Yang Liu

← Primer

HSC · Healing Source Code

一款两百万美元起的疗法——「一次治愈」的定价逻辑与支付学 A Cure Starting at Two Million Dollars — The Pricing Logic and Payment Science of the 'One-Time Cure'

2026-07-31 · Casgevy, Lyfgenia, 基因治疗定价, 疗效支付协议, CMS, 镰状细胞病

2023 年 12 月 8 日,美国 FDA 在同一天批准了两款治疗镰刀型贫血的基因疗法。新闻通稿里最先被引用的,往往不是它们如何编辑基因组,而是紧随其后的一串数字:Vertex 与 CRISPR Therapeutics 的 Casgevy,美国批发采购价 220 万美元;bluebird bio 的 Lyfgenia,310 万美元。同一种病,同一天获批,价格却相差近百万美元。这不是定价失误,而是整个行业第一次被迫回答一个此前从未认真回答过的问题:一次性治愈一种终身疾病,到底该怎么定价、又该怎么付钱。

价格不是成本,是被省下来的钱

一支疫苗或一片药的定价,大体还能和生产成本挂上钩。基因疗法的定价逻辑完全是另一套账。Casgevy 和 Lyfgenia 的原料是患者自己的造血干细胞,加工成本确实高昂,但真正撑起两三百万美元报价的,不是这条生产线,而是”如果不治疗,病人一辈子要花多少钱”这道题。镰刀型贫血患者反复发作的血管闭塞危象、器官损伤、终身输血与随之而来的铁过载治疗,叠加下来的终身医疗支出被业界估算到数百万美元量级——一次性疗法的定价,本质上是在向支付方兜售”提前把这笔长期支出买断”的逻辑。

这套逻辑有一个学术上的检验标准:每挽回一个质量调整生命年(QALY)要花多少钱。美国的非营利机构 Institute for Clinical and Economic Review(ICER)在 2023 年 8 月发布的最终证据报告里,给出了 Casgevy 和 Lyfgenia 的”物有所值”价格区间——只有当两款疗法的定价落在 135 万至 205 万美元之间时,才能达到常规的成本效果阈值。换句话说,两家公司最终挂出的官方定价(220 万与 310 万美元),均已超出了这条学术意义上的”划算线”。ICER 同时提醒,这个区间本身建立在对长期疗效的模型推算之上——支持 Casgevy 获批的关键 3 期试验 CLIMB SCD-121 中,44 名患者接受了 exa-cel 输注,其中 30 人有足够长的随访数据可评估主要终点,29 人(97%)在输注后连续至少 12 个月未再发生血管闭塞危象,中位随访时间只有 19.3 个月(Frangoul et al. 2024, N Engl J Med)。两款疗法上市时都只有这样几年量级的随访数据,谁也不知道这份”治愈”能不能撑过患者的一生。

同一个池子里的价目表

Casgevy 与 Lyfgenia 之所以价格不同,一部分原因写在技术路线里:前者用 CRISPR-Cas9 编辑 BCL11A 的红系增强子,唤醒胎儿血红蛋白;后者用慢病毒载体把一个功能性抗镰变珠蛋白基因整合进基因组,是更传统的基因添加。bluebird bio(该公司已于 2025 年 9 月更名为 Genetix Biotherapeutics Inc.,CMS 官方现行文件已用新名指代 Lyfgenia 的签约方)为 Lyfgenia 更高定价给出的解释是:这一定价”体现了该疗法可能带来的稳健而持久的临床获益,以及减少或消除血管闭塞事件(VOE)对患者医疗资源使用、未来收入与人生机会所产生的终身影响的估值”——一个和 ICER 的模型逻辑互为镜像的说法:同一套”终身成本”账本,不同公司算出了不同的报价。

把镜头拉远一点,会发现这两款药只是一张更长价目表上的两行。同样出自 bluebird bio 的 Zynteglo(治疗输血依赖型 β-地中海贫血)标价 280 万美元;该公司一个月后获批的 Skysona(治疗脑肾上腺脑白质营养不良,2022 年 9 月获批)标价 300 万美元,在它上市的那一刻曾短暂坐上”全球最贵药物”的位置。这个头衔很快又被 CSL Behring 与 uniQure 的血友病 B 基因疗法 Hemgenix 拿走——350 万美元一针,美国 FDA 于 2022 年 11 月批准。CSL Behring 的辩护逻辑与 bluebird 如出一辙:公司测算称,即便定价 350 万美元,Hemgenix 仍能为美国医疗系统在每个患者身上省下 500 万至 580 万美元的长期支出。更早一步蹚出这条路的是 Novartis 旗下 AveXis 治疗脊髓性肌萎缩症的 Zolgensma——2019 年以 212.5 万美元定价上市时,同样引来”是否物有所值”的争论,但也第一次把”能不能分期付”这个问题摆上了台面。

一次性的钱,分期付

真正让这个行业的支付方式发生转向的,不是价格本身,而是围绕价格衍生出的一整套金融工程。Zolgensma 是最早的试验场:AveXis 为保险公司设计了两套方案——五年分期支付,以及一份与疗效挂钩的五年期”outcomes-based agreement”(疗效支付协议),把 212.5 万美元摊成每年约 42.5 万美元,并附加”疗效不达标可部分退款”的条款。bluebird bio 为 Zynteglo 设计的合约走得更远,承诺最高 80% 的退款比例——如果患者未能在约定时间内摆脱输血依赖,厂商要把大部分货款退回去。相比之下,bluebird 自己也承认,Skysona 面对的患者群体太小、病程太复杂,类似的疗效对赌几乎无法设计,索性没有提供这类合约。

这类”先用后买单、按疗效结账”的思路,并非药企临时起意,背后有一条学术脉络。MIT 的 Mark Trusheim 与内科医生出身、时任 Memorial Sloan Kettering 卫生政策与结果研究中心(Center for Health Policy and Outcomes)主任的 Peter Bach,在 2018 年的《JAMA》上提出过一个后来被业界戏称为”Netflix 模式”的方案:最初是为丙肝新药设计——政府或保险方与厂商谈定一笔固定总价,在协议期内不限量地治疗尽可能多的患者,把”按人头付费”变成”按订阅付费”。这条思路后来被移植到基因疗法领域,衍生出各种分期偿付、按年付息的”年金模式”,核心目的都是同一个:把一笔天文数字的一次性支出,拆解成保险体系消化得动的现金流。

这套逻辑在 2024 年被美国联邦政府第一次搬上了国家级的操作台。美国医疗保险与医疗补助服务中心(CMS)推出的 Cell and Gene Therapy(CGT)Access Model,专门针对镰刀型贫血基因疗法设计了一套”国家代理谈判”机制:由 CMS 出面,与 Vertex(Casgevy)和 bluebird bio(Lyfgenia)分别谈定统一的疗效支付协议模板,各州 Medicaid 项目不必各自单独议价,只需在协议框架下换取厂商返利。这套模型的申请窗口在 2024 年 5 月截止,谈判持续到当年 11 月,截至该项目推进阶段,已有 33 个州加上华盛顿特区与波多黎各加入,覆盖全美 84% 享受 Medicaid 保障的镰刀型贫血患者,这批患者的总数被官方估算超过 10 万人。这是”疗效支付”第一次不再是某家药企与某家保险公司之间的私下条款,而变成了一整套由政府背书、跨州通用的公共支付基础设施。

还没解决的部分

分期支付解决的是现金流问题,却解决不了几个更棘手的账。第一,美国的医疗保险高度碎片化:雇主每隔几年更换承保方案是常态,一个患者今天在 A 保险公司付了第一年的分期款,明年跳槽换了工作、换了保险,谁来接盘剩下的四年账单,至今没有行业统一答案——这正是”一次性治愈”和”按年付费”这两种逻辑天然的错位。第二,所有这些疗效支付协议,赌注押在一个尚未被时间验证的假设上:编辑或整合进基因组的改动能维持多久。上市时几年的随访数据支撑不起”终身治愈”这个词该有的确定性,一旦某个远期并发症或疗效衰减在多年后浮现,今天谈定的退款条款很可能早已过期。第三,即便定价机制被打磨得再精巧,这套价目表的门槛本身仍然把绝大多数患者挡在外面——镰刀型贫血的疾病负担集中在撒哈拉以南非洲,而能负担两三百万美元一针的支付体系,几乎只存在于少数几个富裕国家。这道更根本的鸿沟,留到专门讲可及性的那一篇再展开。

价目表上的数字终会被更多年的真实世界数据修正,分期方案也会一版版迭代。但 2023 年那个同一天挂出的两个价签,已经标记出这个领域必须共同回答的问题:当科学第一次能把”治愈”两个字兑现成一针注射,金融和保险体系能不能跟上,决定了这份治愈究竟是少数人的特权,还是一项可以被规模化交付的公共产品。

参考文献

  1. Frangoul H, et al. Exagamglogene Autotemcel for Severe Sickle Cell Disease. N Engl J Med. 2024;390:1649-1662. DOI
  2. Trusheim MR, Cassidy WM, Bach PB. Alternative State-Level Financing for Hepatitis C Treatment — The “Netflix Model”. JAMA. 2018;320(19):1977-1978. DOI
  3. ICER (Institute for Clinical and Economic Review). Gene Therapies for Sickle Cell Disease: Effectiveness and Value — Final Evidence Report. Published Aug 21, 2023. icer.org / Press release
  4. CMS. Cell and Gene Therapy (CGT) Access Model — Frequently Asked Questions & program overview. cms.gov
  5. CMS Innovation Center. Innovation Insight: CMS Model Delivers Access to Sickle Cell Gene Therapy with Expansive List of State Participants. cms.gov
  6. Sidley Austin LLP. Cell and Gene Therapies Access Model on Sickle Cell Disease Announced by U.S. CMS Innovation Center. Feb 2024. sidley.com
  7. Novartis / AveXis. AveXis Announces Innovative Zolgensma Gene Therapy Access Programs for US Payers and Families. novartis.com
  8. Xtalks. Bluebird Bio’s Skysona Receives FDA Approval and Becomes World’s Most Expensive Drug. xtalks.com
  9. BioWorld. With $2.8M price tag, Bluebird’s Zynteglo first cell-based gene therapy for certain beta-thalassemia patients approved in US. bioworld.com
  10. CNN. FDA approves $3.5 million treatment for hemophilia, now the most expensive drug in the world. Nov 23, 2022. cnn.com
  11. Fierce Pharma. Vertex, CRISPR score landmark FDA approval for sickle cell disease gene therapy Casgevy. fiercepharma.com
  12. BioSpace. bluebird bio Details Plans for the Commercial Launch of LYFGENIA Gene Therapy for Patients Ages 12 and Older with Sickle Cell Disease and a History of Vaso-Occlusive Events. biospace.com

On December 8, 2023, the FDA approved two gene therapies for sickle cell disease on the same day. The number press releases led with was rarely how the genome had been edited — it was the figure that followed: Casgevy, from Vertex and CRISPR Therapeutics, carried a U.S. wholesale acquisition cost of $2.2 million; Lyfgenia, from bluebird bio, $3.1 million. Same disease, same approval day, and yet the prices differed by nearly a million dollars. This wasn’t a pricing mistake — it was the moment the industry was forced, for the first time, to seriously answer a question it had never really had to answer before: how do you price a one-time cure for a lifelong disease, and how does anyone pay for it?

The Price Isn’t the Cost — It’s the Money Saved

The price of a vaccine or a pill can still be roughly tied to its manufacturing cost. Gene therapy pricing runs on an entirely different ledger. The raw material for Casgevy and Lyfgenia is the patient’s own hematopoietic stem cells, and the processing costs are genuinely steep — but what actually supports a price tag of two or three million dollars isn’t that manufacturing line. It’s the question of how much a patient would otherwise spend, over an entire lifetime, if the disease went untreated. The recurring vaso-occlusive crises, organ damage, lifelong transfusions, and the iron-overload treatment that comes with them add up, by industry estimates, to a lifetime medical bill in the millions of dollars — and pricing a one-time therapy is, at bottom, selling payers on the logic of buying out that long-term spending in advance.

This logic has an academic yardstick: how much it costs per quality-adjusted life year (QALY) gained. The U.S. nonprofit Institute for Clinical and Economic Review (ICER), in its final evidence report published in August 2023, worked out a “value-based” price range for Casgevy and Lyfgenia — only if the two therapies were priced between $1.35 million and $2.05 million would they meet a conventional cost-effectiveness threshold. In other words, the official list prices the two companies eventually posted — $2.2 million and $3.1 million — both exceed that academically defined line of being “worth it.” ICER also cautioned that this range itself rests on modeled projections of long-term efficacy: in CLIMB SCD-121, the pivotal Phase 3 trial supporting Casgevy’s approval, 44 patients received an exa-cel infusion, 30 of whom had follow-up long enough to be evaluable for the primary endpoint, and 29 (97%) remained free of vaso-occlusive crises for at least 12 consecutive months after infusion, with a median follow-up of only 19.3 months (Frangoul et al. 2024, N Engl J Med). Both therapies reached the market with only a few years’ worth of follow-up data at that scale — nobody actually knows whether this “cure” will hold for the rest of a patient’s life.

A Price List From the Same Pool

Part of why Casgevy and Lyfgenia are priced differently is written into their technology: the former uses CRISPR-Cas9 to edit the erythroid enhancer of BCL11A, waking up fetal hemoglobin; the latter uses a lentiviral vector to integrate a functional anti-sickling globin gene into the genome — a more conventional gene-addition approach. bluebird bio (the company renamed itself Genetix Biotherapeutics Inc. in September 2025, and current CMS documents already refer to Lyfgenia’s contracted party by the new name) set Lyfgenia’s higher price “in recognition of the value the therapy may deliver through robust and sustained clinical benefits and the estimated lifetime impact that reducing or eliminating VOEs may have on patients’ healthcare utilization, future earnings, and life opportunities” — language that mirrors ICER’s own modeling logic: the same “lifetime cost” ledger, priced out differently by different companies.

Pull the lens back further, and these two drugs turn out to be just two lines on a longer price list. Zynteglo (for transfusion-dependent beta-thalassemia), also from bluebird bio, was priced at $2.8 million; the company’s Skysona (for cerebral adrenoleukodystrophy, approved September 2022), which cleared the FDA a month later, was priced at $3 million and briefly held the title of “world’s most expensive drug” the moment it launched. That title was quickly taken by CSL Behring and uniQure’s hemophilia B gene therapy Hemgenix — $3.5 million a dose, approved by the FDA in November 2022. CSL Behring’s justification followed the same logic as bluebird’s: the company calculated that even at $3.5 million, Hemgenix would still save the U.S. healthcare system between $5 million and $5.8 million in long-term costs per patient. The path had been broken a step earlier by Novartis’s AveXis, with Zolgensma for spinal muscular atrophy — when it launched in 2019 priced at $2.125 million, it likewise drew arguments over whether it was “worth it,” but it was also the first to put the question of installment payment on the table.

Paying in Installments for a One-Time Bill

What actually shifted how this industry gets paid wasn’t the price itself, but the whole apparatus of financial engineering built up around it. Zolgensma was the first testing ground: AveXis designed two options for insurers — a five-year installment payment plan, and a five-year “outcomes-based agreement” tied to efficacy, spreading the $2.125 million out to roughly $425,000 a year, with a clause allowing partial refunds if the therapy underperformed. The contract bluebird bio designed for Zynteglo went further, promising a refund of up to 80% — if a patient failed to become transfusion-independent within an agreed window, the manufacturer would return most of the payment. By contrast, bluebird itself acknowledged that Skysona’s patient population was too small and its disease course too complex for a comparable outcomes bet to be designed at all, so it simply didn’t offer this kind of contract.

This idea of paying after the fact, settling accounts by outcome, wasn’t something drug companies improvised on the spot — it has an academic lineage behind it. MIT’s Mark Trusheim, together with Peter Bach — a physician by training, then director of the Center for Health Policy and Outcomes at Memorial Sloan Kettering — proposed a scheme in JAMA in 2018 that the industry later nicknamed the “Netflix model”: designed originally for new hepatitis C drugs, it had the government or an insurer negotiate a fixed total price with the manufacturer, treating as many patients as possible without limit over the life of the agreement — turning “pay per head” into “pay by subscription.” The idea was later transplanted into gene therapy, spinning off various installment and annuity-style repayment schemes, all built toward the same goal: breaking an astronomical one-time expense into cash flows an insurance system can actually digest.

In 2024, this logic was, for the first time, moved onto a national-level operating table by the U.S. federal government. The Cell and Gene Therapy (CGT) Access Model, launched by the Centers for Medicare & Medicaid Services (CMS), built a mechanism specifically for sickle cell gene therapies that amounted to centralized negotiation on states’ behalf: CMS itself negotiated standardized outcomes-based agreement templates with Vertex (for Casgevy) and bluebird bio (for Lyfgenia), so that individual state Medicaid programs wouldn’t have to negotiate separately — they could simply draw on manufacturer rebates within the shared framework. The model’s application window closed in May 2024, and negotiations continued through that November; by the stage the program had reached, 33 states plus Washington, D.C. and Puerto Rico had joined, covering 84% of the sickle cell disease patients on Medicaid nationwide — a patient population officially estimated at more than 100,000. This was the first time “paying by outcome” stopped being a private clause between one drugmaker and one insurer, and became an entire piece of government-backed, cross-state public payment infrastructure.

What’s Still Unresolved

Installment payment solves the cash-flow problem, but it doesn’t solve several thornier ones. First, U.S. health insurance is highly fragmented: employers switching coverage plans every few years is the norm, and if a patient pays the first year’s installment through Insurer A this year and then changes jobs — and insurers — next year, there’s still no industry-wide answer for who picks up the remaining four years of the bill. That’s exactly where the logic of a “one-time cure” and the logic of “pay by year” naturally come apart. Second, all of these outcomes-based agreements are betting on an assumption that time hasn’t yet tested: how long an edit or an integrated gene addition will actually last. A few years of follow-up data at launch can’t support the certainty that the phrase “lifelong cure” implies, and if some late complication or waning of effect surfaces years down the road, the refund clause negotiated today may well have already expired. Third, no matter how finely the pricing mechanism is tuned, the threshold of this price list still locks out the vast majority of patients — the disease burden of sickle cell disease is concentrated in sub-Saharan Africa, while a payment system that can absorb two or three million dollars a shot exists in only a handful of wealthy countries. That deeper divide is left for the piece specifically about access.

The numbers on the price list will eventually be revised by more years of real-world data, and installment schemes will keep being iterated. But the two price tags posted on that same day in 2023 have already marked out the question this field has to answer together: now that science can, for the first time, cash the word “cure” out into a single injection, whether finance and insurance can keep up will decide whether that cure ends up as a privilege for the few, or a public good that can be delivered at scale.

References

  1. Frangoul H, et al. Exagamglogene Autotemcel for Severe Sickle Cell Disease. N Engl J Med. 2024;390:1649-1662. DOI
  2. Trusheim MR, Cassidy WM, Bach PB. Alternative State-Level Financing for Hepatitis C Treatment — The “Netflix Model”. JAMA. 2018;320(19):1977-1978. DOI
  3. ICER (Institute for Clinical and Economic Review). Gene Therapies for Sickle Cell Disease: Effectiveness and Value — Final Evidence Report. Published Aug 21, 2023. icer.org / Press release
  4. CMS. Cell and Gene Therapy (CGT) Access Model — Frequently Asked Questions & program overview. cms.gov
  5. CMS Innovation Center. Innovation Insight: CMS Model Delivers Access to Sickle Cell Gene Therapy with Expansive List of State Participants. cms.gov
  6. Sidley Austin LLP. Cell and Gene Therapies Access Model on Sickle Cell Disease Announced by U.S. CMS Innovation Center. Feb 2024. sidley.com
  7. Novartis / AveXis. AveXis Announces Innovative Zolgensma Gene Therapy Access Programs for US Payers and Families. novartis.com
  8. Xtalks. Bluebird Bio’s Skysona Receives FDA Approval and Becomes World’s Most Expensive Drug. xtalks.com
  9. BioWorld. With $2.8M price tag, Bluebird’s Zynteglo first cell-based gene therapy for certain beta-thalassemia patients approved in US. bioworld.com
  10. CNN. FDA approves $3.5 million treatment for hemophilia, now the most expensive drug in the world. Nov 23, 2022. cnn.com
  11. Fierce Pharma. Vertex, CRISPR score landmark FDA approval for sickle cell disease gene therapy Casgevy. fiercepharma.com
  12. BioSpace. bluebird bio Details Plans for the Commercial Launch of LYFGENIA Gene Therapy for Patients Ages 12 and Older with Sickle Cell Disease and a History of Vaso-Occlusive Events. biospace.com