RF Safety Temperature Limits And Thermal Damage Biology Essay

Published: Last Edited:

This essay has been submitted by a student. This is not an example of the work written by our professional essay writers.

Safety standards for human exposure to radiofrequency electromagnetic fields (RF EMF) are based on dose limits for ambient exposure to RF electric and magnetic fields &/or induced RF EMF inside the body (SAR and current density). There is increasing interest in using temperature as the fundamental dose metric for limiting RF EMF exposures since tissue heating is recognized as a principal mechanism for causing adverse effects. The time of elevated temperature is also an important factor, and time/temperature thresholds for causing adverse tissue effects have been quantified using the CEM43°C exposure parameter. In reviewing the applicability of temperature based limits for RF safety standards we concluded that: 1) Both temperature and exposure time are required to predict thermal tissue damage; 2) The published CEM43 °C data below 43 °C is generally too sparse and uncertain to be reliably used for the generally weak heating induced by RF EMF exposures; 3) Differential sensitivity of tissues argues for the use of more tissue-specific thermal or SAR limits; 4) There are substantial problems in developing compliance tests for temperature based limits; 5) Temperature based limits are feasible and potentially better than RF-EMF limits, but will require more research.

Introduction (EM safety standards).

Tissue heating is a well established biophysical effect and potential hazard for human exposure to radiofrequency (RF) electromagnetic fields (EMF), particularly at frequencies above 100 kHz. Low level non-thermal RF effects have also been reported in the literature and different physical mechanisms have been proposed to explain them, but to date are either implausible, irreproducible or have no obvious health impact [Adair, 2003; Challis, 2005; Vecchia et al., 2009]. Consequently, RF heating is the main rationale for setting limits on human exposure to RF EMF above 100 kHz in the major guidelines and international standards for RF safety [ICNIRP, 1998; IEEE, 2006] .

The level of RF induced tissue heating is commonly quantified by the Specific energy Absorption Rate, SAR, in W/kg. The SAR can be calculated at any point inside the body from the magnitude of the internal electric field, Eint (rms V/m), the tissue conductivity, ³ (S/m), and the tissue mass density, ² (kg/m³):

SAR = ³|Eint²|/²

Basic restriction limits on SAR in safety standards and guidelines are formulated as mass averages in recognition of thermal diffusion effects. A whole body average (WBA) SAR limit is provided for protection against whole body systemic heating effects and localized SAR limits for protection against local tissue heating effects[ICNIRP, 1998; IEEE, 2006]. However SAR is a only a partial indicator of local tissue temperature rise as other factors such as blood perfusion are very important heat transfer mechanisms [McIntosh et al., 2008]. ICNIRP safety standard considers that "Many laboratory studies with rodent and non-human primate models have demonstrated the broad range of tissue damage resulting from either partial body or whole-body heating producing temperature rises in excess of 1-2 ° C" [ICNIRP, 1998]. A further factor to be considered is exposure time. In this paper we focus on the issue of local heating effects and whether exposure limits for this potential hazard would be better expressed on a temperature based limit (temperature rise or its derivative that considers time: CEM 43 °C) rather than a localized SAR limit. Temperature based limits are discussed in the context of the CEM 43 °C concept, the cellular mechanisms and tissue sensitivity to thermal damage. An emphasis is set on nervous system effects and the role of regulation of brain temperature since this site of exposure has been subject to special concern in safety standards

CEM 43 as indicator for thermal tissue damage

The main driving force in the development of thermal dosimetry have been the efforts to use hyperthermia as a treatment to kill cancerous cells aggregated in tumors with minimal damage in surrounding tissue. Thermal cell killing or damage is not only depends on temperature, but also on time of exposure. The many possible combinations of time and temperature prompted researchers to find a method to determine tissue damage thresholds without the need to test every possible combination. Extensive experimental evidence of hyperthermic exposure of cell cultures in vitro and tissues in vivo supports the notion that the time required to produce a specific level of thermal damage or isoeffect is an inverse exponential function of the temperature in a fashion that follows the Arrhenius equation for temperature dependent processes like chemical reactions [Upadhyay, 2006] as illustrated in figure 1 .

The Arrhenius-like relationship between thermal damage, time and temperature allowed 25 years ago to formulate a "thermal dose" quantity by converting any combination of time and temperature to "equivalent minutes" of hyperthermia at a standard reference temperature of 43°C [Dewey, 1994; Sapareto and Dewey, 1984]. The thermal isoeffective dose can be calculated as follows:

CEM 43 °C(min) = tR (43-T)

where CEM 43 °C(min) is the cumulative number of equivalent hyperthermia minutes at 43 °C to produce the same effect; t is the time interval; T is the average temperature during the time interval t, and R is the ratio of the time interval when temperature is decreased by 1 °C. Log R is the slope in the time vs. temperature plot (Fig 1). When R is determined experimentally the activation energy („H) of the process can be calculated from the formula lnR =„H/2T2.. R is frequently assumed to be 0.5 above 43°C and 0.25 below this temperature. When there is temporal variation in the tissue temperature, the CEM 43 °C needs to be calculated for each period bin where the temperature is constant and then a summed over the whole period of exposure [Dewhirst et al., 2003; Sapareto and Dewey, 1984].

The plot of Log t vs T for in vitro data is typically biphasic with a breakpoint around 43 °C where a change in slope is found. For in vivo studies, only few of them contain the information required to derive the Arrhenius data. In human and pig skin the breakpoint was found to be 47 °C and 42.5 °C in mouse [Dewhirst et al., 2003]. The higher breakpoint temperature in human and pig skin, predicts a greater thermal resistance than other tissues although the measurements are suspected to underestimate the actual temperature by 1-6 °C due to the spatial separation from the basal layer where the more important damage occur. In consequence the thermal isoeffective dose to achieve injury in human skin is expected to be overestimated [Dewey, 1994; Dewhirst et al., 2003]. Thus it is uncertain where the breakpoint is in vivo. A second difficulty to calculate the thermal isoeffective dose is the uncertainty of the R value below the breakpoint. For rodents the R value in vivo is between 0.2 and 0.25 and in human skin is estimated to be 0.13 [Dewhirst et al., 2003]. A third problem is that most studies did not evaluate temperatures below 42 °C. To allow for these uncertainties Dewhirst proposed to use a conservative approach to estimate the thresholds for thermal damage for human and other animal tissues. As can be seen from figure 1, this is accomplished by using the parameters derived from rodent tissues that have a lower breakpoint and higher R value below the breakpoint and hence produce thermal isoeffective doses estimates with lower values [Dewhirst et al., 2003].

Cellular mechanisms of thermal damage.

The cellular mechanisms of thermal damage are reviewed by Lepock [Lepock, 2003] and are summarized as follows: The specific biophysical mechanisms of hyperthermic damage are not well known. The underlying molecular events have been inferred from Arrhenius analysis of survival curves of "in vitro" cell cultures. Cell killing requires an activation energy around 120-150 Kcal/mol, a range that overlaps with the activation energy for molecular transitions (100-200 Kcal/mol) which are defined as structural transformations from a more ordered (native) state to a more disordered (denatured) state triggered by a temperature change. The activation energy for metabolic and other enzymatic reactions (3-20 Kcal/mol) is below the cell killing range, thus suggesting they are not the primary cause of thermal damage. Different kinds of transitions can occur in cells but is possible to distinguish them by knowing the temperature range at which they occur. Transitions in membrane lipids are known to occur at 8 °C and no strong evidence for temperatures above 37 °C. DNA and RNA transitions (melting) usually occur around 85-90 °C but in small molecules such as tRNA, protein-RNA complexes and in some instable regions of the larger molecules local melting is possible at lower temperatures. However the transition most likely to be responsible for hyperthermic cell killing is protein denaturation even when other transitions may also play an important role. In mammalian cells denaturation occurs when the temperature is higher than 40-42 °C, although can be lower in conditions of proteotoxic stress. Cell killing can occur when more than 5% of the total cellular protein is denatured and reaches 95% of killing when denaturation is 10%. The cell damage by protein denaturation is due to direct inactivation of protein function (i.e. as an enzyme, membrane receptor, ion transporter, etc) or by disruption of complex structures (i.e., depolymerization of complex structures as the cytoskeleton, protein aggregation and formation of insoluble aggregates). Protein aggregation occurs because denatured proteins expose their hydrophobic regions and they interact with each other. The aggregated proteins may include proteins not directly altered by hyperthermia. Structural damage is more likely to be irreversible [Lepock, 2003].

The thermal damage of proteins produces a chain reaction or domino-like effect leading to a wide spectrum of disrupting consequences. They include DNA damage by disruption of transcription, replication and repair of DNA that cause lethal lesions and makes cells in mitosis and S- phase more sensitive to heat killing and changes in plasma membrane protein distribution and permeability that affect membrane potential. A similar effect on mitochondrial membrane causes a change in the redox status of the cells that result in a burst of free radicals that enhance protein sensitivity to heat [Lepock, 2003; Roti Roti, 2008].

Tissue sensitivity and thresholds for thermal damage


In a comprehensive compilation by Dewhirst [Dewhirst et al., 2003] it was found that thresholds to damage are endpoint dependent, and the sensitivity of the endpoints varies considerably. Arrhenius plot analyses for a wide range of time-temperature combinations have only been studied in human and pig skin, and several rodent tissues. From these it was found that, diverse tissues or different endpoints in the same tissue have dissimilar thresholds but parallel CEM 43 °C curves. However the threshold, so far is the same between diverse species when the same tissue is compared at the same endpoint. In skin of rodents, pigs and humans, it has been noted that around the threshold for damage, a small increase in exposure time produces an abrupt increase in damage incidence [Dewhirst et al., 2003; Law, 1979; Moritz and Henriques, 1947]. Due to the steepness of the transition to damage, safety factors for RF limits are required to avoid localized thermal burns. Current RF safety factors for local heating use 1 °C temperature rise as a reference that seems to be more related to a systemic effect (i.e. whole body heating). However there is no evidence of damage of any human tissue by a local temperature rise of 1 °C above the basal temperature of 37 °C. A further flaw is that no exposure time is considered.

The amount of damage is dependent on time passed between the exposure and the assessment, frequently several hours or days after the hyperthermic exposure, an increased level of damage can be found [Dewhirst et al., 2003]. In clinical skin burns, the phenomenon is known as secondary aggravation or conversion, where the size and deep of the initial injury usually increases afterwards [Mahajan et al., 2006; Penington et al., 2006; Singer et al., 2008]. Nevertheless in assessments with a further delay, restorative events might reverse the damage to some extent [Dewhirst et al., 2003].

The properties of thermal sensitivity described occur above the breakpoint temperature and no clear curve is observed below that value. These findings are generally assumed to be common to all tissues in all species. The sensitivity to thermal damage doesn't follow a clear tissue classification on proliferative potential or tissue type. For example, brain and testis can be considered between the most sensitive tissues to heat damage but differ greatly in their proliferative potential. Peripheral nerves are regarded as highly resistant to heat and spinal cord is in the intermediate level [Dewhirst et al., 2003].

Other factors that have been suggested to play an important role in the observed sensitivities of the tissues include: tissue architecture and kinetics of repair and cell replacement; subtle differences in protein structure, thermotolerance development due to differences in heating rate, acidosis by low blood flow [Dewhirst et al., 2003].

Sensitivity of brain tissues

We present endpoints for brain tissues from the literature reviewed; the rat brain tissue thermal sensitivity is unexpectedly low. In adult rats, a CEM 43 °C as low as 0.12 s has been reported to produce limbic seizures that involve the hippocampus when the CEM 43 °C is calculated from the temperature measured directly in the hippocampus; a higher dose is obtained (23.1 s) when core temperature is used [Ullal et al., 2006; Ullal et al., 1996]. With repeated hyperthermic-induced seizures, histopathological alterations are found in adult rats from a thermal dose of 1.64 s [Ullal et al., 1996]. In rat pups a CEM 43 °C of 3.44 s (calculated from core temperature) can induce seizures [Schuchmann et al., 2008]. Higher thermal doses (i.e. > 85 s) provoke seizures in practically all pups [Dube et al., 2006]. Proneness to seizures is age dependent and it is higher in pups around 10 days after partum and lower in older rats [Schuchmann et al., 2006]. The brain in rats of this age has been considered by some authors, the equivalent in development to a human brain of several months to 3 years age, because the similarity in structural and functional changes taking place [Avishai-Eliner et al., 2002; Dobbing and Sands, 1979]. Pup rats are used as an animal model of febrile seizures that in humans occur mostly in infants [Bender and Baram, 2007; Dube et al., 2009].

In humans, a typical febrile seizure is short and characterized by behavior arrest, confusion or altered consciousness and frequently without the motor phenomena commonly associated to seizures. In rodents an arrest of movement with loss of responsiveness to external stimuli occurs [Dube et al., 2009; Moreno and Furtner, 2009]. Febrile seizures have at least some resemblance to the phenomenon known as "work stoppage" that is the cessation of the ongoing behavior by exposure to microwaves with enough power density to increase core temperature by more than 1 °C. However it is unknown whether the similarity also extends to their causal mechanisms. The mechanisms of work stoppage are poorly understood and is uncertain whether the cause is an integral whole body effect or due to the heating of an specific brain locus [D'Andrea et al., 2003]. On the other side, grand mal seizures and death can be caused by RF when the brain temperature increases by several degrees [Guy and Chou, 1982] thus a continuum between these two endpoints might exist. To our best knowledge there are no studies in other species than rodents to evaluate hyperthermic seizure induction with accurate intracerebral temperature measurements. In dogs and cats the threshold to produce histopathological alterations are CEM 43 °C of 112 and 187 s respectively [Britt et al., 1983; Dewhirst et al., 2003; Lyons et al., 1986]. However other authors report no effects after a CEM 43 °C of 900 s (15 min) in dogs and rabbits [Silberman et al., 1986; Takahashi et al., 1999]. Brain damage from low thermal doses tends to occur with delayed onset [Britt et al., 1983; Lyons et al., 1986; Sharma, 2006; Sinigaglia-Coimbra et al., 2002; Ullal et al., 1996]. There is also a strong synergism between hyperthermia and ischemia to cause brain damage [den Hertog et al., 2007; Linares and Mayer, 2009; Zaremba, 2004]

Sensitivity of other tissues

Other tissues that can be considered highly sensitive are rabbit cornea, mouse testis, dog's urethra and liver, mouse's sclera, choroid and lens, mouse bone marrow, mouse testis, dog's brain. At the other end, one of the most thermally resistant tissues is pig's skin [Dewhirst et al., 2003]. No obvious difference in overall thermal sensitivity of different species is observed but there may be differences for specific tissues. In table I, the lowest CEM 43 °C values reported to produce damaging endpoints in several tissues and species is presented. The equivalent temperature for several exposure times was calculated with the CEM 43 °C equation.

Are there additional cellular mechanisms to explain the effect of hyperthermia on brain tissues?

The mechanisms of hyperthermic seizures remain elusive [Thomas et al., 2009]. The participation of protein denaturation produced by temperature rise in seizures has not been studied to our best knowledge. A mechanism that has been shown to trigger seizures in hyperthermia is the blood alkalosis that results from the hyperventilation provoked by hyperthermia [Dube et al., 2009; Thomas et al., 2009]. The effect of temperature on the gating rate of ion channels is another mechanism likely to be involved since ion channels regulate neuronal excitability [Thomas et al., 2009]. The degree of sensitivity of ion channels to temperature varies; some ion channels are relatively insensitive to temperature changes around the physiological range and other ion channels are very sensitive. The family of the TRP channels contains a special group of ion channels that function as thermoreceptors, due to its sensitivity to temperature changes. The presence or absence of TRP channels can modify the excitability of neurons at basal core temperatures [Talavera et al., 2008]. Brain sodium channels have also been found to be sensitive to temperature and are thought to contribute to hyperthermic seizures [Thomas et al., 2009]. Depending on the age of the animal, seizure type, and duration, cell loss might occur after seizures. Cell death often result from prolonged or repeated seizures and can occur due to the massive glutamate release and subsequent unregulated increase in intracellular calcium [Holmes, 2002]. These mechanisms are independent of protein denaturation

Basal tissue temperature and thermal sensitivity.

CEM 43 °C concept seems to imply that when exposure time is constant, the thermal damage is determined by the absolute temperature reached. However, there are some exceptions. The temperature rise relative to the resting temperature affect the degree of thermal damage in tissue cultures [Dewhirst et al., 2003], and in embryos the temperature threshold to produce teratogenesis and might be important factor to adverse effects in some new born tissues and adult tissues like the hippocampus at least in some species. In embryos, a temperature rise of 2 -2.5 °C above the normal core temperature for more than 1 h is teratogenic. Embryos from species with relatively low resting temperature like humans (37 °C) can suffer from teratogenesis if exposed to temperatures that are innocuous to embryos of species with higher basal temperature (39-39.5 °C) such as sheep and guinea pig [Edwards et al., 2003]. An analogous situation might occur in some brain regions like the rat hippocampus that seem to be more sensible to deleterious effects of heating as shown above and to have a lower resting temperature (35.5 °C) than the core body [Kiyatkin, 2005; Ullal et al., 1996]. Newborn rats have a modest thermoregulatory capacity [Crawshaw, 1980]. Pup rats exhibit high sensitivity to heat (detailed above) and can have a low core basal temperature (around 33.6 ±0.5 °C) depending on the external environmental conditions [Schuchmann et al., 2006]. In both, adult rat hippocampus and pups, the temperature difference is larger for any hyperthermic exposure than for a tissue with basal temperature of 37 °C. It is unclear whether other rat's brain structures with higher basal temperatures (37.3 ±0.01°C) such as the ventral tegmental area or the hypothalamus [Kiyatkin, 2005] are more resistant to hyperthermia than the hippocampus. An extension of the basal temperature argument to species sensitivity would suggest that animals with higher core temperatures should have higher CEM 43 °C thresholds. However there is still not enough evidence that the lower thresholds for adverse effects in rat brain tissues compared to cat, dog and rabbits are due to their lower core temperatures (see table 1).

Moreover, it might be invalid to extrapolate the CEM 43 °C thresholds to adverse endpoints for structures like the brain from rodents to other species when CEM 43 °C is calculated from core temperature because the local basal temperature of the tissue can be different from the core temperature to an extent that give rise to significant differences in thermal dose, as will be shown in detail in the next section. It is remains unclear when and why the basal tissue temperature affects the thermal sensitivity of tissues. The skin frequently has a basal temperature several degrees below the core temperature due to the temperature gradient with the environment and there is no evidence of higher sensitivity of the skin to heat compared to other tissues [Dewhirst et al., 2003]

Regulation of brain temperature: a factor to be considered in brain thermal dose assessments.

Temperature inhomogeneity

The main thermostat for body temperature control is localized at the base of the brain in the hypothalamus. In humans at rest, the average brain temperature is usually around 37 °C and is generally believed to be stable, homogeneous and tightly regulated. However, there is growing evidence that the brain's temperature is neither homogeneous nor stable in a range of 2-3 °C around the average.

In rats, the dorsal cortex is usually 1 - 1.5 °C cooler than the base of the brain [Andersen and Moser, 1995], and furthermore different brain structures in the rat have their own basal temperature [Kiyatkin, 2005]. A spatial temperature gradient has also been found in neurosurgical patients. A difference of up to 1 °C has been reported between epidural and intraventricular temperature with an average of 0.47 °C, and a difference of 0.8 to 1.3 °C has been reported between subcortical and thalamic temperature [Mellergård, 1995].

In both rats and humans a ventro-dorsal temperature gradient is maintained during temperature increases from exercise and temperature decreases from anesthesia [Andersen and Moser, 1995; Mellergård, 1995].

Brain heat sources

At rest and in normothermic environment conditions, the brain's metabolism is thought to be the main source of heat that contributes to brain temperature. The brain's metabolism is characterized by an intense heat production - it spends 20% of the body's oxygen but represents only 2% of the body mass and almost all this energy is converted to heat since no mechanical work is performed. The heat released by the brain is estimated to be 11 W/kg [Yablonskiy et al., 2000]. The removal of this heat depends on blood perfusion, heat conduction and heat exchange of the body surface with the environment.

Consequently inhomogeneities in thermal properties of tissues may play an important role. In the brain, both heat production and blood flow are on average four times larger in gray matter than in white matter [1] .

Under baseline physiological conditions the human brain temperature of the deep regions is calculated to be 0.3-0.4 °C higher than the temperature of the arterial blood by a theoretical model [Sukstanskii and Yablonskiy, 2006]. Measurements in neurosurgical patients show ventricular temperature is on average 0.33 °C higher than core body temperature (rectal), however periods with temperature differences of 0.5 to 1.0 °C occurred in most of the subjects and the maximum difference observed was 2.3 °C [Mellergård, 1995].

During functional brain activity there are disproportional changes in local blood flow compared with metabolic heat production leading to local temperature changes. This has been studied recently by a theoretical analysis that uses a simplified form of Pennes' equation. The human head and the region of influence of a localized heating (due to local alterations in blood flow) are modeled as spheres. It has been found that brain areas where the blood flow increases with the performance of a task (activated areas) there is an increase in brain temperature and conversely a decrease in the areas where the blood flow decreases (deactivated areas). The change in temperature around the region with altered blood flow can extend to surrounding tissue at rest due to heat conduction, typically for a distance of several millimeters. Outside this area the temperature distribution of the brain remains practically unchanged [Sukstanskii and Yablonskiy, 2006].

Near the brain surface there is heat exchange with the environment and when the ambient temperature is below a certain threshold the brain is cooler than the arterial blood, so the changes in blood flow produce the opposite effect on the local temperature in contrast to deep regions. The blood flow near the brain's surface acts like a heater and produce a "temperature shielding" that prevents the extracranial cold from penetrating deep brain structures. When the ambient temperature exceeds a certain threshold, the blood flow's heating disappears and its effect is inverted [Sukstanskii and Yablonskiy, 2006]. This theoretical model is supported by studies were brain thermometry was done by the 1H magnetic resonance spectroscopy (1H MRS) technique and infrared imaging [Ecker et al., 2002; Gorbach et al., 2003; Yablonskiy et al., 2000].

The temperature distribution in the brain of small animals such as rats is much more influenced by environmental temperature than in larger animals (like humans) because the shielding length (2-4 mm) is almost the same as the brain's radius (5 mm). Furthermore, small animals lose more heat to the surrounding environment than larger animals because of a higher surface to volume ratio (10Ã- higher than in adult humans). In the rat brain, this greater heat loss is counteracted by a higher cerebral blood flow which better equilibrates temperature between body and brain and is thought to mainly account for the ¾1 °C lower temperature found in deep brain compared to core temperature. This means that blood flow in small animals is a source of heat like in the superficial brain of bigger animals rather a heat sink as it is usually the case in deeper regions [Zhu et al., 2006]. This implies that whole body heating would produce a heating pattern in the brain that is different between small animals and large organisms like humans

Further support for the notion of blood flow in rodents as a heat source is provided by rat experiments where intense and prolonged stimulation of the hippocampus failed to produce an increase in the temperature of more than 0.6 °C. Widespread activity during paradoxical sleep produced increases of less than 0.3 °C and thus seems to rule out the brain's metabolic heat as the origin of the 1.5 - 2 °C temperature gradient existent in rat's brain, suggesting the basal arteries at the base of the brain are the main heat source [Andersen and Moser, 1995; Moser and Mathiesen, 1996].

However there is contradictory data that brain temperature in rats is significantly higher than the arterial blood under basal conditions and in an emotionally induced hyperthermia [Kiyatkin, 2005]. In animals and humans core temperature recordings show that stressful, emotional or arousing stimuli produce hyperthermia [Briese, 1995; Moltz, 1993]. Arousing stimuli in rats (cage transfer, tail-pinch, social interaction) produce a temperature increase in all of the brain structures measured and in the arterial blood for a period exceeding the stimulation time. The changes in temperature in each brain structure were faster (7-14 s in brain and 20-40 s in arterial blood) and stronger in amplitude than in arterial blood, indicating the intra-brain heat production is the primary cause of this kind of hyperthermia and that blood circulation removes heat more strongly than it delivers heat to brain.

In cage transferred rats (environmental change), the effect is higher, with a brain temperature increase of ¾1.8 °C lasting 2-4 h. [Kiyatkin, 2005; Kiyatkin et al., 2002]. Under physical activity in a thermoneutral environment, heat production by the muscles elevates the core and arterial blood temperature and during the first 10-15 min there is a reduction between the venous and arterial temperature across the brain producing heat storage in the brain and an increase of ¾1 °C in the brain average temperature.

Brain heat dissipation during hyperthermia

In a heat stressing environment, the physical activity keeps the body core and arterial blood temperature increasing and there is a reduction in cerebral blow flow and an increase in the cerebral metabolic rate. In humans at rest and during exercise, brain temperature is higher than in the trunk [Nybo, 2007], in contrast other animals have a special vascular structure called the carotid rete that cools the arterial blood before it enters the brain, thereby protecting against thermal damage under exertional (physical activity) or environmental hyperthermia that may otherwise be lethal. Animals with carotid retes include antelopes, camels, goats, sheep, felids, goat and dog [Baker, 1982; Caputa, 2004; Jessen, 2001]. In the case of dogs at rest, the brain is warmer than arterial blood but when they run the sharp rise in arterial blood temperature is paired with a brain temperature drop of 1.3 °C below carotid temperature due to their rudimentary carotid rete [Baker and Chapman, 1977].

Although humans don't have a carotid rete, other special mechanisms that selectively cool the brain during hyperthermia have been proposed to exist. These mechanisms include a higher sweat capacity compared with the rest of the body, and evaporative heat loss from the upper airways that dissipate 125-175 and 100 W respectively [Cabanac, 1993].

However, the existence of these brain cooling mechanisms remains controversial because surrogate measurements in other sites than the brain have been used in healthy subjects under hyperthermia and when brain measurements have been performed, they are in patients generally without a thermal challenge and with aresponsive capacity that may be compromised due to elevated intracranial pressure and severe circulatory insufficiency [Caputa, 2004; Nybo, 2007; Simon, 2007]. However, at least one study in humans, subdural temperature measurements show a selective cooling by the upper respiratory tract that contributes to a drop of 0.4-0.8 °C on the basal part of frontal lobes under mild hyperthermia [Mariak et al., 1999].


Exposure to high level RF fields can lead to significant absorption of energy and consequent temperature increases. A main objective of RF safety standards and guidelines is to prevent excessive temperature rise in tissue to avoid thermal damage which is highly dependent on temperature level and exposure time as implied by the thermal dose concept, CEM 43 °C. One of the present shortcomings in the derivation of RF exposure limits is that only temperature rise, but not exposure time, is considered. The thermal dose has been well studied in the temperature range between 41-57 °C, but it is still uncertain whether thermal doses are reliable enough to predict damage endpoints at temperatures below 40 °C [Dewhirst et al., 2003], which is the more likely range for RF exposures. This weakness in CEM 43 °C to assess RF exposures is due to four factors. 1) The variability in the R values found at temperatures below 43 °C. 2) It is not clear where the breakpoint is in human tissues [Dewhirst et al., 2003]. The combined uncertainty of the R value and breakpoint below 40 °C can produce a thermal dose inaccuracy of more than four orders of magnitude in difference. The more conservative approach is to assume a breakpoint of 43 °C and an R = 0.25 below the breakpoint as discussed earlier (see figure 1). 3) Estimation of temperature is frequently inaccurate when the core temperature as measured from the colon is assumed to be representative of the tissues in question. From the thermal dose point of view, variations in temperature of 2 °C are important because that translates to 16 fold difference in CEM 43 °C for any fixed exposure time. Thus temperature gradients should always be taken into account to calculate CEM43. In an important proportion of the literature temperature gradients are neglected. 4) It remains undetermined the lowest temperature to produce a given endpoint. CEM 43 °C predicts damage to occur at temperatures in the normothermic range or below if enough exposure time is allowed. The issue is more evident in endpoints with very low thermal dose thresholds (see table 1). For example seizures in rats occur with a CEM 43 °C of few seconds, however the effect occurs only when the core temperature is above 40°C or 37 °C measured in the hippocampus (1.5 °C above rat hippocampal basal temperature). It is clear then that CEM 43 °C thresholds alone are not enough to assess tissue sensitivity and in vivo data doesn't seem to support an infinite extrapolation of the CEM 43 curve toward the lower temperature region.

Another factor to be considered is temperature rise time. In general, temperature increase from RF thermal (SAR) loads does not lead to a sudden increase in temperature, but rather a gradual one. Slow temperature rise produces some degree of thermoresistency in the tissues and reduces the damage that can occur compared with the same thermal dose with a faster temperature rise. Thus a higher thermal dose (by increasing SAR, exposure time or both) may be required to reach the same damage endpoint compared to a fast heating source. At higher RF frequencies, thermoresistence might not occur because the energy is absorbed in relatively small volumes of tissue near the surface of the body and heating can be quite rapid. A further evaluation may be required on the effect of RF frequency on thermal damage on superficial and highly sensitive tissues such as those in the eyes.

There is some evidence that thresholds are the same between diverse species when the same tissue is compared at the same endpoint. Accordingly it is believed that the order of tissue sensitivity is likely to be similar across species. The tissues considered more sensitive are brain, eyes and testis, with the threshold to damage reported to occur at CEM 43 °C below 30 min and the most resistant tissues are muscles, and fat with CEM 43 °C above 80 min. The remaining tissues have CEM 43 °C thresholds between these range of values [Dewhirst et al., 2003]. However the information available is sparse. For many tissues very dissimilar thresholds for damage have been reported and remain unknown whether the cause is the endpoints used, or differences across species. For a given tissue and species, there can be endpoints with more than 1000 fold difference in thermal dose, depending on the degree/kind of damage selected and the time of assessment after exposure. The lowest harmful doses usually occur with low damage levels and delayed assessment time (i.e., days or weeks). A related problem in classifying thermal sensitivity of different tissues is reconciling the different endpoints used. In some cases the endpoints are tissue specific making the comparison complicated. It has also been the case that different tissues have been tested in just one of the several species but not in others precluding a straightforward evaluation.

Another aspect largely unexplored in vivo is the relationship between the degree of damage and thermal dose. In some cases, small increases in damage may require large increases in thermal dose, for example 1.5% damage of fat tissue in pigs requires a CEM 43 °C = 80 minutes but 11% damage requires a CEM 43 °C = 1280 minutes, a 16 fold difference in thermal dose [Adams et al., 1985]. Such degree of differences between endpoints is not unexpected since thermal dose increases exponentially with temperature. However in other cases the change in degree of damage is much steeper. For example the heating of the spinal cord in mice at 42.3 °C for 75 min (CEM 43 °C = 28) produces only minor neurological problems but the same temperature for 92 min (CEM 43 °C = 34.9) is lethal in 50% of animals [Sminia et al., 1987]. The level of damage is an essential aspect because for tissues with higher capacity to recover and resistance to functional disruption, low level damage is less important. On the other hand, tissues like brain have less capacity to recover from damage and thermally induced lesions tend to be more permanent and accumulate (e.g. cataracts) and hence low level damage is important. The relationship between the degree of damage in vivo and thermal dose is required to develop appropriate safety factors to any temperature-based RF restriction.

Besides thresholds for damage, the determination of tissue sensitivity also requires consideration of the endpoints used. No quantitative method has been developed to allow integrating thresholds, endpoints, capacity of the tissue to carry out its functions and recover from a given level of damage in a single value. Consequently current classification of sensitivity of tissues to hyperthermia based on CEM 43 °C contains some degree of subjectivity. However a clear difference can be seen in those tissues in the extremes of thermal dose thresholds and recovery capacity.

Tissue damage has been reported to occur only when there is a temperature rise of more than 2°C above normothermia (37°C for humans) and generally after prolonged exposures for the most sensitive tissues, including brain/BBB, some eye tissues, testis, urethra, spleen, spinal cord and others (see table 1). Other deleterious effects like seizures or epileptogenic activity can have a short exposure onset (and thus remarkable low CEM 43 °C) but they usually require also a temperature rise of more than 2°C [Dube et al., 2006; Dube et al., 2007; Gonzalez-Ramirez et al., 2005; Kwak et al., 2008; Schuchmann et al., 2008; Ullal et al., 1996]. In humans seizures seem to require a higher CEM 43 °C compared to rats, unfortunately no accurate brain thermometry data is available. However, most people with high temperature don't suffer seizures. To our best knowledge, no tissue damage has been reported to occur with temperature increases of 1 °C or less and such possibility seems very unlikely even for the most sensitive tissues. However temperature increases of 1 °C or less might exert transient effects. For example, immediate transitory changes in neurophysiologic function occur with small increases in temperature (< 1 °C ) with no tissue damage [Acar et al., 2009; Tryba and Ramirez, 2004]. These effects can be explained at least partially thought the presence of temperature sensitive channels like TRPs and some Na channels. They seem involved in the modulation of neuronal excitability and might be part of the normal neurophysiology, as is suggested by the participation of TRP channels for instance growth of neuronal projections although their function remain largely unknown [Talavera et al., 2008]. Temperature fluctuations in the physiological range brain may play an important role in brain function because of its effects on the kinetics of biochemical reactions and on receptor geometry, building up of receptor clusters in the membrane, and alterations in protein expression [Agnati et al., 2005]. Thus the 10 W/kg SAR limit for localized RF exposure that produce temperature increases of this magnitude [ICNIRP, 1998; IEEE, 2006; Vecchia et al., 2009] provides a wide safety margin for thermal damage.

There is evidence that sensitivity to thermal damage depends in some tissues of the tissue basal temperature that can be lower than core temperature like in embryos and probably the rat hippocampus. In the cases where this results true, it is invalid to apply temperature and CEM43 safety limits obtained from other species with a higher basal tissue temperature or from temperature estimations that not take into account the existence of temperature gradients across the tissues. Currently there are not satisfactory explanations of the sensitivity of some tissues to damage at temperatures near 37°C. An in vitro study with Chinese hamster cells suggests that thermal damage can occur even at physiological temperatures (37 °C). A 0.2% cell loss per hour was found when the cell generation time (the time it takes for a parent cell to divide into two daughter cells) was compared to the population doubling time. The cell loss at 37 °C coincide with the extrapolation of cell loss Arrhenius curve at higher temperatures and thus it can be speculated that the spontaneous cell loss that occurs in all biological systems might be due to it [Johnson and Pavelec, 1972a; Johnson and Pavelec, 1972b] At 37°C most cells seem to able to endure or repair the damage, and maintain an steady state so damage is not evident. It follows that interference with endurance or repair mechanisms will increase the apparent damage. Exposure of cells to compounds and conditions that exert a proteotoxic stress, has shown that measurable protein denaturation can occur at 37 °C [Lepock, 2003] and thus support the notion of a baseline level of damage that is constantly being repaired. Cell's endurance and repair might be the reason that most tissues doesn't seem to be damaged in tissues in vivo at 37°C. More research is required to determine whether the tissues with higher temperature sensitivity have reduced resources for endurance and repair.

In environmental or physiological conditions that compromise the body's capacity to dissipate heat an additional thermal source may lead to heat buildup in tissues. A 1-2 °C rise in local temperature resulting from environmental loads such as RF energy will add to previous thermal stress and may be relatively more harmful and consequently require larger safety margins. This may happen in very warm and humid environments and can be worsened by hard physical activity. The same can be said about clinical or physiological conditions that impairs the body's thermoregulatory mechanisms. Populations like the elderly have a high risk for dysfunction of the thermoregulatory system (i.e. decrease in workload capacity of the heart, decrease in peripheral blood by reduced vascularity, reduction in the number of sweat glands and sweat gland response flow). The body's ability to respond to extreme heat is compromised further in conditions as hypertension, atherosclerosis, and heart failure [Worfolk, 2000]. An example is the capacity of high intensity RF exposure to exacerbate temperature raise in monkeys with fever [Adair et al., 1997]. Even when these factors have been recognized by safety standards, it may be necessary to quantify the impact of these factors in safety margins in worse case scenarios that are likely to occur.

RF safety standards already contain an implicit temperature rise limit of 1 °C embedded in their SAR restrictions andliterature shows that for tissue damage by local exposures it is a very conservative limit. That fact encourages the development of a thermally based RF standard. However no enough knowledge is available to establish a thermally based safety standard.Thermally based RF limits require thatthe uncertainties around CEM 43 are solved and enough data is available to circumvent the constrains mentioned before. In the meantime SAR limits would benefit if they aim to have temperature and CEM43 defined endpoints to support SAR limits. Compliance issues add to the list of problems associated to thermally based limits……….

In Summary:

1) Both temperature and exposure time are required to predict thermal tissue damage (i.e CEM 43 and the minimum temperature where CEM 43 can predict damage for a particular endpoint). 2) The published CEM43 °C data below 43 °C is generally too sparse and uncertain to be reliably used for the generally weak heating induced by RF EMF exposures; however thermal damage seems to require at least 2-3 °C increase but there is no evidence of damage with temperature increases of 1 °C or less; 3) Differential sensitivity of tissues argues for the use of more tissue-specific thermal or SAR limits; 4) There are substantial problems in developing compliance tests for temperature based limits; 5) Temperature based limits are feasible and potentially better than RF-EMF limits, but will require more research.