Winer, Eric P.

1005 Background: Evidence-based treatments (tx) for metastatic, HER2+ BCBM are limited. We previously found a central nervous system (CNS) objective response rate (ORR) of 8% (95% CI 2-22%) for the irreversible, EGFR/HER2-targeted kinase inhibitor, neratinib. To enhance CNS activity, we evaluated the combination of neratinib + capecitabine in a subsequent cohort, and report results here. Methods: Pts with measurable BCBM (≥ 1 cm in longest dimension) and no prior lapatinib or capecitabine were eligible. All but 3 had CNS progression after local CNS tx. During 21 day cycles, pts received capecitabine 750 mg/m2 twice daily x 14 days followed by 7 days off + neratinib 240 mg orally once daily. Loperamide prophylaxis (16 mg total daily) was recommended during cycle 1. Brain MRI and non-CNS imaging were repeated every 2 cycles until 18 wks, then every 3 cycles. The primary endpoint was composite CNS ORR, requiring all of the following: ≥50% reduction in volumetric sum of target CNS lesions (central review, VORR), no progression of non-target or non-CNS lesions, no new lesions, no escalating steroids, and no progressive neurologic signs/symptoms. We used a two-stage design with hypotheses ORR 15% and 35% (error rates 5% and 20%), responses in ≥5/19 pts to enter 2ndstage; responses in ≥9/35 [26%] pts to be promising. Results: 39 pts enrolled between 4/2014-11/2016 (2 withdrew before tx, 37 analyzed); median age 51, median prior metastatic lines 2 (range 0-6), 65% had prior WBRT. As of 11/15/16, 23 (62%) patients are alive and 7 remain on protocol tx; median number of cycles initiated = 5 (range 1-26). 18 women (49%) had a VORR (95% CI 32-66%, neurologic exams not yet available on all pts). Overall 12-month survival is 63% (95% CI 43%-77%); 4/7 pts still on protocol therapy have not yet reached 6 cycles. No pts had grade 4 toxicity; 18 (49%) had grade 3 toxicity, with diarrhea most common (32%), and 6 pts discontinued tx for toxicity. Conclusions: The combination of neratinib and capecitabine is active for BCBM with VORR in nearly half of pts, supporting further development of the regimen for BCBM. Updated results will be presented at the meeting. Clinical trial information: NCT01494662.

1088 Background: Standard first-line treatment for mTNBC is chemotherapy. However, outcomes are poor, and new treatment options are needed. Cohort B of KEYNOTE-086 (NCT02447003) assessed the safety and antitumor activity of pembrolizumab as first-line therapy for patients (pts) with PD-L1–positive mTNBC. Methods: Men and women with centrally confirmed mTNBC, no prior systemic anticancer therapy for metastatic disease, ECOG PS 0-1, and a tumor PD-L1 combined positive score (CPS) ≥1% received pembrolizumab 200 mg Q3W for 24 mo or until disease progression, intolerable toxicity, or investigator or pt decision. Tumor imaging was performed Q9W for 12 mo and Q12W thereafter. Clinically stable pts with PD could remain on pembrolizumab until PD was confirmed on subsequent assessment. Primary end point was safety. Secondary end points included ORR, DOR, and PFS (RECIST v1.1, central review). Planned enrollment was 80 pts. This analysis included all pts who had ≥18 wk of follow-up as of Nov 10, 2016. Results: 79 of the first 137 pts with PD-L1–evaluable tumors (58%) had PD-L1 CPS ≥1%. Of the first 52 pts enrolled, 100% were women, median age was 53 y, 40% had elevated LDH, 69% had visceral metastases, and 87% received prior (neo)adjuvant therapy. After a median follow-up of 7.0 mo (range 4.4-12.5), 15 (29%) pts remained on pembrolizumab. Treatment-related AEs occurred in 37 (71%) pts, most commonly fatigue (31%), nausea (15%), and diarrhea (13%). 4 (8%) pts experienced 5 grade 3-4 treatment-related AEs: back pain, fatigue, hyponatremia, hypotension, and migraine (n = 1 each). No pts died or discontinued pembrolizumab due to an AE. ORR was 23% (95% CI 14%-36%). Best overall response was CR in 4%, PR in 19%, SD in 17%, PD in 58%, and not assessed in 2%. Median time to response was 8.7 wk (range 8.1-17.7). Median DOR was 8.4 mo (range, 2.1+ to 8.4), with 8 (67%) responses ongoing at cutoff. Median PFS was 2.1 mo (95% CI, 2.0-3.9); estimated 6-mo PFS rate was 29%. Conclusions: Data from the first 52 pts enrolled in KEYNOTE-086 cohort B suggest that pembrolizumab monotherapy has a manageable safety profile and promising antitumor activity as first-line therapy for PD-L1–positive mTNBC. Clinical trial information: NCT02447003.

Abstract Introduction. Male breast cancer (male BC) is a rare disease for which disease management is extrapolated from females. IMBC, an international consortium, which previously reported on clinico-pathological aspects, now reports on molecular subgroups revealed by RNA sequencing and their relation to patient outcome. Methods. Tumor samples from the retrospective MALE BC registry diagnosed between 1990-2010 and with pathology and outcome data (relapse-free- (RFS) and overall survival (OS)) were included (n=699). To allow the discovery of prognostic features, we selected, stratified for known risk factors (TN stage, grade, age at diagnose and adjuvant endocrine treatment), from the cohort 152 cases with poor (RFS <= 4 yrs) and good outcome (RFS > 7yrs) evenly distributed. Here, we report RNA sequencing results of the first 73 cases, 38 with poor and 35 with good outcome. RNA sequencing reads were used to generate gene expression values and to report transcripts carrying driver mutations. Unsupervised clustering identified subgroups and within subgroups differentially expressed genes were identified. The reported prognostic male BC subgroups M1 and M2 (Johansson BCR 2012(14):R31) were also annotated. All identified subgroups were related to outcome using logistic regression (p-value using Wald test). Results. Unsupervised clustering revealed 2 main subgroups of which group 1 was enriched for expression of ER target genes, WNT3 and genes from amplicons known for female BC, e.g. 19p13 (CCNE1), 8q24 (MYC), and 11q13 (CCND1). The biology of the smaller group 2 was less defined but TGFβ2 expression was high as were various kallikreins (KLK) including interestingly KLK3 (prostate specific antigen). Other known amplified regions [(8p11 (FGFR1), 20q13 (ZNF217) and 12q15 (MDM2)] and mutated transcripts [PIK3CA (H1047R/L/Q; E542K, E545K, N345K; 16% of cases), TP53 and SF3B1 (K700E) (2% of cases)] were identified. Profound tumor infiltrate gene expression was present in 5% of cases and one third of cases expressed proliferation markers. Except for TP53, none of these latter characteristics were unevenly distributed among the 2 main subgroups. ER and AR were highly correlated, particularly in group 1. The two main groups could be further subdivided. Group 1 comprised 3 subgroups of which subgroup 1a expressed TFF1/3 and NAT1, well-known ER targets, while subgroups 1b and 1c expressed other ER targets,respectively BEX1 and PITX1. HOXC cluster expression differentiated subgroup 1b from 1a and 1c. None of these intrinsic subgroups were, however, related to RFS. The previously reported M2 subgroup, which largely segregated with subclusters 1a and 1b, was associated with a better RFS than the M1 subgroup (OR=2.9; 95%CI 1.1-7.5; p-value=0.03). Conclusions. 1) Intrinsic subtypes of male BC were revealed and their subgrouping is defined by ER associated subsets of genes. 2) The association of the reported M2 subgroup of male BC with longer RFS was validated; 3) Currently identified biological characteristics of male BC may improve future treatments. The full report on 152 cases including a comparison to female BC will be presented at the conference. This research was funded by Breast Cancer Research Foundation Citation Format: Martens JWM, Sieuwerts A, Ponchet C, Smid M, de Weerd V, Slaets L, Piper T, van Deurzen CHM, Schroder CP, Stangle C, Kloosterman W, van Leeuwen-Stok E, Nilsson C, Vermeij J, Peeters S, Goulioti T, Nowaczyk M, Aebi S, Rubio IT, Kelly C, Bayani J, Porter P, Murray M, Hudis C, Middleton L, Korde L, Ruddy K, Winer E, Bogler O, van den Weyngaert D, dal Lago L, Fraser J, Benstead K, van Asperen C, Linderholm B, Hedenfalk I, Tryfonidis K, Giordano S, Bartlett J, Cardoso F. Molecular subtyping of male breast cancer by the International male breast cancer program (IMBC): EORTC 10085/TBCRC 0-29/BIG 2-07/NABCG/BOOG 2009-04 [abstract]. In: Proceedings of the 2017 San Antonio Breast Cancer Symposium; 2017 Dec 5-9; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2018;78(4 Suppl):Abstract nr PD7-12.

Abstract Background: Stage III triple negative(TN) inflammatory breast cancer(IBC) is associated with a poor prognosis evidenced by a 15 month(mo) median disease free survival(DFS) and overall survival(OS) of 34 mo. The substantial incidence of developing distant metastasis may be due to the prevalence of cancer cells with stem cell-like features (e.g. CD44+/CD24-) in TNIBC. The transcriptional pathway JAK2/STAT3 is associated with the survival of CD44+/CD24- cells, and preclinical data demonstrates overexpression of activated STAT3(pSTAT3) in > 95% of TNIBC. Preclinical studies have shown that ruxolitinib (Incyte®Corporation), an approved JAK1/JAK2 inhibitor, suppresses pSTAT3 in IBC patient derived xenograft models, and when combined with paclitaxel, results in a synergistic reduction in tumor weight. Given the lack of a known therapeutic target in TNIBC, this preoperative proof of principle study exploits the survival mechanism of CD44+/CD24- stem cells prevalent in this disease, by combining ruxolitinib(Rux) with paclitaxel(T) followed by doxorubicin/cyclophosphamide(AC). Methods: Up to 64 pts with newly diagnosed stage III TNIBC (cT4d, any N, M0) are eligible if they have adequate organ function and are willing to undergo 2 research biopsies(rbx) of the affected breast. Following baseline rbx, pts are randomized to a 7 day(d) run-in phase of Rux vs Rux(15 mg bid) + T(80mg/m2/wkx1). A 2nd rbx is obtained after the run-in phase. Pts randomized to RuxT continue to receive a total of Tx12 wks+Rux. Pts randomized to Rux alone, are re-randomized to receive Tx12 wks+Rux vs Tx12wks alone. Following T, all pts receive AC (A-60mg/m2,C-600mg/m2) every 14d x 4. Pts proceed to modified radical mastectomy (MRM) followed by chest wall/regional lymph nodes radiation therapy. Correlatives: To assess the effect of JAK inhibition with Rux on pSTAT3 and STAT3 related gene expression, molecular and genomic markers (e.g. RNA-seq, ChIPseq, FISH) will be determined in each rbx and residual tumor at MRM. The relative frequency and topology of CD44+/CD24- cell population and pSTAT3 expression by IHC will also be assessed in these tumor specimens. IL-6 and CRP plasma concentrations will be measured at baseline, prior to T and AC and prior to MRM. Statistics: The primary endpoint is change in markers of JAK/STAT inhibition. If the proportion of rbx exhibit a biologic response to Rux alone (i.e. change from pSTAT3 expression to pSTAT3 negative) is <10%, then Rux alone is minimally effective on JAK inhibition vs alternative hypothesis that Rux inhibits JAK if the proportion of biologic response is ≥33%. If ≥5/25 rbx treated with Rux alone have a biologic response then the hypothesis that biologic response is ≤10% is rejected with an error rate of 0.098 (target 0.10). If ≤4/25 rbx have a biologic response then the hypothesis that biologic response is ≥33% is rejected with an error rate of 0.05 (target 0.10). Biologic response of rbx with Rux alone will also be compared with the proportion of biologic response to RuxT (33% vs. 66% based upon presumed synergy with RuxT). Secondary endpoints are clinical: pathologic complete response in breast/lymph nodes, Residual Cancer Burden, DFS and OS. Clinical Trial Information: NCT02876302. Citation Format: Overmoyer B, Regan M, Polyak K, Brock J, Van Poznak C, King T, Haddad T, Stearns V, Hwang S, Winer E. TBCRC 039: Phase II study of combination ruxolitinib (INCB018424) with preoperative chemotherapy for triple negative inflammatory breast cancer [abstract]. In: Proceedings of the 2017 San Antonio Breast Cancer Symposium; 2017 Dec 5-9; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2018;78(4 Suppl):Abstract nr OT3-05-01.

Abstract Background: CALGB 40502/NCCTG N063H (Alliance) compared weekly NP or Ix to P; most patients received bevacizumab. Ix was inferior to P, and NP was not superior with a trend toward inferiority. Toxicity was increased in the experimental arms compared to P (Rugo et al, JCO 2015). We report long-term follow-up (FU) of this trial with an unplanned subset analysis in hormone receptor positive (HR+) and triple negative (TNBC) breast cancer. Methods: Patients were randomized 1:1:1 to receive P (90 mg/m2), Ix (16 mg/m2) or NP (150 mg/m2) on a 3 week (wk) on, 1 wk off schedule, stratified by prior adjuvant taxane use and hormone receptor status. B was initially given to all patients, but became optional in 3/2011 and was added to stratification. The primary endpoint was progression-free survival (PFS); secondary endpoints included safety and overall survival (OS). With a target N=900 patients, the study was powered to detect a hazard ratio of 1.36 (median PFS 10 vs 13.6 months). Eligibility included no prior chemotherapy for MBC, >12 mo from adjuvant P and measurable disease. Results: 799 patients were randomized between 11/08 and11/11 (283 to P, 271 to NP, 245 to Ix); 98% received bevacizumab. 68% (546) had HR+ disease, 25% (201) had TNBC. Median FU is 5 years. Median PFS is unchanged at 10.8, 9.2 and 7.4 mo for P, NP and Ix with hazard ratios (95% CIs) of 1.13 (0.94-1.34) and 1.44 (1.2-1.72) for NP and Ix to P, respectively. Median OS was 27.1, 24.2 and 23.6 months for P, NP and Ix with hazard ratios of 1.10 (0.91-1.34) and 1.3 (1.07-1.57) for NP and Ix to P, respectively. The effects of NP vs P on PFS and OS were significantly modified by subtype (interaction p=0.0018 and 0.0073), whereas Ix vs P was unchanged (interaction p's > 0.9, Table). More patients discontinued treatment due to adverse events in the experimental arms (14 vs 27 vs 23% for P, NP and Ix). Table P (mo)NP (mo)NP to P; HR (95% CI)Ix (mo)Ix to P, HR (95% CI)TNBC, PFS6.47.40.79 (0.55-1.12)15.61.39 (0.99-1.96)3HR+, PFS12.29.61.29 (1.04-1.59)181.5 (1.21-1.86)3TNBC, OS15.3210.74 (0.51-1.07)215.11.28(0.9-1.82)4HR+, OS33.226.61.25 (0.99-1.58)225.41.35(1.07-1.714Interaction tests: 1. p=0.0018; 2. p=0.0073; 3. p=0.96; 4. p=0.92 mo: months; HR: hazard ratio·· Conclusion: In patients with chemotherapy-naive MBC, Ix was inferior to P for PFS, and P was better tolerated than either NP or Ix. In this retrospective subset analysis, Ix and NP were inferior to P in HR+ disease, with a suggestion of improved PFS and OS with NP in patients with TNBC. Further investigation is required to explain and validate the subtype specificity seen in this exploratory analysis. Support: U10CA180820, U10CA180821, U10CA180882, U10CA180888. ClinicalTrials.gov Identifier: NCT00785291 Citation Format: Rugo HS, Barry WT, Moreno-Aspitia A, Lyss A, Huebner L, Mayer EL, Naughton M, Layman RM, Carey LA, Somer RA, Toppmeyer D, Velasco M, Perez EA, Hudis CA, Winer E. Long-term follow-up of CALGB 40502/NCCTG N063H (Alliance): A randomized phase III trial of weekly paclitaxel (P) compared to weekly nanoparticle albumin bound nab-Paclitaxel (NP) or ixabepilone (Ix) +/- bevacizumab as first-line therapy for locally recurrent or metastatic breast cancer (MBC) [abstract]. In: Proceedings of the 2017 San Antonio Breast Cancer Symposium; 2017 Dec 5-9; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2018;78(4 Suppl):Abstract nr GS3-06.

Abstract Introduction: MutHER is a phase II trial that demonstrated the anti-tumor activity of the pan-HER inhibitor neratinib in HER2mut, non-amplified MBC. The major challenges to accrue to this trial were the large number of pts to screen for the 2-3% HER2mut population and the high rate (24%) of poor quality tumor DNA for sequencing. The goals of this ctDNA study were: 1) the concordance of HER2mut detected by ctDNA versus tumor testing; 2) the incidence of ctDNA HER2mut in HER2 non-amplified MBC; 3) changes in HER2mut variant allele frequency (VAF) on neratinib therapy. Methods: A sample size of 30 negative (neg) controls was needed to ensure 90% confidence if ctDNA testing has >90% specificity in detecting HER2mut. Thus, plasma from MBC pts obtained at screening for MutHER trial (Neg control: 40 pts without HER2mut on tumor testing; Positive (pos) control: 14 pts with known HER2mut who received neratinib) were subjected to Guardant360 ctDNA 70-gene panel sequencing (all exons of HER2 included). ctDNA from the 14 neratinib treated pts were also analyzed at week (wk) 4 and upon progression. ctDNA data from MBC pts clinically tested at Guardant Health were interrogated for HER2mut incidence. Results: Among the 14 pts with tumor pos for HER2mut, ctDNA sequencing identified the same HER2mut in 11, discrepant HER2mut in 1, and neg in 2. The 2 pts with ctDNA neg for HER2mut had progressive disease (PD) and stable disease (SD > 6 months) on neratinib, respectively. Among the 40 neg controls, 8 were not evaluable (no detectable ctDNA or assay unsuccessful) and all 32 successfully sequenced cases were neg for HER2mut. The sensitivity and specificity of ctDNA for HER2mut detection was 11/14 (79%, 90% CI: 53-94%) and 32/32 (100%, 90% CI: 91-100%), respectively. Among the 11 paired baseline and wk 4 samples, 9 (82%) had lower HER2mut VAFs at wk 4 than at baseline, with 1 complete response (CR), 1 partial response (PR), 5 SD, and 4 PD at wk 8 as best tumor response. Two pts had higher wk 4 ctDNA HER2mut VAFs and both had radiographic PD at wk 8. The absolute HER2mut VAF levels at wk 4 were significantly associated with TTP (Spearman rho=-0.69, p=0.02) and tumor size change (rho=0.67, p=0.05). The HER2mut VAFs were significantly higher at progression than wk 4 in all pts (p<0.01). One pt acquired a new HER2mut T798I, which is analogous to the gate-keeper mutation EGFR T790M. The incidence of HER2mut without amplification in unselected consecutive MBC clinically tested by Guardant360 was 3% (48/1,584), with mutation pattern similar to published tumor testing data. Conclusions: ctDNA sequencing is sensitive and highly specific in detecting HER2mut, offering a non-invasive method to identify pts for trials of HER2mut-targeted therapy. Decreased HER2mut VAFs at wk 4 was observed in 82% of cases, consistent with the on-target effect of neratinib. Increased HER2mut VAFs at wk 4 is a potential early marker of progression. Citation Format: Cynthia Ma, Ron Bose, Feng Gao, Rachel Freedman, Melinda Telli, Gretchen Kimmick, Eric Winer, Michael Naughton, Matthew Goetz, Christy Russell, Debu Tripathy, Melody Cobleigh, Andres Forero, Timothy Pluard, Carey Anders, Shana Thomas, Jill Anderson, Caroline Bumb, Kimberly Banks, Richard Lanman, Richard Bryce, Alshad Lalani, John Pfeifer, Daniel Hays, Mark Pegram, Kimberly Blackwell, Philippe Bedard, Hussam Al-Kateb, Matthew Ellis. Circulating tumor DNA (ctDNA) sequencing for HER2 mutation (HER2mut) screening and response monitoring to neratinib in metastatic breast cancer (MBC) [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2017; 2017 Apr 1-5; Washington, DC. Philadelphia (PA): AACR; Cancer Res 2017;77(13 Suppl):Abstract nr CT011. doi:10.1158/1538-7445.AM2017-CT011

Abstract Background: In vivo synergy with concurrent PI3-Kinase inhibition and PARP inhibition has been observed in BRCA-deficient and BRCA-proficient preclinical models of triple negative breast cancer (TNBC) and ovarian cancer (OC). A phase I trial of the oral pan-class I PI3-Kinase inhibitor BKM120 and the PARP inhibitor olaparib demonstrated anti-cancer activity in TNBC and OC, both in patients with and without germline BRCA1 and BRCA2 (BRCA) mutations. However, CNS toxicity (depression) and liver function test abnormalities limited dose escalation of BKM120 prompting evaluation of the alpha specific PI3-Kinase inhibitor BYL719 (which has no CNS toxicity) in combination with olaparib. Methods: Olaparib was administered twice daily (tablet formulation) and BYL719 daily on a 28-day cycle, both orally. A 3 + 3 dose-escalation design was employed with primary objectives of defining the maximum tolerated dose (MTD) and recommended phase 2 dose of the combination of BYL719 and olaparib, and secondary objectives of defining toxicity, activity, and pharmacokinetic profiles of both agents. Eligibility included recurrent TNBC or high grade serous (HGS) OC, or any histology OC or breast cancer (BC) with presence of a known germline BRCA mutation, performance status of 0-1 and measurable/evaluable cancer. Patients with platinum sensitive or resistant or refractory OC were eligible and prior PARP inhibitor use was allowed. Dose-expansion cohorts at the MTD were enrolled for both BC and OC. Results: 46 patients (16 BC and 30 OC) have been enrolled in the study; 28 patients participated in the dose escalation portion of the study (4 BC and 24 OC). Two patients with OC did not receive study drugs because of ineligibility. MTD was defined as BYL719 200mg once daily and olaparib 200mg twice daily. Dose limiting toxicities included hyperglycemia, rash and fever with decreased neutrophil count. Four patients (3 OC and 1 BC) discontinued protocol therapy because of toxicity (2 for hyperglycemia, 1 for nausea and 1 for allergic reaction). Most common toxicities included nausea, hyperglycemia, fatigue, diarrhea and vomiting. At the MTD, 6 patients with OC and 12 patients with BC were enrolled into a dose expansion cohort. The OC expansion cohort has completed enrollment, while the BC cohort is still enrolling. Among patients with OC who received study drugs (28 patients, 26 (93%) with platinum resistant disease), objective response rate (ORR) by RECIST 1.1 was 36% (10/28 patients, all partial responses (PRs)). Median duration of response was 167 days (range 16-398 days); 5 of 10 patients with PR remain on treatment. ORR was 33% for patients with germline BRCA mutations and 31% for patients without germline BRCA mutations. Among patients without germline BRCA mutations with platinum resistant OC, ORR was 29%. Conclusions: Combined BYL719 and olaparib is feasible, and similar clinical benefit was observed in patients with and without germline BRCA mutations. The activity of this combination in OC patients without germline BRCA mutations and with platinum resistant disease was higher than expected from olaparib monotherapy and warrants further investigation. This work was funded in part by the Stand Up To Cancer Ovarian Dream Team. Clinical trial: NCT01623349. Citation Format: Panagiotis A. Konstantinopoulos, William T. Barry, Michael Birrer, Shannon N. Westin, Sarah Farooq, Karen Cadoo, Christin Whalen, Weixiu Luo, Hui Liu, Carol Aghajanian, David B. Solit, Gordon B. Mills, Barry S. Taylor, Helen Won, Michael F. Berger, Sangeetha Palakurthi, Joyce F. Liu, Lew Cantley, Scott H. Kaufmann, Elizabeth M. Swisher, Alan D. D'Andrea, Eric Winer, Gerburg M. Wulf, Ursula A. Matulonis. Phase I study of the alpha specific PI3-Kinase inhibitor BYL719 and the poly (ADP-Ribose) polymerase (PARP) inhibitor olaparib in recurrent ovarian and breast cancer: Analysis of the dose escalation and ovarian cancer expansion cohort [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2017; 2017 Apr 1-5; Washington, DC. Philadelphia (PA): AACR; Cancer Res 2017;77(13 Suppl):Abstract nr CT008. doi:10.1158/1538-7445.AM2017-CT008

Abstract Purpose: Based on promising preclinical data, we conducted a single-arm phase II trial to assess the clinical benefit rate (CBR) of neratinib, defined as complete/partial response (CR/PR) or stable disease (SD) ≥24 weeks, in HER2mut nonamplified metastatic breast cancer (MBC). Secondary endpoints included progression-free survival (PFS), toxicity, and circulating tumor DNA (ctDNA) HER2mut detection. Experimental Design: Tumor tissue positive for HER2mut was required for eligibility. Neratinib was administered 240 mg daily with prophylactic loperamide. ctDNA sequencing was performed retrospectively for 54 patients (14 positive and 40 negative for tumor HER2mut). Results: Nine of 381 tumors (2.4%) sequenced centrally harbored HER2mut (lobular 7.8% vs. ductal 1.6%; P = 0.026). Thirteen additional HER2mut cases were identified locally. Twenty-one of these 22 HER2mut cases were estrogen receptor positive. Sixteen patients [median age 58 (31–74) years and three (2–10) prior metastatic regimens] received neratinib. The CBR was 31% [90% confidence interval (CI), 13%–55%], including one CR, one PR, and three SD ≥24 weeks. Median PFS was 16 (90% CI, 8–31) weeks. Diarrhea (grade 2, 44%; grade 3, 25%) was the most common adverse event. Baseline ctDNA sequencing identified the same HER2mut in 11 of 14 tumor-positive cases (sensitivity, 79%; 90% CI, 53%–94%) and correctly assigned 32 of 32 informative negative cases (specificity, 100%; 90% CI, 91%–100%). In addition, ctDNA HER2mut variant allele frequency decreased in nine of 11 paired samples at week 4, followed by an increase upon progression. Conclusions: Neratinib is active in HER2mut, nonamplified MBC. ctDNA sequencing offers a noninvasive strategy to identify patients with HER2mut cancers for clinical trial participation. Clin Cancer Res; 23(19); 5687–95. ©2017 AACR.