Ultrasound-Stimulated Microbubbles Enhance Radiation-induced Cell Killing

Background: There has been recent interest in the use of ultrasound-stimulated microbubbles (USMB) as a localised radiosensitiser, with in vivo studies to date demonstrating impressive results. In vitro results have focussed primarily on Human Umbilical Vein Endothelial Cells (HUVEC), and studies using other cell lines have shown varying results. This study aims to further investigate any increases in radiation-induced cell killing in vitro using HUVECs and two carcinoma cell lines; and to study the effect different radiation energies have on cells treated with ultrasound-stimulated microbubbles and radiation. Methods: HUVEC, Metastatic Follicular Thyroid Carcinoma cells (FTC-238) and Non-Small Cell Lung Carcinoma cells (NCI-H727) were treated using a combination of 1.6% v/v microbubbles (Denity®); ~90 seconds of 2MHz ultrasound with a mechanical index of 0.8; and 0-6Gy of kV or MV x-rays. Cell viability was measured using a cell viability assay 72 hours post-treatment, and normalised survival calculated relative to untreated controls. Statistical signicance was determined using a Three-Way Analysis of Variance (ANOVA) after conrming equal variance existed across all groups. Results: All cells treated with combined USMB and radiation demonstrated decreased normalised survival, with statistically signicant effects observed for the NCI-H727 cells. Statistically signicant differences in the effect of kV and MV radiation were also observed for FTC-238 cells. Conclusions: USMB and Radiation Therapy signicantly reduced the viability of NCI-H727 cells when compared to cells exposed to radiation treatment only. Decreased cell viability was also observed in HUVEC and FTC-238 cells, but this was not signicant. Further studies using increased Microbubble concentrations may be required to achieve statistically signicant results as reported elsewhere. Similarly, further research the differing effect of kV and MV x-rays on survival of FTC-238 cells. inuencing these results, such as variations in applied acoustic elds and differences in the cytotoxicity proles for USM-alone. The differences in cell killing elicited by kV and MV x-rays in FTC-238 cells are likely to be due to other factors, as similar observations were not seen in NCI-H727 or HUVEC cells. Given the overall trends observed here, this study demonstrates that this type of research is feasible using clinical hardware, and that further opportunities exist to understand the specic mechanisms involved in the direct radiosensitisation of tumour cells using USMB.


Background
Microbubbles (MBs) have long been used as contrast agents to improve image quality in ultrasound (US) cardiac imaging [1]. Their size is similar to the average erythrocyte, meaning they remain within the endovascular borders during their travel from the site of injection to the site of interest. This feature has resulted in MBs also being investigated for a variety of focussed therapeutic purposes, as they can be 'burst' using higher US intensities once at the site of interest, creating localised biophysical disruptions in the vicinity [2,3], and if loaded with a treatment agent, delivering a localised therapeutic payload [4].

In vivostudies to date
Consequently, there has been recent interest in using Ultrasound-Stimulated Microbubbles (USMB) as localised radiosensitisers of radiation treatment (RT), with in vivo studies demonstrating some promising results. Czarnota et al. [5] reported a statistically signi cant increase in mean animal survival from 19 days with RT alone to 28 days using USMB and RT together (USMB + RT). PC3 human prostate cancer xenografts in mice were treated twice weekly with USMBs in conjunction with 24Gy in 12 fractions of 160kVp x-rays delivered four times a week for three weeks. Daecher et al. [6] reported a mean animal survival of 11 days with RT alone compared to 35 days using USMB + RT in treating human hepatocellular carcinoma xenografts in nude rats. In this study the xenografts were treated with USMBs, followed by a single dose of 5Gy kV x-rays delivered using 4x 1.25Gy beams. Two animals treated with USMB + RT showed complete tumour control, with tumour reduction or stability observed 50 days post-treatment.
Eisenbrey et al. [7] observed a statistically signi cant growth delay for MDA-MB-231 human adenocarcinoma breast xenografts in mice using USMB + RT compared to the control of US and oxygen MBs alone (p = 0.03), or US and RT in the absence of MBs (p = 0.01). Animals were treated with in-house manufactured oxygen or nitrogen USMBs and a single fraction of 5Gy using 310 kV x-rays.
Other in vivo studies have also demonstrated that USMB + RT increased tumour cell death 24 hours after treatment. Al-Mahrouki et al. [8], using clonogenic assays from excised PC3 xenografts in mice, showed decreases in cell survival from 45.5% for 2Gy alone compared to 26.8% for USMB + 2Gy; and 38.2% for 8Gy alone compared to 14.4% for USMB + 8Gy, respectively. Lai et al. [9], using in situ end-labelling (ISEL) staining, reported increased rates of tumour death for MDA-MB-231 xenografts in Swiss nude mice of 3.4and 2.3-fold respectively, for USMB + 2Gy or USMB + 8Gy treatments compared to that of 2Gy or 8Gy RT alone.
It has been proposed that USMB + RT acts as a biophysical vascular disruptor in vivo, acting primarily on endothelial cells to disturb tumoural blood supply [10]. As a result of downstream vascular disruption, overall tumour control can be achieved at much lower radiation doses as microvascular collapse and rapid vascular shut down leads to secondary tumour cell death once these cells become starved of nutrients and oxygen. Consequently, three in vitro studies were centred around human umbilical vein endothelial cells (HUVECs) using 3.3% v/v MBs, 30 seconds of US with a mechanical index of 0.8, and between 0 and 8Gy of 160 kVp x-rays.
Clonogenic assays (normalised to control) demonstrated an increase in cell death from 67% for 8Gy alone (p < 0.014) to 95% for USMB + 8Gy (p < 0.002) [11], and a decrease in cell survival from 8% using 2Gy alone, to 1% using USMB + 2Gy [12]. The third study using HUVECs [13], found that treatment with USMB alone or USMB + RT caused cell membrane distortion, which was not evident in cells treated using RT alone, suggesting multiple mechanisms of cell damage may be achieved through the combined USMB + RT treatment.
In vitro studies to date Other in vitro studies investigating direct tumour cell radiosensitisation using USMB, have also been conducted. Karsha an et al. [14] exposed acute myeloid leukaemia (AML-5) cells to varying concentrations of MBs, differing US pressures and 2-8Gy of 160kVp xrays; and observed that cell viability decreased with increased MB concentration, until maximal death was achieved at ~ 1.6% v/v.
Higher USMB levels did not increase cell killing but rather conferred a slightly protective effect. When AML-5 cells were exposed to 570kPa US Peak Negative Pressure (PNP), and 3.3% v/v USMB, cell viability was 71 ± 7% which was higher than that seen for cells only exposed to 4Gy radiation (55 ± 5%). The order of treatment (i.e. USMB followed by RT compared to RT followed by USMB) was shown to have minimal impact on cell viability. Clonogenic assays were 11% for 2Gy alone vs 2.5% for USMB + 2Gy normalised to control (no treatment). Clonogenic survival rates also fell in another study [15], when PC3 and KHT-C murine brosarcoma cells were treated with USMB + 3Gy (PC3 17% and KHT-C 34%, respectively) compared to RT-alone (PC3 50% and KHT-C 75%, respectively). Treatment order and time between treatments also impacted clonogenic survival, with greatest decreases in clonogenic viability seen when PC3 cells were exposed to USMB followed by immediate 3Gy treatment (17%), compared to that of KHT-C cells treated with USMB followed by a 3-hour gap before being exposed to 3Gy (30%).
Deng et al. [16] treated NCE-2 nasopharyngeal carcinoma cells with 3% v/v per uropropane-based USMBs, administered 24 hours prior to exposure to 2Gy or 8Gy RT. Cell viability was measured using the 3-(4, 5-dimethyl-2-thiazolyl)-2, 5-diphenyl-2-H-tetrazolium bromide (MTT) assay 24, 48 and 72 hours post-treatment. Signi cant (p < 0.01) decreases in viability for NCE-2 cells treated with USMB + 2Gy compared to 2Gy was observed at 48 hours; and at 72 hours USMB + RT signi cantly increased cell death for both radiation doses. However, in a study of human pharyngeal squamous carcinoma (FaDU) cells, no differences between cells treated with USMB + RT and RT alone were observed in clonogenic assays [17]. In this study the cells were exposed to 1µg/ml Cisplatin and USMB (700µl of SonoVue™) 24 hours before being treated with varying doses of MV γ-rays (2, 4, 6, 8 or 10Gy). The addition of USMB did not enhance any nuclear damage to that elicited by RT alone in these cells.
Given the variation between results reported for in vivo studies compared to in vitro, the purpose of this study was to investigate the role of USMBs in directly enhancing radiation-induced tumour cell killing in vitro. Two tumour cell lines not previously investigated before were used to determine whether direct USMB radiosensitisation could be achieved for both a primary and metastatic cell line.
HUVECs were also investigated to allow for comparisons to existing published data. De nity® MBs were selected as they are the only commercially available MB approved for clinical use in Australia, and clinical hardware was used to provide US simulation.
Both kV and MV x-rays were used to investigate if different radiation energies enhance the cytotoxic effect of USMBs, as previously it was shown that MV energies did not [17]. Salisbury, United Kingdom) as catalogue numbers 94060902 and 94060303 respectively, and were purchased from CellBank Australia (Westmead, Australia). All cell lines have been previously described [18][19][20].

Cell lines and Culture Conditions
Cells were maintained according to the manufacturer's instructions and incubated at 37 o C, 5% CO 2 and 95% humidity. HUVEC cells were cultured using EGM-2 Endothelial Cell Growth Medium-2 BulletKit™ from Lonza (catalogue number CC-13162

Plating and MTS Assay optimisation
Cell Proliferation Assays were initially optimised in the absence of treatments using CellTiter 96® Aqueous One Solution Cell Proliferation Assay (catalogue number G3582, Promega, Australia). The MTS tetrazolium compound (3-(4,5-dimethylthiazol-2-yl)-5-(3-carboxymethoxyphenyl)-2-(4-sulfophenyl)-2H-tetrazolium) is similar to the MTT compound used in previous studies [16], and is bioreduced by viable cells into a coloured formazan product that can be analysed via colorimetry. Cells are incubated with the reagent for between 1-4 hours, and then absorbance readings taken at 490nm in a 96-well plate reader. The amount of formazan product measured by absorbance is directly proportional to the number of viable cells in the culture [21].
Cell seeding densities and MTS development times were optimised to achieve 80% con uence at 72 hours post-treatment for each cell line, as summarised in Table 1. This level of con uence was chosen in order to avoid any potential effects of contact-dependent cell signalling or contact inhibition, and 72 hours is consistent with the manufacturer's instructions for cell proliferation studies and other experiments using this assay [21][22][23] or the MTT assay [16]. Tullamarine, Australia). Cells at the designated seeding densities were allowed to adhere overnight in their respective asks for ~14-16 hours before treatment. Eight asks were prepared for each experiment -three controls (no treatment), one USMB-alone (USMB), and two asks for each radiation dose level -one with USMB and one without (ie. 3Gy, USMB+3Gy; 6Gy, and USMB+6Gy). Each experiment was performed in triplicate, with separate experiments run for kV and MV x-ray energies using the HUVEC and FTC-238 cell lines. Concurrent experiments were undertaken for NCI-H727 cells where 12 asks were prepared for each experiment -four for kV and four for MV in addition to four 0Gy asks (ie. three controls and one USMB). This experiment was also performed in triplicate.

Pre-Treatment Preparation
Two hours prior to treatment, cells grown in T25cm 2 asks (volumes used for the T12.5cm 2 asks were half that described here) were washed using 5ml Phosphate-Buffered Saline (PBS) (Thermo Fisher, catalogue number 70011069), and detached from their adherent state using 2ml 0.25% Trypsin-EDTA (Thermo Fisher, catalogue number 25200056). Both ask sizes were then lled with 30ml of cell-line speci c complete media. De nity® Microbubbles (Lantheus Medical Imaging, Inc, supplied by Global Medical Solutions, Keilor Park, Australia) were activated using the Vialmix® activation device (Lantheus Medical Imaging, Inc) following the manufacturer's instructions [24], and 480µl added to the relevant treatment asks to give a nal microbubble concentration of 1.6% v/v, which was close to that described in other in vitro studies [11][12][13][14][15][16].
The culture asks including the untreated controls were then transported offsite to the treatment facility at either the Australian Radiation Protection and Nuclear Safety Agency (ARPANSA) (Yallambie, Australia) or the GenesisCare Epping Radiation Oncology Centre (EPROC) (Epping, Australia). Travel time to and from these facilities was around 20min each way, with the cells out of the 37°C incubator for ~90min.

Ultrasound Treatment
Once onsite at the treatment facility, Ultrasound sonication was applied using the LOGIQ i Portable Ultrasound with the 4C-RS transducer (GE Healthcare, Paramatta, Australia). Here, 2MHz frequency ultrasound was applied directly to the ask using a coupling gel, with a focal point of 3.25cm Focal Point and depth of 4cm to achieve a Mechanical Index of 0.9. These settings were the closest match possible to the mechanical index of 0.8 reported in previous studies [11][12][13]. The transducer was moved across the ask for 90 seconds until the opaque, milky-white microbubbles had burst and the media had returned to its normal transparency. Previous preliminary studies using colorimetry and microscopy had shown that this visual change correlated with the bursting of the microbubbles (Data not shown). The asks were then immediately exposed to either 0Gy, 3Gy or 6Gy of kV or MV x-rays. The three control asks were left untreated, whilst the USMB only ask received ultrasound sonication in the absence of ionising radiation.
Radiation Treatment 6MV x-rays were delivered using either a Varian iX (EPROC) or an Elekta Synergy Linear Accelerator (ARPANSA), with the same linac used for each repeat of an individual experiment. Flasks were laid at on the treatment couch on top of 10cm of solid water, with a source-to-surface distance (SSD) of 100cm set to the top of the solid water from a gantry angle of 0 o . Next 2cm of solid water was then placed on top of the asks to provide a build-up region, and both the RT±USMB asks irradiated simultaneously using a 20x20cm eld size to cover both asks. A schematic diagram of this setup is shown in Figure 1.
100kVp x-rays (average energy 42.5kV) were delivered using a Model X80 X-Ray Beam irradiator by Hopewell Designs (ARPANSA). A 100cm SSD was set to the baseplate of the unit before the RT±USMB asks were taped vertically to the baseplate (see Figure 2). A 10cm circle eld size was used to expose both asks concurrently. As the dose rate from this unit was 6mGy/s, the 3Gy and 6Gy doses took 491.8 and 983.6 seconds to deliver respectively, which was the determining factor in the selection of x-ray doses used in this study. The 3Gy dose was chosen as an approximate clinical fractional dose consistent with that used in an earlier study [15], while the 6Gy dose selected as the highest dose deliverable in a reasonable timeframe within the limitations of the 6mGy/s dose rate. Radiation eld uniformity and coverage was validated in the x-and y-planes for all radiation setups using Gafchromic lm (data not shown).

Post-Treatment
Following treatments, the asks were transported back to the laboratory where they were returned into the incubator for a few minutes whilst the laminar ow hood was prepared. Once asks were examined under the microscope to con rm cells were still detached, ask contents were transferred to a 50ml tube and centrifuged (200g for 5min) to remove any residual treatment compounds. Cells were then resuspended in a T25cm 2 ask containing 10ml complete media and returned to the incubator. After 24 hours, the tissue culture media was replenished, and 48 hours after which cell viability was measured using MTS assay.

Cell viability measurements
At ~72 hours after treatment, the media was removed from each ask, and replaced with 1ml supplement free (ie. Incomplete) tissue culture media and 200µl MTS at a ratio of 5:1 per the manufacturer's instructions [21].

Statistical Analysis
Overall normalised survival was averaged across the three repeats of the same experiment, and the S.D. calculated using Excel. The key assumptions of the ANOVA test were initially validated in SPSS V26 (International Business Machines Corporation (IBM), USA), using the Shapiro-Wilk (SW) test to con rm normality of distributions; Levine's test to con rm homogeneity of Error Variances (LHEV); and outliers identi ed visually on the box-and-whisker plot as agged by the software [25,26]. Statistical signi cance between treatment groups was then determined via a Three-Way ANOVA using a 2x3x2 design to report on the two levels of USMB (ie. presence or absence), three radiation dose levels of 0, 3 and 6Gy, and two radiation energies of kV and MV x-rays. Post-Hoc analysis included pair-wise t-tests, with p values <0.05 reported as statistically signi cant [27]. Where LHEV revealed unequal variances, data was re-analysed using a One-Way Welch's ANOVA with Games-Howell Post-Hoc testing after re-expressing the 2x3x2 design as a one-way subprogram [26,28].
A coe cient of variation was calculated using the standard error of the Raw Absorbance values from each experiment and dividing this by the overall average. This value is used to give an indication of any variation arising from differences in plating between experiments [27].

Results
The effect of USMB on the viability of cells exposed to either kV or MV radiation was examined using three different cell lines to observe if ultrasound stimulated microbubbles directly enhance radiation-induced cell killing in vitro. NCI-H727 cells are from a primary non-small cell lung carcinoma (NSCLC), while FTC-238 cells were from a metastatic follicular thyroid carcinoma deposit in lung tissue. Human umbilical vein endothelial cells (HUVECs) were chosen to allow for comparisons to previously published results [11,13].
Plating between untreated control asks was fairly uniform, with coe cients of variation for each experiment ranging from 6-11% as summarised in Table 2, suggesting reasonable consistency in cell numbers was achieved for each repeat of the same experiment. Three control asks were used in each experiment, and the cell number determined using the MTS assay. The results expressed as the mean ± SD for 3 replicates for each experiment. The overall average for the experimental repeats is expressed as the mean ± SEM for 9 replicates. The coe cient of variation for each experiment was calculated as SEM / Overall average.

NCI-H727 cells
Statistically signi cant decreases in irradiated cell survival were demonstrated across all RT dose levels and energies through the addition of USMB (Figure 3). Results were very highly signi cant (p<0.001) for USMB-alone in the absence of RT, with the highest Partial Eta Squared (η p 2 ) value of 0.764 suggesting the presence of USMB had a large effect on survival. The η p 2 for the Radiation dose was 0.728 and for radiation energy it was 0.028, with the later suggesting radiation energy did not in uence normalised survival for the NCI-H727 cells. This was further supported when no statistical difference between kV or MV radiation groups at any dose level was observed (p=0.972, η p 2 <0.001 for 3Gy, p=0.225, η p 2 =0.058 for 6Gy). Three-Way ANOVA revealed equal variance across the entire dataset (LHEV=0.097).

FTC-238 cells
Decreases in survival when the irradiated cells were pre-treated with USMB was more pronounced in FTC-238 cells exposed to kV compared to MV x-rays ( Figure 4). These decreases were not signi cant at any dose level. Between the 2 cohorts, kV radiation had an overall greater impact on cell survival than did MV radiation, however there was no statistically signi cant difference observed between equivalent groups (eg. The mean difference between 3Gy kV-alone vs 3Gy MV-alone was -0.13, p=0.913; and between 6Gy kV-alone and 6Gy MV-alone was also -0.13, p=0.523). Unirradiated cells treated with USMB-alone demonstrated no signi cant impact on survival compared to control in the either the kV or MV cohorts. The exclusion of an outlier observed within the 0GyMVcontrol group resulted in LHEV=2.417, p=0.023, hence a One-Way ANOVA with Games-Howell Post-Hoc testing was used for the analysis of this cell line after the Welch's test of Equality of Means was con rmed as statistically signi cant (p<0.001). SW testing revealed the USMB+6Gy-MV subgroup was non-normally distributed (p<0.0001), however this was only the case for one subgroup, and there is literature to support the robustness of the ANVOA to violations of normality [29,30].

HUVEC cells
When the unirradiated HUVEC cells were treated with USMB there was no signi cant effect on cell viability ( Figure 5). There was a slight, non-signi cant decrease in survival when the irradiated cells were pre-treated with USMB. This effect was similar irrespective of whether the cells were exposed to kV or MV ionising radiation. At equivalent radiation dose levels, cell survival was slightly higher in those cells exposed to MV x-rays compared to kV x-rays, however this was not signi cant. LHEV revealed unequal variances, so these cells were analysed using a One-Way ANOVA with Games-Howell Post-Hoc testing. The resulting Welch's test of Equality of Means was highly statistically signi cant (p<0.001), overcoming the limitations of the initial LHEV result. A single violation of the normality assumption was again observed, this time for the 6GykV-alone subgroup so additional caution was applied in the interpretation of these results.

Discussion
USMB had a signi cant effect on the survival of irradiated NCI-H727 cells. This effect could be cell-type dependent, which has been suggested in an earlier study [10]. These cell-type-speci c responses to USMB + RT have been linked to ASMase gene expression and ceramide signalling associated with apoptosis. This was demonstrated by No ele et al. [12] using ASMase de cient (−/−) astrocytes and ASMase (+/+) astrocytes treated with Sphingosine-1-Phosphate (S1P) to counteract the ceramide-mediated apoptosis pathway.
ASMase (+/+) astrocytes were more sensitive to USMB + RT compared to ASMase (−/−) or S1P-ASMase (+/+) astrocytes where ASMase and Ceramide production had been genetically or chemically inhibited. As NCI-H727 cells elicited a strong response to USMB + RT, changes in ASMase gene expression and ceramide levels produced in response to this treatment should be investigated to determine whether this could be used as an indicator of cell-sensitivity to such treatments. In a study of pre-operative blood serum samples extracted from 61 patients diagnosed with NSCLC, signi cantly increased levels of ASMase activity was detected compared to healthy controls, which may explain why the NSCLC cell line investigated here demonstrated such a strong response compared to the other two cell lines [31]. However HUVEC cells have also been reported to have ~ 20-fold higher levels of ASMase compared to other cells [32], and so should have also demonstrated decreased viability when treated using USMB + RT in this study if ASMase gene expression is the primary mechanism involved. Further investigations on what other factors may be responsible for the dramatic effect USMB have on the viability of irradiated NCI-727 cells are underway.
These factors under investigation include the use of T12.5cm 2 asks, as the acoustic eld applied to the NCI-H727 cells may have been different than that applied to the HUVEC and FTC-238 cells treated in T25cm 2 asks. Kinoshita et al. [33] demonstrated that cell viability was substantially decreased in the presence of standing waves, and the interaction of the US eld with the T12.5cm 2 ask may have created such a phenomenon which was absent in the T25cm 2 asks. It is also interesting to note that Kinoshita et al. reported differences in cell viability when cells were exposed to USMB in a suspended state compared to an adherent state.
Lammertink et al. [17] also exposed the cells in an adherent state, further introducing yet another point of difference to all other research using USMB + RT which may also explain why their results were so different.
Although slight dose enhancement was noted for HUVECs treated with USMB + RT compared to that of RT, these were not signi cant and less than that seen in previous studies [11][12][13]. This difference could be related to the experimental setup, where different cell numbers and culture vessels; MB concentrations; US applications; and x-ray elds were used. The 1.6% v/v MB concentration used in this current study was less than the 3.3% v/v used in previous studies on HUVEC [11][12][13]. MB concentrations was shown to be optimal at 1.6% v/v in enhancing the effect of RT on Acute Myeloid Leukaemia cells [14]; however Lammertink et al. [17] observed no signi cant difference in the survival of FaDU cells when MB at 7% v/v were added to cells exposed to RT. Unlike HUVECs, the effect of MB on the cytotoxic effect of RT has not been examined for either FTC-238 and NCI-H727 cells, and as such 1.6% v/v was chosen as the MB concentration to use in these experiments. As the effects elicited with 1.6% v/v MB on the cytotoxic effect of RT on HUVEC cells reported here was less than those seen in earlier studies [11][12][13], repeat experiments using 3.3% v/v is now warranted.
The FTC-238 cells are derived from a metastatic tumour deposit of a follicular thyroid carcinoma. The addition of MB at 1.6% v/v did not have a signi cant effect on the viability of these secondary tumour cells when exposed to either kV or MV radiation. Of interest was that at the same dose (3 or 6Gy), kV radiation was signi cantly more cytotoxic than MV radiation. This difference between the cytotoxic effect elicited by kV and MV radiation was not seen in HUVEC or NCI-H727 cells. It is unlikely that such responses to differences in radiation energy are cell-type dependent and is more likely due to the differences in raw absorbance readings recorded for the control asks, which varied by ~ 26% between the two x-ray treatment types for the FTC-238 cell line (Table 2). This could have arisen due to differences in travel times to the various treatment facilities resulting in these cells being out of the incubator for different lengths of time. Subsequent research is required to investigate the impact of this further. In the study conducted by Lammertink et al. [17], FaDU cells were exposed to MV x-rays and the survival curves for cells treated with USMB + RT compared to that of RT were similar. This differs to other studies where cells exposed to kV x-rays were treated with USMB and there was a reduction in cell survival rates [14][15][16]. In our study, asks of FTC-238 cells were exposed to kV and MV x-rays at different times and it is unknown if a similar result would be obtained if the asks were irradiated by both forms of x-rays concurrently, as done for the NCI-H727 cells. We observed that in the NCI-727 cells, there was minimal difference in survivals at similar dose levels for kV and MV x-rays, compared to the differences seen when the different cohorts were treated separately for FTC-238 cells.
However, the most likely explanation for the NCI-H727 results here is that these cells are inherently more sensitive to the effects of USMB, which is supported by the very highly signi cant difference (p < 0.001) in normalised survival between the control and USMB only cohorts. Where the HUVEC and FTC-238 cells may bene t from increases in MB concentration, further experiments at lower concentrations for the NCI-H727 cells may be required to establishing toxicity pro les for US, MB and USMB-alone in the absence of RT.

Conclusions
While USMB did enhance radiation-induced cell killing in vitro in the cell lines examined it was only statistically signi cant for NCI-H727 non-small cell lung carcinoma cells. This suggests that enhancement is most likely cell-type dependent, which is consistent with previous ndings in the literature. However, there may also be other factors in uencing these results, such as variations in applied acoustic elds and differences in the cytotoxicity pro les for USM-alone. The differences in cell killing elicited by kV and MV x-rays in FTC-238 cells are likely to be due to other factors, as similar observations were not seen in NCI-H727 or HUVEC cells. Given the overall trends observed here, this study demonstrates that this type of research is feasible using clinical hardware, and that further opportunities exist to understand the speci c mechanisms involved in the direct radiosensitisation of tumour cells using