Development of genetic prognostication models in newly diagnosed chronic lymphocytic leukaemia (CLL) patients

慢性淋巴细胞白血病(CLL)是欧洲血统的人中最常见的白血病，仅在英国每天就有10多例新病例。慢性淋巴细胞性白血病具有高度多样化的临床过程，大多数患者被诊断为早期无症状疾病，最初不需要治疗。一些患者与无症状的疾病生活了几年，而另一些患者病情进展很快，需要治疗。CLL是一种固有的不治之症，是死亡和发病率的重要原因，包括复发感染的高风险。CLL治疗已经被高效和耐受性良好的B细胞受体信号通路抑制剂(BCRi)所改变，这些药物可以改善晚期疾病患者的预后。鉴于在治疗有症状的慢性淋巴细胞白血病方面取得了成功，最近的重点转移到了研究预防性治疗是否可以改善早期无症状疾病但有高进展风险的患者的预后。德国CLL12试验的初步结果表明，早期使用BCRi ibrutinib治疗的高危CLL患者的预后有显著改善，早期治疗的高危患者的死亡时间延长4-5倍。尽管取得了这一成功，但目前的预测模型并不充分，只能识别少数高危患者。该项目的目的是为早期CLL开发准确的预测模型，以便识别可能从早期治疗中受益的高危患者。为此，我们最近发表了一项利用早期CLL病例的全基因组关联研究，并确定了两种与高危CLL显著相关的常见生殖系遗传变异(Lin等人，2021，《自然通讯》，12：665)。DOI：10.1038/s41467020-20822-9)。这些变异具有与已建立的临床标记相当的预后价值，并提供了概念证明，即纳入胚系遗传标记可以显著改善大多数CLL患者的预后模型。我们最近将我们的CLL队列扩大到大约2000名患者，中位随访时间为15年，其中包括感染和出血事件以及死亡的数据。除了识别预测高危疾病的新的种系遗传变异外，该项目还将使用机器学习来从我们小组已经生成的高密度阵列数据中识别白血病特有的插入和缺失(体细胞改变)。这个项目还将提供一个使用基于细胞的模型从功能上询问新的体细胞变化的机会。关键是，这将是第一次将患者生殖系和白血病体细胞遗传数据与已建立的临床标记结合在一起，用于早期CLL患者的准确预测。预计将所产生的模型应用于临床实践将改善数以万计的慢性淋巴细胞性白血病患者的预后，并可能导致采用类似的方法来改善其他癌症的预后。

Chronic lymphocytic leukaemia (CLL) is the most common leukaemia in people of European ancestry with more than 10 new cases per day in the UK alone. CLL has a highly heterogeneous clinical course and most patients are diagnosed with early stage asymptomatic disease that does not initially require treatment. Some patients live with asymptomatic disease for several years while others progress quickly requiring treatment. CLL is inherently incurable and a significant cause of mortality and morbidity, including high risk of recurrent infections.CLL therapy has been transformed by highly effective and well tolerated B-cell receptor signalling pathway inhibitors (BCRi) which improve patient outcomes for those with advanced disease. Given the success in treating symptomatic CLL, focus has recently shifted towards addressing whether pre-emptive treatment can improve outcomes for patients with early-stage asymptomatic disease but at high-risk of progressing. Preliminary results from the German CLL12 trial report remarkable improvements in outcomes for high-risk CLL patients treated early with the BCRi ibrutinib, where time to death was 4-5 times longer in high-risk patients treated early. Despite this success, current prognostication models are inadequate and identify only a minority of high-risk patients.The aim of this project is to develop accurate prognostication models for early-stage CLL in order to identify high-risk patients who might benefit from earlier treatment. To this end, we recently published a genome-wide association study utilising early-stage CLL cases and identified two common germline genetic variants that significantly associate with high-risk CLL (Lin et al, 2021, Nature Communications, 12: 665. doi: 10.1038/s41467-020-20822-9). These variants have prognostic value equivalent to established clinical markers and provide proof of concept that the incorporation of germline genetic markers can significantly improve prognostication models for the majority of CLL patients.We have recently expanded our CLL cohort to approximately 2000 patients with a median follow-up of 15 years that includes data on infections and bleeding events as well as death. In addition to identifying new germline genetic variants predicting high-risk disease, this project will use machine learning to identify leukaemia-specific insertions and deletions (somatic alterations) from high-density array data already generated by our group. This project will also provide an opportunity to functionally interrogate novel somatic alternations using cell-based models.Critically, this will be the first study to incorporate patient germline and leukaemia somatic genetic data along with established clinical markers for accurate prognostication in early-stage CLL patients. Implementation of the resulting model into clinical practise is predicted to improve outcomes for the tens of thousands of patients living with CLL and could also lead to similar approaches being adopted to improve prognostication in other cancers.