Table 1.
Data describing patients' history of disease, muscle and cognitive functions (quantitative variables are mean±SD).
Figure 1.
DMD patients were assigned to 4 clusters in the final hierarchical cluster analysis (linkage: Ward's method, metric: squared Euclidean distance) using CatPCA-derived XY coordinates in a plane defined by the cognition and motor function principal components (axes) (Cronbach's α: 0.859 and 0.721, respectively). Cognition was always altered to various extent in patients with early infantile DMD (A) and classical DMD (B), but was preserved in patients with moderate (C) and severe (D) pure motor DMD who differed markedly in motor function impairment. Overall, cluster A patients were most severely affected.
Table 2.
Variables' contributions to the two principal components (Cronbach's α, a measure by default for the internal consistency of categorical principal components in the SPSS software, is 0.891 and 0.721), absolute values above 0.500 were retained to define the principal components.
Figure 2.
Variables constituting the “cognition” (left) and the “motor function” (right) principal components.
Group A patients manifested early in life by psychomotor delay. “Cognitive status” measures increased from cluster A to D (the following classification was used: (I) severely mentally retarded; (II) mildly mentally retarded; (III) borderline; (IV) normal). This was also supported by an identical increase of “maximal education level” and an inversely symmetrical decrease of “school delay” (p<0.00001 for both, data not shown). Accordingly, 86% of D patients attended an ordinary educational establishment vs. 38% of C, 26% of B and 21% of A patients (all p<0.007 by Fischer exact test). A and C patients had values at each extremity of the motor function spectrum, also comprising the age of becoming “wheelchair-ridden” (p<0.00001, not shown). “Delay of diagnosis” contributed to the integrity of the second principal component and, notably, to the delineation of cluster D (indicated p values are obtained by Kruskal-Wallis test; clusters were compared by a multiple range post-test with p<0.05).
Figure 3.
Muscle strength variables not used to constitute the model.
C patients had better muscle strength than A, B, or D patients and a retarded onset of scoliosis. These motor parameters, not used in the multivariate analysis, support the validity of the model. “Age at onset of physical therapy” trailed behind the evolution of muscle strength and motor function: no clear benefit from therapy could be demonstrated in this series (indicated p values are obtained by Kruskal-Wallis test; clusters were compared by a multiple range post-test with p<0.05).
Figure 4.
Pulmonary and cardiac function variables not used to constitute the model.
Group A patients had poorest respiratory and myocardial functions and highest serum CK levels at diagnosis. These variables, not used in the multivariate model, also corroborate its predictive relevance (indicated p values are obtained by Kruskal-Wallis test; clusters were compared by a multiple range post-test with p<0.05).
Figure 5.
Genotype/phenotype correlations.
Proportion of patients with a mutation upstream to exon 30 steadily increased from group A to D. This ascent correlated with spared cognition (mental status: p<0.0003) but not with motor function (age at ambulation loss: NS) (Fisher's exact test). Expectedly, the 3 patients with mutation after exon 63 affecting the brain specific DP71 transcript were classified in group A.
Table 3.
Simplified indicators of congenital DMD (cluster A) derived from series 1.
Table 4.
Simplified indicators of moderate pure motor DMD (cluster C) derived from series 1.
Table 5.
Simplified indicators applied to congenital and to moderate pure motor DMD in series 2.