Introduction
Sepsis is defined as an acute, life threatening organ dysfunction caused by a dysregulated immune response to a microbial assault [
1]. It is one of the leading causes of mortality in industrialized nations, affecting millions of individuals per year [
2]. The etiologies of the dysregulated immune response might be diverse, e.g., bacterial assault as well as fungal or viral infection [
3,
4]. In addition many confounding factors can influence the immune response such as age, trauma or chronic diseases [
5] making the immune response a non-linear, complex system [
6]. Nonetheless, it is well recognized that pro- and anti-inflammatory cytokines, such as IL-6 or IL-10, are good predictors of clinical outcome, suggesting the influence of the immune response on the survival of the patient [
7‐
9]. So far, clinical trials that attempted to modulate the dysregulated immune response have largely failed [
10,
11]. Recently, the pre-emptive treatment of the human cytomegalovirus (HCMV) has been discussed in the context of sepsis [
12,
13].
HCMV is a member of the herpes virus family and is endemic in humans with an increasing rate of seropositivity (IgG +) according to age [
14] and hygienic standards [
14]. In Europe, the seropositivity rate of HCMV is around 60–70% which is reached in the seventh decade of life [
15], rendering HCMV infections relevant in the major parts of the population. While infections in adult immunocompetent individuals frequently progress without overt disease or only mild symptoms, one of the major characteristics of HCMV is its ability to establish latency in the host. Latency is defined by the presence of the replication competent viral DNA in bone marrow hematopoietic progenitor cells [
16] and tissue endothelial cells [
17] as potential reservoir below the level of immune detection, while most viral translational products/proteins are absent [
18]. This state does not cause an active disease. Hence, IgG seropositivity for HCMV is the Gold Standard bio-marker for a latent virus infection as it indicates a prior infection with HCMV and therefore latency of the virus. This latent infection persists for the lifetime of the host [
19] and can be followed by a potentially dangerous HCMV reactivation during conditions of impaired immunity [
20,
21] which is well known and investigated in numerous trials [
20‐
23]. While the dogma in intensive care medicine is that HCMV reactivation during critical illness is potentially life threatening, little attention has been paid to the latency (i.e., the seropositivity), despite the fact that latency significantly alters the immune response [
24] and the composition of the immune system [
25].
Therefore, the impact of HCMV latency on the immune system has been suggested and discussed [
18] but not sufficiently investigated in critically ill patients. Thus, further research is needed to evaluate the effects of the latent HCMV infection on the hosts’ immunity.
To achieve a state of latency, HCMV counteracts its host immune response in multiple ways. Disturbing the cellular presentation of antigens to T- and NK-cells [
26] as well as the modulation of the cellular signaling toward apoptosis [
27] preserves the virus’ latent survival but also alters the steady state of the host. An extended cellular surface maintenance of TLR-4 and TLR-5 on CMV-infected cells as well as an amplified intracellular proinflammatory cascade with the phosphorylation of IκB-α and NF-κB and significantly promoted TNF-α, IL-6, and IL-8 expression was described in a macrophage model [
28]. These findings suggest a HCMV-induced impairment of the immune system shifting the balance toward a more pronounced inflammation. This might significantly affect the patient when suffering from an infection or even sepsis later in life.
In light of this, we asked (1) if we can find an altered inflammatory response during sepsis associated with the HCMV serostatus and (2) if the HCMV serostatus impacts the 30-day survival of sepsis patients.
Material and methods
Study design and cohort
The SepsisDataNet.NRW study [
29] (German Clinical Trial Registry No. DRKS00018871;
http://www.sepsisdatanet.nrw) prospectively enrolled patients fulfilling the SEPSIS-3 criteria in a multicentric approach from the ICUs of seven different hospitals (university hospitals or tertiary care hospitals) in the German state of North Rhine-Westphalia. This study was approved by the Ethics Committee of the Medical Faculty of Ruhr-University Bochum (Registration No. 18-6606—BR) or the responsible ethics committee of each respective study center. The patients were recruited after obtaining written informed consent from the 1st of March 2018 to the 31th of May 2022. This study included adult patients with a sepsis diagnosis within the previous 36 h according to the current SEPSIS-3 definition (suspected/proven infection and an increase in the Sequential Organ Failure Assessment (SOFA) score by two points or more). The cohort comprised of mixed surgical and medical patients admitted to the ICU. The exclusion criteria were as follows: (1) age of < 18 years old at the time of ICU admission, (2) withdrawal or withhold of consent, and (3) withdrawal of treatment. Patients with unknown 30-day survival status were excluded from further analysis. It is noteworthy that immunosuppression was not an exclusion criterion as we did not focus on re-activation of HCMV.
Clinical data and patients’ characteristics
Electronic medical data including vitals, laboratory values, point of care diagnostics, demographics, and the length of ICU stay were captured in a comprehensive database (CentraXX software, Kairos GmbH, Bochum, Germany) following pseudonymization according to the obligations of the ethics committee. Missing data was augmented by the individual investigation of patient records at each corresponding clinic by an experienced physician and, where appropriate, completed by including the data from ± 12 h of the onset of sepsis. SAPS-2 DRG as well as SOFA score was manually calculated by an experienced physician at each recruitment site. All patients were treated according to the current sepsis guidelines.
HCMV serostatus (IgG) and reactivation (IgM and DNA)
HCMV IgG and IgM
The patient sera were evaluated for the presence of HCMV IgG at Day 1 and IgM at Days 1 and 8 after enrollment using the SERION ELISA Classic Cytomegalovirus IgG/IgM Kit (Institut Virion\Serion GmbH, Würzburg, Germany). One hundred µL of diluted samples (1:40) and respective controls were applied into micro test wells and incubated for 60 min at 37°C. Afterwards, the samples were washed three times and incubated with 100 µL substrate solution for at least 20 min at 37°C. The reactions were stopped using 100 µL of stopping solution, and the optical densities (OD) were determined using a microplate reader (Sunrise Microplate Reader\Tecan, Maennedorf, Switzerland). The OD values were normalized to a standard and the units were calculated. The IgM samples were classified as positive when 15 or more units were detected, while the IgG assays were classified as positive when 35 or more units were detected.
HCMV DNA
Whole blood DNA was used to evaluate the concentration of HCMV DNA at Days 1, 4, and 8. Table
1 shows the primer/probe combination. DNA was subjected to a qPCR analysis using TaqMan Universal Mastermix (Applied Biosystems) in accordance with the manufacturer’s recommendations. The threshold cycles (Ct) were determined and the samples were identified as positive if Ct was < 40. The samples between Ct 35 and 40 were tested a second time to filter out false positives.
Table 1
Primer and probe combination for the evaluation of HCMV re-activation
Forward primer | 5 ‘-ATAGGAGGCGCCACGTATTC-3 ‘ | Biomers |
Reverse primer | 5 ‘-TACCCCCTATCGCGTGTGTTC-3 ‘ | Biomers |
Probe | 5’-FAM-CGTTTCGTCGTAGCTACGCTTACAT-TAMRA-3’ | Biomers |
Cytokine concentrations
As part of the SepsisDataNet.NRW study, the biomaterials (serum) were collected at Days 1, 4, and 8 after recruitment. The serum samples collected at Day 1 were used to quantify the concentration of 13 cytokines at the time of recruitment. The LegendPlex Human Inflammation Panel 1 (Biolegend, San Diego) was used in accordance with the manufacturer’s instructions. Briefly, the serum samples were incubated with LegendPlex beads for antigen capture, washed, and incubated with detection antibodies. After additional washing, the beads were measured using a flow cytometer (Canto II, BD Biosciences, CA), and cytokine concentration was quantified using a standard curve. If the recorded concentration of a cytokine was below the lower limit of detection (LOD), the value was treated as 0 ng/mL. If a value was recorded as higher than the upper LOD, it was treated as the upper LOD.
Antibodies for flow cytometry
All antibodies are from BioLegend (San Diego) unless otherwise noted: CD16-APC-fire750, clone: 3G8; CD56-AF647, clone: NCAM; CD45RA-BV650, clone: HI100; CCR7-Pe/Dazzle 594, clone: G043H7; CD8-PeCy7, clone: SK1; CD45-A488, clone: SD1; CD4-AF700, clone: OKT4; CD19-BV605, clone: HIB19; CD3-BV785, clone: OKT3; HLADR-PE, clone: L243 (BD Biosciences, CA); CD14-PerCP-Cy5.5, clone: MφP9 (BD Biosciences, CA); Zombie Aqua™ Fixable Viability Kit.
Immunophenotyping
25 µl EDTA-treated whole blood was stained with 25 µl master mix, containing the optimal concentrations of each antibody, for 10 min at room temperature in the dark. Erythrocytes were lyzed using RBC Lysis Buffer (BioLegend, San Diego) for 10 min at room temperature in the dark and samples were immediately acquired on a CytoFlex flow cytometer (Beckman Coulter, Brea). Quality control was performed daily using the recommended CytoFlex Daily QC Fluorospheres (Beckman Coulter, Brea). No modification to the compensation matrix was required throughout the study.
Plasma proteomics
The plasma samples from 171 patients who were enrolled in the participating centers in Bochum and Bonn were digested according to the SP3 protocol with slight modifications. Briefly, 100 µg protein was purified using paramagnetic beads (Cytiva Sera-Mag Carboxyl-Magnet-Beads, GE Healthcare, Chicago, IL) and digested overnight using trypsin (SERVA Electrophoresis, Heidelberg, Germany). Subsequently, 300 ng tryptic peptides per sample were analyzed using an Ultimate 3000 RSLCnano HPLC coupled online to either an Orbitrap QExactive, Orbitrap QExactive HF, or Orbitrap Fusion Lumos mass spectrometer (all Thermo Scientific, Bremen, Germany). In total, 306 samples were analyzed and distributed over five batches and separated by either a 96-min (Batch 1) or 38-min (Batches 2–5) LC gradient. The mass spectrometers were operated in data-independent acquisition mode. Spectral libraries were generated with FragPipe (v.17.1) and protein quantification was conducted using DIA-NN (v.1.8) [
30]. The Uniprot/SwissProt database restricted to homo-sapiens (release 01_2022; 20,386 entries) was used for protein identification. The resulting protein intensities were first normalized using the LOESS method [
31]. The subsequent cross-batch normalization was based on linear regression models. A detailed description of the applied methods can be found in the Additional file.
Statistics
Statistical analyses were performed using the IBM® SPSS Statistics software version 28. The statistical differences of the categorical variables were determined using Fisher’s Exact Test, while the continuous variables were tested using Student’s t-test and the Mann–Whitney U test. Normally distributed variables were compared using Student’s t-test, while non-normally distributed variables were compared using the Mann–Whitney U test. The impact of a continuous variable on the 30-day survival was analyzed by Receiver Operator Characteristics, Log Rank Tests (visualized by Kaplan–Meier curves), and multivariate Cox regression analysis. First, the best predictive cutoff for each variable was determined by ROC and the calculation of Youden’s index. Accordingly, we defined a cutoff of 440 pg/ml for IL-6, 13.5 pg/mL for IL-10, and 3.43ng/ml for PCT. If not otherwise stated, p-values below 5% were considered significant. Proteomics data were processed using R (v.4.2.1; r-project.org). Statistical differences between the groups were assessed based on normalized protein intensities using Student’s t-test. The resulting p-values were FDR-corrected according to Benjamini-Hochberg. Relative changes were calculated as ratios of means based on delogarithmized intensities.
Discussion
With this study we provide a novel and alternative view on the ongoing debate, whether or not HCMV re-activation is detrimental for the outcome of sepsis. We propose that HCMV latency, rather than re-activation could be the real culprit. HCMV latency refers to the period when the virus is present in a silent, inactive form within the host. During latency, infectious virus particles are not produced by the virus. Instead, the viral genome is maintained within the host cell's nucleus, and only a limited set of viral genes is expressed at low levels [
32]. This limited gene expression helps the virus evade the immune system and persist in the host. Because of absence of virions in the latent state the best way to detect HCMV latency is a positive HCMV IgG serostatus. Our data demonstrate a strong association between this HCMV serostatus and sepsis survival. HCMV seropositivity was highly predictive for worsened outcomes in sepsis. This suggests that HCMV serostatus should be considered when evaluating the predictive value of common biomarkers and cytokines in sepsis. As latent HCMV infections, which are the reason for HCMV seropositivity, affect the majority of mankind [
14], our findings are relevant to most adult sepsis patients. In this context, the impact of HCMV latency on the complex interplay of human immunity based on non-replicating viruses has been discussed [
18,
33,
34]. Brodin et al. even go so far as to call HCMV latency the strongest non-heritable immune modulator in humans [
35]. In vitro inflammation in HCMV-infected human macrophages resulted in an increased expression of toll-like receptors TLR-4 and TLR-5 at the cell surface, potentiating the susceptibility and downstream activation of the cellular response to LPS stimulation [
28]. In vivo, after vaccination against influenza, the HCMV-seropositive individuals presented an enhanced antibody response and elevated circulating IFN-γ levels compared with the HCMV-seronegative individuals [
36]. In contrast to this HCMV seropositive patients exhibit a lower number of Influenca specific CD8 T-cells than seronegative patients [
37]. Overall, these findings showed a modified inflammatory response in HCMV-seropositive individuals. Usually these changes are attributed to re-activation of HCMV, a dogma that exists in intensive care medicine for some time. In line with this, while HCMV reactivation has been intensely studied in sepsis patients for more than 30 years [
13,
38,
39], only few studies investigated the HCMV serostatus and its relevance in infectious diseases. One retrospective study [
40] that focused on critically ill patients in general could not demonstrate a relevant impact on the patients’ survival. However, as that patient cohort only consisted of around 60% sepsis patients (SEPSIS-1 definition), only limited conclusions could be drawn regarding sepsis. To our knowledge, the impact of the HCMV serostatus on patients suffering from sepsis (according to SEPSIS-3) has not yet been investigated.
Therefore, we performed a post hoc analysis of the multicentric SepsisDataNet.NRW study that prospectively enrolled patients suffering from sepsis. In this study cohort, we found the expected proportion of HCMV-positive patients (63%) and could for the first time show that these patients had a worse prognosis than the HCMV seronegative patients. Of note, seropositive patients were slightly older and sicker than seronegative patients, which could theoretically explain the observed effects on survival as confounders. From a different perspective, one could speculate that the higher severity of the disease might also be a consequence of HCMV latency.
As reactivation might also be an important confounder, we analyzed the HCMV reactivation both by qPCR and by measuring IgM and could not detect a relevant number of reactivating patients even until Day 8. Due to the post hoc nature of our study, we cannot make a definite statement on the impact of later HCMV reactivation. However, a recent study evaluating HCMV reactivation in sepsis patients determined the median time to reactivation at seven days with a total reactivation of 18.3% in their cohort [
41]. On the other hand, other studies found a meaningful reactivation to occur by Day 21 [
20]. This high heterogeneity makes it difficult to accurately rule out HCMV reactivation as a confounder. However, as the effect of HCMV serostatus on patient survival is already significant at Day 8 (
p = 0.014), we postulated nonetheless that CMV reactivation, even if it is expected later in the course of the disease, would not change our findings. Thus, we concluded that HCMV reactivation is not an important confounder in our cohort in relation to survival. Based on our findings, we conclude that the impact of HCMV latency on the immune reaction is at least partially responsible for the observed effects of the serostatus on survival. Obviously, there are a range of possible confounders which, due to the post-hoc nature of this analysis we could not correct for. In the SepsisDataNet.NRW study the socio-economic status, which is a major contributor to HCMV seropositivity was not assessed, neither were other possible confounders such as ethnicity. However, to our knowledge these confounders do not influence survival in sepsis, so they are unlikely to explain our findings. Another possible confounder for HCMV serostatus as well as an altered immune response is age. However, in our Cox Regression analysis the effect of the cytokines was independent of age and disease severity (SOFA score). Additionally, it is important to acknowledge that the representation of HCMV seronegative patients in any representative cohort, including ours, is naturally limited due to the statistical distribution. Consequently, the power of the HCMV seronegative cohort is lower. However, the effect size observed in the seropositive cohort is substantial enough to be detectable even in the seronegative cohort. Therefore, we are confident that the impact of the immune reaction is largely limited to the seropositive cohort.
In order to investigate the implications of HCMV-induced changes to the human immunity, we compared the cytokine levels of seropositive- and -negative patients during the study inclusion period. Only the IL-8 serum concentration was found to be significantly different at Day 1. This is consistent with previous literature suggesting the elevated IL-8 levels in HCMV-seropositive patients [
28]. Interestingly, as seropositive patients show a higher number of NKT-cells, one could expect an increase of IFN-γ [
42]. However, as cytokines are produced by many different cells it is not un-surprising to not see correlations with the HCMV serostatus. While the serum levels of all other measured cytokines as well as a mass spectrometry-based analysis of the plasma proteome showed no significant difference with regard to the HCMV serostatus, we indicated that the expected predictive value of IL-6 [
7] and IL-10 [
43] on the patients’ survival was associated with the HCMV serostatus. Serum PCT showed an analogous association. For all tested cytokine-concentrations, the best discriminating cutoff-value was determined calculating the Youden-Index. Beyond this, our cutoff values for IL-6 and IL-10 resemble reported values [
44,
45].
This is surprising considering that the predictive value for IL-6, IL-10, and other immunological marker proteins is described in many studies in prestigious journals [
7‐
9]. Our data do not contradict these earlier findings but rather suggest that the predictive power of those cytokines is based solely on HCMV-positive individuals, which frequently accounts for the major parts of the study population. This also appears notable as the interpretation of these often-used markers might have to be seen in a different perspective from now on. In line with this, we found multiple differentially abundant plasma proteins, such as CRP, vWF [
46], or lysozyme [
47], in HCMV-seropositive patients to be associated with the survival status, which is not observed in HCMV-seronegative individuals. These data show that protein networks related to reactive oxygen species, complement activation and apoptotic cell clearance are stronger activated in HCMV seropositive patients succumbing to the disease. One possible implication from these data is that immunomodulation (e.g., using hemadsorption [
48] or anti IL-6 antibodies [
49]) might only be proven beneficial for HCMV-seropositive patients.
From a clinical point of view, we also considered the common biomarker serum procalcitonin. Interestingly, we found a similar effect to the one seen in cytokines, restricting the predictive power of PCT [
50] to only CMV-seropositive patients. In light of this, we propose to re-evaluate common clinical biomarkers in sepsis with regard to HCMV serostatus. Although our data are retrospective in nature, it is tempting to speculate on possible mechanisms, including a HCMV latency-related change in the immune reaction. However, further research will be needed to determine the underlying relationships.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.