To the Editor—Recently published data suggest that cytomegalovirus (CMV) DNAemia, even when present at magnitudes below the threshold levels commonly used to trigger the inception of preemptive antiviral therapy (PET) [1] is associated with increased overall and nonrelapse mortality in allogeneic hematopoietic stem cell transplant recipients (allo-HSCT) [2], although this extent remains controversial [3]. Elucidation of this issue is of paramount relevance because the use of PET would have to be discouraged were the veracity of this assumption proved. Hill et al. [4] recently reported that the area under a curve (AUC) for plasma CMV DNAemia was independently associated with overall mortality. Here we provide data that challenge this finding and suggest that methodological differences in calculating AUCs may account, at least in part, for this apparent discrepancy. This retrospective study included a total of 196 consecutive patients with hematological malignancies who underwent T-cell replete allo-HSCT from May 2010 to April 2017. CMV DNA in plasma was quantified using real-time polymerase chain reaction (PCR) assays. A total of 142 (72.4%) patients developed 1 (n = 76), 2 (n = 41), 3 (n = 20), or 4 (n = 5) episodes of CMV DNAemia. Overall, 101 patients (71.1%) required 1 or more courses of antiviral therapy. The median CMV DNA log₁₀ load AUC_0-180 and AUC_t were 4.38 log₁₀ copies × days × mL⁻¹ (range, 2.46 to 7.28) and 4.50 log₁₀ copies × days × ml⁻¹ (range, 2.46 to 7.28), respectively. Patients treated with antivirals displayed higher AUC_t values than those who spontaneously resolved CMV DNAemia (median, 4.79 log₁₀ vs. 3.61 log₁₀; P = < .001; Mann-Whitney U test). To assess the potential association between CMV DNA log₁₀ load AUCs and mortality, we arbitrarily stratified AUC log₁₀ values into 4 categories (2–3, 3–4, 4–5, >5). The cumulative incidence of overall mortality both by day 180 and by day 365 after transplantation was not associated with the magnitude of the respective CMV DNA log₁₀ load AUC (footnote Table 1). Cox models identified the use of anti- thymocyte globulin (ATG) in the conditioning regimen, the occurrence of severe aGvHD (grades III to IV), and treatment with antivirals (ganciclovir in most cases and only for mortality by day 365), but not CMV DNAemia AUCs as factors that were independently associated with overall mortality (Table 1). Although dissimilarities between the study of Hill et al. [4] and ours in terms of patient characteristics, transplant modalities, sample size (ours was smaller), the variables introduced into and adjusted for in the Cox models, the overall CMV DNA monitoring period length as well as in the criteria adopted to initiate preemptive antiviral therapy may contribute to explaining this discrepancy, we suggest that our use of a different mathematical approach to calculate AUCs [4, 5] may also account for this apparent contradiction. We used the curve-trapezoid rule method [6], which is classically employed in drug pharmacokinetic and pharmacodynamic studies. Thus, in view of the divergent nature of our data compared to previously published data [4] and the relevance of its clinical significance, further well-powered prospective studies are warranted to true relationship between these factors.

中文翻译：

---

同种异体造血干细胞移植后巨细胞病毒DNAemia负担和死亡率：基于曲线的研究方法下的区域

致编辑—最近发表的数据表明，即使巨细胞病毒（CMV）DNA血症的水平低于通常用于触发先发性抗病毒治疗（PET）的阈值水平[1]，也与同种异体造血干细胞的整体和非复发死亡率增加相关移植受者（allo-HSCT）[2]，尽管这一程度仍存在争议[3]。阐明此问题至关重要，因为如果证明这一假设的准确性，则不鼓励使用PET。希尔等。[4]最近报道血浆CMV DNAemia的曲线下面积（AUC）与总体死亡率独立相关。在这里，我们提供的数据对这一发现提出了挑战，并建议在计算AUC时方法上的差异至少可以部分解释为：对于这种明显的差异。这项回顾性研究包括从2010年5月至2017年4月接受T细胞补充all-HSCT的196例连续的血液系统恶性肿瘤患者。使用实时聚合酶链反应（PCR）分析对血浆中的CMV DNA进行定量。共有142名（72.4％）患者发生了1例（n = 76），2例（n = 41），3例（n = 20）或4例（n = 5）的CMV DNA血症发作。总体而言，有101名患者（占71.1％）需要接受1疗程或更多疗程的抗病毒治疗。中值CMV DNA对数 或CMV DNAemia发生4次（n = 5）。总体而言，有101名患者（占71.1％）需要接受1疗程或更多疗程的抗病毒治疗。中值CMV DNA对数 或CMV DNAemia发生4次（n = 5）。总体而言，有101名患者（占71.1％）需要接受1或更多疗程的抗病毒治疗。中值CMV DNA对数₁₀载荷AUC _0-180和AUC _t分别为4.38 log ₁₀份×天×mL ^-1（范围2.46至7.28）和4.50 log ₁₀份×天×ml ^-1（范围2.46至7.28）。抗病毒药物治疗的患者显示的AUC _t值高于自发解决CMV DNA血症的患者（中位数，4.79 log ₁₀ vs. 3.61 log ₁₀；P = <0.001； Mann-Whitney U检验）。为了评估CMV DNA log ₁₀负荷AUC与死亡率之间的潜在关联，我们任意分层了AUC log ₁₀值分为4类（2-3、3-4、4-5，> 5）。移植后180天和365天总死亡率的累积发生率与相应CMV DNA log ₁₀的大小无关加载AUC（脚注表1）。Cox模型确定了在调理方案中使用抗胸腺细胞球蛋白（ATG），发生严重的aGvHD（III至IV级）和使用抗病毒药治疗（大多数情况下使用更昔洛韦，并且仅在365天时可导致死亡率）， CMV DNAemia AUCs是与总死亡率独立相关的因素（表1）。尽管希尔等人的研究之间存在差异。[4]和我们在患者特征，移植方式，样本量（我们的较小），在Cox模型中引入和调整的变量，整体CMV DNA监测期长度以及采用的启动标准方面抢先抗病毒治疗可能有助于解释这种差异，我们建议我们使用不同的数学方法来计算AUC [4，[5]也可能解释了这种明显的矛盾。我们使用了曲线梯形法则[6]，该方法通常用于药物药代动力学和药效学研究。因此，鉴于我们的数据与先前发表的数据相比具有不同的性质[4]，以及其临床意义的相关性，有必要对这些因素之间的真实关系进行有力的前瞻性研究。