Herpes simplex virus and Cytomegalovirus reactivation among severe ARDS patients under veno-venous ECMO

Herpes simplex virus (HSV) and Cytomegalovirus (CMV) belong to the Herpesviridae family and are characterized by an often asymptomatic primo-infection generally during childhood followed by a latency phase. In immunocompromised subjects, Herpesviridae are common viral causes of opportunistic infections. But HSV and CMV reactivations are also frequently reported in intensive care unit (ICU) non-immunocompromised patients [1, 2]. Reactivation ranges from 13 to 64% and 15 to 45% for HSV and CMV, respectively [3, 4]. Herpesviridae reactivation in immunocompetent ICU patients is associated with poorer outcome [5]. HSV pulmonary reactivation has been described to be associated with a longer mechanical ventilation (MV) duration, ICU stay and mortality [2, 6, 7]. CMV reactivation is also associated with a higher mortality, MV duration and ICU length of stay [8]. In particular, CMV has been identified as a cause of persistent acute respiratory distress syndrome (ARDS) [9] and has also been shown to increase the mortality in ARDS patients [10]. However, despite these associations, the debate on the proper pathogen role of Herpesviridae rather than being a witness of patients’ severity is still ongoing. Studies failed to demonstrate that CMV prophylaxis was able to decrease Il-6 plasma levels in CMV seropositive critically ill patients [11] or to decrease mortality [12]. The role of Herpesviridae pre-emptive treatment among ICU patients has been recently evaluated in a randomized controlled trial (RCT) (NCT 02152358). The data concerning HSV showed that preemptive acyclovir did not decrease the duration of MV although a trend towards lower mortality was found in treated patients [13].

The most frequent risk factors for CMV and HSV reactivation in the ICU are patients severity, sepsis, prolonged MV [14], high-dose corticosteroid therapy, acute renal failure or massive transfusion [15], with a strong association for MV and sepsis [16]. Patients under veno-venous extracorporeal membrane oxygenation (VV ECMO) for severe ARDS [17] often combine several or all of these risk factors [18]. Despite the uncertainties regarding the exact role of Herpesviridae reactivation in immunocompetent critically ill patients, it might add to the pulmonary pathology in patients with ARDS. In experimental studies, CMV reactivation led to increased pulmonary fibrosis [19] and accessing bacterial pneumonia [20]. These findings suggest that Herpesviridae-related pulmonary pathology may be causally linked to the clinical disease course following ARDS onset, especially in the most severely ill patients who require prolonged mechanical ventilation, and might particularly concern patients under ECMO. However, despite the tight link that seems to exist between Herpesviridae, mechanical ventilation and ARDS, no study has investigated the occurrence of HSV and/or CMV reactivation in patients under VV ECMO. In this study, we aimed to assess the frequency of herpesviruses reactivation during ECMO course and to determine its impact on patients’ prognosis.

During the study period, 181 patients were admitted to our ICU for severe ARDS requiring a VV ECMO for 2 days or more (see flowchart, Fig. 1). Of these, 58 patients were excluded because of immunosuppression (44 patients), HSV/CMV reactivation at the time of ECMO implementation (10 patients) or acyclovir/ganciclovir treatment before ECMO (4 patients). Among the 123 patients included, 67 patients (54%) experienced HSV and/or CMV reactivation during the ICU stay and 56 (46%) were free from HSV/CMV reactivation at the time of ICU discharge or death.

Until today, no data have been published concerning Herpesviridae reactivation in ICU patients under VV ECMO for severe ARDS. In this retrospective study covering a 5-year period, we found that HSV/CMV reactivation was frequent and concerned more than half non-immunocompromised patients, which is higher than that described in previous studies including all ICU patients [4, 8, 14]. This might be explained by several reasons: the use of PCR to diagnose reactivation with a higher sensitivity than older technics, the age of our cohort of patients (with a high probability of seropositivity for HSV and CMV at ICU admission) and the frequency of sepsis with a probable induced “immunoparalysis” [24]. In our cohort, HSV reactivation occurred earlier than CMV reactivation and the median time of reactivation for both viruses was comparable to what is described in “non-ECMO” patients [25]. CMV viral loads in blood and BAL were high in almost all patients. Elevated CMV viremia is associated with a higher risk of death or prolonged hospitalization [4].

Patients included were comparable except for the duration of MV before ECMO that was longer in the reactivation group. It is well known that MV is a risk factor for Herpesviridae reactivation with a strong association for CMV [16].

We found that Herpesviridae reactivation was associated with a prolonged MV, this association persisting in multi-variate analysis. We also found in these patients a prolonged ECMO duration, ICU, and hospital stay, although not confirmed in multivariate analysis. In a recently published meta-analysis, Li et al. [8] showed that CMV reactivation was associated with an increase of 9 days in MV and a 12 days increase in ICU stay. These results confirmed those published by Limaye et al. [4], which showed that CMV viremia among ICU patients was associated with a higher risk of death or prolonged ICU stay > 30 days. Similarly, in a case–control study [15], CMV reactivation was associated with a prolonged duration of MV and ICU stay. In a specific population of ARDS patients, Ong et al. [10] demonstrated that patients with CMV reactivation had a 15 (10–26) days median duration of MV as compared to 8 (6–12) days for non-reactivated patients. ICU length of stay was also longer [16 (11–28) vs. 9 (7–14) days] for reactivated patients. Same results have been published concerning HSV reactivation [2, 6], especially during ARDS. Our findings suggest that Herpesviridae reactivation is associated with worse outcomes for ARDS patients including when they are under ECMO. When examining the impact of each virus separately, we found that both HSV and CMV were associated with a prolonged MV, and also ECMO duration for HSV. Co-reactivation had a negative effect not only on MV and ECMO duration but also on ICU and hospital stay, as compared to patients free from reactivation or with only one virus. HSV was independently associated with a longer duration of MV and hospital length of stay but, surprisingly, a shorter ICU stay. These results highlight the potential negative role of HSV in ARDS patients under ECMO. Very recently, Luyt et al. [13] showed that preemptive treatment with acyclovir, compared to placebo, for mechanically ventilated patients with oropharyngeal HSV reactivation, was not associated with shorter MV duration. However, a trend towards lower day-60 mortality was observed in the acyclovir group. In our cohort, more than half of the patients were treated after the diagnosis of viral reactivation. Treatment with acyclovir or ganciclovir did not improve the outcomes, with a trend for longer duration of MV in the sub-group of treated patients. These results might be explained by the fact that anti-viral treatment was decided by clinicians more frequently in case of worsening respiratory status, persisting fever or end-organ HSV/CMV disease, and so reserved for the most severe patients. We did not find any increase in renal failure in patients receiving antiviral drugs, which was also noticed in Luyt et al.’s study [13]. However, we cannot exclude any other side effects.

Our study has some limitations. First, the retrospective design of our cohort, counterbalanced by the important number of patients included during this 5-year period. Second, the applicability of our results to the general population of patients under ECMO must be considered cautiously considering the high rate of patients treated with antiviral drugs after reactivation. However, in non-EMCO patients, routine screening of Herpesviridae has been reported as well as the use of antiviral treatment despite the lack of recommendation [3, 15, 16]. Third, few patients developed an isolated CMV reactivation. This precludes to conclude clearly on the specific impact of CMV in our cohort of patients. Fourth, our methods do not prevent competing risks. In particular, the difference in MV duration between reactivated and non-reactivated patients might have been influenced by the high mortality reported. However, this mortality was similar in both groups and the difference of MV duration persisted when considering only the patients discharged alive from the ICU.

Finally, despite the statistical association, it is not possible to conclude whether Herpesviridae reactivation is directly responsible for worse clinical outcomes or if it is a consequence and a witness of the severity of the disease, as in non-ECMO populations [3].