Cytomegalovirus in intensive care

P. Griffiths
DOI: https://doi.org/10.1002/rmv.646
2010-01-01
Reviews in Medical Virology
Abstract:People can require intensive care for a variety of reasons: burns, trauma, myocardial infarction, etc. and many of these patients have a stormy clinical course with a high mortality and/or a protracted stay in the intensive care unit (ICU). The development of sepsis and the need for mechanical ventilation are well-recognised complications associated with an adverse clinical outcome. In cases of sepsis, a pro-inflammatory Th1 cytokine response of IFN , TNF and IL2 is reported early after the insult, followed by a later immunosuppressive phase including IL10 and IL4 [1,2]. Indeed, high concentrations of IL10 are a risk factor for a generally poor prognosis [3]. At autopsy, decreased numbers of B-lymphocytes, CD4 Tlymphocytes and follicular dendritic cells are reported, profoundly impairing the function of the adaptive immune system [2]. Over the past decade or so, a series of studies have linked the presence of CMV with adverse outcomes such as increased length of stay and/ or mortality [4–8]. While these studies have been innovative, they have suffered from some disadvantages, for example not all the studies defined CMV serostatus at baseline. In addition, some used PCR to detect CMV whereas others used antigenaemia or culture, while no studies used fully quantitative real time PCR. The number of patients studied was also generally small and the studies were not blinded. Indeed, some of the results were available in real time so that some patients were given ganciclovir which may have altered the natural history of this infection. A recent systematic review [9] and another meta-analysis [10] describe the differences between these and other published studies, highlighting the great variability in the underlying reasons for patients being admitted into ICU. Nine studies were prospective and four retrospective with variable sampling for CMV, always from blood but sometimes from urine and respiratory secretions as well. Not surprisingly, CMV was detected more frequently when the studies recruited only CMV seropositive patients and when PCR or antigenaemia were used instead of cell culture [9,10]. There was no decrease in the detection of CMV when studies before or after 2001 were considered showing that CMV infection has been endemic for some time in ICUs [10]. The odds ratio for all-cause mortality among patients in whom CMV was detected, compared to those in whom it was absent, was 1.9 [10]. One study justifies discussion in depth [11]. Investigators in Seattle recruited patients from six ICUs in two separate hospitals. They performed daily screening of new admissions to exclude transfers within the hospital, to exclude CMV seronegatives and to exclude any who were known to be immunocompromised. Plasma PCR was tested three times per week and the results were available only retrospectively so that natural history as seen by the clinicians could not be affected. Importantly, the assay was fully quantitative and used appropriate negative and positive controls. A clinically relevant composite end point of requiring hospitalisation or having died by day 30 post admission was used. In total, 1954 samples were tested from 120 patients for a median of 11 per patient (range 1–89). The results showed that CMV viraemia was detectable in 33% of patients and viraemia greater than 1000 genomes/ ml of plasma in 20% of patients. The median maximum viral load was 3.3 logs. CMV first became detectable at a median of 12 days after admission and the median duration of viraemia was 17 days. There were differences between patients admitted to the different types of ICUs with those admitted to the burn ICU having the greatest incidence of CMV viraemia. A multivariable analysis showed that being admitted to the burn ICU, requiring mechanical ventilation and developing a major infection were all risk factors for hospitalisation or death by day 30. In addition, the presence of CMV viraemia and especially CMV of high viral load was an additional and independent risk factor for poor outcome. E D I T O R I A L
What problem does this paper attempt to address?