Understanding what happens in the brain during PMDD symptoms is a key step toward better treatments. This study used brain imaging (fMRI) to compare how the brains of 29 people with PMDD and 27 people without it responded to emotional faces at two points in their menstrual cycle — once during the symptom-free follicular phase and once during the late-luteal phase when PMDD symptoms peak.
During the luteal phase, people with PMDD showed greater activity in many brain areas involved in processing emotions, including the insula, cingulate cortex, and cerebellum. Importantly, these differences were only present during the luteal phase — during the follicular phase, the two groups looked the same. This fits with the lived experience of PMDD, where symptoms are tied to a specific part of the cycle.
One especially interesting finding involved two brain chemicals called allopregnanolone (ALLO) and isoallopregnanolone (ISO). Both are natural byproducts of progesterone that affect a calming brain system called GABA. The researchers found that the balance between these two chemicals related to brain activity in the amygdala — a key emotion center — in opposite ways for people with and without PMDD. In the PMDD group, a higher ISO-to-ALLO ratio was linked to more brain activation, while in controls it was linked to less. This supports the idea that PMDD isn't about having abnormal hormone levels (both groups had similar levels) but rather about the brain responding differently to normal hormonal changes.
The study also found that greater cerebellar brain activity during the luteal phase was linked to more severe anxiety symptoms in the PMDD group. While this research doesn't lead directly to new treatments yet, it deepens our understanding of why the luteal phase feels so different for people with PMDD and highlights specific brain pathways that future therapies might target.
Key findings
- Participants with PMDD (N=29) exhibited greater brain activity in emotion-processing regions (bilateral insula, ACC, PCC, precuneus, MFG, SFG, SMA, dorsal striatum, thalamus, cerebellum) compared to controls (N=27) during the late-luteal phase (pFWE < 0.05), with no group differences in the mid-follicular phase
- The ratio of ISO/ALLO serum levels was differentially associated with brain activity in the right parahippocampal gyrus/amygdala between groups: a positive relationship was found in PMDD subjects while a negative relationship was observed in controls (pFWE < 0.05)
- Emotion-induced cerebellar activity (lobules V-VI and vermis I-IV) was positively associated with anxiety symptom severity in PMDD subjects during the late-luteal phase (R² = 0.486 and R² = 0.395 respectively, pFWE < 0.05)
- In the late-luteal phase, PMDD subjects were less accurate (p = 0.006) but faster (p = 0.046) than controls on the face-matching emotion task
- No group differences in serum levels of progesterone, estradiol, ALLO, or ISO were found across the menstrual cycle, consistent with the hypothesis that PMDD involves altered sensitivity to normal neurosteroid levels rather than abnormal hormone concentrations
- Right amygdala showed increased activity at trend-level (pFWE < 0.10) in PMDD compared to controls during the late-luteal phase in small volume corrected analyses
Methods, briefly
Cross-sectional fMRI study with N=56 participants (29 PMDD, 27 controls) scanned during both mid-follicular and late-luteal menstrual cycle phases. PMDD diagnosed prospectively over minimum two cycles using DRSP ratings. Participants completed an emotional face discrimination task during fMRI (3T scanner). Serum levels of progesterone, estradiol, allopregnanolone (ALLO), and isoallopregnanolone (ISO) were measured at each session. Non-parametric permutation testing (FSL Randomise, 5000 permutations) was used with TFCE correction. Both ROI (bilateral amygdala) and whole-brain analyses were conducted. Session order was counterbalanced across groups.
Limitations to keep in mind
- DRSP symptom ratings used for brain-symptom correlation analyses were collected during screening cycles, not during the actual scanning cycles
- Serum neurosteroid levels may not reflect local brain concentrations, as steroids accumulate non-uniformly across brain regions and can be synthesized de novo in neural tissue
- High proportion of ISO values below the limit of quantification in the mid-follicular phase (72.4% PMDD, 59.3% controls) limited ISO analyses to the late-luteal phase only
- Some inconsistencies with prior PMDD neuroimaging literature in direction of effects for certain brain regions (e.g., increased rather than decreased ACC activity)
- Potential confound of psychiatric history, which tended to differ between groups (31% PMDD vs 11.1% controls), though additional analyses showed results were not significantly influenced by this variable
- Interpretational difficulties inherent in using hormone ratio variables
The latest PMDD research, in your pocket
The app’s research feed is updated daily with newly published, peer-reviewed PMDD work - summarized in plain language.