Genotype/Phenotype Relationship: Lessons From 137 Patients With PMM2-CDG

Abstract

We report on the largest single dataset of patients with PMM2-CDG enrolled in an ongoing international, multicenter natural history study collecting genetic, clinical, and biological information to evaluate similarities with previous studies, report on novel findings, and, additionally, examine potential genotype/phenotype correlations. A total of 137 participants had complete genotype information, representing 60 unique variants, of which the most common were found to be p.Arg141His in 58.4% (n = 80) of participants, followed by p.Pro113Leu (21.2%, n = 29), and p.Phe119Leu (12.4%, n = 17), consistent with previous studies. Interestingly, six new variants were reported, comprised of five missense variants (p.Pro20Leu, p.Tyr64Ser, p.Phe68Cys, p.Tyr76His, and p.Arg238His) and one frameshift (c.696del p.Ala233Argfs∗100). Patient phenotypes were characterized via the Nijmegen Progression CDG Rating Scale (NPCRS), together with biochemical parameters, the most consistently dysregulated of which were coagulation factors, specifically antithrombin (below normal in 79.5%, 93 of 117), in addition to Factor XI and protein C activity. Patient genotypes were classified based upon the predicted pathogenetic mechanism of disease-associated mutations, of which most were found in the catalysis/activation, folding, or dimerization regions of the PMM2 enzyme. Two different approaches were used to uncover genotype/phenotype relationships. The first characterized genotype only by the predicted pathogenic mechanisms and uncovered associated changes in biochemical parameters, not apparent using only NPCRS, involving catalysis/activation, dimerization, folding, and no protein variants. The second approach characterized genotype by the predicted pathogenic mechanism and/or individual variants when paired with a subset of severe nonfunctioning variants and uncovered correlations with both NPCRS and biochemical parameters, demonstrating that p.Cys241Ser was associated with milder disease, while p.Val231Met, dimerization, and folding variants with more severe disease. Although determining comprehensive genotype/phenotype relationships has previously proven challenging for PMM2-CDG, the larger sample size, plus inclusion of biochemical parameters in the current study, has provided new insights into the interplay of genetics with disease. Trial Registration: NCT03173300.

Keywords: PMM2; congenital disorders of glycosylation; genetic variations; genotype/phenotype correlations; inherited metabolic disorders; natural history; phosphomannomutase 2-CDG.

Figure 1

Heat map showing the scores for all participants as indicated by each rectangle, which represents one participant for each NPCRS item (a higher score reflects more disease severity). Thus, when the scores are viewed together, the severity of impairment can be seen. For example, mobility tends to be mostly red and orange, indicating severe mobility impairment in most patients.

Figure 2

Structure of PMM2 by variant analysis using the model from Briso-Montiano et al. [23]. PyMol2 software has been used to highlight the mutated residues. (a) Front view of the dimer with the 60 pathogenic variants in colors according to their categorization (see (b)). (b) Categorization of the 60 pathogenic variants identified in the natural history study.

Figure 3

Coagulation parameters by activity categories for categories with at least four participants. (a) Antithrombin activity, (b) protein C activity, (c) Factor XI activity, and (d) Factor IX levels.

Figure 4

Liver function by activity categories for categories with at least four participants. (a) Alkaline phosphatase, (b) alanine transaminase, (c) aspartate aminotransferase, and (d) ceruloplasmin.

Figure 5

Heat map showing the genotype/phenotype relationships found in the current study.