Clinical Chemistry 63:2 Molecular Diagnostics and Genetics

513–524 (2017)

Universal Haplotype-Based Noninvasive Prenatal

Testing for Single Gene Diseases
Winnie W.I. Hui,1 Peiyong Jiang,1,2 Yu K. Tong,1 Wing-Shan Lee,1 Yvonne K.Y. Cheng,3 Maria I. New,4
Rezan A. Kadir,5,6 K.C. Allen Chan,1,2 Tak Y. Leung,3 Y.M. Dennis Lo,1,2 and Rossa W.K. Chiu1,2*

BACKGROUND: Researchers have developed approaches approach is universally applicable to pregnancies at risk
for the noninvasive prenatal testing of single gene dis- for the inheritance of a single gene disease.
eases. One approach that allows for the noninvasive as- © 2016 American Association for Clinical Chemistry
sessment of both maternally and paternally inherited
mutations involves the analysis of single nucleotide poly-
morphisms (SNPs) in maternal plasma DNA with refer- The presence of cell-free fetal DNA in maternal plasma
ence to parental haplotype information. In the past, pa- (1 ) offers a noninvasive approach for prenatal diagnosis.
rental haplotypes were resolved by complex experimental Maternal plasma DNA analysis for the screening of com-
methods or inferential approaches, such as through the mon fetal chromosomal aneuploidies has been achieved
analysis of DNA from other affected family members. with high degree of accuracy (2, 3 ) resulting in substan-
Recently, microfluidics-based linked-read sequencing tial reductions in the number of invasive prenatal diag-
technology has become available and allows the direct nostic procedures performed.
haplotype phasing of the whole genome rapidly. We ex- Apart from fetal aneuploidies, single gene disease is
plored the feasibility of applying this direct haplotyping the other reason why some pregnant women consider
technology in noninvasive prenatal testing. prenatal diagnosis. Since fetal DNA is present in a back-
ground of maternal DNA (4 ), early work for the nonin-
METHODS: We first resolved the haplotypes of parental vasive determination of single gene disease inheritance
genomes with the use of linked-read sequencing technol- focused on the analysis of paternally transmitted fetal-
ogy. Then, we identified SNPs within and flanking the specific sequences or mutations that could be distin-
genes of interest in maternal plasma DNA by targeted guished from the maternal genome. For example, the
sequencing. Finally, we applied relative haplotype dosage detection of chromosome Y sequences in maternal
analysis to deduce the mutation inheritance status of the plasma allowed accurate fetal sex determination and
fetus. hence served as a means to evaluate the risk of a fetus for
having a sex-linked disorder (5–7 ). The presence or ab-
sence of paternally inherited mutant alleles in maternal
RESULTS: Haplotype phasing and relative haplotype dos-
plasma has been applied to the noninvasive assessment of
age analysis of 12 out of 13 families were successfully paternally inherited autosomal dominant diseases or for
achieved. The mutational status of these 12 fetuses was the exclusion of the fetus being affected by an autosomal
correctly classified. recessive disease (8 –10 ).
To assess the fetal inheritance of maternally trans-
CONCLUSIONS: High-throughput linked-read sequencing mitted mutations, approaches have been developed to
followed by maternal plasma-based relative haplotype compare the relative amounts of the mutant and wild-
dosage analysis represents a streamlined approach for type alleles or haplotypes in maternal plasma. The rela-
noninvasive prenatal testing of inherited single gene dis- tive mutation dosage approach directly measures the
eases. The approach bypasses the need for mutation- number of DNA molecules in maternal plasma that carry
specific assays and is not dependent on the availability of the mutant or wild-type alleles. For a mother who is a
DNA from other affected family members. Thus, the carrier of a mutation, equal amounts or skewed amounts

1
Department of Chemical Pathology, The Chinese University of Hong Kong, Hong Kong * Address correspondence to this author at: Department of Chemical Pathology, Prince of
SAR, China; 2 Centre for Research into Circulating Fetal Nucleic Acids, Li Ka Shing Insti- Wales Hospital, 30-32 Ngan Shing St., New Territories, Hong Kong SAR, China. Fax
tute of Health Sciences, The Chinese University of Hong Kong, Shatin, New Territories, +852-2636-5090; e-mail rossachiu@cuhk.edu.hk.
Hong Kong SAR, China; 3 Department of Obstetrics and Gynaecology, The Chinese Uni- Received October 18, 2016; accepted November 17, 2016.
versity of Hong Kong, Hong Kong SAR, China; 4 Department of Pediatrics, Icahn School of Previously published online at DOI: 10.1373/clinchem.2016.268375
Medicine at Mount Sinai, NY; 5 Department of Obstetrics and Gynaecology, Royal Free © 2016 American Association for Clinical Chemistry
London NHS Foundation Trust, London, UK; 6 Katharine Dormandy Haemophilia Centre
and Thrombosis Unit, Royal Free London NHS Foundation Trust, London, UK.

513
between the 2 alleles in maternal plasma would provide for the noninvasive prenatal assessment of a number of
an indication of whether the fetus is heterozygous or autosomal and X-linked diseases, showing that this
homozygous for either allele, respectively (11, 12 ). streamlined approach enabled noninvasive detection of
The relative haplotype dosage (RHDO)7 approach, single gene disease inheritance without the need to design
on the other hand, allows the deduction of the fetal ge- bespoke assays to assess mutations on a case-by-case basis
notype by measuring the relative counts of single nucle- (22, 23 ) and only required the use of specimens from the
otide polymorphism (SNP) alleles on haplotypes linked parents.
with the mutant allele and wild-type allele in maternal
plasma DNA (13 ). This method allows the indirect mea- Materials and Methods
surement of mutations that are more challenging to be
detected by direct mutation-specific assays, such as gene GENERAL PRINCIPLES
deletion, inversion, mutations in repetitive elements and Parental haplotypes were first determined using
homologous genes (14 ). The RHDO method could be microfluidics-based linked-read sequencing (21 ) on
applied in a genome-wide (13 ) or a targeted fashion spec-
blood cell DNA obtained from the pregnant woman and
ifying the analysis for particular loci (14, 15 ).
her male partner. Maternal plasma DNA were then sub-
In RHDO analysis, maternal haplotype information
jected to targeted sequencing and SNP alleles located
is required. However, haplotype phasing strategies used
in previous studies were complicated and laborious. upstream and downstream of a disease locus were identi-
Methods to determine haplotype information include in- fied. The haplotype origin of each SNP allele is deduced.
ferential statistical analysis and direct experimental tech- A statistical comparison between the abundance of
niques. By genotyping genomic DNA of trios, including plasma DNA molecules derived from the 2 maternal hap-
the father, mother and an affected proband in the family, lotypes was performed. Another statistical comparison
SNPs linked with mutation sites could be identified and between the abundance of plasma DNA molecules de-
thus haplotypes could be deduced (14 ). This approach rived from the 2 paternal haplotypes was performed. The
restricts the application of the testing to families with a fetal genotype was then deduced based on the 2 sets of
previously affected family member whose DNA is avail- statistical results. Design of the target capture probes for
able. Alternatively, haplotypes could be preconstructed targeted sequencing is provided in the Data Supplement
by population-based inference (16 ) or reconstructed that accompanies the online version of this article at
from genomic DNA of an individual by methods such http://www.clinchem.org/content/vol63/issue2.
as clone pool dilution sequencing (17 ), contiguity-
preserving transposition sequencing (18 ) and HaploSeq SAMPLE COLLECTION
(19 ). However, these techniques require intricate exper- Patients were recruited at the Prince of Wales Hospital,
imental protocols or reagents that are not yet widely com- Hong Kong; Royal Free Hospital, London, UK; and
mercially available (20 ). Mount Sinai School of Medicine, New York, NY, with
Recently, a direct haplotype phasing approach using informed consent. Ethics approvals from all respective
microfluidics-based linked-read sequencing technology institutional boards were granted. 5–10 mL maternal
became available (21 ). Long input DNA molecules are blood samples were collected before any invasive proce-
partitioned into droplets and transformed into short bar- dures during pregnancy. Paternal and maternal blood
coded fragments for sequencing. Identical barcodes are samples were centrifuged at 1600g for 10 min at 4 °C,
used to identify short fragments that originate from the
and the plasma portion was recentrifuged at 16000g for
same long genomic DNA molecule. Computational al-
10 min at 4 °C (24 ). Plasma, buffy coat and genomic
gorithm is then used to reconstruct the short-read se-
DNA were transferred to the Hong Kong laboratory at
quencing data into long-range haplotype information.
In this study, we applied linked-read sequencing the Department of Chemical Pathology at The Chinese
technology to directly generate haplotype-resolved ge- University of Hong Kong for analysis. The paternal and
nome sequence from parental DNA. Maternal plasma maternal buffy coat DNA processing and the plasma
DNA sequencing data were interpreted with the parental DNA processing are described in the online Supplemen-
haplotype information to deduce the mutational status of tal Methods file.
the fetus. We assessed the feasibility of using this protocol
SEQUENCE READ ALIGNMENT AND MEASUREMENT OF
FRACTIONAL FETAL DNA CONCENTRATION
Information describing the sequence read alignment
7
and the measurement of fractional fetal DNA concen-
Nonstandard abbreviations: RHDO, relative haplotype dosage; SNP, single nucleotide
polymorphism; KS, Kolmogorov–Smirnov; SPRT, sequential probability ratio test; CAH, tration is provided in the online Supplemental Meth-
congenital adrenal hyperplasia; EVC, Ellis-van Creveld syndrome; chr, chromosome. ods file.

514 Clinical Chemistry 63:2 (2017)

Haplotype-Based NIPT for Single Gene Diseases

Fig. 1. Schematic diagram of the principle of haplotype phasing.

Long DNA molecules were partitioned into gel beads and amplified by unique 10xTM barcoded primers. Barcode-tagged short reads were
sequenced. Sequenced reads that shared the same barcode with the reads carrying mutant allele were linked (mutant-linked barcode reads)
and phased to the same haplotype. Wild-type–linked barcode reads were phased to the opposite haplotype. SNP information on the mutant-
or wild-type–linked haplotype were extracted for RHDO analysis.

HAPLOTYPE PHASING quencing data, barcode information of each sequenced

Haplotype phasing of genomic DNA was achieved by read was used to link the short sequenced reads into orig-
linking short read sequencing data to provide long range inal long input molecules. With sufficient dilution, the
genetic information (Fig. 1). From the buffy coat se- chance of having 2 distinct long DNA molecules that

Clinical Chemistry 63:2 (2017) 515

cover a genomic locus with opposing haplotype in the maternal wild-type–linked haplotype, i.e., haplotype
same gel bead is very low. Therefore, reads that shared the linked with the wild-type allele. If the fetus had inher-
same barcode with the ones carrying mutant alleles were ited the wild-type haplotype, an overrepresentation of
phased to the same haplotype (termed HapI or mutant- wild-type–linked haplotype would be observed. On
linked haplotype) and those which shared the same bar- the other hand, if the fetus had inherited the mutant
code with the ones carrying wild-type alleles were phased allele, both haplotypes would be equally represented.
to the opposite haplotype (termed HapII or wild-type– RHDO analysis involved a statistical evaluation of
linked haplotype). SNPs linked on the same haplotypes dosage balance or imbalance between alleles to deter-
with the mutant and wild-type alleles were noted and mine the haplotype block inherited (13 ).
were used for the subsequent maternal plasma DNA Further information about the RHDO analysis of
RHDO analysis. maternally transmitted autosomal mutations can be
found in the online Supplemental Methods file.
KOLMOGOROV–SMIRNOV TEST OF PATERNALLY
TRANSMITTED AUTOSOMAL MUTATIONS STATISTICAL ANALYSIS FOR THE ASSESSMENT OF X-LINKED
Informative SNPs where the mother was homozygous INHERITANCE
and the father was heterozygous were analyzed. If the Informative SNPs on chromosome X where mother was
fetus had inherited the mutation from father, the heterozygous were analyzed. If the fetus had inherited the
paternal-specific SNP alleles that could be detected in mutation, there would be an overrepresentation of
maternal plasma would belong to the paternal mutant- mutant-linked haplotype in maternal plasma DNA anal-
linked haplotype as identified by the haplotype analysis ysis. If the fetus had inherited the wild-type allele, there
of the paternal DNA. Kolmogorov–Smirnov (KS) test would be an underrepresentation of mutant-linked hap-
was applied to determine whether there was a statistical lotype. SPRT was used to test the 2 alternative hypothe-
difference of allelic counts between the 2 paternal haplo- ses: (a) the mutant allele was overrepresented when com-
types. Read counts of paternal-specific alleles between pared to the wild-type allele, and (b) the mutant allele was
paternal haplotypes were respectively accumulated until a underrepresented when compared to the wild-type allele
mutant-linked haplotype or a wild-type–linked haplo- (12 ).
type was classified (14 ). To minimize stochastic influ-
ences, the haplotype block was required to fit certain Results
criteria: the number of SNPs in test region ⱖ25; the
cumulative difference between 2 haplotypes ⬎0.53%; STUDY PARTICIPANTS
and the P value of the KS test ⬍0.05 (14). With the use Thirteen families at risk for a fetus with congenital adre-
of the KS test, we could minimize the analytical errors nal hyperplasia (CAH), ␤-thalassemia, Ellis-van Creveld
caused by sequencing errors by providing an unbiased syndrome (EVC), hemophilia or Hunter syndrome were
statistical comparison between the 2 haplotypes. As a recruited. Except for the pregnancy affected by EVC,
result, locations of recombination events that may occur each of the recruited families had a known family history
between paternal haplotypes could be pinpointed with of the disease for which conventional prenatal diagnosis
higher precision. was sought. For the EVC case, ultrasound examination
revealed multiple structural abnormalities that led to the
RHDO ANALYSIS OF MATERNALLY TRANSMITTED suspicion of EVC. The disease status of the fetus was
AUTOSOMAL MUTATIONS determined by conventional prenatal assessment based
RHDO analysis based on sequential probability ratio on mutational analysis of the parental DNA and the fetal
test (SPRT) classification was performed to deduce the DNA obtained by chorionic villus sampling or amnio-
fetal inheritance of the maternally transmitted muta- centesis or after delivery by cord blood or newborn DNA
tions (13, 14 ). Informative SNPs where mother was analysis. The mutational status of the studied cases is
heterozygous and father was homozygous were ana- listed in Table 1.
lyzed. Each SNP was classified as type ␣ or type ␤. For
type ␣ SNPs, the paternal alleles were identical to the SEQUENCING
maternal alleles on the maternal mutant-linked haplo- For the CAH families, linked short reads were pre-
type. If the fetus had inherited the mutant allele, an pared from the parental buffy coat DNA that were
overrepresentation of mutant-linked haplotype would target captured and sequenced to an average of 646-
be observed in maternal plasma DNA. In contrast, if fold haploid human coverage. For the other families,
the fetus had inherited the wild-type allele, there genome-wide sequencing of the linked short reads pre-
would be no overrepresentation of either one of the pared from the parental buffy coat DNA was per-
maternal haplotypes. For type ␤ SNPs, the paternal formed to a mean of 34-fold haploid coverage. N50
alleles were identical to the maternal alleles on the phase block length of the parental DNA samples

516 Clinical Chemistry 63:2 (2017)

 Table 1. Clinical information of the recruited families.

Genotypes Gestation
Location Fetal age Fetal DNA
Family Diseases Gene of gene Mother Father Fetus sex (weeks)a concentration (%)
A CAH CYP21A2 chr6 del/nLb Int2/nL nL/nL F 8 1/7 5.4
B CAH CYP21A2 chr6 Int2/nL Ex3/nL Ex3/Int2 M 11 1/7 8.8
C CAH CYP21A2 chr6 del/nL del/nL nL/nL F 9 6/7 8.5
D CAH CYP21A2 chr6 del/nL L307frameshift/R356W L307fs/del F 8 2/7 15.3
Haplotype-Based NIPT for Single Gene Diseases

E Beta thalassemia HBB chr11 c.316-197C>T/nL c.126_129delCTTT/nL c.316-197C>T/nL F 17 3/7 4.7

F EVC EVC and EVC2 chr4 c.1006-12A>G/nL in c.871-2_894del/nL c.1006-12A>G/c.871-2_894del F 20 3/7 14.7
EVC2 and c.83T>C/nL in EVC2 in EVC2 and c.83T>C/nL
in EVC in EVC
G Hemophilia A F8 chrX c.1898T>G/nL N/A nL M 21 4/7 23.1
H Hemophilia A F8 chrX c.71A>G/nL N/A nL M 34 11.8
I Hemophilia A F8 chrX c.1292T>C/nL N/A c.1292T>C M 31 12.3
J Hemophilia B F9 chrX c.802T>A/nL N/A c.802T>A M 21 2/7 22.2
K Hemophilia B F9 chrX c.509G>A/nL N/A nL M 30 6/7 10.1
L Hemophilia B F9 chrX c.391 + 10T>G/nL N/A c.391 + 10T>G M 2nd trimester 21.1
M Hunter syndrome IDS and IDS2 chrX Heterozygous IDS/IDS2 N/A Hemizygous IDS/IDS2 M 13 1/7 12.6
rearrangement rearrangement

a
At the time of blood sampling for analysis.
b
del, 30-kb large gene deletion; Int2, intron 2, c.293-13A/C>G; ex3, exon3, c.332_339del; nL, normal allele; N/A, not applicable.

Clinical Chemistry 63:2 (2017) 517

 Table 2. Haplotype phasing data.

Length of phase
Phase block across block across No. of SNPs
Family Sample target region target region (bases) across target region

A Mother chr6:27979631-32679591 4 699 960 4519

Father chr6:24958611-32429077 7 470 466 4631
B Mother chr6:24958611-32414273 7 455 662 4079
Father chr6:24985475-32414273 7 428 798 3595
C Mother chr6:24957224-38920455 13 963 231 9106
Father chr6:29293472-32431785 3 138 313 2774
D Mother chr6:24958611-32412539 7 453 928 4152
Father chr6:24957224-32456672 7 499 448 3891
E Mother chr11:3628799-12689249 9 060 450 9468
Father chr11:1868168-22588088 20 719 920 17 163
F Mother chr4:3570130-8737592 5 167 462 5292
Father chr4:3671398-7773524 4 102 126 4274
G Mother chrX:153020246-154404181 1 383 935 448
H Mother chrX:154210627-155173951 963 324 23
I Mother chrX:147798372-155025884 7 227 512 3188
J Mother chrX:135540724-139774089 4 233 365 1163
K Mother chrX:134856212-153728057 18 871 845 8767
L Mother chrX:130630587-146943274 16 312 687 7354
M Mother chrX:147662589-152231103 4 568 514 1235

ranged from 3–14 Mb with ⬎94% of SNPs phased. family A member 2 (CYP21A2)8 locus (Table 1). A ma-
N50 is an indicator of haplotyping performance and ternal blood sample was collected at the gestational age of
defined as the block length at which the sum of block 8 weeks and 1 day. We first resolved the haplotype of the
length of that block and larger blocks represents 50% parents from linked-read sequencing data of the parental
of the overall phased sequence (20, 21 ). The mean buffy coat DNA. For maternal haplotyping, we phased
sequencing depth of maternal plasma DNA was 275- the SNP alleles detected from sequenced reads that
fold. Key summary statistics of the sequencing data are showed the same barcode as the wild-type allele without
shown in online Supplemental Tables 1 and 2. the 30-kb deletion as haplotype II. Then, we inferred the
other alleles as derived from the haplotype linked with
PRENATAL ASSESSMENT FOR AUTOSOMAL RECESSIVE the 30-kb deletion (Fig. 3A). The phased maternal hap-
DISEASES lotype block across the target gene was around 4.7 Mb in
Families A to F each presented for prenatal assessment of length and contained 4519 informative SNPs for subse-
an autosomal recessive disease. The mutant-linked and quent maternal plasma RHDO analysis (Table 2). The
the wild-type–linked haplotypes for the mother as well as paternal point mutation was located on chromosome
the father were successfully determined for each case (Ta- (chr) 6, at genomic coordinate 32006858 (GRCh37/
ble 2 and online Supplemental Table 1). The fetal inher- hg19). Sequenced reads that shared the same barcodes
itance of the maternal and paternal haplotypes was deter- with the ones containing the paternal mutant allele were
mined through statistical comparisons between the phased to one haplotype (HapIII). Those shared the same
maternal plasma DNA sequencing reads. Results for each barcodes with reads carrying wild-type alleles were
case are shown in Fig. 2. The deduced fetal genotypes
were concordant with the results of the conventional di-
agnostic tests. 8
Human genes: CYP21A2, cytochrome P450 family 21 subfamily A member 2; IDS, idu-
As an illustration, the father in family A was a carrier ronate 2-sulfatase; IDSP1 or IDS2, iduronate 2-sulfatase pseudogene 1; CYP21A1P, cy-
tochrome P450 family 21 subfamily A member 1, pseudogene; F8, coagulation factor
of a point mutation while the mother was a carrier of a VIII; EVC2, EvC ciliary complex subunit 2; EVC, EvC ciliary complex subunit 1; HBB, he-
30-kb deletion at the cytochrome P450 family 21 sub- moglobin subunit beta; F9, coagulation factor IX.

518 Clinical Chemistry 63:2 (2017)

Haplotype-Based NIPT for Single Gene Diseases

Fig. 2. Fetal haplotype analyses in families A to F.

The analysis started from the SNPs flanking the mutation site and then extended towards the telomeric and centromeric directions. Haplotype
block is denoted by an arrow. The tail and tip of the arrow indicate the start and end positions of a haplotype block. The fetal inheritance of which
maternal haplotype was determined by RHDO analysis. The fetal inheritance of which paternal haplotype was determined by KS test analysis.
A red arrow indicates the inheritance of the mutant-linked haplotype and a blue arrow indicates the inheritance of wild-type–linked haplotype.
RHDO analysis was only performed for the EvC ciliary complex subunit 2 (EVC2) locus for Family F. EVC syndrome is an autosomal recessive
disease caused by mutations in the EvC ciliary complex subunit 1 (EVC) or EVC2 genes and both parents were carriers for mutations on EVC2.

phased to the opposite haplotype (HapIV) (Fig. 3B). The classifications. For type ␤ SNPs, an overrepresentation of
phased haplotype block across the target gene was around wild-type–linked haplotype was observed in 2 SPRT clas-
7.5 Mb in length and contained 4631 informative SNPs sifications. Both analyses indicated that the fetus had in-
for subsequent maternal plasma KS test analysis (Table 2). herited the wild-type–linked haplotype from the mother.
To determine the fetal inheritance of the maternal To determine the fetal inheritance of the paternal
mutation, we counted the number of plasma DNA mol- mutation, 2863 informative SNPs within the targeted
ecules carrying informative SNP alleles. Then, we evalu- CYP21A2 region were detected in maternal plasma. 65
ated the haplotype dosage balance or imbalance of type ␣ KS tests were done across the locus (see online Supple-
and type ␤ SNPs with SPRT classification and deduced mental Table 3). Each KS test reached statistical signifi-
the haplotype block inherited by the fetus as stated in the cance (interquartile range of the P values ⫽ 10⫺10–10⫺4;
Materials and Methods section. A total of 108 type ␣ minimal cumulative difference between 2 haplotypes
SNPs and 92 type ␤ SNPs were identified and they were ⬎0.53%) indicating that there were more paternal-
counted separately in the SPRT classification (see online specific alleles on the wild-type–linked haplotype than
Supplemental Table 3). For type ␣ SNPs, an equal rep- those on the mutant-linked haplotype in maternal
resentation of both haplotypes was observed in 6 SPRT plasma. The KS test analysis supported the conclusion

Clinical Chemistry 63:2 (2017) 519

 Fig. 3. Haplotype linkage to mutation site in family A.
(A) The mother was a carrier of a 30 kb deletion. SNPs within sequenced reads that shared the same barcode as reads without the 30 kb deletion
was considered to be linked to HapII, the wild-type–linked haplotype. Reads containing the alternative alleles were assigned to HapI, mutant-
linked haplotype. (B) The father was a carrier of a point mutation. SNPs found on the mutant-linked reads were phased to one haplotype
(HapIII) and those on the wild-type–linked reads were phased to another (HapIV).

520 Clinical Chemistry 63:2 (2017)

Haplotype-Based NIPT for Single Gene Diseases

that the fetus had inherited the wild-type–linked haplo-

type from the father. We therefore concluded that the
fetus did not inherit any of the parental mutations and
was not affected by CAH.
The same processes were applied to families B to F
and the deduced fetal genotypes and hence the disease
status were concordant with the conventional prenatal
diagnostic results. It is particularly noteworthy that a
change in RHDO inheritance was observed in the plasma
DNA data for family B and F (Fig. 2). In family B, the
maternal haplotype inherited by the fetus deduced from
the RHDO analysis changed from wild-type–linked to
mutant-linked at around 28 –30 Mb on chromosome 6.
In family F, there is a shift in the deduced paternal hap-
lotype inherited by the fetus from wild-type–linked to
mutant-linked at around 5–5.5 Mb on chromosome 4.
In the Figs., a change in the color of the arrows between
blue and red indicates the location where a recombina-
tion is suspected. The suspected recombinations were
confirmed by sequencing the chorionic villus and amni- Fig. 4. Fetal haplotype analyses in families G to M.
otic fluid samples, respectively. Since males were hemizygous for chromosome X, we only ana-
lyzed maternal inheritance of the X-linked mutations. The analysis
PRENATAL ASSESSMENT OF X-LINKED DISEASES
started from the SNPs flanking the mutation site and then ex-
Families G to L had a family history of hemophilia A or tended towards the telomeric and centromeric directions. The
B. Family M had a family history for Hunter syndrome. haplotype block inherited is denoted by an arrow. The tail
Since males are hemizygous for chromosome X, only ma- and tip of the arrow indicate the start and end position of a haplo-
ternal haplotype analysis and fetal inheritance of the ma- type block. The fetal inheritance of which maternal haplotype was
ternal X-linked mutations were performed (Fig. 4). In classified by RHDO analysis. A red arrow infers that an overrepre-
family G, the mother was a carrier of a point mutation on sentation of mutant-linked SNP alleles in the maternal plasma
coagulation factor VIII (F8). Haplotypes were con- DNA was classified and indicates that the fetus had inherited the
structed from heterozygous SNPs on chromosome X de- mutant-linked haplotype. A blue arrow infers that an overrepre-
tected from maternal genomic DNA and linkage to the sentation of wild-type–linked SNP alleles in the maternal plasma
mutant or wild-type allele was determined. The length of DNA was classified and indicates that the fetus had inherited the
the reconstructed haplotype block was 1.4 Mb and con- wild-type–linked haplotype.
tained 448 informative SNPs for inheritance analysis.
Because the maternal DNA was subjected to genome-
wide sequencing while targeted sequencing was per- A recombination event was suspected from the ma-
formed for the maternal plasma sample, only 6 of the ternal plasma DNA analysis performed for family I. The
informative SNPs were detected within the target region recombination was subsequently confirmed by targeted
in maternal plasma. Nonetheless, 1 SPRT classification sequencing of placental DNA. Maternal haplotype anal-
spanning the mutation site was achieved. The result ysis and maternal plasma RHDO assessment were suc-
showed an underrepresentation of informative SNPs cessfully performed for families I to L. The deduced fetal
linked with the mutant allele and indicated that the fetus genotypes were concordant with the conventional diag-
had inherited the wild-type allele from the mother. nostic results.
In family H, the maternal haplotype was successfully In family M, the mother was heterozygous for an
resolved. However, this particular mutation was in an iduronate 2-sulfatase (IDS/IDS2) gene rearrangement.
SNP-depleted repeat region and our capture probes were PCR amplification and restriction fragment length poly-
not specifically designed to target regions spanning this morphism analysis of maternal DNA and chorionic villi
mutation site. Also, the maternal plasma volume for DNA identified a recombination that juxtaposed IDS
DNA extraction was only 0.75 mL, which was much intron 7 and IDS2 intron 7 (25, 26 ). Because of the
lower than an average of 3.68 mL plasma for the other intragenic rearrangement, there would be more short se-
samples, and this may reduce the DNA amount for quenced reads connecting the distant genomic regions.
RHDO analysis. There were therefore not enough infor- Thus, the paired ends of the sequenced maternal DNA
mative SNP data for RHDO classification. molecules that contained the recombinant would appear

Clinical Chemistry 63:2 (2017) 521

to be as long as HMW (high molecular weight) DNA turnaround time during pregnancy. Sometimes, mutation-
molecules when mapped to the reference genome. We specific assays cannot be as readily developed for some chal-
used this feature to assign SNPs to the respective haplo- lenging genomic loci (e.g., repetitive regions, existence of
types, namely SNP alleles associated with the apparently homologous genes) or for certain mutations (deletions, in-
long DNA molecules were assigned to the mutant-linked versions, gene recombinants). CYP21A2 is one such exam-
haplotype. The opposite SNP alleles were then assigned ple. The sequences of CYP21A2 share high homology with
to the wild-type–linked haplotype (see online Supple- the pseudogene cytochrome P450 family 21 subfamily A
mental Fig. 1). From RHDO analysis of maternal plasma member 1, pseudogene (CYP21A1P). Because the fetal ge-
DNA, there was an overrepresentation of mutant-linked notype was inferred from the SNP allelic ratios in maternal
SNP alleles and this indicated that the fetus had inherited plasma, assays tailor-made for the CYP21A2 mutations were
the mutant allele from the mother. The result was con- not needed.
cordant with the clinical diagnosis and the chorionic villi Instead, a series of probes for the target capture of
analysis. SNPs surrounding of a group of clinically important sin-
gle gene disease loci could be prestocked in the labora-
Discussion tory. The scale of the testing could be varied depending
on clinical needs. For example, one may elect to use only
In this study, we used a direct haplotyping method to target capture probes designed for the assessment of 1
resolve the parental haplotypes across disease loci, disease locus at a time. This strategy is suitable for the
which were then used to interpret targeted sequencing assessment of high risk pregnancies either with a family
data obtained from maternal plasma DNA. Using this history for a specific single gene disease or had been iden-
approach, the fetal mutation profiles in 12 of 13 fam- tified to be mutation carriers through screening programs
ilies, at risk for a range of single gene diseases, were (27 ). Alternatively, target capture probes relevant for sev-
successfully deduced. eral disease loci could be pooled and be analyzed concur-
Our previous study showed that the entire fetal ge- rently. This alternative strategy is useful when there are a
nome is present in maternal plasma in a constant relative number of gene loci to be tested, such as for the purpose
proportion to maternal DNA (13 ). It was therefore pos- of investigating fetal abnormalities, like congenital car-
sible to noninvasively determine the fetal inheritance sta- diac defects, detected by ultrasonography.
tus in a genome-wide scale by combining parental hap- There is also the potential to apply this noninvasive
lotype information with massively parallel sequencing testing approach in the public health setting aimed at the
data of maternal plasma DNA. Since whole-genome hap- prenatal management of diseases that are of high preva-
lotyping technologies were not mature in the past, hap- lence in the community, for example cystic fibrosis, sickle
lotype information was derived from analyzing samples cell anemia or the thalassemias, or diseases that would
of related family members such as a proband (14 ). How- benefit from prenatal (28 ) or early neonatal treatment.
ever, this meant that for most practical purposes, the When used as a public health screening tool, the capture
approach could only be applied to families where DNA probes are first used for carrier identification (29 ) where
from a previously affected member was available. With the linked-read sequencing of the parental DNA is used
the use of linked-read sequencing for direct haplotyping, to determine the parental mutations and haplotype struc-
one can use the RHDO approach for noninvasive prena- tures. The same probes are then used for the target capture
tal testing in families where no proband sample is avail- of maternal plasma DNA for haplotype-based fetal geno-
able. Indeed, here we showed that haplotyping of the type assessment. Thus, the workflow for the prenatal
parental DNA was achieved for all 13 families. We screening and detection of single gene diseases would be
showed that this direct whole-genome haplotyping much more streamlined.
method circumvented the need to analyze samples from For the haplotype-based approach to be success-
related family members affected with the disease. This ful, some requirements must be fulfilled. Noninvasive
new development not only means that the cost of the deduction of the fetal genotype can only be achieved if
analysis has reduced, it also means that noninvasive fetal the maternal plasma DNA data surrounding the dis-
genotyping could potentially be applied to most at-risk ease locus are adequate to allow statistically significant
pregnancies. dosage assessments between the parental haplotypes.
Another advantage of haplotype-based methods The amount of sequence information needed is depen-
over direct mutational analysis is that one could infer the dent on the fetal DNA fraction, the number of infor-
fetal inheritance through quantitative assessment of in- mative SNPs and the sequencing depth. For example,
formative SNP alleles in maternal plasma, obviating the we could not provide fetal inheritance assessment for
need for tailor-made mutation-specific assays (22, 23 ). family H because of the inadequate number of infor-
Such tailor-made assays need to be optimized in good mative SNPs as well as low plasma DNA sequence
time to meet the requirements for a clinically acceptable coverage of the highly repetitive region. Perhaps, the

522 Clinical Chemistry 63:2 (2017)

Haplotype-Based NIPT for Single Gene Diseases

capture probes targeting this locus could be redesigned ments: (a) significant contributions to the conception and design, acquisi-
to capture more SNPs. Computational simulation tion of data, or analysis and interpretation of data; (b) drafting or revising
the article for intellectual content; and (c) final approval of the published
showed that if the number of SNPs reached 1000 with
article.
200-fold sequencing depth, a statistical confident
RHDO classifications can be generated even with low Authors’ Disclosures or Potential Conflicts of Interest: Upon man-
fractional fetal DNA concentrations (14 ). In this uscript submission, all authors completed the author disclosure form. Dis-
closures and/or potential conflicts of interest:
study, we can classify a sample of fractional fetal DNA
concentration as low as 4.7%. One should also be Employment or Leadership: K.C.A. Chan, Xcelom and Cirina;
mindful of the effect of recombination as detected in 3 Y.M.D. Lo, Clinical Chemistry, AACC, Xcelom, and Cirina; R.W.K.
Chiu, Xcelom and Cirina.
cases in this study. A recombination event may result
Consultant or Advisory Role: K.C.A. Chan, Xcelom; R.W.K. Chiu,
in incorrect fetal genotype classification if it occurs at Xcelom; P. Jiang, Cirina.
a genomic location near the mutation. Stock Ownership: K.C.A. Chan, Xcelom and Cirina; R.W.K. Chiu,
The protocol described in this study can readily be Xcelom and Cirina; Y.M.D. Lo, Xcelom and Cirina.
employed to many cases and the turnaround time re- Honoraria: R.W.K. Chiu, Illumina.
quired is about 1–2 weeks. We have demonstrated that Research Funding: Research Grants Council of the Hong Kong SAR
Government under the Theme-based research scheme (T12-403/15-
the approach was applicable to a variety of single gene
N); Y.M.D. Lo, Hong Kong Research Grants Council Theme-Based
diseases. We believe that we have developed an approach Research Scheme T12-403/15-N, Sequenom, Sponsored Research
that could be universally applied as a generic protocol for Agreement.
the noninvasive assessment of fetal single gene diseases. Expert Testimony: None declared.
This development could make noninvasive prenatal as- Patents: K.C.A. Chan, US patent 8 467 976; Y.M.D. Lo, US patent
sessment of fetal single gene diseases more widely 8 467 976; R.W.K. Chiu, US patent 8 467 976.
adopted. Role of Sponsor: The funding organizations played no role in the
design of study, choice of enrolled patients, review and interpretation of
data, and final approval of manuscript.

Author Contributions: All authors confirmed they have contributed to Acknowledgments: We thank Dr. Liz Yuen for helpful discussion. We
the intellectual content of this paper and have met the following 3 require- also thank Yongjie Jin for technical assistance.

References
1. Lo YM, Corbetta N, Chamberlain PF, Rai V, Sargent IL, Prenatal DNA diagnosis of a single-gene disorder from 17. Kitzman JO, Mackenzie AP, Adey A, Hiatt JB, Patward-
Redman CW, Wainscoat JS. Presence of fetal DNA in maternal plasma. Lancet 2000;356:1170. han RP, Sudmant PH, et al. Haplotype-resolved ge-
maternal plasma and serum. Lancet 1997;350:485–7. 10. Chiu RW, Lau TK, Leung TN, Chow KC, Chui DH, Lo YM. nome sequencing of a Gujarati Indian individual. Nat
2. Chiu RW, Akolekar R, Zheng YW, Leung TY, Sun H, Chan Prenatal exclusion of beta thalassaemia major by ex- Biotechnol 2011;29:59–63.
KC, et al. Non-invasive prenatal assessment of trisomy amination of maternal plasma. Lancet 2002;360:998– 18. Amini S, Pushkarev D, Christiansen L, Kostem E,
21 by multiplexed maternal plasma DNA sequencing: 1000. Royce T, Turk C, et al. Haplotype-resolved whole-
large scale validity study. BMJ 2011;342:c7401. 11. Lun FM, Tsui NB, Chan KC, Leung TY, Lau TK, Charoen- genome sequencing by contiguity-preserving trans-
3. McCullough RM, Almasri EA, Guan X, Geis JA, Hicks SC, kwan P, et al. Noninvasive prenatal diagnosis of mono- position and combinatorial indexing. Nat Genet
Mazloom AR, et al. Non-invasive prenatal chromosomal genic diseases by digital size selection and relative mu- 2014;46:1343–9.
aneuploidy testing– clinical experience: 100 000 clini- tation dosage on DNA in maternal plasma. Proc Natl 19. Selvaraj S, R Dixon J, Bansal V, Ren B. Whole-genome
cal samples. PLoS One 2014;9:e109173. Acad Sci 2008;105:19920–5. haplotype reconstruction using proximity-ligation and
4. Lun FM, Chiu RW, Chan KC, Leung TY, Lau TK, Lo YM. 12. Tsui NB, Kadir RA, Chan KC, Chi C, Mellars G, Tudden- shotgun sequencing. Nat Biotechnol 2013;31:1111–
Microfluidics digital PCR reveals a higher than expected ham EG, et al. Noninvasive prenatal diagnosis of hemo- 8.
fraction of fetal DNA in maternal plasma. Clin Chem philia by microfluidics digital PCR analysis of maternal 20. Snyder MW, Adey A, Kitzman JO, Shendure J.
2008;54:1664–72. plasma DNA. Blood 2011;117:3684–91. Haplotype-resolved genome sequencing: experimen-
5. Lo YM, Tein MS, Lau TK, Haines CJ, Leung TN, Poon PM, 13. Lo YM, Chan KC, Sun H, Chen EZ, Jiang P, Lun FM, et al. tal methods and applications. Nat Rev Genet 2015;16:
et al. Quantitative analysis of fetal DNA in maternal Maternal plasma DNA sequencing reveals the genome- 344–58.
plasma and serum: implications for noninvasive prena- wide genetic and mutational profile of the fetus. Sci 21. Zheng GX, Lau BT, Schnall-Levin M, Jarosz M, Bell JM,
tal diagnosis. Am J Hum Genet 1998;62:768–75. Transl Med 2010;2:61ra91. Hindson CM, et al. Haplotyping germline and cancer
6. Costa JM, Benachi A, Gautier E. New strategy for prena- 14. New MI, Tong YK, Yuen T, Jiang P, Pina C, Chan KC, genomes with high-throughput linked-read sequenc-
tal diagnosis of X-linked disorders. N Engl J Med 2002; et al. Noninvasive prenatal diagnosis of congenital ing. Nat Biotechnol 2016;34:303–11.
346:1502. adrenal hyperplasia using cell-free fetal DNA in ma- 22. Lench N, Barrett A, Fielding S, McKay F, Hill M, Jenkins
7. Bustamante-Aragones A, Rodriguez de Alba M, ternal plasma. J Clin Endocrinol Metab 2014;99: L, et al. The clinical implementation of non-invasive
Gonzalez-Gonzalez C, Trujillo-Tiebas MJ, Diego-Alvarez E1022–30. prenatal diagnosis for single-gene disorders: chal-
D, Vallespin E, et al. Foetal sex determination in mater- 15. Lam KW, Jiang P, Liao GJ, Chan KC, Leung TY, Chiu RW, lenges and progress made. Prenat Diagn 2013;33:
nal blood from the seventh week of gestation and its Lo YM. Noninvasive prenatal diagnosis of monogenic 555–62.
role in diagnosing haemophilia in the foetuses of fe- diseases by targeted massively parallel sequencing of 23. Verhoef TI, Hill M, Drury S, Mason S, Jenkins L, Morris S,
male carriers. Haemophilia 2008;14:593–8. maternal plasma: application to beta-thalassemia. Clin Chitty LS. Non-invasive prenatal diagnosis (NIPD) for
8. Lo YM, Hjelm NM, Fidler C, Sargent IL, Murphy MF, Chem 2012;58:1467–75. single gene disorders: cost analysis of NIPD and inva-
Chamberlain PF, et al. Prenatal diagnosis of fetal RhD 16. Zeevi DA, Altarescu G, Weinberg-Shukron A, Zahdeh F, sive testing pathways. Prenat Diagn 2016;36:636–42.
status by molecular analysis of maternal plasma. N Engl Dinur T, Chicco G, et al. Proof-of-principle rapid nonin- 24. Chiu RW, Poon LL, Lau TK, Leung TN, Wong EM, Lo YM.
J Med 1998;339:1734–8. vasive prenatal diagnosis of autosomal recessive Effects of blood-processing protocols on fetal and total
9. Saito H, Sekizawa A, Morimoto T, Suzuki M, Yanaihara T. founder mutations. J Clin Invest 2015;125:3757–65. DNA quantification in maternal plasma. Clin Chem

Clinical Chemistry 63:2 (2017) 523

 2001;47:1607–13. 26. Bondeson ML, Dahl N, Malmgren H, Kleijer WJ, Ton- 28. New MI, Abraham M, Yuen T, Lekarev O. An update on
25. Lualdi S, Regis S, Di Rocco M, Corsolini F, Stroppiano M, nesen T, Carlberg BM, Pettersson U. Inversion of the prenatal diagnosis and treatment of congenital adrenal
Antuzzi D, Filocamo M. Characterization of iduronate-2- IDS gene resulting from recombination with IDS- hyperplasia. Semin Reprod Med 2012;30:396–9.
sulfatase gene-pseudogene recombinations in eight related sequences is a common cause of the Hunter 29. Bell CJ, Dinwiddie DL, Miller NA, Hateley SL, Ganusova
patients with mucopolysaccharidosis type II revealed syndrome. Hum Mol Genet 1995;4:615–21. EE, Mudge J, et al. Carrier testing for severe childhood
by a rapid PCR-based method. Hum Mutat 2005;25: 27. Samavat A, Modell B. Iranian national thalassaemia recessive diseases by next-generation sequencing. Sci
491–7. screening programme. BMJ 2004;329:1134–7. Transl Med 2011;3:65ra4.

524 Clinical Chemistry 63:2 (2017)