To the editor:

Hematologic malignancies encompass a group of heterogenous diseases with variable effects on immune function, and the degree of immune dysfunction may be further exacerbated by therapies used to treat specific diseases. Emerging real-world data suggest that patients with hematologic malignancy, particularly B-cell malignancies and those receiving B-cell depleting therapies, likely do not elicit robust immunologic responses following COVID-19 vaccination; however, existing studies include modest sample sizes and uncertainty remains. Thus, we performed a systematic review and meta-analysis aggregating data on anti-SARS-CoV-2 IgG antibody seroresponse (SR) to COVID-19 vaccination in patients with hematologic malignancy.

PubMed and EMBASE were searched from January 1, 2021 to November 4, 2021, to identify studies of vaccine immunogenicity following COVID-19 vaccination in patients with hematologic malignancy (Supplementary Methods; Tables S1 and S2; Figure S1). The primary outcomes were pooled SR in all studies, and pooled relative benefit ratio (RB) compared to controls in studies with a comparator group. Secondary outcomes were pooled SR and pooled RB by hematologic malignancy subtype and treatment status and type. Pooled estimates, RBs, along with 95% confidence intervals (CIs) were calculated using a random-effects model using MetaXL and Review Manager 5.4. Further details on data extraction and synthesis are provided in the Supplementary Methods.

We identified 2205 unique publications, of which 64 studies met inclusion criteria, comprising 8546 adult patients with hematological malignancy (Table S3). Full results are provided in Supplementary Results. Figure 1 provides a visual depiction of pooled SR and RB for outcomes. Pooled SR of all included patients was 59% (95% CI 55%–64%, with considerable heterogeneity I² 95%; Figure S2). RB when compared to controls (either health care workers, healthy volunteers, or age-matched cancer-free controls) in available reports was 0.61 (95% CI 0.55–0.66, I² 91%; Figure S3). SR varied according to hematologic malignancy subtype with better responses seen in myeloid malignancies (SR for myeloproliferative neoplasms [MPN] 81%, 95% CI 72%–89%; SR for myelodysplastic syndrome [MDS] 63%, 95% CI 47%–78%; SR for acute leukemias 83%, 95% CI 77%–89%), and lower responses observed in lymphoid malignancies (SR for chronic lymphocytic leukemia [CLL] 44%, 95% CI 35%–53%; SR for lymphoid malignancies excluding CLL 52%, 95% CI 44%–61%; SR for plasma cell dyscrasias [PCD] 72%, 95% CI 64%–79%) (Figures S9-S14). Patients with history of stem cell transplant (SCT) had good SR of 79% (95% CI 75–82), irrespective of allogeneic (78%) or autologous (88%) SCT (Figures S15). In contrast, patients with a history of chimeric antigen receptor T-cell (CAR-T) therapy had poor SR of 33% (95% CI 18–49) (Figures S16). RB from studies with comparators showed similar findings (Figures S17–S19).

Serologic responses were abrogated by cancer treatment, with SR in patients receiving treatment 47% (95% CI 36%–58%) compared to untreated SR 83% (95% CI 75%–90%), and RB of 0.38 (95% CI 0.25–0.58; Figure S4). This was particularly notable for prior anti-CD20 therapy with RB 0.16 (95% CI 0.08–0.30) when comparing receipt of anti-CD20 therapy <9–12 months of vaccination to >9–12 months after vaccination (Table S4 and Figure S5). Similarly, poor SR was observed in patients receiving novel targeted therapies Bruton tyrosine kinase inhibitor (BTKi) or venetoclax (37%, 95% CI 22%–54%) and anti-CD38 therapy (48%, 95% CI 27%–68%) (Figures S20 and S21).

There were 10 studies reporting on T-cell responses and conflicting results were observed. While some studies demonstrated concordance between reduced T-cell response and low antibody response, particularly in anti-CD20 treatment patients, others demonstrated the presence of T-cell responses in patients without antibody response (Table S5).

We identified three studies reporting on SR following booster (third) dose in hematologic malignancy patients. These studies demonstrated that a booster dose could achieve SR up to 55% of patients who were seronegative following initial vaccination series.¹

Overall, our data support the rapidly emerging evidence demonstrating impaired humoral response following vaccination in hematologic malignancy patients. Patients with B-cell malignancies, particularly CLL and lymphoma, had the lowest SR. This likely reflects the disease-specific biology which causes underlying immune dysfunction in many patients as well as therapies used to treat lymphoid malignancies, including B-cell depleting therapies such as anti-CD20 and BTK inhibitors. Indeed, patients receiving treatment with B-cell depleting therapy had markedly impaired antibody responses compared to noncancer controls and disease patients, with reported SR ranging from 0% to 25% for those who received anti-CD20 within 3 months of vaccination. Response rates following COVID-19 vaccination improved with the passage of time, with higher serologic responses observed in those who received vaccination more than 9–12 months following anti-CD20 therapy. These findings are in keeping with prior studies suggesting that B-cell reconstitution following anti-CD20 antibody treatment requires 9–12 months.² Similarly, studies reporting on oral B-cell depleting therapies also observed impaired antibody response, with reported SR ranging from 14% to 57%. Importantly, seroconversion from seronegative to seropositive following booster vaccinations was evaluated in several studies, support the role of this strategy in achieving seroconversion. Furthermore, a study evaluating the kinetics of antibody titers demonstrated a rapid decline in titers from 36 days onward, and resulted in conversion from seropositive to seronegative in patients with hematologic malignancy while SR was conserved in patients with solid malignancies.³ Collectively, these data support the use of booster doses to achieve optimal serologic response in patients with hematologic malignancy.

Although PCD can also suppress the immune system and affect normal B-cell function, patients with PCD had higher responses compared to lymphoma and CLL. However, studies showed conflicting results in terms of effect of antimyeloma therapy on serologic response. Several reports observed impaired antibody response with anti-CD38 therapy, anti-B-cell maturation antigen therapy, along with number of lines of therapy, while other studies did not demonstrate a significant difference in SR when comparing treated versus untreated patients. We postulate that these conflicting results may be due to difficulty discerning specific treatment regimen effects on serological response in MM patients given that anti-myeloma therapy are usually given in combination incorporating several drug classes (immunomodulators, proteasome inhibitors, alkylating agents, steroids, and anti-CD38 therapy).

Patients with myeloid malignancies and their associated treatment such as tyrosine kinase inhibitors, did not have blunted SR. SCT recipients generally attained moderate to higher SR ranging from 50% to 89%, although studies suggest reduced SR within 1 year of transplantation (SR ranged between 20%–54% within 1 year of SCT vs. 80%–91% ≥1 year of SCT as reported from two studies included in this review). Albeit limited by sample size, CAR-T therapy was associated with very poor serologic response. For these patients, ASH and the American Society of Transplantation and Cellular Therapy advised that COVID-19 vaccines should be offered to patients 3 months or later following SCT and CART-T therapy.

There are several limitations to consider when interpreting results of this study. We observed significant heterogeneity in reported outcomes. This is likely due to several disease- and treatment-related factors including heterogeneous disease biology impacting on humoral and cellular immune system, disease status, and type of therapy received, particularly B-cell depleting therapies. To explore the heterogeneity, we conducted subgroup analysis based on hematologic malignancy subtype; heterogeneity was reduced for acute leukemia and SCT but remained high for other subtypes. Similarly, when analyzing data based on treatment status, heterogeneity only slightly reduced. As such, our pooled SR estimates should be interpreted with caution and highlight the need for larger and more robust studies. Another limitation is that most studies included in this systematic review measured SR by using anti-SARS-CoV-2 spike protein IgG, with only a small number of studies measuring neutralizing antibody response. Although neutralizing antibody response is the gold standard for humoral response, with higher levels inferring protection, recent studies demonstrate a high degree of correlation between neutralizing antibody titers and IgG antibodies in both convalescent and vaccinated individuals.⁴ Furthermore, among fully vaccinated healthcare workers, breakthrough infections correlate with lower levels of both anti-spike IgG antibodies and neutralizing antibodies, compared to matched uninfected controls,⁵ supporting the importance of serologic response in protective immunity against COVID-19. As outlined above, studies correlating humoral responses with T-cell responses showed conflicting results with some studies demonstrating concordance (double negativity) while others demonstrated presence of T-cell responses in patients without humoral response. As such, additional studies are needed to evaluate the relative importance of antibody and cellular responses to COVID-19 infection protection, and whether T-cell responses are sufficient to decrease severity of COVID-19 disease in those without humoral response.

In summary, in this meta-analysis aggregating SR following COVID-19 vaccination in patients with hematologic malignancy, the lowest response was observed in lymphoid malignancies, particularly those treated with anti-CD20 therapy, and other B-cell depleting therapies. Emerging data correlating neutralizing antibody response to anti-Sars-CoV-2 antibody levels and infection risk suggest that hematologic malignancy patients without adequate antibody levels remain at high risk of COVID-19 infection. Additional studies are urgently needed to determine whether immunologic response can be improved with tailored dosing and booster vaccination doses. Furthermore, therapies such as anti-COVID-19 monoclonal antibodies and convalescent serum should be evaluated in hematologic malignancy patients as prophylactic and treatment modalities, particularly for those unable to mount an immunologic response to vaccination.⁶

致编辑：

血液系统恶性肿瘤包括一组对免疫功能有不同影响的异质性疾病，免疫功能障碍的程度可能会因用于治疗特定疾病的疗法而进一步恶化。新出现的真实世界数据表明，患有血液系统恶性肿瘤的患者，尤其是 B 细胞恶性肿瘤和接受 B 细胞耗竭疗法的患者，在接种 COVID-19 疫苗后可能不会引发强烈的免疫反应；然而，现有研究包括适度的样本量和不确定性。因此，我们进行了系统回顾和荟萃分析，汇总了血液系统恶性肿瘤患者对 COVID-19 疫苗接种的抗 SARS-CoV-2 IgG 抗体血清反应 (SR) 的数据。

在 2021 年 1 月 1 日至 2021 年 11 月 4 日期间对 PubMed 和 EMBASE 进行了搜索，以确定血液系统恶性肿瘤患者接种 COVID-19 疫苗后的疫苗免疫原性研究（补充方法；表 S1 和 S2；图 S1）。主要结果是所有研究中的汇总 SR，以及与比较组研究中的对照组相比的汇总相对效益比 (RB)。次要结果是按血液恶性肿瘤亚型和治疗状态和类型汇总的 SR 和汇总 RB。使用 MetaXL 和 Review Manager 5.4 使用随机效应模型计算汇总估计值、RB 以及 95% 置信区间 (CI)。补充方法中提供了有关数据提取和合成的更多详细信息。

我们确定了 2205 篇独特的出版物，其中 64 项研究符合纳入标准，包括 8546 名患有血液系统恶性肿瘤的成年患者（表 S3）。完整结果在补充结果中提供。图 1 提供了结果汇总 SR 和 RB 的可视化描述。所有纳入患者的汇总 SR 为 59%（95% CI 55%–64%，具有相当大的异质性I ² 95%；图 S2）。与现有报告中的对照组（医护人员、健康志愿者或年龄匹配的无癌症对照组）相比，RB 为 0.61（95% CI 0.55–0.66，I ²91%；图 S3)。SR 因血液系统恶性肿瘤亚型而异，在髓系恶性肿瘤中反应更好（骨髓增生性肿瘤 [MPN] SR 81%、95% CI 72%–89%；骨髓增生异常综合征 [MDS] SR 63%、95% CI 47%– 78%；急性白血病的 SR 83%，95% CI 77%–89%），并且在淋巴恶性肿瘤中观察到较低的反应（慢性淋巴细胞白血病 [CLL] 的 SR 44%，95% CI 35%–53%；SR 用于淋巴恶性肿瘤，不包括 CLL 52%，95% CI 44%–61%；浆细胞恶液质 [PCD] 的 SR 72%，95% CI 64%–79%）（图 S9-S14）。有干细胞移植 (SCT) 病史的患者的 SR 为 79% (95% CI 75-82)，无论是同种异体 (78%) 还是自体 (88%) SCT（图 S15）。相比之下，有嵌合抗原受体 T 细胞 (CAR-T) 治疗史的患者 SR 较差，为 33% (95% CI 18-49)（图 S16）。

血清学反应被癌症治疗消除，接受治疗的患者的 SR 为 47%（95% CI 36%–58%），而未治疗的 SR 为 83%（95% CI 75%–90%），RB 为 0.38（95% CI 0.25–0.58；图 S4)。当比较接受抗 CD20 治疗<9-12 个月的疫苗接种与>9-12 个月的疫苗接种后（表 S4 和图 S5 ）。同样，在接受新型靶向治疗 Bruton 酪氨酸激酶抑制剂 (BTKi) 或 venetoclax (37%, 95% CI 22%–54%) 和抗 CD38 治疗 (48%, 95% CI 27%–68) 的患者中观察到较差的 SR %)（图 S20 和 S21）。

有 10 项研究报告了 T 细胞反应，但观察到了相互矛盾的结果。虽然一些研究表明 T 细胞反应降低和抗体反应低之间存在一致性，特别是在抗 CD20 治疗患者中，但其他研究表明在没有抗体反应的患者中存在 T 细胞反应（表 S5）。

我们确定了三项研究报告了血液系统恶性肿瘤患者加强（第三次）剂量后的 SR。这些研究表明，在初始疫苗系列接种后血清阴性的患者中，加强剂量可以达到高达 55% 的 SR。¹

总体而言，我们的数据支持迅速出现的证据，证明血液系统恶性肿瘤患者接种疫苗后体液反应受损。B 细胞恶性肿瘤患者，尤其是 CLL 和淋巴瘤患者的 SR 最低。这可能反映了导致许多患者潜在免疫功能障碍的疾病特异性生物学以及用于治疗淋巴恶性肿瘤的疗法，包括 B 细胞消耗疗法，例如抗 CD20 和 BTK 抑制剂。事实上，与非癌症对照和疾病患者相比，接受 B 细胞耗竭疗法治疗的患者的抗体反应明显受损，据报道，在接种疫苗后 3 个月内接受抗 CD20 治疗的患者的 SR 范围为 0% 至 25%。COVID-19 疫苗接种后的反应率随着时间的推移而提高，在抗 CD20 治疗后 9-12 个月以上接受疫苗接种的人中观察到更高的血清学反应。这些发现与先前的研究一致，这些研究表明抗 CD20 抗体治疗后的 B 细胞重建需要 9-12 个月。²同样，关于口服 B 细胞消耗疗法的研究也观察到抗体反应受损，报告的 SR 范围为 14% 至 57%。重要的是，几项研究评估了加强疫苗接种后从血清阴性到血清阳性的血清转化，支持该策略在实现血清转化中的作用。此外，一项评估抗体滴度动力学的研究表明，从 36 天起，滴度迅速下降，导致血液系统恶性肿瘤患者从血清阳性转为血清阴性，而实体恶性肿瘤患者 SR 保持不变。³总的来说，这些数据支持使用加强剂量来实现血液系统恶性肿瘤患者的最佳血清学反应。

虽然 PCD 也可以抑制免疫系统并影响正常的 B 细胞功能，但与淋巴瘤和 CLL 相比，PCD 患者的反应更高。然而，就抗骨髓瘤治疗对血清学反应的影响而言，研究显示出相互矛盾的结果。几份报告观察到抗 CD38 治疗、抗 B 细胞成熟抗原治疗以及治疗线数量受损的抗体反应，而其他研究在比较治疗与未治疗患者时未显示 SR 的显着差异。我们假设这些相互矛盾的结果可能是由于难以辨别特定治疗方案对 MM 患者血清学反应的影响，因为抗骨髓瘤治疗通常结合多种药物类别（免疫调节剂、蛋白酶体抑制剂、烷化剂、

髓系恶性肿瘤患者及其相关治疗（如酪氨酸激酶抑制剂）没有减弱 SR。尽管研究表明移植后 1 年内 SR 降低（SCT 1 年内 SR 范围在 20%–54% 与 80%–91% ≥1 年），但 SCT 受者通常达到中度至更高的 SR本综述中包含的两项研究报告的 SCT）。尽管受样本量的限制，CAR-T 疗法与非常差的血清学反应相关。对于这些患者，ASH 和美国移植和细胞治疗学会建议，应在 SCT 和 CART-T 治疗后 3 个月或更晚向患者提供 COVID-19 疫苗。

在解释本研究的结果时需要考虑几个限制。我们观察到报告结果的显着异质性。这可能是由于多种疾病和治疗相关因素，包括影响体液和细胞免疫系统的异质疾病生物学、疾病状态和接受的治疗类型，特别是 B 细胞耗竭疗法。为了探索异质性，我们根据血液系统恶性肿瘤亚型进行了亚组分析；急性白血病和 SCT 的异质性降低，但其他亚型的异质性仍然很高。同样，在根据治疗状态分析数据时，异质性仅略有降低。因此，我们的汇总 SR 估计值应谨慎解释，并强调需要进行更大规模、更稳健的研究。另一个限制是，本系统评价中包含的大多数研究都是通过使用抗 SARS-CoV-2 刺突蛋白 IgG 来测量 SR，只有少数研究测量了中和抗体反应。尽管中和抗体反应是体液反应的金标准，具有更高水平的推断保护作用，但最近的研究表明，在恢复期和接种疫苗的个体中，中和抗体滴度与 IgG 抗体之间存在高度相关性。⁴此外，在完全接种疫苗的医护人员中，与匹配的未感染对照相比，突破性感染与较低水平的抗尖峰 IgG 抗体和中和抗体相关， ⁵支持血清学反应在针对 COVID-19 的保护性免疫中的重要性。如上所述，将体液反应与 T 细胞反应相关的研究显示出相互矛盾的结果，一些研究证明了一致性（双重否定性），而另一些研究表明在没有体液反应的患者中存在 T 细胞反应。因此，需要更多的研究来评估抗体和细胞反应对 COVID-19 感染保护的相对重要性，以及 T 细胞反应是否足以降低没有体液反应的 COVID-19 疾病的严重程度。

总之，在这项汇总血液系统恶性肿瘤患者 COVID-19 疫苗接种后 SR 的荟萃分析中，在淋巴恶性肿瘤中观察到最低反应，特别是那些接受抗 CD20 治疗和其他 B 细胞消耗疗法治疗的患者。将中和抗体对抗 Sars-CoV-2 抗体水平和感染风险的反应相关的新数据表明，没有足够抗体水平的血液系统恶性肿瘤患者仍然处于 COVID-19 感染的高风险中。迫切需要进一步的研究来确定是否可以通过定制剂量和加强疫苗接种来改善免疫反应。此外，应在血液系统恶性肿瘤患者中评估抗 COVID-19 单克隆抗体和恢复期血清等疗法作为预防和治疗方式，⁶