Epstein–Barr virus (EBV) is associated with several malignancies including nasopharyngeal carcinoma, a high incidence tumor in Chinese populations, in which tumor cells express the two EBV antigens EB nuclear antigen 1 (EBNA1) and latent membrane protein 2 (LMP2). Here, we report the phase I trial of a recombinant vaccinia virus, MVA-EL, which encodes an EBNA1/LMP2 fusion protein designed to boost T-cell immunity to these antigens. The vaccine was delivered to Hong Kong patients with nasopharyngeal carcinoma to determine a safe and immunogenic dose. The patients, all in remission more than 12 weeks after primary therapy, received three intradermal MVA-EL vaccinations at three weekly intervals, using five escalating dose levels between 5 × 107 and 5 × 108 plaque-forming unit (pfu). Blood samples were taken during prescreening, immediately before vaccination, one week afterward and at intervals up to one year later. Immunogenicity was tested by IFN-γ ELIspot assays using complete EBNA1 and LMP2 15-mer peptide mixes and known epitope peptides relevant to patient MHC type. Eighteen patients were treated, three per dose level one to four and six at the highest dose, without dose-limiting toxicity. T-cell responses to one or both vaccine antigens were increased in 15 of 18 patients and, in many cases, were mapped to known CD4 and CD8 epitopes in EBNA1 and/or LMP2. The range of these responses suggested a direct relationship with vaccine dose, with all six patients at the highest dose level giving strong EBNA1/LMP2 responses. We concluded that MVA-EL is both safe and immunogenic, allowing the highest dose to be forwarded to phase II studies examining clinical benefit. Cancer Res; 73(6); 1676–88. ©2012 AACR.

Epstein–Barr virus (EBV) persists in most individuals as a life-long asymptomatic infection, with both lytic virus replication in the oropharynx and latent growth-transforming infections in the B-lymphoid system being kept in check by immune T-cell surveillance (1). Despite the usually asymptomatic nature of EBV carriage, the virus has potent growth-transforming ability in vitro and oncogenic potential in vivo, causing fatal B-lymphoproliferative disease lesions in the immunosuppressed and being strongly linked to several malignancies; these include endemic Burkitt lymphoma, a subset of Hodgkin lymphomas, certain aggressive T/NK cell lymphomas, and an epithelial tumor, nasopharyngeal carcinoma (NPC; ref. 2). Of these, nasopharyngeal carcinoma is the most important in world health terms, occurring worldwide but at particularly high incidence throughout South East Asia, especially among Chinese people (3); thus, the Hong Kong Cancer Registry 2009 data report a male age-standardized incidence of 14.6/100,000, ranking nasopharyngeal carcinoma the 6th most common cancer in men in that region. Despite radiotherapy and (in meta-analyses) additional chemotherapy improving outcomes, 2- and 5-year disease-free survival rates are still only 63% and 52%, respectively, and distant metastases account for more than 40% of recurrences (4), at which point the prognosis is very poor (5, 6). New treatment modalities are therefore needed in both the adjuvant and salvage settings (7), particularly approaches designed for populations at highest risk.

The consistent presence of EBV in all nasopharyngeal carcinoma cells opens up the possibility of an immunotherapeutic approach, exploiting the potential of the immune T-cell system to recognize tumor cells through their expression of viral antigens. In that regard, most nasopharyngeal carcinomas express just 2 latent cycle proteins, the EB nuclear antigen 1 (EBNA1), a sequence-specific DNA-binding protein involved in maintenance of the episomal virus genome, and latent membrane protein 2 (LMP2), a membrane-signaling protein with growth-promoting activity in epithelial cells (2, 8, 9). From the study of healthy individuals, these proteins do not constitute dominant targets of the T-cell response induced by natural EBV infection; nevertheless, they do contain a number of subdominant CD8 target epitopes (mainly in LMP2) and several CD4 epitopes (mainly in EBNA1), presented by individual HLA I and HLA II alleles, respectively (1). Indeed, there is already some preliminary evidence both from a dendritic cell–based vaccine trial (10) and from subsequent adoptive T-cell therapy trials (11–15) that boosting CD8 immunity to LMP2 epitopes may be of clinical benefit in nasopharyngeal carcinoma. The present approach is distinct from the above and is based on vaccination with a modified vaccinia Ankara (MVA) recombinant vector expressing the tumor-associated viral antigens; unlike the above cell-based protocols, such a procedure is not limited to specialized centers but has the potential for delivery on a large scale. The vaccine virus, MVA-EL, was constructed using sequences cloned from a typical Chinese EBV strain and encodes a functionally inactive fusion protein containing the C-terminal half of EBNA1 and full-length LMP2A (16). Note that the C-terminal EBNA1 fragment includes almost all known epitopes in this protein (1) but lacks the gly-ala repeat domain that partly impairs presentation of cis-linked sequences to CD8+ T cells (17–19).

To date MVA or similar poxvirus-based therapeutic vaccines expressing cellular target antigens have been tested in certain nonvirus-associated cancers; these trials almost all involve Caucasian patient cohorts and most focus only on CD8+ T cell responses, indeed often on responses restricted through just one common Caucasian HLA I allele, HLA-A*02:01 (20, 21). In contrast, the present trial has several distinct features. First, the vaccine antigens are viral proteins against which patients were likely to have some (albeit low) preexisting immunity (22, 23) Second, the analysis focuses on both CD4+ and CD8+ T cell responses to these antigens, as both arms of the T-cell system are likely to be important for effective tumor control in the longer term (24–26). Third, the vaccine is specifically designed for use in the high-risk Chinese population, with its own distinctive array of HLA I and HLA II alleles that will determine target epitope choice (23, 27–29). Thus, testing MVA-EL in patients with nasopharyngeal carcinoma is important not only because it addresses an urgent clinical need but also because it has the potential to provide lessons of general relevance for tumor vaccination strategies.

Here, we report the results of a first trial of MVA-EL vaccination in Chinese patients with nasopharyngeal carcinoma, focusing on patients who were in remission following standard therapy. The objectives were first to determine tolerability of the vaccine across a range of doses, and second to determine its capacity to induce EBNA1- and/or LMP2-specific T-cell responses, using natural fluctuations in immunity against irrelevant viral targets as an internal control. In the absence of safety concerns, the scale, specificity, and diversity of vaccine responses were the critical factors determining dose selection and whether to proceed to further investigation.

Patients

This was a phase I dose-escalation trial of MVA-EL vaccination for patients with EBV-associated nasopharyngeal carcinoma. Eligible patients had histologically poorly differentiated nasopharyngeal carcinoma, either confirmed as EBV-positive by in situ hybridization for the noncoding EBER RNAs or with characteristically raised serum immunoglobulin A (IgA) to EBV capsid antigen. All patients 12 weeks or more following completion of first-line treatment, were in remission and satisfied the following inclusion criteria: 18 years or more; free from toxicities >grade 1; using adequate birth control; performance status 0 to 1; life expectancy more than 4 months; alkaline phosphatase and alanine aminotransferase (ALT) <1.5 × ULN; creatinine clearance >50 mL/min, hemoglobin >10.0 g/dL; lymphocytes >1.0 × 109/L; neutrophils ≥1.5 × 109/L; platelets ≥100 × 109/L; lacking known active hepatitis B, hepatitis C, or HIV infection, autoimmune or skin disease requiring therapy, active infection, severe egg allergy, splenic dysfunction, previous myeloablative, or current immunosuppressive therapy.

The protocol (clinicaltrials.gov NCT01256853) was approved by the Institutional Review Board. Conduct was ICH-GCP-compliant and all patients gave written informed consent.

Procedures

Patients received 3 intradermal vaccinations of MVA-EL (Impfstoffwerk Dessau-Tornau) delivered at 3-week intervals. Sequential cohorts of 3 patients received doses of either 5 × 107, 1 × 108, 2 × 108, or 3.3 × 108 plaque-forming units (pfu) per vaccination (dose levels 1–4, respectively). To gain additional immune data, the highest dose cohort, dose level 05 receiving 5 × 108 pfu per vaccination, comprised 6 patients. Throughout the text, patients are identified by the code xxyy, in which xx is the vaccine dose level from 01 to 05 and yy is the patient number from 01 to 18. Toxicity was graded according to the National Cancer Institute Common Terminology Criteria for Adverse Events (CTCAE) Version 3.0 and using a protocol specific grading of local injection site reactions. Blood was collected at prescreening, immediately before each vaccination (day1; cycle 1, 2, or 3), 7 days after each vaccination (day 8; cycle 1, 2, or 3), and at weeks 10, 11, 14, and 6 and 12 months following the start of the vaccine course. Peripheral blood mononuclear cell (PBMC) and plasma samples were cryopreserved on each occasion, and T-cell response assays on an individual patient were conducted on simultaneously thawed PBMC samples. ELIspot assays of T-cell responses were conducted either against overlapping 15-mer peptides pools (pepmixes) spanning an antigen's primary sequence or against defined epitopes peptides. Antigen-specific recognition was defined as numbers of spot-forming cells (sfc) against a viral antigen ≥10 sfc/well and ≥2-fold greater than that against a negative control (actin) pepmix, epitope-specific recognition as a reading ≥10 sfc/well and ≥2-fold greater than against the dimethyl sulfoxide (DMSO) solvent control. Mean spot counts for negative control wells (actin or DMSO) were subtracted from those for test wells (viral antigen pepmix or peptides, respectively) to generate mean adjusted readings. Vaccine responses were defined as antigen or epitope recognition for which the adjusted counts postvaccine cycle 2 or 3 were ≥2-fold higher than for prevaccination samples. EBV DNA loads in plasma were assayed by quantitative PCR (qPCR); EBNA1 and vaccinia-specific antibodies were measured before and after vaccination using standard assays. Full details of all procedures are provided in Supplementary Data.

Design and statistical analysis

The primary objectives were (i) to determine safety and to characterize the toxicity profile of MVA-EL vaccine, and (ii) to describe changes in the frequency of functional T-cell responses to EBNA1 and LMP2. The secondary objective was to assess changes in levels of EBV genomes in plasma.

In all ELIspot assays, criteria to define antigen recognition and vaccine response were predetermined and systematically applied. Exploratory analysis of pepmix assay readings was undertaken across all patients using repeated measures ANOVA to test whether mean adjusted readings at the 3 time points differed significantly. For antigens in which mean adjusted readings at the 3 time points were significantly different, adjusted readings at each later time point were compared with those for prevaccination samples using the Dunnett multiple comparison test. The relationship between size of response and vaccine dose was analyzed by linear regression. All statistical testing was undertaken using GraphPad Prism 5.

Table 1 gives the clinical history of the 18 Chinese patients with nasopharyngeal carcinoma (15 male and 3 female), who participated in the trial between November 2006 and April 2009. All had received prior radiotherapy and 14 of 18 also chemotherapy for their disease. The median duration from last treatment to first vaccination was 20 weeks (range, 14–42 weeks) and all patients were clinically disease-free on entry into the trial. Each of these 18 individuals completed their 3 vaccinations at the planned dose and the trial proceeded through the 5 levels of increasing vaccine dose (5 × 107 to 5 × 108 pfu/vaccination) with no evidence of dose-limiting toxicity. As detailed in Supplementary Table S1, related adverse events mainly involved injection site reactions, seen at grade 1 in most patients but extending to grade 2 in 2 cases, and transiently to grade 3 in 1 individual. Fatigue, flu-like symptoms, and arthralgia were observed occasionally at all vaccine doses, whereas myalgia was only reported for patients on dose level 4 to 5.

Table 1.

Patients, diagnoses, and previous treatments

Patient No.aAge, ySexStagebRadiotherapyChemotherapycTime since treatment, wks
0101 41 I (T1N0M0) 6,600 cGy in 33 fractions, (NP boost by intubation) 1,800 cGy in 4 fractions NA 15 
0102 57 II (T2bN0M0) 6,600 cGy in 33 fractions, (right PP boost) 1,400 cGy in 7 fractions NA 16 
0103 55 III (T2bN2M0) 6,600 cGy in 33 fractions, (bilateral PP boost) 1,000 cGy in 5 fractions, (left PP boost) 400 cGy in 2 fractions Cisplatin (6 cycles) 15 
0204 59 II (T2bN1M0) 6,600 cGy in 33 fractions, (NP boost) 800 cGy in 4 fractions NA 22 
0205 46 III (T2bN2M0) 6,600 cGy in 33 fractions, (NP boost) 800 cGy in 4 fractions Cisplatin (6 cycles) 19 
0206 46 IV (T2bN3M0) 6,600 cGy in 33 fractions, (left PP boost) 1,400 cGy in 7 fractions Cisplatin (6 cycles) 27 
0307 53 III (T3N2M0) 6,600 cGy in 33 fractions, (NP boost) 4 fractions 800 cGy Cisplatin (7 cycles) 30 
0308 40 III (T2bN2M0) 6,600 cGy in 33 fractions, (NP boost) 800 cGy in 4 fractions, (LN boost) 750 cGy in 2 fractions Cisplatin (7 cycles) 33 
0309 47 IV (T4N1M0) 7,000 cGy in 35 fractions Induction carboplatin+paclitaxel (2 cycles), cisplatin (7 cycles) 22 
0410 50 IV (T4N0M0) 7,000 cGy in 35 fractions Induction cisplatin +gemcitabine (2 cycles), cisplatin (6 cycles) 39 
0411 57 III (T3N0M0) 7,000 cGy in 35 fractions Cisplatin (6 cycles) 17 
0412 57 III (T3N0 M0) 7,000 cGy in 35 fractions Cisplatin (4 cycles) 14 
0513 62 II (T2bN1M0) 7,000 cGy in 35 fractions Cisplatin (5 cycles) 14 
0514 42 IV (T4N1M0) 7,000 cGy in 35 fractions, (LN boost) 750 cGy in 2 fractions Cisplatin (6 cycles) 15 
0515 55 II (T2bN1M0) 7,000 cGy in 35 fractions Cisplatin (6 cycles) 21 
0516 36 II (T2bN0M0) 7,000 cGy in 35 fractions NA 18 
0517 63 IV (T4N1M0) 7,000 cGy in 35 fractions Cisplatin (6 cycles) 42 
0518 55 II (T2bN1M0) 7,000 cGy in 35 fractions Cisplatin (6 cycles) 27 
Patient No.aAge, ySexStagebRadiotherapyChemotherapycTime since treatment, wks
0101 41 I (T1N0M0) 6,600 cGy in 33 fractions, (NP boost by intubation) 1,800 cGy in 4 fractions NA 15 
0102 57 II (T2bN0M0) 6,600 cGy in 33 fractions, (right PP boost) 1,400 cGy in 7 fractions NA 16 
0103 55 III (T2bN2M0) 6,600 cGy in 33 fractions, (bilateral PP boost) 1,000 cGy in 5 fractions, (left PP boost) 400 cGy in 2 fractions Cisplatin (6 cycles) 15 
0204 59 II (T2bN1M0) 6,600 cGy in 33 fractions, (NP boost) 800 cGy in 4 fractions NA 22 
0205 46 III (T2bN2M0) 6,600 cGy in 33 fractions, (NP boost) 800 cGy in 4 fractions Cisplatin (6 cycles) 19 
0206 46 IV (T2bN3M0) 6,600 cGy in 33 fractions, (left PP boost) 1,400 cGy in 7 fractions Cisplatin (6 cycles) 27 
0307 53 III (T3N2M0) 6,600 cGy in 33 fractions, (NP boost) 4 fractions 800 cGy Cisplatin (7 cycles) 30 
0308 40 III (T2bN2M0) 6,600 cGy in 33 fractions, (NP boost) 800 cGy in 4 fractions, (LN boost) 750 cGy in 2 fractions Cisplatin (7 cycles) 33 
0309 47 IV (T4N1M0) 7,000 cGy in 35 fractions Induction carboplatin+paclitaxel (2 cycles), cisplatin (7 cycles) 22 
0410 50 IV (T4N0M0) 7,000 cGy in 35 fractions Induction cisplatin +gemcitabine (2 cycles), cisplatin (6 cycles) 39 
0411 57 III (T3N0M0) 7,000 cGy in 35 fractions Cisplatin (6 cycles) 17 
0412 57 III (T3N0 M0) 7,000 cGy in 35 fractions Cisplatin (4 cycles) 14 
0513 62 II (T2bN1M0) 7,000 cGy in 35 fractions Cisplatin (5 cycles) 14 
0514 42 IV (T4N1M0) 7,000 cGy in 35 fractions, (LN boost) 750 cGy in 2 fractions Cisplatin (6 cycles) 15 
0515 55 II (T2bN1M0) 7,000 cGy in 35 fractions Cisplatin (6 cycles) 21 
0516 36 II (T2bN0M0) 7,000 cGy in 35 fractions NA 18 
0517 63 IV (T4N1M0) 7,000 cGy in 35 fractions Cisplatin (6 cycles) 42 
0518 55 II (T2bN1M0) 7,000 cGy in 35 fractions Cisplatin (6 cycles) 27 

Abbreviations: NP, nasopharynx; PP, parapharynx; LN, lymph node.

aPatient identifiers are denoted by the prefix xxyy, in which xx is the dose level from 01 to 05 and yy is the patient number from 01 to 18.

bAccording to Unio Internationale Contra Cancrum/American Joint Committee on Cancer (UICC/AJCC) 1997 cancer staging manual.

cCisplatin 40 mg/m2 was given weekly concurrently with radiotherapy (except as induction chemotherapy).

The vaccine-coded EBNA1/LMP2 fusion protein, shown diagrammatically in Fig. 1A, contains a 280-amino acid sequence from the EBNA1 C-terminus and the full 497 amino acids of LMP2A. To look for vaccine responses, PBMC samples cryopreserved from individual patients before vaccination, 1 week after the second vaccination (cycle 2; day 8) and within 4 weeks after the third vaccination (cycle 3; usually day 22) were tested in ELIspot assays against separate pepmixes encompassing the primary sequences of EBNA1 and LMP2. As controls, these assays tested the same PBMCs against separate pepmixes covering the sequences of actin (a self-antigen used to assess assay background), of EBNA3A (an EBV latent protein not encoded by the vaccine) and of the influenza matrix and nucleoproteins (FLU, an unrelated virus against which patients are expected to have preexisting immunity). Note that these assays on total PBMCs detect the sum of both CD4+ and CD8+ T-cell responses to epitope peptides within each pepmix. Figure 1B shows examples of the results. Patients 0516 and 0410 were typical of many vaccinees in that the prevaccine sample revealed low but detectable preexisting memory responses to both EBNA1 and LMP2, as well as responses to EBNA3A and FLU. Importantly, the responses to EBNA1 and LMP2 were clearly increased after the second and third vaccinations, whereas those to the control antigens were not. In contrast, patient 0206 was one of a minority in which we were unable to detect any significant prevaccine response to EBNA1 or LMP2 (nor, in this case, to EBNA3A). Here, vaccination again induced a response to both EBNA1 and LMP2 but not to the EBNA3A control; interestingly, this patient also provided a rare example in which the preexisting FLU response was coincidentally raised in 1 of 2 postvaccine samples.

Figure 1.

T-cell responses in patients with nasopharyngeal carcinoma stimulated by MVA-EL vaccination. A, design of the EBNA1 LMP2 chimeric gene insert in the MVA-EL vaccine. DNA encoding the carboxy terminal half of EBNA1 (dark gray) was fused to the full-length LMP2 gene (light gray) to generate the fusion gene inserted into MVA. Locations of defined CD8 and CD4 T-cell epitopes studied in ELIspot assays are shown as black or white lines, respectively. Two tyrosine residues within the LMP2 part of the fusion were mutated to phenylalanine to abrogate LMP2 signaling function. B, representative results from three patients using the whole-antigen pepmix ELIspot assay. PBMC samples taken at three different time points (screening, C2D8, and C3D22) were each stimulated with EBNA1 and LMP2 pepmixes in triplicate wells. Actin pepmix served as the negative control indicating background T-cell activity in the assay. Cells were also tested with EBNA3A and influenza pepmixes; these antigens are not present in the MVA-EBNA1/LMP2 vaccine and therefore measure nonspecific amplification of T-cell responses. In all three patients, the T-cell response to EBNA1 and LMP2 is increased at C2D8 following 2 cycles of vaccine. These responses increased further at C3D22 for patients receiving lower doses of vaccine (patients 0206 and 0410), whereas no further amplification was seen for the patient receiving the highest dose (patient 0513). Note that for patient 0206, the influenza response is increased at C3D22 but is unchanged at the earlier time point, possibly indicating seasonal infection with influenza. Across the trial, only two patients had altered influenza T-cell response compared with 15 patients with increased EBNA1/LMP2 responses.

Figure 1.

T-cell responses in patients with nasopharyngeal carcinoma stimulated by MVA-EL vaccination. A, design of the EBNA1 LMP2 chimeric gene insert in the MVA-EL vaccine. DNA encoding the carboxy terminal half of EBNA1 (dark gray) was fused to the full-length LMP2 gene (light gray) to generate the fusion gene inserted into MVA. Locations of defined CD8 and CD4 T-cell epitopes studied in ELIspot assays are shown as black or white lines, respectively. Two tyrosine residues within the LMP2 part of the fusion were mutated to phenylalanine to abrogate LMP2 signaling function. B, representative results from three patients using the whole-antigen pepmix ELIspot assay. PBMC samples taken at three different time points (screening, C2D8, and C3D22) were each stimulated with EBNA1 and LMP2 pepmixes in triplicate wells. Actin pepmix served as the negative control indicating background T-cell activity in the assay. Cells were also tested with EBNA3A and influenza pepmixes; these antigens are not present in the MVA-EBNA1/LMP2 vaccine and therefore measure nonspecific amplification of T-cell responses. In all three patients, the T-cell response to EBNA1 and LMP2 is increased at C2D8 following 2 cycles of vaccine. These responses increased further at C3D22 for patients receiving lower doses of vaccine (patients 0206 and 0410), whereas no further amplification was seen for the patient receiving the highest dose (patient 0513). Note that for patient 0206, the influenza response is increased at C3D22 but is unchanged at the earlier time point, possibly indicating seasonal infection with influenza. Across the trial, only two patients had altered influenza T-cell response compared with 15 patients with increased EBNA1/LMP2 responses.

Close modal

The numbers of patients showing significant increases in responsiveness following vaccination at the different dose levels are shown in Table 2. Responses to EBNA1 were seen in 12 patients after 2 and 3 vaccine cycles, whereas LMP2 responses were seen in 10 patients after 2 and in 9 patients after 3 cycles. In contrast, only 4 of 18 patients showed a coincident increase in responses to one or other of the control antigens (EBNA3A or FLU) and in all but one case the increase was only seen in 1 of 2 postvaccine samples. This clear difference between vaccine-coded and control antigen responses was observed despite the fact that, before vaccination (see hatched columns; Supplementary Fig. S1), the incidence of patients with low-level preexisting immunity to EBNA1 (15 of 18) and LMP2 (12 of 18) was not dissimilar to the numbers with preexisting immunity to EBNA3A (9 of 18) or FLU (14 of 18). The individual pepmix assay results from all 18 patients are shown as histograms in Supplementary Fig. S1, those responses that are significantly raised above prevaccine levels after 2 (gray columns) or 3 (black columns) being identified by asterisks. Vaccination was associated with increased responses to both EBNA1 and LMP2 in 11 patients, to EBNA1 only in 3 patients, and to LMP2 only in 1 case. Overall, therefore, 15 of 18 patients responded to 1 or other vaccine-coded antigen, in most cases with responses being apparent both after 2 and after 3 vaccine cycles.

Table 2.

Numbers of patients with ≥2-fold increases in antigen-specific responders across vaccination according to vaccine dose level

Following cycle 2 vs. prevaccinationFollowing cycle 3 vs. prevaccination
EBNA1LMP2EBNA3AFLUEBNA1LMP2EBNA3AFLU
2-Fold amplification of immune response Dose level 1 1/3 0/3 0/3 0/3 1/3 1/3 0/3 0/3 
 Dose level 2 1/3 1/3 0/3 0/3 2/3 1/3 0/3 1/3 
 Dose level 3 3/3 1/3 0/3 0/3 2/3 1/3 0/3 0/3 
 Dose level 4 1/3 3/3 0/3 0/3 2/3 2/3 0/3 0/3 
 Dose level 5 6/6 5/6 0/6 1/6 5/6 4/6 2/6 1/6 
 All patients 12/18 10/18 0/18 1/18 12/18 9/18 2/18 2/18 
Following cycle 2 vs. prevaccinationFollowing cycle 3 vs. prevaccination
EBNA1LMP2EBNA3AFLUEBNA1LMP2EBNA3AFLU
2-Fold amplification of immune response Dose level 1 1/3 0/3 0/3 0/3 1/3 1/3 0/3 0/3 
 Dose level 2 1/3 1/3 0/3 0/3 2/3 1/3 0/3 1/3 
 Dose level 3 3/3 1/3 0/3 0/3 2/3 1/3 0/3 0/3 
 Dose level 4 1/3 3/3 0/3 0/3 2/3 2/3 0/3 0/3 
 Dose level 5 6/6 5/6 0/6 1/6 5/6 4/6 2/6 1/6 
 All patients 12/18 10/18 0/18 1/18 12/18 9/18 2/18 2/18 

NOTE: Cryopreserved PBMC from three time points were thawed and tested in ELIspot assays against overlapping peptides covering the whole sequence of actin, EBNA1, LMP2, EBNA 3, and influenza antigen (FLU) at 3 × 105/well. Adjusted spot counts were the spot counts for the viral antigens minus the spot count for actin. Postvaccination, adjusted readings ≥2-fold higher than for prevaccination samples defined an immune response.

About the absolute size of these vaccine-induced changes, Fig. 2A combines data from all 18 patients and shows, for each pepmix antigen, the mean difference [±95% confidence intervals (CI)] in ELIspot counts per well after 2 and 3 vaccine cycles and the corresponding prevaccine value. Clearly, there were statistically significant increases in reactivity to the vaccine-coded EBNA1 and LMP2 antigens but no significant change in reactivity to the control EBNA3A and FLU antigens. It is also instructive to express these increases relative to the preexisting levels of EBNA1-specific and LMP2-specific immunity seen in the prevaccine pepmix counts. Again, combining data from all 18 patients, reactivity to EBNA1 was increased by means of 3.5-fold (range, 0.0–8.8) and 3.2-fold (0.5 to 8.2) after 2 and 3 vaccine cycles respectively, and reactivity to LMP2 by means of 3.7-fold (0.0–10.5) and 3.9-fold (0.4–16.6) over the same period. It is also apparent from Table 2 that the proportion of patients mounting significant responses to EBNA1 and/or LMP2 was greater at dose level 5 than at the lower dose levels 1 to 4, implying a dose–response relationship. This was further explored by plotting response size (the difference in adjusted readings between pre- and postvaccination levels) against vaccine dose. The results, shown in Fig. 2B, indeed suggested a linear relationship for both EBNA1 and LMP2 responses following cycle 2 and for LMP2 responses following cycle 3 (Fig. 2B).

Figure 2.

Summary of whole-antigen pepmix responses for all patients in the trial. A, the mean change in ELIspot readings for all patients that occurred after two (diamond) or three (cross) cycles of vaccination are shown along with the 95% CIs. Results are expressed as sfc/3 × 105 PBMC. First, a repeated measures ANOVA was used to test whether the mean pepmix ELIspot readings (adjusted by subtracting background counts against actin) at the three time points were significantly different from each other: EBNA 1, P = 0.0001; LMP2, P = 0.002; EBNA 3A, P = 0.40; FLU, P = 0.27. Using the Dunnett multiple comparison test, statistically significant increases (P < 0.05 as shown) were observed for vaccine-encoded antigens EBNA1 and LMP2 but not for control antigens EBNA3A or FLU. B, higher vaccine doses elicit stronger EBNA1 and LMP2 T-cell responses. Responses obtained in whole-antigen pepmix assays from all 18 vaccinated patients are plotted against vaccine dose. The change in the adjusted spot count/3 × 105 PBMC in the prevaccination samples versus the samples after cycle 2 and 3, respectively, is plotted against the vaccine dose given in pfu. The linear regression lines with 95% CIs are shown with an estimate of goodness of fit of the line to the data (R2) and the P value testing the null hypothesis that the slope of the line does not differ significantly from 0.

Figure 2.

Summary of whole-antigen pepmix responses for all patients in the trial. A, the mean change in ELIspot readings for all patients that occurred after two (diamond) or three (cross) cycles of vaccination are shown along with the 95% CIs. Results are expressed as sfc/3 × 105 PBMC. First, a repeated measures ANOVA was used to test whether the mean pepmix ELIspot readings (adjusted by subtracting background counts against actin) at the three time points were significantly different from each other: EBNA 1, P = 0.0001; LMP2, P = 0.002; EBNA 3A, P = 0.40; FLU, P = 0.27. Using the Dunnett multiple comparison test, statistically significant increases (P < 0.05 as shown) were observed for vaccine-encoded antigens EBNA1 and LMP2 but not for control antigens EBNA3A or FLU. B, higher vaccine doses elicit stronger EBNA1 and LMP2 T-cell responses. Responses obtained in whole-antigen pepmix assays from all 18 vaccinated patients are plotted against vaccine dose. The change in the adjusted spot count/3 × 105 PBMC in the prevaccination samples versus the samples after cycle 2 and 3, respectively, is plotted against the vaccine dose given in pfu. The linear regression lines with 95% CIs are shown with an estimate of goodness of fit of the line to the data (R2) and the P value testing the null hypothesis that the slope of the line does not differ significantly from 0.

Close modal

While vaccine dose clearly seemed to influence both the incidence and size of responses to the MVA-EL vaccine, we considered 2 other possible factors that could be involved. The first factor was regulatory T cells (Treg). Multiparametric flow-cytometric analysis of PBMCs from 6 representative patients showed the vaccine-induced increases in EBNA1- and/or LMP2-specific T-cell responses seen in ELIspot assays occurred in patients regardless of whether Treg frequencies were normal or elevated in their peripheral blood (Supplementary Fig. S2). These data also showed that for 3 of 6 patients Treg frequency was unchanged by vaccination; a small increase was observed in 2 patients and a small decrease occurred in another. Focusing on 2 patients, 0517 and 0518, for whom long-term (6 month) immune data was available and who received the same vaccine dose, we noted it was patient 0518, who had the lower frequency of Tregs, in whom the EBNA1- and LMP2-specific T-cell responses were sustained. Indeed, the EBNA1-specific response in this patient was readily detectable 6 months after vaccination (Supplementary Table S3). The second factor was the prior status of the patient with respect to vaccinia exposure from smallpox vaccination. We therefore screened all 18 patients and identified 4 individuals who were negative for antivaccinia antibodies; these included patients receiving the MVA-EL vaccine across a range of doses from level 1 (patient 0101), through 3 (patient 0308) to level 5 (patients 0513 and 0514). Figure 3A shows the size of vaccine-induced T-cell responses to EBNA1 and LMP2 pepmixes in these 4 patients versus the 14 patients with serologic evidence of prior vaccinia exposure. No significant differences were apparent. Note that nonreplicating MVA-based vaccines such as MVA-EL are primarily designed to induce T cell rather than antibody responses to the vaccine-coded antigen; nevertheless, such vaccines would be expected to induce a serologic response to structural proteins of the MVA virion. Indeed, we confirmed that antivaccinia antibodies were induced in the 4 previously vector-naïve recipients and were boosted in those with preexisting immunity (data not shown). Interestingly, as shown in Fig. 3B, we also noted that the mean anti-EBNA1 antibody titer in the vaccine recipients was raised slightly above the prevaccine mean, although the increase only achieved statistical significance in the sample taken soon after completing the vaccine course, and not 6 or 12 months later. Again, there was no apparent difference in the antibody response to EBNA1 in vaccinia-naïve and vaccinia-immune individuals.

Figure 3.

Preexisting immunity to the vector does not affect the size of immune response to the vaccine-coded EBV antigens. A, the difference between whole-antigen pepmix response measured by ELIspot after two cycles of vaccination and that, at prevaccination, is plotted against seropositivity for vaccinia-specific immunoglobulin before MVA-EL vaccination. The means are shown as horizontal bars and are expressed as sfc/3 × 105 PBMC. B, EBNA1 antibody responses before and after MVA-EL vaccination. Using data from all patients, prevaccination anti-EBNA1 IgG titers were compared with those measured using samples from C3 D29, 6 and 12 months postvaccination. Patients who were vaccinia seronegative before MVA-EL vaccination are represented by open symbols. The mean values (horizontal bars) were significantly different (P = 0.025 repeated measures ANOVA). Compared with the prevaccination control sample, the mean difference in reading was significant at the 5% level only for the samples taken at C3 D29 but not at 6 and 12 months (the Dunnett multiple comparison test).

Figure 3.

Preexisting immunity to the vector does not affect the size of immune response to the vaccine-coded EBV antigens. A, the difference between whole-antigen pepmix response measured by ELIspot after two cycles of vaccination and that, at prevaccination, is plotted against seropositivity for vaccinia-specific immunoglobulin before MVA-EL vaccination. The means are shown as horizontal bars and are expressed as sfc/3 × 105 PBMC. B, EBNA1 antibody responses before and after MVA-EL vaccination. Using data from all patients, prevaccination anti-EBNA1 IgG titers were compared with those measured using samples from C3 D29, 6 and 12 months postvaccination. Patients who were vaccinia seronegative before MVA-EL vaccination are represented by open symbols. The mean values (horizontal bars) were significantly different (P = 0.025 repeated measures ANOVA). Compared with the prevaccination control sample, the mean difference in reading was significant at the 5% level only for the samples taken at C3 D29 but not at 6 and 12 months (the Dunnett multiple comparison test).

Close modal

We then sought to further analyze the vaccine-induced T-cell responses seen in the pepmix assays. Previous work has identified a number of peptide epitopes within the primary sequences of EBNA1 and LMP2 recognized by CD4+ and CD8+ T cells (1, 23, 27–29). Figure 1A shows the location of 6 known CD4 epitopes and 12 known CD8 epitopes lying within the EBNA1-LMP2 vaccine antigen; their coordinates, amino acid sequences, and where available, their HLA restrictions are detailed in Supplementary Table S2. Additional PBMC aliquots were tested in ELIspot assays against epitope peptides relevant to their HLA I and HLA II types, either using the peptides individually, or where cell numbers were limiting, against small pools of relevant CD4 or CD8 epitopes.

Figure 4A shows the epitope-specific ELIspot assay plates from one such patient, 0206 (HLA-A*11:01, A*02:03 and HLA-DP5-positive), who had already shown evidence of vaccine-induced EBNA1 and LMP2 responses in the pepmix assays (see Fig. 1 and Supplementary Fig. S1). In line with that LMP2 pepmix response, this patient gave evidence of responses to 2 known LMP2-derived CD8 epitopes, SSC and LLS, that are restricted by the A*11:01 and A*02:03 alleles, respectively; likewise, in line with the EBNA1 pepmix response, this patient responded to a CD8 epitope, VLK/A*02:03, and a CD4 epitope, VFLQ/DP5, both derived from EBNA1. These results, and corresponding data from epitope peptide screening on 2 further vaccine recipients, patients 0513 and 0516, are shown as graphs in Fig. 4B, below histograms of the same patients' pepmix responses. Patients 0513 and 0516 had been identified as responding to LMP2 in pepmix assays; both individuals were HLA-A*24:02–positive and indeed both responded to 2 LMP2-derived CD8 epitopes restricted by this allele, PYL and TYG. Likewise, both patients had shown an EBNA1 pepmix response and epitope peptide screening confirmed that both responded to one or more relevant EBNA1-derived CD4 epitopes.

Figure 4.

MVA-EL stimulates CD8 and CD4 immune responses to multiple EBNA1 and LMP2 epitopes within an individual. A, shown are results obtained testing PBMCs from multiple time points from a single patient (0206) in an ELIspot assay using defined EBNA and LMP2 epitope peptides. Epitope sequences are denoted as being restricted through MHC class I (3 residue codes) and MHC class II (4 residue codes). DMSO solvent was included as a negative control and phytohemagglutinin (PHA) as a positive control. For most time points, sufficient cells were available for testing with each peptide or control in triplicate; to aid clarity only one representative well is shown in the figure. Following vaccination, epitope specific T-cell responses can be detected within 8 days and increase in magnitude, peaking 8 days after the third dose of vaccine (C3D8) and still detectable up to a further 50 days later (C3D51). Note that the EBNA1 and LMP2-specific T-cell responses detected in the whole-antigen ELIspot assay (see Fig. 1A) map to known CD8 epitopes in LMP2 and to known CD8 and CD4 epitopes in EBNA1. B, MVA-EL stimulates broad immune responses in multiple patients. Top, responses observed for three patients to whole antigens (pepmix); actin control (open boxes), EBNA1 (gray) and LMP2 (black) using samples obtained at three time points. The remaining panels show the results of ELIspot assays against epitope peptides within LMP2 and EBNA1, respectively, across a wider range of time points. These panels show the adjusted readings, for example, the difference between the epitope peptide and DMSO control spot counts at the same time points and are expressed as sfc/3 × 105 PBMC. For patients 0206 and 0513, pools of two LMP2 peptides were used (Supplementary Table S3); subsequent assays assigned T-cell responses to the individual epitope peptides indicated on the figure.

Figure 4.

MVA-EL stimulates CD8 and CD4 immune responses to multiple EBNA1 and LMP2 epitopes within an individual. A, shown are results obtained testing PBMCs from multiple time points from a single patient (0206) in an ELIspot assay using defined EBNA and LMP2 epitope peptides. Epitope sequences are denoted as being restricted through MHC class I (3 residue codes) and MHC class II (4 residue codes). DMSO solvent was included as a negative control and phytohemagglutinin (PHA) as a positive control. For most time points, sufficient cells were available for testing with each peptide or control in triplicate; to aid clarity only one representative well is shown in the figure. Following vaccination, epitope specific T-cell responses can be detected within 8 days and increase in magnitude, peaking 8 days after the third dose of vaccine (C3D8) and still detectable up to a further 50 days later (C3D51). Note that the EBNA1 and LMP2-specific T-cell responses detected in the whole-antigen ELIspot assay (see Fig. 1A) map to known CD8 epitopes in LMP2 and to known CD8 and CD4 epitopes in EBNA1. B, MVA-EL stimulates broad immune responses in multiple patients. Top, responses observed for three patients to whole antigens (pepmix); actin control (open boxes), EBNA1 (gray) and LMP2 (black) using samples obtained at three time points. The remaining panels show the results of ELIspot assays against epitope peptides within LMP2 and EBNA1, respectively, across a wider range of time points. These panels show the adjusted readings, for example, the difference between the epitope peptide and DMSO control spot counts at the same time points and are expressed as sfc/3 × 105 PBMC. For patients 0206 and 0513, pools of two LMP2 peptides were used (Supplementary Table S3); subsequent assays assigned T-cell responses to the individual epitope peptides indicated on the figure.

Close modal

The detailed results obtained from all patients screened in epitope peptide assays are recorded in Supplementary Table S3. Overall 11 of 18 patients gave evidence of epitope responses appropriate to their HLA type in the period during the vaccine course (up to, but not including, week 14). Responses were again seen most consistently in individuals receiving the highest vaccine dose. Six-month follow-up samples were available for 12 patients, who had an EBNA1 and/or LMP2 response during vaccination allowing the durability of these responses to be measured. Persistence of previously identified vaccine responses was not observed for the 6 patients vaccinated at dose levels 1 to 4 but were identified in 3 of 6 patients at the highest dose level.

All patients were also screened by qPCR for the presence of EBV DNA in plasma; this not only provides a surrogate measure of tumor burden in nasopharyngeal carcinoma (30) but also shows a transient increase as tumor cells lyse in vivo immediately following radio/chemotherapy (31). In this trial, all patients were clinically disease-free at the time of vaccination, and so, not surprisingly, most patients' plasma samples proved to be EBV DNA-negative both before and 1 month after vaccination. However, 1 patient (0517) did have low but detectable prevaccination levels of plasma EBV DNA. Interestingly, as shown in Fig. 5, this patient made a significant T-cell response to both EBNA1 and LMP2 pepmixes (and to at least 1 EBNA1 CD4 epitope) during the course of vaccination, though this seemed to fall subsequently. Immediately following the peak of this immunologic response, EBV genome levels in plasma increased to a peak at 10 weeks and then decreased to undetectable levels in the 6- and 12-month samples. After 19 months, however, this patient experienced nodal relapse, with plasma EBV DNA again detectable.

Figure 5.

Time course of EBV genomes in plasma in relation to immune responses for patient 0517. Events during screening, on and after trial participation are shown. Top, whole-antigen pepmix ELIspot data showing sfc/3 × 105 PBMC recognizing EBNA1 (open triangles) and LMP2 (closed triangles) pepmixes. Middle, ELIspot data showing sfc/3 × 105 PBMC recognizing a pool of three EBNA1 CD4+ T-cell epitopes (PQCR+LRVL+NPKF). Bottom, EBV genome copies/mL plasma, the lower limit of detection of the assay is 500 EBV genomes/mL plasma (shown as horizontal broken line). Times of vaccination are shown as arrows below the x-axis and the timescale shown on this axis applies to all three graphs. Low but stable levels of EBV DNA in the patient's plasma then increased to a peak immediately after vaccination before decreasing to undetectable levels This peak was preceded by vaccine-stimulated increases in EBNA1 and LMP2 immune responses. Much later (67 weeks), this patient experienced a parotid lymph node relapse (arrow). At time of relapse, the EBV level was 1,637 copies/mL but is not plotted because it was measured in a different center using a different assay.

Figure 5.

Time course of EBV genomes in plasma in relation to immune responses for patient 0517. Events during screening, on and after trial participation are shown. Top, whole-antigen pepmix ELIspot data showing sfc/3 × 105 PBMC recognizing EBNA1 (open triangles) and LMP2 (closed triangles) pepmixes. Middle, ELIspot data showing sfc/3 × 105 PBMC recognizing a pool of three EBNA1 CD4+ T-cell epitopes (PQCR+LRVL+NPKF). Bottom, EBV genome copies/mL plasma, the lower limit of detection of the assay is 500 EBV genomes/mL plasma (shown as horizontal broken line). Times of vaccination are shown as arrows below the x-axis and the timescale shown on this axis applies to all three graphs. Low but stable levels of EBV DNA in the patient's plasma then increased to a peak immediately after vaccination before decreasing to undetectable levels This peak was preceded by vaccine-stimulated increases in EBNA1 and LMP2 immune responses. Much later (67 weeks), this patient experienced a parotid lymph node relapse (arrow). At time of relapse, the EBV level was 1,637 copies/mL but is not plotted because it was measured in a different center using a different assay.

Close modal

Virus-associated malignancies offer an important model for the development of T cell–based cancer therapies, and the example of nasopharyngeal carcinoma is particularly interesting in several respects. On the one hand, nasopharyngeal carcinoma cells consistently carry the EBV genome and in most cases are also positive for both HLA I and HLA II molecules at the cell surface (29, 32, 33), implying visibility to both CD4+ and CD8+ T cells. On the other hand, virus antigen expression in nasopharyngeal carcinoma cells is limited to EBNA1 and LMP2 (and in a minority of cases, also LMP1) and none of these proteins are dominant targets of the natural virus-induced T-cell response (1). Nevertheless, most patients with nasopharyngeal carcinoma do have low levels of CD4 and CD8 memory T cells specific for EBNA1 and/or LMP2 in the circulation (22, 23), and indeed low numbers of CD8+ T cells with relevant reactivities have been detected in at least some tumor biopsies (29, 34). Furthermore, adoptive transfer of autologous T cells, expanded in vitro to enrich for EBNA1/LMP2–specific reactivities, has given early indications that nasopharyngeal carcinoma may be accessible to attack by appropriately activated, antigen-specific cells (11–15). The present work set out to explore whether one might be able to reactivate patients' immunity to the EBNA1 and LMP2 antigens by vaccination rather than by adoptive T-cell transfer, thereby opening up the possibility of a second immunotherapeutic strategy for nasopharyngeal carcinoma, indeed one that would be deliverable to large numbers of patients without requiring specialist laboratory support.

The results of this first vaccine trial carried out on patients with nasopharyngeal carcinoma in disease remission following conventional therapy, show that the MVA-EL vaccine is both safe and immunogenic. On the issue of safety, vaccine doses up to 5 × 108 pfu, given intradermally on 3 successive occasions at 3 weekly intervals, were well tolerated even in patients who were recovering from potentially immunosuppressive multicycle chemotherapy 4 to 10 months previously. Adverse events were limited to mainly mild injection site and transient systemic reactions, similar to reported patterns for other recombinant MVA vaccines up to 109 pfu (21, 35–37; Supplementary Table S1). This safety profile bodes well for future development of the MVA-EL–based vaccine approach as an immunotherapy for nasopharyngeal carcinoma.

About immunogenicity, from the evidence of pepmix assays the vaccine specifically increased circulating T-cell responses to EBNA1 and/or LMP2 in 15 of 18 patients. In most cases, responses to these antigens were detectable prevaccination, albeit at low levels, indicating that the vaccine was largely boosting rather than priming immunity. There were occasional examples of EBNA1 and/or LMP2 responses in the apparent absence of significant preexisting immunity (see patients 0308 and 0412; Supplementary Fig. S1), but it may well be that memory cell numbers in these individuals before vaccination were simply below the level of detection of the pepmix assay. Combining data from all 18 patients, mean levels of both EBNA1- and LMP2-specific T-cell numbers in the blood were increased 3- to 4-fold by vaccination. This effect was specific as idiosyncratic increases in preexisting immunity to control antigens (EBNA3A and FLU) were observed only occasionally, in 4 of 18 patients, and overall there was no significant change in their mean response levels. About the absolute size of responses to the vaccine-coded antigens, the mean increases in EBNA1 and LMP2 pepmix counts (Fig. 2A), recalculated as 230 and 156 sfc/106 PBMC, respectively, compare well against those observed with other MVA recombinant vaccines, being similar to those for the oncofetal antigen 5T4 in patients with colorectal cancer with concurrent chemotherapy (38) and exceeding those seen for other MVA cancer vaccines (21, 39). Higher frequency responses have been observed with MVA vaccines against malaria or tuberculosis; however, those studies used primed healthy volunteers and focused on recall responses to some of those organism's most immunogenic proteins (40, 41), not (as in the present work) recent chemotherapy recipients responding to naturally subdominant EBV antigens. Perhaps, more importantly, the size of MVA-EL–induced responses in the blood was similar or higher than measured for LMP2-specific effector frequencies following adoptive transfer of in vitro-expanded T-cell preparations into patients with nasopharyngeal carcinoma and Hodgkin lymphoma (11, 14, 42).

An important trial objective was to determine the dose for subsequent efficacy trials. Selection of 5 × 108 pfu is supported by several pieces of evidence. Thus, a greater proportion of responders in pepmix assays were observed at the highest than the lower 4 doses and the size of response also increased with dose (Table 2, Fig. 2B). Likewise, defined epitope-specific responses were detected in only 5 of 12 patients receiving dose levels 1 to 4 but were present, typically at higher levels, in all 6 patients at the highest dose (Supplementary Table S3). Although the small sample prevented multivariate analysis for all possible factors influencing responsiveness (e.g., HLA type, age, nasopharyngeal carcinoma treatment, and prevaccine immunity), univariate regression modeling also suggested a relationship between vaccine dose and immune response. Interestingly, this dose/response relationship was more apparent after 2 rather than 3 vaccine cycles, raising the possibility that vaccine dose may be less critical with repeated exposure. It is still a question of debate whether preexisting antibodies to the vaccinia vector itself, particularly virus-neutralizing antibodies, might limit the effectiveness of MVA-based vaccination. Our data on this point are limited as only 4 of 18 vaccine recipients were antivaccinia antibody-negative. However, we saw no evidence that immune responses to the vaccine-encoded EBNA1 and LMP2 antigens were significantly different in vector-naïve versus vector-immune individuals (Fig. 3).

The pepmix assays provide an important overall measure of T-cell immunity to a specific antigen but do not discriminate between CD4+ and CD8+ responses and give no information on specific epitope choice. In a second set of assays, therefore, we took advantage of existing epitope maps of the EBNA1 and LMP2 antigens (1) to re-examine the vaccine-induced responses in ELIspot assays using defined epitope peptides relevant to patients' HLA types. This showed that overall responses detected in pepmix assays often included reactivities to known CD8 epitopes in LMP2 or EBNA1 epitopes plus known CD4 epitopes in EBNA1. This ability to stimulate both CD4+ and CD8+ T cells may reflect the EBNA1/LMP2 fusion protein's ability to access both HLA class I and class II processing pathways within MVA-EL–infected antigen-presenting cells (16). Certainly, this coinduction of CD4 and CD8 responses by the vaccine is encouraging, as both T-cell subsets are likely to be required for effective vaccine-induced tumor surveillance (24–26). In this regard, we noted that varying response kinetics were observed for different epitopes; these included (i) stable responses postcycle 1, (ii) early responses diminishing immediately before cycle 2 and boosted on cycle 2 and 3, and (iii) responses rising and falling with successive vaccinations (Fig. 4). These different patterns in the blood may reflect the complex interplay between boosting responses, mobilization of T lymphocytes from the circulation into tissue, and possibly effector cell exhaustion, all factors that remain to be dissected in the context of cancer vaccination. It was also interesting to note occasional examples of patients with EBNA1 and/or LMP2 pepmix responses that could not be mapped in the epitope peptide assays (see patient 0101 for both antigens, patient 0514 for LMP2); this might be because not all potentially relevant epitopes could be included in all assays, but may equally well indicate that new target epitopes remain to be discovered in these proteins. Combining both pepmix and defined epitope screenings would therefore seem to be the best strategy for therapeutic vaccine trials, particularly where these involve patients not selected by HLA type and/or not of Caucasian origin. Variation in HLA alleles and suballeles, both within and between human populations, can dictate novel patterns of epitope choice that are distinct from previously mapped reactivities. Using as an example the different distribution of A2 suballeles prevalent in Chinese versus Caucasian populations, in the present work vaccine-induced reactivities were more often seen against recently identified HLA-A*02:03–restricted peptides in EBNA1 (VLK) and LMP2 (LLS) than to the better known LMP2 epitopes presented by the A*02:01 allele.

About the therapeutic potential of this vaccine, we would stress that patients in the current trial were in remission and it was not the purpose to measure clinical effect. However, it was interesting to observe that the 1 patient who had a detectable load of EBV DNA in plasma as a marker of tumor burden before vaccination (patient 0517) showed a transient spike in EBV levels following the development of a vaccine-induced T-cell response (Fig. 5A). A similar observation was made in a patient with nasopharyngeal carcinoma treated with EBV-specific T cells (11), and transient increases in EBV DNA are well recognized as an indicator of tumor cell killing in patients receiving conventional therapies (31). The example of patient 0517 is also interesting because this patient suffered a late recurrence of tumor some time after the vaccine-induced T-cell response had subsided. In fact, only 3 of 6 patients given the highest vaccine dose (patients 0515, 0516, and 0518) showed a sustained increase in circulating EBNA1 and/or LMP2-specific memory cell numbers present in the postvaccine bleeds at 6 months. While an immediate boosting of tumor-specific immunity by MVA-EL vaccination may be valuable in a therapeutic context, achieving a durable increase in these responses is an important long-term goal. Extending the vaccine course, either using the same vector or alternating with a heterologous vector expressing the EBNA1/LMP2 fusion antigen, would be one approach in that regard. Another possible factor determining long-term vaccine-induced immunity could be Tregs. In agreement with a previous study (44), we found the frequency of Tregs was raised in the blood of some patients with nasopharyngeal carcinoma. Interestingly, upon comparing 2 patients for whom Treg frequencies and long-term immune data were available (and who received the same dose of vaccine), we found differences in the durability of the vaccine-stimulated immune response inversely correlated with Treg frequency. EBNA1- and LMP2-specific T-cell responses were sustained in patient 0518 (with a vaccine-boosted EBNA1-specific response being detectable 6 months after vaccination) but were transient in patient 0517 in whom the frequency of Tregs was higher. However, confirming whether Tregs are indeed responsible for the differences in durability of the vaccine-induced T-cell response we observed, this will require a larger number of patients to be analyzed.

In summary, these findings and another trial of MVA-EL in United Kingdom patients (article in preparation), clearly show the immunogenicity of the vaccine and point to its potential as an adjuvant treatment of nasopharyngeal carcinoma when combined with conventional therapies. The work has also allowed the highest and most consistently immunogenic vaccine dose, 5 × 108 pfu, to be chosen for an ongoing phase II trial determining the clinical response rate to MVA-EL in the setting of relapsed or metastatic disease. Demonstration of the vaccine's immunogenicity in the context of nasopharyngeal carcinoma opens up its potential applicability to a wider range of EBV-associated tumors. Thus, EBV-positive Hodgkin lymphoma expresses both EBNA1 and LMP2 (as well as LMP1; ref. 2), whereas EBV-positive T/NK lymphoma, another malignancy seen most commonly in South East Asia, expresses EBNA1 and a shortened LMP2, which, though lacking N-terminal sequences, retains all of the currently known T-cell target epitopes (43). In each of these settings, the vaccine might be used to maintain remission following primary treatment or as treatment of relapsed disease, either alone or in combination with adoptive T-cell therapy or other modalities.

No potential conflicts of interest were disclosed.

Conception and design: E.P. Hui, G.S. Taylor, A.B. Rickinson, N.M. Steven, A.T.C. Chan

Development of methodology: E.P. Hui, G.S. Taylor, H. Jia, N.M. Steven, A.T.C. Chan

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): E.P. Hui, G.S. Taylor, H. Jia, B.B.Y. Ma, S.L. Chan, R. Ho, W.L. Wong, S. Wilson, B.F. Johnson, A.T.C. Chan

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): E.P. Hui, G.S. Taylor, H. Jia, B.F. Johnson, D.D. Stocken, A.B. Rickinson, N.M. Steven, A.T.C. Chan

Writing, review, and/or revision of the manuscript: E.P. Hui, G.S. Taylor, B.B.Y. Ma, S.L. Chan, R. Ho, W.L. Wong, C. Edwards, D.D. Stocken, A.B. Rickinson, N.M. Steven, A.T.C. Chan

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): E.P. Hui, B.B.Y. Ma, S.L. Chan, R. Ho, W.L. Wong, A.T.C. Chan

Study supervision: E.P. Hui, G.S. Taylor, C. Edwards, A.B. Rickinson, N.M. Steven, A.T.C. Chan

This study was supported by grants from Cancer Research UK, Research Grant Council of Hong Kong (CUHK 460708), and Hong Kong Cancer Fund.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

1.
Hislop
AD
,
Taylor
GS
,
Sauce
D
,
Rickinson
AB
. 
Cellular responses to viral infection in humans: lessons from Epstein–Barr virus
.
Annu Rev Immunol
2007
;
25
:
587
617
.
2.
Young
LS
,
Rickinson
AB
. 
Epstein–Barr virus: 40 years on
.
Nat Rev Cancer
2004
;
4
:
757
68
.
3.
Parkin
DM
. 
The global health burden of infection-associated cancers in the year 2002
.
Int J Cancer
2006
;
118
:
3030
44
.
4.
Baujat
B
,
Audry
H
,
Bourhis
J
,
Chan
AT
,
Onat
H
,
Chua
DT
, et al
Chemotherapy as an adjunct to radiotherapy in locally advanced nasopharyngeal carcinoma
.
Cochrane Database Syst Rev
2006
:
CD004329
.
5.
Lu
H
,
Peng
L
,
Yuan
X
,
Hao
Y
,
Lu
Z
,
Chen
J
, et al
Concurrent chemoradiotherapy in locally advanced nasopharyngeal carcinoma: a treatment paradigm also applicable to patients in Southeast Asia
.
Cancer Treat Rev
2009
;
35
:
345
53
.
6.
Bensouda
Y
,
Kaikani
W
,
Ahbeddou
N
,
Rahhali
R
,
Jabri
M
,
Mrabti
H
, et al
Treatment for metastatic nasopharyngeal carcinoma
.
Eur Ann Otorhinolaryngol Head Neck Dis
2011
;
128
:
79
85
.
7.
Razak
AR
,
Siu
LL
,
Liu
FF
,
Ito
E
,
O'Sullivan
B
,
Chan
K
. 
Nasopharyngeal carcinoma: the next challenges
.
Eur J Cancer
2010
;
46
:
1967
78
.
8.
Brooks
L
,
Yao
QY
,
Rickinson
AB
,
Young
LS
. 
Epstein–Barr virus latent gene transcription in nasopharyngeal carcinoma cells: coexpression of EBNA1, LMP1, and LMP2 transcripts
.
J Virol
1992
;
66
:
2689
97
.
9.
Bell
AI
,
Groves
K
,
Kelly
GL
,
Croom-Carter
D
,
Hui
E
,
Chan
AT
, et al
Analysis of Epstein–Barr virus latent gene expression in endemic Burkitt's lymphoma and nasopharyngeal carcinoma tumour cells by using quantitative real-time PCR assays
.
J Gen Virol
2006
;
87
:
2885
90
.
10.
Lin
CL
,
Lo
WF
,
Lee
TH
,
Ren
Y
,
Hwang
SL
,
Cheng
YF
, et al
Immunization with Epstein–Barr Virus (EBV) peptide-pulsed dendritic cells induces functional CD8+ T-cell immunity and may lead to tumor regression in patients with EBV-positive nasopharyngeal carcinoma
.
Cancer Res
2002
;
62
:
6952
8
.
11.
Comoli
P
,
Pedrazzoli
P
,
Maccario
R
,
Basso
S
,
Carminati
O
,
Labirio
M
, et al
Cell therapy of stage IV nasopharyngeal carcinoma with autologous Epstein–Barr virus-targeted cytotoxic T lymphocytes
.
J Clin Oncol
2005
;
23
:
8942
9
.
12.
Louis
CU
,
Straathof
K
,
Bollard
CM
,
Ennamuri
S
,
Gerken
C
,
Lopez
TT
, et al
Adoptive transfer of EBV-specific T cells results in sustained clinical responses in patients with locoregional nasopharyngeal carcinoma
.
J Immunother
2010
;
33
:
983
90
.
13.
Louis
CU
,
Straathof
K
,
Bollard
CM
,
Gerken
C
,
Huls
MH
,
Gresik
MV
, et al
Enhancing the in vivo expansion of adoptively transferred EBV-specific CTL with lymphodepleting CD45 monoclonal antibodies in NPC patients
.
Blood
2009
;
113
:
2442
50
.
14.
Straathof
KC
,
Bollard
CM
,
Popat
U
,
Huls
MH
,
Lopez
T
,
Morriss
MC
, et al
Treatment of nasopharyngeal carcinoma with Epstein–Barr virus–specific T lymphocytes
.
Blood
2005
;
105
:
1898
904
.
15.
Smith
C
,
Tsang
J
,
Beagley
L
,
Chua
D
,
Lee
V
,
Li
V
, et al
Effective treatment of metastatic forms of Epstein–Barr virus-associated nasopharyngeal carcinoma with a novel adenovirus-based adoptive immunotherapy
.
Cancer Res
2012
;
72
:
1116
25
.
16.
Taylor
GS
,
Haigh
TA
,
Gudgeon
NH
,
Phelps
RJ
,
Lee
SP
,
Steven
NM
, et al
Dual stimulation of Epstein–Barr Virus (EBV)-specific CD4+- and CD8+-T-cell responses by a chimeric antigen construct: potential therapeutic vaccine for EBV-positive nasopharyngeal carcinoma
.
J Virol
2004
;
78
:
768
78
.
17.
Voo
KS
,
Fu
T
,
Wang
HY
,
Tellam
J
,
Heslop
HE
,
Brenner
MK
, et al
Evidence for the presentation of major histocompatibility complex class I-restricted Epstein–Barr virus nuclear antigen 1 peptides to CD8+ T lymphocytes
.
J Exp Med
2004
;
199
:
459
70
.
18.
Tellam
J
,
Connolly
G
,
Green
KJ
,
Miles
JJ
,
Moss
DJ
,
Burrows
SR
, et al
Endogenous presentation of CD8+ T cell epitopes from Epstein–Barr virus-encoded nuclear antigen 1
.
J Exp Med
2004
;
199
:
1421
31
.
19.
Lee
SP
,
Brooks
JM
,
Al-Jarrah
H
,
Thomas
WA
,
Haigh
TA
,
Taylor
GS
, et al
CD8 T cell recognition of endogenously expressed Epstein–Barr virus nuclear antigen 1
.
J Exp Med
2004
;
199
:
1409
20
.
20.
Smith
CL
,
Dunbar
PR
,
Mirza
F
,
Palmowski
MJ
,
Shepherd
D
,
Gilbert
SC
, et al
Recombinant modified vaccinia Ankara primes functionally activated CTL specific for a melanoma tumor antigen epitope in melanoma patients with a high risk of disease recurrence
.
Int J Cancer
2005
;
113
:
259
66
.
21.
Dangoor
A
,
Lorigan
P
,
Keilholz
U
,
Schadendorf
D
,
Harris
A
,
Ottensmeier
C
, et al
Clinical and immunological responses in metastatic melanoma patients vaccinated with a high-dose poly-epitope vaccine
.
Cancer Immunol Immunother
2010
;
59
:
863
73
.
22.
Fogg
MH
,
Wirth
LJ
,
Posner
M
,
Wang
F
. 
Decreased EBNA-1-specific CD8+ T cells in patients with Epstein–Barr virus-associated nasopharyngeal carcinoma
.
Proc Natl Acad Sci U S A
2009
;
106
:
3318
23
.
23.
Lin
X
,
Gudgeon
NH
,
Hui
EP
,
Jia
H
,
Qun
X
,
Taylor
GS
, et al
CD4 and CD8 T cell responses to tumour-associated Epstein–Barr virus antigens in nasopharyngeal carcinoma patients
.
Cancer Immunol Immunother
2008
;
57
:
963
75
.
24.
Sun
JC
,
Bevan
MJ
. 
Defective CD8 T cell memory following acute infection without CD4 T cell help
.
Science
2003
;
300
:
339
42
.
25.
Shedlock
DJ
,
Shen
H
. 
Requirement for CD4 T cell help in generating functional CD8 T cell memory
.
Science
2003
;
300
:
337
9
.
26.
Khanolkar
A
,
Fuller
MJ
,
Zajac
AJ
. 
CD4 T cell-dependent CD8 T cell maturation
.
J Immunol
2004
;
172
:
2834
44
.
27.
Straathof
KC
,
Leen
AM
,
Buza
EL
,
Taylor
G
,
Huls
MH
,
Heslop
HE
, et al
Characterization of latent membrane protein 2 specificity in CTL lines from patients with EBV-positive nasopharyngeal carcinoma and lymphoma
.
J Immunol
2005
;
175
:
4137
47
.
28.
Tsang
CW
,
Lin
X
,
Gudgeon
NH
,
Taylor
GS
,
Jia
H
,
Hui
EP
, et al
CD4+ T-cell responses to Epstein–Barr virus nuclear antigen EBNA1 in Chinese populations are highly focused on novel C-terminal domain-derived epitopes
.
J Virol
2006
;
80
:
8263
6
.
29.
Lee
SP
,
Chan
AT
,
Cheung
ST
,
Thomas
WA
,
CroomCarter
D
,
Dawson
CW
, et al
CTL control of EBV in nasopharyngeal carcinoma (NPC): EBV-specific CTL responses in the blood and tumors of NPC patients and the antigen-processing function of the tumor cells
.
J Immunol
2000
;
165
:
573
82
.
30.
Chan
KC
,
Lo
YM
. 
Circulating EBV DNA as a tumor marker for nasopharyngeal carcinoma
.
Semin Cancer Biol
2002
;
12
:
489
96
.
31.
Lo
YM
,
Leung
SF
,
Chan
LY
,
Chan
AT
,
Lo
KW
,
Johnson
PJ
, et al
Kinetics of plasma Epstein–Barr virus DNA during radiation therapy for nasopharyngeal carcinoma
.
Cancer Res
2000
;
60
:
2351
5
.
32.
Khanna
R
,
Busson
P
,
Burrows
SR
,
Raffoux
C
,
Moss
DJ
,
Nicholls
JM
, et al
Molecular characterization of antigen-processing function in nasopharyngeal carcinoma (NPC): evidence for efficient presentation of Epstein–Barr virus cytotoxic T-cell epitopes by NPC cells
.
Cancer Res
1998
;
58
:
310
4
.
33.
Yao
Y
,
Minter
HA
,
Chen
X
,
Reynolds
GM
,
Bromley
M
,
Arrand
JR
. 
Heterogeneity of HLA and EBER expression in Epstein–Barr virus-associated nasopharyngeal carcinoma
.
Int J Cancer
2000
;
88
:
949
55
.
34.
Li
J
,
Zeng
XH
,
Mo
HY
,
Rolen
U
,
Gao
YF
,
Zhang
XS
, et al
Functional inactivation of EBV-specific T-lymphocytes in nasopharyngeal carcinoma: implications for tumor immunotherapy
.
PLoS ONE
2007
;
2
:
e1122
.
35.
Harrop
R
,
Connolly
N
,
Redchenko
I
,
Valle
J
,
Saunders
M
,
Ryan
MG
, et al
Vaccination of colorectal cancer patients with modified vaccinia Ankara delivering the tumor antigen 5T4 (TroVax) induces immune responses which correlate with disease control: a phase I/II trial
.
Clin Cancer Res
2006
;
12
:
3416
24
.
36.
Bejon
P
,
Peshu
N
,
Gilbert
SC
,
Lowe
BS
,
Molyneux
CS
,
Forsdyke
J
, et al
Safety profile of the viral vectors of attenuated fowlpox strain FP9 and modified vaccinia virus Ankara recombinant for either of 2 preerythrocytic malaria antigens, ME-TRAP or the circumsporozoite protein, in children and adults in Kenya
.
Clin Infect Dis
2006
;
42
:
1102
10
.
37.
Dunachie
SJ
,
Walther
M
,
Epstein
JE
,
Keating
S
,
Berthoud
T
,
Andrews
L
, et al
A DNA prime-modified vaccinia virus ankara boost vaccine encoding thrombospondin-related adhesion protein but not circumsporozoite protein partially protects healthy malaria-naive adults against Plasmodium falciparum sporozoite challenge
.
Infect Immun
2006
;
74
:
5933
42
.
38.
Harrop
R
,
Drury
N
,
Shingler
W
,
Chikoti
P
,
Redchenko
I
,
Carroll
MW
, et al
Vaccination of colorectal cancer patients with TroVax given alongside chemotherapy (5-fluorouracil, leukovorin and irinotecan) is safe and induces potent immune responses
.
Cancer Immunol Immunother
2008
;
57
:
977
86
.
39.
Ramlau
R
,
Quoix
E
,
Rolski
J
,
Pless
M
,
Lena
H
,
Levy
E
, et al
A phase II study of Tg4010 (Mva-Muc1-Il2) in association with chemotherapy in patients with stage III/IV non–small cell lung cancer
.
J Thorac Oncol
2008
;
3
:
735
44
.
40.
McShane
H
,
Pathan
AA
,
Sander
CR
,
Keating
SM
,
Gilbert
SC
,
Huygen
K
, et al
Recombinant modified vaccinia virus Ankara expressing antigen 85A boosts BCG-primed and naturally acquired antimycobacterial immunity in humans
.
Nat Med
2004
;
10
:
1240
4
.
41.
McConkey
SJ
,
Reece
WH
,
Moorthy
VS
,
Webster
D
,
Dunachie
S
,
Butcher
G
, et al
Enhanced T-cell immunogenicity of plasmid DNA vaccines boosted by recombinant modified vaccinia virus Ankara in humans
.
Nat Med
2003
;
9
:
729
35
.
42.
Bollard
CM
,
Aguilar
L
,
Straathof
KC
,
Gahn
B
,
Huls
MH
,
Rousseau
A
, et al
Cytotoxic T lymphocyte therapy for Epstein–Barr virus+ Hodgkin's disease
.
J Exp Med
2004
;
200
:
1623
33
.
43.
Fox
CP
,
Haigh
TA
,
Taylor
GS
,
Long
HM
,
Lee
SP
,
Shannon-Lowe
C
, et al
A novel latent membrane 2 transcript expressed in Epstein–Barr virus-positive NK- and T-cell lymphoproliferative disease encodes a target for cellular immunotherapy
.
Blood
2010
;
116
:
3695
704
.
44.
Lau
KM
,
Cheng
SH
,
Lo
KW
,
Lee
SA
,
Woo
JK
,
van Hasselt
CA
, et al
Increase in circulating Foxp3+CD4+CD25(high) regulatory T cells in nasopharyngeal carcinoma patients
.
Br J Cancer
2007
96
:
617
22
.