Effects of an Oral Contraceptive on Dynamic Brain States and Network Modularity in a Serial Single-Subject Study

Abstract

Hormonal contraceptive drugs are used by adolescent and adult women worldwide. Increasing evidence from human neuroimaging research indicates that oral contraceptives can alter regional functional brain connectivity and brain chemistry. However, questions remain regarding static whole-brain and dynamic network-wise functional connectivity changes. A healthy woman (23 years old) was scanned every day over 30 consecutive days during a naturally occurring menstrual cycle and again a year later while using a combined hormonal contraceptive. Here we calculated graph theory-derived, whole-brain, network-level measures (modularity and system segregation) and global brain connectivity (characteristic path length) as well as dynamic functional brain connectivity using Leading Eigenvector Dynamic Analysis and diametrical clustering. These metrics were calculated for each scan session during the serial sampling periods to compare metrics between the subject's natural and contraceptive cycles. Modularity, system segregation, and characteristic path length were statistically significantly higher across the natural compared to contraceptive cycle scans. We also observed a shift in the prevalence of two discrete brain states when using the contraceptive. Our results suggest a more network-structured brain connectivity architecture during the natural cycle, whereas oral contraceptive use is associated with a generally increased connectivity structure evidenced by lower characteristic path length. The results of this repeated, single-subject analysis allude to the possible effects of oral contraceptives on brain-wide connectivity, which should be evaluated in a cohort to resolve the extent to which these effects generalize across the population and the possible impact of a year-long period between conditions.

Keywords: brain modularity; dynamic functional connectivity (dFC); functional connectivity (FC); functional magnetic resonance imaging (fMRI); hormonal contraceptive; menstrual cycle; oral contraceptive (OC); steroid hormones.

FIGURE 1

Hormonal timelines in each of the two conditions. The subject’s hormone levels during the natural cycle and while taking the oral contraceptive in the order of experiment days. Modified from Taylor et al. (2020).

FIGURE 2

Static functional connectivity comparisons between naturally cycling and oral contraceptive (OC) condition scans. Panels (A,B) show Tukey’s boxplots representing system segregation and weighted characteristic path length values (CPL), respectively. The horizontal lines represent medians, and the vertical limits of the colored area represent the first and third quartiles. Notches represent 95% confidence intervals for comparing medians, calculated as $\frac{\mathbf{1.58}\times \mathrm{IQR}}{\sqrt{\mathbf{n}}}$. Panels (C,D) show modularity and binarized CPL values reported for each condition at a range of sparsity threshold values between 0.1 and 0.25 and are plotted as mean ± SD.

FIGURE 3

Two brain states for k = 7 identified as having significantly elevated fractional occurrence when oral contraceptive medication is used compared to a natural menstrual cycle (A), and opposite (B), and summary p-values across the range of k (C). In the connectivity visualizations, nodes are only shown if the corresponding centroid element has a strength of at least 50% of the maximum absolute loading. In panel (C), green triangles correspond to state 1, purple triangles to state 2, and the dotted black line corresponds to the within-k Bonferroni-corrected threshold for statistical significance.