The present study was performed according to the protocols approved by the Ethics Committee of Tohoku University Graduate School of Medicine (2016–1-071 and 2017–1-581). All the patients included in this study provided written informed consent. This study was conducted between 2016 and 2018. The treatment conditions for the pilot trial of LIPUS therapy (2018 ~ 2022) were determined after comprehensive evaluation, including the results of this study. This study was supported in part by grants from the Japan Agency for Medical Research and Development (16lm0103007j0005 and 17lk1403011h0001).

Optimal sound field setting of LIPUS for intracranial therapy

It has been reported that Aβ accumulates throughout the brain beyond the hippocampus as the AD progresses [15]. Thus, we aimed to apply LIPUS to the whole brain from bilateral temporal bones with the thinnest thickness in the human skull, and designed a special probe that could achieve a sufficient sound field. For LIPUS therapy, we used a multifunction generator (WF1974; NF Corporation Yokohama, Japan) with a bipolar amplifier (BA4825; NF Corporation Yokohama, Japan), and a planer ultrasound probe (Honda Electronics Co., Aichi, Japan). As an estimation of human skull volume, the radius of the curvature of the probe was set based on the Japanese skull size database (Agency of Industrial Science and Technology, Tokyo, Japan). In addition, the distance from the temporal bone to the hippocampus was determined as 75–85 mm in adult patients who underwent head MRI at Tohoku University Hospital. Thus, we designed the probe so that the sound pressure at the depth of 8 mm passed after the temporal bone from the surface to be within the therapeutic range (amplitude 0.1–0.5 MPa) (Fig. 1). The commercially available Acoustic Intensity Measurement System (Eastek Corporation, Tokyo, Japan) and PVDF Needle Hydrophone (Eastek Corporation, Tokyo, Japan) were used as sound field measurement devices. The distance from the LIPUS transducer to the hydrophone was 80 mm. In addition, for safety reasons, we set the sound pressure so as not to exceed \(_}}\) (spatial peak temporal average intensity) 240 mW/cm2 in the cranium based on the Japanese Industrial Standards.

Fig. 1

Sound pressure distribution of LIPUS therapy Sound field analysis of the ultrasound probe using the Acoustic Intensity Measurement System is shown. The energy just below the probe is the highest. We designed the ultrasound probe so that the sound pressure at a depth of 8 mm passing after the temporal bone would be within the therapeutic range (amplitude 0.5-1.0 MPa)

Relationships between bone density/thickness and ultrasound transmittance

Next, since the backbone data of LIPUS therapy are based on previous evaluation of the thin mouse skull, we needed to modify the therapeutic condition of LIPUS so that it could penetrate the human skull and exert a sufficient therapeutic effect on the brain. In this study, we used the skulls of 20 patients who underwent an autopsy at the Anatomy Department of Tohoku University Hospital in 2016. Written consent was obtained from the families of all autopsy patients who donated skulls. Since brain tissue has steady blood perfusion, sound field distribution of LIPUS was thought to be stabilized in a few minutes. Thus, by submerging the human skull in a water tank, we used a hydrophone sensor to examine how much LIPUS penetrated the skull when irradiated from the temporal bone. In addition, the temporal bone thickness \((\Delta _}})\) and bone density \((\uprho )\) were treated as variables that could affect the transmittance, and the correlation with the transmittance was ascertained.

The acoustic impedance of bone \(_}}\) is given by \(_}}=_}}_}}\), and the acoustic impedance of skin and brain soft tissue (almost the same as water) \(_}}\) is given by \(_}}=_}}_}}\). Where the sound velocity in the bone is \(_}}\), and its density is \(_}}\), the sound velocity of water in the skin and brain soft tissue is \(_}}\), and its density is \(_}}\). Transmittance of incident sound pressure from skin to bone \(_\) and transmittance of incident sound pressure from the bone to brain soft tissue \(_\) was given as follow:

On the other hand, when the propagation attenuation coefficient in the bone \(\mathrm}_}}\), the propagation attenuation coefficient in soft tissue (water) \(}_}}\), skin thickness \(\Delta _}1}\), bone thickness \(\Delta _}}\), and brain soft tissue distance \(\Delta _}3}=80 }\), the sound pressure \(_\) at a depth of 80 mm in the brain can be obtained as follows:

$$\beginc} = P_ \cdot exp( - \alpha _}} \Delta h_}1}} ) \cdot T_} \cdot exp( - \alpha _}} \Delta h_}} ) \cdot T_} \cdot exp( - \alpha _}} \Delta h_}3}} ),} \\ \end$$

(1)

where the sound pressure just below the probe is \(_\). Transmittance corresponds to \(_/_\).

Transform this equation as follows:

$$\beginc} }} }}} \right) = \left( }} \Delta h_}1}} - \alpha _}} \Delta h_}} - \alpha _}} \Delta h_}3}} } \right) + ln\left( } \cdot T_} } \right)} \\ - \alpha _}} \Delta h_}} + ln\left( } \cdot T_} } \right)} \\ \end$$

(2)

Here, \(_=-}_}}\Delta _}1}-}_}}\Delta _}3}\) is constant in this experiment.

And, since \(_\gg _\) holds,

$$_=\frac_}_+_}\sim \frac_}_}=2$$

$$_=\frac_}_+_}\sim \frac_}_}$$

$$_\cdot _\sim \frac_}_}=4\frac__}__}=\frac_}__}$$

\(_=4__\) is constant in this experiment.

Inserting these results into Eq. (2) and expressing the dependence of \(}_}}\) on frequency \(f\) as \(}_}}\left(f\right)\),

$$\beginc} }} }}} \right)\sim C_ + ln(C_ ) - \alpha _}} \left( f \right) \cdot \Delta h_}} - ln\left( } \right) - ln\left( } \right),} \\ - \alpha _}} \left( f \right) \cdot \Delta h_}} - ln\left( } \right) - ln\left( } \right),} \\ \\ \end$$

(3)

Here, \(_=_+}(_)\) is constant in this experiment.

Since \(}\left(_\right)\) and \(}\left(_\right)\) are logarithms, it could be assumed that they hardly change. Also, \(}_}}\left(f\right)\) generally increases as the frequency \(f\) increases. The irradiation condition using 1.875 MHz was the standard treatment condition for the heart (angina pectoris) through the intercostal space [2, 3]. However, we found in the preliminary study that almost no transmission occurred with 1.875 MHz through the thickness of the human skull (data not shown). Thus, in order to enhance LIPUS transmittance through human skull, we examined longer wave lengths, i.e., 0.5 and 1.0 MHz.

In vitro experiments to determine optimal irradiation conditions

In order to modify the treatment conditions for mice to those for humans, it seemed to be necessary to adjust not only the frequency but also the duty cycle and sound pressure. Thus, we next performed LIPUS irradiation experiments using cultured vascular endothelial cells and examined mRNA expression of vascular endothelial growth factor (VEGF), fibroblast growth factor 2 (FGF2), and endothelial nitric oxide synthase (eNOS). We previously reported that LIPUS upregulated not only eNOS but also cell growth factors such as VEGF and FGF2 [2, 3]. In this study, we evaluated these growth factors and confirmed the effectiveness of LIPUS. We irradiated endothelial cells with LIPUS through an agar phantom gel as previously reported [2, 3]. The culture dish for cells was filled with culture medium, and an ultrasound probe was placed to prevent air bubbles from entering between them, and irradiation was performed while keeping the probe clean. The center frequencies of the ultrasound probes used in this study were 0.5, 1.0, 1.5, and 1.875 MHz depending on the experimental system. We only used probes that met the standard of variation within ± 8% based on the ultrasonic output of 72 mW. The ultrasound output was measured using a high-performance electronic balance-type low-ultrasound-power measuring device (UPM-DT-1E; Eastek Corporation, Tokyo, Japan). The probe shape was a circular planar element (diameter 26 mm, 28 mm). The attenuation coefficient \(}_}}\) of the phantom gel was almost comparable to that of living cells (e.g., muscles, fat, and blood). Human umbilical vein endothelial cells (HUVEC) from a single donor (Lonza, Basel, Switzerland) were cultured in a complete endothelial medium (EGM-2 BulletKit, Lonza). The cells were used at passages three to five and were maintained in EGM-2. Twenty-four hours before LIPUS treatment, the cells (1 × 105) were re-suspended in a 2-ml tube with EGM (Lonza). They were exposed to LIPUS under the following irradiation conditions (frequency: 0.5, 1.0, and 1.875 MHz; duty cycle: 1, 5, 10, 20, and 40%; amplitude: 0.05, 0.1, 0.15, 0.25, 0.5, 1.08, and 2.2 MPa) for 20 min (n = 12 each). Since heat generation on the element surface cannot be ignored at a duty ratio of 40% or more, we only examined the duty ratio under conditions of 1, 20, and 40%. Regarding sound pressure, referring to our previous reports, it was thought that sound pressure over 2.2 MPa is highly cytotoxic [2, 3]. The pulse repetition rate (PRT) was fixed at 320 μs to comply with previous reports [2, 3] and safety standards [16]. After irradiation, the cells were stored for 6 h in the same medium before RNA extraction. mRNA was extracted using the RNeasy Plus Mini kit (QIAGEN). mRNA (600 ng) was reverse- transcribed using a QuantiTect Reverse Transcription kit (QIAGEN). Real-time PCR was performed using the Real-Time Detection System (Bio-Rad Laboratories). cDNA was synthesized by using PrimeScript RT Master Mix (Takara Bio Inc., Tokyo, Japan). The primer sequences were as follows: VEGFA, (Forward) 5’-GAGCCTTGCCTTGCTGCTCTA-3’ and (Reverse) 5’-CACCAGGGTCTCGATTGGATG-3’; FGF2, (Forward) 5′- GTGTGCTAACCGTTACCTGGCTATG-3′ and (Reverse) 5′- CCAGTTCGTTTCAGTGCCACA-3′; eNOS, (Forward) 5’-AAAGACAAGGCAGCAGTGGAAAT-3’ and (Reverse) 5’-TCCACGATGGTGACTTTGGCTA-3’; GAPDH, (Forward) 5’-GCACCGTCAAGGCTGAGAAC-3’ and (Reverse) 5’-TGGTGAAGACGCCAGTGGA-3’, all of which were designed with the Perfect Real Time Support System (Takara Bio Inc.). For the final validation, the irradiation conditions considered to be the optimal conditions for LIPUS treatment were applied to cerebral vascular endothelial cells (Lonza), and VEGF mRNA expression was also evaluated.

Statistical analysis

Results are shown as mean ± standard deviation (SD) for all experiments. We used Student’s t-test followed by Bonferroni multiple comparisons and 2-way ANOVA with Turkey’s HSD multiple comparison test to compare mean values (GraphPad Prism Software, San Diego, CA). Differences with a P-value < 0.05 were considered to be statistically significant.