Pharmacological treatments are often curbed by intolerable side effects or low effectiveness of drugs (Lyons and Pahwa, 2004; Bystritsky, 2006; Zesiewicz et al., 2010; Al-Harbi, 2012; Elias et al., 2013; Elias et al., 2016; Jaffe et al., 2019). Consequently, millions of patients remain resistant to treatments and suffer from poor quality of life. There is a critical need for approaches that deliver medication selectively into the desired target in the body at a high concentration while sparing surrounding tissues.

Pioneering work in this domain utilized temperature-sensitive liposomes (Needham and Dewhirst, 2001) that can be activated by heat or radiation. However, localized heating at depth is challenging to achieve safely and in a controlled manner. Due to this issue, recent efforts have shifted to using low-intensity ultrasound as a safe and practical form of energy for localized drug release. Ultrasound can be remotely focused into a biological target at depth. Impacting circulating drug carriers, ultrasound triggers targeted release with minimal off-target effects.

Several groups have shown that ultrasound can trigger drug release from nano-sized structures stabilized with biocompatible polymeric shells (Rapoport et al., 2007; Rapoport et al., 2009; Rapoport et al., 2011; Rapoport et al., 2015; Rapoport, 2016; Airan et al., 2017; Boissenot et al., 2017; Wang et al., 2018; Wu et al., 2018; Zhong et al., 2019; Lea-Banks et al., 2021; Lea-Banks and Hynynen, 2021; Somaglino et al., 2021), and we have observed effective release in the brain of non-human primates (Wilson et al., 2024).

These nano-sized structures have been commonly filled with perfluorocarbon (PFC) cores. PFCs are highly inert and bestow the resulting nanodroplets with sensitivity to ultrasound. When exposed to ultrasound, the PFC core has been hypothesized to change phase from liquid to gas, greatly expanding in volume, and thus mediating drug release (Kripfgans et al., 2000; Sheeran et al., 2013; Doinikov et al., 2014; Shpak et al., 2014; Rapoport, 2016). Harnessing this understanding, a majority of previous studies used nanodroplets with PFC boiling points below body temperature (Rapoport et al., 2007; Rapoport et al., 2009; Rapoport et al., 2011; Sheeran et al., 2012; Rapoport et al., 2015; Rapoport, 2016; Airan et al., 2017; Wang et al., 2018; Wu et al., 2018; Zhong et al., 2019; Lea-Banks et al., 2021; Lea-Banks and Hynynen, 2021). However, successful translation of these nanocarriers has been hampered by their instability. By contrast, PFCs with high boiling points are considered to be highly safe and stable and have been well-studied for use in large quantities (Castro and Briceno, 2010; Li et al., 2018). Nonetheless, such nanodroplets have provided relatively ineffective release (Boissenot et al., 2017; Somaglino et al., 2021). In this study, we sought to identify the ultrasound parameters necessary to trigger drug release from high boiling point nanodroplets. We further sought to investigate the biocompatibility and safety of these stable nanodroplets in a nonhuman primate over a repeated dosing regimen.

In particular, this study assessed whether low-frequency ultrasound could enable drug release from stable, high boiling point drug carriers. According to the prevailing hypothesis, ultrasound delivered at the target induces vaporization of the PFC core (Sheeran et al., 2013; Doinikov et al., 2014; Shpak et al., 2014; Rapoport, 2016). This vaporization may be related to either ultrasound’s mechanical or thermal effects, which vary depending on frequency. High frequencies are more likely to engage thermal mechanisms by depositing more energy at the target. PFCs with high boiling points are unlikely to be influenced strongly by thermal effects and, thus, are expected to release drugs in response to low-frequency ultrasound more effectively. By contrast, high-frequency ultrasound, which has been used predominantly thus far, is likely to favor low boiling point PFCs. In this paper, we demonstrate that the combination of nanodroplet formulation and ultrasound parameters could provide remotely controlled, effective, and safe drug delivery. We also describe the procedure to produce the nanoparticles and release results consistently.

This study investigated the effectiveness and safety of drug release from biocompatible nanodroplets triggered by focused ultrasound. Uniquely, we evaluated the release efficacy from nanodroplets filled with PFCs with distinct boiling points, including boiling points above the body temperature, and investigated the relevant ultrasound parameters that activate them. We found that low frequency ultrasound (300 kHz in particular) was more effective at mediating release in general and that low frequencies are paramount for effective release from PFCs with a high boiling point.

The core has been hypothesized to be a critical factor in governing the effectiveness of ultrasound-based drug release (Rapoport et al., 2011; Wu et al., 2018; Somaglino et al., 2021). Indeed, our 900 kHz data show (Figure 1B, red) that at the range of commonly applied ultrasound frequencies (about 1 MHz and greater), the core is a key determinant of release effectiveness. We found that the higher the boiling point of the encapsulated PFC, the lower the release effectiveness. This finding is in accord with previous results (Zhang et al., 2010; Doinikov et al., 2014; Rapoport, 2016; Wu et al., 2018). Critically, we found that lowering the frequency to 300 kHz increased the release effectiveness at a general, core-independent level (Figure 1). This effect can also be found in a study that used PFP (Zhong et al., 2019). The overall effect of decreasing frequency may be enhanced by the increase in focal volume, but the differences between cores at each frequency remain informative. The application of low-frequency ultrasound thus opens the path to using high boiling point cores (e.g., PFOB) as release actuators. A frequency of 300 kHz provides spatial resolution on the order of several millimeters, which approximates the focal volume of the drug released in tissue by ultrasound (Wang et al., 2018; Lea-Banks and Hynynen, 2021). While the focal size is larger than ultrasound of higher frequencies, the focal size remains applicable for confined targets, including in the brain (Weniger et al., 2006; Bette et al., 2018).

This work uniquely evaluated the safety of ultrasound responsive PFC nanodroplets over 6 weeks of repeated dosing in a nonhuman primate. This type of dosing regime may be necessary for long-term treatments. The minimal impact on measures of toxicity in a nonhuman primate over this time period are a key indicator of potential for clinical translation. Indeed, only blood glucose was found to show a statistically significant change at 1 day after the nanodroplet administration. The level stabilized at day 7. This change in blood glucose may be attributed to diet—the monkey received sweet juice as a reward on the days of the nanodroplet administration. Since we did not fast the animals in these experiments, we could not reliably use the blood glucose level as a metric of pancreatic inflammation (Sun et al., 2019). Without fasting, most of the circulating glucose is attributed to diet (McMillin, 1990).

This longer-term analysis aligns with our previous work, which quantified the immediate response to nanodroplet administration (Wilson et al., 2024). That study found a modest and transient increase in white blood cell count. The increase was within the normal range. We observed no significant change in white blood cell count in this study, which indicates that any immune system activation resolves within the first day following administration.

The majority of studies that applied ultrasound to PFC-based nanodroplets used PFCs with boiling point below the body temperature (Rapoport et al., 2007; Rapoport et al., 2009; Rapoport et al., 2011; Sheeran et al., 2012; Rapoport et al., 2015; Rapoport, 2016; Airan et al., 2017; Wang et al., 2018; Wu et al., 2018; Zhong et al., 2019; Lea-Banks et al., 2021; Lea-Banks and Hynynen, 2021). Although the Laplace pressure of sub-micron particles can effectively increase the boiling point when encapsulated (Rapoport, 2012), PFCs with low boiling points may suffer from instability issues, which has raised safety concerns. Indeed, our data show that PFP-based nanodroplets not exposed to ultrasound spontaneously increase in size over a short time window, whereas PFOB nanodroplets remain stable (Figure 3). The increase in size can be attributed to either Ostwald ripening or droplet vaporization. Ostwald ripening in emulsions is the process of larger droplets becoming larger by diffusion of the PFC from smaller droplets. This effect is diminished for higher molecular weights (Kabalnov and Shchukin, 1992). This applies to our findings—PFOB has a molecular weight of 498.97 g/mol while PFP is only 288.04 g/mol. Spontaneous vaporization of the PFP core may also be at play, since the boiling point of PFP is close to room temperature. Further, low boiling point PFCs can form persistent microbubbles after vaporization by ultrasound (Rapoport et al., 2009; Rapoport, 2016; Lea-Banks et al., 2021). A combination of these effects may also be at play in the observed increase in size: Ostwald ripening may lower the vaporization threshold of the PFC core by reducing the Laplace pressure. These data suggest that PFP nanodroplets, if at all, should be used immediately following their production, or else be stabilized, as accomplished previously by freezing with added glycerin (Wang et al., 2018). It is nonetheless unclear if PFP nanodroplets would expand more quickly than they are cleared from the bloodstream in vivo; more work is required on this matter. Our finding that PFOB can be used for effective release from nanodroplets at low ultrasound frequencies (Figure 1; Figure 2) should mitigate the boiling-point associated concerns. Indeed, PFOB-based products including LiquiVent—an oxygen-carrying liquid drug, and Oxygen (both Alliance Pharmaceutical Corporation, San Diego, CA, United States of America)—a blood substitution agent (Cohn and Cushing, 2009; Castro and Briceno, 2010; Li et al., 2018) have been thoroughly studied for use in liter-quantities.

Perfluorocarbons are cleared via the reticuloendothelial system (RES), primarily by macrophage phagocytosis, and ultimately eliminated via exhalation from the lungs (Flaim, 1994). The clearance half-life of PFOB is estimated to be 3 days (Riess, 1988; Flaim, 1994). In clinical trials, side effects are typically transient and primarily associated with activation of macrophages at doses much higher than used here (Flaim, 1994; Flaim et al., 1994). PFCs have been studied at doses starting around 2 g/kg for blood substitution applications, while our maximum dose was only 0.14 g/kg.

The in vitro release data presented here also contribute to the understanding of the release mechanism. Thus far, the predominant hypothesis of action has been the vaporization of the PFC droplet upon the impact of focused ultrasound (Rapoport et al., 2011; Sheeran et al., 2013; Doinikov et al., 2014; Shpak et al., 2014; Wu et al., 2018; Somaglino et al., 2021). In this mechanism, the thermal and mechanical aspects of propagating ultrasound exceed the vaporization threshold governed by the PFC boiling point and the Laplace pressure. The PFC core increases in size (up to a 5-fold increase in diameter upon vaporization (Rapoport, 2012)). This increase in diameter is expected to increase the rate of drug diffusion through the polymer shell, contributing, in addition to any superimposed effects of ultrasound, to drug release. Indeed, our data at the common range of ultrasound frequencies, 900 kHz, provide two lines of evidence in support of this hypothesis. First, we found that the release increases with decreasing boiling point (Figure 1B), as expected from the vaporization hypothesis. Second, thermal energy delivered by ultrasound is known to be proportional to pressure squared (Cobbold, 2006). We indeed found a quadratic dependence of the release on pressure (Figure 2B).

Our and a previous study (Zhong et al., 2019) suggest that ultrasound of frequencies lower than those in the common diagnostic range may mediate more effective drug release. Lower frequencies are known to accentuate mechanical effects, which can take two forms. First, ultrasound can induce cavitation, the formation of gaseous nuclei from dissolved gasses under the low-pressure epochs of the ultrasound pressure wave (Apfel and Holland, 1991). Cavitation can provide useful mechanical forces until it exceeds a threshold at which the formed gaseous nuclei collapse and cause damage (Apfel and Holland, 1991). From the FDA’s 510(k) Track 3 standpoint (FDA, 2019), cavitation is unlikely to occur for mechanical index—defined as ultrasound pressure divided by the square root of frequency—values below 1.9. In our hands, a 300 kHz, 1.0 MPa pressure at target yields a mechanical index of 1.83. Despite being below the value of 1.9, this pressure level already drives significant release (Figure 2A). Thus, if cavitation is involved, it likely constitutes one of multiple mechanisms. Another candidate mechanism for drug release is particle displacement. The maximal displacement of a particle in the ultrasound path, ξ _m , is linearly proportional to the ultrasound pressure amplitude P: ξ m = P Z ∫ 0 T / 2 ⁡ sin ( 2 π F t ) d t = P Z π F , where F is the ultrasound frequency and Z the acoustic impedance. Indeed, our 300 kHz data show a linear dependence of the release on pressure (Figure 2A). This mechanism is supported by high-speed imaging, which did not detect a persistent phase change despite effective release in rodents (Zhong et al., 2019). Together, our findings indicate that both thermal and mechanical effects can be at play, depending on the applied ultrasound frequency and the PFC core selected. Higher frequencies accentuate thermal effects (PFC vaporization) and require low-boiling point cores, whereas lower frequencies accentuate mechanical effects (cavitation or cycle-by-cycle particle displacement) and enable the use of stable, high-boiling point cores.

The mechanical nature of the drug release effect at low frequencies is beneficial for clinical translation, as no heating of the target tissue is required. Indeed, we have observed no signs of damage in diagnostic MRI at the ultrasound target with 30 ms ultrasound pulses at 1.5 MPa in nonhuman primates (Wilson et al., 2024). To minimize the potential for heating, clinical studies may use shorter pulses and lower duty cycles, which decreases the deposited energy. The risk of heating can also be reduced using an approach that corrects for the skull and hair to deliver a predictable dose of ultrasound (Riis et al., 2023; Riis et al., 2024). To further reduce the risk of unwanted heating in clinical studies, MR thermometry could be used to monitor for potential temperature rise at the focus and throughout the brain. Cavitation detection systems have also be used to identify the minimum threshold to trigger drug release (Lea-Banks and Hynynen, 2021), thus minimizing the delivered energy.

Interestingly, at 300 kHz, we observed a marginal decrease in drug release from DFP and PFOB nanodroplets at the highest ultrasound pressure tested (Figure 2A). This effect has been also observed in another study (Lea-Banks and Hynynen, 2021). These high pressures may lead to cavitation—the formation of bubbles—which may consequently block the propagation of ultrasound into the target (Lea-Banks and Hynynen, 2021). PFP did not exhibit this effect, possibly due to the lower boiling point—complete vaporization and subsequent collapse of PFP bubbles may prevent shielding effects with the repeated pulsing scheme used here.

The study has four limitations. First, we tested the effects of ultrasound frequency and pressure, but not the ultrasound waveform. For instance, we have not tested the effects of the duty cycle and only used a fixed value of 10%. Increases in the duty cycle are known to lead to stronger, cumulative effects (Airan et al., 2017; Zhong et al., 2019). Optimizing the ultrasound parameters would help to maximize safety and efficacy of drug release by avoiding unneeded increases in ultrasound pressure or duration. For example, pulse duration has been evaluated systematically, indicating that pulses longer than 50 ms reach diminishing returns (Airan et al., 2017). Second, we quantified the release of nanodroplets using a single copolymer shell (PEG:PDLLA). This limitation is mitigated by a previous study that found little variability in the release due to specific forms of the polymeric shell (Zhong et al., 2019). However, different polymer shells may change the interactions of the nanodroplets and the immune system. This has substantial implications on the drug released in vivo. For example, longer PEG chains have been used to extend blood circulation time (Pata et al., 2003). Further, while nanodroplet size did not affect drug release or encapsulation rates in this study, it may affect the pharmacokinetics. Future work in vivo could help optimize the composition of these droplets. Third, the selection of parameters for the drug release study limits our ability to make definitive conclusions about the mechanism of action. As mentioned previously, we uncovered the involvement of a mechanical effect in the drug release from high-boiling point PFCs, but whether the effect is attributed to cavitation, particle displacement, or another mechanism remains unclear. Studies of the acoustic emissions from the nanodroplets could quantify the likelihood of cavitation (Xu et al., 2020; Welch et al., 2022), and high-speed imaging studies could identify the presence of vaporized droplets (Sheeran et al., 2013; Doinikov et al., 2014; Puett et al., 2014). Finally, the in vitro drug release assay does not fully replicate the conditions in vivo, which we have evaluated in a dedicated study (Wilson et al., 2024). Blood contains many proteins and cells which may interfere with the function of the nanodroplets or diffusion of drug after activation, and uptake into brain tissue may differ from uptake into hexane. The pharmacokinetics of the nanodroplets are also a critical in vivo factor that was not addressed here. However, past studies by our group (Wilson et al., 2024) and others (Wang et al., 2018; Lea-Banks et al., 2020) have found a half-life of several minutes in circulation, which has allowed sufficient drug release to modulate behavior (Lea-Banks et al., 2020; Wilson et al., 2024). Additional studies have mapped the extent of drug release in the brain from PFC nanodroplets, indicating a precisely targeted region of release (Lea-Banks and Hynynen, 2021). Nonetheless, further work will be needed to accurately predict the amount of drug release at the target. While the absolute values for drug release presented here may not translate directly to in vivo situations, we have emphasized the relative comparisons between nanodroplet compositions and the ultrasound frequencies.

In summary, we found that low-frequency ultrasound can effectively release drugs from stable PFC-based nanocarriers. This study informs the use of specific PFC-based nanodroplets and ultrasound parameters for effective, safe, and targeted drug release in humans while providing evidence of biocompatibility in nonhuman primates over an extended period of time. To facilitate the translation into applications in humans, we report the parameters critical for robust assembly and production of the nanoparticles. Further work is still needed to more accurately predict or measure the drug dose administered at target in vivo, which depends significantly on the quantity of nanodroplets present at the target. Specifically, the field would benefit from more detailed pharmacokinetics studies and real-time feedback systems for nanodroplet concentration and drug release effectiveness. This approach has a variety of clinical applications depending on the drug loaded and the target region. A short-lasting psychoactive drug such as propofol could be used to systematically modulate specific parts of the brain in a diagnostic capacity, while longer-lasting drugs such as ketamine could then induce neuroplastic changes. Outside of neuromodulation, there are many applications which would benefit from localized release of drugs, including glioblastoma where localized release of chemotherapeutics could reduce the overall dose required, or organ transplants, where immunosuppresants could be localized to the transplanted organ. The identification of stable drug carriers and compatible ultrasound parameters for effective release provided here lower the barrier for clinical translation of these approaches.