第一部分RHOF在原发纵隔大B细胞淋巴瘤与弥漫性大B细胞淋巴瘤非特殊型中的鉴别诊断及预后作用探究 

目的：大B细胞淋巴瘤分类众多，包括发病率最高的弥漫性大B细胞淋巴瘤非特殊型（DLBCL-NOS）和其他17种类别。原发纵隔大B细胞淋巴瘤（PMBL）是一类具有独特临床、病理及分子特征的大B细胞淋巴瘤分型。由于PMBL的部分组织学/分子特征与其他淋巴瘤具有重叠之处且发病率低导致临床数据有限，因此在实际诊疗过程中，PMBL的诊断尤其与DLBCL-NOS累及纵隔的鉴别诊断具有一定挑战性。现有研究表明，Rho鸟苷三磷酸酶家族分子与B细胞发育密切相关，其中RHOF在B细胞源性肿瘤细胞中表达上调。然而，RHOF在大B细胞淋巴瘤中的研究尚处空白。因此，本研究基于前期结果拟从转录及蛋白水平探究RHOF在PMBL与DLBCL-NOS中的表达情况，并评估其临床意义。

方法：本研究纳入从GEO数据库获取的2组数据集（GSE11318和GSE87371）中共51例PMBL及375例DLBCL-NOS数据用于转录水平分析；同时回顾性收集于本院接受诊治的65例PMBL和117例DLBCL-NOS组织样本及临床资料，通过免疫组织化学染色从蛋白水平对RHOF在PMBL及DLBCL-NOS中的表达情况及临床意义进行进一步验证。

结果：在转录及蛋白表达水平，RHOF在PMBL中的表达均显著高于DLBCL-NOS（P＜0.001），免疫组化染色结果显示RHOF在PMBL中的阳性染色率为93.8%（61/65），高表达率为81.5%（53/65）。ROC曲线分析提示RHOF高表达对于PMBL及DLBCL-NOS具有较高的鉴别诊断能力（转录水平：AUC=0.913-0.94；蛋白水平：AUC=0.878）。通过与常用的PMBL诊断类标志物CD23、CD30及PD-L1的蛋白表达比较发现，RHOF高表达是一种更具灵敏性及特异性的PMBL诊断标志物（AUC（RHOF）=0.878＞AUC（CD23）=0.818＞AUC（CD30）=0.756＞AUC（PD-L1）=0.590）。基于CD23、CD30及RHOF建立的联合诊断模型对PMBL的诊断灵敏度高于上述任何单个标志物（AUC=0.939，敏感度=0.985，特异度=0.721）。另外，RHOF的表达情况在PMBL与DLBCL-NOS同时伴CD23和CD30阳性的病例中是相似的。在预后方面，RHOF转录水平高表达与PMBL较短的总生存期（OS）呈显著相关性（P=0.00037），RHOF蛋白水平高表达与PMBL较短的OS及无进展生存期（PFS）呈显著相关性（P=0.034；P=0.034）。RHOF蛋白高表达与原发肿瘤大小≥10cm （P=0.006）、分期（Ⅲ-Ⅳ期）（P=0.012）、IPI评分（≥3分）（P=0.029）及一线治疗后复发/进展（P=0.013）显著相关。

结论：本研究表明，RHOF在鉴别诊断PMBL及DLBCL-NOS方面显示出良好的、优于现有标志物的能力。RHOF在PMBL中的高表达与该疾病的诊断、预后和临床病理特征密切相关。此外，对RHOF在PMBL中表达特点的研究，有望为PMBL的治疗和预后评估提供新的靶点和思路。

关键词：RHOF；原发纵隔大B细胞淋巴瘤；弥漫性大B细胞淋巴瘤非特殊型；鉴别诊断；预后

第二部分 基于单细胞转录组分析原发纵隔大B细胞淋巴瘤的瘤内异质性及肿瘤微环境特征

目的：在大B细胞淋巴瘤中，不同于最常见的弥漫性大B细胞淋巴瘤（DLBCL），原发纵隔大B细胞淋巴瘤（PMBL）是一种具有独特分子特征的侵袭性淋巴瘤，关于其瘤内异质性、细胞起源及肿瘤微环境的作用尚不明确。本研究旨在通过单细胞转录组测序技术，比较PMBL与DLBCL和反应性淋巴结（rLN）的组分差异，系统解析PMBL的恶性B细胞异质性、肿瘤微环境特征及细胞间互作通路，为PMBL的分子分型、起源探索及靶向治疗提供理论依据。

方法：收集4例PMBL患者的肿瘤组织样本，通过石蜡包埋组织单细胞RNA测序技术获取PMBL单细胞转录组数据，并整合来自heiDATA数据库的3例DLBCL和3例rLN的公开单细胞测序数据集。利用Seurat（v5.0）进行数据质控、标准化、批次校正及降维聚类，通过CopyKAT区分恶性和非恶性B细胞，采用GSVA、CytoTRACE2、Monocle3和CellChat分析细胞功能异质性、分化轨迹及细胞间通讯。

结果：本研究共获得50145个高质量单细胞数据用于分析。PMBL肿瘤微环境中髓系细胞浸润占比（16.18%）明显高于DLBCL组（0.49%），且具有独特的成纤维细胞成分（7.29%），而DLBCL及rLN中缺乏成纤维细胞成分。PMBL恶性B细胞可分为5个亚群（B cells-1至5），其中B cells-1为优势亚群，显示出代谢活跃且显著激活mTORC1信号通路的特点，B cells-2为增殖活跃且MYC通路激活的肿瘤祖细胞，B Cells-3至5分别具有免疫调节、抗原呈递和初始分化的特征。PMBL肿瘤微环境中的成纤维细胞可分为4类：基质型、免疫调节型（优势亚群）、肿瘤样及血管相关型。成纤维细胞主要通过COL1A1/COL6A3-CD44和LAMA2/LAMB1-CD44与恶性B细胞互作，并通过COMPLEMENT和MIF通路调控髓系细胞功能，形成基质-免疫协同的肿瘤微环境。在PMBL中，T细胞分为CD8+中央记忆型（优势亚群）、细胞毒性效应型、增殖性、辅助性等4类亚群；髓系细胞被分为经典树突状细胞1型、浆细胞样树突状细胞、活化/肿瘤相关树突状细胞、基质重塑型巨噬细胞（优势亚群）及促炎型巨噬细胞。

结论：本研究首次系统描绘了PMBL的单细胞转录组图谱，揭示了其恶性B细胞的分子异质性及与成纤维细胞主导的微环境互作模式，也为PMBL的细胞起源提供了新证据。靶向MYC/mTOR通路、成纤维细胞亚群特异性标记或免疫检查点干预可能改善PMBL治疗策略，尤其是为复发/难治患者提供了可能的治疗新思路。

关键词：原发纵隔大B细胞淋巴瘤；单细胞测序；肿瘤异质性；肿瘤微环境

第三部分 基于免疫微环境探究含PD-1单抗方案治疗复发/难治性大B细胞淋巴瘤的疗效预测因素

目的：复发/难治性（R/R）大B细胞淋巴瘤（LBCL）患者预后较差，免疫治疗（如含PD-1单抗方案）为这类患者提供了治疗选择。根据临床试验结果，PD-1单抗已被美国食品及药物管理局批准用于治疗R/R原发纵隔大B细胞淋巴瘤（PMBL）患者；在实际诊疗过程中，含PD-1单抗方案也已应用于弥漫性大B细胞淋巴瘤（DLBCL）患者的后线治疗。免疫微环境在肿瘤免疫治疗中起着关键作用，可能影响治疗反应和生存。本研究旨在探讨免疫微环境中细胞成分、免疫检查点标志物表达情况及免疫分型与R/R LBCL患者免疫治疗疗效之间的关系，为个性化治疗及分层提供依据。

方法：本研究回顾性收集了2007年8月至2022年3月在本院首次就诊并在后续接受PD-1单抗联合利妥昔单抗作为挽救治疗方案的46例R/R DLBCL和9例R/R PMBL患者的临床病理资料。经样本评估，对33例R/R DLBCL和7例R/R PMBL进行多重免疫荧光染色实验，包括Panel 1（PAX5、CD4、CD8、PD-1及PD-L1）和Panel 2（CD68、CD163、HLA-DR、SPARC及PD-L1）。使用HALO图像分析软件计算各表型细胞密度并比较不同疗效分组中的细胞成分、免疫检查点标志物表达及免疫分型差异。

结果：分组分析显示，R/R DLBCL缓解组与年龄≤60岁（P=0.034）、Ki-67指数＜70%（P=0.037）、伴BCL-2/C-MYC双表达（P=0.025）及一线治疗有效（P=0.012）显著相关。分子特征方面，TP53突变（61.3%）、KMT2D（25.8%）和MYD88（22.6%）为R/R DLBCL患者的高频突变；NOTCH1和B2-MG突变仅见于疗效不佳组，而TET2和EP300突变仅见于缓解组。多因素分析证实，免疫治疗疗效差、MYC易位及一线治疗未达完全缓解是治疗后OS的独立危险因素。免疫微环境细胞分析表明，CD8+T细胞（P＜0.001）和CD68+SPARC+巨噬细胞（P＜0.01）密度在缓解组显著升高，而M2型巨噬细胞密度在疗效不佳组显著升高（P＜0.05）。PD-L1在B细胞及巨噬细胞上的表达均与免疫治疗疗效显著相关，另外，当TPS≥10.2%（AUC=0.865）或CPS≥5.6（AUC=0.901）时，R/R DLBCL患者行PD-1单抗联合利妥昔单抗治疗效果更好。基于PD-L1+细胞和CD8+T细胞密度的微环境分型显示，不同免疫微环境分型的患者免疫治疗应答率显著不同，PD-L1高/CD8高型（Type I）患者缓解率最高，PD-L1低/CD8低型（Type II）最低。Type I型（P＜0.0001）及PD-L1高/CD8低或PD-L1低/CD8高型（Type III）（P=0.027）病例的治疗后PFS均显著优于Type II型。R/R PMBL因样本量小未能深入分析，但部分结果显示出与R/R DLBCL分析结果相似的趋势，一定程度上起到了验证的作用。

结论：R/R LBCL行PD-1单抗联合利妥昔单抗的疗效与免疫微环境特征密切相关，如CD8+T细胞、CD68+SPARC+巨噬细胞高密度浸润及PD-L1高表达提示具有治疗优势。基于PD-L1+细胞和CD8+T细胞的免疫微环境分型显示出可作为疗效预测工具为临床决策提供指导的作用。本研究为R/R LBCL免疫治疗精准分层提供了免疫微环境层面的依据，并助力个性化治疗策略的制定，但仍需进一步扩大队列验证相关结论。

Part Ⅰ Exploration of the Differential Diagnosis and Prognostic Role of RHOF in Primary Mediastinal Large B-Cell Lymphoma and Diffuse Large B-Cell Lymphoma, Not Otherwise Specified

Objective: Large B-cell lymphomas encompass diverse subtypes, including the most common diffuse large B-cell lymphoma, not otherwise specified (DLBCL-NOS), and 17 other categories. Primary mediastinal large B-cell lymphoma (PMBL) is a distinct subtype with unique clinical, pathological, and molecular characteristics. Due to overlapping histologic/molecular features with other lymphomas and limited clinical data caused by its low incidence, differentiating PMBL from DLBCL-NOS with mediastinal involvement remains challenging in clinical practice. Current evidence suggests that Rho GTPase family members are closely associated with B-cell development, and RHOF is upregulated in B-cell-derived malignancies. However, the role of RHOF in large B-cell lymphomas remains unexplored. This study investigates the expression and clinical significance of RHOF in PMBL and DLBCL at both transcriptional and protein levels.

Methods: Transcriptomic data (51 PMBL and 375 DLBCL-NOS cases) from two GEO datasets (GSE11318和GSE87371) were analyzed. Additionally, 65 PMBL and 117 DLBCL-NOS tissue samples with clinical data from our hospital were retrospectively collected. The expression and clinical significance of RHOF in PMBL and DLBCL-NOS were further verified by immunohistochemistry (IHC).

Results: RHOF expression was significantly higher in PMBL than in DLBCL-NOS at both transcriptional and protein levels (P < 0.001). IHC revealed RHOF positivity in 93.8% (61/65) of PMBL cases, with high expression in 81.5% (53/65). ROC analysis demonstrated strong diagnostic utility for RHOF in distinguishing PMBL from DLBCL-NOS (transcriptional AUC = 0.913–0.94; protein AUC = 0.878). Compared to conventional PMBL markers (CD23, CD30, PD-L1), RHOF showed superior sensitivity and specificity (AUC(RHOF)=0.878＞AUC(CD23)=0.818＞AUC(CD30)=0.756＞AUC(PD-L1)=0.590). A diagnostic model combining CD23, CD30, and RHOF achieved higher sensitivity (AUC=0.939, sensitivity=0.985, specificity=0.721). Notably, RHOF expression remained consistent in PMBL and DLBCL-NOS cases co-expressing CD23/CD30. Prognostically, high expression of RHOF transcription level was significantly correlated with poor overall survival (OS) of PMBL (P=0.00037), and high expression of RHOF protein level was significantly correlated with poor OS and progression-free survival (PFS) of PMBL (P=0.034; P=0.034). High RHOF protein expression was associated with tumor size ≥10 cm (P = 0.006), advanced stage (III–IV; P = 0.012), high IPI score (≥3; P = 0.029), and recurrence/progression after first-line treatment (P=0.013).

Conclusions: RHOF exhibits superior diagnostic performance over existing markers in differentiating PMBL from DLBCL-NOS. Its high expression in PMBL is closely linked to diagnosis, prognosis, and clinicopathological features. In addition, the study on the expression characteristics of RHOF in PMBL is expected to provide new target and biomarker for the treatment and prognosis evaluation of PMBL.

Keywords: RHOF; Primary mediastinal large B-cell lymphoma; Diffuse large B-cell lymphoma, not otherwise specified; Differential diagnosis; Prognosis

Part Ⅱ Single-Cell Transcriptomic Analysis Reveals Intratumoral Heterogeneity and Tumor Microenvironment Features in Primary Mediastinal Large B-Cell Lymphoma

Objective: Among large B-cell lymphomas, unlike the most common diffuse large B-cell lymphoma (DLBCL), primary mediastinal large B-cell lymphoma (PMBL) is an aggressive lymphoma with unique molecular characteristics, and the role of intratumoral heterogeneity, cell origin, and tumor microenvironment (TME) remains unclear. The purpose of this study was to compare the composition differences between PMBL and DLBCL and reactive lymph node (rLN), and analyze the malignant B-cell heterogeneity, tumor microenvironment characteristics and cell-cell interaction pathways of PMBL using single-cell RNA sequencing (scRNA-seq), so as to provide theoretical basis for molecular typing, origin exploration and targeted therapy of PMBL.

Methods: scRNA-seq data from 4 PMBL paraffin-embedded tissues were integrated with public scRNA-seq data sets of 3 DLBCL and 3 rLN from heiDATA database.  Seurat (v5.0) performed quality control, normalization, batch correction, and dimension reduction clustering. CopyKAT distinguished malignant/non-malignant B cells. GSVA, CytoTRACE2, Monocle3, and CellChat analyzed functional heterogeneity, differentiation trajectories, and cell-cell communication.

Results: A total of 50145 high-quality cells were analyzed. PMBL TME exhibited higher myeloid cell infiltration (16.18% vs. DLBCL: 0.49%) and unique fibroblast populations (7.29%), while there was a lack of fibroblast component in DLBCL and rLN. PMBL malignant B cells can be divided into 5 subgroups (B cells-1 to 5), in which B cells-1 is the dominant subgroup, showing active metabolism and significantly activating mTORC1 signaling pathway, and B cells-2 is the tumor progenitor cells with active proliferation and MYC pathway activation. B Cells-3 to 5 were characterized by immune regulation, antigen presentation and initial differentiation, respectively. Fibroblasts in the PMBL TME included matrix, Immune-regulatory (dominant subgroup), tumor-like, and vascular-associated types. Fibroblasts interact with malignant B cells mainly through COL1A1-CD44/LAMA2-CD44, and regulate the function of myeloid cells through COMPLEMENT and MIF pathways, forming a stroma-immune synergistic TME. In PMBL, T cells were classified into CD8+ central memory (dominant subgroup), cytotoxic effector, proliferative and helper subgroups. Myeloid cells were classified into classical dendritic cell type 1, plasmacytoid dendritic cells, activated/tumor-associated dendritic cells, tissue-remodeling macrophage (dominant subgroup), and pro-inflammatory macrophage.

Conclusions: This study delineates the first single-cell atlas of PMBL, revealing the molecular heterogeneity of malignant B cells, fibroblast-driven TME interactions and  novel insights into cellular origins. Interventions targeting the MYC/mTOR pathway, fibroblast subpopulation specific markers, or immune checkpoints may improve treatment strategies for PMBL, especially for relapsed/refractory patients.

Keywords: Primary mediastinal large B-cell lymphoma; Single-cell RNA sequencing; Tumor heterogeneity; Tumor microenvironment

Part Ⅲ Immune Microenvironment-Based Biomarkers for Predicting Efficacy of PD-1 Monoclonal Antibody Containing Regimens in Relapsed/Refractory Large B-Cell Lymphoma

Objective: Patients with relapsed/refractory (R/R) large B-cell lymphoma (LBCL) have poor prognoses, and PD-1 monoclonal antibody (mAb) containing immunotherapy offers a therapeutic option. Based on clinical trial results, PD-1 mAb is currently approved by the U.S. Food and Drug Administration for the treatment of patients with R/R primary mediastinal large B-cell lymphoma (PMBL). In the actual treatment process, PD-1 mAb regimen has also been applied to the posterior line treatment of patients with diffuse large B-cell lymphoma (DLBCL). The immune microenvironment (IME) plays a critical role in tumor immunotherapy, potentially influencing treatment response and survival. This study explores associations between IME components, immune checkpoint markers, immune subtypes, and clinical outcomes in R/R LBCL patients receiving PD-1 mAb.

Methods: This study retrospectively collected clinical and pathological data from 46 R/R DLBCL and 9 R/R PMBL who first visited our hospital from August 2007 to March 2022 and subsequently received PD-1 mAb combined with rituximab as salvage treatment. Multiplex immunofluorescence (Panel 1: PAX5, CD4, CD8, PD-1, PD-L1; Panel 2: CD68, CD163, HLA-DR, SPARC, PD-L1) was performed on 33 R/R DLBCL and 7 R/R PMBL samples. HALO software quantified cell densities and immune checkpoint expression.

Results: In R/R DLBCL, the analysis revealed that the response group (PR/CR) was significantly associated with age ≤60 years (P=0.034), Ki-67 index <70% (P=0.037), BCL-2/C-MYC double expression (P=0.025), and response to first-line treatment (P=0.012). In terms of molecular characteristics, TP53 mutations (61.3%), KMT2D (25.8%), and MYD88 (22.6%) were frequent mutations in R/R DLBCL patients. NOTCH1 and B2-MG mutations were exclusive to non-responders, while TET2 and EP300 mutations occurred only in responders. Multivariate analysis identified poor immunotherapy response, MYC rearrangement, and non-complete remission of first-line therapy were independent risk factors for OS after immunotherapy. IME analysis revealed the density of CD8+T cells (P < 0.001) and CD68+SPARC+ macrophages (P < 0.01) was significantly increased in responders, whereas M2 macrophages were enriched in non-responders (P < 0.05). The expression of PD-L1 on both B cells and macrophages was significantly correlated with the efficacy of immunotherapy. In addition, when TPS≥10.2% (AUC=0.865) or CPS≥5.6 (AUC=0.901), patients with R/R DLBCL were better treated with PD-1 mAb combined with rituximab. IME typing based on the density of PD-L1+ cells and CD8+T cells showed that patients with high PD-L1 / high CD8 (Type I) had the highest response rate, and patients with low PD-L1 / low CD8 (Type II) had the lowest response rate. Patients with Type I (P < 0.0001), high PD-L1 / low CD8 or low PD-L1 / high CD8 (Type III) (P=0.027) had significantly better PFS after treatment than those with Type II. Due to the small sample size, R/R PMBL could not be further analyzed, but some results showed a similar trend to the results of R/R DLBCL analysis, which played a validation role to a certain extent.

Conclusions: The efficacy of PD-1 mAb combined with rituximab in R/R LBCL is closely related to the characteristics of the IME, such as CD8+T cells, CD68+SPARC+ macrophage high density and high expression of PD-L1 suggest therapeutic advantages. IME typing based on PD-L1+ cells and CD8+T cells has shown potential as a predictive tool for clinical decision making. This study provides a basis for the precise stratification of R/R LBCL immunotherapy at the IME level, and facilitates the formulation of personalized treatment strategies. However, it is still necessary to further expand the cohort to verify relevant conclusions.