Immunity, Homing and Efficacy of Allogeneic Adoptive Immunotherapy for Posttransplant Lymphoproliferative Disorders

Authors


* Corresponding author: Dr. Maher K. Gandhi, maherG@qimr.edu.au

Abstract

Adoptive immunotherapy using autologous Epstein-Barr virus (EBV)-specific cytotoxic T-lymphocytes (auto-CTL) can regress posttransplant lymphoproliferative disorders (PTLD). Widespread applicability of auto-CTL remains constrained. Generation is time-consuming, and auto-CTL cannot be established in patients treated with the B-cell depleting antibody rituximab. By contrast, pregenerated allogeneic CTL (allo-CTL) offers immediate accessibility. Allo-CTL has previously shown efficacy in “early” polyclonal- PTLD. We treated three patients with aggressive, advanced monoclonal-PTLD following solid-organ transplantation. All were refractory to at least three prior therapies. Despite HLA disparity, there was negligible toxicity, with early in vivo antiviral efficacy and reconstitution of EBV peptide-specific immunity. Two patients attained complete remission (CR). One remains in CR 17 months following therapy, coincident with persistence of donor-derived tumor targeted EBV-specific CTL; the other died of non-PTLD related pathology. In the third patient, autopsy demonstrated homing of allo-CTL at the tumor site. Larger prospective studies of EBV-specific allo-CTL in PTLD are warranted.

Introduction

Iatrogenic immunosuppression is a major risk factor for the development of Epstein-Barr Virus (EBV) positive posttransplantation lymphoproliferative disorders (PTLD) following solid-organ transplantation (SOT) (1–3). Restoration of EBV-specific T-cell immunity by autologous adoptive immunotherapy (auto-CTL) is frequently associated with tumor regression (4–6). Auto-CTL production from immunosuppressed patients is time consuming, and its use not widely adopted. In vitro auto-CTL generation involves stimulation of the patient's lymphocytes with their own EBV-transformed lymphoblastoid B-cell line (LCL). Thus auto-CTL cannot be established in patients already treated with the B-cell depleting antibody rituximab, a therapy of proven value in PTLD (7). HIV-associated lymphomas are frequently EBV-positive (1). Auto-CTL generation in these patients poses risks to laboratory workers. In poststem cell transplantation PTLD, CTL generated from the stem cell donor can be efficacious (6). However, this source is unavailable to SOT recipients. A strategy which overcomes these limitations is to use partially HLA-matched unrelated allogeneic cytotoxic T-lymphocytes (allo-CTL) from healthy immunocompetent EBV-seropositive subjects that have been cryopreserved into a pregenerated CTL “bank.”. This provides immediate access, with CTL easily transported to patients irrespective of geography. Previously, we showed allo-CTL could be used to treat “early” polyclonal hyperplastic PTLD after SOT (8). Here, we extend the application to aggressive, advanced monoclonal-PTLD.

Materials and Methods

Patients

We administered allo-CTL to 3 patients (P1-3) with histologically confirmed monoclonal-PTLD. Following full clinical assessment including systemic radiological staging by computerized tomography scan, all were found to have highly aggressive, poor-risk PTLD that was relapsed or refractory to initial therapies. Patient characteristics are outlined in Table 1. Informed consent was provided. All three patients were treated under the Australian Special Access Scheme, which permits access to non-licensed medicines on compassionate grounds.

Table 1.  Patient, treatment and response characteristics of P1-3
 Gender/SOT/Time to PTLD/Age at PTLD/Donor: Recipient EBV statusPresentationECOG/Prognostic scoreAllo-CTL EBV epitope specificityCTL dates/side-effect profile/HLA matchOther therapies (with dates)EBV viral copy number/mL plasma*RadiologyOutcome
  1. Chemotherapies used: Arm ‘B’ of the HCVAD regimen consists of i.v. methotrexate (MTX) 1 g/m2 x1 and i.v. Ara-C 3 g/m2× 4. Memorial Sloan Kettering (MSK) protocol comprises five courses of i.v. MTX 3.5 g/m2 and i.v. vincristine 1.4 mg/m2 with p.o. procarbazine 500 mg/m2 for courses 1, 3 and 5 and alternate weekly intrathecal methotrexate 12 mg. All rituximab infusions were 375 mg/m2. Definitions: SOT solid organ transplant; PTLD posttransplantation lymphoproliferative disorder; ECOG Eastern Cooperative Group performance status; PCNSL primary CNS lymphomas; EBER-ISH EBV-encoded RNA in situ hybridization; LMP1 Latent membrane protein 1; BZLF1 BamHI-fragment-Z first leftward reading frame; Age-adjusted IPI Age-adjusted International prognostic index, composed of elevated serum lactate dehydrogenase (LDH), performance status (PS) and clinical stage; PCNSL PS PCNSL prognostic score composed of age more than 60 years, PS more than 1, elevated LDH, high cerebrospinal fluid protein concentration, and involvement of deep regions of the brain; IS immunosuppression, WBRT Whole-brain radiotherapy; CR Complete remission; MRI Magnetic Resonance Imaging; CT Computerized Tomography; ET Essential Thrombocythemia; DLBCL Diffuse Large B-cell Lymphoma; GvHD graft versus host disease. *For P1 and P2, no EBV-DNA was detectable immediately prior to each CTL infusion except in P2: 32 copies per ml before the fifth (#5) infusion. **Temperature rise to 37.5°C and facial flushing after 1st infusion, lasting 30 min. ***In patient 1, at initial presentation, reduced immunosuppression consisted of 5 mg of prednisolone during chemotherapy. Subsequently, immunosuppression was maintained with dexamethasone 4 mg daily and cyclosporine A (CSA) 50 mg twice daily until relapse, at which point she was switched to 5 mg of prednisolone. In patient 2, reduced immunosuppression consisted of sirolimus down from 3 mg to 1 mg daily, prednisolone from 15 mg to 5 mg and cessation of mycophenolate. In patient 3, initially reduced immunosuppression consisted of halving the dose of cyclosporine A dose (aiming to achieve levels between 50 and 100 μg/L). Prednisolone dose was continued at 20 mg daily. Later (but still prior to allo-CTL infusion) cyclosporine A was substituted for sirolimus 1 mg twice daily.

  2. ****Serial MRI's indicated an objective ongoing partial response to allo-CTL up to and including the seventh CTL, with CR1 documented following whole brain radiotherapy (WBRT) and the eighth infusion. *****Conventional chemotherapy was withheld prior to this at physician discretion, as patient judged to be intolerant of chemotherapy.

P1Female/Kidney/2 years/18 years/D+: R-.1. Polymorphic B-cell PTLD: PCNSL right cerebral hemisphere.1. ECOG 3/PCNSL PS 2EBNA3.1. No CTL given.1. Reduced IS***; #1b HCVAD; #5 MSK protocol; 45Gy WBRT. 1. MRI: CR11. CR1 of 2.5 mo.
 2. Relapse in bilateral cerebral & cerebellar hemispheres.2. ECOG 4/PCNSL PS 3 2. Days 0, +7, +14, +21/No GvHD**/3/6 match2. Reduced IS***; 1 g/m2 MTX day-9; rituximab x4, days −4, +5, +28, +35.2. Day 0 nil; #1 366; #2 844; #3 156; #4 128; days +34, +45, +62, +103, +194, +249.2. MRI: days +17: PR, +28: CR2.2. Ongoing CR2, 17mo. following 1st CTL.
P2Female/Heart-lung/3 years/52 years/D unknown: R+.Monomorphic B-cell PTLD (DLBCL): PCNSL bilateral cerebral hemispheres.ECOG 4/PCNSL PS 4LMP1; LMP2; EBNA3; BZLF1.Days 0, +8, +14, +22, +35, +42, +56, +78/No GvHD/4/6 match.Reduced IS***; 1 g/m2 MTX days -31, -18; rituximab x8, days -28, -20, -13, -6, +1, +8, +15, +22; 30 Gy WBRT day +65 to +85.Day -29 121; day 0 nil; #1 14136; #2 21152; #3 9455; #4 29518; #5 15384; #6 17202; #7 11344; #8 nil; day +79 nil; day +106 nil.MRI: days +12, +26, +57: ongoing response****; +110: CR1.Course complicated by recurrent CMV gastritis and gastric haemorrhage. Died day +113. Autopsy confirmed CR1.
P3Female/Lung/3 mo./58 years/D-: R+.Monomorphic B-cell PTLD (DLBCL): Hilar, mediastinal and bilateral lung involvement.ECOG 3/Age-adjusted IPI 3.EBNA1.Day 0 only. No GvHD/3/6 match.Reduced IS***; Ganciclovir; Rituximab x7 days -69, -62, -55, -48, -41, -34, +7; vincristine 2 mg day +7, prednisolone 100 mg days +7 to 11*****.Day 0 191703; #1 504254; day +11 1142185.CT: Progressive disease.Death from respiratory/renal failure day +11.

Blood samples

Blood was taken immediately prior to each infusion, 1 h postinfusion and then at periodic intervals following last infusion. Peripheral blood mononuclear cells (PBMC) were prepared by gradient centrifugation (Ficoll-Paque; Amersham Biosciences, Sweden). PBMC for phenotyping, functional T-cell assays and DNA extraction (viral load) were stored by controlled rate freezing in liquid nitrogen. Plasma samples were stored at −20°C.

EBV tissue staining

All assays were carried out on sections of routinely fixed, paraffin-embedded material. BZLF1, LMP1 immunohistochemistry (IHC) and EBER in situ hybridization assays were performed as previously published (9). The distribution of reactive cells was assessed by a morphologist on matching haematoxylin and eosin slides prior to enable interpretation. For IHC, following antigen retrieval with trypsin, the test antibody or isotype control was used to detect the presence or absence of LMP1 or BZLF1 (Dako, CA, USA). Positive reactivity was detected as per manufacturer's instructions incorporating DAB as the chromogenic substrate (Envision kit, Dako, CA, USA). The EBER in situ hybridization assay utilized a commercially available hybridization kit (Dako, CA, USA). A fluoresceinated poly-dT oligonucleotide probe was used to hybridize to polyA mRNA in the tissue sections to serve as a control for tissue and mRNA integrity. For all stains, a known case of EBV-positive posttransplant lymphoproliferative disorder (responsive to auto-CTL) was used as a positive control.

Quantitative real time PCR determination of the EBV-DNA load in the plasma and PBMC

Real time PCR was performed as previously described (10). PCR primers and probes selected from within the BALF5 gene encoding EBV-DNA polymerase were synthesized by PE Applied Biosystems (Foster City, CA). For each sample, assays were performed in triplicate, and for each analysis, a calibration curve was run with serially diluted EBV-DNA extracted from the cell line Namalwa and a standard curve of the CT values was constructed. The CT values from test samples were then compared to the standard curve and copy number determined automatically. Multiple no template wells were included in each assay as negative controls.

Flow cytometry

Monoclonal antibodies (mAb) conjugated to Fluorescein isothiocyanate (FITC), phycoerythrin (PE), peridinin chlorophyll protein (PerCP) and allophycocyanin (APC) were used to detect the presence of the surface markers: CD3, CD4 and CD8. Appropriate isotype controls were included. For direct visualization of MHC class I EBV-specific peptides, we incubated cells with peptide-MHC class I pentamers (Proimmune, Oxford, UK) pre-titrated concentrations at +4°C for 30 minutes. Cells were then washed in FACS buffer followed by staining for surface markers. PBMC from healthy seropositive HLA mismatched subjects were used as negative controls where relevant. Cells were acquired on a FACSCanto (Becton Dickinson) cytometer and results analyzed using FloJo software (Tree Star, Inc., Stanford, USA).

Effector T-cell assays

HLA class I viral peptide-specific effector T-cell function was determined by intracellular cytokine staining IFN-gamma as previously described (9). PBMC were incubated with EBV synthetic peptides (11) (or no peptide, or phorbol myristate acetate [50 nanograms/ml] and ionomycin [1 μg/mL] as respective negative and positive controls) at 1 μgram/mL for 1 h. 10 μg/mL Brefeldin A was then added for a further 4 h. Cells were washed with FACS buffer and fixed in 40 μL of 1% paraformaldehyde for 20 min at room temperature. Following two washes, cells were stained with IFN-gamma and CD3, CD8 surface antibodies, and PBS with 0.5% saponin at 4°C overnight. Cells were washed once before acquisition. To minimize intra-assay variability, all samples from a particular subject were generally assayed at one time. In addition, the same operator (UD) performed all the intracellular cytokine assays in this study.

Chromium release assays

Polyclonal HLA class I viral peptide-specific cytotoxic T lymphocytes (CTL) lines were established according to previously published methods (12). Briefly, autologous stimulator PBMC were pulsed with viral peptide (at 20 μg/mL) for 1 h at 37°C and then washed twice, before co-incubating with responder PBMC (responder-to-stimulator ratio of 2:1) in RPMI 1640 (Gibco Invitrogen, Carlsbad, CA) supplemented with 2 mM L-glutamine, 100 IU/mL penicillin, and 100 μg/mL streptomycin plus 10% fetal calf serum (culture medium). After three days, culture media was supplemented with interleukin-2 (IL-2) at 20 IU/mL and the cells were further expanded. These lymphocytes were restimulated on day 7 with gamma-irradiated (2000 rad) autologous peptide-labeled PBMC. Media was refreshed every 3–4 days, and IL-2 added every seven days. After 21 days of in vitro culture, cells were used as polyclonal effectors in a standard 51Cr-release assay against peptide-sensitized autologous phytohemagglutinin (PHA) blasts (12). To minimize intra-assay variability, the same operator (UD) performed all the T-cell culture and chromium release assays in this study.

Allogeneic EBV-specific CTL generation and infusion

AlloCTL were generated as previously described (13) and cryopreserved in aliquots of 20 × 106 cells per vial. EBV-specificity of the CTL line was indicated by strong lysis of autologous LCL and little or no reactivity (<10%) against autologous PHA blasts, HLA-mismatched LCL and the human erythroleukemia cell line (K562). AlloCTL were given at 2 × 106 per kilogram body weight, rounded to the nearest vial. AlloCTL were thawed, washed and then resuspended in normal saline containing 10% human albumin solution, and re-infused over 5 min. Prior to the first infusion, premedication consisted of 1 g oral Paracetamol, and 12.5 mg intravenous Phenergan. Phenergan was omitted following subsequent infusions if no reaction occurred. The protocol was for patients to initially receive four times weekly infusions, followed by clinical and radiological reassessment. Patients with residual but responsive disease were permitted a further four times weekly infusions.

Results

Despite HLA disparity, allo-CTL infusions were associated with negligible immediate toxicity. No evidence of graft versus host disease (GvHD) was observed in patient 1; however, the deaths of patients 2 and 3 prevent any conclusion as to whether GvHD may subsequently have developed (Table 1). Larger studies are required for definitive statements regarding the safety profile. Following infusion, relevant patient EBV peptide-specific T-cell responses were characterized. PTLD expresses the EBV-latency genes latent membrane proteins 1 and 2 (LMP1/2), and the EBV nuclear antigens (EBNA) 1-6 (1). LMP1 was detectable in all three patients. The early lytic antigen BamHI-fragment-Z first-leftward reading frame (BZLF1) is also frequently expressed (14) and was present in P2 but not P1 or P3. In all patients, no CD3+ CD8+ EBV-peptide-specific T cells were present prior to the first infusion, but after allo-CTL there was restoration of EBV-specific CD8+ T cells presented by a shared HLA allele. With P1, both recipient and donor shared HLA B8, which is known to present the EBNA3FLRGRAYGL peptide. HLA typing performed on cultured EBNA3FLRGRAYGL peptide-specific T cells (using irradiated recipient peptide-labeled PBMC as stimulators) from day +194 blood, demonstrated cells of donor origin. P1 was maintained on 5 mg prednisolone following CR1. The low-dose of immunosuppression may have permitted allo-CTL engraftment. CRA performed on CTL cultured from sequential blood samples, confirmed the sustained cytolytic function of EBNA3FLRGRAYGL peptide-specific T cells (Figure 1).

Figure 1.

EBV-peptide-specific CTL function following allo-CTL infusion. Patient 1. Left panel shows assays of EBV-peptide-specific CTL function at four time points (labeled 1–4) following allo-CTL infusion. Blood was sampled at day +21, 1 h following the fourth allogeneic CTL infusion (1), and then at days +62 (2), +194 (3) and +249 (4). EBV-EBNA3FLRGRAYGL peptide-stimulated polyclonal CTL were generated by three weeks culture, involving supplementation with interleukin-2, and weekly stimulation with irradiated recipient peptide-labelled PBMC. CTL were tested for cytotoxicity against recipient PHA blasts presensitized with either EBNA3FLRGRAYGL or a no peptide (1-4NP) control. A range of effector to target (E:T) ratios were use in a standard chromium-release assay. SD was <10% at all ratios. Right panel uses HLA class I EBV-specific pentamer analysis to show a population of in vitro expanded CD8+ EBNA3FLRGRAYGL peptide-specific T cells from a blood sample taken at day +249. 0.717% CD8+ EBNA3FLRGRAYGL T cells were present within the culture. By ex vivo analysis and in vitro culture, CD8+ EBNA3FLRGRAYGL peptide-specific T cells were not detectable prior to the first infusion.

Real time PCR was used to quantify EBV-DNA (10). In all patients, cellular EBV-DNA remained undetectable at all time points. This agrees with previous reports that circulating cellular viral copy number falls following rituximab, even if the tumor is progressing (15). In EBV-positive Hodgkin's Lymphoma there is evidence that plasma viral load (released from apoptotic circulating tumor cells) is a reflection of tumor load (10). On the assumption that cell-free viral DNA is similarly informative in PTLD, we performed serial measurements of plasma EBV-DNA on our patients. For P1 and 2 (with non-systemic PTLD), prior to each CTL infusion no plasma EBV-DNA was detectable in the systemic circulation. Strikingly, so long as PTLD persisted, virus was consistently present one hour postinfusion in the systemic circulation (Table 1). Detectable plasma EBV-DNA postinfusion may also indicate virus released from CTL contaminated with the EBV laboratory strain B95.8. However, PCR amplification assays using primers specific for B95.8 (5), showed that peaks were due to lysis of wild-type virus present within the tumor. In P2, the first seven infusions were associated with a spike in viral load (Figure 2). Serial MRI's indicated an objective ongoing partial response to allo-CTL up to and including the seventh CTL, with complete remission (CR) induced by whole brain radiotherapy (WBRT). Consistent with CR, virus was not detected one hour following the eighth CTL infusion, which was administered 13 days into a 20-day course of fractionated WBRT.

Figure 2.

Fluctuations in plasma EBV viral load demonstrating In-vivo efficacy of EBV-specific allo-CTL. Patient 2. Blood was sampled at the time points indicated, using primers for BamHI-fragment-A fifth leftward reading frame [7]. PCR amplification of plasma samples taken 1 h following infusion, indicated that viral load peaks were due to lysis of wild-type virus present within the tumor, and not viral contamination of CTL infusion. Serial MRI's indicated an objective ongoing partial response to allo-CTL up to and including the 7th CTL, with complete remission (CR) induced by whole brain radiotherapy (WBRT). Consistent with CR, virus was not detected 1 h following the 8th CTL infusion, which was administered 13 days into a 20-day course of fractionated WBRT.

Despite the poor prognostic outlook of primary central nervous system lymphomas (PCNSL) (16), P1 (Figure 3) and two went into CR. Owing to concurrent therapies, no definitive conclusions regarding clinical efficacy can be drawn, but certain objective observations are valid. In P1, second CR (CR2) is currently seven times the duration of CR1. P1 and two received systemic rituximab. Previous data suggests systemic rituximab has modest activity for parenchymal PCNSL, and cerebrospinal fluid levels following intravenous administration are 0.1% of serum levels associated with therapeutic efficacy (17). By contrast, activated EBV-specific T cells are able to cross the blood-brain barrier (18, 19). Although the time course of response relative to therapy, the low-dose chemotherapy administered, the restoration and persistence (of donor-derived) EBV-specific T-cell immunity and their antiviral efficacy are all suggestive that allo-CTL were significant in the induction and maintenance of remission, this series does not provide definitive proof of efficacy. A case of EBV-positive PCNSL responsive to allo-CTL has been reported in a child with primary immune deficiency (19) (i.e. in receipt of no iatrogenic immunosuppression), but P1 and two are the first reported PCNSL-PTLD's responding to allo-CTL. Although P2 had ongoing response with allo-CTL, CR was only documented once WBRT had commenced, indicating the importance of combined modality therapy, a point highlighted with auto-CTL (4). In view of recent innovations in multiviral specific CTL (20), it should be noted that P2 succumbed to complications from cytomegalovirus whilst in CR.

Figure 3.

Cerebral magnetic resonance imaging (MRI) scans. Patient 1. Left panel shows axial MRI scan showing abnormal T2 flair signal, with lesions in bilateral basal ganglia and left thalamus at relapse. These lesions were gadolinium contrast enhancing compatible with relapsed PTLD. Right panel shows T2 flair MRI images at six months following allo-CTL. The (contrast enhancing) lesions have resolved, but an abnormal signal most marked in the right anterior and posterior ventricular horns with a dilated ventricular system remains. The latter changes have remained stable over sequential scans, were present prior to relapse, and are believed to be a result of prior chemoradiotherapy.

P3 had a highly aggressive systemic PTLD that was unresponsive to antivirals, reduced immunosuppression and rituximab. She received a single infusion of allo-CTL from a male subject. One hour postinfusion, the EBV-DNA was almost treble the pre-infusion value. Although the kinetics of viral load indicate that allo-CTL induced tumor lysis, she succumbed to progressive respiratory failure secondary to lung infiltration with rapidly progressive PTLD. Figure 4 demonstrates scattered CD8+ and XY cells admixed within the recipient's B-cell PTLD. To our knowledge, this is the first report of tumor infiltration by allo-CTL.

Figure 4.

Tumor Homing of EBV-specific allo-CTL. Patient 3. Immunohistochemistry (IHC) and florescent in situ hybridization (FISH) in left and right panels respectively demonstrating scattered CD8+ and XY cells admixed within the B-cell PTLD, taken at autopsy. The allo-CTL donor was male, and the recipient was female. The PTLD was of recipient origin. The IHC image was visualized with an Olympus CX41 at ×40 objective, taken with a Nikon Coolpix 5700, and analyzed with Microsoft Photo Editor. FISH was performed using the SRY/CEP X dual colour commercial probe (Vysis, Ilinois, USA).

We demonstrate safety, immunity, homing and efficacy of allo-CTL for aggressive monoclonal-PTLD after SOT. Although regulatory obstacles regarding cellular therapies will need to be addressed, larger prospective studies of allo-CTL are warranted.

Acknowledgments

The study was supported by the Queensland Department of State Development and Innovation, and Cancer Research United Kingdom. (M.K. Gandhi has a National Health and Medical Research Council (Australia) Clinical Career Development Award.) We thank Sue Godwin and Tanya Graham for their kind help in coordinating the clinical aspects of this study, Geraldine Bollard for assistance with the CTL infusion (P3), Barbara Savoldo for supplying the B95.8 primer sequence, Thomas Robertson and Tracey Scott for the help with the SRY probe, and Prof Rajiv Khanna for critically reading the manuscript.

Ancillary