Multiparameter flow cytometry (MFC)-based MRD testing is increasingly utilized in AML, but methodologies and assay characteristics vary substantially [1,2,3]. Nonetheless, following regulatory authorities’ guidance MRD assays be validated 1-log below the chosen threshold for clinical decision-making, the European LeukemiaNet (ELN) MRD Working Party presently defines MRD positivity uniformly at 0.1% [2]. Yet, agnostic to assay specifications, risk stratification may be suboptimal. We examined this possibility using 4 well-annotated cohorts of AML patients in morphologic remission in which MRD results did not guide treatment decisions.

Our dataset included patients from GIMEMA (n = 182, from EORTC/GIMEMA AML-10 and AML-12), HOVON/SAKK (n = 171, from HOVON/SAKK 42A), MRC/NCRI (n = 846, from AML17), and Seattle (n = 861; adults allografted in first morphologic remission). All patients were treated on IRB-approved research protocols or standard treatment protocols and gave consent in accordance with the Declaration of Helsinki; for details, see online supplement. For GIMEMA, cases expressing a leukemia-associated immunophenotype (LAIP) were re-analyzed with a combination of 6–8 markers to define a LAIP fingerprint for MRD tracking [4,5,6,7]. For HOVON/SAKK, a suitable LAIP comprising ≥10% of the CD45^dim blasts was identified at diagnosis using a 4-color assay and monitored during follow-up. For MRC/NCRI, MFC-MRD testing was centrally performed with a standardized 7/8-color assay that screened for “different-from-normal” LAIPs pretreatment and tracked these (∼0.02–0.05% sensitivity) but also applied the different-from-normal approach in follow-up samples to detect LAIP changes (∼0.05–0.1% sensitivity). Only samples with pretreatment LAIPs could be reported as MRD^neg. For Seattle, pre-transplant bone marrows were subjected to 10-color MFC-MRD testing with a panel of 3 antibody combinations, collecting up to 1 million events/tube [8, 9]. MRD was identified using a “different-from-normal” approach, with detection sensitivity to a level of 0.1% in most cases and at lower levels in progressively smaller patients subsets [8, 9]. For all cohorts, only samples of sufficient quality for MRD testing were included in our analyses, which were conducted under a research protocol approved by the Fred Hutchinson Cancer Center’s Institutional Review Board.

基于多参数流式细胞术 (MFC) 的 MRD 检测越来越多地用于 AML，但方法和检测特征差异很大 [1,2,3]。尽管如此，根据监管机构的指导，MRD 检测应验证为低于临床决策所选阈值 1 个对数，欧洲白血病网 (ELN) MRD 工作组目前将 MRD 阳性率统一定义为 0.1% [2]。然而，由于与检测规范无关，风险分层可能不是最理想的。我们使用 4 个形态学缓解的 AML 患者队列研究了这种可能性，其中 MRD 结果不能指导治疗决策。

我们的数据集包括来自 GIMEMA（ n = 182，来自 EORTC/GIMEMA AML-10 和 AML-12）、HOVON/SAKK（ n = 171，来自 HOVON/SAKK 42A）、MRC/NCRI（ n = 846，来自 AML17）的患者和西雅图（ n = 861；首次形态学缓解时同种异体移植的成人）。所有患者均按照IRB批准的研究方案或标准治疗方案进行治疗，并根据赫尔辛基宣言给予同意；详情请参见在线补充。对于 GIMEMA，使用 6-8 个标记的组合重新分析表达白血病相关免疫表型 (LAIP) 的病例，以定义用于 MRD 跟踪的 LAIP 指纹 [4,5,6,7]。对于 HOVON/SAKK，在诊断时使用 4 色测定鉴定出包含 ≥10% CD45^暗母细胞的合适 LAIP，并在随访期间进行监测。对于 MRC/NCRI，MFC-MRD 测试集中使用标准化 7/8 色测定进行，该测定筛选“不同于正常”的 LAIP 预处理并跟踪这些（ ~ 0.02-0.05% 灵敏度），但也应用了不同的在后续样本中采用非正常方法来检测 LAIP 变化（ ~ 0.05–0.1% 灵敏度）。只有经过预处理 LAIP 的样品才能报告为 MRD ^neg 。在西雅图，使用一组 3 种抗体组合对移植前骨髓进行 10 色 MFC-MRD 测试，每管收集多达 100 万个事件 [8, 9]。 MRD 使用“不同于正常”的方法进行识别，在大多数情况下检测灵敏度为 0.1%，在逐渐缩小的患者亚群中检测灵敏度较低 [8, 9]。 对于所有队列，我们​​的分析中仅包含足以进行 MRD 检测的样本，分析是根据 Fred Hutchinson 癌症中心机构审查委员会批准的研究方案进行的。