Ethical clearance for this study was obtained from the Ethics Committee for Animal Experiment of the Second Military Medical University. The experiment protocol was carried out in accordance with internationally accepted humane standards (Russell and Burch, 1959).

Thirteen Bama swine were received by the animal husbandry facility of the university. They were males, castrated 2 weeks after they were born to avoid the influence of hormonal effects. The animals were allowed to acclimatize individually to general laboratory temperature of 23 ± 1°C and humidity of 50–65% and provided standard laboratory swine meal, in portions of 2% of their body weight daily and water was made available ad libitum. They weighed between 20 and 25 (22.0 ± 1.2) kg at the beginning of the experiment and were confirmed free of systemic infections or skin diseases.

The animals were singly utilized in the experiment. Four days before hyperbaric exposure, they were trained to walk on a treadmill once a day for 2 days. Two days before hyperbaric exposure, evoked potentials were measured and normal skin samples were collected. One day prior to hyperbaric exposure, a thorough skin examination was performed to find contusions, abrasions and lacerations and the hair of all chosen sites was trimmed. Structure and blood flow of heart were measured. One hour before the exposure, skin temperature, and cutaneous ultrasound (for structure and blood flow) were recorded at trimmed sites.

The swine were placed one at a time in a 1,000 L animal compression chamber (DCW150, Yangyuan, Shanghai, China) with view ports to observe the animal inside. The chamber was pressurized with air to 40 msw and maintained for 35 min before decompression. Compression was performed in 5 min which began at 5 msw/min to minimize the possible discomfort to the swine. Chamber oxygen (O₂), carbon dioxide (CO₂), temperature, and relative humidity were continuously monitored and maintained at 21%, <0.3%, 22–24°C, and 65–75%, respectively. Decompression was conducted in linear segments of 5 msw/min from 40 to 30 msw, 4 msw/min from 30 to 20 msw, 3.3 msw/min from 20 to 10 msw, and 2.9 msw/min from 10 to 0 msw.

After surfacing, the skin was thoroughly examined for up to 36 h. The observations were continuous in the first 6 h and periodic with 1 h in next 18 h. In the last 12, 2–3 h interval of observations was arranged. The appearance, evolution, locations, and dimensions of lesions were recorded on a swine-shape figure (Figure 1). Lesions were measured by the palm of a single experimenter, which is similar to the estimation of burn surface area. The swine body surface area was calculated by Meeh-Rubner equation Area = 0.0974 × Weight2/3 (Quiring, 1955).

Based on our previous experimental results, several sites were selected as main observational regions including the four limbs, neck, chest, abdomen, waist, and back. Skin temperature was recorded using infrared thermometer (JXB182, Berrcom, China) with 0.1°C resolution. Skin thickness from the keratin cell layer to the dermis layer was measured using an Apogee 1000 ultrasound unit and a 14 MHz transducer (L8L38C, SIUI, Shantou, China) in color Doppler mode to image skin vessels. Both measurements were repeated during the evolution of lesions.

Heart structure was evaluated using transthoracic echocardiography (TTE) (P3f14C, SIUI, Shantou, China). A left ventricular long axis view was obtained, and the probe position was then adjusted to obtain an apical four chamber view. In the latter view an examination (incorporating observation of gross structure and use of color flow Doppler) was made for structural abnormalities in the atrial and ventricular septum. Although no formal bubble contrast studies were conducted, the real time repeat observations of venous and arterial bubbles after decompression served as an additional check for the presence of right to left shunt.

Bubbles in heart chambers were detected extrathoracically by the same machine and detector described above. Detection was repeated at 30, 60, 90 min, 2, 3, 4, and 6 h following surfacing, each lasting for 2 min. Left ventricular long axis view was found as the datum plane first, in which left atrium (LA), left ventricle (LV), and aorta (AO) can be seen clearly. Then the probe was adjusted to the aortic root short axis as the final view for detection. In this view, right ventricular outflow tract (RVOT), pulmonary artery (PA), and AO were presented. No detection was performed after 6 h due to the necessary oxygen breathing during anesthesia for evoked potential detection. Bubbles in ultrasound images were scored by Eftedal-Brubakk grading scale (Eftedal and Brubakk, 1997).

The two training sessions before hyperbaric exposure were defined as complete when each animal walked comfortably on a treadmill (YS-C600, KunYu, Henan, China) for 5 min at 1.6 km/h. At 2 and 6 h post-dive, motor function were tested on the same treadmill. The speed was gradually increased to 1.6 km/h at 0.4 km/h increments over 30 s, and the animal walked for 5 min at the highest speed, which earned a score of 5. If the animals could not stand or the endpoint on the treadmill was not achieved, the 5-point Tarlov score was adopted to grade motor function (Mahon et al., 2010).

Evoked potential detection was performed 2 days prior to, and 6 h after, the hyperbaric exposure. Animals fasted for 12 h before anesthesia, which was induced by intramuscular injection of 0.05 mg/kg atropine and 0.1 ml/kg Sumianxin and maintained with inhaled isoflurane (6%) via endotracheal intubation and an anesthetic machine (WATO EX-20 Vet, Mindray, Shenzhen, China). Sensory evoked potential (SEP) was measured using electromyography and evoked potential instruments (NDI-094, Haishen, Shanghai, China). Stimulating electrodes were placed into the right and left ankles to stimulate tibial nerves and electroneurographic signals were collected from first lumbar vertebra and head by receiving electrodes, which were defined as spinal somatosensory evoked potential (SSEP) and cortical somatosensory evoked potential (CSEP), respectively.

After each evoked potential detection with the animal still under anesthesia, skin samples were collected from selected areas and the center of stage IV lesions (described in Results)/adjacent non-affected sites. Biopsy sites were cleaned thoroughly with povidone iodine and saline. Full thickness skin biopsies were obtained, 3 × 1 cm spindle shaped. Excess fat from the biopsy was removed and 0.5 g of skin tissue separated from each specimen was stored at −80°C prior to NO determination using a total nitric oxide assay kit (Beyotime, Shanghai, China). The rest of the skin was fixed in 10% formaldehyde solution for 48 h before being mounted in paraffin, longitudinally sectioned, stained with Haematoxylin and Eosin (HE), and observed and photographed using a microscope (Eclipse55i, Nikon, Japan).

All data are presented as mean ± SD. The Pearson correlation coefficient was derived to evaluate the relation between bubble load, skin lesion area, and latent time to appearance. The changes in skin temperature and thickness were tested using one-way ANOVA followed by post-hoc Student Newman–Keuls test. The changes in NO level were tested by Dunnett's test. Latent time in SEP was tested for significance with a two-tailed Student's t-tests. Significance in all cases was accepted at P ≤ 0.05.

Eleven swine survived the experiments, and all developed skin lesions. The lesions showed obvious temporal patterns of change. The erythema, homogeneous purple-red color of the lesion gradually progressed into patterns of marbling or scattered lesions and completely disappeared without leaving any residual macroscopic skin changes. We therefore proposed a graduated system for DCS skin lesions. Not all lesions displayed each feature in the sequence, but typical lesion development can be described using stages shown in the caption of Figure 2, and chronological, exemplary appearances are shown in Figure 2.

Each swine had at least two lesions located in different sites along the body. All stages appeared in most of the lesions. In each lesion, stage III was the most serious appearance and the area was measured. 10–20% lesions in each animal were mild, the lesions faded quickly within 1 h after Stage I and II. Hence, the mild lesions did not show Stage III manifestation and were not counted in the total lesion area. Occurrence of Stage III lesions in the body sites are listed in Table 1.

Skin lesions appeared quickly following surfacing (13 ± 12 min) and almost simultaneously in different body sites. The speed of developing into stages varied among animals and lesions. Stage I and II usually existed for a relatively short time only, while stages III–VI took longer to evolve. Some lesions, especially in dorsal skin, remained at stage IV even after 24 h from surfacing. The mean duration for each stage is shown in Figure 3A from the 44 lesion s listed in Table 1, which show considerable variation. Another five lesions were distributed in sites other than the selected and trimmed areas.

The latency to stage III lesion can be taken as a parameter to reflect lesion severity as it correlates well with lesion area. As shown in Figure 3B, the shorter the latent time to stage III lesion, the larger the lesion area (r = −0.93, P = 0.000).

From the apical four chamber view, all swine had complete structure of ventricular septal and interatrial septum, and no patent foramen ovale and no abnormal blood flow were observed.

Figure 4 shows bubbles and the scores detected from the right hearts in the 11 survivors. Gas bubbles could be clearly observed (moving bright spots) in right ventricular outflow tract RVOT and PA in ultrasound images. No bubbles were observed in AO, LV, and LA in all the surviving swine. During the observation period, bubble amount was greatest at the first detection performed at 30 min following decompression, which showed a white-out image that single bubbles could not be discriminated in most of the swine. Bubble grades gradually decreased and disappeared within 6 h in 9 out of the 11 swine. The average bubble score for each pig was calculated from the area under the curve (similar to that shown in Figure 4C) divided by 6 h, and was regarded as bubble load for each animal.

There were significant correlations between bubble load and maximum lesion area ratio, latency to stage III lesion or total duration of stage III and IV. As shown in Figures 4D–F, the heavier the bubble load, the higher the lesion area ratio (r = 0.76, P = 0.007), the earlier the occurrence of stage III lesions (r = −0.83, P = 0.002), and the longer existence of cutaneous lesions (r = 0.79, P = 0.004). According to the area and latency of Stage III lesions, the bubble grade could be estimated (Table 2).

Swine skin thickness from the squamous keratin layer to the dermis varied across body areas including the thinnest abdomen (1.4 ± 0.1 mm) and thickest waist and back (3.7 ± 0.2 mm). Figure 5 shows the change in lesions with stages, increasing before Stage III and a gradual recovery from Stage IV to VI. The maximum thickness appeared around stage III to IV.

Generally, vessels could not be detected by color Doppler scanning in skin layer and were also hardly to be found in subcutaneous tissue. In lesion skin, especially in Stage III, in the lateral neck subcutaneous tissue, vessel images were not grossly apparent (Figure 6).

Surface temperature was slightly different in different body areas. However, changes showed no statistically significant variation during the six stages on all lesion surfaces (P = 0.880).

As shown in Figure 7, NO level increased significantly in lesion skin tissues compared to non-affected tissues (P = 0.002) or normal skins before modeling (P = 0.001). No difference was found between non-affected and normal skin (P = 0.611).

Congestion was the most common finding in skin lesions, with red blood cells (RBCs) were observed clogging in the dermal capillaries. Other changes included dilatation, hemorrhage, and neutrophil infiltrates (Figures not present). In non-affected skin from the decompressed swine, only dilatation was occasionally observed. There was no abnormal findings in normal control skin. A summary of histologic findings is shown in Table 3.

Assessment of motor function at 2 and 6 h after decompression yielded a score of 5 in all survivors. Pre-dive and post-dive SSEP and CSEP were compared, with no significant changes (P > 0.05, Figure 8). Histological examination of spinal cord was also performed with no abnormal findings (data not shown). No other neurological abnormality was observed.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.