Our study revealed that Tibetan males had larger brains than Tibetan females in both global GM volume and WM volume, which was consistent with that found in lowlanders such as the Germans, Americans, Koreans, Swiss, and Australians  as well as in our present Chinese Han subjects. Tibetan females had a larger proportion of GM volume than males, which was also found in previous studies on lowlanders [29, 30] but not in our present study on Han subjects. Moreover, Tibetan females had smaller regional cortical GM volume than males, with GM volume in the left pars opercularis in males and females had a significant positive correlation with forward digit span performance. In contrast, Tibetan females had larger regional CT than males, and moreover, in females, CT values in these regions had significant negative correlations with altitude.
In our study, the accurate brain sites showing sex differences of CT were different between Tibetans and Han subjects, but they were all within the regions found in the Germans aged 24.3 ± 4.3 years , Koreans aged 19-36 years , and Americans aged 7-87 years , with significantly thicker cortices in females. In our study, although we got an opposite results in regional GM volume between Tibetans and Han, but the smaller regions in Tibetan females and the larger regions in Han females all overlapped and were consistent with the regions found in the Germans aged 20-30 years . In addition, the regions that showed higher CT in the right superior temporal gyrus and left occipital lobe in our Han subjects also showed larger GM volume in America females aged 25.1 ± 4.5 years . In the present study, we have also analyzed brain structural differences between Han males and Tibetan males and brain structural differences between Han females and Tibetan females, and the results showed that the sex differences of brains were consistent with that found in our previous study in Tibetans and Han subjects .
Our study found that Tibetan females had significantly thicker regional CT than males and the correlation of regional CT with altitude existed in female but not male Tibetans by both regional and global analyses. (1) Firstly, these results suggest that HA environmental factors may affect easily on brain developments in female residents. In agreement with our results, depression and anxiety behaviors have been observed to increase with altitude in female but not male rats . In our previous study on sea-level residents after 4-week exposure on the Qinghai-Tibet Plateau, both males and females showed significant increases in cerebral iron deposition in the deep nuclei of brains, while the increased proportion of females (4%) was greater than males (2%) . Baum et al.  found that chronic intermittent hypoxia induced a higher Fosb gene expression in females than in males, reflecting stronger neuroplastic dynamics. (2) Secondly, these results suggest that females may have a better capacity to adapt to hypoxia. Some clinic and experimental data support this suggestion. When both male and female rats were reared at an altitude, red blood cell count, haematocrit, and plasma erythropoietin levels were lower in females than in males . A lot of laboratory studies showed that female animals with cerebral hypoxia-ischemia were less adversely affected relative to comparably injured males [14, 22, 44]. Clinical data also suggests females with cerebral hypoxia-ischemia exhibited less severe behavioral deficits compared to males . Cohort studies have demonstrated a higher vulnerability in males towards neonatal ischemic and/or hypoxic-ischemic injury. Male brains are poorly repaired after neonatal hypoxia-ischemia, and males have an increased incidence of long-term cognitive deficits . Female resistance to hypoxia can explain the lower female total mortality rate in infancy, childhood and adulthood .
Several studies have reported sexual dimorphism of the proportion of major cranial tissue compartments in the brain. In our study, Tibetan females showed a larger proportion of GM volume than males, which was consistent with the findings in the Americans  and Germans . However, consistent with our findings in Han subjects, two studies failed to detect any sex differences in the Americans [47, 48]. Others observed both higher GM and WM proportions in males (reviewed by Luders Toga ). Moreover, in our study, both Tibetan males and females had a larger proportion of GM volume than Han males and females. Taken together, these differences may be due to racial factor.
Allelic variation of natural selective genes for HA adaptations may account for discrepant pattern of sex differences exists between Tibetan and Han brains. Genome-wide scans has reported evidence for positive natural selection at the Egl nine homolog 1 (EGLN1), endothelial PAS domain-containing protein 1 (EPAS1), and peroxisome proliferator-activated receptor alpha (PPARa) loci in the Tibetan on Qinghai-Tibet Plateau [36, 37]. All of these genes are associated with the hypoxia-inducible transcription factor (HIF) pathway. EPAS1 gene encodes the HIF-2α subunit of HIF complex. Variation at the EGLN1 locus is associated with protection against polycythemia in Tibetans at HA , and single nucleotide polymorphism at the EPAS1 locus is associated with hemoglobin levels in Tibetans [36, 50]. EPAS1 plays an important role in vascular remodeling . HIF-2α also mediates the transcriptional activation of EPO expression in astrocytes, and thus promotes astrocytic paracrine-dependent neuronal survival during ischemia . EGLN1 through encoding HIF prolyl 4-hydroxylase 2 (PHD2) plays a critical role in glucose metabolism . PHD2 deficiency induces vascular remodelin . PHD is highly expressed in the cortex . It is shown to regulate synaptic density and alter cell migration  and is involved in axon rewiring following a brain injury by regulating neurite elongation of cortical neurons . PPARa is involved in neuronal proliferation, differentiation, and apoptosis .
The greater capacity for females to adapt to hypoxia may be related to the effects of circulating estrogen and progesterone. These two hormones have been shown greater in the females living at HA than the females who resident at lowlands . The resistance of females to ischemia is acquired after puberty  and is lost after menopause, which is in accordance with the protective effects of estrogen . Exogenous administration of estrogen has been shown to reduce ischemia-induced cerebral injury , and the protective effect may be through preventing neuron death  and related to its antioxidant properties . Estrogen can also increase regional cerebral blood flow (CBF) [64-66] and correlates directly with CBF velocity . Females have higher CBF compared with males in the left inferior frontal gyrus, bilateral middle temporal gyri, and left superior temporal gyrus [68, 69]. In addition, in the animal models of neonatal hypoxia-ischemia, males were more sensitive to mitochondrial dysfunction, with the increased mitochondrial permeability on the inner and outer membranes leading to a high amount of released proteins as compared to females . Taken together, the increased blood sex hormones, increased CBF, and relatively little mitochondrial permeability may contribute to female resistance to hypoxia. Females also have a better capacity to adapt to cold. For example, the vascular response to coldness at HA was smaller in females compared with males ; cold decreased the fatigue index of a sustained 2-min maximal voluntary contraction in males but not in females ; a significant benefit of temperature reduction in hypoxia ischemia was found in females but not in males .
In our study, sex differences of brains were found in the young adult residents at HA, which could be different from that in children or older peoples, as age-associated changes are sex-specific. A study has shown that men experienced greater volume decrement across age-groups than women, particularly in the dorsolateral prefrontal regions . Another MRI studied on healthy adults aged range 18-80 years showed that the greatest amount of atrophy in elderly men was in the left hemisphere, whereas in women age effect was symmetric .
In our study, females showed significant decreases of GM volume in the left pars opercularis and pars triangularis of Broca’s area, and GM volume in the left pars opercularis in Tibetans had a significant positive correlation with forward digit span performance. Previous study has found an association between the impaired forward digit span performance and the ischemia in pars opercularis . Moreover, stimulation of left Broca's area interfered with digit span, producing significantly more item than order errors . Forward digit span is a kind of verbal phonological short-term memory. MRI studies on articulatory suppression indicated that pars opercularis was involved in phonological short term-memory [77, 78]. Patients, who had a disability to make phonological judgements, showed lesions in the left Broca’s area . Therefore, the decreased GM volume of left pars opercularis of Broca area may be associated with poor phonological short term-memory in Tibetan females.
In comparison with Han population, both Tibetan males and Tibetan females had lower scores in almost all items of behavioral tests. Considering all the behaviors were originally designed for the test of Han population and cultural difference between Tibetans and Han population is real, we cannot reach the conclusion that the Tibetans performed worse than Han population.
The limitation in our study is that the inter-ethnic differences in brain structures can be large. We cannot draw a conclusion that gene and developmental environment which one dominantly determined the different pattern of sex differences between Tibetan and Han brains. Brain morphological differences between populations of different origins have been found in the early neonate life  and in adults in terms of whole brain and region-specific volume [81-83]. Our previous study has revealed that, compared with Han subjects living at lowlands, Tibetans living on the Qinghai-Tibetan Plateau were associated with structural modifications in cortical thicknesses, curvature, and sulcus . Therefore, the global brain differences between these two populations may underlie the different pattern of sex differences between Tibetan and Han brains.