Sociodemographic variables
No significant differences between groups emerged in terms of age F(2, 38) = .613, p = .547; sexual orientation 𝛘² (4) = 7.129, p = .129; marital status 𝛘² (6) = 7.199, p = .333; level of education 𝛘² (6) = 11.700, p = .069; employment status 𝛘² (2) = 3.632, p = .163; handness 𝛘² (2) = .084, p = .959; and IQ F(2, 35) = 1.615, p = .213.
As expected, significant differences emerged in testosterone F(2, 37) = 10.325, p = <.00001, and in estradiol levels F(2, 37) = 3.448, p = .042. Moreover, a trend was found in progesterone levels F(2, 37) = 3.028, p = .061. Post hoc Bonferroni correction tests revealed a significant difference in testosterone between cis women (M = .36, SD = .09) and trans men (M = 5.95, SD = 5.18) p = .001, and between cis women and cis men (M = 6.27, SD = 2.48) p = .001.
IC analysis and dual regression
Dual regression analysis showed statistically different FC in IC7, IC11, and IC20.
IC7 included a cluster in precentral gyrus (peak at x, y, z: 75, 66, 46, cluster size = 169 voxels) (Figure 1) and showed in trans men a weaker connectivity than cisgender men (p <.01).
IC11 included a cluster in the subcallosal cortex (peak at x, y, z: 43, 78, 35, cluster size = 664 voxels) (Figure 2), a cluster in paracingulate gyrus (peak at x, y, z: 44, 89, 43, cluster size = 170 voxels) (Figure 3), a cluster in temporal pole (peak at x, y, z: 66, 71, 27, cluster size = 142 voxels) (Figure 4), and showed in trans men a weaker FC than cisgender males (p <.01).
IC20 included a cluster in the cingulate gyrus (peak at x, y, z: 44, 60, 53, cluster size: 157 voxels) (Figure 5) and showed in trans men a weaker FC than cis men (p <.01).
Behavioral data
The ANOVA analyses showed significant differences in Digit Span Backward F(2, 30) = 6.856, p = .004, Corsi Span Forward F(2, 30) = 5.693, p = .008, and Corsi Span Backward F(2, 30) = 5.028, p = .013, all higher in cis men than cis women and trans men. No other statistically significant differences were found among the three groups in visuo-spatial, attentions, executive, and language functions.
Post-hoc analysis of Digit Span BW data showed that trans men (p = .022) and cis women (p = .005) performed worse than cis men. We found similar results with Digit Span FW, even if not significant (p = 0.055). Equally, in Corsi Span FW trans men (p = .029) and cis women (p = .014) performed worse than cis men. However, in Corsi Span BW only trans men performed worse than cis men (p = .015). No statistically differences were found in Corsi Span BW between cis men and cis women.
ICA, behavioral data, and hormones
Pearson correlation analysis revealed in ICA 7 a negative correlation between precentral gyrus and TMT in trans men r = -.597, p = .04.
In trans men a cluster involving the subcallosal cortex (ICA 11) was negatively correlated with Stroop IT r = -.626, p = .03; a cluster involving the paracingulate gyrus (ICA 11) was negatively correlated with Stroop IT r = -.591, p = .043; in ICA 20 a cluster involving the cingulate gyrus was positively correlated with FAS r = .713, p = .009. In cisgender women, the precentral gyrus (ICA 7) was negatively correlated with Corsi BW r = -.697, p = .025; furthermore, a cluster involving the cingulate gyrus (ICA 20) was positively correlated with Stroop IE r = .805, p = .005.
In trans men a negative correlation between the precentral gyrus (ICA 7) and estradiol was highlighted r = -.654, p = .004 together with a trend with testosterone r = .459, p = .064, and progesterone r = -.475, p =.054. Moreover, a cluster involving the subcallosal cortex (ICA 11) was positively correlated with testosterone r = .718, p = .001, and negatively with estradiol r = -.602, p = .011; finally, a positive correlation between the cluster involving the paracingulate gyrus (IC 11) and testosterone was found r = .592, p = .012. In cisgender men participants, no significant correlations were found.