Supplementary Material for: PDIA5 and ARFIP1 as immunogenetic biomarkers and therapeutic targets in pancreatic neuroendocrine neoplasms: a multi-omics study integrating MR, gene expression microarray and single-cell transcriptomics

Introduction: Pancreatic neuroendocrine tumors (PanNENs) are characterized by significant clinical heterogeneity and limited therapeutic options, particularly in metastatic or recurrent cases. Identifying actionable molecular targets and biomarkers is essential for improving patient outcomes.
Methods: We employed a multi-omics approach to identify biomarkers and therapeutic targets for PanNENs. Two-sample Mendelian randomization (TSMR) was conducted using plasma proteome data from Zheng et al. and deCODE. Differential expression and weighted gene co-expression network analyses (WGCNA) were performed on the GSE73338 dataset to prioritize biomarkers. Validation included cis-eQTL data from eQTLGen and summary data-based MR with heterogeneity in dependent instrument (HEIDI) tests. Immune mediation analysis was performed, followed by transcriptomic validation using gene expression microarrays (GSE43795) and single-cell RNA sequencing (GSE256136) data. Immune infiltration was assessed using CIBERSORT, and protein expression was validated using immunohistochemistry (IHC). Pan-cancer analysis was conducted using TCGA and GTEx data, and diagnostic performance was evaluated using ROC curves and nomograms. Results: PDIA5 and ARFIP1 were identified as potential biomarkers and therapeutic targets for PanNENs. Both proteins were upregulated in PanNEN tissues and showed consistent associations with PanNEN risk through TSMR and SMR analyses. Immune mediation analysis suggested their involvement in immune modulation. Pan-cancer analysis revealed their overexpression in multiple cancer types. Diagnostic performance, evaluated using ROC curves, demonstrated the strong potential of PDIA5 and ARFIP1 in PanNEN diagnosis. Conclusion: PDIA5 and ARFIP1 are promising biomarkers and therapeutic targets for PanNEN. Further validation and clinical exploration are required.

引言：胰腺神经内分泌肿瘤（Pancreatic neuroendocrine tumors, PanNENs）具有显著的临床异质性，且治疗选择有限，在转移性或复发性病例中这一问题尤为突出。识别可靶向的分子靶点与生物标志物，对改善患者预后至关重要。
方法：本研究采用多组学方法，旨在筛选胰腺神经内分泌肿瘤的生物标志物与治疗靶点。我们利用郑等人及deCODE研究的血浆蛋白质组数据开展双样本孟德尔随机化（Two-sample Mendelian randomization, TSMR）分析；针对GSE73338数据集进行差异表达分析与加权基因共表达网络分析（weighted gene co-expression network analyses, WGCNA），以优先筛选潜在生物标志物。验证环节涵盖来自eQTLGen的顺式表达数量性状位点（cis-eQTL）数据、基于汇总数据的孟德尔随机化分析及依赖工具异质性检验（heterogeneity in dependent instrument, HEIDI）。随后开展免疫介导分析，并通过基因表达微阵列（GSE43795）与单细胞RNA测序（GSE256136）数据完成转录组验证。采用CIBERSORT评估免疫浸润情况，通过免疫组化（immunohistochemistry, IHC）验证蛋白表达水平。利用TCGA与GTEx数据集开展泛癌分析，并采用ROC曲线与列线图（nomograms）评估诊断性能。
结果：本研究鉴定出PDIA5与ARFIP1作为胰腺神经内分泌肿瘤潜在的生物标志物与治疗靶点。两种蛋白在胰腺神经内分泌肿瘤组织中均呈上调表达，且通过TSMR与SMR分析证实其与胰腺神经内分泌肿瘤风险存在稳定关联。免疫介导分析提示二者参与免疫调节过程。泛癌分析显示，二者在多种癌症类型中均存在过表达现象。ROC曲线评估的诊断性能表明，PDIA5与ARFIP1在胰腺神经内分泌肿瘤诊断中具有良好应用潜力。
结论：PDIA5与ARFIP1是胰腺神经内分泌肿瘤极具前景的生物标志物与治疗靶点，尚需开展进一步验证与临床探索。