Molecular classification and dynamic prognostication of Diffuse Large B Cell Lymphoma (DLBCL) by circulating cell-free DNA sequencing

弥漫性大b细胞淋巴瘤占非霍奇金淋巴瘤的大多数病例，通常是一种可治愈的疾病。然而，超过三分之一的患者不能通过标准的免疫化疗治愈，复发后的预后很差。目前采用国际预后指数（IPI）进行风险分层。然而，这种风险分层缺乏歧视性。因此，即使在不同的风险群体中，预后也有很大差异，而且高度依赖于对治疗的反应。因此，目前DLBCL的风险适应治疗非常有限，导致低风险患者的过度治疗和高风险患者的治疗不足。最近，使用下一代测序（NGS）开发了更复杂的依赖于遗传定义集群的风险分类。然而，这些分类所需的方法昂贵，技术复杂，只有部分标准化，并且需要肿瘤活检材料作为先决条件。包括dlbcl患者在内的癌症患者外周血中含有游离肿瘤DNA片段，这种游离DNA可以很容易地通过CAPP-Seq （cancer Personalized Profiling by deep sequencing）提取和分析。我们的目的是通过CAPP-Seq从DLBCL患者的外周血中识别预后的复发性遗传畸变。在此基础上，我们将开发纯粹基于液体活检的DLBCL分子风险分类。接下来，将在一组复发/难治性DLBCL患者中验证确定的高风险特征。我们假设高危特征将在这组患者中得到丰富。此外，我们将在DLBCL的原生小鼠模型中验证检测到的高风险特征。最后，CAPP-Seq将应用于开始全身治疗后的早期随访样本，在分子基础上进行高灵敏度的最小残留疾病评估。结合初始风险模型，这将产生一个动态的、自适应的风险分类。综上所述，我们计划建立一个基于复发性遗传畸变、早期分子反应和临床数据的DLBCL风险模型，对DLBCL患者进行动态风险评估。该模型可能有助于未来开发治疗算法，包括高风险患者的早期适应治疗，例如在常规治疗中添加CAR-T细胞、双特异性抗体或靶向药物等新型治疗方式。我们相信，这将最终改善DLBCL的治疗结果，从而提高患者的生存率。

Diffuse large B-cell Lymphoma accounts for most cases of Non Hodgkin Lymphoma and is generally a curable disease. However, more than one third of patients cannot be cured by standard immunochemotherapy and the prognosis at relapse is dismal. Risk stratification is currently carried out by the international prognostic index (IPI). However, this risk stratification lacks discriminatory power. Therefore, the prognosis differs significantly even within the different risk groups and is furthermore highly dependent on the response to therapy. Consequently, risk-adapted treatment in DLBCL is currently very limited leading to over-treatment of low-risk and under-treatment of high-risk patients.Recently, more sophisticated risk classifications relying on genetically defined clusters were developed using next-generation sequencing (NGS). However, the methods needed for these classifications are costly, technically complicated, only partly standardized and need tumor biopsy material as a prerequisite. Peripheral blood of cancer patients including DLBCL-patients contains fragments of cell-free tumor-DNA This cell-free DNA can easily be extracted and analyzed through Cancer Personalized Profiling by deep sequencing (CAPP-Seq). Our aim is to identify recurrent genetic aberrations by CAPP-Seq from peripheral blood of DLBCL patients that are prognostic. Building on that, we will develop a molecular risk classification of DLBCL based purely on a liquid biopsy. Next, an identified high-risk signature will be validated within a set of relapsed / refractory DLBCL patients. We hypothesize that the high-risk signature will be enriched within this patient set. Additionally, we will validate the detected high-risk signature within autochthonous mouse models of DLBCL. Lastly, CAPP-Seq will be applied to early follow-up samples after initiation of systemic therapy to perform minimal residual disease assessment on a molecular basis with high sensitivity. Combined with the initial risk model, this will result in a dynamic, adaptive risk classification.In summary, we plan to establish a DLBCL risk model allowing for dynamic risk assessment in DLBCL patients based on recurrent genetic aberrations, early molecular response and clinical data. This model will likely aid future efforts to develop treatment algorithms that involve early adaption of treatment in high-risk patients, for example by adding novel treatment modalities such as CAR-T cells, bispecific antibodies or targeted agents to conventional treatment. We are convinced that this will ultimately improve treatment outcome in DLBCL and consequently patient survival.