HMGB1, nucleosomes and sRAGE as new prognostic serum markers after multiple trauma

Introduction

Trauma represents the leading cause of mortality for people younger than 45 years in the United States and worldwide [1]. The correct treatment of multiple trauma patients is still a medical challenge, especially during the first hours. There is only a narrow timeframe (the “golden hour”) to handle the life-threatening complications that is mostly decisive for the further prognosis [2]. Hence the fast and reliable assessment of the severity of the traumatic lesions and the physical condition is of prime importance. It should be estimated several times, to detect changes and complications at the earliest time possible. Currently, monitoring of vital function and different trauma scores based on clinical parameters are used most frequently. Many complications are caused by posttraumatic immunologic changes. Endothelial cell damage, disseminated intravascular coagulation, accumulation of leukocytes and microcirculatory dysfunction lead finally to apoptosis and necrosis of parenchymal cells [3]. Those conditions can also cause the dysfunction and failure of remote organs that themselves had not directly been injured [3]. Some frequent complications like multiple organ dysfunction syndrome or systemic inflammatory response syndrome are early mirrored by biochemical changes in the blood – often in a clinical still inconspicuous stage [4]. A better understanding of the pathological processes and the availability of clinical reliable indicators and prognostic markers is highly needed for better individual adaptation of future therapies [3]. Blood levels of cell death parameters such as nucleosomes and DNA as well as immunologic mediators such as high mobility group box 1 (HMGB1) and the receptor of advanced glycation end products (RAGE) might provide valuable prognostic information.

Nucleosomes are structural units, composed of a central histone octamer and DNA, that organize the eukaryotic genome [5]. They are released from dying and stressed cells into the blood circulation [6]. Concentrations of circulating DNA are found to be elevated in acute diseases like trauma [7, 8], stroke [9], and sepsis [10], as well as in cancers and autoimmune diseases [6].

The nuclear protein HMGB1 has be known for decades as nuclear DNA binding protein that stabilizes nucleosome structure and facilitates DNA transcription [11]. It is passively released during cell death and can also be secreted by activated macrophages and monocytes, to act as signaling molecule [12, 13]. Recently, plasma HMGB1 was reported to be released early after traumatic injury and may be integral to the early inflammatory response [13–15]. In the blood, HMGB1 acts as damage associated molecular pattern (DAMP) that leads to the maturation of dendritic cells and the stimulation of macrophages and T-cells [13, 14, 16]. Those effects have been observed to be even stronger when HMGB1 is present in complexes with nucleosomes, lipopolysaccharides or other binding partners [13, 17, 18].

Besides the toll like receptors 2 and 4, RAGE is an important receptor and binding partner for HMGB1 on those immune cells [12–14]. RAGE is a transmembrane, cell surface, multiligand receptor that mediates chemotaxis and stimulation of cell growth, differention of immune cells, migration of immune cells, and upregulation of cell surface receptors, including RAGE itself [16, 19–21]. RAGE has also a secretory splice isoform, that lost its transmembrane domain and therefore circulates as soluble form in the blood [21]. The functions of soluble receptor of advanced glycation end products (sRAGE) are not fully understood. It might be that it reflects only the expression rate on the cellular surface or that it acts as a decoy receptor to neutralize the effects of HMGB1. Thus, the soluble form may act as endogenous protection factor against RAGE-mediated damage [21].

In the present study, we investigated serum levels of these biomarkers of immunogenic cell death in patients with acute trauma to test their relevance for the estimation of the severity of traumatic lesions and for prognosis of early mortality in trauma patients.

Patients and methods

Results

Data distribution

Of the 164 patients who were included, 64 had SMT 51 moderate multiple trauma (MMT), 24 single femoral neck fracture (FNF) and 25 single AF.

Generally, we observed a considerable difference between the groups, particularly for HMGB1 and sRAGE. Levels of circulating nucleosomes were increased in all groups of diseased patients with medians of 220.5 ng/mL in SMT, 235.0 ng/mL in MMT, 298.0 ng/mL in FNF, and 155.0 ng/mL in AF. There was a considerable overlap of values in diverse groups and only a trend to lower values in AF patients (p=0.083). However, HMGB1 levels were significantly higher in patients with SMT (median 10.8 ng/mL) than with MMT (median 5.9 ng/mL; p=0.005), FNF (median 4.6 ng/mL; p=0.001), and with AF (median 3.6 ng/mL; p<0.001). Similarly, sRAGE levels were significantly higher in patients with SMT (median 2.14 ng/mL) than with MMT (median 1.07 ng/mL; p<0.001), FNF (median 1.07 ng/mL; p=0.003), and with AF (median 0.87 ng/mL; p<0.001) (Figure 1A–C, Table 2).

Figure 1:

Biomarker levels in diverse patient groups.

Distribution and median of serum levels of nucleosomes (A), HMGB1 (B) and RAGE (C) in diverse patient groups. D) Correlation of HMGB1 and nucleosomes (rhombs; R=0.56; p<0.001) and HMGB1 with RAGE values (triangles; R=0.44; p<0.001).

Table 2:

Biomarkers in various disease groups.

		Nucleosomes, ng/mL		HMGB1, ng/mL		sRAGE, ng/mL
	n	Median	Range	Median	Range	Median	Range
SMT	64	220.5	24.9–2195.0	10.8	0.7–260.0	2.14	3.69–500
MMT	51	235.0	13.5–3447.0	5.9	0.6–31.0	1.07	3.27–4.19
p-Value		p=0.48		p=0.005		p<0.001
FNF	24	298.0	37.1–2288.0	4.6	0.2–16.6	1.07	0.41–3.58
p-Value		p=0.47		p=0.001		p=0.003
AF	25	155.0	20.4–3442.0	3.6	1.2–10.9	0.87	0.43–1.96
p-Value		p=0.08		p<0.001		p<0.001
TBI in SMT
GCS≤8	28	262.0	54.9–2195.0	17.0	1.6–260.0	3.50	0.62–500
GCS>8	26	165.5	24.9–1098.0	7.4	0.7–144.0	1.30	0.37–69.3
p-Value		p=0.11		p=0.005		p=0.006
Survival in SMT
Yes	51	175.0	24.9–1922.0	8.1	0.7–53.0	1.49	0.37–69.3
No	13	404.0	74.6–2195.0	31.0	8.0–260.0	3.94	0.80–500
p-Value		p=0.01		p<0.001		p=0.02

Diagnostic significance

Differentiation between the various groups was best for HMGB1 and sRAGE. When SMT was compared with all other groups (MMT, FNF, AF), HMGB1 reached an AUC in ROC curves of 70.4%, sRAGE even 73.9% for identification of SMT. Sensitivities at 95% specificity were 27% for HMGB1 and 48% for sRAGE; at 100% specificity, sensitivities reached 13% and 34%, respectively (Figure 2A). When SMT was compared with controls having no multiple trauma (FNF, AF), HMGB1 reached an AUC in ROC curves of 75.8% and sRAGE of 74.1% for identification of SMT. Sensitivities at 95% specificity were 36% for HMGB1 and 55% for sRAGE; at 100% specificity, sensitivities reached 30% and 39%, respectively (Figure 2B).

Figure 2:

Diagnostic performance of biomarkers.

Receiver-operating characteristic curve for nucleosomes, HMGB1, RAGE for the discrimination of (A) severe multiple injured patients from all other control groups, (B) severe multiple injured patients from patients with femur or ankle fractures.

In the SMT-group, the association of the biomarkers with the severity of disease was tested by correlation with the ISS-score, while GCS-score was used to objectify the severity of TBI. There was a highly significant correlation of HMGB1 (R=0.44; p<0.01) and for sRAGE with ISS-score (R=0.56; p<0.01) while there was no association for circulating nucleosomes (R=0.13; p=0.28) (Table 2). Further, HMGB1 levels were significantly higher in patients with severe TBI (GCS≤8: median 17.0 ng/mL) than in those with mild or moderate TBI (GCS>8: 7.4 ng/mL; p=0.005). Similar results were obtained for sRAGE (GCS≤8: 3.50 ng/mL; GCS>8: 1.30 ng/mL; p=0.006) while there was only a trend to higher values for nucleosomes in more severely diseased patients (GCS≤8: 262.0 ng/mL; GCS>8: 165.5 ng/mL; p=0.115) (Figure 3A–C, Table 2). Best discrimination of TBI groups was found for HMGB1 (AUC 72.3%) and sRAGE (AUC 71.8%) in ROC analyses with sensitivities for severe TBI detection of 50% (HMGB1) and 21% (sRAGE) at 95% specificity (Figure 5A).

Figure 3:

Correlation of biomarkers with traumatic brain injury.

Distribution and median of serum levels of nucleosomes (A), HMGB1 (B), and RAGE (C) in group I patients with diverse grades of head trauma quantified by Glasgow Coma Scale values.

In addition, association of initial biomarker levels with early hospital mortaility was tested. Thirteen patients of the SMT-group died during the first 7 days. Significantly higher values were found in non-surviving (NS) than in surviving (S) patients for all three biomarkers, namely HMGB1 (median NS: 31.0 ng/mL; S: 8.1 ng/mL; p<0.001), sRAGE (NS: 3.95 ng/mL; S: 1.49 ng/mL; p=0.02), and also nucleosomes (NS: 404.0 ng/mL; S: 175.0 ng/mL; p=0.01) (Figure 4A–C, Table 2). Best discrimination of both groups was found for HMGB1 (AUC 90.6%) followed by nucleosomes (AUC 72.7%) and sRAGE (AUC 70.4%) in ROC analyses. HMGB1 reached sensitivities for non-surviving patients of 77% at 95% specificity and 39% at 100% (Figure 5B). Thus, all surviving patients had HMGB1 values below 55.0 ng/mL but 39% of non-surviving patients had considerably higher values. On the other hand, no non-surviving patient had low HMGB1 levels <8.0 ng/mL, however, there were about half of the surviving patients with HMGB1 <8.0 ng/mL that could nonambiguously be identified as survivers (specifity of 47% at 100% sensitivity) (Figure 5B).

Figure 4:

Correlation of biomarkers with prognosis.

Distribution and median of serum levels of nucleosomes (A), HMGB1 (B), and RAGE (C) in group I patients with diverse first-week survival in the hospital.

Figure 5:

Performance of biomarkers for estimation of disease severity and prognosis.

Receiver-operating characteristic curve for nucleosomes, HMGB1, RAGE for the discrimination (A) patients with severe head injuries (GCS≤8) from those with moderate head injuries (GCS>8) and (B) severe multiple injured patients surviving and non-surviving the first week after trauma.

Discussion

According to recent data of the German Society for Trauma Surgery, mortality of severe multiple trauma (SMT; ISS≥16) has declined from 37% to 18% during the last 30 years [27], which is in line with our findings (20.3%) and is explained by the enormous medical progress in trauma management and the widespread establishment of specalized trauma centers. However, unintentional injuries still lead the ranking of death causes in subjects younger than 44 years [1]. A further reduction of mortality needs better understanding of immunologic changes that occur after trauma to better adapt therapies on an individual level and to avoid posttraumatic organ failure [27].

In the present study, we investigated the clinical role of blood levels of circulating cell-death products that are supposed to be immunologically active and therefore are known as immunogenic cell death markers [12, 13, 17, 18, 21]. They are released in diverse pathologic conditions such as cancer, autoimmune and metabolic disease, as well as in sepsis and trauma [18]. Although the resulting immunogenic effect may depend on either or the sum of these biomarkers there are only rare reports on the combined investigation of these proteins on their role as blood biomarkers in any disease. As there is a need for new meaningful and easily measured biomarkers for early estimating prognosis and monitoring the treatment of trauma patients [3, 6, 27], we investigated, to our knowledge for the first time, the panel of these biomarkers in blood of patients with different severities of trauma.

The elevated levels of circulating nucleosomes after trauma correspond with earlier findings on high serum nucleosome levels in various acute and chronic diseases [6]. In cancer, they showed to be valuable for monitoring cytotoxic cancer therapy [6, 28]. Further, they were relevant for the estimation of disease extent and prognosis in patients with sepsis and stroke [9, 10]. Interestingly, values increased after trauma already within the first hours, but only 24 h after cytotoxic therapy [6, 28]. While most of DNA is supposed to circulate in blood in association with histones as nucleosomes [22] other forms such as cell-free DNA were found to be prognostically relevant after trauma as well [7, 8]. Lo et al., found multiple higher plasma DNA concentrations at a median time of 1 h after trauma in patients with adverse outcome, including acute lung injury, acute respiratory distress syndrome and death, than in those who did not develop these serious complications [7]. Margraf et al., investigated cell-free plasma DNA in fibrous extracellular chromatin traps released by stimulated neutrophils and reported cell-free DNA to be a valuable tool for the prediction of an inflammatory second hit, organ failure, and sepsis after trauma [8].

The DAMP HMGB1 is rapidly released after trauma, induces expression and secretion of proinflammatory cytokines and improves phagocytation and cross-presentation of pathogenetic cell death products after binding to toll-like receptors or RAGE [12, 13]. These immune-stimulatory effects are augmented if HMGB1 is associated with binding partners such as lipopolysaccharides or nucleosomes [13]. Further, HMGB1 was shown to be a potent mediator of triggering posttraumatic sterile inflammation. Marked elevations was observed from 2 to 6 h postinjury [14] and correlated with severity of trauma and patient outcome [15]. This is in line with our findings of early increases of serum HMGB1 levels on admission to the hospital and of the strong correlation with patient mortality. Similar results in a different setting were obtained by Beiter et al. [29] who reported on increased HMGB1 and DNA levels within 10 min after starting exhausting running and a rapid clearance from circulation after exercise stop. Interestingly we found a strong correlation of serum nucleosome and HMGB1 levels after trauma which points at a potentially combined release of both markers as observed after local cytotoxic therapies in cancer patients [28, 30]. Apart from diagnostic aspects, HMGB1 was concerned as relevant therapeutic aim in sepsis for application of anti-HMGB1 antibodies reversed sepsis lethality in a mouse model [31].

While cellular functions of RAGE as a receptor of advanced glycation end products and HMGB1 have been investigated extensively, little is known about the relevance of soluble RAGE in circulation. It have been found to be a measurable marker in venous blood sample, but its function as a soluble receptor is not fully understood. As high serum and plasma levels of sRAGE were found to be associated with a reduced risk of coronary artery disease, hypertension, peripheral arterial disease, metabolic syndrome, arthritis, cerebrovascular dementia and cancer [21, 32], a protective function of sRAGE was assumed. In cancer patients, high sRAGE levels were favorable for the patient outcome during cytotoxic therapies as well [18, 28]. One possible explanation is that sRAGE acts as a decoy receptor circulating in the blood that binds and neutralizes DAMPs such as HMGB1 in these situations [13, 21]. In contrast, sRAGE plasma levels were found to be increased within already 30 min after trauma and to be associated with poor prognosis in terms of posttraumatic coagulopathy, hyperfibrinolysis and endothelial cell activation as Cohen et al. [15] recently reported in his study. Similarly, high sRAGE values were non-favorable prognostic markers in patients with severe sepsis [33]. Levels obtained within 24 h after onset of sepsis strongly correlated with 28 day mortality: While mean sRAGE levels were 1.3 ng/mL in survivors, non-survivors had significantly higher levels (2.3 ng/mL). These reports are in line with our findings of higher sRAGE levels being related to more severe trauma disease and poor patient outcome. Generally, sRAGE can derive from alternative splicing of pre-mRNA and subsequent release as endogenous secreted RAGE (esRAGE) or from proteolytic cleavage of surface RAGE (cRAGE), which is both comprised in the determined sRAGE [21]. Application of more specific immunoassays will help to reveal which portion contributes most to elevated sRAGE levels after trauma.

To objectify the severity of trauma, we classified the patients according the well established ISS being aware of its obvious shortcomings: diverse combinations of injury severity and specific injuries may lead to identical ISS values despite different risks of mortality [25, 34]. Nevertheless, HMGB1 and sRAGE levels correlated significantly with this score. Similarly the GCS score was used to assess TBI, which is the most important mortality factor after trauma [2]. It is used for therapeutical decision making and for clinical studies [26]. Once again, high HMGB1 and sRAGE levels were correlated with severe TBI and tendencially also nucleosomes. This corresponds with earlier findings on high amounts of serum nucleosomes after cerebral stroke being correlated with infarction volume and non-favorable patient outcome [9].

Of particular interest is the close relationship of initial biomarker levels with early hospital mortality: High levels of all three biomarkers correlated with poor first-week survival with best discriminative power was found for HMGB1 (AUC 90.6%). Notably, sensitivities of 77% at 95% specificity and 39% at 100% were obtained meaning that all surviving patients had HMGB1 values below 54.0 ng/mL but 39% of non-surviving patients had considerably higher values. Further, no non-surviving patient had low HMGB1 levels <8.0 ng/mL (100% sensitivity) but 47% of the surviving patients. Thus, nonambiguous identification of non-survivors by very high values and of survivors by very low values was possible on the basis of HMGB1 serum levels.

Limitations of the present study rely on the cross-sectional study design that does not elucidate the causal relationship among serum biomarker levels, immune function and mortality. Further, inclusion criteria were set at 6 h after trauma which is a considerably long period in comparison with other studies. Indeed this long time period was chosen due to the delayed admission of control group patients. Multiple trauma patients were – in most cases – already admitted within 1–3 h after the event. It would have been ideal to determine biomarker levels in therapy-naive patients. As the blood drawing was only done in the resuscitation room, prior therapies could not be ruled out. However, we checked nucleosome content in erythrocyte concentrates and did not find elevated levels thereby assuring that there was no additional infusion from external nucleosomes.

It has to be emphasized that all patients were included prospectively, blood drawings and preanalytics until measurement were performed highly controlled according to standard operating procedures (SOPs). Measurements of the biomarkers were done in batches. Additional determination of pool sera and final crossplate check ensured a high inter-assay precision. Clinical and preanalytical data were collected in a consistent and standardized way. All parts of data and sample compilation, laboratory measurements and statistical evaluation were done independently. Notably, patients with single fractures as well as trauma patients of minor grade were used as controls confirming the high relevance of this new marker set for diagnosis and prognosis in multiple trauma patients.

Conclusions

In conclusion this study demonstrates, for the first time, the high diagnostic and prognostic value of the immunogenic cell death biomarker panel consisting of nucleosomes, HMGB1 and sRAGE in multiple trauma patients. Further, HMGB1 and sRAGE have shown to be helpful in classification of severity of trauma. Notably, subgroups of trauma patients with good and poor mortality prognoses could be identified on the basis of HMGB1 levels. If these findings are confirmed by other prospective trials, these markers can be valuable for future management of trauma patients.