背景  

非小细胞肺癌（non-small cell lung cancer, NSCLC）与弥漫大B淋巴瘤（diffuse large B-cell lymphoma, DLBCL）作为临床常见的实体瘤与血液系统恶性肿瘤，其治疗仍面临显著的疗效异质性挑战。尽管免疫检查点抑制剂（immune checkpoint inhibitors, ICIs）在NSCLC治疗中取得了显著疗效，DLBCL标准R-CHOP治疗方案亦显著提高了患者生存率，但两类肿瘤均存在部分患者原发耐药或治疗后复发的临床难题。本研究综合多组学技术，系统评估了NSCLC和DLBCL患者自身抗体及代谢标志物的预后价值及其与免疫微环境的关联机制。

方法  

自身抗体标志物研究方面，第一章NSCLC队列回顾性纳入267例接受ICIs治疗患者的528份纵向血浆样本及30例福尔马林固定的石蜡包埋（formalin-fixed paraffin-embedded，FFPE）样本；第二章DLBCL队列包括336例R-CHOP治疗前血浆样本、55例FFPE样本及90例健康对照的血浆样本。通过HuProt™高密度蛋白芯片进行筛选发现，经靶向微阵列初步验证后采用ELISA最终验证自身抗体。在NSCLC中比较自身抗体与程序性死亡-配体1（programmed death-ligand1, PD-L1）表达对无进展生存期（progression-free survival，PFS）的预测价值；在DLBCL中构建基于自身抗体的风险评分模型，并与国际预后评分（International Prognostic Index，IPI）比较。结合免疫组化、单细胞、空间转录组、多重免疫荧光（multiple immunofluorescence，mIF）和功能实验，验证靶点在信使RNA（messenger RNA，mRNA）及蛋白层面的预测效能，探索其调控免疫微环境的机制。差异自身抗体进一步用于三组人群间（健康对照、NSCLC、DLBCL）通路富集分析。

代谢标志物研究方面，第三章NSCLC队列采集了134例ICIs治疗患者的血浆样本，采用非靶向UPLC-MS/MS技术筛选与免疫治疗疗效相关的代谢通路。在GEO数据集中识别支链氨基酸（branched chain amino acid，BCAA）代谢亚型，在泛癌ICIs队列中构建并验证基于关键BCAA基因的预后模型。mIF、单细胞及空间转录组学验证其在组织水平的表达及与免疫微环境的关联，进一步结合功能实验探究具体调控机制。第四章在DLBCL中整合scRNA-seq、普通转录组、空间转录组、免疫组化及mIF技术。在单细胞层面识别高恶性B细胞及其糖酵解特征，利用hdWGCNA方法揭示与肿瘤进展密切相关的糖酵解通路，并聚焦于免疫抑制型IFN_TAMs亚群的功能机制及预后意义。

结果  

在aNSCLC免疫联合化疗中，识别并验证了治疗前（MAX和DHX29）及治疗后（MAX和TAPBP）与疗效显著相关的自身抗体，均与早期治疗反应和PFS正相关。治疗反应者MAX自身抗体持续升高，而无反应者表现为先升后降的趋势。MAX在蛋白和mRNA层面均具有预后价值，并与干扰素信号通路关键分子IFNGR1/2、适应性免疫信号通路关键分子HLA-A等免疫因子表达正相关。免疫单药队列中发现以HDAC3为代表的五种预后相关自身抗体，HDAC3高表达与PFS缩短、TGFβ1及PD-L1上调、CD8⁺T细胞浸润降低显著相关。HDAC3敲降能够抑制肺癌细胞增殖，HDAC3抑制剂联合抗PD-1治疗可增强CD8⁺T细胞浸润。DLBCL队列中，识别出CREB1、N4BP1、UBAP2和DEAF1四种自身抗体，均与EFS24及较好的PFS和总生存期相关。基于CREB1和N4BP1自身抗体构建的风险评分的预测效能优于IPI评分，具有独立预测价值。CREB1和N4BP1在蛋白和mRNA层面均具有预后价值。功能实验显示CREB1敲降可促进细胞增殖、上调C-myc、下调P50蛋白表达。三组人群间差异自身抗体多富集于代谢通路，提示免疫代谢在肿瘤进展中的关键作用。

NSCLC代谢组学研究发现，血浆L-亮氨酸水平及BCAA代谢通路与免疫治疗疗效显著相关。关键BCAA基因HMGCS1在ICIs队列中表现出独立预后价值，与CD8⁺ T细胞浸润负相关，并参与调控胆固醇代谢及BCAA降解，以上结果通过多组学和功能实验得到了验证。动物实验证实靶向干预HMGCS1可抑制肿瘤进展。在DLBCL中识别出了与恶性程度相关的STMN1、ENO1、PKM和CDK1糖酵解基因，其高表达与高糖酵解IFN_TAMs（CD68⁺CXCL10⁺PD-L1⁺）富集、低水平CD8⁺T细胞浸润及免疫抑制微环境形成相关。IFN_TAMs高表达CD274、PFKFB3和TGFB1，与高恶性DLBCL细胞存在密切的相互作用，其浸润水平及糖酵解基因构建的风险评分均与不良生存结局相关。

结论  

MAX、HDAC3、CREB1和N4BP1等自身抗体可分别作为NSCLC和DLBCL患者预后预测的潜在血液标志物，HMGCS1及糖酵解相关代谢基因在调控肿瘤免疫代谢微环境及影响免疫治疗疗效中发挥关键作用。本研究为肿瘤精准免疫治疗提供了新的预后指标和潜在靶点。

Background

Non-small cell lung cancer (NSCLC) and diffuse large B-cell lymphoma (DLBCL) are among the most common solid tumors and hematologic malignancies in clinical practice, respectively. However, both diseases still face significant challenges due to heterogeneity in treatment response. Despite the notable efficacy of immune checkpoint inhibitors (ICIs) in NSCLC and the substantial survival benefit achieved with the standard R-CHOP regimen in DLBCL, a proportion of patients experience primary resistance or relapse after treatment. This study employed integrative multi-omics approaches to systematically evaluate the prognostic value of autoantibodies and metabolic biomarkers in NSCLC and DLBCL, and to elucidate their mechanistic associations with the tumor immune microenvironment.

Methods

For the investigation of autoantibody biomarkers, the NSCLC cohort retrospectively included 528 longitudinal plasma samples from 267 patients treated with ICIs, along with 30 formalin-fixed paraffin-embedded (FFPE) tissue samples. The DLBCL cohort consisted of 336 pre-treatment plasma samples, 55 FFPE samples, and plasma from 90 healthy controls. Autoantibodies were initially screened using the HuProt™ high-density protein microarray, preliminarily validated with targeted microarrays, and finally confirmed by ELISA. In NSCLC, the predictive value of autoantibodies for progression-free survival (PFS) was compared to that of PD-L1 expression. In DLBCL, a risk score model based on autoantibody expression was constructed and compared with the International Prognostic Index (IPI). Validation of target biomarkers at the mRNA and protein levels and exploration of their roles in modulating the immune microenvironment were conducted through immunohistochemistry, single-cell transcriptomics, spatial transcriptomics, multiple immunofluorescence (mIF), and functional assays. Differentially expressed autoantibodies were further used for pathway enrichment analyses across healthy controls, aNSCLC, and DLBCL populations.

For the analysis of metabolic biomarkers, plasma from 134 NSCLC patients treated with ICIs was analyzed using untargeted UPLC-MS/MS to identify immunotherapy-related metabolic pathways. A branched-chain amino acid (BCAA) metabolic subtype was identified from GEO datasets, and a prognostic model based on key BCAA-related genes was constructed and validated in pan-cancer ICI-treated cohorts. Tissue-level expression and immune microenvironment correlations were assessed by mIF, single-cell, and spatial transcriptomic analyses, and underlying regulatory mechanisms were further explored with functional experiments. In DLBCL, scRNA-seq, bulk transcriptomics, spatial transcriptomics, immunohistochemistry, and mIF were integrated. Highly malignant B cells and their glycolytic features were identified at the single-cell level. Using hdWGCNA analysis, glycolytic pathways closely associated with tumor progression were revealed, focusing on the functional role and prognostic significance of the immunosuppressive IFN_TAMs subpopulation.

Results

In aNSCLC patients receiving chemoimmunotherapy, autoantibodies to MAX and DHX29 (pre-treatment), as well as MAX and TAPBP (post-treatment), were identified and validated as significantly associated with treatment efficacy. These autoantibodies positively correlated with early treatment response and PFS. Sustained elevation of anti-MAX antibodies was observed in responders, whereas a rise-then-fall pattern was noted in non-responders. MAX demonstrated prognostic significance at both protein and mRNA levels, and its expression was positively correlated with immune factors such as IFNGR1/2 and HLA. In the ICI monotherapy cohort, five prognostic autoantibodies including HDAC3 were identified. High HDAC3 expression was significantly associated with shorter PFS, upregulation of TGFβ1 and PD-L1, and reduced CD8⁺ T cell infiltration. HDAC3 knockdown inhibited lung cancer cell proliferation, and its inhibition combined with anti-PD-1 therapy enhanced CD8⁺ T cell infiltration. In DLBCL, four autoantibodies (CREB1, N4BP1, UBAP2, and DEAF1) were associated with improved event-free survival at 24 months (EFS24), PFS, and overall survival. A risk score based on CREB1 and N4BP1 outperformed IPI in prognostic prediction and showed independent prognostic value. Both proteins also demonstrated prognostic relevance at the mRNA and protein levels. Functional experiments revealed that CREB1 knockdown promoted cell proliferation, upregulated C-myc, and downregulated P50 protein expression. Differential autoantibodies between the three groups (healthy controls, aNSCLC, and DLBCL) were enriched in metabolic pathways, underscoring the key role of immunometabolism in tumor progression.

Metabolomic profiling in NSCLC revealed significant associations between plasma L-leucine levels, BCAA metabolic pathways, and immunotherapy efficacy. The key BCAA gene HMGCS1 exhibited independent prognostic value in ICI-treated cohorts, was negatively correlated with CD8⁺ T cell infiltration, and regulated cholesterol metabolism and BCAA degradation, as validated by multi-omics and functional studies. Animal experiments confirmed that targeted inhibition of HMGCS1 could suppress tumor progression. In DLBCL, glycolytic genes such as STMN1, ENO1, PKM, and CDK1 were identified as malignancy-associated and were linked to enrichment of highly glycolytic IFN_TAMs (CD68⁺CXCL10⁺PD-L1⁺), decreased CD8⁺ T cell infiltration, and an immunosuppressive microenvironment. IFN_TAMs showed high expression of CD274, PFKFB3, and TGFB1, with close interactions with high-malignancy DLBCL cells. Their infiltration levels and glycolysis-based risk scores were associated with poor survival outcomes.

Conclusion

Autoantibodies such as MAX, HDAC3, CREB1, and N4BP1 may serve as potential prognostic blood biomarkers in NSCLC and DLBCL. HMGCS1 and glycolysis-related metabolic genes play critical roles in regulating the tumor immunometabolic microenvironment and influencing immunotherapy outcomes. This study provides novel prognostic indicators and potential therapeutic targets for precision immunotherapy in cancer.