抗体偶联药物（ADCs）曾被形象地比喻为肿瘤治疗中的“生物导弹”：抗体负责识别肿瘤细胞，连接子连接抗体与药物，细胞毒性载荷则负责杀伤靶细胞。但随着越来越多ADC进入临床，这一看似简单的“抗体+连接子+载荷”模式正在发生深刻变化。
过去几年，ADC开发已经从单一靶点、单一载荷以及相对随机的偶联方式，逐步迈向双特异性抗体、双载荷、定点偶联、非传统载荷以及数据驱动设计。与此同时，人们对于ADC的认识也从“如何把更强的药送进肿瘤”，转变为“如何同时优化靶点、抗体、连接子、载荷、偶联位点、药物抗体比（DAR）、药代动力学以及可制造性”。
这种范式变化意味着，下一代ADC的竞争不再只是寻找一种更强的载荷或一个更新的靶点，而是围绕整个分子的系统工程。尤其是双特异性ADC、双载荷ADC、定点偶联以及人工智能（AI）等技术的发展，正在进一步改变ADC的可开发性评价、生物工艺开发和生产制造逻辑。
从“高表达靶点”到系统设计：ADC开发逻辑正在改变
早期ADC开发有一个相对直观的逻辑：寻找一个在肿瘤细胞上高表达、在正常组织中低表达的抗原，通过抗体将高效细胞毒性药物递送到肿瘤细胞内部。但临床实践逐渐证明，抗原表达量并不能单独决定ADC是否有效。
理想靶点还需要综合考虑内化效率、抗原脱落、细胞表面持续可利用性、正常组织表达以及肿瘤内部表达异质性。尤其在实体瘤中，致密基质、较高的间质压力以及不均一的抗原表达都会影响ADC在肿瘤组织中的穿透和分布。T-DXd在HER2低表达乳腺癌中的成功就是一个典型案例：它说明即使抗原密度并不高，只要内化、连接子裂解、载荷活性、膜通透性和旁观者效应形成合适组合，仍有可能产生显著的治疗价值。
因此，现代ADC越来越像一个高度耦合的系统。抗体决定“找到谁”和“如何进入”，连接子决定“何时释放”，载荷决定“如何杀伤”，DAR和偶联位点则进一步影响疏水性、聚集、稳定性、清除以及组织分布。任何一个参数发生变化，都可能影响整个分子的药代动力学和治疗窗口。
载荷本身也在快速迭代。早期ADC主要使用美登素类和奥瑞他汀类微管抑制剂，而近年来拓扑异构酶I抑制剂迅速成为重要方向。与此同时，DNA损伤剂、免疫调节剂、STING激动剂、Toll样受体激动剂乃至PROTAC等非传统载荷也开始进入研究视野。
这意味着ADC正在逐渐突破“抗体递送细胞毒性小分子”的传统定义，演变为一种更加广义的抗体介导精准递送平台。
双特异性ADC：从“一个靶点”走向“双重识别”
下一代ADC中非常值得关注的方向之一，是双特异性抗体偶联药物。传统ADC通常依赖一个抗原实现肿瘤识别，但肿瘤并不是由完全一致的细胞构成。治疗过程中，抗原低表达或抗原阴性的细胞可能逃逸，并在选择压力下逐渐成为优势克隆。因此，抗原异质性和抗原丢失一直是ADC产生耐药的重要原因。
双特异性ADC提供了一种新的解决思路：让同一个抗体分子识别两个不同抗原。例如同时识别HER2/TROP2等不同肿瘤相关抗原，可以扩大对异质性肿瘤细胞群的覆盖；另一种思路则是“双表位”设计，即同时结合相同抗原上的两个不同表位，以提高结合、受体聚集和内化效率。这类策略有望降低单一靶点逃逸风险，并改善抗原低表达区域的结合和内化。
但“双靶点”并不意味着简单地把两个抗体功能拼在一起。双特异性ADC需要重新考虑抗体价态、两个结合臂的亲和力、空间构型、内化动力学、靶点共表达程度以及偶联位置。例如，过高的亲和力未必意味着更好的肿瘤穿透；两个靶点是否同时存在于同一个细胞，以及结合后能否高效进入溶酶体，也会直接影响载荷递送。
更重要的是，双特异性结构将进一步放大可开发性与CMC挑战。抗体本身已经具有更复杂的链配对和结构异质性，再叠加连接子和载荷后，需要同时控制聚集、片段、错误配对、DAR、游离载荷、偶联位点以及电荷异质性等质量属性。
因此，下一代双特异性ADC开发必须更早地将“生物学有效性”与“可制造性”结合起来。一个在细胞实验中表现优异的复杂分子，如果表达量低、易聚集、偶联效率不稳定或难以建立稳健的分析方法，最终仍可能在开发过程中遇到瓶颈。
双载荷ADC：一次递送两种武器，能否突破耐药与异质性？
如果说双特异性ADC解决的是“识别谁”的问题，那么双载荷ADC试图解决的则是“如何杀伤”的问题。传统ADC通常在同一抗体上偶联同一种载荷。但肿瘤细胞可能通过上调药物外排泵、改变DNA修复机制、改变细胞周期或细胞内转运等方式逐渐形成耐药。与此同时，同一肿瘤中的不同细胞群对同一种载荷的敏感性也可能不同。
双载荷策略因此受到关注：在同一个抗体骨架上同时偶联两种作用机制不同的药物。例如，将拓扑异构酶抑制剂与另一类细胞毒性载荷组合，使ADC进入肿瘤细胞后从两个不同机制产生杀伤，从理论上降低单一耐药机制导致整体治疗失效的可能。例如有研究将Exatecan与RNA聚合酶II抑制剂Triptolide同时装载到一个ADC平台，并采用分选酶介导的定点偶联实现更加精准的载荷化学计量控制。这一思路体现了双载荷ADC发展的核心——重点已经不是简单追求“装更多药”，而是控制载荷A与载荷B分别装多少、装在哪里，以及以什么速度释放。
而这恰恰也是其制造难点。单载荷ADC已经需要控制平均DAR和DAR分布，双载荷ADC则可能需要进一步定义DAR-A、DAR-B、总DAR、两种载荷比例以及各自的位点占有率。理论上平均DAR相同的两个批次，其载荷组成和分布仍可能完全不同。因此，仅用一个平均DAR数值已经难以充分描述产品。
这将直接推动ADC工艺向更精准的定点偶联、酶促偶联和正交偶联发展，同时对纯化和分析方法提出更高要求。工艺开发需要更加严格地控制抗体预处理、还原/氧化状态、反应时间、pH、温度、载荷摩尔比和混合条件；下游则需要有效去除未偶联抗体、游离载荷、聚集体以及异常偶联组分。
分析端也必须面对更多质量属性，包括两种载荷的身份与含量、偶联位置、载荷比例、DAR分布、聚集、电荷异质性、游离药物以及不同载荷的释放动力学等。分子设计越精准，制造过程往往也需要越精准。
定点偶联与可开发性：ADC正在从“平均值控制”走向分子级控制
ADC制造的另一个明显趋势，是从传统随机偶联向更加可控的定点偶联发展。传统赖氨酸或半胱氨酸偶联技术已经支撑了多款成功ADC，但随机偶联往往产生包含不同DAR和不同偶联位点的分子群。不同组分可能具有不同的疏水性、稳定性、药代动力学和毒性，使最终产品实际上成为一个复杂的分子集合。
工程化半胱氨酸、非天然氨基酸、糖链改造以及酶促偶联等定点技术，则能够更加精确地定义“药物装在哪里”和“装多少”，降低产品异质性并改善批间一致性。但“更均一”并不天然等于“更好的药”。事实上，包括T-DM1、T-DXd在内的成功ADC仍使用传统偶联策略。这说明ADC最终表现来自抗原、抗体、连接子、载荷、DAR、药代动力学和肿瘤生物学的综合作用，而不能简单用某一个技术指标判断。
真正发生改变的，是可开发性评价正在前移。载荷往往具有较强疏水性。随着DAR增加，ADC可能出现溶解度下降、聚集增加、黏度变化和构象稳定性下降。通过PEG、磺酸基团等亲水性设计可以改善部分问题，却又可能影响载荷释放、药效甚至制造工艺。因此，未来ADC候选分子的筛选不能只看细胞杀伤活性，还需要同步评估表达量、热稳定性、疏水性、聚集倾向、溶解度、黏度、DAR分布和储存稳定性。
这对生物工艺开发同样提出新的要求。未来ADC工艺不再只是“先生产抗体，再完成偶联”，而需要让上游抗体质量、纯化工艺、偶联反应、流体管理、过滤、制剂和灌装形成更加连续的质量控制链条。尤其对于高活性细胞毒性载荷，还必须同时考虑密闭操作、人员防护、交叉污染控制以及一次性系统材料的化学相容性和可提取物/析出物风险。对于双特异性、双载荷等复杂ADC而言，工艺本身已经成为分子设计能否真正落地的重要组成部分。
AI加入ADC开发：从“试出来”走向“预测—实验—再学习”
随着ADC设计变量不断增加，一个现实问题也越来越突出：可探索的组合空间正在快速膨胀。不同抗原、抗体序列、双特异性构型、连接子、载荷、偶联位点和DAR之间，可以产生数量巨大的组合。依靠传统“逐个设计—逐个实验”的模式，不仅开发周期漫长，而且很多可开发性问题往往直到项目后期才暴露。
AI和计算模型正在改变这一流程。在靶点发现阶段，机器学习可以整合转录组、蛋白质组、基因组以及正常组织表达信息，对潜在细胞表面抗原进行优先级排序；在抗体工程阶段，蛋白质语言模型和结构预测可以辅助分析CDR、抗体—抗原结合以及双特异性抗体链配对；进入ADC设计阶段后，模型还可以用于预测连接子稳定性、载荷相容性、偶联位点以及不同DAR带来的疏水性和稳定性变化。
这使ADC研发开始形成类似“设计—构建—测试—学习”的闭环。实验失败也不再只是被淘汰的数据，而可以重新进入模型，帮助识别导致聚集、低表达、稳定性差或药代异常的结构特征，再指导下一轮设计。
AI同样开始向临床转化延伸。PBPK和QSP等模型能够综合靶点表达、内化速率、旁观者效应、组织分布和器官毒性等因素，辅助首次人体剂量、给药方案以及联合治疗策略的设计；真实世界数据和机器学习也有望帮助识别间质性肺病、眼部毒性和血液学毒性等风险。
不过，AI目前更适合作为缩小设计空间和提高实验效率的工具，而不是实验的替代者。ADC具有高度非线性的体内行为，而高质量、标准化、带明确标签的ADC数据仍然有限。模型在体外数据上获得的规律，也未必能够直接外推至复杂的人体环境。因此，AI真正的价值可能并不是“一次预测出最佳ADC”，而是让计算、实验、生物工艺和临床数据形成持续迭代的开发闭环。
下一代ADC：创新最终要回到“能否制造、能否控制、能否放大”
从双特异性ADC、双载荷ADC到非细胞毒性载荷和AI辅助设计，ADC显然正在进入更加复杂，也更加精准的阶段。未来ADC可能不再有固定模板：一个分子可以同时识别两个靶点，也可以携带两种不同载荷；载荷可能不再负责直接杀死细胞，而是激活免疫、诱导特定蛋白降解或调控疾病相关通路；偶联也将从随机分布逐渐走向更加精准的位点和化学计量控制。
但复杂性本身并不是目的。真正有价值的创新，需要同时回答几个问题：它能否扩大治疗窗口？能否克服肿瘤异质性和耐药？能否形成稳定、可表征的产品？能否建立稳健、可放大的生产工艺？
这也意味着ADC开发正在进一步打破发现、CMC和制造之间的边界。对于下一代复杂偶联药物，可开发性、可制造性和工艺放大需要更早进入候选分子设计阶段；高分辨分析、精准偶联、封闭式流体处理、一次性技术、高活性物料防护以及数字化过程控制，也将成为支持复杂ADC走向产业化的重要基础。
从更长远的角度看，ADC正在从一种药物形式转变为一个融合抗体工程、化学偶联、生物工艺、分析科学、计算建模与人工智能的平台。下一阶段的突破，或许不再来自某一个孤立技术，而来自这些技术能否真正被整合进同一个设计—制造—验证体系。
当“找到正确的靶点”“设计正确的分子”和“稳定地制造这个分子”开始在研发早期被同时考虑，ADC才真正从经验驱动的偶联药物，迈向数据驱动、工程化和可预测的精准治疗平台。
随着ADC向双特异性、双载荷及更加精准的偶联策略发展，其生产制造也对流体处理、偶联反应、高活性物料防护及工艺密闭性提出更高要求。多宁生物依托完善的一站式生物工艺平台，为ADC生产提供覆盖关键工艺环节的产品与技术支持。一次性配储液系统可满足缓冲液、工艺溶液及中间产物的配制与储存需求；一次性生物偶联反应器兼顾混合性能与材料化学相容性，可支持ADC偶联等关键工艺；针对高活性及细胞毒性物料，柔性隔离器为称量、转移及操作提供有效的密闭防护；配套的一次性流体管理解决方案则覆盖物料转移、连接、泵送及封闭式操作。通过一次性技术与系统化工艺集成，多宁生物助力ADC工艺从开发走向更加安全、高效、灵活且可放大的生产制造。
DuoMix® 一次性搅拌系统
多宁生物DuoMix®一次性搅拌系统专为生物制药工艺中的混合、配液及缓冲液制备等应用场景设计，具备优异的可放大性、灵活性和过程可控性。系统支持从实验室研发到商业化生产的规模化放大，确保不同阶段工艺参数的一致性与稳定性。DuoMix®可根据工艺需求配置pH、电导率、温度等多种传感器，实现关键参数的实时监测与精准控制，同时支持夹套控温，满足不同温度条件下的工艺需求。此外，多宁生物可提供专业的计算流体力学（CFD）分析，深入评估混合均一性、剪切力及传质效率，助力客户优化工艺设计与提升生产可靠性，为生物制药全过程提供高效、稳定的一次性混合解决方案。
一次性生物偶联反应器
针对ADC生产过程中对反应稳定性、温控精度及无菌安全性的高要求，多宁生物开发了一次性偶联反应器解决方案。该系统采用专为偶联工艺优化的磁悬浮搅拌技术，实现低摩擦、低热量输出的温和混合，满足长时间反应需求。同时，设备配备定制化搅拌电机绝热材料及五面蜂窝夹套设计，有效降低设备散热及反应放热对偶联效率的影响，提升温控稳定性。通过集成TCU与蠕动泵自动控制，实现精准控温和一体化操作，温度响应可控制在±0.5℃以内。此外，设备支持集成蠕动泵、限压保护及氮气保护设计，可满足小分子偶联过程中的惰性气体保护需求，为ADC药物开发及商业化生产提供安全、高效、灵活的一次性解决方案。
一次性工艺管组
在ADC工艺中，一次性管组有助于降低交叉污染风险、提升操作效率并简化清洁验证流程，是实现高效、合规、可追溯生产的重要保障。多宁生物提供全面的一次性工艺管组产品，涵盖层析、超滤等关键单元操作，广泛适用于ADC等高附加值产品的生产需求。产品支持灵活设计，可定制管路布局、连接接口和组件配置，兼容市场上主流的层析与超滤系统，便于集成现有平台。管组可集成压力、电导、UV等关键在线监测组件，实现对工艺过程的精准控制。所有组件均通过严格的质量验证，确保在压力、流速和流体分布等关键参数上的稳定性与一致性。
一次性灌装组件解决方案
多宁生物提供面向高价值生物药生产的一次性灌装解决方案，覆盖从药液转移、无菌连接到终端灌装的全流程需求。针对ADC生产过程中高活性、高复杂性以及对无菌和材料相容性的严格要求，多宁生物提供经过验证的一次性灌装组件，采用高等级洁净环境预组装及伽马射线灭菌，集成入口连接器、储液袋、蠕动泵管路、灌装管路及灌装针头等关键部件，可灵活匹配不同灌装设备与容器形式。组件选用合规材料，具备优异的生物相容性、化学耐受性及低析出特性，并配套完整法规支持文件，助力ADC等高附加值药物实现安全、高效、可靠的无菌灌装。
DuoIsola™ 柔性隔离器
多宁DuoIsola™柔性隔离器是一款面向高活性、高毒性药物制造场景的新一代密闭防护解决方案，专为OEB≥4级物料操作设计。产品采用模块化柔性膜结构，结合负压控制、多级HEPA过滤及一次性组件设计，实现人员、产品与环境的高效隔离。其具备CPT <10 ng/m³的高水平密闭性能，支持快速部署、灵活配置及低维护运行，可广泛应用于ADC、HPAPI等创新药物研发及生产过程，为制药企业提供安全、高效、智能化的一站式隔离防护平台。
参考文献：
S.Kumari, L.M.Cool, E.Howard, et al., Advances and Future Directions in Antibody–Drug Conjugates: From Paradigm Shifts to Data-Driven Design. Cancers, 2026.
C.S.Barry, The Role of Contract Manufacturing in Antibody–Drug Conjugate (ADC) Drug Substance Development – Strategies for Successful Engagement. AAPS PharmSciTech, 2026.
X.Cai, S.Guo, J.Li, et al., Global landscape and trends of antibody-drug conjugates in oncology: A machine learning–driven decade-long informatics analysis. Journal of Clinical Oncology, 2026.
Y.Li, Y.Zhu, Y.Chen, et al., Advances in antibody-drug conjugates in cancer: latest updates from the 2026 AACR annual meeting. Experimental Hematology & Oncology, 2026.
关于多宁生物
上海多宁生物科技股份有限公司是一家中国领先的一站式生物工艺解决方案提供商，致力于提供生物制药产品从研发到商业化生产的综合解决方案，包括试剂及耗材、仪器设备、服务。公司主要运营生物工艺解决方案、实验室产品及服务两大业务线，通过一站式生物工艺体系帮助合作伙伴实现高效、稳定、质量及成本可控的药物研发及生产流程。
了解更多：www.duoningbio.com
媒体垂询：PR@duoningbio.com
Antibody–drug conjugates (ADCs) were once vividly described as "biological missiles" for cancer treatment: the antibody recognizes tumor cells, the linker connects the antibody to the drug, and the cytotoxic payload kills the target cells. However, as an increasing number of ADCs enter clinical development, this seemingly straightforward "antibody + linker + payload" model is undergoing profound changes.
Over the past few years, ADC development has evolved from single targets, single payloads, and relatively random conjugation approaches toward bispecific antibodies, dual payloads, site-specific conjugation, non-traditional payloads, and data-driven design. At the same time, the focus of ADC development has shifted from "how to deliver a more potent drug into tumors" to "how to simultaneously optimize the target, antibody, linker, payload, conjugation site, drug-to-antibody ratio (DAR), pharmacokinetics, and manufacturability."
This paradigm shift means that competition in next-generation ADCs is no longer simply about identifying a more potent payload or a novel target, but rather about systems engineering of the entire molecule. In particular, advances in bispecific ADCs, dual-payload ADCs, site-specific conjugation, and artificial intelligence (AI) are further reshaping the logic of ADC developability assessment, bioprocess development, and manufacturing.
From "Highly Expressed Targets" to Systems Design: The Logic of ADC Development Is Changing
Early ADC development followed a relatively straightforward rationale: identify an antigen that is highly expressed on tumor cells but minimally expressed in normal tissues, and use an antibody to deliver a highly potent cytotoxic drug into tumor cells. However, clinical experience has increasingly demonstrated that antigen expression level alone does not determine whether an ADC will be effective.
An ideal target must also be evaluated in terms of internalization efficiency, antigen shedding, sustained availability on the cell surface, expression in normal tissues, and intratumoral expression heterogeneity. Particularly in solid tumors, dense extracellular matrices, elevated interstitial pressure, and heterogeneous antigen expression can all affect ADC penetration and distribution within tumor tissues. The success of T-DXd in HER2-low breast cancer is a representative example: it demonstrates that even when antigen density is relatively low, significant therapeutic benefit may still be achieved if internalization, linker cleavage, payload potency, membrane permeability, and the bystander effect are appropriately balanced.
Modern ADCs are therefore increasingly viewed as highly coupled systems. The antibody determines "what to target" and "how to enter the cell," the linker determines "when to release," and the payload determines "how to kill." DAR and conjugation sites further influence hydrophobicity, aggregation, stability, clearance, and tissue distribution. A change in any one of these parameters may affect the pharmacokinetics and therapeutic window of the entire molecule.
Payload technologies are also evolving rapidly. Early ADCs primarily relied on microtubule inhibitors such as maytansinoids and auristatins, while topoisomerase I inhibitors have emerged as an important class of payloads in recent years. At the same time, non-traditional payloads—including DNA-damaging agents, immunomodulators, STING agonists, Toll-like receptor agonists, and even PROTACs—are beginning to attract research interest.
This means that ADCs are gradually moving beyond the traditional definition of "antibody-mediated delivery of cytotoxic small molecules" and evolving into a broader platform for antibody-mediated precision delivery.
Bispecific ADCs: Moving from "One Target" to "Dual Recognition"
One of the most promising directions in next-generation ADC development is bispecific antibody–drug conjugates. Conventional ADCs typically rely on a single antigen for tumor recognition, but tumors are not composed of completely homogeneous cell populations. During treatment, cells with low or absent target antigen expression may escape therapy and gradually become dominant clones under selective pressure. Consequently, antigen heterogeneity and antigen loss have long been important mechanisms of ADC resistance.
Bispecific ADCs offer a new strategy to address this challenge by enabling a single antibody molecule to recognize two different antigens. For example, simultaneously targeting different tumor-associated antigens such as HER2 and TROP2 may broaden coverage across heterogeneous tumor cell populations. Another approach involves "dual-epitope" designs that simultaneously bind two distinct epitopes on the same antigen, potentially enhancing binding, receptor clustering, and internalization. Such strategies may reduce the risk of escape from single-target therapy while improving binding and internalization in regions with low antigen expression.
However, "dual targeting" does not simply mean combining the functions of two antibodies. Bispecific ADC development requires careful reconsideration of antibody valency, the affinity of each binding arm, spatial configuration, internalization kinetics, the degree of target co-expression, and conjugation sites. For example, excessively high affinity does not necessarily translate into better tumor penetration. Whether the two targets are co-expressed on the same cell and whether the resulting complex can efficiently traffic to lysosomes also directly affect payload delivery.
More importantly, bispecific structures further amplify developability and CMC challenges. Bispecific antibodies already involve more complex chain pairing and structural heterogeneity. Once linkers and payloads are introduced, multiple quality attributes must be controlled simultaneously, including aggregation, fragments, mispaired species, DAR, free payload, conjugation sites, and charge heterogeneity.
Therefore, the development of next-generation bispecific ADCs must integrate "biological efficacy" with "manufacturability" at an earlier stage. A complex molecule that performs exceptionally well in cellular assays may still encounter significant development bottlenecks if it has low expression levels, a high tendency to aggregate, unstable conjugation efficiency, or difficulties in establishing robust analytical methods.
Dual-Payload ADCs: Can Delivering Two Weapons at Once Overcome Drug Resistance and Tumor Heterogeneity?
If bispecific ADCs address the question of "what to recognize," dual-payload ADCs attempt to answer the question of "how to kill." Conventional ADCs typically carry a single type of payload on the same antibody. However, tumor cells may gradually develop resistance through mechanisms such as upregulation of drug efflux pumps, alterations in DNA repair pathways, changes in the cell cycle, or modified intracellular trafficking. Meanwhile, different cell populations within the same tumor may also exhibit varying sensitivities to the same payload.
This has driven growing interest in dual-payload strategies, in which two drugs with different mechanisms of action are conjugated to the same antibody scaffold. For example, combining a topoisomerase inhibitor with another class of cytotoxic payload could enable an ADC to kill tumor cells through two distinct mechanisms after internalization, theoretically reducing the likelihood that a single resistance mechanism will cause complete treatment failure. One study, for example, incorporated both exatecan and the RNA polymerase II inhibitor triptolide into a single ADC platform and used sortase-mediated site-specific conjugation to achieve more precise control over payload stoichiometry. This approach illustrates a central principle of dual-payload ADC development: the goal is no longer simply to "load more drugs," but to precisely control how much of payload A and payload B is attached, where each payload is conjugated, and how quickly each is released.
This is also precisely where the manufacturing challenges arise. Single-payload ADCs already require control over average DAR and DAR distribution. Dual-payload ADCs may require further definition of DAR-A, DAR-B, total DAR, the ratio between the two payloads, and site occupancy for each payload. Two batches with the same average DAR could theoretically have completely different payload compositions and distributions. As a result, a single average DAR value is no longer sufficient to fully characterize the product.
This complexity will directly drive ADC processes toward more precise site-specific, enzymatic, and orthogonal conjugation technologies, while imposing higher requirements on purification and analytical methods. Process development will require tighter control over antibody pretreatment, reduction/oxidation state, reaction time, pH, temperature, payload molar ratios, and mixing conditions. Downstream processes must also efficiently remove unconjugated antibodies, free payloads, aggregates, and aberrantly conjugated species.
Analytical development must likewise address an expanding range of quality attributes, including the identity and content of both payloads, conjugation sites, payload ratios, DAR distribution, aggregation, charge heterogeneity, free drug, and the release kinetics of each payload. The more precisely a molecule is designed, the more precise its manufacturing process often needs to become.
Site-Specific Conjugation and Developability: ADCs Are Moving from "Average-Value Control" to Molecular-Level Control
Another clear trend in ADC manufacturing is the shift from traditional random conjugation toward more controlled site-specific conjugation. Conventional lysine- or cysteine-based conjugation technologies have supported the development of several successful ADCs, but random conjugation typically generates heterogeneous molecular populations containing different DARs and conjugation sites. Different species may exhibit distinct hydrophobicity, stability, pharmacokinetics, and toxicity profiles, making the final product essentially a complex mixture of molecular species.
Site-specific technologies based on engineered cysteines, non-natural amino acids, glycan engineering, and enzymatic conjugation can more precisely define "where the drug is attached" and "how much drug is attached," thereby reducing product heterogeneity and improving batch-to-batch consistency. However, "greater homogeneity" does not inherently mean "a better drug." In fact, successful ADCs including T-DM1 and T-DXd still use conventional conjugation strategies. This demonstrates that the ultimate performance of an ADC results from the combined effects of the antigen, antibody, linker, payload, DAR, pharmacokinetics, and tumor biology, rather than being determined by any single technical parameter.
What is truly changing is that developability assessment is moving earlier in the development process. Payloads are often highly hydrophobic. As DAR increases, ADCs may experience reduced solubility, increased aggregation, altered viscosity, and decreased conformational stability. Hydrophilic modifications such as PEGylation or the introduction of sulfonate groups can alleviate some of these issues, but they may also affect payload release, pharmacological activity, and even the manufacturing process. Therefore, future ADC candidate screening cannot focus solely on cell-killing activity. Expression level, thermal stability, hydrophobicity, aggregation propensity, solubility, viscosity, DAR distribution, and storage stability will also need to be evaluated in parallel.
This also places new demands on bioprocess development. Future ADC manufacturing will no longer be simply a matter of "producing the antibody first and then performing conjugation." Instead, upstream antibody quality, purification, conjugation reactions, fluid management, filtration, formulation, and filling will need to form a more continuous and integrated quality-control chain. Particularly for highly potent cytotoxic payloads, manufacturing must simultaneously address closed processing, operator protection, cross-contamination control, chemical compatibility of single-use system materials, and extractables and leachables risks. For complex ADC formats such as bispecific and dual-payload ADCs, the manufacturing process itself has become a critical factor in determining whether an advanced molecular design can be successfully translated into a viable product.
AI Enters ADC Development: Moving from "Trial and Error" to "Predict–Experiment–Learn"
As the number of variables involved in ADC design continues to increase, a practical challenge is becoming increasingly apparent: the design space is expanding rapidly. Different combinations of antigens, antibody sequences, bispecific formats, linkers, payloads, conjugation sites, and drug-to-antibody ratios (DARs) can generate an enormous number of possibilities. Relying on the traditional "design one—test one" approach not only prolongs development timelines, but also means that many developability issues may not emerge until relatively late in development.
AI and computational modeling are beginning to transform this process. During target discovery, machine learning can integrate transcriptomic, proteomic, genomic, and normal-tissue expression data to prioritize potential cell-surface antigens. In antibody engineering, protein language models and structural prediction tools can assist in analyzing complementarity-determining regions (CDRs), antibody–antigen interactions, and chain pairing in bispecific antibodies. At the ADC design stage, computational models can also be used to predict linker stability, payload compatibility, conjugation sites, and changes in hydrophobicity and stability associated with different DARs.
This is enabling ADC development to evolve toward a closed-loop "design–build–test–learn" model. Failed experiments are no longer simply discarded data points; instead, they can be fed back into computational models to help identify structural features associated with aggregation, low expression, poor stability, or abnormal pharmacokinetic behavior, thereby informing the next round of molecular design.
AI is also beginning to extend into clinical translation. Models such as physiologically based pharmacokinetic (PBPK) and quantitative systems pharmacology (QSP) models can integrate factors including target expression, internalization rates, bystander effects, tissue distribution, and organ toxicity to support the design of first-in-human dosing, dosing regimens, and combination treatment strategies. Real-world data and machine learning may also help identify risks associated with interstitial lung disease, ocular toxicity, hematologic toxicity, and other adverse effects.
At present, however, AI is better positioned as a tool for narrowing the design space and improving experimental efficiency rather than replacing experiments. ADCs exhibit highly nonlinear behavior in vivo, while high-quality, standardized, and clearly labeled ADC datasets remain limited. Patterns derived from in vitro data may not necessarily extrapolate directly to the complex human physiological environment. Therefore, the true value of AI may not lie in "predicting the optimal ADC in a single attempt," but rather in enabling computational modeling, experimentation, bioprocess development, and clinical data to form a continuously iterative development loop.
Next-Generation ADCs: Innovation Ultimately Comes Down to Manufacturability, Controllability, and Scalability
From bispecific ADCs and dual-payload ADCs to non-cytotoxic payloads and AI-assisted design, ADCs are clearly entering a more complex—and more precise—stage of development. Future ADCs may no longer follow a fixed template: a single molecule may recognize two targets while simultaneously carrying two different payloads. Payloads may no longer function solely by directly killing cells, but instead activate immune responses, induce targeted protein degradation, or modulate disease-related pathways. Conjugation strategies are also likely to evolve from random distributions toward increasingly precise control of conjugation sites and stoichiometry.
Complexity itself, however, is not the goal. Meaningful innovation must address several fundamental questions: Can it broaden the therapeutic window? Can it overcome tumor heterogeneity and drug resistance? Can it produce a stable and well-characterized product? And can a robust and scalable manufacturing process be established?
This also means that ADC development is increasingly breaking down the traditional boundaries between discovery, CMC, and manufacturing. For next-generation complex conjugated therapeutics, developability, manufacturability, and process scalability need to be incorporated earlier into candidate design. High-resolution analytics, precision conjugation, closed fluid handling, single-use technologies, containment strategies for highly potent materials, and digital process control will also become important foundations for translating increasingly complex ADCs into commercial manufacturing.
From a longer-term perspective, ADCs are evolving from a single drug modality into an integrated platform combining antibody engineering, chemical conjugation, bioprocessing, analytical science, computational modeling, and artificial intelligence. The next major breakthrough may therefore come not from any single technology in isolation, but from how effectively these technologies can be integrated into a unified design–manufacturing–validation framework.
When "identifying the right target," "designing the right molecule," and "manufacturing that molecule consistently" are considered simultaneously from the early stages of development, ADCs can truly evolve from empirically developed conjugated therapeutics into data-driven, engineered, and increasingly predictable precision medicine platforms.
As ADCs evolve toward bispecific formats, dual payloads, and more precise conjugation strategies, their manufacturing processes are placing increasingly stringent requirements on fluid handling, conjugation reactions, containment of highly potent materials, and process closure. Leveraging its comprehensive one-stop bioprocessing platform, Duoning Biotech provides products and technical support covering key steps in ADC manufacturing. Single-use mixing and storage systems meet the requirements for the preparation and storage of buffers, process solutions, and intermediates; single-use bioconjugation reactors combine effective mixing performance with material chemical compatibility to support critical processes such as ADC conjugation; flexible isolators provide effective containment for the weighing, transfer, and handling of highly potent and cytotoxic materials; and complementary single-use fluid management solutions support material transfer, connection, pumping, and closed processing. Through single-use technologies and systematic process integration, Duoning Biotech helps advance ADC processes from development toward safer, more efficient, flexible, and scalable manufacturing.
DuoMix® Single-Use Mixing System
Duoning Biotech’s DuoMix® Single-Use Mixing System is specifically designed for mixing, solution preparation, and buffer preparation applications in biopharmaceutical processes, offering excellent scalability, flexibility, and process controllability. The system supports seamless scale-up from laboratory development to commercial manufacturing, ensuring consistency and stability of process parameters throughout different production stages. DuoMix® can be configured with various sensors, including pH, conductivity, and temperature sensors, according to specific process requirements, enabling real-time monitoring and precise control of critical process parameters. The system also supports jacketed temperature control to meet process requirements under different temperature conditions. In addition, Duoning Biotech provides professional computational fluid dynamics (CFD) analysis to comprehensively evaluate mixing uniformity, shear stress, and mass transfer efficiency. These capabilities help customers optimize process design, enhance manufacturing reliability, and achieve efficient and stable single-use mixing solutions throughout the biopharmaceutical production lifecycle.
Single-use Bioconjugation Reactor
To address the stringent requirements for reaction stability, temperature control accuracy, and aseptic safety during ADC manufacturing, Duoning Biotech has developed a single-use bioconjugation reactor solution. The system adopts a magnetic levitation mixing technology specifically optimized for conjugation processes, enabling gentle mixing with low friction and minimal heat generation to meet the requirements of long-duration reactions. Meanwhile, the reactor is equipped with customized thermal insulation materials for the mixing motor and a five-sided honeycomb jacket design, effectively reducing heat dissipation from the equipment and minimizing the impact of reaction heat release on conjugation efficiency, thereby improving temperature control stability. By integrating TCU and peristaltic pump automatic control, the system enables precise temperature regulation and integrated operation, with temperature response accuracy controlled within ±0.5°C. In addition, the reactor supports integrated peristaltic pumps, pressure-limit protection, and nitrogen protection designs to meet the inert gas protection requirements of small-molecule conjugation processes. It provides a safe, efficient, and flexible single-use solution for ADC drug development and commercial-scale manufacturing.
Single-Use Process Assemblies
In ADC manufacturing processes, single-use process assemblies help reduce the risk of cross-contamination, improve operational efficiency, and simplify cleaning validation requirements, serving as an important foundation for efficient, compliant, and traceable production. We provide a comprehensive portfolio of single-use process assemblies covering critical unit operations such as chromatography and ultrafiltration, supporting the manufacturing requirements of high-value products including ADCs. The assemblies feature flexible design capabilities, allowing customization of tubing layouts, connection interfaces, and component configurations. They are compatible with mainstream chromatography and ultrafiltration platforms available in the market, enabling seamless integration into existing manufacturing systems. The assemblies can be integrated with critical in-line monitoring components, including pressure sensors, conductivity sensors, and UV detectors, enabling precise process monitoring and control. All components undergo rigorous quality verification to ensure stable and consistent performance in key parameters such as pressure tolerance, flow rate, and fluid distribution.
DuoIsola™ Flexible Isolator
Duoning DuoIsola™ Flexible Isolator is a next-generation containment protection solution designed for high-potency and highly toxic drug manufacturing applications. Specifically developed for OEB ≥4 material handling, it features a modular flexible film structure combined with negative pressure control, multi-stage HEPA filtration, and single-use components to achieve effective isolation between operators, products, and the environment. With a high-level containment performance of CPT <10 ng/m³, DuoIsola™ enables rapid deployment, flexible configuration, and low-maintenance operation. It is widely applicable in the R&D and manufacturing processes of innovative therapies such as ADCs and HPAPIs, providing pharmaceutical companies with a safe, efficient, and intelligent one-stop containment platform.
References:
S.Kumari, L.M.Cool, E.Howard, et al., Advances and Future Directions in Antibody–Drug Conjugates: From Paradigm Shifts to Data-Driven Design. Cancers, 2026.
C.S.Barry, The Role of Contract Manufacturing in Antibody–Drug Conjugate (ADC) Drug Substance Development – Strategies for Successful Engagement. AAPS PharmSciTech, 2026.
X.Cai, S.Guo, J.Li, et al., Global landscape and trends of antibody-drug conjugates in oncology: A machine learning–driven decade-long informatics analysis. Journal of Clinical Oncology, 2026.
Y.Li, Y.Zhu, Y.Chen, et al., Advances in antibody-drug conjugates in cancer: latest updates from the 2026 AACR annual meeting. Experimental Hematology & Oncology, 2026.
Shanghai Duoning Biotechnology Co., Ltd. Is a leading one-stop bioprocess solutions provider, dedicated to offering comprehensive solutions for biopharmaceutical products from R&D to commercial manufacturing, including reagents and consumables, instruments and equipment, and services. The company primarily operates two business segments: bioprocess solutions and laboratory products & services. Through its one-stop bioprocess platform, we help partners achieve efficient, stable, and cost- and quality-controlled drug development and manufacturing processes.
Learn more: www.duoningbio.com
Media: PR@duoningbio.com
Swipe left/right to switch languages (CN/EN)
