We leveraged the Cleveland Clinic STARLIT Registry to examine 1474 vPSGs that underwent manual RSWA scoring because of suspected parasomnia. RSWA was scored using chin and bilateral FDS EMG according to AASM criteria [17, 18] to study the association of AD use with RSWA% and RBD diagnosis. We found (1) SSRI and SNRI monotherapy use was associated with a higher percentage of RSWA epoch, and PSGs meeting RBD diagnosis versus non-users; (2) the highest RSWA% estimates among combination groups were observed with SSRI + SNRI, although combination subgroup findings should be interpreted cautiously given small sample sizes.; (3) no association was found between TCA use alone and RSWA%; and (4) the proportion of patients meeting criteria for RBD was higher among those taking ADs compared with non-users.

Prior studies quantifying RSWA in people taking antidepressants were small and did not record upper extremity EMG

The largest study investigating AD effects on RSWA% used qualitative visual inspection of EMG in REM sleep (identifying only the presence/absence of RSWA) in 10,746 PSGs, and excluded TCAs [9]. They found any RSWA in at least one epoch of REM sleep in 12.2% of PSGs of patients using ADs compared with 2.1% in the entire cohort of AD users and non-users. As expected given the younger cohort, the prevalence of RBD, defined by the presence of RSWA with a history DEB (by chart review, vPSG, or pre-PSG questionnaire), was lower among AD users (0.48%) than in the overall cohort (1.0%; p= 0.005), suggesting that while SSRIs and SNRIs are strongly associated with RSWA, but their link to clinical RBD may be less pronounced [9].

Only three prior investigations have quantified RSWA by AD use, all with smaller sample sizes and different recording and scoring criteria. All used the Lapierre-Montplaisir method defining (1) tonic RSWA when chin EMG activity amplitude ≥ 2 times the background or > 10 µV is present for ≥ 50% of the epoch and atonia when it is present for < 50 and (2) phasic chin EMG as the percentage of 2- or 3-s mini epoch according to which adaptation used [12, 21] containing EMG bursts lasting 0.1–10 s with amplitude > 4 times the background; leg movements during REM are also included from bilateral AT EMG as events lasting 0.1–10 s with amplitude > 4 times the background [22]. In addition to the different adaptations of the Montplaisir method, two studies only included serotonergic AD such as sertraline [10, 21], whereas the third included ADs without specifying the class. These methodological differences likely contribute to the variability in reported RSWA percentages [12].

In the first, elevated tonic submental EMG was observed in 9.54% ± 9.06 of REM epochs in 15 AD users versus 2.36% ± 3.88 in 15 matched controls (p = 0.02) [10]. In the second, combined submental and/or AT RSWA activity averaged 36.9% ± 21.3 in 30 AD psychiatric patients compared to 15.9% ± 8.4 in 15 non-user psychiatric patients (p < 0.01) [12]. The third investigated the effect of adding sertraline in 31 patients with depression and found RSWA increased from day 1 to day 56 for both submental (3.4 ± 1.9% vs. 11.4 ± 4.2%; p < 0.001) and AT (6.2 ± 2.1% vs 15.1 ± 6.6%; p< 0.001) EMG [21]. No clinical RBD behaviors accompanied RSWA on video-PSG [21].

Recent studies confirming submentalis and flexor digitorum superficialis best diagnostic marker for isolated RBD

We recorded and scored RSWA using the FDS EMG rather than AT because the AT leads are prone to motion and electrode artifact which can lead to false-positive findings [23]. Recent studies evaluating diagnostic accuracy confirm that manual scoring of phasic/tonic submental + phasic FDS has the best performance for the diagnosis of RBD [23]. Frauscher et al. systematically recorded EMG in 11muscles of 30 patients with RBD, 15 with PD, and 30 matched controls, and reported the discriminative power for identifying RSWA was higher in the upper limb (100% specificity, area under the curve (AUC) 0.987–9.997) than in lower limb muscles (100% specificity, AUC 0.813–0.852) [11]. Furthermore, a 2023 meta-analysis (14 studies, 434 subjects) found that AASM RSWA scoring methods demonstrated the highest diagnostic accuracy for RBD at 92.2%, with a sensitivity of 92% and specificity of 99% while the Lapierre-Montplaisir RSWA scoring system showed a lower accuracy of 84.8% [24].

Possible mechanisms for association between antidepressant use and RSWA

REM sleep atonia in humans is regulated by the glutamate-releasing neurons in the pontine sublaterodorsal nucleus which activate gamma-aminobutyric acid and glycine releasing premotor neurons in the gigantocellular and magnocellular nuclei in the ventral medulla which inhibit spinal motoneurons thereby causing motor paralysis in REM sleep. The observed association between SSRIs/SNRIs and RSWA may be due to activation of dorsal raphe serotonergic and locus coeruleus noradrenergic brainstem nuclei during REM sleep inhibiting normal REM atonia [25, 26]. This aligns with the hypothesis that serotonergic ADs primarily unmask subclinical RBD rather than cause it de novo, as patients with AD-associated RBD demonstrate neurodegenerative markers such as olfactory impairment, color vision deficits, and autonomic dysfunction that cannot be explained by AD use alone [16].

Limitations

These findings may be relevant to RBD research because AD use is common among patients referred for sleep evaluation; however, the clinical selection of this cohort limits generalizability to broader populations.

Strengths of our work include robust RSWA scoring methodology performed by four experienced sleep neurologists using current AASM criteria applied to a large clinical dataset. However, our study has limitations inherent to its cross-sectional design. First, this cohort was derived from a clinical database of vPSGs selected for manual RSWA scoring; the sample was enriched for patients with suspected parasomnia. This selection process likely increases the overall percentage of RSWA compared with an unselected sleep laboratory population. Therefore, the RSWA percentages reported in this study should not be interpreted as prevalence estimates for all patients taking ADs or for all patients undergoing vPSG. Rather, these findings apply specifically to patients who met clinical criteria for RSWA scoring at our center and should be interpreted within that context. Second, causality between AD use and RSWA/RBD diagnosis cannot be inferred. Third, we could not assess the impact of potential confounders such as use of melatonin and other therapies for RBD or medical comorbidities. Fourth, because the annotation software does not retain information distinguishing tonic from phasic EMG activity during label processing, we were unable to separately quantify tonic and phasic RSWA. Lastly, because our sample included a higher proportion of White participants, and the AD group had a higher proportion of females and was slightly younger than the non-user group, our findings may not be generalizable to community or more diverse populations. Importantly, although our methodology represents one of the most rigorous approaches to RSWA quantification to date, the process is highly labor intensive limiting its scalability across centers.