Search

Background: About half of all breast cancers exhibit low HER2 expression. Despite lack of ERBB2 amplification, HER2-low tumors respond to trastuzumab deruxtecan (T-DXd), leading to the NCCN recommendation of T-DXd both for patients with HER2+ and HER2-low metastatic breast cancer (MBC). It remains however unclear if HER2-low represents a distinct molecular entity, as compared to HER2-0 MBC. Here, we compare the genomic landscape of HER2-low versus HER2-0 breast cancers in a large, single institution cohort. Methods: We identified consecutive patients with MBC seen at Dana-Farber Cancer Institute between 07/2013 and 12/2020. Patients were included if they had HER2-negative MBC per ASCO/CAP Guidelines and had undergone next generation sequencing (NGS) testing with a targeted, tumor-only platform (OncoPanel). Based on the HER2 status of the specimen tested by NGS, patients were divided into 2 groups: (i) HER2-low if immunohistochemistry (IHC) 1+ or 2+ non-amplified, or (ii) HER2-0 if IHC 0. Mutations of interest detected on NGS were classified as oncogenic using the OncoKB tool and additional annotation. Genomic profiles of HER2-low and HER2-0 tumors were compared using Chi-Square and Kruskal-Wallis tests. To determine genomic event enrichment between the two HER2 groups, logistic regression models were used, accounting for background rate and estrogen receptor (ER) expression. ERBB2 copy counts were calculated for tumors with recorded histology-estimated purities and copy-number segmentation using a simple model of allelic gain/loss. Results: Among 1847 patients with HER2-negative MBC, 1043 underwent NGS testing on a HER2-low (n=489, 47%) or HER2-0 sample (n=554, 53%). Most samples were metastatic (71%, n=743) while 29% (n=300) were from primary tumors. 73% had ductal histology, 13% were lobular and 14% had mixed or other histology. ER expression was enriched among HER2-low vs. HER2-0 tumors (76% vs. 60%; p< 0.001). Focusing on the most commonly occurring genetic mutations, no major differences were observed in HER2-low vs. HER2-0 tumors, after correcting for ER status (Table 1). Among all mutational events, any mutation in MPL, CYLD, and MAP3K and oncogenic mutations in TP53 and NF1 were more common in HER2-0, while any mutation in MTOR, RAD21, DNMT3A, and PDGFRA were enriched in HER2-low patients, when controlling for ER status and background mutational rate (p< 0.05). However, no mutation reached significance after accounting for multiple hypothesis testing. Similarly, no deep deletion or high amplification CNV events reached significance for either group. Analysis of tumor mutational burden in HER2-low vs. HER-0 tumors revealed no significant differences (median: 7.26 muts/Mb vs. 7.60 muts/Mb, p=1.00), including when accounting for ER status. Finally, among tumors with sufficient tumor purity for ERBB2 copy count analysis (n=374 and 419 for HER2-low and HER2-0, respectively), HER2-low tumors had a significantly higher number of ERBB2 alleles as compared to HER2-0 (< 2 copies, 15.0% vs. 30.9%, 2 copies 67.4% vs. 60.5%, and >2 copies, 17.6% vs. 8.6%; p< 0.001 by Kruskal-Wallis). Conclusions: To our knowledge, this is the largest comprehensive genomic analysis of HER2-low MBC to date. In our cohort of patients with HER2-negative MBC, the genomic landscape of HER2-low and HER2-0 tumors did not differ significantly, apart from a higher number of ERBB2 alleles. These data further support the notion that HER2-low, as currently defined, is not a distinct molecular subtype of breast cancer. Citation Format: Paolo Tarantino, Hersh V. Gupta, Melissa E. Hughes, Janet L. Files, Sarah Strauss, Gregory Kirkner, Anne-Marie Feeney, Yvonne Y. Li, Ana C. Garrido-Castro, Romualdo Barroso-Sousa, Brittany Bychkovsky, Laura MacConaill, Neal Lindeman, Bruce Johnson, Matthew Meyerson, Sheheryar Kabraji, Rinath Jeselsohn, Xintao Qiu, Rong Li, Henry W. Long, Eric Winer, Deborah A. Dillon, Giuseppe Curigliano, Andrew Cherniack, Sara Tolaney, Nancy U. Lin. HER2-05 Comprehensive genomic characterization of HER2-low breast cancer [abstract]. In: Proceedings of the 2022 San Antonio Breast Cancer Symposium; 2022 Dec 6-10; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2023;83(5 Suppl):Abstract nr HER2-05.

Purpose: Brain metastases can occur in up to 50% of patients with metastatic HER2-positive breast cancer. Because patients with active brain metastases were excluded from previous pivotal clinical trials, the central nervous system (CNS) activity of the antibody–drug conjugate trastuzumab deruxtecan (T-DXd) is not well characterized. Experimental Design: We studied how T-DXd affects growth and overall survival in orthotopic patient-derived xenografts (PDX) of HER2-positive and HER2-low breast cancer brain metastases (BCBM). Separately, we evaluated the effects of T-DXd in a retrospective cohort study of 17 patients with stable or active brain metastases. Results: T-DXd inhibited tumor growth and prolonged survival in orthotopic PDX models of HER2-positive (IHC 3+) and HER2-low (IHC 2+/FISH ratio < 2) BCBMs. T-DXd reduced tumor size and prolonged survival in a T-DM1–resistant HER2-positive BCBM PDX model. In a retrospective multi-institutional cohort study of 17 patients with predominantly HER2-positive BCBMs, the CNS objective response rate (ORR) was 73% (11/15) while extracranial response rate was 45% (5/11). In the subset of patients with untreated or progressive BCBM at baseline, the CNS ORR was 70% (7/10). The median time on treatment with T-DXd was 8.9 (1.3–16.2) months, with 42% (7/17) remaining on treatment at data cutoff. Conclusions: T-DXd demonstrates evidence of CNS activity in HER2-positive and HER2-low PDX models of BCBM and preliminary evidence of clinical efficacy in a multi-institution case series of patients with BCBM. Prospective clinical trials to further evaluate CNS activity of T-DXd in patients with active brain metastases are warranted. See related commentary by Soffietti and Pellerino, p. 8

Background: Despite advances in treatment options, outcomes remain poor for many pts with breast cancer brain metastases (BCBMs). Identifying genomic predictors of brain metastasis from primary tumors could lead to better stratification of pts at risk and drive the development of preventative strategies. The objective of this study was to describe the landscape of genomic alterations in primary tumors from pts with MBC who subsequently did or did not develop BCBMs. Methods: We performed a case control study to identify somatic alterations in primary tumors associated with a higher incidence of brain metastases. We reviewed outcomes for 2562 unique MBC patients from a single institution who underwent targeted next-generation DNA sequencing of > 280 cancer-related genes (OncoPanel) from their tumor between July 1, 2013 and December 31, 2020. Pts were included in this analysis if they had at least 2 years of follow-up from date of metastatic diagnosis and OncoPanel testing on a primary breast tumor. We compared single nucleotide variants (oncogenic or likely oncogenic), copy number variation (amplification and deep deletions) and tumor mutation burden in the primary tumors of pts in this cohort. Copy number variation was corrected for Panel version and tumor purity. Wilcoxon rank sum test and Fisher exact test was used to compare genomic differences between groups. False discovery rate was used to correct for multiple hypothesis testing and q < 0.1 was considered significant Results: A total of 369 pts were included in the final analytic cohort. Of these, 115 were diagnosed with brain mets (cases, BM group) and 224 were not (controls, nBM group). The BM group was enriched for patients with HER2-positive breast cancer (33 vs 12.5%), consistent with previous work. In the whole cohort, the most common and clinically significant somatic alterations (oncogenic single nucleotide variants or copy number high amplification or two copy deletion) are shown in Table 1. When adjusting for subtype there were no significantly enriched SNVs in BM vs nBM group. When adjusting for subtype, FGFR1 amplification was significantly enriched in hormone receptor positive HER2 negative (HR+ HER2-) patients with BM (log2 odds ratio 1.22, q < 0.1). Tumor mutation burden was not significantly different in primary tumors between the BM and nBM groups (median TMB 7.3 vs 6.1, Wilcoxon p = 0.08). Pathway analysis combining all subtypes revealed that RTK_RAS pathway (log2 odds ratio 1.64, q value < 0.1) and TP53 pathway (log2 odds ratio 1.15, q value < 0.1) gene sets were significantly enriched in the BM group. When controlling for subtype, pathway analysis revealed that RTK_RAS pathway gene set was significantly enriched in HR+ HER2- BM group (log2 odds ratio 1.36 q < 0.1). Conclusions: In this case control series of patients with metastatic breast cancer with or without brain metastases, we found that primary tumors that are enriched for somatic alterations in the RTK_RAS and TP53 pathway may be associated with higher risk of developing brain metastases. Further validation in larger cohorts is warranted. Table 1. Frequency of somatic alterations in primary tumor by brain metastasis outcome. Citation Format: Sheheryar Kabraji, Yvonne Y. Li, Melissa E. Hughes, Hersh V. Gupta, Lauren Buckley, Janet L. Files, Ayesha Mohammed-Abreu, Anne-Marie Feeney, Greg Kirkner, Ashka Patel, Ana C. Garrido-Castro, Romualdo Barroso-Sousa, Brittany Bychkovsky, Matthew Meyerson, Sara Tolaney, Deborah A. Dillon, Bruce Johnson, Eric Winer, Andrew Cherniack, Nancy U. Lin. Somatic alterations in primary tumors of patients (pts) with metastatic breast cancer (MBC) may predict likelihood of brain metastasis [abstract]. In: Proceedings of the 2022 San Antonio Breast Cancer Symposium; 2022 Dec 6-10; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2023;83(5 Suppl):Abstract nr PD7-07.

Purpose: Up to half of patients (pts) with HER2+ metastatic breast cancer (MBC) will develop BCBM. Trastuzumab deruxtecan (T-DXd) is an antibody-drug conjugate (ADC) with demonstrated efficacy in previously treated pts with HER2+ MBC. However, few pts with stable/treated brain metastases, and no pts with active (untreated or progressive) brain metastases, were included in completed clinical trials of T-DXd. Thus, central nervous system (CNS) efficacy of T-DXd is not well-characterized. Methods: We tested T-DXd in orthotopic pt BCBM-derived xenograft (PDX) models of HER2+, and HER2-low, BCBMs. To further validate T-DXd single agent CNS activity, we described clinical outcomes of T-DXd in a multi-institutional retrospective cohort of 16 patients with BCBM. Consecutive pts who initiated T-DXd between 1 Jan 2020 - 1 Nov 2020 (Duke) or 1 Jan 2020 - 15 June 2020 (Dana-Farber Cancer Institute) were included. Data cut-off date was 31 Dec 2020. CNS response was measured by via central radiology review at each participating institution. Up to 5 CNS target lesions were included. CNS partial response (PR) required >30% reduction in sum of CNS target lesions. Results: Treatment with 10 mg/kg T-DXd significantly prolonged survival in HER2+ BCBM PDX models DFBM-354 (67 vs 154 days, p=0.0018) and DFBM-355 (78 vs 156 days, p=0.0067) vs. vehicle control. We then tested TDX-d in a HER2 low BCBM PDX model (IHC 2+/FISH ratio Table 1.Best CNS Response of study cohortOverall population (n=16)Pts with progressive or untreated CNS disease at baseline (N=9)Pts with stable/treated CNS disease at baseline (N=7)Complete response (CR)0 (0)00 (0)Partial response (PR)10 (63)6 (67)#4 (57)+Stable disease (SD)2 (13)2 (22)0 (0)Progressive disease (PD)1 (6)0 (0)1 (14)No measurable CNS disease at baseline2 (13)0 (0)2 (29)Lost to follow-up1 (6)1 (11)0 (0).#Among these 6 pts, interval between most recent radiation and T-DXd initiation was 14.3 months, 12.4 months, 17 months, 18.2 months, and 8.2 months (one pt was radiation-naïve). +Among these 4 pts, interval between most recent radiation and T-DXd initiation was 19.1 months, 15.1 months, 1.5 months, and 16.9 months, respectively. Citation Format: Sheheryar Kabraji, Jing Ni, Sarah Sammons, Amanda ED Van Swearingen, Yanzhi Wang, Alyssa M Pereslete, Liangge Hsu, Chris Lascola, Heather Moore, Melissa Hughes, Akshara S Raghavendra, Maria Gule-Monroe, Rashmi K Murthy, Eric P Winer, Carey K Anders, Jean J Zhao, Nancy U Lin. Preclinical and clinical efficacy of trastuzumab deruxtecan in breast cancer brain metastases (BCBM) [abstract]. In: Proceedings of the 2021 San Antonio Breast Cancer Symposium; 2021 Dec 7-10; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2022;82(4 Suppl):Abstract nr PD4-05.

Small-molecule targeted therapies have demonstrated outstanding potential in the clinic. These drugs are designed to minimize adverse effects by selectively attacking cancer cells while exerting minimal damage to normal cells. Although initial response to targeted therapies may be high, yielding positive response rates and often improving survival for an important percentage of patients, resistance often limits long-term effectiveness. On the other hand, immunotherapy has demonstrated durable results, yet for a limited number of patients. Growing evidence indicates that some targeted agents can modulate different components of the antitumor immune response. These include immune sensitization by inhibiting tumor cell–intrinsic immune evasion programs or enhancing antigenicity, as well as direct effects on immune effector and immunosuppressive cells. The combination of these two approaches, therefore, has the potential to result in synergistic and durable outcomes for patients. In this review, we focus on the latest advances on integrating immunotherapy with small-molecule targeted inhibitors. In particular, we discuss how specific oncogenic events differentially affect immune response, and the implications of these findings on the rational design of effective combinations of immunotherapy and targeted therapies.

Human tumor growth depends on rapidly dividing cancer cells driving population expansion. Even advanced tumors, however, contain slowly proliferating cancer cells for reasons that remain unclear. Here, we selectively disrupt the ability of rapidly proliferating cancer cells to spawn AKT1low daughter cells that are rare, slowly proliferating, tumor-initiating, and chemotherapy-resistant, using β1-integrin activation and the AKT1-E17K–mutant oncoprotein as experimental tools in vivo. Surprisingly, we find that selective depletion of AKT1low slow proliferators actually reduces the growth of a molecularly diverse panel of human cancer cell xenograft models without globally altering cell proliferation or survival in vivo. Moreover, we find that unusual cancer patients with AKT1-E17K–mutant solid tumors also fail to produce AKT1low quiescent cancer cells and that this correlates with significantly prolonged survival after adjuvant treatment compared with other patients. These findings support a model whereby human solid tumor growth depends on not only rapidly proliferating cancer cells but also on the continuous production of AKT1low slow proliferators. Mol Cancer Ther; 17(1); 254–63. ©2017 AACR.

Ductal carcinoma in situ (DCIS) of the breast precedes the development of invasive breast cancer and reflects the genomic changes and protein expression profile of invasive disease. AKT1low cancer cells (QCC) are a rare, drug-tolerant, epigenetically plastic, and quiescent cancer cell subset that we previously identified at a frequency of 0.5–1% in primary breast tumors using the marker profile: AKTlow/H3K9me2low/HES1high. Here we used quantitative immunofluorescence microscopy with computational image analysis to show that AKT1low QCCs are present in DCIS from patients with and without co-existing invasive breast cancer. These data suggest that a drug-resistant, quiescent cancer cell state is present in premalignant breast lesions prior to the development of invasive disease. These findings warrant further study of whether AKT1low QCCs contribute to invasive tumor development and recurrence, similar to their role in more advanced malignancy.

Uncontrolled cell proliferation is a defining feature of malignancy. Current understanding of proliferation, particularly in humans, derives primarily from studying cells growing rapidly in the non-physiological conditions of cell culture. However, tumors are exceedingly complex admixtures of different cell types and subclonal malignant populations comprising proliferative, non-proliferative, and arrested states that are influenced by physical, metabolic, and molecular conditions. Images of single or small sets of protein markers from fixed tissue samples only provide limited and static views into the nature of these complex states. Here we identify proliferation states and develop a quantitative framework to extract cell cycle dynamics from multiplexed, spatially-resolved tissue images of millions of tumor cells from human cancers and genetically engineered tumors in mice. Across spatial scales, tumors display intrinsic regional variability in proliferation patterns and in the coherence of cell cycle markers in high-dimensional space. Cell cycle dynamics and cell cycle coherence are not solely a function of tumor growth and oncogene expression and rapidly adapt following genetic and therapeutic perturbations. Replacing binary metrics with multivariate traits provides a quantitative framework for extracting multidimensional dynamic information from static images that is broadly applicable to the study of temporal processes within the native architecture of human disease tissues. Citation Format: Giorgio Gaglia, Sheheryar Kabraji, Danae Argyropoulou, Yang Dai, Johann Bergholz, Shannon Coy, Jia-Ren Lin, Eric P. Winer, Deborah Dillon, Jean J. Zhao, Peter K. Sorger, Sandro Santagata. Temporal and spatial topography of cell proliferation in cancer [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2021; 2021 Apr 10-15 and May 17-21. Philadelphia (PA): AACR; Cancer Res 2021;81(13_Suppl):Abstract nr 4.

Approximately 50% of patients with metastatic HER2-positive (HER2+) breast cancer develop brain metastases (BCBMs). To identify predictors of response and resistance to brain-penetrant, small molecule targeted therapies, we performed whole exome sequencing, RNAseq and quantitative immunohistochemistry of HER2+ BCBM tissues from 21 patients. We found that the tumor suppressor p16INK4A (p16) was deficient in the majority of HER2+ BCBMs. Orthotopic HER2+ BCBM patient-derived xenograft (PDX) models with p16 deficiency were resistant to the brain-penetrant HER2 inhibitor, tucatinib or the brain penetrant CDK4/6 inhibitor, abemaciclib, when used as single agents. However, p16-deficiency predicted response to combined tucatinib and abemaciclib in orthotopic HER2+ BCBM PDX models. These data establish the rationale for a biomarker-driven clinical trial of combined CDK4/6- and HER2-targeted agents for patients with HER2+ BCBM. Citation Format: Jing Ni, Sheheryar Kabraji, Shaozhen Xie, Yanzhi Wang, Peichen Pan, Xiaofang He, Zongming Liu, Henry Long, Myles Brown, Eric P. Winer, Deborah Dillon, Nancy Lin, Jean Zhao. p16INK4A-deficiency predicts for response to combined HER2 and CDK4/6 inhibition in HER2+ breast cancer brain metastases [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2021; 2021 Apr 10-15 and May 17-21. Philadelphia (PA): AACR; Cancer Res 2021;81(13_Suppl):Abstract nr 337.

3122 Background: Tumors are complex ecosystems where exogenous and endogenous cues are integrated to either stimulate or inhibit cancer cell proliferation. However, the nature of these complex cell cycle states, their spatial organization, response to perturbation, and implications for clinical outcomes, are poorly characterized in tumor tissues. Methods: We used multiplexed tissue imaging to develop a robust classifier of proliferation, the multivariate proliferation index (MPI), using 513 unique tumors across five cancer types. Next, we used dimensionality reduction analysis to assess how the patterns of cell cycle protein expression in tumors were altered in response to perturbation. Results: The MPI outperforms single markers, like Ki67, when classifying proliferative index across diverse tumor types and reveals the proliferative architecture of tumors in situ. We find that proliferative and non-proliferative cancer cells are organized across microscopic (cell-to-cell) and macroscopic (tissue-level) scales. Both domains are reshaped by therapy, and local clusters of proliferative and non-proliferative tumor cells preferentially neighbor distinct tumor-infiltrating immune cells. We further phenotyped non-proliferating cancer cells using markers of quiescent cancer cells, cancer stem cells, and dormant cancer cells. We found that these types of non-proliferating cancer cells can occupy distinct regions within the same primary tumor. In high-dimensional marker space, populations of proliferative cancer cells express canonical patterns of cell cycle protein markers, a property we refer to as “cell cycle coherence”. Untreated tumors exist in a continuum of coherence states, ranging from optimal coherence, akin to freely cycling cells in culture, to reduced coherence characterized by either cell cycle polarization or non-canonical marker expression. Coherence can be stereotypically altered by induction and abrogation of mitogen signaling in a HER2-driven model of breast cancer. Cell cycle coherence is modulated by neoadjuvant therapy in patients with localized breast cancer, and coherence is associated with disease-free survival after adjuvant therapy in patients with colorectal cancer, mesothelioma and glioblastoma. Conclusions: The MPI robustly defines proliferating and non-proliferating cells in tissues, with immediate implications for clinical practice and research. The coherence metrics capture the diversity of post-treatment cell cycle states directly in clinical samples, a fundamental step in advancing precision medicine. More broadly, replacing binary metrics with multivariate traits provides a quantitative framework to study temporal processes from fixed static images and to investigate the rich spatial biology of human cancers.

2525 Background: Genomic characterization of breast cancer brain metastases (BCBMs) has thus far been limited. The objective of this study was to describe the landscape of genomic alterations in patients (pts) with BCBMs. Methods: Targeted next-generation DNA sequencing of > 300 cancer-related genes (OncoPanel) was prospectively performed on primary and metastatic (met) tumors in 321 pts with a diagnosis of BCBM between August 2016 and April 2019 at Dana-Farber Cancer Institute (table). Enrichment analysis of genomic alterations was performed using a two-sided Fisher exact test and differences in tumor mutation burden (TMB) between groups were assessed using two-sided Mann-Whitney U test. Multiple comparison correction was performed using the Benjamini-Hochberg procedure. Results: All subtypes were represented in BCBM (25 HR+/HER2-; 24 HR+/HER2+; 27 HR-/HER2+; 18 TNBC; 5 unknown; n = 99) and extracranial (EC) samples: (96 HR+/HER2-; 32 HR+/HER2+; 22 HR-/HER2+; 41 TNBC; 31 unknown; n = 222). BCBMs were found most commonly to have mutations or copy number alterations in TP53, ERBB2, PIK3CA, GATA3, PTEN, ESR1, CDH1, BRCA2, ARID1A, BRCA1 (>5% frequency, table). Two pts acquired ERBB2 amplification (amp) between the matched primary breast sample and brain met. In pair-wise comparisons of BCBMs to unmatched primaries or EC mets, only ERBB2 amp was significantly enriched (table, † = adjusted p < 0.05). There was no significant difference in TMB between BCBM and EC mets (median 9.12 vs 7.26, p = 0.15). In contrast, TMB was significantly higher in BCBMs compared to unmatched primaries (median 9.12 vs 7.26, p=0.005). Conclusions: BCBMs display similar mutations and copy number alterations compared to primary tumors and EC mets in pts with BCBM. These data suggest that BCBMs contain actionable genomic alterations that are most often also reflected in EC disease. Alterations in ERBB2, PIK3CA/PTEN, and BRCA1/2 represent potentially targetable alterations in pts with BCBM. [Table: see text]

The brain is the most common site of first metastasis for patients with HER2-positive breast cancer treated with HER2-targeting drugs. However, the development of effective therapies for breast cancer brain metastases (BCBM) is limited by an incomplete understanding of the mechanisms governing drug sensitivity in the central nervous system. Pharmacodynamic data from patients and in vivo models suggest that inadequate drug penetration across the “blood–tumor” barrier is not the whole story. Using HER2-positive BCBMs as a case study, we highlight recent data from orthotopic brain metastasis models that implicate brain-specific drug resistance mechanisms in BCBMs and suggest a translational research paradigm to guide drug development for treatment of BCBMs. Clin Cancer Res; 24(8); 1795–804. ©2018 AACR.

Up to half of patients with metastatic HER2-positive breast cancer will develop brain metastases. Breast cancer brain metastases (BCBM) are a major cause of morbidity and mortality, despite multimodal management including surgery, radiotherapy, and systemic therapies. Therefore, there is an urgent need to develop novel, efficacious therapies. Neratinib is an orally bioavailable, irreversible pan-HER tyrosine kinase inhibitor that is FDA-approved in the extended adjuvant treatment setting for HER2-positive, early breast cancer. Neratinib has only modest activity as a single agent in clinical trials of patients with HER2-positive brain metastases. Though the combination of neratinib and capecitabine results in CNS responses in up to half of patients, patients eventually develop drug resistance, and toxicities have been a concern-thus exploration of alternative neratinib combinations is of significant clinical interest. Ado-trastuzumab emtansine (T-DM1) is an antibody-drug conjugate with reported single-agent activity against HER2-positive BCBM. Here, we used HER2-positive orthotopic patient derived xenograft (PDX) models of BCBM to test if combining neratinib with T-DM1 could improve tumor response. PDX cells are labelled with luciferase to allow tumor growth measurement in vivo. We found that neratinib is able to reduce phosphorylated HER2 in an orthotopic PDX tumor derived from HER2-positive BCBM, indicating that neratinib can cross the BBB and inhibit HER2 activation in BCBM PDX tissues. However, in the HER2-positive DF-BM354 PDX model, single agent neratinib did not block orthotopic tumor growth compared to vehicle control as monitored by bioluminescence measurements. In contrast, combined treatment of neratinib with T-DM1 significantly reduced tumor growth compared to single agent treatment of neratinib or T-DM1 alone at earlier time points. At later time points, the combined treatment is comparable to T-DM1 alone. These data warrant further testing of neratinib alone or in combination with T-DM1 in additional BCBM PDX models to better understand drivers of resistance and susceptibility to HER2-inhibitors in HER2-positive BCBMs. Furthermore, they support the launch of a prospective clinical trial (NCT01494662) to test the efficacy and tolerability of T-DM1 in combination with neratinib in patients with progressive HER2-positive BCBM. Citation Format: Jing Ni, Yanzhi Wang, Irmina Diala, Sheheryar Kabraji, Rachel Freedman, Nancy Lin, Jean Zhao. Preclinical evaluation of neratinib plus T-DM1 in orthotopic PDX models of HER2-positive breast cancer brain metastases [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2019; 2019 Mar 29-Apr 3; Atlanta, GA. Philadelphia (PA): AACR; Cancer Res 2019;79(13 Suppl):Abstract nr 4832.

Purpose: Anaplastic lymphoma kinase (ALK) gene rearrangements define a distinct molecular subset of non–small cell lung cancer (NSCLC). Recently, several case reports and small series have reported that ALK rearrangements can overlap with other oncogenic drivers in NSCLC in crizotinib-naïve and crizotinib-resistant cancers. Experimental Design: We reviewed clinical genotyping data from 1,683 patients with NSCLC and investigated the prevalence of concomitant EGFR or KRAS mutations among patients with ALK-positive NSCLC. We also examined biopsy specimens from 34 patients with ALK-positive NSCLC after the development of resistance to crizotinib. Results: Screening identified 301 (17.8%) EGFR mutations, 465 (27.6%) KRAS mutations, and 75 (4.4%) ALK rearrangements. EGFR mutations and ALK rearrangements were mutually exclusive. Four patients with KRAS mutations were found to have abnormal ALK FISH patterns, most commonly involving isolated 5′ green probes. Sufficient tissue was available for confirmatory ALK immunohistochemistry in 3 cases, all of which were negative for ALK expression. Among patients with ALK-positive NSCLC who acquired resistance to crizotinib, repeat biopsy specimens were ALK FISH positive in 29 of 29 (100%) cases. Secondary mutations in the ALK kinase domain and ALK gene amplification were observed in 7 of 34 (20.6%) and 3 of 29 (10.3%) cases, respectively. No EGFR or KRAS mutations were identified among any of the 25 crizotinib-resistant, ALK-positive patients with sufficient tissue for testing. Conclusions: Functional ALK rearrangements were mutually exclusive with EGFR and KRAS mutations in a large Western patient population. This lack of overlap was also observed in ALK-positive cancers with acquired resistance to crizotinib. Clin Cancer Res; 19(15); 4273–81. ©2013 AACR.

Prostate adenocarcinoma (CaP) patients are classified into low-, intermediate-, and high-risk groups that reflect relative survival categories. While there are accepted treatment regimens for low- and high-risk patients, intermediate-risk patients pose a clinical dilemma, as treatment outcomes are highly variable for these individuals. A better understanding of the factors that regulate the progression of CaP is required to delineate risk. For example, aberrant activation of the Hedgehog (Hh) pathway is implicated in CaP progression. Here, we identify the serine protease inhibitor protease nexin 1 (PN1) as a negative regulator of Hh signaling in prostate. Using human CaP cell lines and a mouse xenograft model of CaP, we demonstrate that PN1 regulates Hh signaling by decreasing protein levels of the Hh ligand Sonic (SHH) and its downstream effectors. Furthermore, we show that SHH expression enhanced tumor growth while overexpression of PN1 inhibited tumor growth and angiogenesis in mice. Finally, using comparative genome hybridization, we found that genetic alterations in Hh pathway genes correlated with worse clinical outcomes in intermediate-risk CaP patients, indicating the importance of this pathway in CaP.

The degree of heterogeneity among cancer stem cells (CSC) remains ill-defined and may hinder effective anti-CSC therapy. Evaluation of oral cancers for such heterogeneity identified two compartments within the CSC pool. One compartment was detected using a reporter for expression of the H3K4me3 demethylase JARID1B to isolate a JARID1Bhigh fraction of cells with stem cell–like function. JARID1Bhigh cells expressed oral CSC markers including CD44 and ALDH1 and showed increased PI3K pathway activation. They were distinguished from a fraction in a G0-like cell-cycle state characterized by low reactive oxygen species and suppressed PI3K/AKT signaling. G0-like cells lacked conventional CSC markers but were primed to acquire stem cell–like function by upregulating JARID1B, which directly mediated transition to a state expressing known oral CSC markers. The transition was regulated by PI3K signals acting upstream of JARID1B expression, resulting in PI3K inhibition depleting JARID1Bhigh cells but expanding the G0-like subset. These findings define a novel developmental relationship between two cell phenotypes that may jointly contribute to CSC maintenance. Expansion of the G0-like subset during targeted depletion of JARID1Bhigh cells implicates it as a candidate therapeutic target within the oral CSC pool. Cancer Res; 76(18); 5538–49. ©2016 AACR.

The mechanisms that allow triple negative breast cancer tumors to survive neoadjuvant chemotherapy are incompletely understood. Evidence suggests that proliferative heterogeneity may contribute to primary chemotherapy resistance in patients with triple negative breast cancer. AKT1low quiescent cancer cells (QCCs) are a quiescent, epigenetically plastic, and chemotherapy resistant subpopulation initially identified in experimental cancer models. Here, we identify QCCs in primary and metastatic human breast tumors using automated, quantitative, immunofluorescence microscopy coupled with computational and statistical analysis. We show that QCCs exist as non-random and heterogeneously distributed clusters within primary tumors. In addition, these QCC clusters persist after treatment with multi-agent, multi-cycle, neoadjuvant chemotherapy in both residual primary tumors as well as nodal and distant metastases in patients with triple negative breast cancer. Together, these data qualify QCCs as a non-genetic mechanism of chemotherapy resistance in triple negative breast cancer patients that warrants further study. Citation Format: Sheheryar Kabraji, Xavier Sole, Ying Huang, Clyde Bango, Michaela Bowden, Aditya Bardia, Dennis Sgroi, Massimo Loda, Sridhar Ramaswamy. AKT1low quiescent cancer cells persist after neoadjuvant chemotherapy in triple-negative breast cancer patients [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 3173. doi:10.1158/1538-7445.AM2017-3173

11579 Background: The mechanisms that allow triple negative breast cancer (TNBC) tumors to survive neoadjuvant chemotherapy (NACT) are incompletely understood. Evidence suggests that proliferative heterogeneity may contribute to primary chemotherapy resistance in patients with localized triple negative breast cancer. However, the detailed characterization of a drug-resistant cancer cell state in residual TNBC tissue after NACT has remained elusive. AKT1lowquiescent cancer cells (QCCs) are a quiescent, epigenetically plastic, and chemotherapy resistant subpopulation initially identified in experimental cancer models. Here, we asked whether AKT1low QCCs actually exist in primary tumors from patients with TNBC and persist after treatment with NACT. Methods: We identified QCCs in primary and metastatic human breast tumors using automated, quantitative, immunofluorescence microscopy coupled with computational and statistical analysis. We obtained pre-treatment biopsy, post-treatment mastectomy, and metastatic specimens from a retrospective cohort of TNBC patients treated with neoadjuvant chemotherapy at Massachusetts General Hospital (n = 25). Using automated quantitative immunofluorescence microscopy, QCCs were identified as AKTlow / H3K9me2low / HES1high cancer cells using prespecified immunofluorescence intensity thresholds. QCCs were represented as 2D and 3D digital tumor maps and QCC percentage (QCC-P) and QCC cluster index (QCC-CI) were determined for each sample. Results: We found that QCCs exist as non-random and heterogeneously distributed clusters within primary tumors. In addition, these QCC clusters are enriched after treatment with multi-agent, multi-cycle, neoadjuvant chemotherapy in both residual primary tumors as well as nodal and distant metastases in patients with triple negative breast cancer. Conclusions: Together, these data qualify QCCs as a non-genetic mechanism of chemotherapy resistance in triple negative breast cancer patients that warrants further study.

8083 Background: Anaplastic lymphoma kinase (ALK) gene rearrangements define a distinct molecular subset of NSCLC. Recently, several studies have reported that ALK+ pts occasionally harbor concomitant mutations in other oncogenic drivers. Methods: We retrospectively analyzed tumor genotyping data from 1,683 pts with NSCLC seen at 3 U.S. centers from 2009 – 2012 to determine rates of overlapping alterations in EGFR, KRAS and ALK. Mutations in EGFR and KRAS were mainly identified using the SNaPshot multiplexed assay (>95% of cases). ALK FISH was performed in all cases. To determine if this prevalence is impacted by crizotinib, we also updated our earlier analysis (Katayama et al., Sci Transl Med, 2012) of a series of repeat biopsy specimens from 34 crizotinib-resistant, ALK+ pts. Resistant specimens were examined using ALK FISH, SNaPshot, and direct sequencing of the ALK tyrosine kinase domain (TKD). Results: Screening identified 301 (17.8%) EGFR mutations, 465 (27.6%) KRAS mutations, and 75 (4.4%) ALK rearrangements. EGFR mutations and ALK rearrangements were mutually exclusive. 4 pts with KRAS mutations also had abnormal ALK FISH patterns, involving isolated 5’ green probes (3/4 cases) and an isolated 3’ red probe that was unusually small (1/4 cases). Sufficient tissue was available for confirmatory ALK immunohistochemistry (clone 5A4, Novacastra, UK) in 3 of these cases, all of which were negative for ALK expression. Among pts with ALK+ NSCLC and acquired crizotinib resistance, repeat biopsy specimens remained ALK fusion positive in 28/28 (100%) cases. Secondary mutations in the ALK TKD (1151Tins, L1196M, G1202R, S1206Y, and G1269A) were identified in 7/34 (20.6%) cases. L1196M was the most common secondary mutation (3/34, 8.8% cases). ALK gene amplification was present in 3/28 (10.71%) pts. No EGFR or KRAS mutations were identified in 23 crizotinib-resistant, ALK+ pts with sufficient tissue for testing. Conclusions: Functional ALK rearrangements were mutually exclusive with EGFR and KRAS mutations in a large Western patient population. This lack of overlap was also observed in ALK+ pts with acquired resistance to crizotinib.

Prostate adenocarcinoma is the most frequent cancer and is the second leading cause of cancer death of men in the Western world. Aberrant activation of the Hedgehog (Hh) pathway is implicated in the growth and progression of prostate cancer. We show that the Hh pathway is inhibited by the serine protease inhibitor, protease nexin 1 (PN1), an extracellular matrix (ECM) protein highly expressed in normal prostate. We demonstrate here that PN1 regulates Hh signalling by reduction of the Hh ligand, Sonic (SHH), and its downstream effectors in tissue culture and in vivo. Conversely, matrix metalloproteinase 9 (MMP9) enhances Hh signalling through degradation of PN1. MRI imaging was used to determine variation in tumour formation in an orthotopic prostate tumour model using PN1 and MMP9 knock-out mice. In PN1-/- animals, SHH was increased and tumor growth was stimulated. In contrast, PN1 levels were elevated in MMP9-/- mice while Hh signalling was reduced with concomitant decreased tumor growth. Finally, genetic alterations in Hh pathway genes identified by comparative genomic hybridization (CGH) from intermediate risk prostate cancer patients correlated with a worse clinical outcome (HR=2.67, p=0.00067), indicating the importance of this pathway in human cancer. Our data identify PN1 as a novel regulator of SHH and its downstream targets allowing PN1 levels to influence prostate tumor growth. Regulation of the PN1-Hh pathway could constitute a significant therapeutic advance in treating cancers that exhibit high expression of Hh markers. These data also provide a mechanism through which MMP-9 can regulate hedgehog pathway signalling in tissues rich in PN1. Citation Format: {Authors}. {Abstract title} [abstract]. In: Proceedings of the 103rd Annual Meeting of the American Association for Cancer Research; 2012 Mar 31-Apr 4; Chicago, IL. Philadelphia (PA): AACR; Cancer Res 2012;72(8 Suppl):Abstract nr 2463. doi:1538-7445.AM2012-2463