Update of BATTLE-2 Trial

Waun Ki Hong Vali Papadimitrakopoulou

UT MD Anderson Cancer Center BATTLE Concepts in 2004

Platform for integrated translational research 1.Personalized Targeted Therapy 2. Novel trial design 3. discovery

Overall Hypotheses 1.Molecular analysis of fresh biopsies can accurately reflect aberrant signaling pathway

2. Matching targeted agents with altered signaling pathways will improve disease control in patients BATTLE-1 Trial (2007- 9 ) Biopsy

Core Biopsy Biomarker EGFR Profile KRAS/BRAF VEGF

RXR/CyclinD1Randomization: Equal  Adaptive

Erlotinib Vandetanib + Bexarotene Sorafenib

Primary endpoint: 8-week disease control Skepticisms

Not feasible to do core biopsies for so many patients

Impossible to get biomarker data within 2 weeks

Impossible to get different drugs from four different companies

Highly controversial adaptive randomization as new concept

Will never complete trial BATTLE Manuscript: Lead Article in Inaugural Issue of Newest AACR Journal

Kim ES*, Herbst RS*, Wistuba II*, Lee JJ*, et al…Lippman SM#, Hong WK#. Cancer Discovery, 2011

*co-first authors #co-senior authors Editorials

The Battle trial: A bold step toward improving the efficiency of biomarker-based drug development

Clinical Trials Game –Changer ?

A New BATTLE in the Evolving War on Cancer

Time Has Come to Raise the Bar in Oncology Trials

Set a new standard for biopsy mandated personalized targeted therapy Impact of BATTLE-1 Trial

Galvanize Whole Field of Precision Medicine:

NCI Match trial, Lung Squamous Master Trial

IMPACT Trial, ATTACK Trial, Winther Trial

BATTLE-2 Trial

7 BATTLE-2 Trial (2012- )

Protocol enrollment Biopsy performed

EML4-ALK Stage 1: (n=200) Fusion or EGFR Μut Adaptive Randomization exclusion by KRAS mut status Rationale + Novelty: Statistical modeling and biomarker selection Rationally designed targeted therapy combinations More emphasis on KRAS targeting Predictive biomarker discovery plan Stage 2: (n=200) Refined Adaptive Randomization “Best” discovery markers/signatures

Erlotinib E+MK-2206 MK-2206+ AZD6244 Sorafenib (AKTi) (MEKi) Discovery Markers: Protein expression p-AKT (Ser473), PTEN, HIF-1a, LKB1 Primary endpoint: 8-week disease control (N = 400) Mutation analysis (Sequenom)/NGS-Foundation Medicine First stage 200 pts completed (6/11-3/14) mRNA pathways activation signatures: Affymetrix® RNA sequencing Update of BATTLE-2

Mutational Landscape

Clinical Outcome

Characteristics of 200 patients in BATTLE-2 (Chemorefractory metastatic NSCLC)

Smoking Status : 86 % Prior chemotherapy > 2 : 75% Prior Erotinib therapy : 37.5% KRAS mutation : 27 % ECOG PS ( 0- 2): 100% : 73% Squamous Carcinoma: 17% Others : 9 %

Identification of Genomic Alterations in NSCLC (Chemotherapy-naïve, resectable tumors)

TCGA Lung Adenocarcinoma TCGA Lung Squamous Cell Carcinoma (Nature, 2014) (Nature, 2012) TP53 CDKN2A PTEN

Frequency PIK3CA KEAP1 MLL2

HLA-A NFE2L2 NOTCH1 RB1 Mutation Matrix in BATTLE-2 patients (N=187) Molecular Alterations in 119 Genes

ATM LRP1B SPOP TET2 PAX5 MLH1 EGFR KDM6A PTEN JAK1 FGF4 GSK3B BRCA1 CTNNB1 SUFU ERBB2 MEN1 WT1 SMARCA4 BCOR PALB2 MSH6 FANCC SF3B1 TP53 STAT4 DNMT3A NCOR1 PIK3R1 SMAD4 ARID1A CSF1R MAP3K13 TBX3 ALK NOTCH3 KDM5C NFE2L2 VHL MLL ERBB3 ATR CDH1 ASXL1 CRBN KEAP1 NF2 BACH1 RAD51 NOTCH1 EP300 PARP4 STAG2 NRAS KRAS GRIN2A SMARCD1 PIK3C3 HRAS FANCA BRCA2 PBRM1 APC RET INHBA MSH2 STK11 RB1 TIPARP EPHA3 RAD52 WISP3 BRAF SETD2 MYCN CDC73 TOP1 PPP2R1A NOTCH4 BCL6 BCORL1 CTCF NOTCH2 CASP8 MET TRRAP RAD51C RAD50 CDK12 MAP2K1 CDKN2A ARID2 PIK3CA KLHL6 FGFR1 GATA1 KDR RUNX1 GATA3 BAP1 CEBPA PDGFRA NF1 MLL2 ROS1 MAP2K4 KDM5A FLT3 SPEN FBXW7 IDH1 CHUK NUP93 PIK3CG HGF BARD1 GNAS CREBBP MAP3K1

Papadimitrakopoulou Copy Number Matrix in BATTLE-2 patients (N=187) Copy-number-alterations in 82 genes

CCND1 CDK6 FGF10 PIK3R1 KIT EMSY JUN KRAS EPHA3 NFKBIA FGF19 MCL1 MYCL1 SMARCA4 NKX2-1 FGF3 MET CCNE1 FGF14 BCL2L2 FGF4 RICTOR HGF SMAD4 IKBKE CDKN2A ARFRP1 IRS2 JAK2 MDM4 CDKN2B TP53 PDGFRA FLT4 ERBB2 PTEN NRAS BRAF CASP8 AKT3 FGFR1 STK11 MYCN NF1 ATM LRP1B MYST3 CCND2 ERBB3 RAD51C PIK3CA ZNF703 FGF23 AXL SUFU SOX2 CRKL FGF6 ERBB4 FGF12 CDK4 MYC ARID2 CDKN2C GNAS EGFR REL RET KDM5C ARAF MDM2 NF2 RB1 CCND3 AR KDR AKT2 PRSS8 ZNF217 KDM6A FANCD2 BCOR

Papadimitrakopoulou Mutational Landscape in BATTLE-2 Patients

CNG/loss prevalence of 187 samples Mutation prevalence of 187 samples

ADE

SCQ TCGA (Chemo-naive) vs BATTLE-2 (Platinum-refractory) : Mutational Evolution

LUAD

* p<.05 LUSC

• Evolution- enrichment in commonly mutated genes (“Trunk” Genetic events) associated with refractory and metastatic NSCLC. • Similar trend for copy number changes (NFKBIA, MYC, MDM4, AKT2, CEBPA, PMS2) Co-occurring Genetic Events are Dominant Determinants of Cluster Membership

TCGA (n=68) Group Group KC KL KP

KRAS

TP53 LKB1 KRAS ATM

KEAP1 CDKN2A

Skoulidis F et al., Cancer Discovery, 2015 Validation in Metastatic, Platinum-Refractory KRAS- Mutant LUADs from the BATTLE-2 trial

BATTLE-2 (n=36) Group KC KL KP Group KRAS TP53 LKB1 ATM KEAP1 KRAS CDKN2A

Skoulidis F et al., Cancer Discovery, 2015 What are the major features of the different KRAS subgroups?

Are they biologically or therapeutically relevant? Higher somatic mutation burden in KP LUADs (TCGA cohort)

A. B. Reduced expression of PD-L1 in KRAS-mutant LKB1-deficient LUADs (MDACC cohort)

A. B.

Skoulidis F et al., 2015 ASCO Annual Meeting Reduced density of intra-tumoral T lymphocytes in KRAS-mutant LKB1-deficient LUACs (MDACC cohort)

Skoulidis F et al., 2015 ASCO Annual Meeting Summary of Mutational Landscape • Enrichment for known driver mutations in chemotherapy-refractory NSCLC • More TP53 mutations in BATTLE-2 patients • Copy number changes are similar both group • BATTLE-2 : Increased frequency of triple mutant KRAS;LKB1;TP53 and KRAS;TP53 • Increased somatic mutation burden in KRAS/TP53 • Reduced expression of PD-L1 and T-cell infiltration in KRAS/LKB1 • New biologically distinct KRAS co-mutational subsets. Update of BATTLE-2

Mutational Landscape

Clinical Outcome

Primary Endpoint

PR SD PD not evaluable total Response 7 83 97 13 187

8 week response Arm1 Arm2 Arm3 Arm4 Total PR 1(5.0%) 3(4.3%) 3(4.9%) 7(3.7%) By arms SD 6(30.0%) 18(50.0%) 34(48.6%) 25(41.0%) 83(44.4%) PD 13(65.0%) 18(50.0%) 33(47.1%) 33(54.1%) 97(51.9%) Non Evaluable 2 6 5 0 13

Yes No not evaluable total 8-wk DC 90(48.1%) 97 13 187

8 week disease control E(1) E+M (2) M+A (3) S Total By arms 8wk DC 7(35.0%) 18(50.0%) 37(52.9%) 28(45.9%) 90(48.1%) No 8wk DC 13(65.0%) 18(50.0%) 33(47.1%) 33(54.1%) 97(51.9%) p (Fisher’s exact .40 .20 .44 test) OS by treatment PFS by treatment

Median PFS: 2 months

P=0.4636 P=0.1652 PFS by KRAS mutation status

Arm 1 Arm 2 KRAS Mut

P=0.3288 P=0.0739

P=0.8919

Arm 3 Arm 4

P=0.3139 P=0.6041 PFS by KRAS mut PFS by KRAS wild

P=0.0422 P=0.1309

Genotype Arm 3 PFS <3 PFS >3 p • MEKi () and sorafenib: mos mos benefit for KRAS mut+ KRAS wt 16 7 • Trend for benefit with MEKi+AKTi for mut+ 8 12 0.069 KRAS mut+, enriched by co-mutations KRAS wt 13 2 KRAS mut+ and MEK/PI3K 7 7 0.05 mut+* *co-mutations examined: PIK3CA, CDKN2A, PTEN, FBXW7, BRAF, ERBB3, MEK1, MEK2, ARAF Erlotinib DCR 35% Erlotinib +MK-2206 DCR 50%

PTEN , TP53 , KRAS ampl, CCND2, FGF23 CDKN2A, TP53 , BRCA2, ampl,ARID1A, amp CCNE1 , AKT 2, ERBB3, KEAP1,, EGFR exon 19 PIK3CA, SOX2 EGFR , NOTCH2 del/TP53mut+ CEBPA, MAP3K13, BCL6 , SETD2 mut

Sorafenib MK-2206+AZD6244 DCR:46% DCR:53%

KRASG12C, RUNX1 mut, ERBB4 EGFR T790M,/L858R, TIPARP, TP53, CL2 , mut, ALK mut, ARAF mut, ATRX, MCL , MYC Ampl, NFKBIA, NKX2-1, AR, BTK, CDK4 AXL, CARD11, FGF6, MED12,TSC2

CDKN2A, FBXW7 , KDM5c , MLL2, TP53mut, CCNE1, AKT2 ,AXL, PIK3CA, SOX2, TP53, ARID1A, BRIP1, CDK8, FGF12, NOTCH2, KLHL6, MAPK3K13, BCL6 CREBBP, EP300 KRAS, STK11 mut, MYC, Ampl,, ATM,AR, SF1R, EPHA3, FANCM, FAT3, GNA13,ampl, ATM, HGF, SETD2,SMO. ATRX, KDR, NOTCH3, PIK3R2. 10/11/2011 12/21/2011 KRASG12C, RUNX1 Treatment: Arm 3 AKTi+MEKi mut+, ALK mut, ARAF KRAS mut in codon 12 (GGT to TGT) Gly to Cys (G12C). mut, ATRX, BTK, CDK4 Model of Multiplicity of Mutations within MAPK pathway

• Multiple alterations within MAPK pathway may drive growth and determine sensitivity . • Response in the setting of KRASG12C/ ARAF, CDK4 mutations • Responses to MEK inhibitors may be driven by a critical number of alterations within the MAPK pathway.

Activation of PI3K/AKT Pathway

• Multiple alterations leading to PI3K/AKT pathway activation including FBXW7, KDM5C associated with response to pan- AKT inhibitor. • FBXW7: is the substrate recognition component of a SCF-type E3 ubiquitin ligase. Targets Notch1, c-Jun, E and mTOR for degradation. Loss of FBXW7 may be a biomarker for human cancers susceptible to treatment with mTOR/AKT inhibitors.

Summary of Clinical Outcome

• 8-Week DCR : 48 %

• PFS by KRAS mut vs wild : No difference across the board

• Potential harmful effect of Erlotinib in KRAS mut

• Trend improvement of PFS in KRAS mut : MEK &AKT i and Sorafenib

• Potential benefit of MEKi+AKTi for KRAS mut with MEK /PI3K Co-mutation • Targeting MAPK pathway by MEK i and PI3K/AKT pathway pan-AKT inhibitor as combination needs further validation