Ultrasound Increases Nanoparticle Delivery by Reducing Intratumoral Pressure and Increasing Transport in Epithelial and Epithelial–Mesenchymal Transition Tumors

The delivery of drugs to solid tumors remains a significant challenge, particularly for stromal tumors characterized by an epithelial–mesenchymal transition (EMT) phenotype. EMT is a developmental process linked to increased cell proliferation and motility that is important for morphogenic processes that require migratory mesenchymal cells to be recruited at distant sites for epithelial tissue differentiation (1). Tumors have co-opted this transdifferentiation program for enhanced cell migration. While epithelial tumors are highly vascularized with tight cell–cell junctions, EMT tumor cells adopt an irregular, elongated shape with weak cell–cell adhesion and a reduction in E-cadherin, cytokeratin 8/18 (CK8/18), and cytokeratin 19 (CK19; ref. 2). In this study, by controlling the confluency and passage number, epithelial and EMT phenotypes were both produced from identical transgenic Met-1 cells (3). Given the low accumulation and poor penetration of nanotherapeutics in EMT tumors, we set out to test whether nanodelivery in epithelial and EMT tumor phenotypes improves with low levels of therapeutic ultrasound exposure. Using positron emission tomography (PET), we have created imaging-based methods to optimize and quantify drug delivery vehicle distribution (4). Classical optical techniques and analyses were used to validate imaging-based results (5).

Nanoparticle-encapsulated chemotherapeutics exploit the enhanced vascular permeability and retention effect of tumors. Liposomes, with a 100-nm diameter, increase the therapeutic index of chemotherapeutic drugs by reducing systemic toxicities and enhancing the drug concentration in target tissues (6). Albumin, with a smaller diameter than liposomes (∼66 kDa, dimensions of 4 × 4 × 14 nm), has been used in a similar fashion to carry therapeutics and study permeability (4, 5). However, as compared with small-molecule therapeutics, delivery using liposomal or albumin-based carriers is limited by high interstitial pressure and heterogeneous vascular permeability (7, 8).

Both hyperthermia and local ablation have been associated with enhanced nanodelivery to the treated region. The concentration of doxorubicin liposomes has been increased several fold to tissues surrounding tumors treated with radiofrequency ablation (9–11). Furthermore, mild hyperthermia has been shown to increase the effective pore size and enhance liposome extravasation (12). Here, we develop analytic techniques based on the Kedem–Katchalsky equation for the net flux of solute across a blood vessel wall to directly assess nanoparticle transport to the endothelium with noninvasive imaging methods (4, 5, 13, 14).

Ultrasound-induced hyperthermia is noninvasive and easily directed to the region of interest. In addition, ultrasound has both mechanical and thermal effects (15, 16) and has been associated with enhancing drug delivery (17), increasing blood supply, activating immune cell trafficking (18–20), inducing mast cell degranulation (21), and prompting macrophage release of fibroblast mitogenic factor for fibroblast proliferation (22). Combining hyperthermia with chemotherapeutic drugs enhances the efficacy of macromolecular and liposomal formulations (23, 24) by increasing tumor blood flow (25) and sensitizing tumor cells to radiation (26). However, there are contradictory reports of the efficacy of ultrasonic enhancement of nanoparticle extravasation (27–29).

Insonation-induced tissue temperature increases depend on the dynamic balance of ultrasound absorption and heat conduction and diffusion in the tissue (30). The resulting thermal effect on tissue increases with increasing temperature and exposure time. Thermal dose is often estimated in terms of the cumulative equivalent minutes at 43°C (CEM43), where CEM43 is defined as tR^(43−T), with t being the time of treatment, T the average temperature during treatment, and R a constant that equals 0.25 for temperatures between 37°C and 43°C (31, 32) and 0.5 above 43°C. Heat damage to tumor tissue and vasculature varies amongst species with human tumors being the most heat resistant. In general, treatments resulting in a CEM43 below 0.5 are considered to be reversible and nonlethal (33).

A primary goal of this work is to determine the mechanism and optimal treatment protocol for ultrasound-enhanced nanodelivery to epithelial and EMT tumors, including peak-negative pressure (PNP) and thermal dose. In addition, we seek to establish the use of imaging-based pharmacokinetic techniques to estimate vascular permeability and nanoparticle accumulation. We have developed chemistries that can be applied to radiolabel lipid, polymer, and albumin-based particles for preclinical and clinical studies (34, 35). Such techniques show promise in facilitating head-to-head comparisons of nanodelivery vehicles and in evaluating techniques (such as ultrasound) to enhance delivery.

All protocols were approved by the University of California, Davis Animal Care and Use Committee. Met-1^fvb-2 cells were cultured as previously described (3). Low passage (P9), highly confluent (120% to 150%) cells produced epithelial tumors. High passage (P16–25), 80% to 85% confluent cells generated EMT tumors. A total of 1 × 10⁵ cells in 0.02 mL were injected into the 4th mammary inguinal fat pads of donor animals. Fresh tumor samples were transplanted into the inguinal mammary fat pad from cell-injected donor animals as previously described (3).

Liposomes were composed of a total of 10 or 15 mg HSPC:cholesterol:DSPE-PEG2K-OMe:BAT-PEG-lipid = 55.5:39:5.0:0.5 mol. Preparation and radiolabeling followed previously described protocols (34). The radiochemical purity of liposomes on TLC was greater than 95%. Isoflurane-anesthetized mice were injected with 1 mg/150 μL of ⁶⁴Cu-liposomes. PET acquisitions were obtained as previously described (4, 23, 34) with a dedicated small animal PET scanner (Focus120, Siemens Medical Solutions, Inc.). In vivo PET scans were obtained for 30 minutes at 0, 3, 6, 18, 28, and 48 hours postinjection. The biodistribution of radioactivity in collected organs was measured in a gamma counter (Perkin-Elmer Life Sciences).

One tumor per animal was insonified before liposome injection. Tumor temperature was either maintained at 37°C ± 0.5°C (no thermal influence) or increased to 42°C for 2, 7, or 18 minutes using feedback-controlled ultrasound as described in the work of Kruse and colleagues (36). The ultrasound pulses consisted of 100-cycle bursts at 1.5 MHz center frequency and 1.1 or 2.4 MPa PNP, with variable pulse repetition frequency (PRF) ranging from 100 Hz up to 5 kHz. Studies were designed on the basis of CEM43 and PNP goals.

Excised tumors were fixed overnight in 10% formalin and dehydrated in 70% ethanol. Four-micron slices were stained with hematoxylin and eosin (H&E) or deparaffinized for immunohistochemistry with the following antibodies: CD31 (Santa Cruz, sc-1506, 1:800), E-cadherin (Abcam, ab53033, 1:1200), N-cadherin (Epitomics, 2019-1, 1:800), CK8/18 (Fitzgerald/RDI, 20R-CP004, 1:8000), CK19 (Epitomics, 3863-1, 1:1000), and vimentin (Epitomics, 2707-1, 1:1000).

Pressure was measured with the wick-in needle method using a 25G needle with a side hole as previously described (37).

The 30 minute clearance of fluorescently labeled albumin was measured as described in detail elsewhere (5, 38). The tissue was dried over 3 days in an oven to obtain a dry weight measurement.

A total of 123 animals with bilateral tumors were studied. Enhanced accumulation was determined by the equation:

Statistical analysis was conducted in collaboration with the Division of Biostatistics, UC Davis School of Medicine (Davis, CA). The maximum accumulation over time and time–activity data acquired over multiple time points was analyzed using a linear mixed model including repeated measures. When analyzing the effect of ultrasound on epithelial and EMT tumors, multiple comparisons were conducted using the Turkey honest significant difference test with a repeated measures mixed model. Least squares means were calculated to test for differences between paired, treated, and control animals after a global F test. A minimum of 7 animals per group were studied with some groups including more than 7 since tumor characterization was not carried out until histology was evaluated.

To differentiate tumor phenotypes, we first characterized epithelial and EMT tumors based on H&E staining for gross anatomy and 5 immunohistochemical markers (E-cadherin, N-cadherin, CK8/18, CK19, and vimentin). Epithelial tumors showed a well-defined margin (Fig. 1A) that displaced the host tissue during proliferation and contained large diameter vascular sinuses and vessels (Fig. 1B). EMT Met-1 tumors were highly invasive (Fig. 1C) with a reduced vascular volume (43.7 ± 15.0 μL/g tissue), as compared with epithelial tumors (107.4 ± 111.6 μL/g tissue, P < 0.02) and the quadriceps muscle (24.5 ± 9.2 μL/g dry tissue, P < 5.0 × 10⁻⁵; Fig. 1D and E). The positivity (N positive/N total) of the immunohistochemical stains was evaluated to determine protein prevalence (Fig. 2A–M). Epithelial tumors (Fig. 2A–F) were characterized by the presence of high levels of E-cadherin, CK8/18, and CK19 expression (0.63 ± 0.01, 0.65 ± 0.06, and 0.32 ± 0.09, respectively) with low vimentin expression levels (0.28 ± 0.07). N-cadherin was highly expressed in both tumor phenotypes (0.68 ± 0.02 and 0.64 ± 0.02) but was bound to the plasma membrane in epithelial tumors versus cytoplasmic in EMT tumors. Vimentin was upregulated in EMT tumors (0.58 ± 0.09), while E-cadherin, CK8/18, and CK19 levels fell below the 20% detection threshold (Fig. 2G–L).

The accumulation of liposomal particles differed between epithelial (N = 39) and EMT (N = 58) phenotypes. In the absence of ultrasound, liposomal accumulation was greater in epithelial (8.2% ± 3.6% ID/g) than EMT tumors (5.5% ± 2.0% ID/g, P < 3.0 × 10⁻⁶; Fig. 2N). The time required for maximum accumulation was longer for epithelial tumors (30.3 ± 9.0 hours) than EMT tumors (23.5 ± 9.0 hours, P < 0.0005; Fig. 2N).

Baseline levels of liposomal accumulation were first assessed in both epithelial and EMT tumor phenotypes. Ultrasound was then applied to induce mild hyperthermia and mechanical stress, with a goal of increasing the transport through the endothelium and thus increasing accumulation (Fig. 3A and B). Over the first 18 hours after intravenous injection, radiolabeled liposomes were highly concentrated within the blood pool, evident in the images within the heart chambers and major blood vessels (Fig. 3C). Within 18 hours, liposome accumulation in tumors was observed and accumulation was greatest for the insonified tumor (Fig. 3C, red arrow). Through region of interest analysis and biodistribution data (Supplementary Fig. S1), the increased accumulation of particles within the insonified tumor was quantified (Fig. 3D), validating the enhanced accumulation in the insonified tumors. Across EMT and epithelial tumors (N = 97), therapeutic ultrasound treatment resulted in a 1.5-fold increase in liposomal particle accumulation over control tumors (Fig. 3C and D; P < 0.001). For these long circulating particles, tumor accumulation peaked between 18 and 28 hours (Fig. 3D) with particle clearance occurring through liver metabolism and urinary and intestinal excretion. Both the maximum accumulation over time and the accumulation at the final 48-hour time point increased in insonified (as compared with control) tumors. Accumulation was greatest when the particles were injected within 1 hour of ultrasound treatment (8.5% ± 2.8% ID/g) as compared with injection times exceeding 1 hour after insonation (6.9% ± 3.1% ID/g, P < 0.01, N = 11 for both <1 and >1 hour; Fig. 3E).

The effect of ultrasound differed with both thermal and mechanical ultrasound dose. Increasing the thermal ultrasound dose (CEM43) from less than 1 to 4.5 enhanced the accumulation of particles in EMT tumors (5.3% ± 1.8%, 5.3% ± 2.6%, 6.1% ± 1.8%, 8.0% ± 4.1%, 9.2% ± 3.0% ID/g tumor for a CEM43 of 0, <1, 1.4, 1.6, and 4.5, respectively). Accumulation in epithelial tumors peaked at approximately 15% ID/cc with CEM43 of 1.4 and a PNP of 1.1 MPa (Figs. 4A and B, 5A). When ultrasound with a PNP of 2.4 MPa was applied with the tumor temperature remaining within ±0.5°C of 37°C, accumulation was not increased in the group tested (EMT tumors; 5.6% ± 1.7% vs. 5.3% ± 2.6% ID/g; Fig. 5B).

The rate of particle extravasation was then evaluated using an image-driven pharmacokinetic model with transport between the vasculature and tumor estimated over the first 28 hours after insonation and injection. Apparent permeability (P_AP) is a measure of particle transport, combining convective and diffusive transport through modified intercellular junctions and specialized transendothelial pathways such as transendothelial gaps and vesiculo-vacuolar organelles (13). Averaged across all insonation parameters, for both epithelial and EMT tumors, transport was increased by approximately 73% with insonation. For epithelial tumors, P_AP increased from 4.1 × 10⁻⁸ ± 2.8 × 10⁻⁸ to 6.7 × 10⁻⁸ ± 4.6 × 10⁻⁸ cm/s (Fig. 5C, P < 0.001). P_AP in EMT tumors was significantly lower than in epithelial tumors prior to insonation (2.7 × 10⁻⁸ ± 1.4 × 10⁻⁸ cm/s, P < 0.005); however, insonation increased P_AP to 5.1 × 10⁻⁸ ± 3.6 × 10⁻⁸ cm/s (P < 1.0 × 10⁻⁵). A general trend of increasing P_AP with increasing CEM43 for EMT tumors was observed with R² = 0.93 (Fig. 5D).

The mechanisms for the effect of ultrasound on particle accumulation were also investigated. Ultrasound reduced the intratumoral pressure and increased P_AP, thereby enhancing the vascular clearance of nanoparticles into the tumor interstitium (Fig. 6A and C). Intratumoral pressure in control tumors averaged 13.6 ± 5.4 mm Hg. As the tumor was heated from body temperature (37°C) to 42°C, intratumoral pressure was unchanged (control pressure normalized to 1 vs. 0.95 ± 0.19). When tumor temperature was maintained at 42°C and insonation continued for 5 minutes (CEM43 ∼1.4, PNP 2.4 MPa) in a live animal, intratumoral pressure declined (1 vs. 0.82 ± 0.24, P < 0.02, N = 15). As expected, intratumoral pressure fell after euthanasia (from 1 to 0.84 ± 0.084, P < 0.005; Fig. 6B). Tumor insonation after euthanasia also resulted in an intratumoral pressure reduction; a significant change was not observed during the 37°C to 42°C incline, but intratumoral pressure dropped during a 5 minute insonation at 42°C (from 0.84 ± 0.084 to 0.76 ± 0.11, P < 0.05, N = 7).

For comparison with classical permeability studies, the clearance of fluorescently labeled albumin was evaluated before and after insonation (Fig. 6C). The 30-minute clearance of albumin from the tumor vasculature into the tumor interstitium was faster in insonified (85.5 ± 40.6 μL/g dry tissue) than control tumors (64.9 ± 33.8 μL/g dry tissue, P < 0.001, N = 18) and greater than quadriceps muscle clearance (8.6 ± 3.1 μL/g dry tissue). Intratumoral water content did not change after ultrasound (Fig. 6D, N = 19), but exceeded that obtained from the quadriceps muscle (4.2 ± 0.4 mL/g), collected as an internal control.

The use of ultrasound to enhance the extravasation of drugs and nanoparticles has been widely reported; however, reports of significant enhancements have been accompanied by reports of failed protocols. Alternatively, reports of enhanced accumulation by radiofrequency ablation or by mild hyperthermia have led to consistent, well-characterized protocols. Here, by labeling liposomes and albumin with a PET or fluorescent tracer, individualized protocols for ultrasound treatment of epithelial and EMT tumors have been developed.

Without ultrasound, we found that the accumulation of liposomes was approximately 1.5 times greater in epithelial, as compared with EMT, tumors. When ultrasound was applied within 1 hour of particle injection, accumulation was enhanced in the insonified, as compared with the contralateral, tumor with a concomitant increase in both the rate and volume of accumulation. Across both epithelial and EMT tumors, maximum and mean accumulation in the tumor interstitium were increased in the insonified tumors. The effect of ultrasound differed with both thermal and mechanical ultrasound indices. Increasing the CEM43 from 1.6 to 4.5 increased the accumulation of particles in EMT tumors to a maximum of 9.2% ± 3.0% ID/g for a CEM43 of 4.5, whereas accumulation peaked in epithelial tumors at 14.7% ± 3.0% ID/g. Given the narrow therapeutic window for current chemotherapeutics, this increase in tumor accumulation could significantly improve efficacy.

Enhanced delivery to the EMT phenotype is of particular interest as such tumors are aggressive and frequently difficult to treat (39). Here, a significant enhancement of accumulation was observed in both the EMT and epithelial models, yet the parameters required to maximize the effect differed with the tumor biology. Accumulation increased in both EMT and epithelial tumors with a PNP of 2.4 MPa; however, a greater thermal dose decreased accumulation in epithelial tumors. The mean vascular volume in epithelial tumors was more than double that in the EMT phenotype. We hypothesize that the decreased accumulation in epithelial tumors with a higher thermal dose results from vascular stasis in the large vessels that are prevalent in the epithelial (but not EMT) phenotype (12). On the contrary, accumulation in the EMT phenotype was greatest at the highest thermal dose evaluated here, where small regions of heat-mediated necrosis were evident. For EMT tumors, the proportional increase in transport with increasing CEM43 likely results from decreased interstitial pressure, increased P_AP, and cell damage (40).

Mechanical effects of ultrasound stem from multiple sources, including cavitation, radiation pressure, and expansion and contraction of space between lipid membrane bilayers (41). Potential benefits of leaflet stretching include stimulation of mechanosensitive membrane proteins, rearrangement of cytoskeletal elements, membrane perforation increasing drug uptake, and increased tissue permeability (42). As shown in Fig. 5A, accumulation was enhanced in epithelial tumors at a PNP of 1.1 and 2.4 MPa; however, accumulation was not maximized in EMT tumors until a PNP of 2.4 MPa was applied. Following the transformation of Met-1 cells from an epithelial to an EMT phenotype, we observed downregulation of E-cadherin, CK8/18, and CK19 expression but upregulation of vimentin. E-cadherin expression is important for maintaining cell–cell contacts and tissue organization (43, 44). Vimentin has a number of critical functions within the cell that involve attachment, migration, and cell signaling (45). In addition, vimentin modulates the negative effects of mechanical and thermal stresses (46, 47). Our results suggest that imposing a mechanical stress by ultrasound (in addition to heat) enhances accumulation in the EMT phenotype, presumably due to weakened cell–cell adhesion and increased vimentin expression.

The mechanisms for the effect of ultrasound on P_AP and particle accumulation were also investigated. First, techniques for the estimation of vascular volume, clearance rate, and water content were validated by measurements for the quadriceps muscle (internal control), where values were similar to those previously reported for C57BL/6 mice (5, 38). Ultrasound reduced the intratumoral pressure and increased P_AP in our tumor model without significant alterations to vascular volume and tissue water content. The measured changes in these 2 parameters likely contributed to the enhanced accumulation of particles within the tumor. Interestingly, ultrasound reduced intratumoral pressure even after euthanasia; therefore, the effect cannot be solely attributed to changes in blood flow, protein expression, or macrophage recruitment. We hypothesize that this reduction is likely due to the effect of insonation and hyperthermia on the extracellular matrix and will evaluate this effect further in future studies. Thermal absorption and mechanical agitation can influence the extracellular matrix, particularly collagen integrity (48). Also, collagen content is known to influence extracellular matrix hydraulic conductivity (49).

PET provided a convenient method to assess the accumulation and apparent vascular permeability over time and across the tumor volume; such measurements can be directly compared with classical estimates of vascular clearance. PET-based pharmacokinetic modeling has previously been developed for small-molecule therapeutics, but nuclear medicine–driven techniques to evaluate the pharmacokinetics of nanoparticles are currently unavailable. While small molecules distribute in seconds to minutes, nanoparticles extravasate from the blood pool in hours to tens of hours, and therefore, nanoparticle distribution at the time of injection provides a direct measurement of organ and tumor vascular volume. With PET, the circulating radioactivity can be measured on the basis of cardiac chamber activity. With estimates of circulating particle concentration and tumor vascular volume, PET can be applied to estimate the rate of nanoparticle extravasation for each acquisition.

In our Supplementary information, we detail the relationship between the P_AP parameter derived through the image-guided model with the classical parameters of solute transport coefficient, clearance, and 30-minute albumin clearance (Cl₃₀). P_AP and Cl₃₀ are related as described by SI Eq6,

where V_T is the tumor volume, r is the mean radius of the blood vessels within the tumor, and η is the plasma volume fraction in the tumor. In our study, both the P_AP (as determined by PET) and the clearance (as determined by fluorescent albumin tracers) were increased as a result of ultrasound. Not surprisingly, because of the difference in diameter, the ultrasound-induced increase in P_AP for liposomes was greater than the induced increase in clearance for albumin (∼30% increase in Cl₃₀ for albumin and ∼70% increase in P_AP for liposomes as assessed by PET, averaged across epithelial and EMT tumors). One limitation of these studies is that we were not able to determine whether the net flux of solute into the tumor interstitium was due to increased extravasation or a decrease in the net clearance from the tumor. Regardless of the mechanism, both albumin clearance and PET imaging of liposomes assess net flux of solute between the tumor vasculature and interstitium and showed the effectiveness of ultrasound to enhance accumulation. A further limitation is that we selected one peak treatment temperature to study (42°C) and the optimal protocol is expected to vary with changes in this treatment temperature.

In summary, we applied ultrasound to increase nanoparticle-based drug extravasation and accumulation in both epithelial and EMT tumors. Using both the thermal and mechanical properties of ultrasound, effective treatment times were shorter than times typically used in mild hyperthermia treatments (5–18 minutes compared with durations of ≥1 hours). In addition, we observed a decrease in intratumoral pressure after only 5 minutes as compared with the reported 6 hours required for interstitial fluid pressure reduction with whole body hyperthermia (50). Our study clearly shows that differences in tumor phenotype must be recognized in treatment planning and suggests that imaging techniques, such as those applied here, may provide the information required to optimize treatment for individual characteristics. Such individual differences can be viewed as an opportunity or obstacle for such treatments. For our ultrasound treatment parameters and a target temperature of 42°C, thermal dose must be limited when treating highly vascular epithelial tumors; whereas delivery can be increased in the poorly vascularized EMT phenotype with a higher mechanical and thermal dose.

No potential conflicts of interest were disclosed.

The authors thank the statistical assistance of Drs. Laurel Becket (PhD) and Judy Li (PhD) as well as the graphical reconstruction by Kathleen Montgomery in Fig. 3.

The work reported here was supported by NIHR01CA134659, NIHR01CA103828, NIHR21EB009434, NHLBI Training Program in Basic and Translational Cardiovascular Science (NIHT32HL86350), Focused Ultrasound Surgery Foundation, and the UC Davis Cancer Center Support Grant (P30CA093373-06).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.