Assessment of sepsis-induced immunosuppression at ICU discharge and 6 months after ICU discharge

Despite progress in the therapeutic management of patients with sepsis and septic shock, hospital mortality of these patients has only slightly decreased in recent decades and remains around 20 and 40%, respectively [1]. While mortality in the first week has decreased, it has increased during subsequent weeks [2]. It is hypothesized that sepsis-induced immunosuppression occurring during this second phase of sepsis explains the “late” mortality, most likely due to increased risk of secondary infections [3].

Sepsis-induced immunosuppression is characterized by alterations affecting both innate and adaptive immunity. Among the biological parameters assessing the innate immune system, the level of the human leukocyte antigen DR expression on the surface of monocytes (mHLA-DR) is one of the best studied, and is recognized as an effective indicator of immunosuppression [4, 5]. Furthermore, a reduction in mHLA-DR is the only immune parameter shown to be an independent factor of nosocomial infection occurrence [6] and of mortality [7]. With regard to adaptive immunity, lymphopenia on the fourth day following the diagnosis of sepsis predicts mortality and may also serve as a biomarker for sepsis-induced immunosuppression [8]. Regulatory T cells (Treg) are less-well studied but present nevertheless well-defined and broad immunosuppressive properties [9‐11]. Thus, in sepsis, the increase in the proportion of Treg cells among the total number of T CD4+ lymphocytes is correlated with poor prognosis [12].

Whereas the occurrence of immunosuppression during ICU stay is well documented, to date, there is almost no data on ICU discharge and on long-term evolution of sepsis-induced immunosuppression in survivors of septic shock. Some data support the hypothesis that sepsis could induce persistent long-term immune dysfunctions even long after the hospital discharge. Indeed, in the longer term, even after total clinical recovery in sepsis survivors, increase in mortality continues. It has been reported mortality of 43% after 1 year [13], 45% after 2 years [14], and 74% after 5 years [15] beyond hospital discharge. Furthermore, an increase of recurrent infections in the year following ICU discharge has also been reported in such patients [16, 17]. Taken together, persistent sepsis-induced immunosuppression at ICU discharge and in long term could be the underlying reason for this delayed mortality in sepsis survivors.

The aim of this study was to assess the persistence of sepsis-induced immunosuppression at the cellular level at ICU discharge and 6 months thereafter by measuring mHLA-DR, CD4+ T cells, and Treg expression in patients admitted for septic shock.

This study found persistent cellular immune alterations in septic shock survivors at ICU discharge but no detectable long-term immune alterations 6 months after ICU discharge in some of them, as evidenced from mHLA-DR expression, CD4+ T cells, and relative proportion of Treg. Although unexpected, these data suggest that sepsis-induced immunosuppression at the cellular level based on these three markers is reversible after a period of time.

Among the biological parameters assessing the innate immune system, mHLA-DR seems to be an effective indicator of immune alterations [4, 5]. Its decreased expression represents one of the only immune marker shown to be an independent factor of occurrence of secondary nosocomial infections [6]. The interest of mHLA-DR expression assessment stands in the fact that this marker is a consequence of the effect of multiple mediators on the inflammatory response [21‐23]. Its routine determination could be an important element of the immunological monitoring of septic patients [24, 25] or even of all patients undergoing an inflammatory injury and at risk of developing secondary infections. Usually, a patient is considered to be immunocompetent when mHLA-DR expression is above 15,000 AB/C. Values comprised between 10,000 and 15,000 AB/C may represent moderate immunosuppression, whereas values lower than 5000 AB/C may represent very deep immune alterations [18] which existed in 80% of patients in the present study, either on day 1–2 or on day 3–4. To our knowledge, there is no data in the medical literature assessing mHLA-DR expression at ICU discharge in patients admitted to the ICU for septic shock which would allow us to determine whether sepsis-induced immunosuppression persists. Herein, three-quarters of the patients have mHLA-DR below 15,000 AB/C at ICU discharge although they are usually free of organ failure at that moment. Furthermore, 12% of patients had mHLA-DR <5000 AB/C at ICU discharge, which represent a deep immune alteration. The immunological failure, not evaluated in routine, may represent a long-term persistent failure in the same way as renal failure in some patients. Among the three patients who died within 6 months post-ICU discharge, one died from cardiac arrest but two died from an infection and had abnormal mHLA-DR at ICU discharge. Interestingly, one of these infections was a CMV ileitis, which corresponds to a latent virus reactivation typically observed in context of deep immunosuppression.

With regard to adaptive immunity, CD4+ T lymphopenia occurring during sepsis is a major component of immunosuppression. It is also associated with higher risk of nosocomial infection and mortality [8, 26, 27]. Drewry et al. [8] reported that persistent lymphopenia on day 4 after the onset of sepsis predicted 28-mortality and 1-year survival. The medical literature on regulatory T cells is less abundant. Monneret et al. [12] reported in 2003 their relative increase in septic patients, which was later confirmed by others [28]. Their immunosuppressive properties affect both the antigen presenting cells of innate immunity and lymphocytes of adaptive immunity [29, 30]. Their presence correlates with a poor prognosis; in sepsis, their relative increase is observed early but persists only in patients dying later [12]. This negative association with survival was also reported in animal models [31]. We also recently reported that the increased proportion of Treg observed in sepsis was correlated with a decrease in lymphocyte proliferation [11]. Thus, Treg participates in the development of lymphocyte anergy described in sepsis which is significantly associated with increased mortality and the occurrence of secondary infections in septic patients [32]. In the present study, almost half of patients had CD4+ T lymphopenia on day 3–4 and this lymphopenia remained present for one-third of patients at ICU discharge although it was not correlated with the ICU length stay. The lack of correlation between immunological parameters and length of stay is probably caused by the small sample size resulting in a lack of power for correlation tests. Few patients (18%) had a percentage of Treg greater than 10%, but all of them had deep loss of mHLA-DR expression (i.e., mHLA-DR <5000 AB/C).

The study population was moderately old which may have affected immunological parameters. However, to our knowledge, modifications of mHLA-DR expression and CD4+ T cell count with age remain moderate [18, 33, 34]. Values found 6 months after ICU discharge were greater than the lower band of the normal range for all but one patient for mHLA-DR expression, and all but one for CD4+ T cell count. Thus, the immunosenescence does not seem to have an impact on the parameters measured.

With regard to immunological data at 6 months post-ICU discharge, the present results at the cellular level are in accordance with those from other investigators [35]. Arens et al. recently challenged the long-term persistence of sepsis-induced immunosuppression hypothesis in a retrospective cohort study composed of 8 sepsis survivors (without having studied immunological data during ICU stay, or at ICU discharge). They reported that immune alterations were minimal 26 months (range 9–52) after the septic event. There was no significant difference in lymphocyte profile (including Treg cell counts) and in expression of mHLA-DR between sepsis survivors and healthy volunteers. The expression of pattern recognition receptors (PRR) on leukocyte surface was not different between the two groups except for TLR-5, which was expressed at a lower level in the sepsis group. Finally, they only observed alterations regarding functional testing; after ex vivo stimulation of whole blood with potent activators, immune cells exhibited a decreased secretion of INF-γ, IL-10, IL-6, and TNF-α. Moreover, 62% of sepsis survivor patients had at least another episode of sepsis within the year prior to inclusion [35].

These findings raise the importance of functional tests, which must remain the reference to assess immune status as they directly measure the ability of cells to respond to pathogens. However, such tests are not always feasible in routine due to methodological difficulties such as a long incubation period, prolonged cellular purification procedures, and absence of standardization. Efforts are needed to facilitate practice of functional tests with the aim that their use becomes as common as that of static biomarkers.

The results presented herein complete two limitations of the study reported by Arens et al. [35]. Firstly, in the previous study, septic patients were not monitored during their ICU stay which means that sepsis-induced immunosuppression was not demonstrated in these patients. Secondly, the delay between septic event and mHLA-DR measurement was very long (median: 26 months), whereas herein it is demonstrated that mHLA-DR expression is already back to normal values as early as 6 months after ICU discharge. Additional studies with immunological monitoring within weeks following ICU discharge are needed to determine the timing of recovery of sepsis-induced immunosuppression.

Nevertheless, the present study has itself some limitations. For instance, only a small number of patients were studied at 6 months because a majority could not or refused to return to our hospital 6 months after the ICU discharge. Reasons for not coming back to the hospital were various. Most of these patients were elderly [median 71 (IQR 64–76) years (data not shown)], could not drive a car, could not take public transportation, and generally, they did not leave their home. This may have generated a selection bias in so far as participating patients were potentially in better clinical condition, which may have influenced results. We were not able to follow the immune status of patients with a bad outcome (those who died within 6 months or who suffered from new infections). This does not exclude the possibility that a sub-group of patients was not analyzed because of prolonged immune suppression. In other words, the analysis concerned only patients with a favorable outcome, whose immune functions returned to normal range. Data from the whole population are lacking, and these results need to be confirmed by a larger cohort. The small sample size also precludes a sub-analysis evaluating the immune status of patients suffering from complications in the long term in comparison with those who do not. However, these 15 patients were representative of the whole population, particularly for immunological parameters at ICU discharge. Thus, it should also be noted that 7.5% of patients alive at ICU discharge died within 6 months. Moreover, two other patients were out of contact and potentially also died. The total population had a favorable outcome because we included only patients who were alive at the time of ICU discharge, while mortality rate in septic shock is reported to reach up to 40% [1]. There is also a number of missing immunological data in particular at ICU discharge. This is due to the fact that measures could not be performed over the weekend or during the week outside the opening hours of the immunology laboratory. It would also have been interesting to have complete clinical data in particular the occurrence of infections since the ICU discharge for all patients who did not come back to the hospital 6 months after their ICU discharge. This could have allowed an exploration of a possible association with their immunological parameters at ICU discharge. Furthermore, we estimated only three immunological parameters, which also constitute a limit of the study.

If one hypothesizes that persisting long-term immune alterations may participate in increased risk for additional infections in septic patients, the present results are somehow unexpected. As opposed to mHLA-DR results, CD4+ T cells results and Treg results, the main result observed in the study reported by Arens et al. [35] is the ex vivo decreased whole blood cytokine secretion in long-term survivors. This is the confirmation that a single marker is likely to not be sufficient to fully monitor patient immune status. In addition, it supports the need to include functional testing in an immune monitoring panel which could detect some persistent immune alterations; additional research is needed to improve our understanding of long-term mortality in septic patients.

This study did not find any persistent immune alterations in septic shock survivors with good clinical outcome 6 months after ICU discharge as evidenced by mHLA-DR expression, CD4+ T cells and Treg count. This does not exclude that immune alterations are present in a sub-group of patients with a bad clinical outcome. Nevertheless, CD4+ lymphopenia and low mHLA-DR expression were still present at ICU discharge for most patients.