home | clinical cancer research · web viewnodal disease right hilum and mediastinum e006 48 2 +-t2...

28
Online only Supplementary Materials Supplementary Methods Whole exome sequencing The diagnostic FFPE blocks were visually inspected and the ones with the most residual tumor were used for DNA isolation. A single H&E tissue section was reviewed by a consultant histopathologist (D M) and a minimum of 2 regions from the tumor were macrodissected using a 1 mm tissue microarray core needle. DNA was extracted from the FFPE tumor cores using the Gene Read kit (Qiagen) according to the manufacturer’s instructions and DNA concentration was measured as previously described. 1 Illumina HiSeq was used to perform whole-exome sequencing on 200-500 ng of pooled tumor DNA from 1-3 regions cored from each FFPE primary tumor block (sequencing was performed as a fee per service by Novogene) at an average deduplicated on-target read depth of 150x for all 49 tumor DNA and 50x for 49 matched germline samples. All sequencing data will be deposited in the European Genome–Phenome Archive on publication. Signatera Workflow Patient specific tumor profiles A clinical grade, strictly-controlled semi-automated lab process was performed by trained personnel with signed-off standard operating protocols and witnessing. The process and reagent/equipment information were captured electronically and uploaded to a database with built-in integrity checks. Quality control (QC) was performed at every step of the workflow (Supplementary Figure S1). Samples and amplicons that did not pass pre-specified QC were excluded from further analyses. A set of 45 SNPs were genotyped in tumor and plasma from each patient to ensure sample concordance. Primary tissue from each patient was sequenced by whole exome sequencing (WES). WES data was used to design bespoke 16-plex assays specific to the tumor mutational signature for each patient. For each target variant identified in the plasma, a confidence score was calculated based on the depth of read for mutant and reference alleles. Blood sample Collection, Cell free DNA extraction and quantification Blood samples were collected in K2-EDTA tubes. Samples were processed within 2 hours of collection by double centrifugation of the blood, first for 10 minutes at 1000g, then the plasma for 10 minutes at 2000g. Plasma was stored at −80 °C. Up to 8 ml of plasma 1

Upload: others

Post on 31-Jan-2021

0 views

Category:

Documents


0 download

TRANSCRIPT

Online only Supplementary Materials

Supplementary Methods

Whole exome sequencing

The diagnostic FFPE blocks were visually inspected and the ones with the most residual tumor were used for DNA isolation. A single H&E tissue section was reviewed by a consultant histopathologist (D M) and a minimum of 2 regions from the tumor were macrodissected using a 1 mm tissue microarray core needle. DNA was extracted from the FFPE tumor cores using the Gene Read kit (Qiagen) according to the manufacturer’s instructions and DNA concentration was measured as previously described.1

Illumina HiSeq was used to perform whole-exome sequencing on 200-500 ng of pooled tumor DNA from 1-3 regions cored from each FFPE primary tumor block (sequencing was performed as a fee per service by Novogene) at an average deduplicated on-target read depth of 150x for all 49 tumor DNA and 50x for 49 matched germline samples. All sequencing data will be deposited in the European Genome–Phenome Archive on publication.

Signatera Workflow

Patient specific tumor profiles

A clinical grade, strictly-controlled semi-automated lab process was performed by trained personnel with signed-off standard operating protocols and witnessing. The process and reagent/equipment information were captured electronically and uploaded to a database with built-in integrity checks. Quality control (QC) was performed at every step of the workflow (Supplementary Figure S1). Samples and amplicons that did not pass pre-specified QC were excluded from further analyses. A set of 45 SNPs were genotyped in tumor and plasma from each patient to ensure sample concordance.

Primary tissue from each patient was sequenced by whole exome sequencing (WES). WES data was used to design bespoke 16-plex assays specific to the tumor mutational signature for each patient. For each target variant identified in the plasma, a confidence score was calculated based on the depth of read for mutant and reference alleles.

Blood sample Collection, Cell free DNA extraction and quantification

Blood samples were collected in K2-EDTA tubes. Samples were processed within 2 hours of collection by double centrifugation of the blood, first for 10 minutes at 1000g, then the plasma for 10 minutes at 2000g. Plasma was stored at −80 °C. Up to 8 ml of plasma per case was available for this study (range, 1–8 ml; median 5 mL). The entire volume of plasma was used for cfDNA extraction using the QIAamp Circulating Nucleic Acid kit (Qiagen) and eluted into 50 μL DNA Suspension Buffer (Sigma). Each cfDNA sample was quantified by Quant-iT High Sensitivity dsDNA Assay Kit (Invitrogen). In 49 patients, cfDNA was isolated from a total of 215 serial plasma samples.

Cell free DNA library preparation

Up to 66 ng (20,000 genome equivalents) of cfDNA from each plasma sample was used as input into the library preparation, as described previously.25 The cfDNA was end-repaired, A-tailed, and ligated with custom adapters. The purified ligation product was amplified for 20 cycles and purified using Ampure XP beads (Agencourt/Beckman Coulter).

Plasma multiplex-PCR NGS workflow

A 16% aliquot of each library was used as input into the associated patient-specific 16-plex PCR reaction. Samples were amplified using patient-specific assay and barcoded, followed by pooling. Sequencing was performed on an Illumina HiSeq 2500 Rapid Run with 50 cycles of paired-end reads using the Illumina Paired End v2 kit with an average read depth of >100,000X per amplicon.

Bioinformatics pipeline

All paired-end reads were merged using Pear software.3 Bases that did not match in forward and reverse reads or had a low quality score were filtered out to minimize sequencing errors. Merged reads were mapped to the hg19 reference genome with Novoalign version 2.3.4 (http://www.novocraft.com/). Amplicons with < 5,000 high quality reads were considered to have failed the QC. QC was performed using an in-house program checking for a wide list of statistics per sample that included total numbers of reads, mapped reads, on-target reads, number of failed targets and average error rate (Figure S5).

Plasma variant calling

A large set of negative control samples (~1000) were pre-processed to build a background error model. For each target variant using mutant and reference alleles depth of read, a confidence score was calculated on the basis of the error model, as described in Abbosh et al. 2017.2 As previously, a plasma sample with at least 2 variants with a confidence score above a predefined algorithm threshold (0.97) was defined as ctDNA positive.

Analytical validation

The analytical validation was performed using two different sets of titration samples built from a) mononucleosomal DNA from cancer cell lines, b) a commercially available mutation mixture “SeraseqTM ctDNA Mutation Mix v2”. Mononucleosomal DNA from three cancer cell lines, including two breast cancer cell lines (HCC2218 & HCC1395) and one lung cancer cell line (NCI-H1395) were titrated into their respective matched normal B lymphoblast-derived cell lines (HCC2218-BL, HCC1395-BL, and NCI-BL1395) matched normal counterparts (ATCC). For each cell line pair, DNA was exome-sequenced, target variants were selected, and two multiplex-PCR primer pools were designed using the Signatera pipeline. Titrations of tumor into normal mononucleosomal DNA were made at average VAFs (based on DNA input) of 1%, 0.5%, 0.3%, 0.1%, 0.05%, 0.03%, 0.01%, 0.005%, 0%. Replicate numbers were from two to nine— increasing with the dilution factor. Due to heterogeneity and aneuploidy in the tumor cell lines, the VAFs of the individual targets differ from the average input tumor fraction. To accurately calculate the nominal VAF of each target in each titration step, a calibration experiment was performed with 10% VAF mixtures. The observed VAFs were then used to calculate input correction factors that were applied to their respective targets in the titration series. Additional negative samples were run using cfDNA isolated from 16 healthy human plasmas (approximately 8 mL each). For the titration series, 66 ng cfDNA (corresponding to 20,000 haploid genome equivalents) were used as input into library preparation; for the plasma cfDNA samples all isolated DNA (13-55 ng) was used as input. These libraries were then run through the plasma workflow (two different primer pools for each titration sample and five primer pools for each cfDNA sample), sequenced, and analyzed with our analysis pipeline.

Accuracy of 100% was achieved by calling all 132 positive samples as ctDNA positive with VAF down to 0.03%, and all 134 negative control samples as ctDNA negative. We calculate the sensitivity for 180 samples with VAF down 0.01% to be 93%. Figure S6A shows the estimated sensitivity for detecting individual mutations in plasma at various levels of tumor concentration by using the confidence threshold. Sensitivity of >99% at mutation frequencies down to 0.1% was achieved. However, in a total of 134 negative samples, 410 negative base changes (out of 107, 078 assessed) and 19 negative INDELs (out of 35,651 possible) were measured above the confidence threshold. Specificity was estimated to be 99.71% for individual mutation detection. Specificity for a 16-plex assay would thus be only 95.5% if a single measurement above the threshold was used to call the plasma sample ctDNA positive. In order to achieve high specificity of > 99.8% we require 2 mutations to be measured above the confidence threshold as previously described.4

The sample-level sensitivity was derived by calculating a binomial probability for detecting at least two clonal mutations at a given ctDNA level, assuming that the majority of the custom panels have between 10 to 16 clonal variants (Figure S6A). Clonality inference analyses of the 49 patients of this study estimated the median number of clonal mutations in our assays to be 13 (Figure S2). As shown in Figure S6A, ctDNA will be detected in samples with ctDNA between 0.01% and 0.02% for > 98% of samples.

Additionally, the Signatera process was also verified using a commercially available standard control, SeraseqTM ctDNA Mutation Mix v2 (SeraCare, Milford, MA). Allele fractions for all targets in the mix were verified by SeraCare by droplet digital PCR. Two 16-plex pools were designed for the SeraCare standard using the Signatera pipeline. As for the analytical validation, the SeraCare standard was titrated at 0.5%, 0.2%, 0.1%, 0.05%, 0.03%, 0.01%, and 0.005% VAF (Figure S6B). Each titration point was run in triplicate and run on both 16-plex pools. Five negative control samples run on both 16-plex pools. 66 ng cfDNA was used as input into the Signatera plasma workflow.

Accuracy of 100% was achieved by calling all 36 positive samples as ctDNA positive with VAF down to 0.01%, and all 10 negative samples as ctDNA negative. Target-level sensitivity of >99% achieved at mutations frequencies down to 0.1%, similar to targets tested in cell-line samples (Figure S6C). There was only one target above the calling threshold in control samples.

Target and sample-level performance estimated based on cell-line derived samples were corroborated using this commercially available SNV standard control (Figures S6A-C). Figure S6D shows a titration curve from this standard for a target that was present in six of the breast cancer cases in this study. This target was called down to below 0.05% in the clinical samples which is firmly within the quantitative detection range for this target (detection down to 0.005%, quantitative detection down to 0.03%).

3

Online-Only Tables

TABLE S1: Clinical characteristics of all 49 patients

Case No

Age at diagnosis

Tumour Grade

ER Status

HER2 status

TNM staging

Time from first sample to relapse (days)

Date of relapse

Site of distant metastasis

Local Recurrence

E001

48

3

-

+

T2 N1 M0

-

NA

NA

E002

56

3

+

+

T2 N2 M0

-

NA

NA

E003

65

2

+

-

T3 N2 M0

133

25/06/2014

Pleural Effusion

E004

50

2

+

-

T2 N3 M0

-

NA

NA

E005

69

2

+

-

T3 N3 M0

263

02/01/2015

Nodal disease right hilum and mediastinum

E006

48

2

+

-

T2 N2 M0

-

NA

NA

E007

67

2

+

+

T2 N3 M0

-

NA

NA

E008

47

2

+

-

T3 N3 M0

-

NA

NA

E009

44

3

+

-

T2 N1 M0

125

28/10/2014

Sternum, pelvis, vertebrae*

E010

38

2

+

-

T2 N2 M0

561

08/01/2016

NA

Sternum

E011

69

2

+

-

T2 N2 M0

-

NA

NA

E012

66

3

+

-

T2 N1 M0

-

NA

NA

E013

80

2

+

-

T3 N2 M0

-

NA

NA

E014

57

2

+

-

T2 N3 M0

-

NA

NA

E015

81

3

-

-

T2 N1 M0

-

NA

NA

E016

80

3

-

+

T2 N2 M0

-

NA

NA

E017

67

3

+

-

T3 N3 M0

1281

31/01/2018

Sternoclavicular joint, skin, lung

E018

55

2

+

-

T3 N3 M0

-

NA

NA

E019

53

3

+

-

T2 N1 M0

-

NA

NA

Case No

Age at diagnosis

Tumour Grade

ER Status

HER2 status

TNM staging

Time from first sample to relapse (days)

Date of relapse

Site of distant metastasis

Local Recurrence

E020

80

2

+

-

T2 N2 M0

-

NA

NA

E021

44

2

+

-

T2 N3 M0

-

NA

NA

E022

54

2

+

-

T2 N2 M0

-

NA

NA

New breast primary in contralateral breast

E023

54

2

+

-

T3 N2 M0

-

NA

NA

E024

78

3

-

-

T2 N1 M0

-

NA

NA

E025

46

3

-

-

T1 N2 M0

-

NA

NA

E026

73

2

+

-

T2 N3 M0

793

17/03/2017

Spine

E027

60

2

+

-

T2 N2 M0

-

NA

NA

E028

76

3

+

+

T1 N2 M0

-

NA

NA

E029

57

3

-

-

T2 N1 M0

442

30/11/2016

Lung

E030

70

3

+

-

T2 N1 M0

-

NA

NA

E031

77

2

+

-

T2 N3 M0

1087

23/05/2018

Skin on right lower back

E032

39

3

+

-

T1 N1 M0

-

NA

NA

E033

57

2

-

-

T1 N0 M0

570

30/12/2016

Intraclavicular fossa and sentinel lymph nodes

E034

66

2

+

-

T1 N3 M0

-

NA

NA

E035

59

2

+

-

T2 N3 M0

-

NA

NA

E036

46

2

+

-

T1 N3 M0

415

13/10/2016

Bone and bladder

E037

78

3

-

+

T2 N3 M0

610

07/06/2017

Lung

E038

63

1

+

-

T2 N2 M0

-

NA

NA

E039

48

3

+

-

T2 N1 M0

-

NA

NA

E040

55

3

-

+

T2 N1 M0

313

05/09/2016

Bone

Case No

Age at diagnosis

Tumour Grade

ER Status

HER2 status

TNM staging

Time from first sample to relapse (days)

Date of relapse

Site of distant metastasis

Local Recurrence

E041

67

3

+

-

T2 N1 M0

-

NA

NA

E042

49

2

+

-

T2 N1 M0

-

NA

NA

E043

41

2

+

-

T3 N1 M0

68

23/05/2016

Liver, Lung, Bone*

E044

46

3

+

-

T2 N1 M0

323

04/02/2016

Local nodes

E045

46

3

+

+

T2 N1 M0

-

NA

NA

E046

57

3

-

-

T2 N2 M0

269

08/02/2016

Bone, liver and pleura

E047

62

3

+

-

T3 N1 M0

117

18/11/2015

Local nodes and intrapulmonal nodes

E048

52

3

-

-

T2 N1 M0

202

31/12/2015

Bone and pleura

E049

45

3

-

-

T2 N1 M0

148

08/10/2016

Not known*

NA, not applicable. Patients marked with an asterisk (*) are deceased.

Table S2 (A-D) Sample visit information for 49 patients including sample visit and treatment

Table S2A: Dates of sample visits for 49 patients

Case ID

Plasma Timepoint (date)

0

1

2

3

4

5

6

7

EoS

E001

15/01/2014

08/10/2014

07/01/2015

24/06/2015

20/01/2016

13/07/2016

04/01/2017

05/07/2017

 

E002

10/02/2014

08/09/2014

09/02/2015

01/09/2015*

18/02/2016

25/08/2016

16/03/2017

 

 

E003

12/02/2014

 

 

 

 

 

 

 

 

E004

14/04/2014

27/10/2014

27/04/2015

19/10/2015

09/06/2016

28/11/2016

 

 

 

E005

14/04/2014

16/10/2014

 

 

 

 

 

 

 

E006

07/05/2014

 

 

09/12/2015*

23/08/2016

01/03/2017

 

 

 

E007

12/05/2014

11/11/2014

 

17/11/2015

26/05/2016

08/12/2016

16/07/2017

 

 

E008

16/05/2014

03/11/2014

20/04/2015

19/10/2015

12/12/2016

 

 

 

E009

25/06/2014

 

 

 

 

 

 

 

 

E010

26/06/2014

10/12/2014

 

 

 

 

 

20/01/2016

E011

09/07/2014

28/01/2015

19/08/2015

06/01/2016

20/07/2016

25/01/2017

06/09/2017

 

 

E012

16/07/2014

14/01/2015

08/07/2015

27/01/2016

27/07/2016

25/01/2017

26/07/2017

 

 

E013

16/07/2014

14/01/2015

15/07/2015

13/01/2016

13/07/2016

25/01/2017

09/08/2017

 

 

E014

16/07/2014

21/01/2015

03/02/2016

12/07/2016

31/01/2017

 

 

 

E015

21/07/2014

14/01/2015

15/07/2015

20/01/2016

20/07/2016

08/02/2017

26/07/2017

 

 

E016

23/07/2014

21/01/2015

29/10/2015

08/06/2016

11/01/2017

05/07/2017

 

 

 

E017

30/07/2014

14/01/2015

08/07/2015

10/02/2016

31/08/2016

22/02/2017

04/07/2017

 

 

E018

11/08/2014

16/02/2015

04/04/2016*

28/11/2016

24/05/2017

 

 

 

E019

27/08/2014

11/02/2015

06/08/2015

16/03/2016

18/10/2016

21/04/2017

 

 

 

E020

04/09/2014

20/04/2015

15/10/2015

14/01/2016

03/11/2016

03/05/2017

 

 

 

E021

09/09/2014

10/03/2015

15/09/2015

24/03/2016

22/09/2016

30/03/2017

 

 

 

E022

10/09/2014

11/03/2015

04/08/2015

 

 

 

 

 

 

E023

11/09/2014

05/02/2015

03/08/2015

08/03/2016

25/08/2016

09/03/2017

 

 

 

E024

24/09/2014

18/03/2015

23/09/2015

18/05/2016

30/11/2016

28/06/2017

 

 

 

E025

05/11/2014

13/05/2015

23/12/2015

22/06/2016

18/01/2017

06/09/2017

 

 

 

E026

14/01/2015

15/07/2015

24/02/2016

13/09/2016

28/02/2017

 

 

 

 

E027

03/02/2015

04/08/2015

04/02/2016

 

 

 

 

 

 

E028

11/02/2015

30/08/2016

21/02/2017

05/09/2017

 

 

 

E029

15/09/2015

17/03/2016

29/09/2016

 

 

 

 

 

12/01/2017

E030

20/05/2015

18/11/2015

18/05/2016

 

19/07/2017

 

 

 

 

E031

01/06/2015

29/10/2015

28/04/2016

26/01/2017

26/07/2017

 

 

 

 

Case ID

Plasma Timepoint (date)

0

1

2

3

4

5

6

7

EoS

E032

03/06/2015

14/10/2015

08/06/2016

21/12/2016

 

 

 

 

 

E033

09/06/2015

22/12/2015

21/07/2016

 

 

 

 

 

09/01/2017

E034

10/06/2015

09/12/2015*

07/06/2016

 

04/07/2017

 

 

 

 

E035

18/06/2015

21/01/2016

09/09/2016

30/01/2017

12/06/2017

 

 

 

 

E036

25/08/2015

24/03/2016*

22/09/2016

 

 

 

 

 

 

E037

06/10/2015

24/05/2017#

 

 

 

 

24/05/2017#

E038

07/10/2015

13/04/2016

19/10/2016*

19/04/2017

 

 

 

 

 

E039

20/10/2015

18/05/2016

16/11/2016

31/05/2017

 

 

 

 

 

E040

28/10/2015

13/04/2016

 

 

 

 

 

19/10/2016

E041

20/01/2016

20/07/2016

08/02/2017

04/08/2017

 

 

 

 

 

E042

03/02/2016

16/08/2016

24/01/2017

 

 

 

 

 

 

E043

16/03/2016

 

 

 

 

 

 

 

 

E044

18/03/2015

23/09/2015*

 

 

 

 

 

17/02/2016

E045

15/12/2015

15/06/2016

14/12/2016

 

 

 

 

 

 

E046

15/05/2015

06/11/2015

12/02/2016

 

 

 

 

 

 

E047

24/07/2015

16/12/2016

 

 

 

 

 

 

 

E048

12/06/2015

04/12/2015

 

 

 

 

 

08/01/2016

E049

13/05/2016

 

 

 

 

 

 

 

 

*Samples omitted due to failed QC.

#Duplicate plasma timepoint

EoS, End of Study.

Table S2B: CA15-3 levels (U/ml) at sample visit

Case ID

CA15-3 levels at plasma timepoint

0

1

2

3

4

5

6

7

EoS

E001

7

20

19

22

20

18

17

20

E002

8

9

8

10

9

9

7

E003

17

E004

18

18

18

18

19

17

E005

22

26

E006

15

11*

14*

11

13

13

E007

15

14

14

13

14

14

E008

17

17

18

17

21*

15

E009

51

E010

13

12

16

E011

9

8

7

8

9

9

ND

E012

9

8

9

8

10

9

10

E013

31

33

23

21

19

57

28

E014

18

17

18

18

20

E015

20

20

22

22

20

23

20

E016

15

13

15

17

16

15

E017

13

13

13

14

15

17

17

E018

5

6

5

7

7

E019

29

23

29

26

29

29

E020

26

51

32

28

29

36

E021

14

14

14

14

14

14

E022

10

10

12

E023

11

13

13

12

13

14

E024

23

23

28

24

25

22

E025

20

20

20

17

19

21

E026

24

31

46

58

104

E027

24

23

20

E028

35

33

29

E029

10

13

10

12

E030

13

13

10

14

E031

34

27

31

32

ND

E032

12

10

11

11

E033

18

21

20

22

E034

26

25

23

27

E035

21

19

24

22

21

E036

25

48

131

E037

20

22

21

E038

16

17

19

16

E039

14

12

11

13

E040

33

35

36

E041

26

23

26

22

E042

5

5

5

E043

65

E044

ND

ND

ND

E045

ND

ND

ND

E046

ND

ND

ND

E047

ND

ND

E048

ND

ND

ND

E049

ND

*No plasma samples.

EoS, End of Study; ND, Not Determined.

Table S2C: Details of patient’s tests and scans

Case ID

Tests/Scans showing lack of metastasis (date)

E001

CT - 12/16, Bone scan - 01/18, LFTs - 07/16, 07/17

E002

CT - 01/18, Bone scan - 03/17, LFTs - 01/16, 03/18

E003

CT - 01/14, Bone scan - 05/13, Liver US - 05/13

E004

CT - 11/15, Liver US - 05/18, LFTs - 05/16, 12/17

E005

LFTs - 04/14

E006

CT - 02/14, Liver US - 05/14, LFTs - 02/14

E007

CT - 02/14, LFTs - 02/14

E008

Bone - 12/13, Liver US - 04/14, LFTs - 05/14

E009

CT - 01/14, Bone scan - 12/13, Chest Xray - 06/14, LFTs - 06/14

E010

Bone scan - 10/13, Chest Xray - 03/14, LFTs - 06/14

E011

CT - 12/14, LFTs - 07/14

E012

Bone scan - 05/15, Chest Xray - 02/15, LFTs - 07/14

E013

Chest Xray - 03/15, Liver US - 07/15, LFTs - 03/14, 11/14, CA15-3 - normal

E014

CT - 03/14, 02/15, LFTs - 03/14, 09/14

E015

No scans but LFTs 07/14 normal

E016

CT - 12/15, Bone scan - 02/15, LFTs - 07/14

E017

CT - 07/14, LFTs - 07/14

E018

CT - 08/14, Bone scan - 08/14, LFTs - 08/14

E019

Chest Xray - 05/14, Liver US - 05/14, LFTs - 08/14

E020

Chest Xray - 06/14, Liver US - 08/15, 03/14, LFTs - 09/14

E021

CT - 07/13, Bone scan - 07/13, LFTs - 09/14

E022

CT - 07/15, Bone scan - 06/15

E023

LFTs - 05/14 (No imaging)

E024

Bone scan - 03/14, LFTs - 09/15

E025

LFTs - 05/15 (no imaging)

E026

Steadily rising CA15-3 since 02/16. LFTs - 06/14, 02/16, Mammogram - 07/15

E027

LFTs - 12/13 (no imaging)

E028

Bone scan - 02/15, LFTs - 02/15

E029

Bone scan - 03/15, 10/16, MRI Bone - 02/15, LFTs - 09/15

E030

Bone scan - 12/15, Chest Xray - 01/14, LFTs - 05/15

E031

CT - 04/17, Bone - 08/17, LFTs - 10/15

E032

Chest Xray - 11/13, Liver US - 10/16, LFTs - transient increase AT (56)

E033

LFTs - 16/15

E034

LFTs - 10/06/15, MRI - 13/11/14

E035

CT - 10/14, Bone scan - 10/14, LFTs - 06/15, MRI - 04/15

E036

CT - 09/13, Bone - 10/13, LFTs - 05/15,09/15

E037

CT - 07/15, Bone scan - 07/15, LFTs - 06/16, 7/15

E038

CT - 08/13, LFTs - 10/15

E039

CT - 03/15, Chest Xray - 07/15, LFTs - 10/15

E040

CT - 08/12, Chest Xray - 11/12, LFTs (raised but diabetic and on statins - deemed normal) - 10/15

E041

CT - 09/15, Chest Xray - 09/15, LFTs - 01/16

E042

CT - 08/14, Bone Scan - 08/14, LFTs - 02/16

E043

CT - 11/14, Bone Scan - 11/14, Chest Xray - 15/01/15, LFTs 03/16

E044

No scans but LFTs 03/15 normal

E045

Mammogram 12/15, LFT - 12/15

E046

CT scan excluded mets at baseline

E047

CT scan excluded mets at baseline

E048

CT scan excluded mets at baseline

E049

CT scan excluded mets at baseline

Table S2D: Chemotherapy status and endocrine treatment regime in 49 patients

Case ID

Previous chemotherapy

Plasma time point

0

1

2

3

4

5

6

7

EoS

E001

Yes

None

None

None

None

None

None

None

None

None

E002

Yes

Letrozole

Letrozole

Letrozole

Letrozole

Letrozole

Letrozole

Letrozole

Letrozole

Letrozole

E003

Yes

Letrozole

Letrozole

E004

Yes

Tamoxifen

Tamoxifen

Tamoxifen

Tamoxifen

Tamoxifen

Tamoxifen

E005

No

Anastrozole

Anastrozole

Anastrozole

E006

Yes

Letrozole

Letrozole

Letrozole

Exemestane

Letrozole

Letrozole

E007

Yes

Anastrozole

Tamoxifen

No visit

Tamoxifen

Tamoxifen

Tamoxifen

Tamoxifen

E008

Yes

Anastrozole/Zoladex

Anastrozole/

Zoladex

Anastrazole

Exemestane

Exemestane

Exemestane

E009

Yes

None

E010

Yes

Tamoxifen/

Zoladex

Tamoxifen/

Zoladex

E011

Yes

Letrozole

Letrozole

Letrozole

Letrozole

Letrozole

Letrozole

E012

Yes

Anastrozole

Anastrozole

Anastrozole

Anastrozole

Anastrozole

Anastrozole

Anastrozole

E013

No

Letrozole

Letrozole

Letrozole

Letrozole

Letrozole

Letrozole

Letrozole

E014

Yes

Letrozole

Letrozole

Missed

Letrozole

Letrozole

Letrozole

E015

No

None

None

None

None

None

None

None

E016

No

Anastrozole

Anastrozole

Anastrozole

Anastrozole

Anastrozole

Anastrozole

E017

Yes

Tamoxifen

Tamoxifen

Tamoxifen

Tamoxifen

Tamoxifen

Tamoxifen

Tamoxifen

E018

Yes

Exemestane

Exemestane

Missed

None

None

None

E019

Yes

None

Tamoxifen

Tamoxifen

Tamoxifen

Letrozole

Letrozole

E020

No

Tamoxifen

Tamoxifen

Tamoxifen

Tamoxifen

Tamoxifen

Tamoxifen

E021

Yes

Tamoxifen

Tamoxifen

Tamoxifen

Tamoxifen

Tamoxifen

Tamoxifen

E022

Yes

Anastrozole

Anastrozole

E023

Yes

Letrozole

Letrozole

Letrozole

Letrozole

Letrozole

Letrozole

E024

Yes

None

None

None

None

None

None

E025

Yes

Tamoxifen

Tamoxifen

Tamoxifen

Tamoxifen

Tamoxifen

Tamoxifen

E026

Yes

Letrozole

Letrozole

Letrozole

Letrozole

Letrozole

E027

Yes

Anastrozole

Anastrozole

Anastrozole

E028

Yes

Letrozole

Missed

Missed

Letrozole

Letrozole

Letrozole

E029

Yes

None

None

None

None

Case ID

Previous chemotherapy

Plasma time point

0

1

2

3

4

5

6

7

EoS

E030

Yes

Letrozole

Letrozole

Letrozole

Letrozole

E031

No

Anastrozole

Anastrozole

Anastrozole

Anastrozole

Anastrozole

E032

Yes

None

None

None

None

E033

Yes

None

None

None

E034

Yes

Letrozole

Letrozole

Letrozole

Letrozole

Letrozole

E035

Yes

Anastrozole

Anastrozole

Anastrozole

Anastrozole

Anastrozole

E036

Yes

Tamoxifen

Tamoxifen

Tamoxifen

E037

No

None

None

E038

Yes

Letrozole

Letrozole

Letrozole

Letrozole

E039

Yes

None

None

Letrozole

Letrozole

E040

Yes

Letrozole

Letrozole

E041

Yes

None

Tamoxifen

Tamoxifen

Tamoxifen

E042

Yes

Tamoxifen

Tamoxifen

Tamoxifen

E043

Yes

Tamoxifen

E044

Yes

Tamoxifen

Tamoxifen

Tamoxifen

E045

Yes

Tamoxifen

Not recorded

Not recorded

E046

Yes

None

None

None

E047

Yes

Endocrine therapy*

Endocrine therapy*

E048

Yes

None

None

None

E049

Yes

None

None

None

‘None’ indicates not on any treatment at that time. ‘Missed’ indicates that particular visit was missed by the patient. ‘No visit’ indicates no visit was made by the patient. Endocrine therapy * indicates on endocrine therapy but not specified.

Table S2E: Herceptin Treatment Regimen in HER2+ Patients

Case ID

Herceptin Treatment

Start Date

End Date

E001

02/2013

01/2014

E002

01/2013

01/2014

E007

05/2013

12/2013

E016

11/2013

N/A

E028

05/2014

05/2015

E037

E040

10/2012

10/2013

E045

12/2014

11/2015

N/A, Not Available.

Tables S3a,b and c – please see attached excel files.

Supplemental Figure Legends:

Figure S1: Quality control was performed at every step of the workflow. Of 215 plasma samples, 208 passed the sample QC process, and out of 784 unique assays designed, 767 passed the assay QC (corresponding to a total of 3237 assays passing out of 3328 across all samples). A. The quantity of cfDNA extracted per plasma sample.  Median extracted cfDNA per mL was measured to be 5.5 ng. B. Quantity of input DNA per library preparation. 1 to 66ng of cfDNA (median=25.02 ng) from each plasma sample was used as input into library prep protocol. QC was performed on purified libraries before proceeding to subsequent steps. C. The depth of read per sample. Assays with sequencing coverage less than 5000x were excluded from analyses (denoted in red). Subsequently, samples with less than 8 passing assays failed sequencing coverage QC. The median depth of read for the assays passed coverage QC was 110,000x. D. Sample concordance QC. In order to track sample integrity, SNP tracers were used to measure concordance between patient’s samples. For each plasma samples, a genotyping concordance score was calculated in comparison to the corresponding matched normal genotyping data. Samples were considered to be from the same patients when at least 85% of their SNPs had identical genotypes. Six plasma samples did not pass the sample concordance QC (denoted in red) and were excluded from ctDNA analyses.

Figure S2: Signatera variant selection strategy for the 49 patient-specific panels A. The VAF distribution of custom panel in each patient’s primary tumour tissue. B. The number of inferred clonal (red) and subclonal (blue) variants in each patient’s custom pool. The median number of clonal variants in the 49 custom panels is 13 out of 16. C. The number of inferred clonal (red) and subclonal (blue) variants in each patient’s primary tumor WES data. Different colors represent different subtypes HR+/HER2- (dark blue), and HER2+ (green), and triple negative (orange). Note that clonality inference is affected by tumor purity. Very few to none clonal variants were identified from WES samples with <20% tumor cell fraction (E042, E012, E004).

Figure S3: Detection of Driver Mutations. All somatic variants identified by WES of tumor DNA were submitted to the cancer genome interpreter (https://www.cancergenomeinterpreter.org/home) to detect TIER 1 and TIER 2 driver gene variants. Of the 49 patients, only 33 (67%) had one or more driver mutations (range of 1-11) detected. For these 33 patients, a second assay was designed to track as many driver variants as possible.

Figure S4: Plasma ctDNA levels for 12 breast cancer patients. A-L. Patient-specific plots of 11 relapsed and one non-relapsed patient. Each patient specific assay was designed to target 16 somatic variants using massively parallel sequencing (median depth >100,000X per target). Mean VAFs are denoted by a dark blue circle and the solid line represent average VAF profile over time. The lead time is calculated by the difference in the date of clinical relapse (red triangle) and ctDNA determined molecular relapse (blue triangle). CA 15-3 levels are graphed over time (teal circles).

Figure S5: Distribution of VAFs and mutant counts. In total, 251 targets were detected in ctDNA positive plasma samples. A. The VAF of detected targets ranged from 0.01%–64%, with a median of 0.82%. The observed mutant VAF and total number of DNA molecules in each sample was used to calculate the number of tumor molecules present in the patient’s plasma sample. B. The number of detected mutant molecules in the 251 positive targets ranged from 1–6500 mutant molecules, with a median of 39 molecules.

Figure S6: Analytical validation and orthogonal control results. Titrations of tumor into normal mononucleosomal DNAs (cell line-derived) were made at average VAFs of 1% to 0.005%. Six primer pools were tested with replicate numbers from two to nine (for each pool)—increasing with the dilution factor. In addition, a commercially available control SNV mixture from SeraCare (SeraseqTM ctDNA Mutation Mix v2) was titrated from 0.5% to 0.005%. Starting allele fractions were confirmed by SeraCare by droplet digital PCR. Two primer pools were tested in triplicate on these mixtures. A. Sample-level performance calculated from the analytical validation. Sample-level sensitivity was calculated when at least two mutations are detected from a set of 16 target variants, assuming majority of patient-specific panels have between 10–16 clonal variants (n = the number of target level datapoints used to calculate sample-level data). Reproducibility was calculated as the %CV of the median VAF of positive targets. B. Sample-level performance calculated from orthogonal control samples from SeraCare. C. Target-level sensitivity of the 16-plex assay was >99% at mutation frequencies down to 0.1%. Circles represent the analytical validation data (n=72-562 per data point, and squares represent SeraCare data (n=90 per datapoint). D. Graph of the SeraCare titration series for target PIK3CA chr3:178952085:A>G demonstrating quantitative detection down 0.03% and single molecule detection down to 0.005%. Closed circles represent mutation detected; Open circles represent no mutation detected. This target was present and detected down to below 0.05% in plasma samples of six patients in this cohort.

Figure S1

Figure S2

Figure S3

Figure S4

Figure S5

Figure S6.

Supplemental References

1.Adamo P, Cowley CM, Neal CP, et al. Profiling tumour heterogeneity through circulating tumour DNA in patients with pancreatic cancer. Oncotarget. 2017;8(50):87221-87233.

2.McGranahan N, Favero F, de Bruin EC, Birkbak NJ, Szallasi Z, Swanton C. Clonal status of actionable driver events and the timing of mutational processes in cancer evolution. Sci Transl Med. 2015;7(283):283ra254.

3.Zhang J, Kobert K, Flouri T, Stamatakis A. PEAR: a fast and accurate Illumina Paired-End reAd mergeR. Bioinformatics. 2014;30(5):614-620.

4.Abbosh C, Birkbak NJ, Wilson GA, et al. Phylogenetic ctDNA analysis depicts early-stage lung cancer evolution. Nature. 2017;545(7655):446-451.