An investigation of a novel broad autism phenotype: increased facial masculinity among parents of children on the autism spectrum

Diana Weiting Tan, Syed Zulqarnain Gilani, Gail A. Alvares, Ajmal Mian, Andrew Whitehouse, Murray T. MayberyProceedings of the Royal Society B Biological Sciences 2022 · DOI

TL;DR

This study investigates the "broad autism phenotype" (BAP) by examining facial masculinity in parents of autistic children. They found that parents of autistic children, both mothers and fathers, exhibit increased facial masculinity compared to parents without a family history of autism, suggesting a heritable physical manifestation of the BAP.

What Problem Does This Paper Address?

Autism Spectrum Disorder (ASD) is a neurodevelopmental condition with a strong genetic component. The "broad autism phenotype" (BAP) describes sub-clinical autistic-like traits observed in relatives of individuals with ASD. While BAP has been primarily studied in terms of behavior and cognition, recent research suggests it may also manifest physically, specifically in facial morphology. Previous studies have found increased facial masculinity in autistic children and their non-autistic siblings. This paper aims to extend this line of inquiry by investigating whether this facial masculinity is also present in the non-autistic parents of autistic children, thereby exploring the potential heritability and maternal/paternal contributions to this physical aspect of the BAP.

Methods

The researchers employed a sophisticated approach involving:

  • Three-dimensional facial photography: High-precision 3D facial scans were captured for all participants.
  • Gender classification and scoring algorithm: A machine learning approach, previously established and validated, was used. This involved:
  • Landmark identification: 21 key facial landmarks were identified on each 3D scan.
  • Distance measurements: Both Euclidean (straight-line) and geodesic (surface) distances between these landmarks were calculated.
  • Feature selection: A Gradient-Based Efficient Feature Selection (GEFS) algorithm was used to identify the most discriminating facial features for gender classification.
  • Linear Discriminant Analysis (LDA): An LDA classifier was trained using the selected features to distinguish between male and female faces.
  • Facial Masculinity Score: A score was computed for each face based on its deviation from the mean male and female facial measurements, ranging from 0 (highly feminine) to 1 (highly masculine).
  • Statistical Analyses:
    • Correlation analyses: To assess potential confounding effects of age and facial area on masculinity scores and distances.
  • ANCOVA: Analysis of Covariance was used to compare facial masculinity scores and distances between the "parents of autistic children" group and the "parents of non-autistic children" group, while controlling for age and facial area.
  • LDA for diagnostic classification: A separate LDA model was used to assess how well the identified facial features could classify the autism diagnostic status of the children.

Datasets

The study utilized data from two groups of participants:

  • Parents of autistic children: 192 individuals (58 males, 134 females) recruited from the Telethon Kids Institute.
  • Parents of non-autistic children (comparison group): 163 individuals (50 males, 113 females) recruited from the general population.

All participants were of Caucasian descent and screened for syndromic disorders or significant facial trauma. The paper does not introduce a new benchmark dataset but rather uses existing methods applied to a novel participant group.

Key Insights and Ideas

  • Facial Masculinity as a Heritable BAP Trait: The study provides strong evidence that increased facial masculinity is not confined to autistic individuals or their siblings but is also present in their parents. This suggests that facial masculinity could be a heritable physical manifestation of the broad autism phenotype.
  • Maternal and Paternal Contributions: The finding of increased facial masculinity in both fathers and mothers of autistic children is significant. It implies that both maternal and paternal genetic or biological pathways may contribute to this physical trait associated with ASD.
  • Sex-Specific Differences in Facial Features: As expected, the study confirmed significant sex differences in facial morphology, with males generally exhibiting more masculinized features than females. However, the key finding of increased masculinity in parents of autistic children held true regardless of sex, highlighting the robustness of the BAP association.
  • Predictive Power of Facial Features: The identified facial features showed a moderate accuracy (74.3%) in classifying the autism diagnostic status of the children, suggesting that facial morphology, in conjunction with other factors, could potentially contribute to early identification or risk assessment.

Key Results

  • Facial Masculinity Scores: Parents of autistic children had significantly higher facial masculinity scores (mean difference of 0.16, p < 0.001) compared to parents of non-autistic children. This effect size was small but statistically significant.
  • Facial Distances: Several facial distances were also significantly larger (more masculine) in parents of autistic children, with effect sizes ranging from small to medium (0.16 ≤ d ≤ .41).
  • Gender Classification Accuracy: The LDA model, using 10 selected facial features, achieved 95.7% accuracy in classifying male and female faces.
  • Autism Diagnosis Classification Accuracy: A separate LDA model using seven key facial variables achieved 74.3% accuracy in classifying the autism diagnostic status of the children.

Limitations and Future Work

  • No Independent Training/Testing Sets: The authors acknowledge that they did not have access to separate datasets for training and testing their classification algorithm, relying instead on iterative division of their sample. They suggest future replication studies are warranted, particularly using publicly available datasets that include family history information.
  • Caucasian Sample: The study participants were restricted to those of Caucasian descent. The authors emphasize the need for future research to investigate whether these findings of facial masculinization associated with autism generalize to other ethnic populations.
  • Modest Diagnostic Classification Accuracy: While 74.3% accuracy is notable, it suggests that facial masculinity is likely one of several facial phenotypes associated with autism. Future work could improve classification by incorporating other facial features, such as facial asymmetry (which has also been linked to ASD in previous studies by the same research group).
  • Simplex vs. Multiplex Families: The authors propose comparing facial masculinity between members of simplex (one family member with ASD) and multiplex (multiple family members with ASD) families to explore how the effect of masculinity might increase with genetic liability for autism.

Summary updated

Abstract

The broad autism phenotype commonly refers to sub-clinical levels of autistic-like behaviour and cognition presented in biological relatives of autistic people. In a recent study, we reported findings suggesting that the broad autism phenotype may also be expressed in facial morphology, specifically increased facial masculinity. Increased facial masculinity has been reported among autistic children, as well as their non-autistic siblings. The present study builds on our previous findings by investigating the presence of increased facial masculinity among non-autistic parents of autistic children. Using a previously established method, a 'facial masculinity score' and several facial distances were calculated for each three-dimensional facial image of 192 parents of autistic children (58 males, 134 females) and 163 age-matched parents of non-autistic children (50 males, 113 females). While controlling for facial area and age, significantly higher masculinity scores and larger (more masculine) facial distances were observed in parents of autistic children relative to the comparison group, with effect sizes ranging from small to medium (0.16 ≤ <i>d</i> ≤ .41), regardless of sex. These findings add to an accumulating evidence base that the broad autism phenotype is expressed in physical characteristics and suggest that both maternal and paternal pathways are implicated in masculinized facial morphology.
BibTeX
@inproceedings{tan2022investigation,
  author = {Diana Weiting Tan and Syed Zulqarnain Gilani and Gail A. Alvares and Ajmal Mian and Andrew Whitehouse and Murray T. Maybery},
  title = {{An investigation of a novel broad autism phenotype: increased facial masculinity among parents of children on the autism spectrum}},
  booktitle = {Proceedings of the Royal Society B Biological Sciences},
  year = {2022},
  doi = {10.1098/rspb.2022.0143},
}
Figure