Numerous studies were able to show gender differences in FP-CIT uptake (Kock et al. 2014; Nobili et al 2013; Eusebio et al.2012; Varrone et al. 2012). The aim of this study is to understand how the choice of the reference VOI affects the final result of this analysis.
In our Institution, FP-CIT SPECT images by 73 patients with essential tremor (39 women and 34 men) were obtained. Before any statistical analysis with SPM8 software, all images were firstly spatially normalized with a template (Kas et al. 2007) and smoothed.
Using Marsbar software’s VOIs (see below) as standard reference, activity normalized images were obtained by calculating the ratio between voxels and VOIs activity.
-”completed” cerebellar (CC) (”cerebellum4-5” + “cerebellum 6” + “cerebellum crus1”)
-”completed” occipital (CO) (all software’s occipital VOI added together)
-superior occipital (SO), medium occipital (MO), inferior occipital (IO)
-medial cerebellar (MC) (”cerebellum 6” + “cerebellum 4-5”)
With the aim to study how different VOIs affect the SPM analysis, we carried out a gender comparison by using the two sample t test. All analyses were age-adjusted.
The comparison done by means of the CC VOI showed 5 clusters of voxels with a higher uptake in women (uncorrected p <0,005 was considered as cut-off in all analyses of this study) which include basal ganglia and various extra-nuclear regions (including motor areas and right insula). Other clusters with higher uptake were also found in men in the same analysis (including right hippocampus, limbic cortex, medial cerebellum and Locus coeruleus).
By using the CO, SO and MO VOI, all the preceding clusters with higher uptake in men were found to be enlarged. No cluster with significant higher uptake in women was found.
By using the MC VOI, clusters that in previous analysis had shown a higher uptake in women now have widened (including the insulae, the occipital cortex bilaterally and the lateral portions of the cerebellum).
Finally, by using IO VOI, dimorphic clusters were found with a higher uptake and a lesser extension in both gender, localised within the same areas as the clusters previously found.
Very different results were obtained in gender comparison by using different VOIs in activity normalization of FP-CIT images, but never “conflicting” one another. This is probably due to gender dimorphism of reference regions (by using a reference area with a higher uptake in a gender, areas with higher uptake in the other gender are easier to find). Literature also partially confirms these results (2012 Eusebio et al.; 2007 Jovanovic et al. with [(11)C] MADAM, a 5-HTT radioligand).
We can presume that in gender comparison the influence of the “gender” variable on reference regions might allow to detect with more statistical power differences between FP-CIT images. We can speculate that this can be true also for other variables.