Introduction
Newborn screening (NBS) is one of the most successful public health initiatives for preventing disability and death. Since the first NBS was started in the 1960s to test for phenylketonuria (PKU), diseases included in NBS have been expanded extensively [
1,
2]. The introduction of tandem mass spectrometry (MS/MS) is one of the most important advances in NBS, because it allows more than 40 metabolic disorders to be detected from dried blood spot (DBS) samples [
3,
4].
Although previous screening methods featured simple, rapid, and convenient procedures, limitations also exist because metabolites or enzymatic activities are often influenced by multiple factors [
5,
6]. Some disorders, such as hearling loss, do not develop evident symptoms during the regular NBS time window; therefore, these disorders could not be detected readily by current NBS methods [
7]. Analysis of disease-associated DNA molecules or genes could be a great supplemental technology for conventional screening, particularly for the disorders without specific metabolites or biomarkers such as severe combined immune deficiency (SCID) and spinal muscular atrophy (SMA) [
8‐
10].
A variety of DNA-based techniques have been applied in NBS of specific disorders. Recent development and rapid implementation of next-generation sequencing (NGS) in the molecular diagnosis of genetic disorders provide further options and opportunities for DNA-based NBS, as exemplifed by NGS-based genetic analysis as a regular diagnostic work-up following an abnormal NBS result [
4,
11]. For disorders with clearly defined pathogenic variations, genetic screening can be used in first-tier tests of SCID [
8,
12], XLA [
9], SMA [
10,
13], fragile X syndrome [
14,
15] and hearing loss [
16]. It has been reported that genetic testing as second-tier analyses improves the diagnostic specificity of CF [
17], MMA [
18], DMD [
19] and NICCD [
20]. In 2013, four genetic screening projects (BabySeq/NBSeq/NC NEXUS/STATseq) based on NGS to explore the clinical utility of whole exome sequencing (WES) and whole genome sequencing (WGS) were implemented in NBS. Although genome-wide genetic screening provides information for significantly more genetic disorders, the interpretation and reporting of genomic variants remain challenging, such as the heteroplasmy of mitochondrial variants and the unknown significance (VUS) variants. In addition, WES and WGS are still relatively expensive and the amount of DNA from dried blood is limited for sequencing library preparation [
21]. Consequently, WES/WGS were not considered suitable for first-tier testing, but could be of merit as second-tier testing according to the programs [
11,
22‐
24]. A more efficient and less expensive approach is warranted for sequence analysis of DNA from DBS to screen for congenital disorders in neonates.
In the present study we designed a newborn genetic sequencing (NBGS) panel based on multiplex PCR to analyze sequences of 134 genes of 74 neonatal inborn disorders that were validated with 287 samples that have molecular diagnosis previously. Furthermore, retrospective studies were performed with 4986 archived DBS for exploring a potential screening strategy that combines conventional NBS (C-NBS) with genetic screening of neonatal inborn disorders, as well as increasing the capacity to screen out diseases not contained in C-NBS.
Materials and methods
Study subjects
A total of 287 samples with molecular diagnoses from previous genetic analyses were enrolled in the validation cohort to validate the performance of the NBGS panel. The retrospective cohort includes two groups of 4986 samples in total, from neonates born between January 2017 and December 2019, who received C-NBS, including inherited metabolic disease screening by MS/MS methods, and congenital hypothyroidism (CH) and glucose-6-phosphate dehydrogenase deficiency (G6PDD) screening by biochemical methods. Of the 4986 samples, group 1 of 2973 were randomly selected from those with a negative initial C-NBS results, and group 2 of 2013 were from samples with a positive result in the initial C-NBS, which received repeat MS/MS analysis. All 4986 samples were analyzed with NBGS, of which 5 samples were excluded from data analysis due to low quality of NGS data and 4981 NGS result were included in the following analysis. In group 2, 53 newborns were C-NBS true positive as confirmed by repeat C-NBS and by clinical manifestation, 8 samples were lost to follow-up, 5 infants with unexplained death, and 1947 were negative by repeat C-NBS and were denoted as MS/MS false-positive cases. The potential capacity of NBGS to reduce C-NBS false-positives and C-NBS false-negatives was estimated in 1947 MS/MS false-positive newborns and the remaining newborns that were undiagnosed clinically, respectively. Also, the value of NBGS to screen disorders out of C-NBS was evaluated in these newborns. The C-NBS result, hearing screening outcome and clinical profiling including newborn gestational age (GA, in weeks), birth weight (BW, in grams) were collected from all newborns. This study was approved by the institutional review board of the ethics committee in Children’s Hospital, Zhejiang University School of Medicine.
Selection of disorders, genes and variants
Seventy four inborn disorders were included in this panel based on the Secretary’s Advisory Committee on Heritable Disorders in Newborns and Children and the Wilson and Jungner criteria [
25,
26], disorders that can be detected by biochemical test were selected with consideration of Recommended Uniform Screening Panel (RUSP) recommendation, and disease with high incidences but cannot be screened by C-NBS in China were also enrolled. The disorders in this panel consist of inherited metabolic diseases screened by MS/MS, G6PD, CH and genetic diseases with high incidences and severe impact on children’s health (Additional file
1: Table S1), including 39 inborn metabolic disorders screened by MS/MS and 35 disorders of skeletal system diseases, hematological system diseases, lysosomal storage disease, etc., of which 33 disorders are not covered by C-NBS. A total of 134 genes were included according to relative prevelance in associated 74 disorders. Known pathogenic variants in 134 genes (Additional file
2: Table S2) were selected as targets of the panel according to one of the following criteria: (1) high frequency in Chinese, Eastern Asian or Asian population; (2) common pathogenic variants in databases including ClinVar and ClinGen; (3) loss of function (LOF) variants in Asian population (≥ 10 allele count in Gnomad) (4) high frequency in local databases. A multiplex PCR panel was designed with each selected variant covered by at least one amplicon.
Library construction and sequence analysis
Genomic DNA were extracted from dried blood spots by Nucleic Acid Automatic Extraction System (Bioer, China). DNA libraries were prepared based on multiplex PCR by using the SLIMamp (StemLoop Inhibition Mediated amplification) method [
27]. High-throughput sequencing was performed with Illumina NextSeq 500 according to the manufacturer’s instruction.
Raw sequencing data were processed using general procedures. Low-quality sequencing reads were removed and the reads were mapped to the NCBI human reference genome (hg19/GRCh37). After variant calling and annotation, potential pathogenic variants were kept for further evaluation. In addition to the pre-selected pathogenic variants, pathogenicity of all other variants were evaluated according to ACMG sequence variant interpretation guideline [
28]. Only pathogenic and likely pathogenic variants were considered for reporting.
IEM panel
Inborn errors of metabolism (IEM) panel is an in-house targeted NGS panel of 86 genes (PAH, PTS, MUT, SLC22A5, etc.) associated with inherited metabolic disorders. Coding exons of target genes were captured using an Agilent High Sensitivity DNA Kit (Agilent, Santa Clara, CA, USA). Libraries generated from enriched DNA were sequenced using the Illumina NextSeq 500 platform (Illumina Inc., San Diego, CA, USA) in paired-end mode.
Statistical analysis
The comparison of the biochemical index, GA and BW of different groups was done by analysis of variance (ANOVA), and statistical analysis between any two groups was performed using a t-test (two-tailed).
Discussion
The compelling problems facing NBS include reducing false-positive and false-negative results, improving the positive detection rate and expanding the spectrum of screened disorders beyond current C-NBS methods [
20,
29‐
34]. We adapted a validated genetic screening panel for newborn genetic screening to achieve the goal. Our results showed that NBGS could reliably detect targeted variants, identify newborns affected by conditions not included in current C-NBS and reduce false-positive rate by C-NBS in our recruited population that highlight the complementarity of both methods.
In C-NBS true-positive newborns, 68% (36/53) of them were identified as positive by NBGS. Several studies have demonstrated the diagnostic rates of NGS panels for IEM-affected infants were approximately 50–59% [
35,
36]. However, a recent study showed that WES had an overall sensitivity of 88% for infants who screened positive by C-NBS, in which 32% of the reported variants were absent from both HGMD and ClinVar [
37], consistent with the sensitivity of the NBGS panel that only reports known mutations. NBGS showed good performance in disorders with high incidence such as G6PDD (100%) and phenylalanine hydroxylase deficiency (PAHD) (83.3%), similar to the detection rates reported in other studies [
32,
38,
39]. For the 17 C-NBS true-positive/NBGS negative cases, the possible reasons include: (1) CH in our study were all identified negative by NBGS, it was likely that the CH associated gene mutation identified in few patient, and genetic mechanism responsible remains to be elucidated, only 51.82% (57/110) CH patients have been reported carried biallelic mutations of 21 candidate genes [
40]. Due to limited genes and hotspot mutations, false-negative result would occur in genetic screening and it was suggested that the conventional screening is more suitable for CH in this study. (2) The NBGS negative P/LP variants or disorder were not targeted in our panel. (3) The VUS variants detected by IEM panel were not included in NBGS because the unclear mechanism and interpretation of VUS was challenging for clinicians in a screening panel. As literature and clinical data evolve, the VUS may become reportable and the panel should be reassessed by then. (4) Larger copy-number variants (CNVs) could not be identified by NBGS. (5) The mechanism of genetic disease was not defined completely and the relevant genes may be undiscovered, and 11 cases were also negative after performed IEM in this study.
In the 77 C-NBS false-positive and false-negative/NBGS positive infants, G6PDD was the most prevalent condition, followed by hearing loss and PAHD. Female carriers of heterozygous
G6PD mutations are at elevated risk of developing acute hemolysis, but could be difficult to recognize by conventional biochemical methods. It has been reported that female carriers identified by genetic screening showed mild phenotypes, which is also observed in our study [
41]. Therefore, combined biochemical methods with genetic screening could improve the identification of G6PDD. Mitochondrial 12S rRNA mutations are related to aminoglycoside induced deafness [
42]. The mutation frequencies of mtDNA m.1557A > G and m.1496 C > T in China were 0.20% and 0.03% respectively [
43]. Genetic screening helps to identify carriers and avoid deafness caused by administration of aminoglycoside antibiotics. For all of 16 PAHD newborns identified by NBGS, the c.158G > A variant was found in trans with a pathogenic or likely pathogenic variant in the
PAH gene. The c.158G > A variant has been reported as a pathogenic variant in hyperphenylalaninemia patients, however, this variant was also considered as "likely benign" due to high population frequency and clinical evidence [
44]. The concentration of phe elevated slightly at regular intervals and no clinical intervention was performed in 16 PAHD. It seems that c.158G > A may relate to mild hyperphenylalaninemia in our enrolled population. The most common mutation of mitochondrial encephalomyopathy is m.3224A > G, however, classical presentations may be absent in some pediatric patients [
45]. Because of the heteroplasmy, it is difficult to evaluate the mutation load in related tissues/organs, thereby making diagnosis more challenging. A total of six newborns were identified with carriers of m.3224A > G by NBGS, we have been followed up these cases until now, and all without any clinical manifestation, maybe it suggested that mitochondrial disorders were not suitable for newborn genetic screening.
Using NBGS, 4 cases were diagnosed with MMA, PCD and WD that had been missed by C-NBS. It has been reported that the determination of MMA by MS/MS has been affected by several factors, and one case was missed due to the concentrations of C3 and C3/C0 decrease with age [
46,
47]. However, unlike metabolites, genetic screening was sufficiently accurate and specific for inborn disorders because it does not vary with age, season or maternal status. WD was not screened in C-NBS due to unavailability of specific markers. Some asymptomatic patients do not receive effective treatment before irreversible injury is present [
48]. In our study, 2 newborns showed normal liver function initially but identified positive by NBGS. Both newborns were diagnosed, and the old sister of NP0439 also was determined soon. It was note-worthy that applied genetic screening to C-NBS could effectively reduce the false-negative cases of C-NBS, and also could identify an additional subgroup of patients that undetectable by C-NBS, and the turn-around time was also decreased compared with conventional method that at the time of their clinical presentation arisen, actually, the turn-around time of C-NBS methods would takes about 1–2 months, one with positive initial screening result and positive multiple retest needs other biochemical test and genetic analysis to confirm. However, the testing period of NBGS is just 10 working days and C-NBS result was also finished during the time, combined NBGS and traditional methods, the turn-around time would decreased to 2–3 weeks. Moreover, expanding the spectrum of disorders that can not be screened by MS/MS could be highly beneficial especially for those with high incidence, treatable and early diagnosis is helpful for their prognosis. In our panel, a total of 33 disorders out of C-NBS could be identified such as WD, SCID and XLA.
Recently, the carrier rate of multiple disorders had been mainly reported in carrier screening of the general population. The carriers for at least one disorder was reported to be 24% [
49], as in our study also showed about 1/3 screened individuals with at least one pathogenic mutation in childhood diseases. It was challenging to provide genetic counseling for such a large amount of all infants but that is meaningful to carriers of genetic disorders to determine risk for future offspring and their parents to have additional children. In the future prospective study, we would offer an "opt-out" option for reporting carrier state when signed informed consent of NBGS testing and the related counseling. If the parents choose to report carrier result and the infant identified as carrier, we can provide genetic counseling for them. The top genes such as
PAH,
SLC22A5,
ACADS and
MUT were most frequently reported for carriers known to have high carrier frequency in reported study [
46], which could verify the relative high incidence of associated inborn disorders [
50,
51]. Some high-frequency mutations of
DUOX2, HBB,
ATP7B gene in our study were not found in previous investigations. This may be attribute to the differences in targeted population, sample size, genetic background, and detected range of mutations [
47,
50]. In addition, although NBGS panel was designed to screen neonates, reporting increased risk for certain conditions may be beneficial, particularly for carrier parents. However, we would recommend additional consents and counseling for reporting these conditions since uncertainty of disease onset and reduced penetrance could cause unnecessary burden for participants and change the risk–benefit balance.
In 1890 C-NBS false-positive/NBGS negative newborns, carriers with
ASS1 and
MCCC1 mutation showed abnormal biochemical results, but all newborns with normal phenotype until now. Generally, it is unlikely that variants of carrier would impact phenotype. A previous study showed 4.3% heterozygous carriers with mild presentations such as in Hemophilia A, CH, and G6PDD [
41]. Given the limited numbers in carriers with phenotype and the limited understanding of their penetrance and functional studies, the relationship of carrier status and phenotypic expression is still ambiguous [
41]. Therefore, NBGS could help to exclude some false positives and to reduce follow-up time and anxiety of heterozygous carriers. Moreover, we also observed GA and BW with significant difference in false-positive newborns and control. The physiological states associated with preterm, LBW, and sick newborns and the treatments received directly affect the reliability of results for many diseases screened in public health [
52‐
54]. Metabolite levels were influenced by birth weight and gestational age [
6,
55], which might increase false positives of C-NBS but could screened out by NBGS. In low BW and small GA newborns, we have not found difference in each subgroups which maybe due to the small amount of preterm infants in our study.
In summary, our results showed that advantages of the NBGS panel as a novel newborn screening test may include: (1) analysis for more preventable and treatable disorders that are not targeted by C-NBS; (2) pre-selected and pre-curated mutations that avoid complex and cumbersome interpretation and reporting; (3) lower-cost and faster turn-around time compared with WES or capture-based NGS panels that need more complex experimental process and data analysis; (4) providing rapid diagnosis of affected newborns after screened positive by C-NBS and improving the cost-effectiveness; (5) reducing false-positive newborns identified by C-NBS. Our data suggest that it may be a more effective strategy for newborn screening to combine NBGS with C-NBS. However, the NBGS test does have limitations. Because the NBGS panel only covers pre-selected known mutations, it will inevitably miss some pathogenic variants, which may lead to false-negative results and low overall detection rates. Although capture-based NGS methods could provide almost complete coverage of coding regions of targeted genes, variants in noncoding regions such as introns and untranslated regions are usually not readily detected, as well as large copy number variations (CNV) and complex structural variations (SV). In addition, because the NBGS test only reports pathogenic and likely pathogenic variants, a significant number of rare variants, particularly missense and splicing variants, will be intepreted as VUS due to lack of reported clinical or research evidence, and thus not reported, as indicated in a recent study [
37]. Nonetheless, if C-NBS and NBGS are concurrently applied, joint ananlysis of biochemical and sequence data would help optimize both tests reciprocally. Further research is needed to evaluate the clinical utility and effectiveness in a larger newborn population to explore the long-term implication of sequencing-based newborn screening.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.