Patients who remain in the intensive care unit for a week often carry two simultaneous warning signs: a measurable weakening of immune defenses and the reappearance of dormant herpesviruses in the blood. A new analysis of 195 critically ill patients suggests that only one of these signals reliably predicts who will fare badly. When researchers from Hospices Civils de Lyon examined both markers side by side at day 7 after ICU admission, they found that reduced expression of a molecule called monocyte HLA-DR was strongly linked to adverse outcomes, whereas the presence of viral DNA in plasma added little prognostic value in this mixed population.

The study, published in Critical Care, was designed to test whether combining the two markers could identify a particularly severe patient subgroup. The idea came from recent work by Rosiewicz and colleagues, who reported that Epstein-Barr virus reactivation occurring in sepsis patients with immunosuppression, defined by reduced monocyte HLA-DR expression, identified a deleterious sepsis endotype. In that cohort, patients showing both immunosuppression and EBV reactivation during the first week after ICU admission had the highest mortality. Because both markers are routinely available in clinical practice, the combination appeared promising as a risk-stratification tool, especially as interest grows in delayed, individualized immunostimulation for patients whose immune systems fail to recover.

To extend this approach beyond EBV and beyond sepsis, the Lyon team drew on the REALISM cohort, which includes a mixed population of ICU patients: those with sepsis, severe trauma, major surgery, and burns. For these patients, both viral reactivation data and monocyte HLA-DR measurements were available. Viral reactivation was assessed by quantitative PCR testing of plasma for four herpesviruses: Epstein-Barr virus, cytomegalovirus, human herpesvirus 6, and herpes simplex virus 1. The researchers deliberately focused on a single time point, day 7 after admission, and included only patients who were still hospitalized in the ICU at that moment. This day-7 landmark analysis was chosen to mirror a real-world clinical scenario, because day 7 corresponds to the current inclusion criterion for initiating interferon-gamma as an immunoadjuvant treatment in ongoing clinical trials evaluating immunostimulation.

Immunosuppression was defined using a monocyte HLA-DR threshold of fewer than 8,000 antibodies bound per cell, a cutoff widely used in clinical trials and recently validated as a robust clinical threshold in more than 1,000 patients with septic shock. Viral reactivation was defined as at least one positive herpesvirus DNAemia at that time point. Applying these definitions to the 195 patients who had both measurements available at day 7, the investigators classified 45 patients as immunosuppressed with viral reactivation, 62 as immunosuppressed without reactivation, 20 as non-immunosuppressed with reactivation, and 68 as having neither marker. The primary endpoint was a composite of ICU-acquired infection and/or 90-day mortality; the authors note that the composite and 90-day mortality were chosen to capture clinically meaningful outcomes in a cohort with relatively low 28-day mortality.

Herpesvirus DNAemia proved frequent and increased progressively over the first week. By day 7, 33 percent of patients had at least one viral reactivation, and among these, 25 percent exhibited reactivation of more than one herpesvirus. These figures are in line with previous observations in septic shock. Yet when the researchers examined whether viral reactivation worsened prognosis, the answer was no. In this cohort, they found no clear evidence that viral reactivation was associated with a worse prognosis in terms of secondary infections or mortality when patients’ immune status was taken into account. Among immunosuppressed patients, no significant differences were observed between those with and without viral reactivation.

The picture changed when the analysis was reversed. Among patients presenting with viral reactivation, outcomes differed significantly according to immunosuppression status, and the same held true among patients without reactivation: outcomes differed significantly between immunosuppressed and non-immunosuppressed patients. A multivariable analysis that adjusted for age, Charlson comorbidity index, and SOFA score at ICU admission as standard confounders showed an unfavorable hazard ratio among patients with immunosuppression. Kaplan-Meier curves depicting the cumulative incidence of unfavorable outcomes over time reinforced the interpretation that immunosuppression, rather than viral reactivation, may be the main driver of the association with adverse outcomes. When the two markers were analyzed separately, low monocyte HLA-DR, below the 8,000 antibodies-per-cell threshold, was significantly associated with adverse outcomes, whereas viral reactivation was not.

The findings were robust across several sensitivity analyses. Results stratified by admission etiology were similar regardless of the group considered, whether sepsis, trauma, surgery, or burns. Additional analyses using 28-day mortality instead of the composite endpoint, a stricter monocyte HLA-DR threshold of fewer than 5,000 antibodies per cell, EBV reactivation alone, or the septic patient subgroup all produced findings consistent with the main result. Taken together, the authors conclude that in this mixed cohort of critically ill patients, they found no evidence that the combination of immunosuppression and herpesvirus reactivation at day 7 identifies a particularly severe patient endotype. Beyond initial disease severity, the association with adverse outcomes appears to be primarily driven by reduced monocyte HLA-DR expression rather than by viral reactivation, consistent with recent evidence supporting the prognostic value of this marker at day 7.

Why do these results diverge from the earlier sepsis study? The authors point to several methodological differences and caution that the two studies were not designed to address the same objectives. Rosiewicz and colleagues evaluated only EBV reactivation in patients with sepsis, whereas the Lyon team investigated four herpesvirus reactivations considered collectively in a mixed ICU population. The number of septic patients in the present cohort was also smaller, 76 compared with 124, so the authors cannot exclude insufficient statistical power in their subgroup analyses restricted to EBV and sepsis to reach statistical significance. The definition of reactivation differed as well: the earlier study counted any positive result during the first week after ICU admission, yielding a reported prevalence of 78 percent, whereas the present analysis specifically assessed viral status at a predefined time point on day 7 using a landmark analysis designed to reflect the clinical decision-making context in which initiation of immunostimulation would be considered. The authors also acknowledge that other potential sites of viral reactivation, particularly in the respiratory tract, were not assessed.

At this stage, the authors describe their findings as exploratory and hypothesis-generating. They suggest that the added value of viral reactivation in blood as a biomarker of adverse outcomes at a given time point may be limited in a mixed ICU population, though this does not preclude its potential value for understanding the underlying pathophysiology. Future studies, they argue, should focus on clinically relevant manifestations of viral reactivation rather than detection of DNAemia alone, or adopt a quantitative approach by evaluating viral load as a continuous variable instead of dichotomizing patients according to the presence or absence of viral DNAemia. From the perspective of patient stratification at day 7 for potential immunostimulatory therapies, and pending further evidence on the clinical significance of viral reactivation, other biomarkers of immunosuppression may be more relevant to combine with monocyte HLA-DR. These include lymphopenia, impaired interferon-gamma release, persistently elevated cytokine concentrations, or an elevated neutrophil-to-lymphocyte ratio. Integrating such markers into a composite score, the authors suggest, may better identify the subphenotype of patients with profound immunosuppression who are most likely to benefit from immunostimulatory treatment.

For clinicians weighing immunostimulation at the end of the first ICU week, the message is that a single immune marker may matter more than a viral one. Monocyte HLA-DR measurement, already available in many centers, appears to carry the prognostic weight in this mixed population, while the mere detection of herpesvirus DNA in plasma at day 7 did not distinguish patients at higher risk of secondary infection or death. Whether quantitative viral loads, respiratory-site sampling, or combined immune scores will change that picture remains a question for future research.

Subject of Research: Prognostic value of immunosuppression and herpesvirus reactivation measured at day 7 after ICU admission in critically ill patients

Article Title: Concomitant assessment of immunosuppression and Herpesvirus reactivation in blood at Day 7 after ICU admission in a mixed population

Article References: Demougeot, Q., Venet, F., & Monneret, G. (2026). Concomitant assessment of immunosuppression and Herpesvirus reactivation in blood at Day 7 after ICU admission in a mixed population. Critical Care, 30(1), Article 562. https://doi.org/10.1186/s13054-026-06437-4

Image Credits: AI Generated

DOI: 10.1186/s13054-026-06437-4

Keywords: ICU-acquired immunosuppression, monocyte HLA-DR, herpesvirus reactivation, Epstein-Barr virus, cytomegalovirus, sepsis, REALISM cohort, viral DNAemia, immunostimulation, prognostic biomarkers, critical care, landmark analysis