Dose-Escalation Study of Amygdalar Transcranial Focused Ultrasound in Healthy Volunteers

Introduction

Determining the optimal intensity of transcranial focused ultrasound (tFUS) needed to effectuate change in neural activity is a critical step in moving towards using tFUS as a safe and effective neuromodulatory tool. While prior studies have established the feasibility of modulating deep brain structures such as the amygdala, thalamus, and entorhinal cortex, these all operated within the Food and Drug Administration’s (FDA) output limit for diagnostic ultrasound [1–4].

In 1985, the United States FDA established exposure limits for diagnostic ultrasound established application-specific intensity limits for diagnostic ultrasound based on the highest output intensities being currently used by devices already in clinical use when the FDA gained regulatory control. These limits were expressed as derated spatial-peak temporal-average intensity (ISPTA.3), where derating applies an assumed soft-tissue attenuation of 0.3 dB/cm/MHz to better estimate in situ exposure. The FDA limits were set as follows: peripheral vascular, 720 mW/cm²; cardiac, 430 mW/cm²; fetal and other, 94 mW/cm²; and ophthalmic, 17 mW/cm² (all are derated intensity spatial peak temporal average (Ispta.3)). Derated acoustic metrics refer to ultrasound output parameters that have been adjusted using a standardized attenuation coefficient (0.3 dB/cm/MHz) to approximate in situ exposure within tissue, rather than free-field measurements obtained in water. In this study, exposure was quantified using FDA-recognized derated intensity metrics, including the spatial-peak temporal-average intensity (Ispta.3), which reflects time-averaged energy deposition and is primarily associated with thermal effects, and the spatial-peak pulse-average intensity (Isppa.3), which reflects peak energy during pulse delivery and is more relevant to mechanical bioeffects. In cooperation with the ultrasonic community, these values were revised in 1992, establishing a new limit of Ispta.3 = 720 mW/cm² for all applications except those involving ophthalmic applications. The vast majority of tFUS work uses the FDA “Track 3” requirements, which require the following: Ispta.3 must not exceed 720 mW/cm², and either the Mechanical Index (MI) must be below 1.9 or the intensity spatial peak pulse average (Isppa.3) must be below 190 W/cm². The Mechanical Index (MI) is a dimensionless parameter defined as the ratio of the derated peak rarefactional pressure to the square root of frequency (MI = Pr.3/√f), and serves as an indicator of the likelihood of cavitation-related mechanical effects. However, these limits apply to diagnostic ultrasound, not therapeutic ultrasound. Nevertheless, many researchers abide by the FDA limits for ease of gaining approval from oversight bodies (such as Institutional Review Boards (IRB)) to conduct their research. Furthermore, many publications over the last two decades have underscored both the safety record of tFUS [5–8] and its ability to non-invasively exert neuromodulatory effects[9, 10]. It is important to note that these guidelines do not imply that these are the safety limits, above which stimulation would be unsafe and that stimulation below these intensities is automatically safe.

While the FDA limit has allowed for ease in the obtaining of regulatory approval from an IRB to proceed with these studies, neuromodulation requires different considerations, as ultrasound delivered at intensities compliant with the limit may evoke non-uniform responses attributable to the stochastic nature of neuronal firing. Accordingly, variability at FDA-compliant outputs should not be taken to imply a lack of efficacy; rather, it suggests that achieving reliable neuromodulatory engagement may require operation within a parameter space not fully encompassed by diagnostic ultrasound limits. In fact, multiple studies have shown the safety of intensities exceeding the FDA limits[11–14]. Our group conducted an open-label study of ultrasound administration to the epileptogenic focus in pre-surgical temporal lobe epilepsy patients who subsequently underwent resection of the epileptogenic zone to treat their refractory epilepsy. In that work, we did not see histological changes in brain tissue when sonicating up to Ispta.3 of 5.76 W/cm² (eight times above the FDA limit)[13]. However, since in-vivo sonication requires transmission through the skin and skull, which inherently causes attenuation and some loss of energy[15], we aimed to first understand the minimum energy required to cause damage in tissue directly. An ex vivo brain tissue sample study was designed to determine safety thresholds for ultrasonic brain stimulation. In that work, pieces of brain tissue from recently deceased cadavers and surgical resections were directly sonicated in a petri dish at varying intensity levels to determine the threshold at which damage starts to appear on histology[11]. No signs of damage were observed in any samples that underwent sonication at or below 14 W/cm² Ispta.0, while inconsistent damage above this intensity included changes such as extravasation of blood vessels and spongiosis. Again, there was no skull to provide for any attenuation of ultrasonic energy. While the aforementioned study involved direct ex-vivo sonication of brain tissue, studies in living humans involve transcranial administration through skin, skull, and hair. It was therefore important to understand if the increased energy deposition associated with higher intensity stimulation (and subsequent energy absorption by the skull) would result in appreciable heating of the skin and skull. A small study in healthy volunteers involved sonication of the right scapula, administered from the dorsal surface[16]. With thermoprobes placed between the ultrasound transducer and the skin, the temperature was measured during stimulation with five different intensity levels up to 14 W/cm² Ispta.3. Even at the maximum stimulation intensity, only 1.5° C of heating were seen, which is a reasonable rise in temperature that is not expected to produce adverse events and which is below the FDA cutoff of 2° C. These findings thus suggest the presence of a safe window of increased energy deposition well above the current FDA energy limit, implying a region of ultrasound parametrization capable of increased neuromodulation and, putatively, increased therapeutic effectiveness, without a concomitant increase in risk profile.

The possibility of employing intensities above the FDA limit is particularly important, since the skull substantially attenuates ultrasound[17], which may greatly limit the effect sizes induced by ultrasonic neuromodulation. As such, at intensities compliant with the FDA limit, there is limited energy delivered to the tissue itself, which limits the neuromodulatory effect. This raises a central question: does an intensity range above the FDA limit permit stronger neuromodulatory effects without compromising safety, and is there a dose–response such that increasing intensity yields more reliable or larger changes in brain activity?

Of course, there needs to be safe, systematic, and incremental explorations of these FDA limit-exceeding intensities. Higher intensities carry the risk of ablative tissue damage or cavitation effects[18]. Perhaps most notable for the field was the likely-cavitation event that occurred at West Virginia University Rockefeller Neuroscience Institute (WVU/RNI), in which a subject enrolled in a study using tFUS for neuromodulation in severe substance use disorder was rendered comatose during treatment[19]. Notably, this study was a “significant-risk” designated study, carried out at an FDA-limit exceeding intensity and mechanical index (MI)[20]. While report of this event was released after our study and analysis had concluded, it nonetheless points to the importance of safely exploring new parameter spaces in controlled, non-significant risk studies.

Consequently, the present study helps to fill this gap through a carefully designed, dose-escalation study, conducted in healthy volunteers. Specifically, using four carefully chosen stimulation paradigms, we tested whether tFUS of the right amygdala at intensities above the FDA limit had any differential effect on a behavioral task previously validated with amygdalar tFUS[1]. Throughout this manuscript, FDA diagnostic limits are referenced in relation to free-field derated transducer output (ISPTA.3), consistent with regulatory definitions, whereas intracranial intensity estimates are provided solely for contextual interpretation and are not used as safety thresholds.

It is known that the amygdala plays a role in various emotional and behavioral responses, and that amygdalar circuits are crucial for anxiety and fear processing and play an essential role in regulating emotions[21, 22]. Given the significant role of the amygdala in emotional processing and its links to psychiatric disorders, investigating clinical interventions aimed at modulating amygdalar activity is key to making forward therapeutic progress. We also acquired clinical MR sequences to monitor for changes indicative of injury caused by tFUS[23]. We hypothesized that higher stimulation intensities would lead to more significant emotional shifts as assessed by changes in the behavioral tasks, while ensuring that the increased intensities would not result in tissue injury as assessed by structural MRI.

Materials and Methods

Results

Discussion

This dose-escalation study demonstrated the initial safety and practicality of applying tFUS to the right amygdala in healthy volunteers, using intensities well above the FDA output limit threshold for diagnostic ultrasound of Ispta.3= 720 mW/cm². Our primary safety aim was evaluated through clinical review of structural MRI, specifically, whether any of the stimulation parameter sets would create changes on structural MRI. This means of safety evaluation was employed to mirror methods used clinically during high-intensity focused ultrasound (HIFU) procedures, during which structural MRI is used to monitor for edema, hemorrhage, or unintended tissue damage during and after sonication[35]. Importantly, across all stimulation sessions, there were no structural changes observed on any MR scans. This held true even for the previously unused in-vivo stimulation conditions (Parameter Set C and D) with an Ispta.3 of 10.08 W/cm², as they did not lead to any adverse effects, despite the delivered intensity corresponding to a free-field derated Ispta.3 approximately 14-fold higher than the FDA diagnostic ultrasound output limit, which is defined at the transducer face and does not account for transcranial attenuation. Additionally, MRI scans of patients who received Parameter Set D, which has a Mechanical Index of 2.8 (exceeding the iTRUSST value of 1.9 [36] ), did not show any of the structural changes, such as edema or microhemorrhage, that are typical with HIFU-like stimulation. Furthermore, there was a lack of any adverse events or serious adverse events during any stimulation session across subjects.

We acknowledge that because CT or pseudo-CT from ultrashort-echo MR imaging was not acquired, skull acoustic properties were estimated using T1-weighted MRI, which may reduce the quantitative accuracy of modeled intracranial intensities. However, because safety conclusions in this study were based on serial neuroradiological MRI assessment rather than simulation-derived exposure metrics, this limitation does not affect the primary safety conclusions. Acoustic modeling was performed solely to demonstrate accurate targeting and obtain a crude estimate of the intracranial effects of stimulation. Additionally, we note the lack of confidence intervals for the estimated stimulation intensities, which is a limitation of the modeling software itself. Despite the lack of pseudo-CT, we note that the estimated ISPTA at the target in all instances falls within the FDA regulatory limit. This suggests that investigators may increase free-field intensities, even substantially beyond current FDA limits as done in this study, safely while continuing to deliver safe stimulation paradigms. Together, these results show a level of safety within the parameter space used in our study.

Although our sample size was small, with ten subjects, these findings still suggest the safety and practicality of these parameters for in vivo neuromodulation studies. They also raise questions about the current FDA intensity guidelines for Ispta, set at 720 mW/cm². Based on the evidence presented in this study and other research indicating that safe tFUS intensities exist beyond this limit, additional systematic investigation into higher-output parameter spaces is desperately needed to determine true safety margins prior to any regulatory reassessment of the existing limits. Such adjustments might also help facilitate the approval process for future studies within this parameter space, possibly making IRB approval easier to obtain, and allowing for faster study initiation and smoother translation of innovative tFUS protocols to clinical applications. Additionally, understanding ultrasound targeting using techniques such as magnetic resonance acoustic radiation force impulse imaging (MR-ARFI) also requires higher stimulation intensities[37].

Of course, further exploration of this parameter space must be done in a safe, controlled, and incremental manner. The WVU/RNI event serves as an unfortunate example of the possible danger when this is not done. In a followup to their initial publication, the authors provide some estimates of the parameters they used[38]. Their highest stimulation pressure in water was 3.5 MPa, which at a frequency of 220 KHz, corresponds to an MI of 7.46. Parameter set D in this study had an in-water MI of 2.8, placing it well below that of the WVU study. As such, these findings and the unfortunate incident underscore the need for systematic safety data at all stimulation intensities, with careful stepwise escalation, rigorous monitoring, and careful interpretation of studies approaching or exceeding these values. We also note that while recent iTRUSST guidance provides guidance on stimulation at biophysical levels that have never shown adverse events, these stimulations may be inadequate to elicit robust neuromodulatory effects. Further, the iTRUSST values should not be assumed to classify intensities above their stated values as inherently dangerous or unsafe. Accordingly, these guidelines should be interpreted as conservative reference points rather than definitive safety thresholds, particularly in the context of emerging neuromodulation applications where effective parameter ranges may extend beyond those historically studied. The present findings, together with prior work, suggest the existence of a broader therapeutic window that remains incompletely characterized, emphasizing the need for carefully designed dose-escalation studies that balance the pursuit of efficacy with rigorous safety oversight. Moving forward, the field would benefit from systematic mapping of this parameter space, integrating empirical safety outcomes, acoustic modeling, and real-time monitoring to more precisely define exposure limits that are both biologically informed and clinically meaningful.

In addition to supporting the structural and regulatory safety of higher-intensity tFUS, our findings also offer preliminary insights into its temporal profile, raising important questions about the duration and persistence of neuromodulatory effects following stimulation. Even with sessions spaced one week apart, there was no cumulative or residual effect on emotional reactivity or baseline emotional state, as measured by pre-stimulation behavioral task results, indicating that, even with significantly higher-intensity tFUS, neuromodulatory effects wore off between stimulation sessions. It is likely that with 10 subjects, we did not have sufficient power to detect an effect. While this further supports the temporal safety of this parameter space, it also points to an incomplete understanding of the temporal effects of tFUS neuromodulation. The precise duration of neuromodulatory effects induced by tFUS remains to be fully characterized, although prior studies in rodents and humans have demonstrated that such effects can persist for days to weeks[26, 39], underscoring the need for continued longitudinal evaluation within this new parameter space. Furthermore, much has been written about the acoustic artifact and its potential as a confounder in any analyses[40]. Unfortunately, this study was not designed to assess for the acoustic artifact, so this is another limitation to the generalization of these results. Lastly, subjective tolerability was not captured using a formal standardized questionnaire. Thus, while these results do demonstrate a preliminary level of safety of this parameter space and point towards further clinical trial testing, we are still unsure of the relationship, if any, between tFUS intensity and transient neuromodulatory effects.

The primary efficacy aim was to test whether increasing the intensity of ultrasound would produce a dose-dependent modulation of affective responses. Regarding our dose-response hypothesis, the results were rather mixed and inconclusive. Notably, no stimulation condition produced a statistically significant pre–post change relative to baseline after correction for multiple comparisons. A significant linear effect was observed in changes in negative valence across stimulation paradigms A to C; however, this effect was not supported by significant omnibus or pairwise comparisons, did not survive correction for multiple comparisons, and was not consistently reflected across related behavioral measures. Accordingly, this finding should be interpreted cautiously and does not provide robust evidence of a dose–response relationship. While the direction of this effect may be consistent with the known role of the right amygdala in negative emotional processing, the lack of statistical robustness and internal consistency precludes a reliable interpretation[41, 42]. Importantly, no participants reported subjective distress or adverse emotional effects. Given the small sample size and use of a healthy cohort, this finding should be considered exploratory and not indicative of maladaptive or countertherapeutic outcomes.

With respect to changes in arousal across paradigms A to C, the DAI and positive arousal changes did not reach statistical significance. Notably, a quadratic effect did not reach statistical significance, although a moderate effect size was observed across changes in negative arousal. While non-linear dose–response curves have been described in other neuromodulation contexts, particularly in studies targeting subcortical structures like the amygdala[12, 43], the present findings are inconclusive and likely reflect variability related to small sample size and limitations of the behavioral task.

When comparing Stimulation Paradigm C, with a high duty cycle (10.08 W/cm² Ispta.3, 70% DC) and Stimulation Paradigm D, with a low duty cycle (10.08 W/cm² Ispta.3, 5% DC), no metric reached a level of significance. It’s likely then that while these paradigms differ theoretically in their predicted neuromodulatory direction (based on high versus low duty cycle), the task used here was not sensitive enough to detect these specific effects, especially given the small sample size in this pilot study. Additionally, it’s possible that the modulation occurred at a neurophysiological level that did not lead to immediate directional changes in arousal or valence. As such, future work should ensure that concurrent neuroimaging or electrophysiology is collected to assess the polarity of the neuromodulation on a more granular level.

When conducting within-paradigm analyses, while several metrics showed nominal significance prior to correction, none survived conservative correction for multiple comparisons (Table 3), with a corrected significance threshold of α= 0.002. Given the exploratory nature of behavioral outcomes and the primary focus on safety, all statistical tests were treated as a single family and corrected using a Bonferroni procedure. Thus, combined with the small sample size, these results should be interpreted cautiously and may reflect limited power to detect small effects. Importantly, the absence of robust effects across paradigms is consistent with the intended safety properties of the system.

Collectively, these results do not provide robust evidence for a dose–response effect of tFUS intensity. Beyond the limitations imposed by the small sample size, it is also important to note that this trial was conducted in healthy subjects, and as such, the primary endpoint was safety rather than therapeutic efficacy. The tasks used are primarily meant for anxious individuals, not for healthy volunteers. Subsequently, the lack of clear and robust neuromodulatory effects was not unexpected. In fact, the lack of significant effects on affective stimuli might be considered a positive, since it suggests that the intensities used in the study did not have adverse effects on the subject’s behavior, adding another successful safety metric to this work. We anticipate that installing these higher intensity tFUS parameters in clinical populations with dysregulated affective circuitry, such as in mood or anxiety disorders, could result in clearer effects, especially in the context of targeting abnormally hyperactive or hypoactive circuits. This is especially relevant given that prior studies were constrained by the FDA’s diagnostic ultrasound intensity limit of Ispta.3 = 720 mW/cm² may have underdelivered energy relative to what is needed for clinically meaningful engagement. Ultimately, replication of these findings in patient populations, with symptom-specific targets and outcomes, will be essential to determine the true translational value of higher-intensity tFUS.

Conclusion

This dose-escalation study provides preliminary evidence that low-intensity neuromodulatory tFUS delivered using free-field outputs above current FDA diagnostic output limits can be applied safely to the right amygdala in healthy volunteers. Across all stimulation sessions, structural MRI and behavioral monitoring revealed no adverse events, edema, hemorrhage, or other tissue disruption, even at intensities exceeding the current FDA diagnostic ultrasound output limit and the iTRUSST recommended maximum. These findings provide empirical data relevant to ongoing discussions about whether diagnostic ultrasound output limits are optimally calibrated for neuromodulation research, while underscoring the need for larger, multi-site studies before any regulatory reconsideration. While exploratory analyses suggested possible dose-dependent effects on affective processing, particularly in negative valence and arousal, these results were inconsistent and likely limited by small sample size and the use of healthy controls. The complete absence of adverse effects, together with preliminary signals of neuromodulatory engagement, supports the continued evaluation of higher-intensity parameters in clinical populations. Future studies with larger samples, patient cohorts, and concurrent functional neuroimaging will be essential for clarifying both the therapeutic potential and mechanistic specificity of tFUS beyond current FDA and iTRUSST limits.

Supplementary Material