Introduction
Cytomegalovirus (CMV) reactivation is observed in 14–41% of intensive care unit (ICU) patients without known prior immune deficiency [
1‐
3] and is associated with increased morbidity and mortality [
4‐
6]. In a previous study, we estimated that the population-attributable fraction of ICU mortality due to CMV reactivation was 23% in patients with acute respiratory distress syndrome (ARDS) [
7]. In a subsequent study among patients with septic shock, we found an effect of CMV reactivation on ICU mortality only in patients with concurrent Epstein–Barr virus reactivation [
8]. Although multiple studies point toward a causal relationship, definitive proof that CMV reactivation worsens clinical outcome is lacking, as most data are also compatible with a scenario in which CMV reactivation is merely a marker of immune suppression in this patient group.
Based on previous studies in ICU patients, there is a clear pathophysiological link between inflammation and immune suppression on the one hand and the subsequent risk of CMV reactivation on the other [
9‐
13]. Markers reflecting impaired functioning of natural killer cells and cytotoxic T cells were predictive of CMV reactivation [
10,
11]. Furthermore, bacterial sepsis and corticosteroids have been identified as clinical risk factors for CMV reactivation [
9,
12,
13]. However, less is known about the reverse association and thus the effects of CMV reactivation on the immune system. Direct cytotoxic effects of CMV on organs have been observed primarily in immunocompromised hosts [
14] but also in previously immunocompetent patients in the ICU [
15]. Moreover, indirect immune-modulating effects are assumed to play a role in the pathogenicity of CMV [
13,
16‐
18].
In vitro analysis revealed multiple mechanisms encoded within the genome of CMV that may contribute to a non-specific inhibition of both cellular and humoral immunity [
19]. Observational clinical studies yielded conflicting results comparing levels of multiple inflammatory markers in patients with and without CMV reactivation [
1,
11,
20]. However, these studies analyzed biomarker responses only immediately upon ICU admission and thus could not assess potential immunological effects due to the onset of CMV reactivation. Nevertheless, cytokine levels were used as a primary (surrogate) endpoint in a recent placebo-controlled randomized control trial in which prophylactic antiviral treatment with ganciclovir failed to reduce interleukin-6 (IL-6) levels [
21]. Hence, definite proof of immune-modulating effects induced by CMV remains to be demonstrated. Naturally, such an effect can be demonstrated only after onset of CMV reactivation. Therefore, this longitudinal study aimed to investigate whether the temporal course of seven host response biomarkers, including both pro- and anti-inflammatory cytokines, in previously immunocompetent ICU patients with sepsis differs between patients with and without CMV reactivation.
Discussion
We performed an explorative study to compare time trends of host response biomarkers in patients with reactivation that were matched to non-reactivating control patients who were either seropositive or seronegative for CMV. Although we initially observed differential trends of IL-1RA and IP-10 in the crude analysis, these differences did not remain in the linear mixed model analysis with adjustment for repeated measurement, loss to follow-up, and confounding. Thus, no overall and independent effect of CMV reactivation on the temporal trends of host response biomarkers following onset of viremia in patients with sepsis could be demonstrated.
The hypothesis of an immune-modulating effect of CMV is based on the observation of increased mortality and morbidity in patients with viremia without organ manifestation of CMV disease [
13,
19]. Proposed mechanisms of such indirect pathogenicity are autoantibody production, enhanced inflammation, vascular damage, and CMV-induced immunosuppression [
17]. Based on this hypothesis and an observed association between plasma markers and mortality in patients with ARDS [
26], IL-6 was used as a surrogate endpoint in a recent randomized controlled trial that evaluated the safety of preventive antiviral treatment in ICU patients [
16]. Our finding that CMV reactivation is not associated with modified IL-6 dynamics questions the suitability of IL-6 as an endpoint in clinical trials evaluating preventive therapy for CMV reactivation in ICU patients. Furthermore, time trends of other immunological biomarkers were not robustly affected by CMV reactivation.
Our study has several strengths. First, to our knowledge, this is the first study with serial measurements of the immune response following (instead of prior to) CMV reactivation. Second, our study design included two matched control groups. Because of the used matched cohort design, we could include only 45 out of 63 patients with CMV reactivation but this loss was compensated by the ability to include controls that were more comparable to those patients. Sepsis patients in the ICU are known to be very heterogeneous [
27,
28]; thus, the matching reduced in theory both confounding and unwanted variation by extraneous factors. Third, by using mixed model analyses, we accounted for correlation of measurements performed within one patient by the use of random effects, which increased the statistical power to identify differences between patient groups. Moreover, this type of analysis takes into account the considerable loss to follow-up of patients and allowed us to estimate an average trend over time based on available data.
Our study also has some limitations. First, this was an explorative study evaluating multiple host response biomarkers. We chose a lower
P value threshold of significance in order to decrease the risk of spurious findings due to multiple testing, but false-negative findings remain an accessory risk to keep in mind also when considering our study sample size. Unfortunately, a formal sample size calculation for this kind of statistical analysis was not possible. Nevertheless, we postulate that possible immunomodulating effects of CMV reactivation seem at most to be rather limited in these patients because no large differences in biomarker levels between matched groups were observed. Second, we analyzed host response biomarkers as standalone markers, which is probably a simplification of the complex immune response. However, large sample sizes are required to perform more advanced network analyses, and the integration of time series in such analyses, to our knowledge, has not been conducted before. We also measured only the plasma concentrations. Since CMV pneumonitis could be an important mediator of the pathological effect of CMV reactivation in critically ill patients, bronchoalveolar lavage samples may be additionally informative but were not available [
16,
29]. Finally, we did not evaluate all potentially relevant biomarkers for CMV reactivation; thus, future studies are needed before an immunomodulating effect of CMV can be ruled out with certainty as an important pathological mechanism in previously immunocompetent ICU patients.
Acknowledgments
We thank the Department of Medical Microbiology and the Multiplex core facility of the Laboratory for Translational Immunology for their logistical support and performance of measurements, the participating ICUs and research nurses of the two medical centers for their help in data acquisition, and all members of the Molecular Diagnosis and Risk Stratification of Sepsis (MARS) consortium for their contribution to this research.
Members of the MARS consortium:
Amsterdam University Medical Centers, University of Amsterdam (NL): Friso M. de Beer, MD; Lieuwe D. J. Bos, MD, PhD; Gerie J. Glas, MD; Arie J. Hoogendijk, PhD; Roosmarijn T. M. van Hooijdonk, MD, PhD; Janneke Horn MD, PhD; Mischa A. Huson, MD, PhD; Nicole P. Juffermans, MD, PhD; Tom van der Poll, MD, PhD; Laura R. A. Schouten, MD; Brendon Scicluna, PhD; Marcus J. Schultz, MD, PhD; Marleen Straat, MD, PhD; Lonneke A. van Vught, MD, PhD; Luuk Wieske, MD, PhD; Maryse A. Wiewel, MD, PhD; Esther Witteveen, MD, PhD.
University Medical Center Utrecht, Utrecht University (NL): Marc J.M. Bonten, MD, PhD; Olaf L. Cremer, MD, PhD; Jos F. Frencken, MD, PhD; K. van de Groep, MD; Peter M.C. Klein Klouwenberg, MD, PharmD, PhD; Maria E. Koster-Brouwer, MSc; David S.Y. Ong, MD, PharmD, PhD; Meri R.J. Varkila MD; Diana M. Verboom, MD.