Expanded carrier screening for inherited genetic disease using exome and genome sequencing
Abstract
The goal of this study was to assess the feasibility of using exome (ES) and genome sequencing (GS) in guiding preconception genetic screening (PCGS) for couples who are planning to conceive by creating a workflow for identifying risk alleles for autosomal recessive (AR) and X-linked (XL) disorders without the constraints of a predetermined, targeted gene panel. There were several limitations and challenges related to reporting and the technical aspects of ES and GS, which are listed in the discussion. We selected 150 couples from a cohort of families (trios) enrolled in a research protocol where the goal was to define the genetic etiology of disease in an affected child. Pre-existing, de-identified parental sequencing data were analyzed to define variants that would place the couple at risk of having a child affected by an AR or XL disorder. We identified 17 families who would be selected for counseling about risk alleles. We noted that only 3 of these at-risk couples would be identified if we limited ourselves to the current ACMG-recommended expanded carrier screening gene panel. ES and GS successfully identified couples who are at risk of having a child with a rare AR or XL disorder that would have been missed by the current recommended guidelines. Current limitations of this approach include ethical concerns, difficulties in reporting results including variant calling due to the rare nature of some of the variants, determining which disorders to report, as well as technical difficulties in detecting certain variants such as repeat expansions.
1 INTRODUCTION
Benjamin Franklin famously said: “An ounce of prevention is worth a pound of cure” (National Archives, 1961). This proverb was initially promulgated in an effort to implement proactive fire prevention methods but has universal application. For rare genetic disorders, preconception carrier screening is one application of Dr. Franklin's proverb that has evolved over the last 50 years from a single-gene, high-risk ethnic approach to the pan-ethnic larger panel screening offered by clinical labs today.
Genetic disorders, though individually rare, are collectively common. It is estimated that there are as many as 10,867 rare diseases affecting over 400 million people worldwide and 33 million in the United States (Lamoreaux et al., 2022). Our ability to diagnose and our understanding of genetic conditions have greatly improved with the advent of next-generation sequencing (NGS). Advances in technology, foundational genomic resources and analytical tools, and access to vast amounts of genotype and phenotype data have led to an increased understanding of the etiology and biology of rare and common diseases, as well as the development of novel therapeutics (Claussnitzer et al., 2020). In addition, the cost of genetic sequencing has dramatically decreased, making direct-to-consumer testing more affordable. In 2003, the cost of sequencing the first person as part of the Human Genome Project was close to $2.7 million. Today, many commercial companies offer whole-genome sequencing for under $1000 (Wetterstrand, 2021).
Despite these advances and the federal incentives for the development of orphan drugs, treatment for genetic diseases remains very limited. According to an October 2021 US Government Accountability Office Report to Congress, there are FDA-approved treatments for around 300 unique rare diseases. Few of these treatments are cures; most are intended to address specific symptoms or lower the risk of potential complications (Dicken, 2021; Garrison et al., 2022).
The economic burden of rare diseases is substantial and includes direct, indirect, and non-medical costs. In a recent Health Affairs article, these costs approached 1 trillion dollars annually (Garrison et al., 2022; Wetterstrand, 2021). Emerging gene therapies to treat genetic disorders are costly. Etranacogene dezaparvovec-drib, a gene therapy for hemophilia B, has a one-time treatment cost of $3.5 million (Naddaf, 2022). The emotional cost and the day-to-day challenges for affected patients and families are difficult to quantify.
One approach to reducing these financial and emotional burdens is prevention through pre-conception carrier screening. This approach increases reproductive autonomy as couples learn about their risks before conception. This allows for more available reproductive options without time constraints as opposed to those whose risk is detected during pregnancy or after the birth of an affected child (Plantinga et al., 2016).
Carrier screening for genetic disorders began in the 1970s and targeted conditions with a high prevalence in specific racial/ethnic groups (Gregg et al., 2021). One successful example was screening for Tay–Sachs disease (TSD), a progressive and fatal neurodegenerative disease. The biochemical evaluation of hexosaminidase A activity was used to screen for TSD carriers in the Ashkenazi Jewish (AJ) population. Screening in the United States, Canada, and Israel was so successful that there was a greater than 90% reduction in TSD in the AJ population over a period of 10 years (Lew et al., 2015). Similar results are documented in reducing hemoglobinopathies in Middle Eastern, Southeast Asian, and Mediterranean populations (Thermo Fisher Scientific, 2023). Barriers to carrier screening include inequitable access to services, reproductive care provider's preference for smaller panels, and ancestry-based screening due to limitations in time for returning of results and follow-up (Ramdaney et al., 2022; Sagaser et al., 2023).
Screening has evolved over the last 50 years from a targeted, single-gene, high-risk ethnic approach to the larger pan-ethnic expanded panels offered by clinical labs today. In 2021, the American College of Medical Genetics (ACMG) created a list of 113 AR and XL disorders that should be offered to anyone in the general population who is pregnant or considering a future pregnancy (Plantinga et al., 2016). This list includes genes that are well characterized in ClinGen and result in a moderate to severe phenotype (as characterized by their 2020 working group). In contrast, the American College of Obstetricians and Gynecologists (ACOG) Committee Opinion developed in 2017 and reaffirmed in 2023 encourages each OB/GYN practitioner to develop their own standard approach, which may include ethnic-specific, pan-ethnic, or expanded carrier screening. Current ACOG recommendation is that all patients considering or who are already pregnant should be screened for cystic fibrosis, spinal muscular atrophy, and hemoglobinopathies. Additional screening can be offered to individuals with an indicated family history or based on specific ethnicities (Romero et al., 2017). The National Society of Genetic Counseling recommends the use of expanded or “equitable” carrier screening as replacement of ethnicity-based carrier screening but does not provide specific recommendations on the number of genes to screen (Sagaser et al., 2023).
Screening has become the standard of care in today's medicine. Screening programs have allowed for the early detection of chronic diseases and certain cancers, and, when done properly, can reduce overall morbidity and mortality. Newborn screening programs have allowed for early identification and treatment of many diseases. The list of conditions proposed by the Health Resources and Services Administration for the recommended uniform newborn screening program continues to expand as more treatments are discovered (Health Resources & Services Administration, 2024). For example, with the development of gene therapy for the early treatment of spinal muscular atrophy (SMA), the disorder was added to the Recommended Uniform Screening Panel (RUSP) list in 2019 and is now available in all 50 states and Washington, D.C (CureSMA, 2024).
Current PCGS programs have their limitations. While they account for well-established ClinGen disorders, they fail to cover the greater than 2300 AR and XL disorders that affect the rare disease community (Kirk et al., 2021). In addition, some clinical carrier screening protocols focus only on more common variants within a gene and fail to provide coverage of less common variants.
This project was inspired by a child who was diagnosed with metachromatic leukodystrophy (MLD) and her family's story. A 2-year-old child presented with signs of neurological decline over the previous 6 months. Based on her clinical examination and brain MRI findings, we suspected MLD, an incurable neurodegenerative disease caused by deficiency of the enzyme arylsulfatase A. This diagnosis was confirmed by biochemical and genetic testing; she had two pathogenic variants in the ARSA gene. Subsequent testing of both parents confirmed that they were each heterozygous for a different pathogenic variant in the ARSA gene. An asymptomatic younger infant sibling also carried both variants placing her at high risk for developing MLD. The younger sibling underwent ex vivo gene therapy followed by autologous stem cell transplantation in Milan, Italy.
The parents had pre-conception genetic screening several years earlier through a commercial lab before having their first child. Based on a 102-gene panel screening test, the couple was counseled that they were at very low reproductive risk for the tested recessive diseases, including MLD (1 in 720,000). In retrospect, this screening only targeted 5 specific variants in the ARSA gene, none of which were found in either parent.
We believe that, in this genomic era where ES and GS are first- or second-tier tests in children with undiagnosed disorders, we should be able to apply similar methods for improved expanded carrier screening. The objectives of this study are (1) to assess the feasibility of applying unbiased whole-genome sequencing technologies for unbiased preconception carrier screening in couples, and (2) to identify problem areas for data analysis and for counseling.
2 MATERIALS AND METHODS
Family trios were enrolled by the authors in a research study approved by the Western Institutional Review Board (WIRB protocol #20120789). An Exempt Research Certification form was filed with the TGen Office of Research Compliance and Quality Management stating that the data would be deidentified and repurposed for this study and that the investigator had no intention of contacting or re-identifying the subjects.
Retrospective analysis of 35 deidentified ES and 115 GS data was performed for 150 couples. The sequence data were previously generated in the above cited research study aimed at identifying the genetic cause of rare disease in a child, using a trio-based approach. For the purpose of this study, the child's data were excluded in the analysis. We combined the variant call format (VCF) files from the parents (maternal and paternal genetic data) and generated data for a “synthetic proband.” The VCF files for this trio, consisting of maternal, paternal, and synthetic proband, was imported into our standard trio template in VarSeq software (Golden Helix). The resulting variant annotation files were filtered for variants with the following parameters (Table 1): retention of variants with a Combined Annotation Dependent Depletion (CADD) score ≥15 (an in silico method used to determine the deleteriousness of SNPs and insertions/deletions), exclusion of variants of uncertain significance (VUS) that were synonymous or found in the 3′-UTR or 5′-UTR of genes, exclusion of variants with more than 5 hemizygotes or homozygotes in the Genome Aggregation Database (gnomAD), exclusion of variants scored as benign or likely benign in the ClinVar database, retention of variants classified as likely pathogenic or pathogenic by ACMG criteria (Richards et al., 2015), and retention of variants only in genes associated with known AR or XL disorders listed in Online Mendelian Inheritance in Man (OMIM) (see Table 1). The retained variants were then manually reviewed to identify risk alleles in the parents that lead to AR (homozygous or compound heterozygous) or XL pathogenic variants in the “synthetic proband.”
| Filtering criteria | Exclusion parameters |
|---|---|
| Deleteriousness of the variant based on in silico analysis | Variants with a CADD score less than 15 were excluded |
| Variants of uncertain significance | VUS that were synonymous or found in the 3′-UTR and 5′-UTR of genes |
| Population frequency of variant | Variants with more than 5 hemizygotes or homozygotes in gnomAD |
| Clinical significance of variant | Variants scored as benign or likely benign in the ClinVar database |
| Associated disease/disorder description | Variants without an OMIM disease/disorder description |
3 RESULTS
From the 150 couples whose genomic data was analyzed, 17 were identified to be at risk of transmitting an autosomal recessive or X-linked inherited disease (see Table 2). Of significance, only three genes from this list are in the ACMG-recommended pan-ethnic screening panel.
| Couple number | Gene name | Disorder | Mode of inheritance | ACMG panel |
|---|---|---|---|---|
| 1 | TMCO1 | Cerebro-facio-thoracic dysplasia | AR | No |
| 2 | COL27A1 | Steel syndrome | AR | No |
| L1CAM | L1 syndrome | X-linked | Yes | |
| 3 | HEXA | Tay–Sachs disease | AR | Yes |
| 4 | PDK3 | Charcot–Marie–Tooth disease | X-linked | No |
| 5 | ABCA4 | Retinal dystrophy early-onset, severe | AR | No |
| 6 | SMS | Snyder–Robinson syndrome | X-linked | No |
| 7 | OTOA | Deafness | AR | No |
| 8 | GALC | Krabbe disease | AR | No |
| 9 | SLC5A5 | Thyroid dyshormonogenesis 1 | AR | No |
| 10 | TNFRSF13B | Common variable immunodeficiency (CVID) | AR | No |
| 11 | OTOA | Deafness | AR | No |
| 12 | LAMA3 | Epidermolysis bullosa | AR | No |
| OTOA | Deafness | AR | No | |
| 13 | CYP27B1 | Vitamin D-dependent rickets | AR | Yes |
| 14 | OCRL | Dent disease-2 | X-linked | No |
| 15 | TAF8 | Neurodevelopmental disorder with severe motor impairment, absent language, cerebral hypomyelination, and brain atrophy (NEDMLHB) | AR | No |
| 16 | GEMIN5 | Neurodevelopmental disorder with cerebellar atrophy and motor dysfunction (NEDCAM) | AR | No |
| 17 | DNAH11 | Primary ciliary dyskinesia-7 (CILD7) | AR | No |
Seventeen genes were identified with pathogenic or likely pathogenic variants. In 14 couples, both parents were heterozygous for variants linked to an AR disorder. Four female parents were heterozygous for variants connected to an XL disorder. Disorders included intellectual disability (TMCO1, L1CAM, HEXA, SMS, GALC, TAF8, GEMIN5), skeletal problems (COL27A1, SMS, CYP27B1), degenerative conditions (HEXA, GALC), peripheral neuropathy (PDK3), eye conditions (ABCA4), deafness (OTOA), respiratory problems (DNAH11), endocrine disorder (SLC5A5), skin conditions (LAMA3), immunodeficiency problems (TNFRSF13B), and kidney disorders (OCRL).
4 DISCUSSION
4.1 Limitations/challenges
While some of the limitations and challenges of using ES and GS for PCGS were more technical in nature, there are also ethical issues, which require additional research. Many of the challenges we noted were related to reporting. For AR disorders, we only reported variants that were found in the same gene in both parents. For instance, if one parent was heterozygous for a variant in CFTR but the partner was not, the variant would not be reported. This can be problematic if one of the partners leaves the relationship and is unaware of their carrier status, which has been a criticism of couple-based carrier screening programs in the past (Grody et al., 2001; Wald, 1991). We also recognize that there is a selection bias for our study population as it is not a true representative of the general population of reproductive-age adults and was generated as a part of a pre-existing study of children with rare disorders.
There may not be a good genotype–phenotype correlation for all variants. Determining which variants to report by phenotype was a challenge and raised additional questions: Do we report variants for mild disease, adult-onset, variable penetrance, deafness, or sex determination? Reporting the data by phenotype raises serious concerns regarding discrimination against those with disabilities and has the potential to lead to the practice of eugenics. For the purpose of this initial study, we chose to exclude variants in genes where the only phenotype was infertility but included variants associated with hearing loss. We recognize that genome-wide screening for all known AR and XL disorders may be controversial. For example, 80% of nonsyndromic genetic forms of hearing loss are caused by biallelic variants (Shearer et al., 1999). Individuals in the Deaf community may define deafness as an “audiological anomaly” rather than a disability and consider this “a form of human variation” “more akin to being a member of a culturally and linguistically diverse group.” (Freeman et al., 2022) Some within the Deaf community have said: “allowing the use of genetic testing for deafness in the reproductive setting expresses a negative view of deafness” (Freeman et al., 2022). This also could be said of other types of disabilities, and it is important to keep in mind the perspectives of these groups. In the context of this paper, we believe it is important to keep the focus on reducing the burden of genetic disease and not diverge into singling out specific traits, sex, or other problems that can be seen as eugenic-like. We recognize that PCGS combined with genetic counseling are geared towards providing the best available information to couples to allow them to make reproductive decisions as they feel appropriate. The ideas of severity of disorder, degree of disability, what is acceptable, all vary from couple to couple, and can be influenced by cultural and religious background and values.
There is the possibility of giving back false-positive results when variants are classified as pathogenic or likely pathogenic by ACMG standards but not historically seen in affected patients. We did not report back variants of uncertain significance (VUS) or the probability of false-negative results, which could give a false sense of security. Likewise, if one partner had a VUS but the other had a P/LP variant, the risk of inheriting the disorder may be missed. ACMG recommends that such a situation be discussed during the pre-test counseling with the option of reporting back VUSs when such a situation arises (Plantinga et al., 2016). By including over 6335 disorders in our analysis, this increases the likelihood of returning results for ultrarare disorders. The unfamiliarity with such disorders may create a large burden for a couple as well as for the provider. The clinical burden is changing for a number of disorders due to the approval of novel gene- or variant-based therapeutics (e.g., SMA, CF). If a previously devastating disorder becomes treatable, how will it affect the way in which we deliver prenatal counseling? (Massie et al., 2014).
In addition, there are several AR disorders, like Fanconi anemia and Gaucher disease, or XL disorders, like Fabry disease, where heterozygotes may be affected, show mild symptoms, or are at an increased risk of developing later-onset disease. With all the issues that were identified, non-directive genetic pre- and post-test counseling is needed to provide education and unbiased information.
4.2 Technical challenges
We also noted many technical challenges, some of which have been previously reported. Certain diseases (spinal muscular atrophy, hemoglobinopathies, Fragile X syndrome, congenital adrenal hyperplasia, Gaucher disease) with a high prevalence have known technical challenges with NGS and require alternative methods to diagnose. Disorders that are caused by tandem repeat expansion, such as Fragile X, are not reliably revealed by NGS (Beauchamp et al., 2018). However, commercial clinical laboratories are working on software tools to tackle this issue (Chen et al., 2020). ES targets only the coding exons plus ~10 bp of flanking non-coding DNA for each exon. Unless specifically indicated, no information about other portions of the gene, such as regulatory domains, deep intronic regions, uncharacterized alternative exons, and chromosomal rearrangements are provided. Many of these issues can be addressed through GS. Furthermore, our ability to detect CNVs due to somatic mosaicism is restricted. It is difficult to differentiate pseudogenes, such as those for GBA and CYP21A2, and to detect structural changes such as the intronic inversion in the F8 gene as seen in hemophilia A, and to meaningfully interpret non-coding variants in GS (Shearer et al., 1999). Mitochondrial DNA requires deep sequencing to determine heteroplasmy levels and as such was not explored in this study.
4.3 Conclusion
This study shows that it is feasible to use ES and GS to develop a pre-conception genetic screen for guiding couples contemplating having a child. The strength of this screening is the ability to provide couples with information on the risk of having a child with one of over six thousand different disorders. As shown by our analysis, many of these disorders would be missed by ACMG or ACOG's current expanded screening recommendations. Limiting the number of disorders on PCGS can be viewed as paternalistic. The argument against including such large numbers of disorders is that providers may feel overwhelmed due to the rarity of some disorders where it may be difficult to counsel couples when the prognosis and natural history of such diseases are limited or unknown. There is also concern that using such an approach could lead to the practice of eugenics. Some in the disability community have raised concerns stating: “using genetic medicine to select against severe impairments will inevitably lead to lower tolerance of minor variations, and ultimately to the rejection of any deviation from the phenotypic norm (or the phenotypic ideal), if genetic enhancements are permitted” (Scully, 2008). Other concerns are who will provide pre- and post-screening counseling. The couples' primary care physician or obstetrician/gynecologist may have limited understanding of the limitations of the screening or how to interpret the results. To address the limited availability of genetic counseling, interactive computer programs have been created to provide pre-test counseling and are shown to be a valuable adjunct to genetic counseling (Green et al., 2005). Additional studies are needed to determine health care providers' willingness to use GS PCGS in their clinic and to determine barriers to implementation. Likewise, additional studies are required with the general public regarding the ethics surrounding PCGS and to determine the emotional impact on and acceptance by couples of receiving results of extremely rare conditions and disorders with a mild phenotype, variable phenotype, late-onset, variable penetrance, deafness, and sex determination.
Our study also highlights the limitations of current pre-conception gene panels. Special analysis tools are required for conditions caused by technically challenging classes of variants (including fragile X and other repeat expansion disorders, deletions/insertions, and others). In cases where the limitations of GS cannot be overcome, alternative diagnostic methods may be needed. An example of this is Friedreich ataxia, where the majority of expanded alleles contain between 600 and 1200 GAA repeats and Southern blot and PCR are used for diagnosis (Bidichandani & Delatycki, 1998; GeneDx, 2024; LabCorp, 2024).
The decision for couples to undergo PCGS is complex. This may start with identifying a clinician who may initiate this process (primary care or OB-Gyn physician) and issues of cost or insurance coverage for testing. Counseling and informed consent is necessary and couples must understand the risks, benefits, and consequences of screening. Pre-test counseling should include a non-biased discussion of the following issues: (a) an understanding of current limitations of PCGS; (b) that a negative screening result does not eliminate the possibility of having a child affected by any condition screened, but it does reduce this risk; (c) a discussion of reproductive options available to couples at risk. In addition, as genetic therapies become available for more genetic disorders, preconception counseling will need to include discussions of the costs and side effects of treatments if the couple decides to conceive and the pregnancy results in an affected child. For example, as with SMA, gene therapy may not be a cure, as the neurodegeneration that occurs prior to treatment is not reversible; thus, couples will need to know these facts before making reproductive decisions. With appropriate counseling, genome-level carrier screening will allow couples to make appropriate reproductive decisions for their family based on thousands of disorders, with a monetary cost that is much less than the available gene therapies.
AUTHOR CONTRIBUTIONS
Conceptualization: K. Ramsey, S. Szelinger, W. W. Grody, V. Narayanan; Data curation: A. Abraham, S. Rangasamy, A. Bonfitto, M. Naymik, M. Huentelman, S. Szelinger; Formal analysis: N. Belnap, K. Ramsey, A. Abraham, A. Ryan, S. Rangasamy, A. Bonfitto, M. Naymik, M. Huentelman, S. Strom, D. Perry, A. Subramaniam, W. W. Grody, S. Szelinger, V. Narayanan; Investigation: N. Belnap, K. Ramsey, A. Abraham, A. Bonfitto, A. Subramaniam, W. W. Grody, V. Narayanan; Writing-original draft: N. Belnap, K. Ramsey, V. Narayanan; Writing-review & editing: N. Belnap, K. Ramsey, A. Abraham, A. Ryan, S. Rangasamy, A. Bonfitto, M. Naymik, M. Huentelman, S. Strom, D. Perry, A. Subramaniam, W. W. Grody, S. Szelinger, V. Narayanan.
ACKNOWLEDGMENTS
We would like to thank the families that provided samples making the analysis of NGS possible.
FUNDING INFORMATION
This project was supported by a Flinn Foundation Accelerator for Medical Technology Seed Grant.
CONFLICT OF INTEREST STATEMENT
The authors declare no conflict of interest.
ETHICS STATEMENT
Human studies and informed consent: The study was part of a research study (dcraig12-003 Genetic Studies of Patients and their Families with Diseases of Unknown Genetic Etiology) approved by the Western Institutional Review Board (WIRB protocol #20120789). Informed consent was obtained from all participants as required by the IRB. An Exempt Research Certification form was filed with the Translational Genomics Research Institute's (TGen) Office of Research Compliance and Quality Management stating that the data would be deidentified and that the investigator had no intention of contacting or re-identifying the subjects.
Animal studies: No non-human animal studies were carried out by the authors for this article.
Open Research
DATA AVAILABILITY STATEMENT
The data supporting this study's findings are available from the corresponding author on reasonable request.
