Poster 29 NCRI, November 2016
Authors
Analytical performance and validation of an
™
enhanced TAm-Seq circulating tumour DNA
sequencing assay across two laboratories
Introduction
Karen Howarth1, John Calaway2, Sarah Smalley1,2 ,Shannon Blais2,
Andrew Lawson1, Greg Jones2, Davina Gale1, Michelle Pugh1,
Mikidache Madi1, Bradley Durham1, Esther Musgrave-Brown1,
Samuel Woodhouse1, Tim Forshew1, Emma Green1, Vincent Plagnol1,
Nitzan Rosenfeld1 and James Clark 1
1
Inivata Ltd, Li Ka Shing Centre, Robinson Way, Cambridge, CB2 0RE, UK
2
Inivata Inc, 7020 Kit Creek Road, Suite 140, Research Triangle Park, NC 27709, USA
Analytical validation performance
UK
A dilution series of Horizon Tru-Q 7 (Medium, AF
0.06%-1.25%) was used to assess LoD. We successfully
detected >97.66% mutations at >0.5% AF, and 90% of
mutations at 0.25% AF across both laboratories. In
addition, we continued to detect mutations down to
0.06% (Fig 5). To assess specificity, we analysed 136
healthy controls, revealing 2 false positive (FP) results
(which amounts to a per base specificity of 99.99982%).
US
100
75
Sensitivity (%)
Circulating tumour DNA (ctDNA) is rapidly becoming established as an invaluable tool to supplement conventional
biopsies for molecular characterization and monitoring of solid cancers, particularly for cancers such as non-small cell
lung cancer (NSCLC), where tumor tissue is limited or unobtainable. As ctDNA may be a small fraction of total cell-free
DNA (cfDNA), sensitive methods are required for its reliable detection. To overcome these challenges, we have developed
an enhanced tagged amplicon deep sequencing method (enhanced TAm-SeqTM). Here we present analytical validation of
this technology across two clinical laboratories based in the UK and US to support its use in clinical applications.
50
25
Fig. 1. Diagram representing the release of
ctDNA into the bloodstream by a tumor. ctDNA
can be distinguished from other cell-free DNA
of non-cancerous origin by the presence of
cancer-specific mutations.
.0
8
.1
5
−0
Fig 5. Sensitivity of eTAm-Seq in a
dilution series across AF range
(0.0625-1.25%).
0.
06
0.
12
−0
.3
1
0.
25
−0
0.
62
0.
5−
1−
1.
25
.0
8
0.
06
−0
.1
5
0.
12
−0
.3
1
−0
0.
25
0.
62
0.
5−
1−
1.
25
0
Allele fraction (%)
What is enhanced TAm-Seq?
Correlation of eTAm-Seq and dPCR results
A
B
AKT1
ESR1
ALK
Pool &
Sequence
HRAS
FGFR1
NRAS
IDH1
FGFR2
IDH2
PIK3CA
CCND1
FGFR3
KIT
PPP2R1A
CDKN2A
FOXL2
KRAS
Horizon TruQ6, low input
30
20
●
●
●
10
●
5
Fig. 6. Comparison of AFs using
eTAm-Seq (boxplot) & Horizon dPCR
(red). Note, according to Horizon
product literature FGFR2 S252W
expected at 1% AF but 1.9% AF
observed by dPCR.
PDGFRA
BRAF
In analysis of undiluted
samples, eTAm-Seq AFs were
compared to AFs based on
Horizon dPCR data, and show
good correlation (Fig 6).
Allele fraction (%)
Enhanced TAm-Seq (eTAm-SeqTM) combines efficient library preparation and statistically-based analysis algorithms to
identify and quantify cancer mutations. Analysis was performed using a 35-gene panel including cancer hotspots, entire
coding regions and copy number variants (Fig 2).
●
3
2
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
1
Horizon TruQ7, medium input
30
20
●
●
●
●
●
●
10
●
5
3
2
●
●
●
1
PTEN
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
MET
STK11
EGFR
GNAQ
MYC
TP53
ERBB2
GNAS
NFE2L2
CA H
GNA11
PIK3
CTNNB1
Fig 2B. eTAm-Seq panel used in this study.
Hotspot regions
Hotspot regions & CNVs
CNVs
Assay validation
We have performed analytical validation of eTAm-Seq to test specificity, sensitivity, limit of detection (LoD) and
inter-operator variability for detection of SNVs and indels. Analysis was performed using plasma DNA from healthy
controls and Horizon Tru-Q reference cell-line DNA carrying mutations in cancer-related genes at known allele
fractions (AFs). eTAm-Seq was performed across two laboratories by 5 different operators, on different days and
sequenced on different NGS runs. One operator performed eTAm-Seq at both laboratories.
cfDNA quantification of NSCLC patient plasma
Fig 7 shows the allele fraction and spectrum of
mutations identified in 153 patients with Stage II-IV
NSCLC, highlighting that 49.2% of mutations
identified by eTAm-Seq in this patient population
had AF ≤1%, with 36.4% having AF ≤0.5%. These
mutations would routinely be missed using less
sensitive assays. A broad spectrum of mutations
across 21 different genes, including common EGFR
(28.1% of patients) and EGFR resistance mutations
(T790M (13.7%), C797S (1.3%)), were identified. This
combined patient dataset is derived from prospective
studies of NSCLC patients in first and second-line
treatment settings, including an EGFR-positive
cohort (n=55) being monitored for treatment
resistance mutations using ctDNA (Remon J et al
Liquid biopsies for molecular profiling of mutations in
non-small cell lung cancer (NSCLC) patients lacking tissue
samples, AACR 2016, New Orleans, US, April 2016).
A
75
50
14.6%
12.6%
11.3% 11.7%
25
8.1%
7.1%
5.8%
0
< 0.1%
0.1−0.5% 0.5%−1%
1−2%
2−5%
5−10%
10−20%
> 20%
Allele fraction
200,000
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
13.7%
5.9%
QNS
Low
Medium
High
Analytical validation performance
High input (AF 0.25%−0.3125%)
F1
0.7%
U2
A
FR
2
0.7%
FG
TA
3
0.7%
GA
AQ
0.7%
GN
0.7%
IDH
2
FR
3
0.7%
FG
0.7%
GN
AS
E2
L2
1.3%
NF
1
1.3%
CC
ND
1.3%
T
1.3%
ME
1.3%
AF
2%
BR
2%
7S
3C
A
PIK
EN
PT
KN
2A
1
K1
ST
Gene
79
0M
Sample number
FR
_T
200
EG
150
AS
100
KR
50
0
2%
79
3.3%
_C
6.5%
S
8.5%
EG
FR
15%
20
NR
A
4,000
2,000
28.1%
40
NN
B1
8,000
60
CT
16,000
46.4%
2
30,000
Fig 7. Frequency of mutations identified in 153 patients with Stage II-IV
NSCLC patients. (A) At different allele fractions. (B) In different genes.
ER
BB
50,000
B
(out of 153 NSCLC cases)
100,000
0
Summary
Low input (AF 2%−2.5%)
Medium input (AF 1%−1.25%)
• ctDNA has potential as a liquid biopsy to supplement conventional tissue biopsies for non-invasive
molecular stratification, monitoring response to treatment and detection of resistance mutations, and
has particular utility in patients where tissue is limited or unobtainable.
100
75
UK
50
Sensitivity (%)
28.8%
●
●
●
0
25
0
100
• Enhanced TAm-Seq is able to detect and quantify cancer mutations in plasma. Analytical validation
studies presented here show that eTAm-Seq is a highly sensitive method, with identification of >97.66%
point mutations at >0.5% AF across both laboratories and a limit of detection (LoD90) of 0.25%.
• Analytical validation across two clinical laboratories in the UK and US showed good concordance,
demonstrating the robustness of the eTAm-Seq assay.
75
50
US
To assess sensitivity, Horizon Tru-Q 6 (Low input,
undiluted, AF majority at 2%-2.5%) and Tru-Q 7 samples
(Medium, undiluted AF majority at 1%-1.25% & High,
diluted to AF 0.25%-0.3125%) were sheared to ~200bp to
mimic fragmented cfDNA. Dilutions were prepared
using Horizon Tru-Q 0 wild-type DNA as diluent. Fig 4
shows eTAm-Seq has high sensitivity, with 100% (95% CI:
99.01%-100%) detected in low input samples; 99.17%
mutations (95% CI: 97.40%-99.85%) detected in Medium ,
and 95.45% mutations (95% CI: 93.09%-97.18%) detected
in High input samples from the UK validation (top).
Comparable results from the US lab are shown below,
with 99.7% (95% CI: 98.6%-99.98%) detected in low input
samples; 100% mutations (95% CI: 98.97%-100%)
detected in Medium , and 92.71% mutations (95% CI:
90.14%-94.77%) detected in High input samples.
Clinical experience using eTAm-Seq
Number of patients
Fig 3. Graph showing DNA yields from 10 mL EDTA blood draws from
Stage II-IV NSCLC patients (n=214 samples). Colors represent samples
with High, Medium and Low DNA classified as described.
300,000
Amplifiable copies per 10 mL of blood
The rationale for determining DNA input amounts used
in validation experiments was based on dPCR
quantification of cfDNA levels routinely obtained from
NSCLC patients in a 10 mL EDTA blood draw. DNA
yields ranged from 1,000-298,350 Amplifiable Copies/10
mL blood. (Fig 3). Based on this, 3 different input
amounts were used - Low (2000 AC, >95% samples),
Medium (8000 AC, >69% samples) and High (16,000 AC,
>40% samples).
Variant
Number of mutation calls
Exon tiling (88–100% coverage)
EG
FR
Fig 2A. Overview of the TAm-Seq® assay. Adapted
from Forshew et al.; Sci Trans Med; 2012;4(136):136ra68.
RET
CD
Barcoding PCR
MED12
53
Amplicon PCR
GATA3
TP
DNA
CHEK2
1047
R
KRA
S G1
3D
EGF
R G7
19S
BRA
F V6
00E
NRA
S Q6
1L
NRA
S Q6
1H
NRA
S Q6
1R
GNA
11 Q
209L
KRA
S G1
2C
KRA
S Q6
1L
IDH1
R132
H
BRA
F V6
00R
IDH2
R140
Q
KRA
S G1
PDG
2S
FRA
D842
V
IDH2
R172
K
ALK
F117
4L
KIT D
816V
IDH1
R132
C
KRA
S G1
2D
KRA
S G1
2A
KRA
S Q6
1H
KRA
S G1
2V
EGF
R L8
61Q
KRA
S G1
2R
PIK3
CA E
545K
EGF
R T7
90M
PIK3
CA E
542K
NRA
S Q6
1K
FGF
R2 S
252W
BRA
F V6
00G
BRA
F V6
00M
BRA
F V6
00K
EGF
R
L858
EGF
R de
R
l746−
A750
●
25
0
1
2
3
All
1
2
3
All
1
2
3
All
Operator
• Analysis of 153 Stage II-IV NSCLC patients show that eTAm-Seq is able to detect mutations at low allele
fractions that would be routinely missed using less sensitive assays.
• These data demonstrate the analytical validity of the enhanced TAm-Seq assay and support its use in
clinical applications.
Fig 4. eTAm-Seq analysis of High, Medium and Low input DNA by 3
operators in both the UK (top) and US (bottom) labs on different days and
sequencing runs to assess sensitivity and inter-operator variability.
©2016 Inivata Ltd.
www.inivata.com