Acute myeloid leukemia (AML) is a biologically and clinically heterogenous disease with diverse genetic abnormalities^1-6 and a wide-ranging white blood cell counts (WBC) at diagnosis.¹ The 2022 European LeukemiaNet (ELN) genetic-risk classification incorporates cytogenetic and selected molecular alterations to define Favorable, Intermediate, and Adverse genetic-risk groups providing valuable prognostic information.¹ Pretreatment WBC levels are also a main prognostic factor for patients with AML;^{7, 8} however, there are no standardized WBC-associated groups that improve prognostication for younger patients. de Jonge et al.⁹ first investigated the impact of WBC on outcome of AML patients harboring NPM1 mutations and FLT3 internal tandem duplications (FLT3-ITD) that were divided into three WBC groups: <20 000/μL, WBC 20 000/μL–100 000/μL, and ≥100 000/μL. We decided to build upon this previous work by analyzing a larger, clinically and molecularly well-characterized cohort of 1121 younger (aged <60 years) adults with de novo AML to define clinically relevant WBC level groups and determine their potential independent prognostic impact and associations with gene-expression profiles.

We defined three WBC groups: low (<10 000/μL, n = 298 patients), intermediate (10 000–49 999/μL, n = 488), and high (≥50 000/μL, n = 335). These groups were chosen because there was no correlation with an observable steady decrease in any analyzed outcome endpoint, including disease-free (DFS), event-free (EFS), or overall (OS) survival, with WBC increasing above 50 000/μL by 10 000/μL increments, nor were there any linear changes in endpoints associated with WBC decreasing below 10 000/μL in 1000/μL increments across the entire cohort (Figure S1A,B). All patients were similarly treated with a cytarabine/anthracycline based induction on frontline Cancer and Leukemia Group B (CALGB)/Alliance for Clinical Trials in Oncology (Alliance) protocols, with details of CALGB treatment protocols provided in the Data S1. No patient included in the analysis received an allogeneic hematopoietic stem-cell transplantation (HSCT) in first complete remission (CR), and patients who underwent an allogeneic HSCT off-study were not included because of incomplete or missing follow-up information. All patients were enrolled on CALGB 8461 (cytogenetic studies), CALGB 9665 (leukemia tissue bank), and CALGB 20202 (molecular studies) companion protocols, with further treatment details provided in the Data S1. The mutational status of 80 total protein-coding genes was determined centrally at The Ohio State University by targeted amplicon sequencing using the MiSeq platform (Illumina)⁶ and further detailed in the Data S1. Analysis of differentially expressed genes within each WBC group was done on peripheral blood samples via total transcriptome RNAseq with subsequent gene set enrichment analyses (GSEA) via Hallmark/Kegg pathways of 40% of our entire cohort. Clinical endpoints were defined according to generally accepted criteria.¹ Baseline characteristics were compared using the Fisher's exact and Wilcoxon rank-sum tests for categorical and continuous variables, respectively.¹⁰ Estimated probabilities of DFS and OS were calculated using the Kaplan–Meier method,¹¹ and the log-rank test evaluated differences between survival distributions. Multivariable logistic regression models were generated for attainment of CR, and multivariable proportional hazards models were constructed for DFS and OS using a limited backward elimination procedure, with variables significant at α = 0.2 from the univariable analyses considered for multivariable analyses. All analyses were performed by the Alliance Statistics and Data Center on a database locked on June 9, 2020 using SAS 9.4 and TIBCO Spotfire S + 8.2.

Pretreatment patient characteristics are shown in Table S1. The median age for the three WBC groups (low, intermediate, and high) was similar (46, 44, and 43 years, respectively), and 55% of patients were male. Patients in the high WBC group had the highest extramedullary disease burden at diagnosis followed by patients in the intermediate and low groups (38% vs. 29% and 12%, respectively; p < .001). Patients in the intermediate WBC group more often had core-binding factor AML (CBF-AML; p < .001), that is, harbored inv(16)(p13.1q22)/t(16;16)(p13.1;q22) or t(8;21)(q22;q22)¹² (Table S2), and the low WBC group patients had complex karyotype more often (p < .001). Patients in the high WBC group presented more frequently with FLT3-ITD than those in the intermediate and, especially, low groups (36% vs. 24% vs. 12%, respectively; p < .001). Among patients with FLT3-ITD, those in the high WBC group had more often high allelic ratio (≥50%) than patients in the intermediate and low groups (30% vs. 17% vs. 5%, respectively; p < .001).

Patients in the intermediate WBC group had a higher CR rate (p < .001), longer DFS (p = .002; Figure S2), EFS (p < .001; Figure 1A), and OS (p < .001; Figure 1B) than patients in the other two WBC groups (Table S3). Intermediate WBC group patients also had higher CR rates (p = .009; Table S4), longer DFS (p = .01; Table S5), EFS (p < .001; Table S6), and OS (p = .01; Table S7) in multivariable modeling, after adjusting for other variables provided in Tables S4–S7.

Among patients categorized according to the 2022 ELN classification, those in the Favorable group who had an intermediate WBC had longer EFS (p < .004; Figure 1C) and were associated by trend with a longer DFS (p = .08; Figure S3) and OS (p = .06; Figure S4) than patients with high WBC. WBC group did not affect patient outcome in the 2022 ELN Intermediate or Adverse groups. Within the 2022 ELN Favorable group, intermediate WBC was associated with longer DFS (p = .04), EFS (p = .002), and OS (p = .005) in patients with CEBPA^bZIP mutations or NPM1 mutations with no FLT3-ITD, but not in patients with CBF-AML (Figure 1D–F). Our data thus suggest it may be time to rethink the paradigm that all AML patients who present with an elevated WBC at diagnosis have a greater probability of relapse or worsened survival. For example, younger adults with AML classified in high WBC group who harbor genetic alterations assigning them to either 2022 ELN Intermediate or Adverse group do not have worse outcomes than other patients assigned to the same 2022 ELN genetic-risk groups who have low or intermediate WBC levels. In contrast, our data suggest that a more aggressive post-remission treatment is needed for patients categorized in the 2022 ELN Favorable group who belong to either high or low WBC group. This is especially relevant for 2022 ELN Favorable patients who carry CEBPA^bZIP mutations or have NPM1 mutations in the absence of FLT3-ITD. Conversely, younger adults with AML and a 2022 ELN Favorable profile belonging to the intermediate WBC group do better than those in the low or high WBC categories. This finding requires further validation and exploration.

We next analyzed the differences in WBC group-associated gene-expression profiles. Patients in the high WBC group had upregulation of the C-X-C motif chemokine ligand 10 (CXCL10) and hypoxia inducible lipid droplet associated (HILPDA) genes, and downregulation of the fibronectin 1 (FN1) and mesothelin (MSLN) genes compared with the intermediate group (Figure S5A). This suggests a predilection for inflammatory cytokine signaling and upregulation of fat storage and fatty acid metabolism in this group, which also mimicked the GSEA results seen in the high WBC group (Figure S6A). Conversely, MMP7 and GZMA were upregulated in the low WBC group compared with the intermediate group (Figure S5B), which may suggest a common catabolic/destructive pathway as also seen in its GSEA with increasing inflammation (i.e., interferon α/γ) (Figure S6B). Importantly, high inflammation risk-score recently emerged as a strong prognosticator of treatment outcome in AML.¹³ While higher inflammation was associated with select molecular features, its prognostic significance was independent from established molecular risk markers.¹³ Our data suggest that the adverse prognostic impact of high WBC may, in part, be mediated by changes in the inflammatory response, thereby supporting exploration of therapies targeting inflammation to mitigate the adverse prognosis.

Given the clinically relevant sequelae of hyperleukocytosis (HL),¹⁴ we also analyzed the extreme WBC ranges of our patient cohort with WBC ≥50 000/μL. We observed that patients with a WBC >200 000/μL had OS similar to OS of patients with a WBC between 50 000/μL and 200 000/μL (Figure S1B). Our analysis is limited by a relatively small number of only 30 patients presenting with a WBC >200 000/μL, but this may suggest that the “extreme” ends of HL may be equivalent to the “lower” ends of clinically defined HL. There is currently mixed data concerning best initial management for patients who present with HL regarding leukapheresis compared to more immediate chemotherapy.^15-19 Currently it is unclear as to how to best manage patients presenting with HL in regard to leukapheresis compared to more immediate chemotherapy. Retrospective data have shown variable outcomes with either approach and there are multiple barriers due to the acuity of these patients in performing a prospective study to better determine selection of initial management. These WBC groups should not be used to determine management for patients with clinical signs or symptoms of leukostasis, but serve as a prognostic tool in conjunction with a physician's clinical expertise to help guide therapy decisions.

In conclusion, we propose three WBC groups to offer additional prognostic information for younger AML patients. Our study revealed that an intermediate WBC group was associated with better outcomes among all patients, with higher CR rates and longer survival in all calculated metrics (DFS, EFS, OS) in multivariable modeling compared with patients in the low and high WBC groups. Patients in the 2022 ELN Favorable group, especially those with non-CBF-AML, benefited from having intermediate WBC levels at diagnosis. Clear differences in the metabolic pathways exist among WBC groups. These WBC groups offer additional prognostic information for younger AML patients.

急性髓性白血病 （AML） 是一种生物学和临床异质性疾病，诊断时具有不同的遗传异常^1-6 和广泛的白细胞计数 （WBC）。¹ 2022 年欧洲白血病网 （ELN） 遗传风险分类结合了细胞遗传学和选定的分子改变，以定义有利、中等和不良遗传风险组，提供有价值的预后信息。¹ 治疗前 WBC 水平也是 AML 患者的主要预后因素;^{7， 8} 然而，没有标准化的 WBC 相关组可以改善年轻患者的预后。de Jonge 等人⁹ 首先研究了 WBC 对携带 NPM1 突变和 FLT3 内部串联重复 （FLT3-ITD） 的 AML 患者预后的影响，这些患者分为三个 WBC 组：<20 000/μL、WBC 20 000/μL–100 000/μL 和 ≥100 000/μL。我们决定在先前工作的基础上，分析一个更大的、临床和分子特征明确的队列，该队列由 1121 名年轻 （<60 岁） 新发 AML 成人组成，以确定临床相关的 WBC 水平组，并确定它们潜在的独立预后影响和与基因表达谱的关联。

我们定义了三个 WBC 组：低 （<10 000/μL，n = 298 名患者）、中 （10 000–49 999/μL，n = 488） 和高 （≥50 000/μL，n = 335 ）。 选择这些组是因为与任何分析结果终点的可观察到的稳定下降无关，包括无病 （DFS）、无事件 （EFS） 或总 （OS） 生存率，WBC 以 10 000/μL 的增量增加到 50 000/μL 以上，并且与整个队列中 WBC 以 1000/μL 的增量降至 10 000/μL 以下相关的终点也没有任何线性变化（图 S1A，所有患者在一线癌症和白血病 B 组 （CALGB）/肿瘤临床试验联盟 （Alliance） 方案中接受了基于阿糖胞苷/蒽环类药物的诱导治疗，数据 S1 中提供了 CALGB 治疗方案的详细信息。纳入分析的患者在首次完全缓解 （CR） 时未接受同种异体造血干细胞移植 （HSCT），并且由于随访信息不完整或缺失，接受同种异体 HSCT 非研究的患者未被纳入。所有患者均接受 CALGB 8461（细胞遗传学研究）、CALGB 9665（白血病组织库）和 CALGB 20202（分子研究）配套方案入组，数据 S1 中提供了进一步的治疗细节。在俄亥俄州立大学，通过使用 MiSeq 平台 （Illumina）⁶ 的靶向扩增子测序集中确定了 80 个总蛋白编码基因的突变状态，并在数据 S1 中进一步详细说明。 通过总转录组 RNAseq 对外周血样本进行每个 WBC 组内差异表达基因的分析，随后通过我们整个队列中 40% 的 Hallmark/Kegg 通路进行基因集富集分析 （GSEA）。临床终点根据普遍接受的标准定义。¹ 分别使用分类变量和连续变量的 Fisher 精确和 Wilcoxon 秩和检验比较基线特征。¹⁰ 使用 Kaplan-Meier 方法计算 DFS 和 OS 的估计概率，¹¹ 对数秩检验评估生存分布之间的差异。为达到 CR 生成多变量 logistic 回归模型，并使用有限的向后消除程序为 DFS 和 OS 构建多变量比例风险模型，其中 α = 0.2 的变量在单变量分析中具有显著性，用于多变量分析。所有分析均由 Alliance Statistics and Data Center 使用 SAS 9.4 和 TIBCO Spotfire S + 8.2 在 2020 年 6 月 9 日锁定的数据库上执行。

治疗前患者特征如表 S1 所示。三个 WBC 组 （低、中和高） 的中位年龄相似 （分别为 46 、 44 和 43 岁），55% 的患者为男性。高 WBC 组患者在诊断时髓外疾病负担最高，其次是中低组患者 （分别为 38% vs. 29% 和 12%;p < .001）.中间 WBC 组患者更常有核心结合因子 AML （CBF-AML;p < .001），即藏匿 inv（16）（p13.1q22）/t（16;16）（第 13.1 页;Q22） 或 T（8;21）（q22;q22）¹² （表 S2），低 WBC 组患者更常具有复杂的核型 （p < .001）。高 WBC 组患者比中间组，尤其是低组患者更频繁地出现 FLT3-ITD （分别为 36% 和 24% 和 12%;p < .001）.在 FLT3-ITD 患者中，高 WBC 组的患者比中间组和低组患者更常具有高等位基因比 （≥50%） （分别为 30% vs. 17% vs. 5%;p < .001）.

中间 WBC 组患者具有较高的 CR 率 （p < .001），较长的 DFS （p = .002;图 S2）、EFS （p < .001;图 1A） 和 OS （p < .001;图 1B） 比其他两个 WBC 组的患者（表 S3）。中间WBC 组患者的 CR 率也较高 （p = .009;表 S4）、较长的 DFS （p = .01;表 S5）、EFS（p < .001;表 S6） 和 OS （p = .01;表 S7），在调整了表 S4-S7 中提供的其他变量后。

在根据 2022 年 ELN 分类分类的患者中，具有中间 WBC 的 Favorable 组中的 EFS 更长（p < .004;图 1C），并且按趋势与较长的 DFS 相关 （p = .08;图 S3）和 OS （p = .06;图 S4）比 WBC 高的患者。WBC 组不影响 2022 年 ELN 中间组或不良事件组的患者预后。在 2022 年 ELN 有利组中，中间 WBC 与较长的 DFS （p = .04）、EFS （p = .002） 和 OS （p = .005） 相关，在无 FLT3-ITD 的 NPM1 突变患者中，但在 CBF-AML 患者中则不相关（图 1D-F）。因此，我们的数据表明，可能是时候重新考虑所有诊断时 WBC 升高的 AML 患者复发或生存率恶化的可能性更大。例如，被归类为高 WBC 组的年轻 AML 患者，如果携带基因改变，将他们分配到 2022 ELN 中间组或不良事件组，其结果并不比分配到相同 2022 ELN 遗传风险组且 WBC 水平低或中等的其他患者更差。相比之下，我们的数据表明，属于 2022 年 ELN 良好组且属于高 WBC 组或低 WBC 组的患者需要更积极的缓解后治疗。这对于携带 CEBPA^bZIP 突变或在没有 FLT2022-ITD 的情况下具有 NPM1 突变的 3 ELN Favorable 患者尤其相关。 相反，属于中间 WBC 组的患有 AML 和 2022 年 ELN 良好特征的年轻人比低或高 WBC 类别中的年轻人表现更好。这一发现需要进一步的验证和探索。

接下来，我们分析了 WBC 组相关基因表达谱的差异。与中间组相比，高 WBC 组患者 C-X-C 基序趋化因子配体 10 （CXCL10） 和缺氧诱导脂滴相关 （HILPDA） 基因上调，纤连蛋白 1 （FN1） 和间皮素 （MSLN） 基因下调 （图 S5A）。这表明该组偏爱炎性细胞因子信号传导以及脂肪储存和脂肪酸代谢的上调，这也模拟了高 WBC 组的 GSEA 结果（图 S6A）。相反，与中间组相比，低 WBC 组的 MMP7 和 GZMA 上调（图 S5B），这可能表明一个共同的分解代谢/破坏性途径，正如在其 GSEA 中也看到的那样，随着炎症的增加（即干扰素 α/γ）（图 S6B）。重要的是，高炎症风险评分最近成为 AML 治疗结果的有力预测指标。¹³ 虽然较高的炎症与选定的分子特征相关，但其预后意义与已建立的分子风险标志物无关。¹³ 我们的数据表明，高 WBC 的不良预后影响可能部分是由炎症反应的变化介导的，从而支持探索针对炎症的疗法以减轻不良预后。

鉴于白细胞增多症 （HL） 的临床相关后遗症，¹⁴ 我们还分析了 WBC ≥ 50 000/μL 的患者队列的极端 WBC 范围。我们观察到 WBC >200 000/μL 患者的 OS 与 WBC 在 50 000/μL 和 200 000/μL 之间的患者的 OS 相似（图 S1B）。我们的分析受到相对少数的 30 名 WBC >200 000/μL 患者的限制，但这可能表明 HL 的“极端”端可能相当于临床定义的 HL 的“下端”。目前关于荷金淋巴瘤患者白细胞去除术与更即时化疗相比的最佳初始管理的数据不一。^15-19 目前尚不清楚与更即时的化疗相比，如何最好地管理出现 HL 的白细胞去除术患者。回顾性数据显示，这两种方法的结果各不相同，并且由于这些患者在进行前瞻性研究以更好地确定初始治疗的选择时存在多种障碍。这些 WBC 组不应用于确定有白细胞淤滞临床体征或症状的患者的治疗，而应作为预后工具，结合医生的临床专业知识，帮助指导治疗决策。

总之，我们提出了三个 WBC 组，为年轻的 AML 患者提供额外的预后信息。我们的研究表明，与低 WBC 和高 WBC 组的患者相比，中间 WBC 组在所有患者中与更好的结果相关，在多变量模型中具有更高的 CR 率和更长的生存率 （DFS、EFS、OS）。2022 年 ELN 有利组的患者，尤其是非 CBF-AML 患者，受益于诊断时具有中等 WBC 水平。WBC 组之间存在明显的代谢途径差异。这些 WBC 组为年轻 AML 患者提供了额外的预后信息。