Clinical, biochemical and molecular phenotype of congenital disorders of glycosylation: long-term follow-up

Congenital disorders of glycosylation (CDG), first reported in 1980, result from defects in the synthesis of glycans and the attachment of glycans to proteins and lipids [1]. According to the current nomenclature, CDG can be classified into defects in protein N-glycosylation, O-glycosylation, glycosphingolipid and glycosylphosphatidylinositol anchor glycosylation defects, and multiple glycosylation pathways defects. Since the description of phosphomannomutase 2 deficiency (PMM2-CDG), more than 130 other CDG subtypes have been reported [2, 3].

So far, single case reports and case series regarding various CDG have been published, but the data regarding follow-up are sparse in the literature. Our study aimed to describe the clinical, biochemical, including serum Tf isoform analysis, as well as molecular features of patients diagnosed with CDG in one referral center, and to present their long-term follow-up.

During 1995–2019 years, a total number of 22,063 serum Tf isoform analysis, have been done in our Institute. Serum Tf isoforms were analyzed by isoelectrofocusing (IEF) agarose gel electrophoresis according to our modification [4, 5] of the method described by Van Eijk et al. [6]. Serum Tf consists of the mixture of isoforms and has two iron binding sites, which was saturated with iron (20 μl serum with 80 μl 0,9% NaCl, 2 μl 10 mM Fe(III) citrate and 2 μl 0.1 M NaHCO₃). Transferrin migrated through 1% agarose gel (45 mg of agarose were added to 4.5 ml of distilled water) with 5% ampholines in a pH range of 5.0–7.0 on a Multiphore 2117 apparatus (LKB) with a modified electrode lid. Tf isoforms were visualized by immunofixation (200 µl polyclonal rabbit anti-human transferrin serum) and 0.5% Coomasie Brillant Blue solution staining, as presented in Fig. 1. The percentage of Tf fractions was assessed densitometrically and carbohydrate deficient transferrin value (% CDT) was measured as a sum of asialo-, monosialo-, and disialo-Tf isoforms.

Serum O-glycoprotein apolipoprotein C-III (ApoC-III) isoforms were analyzed by isoelectrofocusing polyacrylamide gel electrophoresis according to the method described by Wopereis et al. [7]. The activity of phosphomannomutase 2 (PMM2; EC 5.4.2.8) and mannosephosphate isomerase (MPI; EC 5.3.1.8) in fibroblasts were determined according to the method of Van Schaftingen and Jaeken [8]. Phosphoglucomutase 1 (PGM1) activity was assayed spectrophotometrically on cell extracts [9, 10]. Lipid-linked oligosaccharide (LLO) profiles in fibroblasts were analyzed by high-performance liquid chromatography (HPLC) [11‐13].

The molecular analysis using direct sequencing of single genes (PMM2, MPI, PGM1) was performed to confirm the proper biochemical diagnosis. In a group of patients with no diagnosis established based on abovementioned diagnostic procedures, whole-exome sequencing (WES) was applied. The nomenclature of identified variants and patients’ genotype follows the Human Genome Variation Society guidelines (HGVS v 2.0, www.​hgvs.​org/​mutnomen) and referral according to cDNA and protein sequences of various CDG genes followed the Human Gene Mutation Database (HGMD, www.​hgmd.​cf.​ac.​uk).

Finally, 32 patients with molecularly confirmed congenital disorders of N-glycosylation and combined N- and O-hypoglycosylation, diagnosed and followed-up, have been recruited to the study.

The chart review of patients’ medical records concerning the demographics, first presented signs and symptoms, age at diagnosis, diagnostics methods (described below), clinical outcome, as well as biochemical (serum Tf isoforms, aspartate (AST) and alanine (ALT) aminotransferases, international normalized ratio (INR)), and molecular data were collected. Ethical approval was obtained from the Children’s Memorial Health Institute Bioethical Committee, Nr 23/KBE/2020, Warsaw, Poland.

An extended statistical analysis was performed. First, in order to find correlations between % of Tf isoforms, we calculated Pearson’s linear correlation coefficient r between all Tf isoforms. Next, we compared the subcohort of PMM2-CDG-I and non-PMM2-CDG-I patients in terms of the mean of % Tf isoforms. We also performed chi-squared test for independence to establish if diagnostic variables, such as muscle hypotonia, hypothyroidism, nystagmus, etc., are independent of the type of CDG-I disease. Finally, we investigated the effect of treatment administrated in three patients on the % Tf isoforms. We compared means (or single values) measured before treatment with the means of measurement taken after the first administration of mannose administration (in one patient with CDG-I/CDG-II) or galactose administration (two patients with non-PMM2-CDG-I).

The alpha level of significance was chosen 0.05. For computations we used statistical software “R”.

The paper presents the phenotypic and genotypic spectrum of 32 CDG patients during a long-term (over 20 years) observation in one center.

PMM2-CDG was the most common (40%) CDG identified in our study, similarly to the literature [16, 17]. Among the N-hypoglycosylation disorders, ALG6-CDG is reported as the second, while ALG1-CDG as the third most frequent type in the literature [18]. We did not identify any ALG6-CDG patients, whereas SRD5A3-CDG was the second most frequent type. ALG1-CDG, ALG13-CDG and MPI-CDG were comparably frequent.

Prenatal presentation (NIHF) was observed in three (9%) of our patients, including two PMM2-CDG (one died at two months of age). In a recent review by Makhamreh et al. (2020), the most common CDG associated with NIHF were PMM2-CDG, ALG9-CDG, and ALG8-CDG, which follows the line of our results [19]. Our paper is one of the first to report NIHF associated with ATP6AP1-CDG. Recently, thickened nuchal translucency and large fluid filled space with septations along the foetal spine have been described during the first trimester ultrasound in ATP6AP1-CDG foetus [20].

CDG patients frequently present neurological involvement [21]. The main neurological symptom in our patients remained psychomotor retardation and cerebellar ataxia with cerebellar hypoplasia on brain MRI scans, observed in 83% of PMM2-CDG patients and all SRD5A3-CDG patients. This finding confirms the thesis that the cerebellum is regularly involved in PMM2-CDG and SRD5A3-CDG [14, 25‐27]. Epilepsy is also a common symptom (almost one-third of patients with PMM2-CDG in the study by Monin et al. [22]), in our cohort it was observed in ALG13-CDG patients. There are others CDG in which epilepsy is severe and difficult to control, like DPM1-CDG, DPM2-CDG, MPDU1-CDG, ALG2-CDG, ALG12-CDG, ALG8-CDG, ALG9-CDG, ALG11-CDG, and RTF1-CDG [23, 24].

Liver involvement was present in about 22% of all CDG in the last review by Marques-da-Silva et al. (2018) [28]. In our cohort, the predominant liver phenotype was observed in MPI-CDG patients, while other CDG like PMM2-CDG, ATP6AP1-CDG were associated with liver disease. Up to now, 14 patients have been reported with X-linked ATP6AP1 deficiency and the key features were immunodeficiency and liver involvement ranging from a mild elevation of serum transaminases to liver failure [15, 29, 30].

Our results are in agreement with the thesis that selected CDG have unique characteristics, which may facilitate and shorten their recognition. These include: connective tissue involvement in ATP6AP1-CDG, midline malformations in PGM1-CDG, chronic diarrhoea in MPI-CDG, inverted nipples and abnormal fat distribution in PMM2-CDG, cataract/coloboma in SRD5A3-CDG, cerebellar hypoplasia in PMM2-CDG and SRD5A3-CDG [31, 32]. Moreover, some of CDG are noted to manifest specific craniofacial dysmorphy which, if consistent with the entire clinical picture, allows for verification of the diagnosis. It refers especially to PMM2-CDG, manifesting with microcephaly, prominent forehead, flat nasal bridge, thin upper lip and large ears [31]. Recognizable facial features have been described in a number of other CDG [33, 34].

The clinical outcome of PMM2-CDG varies among patients [25]. Inverted nipples and abnormal fat distribution were reported as its characteristic features and were present in the majority of our patients. Pericardial effusion was noted in about 30% of PMM2-CDG patients [35]; in our cohort 50% of PMM2-CDG patients. Cardiac involvement in PMM2-CDG seems to be clinically relevant [35]; in our group five PMM2-CDG patients had cardiomyopathy, while it was a rare symptom in non-PMM2-CDG patients.

Abnormal thyroid function was reported in approximately 75% of PMM2-CDG patients [34]; we observed it with a comparable frequency. It results from an abnormal glycosylation of thyreotropine and thyroid-binding globulin [36].

Renal involvement was presented in 17% of PMM2-CDG patients from our cohort vs 6% of PMM2-CDG patients reported in the literature [36]. Our patients had proteinuria, and one of them congenital nephrotic syndrome. Like in the literature, mild proteinuria is the most common renal abnormality in those patients [37].

Some CDG have peculiar phenotypes being a combination of features. In our cohort, MPI-CDG manifested as purely hepatic (hepatomegaly, elevated liver transaminases) or hepato-intestinal (symptoms as above with recurrent diarrhoea) disease. In SRD5A3-CDG patients, cerebellar ataxia and signs of visual impairment and variable eye malformations, including optic disc hypoplasia, nystagmus, were the characteristic features. ATP6V0A2-CDG, as well as COG7-CDG, were described as cutis laxa syndrome with defective protein glycosylation [30, 38]. About 50% of patients with ATP6V0A2-CDG showed cortical malformations, especially pachygyria, in brain MRI. Our patient with ATP6V0A2-CDG showed cutis laxa, pachygyria, as well as polymicrogyria, which expands the spectrum of brain MRI features.

Isoelectric focusing (IEF) of serum transferrin (Tf) is still the method of choice for the diagnosis of N-glycosylation disorders associated with sialic acid deficiency. Our study showed for the first time that ALG13-CDG Tf IEF was abnormal in two our patients, while in ALG1-CDG one patient had normal Tf IEF. There were no reports on that in the literature. In every patient with clinical and biochemical diagnosis of CDG, molecular analysis has to be performed. It is essential to confirm the diagnosis and also to predict the possible genotype–phenotype correlation. In our study, WES analysis enabled the diagnosis of ALG13-CDG in patients with early-onset drug-resistant epileptic encephalopathy.

The most common pathogenic variants in the PMM2 gene were c.691G > A, p.Val231Met and c.422G > A, p.Arg141His. The latter is also the most commonly reported in the literature [39]. The compound heterozygotes for p.Arg141H is noted in our study, and in the literature as well, were associated with a mild phenotype.

Two deceased patients with ALG1-CDG were compound heterozygous for the c.773C > T, p.Ser258Leu. This variant, if present in the homozygous state, is regarded to be related with an early fatal outcome.

Three out of four patients with SRD5A3-CDG presented with an improvement of motor skills and speech development, despite genetic alteration known to results in poor prognosis -c.292_293del, p.Leu98ValfsX121 (in two of them).

The data regarding follow-up of CDG is sparse in the literature. Our PMM2-CDG and SRD5A3-CDG patients showed an improvement in motor skills as well as no clinical progression of cerebellar symptoms. In ATP6AP1-CDG some symptoms aggravated with age (from sensorineural hearing loss to total deafness, from hair loss to total alopecia). ALG1-CDG forecast rather a poor outcome with death in the first months of life. Patients with MPI-CDG on mannose supplementation improved clinically and their Tf isoforms levels reached values close to the reference range. Patient with PGM1-CDG on galactose supplementation had died despite the improvement of TF isoforms.