Early immunosuppression was associated with poor prognosis in elderly patients with sepsis: secondary analysis of the ETASS study

Background: Although immunosuppression has been investigated in adult septic patients, early immune status remains unclear. In this study, we aimed to assess early immune status in adult patients with sepsis stratied by age and its relevance to hospital mortality. Methods: From post hoc analysis of a multicenter, randomized controlled trial, 273 patients whose levels of monocyte human leukocyte antigen-DR (mHLA-DR) were obtained within 48 hours after onset of sepsis were enrolled. All patients were divided into elderly ( ≥ 60yrs) group and non-elderly (<60yrs) group. Early immune status was evaluated by the percentage of mHLA-DR in total monocytes within 48 hours after onset of sepsis and it was classied as immunosuppression (mHLA-DR ≤ 30%) or non-immunosuppression (>30%). Changes in immune status were assessed by the value change in mHLA-DR on day 3 compared with the rst measurement. Three logistic regression models were conducted to test the associations between early immunosuppression and hospital mortality. We also did a sensitivity analysis to nd out if the denition of early immune status (24 vs. 48 hours after onset of sepsis) affects the outcomes. Results: Of the 181 elderly and 92 non-elderly septic patients, 71 (39.2%) elderly and 25 (27.2%) non-elderly died in hospital. The percentage of early immunosuppression in the elderly was twice of that of the non-elderly patients (32% vs. 16%, p=0.006). Immunosuppressed elderly had higher hospital mortality than the non-immunosuppressed elderly (53.4% vs. 32.5%, p=0.009), but there was no signicant difference in mortality between immunosuppresed non-elderly patients and non-immunosuppressed non-elderly patients (33.5% vs. 26.0%, p=0.541). In all of the three logistic regression models, we found that early immunosuppression was independently associated with increased hospital mortality in elderly, but not in non-elderly patients. Sensitivity analysis further conrmed the denition of early immune status did not affect the outcomes. In addition, immune status improvement on day 3 was associated with reduced hospital mortality in both elderly and non-elderly patients. Conclusion: In adult patients with sepsis, the elderly were more susceptible to early immunosuppression after onset of sepsis. Early immunosuppression was independently associated with poor prognosis in elderly patients. Trial registration: ClinicalTrials.gov


Introduction
Sepsis is de ned as life-threatening organ dysfunction caused by a dysregulated host response to infection [1]. After decades of effort, the mortality rate of sepsis has been decreasing; however, the absolute number of deaths is likely to continue to increase as the incidence of sepsis continues rising [2,3]. Although sepsis has been studied for decades, its pathogenesis remains unclear. A post-mortem study of septic patients demonstrated that patients who died of sepsis were associated with widespread immunosuppression [4]. Subsequently, numerous studies have demonstrated that immunosuppression is the main cause of high mortality in septic patients [5,6].
Decreased expression in monocyte human leukocyte antigen-DR (mHLA-DR) is a key biomarker to assess immune status. Our previous study and other studies have revealed that both mHLA-DR and dynamic change of mHLA-DR was associated with poor prognosis in patients with sepsis [7][8][9]. The percentage of mHLA-DR below 30% was widely accepted as immunoparalysis or immunosuppression [9,10].
Furthermore, mHLA-DR has been applied to select immunosuppressed septic patients in the study of immunostimulant [11]. Therefore, the use of mHLA-DR as a biomarker to assess immune status in septic patients is supported by previous research.
It is well known that the immune function declines with age, so elderly may be more susceptible to early immunosuppression than younger patients [12,13]. Elderly patients are more susceptible to sepsis during hospitalization, and the mortality of elderly patients with sepsis was higher than that of younger patients [14][15][16]. Several studies found immunosuppression in elderly patient increases the risk of death and secondary infection in the course of sepsis [17,18]. Although immunosuppression is associated with poor outcome in elderly septic patients, the timing of immunosuppression remains unclear. Recently, Muszynski et al. revealed that critically ill children with sepsis had immunosuppression from early stage (within 48 hours after onset of sepsis) [19]. However, early immune status of elderly patients with sepsis remains unclear.
In our current study, we aimed to assess early immune status in adult patients with sepsis strati ed by age groups and determine their relevance to hospital mortality.

Study design
The ETASS (E cacy of Thymosin Alpha 1 for Severe Sepsis, ETASS) study was a multi-center, randomized controlled study comparing the effect of thymosin alpha 1 (Tα 1) vs. placebo in patients with severe sepsis [20]. A full description of the methods of the ETASS study, including the full study protocol, case report form, sample size, quality control, and main results can be found in the original paper [20]. In the ETASS study, severe sepsis was de ned as the presence of a proven or suspected infection in at least one site, two or more signs of a systemic in ammatory reaction, and at least one acute sepsis-related organ dysfunction. Therefore, the term 'severe sepsis' in our previous study is approximately equal to the de nition of sepsis in Sepsis 3.0 [1]. Unless otherwise speci ed, 'sepsis' was used to replace 'severe sepsis' in this study. Immunotherapy in the study was de ned as patients have received at least one dose of thymosin alpha 1 in the ETASS study.
In the present study, the primary outcome was to assess early immune status in adult septic patients and its relevance to hospital mortality. All adult septic patients were divided into elderly and non-elderly group. According to China Country Assessment Report on Ageing and Health from World Health Organization in 2015, elderly was de ned as aged 60 years or over [21]. According to previous study, early immune status was de ned as the immune status within 48 hours after onset of sepsis [19]. Immune status was measured by the expression of mHLA-DR because of its proven value in septic patients [22]. Therefore, only those patients with mHLA-DR measured within 48 hours after onset of sepsis was enrolled in this study (273/361). To assess different immune status, we divided early immune status into two categories: immunosuppression (≤ 30%) and non-immunosuppression (mHLA-DR > 30%) [10,23]. Changes in immune status were assessed by the value change in mHLA-DR on day 3 compared with the rst measurement. According to our previous study, a change of mHLA-DR value of 4.8% on day 3 compared to initial measurement allowed discrimination between survivors and non-survivors [8]. Thus, the value change over 4.8% was de ned as immune status improvement, and equal or less than 4.8% was de ned as immune status non-improvement.
In addition, the Acute Physiology and Chronic Health Evaluation II (APACHE II) score with or without the age component and the Sequential Organ Failure Assessment (SOFA) score were recorded and computed during the rst 24 hours after onset of sepsis. When using dichotomous variable for logistic analysis, high SOFA score was de ned as the SOFA score over than 8 according to previous study [24]. Other clinical or laboratory parameters were also recorded at the same time. For prognosis, we assessed hospital mortality, 28-day mortality, ICU mortality, length of ICU stay and mechanical ventilation (MV) support days.
Because we did a second selection of patients from the ETASS study, considering of the possible selection bias, the baseline clinical characteristics between included (273/361) and excluded (88/361) patients was compared.

Statistic methods
Continuous variables with normal distribution were summarized as mean (standard deviation, SD) and compared by t-test; while non-normal distributed variables were described as median (interquartile range, IQR) and compared by the Wilcoxon rank sum test. Categorical data were presented as frequencies and percentages, and compared with Chi-squared tests. Logistic regression analysis was used to evaluate the association between early immunosuppression and mortality, strati ed by age groups. In model 1, the crude odd ratios (ORs) and 95% con dence intervals (CIs) were calculated by entering only the variable for early immunosuppression. In model 2, data were adjusted for sex, age, pre-existing condition, immunotherapy, and SOFA score. Then, we further adjusted for the dichotomous variable of immune status improvement in model 3. We also did a sensitivity analysis in which only patients whose levels of mHLA-DR were obtained within 24 hours after onset of sepsis were included to nd out if the de nition of early immune status (24 hours vs. 48 hours after onset of sepsis) affects the outcomes. A p-value < 0.05 (two tailed) was considered statistically signi cant. All analyses were conducted using IBM SPSS software version 24.0 (IBM Corp., Armonk, NY, USA).

Baseline clinical characteristics of adult patients with sepsis
Of the 361 patients, 273 patients were enrolled in this study, including 181 elderly and 92 non-elderly ( Fig. 1). There was no signi cant difference in most clinical variables between the included and excluded patients except for the prevalence of immunotherapy (54/88 vs. 127/273, p = 0.016). The excluded patients were more likely to receive immunotherapy than those included (Table S1). Baseline clinical characteristic data of included patients are shown in Table 1. The mean APACHE II score without the age component were 18.3 (7.4) in the elderly and 17.5 (6.9) in non-elderly. The mean SOFA score was 8.0 (3.9) in the elderly and 7.5 (3.6) in non-elderly. There was no difference in the severity of sepsis between the elderly and non-elderly group, but the elderly group had a higher percentage of pre-existing conditions than non-elderly (85.1% vs. 70.7%, p = 0.006).

Prognosis of adult patients with sepsis
In our study, 71 (39.2%) elderly and 25 (27.2%) non-elderly died in hospital ( Table 2). The elderly patients received more days of mechanical ventilation support than the non-elderly patients (6.2 vs. 4.8; p = 0.009), but there was no difference in the length of ICU stay days (10.3 vs. 9.0; p = 0.103) between elderly and non-elderly patients ( Table 2).

Immune status in adult patients with sepsis
The percentage of early immunosuppression (mHLA-DR ≤ 30%) in the elderly was twice of that in the non-elderly (32% vs. 16%, p = 0.006) patients ( Fig. 2A). The immunosuppressed elderly patients had higher hospital mortality than the non-immunosuppressed ones (53.4% vs. 32.5%, p = 0.009), but there was no signi cant difference in hospital mortality between the immunosuppressed non-elderly and the non-immunosuppressed non-elderly (33.5% vs. 26.0%, p = 0.541) patients (Fig. 3A). To detect the change of immune status, 239 septic patients (80 non-elderly and 159 elderly) whose mHLA-DR was measured on day 3 were included. In these patients, about half of the elderly (82/159, 52%) and the non-elderly (38/80, 47%) patients had immune status improvement on day 3 (Fig. 2B). We also found that patients with immune status improvement on day 3 had lower hospital mortality than those without immune status improvement in both the elderly (21/82 vs. 35/77) and the non-elderly (4/38 vs. 16/42) group (Fig. 3B).
Early immunosuppression was associated with increased mortality in elderly patients  (Table 3). In addition, we also found that immune status improvement on day 3 was associated with reduced hospital mortality in both elderly (ORs: 0.335; 95% CI 0.159-0.706; p = 0.004) and non-elderly (ORs: 0.131; 95% CI 0.029-0.584; p = 0.008) patients. Values are odds ratios (95% con dence intervals) unless stated otherwise a, Adjusted for sex, age (per 10 years), pre-existing condition, immunotherapy, and SOFA score (high vs. low).
b, Adjusted for covariates in model 2 and the dichotomous variable of immune status improvement on day 3 (yes vs. no).
The de nition of early immune status is a controversial issue, and early immune status was de ned as immune status within 48 hours after onset of sepsis in our study. To evaluate whether the de nition of early immune status was driving our results, we performed a sensitivity analyses in which early immune status was measured within 24 hours after onset of sepsis. Consistently, early immunosuppression was independently associated with increased hospital mortality in elderly (ORs: 5.507; 95% CIs: 1.497 ~ 20.2584; p = 0.010), but not in non-elderly (ORs: 1.536; 95% CI 0.236-10.023; p = 0.654) patients ( Figure  S1).

Discussion
The present study is, to the best of our knowledge, the rst one to evaluate early immune status in elderly septic patients. Our study has a few important ndings. First, our data indicated the elderly had greater risk of developing immunosuppression within 48 hours after onset of sepsis, and the rate was twice of that of the non-elderly. Secondly, early immunosuppression in elderly was associated with poor prognosis, but not in non-elderly patients. Thirdly, immune status improvement was associated with reduced mortality in both elderly and non-elderly.
Immunosuppression was associated with increased mortality and secondary infection, prolonged length of ICU stay and aggravated organ dysfunction in adult and children with sepsis [18,19,25,26]. However, the timing of immunosuppression remains controversial, so it is di cult to determine when to administer immune monitoring and immunotherapy. A recent study demonstrated that septic paediatric patients were immunosuppressed within the rst 48 hours after sepsis, and such early immunosuppression was signi cantly associated with prolonged organ dysfunction time [19]. In our study, about one-third of the elderly (58/181) had early immunosuppression within 48 hours, and more than half of the immunosuppressed patients (31/58) died in hospital. All these results displayed that elderly were more susceptible to immune dysfunction early after onset of sepsis. Therefore, the immune status of elderly patients needs to be monitored from the early stage of sepsis.
Consistent with previous studies, we found elderly patients with sepsis have higher mortality than nonelderly [2,27]. However, it appears that it is not the age per se but rather the associated factors, such as severity of illness or immune status, contribute to the increased mortality rate [28]. In our study, the severity of sepsis (SOFA score and APACHE-score without an age component) was similar in the elderly and non-elderly, but the percentage of early immunosuppression in the elderly was twice of that of the non-elderly patients. We also found that more than half of the immunosuppressed elderly (31/58, 53%) died in hospital, but only one third of the non-elderly (5/15, 33%) died during hospitalization. Furthermore, immunosuppression was associated with increased hospital mortality in the elderly, but not in the nonelderly patients. That is to say, immunosuppression may be responsible for the increased mortality in elderly patients with sepsis.
Numerous studies revealed that mHLADR can be used to predict prognosis and to select immunosuppressed patients who needed immunostimulant [8,9,29]. Monneret and his colleagues previously found that mHLA-DR decreased in septic patients on days 1-2 and days 3-4 after onset of sepsis, but only low mHLA-DR (≤ 30%) on days 3-4 was independently associated with increased 28-day mortality in patients with sepsis [10]. However, Perry and his colleagues reported a different result that mHLADR on days 1-3 cannot help to predict outcome in sepsis [30]. In Perry' research, the median age of septic patients was about 56 with a range from 20 to 84, while in Monneret' study, the median age was 64 (IQR: 48 ~ 75), the age differences in septic patients may be the cause of the opposite outcomes in the two studies. In current study, our results indicated that early low mHLA-DR expression was an independent risk factor for poor outcome in elderly, but not in non-elderly septic patients. Our previous study demonstrated that dynamic change of mHLA-DR was a reliable predictor for mortality in septic patients [8]. Then, we combined early immune status with changes of immune status to evaluate hospital mortality. In this study, early immunosuppression was associated with increased hospital mortality in elderly, and immune status improvement on day 3 was associated with reduced hospital mortality in both elderly and non-elderly patients. Therefore, monitoring of early immune status should be carried out in elderly patients, and it may be bene cial to monitor the dynamic changes of immune status in both elderly and non-elderly patients.
Several limitations should be noted in our study. Firstly, our data came from a clinical study of immunotherapy for patients with sepsis. Considering the interference of immunotherapy in our study, we took immunotherapy as a xed covariate in multivariate logistic regression analysis for mortality and found immunotherapy did not affect the prognosis of elderly patients. Secondly, only 15 non-elderly septic patients with early immunosuppression were included in our study, so a larger study is needed to further verify the results that early immunosuppression was not associated with poor prognosis in nonelderly patients.

Conclusions
In adult patients with sepsis, the elderly were more susceptible to early immunosuppression after onset of sepsis. Early immunosuppression was associated with poor prognosis in elderly patients.