Previous neuroimaging studies

Some previous studies support the hypothesis of a neurobiological basis in the origin of sexual orientation, and have indicated that sexual orientation related structural and functional cerebral differences may exist. See [2], [8] for a comprehensive overview on the topic of sexual orientation related differences in the human brain. In brief, using magnetic resonance imaging (MRI) and positron emission tomography (PET) Savic et al. measured volumetric hemispheric asymmetry and functional connectivity of the amygdala, and found that hemispheric volumes of homosexual men and heterosexual women were symmetric, whereas heterosexual men and homosexual women showed a rightward cerebral asymmetry. In addition, the amygdala connections of the in homosexual men and heterosexual women were found to be more widespread from the left amygdala, whereas heterosexual men and homosexual women the functional connections were more widespread from the right amygdala [11]. Moreover, human pheromone-like compounds activated the anterior hypothalamus of heterosexual women and homosexual men in a similar way, but differently from the subjects of the same sex and opposite sexual orientation [12], [13]. Another MRI study revealed a more male-like structural pattern of grey matter in the perirhinal cortex of homosexual female participants [7]. Furthermore, Kranz et al. found that heterosexual females and homosexual males responded more to photographs of male faces, whereas homosexual females and heterosexual males responded more to photographs showing female faces [14]. It was also reported that compared to heM functional connectivity between a seed region in the left inferior occipital gyrus and the right cuneus was reduced in hoM, and the connectivity between this seed region and the right thalamus and cuneus correlated positively with Kinsey scores [15]. Moreover, during visually evoked sexual arousal different neural circuits were active in hoM and heM: heM showed activation in lingual, hippocampus and parahippocampal gyrus, whereas in hoM the angular gyrus, caudate and pallidum were activated. [16]. Although research pointing to the existence of cerebral differences related to sexual orientation is emerging, knowledge about neuroanatomical characteristics is still poor, and numerous questions about its origin and development still remain unanswered.

Cortical thickness (Cth)

A metric increasingly used to study brain-behavior relationships in multiple clinical conditions and healthy controls, which is also associated with neurocognitive functions, is cortical thickness (Cth) [17]–[19]. The measure of Cth is related to the size, number and density of cells and dendrites in a cortical column [20], can serve as a proxy marker of the integrity of the cerebral cortex [21]–[23] and was shown to be related to brain metabolite concentrations, such as N-acetylaspartate [23]. The genetic coding of Cth seems also to be independent from the genetic coding of cortical surface area and volume [24], [25]. Intriguingly, Bramen et al. found that Cth was associated to circulating testosterone levels differently in adolescent boys (positively) and girls (negatively), and predominantly in regions high in androgen receptors, such as the medial occipital cortex [26]. Furthermore, the corresponding correlation seems to become inversed in the adulthood [27], [28] illustrating the complexity of the sex hormone interaction with sex, stage of development and region of the brain. Nevertheless, the reported preliminary observations are of interest for the research of sexual orientation, since testosterone is believed to have impact on sex differences in brain structure and some functions [29]–[31] implicating a possible link also to sexual orientation [2], [9], [32], [33]. In this respect it is also of note that sex differences in Cth have been reported, primarily in frontal, parietal and occipital areas of the brain [28], [34]–[39], albeit with some inconsistencies between the still limited reports in the literature. For example, using brain size normalized data Luders et al. demonstrated higher cortical thickness in females compared to males wide spread over all lobes, whereas when analyzing unscaled data females had thicker cortices in e.g. lateral superior pre- and postcentral regions, the occipital lobe and in left frontal gyrus. In addition, they showed that males had thicker cortices than females in a region of the left lateral temporal lobe, as also found by Lv et al [38]. Furthermore, Sowell et al. found thicker cortices in a region comprising the right inferior parietal and posterior temporal regions of females compared to males, when not correcting for differences in brain size, and males had thicker cortices than females in right anterior temporal and orbitofrontal regions [34]. In another study, Im et al. demonstrated greater cortical thickness in females in superior parietal and postcentral gyrus, but in contrast to other studies locally thicker cortices in men were not observed [37]. Presumably, some of these inconsistencies may stem from population based and/or methodological differences across studies, and taken together, further in vivo study is required to elucidate the relationship between sex and regional Cth.

The present study

Findings in previous studies suggest that certain cerebral characteristics might differ between homosexual and heterosexual individuals, and that homosexual individuals may have some cerebral features, that are more similar to their opposite sex heterosexual counterpart. Moreover, sex differences in cortical thickness were reported, and testosterone effects are assumed to play a role in sexual dimorphism of the brain, as well as being linked to the development of a certain sexual orientation. Thus, based on previous studies we hypothesized that sexual orientation related differences in regional Cth could be found, most likely in regions considered as sexually dimorphic (reviewed above) and/or regions high in androgen receptors, such as the medial occipital lobe.

To our knowledge, no previous studies investigated Cth differences in relation to sexual orientation. In pursuit of a better understanding of the origin and development of sexual orientation and to expand the knowledge about sexual orientation specific differences in the human brain, we investigated differences in regional cortical thickness between heterosexual men (heM) and homosexual men (hoM). To further explore whether potential differences are present in regions, where sex specific differences have been reported previously, we also compared both groups to heterosexual women (heW).

Participants

Sixty one healthy age-matched hoM (n = 19), heM (n = 21) and heW (n = 21) were recruited from the local community. The Ethics Committee of the Karolinska Institutet Stockholm approved the study, and prior to study all participants signed informed consent forms. heM and hoM differed from heW in biological sex (male/female), whereas heM and hoM only differed in sexual orientation. Sexual orientation was assessed through the Kinsey hetero/homosexual self-identification scale (0: exclusively heterosexual, 6: exclusively homosexual) [40]. Individuals were considered as heterosexual when scoring between 0 and 1, and as homosexual when scoring between 5 and 6 on the Kinsey scale. Participants reported no history of psychiatric disorders, head trauma and head injury resulting in loss of consciousness for more than 10 minutes, HIV/AIDS, Hepatitis-C, chronic pain conditions, vision or hearing problems. In addition, all MRI scans were evaluated by a clinical radiologist to assure that the analyzed cohort show no brain pathology. To assure a balanced number of left and right handed participants in each group handedness was determined using the Edinburgh Handedness Inventory [41]. Participants' characteristics are described in more detail in the results section.

MRI acquisition

3D T1-weighted SPGR (spoiled gradient echo; TE  = 3.1 ms, TR  = 7.9 ms, TI  = 450 ms, FOV  = 24 cm, matrix = 240×240, 176 axial slices, flip angle  = 12 degree) were acquired with 1 mm³ isotropic resolution using a 3-Tesla MRI medical scanner (Discovery 3T GE-MR750, General Electric, Milwaukee, Wisconsin) equipped with an 8-channel phased array receiving coil.

Image processing

Cortical thickness (Cth) measures were obtained using the semi-automated subcortical volumetric segmentation and cortical surface reconstruction methods provided by Freesurfer v5.1. Methodological details of the individual processing steps can be found in [42]–[47]. In brief, the procedure includes intensity normalization, removal of non-brain tissue, segmentation of cortical gray, subcortical white, deep gray matter volumetric structures, triangular tessellation of the gray/white matter interface and the pial surface (white matter/cerebrospinal fluid boundary). Cortical thickness is calculated as the closest distance from the gray/white to the pial surface at each vertex on the reconstructed surfaces. All reconstructions were visually inspected and, where necessary, manual corrections were made by trained operators, including corrections of erroneous skull striping, white matter, grey matter and subcortical segmentation using editing tools provided by Freesurfer. Freesurfer allows an automated parcellation of the cortical surfaces into 34 anatomical regions of interest (ROI) for each hemisphere, segmentation of subcortical ROIs (thalamus, caudate, putamen, hippocampus, amygdala) and cerebellum, see [45], [48] for technical details. For each ROI, Cth and subcortical volume measures were obtained from subjects' native space.

Statistical analyses

Overall group differences in regional Cth and subcortical volumes were tested in separate univariate analyses of covariance (ANCOVA) using SPSS v20, with the MRI outcome measure as dependent variable, age as covariate and group (heM, hoM and heW) as fixed factor. Significant ANCOVA findings for group (p<0.05) were followed up by pairwise group comparisons (t-tests) to understand the nature of the main effect for the group factor. Since the tested outcome measures are correlated to each other, we corrected univariate and pairwise group analyses for multiple comparisons with an adjusted (or partial) Bonferroni method, details on the applied Dubey/Armittage-Parmaris procedure can be found in [49]. Alpha levels (0.05) were adjusted to p = 0.0024 (cortical) and p = 0.0133 (subcortical) using the average inter correlation coefficients among all investigated cortical (r = 0.28) and subcortical ROIs (r = 0.46) for the combined cohort. Those adjusted alpha levels represent the upper boundaries of p values, which can be considered as statistically significant after correction for multiple comparisons. Effect sizes for group mean differences were calculated via Cohen's d. We also tested for effects of years of education and total brain volume, respectively, by entering those variables simultaneous with age as covariates. Before final analysis, distributions of all investigated variables were tested for normality using the one-sample Kolmogorov-Smirnov test within each group separately and in the combined group. We also analyzed sexual orientation related differences in Cth on a vertex level. The methods and results of this secondary analysis are presented in File S1 .

Demographic variables

Groups did not differ in age (heM: 31.9±6.0 years, heW: 33.2±6.1 years, hoM: 33.5±6.1 years), years of education (heM: 15.9±2.0, heM: 17.3±3.3, hoM: 16.4±2.9) or handedness scores (heM: 74±26, heW: 77±26, hoM: 85±27) and participants were primarily right handed. Two participants in the female group had slightly negative scores of −6 and −20, respectively (mixed-handedness). Heterosexual individuals scored in average 0.0±0.5, and homosexual individuals 5.5±0.5 on the Kinsey scale.

Group comparisons of Cth and subcortical volumes

Univariate tests were significant for group differences in Cth in the left middle and superior temporal ROI, in the right inferior temporal, lateral orbitofrontal, pars triangularis, lingual, cuneus and pericalcarine ROI (all 0.0001<p<0.041). Group differences were also significant for subcortical volumes in amygdala, putamen, thalamus and cerebellum (all 0.0001<p<0.004). Results of pairwise group statistics of regions in which group differences were found are shown in Table 1 and 2. MR measures (group means of Cth and subcortical volumes) of all investigated ROIs can be found in Table S1 and Table S2. In pairwise group comparisons of Cth in the left hemisphere, heW demonstrated significantly thinner cortices than hoM in the middle and superior temporal ROI. The same was observed comparing heW to heM, where the difference in the middle temporal ROI (p = 0.004) can be considered as trend after Bonferroni adjustment. No differences were found between heM and hoM. In the right hemisphere, cortices of heW and hoM were thinner compared to heM in the lateral orbitofrontal, as well as in the lingual, pericalcarine and cuneus ROI, where the differences in right lateral orbitofrontal (p = 0.003) and cuneus (p = 0.003) are considered as strong trends after Bonferroni adjustments. In those ROIs no differences were found between heW and hoM. In addition, hoM showed thinner cortices than heM (and same trend compared to heW) in the right pars triangularis and inferior temporal ROI. Where significant differences or trends (p<0.1) were found effect sizes were moderate (0.5–0.79) to strong (≥0.80) [50].

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Group differences in Cth.

https://doi.org/10.1371/journal.pone.0114721.t001

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Group differences in subcortical volumes.

https://doi.org/10.1371/journal.pone.0114721.t002

heW had smaller volumes than both, heM and hoM, in left and right amygdale, putamen and cerebellum. However, in both hemispheres thalamus volumes of both hoM and heW were smaller than in heM, whereas thalamus volumes of hoM and heW did not differ. Controlling for years of education and brain volume, respectively, did not change the reported results of group differences in Cth. The same applies for group comparisons of subcortical volumes, with the exemption: when correcting for brain volume, groups were similar in amygdale, thalamus and putamen volumes, but hoM had still significantly smaller thalamus volumes than heM and heW (all p<0.001). Excluding the two ambidextrous female participants from the analysis did not change the results for observed group differences, neither in Cth nor subcortical volumes.

Study limitations and Outlook

The present findings relate only to male homosexuality, and the generated data cannot be extrapolated to female homosexuality, which requires separate studies to be conducted. The data consisted of average thickness values obtained from relatively large cortical regions defined by the Desikan's atlas. Thus evaluations of group comparisons with higher regional resolution (e.g. vertex-wise analysis), as well as investigations of larger populations will most likely give more detailed and/or additional information about sexual orientation related cortical characteristics.

Furthermore, participants did not report any history of psychiatric disorders; however, we did not administer depression, mood or anxiety measures, which may be related to the morphological measures. Furthermore, differences in body mass index, diet and exercise, as well as smoking status may have influenced the findings. Also here, larger study samples are needed to identify the degree to which those factors affect our results. Although the present data do not provide information about the underlying mechanisms, and may not disentangle genetic, hormonal or other environmental processes, they open a direction for future investigations of the biology of sexual orientation.