A Functional Near-Infrared Spectroscopy Study on the Cortical Haemodynamic Responses During the Maastricht Acute Stress Test

In order to better understand stress responses, neuroimaging studies have investigated the underlying neural correlates of stress. Amongst other brain regions, they highlight the involvement of the prefrontal cortex. The aim of the present study was to explore haemodynamic changes in the prefrontal cortex during the Maastricht Acute Stress Test (MAST) using mobile functional Near-Infrared Spectroscopy (fNIRS), examining the stress response in an ecological environment. The MAST includes a challenging mental arithmic task and a physically stressful ice-water task. In a between-subject design, participants either performed the MAST or a non-stress control condition. FNIRS data were recorded throughout the test. Additionally, subjective stress ratings, heart rate and salivary cortisol were evaluated, confirming a successful stress induction. The fNIRS data indicated significantly increased neural activity of brain regions of the dorsolateral prefrontal cortex (dlPFC) and the orbitofrontal cortex (OFC) in response to the MAST, compared to the control condition. Furthermore, the mental arithmetic task indicated an increase in neural activity in brain regions of the dlPFC and OFC; whereas the physically stressful hand immersion task indicated a lateral decrease of neural activity in the left dlPFC. The study highlights the potential use of mobile fNIRS in clinical and applied (stress) research.


exploring Stress with Mobile fniRS
In order to increase the ecological validity of the examination of the neural responses, a mobile applicable neuroimaging method was recently introduced. Mobile functional Near-Infrared Spectroscopy (fNIRS) explores brain activity by comparable means like fMRI, indirectly measuring the concentrations of oxygenated, deoxygenated and total haemoglobin. Thereby, mobile fNIRS indirectly quantifies brain activity by means of near-infrared light, which is absorbed or reflected by (de)-oxygenated blood. Signal changes in specific brain regions indicate increased or decreased neural activity of this region during the task or situation in which mobile fNIRS is applied. Compared to other neuroimaging methods such as fMRI, the spatial resolution of mobile fNIRS is rather low and limited to the measurement of cortical, near-surface brain regions (1 to 2 centimetres depth) 25 . However, an important advantage of mobile fNIRS is that it is transportable, allowing to measure brain activity in more ecological valid environments and real-life situations such as in supermarkets (in order to investigate merchandising communication strategies) 26 or clinical settings (in order to monitor patients pain and stress levels) 27,28 . Furthermore, mobile fNIRS is less sensitive to external noise such as head movements or eye blinks of the participant compared to for example an electroencephalogram. Additionally, the application of mobile fNIRS is comparably inexpensive, very easy to handle and does not cause discomfort for the participant 27 .
To date, fNIRS has only been used to investigate the neural response of stress in a limited number of studies. For example, Rosenbaum and colleagues assessed cortical neural activity during the Trier Social Stress Test (TSST) 29 using fNIRS and revealed activity changes in the dorsolateral prefrontal cortex, the inferior frontal gyrus and superior parietal cortex 21 . The same research group replicated the involvement of these brain areas in a second study also applying the TSST to induce stress and additionally revealed that high ruminators showed attenuated responses in the inferior frontal and dorsolateral prefrontal cortex 30 . Furthermore, two studies using a mental arithmetic task with negative feedback to stress participants demonstrated reduced activity in the right prefrontal lobe in the stress condition compared to the control condition 31,32 .
Additionally, fNIRS has also be used to investigate the concept of pain. A study by Yücel et al. 33 compared innocuous and noxious electrical stimuli and revealed differences in signal size and profile of the activation in the primary somatosensory cortex contralateral to the stimulus applied between the two types of stimuli. Another study revealed that noxious electrical stimuli elicit reduced activity in brain regions of the frontal lobe 34 . Furthermore another fNIRS study on pain perception using the cold pressure test revealed that the neural activity of the right frontal lobe was correlated to pain thresholds 35 . The studies highlight that the preliminary research results should encourage researchers to examine the use of (mobile) fNIRS in real-life situations, for example to monitor patients in the operating theatre.
the Maastricht Acute Stress test (MASt). Recently, a novel stress test has been developed, the Maastricht Acute Stress Test (MAST) 18 , a stress test, which combines features of the widely applied Trier Social Stress Test (TSST) 36 and the Cold Pressor Test (CPT) 37 , allowing to measure both dimensions of stress -psychosocial (challenging mental task) and physiological (pain processing) stress. The MAST follows a simple and easy to perform protocol and recent studies have shown its effectiveness and reliability to induce stress 38,39 . In the experimental stress condition of the MAST, participants have to alternate between two phases, which are 45-90 seconds long: insert their hand into 0-4 °C cold water and perform a difficult mathematical task. The non-stress control condition alternates between inserting the hand into 36° warm water and simple upwards counting.
The present study is the first to examine the haemodynamic brain response during the MAST, investigating the involvement of prefrontal brain regions during this task. In order to strengthen the results, this research work also included other stress measurements such as salivary cortisol and subjective stress level ratings with visual analogue scales at three time points (before the MAST (T1), directly after the MAST (T2) and 20 minutes after the MAST (T3)). Furthermore, heart rate was measured during baseline measurements and during the MAST. By including several objective and subjective measurements, we receive a complex picture of the stress response. Additionally, as the MAST includes two different phases, the mentally stressful arithmetic task and the physically stressful hand immersion task, we are able to compare the haemodynamic brain responses during the different stressor types. A between subject design was chosen. One group of participants (N = 19) were randomly assigned to the stress condition and performed the original MAST, whereas the other, non-stress group (N = 21) performed the matched control task, which was considered to be not stressful at all.
In line with recent research 14,15,19,21,31 , we hypothesised haemodynamic changes in the prefrontal cortex in response to the stress task (compared to the non-stress task) measured with mobile fNIRS. Moreover, we expect significant differences in heart rate, salivary cortisol and subjective ratings between the groups, ensuring that the stress induction was successful.
physiological data -heart rate. Paired-sample t-tests for each group showed that the experimental group displayed a significant increase in heart rate values, t(20) = 3.709, p = 0.001, with a mean increase of 5.08 bpm (±5.25), whereas the increase (M = 0.9 ± 3.14 bpm) was non-significant for the control group (p = 0.181) (Fig. 1C). Additionally, an independent t-test with the increase in heart rate as the dependent factor revealed a significant difference between groups, t(36) = 2.939, p = 0.006.
Overall, our control measures indicate a successful stress induction on a subjective, endocrinological and physiological level. www.nature.com/scientificreports www.nature.com/scientificreports/ neural data -prefrontal cortex activity patterns. (Pre-)Processing. Before the actual analysis, the collected fNIRS raw data were pre-processed. A low-pass filter (high frequent filter; higher cut of frequency value was set to 0.2 Hz) was applied in order to control for artefacts that might overshadow the measurement of the expected effects. Raw optical density signals were converted to haemoglobin concentration changes using the modified Beer-Lambert law [40][41][42][43] within the NIRx Software package. The parameters used to compute the haemodynamic states were set as follows: the distance of the first channel was set to three centimetres, the wavelengths were specified to values of 760 and 850 nanometre and the associated pathlength factor was set to 7.25 for the wavelength of 760 nm and 6.38 for the wavelength of 850 nm, in accordance with values reported in literature [44][45][46] . As the oxygenated-haemoglobin (oxy-Hb) signal has been shown to correlate with cerebral blood flow better than the deoxygenated signal 47 , the analysis concentrates on the oxy-Hb signal. It should however be evident that mobile fNIRS, in comparison to other neuroimaging methods, is capable of investigating also other types of signals such as the raw light absorption rate, the deoxygenated as well as total haemoglobin concentrations. The activation map threshold was set to a p-value of p < 0.05.

Results
The results indicate bilateral increased neural activity in brain regions of the prefrontal cortex, more precisely of the dorsolateral prefrontal cortex (dlPFC) and the orbitofrontal cortex (OFC), when participants completed the Maastricht Acute Stress Task (MAST) in comparison to participants who completed the non-stress control task. Moreover, increased neural activity in Furthermore, in order to investigate potential neural activity differences between psychosocial and physical stress, the mental arithmetic task and the hand immersion task were separately contrasted to the control group.
The results indicate a significant difference between the neural activity of the mental arithmetic compared to the hand immersion task for specific prefrontal brain regions. The mental arithmetic task provoked increased neural activity of brain regions of the OFC (bilateral) and the left dlPFC (Fig. 3), whereas the hand immersion task, indicate a decreased neural activity of the left dlPFC (Fig. 4). In more detail, for the arithmetic task significant increased activity was revealed for in Channel 7: t(36) = 3.59, p < 0.001, d = 1.

Discussion
The present research work investigated haemodynamic responses during the MAST, aiming to demonstrate that mobile fNIRS might be a promising tool to measure neural responses related to stress induced by mental stress and/or physical pain.
The results indicate a successful stress induction using the MAST on endocrinological, physiological and subjective stress parameters. More precisely, salivary cortisol levels, heart rate and subjective stress values scales revealed significant differences between the MAST and non-stress group. The MAST group displayed significantly higher cortisol levels after the stress task as well as a higher increase in heart rate values compared to the control group. Additionally, the indicated subjective stress level displayed that the experimental group experienced higher stress levels after the MAST than the control group. The increase in saliva cortisol concentration and heart rate related to stress has been repeatedly shown 36,48,49 . The fact that the two groups differ significantly on the cortisol levels only 20 minutes after the MAST is in line with the well-established latency of cortisol being detectable in saliva 50,51 . www.nature.com/scientificreports www.nature.com/scientificreports/ On a neural level, the analysis of the mobile fNIRS data revealed significantly increased neural activity in bilateral dlPFC and the OFC whilst performing the MAST compared to the non-stress condition. Furthermore, significant differences in activity patterns of the dlPFC in the experimental group when contrasting the haemodynamic changes during the mental arithmetic and hand immersion task were shown. The results are in line with previously conducted fMRI and fNIRS studies, displaying that stress induction is associated with neural activity changes in brain regions of the prefrontal cortex 14,21,31,32,52 . Moreover, the involvement of the OFC for the stress response is also in line with previous studies 53,22 . However, the direction of the neural activity changes in the prefrontal brain regions during stress requires further elaboration. The present study revealed increased neural activity in the dlPFC and the OFC during the MAST, which is in line with a recently conducted fNIRS study that reported increased neural prefrontal cortex activity during the TSST 21 . However, other studies have indicated decreased neural activity of the prefrontal cortex and other brain regions during stressful tasks 31,54 . A meta-analysis by Kogler and colleagues 19 on studies that investigated the neural correlates of stress measured with fMRI and positron emission tomography, also highlights inconsistent findings of neural activity patterns in frontal brain regions during stress tasks. Future research should therefore address this issue by comparing different stressors within one study in order to disentangle the direction of the neural activity and the direction of the activity changes in frontal brain regions during stress.
The results of the present study also indicate that the cold pressure task, which is related to stress but is at the same time also associated to pain 55,56 , leads to reduced neural activity in the left dlPFC. In comparison the mental arithmetic task, revealed an increase in neural activity in brain regions of the dlPFC and OFC. The research finding of decreased neural activity during the cold pressure task is supported by a recent study conducted by Aasted and colleagues 34 , which revealed strong decreased neural activity of prefrontal brain regions when exposed to  www.nature.com/scientificreports www.nature.com/scientificreports/ noxious stimuli. It should, however, be noted that the decreased neural activity in the left dlPFC shows a medium effect size with a marginally significant p-value. Nevertheless, we report the effect as we believe that this effect is meaningful and highlights the issue of sufficient sample sizes in neuroscience studies and the importance of reporting effect sizes 57 . The results of the mental arithmetic task, indicating an increase in neural activity of the dlPFC during the mental arithmetic task in comparison to the control task, are in line with previous research findings, highlighting an increased neural activity during mentally stressful arithmetic tasks 58,59 .
The given results specify that the brain response induced by psychosocial and physiological stress can be distinguished, measuring haemodynamic changes with mobile fNIRS. This provides evidence for the usefulness of mobile fNIRS, also in clinical and applied research. This line of thought seems to be very interesting and the aim for future research should consequently be to validate the presented results in follow-up studies, exploring the prospective use of mobile fNIRS in real-life settings such as surgeries 60 . Thereby, it has been proposed that mobile fNIRS could be a potential tool to measure nociceptive activity in the brain during surgery in order to improve anaesthesia 34,61 . Additionally, the use of mobile fNIRS to investigate psychophysiological underpinnings of depressive symptoms is discussed in a recent commentary by Adorni and colleagues 27 . Our results indicate that stress and pain related haemodynamic patterns can be inexpensively measured with a mobile fNIRS device, which might be of huge advantage for clinical and applied research. However, as research in this area is still in its infancy, more research is needed, investigating the concepts of stress and pain in more depth and with more distinct tasks. As the cold pressure task reflects aspects of stress and pain, it might also be desirable to include an independent procedure, which explicitly measures pain, such as noxious stimuli, in order to compare the overlapping and unique neural underpinnings of stress and pain, respectively. In this respect, as noted in the introduction, studies have shown that stress can lead to an attenuation of pain perception 11 . The increase in cortisol levels in response to stress, which is also shown in the present study, may interact with sympathetic and opioid mechanisms involved in central pain processing 62 . Therefore, a follow-up study with distinct blocks of stress and pain is desirable, as we cannot rule out that the mentally and physically stressful phases influenced each other in the present research work. However, the different neural activity for the two types of stressor presented, encourage further research in this area.
The present study faces some additional limitations, which warrant a comment and which might be improved in follow-up studies. For example, the utilised mobile fNIRS headband only covers prefrontal brain regions, allowing to merely explore neural activity of the prefrontal cortical brain regions. As other studies have shown that parietal brain areas are also involved in the perception of stress 21 , it would be desirable to utilise a mobile fNIRS headcap, covering the whole cortex in future studies. Furthermore, only neural activity changes in cortical brain regions can be detected, as mobile fNIRS is not capable of measuring activity in subcortical brain regions 25 . However, as brain regions such as the amygdala and the hippocampus also play an important role in the processing of stress 6,14,52 -in particular for the emotional evaluation of the stressor -information on how these brain regions are involved during the MAST cannot be explored whilst utilising mobile fNIRS. In regard to the given measurement procedure and handling of artefacts, future research might control for movements artefacts in the fNIRS data (e.g. head/jaw movements) and for possible haemodynamic contamination from superficial layers (e.g. skull and scalp), consulting existing research work 63,64 . Furthermore, a comment regarding the non-stressful control condition of the MAST is necessary. The stress and non-stress condition of the mental phase differ in regard to work load as the stress version comprises a difficult arithmetic task and the non-stress condition contains of simple counting. In this respect, research has shown a relationship between heart rate increase and higher cognitive load 65 . Still, Smeets et al. 18 developed their non-stressful control condition carefully, also considering the issue of work load, and have demonstrated that the MAST induced stress reliably in comparison to the control condition. Furthermore, the here revealed haemodynamic changes in the dlPFC and OFC in response to the MAST fit well to previous stress research 14,21,22,31,53 .
In conclusion, the present study is the first to investigate the neural activity of brain regions of the prefrontal cortex whilst performing the MAST utilising mobile fNIRS. The results indicate that the MAST successfully induced stress responses as significant differences between groups in subjective stress ratings, salivary cortisol levels and heart rate increase occurred. Additionally, on a neural level, the mobile fNIRS data indicated significantly increased neural activity in brain regions of the dlPFC and OFC in response to the stressful MAST. Furthermore, a distinguished neural pattern could be revealed indicating significantly decreased neural activity during the physically stressful hand immersion task in the dlPFC and an increase in neural activity in the dlPFC and the OFC during the mentally stressful arithmetic task. Based on these results, future studies should investigate the potential use of mobile fNIRS in clinical and applied settings, for example to monitor patients' stress levels in surgeries in order to improve patients' clinical outcomes.

Methods
participants. Forty participants (5 male) with a mean age of 23.8 years (SD = 4.1) took part in the study. All participants were right handed, had normal vision and no history of neurological disorder. Participants signed an informed written consent before participating in the study. The conducted experiment was approved by the ethics committee of the Faculty of Mathematics and Natural Sciences of the Heinrich-Heine-University Düsseldorf, Germany. 18 is a 10-minutelong acute stress test that includes two different phases: (1) the Cold Pressor Task (CPT), in which participants are instructed to immerse their dominant hand up to the wrist into ice-cold water (0-4 °C) and (2) a challenging mental arithmetic task, which is similar to the mental arithmetic task of the Trier Social Stress Test (TSST) 36 .

Material. Maastricht Acute Stress Test (MAST). The Maastricht Acute Stress Test (MAST)
Participants were asked to sit in front of a computer and instructions were given on screen.
www.nature.com/scientificreports www.nature.com/scientificreports/ In the experimental stress condition, participants were informed to complete multiple trials that differ in their duration and in which they have to immerse their dominant hand into ice-cold water or engage in the mental arithmetic task, which consisted of counting backwards starting at 2043 in steps of 17 as fast and accurate as possible. Negative feedback was given by the experimenter on the accuracy and/or speed of the calculations. Each time participants made a mistake, they were instructed to start again at 2043. Participants could not predict the duration of each trail and furthermore, to increase the stress response, they were told that their performance is recorded on video.
In the non-stress control condition, participants were also informed to complete multiple trials, which differ in their duration. However, in comparison to the experimental condition, in the control condition participants were asked to immerse their dominant hand into (body)warm water (36 °C) and were asked to count forwards slowly from 1 to 20. Participants were told to continue counting until the computer gives the signal to start the next hand immersion trail. No feedback was given by the experimenter.
In reality, the duration of all trails was fixed with the same standardised protocol used for all participants (see Participants were asked to keep their head as still as possible during the MAST and not to talk to the experimenter. Please see Smeets et al. 18 for more information on the MAST. MAST group allocation (stressful experimental group vs. non-stressful control group) was defined as the independent variable. physiological data -heart rate. In order to explore the physiological reaction to the MAST, heart rate was assessed with a NEXUS -4 device (Hasomed, Magdeburg, Germany) and a Blood Volume Pulse Sensor, which was placed on the index finger of the non-dominant hand. Heart rate was recorded throughout the MAST, as well as during the baseline period. Mean heart rate values were calculated from every participant for the baseline period (3 minutes) and the MAST (10 minutes).
Subjective data -subjective stress rating. To evaluate the subjective stress level of the participants, a 10 cm long visual analogue scale was used. The scale ranged from "no stress" (zero point) to "maximal stress" (at 100 mm) and participants marked their stress level with a cross on the provided line in response to the question "How stressed are you just now?" Participants were asked to indicate their stress level at three time points: before the MAST, directly after the MAST and 20 minutes after the MAST. The analogue scale was used to prevent memory effects or other subjective rating biases (e.g. anchoring bias). Stress levels were assessed by measuring the distance from 0 ("no stress"), with higher numbers indicating higher stress levels. endocrinological data -salivary cortisol. In order to assess salivary cortisol, saliva sample were taken using salivettes from Sarstedt (Sarstedt, Germany) and stored at −20 °C until further analysis. For the saliva samples participants were asked to insalivate a cotton swab thoroughly. Saliva samples were taken at three time points: before the MAST, directly after the MAST and 20 minutes after the MAST. Saliva samples were analysed as described elsewhere 66 . Salivary cortisol is an indirect marker of the activation of the hypothalamic-pituitary-adrenal axis. neural data -neural activation patterns. Data collection. Optical signals were recorded on a two-wavelength (760 and 850 nm) continuous-wave fNIRSport-System (NIRx Medical Technologies, Berlin, Germany, http://nirx.net). Data was collected from detectors in parallel at a sampling rate of 7.81 Hz. The optical channels were comprised of eight sources and eight detectors. Optodes and diodes are separated from each other by a distance of three centimetres in order to guarantee signal quality. Participants are fitted with a headband, covering most of the prefrontal cortex in particular bilateral orbitofrontal cortex, bilateral dorsolateral prefrontal cortex and bilateral premotor cortex. In order to ensure that the utilised headband is located according to the anatomical brain structures of the participants, the craniometric point of the nasion, where the top of the nose meets the ridge of the forehead, was used to assure comparability between all tested participants. A schematic representation of the measurement sites, the topographical layout, integrating 22 channels was designed to allow the measurement of the cortical neural activity of brain regions of the PFC (Fig. 6). The NIRS-Star software package (version 14.2) was used to check for signal quality and data collection. Mobile fNIRS data sets of three participants could not be used for the analysis due to malfunction.
Data analysis. Neural data was collected throughout the whole tasks. Additionally, a baseline measurement was implemented before starting the task, resulting in a three-minute baseline measurement in which participants were asked to sit in front of a monitor without any task. Following the same measurement protocol for all www.nature.com/scientificreports www.nature.com/scientificreports/ participants, integrating three hand immersion trial with the length of 60 s and two hand immersion trials with the length of 90 s; as well as two mental arithmetic (or control) trials with the length of 45 s, one mental arithmetic (or control) trial with the length of 60 s and one mental arithmetic (or control) trial with the length of 90 s; participants neural activity was measured utilising mobile fNIRS. For every participant, a general linear model (GLM) was set up to model neural activity during the experiment or control task, integrating all samples of each regressor (trial duration × 7.81 sampling rate; i.e. the 60 s hand immersion trail results in 468,6 samples). The parameters -hand immersion trials and the mental arithmetic (or control) trials -were modelled separately for every time interval, adding up to nine event-related regressors together with an additional error term at the end. For every participant (j) the following GLM was calculated: Yj = β1xj1 + β2xj2 + β3xj3 + β4xj4 + β5xj5 + β6xj6 + β7xj7 + β8xj8 + β9xj9 + εj. Each time course was further corrected for serial correlations such as physiological noise sources, modulating the stimulus onsets convolved by a haemodynamic response function 67 . No contrast was calculated for every participant individually (on within-subject-level). However, in order to investigate the estimated effects on group-level (between-subject-level), two groups -the MAST group and control group -were created and neural activity patterns were contrasted. Thereby, N = 19 participants who completed the MAST and N = 18 participants who completed the non-stressful control task were integrated in the data analysis. procedure. After signing the written informed consent, participants filled in the first subjective stress rating and gave a saliva sample (T1). Then the mobile fNIRS preparation began and the fNIRS cap was placed on the prefrontal cortex of the participant. A baseline fNIRS signal measurement was taken lasting three minutes. Then the MAST test, either in the stress or non-stress condition, began. Participants were randomly allocated to the stress (N = 19) or non-stress (N = 21) condition. As soon as the MAST was finished participants gave their second subjective stress rating and gave a second saliva sample (T2). They were then asked to answer a questionnaire including demographics. Twenty minutes after the completion of the MAST, participants gave their third subjective stress rating and gave a third saliva sample (T3). Participants received either course credits or 10 € for their participation. ethics approval and consent to participate. The study was approved by the ethics committee of the Faculty of Mathematics and Natural Science of the Heinrich-Heine-University in Dusseldorf. The research was conducted in accordance with the Helsinki Declaration. All patients gave their written consent.

Data Availability
The dataset analysed during the current study is available from the corresponding author on reasonable request. Figure 6. The topographical layout of the diodes and optodes of the applied mobile fNIRS system. This image was created using the analyse tool Nirslab (https://nirx.net/nirslab-1).