A cross-species analysis of neuroanatomical covariance sex differences in humans and mice.

Background: Structural covariance within the brain is thought to reflect inter-regional sharing of developmental influences. This hypothesis has proved difficult to test but can be informatively probed by the study of sex differences. Here, we use neuroimaging in humans and mice to study sex-differences in anatomical covariance- asking (1) are there sex differences in structural covariance and (2) do regions that share the same developmental influences, as exhibited by shared sex differences in volume, also show shared sex differences in volume covariance. This study design illuminates both the biology of sex-differences and theoretical models for anatomical covariance- benefitting from tests of inter-species convergence.

Methods: Brain volume correlations for males and females across 255 regions in mice (n = 423) and 378 regions in humans (n = 436) were calculated using volumetric measures obtained from structural MRI. Mean correlations for each sex were compared within species to determine whether covariance sex differences exist. Specific covariances with strong sex differences in each species were identified via permutation tests for statistical significance. Brain maps of regional average structural covariance sex-bias were generated for mice and humans. Regional average structural covariance sex-bias and volumetric sex-bias were correlated to identify whether these features align in their direction of sex-bias.

Results: We find that volumetric structural covariance is stronger in adult females than males for both wild-type mice and healthy human subjects: 98% of comparisons with statistically significant covariance sex differences in mice are female-biased, while 76% of such comparisons are female-biased in humans (q < 0.05). Regional covariance and volumetric sex-biases have weak inverse relationships to each other in both species: volumetrically male-biased regions contain more female-biased covariations, while volumetrically female-biased regions have more male-biased covariations (mice: r = -0.185, p = 0.002; humans: r = -0.189, p = 0.001).

Conclusions: Our results identify a tendency for females to show stronger neuroanatomical covariance across species. These structural covariance sex differences are also partially related to regional sex differences in volume for both species, suggesting that stronger structural covariance in females could be an evolutionarily conserved feature - partially shaped by the same developmental influences that mediate volumetric sex-biases.