Therapeutic Strategies to Overcome Payload Resistance of Trastuzumab Deruxtecan in HER2‐Positive Cancers

ABSTRACT

Antibody–drug conjugates (ADCs) are emerging as a promising class of targeted cancer therapy. Trastuzumab deruxtecan (T‐DXd), a human epidermal growth factor receptor 2 (HER2)‐directed ADC, has demonstrated clinical efficacy in HER2‐positive gastric and breast cancers, as well as in HER2‐mutant non‐small cell lung cancer. However, the development of acquired resistance limits their long‐term efficacy. To elucidate the resistance mechanism, we established T‐DXd‐resistant cell lines derived from HER2‐amplified gastric xenografts (N87 acquired resistance [AR]) and leptomeningeal carcinomatosis (Calu‐3 AR) lung cancer cells. N87 AR cells exhibited cross‐resistance to T‐DXd, payload DXd, and topoisomerase I inhibitor SN‐38 despite preserved HER2 expression and intact drug internalization. As payload resistance‐related molecules, ATP‐binding cassette (ABC) transporter ABCG2 and ABCB1 were markedly upregulated in N87 AR and Calu‐3 cells, respectively. Inhibition of ABCG2 and ABCB1 in N87 AR and Calu‐3 cells, respectively, through siRNA‐mediated knockdown restored T‐DXd sensitivity in both models. As a strategy to overcome resistance, pharmacological inhibitors of ABCG2 and ABCB1 restored the T‐DXd sensitivity of N87 AR and Calu‐3 cells, respectively. Moreover, BB‐1701, a novel HER2‐ADC containing eribulin as a payload, to which N87 AR cells are sensitive, exhibited antitumor effects in N87 AR cells in vitro and in vivo. These findings indicate that ABC transporter‐mediated drug efflux is an important mechanism underlying T‐DXd resistance in HER2‐positive gastric and lung cancer models. Furthermore, our study suggests that both targeting drug efflux pathways and utilizing alternative payloads may be effective strategies for overcoming T‐DXd resistance in HER2‐positive gastric and lung cancers.

Overexpression of ABC transporters, especially ABCG2 and ABCB1, mediates acquired resistance to T‐DXd in HER2‐positive cancers. Inhibition of these transporters restored T‐DXd efficacy, highlighting efflux‐mediated resistance as a potential therapeutic target.

Abbreviations

ABC

ATP‐binding cassette

ABCB1

ATP‐binding cassette subfamily B member 1 (MDR1)

ABCG2

ATP‐binding cassette subfamily G member 2 (BCRP)

ADC

antibody–drug conjugate

AR

acquired resistance

DXd

exatecan derivative

HER2

human epidermal growth factor receptor 2

siRNA

small interfering RNA

SN‐38

7‐ethyl‐10‐hydroxycamptothecin

T‐DXd

trastuzumab deruxtecan

TKI

tyrosine kinase inhibitor

1. Introduction

Antibody–drug conjugates (ADCs) are a class of anticancer agents that couple the target specificity of monoclonal antibodies with the cytotoxic potency of chemotherapy. A typical ADC comprises three components: a monoclonal antibody targeting a tumor‐associated antigen, a cytotoxic payload, and a chemical linker connecting them. This targeted delivery enables selective accumulation of cytotoxic agents within tumor cells, enhancing antitumor efficacy while limiting systemic toxicity relative to conventional chemotherapy [1, 2].

Trastuzumab deruxtecan (T‐DXd) is a human epidermal growth factor receptor 2 (HER2)–directed ADC that has demonstrated robust clinical activity in HER2‐positive gastric and breast cancers and in HER2‐mutant non‐small cell lung cancer (NSCLC) [3, 4, 5]. In DESTINY‐Gastric01, T‐DXd significantly improved objective response rate and overall survival in patients with previously treated HER2‐positive advanced gastric cancer [3]. In addition, activity in HER2‐positive solid tumors has been reported in a phase I study [6]. Despite these advances, a substantial proportion of patients develop acquired resistance that limits the durability of benefit. Elucidating the molecular mechanisms of T‐DXd resistance is therefore essential to optimize current treatment strategies and to guide the development of next‐generation ADCs [7].

Multiple mechanisms contribute to ADC resistance, including downregulation or loss of the target antigen, impaired internalization or intracellular trafficking, altered drug metabolism, and increased efflux of cytotoxic payloads by ATP‐binding cassette (ABC) transporters [8, 9]. For T‐DXd, clinical and preclinical studies have implicated decreased HER2 expression and mutations in TOP1 (which encodes topoisomerase I) as potential mediators of resistance [10, 11].

In this study, we sought to define mechanisms of acquired T‐DXd resistance in HER2‐positive gastric and lung cancers using xenograft‐derived and leptomeningeal dissemination–derived resistant models. We focused on ABC transporter upregulation and evaluated strategies to overcome resistance, including pharmacological blockade of transporters and deployment of HER2‐targeted ADCs with alternative payloads.

2. Materials and Methods

2.1. Establishment of T‐DXd‐Resistant Cell Lines

To generate T‐DXd‐resistant gastric cancer cells (N87 AR), N87 cells (5 × 10⁶) were subcutaneously injected into the flanks of 5‐week‐old male SHO mice (Jackson Laboratory, Japan). T‐DXd (5 mg/kg) was administered intravenously for three cycles, and the relapsed tumors were harvested and dissociated into single cells for culture.

T‐DXd‐resistant lung cancer cells (Calu‐3 AR) were generated by using a meningeal dissemination model. EGFP‐luciferase‐labeled Calu‐3 cells were injected into the cisterna magna. Following three doses of T‐DXd, relapsed tumor cells were collected and cultured ex vivo.

2.2. Live‐Cell Imaging and Quantitative Analysis of Intracellular ADC Uptake

T‐DXd and BB‐1701 were fluorescently labeled with HiLyte Fluor 647 (Dojindo, Japan). The NH2‐reactive dye (succinimidyl ester) covalently labels primary amines on the antibody (lysine residues), and the payloads (DXd/eribulin) were not labeled; therefore, this assay visualizes antibody internalization and early endocytic trafficking rather than long‐term intracellular payload retention. HER2‐amplified N87 and N87 AR cells were seeded on poly‐l‐lysine–coated glass‐bottom dishes (Matsunami) in DMEM with 0.5% FBS and incubated for 24 h. Cells were treated with 10 μL of labeled ADC (∼1 mg/mL, estimated) for 5 min at room temperature, washed once with PBS + 0.5% FBS, and immediately imaged. Time‐lapse imaging was conducted using the APEXVIEW APX100 digital imaging system (Evident, Japan) with a 60× oil objective (NA 1.42) and sCMOS camera. Autofocus tracking minimized focal drift, and images were acquired every 10 min for 2 h (exposure time: 2 s). The first frame was taken ~10 min post‐treatment. Image analysis was performed in ImageJ (NIH). Cytoplasmic puncta were defined as fluorescent foci ≥ 10 arbitrary units above local background and manually counted in ≥ 10 cells per condition. Pseudo‐coloring (ImageJ “Phase” LUT) enhanced visualization. For spatial analysis, the shortest distance between each punctum and the plasma membrane was measured to assess subcellular distribution.

3. Results

3.1. Establishment of T‐DXd Resistant HER2‐Amplified Gastric Cancer Cells

To investigate the mechanisms underlying acquired resistance to T‐DXd, we established a resistant HER2‐positive gastric cancer cell line using a xenograft model. HER2‐amplified N87 cells were subcutaneously inoculated into SHO mice. Mice were intravenously administered T‐DXd (5 mg/kg) and the treatment was repeated after tumor regrowth for a total of three cycles. Tumors that no longer responded were harvested, and the derived cells were designated as N87 acquired resistant (N87 AR) cells (Figure 1A).

FIGURE 1.

Establishment of T‐DXd resistant HER2‐amplified gastric cancer cells in vivo. (A) In vivo induction of T‐DXd resistance. N87 cells were injected subcutaneously into mice and treated with T‐DXd (5 mg/kg, intravenously) three times upon tumor progression. Resistant cells (N87 AR) were harvested and cultured. (B) Cell proliferation curves of N87 and N87 AR by MTT assay. (C) Crystal violet staining of N87 and N87 AR after 72 h exposure to T‐DXd (10 μg/mL). (D–F) N87 AR shows resistance to T‐DXd (D), DXd (E), and SN‐38 (F) based on 72 h MTT assays.

N87 AR cells exhibited a proliferation rate comparable to that of N87 cells in vitro (Figure 1B), and short tandem repeat (STR) analysis confirmed their clonal identity. Crystal violet staining after 72 h of exposure to T‐DXd (10 μg/mL) revealed resistance in N87 AR cells (Figure 1C).

To quantify drug sensitivity, MTT assays were performed after 72 h of exposure to T‐DXd, its cytotoxic payload DXd, and topoisomerase I inhibitor SN‐38 (Figure 1D–F). N87 cells had IC₅₀ values of 0.083 μg/mL for T‐DXd, 3.28 μM for DXd, and 0.013 μM for SN‐38. In contrast, N87 AR cells showed no measurable IC₅₀ values for any of these agents, indicating acquired resistance to T‐DXd, cross‐resistance to its payload, or other topoisomerase I inhibitors.

3.2. HER2 Expression and Dependency Are Maintained

To determine whether resistance involves alterations in HER2 expression or signaling dependency, we assessed HER family protein levels in N87 and N87 AR cells. Western blot analysis revealed no significant differences in the expression of HER2, epidermal growth factor receptor (EGFR), or HER3 (Figure 2A). Flow cytometry showed that surface HER2 levels in N87 AR cells were slightly lower than in parental N87 cells but remained clearly high (Figure 2B).

FIGURE 2.

HER2 dependency of N87 and N87 AR cells. (A) Expression of HER family proteins analyzed by western blot. (B) Surface HER2 levels determined by flow cytometry. (C–E) Cell viability after 72 h exposure to afatinib (C), lapatinib (D) and erlotinib (E). (F) HER2 knockdown via siRNA (10 nM, 48 h) confirmed by western blot. (G) HER2 depletion reduces cell viability in both N87 and N87 AR. **p < 0.005, ****p < 0.001.

Next, we examined sensitivity to HER family tyrosine kinase inhibitors (TKIs), including afatinib (pan‐HER), lapatinib (EGFR/HER2), and erlotinib (EGFR). After 72 h of exposure, IC₅₀ values for afatinib, lapatinib, and erlotinib were 0.026, 0.01, and 0.037 μM in N87 cells and 0.032, 0.011, and 0.050 μM in N87 AR cells, respectively (Figure 2C–E), indicating that TKI sensitivity was preserved.

To confirm HER2 dependency, HER2 knockdown experiments were conducted using siRNAs. Western blot analysis confirmed effective suppression of HER2 expression in both N87 and N87 AR cells (Figure 2F). Viability assays performed 48 h after transfection with 10 nM HER2 siRNA showed a reduction in cell viability to 59.88% (****p < 0.001) in N87 cells and 55.71% (**p < 0.005) in N87 AR cells, compared to control siRNA‐transfected cells (Figure 2G). These data indicate that HER2 signaling remains functionally essential in N87 AR cells despite their acquired T‐DXd resistance.

3.3. Drug Uptake and Subcellular Trafficking of T‐DXd Are Preserved in Resistant Cells

To investigate whether acquired resistance to T‐DXd was attributable to altered intracellular drug delivery, we performed high‐resolution live‐cell fluorescence imaging using HiLyte Fluor 647–conjugated T‐DXd. N87 parental and resistant (N87 AR) cells were incubated with the labeled ADC, and time‐lapse images were captured every 10 min over a 2 h period.

Both cell lines demonstrated progressive intracellular accumulation of fluorescent puncta in a time‐dependent manner, indicative of efficient T‐DXd internalization (Figure 3A). Pseudo‐colored images generated with ImageJ's “Phase” LUT enhanced the contrast of punctate structures. Notably, higher‐magnification views revealed that the fluorescent signals were predominantly localized to the periphery of the cytoplasm in both parental and resistant cells (Figure 3B), suggesting similar subcellular targeting and early endosomal localization.

FIGURE 3.

Intracellular accumulation and spatial distribution of fluorescently labeled T‐DXd in N87 and N87 AR cells. (A) Time‐lapse live‐cell imaging of N87 parental and T‐DXd–resistant (N87 AR) cells following exposure to HiLyte Fluor 647–labeled T‐DXd (fluorophore conjugated to the antibody moiety; payload not labeled). Images were acquired every 10 min for 2 h; representative images at 30‐min intervals are shown. Intracellular drug localization is visualized as red fluorescent puncta. Pseudo‐colored images (bottom panels) were generated using the ImageJ “Phase” LUT to enhance contrast and facilitate puncta detection. Scale bar, 10 μm. (B) Higher‐magnification views showing subcellular distribution of fluorescent puncta localized near the plasma membrane in both cell lines (arrowheads). Scale bar, 5 μm. (C) Quantitative analysis of intracellular antibody uptake over time. Puncta were defined as discrete cytoplasmic foci with mean fluorescence intensity ≥ 10 arbitrary units above background. Data represent the average puncta count per cell at each time point (mean ± SEM; n = 10 cells per group). (D) Spatial analysis of early trafficking, represented as the shortest distance between each fluorescent punctum and the nearest plasma membrane. The top panel illustrates the measurement strategy. No statistically significant differences in subcellular localization were observed between the two cell lines at any time point. Data are shown as mean ± SEM (n = 10 cells per group).

Quantitative analysis confirmed a comparable temporal increase in the number of fluorescent puncta per cell between the two groups (Figure 3C). Although N87 cells exhibited a marginally higher uptake, the difference did not reach statistical significance (n = 10 cells/group, unpaired two‐tailed t‐test). To further assess the intracellular trafficking dynamics, we measured the shortest distance between each fluorescent dot and the nearest plasma membrane. The spatial distribution of internalized T‐DXd was indistinguishable between N87 and N87 AR cells at all time points examined (Figure 3D).

Together, these findings rule out impaired internalization or altered subcellular trafficking as mechanisms underlying T‐DXd resistance in N87 AR cells. These results underscore that despite phenotypic resistance, the initial steps of ADC delivery—including membrane binding, endocytosis, and early trafficking—remain intact, pointing instead to post‐internalization mechanisms such as efflux or payload inactivation as likely contributors to resistance.

3.4. ABCG2 Overexpression Confers T‐DXd Resistance

To elucidate the mechanism underlying payload resistance, whole‐exome sequencing was performed on N87 AR cells, which revealed no TOP1 mutation. Transcriptomic analysis by RNA‐seq revealed an increased expression of multiple ABC transporters, including ABCC11, ABCC6, ABCC3, ABCC2, ABCA7, ABCG2, and ABCB8 (Figure S1A). The expression level of ABCG2 was significantly increased in N87 AR cells compared with that in N87 cells, with a log₂ fold change of +4.3 and an adjusted p‐value of 3.44 × 10⁻⁵². Based on the regularized log expression (RLE) values, ABCG2 expression increased from 20.9 (N87) to 488.6 (N87 AR), corresponding to a 23.4‐fold increase (Figure S1B).

Given that DXd is a known substrate of ABC transporters, the expression levels of ABCG2 (BCRP), ABCB1 (MDR1), and ABCC1 (MRP1) were analyzed. Western blotting and flow cytometry confirmed a marked increase in ABCG2 protein levels in N87 AR cells compared to N87 cells (Figure 4A,B).

FIGURE 4.

ABCG2 overexpression mediates T‐DXd resistance in N87 AR cells. (A) Expression of ABC transporters (ABCB1, ABCG2, ABCC1) by western blot. (B) Flow cytometry analysis of ABCG2 expression. (C) Knockdown of ABCG2 via siRNA (10 nM, 48 h). (D) Restoration of T‐DXd sensitivity in N87 AR after ABCG2 knockdown (72 h MTT assay). (E) Overexpression of ABCG2 in N87 via plasmid transfection (24 h). (F) Decreased T‐DXd sensitivity in ABCG2‐overexpressing N87 cells (72 h MTT assay). ABC, ATP‐binding cassette; T‐DXd, Trastuzumab deruxtecan.

To functionally validate the role of ABCG2 in AR resistance, N87 AR cells were transfected with siRNA targeting ABCG2 (10 nM for 48 h). Western blotting analysis confirmed the effective knockdown of ABCG2 (Figure 4C). In MTT assays, ABCG2 knockdown restored sensitivity to T‐DXd, with an IC₅₀ of 0.17 μg/mL, compared to no measurable IC₅₀ in control siSCR cells (Figure 4D).

Conversely, overexpression of ABCG2 in N87 cells conferred resistance. The IC₅₀ for T‐DXd in control vector‐transfected cells was 0.031 μg/mL, whereas in ABCG2‐overexpressing cells, the IC₅₀ was not reached at the highest tested concentration (Figure 4E,F).

3.5. ABCG2 Inhibitors Restore T‐DXd Sensitivity In Vitro

To explore the effect of chemical inhibition of ABCG2 on resistance, we performed MTT assays with T‐DXd and two ABCG2 inhibitors, Ko143 and KS176. The Bliss synergy scores for these combinations were 17.2 and 18.8, respectively, indicating synergistic effects (Figure S2A,B). Co‐treatment restored T‐DXd sensitivity in N87 AR cells, reducing cell viability to levels comparable to those observed in N87 cells (Figure S2C,D). Crystal violet staining confirmed the restored sensitivity (Figure S2E). In the MTT assay, exposure to T‐DXd alone (10 μg/mL) resulted in a 50.2% viability of N87 cells and 79.5% viability of N87 AR cells. Co‐treatment with Ko143 and KS176 reduced the viability of N87 AR cells to 44.3% and 48.0%, respectively (**p < 0.005, ***p < 0.001, and ****p < 0.0001) (Figure 5A). Similar sensitization was observed for DXd (10 μM), where viability decreased from 86.2% to 40.8% (Ko143) and 45.8% (KS176) (*p < 0.05, **p < 0.005) (Figure 5B). The viability of SN‐38 (10 μM) cells decreased from 72.2% to 20.0% (Ko143) and 19.2% (KS176) (***p < 0.001, ****p < 0.0001) (Figure 5C).

FIGURE 5.

ABCG2 inhibitors reverse T‐DXd resistance in vitro and in vivo. (A–C) Combination of Ko143 or KS176 (5 μM) with T‐DXd (A), DXd (B), or SN‐38 (C) restores drug sensitivity in N87 AR cells (72 h). *p < 0.05, **p < 0.005, ***p < 0.001, ****p < 0.0001. (D) Western blot showing increased levels of cleaved PARP and caspase‐3 in N87 AR cells treated with T‐DXd + Ko143 (48 h). (E) Cleaved caspase 3/7 activity with Ko143 and T‐DXd (24 h). **p < 0.005, ***p < 0.001. (F) In vivo, T‐DXd + Ko143 significantly reduced tumor volume in N87 AR xenografts. *p < 0.05. (G) Expression of apoptotic markers in xenograft tumors treated with T‐DXd ± Ko143. T‐DXd, Trastuzumab deruxtecan.

Western blot analysis revealed an increased expression of cleaved PARP and cleaved caspase‐3 in N87 AR cells after 48 h of co‐treatment with T‐DXd and Ko143 (Figure 5D). Caspase‐3/7 activity assays confirmed the induction of apoptosis, with the activity increasing from 1.2 (T‐DXd alone) to 1.8 (T‐DXd + Ko143) (**p < 0.005, ***p < 0.001) (Figure 5E).

3.6. ABCG2 Inhibition Sensitizes Tumors to T‐DXd In Vivo

To evaluate the therapeutic potential of ABCG2 inhibition in vivo, N87 AR cells were subcutaneously implanted in SHO mice. The mice were randomly divided into two groups (n = 3 each): one group received T‐DXd monotherapy (5 mg/kg, intravenous), and the other received combination therapy with T‐DXd and Ko143 (30 mg/kg, oral, daily). T‐DXd monotherapy did not result in tumor regression, whereas combination therapy significantly reduced the tumor volume (*p < 0.05) (Figure 5F).

No remarkable loss of body weight was observed in the Ko143 group (Figure S3A).

In a separate cohort, tumors were harvested 72 h after treatment with control, T‐DXd alone, or T‐DXd + Ko143. Western blot analysis revealed increased levels of cleaved PARP and cleaved caspase‐3 in tumors from the combination group, indicating enhanced apoptosis (Figure 5G).

3.7. BB‐1701, an Eribulin‐Based HER2‐ADC, Overcomes Resistance

To assess whether alternate HER2‐targeted ADC could overcome resistance in N87 AR cells, we evaluated BB‐1701, a novel eribulin‐conjugated HER2‐ADC [12]. Eribulin is a microtubule inhibitor known to be a substrate of ABCB1, but not ABCG2 [13].

In 72 h MTT assays, eribulin showed potent cytotoxicity in both N87 and N87 AR cells, with IC₅₀ values of 0.059 and 0.026 μM, respectively (Figure 6A). N87 AR cells remained resistant to T‐DXd (IC₅₀ not reached) but were sensitive to BB‐1701, with an IC₅₀ of 0.11 μg/mL (Figure 6B).

FIGURE 6.

BB‐1701 overcomes T‐DXd resistance via an alternative payload. (A) N87 AR cells retain sensitivity to eribulin (72 h MTT). (B) BB‐1701, but not T‐DXd, reduces N87 AR viability (72 h MTT). (C) Western blot analysis of apoptotic proteins after BB‐1701 or T‐DXd treatment (48 h). (D) Cleaved caspase 3/7 activity after 24 h exposure. ***p < 0.001. (E) Tumor volume reduction in N87 AR xenografts treated with BB‐1701 versus T‐DXd. ***p < 0.001. (F) Increased apoptosis in BB‐1701‐treated tumors confirmed by western blot. T‐DXd, Trastuzumab deruxtecan.

After 48 h of treatment with T‐DXd or BB‐1701 (10 μg/mL), western blotting revealed that T‐DXd modestly induced cleaved PARP and cleaved caspase‐3, whereas both markers were strongly upregulated by BB‐1701 in N87 AR cells (Figure 6C). Caspase‐3/7 activity further confirmed the induction of apoptosis, increasing from 1.4 in T‐DXd‐treated cells to 2.6 in BB‐1701‐treated cells (***p < 0.001) (Figure 6D).

In vivo, N87 AR xenografts were subcutaneously established in SHO mice. Tumor‐bearing mice (n = 3 per group) were intravenously administered 5 mg/kg T‐DXd or BB‐1701. BB‐1701 significantly inhibited tumor growth compared to T‐DXd (***p < 0.001) (Figure 6E). No significant loss in body weight was observed in the BB‐1701 group (Figure S3B). Western blotting of tumors harvested 72 h post‐treatment confirmed the higher expression of cleaved PARP and cleaved caspase‐3 in the BB‐1701‐treated group (Figure 6F).

3.8. BB‐1701 Internalization Is Preserved in Resistant Cells

To evaluate whether intracellular trafficking contributes to BB‐1701 sensitivity, HiLyte Fluor 647‐labeled BB‐1701 was applied to N87 and N87 AR cells. Live cell imaging was performed every 10 min for 2 h.

N87 and N87 AR cells exhibited progressive accumulation of intracellular fluorescent puncta, indicating uptake of BB‐1701. Pseudo‐colored images generated using the ImageJ LUT “Phase” mode facilitated visualization of cytoplasmic signals (Figure S4A).

Higher‐magnification images of the cell periphery revealed similar patterns of punctate fluorescence near the plasma membrane in both N87 and N87 AR cells (Figure S4B).

Quantitative analysis of intracellular fluorescence over time showed that both cell lines accumulated BB‐1701 in a time‐dependent manner (Figure S4C). Although N87 cells demonstrated slightly higher overall uptake, the difference was not statistically significant (n = 10 cells per group).

To examine the potential differences in intracellular trafficking, we measured the shortest distance between each fluorescent punctum and the nearest plasma membrane (Figure S4D). The average distances were comparable between the two groups at each time point, suggesting no alteration in subcellular localization associated with resistance.

3.9. Clinical Correlation Between ABCG2 Expression and T‐DXd Response

To assess the clinical relevance of ABCG2 expression, immunohistochemistry was performed on two of the three HER2‐positive gastric cancer patients who received T‐DXd at Kanazawa University Hospital in March 2025 and had available archival tissues.

In Case 1, a patient with HER2 IHC 3+ gastric cancer who received mFOLFOX6 plus trastuzumab preoperatively and weak (1+) but discernible levels of ABCG2 expression were observed. The patient developed liver metastases and received T‐DXd, achieving stable disease for 1 month. In Case 2, another patient with HER2 IHC 3+ gastric cancer with negative ABCG2 expression achieved a partial response to T‐DXd, with a duration of 4.7 months. These observations suggest that higher ABCG2 expression may be associated with poorer clinical response to T‐DXd (Figure S5A). Kaplan–Meier analysis using the TCGA‐STAD dataset showed that high ABCG2 expression was associated with significantly worse overall survival (p = 0.008, Figure S5B), whereas ERBB2 (HER2) expression was not significantly correlated with prognosis (p = 0.1204, Figure S5C).

3.10. ABCB1‐Mediated Resistance in a HER2‐Amplified Lung Cancer Leptomeningeal Metastasis Model

To establish a T‐DXd‐resistant HER2‐amplified lung cancer model, Calu‐3 cells were transduced with an EGFP‐luciferase construct and intrathecally injected into SHO mice to generate a leptomeningeal dissemination model. Tumor progression was monitored using the In Vivo Imaging System (IVIS) following intraperitoneal injection of d‐luciferin. After three doses of T‐DXd (5 mg/kg, intravenously), residual tumor cells were harvested and expanded ex vivo. These cells were designated Calu‐3 AR cells (Figure S6A). The MTT assay showed that the IC₅₀ of T‐DXd was 0.0048 μg/mL in Calu‐3 cells, whereas it exceeded 10 μg/mL in Calu‐3 AR cells (Figure S6B).

Whole‐exome sequencing revealed no mutations in TOP1, a molecular target of DXd. RNA‐seq analysis revealed the elevated expression of several ABC transporters, including ABCB9, ABCA5, ABCB1, ABCA3, ABCC3, and ABCA2 (Figure S7A). ABCB1 expression was significantly increased in Calu‐3 AR cells than in Calu‐3 cells, with a log₂ fold change of +1.6 and an adjusted p‐value of 2.3 × 10⁻⁹. Based on the RLE‐normalized expression values, ABCB1 expression increased from ~75.4 in Calu‐3 cells to ~235.2 in Calu‐3 AR cells, corresponding to a 3.1‐fold upregulation (Figure S7B).

Western blot analysis confirmed strong upregulation of ABCB1 in Calu‐3 AR cells compared to that in parental Calu‐3 cells, while ABCG2 and ABCC1 levels remained unchanged (Figure S8A). To assess functional relevance, 72 h quantitative viability assays were performed with or without the ABCB1 inhibitor tariquidar (5 μM) or verapamil (5 μM). The cell viability following T‐DXd (1 μg/mL) treatment was 26.1% in Calu‐3 cells and 63.1% in Calu‐3 AR cells. When combined with verapamil (5 μM) or tariquidar (5 μM), the viability of the Calu‐3 AR cells decreased to 24.6% and 23.2%, respectively (*p < 0.05, ***p < 0.001, Figure S8B).

Western blot analysis confirmed the effective ABCB1 knockdown in Calu‐3 AR cells following siABCB1 treatment (10 nM, 48 h) (Figure S9A). In a 72 h viability assay after T‐DXd treatment (10 μg/mL), cell survival decreased to 29.5% in the siABCB1‐treated group compared to 52.6% in the siSCR control group (**p < 0.005, Figure S9B).

3.11. Summary of Findings

In our models, T‐DXd resistance was largely associated with upregulation of ABCG2 in HER2‐positive gastric cancer (N87) and ABCB1 in HER2‐amplified lung cancer (Calu‐3) with leptomeningeal dissemination. These alterations are expected to limit the intracellular accumulation of the cytotoxic payload DXd. In these models, resistance can be at least partially reversed using ABC transporter inhibition or HER2‐ADCs such as BB‐1701, which carry alternative non‐substrate payloads (Figure 7).

FIGURE 7.

Summary diagram of the mechanism of action and resistance to T‐DXd via ABCG2 or ABCB1. ABCG2 or ABCB1 overexpression is associated with T‐DXd resistance. T‐DXd, Trastuzumab deruxtecan.

4. Discussion

ADCs deliver cytotoxics to antigen‐defined cells, yet resistance emerges via antigen loss, trafficking defects, payload metabolism, and drug efflux [1, 2, 7, 8, 9]. Using HER2‐amplified gastric (N87) and lung (Calu‐3) systems, we profiled T‐DXd resistance across these axes and found that efflux plays a major role in the resistance architecture in these models.

In N87‐AR, HER2 expression and dependence were preserved; no TOP1 mutations were detected, and quantitative imaging showed intact uptake and early trafficking, arguing against target loss or endocytic failure and implicating post‐internalization mechanisms [10, 11]. In these live‐cell imaging experiments, T‐DXd was labeled on lysine residues of the antibody moiety, and the released DXd payload itself was not visualized. Accordingly, our 2‐h time‐lapse assay is optimized to assess HER2‐mediated uptake and early endo/lysosomal trafficking rather than long‐term intracellular payload retention or efflux.

Transcriptomic and functional assays converged on ABCG2: overexpression conferred resistance, siRNA knockdown restored T‐DXd sensitivity, and Ko143/KS176 re‐sensitized cells, with in vivo enhancement of T‐DXd activity—consistent with the canonical role of ABC transporters and the efflux liability of topoisomerase‐I payloads [14, 15, 16, 17, 18, 19].

Because efflux is payload‐centric, payload switching constitutes a rational counterstrategy. T‐DXd itself overcame T‐DM1 insensitivity, a property linked to a membrane‐permeable payload and bystander effect [20, 21]. Extending this logic beyond TOP1 payloads, BB‐1701—a HER2‐ADC carrying eribulin—retained activity in ABCG2‐high, T‐DXd–resistant settings, increased apoptotic readouts, and suppressed xenografts, aligning with preclinical potency and early clinical signals [12, 22]. While eribulin can interact with ABCB1, its efflux‐susceptibility spectrum differs from ABCG2‐preferred TOP1 payloads, providing a mechanistic rationale for cross‐resistance escape.

Transporter context was tumor‐specific: in the HER2‐amplified leptomeningeal lung model, resistance mapped to ABCB1 without ABCG2 induction. Pharmacologic P‐gp modulation (tariquidar or verapamil) partially restored T‐DXd cytotoxicity, echoing human imaging/pharmacology data but also the long‐standing safety and interaction challenges that have limited routine clinical deployment [23, 24].

Across emerging clinical datasets of topoisomerase‐I–based ADCs, TOP1 mutations have been detected in plasma ctDNA mainly toward the end of treatment and at progression. In a metastatic breast‐cancer cohort, TOP1 mutations emerged at progression after 312–574 days on the first topoisomerase‐I–based ADC, and subsequent ADCs failed more rapidly—consistent with cross‐resistance once TOP1 alterations arose [10, 25]. By contrast, our models developed resistance within 70–120 days, raising the possibility that differences in exposure duration may contribute to an exposure‐dependent hierarchy, in which efflux‐based routes are favored after shorter courses, whereas TOP1‐mediated routes tend to emerge only after prolonged exposure. This proposed hierarchy remains hypothetical and will require validation in longitudinal clinical and preclinical studies.

Clinically, our small exploratory IHC series did not allow firm conclusions; trends were compatible with an efflux‐centered model and aligned with signals observed in public datasets [26]. Taken together, our data position ABCG2/ABCB1 as tractable resistance nodes in HER2‐positive cancers. In patients with intact HER2 but elevated efflux, the rational path is payload‐aware sequencing toward non‐TOP1 HER2‐ADCs, with transient, trial‐based transporter inhibition as a complementary strategy.

Our models are limited to two cell lines—N87 and Calu‐3. Until the findings are validated in other tumor types, such as breast cancer, any clinical correlations should be interpreted as exploratory. In addition, we did not directly quantify long‐term intracellular DXd retention. The lack of a detailed intracellular DXd time‐course represents another limitation of our study, and future work using direct payload imaging or subcellular LC–MS/MS will be required to fully characterize DXd kinetics.

Our study supports the concept that increased expression of ABC transporters is an important mechanism of acquired resistance to T‐DXd in HER2‐positive gastric and lung cancer models. Notably, in these models, this resistance could be at least partially overcome by co‐treatment with ABC transporter inhibitors or by utilizing novel HER2‐directed ADCs that incorporate alternative non‐substrate payloads. However, pharmacologic ABC‐transporter inhibition remains experimental, with drug–drug interactions and safety constraints; routine clinical use is not established and should be limited to prospective trials. By contrast, biomarker‐guided payload selection appears actionable—ABCG2‐high contexts may favor non–TOP1 payloads, whereas ABCB1‐high settings warrant caution with eribulin‐based ADCs. Collectively, these findings highlight the therapeutic potential of efflux‐targeted strategies and rational payload design to overcome ADC resistance.

Author Contributions

Yuya Murase: conceptualization, data curation, formal analysis, methodology, software, validation, visualization, writing – original draft. Shigeki Nanjo: methodology, supervision. Sachiko Arai: formal analysis, validation, visualization. Sota Kondo: formal analysis, visualization, writing – original draft. Hayato Koba: methodology, writing – review and editing. Yifeng Liu: validation. Koji Fukuda: methodology. Shigeki Sato: methodology. Jun Kinoshita: investigation, resources. Noriyuki Inaki: investigation, resources. Tsukasa Ueda: validation. Shunichi Nomura: validation. Yuichi Tambo: writing – review and editing. Takahiro Shimizu: validation. Masafumi Horie: investigation. Daichi Maeda: investigation. Richard W. Wong: formal analysis, methodology, writing – original draft. Kazuyoshi Hosomichi: formal analysis, methodology, software. Takafumi Kobayashi: writing – review and editing. Satoshi Watanabe: writing – review and editing. Kenta Yamamura: writing – review and editing. Noriyuki Ohkura: writing – review and editing. Miki Abo: writing – review and editing. Seiji Yano: conceptualization, data curation, funding acquisition, methodology, project administration, resources, software, supervision, visualization, writing – original draft.

Funding

This research was supported by grants from JSPS KAKENHI [Grant Number 23K27609 (Seiji Yano)] and Bliss Biopharmaceutical (Hangzhou) Co. Ltd.

Disclosure

Seiji Yano is an editorial board member of Cancer Science.

Ethics Statement

All experimental procedures of this study were reviewed and approved by the Ethics Committee of Kanazawa University Hospital (2023–082 [114334]). Informed consent was obtained from all patients or waived by the IRB under institutional guidelines for the use of archived tissues. Animal Studies: All animal experiments were approved by the Institutional Animal Care and Use Committee of Kanazawa University (approval no. AP‐23060) and conducted in accordance with the Guidelines for the Care and Use of Laboratory Animals in Japan.

Conflicts of Interest

Seiji Yano received research funding from Bliss Bio Co. Ltd., Eisai Co., Takeda Pharmaceutical Company Limited, and Janssen Pharmaceutical K.K. Ltd. Seiji Yano has also received lecture fees from Chugai Pharmaceutical Co. Ltd., Pfizer Inc., Eli Lilly Japan K.K., Daiichi Sankyo Co. Ltd., MSD, and Eisai Co. Ltd. Shigeki Nanjo has received speaker fees from Chugai Pharmaceutical, Eli Lilly, Bristol Myers Squibb, AstraZeneca, Taiho Pharmaceutical, Ono Pharmaceutical Co., and Kyowa Kirin, and research grants from MSD K.K. and Revolution Medicines. Hayato Koba has received research funding from JSPS KAKENHI (Grant Number 23K14614). Yuichi Tambo has received research funding from JSPS KAKENHI (Grant Number 22K15577), AstraZeneca K.K., BeiGene, Bristol Myers Squibb (BMS), Daiichi Sankyo Company, Limited, MSD, Regeneron, Ono, and Gilead. Yuichi Tambo has also received lecture fees from Chugai Pharmaceutical Co. Ltd., AstraZeneca K.K., Taiho Pharmaceutical Co. Ltd., MSD, Pfizer Inc., Kyowa Kirin Co. Ltd., Takeda Pharmaceutical Company Limited, BMS, and Daiichi Sankyo Company, Limited. The other authors have not received any funding and declare no conflicts of interest.

Supporting information

Figure S1: RNA sequencing of N87 and N87 AR cells.

Figure S2: Synergistic effects of ABCG2 inhibitors and T‐DXd.

Figure S3: Changes in mice body weight in vivo. 25 20 30 T‐DXd T‐DXd + Ko143 T‐DXd BB‐1701 A. No appreciable changes in body weight are obs.

Figure S4: Intracellular uptake and spatial distribution of the fluorescent‐labeled BB‐1701 in N87 and N87 AR cells.

Figure S5: ABCG2 expression in clinical gastric cancer specimens.

Figure S6: Establishment of T‐DXd resistant HER2‐amplified lung cancer cells in vivo.

Figure S7: RNA sequencing of Calu‐3 and Calu‐3 AR cells. Calu‐3 AR.

Figure S8: ABCB1 inhibitors reverse T‐DXd resistance in vitro.

Figure S9: siABCB1 reverses T‐DXd resistance in vitro.

Table S1: Key resources table.

Data Availability Statement

The data supporting the findings of this study are available within the article and its Supporting Information. Public datasets used for the survival analysis are available from TCGA‐STAD and KMplot as indicated. Additional information is available from the corresponding author upon reasonable request. Details are provided in Supporting Information (Methods S1; Table S1).