Introduction
Severe sepsis is an important cause of admission to intensive care units (ICUs) throughout the world and is characterized by high mortality in adults [
1‐
4]. But the pathogenesis of sepsis is still not clear. Previous reports suggest that the depression of the immune system may contribute to the severity of sepsis. Although the mechanistic and molecular bases for ICU-acquired immunosuppression are not exhaustively established, several features of the condition, including enhanced leukocyte apoptosis, lymphocyte anergy, and deactivated monocyte functions, have already been described [
5‐
7]. mHLA-DR has been suggested to be a reliable marker for estimating immunosuppression. The level of mHLA-DR was significantly decreased during severe sepsis [
8], although little is known about the underlying mechanism. The reduced expression was associated with high mortality and had a predictive value for the prognosis of patients with sepsis [
9‐
12] or the risk of secondary infection [
13‐
16]. However, the association between low mHLA-DR and mortality in severe sepsis has been challenged [
17]. These differences in findings may be partially explained by the fact that immune function is dynamically changing during the clinical course of severe sepsis. In this study, we monitored the expression of mHLA-DR during 1 week to evaluate the predictive power of serial determinations of mHLA-DR as a marker of mortality in severe sepsis. Our hypothesis was that ΔmHLA-DR would be more accurate than mHLA-DR in predicting 28-day mortality in severe sepsis.
Discussion
Sepsis is one of the 10 leading causes of death in critically ill patients in the US. It is the third leading cause of death among patients in non-coronary ICUs [
22,
23]. Severe sepsis develops each year in more than 750,000 people, 215,000 of whom die of the disease [
3], and is considered a disorder partly due to immunosuppression [
5]. The diagnosis of immunosuppression depends on paraclinical parameters because of the absence of specific clinical symptoms. Among these, monocyte expression of human leukocyte antigen type DR (HLA-DR) has been shown to be useful for monitoring immunoparalysis and accepted as a reliable marker for evaluation of immune function [
8,
15,
17]. Also, downregulation of mHLA-DR is generally accepted as a reliable marker for an immune dysfunction in patients with sepsis [
24].
Volk and colleagues [
12] were the first to describe immunoparalysis indicated by mHLA-DR expression in patients with sepsis. Abundant research has demonstrated the reduction of mHLA-DR expression in patients with sepsis [
8,
9,
13]. A recent study indicated that patients in severe trauma present with a transient immunosuppression with decreased mHLA-DR expression. The lack of mHLA-DR recovery between days 3 and 4 and days 1 and 2 is associated with sepsis [
25]. Furthermore, the prognostic value of low HLA-DR expression on monocytes has been elucidated, and the severity of the sepsis and mortality have been correlated with low HLA-DR expression [
8,
10,
15,
26,
27]. In recent years, it has been shown that patients with sepsis-induced immunosuppression were at a higher risk to develop secondary infections [
13]. In a prospective single-center observational trial, Landelle
et al. found that persistently low mHLA-DR expression is independently associated with the development of nosocomial infections [
14,
28]. Monocyte HLA-DR expression has also been successfully applied to monitor immunomodulatory therapies, including medications such as granulocyte/macrophage colony-stimulating factor (GM-CSF) [
29,
30], filgrastim [
31], thymosin alpha 1 [
32], and interferon-gamma [
33], as well as extracorporeal immune interventions such as immunoadsorption treatment and continuous hemodiafiltration [
34,
35].
In contrast, other research suggested that mHLA-DR was not a useful prognostic marker for outcome. A single-center study showed no significant difference in mortality between patients with low HLA-DR expression and those with normal HLA-DR expression [
17]. That study indicated that mHLA-DR did not give satisfactory discriminatory power to assist in an outcome prediction. Another study reported that the low HLA-DR expression was not an independent outcome predictor, because the correlation between outcome and early HLA-DR expression disappeared after adjustment of the severity of illness by Sequential Organ Failure Assessment score or Simplified Acute Physiology Score II [
13]. The contradictory results prompt clinicians to seek a more representative index for the connection between immune status and outcomes.
Although a number of recent studies have adopted static HLA-DR expression as a predictive marker, few of them have addressed the changes of HLA-DR expression during the disease progress. To address the fact that the immune function is dynamically influenced by severe sepsis and to compensate for the drawbacks of previous studies, we measured the expression of mHLA-DR consecutively to find out whether its change over time could predict mortality.
Our results indicated that mHLA-DR was significantly increased in the survivor group with the passage of time, but not in the non-survivor group. The findings were similar to those of Monneret and colleagues [
27]. In a study in septic shock, they found that mHLA-DR expressions were not significantly different between survivors and non-survivors at days 1 and 2. However, at days 3 and 4, the mHLA-DR expression had increased in survivors, but not in non-survivors.
We found that the AUCs of mHLA-DR3 and mHLA-DR7 for 28-day mortality in patients with severe sepsis were 0.629 and 0.598, respectively, with low specificity despite the relatively high sensitivity. In contrast, ΔmHLA-DR was a good predictor for the outcome of severe sepsis. Among patients with severe sepsis, those with increased ΔmHLA-DR expression higher than threshold had markedly lower mortality. ROC curve analysis showed that ΔmHLA-DR3 and ΔmHLA-DR7 were reliable indicators of mortality in severe sepsis with high sensitivity and specificity. Multivariate logistic regression analysis showed that ΔmHLA-DR3, ΔmHLA-DR7, and ΔmHLA-DR7-3 were all associated with a higher mortality. However, the wide range of CIs implied poor precision because of the relatively small size of the cohort. It was also found that ΔmHLA-DR7-3 was not as good as ΔmHLA-DR7 and ΔmHLA-DR3 either in predicting mortality or in representing the difference in change of mHLA-DR between survivors and non-survivors. A possible explanation is that ΔmHLA-DR3 and ΔmHLA-DR7 are calculated from a baseline on day 0 but that ΔmHLA-DR7-3 is calculated from a baseline on day 3, meaning that ΔmHLA-DR7-3 is affected by more confounding factors such as the treatment in the ICU, the deterioration or improvement of disease, and other conditions that have impacts on immune status. Overall, our study suggests that elevated ΔmHLA-DR expression (especially ΔmHLA-DR3 and ΔmHLA-DR7) may be seen as a marker for a gradually recovering immune function during the course of severe sepsis or a positive response to treatment and may indicate a better outcome.
A threshold of 30% is retained to predict mortality in published research that showed that non-survivors had an expression of mHLA-DR of lower than 30% [
9,
36]. However, in this study, there was no significant difference in 28-day mortality between patients with mHLA-DR expression of not more than 30% and those greater than 30% on days 0, 3, and 7, although non-survivors tended to exhibit lower mHLA-DR expression than survivors. At the same time, the present study found that mean expression of mHLA-DR in non-survivors was about 60%. This finding may be explained by the following facts: (a) our study included only surgical patients, who may have a relatively minor degree of immunosuppression compared with other critical patients with severe underlying diseases, and (b) to ensure that severe sepsis was the most likely cause for patients' immunosuppression in our study, we had excluded severely immnunosuppressed patients caused by other factors, including post-transplantation status and immunosuppressive therapy. It is assumed that different patient populations and selection criteria as well as the heterogeneity of septic host [
8] may contribute to the varied findings in different studies. Given the mixed results in different studies, a fixed static value of HLA-DR may not be appropriate to be applied in all patient populations and the dynamic change of HLA-DR seems more reasonable as an index for mortality.
The present study has certain limitations that need to be taken into account. This is a single-center study with a relatively small cohort, and the findings need to be confirmed by a multicenter study. In our study, mHLA-DR was expressed as percentages of HLA-DR-positive monocytes in the total monocyte population, and the measurement was reported to be reproducible with coefficients of variation from precision studies less than 5% [
37]. In spite of this, the value of delta HLA-DR should be separately interpreted because we cannot exclude the variability of measurements by flow cytometry. With the development of techniques, quantitative flow cytometry offers a better means of standardization within and between flow cytometers [
38]. With this method, results become comparable among different laboratories.
Given these limitations, our objective is not to establish a golden standard about delta mHLA-DR for predicting severe sepsis mortality considering the small cohort and heterogeneity in severe sepsis but to remind clinicians that the dynamic change of mHLA-DR may be a better parameter than static values in judging prognosis and evaluating efficacy of immunomodulatory therapies. Actually, similar indices such as delta central venous pressure, delta stroke volume, and lactate clearance are being widely used in the ICU.
Competing interests
The authors declare that they have no competing interests.
Authors' contributions
J-FW and JM designed and performed the research and wrote the manuscript. X-DG designed the research, provided the supportive work and supervision, and wrote the manuscript. JC, BO-Y, M-YC, L-FL, and Y-JL performed the research. A-HL analyzed the data. All authors read and approved the final manuscript.