Succinic semialdehyde dehydrogenase is the final of two enzymatic steps in the catabolism of gamma-aminobutyric acid (GABA) (Pearl et al., 2019). An NAD + -dependent enzyme, SSADH converts succinic semialdehyde to succinic acid prior to its entry into the citric acid cycle (Figure 1). Loss of enzyme activity leads to an excess of succinic semialdehyde which itself may be converted to gamma-hydroxybutyric acid (GHB or 4-hydroxybutyric acid) by the enzyme 4-hydroxybutyrate dehydrogenase. Significant elevations in GHB have been observed in the blood, urine, and cerebrospinal fluid of affected individuals, which is how patients were initially diagnosed with the disorder (Gibson et al., 1990; Struys et al., 2005). However, some reports describe individuals with molecularly and enzymatically confirmed SSADHD and normal or only mildly elevated GHB excretion, as well as elevated GHB in individuals with reportedly normal enzyme activity (Bonham et al., 1994; Pitt et al., 1997; Gibson et al., 1998). A recent study also highlighted that while the level of GHB decreases with age in healthy individuals, the same effect may also be observed in affected individuals (Brown et al., 2019; Brown et al., 2020). Given these findings, GHB may not always serve as an accurate biomarker, thereby necessitating further genetic or enzymatic analysis in cases strongly suspected of having SSADHD. Additionally, while SSADH enzyme assays have historically been utilized in research studies, such testing is not currently widely available for clinical diagnosis (Gibson et al., 1991; Gibson et al., 1994; Hogema et al., 2001).

While the exact patho-mechanism of SSADHD remains unclear, it is commonly assumed that most neurologic symptoms are due to elevations in GABA due to its role as the primary inhibitory neurotransmitter via GABA_A and GABA_B receptors (Reis et al., 2012; Tang et al., 2021). GABA signaling is also known to be critical for developmental neuronal migration and synaptogenesis, as well as playing an important role in neuronal stem cell proliferation, which may explain some of the early onset developmental delays in patients with SSADHD (Wang and Kriegstein, 2009; Pontes et al., 2013). As a GABA metabolite, GHB is thought to act as a neuromodulator at endogenous concentrations and has also been shown to modify the release of other important neurotransmitters like GABA, glutamate, and dopamine (Maitre et al., 2016; Felmlee et al., 2021). GHB exerts its effects by binding to both its own unique receptor (SLC52A2, also called GHBh1), as well as GABA_B receptors (Andriamampandry et al., 2007; Yao et al., 2010; Absalom et al., 2012; Bay et al., 2014; Maitre et al., 2016) and at pharmacologic doses, is known to cause euphoria, drowsiness, and a loss of inhibition. Because of these effects, GHB has been exploited both as a treatment for conditions like narcolepsy and as a drug of abuse, through its effects on dopamine as mediated by its GABA_B receptor binding (White, 2017; Felmlee et al., 2021). Less clear, however, is what role the molecule may play in terms of neuronal development and why individuals with SSADHD do not exhibit the same symptoms as those taking the drug illicitly.

Clinically, patients with SSADHD usually present within the first few years of life with developmental delays, particularly with delayed, limited, or deficient speech acquisition (Pearl et al., 2003b; Pearl et al., 2009). Other early symptoms may include hyporeflexia, ataxia, autism spectrum disorders, sleep disorders, and behavioral disturbances (including anxiety, obsessive-compulsive disorder, inattention/hyperactivity and sometimes aggression) (Pearl et al., 2003b; Gibson et al., 2003; Knerr et al., 2008; Pearl et al., 2009). Medical histories may also reveal early features like neonatal hypotonia, poor feeding, and respiratory distress. Seizures may develop in up to 50% of affected patients, with increasing prevalence with age (Pearl et al., 2011; DiBacco et al., 2019; Pearl et al., 2021). These seizures may be characterized by a variety of semiologies, including generalized tonic-clonic, absence, focal, and in some cases, myoclonic seizures (Pearl et al., 2011). Electroencephalographic (EEG) findings include background abnormalities/slowing, spike discharges, electrographic status epilepticus during slow wave sleep, and photosensitivity (Pearl et al., 2011). Magnetic resonance imaging studies in affected patients have identified increased T2-weighted signal in the globus pallidus, subcortical white matter, subthalamic nuclei, cerebellar dentate nuclei, and brainstem, as well as cerebral and cerebellar atrophy, and delayed myelination (Yalcinkaya et al., 2000; Ziyeh et al., 2002; Afacan et al., 2021). Metabolic strokes have also been rarely described in the setting of acute illnesses (Wang et al., 2016; Yoganathan et al., 2020; Yoganathan et al., 2021). Magnetic resonance spectroscopy studies have identified increased GABA concentrations, as well as elevations of GHB and homocarnosine (Al-Essa et al., 2000; Ethofer et al., 2004; Afacan et al., 2021).

Currently, there is no specific treatment for SSADHD, and most therapies focus on management of symptoms. Seizures, for instance, are treated with standard therapies, though valproate is usually avoided due to its inhibition of residual SSADH enzyme activity (Shinka et al., 2003). Vigabatrin therapy has been trialed extensively due to its known role as an inhibitor of GABA transaminase (Tolman and Faulkner, 2009). While the drug indeed does lead to an increase in GABA and noncommittal decrease in GHB, clinical response to this therapy has been largely mixed with some milder patients exhibiting a positive response while others experience little to no change in symptoms (Gibson et al., 1989; Gibson et al., 1995; Matern et al., 1996; Gropman, 2003; Leuzzi et al., 2007; Vogel et al., 2013). Attempts to improve upon the effects of vigabatrin led to the discovery and development of (3-aminopropyl) (n-butyl)phosphinic acid (known as SGS-742), another GABA_B receptor antagonist. Following pilot studies in mouse models, a human randomized control of the drug was conducted, and though the drug did not cause an increase in adverse events, there was no discernible improvement in cognition or cortical excitability (Schreiber et al., 2021). The amino acid taurine has also been trialed as a potential treatment for SSADHD based on its postulated role as an alternative substrate for the GABA transporter and its abilities to modulate ER-stress (Pearl et al., 2014; Jakaria et al., 2019). In an open-label study, however, there were no statistically significant improvements observed in adaptive behavior, no changes in cerebrospinal fluid (CSF) GABA concentrations, and paradoxically increased cortical excitability on paired-pulse transcranial magnetic stimulation (Pearl et al., 2014; Schreiber et al., 2016). Attempts to develop newer and more efficient therapies continue to this day along with the optimization of supportive therapies (Nylen et al., 2008; Pearl et al., 2009; Vogel et al., 2013; Vogel et al., 2018).

Data from all referenced cases (1 through 16, also referred to as BG-BCM cohort) presented here were derived from studies conducted under clinical diagnostic testing at Baylor Genetics Laboratories (Houston, TX, United States) and Baylor College of Medicine (BCM) Institutional Review Board (IRB) approved protocols H-32701 and H-35388.

EDTA plasma and urine samples from subjects 17 through 24 were obtained from the Succinic Semialdehyde Dehydrogenase (SSADH) Deficiency Biorepository, Rosamund Stone Zander Translational Neuroscience Center, Boston Children’s Hospital (Boston, MA, United States). This study was approved by the Boston Children’s Hospital Institutional Review Board approved protocol IRB 09-02-0043.

Molecular testing for the remaining cases in the BG-BCM cohort included either full gene Sanger sequencing, or exome sequencing performed at Baylor Genetics Laboratories. Molecular testing for cases 11 and 13 was performed at MNG Laboratories (Atlanta, GA, United States).

EDTA-plasma or urine samples were stored in −80 deg C freezer. Pre-analytically, upon collection of blood sample in an EDTA tube, plasma was separated and frozen, and remained frozen until testing. Thawed samples were not accepted for testing. Plasma with evidence of hemolysis was not accepted for testing. Similarly, urine samples were only accepted when received frozen. Samples were kept on ice through the pre-analytical phase of sample handling, processing, and extraction.

Clinical untargeted metabolomic profiles were generated from EDTA plasma derived from venous whole blood and urine obtained from five subjects–case numbers 2, 11, 13, 14, and 15 in the BG-BCM Cohort. To further validate metabolomic findings from the BG-BCM cohort, we acquired additional plasma and urine samples from known individuals with SSADHD (cases 17 through 24) from the SSADH Biobank. Urine samples were normalized per creatinine levels. Metabolomic profiling was performed at Baylor Genetics (Houston, TX, United States) and Metabolon, Inc. (Morrisville, NC, United States) using liquid chromatography mass spectrometry (LC/MS) analysis as previously described (Miller et al., 2015; Glinton et al., 2018; Alaimo et al., 2020; Ford et al., 2020; Liu et al., 2021). Small molecules were extracted from 100 µL of sample in an 80% methanol solution, and samples were run on four independent platforms, each able to measure a different set of metabolites: (1) positive ionization with Waters BEH C18 chromatographic separation of hydrophilic compounds (LC/MS/MS Pos Polar), (2) positive ionization with Waters BEH C18 chromatographic separation of hydrophobic compounds (LC/MS/MS Pos Lipid), (3) negative ionization with Waters BEH C18 optimized conditions (LC/MS/MS Neg), and (4) negative ionization with Waters BEH Amide (HILIC) chromatography (LC/MS/MS Polar). All chromatography was performed using a Waters Acquity UPLC (Waters, Milford, MA) held at 40°C–50 °C. Deuterated internal standards, including 2-pyrrolidinone were used to monitor performance and serve as retention index markers and chosen based on their broad chemical structures, biological variety, and their elution spectrum on each of the arms of the platform. The chemical structures of known metabolites were identified by matching the ions’ chromatographic retention index, nominal mass, and mass spectral fragmentation signatures with reference library entries created from authentic standard metabolites under the identical analytical procedure as the experimental samples. Semiquantitative analysis of metabolites in each sample was achieved by comparing the individual patient sample to a set of invariant anchor specimens included in each batch. Raw spectral intensity values were then normalized to the anchor samples, log transformed, and compared to a normal reference population to generate Z-scores. Results were considered abnormal if the z-score for a compound was equal to or greater than two standard deviations above or below the mean of the control reference population (n = 395). The metabolomic signature of SSADHD was identified based on characteristic perturbations in the concentrations of GABA metabolites and precursors including 2-pyrrolidinone, 4-guanidinobutanoate, succinamic acid, and succinimide (Figure 2) (Kennedy et al., 2019). Simple linear regression analysis of metabolite z-scores was performed to assess their association with age of the affected individuals. The average metabolite z-score of pediatric and adult age-groups were compared statistically using a two-tailed Student’s t-test. Complete metabolomic profiles are provided in Supplementary Table S5 for each case.

ALDH5A1 pathogenic variants were collated from the historical literature, ClinVar (https://www.ncbi.nlm.nih.gov/clinvar/), and an internal SSADH Deficiency Association database (https://www.ssadh.net) to provide the most complete assessment of disease-causing variants in ALDH5A1. Variants were excluded if they could not be confirmed and validated in the Allele Registry (http://reg.clinicalgenome.org/redmine/projects/registry/genboree_registry/landing) or in other human genome reference databases, including ClinVar (https://www.ncbi.nlm.nih.gov/clinvar/) and ClinVar Miner (https://clinvarminer.genetics.utah.edu). The Allele Registry is a genome integration database supported by the ClinGen Project (https://clinicalgenome.org). Nomenclature for all variants follows HGVS standards.

Population-based exome and whole genome data from the Genome Aggregation Database (gnomAD) were interrogated for ALDH5A1 pathogenic/likely pathogenic variants reported as disease-causing variants. All ALDH5A1 variants listed in gnomAD v4.0.0 (https://gnomad.broadinstitute.org) were curated to determine predicted pathogenicity. Two primary transcripts for ALDH5A1 are reported: (1) the canonical transcript, ALDH5A1 NM_17040, which contains 11 exons, and (2) ALDH5A1 NM_001080, which contains 10 exons, is the MANE Select transcript, and is represented in gnomAD v4.0.0. The transcript NM_17040 differs from NM_001080 by the presence of an additional in-frame exon with 13 amino acids in the middle of the protein. Differences in function and expression of these two transcripts are not described.

Protein truncating variants (including frameshifting variants, stop-gains, start-loss, splice site variants at +/-1 and +/-2 location) which are present in gnomAD but not yet reported to be associated with SSADHD in literature or ClinVar database were classified as likely pathogenic. All known pathogenic or likely pathogenic variants that have been reported to be associated with SSADHD either in ClinVar database, published patient data, confirmed variants from our cohort, and the SSADHD Biobank were also identified. Taken together, these novel and known likely pathogenic or pathogenic variants in the general population represented in gnomAD v4.0.0 were used to calculate carrier frequency and to estimate disease prevalence in this population. Additional measures of carrier frequency and disease prevalence were performed by applying stringent computational predictions of variant deleteriousness with a CADD score cut-off value of ≥28.0 for nonsynonymous variants and SpliceAI score (≥0.2) for potentially spliceogenic variants.

Carrier frequency was assessed by determining the minor allele frequency of all pathogenic/likely pathogenic variants across all populations present in the publicly accessible gnomAD database v4.0.0 (https://gnomad.broadinstitute.org/). The minor allele frequency (MAF) of each variant was used to determine overall carrier frequency and disease prevalence, as described previously (Cappuccio et al., 2019; Elsea et al., 2020; Martin et al., 2021). Disease prevalence estimates were determined based on cumulative ALDH5A1 pathogenic/likely pathogenic allele frequency within this database. Data were assumed to be in Hardy-Weinberg equilibrium, and standard Hardy-Weinberg calculations were used to determine carrier frequency and disease.

During the period 2014–2022, we assessed 5,000 plasma, urine, and CSF untargeted metabolomic samples at our laboratory. During the same period, we assessed a total of 17,000 Sanger, exome, or genome sequencing samples. Collectively, we considered these samples as our clinically referred population of 22,000 samples. Among these, we assessed the number of individuals confirmed to be diagnosed with SSADHD and GABAT-D to determine the estimated disease frequency in a clinically referred population.

We identified 16 individuals in the BG-BCM cohort (Cases 1–16, Table 1) with possible or known SSADHD. In addition, we also acquired plasma and urine samples from eight unrelated individuals from the Boston Children’s Hospital SSADHD Biobank (Cases 17–24, Table 1). Of the 24 individuals, 10 (41.6%) were males and 14 (58.3%) were females. At the time of genetic testing or Biobank sample collection and analysis, participant ages ranged from 6 months to 44 years, with a mean age of 11.5 ± 12.66 years and median age 7 years. Among these, 18 individuals were in the pediatric age-group ranging from 6 months to 12 years and six individuals were in the adult age-group ranging from 22 years to 44 years. Among the BG-BCM cohort, the phenotypic presentation or most common reason for referral included developmental delays and hypotonia, while ataxia, seizures, abnormal brain MRI, autistic features, failure to thrive, or stroke were indicated in others (Table 1).

The individuals in the BG-BCM Cohort (Cases 1–16) were identified by surveying over 17,000 clinical exome sequencing and Sanger sequencing databases for cases carrying biallelic, pathogenic, or likely pathogenic variants in ALDH5A1 at Baylor Genetics. Novel variants were curated based on current ACMG/AMP variant interpretation criteria (Richards et al., 2015). This comprehensive review of sequencing data identified 14 individuals with biallelic variants in ALDH5A1 (Table 1). In two individuals (Cases 11 and 13), molecular sequencing was carried out at another laboratory with sequencing information provided to aid in the interpretation of metabolomic analysis. In these instances, the finding of biallelic pathogenic or likely pathogenic variants in ALDH5A1 in the setting of characteristic clinical features and/or elevated levels of GHB were consistent with the diagnosis of SSADHD. The clinical metabolomics database of the laboratory was also surveyed for cases with a significant pattern of perturbations in GABA metabolism analytes. When available, corresponding molecular and metabolomic data were combined with provided clinical information to confirm or rule out SSADHD as a clinical diagnosis. SSADHD Biobank samples (Cases 17–24) were from patients with previously confirmed SSADHD and are shown with ALDH5A1 genotypes and urine GHB level at the time of evaluation (Table 1).

Among the 16 identified cases in the BG-BCM cohort, four individuals harbored homozygous pathogenic or likely pathogenic variants. A total of 19 unique variants were identified in our study, 3 (15.7%) of which are novel. All three variants were classified as pathogenic [c.111_122delinsG (p.Ala38GlyfsTer94); c.768_784del (p.Ile257GlufsTer12); and c.380G>A (p.Trp127Ter)] given that they are predicted to lead to premature translation termination (Supplementary Table S3). A summary of known disease associated variants in SSADHD in BG-BCM cohort and those reported in ClinVar database have been shown for an overview (Figure 3).

ALDH5A1 variants were considered pathogenic (P) or likely pathogenic (LP) if the nucleotide change (1) resulted in a premature termination codon (stop-gain), (2) occurred at a splice site (within +/- 0–2 bp from reported, canonical splice site), (3) resulted in a frameshifting effect on the amino acid sequence, (4) caused loss of initiation codon, or (5) was previously reported in a confirmed, clinically affected individual with SSADH deficiency. Stop-loss variants were considered variants of uncertain significance and not included in the prevalence analysis. Variants were also excluded from prevalence estimates if determined to be benign/likely benign or if frequency in populations was greater than expected for disease. It is known that SSADHD disease is underdiagnosed and as such, is under-reported. Therefore, to better assess the status of disease prevalence and population carrier frequency, in addition to the above-described criteria for variant inclusion, ALDH5A1 variants were further assessed using stringent CADD score prediction value of ≥28, including SpliceAI with a minimum score of >0.20 (https://cadd.gs.washington.edu). These cut-off values were determined by assessing known disease-causing pathogenic variants within these prediction programs and are inclusive of known pathogenic missense variants that are also present in gnomAD. The PHRED score of a reported pathogenic missense variant in ALDH5A1 present in gnomAD v4.0.0 was 20.7–35, with a median CADD score of 25.7. Therefore, a high value of CADD score ≥28.0 was chosen as a best measure to ensure high stringency in selection and inclusion of potentially deleterious but unreported variants in the general population. In addition, SpliceAI score ≥0.2 was also applied, and those variants were selected to include potentially spliceogenic variants for carrier and disease prevalence estimates. This stratified assessment allowed for evaluation of disease prevalence, not only based on strictly known pathogenic/likely pathogenic variants but also on predicted deleterious variants for better estimates of the prevalence boundary.

Variants for assessment and subsequent prevalence calculations included both reported disease-associated ALDH5A1 variants identified in individuals with documented SSADH enzyme deficiency (Supplementary Table S1), as well as pathogenic/likely pathogenic variants present in the populations represented in gnomAD (Supplementary Table S2). Carrier frequency was assessed by determining the minor allele frequency of these variants (Supplementary Tables S1, S2) across all populations present in the publicly accessible gnomAD databases. The minor allele frequency (MAF) of each variant was used to determine overall carrier frequency and disease prevalence, as described previously (Cappuccio et al., 2019; Martin et al., 2021). Disease prevalence estimates were determined based on the cumulative ALDH5A1 pathogenic/likely pathogenic allele frequency. Data were assumed to be in Hardy-Weinberg equilibrium. Standard Hardy-Weinberg calculations were used to determine carrier frequency and disease prevalence. Our analysis estimates that the pan-ethnic prevalence of SSADHD based on reported disease-causing variants and loss-of-function variants in the general population database in gnomAD v4 is one in 564,000. However, when CADD-score based predictions of deleteriousness (CADD score ≥28.0) or splice AI (≥0.2) were applied, the pan-ethnic prevalence was estimated to be as high as one in 223,000 (Table 2).

Using these pathogenicity prediction estimates, among the subpopulations defined based on genetic ancestry in gnomAD v4.0.0, the African/African American subpopulation revealed the highest fold-increase in estimated disease prevalence of 13-fold when compared to estimated prevalence measured with strictly known pathogenic/likely pathogenic or truncating variants in ALDH5A1. Admixed American, East Asian, European (non-Finnish), European (Finnish), South Asian, and ‘remaining subpopulations’ revealed only ∼1.3- to 3.2-fold increase in estimated disease prevalence (Table 2). Amish alleles were not identified; thus, since the allele numbers are limited in gnomAD, conclusions may not be derived regarding disease prevalence in this subpopulation. Similarly, limited numbers of individuals are present in gnomAD representing both the Ashkenazi Jewish and Middle Eastern populations, which prevent firm conclusions regarding prevalence in these subpopulations.

Our analysis estimates that the pan-ethnic prevalence for SSADHD could lie somewhere between 1/223,000 to 1/564,000 based on population allele frequency data from the gnomAD v4.0.0 database (Table 2). This finding is consistent with our previously reported estimate of ∼1/460,000 worldwide (Martin et al., 2021).

Among the clinically referred population of 22,000, we identified six individuals with confirmed GABA-TD and 16 individuals with SSADHD suggesting that the disease frequency of SSADHD in this clinically referred population at our laboratory is one in 1,375 and that of GABA-TD is one in 3,500. The cumulative disease frequency of these GABA metabolism disorders in this clinically referred population at our laboratory during the time period from 2014 to 2022 is estimated to be ∼1/1,100.

Plasma untargeted metabolomic assays were obtained for a total of 13 individuals - five individuals from the BG-BCM Cohort (Cases 2, 11, 13, 14, and 15, Table 1) and eight individuals from the SSADHD Biobank (Cases 17–24; Table 1) and compared to the normal control population. These biomarker profiles were also compared to data generated from individuals with GABA transaminase deficiency (GABA-TD), a disorder that also results from altered GABA metabolism in the brain (Kennedy et al., 2019).

Among pediatric cases (ages ≤18 years, n = 8), all samples (100%) showed significant elevations of both 2-pyrrolidinone (Z-score range +3.12 to +6.50) and 4-guanidinobutanoate (Z-score range +2.61 to +4.03) compared to the control cohort (n = 395) (Figure 4A; Supplementary Table S4). In addition, argininate, a guanidino-group containing derivative of arginine, was elevated in four of seven (57%) pediatric samples (and not reported in one pediatric sample tested on an older version of the LC-MS/MS platform). Among the adult individuals (ages ≥19 years, n = 5), four (80%) showed normal 2-pyrrolidinone levels (Z-score range +0.02 to +1.24), with the exception of one sample (case 22) that showed mildly elevated 2-pyrrolidinone (Z-score +2.11). 4-guanidinobutanoate was found to be high-normal or mildly elevated in all five (100%) of the adult cases (Z-score +1.70–2.38). Z-scores of both 2-pyrrolidinone and 4-guanidinobutanoate were significantly lower in the adult age group compared to the pediatric age group (p = 0.0004; p = 0.002, respectively) (Figures 4B,C). Argininate was also found to be mild to moderately elevated in plasma in 6/12 (50%) samples where this metabolite was reliably detected. Argininate was not reported in one sample (Case 2) that was assayed on a previous version of the LC-MS/MS platform. Z-scores of argininate were not significantly different between the pediatric and adult age-groups (Figure 4D).

Urine metabolomic assays were obtained from 12 individuals–four individuals from the BG-BCM cohort (Cases 11, 13, 14, and 15, Table 1) and eight individuals from the SSADHD Biobank (Cases 17–24; Table 1). In two samples (Cases 14 and 15), 2-pyrrolidinone was not reported in the LC-MS/MS platform used for the assay at the time. For the remaining 10 urine samples, 2-pyrrolidinone Z-scores were well within the normal range (Z-scores ranged from −1.005 to +0.98) (Supplementary Table S4). However, 4-guanidinobutanoate was mildly elevated in 10 of 12 samples (83%, Z-scores ranged from +1.91 to +2.78) regardless of age-group but normal in two adult samples (cases 14 and 15) (Supplementary Table S4). Argininate z-scores were normal in eight of ten (80%) samples where it was reported, and high-normal or mildly elevated in two samples (cases 17 and 18). Argininate was not reported in two samples (cases 14 and 15). Succinimide was found to be mild to moderately elevated in all pediatric urine samples (Z-score +2.09 to +3.16) and normal in all five samples in the adult age-group (Z-score +1.02 to +1.72) (Supplementary Table S4).

Based on all available data, we have comprehensively confirmed the diagnosis of SSADHD in three individuals in BG-BCM cohort with both molecular and metabolomic analyses. Here, we present their clinical histories as a proof-of-concept of the power of this integrated diagnostic approach.

This male patient was referred for testing at 6-month of age because of developmental delay and hypotonia. He was born following an uncomplicated, term pregnancy via vacuum-assisted vaginal delivery to a healthy G2P2 mother. At 2 months of age, concerns were raised due to poor weight gain. At 4.5 months of life, serum bicarbonate was measured and found to be reduced, and he was subsequently diagnosed with renal tubular acidosis and initiated on sodium bicarbonate. A dedicated renal ultrasound was normal. The patient, however, continued to have worsening hypotonia, weight loss, and lethargy. A brain MRI showed mild prominence of frontotemporal subarachnoid spaces though no structural anomalies. Urine organic acid analysis showed elevations of 4-hydroxybutyric acid, 2,4-dihydroxybutyric acid and 3,4-dihydroxybutyric acid. Subsequent exome sequencing analysis identified compound heterozygous, pathogenic variants in ALDH5A1 (NM_001080.3:c.1015–2A>C and NM_001080.3:c.1597G>A (p.Gly533Arg) confirming the diagnosis of succinic semialdehyde dehydrogenase deficiency (Table 1). Untargeted metabolomic profiling in plasma revealed elevations of 2-pyrrolidinone and 4-guanidinobutanoate, consistent with molecular genetic findings and previous biochemical testing (Figure 4A; Supplementary Table S4).

This female patient had been conceived via in vitro fertilization (IVF), and the pregnancy was complicated by twin gestation and the later in utero demise of her twin at 12–14 weeks gestation. The patient was eventually delivered at term (41 weeks) following an uncomplicated vaginal delivery. She was first noted to be hypotonic around 2 weeks of life and was notably delayed in the attainment of gross motor, fine motor, and speech milestones. At 12 months of life, her babbling decreased, and she developed a disinterest in interactive play with unusual hand postures. An MRI of the brain at the time was read as “normal”; later careful reinterpretation, however, noted subtle bilateral increased signal in the globus pallidus. The patient had normal female karyotype, MECP2 sequencing and deletion/duplication analysis, creatine kinase (CK) level, and transferrin isoelectric focusing. Chromosomal microarray had shown only a paternally inherited deletion at 12q22.2 that was felt not to be clinically significant. Urine organic acids analysis at the age 27 months, however, demonstrated significantly elevated GHB at >500 mmol/molCr. Confirmatory Sanger sequencing at 31 months of age (2 years 7 months) revealed two pathogenic variants in ALDH5A1 (confirmed to be in trans): NM_001080.3:c.612G>A (p.Trp204Ter) and NM_001080.3:c.1234C>T (p.Arg412Ter) consistent with her biochemical diagnosis of SSADHD (Table 1). Untargeted metabolomic profiling in plasma revealed elevations of 2-pyrrolidinone, 4-guanidinobutanoate, and argininate consistent with previous molecular genetic and biochemical testing (Figure 4A; Supplementary Table S4).

The patient has had two electroencephalograms (EEG) and has never had clinical or electrographic seizures. She has been diagnosed with perseveration/fixation, attention deficit hyperactivity disorder (ADHD), and obsessive-compulsive disorder (OCD). Given that she has never had seizures, vigabatrin has never been used in her care, though she has had a trial of taurine, lemon balm, and omega-3 DHA gel (all without significant improvement in her symptoms). She does take supplemental creatine with reported improvements in her reading, math skills, and occupational therapy goals.

This patient is the younger brother of Case 13 and was 3 months old when his sister was eventually diagnosed and is now 10 years old. He was born at term (39 weeks) via an uncomplicated vaginal delivery and had a normal newborn course. Because of his siblings’ diagnosis, the patient underwent urine organic analysis at 4.5 months of life which was indeed abnormal with gross elevations in GHB (>500 mmol/molCr). He thereafter had known familial mutation testing and was confirmed to have the same two variants in ALDH5A1 that were identified in his sister (NM_001080.3:c.612G>A (p.Trp204Ter) and NM_001080.3:c.1234C>T (p.Arg412Ter)) (Table 1). He later had a brain MRI that also showed a very subtle increase in T2 signal in the globus pallidus bilaterally with normal MR spectroscopy. Developmentally, the patient had normal muscle tone for the first 4 months of his life before becoming progressively more hypotonic beginning at the age of 6 months. He began to babble at 12 months, crawled at 13 months, and walked at 2.5 years old. He currently has about 5 words and uses sign language as his primary form of communication. He attends a school for the deaf and can sign four to 5-word sentences. Additionally, he uses a mobile device to create sentences through pictures. He has, much like his sister, never had seizures, aggression, sleep disturbances, or self-injurious behaviors, though he has been diagnosed with perseveration/fixation, attention deficit hyperactivity disorder (ADHD), and obsessive-compulsive disorder (OCD). He has never taken vigabatrin but has had a trial of taurine, lemon balm, and omega-3 DHA gel. He is currently taking creatine with reported improvements in his reading, math skills, and occupational therapy goals. Untargeted metabolomic profiling in plasma revealed elevations of 2-pyrrolidinone, 4-guanidinobutanoate, succinamic acid, and argininate consistent with previous molecular genetic and biochemical testing (Figure 4A; Supplementary Table S4).