Of all ICU patients included, 83 patients had a confirmed COVID-19 diagnosis and 11 patients did not have COVID-19 and were used as ICU control patients. We obtained a total of 315 plasma-samples from the 83 patients with confirmed COVID-19, while 18 samples from 11 patients were obtained from the ICU control group. The healthy control group consisted of 15 individuals with a median age of 32 (IQR 24–37) of which 7 were male (47%).
The baseline (day 1) characteristics of the 83 ICU patients are shown in Table 1. Between the COVID-19 and the non-COVID-19 patients, males were overrepresented in the infected group, with 71.1% (p = 0.030), whereas BMI was not different between both ICU groups. Non-COVID-19 patients tended to be older than the COVID-19 patients. To allow comparison with a healthy population, a healthy control group was further included in our analyses.
Table 1
Demographic and baseline characteristics of 94 patients on admission to the Intensive Care Unit. Values are represented as median (IQR) or n (%).The p-value is calculated for continuous parameters with the Mann-Whitney U test, and for categorical parameters the chi-square test; p < 0.05 is considered significant.
|
ICU Covid-19
n = 83
|
ICU Control
n = 11
|
P
|
Age, yrs, median
|
60 (52–71)
|
70 (59–75)
|
0.193
|
Gender, male N
|
59 (71)
|
5 (46)
|
0.030
|
BMI
|
28.8 (25.1–32.7)
|
27.7 (25.4–31.8)
|
0.829
|
Respiratory rate, breaths/min
|
27 (22–35)
|
15 (15–17)
|
0.001
|
Heart rate, beats/min
|
88 (77–100)
|
80 (72–92)
|
0.208
|
MAP, mmHg
|
90 (81–98)
|
80 (67–86)
|
0.004
|
Temperature, °C
|
38.0 (37.4–38.7)
|
36.4 (36.4–36.7)
|
< 0.001
|
Diabetes, yes
|
22 (27)
|
2 (18)
|
0.552
|
Hypertension, yes
|
41 (50)
|
4 (36)
|
0.375
|
Heart failure, yes
|
3 (4)
|
1 (9)
|
0.398
|
Ischemic heart failure, yes
|
8 (10)
|
0 (0)
|
0.282
|
Vessel disease, yes
|
12 (15)
|
1 (9)
|
0.628
|
Malignancy, yes
|
4 (5)
|
11 (100)
|
< 0.001
|
HIPEC surgery, yes
|
0 (0)
|
11 (100)
|
< 0.001
|
Pulmonary disease, yes
|
21 (25)
|
1 (9)
|
0.233
|
Asthma
|
14 (17)
|
1 (9)
|
|
COPD
|
6 (7)
|
|
|
Sarcoidosis
|
1 (1)
|
|
|
Smoker No
|
67 (80.7)
|
9 (81.8)
|
0.721
|
Yes
|
3 (3.6)
|
0 (0)
|
|
Previous
|
10 (12.0)
|
2 (18.2)
|
|
Unknown
|
3 (3.6)
|
|
|
Steroid treatment, yes
|
9 (11)
|
1 (9)
|
0.850
|
ACEi/ARB treatment, yes
|
29 (35)
|
2 (18)
|
0.256
|
Anticoagulant treatment, yes
|
16 (19)
|
5 (46)
|
0.050
|
SAPS-3
|
52 (47–58)
|
54 (49–58)
|
0.568
|
SOFA
|
6 (4–8)
|
6 (6–8)
|
0.980
|
Physiological characteristics are also shown in Table 1. Between the COVID-19 and the non-COVID-19 patients, significant differences were observed in their clinical parameters, with respiratory rate, mean arterial pressure (MAP) and body temperature being higher in the COVID-19 group.
Prior to ICU admittance, malignancy, representing the main indication for ICU admission for HIPEC surgery of the non-COVID-19 group and anticoagulant treatment were significantly higher in the control ICU group (p < 0.001 and p 0.050 respectively). All patients in both patient groups required supplemental inspired oxygen (Supplemental Table 1), but in the ICU control patients no high-flow nasal oxygen (HFNO) or oxygen masks were required, yet the latter group received more vasoactive treatment. No significant differences were present in antibiotic use. Supplementary Tables 2 and 3 presents routine parameters measured for most COVID-19 patients at study inclusion. The PaO2/FiO2 ratio, a clinical indicator of respiratory dysfunction, was on average 149 mmHg (IQR 123–186 mmHg) in the COVID-19 group. Leukocytes were increased in the ICU control patients whereas erythrocytes and thrombocytes were within reference range for both patient groups. A significant rise in CRP, IL-6 (n = 40), procalcitonin, and lactate dehydrogenase (LDH) was seen in the COVID-19 patients, indicative of the ongoing inflammatory response and tissue damage. All these measures had the lower interquartile range above the reference range. D-dimer, indicative of fibrinolytic activation, was also above the reference range, although in the 19 patients where aPTT was measured, the median value was prolonged but within reference range.
Detection of Histone H3, cfDNA and NE in plasma.
Plasma samples from the 94 ICU patients, were analyzed together with those of 15 healthy volunteers (Fig. 1). In the majority of patients (67%), no extracellular H3 could be detected at ICU admission. As compared with plasma samples of the ICU control patients however, the levels of extracellular histone H3 in the COVID-19 patients on day 1 were significantly higher (p = 0.047), as well as compared to the healthy controls (p = 0.025; Fig. 1A), while there was no significant difference between the ICU control and healthy control group (Table 2). The cfDNA values from both ICU groups differ greatly (p < 0.001), with the COVID-19 group presenting 25-times higher levels than of the control group (Fig. 1B). The levels of cfDNA did not differ between the ICU control and healthy control patients (Table 2). There was a highly significant difference between both groups when NE was determined (p < 0.001; Fig. 1C) with NE being virtually absent from the ICU control group and the healthy controls (Table 2).
Table 2
Baseline ICU patients and healthy control plasma parameters expressed as median (IQR).
|
|
ICU COVID-19
n = 83
|
ICU Control
n = 11
|
Healthy Control
n = 15
|
P
|
Histone H3
|
µg/ml
|
0.0
(0.0–0.3)
|
0.0
(0.0–0.0)
|
0.0
(0.0–0.0)
|
0.004
|
cfDNA
|
ng/µl
|
225.8
(125.7–456.3)
|
7.8
(5.9–17.1)
|
4.1
(2.8–5.3)
|
< 0.001
|
Gas6
|
ng/ml
|
24.8
(18.1–33.3)
|
12.1
(9.5–15.9)
|
14.4
(11.0–19.7)
|
< 0.001
|
sAxl
|
ng/ml
|
19.8
(14.6–26.1)
|
13.2
(11.3–17.6)
|
17.3
(13.7–18.4)
|
0.018
|
NE
|
ng/ml
|
71.9
(32.4–134.9)
|
0.0
(0.0–0.0)
|
0.0
(0.0–0.0)
|
< 0.001
|
P-values were calculated with the Kruskal-Wallis test with Dunn’s post-hoc test. |
Detection of GAS6 and sAXL in plasma.
In COVID-19 and non-COVID-19 ICU patients, the GAS6 concentration at study inclusion was 24.8 (18.1–33.3) ng/mL and 12.1 (9.5–15.9) ng/mL respectively, indicating a significant difference between both groups (p < 0.001; Fig. 2A, Table 2). The concentration of the healthy control group 14.4 (11.0–19.7) ng/mL is significantly lower (p < 0.001) compared to the COVID-19 ICU group, while no difference is found with the non-COVID-19 ICU group (Table 2).
The concentration of sAXL at inclusion between the ICU groups being 19.8 (14.6–26.1) ng/mL and 13.2 (11.3–17.6) ng/mL respectively, showed a significant increase (p < 0.001) of the COVID-19 group compared to the non-COVID-19. The concentration of the healthy controls 17.3 (13.7–18.5) ng/mL was lower than that of the COVID-19 group, though this difference was not significant.
Correlations between different plasma markers and laboratory parameters.
Table 3
Correlations between various parameters measured at day 1.
|
N
|
Correlation
|
P
|
H3 vs. cfDNA
|
83
|
0.367
|
0.001
|
H3 vs. sAXL
|
83
|
0.245
|
0.026
|
H3 vs. NE
|
83
|
0.391
|
< 0.001
|
cfDNA vs. Leukocytes
|
79
|
0.235
|
0.037
|
cfDNA vs. Platelets
|
79
|
0.236
|
0.036
|
cfDNA vs AST
|
67
|
0.262
|
0.032
|
cfDNA vs. LDH
|
64
|
0.506
|
< 0.001
|
sAXL vs. Troponin I
|
65
|
0.355
|
0.004
|
GAS6 vs. Procalcitonin
|
72
|
0.246
|
0.038
|
Correlations were calculated with the Spearman’s rank-order correlation test. Correlations were considered significant if P < 0.05, only significant correlations are mentioned here. |
The concentration of H3 correlated with cfDNA and NE concentrations at ICU admission in COVID-19 patients (Table 3). At this point, there was no significant correlation between cfDNA and NE measurements. In contrast, LDH correlated very well with cfDNA (0.506; p < 0.001), which could indicate a common origin in cellular damage. Remarkably, sAXL correlated significantly with the myocardial injury biomarker troponin I in this cohort.
These correlations persisted when all subsequent sampling was included (supplementary Table 4). In addition, other correlations with biochemical parameters indicating organ damage were observed, including H3 and cfDNA with AST, LDH and procalcitonin (weaker for H3 in the latter case; p = 0.029). Interestingly, cfDNA correlated also with D-dimer and CRP when all points were considered, indicators of ongoing coagulation and fibrinolysis. Although NE correlated strongly with H3 as was the case with the initial values, NE did not correlate significantly with other tissue damage parameters, nor with cfDNA, while it correlated significantly with leukocyte and neutrophil cell count, as well as procalcitonin and GAS6 (Supplementary Table 4).
Prognosis and time-dependent parameter development in COVID-19.
At ICU admission, the value of SAPS-3 score (but not SOFA, supplementary Fig. 1) was significantly higher in the group of COVID-19 patients that did not survive (n = 21) compared to those that survived (n = 62; supplementary Table 5). Non-survivors were older, hypertensive (86%), and had vessel or ischemic heart failure more frequently than survivors. Accordingly, they were more frequently treated with anticoagulants and ACEi/ARB treatment. They also suffered more thromboembolisms during ICU stay. From the biochemical measurements performed, non-survivors had higher markers of organ damage, including AST, troponin I and procalcitonin (Supplementary Table 6). In those patients where IL-6 was determined, mean IL-6 concentration in non-survivors (n = 10) was almost 3 times that of survivors (n = 30). Although cfDNA, H3, NE, sAXL and GAS6 were all elevated in COVID-19 positive patients at admission, none of these parameters was a good predictor of final outcome. Indeed, at ICU admission, there was no correlation of any of these parameters with SAPS-3 or SOFA scores.
In 22 patients where several samples could be obtained during their ICU stay, we studied the time-dependent development of the markers. We arbitrarily divided the results in an early sample (day 1–5) and a late sample (day 6–12), using the mean value of the different samples obtained during each period (supplementary Table 7). Platelet counts, in the normal range in the baseline measurements, increased significantly in the late samples, as well as leukocytes (supplementary Table 7). Markers of tissue damage also increased, including ALT and AST. There were no significant differences between early and late values of histone H3, cfDNA or NE, although more samples were positive for H3 in the latter group. No differences were observed for sAXL or GAS6. We furthermore inspected if the correlations observed in the initial samples persisted over the early (day 1–5) and late phase of ICU stay for the 22 COVID-19 patients for which data were available over this complete period of time (Supplementary Table 8). We observed that H3 correlated with cfDNA over both time periods and generally showed persistent correlation with neutrophil counts. Strong correlations were seen between NE levels and neutrophils (0,810 and 0,708 respectively for early and late phase).
Next, we divided the 22 patients according to their final outcome at 30 days (Fig. 3). Those individuals that survived (n = 15) showed a 40% decrease in cfDNA concentration in the late plasma samples. Further, in these patients GAS6 concentration decreased more than 30% in late samples. In contrast, neither cfDNA nor GAS6 decreased in the group of non-survivors (n = 7), with their late values being very similar to the early ones. The same trends were observed in H3 and NE measurements, although in these cases, it did not reach significance. In addition, when regarding the NE concentrations, we found both in the early phase, and in the late phase a significant difference between survivors and non-survivors.
Histone Analysis
Extracellular histone H3 was detected exclusively in samples originating from COVID-19 patients. At ICU admission, 27 out of 83 COVID-19 positive patients (33%) had detectable histone H3 in plasma. When considered the whole group of 315 samples obtained at different time points, histones were detected in 135 (43%) samples obtained from 83 patients.
Table 4
Comparison of histone positive and histone negative samples.
|
Histone Positive
(n = 135)
|
Histone Negative
(n = 180)
|
P
|
Histone Cleavage
|
55 (40.7%)
|
0 (0%)
|
|
Histone H3 (µg/ml)
|
0.59 (0.14–1.39)
|
0.00 (0.00–0.00)
|
< 0.001
|
NE (ng/ml)
|
98.6 (60.0–159.0)
|
53.5 (19.6–109.6)
|
< 0.001
|
cfDNA (ng/µl)
|
555.0 (321.9–829.6)
|
244.8 (137.7–429.1)
|
< 0.001
|
Cleavage of Histone H3 and Plasma levels of Histone H3, NE and cfDNA in all samples of COVID-19 patients (n = 315), expressed as median (IQR). Samples are stratified in two groups depending on histone H3 presence. P-value is calculated with the Mann-Whitney U test; p < 0.05 is considered significant. |
Histone positive samples contained significantly higher NE and cfDNA than those in which no H3 was detected (p < 0,001 and p < 0,001 respectively, Table 4), in line with the strong correlation between these parameters (Table 3 and supplementary table 4). The presence of histone in COVID-19 patients appeared not randomly divided over time (Fig. 3). Although no statistical significance was reached for comparison between early and late survivors/non-survivors, the average histone levels in early survivors were higher than in late survivors, whereas average histone H3 levels in early non-survivors were lower than in late non-survivors.
In several COVID-19 patients, histone H3 showed an additional 13 kDa band on Western blot (Figs. 4A and 4B). This was seen in 37% of histone H3 positive patients at day 1 and overall in 40,7% of all samples tested. The origin of this band is most likely a proteolytic cleavage of the single 15 kDa band. In one patient, only the lower weight band was seen (patient 47, day 7, Fig. 4A), suggesting full H3 proteolysis. In 5 patients, the proteolysis disappeared in samples collected after 6 days or later of ICU stay.
In addition to presence of histones per se, we found that in a subset of 50 randomly selected COVID-19 patients with histone H3 in their plasma, the presence of citrullinated histone and found that 36 of these (72%) were positive for citrullination (Fig. 4C).