Visualization of the inflammatory response to injury by neutrophil phenotype categories

Patients with severe injuries are at risk for (severe) infectious complications. Due to advances in the organization of trauma care, hemorrhage control, surgical approaches and resuscitation, overall mortality after trauma has declined up to 75% in recent decades [1‐3]. With increased early survival rates, up to 30–50% of multitrauma patients develop an infectious complication, which creates new challenges in terms of diagnostics and treatment in longitudinal management of trauma patients [4]. Infections are now the most common complications affecting trauma patients and are an increasing and substantial cause of morbidity, contributing to a mortality rate of 5–8% after trauma [1, 5, 6]. Prior research has shown that the risk of infections after trauma is determined by injury-related tissue damage leading to an inflammatory response initiated by damage-associated molecular patterns (DAMPs) [7, 8]. The innate immune system, in particular the neutrophil compartment, is essential in tissue repair and in the first line of defense against invading pathogens. Therefore, these cells are very sensitive for activation by DAMPs [9‐11]. An imbalance in the neutrophil compartment after trauma and a subsequent dysregulation of the immune response predisposes patients to infections [12].

Recently, it became possible to measure the neutrophil phenotype at the patient’s bedside in the acute setting with an automatic point-of-care flow cytometry approach [13]. In our center, the phenotypical and functional analysis of the neutrophil compartment has been integrated into clinical practice, allowing early immune monitoring to aid diagnosis and prognosis of aberrant innate immune responses. This could help in treatment choices in this very heterogenous patient population.

During acute inflammation, neutrophils can be divided into different subsets, based on the expression of specific surface proteins (CD16/FcγRIII and CD62L/L-selectin) [14]. A correlation was shown between injury severity and the percentage of circulating CD16^dim neutrophils [15]. In addition, we have previously shown that a decreased responsiveness of neutrophils to formyl-peptides (fMLF) immediately after trauma was related to the development of septic shock 5–8 days later [16]. Both parameters provide information on the inflammatory response to injury.

The primary objective of this study was to define immunophenotype categories, based on visual recognition of neutrophil subsets by CD16/CD62L expression, and to investigate by algorithm based analysis if these categories are valuable to interpret the inflammatory response to trauma. The secondary goal was to determine if the immunophenotype categories correlate with potentially pre-existing patient factors, the severity of injuries, and clinical outcome of trauma patients, in terms of infectious complications, duration of hospital admission and mortality.

Included patients were trauma patients of 18 years and older, presented at the trauma bay in the emergency department of the University Medical Center Utrecht (UMCU). The criteria for inclusion were trauma patients with injuries of any severity who underwent diagnostic blood sampling as part of the standard-of-care diagnostic workup during acute trauma care. No exclusion criteria based on patient characteristics nor injury severity were formulated, because the focus of this project was to analyze the immune response as early as possible after trauma, and to investigate if patient and/or injury characteristics affect this inflammatory response. Patients who were discharged from the hospital directly after presentation were identified as a non-significantly injured reference group. Patients transferred from other hospitals to the UMC Utrecht > 24 h after trauma were excluded because the first inflammatory response would not reflect a measurement < 24 h after injury. Patients without interpretable results of measurement on the flow cytometer were also excluded. Examples of uninterpretable results are a too low number of analyzed cells, a malfunction or technical error of the machine (e.g., a clog or bubble in the system, or failed lysis of red blood cells), or missing activation data. Immunocompromised patients were not excluded, because the possible impact of immunosuppressive medication on the inflammatory reaction to trauma has not been investigated earlier.

In total, 417 trauma patients were identified based on trauma registration and measurement on the flow cytometer, of which 380 were included for analysis. After inclusion, one measurement had to be excluded due to an incorrect display of values. For 20 patients, follow-up was not possible, as these patients were transferred to another hospital after admittance and acute care in the UMCU trauma bay. As shown in Fig. 1, three different groups of patients were defined: all trauma patients (group A), who were then divided into: the significantly injured patients, as defined by their admittance to the hospital, (group B) and the patients without significant injury who were not admitted (group C). An overview of patient characteristics is described in Table 1.

Immunophenotype categories based on visual pattern recognition of neutrophil subsets, as measured in a fully automated flow cytometer, are feasible and easily clinical applicable to provide insight into the inflammatory response immediately after trauma. Cluster analysis, performed using FlowSOM, showed that these predefined categories are valuable, as the amount of banded (MC2) and hypersegmented (MC3) neutrophils and the responsiveness to fNLF differed significantly between the different categories (see Table 2). This supports the concept that visual interpretation of 2D FACS dot plots holds valuable information about different neutrophil populations. Visual interpretation can easily be done by a clinician and can hold valuable clinical information about the immune status of trauma patients. However, it would be preferable to analyze visual information in a more objective manner, e.g., by artificial intelligence or with machine learning programs. This could not yet be achieved and requires further research with much larger numbers to make this clinically applicable. On the other hand, the use of these visual categories is a first approach that is useful to interpret flow cytometry results after trauma by the clinician directly at the bedside, in the same manner as interpreting e.g., an X-ray or CT scan.

Our results demonstrated that Categories 3–6 were each related to a more severe level of injury and a greater risk of infectious complications after trauma. This is in line with previous studies based on CD16^dim neutrophils and its clinical consequences (i.e., mortality, hospital stay) [15]. Furthermore, the outcomes of Category 6 confirm previous findings that this is an extreme immunophenotype that occurs only in patients in extremis (e.g., severe shock, traumatic resuscitation, full drowning) and is associated with minimal chance of survival.

As no relation was found between the patient characteristics independent of trauma and the immunophenotype Categories 3–6, it is likely that the appearance of these cells is a direct consequence of the inflammatory response initiated by tissue damage, after an injury-severity threshold is reached. This finding reinforces the fact that the impact of injuries and the subsequent tissue damage determines the inflammatory response, which can be visualized in a fast and non-biased way at the bedside by the characterization of neutrophil subsets in the circulation. Measuring the immunophenotype of severely injured patients immediately after trauma might be the next step to improve personalized and point-of-care decision making in trauma care. Moreover, the course of the inflammatory response is dynamic in time and is driven by both patient characteristics, injury type and severity, timing and extent of surgery. Further determination of these changes over time will increasingly facilitate personalized care of injured patients. As it is now possible to measure the neutrophil functional phenotype at any time point after trauma, the longitudinal course of the inflammatory response and the impact of additional interventions can be followed. In the future, measuring the neutrophil compartment over time could possibly guide decision making regarding timing and extent of immune protective surgery and determine the administration of (preventive) antibiotics [22].

Second, our results showed that patients in Categories 0–2 had similarly low injury severity scores. Thus, it is tempting to speculate that patients in Categories 0–2 display immune profiles that are not significantly initiated by tissue damage and the subsequent release of DAMPs into the circulation. These patients may display such immune phenotype intrinsically, regardless of any injury. Supporting this hypothesis is the finding that, even though the differences were not significant, patients in Category 1 seemed to be more prone to infectious complications than patients in Category 0 and Category 2 (see Table 3). The patient and injury characteristics give no explanation for the different outcome in this group. Further research is warranted into such putative personal immune profiles or individualized immune responses to limited inflammatory stimuli and the factors influencing them. Such inherent profiles can be studied in patients who have completely recovered from their trauma, as ‘before-measurements’ are impossible in trauma patients. If these immune profiles prove to provide insight in a personal immune response and a subsequent risk of infectious complications, it would be of value to assess the immunophenotype for every trauma patient presented at the trauma bay with suspected injuries that require full diagnostic work up.

The patients in Category 1 (with CD62L^dim cells) showed a decreased responsiveness to fNLF and seemed more prone to develop infectious complications compared to patients in Category 0, 2 and 3 (see Table 2). Previous studies demonstrated a relation between a decreased responsiveness towards fNLF immediately after trauma and severe sepsis after a week [23]. In addition, the subset of neutrophils characteristic for Category 1 (CD62L^dim) was found to be less capable of preventing bacterial outgrowth because of impaired intracellular bacterial containment [14]. Together with the refractoriness to stimulation with formyl-peptides, this can lead to a clinical situation that is associated with a higher risk of infectious complications after trauma [16]. The reason why the patients in Category 1 display this phenotype of immune cells remains to be identified.

The immunophenotype found in patients in Category 2 represents a subset of neutrophils that has not yet been described in literature, nor was it noticed in our previous research. Our data demonstrated that patients in this category were younger than patients in the other categories, and none developed infectious complications. Although the small sample size of this category should be kept in mind, this might suggest the presence of a neutrophil subset with a different functionality. It may be that these cells express other activation markers than CD16/CD62L that could give us further information about the cells. The neutrophil activation marker CD11b was analyzed, but no difference in expression of this marker between the immunophenotype categories was observed. To gain more insight into the subset of neutrophils in Category 2, these cells should be sorted, and functional assays must be performed to assess whether or not they show differences in functionality as well. Moreover, the cytospins of these cells should be made to obtain more information about the morphology of these cells. However, these measurements were beyond the scope of this current project.

Earlier studies suggested that neutrophil subtypes are involved in the pathogenesis of infectious complications after trauma [7, 14, 23, 24]. Understanding the functions and characteristics of neutrophil subtypes is important to understand their involvement in the development of infectious complications. Our results demonstrated that patients with extended neutrophil subsets in their blood after trauma were more severely injured and had a higher risk of infectious complications. Prior research described CD16^dim neutrophils as an immature neutrophil population with a higher bacterial containment capacity than other neutrophil subsets [25]. This supports the hypothesis that trauma can lead to massive mobilization of the best functioning (CD16^dim) neutrophils from the bone marrow. However, this might lead to a subsequent imbalance of the neutrophil compartment when this process takes too long, as the bone marrow will be unable to fully compensate in time for the loss of these well-functioning neutrophils [26]. This might lead to a relatively high sensitivity to infectious complications [27]. A second explanation is the partial desensitization of the bone marrow to subsequent stimuli such as intruding (bacterial) pathogens after the first hit of major injury and the recruitment of young neutrophils [20]. In previous research it was shown that this type of immunoparalysis could be simulated in a two-challenge LPS endotoxemia model, where indeed fewer CD16^dim cells were recruited after a second challenge [28]. This mechanism might also be of relevance in trauma patients suffering from infectious complications.

One of the objectives of this project was to determine if the immunophenotype categories correlated with pre-existent patient and injury characteristics. One of the patient characteristics that possibly determine the inflammatory response is age. Earlier research described elevated plasma concentrations of neutrophil stimulating cytokines in a healthy elderly population, contributing to an enhanced inflammatory status created by continuous stimulation of the innate immune system generally referred to as inflamm-ageing [29]. This could lead to a lower capacity of the immune system to liberate the correct numbers of the different neutrophil subsets during acute inflammation. This is in line with studies showing an age-related decrease in phagocytic- and bactericidal capacity in neutrophils [30]. However, in this study no significant differences in immunophenotype categories in the elderly population were observed. Further research into the neutrophil compartment of the elderly trauma patient should be conducted in a larger study population to further investigate these changes in the aging immune system. Furthermore, it could be hypothesized that different types of injury (penetrating or blunt) or even injury in different body regions (e.g., thorax trauma, isolated brain injury or a fractured femur) could lead to a different amplitude of inflammatory reaction to the tissue damage. This was previously shown in a population with either blunt or penetrating thorax trauma [31]. As for the elderly population, further research should be conducted into the reaction of the immune system to different types of injuries in a larger prospective cohort.

This study is unique in the fact that a large amount of reproducible flow data were generated by the easy-to-use point-of-care “load&go” flow cytometry directly after trauma. The studied trauma population is very heterogenic, containing a wide range of severity of injuries. The approach of including all patients admitted at the trauma bay, regardless of injury severity, enabled the analysis of a broad spectrum of patient characteristics in the context of the inflammatory response. ISS was not used as an inclusion criterion, because ISS is a poor predictor of tissue damage, poorly reflects physiological challenges such as shock a drowning, and possibly underestimates the burden of injury due to tissue damage, which is better reflected in the inflammatory response. Furthermore, ISS is calculated mostly after several days. Our data show that, although patients who develop infectious complications have a significantly higher ISS, there is a wide range of outliers, from ISS 1 to ISS 75. This indicated that the ISS did not perform well as a specific predictor for risk of infectious complications. Our hypothesis was that individual patients display an individual inflammatory profile as a response to trauma. However, no correlation was found between host-dependent variables and inflammatory response to trauma. A limitation of this retrospective cohort is that the study population represents the more severely injured patients admitted at the trauma bay of a level one trauma center such as the UMC Utrecht. The included population had a longer in-hospital and intensive care unit stay and the percentage of patients who died in the hospital was higher in the included patients compared to all trauma patients admitted at the UMCU during the year 2020 (see Table 4). These differences can be explained by the exclusion of all patients under the age of 18 years and due to the fact that drawing blood samples is less necessary in patient without severe injuries.

In conclusion, the distribution of neutrophil subsets described in phenotype categories is valuable and easily recognizable for clinicians at the bedside. No correlation was found between patient-specific variables (e.g., age, sex, comorbidity) and inflammatory response as described by neutrophil phenotype categories, indicating that the investigated patient characteristics did not contribute to a specific inflammatory profile. After a certain threshold, injury severity was related to the immunophenotype Categories 3–6. Furthermore, patients in higher categories had a significantly higher risk of developing infectious complications. This interpretation of the immunophenotype of a patient immediately after trauma is a fast and non-biased way of visualizing injury severity and the subsequent inflammatory response at the bedside, which might be the next step to improving personalized and point-of-care decision-making in trauma care.