抗体偶联药物（ADC）通过抗体识别肿瘤相关抗原并递送高效细胞毒载荷，是肿瘤精准治疗的重要药物形式。然而，靶向递送并不必然带来低毒性或宽治疗窗。ADC疗效与毒性能否相对分离，取决于肿瘤与正常组织靶抗原表达差异能否转化为有效暴露差异，并受抗原空间分布、内吞转运、连接子稳定性、载荷扩散、抗原非依赖性摄取及患者筛选等因素影响。围绕疗效-毒性分离，综述ADC靶抗原识别、空间异质性、微环境因素、靶点相关非肿瘤组织毒性及工程化和临床优化策略。ADC优化不应单纯追求更强杀伤，而应通过减少无效毒性暴露、控制载荷释放空间和优化剂量，使肿瘤有效暴露达到治疗阈值，同时使关键正常组织暴露低于不可耐受毒性阈值。建立“靶点表达-药物富集-载荷释放-毒性表型-人群剂量策略”的闭环证据链，有助于ADC靶点筛选、治疗窗评价和临床开发。

Antibody-drug conjugates (ADC) have emerged as a transformative therapeutic modality in precision oncology by selectively delivering highly potent cytotoxic payloads to tumor cells through antibody-mediated recognition of tumor-associated antigens. However, targeted delivery does not automatically translate into a favorable therapeutic window. The extent to which antitumor efficacy can be separated from treatment-related toxicity depends not only on differential antigen expression between malignant and normal tissues, but also on a complex interplay of factors including antigen density and spatial distribution, internalization and intracellular trafficking kinetics, linker stability, payload permeability and bystander effects, target-independent uptake, and patient-specific biological characteristics. This review examined the concept of efficacy-toxicity separation as a central framework for ADC development. Current advances were summarized on target antigen discovery and validation, spatial and temporal heterogeneity of antigen expression, tumor microenvironmental determinants of ADC activity, mechanisms underlying on-target/offtumor and off-target toxicities, and emerging engineering and clinical strategies aimed at improving therapeutic selectivity. Increasing evidence suggests that successful ADC development should not focus solely on maximizing cytotoxic potency, but rather on minimizing non-productive toxic exposure, controlling the spatiotemporal release of payloads, and optimizing patient selection and dosing strategies. The ultimate goal is to achieve sufficient intratumoral payload exposure to exceed efficacy thresholds while maintaining exposure in critical normal tissues below toxicity thresholds. We further propose that an integrated evidence framework linking target expression, tissue drug accumulation, payload release, toxicity phenotypes, biomarker-guided patient stratification, and dose optimization may provide a rational foundation for target prioritization, therapeutic-window assessment, and next-generation ADC development.

R979.1

国家自然科学基金资助项目（82272767）；天津市自然科学基金联合基金资助项目（25JCLZJC00430，25JCLMJC00440）；耀动神州-药学科研能力建设基金（ Z04J2023E095）；天津市医学重点学科建设资助项目（ TJYXZDXK-3-003A）