Case 1:20-cv-01644-RGA Document 1-15 Filed 12/03/20 Page 1 of 9 PageID #: 855
`Case 1:20-cv-01644-RGA Document 1-15 Filed 12/03/20 Page 1 of 9 PageID #: 855
`
`EXHIBIT 15
`
`EXHIBIT 15
`
`
`
`
`
`

`

`Case 1:20-cv-01644-RGA Document 1-15 Filed 12/03/20 Page 2 of 9 PageID #: 856
`
`Articles
`
`Lancet 2007; 369: 474–81
`Published Online
`February 2, 2007
`DOI:10.1016/S0140-
`6736(07)60115-9
`See Comment page 440
`Ravgen Inc, Columbia, MD, USA
`(R Dhallan MD, X Guo PhD,
`S Emche PhD, J Barry BS,
`J Betz BS, K Franz BS, K Gold BS,
`B Vallecillo BS, J Varney BS);
`Department of Obstetrics and
`Gynecology, York Hospital/
`WellSpan Health, York, PA, USA
`(M Damewood MD);
`Department of Maternal Fetal
`Medicine, Lancaster General
`Women and Babies Hospital,
`Lancaster, PA, USA
`(P Bayliss MD); and Whyte
`Hirschboeck Dudek S.C.
`Madison, MI, USA
`(M M Cronin PhD)
`Correspondence to:
`Dr Ravinder Dhallan, Ravgen Inc,
`9241 Rumsey Rd, Columbia,
`MD 21045, USA
`rdhallan@ravgen.com
`
`A non-invasive test for prenatal diagnosis based on fetal
`DNA present in maternal blood: a preliminary study
`
`Ravinder Dhallan, Xin Guo, Sarah Emche, Marian Damewood, Philip Bayliss, Michael Cronin, Julie Barry, Jordan Betz, Kara Franz, Katie Gold,
`Brett Vallecillo, John Varney
`
`Summary
`Background Use of free fetal DNA to diagnose fetal chromosomal abnormalities has been hindered by the inability to
`distinguish fetal DNA from maternal DNA. Our aim was to establish whether single nucleotide polymorphisms
`(SNPs) can be used to distinguish fetal DNA from maternal DNA—and to determine the number of fetal
`chromosomes—in maternal blood samples.
`
`Methods Formaldehyde-treated blood samples from 60 pregnant women and the stated biological fathers were
`analysed. Maternal plasma fractions were quantifi ed at multiple SNPs, and the ratio of the unique fetal allele signal to
`the combined maternal and fetal allele signal calculated. The mean ratios of SNPs on chromosomes 13 and 21 were
`compared to test for potential fetal chromosomal abnormalities.
`
`Findings The mean proportion of free fetal DNA was 34·0% (median 32·5%, range 17·0–93·8). We identifi ed three
`samples with signifi cant diff erences in the fetal DNA ratios for chromosome 13 and chromosome 21, indicative of
`trisomy 21; the remaining 57 samples were deemed to be normal. Amniocentesis or newborn reports from the clinical
`sites confi rmed that the copy number of fetal chromosomes 13 and 21 was established correctly for 58 of the 60 samples,
`identifying 56 of the 57 normal samples, and two of the three trisomy 21 samples. Of the incorrectly identifi ed samples,
`one was a false negative and one was a false positive. The sensitivity and positive predictive value were both 66·7%
`(95% CI 12·5–98·2) and the specifi city and negative predictive values were both 98·2% (89·4–99·9).
`
`Interpretation The copy number of chromosomes of interest can be directly established from maternal plasma. Such
`a non-invasive prenatal test could provide a useful complement to currently used screening tests.
`
`Introduction
`Available protocols for prenatal diagnosis of aneuploidy
`are limited by several factors. Screening tests—eg,
`nuchal translucency and the quadruple screen—are
`non-invasive, but diagnosis requires further invasive
`testing.1 Invasive diagnostic tests—eg, amniocentesis
`and chorionic villus sampling—are about 99% accurate
`in
`identifying
`the
`spectrum of
`chromosomal
`abnormalities, but are associated with increased risks to
`the pregnancy.2–7 Development of non-invasive tests that
`yield diagnostic results would be a useful advancement
`in prenatal care.
`Analysis of fetal cells and free fetal DNA in the maternal
`circulation provides an alternative to existing prenatal
`tests.8–14 The use of free fetal DNA has been reported in
`the diagnosis of achondroplasia and myotonic dystrophy,
`to determine fetal sex,15–17 and in fetal rhesus D
`genotyping.18 However, two major issues have restricted
`the clinical use of analysis of free fetal DNA. First, little
`free fetal DNA exists in the maternal circulation, with
`initial studies reporting a mean of only 3·4% free fetal
`DNA in the late fi rst trimester to mid second trimester.19
`Second, in a heterogeneous mixture of maternal and fetal
`DNA it is diffi cult to distinguish fetal chromosomes of
`clinical interest—eg, chromosomes 13, 18, and 21—from
`maternal chromosomes.
`We have previously reported that careful sample
`processing and the addition of formaldehyde increased
`
`the proportion of free fetal DNA recovered from the
`maternal circulation to about 25%.20 Having increased
`the proportion of free fetal DNA, one major challenge
`remained: how to determine the copy number of fetal
`chromosomes in a heterogeneous mixture of maternal
`and fetal DNA.
`Sequencing of the human genome has led to the
`discovery of variation in base sequences in individuals,
`referred to as single nucleotide polymorphisms (SNPs).
`The use of SNPs for the detection of trisomy 21 has been
`described from amniotic fl uid specimens.21 Amniotic
`fl uid, however, contains a 100% sample of fetal DNA
`compared with the heterogeneous mixture of maternal
`and fetal DNA seen in maternal plasma. We postulated
`that an approach that used multiple SNPs, and
`quantifi cation of an allele ratio for these SNPs in a
`maternal blood sample, could indicate the presence or
`absence of fetal aneuploidy. This technique would require
`neither cell separation nor isolation of free fetal DNA.
`At certain SNP sites, the maternal genome will be
`homozygous for a nucleotide—eg, guanine (G/G)—while
`at the same site, the paternal genome might be
`homozygous for a diff erent nucleotide—eg, thymine
`(T/T; fi gure 1). Since one copy of each chromosome is
`inherited from each parent, the fetal genome will be
`heterozygous (G/T) at the SNP site (fi gure 1). In the
`plasma DNA of the maternal blood sample, the presence
`of thymine at the SNP site represents a unique fetal
`
`474
`
`www.thelancet.com Vol 369 February 10, 2007
`
`

`

`Case 1:20-cv-01644-RGA Document 1-15 Filed 12/03/20 Page 3 of 9 PageID #: 857
`
`Articles
`
`signal in the maternal DNA background (fi gure 1). Our
`aim was to establish a SNP ratio to identify fetal trisomy 21
`from maternal plasma and to develop an approach to
`quantify and compare allele ratios for SNP sites on
`chromosomes 13 and 21.
`
`Methods
`Procedures
`A network of ten clinical sites was established to gather
`specimens. Each site was involved in perinatal diagnosis
`of
`the patients
`involved. Clinical sites received
`institutional review board approval before patient
`enrolment. Patients and the stated biological fathers were
`aged 18 years or older; only patients with singleton
`pregnancies were included. Written informed consent
`was obtained before participation. All test results were
`compared with amniocentesis or newborn reports
`obtained from the clinical sites. The study was done
`between January, 2004, and August, 2006, during which
`time the technology was developed and refi ned, and
`60 samples were gathered and processed sequentially.
`Blood samples were taken from the pregnant women
`and also from the stated biological fathers. About 35 mL
`of blood was gathered from all patients (range 25–50 mL).
`0·225 mL of a 10% neutral buff ered solution containing
`formaldehyde (4% w/v) was added to all tubes of maternal
`blood immediately after the blood was drawn. Specimens
`were picked up the same day at regional clinical sites or
`shipped by commercial carrier for overnight delivery.
`Blood was stored at 4°C until processed. The samples
`were shipped to our central laboratory for analysis.
`Laboratory personnel were masked as to the identity of
`the samples with a numerical coding system.
`Plasma and buff y coat samples were isolated in
`accordance with methods described previously.20 Genomic
`DNA was purifi ed from both the plasma fraction and
`buff y coat of the same maternal blood sample, and the
`buff y coat fraction of the paternal sample, with the
`QIAamp DNA Blood Maxi Kit (Qiagen, Valencia, CA,
`USA). DNA samples were eluted in 2 mL of DNase/
`RNase-free water. Plasma DNA was concentrated to
`50–70 μL with a 10 kDa nominal molecular weight cutoff
`fi lter (Millipore, Bedford, MA, USA). The concentrated
`plasma DNA was split equally into three replicates that
`were processed separately for detection of fetal signals.
`10 μL of each replicate was amplifi ed with the GenomePlex
`Whole Genome Amplifi cation Kit (Sigma-Aldrich USA,
`St Louis, MO, USA).
`Before analysis of the plasma DNA, maternal and
`paternal buff y coat samples were analysed to identify two
`categories of SNPs. In category one, the maternal and
`paternal samples were homozygous for diff erent alleles
`at the SNP site. Therefore, the fetal DNA from the plasma
`was expected to be heterozygous at these sites. In category
`two, the maternal sample was homozygous, and the
`paternal sample was heterozygous at the same SNP site.
`Therefore, there was a 50% chance that the allele
`
`G
`
`T
`
`T
`
`G
`
`T
`
`Maternal
`
`G
`T
`Fetal
`
`Paternal
`
`
`
`A
`
`G
`
`B
`
`C
`
`Maternal buffy coat
`
`Paternal buffy coat
`
`Maternal plasma
`
`T
`
`G
`
`T
`
`G
`
`Figure 1: Inheritance of single nucleotide polymorphisms
`(A) Single nucleotide polymorphisms (SNPs) are the greatest single source of
`natural variation in the human genome and can be used to determine the copy
`number of fetal chromosomes. (B) At certain sites within the fetal genome, the
`inherited paternal allele (“T”) will diff er from the inherited maternal allele (“G”).
`(C) Presence of the paternal allele in the plasma provides a genetic beacon to
`distinguish fetal DNA from maternal DNA.
`
`inherited by the fetus would be diff erent from the
`maternal allele. Only SNPs showing a unique fetal allele
`in the maternal plasma were quantifi ed.
`SNPs were chosen so that their sequences fi t the
`following criteria: the nucleotide at ±2 bp from the SNP
`site matched one of the two alleles located at the SNP site,
`whereas the nucleotide at ±1 bp from the same site
`contained neither of the two alleles located at the SNP
`site.22 For each of the 60 samples, 549 SNPs were analysed
`on chromosome 13 and 570 SNPs were analysed on
`chromosome 21. Chromosome 21 was selected for analysis
`because it is commonly associated with fetal abnormalities,
`whereas chromosome 13 was chosen because it is less
`commonly associated with fetal abnormalities, and thus
`could also serve as a reference chromosome.
`SNPs were amplifi ed from genomic DNA isolated
`from the maternal plasma, the maternal buff y coat, and
`
`www.thelancet.com Vol 369 February 10, 2007
`
`475
`
`

`

`Case 1:20-cv-01644-RGA Document 1-15 Filed 12/03/20 Page 4 of 9 PageID #: 858
`
`Articles
`
`the paternal buff y coat by PCR. For all SNPs, the fi rst
`primer contained a recognition site for the restriction
`enzyme EcoRI and had a biotin tag at the 5´ end. The
`second primer contained a recognition site for a type
`IIS restriction enzyme.22 38 cycles of PCR were done as
`follows: (1) 95°C for 15 min, (2) 94°C for 30 s, (3) 37°C
`for 30 s, (4) 94°C for 30 s, (5) 52°C for 30 s,
`(6) 94°C for 30 s, (7) 58°C for 30 s, (8) repeat steps 6
`and 7 for 37 cycles, and (9) 72°C for 3 min.22 PCR
`products were bound to streptavidin-coated well plates
`for 1 h at 37°C with constant mixing at 400 rev per min
`with a Thermomixer (Eppendorf, Westbury, NY, USA).
`
`24% fetal DNA
`
`58% fetal DNA
`
`94% fetal DNA
`
`Maternal
`
`Paternal Plasma
`
`Maternal Paternal Plasma
`
`Maternal Paternal Plasma
`
`SNP64-24-289
`
`SNP64-51-1096
`
`SNP50-107-1412
`
`Chromosome 13
`
`Chromosome 21
`
`SNP52-92-3150
`
`SNP64-84-684
`
`SNP42-28-1628
`
`Figure 2: SNPs on chromosomes 13 and 21 at various fetal DNA percentages in maternal plasma samples
`In a maternal plasma sample, the inherited paternal allele represents a unique marker, which distinguishes fetal
`DNA from maternal DNA. A homozygous allele signal from maternal genomic DNA (maternal), a homozygous
`allele signal from paternal genomic DNA (paternal) and signals from heterozygous alleles in maternal plasma DNA
`(plasma) are shown. The 94% fetal DNA example is sample 4, which was identifi ed as trisomy 21. By contrast with
`the roughly 1 to 1 ratio of maternal (upper) and unique fetal (lower) alleles in the chromosome 13 panel (top), an
`indication of the presence of an additional, paternally inherited allele, is visible in the chromosome 21 panel
`(bottom), in which the ratio of maternal to unique fetal alleles is greater than 1 to 1.
`
`A Example of a SNP quantified on chromosome
`13 in a sample with trisomy 21
`
`Copies of chromosome 13
` Maternal Fetal
`
`T
`G
`
`G
`
`G
`
`Copies of chromosome 21
` Maternal Fetal
`
`C
`
`Maternal buffy coat
`Paternal buffy coat
`Maternal plasma
`
`T
`
`C
`
`T
`
`G
`
`C
`
`A
`
`G
`
`A
`
`Ratio:
` /
`1T
`3G
`
`Ratio:
` /
`1C
`4A
`
`B Example of a SNP quantified on
`chromosome 21 in a sample with trisomy 21
`
`A
`
`A
`
`A
`
`A
`
`Figure 3: Establishing the copy number of fetal chromosomes through analysis of allele ratios in the maternal
`plasma
`(A) Example of a SNP quantifi ed on chromosome 13 in a sample with trisomy 21; the ratio of unique fetal signal
`(“T”) to the combined maternal and fetal signal (“G”) in the plasma is 1 to 3 and over many SNPs, approaches a
`theoretical mean ratio of 0·333. (B) Example of a SNP quantifi ed on chromosome 21 in a sample with trisomy 21;
`the ratio of unique fetal signal to combined maternal and fetal signal is reduced to 1 to 4 and over many SNPs,
`approaches a theoretical mean ratio of 0·25; consequently, the mean ratio for chromosome 21 will be substantially
`lower than the mean ratio for chromosome 13.
`
`After binding, all wells were washed three times with
`phosphate-buff ered saline.
` Bound PCR products were digested for 1 h at 55°C
`with the appropriate type IIS restriction enzyme to yield
`bound DNA fragments with a specifi c 5´ overhang
`containing the SNP of interest, and a 3´ recessed end.
`Wells were rinsed three times with phosphate-buff ered
`saline. The bound DNA fragments were labelled at 55°C
`for 1 h with a mixture of fl uorescently labelled di-
`deoxynucleotide triphosphates, non-fl uorescently labelled
`deoxynucleotide triphosphates, and a DNA polymerase.
`Labelled PCR products were released from the wells by
`digestion with EcoRI for 1 h at 37°C with constant mixing
`at 400 rev per min.
`3 μL of the resulting digests from each well were
`loaded onto sequencing gels. Gels were run for 1 h at
`65 W constant power. Fluorescently labelled fragments
`were visualised with Typhoon 8600 and Typhoon 9400
`variable mode imagers (Amersham/GE Biosciences,
`Piscataway, NJ, USA). Maternal and paternal DNA
`samples isolated from the corresponding buff y coats
`were loaded next to the three corresponding plasma
`DNA replicates. Bands representing each allele in the
`plasma sample (upper or lower), were delimited and the
`pixel density of each band was quantifi ed with
`ImageQuant version 5.2 (Amersham/GE Biosciences,
`Piscataway, NJ, USA). Pixel density measurements were
`used to calculate a ratio of unique fetal allele signal to
`the combined maternal and fetal allele signal.
`Figure 2 shows examples of category one SNPs on
`chromosomes 13 and 21 at diff erent percentages of fetal
`DNA in the maternal circulation, showing allele signals
`from maternal buff y coat DNA, paternal buff y coat
`DNA, and maternal plasma DNA. The unique fetal
`DNA signal in the plasma matches the paternal allele,
`and is opposite the maternal allele. The ratio of the
`unique fetal allele signal to the combined signal from
`the maternal and fetal alleles was calculated at multiple
`SNP sites on chromosomes 13 and 21. Figure 3 illus-
`trates how
`ratios at
`individual SNP sites on
`chromosome 13 and chromosome 21 were used to
`calculate the ratio of fetal to maternal DNA in the
`maternal plasma. This example is based on the
`quantifi cation of a sample with trisomy 21 containing
`50% fetal DNA in the maternal plasma.
`Figure 3A shows the quantifi cation of a SNP on
`chromosome 13 in a sample with trisomy 21. In this
`example, the fetus inherits one maternal and one
`paternal copy of chromosome 13. Therefore, the ratio of
`unique fetal allele signal to the combined maternal and
`fetal allele signal in the plasma is 1 to 3. At all SNP sites
`on chromosome 13, the ratio would be expected to
`approach 1 to 3, and over multiple SNP sites, the mean
`ratio would be very close to 0·333. Figure 3B shows the
`quantifi cation of a SNP on chromosome 21 in a sample
`with trisomy 21. Because the fetus inherits two maternal
`copies and one paternal copy of chromosome 21, the
`
`476
`
`www.thelancet.com Vol 369 February 10, 2007
`
`

`

`Case 1:20-cv-01644-RGA Document 1-15 Filed 12/03/20 Page 5 of 9 PageID #: 859
`
`Articles
`
`ratio of unique fetal allele signal to the combined
`maternal and fetal allele signal would be expected to be
`1 to 4 (ie, 0·25). A comparison of mean fetal DNA ratios
`for the two chromosomes would show that the mean
`ratio for chromosome 21 is signifi cantly lower than the
`mean ratio for chromosome 13, thus
`indicating
`trisomy 21.
`The fetal DNA ratio (R) was used to calculate the
`percentage (P) of fetal DNA in the plasma with the
`formula: P=(2R/[1+R])×100%. The mean log ratio for all
`SNPs on chromosome 13 and the mean log ratio for all
`SNPs on chromosome 21 from the three independent
`plasma replicates were calculated to test for diff erences
`between the two chromosomes.22,23
`
`Statistical analysis
`All category one and two SNPs that were identifi ed in the
`plasma DNA were quantifi ed, meaning that for every
`sample, a diff erent number of SNPs were quantifi ed per
`chromosome. The natural log of the raw ratios was
`calculated and averaged across replicates, and then
`averaged across SNPs for each chromosome. The mean
`log ratio of fetal DNA between chromosomes 13 and 21
`was compared by a two-tailed Student’s t test allowing for
`unequal variances. The diff erence between the ratio of
`chromosome 13 and chromosome 21 fetal DNA was
`deemed to be signifi cant when the signifi cance level of
`the t test was less than 0·05. This process was repeated
`separately for data from all patient samples. All
`calculations were done with S-Plus version 7.0.
`
`Role of the funding source
`Ravgen, Inc designed and conducted the study; collected,
`managed, analysed, and interpreted the data; and prepared,
`reviewed, and submitted the manuscript. M Damewood
`had access to all the data in this study and takes
`responsibility for the integrity of the data and the integrity
`of the data analysis, and is not an employee of Ravgen,
`Inc. The corresponding author had fi nal responsibility for
`the decision to submit the manuscript for publication.
`
`Results
`The median maternal age was 34 years; the median
`gestational age was 17 weeks and 5 days (table 1). The
`earliest gestational age at which the test was done was
`just over 8 weeks (sample 46; table 1). Eight of the samples
`were drawn in the fi rst trimester.
`The mean proportion of free fetal DNA was 34·0%
`(median 32·5%, range 17·0–93·8; table 2). 51 of 60
`samples had more than 25% free fetal DNA. Of these
`samples, three had 50% or more free fetal DNA.
`Table 2 shows the number of SNPs analysed on each
`chromosome for all samples. All category one and two
`SNPs identifi ed in the plasma DNA were quantifi ed. The
`number of SNPs analysed in the plasma fraction varied
`by sample, but a mean of 22 SNPs were analysed on
`chromosome 13, and 20 SNPs were analysed on
`
`1
`2
`3
`4
`5
`6
`7
`8
`9
`10
`11
`12
`13
`14
`15
`16
`17
`18
`19
`20
`21
`22
`23
`24
`25
`26
`27
`28
`29
`30
`31
`32
`33
`34
`35
`36
`37
`38
`39
`40
`41
`42
`43
`44
`45
`46
`47
`48
`49
`50
`51
`52
`
`Maternal age (years) Gestational age (weeks and days)
`
`36
`27
`34
`43
`20
`41
`30
`37
`35
`29
`37
`37
`39
`27
`34
`40
`41
`36
`35
`31
`33
`29
`25
`35
`22
`34
`36
`38
`30
`30
`36
`37
`31
`34
`42
`24
`32
`31
`35
`38
`34
`34
`30
`32
`27
`30
`33
`34
`34
`33
`34
`18
`
`16 weeks 0 days
`26 weeks 6 days
`18 weeks 1 day
`20 weeks 1 day
`18 weeks 5 days
`16 weeks 4 days
`19 weeks 1 day
`17 weeks 1 day
`16 weeks 3 days
`17 weeks 0 days
`16 weeks 6 days
`19 weeks 1 day
`17 weeks 2 days
`17 weeks 4 days
`17 weeks 3 days
`18 weeks 1 day
`17 weeks 5 days
`17 weeks 5 days
`18 weeks 1 day
`21 weeks 2 days
`15 weeks 6 days
`20 weeks 0 days
`36 weeks 5 days
`17 weeks 0 days
`14 weeks 1 day
`22 weeks 4 days
`35 weeks 6 days
`16 weeks 6 days
`15 weeks 6 days
`11 weeks 4 days
`19 weeks 2 days
`17 weeks 4 days
`16 weeks 3 days
`29 weeks 5 days
`11 weeks 6 days
`13 weeks 6 days
`16 weeks 1 day
`32 weeks 1 day
`18 weeks 6 days
`12 weeks 1 day
`20 weeks 0 day
`21 weeks 2 days
`12 weeks 2 days
`37 weeks 2 days
`26 weeks 1 day
`8 weeks 1 day
`17 weeks 0 days
`27 weeks 5 days
`11 weeks 1 day
`15 weeks 1 day
`18 weeks 3 days
`35 weeks 6 days
`(Continues on next page)
`
`www.thelancet.com Vol 369 February 10, 2007
`
`477
`
`

`

`Case 1:20-cv-01644-RGA Document 1-15 Filed 12/03/20 Page 6 of 9 PageID #: 860
`
`Articles
`
`chromosome 21. Signifi cant diff erences in the ratio of
`fetal to maternal DNA were seen in samples 4 (p=0·04),
`18 (p=0·05), and 31 (p=0·04).
`
`(Continued from previous page)
`53
`31
`54
`21
`55
`33
`56
`43
`57
`31
`58
`34
`59
`26
`60
`36
`Median
`34
`
`16 weeks 5 days
`28 weeks 0 days
`16 weeks 4 days
`36 weeks 6 days
`38 weeks 6 days
`10 weeks 6 days
`26 weeks 1 day
`28 weeks 0 days
`17 weeks 5 days
`
`Table 1: Maternal age and gestational age for the 60 patient samples
`
`Chromosome Number of
`SNPs
`quantifi ed
`
`Ratio of fetal
`to maternal
`DNA
`
`Diff erence in
`fetal DNA
`ratio (13 vs 21)
`
`Percentage
`fetal DNA
`
`p value*
`
`1
`
`2
`
`3
`
`4
`
`5
`
`6
`
`7
`
`8
`
`9
`
`10
`
`11
`
`12
`
`13
`
`14
`
`15
`
`16
`
`17
`
`13
`21
`13
`21
`13
`21
`13
`21
`13
`21
`13
`21
`13
`21
`13
`21
`13
`21
`13
`21
`13
`21
`13
`21
`13
`21
`13
`21
`13
`21
`13
`21
`13
`21
`
`13
`11
`40
`33
`27
`29
`46
`35
`18
`23
`24
`24
`13
`21
`16
`14
`22
`24
`28
`17
`18
`23
`12
`8
`41
`17
`19
`17
`17
`30
`7
`15
`25
`11
`
`0·3209
`0·2249
`0·4032
`0·4453
`0·1959
`0·2219
`0·8826
`1·2446
`0·1489
`0·1514
`0·0928
`0·0811
`0·1468
`0·1528
`0·1217
`0·1315
`0·1528
`0·1436
`0·1486
`0·1919
`0·2335
`0·1732
`0·1504
`0·1288
`0·1576
`0·1996
`0·0994
`0·1093
`0·1347
`0·1151
`0·2126
`0·1599
`0·1478
`0·2174
`
`0·0960
`
`48·6%
`
`–0·0421
`
`57·5%
`
`–0·0260
`
`32·8%
`
`0·15
`
`0·43
`
`0·44
`
`–0·3621
`
`93·8%
`
` 0·04†
`
`–0·0025
`
`25·9%
`
`0·0117
`
`17·0%
`
`–0·0060
`
`25·6%
`
`–0·0098
`
`21·7%
`
`0·0092
`
`26·5%
`
`–0·0433
`
`25·9%
`
`0·0604
`
`37·9%
`
`0·0217
`
`26·2%
`
`–0·0420
`
`27·2%
`
`–0·0099
`
`18·1%
`
`0·0196
`
`0·0528
`
`23·7%
`
`35·1%
`
`–0·0696
`
`25·7%
`
`0·95
`
`0·45
`
`0·88
`
`0·75
`
`0·74
`
`0·27
`
`0·12
`
`0·59
`
`0·16
`
`0·68
`
`0·42
`
`0·19
`
`0·10
`
`(Continues on next page)
`
`Amniocentesis or newborn reports from the clinical
`sites confi rmed
`that
`the copy number of fetal
`chromosomes 13 and 21 was determined correctly for
`58 out of the 60 samples analysed; our method correctly
`identifi ed 56 of the 57 normal samples, and two of the
`three trisomy 21 samples (samples 4 and 31). Sample 18
`was a false positive (p=0·05) on the basis of a negative
`amniocentesis report for the sample. Sample 55, which
`was identifi ed as trisomy 21 by amniocentesis was
`falsely identifi ed as a normal sample by our methods
`(ie, was a false negative; p=0·34). The sensitivity of our
`test was 66·7% (95% CI 12·5–98·2), specifi city was
`98·2% (89·4–99·9), positive predictive value was 66·7%
`(12·5–98·2), and the negative predictive value was
`98·2% (89·4–99·9). Of the 57 true normal samples, the
`median maternal age was 34 years and the median
`gestational age was 17 weeks and 4 days. Of the three
`true abnormal samples, the median maternal age was
`36 years and the median gestational age was 20 weeks
`and 1 day (table 1).
`In sample 4, the mean ratio of fetal DNA for
`chromosome 21 was signifi cantly higher than that for
`chromosome 13, indicating trisomy 21 in the fetus, and
`that the additional copy was inherited from the paternal
`genome (p=0·04; table 2). In sample 31, the mean ratio of
`fetal DNA for chromosome 21 was signifi cantly lower
`than that for chromosome 13, indicating trisomy 21 in
`the fetus, and that the additional copy was inherited from
`the maternal genome (p=0·04; table 2).
`
`Discussion
`By use of this method to detect and quantify fetal DNA in
`maternal plasma, the copy number of fetal chromosomes
`of interest—eg, chromosomes 13 and 21—can be
`determined from maternal blood samples, and does not
`require isolation of fetal cells or free fetal DNA.
`Venipunctures are done routinely in clinical settings and
`present little risk to the mother and fetus.
`Many of the participants that we tested were from the
`patient population for which testing for aneuploidy is
`recommended; the mothers had a median age of
`34 years. The median gestational age of the fetus
`(17 weeks and 5 days) is within the time-frame when
`invasive diagnostic procedures are commonly done.
`The earliest gestational age analysed in this study was
`just over 8 weeks, and resulted
`in the correct
`identifi cation of a normal copy number of chromo-
`somes 13 and 21. Of the eight fi rst-trimester samples
`analysed, all were correctly identifi ed as being normal.
`Thus,
`the methods described herein serve
`the
`appropriate patient population.
`Amniocentesis or newborn reports from the clinical
`sites confi rmed
`that
`the copy number of
`fetal
`chromosomes 13 and 21 was determined correctly for
`58 out of the 60 samples analysed, including two of the
`three cases of trisomy 21. A probability calculation shows
`that the chances of identifying two or more of the three
`
`478
`
`www.thelancet.com Vol 369 February 10, 2007
`
`

`

`Case 1:20-cv-01644-RGA Document 1-15 Filed 12/03/20 Page 7 of 9 PageID #: 861
`
`Articles
`
`19
`
`20
`
`21
`
`22
`
`23
`
`24
`
`25
`
`26
`
`27
`
`28
`
`29
`
`30
`
`–0·0772
`
`21·7%
`
` 0·05†
`
`0·0121
`
`30·9%
`
`0·0291
`
`29·2%
`
`–0·0088
`
`29·9%
`
`0·0108
`
`26·1%
`
`0·0517
`
`51·5%
`
`–0·0256
`
`32·4%
`
`–0·0148
`
`23·8%
`
`0·0438
`
`32·6%
`
`–0·0616
`
`37·2%
`
`–0·0845
`
`26·6%
`
`0·0026
`
`31·0%
`
`0·0350
`
`32·5%
`
`0·78
`
`0·29
`
`0·81
`
`0·67
`
`0·27
`
`0·58
`
`0·70
`
`0·24
`
`0·25
`
`0·16
`
`0·94
`
`0·31
`
`0·0877
`
`42·6%
`
` 0·04†
`
`cases of trisomy 21 in the 60 observations, if the test had
`zero diagnostic power, is 0·005.
`At present, prenatal testing includes a combination of
`available diagnostic and screening tests. Screening tests
`include the fi rst-trimester screen (ultrasound-based
`nuchal translucency measurement with maternal serum
`analyte biochemistry), second-trimester maternal triple
`or quadruple serum analyte biochemistry, and second-
`trimester fetal ultrasound. Screening tests for Down’s
`syndrome are often reported to have a 5% false positive
`rate, with 64–96% of cases of trisomy 21 being
`identifi ed.1,24,25 The goal of any screening test is to increase
`the number of diagnoses made to a maximum while
`reducing the number of false positives to a minimum.
`Such a goal would decrease the number of follow-up
`invasive diagnostic tests—eg, amniocentesis or chorionic
`villus sampling.
`A practice bulletin from the American College of
`Obstetricians and Gynecologists, published in January,
`2007, recommends that all women should be off ered the
`option of genetic screening, irrespective of maternal
`age.25 Assuming that a larger number of women will be
`tested, and given the 5% false positive rate associated
`with available screening methods, this recommendation
`will probably increase the need for invasive diagnostic
`tests that are associated with a higher risk of miscarriage.
`In turn, it follows that there will be a greater demand for
`the development of non-invasive tests that yield diagnostic
`results.
`To achieve a false positive rate of 1%, the detection
`rate of available screening tests drops to a range of
`45–88%.1 On the basis of our results, the SNP ratio
`method described here achieves a false positive rate
`under 2% with a detection rate of 66%. Since we report
`only 60 observations, including three cases of trisomy 21,
`estimates of the sensitivity, specifi city, and positive and
`negative predictive values of the test are preliminary.
`Larger trials are needed to confi rm that test performance
`is reproducible and comparable to available tests.
`Refi nements to the method—eg, quantifying a greater
`number of SNPs and increasing the number of
`reference
`chromosomes—should
`enhance
`test
`performance.
`One of the samples identifi ed as a case of trisomy 21
`by amniocentesis was falsely identifi ed as a normal
`sample using our methods (sample 55, p=0·34).
`Additionally, one of the samples identifi ed by our
`methods as a trisomy 21 sample was a false positive
`(sample 18, p=0·05). Large-scale clinical trials are
`needed to more accurately establish the specifi city and
`sensitivity of the test, and to determine the optimum
`number of SNPs needed to identify chromosomal
`abnormalities at a
`range of
`free
`fetal DNA
`concentrations.
`There is generally less variation in mean fetal DNA
`ratios between chromosomes as more SNPs are
`quantifi ed. For instance, sample 4, which was identifi ed
`
`www.thelancet.com Vol 369 February 10, 2007
`
`(Continued from previous page)
`18
`13
`11
`21
`10
`13
`17
`21
`19
`13
`12
`21
`16
`13
`31
`21
`19
`13
`27
`21
`36
`13
`37
`21
`32
`13
`27
`21
`22
`13
`20
`21
`13
`13
`23
`21
`19
`13
`26
`21
`19
`13
`9
`21
`17
`13
`28
`21
`14
`13
`27
`21
`20
`13
`34
`21
`23
`13
`19
`21
`9
`13
`17
`21
`9
`13
`25
`21
`15
`13
`13
`21
`13
`13
`20
`21
`18
`13
`27
`21
`17
`13
`16
`21
`26
`13
`9
`21
`26
`13
`31
`21
`28
`13
`17
`21
`18
`13
`15
`21
`12
`13
`27
`21
`13
`
`31
`
`32
`
`33
`
`34
`
`35
`
`36
`
`37
`
`38
`
`39
`
`40
`
`41
`
`42
`
`43
`
`0·1218
`0·1990
`0·1828
`0·1707
`0·1712
`0·1421
`0·1759
`0·1847
`0·1504
`0·1396
`0·3465
`0·2948
`0·1932
`0·2188
`0·1351
`0·1499
`0·1947
`0·1509
`0·2286
`0·2902
`0·1532
`0·2377
`0·1835
`0·1809
`0·1938
`0·1588
`0·2704
`0·1827
`0·1674
`0·2125
`0·2050
`0·2223
`0·1840
`0·2070
`0·2043
`0·2221
`0·3329
`0·3546
`0·2980
`0·3389
`0·2765
`0·2561
`0·3320
`0·2380
`0·2755
`0·2929
`0·1601
`0·1295
`0·1219
`0·1275
`0·1378
`0·0954
`
`–0·0450
`
`28·7%
`
`–0·0172
`
`34·0%
`
`–0·0229
`
`31·1%
`
`–0·0179
`
`33·9%
`
`–0·0217
`
`49·9%
`
`–0·0409
`
`45·9%
`
`0·0204
`
`43·3%
`
`0·0940
`
`49·9%
`
`–0·0174
`
`43·2%
`
`0·0306
`
`27·6%
`
`–0·0056
`
`21·7%
`
`0·0424
`
`24·2%
`
`0·43
`
`0·78
`
`0·70
`
`0·73
`
`0·74
`
`0·56
`
`0·76
`
`0·13
`
`0·76
`
`0·35
`
`0·88
`
`0·16
`
`(Continues on next page)
`
`479
`
`

`

`Case 1:20-cv-01644-RGA Document 1-15 Filed 12/03/20 Page 8 of 9 PageID #: 862
`
`Articles
`
`Quantifi cation of fewer SNPs could be suffi cient at
`high percentages of free fetal DNA and more SNPs can
`be analysed when the percentage of free fetal DNA is low.
`To establish the optimum number of SNPs needed for
`diagnosis of trisomy 21 and other clinically relevant
`trisomies is important. Given that eight fi rst-trimester
`samples, including a sample drawn at about 8 weeks’
`gestation, were correctly identifi ed, and that phlebotomy
`is minimally invasive, there exists the possibility of
`analysing follow-up samples to identify a greater number
`of SNPs of the correct pattern, should a low number of
`SNPs of the correct pattern be identifi ed in the initial
`sample.
`In this study, paternal genomic DNA was genotyped to
`reduce the number of SNPs analysed in the plasma
`sample. Inclusion of paternal DNA was strictly for
`reference purposes and is not required for quantitative
`analysis of fetal DNA. Quantitative analysis is achieved
`through a comparison of maternal buff y coat DNA and
`maternal plasma DNA from a single sample of the
`mother’s blood.
`Importantly, the number of reference chromosomes
`that can be studied with this method is not limited. We
`compared chromosome 13 with chromosome 21, but
`any number of chromosomes can be compared,
`including chromosomes 18, X, and Y, together with
`reference chromosomes (eg, chromosomes 15 and 22).
`The scope and accuracy of this method would likely be
`increased by
`comparing a
`larger number of
`chromosomes and a larger number of SNPs on each
`chromosome. Furthermore, the human genome project
`has identifi ed over 3·7 million SNPs to date. Even on
`the smallest human chromosome—chromosome 21—
`about 54 000 genotyped SNPs are available for analysis.
`Clearly, there are additional SNPs that could be added
`to the current test.
`The approach described here, which uses standard
`molecular biology equipment, allows for precise
`analysis of genetic material. Free fetal DNA is directly
`quantifi ed from the heterogeneous mixture of maternal
`and fetal DNA in the maternal plasma. The maternal
`signal in the plasma DNA serves as an internal control,
`thus reducing variability. Rather than calculating
`absolute readings of the fetal-specifi c alleles, the ratio
`of unique fetal-specifi c allele signals to the combined
`maternal and fetal allele signal is determined. Since the
`diff erence between the mean ratios for chromosome 13
`and 21 is directly compared, the eff ect of systematic
`variations originating from experimental processing
`and variations in SNPs in the genome can be kept to a
`minimum.
`Our results show that SNPs can be used to distinguish
`fetal DNA from maternal DNA—and to determine the
`copy number of fetal chromosomes—in maternal blood
`samples. With further refi nement, a prenatal diagnostic
`test based on the methods described here could be a
`useful complement