Autoimmune hemolytic anemia (AIHA) is defined as augmented destruction of erythrocytes by autoimmune mechanisms, usually mediated by autoantibodies against erythrocyte surface antigens.¹ Drug-induced hemolytic anemia is distinct cause of AIHA and is generally mediated by immunological mechanisms.¹ Immune checkpoint inhibitors (ICIs) that are used to treat cancer, in particular PD-1 inhibitors, can induce AIHA.^{1, 2}

The hematopoietic system and red blood cells could be occasionally involved as immune-related adverse events triggered by ICI with a frequency estimated to be around 0.5% of treated patients.³ ICI-AIHAs have only been reported in few patients in cases-series,^{2, 3} and data remain scarce about AIHA subtype distributions (warm vs. cold) and outcome of patients including mortality associated with AIHA and treatment efficacy. Management of patients with ICI-AIHA is based on corticosteroids and eventually rituximab,^{2, 3} and safety and efficacy data are not fully reported. We report here the management and therapeutic outcome of patients treated for ICI-AIHA in a tertiary academic anticancer center.

The study design was an observational retrospective study of adult patients with ICI-AIHA in France. Immune checkpoint inhibitors considered for the study included anti-programmed death cell 1 (PD-1), anti-programmed death cell ligand 1 (PD-L1), and anti-CTLA4. Eligibility criteria were patients aged ≥18 years, with ICI-AIHA who were registered in the pharmacovigilance registry REISAMIC (Registry of Severe Adverse Reactions to Immunomodulatory Monoclonal Antibodies in Cancer)⁴ and those referred to the Gustave Roussy Immunotoxicity board⁵ over the period from June 27, 2014, to April 29, 2022, and those referred to the centers belonging to the French Reference Centers for Adult' Autoimmune Cytopenias (CeReCAI) network from 2014 to 2022 (Figure 1). The AIHA criteria and definitions for this study followed the 2020 first international consensus criteria for AIHA management.⁶ Study objectives, data collection information, statistical analysis, and ethical considerations are detailed in the Appendix S1.

Between June 27, 2014, and April 29, 2022, 21 patients with confirmed ICI-AIHA were included in analysis (Figure 1). Among 2775 patients prospectively included in the REISAMIC registry of Gustave Roussy Centre, two patients have developed ICI-AIHA, and the rate of ICI-AIHA was 0.07% (two out of 2775 ICI-treated patients). Among the 21 patients diagnosed with ICI-AIHA; median age at onset of ICI-AIHA was 70 years (ranging from 29 to 88 years); sex ratio was 2.0 (14 men and 7 women); and tumor types were non-small cell lung carcinoma (n = 8; 38%), melanoma (n = 7; 33%), renal cancer (n = 3; 14%), Hodgkin lymphoma (n = 1; 5%), and other tumor types (n = 2; 10%). Patients with ICI-AIHA mostly received anti-PD-1 immunotherapy (n = 17; 81% of cases) and developed ICI-AIHA after a median time of 28 days (ranging from 11 to 1060 days) (Table S1).

A medical history of lymphoproliferative disorder was found in 9/21 patients (43%), and a pre-existing autoimmune or inflammatory disease was known for three patients (14%). The lymphoproliferative disorders were chronic lymphocytic leukemia (n = 3 patients), marginal zone lymphoma (n = 2 patients), Hodgkin lymphoma (n = 2 patients), Waldenstrom disease (n = 1 patient), and monoclonal gammopathy of undetermined significance (n = 1 patient) (Table S1). Among these nine patients with lymphoproliferative disorder, only one patient had a previously known lymphoproliferative disorder (patient with Hodgkin lymphoma, treated with anti-PD1), the other eight patients had no known history of lymphoproliferative disorder, and the diagnosis of lymphoproliferative disorder was performed and revealed at the same time of ICI-AIHA.

Nadir of hemoglobin was 6.9 g/dL (range 4.2–9.6 g/dL), and 14 patients (71%) required red blood cell transfusions for anemia management, with a median of four packed red cell units per patient (range 1–8). Mortality related to ICI-AIHA was observed in two (10%) patients (Table S2). According to common terminology criteria for adverse events (CTCAEV5), severity of anemia was grade ≥3 in 18 (86%) patients, and the severity of anemia was observed and distributed according to each ICI-AIHA subtype (Figure 1).

All patients had direct antiglobulin test (DAT) performed, and 19 (90%) had a positive test, with an IgG pattern for 11 (52%) and qualified as warm ICI-AIHA, five (24%) were positive for IgM and qualified as cold ICI-AIHA, and three (14%) had a mixed profile and qualified as mixed ICI-AIHA. Two patients (10%) had a negative DAT (Table S1).

Fourteen (66%) patients received corticosteroids alone, and six (29%) patients received corticosteroids and rituximab (Tables S2 and S3). One patient received only palliative therapies without specific treatment for ICI-AIHA and died due to cancer progression before receiving any specific treatment for ICI-AIHA. Overall complete response to corticosteroids was 36% (5/14 patients treated) (Table S2). The efficacy of corticosteroids was observed only in the warm-AIHA subtype (details of therapeutic responses by subtypes of ICI-AIHA are available in Tables S2 and S4). Six (29%) patients received rituximab for ICI-AIHA. The response rate to rituximab was 50% (3/6 patients treated) (Table S2). The efficacy of rituximab was 1/3 and 2/2 of treated patients in the warm-AIHA and cold-AIHA subtypes, respectively.

Remission of ICI-AIHA was achieved in 19 of the 21 patients (90%), and treatment outcomes were complete response in eight (38%) patients and partial response in 11 (52%) patients, with similar response rates among patients treated with corticosteroids or with corticosteroids plus rituximab (Table S2). Evolution of hemoglobin after ICI-AIHA treatments according to corticosteroid or corticosteroid plus rituximab was similar as depicted in Figure 1. Although hemoglobin concentration at ICI-AIHA diagnosis timepoints tended to be lower in patients treated with corticosteroids plus rituximab as compared to corticosteroids alone (Hb 5.8 g/dL vs. 7.3 g/dL; p = .10; Table S3), the hemoglobin recovery at 6 months was similar (Figure 1).

After ICI-AIHA diagnosis, immune checkpoint inhibitor therapy was permanently discontinued in 19 patients (90%). Two (9.5%) of the 21 patients were rechallenged with same ICI after obtaining durable complete remission of ICI-AIHA (Table S2). The two rechallenged patients were treated with corticosteroids alone, and in these two patients, no recurrence of ICI-AIHA was observed with a period after ICI rechallenge of 24 months.

Finally, our study reports a case series of 21 patients diagnosed with ICI-AIHA delineating the subtypes of ICI-AIHA that appears to be mainly warm-AIHA (52%) and cold-AIHA (24%). DAT-negative ICI-AIHA was reported in 10% of cases, which seems to confirm that ICI-AIHA could be more frequently associated with negative DAT (DAT-negative AIHA is thought to be present in <5% of patients in the setting of AIHA outside of ICI¹). Leaf et al.² reported in the ICI-AIHA case series from the USA a negative DAT rate of 28%.

Our study was able to estimate that ICI-AIHA is a rare complication associated with immunotherapy with an incidence rate of 0.07%. The mortality associated with ICI-AIHA was 10% of patients in our study, which indicates that ICI-AIHA was a severe and potential life-threatening complication of immunotherapy.

In all patients in the study, corticosteroids or corticosteroids plus rituximab produced similar results on hemoglobin recovery, with remission of ICI-AIHA in 90% of patients. Overall, our results indicate that ICI-AIHA treatment should consider the AIHA subtypes; warm ICI-AIHA is to be treated with corticosteroids, with addition of rituximab in cases of insufficient response to corticosteroids. In patients with cold ICI-AIHA, corticosteroids do not seem to be effective, and the treatment should be based on rituximab.

In our study, ICI-AIHA was associated with an underlying lymphoproliferative disorder in 43% of patients. Leaf et al.,² in another report of ICI-AIHA, found an associated lymphoproliferative syndrome (most often chronic lymphocytic leukemia or indolent lymphoma) in 29% of their patients. We could discuss the causal factor of AIHA between the immunotherapy treatment or the natural evolution of the underlying lymphoproliferative syndrome, or the conjunction of the two.

In conclusion, main ICI-AIHA subtypes included both warm-AIHA and cold-AIHA. An associated underlying lymphoproliferative disorder was frequently observed and should be recommended to look for. Our study suggests that ICI-AIHA treatment should consider AIHA subtypes: corticosteroids ± rituximab for warm-AIHA and rituximab alone for cold-AIHA.

自身免疫性溶血性贫血（AIHA）被定义为自身免疫机制对红细胞的破坏加剧，通常由针对红细胞表面抗原的自身抗体介导。¹药物引起的溶血性贫血是 AIHA 的独特病因，通常由免疫机制介导。¹用于治疗癌症的免疫检查点抑制剂 (ICIs)，特别是 PD-1 抑制剂，可诱发 AIHA。^{1, 2}

ICI 引发的免疫相关不良事件偶尔会涉及造血系统和红细胞，估计治疗患者的发生频率约为 0.5%。³ ICI-AIHA 仅在病例系列中的少数患者中报告，^2、3并且有关 AIHA 亚型分布（热与冷）和患者结局（包括与 AIHA 相关的死亡率和治疗效果）的数据仍然很少。 ICI-AIHA 患者的管理基于皮质类固醇和最终的利妥昔单抗，^{2, 3}，安全性和有效性数据尚未完全报告。我们在此报告在三级学术抗癌中心接受 ICI-AIHA 治疗的患者的管理和治疗结果。

该研究设计是一项针对法国 ICI-AIHA 成年患者的观察性回顾性研究。该研究考虑的免疫检查点抑制剂包括抗程序性死亡细胞 1 (PD-1)、抗程序性死亡细胞配体 1 (PD-L1) 和抗 CTLA4。资格标准是年龄≥18岁、患有ICI-AIHA、在药物警戒登记处REISAMIC（癌症免疫调节单克隆抗体严重不良反应登记处）⁴登记的患者，以及自6月以来转介至Gustave Roussy免疫毒性委员会^5的患者2014年2月27日至2022年4月29日，以及2014年至2022年属于法国成人自身免疫性血细胞减少症参考中心（CeReCAI）网络的中心（图1）。本研究的 AIHA 标准和定义遵循 2020 年首个 AIHA 管理国际共识标准。⁶研究目标、数据收集信息、统计分析和伦理考虑详见附录 S1。

2014年6月27日至2022年4月29日期间，21名确诊为ICI-AIHA的患者被纳入分析（图1）。在 Gustave Roussy 中心 REISAMIC 登记的 2775 名患者中，有 2 名患者发展为 ICI-AIHA，ICI-AIHA 发生率为 0.07%（2775 名接受 ICI 治疗的患者中有 2 名）。诊断为ICI-AIHA的21名患者中； ICI-AIHA 发病的中位年龄为 70 岁（范围为 29 至 88 岁）；性别比为2.0（男性14名，女性7名）；肿瘤类型为非小细胞肺癌 ( n  = 8; 38%)、黑色素瘤 ( n  = 7; 33%)、肾癌 ( n  = 3; 14%)、霍奇金淋巴瘤 ( n  = 1; 5%)和其他肿瘤类型 ( n  = 2; 10%)。 ICI-AIHA 患者大多接受抗 PD-1 免疫治疗（n  = 17；占病例的 81%），并在中位时间 28 天（范围从 11 至 1060 天）后出现 ICI-AIHA（表 S1）。

9/21 名患者 (43%) 发现有淋巴组织增生性疾病病史，3 名患者 (14%) 已知患有自身免疫性疾病或炎症性疾病。淋巴细胞增殖性疾病为慢性淋巴细胞白血病（n  = 3 名患者）、边缘区淋巴瘤（n  = 2 名患者）、霍奇金淋巴瘤（n  = 2 名患者）、瓦尔登斯特伦病（n  = 1 名患者）和意义未明的单克隆丙种球蛋白病（n = 1 名患者）  = 1 名患者）（表 S1）。在这9名淋巴组织增生性疾病患者中，只有1名患者既往已知淋巴组织增生性疾病（霍奇金淋巴瘤患者，接受抗PD1治疗），其他8名患者没有已知的淋巴组织增生性疾病病史，诊断为淋巴组织增生性疾病并与ICI-AIHA同时揭晓。

血红蛋白最低值为 6.9 g/dL（范围 4.2–9.6 g/dL），14 名患者 (71%) 需要输注红细胞来治疗贫血，每位患者中位数为 4 个浓缩红细胞单位（范围 1–8 ）。在两名 (10%) 患者中观察到与 ICI-AIHA 相关的死亡率（表 S2）。根据不良事件通用术语标准（CTCAEV5），18例（86%）患者贫血严重程度≥3级，并根据每种ICI-AIHA亚型观察和分布贫血严重程度（图1）。

所有患者均进行了直接抗球蛋白试验 (DAT)，其中 19 例 (90%) 检测结果呈阳性，其中 11 例 (52%) 呈 IgG 模式，符合温 ICI-AIHA 条件，5 例 (24%) 呈 IgM 阳性，符合冷 ICI-AIHA 资格，其中 3 名 (14%) 具有混合特征，符合混合 ICI-AIHA 资格。两名患者 (10%) 的 DAT 呈阴性（表 S1）。

14 名 (66%) 患者仅接受皮质类固醇治疗，6 名 (29%) 患者接受皮质类固醇和利妥昔单抗治疗（表 S2 和 S3）。一名患者仅接受姑息治疗，没有针对 ICI-AIHA 进行特异性治疗，并在接受任何 ICI-AIHA 特异性治疗之前因癌症进展而死亡。对皮质类固醇的总体完全缓解率为 36%（5/14 接受治疗的患者）（表 S2）。仅在温热 AIHA 亚型中观察到皮质类固醇的功效（ICI-AIHA 亚型的治疗反应详情见表 S2 和 S4）。六名 (29%) 患者因 ICI-AIHA 接受利妥昔单抗治疗。对利妥昔单抗的反应率为 50%（3/6 的患者接受治疗）（表 S2）。在热 AIHA 和冷 AIHA 亚型中，利妥昔单抗的疗效分别是治疗患者的 1/3 和 2/2。

21 名患者中有 19 名 (90%) 实现了 ICI-AIHA 缓解，治疗结果为 8 名患者 (38%) 完全缓解，11 名患者 (52%) 部分缓解，接受治疗的患者的缓解率相似。皮质类固醇或皮质类固醇加利妥昔单抗（表S2）。根据皮质类固醇或皮质类固醇加利妥昔单抗进行 ICI-AIHA 治疗后血红蛋白的演变与图 1 所示相似。尽管与单独使用皮质类固醇相比，接受皮质类固醇加利妥昔单抗治疗的患者在 ICI-AIHA 诊断时间点的血红蛋白浓度往往较低（Hb） 5.8 g/dL 与 7.3 g/dL；p  = .10；表 S3)，6 个月时的血红蛋白恢复情况相似（图 1）。

ICI-AIHA 诊断后，19 名患者 (90%) 永久停止免疫检查点抑制剂治疗。 21 名患者中有 2 名 (9.5%) 在获得 ICI-AIHA 持久完全缓解后再次接受相同的 ICI 治疗（表 S2）。两名再激发患者仅接受皮质类固醇治疗，在这两名患者中，在 ICI 再激发后 24 个月内未观察到 ICI-AIHA 复发。

最后，我们的研究报告了由 21 名诊断为 ICI-AIHA 的患者组成的病例系列，描述了 ICI-AIHA 的亚型，主要是热 AIHA (52%) 和冷 AIHA (24%)。据报道，10% 的病例呈 DAT 阴性 ICI-AIHA，这似乎证实了 ICI-AIHA 可能更频繁地与 DAT 阴性相关（在以下情况下，DAT 阴性 AIHA 被认为存在于 <5% 的患者中） ICI 之外的 AIHA ¹ )。叶等人。²在美国 ICI-AIHA 病例系列中报道，DAT 阴性率为 28%。

我们的研究能够估计 ICI-AIHA 是一种与免疫治疗相关的罕见并发症，发病率为 0.07%。在我们的研究中，与 ICI-AIHA 相关的死亡率为 10%，这表明 ICI-AIHA 是免疫治疗的一种严重且潜在危及生命的并发症。

在研究中的所有患者中，皮质类固醇或皮质类固醇加利妥昔单抗对血红蛋白恢复产生相似的结果，90% 的患者 ICI-AIHA 得到缓解。总体而言，我们的结果表明 ICI-AIHA 治疗应考虑 AIHA 亚型；温热的 ICI-AIHA 需用皮质类固醇治疗，如果对皮质类固醇反应不足，则加用利妥昔单抗。对于感冒ICI-AIHA患者，皮质类固醇似乎无效，治疗应以利妥昔单抗为主。

在我们的研究中，43% 的患者 ICI-AIHA 与潜在的淋巴增殖性疾病相关。 Leaf 等人²在 ICI-AIHA 的另一份报告中发现，29% 的患者患有相关的淋巴细胞增殖综合征（最常见的是慢性淋巴细胞白血病或惰性淋巴瘤）。我们可以讨论免疫治疗或潜在淋巴增殖综合征的自然演变之间的 AIHA 致病因素，或两者的结合。

总之，主要的ICI-AIHA亚型包括暖型AIHA和冷型AIHA。经常观察到相关的潜在淋巴增殖性疾病，应建议进行检查。我们的研究表明，ICI-AIHA 治疗应考虑 AIHA 亚型：皮质类固醇 ± 利妥昔单抗用于热 AIHA，单独利妥昔单抗用于冷 AIHA。