Prediction Of Therapeutic Outcomes In Dlbcl From Circulating Tumor Dna Dynamics.

JOURNAL OF CLINICAL ONCOLOGY(2016)

引用 4|浏览23
暂无评分
摘要
7511 Background: Responses to initial therapy in diffuse large B-cell lymphoma (DLBCL) are heterogeneous, where patients who fail to respond have poor outcomes. Early identification of failures remains a challenge. We report an empiric method to response prediction utilizing circulating tumor DNA (ctDNA) from as few as one cycle of therapy. Methods: We utilized CAPP-Seq, a next-generation sequencing platform for ctDNA detection (Newman AM et al, Nat Med 2014), to profile 33 patients with DLBCL from Stanford University during induction therapy. Tumor and germline DNA was sequenced to define tumor-specific somatic alterations, which were then followed in plasma DNA. Changes in ctDNA were compared with response rate (PET/CT, Deauville score < 4 defined as CR) and outcomes using a training/validation design. Results: To understand early dynamics of ctDNA, we serially profiled plasma from a training cohort of DLBCL patients throughout their first two cycles of immunochemotherapy (n = 12 patients, 114 samples; median 9/patient). In patients achieving a radiographic response, ctDNA levels declined within days of therapy. At onset of cycle 2 (day 22), patients achieving a CR had a greater reduction in ctDNA than those who did not (3.35-log vs 0.98-log reduction, p = 0.0014). Furthermore, all patients with a > 2-log reduction in ctDNA by cycle 2 achieved a CR (5/5), while those failing to reach this threshold did not (7/7; p = 0.0013). We therefore established a 2-log drop in ctDNA as a threshold to predict response. We tested the ability of this threshold to predict outcomes in an expanded cohort of 33 patients. This threshold significantly predicted CR rate (75% vs 22%, p = 0.01), PFS (p = 0.002) and OS (p = 0.001). Change in ctDNA remained significantly associated with improved PFS in multivariate analysis (p = 0.003) when controlling for IPI, cell of origin, and prior treatment. Finally, in a cohort of patients experiencing relapse, ctDNA heralded radiographic progression by a median of 148 days (n = 8). Conclusions: Circulating tumor DNA in DLBCL changes rapidly during treatment, with > 2-log decline in the first two cycles associated with response and survival. Studies utilizing the dynamics of ctDNA to risk stratify patients and guide therapy are warranted.
更多
查看译文
关键词
tumor dna dynamics,dlbcl
AI 理解论文
溯源树
样例
生成溯源树,研究论文发展脉络
Chat Paper
正在生成论文摘要