and its significant increase in GC tissues compared to controls4,19 further support its role in gastric tumorigenesis. Additionally, a previous study demonstrated that MK2 expression, which regulates inflammation and metastasis, is strongly correlated with MIP1 beta, and can classify GC metastasis with an accuracy of 85.7%.
20

Similarly, AEG1 (metadherin/LYRIC) has been reported as a crucial oncogene in GC, with overexpression correlating with poor prognosis.
21
AEG1 is also closely associated with angiogenesis
22
and GC progression via activation of the eIF4E/cyclin D1 signaling pathway.
23
Our findings suggest that IgG and IgA autoantibodies against MIP1 beta and AEG1 could serve as non-invasive biomarkers, providing an alternative means of detecting their involvement in GC progression.
MMP7, a matrix metalloproteinase, plays a critical role in both GC progression and H. pylori-associated inflammation. In this study, we found that among all tumor-associated antigens, the IgG/IgA ratios of H. pylori-related autoantibodies (HpCagA and Hp0305) showed the strongest correlations with MMP7, suggesting a key link between chronic H. pylori infection and MMP7-driven tumorigenesis. MMP7 is known to contribute to GC invasion, metastasis, and TNM staging.24,25 Its overexpression is an independent prognostic marker, with increased serum levels correlating with poor survival in GC patients.19,26 Additionally, MMP7 has been implicated in the progression of H. pylori-induced premalignant lesions, with its deficiency shown to exacerbate gastric inflammation and promote M1 macrophage polarization.
27
The World Health Organization (WHO) categorized H. pylori as a group 1 carcinogen.
28

H. pylori accounts for more than 60% of gastric cancers,
29
and its eradication is associated with a reduced GC incidence across different clinical scenarios.
30
Our study found that the ratio of H. pylori-related autoantibodies (HpCagA and Hp0305) showed higher correlations to TAAs in gastritis patients than in GC patients, with MMP7 showing the highest correlation in both atrophic and chronic gastritis. Given that MMP7 is both an indicator of tumor aggressiveness and a mediator of H. pylori-associated inflammatory responses, its strong correlation with H. pylori-related autoantibodies in this study suggests that the IgG/IgA ratio of MMP7 may serve as a particularly effective biomarker for distinguishing GC from HC and gastritis.

Strengthening diagnostic performance with IgG/IgA ratios
Given the complementary roles of IgA and IgG autoantibodies, we assessed whether their combined detection could enhance diagnostic performance. We found that integrating IgG/IgA ratios into biomarker panels significantly improved accuracy. For GC versus HCs, the 14-marker panel achieved an AUC of 1 with 100% sensitivity and specificity in the training set. Similarly, the 13-marker panel for distinguishing GC from gastritis reached an AUC of 0.9994 with 98.5% sensitivity and 99.2% specificity (Figure 4(a) and (b)). These findings suggest that IgG/IgA ratios could serve as robust, non-invasive diagnostic indices for GC, potentially outperforming conventional biomarkers. Given these promising diagnostic performances, understanding the underlying variability in IgG and IgA responses is crucial for optimizing biomarker utility.

Understanding IgA and IgG autoantibody variability in GC
In our data, we observed notable variability in IgA and IgG autoantibody responses among GC patients, suggesting that immune profiles differ significantly between individuals. This variability is likely driven by multiple biological factors, including tumor heterogeneity, disease progression, and host immune responses.
The significant differences in IgA and IgG autoantibody profiles observed in GC patients are likely driven by a combination of tumor heterogeneity, immune system modulation, and chronic inflammation. GC is a highly heterogeneous disease with distinct molecular subtypes and variable TAA expression, leading to patient-specific B-cell activation and autoantibody production.
31
As GC progresses, the immune response shifts from a mucosal IgA-dominant response to a systemic IgG-dominant response, reflecting tumor-driven immune modulation.
32
Advanced GC is associated with immune suppression and antigen masking, which can reduce autoantibody diversity.
33

Additionally, H. pylori infection and chronic inflammation contribute to antibody class switching, with localized IgA responses in early-stage disease and systemic IgG activation in advanced disease.
32
The tumor microenvironment further influences immune class switching, where some tumors promote IgA-mediated immunosuppression, while others trigger IgG-driven systemic inflammation.
9
In this study, we focused on IgG and IgA autoantibodies due to their known stability and diagnostic potential. These factors collectively highlight the importance of multi-marker autoantibody panels for accurate GC diagnosis and underscore the need for further research to better understand immune variability in GC patients.

Limitations and future directions
Despite the strong diagnostic performance observed, this study has several limitations. First, the cohort size was relatively small and sourced from only two medical centers, which may limit the generalizability of the findings. Second, we did not use an independent validation cohort due to sample availability constraints and study design limitations. While the current findings are promising, future studies should incorporate larger, multi-center cohorts with independent validation to strengthen clinical applicability. Additionally, expanding the autoantibody panel to include novel targets from different oncogenic pathways may further improve diagnostic accuracy.
A key limitation is the lack of mechanistic insights into why certain IgG/IgA autoantibodies are preferentially produced in GC. Future studies should explore the underlying immunological mechanisms, particularly the influence of the tumor microenvironment on antibody class switching and immune evasion.

Conclusion
This study demonstrates that IgA and IgG autoantibody responses differ significantly in GC patients and that integrating their ratios enhances diagnostic accuracy. MIP1 beta, AEG1, and MMP7 autoantibodies emerged as promising biomarkers, and IgG/IgA ratio-based panels exhibited near-perfect classification of GC versus non-GC cases. These findings provide a strong foundation for advancing autoantibody-based diagnostics for GC, with potential applications in non-invasive screening and early detection.

Supplemental Material

sj-docx-1-cbm-10.1177_18758592251363414 - Supplemental material for Combination detection of IgG- and IgA-related autoantibodies for the early diagnosis of gastric cancer

Supplemental material, sj-docx-1-cbm-10.1177_18758592251363414 for Combination detection of IgG- and IgA-related autoantibodies for the early diagnosis of gastric cancer by Congcong Fu, Tao Wang, Xianzhu Zhou, Jianmin Fang, Yanlin Wang, Yuxin Wang, Xiaomao Yin, Wei Zhu, Hua Dong, Yiqi Du, Shuhong Luo and Ruo-Pan Huang in Cancer Biomarkers