Clinical Chemistry 56:3
`459–463 (2010)
`
`Brief Communications
`
`Maternal Plasma DNA Analysis with
`Massively Parallel Sequencing by
`Ligation for Noninvasive Prenatal
`Diagnosis of Trisomy 21
`
`Rossa W.K. Chiu,1,2 Hao Sun,1,2 Ranjit Akolekar,3
`Christopher Clouser,4 Clarence Lee,4 Kevin McKer-
`nan,4 Daixing Zhou,4 Kypros H. Nicolaides,3 and Y.M.
`Dennis Lo1,2*
`1 Centre for Research into Circulating Fetal Nucleic
`Acids, Li Ka Shing Institute of Health Sciences, and
`2 Department of Chemical Pathology, The Chinese
`University of Hong Kong, Shatin, New Territories,
`Hong Kong SAR, China; 3 Harris Birthright Research
`Centre for Fetal Medicine, King’s College Hospital,
`London, UK; 4 Life Technologies, Beverly, MA; * ad-
`dress correspondence to this author at: Department
`of Chemical Pathology, The Chinese University of
`Hong Kong, Rm. 38061, 1/F, Clinical Sciences Bldg.,
`Prince of Wales Hospital, 30-32 Ngan Shing St., Sha-
`tin, Hong Kong SAR, China. Fax 852-2636-5090;
`e-mail loym@cuhk.edu.hk.
`
`BACKGROUND: Noninvasive prenatal diagnosis of tri-
`somy 21 (T21) has recently been shown to be achiev-
`able by massively parallel sequencing of maternal
`plasma on a sequencing-by-synthesis platform. The
`quantification of several other human chromosomes,
`including chromosomes 18 and 13, has been shown to
`be less precise, however, with quantitative biases re-
`lated to the chromosomal GC content.
`
`METHODS: Maternal plasma DNA from 10 euploid and
`5 T21 pregnancies was sequenced with a sequencing-
`by-ligation approach. We calculated the genomic rep-
`resentations (GRs) of sequenced reads from each chro-
`mosome and their associated measurement CVs and
`compared the GRs of chromosome 21 (chr21) for the
`euploid and T21 pregnancies.
`
`RESULTS: We obtained a median of 12 ⫻ 106 unique
`reads (21% of the total reads) per sample. The GRs
`deviated from those expected for some chromosomes
`but in a manner different from that previously reported
`for the sequencing-by-synthesis approach. Measure-
`ments of the GRs for chromosomes 18 and 13 were less
`precise than for chr21. z Scores of the GR of chr21 were
`increased in the T21 pregnancies, compared with the
`euploid pregnancies.
`
`CONCLUSIONS: Massively parallel sequencing-by-ligation
`of maternal plasma DNA was effective in identifying
`T21 fetuses noninvasively. The quantitative biases ob-
`served among the GRs of certain chromosomes were
`more likely based on analytical factors than biological
`
`factors. Further research is needed to enhance the pre-
`cision for measuring for the representations of chro-
`mosomes 18 and 13.
`
`The possibility of fetal chromosomal aneuploidy, par-
`ticularly trisomy 21 (T21)5 (Down syndrome), is a ma-
`jor reason why couples consider prenatal diagnostic
`studies. Current definitive practices of prenatal diag-
`nosis rely on obtaining fetal genetic material via chori-
`onic villus sampling or amniocentesis, both of which
`have associated risks of fetal miscarriage. In 1997, we
`reported the presence of cell-free fetal DNA in the cir-
`culation of pregnant women (1, 2 ). Analysis of fetal
`DNA in maternal plasma is useful for the prenatal as-
`sessment of sex-linked diseases, Rhesus D genotyping,
`and certain monogenic diseases (3, 4 ).
`The direct diagnosis of fetal chromosomal aneu-
`ploidy via nucleic acid analysis of maternal plasma has
`been more challenging (5 ). Chromosomal aneuploidy
`refers to a quantitative imbalance in the dosage of par-
`ticular chromosomes in a genome. Thus, approaches
`need to be developed to detect the quantitative pertur-
`bations associated with the aneuploid chromosome
`among the fetal DNA molecules within maternal
`plasma. Precise quantification of circulating fetal DNA
`in maternal plasma has proved difficult, however, ow-
`ing to its low fractional and absolute concentrations
`against the high background concentrations of mater-
`nal DNA (6, 7 ). Recently, our group (8 ) and Fan et al.
`(9 ) demonstrated the feasibility of noninvasive prena-
`tal diagnosis of T21 by massively parallel sequencing of
`maternal plasma DNA.
`It is possible to identify the chromosomal origin of
`each sequenced plasma DNA molecule by comparing
`its nucleotide sequence with the reference human ge-
`nome. Because a T21 fetus carries an additional copy of
`chromosome 21 (chr21) in its genome, this copy would
`contribute additional amounts of chr21 DNA frag-
`ments into the maternal plasma. The small increments
`in the proportional amounts, termed the “genomic
`representation” (GR), of chr21 sequences in the plasma
`of women carrying T21 fetuses compared with euploid
`fetuses could be detected with high precision by mas-
`sively parallel sequencing (5, 8 ).
`We reported, however, that the precision for mea-
`suring the GR varied among human chromosomes and
`tended to be worse for chromosomes with GC contents
`at either end of the GC-abundance spectrum (8 ). Fan
`
`5 Nonstandard abbreviations: T21, trisomy 21; chr21, chromosome 21; GR,
`genomic representation; %GR, percentage for a given chromosome of the total
`number of unique reads obtained for the sample.
`
`459
`
`Ariosa Exhibit 1013
`pg. 1
`
`

`

`Brief Communications
`
`et al. (9 ) observed a quantitative bias in the relative
`amounts of plasma DNA molecules sequenced from
`each human chromosome that bore a relationship to
`the chromosome’s GC content. Chromosomes with a
`low GC content were underrepresented, whereas chro-
`mosomes with a high GC content were overrepre-
`sented. Both previous studies (8, 9 ) were performed
`with the Genome Analyzer from Illumina, which uses a
`sequencing-by-synthesis approach (10 ). In particular,
`the measurements of the proportions of plasma
`DNA molecules originating from chromosomes 13
`and 18 were less precise and had greater negative
`biases than those for chr21. Because chromosomes
`13 and 18 are involved in trisomy 13 and trisomy 18,
`respectively, it would be of diagnostic interest to de-
`velop protocols that have more uniform perfor-
`mance across chromosomes.
`We collected 5 mL of blood from 15 women pre-
`senting for first trimester aneuploidy screening: 5
`women with T21 pregnancies (2 female fetuses) and 10
`women with euploid pregnancies (3 female fetuses).
`Maternal plasma DNA libraries were prepared using
`the low-input fragment DNA protocol, without shear-
`ing the DNA (11 ). Massively parallel sequencing-by-
`ligation was performed on a SOLiD™ 3 System (Ap-
`plied Biosystems/Life Technologies) according to the
`manufacturer’s protocol (11 ). Details of the methods
`are given in the Supplemental Data file in the Data
`Supplement that accompanies the online version of
`this Brief Communication at http://www.clinchem.
`org/content/vol56/issue3.
`We obtained a median of 59 ⫻ 106 raw reads
`(range, 31–78 ⫻ 106) from each sample. A median of
`12 ⫻ 106 reads/sample (range, 7–16 ⫻ 106), represent-
`ing 21% (range, 17%–23%) of the raw reads, could be
`aligned uniquely to a single location on the reference
`genome with up to 1 color space mismatch. We termed
`these reads “perfectly matched unique reads” because
`despite the 1 color space mismatch, the correct se-
`quence of the read could be deduced because each se-
`quenced base was interrogated by 2 adjacent colored
`dinucleotides. With a read length of 50 bp, a median of
`12 ⫻ 106 unique reads is equivalent to 20% coverage of
`the haploid human genome. The number of unique
`reads aligned to each human chromosome was ex-
`pressed as a percentage of the total number of unique
`reads obtained for the sample, which we termed the
`“%GR” of that chromosome. We determined the me-
`dian, mean, and SD of the %GR for each chromosome
`of the sequenced maternal plasma samples. The corre-
`sponding values for chr21 were based on the 10 euploid
`samples only, whereas those for the other chromo-
`somes were calculated from all 15 samples.
`We compared the median %GR of each chromo-
`some to that expected for a repeat-masked haploid fe-
`
`460 Clinical Chemistry 56:3 (2010)
`
`male genome, because the majority of DNA molecules
`in maternal plasma originated from a female, i.e., the
`mother. The degree of deviation from the expected
`%GR is estimated by: {[(median of the experimentally
`derived %GR for chromosome N) ⫺ (expected %GR
`for chromosome N)]/(expected %GR for chromosome
`N)} ⫻ 100%. The data are presented in Fig. 1 in the
`online Data Supplement. The CVs for %GR mea-
`surements were calculated for each chromosome; the
`results are presented in Fig. 2 in the online Data
`Supplement.
`For T21 diagnosis, 4 euploid pregnancies of
`male fetuses were selected as the control group. The
`mean and SD values of the chr21 %GR for these 4
`controls were used to determine the chr21 z scores of
`the remaining 11 samples, as previously described
`(8 ). The chr21 z score for a test case is the difference
`between the %GR of chr21 for the test case and the
`mean %GR for the same chromosome of the control
`group, divided by the SD. Because ⫹3 SDs is the
`99.9th percentile from the mean of the reference
`group for a 1-tailed distribution, a z score of ⫹3 was
`used as the cutoff value for determining overrepre-
`sentation of chr21 sequences in a maternal plasma
`sample. The chr21 z scores for the 5 T21 cases were
`8.6, 15.3, 15.6, 18.6, and 19.9. The chr21 z scores for
`the 6 euploid cases were ⫺2.1, ⫺1.5, ⫺1.0, ⫺0.7,
`⫺0.7, and 0.8. z Scores for the other autosomes were
`also calculated; these results are presented in Fig. 1.
`To further confirm the robustness of this ap-
`proach for T21 diagnosis, we used a bootstrapping pro-
`cedure on 9 of the 10 euploid cases at a time as the
`control group to reanalyze the chr21 z scores for the 5
`T21 cases and the remaining euploid case. Because
`there were 10 euploid cases, we had 10 different possi-
`ble combinations of 9 control cases. Fig. 3 in the online
`Data Supplement shows that for all 10 analyses the eu-
`ploid case had a z score ⬍3, whereas, all of the T21 cases
`had z scores ⬎3.
`We previously showed that a small number of se-
`quences would be falsely aligned to chromosome Y,
`even for female fetuses. In the present study, the %GR
`values of chromosome Y were 0.009% for 4 female fe-
`tuses and 0.023% for the remaining female fetus. The
`median %GR of chromosome Y for the male fetuses
`was 0.034% (range, 0.019%– 0.045%). Thus, the %GR
`of chromosome Y for this 1 female fetus was as high as
`that for the male fetuses. The chorionic villus of this
`fetus was negative when tested by a quantitative PCR
`assay targeting the SRY gene (sex determining region
`Y) (6 ). We cannot exclude the possibility of a mix-up
`of the maternal plasma sample with that from a preg-
`nancy with a male fetus. The accuracy of the use of the
`chromosome Y %GR for fetal sex determination would
`need to be further confirmed in future studies.
`
`Ariosa Exhibit 1013
`pg. 2
`
`

`

`Brief Communications
`
`1
`
`2
`
`3
`
`4
`
`1 2 3 4 5 6 7 8 9 101112 131415
`
`1 2 3 4 5 6 7 8 9 101112 131415
`
`1 2 3 4 5 6 7 8 9 101112 131415
`
`1 2 3 4 5 6 7 8 9 1011 12 131415
`
`5
`
`6
`
`7
`
`8
`
`1 2 3 4 5 6 7 8 9 101112 131415
`
`1 2 3 4 5 6 7 8 9 101112 131415
`
`1 2 3 4 5 6 7 8 9 101112 131415
`
`1 2 3 4 5 6 7 8 9 101112 131415
`
`9
`
`01
`
`11
`
`21
`
`1 2 3 4 5 6 7 8 9 101112 131415
`
`1 2 3 4 5 6 7 8 9 101112 131415
`
`1 2 3 4 5 6 7 8 9 101112 131415
`
`1 2 3 4 5 6 7 8 9 101112 131415
`
`31
`
`41
`
`51
`
`61
`
`1 2 3 4 5 6 7 8 9 101112 131415
`
`1 2 3 4 5 6 7 8 9 101112 131415
`
`1 2 3 4 5 6 7 8 9 101112 131415
`
`1 2 3 4 5 6 7 8 9 101112 131415
`
`71
`
`81
`
`91
`
`02
`
`1 2 3 4 5 6 7 8 9 101112 131415
`
`1 2 3 4 5 6 7 8 9 101112 131415
`
`1 2 3 4 5 6 7 8 9 101112 131415
`
`1 2 3 4 5 6 7 8 9 101112 131415
`
`12
`
`22
`
`20
`
`15
`
`10
`
`05
`
`20
`
`15
`
`10
`
`05
`
`20
`
`15
`
`10
`
`05
`
`20
`
`15
`
`10
`
`05
`
`20
`
`15
`
`10
`
`05
`
`20
`
`15
`
`10
`
`05
`
`z Score
`
`z Score
`
`z Score
`
`z Score
`
`z Score
`
`z Score
`
`1 2 3 4 5 6 7 8 9 101112 131415
`
`1 2 3 4 5 6 7 8 9 101112 131415
`
`Fig. 1. z Scores for the autosomes (chromosome number indicated by the number within each panel) in the control
`(cases 1– 4, in black), euploid (cases 5–10, in blue), and T21 (cases 11–15, in red) pregnancies.
`A z score of 3 (dashed line) was used as a cutoff to determine the presence of overrepresentation of sequences from the
`corresponding chromosome.
`
`Clinical Chemistry 56:3 (2010) 461
`
`Ariosa Exhibit 1013
`pg. 3
`
`

`

`Brief Communications
`
`Similarly with the sequencing-by-synthesis ap-
`proach (9 ), the quantitative representation of each
`chromosome was not uniform. It is interesting that un-
`like Fan et al. (9 ), who reported a positive bias for chro-
`mosomes with a high GC content, such as chromo-
`somes 19 and 22, we observed a negative bias (see Fig. 1
`in the online Data Supplement). These data suggest
`that the nonuniform representation is more likely ex-
`plained by analytical factors than biological factors
`(12, 13 ). For example, the amplification efficiency of
`the DNA libraries may be nonuniform across se-
`quences with different GC contents, although the form
`and extent of biases are different among different mas-
`sively parallel sequencing platforms. In fact, Chu et al.
`(14 ) suggested that such PCR-related biases should be
`included as a parameter in the algorithms for calculat-
`ing the chromosomal GR. To compute such sophisti-
`cated algorithms, however, would first require elucida-
`tion of the exact mechanism and extent of analytical
`bias introduced by the sequencing protocols. There are
`key differences between the protocols for the Genome
`Analyzer and the SOLiD 3 System. For clonal amplifi-
`cation of DNA libraries, the former uses solid-phase
`bridge amplification, whereas the latter uses emulsion
`PCR. Interestingly, the sequencing-by-ligation ap-
`proach demonstrated less bias for sequences from
`chromosomes with a low GC content, as exemplified
`by the data for chromosome 13.
`The precision profiles for chromosomes 18 and 13
`are still worse than for chr21. In our previous study (8 ),
`the CVs for measuring the %GR of chromosomes 21,
`18, and 13 ranked (from most to least precise) 3rd,
`10th, and 17th, respectively. In the present study, the
`corresponding rankings are 4th, 8th, and 16th. Thus,
`despite the change in the sequencing platform, the
`%GR values of chromosomes 18 and 13 were still more
`difficult to quantify as precisely as the %GR of chr21.
`Fan et al. (9 ) performed massively parallel sequencing
`of maternal plasma from 1 pregnancy each for trisomy
`18 and trisomy 13 to study the feasibility of their non-
`invasive prenatal diagnosis. Yet, our current and previ-
`ous data (8 ) suggest that noninvasive prenatal diag-
`nosis of trisomy 18 and 13 with this approach would
`
`likely be less precise than for T21. Conversely, the
`sequencing-by-ligation approach appears to be a ro-
`bust method for the direct detection of T21 fetuses.
`Large-scale clinical trials would be needed to confirm
`these findings. Currently, the complete protocol on the
`SOLiD 3 System, from DNA extraction to data inter-
`pretation, takes 7.5 days; however, we obtained 5 times
`more unique reads per sample than in our earlier study,
`when the equivalent processing time was 5 days (8 ).
`Higher sequencing counts would allow the analysis of
`⬎1 sample per reaction chamber if bar codes were
`used. Hence, there is an opportunity for cost reduction
`and increased sample throughput, which would bring
`this new technology closer to clinical application.
`
`Author Contributions: All authors confirmed they have contributed to
`the intellectual content of this paper and have met the following 3 re-
`quirements: (a) significant contributions to the conception and design,
`acquisition of data, or analysis and interpretation of data; (b) drafting
`or revising the article for intellectual content; and (c) final approval of
`the published article.
`
`Authors’ Disclosures of Potential Conflicts of Interest: Upon
`manuscript submission, all authors completed the Disclosures of Poten-
`tial Conflict of Interest form. Potential conflicts of interest:
`
`Employment or Leadership: None declared.
`Consultant or Advisory Role: Y.M.D. Lo, Sequenom.
`Stock Ownership: C. Lee, Life Technologies; K. McKernan, Life
`Technologies; D. Zhou, Life Technologies; Y.M.D. Lo, Sequenom.
`Honoraria: None declared.
`Research Funding: R.W.K. Chiu and Y.M.D. Lo, Life Technolo-
`gies, sponsor of the sequencing runs as part of a grant from the
`Innovation and Technology Fund (ITS/054/09); Y.M.D. Lo, Se-
`quenom.
`Expert Testimony: None declared.
`Other Remuneration: R.W.K. Chiu and Y.M.D. Lo hold patents and
`have filed patent applications on the detection of fetal nucleic acids in
`maternal plasma for noninvasive prenatal diagnosis. Y.M.D. Lo was
`supported by an endowed professorship from the Li Ka Shing
`Foundation.
`
`Role of Sponsor: The funding organizations played no role in the
`design of study, choice of enrolled patients, review and interpretation
`of data, or preparation or approval of manuscript.
`
`Acknowledgments: We thank Elizabeth Levandowsky and Tristen
`Weaver for preparing the libraries and performing the sequencing.
`
`References
`
`1. Lo YMD, Corbetta N, Chamberlain PF, Rai V,
`Sargent IL, Redman CW, Wainscoat JS. Presence
`of fetal DNA in maternal plasma and serum.
`Lancet 1997;350:485–7.
`2. Lo YMD, Chiu RWK. Prenatal diagnosis: progress
`through plasma nucleic acids. Nat Rev Genet
`2007;8:71–7.
`3. Lun FMF, Tsui NBY, Chan KCA, Leung TY, Lau TK,
`Charoenkwan P, et al. Noninvasive prenatal di-
`agnosis of monogenic diseases by digital size
`selection and relative mutation dosage on DNA in
`maternal plasma. Proc Natl Acad Sci U S A 2008;
`
`105:19920 –5.
`4. Wright CF, Burton H. The use of cell-free fetal
`nucleic acids in maternal blood for non-invasive
`prenatal diagnosis. Hum Reprod Update 2009;15:
`139 –51.
`5. Chiu RWK, Cantor CR, Lo YMD. Non-invasive
`prenatal diagnosis by single molecule counting
`technologies. Trends Genet 2009;25:324 –31.
`6. Lo YMD, 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.
`7. Lo YMD, Lun FMF, Chan KCA, Tsui NBY, Chong
`KC, Lau TK, et al. Digital PCR for the molecular
`detection of fetal chromosomal aneuploidy. Proc
`Natl Acad Sci U S A 2007;104:13116 –21.
`8. Chiu RWK, Chan KCA, Gao Y, Lau VYM, 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 U S A 2008;105:
`20458 – 63.
`9. Fan HC, Blumenfeld YJ, Chitkara U, Hudgins L,
`
`462 Clinical Chemistry 56:3 (2010)
`
`Ariosa Exhibit 1013
`pg. 4
`
`

`

`Brief Communications
`
`Quake SR. Noninvasive diagnosis of fetal aneu-
`ploidy by shotgun sequencing DNA from maternal
`blood. Proc Natl Acad Sci U S A 2008;105:
`16266 –71.
`10. Schuster SC. Next-generation sequencing trans-
`forms today’s biology. Nat Methods 2008;5:
`16 – 8.
`11. McKernan KJ, Peckham HE, Costa GL, McLaughlin
`SF, Fu Y, Tsung EF, et al. Sequence and structural
`variation in a human genome uncovered by short-
`read, massively parallel ligation sequencing using
`
`encoding. Genome Res 2009;19:
`
`two-base
`1527– 41.
`12. Hillier LW, Marth GT, Quinlan AR, Dooling D,
`Fewell G, Barnett D, et al. Whole-genome se-
`quencing and variant discovery in C.elegans. Nat
`Methods 2008;5:183– 8.
`13. 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.
`14. Chu T, Bunce K, Hogge WA, Peters DG. Statis-
`
`tical model for whole genome sequencing and
`its application to minimally invasive diagnosis
`of fetal genetic disease. Bioinformatics 2009;
`25:1244 –50.
`
`Previously published online at
`DOI: 10.1373/clinchem.2009.136507
`
`Clinical Chemistry 56:3 (2010) 463
`
`Ariosa Exhibit 1013
`pg. 5
`
`