Development and validation of a haplotype‐free technique for non‐invasive prenatal diagnosis of spinal muscular atrophy

Abstract Objective To develop a technique for non‐invasive prenatal diagnosis of spinal muscular atrophy and validate its performance. Study Design Pregnant women with 1 copy of SMN1 and male fetuses were enrolled. Seventeen women were included in test set A, and 10 of them were selected into test set B randomly and blinded. The two sets were tested independently by two different researchers blinded to fetal genotypes. Fetal DNA fractions were calculated based on the relative proportion of mapped chromosome Y sequencing reads. An algorithm was developed to decide fetal SMN1 copy numbers. Results The concordance rate with the results of MLPA testing of amniocyte DNA was 94.12% in test set A and 90% in set B. For all tests with a classifiable result, the percent of agreement with the results of MLPA testing of amniocyte DNA was up to 100% (25/25). Conclusion We have developed a direct, rapid, and low‐cost technique, which has a potential to be utilized for first‐trimester non‐invasive prenatal diagnosis and screening for spinal muscular atrophy with considerable reliability and feasibility.


| INTRODUC TI ON
Spinal muscular atrophy (SMA) is one of the most common autosomal recessive diseases causing infant mortality, with an incidence around 1 in 10 000 births. 1 About 81%~95% of SMA patients have no detectable exon 7 of the SMN1 gene, 2 which is located in a 1.5-Mb reverse-duplicated region containing multiple copies of homologous sequences. 3 Survival motor neuron 2 (SMN2) gene is also located in the 5q13 region, the coding sequence of which differs from SMN1 only by the 6th nucleotide of exon 7, where a C-to-T transition leads to the alternatively spliced isoform translating the non-functional SMN△7 protein. 4,5 Prenatal diagnosis is an essential prevention for SMA. Conventional procedure involves invasive approaches for fetal genetic materials such as amniocentesis and chorionic villus sampling (CVS), which harbor risks for miscarriage and infection. 6 Non-invasive prenatal diagnosis of SMA in earlier pregnancy would be timely and safer.
The discovery of cell-free DNA (cfDNA) in maternal plasma has enhanced the development of non-invasive prenatal testing (NIPT). 7 Although NIPT for fetal aneuploidies has already been clinically applied, non-invasive prenatal diagnosis for many single-gene disorders remains on the developing stage. For NIPD of SMA, a technique by targeted sequencing of cfDNA in maternal plasma and relative haplotype dosage (RHDO) analysis has been previously published. 8,9 However, this haplotype-based strategy has several limitations. Firstly, there is a rigid demand for DNA of the probands and parents, as well as adequate informative genomic markers beside the SMN1 gene. 10 This limitation restricts the scope of subjects applicable to this test. Secondly, a recombination event may result in incorrect fetal genotype classification if it occurs as a genomic location near the mutation. Thirdly, for de novo SMN1 mutations with a rate that is reported to be high, 11 and for germline mosaicism, haplotyping would fail or come out with false-negative results.
Droplet digital PCR is a technology with high sensitivity, specificity, and accuracy to detect and analyze low-abundance nucleic acids.
Its high resolution is guaranteed by millions of oil droplets generated per test. Utilizing digital PCR, the feasibility of non-invasive prenatal diagnosis (NIPD) for fetal monogenic disorders has been proved in several studies analyzing cfDNA. [12][13][14][15][16] In particular, for maternally inherited single nucleotide mutations, the relative mutation dosage (RMD) analysis based on the sequential probability ratio test (SPRT) has enabled detection of a slight increase in the load of the mutant allele in the maternal plasma of heterozygous carriers. 17 Digital PCR provides an ideal platform for the development of a haplotype-free test strategy for SMA-NIPD.
Unlike most other single-gene disorders, SMA harbors need and potential for a specific design of NIPD technique. The prominent hot spot mutation in the SMN1 gene, which is the loss of exon 7 copies but not point mutations, implies the utilization of single-base targeting strategy but disables the application of regular RMD algorithm.
The pseudogene SMN2 that disturbs quantification of SMN1 proposes a major obstacle. In this article, we present a novel technique that directly analyses SMN1 gene dosage using droplet digital PCR, as well as the results of performance validation.

| Design of probes and primers
TaqMan MGB probes were designed at the 6th nucleotide in exon 7 of SMN1/SMN2 gene and intron 3 of the reference ALB gene and synthesized by Thermo Fisher Scientific. We designed the length of SMN1/SMN2 and ALB amplicons as short as 75 bp and 72 bp to reduce the bias caused by unbalanced PCR amplification in favor of fetal cfDNA, which is generally shorter than maternal cfDNA. 18 Sequences of the probes and primers are listed in the Appendix S1.
Quantitative PCR was conducted for samples with various SMN1/ SMN2 copy numbers to validate the specificity of the probes.

| Droplet digital PCR
RainDrop droplet digital PCR should be performed following standard protocols, through processes including PCR mixture preparation, droplet sourcing, PCRs, and signal sensing. For each PCR, droplets with positive signal for SMN1/ALB should be counted using RainDrop Analyst II software.

| The digital relative SMN1 dosage method
Statistical analysis is essential for the determination of SMN1 copy number from digital PCR data. Based on the principle of Poisson distribution and hypothesis testing, we set up an algorithm called digital relative SMN1 dosage, as specified and illustrated in the Appendix S1. In short, a hypothesis that fetal SMN1 copy number equals 1 is established at first. Next, Pr(observed) value is generated for each test of one sample (one data set), which is determined solely by the number of reaction-positive droplets. Then, through comparing Pr(observed) to the upper and lower thresholds (derived from the number of reaction-positive droplets and FF) under a certain threshold likelihood ratio (a marker of statistical significance with a default value 2, a higher value represents higher reliability), the algorithm would return one of the three possible outcomes: accept the hypothesis (fetal SMN1 copy number = 1)/reject the hypothesis (fetal SMN1 copy number = 0 when n SMN1/ n ALB < 0.5; fetal SMN1 copy number = 2 when n SMN1/ n ALB > 0.5)/an unclassifiable result.

| Participants and sample processing
For the validation, we recruited pregnant women seeking SMA prenatal diagnosis on 16 ~ 22 weeks of gestation for this study from the Hunan Jiahui Genetics Hospital and signed informed consent.
All of the pregnancies had undergone non-invasive prenatal screening for fetal aneuploidies by next-generation sequencing (NGS).
Approval was obtained from the Ethics Committee of The Center for Medical Genetics, School of Life Sciences, Central South University, Hunan, China. For each participant, 6 ~ 10 mL of peripheral blood was collected in BCT Cell-Free DNA Blood Collection Tube (Streck) and 10 mL of blood was collected into k3-EDTA acid tubes. Weeks of gestation when sampling blood are listed in Table 1. Plasma was separated after double centrifugation within 6 hours after collection, one at 1600 g for 10 minutes and the second at 16000 g for 10 minutes. We extracted cell-free DNA from maternal plasma using the QIAamp Circulating Nucleic Acid Kit (Qiagen) following the manufacturer's instructions. The concentrations of cfDNA samples were tested on Qubit (Thermo Fisher Scientific). Amniotic fluid was obtained by amniocentesis, from which fetal genomic DNA was extracted by the phenol-chloroform method.

| Determination of fetal DNA fraction
We determined fetal DNA fraction (FF) of the samples based on the relative proportion of mapped chromosome Y (ChrY) sequencing reads, which is the golden standard method for FF determination. 19 In brief, low-coverage (0.1×) whole-genome sequencing was performed for the cfDNA samples. FF in maternal plasma was calculated by comparing the sequence tag density of ChrY in maternal plasma with the sequence tag density of ChrY in male plasma.

| Validation of probe specificity for SMN1 by quantitative real-time PCR
Results of TaqMan quantitative real-time PCR conducted on genomic DNA samples with various SMN1/SMN2 copy numbers were completely in accordance with the MLPA results. Samples with one or more SMN2 copies and no SMN1 copy did not produce fluorescence signal (FAM) of SMN1, which proved reliable specificity of the designed TaqMan MGB probes.    Note: † n SMN1 /n ALB : It is the only index determining hypothesis testing H1. H1: fetal SMN1 copy number = 0 (in cases that n SMN1 /n ALB < 0.5) or fetal SMN1 copy number = 2 (in cases that n SMN1 /n ALB > 0.5). ‡ Pr(observed): Pr(observed) = n ALB /(n ALB + n SMN1 ). It is a value entirely depending on the data of one single test on one sample. Fetal SMN1 copy number is determined by comparing the value of germline mosaicism/a recombination near the mutation occurred.

| D ISCUSS I ON
The validation results exhibited a considerable accuracy and repeatability of the technique. If shown to be robust in future systematically evaluation in a larger population, it may be a safer and more preferable alternative to traditional invasive prenatal diagnosis for SMA families, with an ability to identify affected fetuses at an earlier gestational age.
On the other hand, the feasibility and adaptability of the technique had also been proved by the tests. In terms of the cost, one run with eight samples on the RainDance platform only requires consumable items priced about $600 (including source chip, sense chip, and carrier oil, and would be even lower on a digital PCR platform other than RainDance) and a total experiment time about 6 hours.
Besides, in the present study, it is a cost-saving way to determine fetal fraction by analyzing the existing NGS data of prenatal screening for fetal aneuploidy, which has been generally used as the firsttier screening assay in clinical practice.  Note: † n SMN1 /n ALB : It is the only index determining hypothesis testing H1. H1: fetal SMN1 copy number = 0 (in cases that n SMN1 /n ALB < 0.5) or fetal SMN1 copy number = 2 (in cases that n SMN1 /n ALB > 0.5). ‡ Pr(observed): Pr(observed) = n ALB /(n ALB + n SMN1 ). It is a value entirely depending on the data of one single test on one sample. Fetal SMN1 copy number is determined by comparing the value of

ACK N OWLED G M ENTS
We thank Liu Yuan for technical supports and useful advice.

CO N FLI C T O F I NTE R E S T
The authors declare no conflict of interest.