The Phase II MutHER Study of Neratinib Alone and in Combination with Fulvestrant in HER2-Mutated, Non-amplified Metastatic Breast Cancer

This article is featured in Highlights of This Issue, p. 1241

HER2 is one of the four HER family receptor tyrosine kinases that regulate cell growth and differentiation (1). HER2 amplification/overexpression is an established therapeutic target in breast cancer (1). However, the majority of breast cancers are without HER2 amplification/overexpression and not currently eligible to receive HER2-targeted drugs. Advances in tumor genome sequencing technology led to the identification of recurrent HER2 mutations (HER2^mut) in approximately 2% of primary (2), and 3%–5% of metastatic HER2 non-amplified breast cancers (3–5). A higher incidence of HER2^mut is observed in the lobular histology (5%–7%; refs. 6–9) which is associated with poor survival (6, 7). The majority of HER2^mut cluster in the tyrosine kinase and extracellular domains, and have oncogenic properties in preclinical models (2). Importantly, tumor cells harboring HER2^mut are sensitive to the antitumor effects of HER2-targeted agents in preclinical models, especially neratinib, a potent irreversible pan-HER inhibitor (2). However, neratinib monotherapy has demonstrated only modest single-agent activity in HER2^mut, non-amplified metastatic breast cancer (MBC), with an objective response rate of 18% and clinical benefit rate (CBR) of approximately 30%, and median progression-free survival (mPFS) between 3.6 and 4 months (9–11). Combination studies to improve therapeutic efficacy are therefore justified.

The primary clinical evidence supporting the cross-talk between the estrogen receptor (ER) pathway and HER2^mut signaling is that these mutations occur predominantly in ER⁺ breast cancer (9). Several studies have shown that HER2^mut is a distinct mechanism of resistance to endocrine therapy (12). HER2^mut leads to reprogrammed ER signaling, ERK and AKT activation (5, 12). Studies of paired treatment-naïve primary and metastatic biopsies or liquid biopsies in patients with ER⁺ MBC have identified acquired HER2^mut coincident with the development of endocrine therapy resistance (3, 5, 12–14). Interestingly, mutual exclusivity between mutations in HER2 and ESR1 has been observed in some studies (3, 5, 12, 15). In preclinical studies, the combination of endocrine therapy and neratinib results in additive antitumor effects (5, 12).

On the basis of the hypothesis that the combination of neratinib and fulvestrant will be more effective than neratinib alone in ER⁺/HER2^mut, non-amplified MBC, we conducted a single-arm phase II study of neratinib plus fulvestrant with two cohorts, fulvestrant (FUL)-treated and FUL-naïve, for patients with ER⁺/HER2^mut, non-amplified MBC to assess the antitumor effects of this combination. An exploratory ER-negative (ER⁻)/HER2^mut cohort was also included for the efficacy of neratinib monotherapy. In addition, circulating tumor DNA (ctDNA) analyses at baseline, on therapy, and at progression were performed to investigate potential predictors of response and to evaluate mutational mechanisms of acquired resistance. Extrapolating data in HER2-amplified disease that dual HER2 targeting is more effective (16, 17), patients were also eligible to continue treatment at progression, with the addition of trastuzumab (if obtained through insurance) to explore its ability to reverse resistance.

Eligible patients included men or premenopausal and postmenopausal women at least 18 years old with cytologically/histologically confirmed breast cancer that is stage IV and HER2 negative (0 or 1+ by IHC or non-amplified by FISH based on American Society of Clinical Oncology/College of American Pathologists guideline on HER2 testing at the time of enrollment), with eligible HER2^mut identified in tumor tissue or ctDNA testing by a Clinical Laboratory Improvement Amendments laboratory. Additional eligibility criteria included Eastern Cooperative Oncology Group (ECOG) performance status 0–2, adequate organ and marrow function, QTc interval ≤450 ms for men or ≤470 ms for women within 2 weeks of registration and a left ventricular ejection fraction (LVEF) within institutional limits of normal. Eligible patients could have measurable or nonmeasureable but evaluable disease by RECIST 1.1. There were no limitations on the number of prior lines of systemic therapy; however, patients were required to have documented disease progression on the most recent disease evaluation. Known brain metastases must have received radiation, be off steroids and stable for ≥3 months. Prior fulvestrant was not allowed for the fulvestrant naïve (FUL-naïve) cohort, but required for the fulvestrant-treated (FUL-treated) cohort. ER positivity from the most recent tumor specimen is required for FUL-naïve and FUL-treated cohorts, unless the tissue source (e.g., bone biopsy) may yield false-negative result, in which case the pathology from an earlier timepoint could be used. Exclusion criteria included prior neratinib, concurrent systemic cancer therapy or other investigational agents, uncontrolled illness or active infection, history of significant cardiac disease, cardiac risk factors, or uncontrolled arrhythmias, active or recent coronary events, active hepatic or biliary disease, symptomatic intrinsic lung disease, women who are pregnant and/or breastfeeding and diarrhea >grade 2. This study was approved by the Institutional Review Board at all participating sites and followed the Declaration of Helsinki and Good Clinical Practice guidelines. Written informed consent was obtained prior to all study procedures.

Patients are enrolled into three cohorts in part II of MutHER trial: FUL-naïve ER⁺, FUL-treated ER⁺ or the ER⁻ cohort. Each of the ER⁺ cohorts were designed on the basis of a Simon minimax two-stage phase II design with the primary endpoint being CBR, defined as the rates of complete/partial response (CR/PR) plus stable disease (SD) ≥24 weeks. In the FUL-naïve cohort, the sample size was determined to be 26 (12 patients in stage 1 and 14 in stage 2) to detect an anticipated CBR of 65% against the null hypothesis of 40% (power = 0.8, alpha = 0.05). For the FUL-treated cohort, a sample size of 41 patients (21 patients in stage 1 and 20 in stage 2) was proposed on the basis of the CBR of 38.5% with neratinib monotherapy in part I (9), to detect an anticipated 55% CBR over the null hypothesis of 35% (power = 0.8, alpha = 0.05). Finally, the sample size of the ER⁻ cohort was set at 10 patients with the hypothesis that two CB observed out of 10 would provide 80% confidence that the “true” rate will be >5%. Secondary objectives included determining PFS, safety and tolerability, and correlation of HER2^mut with histologic subtypes. There is no efficacy comparison among cohorts.

Patients in the FUL-treated or FUL-naïve cohort received neratinib plus fulvestrant and those in the ER⁻ cohort received neratinib monotherapy. Neratinib (240 mg daily by mouth) was administered on a 28-day cycle. Dose escalation of neratinib to 320 mg daily is allowed if no intolerable grade 2 or higher treatment-related adverse events (AE) during cycle 1. Three dose reductions in 40 mg decrements were permitted for toxicities. Prophylactic loperamide was administered as described previously (9). Fulvestrant was administered 500 mg on days 1 and 15 of cycle 1, then on day 1 of each subsequent cycle. Study therapy was continued until disease progression, an intolerable AE, death, or withdraw of consent. At disease progression, patients were offered optional therapy with trastuzumab (intravenously with a loading dose of 6 mg/kg and then 4 mg/kg every 2 weeks) and to continue neratinib or neratinib plus fulvestrant. Treatment cycles were restarted at cycle 1 for AE and disease monitoring. Upon progression, all patients were followed for AEs for 30 days and up to 2 years or until death.

AEs were assessed on day 1 of each cycle according to the NCI Common Terminology Criteria for Adverse Events version 4.0. Peripheral blood was processed for serum, plasma, and plasma for ctDNA at baseline, cycle 1 day 15, cycle 2 day 1, cycle 3 day 1, day 1 of every other cycle and at the end of treatment. Radiologic tumor assessment was required at baseline and every two cycles with response evaluated by RECIST 1.1.

Plasma ctDNA isolation and sequencing by Guardant Health, Inc. (Guardant360) were described previously (9).

CBR was calculated as the proportion of CR, PR, or SD ≥ 24 weeks with 95% exact binomial confidence intervals (CI) by RECIST 1.1. PFS was defined as weeks from treatment initiation to progression or death. The associations between clinical or tumor characteristics and CBR were assessed using Fisher exact tests. The Wilcoxon signed-rank test was applied to compare HER2^mut variant allele frequency (VAF) between the paired baseline and cycle 2 day 1 samples.

All data generated in this study are available upon request from the corresponding author. Specifically, the ctDNA sequencing data provided by Guardant Health, Inc can be found in Supplementary Table S3.

Between September 2015 and October 2020, 40 patients were enrolled to the FUL-treated (n = 24), FUL-naïve (n = 11), and ER⁻ cohorts (n = 5; Table 1). The enrollment to both fulvestrant cohorts was terminated on the basis of the two-stage protocol design, and the ER⁻ cohort was terminated the same time to close the study. Across cohorts, patients had a median age of 63 years (range, 35–82) and three (range, 0–12) prior lines of systemic therapy in the metastatic setting, with 26 (65%) patients having had a CDK4/6 inhibitor (CDK4/6i) and 13 (32.5%) with a prior PI3K/AKT/mTOR inhibitor (mTORi). Seventeen patients (42.5%) were diagnosed with the lobular histology and 37 (92.5%) had visceral metastases. All patients have completed protocol therapy at this report.

All patients were included for AE assessment. The safety profile of neratinib or neratinib plus fulvestrant was consistent with prior studies. Across all cohorts, the most AEs at least possibly related to the study were diarrhea (85%), nausea (53%), fatigue (50%), anorexia (35%), and aspartate aminotransferase (AST) increase (28%; Table 2). The most common grade 3 AE was diarrhea (25%). Other AEs were mostly grade 1 and 2. There were no grade 4 AEs. Neratinib was dose reduced in 6 (15%) patients who received neratinib plus fulvestrant to 200 mg daily, due to nausea/vomiting (n = 2), diarrhea (n = 3), or elevated liver enzymes (n = 1). One patient had a second dose reduction to 160 mg daily due to diarrhea. No patients discontinued therapy due to AE.

Of the 40 patients enrolled, 5 terminated treatment after having received study drugs for 2 days (n = 1, unrelated grade 2 vomiting), 7 days (n = 3, 2 grade 3 diarrhea, 1 grade 2 vomiting), or 14 days (n = 1, unrelated grade 3 dizziness), and subsequently went off trial during cycle 1, therefore not evaluable for response and CBR. Among the remaining 35 (21 FUL-treated, 10 FUL-naïve, 4 ER⁻) evaluable patients (Fig. 1), the CBR was 38% (8/21, 95% CI, 18%–62%), including 1 CR, 4 PR, and 3 SD≥ 24 weeks, in the FUL-treated cohort. The CBR was 30% (3/10, 95% CI, 7%–65%), including 3 PR, in the FUL-naïve cohort. The CBR did not meet the predefined efficacy cutoffs to proceed to stage 2 enrollment in either ER⁺ cohort. The CBR of the exploratory ER⁻ cohort was 25% (1/4, 95% CI, 1%–81%). All patients (n = 40), the intent to treat population, were assessed for PFS. The mPFS was 24 (95% CI, 15.7–31), 20 [95% CI, 8–not available (NA)], and 8.5 (95% CI, 8–NA) weeks, in the FUL-treated, FUL-naïve, and ER⁻ cohort, respectively (Fig. 1). In an exploratory analysis by durations of prior fulvestrant (<6 vs. >6 months) in the FUL-treated cohort, no significant difference in mPFS (P = 0.71) or CBR (P = 0.65) was observed (Supplementary Fig. S1).

The protein domain location of each HER2^mut at enrollment and the corresponding treatment response is illustrated in Fig. 2. Notably only 1 of the 10 patients with the HER2 L755S derived CB, with a SD > 6 months (38 weeks). Interestingly, although this patient was enrolled on the basis of the L755S mutation identified by sequencing of a pleural metastasis, a D769Y mutation was found to be the predominant HER2^mut on the pretreatment ctDNA analysis. The D769Y mutation became undetectable in the ctDNA at cycle 2 day 1, but appeared at cycle 9 day 1 and then further increased at progression. L755 alterations (10 L755S and two L755_ins del alterations) were associated with significantly lower CBR compared with those with other types of HER2^mut [1/12 (8.3%) vs. 11/23 (47.8%), P = 0.03]. In contrast, the presence of an exon 20 insertion mutation was associated with a significantly higher CBR [5/6 (83.3%) vs. 7/29 (24.1), P = 0.01] (Supplementary Table S1).

We also examined the association between clinical or tumor characteristics and CB (Fig. 3; Supplementary Table S2). Among the 35 evaluable patients, 22 (63%) had prior CDK4/6i, 8 (23%) prior mTORi. Neither the number of prior lines of systemic therapy, nor prior CDK4/6i or mTORi impacted CBR. CBR was 45.5% (10/22) and 50% (4/8) in patients who received prior CDK4/6i or mTORi, respectively. Invasive lobular cancer (ILC) represented 37.1% (13/35) patients and was associated with a significantly higher CBR (61.5% vs. 18.2%, P = 0.02). The objective response rate was 38.5% (5/13, 95% CI, 15.1%–67.7%) in those with ILC. Concurrent mutations in TP53 (51.4%), PIK3CA (42.9%), and CDH1 (37.1%) were common, identified either in tumor tissue or ctDNA, but their presence was not associated with CBR. Interestingly ESR1 mutations were observed in 7 patients (all in FUL-treated cohort) and RB1 mutations in 3 patients, also without apparent impact on CBR.

ctDNA analyses with the Guardant 360 assay were performed in all patients at baseline except in four who did not have a sample available (Fig. 4; Supplementary Table S3). Among the remaining 36 patients, HER2^mut was detected in ctDNA in 75% (27/36) of patients. The HER2^mut VAF was significantly lower at cycle 2 day 1 compared with baseline (median = 0.195 at cycle 2 day 1 vs. 3.69 at baseline, P = 0.02; Fig. 4A). HER2^mut VAF rose in 78% (18/23) of patients at progression (PD) when compared with the immediate previous timepoint (Fig. 4A). Acquired HER2^mut were detected in ctDNA at PD in 5 of 23 patients with blood collected at progression (Fig. 4B), including 3 with the T798I gatekeeper mutation known to be associated with resistance to neratinib (18). Interestingly, HER2 L755S was found to be one of the HER2^mut acquired at higher VAF (13%) compared with T798I (4.5%) at progression after a PR on neratinib and fulvestrant in patient 41. Adding trastuzumab at progression successfully eliminated all acquired HER2^mut in ctDNA and induced a PR. Among the 7 patients with ESR1 mutations, HER2^mut was not detected in baseline ctDNA in 1 patient. The baseline ESR1 mutation VAFs were lower compared with that of the HER2^mut in most cases, suggesting subclonality of ESR1 mutation (Supplementary Table S4). Among the 6 patients with both ESR1 and HER2 mutations detected in baseline ctDNA, reductions in HER2^mut VAFs at cycle 2 day 1 was observed in 5, among whom 3 had an increase in ESR1 mutation VAFs at cycle 2 day 1 (Supplementary Table S4; Supplementary Fig. S2). The majority of the patients had increases in the VAFs for both HER2 and ESR1 mutations (Supplementary Table S4; Supplementary Fig. S2).

In an exploratory analysis, trastuzumab was added to the protocol regimen upon progression on neratinib plus fulvestrant in 4 ER⁺ patients (2 each in FUL-treated and FUL-naïve cohorts) or on neratinib monotherapy in 1 patient in the ER⁻ cohort (Fig. 5). Among these 5 patients, 3 had lobular cancer and 2 had ductal cancer. CB was observed in 4 patients (3 PR, 1 SD ≥24 weeks), including all 3 patients with lobular histology, with a mPFS of 28 (95% CI, 18–NA) weeks. Of note, all 4 responsive patients had derived CB previously while on neratinib or neratinib plus fulvestrant (Fig. 5A). ctDNA analysis demonstrated reduction of HER2^mut VAF upon the addition of trastuzumab, followed by a rise at progression (Fig. 5B). Similarly, acquired HER2^mut were observed at progression (Fig. 4B and Fig. 5B).

In this single-arm multi-cohort phase II trial, we evaluated the efficacy of neratinib plus fulvestrant in patients with ER⁺/HER2^mut, HER2 non-amplified MBC, who received prior fulvestrant (FUL-treated cohort) or not (FUL-naïve cohort). Patients with ER⁻/HER2^mut, HER2 non-amplified MBC were enrolled into an exploratory ER⁻ cohort to receive neratinib monotherapy. The CBR and mPFS were 38% (95% CI, 18–62) and 24 (95% CI, 15.7–31) weeks in the FUL-treated cohort and 30% (95% CI, 7–65) and 20 (95% CI, 8–NA) weeks in the FUL-naïve cohort, respectively. Although the trial was not designed for efficacy comparison by cohort, the similar CBR and mPFS observed in the FUL-treated and FUL-naïve cohorts indicates the efficacy of neratinib plus fulvestrant regardless of prior fulvestrant exposure. In addition, an exploratory analysis of the FUL-treated cohort suggested that the mPFS and CBR were not associated with prior duration of fulvestrant therapy (less or more than 6 months). One of 4 patients in the ER⁻ cohort derived CB, with CBR of 25% (95% CI, 1–81) and mPFS 8.5 (95% CI, 8–NA) weeks. Although the study did not meet the predefined CBR criteria to support the superiority of adding fulvestrant to neratinib, responses were observed in patients who received multiple lines of prior therapy regardless of prior CDK4/6i or mTORi. In addition, due to the limitations of a single-arm study, without a neratinib or fulvestrant monotherapy arm, potential benefits for combination therapy with fulvestrant cannot be ruled out.

The results were similar to prior studies of neratinib and fulvestrant in ER⁺/HER2^mut, HER2 non-amplified MBC (11, 19). The SUMMIT trial reported outcomes for 70 patients with ER⁺/HER2^mut, non-amplified MBC sequentially enrolled in the breast cancer basket to receive neratinib monotherapy (n = 23) or neratinib plus fulvestrant (n = 47; ref. 11). Neratinib plus fulvestrant led to CBR (95% CI) of 46.8% (32.1–61.9) and mPFS (95% CI) of 5.4 (3.7–9.2) months, while neratinib monotherapy was associated with a CBR of 30.4% (13.2–52.9) and mPFS of 3.6 (1.8–4.3) months (11). The SUMMIT trial was not designed to compare monotherapy versus combination therapy because patients were sequentially enrolled. A higher percentage of patients in the monotherapy cohort had prior CDK4/6i (43% vs. 12%; ref. 11). In the plasmaMATCH trial, an open-label, multicohort, phase IIa trial of ctDNA testing to direct therapy in advanced breast cancer, patients with HER2^mut detected in ctDNA were enrolled in cohort B to receive neratinib (and in combination with fulvestrant if ER⁺; ref. 19). Similar to the MutHER and SUMMIT trials, the CBR in cohort B was 45% (95% CI, 23–68) and mPFS was 5.4 months (19).

A high percentage (42.5%) of patients in the MutHER trial had ILC, compared with the overall prevalence of 10%–15% of ILC in primary breast cancer. ILC is known to have a higher frequency of HER2^mut which is associated with increased risk of recurrence and worse survival outcomes (6, 7). In an in silico comparison of HER2 non-amplified cases of ER⁺ stage I–III primary ILC (N = 279) and invasive ductal carcinoma (IDC, N = 1301) using METABRIC, The Cancer Genome Atlas, and MSK-IMPACT data, ILC tumors comprised 17.7% of all cases in the dataset but accounted for 47.1% of HER2-mutated cases. Mutations in HER2 were enriched in ILC versus IDC cases (5.7% vs. 1.4%; P < 0.0001). The median overall survival (OS) for patients with HER2-mutant ILC tumors was 66 versus 211 months for HER2 wild-type ILC (P = 0.0001), and 159 versus 166 months (P = 0.733) for IDC tumors (7). HER2 mutational status was a prognostic marker of 10-year OS in ILC (HR = 3.7; 95% CI, 1.2–11.0; P = 0.021), independently of known prognostic clinicopathologic features including lymph node status and tumor grade. The finding that neratinib plus fulvestrant is associated with a higher CBR (61.5% vs. 18.2%; P = 0.02) in ILC compared with non-ILC histology in the MutHER trial is therefore particularly relevant.

An intriguing finding from the MutHER trial is that at the time of progression on neratinib, the addition of trastuzumab induced tumor responses and clinical benefit in 4 of 5 patients. These 5 patients had different HER2^mut, including extracellular domain, and kinase domain, as well as, exon 20 insertion mutations. Interestingly, all 4 patients who derived benefit after adding trastuzumab were responsive to the initial neratinib regimen. The importance of dual HER2 blockade to maximize efficacy is well established for the treatment of HER2-amplified breast cancer (16). Neratinib in combination with trastuzumab demonstrated greater efficacy in reducing HER2 phosphorylation and levels of pAKT and pERK, and enhanced antitumor activity than either drug alone in HER2-amplified breast cancer cells in preclinical studies (17). A CBR of 35.7% (95% CI, 18.6–55.9) was observed in a phase I/II trial of neratinib plus trastuzumab in heavily pretreated patient population with metastatic HER2-amplified breast cancer (20). MutHER trial provides the first evidence that adding trastuzumab could overcome resistance to neratinib and is strongly supportive of dual HER2 blockade in HER2^mut non-amplified breast cancer. Indeed, the SUMMIT basket trial reported an encouraging response rate of 53% (95% CI, 28–77) and median PFS of 9.8 (4.0–not evaluable) months in the preliminary analysis of 17 patients with metastatic HER2^mut, non-amplified breast cancer with the triplet regimen of trastuzumab, neratinib, and fulvestrant (21). The SUMMIT trial has been amended so that patients with ER⁺, HER2^mut MBC who had prior CDK/6i are randomized to receive either fulvestrant, fulvestrant plus trastuzumab, or the triplet of fulvestrant/neratinib/trastuzumab, with results eagerly awaited.

The clinical responses induced by adding trastuzumab at progression on neratinib in MutHER trial was accompanied by a significant reduction or elimination of ctDNA HER2^mut, including those acquired at progression on neratinib. This suggests mechanisms agnostic to specific HER2^mut might be in play. It is well established in HER2-amplified breast cancer cell lines that trastuzumab disrupts ligand-independent HER2/HER3 interaction, leading to HER3 dephosphorylation, degradation, and uncoupling of downstream PI3K activation (22–24). Recent data also indicate the importance of HER3 signaling in mediating the oncogenic effect of HER2^mut (25). In addition, one of the main mechanisms of the antitumor effect is antibody-dependent cell-mediated cytotoxicity (ADCC), mediated by immune effector cells like natural killer cells via interactions of the FcγRs and the Fc domain of trastuzumab (26). Though this mechanism is usually implicated in the context of HER2 amplification/overexpression, Scaltriti and colleagues have reported the accumulation of HER2 at the cell surface by the HER2/EGFR kinase inhibitor lapatinib, which enhanced immune-mediated trastuzumab-dependent cytotoxicity (27). It is plausible that neratinib may share this mechanism in cells with activating HER2^mut, and because the mean plasma half-life of neratinib (and some active metaboliltes) during multiple doses is 14.6 hours (21.6 hours for active metabolite M3), there may be potential for synergism with trastuzumab starting in the first cycle. Furthermore, dual HER2 blockade with trastuzumab and neratinib could overcome resistance in in HER2-amplified breast cancer resistant to either agent (17). These mechanisms potentially explain our results.

The MutHER trial also demonstrates that the protein location of HER2^mut may impact sensitivity to the neratinib regimen. In an exploratory analysis that examined the interaction between HER2 mutation type and CBR, HER2 L755 alterations were associated with significantly lower CBR compared with other HER2^mut (8.3% vs. 47.8%; P = 0.03). The one patient with L755S (identified by DNA sequencing of a pleural metastasis) who derived CB was found to have HER2 D769Y as the predominant HER2^mut in ctDNA. In contrast, exon 20 insertion mutations were associated with a higher CBR. HER2 L755S is the most common HER2^mut in breast cancer, representing approximately 25% of the MutHER population. The reduced efficacy of neratinib or neratinib plus fulvestrant associated with L755S is intriguing; however, should be interpreted with caution due to the small sample size. Response to neratinib in the setting of L755S mutation has been reported in the SUMMIT and plasmaMATCH trials (11, 19). Among those with L755S mutations, the SUMMIT trial reported responses in 2 of 10 treated with neratinib monotherapy and 2 of 8 who received fulvestrant plus neratinib (11), while the plasmaMATCH reported 3 of 10 being responders (19). A meta-analysis of these trial data may be necessary to draw definitive conclusion. Our data are, however, consistent with observations from several preclinical studies that HER2 L755S is associated with resistance to trastuzumab (2) and reduced sensitivity to reversible and irreversible HER2 tyrosine kinase inhibitors (2, 25, 28, 29), likely as a result of higher levels of PI3K/AKT/mTOR and MAPK signaling activity (5, 28). Interestingly, among the 6 patients with HER2 exon 20 insertion mutations in the MutHER trial, 5 (83.3%) derived CB, compared with the CBR of 24.1% (7 of 29) in other patients (P = 0.01), suggesting a particular sensitivity to neratinib. While the small sample size limits the ability to draw any definitive conclusion, the data are consistent with that from the breast cancer basket of the SUMMIT trial, in which responses were observed in 11 of 15 (73%) patients (Fig. 1A; ref. 11). However, it is noteworthy that sensitivity to neratinib by HER2 mutation type is likely tumor type dependent. For example, in the SUMMIT trial, none of the 18 patients with non–small cell lung cancer (NSCLC) harboring exon 20 insertion mutations, which is the most common HER2^mut in NSCLC, responded (10).

Several patients in the MutHER trial acquired HER2^mut, often multiple, at progression. Among these, the HER2 T798I gatekeeper mutation, has been observed previously in patients with HER2^mut MBC progressed on neratinib (9, 11, 18). Preclinical studies confirmed its resistance to neratinib (18). The acquisition of HER2^mut at progression, even after dual HER2 blockade, highlights the HER2^mut as driver events.

Another interesting finding from the MutHER ctDNA analysis is the identification of ESR1 mutations in this patient population. Previous studies demonstrated that HER2 and ESR1 mutations are biologically distinct mechanisms of resistance to hormonal therapy and they were mutually exclusive in tumor tissue analysis (3, 5). Either HER2 or ESR1 mutation was detected in separate metastatic lesions in a patient (3) in a prior study, which may explain our finding of coexisting HER2 and ESR1 mutations in ctDNA as it pools tumor DNA from different metastatic sites. Interestingly, in most cases, the VAFs of ESR1 mutation were much lower than that of the HER2^mut, suggesting subclonality of ESR1-mutated cells. All ESR1 mutations were identified in FUL-treated cohort. The rise of ESR1 mutation VAFs at cycle 2 day 1 in 3 of the 5 patients who had reductions in HER2 mutation suggests resistance to fulvestrant in at least a subset of patients. Other genes that are commonly mutated in the MutHER population included TP53, PIK3CA and CDH1. We did not observe any association of mutations in these genes with CBR of the neratinib regimen. However these exploratory analyses were limited by the small sample size.

In conclusion, the MutHER trial part II demonstrated the activity of neratinib plus fulvestrant in heavily pretreated patients with ER⁺/HER2^mut non-amplified MBC, including those who received prior CDK4/6i or mTORi. Although the study did not meet the predefined statistical cutoff to support the combination of neratinib and fulvestrant relative to neratinib monotherapy, the result should be interpreted in the context of a single-arm study and its associated limitations including the lack of a neratinib monotherapy arm and a randomized design in the ER⁺ cohorts. Our results indicated the particular sensitivity of the poor prognosis HER2^mut ILC to the study regimen and the importance of dual HER2 blockade. These data support further development of the neratinib and trastuzumab combination (with ER blockade if ER⁺) in HER2^mut MBC, especially ILC.

C.X. Ma reports grants from Puma Biotechnology, Department of Defense, and St. Louis Men's Group Against Cancer during the conduct of the study as well as personal fees from Seattle Genetics, AstraZeneca, Puma Biotechnology, Biovica, Athenex, OncoSignal, Eisai, Bayer, Novartis, Sanofi, Inivita, Natera, and Jacobio and grants and personal fees from Pfizer outside the submitted work. R.A. Freedman reports other support from Firefly Health outside the submitted work. T.J. Pluard reports grants and personal fees from Novartis, SeaGen, Pfizer, H3B, Gilead, AstraZeneca, Sermonix, and Daiichi; personal fees from Tempus; and grants from Zymeworks and Olema outside the submitted work. J.R. Nangia reports other support from Paxman Coolets Ltd and Novartis outside the submitted work. J. Lu reports personal fees from Pfizer, Novartis, Radius, Sanofi, and AstraZeneca during the conduct of the study. F. Valdez-Albini reports other support from Puma Biotechnology Inc. and Washington University in St. Louis during the conduct of the study. M. Cobleigh reports grants and personal fees from Seattle Genetics; personal fees from Dragonfly Therapeutics, Macrogenics, Genentech, Immunomedics, and Puma outside the submitted work; in addition, M. Cobleigh has a patent for Genomic Health with royalties paid. N.U. Lin reports grants and personal fees from AstraZeneca, SeaGen, Pfizer, and Olema Pharma; personal fees from Merck, Denali Therapeutics, Prelude Therapeutics, Affinia Therapeutics, Voyager Therapeutics, and Aleta Biopharma; and grants from Novartis, Genentech/Roche, and Zion Pharma outside the submitted work. E.P. Winer reports personal fees from Athenex, Carrick Therapeutics, Genentech/Roche, GSK, Jounce, and Leap Therapeutics and grants from Genentech/Roche outside the submitted work. P.K. Marcom is an employee of Veracyte Inc., took position after work was completed. L. Bucheit reports other support from Guardant Health during the conduct of the study as well as other support from Guardant Health outside the submitted work. R. Bryce reports other support from Puma Biotechnology during the conduct of the study as well as other support from Puma Biotechnology outside the submitted work. A.S. Lalani reports other support from Puma Biotechnology, Inc. during the conduct of the study as well as other support from Puma Biotechnology, Inc. outside the submitted work. M.P. Goetz reports grants from Puma during the conduct of the study as well as personal fees from Research to Practice & Clinical Education Alliance; other support from AstraZeneca, Biovica, Biotheranostics, Blueprint Medicines, Eagle Pharmaceuticals, Lilly, Novartis, Pfizer, and Sermonix; and grants from Lilly, Pfizer, and Sermonix outside the submitted work. G. Kimmick reports other support from TBCRC during the conduct of the study as well as other support from Eisai, Boehringer Ingelheim, Immunomedics, Genomic Health, Agendia, Seattle Genetics, Foundation Medicine, UpToDate, and Springer; grants and other support from Novartis; and grants from Pfizer outside the submitted work. M.D. Pegram reports other support from AstraZeneca/Medimmune outside the submitted work. M.J. Ellis reports other support from Puma during the conduct of the study as well as personal fees from AstraZeneca outside the submitted work; in addition, M.J. Ellis has a patent for Prosigna/PAM50 issued, licensed, and with royalties paid from Veracyte. R. Bose reports grants from Puma Biotechnology during the conduct of the study as well as personal fees from Genentech outside the submitted work. No disclosures were reported by the other authors.

C.X. Ma: Conceptualization, resources, data curation, formal analysis, supervision, funding acquisition, investigation, visualization, methodology, writing–original draft, project administration, writing–review and editing. J. Luo: Data curation, formal analysis, validation, writing–review and editing. R.A. Freedman: Resources, investigation, writing–review and editing. T.J. Pluard: Resources, investigation, writing–review and editing. J.R. Nangia: Resources, investigation, writing–review and editing. J. Lu: Resources, writing–review and editing. F. Valdez-Albini: Resources, investigation, writing–review and editing. M. Cobleigh: Resources, investigation, writing–review and editing. J.M. Jones: Resources, investigation, writing–review and editing. N.U. Lin: Resources, investigation, writing–review and editing. E.P. Winer: Resources, investigation, writing–review and editing. P.K. Marcom: Resources, investigation, writing–review and editing. S. Thomas: Data curation, visualization, writing–original draft, writing–review and editing. J. Anderson: Resources, investigation, writing–review and editing. B. Haas: Resources, data curation, investigation, writing–review and editing. L. Bucheit: Resources, investigation, visualization, writing–review and editing. R. Bryce: Conceptualization, funding acquisition, writing–review and editing. A.S. Lalani: Conceptualization, funding acquisition, writing–review and editing. L.A. Carey: Resources, investigation, writing–review and editing. M.P. Goetz: Resources, investigation, writing–review and editing. F. Gao: Data curation, formal analysis, writing–review and editing. G. Kimmick: Resources, investigation, writing–review and editing. M.D. Pegram: Resources, investigation, writing–review and editing. M.J. Ellis: Resources, investigation, methodology, writing–review and editing. R. Bose: Conceptualization, resources, supervision, funding acquisition, investigation, methodology, writing–original draft, writing–review and editing.

MutHER is an investigator-initiated trial; we thank patients and their families for participation in this study, as well as physicians, nurses, and research and regulatory coordinators for their work. We also thank PUMA Biotechnology (C.X. Ma and R. Bose), the Department of Defense (grant number BC170330, C.X. Ma and R. Bose), Siteman Cancer Center (C.X. Ma), and the St. Louis Men's Group Against Cancer (C.X. Ma) for their generous trial support.

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.