DOI 10.1515/cclm-2012-0281      Clin Chem Lab Med 2013; 51(6): 1141–1154

Review

Yana N. Nepomnyashchaya , Artem V. Artemov , Sergey A. Roumiantsev ,

Alexander G. Roumyantsev and Alex Zhavoronkov *

Non-invasive prenatal diagnostics of aneuploidy using next-generation DNA sequencing technologies, and clinical considerations

Abstract

Rapidly developing next-generation sequencing (NGS)

technologies produce a large amount of data across the

whole human genome and allow a large number of DNA

samples to be analyzed simultaneously. Screening cell-free

fetal DNA (cffDNA) obtained from maternal blood using

NGS technologies has provided new opportunities for

non-invasive prenatal diagnosis (NIPD) of fetal aneuploi-

dies. One of the major challenges to the analysis of fetal

abnormalities is the development of accurate and reliable

algorithms capable of analyzing large numbers of short

sequence reads. Several such algorithms have recently

been developed. Here, we provide a review of recent NGS-

based NIPD methods as well as the available algorithms for

short-read sequence analysis. We furthermore introduce

the practical application of these algorithms for the detec-

tion of different types of fetal aneuploidies, and compare

the performance, cost and complexity of each approach

for clinical deployment. Our review identifies several main

technologies and trends in NGS-based NIPD. The main con-

siderations for clinical development for NIPD and screen-

ing tests using DNA sequencing are: accuracy, intellectual

property, cost and the ability to screen for a wide range of

chromosomal abnormalities and genetic defects. The cost

of the diagnostic test depends on the sequencing method,

diagnostic algorithm and volume of the tests. If the cost of

sequencing equipment and reagents remains at or around

current levels, targeted approaches for sequencing-based

aneuploidy testing and SNP-based methods are preferred.

Keywords: aneuploidy; cell-free fetal DNA; Down syn-

drome; fetal aneuploidies; next-generation sequencing;

non-invasive prenatal diagnosis; prenatal screening.

*Corresponding author: Alex Zhavoronkov, PhD, Bioinformatics and

Medical Information Technology Laboratory, Federal Clinical

Research Center for Pediatric Hematology, Oncology and

Immunology – Experimental and Molecular Medicine, Samory

Mashela 1, Moscow 117198, Russian Federation,

E-mail: alex@biogerontology.org

Yana N. Nepomnyashchaya , Artem V. Artemov , Sergey A. Roumiantsev and Alexander G. Roumyantsev: Bioinformatics and

Medical Information Technology Laboratory , Federal Clinical

Research Center of Pediatric Hematology, Oncology and

Immunology, Moscow , Russian Federation Alex Zhavoronkov: The Biogerontology Research Foundation ,

London , UK

Introduction To date, invasive procedures, such as amniocentesis

and chorionic villus sampling (CVS) have been used

successfully for the detection of fetal aneuploidy in

high-risk pregnancies. However, invasive methods are

associated with significant risks of the fetal loss – a

major adverse consequence of prenatal diagnosis in

obstetric practices [1] . In 1997, a research paper on the

discovery of circulating cffDNA from the Y chromosome

of male fetuses in the maternal blood stream during

pregnancy opened up new horizons for NIPD [2] . Since

then there have been numerous studies describing the

use of cffDNA for NIPD of chromosomal aneuploidy; in

particular, trisomies 21 (Down syndrome), 18 (Edward

syndrome) and 13 (Patau syndrome), all of which used

digital PCR analysis [3 – 5] .

The appearance of cffDNA in the maternal circula-

tion occurs when normal placental cell death causes

the chromosomes to break into short fragments, most of

which are under 300 bp in length [6 – 8] . The proportion

of cffDNA in maternal blood increases as pregnancy pro-

gresses. cffDNA comprises around 5 % – 10 % of the total

cell-free DNA during the first and second trimesters and

can be detected reliably as early as the seventh week

Unauthenticated | 95.165.92.74Download Date | 11/23/13 10:46 PM

1142      Nepomnyashchaya et al.: Non-invasive prenatal diagnostics of aneuploidy using NGS

of gestation [2, 9] . Circulating cffDNA is rapidly cleared

from maternal blood after delivery, except for cases

where small amounts remain, including cells from previ-

ous pregnancies [10] . It has recently been found that the

entire fetal genome, in the form of cffDNA, is present in

maternal blood [11] .

New NGS technologies permit the simultaneous,

or ‘ massively parallel ’ , sequencing of extremely large

quantities of DNA molecules. NGS produces billions

of short sequence reads per instrument run [12] . Mas-

sively parallel sequencing of fetal DNA from maternal

blood has enormous potential, not only for increasing

our understanding of the causes of prenatal genetic dis-

orders in the fetus but also for designing non-invasive

clinical diagnostic tests [11] . At the moment, non-inva-

sive methods do not detect other genetic abnormalities,

such as SNP and indel variations causing Mendelian dis-

orders, and invasive procedures are thus still required.

Non-invasive methods of fetal genotyping are currently

in development, and accurate screening of both domi-

nant and recessive single gene disorders may be possi-

ble in clinical practice in the near future [13] . The pos-

sibility of using massively parallel shotgun sequencing

to detect non-invasive fetal trisomies from maternal

blood by analyzing the relevant chromosomes in locus-

independent assays has been demonstrated [14, 15] , and

recent studies have confirmed this finding [16, 17] . An

alternative approach to sequencing whole genomes for

non-invasive detection of fetal abnormalities would be to

enrich for the region of interest using array capture prior

to sequencing [18 – 20] .

NGS technologies have already begun to show their

remarkable potential for detecting the most common

aneuploidies in live births, including Down, Edward,

and Patau syndromes. These research discoveries have

been translated into clinical tests, resulting in major

benefits for NIPD. In this review, we describe different

approaches to non-invasive detection of prenatal aneu-

ploidy which, for their clinical application, use the NGS

technologies of four innovative companies. The compa-

nies Sequenom, Inc (San Diego, CA, USA) and Verinata

Health, Inc (Redwood City, CA, USA) offer methods based

on collecting and analyzing information across the

entire genome, while the approach of Ariosa Diagnos-

tics, Inc (San Jose, CA, USA) and Natera, Inc (San Carlos,

CA, USA) is based on the selection of only the chromo-

somes of interest. Both approaches are capable of detect-

ing the most common fetal aneuploidies in the popula-

tion. To date of submission of this manuscript, three of

the aforementioned companies – Sequenom, Verinata

Health, and Ariosa Diagnostics – have published clinical

validation studies. These companies have furthermore

launched cffDNA-based non-invasive prenatal laboratory

developed tests (LDT) using next-generation sequencing

through the Clinical Laboratory Improvement Amend-

ments (CLIA) laboratories. Several clinical diagnostics

laboratories in other countries have also launched NIPD

screening tests using NGS, including the Beijing Geno-

mics Institute in China. Although every company has

sponsored clinical validation studies and published their

results in peer-reviewed, high-impact industry journals,

none of the tests are approved by the US Food and Drug

Administration (FDA). Although tests and clinical vali-

dation studies are still ongoing, it is expected that the

diagnostic accuracy, sensitivity, and specificity of these

techniques will be very high.

Limitations of current methods of prenatal diagnostics Many prenatal screening and diagnostic tests have been

developed and introduced into clinical practice. While

standard methods of prenatal care vary, combination

screening using biochemical markers in serum, includ-

ing pregnancy-associated plasma protein A (PAPP-A),

free β -subunit of human chorionic gonadotropin (free

β-hCG), α -fetoprotein (AFP) and unconjugated estriol

(uE3), in conjunction with a sonographic measurement

of nuchal translucency and detection of the presence

or absence of the nasal bone, are commonly used. The

detection rate of non-invasive combination screening

has improved greatly over the past decade to over 90 % in

the first trimester of pregnancy [21] . However, achieving

such a high detection rate requires exceptionally skilled

physicians and strict adherence to protocols. Invasive

diagnostic methods are still recommended for high-risk

pregnancies.

Currently, diagnostic testing requires that the fetal

cells that are to be tested be removed directly from the

uterus using either CVS between 11 and 14 weeks of ges-

tation, or amniocentesis between 15 and 20 weeks. While

using direct fetal material provides over 99 % accuracy,

both of these procedures are invasive and are associated

with an increased risk of transplacental hemorrhage

and spontaneous abortion [22] . The risk of miscarriage

following these procedures can be conservatively esti-

mated at around 1 % [23] , but this varies greatly depend-

ing on a number of human and environmental factors

including the country, the hospital or clinic, and the

physician.

Unauthenticated | 95.165.92.74Download Date | 11/23/13 10:46 PM

Nepomnyashchaya et al.: Non-invasive prenatal diagnostics of aneuploidy using NGS      1143

Whole-genome sequencing approaches for NIPD of fetal aneuploidy In cases where a woman is carrying a fetus with aneu-

ploidy – trisomy 21, e.g., the amount of copies of chromo-

some 21 is expected to be slightly higher in comparison

with other autosomes. Rapidly developing NGS tech-

nologies, which provide a vast amount of data across

the entire genome, appear to be suitable for counting

genome representation and determining the over-rep-

resented chromosomes of interest in the affected fetus.

Two scientific groups have independently shown that,

from cffDNA obtained from maternal blood, NGS can

clearly identify plasma samples from women carrying

aneuploid fetuses by comparing them with samples

taken from women with euploid fetuses [14, 15] . A short

region at one end of each DNA molecule of maternal

plasma was sequenced using synthesis technology on

an Illumina platform, and mapped against the reference

human genome to determine the chromosomal origin of

each sequence. The density of sequenced tags from the

MaternalDNA

FetalDNA

Trisomydetected

Threshold

chr 1

chr i

chr j

…

…

Counting of reads per each chromosome

Mapping of the reads

Figure 1   Schematic illustration of the procedural framework for using massively parallel genomic sequencing for the non-invasive prenatal

detection of fetal chromosomal aneuploidy.

(A) Fetal DNA (green fragments) circulates in maternal plasma as a minor population among a high background of maternal DNA (blue frag-

ments). A sample containing a representative profile of DNA molecules in maternal plasma is obtained. Short fragments of cell-free DNA

are then sequenced by some NGS technique. (B) The chromosomal origin of each sequence is identified by mapping the obtained reads to

the human reference genome. (C) The number of unique sequences mapped to each chromosome are counted and genome representation

of the chromosomes of interest is determined in subsequent analysis. Aneuploidy can be detected by various statistical techniques based

on the number of reads representing the chromosome of interest compared to the other chromosomes. Notice that blue and green circles

schematically representing the reads mapped to each chromosome, originating from maternal and fetal DNA fractions, respectively, are

indistinguishable and only a difference in the total amount of reads can be detected (as shown for ‘ chr j ’ in the example).

chromosome of interest from an aneuploid fetus was

compared with cases of trisomy and euploid pregnancies

(Figure 1 ).

Fan et al. [14] used from 1.3 to 3.2 mL of plasma taken

from pregnant women at risk for fetal aneuploidy. Their

study examined a total of 18 normal and aneuploid preg-

nancies, which included nine cases of trisomy 21, two of

trisomy 18 and one of trisomy 13. Amniocentesis or CVS

was conducted to analyze and confirm the fetal karyotype.

Cell-free plasma DNA was sequenced on the 1G Genome

Analyzer platform (Solexa/Illumina, Inc, San Diego, CA,

USA). Each of the 25 bp reads were mapped against the

non-repeat-masked reference human genome build 36

(hg18) using ELAND from the Solexa data analysis pipe-

line. Sequencing resulted in approximately five million

unique sequence reads per patient sample with, at most,

one mismatch against the human genome. Each set of

patient sequence reads covered approximately 4 % of the

entire human genome. The average sequence read density

for 50-kb windows across the chromosome of interest

was normalized using the median value obtained from

all autosomes in the euploid cases. A 99 % confidence

interval in the distribution sequence read density of the

Unauthenticated | 95.165.92.74Download Date | 11/23/13 10:46 PM

1144      Nepomnyashchaya et al.: Non-invasive prenatal diagnostics of aneuploidy using NGS

chromosome of interest for disomy pregnancies was cal-

culated. The values for normalized sequence tag density

of trisomy 21, 18 and 13 cases were reported to be outside

the upper boundary of the confidence interval, and those

for all disomy 21, 18 and 13 cases were below the boundary,

indicating 100 % sensitivity with a false-positive rate 0 % .

The coverage of chromosome 21 for trisomy 21 pregnan-

cies was on average ∼ 11 % higher than that of the disomy

21 cases. The method described by Fan et al. reveals that

the minimum fraction of fetal DNA that could be detected

by NGS is approximately 2 % .

The distribution of sequence reads across each chro-

mosome for all samples was not uniform. In addition,

the GC content of the sequenced reads of all the samples

was ∼ 10 % higher than the value of the sequenced human

genome [24] . The mean read density of chromosomes,

which was reported to correlate with the GC content of

the chromosomes, was observed to have a positive bias

for chromosomes with high GC content. Interestingly –

and differing from this report – Chiu et al. [25] observed a

negative correlation between chromosome representation

and GC content in their analysis employing massively par-

allel sequencing of fetal DNA from maternal plasma using

ligation technology on the SOLiD ™ 3 platform (Applied

Biosystems/Life Technologies, Carlsbad, CA, USA). In both

sets of data, a strong GC bias and the non-uniform repre-

sentation of read density between chromosomes are likely

to be explained by analytical rather than biological factors

stemming from the sequencing process, which would be

eliminated by the use of appropriate algorithms [26 – 28] .

Importantly, blood samples were collected from 15 to

30 min after amniocentesis or CVS, and the average gesta-

tional period of the aneuploidy pregnancies (20.6 weeks)

was longer than that of the euploid group (13.8 weeks).

Since the proportion of fetal DNA in maternal blood was

reported to increase with the length of gestation age [29,

30] , the above-mentioned factors should be taken into

consideration in future analyses. Nonetheless, this pio-

neering work has shown that NGS could be used as a

non-invasive prenatal diagnostic tool for the quantitative

measurement of a number of chromosomes.

Chiu et al. analyzed cell-free DNA from 5 to 10 mL

of plasma taken from 14 trisomy 21 and 14 euploid preg-

nancies in women at risk for fetal aneuploidy [15] . In the

majority of cases, maternal blood samples were collected

before invasive obstetrics procedures had been carried

out, with comparable median lengths of gestation for both

the euploid and the trisomy case groups. One end of the

clonally expanded copies of each plasma DNA fragment

was sequenced (36 bp reads) and processed by stand-

ard post-sequencing bioinformatic alignment analysis

with the Illumina Genome Analyzer, which uses ELAND

software. The obtained mean of approximately 2 million

unique reads per sample resulted in no mismatches with

the repeat-masked reference human genome (NCBI Build

36, version 48). The number of reads originating from chro-

mosomes of interest was normalized by the total number

of reads generated by the sequencing run. Z-scores repre-

senting the number of standard deviations from the mean

proportion of chromosomes-of-interest reads in a refer-

ence set of euploid cases were calculated for each case. A

statistically significant difference between the parameter

estimated in the test case and that in the reference group

was 99 % under z-score > 3. Z-scores for chromosome 21

were above + 5 (range 5.03 – 25.11) for all 14 trisomy 21 cases,

i.e., at three standard deviations above the reference

established from the male euploid fetuses. In the study,

the method accurately detected all trisomy 21 cases and

produced no false-positive results.

One caveat of using this approach is that the distribu-

tion of reads on each chromosome can vary from sequenc-

ing run to sequencing run; hence, these intra- and

inter-run sequencing variations can increase the overall

variation in the aneuploidy detection metric.

Both studies demonstrated that massively parallel

sequencing can both identify and count DNA fragments

in maternal plasma to detect small quantitative altera-

tions in the genomic distribution of plasma DNA. Methods

that do not require the differentiation between fetal and

maternal DNA are currently being developed, and can be

applied to arbitrarily small concentrations of fetal DNA in

maternal plasma. Since fetal DNA is natively fragmented,

no further fragmentation step for the library preparation

is needed, substantially simplifying subsequent analysis.

Non-invasive NGS detection of the variation in the

number of copied fetal chromosomes in maternal plasma

is already in use on pregnant women in proof-of-concept

studies, and has been brought into clinical application.

A number of subsequent large-scale clinical studies were

carried out by academic research institutions and biotech-

nology companies to validate various approaches to NIPD

using massively parallel genomic sequencing for use in

clinical practice [16, 17, 31 – 33] . Table 1 presents a summary

of these studies for detection of Down syndrome.

Targeted approaches to sequencing-based NIPD of fetal aneuploidy The two methods described above rely on the massively

parallel sequencing of all the DNA in a maternal plasma

Unauthenticated | 95.165.92.74Download Date | 11/23/13 10:46 PM

Nepomnyashchaya et al.: Non-invasive prenatal diagnostics of aneuploidy using NGS      1145

sample without targeting specific chromosomes. However,

several targeted approaches were recently developed

based on the a priori selection of DNA regions for analy-

sis. Compared to sequencing and counting all reads from

all chromosomes, limiting the number of DNA regions to

quantify greatly reduces the effort required to assess the

dosage of a chromosome. Moreover, careful selection of

the regions to quantify can potentially reduce the con-

founding variation in the number of reads per locus by

taking into account only the loci with similar properties,

e.g., GC content or the number of repeats of a particular

sequence in the genome [19, 20] .

Different strategies can be used to select and enrich

for the genomic regions of interest. The enrichment step

should meet two requirements: the method should mini-

mize introductory bias (i.e., the output quantity of a frag-

ment should depend only on its input quantity and not on

other factors, such as the features of its sequence) to allow

further quantitative analysis, and the method should be

capable of dealing with small amounts of sampled DNA

[34] . Using the SureSelect System (Agilent Technologies,

Santa Clara, CA, USA), Liao et al. successfully performed

a 213-fold enrichment of the selected loci of cfDNA from

the blood samples of pregnant women [35] as an example

of an in-solution, separation-based enrichment technique

[34] . The enrichment did not introduce any bias in the

ratio of maternal and fetal DNA.

Sparks et al. [19] describe a method for detecting

chromosome aneuploidy using NGS-sequencing of DNA

enriched by a so-called Digital ANalysis of Selected

Regions amplification-based enrichment assay. The

assay comprises three oligos per analyzed locus. Liga-

tion oligos, “ Left ” , “ Middle ” and “ Right ” can comple-

mentarily hybridize to the respective DNA locus, thus

forming a contiguous chain with two nicks, which

can subsequently be ligated. Besides the fragment,

complementary to the matrix, the “ Left ” and “ Right ”

oligos contain constitutive fragments on their ends,

which are then used for PCR amplification with univer-

sal primers. The authors obtained and sequenced DNA

product. Of the 298 samples analyzed (including 39

trisomy 21 and 7 trisomy 18 samples), all the aneuploid

samples were correctly distinguished from the controls,

the authors concluded to 100 % sensitivity and speci-

ficity. The level of sequencing, covering only 420,000

reads per sample, was nevertheless sufficient to detect

trisomy 21 and trisomy 18 reliably (z statistics exceeded

3.6 in all samples). This level corresponds to < 5 % of

the level required by non-targeted approaches, and

enables multiplexing [the study claims that 96 samples

were processed simultaneously in one HiSeq2000

Full

sam

ple,

n a

Tris

omy

21

sam

ple,

nFa

lse-

posi

tive

rate

(95 %

CI),

%

Sens

itivi

ty (9

5 %

CI),

%

Spec

ifici

ty (9

5 %

CI),

%

Gest

atio

nal a

ge,

rang

e (m

ean)

wee

ks b

Met

hod

Refe

renc

es

18

90

(0

– 3

7)

10

0 (

63

– 1

00

)1

00

(6

3 –

10

0)

10

– 3

5 (

18

.3)

Ma

ss

ive

ly p

ara

lle

l s

eq

ue

nci

ng

Fan

et

al.

, 2

00

8 c [

14

]

28

14

0 (

0 –

27

)1

00

(7

3 –

10

0)

10

0 (

73

– 1

00

)1

1.2

– 2

0.3

(1

4.9

)M

as

siv

ely

pa

rall

el

se

qu

en

cin

gC

hiu

et

al.

, 2

00

8 d [

15

]

47

13

0 (

0 –

13

)1

00

(7

2 –

10

0)

10

0 (

87

– 1

00

)1

0.4

– 2

8.3

(1

5.3

)M

as

siv

ely

pa

rall

el

se

qu

en

cin

gS

eh

ne

rt e

t a

l.,

20

11

c [2

8]

44

93

90

.3 (

0.1

– 1

.5)

10

0 (

89

– 1

00

)9

9.7

(9

8.5

– 9

9.9

)8

.0 –

36

.0 (

16

.0)

4-P

lex

ed

ma

ss

ive

ly p

ara

lle

l s

eq

ue

nci

ng

Eh

rich

et

al.

, 2

01

1 d [

31

]

14

68

62

.1 (

0.6

– 6

.4)

10

0 (

95

– 1

)9

7.9

(9

3.6

– 9

9.4

)1

2.3

– 1

3.5

(1

3.0

)2

-Ple

xe

d m

as

siv

ely

pa

rall

el

se

qu

en

cin

gC

hiu

et

al.

, 2

01

1 d [

32

]

57

18

61

.1 (

0.5

– 2

.4)

79

.1 (

69

– 8

7)

98

.9 (

97

.6 –

99

.5)

8-P

lex

ed

ma

ss

ive

ly p

ara

lle

l s

eq

ue

nci

ng

16

96

21

20

.2 (

0.1

– 0

.7)

98

.6 (

95

.9 –

99

.7)

99

.8 (

99

.3 –

99

.9)

9.2

– 2

1.3

(1

5.3

)4

-Ple

xe

d m

as

siv

ely

pa

rall

el

se

qu

en

cin

gP

alo

ma

ki

et

al.

, 2

01

1 d [

33

]

29

83

90

(0

– 1

.8)

10

0 (

88

.8 –

1)

10

0 (

98

.2 –

10

0)

13

.4 –

35

.4 (

20

.5)

Targ

ete

d m

as

siv

ely

pa

rall

el

se

qu

en

cin

g e

Sp

ark

s e

t a

l.,

20

12

a f [

19

]

16

73

60

(0

– 3

.6)

10

0 (

88

.0 –

10

0)

10

0 (

96

.4 –

10

0)

11

.0 –

36

.1 (

18

.6)

Targ

ete

d m

as

siv

ely

pa

rall

el

se

qu

en

cin

gS

pa

rks

et

al.

, 2

01

2b

f [2

0]

53

28

90

(0

– 1

.1)

10

0 (

94

.8 –

10

0)

10

0 (

98

.9 –

10

0)

10

.0 –

23

.0 (

15

.1)

Ma

ss

ive

ly p

ara

lle

l s

eq

ue

nci

ng

Bia

nch

i e

t a

l.,

20

12

c [1

6]

Tabl

e 1  

Dia

gn

os

tic

pe

rfo

rma

nce

of

ma

tern

al

pla

sm

a D

NA

se

qu

en

cin

g f

or

de

tect

ing

fe

tal

tris

om

y 2

1.

a O

nly

th

e v

ali

da

tio

n s

et

is c

on

sid

ere

d.

b G

es

tati

on

al

ag

e o

f th

e t

ris

om

y 2

1 c

as

es

. F

rom

c Ve

rin

ata

He

alt

h,

Inc

(Re

dw

oo

d C

ity,

CA

, U

SA

). d

Se

qu

en

om

In

c (S

an

Die

go

, C

A,

US

A).

e Ta

rge

t e

nri

chm

en

t o

f

38

4 l

oci

pe

r ch

rom

os

om

e 2

1.

f Ari

os

a D

iag

no

sti

cs,

Inc

(Sa

n J

os

e,

CA

, U

SA

).

Unauthenticated | 95.165.92.74Download Date | 11/23/13 10:46 PM

1146      Nepomnyashchaya et al.: Non-invasive prenatal diagnostics of aneuploidy using NGS

run (Illumina, Inc)]; thus greatly reducing the cost of

analysis.

A method for detecting aneuploidy based on the

assessment of heterozygosity for various polymorphisms

is described by Natera in the US patent application

US20110288780A1 [18] . The key feature of this method is

that it takes the mixture of maternal and fetal DNA obtained

from blood plasma into account separately from the DNA

from one or both parents. As fetal DNA is represented

exclusively in the cell-free fraction, purely maternal DNA

can be extracted from the blood cells – white blood cells in

particular. Paternal DNA can be additionally included in

the analysis to improve the accuracy of the test, although

the test can still be performed based solely on the samples

taken from the pregnant mother. This method implies

that the sequencing of DNA regions is targeted. One of the

patent ’ s claims suggests the enrichment of the explored

polymorphic loci with an assay of pre-circularized probes,

which have been previously been described [36, 37] . Col-

lected sequencing data contains information on the counts

of sequences having each allele at each of the selected

SNP loci. A statistical model estimates a likelihood ratio

for the total number of reads containing one or another

allele for every possible combination of parental geno-

types and for each number of maternal n m and paternal n f

chromosomes inherited by the fetus. This model takes the

fraction of fetal DNA σ as a parameter. First, a maximum

likelihood estimation is performed for the fetal DNA frac-

tion σ , then n m and n f are determined for the chromosomes

of interest using the value of σ estimated in the previous

step. As the approach described builds statistical models

of the euploid state as well as all the other abnormal ploidy

states, it not only finds the most likely ploidy state of the

fetus for each chromosome, but also permits an estima-

tion of the significance of this decision. This allows a sta-

tistically grounded identification of the samples where no

reliable decision can be made without choosing arbitrary

thresholds or estimating them from a training set. Unfor-

tunately, the patent does not describe the results of the

method ’ s performance tests, thus it is difficult to estimate

its clinical efficiency. The sensitivity and specificity of the

method depend on the fetal DNA fraction and the avail-

ability of the paternal genotype. The availability of pater-

nal DNA allows for samples of the maternal peripheral

blood with lower cffDNA content to be accepted for diag-

nosis. Methods using paternal content are likely to enable

NIPD at earlier stages of the pregnancy when the concen-

tration of fetal material is below the minimal threshold

for methods utilizing only maternal samples. Separate

sequencing of purely maternal cellular DNA may further

improve performance. Along with trisomy 21, trisomy 18

and trisomy 13, sex chromosome aneuploidies (e.g., X0,

XXY, XXX, XYY) can also be detected (Natera, personal

communication), which is an important advantage of this

method in light of the high occurrence of these abnormali-

ties. However, the detection performance of the method is

still to be published. The clinical trial of the Prenatal Non-

invasive Aneuploidy Testing Using SNPs supported by the

National Institutes of Health is underway [38] . As it is SNP-

based, the method may need to be tested on patients from

different populations.

The recent SNP-based targeted NIPD methods of

Sparks et al. [20] and Rabinowitz et al. [18] seem to be

highly efficient. As they can be performed on a sequenc-

ing machine with a lower price per run and lower

throughput [e.g., PGM (Ion Torrent/Life Technologies,

Carlsbad, USA) or MiSeq (Illumina, Inc)], these methods

are preferred, especially for average-sized clinics. In

addition to modeling disomy and trisomy, both methods

model other ploidy states, thus enhancing statistical per-

formance. The Natera method can potentially perform

better because it extracts both cell-free DNA and cellu-

lar DNA from the same blood sample of purely maternal

origin. It is also claimed that the Natera method can detect

aneuploidy in sex chromosomes. Nevertheless, perfor-

mance tests for the Natera method have not yet been

published.

Counting statistics for improvements in the sensitivity of the NIPD of fetal aneuploidy Early reports suggest that inaccuracy in measuring

genomic representation is variable [14, 15] . Although the

algorithms used in recently published studies success-

fully classify fetal trisomy 21, they appear to be unable to

effectively detect other aneuploidies, such as trisomy 18

and 13, which would inevitably occur in the population.

However, it has been reported that high throughput mas-

sively parallel sequencing assays using specific bioinfor-

matic algorithms, may enable the non-invasive detection

of any type of fetal aneuploidy [17, 26 – 28] .

The existence of a substantial GC bias in Illumina/

Solexa and ABI/SOLiD sequencing has recently been

shown. This issue limits the sensitivity of measuring

genomic representation in chromosomes [14, 39 – 41] .

Fan and Quake [26] have analyzed sequencing data col-

lected in a previous study [14] , as described earlier. In this

study, a bioinformatic algorithm was developed to remove

Unauthenticated | 95.165.92.74Download Date | 11/23/13 10:46 PM

Nepomnyashchaya et al.: Non-invasive prenatal diagnostics of aneuploidy using NGS      1147

GC-content-dependent artifacts in shotgun sequencing

data by applying weight to each sequenced read based

on the local genomic GC content of large regions of the

human genome with relatively homogeneous GC content

[42] . After calculating the standard z-statistic, z-scores for

chromosomes 18 and 13 were increased in two of the two

trisomy 18 and one of the one trisomy 13 cases, respec-

tively. Consequently, all trisomy 18 and 13 cases were cor-

rectly classified. The algorithms for removal of the effect

of GC bias used in these studies appear to be able to effec-

tively detect cases of trisomy 18 and 13 as well as trisomy

21. Although the classification accuracy for trisomy 18 and

13 improved, there were not enough positive samples to

measure a representative distribution.

Chen et al. [27] have demonstrated the successful use

of two-plex massively parallel plasma DNA sequencing

for NIPD of trisomy 18 and 13 on the Genome Analyzer IIx

platform (Illumina) for a large sample set. Cell-free DNA

from 5 to 10 mL of plasma from 25 trisomy 13, 37 trisomy

18, 86 trisomy 21, one sex chromosome mosaic case and

140 euploid pregnancies was used in the analysis. A

total of 392 cases were analyzed, including 103 cases of

women pregnant with euploid fetuses, which were used

as normal controls for z-score calculation. As previously

described, standard z-scores representing the number of

standard deviations away from the mean proportion of

chromosome 18 and 13 reads in a reference set of euploid

cases were determined for each case [15, 32] . Based on the

previous findings that the statistical power of the mole-

cular counting approach increases with the number of

molecules counted, a mean of approximately 4.6 million

unique reads per sample, without mismatches to the non-

repeat-masked reference human genome, was obtained.

As a result, the classification accuracies for trisomy 18

and 13 were improved. For trisomy 18 detection, 31 of 37

trisomy 18 cases and 247 of 252 non-trisomy 18 cases were

identified correctly, corresponding to sensitivity and spec-

ificity of 83.8 % (95 % CI 67.3 % – 93.2 % ) and 98.0 % (95 % CI

95.2 % – 99.3 % ), respectively. For trisomy 13 detection, 11 of

25 trisomy 13 cases and 247 of 264 non-trisomy 13 cases

were identified correctly, corresponding to improved sen-

sitivity and specificity of 44.0 % (95 % CI 25.0 % – 64.7 % ) and

93.6 % (95 % CI 89.7 % – 96.1 % ), respectively. As the impreci-

sion of measuring the genomic representation of chromo-

somes was shown to be variable and dependent on the GC

content of each chromosome, a statistical approach based

on z-score calculation, but with an additional GC correc-

tion algorithm, has been developed in order to improve

diagnosis of trisomy 18 and 13 [14, 15] . Specifically, all

chromosomes from each sample were first divided into 50

kb bins using bioinformatics. Chen et al. determined the

number of sequenced reads and GC content in each bin

and applied the locally weighted scatter plot smoothing

(LOESS) regression to calculate the predicted (P) value

for each bin. Using the raw read counts (RC raw ) the GC-

corrected read counts (RC GC ) of each bin were calculated

with the correlation factor (F), which was derived from the

median counts of all the bins (M) and the LOESS fit pre-

dicted value by the following equations:

F = M/P (1)

and

RC GC = RC raw × F (2)

The standard z-score was calculated using the

genomic representations of chromosomes 18 and 13, and

all of the trisomy 13 cases (25 out of 25) were clearly identi-

fied. Two hundred and sixty-one out of 264 non-trisomy

13 cases were correctly determined under z-score > 3, indi-

cating 100 % (95 % CI 83.4 % – 100 % ) sensitivity and 98.9 %

(95 % CI 96.4 % – 99.7 % ) specificity of the GC correction

approach. Thirty-four out of 37 trisomy 18 samples and 247

out of 252 non-trisomy 18 cases were classified providing

91.9 % (95 % CI 77.0 % – 97.9 % ) sensitivity and 98.0 % (95 %

CI 95.2 % – 99.3 % ) specificity. This study has shown that

the use of both the mapping of sequence reads against

the non-repeat-masked genome and the GC correction

approaches have improved the accuracy of trisomy 18 and

13 diagnosis. Essentially, correct aneuploidy detection

was achieved by increasing the number of aligned reads

in general and hence performing deeper sequencing.

This bioinformatic method will be able to lower the cost

of NGS in the future. Current algorithms for removing GC

bias improve the precision of measuring the genomic rep-

resentation of chromosomes as well as allow the effective

classification of aneuploidies, most notably trisomy 13.

Larger sample collections are required to further examine

the algorithm for trisomy 18 detection.

An alternative for improving the accuracy of the

detection of chromosomal abnormalities is to develop

an optimized algorithm using a normalized chromo-

some value (NCV) from the sequencing data of the refer-

ence group (training set) of 71 samples with 26 abnormal

karyotypes. This would minimize the intra- and inter-run

sequencing variation as previously described [28] . Short,

single-end reads of each plasma sample were sequenced

on the Genome Analyzer IIx platform and unambigu-

ously mapped to the non-repeat-masked reference

human genome, allowing for up to two base mismatches

during alignment. In the test set, the number of unique

sequence tags varied from approximately 13 × 10 6 – 26 × 10 6 .

To determine cases of fetal aneuploidy, the NCVs for the

Unauthenticated | 95.165.92.74Download Date | 11/23/13 10:46 PM

1148      Nepomnyashchaya et al.: Non-invasive prenatal diagnostics of aneuploidy using NGS

chromosome of interest from the test set were compared

with the respective NCVs for those from the training set.

Initially, in the sequencing data from the training set,

the sequence read density (number of mapped sites) for

the chromosome of interest was normalized to counts

observed on another predetermined chromosome (or set

of chromosomes). Each autosome was considered as a

potential denominator in a ratio of counts with our chro-

mosomes of interest from an unaffected subset of the

training data in order to determine the optimal chromo-

some ratio for each chromosome of interest. Denominator

chromosomes that minimized the variation of the chro-

mosome ratios within and between sequencing runs were

selected. An NCV representing the number of standard

deviations away from the mean chromosome ratios for the

unaffected samples in the training set was determined for

each sample and chromosome of interest, and was calcu-

lated with the following equation:

NCV ij = x ij − μ j / σ j (3)

where μ j and σ

j are the estimated training set mean and

SD, respectively, for the j-th chromosome ratio and x ij is the

observed j-th chromosome ratio for sample i. NCVs > 4.0

classify a chromosome as affected, whereas NCVs < 2.5

specify a chromosome as unaffected, which indicates a

99 % chance of a statistically significant difference in the

assessed parameter for the test set compared with the ref-

erence training set. All eight samples with clinical karyo-

types indicating fetal trisomy 18 were correctly identified,

with NCVs between 8.5 and 22, indicating 100 % (95 % CI

59.7 % – 100 % ) sensitivity. The single trisomy 13 individual

with an NCV of approximately three was classified as a ‘ no

call ’ . Hence, the current algorithm demonstrated a 100 %

accurate classification of samples with cases of trisomy 18.

However, larger sample collections are required in order

to test the algorithm further for the detection of trisomy

13. There were no discernible differences in results with

respect to ethnicity. It was also shown that this approach

is particularly informative in the case of twins. Canick

et al. analyzed 27 samples of multiple gestations, collected

during a large-scale study, and used the algorithm correct-

ing for the GC shift [17, 33, 43] . All seven trisomy 21 cases

and one trisomy 13 case were accurately detected (95 % CI

59 % – 100 % ) with no false-positives. Two triplet pregnan-

cies were analyzed and correctly confirmed as euploid as

well. The study confirmed that the algorithm may be used

to correctly detect trisomies in cases involving multiple,

simultaneous gestations.

To return to the recent study by Sparks et al. [19] , in

the current report the authors introduced Fetal-fraction

Optimized Risk of Trisomy Evaluation – an improved

statistical algorithm for trisomy detection [20] . It esti-

mates the risk of aneuploidy by computing an odds ratio

that compares the probability of observing the outcome

according to a model representing a disomic fetal chro-

mosome and a model representing a trisomic fetal

chromosome. Modeling the case of trisomy represents

a major improvement in the study. Both disomic and

trisomic models are normal distributions of the number

of reads from the chromosome of interest scaled by the

number of reads from a different chromosome. The

mean value for the disomy model was taken as a mean

proportion over the reference disomy samples (pos-

sibly with bootstrapping of samples). For the trisomy

model, adjustment of the mean proportion was based

on the estimated fetal DNA fraction. In order to assess

the fetal DNA fraction for a sample, a set polymorphic

loci was quantified together with the constitutive loci

in the assay. A maximum likelihood estimation of the

fetal DNA fraction was performed for every sample,

based on loci where fetal and maternal genotypes differ.

The standard deviation for both the proportion of reads

mapped on the chromosome of interest and of the frac-

tion of fetal DNA was estimated by bootstrapping the

reference samples and taking into account polymorphic

loci. As a result, of the total of 192 polymorphic and 576

non-polymorphic loci quantified in the samples from

the training set, the polymorphic regions and 384 non-

polymorphic loci showing the highest residual difference

between normal and trisomic samples were selected.

Thirty-six samples of trisomy 21 and 8 trisomy 18 from

167 samples in the test set were identified correctly,

showing 100 % for trisomy 21 (95 % CI 88.0 % – 100 % ) and

for trisomy 18 (95 % CI 59.8 % – 100 % ) detection rate with

a false-positive rate of 0 % for trisomy 21 (95 % CI 0 % –

3.6 % ) and for trisomy 18 (95 % CI 0 % – 3.0 % ). The method

is promising in terms of the small amount of sequenced

reads required and the potential for screening for sub-

chromosomal abnormalities. Nevertheless, a study on

a larger cohort is required. Moreover, as the approach

is SNP-based and the frequency of SNP genotypes can

vary in different populations, further studies should be

conducted to verify whether extending the cohort would

require the number of screened polymorphic loci to be

extended.

Clinical considerations To date three diagnostic companies – Sequenom, Veri-

nata Health, and Ariosa Diagnostics – have published

Unauthenticated | 95.165.92.74Download Date | 11/23/13 10:46 PM

Nepomnyashchaya et al.: Non-invasive prenatal diagnostics of aneuploidy using NGS      1149

the results of sponsored clinical studies validating their

methods and started offering NGS-based NIPD services

commercially as LDTs [44 – 46] .

Sequenom was the first to launch the NIPD screening

test in November 2011. Presently, this non-invasive, LDT

detects the increased number of reads from chromosomes

21, 18 and 13 resulting from whole-genome sequencing. In

particular, the reports by Palomaki et al. [17, 33] provide

a large scale international investigation into the determi-

nation of fetal trisomy 21, 18 and 13 involving around two

thousand validation samples. The sensitivity for detect-

ing cases of trisomy 21 was 98.6 % (95 % CI 95.9 % – 99.7 % )

with a false-positive rate 0.2 % (95 % CI < 0.1 – 0.6). Failure

to obtain results occurred in 0.8 % of cases. Trisomy 18 and

13 detection rates were 100 % (95 % CI 93.9 % – 100 % ) and

91.7 % (95 % CI 61 % – 99 % ), respectively with false-positive

rates of 0.3 % (95 % CI 0.1 % – 0.7 % ) for chromosome 18 and

0.9 % (95 % CI 0.5 % – 1.5 % ) for chromosome 13. The average

gestation period at the time the maternal blood was

sampled was 15 weeks and 3 days. Although these studies

are large scale, 63.6 % samples were excluded from analy-

sis due to poor sample quality, volume or long processing

time. The developed test has a high accuracy for determi-

nation of the most common trisomies in the population

and can be offered in combination with other non-inva-

sive methods for diagnosis aneuploidy as a pre-invasive

procedure for high-risk pregnancies. However, since only

four to eight samples can be analyzed in one sequencing

run, and the processing and analysis of whole-genome

sequencing is required, the test cost is high and thus is a

limitation factor for widespread use of this test for diagno-

sis in a clinical setting.

Verinata Health launched the verifi ™ prenatal test

for diagnosis of trisomy 21, 18 and 13 as early as 10 weeks

gestational age, based on a clinical study which was con-

ducted by Bianchi et al. [16] . Bianchi et al. carried out a

study examining 532 maternal blood samples where every

sample was analyzed for six independent categories in

order to define test performance and determine the pres-

ence of trisomy 21, 18, 13 aneuploidy male, female or

monosomy X. It is the first published study which is able

to detect sex chromosome aneuploidy, including mono-

somy X. The sensitivity for detection of trisomy 21, 18 and

13 was 100 % (95 % CI 95.9 – 100), 97.2 % (95 % CI 85.5 – 99.9)

and 78.6 % (95 % CI 49.2 – 99.9), respectively, with a speci-

ficity of 100 % (95 % CI more than 98.5 – 100) for all trisomy

cases. Also, all chromosomes of the human genome were

analyzed using this approach. Current test determines of

the presence of trisomy 21 with a 100 % detection rate and

may be utilized for trisomy 21 diagnosis along with other

NIPD methods. Bianchi et al. [16] additionally provide the

first report in which the non-invasive diagnosis of trisomy

13 occurs during the first trimester of gestation. However,

because the detection rate of trisomy 13 is low, this test

cannot be used as the sole screening method for Patau

syndrome, and negative results should be confirmed with

further testing.

Ariosa Diagnostics launched the Harmony Prenatal

Test for NIPD of trisomy 21, 18 and 13 using direct sequenc-

ing of selective chromosome regions of interest, which

was developed by Sparks et al. [20] . From a blinded vali-

dation set of 167 individuals, all trisomy 21 and 18 cases

were correctly determined with 100 % sensitivity (95 % CI

88.0 % – 100 % for chromosome 21 and 95 % CI 59.8 % – 100 %

for chromosome 18) and 100 % specificity (95 % CI 96.4 % –

100 % for chromosome 21 and 95 % CI 97.0 % – 100 % for

chromosome 18). Data for the study of trisomy 13 is not

shown. If one relies exclusively on published data, this

test appears to be the most effective in the detection of

fetal aneuploidy. However, only a small sample of posi-

tive trisomy cases were examined in comparison with the

general number of samples examined in the study. One

relevant advantage of this approach is the possibility of

analyzing a huge amount of samples in one sequencing

run, which thus reduces the cost of the test.

CffDNA-based NIPD of aneuploidy using NGS tech-

nologies is the first NIPD method that both appears in

clinical practice and detects trisomy with high accuracy.

Through the continuous development and improve-

ment of algorithms for data sequencing and analysis, it

has been possible to raise the accuracy, sensitivity and

specificity of aneuploidy detection using this approach to

100 % . Today, successfully used non-invasive integrated

screening during the first and second trimesters for fetal

aneuploidy is both safe for the fetus and accurate for diag-

nosis, which is reflected in a high sensitivity of between

72 % and 95 % (5 % false-positive rate) reported during

the last 10 years. However, in some cases, the patient

does not pass all screening tests, necessitating succes-

sive tests, which require a lot of time [47 – 49] . The inde-

pendent conduction of serum-based integrated screening

during the first and second trimesters and the interpre-

tation of results lead to unnecessary invasive procedures

for normal pregnancies in 11 % – 17 % of cases because the

current false-positive rate is high. Although the combina-

tion of ultrasonographic detection, serum markers and

maternal age at first trimester detect trisomy 21 at a rate of

90 % , the false-positive rate is high – approximately 20 %

[50, 51] . One relevant advantage of NGS-based tests for the

diagnosis of aneuploidy is the ability to obtain informa-

tion about the ploidy of the fetus in early pregnancy. The

combination of data obtained from NGS-based testing

Unauthenticated | 95.165.92.74Download Date | 11/23/13 10:46 PM

1150      Nepomnyashchaya et al.: Non-invasive prenatal diagnostics of aneuploidy using NGS

with ultrasonographic detection, serum markers and

maternal age at first trimester may increase the detec-

tion of aneuploidy with a low false-positive rate. After

such screening, women at high risk of fetal aneuploidy

can choose CVS during the first trimester or amniocen-

tesis during the second trimester. This approach would

decrease the likelihood that a woman with a normal fetus

would undergo an invasive procedure.

Since the current NIPD tests are designed to identify

a limited number of aneuploidy, only invasive procedures

are defining multiple fetal chromosome abnormalities

with high accuracy. Although the methods NGS for ana-

lyzing cffDNA have huge diagnostic potential for detecting

all possible aneuploidies in a clinical practice setting, the

screening for other multifactorial birth defects remains a

big challenge. For instance, neural tube defects are one

of the most common defects in the general population,

with both genetic and environmental factors contributing

to their development. Neural tube defects are successfully

detected by measuring the level of an AFP in the amniotic

fluid between 13 and 22 weeks’ gestation [52, 53] .

NGS methods usually have a fixed price per run, thus

a serious reduction of in cost can be achieved by analyz-

ing multiple samples in one sequencing run, while bar-

coding DNA samples in order to determine the origin of

every read obtained. In theory, maximal multiplexing can

dramatically decrease the price. Unfortunately, the time

in which the test can be performed is naturally limited by

the growth of the fetus. That is why it is not always possi-

ble for a clinic to collect hundreds of samples for each run

if runs start approximately every 10 – 20 days (rough esti-

mation, 11 days is the run time for the Illumina HiSeq2000

sequencer. We also believe that the test will be of no use if

it takes more than a month to perform).

We compared some of the prenatal diagnostic methods

based on the equipment they use (Table 2 ). According to

rough estimates of minimal sequencing depth, equip-

ment required, and maximal multiplexing, the Ariosa

Diagnostics method appears to be the most cost-effective.

Targeted approaches can be the only choice if performed

on sequencing machines with lower run price and lower

throughput, like PGM or MiSeq, which seems to be a more

practical choice for a clinic with tens rather than hun-

dreds of pregnancies to analyze every month. For targeted

approaches, the sample-enrichment step of particular

DNA regions should also be taken into account, both in

terms of the cost of the equipment and reagents and in

terms of personnel trained to perform it.

Despite providing high accuracy and sensitivity early

in the pregnancy without risk to the fetus or the mother,

NGS-based NIPD of aneuploidy has several disadvantages

slowing down the propagation into the mainstream clini-

cal use. The long test turnaround time, high reagent and

equipment costs and high percentage of cases, where

the diagnosis cannot be made due to insufficient cffDNA

content or other factors impede mass adoption. When

deciding on the NGS-based NIPD strategy these additional

factors should be considered in addition to the diagnos-

tic sensitivity, specificity and scalability. The concept of

clinical utility may include elements of whether the clini-

cal outcomes are effective and whether its implementa-

tion offers an economically efficient solution compared to

alternative methods [54, 55] .

Conclusions Since the discovery of the cell-free fetal nucleic acid

sequences in maternal peripheral blood, several methods

for highly accurate and highly sensitive aneuploidy testing

using NGS technology either for full genome sequenc-

ing or sequencing of targeted areas of the genome were

developed. Prenatal tests utilizing these methods are

already offered as screening tests for trisomy 21, 18 and

13, reducing the need for risky invasive procedures.

Additional clinical trials are underway to validate these

methods for use as diagnostic tests for both high-risk

pregnancies and screening of the general population.

The final decision on the implementation of a NGS-based

test for NIPD of aneuploidy in clinical practice should

be based on the criteria of high diagnostic accuracy,

Company a Study Sequencing depth, reads

Equipment used Multiplexing

Ariosa diagnostics Sparks et al., 2012 a [19] 204 K – 410 K Illumina HiSeq 2000 96

Sequenom Palomaki et al., 2011 [33] NA Illumina HiSeq 2000 4*8 = 32 (1)

Verinata health Fan et al., 2010 [26] ∼ 10 M Solexa/Illumina 3 × 10 9 /10 × 10 6 = 300

Table 2   Sequencing throughput requirements for selected NIPD methods.

a Clinical study sponsor.

Unauthenticated | 95.165.92.74Download Date | 11/23/13 10:46 PM

Nepomnyashchaya et al.: Non-invasive prenatal diagnostics of aneuploidy using NGS      1151

clinical and cost-effectiveness and the ability to make a

diagnosis even in cases where the content of cffDNA is

low. Furthermore, large-scale validation studies should

be carried out independent from the tests ’ manufactur-

ing companies. Tests implemented in a clinical setting

should not be time consuming, which is very important

in prenatal diagnosis. It is also important to take into

account the nationality of the patients in order to imple-

ment the test in clinics around the world. Tests should

also require a minimal cost of equipment and infrastruc-

ture in order to be available to small laboratories around

the world. Today NGS-based tests for diagnosis of tri-

somies 21, 18 and 13 may be combined with ultrasono-

graphic detection and serum markers for more accurate

diagnosis of fetal aneuploidy, in order to avoid invasive

procedures. Methods utilizing full genome sequencing

allow for accurate detection of other autosomal and sex

aneuploidies, but are limited by the high cost of sequenc-

ing. Sequencing of targeted areas of the genome allows

one to significantly lower the cost of sequencing while

providing high accuracy and sensitivity in diagnosing

common aneuploidies. Methods utilizing parental geno-

types, where DNA from one or both parents is available,

in addition to common trisomy detection, provide for

highly accurate counts of autosomes and sex chromo-

somes and can be performed using significantly cheaper

and easier to operate sequencing equipment. Our review

demonstrated that NGS-based NIPD is a rapidly evolving

field with many research teams developing and commer-

cializing tests using new technologies and performing

large scale clinical trials. As the new NGS technologies

become available, new methods for NIPD will be devel-

oped that allow the analysis of a broader spectrum of

chromosomal abnormalities and genetic diseases, and

cost will be reduced. Several commercial NIPD provid-

ers developed proprietary fetal quantifiers and proto-

cols for increasing diagnostic accuracy of the tests and

these may not be publicly available. All of the reviewed

methods bear equipment, technology, cost, intellectual

property and performance risk; thus, careful consid-

eration should be given to each of these aspects when

deploying or developing NGS-based NIPD in a clinical

setting.

Data sources and method of study selection We searched the PubMed, PubMed Central, Bookshelf,

FreePatentsOnline and ClinicalTtrials.gov databases for

reports published after 1997 using the key words – ‘ pre-

natal diagnostics aneuploidy ’ and ‘ fetal next-generation

DNA sequencing ’ . Both thorough and theoretical reviews

of relevant full-text articles were performed. An extensive

analysis of references and literature sources was con-

ducted for the most relevant publications.

Conflict of interest statement Authors ’ conflict of interest disclosure: The authors stated that there

are no conflicts of interest regarding the publication of this article.

Research funding: None declared.

Employment or leadership: None declared.

Honorarium: None declared.

Received May 4, 2012; accepted August 29, 2012; previously

published online September 29, 2012

References 1. ACOG. ACOG Practice Bulletin No. 88, December 2007. Invasive

prenatal testing for aneuploidy. Obstet Gynecol 2007;110:

1459 – 67.

2. Lo YM, Corbetta N, Chamberlain PF, Rai V, Sargent IL, Redman

CW, et al. Presence of fetal DNA in maternal plasma and serum.

Lancet 1997;350:485 – 7.

3. Fan HC, Quake SR. Detection of aneuploidy with digital

polymerase chain reaction. Anal Chem 2007;79:7576 – 9.

4. Lo YM, Lun FM, Chan KC, Tsui NB, Chong KC, Lau TK, et al.

Digital PCR for the molecular detection of fetal chromosomal

aneuploidy. Proc Natl Acad Sci USA 2007;104:13116 – 21.

5. Fan HC, Blumenfeld YJ, El-Sayed YY, Chueh J, Quake SR.

Microfluidic digital PCR enables rapid prenatal diagnosis of fetal

aneuploidy. Am J Obstet Gynecol 2009;200:e541 – 9.

6. Chan KC, Zhang J, Hui AB, Wong N, Lau TK, Leung TN, et al. Size

distributions of maternal and fetal DNA in maternal plasma. Clin

Chem 2004;50:88 – 92.

7. Li Y, Zimmermann B, Rusterholz C, Kang A, Holzgreve W, Hahn

S. Size separation of circulatory DNA in maternal plasma

permits ready detection of fetal DNA polymorphisms. Clin Chem

2004;50:1002 – 11.

8. Alberry M, Maddocks D, Jones M, Abdel Hadi M, Abdel-Fattah S,

Avent N, et al. Free fetal DNA in maternal plasma in anembryonic

pregnancies: confirmation that the origin is the trophoblast.

Prenat Diagn 2007;27:415 – 8.

9. Lun FM, Chiu RW, Chan KC, Leung TY, Lau TK, Lo YM. Microfluidics

digital PCR reveals a higher than expected fraction of fetal DNA in

maternal plasma. Clin Chem 2008;54:1664 – 72.

Unauthenticated | 95.165.92.74Download Date | 11/23/13 10:46 PM

1152      Nepomnyashchaya et al.: Non-invasive prenatal diagnostics of aneuploidy using NGS

10. Lo YM, Zhang J, Leung TN, Lau TK, Chang AM, Hjelm NM. Rapid

clearance of fetal DNA from maternal plasma. Am J Hum Genet

1999;64:218 – 24.

11. Lo YM, Chan KC, Sun H, Chen EZ, Jiang P, Lun FM, et al. Maternal

plasma DNA sequencing reveals the genome-wide genetic and

mutational profile of the fetus. Sci Transl Med 2010;2:61ra91.

12. Metzker ML. Sequencing technologies – the next generation.

Nat Rev Genet 2010;11:31 – 46.

13. Kitzman JO, Snyder MW, Ventura M, Lewis AP, Qiu R, Simmons

LE, et al. Noninvasive whole-genome sequencing of a human

fetus. Sci Transl Med 2012;4:137ra76.

14. Fan HC, Blumenfeld YJ, Chitkara U, Hudgins L, Quake SR.

Noninvasive diagnosis of fetal aneuploidy by shotgun

sequencing DNA from maternal blood. Proc Natl Acad Sci USA

2008;105:16266 – 71.

15. Chiu RW, Chan KC, Gao Y, Lau VY, Zheng W, Leung TY,

et al. Noninvasive prenatal diagnosis of fetal chromosomal

aneuploidy by massively parallel genomic sequencing of DNA in

maternal plasma. Proc Natl Acad Sci USA 2008;105:20458 – 63.

16. Bianchi DW, Platt LD, Goldberg JD, Abuhamad AZ, Sehnert AJ,

Rava RP. Genome-wide fetal aneuploidy detection by maternal

plasma DNA sequencing. Obstet Gynecol 2012;119:890 – 901.

17. Palomaki GE, Deciu C, Kloza EM, Lambert-Messerlian GM,

Haddow JE, Neveux LM, et al. DNA sequencing of maternal

plasma reliably identifies trisomy 18 and trisomy 13 as well as

Down syndrome: an international collaborative study. Genet

Med 2012;3:296 – 305.

18. Rabinowitz M, Gemelos G, Banjevic M, Ryan A, Demko Z, Hill M,

et al. Methods for non-invasive prenatal ploidy calling. Patent

2011;App:US 2011/0288780.

19. Sparks AB, Wang ET, Struble CA, Barrett W, Stokowski R,

McBride C, et al. Selective analysis of cell-free DNA in maternal

blood for evaluation of fetal trisomy. Prenat Diagn 2012;32:3 – 9.

20. Sparks AB, Struble CA, Wang ET, Song K, Oliphant A. Non-invasive

prenatal detection and selective analysis of cell-free DNA

obtained from maternal blood: evaluation for trisomy 21 and

trisomy 18. Am J Obstet Gynecol 2012;206:319.e1 – 9.

21. Wright D, Spencer K, Kagan KK, T ø rring N, Petersen OB,

Christou A, et al. First-trimester combined screening for trisomy

21 at 7 – 14 weeks ’ gestation. Ultrasound Obstet Gynecol

2010;36:404 – 11.

22. Wilson RD. Amniocentesis and chorionic villus sampling. Curr

Opin Obstet Gynecol 2000;12:810 – 6.

23. Mujezinovic F, Alfirevic Z. Procedure-related complications of

amniocentesis and chorionic villous sampling: a systematic

review. Obstet Gynecol 2007;110:687 – 94.

24. Lander ES, Linton LM, Birren B, Nusbaum C, Zody MC, Baldwin

J, et al. Initial sequencing and analysis of the human genome.

Nature 2001;409:860 – 921.

25. Chiu RW, Sun H, Akolekar R, Clouser C, Lee C, McKernan K,

et al. Maternal plasma DNA analysis with massively parallel

sequencing by ligation for noninvasive prenatal diagnosis of

trisomy 21. Clin Chem 2010;56:459 – 63.

26. 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:e10439.

27. Chen EZ, Chiu RW, Sun H, Akolekar R, Chan KC, Leung TY,

et al. Noninvasive prenatal diagnosis of fetal trisomy 18 and

trisomy 13 by maternal plasma DNA sequencing. PLoS One

2011;6:e21791.

28. Sehnert AJ, Rhees B, Comstock D, de Feo E, Heilek G, Burke J,

et al. Optimal detection of fetal chromosomal abnormalities by

massively parallel DNA sequencing of cell-free fetal DNA from

maternal blood. Clin Chem 2011;57:1042 – 9.

29. Lo YM, Tein MS, Lau TK, Haines CJ, Leung TN, Poon PM, et al.

Quantitative analysis of fetal DNA in maternal plasma and

serum: implications for noninvasive prenatal diagnosis. Am J

Hum Genet 1998;62:768 – 75.

30. Samura O, Miharu N, Hyodo M, Honda H, Ohashi Y, Honda N,

et al. Cell-free fetal DNA in maternal circulation after

amniocentesis. Clin Chem 2003;49:1193 – 5.

31. Ehrich M, Deciu C, Zwiefelhofer T, Tynan JA, Cagasan L, Tim R,

et al. Noninvasive detection of fetal trisomy 21 by sequencing of

DNA in maternal blood: a study in a clinical setting. Am J Obstet

Gynecol 2011;204:205.e1 – 11.

32. Chiu RW, Akolekar R, Zheng YW, Leung TY, Sun H, Chan KC, et al.

Non-invasive prenatal assessment of trisomy 21 by multiplexed

maternal plasma DNA sequencing: large scale validity study.

Br Med J 2011;342:c7401.

33. Palomaki GE, Kloza EM, Lambert-Messerlian GM, Haddow JE,

Neveux LM, Ehrich M, et al. DNA sequencing of maternal plasma

to detect Down syndrome: an international clinical validation

study. Genet Med 2011;13:913 – 20.

34. Liao GJ, Lun FM, Zheng YW, Chan KC, Leung TY, Lau TK, et al.

Targeted massively parallel sequencing of maternal plasma DNA

permits efficient and unbiased detection of fetal alleles. Clin

Chem 2011;57:92 – 101.

35. Gnirke A, Melnikov A, Maguire J, Rogov P, LeProust EM,

Brockman W, et al. Solution hybrid selection with ultra-long

oligonucleotides for massively parallel targeted sequencing.

Nat Biotech 2009;27:82 – 189.

36. Porreca GJ, Zhang K, Li JB, Xie B, Austin D, Vassallo SL, et al.

Multiplex amplification of large sets of human exons. Nat

Methods 2007;4:931 – 6.

37. Turner EH, Lee C, Ng SB, Nickerson DA, Shendure J. Massively

parallel exon capture and library-free resequencing across 16

genomes. Nat Methods 2009;6:315 – 6.

38. Natera, Inc. Available from: http://natera.com. Accessed on 25

June 2012.

39. Dohm JC, Lottaz C, Borodina T, Himmelbauer H. Substantial

biases in ultra-short read data sets from high-throughput DNA

sequencing. Nucleic Acids Res 2008;36:e105.

40. Chiang DY, Getz G, Jaffe DB, O ’ Kelly MJ, Zhao X, Carter SL,

et al. High resolution mapping of copy-number alterations

with massively parallel sequencing. Nat Methods 2009;6:

99 – 103.

41. Chu T, Bunce K, Hogge WA, Peters DG. Statistical model for

whole genome sequencing and its application to minimally

invasive diagnosis of fetal genetic disease. Bioinformatics

2009;25:1244 – 50.

42. Oliver JL, Carpena P, Roman-Roldan R, Mata-Balaguer T, Mejias-

Romero A, Hackenberg M, et al. Isochore chromosome maps of

the human genome. Gene 2002;300:117 – 27.

43. Canick JA, Kloza EM, Lambert-Messerlian GM, Haddow JE,

Ehrich M, van den Boom D, et al. DNA sequencing of

maternal plasma to identify Down syndrome and other

trisomies in multiple gestations. Prenat Diagn 2012;32:

740–4.

44. Sequenom, Inc. Available from: http://www.sequenom.com.

Accessed on 25 June 2012.

Unauthenticated | 95.165.92.74Download Date | 11/23/13 10:46 PM

Nepomnyashchaya et al.: Non-invasive prenatal diagnostics of aneuploidy using NGS      1153

45. Verinata Health, Inc. Available from: http://www.verinata.com

Accessed on 25 June 2012.

46. Ariosa Diagnostics, Inc. Available from: http://www.ariadx.com.

Accessed on 25 June 2012.

47. Canick J. Prenatal screening for trisomy 21: recent advances and

guidelines. Clin Chem Lab Med 2011;50:1003 – 8.

48. Miao ZY, Liu X, Shi TK, Xu Y, Song QH, Tang SH. First trimester,

second trimester, and integrated screening for Down ’ s

syndrome in China. J Med Screen 2012;19:68 – 71.

49. Boyd P, Rounding C, Chamberlain P, Wellesley D, Kurinczuk

J. The evolution of prenatal screening and diagnosis and its

impact on an unselected population over an 18-year period.

BJOG 2012;119:1131 – 40.

50. Platt LD, Greene N, Johnson A, Zachary J, Thom E, Krantz D,

et al. First trimester maternal serum biochemistry and fetal

nuchal translucency screening (BUN) study group. Sequential

pathways of testing after first-trimester screening for trisomy

21. Obstet Gynecol 2004;104:661 – 6.

51. Malone FD, Canick JA, Ball RH, Nyberg DA, Comstock CH,

Bukowski R, et al. First- and second-trimester evaluation of

risk (FASTER) research consortium. First-trimester or second-

trimester screening, or both, for Down ’ s syndrome. N Engl J Med

2005;353:2001 – 11.

52. Crandall BF, Chua C. Detecting neural tube defects by

amniocentesis between 11 and 15 weeks ’ gestation. Prenat

Diagn 1995;15:339 – 43.

53. Barber R, Shalat S, Hendricks K, Joggerst B, Larsen R, Suarez

L, et al. Investigation of folate pathway gene polymorphisms

and the incidence of neural tube defects in a Texas hispanic

population. Mol Genet Metab 2000;70:45 – 52.

54. Ashcroft R. What is clinical effectiveness ? Stud Hist Philos Biol

Biomed Sci 2002;33:219 – 33.

55. Gray A. Critical appraisal of methods: economic evaluation. In:

Dawes M, Davies P, Gray A, Mant J, Seers K, Snowball R, editors.

Evidence-based practice: a primer for healthcare professionals,

2nd ed. London: Elsevier, 2005:121–33.

Dr. Yana N. Nepomnyashchaya, PhD, is a research fellow at the

Laboratory of Bioinformatics and Medical Information Technology at

the Federal Research and Clinical Center for Pediatric Hematology,

Oncology and Immunology. Her primary research interests include

non-invasive prenatal diagnosis of aneuploidy and single gene dis-

orders. She is working on developing a novel optimized algorithm to

estimate the risk of the fetal abnormalities, such as Down syndrome

on the basis of nucleotide sequences data analysis.

Artem Artemov is a research fellow at the Laboratory of Bioinformat-

ics and Medical Information Technology at the Federal Research and

Clinical Center for Pediatric Hematology, Oncology and Immunology

and a PhD student at the Laboratory of Bioinformatics at Moscow

State University (Department of Bioengineering and Bioinformatics),

where he previously received a Master ’ s degree in bioinformatics.

He is mainly interested in epigenetics and non-invasive cell-free

DNA-based diagnostics of fetal abnormalities as well as cancer. He

is currently working on developing a novel optimized algorithm to

estimate the risk of the fetal abnormalities.

Prof. Sergey A. Roumiantsev, MD, PhD, DSc, is Head of the Depart-

ment of Molecular and Experimental Medicine at the Federal Clinical

Research Center of Pediatric Hematology, Oncology and Immunol-

ogy, and a Professor at the Russian National Research Medical Uni-

versity (Department of Oncology and Hematology, Pediatric Faculty).

He has more than 160 scientific publications in medical journals.

His primary research interests include molecular diagnostics,

regenerative medicine, immunobiology and immunopharmacology,

biology of neoplastic growth, properties of leukemic and normal

blood and bone marrow cells, normal and leukemic hematopoiesis,

stem cell banking, angiogenesis, oncogenetic, target therapy, and

blood doping.

Unauthenticated | 95.165.92.74Download Date | 11/23/13 10:46 PM

1154      Nepomnyashchaya et al.: Non-invasive prenatal diagnostics of aneuploidy using NGS

Alexander G. Rumyantsev, academician of RAMS, prof., MD, PhD,

DSc. The director of Federal Clinical Research Center of Pediatric

Hematology, Oncology and Immunology Department of Health and

Social Development. He has published over 650 scientific papers,

including 35 monographs and manuals primarily in prenatal, perina-

tal and neonatal medicine. A.G. Rumyantsev is the chief pediatrician

of the Department of Health in Moscow, a member of the Union of

Pediatrics and the Medical Association of Russia, member of inter-

national organizations, pediatricians, hematologists, and pediatric

oncologists. He is a full member of the Russian Academy of Medical

Sciences and the Academy of Biomedical Department of Natural Sci-

ences. He is also a member of the New York Academy of Sciences.

Alex Zhavoronkov, PhD, is a head of the Bioinformatics and Medical

Information Technology Laboratory at the Federal Clinical Research

Center for Pediatric Hematology, Oncology and Immunology in

Moscow. He is also a director and trustee of the Biogerontology

Research Foundation, a UK-based registered charity supporting

aging research worldwide and a director of the International Aging

Research Portfolio (IARP) knowledge management project. He also

heads NeuroG, a neuroinformatics project intended to assist the

elderly suffering from dementia. His primary research interests

include systems biology of aging, regenerative medicine, next-gen-

eration sequencing, molecular diagnostics and pathway analysis.

He holds two Bachelor Degrees from Queen ’ s University, a Masters

in Biotechnology from Johns Hopkins University, and a PhD in Bio-

physics from the Moscow State University.

Unauthenticated | 95.165.92.74Download Date | 11/23/13 10:46 PM