Thirty-one male Sprague–Dawley rats (8–9 weeks old, Japan SLC, Japan) were used in this experiment. They were fed a standard pellet diet watered ad libitum in a room with a 12-h light/dark cycle at a temperature of 22.5 ± 1.0°C and a relative humidity of 50 ± 20%. The dorsal body hair of the rats were shorn 1–3 days before the start of the experiment. All experimental procedures were conducted in accordance with the ethical guidelines for animal experiments at Kurume University School of Medicine (approval numbers: 2020-174 and 2021-150).

The temperature in three regions (rectum, dorsal skin, and tail skin) and the blood flow (BF) in two regions (dorsal skin and tail skin) were measured in real time using fiber-optic thermometers (FL-2400, Anritsu Meter Co., Ltd., Japan) and Doppler blood flow meters (FLO-C1 TWIN, Omegawave, Tokyo, Japan), respectively. The animals were anesthetized with isoflurane (3%) in 100% oxygen (Narcobit-E, Natsume Seisakusho Co., Ltd., Japan) for placement in an acrylic holder (Rev1, Kyoto Jushi-Seiko, Co., Ltd., Japan) and to set up the following probes: one thermometer probe was inserted into the rectum (2 cm from the anus), and the other two thermometer probes and BF-meter probes were placed in contact with the dorsal skin (below the antenna) and the base of the tail. All probes were made of non-metallic materials. Analog data acquired from the fiber-optic thermometers and Doppler BF meter were digitized using an A/D converter (PL3516, ADInstruments Ltd., New Zealand) at a sampling rate of 200 Hz (20). Because the Doppler BF meter was unable to measure the absolute flow values, the BF was revealed as a value relative to the reference response. This is displayed as the amount of change, with the value at the beginning of the exposure as 100%. BF data were analyzed after removing the upper 0.01 Hz with a low-pass filter.

Whole-body exposure systems were designed to expose the rats to qMMW (Figure 1). The system consists of power supply chain as shown in Figure 1C: a signal generator (JOGSAG1401, SAF Tehnika, Riga, Latvia), an amplifier (AMP6034, Exodus Advanced Communications, Las Vegas, Nevada), a power meter (U8487A, Keysight Technologies, Inc., CA), and a lens antenna. The lens antenna consists of a conical-horn antenna and a dielectric lens which was made of the ultrahigh-molecular-weight polyethylene (relative permittivity of 2.3). The detailed of the developed antenna is described in our previous study (21). The sinusoidal wave at 28 GHz were generated using a signal generator and fed into an amplifier via a cable. The amplified continuous wave (CW) was fed to the lens antenna through a waveguide (WR-28) and power meter, and then radiated as a gaussian distribution on the surface 50 cm from the front of the antenna. The input power (IP) of the CW from the amplifier to the antenna was constantly monitored using a power meter via a 40 dB coupler. Three IP levels to the antenna, namely 0, 14, and 28 W, were selected for the experiment.

Figure 2A illustrates a numerical rat model used to derive absorbed power in the computation (21, 22). The model was developed from CT image of a Sprague–Dawley rat weighing 265 g. The model consists of six anatomical tissues (skin, muscle, fat, bone, brain, and eye). The dielectric properties of the tissues were determined with a Cole–Cole dispersion model (23). The power absorption in the rat model was computed using the finite-difference time-domain (FDTD) methods (24) exposed from the lens antenna (21).

The WBASAR is defined as follows in the guideline (11):

where |E(r)|, σ, and ρ, r denotes the internal electric field (root mean square values), the conductivity, and the tissue density, position vector in the rat model, respectively.

The APD is defined at the body surface (11):

where A denotes the averaging area, z = 0 is the body surface, z_max is the depth from the body surface, which is much larger than the penetration depth.

After setting for the measurements of the physiological parameters, a rat was kept in an acrylic holder without anesthesia and placed 50 cm below the antenna (Figure 1). 28 GHz-qMMW was exposed to the dorsal side of the rat for 40 min individually. Rats were randomly divided into three groups, and each group of rats was exposed at 0 W (n = 8), 14 W (n = 11), and 28 W (n = 12) of IP, respectively. The measurements started 4 min before the beginning of the exposure and lasted until the end of 40 min of exposure. The data, including the pre-exposure data, were recorded on a PC and later analyzed offline. Through the experiment, the room temperature and humidity were kept at 22.5 ± 0.4°C and 44.5 ± 1.3% (mean ± SD), respectively.

The difference in elapsed time per second compared to the sham-exposed group was analyzed nonparametrically using the Kruskal-Wallis test followed by the Steel test, a conservative method because the number of observations was insufficient to confirm the distribution. Estimation of exposure intensities of 28 GHz-qMMW from observations in rat exposure experiments at each exposure region was performed using a linear regression model based on the assumption of a normal distribution of temperature change (p = 0.1 to 0.9 for Shapiro–Wilk test). These tests were performed using Bell Curve for Excel software (ver. 3.21; Social Survey Research Information Co. Ltd.).

Figure 2 shows the distribution of power absorption in the rat model exposed to the lens antenna as an IP of 1.0 W. The peak spatial values of the incident power densities (IPD) and APD averaged over 4 cm² corresponding to an IP of 1.0 W were 22.1 W/m² and 8.5 W/m², respectively, at 50 cm below the center of the antenna. Under the exposure conditions, the relative value of WBASAR was estimated as 0.26 W/kg. The IPD, APD, and WBASAR at each IP level are listed in Table 1.

At the beginning of whole-body exposure, the temperatures of the three regions were not significantly different among the three groups, including the sham-exposed group (n = 8–12 animals per group). The initial temperatures for all rats in dorsal skin, rectum, and tail skin were 34.9 ± 0.9°C, 36.7 ± 0.5°C, and 27.4 ± 1.5°C, respectively.

Local temperature and BF were measured simultaneously during exposure to investigate whether qMMW exposure changes them. Figure 3 shows the trend over time of the temperature response to qMMW exposure. In the two qMMW-exposed groups, the dorsal skin temperature increased immediately after exposure (Figure 3A), whereas there was a delay of 46 s or more in the rectum (Figure 3B) and tail skin (Figure 3C). These differences were reflected in the elapsed time during which a significant increase in temperature was detected compared to the sham-exposed group (Table 2). The elapsed time for dorsal skin temperature was 4 s and 5 s for 28 W and 14 W IP, respectively. In contrast, 49 s and 351 s of elapsed time were required for the tail skin and rectum, respectively, even at 28 W of IP.

These temperature increases persisted until the end of exposure, but the trend over time patterns were different. The dorsal skin temperature had two distinct phases. It approached its maximum value after the first phase of the rapid temperature rise, followed by the second phase of gradual temperature rise. In contrast, the rectal and tail skin temperatures increased almost linearly until the end of the exposure period. Interestingly, the temperature changes in the tail skin were approximately twice as large as those in the dorsal skin at the end of the exposure.

Figure 4 shows the trend over time of the BF under qMMW exposure. Dorsal skin BF (Figure 4A) showed no obvious changes with exposure, even at the maximum intensity. In contrast, tail skin BF increased linearly from the beginning to the end of exposure (Figure 4B). In addition, the degree of increase and elapsed time of tail skin BF were dependent on the exposure intensity. Although changes in the BF were observed early after exposure, the elapsed time was more than 6 min because of the variability of BF signals, which are susceptible to rat movements.

To elucidate whether the increase in local temperature depended on the intensity of qMMW exposure, we analyzed the correlation between these parameters at 6, 12, and 30 min after the beginning of exposure (Figure 5). At first, the normality of the data was confirmed for all combinations of exposure intensity and exposure duration in three regions. The temperature changes in all three regions had a significant relationship with the exposure intensity in each period, and the relationships were indicated as linear regression models (Table 3).

Using these regression models, we estimated the exposure intensities to induce 1°C or 5°C of temperature increase in each region. The dorsal skin was directly irradiated with qMMW; thus, the required exposure intensities were evaluated using a regression model between the APD value and temperature change in the dorsal skin. As a result, it was estimated that 88 W/m² and 488 W/m² of APD could induce 1°C and 5°C of temperature increase, respectively, for 6 min of exposure.

In contrast, rectal and tail skin temperatures were unlikely to be directly altered by qMMW exposure. Therefore, it is suitable to calculate the required intensity using the WBASAR value for 30 min exposure. It was estimated that 4.6 W/kg and 38 W/kg of WBASAR could induce 1°C and 5°C of rectal temperature increase, respectively. In tail skin, 0.2 W/kg and 4.6 W/kg of WBASAR were required for 1°C and 5°C temperature increase, respectively.

Previous physiological studies in rats (19) suggest that heat accumulated in the rat body is dissipated through the tail. Indeed, in the present experiment, the largest temperature change was observed in the tail skin of the three regions we measured (Table 3), even though the tail region was not exposed to much qMMW. This fact reminds that a similar thermoregulation may be at work during dorsal skin heating by qMMW exposure. To explore the mechanism of this heat transfer, we plotted the mean ± SD values of temperature and blood flow changes in each region every 3 min after the beginning of exposure, and examined the correlation between the measured parameters.

As a result, two characteristic linear relationships with relatively little dependence on exposure intensity were found between rectal temperature rise and tail skin BF (Figure 6A), and between tail skin BF and tail skin temperature (Figure 6B). One was that the tail skin BF (BFt) increases in proportion to rectal temperature change (Tr), as shown by the equation BFt = 149.5Tr + 99.8 (R² = 0.721, p < 0.01). This linear model was designated as thermoregulatory model-1 for later use. The other was that the tail skin temperature (Tt) rises in proportion to the change in tail skin BF (BFt), as shown by Tt = 0.036BFt – 3.18 (R² = 0.933, p < 0.01). This second linear model was designated as the thermoregulatory model-2 for later use.

To validate the above two linear regression models in the rat thermoregulatory system, we estimated the increase in tail skin BF and the change in tail temperature based on the increase in body temperature during qMMW exposure and compared the estimated values with those obtained in animal experiments. Because a significant increase in tail BF was observed at least 570 s after the beginning of exposure (Table 2), estimates were made for data at 12 and 30 min of exposure.

First, rectal temperature changes during qMMW exposure were calculated for every 0.5 W/kg in the range of WBASAR from 0 to 8 W/kg, using a model derived from the relationship between WBASAR and rectal temperature change. In the animal experiment, the rectal temperature of the sham-exposed group was maintained below 0.5°C, but spontaneous temperature rises were observed throughout the experimental period. To avoid the influence of such spontaneous changes in the estimation using the regression model, the rectal temperature during the sham exposure was defined here as 0°C. In this case, the equations for rectal temperature change at 12 and 30 min were modified to y = 0.091x_w and y = 0.193x_w, respectively, compared to the regression equations in Table 3. Where y and x_w are rectal temperature change (°C) and WBASAR (W/kg), respectively. Using these equations, the rectal temperature change was estimated to be 0 to 0.7°C and 0 to 1.5°C at 12 and 30 min of exposure, respectively, for WBASAR 0 to 8 W/kg.

Second, BFt was calculated by substituting the above estimated value of rectal temperature change (Tr) into thermoregulatory model-1. As a result, BFt was estimated to be 100 to 209% and 100 to 331% at 12 and 30 min of exposure, respectively, relative to WBASAR 0–8 W/kg (Figure 7A).

Third, the change in tail skin temperature (Tt) was calculated by substituting BFt estimated in the second step just above into thermoregulatory model-2. As a result, Tt were estimated to be 0.4 to 4.2°C and 0.4 to 8.6°C at 12 and 30 min of exposure, respectively, for WBASAR 0 to 8 W/kg (Figure 7B).

Finally, we confirmed that both the estimated BFt and Tt values were close to the experimental values (solid plots in Figure 7), suggesting that model-1 and model-2 may be reasonable systems for increasing tail skin temperature as thermoregulatory mechanisms.

In this study, we subjected rats to whole-body exposure to 28 GHz-qMMW and measured their physiological parameters using a real-time measurement system for biological responses. We succeeded in simultaneously capturing changes in body temperature and skin blood flow in response to qMMW exposure in awake conditions. The temperature changes in each region appeared with respect to their respective elapsed times. The skin temperature on the dorsal side just below the antenna increased rapidly after exposure, followed by an increase in the temperature in other regions. A similar result was observed in anesthetized rats exposed to 35 or 94 GHz MMW (7, 25). These patterns appear to be consistent with the initial accumulation of energy in the dorsal skin, followed by heat transfer to internal structures. This indicates that the energy of the qMMW is absorbed at the skin surface, which is consistent with the fact that MMW penetrates only a few millimeters through the skin (15, 16).

It is important to estimate the intensity of qMMW exposure to increase tissue/organ temperature, which could cause biological effects. In the ICNIRP guideline for local and whole-body exposures to >6 to 300 GHz, skin and core temperatures to avoid health effects are set at 5°C for 6 min of local exposure and 1°C for 30 min of whole-body exposure, respectively (10). However, little information is available regarding the exposure intensities that induce temperature changes in living animals and humans. This is because the current values of exposure intensities in the guidelines were estimated using simulated human models (12, 13).

The present study related local temperature and exposure intensities in a living rat body and allowed the estimation of the exposure intensity required to reach the expected temperature for exposure, at least under the experimental conditions we used. The estimation revealed that the exposure to 488 W/m² of APD was necessary to increase the dorsal skin temperature by 5°C for 6 min. This intensity was twice or more than that simulated (200 W/m²) using the human skin model (16). This suggests that the temperature in the living skin does not increase as much as that in simulated skin, suggesting the occurrence of thermoregulation, such as BF changes. On the other hand, whole body exposure at 4.6 W/kg of WBASAR for 30 min would induce a 1°C increase in rectal temperature in the rat. This intensity was quite similar to that estimated using the human whole-body model (4 W/kg for a 1°C increase in core temperature).

The data presented in Table 3 provide additional information on the following two adverse biological effects. First, in dorsal skin, qMMW exposure at 291 W/m² or more of APD can increase the skin temperature by 5°C for 30 min. At the beginning of exposure, the skin temperature was 34.9 ± 0.9°C. Thus, the final temperature of the dorsal skin was expected to be 40°C. It has been reported that the critical temperature for deep dermal burns was 41.9°C in rat skin (26). Therefore, if the exposure intensity exceeds an APD of 291 W/m², qMMW exposure may induce skin damage. Second, we estimated that qMMW exposure at 38 W/kg of WBASAR increased the rectal temperature by 5°C for 30 min. This suggests that the rectal temperature of the rat reaches 42°C because its initial temperature was 36.7 ± 0.5°C. Several research groups have found a precipitous decrease in arterial blood pressure and subsequent death in rats at approximately 42°C of colonic temperature under MMW exposure (25, 27, 28). Thus, a similar phenomenon may be induced at our estimated exposure intensity of 38 W/kg of WBASAR.

The rat may have used its tail to dissipate the heat accumulated in its body under 28 GHz-qMMW exposure. It is well known that the rat tails play an important role in releasing heat from its body, acting like a radiator because rats do not use sweat glands (19). In animals, the tail, ears, and limbs are important heat-dissipating organs. When the bodies of rats and guinea pigs were electrically heated and the local temperature was examined, the tail temperature showed a marked increase compared to other organs (29). It has also been confirmed that almost 20% of heat is dissipated from the rat tail (30). In general, the temperature of the rat tail skin is the same as the ambient temperature and blood barely flows from the tip when the tail is amputated. However, when the body is warmed and the temperature of the tail is raised, thick blood vessels become visible under the tail skin and blood starts to flow from the tip if the tail is cut off (29). In other words, when the body is warmed, heat is transported to the tail through the BF.

A similar physiological response was observed in the tail following qMMW exposure. qMMW exposure increased skin temperature and BF in the tail. Moreover, the temperature changes in the tail skin were twice as large as those in the dorsal skin (Figures 3A,C), even though the tail skin received little exposure to qMMW. Interestingly, BF did not change in the dorsal skin even when heated to the maximum intensity of qMMW exposure. These results suggest that the heat generated in the rat dorsal skin is transferred to the tail skin via an intentional change in the proportion of BF to maintain or reduce the core body temperature.

Whole-body exposure to 28 GHz-qMMW could be a triggered the activation of thermoregulatory systems via an increase in core temperature in rats. To explore the mechanism of thermoregulation observed in rats exposed to qMMW and estimate the threshold of exposure intensity that drives this regulation, we investigated the correlation between the measured parameters. We found that the three linear regression models were suitable for explaining and mimicking these systems. Namely, the models could demonstrate the following steps: (1) heat generation by whole-body exposure to qMMW (which mainly occurs in the skin of the body surface) accumulates in the internal body and increases the core temperature; (2) the change in core temperature reflects the BF increase in the tail skin; and (3) the BF increase in the tail skin is directly reflected in the tail skin temperature increase and promotes heat dissipation from the tail.

Each of these steps was consistent with the findings of previous physiological experiments. The results from Step 1 were supported by Millenbaugh’s findings using ambient temperature increase and MMW exposure (7, 25), although these studies lack an assessment of the intensity dependence of temperature. Step 2, described as model-1, may involve the central nervous system in rats. It is possible that the hypothalamus senses an increase in core temperature (31, 32) and dilates peripheral blood vessels in the tail (33) through the inactivation of the efferent nerve (sympathetic nerve) (34, 35). Step 3, which was explained as model-2, is also shown in the proportional relationship between BF and temperature in the rat tail directly (36) and indirectly (37). Other mechanisms such as evaporative respiratory heat dissipation (38) may also contribute to thermoregulation under qMMW exposure. However, the estimated values of blood flow and temperature changes in the tail skin were similar to those measured in conscious rats. Our findings suggest that these models are dominant in simulating thermoregulation under qMMW exposure.

According to the estimated results using thermoregulatory model-1 (Figure 6A) and model-2 (Figure 6B), whole-body exposure to qMMW may trigger thermoregulatory systems even at less than 4 W/kg of WBASAR. For example, whole-body qMMW exposure at 2 W/kg could drive the thermoregulatory systems, accompanied by increases in tail BF and tail skin temperature by approximately 50 points and 2.5°C, respectively (Figure 7). Many research groups have used 4 W/kg of WBASAR as one of the criteria or thresholds to evaluate the non-thermal/thermal effects of exposure to microwaves in animal experiments (3, 39–46) because this value of 4 W/kg is a biological basis for the basic restriction mentioned above. In addition, some of these studies found biological effects of whole-body exposure below a WBASAR of 4 W/kg (3, 46). However, the biological responses hypothesized here seem to be reversible and physiological to maintain homeostasis (36). Therefore, it is important to confirm whether the effects of low exposure intensity are adverse or within the physiological range.

In the present study, there are a couple of limitations that should be considered when establishing a thermoregulatory model. First, rats were irradiated with qMMW only from the dorsal side. The abdominal skin of the rat, which was placed on the opposite side of the antenna, was not directly exposed. Therefore, the abdominal skin may contribute to the keep/decrease in core temperature. Second, there was large dispersion in the measurement data, particularly for skin blood flow. This was due to the experimental difficulties caused by the use of conscious rats. Nevertheless, the models obtained for physiological parameters could provide useful information for future experiments because of the similarity in values between the experimental measurements and estimations.