Frontonasal dysmorphology in bipolar disorder by 3D laser surface imaging and geometric morphometrics: comparisons with schizophrenia

Abstract

Any developmental relationship between bipolar disorder and schizophrenia engenders continuing debate. As the brain and face emerge in embryological intimacy, brain dysmorphogenesis is accompanied by facial dysmorphogenesis. 3D laser surface imaging was used to capture the facial surface of 13 male and 14 female patients with bipolar disorder in comparison with 61 male and 75 female control subjects and with 37 male and 32 female patients with schizophrenia. Surface images were analysed using geometric morphometrics and 3D visualisations to identify domains of facial shape that distinguish bipolar patients from controls and bipolar patients from those with schizophrenia. Both male and female bipolar patients evidenced significant facial dysmorphology: common to male and female patients was overall facial widening, increased width of nose, narrowing of mouth and upward displacement of the chin; dysmorphology differed between male and female patients for nose length, lip thickness and tragion height. There were few morphological differences in comparison with schizophrenia patients. That dysmorphology of the frontonasal prominences and related facial regions in bipolar disorder is more similar to than different from that found in schizophrenia indicates some common dysmorphogenesis. Bipolar disorder and schizophrenia might reflect similar insult(s) acting over slightly differing time-frames or slightly differing insult(s) acting over a similar time-frame.

Fig. 1

Visualisation of regression models for facial landmark analyses, separately for (A) male and (B) female bipolar patients vs. controls in (upper left) coronal, (upper right) sagittal and (bottom) axial planes. The Procrustes mean coordinates of facial landmarks for the pooled sample of subjects for each sex are joined by dashed black lines; the bipolar coordinates (the coordinates of a hypothetical bipolar patient of that sex lying along the bipolar–control discrimination axis) are joined by red lines that exaggerate the features of “patientness”, i.e. the difference of the patient from the sample mean, by a factor of approximately 5 to render them visible. Right inset: typical laser surface image (see Hennessy et al., 2007) showing the twenty-six 3D landmarks. Filled circles: the 24 landmarks used in landmark analysis. Open circles with crosses: the 2 landmarks used, in addition to the 24 landmarks, to calculate the pseudo-landmarks. Landmarks: a, soft tissue nasion; b, pronasale; c, sublabiale; d, pogonion; e/f, inner canthus; g/h, outer canthus; i/j, alar crest; k, subnasale; l/m, alare; n/o, columella breakpoint; p/q, christa philtrum; r, labiale superius; s, labiale inferius; t, stomion; u/v, cheilion; w/x, tragion; y/z, otobasion inferius.

Fig. 2

Visualisation of regression models for facial surface analyses, separately for (A) male and (B) female bipolar patients vs. controls. Procrustes mean shape is shown displaced equally in each direction along the bipolar–control discrimination axis. The displacement positions from mean facial shape (middle column) are exaggerated approximately 5-fold. This is equivalent to the patient (left column) — control (right column) dimorphism exaggerated approximately 10-fold.

Fig. 3

Visualisation of displacement vectors for each point on the surface, separately for (A) male and (B) female bipolar patients vs. controls. Left column: change in surface area coded as red [expanded in bipolar disorder relative to controls] or blue [contracted in bipolar disorder relative to controls]; data are coded as 33% percentiles, with darker colours indicating upper percentiles. Centre column: change in angle of displacement vector to surface normal coded as red [angled outward in bipolar disorder relative to controls] or blue [angled inward in bipolar disorder relative to controls]; darker colours code smaller angle to the normal, i.e. displacement vector more perpendicular to the surface. Right column: length of displacement vector, in any direction, coded as red; darker colour codes greater vector length, i.e. greater difference between bipolar disorder and controls.