Immune control of cytomegalovirus reactivation in stem cell transplantation

Autologous stem cell transplantation (SCT) and allogeneic SCT are undertaken for a broad array of malignant and nonmalignant conditions in patients ranging from infants to >70 years in age. As such, patients come to SCT with vastly different viral exposures and infections that are dictated by age, prior therapy, and geographic and socioeconomic factors. Viral infections can be broadly considered in the context of those that are eradicated before SCT, develop as a primary infection after SCT, or those where active infection has been cleared before SCT but latent infection exists and can result in reactivation in settings of immune deficiency. Because of their capacity to exist in a state of latency, herpesviruses represent most of the viral burden of disease in transplantation.

Human cytomegalovirus (HCMV), a ubiquitous pathogen with a worldwide prevalence ranging from 40% to >90%, will be the focus of this review. Although generally asymptomatic in healthy individuals, without prompt intervention CMV disease after SCT is life threatening, such that patients are monitored weekly for reactivation. Despite strict monitoring and the use of preemptive antiviral therapy, significant CMV viremia levels of 501 to 1000 and >1000 IU/mL in the first 2 months after transplant are associated with hazards ratios (HRs) of 21 and 26 respectively for increased risk of death.¹ CMV reactivation is also associated with myelosuppression (reviewed in Randolph-Habecker et al)² and higher rates of invasive fungal infections,³ although it remains unclear whether this is because of CMV itself or antiviral drug-induced neutropenia. The question of whether CMV confers protection against relapse of acute myeloid leukemia has also been widely debated, although the effects are modest,^4,5 and this topic is outside the scope of this review.

Managing CMV reactivation and disease in patients undergoing SCT carries significant morbidity and economic burden. In addition to the significant costs of antiviral treatments, myelosuppression, cytopenia and renal toxicity lead to complex treatment regimens and longer hospitalization. Studies have estimated that CMV infection increases the cost of posttransplant care by US $58 000 to 74 000.⁶ 

The rates of reactivation for double-stranded DNA viruses are reported as 46% to 65% for CMV, BK virus (BKV) and human herpesvirus (HHV)-6 and 9% to 10% for adenovirus and Epstein-Barr virus (EBV), with reactivation of multiple viruses the rule.⁷ Relative to recipients of HLA-matched grafts, viral reactivation occurs with HRs between 2 and 3 in recipients of HLA-mismatched grafts, cord blood transplant (CBT), T-cell depleted grafts or in the setting of severe acute graft-versus-host disease (GVHD).⁷ The level by which “clinically relevant” CMV reactivation is defined varies between institutions, but is generally between 50 and 150 IU/ml, and the impetus to treat at a particular threshold is usually based on the risk of CMV disease within specific transplant populations (eg, lower thresholds for CBT, mismatched and T-cell–depleted vs matched T-cell-replete transplants, and those not receiving letermovir prophylaxis).⁸ Furthermore, peak, mean and change in viral load in the first 5 weeks have recently been shown to be highly predictive surrogates for the development of CMV disease.⁹ 

The risk of viral infection is affected by the type of transplantation. Autologous SCT typically results in a temporary and relatively modest state of immune deficiency compared with allogeneic SCT (allo-SCT), reflecting the lack of concurrent pharmacological immune suppression and the absence of GVHD, that is itself highly immune suppressive. The type of allo-SCT determines the severity and duration of immune deficiency, with CBT with concurrent anti-thymocyte globulin being the most extreme example. Stem cell grafts associated with T-cell depletion, HLA-mismatch or the presence of GVHD requiring significant steroid administration also impact the degree of immune deficiency after SCT (Table 1). CMV reactivation in the setting of T-cell depletion is dependent on the magnitude of T-cell reduction, which in turn reflects the dose and target of the therapeutic agent used (eg, campath vs anti-thymocyte globulin).^10,11

This review will focus on the immunological factors controlling CMV reactivation after SCT because CMV remains the most common clinical infection affecting transplant outcome¹² and carries an unacceptable burden of disease.¹ Throughout the review, we will highlight what is known in the field and what we consider the most critical questions that require addressing to improve the clinical management of CMV in SCT. These unknowns include:

• •

  the immune parameters critical to control CMV infection (latency, reactivation, dissemination, overt replication, and disease) over time;

• •

  the importance of the origin (recipient vs donor) and localization of anti-viral responses;

• •

  the impact of GVHD and associated clinical treatments (eg, high-dose steroids) on the functionality and relevance of specific immune axes; and

• •

  whether antiviral drugs are sufficient to guarantee enduring immunity.

The transition from latency to productive reactivation is one of the least understood but most important facets of CMV biology and a critical step to target to improve the impact of CMV in transplantation. The first step in CMV reactivation requires latent viral genomes to be transcriptionally activated. The process of molecular reactivation is complex and involves regulation of the major immediate-early enhancer promoter (MIEP).¹³ This region is silenced during latency; derepression of MIEP leads to epigenetic changes and increased chromatin access that enable the transcription of critical viral genes, viral replication, and ultimately the production of infectious virus. Recently, it has become apparent that alternative promoter elements within loci other than the MIEP, including the MIE locus, can initiate viral gene expression and may be relevant to reactivation of CMV in CD34⁺ hematopoietic stem cells (HSCs).¹⁴ 

CMV infects both stromal (eg, fibroblast, endothelial, and epithelial) and hematopoietic (eg, dendritic cells [DCs] and macrophages) cell types in vitro, and viral genomes (ie, latent virus) have been detected in CD34⁺ HSCs and CD14⁺ lineage–committed granulocyte-macrophage progenitors. CMV reactivation is influenced by the differentiation state of cells as well as the immune environment, but the factors driving the initiation of viral reactivation remain largely unknown, and inflammation may be important. During transplantation and GVHD, engagement of TLRs by DAMPS results in overt inflammation characterized by the dysregulation of several cytokines including interferons (IFNs), tumor necrosis factor (TNF), and interleukin-6 (IL-6).^15,16 Once CMV has reactivated, several additional host responses are deployed to contain the infection. Type I IFNs potently inhibit viral replication in tissues¹⁷ and can enhance cross-presentation of viral antigens by antigen-presenting cells (APCs).¹⁸ Although the antiviral effects of cytokines have been studied predominantly in vitro, blockade of the IL-6 receptor with tocilizumab has been recently associated with reduced CMV reactivation in recipients of unrelated volunteer donor grafts.¹⁹ Further studies are necessary to ascertain whether this finding is related to direct effects on viral reactivation and dissemination, or reflects effects of IL-6 on virus-specific immunity.

Although primary infection originates from latently infected hematopoietic cells in the donor seropositive/recipient seronegative (D⁺/R⁻) setting, the frequency of hematopoietic progenitors carrying latent viral genomes is extremely low (0.001% to 0.01%)²⁰ and the risk of primary infection in CMV-naive recipients receiving grafts from CMV⁺ donors is low. CMV reactivation typically occurs 4 to 8 weeks after SCT, even when seronegative grafts are transplanted. Most (94% to 100%) myeloid cells are of donor origin by day 28, even after nonmyeloablative conditioning, at least in the periphery.²¹ Although this does not exclude the potential for reactivation to occur from recipient tissue macrophages that are eradicated more slowly after SCT,²² these chimerism kinetics raise the likelihood that reactivation may originate from stromal/nonhematopoietic cells. The importance of stromal cells as a latent viral reservoir is supported by data in mouse models where murine CMV (MCMV) establishes latency in stromal cells of various tissues (reviewed in Reddehase et al²³). Lack of understanding about the origin of reactivating virus, both in relation to cell lineage and target tissue, represents a major limitation to the development of logical therapeutic approaches to prevent reactivation. Addressing and resolving these questions will require sophisticated model systems to track latent and reactivating virus in vivo after transplantation.

Natural killer (NK) cells reconstitute within 1 month after autologous SCT (ASCT), after neutrophil recovery.²⁴ Transplanted virus-specific mature lymphocytes expand quickly in the profound lymphopenia that accompanies SCT and retain memory to antigens to which they were previously exposed. T and B cells reach normal values within 3 to 6 months after ASCT, whereas full functional reconstitution typically takes up to 1 year.²⁵ In contrast to CD8⁺ T cells, CD4⁺ T cells reconstitute more slowly, especially the CD45RA⁺ naive subset, leading to inverted CD4/CD8 ratios for an extended period of time.^24,26 CMV reactivation is rare after ASCT, even after tandem procedures.^27,28 In mouse systems, the loss of both CMV-specific T-cell and humoral immunity is essential for reactivation (ie, CMV-specific T cells or IgG are protective in isolation).²⁹ The fact that at-risk seropositive autografted patients receive grafts containing CMV-specific T cells in the absence of GVHD and GVHD prophylaxis may explain why they are at low risk of reactivation, even in the setting of highly compromised humoral immunity that is characteristic of myeloma treatment in particular.

Immune reconstitution after allo-SCT is complex and compromised by pharmacological immunosuppression and GVHD.^30-32 NK cells reconstitute rapidly after allo-SCT.³³ Delayed immune reconstitution and GVHD are major risk factors for CMV reactivation and dissemination,³⁴ and GVHD prophylaxis significantly impairs the functional recovery of CMV-specific CD4⁺ and CD8⁺ T cells.^35,36 Reconstitution of CD8⁺ T cells is faster than that of CD4⁺ T cells or B cells and can reach normal values as early as 100 days after allo-SCT (reviewed in Ogonek et al);³⁷ thymic involution in aging recipients and transplant-related damage to thymopoiesis after allo-SCT can delay the reconstitution of CD4⁺ T cells by years.³⁸ Thus, CMV reactivation is likely to be controlled by transplanted memory T cells in the short term and stem cell–derived, thymically educated cells in the long term.

Diverse B-cell reconstitution and antibody secretion require help from CD4⁺ T cells. Thus, the combined B-cell and CD4⁺ T-cell deficiency that occurs in allo-SCT results in profound defects in humoral immunity.^31,32,39 IgM reaches normal values 3 to 6 months after allo-SCT, followed by IgG, whereas IgA normalization can take up to 5 years. The risk of acquiring pneumococcal infection remains 30-fold higher compared with the general population up to 10 years after SCT,^39,40 consistent with long-term functional defects in humoral immunity,⁴¹ especially in patients with a history of chronic GVHD.⁴² Thus, delayed recovery of donor immunity and loss of recipient immunoglobulins confer a high risk of opportunistic infections in this patient population.

Calcineurin inhibitor (CNI)-based immune suppression remains the backbone of GVHD prophylaxis protocols. The accompanying broad suppression of T cells by CNI impairs pathogen-specific immunity. T-cell depletion is frequently incorporated in conditioning regimens and is associated with immune deficiency that predisposes to viral reactivation.^10,36 High-dose steroids used to treat GVHD also profoundly suppress immune recovery after allo-SCT.³⁵ 

Alternative approaches, such as the use of rapamycin or posttransplant cyclophosphamide (PTCy), also modify the risk of CMV reactivation. Unexpectedly, rapamycin is associated with a dose-dependent reduction in CMV reactivation.^21,43,44 This finding can be reconciled by the fact that in nontransplant settings rapamycin has been shown to promote the differentiation of antiviral memory CD8⁺ T cells⁴⁵ and modulate immunoglobulin class switching, resulting in heterosubtypic cross-protective humoral responses to influenza.⁴⁶ PTCy regulates GVHD via the depletion of alloreactive T cells, while sparing the regulatory T-cell compartment and enhancing B-cell reconstitution^42,47 and has the potential to indirectly preserve antiviral immunity by reducing the incidence of chronic GVHD and associated immune suppression. Preliminary investigations in SCT with PTCy-based immune suppression reported higher CMV viremia before day 70 posttransplant, but lower levels thereafter,⁴⁸ likely reflecting impaired early antiviral immunity but improved antiviral responses subsequently in the absence of chronic GVHD. Intriguingly, after sibling allo-SCT, PTCy is associated with increased CMV reactivation compared with calcineurin-based immune suppression, and reactivation in PTCy patients is also associated with higher levels of chronic GVHD.⁴⁹ 

The impact of GVHD on the various facets of antiviral immunity and CMV reactivation will be discussed in detail in the sections that follow.

The concerted activities of multiple immune effectors ensure control of CMV infection and reactivation in immunocompetent hosts (Figure 1). Approximately 50% of the population in the United States is CMV seropositive, and this number is substantially higher in developing countries.⁵⁰ Transplantation of a seropositive recipient with a seronegative donor is a major risk factor for CMV reactivation,⁵¹ reflecting the importance of donor-derived antiviral immunity. Consistent with this, transplantation with donor-seropositive grafts is associated with improved recovery of CMV-specific T cells and lower risk of CMV reactivation and disease.⁵² 

The array of immune responses elicited by CMV infection/reactivation, how they reconstitute after SCT and their importance in controlling CMV are outlined below and in Figure 2.

Early after SCT, recipient and donor-derived APCs (eg, DCs, monocyte-macrophages, and B cells) present alloantigens to donor T cells to initiate and amplify GVHD, respectively.⁵³ Donor APCs present exogenously derived viral peptides to CD4⁺ T cells within major histocompatibility complex (MHC) class II and “cross-present” exogenous viral antigens to CD8⁺ T cells within MHC class I.⁵⁴ In mouse models, donor DCs are critical for the generation of cell-mediated immunity and virus control in a primary CMV infection after SCT.⁵⁵ In this setting, DCs are infected and in the absence of GVHD are capable of direct antigen presentation.⁵⁶ In contrast, the APCs and pathways of antigen presentation required to generate immunity during reactivation are unclear. Of note, in preclinical mouse models of both primary CMV infection and reactivation, GVHD and the associated cytokine dysregulation impair the development of DCs and antigen presentation. Both MHC class I and II processes are corrupted, resulting in defective presentation of endogenous and exogenous antigens,^55,57 which is particularly problematic for the effective priming of CMV-specific naive T cells after SCT (eg, seronegative donors).⁵⁵ In addition, defects in MHC class II presentation during acute GVHD result in defective homeostasis of regulatory T cells and are associated with chronic GVHD in murine systems,⁵⁸ which may further impair pathogen-specific immunity.

NK cells recognize and eliminate virally infected cells with high efficiency, and CMV infection has been associated with a series of phenotypic and functional changes in NK cells, both in healthy donors^59,60 and after SCT.⁶¹ The concept of NK cell memory has stemmed from studies of CMV infection in the mouse that demonstrated that a population of NK cells expressing the Ly49H activating receptor (which recognizes the viral protein m157) can “remember” a previous encounter with CMV and mount a more robust secondary response.^62,63 Analogous to findings in the mouse, NK cells expressing the activating receptor NKG2C expand in acute infection and in some patients in association with CMV reactivation during SCT.^64,65 Interestingly, these effects of CMV reactivation on NK cell maturation are predominantly seen in patients who receive stem cells from bone marrow rather than peripheral blood.⁶⁶ 

Recent studies have reported that conditioning has a significant impact on both the reconstitution kinetics and the phenotype of the NK cell repertoire.⁶⁷ Furthermore, GVHD dramatically impairs NK cell expansion in both preclinical systems⁶⁸ and in patients⁶⁹ because of IL-15 sequestration by alloreactive T cells. As a result, both NK-cell–dependent leukemia and CMV-specific immunity are impaired.⁶⁸ IL-15 mimetics have shown promise in the treatment of John Cunningham virus reactivation and progressive multifocal leukoencephalopathy after clinical allo-SCT, in association with NK cell and memory CD8⁺ T-cell expansion.⁷⁰ Therapy with IL-15 receptor agonists may prove relevant for controlling other relapsed and refractory viral reactivations and deserves further study. Interestingly, NK-cell recovery, especially after CBT, has been associated with improved CD4⁺ T-cell reconstitution,⁷¹ suggesting that improving innate responses may assist with other components of CMV immunity.

Within mucosal tissue, type 1 resident innate lymphoid cells (ILC1) play a sentinel role in pathogen responses. In mouse models of CMV infection ILC1 have been shown to assist in controlling early viral replication via IFNγ produced in response to IL-12 secreted by XCR1⁺ DCs that sense CMV.⁷² The role of ILC1 in CMV reactivation after clinical SCT is unknown, but tissue resident lymphocytes, including memory populations of innate cells and T cells, are likely relevant and require further investigation.

Our current understanding of the control of CMV replication and resolution of disease arising from viral reactivation suggests that these processes are dependent on T-cell activities. Recovery from CMV disease after clinical SCT correlates with reconstitution of functional donor CD8⁺ T cells,^35,73-78 providing associative evidence for a crucial role of these cells. Attempts have been made to quantify the number of CMV-specific T cells after transplant to identify patients at high risk of initial or recurrent reactivation. Low numbers of cytokine-secreting CD8⁺ T cells (measured by IFNγ QuantiFERON assays of responses to CMV peptide pools) a week after reactivation associate with complicated (recurrent) reactivation.⁷⁹ Likewise, ELISpot and flow cytometry–based cytokine secretion assays have demonstrated an association between low cell-mediated CMV immunity and clinically significant reactivation in patients.^80,81 Recently, a “protective signature” for clinical CMV reactivation has been defined based on polyfunctional cytokine-secreting CD8⁺ T cells (IL-2, IFNγ, TNF, and inflammatory protein-1β).⁷⁸ MHC class I and II tetramer-based quantification of CMV immunity similarly generated thresholds for defining patients at high risk of recurrent reactivation.^34,82 Although no T-cell assay has entered routine diagnostic clinical practice to date, recent studies confirm the predictive value of commercial ELISpot-based assays, and they may be used more widely in the future to predict patients at high risk of recurrent reactivation.⁸³ 

Further support for the central role of T cells in the control of CMV reactivation comes from the clinical adoptive transfer studies of CMV-specific CD8⁺ cytotoxic T cells expanded in vitro. These cells can generate virus-specific immunity and prevent recurrent CMV infection and disease in patients.^77,84-86 In clinical SCT these approaches are limited by (1) the labor-intensive and lengthy processes required for T-cell manufacturing, (2) the incomplete persistence of functional T cells after transfer, (3) the requirement for a CMV⁺ donor, and (4) the lack of efficacy data for adoptively transferred T cells in the presence of high-dose steroids, as is common in allo-SCT. More recently, isolating antiviral T cells directly from donor blood in the clinic, without ex vivo expansion, on the basis of IFN-γ production,⁸⁵ or using tetramers/streptamers has been trialed.^87,88 Although these methods offer promise, they rely on the transfer of a pool of cells, with only a fraction likely to be capable of improving viral control and often excluding important protective subsets (eg, CD4⁺ T cells). As an alternative to donor-derived CMV-specific T cells, third-party, multivirus-responsive T cells that can be manufactured in a large scale have the potential to provide protection in a timely manner; preliminary clinical studies have confirmed safety and efficacy for this approach^89,90 and definitive placebo-controlled clinical trials are in progress (registered at www.clinicaltrials.gov as NCT04693637).

In addition to the clear importance of CD8⁺ T cells, a role for CD4⁺ T cells is supported by the fact that reconstitution of HCMV-specific CD4⁺ T cells improves viral control.^76,91-93 HCMV-specific CD4⁺ T-cell reconstitution parallels, albeit at a lower level, CD8⁺ T-cell reconstitution, and the magnitude of reconstitution is associated with long-term protection.^34,77 In addition, the clinical efficacy of adoptively transferred CMV-specific T cells is associated with baseline CD4⁺ T-cell reconstitution.⁹⁴ Future studies are needed to investigate the role of CD4⁺ T cells, both as direct antiviral effectors and as important contributors to the efficacy and longevity of CD8⁺ T-cell responses and possibly to antibody-mediated immunity.

CMV reactivation profoundly affects CD8⁺ T-cell reconstitution, expanding CD8⁺CD57⁺ effector memory T cells, with an associated contraction of naive T cells, including recent thymic emigrants. In addition, the diversity of the T-cell receptor repertoire is significantly reduced.⁹⁵ CD4⁺CD57⁺ T cells are also expanded in CMV seropositive donors and further expand in recipients after CMV reactivation.⁹⁶ These CD4⁺ T cells have a cytolytic (granzyme B⁺) Th1 phenotype and are associated with reductions in MHC class II–expressing APCs. Thus, CMV has a dramatic impact on T-cell differentiation and immune competence, suggesting that impaired responses to opportunistic infections other than CMV may explain the increased mortality associated with CMV reactivation.^97,98

The role of antibodies and the B/plasma cells from which they derive in the control of CMV infection has been controversial. Immunocompetent individuals with strong CMV-specific antibody (and T cell) responses are commonly reinfected with multiple CMV strains. Attempts to prevent clinical CMV infection, reactivation or ameliorate CMV disease with polyvalent immunoglobulins have provided limited or ambiguous evidence of efficacy.^12,99-102 A recent meta-analysis of immunoglobulin prophylaxis in patients noted reductions in CMV disease (HR, 0.52), but no differences in CMV infection.¹⁰² Thus, to date there are inconsistent data regarding the role of antibodies in limiting CMV reactivation in clinical bone marrow transplants. In this context, mouse models, the key features of which are summarized in Table 2, have proven valuable, with several findings translated to the clinic.^103,104

In the MCMV mouse model, adoptive transfer of donor CMV-specific memory B cells has been shown to protect immunodeficient hosts from CMV infection,¹⁰⁵ but the potential of this approach after clinical allo-SCT has not been investigated. Recently, the importance of antibodies in limiting CMV reactivation has been demonstrated in novel bona fide mouse models of CMV reactivation after allo-SCT. Although CMV protective immunity in SCT is generally thought to be largely T-cell dependent, CMV reactivation and disease occured only when both T-cell and humoral immune responses were compromised. This finding was unexpected given the limited efficacy of immunoglobulins in the treatment of CMV reactivation in clinical SCT recipients, but is reconciled by the fact that previous approaches largely ignored the genetic variability between HCMV strains. Consistent with the importance of CMV genetic diversity, in these mouse systems, protection from CMV reactivation is optimal when antibodies are matched to a specific CMV strain.²⁹ The characteristics of these protective antibodies should now be examined, and their mode of action defined. In murine systems, antibodies that inhibit cell-to-cell spread of virus appear critical²⁹; however, this possibility has yet to be examined in clinical settings.

An important consideration in the clinical setting of SCT is ensuring the longevity of antibodies. Long-lived and substantial defects in humoral immunity are well documented in patients who undergo SCT and include the loss of recipient-derived pathogen-specific immunoglobulins associated with GVHD.⁴² Similarly, the half-life of IV immunoglobulins is considerably shortened in patients after SCT (from 22 to 6 days), especially in the presence of GVHD.^106,107 Analyses in mouse models of CMV reactivation after allo-SCT have highlighted 2 important requirements for protection: (1) matching antibodies to the infecting strain of virus is critical, thus antiviral antibodies of recipient origin are most effective in preventing viral reactivation and dissemination when the recipient is CMV seropositive, as they are inherently matched to the reactivating viral strains, and (2) the longevity of preexisting recipient-derived antibodies and passively transferred immunoglobulins is a critical determinant of protection.²⁹ Thus, a key consideration in clinical SCT may be to maintain and possibly enhance the durability of antiviral immunoglobulins. Studies to define the relevant mechanisms are possible using recently developed mouse models.²⁹ 

Although the importance of T-cell control of CMV has been established, it is likely that different immunological pathways play distinct roles early (eg, in the first 3 months) vs late (beyond 3 months) after transplant. In this regard, any protection by persisting strain-specific recipient antibodies may be important in the first 1 to 2 months after transplant, particularly when the donor is seronegative. Likewise, NK cells reconstitute quickly after transplant³³ and are possibly important in the early transplant period. As professional donor APCs reconstitute after engraftment, the priming and expansion of donor T cells is likely to occur in the second and third months after transplant,³⁷ during the characteristic period of viral reactivation. Thereafter, protection is most likely afforded by ongoing T-cell control, consistent with the fact that the risk of recurrent disease is highly linked to persistent defects in T-cell immunity after reactivation, as outlined herein.^79,80,82,83 Donor B-cell reconstitution³⁷ and memory differentiation is very slow after allogeneic SCT, particularly in the setting of GVHD;³² therefore, any protection by donor-derived antibodies would be likely to occur late after transplant (ie, >6 months). Recurrent and clinically problematic CMV reactivation typically occurs late after transplant, in the setting of GVHD and ongoing pharmacological suppression. GVHD has profound effects on antigen presentation,⁵⁷ T-cell,⁵⁵ B-cell,³² and NK-cell⁶⁸ differentiation, and so this scenario effectively creates a perfect storm, whereby all arms of CMV immunity are compromised. A schematic overview of this timeline is shown in Figure 2.

CMV infection is particularly problematic because of the unique features that characterize this virus, including: (1) the propensity to infect and establish latency in multiple cell types of both hematopoietic and stromal origin, (2) the efficiency with which the virus can interfere with antiviral immune responses, (3) the existence of numerous viral strains, and (4) the capacity to mutate when placed under selective pressure.

These properties of CMV have made it extremely difficult to generate effective vaccines and/or therapies, despite enormous efforts.

The most critical insights into how to improve on the status quo come from understanding how immunocompetent hosts deal with CMV infection over its protracted course and highlight the need for combination strategies that mobilize multiple facets of the immune response.

Antiviral therapies have revolutionized the control of CMV infection in SCT and will continue to provide an important line of defense.¹⁰⁸ Primary prophylaxis with letermovir has been very successful; however, extended use and cost, the emergence of viral resistance, and late-onset viral disease still pose significant challenges. In addition, preliminary evidence suggests letermovir prophylaxis is associated with delayed CMV-specific cellular reconstitution, possibly caused by reduced antigen exposure.⁸¹ Similar predicaments may be faced with newer drugs, such as maribavir.

The use of vaccines in SCT is hindered by the fact that responses to vaccination are dependent on the functional reconstitution of CD4⁺ T cells and B cells, which is significantly reduced for 2 to 3 years after SCT when compared with healthy populations of comparable age.¹⁰⁹ Furthermore, CMV possesses multiple mechanisms for evading host immune responses, making the development of CMV vaccines difficult. Despite early promise in a phase 2 clinical trial,¹¹⁰ a bivalent DNA vaccine failed to meet end points or to demonstrate efficacy in a randomized placebo-controlled phase 3 study (www.clinicaltrials.gov, NCT01877655). Renewed promise in a vaccine approach has come from clinical studies of a chimeric peptide vaccine targeting the HLA-A201–restricted pp65 CD8 T-cell epitope (PepVax)¹¹¹ and a recombinant attenuated poxvirus expressing 3 immunodominant CMV antigens (Triplex; pp65, IE1, and IE2). In a randomized phase 2 trial, administration of PepVax to seropositive recipients at days 28 and 56 after SCT was associated with reduced rates of CMV events and higher rates of CMV-specific T cells early after SCT.¹¹² Extending these studies to include donor vaccination strategies is likely to be important in the future.

Virus-specific T cells are clearly important in limiting CMV disease, and third-party T cells have eliminated time constraints and HLA-matching restrictions; however, their use remains limited by concurrent GVHD and associated immunosuppressive therapy (eg, steroids).

The use of biologics, in particular antibodies that limit infection/reactivation and dissemination, is likely to represent a viable, nontoxic addition to pharmacological intervention and/or T-cell therapy to guarantee enduring immunity and thus meet the challenges posed by late-onset disease. Several antibody-based approaches are under investigation that principally target the glycoprotein gB and molecular cell entry pathways mediated by the gH/gL/pUL128-pUL130-pUL131A pentameric complex required for viral entry into endothelial and epithelial cells.¹¹³ The concept of antibodies that target specific cell lineages is crucial for the development of vaccines and therapeutic application of CMV immunoglobulins. The mode of action of antibodies is also critical. Neutralizing antibodies are effective against cell-free virus and thus are critical in preventing primary infection. The spread of virus, however, relies largely on cell-associated virus, and thus antibodies that block cell-to-cell spread will most effectively prevent viral dissemination after reactivation. Consistent with this paradigm, inhibition of viral cell-to-cell spread predicts the protective capacity of antibodies in vivo in preclinical models of CMV reactivation after SCT²⁹ and appears relevant for HCMV infection.¹¹⁴ Development of approaches that permit the testing of relevant antibodies in vivo in appropriate SCT models of viral reactivation will be integral to moving the field forward by defining the efficacy of specific antibodies and their mechanism of action.

The capacity to retain (recipient), establish, and maintain (donor) protective antiviral responses, both humoral and cell mediated, that are effective in the context of transplant-related immune suppression (pharmacological intervention and GVHD) and viral escape strategies (strain diversity, immune evasion, and viral resistance) will be necessary to effectively control CMV infection in SCT. Concurrently, transplant approaches that limit T-cell depletion, prevent GVHD, and promote reconstitution of immune memory will be important for the control of CMV in the long term, and new approaches such as naive T-cell depletion (that spare memory T-cell responses) may hold promise in this regard.¹¹⁵ Gut microbiota diversity and antibiotic exposure correlate with GVHD; however, current data are inconsistent with regard to CMV reactivation,^116,117 and interrogation of larger data sets is required.

Given that our understanding of the immunological controls of CMV reactivation remains largely limited to T cells, delineating the additional pathways, both humoral and innate, and their relative contribution requires urgent attention. This information should allow for integration with transplant type, prior donor and recipient viral exposure, and planned antiviral prophylaxis to generate logical risk profiles that guide immunological monitoring and additional therapeutic requirements. Finally, any effects of non–T-cell immunity to CMV need to be understood in total and in the context of potential downstream effects on critical anti-viral T-cell responses.

The authors thank Christopher E. Andoniou and Ping Zhang for contributions in planning the review and critical reading of the manuscript.

The figures were generated by Xavier Y. X. Sng with guidance from C. E. Andoniou and P. Zhang.

Contribution: M.A.D.-E. and G.R.H. wrote, reviewed, and approved the final manuscript.

Conflict-of-interest disclosure: G.R.H. has consulted for Generon Corporation, NapaJen Pharma, and Neoleukin Therapeutics and has received research funding from Roche, Compass Therapeutics, Syndax Pharmaceuticals, Applied Molecular Transport, and iTeos Therapeutics. M.A.D.-E. and G.R.H. are inventors on patent applications describing methods of preventing CMV reactivation.

Correspondence: Mariapia A. Degli-Esposti, Monash Biomedicine Discovery Institute, Monash University, 15 Innovation Walk, Clayton, VIC 3800 Australia; e-mail: mariapia.degli-esposti@monash.edu; and Geoffrey R. Hill, Fred Hutchinson Cancer Research Center, 1100 Fairview Ave N, Seattle, WA 98109; e-mail: grhill@fredhutch.org.