Yang Liu

← Thinking

定向清髓距离替代 busulfan 还有多远 How Far Is Targeted Conditioning from Replacing Busulfan?

2026-06-28 · hsc, gene-therapy, conditioning, hemoglobinopathy, radiotherapy

儿童 exa-cel 给了血液基因治疗一个很清楚的信号: 编辑细胞本身已经可以产生接近“教科书式”的疗效, 但治疗仍然要经过移植医学里最沉重的步骤之一: myeloablative conditioning。问题不是 busulfan 有没有用, 而是如果一个疗法必须先让儿童承受全清髓、严重血液学毒性、VOD/SOS 风险、住院和生育风险, 它就很难从“高端中心里的治愈性疗法”变成更广泛可及的疾病改写工具。

所以, “定向清髓能不能取代 busulfan?” 这个问题现在不能回答成“能”或“不能”。更准确的判断是: 近期 CD45 抗体加 α 粒子的非人灵长类数据, 已经把“血液系统靶向打开 HSC 生态位并支持编辑细胞长期植入”推到一个更接近转化的问题层面; 但它距离儿童 exa-cel 这类临床场景还差三道门: 人体剂量窗、儿童毒性窗、以及放射药供应链和治疗中心可及性。

编辑已经有效, 清髓还在收费

2026 年 6 月, Haydar Frangoul、Franco Locatelli 与 CLIMB THAL-141 / CLIMB SCD-151 研究组在 New England Journal of Medicine 发表了 5-11 岁儿童 exa-cel 的最新临床读数。这个近期读数有两层完全不同的含义。

第一层是疗效。研究纳入 5-11 岁儿童, 15 名输血依赖型 β 地中海贫血儿童和 11 名镰状细胞病儿童接受 exa-cel。可评估到至少 16 个月的 TDT 儿童中, 8/8 实现至少 12 个月输血独立; 可评估的 SCD 儿童中, 8/8 实现至少 12 个月无重度 vaso-occlusive crisis。对 BCL11A erythroid enhancer 的 ex vivo CRISPR-Cas9 编辑, 在这个读数里继续证明了“提高 HbF 可以改写血红蛋白病表型”。

第二层是代价。所有接受治疗的儿童都发生至少一个 3/4 级不良事件。TDT 队列里有 2 例严重 hepatic veno-occlusive disease / sinusoidal obstruction syndrome, 均被评估为与 busulfan 预处理相关, 其中 1 例死亡。

这不是 exa-cel 独有的问题。CASGEVY 的美国标签本身就把“full myeloablative conditioning”放在治疗流程里; 标签中的成人和青少年数据也显示, SCD 和 TDT 患者在 busulfan 预处理和 CASGEVY 输注后, 100% 出现 3/4 级中性粒细胞减少和血小板减少。TDT 队列里 VOD 不是罕见到可以忽略的噪音, 而是 52 名受试者中 5 例的标签级安全信号。标签还明确提醒, conditioning 之后患者可能无法怀孕或生育, 需要提前讨论 fertility preservation。

这说明一个基本事实: 对血液系统基因治疗来说, 编辑效率、HbF 上升和临床缓解只是疗法的一半。另一半是为了让这些细胞住进骨髓, 现有产品仍要借用传统移植时代留下来的清髓逻辑。

busulfan 不是配角, 而是疗法的边界

busulfan 的价值很直接: 它能清空骨髓生态位, 让回输的自体编辑 HSPC 有足够机会长期植入。对于 hemoglobinopathy 这种需要持续产生功能性红细胞的疾病, 短期表达不够, 部分外周血细胞改变也不够; 真正需要的是 HSC 层面的 durable engraftment。

这也是为什么“降低清髓强度”不是一个简单的伦理愿望。清髓少了, 毒性可能下降, 但编辑细胞植入比例可能不够; 植入比例不够, HbF、正常血红蛋白或纠正后红细胞输出可能低于临床阈值。疗法的风险不是线性减少, 而是可能从急性毒性转移为疗效不足、复发输血、再次治疗或救援移植。

所以 busulfan 当前不是一个容易被拿掉的“药物配角”。它是 ex vivo HSC 基因治疗产品的隐性基础设施: 产品制造、住院流程、中心资质、保险支付、患者筛选、长期随访, 都围绕它展开。清髓方案一旦改变, 改变的不是处方里的一行药名, 而是整个产品的风险结构。

CD45-α 粒子真正替代的是什么

2026 年 6 月, Fred Hutch 的 Stefan Radtke、Hans-Peter Kiem、Roland Walter 等人在 Blood 在线发表了 [²¹¹At]astatine 标记人源化 CD45 抗体用于自体 HSPC 基因治疗预处理的非人灵长类研究。这篇近期论文的重要性在于, 它没有只展示“杀掉一些造血细胞”, 而是把问题推进到自体 HSPC 基因治疗最关心的层面: 编辑 HSPC 能不能长期、多谱系、稳定地重建血液系统。

这项研究把人源化 CD45 抗体 HuBC8 与 α 发射核素 astatine-211 标记, 用作 CD45-directed radioimmunotherapy。模型上, 研究者先对动员得到的 CD34+ HSPC 做 multiplex gene editing: 用 adenine base editor 改 HBG promoter 以重新激活 fetal hemoglobin, 同时删除 CD33。随后, 非人灵长类接受 300 或 400 μCi/kg 的 ²¹¹At-CD45-RIT, 再回输编辑细胞。

几个结果值得认真看:

第一, 植入呈剂量依赖。单细胞测序显示, 外周血中 mono- 和 bi-allelic 编辑细胞合计最高可到约 70%, 与骨髓 HSC compartment 被充分替换相符。

第二, 持续性超过 18 个月。对 HSC 基因治疗来说, 这比短期外周血读数更有价值, 因为它指向长期造血干细胞层面的贡献。

第三, 谱系贡献没有明显偏倚。单细胞数据支持 multiplex-edited HSPC 对成熟血液谱系的稳定贡献。

第四, 与历史 TBI 对照相比, 研究中未观察到明显非造血组织毒性, 中性粒细胞和血小板恢复较快, 动物几乎不依赖输血。

这是一组强的灵长类证据。但它仍然不是“可以替代 busulfan”这一临床命题的证明。样本量是 4 只动物; 对照是历史 TBI, 不是儿童 exa-cel 的 busulfan 方案; 疾病背景不是 SCD 或 TDT 儿童; 终点是模型中的编辑细胞重建, 不是人体里的 VOC 消失、输血独立、肝毒性降低、感染减少、生育功能保留和总体治疗可及性改善。

剂量窗比靶点更难

CD45 是一个强靶点, 因为它广泛表达于白细胞和造血系统细胞, 能把杀伤集中到血液系统。α 粒子的射程短、线性能量转移高, 理论上适合把辐射剂量压进骨髓和造血组织, 减少对实体器官的外溢伤害。

但正因为它强, 临床剂量窗不会自动变宽。

第一道问题是 human dosimetry。非人灵长类的 300/400 μCi/kg 不能直接平移到儿童。人体里抗体分布、骨髓负荷、脾脏、肝脏、炎症状态、既往输血和铁负荷, 都可能改变吸收剂量。SCD 和 TDT 患者本身常带有肝脏、脾脏、炎症和血管病变背景, 与健康动物不同。

第二道问题是“足够清髓”的阈值。对于血红蛋白病, 目标不是一点 donor chimerism 或一点编辑信号, 而是达到能长期改变红细胞输出的 HSC 替换水平。CD45-²¹¹At 要在人体里证明的不是“有植入”, 而是在可接受毒性下稳定达到足以替代 busulfan 的植入比例。

第三道问题是 failure mode。busulfan 的毒性很明确, 但它的药代监测、目标 AUC、救援路径和移植中心经验也很成熟。定向放射免疫清髓如果剂量不足, 可能换来植入不足; 如果剂量过高, 可能带来深度免疫清除、感染、迟发血液学恢复、器官吸收剂量或二次恶性风险。临床开发不能只证明“比化疗更优雅”, 必须证明风险比 busulfan 更可控。

CD117 给了人体线索, 也提醒我们不要过度简化

CD45-²¹¹At 不是唯一的定向清髓路线。CD117/c-Kit 是另一个重要方向, 因为它更贴近 HSPC 和 stem cell factor 轴。briquilimab/JSP191 的人类数据已经给出了一部分 proof of concept。

2026 年 5 月, Lori Muffly、Catherine Lee、Andrew Artz 等人在 Blood 发表了 briquilimab/JSP191 的 phase 1 研究: 在高危 AML/MDS 老年患者中, 将 anti-CD117 antibody 加入 nonmyeloablative fludarabine 和低剂量 TBI 的异基因移植预处理。32 名患者中没有 briquilimab infusion reaction、dose-limiting toxicity 或 primary graft failure; 骨髓样本显示 AML/MDS HSPC 平均减少约 62%。这说明 CD117 靶向确实可以进入人体移植预处理流程。

但它同时说明了边界: 这不是单药替代 busulfan, 也不是自体 HSC 基因治疗的直接证据。另一个 SCD/β-thalassemia 的 briquilimab 研究登记方案也仍然和 300 cGy TBI、alemtuzumab、sirolimus 等组合在一起。换句话说, 目前 CD117 更像是在降低传统预处理强度或改造异基因移植流程, 还没有回答“在儿童自体基因治疗里能否独立打开足够生态位”的问题。

这对 CD45-²¹¹At 也有启发: 定向不等于无毒, 靶向不等于可替代。真正需要比较的是完整治疗包: 植入强度、住院时间、感染窗口、血小板恢复、肝毒性、黏膜毒性、生育毒性、长期恶性风险、制造协调和中心可及性。

可及性不是最后一步, 而是产品的一部分

如果 CD45-²¹¹At 未来在人体里跑通, 它仍然不会天然变成一个更容易普及的方案。原因在 ²¹¹At 本身。

Astatine-211 的半衰期约 7.2 小时, 适合靶向 α 治疗, 但也意味着生产、标记、质控、运输和给药必须高度协调。早期综述已经指出, ²¹¹At 的主要限制不是理论成本, 而是需要具备中能 α 粒子束能力的回旋加速器, 全球符合条件的设备数量有限。即便 2026 年的生产网络比 2011 年更成熟, 这个物理约束仍然存在: 它不像 busulfan 那样可以作为常规药品在全球移植中心稳定获得。

这会把可及性问题推到三个层面。

第一是地理可及性。能做 exa-cel 的中心已经不多; 能在自体细胞产品窗口内协调 α 放射药生产和输注的中心会更少。

第二是流程可及性。exa-cel 已经需要动员、采集、制造、质控、冷链、busulfan 预处理、住院和长期随访。换成 CD45-²¹¹At 后, 预处理可能更精准, 但流程会增加放射药 GMP、辐射安全、剂量学和核医学协作。它可能降低生物毒性, 但提高系统复杂性。

第三是商业可及性。一个基因治疗产品如果必须绑定一个短半衰期放射性抗体, 谁负责供应? 谁拥有标签? 谁承担剂量失败、制造延迟或 cell product 延误后的责任? 这不是科学论文里的附属问题, 而是产品能否规模化的一部分。

未来不是去掉清髓, 而是重新定义清髓

血液系统疾病基因治疗的过去, 是从异基因移植继承来的清髓逻辑: 先用化疗或放疗清空骨髓, 再把细胞放进去。现在, exa-cel 这类疗法已经把“细胞能不能被编辑并产生疗效”推到了很高的确定性, 于是清髓从背景步骤变成了主要瓶颈。

未来可能有三条路线。

第一条是优化传统 busulfan: 更好的 PK 监测、更细的风险分层、更强的 VOD 预防和支持治疗。这会改善边际安全性, 但很难改变“全身烷化剂清髓”的本质。

第二条是靶向清髓: CD117 抗体、CD117-ADC、CD45-RIT、免疫毒素、抗体偶联药物或组合预处理。它们的共同目标不是完全无清髓, 而是把杀伤从全身转向 HSC niche 和造血系统。

第三条是绕开 ex vivo HSC 移植: 例如体内递送、体内编辑、选择性扩增或无需深度 niche opening 的策略。但对于 SCD/TDT 这种需要稳定红系输出的疾病, 这条路仍要面对同一个根本问题: 如何让足够多的长期 HSC 或其功能后代被改写。

因此, “取代 busulfan”不是一个单点事件。更可能出现的是分阶段替代: 先在高资源中心的成人或青少年早期试验中证明安全剂量; 再在特定高风险或 busulfan 不适合人群中证明风险收益; 最后才可能进入儿童血红蛋白病一线基因治疗流程。

我的判断

我对这个领域的判断不是押注 CD45-²¹¹At、CD117 或任何单一路线。真正值得看的是: 清髓正在从“移植前的必要伤害”变成一个可以被工程化、比较和定价的产品模块。未来胜出的方案未必是最靶向、最先进或机制最漂亮的方案, 而是能在足够植入、可控毒性和真实可及性之间取得最好综合平衡的方案。

儿童 exa-cel 的疗效越干净, busulfan 的代价就越刺眼。前面提到的这篇刚刚发表于 New England Journal of Medicine 的儿科 exa-cel 读数说明, 编辑疗法已经足以让人相信 disease-modifying 甚至 functional cure 的方向; 但同一组数据也说明, 如果清髓仍然依赖传统全身毒性, 儿童、低资源地区和长期安全性都会成为疗法扩张的硬边界。

CD45-²¹¹At 的灵长类数据是这个方向上很强的一步, 因为它在长期、多谱系、HSC 层面的编辑细胞重建上给出了比普通小鼠实验更接近转化的问题答案。但这只是定向清髓版图中的一个节点。CD117 阻断、ADC、radioimmunotherapy、免疫毒素、低剂量组合预处理、体内 HSC 编辑和选择性扩增, 都是在回答同一个问题: 能不能在不牺牲疗效阈值的前提下, 减少传统清髓对肝脏、黏膜、感染风险、生育和住院资源的索取。

我的当前判断是: busulfan 短期内仍会是自体 HSC 基因治疗最现实、最成熟的清髓基准; 定向清髓还处在“证明能否更好地重新分配风险”的阶段, 而不是“已经替代标准方案”的阶段。这个领域的核心竞争不只是找到一个更聪明的靶点, 而是把植入、毒性、制造、放射药或抗体供应、中心流程和支付方式整合成一个可复制治疗系统。谁能做到这一点, 谁才真正降低了血液基因治疗的门槛。

Sources

Pediatric exa-cel sends a clear signal for blood gene therapy: the edited cells can generate a striking therapeutic effect, but the treatment still depends on one of the heaviest steps inherited from transplant medicine: myeloablative conditioning. The question is not whether busulfan works. The question is whether a therapy that first requires children to accept full marrow ablation, severe hematologic toxicity, VOD/SOS risk, hospitalization, and fertility risk can move from a curative option at specialized centers into a broadly accessible disease-modifying system.

So the question “can targeted conditioning replace busulfan?” should not be answered as a simple yes or no. A more accurate view is this: recent nonhuman primate data for a CD45 antibody linked to an alpha emitter have moved the idea of targeted marrow niche opening and long-term edited-cell engraftment closer to a translational problem. But for pediatric exa-cel-like use cases, three gates remain: human dose window, pediatric safety window, and the availability of the radiopharmaceutical and treatment-center infrastructure.

Editing Has Worked; Conditioning Still Collects the Toll

In June 2026, Haydar Frangoul, Franco Locatelli, and the CLIMB THAL-141 / CLIMB SCD-151 study groups published the latest clinical readout of exa-cel in children aged 5 to 11 years in the New England Journal of Medicine. This recent readout has two very different meanings.

The first is efficacy. The study included 15 children with transfusion-dependent beta-thalassemia and 11 children with sickle cell disease who received exa-cel. Among children with TDT who were evaluable at at least 16 months, 8 of 8 achieved transfusion independence for at least 12 months. Among evaluable children with SCD, 8 of 8 were free from severe vaso-occlusive crises for at least 12 months. Ex vivo CRISPR-Cas9 editing of the BCL11A erythroid enhancer continues to support the biological thesis that raising fetal hemoglobin can reshape the phenotype of hemoglobinopathies.

The second is cost. Every treated child had at least one grade 3 or 4 adverse event. In the TDT cohort, two children had severe hepatic veno-occlusive disease / sinusoidal obstruction syndrome assessed as related to busulfan conditioning, and one of them died.

This is not an exa-cel-only issue. The U.S. label for CASGEVY places full myeloablative conditioning inside the treatment process. In the adult and adolescent label data, 100% of patients with SCD and TDT had grade 3 or 4 neutropenia and thrombocytopenia after busulfan conditioning and CASGEVY infusion. In the TDT cohort, VOD is not a background signal that can be ignored: it appears in 5 of 52 treated patients in the label. The label also warns that after conditioning, patients may not be able to become pregnant or father a child, and should discuss fertility preservation before treatment.

This points to a basic fact: in blood gene therapy, editing efficiency, HbF induction, and clinical remission are only half of the therapy. The other half is the price paid to let those cells live in the marrow. Current products still borrow the conditioning logic of the transplant era.

Busulfan Is Not a Supporting Character

Busulfan has a straightforward value: it clears marrow niches so that reinfused autologous edited HSPCs have a chance to engraft long term. For hemoglobinopathies, which require sustained production of functional red cells, transient expression is not enough. A partial change in peripheral blood cells is not enough. What matters is durable contribution at the HSC level.

That is why reducing conditioning intensity is not a simple ethical preference. Less conditioning may reduce toxicity, but it may also reduce edited-cell engraftment. If engraftment falls below the required threshold, HbF, normal hemoglobin, or corrected red-cell output may not be sufficient. The risk may not shrink linearly; it may move from acute toxicity to inadequate efficacy, recurrent transfusion, retreatment, or rescue transplantation.

Busulfan is therefore not an easy drug to remove from the regimen. It is hidden infrastructure for ex vivo HSC gene therapy: product manufacturing, hospital workflow, center qualification, payer logic, patient selection, and long-term follow-up all orbit around it. Changing conditioning is not changing one line in a prescription. It changes the risk structure of the whole product.

What CD45 Alpha Conditioning Is Really Trying to Replace

In June 2026, Stefan Radtke, Hans-Peter Kiem, Roland Walter, and colleagues at Fred Hutch published in Blood a nonhuman primate study of [211At]astatine-labeled humanized anti-CD45 antibody conditioning for autologous HSPC gene therapy. The importance of this recent paper is that it does not only show depletion of hematopoietic cells. It pushes the problem into the level that matters most for autologous HSPC gene therapy: whether edited HSPCs can rebuild the blood system durably, across lineages, and without obvious bias.

The study used HuBC8, a humanized CD45 antibody, labeled with the alpha emitter astatine-211 as CD45-directed radioimmunotherapy. As a model, mobilized CD34+ HSPCs were multiplex edited: an adenine base editor modified the HBG promoter to reactivate fetal hemoglobin, while CD33 was deleted. Nonhuman primates then received 300 or 400 uCi/kg of 211At-CD45-RIT before reinfusion of edited cells.

Several results matter.

First, engraftment was dose-dependent. Single-cell sequencing showed up to roughly 70% combined monoallelic and biallelic edited cells in blood, consistent with substantial replacement of the marrow stem-cell compartment.

Second, the signal persisted beyond 18 months. For HSC gene therapy, this matters more than short-term peripheral-blood readouts because it points to long-term stem-cell contribution.

Third, lineage contribution appeared stable and unbiased. Single-cell data supported contribution of multiplex-edited HSPCs to mature blood lineages.

Fourth, compared with historical total-body irradiation controls, the treated animals did not show noticeable non-hematopoietic toxicity, had rapid neutrophil and platelet recovery, and were almost entirely transfusion-independent.

This is strong primate evidence. But it is not proof that the approach can replace busulfan clinically. The sample size was four animals. The comparison was against historical TBI controls, not the busulfan regimen used in pediatric exa-cel. The disease context was not children with SCD or TDT. The endpoint was edited-cell reconstitution in a model, not freedom from VOCs, transfusion independence, lower liver toxicity, fewer infections, fertility preservation, or improved access in humans.

The Dose Window Is Harder Than the Target

CD45 is a powerful target because it is broadly expressed on leukocytes and hematopoietic cells, allowing the radiation payload to concentrate in the blood system. Alpha particles have short path length and high linear energy transfer, which makes them theoretically attractive for delivering radiation into marrow and hematopoietic tissue while limiting spillover into solid organs.

But a strong target does not automatically create a wide clinical window.

The first problem is human dosimetry. The 300/400 uCi/kg doses in nonhuman primates cannot be directly carried into children. Antibody distribution, marrow burden, spleen, liver, inflammation, prior transfusion, and iron overload may all change absorbed dose. Patients with SCD and TDT often have hepatic, splenic, inflammatory, and vascular background disease that differs from healthy animals.

The second problem is the threshold for sufficient myeloablation. In hemoglobinopathies, the goal is not a little donor chimerism or a small editing signal. The goal is enough HSC replacement to durably change red-cell output. CD45-211At must prove in humans not simply that it enables engraftment, but that it reaches a busulfan-like engraftment threshold within an acceptable toxicity window.

The third problem is failure mode. Busulfan toxicity is real, but its pharmacokinetic monitoring, target AUC ranges, rescue pathways, and transplant-center experience are mature. If targeted radioimmunotherapy underdoses, engraftment may be insufficient. If it overdoses, the risks may include deep immune depletion, infection, delayed hematologic recovery, organ absorbed dose, or secondary malignancy. Clinical development cannot merely show that the approach is more elegant than chemotherapy. It has to show that the risk is more controllable than busulfan.

CD117 Gives a Human Signal, and a Warning Against Oversimplification

CD45-211At is not the only targeted-conditioning route. CD117/c-Kit is another important direction because it sits closer to the HSPC and stem cell factor axis. Human data for briquilimab/JSP191 already provide part of the proof of concept.

In May 2026, Lori Muffly, Catherine Lee, Andrew Artz, and colleagues published a phase 1 study of briquilimab/JSP191 in Blood. In older adults with high-risk AML/MDS, they added an anti-CD117 antibody to nonmyeloablative fludarabine and low-dose TBI for allogeneic transplant conditioning. Among 32 enrolled patients, there were no briquilimab infusion reactions, dose-limiting toxicities, or primary graft failures. Marrow samples showed an average 62% reduction in AML/MDS HSPCs. This suggests that CD117 targeting can enter human transplant-conditioning workflows.

But it also shows the boundary. This is not single-agent replacement of busulfan, and it is not direct evidence for autologous HSC gene therapy. Another registered briquilimab study in SCD and beta-thalassemia still combines the antibody with 300 cGy TBI, alemtuzumab, and sirolimus. In other words, CD117 currently looks more like a way to reduce traditional conditioning intensity or redesign allogeneic transplant workflows than an answer to whether pediatric autologous gene therapy can open enough marrow niche without busulfan.

The lesson also applies to CD45-211At: targeted does not mean non-toxic, and targeted does not mean clinically replaceable. What must be compared is the whole treatment package: engraftment strength, hospitalization, infection window, platelet recovery, liver toxicity, mucositis, fertility risk, long-term malignancy risk, manufacturing coordination, and access.

Access Is Not the Last Step; It Is Part of the Product

Even if CD45-211At works in humans, it will not automatically become an easier-to-scale regimen. The reason is astatine-211 itself.

Astatine-211 has a half-life of about 7.2 hours, which is useful for targeted alpha therapy but also means that production, labeling, quality control, transport, and dosing must be tightly coordinated. An early review noted that the main constraint for 211At is not theoretical cost but the need for cyclotrons with medium-energy alpha-particle beams; globally, only a limited number of machines have the required characteristics. Even if the production network is more mature in 2026 than it was in 2011, the physical constraint remains: 211At is not like busulfan, a conventional drug that transplant centers can access routinely around the world.

That pushes access into three layers.

First, geographic access. The number of centers that can deliver exa-cel is already limited. The number that can coordinate alpha-radiopharmaceutical production and infusion inside an autologous cell-therapy window will be smaller.

Second, workflow access. Exa-cel already requires mobilization, collection, manufacturing, quality control, cold chain, busulfan conditioning, hospitalization, and long-term follow-up. Replacing busulfan with CD45-211At may make conditioning more precise, but it adds radiopharmaceutical GMP, radiation safety, dosimetry, and nuclear-medicine coordination. It may lower biological toxicity while increasing system complexity.

Third, commercial access. If a gene therapy product must be paired with a short-lived radioactive antibody, who supplies it? Who owns the label? Who carries responsibility for dose failure, manufacturing delay, or mismatch with the cell-product timeline? These are not details outside the paper. They are part of whether the product can scale.

The Future Is Not Removing Conditioning, but Redefining It

The past of blood-disease gene therapy inherited its conditioning logic from allogeneic transplant: use chemotherapy or radiation to clear marrow, then put cells back in. Now, therapies such as exa-cel have pushed the question of whether cells can be edited and work in patients to a high level of certainty. Conditioning has moved from a background step to a central bottleneck.

The future likely has three routes.

The first is optimizing conventional busulfan: better pharmacokinetic monitoring, finer risk stratification, stronger VOD prevention, and supportive care. This may improve marginal safety but will not change the nature of systemic alkylator-based myeloablation.

The second is targeted conditioning: CD117 antibodies, CD117 ADCs, CD45 radioimmunotherapy, immunotoxins, antibody-drug conjugates, or combination regimens. Their shared goal is not truly conditioning-free therapy, but moving the injury from the whole body toward the HSC niche and hematopoietic system.

The third is avoiding ex vivo HSC transplant altogether: in vivo delivery, in vivo editing, selective expansion, or strategies that require less niche opening. But for SCD and TDT, which require stable red-cell output, the same fundamental question remains: how can enough long-term HSCs, or their functional descendants, be rewritten?

Therefore, replacement of busulfan is unlikely to be a single event. It is more likely to unfold in stages: first, dose and safety in adults or adolescents at highly capable centers; then, risk-benefit in selected patients who are poor candidates for busulfan; and only later, possible use in frontline pediatric hemoglobinopathy gene therapy.

My View

My view of this field is not a bet on CD45-211At, CD117, or any single route. The important shift is that conditioning is moving from “necessary injury before transplant” into a product module that can be engineered, compared, and priced. The winning approach may not be the most targeted, the most advanced, or the most elegant mechanistically. It will be the approach that finds the best combined balance among sufficient engraftment, controllable toxicity, and real-world access.

The cleaner pediatric exa-cel efficacy looks, the more visible the cost of busulfan becomes. The newly published New England Journal of Medicine pediatric exa-cel readout discussed above shows that editing-based therapy is credible as a disease-modifying and potentially functional-cure strategy. But the same data also show that if conditioning continues to rely on traditional systemic toxicity, children, lower-resource settings, and long-term safety will remain hard boundaries for expansion.

The CD45-211At primate data are a strong step in this direction because they address long-term, multilineage, HSC-level edited-cell reconstitution in a model closer to translation than ordinary mouse studies. But this is one node in the targeted-conditioning map. CD117 blockade, ADCs, radioimmunotherapy, immunotoxins, low-dose combination conditioning, in vivo HSC editing, and selective expansion are all answering the same question: can we reduce the price paid by the liver, mucosa, infection risk, fertility, and hospital infrastructure without losing the efficacy threshold?

My current judgment is that busulfan will remain the most realistic and mature conditioning benchmark for autologous HSC gene therapy in the near term. Targeted conditioning is still in the stage of proving whether it can redistribute risk better, not in the stage of having replaced the standard regimen. The core competition in this field is not simply finding a smarter target. It is integrating engraftment, toxicity, manufacturing, radiopharmaceutical or antibody supply, center workflow, and payment into a reproducible treatment system. Whoever does that will have lowered the real barrier for blood gene therapy.

Sources