BCKDK deficiency: a treatable neurodevelopmental disease amenable to newborn screening

Abstract

There are few causes of treatable neurodevelopmental diseases described to date. Branched-chain ketoacid dehydrogenase kinase (BCKDK) deficiency causes branched-chain amino acid (BCAA) depletion and is linked to a neurodevelopmental disorder characterized by autism, intellectual disability and microcephaly. We report the largest cohort of patients studied, broadening the phenotypic and genotypic spectrum. Moreover, this is the first study to present newborn screening findings and mid-term clinical outcome.

In this cross-sectional study, patients with a diagnosis of BCKDK deficiency were recruited via investigators’ practices through a MetabERN initiative. Clinical, biochemical and genetic data were collected. Dried blood spot (DBS) newborn screening (NBS) amino acid profiles were retrieved from collaborating centres and compared to a healthy newborn reference population.

Twenty-one patients with BCKDK mutations were included from 13 families. Patients were diagnosed between 8 months and 16 years (mean: 5.8 years, 43% female). At diagnosis, BCAA levels (leucine, valine and isoleucine) were below reference values in plasma and in CSF. All patients had global neurodevelopmental delay; 18/21 had gross motor function (GMF) impairment with GMF III or worse in 5/18, 16/16 intellectual disability, 17/17 language impairment, 12/17 autism spectrum disorder, 9/21 epilepsy, 12/15 clumsiness, 3/21 had sensorineural hearing loss and 4/20 feeding difficulties. No microcephaly was observed at birth, but 17/20 developed microcephaly during follow-up. Regression was reported in six patients. Movement disorder was observed in 3/21 patients: hyperkinetic movements (1), truncal ataxia (1) and dystonia (2). After treatment with a high-protein diet (≥ 2 g/kg/day) and BCAA supplementation (100–250 mg/kg/day), plasma BCAA increased significantly (P < 0.001), motor functions and head circumference stabilized/improved in 13/13 and in 11/15 patients, respectively. Among cases with follow-up data, none of the three patients starting treatment before 2 years of age developed autism at follow-up. The patient with the earliest age of treatment initiation (8 months) showed normal development at 3 years of age. NBS in DBS identified BCAA levels significantly lower than those of the normal population.

This work highlights the potential benefits of dietetic treatment, in particular early introduction of BCAA. Therefore, it is of utmost importance to increase awareness about this treatable disease and consider it as a candidate for early detection by NBS programmes.

BCKDK, intellectual disability, autism spectrum disorder, microcephaly, newborn screening

Introduction

Branched-chain ketoacid dehydrogenase kinase (BCKDK) deficiency (OMIM #614901) was first described by Novarino et al.¹ in 2012 as a Mendelian form of autism with intellectual disability and epilepsy. In this first article, six patients were reported (age range: 5 to 22 years). Autism and intellectual disability were constant symptoms in all patients; half had seizures, and all exhibited low plasma levels of branched-chain amino acids (BCAA). In 2014, two new cases with novel genetic variants and a similar phenotype were described.² No further clinical reports or series of new patients with this rare neurodevelopmental disorder have been described in the literature.

Neurodevelopmental disorders (NDD) affect approximately 62/100,000 people around the world, which suggest that one child out of 160 should be affected.³ Genetic causes of developmental delay are diverse and include around 4525 different conditions in OMIM (https://www.omim.org/). The pathophysiology involved is complex and covers a great spectrum of mechanisms spanning from abnormal gene expression to the particular disturbed cellular process. Treatable NDD are mostly inborn errors of metabolism (IEM) and constitute around 116 different diseases according to a recent review.⁴ An important percentage of them are small-molecule disorders that have frequent multisystem involvement and decompensations triggered by metabolic stress factors, such as intercurrent infections or fasting.⁵ However, among the newest pathophysiological categories of IEM, defects of synthesis and transport of small molecules represent an emerging group of disorders that mimics non-metabolic genetic NDD of diverse severity.⁵ BCKDK deficiency belongs to this latter category of diseases.

Branched-chain ketoacid dehydrogenase (BCKDH) complex is a multimeric mitochondrial enzyme composed by four catalytic subunits E1α, E1β, E2-DBT and E3-DLD.⁶ It is regulated by kinase phosphorylation (inactivation) and phosphatase dephosphorylation (activation).⁷ When BCAA levels are in excess, BCKDH is in its active form and catabolizes the BCAAs that, in high concentration (especially leucine), are toxic for the brain. BCKDH deficiency causes Maple Syrup Urine Disease, where accumulation of BCAAs and their corresponding ketoacids result in a toxic effect on brain function. By contrast, when BCKDH is constantly over-active due to mutations in the regulatory kinase (BCKDK), there is an uninhibited oxidation of BCAAs leading to deficiency of these essential amino acids, which impairs protein synthesis during development and post-natal growth.^1,2

As a MetabERN (the European Reference Network of Inherited Metabolic Disorders) initiative, and with the collaboration of physicians from different centres in Europe, the Middle-East and America, 21 patients were recruited with BCKDK deficiency in order to describe the clinical and biochemical characteristics of the disease, to report the mid-term outcome after treatment and the dried blood spot (DBS) newborn screening (NBS) BCAA profiles results when available.

Materials and methods

Patient recruitment and clinical phenotyping

Twenty-one patients presenting with low BCAA and harbouring pathogenic mutations in the BCKDK gene were recruited via participating physicians in Belgium, Brazil, Germany, Norway, Saudi Arabia, Spain, Turkey and the United Kingdom during 2018–2021. Patient characteristics, demographics, growth, diet, clinical phenotype before and after BCAA treatment, neuroimaging, EEG studies, biochemical studies (amino acid measurements in plasma, CSF and in NBS DBS when available), DNA analyses and treatment outcomes were collected. The severity of intellectual disability (ID) was established with intelligence quotient (IQ) scores (when available) or information on the level of functioning in accordance with the fifth edition of the Diagnostic and Statistical Manual of Mental Disorders (DSM-5), American Psychiatric Association.⁸ Autism spectrum disorder (ASD) was assessed according to DSM-5 autism criteria. Global motor function was scored using the gross motor function (GMF) scale.⁹ Clumsiness was defined as inadequate balance and coordination.

Anthropometric classification

Body mass index (BMI, calculated as kg/m²) was used to assess nutritional status. Anthropometric data were converted to standard deviation scores (BMI Z-score and height Z-score) as recommended by the World Health Organization (WHO).¹⁰ As the WHO charts of head circumference by age and sex group¹⁰ only cover children until 5 years of age, the occipital frontal circumference (OFC) of patients older than 5 years was categorized according to the Fernández et al.¹¹ charts by sex and age that has been validated for Spanish territory to native Caucasian population, Africa immigrants from Maghreb and Sub-Saharan territories and in term newborns. Central and South America from 0 to 18 years old were categorized according to González et al.¹² Normal nutritional state was defined as weight-for-length/height or BMI-for-age Z-score ≥ −2 and <+2 of the median for patients younger than 5 years (60 months), and Z-score ≥ −2 and <+1 if older than 5 years. Microcephaly was defined as an OFC Z-score < −2.

Biochemical studies

Amino acids were quantified by ion-exchange chromatography or by liquid-chromatography coupled to mass spectrometry (LC-MS) in the respective centres, as previously reported.¹³ Participants identified BCAA deficiency by comparing it with the reference values of their local laboratories. In some patients branched-chain ketoacids (BCKAs) were also analysed in urine by gas chromatography–mass spectrometry as previously described.¹⁴

Genetic studies and structural analysis

BCKDK gene mutations were described based on the longest isoform 1 of BCKDK.¹⁵ The pathogenicity of all BCKDK variants were evaluated using the standard American College of Medical Genetics and Genomics guidelines.¹⁶ For missense variants, mutation effects were first studied in silico with online prediction tolls including Varsome (for mutations pathogenicity prediction) and gnomAD (for frequency population).

The three missense mutations His206Pro, Gly367Asp and Leu389Pro in the BCKDK structure, caused by the c.617A>C, c.1100G>A and c.1166T>C changes, respectively, were analysed using PyMOL Molecular Graphics System from Schrödinger, LLC and the previously determined crystal structure of the rat branched-chain α-ketoacid dehydrogenase kinase (PDB ID: 4DZY).

Dietetic data collection

The estimated natural protein intake (in g/kg/day) was retrieved from collaborating participants at baseline, upon diagnosis and at follow-up. Intake of each BCAA supplemented (in mg/kg/day) was collected at diagnosis and at follow-up, plus their frequency of administration per 24 h. A high-protein diet was defined as ≥2 g/kg/day regardless of age.

Newborn screening analyses and statistics

Amino acids and acylcarnitines were extracted from a single 3.2-mm diameter punch from each DBS in Belgium and Germany by using the ‘MassChrom Amino Acids and Acylcarnitines from Dried Blood’ derivatized kit (Chromsystems), while in Norway and Spain the NeoBase Non-derivatized kit (PerkinElmer) was used and quantified by flow injection analysis with ultra-performance liquid chromatography coupled to tandem mass spectrometry (UPLC-MS/MS).

Dried blood spot amino acids and acylcarnitines required for score calculations using the BCKDK post-analytical Collaborative Laboratory Integrated Reports (CLIR) tool were collected and the following data were uploaded to the Tool Runner (CLIR—Login Page (mayo.edu): alanine (ALA), valine (VAL), leucine\isoleucine\OH-proline (XLE), methionine (MET), citrulline (CIT), phenylalanine (PHE), tyrosine (TYR), free carnitine (C0), acetylcarnitine (C2), propionylcarnitine (C3), butyrylcarnitine (C4), isovalerylcarnitine (C5), tetradecanoylcarnitine (C14) and octadecanoylcarnitine (C18:1). The entered values were adjusted for the covariates birthweight (BW; g) and age (hours at NBS sample collection). Site-specific amino acid reference ranges available in CLIR were incorporated into BCKDK tool score calculation for Norwegian and Spanish patients through the use of location-specific covariate adjustments¹⁷; site-specific reference ranges were missing to calculate location-specific covariate adjustments for Belgian and German data. A site-specific version of the BCKDK tool was created for the German location to accommodate missing values for C3 and CIT.

An alternative approach to separate newborns with BCKDK deficiency from healthy controls was based on a previous pilot study performed in CLIR at Mayo Clinic (Rochester, MN, USA), demonstrating the amino acids VAL and XLE together with their ratios to PHE, ALA and TYR best discriminated BCKDK deficiency patients from the rest of NBS DBS samples. To optimize the procedure, all amino acids and ratios sampled from the BCKDK newborn DBS were Box–Cox transformed to correct for skewness, subsequently adjusted for sampling time by polynomial regression in age and finally converted to Z -scores. Z-scores outside Tukey fences (−3.372 and + 3.372; factor k = 2), corresponding to 0.04 and 99.96% percentiles, were considered as statistical outliers and excluded. As VAL and XLE are highly correlated we constructed two more independent variables: X1 = mean Z-score of VAL and XLE, X2 = mean Z-score of the ratios VAL/PHE, XLE/ALA, XLE/PHE and XLE/TYR. X1 is the primary BCAA marker and X2 compensates for other factors that tend to give a general change in the level of amino acids. In a scatterplot between variable X1 and X2 the patients were compared to the reference population. The apparently healthy newborn reference population was made available by The Karolinska Institute, Stockholm (courtesy of Dr Rolf Zetterström). From a total of n = 275 381 samples, a subgroup of n = 55 374 reference babies were selected to establish birthweight (3400–3650 g) and age (48–250 h) ranges to compare with the patient group.

For the statistical analysis, the SPSS 15.0 was employed (IBM Corporation) software. Mann–Whitney U-test was used to compare the plasma and CSF BCAA concentrations before and after treatment.

Standard protocol approvals, registrations and patient consents

All parents or legal representatives of the patients gave written informed consent. This study was approved by the local institutional ethics committee (Sant Juan de Déu Hospital ID number: PIC-131-18) and by the Norwegian Regional Committee for Medical and Health Research Ethics (ID Number 75608/2020) and the local institutional ethics committee at Oslo University Hospital (ID number 20/00681).

Data availability

Epidemiological data, treatment and biochemical data before and after the treatment are available at Brain online on Supplementary material.

Results

Patient cohort, genetic and structural characterization

Twenty-one patients (57% male) from 13 different families with (likely) pathogenic BCKDK variants were included. Mean age of diagnosis was 5.8 years (range 8 months–16.6 years). Twenty of twenty-one patients harboured homozygous BCKDK variants: 11 nonsense mutations, 4 splice defects and 5 missense/in-frame variants. One patient had compound heterozygous variants (nonsense and missense). Sixteen patients were not previously reported during the study period, 10 with novel mutations which included a total of six unique mutations as two of them were recurrent. The remaining six individuals had previously reported mutations. Figure 1 depicts the 10 BCKDK mutations reported in our patient cohort and the proportion of the four genotypic groups. Supplementary Table 1 describes the epidemiological data of our cohort and Table 1 describes the phenotypic features at the diagnosis.

Figure 1

Schematic presentation of BCKDK mutations in 21 patients and proportion of genotypic groups. Mutations variants are represented in the figures.

Table 1

BCKDK phenotype at diagnosis

Phenotype	HPO code	Frequency n (%)
Progressive microcephaly	HP:0000253	17/20 (85)
Global developmental delay	HP:0001263	21 (100)
Motor delay	HP:0001270	21 (100)
Language impairment	HP:0002463	17 (100)
Intellectual disability	HP:0001249	16 (100)
Intellectual disability, severe	HP:0010864	15/16 (93.8)
Developmental regression	HP:0002376	5/8 (62.5)
Behavioural abnormality	HP:0000708	20/21 (95.2)
Autistic behaviour	HP:0000729	14/19 (73.7)
Autism	HP:0000717	11/19 (58)
Self-injurious behaviour	HP:0100716	5/17 (28)
Aggressive behaviour	HP:0000718	8/18 (44.4)
Hyperactivity	HP:0000752	5/21 (23.8)
Restlessness	HP:0000711	6/21 (26.8)
Attention deficit hyperactivity disorder	HP:0007018	4/18 (22.2)
Seizure	HP:0001250	9/21 (42.9)
Bilateral tonic–clonic seizure	HP:0002069	3/18 (16.7)
Generalized myoclonic seizure	HP:0002123	1/18 (5.6)
Typical absence seizure	HP:0011147	2/18 (11.1)
Generalized-onset seizure	HP:0002197	9/21 (42.9)
Focal-onset seizure	HP:0007359	1/21 (4.8)
Interictal epileptiform activity	HP:0011182	12/18 (66.7)
Hypotonia	HP:0001252	5/17 (29.4)
Abnormality of movement	HP:0100022	3/20 (15)
Dystonia	HP:0001332	2/10 (10)
Ataxia	HP:0001251	1/20 (5)
Hyperkinetic movements	HP:0002487	1/20 (5)
Clumsiness	HP:0002312	12/15 (80)
Feeding difficulties	HP:0011968	5/20 (25)
Sensorineural hearing impairment	HP:0000407	3/21 (14.3)
Hyperreflexia	HP:0001347	5/17 (29.4)
Polyneuropathy	HP:0001271	2/17 (11.8)
Abnormal facial shape	HP:0001999	10/21 (47.6)
Full cheeks	HP:0000293	4/21 (19)
Thin upper lip vermilion	HP:0000219	4/21 (19)
Abnormal nasal bridge morphology	HP:0000422	2/21 (9.5)
Hypoplastic philtrum	HP:0005326	2/21 (9.5)
Small forehead	HP:0000350	3/21 (14.3)
Coarse hair	HP:0002208	1/17 (5.9)
Dry skin	HP:0000958	2/17 (11.8)
Inflammatory abnormality of the skin	HP:0011123	3/17 (17.6)
Hydronephrosis	HP:0000126	1/17 (5.9)

Structural analysis revealed several results. Amino acid L389 is located in the dimer interface (Fig. 2A and B) and this mutation has previously been shown to have total loss of kinase activity.² Most likely, the Leu389Pro mutation interferes with BCK dimer formation, which consequently affects the enzymatic activity. Amino acid Gly367 is located close to the potassium ion within the nucleotide binding pocket (Fig. 2C). Our structure analysis shows that a substitution of Gly367 with the much larger and positively charged arginine will introduce steric conflicts within the site that might disturb the nucleotide binding activity. Finally, His206 is located at the end of one of the helices in the four-helix bundle in the N-terminal domain (Fig. 2D). As can be seen from the figure, His206 is engaged in two intramolecular hydrogen bonds with the main chain. Thus, it is reasonable to assume that the His206Pro mutation will destabilize the protein. In addition, proline will induce a kink in the α-helix that will further destabilize the folding of the protein and most likely affect the enzymatic activity. The pathogenicity of the in-frame mutation p.Thr335del caused by the c.999_1001delCAC was only recently demonstrated during preparation of this manuscript.¹⁸ Functional studies of missense variants were not performed. However, all three allelic variants found in our study (c.1100G>A, c.1166T>C and c.617A>C) are located within highly conserved regions in vertebrates and higher species, indicating that changes are probably damaging, as demonstrated by in silico prediction (Fig. 2).

Figure 2

Structural overview of the rat branched-chain α-ketoacid dehydrogenase kinase dimer. (A) The dimeric BDK structure with the Leu389 residue (depicted in red) located in the interface between the two monomers. (B) Close-up view of the Leu389 within the dimeric interface. (C) Close-up view of Gly367 and Thr335 in the vicinity of the nucleotide binding pocket. (D) Close-up view of His206 is located at the end of one of the helices in the four-helix bundle in the N-terminal domain.

Clinical status before treatment

Head circumference and nutritional status (body mass index)

The majority of the patients were born at term (19/21) and all had BW, OFC and length > −2 SD for gestational age. However, at the time of the diagnosis all of them presented a decrease in the OFC score and 16/20 (80%) patients had microcephaly (OFC < −2 SD). At diagnosis, 17/18 had BMI Z-scores within the normal range (Supplementary Table 1).

Neurodevelopmental and behavioural findings

Global developmental delay was reported in all patients. All 17 patients older than 2 years of age had language impairment and nine were non-verbal at diagnosis. The delayed motor milestones noted in all patients included lack of head control, delayed rolling over, unsupported sitting and walking. Nineteen of 21 patients had gross motor function impairment, GMF III and IV score in 5/18. Sixteen of 16 had intellectual disability, 9/21 epilepsy, 12/15 clumsiness, 3/20 sensorineural hearing loss and 4/19 feeding difficulties. In all patients except a single case [Patient (P) 18], behaviour issues were noted: 12/17 (70%) fulfilled the DSM-5 criteria of autism spectrum disorder Three patients had autistic traits (repetitive movements, little cognitive flexibility, poor interaction with peers) but did not fulfil the ASD criteria (two) or data of ASD were not available (one), leaving a total of 15 of 20 (75%) with either autistic traits or ASD. Twelve of 17 patients had other behavioural abnormalities including aggressive behaviour (two), self-injurious behaviour (two), restlessness/hyperactivity (eight) and attention deficit hyperactivity disorder (three). Regression was reported in six of nine patients (there were no available data concerning the neuroregression questionnaire in 12 patients). Two children had motor and language regression, three had only motor regression and one had only language regression. Movement disorders were observed in three of 21 patients: truncal ataxia (one), hyperkinetic movements (one) characterized by uncoordinated movements in limbs and mouth and dystonia (two). One patient showed dystonia in hands (during running) and the other one dystonic tetraparesis. At diagnosis, 12/15 patients had clumsiness. In none of the 14 patients where information was obtainable was an abnormal sleep pattern observed (Table 1 and Supplementary Table 1).

Other clinical parameters

Other clinical features presented in the patient cohort were dry skin (two), skin inflammatory processes (acrodermatitis enteropathica-like; three), coarse hair (one), hydronephrosis (one), hypotonia (four), limb hypertonia (one), mild limb paresis (one), hyperreflexia (three) and motor axonal polyneuropathy (one). Dysmorphic features were reported in some patients including full cheeks (four), thin upper lip vermilion (four), abnormal nasal bridge morphology (two) and hypoplastic philtrum (two).

EEG patterns

Nine of 20 patients had epilepsy. One had focal-onset seizure and the others generalized onset seizures: bilateral tonic–clonic seizures (three), generalized myoclonic seizures (one), typical absence seizures (two). Interictal EEG abnormalities were present in 14/17 patients with 12 presenting epileptiform activity (Supplementary Table 1).

Cerebral MRI findings

Cerebral MRI before diagnosis was reported as normal in 11 of 19 patients. The eight remaining patients had non-specific findings: thin corpus callosum (two), corpus callosum agenesis (two), reduced volume of supratentorial brain parenchyma (one), reduced white matter volume (two), enlarged ventricles (two), delayed subcortical and temporal lobe myelination (two) (Supplementary Table 1).

Biochemical results before treatment

At diagnosis, the average between BCAA concentrations (average of two samples per patient) was below the reference ranges both in plasma and CSF (Supplementary Tables 2 and 3) with the exception of P19 and P20, who had just one sample of CSF and plasma, respectively. In P19 Val and Leu levels were normal but isoleucine concentration was decreased. In P20, Leu and Ile (P20) determinations were at normal levels but valine concentration was decreased.

Diet therapy and outcome measures

Dietetic management and biochemical markers outcome

In order to restore BCAA depletion, a dietetic treatment was started characterized by BCAA supplementation and high total protein intake. The BCAA supplementation was initiated in all but two patients (described later; Supplementary Table 2). The BCAA treatment observational period was median 3.26 years (0.5–10.7 years). At diagnosis, natural protein intake was mean 2.5 g/kg/day (1.7–3.5 g/kg/day) in 17/18 patients. The mean Leu, Val and Ile supplementation started at diagnosis was 100 mg/kg/day (50–757), 100 mg/kg/day (25–368) and 100 mg/kg/day (25–400), respectively, with an overall mean BCAA intake of 100 mg/kg/day (33–508). BCAA supplements were administered mean 3.6 times per day (range: 3–6). At follow-up, natural protein and BCAA supplementation were increased to 2.7 g/kg/day (2–4 g/kg/day) and 206 mg/kg/day (140–2000), respectively, and distributed four times per 24 h (range: 4–7; Supplementary Table 2). Three of eight patients had their protein intake increased from baseline (protein intake ≥ 3 g/kg/day) and there was no change in the protein intake (mg/kg per day) for five patients (data available for 8/21 patients). At follow-up, mean intake of Leu, Val and Ile was 250 mg/kg/day (150–2000), 169 mg/kg/day (120–2000) and 202 mg/kg/day (150–2000), respectively. The ratios of Leu:Val and Leu:Ile from the BCAA supplement at diagnosis and follow-up are shown in Supplementary Table 2. The Leu:Val ratio was 2:1 in 5/21 (24%) and 2/14 (14.3%) while the Leu:Ile ratio was 2:1 in 7/21 (33.3%) and 2/14 (14.3%) at diagnosis and follow-up, respectively. In two patients (P19 and P20), BCAA supplements were not provided due to lack of reimbursement and one patient (P13) had poor compliance. After treatment, plasma Leu, Val and Ile increased significantly (P < 0.001; Fig. 3). In three patients (P11, P12 and P16) BCAA concentration was measured before, during and after a meal that included BCAA supplements. The plasma BCAA concentration was under the reference range before the BCAA intake, showing the highest concentrations 2 h after the meal (Fig. 4). Within 3 and 5 h after the meal, Leu and Ile had returned below reference levels. In CSF a tendency towards higher BCAA concentration was observed after the BCAA treatment period, but it was not possible to reach conclusions as there were few patients with BCAA CSF levels available before and after the treatment period (Fig. 3 and Supplementary Table 3).

Figure 3

Branched chain amino acids (BCAA) before BCAA supplementation and a follow-up. (A) Pretreatment: n = 12/21; post-treatment: n= 5/21, 1/5 patients did not have CSF leucine levels pretreatment. No significant differences between the samples from before and after treatment. Mean of 1.2 samples per patient pretreatment and 1 sample post-treatment. (B) Pretreatment: n = 19/21; post-treatment: n = 14/21; P < 0.001 for leucine, valine and isoleucine independently. Mean of 1.8 samples per patient pretreatment and 6.4 samples post-treatment.

Figure 4

Plasma BCAA before and after a diet meal. (A) Patients: n = 3/12 (P11, P12 and P16). The red line corresponds to the moment of a regular meal plus BCAA intake (minute 0). BCAA supplementation: P11 (18 months) 240 mg/kg; P12 (12 years) 112 mg/kg; P16 (4 years) 100 mg/kg. P16 did not have the data before the diet-meal intake available. Time points of BCAA measure: P11: 30 min before, 150, 285 and 390 min after meal; P12: 30 min before, 150, 270 and 405 min after meal; P16: 120, 240 and 360 min after meal. (B) Controls: 500 mg/kg of Ile (isoleucine) was administered at the moment zero.

Anthropometrics, language, neurodevelopment, behavioural issues and MRI findings

At follow-up, 11 of 15 patients increased or stabilize their OFC Z-scores. In Fig. 5 we show the chart of nine patients. Most patients (10 of 12) maintained BMI Z-scores within the normal range at follow-up. In three patients (P2, P9 and P21) language improvement was reported after treatment. Patients with the earliest treatment introduction (P3, P15) developed verbal language and sentence-building acquisition. Patients who started treatment before 2 years of age (P3, P11 and P15) did not develop autistic features over time (at the moment all of them are older than 3 years). Most patients did not modify behavioural abnormalities (hyperactivity, restlessness, aggressiveness) with the diet. Motor functions improved in 8/13 and stabilized in 5/13 patients. Several patients, in particular those with early treatment, developed gait and fine motor functions. After BCAA treatment, hyperkinetic movements characterized by very energetic and sudden jerks appeared in P13 as the only movement disorder reported in all patients. Three of nine patients reported persistence of seizures after diet treatment and one patient had the first episode of generalized tonic–clonic seizures one month after onset of treatment.

Figure 5

Head circumference chart. Only a few patients have data from head circumference follow-up and although a global tendency towards improvement or stabilization is found, normalization was achieved only in few patients (P2, P11 and P15).

Ten patients had brain MRI performed after treatment. No significant changes were observed.

Detailed information about outcome after treatment is given in Supplementary Table 1.

Newborn screening

In seven BCKDK-deficient patients DBS analyses were available from their respective NBS programmes: Belgium (n = 3), Germany (n = 2), Norway (n = 1) and Spain (n = 1). BCKDK CLIR scores ranged from not informative (n = 2), possible BCKDK (n = 1), likely BCKDK (n = 1) and very likely BCKDK (n = 3) (Table 2).

Table 2

BCKDK-deficient case scores (n = 7) entered into the BCKDK post-analytical interpretive CLIR tool

Patient (number)	Sample age (h)	BW (g)	ALA (DBS µmol/l)	VAL (DBS µmol/l)	XLE (DBS µmol/l)	Case scorea	Percentile rank score of all BCKDK	Interpretation
Spain (15)	48	2600	368 (134–473)	50 (64.7–227)	36 (73.5–228)	412	100	Very likely
Norway (11)	63	1700	351 (174–488)	44 (76.2–243)	38 (84–239)	371	92	Very likely
Germany (2)	61	3670	386 (181–412)	82 (<324)	69 (<294)	62b	21	Possibly
Germany (3)	43	4040	366 (181–412)	38 (<207)	41 (<281)	277b	100	Very likely
Belgium (16)	72	3200	317 (130–358)	44 (65.2–200)	89 (131–253)	0	—	Not informative zero scorec
Belgium 2 (17)	72	3200	400 (130–358)	47 (65.2–200)	84 (131–253)	214	57	Likely
Belgium 3 (18)	96	3600	312 (130–358)	81 (65.2–200)	210 (131–253)	0	—	Not informative zero scorec

DBS reference ranges: Site-specific amino acid reference ranges available in CLIR were incorporated into BCKDK tool score calculation for Norwegian and Spanish patients through the use of location-specific covariate adjustments. For Germany and Belgium, unadjusted local laboratory DBS amino acid reference ranges are given. DBS = dried blood spot.

CLIR score interpretation, available-to-all tool: ≥300: Condition is very likely BCKDK, ≥105 and <300: condition is likely BCKDK, ≥25 and <105: condition is possibly BCKDK, < 25: profile is not informative for BCKDK.

CLIR score interpretation, German site-specific tool: ≥190: Condition is very likely BCKDK, ≥75 and <190: condition is likely BCKDK, ≥1 and <75: condition is possibly.

BCKDK, < 1: profile is not informative for BCKDK.

Zero score: the calculated case score was set to zero as a result of XLE falling above the first percentile of the cumulative reference range.

The alternative calculation of BCKDK DBS amino acid profile based on the previous CLIR pilot study, illustrated in Fig. 6, showed Z-scores separated from the healthy cohort, except for one newborn whose amino acid values were located well within the reference interval.

Figure 6

DBS newborn screening from BCKDK deficiency. Amino acid plot from seven newborns later diagnosed with BCKDK deficiency (orange squares) compared to a subpopulation of healthy newborns (blue diamonds). Algorithm by courtesy of Prof. Piero Rinaldo.

Discussion

Branched-chain amino acids are essential amino acids involved in vital cellular reactions such as protein turnover regulation, autophagy signalling, mitochondrial function and neurotransmitter metabolism. Net transference of a BCAA’s amino group to alpha ketoglutarate forms glutamate and does not occur until BCKAs are irreversibly metabolized by the BCKDH complex.¹⁹ BCAA metabolism is tightly regulated by the inactivating phosphorylation of the complex and subsequent reactivating dephosphorylation. Mutations in genes involved in this process result in the neurometabolic diseases Maple Syrup Urine Disease and BCKDK deficiency. BCKDH deficiency due to mutations in the BCKDK gene results in a hypermetabolism of BCAAs and subsequent low body concentrations.²⁰

We report the largest cohort of BCKDK-deficiency patients studied so far, broadening the phenotypic and genotypic spectrum. Moreover, this is the first study to present newborn screening findings in a subset of the cohort. All BCKDK-deficient patients showed global developmental delay at diagnosis. Seventy-five per cent presented autistic traits or ASD and microcephaly was not present at birth in any of the cases, but appeared post-natally in most patients. Comparison with follow-up protein intake was difficult to interpret at a group level, both because data were missing and patients grew older with less protein requirements needed per recommended daily allowances.²¹ Our findings, however, support a marked difference in clinical outcome depending on whether BCAA supplementation occurred in early development (before 2 years old) or at later stages (beyond 2 years of age). In the three patients where BCAA treatment was initiated <2 years of age, follow-up indicated amelioration of the developmental delay compared to older patients. Opposed to that, patients in whom the treatment was started beyond 2 years of age only modestly responded to BCAA supplementation, in accordance with previous case publications. Additionally, as no natural history of the disease has been previously described, it is not possible to know the influence of physiological brain maturation processes in the improvements found in late-treated patients. Given that intensive and appropriate treatment from birth may have the potential to improve neurodevelopmental outcome, we suggest this condition is amenable to NBS. NBS is enabled based on amino acids measurements, as six of seven patients included in this study showed abnormal NBS DBS BCAA profile.

Prominent impaired cognitive function, autistic traits, abnormal motor development, epilepsy and head circumference stagnation are the main features of this series of patients. Head circumference and motor function were the two main items that improved with treatment. However, only few patients have data from head circumference follow-up and although a global tendency towards improvement or stabilization is found, normalization is achieved only in three patients. Remarkably, motor functions stabilized or improved in all patients and balance and coordination were also prone to improvement. In contrast to that, language acquisition and cognitive profile seemed less amendable, especially relevant when treatment was started late. Regarding epilepsy, we could not reliably evaluate if seizure control was due to the efficacy of the anti-epileptic drugs or to the BCAA supplementation itself. Finally, cognition and neuropsychiatric features did not improve after treatment. However, patients who initiated treatment before 2 years of age did not develop autism over time. Additionally, P15, who had the earlier diagnosis and treatment (8 months), presented normal cognition and almost normal global neurodevelopment when evaluated at 3 years.

One interesting feature not previously reported in BCKDK patients was the presence of acrodermatitis enteropathica-like eczema. Three patients presented with mimicking skin lesions that are found in organic acidaemias or Maple Syrup Urine Disease following a restricted BCAA diet.²²

Biochemically, BCKDK deficiency is characterized by low BCAA concentrations both in plasma and CSF.^1,2 Six of seven cases with available NBS analyses showed low BCAA values when local amino acid reference values were applied (Table 2). The single patient with normal BCAA values, graphically displayed within the reference population (Fig. 6), was homozygous for the same pathogenic mutation (c.999_1001delCAC) as two other patients from the same centre recently published.¹⁸ It remains uncertain whether the amino acid profile from this case resulted from timing of meal versus DBS sampling, analysis error or represents a false negative NBS case. Moreover, the BCKDK CLIR tool calculated zero scores for two of the three patients harbouring (c.999_1001delCAC) based on XLE values falling above the first percentile of the cumulative reference range. The Belgian reference range for XLE trends higher than the majority of sites that have contributed data to CLIR, making it an ideal candidate for location adjustment to bring this range into alignment with the rest and to potentially improve tool performance. However, a location adjustment factor could not be calculated for Belgium because it had not contributed sufficient reference data relative to other contributing locations. This is a limitation of the statistical methods implemented in CLIR and illustrates why contribution of reference data is important for optimal tool performance. Only four of the seven cases thus obtained a significant BCKDK CLIR -tool score (likely or very likely) and three patients would probably have been missed if the score was entirely relied on. However, the BCKDK tool has been developed on the amino acid and acylcarnitine profiles from only six previous cases with an age of collection of 24–48 h and the tool performance would most likely benefit from the availability of a larger NBS BCKDK-deficient profile-cohort. The alternative approach (Fig. 6) clearly discriminated six of seven cases from the healthy newborn cohort. The superior performance could possibly be explained by calculations being tailored to the specific cases as opposed to the CLIR tool, whose adjustments must pertain to a general population.

BCAA concentration was significantly increased in plasma after treatment initiation with a higher-protein diet and BCAA supplementation. However, these results must be interpreted with caution as only 1–3 plasma amino acid (p-AA) profiles were obtained per patient before and after BCAA treatment and the timing of p-AA sample in relation to the meal was not collected, thus limiting the validity of our observation. In normal conditions, BCAA plasmatic concentration fluctuates significantly before, during and after meals, suggesting that frequent feeding is necessary to obtain sufficient levels of BCAA available to the brain (Fig. 4). This fluctuation could be of special relevance on BCKDK-deficiency management. In fact, BCAA dosage, frequency and formula composition in Leu:Val and Leu:Ile ratios may play a crucial role.

Generally, a normal diet provides around double the quantity of Leu compared with Ile and and a third or double more Val.²³ Leu requirements are also higher than Ile and Val. We could therefore suggest that this higher BCAA prescription along with more physiological leucine ratios could better meet patient needs, even with daily BCAA plasma variations. Regarding how the diet replaces BCAA in the brain, in P13 the lumbar puncture was performed on two occasions, during BCAA night drip and during oral diet, with higher CSF BCAA levels achievements after BCAA overnight enteral feeding. Interestingly, he showed an increase in OFC during the period with night BCAA drip. However, when BCAA drip was terminated due to administration burden on the family, the improvement in OFC was lost (Fig. 5 and Supplementary Table 2). These findings point towards the need to develop a treatment that could provide a sustained BCAA release ensuring sufficient CSF levels of these amino acids to improve clinical outcome.

Regarding phenotype–genotype, we could outline a relationship between symptoms onset and mutations severity. Patients with missense or in-frame mutations had their first abnormal neurological symptoms recognized after 10–12 months of age. Moreover, these patients had normal or level I GMF. On the other hand, patients with nonsense or splicing mutations had earlier onset of the symptoms, between 2 and 9 months of age.

This study has some important limitations. First, a selection bias towards the most severe patients is unavoidable as patients were recruited via the physicians’ clinical practice in which global development delay, severe intellectual disability, along with progressive microcephaly and behavioural disorder guided the diagnostic approaches. Therefore, it is likely that the full phenotypic spectrum including milder disease is still unknown. Another shortcoming of our study was the retrospective and cross-sectional design limiting the interpretation of BCAA treatment effect versus natural development on outcome measures of the disease. Not all patients had follow-up data available and laboratory reference values both in plasma as well as in CSF or DBS differed between centres. Moreover, there were variations in diet prescriptions, age at diagnosis and time elapsed between diagnosis and follow-up, which all impaired valid conclusions on the treatment effectiveness. Regarding NBS, it remains unknown if DBS will be sensitive enough to detect all cases or milder phenotypes of BCKDK. A prospective pilot study with BCKDK deficiency identified by NBS is warranted to demonstrate newborns with BCKDK deficiency truly benefit from an early standardized BCAA diet guideline and to prove its potential in altering the natural history of BCKDK deficiency in analogy to PKU. In addition, no uniform nomenclature on phenotypic description was obtainable and neurodevelopmental tests applied varied between centres. Enlightened by this reality, and to further prevent it being a limitation in the disease study, we propose BCKDK-deficiency patients to be included in the Unified European Registry for Inherited Metabolic Disorders (U-IMD registry; https://www.u-imd-registry.org). This will enable prospective and standardized data collection.

In summary, we describe the largest series of BCKDK patients and mid-term outcomes after BCAA supplementation. This inborn error of metabolism can be recognized by low BCAA levels on selective diagnostic screening and may also be detected on DBS used in NBS. We suggest a diet with increased natural protein 3–4 g/kg/day and BCAA supplementations (Leu:Val 2:1 and Leu:Ile 2:1). Patients here reported had around 200:150 mg/kg/day (Leu:Val 2:1 and Leu:Ile 2:1) divided into 4–6 intakes per day. In any case, it should be adapted depending on the levels achieved in every patient. Enteral BCAA feedings or continuous drip during night could be a neuroprotective strategy during infancy, but may be challenging to sustain in practice. We recommend performing an annual 24-hour profile of hourly BCAA in patients to establish peaks and troughs to optimize the BCAA intake individually. This could be done through DBS analysis. If introduced early in life, the diet may have a potential to change the natural history and improve neurodevelopmental outcome, but the efficacy and frequency of distribution remains to be shown in prospective studies. Thus, we propose NBS pilot studies enabling early detection and treatment as the first step to improve outcome in parallel with development of more effective treatments.

Acknowledgements

This work was generated within the European Reference Network for Rare Hereditary Metabolic Disorders (MetabERN). We thank Martin Engvall and Rolf Zetterström, MCC, from Karolinska Institute for sharing NBS population reference values; Piero Rinaldo for advice on the CLIR tool; and all the parents and caregivers on behalf of the patients, for participating in this study.

Funding

A.G.C. is supported by FIS P118/00111, FI21/0073 ‘Instituto de Salud Carlos III (ISCIII)’ and ‘Fondo Europeo de desarrollo regional (FEDER)’.

Competing interests

A.G.C. has received honoraria for research support and lectures from PTC Therapeutics, honoraria for lectures from Biomarin, Immedica and Recordati Rare Diseases Foundation, and is a co-founder of the Hospital Sant Joan de Déu start-up ‘Neuroprotect Life Sciences’.

Supplementary material

Supplementary material is available at Brain online.

Author notes

Trine Tangeraas and Juliana R Constante contributed equally to this work.