First-in-human HER2-targeted Bispecific Antibody KN026 for the Treatment of Patients with HER2-positive Metastatic Breast Cancer: Results from a Phase I Study

This article is featured in Highlights of This Issue, p. 569

Breast cancer is the most commonly occurring cancer in women worldwide; approximately 15% to 20% of cases are characterized with HER2/ERBB2 overexpression or amplification that can be treated effectively with HER2-directed targeted agents (1). In recent decades, the introduction of HER2-targeted therapies, most notably, trastuzumab, pertuzumab, antibody–drug conjugates (ADC; e.g., T-DM1 and DS8201a), and tyrosine kinase inhibitors (TKI; e.g., lapatinib, neratinib, pyrotinib, and tucatinib) have shown dramatic improvements in the prognosis of patients with HER2-positive (HER2⁺) breast cancer (2). Nevertheless, there is still an urgent unmet medical need to develop next-generation anti-HER2 agents to improve treatment effect and reverse drug resistance.

KN026 is a novel bispecific antibody generated from heavy chains of trastuzumab and pertuzumab, with a common light chain (3, 4). It simultaneously binds to two distinct HER2 epitopes, which are the same domains of extracellular region targeted by trastuzumab (domain IV) and pertuzumab (domain II), with a complementary mechanism of action in HER2⁺ tumors. The unique mechanism of its antitumor effects includes increasing tumor cell binding, enhancing blockade of ligand-dependent and ligand-independent tumor growth, and increasing target avidity thus promoting enhanced HER2 receptor internalization. Preclinical data suggest that compared with trastuzumab plus pertuzumab, KN026 retains the antibody-dependent cell-mediated cytotoxicity effect, inhibits proliferation of HER2-overexpressing cancer cells in vitro and in vivo at a similar degree or stronger, and can kill tumor cells that are already resistant to the combination of trastuzumab and pertuzumab (4). In this first-in-human study (KN026-CHN-001; ClinicalTrials.gov identifier: NCT03619681), we aimed to evaluate the safety, tolerability, pharmacokinetics, preliminary efficacy, and potential predictive biomarker activity of KN026 administered as monotherapy in patients with HER2⁺ metastatic breast cancer (MBC) who had progressed on anti-HER2 therapies.

This is a phase I first-in-human, multicenter, open-label, single agent, dose-escalation, and dose-expansion study with the safety, tolerability, and pharmacokinetics of KN026 in patients with HER2⁺ MBC as the primary endpoints. Each patient provided signed informed consent before study enrollment. Patients received intravenous KN026 with ascending doses of 5 mg/kg once weekly, 10 mg/kg once weekly, 20 mg/kg once every 2 weeks, and 30 mg/kg once every 3 weeks until disease progression, unacceptable toxicity, withdrawal of consent, or death. Dose escalation was guided by a traditional “3+3” dose-escalation rule. Once a partial response (PR) was observed from the dose-escalation cohort, the cohort was expanded to enroll an additional 23 to 25 patients and continue safety, efficacy, and biomarker assessments. A dose-limiting toxicity (DLT) was defined as an adverse reaction related to study treatment that is unacceptable due to severity and/or irreversibility and limits further dose escalation. The DLT evaluation period was 28 days for dosing frequency of once weekly and once every 2 weeks, and 21 days for once every 3 weeks. The MTD was defined as the highest dose of KN026 not causing DLT in more than 1 of 6 patients. If MTD cannot be identified, biological effective dose determined by population pharmacokinetic and pharmacodynamic approach was used to determine the recommended phase II doses (RP2D). The study was undertaken in accordance with the International Conference on Harmonization E6 Guidelines for Good Clinical Practice, the ethical principles originating in the Declaration of Helsinki, and all applicable local regulations. The study protocol and informed consent forms were approved by the relevant local independent ethics committees or Institutional Review Boards. All participants provided written informed consent.

Eligible female Chinese patients were 18 to 75 years old and had histologically confirmed HER2⁺ MBC. HER2 positivity status was determined according to the ASCO/CAP 2018 guideline (5). Patients had received at least one prior line of anti-HER2 therapy in the metastatic setting, at least one measurable disease per RECIST guidelines, version 1.1 (RECIST v 1.1), and had a baseline left ventricular ejection fraction (LVEF) of ≥ 55%. Patients were excluded whether they had unstable brain metastasis or malignant meningitis before enrollment, a history of symptomatic interstitial lung disease, a history of cumulative doxorubicin dose exceeding 300 mg/m² or equivalent, or a medically significant cardiac disorder. Additional informed consent was obtained for patients who provided blood and/or archival tumor samples for next-generation sequencing (NGS) analysis.

Safety evaluations were conducted at every treatment cycle, including treatment-related adverse events (TRAE), DLTs (for the dose-escalation stage), clinical laboratory measurements, electrocardiogram (ECG), Eastern Cooperative Oncology Group (ECOG) performance status, vital signs, physical examinations, and frequency of anti-drug antibody (ADA). TRAEs were assessed according to the NCI-Common Terminology Criteria for Adverse Events (version 4.03) and were monitored until 30 days after the last dose.

Tumor imaging assessments were conducted every 6 weeks during first 12 months and every 12 weeks thereafter per Investigator review according to RECIST v1.1 until progressive disease, starting a new antitumor therapy or withdrawal of informed consent. Patients who achieved an objective response had confirmatory scans at least 4 weeks apart.

NGS was performed on genomic DNA and ctDNA of baseline tumor tissue sample and peripheral blood collected from 22 patients who had signed the informed consent prior to the start of the study. Correlation between genomic aberrations and clinical efficacy was analyzed using Fisher exact test for OR and log-rank (Mantel–Cox) test for progression-free survival (PFS) based on the data from this study.

To determine the copy-number status of HER2 and cyclin-dependent kinase 12 (CDK12; +1: low level gene gain; +2: high level gene gain), the multi-omics data across 33 cancer types was downloaded from The Cancer Genome Atlas (TCGA) data portal (https://portal.gdc.cancer.gov/), including somatic mutations, somatic copy-number alterations, mRNA expression, and protein expression and analyzed using threshold calls from GISTIC 2.0. In HER2 subtypes of breast cancer, Fisher exact test was used to compute the significance of difference for any somatic driver mutations identified between samples with and without HER2/CDK12 co-amplification. Studentized t test was performed to identify differentially expressed genes to build pre-ranked gene lists. A volcano plot was generated by the R package, EnhancedVolcano (Blighe and colleagues, 2020). The pre-ranked gene lists were then run against the hallmark gene sets using ClusterProfiler (Yu and colleagues, 2012; https://www.liebertpub.com/doi/10.1089/omi.2011.0118). Pathway scores based on reverse phase protein array (RPPA) data as previously described were further calculated (Li and colleagues, 2017; https://www.cell.com/cancer-cell/fulltext/S1535–6108(17)30005–3). To evaluate the tumor immune microenvironment, the abundance of six immune cell types, including B cells, CD4 T cells, CD8 T cells, neutrophils, macrophages, and dendritic cells, were obtained on the basis fo TIMER (Li and colleagues, 2016; https://genomebiology.biomedcentral.com/articles/10.1186/s13059–016–1028–7; website, http://cistrome.org/TIMER/. The difference of pathways or immune subtypes was assessed by Wilcoxon rank-sum test.

KN026 serum concentrations were determined from samples collected on day 1 of treatment cycles 1 and 2 at pre-dose, 30 minutes after infusion, at the end-of-infusion (EOI), 2, 6, 24, 72, and 168 hours after EOI, and days 1 and 15 of treatment cycles 3, 4, 5, and 6 at pre-dose and at EOI (1 treatment cycle = 28 days for once every 2 weeks dosing and 21 days for once every 3 weeks dosing) using a validated quantitative sandwich electrochemiluminescent (ECL) assay with a lower limit of quantification (LLOQ) of 0.0625 μg/mL. Standard pharmacokinetic parameters were determined using a non-compartmental method (Phoenix WinNonlin; Pharsight version 8.0). Population pharmacokinetic parameters were analyzed using nonlinear mixed effects modeling (NONMEM version 7.4, ICON Development Solutions).

KN026 ADA samples were collected along with pre-dose pharmacokinetic samples. Bridging ECL method was designed to detect the presence of anti-KN026 antibodies in serum. This method comprises a screening assay, confirmatory assay, and a titration assay.

As shown in Supplementary Data S1, SK-BR-3 (Human breast cancer, ATCC) were cultured in McCoy's 5a plus 10%FBS. AU565 (Human breast cancer, ATCC), HCC2218 (Human breast cancer, ATCC), HCC1954 (Human breast cancer, ATCC), KYSE-410 (Human esophageal cancer, DSMZ), NCI-H1781 (Human bronchioloalveolar carcinoma, ATCC), NCI-H2170 (Human squamous NSCLC, PUMC), NCI-N87 (Human gastric cancer, ATCC) were cultured in RPMI1640 plus 10% FBS. OE19 (Human esophageal cancer, DSMZ) were cultured in RPMI1640 plus 10% FBS and 2 mmol/L l-Glutamine. EFM-192A (Human breast cancer, CoBioer) and ZR-75-30 (Human breast cancer, ATCC) were cultured in RPMI1640 plus 20% FBS. All cells were incubated at 37°C, 5% CO₂.

KN026 was diluted with final concentrations of 676.52, 169.13, 42.28, 10.57, 2.64, 0.88, 0.22, 0.055, and 0.014 nmol/L; trastuzumab was diluted with final concentrations of 687.13, 171.78, 42.95, 10.74, 2.68, 0.67, 0.17, 0.042, and 0.01 nmol/L; pertuzumab was diluted with final concentrations of 675.67, 168.92, 42.23, 10.56, 2.64, 0.66, 0.17, 0.041, and 0.01 nmol/L. Dinaciclib (CDK12 inhibitor, Beyotime, SC6628) and positive control cisplatin were diluted with DMSO with final concentrations of 100, 33.33, 11.11, 3.70, 1.23, 0.41, 0.14, 0.046, and 0.015 nmol/L for dinaciclib and final concentrations of 100, 33.33, 11.11, 3.70, 1.23, 0.41, 0.14, 0.046, and 0.015 μmol/L for cisplatin.

Cell suspensions (∼1,500–8,000 cells) were incubated in well plates overnight inside a humidified incubator at 37°C with 5% CO₂. A 10 μL compound solution (10X) was then added to each well (in triplicate) to incubate at 37°C with 5% CO₂ for 3 days. Thereafter, 50 μL CellTiter-Glo Reagent was added to each well to facilitate cell lysis. Luminescence using EnVision 2104 Multi Label Reader was measured and recorded. GraphPad Prism 5.0 software was used to display sigmoidal dose–response surviving rate graph by a nonlinear regression model to calculate the inhibitory concentration at 50% of maximum (EC₅₀). Loewe synergy and antagonism analysis were performed to determine synergistic or antagonistic effects between combination drugs.

Patient demographics and baseline characteristics by dose level and overall are summarized in Table 1. A total of 63 female patients (median age: 54 years; range: 31–69 years) were enrolled from September 2018 to December 2019. The median prior lines of systemic therapies are 3 (range: 1–12) and the median prior lines of HER2-targeting therapies are 2 (range: 1–10) for advanced unresectable or metastatic disease. Of the 63 patients, 36 (57.1%) patients received more than three prior lines of antineoplastic regimens. Sixty-one (96.8%) patients received trastuzumab treatment, 32 (50.8%) patients received anti-HER2 TKI, and 15 (23.8%) patients received anti-HER2-ADC treatment. It should be noted that not until December 2018 and February 2020, pertuzumab and T-DM1 were approved in P.R. China, respectively. Therefore, patients who were not pretreated with pertuzumab or T-DM1 were included in this study. Sixty (95.2%) patients had visceral disease at baseline. Common sites of metastases included lung (55.6%), and liver (28.6%). At the time of the data cut-off date of December 21, 2020, 14 patients remained on the study treatment and 49 patients discontinued treatment due to disease progression (n = 46), withdrawal of consent (n = 2), and TRAEs (n = 1; Fig. 1). There were 22 subjects with evaluable data for biomarker analysis.

All 63 patients were evaluated for safety assessments (Table 2). Median treatment duration was 5.6 months (range: 1.1–18.5 months). No DLTs were observed at all dose levels. KN026 TRAEs of any grade were observed in 56 patients (88.9%). The most common (≥10%) TRAEs were pyrexia (23.8%), diarrhea (22.2%), aspartate aminotransferase increased (22.2%), alanine aminotransferase increased (22.2%), white blood cell count decreased (17.5%), hypokalemia (12.7%), infusion-related reaction (12.7%), neutrophil count decreased (12.7%), and rash (11.1%). In total, 4 patients (6.3%; 2 patients in the 20 mg/kg once every 2 weeks cohort and 2 patients in 30 mg/kg once every 3 weeks cohort, reported grade 3 TRAEs, including infusion-related reaction, transaminases increased, ventricular arrhythmia, and cardiac myxoma. No grade 4 or 5 AEs were reported. TRAEs leading to treatment discontinuation occurred in 1 (1.6%) patient allocated in the 20 mg/kg once every 2 weeks cohort. The patient had an abnormal ECG upon study entry, with a prior medication history of using anthracycline and taxanes. After receiving two doses of KN026 treatment, ECG readings indicated ventricular premature beats (quadruple rhythm). The patient was hospitalized and KN026 treatment was discontinued. At day 30 of the safety follow-up, the patient recovered. Neither dose delays nor dose reductions were reported.

Single- and multiple-dose pharmacokinetics of KN026 following intravenous infusion have been characterized by standard non-compartmental analysis. After reaching the peak drug concentration (C_max), KN026 serum concentrations decreased in a biexponential manner (Fig. 2). A linear dose exposure in maximum concentration (C_max) and area under the concentration–time curve (AUC_0-inf) was observed across the dose range between 5 to 30 mg/kg. Apparent total body clearance at steady state (calculated as dose/AUC_tau) was low (average of 20.5 mL/hour). The average half-lives were 137 (±27) and 200 (±39) hours for 20 mg/kg and 30 mg/kg doses after single dose, respectively. Supplementary Table S1 displays the main pharmacokinetic parameters that were evaluated after the single and multiple doses of KN026.

The population pharmacokinetic analysis dataset included 1,191 KN026 serum concentration observations from 90 patients, including 63 patients from this study. The population pharmacokinetic data showed typical characteristics of a two-compartment dynamics. The typical values of clearance, central and peripheral volume were estimated to be 0.353 L/day, 2.58 L and 2.49 L, and variability in these parameters was relatively low (<25% coefficient of variation). All pharmacokinetic parameters were estimated with good precision (relative SEs <30%; Supplementary Table S2). Baseline body weight and country effects on central volume, and albumin, baseline tumor size, and country effects on clearance were found significant. Simulations showed that both 20 mg/kg every 2 weeks and 30 mg/kg every 3 weeks schedules achieved more than 20 μg/mL target threshold derived from preclinical studies in more than 90% of 1,000 simulated populations. A total of 20 mg/kg every 2 weeks may have higher chance to achieve a higher target threshold of 80 μg/mL for more aggressive tumors (Supplementary Fig. S1A).

Two (3.2%) of the 63 tested patients who were evaluable for post-dose anti-KN026 antibodies were confirmed positive for ADAs. No differences were observed in the pharmacokinetic profiles, safety features, or efficacy outcomes for the 2 patients (data not shown).

All patients were evaluated for response assessments. With a median follow-up of 15 months (range: 12.6–26.8 months), tumor shrinkage was observed in 46 (73.0%) of 63 patients who had measurable baseline lesion and at least one post-baseline scan (Fig. 3A and B). A total of 15 patients (23.8%) achieved a best response of confirmed PR, 1 (1.6%) patient achieved confirmed complete response (CR), 28 (44.4%) had stable disease (SD), and 18 (28.6%) had progressive disease (PD). The confirmed objective response rate (ORR) was 25.4% [95% confidence interval (CI): 15.3–37.9] and the disease control rate (DCR) was 69.8% (95% CI: 57.0–80.8). The median PFS was 5.6 months (95% CI: 4.1–8.2; Table 3) and the 6-month follow-up PFS was 47.7% (95% CI: 34.8–59.6). The median overall survival (OS) was not reached. The 12-month OS rate was 85.8% (95% CI: 73.5–92.6). Two patients previously treated with pertuzumab received KN026 at the RP2Ds and achieved PR thereafter (Supplementary Table S3). For example, one of the responders, who had the relapsed breast cancer, previously received adjuvant chemotherapy and radiotherapy (DFI = 20 months), first-line docetaxel/trastuzumab/pertuzumab (median PFS = 10 months) and second-line capecitabine/lapatinib (median PFS = 7 months). The median PFS of third-line KN026 (30 mg/kg every 3 weeks) for the patient was 6.77 months (Supplementary Fig. S1B).

The categorized efficacy (categorized by type of resistance to trastuzumab (6), hormone receptor status, and prior receipt of anti-HER2 therapy including pertuzumab, anti-HER2 TKI, or anti-HER2 ADC for patients in the RP2D cohorts (N = 57 patients) was summarized in Supplementary Table S3. In detail, trastuzumab primary resistance was defined as progression at first radiological reassessment at 8–12 weeks or within 3 months after trastuzumab with or without chemotherapy in the metastatic setting or new recurrences diagnosed during or within 12 months after adjuvant trastuzumab. Trastuzumab secondary resistance was defined as disease progression after trastuzumab-containing regimens that initially achieved disease response or stabilization at first radiological assessment. In the total cohort of 63 evaluable patients, the categorized efficacy was summarized in Supplementary Table S3.

To explore the molecular mechanisms underlying patients' differential responses to KN026 and to identify promising biomarkers, 22 HER2⁺ patients were recruited on the basis of IHC test (Supplementary Table S4). A NGS panel-based gene test was performed on patient tissue samples and the gene test confirmed that 20 patients had HER2 amplification, while 14 of those same patients had CDK12 co-amplification. Of note, responders were highly enriched with HER2/CDK12 co-amplification (Fig. 4A, ORR of 50% vs. 0%, Fisher exact test, P = 0.05) and achieved longer median PFS (Fig. 4H; 8.2 vs. 2.7 months, log-rank test, P = 0.04), suggesting that HER2/CDK12 co-amplification contributes to more effective clinical outcomes and this feature can be potentially used as a biomarker for KN026 treatment. Using multi-omics data of TCGA, HER2/CDK12 co-amplifications were shown to be widespread across a broad range of cancer types, especially in kidney renal papillary cell carcinoma and lung squamous cell carcinoma. Approximately 30% of patients with breast cancer (ranked as the 9th most prevalent cancer among the 33 cancer types) exhibited HER2/CDK12 co-amplification (Fig. 4B), which increased to approximately 70% in HER2-enriched subtype defined by PAM50 in breast cancer (Fig. 4C). Because 80% of the patients with HER2⁺ subtype breast cancer exhibited HER2 amplification (TCGA Network, 2012; https://www.nature.com/articles/nature11412), this subtype was made the focus in this study. It was found that loss-of-function mutations of well-known driver genes in breast cancer, PTEN, were significantly enriched in this cancer subtype without HER2/CDK12 co-amplification (Supplementary Fig. S2, Fisher exact test, P = 1.3 × 10⁻³). As expected, HER2/CDK12 co-amplification resulted in significantly higher expression levels of the two genes at the mRNA level (Fig. 4D). In addition, compared with the co-amplification-free group, the HER2/CDK12 co-amplification group showed significantly lower activities for cell cycle, mTOR, and DNA repair pathways based on both mRNA and protein expression levels (Fig. 4E). In addition, the co-amplification group showed less abundance of immune cells, including CD8⁺ T cells (P = 1.7 × 10⁻²), neutrophils (P = 1.9 × 10⁻²), and B cells (P = 2.9 × 10⁻²; Fig. 4F) and lower immune activities (Fig. 4G). Taken together, patients with breast cancer with HER2/CDK12 co-amplification have distinct altered molecular profiles, dysregulated pathways, and tumor microenvironments, which may be responsible for the observed differential response to KN026. These results suggest that HER2/CDK12 co-amplification may be a potential predictive biomarker for KN026 treatment response.

According to the baseline NGS information from tissue and peripheral blood ctDNA of 22 patients (Supplementary Fig. S3), gain-of-function HER2 mutations in three patient tissues (p.T862A, p.H878Y, p.R897W) and 2 patient ctDNA (p.T862A and p.H878Y, which was consistent with the corresponding tissue NGS information) were found. The best responses of these 3 patients were 2 PD and 1 size-increased SD. TP53, CDK12, MYC, and PIK3CA were the four most frequently altered genes detected in the tissue of the 22 patients. However, alteration in the transcription factor, MYC, was not common in the ctDNA analysis. The only 2 patients without HER2 amplification detected in tissues had the best response of PD. Those patients with PIK3CA mutation detected in tissue or ctDNA were relatively insensitive to KN026 treatment.

Using CellTiter-Glo Luminescent Cell Viability Assay, KN026 was found to be more effective and potent than trastuzumab + pertuzumab in trastuzumab resistant cell lines with respect to potency and efficacy parameters such as absolute EC₅₀, relative EC_95, and maximum inhibition rate. Consistent with these findings, KN026 was shown to be more effective and potent against HER2/CDK12 co-amplified cell lines than in HER2-amplified/CDK12 non-amplified cell lines (Supplementary Data S1). The maximum inhibition rates of the three combinations including dinaciclib (trastuzumab + dinaciclib, trastuzumab + pertuzumab + dinaciclib, KN026 + dinaciclib) were all very high, and no significant difference of inhibition effect between CDK12 co-amplified and no-co-amplified cell lines were observed.

To the best of our knowledge, this is the world's first proof-of-concept clinical trial focusing on MBC with a bispecific humanized antibody targeting HER2 demonstrating that the first-in-human study of KN026 was well tolerated, with a favorable safety profile, and encouraging preliminary antitumor activity in patients with HER2⁺ MBC who had progressed on at least one prior line of anti-HER2 therapy in the metastatic setting. In this study, the MTD of KN026 was not achieved at doses up to 30 mg/kg every 3 weeks. Analysis of pharmacokinetic properties showed that systemic exposure (C_max and AUC_0-t) and t_1/2 increased with dose escalation. The RP2Ds of KN026 were determined to be 20 mg/kg every 2 weeks or 30 mg/kg every 3 weeks based on safety, clinical responses, and pharmacokinetic modeling and simulation.

Reversible TRAEs of any grade occurred in 56 patients (88.9%). The most common TRAEs were pyrexia, diarrhea, alanine aminotransferase increased, aspartate aminotransferase increased, white blood cell count decreased, hypokalemia, infusion-related reaction, neutrophil count decreased, and rash, among which pyrexia was mainly attributed to infusion-related reactions. The safety profile of KN026 has both similarities and differences with trastuzumab and pertuzumab; that is, the occurrence of infusion-related reactions is common in all these drugs. The cardiac dysfunction characterized by a decline in LVEF is also a common AE in more than 10% of the patients treated with trastuzumab or pertuzumab (7, 8). However, only one single patient in the 30 mg/kg once every 3 weeks cohort experienced a 16% absolute decrease in LVEF from baseline (i.e., from 68% to 52%). The LVEF returned to the absolute decrease from baseline <15% within 8 weeks of withholding KN026 dosing. In addition, 4 patients experienced grade 3 TRAEs and recovered after interventions. Drug discontinuation due to ventricular premature beats was reported in a single patient with a prior history of abnormal electrocardiography.

In this study, KN026 demonstrated very promising antitumor activity with an ORR of 28.1%, a DCR of 71.9%, and a median PFS of 6.8 months at RP2Ds. This is comparable with or even better than the results of a phase II trial of the doublet of trastuzumab and pertuzumab in trastuzumab-pretreated HER2⁺ MBC, which exhibited an ORR of 24.2%, a DCR of 50%, and a median PFS of 5.5 months (9). It is worth mentioning that the KN026 benefit was obtained in the more heavily pretreated setting, in which 29.8% of the patients were categorized as trastuzumab primary resistant, and HER2 TKI and HER2 ADC agents had been used in 50.8% and 23.8% of the patients, respectively. More important, the 2 patients previously treated with pertuzumab achieved PR. It is well known that the combination of trastuzumab and pertuzumab can synergistically inhibit tumor growth in both in vitro and in vivo preclinical models, but the underlying mechanism of the more potent tumor-suppressing effect of KN026 remains to be elucidated (10). Actually, Li and colleagues found that a bispecific antibody targeting two non-overlapping epitopes on HER2 would be effective in cross-linking receptors and thereby promoting the endocytosis and altering the receptor intracellular trafficking pathway from recycling to lysosomal degradation (11). Taken together, proceeding with a randomized, statistically well-designed, head-to-head phase III superior study to compare KN026 with trastuzumab/pertuzumab appears reasonable to be justified especially in the first-line setting.

Exploratory analyses demonstrated that the best responses of KN026 in 3 patients with gain-of-function HER2 mutations were 2 PD and 1 size-increasing SD. In fact, HER2 p.T862A, p.H878Y, p.R897W mutations all lie within the HER2 protein kinase domain. In some preclinical studies, transformed cells expressing HER2 gain-of-function mutations demonstrated resistance to trastuzumab (12). TP53, CDK12, and PIK3CA were among the most frequently altered genes detected in both tissue and ctDNA, which was similar to the findings from Chen and colleagues study (13). Interestingly, all PR patients had HER2/CDK12 co-amplification, which is not surprising as the CDK12 gene itself is located in chromosome 17 (17q12), approximately 200 kb proximal to the HER2 oncogene and is frequently co-amplified with HER2 in breast cancer (14–16). According to our analyses based on TCGA dataset and previous published literature (17, 18), patients with HER2⁺ breast cancer with CDK12 co-amplification have distinct altered molecular profiles, dysregulated pathways (such as PI3K/AKT pathway), and tumor microenvironments, which were critically related to trastuzumab and lapatinib resistance. However, both cell line and clinical data in this study showed that co-amplification of HER2/CDK12 was a promising positive marker to predicting better response to KN026. The biological mechanism behind this and the question of whether CDK12 inhibitors can enhance the effect of KN026 are worthy of further investigation in the future. Patients with PIK3CA mutation responded less favorably to KN026, which is consistent with previous studies showing that mutations in PIK3CA may play a key role in the trastuzumab/pertuzumab resistance pathway and progression of MBC (19–22).

The development of bispecific antibodies against HER2 is actually rapidly evolving areas. Notably, two of the most promising novel agents, ZW25 (23–25) and MBS301 (26), share a very similar design with KN026. ZW25 is a bispecific antibody that simultaneously binds to two epitopes on HER2 with the domains containing the exactly same binding sites of trastuzumab and pertuzumab. In a phase I basket trial, 24 patients with HER2⁺ cancers received ZW25 treatment. These patients previously progressed on a median of 3 prior lines of therapies with 71% having previously received trastuzumab. In 17 evaluable patients, 7 (41%) achieved response and a further 7 (41%) had SD, equating to a DCR of 82% with a median PFS of 6.2 months (22–24). Although the ZW25 trial had a smaller sample size for patients with HER2⁺ breast cancer, the impressive preliminary clinical efficacy results with this agent provided further evidence of the promising future for bispecific antibody immunotherapy approaches.

In conclusion, KN026 is well tolerated, with a manageable safety profile and demonstrated encouraging antitumor activity in patients with HER2⁺ breast cancer who had progressed on anti-HER2 therapies. A phase II study (NCT04165993) evaluating the efficacy of KN026+docetaxel is currently ongoing and a randomized, phase III trial to compare this regimen with trastuzumab/pertuzumab+docetaxel in the first-line setting for HER2⁺ MBC is planned. Prospectively evaluating HER2/CDK12 co-amplification to define patients who benefit more from KN026 is warranted.

X. Peng reports a full-time employee of Precision Scientific. T. Xu reports personal fees from Alphamab Oncology during the conduct of the study. No disclosures were reported by the other authors.

This study is sponsored by Jiangsu Alphamab Biopharmaceuticals.

J. Zhang: Conceptualization, resources, data curation, formal analysis, supervision, investigation, writing–original draft, writing–review and editing. D. Ji: Data curation, formal analysis, validation, writing–review and editing. L. Cai: Resources, data curation, writing–review and editing. H. Yao: Resources, data curation, writing–review and editing. M. Yan: Resources, data curation, writing–review and editing. X. Wang: Resources, data curation, writing–review and editing. W. Shen: Resources, data curation, writing–review and editing. Y. Du: Resources, data curation, writing–review and editing. H. Pang: Resources, data curation, writing–review and editing. X. Lai: Resources, data curation, writing–review and editing. H. Zeng: Resources, data curation, writing–review and editing. J. Huang: Resources, data curation, writing–review and editing. Y. Sun: Resources, data curation, writing–review and editing. X. Peng: Formal analysis, methodology, writing–review and editing. J. Xu: Resources, formal analysis. J. Yang: Resources, formal analysis. F. Yang: Resources, formal analysis. T. Xu: Resources, formal analysis. X. Hu: Conceptualization, resources, data curation, formal analysis, supervision, writing–review and editing.

We thank all the patients who agreed to participate in this study.

This work was supported in part by a grant from the National Natural Science Foundation of China (grant no. 82072915).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.