The study was approved by the Human Ethics Committee of Stockholm (Ref. no.: 2021-00932) and conformed to the standards set by the Declaration of Helsinki. Subjects were informed in detail about the experimental procedure before giving their written consent to participate.

Eight healthy, right-handed men participated [mean (range): age 25 (21–30) years, body mass 81.8 (64.9–95.4) kg, height 180 (163–186) cm, total skinfold thickness 94 (50–143) mm, and right-hand volume 414 (326–459) mL]. The sample size was based on our previous work employing a similar experimental design (Keramidas et al. 2019), using α = 0.05, β = 0.85 and an effect size f = 0.50 (G*power software, Heinrich Heine-Universitat, Dusseldorf, Germany; Faul et al. 2007). Subjects were nonsmokers, normotensive, were not taking any medication, and had no history of cold injury. They were instructed to abstain from alcohol and strenuous exercise for at least 24 h before each trial, to refrain from caffeine during the testing day, and to maintain their sleeping, eating and exercise routines throughout the study period.

The study used a repeated-measures, single-blinded design. Specifically, subjects performed, on four separate occasions (days), a local cold provocation consisting of a 30-min hand immersion in 8 °C water, while they were immersed to the chest in: (i) 35.1 (0.2)°C water and were inhaling normal air [fraction of inspired oxygen (FIO2): 0.21; i.e., normothermic, normoxic trial], (ii) 35.0 (0.5)°C water and were inhaling pure O2 (FIO2: 1.0; i.e., normothermic, hyperoxic trial), (iii) 21.0 (0)°C water and were inhaling normal air (i.e., mild-hypothermic, normoxic trial), and (iv) 21.0 (0.1)°C water and were inhaling pure O2 (i.e., mild-hypothermic, hyperoxic trial). To induce a normothermic state, the initial temperature of the water was set at 35.5 °C (Craig and Dvorak 1966; Keramidas et al. 2019); yet, during the course of each immersion, it was slightly adjusted so that the individual rectal temperature (Trec) remained relatively constant. For each thermal state, no differences were detected between the two breathing conditions with regards to the water temperature (normothermic trial: P = 0.43; hypothermic trial: P = 0.63). The four trials were conducted in a quasi-Latin-square fashion: they were performed in a counterbalanced order, but, for each breathing condition (i.e., normoxia and hyperoxia), the mild-hypothermic trial was performed first, because, in the normothermic trial the immersion period preceding the hand cold provocation was determined by the time required to induce a 0.5 °C drop in Trec in the mild-hypothermic trial (Keramidas et al. 2019). The study was carried out between June and November. Yet, for the individual subject, the four trials, which were performed at the same time of the day, were completed within a 21-day period and were separated by ≥ 3 and ≤ 8 days. The mean (standard deviation; SD) temperature, relative humidity and barometric pressure in the laboratory were 27.5 (0.4)°C, 33 (6)% and 756 (7) mmHg, respectively.

Subjects were always clad in regular swim shorts. During all trials, they were equipped with an oronasal mask, and breathed through a low resistance two-way respiratory valve (Model 2, 700 T-Shape, Hans Rudolph, Inc. Shawnee, USA); the inspiratory side of the valve was connected via respiratory corrugated tubing to a bag filled with the respective pre-mixed humidified gas. Each trial commenced with a 20-min baseline phase, during which subjects remained in a resting, semi-reclining position on a gurney placed next to a tank. In the hyperoxic trials, the 20-min baseline phase comprised 10 min of breathing normoxic gas, followed by breathing 100% O2 until the end of the trial; presumably, the 10-min period in hyperoxia was sufficient to stabilise the arterial partial O2 pressure at an elevated level, as indicated by the slightly enhanced values of capillary oxyhaemoglobin saturation [SpO2; normoxia: 98 (1)%, hyperoxia: 100 (1)%; P < 0.001]. After the 20-min baseline phase, subjects entered the tank, which was filled with stirred water maintained at either 35.1 (0.4) °C (i.e., in the normothermic trials), or 21.0 (0.1)°C (i.e., in the mild-hypothermic trials). They were immersed to the level of the xiphoid process, and remained in a semi-upright sitting position with both arms being supported at the level of the heart, above the water surface. The left hand was exposed to the ambient room temperature throughout. The right hand (i.e., the test hand) was placed in a custom-made, water-perfused, tube-lined mitten, and warm water was circulated through the tubes maintaining the skin temperature of the fingers at 35.5–36.0 °C (KTH4H/B, Panasonic, Aichi, Japan) (Keramidas et al. 2022). The mitten was covering the entire hand, and the most distal portion of the forearm (i.e., it extended up to ~15 cm above the wrist). During the mild-hypothermic trials, subjects rested idle in this position until their Trec dropped by 0.5 °C from baseline (B-WI phase). As mentioned previously, during the normothermic trials, the duration of B-WI phase was similar to that obtained in the mild-hypothermic trial of the respective breathing condition. At the end of each B-WI phase, the right hand was removed from the mitten, was covered with a thin plastic bag, and was immersed up to the ulnar and radial styloids for 30 min in a different tank filled with 8 °C water (H-CWI phase). After the completion of the H-CWI phase, the right hand was removed from the water, and a 15-min spontaneous hand rewarming period ensued (H-RW phase), during which subjects remained in the tank with both arms resting on the arm-support and being exposed to the room temperature. Afterwards, subjects were removed from the tank, placed in a well-insulated sleeping bag on the gurney, and were monitored for a further 30-min period (B-RW phase).

Trec was monitored continuously with a rectal thermistor (Yellow Springs Instruments, Yellow Springs, OH) inserted ~10 cm beyond the anal sphincter. Mean skin temperature (Tsk) was derived from the unweighted average of skin temperatures, recorded with copper-constantan (T-type) thermocouple probes (Physitemp Instruments Inc., Clifton, NJ) at the left side of the body: on the forehead, upper arm, upper and low back, forearm, ring finger, chest, abdomen, thigh, calf, foot, and big toe. Mean body temperature (\(\overline\)b) was calculated using the equation: \(\overline\)b = 0.64 × Trec + 0.36 × Tsk (Burton 1935; Colin et al. 1971; Lenhardt and Sessler 2006). Five additional thermocouples were attached to the middle of the palmar side of the distal phalanx of each finger of the right hand. The average (TF-avg), minimum, and maximum finger temperatures were calculated. A finger skin-temperature elevation of ≥1 °C lasting for ≥3 min was defined as a cold-induced vasodilatation (i.e., CIVD) response (Keramidas et al. 2019). The following CIVD-related parameters were also assessed: (i) the total number of CIVD events, (ii) the temperature amplitude, which was calculated as the difference between the lowest temperature recorded just before the CIVD and the highest temperature reached during the CIVD, and (iii) the duration of each CIVD event. All temperatures were sampled at 1 Hz with a NI USB-6215 data acquisition system, and processed with LabVIEW software (v. 2019, National Instruments, Austin, TX).

Systolic (SAP), diastolic (DAP), and mean (MAP) arterial pressures were measured continuously using finger photoplethysmography (Finometer, Finapres Medical Systems BV, Amsterdam, The Netherlands). The pressure cuff was placed around the middle phalanx of the left middle finger, and the reference pressure transducer positioned at the level of the heart. Heart rate (HR) was derived from the arterial pressure curves as the inverse of the interbeat interval. Local skin blood flux was assessed at a rate of 10 Hz on the palmar side of the distal phalanx of the right (i.e., immersed) index finger and on the dorsal side of the left (i.e., non-immersed) forearm by laser-Doppler flowmetry (VMS-LDF2; Moor Instruments, Axminster, UK) using optic probes (VP1/7; Moor Instruments, UK). Skin blood flux was reported as cutaneous vascular conductance (CVC), calculated as skin blood flux divided by MAP.

SpO2 was monitored with an earlobe pulse oximeter (Radical-7, Masimo, Irvine, CA, USA). During the baseline and just before the H-CWI phase, glucose concentration was measured from a finger capillary blood sample (Accu-Check, Aviva, Roche, Mannheim, Germany).

During the baseline, B-WI (at 10-min intervals), H-CWI (at minutes 1, 2, 3, 4, 5, and every 5 min thereafter), H-RW (at minutes 1, 5, 10, and 15), and B-RW (at 10-min intervals) phases, subjects were asked to provide ratings of their whole-body and right-hand thermal sensation (from 1-cold to 7-hot) and comfort (from 1- comfortable to 4-very uncomfortable). At the same time intervals, the general affective valence (from -5-very bad to + 5-very good), the perceived shivering intensity (from 1-no shivering to 4-heavy shivering), and the local (right hand) pain (from 0-no pain to 10-unbearable pain) were also assessed. Upon arrival at the laboratory during the first session, subjects were introduced to and thoroughly familiarised with all scales.

Data are presented as the mean of each phase. Baseline values were calculated as the average of the final 10 min of the 20-min baseline phase. Normality of distribution for all datasets was assessed using the Shapiro–Wilk test. A two-way [thermal status (mild hypothermia vs normothermia) × breathing condition (normoxia vs hyperoxia)] repeated-measures analysis of variance (ANOVA) was used for all physiological variables. Mauchly’s test was conducted to assess the sphericity and, if necessary, the Greenhouse–Geiser ɛ correction was used to adjust the degrees of freedom. When an ANOVA revealed significant effects, multiple pairwise comparisons were performed with Newman-Keuls post hoc test. A paired sample Student’s t-test was used to assess differences in the duration of B-WI phase between the normoxic and hyperoxic trials. Differences in the incidence of CIVD events were determined with the Chi-square test. Differences in the perceptual responses were evaluated with Friedman’s test, followed by a Wilcoxon test. The relationship between local pain and finger skin temperature was assessed with a Spearman rank correlation coefficient (rs); where the magnitudes of correlations were interpreted qualitatively using Cohen’s scale: r < 0.1: trivial, 0.1–0.3: small, 0.3–0.5: moderate, and >0.5: large. Statistical analyses were conducted using Statistica 8.0 (StatSoft, Tulsa, OK), and figures were produced using Prism 10.0 (GraphPad Software Inc., San Diego, CA, USA). Unless otherwise stated, data are presented as mean values with (SD). The α level of significance was set a priori at 0.05.