Blood-based Biomarkers for Lung Cancer Tony Mok MD Li Shu Fan Medical Foundation Professor of Clinical Oncology The Chinese University of Hong Kong

How solid is liquid biopsy? =

Source of plasma DNA -tumor necrosis/apotosis -tumor exosome/microvesicles -normal tissue apotosis -inflammatory cell

Technologies for tissue samples are potentially applicable to plasma cf DNA McLarty et al MOJ Cell Science & Report 2015

Potential application Pathogenesis Symptoms treatment selection Resistance to treatment death Early detection Diagnosis and treatment selection Monitoring Mechanism of resistance and treatment

Potential application Pathogenesis Symptoms treatment selection Resistance to treatment death Early detection Diagnosis and treatment selection Monitoring Mechanism of resistance and treatment

We need to find EGFR mutation at time of diagnosis of adenocarcinoma Confirming the superior efficacy of EGFR TKI in patients with EGFR mutations Author Study N (EGFR mut +) RR Median PFS Mok et al IPASS 132 71.2% vs 47.3 9.8 vs 6.4 months Lee et al First-SIGNAL 27 84.6% vs 37.5% 8.4 vs 6.7 months Mitsudomi et al WJTOG 3405 86 62.1% vs 32.2% 9.2 vs 6.3 months Maemondo et al NEJGSG002 114 73.7% vs 30.7% 10.8 vs 5.4 months Zhou et al OPTIMAL 154 83% vs 36% 13.1 vs 4.6 months Rosell et al EURTAC 135 56% vs 18% 9.2 vs 4.8 months Wu et al LUX Lung 6 364 67% vs 23% 11.0 vs 5.6 months We need to find EGFR mutation at time of diagnosis of adenocarcinoma Mok et al NEJM 2009, Lee et al WCLC 2009, Mitsudomi et al Lancet Oncology 2010, Maemondo NEJM 2010 Zhou et al Lancet Oncol 2010

PROFILE 1014: Crizotinib Superior to Pemetrexed-based Chemotherapy in Prolonging PFSa PFS probability (%) 100 80 60 40 20 0 5 10 15 20 25 30 35 Time (months) 172 120 65 38 19 7 1 0 171 105 36 12 2 1 0 0 No. at risk Crizotinib Chemotherapy Crizotinib (N=172) Chemotherapy Events, n (%) 100 (58) 137 (80) Median, months 10.9 7.0 HR (95% CI) 0.45 (0.35−0.60) Pb <0.0001 We need to find ALK re-arrangement at time of diagnosis of adenocarcinoma Solomon & Mok et al NEJM 2015

Other Targetable Mutation Gene Alteration Histology Frequency Inhibitor BRAF Mutation, fusion ADC 1-3% Vemurafenib, dabrafenib, trametinib MET Amplification, and exon14 splicing 2-4% Tivantinib, cabozantinib, INC280, onartuzumab RET Fusion 1% Carbozantinib , sunitinib, sorafenib, lenvatinib, vandetanib HER2 Mutation Neratinib, afatinib, lapatinib, dacomitinib ROS1 <1% Crizoitinib KRAS mutation 15-25% SML-8-73-1, (Selumetinib, trametinib) FGFR1 amplification SCC 19% Lucitanib, Nintedanib, dovitinib, AZD4547 FGFR2-3 3% FGFR1-3 3.5% DDR2 4% Dasatinib Debatable on the need to identify them at time for first line treatment

EGFR mutation testing in Asia 2011 Country (N diagnosed with NSCLC*) Proportion tested for EGFR mutations % (95% CI†) Proportion of males/females, smokers and non-smokers, and histological subtypes that were tested for EGFR mutations‡ Gender Smoking status Histology Males / Females (%) Current + ex-smoker / Never smoker (%) ADC / All non-ADC / Only SCC (%) Total (22,193) 31.8 (31.2–32.5) 26.9 / 40.2 47.0 / 57.4 50.4 / 12.5 / 12.5 China (12,086§) 18.3 (17.6–19.0) 15.2 / 25.3 N.D. 30.3 / 8.0 / 9.4 Hong Kong (795) 42.0 (38.6–45.5) 36.2 / 52.3 34.1 / 52.1 55.4 / 9.0 / 6.4 Japan (2,379) 64.8 (62.9–66.7) 63.6 / 67.0 68.8 / 68.3 69.2 / 55.0 / 50.3 Korea (3,794) 33.5 (32.0–35.0) 26.1 / 38.1 27.1 / 42.9 62.7 / 9.8 / 8.3 Taiwan (2,890) 54.3 (52.5–56.1) 47.1 / 64.3 37.0 / 56.8 69.3 / 15.5 / 8.5 Thailand (249§) 57.8 (51.6–63.8) 51.6 / 69.3 49.5 / 84.7 83.6 / 7.1 / 6.9

cobas® EGFR _ blood Test _ Kit Components Utilizing most of the reagents in the cobas EGFR_FFPET test and requiring additional reagents and the blood-specific data analysis software Blood ctDNA Preparation Kit Cobas EGFR_Blood ctDNA SP cobas cell-free DNA Preparation Kit (To be used for other blood based assays) 2 ml Plasma cobas EGFR _ blood Test cobas 4800 v 2.0 cobas EGFR _ Blood HPEA x25 PK x2 PBB x6 Additional reagents added to cobas DNA preparation Kit Tube FAM HEX JA270 1 EX 19Del S768I 2 L858R T790M 3 G719X L861Q* EX 20Ins Blood-specific cutoffs; Blood-specific data analysis software *New primer and probe for L861Q Mok et al ASCO 2013

FASTACT-2 (MO22201; CTONG0902) study design Screening Study treatment Maintenance phase Gemcitabine 1,250mg/m2 (d1, 8) + carboplatin AUC=5 or cisplatin 75mg/m2 (d1) + erlotinib 150mg/day (d15–28); q4wks x 6 cycles GC-erlotinib (n=226) Erlotinib 150mg/day PD Previously untreated stage IIIB/IV NSCLC, PS 0/1 (n=451) 1:1; stratified by stage, histology, smoking status and chemo regimen R Gemcitabine 1,250mg/m2 (d1, 8) + carboplatin AUC=5 or cisplatin 75mg/m2 (d1) + placebo (d15–28); q4wks x 6 cycles GC-placebo (n=225) Placebo PD Erlotinib 150mg/day Primary endpoint: PFS with IRC confirmation Secondary endpoints: subgroup analyses, OS in all patients and subgroups, ORR, duration of response, TTP, NPR at 16 weeks, safety, QoL Wu and Mok Lancet Oncology 2013 12 12

EGFR Mutation Analysis using cobas 4800_blood test TISSUE SAMPLES 397 (88%) patients consented 301 (66.7%) samples available 241 (53.4%) samples analyzable 224 patients with matched tumor and plasma samples PLASMA SAMPLES 451 (100%) patients consented 427 (94.6%) samples available 427 (94.6%) samples analyzable

Concordance between tumor and plasma samples Total of 224 patients had both tumor and baseline plasma samples with available EGFR mutation analysis results (Table 3) Sensitivity: 77% (69/90) Specificity: 96% (129/134) Positive predictive value: 93% (69/74) Negative predictive value: 86% (129/150) Overall concordance: 88% (198/224) EGFR Activating Mutations p-EGFR Mut+ (Plasma) p-EGFR Mut- Total t-EGFR Mut+ (Tumor) 69 21 90 t-EGFR Mut- (Tumor) 5 129 134 74 150 224

PFS of p-EGFR and t-EGFR mut+ patients

PFS of p-EGFR and t-EGFR mut- patients

Therascreen in plasma/serum samples from LUX Lung 3 and 6

Droplet digital PCR (ddPCR) Hindson et al. Analytical chemistry 2011

Positive result on exon 19 deletion assay 20130508 Plasma sample 214 Mutant concentration:72 copies/ml plasma Fraction concentration : 3.2%

Negative result on exon 19 deletion assay 20130508

Tumor sample: COBAS EGFR Mutation Test Analyzing plasma and tumor sample from ASPIRATION study and matched control (n=197) Tumor sample: COBAS EGFR Mutation Test Plasma sample: Droplet digital PCR In formalin-fixed, paraffin-embedded tissue (FFPET) specimens, the cobas® EGFR test can detect <5% mutant sequence copies in a background of wild-type DNA.*  Lee et al WCLC 2013

Diagnostic utility of digital PCR for detection of EGFR mutation POS in plasma NEG in plasma POS in tumor 117 27 144 NEG in tumor 53 80 197 Droplet digital PCR Sensitivity 81% Specificity 100% Positive Predictive Value Concordance 86%

Meta-analysis on cf DNA for EGFR mutation 3110 subjects (27 studies) Tissue EGFR mutation status as gold standard Pooled Analysis Sensitivity 62% Specificity 96% Diagnostic odd ratio Area under summary ROC 0.91 Sensitivity is technology dependent, eg BEAM vs Sanger High specificity implies clinical applicability for selection of first line EGFR TKI Qiu et al Can Epi Biomarker Prev 2015 Jan;24(1):206-12.

Plasma DNA positive for EGFR exon 21 L858R Jan 7 at 5:00pm Plasma DNA positive for EGFR exon 21 L858R Jan 8 at 1:00pm CT scan guided lung biopsy Jan 8 at 7:00pm Start EGFR TKI Jan 6, 2015

Jan 8 2015 Jan 15, 2015

Potential application Pathogenesis Symptoms treatment selection Resistance to treatment death Early detection Diagnosis and treatment selection Monitoring Mechanism of resistance and treatment

Serial plasma samples at baseline, C3 and PD BASELINE TISSUE SAMPLES 238 (52.8%) patients with matched tumour and plasma results 397 (88%) patients consented 268 (59.4%) samples available 241 (53.4%) samples analysable PLASMA SAMPLES 447 (99.1%) baseline samples available 447 (99.1%) baseline samples analysable 305 (67.6%) Patients with plasma results at all three time points 451 (100%) patients consented 362 (80.3%) C3 samples available 362 (80.3%) C3 samples analysable 376 (83.4%) PD samples available 376 (83.4%) PD samples analysable

Dynamic mutant DNA change during therapy Patients treated with GC+P Patients treated with GC+E 100,000 100,000 10,000 10,000 1,000 1,000 Mutant DNA copy/mL of plasma 100 Mutant DNA copy/mL of plasma 100 10 10 1 1 Not detectable Not detectable Baseline C3 PD Baseline C3 PD

Positive versus negative pEGFR mut status at C3 (both treatment arms combined) pEGFR mut+ at C3 (n=42) ORR = 33% (14/42) OR=3.93 (95% CI: 1.78–8.66); p=0.0007 pEGFR mut+ at baseline (n=122) pEGFR mut– at C3 (n=80) ORR = 66% (53/80) ORR = objective response rate; OR = odds ratio

Association between pEGFR mut+ at C3 and PFS/OS (both treatment arms combined) C3 mut+ C3 mut+ C3 mut– C3 mut– Median=7.2 months (95% CI: 6.0–7.8) Median=12.0 months (95% CI: 9.6–16.5) HR=0.32 (95% CI: 0.21–0.48); p<0.0001 Median=18.2 months (95% CI: 14.2–27.4) Median=31.9 months (95% CI: 23.5–undefined) HR=0.51 (95% CI: 0.31–0.84); p=0.0066 1.0 0.8 0.6 0.4 0.2 1.0 0.8 0.6 0.4 0.2 PFS probability OS probability 7.2 12.0 18.2 31.9 0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 34 36 Time (months) Time (months) Patients, n Patients, n C3 mut+ 42 42 35 28 14 7 6 4 1 1 1 1 0 0 0 0 0 C3 mut– 80 80 77 65 59 47 40 34 32 28 23 19 13 10 7 3 0 C3 mut+ 42 42 42 41 37 32 30 28 23 21 18 14 14 12 9 4 3 2 C3 mut– 80 80 80 77 77 77 76 71 68 64 59 52 38 29 22 12 3 1 0 0 Positive pEGFR at baseline followed by negative pEGFR at C3 is associated with improved outcomes; patients positive at baseline and still positive at C3 experienced worse outcomes OS = overall survival

Detection of plasma EGFR mutations: summary Detection of EGFR mutations by UD-NGS EGFR status Baseline Progression Mutated 31 (72%) 11 (76%) Wild type 12 (28%) 4 (24%) Total 43 (100%) 15 (100%) Sensitivity: 72% Specificity: 100% Detection of EGFR mutations by cobas® test EGFR status Baseline Progression Mutated 30 (70%) 11 (73%) Wild type 13 (30%) 4 (27%) Total 43 (100%) 15 (100%) Sensitivity: 71% Specificity: 100% Marchetti, et al. WCLC 2015

EGFR SQI on erlotinib therapy in EGFR Mut+ Rapid responders Time (days) EGFR SQI Slow responders Time (days) EGFR SQI In 70% of patients EGFR SQI reduced by >50% at 14 days Patients had no T790M early mutations Two patients showed an increase in T790M mutations Slow responders may be more prone to developing resistance SQI, semi-quantitative index (% of mutant compared to WT) Marchetti, et al. WCLC 2015

Potential application Pathogenesis Symptoms treatment selection Resistance to treatment death Early detection Diagnosis and treatment selection Monitoring Mechanism of resistance and treatment

No corresponding rebiopsy tumor for T790M cfDNA for T790M Digital PCR (Rui Chen et al): Studied 135 patients with acquired resistance to TKI ARMS Digital PCR Pre-TKI (N=109) 5% T790M pos 30% T790M pos Post-TKI (N=135) 25% T790M pos 43% T790M pos No corresponding rebiopsy tumor for T790M Chen et al, WCLC, 2013

Serial monitoring of T790M from plasma DNA 47% found to have T790M Droplet Digital PCR Zhang et al ASCO 2014

Serial change in T790M by ddPCR Zhang et al ASCO 2014

Impact of post-PD treatment on plasma DNA for T790M

TIGER X: Plasma Testing for T790M using BEAMing When inadequate tissue specimens are factored in, plasma testing identifies as many patients as T790M+ as tissue testing T790M tissue-plasma+ are not false positives – T790M confirmed in plasma on subsequent testing in 5/7 samples studies Tissue Total Positive Negative Inadequate tissue Plasma 155 23 12 190 37 8 57 192 35 20 247 Tissue as reference: Positive percent agreement T790M 81% (155/192) Activating mutations 87% (193/221) Sequest et al ASCO 2015

T790M Plasma Testing is a Viable Alternative to Tissue Testing Objective response rate for 188 evaluable patients with both central T790M tissue test result and plasma T790M result Plasma T790M Tissue T790M + - 55% (72/130) 43% (13/30) 53% (85/160) − 35% (6/17) 27% (3/11) 32% (9/28) (78/147) 39% (16/41) Similar ORR observed when detecting T790M in either tissue or plasma Not all patients with progression on first-line TKI are candidates for tissue re-biopsy

Targeting EGFR mutation versus NGS

Guardant 360 Gene selection based on clinical utility - actionability 2015 Complete* or Critical Exon Coverage in 68 Genes All Exons and partial introns Covered (Copy Number Variations in Bold/ INDELS) Point Mutations AMPLIFICATIONs Fusions EGFR exon 19 deletions 38 patient samples were Analyzed using this panel Partial Coverage (“Hot” Exons): INDELS Valifor et al WCLC 2015

Mutations were detected in majority of patients Are these alternation clinically relevant? 78% (n=42/54) patients had at least 1 alteration

Distribution of alterations in 42 mutation positive Distribution of alterations in 42 mutation positive* patients (n mutations=115) *Non-synonymous only

Patients who really benefited from molecular testing 54 lung cancer cases with ctDNA analysis 42 with 1 or more alteration 7/42 (17%) FDA approved therapy 5 exon 19 deletion -2 with concurrent T790M mutations 2 L858R mutation 17/42 (40%) Therapy approved in other disease 11/42 (26%) Clinical trials available 12 no alterations identified Patients who really benefited from molecular testing EGFR 7/ALK 5/MET 2/BRAF 1 = 15 ALK 5 MET 2 BRAF 1 CDKNA2 5 KIT 2 BRCA1 2

NGS is to eat the whole genomic pie in one go

Alternative approach is to rule out EGFR mutation first

Prospective study of plasma genotyping in advanced NSCLC Eligible patients – adv. NSCLC Newly diagnosed (Cohort 1) n=120 EGFR resistance (cohort 2) n=60 Excluded – missed blood draw Cohort 1: n=3 Cohort 2: n=1 Cohort 1 – newly diagnosed Plasma ddPCR for EGFR exon 19 del/L858R and KRAS G12X n=117 Cohort 2 – EGFR resistance Plasma ddPCR for EGFR T790M/ exon 19 del/L858R n=59 Excluded – failed tissue genotyping EGFR: n=2 KRAS: n=30 Excluded – failed tissue genotyping T790M: n=5 KRAS G12X Tissue genotyping complete n=87 (26 KRAS G12X mutant) EGFR exon 19/L858R Tissue genotyping complete Cohort 1: n=115 Cohort 2: n=59 (87 EGFR mutant) EGFR T790M Tissue genotyping complete n=54 (35 T790M mutant) Pilot study of sequential plasma NGS, n=46 Sacher et al WCLC 2015

Positive Predictive Value ddPCR for EGFR and KRAS Assay Specificity Positive Predictive Value Sensitivity EGFR exon 19 del & L858R Cohort 1 99% 95% 77% Cohort 2 100% 80% Overall 98% 79% exon 19 del 92% 85% 81% 82% L858R 69% 78% 74% EGFR T790M 63% KRAS G12X 64% ddPCR from plasma cfDNA in ASPIRATIOM study Park et al ESMO 2014

Sequential plasma ddPCR and NGS Complex genomic alterations EGFR/KRAS mutant Study ID Tissue Genotype Plasma ddPCR Plasma NGS 15 ROS1 Negative Detected 18 ALK 22 EGFR G719A 36 81 PIK3CA E545K 107 RET, p53 127 130 EGFR exon 20 ins, p53 133 KRAS Q61H 137 RET 145 EGFR exon 20 ins 202 250 EGFR exon 19 del/ins 258 None 89 KRAS G13D 108 BRAF V600E 115 KRAS Q61L 169 209 Study ID Tissue Genotype Plasma ddPCR Plasma NGS 1 KRAS G12C Detected 4 EGFR exon 19 del 11 17 39 44 53 70 74 81 91 95 105 116 EGFR exon 19 120 EGFR exon 19/T790M 179 195 200 232 EGFR L858R/T790M 48 EGFR L858R Negative 244 8 28 61 45 94 KRAS G12V 109 Table of sensitivity in rare genotypes and overall cohort. Plasma NGS detected a diversity of complex genomic alterations in patients with negative plasma ddPCR (n=19) with 74% sensitivity. This assay was able to detect EGFR/KRAS mutations confirmed by orthogonal plasma ddPCR with 78% sensitivity (n=27). Plasma NGS detected 13/19 (74%) complex genomic alternation and 8/13 patients had FDA approved treatment

This approach is to take a fast big bite of the genomic pie with ddPCR

Possible paradigm -ive + Patient with newly diagnosed adenocarcinoma 3 to 4 days + ddPCR for EGFR EGFR TKI 10 to 14 days -ive NGS panel (tissue or blood)

Summary Finding EGFR mutation from cfDNA Monitoring progress Finding EGFR mutation from cfDNA is feasible Sensitivity 70 to 80%, highly specific ddPCR or BEAM Monitoring progress Potentially could be used to monitor the EGFR mutation status, but clinical relevance is lacking Detecting T790M Digital PCR is relatively sensitive BEAMing for T790M is associated with clinical response NGS Multiple technology platform is being developed Sensible to perform NGS only in patients without EGFR mutation in plasma

A liquid can be very solid Liquid Crystal