Field of the Invention

The invention provides a non-invasive prenatal in vitro screening method for an autosomal recessive disorder, such as cystic fibrosis (CF), in a fetus using cell free DNA from a maternal sample.

Background to the Invention

Cystic fibrosis (CF) is a severe monogenic autosomal recessive inherited disorder caused by mutations in the CFTR gene on chromosome 7. In Caucasian and related populations, AF508 is the most common mutation but over 1 ,000 others have been documented. ¹ In Europe the prevalence is about 1 in 2500 live births ² ' ³ , with up to about 1 in 25 people being a carrier. ⁴ Current prenatal screening for CF tests couples to identify those who are both CF carriers and then offers an invasive diagnostic procedure (amniocentesis or chorionic villus sampling) to expectant mothers, one in four of whom have an affected pregnancy. ⁵ ' ^6,7,8 Detecting paternal CF mutations in cell-free DNA from maternal plasma has been described, but is not an effective screening method because it fails when the paternal and maternal CF mutations are the same ⁹ ' ¹⁰ which arises in about 50% of cases. Where the maternal and paternal mutations are different the screening task is to distinguish affected pregnancies from those in which the fetus is an unaffected carrier or non-carrier.

Summary of the Invention

The present invention provides a non-invasive prenatal method of screening for an autosomal recessive disorder, such as CF, in a fetus. A prior art method of non-invasive prenatal screening for CF by testing cell-free DNA from maternal plasma is not an effective screening method because it relies on the paternal CF mutation being different from the maternal CF mutation and therefore requires parental CF carrier testing and fails when the paternal and maternal CF mutations are the same. The present method overcomes this problem and does not require parental CF carrier testing. It relies on sequencing and counting DNA fragments and targeting predetermined mutation sites in the cell-free DNA of maternal plasma or serum. Also, using computer modelling, we estimate the number of mutation sites that need to be sequenced and counted to achieve near perfect discrimination between affected and unaffected pregnancies with a very low false-positive rate.

In addition, we provide an efficient method for screening for affected pregnancies where the maternal and paternal mutations inherited by the fetus are different (i.e. the fetus is a

"compound heterozygote"). The method therefore complements other possible methods for prenatal diagnosis of cystic fibrosis and other monogenic recessive genetic disorders in cases where the mutations in the mother and father are identical and the fetus is at risk of inheriting a pair of identical mutations in simple homozygous form. Claims for such methods involving pairs of identical mutations may be found, for example in US publication no. US2012/108460 and US publication no. US2016/053320. However, even in populations such as the European population or the population of the United States where a particular mutation (p.deltaF508) in the CFTR gene occurs frequently, compound heterozygotes account for about half of all affected pregnancies. Thus, there is a requirement for the present invention.

Accordingly, the present invention provides a non-invasive prenatal screening method for an autosomal recessive disorder in a fetus using cell- free DNA (cfDNA) from a maternal plasma or serum sample comprising:

(a) determining that two different mutant DNA sequences which, when present as a compound heterozygote in an individual cause the disorder, are present in the maternal plasma or serum;

(b) for the predominant (maternally-derived) mutation, performing DNA sequence analysis of a number of DNA molecules that are the product of amplification of the DNA region comprising the mutation (the "mutation site");

(c) calculating the ratio of the number of sequenced DNA molecules comprising the predominant (the maternally-derived) mutation to the number of sequenced DNA molecules that do not comprise the predominant mutation at that mutation site (the "ratio"), or calculating the ratio of the number of sequenced DNA molecules that do not comprise the predominant mutation to the number of sequenced DNA molecules comprising the predominant mutation (the "inverse ratio"); and

(d) comparing the ratio or inverse ratio to a predetermined cut-off value and

thereby determining whether the sample is test-positive for a fetus affected with the disorder. Also provided is a computer-implemented method of non-invasive prenatal screening for an autosomal recessive disorder in a fetus using cell-free DNA (cfDNA) from a maternal plasma or serum sample, wherein two different mutant DNA sequences which, when present as a compound heterozygote in an individual cause the disorder, are present in the maternal plasma or serum, comprising:

(a) inputting to a computer system data concerning the number of DNA molecules that are the product of amplification of the DNA region comprising the predominant (maternally-derived) mutation and the number of those DNA molecules that comprise and/or do not comprise the mutation at the mutation site;

(b) determining the ratio of the number of sequenced DNA molecules comprising the predominant (maternally-derived) mutation to the number of sequenced DNA molecules that do not comprise the predominant mutation at that mutation site (the "ratio"), or determining the ratio of the number of sequenced DNA molecules that do not comprise the predominant mutation to the number of sequenced DNA molecules comprising the predominant mutation (the "inverse ratio"); and

(c) comparing the ratio or inverse ratio to a predetermined cut-off value and

thereby determining whether the sample is test-positive for a fetus affected with the disorder.

Further provided is:

- a computer program comprising program code means that, when executed on a

computer system, instruct the computer system to perform all the steps of the computer-implemented method of the invention;

- a computer storage medium comprising the computer program of the invention; and

- a computer system arranged to perform a method according to the invention,

comprising:

(a) means for receiving data concerning the number of DNA molecules that are the product of amplification of the DNA region comprising the predominant (maternally- derived) mutation and the number of those DNA molecules that comprise and/or do not comprise the mutation at the mutation site;

(b) a module for determining the ratio of the number of sequenced DNA molecules comprising the predominant (maternally-derived) mutation to the number of sequenced DNA molecules that do not comprise the predominant mutation at that mutation site (the "ratio"), or determining the ratio of the number of sequenced DNA molecules that do not comprise the predominant mutation to the number of sequenced DNA molecules comprising the predominant mutation (the "inverse ratio"); and (c) a module for comparing the ratio or inverse ratio to a predetermined cut-off value and thereby determining whether the sample is test-positive for a fetus affected with the disorder.

The present invention also provides a non-invasive prenatal in vitro screening method for an autosomal recessive disorder, such as cystic fibrosis (CF), in a fetus using cell-free DNA (cfDNA) from a maternal plasma or serum sample, comprising:

(a) performing targeted amplification of one or more DNA regions of the cfDNA, wherein the or each DNA region comprises at least one site where an autosomal recessive mutation may occur (a "mutation site"), wherein the mutation is one which, when present homozygously in an individual, causes the disorder;

(b) performing DNA sequence analysis of a number of DNA molecules ("n") that are the product of amplification for the or each DNA region and determining the number of those DNA molecules that comprise and/or do not comprise a mutation at the or each mutation site, wherein n is:

(i) more than or equal to N x 32,000; or

(ii) less than or equal to N x 640,000; or

(iii) more than or equal to N x 32,000 but less than or equal to N x 640,000;

and wherein N is the number of different mutation sites used in the method;

(c) for a mutation site, calculating the ratio of the number of sequenced DNA molecules comprising the mutation to the number of sequenced DNA molecules that do not comprise the mutation (the "ratio"), or calculating the ratio of the number of sequenced DNA molecules that do not comprise the mutation to the number of sequenced DNA molecules comprising the mutation (the "inverse ratio"); and

(d) for the mutation site, comparing the calculated ratio or inverse ratio to a predetermined cut-off value and thereby determining whether the sample is test-positive for a fetus affected with the disorder, wherein the sample is test-positive for a fetus affected with the disorder if the calculated ratio exceeds a predetermined cut-off value or if the inverse ratio is less than the predetermined cut-off value.

Also provided is a computer-implemented method of non-invasive prenatal screening for an autosomal recessive disorder in a fetus using cell-free DNA (cfDNA) from a maternal plasma or serum sample, wherein a single autosomal recessive mutation causing the disorder is present, or two different autosomal recessive mutations are present, in the maternal plasma or serum, comprising:

(a) inputting to a computer system data concerning the number of DNA molecules that are the product of amplification of a DNA region comprising the mutation and the number of those DNA molecules that comprise and/or do not comprise the mutation at the mutation site;

(b) determining the ratio of the number of sequenced DNA molecules comprising the mutation to the number of sequenced DNA molecules that do not comprise the mutation at that mutation site (the "ratio"), or determining the ratio of the number of sequenced DNA molecules that do not comprise the mutation to the number of sequenced DNA molecules comprising the mutation (the "inverse ratio"); and

(c) comparing the ratio or inverse ratio to a predetermined cut-off value and

thereby determining whether the sample is test-positive for a fetus affected with the disorder.

Further provided is:

- a computer program comprising program code means that, when executed on a

computer system, instruct the computer system to perform all the steps of the computer-implemented method of the invention;

- a computer storage medium comprising the computer program of the invention; and

- a computer system arranged to perform a method according to claim 40, comprising:

(a) means for receiving data concerning the number of DNA molecules that are the product of amplification of the DNA region comprising the mutation and the number of those DNA molecules that comprise and/or do not comprise the mutation at the mutation site;

(b) a module for determining the ratio of the number of sequenced DNA molecules comprising the mutation to the number of sequenced DNA molecules that do not comprise the mutation at that mutation site (the "ratio"), or determining the ratio of the number of sequenced DNA molecules that do not comprise the mutation to the number of sequenced DNA molecules comprising the mutation (the "inverse ratio"); and (c) a module for comparing the ratio or inverse ratio to a predetermined cut-off value and thereby determining whether the sample is test-positive for a fetus affected with the disorder. Brief Description of the Figures

Figure 1 : Derivation of the expected (mean) percentage DNA fragments with a CF mutation in CF affected and unaffected pregnancies if the fetal fraction is 10%, in scenarios A, B, C and D. In A, both parents are CF mutation carriers. In B, the mother is a CF mutation carrier. In C, the father is a CF mutation carrier. In D, neither parent is a CF mutation carrier. Rows below the inheritance diagram provide (i) the expected percentage of DNA fragments from the mother with the CF mutation; (ii) the expected percentage of DNA fragments from the mother without the CF mutation; (iii) the expected percentage of DNA fragments from the fetus with the CF mutation (10* = 5 inherited from the mother + 5 from the father); (iv) the expected percentage of DNA fragments from the fetus without the CF mutation (10* = 5 inherited from the mother + 5 from the father); (v) the expected ratio of DNA fragments with the CF mutation to DNA fragments without the CF mutation; and (vi) the expected percentage of DNA fragments with a CF mutation (a value of 55% means that the fetus is affected and a value of 50% and below means that the fetus is unaffected).

Figure 2: Derivation of the percentage of DNA fragments with a CF mutation in maternal plasma where two different CF mutations are identified (XI and X2) and the fetal fraction is 10%. In maternal plasma the predominant mutation is always from the mother because there is no direct contribution from the father. The expected percentage of CF mutations at the predominant CF mutation site in an affected pregnancy is 50% and 45% in an unaffected pregnancy. Rows below the inheritance diagram provide (i) the expected percentage of DNA fragments from the mother with the XI (the predominant) CF mutation; (ii) the expected percentage of DNA fragments from the mother without the XI CF mutation; (iii) the expected percentage of DNA fragments from the fetus with the XI CF mutation; (iv) the expected percentage of DNA fragments from the fetus without the XI CF mutation; (v) the expected ratio of DNA fragments with the XI CF mutation to DNA fragments without the XI CF mutation; and (vi) the expected percentage of DNA fragments with a XI CF mutation (a value of 50% means that the fetus is affected and an expected value of 45% means that the fetus is unaffected). Figure 3: Relative frequency distributions of percent DNA fragments with a CF mutation in affected and unaffected pregnancies according to fetal fraction and the number of DNA fragments sequenced that include the relevant mutation site. The chart on the left is for a 10% fetal fraction of the maternal sample and the chart on the right is for a 4% fetal fraction of the maternal sample. The x axis is the percentage of DNA fragments with a CF mutation and the y axes are the distributions for each of 6 values for the number of DNA fragments sequenced that include a mutation site.

Figure 4: Relative distributions of percent DNA fragments with a CF mutation in affected and unaffected pregnancies according to fetal fraction (32,000 to 64,000 DNA fragments sequenced that include the relative mutation site). The x axis is the percentage of DNA fragments with a CF mutation and the y axes are the distributions for each of 5 fetal fraction percentages of the maternal sample. Figure 5: Two different CF mutations found: relative distributions of percent DNA fragments with the predominant cystic fibrosis (CF) mutation in affected and unaffected pregnancies according to fetal fraction (32,000 targeted DNA fragments sequenced). The x axis is the percentage of DNA fragments with a CF mutation and the y axes are the distributions for each of 4 fetal fraction percentages of the maternal sample.

Detailed Description

The present invention provides a method of screening for an autosomal recessive disorder, such as CF, in the fetus based on counting mutation sites in maternal plasma or serum DNA, particularly in the situation where two different disorder causing mutations (alleles) are inherited by the fetus, one being of maternal origin and the other being of paternal origin. The invention also provides a method of screening for an autosomal recessive disorder, such as CF, based on counting mutation sites regardless of their maternal or paternal origin. An advantage of the invention is that it does not require parental carrier testing. Also, it achieves the same detection rate as currently achievable with existing methods, but with a much reduced false-positive rate and therefore a corresponding reduction in the number of diagnostic invasive tests that would be required.

The method of the invention may be used to screen for any autosomal recessive disorder in the fetus. For example, the method of the invention may be used to screen for cystic fibrosis (CF), sickle-cell disease, Tay-Sachs disease, autosomal recessive polycystic kidney disease (ARPKD) or phenylketonuria (PKU). The method of the invention is described herein with reference to screening for CF and CF mutations, but the method could be carried out for any other autosomal recessive disorder by simply substituting the CF mutation(s) with mutation(s) that are linked with or indicative of the autosomal recessive disorder in question. However preferably, the method is used to screen for CF.

Plasma or serum DNA consists of relatively short fragments predominantly of lOObp - 200bp. In the method of the invention, primer pairs may be designed that hybridise with target sites, for example no further apart than about 150bp, in the gene that is relevant to the disorder

(such as the CFTR gene in the case of CF) for polymerase chain reaction amplification (PCR) of short DNA regions known to include mutations that cause the disorder or are otherwise indicative of the disorder (referred to herein as autosomal recessive mutations, or more specifically as CF mutations in the case of CF). Amplification is followed by DNA sequencing of the amplified products, and the number of DNA fragments with and without a mutation are counted. CF is a recessive genetic disorder. Thus, an affected pregnancy is a pregnancy with a fetus that has a CF mutation on each of the pair of chromosomes 7; other pregnancies, including those with fetuses that are CF carriers, are designated unaffected. The same inheritance considerations apply to other autosomal recessive disorders.

As described in more detail in the Example, using computer modelling we estimated the number of maternal plasma DNA fragments to be sequenced and counted to distinguish affected and unaffected pregnancies, and hence estimate screening performance. We found that, if 3% or more of maternal plasma DNA fragments are of fetal origin (fetal fraction), sequencing and counting 32,000 to 64,000 f agments that include a CF mutation site clearly separates the distributions in affected and unaffected pregnancies. For the 13 most common CF mutations, accounting for 81.5% of CF mutations in the population, 416,000 to 832,000 fragments (13 x 32,000 to 64,000) need to be counted, and with a cut-off of 51% of mutant CF fragments per site, the estimated screening detection rate (sensitivity) is 66%, the false- positive rate <0.002%, and the odds of being affected given a positive result is 16: 1 or more.

Knowledge of the number of DNA fragments that need to be sequenced and/or counted to distinguish affected and unaffected pregnancies is useful because it avoids the cost of sequencing more fragments than is necessary, thereby saving sequencing costs. The number of DNA fragments to be sequenced and/or counted forms a part of the method of the invention.

Thus, a method of the invention is a non-invasive prenatal in vitro screening method for an autosomal recessive disorder, such as CF, in a fetus using cell-free DNA (cfDNA) from a maternal plasma or serum sample. The method comprises:

(a) performing targeted amplification of one or more DNA regions of the cfDNA, wherein the or each DNA region comprises at least one site where an autosomal recessive mutation may occur (a "mutation site"), wherein the mutation is one which, when present homozygously in an individual, causes the disorder;

(b) performing DNA sequence analysis of a number of DNA molecules ("n") that are the product of amplification for the or each DNA region and determining the number of those DNA molecules that comprise and/or do not comprise a mutation at the or each mutation site, wherein n is:

(i) more than or equal to N x 32,000; or

(ii) less than or equal to N x 640,000; or

(iii) more than or equal to N x 32,000 but less than or equal to N x 640,000;

and wherein N is the number of different mutation sites used in the method;

(c) for a mutation site, calculating the ratio of the number of sequenced DNA molecules comprising the mutation to the number of sequenced DNA molecules that do not comprise the mutation (the "ratio"), or calculating the ratio of the number of sequenced DNA molecules that do not comprise the mutation to the number of sequenced DNA molecules comprising the mutation (the "inverse ratio"); and

(d) for the mutation site, comparing the calculated ratio or inverse ratio to a predetermined cut-off value and thereby determining whether the sample is test-positive for a fetus affected with the disorder, wherein the sample is test-positive for a fetus affected with the disorder if the calculated ratio exceeds a predetermined cut-off value or if the inverse ratio is less than the predetermined cut-off value.

A preferred method of the invention comprises:

(a) obtaining cfDNA derived from a maternal plasma or serum sample;

(b) performing targeted amplification of one or more DNA regions of the cfDNA, wherein the or each DNA region comprises at least one site where an autosomal recessive mutation may occur (a "mutation site"), wherein the mutation is one which, when present homozygously in an individual, causes the disorder; (c) optionally performing targeted amplification of one or more further predetermined DNA regions of the cfDNA, wherein the or each DNA region comprises one or more mutation sites that are different to the mutation site in step (b);

(d) performing DNA sequence analysis of a number of DNA molecules ("n") that are the product of amplification for the or each DNA region and determining the number of those DNA molecules that comprise and/or do not comprise a mutation at the or each mutation site, wherein n is:

(i) more than or equal to N x 32,000; or

(ii) less than or equal to N x 640,000; or

(iii) more than or equal to N x 32,000 but less than or equal to N x 640,000; and wherein N is the number of different mutation sites used in the method;

(e) for a mutation site, calculating the ratio of the number of sequenced DNA

molecules comprising the mutation to the number of sequenced DNA molecules that do not comprise the mutation (the "ratio"), or calculating the ratio of the number of sequenced DNA molecules that do not comprise the mutation to the number of sequenced DNA molecules comprising the mutation (the "inverse ratio"); and

(f) for the mutation site, comparing the calculated ratio or inverse ratio to a

predetermined cut-off value and thereby determining whether the sample is test- positive for a fetus affected with the disorder, wherein the sample is test-positive for a fetus affected with the disorder if the calculated ratio exceeds a predetermined cut-off value or if the inverse ratio is less than the predetermined cut-off value.

The autosomal recessive disorder being screened may, for example, be CF. Thus, in that case, the method may comprise:

(a) performing targeted amplification of one or more DNA regions of the cfDNA, wherein the or each DNA region comprises at least one site where a CF mutation may occur (a "CF mutation site");

(b) performing DNA sequence analysis of a number of DNA molecules ("n") that are the product of amplification for the or each DNA region and determining the number of those DNA molecules that comprise and/or do not comprise a CF mutation at the or each CF mutation site, wherein n is:

(i) more than or equal to N x 32,000; or

(ii) less than or equal to N x 640,000; or

(iii) more than or equal to N x 32,000 but less than or equal to N x 640,000; and wherein N is the number of different CF mutation sites used in the method;

(c) for a CF mutation site, calculating the ratio of the number of sequenced DNA

molecules comprising the CF mutation to the number of sequenced DNA molecules that do not comprise the CF mutation (the "ratio"), or calculating the ratio of the number of sequenced DNA molecules that do not comprise the CF mutation to the number of sequenced DNA molecules comprising the CF mutation (the "inverse ratio"); and

(d) for the CF mutation site, comparing the calculated ratio or inverse ratio to a

predetermined cut-off value and thereby determining whether the sample is test- positive for a fetus affected with CF, wherein the sample is test-positive for a fetus affected with CF if the calculated ratio exceeds a predetermined cut-off value or if the inverse ratio is less than the predetermined cut-off value.

A preferred method of the invention comprises:

(a) obtaining cfDNA derived from a maternal plasma or serum sample;

(b) performing targeted amplification of a predetermined DNA region of the cfDNA, wherein the DNA region comprises at least one site where a CF mutation may occur (a "CF mutation site");

(c) optionally performing targeted amplification of one or more further predetermined DNA regions of the cfDNA, wherein the or each DNA region comprises one or more CF mutation sites that are different to the CF mutation site in step (b);

(d) performing DNA sequence analysis of a number of DNA molecules ("n") that are the product of amplification for the or each DNA region and determining the number of those DNA molecules that comprise and/or do not comprise a CF mutation at the or each CF mutation site, wherein n is:

(i) more than or equal to N x 32,000; or

(ii) less than or equal to N x 640,000; or

(iii) more than or equal to N x 32,000 but less than or equal to N x 640,000; and wherein N is the number of different CF mutation sites used in the method;

(e) for a CF mutation site, calculating the ratio of the number of sequenced DNA molecules comprising the CF mutation to the number of sequenced DNA molecules that do not comprise the CF mutation (the "ratio"), or calculating the ratio of the number of sequenced DNA molecules that do not comprise the CF mutation to the number of sequenced DNA molecules comprising the CF mutation (the "inverse ratio"); and (f) for the CF mutation site, comparing the calculated ratio or inverse ratio to a predetermined cut-off value and thereby determining whether the sample is test- positive for a fetus affected with CF, wherein the sample is test-positive for a fetus affected with CF if the calculated ratio exceeds a predetermined cut-off value or if the inverse ratio is less than the predetermined cut-off value.

As indicated above, an important aspect to the invention is that it provides a method of screening for an autosomal recessive disorder in the fetus based on counting mutation sites in maternal plasma or serum DNA in the situation where two different mutations (alleles) are inherited by the fetus, one being of maternal origin and the other being of paternal origin. In the case of pregnancies for which two different CF mutations are found in maternal plasma or serum, it is necessary to calculate the ratio (or inverse ratio) and to only compare the ratio (or inverse ratio) to a predetermined cut-off value, for the predominant CF mutation alone, i.e. the maternally-derived mutation.

Thus, in the case of two different CF mutations, one of which is a predominant (maternally- derived) CF mutation and the other is a paternally-derived mutation, the method especially comprises:

(a) performing targeted amplification of one or more DNA regions of the cfDNA, wherein the or each DNA region comprises at least one site where a CF mutation may occur (a "CF mutation site");

(b) performing DNA sequence analysis of a number of DNA molecules ("n") that are the product of amplification for the or each DNA region and determining the number of those DNA molecules that comprise and/or do not comprise a CF mutation at the or each CF mutation site, wherein n is:

(i) more than or equal to N x 32,000; or

(ii) less than or equal to N x 640,000; or

(iii) more than or equal to N x 32,000 but less than or equal to N x 640,000;

and wherein N is the number of different CF mutation sites used in the method;

(c) for a CF mutation site, calculating the ratio of the number of sequenced DNA

molecules comprising the predominant (i.e. maternal) CF mutation to the number of sequenced DNA molecules that do not comprise the predominant CF mutation (the "ratio"), or calculating the ratio of the number of sequenced DNA molecules that do not comprise the predominant (i.e. maternal) CF mutation to the number of sequenced DNA molecules comprising the predominant CF mutation (the "inverse ratio"); and (d) for the CF mutation site, comparing the calculated ratio or inverse ratio to a predetermined cut-off value and thereby determining whether the sample is test- positive for a fetus affected with CF, wherein the sample is test-positive for a fetus affected with CF if the calculated ratio exceeds a predetermined cut-off value or if the inverse ratio is less than the predetermined cut-off value.

Thus, the invention provides a non-invasive prenatal in vitro screening method for cystic fibrosis (CF) in a fetus using cell-free DNA (cfDNA) from a maternal plasma or serum sample, comprising:

(a) performing targeted amplification of one or more DNA regions of the cfDNA, wherein the or each DNA region comprises at least one site where a CF mutation may occur (a "CF mutation site");

(b) performing DNA sequence analysis of a number of DNA molecules ("n") that are the product of amplification for the or each DNA region and determining the number of those DNA molecules that comprise and/or do not comprise a CF mutation at the or each CF mutation site, wherein n is:

(1) more than or equal to N x 32,000; or

(ii) less than or equal to N x 640,000; or

(iii) more than or equal to N x 32,000 but less than or equal to N x 640,000;

and wherein N is the number of different CF mutation sites used in the method;

(c) (1) when one or no CF mutations are found, calculating the ratio of the number of sequenced DNA molecules comprising the CF mutation to the number of sequenced DNA molecules that do not comprise the CF mutation (the "ratio"), or calculating the ratio of the number of sequenced DNA molecules that do not comprise the CF mutation to the number of sequenced DNA molecules comprising the CF mutation

(the "inverse ratio"); or

(2) when two different CF mutations are found, calculating the ratio of the number of sequenced DNA molecules comprising the predominant (i.e. the maternal) CF mutation to the number of sequenced DNA molecules that do not comprise the predominant CF mutation (the "ratio"), or calculating the ratio of the number of sequenced DNA molecules that do not comprise the predominant CF mutation to the number of sequenced DNA molecules comprising the predominant CF mutation (the "inverse ratio"); and

(d) comparing the calculated ratio or inverse ratio to a predetermined cut-off value and thereby determining whether the sample is test-positive for a fetus affected with CF, wherein the sample is test-positive for a fetus affected with CF if the calculated ratio exceeds a predetermined cut-off value or if the inverse ratio is less than the predetermined cut-off value.

Also provided is a non-invasive prenatal screening method for an autosomal recessive disorder in a fetus using cell-free DNA (cfDNA) from a maternal plasma or serum sample comprising:

(a) determining that two different mutant DNA sequences which, when present as a compound heterozygote in an individual cause the disorder, are present in the maternal plasma or serum;

(b) for the predominant (maternally-derived) mutation, performing DNA sequence analysis of a number of DNA molecules that are the product of amplification of the DNA region comprising the mutation (the "mutation site");

(c) calculating the ratio of the number of sequenced DNA molecules comprising the predominant (maternally-derived) mutation to the number of sequenced DNA molecules that do not comprise the predominant mutation at that mutation site (the "ratio"), or calculating the ratio of the number of sequenced DNA molecules that do not comprise the predominant mutation to the number of sequenced DNA molecules comprising the predominant mutation (the "inverse ratio"); and

(d) comparing the calculated ratio or inverse ratio to a predetermined cut-off value and thereby determining whether the sample is test-positive for a fetus affected with the disorder.

Step (a) of the method may comprise (i) performing targeted amplification of two or more DNA regions of the cfDNA, wherein the or each DNA region comprises at least one site where a mutation causing the disorder may occur (a "mutation site"); and/or (ii) performing DNA sequence analysis of the amplified DNA regions to determine if two or more DNA sites are mutated in the amplified regions. Thus, a method of the invention comprises:

(a) performing targeted amplification of two or more DNA regions of the cfDNA, wherein the or each DNA region comprises at least one site where a mutation causing the disorder may occur (a "mutation site");

(b) performing DNA sequence analysis of the amplified DNA regions to determine if two or more DNA sites are mutated in the amplified regions and determining that two different mutant DNA sequences which, when present as a compound heterozygote in an individual cause the disorder, are present in the maternal plasma or serum;

(c) for the predominant (maternally-derived) mutation, performing DNA sequence analysis of a number of DNA molecules that are the product of amplification of the DNA region comprising the mutation (the "mutation site");

(d) calculating the ratio of the number of sequenced DNA molecules comprising the predominant (maternally-derived) mutation to the number of sequenced DNA molecules that do not comprise the predominant mutation at that mutation site (the "ratio"), or calculating the ratio of the number of sequenced DNA molecules that do not comprise the predominant mutation to the number of sequenced DNA molecules comprising the predominant mutation (the "inverse ratio"); and

(e) comparing the calculated ratio or inverse ratio to a predetermined cut-off value and thereby determining whether the sample is test-positive for a fetus affected with the disorder.

The disorder may be cystic fibrosis. Thus, the method may comprise:

(a) determining that two different mutant DNA sequences which, when present as a compound heterozygote in an individual cause CF, are present in the maternal plasma or serum;

(b) for the predominant (maternally-derived) CF mutation, performing DNA sequence analysis of a number of DNA molecules that are the product of amplification of the DNA region comprising the CF mutation (the "CF mutation site");

(c) calculating the ratio of the number of sequenced DNA molecules comprising the predominant (the maternally-derived) CF mutation to the number of sequenced DNA molecules that do not comprise the predominant CF mutation at that CF mutation site (the "ratio"), or calculating the ratio of the number of sequenced DNA molecules that do not comprise the predominant CF mutation to the number of sequenced DNA molecules comprising the predominant CF mutation (the "inverse ratio"); and

(d) comparing the calculated ratio or inverse ratio to a predetermined cut-off value and thereby determining whether the sample is test-positive for a fetus affected with CF.

Preferably, the method comprises: (a) performing targeted amplification of two or more DNA regions of the cfDNA, wherein the or each DNA region comprises at least one site where a CF mutation causing the disorder may occur (a "CF mutation site");

(b) performing DNA sequence analysis of the amplified DNA regions to determine if two or more DNA sites are mutated in the amplified regions and determining that two different mutant DNA sequences which, when present as a compound heterozygote in an individual cause CF, are present in the maternal plasma or serum;

(c) for the predominant (maternally-derived) mutation, performing DNA sequence analysis of a number of DNA molecules that are the product of amplification of the DNA region comprising the mutation (the "CF mutation site");

(d) calculating the ratio of the number of sequenced DNA molecules comprising the predominant (the maternally-derived) CF mutation to the number of sequenced DNA molecules that do not comprise the predominant CF mutation at that mutation site (the "ratio"), or calculating the ratio of the number of sequenced

DNA molecules that do not comprise the predominant CF mutation to the number of sequenced DNA molecules comprising the predominant CF mutation (the "inverse ratio"); and

(e) comparing the calculated ratio or inverse ratio to a predetermined cut-off value and thereby determining whether the sample is test-positive for a fetus affected with the disorder.

The method of the invention has certain advantages over previously described methods of screening for autosomal recessive disorders, such as CF, in a fetus, including that it is simpler than previously described methods. For example, the method of the invention preferably does not comprise:

(i) maternal and/or paternal mutation (e.g CF mutation) carrier testing; and/or

(ii) a step of deducing whether a mutant allele (e.g. a CF mutant allele) is inherited from the mother or the father; and/or

(iii) a step of enriching the fetal DNA fraction of the maternal sample; and/or

(iv) bias towards counting DNA molecules derived from the fetus, for example by digital Nucleic Acid Size Selection (NASS); and/or

(v) analysing a mutation site outside of the relevant gene (such as outside of the CFTR gene in the case of screening for CF) locus; and/or (vi) haplotype analysis; and/or

(vii) relative haplotype dosage analysis (RHDA), for example using Single Nucleotide Polymorphisms (SNPs). A method of the invention preferably does not comprise any one or more of (i) to (vii) above, and more preferably does not comprise any of (i) to (vii).

Mutations The method of the invention comprises performing targeted amplification and DNA sequence analysis to screen the sample for one or more potential mutations, such as CF mutations in the case of screening for CF. The method comprises performing targeted amplification of one or more DNA regions of the cfDNA, wherein the or each DNA region comprises at least one "mutation site" and then counting the number of DNA molecules that comprise the mutation. An "autosomal recessive mutation site" is a site in the DNA where an autosomal recessive mutation may occur. More specifically, it is a site in the DNA where a mutation may occur to cause a mutant allele that, in homozygous form, results in CF. Alternatively, inheritance of two different autosomal recessive mutations (i.e. at different sites within the relevant gene) may result in the autosomal recessive disorder (a 'compound heterozygote'). In the case of CF, a "CF mutation site" is a site in the DNA where a CF mutation may occur. More specifically, it is a site in the DNA where a CF mutation may occur to cause a mutant allele that, in homozygous form, results in CF. Alternatively, inheritance of two different CF mutations (i.e. at different sites within the CFTR gene) may result in CF (a 'compound heterozygote').

Thus, the method of the invention comprises screening the sample for an autosomal recessive mutation or an autosomal recessive mutant allele at a autosomal recessive mutation site. More specifically, in the case of screening for CF, the method comprises screening the sample for a CF mutation or a CF mutant allele at a CF mutation site. An autosomal recessive mutation site (e.g. CF mutation site) may be in the normal (non-mutant) form or in the mutant form and the method of the invention involves determining the relative amount of the two forms in the maternal plasma or serum sample.

A large number of CF mutations have been documented ¹ . The invention may be applied to any suitable CF mutation. For example, a CF mutation used in the invention may be any mutation known in the literature or in a CF mutation database (for example at www.genet.sickkids.on.ca/cftr/). The CF mutation site is preferably in the CFTR gene. In one preferred instance, the CF mutation screened for at a mutation site is a point mutation, particularly a point mutation which is responsible for CF. In one instance, the mutation may therefore be a base substitution, for instance the mutation may be a single base mutation, preferably a single base substitution. In another instance, the mutation may be a deletion, for example a deletion of under 100 bp, under 75 bp, under 50 bp, under 40 bp, under 30 bp, under 20 bp, under 10, 9, 8, 7 or 6 bases in size. The deletion may be, in some instances, 5, 4, 3, 2, or 1 bases in size and in one preferred instance is a single base deletion. The mutation may also be an inversion, for instance of any of the sizes specified. The mutation may be a duplication, for instance, a duplication of any such lengths. The mutation may, for instance, in some cases (a) bring about an amino acid change, (b) result in a stop codon causing premature translation, or (c) result in a change in RNA splicing. In a preferred instance, the mutation may be that responsible for the disorder, such as responsible for CF. In other instances, the mutation or polymorphism may be one closely linked to the disease causing mutation, for instance in linkage disequilibrium with the disease causing mutation. In one instance, the genetic marker being used is a single nucleotide polymorphism (SNP) associated with the disorder. Any suitable SNP known to be associated with the disease may be employed. Therefore the SNP may be in linkage disequilibrium with a disease causing mutation, but the SNP might not be a disease causing mutation itself.

It would be most effective to use one or more mutations sites that are commonly mutated in people affected with the disorder in the population. Examples of common CF mutation sites are those in Table 1. The CF mutation screened for in the method of the invention may be any of the CF mutations in Table 1. Thus, the CF mutation site may be selected from AF508, G542X, G551D, 3120 + 1 G>A, W1282X, N1303K, R553X, 621 + 1 G>T, 1717 - 1G>A, 3849 + 10Kb OT, 1898 + 1 G>T, ΔΙ507 and 2789 + 5 G>A. In a preferred instance, the CF mutation site is one or more of the most common CF mutation sites, for example one or more of the first six mutation sites in Table 1. Thus, in a particularly preferred instance, the CF mutation site is selected from AF508, G542X, G551D, 3120 + 1 G>A, W1282X and N1303K. The CF mutation site AF508 is particularly preferred.

Any number of different mutation sites may be used in the method of the invention.

However, the greater the number of different mutation sites used, the more likely the method would be an effective screening method across the general population. The method of the invention comprises performing targeted amplification of one or more DNA regions of the cfDNA, wherein the one or each DNA region comprises at least one mutation site. The number of mutation sites used in the test may be any number but more than one mutation is preferred, particularly as the mother and father may carry different mutations In step (a) of the method, the method preferably comprises performing targeted amplification of a DNA region of the cfDNA, wherein the DNA region comprises at least one site where a first mutation may occur (a "first mutation site"); and performing targeted amplification of one or more further DNA regions of the cfDNA, wherein the or each further DNA region comprises one or more mutation sites that are different to the first mutation site.

The method preferably comprises screening more than one mutation site, i.e. the method of the invention preferably comprises using a panel of mutation sites. Thus, the number of mutation sites (N), such as CF mutation sites, is preferably 2 or more, 3 or more, 4 or more, 5 or more, 6 or more, 7 or more, 8 or more, 9 or more, 10 or more, 11 or more, 12 or more, or even more preferably 13 or more. In a particularly preferred instance, the method comprises screening for multiple CF mutation sites including one or more of the mutation sites in Table 1. For example, the method preferably comprises screening for AF508 in combination with 1 or more, 2 or more, 3 or more, 4 or more, 5 or more, 6 or more, 7 or more, 8 or more, 9 or more, 10 or more, 11 or more, or 12 or more other CF mutation sites, such as any of the mutation sites in Table 1. Typically, N is 2 to 13, such as 4 to 10 or 6 to 8. In a particularly preferred instance, the method comprises screening 13 CF mutation sites, such as at all of the CF mutation sites in Table 1, i.e. AF508, G542X, G551D, 3120 + 1 G>A, W1282X, N1303K, R553X, 621 + 1 G>T, 1717 - 1G>A, 3849 + 10Kb OT, 1898 + 1 G>T, ΔΙ507 and 2789 + 5 G>A. In a further preferred instance, the method comprises screening for a 23 mutation panel, such as the 23 mutation panel described in reference 36 cited herein.

DNA sequencing analysis may reveal that there is one mutation, i.e. one mutant allele in the maternal plasma or serum sample. Alternatively, it may reveal that there are no mutations at the mutation site. Another scenario is that the analysis may reveal that there are two different mutations, i.e. two different mutant alleles, at different mutation sites in the maternal plasma or serum sample. In this latter scenario, one of the mutant alleles is derived from the mother and the other is derived from the father. Thus, when the maternally-derived and paternally-derived autosomal recessive mutations are different, two different mutations will be found amongst the DNA fragments in the maternal plasma or serum sample, for example by DNA sequencing analysis.

Specifically, it may be determined that two different mutant DNA sequences which, when present as a compound heterozygote in an individual cause the disorder, are present in the maternal plasma or serum. In such a situation, as discussed elsewhere herein, it is only necessary to precisely quantify the predominant mutation in the plasma or serum sample, as illustrated by Figure 2. The predominant mutation always comes from the mother because the father does not make a direct contribution to maternal plasma or serum DNA. Thus, in the method of the invention, when two different mutations are found in maternal plasma or serum, the method comprises quantifying the predominant (i.e. the maternally- derived) mutation. The method may comprise quantifying the predominant (i.e. the maternally-derived) mutation only. Thus, in a preferred method of the invention,

maternally-derived and paternally-derived mutant alleles for the mutation are different.

In the case of screening for CF, the maternally-derived and paternally-derived mutant alleles for the CF mutation are different.

The method of the invention is, however, particularly useful in that it is not necessary for the paternally-derived mutant allele to be different from the maternally-derived mutant allele for the method to be effective. The method can be performed regardless of whether the maternally-derived and paternally derived mutant alleles at the autosomal recessive mutation (e.g. CF mutation) site are identical or whether they differ. Thus, in the method of the invention, the maternally-derived and paternally-derived mutant alleles for the autosomal recessive mutation (e.g. CF mutation) can be the same or different. If they are the same, i.e. in the case of a single mutation being present, the method may comprise determining that one mutant DNA sequence, which when present homozygously in an individual causes the disorder, is present in the maternal plasma or serum sample. If they are different, i.e. in the case of two different mutations being present, the method may comprise determining that two different mutant DNA sequences, which when present as a compound heterozygote in an individual cause the disorder, are present in the maternal plasma or serum sample. In a preferred instance, the maternally-derived and paternally-derived mutant alleles for the autosomal recessive mutation (e.g. CF mutation) are different. Sample

The method comprises analysis of cell-free DNA (cfDNA) molecules in blood, plasma or serum samples obtained from maternal blood of a pregnant female during gestation of the fetus. The pregnant female will commonly be a human subject. However the invention is also applicable to other mammalian subjects and could be applied in a veterinary or agricultural setting.

The method may comprise the step of obtaining the sample of blood, plasma or serum from the pregnant female. The method is preferably performed on DNA collected using a noninvasive approach. The sample may be obtained by routine blood draw from the pregnant female. However, more commonly, the sample will have already been obtained from the pregnant female and sent for testing. The sample may have been obtained at any stage after about 8 weeks of gestation of the fetus. However, it can be helpful to know the outcome of the test as early on in pregnancy as possible. Thus, the sample may have been obtained from a pregnant female in the first, second or third trimester, but more typically the first or second trimester. The sample may have been obtained at 8 weeks of gestation or later, 15 weeks of gestation or later, 25 weeks of gestation or later, or 30 weeks of gestation or later. The volume of sample obtained is any suitable volume. The volume is typically 1 to 20 ml, such as 1 to 5 ml, 5 to 15 ml or 10 to 20 ml. The volume is preferably at least 14 ml, such as 14 to 20 ml.

For convenience, the same maternal blood sample collected for prenatal Down syndrome screening (for example at 11-13 weeks' gestation) can be used. In this scenario, cfDNA is prepared from the plasma, some of which is used for Down syndrome screening, and some for CF screening. Thus, in one aspect of the invention, the sample is also screened for Down syndrome. The method may comprise the step of obtaining, providing or isolating a sample of cfDNA from blood, serum or plasma from a pregnant female. The blood, serum or plasma sample obtained from the female comprises both maternal cfDNA and cfDNA originating from the fetus. Nucleic acid can be routinely derived, obtained or extracted from the sample. Methods for processing samples containing nucleic acids, extracting nucleic acids and/or purifying nucleic acids for use in detection methods are well-known in the art. Total nucleic acid may be isolated or DNA and RNA may be isolated separately.

In one embodiment of the method, maternal and/or paternal DNA from any suitable source such as blood or sputum may be analysed by methods known in the art to determine if one or both parents are carriers of a mutation, prior to performing the fetal screening method of the invention.

Typically, a sample is processed in an appropriate manner such that nucleic acid is provided in a convenient form for performing amplification. The cfDNA will typically need to be denatured to convert the double-stranded DNA into single-stranded form. Thus, the method may comprise a step of denaturing the cell-free DNA to single strands before the

amplification step. Methods of denaturing double-stranded DNA are well known in the art. The cfDNA obtained from blood, plasma or serum may be of any length, but will typically be at least 20, 25, 30, 40, 50, 75, 100 or 150 nucleotides in length. Typically it will be in the range of 50 to 400, 80 to 350, or 100 to 200 nucleotides in length.

In one instance, the sample is treated quickly to help avoid DNA degradation and in particular cfDNA degradation. Hence, in a preferred instance, the sample is processed ready for analysis within 24 hours of being obtained, for instance, within 20 hours, 18 hours, 16 hours, 12 hours, 8 hours, 7 hours, 6 hours, 5 hours, 4 hours, 3 hours, 2 hours or 1 hour. In one particularly preferred instance, the sample is processed within about 7 hours and in particular within about 6 hours. It may be that the sample is not immediately analyzed, but is stored under conditions that prevent or minimize further degradation of the sample, for instance by freezing the sample, such as at -80°C.

Alternatively, or additionally, samples may be collected in a way that helps prevent or reduce nucleic acid degradation and in particular degradation of cell free nucleic acids. The skilled person will be aware of steps or procedures that may be taken. For instance, blood samples may be collected in receptacles that help prevent nucleic acid degradation and in particular degradation of cell free nucleic acids in samples. Such vessels may comprise a preservative that inhibits nucleic acid degradation, an inhibitor of metabolism and/or an inhibitor of an enzyme responsible for nucleic acid degradation. In a particularly preferred instance, the DNA sample is analyzed to confirm the presence of fetal DNA. This may be carried out using any suitable technique known the skilled person, for example by analysis of the presence of a fetal marker. Analysis for the presence of a fetal marker to confirm the presence of fetal DNA may be performed separately or as part of the method of the invention. Any suitable marker may be analyzed to confirm the presence of fetal DNA in the sample, with one instance of a possible marker being the RASSF1 A marker. In one preferred instance, the marker employed to detect fetal DNA is any suitable single nucleotide polymorphisms (SNPs) that the fetus has inherited from the father which is not present in the mother. In the case of a male fetus the Y chromosome can be used, as maternal DNA will lack any Y chromosome sequences, so the presence of any Y chromosome sequence can be used to confirm the presence of fetal DNA and/or to quantify the amount of fetal DNA. In other instances polymorphic short tandem repeats, SNPs and/or indel markers may be employed, such as a panel of any of those markers. In a further preferred instance, the amount of DNA present in the sample may be quantified. Again, the skilled person would be able to perform this aspect using any known suitable technique. The method may quantify the amount of fetal DNA present. For instance, the amount of DNA present may be quantified by PCR, for example by real time PCR or any other suitable technique, and such an approach may be used to quantify the amount of fetal DNA, for example using any of the fetal DNA markers discussed herein. In one instance, the method employed to quantify the amount of fetal DNA may be that described in Nygren AO et ah, "Quantification of fetal DNA by use of methylation-based DNA discrimination." Clin Chem 2010;56:1627-35; or that described in Struble CA, et al., "Fetal fraction estimate in twin pregnancies using directed cell-free DNA analysis." Fetal Diagn Ther 2014;35:199-203 (the entirety of both are incorporated by reference, including the specific methods described).

In a preferred instance, the percentage of maternal plasma DNA fragments that is of fetal origin in the sample (the fetal fraction of the sample) is at least 2%, such as at least 3%, or at least 4%, or more preferably at least 10%. The method of the invention preferably comprises determining that the fetal fraction is at least 2%, or at least 3%, or at least 4%, or at least 10%. In a particularly preferred instance, the method comprises determining that the fetal fraction is at least 2%. Targeted Amplification

A targeted amplification method is used to amplify the DNA region comprising the autosomal recessive mutation site for subsequent sequencing. Suitable techniques for targeted amplification that may be used in the present invention are well known in the art, including target capture by hybridisation. ¹⁵ ' ^{16, 17} ' ¹⁸

In a particularly preferred instance, the amplification method employed in the present invention is PCR. Any suitable primer pairs may be employed in the invention. In one instance, a DNA region (or "amplicon") comprising a mutation site is chosen to be amplified and primer pairs designed accordingly. Alternatively, suitable primer pairs may already be known or available. In a particularly preferred instance, the amplification will result in a population of amplification products that is representative of the starting template DNA, in particular in reference to the incidence of maternal and fetal template molecules within the original sample.

The amplicon will comprise the autosomal recessive mutation site or sites being assessed, for example CF mutation site(s). By sequencing the PCR product it is then typically possible to determine the representation of particular alleles for the region in the PCR product and so in the original sample. By assessing the number of PCR products with the mutation compared with the number of PCR products without the mutation, it is typically possible to determine whether the sample is test-positive for a fetus affected with the disorder, such as CF, or not.

As described above, it is possible to screen any number of mutation sites in the method of the invention and the method preferably comprises screening multiple mutation sites. Thus, more than one amplicon may be chosen with different amplicons encompassing different mutation sites. An amplicon may encompass more than one mutation site. Also, in a preferred instance, more than one amplicon will be amplified for a mutation site. For instance, there may be a plurality of amplicons and in particular overlapping amplicons where each amplicon comprises the mutation site. By adopting such a strategy the different amplicons can be used to confirm further the result given by the other amplicons. In one instance, 2, 3, 4, 5, 6 or 7 amplicons may be amplified or at least such numbers. In a preferred instance, 3, 4 or 5 amplicons may be amplified and in particular 3 amplicons, for a mutation site. In the case of screening multiple different mutation sites, it may be that for each mutation site there are overlapping amplicons. In a preferred instance, it may be that the amplicon is under 200 bp in length, for instance, under 180 bp, 160 bp, or 150 bp in length, in some cases under 140 bp, 130 bp, 125 bp or 120 bp in length. It may be that the amplicon is under 110 bp, 100 bp, or 90 bp in length. In some cases all the amplicons will be under such length.

In a particularly preferred instance, more than one primer pair may be employed for each amplicon. For instance, a first primer pair may be used to perform an initial amplification of the amplicon, then a second primer pair may be employed to amplify the amplicon further. Such a two step approach may help ensure that the allelic frequency is as close as possible in the end PCR product to that in the original template. It may be that the second primer pair employed in the second amplification introduces additional sequences to those amplified from the template. For instance, the second primer set may introduce additional sequences to help facilitate analysis of the sequence data obtained.

In a further preferred aspect, at least one of the primers in each pair comprises a "bar code" or index sequence allowing identification of a specific amplicon. By providing the same primer pair, but with different "bar code" sequences present, it is possible to allocate a particular bar code to a particular maternal subject and then analyze the samples from the different maternal subjects simultaneously, because the results for each maternal subject are denoted by a particular bar code. It is also possible to have a bar code where one part of the sequence is unique to the amplicon, the other denoting which maternal subject the bar code comes from. Such an approach may allow pooling of samples for sequencing. Any suitable PCR method known in the art can be used in the method of the invention. The skilled person can determine a suitable protocol and reagents to use. For example, any suitable polymerase may be used in the amplification, preferably the polymerase is one with a low error rate. Any suitable cycling conditions may be employed in the PCR and the conditions may be tailored to the specific primers employed.

Amplifying targeted DNA regions of interest by conventional PCR can suffer from the problem that the ratio of mutant to non-mutant DNA sequences in maternal plasma will vary between samples as a result of the PCR. However, the use of barcodes as unique molecular identifiers of plasma DNA fragments can adjust for this and avoid the problem. ^31,32,33 Thus, in a preferred aspect of the invention, plasma DNA fragments are first labelled with barcoded adaptors (unique molecular identifiers) so that each fragment has a unique identifying sequence. This can be carried out before targeted amplification of the labelled fragments. Methods for labelling DNA fragments with barcodes (unique molecular identifiers) are well known in the art to the skilled person and are described in references 31 to 33 cited herein, for example. These methods include the introduction of exogenous sequences through PCR or ligation. The individual diversity of sequences is preserved without error ³¹ . Subsequent amplification, sequencing, counting and correction for under- and over-amplification (enabled using the barcodes/unique molecular identifiers) can then be performed. ^31>32>33 Correcting for under- and over-amplification using the barcodes would be routine for the skilled person. ^31>32 · ³³

Sequencing & Analysis The method of the invention comprises performing DNA sequence analysis of DNA molecules that are the product of amplification of a DNA region (amplicon) comprising a mutation site (such as a CF mutation site). The method comprises sequencing the products of amplification and counting the number of products with a mutation and/or the number of products without a mutation. The numbers may be corrected for under- or over-amplification by use of barcodes or unique molecular identifiers as mentioned above. The number of DNA molecules that comprise the mutation is compared to the number of DNA molecules that do not comprise the mutation, preferably by calculating a ratio and determining whether the ratio exceeds a predetermined cut-off value. The aim is to determine, with statistical precision presented herein, whether the number of DNA molecules that comprise the mutation exceeds the number of DNA molecules that do not comprise the mutation, and thereby determining whether the sample is test-positive for a fetus affected with the autosomal recessive disorder, such as CF.

In one instance, after amplification, the sample may be purified to help in subsequent analysis. The sample may be analyzed to help normalize all the samples to the same molarity. Any suitable approach though may be employed in such clean-up and normalization. Corrections may also be made for under- or over-amplification by use of barcodes or unique molecular identifiers. Samples may be pooled prior to sequencing. For instance, where primers carry an index/bar code specific for an amplicon and/or maternal subject that means that pooled samples may be sequenced together. For instance, all of the amplicons generated may be pooled together and sequenced. In some cases, samples may be pooled from 1 to 10, 2 to 8, 3 to 6 and in particular 5 maternal subject samples. In some instances, all of the amplicons from such numbers of maternal subject may be pooled prior to sequencing.

A useful aspect of the invention is the determination by the inventors of the minimum and/or maximum number of products of amplification that should be sequenced and counted for good screening performance. The method of the invention comprises performing DNA sequence analysis of a number of DNA molecules ("n") that are the product of amplification for the or each DNA region comprising a mutation site and determining the number of those DNA molecules that comprise and/or do not comprise a mutation at the mutation site. For a single autosomal recessive mutation site, such as a single CF mutation site, (i.e. N = 1), the number of DNA molecules ("n") that are sequenced is preferably more than or equal to 32,000. In one instance, n is more than or equal to 40,000, or more than or equal to 48,000, or more than or equal to 56,000. In a particularly preferred instance, n is more than or equal to 64,000, for example more than or equal to 70,000, more than or equal to 80,000, more than or equal to 90,000, or more than or equal to 100,000.

Using n is more than or equal to 32,000 or more than or equal to 64,000 is useful for a fetal fraction as low as 2%. Thus, in one instance, the method comprises determining that the fetal fraction is at least 2% and performing DNA sequencing of at least about 32,000, such as at least about 64,000 DNA molecules, such as at least about 70,000, at least about 80,000, at least about 90,000 or at least about 100,000 DNA molecules. The method may comprise determining that the fetal fraction is at least 3% or at least 4% and performing DNA sequencing of at least about 32,000, such as at least about 64,000 DNA molecules, such as at least about 70,000, 80,000, 90,000 or at least about 100,000 DNA molecules. At higher fetal fractions it would be possible to sequence a lower number of DNA molecules. For example, the method may comprise determining that the fetal fraction is at least 10% and performing DNA sequencing of at least about 8,000 DNA molecules, such as at least about 16,000, at least about 24,000, at least about 32,000, or at least about 64,000 DNA molecules. The value of n can be as low as possible to reduce sequencing costs whilst maintaining a good screening performance. Thus, n is typically less than or equal to 1 million, less than or equal to 900,000, less than or equal to 800,000 or less than or equal to 700,000. Preferably, n is less than or equal to 640,000, or more preferably lower, such as less than or equal to 600,000, less than or equal to 500,000, less than or equal to 400,000, less than or equal to 200,000, less than or equal to 150,000, less than or equal to 140,000, or less than or equal to 130,000.

In a preferred method, for one mutation site (i.e. N = 1), n is more than or equal to 32,000 but less than or equal to 640,000, i.e. n is a number in the range of 32,000 to 640,000. Preferably, n is a number in the range of 32,000 to 600,000; 32,000 to 500,000; 32,000 to 400,000;

32,000 to 200,000; 32,000 to 150,000; 32,000 to 140,000; 32,000 to 130,000; 40,000 to 640,000; 48,000 to 640,000; 56,000 to 640,000; 64,000 to 640,000; 70,000 to 640,000; 80,000 to 640,000; 90,000 to 640,000; or 100,000 to 640,000. More preferably, n is a number in the range of 40,000 to 500,000, such as 50,000 to 400,000; 60,000 to 300,000; 64,000 to 128,000. In a particularly preferred instance, n is a number in the range of 32,000 to 128,000, such as 64,000 to 128,000. This is particularly preferred for a fetal fraction of 2%. Thus, the method of the invention may comprise determining that the fetal fraction is at least 2% and performing DNA sequence analysis of n DNA molecules, wherein n is a number in the range of 32,000 to 128,000, such as 64,000 to 128,000.

As explained above, multiple mutation sites can be screened. In the method of the invention, the greater the number of mutation sites that are screened, the greater the number of DNA molecules (n) that are sequenced. In the method of the invention, the number of DNA molecules (n) that are sequenced is:

(i) more than or equal to N x 32,000; or

(ii) less than or equal to N x 640,000; or

(iii) more than or equal to N x 32,000 but less than or equal to N x 640,000;

wherein N is the number of different mutation sites used in the method. Thus, in one instance, the number of DNA molecules ("n") that are sequenced is more than or equal to N x 32,000. In one instance, n is more than or equal to N x 40,000, or more than or equal to N x 48,000, or more than or equal to N x 56,000. In a particularly preferred instance, n is more than or equal to N x 64,000, for example more than or equal to N x 70,000, more than or equal to N x 80,000, more than or equal to N x 90,000, or more than or equal to N x 100,000. Using n is more than or equal to N x 32,000, such as more than or equal to N x 64,000, is useful for a fetal fraction as low as 2%. Thus, in one instance, the method comprises determining that the fetal fraction is at least 2% and performing DNA sequencing of at least about N x 32,000, such as least about N x 64,000 DNA molecules, such as at least about N x 70,000, at least about N x 80,000, at least about N x 90,000 or at least about N x 100,000 DNA molecules. The method may comprise determining that the fetal fraction is at least 3% or at least 4% and performing DNA sequencing of at least about N x 32,000, such as at least about N x 64,000 DNA molecules, such as at least about N x 70,000, N x 80,000, N x 90,000 or at least about N x 100,000 DNA molecules. At higher fetal fractions it would be possible to sequence a lower number of DNA molecules. For example, the method may comprise determining that the fetal fraction is at least 10% and performing DNA sequencing of at least about N x 8,000 DNA molecules, such as at least about N x 16,000, at least about N x 24,000, at least about N x 32,000, or at least about N x 64,000 DNA molecules.

The value of n can be as low as possible to reduce sequencing costs whilst maintaining a good screening performance. Thus, n is typically less than or equal to N x 1 million, less than or equal to N x 900,000, less than or equal to N x 800,000 or less than or equal to N x 700,000. Preferably, n is less than or equal to N x 640,000, or more preferably lower, such as less than or equal to N x 600,000, less than or equal to N x 500,000, less than or equal to N x 400,000, less than or equal to N x 200,000, less than or equal to N x 150,000, less than or equal to N x 140,000, or less than or equal to N x 130,000.

In a preferred method, n is more than or equal to N x 32,000 but less than or equal to N x 640,000, i.e. n is a number in the range of N x 32,000 to N x 640,000. Preferably, n is a number in the range of N x 32,000 to N x 600,000; N x 32,000 to N x 500,000; N x 32,000 to N x 400,000; N x 32,000 to N x 200,000; N x 32,000 to N x 150,000; N x 32,000 to N x 140,000; N x 32,000 to N x 130,000; N x 40,000 to N x 640,000; N x 48,000 to N x 640,000; N x 56,000 to 640,000; N x 64,000 to N x 640,000; N x 70,000 to N x 640,000; N x 80,000 to N x 640,000; N x 90,000 to N x 640,000; or N x 100,000 to N x 640,000. More preferably, n is a number in the range of N x 40,000 to N x 500,000, such as N x 50,000 to N x 400,000; N x 60,000 to N x 300,000; N x 64,000 to N x 128,000. In a particularly preferred instance, n is a number in the range of N x 32,000 to N x 128,000, such as in the range N x 64,000 to N x 128,000. This is particularly preferred for a fetal fraction of 2%. Thus, the method of the invention may comprise determining that the fetal fraction is at least 2% and performing DNA sequence analysis of n DNA molecules, wherein n is a number in the range of N x 32,000 to N x 128,000, such as in the range N x 64,000 to N x 128,000.

Thus, for whatever number of mutation sites (N) used, one simply substitutes that number for N in the formulae above. For example, if the number of mutation sites (N) is 2, n is: more than or equal to 2 x 32,000, i.e. more than or equal to 64,000; or less than or equal to 2 x 640,000, i.e. less than or equal to 1,280,000; or more than or equal to 2 x 32,000 but less than or equal to 2 x 640,000; i.e.

than or equal to 64,000 but less than or equal to 1,280,000.

Thus, in one instance, if the number of mutation sites (N) is 2, the number of DNA molecules (n) sequenced and counted is preferably in the range of 2 x 32,000 to 2 x 640,000, which is the range of 64,000 to 1,280,000. More preferably, when N is 2, n is in the range of 2 x 32,000 to 2 x 128, 000, which is the range of 64,000 to 256,000. More preferably, n is in the range of 2 x 64,000 to 2 x 128,000, which is the range of 128,000 to 256,000.

In another instance, if the number of mutation sites (N) is 6, the number of DNA molecules (n) sequenced and counted is preferably in the range of 6 x 32,000 to 6 x 640,000, which is the range of 132,000 to 3,840,000. More preferably, when N is 6, n is in the range of 6 x

32,000 to 6 x 128,000, which is the range of 192,000 to 768,000. More preferably, n is in the range of 6 x 64,000 to 6 x 128,000, which is the range of 384,000 to 768,000.

In a further instance, if the number of mutation sites (N) is 13, the number of DNA molecules (n) sequenced and counted is preferably in the range of 13 x 32,000 to 13 x 640,000, which is the range of 416,000 to 8,320,000. More preferably, when N is 13, n is in the range of 13 x 32,000 to 13 x 128,000, which is the range of 416,000 to 1,664,000. More preferably, n is in the range of 13 x 64,000 to 13 x 128,000, which is the range of 832,000 to 1,664,000. Any suitable sequencing technique may be employed, particularly those that allow

simultaneous sequencing of a plurality of molecules, including those that allow high throughput and ultrahigh throughput. In one especially preferred instance, a Next Generation Sequencing (NGS) technique is employed to perform the sequencing. Examples of sequencing techniques which may be employed include sequencing by pyrosequencing, for instance the synthesis Roche/454 system; fluorescence based sequencing, such as Illumina sequencing or Intelligent Biosystems sequencing; Ion Torrent (H+ ion detection) sequencing; fluorescence-single molecule sequencing, such as Heliscope sequencing; DNA Nanoball array with CP AL sequencing (Combinatorial probe anchor ligation); complete genomics sequencing; sequencing by ligation, such as SOLiD (based on Polony); and single Molecule Real time sequencing (SMRT) such as Pac Bio sequencing.

Further examples of possible sequencing approaches which may be employed, include electron microscope sequencing - Electron Optica, ZS Genetics; sequencing by synthesis

(recording pH and or temperature), such as Genapsys sequencing; Nanopore - biological and solid state - such as the systems of Genia, IBM Roche, ONT, Nabsys, Noblegen; sequencing by hybridisation - such as the GnuBio system; Optical Imaging - such as the Lightspeed Genomics sequencing; charge detection in a Nanowire - such as the Quantum Dx sequencing; Atomic force microscopy - such as the Reveo sequencing system; and sequencing by expansion - such as the Stratos Genomics system.

In one especially preferred embodiment, the sequencing technique is one based on reversible dye-terminators and/or engineered polymerases and in particular is Illumina sequencing. In Illumina sequencing typically single molecules of DNA are attached to a flat surface, amplified in situ and used as templates for synthetic sequencing with fluorescent reversible terminator deoxyribonucleotides. Images of the surface may then be analyzed to generate a sequence. In a further preferred instance, a bench top sequencer is employed and in particular one which performs NGS and preferably one that performs Illumina sequencing. Illustrative examples of sequencers which may be employed include the HiSeq 2500/1500, HiSeq 2000/1000, the Illumina Genome Analyzer and the Illumina Miseq. Any suitable sequencer may be employed.

The sequence data obtained is then analyzed by any suitable technique. For instance, the analysis will typically allow sequences to be assigned to a particular amplicon and/or maternal subject. The analysis allows the determination of the number of DNA molecules that comprise a mutation at the mutation site and/or the number of DNA molecules that do not comprise a mutation at the mutation site. Then, for a mutation site, the method comprises determining whether the number of DNA molecules that comprise the mutation exceeds the number of DNA molecules that do not comprise the mutation and thereby determining whether the sample is test-positive for a fetus affected with the autosomal recessive disorder. This step comprises:

(a) for a mutation site, calculating the ratio of the number of sequenced DNA

molecules comprising the mutation to the number of sequenced DNA molecules that do not comprise the mutation (the "ratio"), or calculating the ratio of the number of sequenced DNA molecules that do not comprise the mutation to the number of sequenced DNA molecules comprising the mutation (the "inverse ratio"); and

(b) for the mutation site, comparing the calculated ratio or inverse ratio to a

predetermined cut-off value and thereby determining whether the sample is test- positive for a fetus affected with the disorder, wherein the sample is test-positive for a fetus affected with the disorder if the calculated ratio exceeds a

predetermined cut-off value or if the inverse ratio is less than the predetermined cut-off value.

As described herein, in the situation where two different mutations are found in maternal plasma or serum it is only necessary to quantify the predominant (i.e. maternal) mutation. The presence of a high and low abundance of mutations will be evident, with the higher abundance mutation being the maternal one. For the more abundant mutation, the sequence data obtained is analyzed by any suitable technique. For instance, the analysis will typically allow sequences to be assigned to a particular amplicon and/or maternal subject. The analysis allows the determination of the number of DNA molecules that comprise a mutation at the mutation site and/or the number of DNA molecules that do not comprise a mutation at the mutation site. Then, for a mutation site, the method comprises determining whether the number of DNA molecules that comprise the predominant mutation exceeds the number of DNA molecules that do not comprise the predominant mutation and thereby determining whether the sample is test-positive for a fetus affected with the autosomal recessive disorder. This step comprises:

(a) calculating the ratio of the number of sequenced DNA molecules comprising the predominant mutation to the number of sequenced DNA molecules that do not comprise the predominant mutation (the "ratio"), or calculating the ratio of the number of sequenced DNA molecules that do not comprise the predominant mutation to the number of sequenced DNA molecules comprising the predominant mutation (the "inverse ratio"); and

(b) comparing the calculated ratio or inverse ratio to a predetermined cut-off value and thereby determining whether the sample is test-positive for a fetus affected withthe disorder, wherein the sample is test-positive for a fetus affected with the disorder if the calculated ratio exceeds a predetermined cut-off value or if the inverse ratio is less than the predetermined cut-off value.

If the calculated ratio exceeds a predetermined cut-off value or if the inverse ratio is less than the predetermined cut-off value, the sample is test-positive for a fetus affected with the disorder. Conversely, if the calculated ratio is less than a predetermined cut-off value or if the inverse ratio exceeds the predetermined cut-off value, the sample is test-negative for a fetus affected with the disorder. An informative cut-off value can be readily chosen by a person skilled in the art. In general, for example when one or no mutation is found, the predetermined cut-off value may be 50, 51, 52, 53, 54 or 55 %, for example, but is preferably 51 %. However in the case of two different mutations, one of which is maternally-derived and the other paternally-derived, the predetermined cut-off value for the maternally derived CF mutation site in maternal plasma or serum may be 45, 46, 47, 48 49 or 50%, for example, but is preferably 48% or 49%. The cutoff may vary depending on the percentage fetal fraction of the maternal plasma or serum. Thus, the cut-off is preferably 46 to 49% with a fetal fraction of 10% or more; and/or 48 to 49% with a fetal fraction of less than 10% ; and/or 49% with a fetal fraction of 4% or less. Preferably the cut off is 46% with a fetal fraction of 10% or more, or preferably 49% with a fetal fraction of 4% or less. In general, a lower cut-off value results in the test detecting more true cases, i.e. a higher Detection Rate, but this is achieved at the expense of a higher False Positive Rate. For implementation in practice, a cut-off value is chosen on the basis of providing screening performance characteristics, notably Detection Rate and False Positive Rate (Table 2), that are judged adequate for a practical test, i.e. providing a useful Detection Rate without an unduly high False Positive Rate, thus achieving an acceptable odds of being affected given a positive test result. Thus, in one instance, a ratio above 51 % or an inverse ratio below 51 % is indicative that that sample is test-positive for a fetus affected with the disorder such as CF. Conversely, a ratio below 51 % or an inverse ratio above 51 % is indicative that that sample is test-negative for a fetus affected with the disorder such as CF. However in the case of two different mutations, one of which is maternally-derived and the other paternally-derived, a ratio above 48 % or an inverse ratio below 48 % is indicative that that sample is test-positive for a fetus affected with the disorder such as CF. Conversely, a ratio below 48 % or an inverse ratio above 48 % is indicative that that sample is test-negative for a fetus affected with the disorder such as CF. Thus, in one instance for example, where two mutations are found in the maternal plasma or serum and the fetal fraction is 4%, a ratio above 49% or an inverse ratio below 49% is indicative that that sample is test-positive for a fetus affected with the disorder. Conversely, a ratio below 49% or an inverse ratio above 49% is indicative that that sample is test-negative for a fetus affected with the disorder.

The calculation of the ratio and/or the comparison to a cut-off step may be carried out manually, but preferably is carried out using a computer and a suitable computer program. The invention also provides a computer-implemented method of non-invasive prenatal screening for an autosomal recessive disorder in a fetus using cell-free DNA (cfDNA) from a maternal plasma or serum sample, wherein two different mutant DNA sequences which, when present as a compound heterozygote in an individual cause the disorder, are present in the maternal plasma or serum, comprising:

(a) inputting to a computer system data concerning the number of DNA molecules that are the product of amplification of the DNA region comprising the predominant (maternally-derived) mutation and the number of those DNA molecules that comprise and/or do not comprise the mutation at the mutation site;

(b) determining the ratio of the number of sequenced DNA molecules comprising the predominant (maternally-derived) mutation to the number of sequenced DNA molecules that do not comprise the predominant mutation at that mutation site (the "ratio"), or calculating the ratio of the number of sequenced DNA molecules that do not comprise the predominant mutation to the number of sequenced DNA molecules comprising the predominant mutation (the "inverse ratio"); and

(c) comparing the ratio or inverse ratio to a predetermined cut-off value and

thereby determining whether the sample is test-positive for a fetus affected with the disorder.

The method of the invention in the case of a single mutation being present, i.e. where one mutant DNA sequence which, when present homozygously in an individual causes the disorder, is found in maternal plasma or serum, the invention can be similarly computerised. Thus, also provided is a computer-implemented method of non-invasive prenatal screening for an autosomal recessive disorder in a fetus using cell-free DNA (cfDNA) from a maternal plasma or serum sample, wherein a single autosomal recessive mutation causing the disorder is present, or two different autosomal recessive mutations are present, in the maternal plasma or serum, comprising:

(a) inputting to a computer system data concerning the number of DNA molecules that are the product of amplification of a DNA region comprising the mutation and the number of those DNA molecules that comprise and/or do not comprise the mutation at the mutation site;

(b) determining the ratio of the number of sequenced DNA molecules comprising the mutation to the number of sequenced DNA molecules that do not comprise the mutation at that mutation site (the "ratio"), or determining the ratio of the number of sequenced DNA molecules that do not comprise the mutation to the number of sequenced DNA molecules comprising the mutation (the "inverse ratio"); and

(c) comparing the ratio or inverse ratio to a predetermined cut-off value and thereby determining whether the sample is test-positive for a fetus affected with the disorder. Also provided is a computer program comprising program code mean that, when executed on a computer system, instruct the computer system to perform all the steps of the computer- implemented method of the invention. Further provided is a computer storage medium comprising the computer program of the invention. Also provided is a computer system arranged to perform a method of the invention.

A positive result may be obtained using any suitable mutation site and any number of different mutation sites. If using more than one mutation site, the results would be combined. In the case of two different mutations being found in the maternal plasma or serum (one maternally derived and the other paternally derived) a positive result for any two different mutation sites means that the sample is test-positive for a fetus affected with the disorder. For example, one pair of mutation sites may give a negative result and a different pair of mutation sites may give a positive result. In that example, the sample would be determined to be test- positive for a fetus affected with the disorder. The positive or negative result may then be provided, for example in the form of a report and, optionally, further advice may be given.

Example

Methods

Cell-free DNA is prepared from maternal plasma or serum. Plasma DNA consists of relatively short fragments predominantly of lOObp -200bp. Primer pairs are designed that hybridise with target sites, preferably no further apart than about 150bp, in the CFTR gene for polymerase chain reaction amplification of short DNA regions known to include the most common CF mutations. Techniques for such targeted amplification are known in the art ¹⁴ , including target capture by hybridisation. ¹⁵ ' ^16,17 ' ¹⁸ Amplification is followed by DNA sequencing of the amplified products, and the number of DNA fragments with and without a CF mutation are counted. An affected pregnancy is a pregnancy with a fetus that has a CF mutation on each of the pair of chromosomes 7; other pregnancies, including those with fetuses that are CF carriers, are designated unaffected.

Computer modelling was carried out to estimate, for any particular CF mutation site, the relative distributions of the proportion of DNA fragments with a CF mutation in affected and unaffected pregnancies using random sampling without replacement and Gaussian

distributions.

These expected percentages depend on whether one CF mutation is found in the DNA analysis or whether two different CF mutations are found. Thus:

(i) Pregnancies with either one or no CF mutations in the maternal plasma. In this situation, for example, in an affected pregnancy with a fetal fraction of 10% the mother's plasma will, in expectation, contribute 45% of DNA fragments with a CF mutation and the fetus 10%, ie a total of 55%. We then estimated the distribution of the percentages in affected and unaffected pregnancies for specific fetal fractions using Gaussian distributions with a mean m and a standard deviation m X (100 - m)/?!, where n is the number of DNA fragments in the mutation site that are sequenced. The mean of the distribution in affected pregnancies is (lOQ-ff) /2 (maternal contribution to all CF fragments) + (the fetal contribution), where ff is the fetal fraction (percentage of maternal plasma fragments of fetal origin). There are four distributions in unaffected pregnancies; (i) if neither parent is a carrier the mean is zero and the standard deviation is zero, (ii) if the mother but not the fetus is a carrier the mean is (\00-ff)/ 2, (iii) if the fetus is a carrier, with the mutation inherited from the mother the mean is (100-f )/ 2 +ff/2, (iv) if the fetus is a carrier with the mutation inherited from the father the mean is ffll.

(i) Pregnancies with two different CF mutations in the maternal plasma.

In this situation, only the predominant CF mutation is informative (which is always the mutation inherited from the mother). For example, in an affected pregnancy with a fetal fraction of 10%, the mother's plasma will, in expectation, contribute 45% of DNA fragments with a CF mutation and the fetus 10%, half of which is from the father and can be

disregarded, i.e. a total of 50% of DNA fragments a the relevant site have a CF mutation. As above, we estimated the distribution of percentages in affected and unaffected pregnancies for specified fetal fractions. The mean of the distribution in affected pregnancies is (100-ff)/2 + ff/2 and is (100-ff)/2 if the fetus is a carrier.

The distributions were derived for increasing number of DNA fragments counted (because the more counted, the larger N, and the smaller the standard error) and for different fetal fractions (because the larger the fetal fraction, the more separated the distributions) to determine the minimum number of counts needed to obtain complete or near complete separation of the distribution in affected and unaffected pregnancies.

We then selected a panel consisting of 13 CF mutations that includes CF mutations most commonly found in Europe and North America ¹ ' ¹⁰ ' ^19,20 , each of which accounts for at least 0.3% of people with a CF mutation in the population and determined the revised number of counts needed to achieve such separation using 13 mutations. A positive result is one in which the percentage of CF mutations is equal to or greater than a specified cut-off and a screen negative result is one with values below the cut-off. The detection rate (sensitivity) is the proportion of affected pregnancies with a positive result and the false-positive rate is the proportion of unaffected pregnancies with a positive result.

The detection rate was estimated from the proportion of the total area under the distribution of percentage of DNA fragments with a given CF mutation in affected (CF) pregnancies above a specified cut-off level multiplied by the proportion of all CF mutations due to the given mutation in the population.

The false-positive rate was estimated from the proportion of the total area under the distributions of percentage of DNA fragments with a given CF mutation in unaffected pregnancies above the specified cut-off level multiplied by the proportion of all CF mutations due to the given mutation in the population.

Confined placental mosaicism involving trisomy 7, which has an estimated prevalence of 0.2% ^21,22 , has a small influence on the false-positive rate because the "fetal" DNA is, in fact, placental. ²³ If the fetus is a CF carrier and has inherited the CF mutation from the mother, the distribution of the percentage of DNA counts with a CF mutation will depend on whether the extra copy of chromosome 7 in the placenta has the CF mutation; if it does, the distribution will be shifted to the right (i.e. be a little higher), but if not, it will be shifted to the left (i.e. a little lower) and the amount of the shifts will depend on the fetal fraction. The percentage points shift is ± 1/6 x ff, as each percentage point increase in the fetal fraction produces a shift of ± 1/6. For example if the fetal fraction is 9%, then if the fetal genotype is OXX (where X represents presence of the CF mutation and 0 its absence) the fetus contributes 6% of DNA fragments with a CF mutation and 3% without (to sum to 9%). The mother contributes equal numbers of each, hence 45.5% with a CF mutation and 45.5% without, yielding a total of 51.5% with a CF mutation, or 48.5% if the fetal genotype is 00X. With a fetal fraction of 10% the corresponding figures are 51 4/6 and 48 2/6, ie. 1/6 greater in both examples. This calculation is based on all the placenta being mosaic, not just part of it. Consequently our estimates of the false-positive rates may be somewhat higher than those found in practice.

The odds of being affected given a positive result was estimated from the detection rate divided by the false-positive rate times the prevalence of CF expressed as an odds. The pregnancy prevalence of CF is 1 in 2500, or 1 :2499 as an odds. The odds of being affected given a positive result is therefore detection rate/false-positive rate x 1 :2499.

Results

Figure 1 shows the derivation of the percentage of DNA fragments with a CF mutation and the corresponding status of the fetus if the fetal fraction is 10% (affected (XX), carrier (OX), not a carrier(00)) according to parental CF carrier status. If the fetus is affected (i.e. has CF) the fetal component of the plasma DNA is, in expectation, 55%. If the fetus is unaffected and is not a carrier it is 45% or 0% (depending on the parental carrier status). If the fetus is a CF carrier it is 50% or 5% (again depending on the parental carrier status). In this way, affected pregnancies are distinguished from others and a result of 55%, that can statistically be separated from the expected 50% or less, defines a positive screening result.

Figure 2 shows the derivation of the percentage of DNA fragments with a CF mutation in maternal plasma where two different CF mutations are identified (XI and X2). In maternal plasma the predominant mutation is always from the mother because there is no direct contribution from the father. The expected percentage of CF mutations at the predominant CF mutation site in an affected pregnancy is 50% and 45% in an unaffected pregnancy.

Figure 3 shows the estimated relative distributions of DNA fragments with one CF mutation according to fetal fractions (10%, a typical value, and 4%, a lower limit typically used in prenatal DNA Down syndrome screening) ^24,25,26 , the number of DNA fragments sequenced that include the mutation site, and the status of the fetus (affected or unaffected). With a fetal fraction of 10%, counting 8,000 sequenced DNA fragments gives almost complete separation of the relative distributions for each of the three possible fetal genotypes, with complete (or near complete) discrimination, and consequently a very low false positive rate. With a fetal fraction of 4%; counting 64,000 fragments gives good discrimination between affected and unaffected pregnancies. Figure 3 shows that with a fetal fraction of 3%, there is also good discrimination with 64,000 DNA fragments counted, but not with a fetal fraction of 2%.

Figure 5 shows the estimated relative distribution of DNA fragments with two different mutations found in the maternal plasma when 32,000 fragments are counted per site according to fetal fraction. When two different mutations are found, the mean for the predominant CF mutation in affected pregnancies is always 50% regardless of the fetal fraction. This can be seen from Figure 2 where the fetal fraction is 10%. Were the fetal fraction to be 4%, the contributions from the mother and the fetus would still sum to 50% (2% + 48% instead of 5% + 45% when the fetal fraction is 10%). In an unaffected pregnancy in which the fetus is a carrier, the mean increases towards 50% with decreasing fetal fraction and consequently the cut-off to determine a positive test result is dependent on the fetal fraction.

The screening method is robust to sequencing errors because the number of DNA fragments sequenced is large. The distributions in Figures 3 and 4 relate to one CF mutation. Consequently if N CF mutations are used in the test, the number of DNA fragments to be sequenced and counted is multiplied by N.

Table 1, that specifies the selected 13 CF mutations, shows their prevalence in the population. Together, they account for an estimated 81.5% of people with a CF mutation. It follows that the CF detection rate (proportion of CF pregnancies detected) is 66.4%, using the CF mutations in Table 1, calculated from 81.5% x 81.5%, because for a fetus to be affected it must have two CF mutations, one from each parent, assuming random mating. Table 2 shows the estimated screening performance according to the screening cut-off (expressed as the percentage of DNA fragments with a CF mutation), fetal fraction and according to the number of CF mutations included in the test. The cut-off of choice is 51 % when one or no CF mutation is found in the maternal plasma sample. When two CF mutations are found in the maternal plasma the cut-off will vary, for example 46% with a fetal fraction of 10% or 49% with a fetal fraction of 4%. When the panel of 13 mutations is being tested for, the detection rate is 66% (limited on account of the number of mutations used in the test, not by the DNA analysis) and the false-positive rate is <0.001% if the fetal fraction is 4% or greater, with an odds of being affected with a positive result of 140:1. If the fetal fraction is 10% or more the odds decrease to 16: 1. The increase in the odds of being affected given a positive result with decreasing fetal fraction shown in Table 2 appears paradoxical but is explained because a smaller fetal fraction results in both a reduced separation of the distributions (shown for example in Figure 2) and also the opposite effect because of the reduced effect of placental mosaicism. If the fetal fraction is >2%, the odds of being affected given a positive result declines with increasing fetal fraction and decreases at lower levels. Table 2 also shows that the detection rate is reduced to 33% with a fetal fraction of 2%, thus setting a practical lower limit of 3%; less than 1% of pregnancies have a fetal fraction less than 3%. ²⁴ To achieve the 66% detection rate shown in Table 2 requires the use of 13 CF mutations in the test and an estimated 416,000 targeted DNA fragments need to be counted (13 mutations x 32,000 fragments per mutation).

Discussion

The present method of screening regards screening in the same way as prenatal screening for neural tube defects, or Down syndrome. Each pregnancy is considered as a fresh screening opportunity, and the assessment of the screening test is based on quantifying DNA markers of the disorder being screened for, instead of the concentration of, for example, a marker protein. It is not necessary to identify carriers, which is a considerable benefit from a screening perspective, because almost all carriers will never have an affected pregnancy. The fact that the disorder being screened for is inherited is, from the screening perspective, irrelevant. The method is also unaffected by women having a different partner in a subsequent pregnancy.

In the present method of screening, there is no need to carry out tests to determine the carrier status of parents, because the screening can be performed using only a prenatal maternal plasma sample. The present method therefore simplifies the screening process and maintains detection rates achieved in screening based on parental carrier testing. Parental carrier testing identifies carrier couples (prevalence 4% x 4% = 0.16%, assuming random mating) and, among these, three out of four are false-positive (0.16% x 0.75 = 0.12%), 60 times higher than the present method with its false-positive rate of 0.002% (Table 2). Consequently, the odds of being affected with a positive result is about 50 times higher (16: 1 v 1 :3) which means that about three quarters of invasive diagnostic tests are avoided. It also has the benefit that nonpaternity, the rate of which varies among populations, but is typically 2% ²⁸ , becomes irrelevant.

Maternal plasma typically contains about 2400 haploid (single duplex DNA strand) whole genome equivalents/ml. ^34,35 Therefore to count 32,000 targeted DNA fragments containing a CF mutation site requires about a 14 ml plasma sample, which contains more than 25 billion DNA fragments.

As the number of different CF mutations used in the test increases, the number of DNA fragments that need to be counted also increases. The estimated number of DNA fragments to be counted using 13 CF mutations is 416,000 (13 x 32,000); a large number but only about four percent of the approximately 10 million required for unselected DNA-based screening that does not target selected DNA regions as is the case in antenatal screening for Down syndrome, trisomy 18, and trisomy 13. ²⁹ ' ³⁰ Increasing prenatal CF screening to DNA-based screening adds an extra step to amplify the selected CF DNA sites prior to sequencing that can readily be included in laboratory procedure. Different CF mutations can be tested simultaneously without requiring a larger maternal plasma sample because unique molecular identifiers can be used in multiplex amplification. ³² The selection of a screening cut-off involves making a judgment to maximize the detection rate, minimize the false positive rate, and achieve an acceptably high odds of being affected given a positive result. Table 2 suggests that a cut-off of DNA fragments with a CF mutation of 51% would be reasonable. At a 10% fetal fraction it achieves a 66% detection rate with a very low false-positive rate of 2 per 100,000 (or 0.002%). The selection of a higher cut-off of 52% results in a loss of detection with low fetal fractions even though the false-positive rate is even lower. The use of a lower cut-off of 50% retains detection but at the cost of a much increased false-positive rate. Knowledge of the minimum number of DNA fragments that need to be sequenced is useful because it avoids the cost of sequencing more fragments than is necessary. Given that the sequencing costs are reduced to about a tenth or less of those required for Down syndrome screening, and that the sequencing costs of a test are about half the total cost, the CF test should cost about half the cost of a Down syndrome DNA screening test.

The selection of the 13 most common CF mutations is somewhat arbitrary, and any programme adopting the screening method described here would need to take a view on the efficacy and cost-effectiveness of increasing or decreasing the number of CF mutations included in the test. The detection rate could be increased by increasing the number of CF mutations included in the test at the cost of having to increase the number of DNA fragments that would have to be counted, and with diminishing incremental gains in detection for each CF mutation added.

The American College of Medical Genetics/American College of Obstetricians and

Gynecologists recommend a 23 mutation panel for general population carrier testing. ³⁶ These 23 mutations account for 84% of carriers across all ethnic groups in a US population. ³⁶ The detection rate would then be increased from 66% using the 13 mutation panel to 71% using the 23 mutation panel, at a cost of having to increase the number of DNA fragments that would have to be sequenced and counted by 77% [(23/13)-l].

The method would be suitable for known CF carrier couples, provided the CF mutations in the parents were included in the mutation sites used in the method. Then all affected pregnancies would be identified in such couples and amniocentesis or CVS avoided in nearly all cases in which the pregnancy is unaffected, so about 3 in every 4 women could avoid an invasive diagnostic procedure. The screening method would incidentally identify the carrier status of the fetus and most parents. From a screening perspective, there is no reason to regard carriers as screen positive and no justification to report them as part of a screening programme adopting this method. It is not the purpose of screening to identify carriers of autosomal recessive disorders when being a carrier is of minor or no medical consequence.

The carrier status of some fathers cannot be determined, as illustrated by reference to Figure 1. When the percentage of fragments with a CF mutation in the maternal plasma is 0% it is not possible to distinguish case C from D where the former relates to the father being a carrier and the other does not. This also applies when the percentage is 45% or 50%. These cases combined contribute to just over 50% of fathers who are CF carriers. Since the identification of the carrier status of mothers only is feasible, this information could be used as a possible modification to the proposed screening strategy. If there were accessible record linkage data that identified women who had been tested in a previous pregnancy, those found to be non- carriers would not need to be tested again since their fetus could not be affected.

The principles and characteristics of the test described here could also be applied to other autosomal recessive disorders, such as Tay-Sachs disease, and to other inherited disorders in which gene dosage is important. The proposed method also detects de novo mutations provided they are included in the mutation set specified in any given screening programme. The method described here could be incorporated into a screening strategy in which the maternal carrier status is determined for a panel of autosomal recessive disorders as described above for CF, using maternal DNA from saliva, or another convenient source such as white blood cells. Mothers identified as carriers for any particular disorder would then be screened by the method described here.

Conclusion Prenatal maternal plasma or serum DNA screening for autosomal recessive disorders such as CF has several advantages over current screening, which is based on parental carrier testing. The estimated false-positive rate is 60 times lower, there is no loss in the detection of affected pregnancies, parental carrier testing is not needed, and about three quarters of invasive diagnostic procedures are avoided. Table 1. Panel of 13 common CF mutations. Data from Bobadilla et al. 2002 ¹ , Palomaki et al. 2002 ²⁰ , Hill et al. 2015 ¹⁰ . Mutation names are 'legacy names'

(www.genet.sickkids.on.ca/cftr/) that correspond to those in cited references. Updated nomenclature and numbering for nucleotides can be found in Bareil et al. (2010) ²⁷

Percent of people

Location of CF with a CF mutation

Mutation name mutation that can be detected

AF508 exon 10 68.6

G542X exon 11 2.4

G551D exon 11 2.1

3120 + 1 G>A intron 16 1.5

W1282X exon 20 1.4

N1303K exon 21 1.3

R553X exon 11 0.9

621 + 1 G>T intron 4 0.9

1717 - 1G>A intron 10 0.7

3849 + lOKb OT intron 19 0.7

1898 + 1 G>T intron 12 0.4

ΔΙ507 exon 10 0.3

2789 + 5 G>A intron 14b 0.3

Total 81.5

Table 2: Screening performance according to cut-off of percentage of DNA fragments with a CF mutation, number of CF mutations in test and feta fraction (32,000 DNA fragments sequenced per mutation included in the test)

Cut-off

(% DNA fragments with CF

mutation) Number of CF mutations included in test

Number of CF mutations found in

maternal plasma sample Most common (AF508) 6 most common* 13 most common*

2

Fetal (predominant

fraction mutation

(%) O or 1 only) DR (%) FPR (%) OAPR DR (%) FPR (%) OAPR DR (%) FPR (%) OAPR

20 50 40 47 0.7 1:36 60 0.8 1:32 66 0.8 1:31

51 41 47 0.001 14:1 60 0.002 15:1 66 0.002 16:1

52 42 47 0.001 14:1 60 0.002 15:1 66 0.002 16:1

10 50 45 47 0.7 1:36 60 0.8 1:32 66 0.8 1:31

51 46 47 0.001 14:1 60 0.002 16:1 66 0.002 16:1

52 47 47 <0.001 110:1 60 <0.001 120:1 66 <0.001 130:1

4 50 48 47 0.7 1:36 60 0.8 1:32 66 0.8 1:31

51 49 47 <0.001 1.10:1 60 O.001 130:1 66 <0.001 140:1

52 50 24 0 - 30 0 - 33 0 -

3 50 48.5 47 0.7 1:36 60 0.8 1:32 66 0.8 1:31

51 49.5 45 <0.001 350:1 58 <0.001 390:1 64 <0.001 410:1

52 50.5 2 0 - 2 0 - 2 0 -

2 50 49 47 0.7 1:37 60 0.8 1:33 66 0.8 1:31

51 50 24 <0.001 710:1 30 <0.001 800:1 33 <0.001 840:1

52 51 0 0 - 0 0 - 0 0 -

^♦ See Table 1

**Cut-off set to provide same screening performance as when 1 or no CF mutation is found. DR = detection rate; FPR = false-positive rate; OAPR = odds of being affected given a positive result (given to 2 significant figures)

References

1. Bobadilla JL, Macek M Jr, Fine JP, Farrell PM. Cystic fibrosis: a worldwide analysis of CFTR mutations - correlaton with incidence data and application to screening. Hum Mutat 2002;19, 575-606.

2. Farrell PM. The prevalence of cystic fibrosis in the European Union. J Cystic Fibrosis 2008; 7, 450-453.

3. Southern KW, Munck A, Pollitt R et al. A survey of newborn screening for cystic fibrosis in Europe. J Cystic Fibrosis 2007; 6, 57-65.

4. Romeo G, Devoto M, Galietta LJV. Why is cystic fibrosis so frequent? Hum Genet 1989; 84, 1-5.

5. Wald NJ. Couple screening for cystic fibrosis. Lancet 1991; 383, 1318-139.

6. Wald NJ, George L, Wald N, MacKenzie IZ. Further observations in connection with couple screening for cystic fibrosis. Prenat Diagn 1995; 15, 589-90.

7. Haddow JE, Bradley LA, Palomaki GE et al. Issues in implementing prenatal screening for cystic fibrosis: results of a working conference. Genet Med 1999; 4, 129-35.

8. Wald NJ, Morris JK, Rodeck CH et al. Cystic fibrosis: selecting the prenatal screening strategy of choice. Prenat Diagn 2003 ;23, 474-483.

9. Debrand E, Lykoudi A, Bradshaw E, Allen SK. A non-invasive droplet digital PCR (ddPCR) assay to detect paternal CFTR mutations in the cell-free DNA (cffDNA) of three pregnancies at risk of cystic fibrosis via compound heterozygosity. PloS ONE 2015; 10, e0142729.

10. Hill M, Twiss P, Verhoef T et al. Non-invasive prenatal diagnosis for cystic fibrosis: detection of paternal mutations, exploration of patient preferences and cost analysis. Prenat Diagn 2015; 23, 1-9.

11. Chiu RWK, Chan KCA, Gao Y et al. Noninvasive prenatal diagnosis of fetal

chromosomal aneuploidy by massively parallel genomic sequencing of DNA in maternal plasma. Proc Natal Acad Sci USA 2008; 105, 20458-20463.

12. Fan HC, Blumenfeld YJ, Chitkara U et al. Noninvasive diagnosis of fetal aneuploidy by shotgun sequencing DNA from maternal blood. Proc Natl Acad Sci USA 2008; 105, 16266- 71.

13. Lo YMD, Chiu RWK. Genomic analysis of fetal nucleic acids in maternal blood. Annu Rev Genomics Hum Genet 2012;13, 285-306. 14. Lun FMF, Tsui NB, Chan KCA et al. Noninvasive prenatal diagnosis of monogenic diseases by digital size selection and relative mutation dosage on DNA in maternal plasma. Proc Natal Acad Sci USA 2008; 105, 19920-19925.

15. Lam K-WG, Jiang P, Liao GJW et al. Noninvasive prenatal diagnosis of monogenic diseases by targeted massively parallel sequencing of maternal plasma: application to β- thalassaemia. Clin Chemistry 2012; 58, 1467-1475.

16. New MI, Tong YK, Yuen T et al. Noninvasive prenatal diagnosis of congenital adrenal hyperplasia using cell-free fetal DNA in maternal plasma. J Clin Endocrinol Metab 2014; 99, E1022-E1030.

17. Liao GJW, Lun FMF, Zheng YWL et al. Targeted massively parallel sequencing of maternal plasma DNA permits efficient and unbiased detection of fetal alleles. Clin Chem 2011;57, 92-101.

18. Liao GJW, Chan KCA, Jiang P et al. Noninvasive prenatal diagnosis of fetal trisomy 21 by allelic ratio analysis using targeted massively parallel sequencing of maternal plasma DNA. PLoS ONE 2012;7, e38154.

19. Palomaki GE, FitzSimmons SC, Haddow JE. Clinical sensitivity of prenatal screening for cystic fibrosis via CFTR carrier testing in a United States panethnic population. Genet Med 2004; 6, 405-14.

20. Palomaki GE, Haddow JE, Bradley LA, FitzSimmons SC. Updated assessment of cystic fibrosis mutation frequencies in non-Hispanic Caucasians. Genet Med 2002; 4, 90-94.

21. Kalousek DK, Vekemans M. Confined placental mosaicism. J Med Genet 1996; 33, 529- 533.

22. Grati FR, Grimi B, Frascoli G et al. Confirmation of mosaicism and uniparental disomy in amniocytes, after detection of mosaic chromosome abnormalities in chorionic villi. Eur J Hum Genet 2006; 14, 282-288.

23. Hochstenbach R, Nikkels PGJ, Elferink MG et al. Cell-free fetal DNA in the maternal circulation originates from the cytotrophoblast: proof from a unique case. Clinical Case Reports 2015; 3, 489-491.

24. Canick JA, Palomaki GE, Kloza EM et al. The impact of maternal plasma DNA fetal fraction on next generation sequencing tests for common fetal aneuploidies. Prenat Diagn 2013;33, 667-674.

25 Hudecova I, Sahota D, Heung MMS et al. Maternal plasma fetal DNA fractions in pregnancies with low and high risks for fetal chromosomal aneuploidies. PLoS ONE 2014; 9, e88484. 26. Palomaki GE, Kloza EM, Lambert-Messerlian GM et al. DNA sequencing of maternal plasma to detect Down syndrome: an international clinical validation study. Genet Med 2011;13, 913-920.

27. Bareil C, Theze C, Beroud C et al. UMD-CF77.: a database dedicated to CF and CFTR- related disorders. Hum Mutat 2010; 9, 1011-1019.

28. King TE, Jobling MA. Founders, drift and infidelity: the relationship between Y chromosome diversity and patrilineal surnames. Mol Biol Evol 2009;26, 1093-1102.

29. Chiu RWK, Lo YDM. Noninvasive prenatal diagnosis empowered by high-throughput sequencing. Prenat Diagn 2012; 32, 401-406.

30. Dar P, Shani H, Evans MI. Cell-free DNA: comparison of technologies. Clin Lab Med 2016; 36, 199-211.

31. Kivioja T, Vaharautio A, Karlsson K et al. Counting absolute numbers of molecules using unique molecular identifiers. Nature Methods 2012; 9,72-74

32. Stahlberg A, Krzyzanowski PM, Jackson JB et al. Simple, multiplexed, PCR-based barcoding of DNA enables sensitive mutation detection in liquid biopsies during sequencing. Nucl Acids Res 2016;44, el 05

33. Kinde I, Wu J, Papadopoulos N et al. Detection and quantification of rare mutations with massively parallel sequencing. Proc Natl Acad Sci USA 2011 ;108,9530-9535

34. Manokhina I, Singh TK, Penaherrera M et al. Quantification of cell-free DNA in normal and complicated pregnancies: overcoming biological and technical issues. PLoS ONE 2014; 9,el01500

35. Bischoff FZ, Lewis DE, Simpson JL. Cell-free fetal DNA in maternal blood: kinetics, source and structure. Human Reproduction Update 2004;ll,59-67

36. Watson MS, Cutting GR, Desnick RJ et al. Cystic fibrosis population carrier screening: 2004 revision of American College of Medical Genetics mutation panel. Genet Med 2004;6, 387-391

FURTHER ASPECTS

1. A non-invasive prenatal in vitro screening method for cystic fibrosis (CF) in a fetus using cell-free DNA (cfDNA) from a maternal sample, comprising:

(a) performing targeted amplification of one or more DNA regions of the cfDNA,

wherein the or each DNA region comprises at least one site where a CF mutation may occur (a "CF mutation site");

(b) performing DNA sequence analysis of a number of DNA molecules ("n") that are the product of amplification for the or each DNA region and determining the number of those DNA molecules that comprise and/or do not comprise a CF mutation at the or each CF mutation site, wherein n is:

(i) more than or equal to N x 32,000; or

(ii) less than or equal to N x 640,000; or

(iii) more than or equal to N x 32,000 but less than or equal to N x 640,000;

and wherein N is the number of different CF mutation sites used in the method;

(c) for a CF mutation site, calculating the ratio of the number of sequenced DNA

molecules comprising the CF mutation to the number of sequenced DNA molecules that do not comprise the CF mutation (the "ratio"), or calculating the ratio of the number of sequenced DNA molecules that do not comprise the CF mutation to the number of sequenced DNA molecules comprising the CF mutation (the "inverse ratio"); and

(d) for the CF mutation site, comparing the calculated ratio or inverse ratio to a

predetermined cut-off value and thereby determining whether the sample is test- positive for a fetus affected with CF, wherein the sample is test-positive for a fetus affected with CF if the calculated ratio exceeds a predetermined cut-off value or if the inverse ratio is less than the predetermined cut-off value.

2. The method according to aspect 1 , wherein one mutation is found or no mutations are found.

3. The method according to aspect 2, wherein the cut-off value for DNA fragments with a CF mutation is 51% or more. 4. The method according to aspect 1 , wherein two different mutations are found, one of which is a predominant (maternally-derived) CF mutation and the other is a paternally- derived mutation, the method comprises:

(c) calculating the ratio of the number of sequenced DNA molecules comprising the predominant (the maternally-derived) CF mutation to the number of sequenced DNA molecules that do not comprise the predominant CF mutation (the "ratio"), or calculating the ratio of the number of sequenced DNA molecules that do not comprise the predominant CF mutation to the number of sequenced DNA molecules comprising the predominant CF mutation (the "inverse ratio"); and

(d) comparing the calculated ratio or inverse ratio to a predetermined cut-off value and thereby determining whether the sample is test-positive for a fetus affected with CF, wherein the sample is test-positive for a fetus affected with CF if the calculated ratio exceeds a predetermined cut-off value or if the inverse ratio is less than the predetermined cut-off value.

5. The method according to aspect 4, wherein the cut-off value for DNA fragments with a CF mutation is 46% with a fetal fraction of 10% or more, or 49% with a fetal fraction of 4% or less.

6. The method according to any one of the preceding aspects, further comprising the step of obtaining cfDNA derived from a maternal sample and/or denaturing the cfDNA derived from a maternal sample to single strands before step (a).

7. The method according to any one of the preceding aspects, wherein step (a) comprises performing targeted amplification of a DNA region of the cfDNA, wherein the DNA region comprises at least one site where a first CF mutation may occur (a "first CF mutation site"); and performing targeted amplification of one or more further DNA regions of the cfDNA, wherein the or each further DNA region comprises one or more CF mutation sites that are different to the first CF mutation site.

8. The method according to any one of the preceding aspects, wherein n is more than or equal to N x 64,000. 9. The method according to any one of the preceding aspects, wherein n is less than or equal to N x 128,000.

10. The method according to any one of the preceding aspects, wherein the percentage of maternal plasma DNA fragments that is of fetal origin in the sample is at least 2%, or at least 3%.

11. The method according to any one of the preceding aspects, wherein the number of CF mutation sites (N) is 2 or more, 4 or more, 6 or more, 8 or more, or 10 or more.

12. The method according to any one of the preceding aspects, wherein (i) N is 13 and n is more than or equal to 416,000 but less than or equal to 1,664,000; or (ii) N is 13 and n is more than or equal to 832,000 but less than or equal to 1,664,000.

13. The method according to any one of the preceding aspects, wherein the one or more CF mutation sites are selected from the mutations in Table 1.

14. The method according to any one of the preceding aspects, wherein the one or more CF mutation sites comprise AF508.

15. The method according to any one of the preceding aspects, wherein the one or more CF mutation sites comprise all of the mutation sites in Table 1.

16. The method according to any one of the preceding aspects, wherein DNA fragments in the sample are labelled with barcodes (unique molecular identifiers) before targeted amplification.

17. The method according to any one of the preceding aspects, wherein the method does not comprise:

(i) maternal and/or paternal CF carrier testing; and/or

(ii) a step of deducing whether a CF mutant allele is inherited from the mother or the father; and/or

(iii) a step of enriching the fetal DNA fraction of the maternal sample; and/or (iv) bias towards counting DNA molecules derived from the fetus, for example by digital Nucleic Acid Size Selection (NASS); and/or

(v) analysing a mutation outside of the CFTR gene locus in order to determine whether the sample is test-positive for a fetus affected with CF; and/or

(vi) haplotype analysis; and/or

(vii) relative haplotype dosage analysis (RHDA) using Single Nucleotide

Polymorphisms (SNPs).