Characteristic Odor Emanating From Skin During Emotional Tension

Masako Katsuyama (  masako.katsuyama@shiseido.com ) Shiseido Global Innovation Center, SHISEIDO CO. LTD Tomomi Narita Shiseido Global Innovation Center, SHISEIDO CO. LTD Masaya Nakashima Shiseido Global Innovation Center, SHISEIDO CO. LTD Kentaro Kusaba Shiseido Global Innovation Center, SHISEIDO CO. LTD Masatoshi Ochiai Shiseido Global Innovation Center, SHISEIDO CO. LTD Naomi Kunizawa Shiseido Global Innovation Center, SHISEIDO CO. LTD Akihiro Kawaraya Takasago International Corporation, Corporate Research & Development Division Yukari Kuwahara Takasago International Corporation, Corporate Research & Development Division


Introduction
Human emotions appear in motions, voices, facial expressions, and other forms, and people perceive and understand such emotions. Frequently, a person's emotion can be perceived through non-verbal indicators. The study focuses on the effects of body odor as a form of non-verbal communication.
Scholars report two main routes for the generation of body odor. The rst is degeneration by oxidation that occurs through the interaction between microorganisms and sebum/sweat on the surface of the skin [1,2]. The second is odor generated through the skin from inside the body. This type of odor constitutes transcutaneous blood volatile organic compounds or skin gas [3][4][5]. For a long time, scholars recognize the fact that body odor changes with physical condition, such as during illnesses. Recent studies propose that skin gas generated by the body is an accurate method of determining the state of the body along with other biological indicators, such as saliva, urine, and blood [6]. Since then, research in this eld has advanced signi cantly.
For example, in the medical eld, acetone content found in the skin gas of diabetes patients was discovered to be higher than that of healthy individuals [7,8]. In lieu of this nding, interventions for the management of various disease states are being examined. Similarly, many studies reported that dogs can detect the presence of early-stage cancer in their owners through odor. Consequently, dogs are being trained for cancer detection [9,10]. In the same manner, the current researchers investigated the relationship between skin gas and the effects of diet, physical condition, and aging in healthy volunteers, instead of people suffering from illnesses or disorders. The results indicated that aged people emit 2nonenal [11], which occurs by the oxidation of sebum in elderly individuals, and people with constipation emit high levels of p-cresol [12].
Based on the abovementioned ndings, skin gas is deemed a useful means of investigating the condition of the body, including diseases and general physical condition. The study further hypothesized that psychological changes may alter skin gas. Therefore, the study aimed to investigate changes in skin-gas components during stress-inducing sessions.
The results of the study indicate that humans reproducibly emit discernible odors that contain allyl mercaptan and dimethyl trisul de through the skin during interviews that induce psychological tension. Further, it was found that when other people smell this odor, they undergo psychological tension, confusion, and tiredness. In particular, skin gas may contain ingredients that convey human psychological conditions.

Physiological responses and skin-gas characteristics under emotional tension
To examine changes in skin-gas composition as a result of psychological changes, the study investigated physiological responses and changes in the components of skin gas gathered from the hands ( Fig. 1) of subjects during interview sessions that were intended to induce stress (see the Methods section). For one physiological response, heart-rate "R wave" intervals were calculated using electrocardiogram data. The low frequency (LF)/high frequency (HF) during the interviews was signi cantly higher (p < 0.01) and sympathetically dominant when compared before and after the interviews (Fig. 2). For another physiological response, salivary cortisol was quanti ed using ELISA. A signi cant increase (p < 0.01) in salivary cortisol levels was observed after the interviews compared with baseline values recorded during home rest (Fig. 3).
The LF/HF during the interview was signi cantly higher than those of previous and subsequent measurements, which indicates that the sympathetic nervous system was dominant (**p < 0.01 after Bonferroni correction).

Sensory assessment of skin gas
Collected skin gases were smelled by a panel of four experts on odor to compare the differences in odor during stress and at rest. A characteristic odor similar to sulfur-containing compounds was recognized in all gas samples collected during the interviews, whereas none existed in the samples taken at baseline.
Additionally, the intensity of this characteristic odor was positively correlated with the rate of increase in sympathetic nerve activity during the interview (compared with values before the interview; Spearman's correlation coe cient, r = 0.66; p < 0.01; Fig. 4). Furthermore, in a test to con rm this phenomenon in non-Japanese women, this characteristic odor was detected in the gases collected during tension in all subjects (Japanese: 4 men; Chinese: 1 woman and 2 men; and French: 2 men). Conversely, no sulfur-like odor was observed in samples taken during elevated heart rates after exercising on a bicycle. (Fig. 5) Identifying characteristic odorants To identify the main components of characteristic odors, the collected gases were analyzed using gas chromatography-olfactometry (GC-O) and gas chromatography-mass spectrometry (GC-MS) analyses.
Analysis i: One of the characteristic odorants similar to the typical stress odor was detected and identi ed as dimethyl trisul de (DMTS) by GC/MS-O analysis (Fig. 6a).
Analysis ii: In addition to DMTS, a characteristic odorant was detected and estimated as propyl mercaptan (PM) or allyl mercaptan (AM) by GC-O analysis (retention time and odor).
Analysis iii: After comparing retention time and odor between the target odorant with PM and AM using GC-O analysis, the result indicates that retention time and odor of collected skin gas were in good agreement with those of the AM standard. Furthermore, AM was detected in the sample through GC-MS (SIM) analysis (Fig. 6b).
The results suggested that DMTS and AM were major contributors to the characteristic odor, which is similar to typical sulfur-containing compounds, in gas emanating from the skin during the stress-inducing interviews.

Psychological effects of tension odor
The study found that sni ng the tension model odor signi cantly increased the psychological scores for "tension-anxiety" (p < 0.01), "confusion-bewilderment" (p < 0.01), and "fatigue-inertia" (p = 0.12) among seven items that represent the psychological states of the subjects (Fig. 7).

Discussion
Scholars recently noted that humans release a wide variety of gases from the skin, whose components may act as a barometer for indicating the state of the body.
For example, persons with diabetes characteristically release acetone, whereas those with liver diseases typically emit ammonia. Moreover, the study con rmed that the human skin releases speci c components due to changes in the body unrelated to illnesses, such as aging and constipation.
Alternatively, studies reported body odor in samples collected mainly from the armpit, which is caused by various emotional states. Majority of the studies examined the reactions of dogs and humans who smell human odors of different states [13][14][15]. Reports of human odors estimated by statistical methods do not indicate how such odor affects humans [16]. Therefore, this study hypothesized that changes in psychological state chemically affect skin gas and its components affect human psychology. This study was conducted with the aim of identifying the main components of skin gas released during certain emotional states and examining their psychological effects. In the middle of various psychological changes, the study opted to induce mild tension in a manner that is easy to control and with reference to the Trier social stress test method [17,18].
Collecting body odor from humans can be conducted using various methods, such as armpits using cotton balls [13][14][15], clothing [2,19], and the back and neck [20]. However, such methods contain a mixture of bacterial metabolites of sebum and/or apocrine gland secretions in addition to the odor that comes directly from the body. Thus, the study examined the speci c relationship between changes in skin-gas components and the body. For this reason, samples were collected from the hand because it contains less sebaceous glands and no apocrine glands [21].
During the interviews with instigated tension, physiological indicators before, during, and after the interview were measured to con rm the subject's condition. The results indicate that the LF/HF values during the interviews were signi cantly higher than those before and after the interviews, which indicate sympathetic dominance. Cortisol levels of saliva during the interviews were also signi cantly higher than those collected at baseline at the same time on another day. In particular, when the effects of the circadian rhythm [22] on cortisol was canceled and compared, the nding demonstrated that the stress level was high. The results suggested that the subjects were psychologically tense during the tensioninducing interviews. Additionally, sensory assessment was performed to con rm changes in odor components as a result of the altered psychological states. Consequently, a common characteristic odor that smells similar to sulfur-containing compounds was recognized in all samples of skin gas collected during the interviews despite their short duration (20 min). Notably, the intensity of this characteristic odor was positively correlated with the rate of increase in sympathetic activity during the interviews. Moreover, a positive correlation was found between heart-rate variability and odor intensity. On another test, this phenomenon was also observed in men or non-Japanese individuals while they were feeling tensed. These ndings suggest that with greater emotional tension, characteristic odor is emitted more strongly from humans, and this phenomenon may be an instinctive physiological reaction among humans. Conversely, the characteristic odor was unrecognized in the results of the sensory evaluation of skin-gas samples collected during a bike exercise, which featured an elevated heart rate (> 140bpm for 20 min) (Fig. 5). This result suggests that a mere increase in heart rate and accompanying increase in thermal sweating was not the direct cause of the characteristic odor released by the skin.
The characteristic odor was clearly recognized through sensory evaluation. However, determining the target compounds was extremely di cult due to the very low concentrations. Thus, the study searched for effective sample preparation methods and analytical conditions and nally succeeded in identifying the target compounds as DMTS and AM. The successful identi cation of the two sulfur-like compounds rendered possible the arti cial recreation of the model tension-stress odor. Moreover, the study examined the psychological effects of subjects directly inhaling and smelling the odor. The second edition of the short version of the Pro le of Mood States (POMS 2) was used to evaluate psychological status. POMS 2 is a widely used self-report questionnaire for the rapid assessment of mood states, including transient, uctuating feelings, and enduring affect states. The higher the T scores for anger-hostility (AH), confusion-bewilderment (CB), depression-dejection (DD), fatigue-inertia (FI), and tension-anxiety (TA), the more negative the mood. Conversely, positive mood states, such as vigor-activity (VA) and friendliness (F), indicate that the higher the T score, the better the mood [23]. POMS is also used in various research elds to measure mood disturbance and/or fatigue [24,25]. The results showed that sni ng the model's tension-stress odor increased negative emotions such as tension-anxiety (TA), fatigue-inertia (FI), and confusion-confusion (CB). This nding suggests that the odor emanating from emotionally stressed people may exert a psychological impact on people within their immediate vicinity. If odors released from people are a profound means of communication, then this nding may have a great impact on various elds. Thus, the elucidation of the mechanism and signi cance of this characteristic odor development is a subject for future research. Moreover, the current nding that emotional changes may in uence body odor produced from areas apart from the apocrine glands, which is speci c to axillary odor, may indicate new concepts and potential applications in the eld of biogas research. Thus, the study aims to proceed with the development of a practical fragrance material as it endeavors to elucidate the odor generation mechanism. The study intends to eliminate the discomfort that may be cause by a tension-induced odor by formulating a scent that harmonizes with this odor and removes its negative psychological effects instead of covering it with a strong odor. This research is not only an etiquette issue in interpersonal relationships but is also applicable to the personal management of mental health and smooth communication in social interaction in the case of a serious COVID-19 pandemic. Conversely, in the rescue of victims of natural and man-made disasters, consideration is given to the use of human body odor as a clue for those who are still alive [25][26][27]. The perspective of tension-and stress-induced human odor in the eld of rescue can be extremely helpful in detecting humans in crisis. Such applications hold great potential. Therefore, further research is required.

Effects of tension on skin gases Subject
The subjects of the main study consisted of 40 healthy, non-smoking Japanese women aged 35-44 years. In the con rmation test of skin gas odor, four Japanese men, three Chinese men and women, and two French men in the 20-30 years age group who were healthy and non-smoking participated in this study. Prior to the study, the subjects were informed of the purpose, methods, anticipated clinical bene ts, and disadvantages of the study. All participants provided written informed consent. The study was conducted in compliance with ethical principles based on the Declaration of Helsinki. Furthermore, the study was conducted with the approval of the Shiseido Ethics Committee (Approval Number: C01402, C01555, C01577, C01824).
Restrictions for participation in the study were as follows: (a) the subjects did not consume odorous food 1 day prior to the study and (b) the subjects did not make contact with odor-intense objects. Lastly, (c) they were instructed not to wear any fragrance.
Test methods and skin-gas collection methods Samples of skin gas were collected in an air-controlled room. Samples were extracted from each subject's non-dominant hand in two situations: at rest (sitting on a chair and relaxing while reading a magazine) and under stress (answering questions for 20 min from an interviewer they did not know). The subject's hand was washed with unscented soap prior to testing. The washed hand was covered with a sampling bag made of polyvinyl uoride polymer resin lm (Tedlar®) with a single-ended mini valve. Nitrogen gas (500 mL) was added after degassing the air in the bag, which was closed and left for 20 min (Fig. 1).
Afterward, gas obtained from the sampling bags was recovered, placed in a storage bag (ANALYTIC-BARRIER™, GL Science Co., Ltd.) using connected silicone tubes. The gas was used as a sample.

Physiological index measurement
Autonomic status Four electrodes were placed under the left and right subclavian bones and the left and right tenth ribs, whereas electrocardiographic activity was continuously measured by an electrocardiogram monitor (Ledar Circ, Sumitomo Dainippon Pharmaceutical).
Based on the electrocardiogram data, a power spectrum analysis of the heartbeat "R wave" intervals was performed using an autonomic nervous system activity analysis software (Sumitomo Dainippon Pharmaceutical, FLUCLET®). The intensities of the HF (ranging from 0.2 to 2 Hz) and LF (ranging from 0.04 to 0.15 Hz) components were determined. Additionally, the LF/HF component was used as an activity index of the sympathetic nervous system.

Salivary cortisol measurements
Saliva samples were collected twice. The rst measurement denotes samples collected immediately after the interview, whereas the second measurement refers to samples collected at baseline during home rest at the same time frame as the interview but on a different day. This process was observed to avoid cortisol changes resulting from the circadian rhythm. Salivary cortisol was quanti ed using ELISA (Cortisol Salivary Immunoassay Kit Salimetrics LLC).

Odor assessment
The skin-gas samples collected at the relaxed and tense states were subjected to sensory evaluations by four experts to identify differences in odor. The experts noted a characteristic smell in the tension samples, which they could not detect in the relaxed sample. A ve-point Likert-type scale was used to evaluate the intensity of the characteristic smell. The gases containing this smell in the tension samples, which were not identi ed in the control samples, were subjected to the following analysis for component identi cation.

Identifying the components of the characteristic odor
To identify the main components of the characteristic odor, collected gas samples were analyzed in the following order:

Psychological effect
The study developed a model odor using the components identi ed in the previous section and examined their psychological impact on other people.

Subjects and restrictions
The subjects consisted of 33 Japanese women aged 35-44 years. Prior to the study, the subjects were informed of the purpose, methods, anticipated clinical bene ts, and disadvantages of the study. Written informed consent was obtained from all participants. The study was conducted in compliance with ethical principles based on the Declaration of Helsinki. Furthermore, the study was conducted with the approval of the Shiseido Ethics Committee. Restrictions for participation in the study were as follows. (a) The subjects did not consume odorous food 1 day before the study. (b) The subjects did not make contact with odor-intense objects. They were instructed not to wear any fragrances.

Assessing the psychological state
The model tension-stress odor, whose concentration was adjusted according to the odor threshold, was used as test samples. The study rst con rmed the threshold per subject prior to assessing their psychological states.
The model tension-stress odor was presented to the subjects, who continuously sniffed it for 2 min. The participants then completed the validated questionnaires: The psychological state of the subjects before and after sni ng was measured using POMS 2, which contains 35 questions that assess the following moods: AH, CB, DD, FI, TA, VA, and F.

Statistical analysis
The results of the test samples and each experiment are presented as mean ± SD. Paired Student's t-test was used to verify of the signi cance of the results of salivary cortisol levels. Bonferroni's method was used for the signi cance tests of the activity of the sympathetic nervous system. The relationship between odor intensity and autonomic nervous activity was assessed using pairwise non-parametric correlations (Spearman's). Lastly, Wilcoxon's signed-rank test was used to determine signi cant differences in the results of the psychological index. A p value < 0.05 on both sides was considered statistically signi cant. Figure 1 Collection bag for skin gas emanating from subjects' hand. Hands washed with unscented soap were covered with sampling bags made from a uorinated ethylene-propylene copolymer material. The air inside was replaced with nitrogen and recovered after a certain period of time.  Relationship between intensity of tension-stress odor and autonomic nervous activity during the interview. A panel of four experts on odor evaluated the collected skin-gas samples to compare differences between odors emanated during stress and at baseline. Based on all interview samples, the characteristic sulfur-smelling compound odor was discovered. Further, the intensity of this characteristic odor was positively correlated with the rate of increase in sympathetic nerve activity during the interview compared with before the interview (Spearman's correlation coe cient: r = 0.66 (p < 0.01)).

Figure 5
Relationship between the intensity of tension-stress odor during exercise and changes in heart rate. A panel of four experts on scent evaluated the collected skin-gas samples to compare the difference in odor between exercise and rest. No characteristic sulfur compound odor was observed in any of the samples when heart rate increased because of exercise.  Psychological results. The model tension odor, which was formulated according to the threshold, was continuously sniffed for 2 min. POMS 2 was used to measure the subjects' psychological status at baseline and immediately after sni ng the sample. The study observed signi cant increases in "tensionanxiety" (p < 0.01), "confusion-bewilderment" (p < 0.01), and "fatigue-inertia" (p = 0.012). Note: AH: anger-hostility, CB: confusion-bewilderment, DD: depression-dejection, FI: fatigue-inertia, TA: tensionanxiety, VA: vigor-activity, and F: friendliness.