Abstract
Background/Aim: Paclitaxel (PTX), a widely-used chemotherapeutic agent, exhibits a high rate of plasma protein binding, which severely limits its bioavailability and reduces therapeutic efficacy. This study explored a novel strategy using low-intensity, non-thermal focused ultrasound (FUS) to locally disrupt PTX-albumin binding, thereby enhancing drug delivery and tumoricidal efficacy at tumor sites without increasing systemic toxicity.
Materials and Methods: We applied sonication (600 kHz) with varying pulse durations and duty cycles to OVCAR3 cell constructs in vitro and identified the parameters that maximally enhanced PTX uptake and induced tumor cell death. Intracellular PTX concentrations and cell viability were quantified across the conditions. The optimized FUS parameters were then applied to a mouse xenograft model of ovarian cancer using athymic nu/nu mice. Luciferase-expressing OVCAR3 tumor growth was longitudinally monitored using bioluminescence imaging.
Results: The sonication parameters (70% duty cycle and 100 ms pulse duration), applied using 3 W/cm2 spatial peak temporal average intensity, optimally enhanced intracellular PTX uptake and increased cell death, independent of thermal or flow-related effects. In vivo, a single FUS treatment nearly doubled intratumoral PTX levels, without altering serum concentration. Repeated FUS sessions combined with PTX treatments over two weeks significantly suppressed tumor growth, compared to no treatment, PTX alone, or FUS alone. Histological analysis in PTX-treated groups showed that FUS did not cause additional damage to the liver, kidney, or surrounding tissues, nor did it affect peripheral blood markers of liver and kidney function.
Conclusion: FUS can reversibly unbind PTX from albumin, increasing its bioavailability specifically at tumor sites. This targeted approach enhances chemotherapeutic effectiveness without elevating systemic toxicity or causing off-target damage, highlighting FUS as a promising adjuvant strategy for improving anticancer drug delivery in solid tumors.
- Low intensity pulsed ultrasound
- therapeutic ultrasound
- ovarian cancer
- paclitaxel
- chemotherapy
- plasma protein binding
Introduction
Ovarian cancer (OC) is the fifth leading cause of cancer-related deaths in the U.S., accounting for ~13,000 deaths and nearly 20,000 new cases annually (1). Early diagnosis is often difficult due to the absence of symptoms in the early stages, and ~70% of cases are diagnosed at advanced stages, leading to poor prognosis and high recurrence rates (2). Among the first-line treatments, paclitaxel (PTX), often in combination with platinum-based agents such as carboplatin, is a widely accepted chemotherapeutic agent following cytoreduction surgery (3). PTX acts by stabilizing cellular microtubules, thereby inducing apoptosis in proliferating tumor cells (4). However, its therapeutic effectiveness is significantly limited by high binding affinity to plasma proteins such as albumin (>90%) (5). The ester side chain of PTX non-covalently binds to hydrophobic pockets in albumin (subdomains IIA) (6), reducing the concentration of unbound drug available for tumor uptake and ultimately decreasing therapeutic efficacy.
The bioavailability and biodistribution of chemotherapeutic agents are critical determinants of their therapeutic success. Traditional strategies to overcome plasma protein binding (PPB) include the use of drug delivery vehicles like liposomes, nanoparticles, or albumin-bound formulations (e.g., nab-paclitaxel) (7-9). While these approaches can improve drug delivery, they often suffer from variable efficacy, increased production complexity, and lengthy development times before clinical implementation. Moreover, these systemic interventions lack the spatial specificity required to ensure that enhanced delivery occurs exclusively at the tumor site, thereby posing a risk of off-target effects and systemic toxicity. Although increasing systemic doses can improve delivery, it also raises the likelihood of severe side effects, including nephrotoxicity, vasculitis, and peripheral neuropathy (10-16). This challenge highlights the need for a novel, noninvasive technique capable of regionally disrupting drug-albumin binding, thereby increasing the local concentration of free drug molecules without altering the systemic dose.
Advances in focused ultrasound (FUS) techniques have allowed for the noninvasive delivery of acoustic energy to a small region of biological tissue measuring only a few millimeters in diameter, with flexibility in controlling the depth as well as the size of the acoustic focus (17-21). When continually administered at high intensity (typically >5 kW/cm2), FUS can thermally ablate solid tumors – a process known as high-intensity focused ultrasound (HIFU) ablation (22, 23). However, this thermal approach can cause irreversible damage to normal tissues along the acoustic path, necessitating careful image guidance and temperature monitoring (23-25). Alternatively, when combined with exogenously administered microbubbles, FUS can enhance chemotherapeutic drug penetration into tumors by inducing local microvascular injury through acoustic cavitation (26, 27). While this method improves drug delivery, vascular damage may carry the risk of promoting metastatic niches (28, 29).
Application of ultrasound, given at low intensity (well below the level that can elevate tissue temperature or induce cavitation) in pulsed manner, has been shown to unbind phenytoin (~90% albumin binding) (30, 31), lidocaine (~60% α1-acid glycoprotein binding) (32), and finasteride (>90% albumin binding) (33) from plasma proteins, thereby increasing their pharmacological efficacy. Building on this evidence, we previously demonstrated that the pulsed application of FUS to the tumor and its surrounding vasculature can disrupt cisplatin’s binding with albumin, enhancing its intratumoral delivery and reducing the volume and growth of xenografted cervical tumors in mice (34). In this study, we investigated the potential of low-intensity, non-thermal FUS to unbind PTX from albumin within the tumor vasculature to increase the local concentration of free drug available for tumor uptake. We hypothesized that FUS can reversibly disrupt PTX–albumin complexes at the tumor site, thereby locally increasing PTX bioavailability and enhancing its chemotherapeutic effects. This disruption is expected to occur only within the acoustic focus during sonication, with effects rapidly reversing outside the focal area, allowing systemic PTX levels to remain unchanged without elevating systemic toxicity.
To test this hypothesis, we conducted a series of in vitro and in vivo experiments using the human-derived NIH:OVCAR3 ovarian cancer cell line (hereafter referred to as OVCAR3), a high-grade serous ovarian carcinoma (HGSOC) cell line. The in vitro studies were designed to optimize the acoustic parameters – specifically pulse duration and duty cycle – that most effectively enhanced PTX uptake and induced tumor cell death without producing thermal effects. Intracellular PTX concentrations and cell viability were quantified using fluorescence colorimetric assays across a range of sonication conditions. Additionally, equilibrium dialysis (ED) was used to confirm FUS-induced unbinding of PTX from albumin. The optimized FUS parameters were subsequently applied in vivo using the mouse orthotopic xenograft model of OC. Tumor growth and localization were longitudinally monitored using bioluminescence imaging (BLI) of luciferase-expressing OVCAR3 tumors. Mice were randomized into four treatment groups to evaluate the individual and combined effects of repeated PTX and FUS administration: 1) PTX only (Us−/PTX+), 2) FUS only (Us+/PTX−), 3) combined FUS and PTX (Us+/PTX+), and 4) untreated control (Us−/PTX−). Quantitative outcome measures included tumor weight at sacrifice. Additionally, the effects of a single FUS session on intratumoral and serum PTX levels were assessed to evaluate localized drug delivery.
Materials and Methods
FUS device. We used a lab-built, single-element FUS transducer (Figure 1A) operating at a fundamental frequency of 600 kHz. A plano-concave polyetherimide acoustic lens with a 20 mm radius-of-curvature was attached to the piezoceramic disc (19.1 mm diameter, lead zirconate titanate; APC International Ltd., Mackeyville, PA, USA), which generated an acoustic focus located 21 mm from the transducer’s exit plane [~3 mm in diameter and ~11 mm in length, shown at full width at half maximum (FWHM) intensity, Figure 1B]. Sinusoidal signals from a function generator (33210A; Keysight, Santa Rosa, CA, USA) were amplified using a 10 W linear amplifier (Sonomo 500; Electronics and Innovation Ltd, Rochester, NY, USA) with acoustic impedance matching. The input voltage was calibrated to acoustic intensity and pressure at the focal location using a calibrated hydrophone (Onda Corporation, Sunnyvale, CA, USA). Acoustic coupling between the transducer and the skin was achieved using a compressible polyvinyl alcohol (PVA) hydrogel (9% by weight, prepared with two freeze-thaw cycles, 18h:6h=freeze:thaw), which was molded to fit inside a 3D-printed plastic guiding cone (16 mm in height, Figure 1A). The setup allowed the placement of acoustic focus up to 5 mm depth below the scalp surface.
In vitro and in vivo experimental setups. (A) 600 kHz focused ultrasound (FUS) transducer with an acoustic coupling hydrogel and the cone-shaped housing, (B) acoustic pressure map longitudinal to the sonication direction (left) and transversing the acoustic focus (dotted line, right). Bar: 5 mm, (C) OVCAR3 layers on the membrane insert. Bar: 100 μm, (D) BSA-containing chamber that is used to sonicate the tumor model of OVCAR3, (E) A BLI chamber compatible with operations in a biosafety cabinet, and (F) the animal platform inside the chamber for tumor imaging.
OVCAR3 and OVCAR3-luc cell lines. OVCAR3 cells (HTB-161, ATCC, Manassas, VA, USA) were cultured for approximately 5 days in RPMI-1640 medium containing 1% penicillin and streptomycin, 0.01 mg/ml human recombinant insulin (12-585-014, Gibco, Fisher Scientific, Waltham, MA, USA), and 20% synthetic fetal serum (Fetalgro, RM Bio, Missoula, MT, USA) to establish both in vitro and xenograft tumor models. For the assessment of BLI-guided FUS and tumor size monitoring, a luciferase-transfected OVCAR3 cell line (OVCAR3-luc; VLU1B033; Editgene, City of Industry, CA, USA) was obtained and cultured using the same media. To enrich the luciferase-positive cell population, puromycin (1.0 μg/ml; Sigma, Burlington, MA, USA) was added every 8–10 passages and maintained at 0.5 μg/ml thereafter. All cells used for implantation were confirmed negative for Corynebacterium bovis (C. bovis) and lymphocytic choriomeningitis virus (LCMV).
In vitro assessment of sonication parameter-dependent PTX delivery. To examine the impact of varying FUS sonication parameters on PTX unbinding and subsequent enhancement of its delivery to OVCAR3 cells, we first conducted an in vitro study. The OVCAR3 cells were seeded onto transmembrane inserts (25 mm diameter polyester membrane with 0.4 μm porosity and 50 μm thickness, ThermoFisher, Waltham, MA, USA) at a density of 0.25×106 cells in 1 ml of culture media. After 1 week of culture (OVCAR3 layers formed on the membrane insert, Figure 1C), the in vitro tumor construct was immersed in media containing 45 mg/ml bovine serum albumin (BSA, A8806, Sigma) and 0.5 μg/ml fluorescence-labeled PTX (fPTX, Oregon Green 488, P22310, Thermo Fisher), using the sterilized sonication chamber (Figure 1D). fPTX was used to enable colorimetric-based quantification, which offers greater detection sensitivity than direct PTX extraction and analysis via liquid chromatography–mass spectrometry (LC-MS/MS), particularly given the limited cell number in the tumor constructs. Sonication was applied to the center of the construct through the optically and acoustically transparent membrane, with the setup maintained inside an incubator at 36.5°C. The inner wall of the sonication chamber was lined with rubber pads to absorb reflected acoustic waves.
A total of six different sonication pulsing parameter sets (Table I), combining two duty cycles (DCs, 50% and 70%) and three pulse durations [PDs: 50, 75, and 100 ms, with pulse repetition frequencies (PRFs) adjusted accordingly], were applied to OVCAR3 tumor construct for 30 min (n=8 per parameter). A 100% DC was avoided due to its lower unbinding efficacy and to minimize the risk of heat deposition (30). The acoustic intensity was held constant at 3 W/cm2 spatial-peak temporal-average intensity (ISPTA), with the corresponding spatial-peak pulse-average intensity (ISPPA) adjusted for each DC (Table I). All intensities were below the mechanical index (MI) limit of 1.9, set for clinical ultrasound imaging (35-37). The fPTX-BSA bath was replenished between measurements, and bath temperature was recorded before and after each FUS session using an infrared thermal camera (C3; Teledyne FLIR; Wilsonville, OR, USA). As all conditions shared the same ISPTA of 3 W/cm2, temperature was measured once across the six conditions.
Six sets of acoustic parameters.
To evaluate the impact of PTX delivery on OVCAR3 death, parallel sets of in vitro tumor model were prepared using non-luciferase OVCAR3 (to avoid confounding interference of fluorescence), and the experiment was repeated for eight samples for each parameter set. Following sonication, five of the eight membrane inserts per group were rinsed with phosphate-buffered saline (PBS; pH 7.4, 10023, Gibco, ThermoFisher) and incubated in fresh medium for an additional 24 h. The remaining three inserts were cultured in PTX-containing medium to evaluate the impact of continued drug uptake. Cell death was assessed using the ethidium homodimer-1 (EthD-1) assay (E1169; Invitrogen, Carlsbad, CA, USA).
Colorimetric assessment of fPTX. To estimate the concentration of fPTX uptake in the in vitro tumor models, we employed a colorimetric calibration based on known concentrations of fPTX. First, fPTX was dissolved in ethanol (200 μg/ml), then added to the culture media and aliquoted into transmembrane inserts at four concentrations: 0, 0.1, 0.3, and 0.5 μg/ml. The samples were imaged using a microscope-mounted camera (AmScope MU2003-Bi, 200 MP CMOS sensor; AmScope, Irvine, CA, USA) at 15-s exposure time and ×5 gain, through a ×10 objective lens (TS100; Nikon, Tokyo, Japan). ImageJ software (National Institute of Health, Bethesda, MD, USA) was used to extract the green color channel in 8-bit format, and signal intensities were measured from a 100-pixel square region of interest (ROI), generating a linear calibration curve with respect to concentration.
Confirmation of fPTX-albumin unbinding via equilibrium dialysis. We performed equilibrium dialysis (ED) to measure the concentration of fPTX that was unbound from albumin and diffused into dialysis cassettes (Slide-A-Lyzer; 7-kDa molecular weight cutoff; Thermo Fisher). A 180 ml volume of the fPTX-BSA solution was added to the chamber (Figure 1D). The solution was not degassed and had a dissolved oxygen level of 6 ppm, measured using a dissolved oxygen assay kit (K-7512; CHEMetrics, Midland, VA, USA). After hydration, 0.5 ml of normal saline (NS, without BSA or fPTX) was introduced into the dialysis cassettes. One cassette was positioned 1 mm posterior to the center of the sonication focus, while the other was placed outside the sonication field as a control. The sonication parameter that yielded the highest cellular uptake (70% duty cycle, 100 ms pulse duration, 3 W/cm2 ISPTA; see Results section) was applied for one hour. The experiment was conducted at room temperature (26.5±0.2°C), with the solution equilibrated to ambient conditions. A total of ten measurements were taken, with the fPTX-BSA bath replenished between trials.
Bioluminescence imaging for FUS guidance. All experiments involving animals were conducted with approval from and according to the rules set forth by the Institutional Animal Care and Use Committee (IACUC). A BLI chamber was developed to guide the location of FUS sonication to luciferase-labeled OC tumor (Figure 1E). The mouse positioning platform was mounted to a vertical stage via a robotic gimbal [90° rotation for horizontal and lateral views, programmed using operational scripts for Arduino Uno microcontroller board (Arduino, Somerville, MA, USA); Figure 1F]. The platform included an anesthesia mask, and fabric tape was used to secure the mouse during rotation. It was enclosed in a light-proof plastic housing equipped with thermal padding to maintain body temperature during BLI imaging. Both bright-field and BLI images were acquired at the same location using a digital camera with a wide-field CMOS sensor (23.4×15.6 mm, 4,592×3,056 pixels; NEX-5; Sony, Tokyo, Japan).
To enable tumor targeting based on BLI data, the spatial reference coordinates between the transducer location and the planar reference coordinates (x–y) of the 3-axis robotic stage were co-registered to the animal platform using a known reference object (a ruler) visible in the bright-field image. Tumor coordinates identified in the BLI image were then converted to robotic stage coordinates and input to the 3-axis robotic stage (Newmark Systems, Rancho Santa Margarita, CA, USA) to align the FUS focus with the tumor. Control scripts were written in Visual Basic (Microsoft, Redmond, WA, USA) using the Active-X Toolkit (Galil Motion Control, Inc., Rocklin, CA, USA). The user-defined spatial coordinates determined the transducer’s path and step size. The brushless servo motor robotic stage achieves a maximum speed of 10 cm/s with 15 μm accuracy. Separately, another FUS platform was developed, enabling simultaneous sonication of two animals using two FUS transducers mounted on lockable articulating arms (RS series, Tether Tools, B&H Photo, New York, NY, USA). All equipment were designed to be compatible with biosafety cabinet use, allowing experiments on immunocompromised mice (e.g., athymic nu/nu).
Determining the absence of flow-related effects. Low-intensity ultrasound, as routinely adopted in clinical applications of Doppler ultrasound and microvascular flow imaging, does not impact vascular flow. However, since the proposed FUS-mediated disruption of PTX PPB will occur within the vasculature, the putative contributions from ultrasound-induced convective acoustic streaming or other unknown vascular-related phenomena (e.g., vasodilation) need to be examined. Thus, we used laser speckle contrast imaging (LSCI) technique (38) with a near-infrared 785 nm laser (RFLSI-ZW; RWD Life science, Sugar Land, TX, USA) to noninvasively evaluate the vascular flow before and during the sonication. FUS was dorsally applied toward the abdomen of normal phenotypic female mice (C57BL/6, n=2; Charles River Laboratories, Wilmington, MA, USA) for 5 min (600 kHz FF, 100 ms PD, 70% DC, 3 W/cm2 ISPTA), and the dermal vascular flow was ventrally measured for 30 s (3 Hz sampling rate).
OVCAR3 xenograft model of ovarian cancer. Female athymic nu/nu mice (NU/J, homozygous for Foxn1, Jackson Laboratory, Bar Harbor, ME, USA) were selected due to their lack of functional T-cell response (39, 40). Prepared OVCAR3-luc cells [2×104 in 10 μl of Hanks’ Balanced Salt Solution (HBSS)/Matrigel, 1:1 ratio] were unilaterally implanted into the ovarian bursa of 38 mice (41), with the implantation side randomized and balanced across groups. Procedures were performed under inhalation anesthesia (isoflurane). A gas-tight 10 μl syringe with a 30 G needle (Hamilton, Reno, NV, USA) was used for cell injection. Additional twelve mice received unlabeled OVCAR3 cells to evaluate fluorescence-tagged paclitaxel (fPTX) uptake influenced by a single FUS session. After surgical wound closure (Vicryl 7-0; ETHJ488G; Ethicon, Raritan, NJ, USA), animals were allowed to recover for two weeks. All mice were monitored two to three times per week for health status (weight and behavior), with at least one day between monitoring sessions.
Sonication parameter and skin temperature monitoring. We used the sonication parameters optimized from in vitro experiment for the xenograft OC mouse model experiments. All sonication was delivered for 30 min with application of ultrasound gel on the skin (Aquasonic 100; Parker lab, Fairfield, NJ, USA). Skin shaving was unnecessary due to the hairless nature of athymic nu/nu mice. Although sonication was applied at a low intensity that does not elevate tissue temperature, we measured skin temperature before and after each FUS session using an infrared thermal camera (C3, Teledyne FLIR). No change in skin temperature was observed before and after sonication, with measurements consistently ranging from 26-28°C (n=10). Thus, this procedure was omitted in the subsequent experiments.
Assessment of intratumoral PTX uptake and serum PTX level resulting from a single FUS session. After a three-week tumor development period, mice that underwent OVCAR3 xenograft procedure (n=12) were assigned to one of two treatment groups: 1) administration of fPTX alone (1 μg/g body weight, n=6), or 2) administration of fPTX followed by FUS applied to the tumor site (n=6). FUS was administered approximately five min after fPTX injection under ultrasound imaging guidance (4 MHz, B-mode, CMS, Contec Medical, Hebei, PR China), with the tumor location confirmed visually as a subcutaneous bump. Body weights were equivalent between the two groups (FUS+ vs. FUS−=24.8±0.6 g vs. 25.0±0.4 g; two-tailed t-test, p=0.82). All animals were sacrificed three hours post-injection, and ~0.5 ml of blood was collected transcardially to assess serum fPTX levels. Blood samples were allowed to clot at room temperature for 30 min and then centrifuged at 2,600 × g for 10 min in serum isolation tubes (Serum Gel CAT, 1.1 ml; Sarstedt, Nümbrecht, Germany) to recover serum. Colorimetric calibration was performed on a glass slide using 20 μl aliquots of known fPTX concentrations. Tumors were harvested immediately and subjected to fluorescence imaging within a 30-min window, using the same imaging protocol as in calibration. Tumor regions were segmented by applying an intensity threshold to isolate the top 50% of the signal spectrum, and mean fluorescence intensity was calculated to estimate fPTX uptake. Serum fPTX concentrations were measured by analyzing the mean intensity within a circular ROI containing 100,000 pixels.
Enhancing the antitumoral effects of PTX using FUS. Inhalation anesthesia (isoflurane) was used for all procedures. OVCAR3-luc cells were unilaterally implanted into the ovarian bursa, with the implantation side randomized and balanced across groups. Tumor location was monitored weekly using BLI, starting two weeks after OVCAR3 injection. D-luciferin potassium salt (Syd Lab, Natick, MA, USA) was dissolved in Dulbecco’s phosphate-buffered saline (DPBS) and administered via intraperitoneal (i.p.) injection at a dose of 150 μg/g body weight. After an incubation period of 8–10 min to allow luminescence expression, the mouse was placed in the BLI chamber, and images were acquired with a 5-min exposure time. Corresponding bright-field images were also captured from the same field of view. Among the 38 mice that received OVCAR3-luc cells, 37 developed BLI-detectable tumor masses by week 3.
Tumor-bearing mice were randomly assigned to four groups three weeks post-implantation and underwent the following procedures over a two-week period: 1) receiving PTX only (tail vein injection, 2 μg/g body weight, twice per week with >2-day intervals; Us−/PTX+, n=9), 2) receiving the same PTX dose (i.v.) followed by focused ultrasound (FUS) targeted to the tumor site within 5 min of injection (Us+/PTX+, n=11), 3) receiving FUS alone (same duration and frequency as Group 2) without PTX (Us+/PTX−, n=9), and 4) receiving no treatment (Us−/PTX−, n=9). PTX, initially dissolved in ethyl alcohol, was diluted in normal saline (NS) to a final concentration of 0.5 μg/μl. The injection duration (~1 min) was standardized across all groups. One mouse in the Us+/PTX+ group (out of 11) died early due to an unknown cause.
Thermal dose calculation. A numerical simulation was performed using the HIFU Simulator (v1.2) (42), which solves the axisymmetric Khokhlov–Zabolotskaya–Kuznetsov (KZK) equation to compute tissue temperature changes via bioheat transfer modeling. The optimized FUS pulse parameter was used for the simulation, assuming an initial organ temperature of 36.5°C. We used the following parameters: ultrasound frequency of 600 kHz, sound speed of 1,590 m/s, and medium density of 1,055 kg/m3. Bioheat transfer parameters included specific heat capacity of 1,590 J/kg/K, thermal conductivity of 1,055 W/m/K, and blood perfusion rate of 0.011 kg/m3/s (33). The maximum simulated temperature reached 36.5003°C at approximately 200 s after the onset of sonication. Together with the temperature measurements of the BSA-media bath, these results suggest that thermal effects can be excluded from the observed outcomes.
BLI data analysis. All image processing was performed using ImageJ software (NIH). To enhance the signal-to-noise ratio, acquired images were resized to 50% of their original resolution using bilinear interpolation. For consistent visualization of BLI intensities across animals and time points, 8-bit green channel images were uniformly rescaled to a fixed intensity range (2-46), determined based on the overall intensity range across all mice and BLI sessions. ROIs were defined to encompass the surgical implantation site and subsequently thresholded using the top 10% of intensity values. A pseudo-color ‘jet’ lookup table (LUT) was applied to visualize luciferase activity originating from the tumor.
Hematologic evaluation of liver/kidney function. Blood samples were collected from the PTX-treated groups at the end of the 2-week intervention during the sacrifice procedure (Us+/PTX+, n=10; Us−/PTX+, n=9). After a 30 min resting period, serum was isolated using serum-gel separation tubes (Serum-Gel Push Cap Micro Tube, 1.1 ml, Sarstedt Inc.) and centrifuged at 3,200 × g for 10 min. Liver function was assessed by measuring aspartate aminotransferase (AST) and alanine aminotransferase (ALT) levels, while kidney function was evaluated using blood urea nitrogen (BUN).
FUS effects on normal phenotype mice and histological analysis. The effect of FUS in the absence of a tumor and PTX was evaluated in a separate group of normal phenotype mice (C57BL6, n=10) without OVCAR3 implantation. After identifying the location of the ovary unilaterally using ultrasound imaging, FUS was applied unilaterally using the same procedure as in the tumor-bearing mice. Ovaries were harvested at various time points (acute, n=4; 1 week, n=3; 1 month, n=3) post-sonication. In addition to tissues harvested from normal phenotype mice, ovarian tumor mass, liver, and kidney from tumor-bearing animals underwent H&E staining to detect potential hemorrhage (indicative of compromised vascular integrity).
Results
In vitro assessment of FUS parameter-dependent fPTX uptake and OVCAR3 death. The temperature of the bath was 36.6±0.1°C before sonication and 36.5±0.1°C after sonication (Wilcoxon signed-rank test, one-tailed, n=6, z=−1.35, p>0.05), indicating that sonication did not cause any thermal effects. The intracellular fPTX concentration outside the acoustic focus (labeled ‘Outside’ in Figure 2A) ranged from 0.92±0.13 μg/ml (50% DC and 50 ms PD condition) to 0.94±0.12 μg/ml (50% DC and 75 ms PD condition). A one-way ANOVA was performed to compare the effect of sonication parameters on cellular fPTX concentration, revealing absence of differences across the groups [F(6, 41)=0.05, p=0.99]. In contrast, fPTX uptake was significantly higher at the acoustic focus (labeled ‘Focus’ in Figure 2A), with the 70% DC and 100 ms PD condition yielding the highest fPTX concentration of 1.52±0.07 μg/ml among the tested parameters (post hoc pairwise comparisons with other sonication conditions via t-test, all p<0.05), more than a 45% increase compared to the non-focal regions.
Results from in vitro OVCAR3 tumor model. (A) Intracellular fluorescence-labeled paclitaxel (fPTX) concentration measured using different focused ultrasound (FUS) parameters at the acoustic focus (green bars) and outside the focus (red bars). Brackets indicate statistically significant differences (p<0.05); the bracket marked “(!)” denotes that the 70% DC, 100 ms PD condition yielded p<0.01 compared to all other conditions (for brevity). (B) OVCAR3 cell death rate (%) under the same FUS conditions, measured at the focus (green bars) and outside the focus (red bars). PD: Pulse duration; DC: duty cycle; Error bar: standard error.
Cells from the inserts that were cultured in PTX-containing media after sonication (n=3) were completely detached from the membrane (presumably due to cell death), preventing further analysis. Therefore, the results from the inserts that were replenished with fresh, non-PTX media (n=5) are reported herein. The percentage of OVCAR3 cell death (24 h post-sonication; Figure 2B) followed a trend similar to that of intracellular PTX concentration. Outside the acoustic focus, the cell death rate ranged from 2.0±0.8% (70% DC, 75 ms PD) to 2.1±0.6% (70% DC, 100 ms PD), with no significant differences across sonication conditions [Kruskal-Wallis test, H(5, 24)=0.71, p=0.98]. In contrast, significantly higher cell death was observed at the sonicated region, with the 70% DC and 100 ms PD condition producing the greatest effect (14.1±3.1%). This represented approximately a 7-fold increase compared to the value outside the focus (Wilcoxon signed-rank test, one-tailed, n=5, z=−2.02, p<0.05).
We also found that the concentration of fPTX dialyzed to the ED cassette located at the sonication focus (0.27±0.04 μg/ml, n=6) almost doubled (>96%) compared to the value obtained from the cassette located outside of focus (0.14±0.01 μg/ml, paired t-test, t=13.3, p<0.001). These findings are consistent with our previous work demonstrating ultrasound-mediated unbinding of phenytoin from albumin using pulsed ultrasound (31), where a 70% duty cycle yielded greater unbinding efficiency than a 50% duty cycle, with all effects being non-thermal in nature.
Assessment of flow-related effects of FUS via LSCI. LSCI that examined cutaneous vascular flow during the application of FUS showed that the average flow flux at the sonication target (Figure 3A and B, dotted circle) remained unchanged before (885.1±32.6, a.u.) and during FUS application (881.0±44.2, a.u.; Figure 3C). These results indicate that low-intensity FUS is unlikely to introduce flow-related confounders, such as ultrasound-induced convective acoustic streaming or vasodilation, into the study.
An exemplar laser speckle contrast imaging (LSCI) result. Relative flow flux image, in arbitrary unit (a.u.) (A) before and (B) during the application of focused ultrasound (FUS) to mouse abdomen. Dotted circles indicate the site of acoustic focus. Bar=5 mm. (C) The time course of flow flux signal intensity before and during sonication (10 s-sampling).
Assessment of intratumoral PTX uptake and serum PTX level resulting from a single FUS session. In the absence of sonication, intratumoral PTX concentration measured 0.27±0.06 μg/ml. FUS application significantly increased this value by 96.7%, reaching 0.53±0.05 μg/ml (paired t-test, d.f.=5, t=3.2, p=0.005, Figure 4A). In contrast, the serum concentration of fPTX remained equivalent between groups (US−:US+=0.13±0.02: 0.11±0.03 μg/ml, mean±standard error; two-tailed t-test, d.f.=5, p=0.59, Figure 4B). While a small fraction of albumin-bound PTX may be transported into the tumor (e.g., in the form of nab-PTX), distinguishing between the intratumoral distribution of bound versus unbound PTX is extremely challenging, particularly following a single, short-duration FUS session (30 min). Thus, assuming an equivalent rate of tumor uptake for albumin-bound portion of PTX in both FUS+ and FUS− conditions, any observed differences in total intratumoral PTX concentration can be primarily attributed to the fraction of PTX that is unbound by FUS application.
Evaluation of the intratumoral and serum fluorescence-labeled paclitaxel (fPTX) concentration impacted by focused ultrasound (FUS) application. (A) Intratumoral fPTX concentration increased notably in the FUS+ condition while (B) serum fPTX concentrations are equivalent between the control non-sonication condition (FUS−, n=6 each). Error bar: Standard error; n.s. not significant.
Enhancing the antitumoral effects of PTX by FUS - body weight and behavior. Animal body weight was monitored from the time of OVCAR3 implantation (week 0) until sacrifice (week 5), as shown in Figure 5. Following a slight weight loss immediately after implantation, all animals either gained or maintained their weight throughout the study. No significant differences in body weight were observed between groups (two-tailed t-test, all p>0.3), although a mild weight reduction was noted in the PTX+ groups, as anticipated. All animals, including the normal phenotype mice, displayed normal behavior and body condition throughout the experimental period, except for one mouse in the Us−/PTX− group, which developed a skin rash at week 4, post-cell inoculation. Veterinary evaluation did not identify a clear cause while C. bovis testing (for skin infection) returned negative. The rash began resolving immediately and did not affect the animal’s general condition (activity level and weight).
Body weight changes. Averaged body weight across all four conditions across five weeks after the OVCAR3 implantation. Error bar: Standard error.
Analysis of BLI data. Representative composite BLI images from the four experimental groups are shown in Figure 6. Tumor localization at week 3 post-OVCAR3 implantation is clearly visible. However, we found that most animals developed large BL-insensitive masses at the tumor site, resulting in poorly defined BLI tumor margins, sometime forming a ‘ring’-like bioluminescence. Subsequent histological analysis and tumor harvesting revealed the presence of fat pad development surrounding the tumors (also confirmed by microscopy), in many cases much greater than the tumor itself. In several animals, the adipose tissue reached substantial sizes, with thickness exceeding 6 mm.
Representative bioluminescence imaging (BLI) of tumor progression across the two-week treatment period. Images show tumor localization and signal intensity at baseline (Treatment Week 0) and one and two weeks following the start of treatment. Pseudo-color scale ranges from 2 to 46. Experimental groups include: paclitaxel only (Us−/PTX+), focused ultrasound only (Us+/PTX−), combined FUS and PTX (Us+/PTX+), and untreated control (Us−/PTX−). Scale bar=10 mm.
The number of bioluminescence (BL)-active pixels measured at pre-treatment (‘Tx Wk 0’), 1 week post-treatment (‘Tx Wk 1’), and 2 weeks post-treatment (‘Tx Wk 2’; Figure 7A) did not differ significantly across groups or time points (pairwise comparisons, two-tailed t-test, all p>0.05). Similarly, the average BL intensity within active pixels remained unchanged across groups and time points (Figure 7B). We note that both measures exhibited relatively high inter-subject variability. To account for the individual variabilities in tumor BLI, the values obtained at treatment weeks 1 and 2 are expressed as percentage changes relative to the corresponding pre-treatment baseline values. After one week of treatment, the Us+/PTX+ group showed a significant reduction in the number of BL-active pixels compared to the Us−/PTX− group (Figure 7C, marked as ‘*’, t-test, t>2.67, p<0.01). By week 2, the Us+/PTX+ group exhibited a significantly greater reduction in BL-active pixel count compared to all other groups (marked as ‘**’ compared to Us+/PTX− and marked as ‘***’ compared to the Us−/PTX+ at week 2 of the treatment) while a moderate reduction was observed in the Us−/PTX+ group at week 1. Both PTX− groups showed higher tumor BL signal intensity relative to the PTX+ groups. Although a slight (~5%) reduction in average BL intensity was observed across all groups at week 1 (Figure 7D), the Us+/PTX+ group showed the largest decrease (~18%) by week 2, although this change did not reach statistical significance.
BLI analysis across the treatment conditions. (A) Averaged BL-active pixel counts (in Us+/PTX+ groups, n=10; otherwise, n=9). (B) The mean BL intensities across pre-treatment (‘Tx Wk 0’), 1-week, and 2-week post-treatments (‘Tx Wk 1’ and ‘Tx Wk 2’, respectively). The percentage of (C) BL-active pixel counts, and (D) the mean BL intensities relative to the corresponding pre-treatment baseline values. Error bars: standard error. *p<0.05 compared to the Us−/PTX−, **p<0.01 compared to the Us+/PTX−, and ***p<0.02 compared to the Us−/PTX+ group.
Tumor weight and hematologic evaluation of liver/kidney function. Due to the presence of fat pad surrounding the tumor, tumor weight was used as an additional endpoint measure to evaluate treatment efficacy across groups. The weight of the ovary contralateral to the xenograft averaged 3.95±1.35 mg (n=10). In the control group that received neither FUS nor PTX (Us−/PTX−), the harvested tumor weight averaged 38.7±23.1 mg (n=9). Slightly lower tumor weights were observed in the FUS-only (Us+/PTX−) and PTX-only (Us−/PTX+) groups, but these reductions were not statistically significant (Us+/PTX−: 31.2±27.3 mg, p=0.27; Us−/PTX+: 30.8±21.5 mg, p=0.23; two-tailed t-test, n=9 per group). In contrast, the Us+/PTX+ group exhibited a significantly lower tumor weight, averaging 15.6±7.0 mg (n=10), representing approximately a 50% reduction compared to the other conditions (one-tailed t-test, all t<1.7, p<0.05, Figure 8A).
Tumor weight and outcomes of functional blood tests from paclitaxel (PTX)+ groups. (A) Tumor weight across different experimental groups. Individual data points are overlaid on the bars. (B) Plasma levels of alanine aminotransferase (ALT), aspartate aminotransferase (AST), and blood urea nitrogen (BUN). Brackets indicate statistically significant differences (p<0.05, one-tailed t-test). Gray bars represent the normal physiological ranges. Error bars: Standard error.
Group-averaged values for AST, ALT, and BUN among the PTX+ groups were within the normal physiological range (Figure 8B), indicating that FUS did not exacerbate systemic toxicity associated with PTX treatment. Notably, the Us+/PTX+ group showed lower – but still physiologically normal – AST and BUN levels compared to the Us−/PTX+ group (p=0.01, one-tailed t-test, n=10 per group). A few animals in the Us−/PTX+ group exhibited elevated ALT and AST levels exceeding the normal range, suggesting mild liver impairment consistent with known PTX side effects. However, these elevations were not indicative of fatal or severe toxicity.
Histological assessment. Normal histological features were observed in liver sections across all experimental groups; however, animals treated with PTX exhibited renal tubular injuries characterized by microhemorrhages (indicated by arrows in Figure 9), a form of renal damage commonly associated with PTX administration. H&E histology of ovarian tumors from xenografted animals, including ovaries from mice without OVCAR3 implantation, is shown in Figure 10. In the OVCAR3-xenografted animals, particularly in the Us−/PTX− group, multiple hallmarks of active tumor formation were evident, including: 1) densely cellular, poorly organized areas with minimal stroma, characteristic of epithelial tumor nodules; 2) replacement of normal follicular structures (intact follicles with oocytes) by abnormal neoplastic tissue, indicating tumor overgrowth; 3) hyperchromatic, crowded, and variably sized nuclei within the tumor mass, consistent with high-grade malignancy; and 4) pale and fibrous regions representing reactive stroma or compressed ovarian tissue, often seen at the invasive tumor front as desmoplastic stroma or tissue compression. No evidence of metastatic activity was observed in any group.
Representative H&E histology from experimental groups. (Top row) Liver sections showing normal histological features across groups. (Bottom row) Kidney sections showing normal histology in paclitaxel (PTX)− groups, whereas PTX administration resulted in renal damage characterized by microhemorrhages (indicated by arrows). Scale bar=100 μm.
Representative H&E histology of ovaries from unaffected mice and experimental groups. Solid cellular masses are indicated by arrows, and loss of normal ovarian architecture is marked by arrowheads. Scale bars=200 μm.
Discussion
The present study demonstrates that the application of low-intensity, non-thermal FUS to a mouse xenograft model of OC tumor enhanced the efficacy of PTX chemotherapy by locally unbinding PTX from plasma proteins. This approach leverages the biophysical capacity of ultrasound to transiently disrupt drug-protein interactions without altering systemic drug concentration, thereby enabling localized drug delivery. Our findings establish that this method significantly increases intratumoral drug uptake and amplifies therapeutic efficacy while avoiding off-target toxicity.
The optimization of acoustic parameters for efficient PTX unbinding, achieved through in vitro OVCAR3 cell culture construct, was essential in choosing the pulsing scheme, 100 ms PD given at 7 Hz PRF (70% duty cycle) at ISPPA of 4.28 W/cm2, which yielded the highest level of PTX uptake to OVCAR3 cells. One of the key observations was that just 30 min of sonication using this parameter set resulted in a more than 45% increase in intracellular PTX uptake, which also led to a dramatic seven-fold increase in cell death compared to the unsonicated controls. The increased level of unbound level of PTX was also confirmed by ED, which almost doubled (>96%) the concentration compared to the value obtained from the ED cassette located outside of focus. The temperature of the media bath did not change, indicating that sonication did not cause any thermal effects. These findings are consistent with our previous work, where 70% DC yielded greater unbinding efficiency compared to 50% DC, with all effects being non-thermal in nature (31). Prior to probing the efficacy of the technique in the murine cancer model, the potential impact of sonication on vascular flow was verified using LSCI, which demonstrated that FUS did not induce flow-related effects such as acoustic streaming or vasodilation. This reinforced the mechanism behind the technique whereby the observed PTX delivery enhancement is due to molecular unbinding and not secondary circulatory phenomena.
In vivo assessment of intratumoral and serum concentrations of PTX following a single session of FUS revealed that FUS nearly doubled PTX uptake in the tumor, while serum concentrations remained unchanged. These findings support the hypothesis that FUS facilitates transient and localized unbinding of PTX from plasma proteins, thereby enhancing tumor-specific drug delivery. This observation aligns with the non-covalent PPB nature of PTX to albumin (6), wherein unbinding occurs only at the focal site and reverts rapidly outside. However, the duration of the unbound drug state following FUS remains unclear. While we anticipate the duration of unbound state to be short (in the order of min or less), longitudinal monitoring of total PTX concentration (i.e., bound + unbound) after FUS will be essential to characterize the binding dynamics. To address this, total PTX levels can be measured in a separate cohort of animals (through non-survival protocol) using an acetonitrile-mediated protein precipitation method. In parallel, unbound PTX can be isolated by ultrafiltration with a 30 kDa molecular weight cutoff membrane and quantified by liquid chromatography–mass spectrometry (LC-MS/MS).
We unexpectedly found that BLI signals from the luciferase-labeled tumor were often obscured by the presence of a thick fat pad surrounding the tumor. This introduces important technical considerations when interpreting longitudinal in vivo imaging data. Although BLI is a widely used and highly sensitive method for noninvasive tumor burden assessment in small animal models, its signal intensity can be significantly attenuated by overlying tissues, especially those with high lipid content (43). Despite this limitation, the area reduction in BL-active tumor from the Us+/PTX+ group remained lower than that of other groups, indicating a consistent and robust reduction in viable tumor cell population, even after accounting for photon attenuation. The scaled number of BL-sensitive pixels comparing pre-treatment and treatment progression (Figure 7C), along with tumor weight assessments at endpoint (Figure 8A), collectively indicate that the combination of FUS and PTX (Us+/PTX+) significantly reduces tumor burden compared to control and monotherapy groups. These findings further support the conclusion that FUS-mediated enhancement of PTX delivery results in a true reduction in tumor burden.
In normal phenotype mice, FUS did not alter H&E-based histology of the sonicated ovarian tissue. In xenografted mice, FUS did not induce additional damage to liver or kidney function nor did it promote metastasis. OC metastasis was not observed in any animal during the 5-week experimental period, consistent with previous findings that cervical cancer xenografts, such as SiHa cells, also typically do not metastasize within this timeframe (44). Histological and hematological analyses further confirmed the safety of FUS treatment despite microhemorrhages found in the kidney cortex in PTX+ groups. Liver and kidney function tests (AST, ALT, BUN) remained within normal limits in PTX-treated animals. These results indicate that FUS does not elevate the systemic toxicity, which is consistent with the observation that serum PTX concentrations were unaffected by sonication. These safety results are critical for validating the translational potential of this technique, particularly in the context of repeated clinical applications.
Importantly, our study challenges the traditional view of PPB as an unmodifiable barrier to drug delivery. While previous approaches focused on modifying the drug or carrier system, our findings suggest that external physical energy, such as ultrasound, can be used to transiently and locally alter drug-protein interactions. This represents a paradigm shift in drug delivery, introducing the possibility of spatially controlled, reversible, and noninvasive modulation of drug bioavailability. Notably, the observed enhancement in drug delivery and anti-tumor efficacy was achieved with a single 30-min FUS session, highlighting the method’s practicality and potential for clinical translation. Furthermore, the optimized acoustic parameters used in this study fall well within established safety limits for diagnostic ultrasound, supporting the feasibility of rapid integration into clinical workflows.
Despite the promising outcomes demonstrated in this study, several limitations must be acknowledged, particularly regarding the application of this FUS-mediated drug delivery strategy to late-stage, diffusely disseminated OC. In advanced disease stage, OC often manifests with widespread peritoneal metastases, ascites, and microscopic spreads on visceral and serosal surfaces (45, 46), all of which pose substantial challenges for focal therapies like FUS. The current approach relies on spatial targeting of tumor masses, which is feasible in well-defined localized lesions but less practical when the disease is diffusely spread across the peritoneal cavity. Moreover, anatomical variability, organ motion (e.g., bowel peristalsis), and sound absorption/interference from gas-filled structures (e.g., intestine) can hinder effective ultrasound propagation and focusing in deep or irregularly located metastatic sites. Moreover, the presence of ascitic fluid may alter acoustic impedance and degrade the precision and resolution of FUS delivery. These limitations underscore the need for further technological advancement, such as multi-focal or volumetric sonication strategies, to broaden the clinical applicability of this technique in treating advanced-stage OC.
In summary, this study introduces a novel application of low-intensity, non-thermal FUS to modulate PPB and enhance the local delivery of unmodified PTX in a mouse model of OC, achieving the spatial selectivity previously unattainable with conventional drug formulations. We demonstrate that this approach significantly increases tumor drug uptake and therapeutic efficacy without elevating systemic toxicity. These findings provide a compelling rationale for further investigation and clinical development of FUS-mediated PPB disruption as an adjunct to conventional chemotherapy. They also align with the overarching goal of precision oncology, emphasizing spatial and temporal control over drug action. Notably, the rapid reversibility of the unbinding effect – evidenced by unchanged systemic PTX levels – underscores the controllability and safety of this method, a critical advantage for drugs with narrow therapeutic indices or dose-limiting off-target effects. This strategy holds promise for integration with real-time imaging platforms to enable adaptive, image-guided therapy delivery. Beyond OC, FUS-enhanced chemotherapy may be broadly applicable to other solid tumors where drug efficacy is constrained by protein binding. Future work should elucidate the molecular mechanisms underlying FUS-induced unbinding and evaluate its compatibility with other drug classes and delivery systems. Taken together, our findings support the translational potential of this technology and its capacity to redefine localized chemotherapeutic strategies.
Acknowledgements
The Authors thank Ms. Minjee Seo for her assistance in numerical simulation and Mrs. Kathleen Hasselblatt for her help in preparing the OVCAR3-luc cell line. Ellen Buckley Jordan at Comparative Pathology Laboratory of Massachusetts Institute of Technology is gratefully acknowledged for her help in analyzing blood samples.
Footnotes
Authors’ Contributions
Yoo S-S., Zhang Y, and Elias K conceived and designed the experiments. Yoo S-S., Banish K, Fahmi A, Glasner C, Yoon K, Zhang Y, and Elias K performed the experiment and collected the data. Yoo S-S., Banish K, Fahmi A, Glasner C, and Yoon K analyzed the data. Yoo S-S., Banish K, Fahmi A, Glasner C, Yoon K, and Betharia S drafted and edited the paper. Elias K and Horowitz N reviewed and edited the paper.
Conflicts of Interest
The Authors have no conflicts of interest to declare in relation to this study.
Funding
The project was funded by the grant from Focused Ultrasound Surgery Foundation.
Artificial Intelligence (AI) Disclosure
During the preparation of this manuscript, a large language model (ChatGPT, version 4.1, San Francisco, CA, USA) was used solely for language editing and stylistic improvements in select paragraphs. No sections involving the generation, analysis, or interpretation of research data were produced by generative AI. All scientific content was created and verified by the authors. Furthermore, no figures or visual data were generated or modified using generative AI or machine learning–based image enhancement tools.
- Received July 23, 2025.
- Revision received July 31, 2025.
- Accepted August 1, 2025.
- Copyright © 2025 The Author(s). Published by the International Institute of Anticancer Research.
This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited.
References
In this issue
Jump to section
Related Articles
Cited By...
- No citing articles found.