Characterizing EBV-associated lymphoproliferative diseases and the role of myeloid-derived suppressor cells

Epstein-Barr virus (EBV) colonizes the B-cell system, initially as a growth-transforming infection with expression of the full set of latent cycle genes, followed by transition to an antigenically silent latency within memory B cells. Primary infection usually occurs during childhood and is almost always asymptomatic. Delay in primary infection until young adulthood can result in infectious mononucleosis (IM), an acute but self-resolving illness whose febrile symptoms appear to be a product of the host’s exaggerated cellular immune response. That response is initiated by natural killer cells (NK), followed by robust CD4⁺ and particularly CD8⁺ T-cell responses to lytic and latent cycle viral antigens.^1,2  However, when T-cell function is impaired, for example, transplant recipients on immunosuppressive drugs, failure to control latent growth-transforming infection can lead to posttransplant B lymphoproliferative disease (LPD).³ 

By contrast, primary infection in a rare minority of individuals leads to a chronic, nonresolving condition termed chronic active EBV (CAEBV) encompassing many of the clinical features of IM, such as fever, splenomegaly, hepatitis, and lymphadenopathy. In CAEBV, however, these symptoms are severe and persist for at least 3 months accompanied by very high EBV DNA load in the blood and extensive organ infiltration by EBV-infected cells. CAEBV can develop into a spectrum of clinical conditions. Patients with indolent disease can remain stable over many years, whereas others display a more rapidly progressive disease with serious complications, including hemophagocytic lymphohistiocytosis (HLH), hepatic failure, and gastrointestinal ulceration/perforation.

In Western countries, CAEBV is thought to be predominantly associated with EBV infection of B cells. Previously reported cases of B-cell CAEBV identified defective cytotoxic T lymphocytes (CTLs)⁴  or NK cells,⁵  in some cases linked to inactivating mutations in the perforin⁶  or GATA2 gene⁷ ; such impairment could explain the chronic nature of the disease. In contrast, CAEBV in South East Asia, Central, and South America is predominantly associated with EBV infection of T cells and NK cells. In these patients, the clinical features, natural history, and prognosis are associated with the phenotypic identity of the infected lineage; patients with T-cell CAEBV exhibit more aggressive disease and significantly inferior survival, whereas patients with NK-cell CAEBV more often have milder symptoms and a more favorable clinical outcome.^8,9  However, NK-cell CAEBV seems to have a greater propensity (23.1%) for transformation into aggressive NK-cell leukemia or extranodal NK/T-cell lymphoma.^8-10  Thus, the morbidity and mortality conferred by this clinically heterogeneous spectrum of EBV-associated lymphoproliferations are substantial. The high mortality is likely attributable to the paucity of effective treatment options. Immunomodulatory agents are largely ineffective and, although corticosteroids and other immunosuppressive agents temporarily ameliorate clinical symptoms, patients usually succumb to progressive disease. Currently, allogeneic stem cell transplantation is the only established curative treatment.^11-13 

The pathogenesis of T/NK CAEBV is poorly understood. In contrast to B-cell CAEBV, patients with T/NK-cell CAEBV are not thought to have a preexisting immune defect,¹⁰  yet infected cells persist in blood and tissues long term. These EBV-infected T/NK cells exhibit a latency I/II viral gene expression profile (noncoding EBV-encoded RNAs [EBERs], EBNA1 ± LMP1, and LMP2a/TR)¹⁴  and should therefore elicit a T-cell response. This suggests that T-cell responses are absent or in some way functionally inhibited. This hypothesis is supported by clinical experience in this context, whereby infusions of third-party EBV-specific CTLs have proven ineffective against progressive disease,¹⁵  suggesting an inhibitory mechanism is at play.

To address the many issues raised by the different presentations of CAEBV and EBV-associated LPDs, we developed a protocol, applicable to blood cell preparations ex vivo, to inform patient management while helping to understand disease pathogenesis. Here, we describe a protocol that combines EBER in situ hybridization (ISH), allowing visualization and enumeration of EBV-infected cells, with cell surface and intracellular staining for a whole range of phenotypic and functional markers relevant to B, T, or NK cells and their subsets. This protocol was applied to 5 UK patients, all initially diagnosed as CAEBV or as HLH, which were found to be EBV related. Two proved to have B-cell infection; 3 had monoclonal T-cell or NK-cell infections, whose progression could be followed despite attempts at therapy. Surprisingly, this same study revealed another unexpected but consistent feature of blood samples from patients with T/NK CAEBV, the presence of large numbers of circulating myeloid-derived suppressor cells (MDSCs) with the ability to suppress T-cell responses in in vitro assays. This points to a potential link between MDSC-mediated T-cell inhibition and the chronic progressive nature of CAEBV.

This research received institutional approval from the University of Birmingham, and studies of clinical material received ethical approval from the South Birmingham NHS Research Ethics Service Committee (07/H1208/62). All patients and healthy volunteers provided written informed consent in accordance with the Declaration of Helsinki.

Peripheral blood mononuclear cells (PBMCs; 5 × 10⁶) were stained using Fixable Viability Dye eFluor-450 or eFluor-780 (Thermo Fisher Scientific) for 30 minutes at 4°C. Cell surface staining was then performed with fluorophore-conjugated antibodies (Table 1) for 30 minutes at 4°C. MDSCs were preincubated with purified human immunoglobulin G (5 µg/mL) for 15 minutes at 4°C prior to addition of monoclonal antibodies. For intracellular staining, cells were fixed and permeabilized using the TrueNuclear Transcription Factor Buffer Set (Biolegend) according to the manufacturer’s instructions and then labeled with antibodies (Table 1) for 30 minutes at room temperature. Fluorescence data were acquired using an LSR II or LSRFortessa X-20 (Becton Dickinson) and analyzed in FlowJo software version 7.6.5 (Tree Star).

FlowRNA experiments were performed on 1 to 3 × 10⁶ PBMC (PrimeFlow RNA, Thermo Fisher) according to the manufacturer instructions. A full description of the protocol can be found in the supplemental Data, available on the Blood Web site. Briefly, cells were stained using Fixable Viability Dye eFluor-450 and then for cell surface markers (Table 1). Cells were fixed, permeabilized, and intracellular staining performed in the presence of RNase inhibitors. Cells were fixed again, and then target RNA was hybridized using Target Probe sets (custom designed by PrimeFlow RNA, Thermo Fisher Scientific) for EBER-1/2 and β2-microglobulin (B2M). Briefly, the probes were hybridized at 40°C for 3 hours, washed, and then subjected to sequential hybridization at 40°C using the “PreAmplifier,” “Amplifier,” and fluorescent label (type 1 probes, AlexaFluor-647; type 2 probes, AlexaFluor-488). Cells were acquired using an LSRII or LSRFortessa X-20 and analyzed in FlowJo software, version 10.

Plasma was collected following density centrifugation (Lymphoprep; Stemcell Technologies) of fresh EDTA blood samples. Plasma was immediately snap-frozen in liquid nitrogen and stored at −80°C until use. Cytokines in plasma (interleukin-1β [IL-1β], IL-4, IL-6, IL-10, IL-17A, interferon-γ [IFN-γ], granulocyte-macrophage colony-stimulating factor [GM-CSF], tumor necrosis factor-α [TNF-α]) were measured using a custom ProcartaPlex bead-based immunoassay (Thermo Fisher Scientific) according to manufacturer’s instructions. Arginase-1 in plasma was measured by ProcartaPlex Singleplex immunoassay. Mean fluorescence intensity (MIF) of each analyte was detected using a Luminex 200 instrument (R&D Systems, United Kingdom).

CD15⁺ myeloid-derived suppressor cells of granulocyte lineage (G-MDSCs) were isolated from PBMC using anti-human CD15-microbeads, and T lymphocytes were isolated by negative selection following removal of CD15⁺ cell (Miltenyi Biotech). T cells (2 × 10⁵/well) were cultured in 96-well flat-bottom plates coated with anti-CD3 (OKT3; 3 µg/mL) and anti-CD28 (2 µg/mL; eBioscience), in 200 µL complete medium supplemented with 0.1% β-mercaptoethanol (Thermo Fisher Scientific). Cells were incubated for 4 days at 37°C, 5% CO₂, and proliferation was determined by ³H-thymidine (Perkin Elmer Life Sciences) incorporation using a TopCount NXT scintillation counter (Perkin Elmer). MDSC-mediated suppression was determined by coculture with T cells. Results are expressed as a percentage of T-cell proliferation driven by CD3/28 ligation in the presence of MDSCs, relative to T-cell proliferation in their absence (100%).

Statistical tests used are indicated in relevant sections. All experiments where tests have been applied were performed on at least 3 donors. Tests were considered statistically significant if P < .05.

To identify the lymphocyte subset infected by EBV, we established a multicolor flow cytometry-based assay (FlowRNA) to simultaneously detect the abundantly expressed viral noncoding RNAs (EBER-1/2) present in every infected cell, together with lymphocyte subset markers CD3, CD4, CD8, CD19, and CD56 on 1 to 2 × 10⁶ fresh mononuclear leukocytes (PBMC). The assay (Figure 1A) involved cell surface staining, cell fixation, permeabilization, intracellular staining, and then ISH for the viral EBERs, and for B2M transcripts as a positive control. The cells were analyzed by flow cytometry; lymphocytes were identified by forward vs side scatter and CD14⁺ monocytes, and dead cells and cell doublets were excluded from all subsequent analysis (Figure 1B). The assay was initially validated by adding decreasing numbers of EBV⁺/GFP⁺ Akata cells to EBV⁻/GFP⁻ Akata cells. The dual EBER⁺/GFP⁺ populations matched the input cell number and demonstrated assay sensitivity and specificity (Figure 1C). EBV-infected lymphocyte subsets were not identified in the blood of healthy EBV carriers (n = 10), where the percentage of infected B cells was below the threshold of detection (Figure 1D). However, the assay unambiguously identified EBV-infected subsets in previously frozen reference samples from patients with EBV-associated LPDs. In the case shown, EBV was detected in CD3⁺/CD8⁺ T cells with a small fraction detected in CD19⁺ B cells (Figure 1D). Virus loads of ∼8 × 10⁴ virus copies per milliliter blood were reliably detected by FlowRNA provided at least 500 × 10⁵ PBMCs were analyzed.

The clinical data for all 5 patients are summarized in Table 2 (and supplemental Data). Each patient presenting with EBV-associated LPDs had high EBV loads measured in whole blood, enabling the identification and enumeration of infected lymphocyte subsets directly from blood. The FlowRNA revealed clear populations of EBER-positive (EBER^POS) and EBER-negative (EBER^NEG) lymphocytes; no other cell types, including myeloid cells, were EBER^POS. In patients 1 and 2, the EBER^POS cells were identified as CD19⁺ B cells (Figure 2A), where ∼1% of total B cells were EBER^POS. Although this appears to be relatively low, detection of EBV in 1% of B cells is remarkably high, equivalent to the number of EBV-infected B cells in the blood of acute IM patients.¹⁶  In contrast, the numbers of EBV-infected B cells in healthy carriers is on average 1 B cell in 10 000,¹⁷  below the threshold of detection of the assay (Figure 1D).

In patients 3 and 4, FlowRNA revealed EBV infection of the CD4⁺ T cells (Figure 2B). The percentage of EBER^POS cells within the total CD4⁺ T-cell population was high for both patients: 49% from patient 3 and 40% from patient 4. Patient 5 had a more complex profile of non–B-cell infection. EBV was detected in at least 80% of total CD56⁺ NK cells, 100% of which were CD56^high (Figure 2B). EBV was also detected in CD4⁺, CD8⁺ T cells, and CD19⁺ B cells.

Notably, the level of CD3 expression was downregulated on EBER^POS T cells but not EBER^NEG T cells, shown by MFI (Figure 2B). The CD3 MFI was consistently lower on EBER^POS T cells from patient 3 and 5 compared with EBER^NEG T cells (patient 3: EBER^POS 446 vs EBER^NEG 614; patient 5: CD4⁺: EBER^POS 1240 vs EBER^NEG 2019; and CD8⁺ MFI: EBER^POS 459 vs EBER^NEG 1204). CD3 expression on the EBER^POS CD4⁺ T cells from patient 4 was significantly lower than the EBER^NEG cells (EBER^POS 210 vs EBER^NEG 1054).

The EBV-associated T/NK-cell LPDs are considered monoclonal/oligoclonal proliferations of EBV-infected T/NK cells due to the presence of a monoclonal virus.¹⁸  To extend these findings, we investigated T-cell receptor (TCR) clonality of the EBER^POS vs the EBER^NEG CD4⁺ T cells using a panel of monoclonal antibodies directed against Vβ epitopes to delineate the TCRαβ repertoire. The TCR-BV usage in the EBER^POS CD4⁺ T cells from patients 3 and 4 and the EBER^POS CD8⁺ T cells from patient 5 revealed the presence of a monoclonal EBV proliferation; TCR-BV8, TCR-BV12, and TCR-BV13.2, respectively (Figure 2C). However, the EBER^POS CD4⁺ T cells from patient 5 were a clonally diverse EBV-positive T-cell population. In contrast, the total EBER^NEG CD4⁺ and CD8⁺ T cells from each patient retained a heterogeneous spread of TCR-BV usage (Figure 2C). Single-cell cloning of the EBER^POS and EBER^NEG CD4⁺ T cells from patient 4 and subsequent sequencing of the TCR Vα and Vβ chains confirmed the TCR-BV8 had identical CDR3 DNA sequences (supplemental Figure 1).

The adaptability of FlowRNA enabled the simultaneous comparison of phenotype and function between EBER^POS and EBER^NEG cells directly from patient blood, with matched lymphocyte subsets from healthy controls (n = 6). We developed FlowRNA panels to examine markers of T-cell subsets, activation, proliferative status, and functional capacity to determine how EBV alters the phenotype and function of infected cells. We added CD45RA and CCR7 to discriminate between naive (CD45RA⁺, CCR7⁺), central memory (CD45RA⁻, CCR7⁺), effector memory (CD45RA⁻, CCR7−), and terminally differentiated effector (CD45RA⁺, CCR7−) CD4⁺ T cells; CD38 and Ki67 to examine activation and proliferative status; Foxp3 and CD25 to identify regulatory T cells (Tregs); plus IFN-γ and TNF-α to examine functional capacity.

Interestingly, EBV infection appeared to be confined to memory T cells. The EBER^POS CD4⁺ T cells from patients 3 and 4 expressed neither CD45RA nor CCR7, consistent with memory (Figure 3 top row). In contrast, the EBER^NEG CD4⁺ T-cell population maintained naive, central memory, effector memory, and terminally differentiated effector CD4⁺ T-cell subsets at similar ratios to those observed in healthy donors; 2 representative donors are shown (Figure 3A). These EBER^POS CD4⁺ T cells also showed greater levels of activation and proliferation (CD38 and Ki67) than the EBER^NEG CD4⁺ and healthy donor CD4⁺ T cells. In patient 3, >60% of EBER^POS cells expressed CD38, with at least 15% of these coexpressing Ki67. In patient 4, the majority of EBER^POS cells expressed high levels of CD38 and coexpressed Ki67 (Figure 3B). Such findings suggest that EBV may be driving the activation and proliferation of infected cells.

Examination of markers of Tregs, FoxP3 and CD25, revealed the EBER^POS cells were not Tregs. Furthermore, there was no evidence of increased numbers of Tregs in the EBER^NEG population, compared with healthy donors (range 3% to 4%) (Figure 3C). However, EBV appeared to significantly alter the cytokine expression profile of infected cells in patients 3 and 4. Patient and healthy donor PBMCs were stimulated for 6 hours with phorbol 12-myristate 13-acetate (PMA) and ionomycin in the presence of brefeldin A, followed by intracellular cytokine staining. EBER^POS CD4⁺ T cells exhibited remarkably different cytokine expression profiles. Although the majority (>70%) of EBER^POS cells from patient 3 produced IL-17A and IFN-γ, the majority (>80%) of EBER^POS cells from patient 4 produced TNF-α and/or IFN-γ (Figure 3D) but little IL-4 and IL-17 (data not shown). In comparison, the EBER^NEG CD4⁺ T cells from both patients exhibited a cytokine expression profile broadly similar to that of healthy donor CD4⁺ T cells (Figure 3D), with low abundance of IL-4 and IL-17A expressing cells (data not shown).

To investigate whether the EBV infected CD4⁺ T-cell clones were EBV specific, we tested their capacity to produce IFN-γ in response to stimulation with a panel of HLA II–matched lymphoblastoid cell lines . In these analyses, no response was seen (data not shown); however, further studies are required to determine the specificity of the EBV-infected CD4⁺ T cells.

We received blood from patient 3 following a diagnosis of HLH; the peripheral blood EBV load was 673 035 EBV DNA copies per milliliter, and the EBER^POS cells accounted for 17% of their total CD4⁺ T cells (Figure 4A). The patient was treated with corticosteroids and etoposide for the next 6 weeks, based on the HLH-2004 protocol.¹⁹  After this initial phase of therapy (8 weeks posttreatment), although the clinical and laboratory features of HLH had resolved, EBV load was consistently >10⁶ copies per milliliter and the proportion of EBER^POS CD4⁺ T cells had increased >fourfold to 72% (Figure 4A). The total white blood cell (WBC) and lymphocyte counts for pretreatment and posttreatment samples were within the normal ranges, and the counts were very similar (WBC: 4.1 cells per μL vs 8.0 cells per μL for pre- vs posttreatment; lymphocytes: 1.7 cells per μL vs 1.6 cells per μL for pre- vs posttreatment). The patient subsequently underwent an alemtuzumab-based, reduced-intensity conditioned HSCT from a matched unrelated donor. Prior to transplant, the proportion of CD4⁺ T cells infected with EBV remained high, at ∼50%. Immediately following HSCT, EBV became undetectable in peripheral blood and remained so for >18 months posttransplant, aside from 1 episode of low-level reactivation where we confirmed the virus was exclusively in the B cells, successfully treated with rituximab. However, the patient’s EBV DNA load began to rise rapidly at 21 months posttransplant (from 34 950 to 684 750 copies per milliliter over a 3-week period). At 22 months posttransplant, we confirmed reemergence of EBV-infected CD4⁺ T cells in the peripheral blood (Figure 4B). The patient died ∼6 weeks after this date. The CD4⁺ T-cell population comprised 40% of cells positive for TCR-BV8 (Figure 4C), suggesting that the original clone of EBV-infected CD4⁺ T cells had persisted following transplantation and subsequently recrudesced, leading to disease relapse.

Patient 4 initially presented with reactive lymphadenopathy, gastritis with benign histological appearances, and high EBV load. After 6 months, a gastric biopsy demonstrated a dense infiltration of mature T cells, predominantly EBV⁺ CD4⁺ and an EBV load of 1 393 353 EBV DNA copies per milliliter. We received our initial sample at this point where EBER^POS cells accounted for 16% of their total CD4⁺ T cells. He received steroid treatment followed by GDP (gemcitabine, dexamethasone, cisplatin) chemotherapy with some initial clinical improvement. Subsequent EBV load showed a climb to >8 000 000 EBV DNA copies per milliliter, and the proportion of EBER^POS CD4⁺ T cells increased 2.5-fold to 40% (Figure 4A). He was planned for SMILE (dexamethasone, methotrexate, ifosfamide, asparaginase, etoposide) chemotherapy but became acutely unwell and died due to bowel ischemia.

At initial presentation with HLH, the vast majority of EBER^POS cells were CD56^high NK cells, with small additional populations of EBER^POS CD4⁺ and CD8⁺ T cells. We therefore examined the phenotype and function of all infected subsets and their EBER^NEG counterparts. As observed in patients 3 and 4, EBV infection of the CD4⁺ and CD8⁺ T cells from patient 5 was restricted to cells without CD45RA and CCR7 expression, consistent with a memory phenotype (Figure 5A,E); the EBER^NEG T cells and matched healthy controls maintained the expected distribution of naive and memory CD4⁺ T-cell populations. However, the EBER^POS CD4⁺ T cells completely lacked expression of the activation/proliferation markers CD38 and Ki67, with expression matching that of the EBER^NEG CD4⁺ T cells and matched healthy controls, suggesting EBV had not activated these infected cells. Furthermore, the EBER^POS CD8⁺ T cells exhibited low levels of activation but little evidence of proliferation (Figure 5B,F).

The EBER^POS CD4⁺ were not Tregs (CD25⁻, FoxP3⁻), and the proportion of Tregs within the EBER^NEG or healthy control populations was not increased (Figure 5C,G). However, the majority of EBER^POS CD4⁺ T cells expressed TNF-α (± IFN-γ) with only a subset expressing neither cytokine (Figure 5D,H). All the EBER^POS CD8⁺ T cells expressed IFN-γ and TNF-α, or IFN-γ alone. In comparison, over half the EBER^NEG CD8⁺ T-cells expressed neither cytokine.

Finally, we examined the same markers on the NK cells. The EBER^POSCD56^high NK cells lacked expression of CD45RA and CCR7 (Figure 5I), unlike the EBER^NEG patient and healthy donor NK cells, which expressed CD45RA alone. The entire populations of EBER^POS and EBER^NEG NK cells were activated (CD38⁺), although only the EBER^POS cells coexpressed Ki67 (Figure 5J), and the majority of EBER^POS and EBER^NEG NK cells expressed either TNF-α alone or in combination with IFN-γ. In contrast, ∼50% of healthy donor NK cells expressed neither cytokine (Figure 5K). The CD56^high NK cells were CD16^neg (data not shown).

Analysis of fresh patient blood revealed large expansions of cells resembling neutrophils that copurified with PBMCs following density centrifugation. Such cells were present at low abundance in healthy donor blood (n = 6, <0.5% of PBMC) but outnumbered lymphocytes in 3 of the 5 cases (Figure 6A-B). These cells had nuclear morphology typical of cells undergoing granulopoiesis (Figure 6C) and expressed CD11b⁺, CD15⁺, CD16⁺, CD33⁺ but CD14⁻ and HLA class II (Figure 6D), confirming their identity as G-MDSCs.

Numerous growth factors and cytokines are thought to drive the expansion and suppressive activity of MDSCs. Luminex assays performed on patient plasma revealed the presence of high concentrations of GM-CSF, IL-1β, IL-6, and TNF-α (Figure 6E) as well as IL-4, IFN-γ, and IL-10 (data not shown) relative to healthy individuals. Additional analysis of plasma from a banked collection of EBV-associated HLH patients (supplemental Table 2) also showed increased concentrations of these same cytokines (Figure 6E). Furthermore, analysis of tissue culture medium from EBV-infected but not EBV-negative T-cell clones established from CAEBV patients contained high concentrations of these cytokines (supplemental Figure 3), suggesting the EBV-infected cells were producing the majority of the cytokines. Interestingly, we cloned EBV-specific CTLs from 1 patient that showed recognition of LMP1 epitopes, exhibiting epitope-specific degranulation and ability to kill autologous cells presenting these epitopes (supplemental Figure 2).

MDSCs are thought to suppress effector T-cell growth and function through several mechanisms, including production of arginase-1. T cells depend on l-arginine for proliferation, and function is particularly sensitive to arginine depletion by arginase-1. Luminex analysis showed an increase in the concentration of arginase-1 in patient plasma compared with healthy donors, where the concentration was below the level of detection (Figure 6F).

MDSCs are highly labile and cryosensitive; therefore, functional studies were performed on freshly isolated patient MDSCs. Ex vivo MDSCs were cocultured with allogeneic T cells prelabeled with tritiated thymidine to analyze the growth of the T cells in the presence of MDSCs. T-cell growth was increasingly suppressed when cocultured with increasing MDSCs at T-cell:MDSC ratios from 1:0.125 up to 1:1, compared with T cells cultured alone (Figure 6G). Furthermore, T cells cocultured with MDSCs at a 1:1 ratio exhibited significant growth inhibition when compared with culture with MDSCs from healthy individuals (P = .0377) and to T-cell growth alone (P = .0206) (Figure 6H).

FlowRNA has significant clinical utility. The assay is particularly useful in confirming the differential diagnosis of an EBV-associated B-cell vs T/NK-cell LPD. This timely confirmation is important when considering targeted treatment options, monitoring treatment outcomes, and identifying early relapse. This potential was clearly demonstrated for 1 patient who received HSCT due to EBV-infected CD4⁺ T cells. Following HSCT, FlowRNA identified an expansion of EBV-infected B cells, which was successfully treated with rituximab. The FlowRNA subsequently detected disease relapse arising from a reemergence of the original EBER^POS CD4⁺ T-cell clone.

FlowRNA has provided fundamental insights into disease pathogenesis. Although we and others have previously reported infection of >1 lymphocyte subset in patients with T/NK-cell LPDs,^10,20  it is not currently known if all infected subsets contribute to the disease. Our data suggest perhaps not. When comparing the activation status of the EBER^POS CD4⁺ T cells from patients 3, 4, and 5, the CD4⁺ T cells from patient 5 showed a lack of activation, no difference in cytokine expression between the EBER^POS, EBER^NEG, and healthy donor CD4⁺ T cells, and lack of clonality. In the context of multiple subset infection, these EBER^POS CD4⁺ T cells may be bystander infections that contribute little to the overall disease pathogenesis and may represent early infection of progenitor cells before TCR rearrangement.

Decreases in CD3 and TCR expression were noted on all EBER^POS T cells relative to their uninfected counterparts. As exemplified in patient 4, high level of activation is known to decrease cell-surface levels of CD3 and TCR, and indeed these cells coexpressed high levels of activation markers, CD38 and Ki67. Interestingly, EBV infection appears to be restricted to CD45RA⁻/CCR7− memory T cells, reminiscent of B-cell infection where EBV drives germinal center-independent mutagenesis and clonal evolution of immunoglobulin genes in naive B cells to make them resemble memory B cells.²¹  Perhaps EBV also drives TCR-independent differentiation of memory T cells.

Although the clonal EBER^POS CD4⁺ T cells were consistent in their differentiation and activation status, their cytokine expression profile differed. Although we expected these differences would cosegregate with clinical presentation, this was not the case. Instead, infected CD4⁺ T cells exhibited differing cytokine expression profiles, irrespective of the disease. This suggests that EBV infection does not inhibit the functional differentiation of CD4⁺ T cells. Interestingly, the uninfected CD4⁺ T cells from patients and healthy donors were remarkably similar in phenotype and function, suggesting uninfected cells were unaffected by the chronic inflammatory microenvironment.

Understanding how EBV-infected T and NK cells evade immune surveillance is of fundamental importance, particularly in the context of limited effective therapeutic options. MDSCs play a pivotal role in suppressing host immunity. Although most intensively studied in cancer, MDSCs are also generated during chronic viral infection (including hepatitis C virus, HIV, and hepatitis B virus), where they modulate the immunopathology and inhibit antiviral T-cell responses.^22-24  Although evidence presented in this report demonstrates in vitro inhibition of T-cell growth by patients’ MDSCs, it is entirely possible that the MDSCs may also inhibit the antiviral responses, as reported in chronic infection scenarios, enabling persistence of EBV-infected cells; however, this does require further investigation. Mechanisms of MDSC-mediated inhibition of T-cell growth and function are known to include arginase-1, reactive oxygen, and nitrogen species. These mechanisms can be blocked by inhibitors, resulting in expansion of T cells in the presence of MDSCs. Although the lack of patient samples precluded these inhibition experiments, raised concentrations of arginase-1 in patient plasma suggest these mechanisms may play at least some role in the inhibition of T-cell growth and function in these patients.

The activation and expansion of MDSCs are achieved, at least in part, by expression of growth factors (GM-CSF, granulocyte colony-stimulating factor) and proinflammatory cytokines (IL-1β, IL-6, TNF-α).^25,26  These factors are often highly induced in chronic infection and stimulate the activation of myeloid progenitors, but may also drive the expansion and activation of MDSCs.^27,28 

This study has shown EBV-infected T/NK cells appeared to be refractory to treatment with the HLH2004 regimen containing etoposide, dexamethasone, and cyclosporin A. For our evaluable patients, the number of EBV-infected cells increased in number, whereas the total WBC and lymphocyte counts remained within the normal range and both the pre- and the posttreatment samples were numerically similar. The expansion of infected cells following treatment suggests that a subset of infected cells was resistant to etoposide and was able to rapidly grow following the end of treatment. However, further studies are required to determine if infected cells expanded throughout the course of treatment or after treatment stopped. Corticosteroid treatment, usually dexamethasone or methylprednisolone, is a central component of commonly used protocols for EBV-associated HLH and CAEBV. The rationale for steroid-containing regimens is suppression of the transcription of T-cell effector cytokines. However, an important consequence is the loss of effector T cells. These effector T cells may play a role in controlling the outgrowth of EBV-infected cells, and their loss may contribute to the expansion of infected cells following treatment. Therefore, one of 2 mechanisms could be restricting the outgrowth of EBV-infected T/NK cells: the MDSCs themselves or residual activity of endogenous EBV-specific CTLs. In either scenario, corticosteroids would disrupt this restriction, leading to unchecked expansion of infected T/NK cells.

Although only demonstrated in a small number of patients, these data have major implications for therapy. The FlowRNA assay will enable a thorough examination of how the number and function of EBV-infected T/NK-cells impact disease outcome and relapse following therapy and provide a better understanding of the implications of using agents such as etoposide and corticosteroids as first-line therapy. Furthermore, the presence of MDSCs in other chronic viral infections has a significant impact on the disease outcome, particularly when considering the use of cell-based therapies. A thorough understanding of the role of MDSCs in EBV-associated HLH and CAEBV will enable a more targeted approach to treat these patients without the need for resorting to stem cell transplants.

This work was supported by grants from the Medical Research Council, MR/N023781/1 (C.S.-L.) and the Histiocytosis Society (C.S.-L.).

Contribution: P.J.C., L.G., H.P., G.R., C.D.S., F.M., H.L., and C.S.-L. performed experiments; P.J.C., C.P.F., H.L., and C.S.-L. conceived the experimental approach; P.J.C., C.P.F., and C.S.-L. wrote the paper; and C.P.F., L.G., and D.L. provided clinical input.

Conflict-of-interest disclosure: The authors declare no competing financial interests.

Correspondence: Claire Shannon-Lowe, Institute for Immunology and Immunotherapy, University of Birmingham, Vincent Dr, Birmingham B15 2TT, United Kingdom; e-mail: c.shannonlowe@bham.ac.uk.