Resistance to antibody-drug conjugates in breast cancer: mechanisms and solutions

5.1.1 Antibody-mediated resistance

Reduced target expression is a common mechanism for inadequate antigen-antibody binding. For instance, the correlation between higher HER2 levels and greater T-DM1 efficacy was observed in a phase II trial [61]. Decreased HER2 expression and reduced binding were noticed in T-DM1-resistant cell lines from separate studies, which indicated HER2 loss as a mechanism of T-DM1 resistance [62, 63]. The absence of target expression has also been noticed in a TNBC patient with de novo resistance to IMMU-132, who showed undetectable expression of TROP2. In the same study, mutated TACSTD2 (encoding TROP2) was detected in another patient with acquired resistance to IMMU-132, which led to reduced binding due to an altered subcellular localization of TROP2 [64]. Interestingly, although the appearance of TROP2 is necessary, higher level of TROP2 does not necessarily result in a better response [65]. Low levels of antigen expression were also noticed in other cancer types. Downregulation of CD30 induced brentuximab vedotin-resistant in Hodgkin lymphoma [66], and reduced CD33 expression level on leukemic cells was associated with worse outcomes of gemtuzumab ozogamicin in acute myeloid leukemia [67].

5.1.2 Impaired drug trafficking

Compromised endocytosis is considered another mechanism of ADC resistance. Many proteins are essential in the ADC cellular internalization process. Endophilin A2 (Endo II) is a scaffolding protein and is involved in clathrin-independent endocytosis. Impaired Endo II expression is related to decreased HER2 internalization and reduced response to T-DM1 in HER2+ breast cancer models [68]. Aberrant caveolae-mediated endocytosis has also been noticed in T-DM1-resistant cell lines. Enhanced caveolae-mediated endocytosis was noticed in a gastric cancer N87 cell line with acquired T-DM1 resistance, possibly due to caveolae not favoring drug trafficking to lysosomes [69]. However, knockdown of caveolin-1 did not appear to restore the sensitivity to T-DM1. In another study, upregulated caveolin-1 appeared to improve T-DM1 activity in a different cell line (BT-474) [70]. Furthermore, hypoxia-induced translocation of caveolin-1 from vesicles to the plasma membrane was suggested to be the possible mechanism of reduced trastuzumab internalization in the hypoxia microenvironment in breast cancer SKBR3 cells [71]. In the study conducted by Sung et al. [69], they noticed that the ADC:CAV1 ratio differs in different cell lines, which indicates the caveolae-mediated endocytosis weighs differently, and the impact of CAV-1 level. In conclusion, the impact of caveolae-mediated endocytosis in resistance to T-DM1 remains debatable possibly related to the leading endocytosis pathway of the target antigen in a certain cell line.

Except for impaired intracellular trafficking, the way of ADCs trafficking in the circulation system also contribute to drug resistance. T-DM1 is a large hydrophilic molecule and is difficult to diffuse through the blood-brain barrier, which shares a similar tissue distribution with trastuzumab. However, in preclinical studies, T-DM1 demonstrated active in trastuzumab-resistant mouse models with breast cancer brain lesions due to the cytotoxicy of the payload component, DM1 [72]. However, patients with brain metastases did not achieve improved mPFS in T-DM1 group compared to the control group in the TH3RESA trial [73]. Comparable results were also reached in the KATHin ERINE trial, where similar central nervous system recurrence rates were noticed with or without T-DM1 [41]. Notably, breast cancer with brain metastases appears to be better managed using DS-8201a. In the DESTINY-Breast03 trial, patients with brain metastases benefited more with DS-8201a than T-DM1 [53]. Encouraging results were also yielded in the phase II TUXEDO-1 trial, 11 out of 15 patients with brain metastases respond to DS-8201a treatment, including 70% had received prior T-DM1 treatment (NCT04752059) [74].

5.1.3 Disrupted lysosomal function

The lysosomal degradation of ADCs depends on the acidic lysosomal environment and active lysosomal enzymes. Lysosomal alkalization and impaired lysosomal proteolytic enzyme activity were identified in a T-DM1-resistant breast cancer cell line [75]. The cleavage of non-cleavable ADCs depends on the activity of lysosomal enzymes which require a highly acidic environment. The acidic environment is ensured by V-ATPase, a proton pump that regulates lysosomal acidification [76]. Aberrant V-ATPase activity was tested in T-DM1-resistant N87 gastric cancer cell line [77]. The inhibition of V-ATPase by bafilomycin A1 reduced the production of active metabolites of T-DM1 and thus the cytotoxicity of T-DM1 in N87 gastric cancer cells but not in T-DM1-resistant N87 cells [77]. A comparable situation was also noticed in a T-DM1-resistant breast cancer cell line, where lysosomal alkalization and impaired lysosomal proteolytic enzyme activity were observed in BT474 cells [75].

Cytotoxic payloads need to be transported across the lysosomal membrane to exert their cytotoxic effect. As for ADCs incorporated with membrane-impermeable payloads, transporters are required for payload release. SLC46A3 belongs to the solute carrier (SLC) transporter family and serves as a transporter on the lysosomal membrane of maytansine-based catabolites. Decreased levels of SLC46A3 lead to an accumulation of maytansine-based catabolites in lysosomes, while ADCs carrying auristatin-based monomethylauristatin F (MMAF) remain unaffected [21]. Loss of SLC46A3 conferring resistance to T-DM1 was also confirmed in a different study using T-DM1-resistant BT-474M1 cells [63]. A similar mechanism was also noticed in brentuximab vedotin for the treatment of lymphoma. Lysosomal multidrug-resistance protein 1 (MDR1) mediates the efflux of monomethyl auristatin E (MMAE) across the lysosomal membrane. It was inferred that the inhibition of lysosomal MDR1 enhanced the cytotoxicity of brentuximab vedotin in the Hodgkin lymphoma KM-H2 cell line [78].

5.1.4 Payload-related resistance

Payloads are the main agent for the antitumor activity of ADCs. Some cells may develop resistance by upregulating drug efflux pumps and interrupting drug deposition. In a study of a non-Hodgkin lymphoma cell line that was made resistant to anti-CD22-vc-MMAE and anti-CD79b-vc-MMAE, upregulated MDR1 (encoded by ABCB1) expression was identified to be responsible for resistance to vc-MMAE-based conjugates [79]. Likewise, since vc-MMAE-based conjugates are utilized in breast cancer, the overexpression of MDR1 might also be a resistance factor for ADCs in breast cancer. Maytansinoids are another class of substrates for drug efflux transporters. Overexpression of multidrug resistance-associated protein 1 (MRP1, encoded by ABCC1), MRP2 (ABCC2), and MDR1 was demonstrated in different T-DM1-resistant cell lines, and their sensitivity could be restored by the concomitant use of MRP1, MRP2, and MDR1 inhibitors, respectively [62, 63, 80]. Similar mechanisms are also noticed in IMMU-132. Overexpression of breast cancer resistance protein (BCRP, encoded by ABCG2) was verified in IMMU-132-insensitive breast cancer cell lines, and the inhibition of ABCG2 helped to overcome resistance [81].

Tumor cells can also avoid cytotoxic effects via altered targets of payloads. A specific point mutation in TOPI was identified in TNBC patients who were resistant to IMMU-132, which has been previously described and is known to induce resistance to clinical TOPI inhibitors [64]. Other than TOPI mutations that directly disrupt the target of SN-38, the proficiency of the homologous recombinational repair (HRR) pathway is also related to IMMU-132 resistance by compensating for DNA damage caused by SN-38 [82]. Regarding ADCs with anti-mitotic agents like T-DM1, modifications in the microtubule/tubulin complex were found in T-DM1-resistant MDA-MB-361 cells [83].

Anti-mitotic agents arrest target cells at the G2/M phase of the cell cycle and lead to apoptosis. Cyclin B1/cyclin-dependent kinase 1 (CDK1) complex is essential for cell mitosis and mitotic catastrophe [84]. Some T-DM1-resistant cells manage to escape mitotic catastrophe and apoptosis through defective cyclin B1. This mechanism was verified in HER2+ breast cancer cell lines with various levels of acquired T-DM1 resistance, where T-DM1 failed to induce the upregulation of cyclin B1 and the consequent CDK1 activation. In addition, cyclin B1 knockdown induced T-DM1 resistance, while upregulation of cyclin B1 partially sensitized the resistant cells [85]. Moreover, the inability to eliminate genetically unstable cells might further increase tumor malignancy [86]. Besides cyclin B1, Bcl-2/Bcl-xl also participate in cell cycle regulation. Overexpression of Bcl-2/Bcl-xl is associated with resistance to gemtuzumab ozogamicin in acute myeloid leukemia and anti-CD79b-vc-MMAE in NHL cell lines [87, 88]. The inhibition of polo-like kinase 1 (PLK1) can rescue T-DM1 resistance via CDK1-dependent Bcl-2 phosphorylation [89], and the inhibition of Bcl-2/Bcl-xl significantly enhanced the antitumor activity of T-DM1 in PDX models that are either T-DM1-sensitive or T-DM1-resistant [90].

5.2.1 Impaired binding of HER2

The binding of mAbs to HER2 is related to multiple factors. HER2 shedding generates soluble truncated extracellular domain of HER2 molecules and leaves the 95-kDa intracellular domain, p95HER2 [91]. High level of p95HER2 is associated with resistance to trastuzumab and is also expected to confer to T-DM1 resistance. Furthermore, a membrane-bound glycoprotein mucin 4 (MUC4) was also considered a source of T-DM1 resistance. The tumor necrosis factor α (TNFα)-induced upregulation of MUC4 impaired binding and inhibited ADCC by masking the epitope of trastuzumab in HER2, while silencing MUC4 can restore sensitivity to T-DM1. This clinical relevance was proved in patients that the MUC4-positive tumors were associated with unfavorable DFS [92].

5.2.2 HER2 heterogeneity

HER2 intratumoral heterogeneity is referred to as different HER2 expression or amplification status within the same tumor, which can be identified by an area with HER2 amplification in >5% but <50% of tumor cells, or a HER2-negative area by fluorescence in situ hybridization (FISH) [93, 94]. Intratumoral HER2 heterogeneity presents more in breast cancers with low-grade HER2 amplification and equivocal HER2 expression [95].

T-DM1 showed limited efficacy when encountering tumors with heterogeneous HER2 expression. As noticed in the KRISTINE trial, none of the patients with HER2 intratumoral heterogeneity responded to T-DM1 therapy; and a lower pCR was more prevalent in tumors with heterogeneous HER2 expression than homogenous in the exploratory biomarker analysis [96]. This underlying association was also investigated in a phase II trial in early breast cancer patients, based on which HER2 heterogeneity is suggested to be related to T-DM1 resistance, where the fraction of HER2-nonamplified cells was the critical factor [93]. This mechanism was studied in vitro, where HER2 heterogeneity was mimicked by coculturing HER2+ KPL-4 cells and HER2- MDA-MB-468 cells. T-DM1 failed to suppress HER2- cells, while DS-8201a induced cytotoxicity to HER2- cells which neighbor HER2+ cells due to its effective bystander effect, which is also confirmed using phosphor-integrated dots imaging [97]. According to a recent retrospective clinical study, patients with various statuses of HER2 (heterogenous, reduced, and loss) benefited from DS-8201a [98].

It was reported that HER2 heterogeneity was related to centromere 17 copy number gain, possibly associated with chromosomal aneuploidy [99], whereas chromosomal aneuploidy was also noticed in T-DM1-resistant cells [83]. As a consequence of chromosomal instability, chromosomal aneuploidy is a known cause of multidrug resistance in cancer and an indication of poor prognosis [100]. Altogether, chromosomal aneuploidy might also contribute to T-DM1 resistance.

5.2.3 Dysregulated downstream signaling pathways of HER2

mAbs in ADCs also exert intrinsic cytotoxic effects apart from binding to the target antigen. The phosphatidylinositol-3-kinase (PI3K)/Akt and the mammalian target of rapamycin (mTOR) signaling pathway is an important HER2 downstream pathway that is related to cancer development. Upregulation of this pathway is a known mechanism of resistance to anti-HER2 therapy. The tumor suppressor gene, phosphatase and tensin homolog deleted on chromosome ten (PTEN), negatively regulates the PI3K signaling pathway. Decreased PTEN expression in T-DM1-resistant BT-474M1 cells was also noticed, and adding pan-PI3K inhibitor, GDC-0941, resensitized T-DM1-resistant BT-474M1 cells [63]. In the EMILIA trial, patients with absent or decreased tumor PTEN expression benefited more from T-DM1 than capecitabine and lapatinib compared to patients with normal or increased tumor PTEN expression [101].

In the WSG ADAPT TP trial (NCT01779206), phosphatidylinositol-4,5-bisphosphate 3-kinase, catalytic subunit alpha (PIK3CA) mutation was significantly associated with poor prognosis in early HER2+/hormone receptor+ breast cancer [102]. With respect to advanced HER2+ breast cancer, PIK3CA mutation status was also considered a prognostic factor for PFS in the MARIANNE trial; however, this impact was contradictory to the results from both the EMILIA trial and TH3RESA trial, where the clinical efficacy of T-DM1 treatment is irrelevant to PIK3CA mutation status [101, 103]. Given the conflicting results among trials, the investigators hypothesized that due to the co-occurrence of HER2 amplification and PIK3CA mutation following prior treatments, a relatively smaller scale of tumor cells with both HER2- and PIK3CA mutation were presented in pretreated tumors than treatment-naive tumors, even though these tumors were all identified as HER2+ and PIK3CA-mutated as a whole. Hence the impact of PIK3CA mutation status was insignificance in the EMILIA trial and the TH3RESA trial where T-DM1 was used as a second- or later-line treatment [104]. The relationship between PIK3CA mutation and T-DM1 sensitivity still requires further exploration.

Moreover, since other tyrosine kinases share common downstream signaling molecules with HER2, their abnormalities also contribute to T-DM1 resistance. A study investigating the combination therapy of pertuzumab and T-DM1 suggested that the HER3 ligand, heregulin (NRG-1b), reduced cytotoxic activity of T-DM1 in some cell lines by inducing HER2-HER3 heterodimerization and activating downstream PI3K signaling pathways [105]. Receptor tyrosine kinase-like orphan receptor 1 (ROR1) is a receptor tyrosine kinase-like orphan receptor. It was reported that ROR1-positive cells showed increased T-DM1 resistance than ROR1-negative cells [106]. ROR1 overexpression induced by T-DM1 treatment, in turn, leads to T-DM1 resistance through Hippo/YAP pathway and the increase in stemness and self-renewal properties [106]. YES Proto-Oncogene 1 (YES1) is a non-receptor tyrosine kinase whose amplification leads to resistance to HER2-targeted drugs, including T-DM1. The addition of YES1 inhibitor dasatinib could resensitize the resistant cells [107].

5.2.4 HER2 instability

It is essential for ADCs to be transported to lysosomes, where they undergo lysosomal degradation and release cytotoxic payloads. HER2 is endocytosis-deficient with rapid recycling [108]. Insufficient lysosomal trafficking due to excess endosomal recycling has been noticed in T-DM1-resistant JIMT-1 cells [62]. Fewer ADCs were transported to lysosomes since more were shunted to the recycling pathway and traveled back to the plasma membrane.

The membrane stability and signal transduction of HER2 are also in close association with the heat shock protein 90 (HSP90) chaperone machinery. Hyperactive HER2 on HER2+ breast cancer cells require chaperoning by HSP90 to maintain their stability, and the inhibition of HSP90 promotes HER2 to undergo ubiquitin-dependent degradation [109]. The irreversible tyrosine kinase inhibitor neratinib has shown to induce HER2 endosomal-lysosomal endocytosis by dissociating HSP90 from HER2 and triggering ubiquitylation [110]. This mechanism was further applied in improving ADC potency, where irreversible pan-HER kinase inhibitors such as neratinib and afatinib were used in combination with T-DM1. This combination enhanced T-DM1 activity in vitro and was also verified in a patient with HER2-amplified breast cancer enrolled in a clinical trial (NCT01494662) who experienced partial response upon T-DM1 progression [111].

6.2.1 Combine with chemotherapeutics/targeted therapies

The inability of ADCs to elicit antitumor effects in tumors with target loss or target heterogeneity implies the importance of systemic chemotherapy that targets all tumor cells irrespective of target expression [122]. The combination of T-DM1 and docetaxel was efficacious, yet toxic and induced severe adverse events occur in nearly half of the patients [123].

Dual HER2 blockade appears to be more effective than monotherapy and can overcome resistance possibly by inducing efficient internalization and degradation of HER2 [124, 125]. Pertuzumab binds to HER2 and inhibits HER2 dimerization, particularly HER2-HER3 heterodimerization. The co-administration of pertuzumab and T-DM1 resulted in synergistic inhibition in HER2+ tumor models [105]. However, a pooled analysis of the efficacy of adding pertuzumab to T-DM1 showed a non-inferior but not superior clinical outcome [126].

Tyrosine kinase inhibitors (TKIs) bind to the intracellular domain of HER and are thus expected to overcome resistance related to HER2 shedding. Lapatinib is a dual TKI that binds to EGFR/HER2 reversibly. Concomitant use of lapatinib plus T-DM1 and chemotherapy yielded significant efficacy results in both early and advanced HER2+ breast cancer [127, 128]. Neratinib is a second-generation TKI that irreversibly binds to EGFR, HER2, and HER4. The addition of neratinib to T-DM1 has shown preliminary clinical efficacy in a dose-escalation study (NCT02236000) [129]. Tucatinib, a selective HER2 inhibitor, is also being studied in combination with T-DM1 in a phase III trial (HER2CLIMB-02 [NCT03975647]) in patients with pretreated metastatic HER2+ breast cancer.

Cyclin D1 and CDK4/6 are essential for tumor cell proliferation in HER2+ breast cancer. Other than suppressing Rb phosphorylation and inducing cell cycle arrest, CDK4/6 inhibitors also cause suppression of HER downstream mTORC1/S6K/S6RP signaling pathways, thereby disinhibiting HER family phosphorylation and resensitizing HER2-resistant cells to HER2 blockade [130]. The addition of CDK4/6 inhibitors can aid anti-HER2 therapies and induce suppression of HER2+ tumor cells that are refractory to T-DM1 in a preclinical study [131]. The combination of ribociclib and T-DM1 has acheived promising PFS results independent of the prior use of T-DM1 in heavily pretreated patients with HER2+ breast cancer [132]. A phase I/Ib study (NCT01976169) also verified that the addition of palbociclib resensitized patients who were resistant to T-DM1 and other anti-HER2 treatments [133].

Cotreatment of T-DM1 and PI3Kα inhibitors such as alpelisib and taselisib may enhance the activity of T-DM1 by working synergistically to inhibit the PI3K pathway. In a phase I trial investigating alpelisib plus T-DM1 in HER2+ metastatic breast cancer patients (NCT02038010), this combinational therapy appeared to be able to overcome T-DM1 resistance [134].

More potential combinational therapies containing HER2-targeted ADCs have been studied in cell lines and animal models, including co-administration with TNF-α inhibitor, drug efflux pump inhibitors, PLK1 inhibitor volasertib, and HSP90 inhibitor geldanamycin. These inhibitors can help restore sensitivity to T-DM1 by counteracting the corresponding resistance mechanisms, but more investigations are needed before clinical application.

ADC-containing combination therapies are also being researched in TNBC. Poly(ADP-ribose) polymerase 1 (PARP1) inhibitors specifically target PARP which is involved in base excision repair for DNA single-chain breaks. PARP1 inhibitors are indicated for patients with BRCA1/2-mutated breast cancer [135]. The combination of IMMU-132 and PARP inhibitors has been confirmed to have enhanced and synergistic effects in IMMU-132-resistant TNBC tumor models compared to monotherapies [136]. The combination of IMMU-132 and talazoparib is currently being studied in a phase Ib/II study in patients with metastatic TNBC (NCT04039230).

6.2.2 Combine with Immunotherapy

Immune checkpoint inhibitors, such as antibodies against immune inhibitory receptors cytotoxic T-lymphocyte antigen-4, programmed death-1 (PD-1), and programmed death ligand 1 (PD-L1), unleash the inhibition of T cells and activate antitumor immunity; while ADCs mediate antitumor immunity via activation of antigen-presenting cells and mediating intrinsic immunogenic tumor cell death [137]. It was noticed that stromal tumor-infiltrating lymphocytes increased in response to T-DM1, which promoted the augmentation of tumor-specific immunity. The potential for ADCs to act synergistically with immune checkpoint inhibitors to overcome or prevent resistance is rather inspiring [138, 139]. The combination of T-DM1 and anti-PD-1 mAbs appeared to be more efficacious than monotherapies in preclinical models. However, the addition of atezolizumab did not improve PFS and was related to more severe toxicity than the monotherapy with T-DM1 in the KATE2 trial (NCT02924883) [140]; even so, a possible OS benefit was noticed in PD-L1+ patients. This combination is now being explored specifically in patients with HER2+ and PD-L1+ metastatic breast cancer [141]. Another anti-PD-L1 antibody, pembrolizumab, is also being investigated with T-DM1 and DS-8201a.

The combinations of ADCs and immune checkpoint inhibitors are also being studied in TNBC. IMMU-132 and anti-PD-1 antibodies, such as atezolizumab and pembrolizumab, are now being investigated in multiple clinical trials (NCT04468061, NCT04434040), which are expected to optimize the treatment of TNBC.