3.10

4.30

8.70

5.306.00

6.40

4.30

5.156.40

8.60

7.45

4.30

5.30

4.25

0.00

2.00

4.00

6.00

8.00

10.00

12.00

0 3 6 9 12 15

TMB

med

ain

Brain Breast CervicalColorectal Endometrium Esoph.& GastricGIST & soft Head & neck LiverLung Melanoma OvaryPancreas Thyroid

Identification of Lung Cancer Mutational Signatures and Tumor Drivers Associated with Specific Bimodal PD-L1/TMB Status

Methods424 FFPE clinical samples from lung cancer patients were analyzed using a CLIA-validated NGS-based assay that interrogates SNVs, indels using a 323 gene panel and by IHC for PD-L1 using theFDA approved PharmDx assay. TMB (mutations/Mb) is categorized as low (≤7), intermediate(7<TMB≤15) and high: (TMB> 15). NGS results were paired with PD-L1 status which was definedby tumor proportion scores (TPS) as: negative (TPS<1%), Low expressing (≥1-49%) and High(≥50%). In silico analyses were also performed on 5939 lung cancer samples from publicdatabases. The study was approved by Neogenomics Institution’s Ethics Board and external IRB,approval number 420160280.

NeoTYPE Discovery 323 + MSI + TMB Assay Clinical Validation Summary

Figure 1. NeoTYPE Discovery 323 +MSI +TMB assay.

145 solid tumor oncogenes and tumor suppressors 41 Homologous Recombination Repair (HRR) and

DNA damage repair genes 40 NCI-MATCH trial arms directly addressed Cancer pathways: EGFR/ RAS/ RAF MAPK/ PI3K/

MTOR, CDK/ Rb/ FGF and other RTK signaling cascades* Full gene list available upon request

Tumor Mutation Burden

TMB scores TMB category

Mut/Mb

≤7 Low

>7, <15 Intermediate

≥15 High

NeoTYPE Discovery 323 + MSI + TMB

Panel size 323 genes (all exons)

Genomic size 1 Mb

Chemistry Amplicon based

Sequencing platform Novaseq 6000

Microsatellites 27 target regions

Validated samples FFPE block/slides

Min tumor content 20% (30% MSI)

Recommended DNA input 100 ng

Minimum DNA input 50 ng

Type of test CLIA validated LDT

Reimbursement Status CMS covered

SNV/ InDel Clinical Performancevariants/

parameterSensitivity

(5% AF)Specificity Reprod.

SNVs 98.60% 98.10%99.2%

InDel 96.80% 94.30%

Tumor Mutation Burden Clinical Performance

metricAccuracy vs. WES

Reprod.intra assay

Reprod. inter assay

TMB score91.4% Pearson

coeff. - -TMB category

100%100% 100%

Microsatellite instability Clinical PerformanceScore Sensitivity Specificity Reprod.

MSI (p>0.95) 87.50% 100% 100%

Technical Assay performanceAverage coverage 1908XMinimum average coverage ≥500X

Limit of Detection SNVs3% hotspots, 5% all

others

Limit of detection InDel3% hotspots, 5% all

othersAnalytical Specifiticy 99.9996%

Analytical Sensitivity (LOD: AF, 95% CI)

2.5% (SNVs), 2.8% (InDel)

reproducibility intra-assay 98.66%reproducibility inter-assay 99.19%

NSCLC

Figure 2. Tumor Mutation Burden (TMB) across cancer types. TMB median of serially tested NSCLC tumors without any preselection criteria and compared with other tumor types processed in the same manner in the same period. In addition, all lung cancer tumors were Microsatellite Stable (MSS) as determined according to the assay specifications. Cervical and lung cancers presented the highest median values off all tumor sites. Sphere size represents the sample size for each cancer type (n=784)

NSCLC Tumor Mutation Burden

Min 0

25% lower quartile 5.3

median 8.6

75% upper Quartile 16

max 126.2

NSCLC Tumor Mutation Burden

TMB scores Category%

Samples

TMB≤7 Low 38.6 %

7<TMB <15 Intermediate 34.7 %

TMB≥15 High 26.7 %

26%

44%

31%

0%

10%

20%

30%

40%

50%

<1% 1%-49% 50-100%

NSC

LC p

atie

nts

(%

)

PDL1 TPS Categories

0

50

100

PD

-L1

TP

S (

%)

Samples

PDL1 Expression (22C3, FDA CDx)

R² = -0.266

0

20

40

60

80

100

0 50 100 150

PD

-L1

exp

resi

on

TMB

PD-L1 vs TMB correlation

PD-L1 - TMB Correlation in NSCLC

TMB

TMB-LowTMB-

Interm. TMB-High

Total

PD-L1

PD-L1 Neg. 40 37 31 108

PD-L1-Positive

73 67 46 186

PD-L1-High

51 43 36 130

Total 164 147 113 424

Tumor Mutation Burden in the Lung Cancer Cohort

Neg Positive High

Median=7%

Expression:

Figure 3. Mutation signatures and correlations between TMB and PD-L1 expression ina cohort of 424 NSCLC patients. TMB and PD-L1 expression lineal regression wasperformed. Samples were divided according to combined TMB and PD-L1 results. Thetop 12 most frequently mutated genes in each subcategory were plotted.

TPS:

Cervical

29%

24%20%

12%8%10%8% 8% 8%

4% 4% 6%

0%5%

10%15%20%25%30%35%

TP5

3

KR

AS

NO

TCH

2

EGFR

MET

RA

NB

P2

AR

ID1

A

PIK

3C

2B

TOP

2A

FAT1

FLT4

KM

T2A

mu

tate

d s

amp

les

(%)

PD-L1 High & TMB Low (n=51)

65%

42%

65%

29%23%19%16%19%16%13%16%16%

0%

10%

20%

30%

40%

50%

60%

70%

TP5

3

LRP

1B

SPTA

1

SMAR…

GN

AS

ALK

FGFR

2

SLIT

2

RO

S1

AM

ER1

FAT1

MED

12

mu

tate

d t

um

ors

(%

)

PD-L1 Neg. & TMB High (n=31)

28%

18%20%

10%10%8%

5%8% 8% 8%

5% 5%

0%

5%

10%

15%

20%

25%

30%

EGFR

KR

AS

TP5

3

RA

NB

P2

RB

M1

0

CTN

NB

1

SETD

2

SMA

RC

A4

SPEN

STK

11

AR

ID1

A

AR

ID1

Bmu

tate

d s

amp

les

(%

)

PD-L1 Neg. & TMB Low (n=40)

75%

58%

39%36%36%33%33%28%28%25%22%

17%

0%10%20%30%40%50%60%70%80%

TP5

3

LRP

1B

GN

AS

KR

AS

RU

NX

1T1

SPTA

1

KM

T2C

FAT1

KM

T2D

NTR

K3

AR

ID1

A

BR

CA

2mu

tate

d s

amp

les

(%)

PD-L1 High & TMB High (n=36)

Genomic Signatures Define Combined PD-L1 & TMB neg/Low vs High Categories

MEK 1/2

BRAF class 1/2/3ARAF/ RAF1

Figure 6. Genomic signatures on NSCLC patients with intermediate Mutation Burden classified by PD-L1 expression. The 12 most mutated genes associated with varying PDL1 expression are shown. Kaplan-Meyer regressions show progression free survival (left) or Overall Survival (right) on patients with EGFR vs. KEAP1 or STK11 mutant tumors publically available with or without alterations on these genes (cbioportal). Samples with overlapping mutation on these genes are excluded.

Figure 5. Genomic signatures on NSCLC samples divided by combined Tumor Mutation Burden (TMB) and PD-L1 expression. The top-12 most frequently mutated genes in each subcategory were included. Mutated samples were counted only once even if more than 1 mutation in the gene was detected. The genes were ordered by the total number of mutations detected. Overall survival from 5939 NSCLC patients with mutations vs wt status on the top 5 mutated genes on PDL1 Neg &TMB High, was estimated using Kaplan-Meyer regressions using cbioportal.

Results We found poor correlation between PD-L1 expression and TMB in NSCLC (r2=0.266) and

identified gene mutation signatures specific to groups defined by PD-L1 expression

combined with TMB scores.

KRAS mutations are constant across PDL1 TPS and less frequent on TMB High tumors, while

EGFR mutation frequency negatively correlates to TMB and less to PD-L1 TPS..

In PD-L1 Low/TMB High tumors (n=46) EGFR and KRAS were found mutated on 15% and 9%

of samples, respectively (not part of the top 12 genes), while NTRK and NOTCH signaling are

altered on 37% and 52%, respectively (65% combined).

In tumors with an intermediate TMB (7<TMB≤15), we find that KRAS mutations are more

frequent (43%) in PDL-1 High expressing tumors..

EGFR alterations have highest frequency on TMB intermediate/ PD-L1 Low tumors.

KEAP1 mutations, which have been recently proposed as poor prognosis biomarkers upon

immunotherapy and other therapies in NSCLC, are found on up to 22% of lung tumors with

TMB intermediate, including on 17% of tumors with TMB-intermediate and high PD-L1

expression.

STK11 mutations on TMB intermediate tumors are present on 16% of PDL1 negative tumors

but are rare (2%) on High PDL-1 expressing tumors.

KEAP1 mutations predict better PFS than EGFR and STK11, but EGFR mutations predict

longer OS than KEAP1 or STK11 according to publically available data

Mutations on chromatin remodeling /DNA repair SMARCA4 and ARID1A/B genes are

frequently found on the TMB intermediate group.

We also identified a 5 genes set specifically mutated on PDL1-Neg/TMB High tumors with

potential prognostic/predictive value.

Figure 4. EGFR and KRAS alterations. Mutation frequency of the 2 genes was segregated by TMB or PD-L1 categories.

EGFR/KRAS Signaling in NSCLC Cohort

Conclusions Genomic alteration signatures might define subsets of lung cancer tumors with no PD-L1

expression to complement TMB and PD-L1 on the selection criteria for patients whom maybenefit from checkpoint inhibitors.

Further studies are needed to evaluate specific signatures or single genes as biomarkers ofimmunotherapy response in NSCLC patients as well as to identify the molecularmechanisms enabling cancer cells therapy resistance and disease progression.

0%

5%

10%

15%

20%

25%

30%

Low Interm. High Neg. Pos. High

TMB PD-L1

mu

tate

d s

amp

les

(%)

EGFR /KRAS mutations vs. TMB / PDL1

EGFR KRAS

Fernando J. Lopez-Diaz*, Lauryn Keeler, Sally Agersborg, Lawrence Weiss, Vincent Funari. Neogenomics Laboratories, Carlsbad, CA. *fernando.lopezdiaz@neogenomics.com

HRR

DDR

Neo_323

NCI-MATCH

65%

30%22%19%19%16%16%16%14%14%14%14%

0%

20%

40%

60%

80%

TP5

3

LRP

1B

KEA

P1

KR

AS

FAT1

SPTA

1

KM

T2D

STK

11

RB

M1

0

KM

T2A

SMA

RC

A4

PTC

H1

mu

tate

d s

amp

les

(%)

TMB Int. & PD-L1 Neg. (n=37)

67%43%

21%19%19%17%17%14%14%14%14%14%

0%

20%

40%

60%

80%

TP5

3

KR

AS

LRP

1B

CD

KN

2A

RB

M1

0

FAT1

KEA

P1

NO

TCH

2

NF1

SPTA

1

AR

ID1

B

AR

ID1

A

mu

tate

d s

amp

les

(%)

TMB Int. & PD-L1 high (n=43)

60%

30%24%15%15%13%12%12% 9% 12%10% 9%

0%

20%

40%

60%

80%

TP5

3

LRP

1B

KR

AS

EGFR

KEA

P1

SPTA

1

KM

T2D

CD

KN

2A

ATR

X

NO

TCH

2

SLIT

2

ATM

Mu

tate

d s

amp

les

(%)

TMB Int. & PD-L1 Low (n=67)

24%21%20%

14%

7% 7% 6% 5% 5% 5% 4% 4%

0%

5%

10%

15%

20%

25%

30%

TP5

3

EGFR

KR

AS

NOTC

…

RA

NB

P2

STK

11

RB

M1

0

KEA

P1

LRP

1B

AR

ID1

A

CD

KN

2A

BR

CA

2mu

tate

d s

amp

les

(%)

TMB Low (n=164)

63%

34%28%

18%18%16%14%11%14%14%12%11%

0%

20%

40%

60%

80%

TP5

3

LRP

1B

KR

AS

KEA

P1

SPTA

1

FAT1

RB

M1

0

EGFR

KM

T2D

NO

TCH

2

AR

ID1

A

CD

KN

2A

mu

tate

d s

amp

les

(%) TMB Int. (n=147)

29%22%

19%12%12%12%12%10% 9% 9% 9% 9%

0%

10%

20%

30%

40%

TP5

3

LRP

1B

SPTA

1

NOTC

…

KM

T2D

KEA

P1

NTR

K3

NOTC

…

RUNX…

RANB…

EPH

A3

NF1

mu

tate

d s

amp

les

(%) TMB High (n=113)

50%

28%

18%16%17%15%11%12%11%11%11%11%

0%

20%

40%

60%

TP5

3

LRP

1B

SPTA

1

SMA…

KR

AS

EGFR

RBM…

FAT1

KMT…

KEA

P1

NOT…

STK

11

mu

tate

d s

amp

les

(%)

PDL1 Neg (n=108)

48%

27%17%17%19%18%15%12%10% 9% 11% 9%

0%

20%

40%

60%

TP5

3

LRP

1B

SPTA

1

EGFR

KR

AS

NOTC

…

KEA

P1

KM

T2D

RANB…

PR

KD

C

CDKN…

RB

M1

0

mu

tate

d s

amp

les

(%) PD-L1 Pos. (n=186)

32%

20%19%13%12%13%12%

8% 12%12% 9% 10%

0%

10%

20%

30%

40%

TP5

3

LRP

1B

KR

AS

SPTA

1

FAT1

NOTC

…

GN

AS

NTR

K3

KM

T2C

KM

T2D

PIK3C…

AR

ID1

Amu

tate

d s

amp

les

(%) PD-L1 High (n=130)

The Genomic Signatures on Different TMB Intermediate NSCLC Scenarios

5-genespredictor

TP53

LRP1B

SPTA1

SMARCA4

GNAS

72%

54%

41%

30% 28% 26% 24% 24% 22% 22% 22% 22%

0%10%20%30%40%50%60%70%80%

TP5

3

LRP

1B

SPTA

1

NO

TCH

2

NTR

K3

KEA

P1

KM

T2D

NO

TCH

3

RA

NB

P2

RU

NX

1T1

EPH

A3

NF1

mu

tate

d s

amp

les

(%)

PD-L1 Pos. & TMB High (n=46)

26% 25%

18% 17%

8% 8% 8% 8% 8% 7% 7% 7%

0%

5%

10%

15%

20%

25%

30%

EGFR

TP5

3

KR

AS

NO

TCH

2

CD

KN

2A

LRP

1B

STK

11

RA

NB

P2

KEA

P1

PIK

3C

2B

RB

M1

0

EPH

A7

Mu

tate

d s

amp

les

(%)

PD-L1 Pos. & TMB Low (n=72)

PFS Overall Survival Overall Survival

EGFR EGFR

EGFRKEAP1

STK11

KEAP1STK11

Overall Survival

Altered

Non-altered

DataBackgroundPD-L1 expression and Tumor Mutation Burden (TMB) have independently emergedas prospective biomarkers of response to anti PD1-/PDL1 checkpoint inhibitors.Combined use of TMB, PD-L1 protein levels has been proposed. However, how thetumor genomic landscape interplays with the tumor microenvironment in definingparticular predictive therapy response statuses is not clear. Moreover there is stilllittle real world data available regarding the combined prevalence of thesebiomarkers in specific genomic landscapes.