CMV: when bad viruses turn good

Cytomegalovirus (CMV) has had a reputation for causing morbidity and mortality after allogeneic stem cell transplantation (SCT). In this issue of Blood, Elmaagacli et al find an unexpected favorable association of a low rate of leukemic relapse in acute myeloid leukemia patients who reactivate CMV in the first few weeks of SCT.¹ 

Every now and then retrospective analyses of SCT data bring up unexpected and counterintuitive findings. This is the case in the article by Elmaagacli and colleagues from the SCT group in Essen, Germany. Analyzing 266 consecutive patients with acute myeloblastic leukemia (AML) who received SCT from HLA-identical relatives or unrelated donors between 1997 and 2009, they found an unusual association between early CMV reactivation and transplantation outcome. Seventy-seven patients developing their first CMV pp65 antigenemia at a median of 6 weeks after transplantation were found to have a remarkably low risk of leukemic relapse (9% at 10 years after SCT) compared with a 42% risk in 189 patients not reactivating CMV. Furthermore, they found that, far from being a risk for increased transplantation-related mortality, the occurrence of CMV reactivation was not deleterious for survival. In support of a specific effect of CMV they found that positive CMV serology in donor or patient was itself protective against relapse while reactivation of other viruses had no impact.

Historically, CMV disease has been a major complication of allogeneic SCT.²  In the days before high-sensitivity monitoring of CMV antigenemia by PCR for pp65 protein and preemptive treatment of CMV disease with ganciclovir or foscarnet, CMV pneumonitis accounted for up to 20% mortality after transplantation for leukemia.³  Even today, when death from CMV pneumonitis is rare, we regard viral reactivation as bad, leading to CMV disease if not controlled by antivirals, which in their turn cause cytopenia and renal damage. CMV reactivation implies immunodeficiency, loss of control of a resident DNA virus, and breakdown of immunosurveillance against residual leukemia. The detection of the virus early after transplantation might be expected to be associated with an increased risk of relapse. Indeed, an earlier study from the National Institutes of Health demonstrated that persisting pp65 antigenemia in the first 3 months after SCT was associated with defective T-cell replication against CMV peptides, an increased risk of leukemic relapse from the 3-month landmark, and a higher transplantation-related mortality.⁴ 

The findings from Elmaagacli et al therefore fly in the face of established perceptions. Unexpected findings merit special scrutiny if they are to be validated, and in this article the authors have gone to extensive lengths to support their conclusions. An obvious confounding factor is that CMV reactivation is closely linked to the occurrence of acute graft-versus-host disease (GVHD), which in turn implies a graft-versus-leukemia effect; indeed, it has been suggested that CMV reactivation is a trigger for GVHD development.⁵  In this series, grade II-IV acute GVHD doubled the risk of CMV reactivation and preceded pp65 antigenemia in 90% of cases. However, within the group of 187 individuals with grade II-IV acute GVHD, CMV reactivation still had an independent impact on relapse: in 77 reactivators the relapse rate was only 9% compared with a 38% relapse rate in 189 nonreactivators. The same benefit for CMV reactivation held true for chronic GVHD, considered to be an important long-term control of residual leukemia. In careful multivariate analysis CMV reactivation remained an independent variable alongside established risk factors for relapse and survival such as GVHD and leukemia risk group.

Elmaagacli and colleagues propose several explanations for these results (see figure). First, myeloid cells are a reservoir for CMV. It is possible that viral reactivation in AML blasts makes them a target for subsequent attack by virus-specific cytotoxic T cells (CTLs). Alternatively, CMV reactivation in AML cells may up-regulate leukocyte fixation antigen-3 (LFA-3) expression and enhance natural killer (NK)–cell cytotoxicity against the leukemia.⁶  Last, the virus could have a direct cytotoxic effect on the AML cell: a virus-versus-leukemia effect. Of these possibilities the idea that CMV renders the leukemia a target for immune attack seems the more likely. CMV reactivation is a trigger for expansion of potent CTLs targeting the virus,⁷  and while antivirals are routinely used to suppress viremia it is clear from the observation that antigenemia falls as CMV-specific T-cell frequencies rise and the effectiveness of virus-specific T-cell therapy that the immune response is central to controlling the virus. There is some evidence that AML is especially susceptible to allogeneic effects of NK cells after SCT, and an NK-targeting of CMV-infected AML blasts is also conceivable. These hypotheses are testable: we can explore the relationship between the rise of CMV-reactive CTL and AML relapse, which should correlate even better with relapse control than pp65 antigenemia. In vitro studies of the interaction of NK cells and CMV-specific CTLs with AML cells infected with CMV are indicated. First, however, this article challenges us to review our own databases to determine whether they validate these remarkable and significant findings.