P5-12-03 GENOME COPY NUMBER ENTROPY AS PREDICTOR OF RESPONSE FOR NEOADJUVANT THERAPY IN EARLY BREAST CANCER Emilio Alba1,2,15, Oscar M. Rueda3, Ana Lluch2,4,15, Joan Albanell2,5, Suet-Feung Chin3, Jose Ignacio Chacón López-Muñiz2,6, Lourdes Calvo2,7, Juan de la Haba-Rodriguez2,8,15, Begoña Bermejo2,4,15, Nuria Ribelles1, Pedro Sánchez Rovira2,9, Arrate Plazaola2,10, San Antonio Breast Agustí Barnadas2,11, Beatriz Cirauqui2,12, Manuel Ramos Vázquez2,13, Angels Arcusa2,14, Eva Carrasco2, Jesús Herranz2, Massimo Chiesa2, Rosalía Caballero2, Ángela Santonja1,3, Federico Rojo2,15,16, Carlos Caldas3 1Instituto de Investigación Biomédica de Málaga (IBIMA) - Hospital Clínico Universitario Virgen de la Victoria, Málaga, Spain, 2GEICAM Spanish Breast Cancer Group, San Sebastián de los Reyes, Madrid, Spain, 3Cancer Research UK Cambridge Institute, Li Ka Shing Centre, University of Cambridge, Cancer Symposium Robinson Way, Cambridge CB2 0RE, UK, 4Hospital Clínico Universitario de Valencia, Valencia, Spain, 5Hospital del Mar, Barcelona, Spain, 6Hospital Virgen de la Salud, Toledo, Spain, 7Complejo Hospitalario Universitario de A Coruña, A Coruña, Spain, 8Instituto Maimónides de Investigación Biomédica de Córdoba (IMIBIC)–H. Universitario Reina Sofía, Universidad de Córdoba, Córdoba, Spain, 9Complejo Hospitalario de Jaén, Jaén, Spain, 10Onkologikoa, San Sebastián, Spain, 11Hospital de la Santa Creu i Sant Pau, Barcelona, Spain, 12Hospital Germans Trias i Pujol, Barcelona, Spain, 13Centro December 4-8, 2018 Oncológico de Galicia, A Coruña, Spain, 14Consorci Sanitari de Terrassa, Barcelona, Spain, 15Centro de Investigación Biomédica en Red de Oncología, CIBERONC-ISCIII, Spain, 16Fundación Jiménez Díaz, Madrid, Spain

INTRODUCTION A total of 204 tumors (100 pre- and 104 post-treatment, 78 paired) from 165 patients from the GEICAM/2006 - 03 (113 RESULTS patients (68%); 81 treated with CT and 32 with HT) and GEICAM/2006-14 trials (52 patients (32%); 26 by ttm arm) were • Copy Number Alterations (CNAs) represent changes in copy number of genomic regions included in the CNAs analysis (Tab. 1). CNA 1 Tab.2 Comparative analysis of CNA (GAIN/LOSS) p-value that arise in somatic cells as direct consequence of chromosomal instability (1). 2 Regions VARIABLE N (%) the CNA-regions revealed 11 (Start-End) pre- vs post-ttm tumor % adj. BH 3 # Tumor size 4 regions with CNAs (loss or • The presence of CNAs may affect the function of onco- and tumor-suppressor genes and 5 115 3 (187400001-189600001) GAIN (19% vs 43%) 0.0364 cT1+cT2 123 (75%) 6 gain) with a statistically therefore can be involved in tumor development and evolution (2) 7 >cT3 42 (25%) significant different distribution 172 6 (16300001-17900001) GAIN (8% vs 25%) 0.0431 8 Lymph nodes 9 between the 100 pre- and 104 175 6 (33400001-33500001) GAIN (5% vs 17%) 0.0466 • At the same time, the genomic imbalance due to CNAs can play a role in the response to pN0 94 (57%) 10 176 6 (33600001-34100001) GAIN (5% vs 18%) 0.0255 11 post-treatment tumors (p-value anti-cancer therapy and ultimately determine their efficacy (3) pN1-2-3 68 (41%) 12 177 6 (34200001-40900001) GAIN (5% vs 20%) 0.0149 13 adj. BH <0.05; p-value adjusted Unknown 3 (2%) 14 178 6 (41000001-41100001) GAIN (6% vs 18%) 0.0449

Chromosome # Chromosome for multiple comparison Tumor grade 15 16 according to Benjamini- 245 8 (400001-6100001) LOSS (20% vs 50%) 0.0033 OBJECTIVE G1+G2 91 (55%) 17 18 Hochberg). 356 10 (129900001-135300001) LOSS (8% vs 29%) 0.0214 G3 66 (40%) 19 418 11 (130800001-134700001) LOSS (20% vs 44%) 0.0253 We aimed to identify tumoral genomic specific alterations (CNAs) that may predict response to GX 8 (5%) 20 neoadjuvant treatment in the GEICAM/2006-03 and GEICAM/2006-14 clinical trials 21 454 12 (125600001-131900001) LOSS (11% vs 46%) 0 Ki67 50% 22 X 455 12 (132200001-133700001) LOSS (9% vs 30%) 0.0045 Low (<50) 90 (55%)

High (≥50) 61 (37%) 0.0e+00 5.0e+07 1.0e+08 1.5e+08 2.0e+08 2.5e+08 MATERIALS AND METHODS Subtype Base # Luminal 64 (39%) Fig.1 We identified 672 CNA-regions with the “CGH-Regions” tool covering the 1 2 3 4 1 2 3 4 Tab.3 and Fig.5 Association with pCR of of 21q22.12 locus in the • Patients and tumor samples: GEICAM/2006-03 (NCT00432172) HER2-negative patients Her2+ 52 (31%) whole genome upon analysis of CNAs data. Regions are displayed as a bi-colored Hatzis neoadjuvant dataset (GSE25066) (9). Genes whose expression were significantly were selectively treated according to clinical subtypes: triple negative (TN) patients were Triple Negative 49 (30%) segments, the lower and upper part of which correspond to the proportions of increased in the cohort of patients that reached pCR (p-value adjusted BH <0.05 and t- treated with standard taxane/anthracycline-based chemotherapy (TA-CT) +/- carboplatin, Tab.1 Main clinical-pathological samples with a loss (red) or gain (green), respectively. Regions were then categorized test<0) are reported in red. These data indirectly point to a putative role in neoadjuvant while luminal patients were randomized to TA-CT vs. hormonotherapy (HT). GEICAM/2006- patients’ characteristics according to their alteration status as gain, normal and loss. treatment response of specific genes localized in the 21q22.12 locus. A brief description 14 (NCT00841828) HER2+ patients received TA-CT plus anti-HER2 therapy (Trastuzumab of published evidences supporting the relation of these genes with BC is reported, or Lapatinib). Pathological complete response (pCR) in breast and axilla was used to together with a scheme showing their localization in the locus. measure treatment (ttm) response, according to Miller & Payne criteria. Available pre- and Genomic location CNA Region Tumors Tumors (Chr: start-end) # p-val pCR=YES pCR=NO post-ttm tumors were analyzed from patients included in both clinical trials. (%) (%) pCR locus 11q12 association Gene Evidences of association with BC p-val • CNAs identification: Tumoral DNA extracted from FFPE samples was used to generate t-test adj BH sequencing libraries. Single-read 50bp whole-genome sequencing (WGS) was performed on Chromatin Assembly Factor 1 subunit a HiSeq4000 (Illumina). Data were pre-processed and normalized with “QDNAseq“ package locus 16q22 CHAF1B 4.7e-08 -6,47 B: associated with proliferation and bad (4). Segmentation was implemented with “DNA Copy” software (5) while “CGH call” (6) was prognosis (10, 11, 12) used for copy number call. Recurrent altered genomic regions were defined as CNAs CBR3 2.3e-05 -5,07 Carbonyl reductases (1 & 3): lower activity are associated with increased CT adopting 3 categories: Loss (single/double), Normal and Gain (gain/amplification). CBR1 0.0006 -4,21 efficacy (13, 14) Regulator of Calcineurin 1: growth and • CNAs Regions: CNAs data were independently analyzed with “CGH regions” metastasis suppressor that functions in RCAN1 0.0079 -3,36 part through NFE2L3 , (Bioconductor) and “GISTIC2.0” tools (7, 8) in order to identify specific genomic regions that locus 21q22 overexpressed in BC (15,16) show significant copy number alterations amongst samples. Wilcoxon test was used to Solute Carrier Family 5 Member 3: analyze the frequency of altered regions between pre- and post-treatment populations and Prevents intracellular accumulation of SLC5A3 0.0286 -2,87 logistic regression analyses were used to explore their association with treatment response myo-inositol – no described connection with cancer (pCR). Genes localized in significant genomic regions were identified according to the “hg19” RUNX1 transcription factor: tumor CGH alteration: genome assembly. RUNX1 0.0008 4,11 suppressor role in ER+ lum BC and Fig.4 Analysis of CNAs data with “GISTIC 2.0” tool identifies 68 amplified oncogenic role in TNBC (17) • Validation in the Hatzis dataset (GSE25066): gene expression of CNA-regions associated Fig.3 We identified 8 CNA-regions (“CGH regions” tool) whose alterations were significantly associated with treatment response (pCR): gain of regions #374, 653, 654, (red peaks) and 50 deleted loci (blue peaks). The amplified locus 21q22.12 with pCR was analyzed in Hatzis microarray-based dataset including HER2-neg breast Other genes found in this locus not associated with pCR are: SETD4, ATP5O, DOPEY2, ITSN1, (marked with “ “) corresponds to the amplified CNA-regions #653 and 654 cancer (BC) patients treated with neoadjuvant AT-CT (9). 655 or 656 were associated with pCR while loss of regions #533, 534 or 535 were RUNX1-IT1, MORC3, KCNE1, KCNE2, LINC00160, MRPS6, LINC00310, MIR802, associated with no-pCR (p-value adj. BH<0.05) previously shown to be associated with pCR. LOC100133286, CBR3-AS1

CONCLUSIONS REFERENCES ACKNOWLEDGMENTS This presentation is the 1. Tang et al., Cell. 2013; 152(3): 394–405. 10. Polo et al. Cancer Res. 2004, 64(7): 2371-81 intellectual property of the 2. Negrini et al., Nat Rev Mol Cell Biol. 2010;11(3):220-8 11. Polo et al., Histopathology. 2010; 57(5):716-24. • Genomic aberrations landscape is modulated by neoadjuvant therapy in 3. Vargas-Rondón et al., Cancers, 2017; 10(1):4 12. Montes de Oca et al., Mol Oncol, 9(3): 657-74 We thank all investigators and patients author/presenter. Contact at: Breast Cancer. 4. Scheinin et al., Genome Res. 2014; 24: 2022–2032. 13. Jo et al., Antioxid Redox Signal, 2017 26(2):70-83. participating in the GEICAM/2006-03 5. Seshan et al., 2018 R package version 1.54.0 14. Fan et al., Pharmacogenet Genomics, 18(7): 621-31 secretaria- and GEICAM/2006-14 studies. [email protected] • 6. van de Wiel et al., Bioinformatics. 2007; 23(7):892-4 15. Wang et al., JCI Insight; 2017, 2(5): e90651 QR 21q22.12 gene locus amplification might be associated with neoadjuvant 7. van de Wiel SV&M (2009). R package version 1.38.0 16. Rhee et al., Genomics, 2008; 92(6): 419-28 This work was partially supported by for permission to reprint and/or distribute. therapy response in Breast Cancer. 8. Mermel CH et al., Genome Biol. 2011;12(4):R41 17. Mercado-Matos et al., Oncotarget. 2017; 8:36934-5 Ferrer inCode. 9. Hatzis et al., JAMA 2011, 305(18) 1873-81