The face signature of fibrodysplasia ossificans progressiva

Abstract

Fibrodysplasia ossificans progressiva (FOP) causes extensive heterotopic bone formation due to heterozygous mutations in the glycine-serine activation domain of ACVR1 (ALK2), a bone morphogenetic protein type I receptor. Anecdotal observations of facial similarity have been made by clinicians and parents, but no objective quantitative analysis of the faces of FOP patients has ever been undertaken. We delineated the common facial characteristics of 55 individuals with molecularly confirmed FOP by analyzing their face signature (face shape difference normalized against age and sex matched controls) and associated face signature graphs (with face signatures as vertices and adjacency corresponding to greatest similarity). Our analysis identified 10 affected individuals whose face signature is more homogeneous than others with FOP. This distinct subgroup showed the previously identified reduced mandible as well as newly identified features: underdevelopment of the upper orbit/supra-orbital ridge; infra-orbital prominence; and, low-set ears. These findings strongly suggest that the canonical FOP mutation variably affects the postnatal morphogenesis of the normotopic cranial skeleton in the upper midface and mandible and may have important diagnostic and functional implications.

Figure 1

A: Landmarks used in dense surface model construction; B and C: construction of reference normal (black arrow) to plane containing line exocanthion-exocanthion and reference line (red) joining mid-points of exocanthion-exocanthion and preaurale-preaurale; D: estimate of vertical displacement (white arrow) of preaurale-preaurale for an individual compared to the overall same-sex mean control (projection of broken red line on to black arrowed line); E and F: construction of diagonal face depth (double arrowed white line) to estimate mandibular underdevelopment.

Figure 2

Face signature clusters of FOP patients

Figure 3

Face signature graph for FOP patients

Figure 4

Face signature graph for controls and FOP patients

Figure 5

Colored coded version of face signature graph for controls and FOP patients

Figure 6

Collapsed version of face signature graph for controls and FOP patients. Control and FOP signatures are shown respectively as empty and black filled circles

Figure 7

A: For DSM_male, PC1 (in standard deviations) is plotted against age (in years). The PC1 axis is annotated with the model’s mean face morphed to its shape at −3, 0 and +3 st devs. A running mean value of PC1 for 50 consecutively aged controls approximates mean facial growth (continuous grey line). Outliers are labelled. B: For DSM_male, PC2 (in standard deviations) is plotted against age (in years). The PC1 axis is annotated with the model’s mean face morphed to its shape at −5, −3, 0 and +3 st devs. A running mean value of PC2 for 50 consecutively aged controls approximates mean facial growth (continuous grey line). C: For DSM_male, horizontal mandible reduction is plotted against age (in years). Outliers are labelled. D: Normalized male vertical ear displacement (std. dev.) is plotted against age (years). Outliers are labelled.

Figure 8

A: For DSM_female, PC1 (in standard deviations) is plotted against age (in years). The PC1 axis is annotated with the model’s mean face morphed to its shape at −3, 0 and +3 st devs. A running mean value of PC1 for 50 consecutively aged controls approximates mean facial growth (continuous grey line). Outliers are labelled. B: For DSM_female, PC2 (in standard deviations) is plotted against age (in years). The PC1 axis is annotated with the model’s mean face morphed to its shape at −5, −3, 0 and +3 st devs. A running mean value of PC2 for 50 consecutively aged controls approximates mean facial growth (continuous grey line). C: For DSM_female, horizontal mandible reduction is plotted against age (in years). Outliers are labelled. D: Normalized female vertical ear displacement (std. dev.) is plotted against age (years). Outliers are labelled.