To the Editor:

Somatic mutations have been reported in 80%-90% of myelodysplastic syndrome (MDS) using next-generation sequencing (NGS) gene panels interrogating at least 30 MDS-associated genes.^{1, 2} Next-generation sequencing studies are increasingly being used in the diagnosis of MDS, to confirm or exclude a clonal process. However, about 10%-20% MDS cases are reported to have no detectable somatic mutations on typical NGS sequencing panels.^{1, 2} Given the subjectivity in assessing dysplasia and myriad non-MDS causes of cytopenia and dysplasia, there is skepticism if cases that lack evidence of clonality by both cytogenetics and NGS testing represent true MDS. Indeed, NGS study on peripheral blood may be used to screen cytopenic patients, with the assumption that a negative result makes MDS highly unlikely.³ Most large MDS sequencing studies involve multi-institutional databases in which clinical details to assess the veracity of the diagnosis are limited and thus the clinicopathologic and genetic features and natural history of these mutation-negative MDS cases are largely unknown.

We reviewed the pathologic and clinical features of MDS cases (excluding therapy-related disease) from the pathology archives at three large academic medical centers between 2014 and 2020, during which targeted NGS panels (including 48 genes tested in all patients) were performed as a part of the clinical work-up (Table S1). The limit of detection was generally 1% for point mutations and 2%-5% for indels. Only pathogenic or likely pathogenic mutations were included in the analysis. For mutation-negative patients, a diagnosis of MDS required the presence of persistent or progressive cytopenia(s) as well as significant bone marrow (BM) dysplasia. Cases initially called MDS but later proven to be secondary cytopenias due to other processes, or in which cytopenias spontaneously resolved, were excluded. This study was approved by the Institutional Review Boards of all participating institutions. Statistical analysis was performed using GraphPad Prism (GraphPad software, San Diego, CA) and Xlstat, with significance set at a p value < 0.05 (two-sided). In patients with NGS performed after the time of initial diagnosis, overall survival (OS) was measured from the date of NGS; five patients had previously received either hypomethylating agents or lenalidomide, but all showed persistent MDS of the same disease subtype.

The cohort consisted of 317 MDS patients, including 61 with no detectable mutations (MDS-NDM). The incidence of mutation-negative MDS was 10.3% based on the consecutively retrieved MDS cases (2016–2018). The mutation heat map (Figure S1) illustrates the mutation distributions and their associations with karyotype and IPSS-R group for the 256 cases with mutations. Details of the 61 MDS-NDM cases are provided in Table S2. Twenty-six (43%) had an abnormal karyotype, including six cases with isolated del(5q) and six with complex karyotype (≥ 4 independent abnormalities). Twenty-three (38%) had increased BM blasts. Of the 38 patients with no increased BM blasts, 18 had a normal karyotype or -Y. In the latter cases, the diagnosis of MDS was based solely on the presence of significant dysplasia in the context of persistent or progressive unexplained cytopenia. Immunophenotypic aberrancies were detected by flow cytometry in 9/10 MDS-NDM cases lacking cytogenetic abnormalities or increased blasts. Follow up sequencing (prior to any SCT) was performed on 26 MDS-NDM patients, of which 18 continued to show no mutations on one or more subsequent NGS tests. The newly detected mutations found in eight patients were mostly present at a low VAF (< 5%), often at or near the limit of detection of the assay.

The 61 MDS-NDM patients are compared to the 256 MDS patients with one or more mutation (MDS-mut) in Table 1. The MDS-NDM patients were younger (p < 0.001) and showed different distribution of WHO subtypes, with more frequent MDS-SLD (p = 0.04) and MDS-del5q (p = 0.02) and less frequent MDS-RS (p < 0.001). The incidence of abnormal karyotypes was similar to MDS-mut. When limiting the analysis to the 176 cases with normal/-Y karyotypes, similar differences in patient age (p = 0.009) and WHO subtypes (p = 0.004) were observed between MDS-NDM and MDS-Mut (Table 1). A higher proportion of MDS-NDM patients received SCT (41% vs. 22%, p = 0.005), likely reflecting their younger age compared to MDS-mut patients. During the follow up period (median 26 months), 11% of MDS-NDM (including 5/35, 14% of the patients with normal/-Y karyotypes) and 16% of MDS-Mut patients progressed to AML (p = 0.43). The MDS-NDM patients had significantly longer OS (p = 0.008) and AML-free survival (p = 0.015) compared to MDS-Mut patients (Figure S2A,B), however there was no difference in OS (p = 0.18) or AML-free survival (p = 0.38) between MDS-NDM and MDS-Mut when limiting the analysis to patients with normal/-Y karyotypes (Figure S2C,D). On multivariable analysis, presence of an SF3B1 mutation in the absence of adverse mutations (HR 0.424, p = 0.008) and SCT (HR 0.294, p < 0.001) were favorable factors, whereas, IPSS-R, high/very high (HR 3.585, p < 0.001) or intermediate (HR 1.853, p = 0.004)(compared to IPSS-R low/very low) were unfavorable factors for OS. MDS-NDM (HR 0.592, p = 0.059) showed a trend towards favorable OS, while the presence of adverse mutations (TP53, EZH2, NPM1, or RUNX1)⁴ showed a trend towards unfavorable OS (HR 1.318, p = 0.131). However, when limiting the analysis to patients with normal/-Y karyotypes, MDS-NDM did not show a significant effect on OS independent from IPSS-R, SCT, and SF3B1 or adverse mutations (data not shown).

In this study, we examined a cohort of 61 patients diagnosed with de novo MDS based on clinicopathologic data but lacking any pathogenic mutations on standard clinical NGS panels, including 35 with normal/-Y karyotype, of which 18 lacked increased blasts. Putative MDS cases that lack MDS-defining cytogenetics, somatic mutations, or increased blasts are the most challenging to diagnose, since neither cytopenias nor morphologic dysplasia is specific for MDS. We excluded some patients who were initially diagnosed with MDS, but in whom cytopenia(s) and dysplasia resolved at follow-up, underscoring the challenges in initial diagnosis of MDS without genetic evidence of clonality or increased blasts. Although the NGS panels used were large and comprehensive, we acknowledge that the absence of mutations in this study was based on clinical NGS platforms, not whole genome sequencing. Furthermore, mutation detection in clinical NGS platforms might be limited by the regions covered and variable depth of sequencing, the sensitivity of the assays, as well as the individual genes tested. We did identify four cases in which low-level mutated sequences were detected in the initial NGS, but were not callable, reflecting either technical limitations or a very small clone below the limit of reliable detection on clinical NGS panels. Of note, most of the MDS-NDM patients with followup sequencing data continued to show no detectable mutations, in line with the initial negative NGS results.

Compared to MDS with at least one mutation (MDS-mut), MDS-NDM patients were younger, less anemic, had lower BM cellularity and included fewer cases with RS. Some of these differences were likely attributed to certain specific mutations, such as SF3B1 mutation associated with RS and anemia. Although the proportion of cases with an abnormal karyotype was not different from MDS-mut, MDS with isolated del(5q) was more common in MDS-NDM than in MDS-mut (8% vs. 2%). Del(5q), either isolated or in a non-complex karyotype, is known to harbor no or fewer mutations than other subtypes of MDS.⁵ It is possible that leukemogenesis in some MDS-NDM cases might differ from MDS-mut, especially from those carrying well-defined functional driver mutations. In a previous study of MDS cases in which BM CD34+ cells showed preserved B-cell differentiation,⁶ there was a relatively high proportion of unmutated cases, suggesting that the MDS clone might start in later myeloid-committed cells in MDS-NDM. We found a similar IPSS-R distribution between MDS-NDM and MDS-mut groups and on multivariable analysis, lack of detectable mutations showed a trend towards favorable OS, although this was not seen when limiting the analysis to the subset of patients with normal/-Y karyotype. Of note, an increasing mutation burden has been shown to associate with worsening prognosis in MDS.^{1, 4}

In summary, we show that using clinical NGS platforms covering a large panel of genes commonly mutated in hematopoietic neoplasms, approximately 10% of bona fide de novo MDS cases lack detectable pathogenic mutations. About one-third of these cases lacked cytogenetic abnormalities or increased blasts, and in these cases the diagnoses relied solely on persistent cytopenia(s) and dysplasia. This subset of cases may contribute to the approximately 5% false-negative incidence in peripheral blood NGS screening of cytopenic patients for MDS.³ Flow cytometry can provide additional immunophenotypical support to confirm an MDS diagnosis when genetic proof of clonality is lacking. MDS-NDM has more favorable OS and AML-free survival than MDS-Mut, however, a significant subset of patients do progress to AML or die of BM failure. This similarity to MDS-Mut in clinical presentation and natural history raises the possibility that MDS-NDM cases may harbor mutations in genes not covered in standard sequencing panels, or not detected in clinical NGS panels due to technical limitations. Alternatively, other mechanisms driving altered DNA methylation and/or transcriptional reprogramming warrant further investigation. Our data indicate that a lack of pathogenic mutations detected by standard clinical NGS panels in a patient with persistent cytopenia should not exclude a diagnosis of MDS if there is sufficient morphologic dysplasia and the absence of competing etiologies for persistent cytopenia.

致编辑：

据报道，80%-90% 的骨髓增生异常综合征 (MDS) 中存在体细胞突变，使用下一代测序 (NGS) 基因组询问至少 30 个 MDS 相关基因。^{1, 2}新一代测序研究越来越多地用于 MDS 的诊断，以确认或排除克隆过程。然而，据报道，大约 10%-20% 的 MDS 病例在典型的 NGS 测序面板上没有可检测到的体细胞突变。^1、2考虑到评估异常增生和导致血细胞减少和异常增生的无数非 MDS 原因的主观性，如果细胞遗传学和 NGS 检测均缺乏克隆性证据的病例代表真正的 MDS，则存在怀疑。事实上，外周血的 NGS 研究可用于筛查血细胞减少症患者，假设阴性结果使得 MDS 极不可能发生。³大多数大型 MDS 测序研究涉及多机构数据库，其中评估诊断准确性的临床细节有限，因此这些突变阴性 MDS 病例的临床病理学和遗传特征以及自然史在很大程度上是未知的。

我们回顾了 2014 年至 2020 年三个大型学术医疗中心的病理档案中 MDS 病例（不包括治疗相关疾病）的病理和临床特征，在此期间进行了靶向 NGS 面板（包括在所有患者中测试的 48 个基因）作为临床检查的一部分（表 S1）。点突变的检测限通常为 1%，插入缺失的检测限为 2%-5%。分析中仅包括致病性或可能的致病性突变。对于突变阴性患者，MDS 的诊断需要存在持续性或进行性血细胞减少以及显着的骨髓 (BM) 发育不良。最初称为 MDS 但后来证明是由于其他过程引起的继发性血细胞减少症或血细胞减少症自发解决的病例被排除在外。这项研究得到了所有参与机构的机构审查委员会的批准。使用 GraphPad Prism（GraphPad 软件，圣地亚哥，加利福尼亚州）和 Xlstat 进行统计分析，显着性设置为p值 < 0.05（两侧）。在初始诊断后进行 NGS 的患者中，从 NGS 之日起测量总生存期 (OS)；5 名患者之前曾接受过低甲基化药物或来那度胺治疗，但均表现出相同疾病亚型的持续性 MDS。

该队列由 317 名 MDS 患者组成，其中 61 名没有可检测的突变 (MDS-NDM)。根据连续检索到的 MDS 病例（2016-2018），突变阴性 MDS 的发生率为 10.3%。突变热图（图 S1）说明了 256 例突变病例的突变分布及其与核型和 IPSS-R 组的关联。表 S2 提供了 61 个 MDS-NDM 案例的详细信息。26 例 (43%) 有异常核型，其中 6 例为孤立性 del(5q)，6 例为复杂核型（≥ 4 个独立异常）。二十三个 (38%) 增加了 BM 爆炸。在 38 名未增加 BM 原始细胞的患者中，18 名具有正常核型或 -Y。在后一种情况下，MDS 的诊断仅基于在持续或进行性不明原因的血细胞减少的情况下存在显着的发育异常。在缺乏细胞遗传学异常或原始细胞增加的 9/10 MDS-NDM 病例中，通过流式细胞术检测到免疫表型异常。对 26 名 MDS-NDM 患者进行了后续测序（在任何 SCT 之前），其中 18 名在一项或多项后续 NGS 测试中继续显示没有突变。在 8 名患者中发现的新检测到的突变大多以低 VAF (< 5%) 存在，通常处于或接近检测的检测限。其中 18 个在随后的一项或多项 NGS 测试中继续显示没有突变。在 8 名患者中发现的新检测到的突变大多以低 VAF (< 5%) 存在，通常处于或接近检测的检测限。其中 18 个在随后的一项或多项 NGS 测试中继续显示没有突变。在 8 名患者中发现的新检测到的突变大多以低 VAF (< 5%) 存在，通常处于或接近检测的检测限。

61 MDS-NDM患者进行比较，以在表1中的256名MDS患者具有一个或多个突变（MDS-MUT）的MDS-NDM患者更年轻（p  <0.001）并且显示WHO亚型的不同分布，具有更频繁的MDS-SLD ( p  = 0.04) 和 MDS-del5q ( p  = 0.02) 以及频率较低的 MDS-RS ( p  < 0.001)。异常核型的发生率与MDS-mut相似。当将分析限制在 176 例具有正常/-Y 核型的病例时， 在 MDS-NDM 和 MDS-Mut 之间观察到类似的患者年龄 ( p  = 0.009) 和 WHO 亚型 ( p = 0.004) 差异（表 1）。接受 SCT 的 MDS-NDM 患者比例更高（41% 对 22%，p = 0.005)，可能反映了与 MDS-mut 患者相比他们的年龄较小。在随访期间（中位 26 个月），11% 的 MDS-NDM（包括 5/35，14% 的正常/-Y 核型患者）和 16% 的 MDS-Mut 患者进展为 AML（p  = 0.43 ）。 与 MDS-Mut 患者相比，MDS-NDM 患者的 OS（p  = 0.008）和无 AML 生存期（p = 0.015）显着延长（图 S2A、B），但 OS（p  = 0.18）或 当将分析限制在具有正常/-Y 核型的患者时，MDS-NDM 和 MDS-Mut 之间的无 AML 存活率（p = 0.38）（图 S2C、D）。在多变量分析中，SF3B1 的存在不存在不利突变（HR 0.424，p  = 0.008）和SCT（HR 0.294，p  < 0.001）的突变是有利因素，而IPSS-R，高/非常高（HR 3.585，p  < 0.001）或中等（ HR 1.853, p  = 0.004)（与 IPSS-R 低/非常低相比）是 OS 的不利因素。MDS-NDM (HR 0.592, p  = 0.059) 显示出有利 OS 的趋势，而不利突变（TP53、EZH2、NPM1或RUNX1）的存在⁴显示出不利 OS 的趋势（HR 1.318，p = 0.131)。然而，当将分析限制为具有正常/-Y 核型的患者时，MDS-NDM 没有显示出对 OS 的显着影响，独立于 IPSS-R、SCT 和SF3B1或不利突变（数据未显示）。

在这项研究中，我们检查了 61 名根据临床病理数据诊断为新发 MDS 但在标准临床 NGS 面板上缺乏任何致病突变的患者队列，其中 35 名具有正常/-Y 核型，其中 18 名缺乏原始细胞增多。缺乏 MDS 定义的细胞遗传学、体细胞突变或原始细胞增加的推定 MDS 病例诊断最具挑战性，因为血细胞减少症和形态学发育不良都不是 MDS 的特异性病例。我们排除了一些最初被诊断为 MDS 的患者，但这些患者的血细胞减少症和发育不良在随访中得到解决，这强调了在没有克隆性或原始细胞增加的遗传证据的 MDS 初始诊断中的挑战。尽管使用的 NGS 面板庞大而全面，但我们承认本研究中没有突变是基于临床 NGS 平台，不是全基因组测序。此外，临床 NGS 平台中的突变检测可能受到覆盖区域和可变测序深度、检测灵敏度以及测试的单个基因的限制。我们确实确定了四个案例，其中在初始 NGS 中检测到低水平突变序列，但无法调用，这反映了技术限制或低于临床 NGS 面板可靠检测限制的非常小的克隆。值得注意的是，大多数具有后续测序数据的 MDS-NDM 患者继续显示没有可检测的突变，这与最初的阴性 NGS 结果一致。以及测试的单个基因。我们确实确定了四个案例，其中在初始 NGS 中检测到低水平突变序列，但无法调用，这反映了技术限制或低于临床 NGS 面板可靠检测限制的非常小的克隆。值得注意的是，大多数具有后续测序数据的 MDS-NDM 患者继续显示没有可检测的突变，这与最初的阴性 NGS 结果一致。以及测试的单个基因。我们确实确定了四个案例，其中在初始 NGS 中检测到低水平突变序列，但无法调用，这反映了技术限制或低于临床 NGS 面板可靠检测限制的非常小的克隆。值得注意的是，大多数具有后续测序数据的 MDS-NDM 患者继续显示没有可检测的突变，这与最初的阴性 NGS 结果一致。

与具有至少一个突变 (MDS-mut) 的 MDS 相比，MDS-NDM 患者更年轻，贫血更少，BM 细胞结构更低，并且包括更少的 RS 病例。其中一些差异可能归因于某些特定突变，例如与 RS 和贫血相关的SF3B1突变。尽管具有异常核型的病例比例与 MDS-mut 没有区别，但在 MDS-NDM 中具有孤立性 del(5q) 的 MDS 比在 MDS-mut 中更常见（8% 对 2%）。Del(5q)，无论是孤立的还是非复杂的核型，已知没有或比其他 MDS 亚型的突变更少。⁵某些 MDS-NDM 病例中的白血病发生可能与 MDS-mut 不同，尤其是那些携带明确功能驱动突变的病例。在之前对 MDS 病例的研究中，BM CD34+ 细胞显示出保留的 B 细胞分化⁶，未突变病例的比例相对较高，这表明 MDS 克隆可能始于 MDS-NDM 中较晚的髓样定型细胞。我们发现 MDS-NDM 和 MDS-mut 组之间存在类似的 IPSS-R 分布，并且在多变量分析中，缺乏可检测的突变显示出有利于 OS 的趋势，尽管在将分析限制在正常/ -Y 核型。值得注意的是，增加的突变负荷已被证明与 MDS 的预后恶化有关。^1、4

总之，我们表明，使用涵盖造血肿瘤中常见突变的大量基因的临床 NGS 平台，大约 10% 的真正的新发 MDS 病例缺乏可检测的致病突变。这些病例中约有三分之一没有细胞遗传学异常或原始细胞增多，在这些病例中，诊断仅依赖于持续的血细胞减少和发育不良。这部分病例可能导致对 MDS 的血细胞减少患者进行外周血 NGS 筛查时出现约 5% 的假阴性发生率。³当缺乏克隆性的遗传证据时，流式细胞术可以提供额外的免疫表型支持以确认 MDS 诊断。MDS-NDM 比 MDS-Mut 具有更有利的 OS 和无 AML 生存率，但是，很大一部分患者确实会进展为 AML 或死于 BM 失败。这种在临床表现和自然史中与 MDS-Mut 的相似性提出了一种可能性，即 MDS-NDM 病例可能含有标准测序面板中未涵盖的基因突变，或者由于技术限制而未在临床 NGS 面板中检测到。或者，驱动改变的 DNA 甲基化和/或转录重编程的其他机制值得进一步研究。