Karyotype-level noninvasive prenatal testing by sequencing of circulating cell-free DNA from maternal plasma

Issue link:

Contents of this Issue


Page 0 of 0

Genomic Location (Mb) Normalized Bin Count Event Size Sensitivity CHR EVENTS EVENT SIZE (MB) N 28M REadS /SaMpLE 96M REadS /SaMpLE 9 p24.3p13.1 Duplication 38.6 1 22 q12.1q12.3 Deletion 7.9 1 5 Trisomy 5 180.7 1 7 Trisomy 7 159.0 4 11 Trisomy 1 1 134.8 1 22 Trisomy 22 33.9 3 Number of Events percentage 31-41513R1.0 0515 INtroductIoN Noninvasive prenatal testing (NIPT) has become an accepted method for the detection of trisomy 21, 18 and 13 in patients at high risk to carry a fetus with one of these chromosomal abnormalities. 1-3 NIPT can also be used to detect other aneuploidies and sub-chromosomal copy number variations. 4-5 With the inevitable progress of sequencing technologies, it is likely that sequencing-based NIPT will offer more information. We present studies that support the concept of NIPT delivering genome-wide unbalanced copy number results, analogous to what can be detected by cytogenetic G-band karyotyping. Methods theoretical Model: A series of simulations using z-scores was performed to determine the theoretical limits of sensitivity in detecting a sub-chromosomal aberration of a given size at a fixed specificity and for a given number of reads that are independently sampled from the genome. The model assumes an a-priori knowledge of the aberration location. An aberration is detected if the absolute z-score is ≥ 3.95. To determine the proportion of reads deleted, we used a fetal fraction distribution based on historic measurements in maternal plasma. The only source of variability in this simulation was Poisson sampling (no biological or technical variability is considered). In Silico spike-In Model: 2,475 unique sub chromosomal abnormalities ≥ 1Mb from the International Society of Cytogenetic Array (ISCA) database were used for in silico spike-in simulations. These events were spiked into 1070 maternal plasma samples with sequencing read of ~28M reads/sample to obtain 2.6M events. The spike-in was achieved by deletion or addition of RAW bin-counts with an amplitude proportional to autosomal fetal DNA fraction. A simulated event then was subjected to a signal processing algorithm for detection without a priori knowledge of the event location. The event was classified as positive if the following conditions were met: |Z| ≥ 3.95, log ( P ( copy-number <> 2 | amplitude) / P (copy-number=2))>0, and a boot strap confidence of ≥ 0.99. Genomic dNA Analytical Model: 163 genomic DNAs (gDNAs) were confirmed to have 221 subchromosomal aberrations ≥1.25 Mb by microarray and karyotype. DNA was sonicated and fractionated in order to generate fragments between 120-180 bp. These DNA fragments were then used as a template for library preparation. gDNA libraries were then diluted with non-pregnant female plasma libraries to model fetal fractions between 4-15%. Pure gDNA libraries and mixed libraries were then sequenced on a HiSeq2000 or 2500, targeting a read depth of approximately 32M reads/sample. Maternal Plasma: A total of 1 152 clinical plasma samples were collected using IRB-approved clinical protocols. Birth outcome was available for 969 low risk samples and karyotype was available for 183 high risk samples. DNA extraction, library preparation, sequencing, and data normalization were performed as described by Zhao et al. 5 Fetal fraction was determined by a genome-wide sequencing count-based approach. 6 Fetal subchromosomal abnormalities were confirmed by higher coverage sequencing at 96M reads/sample. Amin R. Mazloom 1 , Tong Liu 1 , Roy Leowitz 1 , Yijin Wu 1 , Grant Hogg 1 , John Tynan 1 , Timothy Burcham 2 , Dirk van den Boom 2 , Mathias Ehrich 2 1 Sequenom Laboratories™, San Diego, California, 92121, 2 Sequenom, Inc., San Diego, California, 92121 resuLts coNcLusIoNs • Using genome-wide sequencing results from plasma DNA, we were able to model in silico genome-wide events. These simulations predict >90% sensitivity for ≥7 Mb and >95% for events ≥10 Mb. • Using high coverage sequencing for confirmation, specificity was determined to be >99.9% (95% CI 99.5-100%). Karyotype-level noninvasive prenatal testing by sequencing of circulating cell-free dNA from maternal plasma Figure 4. Determination of specificity for detection of subchromosomal abnormalities in maternal plasma sequenced at 28M reads/sample. • Both CVS testing and NIPT analyze fetal DNA derived from the placenta that can have confined placental mosaicism in 1% of cases. 7 • To determine specificity, sequencing results averaging ~28M reads/sample from 1064 maternal plasma samples were analyzed. • 11 of 11 detected events ≥7 Mb were confirmed by higher coverage sequencing at ~96M reads/sample, resulting in a >99.9% (95% CI 99.5-100%) relative specificity. Figure 3. Detection of a 14.7 Mb deletion at varying levels of fetal fraction. • Example of subchromosomal deletion detection in gDNA mixture models. This sample has a 5p15.31p14.3 deletion according to microarray. • Subchromosomal events, highlighted in red, are first identified by circular binary segmentation. 5 A decision tree algorithm is then used to classify events as positive events (|z-score|>3.95, LOR>0). • Ideograms from sequencing confirm a deletion in 5p15.31p14.3 in close agreement with the array finding. Figure 5. Examples of abnormalities detected by sequencing of maternal plasma. • Ideograms for a 7.9 Mb deletion on Chr22 (Top) and 50kb whole genome traces for a trisomy 7 (Bottom). Plots are shown for both 28M and 92M reads/sample. • The trisomy 7 fetal fraction is 3% based on the event z-score, and is likely mosaic as the sample fetal fraction was measured at 13.7%. Specificity for detection in Maternal plasma detection in a Mixture Model Examples of Events detected by Sequencing in Maternal plasma refereNces 1. Palomaki GE, et al. DNA sequencing of maternal plasma to detect Down syndrome: an international clinical validation study. Genet Med, 2011. 13(11):913-920 2. Jensen TJ, et al. High-throughput massively parallel sequencing for fetal aneuploidy detection from maternal plasma. PLoS One, 2013. 8(3):e57381. 3. Palomaki GE, et al. DNA sequencing of maternal plasma reliably identifies trisomy 18 and trisomy 13 as well as Down syndrome: an international collaborative study. Genet Med, 2012. 14(3):296-305. 4. Mazloom A, et al. Noninvasive prenatal detection of sex chromosome aneuploidies by sequencing circulating cell-free DNA from maternal plasma. Prenatal Diagn, 2013. 33(6):591-597. 5. Zhao C, et al. Detection of fetal subchromosomal abnormalities by sequencing circulating cell-free DNA from maternal plasma. Clin Chem, 2015. 61(4),608-616. 6. Kim SK, et al. Determination of fetal DNA fraction from the plasma of pregnant women using sequencing read counts. Prenatal Diagn, 2015. Accepted. 7. Hahnemann JM, et al. Accuracy of cytogenetic findings on Chorionic Villus Sampling (CVS)—diagnostic consequences of CVS Mosaicism and non-Mosaic Discrepancy in Centres contributing to EUCROMIC 1986–1992. Prenatal Diagn, 1997. 17(9):801-820. Figure 1. Determination of sensitivity for sub- chromosomal abnormalities using a theoretical and an in silico spike-in model. • The theoretical model assumes an a priori knowledge of the MD location and does not take into account biological or technical variability. (Blue Curve) • At 32M reads/sample, the maximum achievable theoretical sensitivity is 100% for MDs ≥ 7Mb. • The in silico spike-in model does not use an a priori knowledge of the MD location and factors in both biological and technical variability. (Orange Curve) • At 28M reads/sample, there is a >90% sensitivity for MDs ≥ 7Mb and >95% sensitivity for MDs ≥ 10 Mb. Figure 2. Determination of sensitivity for sub- chromosomal abnormalities using pure gDNA. • Distribution of sizes for subchromosomal abnormalities in gDNA samples. There are 150 events between 1.25-7 Mb and 71 events larger than 7 Mb. (Top) • Sensitivity for different size subchromosomal ab- normalities. Sensitivity is 98.6% for >1.25 Mb events and 100% for >3 Mb events. (Bottom) • Poor mappability, around the centromere or telomere, is the primary reason for the few undetected events. Theoretical and In Silico Modeling detection in 100% Genomic dNa

Articles in this issue

view archives of sequenom - Karyotype-level noninvasive prenatal testing by sequencing of circulating cell-free DNA from maternal plasma