Executive Summary / Key Takeaways
Antibody–drug conjugates (ADCs) — engineered therapeutics that couple a tumor-targeting monoclonal antibody to a potent cytotoxic payload via a chemical linker — represent one of the most rapidly expanding drug classes in oncology. Despite remarkable clinical successes, primary and acquired resistance limits durable benefit in a substantial proportion of patients. Resistance is multifactorial and operates across every step of the ADC life cycle: antigen expression and heterogeneity, antibody binding and internalization, intracellular trafficking, lysosomal processing, payload release, drug efflux, payload-target adaptation, and tumor microenvironment (TME) barriers. Critically, resistance mechanisms are not uniform — they differ materially by target antigen, linker type (cleavable versus non-cleavable), payload class, drug-to-antibody ratio (DAR), and membrane permeability. Understanding the mechanistic landscape of ADC failure is essential for patient selection, rational sequencing, and the design of next-generation therapeutic strategies.
1. Antigen Expression Heterogeneity: The Upstream Bottleneck
Antigen-level alterations represent the most proximal resistance mechanism. Tumor heterogeneity — both spatial (varying antigen density across tumor regions) and temporal (evolving expression before and after ADC exposure) — directly impairs drug delivery. In the KRISTINE trial of trastuzumab emtansine (T-DM1) combined with pertuzumab, patients with high HER2 heterogeneity (defined as HER2-negative areas in ≥10% of tumor cells by fluorescence in situ hybridization, FISH) achieved a pathologic complete response (pCR) rate of 0%, and experienced shorter progression-free survival (PFS) and overall survival (OS) 1216. Similarly, the ZEPHIR trial demonstrated that high tumor heterogeneity on HER2-positron emission tomography (PET) correlated with shorter time to treatment failure 12.
The DAISY trial of trastuzumab deruxtecan (T-DXd) further delineated a dose–response relationship between antigen expression and clinical outcome: objective response rates (ORR) were 70.6% in HER2-positive disease, 37.5% in HER2-low, and 29.7% in HER2-negative disease. Critically, non-responder tumors were enriched for clusters with low HER2 staining and stromal enrichment (cluster 6), indicating that spatial antigen architecture — not just average expression — is a resistance determinant 3. Dynamic HER2 status evolution carries independent prognostic significance: in a real-world analysis of 191 metastatic breast cancer patients treated with T-DXd, those whose disease progressed from HER2-low to HER2-zero had a median time-to-next-treatment (TTNT) of only 3.0 months (95% CI: 1.7–5.6), compared to 9.4 months in patients who maintained HER2-low status (p = 0.006) 13.
Beyond HER2, Nectin-4 downregulation during metastatic spread of urothelial carcinoma correlates with enfortumab vedotin (EV) resistance, and TROP-2 downregulation has been observed in triple-negative breast cancer (TNBC) cells exposed to sacituzumab govitecan (SG) 121416. Antigen masking (e.g., by mucin MUC4 shielding HER2) and accumulation of truncated extracellular domains (e.g., p95HER2) further compromise antibody binding 1216.
2. Internalization, Intracellular Trafficking, and Lysosomal Processing
Even when antigen engagement occurs successfully, resistance can arise at multiple downstream steps. Efficient ADC internalization requires clathrin-mediated endocytosis, followed by sequential trafficking through early endosomes to lysosomes where payload is liberated. Any disruption in this cascade attenuates cytotoxic throughput 216.
Caveolin-1 (CAV-1) overexpression redirects T-DM1 internalization from the canonical clathrin pathway to caveolin-coated vesicles, reducing lysosomal colocalization and decreasing drug sensitivity 1216. Loss of endophilin A2 (encoded by SH3GL1) similarly reduces HER2 internalization and T-DM1 cytotoxicity 12. Some resistant endosomes undergo recycling, returning ADC complexes to the cell membrane before payload liberation 16.
Within the lysosome, payload release depends on acidic pH and cysteine protease activity. The TARSC (target-responsive subcellular catabolism) framework demonstrated preclinically that: inhibition of clathrin-mediated endocytosis with chlorpromazine, disruption of lysosomal acidification with bafilomycin A1, or blockade of cysteine proteases each substantially reduce the key T-DM1 catabolite lys-MCC-DM1, confirming that each step constitutes an independent resistance node 2. SLC46A3 — a lysosomal transporter that exports maytansine-based catabolites into the cytoplasm — is another critical vulnerability: its silencing leads to catabolite accumulation within lysosomes and drug failure, particularly for ADCs with non-cleavable linkers 1216. T-DM1-resistant clones exhibiting elevated lysosomal pH, disturbed proteolytic activity, and T-DM1 accumulation within lysosomes have been characterized in BT-474 breast cancer models 12.
3. Payload Release, Drug Efflux, and Multidrug Resistance
Following lysosomal cleavage, released payloads must reach intracellular targets without being expelled by adenosine triphosphate (ATP)-binding cassette (ABC) transporters. This represents one of the most clinically actionable resistance mechanisms. Tubulin-binding payloads — including monomethyl auristatin E (MMAE, used in brentuximab vedotin and EV), monomethyl auristatin F (MMAF), and maytansinoids (DM1, DM4) — are recognized substrates of ABCB1 (MDR1/P-glycoprotein), ABCC2, and ABCG2 11216. Preclinical studies in neuroblastoma models confirmed that ABCB1-high cell lines display significantly elevated resistance to these payload classes, reversed by the ABCB1 inhibitor tariquidar, whereas DNA-binding payloads (e.g., pyrrolobenzodiazepine [PBD] dimers, PNU-159682) are not efflux substrates and remain unaffected 1. In HER2-positive gastric cancer models, upregulation of ABCC2 and ABCG2 drove T-DM1 resistance, and this was reversed by the transporter inhibitor MK571 12.
Importantly, intact ADCs are generally not efflux substrates — efflux becomes relevant only after lysosomal processing releases membrane-permeable catabolites. This means that linker chemistry, payload hydrophobicity, and release kinetics all modulate how significantly efflux activity impacts ADC potency 16. Hydrophilic linkers (e.g., PEG 4 Mal) show greater activity against MDR1-expressing tumors than non-polar linkers (e.g., SMCC) 12. Clinical translation of this mechanism was suggested by the observation that ABCC1 upregulation after T-DM1 exposure has been linked to poorer OS following T-DXd, implying cross-resistance patterns persist across payload generations 3.
4. Payload-Target Mutations and Downstream Pathway Adaptations
Resistance can also emerge through alterations in the intracellular target of the payload. For topoisomerase I inhibitor (TOP1i)-based ADCs such as T-DXd and datopotamab deruxtecan (Dato-DXd), loss-of-function mutations in SLX4 (a DNA-damage repair regulatory gene) were enriched at progression on T-DXd in the DAISY trial 312. TOP1 point mutations (e.g., E418K, frameshifts) reduce the enzyme's affinity for the DNA-drug complex, attenuating DXd and SN-38 efficacy 16. RB1 mutations have been reported in NSCLC patients with acquired T-DXd resistance 3. For microtubule-binding payloads such as MMAE, tubulin mutations can similarly compromise ADC cytotoxicity 16.
Downstream survival signaling further buffers against payload injury. Activation of the PI3K/AKT/mTOR pathway via PIK3CA mutations or PTEN loss reduces ADC responsiveness, although exploratory biomarker analysis of the EMILIA trial found that T-DM1 maintained OS and PFS benefit regardless of PIK3CA mutation status, suggesting some degree of pathway independence 12. BCL-2 and BCL-XL overexpression — primarily documented in hematologic malignancies — correlates with resistance to brentuximab vedotin (BV) and gemtuzumab ozogamicin 1219. TP53 mutations at baseline have emerged as a strong independent predictor of shorter TTNT in patients receiving T-DXd (multivariable hazard ratio [HR] = 4.02; p < 0.001) 13.
5. Tumor Microenvironment and Immune-Mediated Factors
The TME modulates ADC delivery and activity through multiple mechanisms. Cancer-associated fibroblasts (CAFs) and extracellular matrix deposition create stromal barriers to ADC penetration 3. Extracellular cathepsin L (CTSL) activity in the TME can paradoxically promote T-DXd efficacy in HER2-low/negative tumors by enabling linker cleavage independent of surface receptor expression 316. Conversely, CD47 upregulation on tumor cells can counteract ADC-induced immunogenic cell death (ICD), and preclinical combination of T-DXd with CD47 blockade restores antitumor immune activity 3.
Immune checkpoint inhibitor (ICI) combinations have been explored to leverage ADC-induced ICD. T-DXd promotes higher release of HMGB1 and calreticulin exposure — hallmarks of ICD — compared to T-DM1 311. A phase Ib trial of T-DXd plus nivolumab in pretreated HER2-expressing metastatic breast cancer reported confirmed ORRs of 59.4% (HER2+) and 37.5% (HER2-low) 13. Phase II/III trials (KATE2, BEGONIA, ASCENT-04/KEYNOTE-D19) show variable PFS benefit from ADC-ICI combinations, with the most consistent signals in PD-L1-positive or immune-rich tumors 313.
For BV, the clearest real-world evidence of immune context superiority came from a phase III trial in advanced classical Hodgkin lymphoma (cHL) in which nivolumab plus AVD (N-AVD) outperformed BV-AVD with a 2-year PFS of 92% versus 83%, suggesting that immune checkpoint blockade may provide a more durable platform than CD30-MMAE delivery in certain disease settings 19.
6. ADC Design Features That Define Resistance Profiles
Resistance profiles differ systematically by ADC design. Cleavable linkers (used in T-DXd, SG, Dato-DXd) enable payload release in the lysosome and support bystander killing of neighboring antigen-low cells — an advantage in heterogeneous tumors — but premature extracellular cleavage by TME proteases can reduce selectivity. Non-cleavable linkers (used in T-DM1) require complete intracellular catabolism and are entirely dependent on intact SLC46A3-mediated transport, making them highly vulnerable to lysosomal dysfunction 512. Higher DAR (T-DXd DAR 8:1 versus T-DM1 DAR 3.5:1) may partially compensate for efflux, but excessively high DARs accelerate antibody clearance, potentially reducing efficacy 12. Membrane-permeable payloads (e.g., MMAE, DXd) confer bystander activity but also efflux susceptibility, while charged payloads (e.g., MMAF) have minimal bystander effect but resist efflux 8. Novel auristatins combining both properties through hydrophobic N-terminal modification have been described preclinically 8. Payload class remains the strongest single determinant of resistance pathway: in a cross-payload neuroblastoma study, DNA-binding payloads had mean IC₅₀ of 25.6 pmol/L versus 943 pmol/L for tubulin-binding drugs, and efflux resistance was entirely absent for non-efflux-substrate payloads 1.
7. Biomarkers for Predicting and Monitoring Resistance
Static immunohistochemistry (IHC) has "largely failed to predict T-DXd activity" across the HER2 expression spectrum 13. Quantitative assays substantially outperform IHC: high-sensitivity HER2 (HS-HER2, measured in attomole/mm²) was continuously associated with TTNT (HR per 5-unit increment: 0.77; p < 0.001) and OS (HR: 0.79; p < 0.001); reverse-phase protein array (RPPA)-derived total HER2 was significantly associated with OS (HR 0.89; p = 0.015); and the HER2DX transcriptomic score independently predicted benefit in both HER2-positive and HER2-negative subsets 13. Dynamic plasma-based biomarkers offer noninvasive longitudinal monitoring: DNADX tumor fraction and HER2-signature subtypes significantly predicted TTNT and OS (p < 0.001), and post-treatment emergence of mutations in ARID1B, NFE2L2, FGFR1, and USP9X has been detected by serial circulating tumor DNA (ctDNA) profiling, suggesting adaptive resistance trajectories 13. TROP2 internalization status (QCS score) predicted datopotamab deruxtecan outcomes in TROPION-Lung01, highlighting target-specific internalization as a clinically measurable biomarker 3.
8. Strategies to Overcome Resistance
Several evidence-based strategies have emerged. Payload-class switching is the most validated: the DESTINY-Breast02 phase III trial established that T-DXd (TOP1i payload) significantly improved outcomes in patients who progressed on T-DM1 (maytansinoid payload), the first randomized evidence that one ADC can overcome resistance to another 13. Patient-derived T-DM1-resistant models retained full sensitivity to SYD985 (cleavable linker, more potent duocarmycin payload) and exatecan-based ADCs 613. Next-generation linker chemistries — including polysarcosine hydrophobicity-masking (PSAR) that stabilizes the linker and reduces clearance — enhanced bystander killing versus T-DXd in preclinical gastric cancer models 13. pH-sensitive peptide conjugates (e.g., CBX-12 using pHLIP-exatecan) provided antigen-independent tumor targeting, achieved complete xenograft regressions with minimal bone marrow toxicity, and synergized with PARP inhibitor talazoparib in HER2-negative models 13. Bispecific/biparatopic ADCs targeting two non-overlapping HER2 epitopes (e.g., zanidatamab zovodotin [ZW49]) enhance antigen clustering and internalization, showing a 31% ORR and 70% disease control rate in first-in-human trials 12. Molecular glue-antibody conjugates (MACs), which use targeted protein degraders as payloads, entered clinical trials by 2025 and represent a mechanistically distinct approach to circumventing classical payload resistance 9. For combination strategies, ADC-ICI combinations showed promising early signals, and ADC-PARP inhibitor combinations appear feasible with improved ADC selectivity, though phase III validation across indications remains ongoing 13.
Standardized Table: ADC Resistance Mechanisms, Clinical Relevance, Biomarkers, and Potential Solutions
| Resistance Mechanism | Biological Basis | Clinical Examples | Biomarkers | Mitigation Strategies |
|---|---|---|---|---|
| Antigen downregulation / heterogeneity | Clonal selection of antigen-low cells; antigen shedding; epitope masking (MUC4, p95HER2); dynamic temporal loss | HER2-low → HER2-zero: TTNT 3.0 vs 9.4 months on T-DXd; KRISTINE: 0% pCR with high HER2 heterogeneity; Nectin-4 loss in urothelial carcinoma metastasis; TROP-2 downregulation in TNBC after SG 31216 | HS-HER2, RPPA HER2, HER2DX, spatial IHC, HER2-PET, single-cell RNA-seq, ctDNA | Quantitative antigen assays; bispecific/biparatopic ADCs; bystander-active payloads; serial biopsy |
| Impaired internalization / trafficking | CAV-1-mediated rerouting away from clathrin pathway; SH3GL1 loss; HSP90 inhibition-induced receptor instability | T-DM1 resistance in HER2+ cells with CAV-1 overexpression; SH3GL1 knockdown reduces T-DM1 cytotoxicity 1216 | CAV-1 expression, SH3GL1 levels, flow cytometry internalization assays, EGFR/HER2 dimerization state | Biparatopic antibodies; TKIs (lapatinib, neratinib) to enhance internalization; cleavable linkers |
| Lysosomal dysfunction / processing defect | Elevated lysosomal pH; reduced cathepsin activity; SLC46A3 loss; impaired acidification | T-DM1-resistant BT-474M1 cells: SLC46A3 silencing → catabolite retention; elevated lysosomal pH and reduced cathepsin activity in resistant clones 21216 | SLC46A3, cathepsin activity assays, lysosomal pH probes | Cleavable linkers; acidifying nanoparticles; alternative protease-independent release mechanisms |
| Drug efflux (ABC transporters / MDR) | ABCB1/MDR1, ABCC2, ABCG2 upregulation exports membrane-permeable payloads | ABCB1-high neuroblastoma cells: efflux of MMAE, DM1, DM4 (reversed by tariquidar); ABCC2/ABCG2 drive T-DM1 resistance in gastric cancer; ABCC1 post-T-DM1 linked to worse T-DXd OS 1312 | ABCB1/MDR1 mRNA/protein, functional efflux assays, payload substrate prediction | Non-efflux-substrate payloads (PBD, duocarmycin); hydrophilic linkers (PEG 4 Mal); higher DAR; payload-class switching |
| Payload-target mutations | TOP1 point mutations (E418K, frameshifts); tubulin alterations; RB1 loss | DAISY: SLX4 mutations enriched at T-DXd progression; NSCLC: RB1 mutations in T-DXd-resistant cases; TOP1 mutations causing cross-resistance to subsequent TOP1i 316 | ctDNA TOP1/RB1/SLX4 sequencing, tumor biopsy genotyping | Payload-class switching; PARP inhibitor + TOP1i ADC combinations; dual-payload ADCs |
| Downstream apoptosis / signaling escape | BCL-2/BCL-XL overexpression; PI3K/AKT/mTOR activation; TP53 mutation; cyclin B1 loss; Wnt/β-catenin activation | BCL-2/BCL-XL: resistance to BV and gemtuzumab ozogamicin; TP53 mutation: HR 4.02 for shorter TTNT on T-DXd; cyclin B1 loss in T-DM1-resistant cells 121319 | BCL-2/BCL-XL IHC, TP53 ctDNA, PI3K/PTEN mutation status, cyclin B1 expression | BCL-2 inhibitors (venetoclax); PI3K/mTOR inhibitors; biomarker-stratified trial designs |
| Tumor microenvironment barriers | CAF-mediated rescue; stromal matrix deposition; CD47 upregulation; reduced immune infiltration | DAISY cluster 6: low HER2 + stromal enrichment in non-responders; CD47 upregulation counteracts T-DXd ICD 313 | CAF markers (α-SMA, FAP), TIL density, CD47 expression, PD-L1 IHC, immune transcriptomics | ADC + ICI (T-DXd + nivolumab: ORR 59.4% HER2+); CD47 blockade; anti-stromal agents |
| Antigen-independent TME release / linker instability | Premature extracellular payload release by TME proteases; maleimide deconjugation in plasma | Dato-DXd: variable TROP2 internalization by tumor type; plasma linker deconjugation observed for some ADCs 313 | Linker stability assays, QCS internalization score, DAR pharmacokinetics | Stabilized linkers; glucuronidase-cleavable linkers; site-specific conjugation; optimized DAR (2–4) |
| ADC design limitations: DAR, bystander, Fc | High DAR → accelerated clearance; low bystander activity in heterogeneous tumors; Fc-silent engineering reduces ADCC/CDC | T-DM1 (non-cleavable, no bystander): disadvantaged in HER2-heterogeneous tumors; Fc-silent ADCs: weaker immune effector activation 51116 | DAR by mass spectrometry, pharmacokinetic profiling, functional ADCC/CDC assays | Site-specific conjugation; bystander-active cleavable payloads; Fc-optimized ADCs; dual-payload formats |
2026 Outlook and Implications for ADC Development and Clinical Practice
As of mid-2026, the ADC landscape encompasses over 17 approved agents globally, with an extensive pipeline including dual-payload formats, bispecific ADCs, and molecular glue conjugates 912. Several principles should guide clinical practice and development strategy:
First, static IHC-based biomarkers are insufficient for predicting durable ADC benefit. Quantitative platforms — including HS-HER2, RPPA, HER2DX transcriptomics, and plasma ctDNA signatures — substantially outperform conventional assays and warrant prospective integration into trial design and clinical decision-making 13.
Second, payload-class switching is the most validated resistance-circumvention strategy to date. DESTINY-Breast02 established the first randomized proof that T-DXd overcomes T-DM1 resistance, and preclinical data support that exatecan-based ADCs can overcome resistance to DXd-class agents in some models 13. However, cross-resistance between ADCs using the same payload mechanism may limit the benefit of same-class sequencing.
Third, combination strategies offer the most promising path to durability, particularly ADC plus ICI in immune-responsive tumors and ADC plus PARP inhibitor in homologous recombination-deficient (HRD) cancers. The TROPION-Breast01 phase III trial demonstrated that while Dato-DXd significantly improved PFS over chemotherapy (HR 0.63; p < 0.0001), OS was not significantly improved (HR 1.01), with post-progression ADC access in the control arm confounding long-term outcomes — underscoring the need for rational first- and second-line sequencing strategies 17.
Fourth, serial biomarker monitoring — particularly ctDNA for TP53 status, HER2 signature evolution, and emerging resistance mutations — represents an evolving clinical tool that may allow earlier detection of resistance and more timely therapeutic adjustment 13.
The convergence of mechanistic understanding, next-generation ADC engineering, and multi-omic biomarker development positions the field to meaningfully extend the durable benefit of ADC therapy in the years ahead. For clinicians, translational researchers, and development teams alike, a granular understanding of the resistance landscape is no longer optional — it is foundational to realizing the full potential of this transformative therapeutic class.