Cytomegalovirus Reactivation in “Immunocompetent” Patients: A Call for Scientific Prophylaxis

Cytomegalovirus (CMV) is a well-known pathogen in immunosuppressed patients and has been receiving increasing attention as a potential pathogen in critically ill patients with sepsis. After primary infection, this betaherpesvirus is not eradicated but establishes life-long infection in its host. CMV is dispersed and becomes dormant in multiple end organs, a state also referred to as "latency," and can later be reactivated by a number of different stimuli, including immunosuppression and inflammation (reviewed in 1). Approximately 60% of people in the United States have been infected with CMV by the age of 6 years 2, and thus a large population harbor latent virus, making them "at risk" for reactivation. During critical illness—and specifically sepsis—CMV is reactivated in ∼30% of these latently infected individuals, a finding that has now been reproduced independently by 4 different groups 3–6. The molecular basis of this septic stimulus has been evaluated by combining animal models of sepsis and CMV reactivation and shows that sepsis and its inflammatory cascade can trigger CMV reactivation 7, 8. In this issue of the Journal, von Müller et al. 9 have gone beyond simple epidemiological documentation of viral activity and have begun to evaluate the important immunological events that might contribute to, or result from, these reactivation episodes. Reactivation in this population has been previously presumed to be a consequence of the severity of underlying immune compromise from septic disease, the intrinsic viral load, and perhaps the severity of the reactivation stimulus. Critically ill patients with sepsis are known to be immunocompromised by their illness, and although this is not deliberate impairment of the immune system, as seen in immunosuppressed individuals, it has been presumed that this immunocompromise predisposes them to CMV reactivation. Data in this issue suggest that immunocompromise in the formof NK cell dysfunction might set the stage for viral reactivation. NK cell activity is critical to viral control after acute infection as well as to control of reactivation 10–12. All patients with sepsis demonstrated suppression of NK cell responses, and, furthermore, interleukin-2 was unable to restore NK cell function in these patients. Unfortunately, it was not as simple as that. Although all patients had NK cell dysfunction, only ∼30% had reactivated virus. The reasons why all patients did not have reactivated virus might lie with other aforementioned contributors to reactivation, namely underlying viral load or strength of stimulus. Using animal models, my colleagues and I as well as other investigators have noted that underlying viral load is directly proportional to the ability to reactivate virus from latency with a septic stimulus (C.H.C., unpublished data, and 13). In addition, location of the reactivation stimulus (e.g., peritoneum vs. bloodstream) influences the ability to reactivate virus 8. Neither of these parameters was accounted for in von Müller et al.'s present report or in other published data, but undoubtedly these factors contribute to the propensity of virus reactivation in an individual. Thus, the possibility that NK cell dysfunction predisposes to CMV reactivation exists but will require further study. Although these study patients had a demonstrable defect in innate immunity, unlike most clinically immunosuppressed patients, they appeared to maintain T cell function. Data suggested that most patients with CMV reactivation maintained T cell responsiveness to both CMV and staphylococcal enterotoxin B (SEB). Although their T cell responses to SEB might be a limited representation of what is occurring more globally in the host immune system, the fact that functional T cell responses to multiple antigens are intact suggests that T cell impairment is not required for reactivation to occur. This is a significant departure from current thinking. Most current data suggest that T cells play a major role in the maintenance of latency 14. That most of their patients with sepsis were capable of and successful at mounting T cell responses to CMV explains why the viral infections were controlled in these patients but warrants reevaluation of the importance of both NK and T cells in maintaining latency. Because these reactivation episodes are "controlled" by the immune system, one might argue that they are of no clinical consequence. Indeed, as best as can be told, patients with reactivation do not appear to be dying of fulminant CMV disease. Nonetheless, clinical studies published to date have demonstrated surprisingly consistent morbidity in these patients 3–6, 15, 16. Nonimmunosuppressed critically ill patients with CMV reactivation require increased duration of mechanical ventilation, prolonged hospitalization, and may have worsened survival 3–6, 15, 16. Despite this circumstantial evidence, definitive causal data demonstrating CMV reactivation as a pathogen or innocent bystander are lacking. Thus, although CMV is a well-accepted pathogen in immunosuppressed patients, there remains skepticism that CMV is a pathogen in nonimmunosuppressed critically ill patients. There are, however, data to support the argument of pathogenicity. One reservoir of latent CMV is the lungs 17, and this is a consistent site of reactivation 3–6, 15, 16. Recent work in animals has suggested that pulmonary CMV reactivation induced by sepsis causes an exaggerated inflammatory response—that is, both stronger at onset and more prolonged than the inflammatory responses seen in noninfected mice 17. This inflammatory response is substantial enough to cause pulmonary fibrosis in reactivated mice 18. Data presented in this issue of the Journal might help to explain this exaggerated immune response. In the setting of intact T cell function, CMV reactivation might actually be more injurious than in those with T cell impairment. Induction of a prominent antiviral inflammatory response could result in pulmonary injury, and this might explain the prolonged durations of respiratory failure seen in patients with sepsis who experience reactivation. This leaves us with a dilemma: what to do with these patients? There are few data that support treatment of infection/reactivation in nonimmunosuppressed patients, and anecdotal data suggest that, after reactivation has been established, therapy is ineffective at reducing morbidity 4, 6, 16. In addition, data presented in the present issue of the Journal by von Müller et al. suggest that most reactivation episodes are controlled and resolve without therapy. As previously mentioned, current animal data suggest that the occurrence of reactivation—and, more importantly, the immune responses to these reactivation events—are what may cause injury 18. Fortunately, antiviral prophylaxis appears to prevent both reactivation and its consequent pulmonary injury 18. Supporting clinical data from transplant recipients also suggest that prophylaxis strategies are most effective at reducing morbidities associated with CMV infection or reactivation 19. Thus, evidence suggesting pathogenicity should no longer be ignored and, taken together, suggests that the most effective therapy will be prophylaxis in those at risk. Studies of prophylaxis of CMV reactivation in nontransplant patients will need to be performed deliberately and with some caution for at least 2 reasons. First, patients with sepsis are among our most ill and currently available antiviral drugs effective against CMV are not innocuous. Second, this question of efficacy needs to be approached and hopefully answered scientifically. "At-risk" populations will need to be carefully defined and at this time should include patients with sepsis and latent infection. The safest and most effective treatment strategies should be defined using animal models of sepsis and CMV reactivation 7. Furthermore, studies will need to monitor viral load, to determine whether patients with low-level antigenemia require therapy and to help define response to therapy. At this juncture, if antiviral therapy simply becomes the standard of care without proper scientific evidence, we will be faced with the same dilemma that has befallen transplant clinicians treating CMV infection. It has taken nearly 2 decades to confirm the effectiveness of CMV prophylaxis because properly controlled prophylaxis trials were not performed from the beginning. Because monitoring of viral load will play a pivotal role in judging response to therapy in a prophylaxis trial, one final point on monitoring should be made. The work of von Müller et al. in this issue supports previous observations that CMV antigenemia or DNAemia might not be the most sensitive methods for detection of CMV reactivation episodes. In their study, von Müller et al. have defined 3 distinct populations of patients with sepsis and latent CMV at risk for reactivation. One group had undetectable pp65 and no changes in CMV-reactive T cells, which likely represent those without CMV reactivation. A second group had reactivation resulting in detectable pp65 antigenemia, which was then presumably limited by T cell responses. The last group was negative for pp65 antigenemia yet developed CMV-specific T cell responses. This group likely represents patients who had reactivation that escaped detection because of the sensitivity of pp65 antigenemia or that T cell responses controlled before detectable antigenemia occurred. It has been previously observed that the vigor with which the immune system responds to CMV is usually inversely related to the viral load detected 20. Future studies of CMV reactivation should thus include measurement of CMV-specific T cell expansion as described by the authors or other, similar methods that use CMVspecific tetramers to measure T cell responses 21–23. It is intriguing that CMV, a well-known pathogen in immunosuppressed patients, is reactivating in a subset of critically ill patients, who appear to do worse than those without reactivation. Is this coincidence? Likely not. The time has come to acknowledge the "elephant in our living room" 24, p. 1153, but it must be addressed by carefully designed trials with antiviral prophylaxis in patients at risk for reactivation.