- Email: [email protected]
Targeting EGFR exon 20 insertions in non-small cell lung cancer by exploiting a dependency on parallel SRC signalling
abstracts
Annals of Oncology 7O
Contrasting the drivers of response to immunotherapy across solid tumour types: Results from analysis of > 1000 cases
K. Litchfield1, C. Swanton2, S. Turajlic3, N. McGranahan4, S. Quezada5 1 Swanton Lab, The Francis Crick Institute, London, UK, 2Translational Cancer Therapeutics, The Francis Crick Institute, London, UK, 3Royal Marsden Hospital NHS Foundation Trust, London, UK, 4Cancer Research UK, London, UK, 5Cancer Institute, University College London, London, UK
8O
Targeting EGFR exon 20 insertions in non-small cell lung cancer by exploiting a dependency on parallel SRC signalling
S. Vyse1, N. Chen2, J. Oddy1, P. Harrison1, A. Le2, A. Estrada-Bernal2, M. Hermsen3, R. Doebele2, P. Huang1 1 Molecular Pathology Department, The Institute of Cancer Research (ICR), London, UK, 2 University of Colorado School of Medicine, Aurora, CO, USA, 3Hospital Universitario Central de Asturias, Oviedo, Spain Background: Inframe insertions in exon 20 of the epidermal growth factor receptor (EGFR) gene are oncogenic drivers in non-small cell lung cancer (NSCLC). However, unlike more common EGFR mutations, low frequency EGFR exon 20 insertions are resistant to clinically approved EGFR inhibitors and are associated with poor patient prognosis. There is an unmet clinical need to identify novel therapies that will be effective for patients with EGFR exon 20 insertions. Methods: Cell viability of three patient-derived NSCLC cell line models harbouring EGFR exon 20 insertions, CUTO14 (A767_V769dupASV), CUTO17 (N771_H773dup NPH) and CUTO18 (S768_D770dupSVD) was assessed after treatment with a chemical compound screen targeting cancer-associated pathways. Western blotting was used to probe cell signalling following dasatinib treatment. Mutant SRC expression constructs were introduced into cell lines via lentiviral transduction and depletion of endogenous SRC levels was achieved using RNAi. Models of acquired drug resistance were developed by long-term drug exposure. Results: Out of 58 screened compounds, the tyrosine kinase inhibitor dasatinib was a shared hit across all EGFR exon 20 insertion models and was superior to the EGFR inhibitor gefitinib in cell viability and apoptosis assays. Dasatinib sensitivity was rescued by expression of a gatekeeper mutant SRC that does not bind dasatinib, whilst all models were sensitive to RNAi-mediated depletion of endogenous SRC expression. Importantly, models of acquired dasatinib resistance maintain sensitivity to poziotinib, an EGFR inhibitor that can target EGFR exon 20 insertions. Similarly, dasatinib effectively inhibits cell viability of a model of acquired resistance to poziotinib, indicating two independent, targetable signalling nodes. Combined treatment with dasatinib and poziotinib prevented the emergence of drug resistance. Conclusions: EGFR exon 20 insertion lung cancer models are sensitive to inhibition of parallel SRC signalling via dasatinib treatment. In a patient population with limited therapeutic options, an upfront rational combination of dasatinib with EGFR inhibitors that target EGFR exon 20 insertions has the potential to achieve durable responses. Legal entity responsible for the study: Paul Huang.
Volume 30 | Supplement 7 | November 2019
Expenses, Licensing / Royalties: Genentech/Roche; Advisory / Consultancy, Travel / Accommodation / Expenses, Licensing / Royalties: Ignyta; Licensing / Royalties: Foundation Medicine; Advisory / Consultancy, Licensing / Royalties: Loxo Oncology; Advisory / Consultancy: Bayer; Advisory / Consultancy, Shareholder / Stockholder / Stock options, Licensing / Royalties: Rain Therapeutics; Honoraria (self), Advisory / Consultancy: Takeda/Millenium; Licensing / Royalties: Abbot Molecular. All other authors have declared no conflicts of interest.
9P
Plasma gene conversions after one cycle (C) abiraterone acetate (AA) for metastatic castration-resistant prostate cancer (mCRPC): A biomarker analysis of a multi-centre, international trial
A.K. Jayaram1, D. Shen2, A. Wingate1, D. Wetterskog1, C. Sternberg3, R. Jones4, A. Berruti5, F. Lefresne6, M. Lahaye6, S. Thomas2, S. Joshi7, M. Gormley2, B. Tombal8, A. Merseburger9, D. Ricci2, G. Attard1 1 Treatment Resistance Department of Oncology, UCL Cancer Institute, London, UK, 2 Janssen Research and Development, Spring House, PA, USA, 3Englander Institute for Precision Medicine, Weill Cornell Medicine, New York, NY, USA, 4The Beatson West Scotland Cancer Centre, University of Glasgow, Glasgow, UK, 5Department of Medical and Surgical Specialties, Radiological Sciences and Public Health, University of Brescia, Spedali Civili Hospital, Brescia, Italy, 6Janssen Research and Development, Beerse, Belgium, 7HireGenics, Duluth, GA, USA, 8Institut de Recherche Clinique, Universite´ Catholique de Louvain, Brussels, Belgium, 9Department Urology, University Hospital Schleswig-Holstein, Campus Lu¨beck, Lubeck, Germany Background: A number of genomic alterations detected in plasma DNA have been associated with worse outcome in mCRPC (Jayaram et al Cancer Discov). We hypothesized that patients (pts) who harbored a genomic alteration that decreased after 1C of treatment derive treatment benefit and this would distinguish them from truly resistant pts. Methods: Plasma DNA (128 C1 day (D) 1,134 C2 D1, and 41 progression [PD] from chemotherapy-naı¨ve mCRPC pts in a phase 2 study of AA (NCT01867710), recently reported (Attard et al, Jama Onc) were subjected to custom targeted-capture NGS. Assay was optimised and validated to detect pathogenic point mutations (PM), deletions and copy number alterations (CNA) inAR, TP53, RB1, PIK3CA and DNA repair deficient genes (DRD): BRCA1, BRCA2, FANCA, ATM, CHEK2, HDAC2, BRIP1, and PALB2. Pts were followed up for overall survival (OS) and radiographic progressionfree survival (rPFS) (48 months). Results: Pts were classified into 4 groups based on whether a gene alteration was detectable (þ) or not (-) at C1D1 and C2D1 respectively. At C1D1, 49 pts (37.5%) had þ alterations. Pts who converted from þ to - (þ/-) had similar outcomes as pts who remained - (-/-) and those that did not convert had worst outcomes (þ/þ) (Table). In matched C1D1 and PD samples pts with ARgain (G) at C1D1 were more ARG at PD (p ¼ <0.01) while ARPM were only detected in PD samples that were ARnormal (N) at C1D1. DRD þ pts at C1D1 were more likely DRD þ at PD (p ¼ <0.1).
Table: 9P Gene conversion (N)
C1D1 þ vs TP53 -/- (85) vs þ/- (16) þ/ þ (8) vs þ/AR -/-(88) vs þ/-(16) þ/ þ (7) vs þ/RB1 -/-(92) vs þ/- (12) þ/þ6) vs þ/DRD -/-(96) vs þ/-(10) þ/ þ (4) vs þ/-
<0.01 2.0 1.23 3.30 <0.01 2.5 1.51 4.27
60.7
0.23 0.07
69.6
0.46 1.3 0.61 2.7 0.09 <0.01 3.3 0.9 11.4 0.07
66.67
0.79 1.1 0.5 2.4 0.37 1.4 0.60 3.27 <0.01 5.7 1.11 29.1 <0.01 5.6 0.92 34.26
73.3
0.85 0.03
0.9 0.44 2.0 1.0 1.0 0.43 2.36 4.4 0.29 69.1 <0.01 4.7 0.72 30.39
0.02 0.29
2.7 0.65 11.37 0.06 1.9 0.51 6.85 0.7
PIK3CA þ/ þ (100) vs þ/-(5) 50 þ/ þ (5) vs þ/-
1.4 0.74 2.77 0.7 1.2 0.55 2.51 2.2 0.73 6.7 <0.01 4.2 1.10 16.44 1.8 0.79 3.91 2.5 0.65 9.84
2.6 0.54 12.33 1.3 0.32 5.2
Conclusions: These findings suggest that tracking gene aberrations in plasma DNA could be an early marker of treatment efficacy. Clinical trial identification: NCT01867710.
doi:10.1093/annonc/mdz413 | vii3
Downloaded from https://academic.oup.com/annonc/article-abstract/30/Supplement_7/mdz413.013/5613297 by guest on 23 December 2019
Background: Multiple genomic and transciptomic biomarkers have been associated with response to immune checkpoint inhibitor (CPI) therapy. Emerging evidence suggests that each solid tumour type has a unique mix of factors determining CPI response, reflecting the subtle differences in antigen repertoire and immune microenvironment across histologies. Compiling large-scale sequencing datasets of patients treated with CPI therapy, from a range of solid tumour types, allows detailed comparison of the contrasting immune drivers per histology. Understanding these differences enhances our understanding of the pathways influencing CPI response, which may be of utility for therapeutic and biomarker development. Methods: We compiled data from 13 CPI treated cohorts, across 6 solid tumour types, encompassing 1,453 patients (n ¼ 1,453 with exome data, n ¼ 674 with RNAseq data). All raw data was accessed, and reprocessed through a standardised state of the art bioinformatics pipeline. A comphrehensive range of genomic & transcriptomic biomarker metrics were derived across the cohort. A combined predictive model was constructred encompassing all biomarkers, & the importance weighting was calculated for each biomarker, in each tumour type. Results: TMB was found to be a universal predictor of response across all tumour types, except for renal cell carcinoma (RCC). Instead CPI response in RCC appears to be strongly driven by expression of human endogeneuos retroviruses (hERV). In malignant melanoma, while TMB (nsSNVs) was associated with CPI response, the number of expressed indel mutations was found to be a stronger predictor. Shared antigen expression also demonstrated tumour specific predictive patterns. A signature of high immune inflitatation was found to be another universal predictor of response across multiple tumour types, however differences in the varying importance of immune cell subsets across histologies was observed. The rate of HLA LOH, and other immune evasion mechanisms also varied dramatically by cancer type. Conclusions: The determinants of immunotherapy response vary across solid tumour types, offering unique insight into both tumour intrinsic and extrinsic drivers of immunogenicity. Legal entity responsible for the study: The Francis Crick Institute. Funding: Has not received any funding. Disclosure: All authors have declared no conflicts of interest.
Funding: Institute of Cancer Research, Cancer Research UK. Disclosure: R. Doebele: Honoraria (self), Advisory / Consultancy, Travel / Accommodation /
Annals of Oncology 7O
Contrasting the drivers of response to immunotherapy across solid tumour types: Results from analysis of > 1000 cases
K. Litchfield1, C. Swanton2, S. Turajlic3, N. McGranahan4, S. Quezada5 1 Swanton Lab, The Francis Crick Institute, London, UK, 2Translational Cancer Therapeutics, The Francis Crick Institute, London, UK, 3Royal Marsden Hospital NHS Foundation Trust, London, UK, 4Cancer Research UK, London, UK, 5Cancer Institute, University College London, London, UK
8O
Targeting EGFR exon 20 insertions in non-small cell lung cancer by exploiting a dependency on parallel SRC signalling
S. Vyse1, N. Chen2, J. Oddy1, P. Harrison1, A. Le2, A. Estrada-Bernal2, M. Hermsen3, R. Doebele2, P. Huang1 1 Molecular Pathology Department, The Institute of Cancer Research (ICR), London, UK, 2 University of Colorado School of Medicine, Aurora, CO, USA, 3Hospital Universitario Central de Asturias, Oviedo, Spain Background: Inframe insertions in exon 20 of the epidermal growth factor receptor (EGFR) gene are oncogenic drivers in non-small cell lung cancer (NSCLC). However, unlike more common EGFR mutations, low frequency EGFR exon 20 insertions are resistant to clinically approved EGFR inhibitors and are associated with poor patient prognosis. There is an unmet clinical need to identify novel therapies that will be effective for patients with EGFR exon 20 insertions. Methods: Cell viability of three patient-derived NSCLC cell line models harbouring EGFR exon 20 insertions, CUTO14 (A767_V769dupASV), CUTO17 (N771_H773dup NPH) and CUTO18 (S768_D770dupSVD) was assessed after treatment with a chemical compound screen targeting cancer-associated pathways. Western blotting was used to probe cell signalling following dasatinib treatment. Mutant SRC expression constructs were introduced into cell lines via lentiviral transduction and depletion of endogenous SRC levels was achieved using RNAi. Models of acquired drug resistance were developed by long-term drug exposure. Results: Out of 58 screened compounds, the tyrosine kinase inhibitor dasatinib was a shared hit across all EGFR exon 20 insertion models and was superior to the EGFR inhibitor gefitinib in cell viability and apoptosis assays. Dasatinib sensitivity was rescued by expression of a gatekeeper mutant SRC that does not bind dasatinib, whilst all models were sensitive to RNAi-mediated depletion of endogenous SRC expression. Importantly, models of acquired dasatinib resistance maintain sensitivity to poziotinib, an EGFR inhibitor that can target EGFR exon 20 insertions. Similarly, dasatinib effectively inhibits cell viability of a model of acquired resistance to poziotinib, indicating two independent, targetable signalling nodes. Combined treatment with dasatinib and poziotinib prevented the emergence of drug resistance. Conclusions: EGFR exon 20 insertion lung cancer models are sensitive to inhibition of parallel SRC signalling via dasatinib treatment. In a patient population with limited therapeutic options, an upfront rational combination of dasatinib with EGFR inhibitors that target EGFR exon 20 insertions has the potential to achieve durable responses. Legal entity responsible for the study: Paul Huang.
Volume 30 | Supplement 7 | November 2019
Expenses, Licensing / Royalties: Genentech/Roche; Advisory / Consultancy, Travel / Accommodation / Expenses, Licensing / Royalties: Ignyta; Licensing / Royalties: Foundation Medicine; Advisory / Consultancy, Licensing / Royalties: Loxo Oncology; Advisory / Consultancy: Bayer; Advisory / Consultancy, Shareholder / Stockholder / Stock options, Licensing / Royalties: Rain Therapeutics; Honoraria (self), Advisory / Consultancy: Takeda/Millenium; Licensing / Royalties: Abbot Molecular. All other authors have declared no conflicts of interest.
9P
Plasma gene conversions after one cycle (C) abiraterone acetate (AA) for metastatic castration-resistant prostate cancer (mCRPC): A biomarker analysis of a multi-centre, international trial
A.K. Jayaram1, D. Shen2, A. Wingate1, D. Wetterskog1, C. Sternberg3, R. Jones4, A. Berruti5, F. Lefresne6, M. Lahaye6, S. Thomas2, S. Joshi7, M. Gormley2, B. Tombal8, A. Merseburger9, D. Ricci2, G. Attard1 1 Treatment Resistance Department of Oncology, UCL Cancer Institute, London, UK, 2 Janssen Research and Development, Spring House, PA, USA, 3Englander Institute for Precision Medicine, Weill Cornell Medicine, New York, NY, USA, 4The Beatson West Scotland Cancer Centre, University of Glasgow, Glasgow, UK, 5Department of Medical and Surgical Specialties, Radiological Sciences and Public Health, University of Brescia, Spedali Civili Hospital, Brescia, Italy, 6Janssen Research and Development, Beerse, Belgium, 7HireGenics, Duluth, GA, USA, 8Institut de Recherche Clinique, Universite´ Catholique de Louvain, Brussels, Belgium, 9Department Urology, University Hospital Schleswig-Holstein, Campus Lu¨beck, Lubeck, Germany Background: A number of genomic alterations detected in plasma DNA have been associated with worse outcome in mCRPC (Jayaram et al Cancer Discov). We hypothesized that patients (pts) who harbored a genomic alteration that decreased after 1C of treatment derive treatment benefit and this would distinguish them from truly resistant pts. Methods: Plasma DNA (128 C1 day (D) 1,134 C2 D1, and 41 progression [PD] from chemotherapy-naı¨ve mCRPC pts in a phase 2 study of AA (NCT01867710), recently reported (Attard et al, Jama Onc) were subjected to custom targeted-capture NGS. Assay was optimised and validated to detect pathogenic point mutations (PM), deletions and copy number alterations (CNA) inAR, TP53, RB1, PIK3CA and DNA repair deficient genes (DRD): BRCA1, BRCA2, FANCA, ATM, CHEK2, HDAC2, BRIP1, and PALB2. Pts were followed up for overall survival (OS) and radiographic progressionfree survival (rPFS) (48 months). Results: Pts were classified into 4 groups based on whether a gene alteration was detectable (þ) or not (-) at C1D1 and C2D1 respectively. At C1D1, 49 pts (37.5%) had þ alterations. Pts who converted from þ to - (þ/-) had similar outcomes as pts who remained - (-/-) and those that did not convert had worst outcomes (þ/þ) (Table). In matched C1D1 and PD samples pts with ARgain (G) at C1D1 were more ARG at PD (p ¼ <0.01) while ARPM were only detected in PD samples that were ARnormal (N) at C1D1. DRD þ pts at C1D1 were more likely DRD þ at PD (p ¼ <0.1).
Table: 9P Gene conversion (N)
C1D1 þ vs TP53 -/- (85) vs þ/- (16) þ/ þ (8) vs þ/AR -/-(88) vs þ/-(16) þ/ þ (7) vs þ/RB1 -/-(92) vs þ/- (12) þ/þ6) vs þ/DRD -/-(96) vs þ/-(10) þ/ þ (4) vs þ/-
<0.01 2.0 1.23 3.30 <0.01 2.5 1.51 4.27
60.7
0.23 0.07
69.6
0.46 1.3 0.61 2.7 0.09 <0.01 3.3 0.9 11.4 0.07
66.67
0.79 1.1 0.5 2.4 0.37 1.4 0.60 3.27 <0.01 5.7 1.11 29.1 <0.01 5.6 0.92 34.26
73.3
0.85 0.03
0.9 0.44 2.0 1.0 1.0 0.43 2.36 4.4 0.29 69.1 <0.01 4.7 0.72 30.39
0.02 0.29
2.7 0.65 11.37 0.06 1.9 0.51 6.85 0.7
PIK3CA þ/ þ (100) vs þ/-(5) 50 þ/ þ (5) vs þ/-
1.4 0.74 2.77 0.7 1.2 0.55 2.51 2.2 0.73 6.7 <0.01 4.2 1.10 16.44 1.8 0.79 3.91 2.5 0.65 9.84
2.6 0.54 12.33 1.3 0.32 5.2
Conclusions: These findings suggest that tracking gene aberrations in plasma DNA could be an early marker of treatment efficacy. Clinical trial identification: NCT01867710.
doi:10.1093/annonc/mdz413 | vii3
Downloaded from https://academic.oup.com/annonc/article-abstract/30/Supplement_7/mdz413.013/5613297 by guest on 23 December 2019
Background: Multiple genomic and transciptomic biomarkers have been associated with response to immune checkpoint inhibitor (CPI) therapy. Emerging evidence suggests that each solid tumour type has a unique mix of factors determining CPI response, reflecting the subtle differences in antigen repertoire and immune microenvironment across histologies. Compiling large-scale sequencing datasets of patients treated with CPI therapy, from a range of solid tumour types, allows detailed comparison of the contrasting immune drivers per histology. Understanding these differences enhances our understanding of the pathways influencing CPI response, which may be of utility for therapeutic and biomarker development. Methods: We compiled data from 13 CPI treated cohorts, across 6 solid tumour types, encompassing 1,453 patients (n ¼ 1,453 with exome data, n ¼ 674 with RNAseq data). All raw data was accessed, and reprocessed through a standardised state of the art bioinformatics pipeline. A comphrehensive range of genomic & transcriptomic biomarker metrics were derived across the cohort. A combined predictive model was constructred encompassing all biomarkers, & the importance weighting was calculated for each biomarker, in each tumour type. Results: TMB was found to be a universal predictor of response across all tumour types, except for renal cell carcinoma (RCC). Instead CPI response in RCC appears to be strongly driven by expression of human endogeneuos retroviruses (hERV). In malignant melanoma, while TMB (nsSNVs) was associated with CPI response, the number of expressed indel mutations was found to be a stronger predictor. Shared antigen expression also demonstrated tumour specific predictive patterns. A signature of high immune inflitatation was found to be another universal predictor of response across multiple tumour types, however differences in the varying importance of immune cell subsets across histologies was observed. The rate of HLA LOH, and other immune evasion mechanisms also varied dramatically by cancer type. Conclusions: The determinants of immunotherapy response vary across solid tumour types, offering unique insight into both tumour intrinsic and extrinsic drivers of immunogenicity. Legal entity responsible for the study: The Francis Crick Institute. Funding: Has not received any funding. Disclosure: All authors have declared no conflicts of interest.
Funding: Institute of Cancer Research, Cancer Research UK. Disclosure: R. Doebele: Honoraria (self), Advisory / Consultancy, Travel / Accommodation /