Safe adoptive transfer of virus-specific T-cell immunity for the treatment of systemic adenovirus infection after allogeneic stem cell transplantation


Dr Tobias Feuchtinger, Department of Paediatric Haematology/Oncology, University Children's Hospital, Eberhard Karls University Hoppe-Seyler Strasse 1, D-72076 Tuebingen, Germany.


During periods of immunosuppression, such as postallogeneic stem cell transplantation (SCT), patients are at significant risk for severe viral infections. Human adenovirus (HAdV) infection is a serious complication post-SCT, especially in children. Virus-specific T cells are essential for the clearance of HAdV, as antiviral chemotherapy has revealed limited success. We present feasibility data for a new treatment option using virus-specific donor T cells for adoptive transfer of immunity to patients with HAdV-infection/reactivation. Virus-specific donor T cells were isolated and infused into nine children with systemic HAdV infection after SCT. Isolation was based on γ-interferon (IFN-γ) secretion after short in vitro stimulation with viral antigen, resulting in a combination of CD4+ and CD8+ T cells. 1·2–50 × 103/kg T cells were infused for adoptive transfer. Isolated cells showed high specificity and markedly reduced alloreactivity in vitro. Adoptive transfer of HAdV-specific immunity was successful in five of six evaluable patients, documented by a dose-independent and sustained in vivo expansion of HAdV-specific T cells, associated with a durable clearance/decrease of viral copies. T-cell infusion was well tolerated in all nine patients, except one case with graft-versus-host disease II of the skin. In conclusion, induction of a specific T-cell response through adoptive transfer was feasible and effective. When performed early in the course of infection, adoptive T-cell transfer may protect from HAdV-related complications.

Allogeneic stem cell transplantation (SCT) is widely used in the management of malignant and non-malignant diseases. To prevent immunological rejection of the graft, the patient has to be immunosuppressed at the time of transfer of the stem cell inoculum. The velocity of immune reconstitution following allogeneic SCT depends on the conditioning regimen, donor type and the extent of T-cell depletion in the graft, which is necessary in human leucocyte antigen (HLA) mismatched donors to prevent graft-versus-host disease (GvHD; Handgretinger et al, 2001; Lang et al, 2003). Moreover, only a few donor-derived thymocytes exit the thymus until 3–6 months after transplantation (Lewin et al, 2002; Bahceci et al, 2003). This leaves a considerable period when the host is severely deficient in T-cell immunity. Viral infections are therefore one of the major causes of morbidity and mortality, especially in patients who receive a haematopoietic stem cell transplant from an unrelated or mismatched donor (Handgretinger et al, 2001; Lang et al, 2003; Mohty et al, 2003). In most cases, viral infections result from reactivation of latent viruses, such as cytomegalovirus (CMV), human adenovirus (HAdV) and Epstein–Barr virus (EBV). In children, adenovirus has become the most common viral pathogen responsible for significant post-transplantation morbidity and mortality (Flomenberg et al, 1994; Lion et al, 2003; Walls et al, 2003). Increased frequencies of severe HAdV infections have also been detected in solid organ transplant recipients (Shirali et al, 2001) and human immunodeficiency virus (HIV)-positive patients (Kojaoghlanian et al, 2003). Importantly, an increased risk of adenovirus infection has been correlated with the lack of endogenous T-cell immunity, capable of controlling such infectious agents (Chakrabarti et al, 2002; Feuchtinger et al, 2005; Heemskerk et al, 2005). Current antiviral agents are often limited by weak activity against adenovirus and by side effects, including suppression of bone-marrow function and renal toxicity (Bordigoni et al, 2001; Gavin & Katz, 2002; Ljungman et al, 2003; Lankester et al, 2004; Leruez-Ville et al, 2004).

So far, little is known about the kinetics and the nature of immune responses against HAdV in vivo (Flomenberg et al, 1995; Olive et al, 2002; Leen et al, 2004). Recently, lymphocyte reconstitution and an increase in lymphocyte counts during the first weeks after infection has been shown to play a crucial role in clearance of HAdV viraemia and survival of the host (Chakrabarti et al, 2002; Heemskerk et al, 2005). We have previously detected HAdV-specific T cells in children after SCT and demonstrated that HAdV-specific T cells are protective against HAdV disease after allogeneic SCT (Feuchtinger et al, 2005). It is now appreciated that T-cell reconstitution is required for the control of HAdV infections and that drug therapy might limit, but not clear the infection.

These observations have led to efforts to reconstitute T cells in order to provide physiological protection against infection. This involves induction of a virus-specific T-cell response in the patient by direct infusion of T cells, a procedure known as adoptive transfer. So far, cellular immunotherapy has been directed almost exclusively against two herpes viruses, CMV and EBV (Walter et al, 1995; Rooney et al, 1998; Einsele et al, 2002). Pioneering work by Riddell et al (1992) showed that adoptive transfer of CMV-specific CD8+ T-cell clones into patients at risk of CMV disease protected the patients from CMV-related complications. To date, there have been no clinical trials using adenovirus-specific T cells. However, as a proof of principle for the use of T-cell therapy in cases of adenoviral infection, seven of nine reported cases of unselected donor lymphocyte infusions (DLI) in HAdV infection post-SCT have been successful in reducing viral replication (Hromas et al, 1994; Howard et al, 1999; Chakrabarti et al, 2000; Bordigoni et al, 2001; Chakrabarti et al, 2002). However, unselected donor lymphocyte infusions are associated with a high risk of GvHD. Additional evidence is provided by the fact that withdrawal of immunosuppression has a beneficial effect on the outcome of infection (La Rosa et al, 2001; Chakrabarti et al, 2002). The effect of immunosuppression and the successful treatment of severe adenovirus infection with DLI therefore support the rationale for adoptive transfer of adenovirus-specific T cells.

Based on these observations, an urgent clinical demand for donor lymphocyte preparations with enriched adenovirus-specific T cells and reduced alloreactivity has been expressed (Leen & Rooney, 2005; Moss & Rickinson, 2005). We have previously described a protocol using the antigen-specific γ-interferon (IFN-γ) secretion of T cells to isolate adenovirus-reactive T cells (Feuchtinger et al, 2004). This approach offers several advantages, since the method is easy, fast and can be readily standardised with various malignant and infectious antigens for antigen-specific cellular immunotherapy approaches (Van Driessche et al, 2005). Moreover, there is a reduced risk of GvHD compared with unmanipulated lymphocyte infusions. The functional characterisation of specificity and reduced alloreactive potential was determined to evaluate this technique as an option for adoptive immunotherapy of adenovirus infection post-transplant. The present work investigated the feasibility and efficacy of a specific immunotherapy with IFN-γ secreting T cells after ex-vivo stimulation with viral antigen. To our knowledge, this is the first report of adoptive transfer with specific T cells in patients with systemic adenovirus infection and the first clinical application of cytokine secreting cells after in vitro stimulation.

Patients, materials and methods


Stem cell transplantation recipients with systemic adenovirus infection and failing antiviral chemotherapy, as defined by the persistence or recurrence of HAdV DNA in peripheral blood and/or stool after 2 weeks of antiviral chemotherapy and lacking HAdV-specific T cells, were eligible for immunotherapy. T-cell transfer was performed if a sufficient HAdV-specific T-cell response was detectable in the donor (>0·01% of vital T cells). Patients were closely monitored for acute side effects during the first 24 h following T-cell transfer and for acute and chronic GvHD later on. The efficacy of T-cell therapy was documented by clinical data, reduction in the virus load and in vivo expansion of HAdV-specific T cells in peripheral blood. Patients were treated in three different centres (1, 6 and 2, 3, 4, 5, 8 and 7, 9 respectively). This work was performed in accordance with the declaration of Helsinki and the guidelines of the local human research ethics committee. Informed consent was obtained from all patients and/or their parents. Infection was defined as two consecutive positive results in one of the listed diagnostic tests. Adenovirus disease was defined as the detection of viral antigen and/or DNA together with appropriate symptoms in the absence of any other recognizable cause.

Conditioning regimen, T-cell depletion and supportive care

All patients underwent a myeloablative conditioning regimen (see Table I) based either on total-body irradiation (TBI, 12 Gy) or busulphan 16 mg/kg (except patient 9, see Table I) in combination with anti-thymocyte globulin (ATG Fresenius rabbit) or alemtuzumab (Campath). Stem cell donors were HLA-matched (10/10 alleles) unrelated (MUD n = 3), mismatched unrelated (MMUD n = 4), one haploidentical, parental donor and one from the HLA-matched father. In six cases, T-cell depletion of the stem cell graft was done through CD34+-selection using the CliniMACS® device (Miltenyi Biotech GmbH, Bergisch-Gladbach, Germany), in order to avoid severe GvHD. Two of these patients received an add-back of 107 CD3+ lymphocytes/kg at day zero. No additional donor lymphocytes were administered. Pharmacological GvHD prophylaxis was performed either with methotrexate and ciclosporine or mycophenolate mofetil (see Table I).

Table I.   Patient and transplant characteristics.
 Patient no.
  1. Prophylaxis with ribavirin had already been started when conditioning was begun.

  2. CR, complete remission; ALL, acute lymphoblastic leukemia; AML, acute myeloid leukemia; B-NHL, B-cell non-Hodgkin lymphoma; HLH, haemophagocytic lymphohistiocytosis; DEA, dyserythropoietic anaemia; MMUD, mismatched unrelated donor; MUD, matched unrelated donor; MRD, matched related donor (father); PBSC, CD34+ selected (CliniMACS) peripheral blood stem cells; BM, bone marrow; TBI, total body irradiation; TLI, total lymphoid irradiation; Gy, gray; TT, thiotepa; VP16, etoposide; Flu, fludarabine; Mel, melphalan; ATG, anti-thymocyte globulin Fresenius rabbit; Bu, busulphan; Cy, cyclophosphamide; Cyt, cytarabin; Am, amsacrine; MTX, methotrexate; CSA, cyclosporin; MMF, mycophenolatemofetil; GvHD, graft-versus-host disease; HAdV, human adenovirus; CMV, cytomegalovirus; EBV, Epstein–Barr Virus; PTLD, post-transplant lymphoproliferative disease.

  3. *Patient 6 received prophylactic use of ribavirin (i.v., 3 × 5 mg/kg/d) for prevention of HAdV infection was performed.

Age (years)111096571811
GraftCD34+ PBSC +107 T cells/kgCD34+ PBSCBMCD34+PBSCBMCD34+ PBSC +107 T cells/kgCD34+PBSCPBSCCD34+PBSC
No. mismatch030200223
ConditioningTBI 12 Gy
TT 10 mg/kg
VP16 40mg/kg
Bu 16 mg/kg
Cy 120 mg/kg
TBI 12 Gy
VP16 60 mg/kg
TBI 12 Gy
Cy 120 mg/kg
VP16 40 mg/kg
Flu 180/m2
Mel 140/m2
Campath TLI 2 Gy
Bu 16 mg/kg
Cy120 mg/kg
Bu 20 mg/kg
Cy 120 mg/kg
Mel 140 mg/m2
Flu 180/m2
Mel 140/m2
TLI 2 Gy
Flu 120 mg/m2
Cyt 8000 mg/m2
Am 400 mg/m2
TBI 4 Gy
Cy 200 mg/kg
GvHD prior to T-cell immunotherapyChronic GvHD III of skinNoNoNoGvHD II (skin acute)NoNoGvHD II (skin acute)GvHD IV (intestine)
Day lymphs > 0·3 × 109/l+58Not reached+44Not reachedNot reached+74+12+61+18
Day CD3 > 0·1 × 109/lNot reachedNot reached+44Not reachedNot reached+74+20+47+33
First day of positive HAdV PCR in stool+168+17+54−7+170+4−8+47
HAdV DNA copies/ml prior to immunotherapy
 Stool (copies/ml)Positive8 × 1058 × 1092 × 1094 × 1075·3 × 108Positive2 × 1096 × 102
 Blood (copies/ml)4·5 × 1076 × 1074 × 1047 × 1037 × 1023·3 × 103Positive, not quantified8 × 102No
Positive HAdV PCR in other body fluidsUrine, throat swabMSF, urine, pleural effusionNoNoUrine pos 2 × 103NoThroat swabEye smearNo
Clinical symptoms prior to immunotherapyMultiorgan failure, lung, heart, kidney, liver, bone marrowFever, haemorrhagic enteritis, hepatitis, pleuritis, carditis, encephalitis, retinitisNoneNoneFever, enteritis, hepatitis, neutropeniaDiarrhoeaLung infiltrates, cardio-respiratory insufficiencyEnteritis, feverMassive diarrhoea, pneumonia, pleural effusion
Antiviral chemotherapy prior to immunotherapyRibavirin, cidofovirGanciclovir, cidofovirCidofovirRibavirin, cidofovirRibavirin, cidofovirRibavirin,* cidofovirValacyclovir, cidofovirRibavirin, cidofovirCidofovir
Additional infectionsHHV-6, BKVCMV viraemia (low)EBVNoCMV viraemia (low)CMV, EBV, HHV-6EBV-PTLDCMVEBV
Lymphocytes and subsets prior to immunotherapy
 Lymphocytes × 109/l0·2130·110·310·280·0500·0210·550·651060
 CD3+ × 109/l0·0390·0600·260·0130·0120·003nt0·475nt
 CD4+CD3+ × 109/l0·0060·0050·0330·009nt0·001nt0·19nt
 CD8+CD3+ × 109/l0·020·050·20·002nt0·001nt0·285nt
 CD56+ × 109/l0·0580·0450·050·10·0340·011nt0·15nt
 CD19+ × 109/l0·020·00100·16500nt0·02nt

Surveillance of adenovirus infection and immune reconstitution

The surveillance of infectious complications post-transplant was performed by stool antigen enzyme-linked immunosorbent assay, stool and blood polymerase chain reaction (PCR), as well as virus isolation from urine and throat swabs. Prospective virus screening in peripheral blood and stool was performed at weekly intervals until day +100 after transplantation. Urine and throat samples were obtained in cases of clinical suspicion. After day +100, tests were carried out only in cases of clinically suspected viral infection. Viral load was monitored by quantitative HAdV PCR, using hexon-specific primers with different protocols. In patients 1 and 6, quantification was done in plasma, whereas quantitative HAdV PCR was performed in blood in the remaining patients, as described elsewhere (Heim et al, 2003; Lion et al, 2003). Patients with adenovirus-DNAemia received preemptive treatment with cidofovir (5 mg/kg/week according to a protocol requiring continuation of treatment until attainment of two negative PCR results). All patients received cidofovir for at least 2 weeks prior to adoptive T-cell transfer. In four cases additional ribavirin was administered. Reconstitution of CD3+, CD4+, CD8+, CD19+, and CD56+ lymphocytes was monitored weekly by flow cytometry until the beginning of T-cell recovery.

Detection of adenovirus-specific T cells

Human adenovirus-specific T cells were detected as described recently (Feuchtinger et al, 2005). Briefly, peripheral blood mononuclear cells (PBMC) were stimulated with type C adenoviral antigen (BioWhittaker, Verviers, Belgium) ex-vivo and T cells with antigen-specific secretion of IFN-γ were detected on the following day. Flow cytometric assessment of IFN-γ secretion of vital T cells was carried out by detection of IFN-γ exocytosis (Miltenyi Biotech GmbH, Bergisch-Gladbach, Germany) and by intracellular cytokine staining. Surface staining was performed using saturating conditions of the following antibodies: anti-CD4, anti-CD8 (clones SK3 or SK1), anti-IFN-γ (clone 25723.11), anti-CD3 (clone SK7), all from Becton Dickinson (BD) or BD Pharmingen (Heidelberg, Germany). At least 100 000 lymphoid cells were analysed on a FACS-Calibur with cellquest software (both from BD). Although control antigens did not usually stimulate any IFN-γ production, the percentage of specific T cells was calculated by subtraction of the frequency obtained by the respective negative control.

Isolation of adenovirus-specific T cells

Generation of HAdV-specific T cells was performed as described (Feuchtinger et al, 2004). After obtaining informed consent, PBMC were obtained from the donor and isolated by Ficoll/Paque (Biochrome, Berlin, Germany) density gradient centrifugation of heparinised blood from healthy donors, diluted at 1 × 107 cells/ml culture medium [RPMI 1640 medium (Biochrome, Berlin, Germany) +10% human AB serum] and stimulated with adenoviral antigen type C. The antigen was negative for adventitious infectious agents, since no good medical practice-grade antigens are commercially available. In total, 0·1–1 × 109 PBMC were stimulated for 16 h in a 37°C humidified incubator. Magnetic enrichment of cytokine secreting cells was performed using the Cytokine Secretion System® and the CliniMACS® device (both Miltenyi Biotec, Bergisch Gladbach, Germany). Briefly, cells were labelled with anti-IFN-γ monoclonal antibody conjugated to leucocyte-specific (CD45) antibody, diluted, and incubated for the IFN-γ capture period (45 min). Thereafter, cells were magnetically labelled with anti- IFN-γ microbeads and applied onto the magnetic column. The column was rinsed and the retained cells eluted after removing the column from the magnetic field. No further in vitro expansion was done prior to adoptive T-cell transfer.

Functional analysis of generated T cells

For functional in vitro characterisation of the HAdV-specific T cells, human foreskin fibroblasts were infected as target cells with adenovirus mixtype 2 and 5, using titred TCID50 (the quantity of virus that, when inoculated onto a number of susceptible tissue cultures, will infect 50% of the individual cultures) of 1000 to 10 000. Fibroblasts were matched with effector cells for at least one HLA-I allele. The HAdV strain had been isolated from the plasma of a patient post-SCT. Infection was confirmed by documentation of adenovirus hexon antigen expression by immunohistochemistry using a monoclonal antibody (Virion GmbH, Munich, Germany). Cytotoxicity assays were done as described previously (Feuchtinger et al, 2004). Briefly, target cells were labelled with 2·5μg/ml BATDA-ligand for 30 min at 37°C and subsequently washed five times. Effector to target cell ratios were determined in titred amounts from 20:1 to 2·5:1 with 5 × 104 target cells/ml in a 96-well plate. Specific lysis was measured in triplicate using Europium solution in a multiple scintilation counter, according to the manufacturer's instructions (all from Wallac, Perkin-Elmer, Turku, Finland). Detection of cytokines in supernatants after ex-vivo stimulation of donor mononuclear cells was done using the TH1/2Cytometric bead array from BectonDickinson (Heidelberg, Germany) according to the manufacturer's recommendation. Ex-vivo stimulation was carried out under the same conditions as described for the detection of specific T cells. Alloreactive T cells were detected by co-culture of HAdV-specific T cells with autologous, HLA-matched unrelated and HLA-mismatched mixed lymphocyte cultures. After 1 week, HAdV-specific T cells were restimulated with the same irradiated autologous or allogeneic PBMC and analysed for alloreactive T cell activation by intracellular detection of IFN-γin vitro proliferation was detected with CFSE (carboxyfluorescein diacetate succinimidyl ester, Molecular Probes, Eugene, OR, USA) according to a recently published protocol (De Rosa et al, 2003).


Generation and in vitro analysis of adenovirus-specific T cells

The HAdV-specific T-cell response was 1·1% (mean ± 1% SD) of total donor T cells prior to the isolation of specific T cells. To obtain HAdV-specific T cells, we stimulated mononuclear cells isolated from 200 to 500 mL peripheral blood of the stem cell donor with HAdV antigen. In all nine cases this procedure efficiently increased the percentage of HAdV-specific T cells to 45·7% (mean ± 24% SD) as determined by flow cytometry (see Fig. 1). The isolated specific T cells contained both CD4+ (63·2% mean ± 10% SD) and CD8+ (28·7% mean ± 8% SD) T cells. The amount and purity of isolated T cells was dependent on the percentage of HAdV-specific T cells in the donor. In the positive fraction a median 54% were T cells. The contaminating, non-T cells in the positive fraction, comprised of monocytes and natural killer (NK)-cells. Since the absolute cell count after the isolation procedure was rather low 2·8 × 106 (mean, viable cells), an in vitro expansion of T cells with interleukin (IL)-2 and irradiated autologous PBMC was necessary for in vitro analysis of alloreactivity and cytotoxic function. For characterisation of the cytokine pattern induced by adenovirus antigen, culture supernatants indicated a T-helper 1 response with predominance of IFN-γ, tumour necrosis factor-α (TNF-α) and IL-2 (see Fig. 2). Specific cytotoxic activity of isolated T cells was tested against infected target cells. Specific lysis of infected target cells by HAdV-specific T cells was detected at an effector to target cell ratio between 20:1 and 2·5:1 and compared with uninfected target cells (see Fig. 3). The proliferative potential of isolated, specific T cells was confirmed by a CFSE-assay, in which effector function (IFN-γ secretion) and proliferation was simultaneously confirmed. Almost all proliferated T cells secreted IFN-γ in response to restimulation with viral antigen and hence retain their specific effector function after expansion (see Fig. 4). For detection of residual alloreactive T cells within the isolated T cell graft, co-culture of HAdV-specific T cells was carried out in autologous, HLA-matched unrelated and HLA-mismatched mixed lymphocyte cultures, followed by restimulation and flow cytometric detection of alloreactive T-cell activation by intracellular detection of IFN-γ. In an HLA-matched mixed lymphocyte culture (MLC) no alloreactive T-cell activation was detectable when compared with negative controls. In a mismatched third party MLC, residual alloreactive T-cell activation to 0·28% (mean ± 0·02% SD) of vital T cells was detectable. This residual alloreactive T-cell activation in third party MLC was detected among both CD4+ and CD8+ T cells (see Fig. 5).

Figure 1.

 Flow cytometric analysis prior to and postisolation of human adenovirus (HAdV)-specific T cells. Donor peripheral blood mononuclear cells were exposed to HAdV antigen ex-vivo. γ-interferon secretion of T cells was examined consecutively by flow cytometry as a marker of specific T-cell activation (pre isolation). These adenovirus-specific T cells were isolated through a magnetic column (positive and negative fraction = postisolation). For exclusion of unspecific activation a negative control was analysed (MOCK control). The fraction of γ-interferon positive T cells are shown in as a percentage in the upper right quadrant of each dot plot. Flow-cytometric analysis of all dot plots included an initial gate on T cells (sideways scatter versus CD3; dot plot not shown). In the positive fraction a median 54% were T cells and only small samples with few cells were available for flow cytometric analysis. In all other samples 100 000 cells were aquired for the analysis in the flow cytometric live gate. Contaminating, non-T cells in the positive fraction, comprised monocytes and NK-cells.

Figure 2.

 Characterisation of human adenovirus (HAdV)-specific cytokine profile after ex-vivo stimulation. For characterisation of the T-helper 1 and T-helper 2 response pattern induced by adenovirus antigen, cytokine profiles were investigate. Donor mononuclear cells were stimulated ex-vivo with adenoviral antigen and culture supernatants subsequently analysed. The predominance of IFN-γ, TNF-α and IL-2 indicates a T-helper 1 response, whereas typical T-helper 2 cytokines like IL-4, IL5 and IL-10 show no significant increase to negative (MOCK) control. These lymphokine producing T-helper 1 cells are especially effective against intracellular infections like viruses.

Figure 3.

 Human adenovirus (HAdV)-specific T cells kill HAdV-infected allogeneic fibroblasts in vitro. The functional relevance of isolated T cells was tested by analysis of cytolytic activity in vitro against infected target cells. Allogeneic human fibroblasts were infected as target cells with an adenovirus strain, isolated from one of the patients, in order to resemble the situation after an allogeneic stem cell transplantation in vivo. Infection was confirmed by immunohistochemical detection of the adenoviral hexon protein by a monoclonal antibody (right panel, original magnification × 100). Specific lysis of infected target cells by HAdV-specific T cells (n = 3) was detected at an effector to target cell ratio between 20:1 and 2·5:1 and compared with uninfected target cells for exclusion of alloreactive lysis.

Figure 4.

 Human adenovirus (HAdV)-specific T cells show antigen-specific activation and proliferation in vitro. Effector function (IFN-γ secretion) and proliferation potential of T cells is essential for immunotherapy and both should, therefore, be retained. After the initial antigen contact, cells were pulsed with CFSE, cultured for 6 d, restimulated over night and then analysed for proliferation and intracellular IFN-γ expression by flow cytometry. Panel A shows the flow cytometric gates that were applied to the dot plots in panel B. The left dot plots in panel B. show CD4+ CD3+ T cells and the two right dot plots show CD8+ CD3+ T cells. Both upper dot plots in panel B. show cells stimulated with HAdV and both lower dot plots show cells stimulated with MOCK controls (neg. control). In both upper dot plots (panel B), the proliferating cells (that have lost CFSE), seen in the upper left quadrants, almost all secrete IFN-γ in response to restimulation with viral antigen and hence retain their specific effector function and meet the requirements for in vivo-expansion after immunotherapy.

Figure 5.

 Alloreactive T-cell activation of adenovirus-specific T-cell lines. Beside specificity, the main challenge for a specific T-cell immunotherapy is the reduction of alloreactivity and hence the risk of graft-versus-host disease (GvHD)-induction. Alloreactive T cells were detected by co-culture of human adenovirus (HAdV)-specific T cells with autologous, HLA-matched unrelated and HLA-mismatched mixed lymphocyte cultures (n = 3). After one week HAdV-specific T cells were restimulated and analysed in a flow cytometer for alloreactive T-cell activation by intracellular detection of IFN-γ. In an HLA-matched mixed lymphocyte culture (MLC) no alloreactive T-cell activation was detectable when compared with negative controls. In a mismatched third party MLC a residual alloreactive T-cell activation to 0·3% of vital T cells was detectable.

Feasibility and side effects of cellular therapy

Generation of a specific T-cell product was successful in all nine donors. Since the absolute cell count after the isolation was low usually all isolated cells were infused, leading to different cell doses (see Table II). No acute clinical side effects were documented after infusion of the adenovirus-specific T-cell product. A slight elevation of C-reactive protein levels in blood was seen in two patients and bilirubin levels increased in one patient (see Table II). One patient (patient 1) had a pre-existing chronic GvHD of the skin. In this case an aggravation of the skin GvHD was seen 10–14 d after adoptive T-cell transfer, which may have been related to the infused T-cell product. The aggravation of the skin GvHD was successfully treated with steroids, however, the patient eventually died from the pre-existing multi-organ dysfunction of liver, kidney, heart and lung. Patient 7 died of a pre-existing myocardial infarction, which was confirmed by autopsy.

Table II.   Adoptive transfer of HAdV-specific T cells and patient follow-up.
 Patient no.
  1. PCR, polymerase chain reaction; CSF, cerebrospinal fluid. For abbreviations see Table I.

  2. *For example, rash, nausea, pain, elevation of body temperature, hypotension.

  3. †C-reactive protein 1·1–1·3 mg/l, Bilirubin 359–444 μmol/l.

  4. ‡C-reactive protein 2·4–1·8 mg/l.

  5. §Patient received high-dose steroids because of relapse of NHL on day +98.

T-cell transfer
 Day post-SCT+378+152+77+97+84+53+40+62+64
 CD3+ cells/kg500020 000600050 00025 000300012 70012003300
Acute clinical side effects*NoNoNoNoNoNoNoNoNo
Deterioration in laboratory resultsNoCRP, bilirubin†NoNoCRP‡NoNoNoNo
GvHD postimmunotherapySkin GvHD II post- T-cell transferNoNoNoNoNoNoNoGvHD improved from grades IV to II
In vivo expansion of specific T cells postimmunotherapyYesNoYesNot evaluableNot evaluableYesNot evaluableYesYes
Clinical course postimmunotherapyMultiorgan failure, lung, heart, kidney, liver, bone marrowGeneralised lethal HAdV-infectionClearance of HAdVDeath because of NHL relapseDeath because of HAdV-infectionClearance of HAdVDied 1 d later because of pre-existing myocardial infarctionClearance of HAdVClearance of HAdV
Additional antiviral treatment postimmunotherapyGanciclovir, cidofovirGanciclovir, cidofovirRibavirin, cidofovirRibavirin, cidofovirRibavirin, cidofovirRibavirin RibavirinCidofovir
HAdV infection 7 d postimmunotherapy
 Change of HAdV copies/ml blood↓1 log↓×0·16↑1 log↑2 log↑×4·28Clearance o f HAdV ↑0·4 logNeg
 Change of HAdV  copies/ml stoolHAdV PCR positive↑2 log↓×0·75No changeNo changeHAdV PCR positive ↓4 log↓× 0·8
 Other body fluidsUrine, throat swab, stool positiveCSF pos, urine pos 3 × 105Urine pos 3 × 104  No  No
HAdV infection 14–20 d postimmunotherapy
 Change of HAdV copies/ml blood↓2 log↑×3↓2 log↑3 log§Not evaluableClearance of HAdV Clearance of HAdVNeg
 Other body fluidsUrine, throat swab, stool positiveStool, CSF, urine positive 3 × 107Stool positiveStool positive Stool neg. Stool pos 3 × 107Clearance of HAdV in stool
HAdV infection 30 d postimmunotherapy
 Course of HAdV infectionDeath because of HAdV-infectionNot evaluableClearance of HAdVNot evaluableNot evaluableClearance of HAdVNot evaluableClearance in blood. stool pos 3 × 105Clearance of HAdV

Reconstitution of adenovirus-specific T cells after adoptive T-cell transfer and outcome

Adoptive transfer of adenovirus-specific T cells was performed in order to induce a virus-directed immune response in vivo. The evaluation of the efficacy of this approach therefore included analysis of in vivo expansion of adenovirus-specific T cells after adoptive transfer. Prior to adoptive T-cell transfer (2–3 d) none of the patients had specific T cells detectable in peripheral blood. Hence, later detection of these T cells in peripheral blood was attributed to the therapeutic intervention. The course of infection was evaluated by clinical data and by quantitative analysis of viral DNA in serial blood and stool specimens. The in vivo expansion of adenovirus-specific T cells in peripheral blood post-transfer is shown in Figs 6 and 7. All patients had an increase of the viral load under antiviral chemotherapy prior to adoptive T-cell transfer. Antiviral chemotherapy with cidofovir alone or in combination with ribavirin was continued after adoptive T-cell transfer. Adoptive transfer of HAdV-specific immunity was successful in five of six evaluable patients. A significant decrease of viral DNA in peripheral blood and stool was detectable in all five patients (patients 1, 3, 6, 8 and 9) with an in vivo expansion of specific T cells (Fig. 6). A sustained HAdV-specific T-cell response was detected after 4–6 months post-T-cell transfer in the three patients in which follow-up was possible (%CD3+IFN-γ+ cells: patient 3, 0·5%; patient 6, 0·25%; patient 8, 0·08%). In the remaining three patients without detectable HAdV-specific T cells, viral DNA in blood further increased or remained unchanged (patients 2, 4 and 5) after T-cell transfer. Three patients were not evaluable for in vivo expansion of adenovirus-specific T cells as a response to adoptive T-cell transfer (patients 4, 5 and 7) for the following reasons: patient 4 experienced a relapse of non-Hodgkin lymphoma (NHL) and received high-dose steroids and chemotherapy. Patient 5 died of pre-existing multi-organ failure caused by adenovirus infection 10 d after T-cell transfer. Patient 7 died 1 d postadoptive transfer (see side effects above and Fig. 6). The in vivo expansion of adenovirus-specific T cells after adoptive T-cell transfer was independent of the infused T-cell dose. All patients with successful induction of a T-cell response mediated by the adoptive T-cell transfer (patients 1, 3, 6, 8 and 9) received remarkably low amounts of T cells (1200–6000 T cells/kg recipient body weight), whereas those without in vivo expansion of adenovirus-specific T cells received higher T cells numbers (20 000–50 000 T cells/kg).

Figure 6.

 Stable in vivo expansion of adenovirus-specific T cells after adoptive transfer is associated with a sustained clearance of adenoviral load. The follow-up of in vivo expansion of specific T cells and the course of the viral load after adoptive immunotherapy with adenovirus-specific donor T cells is shown. All patients had no specific T cells prior to adoptive T-cell transfer (2–3 d) and all had increasing viral load despite pharmacological treatment. Panel A shows the T-cell response of those patients with in vivo expansion of specific T cells and panel B shows the corresponding DNA-load of these patients. Adoptive transfer of HAdV-specific immunity was successful in five of six evaluable patients. Adenovirus-specific T cells emerged the first time in peripheral blood after adoptive transfer, increased and decreased again after successful containment of the infection. In all five patients with a detectable T-cell response in peripheral blood postadoptive immunotherapy, a significant reduction or clearance of adenovirus was detected. Four of them cleared the infection and survived (patients 3, 6, 8 and 9). Panels C and D show those patients without a specific T cell response postadoptive T-cell transfer. Only patient 2 showed a failure of in vivo expansion postadoptive transfer. The absence of a virus-specific T-cell expansion (C) was associated with an increasing/unchanged viral DNA-load in peripheral blood (D). The course of infection is documented by viraemia in patients 1–8 and by copies/ml stool in patient 9. Patients 1, 5 and 7 died from adenovirus-associated, pre-existing multi-organ failure; patient 4 died from relapsed NHL.

Figure 7.

 Flow cytometric analysis from one patient prior to and postadoptive T-cell transfer. The plots show the flow cytometric analysis for adenovirus-specific T cells in patient 6. Prior to adoptive T-cell transfer no IFN-γ secreting T cells were detected in response to adenovirus antigen ex-vivo (upper row). After adoptive T-cell transfer, adenovirus-specific T cells expanded in vivo and were detectable in peripheral blood (lower row). The fraction of IFN-γ positive T cells are shown as a percentage of vital T cells in the upper right quadrant of each dot plot.


Viral infections still have a major impact on the outcome of SCT despite prophylactic and pre-emptive treatment approaches with antiviral chemotherapy (Legrand et al, 2001; Ljungman et al, 2003; Lankester et al, 2004) and intensive monitoring of viral load in peripheral blood (Lion et al, 2003; Leruez-Ville et al, 2004). In particular, children with impaired immune recovery have a higher mortality resulting from adenovirus disease (Chakrabarti et al, 2002; Feuchtinger et al, 2005; Heemskerk et al, 2005; Kampmann et al, 2005). By contrast, some patients with adenovirus DNAemia detectable by PCR can clear the virus without antiviral therapy (Walls et al, 2005). It has been shown that T-cell immunity is crucial for protection against HAdV infection or reactivation (Feuchtinger et al, 2005; Heemskerk et al, 2005). We have therefore established a protocol (Feuchtinger et al, 2004) for the enrichment of specific T cells with effector function (cytokine secretion) and proliferation capacity (see Figs 2 and 4), in order to induce and maintain a T-cell response in the host. This pilot feasibility data showed that, after adoptive T-cell transfer, in vivo expansion takes place, associated with a reduction or even clearance of the viral load and a clinical improvement of HAdV-disease. The successful transfer of specific immunity may protect the host from severe sequelae of adenoviral disease, as the presence of a HAdV-specific T-cell response in the host is associated with an improved outcome post-SCT (Feuchtinger et al, 2005; Heemskerk et al, 2005). After isolation and expansion the adenovirus-specific T cells retained their specific effector function (see Fig. 4) and specific cytolytic activity in vitro (see Fig. 3). The in vitro expansion of enriched cells required for functional assays could potentially alter the properties of the T-cell product, in comparison to directly infused T cells. However, the experimental data supported the in vivo expansion after immunotherapy. T cells generated with group C adenovirus have been shown to exert distinct crossreactivity patterns, recognising HAdV serotypes from the three subgroups A, B, and C (Heemskerk et al, 2003). A serotype-independent reactivity in patients post-SCT against the HAdV-antigen used for detection and isolation, indicate crossreactive viral epitopes in the antigen (Feuchtinger et al, 2004, 2005). It is of interest that the two patients with HAdV type A infection (patients 5 and 8) successfully cleared the virus - presumably because of cross-reactivity – although the specific T cells were generated with HAdV type C. High levels of HAdV-DNA load did not seem to impair T cell proliferation in vivo (see Fig. 6, patients 1 and 3), in contrast to the myelosuppressive effects of CMV-viraemia.

Although the number of patients is small, it is remarkable that infection could be improved or even cleared in those patients who showed an in vivo expansion of specific T cells. In surviving patients, a stable HAdV-specific T-cell response was induced, leading to a sustained clearance of viral load. Anti-viral chemotherapy remained unchanged before and after adoptive T-cell transfer. We therefore attributed the good clinical course of these patients to the adoptive immunotherapy. By contrast, in patients lacking in vivo expansion of specific T cells, the infection was aggravated, leading to an increase of the viral load with fatal outcome (see Fig. 6). In one patient adoptive transfer was unsuccessful and in the remaining three patients in vivo expansion was not possible, because of relapsed NHL (patient 4) and death (patients 5 and 7) shortly after immunotherapy (see Table II).

If, during antiviral chemotherapy no sufficient HAdV-specific T-cell immunity develops, the risk of HAdV disease is high (Heemskerk et al, 2005). In the absence of an orchestrated T-cell response with CD4+ T-cell help, the CD8+ T-cell response becomes functionally impaired and also decreases quantitatively over time. This results in deficient CD8+ T-cell memory and recall responses and diminished protective immunity (Sun et al, 2004). In patients with CMV disease after SCT, adoptive transfer of both CD4+ and CD8+ T cells is essential to maintain a virus-specific T-cell response in peripheral blood after adoptive T-cell transfer and can protect patients from CMV-related complications (Riddell et al, 1992; Walter et al, 1995; Einsele et al, 2002). Based on this experience we performed adoptive immunotherapy with an approach that resulted in a polyclonal T-cell graft, containing enriched numbers of HAdV-specific CD4+ and CD8+ T cells. This was in contrast to alternative approaches for virus-specific T-cell therapy based either on the use of tetramers, which lead to pure CD8+ T-cell populations, or repetitive stimulation cycles with viral protein in vitro, which result mainly in CD4+ T-cell populations. Infused T cells were not expanded in vitro, since the progressive acquisition of terminal effector properties in vitro could be associated with progressively impaired in vivo T-cell activation, proliferation and survival (Gattinoni et al, 2005).

Interestingly, the efficacy of the adoptive T-cell transfer was independent from the T-cell dose, suggesting that even low numbers of transferred HAdV-specific T cells are able to expand in vivo in the presence of HAdV viraemia. T-cell activation and proliferation of HAdV-specific T cells both have been sustained in vivo. This supports the rationale that induction of an immune response is possible with very few T cells and substitution of the immune response with high cell numbers is not absolutely necessary. The adoptive transfer of T-cell immunity was independent from the absolute lymphocyte count at the day of T-cell transfer (see Table I), however, we cannot exclude that pre-existing T cells contributed to the T-cell response postadoptive transfer. The favourable outcome of the infection in patients with in vivo expansion of the transferred T cells may have been related to the early onset of adoptive T-cell transfer during the course of infection, when the immune system has time to limit the infection and the organ damage caused by the virus. Patients who showed no in vivo expansion of adenovirus-specific T cells finally died from adenovirus infection or suffered from multi-organ dysfunction before the onset of immunotherapy. Based on this observation, we suggest that, in cases with increasing viral load, a transfer of specific T cells before the onset of symptoms.

The major challenge for the safety of adoptive T-cell immunotherapy is GvHD. The isolation procedure is an enrichment of virus-specific T cells resulting in a mixed population with a potential contamination of alloreactive T cells. Alloreactive T cells carry the risk of GvHD induction and aggravation of viral infections, while virus-specific T cells are expected to improve the HAdV infection. As alloreactivity of isolated HAdV-specific T cells has been markedly decreased in in vitro experiments, low doses of HAdV-specific T cells are expected to carry minimal risk of inducing GvHD. Acute GvHD of the skin in patient 1 was attributable to the pre-existing chronic GvHD of the skin. All other patients did not show any signs of GvHD in response to immunotherapy, despite of unrelated or even mismatched unrelated donors. In patient 9, GvHD improved together with the clearance of infection. While patients with GvHD are at a particularly high-risk for viral infections, the alloreactive potential of the adoptive T-cell therapy is critically important. The T-cell doses used in this study therefore correspond to those applicable even in patients with a mismatched donor.

In conclusion, specific T-cell immunotherapy was performed as a new treatment approach in nine children with systemic HAdV-infection after allogeneic SCT, who would otherwise have a poor prognosis. Immunotherapy was shown to be feasible with no acute toxicities or increased risk of GvHD induction and resulted in successful transfer of specific immunity to the host. In vivo expansion of specific T cells was independent of the T-cell dose and may have protected from HAdV-related complications. Proof of principle for this approach was demonstrated in five out of six evaluable patients. The method could be easily applied to other infectious antigens and to anti-tumour immunotherapy protocols. Multi centre phase II clinical trials are now required to further investigate the therapeutic potential of this approach.


We thank the nurses and physicians working at the paediatric stem cell transplant programme in Tuebingen, Vienna and Munich for their excellent care of the children and for their support, Christiane Braun and Desiree Schelling for their excellent technical assistance and David Martin, Georg Rauser, Regina Alex and Hermann Einsele for helpful advice. We also want to express warm thanks to Dietrich Niethammer for his continuous support and advice. Last but not least we want to thank the Wilhelm-Sander-Stiftung, the Deutsche Forschungsgemeinschaft (SFB 685) and the Fortuen Program (University Tuebingen) for their financial contribution through grants to TF.

Statement of authorship

The design of this scientific work was done by TF, RH and PL; the isolation procedures and the experimental work was performed by TF, CR and MS; responsibility for the care of the patients was taken by SM and CP in Vienna, MF and FS in Munich and PL, RH and TF in Tuebingen. The virus diagnostics were done by TL, RB and GJ. Target cell infection was done by KH and RL generated the human serum. The manuscript was written by TF and critically reviewed by all authors.

Research support

This work was supported by grants from the Wilhelm-Sander-Stiftung, German National Science Foundation (SFB 685) and the Fortuen Program, Tuebingen University (all to TF).