Third-party virus-specific T cells eradicate adenoviraemia but trigger bystander graft-versus-host disease


The adoptive transfer of T cells with anti-viral properties against pathogens, such as cytomegalovirus (CMV) and Epstein–Barr virus (EBV), from allogeneic donors to recipients has been shown to be highly effective, both as prophylaxis and treatment for viral reactivation following haematopoietic stem cell transplantation (HSCT) (Rooney et al, 1998; Peggs et al, 2003; Leen et al, 2009). Conventional procedures for the generation of virus-specific T cells require weeks of cell culture and most studies have therefore relied on the elective production and pre-emptive transfer of cells. Recently, tri-specific donor T cell populations with specificity against EBV, CMV and adenovirus (ADV) have been used in this manner (Leen et al, 2006). As an elective strategy this approach is highly effective, but is very labour intensive, time consuming and costly. An alternative approach, which allows rapid identification and enrichment of virus-specific donor cells, relies on the detection of gamma-interferon- (IFN-γ) responses to pathogens following ex-vivo stimulation and offers more targeted therapy. A bi-specific antibody is used to ‘capture and fix’ IFN-γ as it is released from stimulated cells and magnetic bead selection bead is then used to isolate responding cells (Chatziandreou et al, 2006). The IFN-γ capture strategy allows rapid selection of both CD4+ and CD8+ T cells and has recently been tested in pilot studies in the UK (Mackinnon et al, 2008) and Germany (Feuchtinger et al, 2006). We recently deployed this technology to identify and select virus-specific T cells from a third-party donor to treat intractable ADV viraemia following mis-matched unrelated donor stem cell transplantation.

The patient, a 7-year-old girl who had suffered relapse of acute lymphoblastic leukaemia (ALL) 2 years after successful treatment with regimen A of the UK-ALL 2003 protocol, had undergone a 1C-mismatched unrelated donor peripheral blood HSCT. She received conditioning with total body irradiation (1440 cGy), cyclophosphamide (200 mg/kg) and alemtuzemab (1 mg/kg). Graft-versus-host disease (GVHD) prophylaxis had comprised cyclosporin and mycophenolate mofetil. In the early post-transplant period, adenoviraemia was treated with cidofovir but, in the context of poor T cell immunity, extremely high viral loads developed (Fig 1). Despite withdrawal of all immunosuppression the patient remained T cell lymphopenic (CD3 0·070 × 109/l) and intractable viraemia persisted.

Figure 1.

 The Adenoviral load is shown peaking at levels above 107/ml and remained grossly elevated despite withdrawal of immunosuppression and Cidofovir therapy. The patient was profoundly lymphopenic, with barely detectable CD4 T cells for the first 100 d. Third party, haploidentical Adenovirus (ADV) T cells were administered on day 104 at a dose of 104/kg, and over the next 4–6 weeks there was clearance of ADV in the peripheral blood. During this period there was also a rapid rise in T cell number and the patient developed skin GVHD (stage III) and liver GVHD (stage IV) with profound cholestasis and grossly elevated serum biluribin (SBR).

Screening of the original donor’s peripheral blood lymphocytes (PBLs) revealed absent IFN-γ response against ADV hexon protein. In contrast, a haploidentical family donor was found to mount a strong IFN-γ response (Fig 2A). Peripheral blood was collected (500 ml) and the mononuclear cell fraction isolated by Ficoll gradient centrifugation. Cells secreting IFN-γ in response to hexon stimulation were captured using the Miltenyi CliniMacs system and infused at a dose of 104/kg the same day. Over a period of 4–6 weeks the high level adenoviraemia resolved (Fig 1), and during this period there was a notable recovery of CD4+ and CD8+ T lymphocyte numbers in the peripheral blood. Although chimerism analysis revealed that 99% of peripheral blood CD3+ T cells were derived from the original donor, a third-party haploidentical signal (<1%) was also detectable. Expression of HLA-A2 by haploidentical cells, but not by unrelated donor or recipient cells, allowed tracking of the ADV-specific populations by flow cytometry. Remarkably, 6 weeks after infusion, challenge of lymphocytes with hexon in vitro revealed that almost all the responding cells were expressing HLA-A2, suggesting that these cells were exclusively responsible for viral clearance (Fig 2B). Skin and liver GVHD developed during this 6-week period. Although the skin GVHD was self-limiting (Stage III), there was extensive damage to the biliary tract resulting in vanishing bile duct syndrome and protracted cholestasis. Liver biopsy confirmed relative paucity of interlobular bile ducts associated with a mild portal tract T-lymphocytic infiltrate. Fluorescence in situ hybridization (FISH) detected only original donor (female, XX) cells and no haploidentical (male, XY) T cells in the liver. Despite treatment with methylprednisolone, tacrolimus and mesenchymal cell therapy, cholestasis persisted and liver GVHD led to progressively deteriorating liver function. ADV-specific responses, mediated by HLA-A2 haploidentical cells, remained detectable in the peripheral blood during this period and adenoviraemia was controlled without additional antiviral therapy. Unfortunately, 6 months after the original transplant procedure the patient succumbed after developing severe pneumonitis concomitant with re-activation of CMV.

Figure 2.

 (A) Flow cytometry of peripheral blood lymphocytes stained for expression of T cell subset markers CD4 or CD8, and capture of IFN-γ at the cell surface following overnight stimulation with Hexon protein. The original matched unrelated donor (MUD) failed to mount a response, whereas the paternal haploidentical sample (Haplo) showed a strong response. (B) Differential expression of HLA-A2 and XY FISH studies enabled the identification of third party cells (XY, HLA-A2). (i) Six weeks after infusion, overnight stimulation of peripheral blood lymphocytes with Hexon revealed potent responses mediated almost exclusively by the haploidentical donor, HLA-A2 positive cells. (ii) Liver biopsy revealed relative paucity of interlobular bile ducts with a portal tract lymphocytic infiltrate, consistent with GVHD (‘vanishing bile duct syndrome’) (Haematoxylin & Eosin ×100). (iii) FISH studies detected XX cells from the unrelated donor, but no XY haploidentical cells (×600; X-centromere labelled red; Y-probe labelled green; no green Y signals are present; hybridization and washes were carried out according to standard Abbot protocols and imaged using an Applied Imaging Cytovision system using an Olympus BX61 microscope). Analysis of 100 cells across the tissue section found two X centromere signals from all the cells screened and the entire slide was screened to exclude the presence of any XY centromeres. Thus, only female lymphocytes from the original unrelated donor were detected in the liver.

ADV morbidity includes pneumonitis, hepatitis and colitis after HSCT. In a previous prospective study of 155 paediatric patients, we detected ADV infection in 41% of patients, with 17% becoming viraemic (Kampmann et al, 2005). We can report that haploidentical cells, responsive to ADV, can be rapidly selected and administered to provide rescue therapy for refractory adenoviraemia. However, administration of potent antiviral T cells in the context of extremely high viral loads carries a risk of inflammatory complications and organ damage. It was anticipated that tissue damage during virus clearance phase could arise and would be directly mediated by the introduced virus-specific populations, either as part of direct recognition of ADV antigen or following allo-recognition of mismatched HLA antigens. In the event we found no evidence of haploidentical populations in the liver, the site of the most extensive tissue damage. It appears that effector responses mediated by cells from the original donor, which were 1-antigen HLA-mismatched, only become problematic when ‘kick-started’ by the infusion of highly specific anti-viral cells in the context of extremely high viral loads. Proliferation of bystander T cell populations, arising in association with antigen-specific T cell responses has been previously described in animal models (Tough et al, 1996) and in humans after vaccination (Di Genova et al, 2006). In the post-transplant setting, with a lymphopenic host, homeostatic expansion usually occurs over a period of weeks or months. Here, there was rapid and accelerated donor T cell expansion following infusion of a relatively small number of third party, anti-viral T cells. In the context of extremely high viral loads, we postulate that cytokines and other factors released during viral clearance stimulated the bystander phenomenon, and therefore advocate early intervention with cellular therapy upon detection of viraemia and before viral disease becomes established.


This work was supported by the Leukaemia and Lymphoma Research Fund, Rowntree Trust and was undertaken at GOSH/UCL Institute of Child Health which receives funding from the Department of Health’s NIHR Biomedical Research Centre’s funding scheme. We are grateful to Mr Steve Chatters, Principal Clinical Cytogeneticist, Great Ormond Street Hospital for FISH analysis. The procedure received ethical approval from the local research and ethics committee and was performed with informed consent.