Primer

2019年的研究发现,健康人群中最高达78%的人已携带能识别Cas9的抗体或T细胞——这套源自两种常见致病菌的免疫记忆,在体外一次性RNP编辑里被巧妙避开,却在追求长期体内表达的下一代基因编辑疗法门前显出真实分量。

A 2019 study found that up to 78% of healthy people already carry antibodies or T cells that recognize Cas9 — immune memory left by two common pathogenic bacteria that ex vivo, one-time RNP editing quietly sidesteps, but that shows its real weight at the door of next-generation therapies aiming for sustained in vivo expression.

全球第一款获批的CRISPR疗法Casgevy如何编辑BCL11A关掉致病开关、两项3期试验交出97%和91%的达标率,以及220万美元一次的治疗为何近两年后才真正开始放量。

How Casgevy, the world's first approved CRISPR therapy, edits BCL11A to silence the switch that shuts off fetal hemoglobin, why two phase 3 trials produced 97% and 91% response rates, and why a $2.2 million one-time treatment took nearly two years to actually start reaching patients.

Casgevy 标价 220 万美元,但真正烧钱的不是编辑那一步:从动员采集到放行检验,这条 ex vivo 制造供应链两头重、中间轻,不少病人常常倒在编辑开始之前。

Casgevy is priced at $2.2 million, but the money mostly isn't going to the editing step — across this ex vivo manufacturing supply chain, from mobilization and collection to release testing, the load sits at both ends and light in the middle, and many patients drop out before editing even begins.

“从G-CSF被禁用、plerixafor动员采集15人仅9人达标,到白消安清髓、27天中性粒细胞与35天血小板engraftment,再到FDA要求的十五年随访——这篇把exa-cel、lovo-cel”一针治愈”背后完整的移植医学全流程摆回台面。”

From G-CSF being off-limits and plerixafor mobilization hitting its collection threshold in only 9 of 15 patients, to busulfan conditioning with 27-day neutrophil and 35-day platelet engraftment, to the FDA's required fifteen years of follow-up — this piece puts the full transplant-medicine process behind exa-cel and lovo-cel's "one-shot cure" back on the table.

In vivo 基因编辑不清髓,骨髓龛位仍被原住民细胞占满——范可尼贫血、ADA-SCID 里疾病本身给了编辑后细胞天然的生存优势,而镰状细胞病、地中海贫血这类没有天然优势的病,只能靠 MGMT 药物选择等工程手段人为制造竞争优势。

In vivo gene editing skips conditioning, so the marrow niches stay full of untouched native cells — Fanconi anemia and ADA-SCID let the disease itself hand corrected cells a natural survival edge, while sickle cell disease and thalassemia, which carry no such edge, have to rely on engineered tools like MGMT drug selection instead.

1999到2002年间,γ逆转录病毒疗法让近二十名X-SCID婴儿中的大多数重获免疫系统,但巴黎九名患儿中四人后来患上白血病——这篇讲整合型病毒载体如何从LMO2那场事故里,一步步把风险收窄进自失活设计。

Between 1999 and 2002, gammaretroviral therapy restored immunity in most of nearly twenty infants with X-SCID, but four of nine treated children in the Paris cohort later developed leukemia — this piece follows how integrating viral vectors narrowed that risk, step by step, into the self-inactivating design behind today's therapies.

2023 年 12 月同一天,两款镰刀型贫血基因疗法定价相差近百万美元——ICER 算出的物有所值区间其实低于两家公司的官方定价,CMS 在 2024 年把疗效支付协议做成了全国统一的公共支付基础设施。

On the same December 2023 day, two sickle cell gene therapies were priced nearly a million dollars apart — ICER's own value-based range actually falls below both companies' list prices, and in 2024 CMS turned outcomes-based agreements into a nationwide public payment infrastructure.

同一台编辑机器放到不同的病上,「够」的定义可以完全不同——疗效阈值写在红细胞的寿命、淋巴细胞的扩张和酶的扩散半径里,而不写在编辑效率里。

The same editing machine, applied to different diseases, meets completely different definitions of enough — the correction threshold is written in red-cell lifespan, lymphocyte expansion and the diffusion radius of an enzyme, not in editing efficiency.

表观编辑不改一个碱基就能把基因关掉,并让沉默随细胞分裂传下去;它在小鼠肝脏和食蟹猴身上已经跑通,在造血干细胞里却仍停留在人类原代细胞加异种移植这一层。

Epigenetic editing switches a gene off without changing a single base and lets the silencing pass down through cell division; it has worked in mouse liver and in cynomolgus monkeys, but in hematopoietic stem cells the evidence still stops at human primary cells plus xenotransplanted mice.

小鼠体内的长期造血干细胞平均每 57 天才分裂一次,而这份「什么都不做」不是待机,是被 CDK inhibitor、代谢通路与 autophagy 主动按住的工作状态——基因治疗要改写它们,第一步就得先把它们叫醒,而叫醒本身就要付代价。

Long-term HSCs in mice divide roughly once every 57 days, and that stillness is not a power-saving mode but a state actively held in place by CDK inhibitors, metabolic pathways and autophagy — which is why gene therapy has to wake them up, and why waking them up has a price.

同一层脂质外壳让体内 base editing 在肝脏里成真,靠的是搭上 ApoE 那班车;而今天想让它去骨髓找造血干细胞,最难对付的恰恰是这项被精心训练出来的天赋。

The same lipid shell that made in vivo base editing real in the liver got there by hitching a ride on ApoE; now that the field wants it to find hematopoietic stem cells in the bone marrow, that carefully trained gift is the hardest thing to overcome.

为造一剂 CASGEVY,说明书建议先从病人身上取走每公斤至少 20×10⁶ 个 CD34⁺ 细胞,而最终回输的最低剂量只有 3×10⁶/kg;在镰刀型贫血的注册试验里,有十分之一的人因为细胞采集不足而没能接受治疗。

To manufacture one dose of CASGEVY, the label recommends collecting at least 20×10⁶ CD34⁺ cells/kg from the patient, while the minimum infused dose is only 3×10⁶/kg; in the sickle cell registration trial, one in ten patients never received the therapy because too few cells could be collected.

柏林病人靠一位缺了 32 个碱基的供者摆脱了 HIV;而后来出现的野生型与杂合供者病例正在提示,功劳可能有一半属于那场把整套免疫系统连根换掉的移植。

The Berlin patient was freed from HIV by a donor missing 32 bases; the wild-type and heterozygous donor cases that followed suggest half the credit may belong to the transplant that replaced his entire immune system.

1968 年,Hurler 与 Hunter 病人的成纤维细胞被混进同一个培养皿,两边的缺陷都被纠正了——溶酶体里的酶,可以借给邻居用。这条逻辑后来变成了把产酶细胞种进骨髓、再把酶一路送进脑脊液的疗法。

In 1968, fibroblasts from a Hurler patient and a Hunter patient were mixed in one dish and both defects were corrected—lysosomal enzymes can be lent to the neighbors. That logic eventually became a therapy that plants enzyme-producing cells in the bone marrow and delivers enzyme all the way into the cerebrospinal fluid.

供者 T 细胞与受者免疫的双向屏障,曾把半相合供者挡在候选名单的最后一行;把 cyclophosphamide 挪到移植后第 3、4 天再给,重新画了这条线——代价也随之换了科室。

The two-way barrier between donor T cells and recipient immunity long kept haploidentical donors on the last line of the candidate list; giving cyclophosphamide on days 3 and 4 after transplantation redrew that line — and moved the cost to another department.

让细胞对免疫系统隐形这件事,已经从小鼠走到猴子、走到人:异体胰岛与 CAR-T 都进了临床。造血干细胞不在这份名单上,因为它的后代就是免疫系统本身。

Making cells invisible to the immune system has moved from mice to macaques to human patients; allogeneic islets and CAR-T are already in the clinic. Hematopoietic stem cells are not on that list, because their progeny are the immune system itself.

编辑效率是幸存者报出来的成绩。从核酸酶剪断 DNA 的那一刻起,p53 就开始决定哪些细胞留下——而它最管不住的,恰恰是 p53 自己已经失能的那些。

Editing efficiency is a score reported by survivors. From the moment a nuclease cuts DNA, p53 begins deciding which cells stay — and the ones it can least restrain are precisely those in which p53 itself has already failed.

七十年过去,造血干细胞仍然没有身份证:所有标记给的都只是概率,唯一说了算的,还是把它放进一个身体,然后等。

Seventy years on, the hematopoietic stem cell still has no ID card: every marker gives only a probability, and the only thing that settles the question is still to put the cell into a body and wait.

AAV 拿下了眼睛、肝脏、肌肉和大脑,却在骨髓前止步;可同一个载体在培养皿里,又是精准改写造血干细胞的首选工具——同一批属性,换个场景就从致命变成刚刚好。

AAV has conquered the eye, the liver, muscle and the brain, yet stops at the door of the bone marrow; and yet in the dish, that same vector is the tool of choice for precisely rewriting hematopoietic stem cells—the same properties, flipped from fatal to just right by a change of setting.

基因治疗的前半程用镊子,后半程用推土机——CD117 与 CD45 抗体想证明,骨髓不是需要被烧平的战场,而是一栋有门牌号的楼。

Gene therapy uses tweezers for the first half and a bulldozer for the second—CD117 and CD45 antibodies aim to show that bone marrow is not a battlefield to be burned flat, but a building with door numbers.

六十五岁以上,每十个人里就有一个,血液里长出了可检出的优势克隆——其中绝大多数终生无事。但它意味着基因治疗动手的那片组织,从来就不是一张白纸。

Past sixty-five, roughly one person in ten has grown a detectable dominant clone in their blood—and most of them live their whole lives untroubled by it. But it means the tissue gene therapy operates on was never a blank page.

培养箱能可靠地给你更多细胞,却不能可靠地给你更多干细胞——五十倍与十七倍之间的那道缝隙,是 HSC 扩增二十年来真正的问题。

An incubator can reliably give you more cells; it cannot reliably give you more stem cells—the gap between fifty-fold and seventeen-fold is the real problem HSC expansion has been working on for twenty years.

镰状细胞病的负担压在撒哈拉以南非洲,能治愈它的 CRISPR 疗法却只长在高收入国家的医疗体系里——挡住它的主要不是药价,而是疗法的形态。

Sickle cell disease's burden falls on sub-Saharan Africa, while the CRISPR cure grows only inside the health systems of high-income countries—what stands in the way is mainly not the price of the drug, but the physical form of the therapy.

同一个坏掉的字母,三条基因治疗路线:加一段抗镰变 β-globin、敲 BCL11A 增强子重启胎儿血红蛋白、或把 HbS 直接改成良性变体——镰刀型贫血的治愈之路怎么分岔。

One broken letter, three gene-therapy routes: add an anti-sickling β-globin, knock out the BCL11A enhancer to reawaken fetal hemoglobin, or edit HbS directly into a benign variant — how the road to curing sickle cell disease forks.

早期 X-SCID 基因治疗救回了必死的孩子,却有患儿因载体激活 LMO2 而患上白血病——这场事故如何逼出自失活载体,重写了整个领域的安全范式。

Early X-SCID gene therapy rescued children who would otherwise have died — yet some of them developed leukemia when the vector activated LMO2. This is how that catastrophe forced the self-inactivating vector into being and rewrote the entire field's safety paradigm.

碱基编辑只能改四类点突变;它的近亲 prime editing 几乎能改写任意致病变异,却为何一再卡在写进长期造血干细胞这一步?

Base editing can only fix four classes of point mutation; its cousin prime editing can rewrite almost any pathogenic variant — so why does it keep stalling at the step of writing itself into long-term hematopoietic stem cells?

体内直接编辑造血干细胞,是把手术级移植压缩成打一针的圣杯——难在递送,以及它离那一针还差什么。

Editing hematopoietic stem cells directly inside the body is the holy grail that would shrink a surgery-grade transplant into a single injection — the hard part is delivery, and this is what still stands between it and that one shot.

不切断双链、直接把一个碱基改写:base editing 如何把基因编辑从破坏后修复变成化学改写,以及它在造血干细胞里的机会与新风险。

Rewriting a single base without cutting the double strand: how base editing turns gene editing from break-then-repair into chemical rewriting, and what that means — in opportunity and in new risk — for hematopoietic stem cells.

从一个 GWAS 统计信号,到第一个获批的 CRISPR 疗法——BCL11A 如何成为血红蛋白病治疗的枢纽,以及它示范的选靶范式。

From a single GWAS statistical signal to the first approved CRISPR therapy—how BCL11A became the linchpin of hemoglobinopathy treatment, and the target-selection paradigm it demonstrates.