【目的】 套细胞淋巴瘤（MCL）是侵袭性非霍奇金淋巴瘤（NHL）的一种罕见亚型，决定其临床预后的遗传学因素尚不明确。目前尚没有MCL的基因组学和转录组学的整合研究，缺乏对疾病的整体认识。且MCL在临床及遗传学存在高度异质性，患者对治疗的反应及预后也相差很大，基于遗传学特点对其区分亚型有助于临床危险度分层并指导精准治疗。

【方法】 该研究对134例MCL患者的152个DNA样本进行了全外显子测序（WES），其中包括123例初治和11例复发的患者，以及16例有初诊和复发续贯样本的患者。其中95例患者接受了标准的基于大剂量阿糖胞苷的免疫化疗，48个样本具有匹配的RNA测序数据。研究者们使用GATK，MutSig2CV和GISTIC2.0识别体细胞突变和拷贝数变化，应用MutationalPatterns定义突变特征，并利用ABSOLUTE和PhylogicNDT确定患者的克隆进化的模式；并应用NNMF非负矩阵分解聚类法定义MCL的不同分子亚组，进一步探索不同亚组患者的基因表达谱和临床预后。

【结果】 每个样本体细胞突变的中位数为29（范围8-72）。该研究共鉴定出34个重现性突变基因，其中包含报道过的驱动突变（TP53，ATM，CCND1，KMT2D，NSD2，SMARCA4，ARID1A，NOTCH1，NOTCH2，BIRC3，TRAF2，UBR5）和新发现的突变（SP140，SVEP1，LRP1B，LRP2，PCDH10）。并检测到7个区域的染色体拷贝数增加和13个区域染色体缺失（频率>10％，q <0.1）。基于无进展生存期（PFS）和总生存期（OS）的多因素COX模型分析显示，IGHV突变状态，MIPI风险分层，TP53突变/del(17p13)，SP140突变/del(2q36)以及NOTCH1、SMARCA4、PCDH10突变和del(9p)，均具有显著的独立预后意义。

研究进一步探究了MCL遗传事件的克隆状态和克隆演变模式。Del(11q22)和del(9p)倾向于表现为主克隆，可能是疾病发生的早期事件，而NSD2、LRP1B和CTNNA2中的突变更可能是亚克隆（q <0.05）。通过对复发前后遗传学异常的动态变化将 MCL患者分为三类：69％（11/16）的患者具有显著的克隆演变（CCF变化>0.5）；25%患者具有轻度的克隆演变（0.2≤CCF变化≤0.5）；6%患者无克隆演变（CCF变化<0.2）。具有显著克隆演变的患者与轻度或无克隆演变患者相比，生存时间明显减少（总生存时间为47.5个月 vs. 未达到，P=0.041；复发后生存时间为17.1个月 vs. 未达到，P=0.023）。

通过对上述重现性体细胞突变和拷贝数变化进行整合分析，将MCL分为四个分子亚型（Cluster，C），每个亚型具有不同的基因表达谱和临床特征。C1和C2-4分别与MCL的惰性和侵袭性临床特征相关；且两种类型MCL的细胞起源不同，C1和C2-4分别具有丰富的记忆B细胞和CCR6阴性生发中心明区B细胞的基因表达特征。其中C1亚型临床上大多表现为白血病样非结性惰性MCL，具有IGHV突变、SOX11阴性、CCND1突变、amp(11q13)的特征，转录组数据显示此群患者BCR信号通路明显激活。C2亚型以del(11q22）、del(1p21)、ATM突变为主要特征，并且伴有NF-κB和DNA修复信号通路的上调。C3亚型多表现为SP140、NOTCH1和NSD2突变，并伴有明显的NF-κB和BCR信号通路下调。C4亚型的母细胞型或多形性MCL发生率最高（23.7％，P=0.016），常伴有del(17p)、del(13q)、del(9p)以及TP53和TRAF2突变；该亚型还表现为的MYC信号通路的激活及增殖相关基因的过表达。重要的是，这四个分子亚型的患者表现出明显的预后差异，中位PFS在C1亚型中尚未达到，C2亚型为41.2个月，C3亚型为30.7个月，C4亚型为16.1个月（P<0.001）。

【结论】该研究首次对MCL的基因组学和转录组学进行整合分析，揭示了MCL遗传特征图谱，并对MCL患者根据遗传学特征进行分子分型，不同亚型患者表现出特征性基因表达谱和临床预后。该研究揭示了与临床预后相关的遗传学特征，并将指导后续MCL治疗策略的开发。

Introduction Mantle cell lymphoma (MCL) is a phenotypically and genetically heterogeneous malignancy in which the genetic alterations determining clinical behavior are not fully understood. The genetic heterogeneity in MCL motivates us to define different genetic subsets, to delineates clonal evolution patterns and their impacts on clinical outcomes. 

Methods We performed whole-exome sequencing (WES) on 152 DNA samples derived from 134 MCL patients. This cohort includes 123 untreated and 11 relapsed patients, in which 18 with longitudinal samples. 95 patients received high dose cytarabine-based immunochemotherapy from BDH-MCL01 clinical trial (NCT02858804). 89 samples with matched normal, 115 samples derived from purified cryopreserved peripheral blood or bone marrow cells and 42 samples from FFPE sections. Within all these samples, 48 with matched RNA sequencing (RNA-seq) data for which 42 from untreated and 6 from relapsed patients. We used GATK, MutSig2CV, and GISTIC2.0 to identify driver genetic lesions, applied MutationalPatterns to define mutational signatures, and utilized ABSOLUTE and PhylogicNDT to determine pattern of clonal evolution. We applied Non-negative matrix factorization consensus clustering to identify molecular subgroups and evaluated association between genomic alterations with clinical outcomes (PFS and OS).

Results The median burden of non-silent mutations was 29 per sample (range 8-72). 34 recurrently mutated genes were identified, containing previously reported driver mutations (ATM, TP53, NSD2, KMT2D, CCND1, SMARCA4, ARID1A, NOTCH2, NOTCH1, BIRC3, UBR5, TRAF2) and novel mutations (SP140, PCLO, SVEP1, LRP1B, PCDH10, LRP2, CTNNA2, VCAN, KMT2C, TACC2). 7 copy number gain and 13 copy number loss regions were detected as recurrent somatic copy number alterations (SCNAs, frequency>10%, q<0.1). A multivariate Cox model of PFS and OS revealed that TP53 mutation/del(17p13), SP140 mutation/del(2q36), mutations in NOTCH1, PCDH10, and del(9p) remained prognostic value, independent of MIPI risk, IGHV mutation and other genetic alterations.

We defined the clonal status of genetic alterations and clonal evolution pattern. Del (11q22) and del (9p) tend to be clonal while mutations in NSD2, LRP1B, CTNNA2 were more likely to be subclonal events (q<0.05). Clonality analysis enabled inference of temporal relationships between pairs of alterations. Then we further determined clonal evolution pattern by measuring the dynamic changes of fractions of cancer cells harboring each genetic alteration. 11 of 16 longitudinal samples had extreme clonal evolution (69%, CCF change > 0.5), 4 with modest evolution (0.2 ≤ CCF change ≤ 0.5), and 1 without evolution (CCF change < 0.2). Patients with extreme evolution had significantly inferior survival than with those with modest or no evolution (median survival from first sampling was 47.5 months vs not reached, p=0.041; from second sampling was 17.1 months vs not reached, p=0.023).

We classified MCL into four subsets based on genetic alterations, each with distinct gene expression profiles and clinical behavior. Cluster 1 (C1) was associated with leukemic non-nodal MCL while Clusters 2, 3, and 4 (C-4) were associated with classical MCL. Consistent with different cellular origins for different types of MCL, C1 was enriched for gene expression signatures of memory B cells and C2, C3, and C4 appeared to have a signature of CCR6 negative light zone B cell. 1) C1 (16.4%) was featured with mutated IGHV, negative expression of SOX11, CCND1 mutation, amp (11q13) as well as with active BCR signaling gene expression and an indolent clinical course. 2) C2 (23.1%) was enriched with del (11q22), del (1p21) and ATM mutation and harbored no TP53 mutation and del(17p). Patients in C2 were upregulated for gene expression in NF-κB and DNA repair pathways. 3) C3 (32.1%) was characterized with mutations in SP140, NOTCH1 and NSD2, del (6p) and amp (13q), and downregulation of gene expression in NF-κB, BCR signaling, MYC and inflammatory pathways. 4) C4 (28.4%) harbored del (17p), del (13q), del (9p) as well as mutations in TP53 and TRAF2. Interestingly, C4 was associated with a higher incidence of Blastoid or pleomorphic MCL (23.7%, p=0.016) and gene expression signatures of hyperproliferation and MYC pathway activation. Patients with the C1 subtype had a more favorable outcome than those with C2, C3, and C4 subtypes; the median PFS for the four subtypes were not reach, 41.2, 30.7, and 16.1 respectively. Differences in survival of patients whose diseases reflected each of the 4 subtypes also remained significant among patients who received the REDOCH/RDHAP regimen.

Conclusions In summary, the integrative genomics and transcriptomics analysis described herein has greatly expanded our knowledge regarding the strong associations among genetic clusters, oncogenic pathways, and clinical outcomes. This study provides a framework to assess unappreciated genetic heterogeneity in the clinically defined subtypes of MCLs and forms the basis for designing precision therapies for aggressive MCL with genetic factors and oncogenic pathways as testable targets. Our dataset serves as a valuable genetic and transcriptomic resource for this rare and aggressive type of non-Hodgkin lymphoma. The outcome-associated genetic subsets will guide the choice of therapies in patients with the greatest need.