6.1. Background: Neuroimaging and Brain Connectivity
One of the major challenges in neuroscience has been to connect the anatomy of the brain to its functions. Since the late 1990s, neuroimaging has become the predominant technique to study the functional architecture of the brain. Many different neuroimaging methods exist; however, functional magnetic resonance imaging (fMRI) and electroencephalogram (EEG) are the predominant methods used in relevant studies, and so these two are briefly described below.
Changes in oxygen usage and regional cerebral blood flow mirror the degree of activity of regional neurons both during task performance and in the resting state [83
]. The fMRI technique uses the magnetic properties of oxygenated and deoxygenated hemoglobin to create images of changing blood flow in the brain associated with neural activity [84
]. These signal changes are referred to as blood oxygenation level-dependent (BOLD) contrast [83
]. Depending on the clinical or scientific question, fMRI images can be used to investigate differences between clinical groups or to relate brain responses to environmental or contextual changes. The main limitations of fMRI are that it detects neural activity indirectly through the associated hemodynamic variations, and the general noisiness of the signals [84
The EEG is a mostly non-invasive electrophysiological monitoring method that records the spontaneous electrical activity of the brain over time. Through multiple electrodes placed on the scalp, the EEG measures voltage fluctuations resulting from ionic current within the firing neurons of the brain [86
]. Compared to fMRI, EEG is a much cheaper and simpler method for brain activity measurement, and it is widely used to diagnose epilepsy and other brain disorders. The EEG is also a popular tool in neuroscience, cognitive science, cognitive psychology, neurolinguistics, and psychophysiological research. However, one of the big disadvantages of the EEG is that it is hard to determine the exact location in the brain from which the electrical activity originates. Combining the EEG-fMRI techniques has been suggested to be beneficial for obtaining the most informative results [84
Generally, neuroimaging studies are designed to investigate one of the two aspects of brain functions: localization or connectivity. When investigating localization, the goal of the study is to identify a functionally specialized region of the brain that is activated in response to a specific input or behavior. However, as most of the brain’s functions are based on coordinated interactions of large numbers of neurons distributed across different brain areas, the second concept, connectivity, is of increasing interest for both psychiatrists and neuroscientists. Connectivity analyses investigate the way in which brain regions communicate with one another [87
]. The methods that are used for studying brain connectivity largely depend on the type of connectivity of interest. Structural connectivity analyses typically concentrate on identifying fiber bundles connecting different regions of the brain by using diffusion magnetic resonance imaging (MRI). Structural connectivity often reflects the functional connectivity [88
], which is defined as a solely statistical dependency among remote neurophysiological events. Functional connectivity is typically inferred by the temporal correlation between fMRI BOLD or EEG signals acquired from separate regions of the brain. Effective connectivity refers to causal dynamics between different brain nodes (the influence one node exert over another), observed by using either BOLD or EEG signals [89
“Brain Network” is an important concept in exploratory neuroimaging studies. Brain Networks are typically defined as brain regions that show functional connectivity (statistically correlated BOLD or other signal fluctuations detected by fMRI or EEG) under specific conditions. Brain imaging studies often concentrate on detecting certain context- or environment-related changes in specific brain networks. Many different brain networks have been identified. For a more detailed review, the reader is referred to Wig 2017 [90
Functional connectivity analyses are increasingly used to study brain connectivity patterns in both clinical and research contexts. Depending on the research goals, different methodological approaches are used. In many studies, the researchers use carefully designed behavioral tasks that the subjects perform during the brain scanning process. However, since a number of clinical populations are unable to perform tasks for a variety of reasons but can still be studied at rest, the resting state fMRI (rs-fMRI) has emerged as a valuable method for studying brain connectivity. In rs-fMRI studies, participants are not required to perform any motor or cognitive task or to pay attention to any particular stimulus, but are instead instructed to clear their minds and not to engage in specific thoughts or visual images [83
]. Rs-fMRI functional connectivity analyses thus revolve around detecting temporally correlated BOLD signals from distinct regions of the brain in the absence of any cognitive task [83
]. Alterations in resting-state functional connectivity were shown to be associated with a wide variety of brain and behavioral disorders, such as attention deficit hyperactivity disorder (ADHD) [91
], Alzheimer’s disease [92
], obsessive compulsive disorder [93
], depression [94
], bipolar disorder [95
], and schizophrenia [96
]. Furthermore, functional connectivity analyses have revealed a discoordination in task-related networks in patients with generalized anxiety disorder [97
], social anxiety disorder in adolescents [98
] and adults [99
], as well as specific phobia [100
], pointing to a dysregulation of the fronto-amygdalar interplay [101
6.2. Altered Brain Function in Hot Conditions–Evidence from Neuroimaging Studies
In 2013, Liu et al. [102
] studied the effects of passive heat exposure on the neural mechanisms of the human attention network. The human attention network is thought to be composed of three separate yet intimately interacting networks that control the alerting, orienting, and executive functions [103
]. The alerting network maintains a state of high sensitivity during task performance over a short period of time. It engages the frontal and parietal brain regions and thalamus and is thought to be regulated by the norepinephrine system [102
]. The orienting network entails selective adjustments that direct attention toward a particular modality, spatial location, or stimulus feature [103
]. For example, the orienting network allows attention to be disengaged from one focus and to be re-engaged on a new pathway of interest. Activation of prefrontal and parietal areas of the brain, including the superior parietal cortex and temporal parietal junction, underlies the orienting function. The orienting network is believed to be regulated by the cholinergic (acetylcholine-transmitting) systems in the basal forebrain [102
]. The executive control network is associated with monitoring for and resolving conflicting mental states. This network activates the frontal lobe, especially in the midline frontal areas, such as the anterior cingulate cortex (associated with monitoring), and the lateral prefrontal cortex (associated with inhibiting irrelevant responses). Dopamine binding in the anterior cingulate cortex is thought to drive the executive function [102
In the study by Liu et al. [102
], the function of all three networks comprising the attention network was investigated during cognitive performance and under either temperate (25 °C) or hot (50 °C) conditions. The performance results suggested that only executive function was impaired by the heat exposure, whereas orienting and alerting were unaffected. However, the fMRI analyses showed that all three of the networks were significantly affected in hot conditions.
The significant impairment of executive function during the performance test suggests that the loss of being able to handle cognitive conflicts may be the most noticeable behavioral change in hyperthermic conditions. The authors suggested that the physiological pathway possibly explaining this impairment is mediated by the body temperature-induced increase in plasma serotonin levels, which results in the inhibition of central dopamine production. As dopamine is believed to regulate the activation of the executive brain network, such inhibition may lead to impaired executive function. However, a the orienting and alerting functions were clearly affected by the heat exposure in the fMRI analysis but not in the performance tests, the authors concluded that a compensatory pathway involving other brain regions may become activated under stressful conditions, such as heat exposure.
Since neuronal synchrony and connectivity is impaired in several mental disorders, such compensatory activation of various brain regions may not be possible in this group of patients and may therefore increase their vulnerability during prolonged heatwaves. Furthermore, as the attention network, and especially the executive function, has been found to be deficient in several groups of mental health patients, such as sufferers of depression [105
] and ADHD [106
], heat exposure may further burden an already weakened function.
Shibasaki et al. [107
] assessed the effects of heat stress on the cognitive processing of human motor control of executive and inhibitory responses. These responses were tested with a somatosensory Go/No-Go task, in which the participant received cutaneous “Go” and “No-go” signals and was asked to respond by pushing a button as quickly as possible, but only after presentation of the Go signal. By using EEG recordings, they demonstrated that the neural activities of both response execution and response inhibition were decreased during heat exposure. Whereas facial cooling quickly returned cerebral blood flow and the self-reported thermal comfort to baseline levels, the executive and inhibitory processing remained reduced for a longer period. Whole body cooling recovered executive processing, whereas inhibitory processing remained reduced. These results suggest that heat stress likely affects cognitive processing in the brain even after thermal comfort is regained.
Sun et al. [82
] studied the effects of passive heat exposure on human brain functional connectivity patterns. Two different kinds of analysis methods, seed-based and node-based, were used to process the data from fMRI resting state functional connectivity analyses (rs-fMRI). The difference between these methods is as follows. In the seed-based correlation analyses, one region of interest (ROI) is defined in the brain, and a map, describing the strength of functional connectivity between the ROI and the rest of the brain regions, is obtained [87
]. In the node-based analyses, the entire brain, or the entire explored area, is divided into “nodes”, i.e., regions that are considered to be functionally homogeneous, and the connectivity between the nodes is measured by pair-wise correlations.
The posterior cingulate cortex and precuneus (PCC/PCu) was defined as the ROI in the seed-based analyses by Sun et al. [64
]. This area was chosen mainly for being a key metabolic area in the human default mode network (DMN), which is a resting state network consisting of regions that typically show slower activity during task performance than during the state of rest [88
]. This area was also chosen, in part, because an abnormal PCC/PCu connectivity has been previously found to be associated with several neuropsychiatric disorders [91
]. In the node-based connectivity analyses, the brain was divided into 90 separate regions and the correlation strength between each pair of these was estimated. Similar to the [102
], the temperature in the heat-exposed group was kept constant at 50 °C and at 25 °C in the control group.
The results from the seed-based analysis showed a general trend toward significantly decreased connectivity between the PCC/PCu region and a number of brain areas when exposed to heat, such as the medial orbitofrontal cortex (mOFC) and medial temporal cortex [82
]. mOFC is a region in the frontal lobes that is thought to be involved in self-evaluation and decision-making, including stimulus-reward associations and reinforcement of behavior [108
]. One of the few increased connectivities in the seed-based analysis, detected between the PCC/PCu and the bilateral orbital superior frontal gyrus (associated with self-awareness), was suggested to be a compensatory reallocation of the mOFC cognitive resources [82
]. Regions in the partial medial temporal lobe (especially the hippocampus and parahippocampal gyrus) are essential for normal human learning, memory, and spatial location. According to several earlier reports, both learning and memory can be impaired by heat-exposure in the performance tests [110
]. The findings of Sun et al. regarding the heat-induced decrease in the connectivity between the ROI and the regions in partial medial temporal lobe thus provided neuroimaging evidence for previous cognition studies [82
]. In addition, the within-region connectivity of the ROI itself was affected by heat, suggesting abnormalities in the function of this region during hyperthermia. This finding may explain some previous evidence from behavioral studies suggesting impaired self-referential cognitive ability during hyperthermia [112
In the node-based analysis, 65 disturbed functional connectivities, 50 of which decreased (77%) and 15 increased, were found in the heat-exposed conditions compared with the temperate conditions [82
]. Similar to the seed-based analysis, the most heat-affected region was the mOFC. In addition, heat exposure decreased the connectivity patterns in the temporal and occipital lobes. Increased connectivity in association with heat exposure was mainly located within the limbic system and was suggested to be related to the sensation of thermal discomfort and activation of thermoregulatory mechanisms [82
]. Although the physiological and psychological consequences of the detected connectivity disturbances are still a matter of speculation, heat exposure clearly affects both inter- and intra-regional information processing and transmission, and thereby influences brain function and cognitive behavior through many different pathways [82
The findings of Sun et al. are interesting from several points of view. Firstly, by demonstrating altered brain connectivity patterns associated with many cognitive functions in healthy individuals in hot environments, they provided neurological evidence for earlier findings from performance tests. Secondly, they provide a basis for speculation about the nature of vulnerability in mental health patients, as several of the psychiatric disorders entail disrupted brain connectivity without being exposed to heat. For example, decreased connectivity between PCC/PCu and mOFC, similar to what was reported by Sun et al., during heat-exposure in healthy individuals, has been previously observed in individuals with social anxiety disorder under unmanipulated temperature conditions [101
]. Furthermore, lower connectivity between PCC/PCu and the medial temporal lobe, also observed by Sun et al., has been described in individuals with Alzheimer’s disease [92
]. Adding the burden of high temperatures to already disrupted neurological functions may have a deleterious effect on the capacity to execute appropriate behavioral and cognitive responses, and may therefore influence the morbidity and mortality of mental health patients during heat waves.
Further evidence for heat-related changes in central neurological networks was reported by Qian et al. [113
]. In this study, the influence of passive heat exposure (50 °C) on human brain functional network topological patterns was examined using a node-based method called the graph theory analysis. The ‘graph’ in this case expresses the network matrix of nodal connections. As the measures of this analysis method contained several specific terms that are most likely not recognized by non-specialists, we will describe the most common estimates bellow.
Minimum path length is a central characteristic in graph theory analysis that expresses the smallest number of nodal connectivity needed to travel from node A to B. The clustering coefficient of a node is a measure of the number of other nodes that are connected to it, which also are connected to each other. A high clustering coefficient means then that if nodes A and B are both connected to C, they are also likely to be connected to each other [87
]. By using the local estimates of minimum path length and clustering coefficient, the mean path length and mean clustering coefficient can be calculated for the whole graph. Global efficiency, which is inversely proportional to the average minimum path length, is used to express the efficiency of the graph as a whole. Furthermore, small worldness in the brain implies that the nodes are densely connected locally, with a few long-distance connections. This kind of organizational pathway requires low minimum path lengths and has a high clustering coefficient. The small worldness of a network is estimated by comparing the small-world average minimum path length and average clustering coefficient with a randomly connected network [87
Qian et al. [93
] found that brain topological patterns differed between the control and heat-exposed groups. The clustering coefficients and small-worldness significantly decreased in the heat-exposed group, implying lower functional network connectivity. In addition, the network efficiency analyses revealed significantly decreased local and global efficiency in hypothermic conditions, entailing lower efficiency in transmitting information in both local and global level brain networks. Altogether, these alterations might result in brain functional disorders and cognitive performance decline [113
]. Furthermore, the decrease in clustering coefficient, lower local efficiency, but maintained shortest path length suggest that functional networks during hyperthermia shift from an efficient small-worldness mode to a more randomized network with a less optimal topological organization. Disruptions in the topological organization and reduced small-worldness have also been associated with, for example, schizophrenic [114
] and bipolar disorders [115
Altered distribution in neural activity during heat stress was furthermore supported by a study investigating the effects of environmental heat on patterns of cerebral blood flow [116
]. As discussed above, one of the central adaptive physiological responses to heat is peripheral vasodilation, resulting in markedly increased skin blood flow. However, this process was speculated to lead to a competition of blood supply between the periphery and the core (the latter including the brain), necessitating the prioritization of certain functions and brain regions by central mechanisms. Supporting this theory, several studies reported that the mean blood velocity in the cerebral artery decreases during heat exposure [107
]. However, other (mostly older) studies have provided evidence that the cerebral circulatory system has an inherent ability to maintain cerebral blood flow (CBF) stable, and is therefore able to meet normal neural activity demands during environmental and physiological variation [120
]. Qian et al. [116
] quantified the resting state CBF under hot and temperate conditions (50 °C versus 25 °C) by using arterial spin labeling (ASL) MRI. ASL studies require that arterial blood water is first magnetically labeled (for example, by applying a radiofrequency inversion pulse) and then imaged by MR. After a 60 min exposure, Qian et al. [96
] detected a non-significant tendency toward decreased general CBF in the heat-exposed group. In addition, they found significant regional blood redistribution in several brain regions, including the prefrontal cortex, somatosensory areas, and the limbic system in the heat-stressed individuals. Induced by increased body temperatures, the blood flow increased in the dorsal hypothalamus, posterior cingulate gyrus, and right middle cingulate, reflecting an increased regional blood demand for central thermoregulation activity, and a psychological sensation of discomfort in heat-exposed persons [116
]. At the same time, decreased CBF in parts of the limbic system (such as, the parahippocampal gyrus, hippocampus, and amygdala) were mirrored by altered mood states, such as increased nervousness and anger detected during performance tests. This study provided evidence that the reorganization of neuronal activity in the brain during heat stress is most likely adaptive and based on prioritizing the most important task for the moment. However, if the heat exposure becomes more long-term, the resulting changes in cognitive and mood function may possibly increase the vulnerability of persons with already weakened mental health.
The above studies provided evidence for altered brain activity and connectivity during heat exposure. However, as the experimental procedures entail only acute effects caused by short periods of rather high temperature exposure, they are unlikely to fully capture the effects of long-lasting heat waves. As the brain dysfunctions caused by heat in many cases seem to be similar to those caused by different mental diseases, we can suspect that heat may, at certain cases, have an exacerbating effect on disease symptoms.