- Email: [email protected]
Clinical validation of a noninvasive prenatal test for genomewide detection of fetal copy number variants
Accepted Manuscript Clinical Validation of a Non-Invasive Prenatal Test for Genome-Wide Detection of Fetal Copy Number Variants Roy B. Lefkowitz, PhD, John A. Tynan, PhD, Tong Liu, PhD, Yijin Wu, PhD, Amin R. Mazloom, PhD, Eyad Almasri, MS, Grant Hogg, MS, Vach Angkachatchai, PhD, Chen Zhao, PhD, Daniel S. Grosu, MD, Graham Mclennan, MS, Mathias Ehrich, MD PII:
S0002-9378(16)00318-5
DOI:
10.1016/j.ajog.2016.02.030
Reference:
YMOB 10954
To appear in:
American Journal of Obstetrics and Gynecology
Received Date: 10 December 2015 Revised Date:
30 January 2016
Accepted Date: 11 February 2016
Please cite this article as: Lefkowitz RB, Tynan JA, Liu T, Wu Y, Mazloom AR, Almasri E, Hogg G, Angkachatchai V, Zhao C, Grosu DS, Mclennan G, Ehrich M, Clinical Validation of a Non-Invasive Prenatal Test for Genome-Wide Detection of Fetal Copy Number Variants, American Journal of Obstetrics and Gynecology (2016), doi: 10.1016/j.ajog.2016.02.030. This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
ACCEPTED MANUSCRIPT
Clinical Validation of a Non-Invasive Prenatal Test for Genome-Wide Detection of
2
Fetal Copy Number Variants
3 4 5 6
Roy B. LEFKOWITZ, PhD1, John A. TYNAN, PhD1, Tong LIU, PhD, 1, Yijin WU, PhD1, Amin R. MAZLOOM, PhD1, Eyad ALMASRI, MS1, Grant HOGG, MS1, Vach ANGKACHATCHAI, PhD1,Chen ZHAO, PhD1,3, Daniel S. GROSU, MD2, Graham MCLENNAN, MS2, Mathias EHRICH, MD2†
7
1
8
2
9
3
RI PT
1
Sequenom Laboratories, 3595 John Hopkins Ct, San Diego, CA, USA
Current affiliation - Illumina, San Diego, CA, USA
M AN U
10
SC
Sequenom Inc., 3595 John Hopkins Ct, San Diego, CA, USA
11
Authors are/were employed and own stock or stock options in Sequenom, Inc.
12
This study was funded by Sequenom, Inc.
13 14
†
15
Dr. Mathias Ehrich, MD
16
Sequenom Laboratories
17
3595 John Hopkins Court, San Diego, CA, 92121
18
Phone: 858-202-9068
19
Email: [email protected]
20
Abstract word count: 297
21
Main text word count: 2997
22
In this study we expand non-invasive prenatal testing to the entire genome and demonstrate
23
high sensitivity and specificity while adding genome-wide clinically relevant content.
24
Running Head: Clinical Validation of Genome-Wide Cell-Free DNA Testing
AC C
EP
TE D
To whom correspondence should be addressed:
ACCEPTED MANUSCRIPT
Keywords: non-invasive prenatal testing, genome-wide, cell-free DNA, chromosomal
26
copy number variant, sub-chromosomal copy number variant, microdeletions
27
Abbreviations: bootstrap confidence level (BCL); chorionic villus sampling (CVS);
28
chromosomal abnormality decision tree (CADET); cell-free DNA (cfDNA); circular binary
29
segmentation (CBS); copy number variants (CNVs); non-invasive prenatal testing
30
(NIPT); sex chromosomal aneuploidies (SCAs); triomsy 21 (T21); trisomy 18 (T18);
31
trisomy 13 (T13).
32 33
1. ABSTRACT
34
Background: Current cell-free DNA assessment of fetal chromosomes does not analyze and
35
report on all chromosomes. Hence, a significant proportion of fetal chromosomal abnormalities
36
are not detectable by current non-invasive methods. Here we report the clinical validation of a
37
novel non-invasive prenatal test designed to detect genome-wide gains and losses of
38
chromosomal material ≥7 Mb and losses associated with specific deletions <7 Mb.
39
Objective: The objective of this study is to provide a clinical validation of the sensitivity and
40
specificity of a novel non-invasive prenatal test for detection of genome-wide abnormalities.
41
Study Design: This retrospective, blinded study included maternal plasma collected from 1222
42
study subjects with pregnancies at increased risk for fetal chromosomal abnormalities that were
43
assessed for trisomy 21, trisomy 18 , trisomy 13 , sex chromosome aneuploidies , fetal sex,
44
genome-wide copy number variants 7 Mb and larger, and select deletions smaller than 7 Mb.
45
Performance was assessed by comparing test results with findings from G-band karyotyping,
46
microarray data, or high coverage sequencing.
AC C
EP
TE D
M AN U
SC
RI PT
25
ACCEPTED MANUSCRIPT
Results: Clinical sensitivity within this study was determined to be 100% for trisomy 21 (95%
48
confidence interval 94.6-100%), trisomy 18 (95% confidence interval 84.4-100%), trisomy 13
49
(95% confidence interval 74.7-100%), and sex chromosome aneuploidies (95% confidence
50
interval 84-100%), and 97.7% for genome-wide copy number variants (95% confidence interval
51
86.2-99.9%). Clinical specificity within this study was determined to be 100% for trisomy 21
52
(95% confidence interval 99.6-100%), trisomy 18 (95% confidence interval 99.6-100%), and
53
trisomy 13 (95% confidence interval 99.6-100%), and 99.9% for sex chromosome aneuploidies
54
and copy number variants (95% confidence interval 99.4-100% for both). Fetal sex classification
55
had an accuracy of 99.6% (95% confidence interval of 98.9-99.8%).
56
Conclusion: This study has demonstrated that genome-wide non-invasive prenatal testing for
57
fetal chromosomal abnormalities can provide high resolution, sensitive, and specific detection of
58
a wide range of sub-chromosomal and whole chromosomal abnormalities that were previously
59
only detectable by invasive karyotype analysis. In some instances, this non-invasive prenatal
60
test also provided additional clarification about the origin of genetic material that had not been
61
identified by invasive karyotype analysis.
62 63
2. INTRODUCTION
64
Since its introduction in 2011, non-invasive prenatal tests (NIPT) have had a significant impact
65
on prenatal care. In only four years, NIPT has evolved into a standard option for high-risk
66
pregnancies 1. Content has also evolved from exclusive trisomy 21 (T21) testing to include
67
trisomy 18 (T18), trisomy 13 (T13), sex chromosome aneuploidies (SCAs), and select
68
microdeletions. This ‘standard’ content can be expected to detect 80-83% of chromosomal
69
abnormalities detected by karyotyping in a general screening population 2-4, however this leaves
70
a gap of approximately 17-20% of alternative chromosomal/sub-chromosomal abnormalities not
71
detected. Consequently, obtaining comprehensive information about the genetic makeup of the
AC C
EP
TE D
M AN U
SC
RI PT
47
ACCEPTED MANUSCRIPT
fetus requires an invasive procedure. To overcome these limitations, NIPT could be extended to
73
cover the entire genome. However, it is challenging to maintain a very high specificity and
74
positive predictive value when interrogating all accessible regions in the genome 5. In previous
75
reports, we have overcome these technical hurdles 6. Furthermore, a recent study by Yin et al.
76
demonstrated feasibility for non-invasive genome-wide detection of sub-chromosomal
77
abnormalities 7. In this report, we have improved the assay and statistical methods to enable
78
comprehensive genome-wide detection of copy number variants (CNVs) ≥7 Mb. We present
79
results of a large blinded clinical study of more than 1200 samples including more than 100
80
samples with common aneuploidies detectable by traditional NIPT and over 30 samples
81
affected by sub-chromosomal CNVs.
82
3. MATERIALS and METHODS
83 84
STUDY DESIGN
85
risk for fetal aneuploidy based on advanced maternal age ≥35, a positive serum screen, an
86
abnormal ultrasound finding, and/or a history of aneuploidy. Archived samples were selected for
87
inclusion in the study by an unblinded internal third party according to the requirements
88
documented in the study plan. Samples were then blind-coded to all operators and the analysts
89
who processed the samples. After sequencing, an automated bioinformatics analysis was
90
performed to detect whole chromosome aneuploidies and sub-chromosomal CNVs. Results
91
were compiled electronically and were reviewed by a subject matter expert who assigned the
92
final classification. This manual review mimics the process in the clinical laboratory, where
93
cases are reviewed by a laboratory director before a result is signed out. Completed
94
classification results were provided to the internal third party for determination of concordance.
95
Analyzed samples had confirmation of positive or negative events by either G-band karyotype or
M AN U
SC
RI PT
72
AC C
EP
TE D
This blinded, retrospective clinical study included samples from women considered at increased
ACCEPTED MANUSCRIPT
microarray findings from samples collected through either CVS or amniocentesis. Circulating
97
cell-free “fetal” DNA is believed to originate largely from placental trophoblasts. Genetic
98
differences between the fetus and the placenta can occur (e.g. confined placental mosaicism),
99
leading to discordance between NIPT results and cytogenetic studies on amniocytes or
RI PT
96
postnatally obtained samples 8. Results from CVS by chromosomal microarray were thus
101
considered the most accurate ground truth. Therefore, discordant results originating from
102
amniocytes (karyotype or microarray) were resolved by sequencing at high coverage (an
103
average of 226 million reads per sample). Sequencing depth has been shown as the limiting
104
factor in NIPT methods, with increased depth allowing improved detection of events in samples
105
with lower fetal fractions or improved detection of smaller events 9. High coverage sequencing
106
has been used in multiple studies to unambiguously identify sub-chromosomal events 6,10-12 and
107
was used here as a reference for performance evaluation in discrepant amniocentesis samples.
108
Details of the sample demographics are described in Table 1. Indications for invasive testing are
109
described in Table 2.
110 111
SAMPLE COLLECTION
112
Review Board (IRB) approved protocols with a small subset (9 samples) comprising remnant
113
plasma samples collected from previously consented patients in accordance with the FDA
114
Guidance on Informed Consent for In Vitro Diagnostic Devices Using Leftover Human
115
Specimens that are Not Individually Identifiable (25 April 2006). Samples from two of the
116
protocols (Compass IRB #00508 and WIRB #20120148) were collected from high risk pregnant
117
subjects prior to undergoing a confirmatory invasive procedure (1189 samples). Samples from
118
the other two protocols (Compass IRB #00351 and Columbia University IRB #AAAN9002) came
119
from subjects who were enrolled in the studies after receiving the fetal karyotype and/or
120
microarray results of a confirmatory invasive procedure (24 samples). All subjects provided
TE D
M AN U
SC
100
AC C
EP
In total, 1222 maternal plasma samples had been previously collected using four Investigational
ACCEPTED MANUSCRIPT
written informed consent prior to undergoing any study related procedures.A total of 5321 high
122
risk subjects were recruited into the 4 clinical studies indicated above at the time of sample
123
selection. To be eligible for inclusion into this study, subjects had to have met all protocol
124
inclusion and no exclusion criteria, have fetal outcome determined by karyotype and/or
125
microarray and have at least one plasma aliquot of ≥3.5mL obtained from whole blood collected
126
in a Streck BCT. There was no sample selection preference based on high risk indication. All
127
subjects meeting these selection criteria that had an abnormal fetal outcome as needed for the
128
study were identified and pulled from inventory. These were then supplemented with randomly
129
selected subjects with samples meeting the same selection criteria but with a normal karyotype
130
to reach the total of 1222 samples for testing.
131 132
LIBRARY PREPARATION, SEQUENCING, AND ANALYTICAL METHODS
133
increase signal, sequencing depth for this analysis was increased to target 32M reads per
134
sample. Sequencing reads were aligned to hg19 using Bowtie 2 14. The genome was then
135
partitioned into 50-kbp non-overlapping segments and the total number of reads per segment
136
was determined, by counting the number of reads with 5’ ends overlapping with a segment.
137
Segments with high read-count variability or low map-ability were excluded. The 50 kbp read-
138
counts were then normalized to remove coverage and GC biases, and other higher-order
139
artifacts using the methods previously described in Zhao et al 6.
140
The presence of fetal DNA was quantified using the regional counts of whole genome single-
141
end sequencing data as described by Kim et al. 15.
142 143
GENOME-WIDE DETECTION OF ABNORMALITIES
144
segmenting each chromosome into contiguous regions of equal copy number 16. A segment-
145
merging algorithm was then used to compensate for over segmentation by CBS when the
M AN U
SC
RI PT
121
AC C
EP
TE D
Libraries were prepared and quantified as described by Tynan et al.13. To reduce noise and
Circular binary segmentation (CBS) was used to identify CNVs throughout the entire genome by
ACCEPTED MANUSCRIPT
signal-to-noise ratio (SNR) was low 17. Z-scores were calculated for both CBS-identified CNVs
147
and whole chromosome variants by comparing the signal amplitude with a reference set of
148
samples in the same region. The measured Z-scores form part of an enhanced version of
149
Chromosomal Aberration DEcision Tree (CADET) previously described in detail in Zhao et al 6.
150
To further improve specificity of CNV detection, bootstrap analysis was performed as an
151
additional measure for the confidence of the candidate CNVs. The within sample read count
152
variability was compared to a normal population (represented by 371 euploid samples) and
153
quantified by bootstrap confidence level (BCL). In order to assess within sample variability, the
154
bootstrap resampling described below was applied to every candidate CNV.
155
For each identified segment within the CNV, the median shift of segment fraction from the
156
normal level across the chromosome was calculated. This median shift was then corrected to
157
create a read count baseline for bootstrapping. Next, a bootstrapped segment of the same
158
segment length as the candidate CNV was randomly sampled with replacement from the
159
baseline read counts. The median shift was then applied to this bootstrapped fragment. The
160
segment fraction of the bootstrapped fragment was calculated as follows:
TE D
M AN U
SC
RI PT
146
EP
=
∑ ℎ ∑ ℎ
This process was repeated 1000 times to generate a bootstrap distribution of segment fractions
162
for an affected population. A normal reference distribution was created based on the segment
163
fraction of the same location as the candidate CNV in 371 euploid samples. A threshold was
164
then calculated as the segment fraction that was at least 3.95 median absolute deviations away
165
from the median segment fraction of the reference distribution. Lastly, the BCL was calculated
166
as the proportion of bootstrap segments whose fractions had absolute z-statistics above the
167
significance threshold.
AC C
161
ACCEPTED MANUSCRIPT
168
A whole chromosome or sub-chromosomal abnormality is detected as described in Zhao et al.
169 170
4. RESULTS
171
samples were excluded because they had no or insufficient karyotype or microarray information
172
(see Figure 1) and three samples were excluded because of confirmed mosaicism.
173
Forty-two (42) of the remaining 1208 samples were flagged as non-reportable using quality
174
criteria that had been established prior to analysis, leaving 1166 reportable samples for
175
analyses comprising 13% (n=153) samples with common whole chromosome 21, 18, 13, X, or
176
Y aneuploidies and 3.6% (n=43) samples with sub-chromosomal CNVs or rarer whole
177
chromosome aneuploidies. Technical failure criteria included but were not limited to: low library
178
concentration, low raw autosomal counts, high GC bias, poor normalization, and high bin
179
variability (see Figure 1). Biological failure criteria included low fetal fraction (less than 4%) and
180
large maternal CNV events. The most common reason for failure was low fetal fraction (n= 11).
181
During review of the data, one sample was signed out as T18 (and was included in the analyzed
182
cohort), even though it did not meet the autosomal count minimum. This sample had sufficient
183
counts for the determination of standard aneuploidies, but not sufficient counts for the detection
184
of sub-chromosomal CNVs. The 42 non-reportable samples showed no evidence of enrichment
185
for whole-chromosome/sub-chromosomal positive samples (5 positive of 42 non-reportable
186
samples vs. 204 positive of 1166 reportable samples). Details of non-reportable and excluded
187
samples are provided in Supplemental Table 2. Concordant with previous studies, the overall
188
reportable rate on first aliquots of maternal plasma was 96.5%.The overall no call rate per
189
patient is expected to be approximately 1% when a second aliquot is available, based on
190
published clinical laboratory experience 18.
AC C
EP
TE D
M AN U
SC
RI PT
The study comprised a total of 1222 maternal plasma samples. After unblinding, eleven
ACCEPTED MANUSCRIPT
191 192
T21, T18, and T13 Detection
193
samples (see Figure 1). All euploid, T21, T18, and T13 samples (determined by invasive
194
diagnostic procedures) were classified correctly by NIPT. One sample was classified by NIPT
195
as T21 but had a normal (46, XX) karyotype by amniocentesis (see Supplemental Table 1). This
196
discrepancy was adjudicated through high coverage ‘uniplex’ sequencing (typically with >180
197
million reads) according to our study plan (see Supplemental Figure 1D). The sample showed
198
16.3% fetal fraction, but the gain of genetic material from chromosome 21 was concordant with
199
6.5% fetal fraction. This is suggestive of confined placental mosaicism, a relatively common
200
cause of discordant results between amniocyte results and CVS or placentally-derived cell free
201
DNA analysis 8. Table 3 summarizes the performance for T21, T18, and T13. For T21, the
202
sensitivity was 100% (95% CI: 94.6%-100%) and the specificity was 100% (95% CI: 99.6%-
203
100%). For T18, the sensitivity was 100% (95% CI: 84.4%-100%) and the specificity was 100%
204
(95% CI: 99.6%-100%). For T13, the sensitivity was 100% (95% CI: 74.7%-100%) and the
205
specificity was 100% (95% CI: 99.6%-100%).
206 207
SCA Detection
208
were flagged as non-reportable specifically for SCA classification based on thresholds for the
209
chromosome X z-score and chromosome Y z-score as described by Mazloom et al. 19. There
210
was also one additional sample with an apparent maternal XXX that was flagged as non-
211
reportable for SCA because the maternal event distorted the sex chromosomal representation
212
to a degree that fetal events could not be classified. Among the remaining 1144 samples
213
reportable for SCAs, there were 7 discordant positives that were classified as normal (46, XX)
214
by karyotype and as XO by sequencing at 6-plex (see Supplemental Table 1). In all discordant
215
cases, the karyotype had been obtained from amniocyte samples. This phenomenon is well
216
described and may be attributed to varying levels of placental or maternal mosaicism 20. Uniplex
TE D
M AN U
SC
RI PT
Among the 1166 reportable samples, there were 85 T21 samples, 27 T18 samples, and 15 T13
AC C
EP
In the 1166 samples reportable for all chromosomal abnormalities, there were 21 samples that
ACCEPTED MANUSCRIPT
sequencing confirmed 6 of the 7 XO samples. The 7th sample had a non-reportable result at
218
uniplex coverage, hence the existing amniocentesis result was used as truth resulting in one
219
false positive assignment. Overall, the sensitivity for SCA was 100% (95% CI: 84.0%-100%)
220
with a specificity of 99.9% (95% CI: 99.4%-100%) (see Table 3).
221 222
Genome Wide Detection of CNVs
223
chromosome abnormalities other than T13, T18, T21, and SCAs); as well as select
224
microdeletions smaller than 7 Mb 5. Among the 1166 samples reportable for sub-chromosomal
225
abnormalities, there were 43 samples that had positive results for a variety of CNV aberrations:
226
Wolf-Hirschhorn syndrome deletions, DiGeorge syndrome deletions, Prader-Willi/Angelman
227
syndrome deletions, Cri du Chat syndrome deletions, and a variety of ≥7 Mb CNVs including
228
both sub-chromosomal CNVs and whole chromosome trisomies. Several examples of NIPT
229
detected CNVs confirmed by microarray or karyotype are shown in Figure 2.
230
Overall, the sensitivity for detection of whole chromosome and sub-chromosomal abnormalities
231
other than T13, T18, T21, and SCAs was 97.7% (95% CI: 86.2%-99.9%) and the specificity was
232
99.9% (95% CI: 99.4%-100%) (see Table 3). One case was clearly mosaic for T22 by both
233
standard coverage and uniplex sequencing (see Supplemental Table 1 and Supplement Figure
234
1E), but was classified as normal by microarray analysis. Because the invasive diagnosis came
235
from microarray analysis of cells derived from CVS, our study design considered this as the gold
236
standard, and this outcome was considered a discordant positive. Another case showed no gain
237
or loss of genetic material with both standard and uniplex sequencing for a sample that had a
238
46, XX, der(12)t(12;19)(p13.1;q13.1) karyotype (see Supplemental Figure 2B). This outcome
239
was considered as a discordant negative given that the invasive procedure was CVS. A set of 7
240
samples were classified as full chromosomal trisomies, and because the karyotype or
241
microarray results were derived from amniocytes in these cases, uniplex sequencing was
RI PT
217
AC C
EP
TE D
M AN U
SC
The test was also designed to detect CNVs equal to or larger than 7 Mb (including whole
ACCEPTED MANUSCRIPT
performed for adjudication per the study design. The trisomy finding was confirmed in each case
243
by uniplex sequencing, and these findings were considered as concordant positives
244
(Supplemental Figure 1).
245
In addition to the high accuracy demonstrated in the results, in some cases NIPT could also
246
provide clarification about the origin of extra genetic material when the G-banding pattern was
247
not sufficiently clear. In one case, the amniocyte karyotype finding, 46, XX, der(5)t(5;?)(p15.3;?),
248
indicated a deletion on chromosome 5 and a duplication of unknown origin (see Figure 3). NIPT
249
identified an 18.6 Mb duplication representing the entire short arm of chromosome 17, indicating
250
this fetal tissue possessed a trisomy of chromosome 17p that was not clarified by standard
251
karyotype. Trisomy 17p is associated with developmental delay, growth retardation, hypotonia,
252
digital abnormalities, congenital heart defects, and distinctive facial features 21. Additional cases
253
are shown in Supplemental Figures 3.
254 255
5. COMMENT
256
baby have multiple options today – ranging from non-invasive tests that screen for select
257
chromosomal abnormalities to invasive procedures including karyotype and microarray testing
258
that can deliver the most comprehensive genomic assessment. Current NIPT methods provide
259
information about a limited set of conditions, that typically include T21, T18, T13 and SCAs. In
260
this study we expand testing to the entire genome. The results demonstrate that high sensitivity
261
and specificity were maintained while adding genome-wide clinically relevant content.
262
Most NIPT tests suffer from the problem of multiple hypothesis testing; when multiple regions of
263
the genome are tested independently, false positive rates become additive 22. This can result in
264
unacceptably high false positive rates and lead to maternal anxiety and additional unnecessary
265
clinical testing. Methods that target individual regions of the genome cannot overcome these
TE D
M AN U
SC
RI PT
242
AC C
EP
Couples and families who are seeking prenatal information about the genetic health of their
ACCEPTED MANUSCRIPT
limitations because they are part of the design of the test 23. We have shown that these
267
limitations can be overcome by using a genome-wide approach, as described here and
268
previously 6. In this study we demonstrate that the test can provide excellent performance for
269
CNVs ≥7Mb and select microdeletions <7Mb, across the genome.
270
This study does have limitations. Although we analyze 2- to 6-fold more samples with sub-
271
chromosomal CNVs than previous studies 6,11,23, the total number of affected samples is still
272
relatively small compared to studies validating performance for detection common whole-
273
chromosome aneuploidies 24. It remains challenging to statistically validate detection of rare
274
individual chromosomal variations. As such we utilize samples with CNVs of varying size and
275
location throughout the genome with varying frequency of individual occurrence. This study
276
design with its high positivity rate allows only estimation of PPV.. As sensitivity and specificity
277
values have proven stable in previous studies when applied to the real world samples, PPV can
278
be considered a function of incidence rate. Incidence estimates of larger karyotype detectable
279
genome-wide sub-chromosomal CNVs vary but are typically considered between 1 in 200 and 1
280
in 40025. Modeled PPVs for these incidence rates would therefore be between 68% and 83%
281
(see Supplemental Figure 4). Due to the high sensitivity observed, negative predictive values
282
are expected to be well over 99.9% when incidence rates are less than 1 in 200. As our study
283
utilized only one aliquot for analysis, our reportable rate was 96.5%, with an expected reportable
284
rate of ~99% with availability of two aliquots18. Of the 11 samples not reported due to the
285
biological effect of low fetal fraction, all had normal karyotype result.
SC
M AN U
TE D
EP
AC C
286
RI PT
266
Another challenge commonly observed in novel technologies with increased performance is
287
the compilation of samples with appropriate ground truth. In our study we face the technical
288
challenge of comparing sequencing versus G-band karyotype or microarray, as well as the
289
biological aspect of cfDNA versus amniocyte or CVS. We have used a method that prioritizes
290
CVS over amniocentesis because of the placental origin of cell free DNA 8. Factors that can
ACCEPTED MANUSCRIPT
lead to discordant results include confined placental mosaicism, maternal mosaicism, vanishing
292
twins, undetected tumors, and technical errors. Consequently our method of evaluation has the
293
potential to overestimate performance with regard to fetal outcome. However, knowledge about
294
the genetic health of the placenta might in itself be advantageous when guiding the pregnancy,
295
for example in the case of intra uterine growth restriction. Prospective studies using the method
296
described here with follow-up of birth outcome will further enhance understanding of clinical
297
effectiveness.
298
An interesting observation in this study was the sometimes subjective nature of karyotyping.
299
When G-banding is used as an analytical method, the optical resolution is occasionally not
300
sufficient to determine the origin of additional genetic material, making a clinical interpretation
301
more difficult. These ambiguities are eliminated when using NIPT because the test requires
302
sequencing alignment to the genome before DNA gains and losses are determined. However,
303
G-banding will, for the foreseeable future, be the superior methodology to determine copy
304
number neutral structural changes.
305
In pregnancies that can benefit from additional information, this test provides more clinically
306
relevant results than previous NIPT options. However, its role as a follow-up test to abnormal
307
ultrasound findings or as a general population screen will likely be debated for the foreseeable
308
future. It has been argued that the incidence of clinically relevant sub-chromosomal CNVs could
309
be as high as 1.7% in a general population 25, and early microarray data 26 indicate that as many
310
as 30% of sub-chromosomal CNVs could be >7Mb. In consequence, the incidence for these
311
large CNVs may be around 0.5% in the general population, substantially higher than the
312
incidence of T21.
313
In summary, this clinical study provides validation for an approach that extends the clinical
314
validity of cfDNA testing, now providing detection of T21, T18, T13, SCAs, fetal sex, and
AC C
EP
TE D
M AN U
SC
RI PT
291
ACCEPTED MANUSCRIPT
genome-wide detection of sub-chromosomal and whole chromosomal abnormalities. Overall,
316
sub-chromosomal abnormalities and aneuploidies other than T13, T18, T21, and SCAs were
317
detected with a combined clinical sensitivity and specificity of 97.7% (95%CI 86.2-99.9%) and
318
99.9% (95%CI 99.4-100%), respectively. This enables comprehensive non-invasive
319
chromosomal assessment that was previously available only by karyotype, and in some cases,
320
may clarify cryptic findings otherwise identifiable only by microarray.
RI PT
315
SC
321
Acknowledgements
323
The authors would like to acknowledge the efforts of those who aided completion of this
324
study: Nathan Faulkner and Ashley Diaz for sample procurement; Sarah Ahn,
325
Christine Safanayong, Haiping Lu, Joshua White, Lisa Chamberlain for sample
326
processing and DNA extraction; David Wong, Jana Schroth, Marc Wycoco, Jesse Fox
327
for library preparation and sequencing, Pat Liu for bioinformatics pipeline processing.
328
We would also like to acknowledge the contribution of samples by Ronald Wapner at
329
Columbia University.
AC C
EP
TE D
M AN U
322
ACCEPTED MANUSCRIPT
6. REFERENCES
331 332
1.
ACOG. Cell-free DNA screening for fetal aneuploidy. Committee Opinion No. 640. Obstetrics & Gynecology. 2015;126:e31-37.
333 334 335 336
2.
Norton ME, Wapner RJ, Kuppermann M, Jelliffe-Pawlowski LL, Currier RJ. Cell free DNA analysis vs sequential screening as primary testing considering all fetal chromosomal abnormalities. 35th Annual Meeting of the Society for Maternal-Fetal Medicine: The Pregnancy Meeting. 02 February 2015 - 07 February 2015 2015;212(1):S2.
337 338 339
3.
Norton ME, Jelliffe-Pawlowski LL, Currier RJ. Chromosome abnormalities detected by current prenatal screening and noninvasive prenatal testing. Obstetrics & Gynecology. 2014;124(5):979-986.
340 341 342
4.
Wellesley D, Dolk H, Boyd PA, et al. Rare chromosome abnormalities, prevalence and prenatal diagnosis rates from population-based congenital anomaly registers in Europe. European Journal of Human Genetics. 2012;20(5):521-526.
343 344
5.
Helgeson J, Wardrop J, Boomer T, et al. Clinical outcome of subchromosomal events detected by whole‐ ‐genome noninvasive prenatal testing. Prenatal Diagnosis. 2015;35(10):999-1004.
345 346 347
6.
Zhao C, Tynan J, Ehrich M, et al. Detection of Fetal Subchromosomal Abnormalities by Sequencing Circulating Cell-Free DNA from Maternal Plasma. Clinical Chemistry. 2015;61(4):608-616.
348 349 350
7.
Yin A-h, Peng C-f, Zhao X, et al. Noninvasive detection of fetal subchromosomal abnormalities by semiconductor sequencing of maternal plasma DNA. Proceedings of the National Academy of Sciences. 2015:201518151.
351 352 353
8.
Grati FR, Malvestiti F, Ferreira JC, et al. Fetoplacental mosaicism: potential implications for false-positive and false-negative noninvasive prenatal screening results. Genetics in Medicine. 2014;16(8):620-624.
354 355 356
9.
Fan HC, Quake SR. Sensitivity of noninvasive prenatal detection of fetal aneuploidy from maternal plasma using shotgun sequencing is limited only by counting statistics. PLoS One. 2010;5(5):e10439.
357 358
10.
Jensen TJ, Kim SK, van den Boom D, Deciu C, Ehrich M. Noninvasive Detection of a Balanced Fetal Translocation from Maternal Plasma. Clinical Chemistry. 2014;60(10):1298-1305.
359 360 361
11.
362 363
12.
Peters D, Chu T, Yatsenko SA, et al. Noninvasive prenatal diagnosis of a fetal microdeletion syndrome. New England Journal of Medicine. 2011;365(19):1847-1848.
364 365 366
13.
Tynan J, Kim S, Mazloom A, et al. Application of risk score analysis to low‐ ‐coverage whole genome sequencing data for the noninvasive detection of trisomy 21, trisomy 18, and trisomy 13. Prenatal Diagnosis. 2015.
AC C
EP
TE D
M AN U
SC
RI PT
330
Srinivasan A, Bianchi DW, Huang H, Sehnert AJ, Rava RP. Noninvasive detection of fetal subchromosome abnormalities via deep sequencing of maternal plasma. The American Journal of Human Genetics. 2013;92(2):167-176.
ACCEPTED MANUSCRIPT
14.
Langmead B, Salzberg SL. Fast gapped-read alignment with Bowtie 2. Nature Methods. 2012;9(4):357-359.
369 370
15.
Kim SK, Hannum G, Geis J, et al. Determination of fetal DNA fraction from the plasma of pregnant women using sequence read counts. Prenatal Diagnosis. 2015;35(8):810-815.
371 372
16.
Olshen AB, Venkatraman E, Lucito R, Wigler M. Circular binary segmentation for the analysis of array‐ ‐based DNA copy number data. Biostatistics. 2004;5(4):557-572.
373 374
17.
Willenbrock H, Fridlyand J. A comparison study: applying segmentation to array CGH data for downstream analyses. Bioinformatics. 2005;21(22):4084-4091.
375 376
18.
McCullough RM, Almasri EA, Guan X, et al. Non-invasive prenatal chromosomal aneuploidy testing-clinical experience: 100,000 clinical samples. PLoS One. 2014;9(10):e109173.
377 378 379
19.
Mazloom AR, Džakula Ž, Oeth P, et al. Noninvasive prenatal detection of sex chromosomal aneuploidies by sequencing circulating cell‐ ‐free DNA from maternal plasma. Prenatal Diagnosis. 2013;33(6):591-597.
380 381 382
20.
Wang Y, Chen Y, Tian F, et al. Maternal mosaicism is a significant contributor to discordant sex chromosomal aneuploidies associated with noninvasive prenatal testing. Clinical Chemistry. 2014;60(1):251-259.
383 384
21.
Jinno Y, Matsuda I, Kajii T. Trisomy 17p due to a t(5; 17)(p15; p11) pat translocation. Annales de Genetique. 1981;25(2):123-125.
385
22.
Miller RG. Simultaneous statistical inference. 2nd ed. New York: Springer-Verlag; 1981.
386 387 388
23.
Wapner RJ, Babiarz JE, Levy B, et al. Expanding the scope of noninvasive prenatal testing: detection of fetal microdeletion syndromes. American Journal of Obstetrics and Gynecology. 2015;212(3):332.e331-339.
389 390 391
24.
Palomaki GE, Kloza EM, Lambert-Messerlian GM, et al. DNA sequencing of maternal plasma to detect Down syndrome: an international clinical validation study. Genetics in Medicine. 2011;13(11):913-920.
392 393
25.
Wapner RJ, Martin CL, Levy B, et al. Chromosomal microarray versus karyotyping for prenatal diagnosis. New England Journal of Medicine. 2012;367(23):2175-2184.
394 395 396
26.
398 399 400 401
SC
M AN U
TE D
EP
AC C
397
RI PT
367 368
Shaffer LG, Dabell MP, Fisher AJ, et al. Experience with microarray‐ ‐based comparative genomic hybridization for prenatal diagnosis in over 5000 pregnancies. Prenatal Diagnosis. 2012;32(10):976-985.
ACCEPTED MANUSCRIPT
402 403
TABLES
404
Table 1. Demographic and pregnancy-related data. Median
Maternal Age (Yrs) (Ntotal = 1177)
36.0
Gestational Age (Wks) (Ntotal =1183)
17
Maternal Weight (Lbs) (Ntotal =1168)
150 Percent
CVS
14.5
17.8-47 8-38
93-366
(Naffected/Ntotal) (175/1203)
SC
Procedure
Range
RI PT
Demographic
Amniocentesis Both
(1025/1203)
0.2
(3/1203)
Percent
(Naffected/Ntotal)
90.4
(1089/1205)
5.8
(70/1205)
M AN U
Confirmation
85.2
Karyotype Microarray
Both 3.8 (46/1205) Note: Choice of invasive procedure, choice of diagnostic test, and demographic data were not available for all 1208 samples included in the study. Ntotal refers to the number of samples where that data was available.
406
Table 2. Indications for testing. Euploid (n=1009)
T21 (n=87)
T18 (n=29)
T13 (n=15)
SCA (n=26)
Positive Serum Screening
404(40%)
36(41.4%)
13(44.8%)
4(26.7%)
3(11.5%)
Maternal Age > 35 Yrs
418(41.4%) 28(32.2%)
10(34.5%)
4(26.7%)
2(7.7%)
Ultrasound Abnormality
152(15.1%) 34(39.1%)
6(20.7%)
10(66.7%)
16(61.5%)
EP
Indication
AC C
407
TE D
405
CNV (n=44) 16(36.4%) 10(22.7%) 22(50%)
Total (n=1210) 476(39.3%) 472(39%) 239(19.8%)
Family History
33(3.3%)
4(4.6%)
1(3.4%)
0(0%)
1(3.8%)
1(2.3%)
40(3.3%)
Not Specified
85(8.4%)
11(12.6%)
5(17.2%)
2(13.3%)
4(15.4%)
2(4.5%)
109(9%)
Note: Each cell lists the count and percentage of the indication within a chromosome category. Samples may have multiple indications for testing and therefore do not sum to 100%. As 2 of the T18 samples also had an SCA, the total number of abnormalities and euploid samples sum to 1210.
ACCEPTED MANUSCRIPT
408
Table 3. Clinical performance for the indicated abnormalities and fetal sex. Concordant Discordant Concordant Discordant Positive Positive Negative Negative 85
0
1081
0
T18
27
0
1139
0
T13
15
0
1151
0
SCA
26
1
1117
CNVs**
42
1
1122
0
1
Concordant Discordant Concordant Discordant Male Male Female Female
Analyte Fetal Sex
583
4
578
1
Specificity (95% CI*)
100%
100%
(94.6%-100%)
(99.6%-100%)
100%
100%
(84.4%-100%)
(99.6%-100%)
100%
100%
(74.7%-100%)
(99.6%-100%)
SC
T21
Sensitivity (95% CI*)
RI PT
Abnormality
M AN U
409
100%
99.9%
(84.0%-100%)
(99.4%-100%)
97.7%
99.9%
(86.2%-99.9%)
(99.4%-100%)
Accuracy (95% CI*) 99.6% (98.9%-99.8%)
412 413 414 415 416 417 418 419 420
EP
411
AC C
410
TE D
* - CI indicates ‘confidence inter val’ **- CNVs include 8 samples with detected whole chromosome trisomies, and 35 samples with subchromosomal CNVs
ACCEPTED MANUSCRIPT
FIGURE LEGENDS
424 425 426 427 428
Figure 1. Flow chart showing sample exclusions and reasons for non-reportable samples. One of the trisomy 18 samples was also an XXY sample and one of the other trisomy 18 samples was also an X sample. For samples that had discordant results for sequencing versus karyotype or microarray derived from testing of amniocytes, uniplex sequencing was used for confirmation.
RI PT
422 423
429
SC
Figure 2. Examples of karyotype- and microarray-detected CNVs detected by sequencing. A) Chromosome 22 ideogram showing sequencing-based detection of a 22q11.21 deletion (DiGeorge syndrome) that was confirmed by karyotype and microarray. B) Chromosome 18 ideogram for a sample with a karyotype 46, XY, i(18)(q10). C) Chromosome 7 and 13 ideograms for a sample with a karyotype 46, XX, der(7)t(7;13)(q36;q22).
M AN U
430 431 432 433 434 435 436 437 438 439 440
Figure 3. Clarification of karyotype findings by sequencing of maternal plasma. Normal regions are colored blue while deletions and duplications that are detected with high bootstrap confidence levels (>0.99) are colored red. Low bootstrap confidence events are colored yellow. Chromosome 5 & 17 ideograms are shown for a sample with a karyotype of 46, XX, der(5)t(5;?)(p15.3;?). The unidentified duplication on the karyotype is from chromosome 17.
444 445 446 447
EP
443
AC C
442
TE D
441
AC C
EP
TE D
M AN U
SC
RI PT
ACCEPTED MANUSCRIPT
AC C
EP
TE D
M AN U
SC
RI PT
ACCEPTED MANUSCRIPT
AC C
EP
TE D
M AN U
SC
RI PT
ACCEPTED MANUSCRIPT
AC C
EP
TE D
M AN U
SC
RI PT
ACCEPTED MANUSCRIPT
AC C
EP
TE D
M AN U
SC
RI PT
ACCEPTED MANUSCRIPT
AC C
EP
TE D
M AN U
SC
RI PT
ACCEPTED MANUSCRIPT
AC C
EP
TE D
M AN U
SC
RI PT
ACCEPTED MANUSCRIPT
AC C
EP
TE D
M AN U
SC
RI PT
ACCEPTED MANUSCRIPT
ACCEPTED MANUSCRIPT
Supplemental Table 2. Breakdown of excluded and non-reported samples.
Biological exclusions
Fetal fraction
Maternal event Low aligned counts
CADET
0.093 0.114 0.081 0.148
54039
0.098
89063 73189 55348 53720 54528 89588 89473 77411 77188 89311 53415 53801 53103 55197 45424
0.096 0.089 0.040 0.036 0.039 0.036 0.032 0.019 0.028 0.037 0.027 0.036 0.037 0.065 0.149
GC warning
Lab Director review
RI PT
72938 71894 54681 72288
77478
0.079
77065 77066 77068 77042 77340 55462 89583 77040 93693 77392 89526 93679 89152 53815 73161 73017 71006 77075 54320 54212 77067 53061
0.087 0.092 0.083 0.109 0.131 0.078 0.082 0.058 0.112 0.091 0.105 0.074 0.076 0.077 0.161 0.158 0.072 0.104 0.048 0.147 0.072 0.043
AC C
Technical exclusions
High variability
0.094
Normal Normal Normal Other: T16 by NIPT. SCA 46-XY Normal karyotype; low level mosaic T21 by FISH Other: 46,XX,dup(1)(q21q32) - mosaic Other: 47,XX,+i(12)(p10)[13]/46,XX[2].nuc ish (ETV6x4,AML 1/20[25/25] - tetrasomy 12p mosaicism Other: 46,XY,var(18)(q11.2) Other: 46XX, 22qvar Other: 47, XX, +mar Other: 47, XX, +mar Other: 47, XX, +mar.ish idic(15;15)(q13;q13)(D15Z1++, SNRPN++) Other: 47, XX, +ESAC considered normal female Other: 47, XY, +ESAC Normal Normal Normal Normal Normal Normal Normal Normal Normal Normal Normal Normal T21 Normal
SC
Vague karyotype result
70600
Karyo/array Outcome
M AN U
Mosaicism
53063 77320 53863 RAS0027851 54172 77168
TE D
Exclusions
Missing invasive confirmation
Fetal fraction 0.130 0.079 0.068 0.065 0.093 0.247
SampleID
EP
Reason
T18 T21 Normal Normal Normal Normal Normal Normal Normal Normal Normal Normal Normal Normal Normal Normal Other: 46xx, add(18)(q21.3) Normal Normal Normal Normal Normal
ACCEPTED MANUSCRIPT
89529 77445 55123 89185 89607 53057 55444 54458 55220 93678 70268 53109 55454 55064 77285 55260
0.106 0.232 0.084 0.068 0.076 0.057 0.041 0.215 0.068 0.145 0.094 0.087 0.047 0.099 0.120 0.154
77157 73164 45364 89582 77476 89220
EP AC C
Normal Normal Normal Normal Normal Normal Normal Normal Normal Normal Normal Normal Normal Normal Normal Normal Other: 46,XX,der(18)(qter-q11.3-qter). Additional material added to 18p. Other: abnormal female karyotype with chromosome 13q deletion T18 T18 T21 T21
RI PT
0.112
SC
54227
Normal Normal Normal Other: 48,XXX,+18 Normal Normal
M AN U
0.064 0.094 0.075 0.092 0.148
0.042 0.085
0.102 0.099 0.108 0.130
TE D
Non-reportable for sex chromosome aneuploidy
Library failure
89359 89149 53420 55112 77347
S0002-9378(16)00318-5
DOI:
10.1016/j.ajog.2016.02.030
Reference:
YMOB 10954
To appear in:
American Journal of Obstetrics and Gynecology
Received Date: 10 December 2015 Revised Date:
30 January 2016
Accepted Date: 11 February 2016
Please cite this article as: Lefkowitz RB, Tynan JA, Liu T, Wu Y, Mazloom AR, Almasri E, Hogg G, Angkachatchai V, Zhao C, Grosu DS, Mclennan G, Ehrich M, Clinical Validation of a Non-Invasive Prenatal Test for Genome-Wide Detection of Fetal Copy Number Variants, American Journal of Obstetrics and Gynecology (2016), doi: 10.1016/j.ajog.2016.02.030. This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
ACCEPTED MANUSCRIPT
Clinical Validation of a Non-Invasive Prenatal Test for Genome-Wide Detection of
2
Fetal Copy Number Variants
3 4 5 6
Roy B. LEFKOWITZ, PhD1, John A. TYNAN, PhD1, Tong LIU, PhD, 1, Yijin WU, PhD1, Amin R. MAZLOOM, PhD1, Eyad ALMASRI, MS1, Grant HOGG, MS1, Vach ANGKACHATCHAI, PhD1,Chen ZHAO, PhD1,3, Daniel S. GROSU, MD2, Graham MCLENNAN, MS2, Mathias EHRICH, MD2†
7
1
8
2
9
3
RI PT
1
Sequenom Laboratories, 3595 John Hopkins Ct, San Diego, CA, USA
Current affiliation - Illumina, San Diego, CA, USA
M AN U
10
SC
Sequenom Inc., 3595 John Hopkins Ct, San Diego, CA, USA
11
Authors are/were employed and own stock or stock options in Sequenom, Inc.
12
This study was funded by Sequenom, Inc.
13 14
†
15
Dr. Mathias Ehrich, MD
16
Sequenom Laboratories
17
3595 John Hopkins Court, San Diego, CA, 92121
18
Phone: 858-202-9068
19
Email: [email protected]
20
Abstract word count: 297
21
Main text word count: 2997
22
In this study we expand non-invasive prenatal testing to the entire genome and demonstrate
23
high sensitivity and specificity while adding genome-wide clinically relevant content.
24
Running Head: Clinical Validation of Genome-Wide Cell-Free DNA Testing
AC C
EP
TE D
To whom correspondence should be addressed:
ACCEPTED MANUSCRIPT
Keywords: non-invasive prenatal testing, genome-wide, cell-free DNA, chromosomal
26
copy number variant, sub-chromosomal copy number variant, microdeletions
27
Abbreviations: bootstrap confidence level (BCL); chorionic villus sampling (CVS);
28
chromosomal abnormality decision tree (CADET); cell-free DNA (cfDNA); circular binary
29
segmentation (CBS); copy number variants (CNVs); non-invasive prenatal testing
30
(NIPT); sex chromosomal aneuploidies (SCAs); triomsy 21 (T21); trisomy 18 (T18);
31
trisomy 13 (T13).
32 33
1. ABSTRACT
34
Background: Current cell-free DNA assessment of fetal chromosomes does not analyze and
35
report on all chromosomes. Hence, a significant proportion of fetal chromosomal abnormalities
36
are not detectable by current non-invasive methods. Here we report the clinical validation of a
37
novel non-invasive prenatal test designed to detect genome-wide gains and losses of
38
chromosomal material ≥7 Mb and losses associated with specific deletions <7 Mb.
39
Objective: The objective of this study is to provide a clinical validation of the sensitivity and
40
specificity of a novel non-invasive prenatal test for detection of genome-wide abnormalities.
41
Study Design: This retrospective, blinded study included maternal plasma collected from 1222
42
study subjects with pregnancies at increased risk for fetal chromosomal abnormalities that were
43
assessed for trisomy 21, trisomy 18 , trisomy 13 , sex chromosome aneuploidies , fetal sex,
44
genome-wide copy number variants 7 Mb and larger, and select deletions smaller than 7 Mb.
45
Performance was assessed by comparing test results with findings from G-band karyotyping,
46
microarray data, or high coverage sequencing.
AC C
EP
TE D
M AN U
SC
RI PT
25
ACCEPTED MANUSCRIPT
Results: Clinical sensitivity within this study was determined to be 100% for trisomy 21 (95%
48
confidence interval 94.6-100%), trisomy 18 (95% confidence interval 84.4-100%), trisomy 13
49
(95% confidence interval 74.7-100%), and sex chromosome aneuploidies (95% confidence
50
interval 84-100%), and 97.7% for genome-wide copy number variants (95% confidence interval
51
86.2-99.9%). Clinical specificity within this study was determined to be 100% for trisomy 21
52
(95% confidence interval 99.6-100%), trisomy 18 (95% confidence interval 99.6-100%), and
53
trisomy 13 (95% confidence interval 99.6-100%), and 99.9% for sex chromosome aneuploidies
54
and copy number variants (95% confidence interval 99.4-100% for both). Fetal sex classification
55
had an accuracy of 99.6% (95% confidence interval of 98.9-99.8%).
56
Conclusion: This study has demonstrated that genome-wide non-invasive prenatal testing for
57
fetal chromosomal abnormalities can provide high resolution, sensitive, and specific detection of
58
a wide range of sub-chromosomal and whole chromosomal abnormalities that were previously
59
only detectable by invasive karyotype analysis. In some instances, this non-invasive prenatal
60
test also provided additional clarification about the origin of genetic material that had not been
61
identified by invasive karyotype analysis.
62 63
2. INTRODUCTION
64
Since its introduction in 2011, non-invasive prenatal tests (NIPT) have had a significant impact
65
on prenatal care. In only four years, NIPT has evolved into a standard option for high-risk
66
pregnancies 1. Content has also evolved from exclusive trisomy 21 (T21) testing to include
67
trisomy 18 (T18), trisomy 13 (T13), sex chromosome aneuploidies (SCAs), and select
68
microdeletions. This ‘standard’ content can be expected to detect 80-83% of chromosomal
69
abnormalities detected by karyotyping in a general screening population 2-4, however this leaves
70
a gap of approximately 17-20% of alternative chromosomal/sub-chromosomal abnormalities not
71
detected. Consequently, obtaining comprehensive information about the genetic makeup of the
AC C
EP
TE D
M AN U
SC
RI PT
47
ACCEPTED MANUSCRIPT
fetus requires an invasive procedure. To overcome these limitations, NIPT could be extended to
73
cover the entire genome. However, it is challenging to maintain a very high specificity and
74
positive predictive value when interrogating all accessible regions in the genome 5. In previous
75
reports, we have overcome these technical hurdles 6. Furthermore, a recent study by Yin et al.
76
demonstrated feasibility for non-invasive genome-wide detection of sub-chromosomal
77
abnormalities 7. In this report, we have improved the assay and statistical methods to enable
78
comprehensive genome-wide detection of copy number variants (CNVs) ≥7 Mb. We present
79
results of a large blinded clinical study of more than 1200 samples including more than 100
80
samples with common aneuploidies detectable by traditional NIPT and over 30 samples
81
affected by sub-chromosomal CNVs.
82
3. MATERIALS and METHODS
83 84
STUDY DESIGN
85
risk for fetal aneuploidy based on advanced maternal age ≥35, a positive serum screen, an
86
abnormal ultrasound finding, and/or a history of aneuploidy. Archived samples were selected for
87
inclusion in the study by an unblinded internal third party according to the requirements
88
documented in the study plan. Samples were then blind-coded to all operators and the analysts
89
who processed the samples. After sequencing, an automated bioinformatics analysis was
90
performed to detect whole chromosome aneuploidies and sub-chromosomal CNVs. Results
91
were compiled electronically and were reviewed by a subject matter expert who assigned the
92
final classification. This manual review mimics the process in the clinical laboratory, where
93
cases are reviewed by a laboratory director before a result is signed out. Completed
94
classification results were provided to the internal third party for determination of concordance.
95
Analyzed samples had confirmation of positive or negative events by either G-band karyotype or
M AN U
SC
RI PT
72
AC C
EP
TE D
This blinded, retrospective clinical study included samples from women considered at increased
ACCEPTED MANUSCRIPT
microarray findings from samples collected through either CVS or amniocentesis. Circulating
97
cell-free “fetal” DNA is believed to originate largely from placental trophoblasts. Genetic
98
differences between the fetus and the placenta can occur (e.g. confined placental mosaicism),
99
leading to discordance between NIPT results and cytogenetic studies on amniocytes or
RI PT
96
postnatally obtained samples 8. Results from CVS by chromosomal microarray were thus
101
considered the most accurate ground truth. Therefore, discordant results originating from
102
amniocytes (karyotype or microarray) were resolved by sequencing at high coverage (an
103
average of 226 million reads per sample). Sequencing depth has been shown as the limiting
104
factor in NIPT methods, with increased depth allowing improved detection of events in samples
105
with lower fetal fractions or improved detection of smaller events 9. High coverage sequencing
106
has been used in multiple studies to unambiguously identify sub-chromosomal events 6,10-12 and
107
was used here as a reference for performance evaluation in discrepant amniocentesis samples.
108
Details of the sample demographics are described in Table 1. Indications for invasive testing are
109
described in Table 2.
110 111
SAMPLE COLLECTION
112
Review Board (IRB) approved protocols with a small subset (9 samples) comprising remnant
113
plasma samples collected from previously consented patients in accordance with the FDA
114
Guidance on Informed Consent for In Vitro Diagnostic Devices Using Leftover Human
115
Specimens that are Not Individually Identifiable (25 April 2006). Samples from two of the
116
protocols (Compass IRB #00508 and WIRB #20120148) were collected from high risk pregnant
117
subjects prior to undergoing a confirmatory invasive procedure (1189 samples). Samples from
118
the other two protocols (Compass IRB #00351 and Columbia University IRB #AAAN9002) came
119
from subjects who were enrolled in the studies after receiving the fetal karyotype and/or
120
microarray results of a confirmatory invasive procedure (24 samples). All subjects provided
TE D
M AN U
SC
100
AC C
EP
In total, 1222 maternal plasma samples had been previously collected using four Investigational
ACCEPTED MANUSCRIPT
written informed consent prior to undergoing any study related procedures.A total of 5321 high
122
risk subjects were recruited into the 4 clinical studies indicated above at the time of sample
123
selection. To be eligible for inclusion into this study, subjects had to have met all protocol
124
inclusion and no exclusion criteria, have fetal outcome determined by karyotype and/or
125
microarray and have at least one plasma aliquot of ≥3.5mL obtained from whole blood collected
126
in a Streck BCT. There was no sample selection preference based on high risk indication. All
127
subjects meeting these selection criteria that had an abnormal fetal outcome as needed for the
128
study were identified and pulled from inventory. These were then supplemented with randomly
129
selected subjects with samples meeting the same selection criteria but with a normal karyotype
130
to reach the total of 1222 samples for testing.
131 132
LIBRARY PREPARATION, SEQUENCING, AND ANALYTICAL METHODS
133
increase signal, sequencing depth for this analysis was increased to target 32M reads per
134
sample. Sequencing reads were aligned to hg19 using Bowtie 2 14. The genome was then
135
partitioned into 50-kbp non-overlapping segments and the total number of reads per segment
136
was determined, by counting the number of reads with 5’ ends overlapping with a segment.
137
Segments with high read-count variability or low map-ability were excluded. The 50 kbp read-
138
counts were then normalized to remove coverage and GC biases, and other higher-order
139
artifacts using the methods previously described in Zhao et al 6.
140
The presence of fetal DNA was quantified using the regional counts of whole genome single-
141
end sequencing data as described by Kim et al. 15.
142 143
GENOME-WIDE DETECTION OF ABNORMALITIES
144
segmenting each chromosome into contiguous regions of equal copy number 16. A segment-
145
merging algorithm was then used to compensate for over segmentation by CBS when the
M AN U
SC
RI PT
121
AC C
EP
TE D
Libraries were prepared and quantified as described by Tynan et al.13. To reduce noise and
Circular binary segmentation (CBS) was used to identify CNVs throughout the entire genome by
ACCEPTED MANUSCRIPT
signal-to-noise ratio (SNR) was low 17. Z-scores were calculated for both CBS-identified CNVs
147
and whole chromosome variants by comparing the signal amplitude with a reference set of
148
samples in the same region. The measured Z-scores form part of an enhanced version of
149
Chromosomal Aberration DEcision Tree (CADET) previously described in detail in Zhao et al 6.
150
To further improve specificity of CNV detection, bootstrap analysis was performed as an
151
additional measure for the confidence of the candidate CNVs. The within sample read count
152
variability was compared to a normal population (represented by 371 euploid samples) and
153
quantified by bootstrap confidence level (BCL). In order to assess within sample variability, the
154
bootstrap resampling described below was applied to every candidate CNV.
155
For each identified segment within the CNV, the median shift of segment fraction from the
156
normal level across the chromosome was calculated. This median shift was then corrected to
157
create a read count baseline for bootstrapping. Next, a bootstrapped segment of the same
158
segment length as the candidate CNV was randomly sampled with replacement from the
159
baseline read counts. The median shift was then applied to this bootstrapped fragment. The
160
segment fraction of the bootstrapped fragment was calculated as follows:
TE D
M AN U
SC
RI PT
146
EP
=
∑ ℎ ∑ ℎ
This process was repeated 1000 times to generate a bootstrap distribution of segment fractions
162
for an affected population. A normal reference distribution was created based on the segment
163
fraction of the same location as the candidate CNV in 371 euploid samples. A threshold was
164
then calculated as the segment fraction that was at least 3.95 median absolute deviations away
165
from the median segment fraction of the reference distribution. Lastly, the BCL was calculated
166
as the proportion of bootstrap segments whose fractions had absolute z-statistics above the
167
significance threshold.
AC C
161
ACCEPTED MANUSCRIPT
168
A whole chromosome or sub-chromosomal abnormality is detected as described in Zhao et al.
169 170
4. RESULTS
171
samples were excluded because they had no or insufficient karyotype or microarray information
172
(see Figure 1) and three samples were excluded because of confirmed mosaicism.
173
Forty-two (42) of the remaining 1208 samples were flagged as non-reportable using quality
174
criteria that had been established prior to analysis, leaving 1166 reportable samples for
175
analyses comprising 13% (n=153) samples with common whole chromosome 21, 18, 13, X, or
176
Y aneuploidies and 3.6% (n=43) samples with sub-chromosomal CNVs or rarer whole
177
chromosome aneuploidies. Technical failure criteria included but were not limited to: low library
178
concentration, low raw autosomal counts, high GC bias, poor normalization, and high bin
179
variability (see Figure 1). Biological failure criteria included low fetal fraction (less than 4%) and
180
large maternal CNV events. The most common reason for failure was low fetal fraction (n= 11).
181
During review of the data, one sample was signed out as T18 (and was included in the analyzed
182
cohort), even though it did not meet the autosomal count minimum. This sample had sufficient
183
counts for the determination of standard aneuploidies, but not sufficient counts for the detection
184
of sub-chromosomal CNVs. The 42 non-reportable samples showed no evidence of enrichment
185
for whole-chromosome/sub-chromosomal positive samples (5 positive of 42 non-reportable
186
samples vs. 204 positive of 1166 reportable samples). Details of non-reportable and excluded
187
samples are provided in Supplemental Table 2. Concordant with previous studies, the overall
188
reportable rate on first aliquots of maternal plasma was 96.5%.The overall no call rate per
189
patient is expected to be approximately 1% when a second aliquot is available, based on
190
published clinical laboratory experience 18.
AC C
EP
TE D
M AN U
SC
RI PT
The study comprised a total of 1222 maternal plasma samples. After unblinding, eleven
ACCEPTED MANUSCRIPT
191 192
T21, T18, and T13 Detection
193
samples (see Figure 1). All euploid, T21, T18, and T13 samples (determined by invasive
194
diagnostic procedures) were classified correctly by NIPT. One sample was classified by NIPT
195
as T21 but had a normal (46, XX) karyotype by amniocentesis (see Supplemental Table 1). This
196
discrepancy was adjudicated through high coverage ‘uniplex’ sequencing (typically with >180
197
million reads) according to our study plan (see Supplemental Figure 1D). The sample showed
198
16.3% fetal fraction, but the gain of genetic material from chromosome 21 was concordant with
199
6.5% fetal fraction. This is suggestive of confined placental mosaicism, a relatively common
200
cause of discordant results between amniocyte results and CVS or placentally-derived cell free
201
DNA analysis 8. Table 3 summarizes the performance for T21, T18, and T13. For T21, the
202
sensitivity was 100% (95% CI: 94.6%-100%) and the specificity was 100% (95% CI: 99.6%-
203
100%). For T18, the sensitivity was 100% (95% CI: 84.4%-100%) and the specificity was 100%
204
(95% CI: 99.6%-100%). For T13, the sensitivity was 100% (95% CI: 74.7%-100%) and the
205
specificity was 100% (95% CI: 99.6%-100%).
206 207
SCA Detection
208
were flagged as non-reportable specifically for SCA classification based on thresholds for the
209
chromosome X z-score and chromosome Y z-score as described by Mazloom et al. 19. There
210
was also one additional sample with an apparent maternal XXX that was flagged as non-
211
reportable for SCA because the maternal event distorted the sex chromosomal representation
212
to a degree that fetal events could not be classified. Among the remaining 1144 samples
213
reportable for SCAs, there were 7 discordant positives that were classified as normal (46, XX)
214
by karyotype and as XO by sequencing at 6-plex (see Supplemental Table 1). In all discordant
215
cases, the karyotype had been obtained from amniocyte samples. This phenomenon is well
216
described and may be attributed to varying levels of placental or maternal mosaicism 20. Uniplex
TE D
M AN U
SC
RI PT
Among the 1166 reportable samples, there were 85 T21 samples, 27 T18 samples, and 15 T13
AC C
EP
In the 1166 samples reportable for all chromosomal abnormalities, there were 21 samples that
ACCEPTED MANUSCRIPT
sequencing confirmed 6 of the 7 XO samples. The 7th sample had a non-reportable result at
218
uniplex coverage, hence the existing amniocentesis result was used as truth resulting in one
219
false positive assignment. Overall, the sensitivity for SCA was 100% (95% CI: 84.0%-100%)
220
with a specificity of 99.9% (95% CI: 99.4%-100%) (see Table 3).
221 222
Genome Wide Detection of CNVs
223
chromosome abnormalities other than T13, T18, T21, and SCAs); as well as select
224
microdeletions smaller than 7 Mb 5. Among the 1166 samples reportable for sub-chromosomal
225
abnormalities, there were 43 samples that had positive results for a variety of CNV aberrations:
226
Wolf-Hirschhorn syndrome deletions, DiGeorge syndrome deletions, Prader-Willi/Angelman
227
syndrome deletions, Cri du Chat syndrome deletions, and a variety of ≥7 Mb CNVs including
228
both sub-chromosomal CNVs and whole chromosome trisomies. Several examples of NIPT
229
detected CNVs confirmed by microarray or karyotype are shown in Figure 2.
230
Overall, the sensitivity for detection of whole chromosome and sub-chromosomal abnormalities
231
other than T13, T18, T21, and SCAs was 97.7% (95% CI: 86.2%-99.9%) and the specificity was
232
99.9% (95% CI: 99.4%-100%) (see Table 3). One case was clearly mosaic for T22 by both
233
standard coverage and uniplex sequencing (see Supplemental Table 1 and Supplement Figure
234
1E), but was classified as normal by microarray analysis. Because the invasive diagnosis came
235
from microarray analysis of cells derived from CVS, our study design considered this as the gold
236
standard, and this outcome was considered a discordant positive. Another case showed no gain
237
or loss of genetic material with both standard and uniplex sequencing for a sample that had a
238
46, XX, der(12)t(12;19)(p13.1;q13.1) karyotype (see Supplemental Figure 2B). This outcome
239
was considered as a discordant negative given that the invasive procedure was CVS. A set of 7
240
samples were classified as full chromosomal trisomies, and because the karyotype or
241
microarray results were derived from amniocytes in these cases, uniplex sequencing was
RI PT
217
AC C
EP
TE D
M AN U
SC
The test was also designed to detect CNVs equal to or larger than 7 Mb (including whole
ACCEPTED MANUSCRIPT
performed for adjudication per the study design. The trisomy finding was confirmed in each case
243
by uniplex sequencing, and these findings were considered as concordant positives
244
(Supplemental Figure 1).
245
In addition to the high accuracy demonstrated in the results, in some cases NIPT could also
246
provide clarification about the origin of extra genetic material when the G-banding pattern was
247
not sufficiently clear. In one case, the amniocyte karyotype finding, 46, XX, der(5)t(5;?)(p15.3;?),
248
indicated a deletion on chromosome 5 and a duplication of unknown origin (see Figure 3). NIPT
249
identified an 18.6 Mb duplication representing the entire short arm of chromosome 17, indicating
250
this fetal tissue possessed a trisomy of chromosome 17p that was not clarified by standard
251
karyotype. Trisomy 17p is associated with developmental delay, growth retardation, hypotonia,
252
digital abnormalities, congenital heart defects, and distinctive facial features 21. Additional cases
253
are shown in Supplemental Figures 3.
254 255
5. COMMENT
256
baby have multiple options today – ranging from non-invasive tests that screen for select
257
chromosomal abnormalities to invasive procedures including karyotype and microarray testing
258
that can deliver the most comprehensive genomic assessment. Current NIPT methods provide
259
information about a limited set of conditions, that typically include T21, T18, T13 and SCAs. In
260
this study we expand testing to the entire genome. The results demonstrate that high sensitivity
261
and specificity were maintained while adding genome-wide clinically relevant content.
262
Most NIPT tests suffer from the problem of multiple hypothesis testing; when multiple regions of
263
the genome are tested independently, false positive rates become additive 22. This can result in
264
unacceptably high false positive rates and lead to maternal anxiety and additional unnecessary
265
clinical testing. Methods that target individual regions of the genome cannot overcome these
TE D
M AN U
SC
RI PT
242
AC C
EP
Couples and families who are seeking prenatal information about the genetic health of their
ACCEPTED MANUSCRIPT
limitations because they are part of the design of the test 23. We have shown that these
267
limitations can be overcome by using a genome-wide approach, as described here and
268
previously 6. In this study we demonstrate that the test can provide excellent performance for
269
CNVs ≥7Mb and select microdeletions <7Mb, across the genome.
270
This study does have limitations. Although we analyze 2- to 6-fold more samples with sub-
271
chromosomal CNVs than previous studies 6,11,23, the total number of affected samples is still
272
relatively small compared to studies validating performance for detection common whole-
273
chromosome aneuploidies 24. It remains challenging to statistically validate detection of rare
274
individual chromosomal variations. As such we utilize samples with CNVs of varying size and
275
location throughout the genome with varying frequency of individual occurrence. This study
276
design with its high positivity rate allows only estimation of PPV.. As sensitivity and specificity
277
values have proven stable in previous studies when applied to the real world samples, PPV can
278
be considered a function of incidence rate. Incidence estimates of larger karyotype detectable
279
genome-wide sub-chromosomal CNVs vary but are typically considered between 1 in 200 and 1
280
in 40025. Modeled PPVs for these incidence rates would therefore be between 68% and 83%
281
(see Supplemental Figure 4). Due to the high sensitivity observed, negative predictive values
282
are expected to be well over 99.9% when incidence rates are less than 1 in 200. As our study
283
utilized only one aliquot for analysis, our reportable rate was 96.5%, with an expected reportable
284
rate of ~99% with availability of two aliquots18. Of the 11 samples not reported due to the
285
biological effect of low fetal fraction, all had normal karyotype result.
SC
M AN U
TE D
EP
AC C
286
RI PT
266
Another challenge commonly observed in novel technologies with increased performance is
287
the compilation of samples with appropriate ground truth. In our study we face the technical
288
challenge of comparing sequencing versus G-band karyotype or microarray, as well as the
289
biological aspect of cfDNA versus amniocyte or CVS. We have used a method that prioritizes
290
CVS over amniocentesis because of the placental origin of cell free DNA 8. Factors that can
ACCEPTED MANUSCRIPT
lead to discordant results include confined placental mosaicism, maternal mosaicism, vanishing
292
twins, undetected tumors, and technical errors. Consequently our method of evaluation has the
293
potential to overestimate performance with regard to fetal outcome. However, knowledge about
294
the genetic health of the placenta might in itself be advantageous when guiding the pregnancy,
295
for example in the case of intra uterine growth restriction. Prospective studies using the method
296
described here with follow-up of birth outcome will further enhance understanding of clinical
297
effectiveness.
298
An interesting observation in this study was the sometimes subjective nature of karyotyping.
299
When G-banding is used as an analytical method, the optical resolution is occasionally not
300
sufficient to determine the origin of additional genetic material, making a clinical interpretation
301
more difficult. These ambiguities are eliminated when using NIPT because the test requires
302
sequencing alignment to the genome before DNA gains and losses are determined. However,
303
G-banding will, for the foreseeable future, be the superior methodology to determine copy
304
number neutral structural changes.
305
In pregnancies that can benefit from additional information, this test provides more clinically
306
relevant results than previous NIPT options. However, its role as a follow-up test to abnormal
307
ultrasound findings or as a general population screen will likely be debated for the foreseeable
308
future. It has been argued that the incidence of clinically relevant sub-chromosomal CNVs could
309
be as high as 1.7% in a general population 25, and early microarray data 26 indicate that as many
310
as 30% of sub-chromosomal CNVs could be >7Mb. In consequence, the incidence for these
311
large CNVs may be around 0.5% in the general population, substantially higher than the
312
incidence of T21.
313
In summary, this clinical study provides validation for an approach that extends the clinical
314
validity of cfDNA testing, now providing detection of T21, T18, T13, SCAs, fetal sex, and
AC C
EP
TE D
M AN U
SC
RI PT
291
ACCEPTED MANUSCRIPT
genome-wide detection of sub-chromosomal and whole chromosomal abnormalities. Overall,
316
sub-chromosomal abnormalities and aneuploidies other than T13, T18, T21, and SCAs were
317
detected with a combined clinical sensitivity and specificity of 97.7% (95%CI 86.2-99.9%) and
318
99.9% (95%CI 99.4-100%), respectively. This enables comprehensive non-invasive
319
chromosomal assessment that was previously available only by karyotype, and in some cases,
320
may clarify cryptic findings otherwise identifiable only by microarray.
RI PT
315
SC
321
Acknowledgements
323
The authors would like to acknowledge the efforts of those who aided completion of this
324
study: Nathan Faulkner and Ashley Diaz for sample procurement; Sarah Ahn,
325
Christine Safanayong, Haiping Lu, Joshua White, Lisa Chamberlain for sample
326
processing and DNA extraction; David Wong, Jana Schroth, Marc Wycoco, Jesse Fox
327
for library preparation and sequencing, Pat Liu for bioinformatics pipeline processing.
328
We would also like to acknowledge the contribution of samples by Ronald Wapner at
329
Columbia University.
AC C
EP
TE D
M AN U
322
ACCEPTED MANUSCRIPT
6. REFERENCES
331 332
1.
ACOG. Cell-free DNA screening for fetal aneuploidy. Committee Opinion No. 640. Obstetrics & Gynecology. 2015;126:e31-37.
333 334 335 336
2.
Norton ME, Wapner RJ, Kuppermann M, Jelliffe-Pawlowski LL, Currier RJ. Cell free DNA analysis vs sequential screening as primary testing considering all fetal chromosomal abnormalities. 35th Annual Meeting of the Society for Maternal-Fetal Medicine: The Pregnancy Meeting. 02 February 2015 - 07 February 2015 2015;212(1):S2.
337 338 339
3.
Norton ME, Jelliffe-Pawlowski LL, Currier RJ. Chromosome abnormalities detected by current prenatal screening and noninvasive prenatal testing. Obstetrics & Gynecology. 2014;124(5):979-986.
340 341 342
4.
Wellesley D, Dolk H, Boyd PA, et al. Rare chromosome abnormalities, prevalence and prenatal diagnosis rates from population-based congenital anomaly registers in Europe. European Journal of Human Genetics. 2012;20(5):521-526.
343 344
5.
Helgeson J, Wardrop J, Boomer T, et al. Clinical outcome of subchromosomal events detected by whole‐ ‐genome noninvasive prenatal testing. Prenatal Diagnosis. 2015;35(10):999-1004.
345 346 347
6.
Zhao C, Tynan J, Ehrich M, et al. Detection of Fetal Subchromosomal Abnormalities by Sequencing Circulating Cell-Free DNA from Maternal Plasma. Clinical Chemistry. 2015;61(4):608-616.
348 349 350
7.
Yin A-h, Peng C-f, Zhao X, et al. Noninvasive detection of fetal subchromosomal abnormalities by semiconductor sequencing of maternal plasma DNA. Proceedings of the National Academy of Sciences. 2015:201518151.
351 352 353
8.
Grati FR, Malvestiti F, Ferreira JC, et al. Fetoplacental mosaicism: potential implications for false-positive and false-negative noninvasive prenatal screening results. Genetics in Medicine. 2014;16(8):620-624.
354 355 356
9.
Fan HC, Quake SR. Sensitivity of noninvasive prenatal detection of fetal aneuploidy from maternal plasma using shotgun sequencing is limited only by counting statistics. PLoS One. 2010;5(5):e10439.
357 358
10.
Jensen TJ, Kim SK, van den Boom D, Deciu C, Ehrich M. Noninvasive Detection of a Balanced Fetal Translocation from Maternal Plasma. Clinical Chemistry. 2014;60(10):1298-1305.
359 360 361
11.
362 363
12.
Peters D, Chu T, Yatsenko SA, et al. Noninvasive prenatal diagnosis of a fetal microdeletion syndrome. New England Journal of Medicine. 2011;365(19):1847-1848.
364 365 366
13.
Tynan J, Kim S, Mazloom A, et al. Application of risk score analysis to low‐ ‐coverage whole genome sequencing data for the noninvasive detection of trisomy 21, trisomy 18, and trisomy 13. Prenatal Diagnosis. 2015.
AC C
EP
TE D
M AN U
SC
RI PT
330
Srinivasan A, Bianchi DW, Huang H, Sehnert AJ, Rava RP. Noninvasive detection of fetal subchromosome abnormalities via deep sequencing of maternal plasma. The American Journal of Human Genetics. 2013;92(2):167-176.
ACCEPTED MANUSCRIPT
14.
Langmead B, Salzberg SL. Fast gapped-read alignment with Bowtie 2. Nature Methods. 2012;9(4):357-359.
369 370
15.
Kim SK, Hannum G, Geis J, et al. Determination of fetal DNA fraction from the plasma of pregnant women using sequence read counts. Prenatal Diagnosis. 2015;35(8):810-815.
371 372
16.
Olshen AB, Venkatraman E, Lucito R, Wigler M. Circular binary segmentation for the analysis of array‐ ‐based DNA copy number data. Biostatistics. 2004;5(4):557-572.
373 374
17.
Willenbrock H, Fridlyand J. A comparison study: applying segmentation to array CGH data for downstream analyses. Bioinformatics. 2005;21(22):4084-4091.
375 376
18.
McCullough RM, Almasri EA, Guan X, et al. Non-invasive prenatal chromosomal aneuploidy testing-clinical experience: 100,000 clinical samples. PLoS One. 2014;9(10):e109173.
377 378 379
19.
Mazloom AR, Džakula Ž, Oeth P, et al. Noninvasive prenatal detection of sex chromosomal aneuploidies by sequencing circulating cell‐ ‐free DNA from maternal plasma. Prenatal Diagnosis. 2013;33(6):591-597.
380 381 382
20.
Wang Y, Chen Y, Tian F, et al. Maternal mosaicism is a significant contributor to discordant sex chromosomal aneuploidies associated with noninvasive prenatal testing. Clinical Chemistry. 2014;60(1):251-259.
383 384
21.
Jinno Y, Matsuda I, Kajii T. Trisomy 17p due to a t(5; 17)(p15; p11) pat translocation. Annales de Genetique. 1981;25(2):123-125.
385
22.
Miller RG. Simultaneous statistical inference. 2nd ed. New York: Springer-Verlag; 1981.
386 387 388
23.
Wapner RJ, Babiarz JE, Levy B, et al. Expanding the scope of noninvasive prenatal testing: detection of fetal microdeletion syndromes. American Journal of Obstetrics and Gynecology. 2015;212(3):332.e331-339.
389 390 391
24.
Palomaki GE, Kloza EM, Lambert-Messerlian GM, et al. DNA sequencing of maternal plasma to detect Down syndrome: an international clinical validation study. Genetics in Medicine. 2011;13(11):913-920.
392 393
25.
Wapner RJ, Martin CL, Levy B, et al. Chromosomal microarray versus karyotyping for prenatal diagnosis. New England Journal of Medicine. 2012;367(23):2175-2184.
394 395 396
26.
398 399 400 401
SC
M AN U
TE D
EP
AC C
397
RI PT
367 368
Shaffer LG, Dabell MP, Fisher AJ, et al. Experience with microarray‐ ‐based comparative genomic hybridization for prenatal diagnosis in over 5000 pregnancies. Prenatal Diagnosis. 2012;32(10):976-985.
ACCEPTED MANUSCRIPT
402 403
TABLES
404
Table 1. Demographic and pregnancy-related data. Median
Maternal Age (Yrs) (Ntotal = 1177)
36.0
Gestational Age (Wks) (Ntotal =1183)
17
Maternal Weight (Lbs) (Ntotal =1168)
150 Percent
CVS
14.5
17.8-47 8-38
93-366
(Naffected/Ntotal) (175/1203)
SC
Procedure
Range
RI PT
Demographic
Amniocentesis Both
(1025/1203)
0.2
(3/1203)
Percent
(Naffected/Ntotal)
90.4
(1089/1205)
5.8
(70/1205)
M AN U
Confirmation
85.2
Karyotype Microarray
Both 3.8 (46/1205) Note: Choice of invasive procedure, choice of diagnostic test, and demographic data were not available for all 1208 samples included in the study. Ntotal refers to the number of samples where that data was available.
406
Table 2. Indications for testing. Euploid (n=1009)
T21 (n=87)
T18 (n=29)
T13 (n=15)
SCA (n=26)
Positive Serum Screening
404(40%)
36(41.4%)
13(44.8%)
4(26.7%)
3(11.5%)
Maternal Age > 35 Yrs
418(41.4%) 28(32.2%)
10(34.5%)
4(26.7%)
2(7.7%)
Ultrasound Abnormality
152(15.1%) 34(39.1%)
6(20.7%)
10(66.7%)
16(61.5%)
EP
Indication
AC C
407
TE D
405
CNV (n=44) 16(36.4%) 10(22.7%) 22(50%)
Total (n=1210) 476(39.3%) 472(39%) 239(19.8%)
Family History
33(3.3%)
4(4.6%)
1(3.4%)
0(0%)
1(3.8%)
1(2.3%)
40(3.3%)
Not Specified
85(8.4%)
11(12.6%)
5(17.2%)
2(13.3%)
4(15.4%)
2(4.5%)
109(9%)
Note: Each cell lists the count and percentage of the indication within a chromosome category. Samples may have multiple indications for testing and therefore do not sum to 100%. As 2 of the T18 samples also had an SCA, the total number of abnormalities and euploid samples sum to 1210.
ACCEPTED MANUSCRIPT
408
Table 3. Clinical performance for the indicated abnormalities and fetal sex. Concordant Discordant Concordant Discordant Positive Positive Negative Negative 85
0
1081
0
T18
27
0
1139
0
T13
15
0
1151
0
SCA
26
1
1117
CNVs**
42
1
1122
0
1
Concordant Discordant Concordant Discordant Male Male Female Female
Analyte Fetal Sex
583
4
578
1
Specificity (95% CI*)
100%
100%
(94.6%-100%)
(99.6%-100%)
100%
100%
(84.4%-100%)
(99.6%-100%)
100%
100%
(74.7%-100%)
(99.6%-100%)
SC
T21
Sensitivity (95% CI*)
RI PT
Abnormality
M AN U
409
100%
99.9%
(84.0%-100%)
(99.4%-100%)
97.7%
99.9%
(86.2%-99.9%)
(99.4%-100%)
Accuracy (95% CI*) 99.6% (98.9%-99.8%)
412 413 414 415 416 417 418 419 420
EP
411
AC C
410
TE D
* - CI indicates ‘confidence inter val’ **- CNVs include 8 samples with detected whole chromosome trisomies, and 35 samples with subchromosomal CNVs
ACCEPTED MANUSCRIPT
FIGURE LEGENDS
424 425 426 427 428
Figure 1. Flow chart showing sample exclusions and reasons for non-reportable samples. One of the trisomy 18 samples was also an XXY sample and one of the other trisomy 18 samples was also an X sample. For samples that had discordant results for sequencing versus karyotype or microarray derived from testing of amniocytes, uniplex sequencing was used for confirmation.
RI PT
422 423
429
SC
Figure 2. Examples of karyotype- and microarray-detected CNVs detected by sequencing. A) Chromosome 22 ideogram showing sequencing-based detection of a 22q11.21 deletion (DiGeorge syndrome) that was confirmed by karyotype and microarray. B) Chromosome 18 ideogram for a sample with a karyotype 46, XY, i(18)(q10). C) Chromosome 7 and 13 ideograms for a sample with a karyotype 46, XX, der(7)t(7;13)(q36;q22).
M AN U
430 431 432 433 434 435 436 437 438 439 440
Figure 3. Clarification of karyotype findings by sequencing of maternal plasma. Normal regions are colored blue while deletions and duplications that are detected with high bootstrap confidence levels (>0.99) are colored red. Low bootstrap confidence events are colored yellow. Chromosome 5 & 17 ideograms are shown for a sample with a karyotype of 46, XX, der(5)t(5;?)(p15.3;?). The unidentified duplication on the karyotype is from chromosome 17.
444 445 446 447
EP
443
AC C
442
TE D
441
AC C
EP
TE D
M AN U
SC
RI PT
ACCEPTED MANUSCRIPT
AC C
EP
TE D
M AN U
SC
RI PT
ACCEPTED MANUSCRIPT
AC C
EP
TE D
M AN U
SC
RI PT
ACCEPTED MANUSCRIPT
AC C
EP
TE D
M AN U
SC
RI PT
ACCEPTED MANUSCRIPT
AC C
EP
TE D
M AN U
SC
RI PT
ACCEPTED MANUSCRIPT
AC C
EP
TE D
M AN U
SC
RI PT
ACCEPTED MANUSCRIPT
AC C
EP
TE D
M AN U
SC
RI PT
ACCEPTED MANUSCRIPT
AC C
EP
TE D
M AN U
SC
RI PT
ACCEPTED MANUSCRIPT
ACCEPTED MANUSCRIPT
Supplemental Table 2. Breakdown of excluded and non-reported samples.
Biological exclusions
Fetal fraction
Maternal event Low aligned counts
CADET
0.093 0.114 0.081 0.148
54039
0.098
89063 73189 55348 53720 54528 89588 89473 77411 77188 89311 53415 53801 53103 55197 45424
0.096 0.089 0.040 0.036 0.039 0.036 0.032 0.019 0.028 0.037 0.027 0.036 0.037 0.065 0.149
GC warning
Lab Director review
RI PT
72938 71894 54681 72288
77478
0.079
77065 77066 77068 77042 77340 55462 89583 77040 93693 77392 89526 93679 89152 53815 73161 73017 71006 77075 54320 54212 77067 53061
0.087 0.092 0.083 0.109 0.131 0.078 0.082 0.058 0.112 0.091 0.105 0.074 0.076 0.077 0.161 0.158 0.072 0.104 0.048 0.147 0.072 0.043
AC C
Technical exclusions
High variability
0.094
Normal Normal Normal Other: T16 by NIPT. SCA 46-XY Normal karyotype; low level mosaic T21 by FISH Other: 46,XX,dup(1)(q21q32) - mosaic Other: 47,XX,+i(12)(p10)[13]/46,XX[2].nuc ish (ETV6x4,AML 1/20[25/25] - tetrasomy 12p mosaicism Other: 46,XY,var(18)(q11.2) Other: 46XX, 22qvar Other: 47, XX, +mar Other: 47, XX, +mar Other: 47, XX, +mar.ish idic(15;15)(q13;q13)(D15Z1++, SNRPN++) Other: 47, XX, +ESAC considered normal female Other: 47, XY, +ESAC Normal Normal Normal Normal Normal Normal Normal Normal Normal Normal Normal Normal T21 Normal
SC
Vague karyotype result
70600
Karyo/array Outcome
M AN U
Mosaicism
53063 77320 53863 RAS0027851 54172 77168
TE D
Exclusions
Missing invasive confirmation
Fetal fraction 0.130 0.079 0.068 0.065 0.093 0.247
SampleID
EP
Reason
T18 T21 Normal Normal Normal Normal Normal Normal Normal Normal Normal Normal Normal Normal Normal Normal Other: 46xx, add(18)(q21.3) Normal Normal Normal Normal Normal
ACCEPTED MANUSCRIPT
89529 77445 55123 89185 89607 53057 55444 54458 55220 93678 70268 53109 55454 55064 77285 55260
0.106 0.232 0.084 0.068 0.076 0.057 0.041 0.215 0.068 0.145 0.094 0.087 0.047 0.099 0.120 0.154
77157 73164 45364 89582 77476 89220
EP AC C
Normal Normal Normal Normal Normal Normal Normal Normal Normal Normal Normal Normal Normal Normal Normal Normal Other: 46,XX,der(18)(qter-q11.3-qter). Additional material added to 18p. Other: abnormal female karyotype with chromosome 13q deletion T18 T18 T21 T21
RI PT
0.112
SC
54227
Normal Normal Normal Other: 48,XXX,+18 Normal Normal
M AN U
0.064 0.094 0.075 0.092 0.148
0.042 0.085
0.102 0.099 0.108 0.130
TE D
Non-reportable for sex chromosome aneuploidy
Library failure
89359 89149 53420 55112 77347