Avenio Ctdna Analysis Kits Performance Across Illumina Sequencing Platforms
Avenio Ctdna Analysis Kits Performance Across Illumina Sequencing Platforms
Introduction
The AVENIO ctDNA Analysis Kits are verified to produce In this study we compare the performance of the AVENIO
enriched circulating tumor DNA libraries, ready to ctDNA Analysis Kits across three widely available Illumina
be sequenced on the Illumina NextSeq 500 and 550 sequencing platforms: Illumina NextSeq 500, HiSeq 4000,
sequencing platforms. The analysis kits enable the detection and HiSeq 2500. Ninety sequencing libraries, containing
of four mutation classes: single nucleotide variants (SNVs), variants with low allele frequencies, were sequenced using
copy number variants (CNVs), insertions and deletions the three sequencing platforms, and analyzed with the
(Indels), and fusions. Three kits are available to sequence AVENIO Oncology Analysis Software version 1.1.0. The
selected genes of interest (AVENIO Targeted Kit, Expanded resulting sequencing metrics and detected variants were
Kit, and Surveillance Kit), all demonstrating high sensitivity used to evaluate assay performance across the Illumina
and positive predictive value (PPV). sequencing platforms for each of the three AVENIO ctDNA
Analysis Kits.
Methods
DNA from cell lines or cell free DNA (cfDNA) extracted contain low variant allele frequencies (AF), listed in
from plasma donors were blended to create fixed Table 1. Indel samples consisted of blends of multiple
proportions of known variants. Horizon Discovery created cell lines with known insertions and deletions. Normal
customer cell lines and verified them by digital droplet PCR cfDNA from unique plasma donors were used to
to contain multiple key SNVs, fusions, or CNVs. NA12878 determine PPV.
(Coriell Institute) or cfDNA were used to dilute cell lines to
Mutation Class Allele Frequency or Copy Number Mixture Description Number of Samples AVENIO Analysis Kit
Results
Sequencing Metrics Analysis: Key sequencing The expected values for on-target rate, unique depth,
metrics of the sequencing libraries are plotted in Figure 1 uniformity and error rate are achieved by the AVENIO
for each of the Analysis Kits, for each Illumina sequencing ctDNA Analysis kits, at consistent and high performing
platform. Median values are indicated. Metric definitions values. As expected, the smaller Targeted Panel (~80kb)
are as follows: On-Target Rate: the percentage of reads attains greater unique depth than the larger Expanded
in the intended target region. Unique Depth: the median and Surveillance panels (~200kb each), with all panels
read depth after removing duplicated reads. Unique generating depths at very high levels. The error rates
depth is a key metric, indicating the number of original across all sequencing platforms were comparable and all
genomic equivalents recovered through the library prep well below 0.1%. Such low error rate is achieved through
and sequencing process. Uniformity: the percentage of the analysis software’s Integrated Digital Error Suppression
positions whose unique depth fall within two-folds of the (iDES).1 Overall, the Illumina HiSeq 4000 and HiSeq 2500
median unique depth. Error Rate: empirically estimated as sequencing platforms are able to generate sequencing
the number of non-reference over all reference base calls metrics similar to the NextSeq 500/550 system.
at < 5% AF, after all molecular and bioinformatic error
suppression is applied.
80 10000
% Reads On-Target
Unique Depth
8000
60
6000
40
4000
20
2000
0 0
Targeted Kit Expanded Kit Surveillance Kit Targeted Kit Expanded Kit Surveillance Kit
100 0.05
80 0.04
% Error Rate
60 0.03
40 0.02
20 0.01
0 0.00
Targeted Kit Expanded Kit Surveillance Kit Targeted Kit Expanded Kit Surveillance Kit
Indels: 5 indels at 1% AF were evaluated for sensitivity. All sequencing platforms achieved 100% sensitivity, with PPV >99.9%.
Indels
Sensitivity PPV
AVENIO Analysis Kit NextSeq 500 HiSeq 4000 HiSeq 2500 NextSeq 500 HiSeq 4000 HiSeq 2500
Targeted Kit 100% 100% 100% >99.9% >99.9% >99.9%
Expanded Kit 100% 100% 100% >99.9% >99.9% >99.9%
Surveillance Kit 100% 100% 100% >99.9% >99.9% >99.9%
Fusions: Fusion samples were analyzed for the detection of fusions at 1% AF. All panels were evaluated for the detection
of EML4-ALK, RET-CCDC6, and SLC34A2-ROS1. In addition, the Expanded Kit also included a TPM3-NTRK1 fusion. All
fusions were detected in all samples at a sensitivity of 100% across sequencing platforms. Note that although all platforms
yielded good PPV, in this sample set, the Expanded Kit on the HiSeq 2500 system had slightly lower PPV based on
replicates of 3 samples per test condition.
Fusions
Sensitivity PPV
AVENIO Analysis Kit NextSeq 500 HiSeq 4000 HiSeq 2500 NextSeq 500 HiSeq 4000 HiSeq 2500
Targeted Kit 100% 100% 100% >99.9% >99.9% >99.9%
Expanded Kit 100% 100% 100% >99.9% >99.9% 96.8%
Surveillance Kit 100% 100% 100% >99.9% >99.9% >99.9%
CNVs: CNV samples were analyzed for sensitivity with 2.3 copies of MET and EGFR and 2.6 copies of ERBB2. In this
study, when evaluating all three genes, the NextSeq 500 system had slightly higher sensitivity than the HiSeq 4000 and
HiSeq 2500 systems, with a replicate size of 3 samples per condition. Impressively, in this study the assay was able to
detect EGFR and ERBB2 at levels lower than the stated limits of detection of 3.0 and 4.5, respectively for those genes. The
AVENIO Oncology Analysis Software version 1.1.0 is aimed to have a CNV caller with an emphasis on high specificity, and
is able to achieve PPV of >99.9% across all platforms.
Conclusion
In this study, the AVENIO ctDNA Analysis Kits created lines with known mutations. Note, it is important that the
libraries that, when sequenced on multiple Illumina user loads the sequencer at a concentration that yields the
platforms, achieved high sensitivity and PPV for all four instrument’s recommended cluster density. Also, for each
mutation classes. The AVENIO ctDNA Analysis Kits were sequencing platform, the total number of samples loaded
evaluated across three Illumina sequencing platforms, the on the sequencer should be considered. To obtain optimal
NextSeq 500, HiSeq 4000, and HiSeq 2500. The expected mutation detection sensitivity, aim for each sample to
sequencing metrics of on-target rate, unique depth, receive sequencing coverage of 40 million paired-end reads
uniformity, and error rate were achieved across all three (20 million clusters) for the Targeted Panel and 60 million
platforms in an equivalent manner, extending the use of paired-end reads (30 million clusters) for the Expanded
the AVENIO ctDNA Analysis Kits to additional sequencing and Surveillance Panels. In conclusion the AVENIO ctDNA
platforms. This study serves as an example for the type of Analysis Kits achieved similar high performance on all
results achievable with the AVENIO ctDNA Analysis Kits, three platforms.
using a blend of samples consisting of cfDNA and cell
1. N
ewman A, Lovejoy A, Klass D et al. Integrated digital error suppression for improved detection
of circulating tumor DNA. Nat Biotechnol. 2016; 34(5):547-555. doi:10.1038/nbt.3520.
Published by:
sequencing.roche.com/ctdna
© 2018 Roche Sequencing Solutions, Inc. All rights reserved. SEQ100355 10/2018