Expanding Access to Noninvasive Prenatal Diagnosis for Monogenic Conditions to Consanguineous Families

Abstract

Background

Cell-free fetal DNA exists within the maternal bloodstream during pregnancy and provides a means for noninvasive prenatal diagnosis (NIPD). Our accredited clinical service offers definitive NIPD for several autosomal recessive (AR) and X-linked conditions using relative haplotype dosage analysis (RHDO). RHDO involves next-generation sequencing (NGS) of thousands of common single nucleotide polymorphism (SNPs) surrounding the gene of interest in the parents and an affected or unaffected offspring to conduct haplotype phasing of the high- and low-risk alleles. NGS is carried out in parallel on the maternal cell-free DNA, and fetal inheritance is predicted using sensitive dosage calculations performed at sites where the parental genotypes differ. RHDO is not currently offered to consanguineous couples owing to the shared haplotype between parents. Here we test the expansion of RHDO for AR monogenic conditions to include consanguineous couples.

Methods

The existing sequential probability ratio test analysis pipeline was modified to apply to SNPs where both parents are heterozygous for the same genotype. Quality control thresholds were developed using 33 nonconsanguineous cases. The performance of the adapted RHDO pipeline was tested on 8 consanguineous cases.

Results

The correct fetal genotype was predicted by our revised RHDO approach in all conclusive cases with known genotypes (n = 5). Haplotype block classification accuracies of 94.5% and 93.9% were obtained for the nonconsanguineous and consanguineous case cohorts, respectively.

Conclusions

Our modified RHDO pipeline correctly predicts the genotype in fetuses from consanguineous families, allowing the potential to expand access to NIPD services for these families.

Introduction

Prenatal diagnosis offers informed options for pregnancy and perinatal management, whether this is in preparing for the birth of an affected child and planning postnatal care or electing to terminate the pregnancy. Historically, the only option for prenatal diagnosis involved invasive testing via amniocentesis or chorionic villus sampling. Advances in prenatal diagnosis have led to the clinical implementation of less invasive methods of accessing fetal DNA, simply requiring a peripheral blood sample from the mother (1, 2).

Detection of maternally inherited variants in the cell-free DNA (cfDNA) of maternal plasma is complex owing to the high background of genomic material derived from the mother herself. Noninvasive prenatal diagnosis (NIPD) of rare maternally inherited monogenic conditions can be achieved using highly sensitive quantity-based approaches, such as relative haplotype dosage analysis (RHDO) (3) and relative mutation dosage. RHDO indirectly predicts fetal inheritance of the high- or low-risk parental alleles, irrespective of the familial pathogenic variant(s) present, whereas relative mutation dosage is targeted to the pathogenic variant itself. The major benefit of RHDO over relative mutation dosage is that the approach can be applied to a cohort of parents at risk of the same monogenic condition, regardless of the pathogenic variant(s) carried. RHDO testing is clinically available in an accredited service or in development for clinical implementation, for several rare autosomal recessive (AR) and X-linked monogenic conditions (4–9).

The RHDO methodology was first described by Lo and colleagues in 2010 and involves haplotype phasing of the high- and low-risk parental alleles followed by fetal genotype prediction (3). Haplotype phasing is carried out by next-generation sequencing (NGS) of several megabases of genomic DNA spanning the target gene in the mother, father, and a previous affected or unaffected non-carrier offspring (i.e., the proband) (Fig. 1A and B). Within the target region, common single nucleotide polymorphisms (SNPs) are identified and phased with the high- or low-risk parental alleles, relative to the status of the proband (i.e., homozygous affected or unaffected). Deep sequencing of the maternal cfDNA is carried out in parallel with the parental and proband genomic samples and the fetal genotype predicted using sensitive dosage-based measurements of the sequencing reads at the “informative SNP” loci, which are those where the parental genotypes differ (Fig. 1B, Supplemental Table 1) (3). Haplotype I refers to the parental genotype that is inherited by the proband and is used to classify the SNPs detected in the cfDNA into 5 major categories (3). For RHDO analysis, types 3 and 4 SNPs are used to ascertain fetal inheritance of the paternal and maternal alleles, respectively (Supplemental Table 1). To date, type 5 SNPs (i.e., sites where the parents are both heterozygous for the same genotype) have not been deemed useful for conventional RHDO testing (Fig. 1C, Supplemental Table 1).

Fig. 1.

Type 5 SNP SPRT RHDO approach to NIPD for autosomal recessive conditions in consanguineous families. (A, B), Haplotype phasing in consanguineous couples at risk of an AR condition by NGS; (C), type 5 SNPs: “A” and “B” are associated with the high- and low-risk alleles, respectively; (D) SPRT conducted in both directions and haplotype block calling as homozygous affected, homozygous unaffected, or heterozygous when statistical significance is reached. A representation of SPRT calls within one haplotype block is shown.

Abbreviations: hap1, haplotype 1 (SNPs associated with the maternal allele inherited by the homozygous proband); hap2, haplotype 2 (SNPs associated with the maternal allele not inherited by the homozygous proband); fwd, forward SPRT; rev, reverse SPRT; q0, null hypothesis of heterozygous fetal genotype (NGS read ratio, 0.5); q1, alternative hypothesis of homozygous fetal genotype (NGS read ratio = 0.5 + FF/2). Color figure available at https://academic.oup.com/clinchem.

The sequential probability ratio test (SPRT) is commonly used to carry out RHDO analysis on type 4 SNPs for fetal genotype prediction of the maternal allele and is the method employed in our accredited clinical service (3–9). The small but significant risk of one or more meiotic recombination events occurring within the region of interest has the potential to reduce the overall classification accuracy (10, 11). As such, SPRT is carried out incrementally along the target region in both directions, resulting in a series of haplotype blocks that are assigned to a heterozygous, homozygous high-risk, or homozygous low-risk category (Fig. 1D) (3–5, 7, 8). A deviation in expected type 3 SNP read depth or type 4 SNP haplotype block calls could indicate paternal or maternal allele recombination, respectively. Detection of meiotic recombination using RHDO has been reported previously (4, 5, 7), thus it is imperative that stringent quality control (QC) metrics are included in the analysis pipeline to avoid reporting of incorrect results to patients (7). Provided no recombination events are detected and that concordant haplotype blocks have been called on either side of the target gene, the most likely fetal genotype is determined based on the identity of the majority haplotype block call.

RHDO is not currently offered to consanguineous couples owing to the high likelihood of the parents sharing a common pathogenic haplotype, thereby significantly diminishing the number of informative SNPs available for analysis (3). In cases where there is a high degree of consanguinity, however, there is expected to be an abundance of type 5 SNPs (Fig. 1C, Supplemental Table 1). The application of type 5 SNPs for consanguineous couples, or AR conditions with a strong founder effect, has thus been proposed as a possible expansion to RHDO testing (3, 12).

Here we describe expanding access to NIPD for consanguineous families through the development and validation of a modified RHDO SPRT approach using type 5 SNPs for fetal genotype prediction.

Methods

Ethics Approval

Ethics approval was not required for this study as UK National Health Service laboratory practice allows the use of anonymized data for service development and validation.

Study Setting

Cases for this retrospective study included samples from couples seeking RHDO through the accredited UK clinical service (n = 33) and samples that had been used for validation of RHDO tests in our laboratory to allow clinical implementation (n = 8, all HBB gene cases). The indications and associated genes for which couples were carriers of a pathogenic variant included congenital adrenal hyperplasia [(CAH) CYP21A2], cystic fibrosis [(CF) CFTR], sickle-cell disease [(SCD) HBB], β-thalassaemia [(BTHAL) HBB], and spinal muscular atrophy [(SMA) SMN1]. To be eligible for testing, the couples had to be at 25% risk of having an affected child with one of these conditions, with no known close relation to each other. Even so, undisclosed consanguinity was later detected in 8 families. Cases were divided into 2 cohorts: (a) “training”/nonconsanguineous cases (n = 33) (Table 1, Supplemental Table 2) and (b) “test”/consanguineous cases (n = 8) (Table 2). The confirmed fetal genotypes were known for all cases in the training cohort and for 6 of 8 cases in the test cohort. Test case CAH_3 was lost to follow-up, and the pregnancy for test case CF_21 was terminated without further testing following a Trisomy 21 result. Case SMA_8 had an unaffected amniocentesis result; however, the polymerase chain reaction-restriction enzyme digest method used could not distinguish between an unaffected and an unaffected carrier genotype. The fetal medicine unit confirmed an unaffected postnatal phenotype for this patient. All 8 cases in the test cohort were included in this study to demonstrate the applicability of our test for consanguineous families, although performance metrics were only evaluated on the basis of the 6 known outcome cases.

Table 1.

Training cohort case information and type 5 SNP SPRT performance evaluation by condition.

Gene	Condition	Cases	Concordant with known fetal genotype	Passed type 5 SNP QC	QC passed average haplotype block class. acc. (%)	Hom low-risk	Hom high-risk	Het
CFTR	Cystic fibrosis	20	20	19	95.6	9	6	5
HBB	β-thalassaemia	3	3	3	94.5	0	0	3
HBB	Sickle cell disease	5	5	4	92.8	2	2	1
SMN1	Spinal muscular atrophy	5	5	4	91.6	2	0	3
Total/average		33	33	30	94.5	13	8	12

Abbreviations: Class. acc., classification accuracy; Hom, homozygous; Het, heterozygous.

Table 2.

Test cohort case information and type 5 SNP SPRT performance evaluation.

Case ID	Gene	Condition	Gestation (weeks + days)	Type 5 SNP SPRT RHDO prediction	Fetal genotype	Type 5 SNPs	Haplotype calls	Fetal fraction (%)	Class. acc./majority block call (%)	QC result
CF_21	CFTR	Cystic fibrosis	11 + 0	Heterozygous	Unknowna	405	38	9.2	97.4a	Pass
CF_22			8 + 3	Hom high-risk	Affected	219	6	2.9	100.0	Pass
CAH_1	CYP21A2	Congenital adrenal hyperplasia	25 + 6	Heterozygous	Unaffected carrier	94	64	14.8	82.8	Pass
CAH_2			13 + 3	Heterozygous	Unaffected carrier	152	13	9.0	84.6	Inconclusive
CAH_3			12 + 1	Hom high-risk	Unknowna	408	4	2.2	100.0a	Inconclusive
SMA_6	SMN1	Spinal muscular atrophy	12 + 2	Hom high-risk	Affected	569	128	7.2	86.7	Pass
SMA_7			unknown	Heterozygous	Unaffected carrier	355	7	4.1	100.0	Pass
SMA_8			9 + 0	Heterozygous	Unaffectedb	360	27	6.6	100.0	Pass

Abbreviations: ID, identifier; Hom, homozygous; Class. Acc., classification accuracy.

^aOutcome unknown; majority haplotype block call used to calculate classification accuracy.

^bPhenotypic outcome confirmed to be unaffected; carrier status not reported.

Fetal Genotyping Using Type 5 SNPs

Sample Processing

Sample preparation, NGS, and haplotype phasing were carried out in 2 accredited centers in England, as previously described (4, 5, 7, 13). Additional information regarding the inclusion criteria and number of SNPs targeted per capture region can be found in the Supplemental Methods and Supplemental Table 3. The “proband” sample used was obtained from a noncarrier unaffected sibling for five cases (CF_6, CF_11, CF_14, CF_19, and SCD_1), or from an invasive sample obtained from the same pregnancy for 3 cases (SCD_3, SCD_4, and SCD_5). Proband samples used for all other cases were obtained from an affected sibling.

Fetal Fraction Calculation

For conventional RHDO testing, the fetal fraction (FF) was calculated using either type 1 or type 3 SNPs (Supplemental Table 1) as previously described (3, 14). This method was used for all cases in the training cohort. As there are no type 1 SNPs obtained for consanguineous cases, a modified FF calculation approach was required for the test cohort. Each RHDO test capture panel is designed to target 3 genes simultaneously, allowing for SNPs in the other genes to be used for FF calculation. Type 3 SNPs were used for FF calculation in consanguineous cases where the paternal halotype II allele was detected in the cfDNA (i.e., the haplotype not inherited by the proband; Fig. 1B). However, in cases where the paternal halotype I allele was detected in the cfDNA, the FF was calculated using informative type 1 or 3 SNPs from the 2 other genes on the capture panel.

Data Analysis Using SPRT

An adapted RHDO SPRT R script was applied to the type 5 SNPs of all cases and analysis carried out as previously described (Fig. 1D, Supplemental Methods) (3, 7). Classification accuracy was calculated as the number of correct haplotype block calls as a percentage of the total number of calls (Supplemental Methods). The majority haplotype block call was used in the calculation in cases where the fetal genotype outcome was unknown. Conclusive results were those that had concordant haplotype blocks on both sides of the target gene and where the number of type 5 SNPs and haplotype blocks met the thresholds determined using our training cohort (Supplemental Table 4). Importantly, even if a result passes the QC metrics, more than one trained analyst is required to make the final decision as to whether or not the result is reportable to the patient (e.g., where nonconcordant haplotype block calls are made in the vicinity of the target gene).

Statistics

Unpaired two-tailed Student t tests were performed to compare the means of 2 groups at a confidence interval of 95% (α = 0.05) using GraphPad Prism software (v10.0.2) (μ = mean; σ_x̅ = standard error of the mean).

Results

Test Development Using Nonconsanguineous Cases

A revised type 5 SNP SPRT approach to RHDO for fetal genotype prediction was developed using a training cohort of pregnancies at risk of CF (n = 20), BTHAL (n = 3), SCD (n = 5), and SMA (n = 5) (Table 1). The gestational ages at sampling ranged from 9 to 19 weeks (n = 1 unreported) and fetal fractions 3.2% to 20.7% (Supplemental Table 2). All type 5 SNP SPRT fetal genotype predictions were concordant with the reported clinical RHDO test result (Table 1). Even so, applying the type 4 SNP QC metrics currently used in our accredited clinical service (Supplemental Table 4) resulted in the inconclusive classification of 6/33 (18%) cases (CF_3, CF_6, CF_15, SCD_5, SMA_1, and SMA_2).

Type 5 SNP SPRT QC Metrics

Significantly fewer type 5 SNPs (μ = 244.45, σ_x̅ = 20.43) than type 4 SNPs (μ = 615.82, σ_x̅ = 27.93) were detected on average for cases in the training cohort (P < 0.0001, data not shown). As a result, fewer haplotype blocks were called for the type 5 SNPs (μ = 75.18, σ_x̅ = 13.59) compared to type 4 SNPs (μ = 160.18, σ_x̅ = 22.71; P < 0.01, data not shown). The reduced numbers in each of these parameters necessitated the adjustment of QC thresholds for type 5 SNP SPRT result analysis (Supplemental Table 4). The performance of the type 5 SNP SPRT was assessed on a case-by-case basis and the minimum criteria required for confidence in a conclusive result determined for use in this study (Supplemental Table 4).

Type 5 SNP SPRT Performance

Conclusive results were obtained when applying the type 5 SNP QC criteria in 30/33 (91%) cases (Table 1). For those that passed the type 5 SNP QC assessment, the average classification accuracy for all indication groups was 94.5% (Table 1). Cases CF_3 and SMA_1 lacked concordant haplotype block calls on 1 side of the target gene (Fig. 2A, Supplemental Table 5), and case SCD_5 did not pass the type 5 SNP QC threshold requirements (Fig. 2A and Supplemental Tables 2 and 4). Notably, these 3 inconclusive cases were among those with the 4 lowest FF values (3.2%, 3.7%, and 4.9%, respectively; Fig. 2B, Supplemental Table 2). The case with the third lowest FF (CF_6, FF = 4.7%) had a similar number of haplotype block calls (n = 14) but a higher classification accuracy (92.6%), thus passing QC (Supplemental Table 2).

Fig. 2.

Type 5 SNP SPRT performance and fetal fraction scatter plots. Type 5 SNP classification accuracies for the “training” (A, B) and “test” (C, D) cohorts relative to the number of haplotype blocks (A, C) or fetal fraction (B, D). Grey areas delineate the QC thresholds. Orange and red datapoints indicate cases lacking concordant haplotype blocks on both sides of the gene or those that did not meet QC thresholds, respectively. Color figure available at https://academic.oup.com/clinchem.

Combined Type 4 and Type 5 SNP SPRT Analysis

Case CF_20 (gestation = 11 + 6, FF = 6.8%; Supplemental Table 2) was a repeat sample obtained after a previously inconclusive RHDO test carried out at an earlier gestation of 9 weeks and 5 days (FF = 4.3%, data not shown). Analysis of the original test data using the type 5 SNP SPRT approach results in a conclusive homozygous high-risk prediction, with a classification accuracy of 100% and 15 haplotype block calls, while the type 4 SNP SPRT result is inconclusive owing to a classification accuracy of 86.2% and 29 haplotype blocks called (Supplemental Table 4). Both type 4 and type 5 SNP SPRT results were conclusive for case CF_20.

Detection of Recombination Events

Paternal and maternal allele recombination events were detected during routine clinical testing for cases BTHAL_1 and SMA_5, respectively. However, sufficient numbers of informative SNPs and haplotype block calls were obtained between the gene of interest and the predicted recombination breakpoint in both cases for conclusive RHDO testing to be carried out. As the presence of these recombination sites was known prior to this study, the target window for our type 5 SNP SPRT analysis was limited to the gene-containing region upstream of the putative breakpoint for each case. Conclusive results were obtained for both cases (Table 1). Importantly, when our type 5 SNP SPRT analysis was applied to the entire genomic target region, a change in block identity around the predetermined recombination locus was clearly detectable in both cases (Fig. 3).

Fig. 3.

Detection of known recombination events by changes in type 5 SNP SPRT haplotype block call identity. Type 5, 4A, and 4B SNP genomic coordinates and haplotype block calls for BTHAL_1 with suspected paternal recombination downstream of HBB (A) or for SMA_5 with suspected maternal recombination downstream of SMN1 (B). Triangles and circles = forward and reverse SPRT haplotype block call loci, respectively; shaded area = gene locus; dotted line = predicted recombination site; grey datapoints = those excluded from classification accuracy calculations.

Abbreviations: Hom Hap1, homozygous for the SNP inherited by the affected proband (i.e., the high-risk allele); Het, heterozygous; Hom Hap2, homozygous for the SNP not inherited by the affected proband (i.e., the low-risk allele). Color figure available at https://academic.oup.com/clinchem.

Test Performance Evaluation Using Consanguineous Cases

Type 5 SNP SPRT Performance

The gestational ages of the cases in the test cohort ranged from 8 to 26 weeks (n = 1 unreported), with FFs from 2.2% to 14.8% (Table 2). Applying the type 5 SNP SPRT QC metrics resulted in conclusive results in 6 of 8 (75%) cases, although confirmatory testing was not available for 2 of the cases (Table 2, Fig. 2C, Fig. 4). All 6 of those with known outcomes had concordant type 5 SNP SPRT fetal genotype predictions, and 5 of these were determined to be conclusive, resulting in a sensitivity of 83.33% and a specificity of 100%, respectively (Table 2). Type 5 SNP SPRT haplotype block classification accuracy was determined to be 93.9% for all conclusive cases with known outcomes (n = 5) (Table 2). The type 5 SNP SPRT fetal genotype prediction for case SMA_8 was determined to be heterozygous, which is consistent with the unaffected invasive test result and postnatal phenotype (Table 2, Fig. 4C). All 8 test cases had concordant haplotype blocks on either side of the target gene (Fig. 4, Supplemental Table 6). Case CF_22 had a FF of 2.9% and only 1 haplotype block call downstream of the gene (Table 2, Fig. 4A, Supplemental Table 6). As the classification accuracy of CF_22 was 100% and there were a total of 6 haplotype blocks called, this case passed the QC filters (Table 2, Fig. 2C, Supplemental Table 4). The 2 inconclusive cases were at risk of CAH (CAH_2 and CAH_3; Table 2, Fig. 2C), and neither met the type 5 SNP QC threshold requirements (Table 2, Fig. 2C, Supplemental Table 4). The FF for case CAH_3 was the lowest across both cohorts at 2.2% and for CAH_2 was 9.0% (Table 2, Fig. 2D).

Fig. 4.

Visualization of type 5 SNP SPRT haplotype block calls across the genomic target region for cases within the test cohort. Type 5 SNP genomic coordinates and haplotype block calls for cases of consanguineous couples at risk of CF (A), CAH (B), or SMA (C). Triangles and circles = forward and reverse SPRT haplotype block call loci, respectively.

Abbreviations: Hom Hap1, homozygous for the SNP inherited by the affected proband (i.e., the high-risk allele); Het, heterozygous; Hom Hap2, homozygous for the SNP not inherited by the affected proband (i.e., the low-risk allele); ^$, phenotypic outcome confirmed to be unaffected, carrier status not reported. Color figure available at https://academic.oup.com/clinchem.

The Effect of Pathogenic Gene Copy Number Variants

The parental genetic aberrations for case CAH_1 involved 2 heterozygous pathogenic single nucleotide variants in one copy of the CYP21A2 gene, alongside a normal duplicated copy of the gene on the same allele. Each parent had a total of 3 copies of the CYP21A2 gene and the affected proband 4. The paternal low-risk allele was found to be inherited by the fetus as determined by the presence of type 3AD SNPs (Supplemental Table 1), and the overall type 5 SNP SPRT analysis correctly predicted the fetus to be an unaffected carrier. Interestingly, a small number of incorrect homozygous high-risk calls were made surrounding the CYP21A2 gene locus (Fig. 4B). As the fetus and mother were both carriers of the pathogenic allele, the cfDNA would have contained 6 copies of the CYP21A2 gene, thereby possibly resulting in a confounding enrichment of NGS read counts at this locus.

Discussion

Here we have developed a new approach to NIPD for couples at risk of an AR condition using SPRT on SNPs where the parents are heterozygous for the same genotype. This study included 41 families, of which 8 were consanguineous and not currently eligible for RHDO testing through the accredited clinical service. The type 5 SNP SPRT methodology was able to predict the fetal genotype accurately, regardless of the degree of shared parental ancestry, with an average classification accuracy of >93% for all cases with known outcomes (n = 39). A total of 5/41 cases were classified as inconclusive either for lacking concordant haplotype blocks on both sides of the gene (n = 2) or for not meeting the type 5 SNP SPRT QC requirements (n = 3). Importantly, stringent metrics for type 5 SNP analysis have been proposed in this study to ensure only results with the highest confidence of accuracy are reported to patients. For this reason, a repeat sample would be requested for the cases that did not meet the QC requirements and where the paternal high-risk allele has been detected, an approach shown to be effective at resolving RHDO diagnoses in the past (7). It is important to note that these are proposed thresholds and that, with increasing numbers of consanguineous cases analyzed in the future, the metrics should be reassessed to ensure the highest level of accuracy in the reported result. It is also possible that the use of type 4A SNP SPRT may support an inconclusive type 5 SNP SPRT result, although larger cohorts are required in future studies to fully investigate the utility of this. Equally, in nonconsanguineous cases with low type 4 SNP classification accuracies, including the type 5 SNP analysis may allow a reportable result. This was demonstrated in case CF_20 whereby retrospective analysis of the previously inconclusive test data using type 5 SNP SPRT provided a correct conclusive result. We therefore recommend the routine use of type 5 SNP analysis in RHDO testing to further strengthen current analysis strategies and potentially reduce the need for resampling.

We obtained conclusive RHDO results using our type 5 SNP SPRT approach even at FFs as low as 2.9% (CF_22) and 4.1% (SMA_7); however, other cases with FFs of <4.5% (n = 3) were all determined to be inconclusive (Table 2, Supplemental Table 2). The number of type 5 SNPs in these cases were all, in theory, sufficient to be able to obtain >5 haplotype block calls. The QC metrics defined in this study were therefore based on the number of haplotype blocks called (with concordance on either side of the target gene) and the overall classification accuracy rather than the FF or number of type 5 SNPs alone. Improving the sensitivity of the RHDO approach to NIPD at low FF values remains an ongoing challenge.

Our type 5 SNP SPRT analysis was able to detect the presence of both maternally and paternally inherited meiotic recombination events (Fig. 3). A current limitation to both the accredited RHDO test as well as our novel approach, however, is the inability to determine whether meiotic recombination events have occurred in the fetus or the proband that was used for haplotype phasing. Such events, however, do not pose a threat to result interpretation provided that they are reliably detected and that the region spanning the pathogenic variant(s) contains sufficient informative SNPs to obtain a statistically significant fetal genotype prediction. Even so, the risk of an undetected double recombination event in regions of low SNP coverage still exists and should be taken into consideration when counseling parents regarding this approach to prenatal diagnosis. Future development of advanced sequencing technologies, such as long- and linked-read sequencing (15–18), could provide further confidence in RHDO results and potentially increase the number of reportable cases where a recombination event has occurred at, or near, the target gene locus.

Another important potential confounding factor in RHDO analysis, regardless of the SNP type used, is the possibility of gene copy number variations resulting in incorrect haplotype block calls, thereby reducing the overall classification accuracy. This is evidenced by the incorrect homozygous for haplotype 1 (Hom Hap 1) calls made around the CYP21A2 gene locus in case CAH_1, which was likely a result of the known monoallelic gene duplication carried by the mother as well as the fetus (Fig. 4B). If the mother is a carrier of a pathogenic copy number variant, then exclusion of this region from analysis should be considered.

We have shown that type 5 SNPs can be used for fetal genotype prediction by RHDO in families that are consanguineous with a high degree of accuracy. These methods will be incorporated into the NIPD clinical service we deliver, allowing access to NIPD for monogenic AR conditions to consanguineous families. Furthermore, our approach offers additional analytical support that can be easily integrated into currently available clinical service pipelines to further improve the robustness of analysis for all couples undergoing NIPD for monogenic AR conditions.

Supplemental Material

Supplemental material is available at Clinical Chemistry online.

Nonstandard Abbreviations

cfDNA, cell-free DNA; NIPD, noninvasive prenatal diagnosis; RHDO, relative haplotype dosage analysis; AR, autosomal recessive; NGS, next-generation sequencing; SNP, single nucleotide polymorphism; SPRT, sequential probability ratio test; QC, quality control; CAH, congenital adrenal hyperplasia; CF, cystic fibrosis; SCD, sickle cell disease; BTHAL, β-thalassaemia; SMA, spinal muscular atrophy; FF, fetal fraction.

Human Genes

CYP21A2, cytochrome P450 family 21 subfamily A member 2; CFTR, cystic fibrosis transmembrane conductance regulator; HBB, hemoglobin subunit beta; SMN1, survival of motor neuron 1.

Author Contributions

The corresponding author takes full responsibility that all authors on this publication have met the following required criteria of eligibility for authorship: (a) significant contributions to the conception and design, acquisition of data, or analysis and interpretation of data; (b) drafting or revising the article for intellectual content; (c) final approval of the published article; and (d) agreement to be accountable for all aspects of the article thus ensuring that questions related to the accuracy or integrity of any part of the article are appropriately investigated and resolved. Nobody who qualifies for authorship has been omitted from the list.

Britt Hanson (Conceptualization-Supporting, Formal analysis-Lead, Methodology-Equal, Writing—original draft-Lead), Joe Shaw (Methodology-Supporting, Writing—review & editing-Supporting), Nikita Povarnitsyn (Formal analysis-Supporting, Methodology-Supporting, Writing—review & editing-Supporting), Benjamin Bowns (Data curation-Supporting, Writing—review & editing-Supporting), Elizabeth Young (Data curation-Supporting, Writing—review & editing-Supporting), Amy Gerrish (Data curation-Supporting, Writing—review & editing-Supporting), Stephanie Allen (Data curation-Supporting, Writing—review & editing-Supporting), Elizabeth Scotchman (Data curation-Supporting, Writing—review & editing-Supporting), Lyn Chitty (Conceptualization-Supporting, Data curation-Supporting, Funding acquisition-Lead, Methodology-Supporting, Resources-Lead, Supervision-Equal, Writing—review & editing-Equal), and Natalie Chandler (Conceptualization-Lead, Data curation-Lead, Formal analysis-Supporting, Methodology-Equal, Supervision-Equal, Writing—original draft-Supporting, Writing—review & editing-Equal)

Authors’ Disclosures or Potential Conflicts of Interest

Upon manuscript submission, all authors completed the author disclosure form.

Research Funding

This research is funded by the National Institute for Health and Care Research Biomedical Research Centre at Great Ormond Street Hospital, London, United Kingdom.

Disclosures

None declared.

Role of Sponsor

The funding organization played no role in the design of study, choice of enrolled patients, review and interpretation of data, preparation of manuscript, or final approval of manuscript.

Acknowledgments

We would like to thank the families for their involvement and consenting to this publication and all of the clinicians involved in the families’ care. For the testing, we thank the NIPD and Bioinformatics teams at North Thames Genomic Laboratory Hub.

Author notes

Disclaimer: The views expressed are those of the authors and not necessarily those of the National Health Service, the National Institute for Health and Care Research, or the UK Department of Health and Social Care.

Previous presentation: Poster presented at the International Society for Prenatal Diagnosis, Edinburgh, June 2023.