Brain charts for neuroanatomical sex differences across the human lifespan

Abstract Population-level sex biases in human neuroanatomy have been highly debated, with prior literature often limited by narrow age windows and simplistic assumptions about how brain regions scale with sex differences in average total brain size. Recent work leveraging massive, global datasets to flexibly chart normative brain structures across the lifespan provides a framework to overcome these hurdles. Here, using magnetic resonance imaging (MRI) data from over 100,000 individuals (51.7% F) from mid-gestation to 99 years, we chart normative population trajectories of 241 structural features capturing age-varying sex effects. We use split-half cross-validation, fitting generalized additive models for location, shape, and scale with nonlinear scaling terms to test sex’s impact on each feature’s distribution and map changes in sex biases over time. Our results reveal replicably significant age-varying sex biases in nearly all brain structures, and show that sex’s effects survive nonlinear correction for total brain size. We find that after correcting for total brain size, regions are equally likely to be larger in males versus females, while males tend to show higher interindividual variability in a majority of regions. Probing temporal dynamics reveals that sex biases tend to increase with age, with male and female trajectories diverging across the lifespan. Finally, we demonstrate that normative scores from these models of age-varying sex effects are sensitive to case-control differences in six neuropsychiatric disorders. This work resolves several existing methodological issues to establish and quantify population-level sex biases in the largest-to-date study of lifespan sex-biases in the human brain. Significance Statement While males’ and females’ brains are more alike than different, understanding population-level sex biases is important for ensuring that research benefits everyone. Here, we build on recent advances in modeling growth trajectories, using data from over 100,000 individuals to overcome limitations in prior studies of sex differences in brain structure. Lifespan statistical models of each region demonstrate: 1) differences in male and female normative trajectories vary with age; and 2) sex biases in brain regions are not just caused by differences in overall brain size. We also show that recognizing sex biases in the population is important for accurately mapping brain correlates of psychiatric diseases, and provide a resource for future researchers to benchmark their own neuroimaging data. Competing Interest Statement M.G., J.S., and A.A.B have an inventorship interest in intellectual property licensed to Centile Bioscience by the Children's Hospital of Philadelphia. J.S., R.A.I.B., and A.A.B. hold shares in and J.S. is director of Centile Bioscience. R.T.S. has received consulting income from Octave Bioscience and compensation for scientific reviewing from the American Medical Association. Footnotes Competing Interest Statement M.G., J.S., and A.A.B have an inventorship interest in intellectual property licensed to Centile Bioscience by the Children’s Hospital of Philadelphia. J.S., R.A.I.B., and A.A.B. hold shares in and J.S. is director of Centile Bioscience. R.T.S. has received consulting income from Octave Bioscience and compensation for scientific reviewing from the American Medical Association.