Viral DNAemia and DNA Virus Seropositivity and Mortality in Pediatric Sepsis

Key Points Question Are viral DNAemia and DNA virus seropositivity associated with increased mortality in pediatric sepsis? Findings In this multicenter cohort study of 401 children with severe sepsis, viral DNAemia was common, with at least 1 virus detected in 191 patients (48%). Cytomegalovirus, adenovirus, BK polyomavirus, and human herpesvirus 6 DNAemia and Epstein-Barr virus seropositivity were associated with increased mortality. Meaning Further study is warranted to determine whether cytomegalovirus, adenovirus, BK polyomavirus, and human herpesvirus 6 DNAemia and Epstein-Barr virus seropositivity only reflect mortality risk or contribute to mortality in children with sepsis.


Introduction
Sepsis remains the leading cause of pediatric mortality worldwide.While sepsis recognition campaigns have improved early mortality, 1,2 ongoing contributors to mortality are unclear.4][5][6] Detection of viral DNA in the bloodstream (viral DNAemia), presumed due to reactivation of latent infection, is an independent predictive factor associated with mortality in immunocompromised patients in the ICU. 7][10][11][12][13][14] While mechanisms triggering reactivation are not fully defined, conditions found in critical illness-including new infection, physiologic stress, immunosuppression, and trauma-have been implicated. 10,15,16rsistent and latent herpesvirus infections have been associated with human disease.8][19][20] Persistent and latent cytomegalovirus (CMV) infection has been associated with risk of cardiovascular disease, including ischemic heart disease, stroke, and cardiovascular death. 21,22Conversely, some evidence suggests that latent DNA virus infection increases resistance to other infections.For example, murine studies 23,24 have shown that herpesvirus latency upregulates innate immunity and shields against certain bacterial infections, and that gammaherpesvirus latency protects against inflammatory responses in acute adenoviral disease, suggesting the association between host and latent virus may be symbiotic or even protective.
Previously, Davila et al 8 studied viral DNAemia among patients in a single-center PICU.The investigators detected plasma DNAemia with multiple viruses and found associations with immunocompromised state, risk of secondary infection, and longer PICU stay and a trend toward increased mortality.Our present multicenter cohort investigation uses mortality as the primary outcome.We further measured selected herpesvirus antibody levels to elucidate a role for reactivation.We used biobanked samples collected from a parent study investigating pediatric sepsis phenotypes. 25

Enrollment Criteria and Sepsis Definition
The parent study 25  abnormal heart rate, breathing, temperature, and/or white blood cell count with organ failure. 25munocompromised state at time of sepsis onset was based on underlying diseases or immunosuppressive treatments.Blood was collected from all enrolled patients twice per week to 28 days of admission or until death or PICU discharge.As a secondary analysis of a prior prospective study, our cohort study followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guideline.

Serologic Testing
The earliest available plasma sample day from each patient's PICU stay (sample 2 from the serial collection used for 275 patients, sample 1 used for 125 patients, sample 3 used for 1 patient) was simultaneously tested for both DNAemia of the 8 viruses and IgG antibodies to CMV, EBV (viral capsid antigen), HSV-1, and HHV-6.Testing was performed using multiplex flow immunoassays (BioPlex 2200 system; BioRad) except for HHV-6, which was tested using an indirect fluorescence assay (SCIMEDX Corporation).Results were expressed as positive, negative, or equivocal.When sufficient volume was available, tests with equivocal results were repeated.Patients younger than 18 months or who received intravenous immunoglobulin (IVIG) prior to testing were excluded from analysis of serologic results due to possible false-positive results from transplacental transfer of maternal antibody or from IVIG.

Patient Classifications
Patients were defined as having viral DNAemia based on qPCR detection of 1 of the 8 viruses tested in 1 or more plasma samples.Because of its frequency and uncertain relationship to disease, TTV was not included in most analyses.Patients were defined as being seropositive based on detection of virus-specific IgG.For the 4 herpesviruses (CMV, EBV, HSV-1, and HHV-6), results of combined qPCR and serologic testing classified patients among 4 groups: no infection (IgG negative and DNAemia negative), acute infection (IgG negative and DNAemia positive), presumed reactivated infection (IgG positive and DNAemia positive), or latent infection without reactivation (IgG positive and DNAemia negative).

Statistical Analysis
Data were collected from 2015 to 2018.Samples were assayed from 2019 to 2022.Data were analyzed from 2022 to 2023.Demographic characteristics, illness severity, PICU admission diagnoses, and specific medical history categories were summarized by immunocompromised state using counts (percentages) and medians (IQR) (Table ).Proportions of patients with positive viral DNAemia and seropositivity were calculated among patients with testing results available.Mortality in the PICU was considered the primary outcome.Univariate analysis was used to identify variables associated with both mortality and presence of viral DNAemia or seropositivity for subsequent multivariate analysis.Adjusted odds ratios (AORs) and 95% CIs estimating the increase in odds of mortality associated with DNAemia were adjusted for age, Pediatric Risk of Mortality (PRISM) score at ICU admission, whether or not the patient was deemed healthy prior to hospitalization, and immunocompromised status.The AOR for increased mortality associated with seropositivity was additionally adjusted for receipt of blood products in the PICU prior to sampling for antibody  d The worst physiologic values 2 hours prior to intensive care unit (ICU) admission through 4 hours post ICU admission were used to calculate PRISM.Scores range from 1 to 26, with higher scores indicating higher mortality risk.
e Patients may have had multiple documented infections at eligibility; counts are not mutually exclusive.
f Viral infection(s) at eligibility was determined by respective research coordinator medical record review and does not include results of investigative viral testing subsequently performed for this study.
g Four patients were listed as both previously healthy and immunocompromised by the site research coordinator.Since no further information was available, patients were reported and analyzed as designated.

Study Population
The study population of 401 patients is described in the

Viral DNAemia
Figure 1A shows the occurrence of viral DNAemia and Figure 1B shows the number of viral DNAemias detected in patients who were or were not immunocompromised at the time of sepsis onset. Excluding

Viral Serology
After excluding 108 patients who were younger than 18 months and 29 who received IVIG prior to their sample being obtained, 264 patients remained for analysis of herpesvirus serologic testing.Viral

Combined qPCR and Serology for Herpesviruses
A detailed breakdown of results of qPCR and serologic testing is provided in eTable 3 in Supplement 1.

Mortality
Of the 401 patients, 44 (11.0%) died while in the PICU.eTable 4 in Supplement 1 shows differences in univariate analysis between those who did or did not die.Patients who died were older, although the difference was not statistically significant (median age, 7.9 [IQR, 1.2-14.5]vs 5.4 [IQR, 1.4-12.0]years; P = .08),and had higher baseline PRISM scores 26,27

Viral DNAemia and Mortality
Adjusted mortality according to the presence of viral DNAemia is shown in A detailed breakdown of patient characteristics and all statistical comparisons can be found in eTable 1 in Supplement 1.The analysis of the number of viruses detected via polymerase chain reaction analysis excludes torque teno virus (TTV).Patients younger than 18 months or those who had intravenous immunoglobulin administration on the same day or before serology sample collection are excluded from serology analyses, leaving a sample size of 264.BkPyV indicates BK polyomavirus; CMV, cytomegalovirus; EBV, Epstein-Barr virus; HHV-6, human herpesvirus 6; and HSV-1, herpes simplex virus 1. a P Յ .01.

Viral Serology and Mortality
As shown in Figure 2  were EBV seropositive.Mortality was higher in patients with more than 1 herpesvirus IgG detected (12 of 65 [18.5%] in patients who were and 12 of 113 [10.6%] in patients who were not immunocompromised at sepsis onset) (Figure 3B).

Combined qPCR and Serology Association With Mortality
Mortality among all patient classifications described in the Methods section is displayed in Figure 4,

Discussion
Previously published single-center studies have reported frequent detection of viral DNAemia in adult 10,12,13,[28][29][30] and pediatric 8,11 patients with sepsis that is correlated with immunocompromised state and severity of illness.Our present large multicenter cohort study validates and extends those findings.We report herein that viral DNAemia was common in pediatric patients admitted to the PICU with severe sepsis, including those who were and were not immunocompromised at the time of sepsis onset.In both groups, mortality increased with increasing number of viral DNAs detected in b P < .001,no infection vs reactivated infection.
c P = .008,no infection vs reactivated infection.
d P = .001,no infection vs latent infection.
Analysis of herpesvirus serologic status in combination with measurements of viral DNAemia allowed separation into 4 distinct groups: uninfected, acute infection, latent infection, and presumed reactivation.DNAemia with any of the 4 herpesviruses was much more common in those who were seropositive for that virus compared with those who were seronegative, consistent with reactivation, and suggesting that primary infection in this setting was uncommon.We also found that seropositivity for EBV but not for any of the other 3 herpesviruses was associated with increased mortality in multivariate analysis.Only 2.0% of patients who were seronegative for EBV died compared with 15.1% of those who were seropositive for EBV.Possible explanations may be that EBV seropositivity is a biomarker for more severe illness that may not have been fully accounted for by the multivariate analysis or, alternatively, that being latently infected by EBV renders an individual more vulnerable to mortality from sepsis.
In addition to expanding the number of study participants compared with previous pediatric studies, this study also expanded the number of DNA viruses tested, adding BKPyV and B19V to those tested by Davila et al. 8 As expected, not all DNA viruses behaved similarly.While CMV, HAdV, HHV-6, and BKPyV DNAemia conferred a significant risk of mortality, plasma DNAemia with HSV-1 and B19V were detected less often than the other viruses and were not associated with mortality.
[33][34] Torque teno virus is commonly detected in patients in the ICU, 9 and studies have suggested a direct correlation of TTV level with degree of T-cell dysfunction in patients undergoing hematopoietic cell transplant 35 and solid organ transplant. 36Previous studies have not examined the association between TTV and mortality; our present study did not find higher levels of TTV associated with mortality.
The origin of viral DNAemia is currently not fully understood.We presumed that detection of Another important outstanding question is whether viral reactivation has adverse effects on patients with sepsis and possibly contributes to adverse outcomes, including death.A direct way to assess contributions of viral reactivation to outcomes of sepsis is to evaluate effects of antiviral treatment.Limaye et al 37 performed a prospective double-blind, placebo-controlled randomized trial of ganciclovir in adult patients undergoing critical care who were at risk for CMV reactivation by virtue of being CMV seropositive and had either sepsis or trauma with respiratory failure.Antiviral therapy did not reduce plasma interleukin 6 levels but was associated with a reduction in CMV DNAemia and an increase in ventilator-free days.There was no impact on mortality.Papazian et al 38 performed a similar double-blind, placebo-controlled randomized trial of 76 adults with CMV reactivation requiring mechanical ventilation and demonstrated no benefit in ventilator-free days in the group treated with preemptive ganciclovir.These results suggest that reactivated viral infection could be a risk biomarker for severity of sepsis rather than a direct contributor.

Strengths and Limitations
Strengths of our study include a large multicenter population of children with sepsis and the use of serologic testing to evaluate whether herpesvirus DNAemia was a result of acute infection or reactivation from latency and to define a possible role for latent infection in sepsis pathogenesis.Our study also has important limitations.The number of children who died was small compared with deaths in studies of adults, leading to relatively wide 95% CIs in multivariable analyses.We lacked access to samples to assess frequency of viral DNAemia and seropositivity in non-critically ill children.
Nevertheless, the herpesvirus seropositivity rates we found in our assayable pediatric sepsis samples were similar to those previously reported in age-matched healthy pediatric patients, [39][40][41][42][43][44][45][46]  that seropositive status may not be increased in children with sepsis as much as it is increased in children with higher proclivity to die of sepsis.We were unable to assess whether DNA virus reactivation occurred in specific locations such as the respiratory tract, which might not be detected in plasma. 16,29,47This could have led us to an underestimation of the frequency of DNA virus reactivation.We also lacked access to data regarding blood products received before the PICU stay.
Although blood-bank filtering techniques all but eliminate passive transfer of viral DNAemia during transfusion, IgG antibody transfer can occur.Larger studies designed to account for blood product receipt prior to PICU admission and sepsis onset are needed to validate our findings.

Conclusions
Our cohort study found that viral DNAemia was both common in pediatric patients with severe sepsis and independently associated with mortality.Cytomegalovirus, HAdV, BKPyV, and HHV-6 DNAemia were each significantly associated with mortality.Detection of viral DNAemia with multiple viruses during PICU admission increased the risk of mortality in patients who were or were not immunocompromised at the time of sepsis onset.Seropositivity for EBV but not for the other 3 herpesviruses studied was also associated with increased mortality and was not altered by whether or not the patient had EBV DNAemia.Our study does not address any causal nature of these associations, but our findings do identify a group of children at risk for sepsis mortality.While unrecognized confounding may explain the associations, active or latent infection with DNA viruses may reflect severity of illness or exert an adverse effect on the ability of children to respond to sepsis.Further investigation of underlying biology of these viral DNA infections in children with sepsis is warranted.
of the present cohort study included 401 patients from 9 PICUs in the Eunice Kennedy Shriver National Institutes of Child Health and Human Development Collaborative Pediatric Critical Care Research Network approved by the central institutional review board of the University of Utah.Enrollment criteria included severe sepsis diagnosis, presence of a central venous or arterial catheter, and age range from 44 weeks of gestation to younger than 18 years.Written informed consent was obtained from 1 or more parents or guardians for each child.Assent was obtained when the child was able.Parents self-identified child sex, race, ethnicity, and gender voluntarily.Race and ethnicity are considered important because they can affect outcome in children in relation to health disparity.Our research network was specifically conceived and designed by the Eunice Kennedy Shriver National Institutes of Child Health and Human Development to represent all racial and ethnic groups for equity in research participation.Severe sepsis was identified by suspected infection with

a
Unless otherwise indicated, data are expressed as No. (%) of patients.Percentages have been rounded and may not total 100.b Conditions included hematologic malignant neoplasm (n = 35), solid tumor (n = 16), hematopoietic stem cell transplant (n = 23), solid organ transplant (n = 10), chronic corticosteroid use (n = 59), congenital immunodeficiency (n = 10), rheumatologic disease (n = 7), functional or anatomic asplenia (n = 3), or other (n = 19).Patients could have more than 1 diagnosis; hence the total is greater than the total number of immunocompromised patients.c Includes self-identified American Indian or Alaska Native, Asian, and Native Hawaiian or Other Pacific Islander according to the National Institutes of Health demographic sheet provided to parents at the time of consent.
seropositivity was common, ranging from 82 of 246 (33.3%) for HSV-1 to 107 of 254 (42.1%) for CMV, 152 of 251 (60.6%) for EBV, and 253 of 257 (98.4%) for HHV-6.Percentage positivity for each virus by immunocompromised status are shown in Figure1C, and percentage of viral IgGs detected are shown in Figure1D.Seventy-seven patients were seropositive for 1 virus, 72 for 2, 51 for 3, and 55 for all 4 herpesviruses.Patients who were immunocompromised at sepsis onset were more likely than nonimmunocompromised patients to be seropositive for EBV, CMV, and HSV-1.Associations between herpesvirus serologic status and demographic characteristics, illness severity, presence of infection at study entry, and PICU admission diagnoses are shown in eTable 2 in Supplement 1.

Figure 3 .
Figure 3. Pediatric Intensive Care Unit (PICU) Mortality by Number of Viruses Detected by Plasma Quantitative Real-Time Polymerase Chain Reaction (qPCR) Analysis and Herpesvirus Serology

Figure 4 .
Figure 4. Association Among Plasma Quantitative Real-Time Polymerase Chain Reaction Analysis, Viral Seropositivity, and Mortality

viral
DNAemia was related to reactivation from latency with new viral replication and production of infectious virus, possibly resulting from sepsis-induced immune suppression.However, detection of viral DNA could also result from release of nonviable viral DNA from latently infected cells that are damaged during sepsis.Assays that detect viable virus or other indicators of viral replication, such as viral messenger RNAs or viral proteins associated with active infection, could help clarify this issue.

JAMA Network Open | Critical Care Medicine Viral
DNAemia and DNA Virus Seropositivity and Mortality in Pediatric Sepsis Abbreviations: PICU, pediatric intensive care unit; PRISM, Pediatric Risk of Mortality.
Viral DNAemia and DNA Virus Seropositivity and Mortality in Pediatric Sepsis All logistic regression analyses were performed in SAS, version 9.4 (SAS Institute).Select figures were generated using R, version 4.2.1, and package ggplot (R Project for Statistical Computing).P values were based on a 2-sided alternative and considered significant if less than .05.No adjustments for multiple comparisons were made.

Table .
chromosomal abnormality, and metabolic or mitochondrial disease.Investigative viral testing performed for this study was not used for defining baseline disease state.
Viral DNAemia and DNA Virus Seropositivity and Mortality in Pediatric Sepsis Detection of DNAemia with 3 or more viruses increased risk of death with morality in 10 of 28 patients (35.7%).Higher viral loads for each individual virus, including TTV, were not associated with increased mortality.
JAMA Network Open.2024;7(2):e240383.doi:10.1001/jamanetworkopen.2024.0383(Reprinted) February 26, 2024 6/13 Downloaded from jamanetwork.comby guest on 03/04/2024 a more detailed breakdown in eTables 5 to 8 in Supplement 1.For each virus, mortality was higher in seropositive patients compared with those who were seronegative, with low mortality in Patients who were seropositive for CMV or HHV-6 each had higher mortality if they also had DNAemia with the corresponding virus.In contrast, EBV-seropositive patients had the same mortality (17 of 113 [15.0%] and 6 of 39[15.4%]),regardless of the presence of EBV DNAemia.eTables 9 to 12 in Supplement 1 show ferritin and C-reactive protein levels.

implying JAMA Network Open | Critical Care Medicine Viral
DNAemia and DNA Virus Seropositivity and Mortality in Pediatric Sepsis JAMA Network Open.2024;7(2):e240383.doi:10.1001/jamanetworkopen.2024.0383(Reprinted) February 26, 2024 9/13 Downloaded from jamanetwork.comby guest on 03/04/2024 Patient Characteristics at ICU Admission by Virus Detection eTable 2. Patient Characteristics at ICU Admission by Results of Viral Serologic Testing eTable 3. Comparisons of Herpesvirus PCR and Serology Status eTable 4. Univariable Analysis of Admission Demographic and Clinical Factors Associated With Mortality eTable 5. Comparisons of EBV PCR and Serology Status and Mortality eTable 6. Comparisons of CMV PCR and Serology Status and Mortality eTable 7. Comparisons of HSV-1 PCR and Serology Status and Mortality eTable 8. Comparisons of HHV-6 PCR and Serology Status and Mortality eTable 9. Comparison of Ferritin, CRP, and PRISM Across EBV Virus Positivity Cohorts in Vital Status Categories eTable 10.Comparison of Ferritin, CRP, and PRISM Across CMV Virus Positivity Cohorts in Vital Status Categories eTable 11.Comparison of Ferritin, CRP, and PRISM Across HSV-1 Virus Positivity Cohorts in Vital Status Categories eTable 12.Comparison of Ferritin, CRP, and PRISM Across HHV-6 Virus Positivity Cohorts in Vital Status Categories