cancerres.aacrjournals.org€¦ · web view, ctdna dynamics revealed an increase in levels of...

Supplementary Data for

Dynamics of tumor and immune responses during immune checkpoint

blockade in non-small cell lung cancer

Valsamo Anagnostou1,2#*, Patrick M. Forde1,2*, James R. White1, Noushin Niknafs1, Carolyn

Hruban1, Jarushka Naidoo1,2, Kristen Marrone1,2, I.K. Ashok Sivakumar1,3,4, Daniel Bruhm1,

Samuel Rosner5, Jillian Phallen1, Alessandro Leal1, Vilmos Adleff1, Kellie N. Smith1,2, Tricia

Cottrell1,6, Lamia Rhymee1, Doreen Palsgrove1, Christine L. Hann1, Benjamin Levy1, Josephine

Feliciano1, Christos Georgiades7, Franco Verde7, Peter Illei1,2,6, Qing Kay Li1,6, Edward

Gabrielson1,6, Malcolm V. Brock8, James M. Isbell9, Jennifer L. Sauter10, Janis Taube1,2,6, Robert

B. Scharpf1, Rachel Karchin1,3, Drew M. Pardoll1,2, Jamie E. Chaft11, Matthew D. Hellmann11,

Julie R. Brahmer1,2 and Victor E. Velculescu1,2,3#

1The Sidney Kimmel Comprehensive Cancer Center, Johns Hopkins University School of Medicine, Baltimore, MD 21287, USA

2The Bloomberg-Kimmel Institute for Cancer Immunotherapy, Johns Hopkins University School of Medicine, Baltimore, MD 21287, USA

3Institute for Computational Medicine, Johns Hopkins University, Baltimore, MD 21204, USA.

4Applied Physics Laboratory, Laurel, MD 20723, USA

5Department of Internal Medicine, Johns Hopkins Bayview Medical Center, Baltimore, MD 21224, USA

6Department of Pathology, Johns Hopkins University School of Medicine, Baltimore, MD 21287, USA

7Department of Radiology, Johns Hopkins University School of Medicine, Baltimore, MD 21287, USA

8Department of Surgery, Johns Hopkins University School of Medicine, Baltimore, MD 21287, USA

9 Thoracic Service, Department of Surgery, Memorial Sloan Kettering Cancer Center, New York, New York, NY 10065, USA.

10Department of Pathology, Memorial Sloan Kettering Cancer Center, New York, NY 10065, USA.

1

11Thoracic Oncology Service, Department of Medicine, Memorial Sloan Kettering Cancer Center and Weill Cornell Medical College, New York, NY 10065, USA

* These authors contributed equally to this work.# Corresponding authors

2

Supplementary Figure S1

0

30

60

90

0%

2%

4%

6%

8%

Pre-

ther

apy

Wee

k 3

Wee

k 6

Wee

k 9

Wee

k 15

Wee

k 18

Wee

k 24

0

5

10

15

0.00%

0.04%

0.08%

0.12%

Base

line

W4

W9

W21

W32

W37

CGLU160

0

40

80

120

0%

1%

2%

3%Ba

selin

e

W4

W8

W17

W20

W29

CGLU168KRAS 12G>DRECIST SLD

7.3mo12.6mo

PFSOS

TMB=411

STK11 272Y>XRECIST SLDNew mets

13.6mo13.6mo

PFSOS

TMB=50

TP53 179H>RRECIST SLD

10.7mo21.3mo

PFSOS

TMB=161

12.3mo12.8mo

PFSOS

TMB=368

KRAS 12G>CTP53 P72Rf*76RECIST SLD

A

B

MAF

(%)

RECI

ST S

LD (m

m)

MAF

(%)

RECI

ST S

LD (m

m)

0

10

20

30

40

0.0%

0.2%

0.4%

0.6%

Base

line

W6

W8

W17

W31

CGLU211D

MAF

(%)

RECI

ST S

LD (m

m)

0

40

80

120

0.0%

0.2%

0.4%

Base

line

W6

W8

W18

W41

CGLU212E

MAF

(%)

RECI

ST S

LD (m

m)

G

0

10

20

30

0.0%

0.3%

0.6%

0.9%

W44

W51

W79

W11

0

CGLU135

TP53 240E>XRECIST SLD

23.7mo38.7mo

PFSOS

TMB=411MAF

(%)

RECI

ST S

LD (m

m)

MAF

(%)

RECI

ST S

LD (m

m)

CGLU347KRAS 13G>CRECIST SLD

5mo12mo

PFSOS 0

30

60

0%

1%

2%

3%

4%

Wee

k 3

Wee

k 6

Wee

k 9

Wee

k 13

Wee

k 16

Wee

k 22

Wee

k 48

KRAS 12G>CRECIST SLD

12mo12mo

PFSOS

MAF

(%)

RECI

ST S

LD (m

m)

CGLU351C F

0

60

120

180

0%

3%

6%

9%

Pre-

ther

apy

Wee

k 4

Wee

k 7

Wee

k 13

Wee

k 19

Wee

k 22

APC 2272R>XTP53 D259Efs*3RB1 533E>XSMAD4 499C>YRECIST SLD

14mo14mo

PFSOS

TMB=846

MAF

(%)

RECI

ST S

LD (m

m)

CGLU337H

ctDNA clonal dynamics during anti-PD1 for patients with a clinical response. For patients with a molecular response, ctDNA levels decreased to undetectable levels early after anti-PD1 therapy initiation (A-H), followed by a rise in levels of tumor specific variants at the time of molecular acquired resistance (A, C, F, G). For CGLU135, where baseline blood was not available for analysis, ctDNA levels rose at the time of acquired resistance compared to the time of response (B). Changes in RECIST tumor burden, shown on the secondary vertical axis of each graph, indicated stable disease for the majority of molecular responders and were therefore not reflective of each patient’s clinical benefit from immune checkpoint blockade. These tumors harbored a wide range of sequence alterations (TMB; 50-846) indicating that TMB alone may be not sufficient to predict response. MAF; mutant allele fraction, SLD; sum of longest diameters, TMB; tumor mutation burden, PFS; progression-free survival, OS; overall survival.

3


0

20

40

60

80

0%

2%

4%

6%

8%

10%Ba

selin

e

W4

W9

W13

CGLU115

0

60

120

180

0%

1%

2%

3%

4%

Base

line

W4

W6

CGLU203

*

AR 504Y>XRECIST SLDNew mets

3.2mo3.8mo

PFSOS

TMB=285

KRAS 12G>CRECIST SLDNew bone mets

3.9mo5.6mo

PFSOS

*

TMB=90

A

B

MAF

(%)

RECI

ST SL

D (m

m)

MAF

(%)

RECI

ST SL

D (m

m)

0

10

20

30

40

0%

1%

2%

3%

4%

Base

line

W4

W6

W11

W16

CGLU159

0

20

40

60

0%

5%

10%

15%

Base

line

W4

W8

W9

CGLU243

TP53 192Q>XRECIST SLDNew mets

3.9mo5.6mo

PFSOS

TMB=65

EGFR L747_T751delRB1 T345Lfs*4RECIST SLDNew mets

2.4mo11.4mo

PFSOS

TMB=42

D

E

MAF

(%)

RECI

ST SL

D (m

m)

MAF

(%)

RECI

ST SL

D (m

m)

*

0

10

20

30

40

0%

1%

2%

3%

Base

line

W4

W8

W18

W28

TP53 244GGMN>DRET 3039+1G>CRECIST SLDLost to follow-up

CGLU162

7mo7mo

PFSOS

TMB=169

*

MAF

(%)

C

RECI

ST SL

D (m

m)

0

50

100

0.0%

0.8%

1.6%

2.4%

Pre-

ther

apy

Wee

k 3

Wee

k 4

Wee

k 8

Wee

k 10

3mo3mo

PFSOS

TMB=191

PTEN 240_253+6delTP53 163Y>CRECIST SLD

CGLU348

MAF

(%)

RECI

ST SL

D (m

m)

G

0

40

80

120

0%

5%

10%

15%

Pre-

ther

apy

Wee

k 1

Wee

k 8

Wee

k 14

Wee

k 16

Wee

k 19

Wee

k 27

KRAS 12G>CRECIST SLDNew mets

6.6mo13.4mo

PFSOS

RECI

ST SL

D (m

m)

MAF

(%)

CGLU340H

0

30

60

90

0.0%

0.4%

0.8%

Pre-

ther

apy

Wee

k 3

Wee

k 4

Wee

k 6

Wee

k 9

Wee

k 11

Wee

k 15

Wee

k 21

KRAS 61Q>HRECIST SLDNew LN mets

5.2mo13mo

PFSOS

MAF

(%)

RECI

ST SL

D (m

m)

F CGLU357

0

40

80

120

0%

1%

2%

3%

4%

Base

line

W7

W8

W14

W15

MAF

(%)

CGLU117TP53 193H>RNTRK3 483H>NRB1 344Q>XRECIST SLD

7.8mo13.8mo

PFSOS

TMB=296

I

RECI

ST SL

D (m

m)

ctDNA clonal dynamics during anti-PD1 for patients with molecular primary resistance. ctDNA levels continued to rise or showed no significant changes from baseline for molecular non-responders (A-I). Detection of ctDNA molecular resistance preceded disease progression determined by RECIST response assessments by an average of 5.5 weeks. Tumor mutation burden did not accurately predict response as a high mutation burden was observed for patients CGLU117, CGLU115, CGLU162 and CGLU348 (A, C, G, I). MAF; mutant allele fraction, SLD; sum of longest diameters, TMB; tumor mutation burden, PFS; progression-free survival, OS; overall survival.

4

Supplementary Figure S3Ti

me

(wee

ks)

0

15

30

45

1 2

P=0.004

Time to best CT response (RECIST 1.1)

Time to molecular response (ctDNA)

ctDNA molecular responses precede radiologic responses. Detection of molecular responses was achieved within 6.7 weeks, on average 8.7 weeks earlier than conventional RECIST1.1 response assessments (6.7 vs 15.4 weeks, p=0.004). Patient CGLU337 was excluded from this graph, as we could not determine time to best response by radiologic imaging.

5


Molecular Responders

Molecular Progressors

Molecular Responders

Molecular Progressors

P=0.002 P=0.008

A B

Molecular responses correlate with a favorable prognosis. Patients with a molecular response had a longer progression-free (A) overall survival (B) compared to molecular non-responders (Mann Whitney P=0.002 and P=0.008 respectively).

6


Duration of Molecular Response (Weeks)

Ove

rall

Surv

ival

Duration of Molecular Response (Weeks)

Prog

ress

ion-

free

Surv

ival

A B

Duration of molecular response correlates with progression-free and overall survival. Minimum duration of molecular response was retrieved for 7 patients who achieved a molecular response but ultimately developed acquired resistance. Duration of molecular response highly correlated with progression-free (A) and overall survival (B) for these patients (Pearson’s coefficient R2=0.86 and R2=0.92 respectively).

7


P=0.04

PRSDPD

Disease prognostication by radiologic imaging at first assessment. Patients with partial response had a longer overall survival compared to patients with stable disease and progressive disease (log rank P=0.04). The disease prognostication by imaging was inferior to molecular response assessments. PR; partial response, SD; stable disease, PD; disease progression.

8


Undetectable ctDNADetectable ctDNA

P=0.01

Differential clinical outcome by molecular responses in patients with radiologically stable disease. Patients with stable disease (n=12) were more accurately classified in terms of clinical outcome based on their molecular response pattern. Within this group, patients with a molecular response (n=5) had a significantly longer progression-free survival compared to molecular non-responders (n=7; log rank P=0.01).

9


High TMBLow TMB

P=0.063 P=0.027

High TMBLow TMB

A B

Disease prognostication by tumor mutation burden. Patients with higher tumor mutation burden (TMB) had a marginally longer PFS (A, log rank p=0.063) and significantly longer overall survival compared to tumors with low TMB (B, log rank p=0.027).

10


Clinical BenefitMolecular Response

Response (RECIST)

Clonal Tumor Mutation Burden

850

P=0.027

A BcTMBPRSDPDctDNA ResponsectDNA ResistanceDurable Clinical BenefitNon-durable Clinical Benefit

ctDNA molecular response more accurately distinguish clinical responders from non-responders compared to clonal tumor mutation burden. Patients with a molecular response (shown in green) achieved durable clinical benefit from immune checkpoint blockade compared to molecular progressors (shown in blue) independent of the clonal tumor mutation burden (A). Representative examples of patients with a high mutation burden with a molecular progression signature and dismal prognosis are highlighted with black arrows and vice versa, a patient with molecular response achieving durable clinical benefit despite a low tumor mutation burden is highlighted with a red arrow. Survival analysis revealed that patients with a ctDNA molecular response had a longer overall survival independent of the tumor mutation burden (B). cTMB; clonal tumor mutation burden, PR; partial response, SD; stable disease, PD; disease progression.

11


0%

40%

80%

120%

0.0%

0.1%

0.2%

Pre-therapy Week 2 Week 4 Week 9

Tum

or B

urde

n

Mut

ant A

llele

Fra

ction

TP53 743G>T% tumor burden at resection% tumor burden by RECIST

0%

40%

80%

120%

0.0%

0.1%

0.2%

0.3%

0.4%

Pre-therapy Week 4 Week 5 Week 8

Tum

or B

urde

n

Mut

ant A

llele

Fra

ction

ALK V7751% tumor burden at resection% tumor burden by RECIST

0%

40%

80%

120%

0.0%

0.1%

0.2%

0.3%

0.4%

Pre-therapy Week 2 Week 4

RB1 K427*% tumor burden at resection% tumor burden by RECIST

Mut

ant A

llele

Fra

ction

Molecular Responder

Tum

or B

urde

n

Molecular Responder

0%

40%

80%

120%

0.0%

0.1%

0.2%


Tum

or B

urde

n

Mut

ant A

llele

Fra

ction

TP53 919+1G>T% tumor burden at resection% tumor burden by RECIST

Molecular Responder

0%

40%

80%

120%

0.0%

1.0%

2.0%

3.0%


Tum

or B

urde

n

Mut

ant A

llele

Fra

ction

KRAS G12CSTK11 E223Rfs*43% tumor burden at resection% tumor burden by RECIST

Molecular Non-Responder

Molecular Responder

A B

C D

E

Molecular responses mirror pathologic responses to anti-PD1 therapy in early stage operable NSCLC. Patients with major pathologic response (<10% residual tumor at the time of resection) and tumors with partial pathologic response (>30% reduction in tumor burden at the time of resection) were also identified as molecular responders by ctDNA dynamics (A-D). In contrast, for a patient with 75% residual tumor at the time of resection (E), ctDNA dynamics revealed an increase in levels of tumor specific mutations, consistent with a molecular and pathologic nonresponse. Radiologic imaging failed to accurately determine the therapeutic effect as an assessment of stable disease was established for all patients. RECIST tumor burden dynamics are plotted on the secondary vertical axis of each graph. Timing of anti-PD1 therapy, CT imaging and surgery is shown for each patient at the bottom of each graph.

12


0

0.5

1

1.5Ba

selin

e

W18

W30

A

0

0.5

1

1.5

2

W44

W11

0

CGLU127

0

0.2

0.4

0.6

0.8

1

W26

W34

CGLU135

CGLU161

Prod

uctiv

e Fr

eque

ncy (

%)

Prod

uctiv

e Fr

eque

ncy (

%)

Prod

uctiv

e Fr

eque

ncy (

%)

B

D

0

0.5

1

1.5

2

Base

line

W8

W37

E CGLU117Pr

oduc

tive

Freq

uenc

y (%

)

0

0.5

1

1.5

Base

line

Wee

k 6

Wee

k 8

Wee

k 18

Prod

uctiv

e Fr

eque

ncy (

%)

CGLU212C

TCR clonal dynamics during response and acquired resistance to anti-PD1. Intratumoral TCR clonotypic amplifications were observed in peripheral blood at the time of radiographic response compared to baseline (A, C, E) followed by clone contractions at the time of acquired resistance (A, E). For patients CGLU161 and CGLU135, where baseline blood was not available, a significant decrease in productive frequencies of intratumoral clones was observed in the peripheral blood at the time of acquired resistance compared to radiographic response (B, C). Average productive frequencies of the statistically significant differentially abundant TCR clones are shown for each individual patient, error bars represent standard error of the mean.

13


0

0.5

1

1.5

2

Base

line

Wee

k 4

CGLU115

0

2

4

6

8

Base

line

W4

W11

W16

CGLU159A D

Prod

uctiv

e Fr

eque

ncy (

%)

Prod

uctiv

e Fr

eque

ncy (

%)

0

0.5

1

1.5

2

2.5

Base

line

W4

W8

W10

W18

W24

W28

CGLU162

Prod

uctiv

e Fr

eque

ncy (

%)

B

0

0.5

1

1.5

2

2.5

Base

line

W4

Prod

uctiv

e Fr

eque

ncy (

%)

CGLU203

0

0.5

1

1.5

2

Base

line

Wee

k 4

Wee

k 8Pr

oduc

tive

Freq

uenc

y (%

)

CGLU243

E

C

TCR clonal dynamics for patients with clinical primary resistance. There were no clones with statistically significant differential abundance between baseline and week 4 for patient CGLU115 with primary resistance to anti-PD1 therapy (A). For patients CGLU162 (B), CGLU243 (C) and CGLU203 (E) clonotypic expansions were observed in peripheral blood, however this pattern was not consistent with the ctDNA molecular response or the clinical outcome. For patient CGLU159 transient clonotypic TCR amplifications were detected in the peripheral blood followed by clone contraction at the time of clinical primary resistance determined by radiologic imaging (D). Average productive frequencies of the statistically significant differentially abundant TCR clones are shown for each individual patient, error bars represent standard error of the mean.

14

Supplementary Figure S13CGLU111

Frac

tion

of T

CR p

rodu

ctive

re-a

rran

gem

ents

CGLU121

Frac

tion

of T

CR p

rodu

ctive

re-a

rran

gem

ents

Not significantly expanded clones

Differential VJ gene usage for a patient with clinical response compared to a patient with clinical primary resistance. For patient CGLU111, who had a sustained response to anti-PD1 therapy, differential V and J gene usage is shown between baseline and time of radiographic response (week 18). Usage of V and J gene segments increased at the time of radiologic response (A). In contrast, for patient CGLU121, who developed primary resistance to anti-PD1 therapy, we did not observe any changes in V and J gene usage between baseline and week 4 (B). Clones with significant expansions are colored, TCR clonotypes with no significant expansion between the two time points are shown in gray.

15


CGLU206 CGLU205

CGLU221 CGLU222

Conc

entr

ation

(pg/

ml)

Conc

entr

ation

(pg/

ml)

Conc

entr

ation

(pg/

ml)

Conc

entr

ation

(pg/

ml)

Dynamic changes in pro-inflammatory and immunosuppressive cytokines in early stage NSCLC patients with differential responses to anti-PD1 therapy. Cytokines were measured by a multiplexed immunoassay at baseline and 2-6 weeks on anti-PD1 therapy for a patient with a major pathologic response (CGLU206), 2 patients with partial pathologic response (CGLU205, CGLU221) and a patient with pathologic nonresponse (CGLU222). We did not identify any differences in cytokine levels among patients with differential responses to anti-PD1 therapy. Differences in cytokine concentration were evaluated by T test.

16

cancerres.aacrjournals.org€¦ · web view, ctdna dynamics revealed an increase in levels of...

Documents