Between June 2016 and June 2019, 44 patients who were admitted to the intensive care unit (ICU) of the General Hospital of the Southern Theatre Command of the People's Liberation Army (PLA) of China, a tertiary hospital, within 24 h of HS occurrence were prospectively enrolled. Patients were diagnosed with HS according to the Expert Consensus on Diagnosis and Treatment of Heat Stroke in China (2019) (19), promulgated by the Expert Group on Prevention and Treatment of Heat Stroke and Critical Care Committee of the PLA of China. The diagnostic criteria were as follows: A medical history of i) exposure to high temperature and high humidity or ii) high intensity exercise; clinical manifestations of i) central nervous system dysfunction (e.g., coma, convulsion, delirium or abnormal behavior), ii) core body temperature >40˚C and iii) multiple (≥2) organ dysfunction that could not be explained by other etiologies. Patients with malignant tumors, chronic liver or kidney diseases, chronic cardiac insufficiency (New York grades 3-4), chronic pulmonary insufficiency, underlying central nervous system disease or metabolic disorders were excluded. Patients were categorized as survivors (n=36) and non-survivors (n=8). Healthy subjects from the physical examination center (15) were enrolled as the control group. Baseline demographic characteristics were recorded on ICU admission (day 1). Body temperature was measured using an ear thermometer on day 1. Blood samples were collected on day 1 and after 4 days of treatment (day 4). Biochemical parameters (lactate, albumin, alanine aminotransferase, aspartate aminotransferase, urea nitrogen, creatine kinase, creatine kinase-myocardial band, creatinine, cardiac troponin I, fibrin degradation product, fibrin, international normalized ratio, myoglobin, partial pressure of arterial oxygen, procalcitonin, platelet, prothrombin time, total bilirubin, white blood cell count and D-dimer levels were assessed) reflecting organ function were collected from patients at the Laboratory of the General Hospital of the Southern Theatre Command of the People's Liberation Army (PLA) of China. Acute Physiology and Chronic Health Enquiry (APACHE) II (20) and Sequential Organ Failure Assessment (SOFA) (21) scores were also recorded. Written informed consent was obtained from all patients or their representatives. This study was approved by the Medical Ethics Review Committee of the General Hospital of the Southern Theater Command of the PLA of China.

A 10 ml volume of blood was collected from the peripheral vein of each patient using ethylenediaminetetraacetic acid anticoagulant tubes. Blood collection tubes were left standing vertically at 22-27˚C for 30 min. Whole blood was then centrifuged at 2,500 x g at 4˚C for 10 min to isolate the plasma. Each plasma sample was diluted with the equivalent volume of phosphate-buffered saline (PBS) and subjected to three rounds of centrifugation at 4˚C (2,500 x g for 30 min, 12,000 x g for 45 min and 110,000 x g for 2 h). Finally, the precipitate (exosomes) was resuspended in 50-200 µl of PBS and stored at -80˚C, as previously described (12).

Exosome samples were obtained from a healthy individual, a patient with mild HS (HS without multiple organ failure), and a patient with a severe HS (HS complicated with multiple organ failure) (randomly selected) and prepared for transmission electron microscopy (TEM) evaluation. Exosomes were precipitated in 50-100 µl 2% paraformaldehyde (diluted in PBS at 4˚C). A 5 µl volume of exosome suspension was added to a formvar-coated copper grid (Mecalab Ltée), incubated for 30 min, washed in 100 µl of PBS, fixed in 2% paraformaldehyde for 10 min and then stained with 2% uranyl acetate dissolved in 50% ethanol for 15 min (all at 4˚C). The samples were then visualized using a Philips CM10 transmission electron microscope (model no. JEM-2100F; Philips Healthcare).

To determine the level and size distribution of the isolated exosomes, nanoparticle tracking analysis (NTA) was performed using the NanoSight NS3000 (Malvern Panalytical; Spectris plc), according to the manufacturer's instructions. Briefly, the exosome samples were diluted to a final ratio of 1:5,000 in sterile PBS, and each sample was analyzed three times, each for 60 sec, using the NanoSight's automatic analysis settings.

Proteins extracted from the plasma exosomes using RIPA lysis buffer (Biosharp Life Sciences; cat. no. BL504A) with a protease inhibitor (Biosharp Life Sciences; cat. no. BL612A) and phenylmethanesulfonyl (Biosharp Life Sciences; cat. no. BL507A) at a ratio of 100:1:1. The protein concentration were determined using a BCA test kit (Biosharp Life Sciences; cat. no. BL521A). Protein (10 µg/lane) was loaded and isolated using 10% sodium dodecyl sulfate-polyacrylamide gel electrophoresis and electro-transferred to a 0.2 µm nitrocellulose membrane. After blocking the nonspecific binding sites with bovine serum albumin (Gibco; Thermo Fisher Scientific, Inc.) for 1 h at 25˚C, the membrane was incubated with primary antibodies (diluted 1:1,000) at 4˚C overnight and then incubated with HRP-conjugated secondary antibodies (Goat Anti-Mouse IgG H&L; cat. no. ab6789; 1:5,000) at 25˚C for 1 h. The antibodies used were the following: Anti-CD63 (cat. no. ab216130), anti-CD9 (cat. no. ab223052), anti-CD81 (cat. no. ab155760), anti-Tsg-101 (cat. no. ab30871) and anti-GAPDH (cat. no. ab8245) and were all obtained from Abcam.

Histone H3 levels in the plasma exosomes were measured using a commercial ELISA kit (Human Histone H3 ELISA kit; cat. no. MM-51796H; Jiangsu Enzyme Free Industrial Co., Ltd.). The protocol provided by the manufacturer was followed without modifications.

A kurtosis test was used to check the data for normality. Normally distributed data are presented as the mean ± standard deviation, while non-normally distributed data were presented as medians ± interquartile ranges. Each statistical analysis involving two groups was performed using a two-tailed Student's t-test for normally distributed data, Kruskal-Wallis H test for non-normal data and Fisher's exact test for categorical data. Mixed two-way ANOVA with both within-subjects (day) and between-subjects (patient group) factors with a post hoc test (simple main effects with a Bonferroni correction) was utilized to compare more than two groups. For categorical data, McNemar's test was used for comparisons within groups. Correlations were analyzed using the Spearman's rank correlation test. The discriminated ability of survivors and non-survivors by plasma exosome histone H3 was examined by receiver operator characteristic (ROC) curve analysis with comparisons to plasma alanine aminotransferase (ALT) and aspartate aminotransferase (AST). A P-value of <0.05 was considered statistically significant. All tests were performed using GraphPad software version 7.0 (GraphPad Software, Inc.).

Blood samples from 44 patients clinically diagnosed with HS and from 15 healthy controls were collected. All patients with HS and healthy controls were male and had few underlying diseases. The mean patient age was comparable between healthy controls and HS patients 25±6 (range 17-40) vs. 21±4 (range 16-38) years]. Clinical and laboratory data are summarized in Table I, Table II and Table III. The overall 28-day mortality rate in patients with HS was 18.2% (8/44). The duration of ICU stay was significantly longer in non-survivors than in survivors (median: 10 vs. 7 days; P<0.001). Body temperatures on admission were similar between the HS survivor and non-survivor groups (38.58±1.62 and 39.11±1.81˚C, respectively) and were higher in these groups than in healthy controls (36.42±0.46˚C; both P<0.001). The majority of the patients were cooled prior to admission and the cooling strategies are presented in Table II. On day 4, significant resolution of hyperthermia was observed in both survivors and non-survivors (both P<0.001 vs. day 1; Table I).

Based on the results of routine biochemical examinations performed on day 1, few of the routine biochemical parameters reflecting organ dysfunction were significantly different between non-survivors and survivors. These were hemodynamic instability (as measured by an increased need for vasoactive drugs and increased lactate levels), respiratory impairment (an increased need for mechanical ventilation), need for continuous renal replacement therapy and presence of coagulation abnormalities [prolonged international normalized ratio (INR), increased fibrinogen degradation product (FDP) levels and decreased fibrinogen levels], rhabdomyolysis (increased creatine kinase and myoglobin levels) and central nervous dysfunction [decreased Glasgow coma scale (GCS) score]. Meanwhile, there were no significant differences between survivors and non-survivors in terms of heart rate (HR), mean arterial pressure, PaO₂/FiO₂ ratio, liver injury indices [total bilirubin (TBil) level and ALT]/AST ratio], renal function [creatinine levels, blood urea nitrogen (BUN) levels, and urine output], prothrombin and D-dimer levels, cardiac injury index [cardiac troponin-I (cTnI)], and levels of the inflammatory marker procalcitonin. Myoglobin levels were significantly higher in HS survivors than in healthy controls but were lower in non-survivors than in survivors. Notably, TBil levels, renal parameters, and D-dimer and cTnI levels did not appear to be significantly different between non-survivors and survivors until day 4. On day 1, both the APACHE II and SOFA scores, which were used to assess overall disease severity, were not significantly different between non-survivors and survivors, whereas SOFA scores were higher among non-survivors only on day 4. Hemodynamic indices, mechanical ventilation (MV) ratios, TBil levels, and renal parameters in non-survivors deteriorated from day 1 to day 4, while the remaining indicators of organ function remained unchanged. Among survivors, most variables were not significantly different between day 1 and 4, except for white blood cell count and procalcitonin levels (Table I, Table II and Table III).

TEM was performed for the isolated plasma exosome samples obtained from healthy controls and surviving and non-surviving patients with HS. Double-membrane vesicle-like structures of ~100 nm in diameter were observed in the samples from each group (Fig. 1A). According to the NTA results, the level of plasma exosomes was higher in both the survived and non-survived HS group than in the control group (6.10x10⁹ and 3.37x10⁹ vs. 0.66x10⁹ particles/ml), while the diameter distribution was comparable between the two groups, the peak value was 76 nm in the control group vs. 78 nm for the survived HS group and 67 nm for the non-survived HS group; the proportion of particles with a diameter of 30-200 nm was 94.7% in the control group vs. 78.2% for the survived HS group and 97.1% for the non-survived HS group (Fig. 1B). The proportion of particles with a diameter >200 nm was much higher in the HS group. It can be hypothesized that this may be due to the increased presence of contaminant microparticles/microvesicles in the HS group as a result of increased cell injury/death. Western blot analysis indicated that the characteristic exosomal surface markers CD9, CD63 and CD81(22) were positively expressed in both the control and survived and non-survived HS groups (Fig. 1C). Therefore, the characteristics of the extracted vesicles were consistent with those of plasma exosomes.

The levels of histone H3 in exosomes and free histone H3 in plasma were detected by ELISA. Histone H3 levels in plasma exosomes on days 1 and 4 after admission were significantly higher in HS non-survivors than in both HS survivors and healthy controls (all P<0.001). On admission, a 2.0-fold increase in histone H3 levels was observed in non-survivors in comparison with survivors. However, there were no significant changes in plasma exosome levels of histone H3 between survivors and healthy controls (P>0.05). Additionally, on day 4, histone H3 levels in plasma exosomes significantly increased in non-survivors (P<0.001) but decreased, though not significantly in survivors (P>0.05), compared with levels on day 1. Moreover, levels of free histone H3 in plasma on days 1 and 4 were significantly higher in HS non-survivors than in healthy controls (all P<0.001). However, no significant changes in plasma exosomal levels of histone H3 were observed between survivors and healthy controls (P>0.05) on both days 1 and 4. On day 4, levels of free histone H3 in plasma were elevated in non-survivors (P<0.001) but remained unchanged in survivors (P>0.05), compared with levels on day 1 (Fig. 2).

The histone H3 levels in plasma exosomes on ICU admission were significantly positively correlated with numerous organ function parameters, including HR, lactate levels, ALT levels, creatinine levels, BUN levels, prothrombin levels, INR, and D-dimer levels, as well as APACHE II and SOFA scores on days 1 and 4; AST, TBil, myoglobin, and troponin-I levels on day 1; and FDP levels on day 4. By contrast, the plasma exosomal levels of histone H3 were negatively correlated with PaO₂/FiO₂ ratios, urine output, fibrinogen levels, platelet counts, and GCS scores on both days 1 and 4 and with albumin levels on day 4 (Table IV). However, all correlations between plasma exosomal levels of histone H3 and organ function indicators on day 1 were poor (all r<0.6), except for that with D-dimer levels (r=0.78), whereas the correlations between plasma exosomal levels of histone H3 and organ function indicators improved on day 4.

The area under the receiver operating characteristic (ROC) curve of plasma exosomal levels of histone H3 for discriminating between non-survivors and survivors was 0.9668 [95% confidence interval (CI), 0.9231-1.014; P<0.001]. At a cutoff value of 307 pg/100 µg, the sensitivity and specificity for predicting mortality risk were 95 and 91.67%, respectively (Fig. 3). The area under the ROC of exosome histone H3 was higher than both plasma AST (0.7882; 95% CI 0.5602-1.016; P=0.01158) and similar with ALT (0.9028; 95% CI, 0.8121-0.9935; P<0.001; Fig. 3).