Moderna Phase 3 Melanoma Trial Meets All Primary Endpoints

Moderna黑色素瘤3期临床试验达到所有主要终点 个性化mRNA疫苗联合Keytruda通过解决细胞内T细胞靶向问题，将癌症复发风险减半并预防远处转移。 Moderna首席执行官Stéphane Bancel讨论了个性化mRNA癌症疗法联合Merck旗下Keytruda所取得的积极3期临床试验结果，该疗法在高危黑色素瘤患者中实现了约80%的五年无病生存率。该疗法通过对患者的肿瘤和健康细胞进行测序，将多达34种患者特异性新抗原编码进定制mRNA中，以此训练T细胞精准靶向癌症特征。在克服了过去二十年该行业诸多失败之后，Moderna将LNP直接递送至淋巴结以推动细胞内抗原呈递。展望以2027年为目标的上市计划，该公司目前正将这一平台拓展至肺癌、胰腺癌以及自身免疫性疾病的治疗。 Moderna 与默克（Merck）联合研发的个体化 mRNA 癌症疫苗联合 Keytruda 疗法在针对高危黑色素瘤的 3 期临床试验中取得积极结果，成功达到无复发生存期（RFS）和无远处转移生存期（DMFS）终点。 (2:48) 2 期临床数据显示，该联合疗法相比单用 Keytruda 将复发或死亡风险降低了约 50%，在治疗 5 年后约 80% 的患者仍保持无病状态。 (2:48) 该疗法结合了 Keytruda 激活 T 细胞与个性化 mRNA 疫苗精准引导 T 细胞识别肿瘤靶点的双重机制，弥补了单药治疗中 40% 患者无长期应答且易引发自身免疫疾病的局限。 (5:46) 以往 20 余年间上千项癌症疫苗试验均告失败，主因在于细胞外蛋白递送低效且依赖共享抗原；而肿瘤新抗原中约 90% 因人而异，需借助 LNP 递送 mRNA 至抗原呈递细胞实现内源性表达。 (11:16) 该疫苗通过比对患者肿瘤与正常组织 DNA 识别独特突变，利用算法筛选最多 34 种新抗原编码进单一 mRNA 分子；目前 3 期试验成果仅基于 1.0 版算法，后续将结合未应答者样本数据持续迭代。 (15:09)

## Background 该视频由投资机构 a16z 于 2026 年 9 月 2 日在其官方 YouTube 频道发布。视频简介指出，a16z 普通合伙人 Jorge Conde 专访了 Moderna 首席执行官 Stéphane Bancel，共同探讨 mRNA 技术的重大历史性里程碑——Moderna 与默沙东（Merck）联合研发的个体化黑色素瘤癌症疫苗取得积极的 III 期临床结果。访谈详细剖析了该疗法如何通过全基因组比对锁定关键突变并编码至多 34 个新抗原以再教育免疫系统，探讨了约 42 天交付周期的工程与制造挑战，并展望了从黑色素瘤向肺癌、胰腺癌以及自身免疫疾病的管线延伸。 ## TL;DR Moderna 与默沙东联合推进的个体化 mRNA 肿瘤疫苗在黑色素瘤 III 期临床试验中斩获阳性结果，终结了癌症疫苗领域长达二十年、上千次试验失败的困局，并在随访中将术后复发风险大幅降低。面对不同患者间肿瘤新抗原差异高达 90% 的生物学现实，Moderna 依托专有算法从患者基因测序比对中挑选出最具免疫原性的 34 个突变，并将其编码入单条长 mRNA 中，实现体内原位直接翻译与全新特异性 T 细胞的激活。在工业工程与监管端，团队摒弃质粒发酵，采用无细胞全水相酶促合成技术，将“针头到针头”的生产交付周期压缩至约 42 天，并以工艺类审评（Process BLA）确立了个体化制药的标准路径。基于这一成功范式，Moderna 正在向早期肺癌单药治疗、难治性胰腺癌联合疗法以及罕见遗传病和个体化自身免疫疾病全面铺开管线。 ## Atmosphere & style 整场对谈洋溢着严谨务实的硬核工程思维与激动人心的医学探索氛围，Jorge Conde 与 Stéphane Bancel 间的交流默契高效，跳过了泛泛的概念炒作，直切底层免疫机制与规模化制造的生死线。Bancel 展现出极具辨识度的“工程师掌舵者”特质，既能用“开闸放猛犬与精准训犬师”的比喻深入浅出地拆解联合疗法逻辑，又能在谈及洁净室占地面积与周转周期时展现出近乎偏执的运营敏锐度。对话中既有攻克二十年医学死局后的自豪与释怀，更有坦言当前疫苗仅是“未来医学史上最差一版”的极度清醒，让受众真切感受到前沿科技范式转移时刻的理性力量与生命温度。 ## Bulletin - [02:48](https://www.youtube.com/watch?v=yTNMEYeCgWw&t=168s) Moderna 与默沙东联合开发的个体化 mRNA 癌症疫苗在黑色素瘤 III 期临床试验中取得积极结果，首次中期分析即同时达成主要终点 RFS（无复发生存期）与次要终点 DMFS（无远处转移生存期），终结了癌症疫苗领域长达 20 年、千次试验接连失败的历史。 - [11:48](https://www.youtube.com/watch?v=yTNMEYeCgWw&t=708s) 临床数据显示不同患者间肿瘤新抗原差异率高达约 90%，印证了通用型共享抗原的局限性；Moderna 通过比对患者全基因组测序筛选出最多 34 个特异突变进行单人定制编码。 - [20:17](https://www.youtube.com/watch?v=yTNMEYeCgWw&t=1217s) Moderna 彻底摒弃传统大肠杆菌质粒发酵路线，采用纯水相无细胞酶促催化合成制造体系，已将活检取样到医院交付的针头到针头（needle-to-needle）周期压缩至约 42 天。 - [27:08](https://www.youtube.com/watch?v=yTNMEYeCgWw&t=1628s) 监管路径实现范式革新：FDA 并非针对每次成分不同的单一化学实体进行审批，而是将其作为工艺类生物制品许可（Process BLA）严格审评端到端生产工艺的稳健性与可重复性。 - [30:05](https://www.youtube.com/watch?v=yTNMEYeCgWw&t=1805s) Moderna 正在启动三大扩展方向，包括向非小细胞肺癌、肾癌、膀胱癌延伸，在 I 期早期肺癌开展无免疫检查点联合的 mRNA 单药 III 期临床，并利用能够诱导 de novo 新生 T 细胞的特性攻坚胰腺癌等“冷肿瘤”。 ## Key takeaways - [02:48](https://www.youtube.com/watch?v=yTNMEYeCgWw&t=168s) Moderna 与默沙东的个体化黑色素瘤联合疗法在 III 期临床首次中期分析中同时达到 RFS 和 DMFS 终点，II 期长期随访显示联合 Keytruda 比单药降低约 50% 复发风险且 5 年无病生存率达约 80%。 - [05:46](https://www.youtube.com/watch?v=yTNMEYeCgWw&t=346s) 单用免疫检查点抑制剂（如 Keytruda）仅能带来约 60% 的 5 年无病生存率，且无差别松开刹车易诱发终身性自身免疫病；mRNA 疫苗扮演“精准训犬师”角色，二者机制正交互补。 - [09:23](https://www.youtube.com/watch?v=yTNMEYeCgWw&t=563s) 癌症疫苗虽在术后以治疗手段介入，但其本质是向免疫系统呈递叛变细胞特征以重新校准免疫监视，从而防止肿瘤复发与死灰复燃。 - [11:48](https://www.youtube.com/watch?v=yTNMEYeCgWw&t=708s) 传统重组蛋白疫苗停留在血液中无法激发强效反应，而肌注的 mRNA 能够穿透淋巴结直接进入抗原递呈细胞（APC）内部实现原生翻译呈递，这是技术破局的生物学根基。 - [11:48](https://www.youtube.com/watch?v=yTNMEYeCgWw&t=708s) 基因测序揭示人与人之间肿瘤新抗原重叠率不足 10%，证明行业过去押注“共享抗原”的方向走偏，必须针对单人 34 个特异新抗原进行全定制化合成。 - [15:40](https://www.youtube.com/watch?v=yTNMEYeCgWw&t=940s) 目前取得优异临床数据的抗原预测算法仍是运行了 10 年之久的“1.0 版本”，Moderna 正利用 AI 对那 20% 未响应患者的分子数据进行攻坚，以驱动后续算法升级。 - [20:17](https://www.youtube.com/watch?v=yTNMEYeCgWw&t=1217s) 不同于 CAR-T 必须在体外对活细胞进行扩增重编程，mRNA 是纯信息分子，前端完成基因测序后仅需传输电子数据即可启动工厂无细胞合成。 - [22:08](https://www.youtube.com/watch?v=yTNMEYeCgWw&t=1328s) 端到端“针头到针头”交付周期已达约 42 天，全合成水相酶促反应大幅缩减了反应器体积，使洁净室单位面积产能成倍攀升，单厂即可满足全美黑色素瘤需求。 - [25:35](https://www.youtube.com/watch?v=yTNMEYeCgWw&t=1535s) 针对 N=1 极客自制药现象，工业化生产线通过严格的 cGMP 规范规避内毒素污染与致命脓毒症，以极高安全性和低门槛构筑了规模化制药护城河。 - [27:08](https://www.youtube.com/watch?v=yTNMEYeCgWw&t=1628s) 监管机制从传统审核固定分子式的“产品类 BLA”成功转向审核制造流程一致性的“工艺类 BLA”，核心考察算法与生产线面对相同输入时输出结果的稳健度与重复性。 - [30:05](https://www.youtube.com/watch?v=yTNMEYeCgWw&t=1805s) 患者治疗前后血液检验证实，mRNA 疫苗能够在体内无中生有（de novo）地诱导产生此前不存在的全新肿瘤特异性 T 细胞。 - [32:00](https://www.youtube.com/watch?v=yTNMEYeCgWw&t=1920s) 针对早期癌症患者，Moderna 规划开展 I 期肺癌的 mRNA 单药辅助治疗 III 期试验，凭借普通疫苗级的低毒副作用填补检查点抑制剂无法覆盖的早筛人群。 - [32:00](https://www.youtube.com/watch?v=yTNMEYeCgWw&t=1920s) 针对免疫检查点药物完全无效的胰腺癌与胃癌等“冷肿瘤”，Moderna 探索将其 de novo T 细胞激活机制与新型靶向药物（如 KRAS 抑制剂）联用以打破僵局。 - [35:08](https://www.youtube.com/watch?v=yTNMEYeCgWw&t=2108s) mRNA 技术平台的演进路径正在从大流行传染病扩展至肿瘤学、儿童罕见肝脏遗传病，并进一步进军特异性清除异常免疫细胞的“个体化自身免疫疾病”前沿根治领域。 ## Detailed walkthrough ### [00:00](https://www.youtube.com/watch?v=yTNMEYeCgWw&t=0s) 引言与技术预览：破解癌症疫苗20年困局 在对话开篇的精彩剪辑中，Moderna 首席执行官 Stéphane Bancel 开门见山地指出了这一里程碑事件的分量：这是整个人类医学史上第一次证实“癌症疫苗”切实有效。在过去的20多年里，整个肿瘤免疫学界开展了超过1000次相关临床试验，但几乎全部以失败告终。主持人 Jorge Conde 随即追问，究竟是什么让 mRNA 技术脱颖而出，赋予了免疫系统前所未有的学习能力？ Bancel 解释了其底层的生物学逻辑：人体无时无刻不在产生癌细胞，正常情况下训练有素的免疫系统能够在极早期识别并清除它们；然而一旦肿瘤不受控地长大，免疫系统便已“失明”，此时的核心挑战就是如何“重新教会”免疫系统发起攻击。Moderna 的做法是从患者体内获取肿瘤活检组织，对其 DNA 每一个碱基字母进行全基因组测序，同时采集患者的健康体细胞做相同的测序比对。算法会逐字、逐核苷酸对比两者的差异，精确筛选出最关键的突变特征（新抗原），并将其编码进一条定制的 mRNA 中。当这种个性化疫苗注射入体内后，它能精准教会免疫系统认出此前漏掉的肿瘤专属特征。面对 Jorge 关于“剂量因人而异、药监部门如何审批”的疑问，Bancel 坦言，这正是监管层面最富挑战性的监管创新所在。 --- ### [00:59](https://www.youtube.com/watch?v=yTNMEYeCgWw&t=59s) 从新冠到癌症：Moderna 的技术跃迁 a16z 生物与健康投资合伙人 Jorge Conde 正式开场，热情欢迎 Stéphane Bancel 重返播客。长线听众对两人的对话并不陌生——早在2020年12月全球抗疫最为紧要的关头，Bancel 就曾做客节目，深度复盘了 Moderna 如何以不可思议的工业化速度对新冠病毒基因序列作出响应、迅速“打印”出拯救数百万人的 mRNA 疫苗，当时那期播客被命名为《制造疫苗的机器》（The Machine That Made the Vaccine）。 转眼到了2026年8月，两人再度聚首，是因为 Moderna 在利用 mRNA 攻克癌症领域取得了重大历史性突破。正如 Jorge 所言，Moderna 再次向世界证明，mRNA 绝非一次性的“抗疫工具”，而是一个具备无限延展性的底层工程化平台。在经历了十余年的深耕后，Moderna 与默沙东（Merck）联合推进的个性化黑色素瘤治疗方案终于迎来关键战果。 --- ### [02:48](https://www.youtube.com/watch?v=yTNMEYeCgWw&t=168s) 黑色素瘤三期临床传捷报：首个击败 Keytruda 的新疗法 Bancel 动情地回顾了这一艰辛历程：在历经长达10年针对个性化肿瘤 mRNA 疗法的攻关后，Moderna 与默沙东在前一周正式宣布，双方联合开展的黑色素瘤 III 期临床试验取得了积极结果。这项里程碑创造了多个“行业第一”——不仅是全球首个被证实有效的癌症疫苗，更是人类有史以来首次在黑色素瘤领域找到一种能够显著优于“单用 Keytruda（帕博利珠单抗）”的治疗方案。 令团队倍感惊喜的是中期分析的亮眼数据。该试验不仅顺利达到了无复发生存期（Recurrence-Free Survival, RFS；即延缓癌症复发或死亡）这一主要终点，更在首次中期分析中意外达到了通常需要更长时间成熟的次要终点——无远处转移生存期（Distant Metastasis-Free Survival, DMFS）。为了给听众提供更直观的数据纵深，Bancel 引用了2026年春季美国临床肿瘤学会（ASCO）年会上公布的 II 期临床长期随访结果：在术后接受随访的患者中，联合方案相比单用 Keytruda 降低了约50%的复发风险；在长达5年的随访节点上（肿瘤学界通常将5年无病生存视为临床治愈），接受联合治疗的患者中有高达80%依然保持无病状态（disease-free）。 - **里程碑意义**：终结了癌症疫苗领域20余年、上千次试验连续告败的阴霾。 - **临床终点达成**：在首次中期分析中同时拿下主要终点（RFS）和次要终点（DMFS）。 - **长期治愈信号**：II 期临床5年数据显示约80%患者无病存活，显著优于既往标准治疗。 - **商业化准备**：位于马萨诸塞州的专属超级工厂已准备就绪，公司正与监管机构紧密沟通，力争在2027年让该药物正式获批上市。 --- ### [05:46](https://www.youtube.com/watch?v=yTNMEYeCgWw&t=346s) Keytruda 的局限与“开闸放狗”比喻：为何需要结合 mRNA 疫苗 针对非生物医药背景的听众，Jorge 抛出了一个核心问题：作为当代肿瘤免疫治疗基石的“抗癌神药”Keytruda 究竟是什么？为什么单独使用它远远不够，必须辅以 Moderna 的个体化 mRNA 疫苗？ Bancel 用了一个通俗生动的比喻来化解复杂的免疫生物学概念：免疫检查点抑制剂（如 Keytruda）的作用就像是“打开门栓放猛犬”（open the gates for letting the dogs out），猛犬代表体内被唤醒的 T 细胞，松开刹车后让它们去扑咬癌细胞。当免疫检查点药物奏效时，效果堪称奇迹，但根据已公布的 III 期临床数据，5年后依然无病的患者仅有约 60%。这意味着足足有 40% 的患者即使经历了严苛的治疗，身体依然无法战胜癌症。更痛苦的是，免疫疗法由于无差别激活免疫，其副作用完全不同于传统的化放疗，而是引发严重的自身免疫性疾病——许多患者在接受 Keytruda 或同类药物后，不得不终身承受 1 型糖尿病、系统性红斑狼疮（lupus）或克罗恩病（Crohn's disease）的折磨。“试想一下，那 40% 无应答的患者不仅未能从抗癌中获益，反而白白承受了自身免疫病的摧残，”Bancel 强调。 这也正是 Moderna 在2015至2016年决定与默沙东达成战略合作的初衷。Moderna 在早期传染病疫苗的研发中积累了极其深厚的免疫调控经验，团队意识到，必须建立一种与免疫检查点完全“正交”（orthogonal）、互补的全新机制：Keytruda 负责“松开犬链”，但放出来的狗缺乏辨识方向的能力，甚至可能盲目撕咬健康器官；而 Moderna 的个性化 mRNA 疫苗则在分子层面充当了“训犬师”，精准给 T 细胞嗅闻患者肿瘤专属的新抗原特征，明确指引它攻击癌细胞。 ```mermaid flowchart TD accTitle: 肿瘤免疫联合疗法机制对比 accDescr: 对比单独使用 Keytruda 与结合 Moderna 个性化 mRNA 疫苗的不同作用路径及效果差异 subgraph Baseline["单用 Keytruda (免疫检查点阻断)"] A["松开免疫刹车<br/>('开门放狗')"] --> B["T细胞被无差别释放"] B --> C1["60% 患者长期获益"] B --> C2["40% 无应答 + 自身免疫病风险<br/>(糖尿病、红斑狼疮等)"] end subgraph Combo["联合疗法 (Keytruda + mRNA 个性化疫苗)"] D["肿瘤活检与健康细胞比对"] --> E["算法锁定高达34个核心突变"] E --> F["mRNA 疫苗编码新抗原<br/>('精准训犬')"] F --> G["特异性激活针对肿瘤的 T 细胞"] G --> H["与 Keytruda 协同精准杀灭肿瘤<br/>(5年无病率提升至约 80%)"] end ``` --- ### [09:23](https://www.youtube.com/watch?v=yTNMEYeCgWw&t=563s) 重新定义“疫苗”：从预防病毒到阻止癌症复发 Jorge 将话题转向了概念辨析：在大众传统认知中，“疫苗”（Vaccine）向来指的是在患病前注射以“预防”外来感染（如麻疹或流感）；然而面对这些体内已经确诊肿瘤的患者，Moderna 研发的却是一种治疗性干预手段，它预防的不是癌症的初次发生，而是“癌症的死灰复燃”（return of cancer）。这种机制在疗法层面上同样展现出了极具颠覆性的构想。 Bancel 确认了这一定义的精准性，并解释道，“癌症疫苗”这一术语并非 Moderna 的独创，而是过去二十多年里行业沿用下来的统称。其核心共性在于底层逻辑的一致性——它们都是在“向免疫系统传授新认知”（teaching the immune system）。接种新冠或流感疫苗，是在病毒入侵前让免疫系统提前识破外敌；而癌症个性化疫苗，则是把因肿瘤伪装而未被识别的叛变细胞基因签名，重新翻译、标记并呈递给免疫系统，从而筑牢防止肿瘤卷土重来的防线。 ### [10:27](https://www.youtube.com/watch?v=yTNMEYeCgWw&t=627s) 重新校准免疫系统：为什么术后疗法仍被称为“疫苗” 人体内无时无刻不在产生癌细胞，无论是外界致癌因素的影响，还是细胞正常复制过程中DNA突变产生的错误。健康状态下，人体的免疫系统受过严密训练，能在极早期察觉并清除这些突变细胞。然而，一旦肿瘤成功躲过监视开始生长，核心挑战就变成了如何“重新校准”（retune）免疫系统，让它补上当初漏掉的危险信号。这就是为什么即使该疗法属于肿瘤切除或发病后的治疗性干预（therapeutic treatment），科学界依然将其称为“疫苗”——其本质是对人体免疫系统进行再教育。 面对过去二十年间全行业进行的大约一千次临床试验全部折戟的惨痛历史，Moderna 与默沙东（Merck）的联合研发能够破局，关键在于两项决定性差异：其一是独特的 mRNA 底层递送技术，其二则是坚持“一次只针对一名患者”的极致个体化设计。 ```mermaid flowchart TD accTitle: 传统肿瘤疫苗与Moderna mRNA疫苗免疫递送机制对比 accDescr: 展现外源重组蛋白进入血液循环与mRNA直接进入淋巴结APC细胞从内部递呈抗原的差异 subgraph Traditional["传统技术路径 (大多失败)"] T1["体外反应器制备<br/>重组蛋白 / 多肽"] --> T2["注射进入血液循环"] T2 --> T3["被动接触免疫细胞<br/>(非内源性加工，免疫原性弱)"] end subgraph Moderna["Moderna mRNA 个体化路径"] M1["针对单人定制的<br/>mRNA 分子"] --> M2["肌肉注射递送至淋巴结"] M2 --> M3["直接进入抗原递呈细胞 (APC)"] M3 --> M4["APC 细胞内部原位翻译<br/>并完成高质量抗原递呈"] end ``` ### [11:48](https://www.youtube.com/watch?v=yTNMEYeCgWw&t=708s) mRNA 的内在优势与 90% 的抗原异质性真相 在底层生物学机制上，Moderna 早在 2015 年便与瑞典卡罗林斯卡学院（Karolinska Institutet）合作发表了关键证据：当注射 Moderna 专有的 mRNA 药物（无论是新冠疫苗、流感疫苗还是肿瘤治疗疫苗）至肌肉后，mRNA 分子会顺着淋巴系统进入局部淋巴结，并直接穿透进入核心免疫细胞——抗原递呈细胞（APC, Antigen Presenting Cells）。mRNA 在 APC 内部被直接翻译，由内而外地向免疫系统完成抗原递呈。这与以往在生物反应器中合成的外源多肽或重组蛋白质疫苗有着天壤之别——重组蛋白注入后仅在血液中漂流循环，免疫系统虽能接触，但无法形成像胞内原生表达那样强效、持久的“由内而外”（from within）的抗原识别。 在抗原选择策略上，过去的研发受制于成本与制造工艺，大多试图押注“共享抗原”（shared antigen，即不同患者共有的癌细胞靶点）。但癌症本质上是一种 DNA 疾病，伴随过去二十年测序成本呈断崖式下跌，Moderna 采取了全新的个体化策略： - 对患者手术切除的肿瘤组织进行活检，测定整整 3 GB 的完整基因组数据。 - 提取患者健康细胞进行全基因组测序，开展逐个字母、逐个核苷酸（nucleotide by nucleotide）的严格比对。 - 从数以百计或千计的突变中，依托专有算法筛选出最具有免疫原性的多达 34 个新抗原（neoantigens），串联编入一条完整的长 mRNA 分子中，在约 30 天内完成制造并实施肌内注射。 - 临床试验数据（包括此前在 ASCO 发布的 II 期数据以及一致的 III 期机制）证明了一个此前未知的惊人事实：**不同患者之间肿瘤新抗原的差异率高达 90% 左右**。这意味着人与人之间的重叠抗原微乎其微，唯有个体化定制才有可能真正击中肿瘤。 ### [15:40](https://www.youtube.com/watch?v=yTNMEYeCgWw&t=940s) 算法黑盒与 1.0 版本：为何现在的疫苗在未来是“最差的一版” 在挑选最具免疫杀伤力的 34 个抗原时，Moderna 的算法核心融合了肿瘤学与免疫学的顶尖研究成果、公共数据库、内部积累的专有临床数据，以及与诊断学和细胞治疗公司合作所获取的 T 细胞受体图谱（T-cell mapping）。但值得注意的是，刚刚公布出色 III 期临床数据的这款管线（intismeran），其底层的抗原预测算法实际上与 II 期乃至 I 期完全相同——它是一个已经在实践中运行了 10 年之久的“1.0 版本”算法。 即便采用这套 10 年前的 1.0 算法，在 II 期临床随访中，已有 80% 的高危黑色素瘤患者在 5 年后仍保持无病生存（disease-free）。然而仍有 20% 的患者未见响应。Moderna 团队目前正调取 III 期试验患者的完整血液样本与测序数据，深入挖掘有效响应与无响应背后的分子机理： > “就像我们谈到 AI 时常开的一个玩笑：今天我们看到的 AI，绝对是你这辈子能见到的最差的一版。对于 intismeran 也是一模一样的道理——Moderna 现在拿出的版本，将是整个医学史往后能见到的‘最差的一版’。” 未来随着算法从 1.0 升级到 2.0，甚至有望在更难啃的“冷肿瘤”如胰腺癌（对现有免疫疗法极不敏感）中打开局面。任何算法的迭代未来都将在严格受控的条件下向监管机构（如 FDA）备案并验证，确保在提高应答率的同时不损失既有疗效。 ### [18:12](https://www.youtube.com/watch?v=yTNMEYeCgWw&t=1092s) 工业工程思维：超越 CAR-T 复杂度的制造硬仗 Jorge Conde 与 Stéphane Bancel 相识已久，他谈及 Bancel 骨子里始终是一位“工程师思维”（engineer at heart）的掌舵者，从 Moderna 创办伊始就近乎偏执地追求工艺流程、供应链与运营效率的最大化。因为面对 90% 抗原毫无重叠的现实，要为全球每一个癌症患者量身打造一剂独一无二的药品，工艺与运营的敏捷性是整个商业模式能否成立的生死线。 Jorge Conde 以肿瘤领域的另一大个体化突破 CAR-T 细胞疗法作为参照对比： - **CAR-T 疗法**：提取患者自身的免疫细胞（T 细胞）到体外，在实验室中利用基因工程重新编程其嵌合抗原受体，使其具备识破肿瘤的能力，扩增后再回输到患者体内。其流程涉及繁琐的活细胞体外扩增与严格的冷链质控。 - **个体化 mRNA 疫苗**：同样始于患者活检组织样本，但无需在体外对活细胞进行复杂繁琐的工程改造，而是通过全外显子/全基因组测序、算法快速识别突变密码，并在工业化流水线上将序列转化为专属 mRNA 制剂。在患者体内完成针对多达 34 个靶点的原位免疫训练。如何压缩从活检取样到医院回输的“血管到血管”（vein to vein）时间，构成了商业化扩张中最大的工程壁垒。 ### [20:17](https://www.youtube.com/watch?v=yTNMEYeCgWw&t=1217s) “针头到针头”的生产周期：mRNA 与 CAR-T 的本质差异 谈及个性化癌症疫苗的交付时效，主持人 Jorge Conde 提出了“静脉到静脉”（vein-to-vein）的周期问题：从获取患者肿瘤样本到制成个性化疫苗回输体内，中间需要满足哪些条件？Moderna 首席执行官 Stéphane Bancel 纠正并使用了“针头到针头”（needle-to-needle）的表述。他透露，目前整个周期大约为 42 天——即从穿刺活检查取肿瘤样本，到量身定制的疫苗成品送达医院备用。随着自动化、机器人技术以及流程效率的持续优化，这一时间还将进一步缩减。 Stéphane 重点强调了 mRNA 疫苗与细胞疗法（如 CAR-T）在底层逻辑上的根本不同。CAR-T 疗法必须从患者体内采集免疫细胞，送至工厂的大型生物反应器中进行体外（ex vivo）基因重编程，再冷链运回医院。而 mRNA 的美妙之处在于它本质上是“信息分子”（information molecule）。Moderna 无需物理运输细胞，只需测序实验室发来一个数据文件，其中包含患者健康细胞与癌细胞的基因组序列对比。 ```mermaid flowchart LR accTitle: 从肿瘤信息到个性化 mRNA 疫苗的合成制造流 accDescr: 展示了从基因测序数据获取，到全酶法无细胞合成，再到脂质纳米颗粒包裹并交付医院的端到端生产流程。 subgraph Data["信息输入"] A["健康细胞与癌细胞<br/>测序数据（数据文件）"] end subgraph Synthesis["纯合成酶促反应"] B["无质粒/全合成<br/>DNA 模板制造"] C["体外转录合成 RNA<br/>（微型水相催化）"] D["LNP 脂质纳米颗粒包裹"] end subgraph Delivery["临床交付"] E["个性化疫苗送达医院<br/>（针头到针头约 42 天）"] end A --> B B --> C C --> D D --> E ``` 在制造端，Moderna 完全抛弃了利用大肠杆菌扩增质粒 DNA 的传统生物发酵路径，转而采用全合成、水相中的酶促反应制备 DNA 模板，再转录出 RNA 并用脂质纳米颗粒（LNP）包裹。传统重组生物制药（如 CAR-T）之所以需要巨大体积的生物反应器，是因为活细胞一旦受到过度挤压就会死亡；而 Moderna 的无细胞酶促催化体积微小，反应器极致紧凑，其工程逻辑更偏向化学小分子药物而非传统大分子生物药。 - **42 天交付周期**：活检取样到疫苗送达医院（针头到针头），未来随着自动化升级仍有压缩空间。 - **信息分子特性**：实体样本在前端完成测序后，仅凭电子数据即可启动工厂合成，无需长途活细胞物流。 - **无质粒合成革命**：彻底摆脱大肠杆菌发酵质粒的传统流程，实现全水相酶促催化生产。 --- ### [22:08](https://www.youtube.com/watch?v=yTNMEYeCgWw&t=1328s) 硬件迭代、洁净室平米成本与产能规模化 为了在推进临床前实现“一人一药”的定制生产，Moderna 早在开展 I 期临床试验之前就必须自研微型化制造设备。Stéphane 坦言，第一代机器体型巨大且笨重，外形宛如一台“大型美式冰箱”。当时管理层对工程团队的明确要求是：优先保证产品质量稳定，杜绝临床中出现假阴性，但暂不追求效率。“如果科学机制在临床上根本行不通，花五年时间造一台精美绝伦的优化机器人没有任何意义；但如果机器不稳定导致试验失败，把有效的科学当成无效，那将是对人类的灾难。” 当他们拿到显示出明确疗效的 II 期中期临床数据（现已拥有随访长达 5 年的优异疗效持久性数据）后，管理层立即调集顶尖工程力量重塑设备。Stéphane 戏称自己视察马萨诸塞州马尔堡（Marlborough）工厂时是个“讨人嫌的监工”，因为他脑子里全被两个指标占据：周转周期（Cycle Time）和占地面积（Square Inches）。在洁净室空间固定的前提下，若能将机器体积大幅压缩，使单位面积内容纳的生产机组提升 2 倍甚至 10 倍，并把计算服务器等外围硬件移出洁净室，就能成倍拉高产能通量，从而大幅摊薄固定资产折旧与制造公摊成本。 在被问及产能与药物销售成本（COGS）时，Stéphane 解释道： - **现行与规划产能**：伴随正在开展的 9 项临床试验，Moderna 已经累计生产了数千剂个性化疫苗；马尔堡工厂的设计产能可达数万剂，伴随技术迭代通量还将继续攀升，单厂即可完全覆盖黑色素瘤市场的用药需求。 - **商业定价与 COGS**：目前尚未公开披露。Moderna 将与合作伙伴默沙东（Merck）一道在公布关键临床数据后，基于临床创造的真实价值同医保及支付方（Payers）展开深入沟通。 --- ### [25:35](https://www.youtube.com/watch?v=yTNMEYeCgWw&t=1535s) N=1 极客自救 vs 工业化制药的安全性护城河 Jorge 提到 GitLab 创始人 Sid Sijbrandij 面对自身骨肉瘤时开启“创始人模式”（Founder Mode）自研个性化疗法的著名案例，并询问：未来这种 N=1 的自救模式是否会并行泛滥，还是此类患者最终都会走向 Moderna？ Stéphane 明确表示，绝大多数患者最终一定会选择 Moderna 这样的工业化平台，核心原因在于“安全”与“便利”。注射类生物制剂的生产极度严苛，存在严重的杂质与污染风险。哪怕药剂中混入单个细菌拷贝，注入患者体内就极易诱发致命的脓毒症（Sepsis）；多步骤手工合成极易出现人工差错。而通过 FDA 严格的 cGMP（动态药品生产质量管理规范）验证的工业化生产线，能够从根本上规避这些致命缺陷。 Stéphane 打了一个生动的比喻：“就像原始人身处没有刀具店的山洞、为了喂饱家人不得不亲手磨制第一把石刀一样；但当街角开起了一家售卖高品质刀具的现成商店，作为人类，你自然会把宝贵的时间省下来去做别的事情。”规模化工业制造一旦将高品质标准品的门槛打下来，个人手工造药的必要性便不复存在。 --- ### [27:08](https://www.youtube.com/watch?v=yTNMEYeCgWw&t=1628s) 监管路径重构：从批准“特定产品”到审评“生产工艺” 针对“每一剂药成分各不相同，监管部门究竟在审评什么”的核心疑虑，Stéphane 详细拆解了个性化药物的监管科学逻辑。他指出，监管先例早已存在——CAR-T 疗法也是依据工艺管线获批，Moderna 的个体化新抗原疗法（INT）同样是作为“工艺类生物制品许可申请”（Process BLA），而非传统单一化合物的“产品类 BLA”（Product BLA）。 Moderna 此前已有 5 款产品按传统产品路径获批上市，但在该项目上，早从最初提交试验性新药申请（IND）起，走的就是工艺类审查。过去十多年间，Moderna 始终与 FDA 保持常态化的高频技术沟通： 1. **全流程沟通机制**：从评估早期进入临床的安全可控性，到每个 III 期临床试验前夕召开 Phase 2 结束会议（End-of-Phase 2 Meetings），双方就试验设计、伴随算法及生产规程提前达成严密共识。 2. **监管的核心关切**：面对技术盲盒，FDA 最关切的底层逻辑在于工艺的稳健性（Robustness）——即如果在黑盒输入端给予完全相同的肿瘤活检与健康血液样本，整个自动化生产系统在输出端是否能够绝对可重复地制造出完全相同质量规格的个性化制剂。 向监管机构充分证明了从样本数据输入到最终制剂输出的系统确定性与可重复性后，一条针对个性化“一人一药”的现代监管通道已经完全成型。随后，Jorge 顺势将话题引向黑色素瘤之外的未来版图：mRNA 癌症疫苗究竟能攻克多大范围的癌种？Stéphane 随即给出了极具雄心的回应。 ### [30:05](https://www.youtube.com/watch?v=yTNMEYeCgWw&t=1805s) 证明新生 T 细胞的产生与肿瘤管线的三大拓展方向 Stéphane Bancel 强调，Moderna 已经向临床学术界和他们自己确凿证实了一件事：mRNA 技术不仅能扩增现有的 T 细胞，更能“无中生有”（de novo）地在人体内激发出此前根本不存在的全新 T 细胞，专门识别 mRNA 所编码的新抗原并精准靶向肿瘤。研究团队通过检测黑色素瘤患者治疗前后的血液样本，直接拿到了这一免疫学证据。以此为根基，Moderna 在黑色素瘤（入组覆盖 II 期、III 期及 IV 期患者）之后，确立了向外拓展的三条清晰战略方向： ```mermaid flowchart TD accTitle: mRNA癌症疫苗临床拓展三大路径 accDescr: 展示Moderna基于mRNA新抗原疫苗扩展的三大临床方向，包括与PD-1协同、早期单药治疗及冷肿瘤联合用药。 subgraph S1["方向一：协同推进"] A["Keytruda 获批且有效的癌种"] --> A1["非小细胞肺癌 (Phase 3)"] A --> A2["肾癌 / 膀胱癌 (Phase 2)"] end subgraph S2["方向二：疾病早期推进"] B["免疫检查点不适用的极早期"] --> B1["I期肺癌单药治疗 (无检查点)"] end subgraph S3["方向三：攻坚冷肿瘤"] C["检查点药物无效的癌种"] --> C1["胰腺癌 / 胃癌"] C1 --> C2["探索与其他靶向药联合 (如 KRAS)"] end ``` - **方向一：进军 Keytruda（帕博利珠单抗）已有显著疗效的适应症**。由于 PD-1 抑制剂与个性化 mRNA 肿瘤疫苗的作用机理完全正交（Orthogonal），二者结合具备天然的协同增效潜力。目前非小细胞肺癌已进入临床 III 期，肾癌和膀胱癌处于临床 II 期。虽然具体疗效提升比例（是如黑色素瘤 II 期中的 50%，还是 30% 到 40%）必须等待试验数据揭晓，但整体管线推进极其迅猛。 - **方向二：布局免疫检查点药物因毒副反应无法覆盖的早期病程**。对于极早期的癌症，患者若使用免疫检查点抑制剂，常面临终身性自身免疫疾病等严重副作用，因此常规临床更倾向于仅进行术后观察。 - **方向三：攻坚免疫检查点抑制剂完全无效的“冷肿瘤”**。胰腺癌与胃癌在过往临床中对 Keytruda 这类免疫检查点药物几乎完全无反应，临床试验均以阴性告终。但鉴于 mRNA 能够生成全新的 de novo T 细胞，Moderna 决定主动挑战这类难治性癌种。 ### [32:00](https://www.youtube.com/watch?v=yTNMEYeCgWw&t=1920s) 早期肺癌单药治疗构想与难治性癌种的联合策略 在谈到疾病早期的单药治疗时，Stéphane 披露了一项重大临床动作：Moderna 计划启动一项针对 I 期肺癌患者的临床 III 期试验，完全作为单药疗法（Monotherapy）使用，不联合任何免疫检查点药物。针对前吸烟者等高危人群，通过定期的胸部低剂量 X 光或 CT 筛查，很容易在 I 期捕捉到肺部肿瘤。患者在接受常规手术切除肿瘤组织后，仅接种为其定制的 mRNA 疫苗；其副作用如同常规普通疫苗一样轻微，通常患者仅会感到约一天的疲劳，“在癌症治疗领域，这可以说是非常理想的副作用表现”。这种极佳的安全性特征，使得它能够填补检查点抑制剂无法进入的早期辅助治疗空白。 针对胰腺癌这类棘手癌种，Stéphane 展现出了极大的科研热情。既然其作用机理不同于检查点药物，就有理由去尝试单药以及新型靶向药物的联合。他特别提到，Revolution Medicines 刚在基于 KRAS 突变的胰腺癌靶向治疗上取得了重大突破并展现出优异的生存期改善；将个性化 mRNA 疫苗与此类靶向药物联合，机制正交且互补，“从科学逻辑上讲完全行得通”。Moderna 期待能够尽快从胰腺癌临床试验中观察到积极信号，从而探索这种强强联合的疗法。 ### [35:08](https://www.youtube.com/watch?v=yTNMEYeCgWw&t=2108s) 算法迭代、罕见病突破与自身免疫疾病的新前沿 面对当前在黑色素瘤取得的佳绩，Stéphane 保持着清醒与紧迫感，指出目前的产品只是“1.0 版本”。他正在严厉督促研发团队跳出传统框架，利用 AI 全面挖掘海量的临床数据。他的关注核心不是成功的案例，而是失败的少数：“我希望团队能够极度专注于那 20% 在五年随访中依然没有响应的患者。因为你永远能从失败中汲取比成功更多的教训。”弄清楚这部分患者为何无效，将反哺算法与技术平台的持续微调升级。 Jorge Conde 感慨，Moderna 将一个最初并非专门为大流行设计的平台迅速投入应对 COVID-19、挽救了千百万人，如今又将其成功带回最初深耕的肿瘤领域。Stéphane 顺势展望了技术的下一站： - **罕见遗传病**：针对儿童罕见肝脏遗传病的晚期关键性临床试验（Pivotal Study）预计年内读出数据；前期 I/II 期试验中，患儿接受药物治疗长达三年，生存和身体状况表现极佳。 - **自身免疫性疾病的根治探索**：Moderna 在年度科学日（Science Day）宣布将 mRNA 平台引向自身免疫疾病。不同于大型药企长期以来仅针对下游症状的对症治疗，Moderna 意图直击根本病因。 - **通用型与个体化双轨并行**：除了研发普适性的自身免疫调控分子，实验室中最令 Stéphane 兴奋的前沿方向是“个体化自身免疫疗法”——精准识别并特异性清除那些将自身组织误判为外来抗原的异常免疫细胞，从源头上切除免疫失序的根源。 节目最后，Jorge 赞叹从传染病到癌症再到自身免疫疾病的技术飞跃，并向 Stéphane 发出邀请，期待未来在自身免疫取得突破时再度做客录制“三部曲”的终篇完结章。 ## Notable quotes > [05:46](https://www.youtube.com/watch?v=yTNMEYeCgWw&t=346s) 打开门栓放猛犬 — Stéphane Bancel > [05:46](https://www.youtube.com/watch?v=yTNMEYeCgWw&t=346s) 试想一下，那 40% 无应答的患者不仅未能从抗癌中获益，反而白白承受了自身免疫病的摧残 — Stéphane Bancel > [15:40](https://www.youtube.com/watch?v=yTNMEYeCgWw&t=940s) 就像我们谈到 AI 时常开的一个玩笑：今天我们看到的 AI，绝对是你这辈子能见到的最差的一版。对于 intismeran 也是一模一样的道理——Moderna 现在拿出的版本，将是整个医学史往后能见到的‘最差的一版’。 — Stéphane Bancel > [22:08](https://www.youtube.com/watch?v=yTNMEYeCgWw&t=1328s) 如果科学机制在临床上根本行不通，花五年时间造一台精美绝伦的优化机器人没有任何意义；但如果机器不稳定导致试验失败，把有效的科学当成无效，那将是对人类的灾难。 — Stéphane Bancel > [25:35](https://www.youtube.com/watch?v=yTNMEYeCgWw&t=1535s) 就像原始人身处没有刀具店的山洞、为了喂饱家人不得不亲手磨制第一把石刀一样；但当街角开起了一家售卖高品质刀具的现成商店，作为人类，你自然会把宝贵的时间省下来去做别的事情。 — Stéphane Bancel > [32:00](https://www.youtube.com/watch?v=yTNMEYeCgWw&t=1920s) 在癌症治疗领域，这可以说是非常理想的副作用表现 — Stéphane Bancel > [35:08](https://www.youtube.com/watch?v=yTNMEYeCgWw&t=2108s) 我希望团队能够极度专注于那 20% 在五年随访中依然没有响应的患者。因为你永远能从失败中汲取比成功更多的教训。 — Stéphane Bancel ## Mentioned Stéphane Bancel, Jorge Conde, Moderna, 默沙东 (Merck), Keytruda (帕博利珠单抗), intismeran, a16z, FDA, 卡罗林斯卡学院 (Karolinska Institutet), ASCO (美国临床肿瘤学会), CAR-T, Sid Sijbrandij, GitLab, Revolution Medicines ## Video description a16z General Partner Jorge Conde sits down with Moderna CEO Stéphane Bancel to discuss a major milestone for mRNA technology: positive Phase 3 results from Moderna and Merck’s individualized treatment for melanoma, after more than a decade of work on personalized cancer vaccines. Stéphane explains how the treatment works by sequencing an individual patient’s tumor and healthy cells, identifying the mutations most relevant to their cancer, and encoding up to 34 of them into an mRNA designed specifically for that patient. Rather than simply unleashing the immune system, the goal is to teach it exactly what to recognize and attack. They also unpack the engineering challenge of manufacturing a different medicine for every patient, how Moderna has brought the process down to roughly 42 days from biopsy to treatment, and what it would take to manufacture personalized medicines at scale. Finally, Stéphane looks beyond melanoma to lung, kidney, bladder, pancreatic, and gastric cancers, as well as Moderna’s longer-term work applying mRNA to rare genetic and autoimmune diseases. Timestamps: 00:00 - Intro 00:59 - How Moderna Went From COVID to Cancer 02:48 - The Phase 3 News: mRNA Cancer Vaccine + Keytruda in Melanoma 05:46 - What Is Keytruda & Why Isn't It Enough Alone? 09:23 - Preventing the Return of Cancer: What "Vaccine" Really Means Here 11:16 - Why 20 Years and 1,000 Trials Failed Before mRNA Worked 15:09 - Inside the Algorithm: How Moderna Picks the 34 Neoantigens 18:48 - The Operational Lift of Making a Vaccine for Every Patient 20:16 - Vein to Vein: How Fast Can You Deliver a Personalized Vaccine? 24:33 - Manufacturing at Scale & the COGS Question 27:08 - How Do You Regulate a Medicine That's Different Every Time? 29:37 - Beyond Melanoma: Which Cancers Are Next? 35:58 - The Full Arc: From COVID to Cancer to Autoimmune Disease Resources: Follow Stéphane Bancel on LinkedIn: https://www.linkedin.com/in/st%C3%A9phane-bancel-8185251/ Follow Jorge Conde on X: https://x.com/JorgeCondeBio Stay Updated: If you enjoyed this episode, be sure to like, subscribe, and share with your friends! Find a16z on X: https://twitter.com/a16z Find a16z on LinkedIn: https://www.linkedin.com/company/a16z Listen to the a16z Show on Spotify: https://open.spotify.com/show/5bC65RDvs3oxnLyqqvkUYX Listen to the a16z Show on Apple Podcasts: https://podcasts.apple.com/us/podcast/a16z-podcast/id842818711 Follow our host: https://x.com/eriktorenberg Please note that the content here is for informational purposes only; should NOT be taken as legal, business, tax, or investment advice or be used to evaluate any investment or security; and is not directed at any investors or potential investors in any a16z fund. a16z and its affiliates may maintain investments in the companies discussed. For more details please see http://a16z.com/disclosures. [Source](https://www.youtube.com/watch?v=yTNMEYeCgWw)

[5:46] Stéphane Bancel: 打开门栓放猛犬 [5:46] Stéphane Bancel: 试想一下，那 40% 无应答的患者不仅未能从抗癌中获益，反而白白承受了自身免疫病的摧残 [15:40] Stéphane Bancel: 就像我们谈到 AI 时常开的一个玩笑：今天我们看到的 AI，绝对是你这辈子能见到的最差的一版。对于 intismeran 也是一模一样的道理——Moderna 现在拿出的版本，将是整个医学史往后能见到的‘最差的一版’。 [22:08] Stéphane Bancel: 如果科学机制在临床上根本行不通，花五年时间造一台精美绝伦的优化机器人没有任何意义；但如果机器不稳定导致试验失败，把有效的科学当成无效，那将是对人类的灾难。 [25:35] Stéphane Bancel: 就像原始人身处没有刀具店的山洞、为了喂饱家人不得不亲手磨制第一把石刀一样；但当街角开起了一家售卖高品质刀具的现成商店，作为人类，你自然会把宝贵的时间省下来去做别的事情。 [32:00] Stéphane Bancel: 在癌症治疗领域，这可以说是非常理想的副作用表现 [35:08] Stéphane Bancel: 我希望团队能够极度专注于那 20% 在五年随访中依然没有响应的患者。因为你永远能从失败中汲取比成功更多的教训。

[2:48] Stéphane Bancel: It's a first of many—it's the first time that there is an agent in melanoma that is better than Keytruda alone. [2:48] Stéphane Bancel: Five years after treatment... in oncology five years is considered by doctors like a cure, and so it's a big, big deal. [5:46] Stéphane Bancel: Keytruda basically unleashes the dogs, but they go a bit randomly... whereas Moderna's product is able at the molecular level inside your immune system to very specifically teach your T-cell: this is what you need to look for

[0:00] It's the first time there is a cancer [0:02] vaccine working. The field has been [0:03] doing that for 20 plus years, more than [0:05] a thousand clinical trials that have all [0:07] failed. [0:08] >> What was different this time? What is it [0:10] about mRNA technology that enables the [0:13] immune system to learn in a way that [0:15] other approaches were unable? [0:17] >> We all have cancer cells all the time in [0:19] our body. Our immune system is very well [0:21] trained. Basically notice those cancer [0:23] cells very early and get rid of them. [0:25] But if your cancer grows, then the [0:28] question is how can you retach immune [0:29] system? We're going to basically take a [0:31] biopsy of your tumor. We're going to [0:33] read all the letters of its DNA. And [0:34] then we're going to do the same things [0:36] with a healthy cell of your body. And [0:37] we're going to literally compare letter [0:39] by letter, nucleotide by nucleotide. And [0:41] then we're going to use an algorithm to [0:43] identify which one of those mutations [0:45] are the most relevant. And so when this [0:47] is injected in your body, it teaches the [0:48] immune system the signature of your [0:50] cancer cell that it missed. [0:52] >> What gets regulated here? Because every [0:54] dose is different. So obviously every [0:55] dose doesn't get approved. [0:57] >> Busy feet. [0:59] >> Hi. Um, welcome to the A16Z podcast. Uh, [1:03] I'm Horge Kai, a general partner on the [1:06] A6Z bion health team. I am thrilled [1:08] today to welcome back Madna CEO Stfan [1:12] Bonsel. Um, for folks that have been [1:15] longtime listeners may recall, uh, Stfan [1:18] joined us on the A6Z podcast back in [1:21] December of 2020. [1:23] um when we were talking about all of the [1:26] work Madna did to uh bring us the the [1:30] mRNA COVID vaccine and at the time um if [1:35] you can go back and listen to that [1:36] episode you'll hear um how quickly um [1:40] Madna was able to react to the existence [1:44] of the virus to analyze it uh to [1:47] essentially print a vaccine um that [1:50] would protect people against an uh [1:53] against the the the co 19 uh vaccine and [1:56] hence put us in a much better position [1:58] uh with respect to the to the co [2:00] pandemic. We titled that episode the the [2:03] machine that made the vaccine and the [2:06] reason why we're talking again today in [2:08] August 2026 [2:11] uh is because Madna has uh put out news [2:15] uh on a really big advancement on what [2:18] they can do with mRNA technology when it [2:20] comes to cancer. And so I want to hand [2:23] it over to you uh Stefan uh again a very [2:27] big welcome. Thank you for coming back. [2:29] But maybe the place to start is let's uh [2:31] let's lead with the news. Um you uh [2:34] Madna and Merc uh announced uh earlier [2:37] this month in August that uh you had uh [2:41] uh conducted a phase three clinical [2:43] trial in melanoma and have gotten uh [2:46] very encouraging results. So why don't I [2:49] hand it over to you and tell us uh what [2:51] have you announced? What have you seen [2:52] in this phase three trial and then I do [2:54] want to dig in into what Madna is doing [2:57] in cancer. [2:58] >> Wonderful. So Roy, thank you so much for [3:00] having us back. We're very happy to be [3:02] to be with you. So indeed uh last week [3:05] now uh we shared the news with our [3:08] colleagues at Merc that after working 10 [3:10] years on a individualized uh treatment [3:14] against cancer using MR&A technology [3:17] that the phase 3 was positive it's a [3:20] it's a first of many it's a first time [3:22] that there is an agent in melanoma that [3:25] is better than kudra alone so it's a big [3:28] deal for patient of course with melanoma [3:31] it's the first time there is cancer the [3:33] vaccine working. Uh the field as you [3:35] know has been doing that for 20 plus [3:37] years. I think more than a thousand [3:39] clinical trials that have all failed. Um [3:43] and um if you look at the phase two data [3:46] because what we announced last week is [3:49] that we met the primary endpoint of the [3:53] study which was recurrence free survival [3:55] i.e. people having the cancer back or [3:57] dying [3:58] >> that we met and then to our own surprise [4:01] because this was the first interim [4:03] analysis. This is not the end of a [4:05] study. It's a first interal of a study [4:08] and what was a surprise even to us is we [4:10] met the secondary endpoint which was [4:13] distant metastasis free survival which [4:16] is of course takes more time to mature [4:18] because it means that you have distance [4:20] metastasis from your primary tumor and [4:22] we also made that uh endpoint which was [4:25] great again unexpected from our side but [4:28] again means it is really good. Um we [4:31] will share the data very soon at a big [4:33] medical oncology conference which is as [4:35] you know what the field does. But just [4:37] to give you a sense maybe to orient kind [4:39] of directionally we showed at ASCO the [4:42] big you know oncology conference in the [4:45] spring of 2026 a few months ago that our [4:48] phase 2 study which was also randomized [4:50] against ketra alone. So the same type of [4:53] study showed that around 50% of people [4:59] had recurrence free survival versus [5:01] people only getting kudra and this is 5 [5:03] years after treatment and as you know [5:06] incology five years is considered by [5:08] doctor like a cure and so it's a big big [5:11] deal and if you look at the data from [5:13] organ that phase two you had around 80% [5:15] of people that were disease free five [5:18] years after their treatment and their [5:20] surgery of melanoma [5:22] And so we're very excited about what it [5:23] means for the field. Uh we are working [5:26] very hard already with regulators to [5:28] file so that the drug can be available [5:29] to patient as soon as possible and we [5:31] hope in 2027 and we're going to try to [5:34] make it as fast as we can. Uh the [5:36] factory is ready in Massachusetts but we [5:38] have figured out how to make and scale a [5:42] product for every human being at the at [5:44] the time. [5:45] >> Well, there's I mean there's so much to [5:47] to unpack here. So I um this is an [5:50] extraordinary uh advance uh for the the [5:53] treatment of in this case melanoma and [5:55] hopefully over time cancers more [5:57] broadly. Uh so let's unpack some of this [5:59] if we could Stefan. Uh the first one um [6:02] uh uh talk to us uh for folks that are [6:04] less familiar with cancer treatments. Uh [6:07] why is Kuda what is Kruda and why is [6:10] Kruda alone not sufficient? Why was [6:13] Madna necess or the the the this [6:16] particular mRNA cancer vaccine necessary [6:18] in conjunction with Kruda? [6:20] >> Sure. So if you look at Kudra which is [6:21] kind of one of the leading imunotherapy [6:24] that people might have heard of is [6:26] basically if I oversimplify for [6:28] nonbiologist it's basically a molecule [6:31] that basically open the gates for [6:33] letting the dogs out that are going to [6:35] go and attack your cancer from your [6:38] immune system if you want. Um the [6:42] the thing about checkpoints is when they [6:44] work they're fantastic because those [6:46] people 5 years after treatment are in [6:49] the cured in the sense they have no [6:52] remaining disease. Uh but only 60% of [6:56] people are disease free after five years [6:59] if you look at the phase three published [7:01] data of kudra. So again for those 60% of [7:04] people it's amazing but it means that [7:06] there's 40% of people where you go [7:08] through the treatment you're fighting [7:10] for cancer and the treatment doesn't [7:12] really work for you and also what is [7:14] difficult is those treatments are [7:16] wonderful but they a lot of time come [7:18] with very serious side effects very [7:20] different from chemotherapy or [7:22] radiotherapy the side effect of [7:24] imunotherapy are mostly immune disease [7:27] >> so you see people if you look at the [7:29] label or the clinical studies people [7:31] that get checkpoints Whether it's Kudra [7:33] or any other type of checkpoints from [7:35] other companies, they end up having type [7:37] 1 diabetes, lupus, chron disease, these [7:40] type of things. Of course, it's better [7:41] to have those disease than of course [7:43] being dead from your cancer, which is [7:45] why those have become standard of care. [7:47] But think about those 40% of people who [7:50] don't respond to checkpoint. They get an [7:52] autoimmune disease most of the time and [7:54] they don't get the benefit of a [7:55] medicine. And so what we try to do with [7:57] Merc and is really a technology that [7:59] emerged from Mona Labs back in 2015 [8:03] 2016 when we did a partnership with Merc [8:05] as we're looking for one of the best [8:07] company in imotherapy to partner and to [8:10] have a complement approach the idea that [8:12] we have at the time using our infectious [8:14] disease learning from our infectious [8:17] disease vaccine. We learn a lot about [8:18] the immune system and how mRNA interact [8:21] with the immune system. We thought we [8:23] could develop a mechanism of action that [8:27] was totally orthogonal. So very [8:28] different from imunotherapy because you [8:31] will be able to start from a sequence of [8:33] your [8:35] tumor. [8:36] >> Mhm. [8:37] >> So that we could design a product to [8:39] basically teach if I go back to the dog [8:40] analogy to teach those dogs what to look [8:44] for very specifically. And that's really [8:47] the the beauty of about our technology [8:49] is if you think about K basically [8:51] unleash the dogs but they go a bit [8:53] randomly sometime it's your immune [8:55] system whereas Mona product is able at [8:59] the molecular level inside your immune [9:01] system to very specifically teach your [9:04] T- cell this is what you need to look [9:05] for and that thing is actually on your [9:08] cancer cell so that those tea cells go [9:11] and basically attack your cancer cells. [9:14] >> Okay. Okay. So now that that's where the [9:16] the personalized cancer fixing um uh [9:19] technology will make a big impact on the [9:22] treatment of melanoma. So that we'll [9:24] come back to the personalized piece [9:25] because I think that's just fascinating [9:27] not only from a technological standpoint [9:29] but just from an operational standpoint. [9:31] Uh so I do want to come back to that but [9:32] let's focus on the word vaccine. So you [9:35] know you know typically when you think [9:37] of the word vaccine uh it is to prevent [9:39] a disease. Um here in this context uh [9:43] these patients already have cancer. So [9:45] what in a sense you are preventing [9:47] something. You're not preventing the [9:48] cancer. What you're preventing is the [9:50] return of the cancer. Um which is in and [9:54] of itself an extraordinary thing to [9:56] think about from a from a therapeutic [9:58] intervention perspective. Um so number [10:01] one is that is that a fair [10:02] characterization of how [10:03] >> characterization and the reason the [10:05] field has used the word vaccine. It was [10:07] not us using it. who just follow the [10:08] field as I mentioned there been a [10:10] thousand plus clinical trial is because [10:12] it's about teaching your immune system [10:14] if you think about a COVID or flu shot [10:17] you teach your immune system before you [10:19] get the virus infecting your body here [10:22] you teach your immune system not about [10:24] the virus but about basically the cancer [10:27] signal that your immune system missed [10:30] because what we know today in the field [10:32] is that we all have cancer cells all the [10:34] time in our body [10:36] >> whether it's outside factors or Just as [10:37] you have cell replication and you have [10:39] mistakes that happen that create [10:41] mutation of DNA that are cancer cells, [10:44] our immune system is very well trained [10:46] to basically notice those cancer cells [10:49] very early and get rid of them. But if [10:53] your cancer grows then the question is [10:56] how can you retune system? So I think [10:57] that's why the field use the word [10:58] vaccine for that approach as well. Even [11:00] as you said it's a therapeutic treatment [11:02] approach post cancer it's about the [11:05] teaching of the immune system. [11:07] Okay. And so as you point out, the field [11:09] has tried this many times before. A [11:11] thousand on the order of a thousand [11:13] clinical trials um have have tested this [11:16] theory. They've all failed. Uh you and [11:19] Madna and Merc have succeeded here. Um [11:23] what was different this time? What is it [11:25] about mRNA technology that enables the [11:28] immune system to learn in a way that [11:30] other approaches uh were unable uh uh to [11:34] be successful? Yeah, I think there's [11:35] there are two components. Uh, one is the [11:37] amount technology and I think the other [11:40] one is the individualization by [11:43] designing a product for one human at a [11:45] time. So, let me go through those two. [11:48] On the technology of MRNA side, what we [11:51] have known and published [11:54] uh actually with Kolinska as you know [11:56] the institute in Sweden that give the [11:58] novel brao medicine back in 2015 is that [12:01] with all technology I cannot speak about [12:03] other mRNA companies that have different [12:05] mas different lipids and so on but with [12:07] all technology when we inject our mRNA [12:09] in a muscle whether it's a co shot or [12:12] flu shot or this cancer treatment [12:15] basically the mRNA goes down to your [12:17] lymph node [12:19] and enters the APCs the antigen [12:21] presenting cell which as you know are [12:23] key component of your immune cells and [12:26] the mRNA gets inside the APCs. This [12:28] we've demonstrated and proven at the [12:31] time of Kolinskar and then it basically [12:34] translate the message contained in the [12:35] mRNA inside the APCs inside your immune [12:38] cells and presented from within. So we [12:42] think it's a very important [12:43] differentiation from most of the [12:45] previous vaccine in cancer in the field [12:47] that were made by protein or peptide [12:50] that basically are made in reactors are [12:51] injected in the in the in the patient [12:54] but I basically turn into the blood [12:56] because as you know recombinant or [12:57] protein when you inject them they just [12:59] go into your blood and they turn around [13:00] so your immune system sees them but not [13:03] in the same manner from within as it [13:06] done with the mRNA. So we think that's [13:08] one very important component of uh [13:11] immune presentation if that makes sense. [13:13] The other component is really [13:14] individualization. In the past a lot of [13:17] time people have tried with non mRNA I [13:19] protein technology or peptide but also [13:22] we tried a shared antigen [13:24] whereas here what we said is because DNA [13:28] is a disease sorry cancer is a disease [13:30] of DNA. [13:31] >> Mhm. What we're going to do here because [13:33] the cost of sequencing have dropped so [13:35] much in the last 20 years is we're going [13:38] to basically take a biopsy of your [13:40] tumor. We're going to read all the [13:42] letters of its DNA, the 3 gigabyte of of [13:46] these genes. And then we're going to do [13:48] the same things with the healthy cell of [13:49] your body. And we're going to literally [13:51] compare letter by letter nucleotide by [13:54] nucleotide. And then we're going to use [13:56] an algorithm to identify over your [13:59] hundreds or thousands of mutation which [14:01] one based on the current knowledge of a [14:03] field of imunology and cancer which one [14:05] of those mutations are the most [14:07] relevant. We select the 34 that we [14:10] believe are the most relevant and we [14:11] stitch them together into one big mRNA [14:13] molecule which we make in 30ish days for [14:16] you that is injected then in a hospital [14:19] intramuscularly. And so when this is [14:21] injected in your body, basically it [14:23] teaches the immune system the signature [14:25] of your cancer cell that it missed. Not [14:28] the signature of every other patient [14:30] with a shared antigen, but the very [14:33] specific signature of your cancer cell. [14:35] And what we we showed at ASCO and we [14:37] published from a phase 2 study, but we [14:39] believe it's the same thing in the phase [14:40] three because it's mechanistic, is that [14:43] around 990% [14:45] of the antigen are different patient to [14:48] patient. [14:49] Because when we started we had no idea [14:51] because again the field came from shared [14:52] antigen. So when we started like we have [14:54] no idea we're going to get 2% 5% 90% of [14:57] of same antigen across all the patient [14:59] actually 90% of antigen are different [15:03] from a human to another one. [15:05] >> Wow. So the only way this can work is [15:07] through personalization individualiz. [15:09] Exactly. [15:09] >> You know assuming that that carries [15:11] over. Um, and so in that regard, uh, uh, [15:14] you know, if you're doing, let's say, [15:16] this normal to tumor comparison of of of [15:19] the genome, you find the differences. [15:21] I'm curious how you arrive at 34 is up [15:24] to 34 as the right number. Um, I'm sure [15:27] there's a very good reason for that. Um, [15:29] but what is the algorithm that enables [15:31] you to do that? Is this something that's [15:33] proprietary to Madna? Is this something [15:35] that is known within the field? Help us [15:37] understand like help us look into that [15:39] black box. [15:40] >> Sure. So I want to share a little bit [15:41] about the blackbox not too much because [15:43] there's a lot of knowhow and and things [15:45] are very confidential to us but [15:47] basically we started with of course what [15:49] is known in the field and so we [15:51] basically use a lot of database and [15:53] publication and a lot of scientists and [15:56] doctors kind of bestin-class in [15:58] iminology and in oncology and then from [16:02] that starting point we use a lot of [16:04] internal data that we generated over [16:06] time we also partner with some companies [16:08] that because they are having the [16:09] diagnostics space or they are let's say [16:12] cell therapy and other space in ocology [16:14] had access to a lot of data a lot of T- [16:16] cell mapping and so on that were very [16:18] useful for the learning the the thing [16:21] that is interesting about the data we [16:24] shot last week is this is what I [16:25] consider in this man version 1.0 zero [16:30] because the algorithm from the phase [16:31] three was the same from the phase two [16:33] was the same from phase one but because [16:35] we've been doing this for 10 years it's [16:37] a 10y year old algorithm [16:39] so as you look at the data it works [16:40] pretty well right as we said about the [16:42] phase two data 80% of people are disease [16:46] free after 5 years it's amazing for [16:47] those patients but there's still 20% of [16:50] patients that don't respond and so one [16:52] of the thing we're going to be doing now [16:54] that we have access to a phase three [16:56] patient data and samples [16:58] is to go back and mine that data to [17:01] figure out why some patient responded [17:04] and why some other patient did not [17:06] respond because we access to all their [17:07] blood sample the sequence everything and [17:09] we're going to try to see can we improve [17:11] the algorithm uh and we will go to CFDA [17:14] if we find scientific reason why we [17:17] should change the algorithm to go from [17:18] let's say 1.0 to a 2.0 zero algorithm [17:21] and then change it. Of course, we have [17:22] to do that in a very controlled way to [17:24] ensure we don't lose efficacy [17:27] very obviously. [17:28] But the way I think about it is like [17:30] when we talk about AI, it's we always [17:32] joke that the current version of AI is [17:34] the worst we're going to see in our [17:35] lifetime. Well, it's exactly the same [17:36] for inaran which is the current version [17:38] of inaran that moana is the worst [17:40] version of inismaran you're going to see [17:42] for the rest of medical history. And so [17:44] that gave me a lot of hope not only in [17:47] melanoma for those 20% of patient that [17:49] don't respond but also for potentially [17:51] over tumors that have been really hard [17:53] in a field like you know uh pancreas [17:56] cancer and others where imotherapy [17:57] doesn't work. We want to be able to [17:59] learn a lot about the technology using [18:02] also what the field has learned in the [18:04] last 10 years because the field has [18:05] learned a lot as you know this is even [18:07] not in inspir 1.0 so that's why I'm so [18:09] excited about what's coming next. [18:12] >> That's fantastic. So, um, looking back, [18:15] uh, you and I have had, we've known each [18:18] other for a very long time. Uh, I won't [18:20] depress you or me by saying how long, [18:23] >> but you were very young. You in [18:24] kindergarten, [laughter] [18:26] >> but so I've had the benefit of of of [18:29] seeing the Madna from the earliest days. [18:31] >> Yes. And one thing that is true that was [18:34] true then is true today is um you are [18:37] well first of all you are an engineer at [18:39] heart and um you have from the very very [18:43] beginning been obsessed with process [18:45] with operations with being efficient. [18:48] Um, and those things need to be [18:50] absolutely true if you're going to [18:52] attempt to do what you're trying to do [18:54] here with personalized cancer vaccines [18:56] and make a a you know a medicine for [18:59] each individual patient precisely [19:01] because in 90% of the cases there's no [19:03] overlap uh uh in terms of the antigens. [19:06] Um, can you talk us uh talk walk us [19:09] through a bit the operational lift that [19:12] is required here that you've already had [19:14] to do to even run the trial, but that [19:16] you would have to do if you eventually [19:18] commercialize um this product. [19:20] >> Sure. [19:21] >> Um and and let me maybe maybe one way to [19:23] frame it is I think a lot of people [19:25] think about the other big personalized [19:27] therapy that exists in cancer is cartis [19:29] cell therapy, right? And in that case, [19:32] the thing that and CARTT cell therapy [19:34] for folks that may not be familiar is [19:35] this idea that you take you take a [19:38] patient's tumor um and then you take the [19:40] patient's immune cells, you take them [19:42] out of the body and you essentially [19:44] reprogram the immune cells um and and [19:46] re-engineer them to be reactive to the [19:48] tumor and put them back into the [19:50] patient. I'm oversimplifying of course [19:52] um but that's you know cartis cell [19:54] therapy in a in a very sort of a simple [19:57] nutshell. In this case, you're doing in [19:59] some ways things that are very similar, [20:00] right? You're you're taking a piece of [20:02] the tumor and that you have in the form [20:04] of of a biopsy presumably um and and and [20:07] you're trying to teach the um you're [20:09] trying to sequence the tumor to generate [20:11] a vaccine that is very specific uh to [20:14] that patient's uh tumor. How do you [20:17] think about um essentially the veinto [20:19] vein tie? like what needs to be true [20:22] from the moment you you know sort of see [20:24] a patient and get access to the tumor to [20:27] the moment that patient receives their [20:29] personalized vaccine. [20:30] >> Sure. So the time is around 42 days now [20:34] needle to needle. So from taking the [20:36] biopsy to getting the vaccine in [20:38] hospital ready for you. I think we're [20:40] going to be able to improve that as we [20:41] still have a lot of efficiencies to work [20:43] on and automations and robotics. Uh the [20:48] place where this is very different from [20:50] carti [20:51] is that we don't have to take your [20:54] immune cells we program them xvivo in a [20:57] reactor in a fac in our factory and send [20:59] them back to your hospital. The only [21:02] thing we need is the information as you [21:04] and I talked about the beautiful thing [21:06] of ma it's an information molecule and [21:08] so basically what we get from the lab is [21:10] the sequence of your healthy cells and [21:13] the sequence of your cancer cells. So we [21:15] just get a file uh and then we use that [21:18] information to basically make the DNA [21:21] but now we don't make it with plasmid [21:24] growing ecoli or whatever we make it all [21:26] synthetic so it's all enzyatic in liquid [21:29] in water then we make the the RNA from [21:32] the the template then we put the lipid [21:34] around it and because of that and that [21:37] it's a synthetic process it's much more [21:39] like small molecule than a large [21:42] molecule you think about carti for me [21:43] the analogy is the re combinant world [21:46] >> where you have no sales and big reactors [21:48] and you have big volumes because as you [21:50] know the the reason you have big volumes [21:52] in biotech industry is if you compress [21:54] the cells too much they die you have the [21:56] same issue with carti so it's all [21:57] everything is big whereas here because [22:00] it's all in water and it's all enzyatic [22:02] meaning it just it's very catalytic the [22:04] reactors are very very tiny and so what [22:08] we been doing since before the clinic [22:11] because as you said we had to start [22:12] developing the technology to [22:14] individualize it one human at a time to [22:16] even do a phase one study, right? Is we [22:20] shrunk everything down. So the first [22:22] version of a machine that looks like a [22:23] big American fridge was a bit big and [22:25] cranky because we told the team make it [22:27] good enough so you have good quality but [22:30] we're not solving for efficiencies yet [22:32] because if it doesn't work in the [22:33] clinic, what's the point of wasting five [22:35] years making a beautiful amazing [22:37] optimized robot if science doesn't work, [22:40] right? So we told the team, make it good [22:42] so that we have no quality issue and we [22:44] don't have a false experiment, a false [22:46] negative in the clinic because it would [22:48] be terrible for patients if you build a [22:50] robot kind of working but not working. [22:53] You run the study and it tells you the [22:55] science doesn't work and you don't if it [22:58] should have worked, right? That would be [22:59] a disaster for humanity obviously. And [23:01] so the team did exactly that. They [23:02] developed a robot that was good in term [23:05] of quality but it was not very [23:07] efficient. uh and when we got the phase [23:10] two data that this was working the first [23:13] inter phase two data at only two years [23:15] now we have five years of data maturing [23:17] beautifully for duration of efficacy we [23:20] told the team okay now we're behind and [23:23] knew this was going to happen in case of [23:24] success which is a happy program and so [23:26] we dedicate a lot of very smart [23:28] engineers to think about okay now how do [23:30] you make a very efficient machine that [23:32] you can compress even the volume of a [23:34] machine because one of the important [23:36] vector of course is time as you [23:38] mentioned cycle time from needle to [23:40] needle, but also cost. And so one way to [23:44] reduce the cost is reduce the footprint [23:46] on the floor because if you have a fixed [23:48] envelope of a clean room facility and [23:52] you can put let's 2x or 10x more [23:54] machines in that surface area, of course [23:57] you're going to get a bigger throughput [23:58] and a much lower price on your fixed [24:00] cost. And so we are obsessed about cycle [24:03] time because the more minimal more you [24:05] can reduce cycle time the more you can [24:06] get a turnaround let's say in a year on [24:08] your assets and the second vector I'm [24:10] obsessed about is square inches [24:13] [laughter] [24:14] and literally I'm always a pain but [24:16] sorry when I go to the factory to look [24:18] you know at all the space we can save [24:20] how we can be creative how even we can [24:23] put some compute out of the clean rooms [24:25] just to shrink things as much as you can [24:28] because a huge impact on cost of the [24:31] product at the end of the day [24:33] >> and had um at scale um how many roughly [24:38] how many for let's say let's just focus [24:40] on melanoma how many doses would you [24:42] need to produce in a given year [24:44] >> so we've already got thousands of doses [24:47] because of nine clinical studies that [24:49] are ongoing [24:50] >> uh the facility will be able to make the [24:52] one in Marboro mass at tens of thousands [24:55] of doses [24:56] u and then as we keep improving the [24:59] technology that number is going to go up [25:00] in the same facility and then we might [25:02] need to build several other facilities [25:05] but if you look at the incidence of [25:06] melanoma if you tens of thousand of [25:09] doses you're going to cover you're going [25:10] to easily the melanoma market [25:14] >> yeah I I can believe that um have you [25:17] disclosed how you think about cogs and [25:20] price or is that something that [25:21] >> we have not disclosed yet uh we need [25:23] first to disclose the data with our mer [25:25] colleagues we need to engage with the [25:27] payers once we can share the data with [25:30] in term of what is the value being [25:31] driven there and so on. Um but this has [25:34] not been discussed yet. [25:35] >> So maybe one place to focus is uh on [25:38] this concept of personalization. Uh I'm [25:41] sure you've seen uh the story of the of [25:43] the GitLab founder uh who went founder [25:46] mode on his on his own uh osteocaroma. [25:49] Um uh number one uh do the future SIDS [25:54] of the world um come to Madna or is is [25:58] this end of one phenomena something that [26:00] you think will just happen and exist in [26:02] parallel. [26:03] >> So I think they will come most of them [26:05] to Mona because he will just going to be [26:08] easier and safer [26:10] because as you know uh making an [26:13] injectable product always carry risk of [26:15] contamination of a product. If you [26:17] inject to somebody a product that has [26:19] even one copy of bacteria, you might [26:20] give the patient sepsis, then you always [26:23] have a question of quality because when [26:25] you have a multi-step process, a mistake [26:27] can happen. And of course, if it's [26:30] industrialized and has been validated in [26:32] term of good manufacturing practice kind [26:34] of FDA standard, you have much less [26:36] chance of this happening. So, it's a bit [26:38] like every tools in life, which is, you [26:41] know, do you make your first knife [26:42] because there's no knife store and you [26:43] are in a cave and you need to feed your [26:45] family? Yes, of course you make your [26:47] first knife because you have to feed [26:48] your family, right? But when you have a [26:50] a store making high quality knives down [26:52] the street, you're going to use your [26:54] time to do something else. So I think [26:55] it's a bit of the same phenomena which [26:57] is like in any technology which is when [26:59] you have industrial scale of high [27:02] quality product, you use your time as a [27:04] human to do something else with your [27:06] time, right? [27:08] >> And in that world um how does the how [27:10] does the regulatory envir uh the [27:12] regulatory apparatus function here? So [27:15] in other words, you you mentioned [27:17] earlier uh you along with Merc will [27:19] prepare um a regulatory filing soon. Um [27:24] what gets regulated here? Because every [27:25] dose is different. So obviously every [27:27] dose doesn't get approved. Is it is it [27:30] obviously the the the process for [27:32] synthesizing mRNA? Is it the algorithm? [27:34] Is it a combination of the the entire [27:36] system? Help us uh understand that and [27:38] build intuition around how we think that [27:41] these kinds of personalized medicines [27:43] will be regulated in the future. Yes. So [27:45] the good news is there are precedents as [27:46] you mentioned Karti. Karti was also [27:49] approved in the same way as we believe [27:51] Intaran will be which is as a process [27:53] BLA uh not a product BLA. So as you know [27:57] Mon has five product approved. So that's [28:00] really product approval on this one. the [28:03] whole process since we started in the [28:04] clinic the IND was a process IND [28:07] >> because we had to ask the FDA can we go [28:08] to the clinic do you think it's safe and [28:10] do you think we have a good control of a [28:12] process so we can do safely a phase one [28:15] study so we already had that discussion [28:17] just to go into the clinic years ago and [28:19] then before we started every phase three [28:21] we need to have end of phase two meeting [28:23] and agree the design of a study this the [28:27] manufacturing protocol with the FDA so [28:30] those discussions have happened for [28:31] years and So it's not like we have not [28:33] talked to FD for the last 10 years I'm [28:34] going to show up at their you know front [28:37] door you know in a week or two and tell [28:39] them this is a new product and they're [28:41] going like what is this there's been a [28:43] lot of discussion a lot of engagement [28:45] several time we've had technical [28:47] question on the manufacturing front [28:50] where we basically requested additional [28:52] meetings to ask the guidance to also [28:55] educate them on the technology what [28:56] we've learned and so on. So there's [28:58] already been a lot of discussions and [29:00] there's a very clear regulatory pathway [29:02] in ter of approving the entire process. [29:05] Basically, what the FDA wants to know, [29:07] which is very legitimate, which I would [29:08] want for my own family sake, obviously, [29:11] which is if you give if you get the same [29:14] sample at the beginning, the tumor and [29:16] the blood, do you get the same product [29:18] make at the end of a big black box? And [29:21] that's what we have to demonstrate first [29:23] to ourself and then with the data to the [29:25] FDA so that we have really robustness [29:28] overall process. So if we have a same [29:30] input, we're going to get the same [29:31] output going to patient as an [29:34] individualized medicine. [29:37] >> How do we think about uh moving beyond [29:40] or how do you all think about moving [29:42] beyond melanoma? Uh is this an approach [29:44] that's going to be uh applicable to a [29:46] broad range of cancers? Are there [29:48] cancers that are much more likely where [29:49] this is going to be a viable option [29:52] versus others? uh and sort of what's [29:54] your u I'll use the word what's the [29:56] ambition here for where uh cancer [29:59] vaccines can can happen [30:00] >> the ambition is pretty big because we [30:02] believe we have demonstrated at least to [30:05] ourselves and we hope to the world [30:07] there's always will be skeptics but [30:09] that's that's always true that we are [30:12] able to create an educo education of T [30:18] cell and this we showed it even at this [30:20] year we took the blood before treatment [30:22] after treatment of cancer patient [30:24] melanoma we technology and we showed [30:26] that don't think we have an expansion of [30:28] a tea cells but we have denovo new tea [30:31] cells being created that recognize what [30:34] we code in the mRNA that was not in the [30:36] patient's body before the treatment so [30:39] we really in my book have proven to [30:41] ourselves and to the clinical community [30:44] that moderniz technology can develop [30:48] teach the immune system to develop new T [30:50] cell to go attack your cancer [30:53] So based on that there are basically I [30:55] would say three different vectors we're [30:57] going after in term of expansion from [31:00] melanoma. So this study to remind people [31:02] was a cancer patient in stage two, stage [31:06] three and stage four that were enrolled [31:08] in that phase three study. So what we're [31:10] doing is we're going first everywhere [31:12] where kudra works because as I told you [31:15] we believe the mechanism of action of a [31:17] PD1 and modern scientist man are totally [31:20] orthogonal. So we think these allow to [31:23] have synergistic element and performance [31:26] or efficacy for the patients. And so we [31:28] are in phase three for lung. Uh we're in [31:31] phase two for kidney cancer, bladder [31:33] cancer. So we have a world slew of [31:35] studies ongoing where the world knows [31:38] that kudra works because keta has been [31:40] approved there. And we believe you're [31:42] going to see a material improvement [31:43] versus ketra alone. Before you run the [31:45] clinical experiment, it's impossible to [31:47] know are you going to get 50% like we [31:49] saw in the phase two of melanoma or you [31:51] going to get 30% or 40% or another [31:54] number. We have to run the study. So [31:56] these are a lot ongoing. The second [31:58] vector is to go early in disease where [32:02] checkpoint work and the best example is [32:05] we announced in the spring of 2026 [32:08] starting a phase three study for patient [32:11] with stage one lung cancer. But as in so [32:15] modernized product as a monotherapy [32:18] without checkpoint. [32:20] >> Wow. [32:20] >> And we are doing that because we believe [32:22] when you go early in disease checkpoints [32:24] are not used because of a side effect [32:25] that they bring. [32:27] >> Mhm. Because if you have stage one [32:28] cancer, the medical field thinks it's [32:32] worth monitoring your cancer versus [32:34] giving you a checkpoint because not [32:37] everybody's going to respond. But [32:39] everybody's going to get pretty serious [32:41] lifelong side effects like autoimmune [32:43] disease. But what if you could have an [32:46] MRNA made for cancer patient that has [32:49] lung disease stage one, which you can [32:51] find easily with X-ray. Let's say in [32:54] former smokers is the easiest target [32:56] population. Just screen regularly with [32:58] X-ray your former smokers and if you do [33:00] it regularly you're going to go from not [33:02] seeing the cancer to seeing the cancer [33:04] and then the idea is do you do a surgery [33:07] which is to the side of care and you [33:09] give in monotherapy which the side [33:12] effect is similar to a vaccine you know [33:14] you might feel tired for a day but [33:16] that's it. So in cancer it's a pretty [33:18] cool type of side effect right uh and [33:21] that's another approach we have in term [33:23] of clinical studies uh where checkpoints [33:26] are not available today the third vector [33:29] is where checkpoint don't work so of [33:31] course it's where you have a highest [33:32] risk [33:34] >> but because the mechanism of action is [33:36] different from a checkpoint we and M [33:40] believe that there's a very good [33:41] scientific rational to go try so two [33:43] places we're trying right now is [33:45] pancreas cancer [33:47] and also gastric cancer. Those two [33:49] cancer type checkpoints and ketra do not [33:52] work. The clinical studies have been run [33:54] in the past and they were negative. But [33:57] we think because again the mechanism of [33:59] action is different from checkpoint and [34:02] we know now that we have a proof that we [34:04] can create denovo T cell. We think it's [34:06] an experiment worth running. If we have [34:09] good signal we will think about [34:10] combination. And as you know literally [34:12] yesterday you know revolution medicine [34:14] had a wonderful new medicine approved [34:17] for pancras cancer using the kas [34:20] mutation. What if you could combine that [34:23] medicine and in this man those are very [34:25] orthogonal mechanism of action. For me [34:28] it makes no scientific sense that if it [34:30] is work and we should know soon in in [34:33] pediatric sorry in in pancreas cancer [34:37] by itself and of course the revolution [34:40] medicine uh does great improvement of [34:43] survival in pancreas cancer. If you [34:46] combine those two things we believe it [34:48] would have benefit again you need to run [34:49] the clinical experiment to see how much [34:51] but that's the type of things we're [34:52] going to want to do. So if you think [34:54] about in tisan the modern medicine is [34:56] going to be used with kidra it's going [34:58] to be used early in diseases where kidra [35:00] and it might be used in places where [35:01] kidra doesn't work but with over agents [35:06] >> well that that gives a lot of uh reason [35:09] for I think uh cancer patients and their [35:11] families to have a lot of hope [35:13] >> for the future of of of novel therapies [35:17] in this field. [35:17] >> Yes and and on top of of what we just [35:19] said remember this is inan 1.0 zero. So [35:23] what I saw is very powerful and I'm [35:24] really pushing our team uh to think [35:27] really outside the box and to do a lot [35:29] of analysis and to use AI to look at [35:30] that gigantic set of data that we have [35:32] which is what are the things we can [35:34] learn from the clinical studies to [35:37] understand about the people that did not [35:39] respond because I think you always learn [35:41] more from things that don't work and [35:43] things that work. So I want to obsess [35:44] about the 20% of patient that do not [35:46] respond 5 years out so we can understand [35:49] why they did not respond and can we to [35:51] trick anything in the algorithm or in [35:54] the technology to be able to help them. [35:58] Wow, that's remarkable. And I just to [36:00] just to wrap, I think, uh it's [36:03] remarkable to see how you were able to [36:04] take a technology platform that [36:07] originally wasn't built for a uh you [36:10] know, a pandemic, uh pointed at a [36:13] pandemic, create a vaccine for millions [36:16] and millions and millions of people, and [36:18] essentially point it back towards uh [36:21] treating some of the diseases that you [36:22] had originally intended uh 10 years [36:24] later as as you described. It's very [36:26] remarkable and the piece that's going to [36:27] be exciting or maybe to close is before [36:30] the end of the year we should have our [36:32] pivotal study so late stage study for [36:36] rare genetic disease for kids that have [36:38] rare genetic disease of a liver. So it's [36:40] another vertical which we pointing the [36:42] technology the phase one two have shown [36:44] kids three years on drugs doing [36:45] fantastic. So we'll see when we get that [36:48] data and in June we had our annual [36:50] science day and we announced that the [36:52] next mountain where we are appointing [36:55] mRNA platform is autoimmune disease [36:57] because if you think about it we've [36:59] learned a lot for infectious disease [37:00] which are know mediated by the immune [37:03] system cancer we just been talking a lot [37:05] about immune system. So we learned so [37:07] much about the immune system that we [37:08] think we have some very novel approach [37:11] on how to treat the root cause of [37:13] autoimmune disease not the symptoms [37:16] which is what the farmer industry has [37:17] been doing. It's of course very helpful [37:19] to patient to treat the symptoms so we [37:21] can have higher quality of life but it [37:23] doesn't treat the root cause and we [37:24] think we might have find ways to use the [37:27] immune system to treat the root cause of [37:29] autoimmune disease. So there's still a [37:31] lot of ways to point the platform. So [37:33] we're quite exciting about what's what's [37:34] to come. Would the theory there be that [37:37] you'd have personalized autoimmune [37:39] modulators or would this be more a [37:41] product or more process? [37:43] >> So, we we're doing both. So, what we [37:45] presented in the in the spring was a [37:47] product that would be the same for [37:48] everybody. But what I'm the most excited [37:51] about which is in the lab still is the [37:54] ability to do individualized autoimmune [37:56] treatment where you target directly to [37:59] the immune cells that are attacking your [38:00] body as self when you have an autoimmune [38:03] disease and to have basically part of [38:04] your immune system going attacking those [38:07] immune cells that are out of order so [38:09] that you able to take the symptom of [38:12] immune disease out. Again, it's still [38:14] early days but that's what I'm excited [38:16] about today. [38:18] Well, going from infectious disease to [38:20] cancer to eventually autoimmune disease, [38:23] we would love to have you back on the [38:25] podcast to film episode 3 and complete [38:28] the trilogy when you're ready. [38:30] >> Wonderful. [38:30] >> Stefan, thank you so much for joining us [38:32] on the A6 and Z podcast. It's always, as [38:35] always, it's great to see you [38:36] >> and congratulations