Interleukin‐18 and tumor necrosis factor‐α are elevated in solid organ transplant recipients with possible cytomegalovirus end‐organ disease

End‐organ cytomegalovirus (CMV) disease can be life threatening to solid organ transplant recipients. Diagnosis is often complicated by variation in amount of CMV DNA in plasma and the need for an invasive procedure to obtain a biopsy of the suspected affected organ, which can delay recognition and treatment. Several inflammatory cytokines are elevated in CMV disease, and the purpose of this study was to determine if they could be used to distinguish solid organ transplant recipients with CMV DNAemia alone from those with possible end‐organ CMV disease. We found that regardless of pre‐transplant CMV serostatus, plasma interleukin (IL)‐18, tumor necrosis factor‐α (TNF‐α), and amount of CMV DNA in plasma were increased in possible end‐organ CMV disease, with elevated IL‐18 associated with increased odds of possible end‐organ CMV disease even after adjusting for amount of CMV DNA. These findings highlight IL‐18 and TNF‐α as potential non‐invasive markers of possible end‐organ CMV disease regardless of transplanted organ or serostatus in solid organ transplant recipients.

replication leading to clinical signs or symptoms, can progress to CMV syndrome and/or end-organ disease. CMV syndrome is often characterized by malaise, fever, or laboratory abnormalities in the setting of detection of CMV DNA in plasma, while end-organ CMV (EO-CMV) indicates the presence of replicating CMV in an organ (gastrointestinal tract, liver, lung, etc.) leading to direct viral-induced damage. 1,6 Distinguishing between CMV syndrome and EO-CMV is important. While CMV syndrome can often be treated with oral valganciclovir alone, EO-CMV more often requires intravenous ganciclovir, especially in organ-or life-threatening cases, and often requires longer courses of therapy. 2 The gold standard for diagnosing EO-CMV remains documenting evidence of CMV by histopathology, viral culture, DNA hybridization, or immunohistochemistry via biopsy of the suspected infected organ. 1,7 This often necessitates an invasive procedure such as a colonoscopy, endoscopy, or bronchoscopy, which are expensive and confer additional risk to the patient. Clinicians therefore sometimes prefer to treat as "CMV colitis," for example, without formally documenting this diagnosis with a colonoscopy and biopsy. Though CMV DNA is often detected in plasma from kidney and liver transplant recipients with gastrointestinal EO-CMV, this is not always the case, particularly among R+ patients. 8 Hence, a rapid, non-invasive method of diagnosing EO-CMV is desirable.
A hallmark of CMV infection and replication in epithelial and myeloid cells is the induction of a number of innate immunity pathways. [9][10][11] One such innate signaling pathway is the inflammasome, which, when activated, leads to an inflammatory type of cell death called pyroptosis and the release of the proinflammatory cytokines interleukin (IL)-1β and IL-18. 12 CMV infection leads directly to activation of the inflammasome, and IL-18 is elevated in D+/R− kidney transplant recipients who develop CMV DNAemia without prophylaxis. 13,14 Plasma IL-18 levels are a sensitive indicator of acute infection with other chronic viruses, such as HIV and hepatitis C virus (HCV), 15,16 and begin to rise prior to alanine transaminase levels in acute HCV infection. 17 Therefore, we sought to determine if IL-18 or other cytokines produced by innate immune recognition of CMV would distinguish SOTRs with EO-CMV from those with CMV DNAemia but not EO-CMV. We report that elevated levels of IL-18 and TNF-α are markers of possible EO-CMV and may prove useful in distinguishing this disease entity from more mild forms of CMV disease.

Clinical data collection
Clinical and demographic data were extracted from the electronic health record (EHR) via manual chart review as outlined in Table 1

Cytokine measurement
Residual plasma from peak CMV DNAemia time points and unde- thaw. If an analyte signal was below background, it was set to 0 and if detectable, but below the manufacturer's internal lower limit of detection, to the lower limit of detection.

Statistical analysis
Data were analyzed using the statistical computing software R ver-  test was applied to cytokine values to determine differences between peak and undetectable CMV DNAemia. Generalized linear models with logit link function and binomial distribution were used to generate odds ratios (OR) and 95% confidence intervals (CI) for possible EO-CMV.
Two-sided p-values <.05 were considered significant. An exploratory analysis comparing models was performed by generating predictions of the original dataset using the generalized linear models, and modeling the discrimination of those predictions by calculating the area under the receiver operator curve (AUC) using the pROC and lmtest packages in R. 18,19 Linear regression models were generated to compare possible correlation between time from transplant and cytokine levels.

Cohort characteristics
A total of 275 SOTRs had CMV DNAemia within the health system during the study period. Of those, 44 had residual plasma stored in the microbiology lab, and 29 of those patients were determined to have CMV DNAemia not attributable to secondary reactivation in the context of another infectious process ( Figure 1). Seven (24%) had possi-    Figure 3A). Given that CMV DNA can be low or undetectable in D+/R+ liver transplant recipients with gastrointestinal EO-CMV, 8 we also examined the R+ subset in the cohort and found that IFN-γ, IL-10, IL-12p70, IL-18, and TNF-α increase during CMV DNAemia in patients that were CMV R+ prior to transplant ( Figure 3B). This also suggests that these cytokines are elevated during CMV reactivation, and not just primary infection. 13

Elevated levels of IL-18 and TNF-α are associated with possible CMV end-organ disease
To determine if any of these cytokines were elevated in possible EO-CMV compared to DNAemia alone, differences in the median value of these cytokines at peak CMV DNAemia between the two disease subgroups were compared. Levels of IL-18, IL-1β, IL-4, and TNF-α at peak CMV DNAemia were significantly higher in the possible EO-CMV group compared to the DNAemia alone group (Figure 4). There was no significant difference in IFN-γ, IL-10, IL-12p70, IL-13, IL-2, IL-6, or IL-8 between the groups. Furthermore, there was no significant correlation between months since transplant and peak cytokine levels ( Figure S1).

DISCUSSION
Here we present evidence that elevations of IL-18 and TNF-α are asso-  Note: Unadjusted odds ratios and 95% CI were generated using a generalized linear model as described in materials and methods. Adjusted odds ratios and 95% CI were generated using a model containing both Log 2 IL-18 and Log 10 CMV. study was over 10,000 pg/ml while the max value of IL-1β was less than 1 pg/ml). This is not unique to the SOT population and reflects previous studies that demonstrate IL-18 is a more robust measure of inflamma-some activation in plasma. 16,17 Furthermore, there may be a specific relationship between IL-18 and CMV disease in SOTRs as was pointed out by a recent study that found certain polymorphisms in the IL-18 promoter were associated with CMV DNAemia after stopping prophylaxis in renal transplant recipients. 24 While IFN-γ and IL-10 were not significantly different between those with CMV DNAemia and those with possible EO-CMV, they were elevated during CMV DNAemia overall. This is consistent with previous literature that found these cytokines are elevated during CMV DNAemia and/or disease. 13,25 However, these cytokines are more commonly associated with adaptive immunity rather than innate immunity. 26,27 While there were some outliers in the cytokines measured in both groups, there was significant heterogeneity among the outliers, meaning that an outlier in one cytokine was rarely an outlier in another. However, there was one patient in the pEO-CMV group that had the high-  Despite these limitations, this study raises the possibility that a noninvasive measurement of one or more of these cytokines, in combination with plasma CMV DNA, may prove useful in the clinical evaluation of patients with possible EO-CMV. Under circumstances where a tissue biopsy is impractical or contraindicated, these assays could lend weight to clinical decision-making and formulation of treatment plans.
This conclusion is both supported by the novel data presented here, as well as by a priori biological plausibility related to the important role of IL-18 and TNF-α in CMV infection. Future prospective studies with biopsy-proven EO-CMV would be helpful to confirm these findings.

CONFLICT OF INTEREST
The authors declare that there is no relevant conflict of interest to disclose.

AUTHOR CONTRIBUTIONS
Andrew H. Karaba

DATA AVAILABILITY STATEMENT
Reasonable requests to the corresponding author for deidentified data will be granted.