Abstract P3-05-03: Metaplastic breast cancers: Genomic profile, mutational burden and TILs

Background: Metaplastic breast cancers (MPBC) are rare, typically triple negative aggressive tumors composed of both, adenocarcinoma and metaplastic elements. Recent evidence that TNBC and MPBC can respond vigorously to immune checkpoint inhibitor therapy (Adams et al, ASCO 2017 and npj Breast Cancer 2017) has prompted the following comprehensive genomic profiling (CGP) and histopathologic assessment of tumor infiltrating lymphocytes (TIL) designed to uncover potential biomarkers of immunotherapy response for MPBC, including mutational burden, Microsatellite Instability (MSI) status and gene amplification of 9p21.4 (or CD274, which includes the PD-L1 locus).

Methods: 12,214 locally aggressive, relapsed and metastatic breast malignancies (mBM) were subjected to CGP using DNA extracted from 40 µm of FFPE sections and adaptor ligation-based libraries to a mean coverage depth >650X for up to 315 cancer-related genes. The results were analyzed for all classes of genomic alterations (GA) including base substitutions, insertions and deletions, select rearrangements, and copy number changes. Tumor mutational burden (TMB) was determined on 1.1 Mbp of sequenced DNA. MSI status was determined by an algorithm based on the sequencing results. TIL were assessed on archived H&E tumor sections and enumerated per guidelines established by the TIL Working Group (Salgado, Ann Oncol 2015) in a subset of MPBC with the highest TMB and compared with low TMBC cases.

Results: 165 of mBM cases were MPBC (1.4%) and are included in this study. All patients were female with a median age of 60 (range 24-86). 165 of the MPBC cases (100%) harbored a wide variety of GA involving more than 100 individual genes. The most common GA were identified in TP53 (65%), followed by PIK3CA (37%). No cases of MSI hi status (0/103) and only one case with amplification of 9p21.4 (1/165, 0.6%) were observed. Most MPBC had a low mutational burden, with a median TMB of 2.7 mutations/Mb (range 0-39.6). Only 11/165 tumors (6.7%) were found to have a TMB over 10 mutations/Mb, including 3 cases (1.8%) with TMB >20. Tumor sections were available for TIL review from 9/11 cases with highest TMB, as well as 11 control cases with lowest TMB. TIL were more frequently observed in high versus low TMB MPBC, with median TIL percentage of 40 and 20 (range 10-80 and range 10-60), respectively, although this difference was not statistically significant (Wilcoxon rank-sum test, p=0.15).

Conclusions: Genomic profiling in the largest cohort of MPBC reported to date confirms that MPBC is enriched for TP53 and PIK3CA mutations and many tumors harbor targetable GA. The frequently observed tumoral PD-L1 expression in MPBC is not based on gene amplification as amplification of 9p21.4 is rare. Most tumors had a low mutation burden, and no significant association of TIL with TMB was observed, suggesting additional processes underlying MPBC immunogenicity.

Citation Format: Taff J, Suh J, Singh B, Denkert C, Troxel AB, Ross JS, Adams S. Metaplastic breast cancers: Genomic profile, mutational burden and TILs [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 P3-05-03.

Abstract P5-21-20: Integrating comprehensive genomic profiling with treatment decisions – Experience gained while treating 139 advanced breast carcinomas

Background: Comprehensive Genomic Profiling (CGP) using next-generation sequencing (NGS) technology can provide insight into potentially clinically relevant genomic alterations (CRGA) within a patient's breast cancer. For example, HER2 amplification status and targetable short variants (SV), acquired ESR1 or BRCA1/2 resistance mutations, and the presence of targetable alterations in the PI3K kinase. We retrospectively reviewed CGP results and subsequent outcomes at one cancer center to illustrate the experience of using molecular subtyping to inform treatment decisions.

Methods: DNA extracted from FFPE tumor tissue or blood samples obtained during routine clinical care for patients (n=139) with predominantly relapsed, refractory or metastatic breast cancer was analyzed by hybrid-capture, NGS for all classes of GA: 1. base substitutions, 2. insertion and deletions, 3. rearrangements, and 4. copy number changes. Treatment decisions based on comprehensive genomic profiles were captured retrospectively. Tumor mutational burden (TMB), scored as mutations (mut)/Mb, was calculated on 0.8-1.2 Mb of sequenced DNA. Alterations affecting the ERBB family included amplification of or oncogenic mutations in ERBB2 (HER2), ERBB3, and EGFR.

Results: From Jan 2013 to May 2017, FFPE tissue samples for 136 patients with advanced breast cancer were analyzed by CGP and 3 additional patients had circulating tumor DNA analyzed for alterations; 11 patients received profiling on multiple biopsies. Tumors analyzed were carcinomas (Ca) NOS (n=84), invasive ductal Ca (n=46), invasive lobular Ca (n=7), a neuroendocrine Ca, and a phyllodes tumor. In total, 118/139 (84.9%) samples harbored CRGA in a targetable pathway: PI3K/MTOR (n=67; 48.2%), CDK cell-cycle (n=40; 28.8%), ERBB family (n=24; 17.3%), FGFR (n=24; 17.3%), ESR1 (n=16; 11.5%), homologous repair (HRD)( n=14; 10.1%), and RAS/RAF/MEK (n=11; 7.9%). Targetable alterations in other cancer-related kinases were found in 10 (7.2%) samples and 10 (7.2%) samples were TMB high (≥20 mut/Mb) or had CD274 (PD-L1) amplification. There were 3 patients (2.1%) with HER2 short variants detected in the absence of ERBB2 amplification; these patients may respond to HER2-targeted therapies but would be HER2-negative by IHC. Many samples had alterations in ≥1 pathway, and overlap is particularly high for the CDK and FGFR pathways (12 samples). Alterations in pathways targeted by MTOR inhibitors, HER2-targeted therapies, or the CDK inhibitors were found in 93/136 (66.9%) tumors. Evaluation of outcomes for these 139 patients is ongoing and will be presented.

Conclusions: Genomic profiling of breast carcinomas, using either tissue or liquid biopsies, provides potentially actionable information to guide treatment decisions. Overall, 84.9% of patient samples harbored oncogenic alterations in a targetable pathway, with two-thirds of tumors having alterations in pathways targeted by therapies with FDA approval for breast cancer and 7.2% of patients having high levels of TMB or amplification of PD-L1, suggesting that checkpoint inhibitors may be relevant options.

Citation Format: Mahtani R, Gay LM, Chung J, Hartmaier R, Sokol E, Elvin JA, Daniel S, Ramkissoon S, Severson E, Suh J, Vergilio J-A, Stephens PJ, Ross JS. Integrating comprehensive genomic profiling with treatment decisions – Experience gained while treating 139 advanced breast carcinomas [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 P5-21-20.

Abstract PD8-01: CDH1 mutated classic and pleomorphic invasive lobular breast carcinomas differ in genomic signatures and opportunities for targeted and immunotherapies

Background: Typically defined by negative IHC staining for E-cadherin, classic (CILC) and pleomorphic (PILC) are often combined as a single breast cancer subtype. We queried whether patients with relapsed metastatic disease, mCILC and mPILC, would harbor contrasting genomic alterations (GA)and that molecular information could further differentiate the 2 tumor types and thereby influence therapy selection.

Methods: DNA was extracted from 40 µm of FFPE sections of 10,784 invasive breast carcinomas. 454 (4%) CDH1 mutated mILC were selected including 428 classic mCILC (94%) and 26 mPLIC (6%) subtypes. Comprehensive genomic profiling (CGP) was performed on hybridization-captured, adaptor ligation-based libraries to a mean coverage depth >600X for up to 315 cancer-related genes. Tumor mutational burden (TMB) was determined on 1.1 Mbp of sequenced DNA.

Results: mCILC and mPILC patients featured a median age of 63 years (Table). Slide based ER+ status and HER2+ status was significantly different in both groups (P<0.0001). The frequency of base substitutions in ESR1 was significantly higher in mCILC, and this difference was also significantly higher in mCILC metastasis biopsies exposed to hormonal therapy than in pre-treatment primary tumors (P<0.0001). ERBB2 (HER2) GA (amp + non-amp) detected by CGP were higher in mPILC than mCILC in both pre-and post-treatment samples (P<0.0001 for both). The ERBB2 GA frequency was nearly twice as high after hormonal therapy in both mCILC and mPILC. ESR1 and ERBB2 GA were mutually exclusive overall and especially in the mCILC group. PIK3CA GA were the most frequent GA in both mCILC and mPILC. TP53 GA were significantly more frequent in mPILC than mCILC. At 19%, the frequency of TMB > 15 mutations/MB in mPILC was more than twice as frequent than in mCILC (P=0.046). All (100%) of both the CILC and PILC groups were negative for mis-match repair deficiency or MSI high status. mCILC and mPILC patients with post primary therapy associated ESR1 and ERBB2 GA responding to targeted and immunotherapies will be presented.

Contrasting Clinical and Genomic Features of CILC and PILC Classic CILC (428 cases)Pleomorphic PILC (26 cases)Median Age6363*ER+98%74%*HER2 IHC/FISH+12 (3%)6 (22%)ESR1 GA Primary Pre-Rx6%0%ESR1 GA Metastatic Post-Rx17%0%ERBB2 GA Primary Pre-Rx7%18%ERBB2 GA Metastatic Post-Rx12%34%Other Significant GAPIK3CA (55%), CCND1 (21%), TP53 (17%), ARID1A, AKT3, MDM4, PTEN (all 11%)PIK3CA (58%), TP53 (30%), AKT1 22%), FGFR4, CCND1, PTEN (all 17%)TMB median (mut/Mb)2.73.6TMB > 15%8%19%*when clinical status available

Conclusions: CGP of mCILC and mPILC reveals significant differences in the panorama of GA both in pre-treatment primary and metastatic disease lesions especially in therapy-impacting GA in ESR1 and ERBB2. mCILC is more often driven by ESR1 GA and mPILC by ERBB2 GA. Although both mCILC and mPILC feature subsets of tumors with high TMB, this is more frequent for mPILC likely indicating different potentials for immunotherapies to benefit these patients.

Citation Format: Ross JS, Chung J, Elvin JE, Vergilio J-A, Ramkissoon S, Suh J, Severson E, Daniel S, Frampton GM, Fabrizio D, Hartmaier RJ, Albacker LA, Ali SM, Schrock AB, Miller VA, Stephens PJ, Gay LM. CDH1 mutated classic and pleomorphic invasive lobular breast carcinomas differ in genomic signatures and opportunities for targeted and immunotherapies [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 PD8-01.

Abstract P1-05-07: Comprehensive genomic profiling of clinically malignant phyllodes tumors of the breast reveals frequent mutation of NF1 and other genes associated with PI3K and RAS pathway activation

Background: Malignant or metastatic breast phyllodes tumors (MPT) are exceptionally rare, and the underlying genomic drivers are still being elucidated. Recent studies report frequent mutations in the RAS and PI3K pathways but have not commonly reported mutations in NF1. Comprehensive genomic profiling (CGP) can measure mutation load (TMB) and identifies all four classes of oncogenic alterations, including rearrangements and copy number loss that commonly affect tumor suppressors such as NF1, and can direct personalized treatment strategies.

Methods: CGP using hybridization capture of 3,769 exons from up to 315 cancer-related genes and select introns of 28 genes commonly rearranged in cancer was applied to ≥50ng of DNA extracted from 21 consecutive MPT and sequenced to high, uniform median coverage (>400X). TMB was determined as mutations/Mb on 1.1 Mb of sequenced DNA.

Results: The 21 MPT featured a median age of 51 yrs (range 14-70 yrs). CGP was performed on the primary MPT in 15 cases and on metastasis biopsies in 6 cases. TMB for all MPT was low (<10 mut/Mb), and all evaluable tumors (17/21) were microsatellite stable (MSS). The most commonly mutated genes were TP53 (57.1%), TERT (56.3%), NF1 (52.4%), MED12 (38.1%), CDKN2A/B (33.3%), and MLL2 (33.3%). 19/21 (90.5%) MPT harbored clinically relevant genomic alterations (CRGA) associated with therapies available on the market or under investigation in late stage clinical trials. Additional alterations in the PI3K/AKT/MTOR, RAS/RAF/MEK, and FGFR pathways were identified (see table); the PI3K/ATK/MTOR pathway was mutated in 10/21 (47.6%) of samples. Although CDKN2A/B loss was found in 6/11 tumors with NF1 mutation and only 1/10 NF1 wild-type samples, the co-occurrence was not significant (p<0.07). No significant correlation exists between the occurrence of NF1 mutations and mutation of MED12, TERT, the PI3K pathway, or other genes in the RAS/RAF pathway (NRAS, BRAF, EGFR). Targetable KIAA1549-BRAF or FGFR3-TACC3 fusions were identified in 2/21 (9.5%) tumors. Responses to targeted treatments will be presented.

Conclusions: More than 90% of MPT feature CRGA, including alteration of NF1, which was by far the most common targetable GA in this study. 52.4% of MPT had alterations predicted to result in loss of NF1 activity. NF1 mutation does not significantly co-occur with mutations in any other gene or pathway commonly altered in MPT. Other tumors with underlying NF1 mutations have responded to the MEK inhibitor selumetinib, suggesting MEK inhibitors may be relevant for the treatment of MPT. Other targetable alterations, including known gene fusions, are common in MPT. Thus, MPT may benefit from combination targeted therapy, warranting further investigation in the clinical trial setting.

  Total Mutation NumberPathwayTotal Cases (n=21)Short VariantsCopy NumberRearrangementsRAS/RAF/MEK    NF111 (52.4%)722BRAF3 (14.3%)301NRAS2 (9.5%)300PI3K/AKT/MTOR    PIK3CA4 (19%)310PTEN4 (19%)130STK112 (9.5%)110AKT11 (4.8%)100FBXW71 (4.8%)001TSC21 (4.8%)010PIK3R11 (4.8%)100FGFR    FGFR11 (4.8%)100FGFR31 (4.8%)001Other    EGFR2 (9.5%)020BRCA21 (4.8%)100PDGFRA1 (4.8%)010KIT1 (4.8%)010

Citation Format: Gay LM, Elvin JA, Vergilio J-A, Suh J, Ramkissoon S, Ali S, Schrock A, Hirshfield K, Ganesan S, Miller VA, Stephens PJ, Ross JS. Comprehensive genomic profiling of clinically malignant phyllodes tumors of the breast reveals frequent mutation of NF1 and other genes associated with PI3K and RAS pathway activation [abstract]. In: Proceedings of the 2016 San Antonio Breast Cancer Symposium; 2016 Dec 6-10; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2017;77(4 Suppl):Abstract nr P1-05-07.

Abstract P1-05-08: Comprehensive genomic profiling of 8,654 breast carcinoma reveals therapeutically targetable molecular subtypes beyond those defined by hormone-receptor expression

Background: Breast carcinomas (BC) are commonly classified into 4 subtypes based on hormone receptor expression: basal, luminal A, luminal B, and HER2 overexpressed. Comprehensive genomic profiling (CGP) reveals targetable genomic alterations (GA) across all four mutation classes, as well measuring tumor mutational burden (TMB), and can redefine BC classification into therapeutically relevant subtypes. Testing with immunohistochemistry or hotspot testing can miss a substantial number of targetable alterations and cannot measure TMB.

Methods: DNA was extracted from 40 µm of FFPE sections for 8654 consecutive BCs. CGP was performed on hybridization-captured, adaptor ligation-based libraries (mean coverage >500X) for up to 315 cancer-related genes and select introns from up to 28 genes frequently rearranged in cancer. Sequences were analyzed for substitutions, small insertions/deletions, copy number changes, and rearrangements. TMB was determined by counting non-driver, non-germline alterations across 1.1 Mbp of sequenced DNA. Clinically relevant GA (CRGA) are GA linked to therapies on the market or under evaluation in clinical trials. Immunotherapy (IO) sensitivity is defined as TMB >20 mut/Mbp or mutation of specific DNA repair pathways.

Results: The table below outlines 7 distinct functional or signal transduction pathways commonly altered in BC. Several are targetable with therapies that are FDA approved for an oncology indication. Mutations can also be found in other targetable kinases such as RET, ROS1, and RAF. 6959 (80.4%) tumors harbor a GA in at least one pathway, and 2697 (31.2%) BC harbor alterations in just one pathway (unique cases). Only 9.8% of BC would be HER2-positive by IHC. Almost 4% (352/8654) of cases harbor rearrangements or gene fusions that may not be detectable with other assays. Mutations in ESR1 characterize an eighth category of tumors with acquired resistance to endocrine therapy; 796/8654 (9%) samples harbor ESR1 alterations.

Conclusions: CGP can identify CRGA and TMB that can stratify tumors by predicted sensitivity to a variety of therapies, including HER2- or mTOR-targeted therapies, immunotherapies, and other kinase inhibitors. 80% of BC harbor targetable GA, and 30% of samples harbor mutations in only one pathway. CGP can provide crucial information for identifying which of several treatment modalities is most appropriate for these 30% of patients. High levels of TMB and most GA would not be identified by IHC or hotspot testing, but can be detected by next-generation sequencing. CGP is a powerful tool for guiding treatment across therapeutically distinct, but targetable, pathways.

 PI3K/AKT/mTOR pathwayFGFR pathwayCDK pathwayERBB pathwayHR deficientIO sensitiveOther kinasesTotal Cases43752650268512941266419424% Total Cases51%31%31%15%15%5%5%Unique Cases14422262312743094858% Unique Cases17%3%3%3%4%1%1%TherapiesEverolimus, TemsirolimusPazopanib, PonatinbPalbociclibTrastuzumab, Pertuzumab, Afatinib, Lapatinib, NeratinibOlaparibPembrolizumab, Nivolumab, Atezolizumab, IpilumumabSorafenib, Regorafenib, Dabrafenib, Vemurafenib, Crizotinib, Cabozantinib, Sunitinib

Citation Format: Ross JS, Gay LM, Elvin JA, Suh J, Vergilio J-A, Ramkissoon S, Schrock A, Ali S, Miller VA, Stephens PJ. Comprehensive genomic profiling of 8,654 breast carcinoma reveals therapeutically targetable molecular subtypes beyond those defined by hormone-receptor expression [abstract]. In: Proceedings of the 2016 San Antonio Breast Cancer Symposium; 2016 Dec 6-10; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2017;77(4 Suppl):Abstract nr P1-05-08.

Role of Tetra Amino Acid Motif Properties on the Function of Protease-Activatable Viral Vectors

Protease-activatable viruses (PAV) based on adeno-associated virus have previously been generated for gene delivery to pathological sites characterized by elevated extracellular proteases. "Peptide locks", composed of a tetra-aspartic acid motif flanked by protease cleavage sequences, were inserted into the virus capsid to inhibit virus-host cell receptor binding and transduction. In the presence of proteases, the peptide locks are cleaved off the capsid, restoring the virus' ability to bind cells and deliver cargo. Although promising, questions remained regarding how the peptide locks prevented cell binding. In particular, it was unclear if the tetra-amino acid (4AA) motif blocks receptor binding via electrostatic repulsion or steric obstruction. To explore this question, we generated a panel of PAVs with lock designs incorporating altered 4AA motifs, each wielding various chemical properties (negative, positive, uncharged polar, and hydrophobic) and characterized the resultant PAV candidates. Notably, all mutants display reduced receptor binding and decreased transduction effciency in the absence of proteases, suggesting simple electrostatics between heparin and the D4 motif do not play an exclusive role in obstructing virus-receptor binding. Even small hydrophobic (A4) and uncharged polar (SGGS) motifs confer a reduction in heparin binding compared to the wild type. Furthermore, both uncharged polar N4 and Q4 mutants (comparable in size to the D4 and E4 motifs respectively, but lacking the negative charge) demonstrate partial ablation of heparin binding. Collectively, these results support a possible dual mechanism of PAV lock operation, where steric hindrance and electrostatics make nonredundant contributions to the disruption of virus-receptor interactions. Finally, because of high virus titer production and superior capsid stability, only the negatively charged 4AA motifs remain viable design choices for PAV construction. Future studies probing the structure-function relationship of PAVs will further expand its promise as a gene delivery vector able to target diseased tissues exhibiting elevated extracellular proteases.

Biocomputing nanoplatforms as therapeutics and diagnostics

Biocomputing nanoplatforms are designed to detect and integrate single or multiple inputs under defined algorithms, such as Boolean logic gates, and generate functionally useful outputs, such as delivery of therapeutics or release of optically detectable signals. Using sensing modules composed of small molecules, polymers, nucleic acids, or proteins/peptides, nanoplatforms have been programmed to detect and process extrinsic stimuli, such as magnetic fields or light, or intrinsic stimuli, such as nucleic acids, enzymes, or pH. Stimulus detection can be transduced by the nanomaterial via three different mechanisms: system assembly, system disassembly, or system transformation. The increasingly sophisticated suite of biocomputing nanoplatforms may be invaluable for a multitude of applications, including medical diagnostics, biomedical imaging, environmental monitoring, and delivery of therapeutics to target cell populations.