How does circadian rhythm disruption change amygdala gray matter and connectivity?

A 15-day shift-work model in 11 men

The study asked whether acute circadian rhythm disruption changes the structure and function of the amygdala, a subcortical structure the paper describes as central to processing fear, anxiety, depression, and social cognition. To test this, the researchers built a controlled laboratory model of shift work rather than observing people in their normal lives.

Eleven healthy adult male volunteers (mean age 23.2 ± 1.9 years) lived in a controlled environment with no natural light. After a 7-day baseline on a fixed 24-hour schedule, they followed a rotating shift schedule for 15 days, with work-rest cycles compressed to under 10 hours. They then returned to a fixed 24-hour schedule for a 30-day recovery period. Medication, smoking, alcohol, and caffeinated drinks were prohibited throughout.

Sleep quality and mood were measured with the Pittsburgh Sleep Quality Index (PSQI), Zung's Self-Rating Anxiety Scale (SAS), and Zung's Self-Rating Depression Scale (SDS) before the experiment, after the 15 days, and at the 30-day follow-up. Brain structure and function were measured with 7 T MRI, an ultra-high-field scanner chosen for its spatial resolution, two days before the experiment and one day after it ended. The design is longitudinal and within-participant: each person is compared with himself, not with a separate control group.

Sleep and mood worsened, then returned to baseline

Most participants moved in the expected direction: PSQI scores rose in 9 of 11, SAS scores in 8 of 11, and SDS scores in 9 of 11 after the 15-day experiment. Two participants developed mild anxiety and two developed mild depression, according to the paper.

A repeated-measures one-way ANOVA across the three time points found significant main effects for PSQI (F = 10.10, P = 0.009), SAS (F = 6.31, P = 0.019), and SDS (F = 7.31, P = 0.019). Post hoc comparisons showed that PSQI (Pfdr = 0.017), SAS (Pfdr = 0.041), and SDS (Pfdr = 0.032) scores were significantly higher after the experiment than at baseline. By the 30-day follow-up, none of these differences remained significant (Pfdr > 0.05).

This matters for reading the brain results: the questionnaire changes establish that the intervention produced measurable sleep and emotional disturbance, and that these symptoms were reversible at the group level within the follow-up window. They do not establish that the amygdala changes caused the mood changes.

Gray matter fell only in the left amygdala

The central structural result is lateralized. Compared with baseline, gray matter volume (GMV) decreased significantly in the left amygdala (Wilcoxon's r = -0.804, P = 0.005) and in three of its subregions: the centromedial (CM) nucleus (r = -0.831, Pfdr = 0.009), the laterobasal (LB) nucleus (r = -0.751, Pfdr = 0.009), and the superficial (SF) nucleus (r = -0.804, Pfdr = 0.007).

The right side behaved differently. No significant GMV change was detected in the right amygdala or in its CM, LB, and SF subregions (all Pfdr > 0.05). Because the left side shrank while the right did not, the right-to-left GMV ratio increased significantly for the whole amygdala (r = 0.858, P = 0.002) and for CM (Pfdr = 0.004), LB (Pfdr = 0.003), and SF (Pfdr = 0.003). The paper stresses that this ratio shift was driven by left-sided loss, not by right-sided growth.

The subregion breakdown is not decorative. The paper treats CM as a key output region involved in attention and cortical vigilance, LB as the main recipient of sensory input, and SF as involved in olfactory-related processing. A volume change confined to the left side across all three nuclei therefore points to a broad left-amygdala effect rather than a single-nucleus signal.

Worse sleep quality tracked greater left amygdala loss

The study then asked whether the amount of structural change moved with the amount of symptom change. The difference in PSQI scores before and after the experiment correlated negatively with the difference in left amygdala GMV (Spearman's ρ = -0.746, P = 0.008): participants whose sleep quality deteriorated more showed larger left amygdala volume reductions.

At the subregion level, the same direction appeared but did not reach the corrected threshold: CM (ρ = -0.622, Pfdr = 0.061), LB (ρ = -0.696, Pfdr = 0.052), and SF (ρ = -0.536, Pfdr = 0.090). The paper also reports no significant correlations between right amygdala GMV differences and changes in SAS or SDS scores (Pfdr > 0.05).

Two cautions belong here rather than at the end. First, this is a correlation across 11 people, so it does not show that volume loss caused poorer sleep or the reverse. Second, the leave-one-out check found that the left amygdala and left LB associations with PSQI stayed significant, the left CM association approached significance (maximum P = 0.078), and the left SF association did not (maximum P = 0.260). The whole-amygdala correlation is the most stable of these relationships.

Prefrontal connectivity dropped just short of significance

The functional side of the study examined connectivity between the amygdala and the medial prefrontal cortex (mPFC), a region the paper describes as exerting top-down regulatory control over amygdala activity. Connectivity between the whole left amygdala and left mPFC did not change significantly, and neither did the corresponding right-hemisphere measure.

The subregion analysis produced the study's most tentative functional result: a nearly significant decrease in connectivity between the left mPFC and the left CM amygdala (Wilcoxon's r = -0.697, Pfdr = 0.055). Connectivity between the left mPFC and left LB (Pfdr = 0.365) and left SF (Pfdr = 0.152) showed no significant change, and no right-side subregion showed a significant change (all Pfdr > 0.05).

Because Pfdr = 0.055 crosses the conventional 0.05 threshold in the wrong direction, this finding should be read as a signal worth following up, not as an established disconnect. The authors connect it to prior work reporting disrupted amygdala-mPFC connectivity after sleep deprivation, but the present data alone do not confirm that mechanism.

Why the left side may be more vulnerable

The lateralized result raises a question the study cannot answer directly: why the left amygdala and not the right? The authors offer a hypothesis rather than a demonstrated mechanism. They propose that under rhythm disruption the left amygdala is more prone to sustained arousal and higher metabolic activity than the right, citing prior work on asymmetric brain responses to sleep loss.

A second hypothesis concerns the volume measurement itself. The authors note that sleep deprivation can produce detectable, reversible GMV changes and that differences in brain tissue hydration between sleep and wakefulness have been proposed as one explanation. In this study, salivary histidine was significantly elevated on days 10, 13, and 16 compared with baseline. The authors suggest upregulated histidine metabolism could contribute to an abnormal arousal state that reduces tissue hydration and produces measurable GMV change, but they explicitly state this hypothesis needs testing.

These are author interpretations, not results of the experiment. They matter because they change what a GMV reduction means: it may reflect a rapid, partly reversible tissue or fluid shift rather than loss of neurons. The 30-day questionnaire recovery does not resolve this, because no follow-up MRI was reported in the evidence available.

What these findings do and do not cover

The sample is small and deliberately narrow: 11 healthy young men, all right-handed, screened to exclude psychiatric, neurological, and sleep disorders and recent circadian-disrupting work. The authors state that men were chosen to remove hormonal-cycle variability and sex-related confounding, and they call for future work in women. Any generalization to women, older adults, or people with existing mood disorders is outside what this study tested.

The intervention is also specific. It is an experimentally induced, laboratory-based model of acute disruption with compressed work-rest cycles, not chronic insomnia or years of night-shift work. The reported effects therefore describe a 15-day exposure in a highly controlled setting, with diet, light, temperature, noise, and activity standardized.

The statistical defense is worth noting alongside the small n. The authors report power above 0.9 given the large observed effect sizes, and they describe bootstrap, permutation, and leave-one-out checks as consistent with the main results. Those checks support robustness to individual participants, but they do not turn a sample of 11 into a population estimate.

The left amygdala is the result to watch

If one finding from this study deserves priority, it is the left-lateralized structural change. The left amygdala and all three of its subregions lost gray matter volume after 15 days of circadian disruption, the right side did not change, and the magnitude of left amygdala loss tracked the magnitude of sleep-quality worsening. The connectivity result sits below that level of confidence at Pfdr = 0.055, and the mechanistic explanations for laterality and for what a GMV reduction physically represents remain hypotheses.

The practical reading is a comparison, not a recommendation: in this model, structural change was measurable on the left after roughly two weeks of disrupted schedule, while the right amygdala and the whole-amygdala connectivity measures were not. That asymmetry, and its coupling to PSQI change, is the specific claim a follow-up study would need to test in women, in chronic shift workers, and with imaging after recovery to see whether the volume loss reverses.

placeholder

placeholder

Sources used in this answer