Adult autoimmune hemolytic anemia (AIHA) is a rare but potentially life-threatening acquired autoimmune disease in which autoantibodies directed toward antigens of autologous red blood cells (RBC) membrane lead to their accelerated destruction. Corticosteroids are the cornerstone first-line therapy for primary warm AIHA (wAIHA) and rituximab is commonly used off-label as a second-line option in most countries,¹ whereas for patients with cold agglutinin disease (CAD) who need to be treated, corticosteroids are of little efficacy and treatment with rituximab alone or in combination with bendamustine may be used depending on patients' age and comorbidities.¹ Adult patients with severe AIHA who are admitted in the intensive care unit (ICU) have a short-term mortality rate of 13% in ICU and of 30% after 1 year of follow-up.² In such severe, life-threatening cases of AIHA, the off-label use of IVIg is often considered^1-3 by analogy with immune thrombocytopenia (ITP) and other autoantibody-mediated autoimmune diseases. Based on its pathophysiology AIHA can be seen as a good candidate for the use of IVIg, but there is actually only little evidence supporting the efficacy of IVIg in this setting, as the sole large retrospective series focused on this topic was published in 1993.⁴

We report here the results of an observational multicenter retrospective (2013–2021) study. The aim of the study was to assess the immediate efficacy and safety of IVIg used as a “rescue” therapy for the management of adult AIHA and to identify some predicting factors of response. Most of the patients were identified via the CARMEN-FRANCE AIHA registry (NCT02877706), a prospective, multicenter, nationwide registry set up in 2016 for adult patients with a new diagnosis of AIHA. To be included in the study, patients had to (1) be ≥18 years old; (2) have a diagnosis of AIHA defined as hemoglobin level <12 g/dL, with ≥2 features of hemolysis (i.e., low haptoglobin level and/or elevated lactate dehydrogenase (LDH) level and/or elevated indirect bilirubin level), and a positive direct antiglobulin test (DAT) with no other underlying cause of acquired or hereditary hemolytic anemia; and (3) have received at least one course of IVIg for the management of AIHA. Patients with primary or secondary wAIHA, CAD, mixed AIHA, or Evans syndrome based on consensual criteria¹ could be included. Patients with DAT-negative AIHA were excluded as well as those with Evans syndrome who received IVIg specifically for the management of ITP. Previous treatment lines were defined by every type of treatment (erythropoietin, corticosteroids, immunosuppressors, etc.) received for treating AIHA before the first administration of IVIg except for RBC transfusions which were considered separately. Concomitant therapies were defined either as ongoing treatments at the time of IVIg administration or treatments initiated/administered for AIHA within 14 days after IVIg administration.

The primary endpoint was to assess the overall response rate (ORR) to IVIg on day 7. Response (R) was defined as an increase of the Hb level ≥2 g/dL on day 7 ± 1 post-IVIg compared to the baseline Hb level (i.e., on the day of the first IVIg administration), in the absence of any transfusion within 7 days after IVIg. A good response (GR) was defined as a Hb level ≥10 g/dL on day 7 ± 1 post IVIg, with at least a 2 g increase from baseline and in the absence of any transfusion within 7 days after IVIg. Patients with an Hb increase <2 g/dL and/or those who were transfused within 7 days after IVIg were considered as nonresponders (NR). Based on the same response criteria, the overall response rate (R + GR) was also assessed at day 14 ± 1 when data were available. Patients who received RBC transfusion between day 7 and day 14 were considered as NR on day 14. Continuous variables were presented as median (min-max). Categorical variables were expressed as numbers and percentages (%). The reference date for day 0 (D0) was defined as the date of the first administration of the first course of IVIg.

The primary outcome variable was calculated based on the change in Hb levels between D0 and D7 post-IVIg. For assessing the variables associated with R achievement (vs. NR) at D7, we conducted logistic regression model. The following variables were included in the model for univariate analyses: the time elapsed since between AIHA onset and the first IVIg administration patient age <or ≥60 years, gender, type of autoimmune hemolytic anemia (i.e., cold, warm, or mixed-AIHA), primary or secondary form of AIHA, severe form (Hb < 6 g/dL at D0), normal reticulocyte count (<120 × 10⁹/L), positive DAT for the C3 fraction of complement, identification of an infectious trigger for AIHA, newly diagnosed versus relapsed AIHA, concurrent use of corticosteroids or other concomitant treatments, and the number of transfusions received from D0 and D7. A significance threshold of p < 0.05 was applied for all statistical tests.

In total, among the 78 patients who were initially screened, 34 patients from 14 centers fulfilling eligibility criteria whose main baseline characteristics are described in Table 1 were eventually included (Figure S1). Patients had received a median of 1 (0–8) previous treatment lines prior to IVIg and up to a median of 5 (1–12) at the end of follow-up (Table 1). In total, 21 patients (62%) were transfused at least once prior to IVIg administration (median number of packed-RBC: 2 [1–15]). At time of IVIg administration, 82% of the patients were on corticosteroids and 41% were on weekly subcutaneous recombinant Epo.

Overall, 11 patients (32.4%) responded to IVIg on day 7, including two with GR and nine with R. Thirteen patients (38.2%) were transfused between D0 and D7 and were therefore considered NR. On day 14 after IVIg first administration, data on the Hb level were available for 28 patients (82.4%) and the ORR was 57%, including six patients who achieved a GR and 10 who achieved R. Different patterns of response (GR, R, and NR) observed after IVIg administration are illustrated in Figure S2.

On univariate analysis (Table S1) as well as multivariate analysis (data not shown), the only factor that was significantly associated with a higher response rate to IVIg was the early time of IVIg administration after AIHA onset (ORR 14.571 [1.866–338.736], p = 0.030). Regarding safety, no case of AIHA exacerbation due to IVIg-induced hemolysis was observed.

One case of thromboembolic event with a good outcome (partial thrombosis of the splenic artery followed by a thrombosis of the inferior vena cava and pulmonary embolism) occurred 5 days after IVIg administration in a 60-year-old man.

In the present retrospective series on 34 adult patients with severe and/or transfusion-dependent AIHA treated with IVIg, the ORR was 32.4% on Day 7. The safety of IVIg was good as only one potentially related thromboembolic event was observed, in a patient with recent COVID-19 and active hemolysis. To the best of our knowledge, this is the largest study focused on this topic since 1993.⁴ Among responders, different patterns of response to IVIg were observed including an immediate and clear increase in the Hb level suggesting an FcR-blockade like mechanism in few of them. The only parameter associated with a better response to IVIg was the early administration of IVIg in newly diagnosed AIHA. In the previous pilot study on 37 patients (including 32 adults and five children) reported in 1993, the ORR (i.e., increase of Hb level of at least 2 g/dL within 10 days) to IVIg was 39.7%.⁴ However, the authors did not specify whether some patients had been transfused during this period of time⁴ and interestingly higher doses of IVIg (5–7 g/kg) were not associated with a higher efficacy. Furthermore, two other small observational studies with respectively five and 17 patients with AIHA did not find any evidence for IVIg efficacy, with an unchanged survival rate of 51Cr-labeled autologous RBC in 4/5 patients studied.^{5, 6} Our study has some limitations: first, due to its retrospective and uncontrolled design, some data could have been missed, and memorization bias could not be excluded for the few patients recruited outside the CARMEN registry. Moreover, since most of the patients included had severe AIHAs with a relatively high (14%) rate of mixed AIHAs which are known to be more severe,¹ and were receiving concomitant therapies, efficacy of IVIg was difficult to assess in some patients and the results cannot be extrapolated to every type of AIHA. Lastly, the relative low number of patients included may have reduced the chance to identify some parameters that could be associated with a higher response rate and in particular the DAT pattern and the type of AIHA (i.e., warm, mixed or cold).

In conclusion, based on this retrospective observational study, we confirm that in addition to standard of care, the off-label use of IVIg may be helpful in one third of adults managed for severe AIHA. In the absence of any clear predicting factor of response, and taking into account the cost and the recurrent problems of IVIg shortage, the use of IVIg in this setting should be limited and well weighed.

成人自身免疫性溶血性贫血（AIHA）是一种罕见但可能危及生命的获得性自身免疫性疾病，其中针对自体红细胞（RBC）膜抗原的自身抗体导致其加速破坏。皮质类固醇是原发性温性 AIHA (wAIHA) 的基础一线治疗，而利妥昔单抗在大多数国家通常是标签外使用的二线选择，¹而对于需要治疗的冷凝集素病 (CAD) 患者，皮质类固醇的疗效不大，根据患者的年龄和合并症，可以单独使用利妥昔单抗或与苯达莫司汀联合治疗。¹入住重症监护病房 (ICU) 的成年严重 AIHA 患者，ICU 内的短期死亡率为 13%，随访 1 年后为 30%。²在这种严重、危及生命的 AIHA 病例中，IVIg 的超说明书使用通常被认为是^1-3，与免疫性血小板减少症 (ITP) 和其他自身抗体介导的自身免疫性疾病类比。根据其病理生理学，AIHA 可以被视为使用 IVIg 的良好候选者，但实际上只有很少的证据支持 IVIg 在这种情况下的功效，因为专注于该主题的唯一大型回顾性系列于 1993 年发表。⁴

我们在此报告一项观察性多中心回顾性（2013-2021）研究的结果。该研究的目的是评估 IVIg 用作治疗成人 AIHA 的“救援”疗法的即时疗效和安全性，并确定一些反应的预测因素。大多数患者是通过 CARMEN-FRANCE AIHA 登记处 (NCT02877706) 确定的，这是一个前瞻性、多中心、全国性的登记处，成立于 2016 年，针对新诊断为 AIHA 的成年患者。要纳入研究，患者必须 (1) ≥ 18 岁； (2) 诊断为 AIHA，定义为血红蛋白水平 <12 g/dL，且具有≥2 个溶血特征（即触珠蛋白水平低和/或乳酸脱氢酶 (LDH) 水平升高和/或间接胆红素水平升高），并且直接抗球蛋白试验（DAT）呈阳性，且无其他导致获得性或遗传性溶血性贫血的根本原因； (3) 已接受至少一个疗程的 IVIg 治疗 AIHA。根据共识标准¹ ，患有原发性或继发性 wAIHA、CAD、混合型 AIHA 或 Evans 综合征的患者可纳入其中。 DAT 阴性 AIHA 患者以及专门接受 IVIg 治疗 ITP 的 Evans 综合征患者被排除在外。先前的治疗线由首次给予 IVIg 之前接受的用于治疗 AIHA 的每种治疗类型（促红细胞生成素、皮质类固醇、免疫抑制剂等）定义，但红细胞输注除外，红细胞输注需单独考虑。伴随治疗被定义为在给予 IVIg 时正在进行的治疗或在给予 IVIg 后 14 天内针对 AIHA 开始/给予的治疗。

主要终点是评估第 7 天对 IVIg 的总体缓解率 (ORR)。缓解 (R) 定义为与基线 Hb 相比，IVIg 后第 7 ± 1 天 Hb 水平增加 ≥2 g/dL IVIg 后 7 天内未进行任何输血的情况下的水平（即首次 IVIg 给药当天）。良好反应 (GR) 的定义是 IVIg 后第 7 ± 1 天的 Hb 水平≥10 g/dL，与基线相比至少增加 2 g，且 IVIg 后 7 天内没有任何输血。 Hb 增加<2 g/dL 的患者和/或 IVIg 后 7 天内输血的患者被视为无反应者 (NR)。基于相同的缓解标准，还在第 14 ± 1 天（当数据可用时）评估了总体缓解率 (R + GR)。在第 7 天和第 14 天之间接受红细胞输注的患者在第 14 天被视为 NR。连续变量以中位数（最小-最大）表示。分类变量以数字和百分比 (%) 表示。第 0 天 (D0) 的参考日期定义为首次施用第一个疗程 IVIg 的日期。

主要结果变量是根据 IVIg 后 D0 和 D7 之间 Hb 水平的变化计算的。为了评估与 D7 时的 R 成绩（与 NR）相关的变量，我们进行了逻辑回归模型。单变量分析模型中包含以下变量： 自 AIHA 发病到首次 IVIg 给药之间的时间间隔 患者年龄 < 或 ≥ 60 岁、性别、自身免疫性溶血性贫血类型（即冷型、温型或混合型 AIHA） )、原发性或继发性 AIHA、严重型（D0 时 Hb < 6 g/dL）、正常网织红细胞计数 (<120 × 10 ⁹ /L)、补体 C3 部分 DAT 阳性、识别感染触发因素AIHA、新诊断与复发的 AIHA、同时使用皮质类固醇或其他联合治疗，以及从 D0 和 D7 接受的输血次数。 所有统计检验均采用p < 0.05的显着性阈值。

总共，在最初筛选的 78 名患者中，最终纳入了来自 14 个中心、符合资格标准的 34 名患者，其主要基线特征如表 1 所示（图 S1）。患者在 IVIg 之前接受过中位数为 1 (0–8) 的治疗线，在随访结束时接受的治疗线中位数为 5 (1–12)（表 1）。总共有 21 名患者 (62%) 在 IVIg 给药前至少接受过一次输血（浓缩红细胞中位数：2 [1-15]）。在给予 IVIg 时，82% 的患者服用皮质类固醇，41% 的患者每周皮下注射重组 Epo。

总体而言，11 名患者 (32.4%) 在第 7 天对 IVIg 有反应，其中 2 名患者为 GR，9 名患者为 R。13 名患者 (38.2%) 在 D0 和 D7 之间接受了输血，因此被认为是 NR。在 IVIg 首次给药后第 14 天，可获得 28 名患者 (82.4%) 的 Hb 水平数据，ORR 为 57%，其中 6 名患者达到 GR，10 名患者达到 R。不同的反应模式（GR、R和 NR）在 IVIg 给药后观察到的结果如图 S2 所示。

在单变量分析（表 S1）和多变量分析（数据未显示）中，与 IVIg 较高反应率显着相关的唯一因素是 AIHA 发病后 IVIg 给药的早期时间（ORR 14.571 [1.866–338.736] ，p  = 0.030）。关于安全性，没有观察到由于 IVIg 引起的溶血而导致 AIHA 恶化的病例。

一名 60 岁男性在注射 IVIg 后 5 天发生一例结局良好的血栓栓塞事件（脾动脉部分血栓形成，随后是下腔静脉血栓形成和肺栓塞）。

在目前对 34 名患有严重和/或输血依赖性 AIHA 的成人患者进行 IVIg 治疗的回顾性系列中，第 7 天的 ORR 为 32.4%。IVIg 的安全性良好，因为在一名患者中仅观察到一种潜在相关的血栓栓塞事件最近患有新冠肺炎 (COVID-19) 和活动性溶血。据我们所知，这是自 1993 年以来关注这一主题的最大规模的研究^。4在应答者中，观察到对 IVIg 的不同反应模式，包括 Hb 水平立即明显升高，表明少数人存在类似 FcR 阻断的机制。其中。与 IVIg 更好反应相关的唯一参数是在新诊断的 AIHA 中尽早给予 IVIg。在 1993 年报告的先前针对 37 名患者（包括 32 名成人和 5 名儿童）的试点研究中，IVIg 的 ORR（即 10 天内 Hb 水平增加至少 2 g/dL）为 39.7%。⁴然而，作者没有具体说明是否有一些患者在此期间接受过输血⁴，有趣的是，较高剂量的 IVIg（5-7 g/kg）与较高的疗效并不相关。此外，另外两项小型观察性研究分别针对 5 名和 17 名 AIHA 患者进行，没有发现任何 IVIg 疗效的证据，在所研究的 4/5 患者中，51Cr 标记的自体红细胞的生存率没有变化。^{5, 6}我们的研究有一些局限性：首先，由于其回顾性和不受控制的设计，可能会遗漏一些数据，并且不能排除在 CARMEN 注册中心之外招募的少数患者的记忆偏差。此外，由于大多数纳入的患者患有严重的 AIHA，其中混合型 AIHA 的发生率相对较高（14%），已知这种情况更为严重，¹并且正在接受伴随治疗，因此很难评估 IVIg 在某些患者中的疗效，并且结果不能推断到每种类型的 AIHA。最后，纳入的患者数量相对较少可能减少了识别一些可能与较高反应率相关的参数的机会，特别是 DAT 模式和 AIHA 类型（即热型、混合型或冷型）。

总之，根据这项回顾性观察研究，我们确认，除了标准护理之外，超说明书使用 IVIg 可能对三分之一患有严重 AIHA 的成人有所帮助。在缺乏任何明确的反应预测因素的情况下，并考虑到成本和反复出现的 IVIg 短缺问题，在这种情况下 IVIg 的使用应受到限制并充分权衡。