deeplex-myctb, an all-in-one test for diagnosis and ...€¦ · rrl, rplc etha, inha, fabg1...
TRANSCRIPT
15-10-29 CONFIDENTIAL 15-10-29 CONFIDENTIAL
Genoscreen 1, rue du Professeur Calmette 59000 LILLE
Phone: +33 (0)3 20 87 71 53 - Fax : +33 (0)3 3 20 87 72 64 www.genoscreen.com - [email protected]
Deeplex-MycTB, an all-in-one test for diagnosis and surveillance of drug-resistant TB
New Diagnostics Working Group Meeting 2016
Liverpool, October, 26, 2016
• Estimated 480,000 new MDR-TB cases, incl. 46,000 XDR-TB cases; ~190,000 deaths
• Less than 30% diagnosed; less than 50% of those diagnosed receive appropriate treatment
MDR- and XDR- tuberculosis in 2014
“ … serious detection and treatment gaps remain.
Intensified efforts to close these gaps are urgently
needed.”
WGS of 3651 MTBC isolates (940 by Genoscreen)
Training set of 2099 genomes correlated with phenotypes: • 120 mutations as resistance determinant and 772 as benign in
14/23 candidate genes
Validation set of 1552 genomes using this catalog : • 89,2 % predicted phenotypes, 92,3% sensitivity and 98,4% specificity • Superior sensitivity compared to 3 line probe assays
Resistance determinants in gene targets
• WGS promising for prediction of drug susceptibility and resistance • Determinants concentrated in 14 candidate genes
Multicentric study on WGS from newly positive culture: • Prediction of species and drug susceptibility
with 93% accuracy • Full WGS Dx, incl. genetic relatedness, median
21 days faster than classical Dx • 7% less costly annually than current workflow
• Primary culture still needed : delayed Dx • WGS on clinical samples: low multiplexing/coverage
depth or capture system (Brown, JCM, 2015) not cost-effective (yet)
Deeplex®-MycTB, an all-in-one NGS-based diagnostic test for M. tuberculosis
DNA extraction
Multiplex PCR
Library
preparation
Deep sequencing
Secured cloud-based
analysis
Visualisation,
reporting
*Photo credit: WHO, The Natural History of Pulmonary Tuberculosis, Facilitator Guide’, 2001
• Targeted NGS of single 24-plex amplification of main resistance targets, plus species
identification and MTBC genotyping targets
• Deep sequencing for sensitive detection of heteroresistance
• Scalable up to 50 samples per MiSeq run, 70 samples per Miniseq run
• Fast, easy-to-use NGS data analysis and reporting on highly secured, high performance cloud
36-48h turnaround time, 4h hands-on time
Clinical sample
Rifampicin Fluoroquinolones
Pyrazinamide
Isoniazid
Ethambutol
Linezolid
Kanamycin
Capreomycin
Amikacin
Streptomycin
Bedaquiline, Clofazimine
Ethionamide
eis
ahpC, fabG1, katG, inhA
embB
pncA
rpoB gyrA, gyrB
gidB, rpsL
rrs
tlyA
Mycobacterial species
Spoligotyping CRISPR/DR region
hsp65
rv0678
rrl, rplC
ethA, inhA, fabG1
First-line drugs Second-line drugs
New 2nd line drugs for MTBC
Total targets >13,000bp
Deeplex®-MycTB: targets
Species identification
MTBC genotyping
+ Phylogenetic SNPs Underlined: full genes + Internal DNA control
Deeplex®-MycTB analysis
Cloud-based analysis
Visualisation,
reporting
Calibrated pipeline
Metagenomics
Mycobacterium hsp65 DB
(Dai et al., 2011) Spoligotyping DB + phylogenetic SNPs
Resistance-associated mutation
DBs
• PhyResSE, Feuerriegel et al. 2014 • Walker et al., 2015 • Miotto et al., 2015 • ReSeqTB, CRyPTIC (when
accessible)
Low frequency variant detection, depending on target position and coverage depth
Mycobacterial & non-mycobacterial species
M. tuberculosis
MTBC
R. mucilaginosa
Gordoniaceae
• SITVITWEB, Demay, 2012 • PhyResSE, Feuerriegel, 2014
Deeplex®-MycTB: Target coverage depth and detection limit
0
2000
4000
6000
8000
10000
12000
rrl
eis
emb
B
eth
A
gid
B
gyrA
gyrB
inh
A
fab
G1
katG
pn
cA
rplC
rpo
B
rpsL rrs
tlyA
ahp
C
Rv0
67
8
hsp
65
1ng – 105 genomes
0
2000
4000
6000
8000
10000
12000
rrl
eis
emb
B
eth
A
gid
B
gyrA
gyrB
inh
A
fab
G1
katG
pn
cA
rplC
rpo
B
rpsL rrs
tlyA
ahp
C
Rv0
67
8
hsp
65
100pg – 104 genomes
0
2000
4000
6000
8000
10000
12000
rrl
eis
emb
B
eth
A
gid
B
gyrA
gyrB
inh
A
fab
G1
katG
pn
cA
rplC
rpo
B
rpsL rrs
tlyA
ahp
C
Rv0
67
8
hsp
65
10pg – 103 genomes
0
2000
4000
6000
8000
10000
12000
rrl
eis
emb
B
eth
A
gid
B
gyrA
gyrB
inh
A
fab
G1
katG
pn
cA
rplC
rpo
B
rpsL rrs
tlyA
ahp
C
Rv0
67
8
hsp
65
1pg – 102 genomes
0
2000
4000
6000
8000
10000
12000
rrl
eis
emb
B
eth
A
gid
B
gyrA
gyrB
inh
A
fab
G1
katG
pn
cA
rplC
rpo
B
rpsL rrs
tlyA
ahp
C
Rv0
67
8
hsp
65
100fg – 101 genomes
0
2000
4000
6000
8000
10000
12000
rrl
eis
emb
B
eth
A
gid
B
gyrA
gyrB
inh
A
fab
G1
katG
pn
cA
rplC
rpo
B
rpsL rrs
tlyA
ahp
C
Rv0
67
8
hsp
65
10fg – 100 genomes
On-target average coverage depths of 100x-1000x, down to 100-10 genomes
Microscopy detection
0
10
20
30
40
50
60
70
80
90
100
1 16 31 46 61 76 91 106121136151166181196211226241256271286301316331346361376391
0
1000
2000
3000
4000
5000
6000
7000
8000
9000
Cover
age
dep
th
Deeplex®-MycTB: smear neg/culture pos MTBC sample
Drug resistance targets
hsp65 sequence: full MTB match
MTBC strain spoligotype: ST53-T1, Euro-American
Co
ver
age
dep
th
% m
atch
vs
MT
B t
ype
sequen
ce
hsp65 sequence position
MTBC
Int. Ctrl
Mycobacterium
Metagenomics: MTBC
0
0.5
1
1.5
2
2.5
3
3.5
4
4.5
5
0 50 100 150 200 250 300 350 400
Deeplex®-MycTB: variant detection level
rpoB
% 1
st v
aria
nt
nucleotide position
threshold
0
20
40
60
80
100
120
0 50 100 150 200 250 300 350 400 450
rpoB
0
0,5
1
1,5
2
2,5
3
3,5
4
4,5
5
0 50 100 150 200 250 300 350 400 450
Zoom in
Hithicked
variant
Probable
mixed
strains
Known
determinants
New predicted
determinants
• Calibration by e.g. analysis of 220 drug susceptible and MDR isolates, plus internal controls
• Background nucleotide variation, incl. amplification and sequencing errors plus background biological
variation below 2% except in few positions
• Likely sufficient in many cases as 1% phenotypically resistant might be an underestimate due to
frequent counter-selection of resistant in culture due to fitness cost
In South African MDR strains not in Xpert
RRDR
nucleotide position
zoom in
Deeplex®-MycTB: semi-quantitative detection of variants
Mu
tant
vs
wt
pro
port
ion
Mixtures of Asp94Gly:wt in gyrA (coll. ITG, Antwerp) in triplicated Deeplex tests
0
10
20
30
40
50
60
70
80
90
100
R100S0 R50S50 R40S60 R30S70 R20S80 R10S90 R5S95 R1S99 R0S100 H37Rv0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
R1S99 R0S100 H37Rv
Zoom in
Semi-quantitative detection down to close to 1%
Deeplex®-MycTB evaluation and beta-testing
• In-house evaluation on clinical specimens:
• Effective on most smear-pos and on ≈25 % smear-neg, culture-pos specimens
• Evaluation of sensitivity and specificity on WHO-TDR MTBC strain bank:
• 229 MTBC samples of diverse phenotypic and genetic resistance profiles and geographic origins
• Previously used to evaluate the CEPHEID GeneXpert MTB/RIF
• 550 clinical samples from WHO drug resistance surveys in Djibouti and DRC (coll. WHO Supranational
Reference Labs D. Cirillo, Milano, and B. De Jong, Antwerp)
Deeplex®-MycTB evaluation and beta-testing
• Preliminary Deeplex results on DNA extracts from clinical samples from Djibouti:
• Analysis done on equivalent of 100 µl of clinical sample (vs 500-1,000 µl by Xpert)
• 270/300 samples with sufficient on-target coverage on first run (7-9/49 invalid by Hain tests)
• 15 predicted MDR-TB
• Among 263 Rif –S by Xpert, only one discordance (with RpoB Ser531Leu, Ile561Val) among
samples with Deeplex results
• Among 36 Rif-R by Xpert, 27 confirmed and 7 Rif-S by Deeplex (2 with mutations but conflictive
prediction depending on allele calling DB) confirmed by Sanger sequencing and Hain tests
Deeplex®-MycTB deployment and use
• Application on high performance, highly secured cloud-based platform, by early 2017
• End-to-end integration from sequencers to computing centers to end-users PC/tablet
(analysis/report) and smartphone (report)
• Private supercomputer datacenters, with local data processing in major countries in
Europe, Asia and US; fully compliant with European and US regulation
• Target: max. 1 hour/NGS run
• Profiled as first-line diagnostic test, but depending on drug resistance incidence, may be used
after e.g. Xpert-based triage
• High-performance test most appropriate in regional, centralized laboratories
EDCTP2 H2020 “DIAMA” Clinical Trial (2016 on):
• 9 African + 4 EU partners
• Deeplex MycTB vs phenotypic, SNP-based molecular tests and patient outcomes
Prospective Deeplex®-MycTB evaluation
• Cyril Gaudin • Caroline Allix-Beguec • Stéphanie Duthoy • Yannick Laurent • Mathilde Mairey • Mihanta Ramaroson • Philip Supply
Acknowledgements
• Vanessa Mathys
• Nadine Lemaitre • Guy Delcroix
• Bouke de Jong • Leen Rigouts • Michèle Driesen
Research Center Borstel
• Silke Feuerriegel • Stefan Niemann
• Timothy Walker
• Louise Pankhurst
• Tim Peto and Derrick Crook
• Daniela Cirillo • Andrea Cabibbe • Elisa Tagliani
0
20000
40000
60000
80000
100000
120000
ahp
C CI
eis
emb
B
eth
A
fab
G1
gid
B
gyrA
gyrB
hsp
65
inh
A
katG
pn
cA
rplC
rpo
B
rpsL
rrl3
rrs1
rrs2
Rv0
67
8
tlyA
Cover
age
dep
th
internal control
0
5000
10000
15000
20000
25000
30000
35000
40000
ahp
C CI
eis
emb
B
eth
A
fab
G1
gid
B
gyrA
gyrB
hsp
65
inh
A
katG
pn
cA
rplC
rpo
B
rpsL
rrl3
rrs1
rrs2
Rv0
67
8
tlyA
0
1000
2000
3000
4000
5000
6000
7000
8000
9000
ahp
C CI
eis
emb
B
eth
A
fab
G1
gid
B
gyrA
gyrB
hsp
65
inh
A
katG
pn
cA
rplC
rpo
B
rpsL
rrl3
rrs1
rrs2
Rv0
67
8
tlyA
0
10
20
30
40
50
60
70
80
90
100
ahp
C CI
eis
emb
B
eth
A
fab
G1
gid
B
gyrA
gyrB
hsp
65
inh
A
katG
pn
cA
rplC
rpo
B
rpsL
rrl3
rrs1
rrs2
Rv0
67
8
tlyA
NTM+ sample
internal control
MTBC+ sample
Cover
age
dep
th
Deeplex®-MycTB: resistance target coverage depth
Species identification
• Target coverage depth of several 100-folds to several 1000-folds • Bacterial load dependent