All patients with a diagnosis of MATD and carriers based on newborn screening (NBS) program, genetic testing, and follow-up between November 2010 and December 2021 in Zhejiang Neonatal Screening Center, in which a total of 4,065,644 neonates were screened covering 364 maternity units in 90 counties. All procedures followed were in accordance with approval from the Research Ethics Committees, the Children’s Hospital of Zhejiang University School of Medicine. Patient information was tabulated without individual identifiers.

Tandem mass spectrometry (MS/MS) was used to detect the concentrations of amino acids and acylcarnitine with dry-spot-blood samples in our NBS center. The reference value of methionine (Met) is 7.18–41.35 μmol/L and the reference range of Phenylalanine (Phe) was 23.3–100 μmol/L. Based on NBS screening data, the cut-off value for Met was set to 82.7 μmol/L, an upper limit of normal values in the 99.9th centile healthy newborns. The cut-off value for the ratio of Met/Phe was 1.1. Plasma total homocysteine (tHcy) was considered to be elevated at a concentration above 15 μmol/L. Newborns were recalled due to a higher Met concentration of more than 82.7 μmol/L alone or due to a higher ratio of Met/Phe. If the second test was also positive (Met >82.7 μmol/L or Met/Phe >1.1), the newborn was referred for a gene test. Cases associated with cystathionine β-synthase deficiency, tyrosinemia type I, or severe liver diseases were excluded from this study. No other positive results of blood physiological indicators for IEME were detected in MATD patients.

Next-generation sequencing was carried out by BGI (Shenzheng, China) or by MyGenostics (Beijing, China) during 2010–2015. Other samples were analyzed with a targeted gene panel by next-generation sequencing (NGS) in our Genetic Diagnostic Laboratory at the Children’s Hospital as previously described (Lin et al., 2020) and contained 166 genes involved in an inborn error of metabolism (IEMD, including causative genes associated with MATD (MAT1A, GNMT, and AHCY) and Met-related IEMD, such as homocystinuria (CBS, MTHFR, ABCD4, LMBRD1, MMACHC, MMADHC, CD320, MLYCD, etc.). Sanger sequencing was performed to verify variants detected in the panel. Novel missense variants were evaluated with several automatic tools (SIFT, PolyPhen-2, MutationTaster, CADD, etc.). Variants leading to truncated protein were classified as pathogenic (P) or likely pathogenic (LP) according to the recommendations of the American College of Medical Genetics and Genomics (ACMG) (Richards et al., 2015).

All positive cases were followed up by specialists at the Department of Genetics and Metabolism. Regular intervals ranged from every 1–3 months, depending on the severity of manifestations, and were followed for cases at 1 year and every 3–6 months for cases beyond 1 year. Data from profiles of serum amino acids (including methionine), tHcy, and routine blood tests for the functions of the liver and kidney were monitored regularly during the follow-up. Other indicators, such as blood gases and electrolytes, serum glucose, and lactic acid, were collected irregularly. Bayley Scales of Infant Development, or the Ages-Stages Questionnaire (ASQ) or Gesell Developmental Schedule were used to assess the developmental status of MATD patients. Length/height-for-age and weight-for-age standards were according to China growth standards (0–3 years old) and WHO Child Growth Standards (>3 years old). Some special patients with severe clinical abnormalities or with extreme elevation of plasma Met (>800 μmol/L) were referred to undergo a brain MRI (magnetic resonance imaging) scan. If suspected cases harbored an unreported variant of MAT1A, the Met value and the ratio of Met/Phe from one parent carrying the same variant were detected.

Interventions were considered when plasma Met was >500 µmol/Lin infants. A low-methionine formula is usually used. Vitamin B6 and betaine were considered when tHcy increased slightly twice or exceeded 30 μmol/L. One patient was given oral SAMe. One patient underwent liver transplantation.

Met values were illustrated using GraphPad Prism 8.0. A 2-tail paired or unpaired Student’s t-test or ordinary one-way ANOVA was performed for comparisons. p < .05 was considered significant.

There were 15 unrelated MATD patients (6 male and 9 female patients) diagnosed with a monoallelic variant of MAT1A, of which 11 individuals (P5-P15) carried c.791G>A/p. R264H and four individuals carried c.776C>T/p. A259V. Patient 1 (P1) and patient 9 (P9) were born prematurely, while their mothers’ pregnancies were uneventful to that point. Consistent with previous reports (Couce et al., 2008; Martins et al., 2012; Couce et al., 2013), variants c.791G>A and c.776C>T are common in the AD type, with frequencies of 73% (11/15) and 27% (4/15) respectively. Patients harboring c. 776C>T had an initial Met range of 86–132 μmol/L, with a Met/Phe ratio of 1.13–1.85 and normal tHcy. Patients harboring c.791G>A had an initial Met range of 67–339 μmol/L, with a Met/Phe ratio of .92–3.02. However, slight elevations in plasma tHcy were detected among five patients harboring c.791G>A. Patients 1–15 were followed up for 4–111 months with Met levels fluctuating from 36 to 441 μmol/L without significant clinical symptoms. Each of them had normal growth and development without drug treatment during follow-up, while a transient decrease in serum zinc was observed in P7 at 2 years old and P10 at 6 months old. The mean initial Met of these 15 patients was 129 μmol/L, and the mean plasma Met during follow-up was 147 μmol/L (Figure 1B). All AD patients were healthy and had no clinical sequelae.

Twenty cases (17 male and 3 female patients) of AR-type MATD were diagnosed with persistently high plasma Met concentrations and compound heterozygous or homozygous variants of MAT1A without other pathogenic genes in this metabolic disorder. All patients were full-term infants. Compared to AD patients, AR patients have more complicated phenotypes and genotypes. The initial Met concentrations in 20 AR cases ranged from 60 to 332 μmol/L, with a mean value of 157 μmol/L (Table 1). No significant difference was exhibited in the initial Met concentrations between the AD and AR groups. However, the plasma Met values, during follow-up, had a wider range from 41 to 1408 μmol/L with a mean value of 479 μmol/L (Figure 1B), which was higher than that of Met in either AD type (147 μmol/L) or carriers (80 μmol/L). During follow-up, increased Met levels were usually paralleled with elevated Met/Phe ratio (.94–44.7) and tHcy (13.0–65.5 μmol/L). Ten patients had plasma Met levels exceeding 800 μmol/L at least once during follow-up (Figure 1C) and then intervened with low-Met formula, diets appropriate for their ages without an excess of proteins, or drug administration. According to their latest evaluation, they had normal growth and development. A wider range of phenotypes were presented in these 10 patients, such as increased liver enzymes, verbal difficulty, motor delay, advanced bone age, developmental delay, and white matter lesions (Table 1). All these clinical symptoms were temporarily observed. Additionally, acidosis, hyperammonemia, or hypoglycemia were absent during follow-up, irrespective of infection. However, we also found asymptomatic cases harboring biallelic MAT1A variants. For example, P34, harboring novel compound heterozygous variants c.547C>G/p. Q183E and c.242G>A/p. R81Q had an initial Met 60 μmol/L that was below the cut-off value (81.7 μmol/L), a negative result according to biochemical indicators in NBS.

As MATD is a rare metabolic disorder with a limited population, detailed records of patients with neurological abnormalities are important evidence for better management of other cases in the future. Here, five patients with a Met value >800 μmol presented as follows: patients P16, P25, P27, and P31 had neurological abnormalities, and P35 carried 3 alleles with the same allele (c.695C>T/p. P232L) in P16.

Patient 16 clinically best fits with the autosomal recessive biallelic population that only one de novo variant c.695 C>T/p. P232L identified. Further evaluation of the intronic allele is needed. He had an initial Met 257 μmol/L, a high ratio of Met/Phe 4.58, and elevated plasma tHcy (59 μmol/L) with persistent hypermethioninaemia during follow-up. The recall Met was increased to 1283 μmol/L in the first month of life. Immediately, a diet with restricted protein and vitamin B6 was prescribed. At 2 months of age, the low-Met formula was added at a ratio of 5:1 to ordinary infant milk powder. The plasma Met levels during treatment were maintained at 403–599 μmol/L with a slight elevation of tHcy 17.6–19.0 μmol/L. At 20 months old, he had an intelligence score of 82 and an action score of 92 according to the Bailey Scale of Infant Development (BSID). Due to difficulties in persisting in a low-met restriction diet when grown, Met levels fluctuated within 1110–1165 μmol/L after 3 years of age. At age 7 years 9 m, brain MRI examination revealed a large range of abnormally high intensity in T2-weighted images with a reduced diffusion of white matter (Figure 2A), which has been observed in patients with phenylketonuria and glutaric aciduria type I (Manara et al., 2009; Tsai et al., 2017). We then were told that he had liver transplantation at the age of 8 years 1 m in another hospital. After transplantation, he picked up the normal diet and had a plasma Met of 74.2 μmol/L, 78.5 μmol/L, and 109.6 μmol/L at ages 8 years 4 m, 8 years 7 m, and 10 years 11 m, respectively. The mean plasma Met dramatically dropped from 844.1 μmol/L to 96.9 μmol/L (Figure 2B). Strikingly, the white matter lesions were gradually reversed to normal after the operation.

Patient 25 was homozygous for a novel variant c.875G>T/p. R292L with transient verbal difficulty and weakness in the right foot when walking. She had a remarkable increase in Met levels of 921 μmol/L when recalled for confirmation and then started on a mixture of low-Met formula and breastfeeding with betaine and folate. The Met value dropped to 599 μmol/L at 8 m of age. However, she could not adhere to the Met-restricted diet after 1 year of age. Fluctuated Met (628–850 μmol/L) values were observed with intermittent interventions. However, her development status was normal on the ASQ test, and both height and weight were P75 at follow-up at the age of 42 m.

Patient 27 was homozygous for a reported variant c.769G>A/p. G257R with persistent hypermethioninaemia. During the first 3 months of life, his Met values were persistent at 1069, 1097, and 1408 μmol/L, respectively, with a low-Met formula diet. At the age of 10 m, he had a motor delay with a gross motor score of 10 and a fine motor score of five on the ASQ evaluation. After SAMe supplementation and rehabilitation training, his ASQ score climbed to normal at 48 m of age, with a gross motor score of 60 and a body weight of P50. However, MRI showed mild hyperintense T2-weighted images in the white matter of cerebral hemispheres and widened bilateral frontotemporal extracerebral space (Figure 2C) at 4 years 11 m of age with a Met level of 839 μmol/L.

Patient 31 had compound heterozygous variants, c.769G>A/p. G257R and c.875G>T/p. R292L. With a Met-restricted diet, his Met was decreased from 977 μmol/L at age 1 m to 159 μmol/L at age 5 m and then manifested <500 mol/L. A transient development delay was observed at age 2 years, with a developmental quotient of 68 by the Gesell system, but disappeared at follow-up with a P50 score in both his height and weight at 5 years 10 m.

Patient 35 harbored three alleles, a paternal c.695C>T/p. P232L variant and two novel maternal variants c.608T>A/p. I203N and c.922G>C/p. A308P, with an initial Met of 183 mol/L. his Met value was 600 μmol/L and then increased to 808 μmol/L at the age of 2 m. Given low Met intake and betaine treatment, the Met fluctuated between 543-595 μmol/L until the age of 13 m, with normal growth and development. However, his Met increased up to 713–982 μmol/L after the protein restriction diet was discontinuous. Both his height and weight were P75 as of follow-up at the age of 108 m.

Genetic characteristics in the AR type were more heterogeneous than the dominant type. Biallelic variants were genotyped in 18 patients, and three alleles occurred in P35. Sixteen patients harbored compound heterozygotes, while only 3 patients carried homozygotes, homozygous c.1070C>T/p. R357L in P22, homozygous c.875G>T/p. R292L in P25, and homozygous c.769G>A/p. G257R in P27 (Table 1). Surprisingly, each patient had a unique genotype. A total of 26 variants were identified, including 24 missense variants and two truncating variants (c.38dupT/p. L13Lfs*15 and c.765_768del CCAG/p. P255Pfs*35) and distributed randomly in genes and protein domains, except in exons 2 and 9 (Figure 3A). Notably, 17 novel variants in patients were found and predicted by automatic tools (Supplementary Table S2), of which c.242G>A/p. R81Q occurred in two asymptomatic individuals (P33 and P34), indicating a benign protein function as prediction in silico. However, the novel truncating variants were postulated to be pathogenic (P) or likely pathogenic (LP) according to ACMG recommendations (Richards et al., 2015). Most variants had high conservation in mammals (Figure 3B). Variants distributed in patients as c.188G>T/p. G63V in five individuals, c.895C>T/p. R299G in four individuals, c.769G>A, c.875G>T, and c.242G>A in two individuals each, and others appeared once.

Thirty-one carriers (21 females and 10 males) were detected by the cutoff index in NBS, as shown in Table 2. Three carriers, C2, C14, and C25, were born prematurely. As shown in Figure 1, the initial Met values of each carrier were <100 μmol/L, ranging from 45 to 95 μmol/L with a mean value of 73 μmol/L. Six previously reported variants occurred in 18 individuals, distributed as c.1070C>T in eight individuals, c.769G>A in five individuals, c.773A>G in two individuals, and c.188G>T, c.271G>C and c.964A>G each in one child. Intriguingly, 8 novel variants were detected in 13 individuals, including one splicing and eight missense variants, of which c.181A>C/p.K61Q appeared in both carrier (C6) and patient (P20) as a recessive variant (Supplementary Table S2). As it is difficult to distinguish whether the novel variants were pathogenic in the dominant type or not, parents were advised to perform gene testing and Met assays to check whether parents harboring the same novel variant have a normal Met value. As shown in Table 2, parents’ Met values harboring novel variants ranged from 19-50 μmol/L with a low ratio of Met/Phe (<1.1) and were healthy, indicating those 6 novel variants, c.67G>A, c.181A>C, c.251G>C, c.292 + 5G>A, c.748T>C, c.748T>C, and c.855G>C, were transmitted as recessive type. Unfortunately, the parents harboring c.179G>A/p. C60Y or c.180C>G/p. C60W did not consent to the test. Cautiously, to avoid missing any AD cases, all carriers were followed up until the Met value was <50 μmol/L, or the Met value decreased continuously at the next three tests after NBS. Interestingly, most of the carriers had normal Met values <50 μmol/L by the age of 10 m, and their growth and development statuses were normal and healthy.