Wilms' tumour antigen 1 Immunity via DNA fusion gene vaccination in haematological malignancies by intramuscular injection followed by intramuscular electroporation: a Phase II non-randomised clinical trial (WIN)

Chronic myeloid leukaemia

Chronic myeloid leukaemia (CML) is a clonal disease of the haematopoietic stem cells (HSCs) in which a reciprocal translocation, t(9;22)(q34;q11), known as the Philadelphia chromosome, results in a fusion gene, BCR–ABL (breakpoint cluster region–Abelson murine leukaemia viral oncogene homolog 1), which in turn expresses an activated tyrosine kinase, and is regarded as the initiating lesion of CML. 3,4 Until quite recently, the only treatment to offer the possibility of long-term disease-free survival was allogeneic stem cell transplantation (alloSCT), the ‘curative’ effect of which is mediated in large part through the alloimmune graft-versus-leukaemia (GVL) effect. 5 However, alloSCT carries a substantial risk of mortality and is only available to a minority of patients. Tyrosine kinase inhibitors (TKIs), notably imatinib, have replaced alloSCT as first-line therapy for CML owing to their lower toxicity and impressive efficacy. Although > 85% of imatinib-treated patients with chronic-phase CML (CML-CP) achieve a complete cytogenetic response (CCyR), the majority of patients have persisting molecular disease, as assessed by quantitative polymerase chain reaction (qPCR) for BCR–ABL transcripts, and almost all will relapse following imatinib withdrawal. 6,7 Functional leukaemia CD34+ (cluster of differentiation 34-positive) progenitor cells have been identified in such patients within CCyR, suggesting the presence of a reservoir of leukaemia cells resistant to TKIs. 8 Furthermore, the durability of these responses has not yet been established. In contrast, long-term survivors of alloSCT very rarely have any detectable molecular disease, indicating that all leukaemia cells must be susceptible to immune destruction (GVL effect). Therefore, novel strategies to eradicate quiescent CML stem cells are required, especially because these cells provide a reservoir for disease relapse.

The immunological effect of alloSCT and donor lymphocyte infusions (DLIs) suggests that an approach based on the amplification of the patient’s own immune response to the disease could add to the responses seen after treatment with the TKI.

Based on our own data we argue here that vaccinating against the Wilms’ tumour antigen 1 (WT1), using deoxyribonucleic acid (DNA) vaccination, is an attractive choice for delivering this immune attack. 9,10 The validity of WT1, as a target for immunotherapy in CML, was recently shown in work published by Yong et al. [in John Barrett’s group at the National Institutes of Health (USA)]. 11 This group studied the expression of leukaemia-associated antigens, including WT1, within the CD34+ primitive stem cell and committed progenitor cell pools in CML patients. WT1 is significantly overexpressed in all CD34+ subpopulations in CML encompassing the most primitive HSCs to the most mature cells,11 which escape control by imatinib. Taken in the context of these clinical data and that from other groups, which show that even suboptimal vaccination with peptide can have clinical effects,11,12 these data strongly suggest that active immunotherapy other than allotransplantation holds significant promise by the induction of tumour antigen-specific CD8+ (cluster of differentiation 8-positive) T lymphocytes (T cells) without adding toxicity.

Clearly it is critical to choose the best clinical setting in which to vaccinate, as previous data have shown that the effect of TKIs as a drug class on the immune system is variable,13 and can be either suppressive or stimulatory. For imatinib specifically, in vivo data show that it can be immunostimulatory, supporting our proposed trial, both in murine14,15 as well as human16–18 studies. Furthermore, Wang et al. 14 demonstrated that in vivo treatment with imatinib not only prevented the induction of tolerance, while preserving responsiveness to a subsequent immunisation, but, critically, enhanced vaccine efficacy. In patients, low-frequency CD8+ T-cell responses to four leukaemia-associated antigens (ABL kinase, proteinase 3, telomerase and WT1) were detected in CML patients on imatinib and show the immune system’s ability to respond to leukaemia-associated antigens in the presence of imatinib. 19 It is therefore unsurprising that two vaccine studies using BCR–ABL peptides in patients with CML treated with imatinib clearly demonstrated the successful induction of CD8+ and CD4+ (cluster of differentiation 4-positive) T cells against the vaccine, even with a suboptimal peptide vaccine approach. 20,21 Bocchia et al. 20 found that antileukaemia T-cell responses could be stimulated after vaccination in 9 out of 14 patients. In the EPIC trial, T-cell responses to CD4+ T-cell responses against the vaccine were seen in all patients and 14 of 19 patients developed T-cell responses to BCR–ABL peptides. 21 Prospective analysis of immune responses to vaccination against influenza A (H1N1, 2009 strain) and Streptococcus pneumoniae (Klein 1884) Chester 1901 in 50 CML-CP patients treated with imatinib (Glivec^®, Novartis Pharmaceuticals UK Ltd), dasatinib (Sprycel^®, Bristol-Myers Squibb) or nilotinib (Tasigna^®, Novartis Pharmaceuticals UK Ltd) and 15 healthy controls was recently performed. 22 Significant CD8+ and CD4+ T-cell responses against flu were induced in patients with CML-CP on TKIs following vaccination and there was no significant difference in the vaccine-induced T-cell response between CML-CP patients on TKIs and healthy controls (manuscript in preparation). These data strongly support that vaccination of patients on stable doses of imatinib will induce immune responses.

Acute myeloid leukaemia

Acute myeloid leukaemia (AML) is a disease of older adults with a median age of 68 years23 and an incidence of 8–12 per 100,000 population. Advances in the understanding of the pathophysiology of AML have not yet led to major improvements in disease-free and overall survival of adults with this disease. Only about one-third of those aged between 18 and 60 years who are diagnosed with AML can be cured; disease-free survival is rare and current therapy is devastating in older adults. Treatment of AML involves chemotherapy with high remission rates in up to 85% of patients; however, remissions are often short lived and > 70% of patients will progress and die from their disease within 2 years (see Figure 1). 24 Treatment also causes significant morbidity and mortality. AlloSCT from a compatible donor carries a 20–75% chance of long-term disease-free survival depending on whether the transplant is performed in remission or with residual disease. Death from relapse is the most common cause of treatment failure following transplant. At this point, a minority of patients respond to chemotherapy and DLIs, but remission rates are around 15% with only a fraction being durable. 17,18 There is, therefore, a need to devise better treatments for AML.

In AML, WT1 has been established as a marker for minimal residual disease (MRD). 25 Additionally, WT1 gene expression has been suggested to carry adverse prognostic implications in AML based on data from a number of studies. 26,27 A recent trial by the European LeukemiaNet defined and standardised a WT1 real-time qPCR assay as a marker for MRD monitoring and risk stratification in AML. 28 We intend to exploit this for the proposed trial of WT1 vaccination. As in CML, peptide vaccination has been tested with some success12,29–33 and the data support that active immunotherapy other than allotransplantation holds significant promise by the induction of tumour antigen-specific CD8+ T cells without added toxicity.

The purpose of the trial was to build on an established programme of DNA fusion gene vaccination delivered by intramuscular injection and exploiting this unique experience with electroporation, to induce durable immune responses with the aim of controlling the disease by precision attack of the tumour by CD8+ T cells. The aim of the trial was to evaluate an identical vaccine strategy in two parallel settings with the purpose of identifying the most promising context for eventual Phase III testing. The hypothesis was that molecular and clinical responses, induced by T cells can be predicted by increases in the number of CD8+ T cells, specific for the vaccine-encoded T-cell epitopes.

Trialling two patient groups will maximise the knowledge gained from this vaccine trial. Patients with CML will allow a direct and objective assessment of the antileukaemia effect of vaccination at the molecular level by BCR–ABL and WT1 monitoring. Patients with AML offer a difficult challenge to haematologists. The advantage of including this patient group is twofold: (1) the antileukaemia effect of vaccination can be assessed objectively by measuring WT1 gene expression levels; and (2), more importantly, data can be collected on the highly clinically relevant question of whether or not vaccination will prevent relapse in this patient group.

Selection of patients for vaccine therapy

Novel therapies are often first introduced in patient groups who have failed all conventional treatment options and have far advanced or metastatic disease. This strategy is inappropriate for vaccine treatments, which depend upon an intact well-functioning immune system, known to be severely impaired in advanced cancers. The cohort to be studied here has therefore been chosen to reflect this conclusion.

Immunotherapy in haematological malignancies targeting Wilms’ tumour antigen 1

DNA fusion vaccines were initially developed to treat B-cell malignancies,34 which showed that fusion of the microbial sequence, fragment C from the tetanus toxin (FrC) to idiotypic tumour antigen provided the T-cell help required to induce humoral35 and CD4+ T-cell responses in preclinical models. 36 Early clinical testing was undertaken in a Phase I/II dose escalation trial [LIFTT trial; Gene Therapy Advisory Committee (GTAC) 029A], with individual idiotypic DNA fusion vaccines to treat patients with follicular lymphoma. The vaccine was safe and 14 out of 18 patients showed an antibody and/or CD4+ T-cell responses against the FrC portion of the fusion gene. Encouragingly, 6 out of 16 patients showed responses to the tumour-specific idiotypic antigen (manuscript in preparation). There was no evidence of a dose response for doses ranging from 500 µg/dose to 2500 µg/dose. 9 Overall, however, the levels of response were relatively low and improvements were sought.

An important development has been electroporation, which has been shown to dramatically increase DNA vaccine performance in mice37 and rhesus macaques,38 and this method of delivery was used in our clinical trial in patients with prostate cancer. We found clear evidence for amplification of antibody and CD4+ T-cell responses in patients. 39 For induction of CD8+ T-cell responses, the vaccine design was modified by reducing the FrC sequence to a single domain (p.DOM; plasmid domain 1 from FrC). This decreased the potential for peptide competition but retained the major histocompatibility complex class II-restricted peptide p30. 39 An epitope-specific sequence was then inserted at the C-terminus of FrC to aid processing/presentation. In multiple models, this p.DOM epitope design (Figure 1a) was able to induce high levels of epitope-specific CD8+ T cells. 9

Importantly, provision of high levels of T-cell help enables induction of immune responses in tolerant settings. 9,39

The preclinical data appear to predict response in humans. 10 For patients with relapsed prostate cancer, a p.DOM epitope design incorporating a peptide sequence from the prostate-specific membrane antigen (PSMA; GTAC 089) has induced high levels of epitope-specific interferon gamma (IFN-γ), producing CD8+ T-cell responses in 67% (10/15) of patients. 40 Data from the 10 patients in the group receiving the lowest dose levels of DNA and DNA/electroporation are shown in Figure 1b. This was the first ever trial to exploit delivery of DNA by electroporation and it was found this approach was safe and readily accepted by patients. 10 Responses were robust and persistent over many months to the end of follow-up in the trial at 18 months (Figure 1b).

FIGURE 1.

Vaccination of patients with the p.DOM epitope vaccine. (a) The p.DOM epitope vaccine consists of a DNA plasmid backbone incorporating cytosine–phosphate–guanine sites. The first domain of the tetanus toxin (DOM; TT865–1120) provides T-cell help when linked to a tumour-associated nucleotide sequence encoding the human leucocyte antigen class I binding epitope of interest. This format allows the appropriate processing and presentation of the peptide. (b) Spots per million peripheral blood mononuclear cell producing IFN-γ in human leucocyte antigen A2-positive patients treated with three monthly doses of DNA (p.DOM.PSMA27). The figure shows data from the first dose cohort, analysed in a cultured.

Figure 2 illustrates the CD8+ T-cell analyses in more detail. In Figure 2a and 2b two non-responders are shown, one of whom (Figure 2b) had pre-existing levels of PSMA peptide 27-specific T cells at baseline. It is interesting to note that these cells appear to leave the circulation post vaccination and become visible again after the first booster injection at 6 months. Further data are required to allow interpretation of this observation. In Figure 2c and 2d, two of the six responders at dose level 1 are shown. The patient in Figure 2c was treated with DNA alone followed by DNA delivered by electroporation, whereas the patient in Figure 2d was treated with DNA/electroporation on five occasions.

FIGURE 2.

The CD8+ T-cell responses to DNA vaccination analysed over time by enzyme-linked immunospot. (a) and (b) show data on two out of four non-responders, of which the patient in (b) shows a low-level CD8+ T-cell response to the PSMA peptide 27 at baseline. As there is no significant increase in levels of PMBCs producing IFN-γ above the baseline, this patient has been classified as a non-responder; (c) and (d) show examples of patients that have significantly increased levels of PMBCs producing IFN-γ compared with baseline levels and to the human immunodeficiency virus-negative control. (n = 6 in the first dose cohort.) HIV, human immunodeficiency virus; PBMC, peripheral blood mononuclear cell.

The effectiveness of the p.DOM epitope design for the treatment of myeloid malignancies has been explored based on these clinical results. WT1 has emerged as one of the most promising targets for immunotherapy of haematological malignancies including CML, AML and myelodysplastic syndromes. 29,41–43 It is also a potential target for the treatment of solid tumours. 43–46 Despite its ubiquitous expression during embryogenesis, WT1 expression in normal individuals is limited to renal podocytes, gonadal cells and a small proportion of CD34+ cells,47–50 where expression is significantly lower (10- to 100-fold). 47 This could raise a concern about autoimmunity, but, reassuringly, the available data document selectivity of attack against tumour cells, sparing the CD34+ cells51,52 and without any evidence of renal or other autoimmune toxicity in murine models53–55 or patients. 29,41–43

The WT1 peptide vaccines have been tested both in preclinical models51,52,56 and in clinical trials. 12,41–43 The data from clinical trials document that T-cell responses can be induced in patients and confirm the presence of an expandable CD8+ T-cell repertoire. Importantly, the ability of peptide vaccines to induce measurable clinical responses has been documented. However, a key problem with class I-restricted peptide vaccines is the inability of this approach to provide linked CD4+ T-cell help, which is crucial for the maintenance of tumour antigen-specific CD8+ T-cell populations. In the clinic, this is visible in poor persistence of the detected CD8+ T-cell responses. In contrast, it was that the p.DOM epitope fusion vaccines appeared to be able to deliver CD8+ T-cell responses, which show long-term persistence (see Figures 1b and 2c and 2d).

Recently, three domain 1 from fragment C of tetanus toxin (DOM) epitope vaccines were evaluated, each encoding a different, previously described, WT1-derived, human leucocyte antigen A2 (HLA A2)-restricted peptide. 56 All were able to induce CD8+ T-cell responses in ‘humanised’, and presumably tolerised, mice expressing HLA A2 and these killed human WT1-positive (WT1+), HHD+ leukaemia cells ex vivo. A direct comparison with a WT1 peptide vaccine (plus T-cell help and adjuvant) showed a clear superiority of the DNA fusion vaccine. 56 In parallel, we showed that low numbers of human WT1 peptide-specific T cells could be expanded in vitro to kill WT1+ HLA A2+ leukaemia cells. The WT1 peptide 37 (WT1-37) and WT1 peptide 126 (WT1-126) peptides were selected for current studies. We have already documented clinically the ability of p.DOM epitope vaccines to induce cytotoxic T lymphocytes (CTLs) and anticipate that dual attack against more than one epitope will provide added clinical benefit. 9,10 Vaccination with p.DOM–WT1-37 and p.DOM–WT1-126 into different locations will allow us to avoid antigenic competition. Given the clear effect on the response to the FrC portion of the vaccine in the prostate trial, electroporation was used as a delivery strategy. 9,10,56

Significant change to trial design

The original trial design used Simon’s optimal Phase II trial design. 57,58 This allowed us to undertake a ‘start/stop’ evaluation once 12 patients had been enrolled into each vaccination group of the study and had been evaluated to 6 months (molecular monitoring). An interim analysis of these CML patients’ molecular data (BCR–ABL and WT1) was to be carried out including up to 6 months of data. If a molecular response (change in transcript level of BCR–ABL or WT1) was observed the CML arm may recruit an additional 25 participants. It was designed that the AML arm would also be opened to recruitment at this point.

It became evident early on in the trial that recruitment of the CML patients was significantly behind target. At the same time, new data emerged which illustrated that immune responses detected in the blood appear to evolve by 6 months post vaccination in most patients. 56 These data were not available when the WT1 Immunity via DNA (WIN) trial protocol was developed. Clinical responses (reduction in BCR–ABL transcript levels) are therefore also expected to happen late which would necessitate a halt in recruitment to ‘observe’ outcomes in the first stage prior to initiating the second stage.

This new information manifested to protocol amendment six. The following changes were included in this amendment to increase the available eligible CML patient population and help speed up recruitment:

• The AML arm was to be opened, independent of the CML result, as a result of the new data that had emerged illustrating that immune responses detected in the blood appear to evolve by 6 months in most patients.

• Patients on any TKI became eligible (rather than just those on imatinib).

• The trial design was changed from Simon’s two-stage design58 to A’Hern’s single-stage design. 59

• Amendment to the sample size for the trial to observe a minimum of 4 out of 32 CML patients who are molecular responders in order to provide evidence that the vaccine warrants further investigation. This is an increase proportionally from 10.8% (4/37) to 12.5% (4/32) and allowed all patients to receive all 12 vaccinations rather than only receive an additional six based on response being observed.

Early closure of the trial

The CML arm of the trial was terminated early by the funders because of poor recruitment. This meant that recruitment of the AML arm could not be implemented until an assessment was done to secure funding to continue with this arm of the trial. The outcome of this assessment was to close the trial. The Research Ethics Committee and the Medicines and Healthcare products Regulatory Agency (MHRA) were notified on the 3 April 2014.

The study met its primary decision-making target with one major molecular response in BCR–ABL transcript levels and, in parallel, new data emerged which illustrated that immune responses detected in the blood appear to evolve by 6 months post vaccination in most patients. 56 The early onset of the major molecular response suggests that vaccination has achieved this molecular event. In parallel, a WT1 molecular response was observed.

A significant amount of time and effort was utilised in an attempt to overcome multiple hurdles that prevented successful recruitment, but despite this the study did not complete recruitment. The failure to achieve the recruitment target was exclusively driven by the lack of recruitment of CML patients.

Some of the major hurdles which contributed to the failure of the studies are as follows:

• The agreement and enthusiasm of the principal investigator (PI) for the centre at Hammersmith to participate in the study was not matched by a main member of the team to recruit to the study, leading to a lack of recruitment from the centre where the largest number of CML patients was expected.

• At the feasibility stage, the expert haemato-oncologist (PI) at Hammersmith assessed the cohort size of eligible patients to be 500 patients with CML who would fulfil the entry criteria for the study. This meant that 70% of the recruitment would come from the cohort size at this centre. Unfortunately, neither the cohort size nor recruitment materialised and this was key to the success of the study.

• Apparent lack of clinical efficacy by the vaccination and, therefore, less interest in the study.

• The unexpected loss of the PI to a US centre, leaving the study unsupervised in the key centre.

The following key reasons contributed to the failure to achieve the recruitment target:

• Extensive attempts were made by the chief investigator via the National Institute for Health Research (NIHR) clinical studies group and by visiting national key centres, which mange CML, to recruit new centres to the study. Although there was significant academic interest for this study, it competed directly with an ongoing study in CML, precluding their participation.

• As highlighted above, the NIHR Efficacy and Mechanism Evaluation (EME) programme did not support recruitment of AML patients. Therefore, recruitment of this study arm was stopped and no patients were vaccinated; it was deemed unethical to recruit one or two patients to the AML arm and then stop the trial.

This was disappointing as the robust induction of FrC responses (10/10 evaluable patients) and WT1 T-cell responses in 7 out of 10 evaluable patients support that the preclinical data link to immunological outcomes as predicted. Overall, the data confirmed the immunogenicity and safety of the vaccine.

Changes to original protocol

A summary of the changes to the original protocol is given in Table 1.

Inclusion criteria

Exclusion criteria

Vaccination schedule

The DNA vaccine was administered 12 times. Patients received the vaccine at 4-weekly intervals into separate sites for the first 6 months, followed by vaccinations every 3 months up to a maximum of 24 months. Vaccines were injected intramuscularly and followed by intramuscular electroporation. Pain assessments were conducted immediately after vaccination and at 48 hours post vaccination.

The first HLA A2+ patient recruited at each site was evaluated 48 hours after administration of the first vaccination, before additional doses were given or before additional patients were vaccinated at that site.

Recruitment and informed consent

The PIs identified potential eligible patients from their existing patient population either during routine consultation or from a database search. Patients identified from their database were approached for the trial at their next clinic appointment. Patients who were interested in participating in the trial were provided with the patient information leaflet and signed the informed consent form prior to enrolment into the trial.

Registration

After written informed consent was obtained from the patient and before screening commenced, sites registered the patient with the Southampton Clinical Trials Unit (SCTU) to obtain the unique patient identification number. The patient’s eligibility was checked during the registration process to ensure that only patients fulfilling the eligibility criteria were registered. Subsequently, the patient identification number was assigned.

Data collection and management

Sites entered trial-specific data, as specified in the protocol, onto paper case report forms (pCRFs). Completed pCRFs were sent to the SCTU, which was responsible for the data management of the trial. Data were transcribed from pCRFs into an InForm database (InForm version 5.0, ORACLE) at the SCTU. A range of data validation checks were carried out within both InForm and SAS version 9.3 (SAS Institute Inc., Cary, NC, USA) or above to minimise incorrect or missing data.

Molecular samples were sent to Haematology department at Hammersmith Hospital (a clinical pathology-accredited MRD laboratory) for molecular monitoring (qPCR for BCR–ABL/WT1 in CML and WT1 in AML). This is an accepted and routine test for monitoring of this disease. Results from the analysed samples were regularly sent back to SCTU for statistical analysis.

Immunological analyses for vaccine responses, including leukapheresis samples and bone marrow samples, were processed and frozen locally according to an agreed standard operating procedure and stored in liquid nitrogen. Samples were transported to the Cancer Sciences Division (Southampton General Hospital, Southampton, UK) in dry ice and using a temperature logger once a sufficient number or samples had been collected locally (Hammersmith Hospital, Exeter, UK).

Source data verification was undertaken during site monitoring visits, in accordance with the SCTU’s trial monitoring plan. Sites were visited at least once during the trial. At least one monitoring visit was undertaken for each participating site. A total of three planned monitoring visits and one triggered monitoring visit were carried out.

Molecular response

Immunological response

Primary outcome

Secondary outcomes

Original sample size

For sample size calculation, we used Simon’s optimal Phase-II trial design for clinical development of therapeutic cancer vaccines. 60,61 This would allow us to undertake a ‘start/stop’ evaluation once 12 patients had been enrolled into each vaccination group of the study and had been evaluated to 6 months (molecular monitoring).

If one or more responders were observed in the 12 initial patients, this particular vaccination group would be extended to 37 patients. This would allow the study to distinguish between p₀ = 5% (standard response) and p₁ = 20% (expected response) with α = 0.1% (two-sided) and β = 0.10, where p₀ is the probability of a clinically uninteresting true response rate and p₁ is the probability of a sufficiently promising true response rate. This gives a less than 10% chance of rejecting a useful vaccine, with error probability limits of α 0.10 and β < 0.10, even if the true response rate were to be p₁ = 20%.

If no molecular responses were seen in the first 12 CML or AML patients, recruitment would have ceased for this patient group, as there will not be sufficient clinical interest to pursue the trial further. This optimal design had an expected sample size under H₀ (the null hypothesis) of 24 and a maximum sample size of 37 in each patient group (AML and CML).

Revised sample size

The sample size was revised in line with the change in trial design from Simon’s two-stage60 to A’Hern’s one-stage trial design. 59 This was approved in version 6 of the protocol on 8 November 2012 (see Table 1).

Analysis populations

Preliminary analyses

Disposition of patients and follow-up information on the ITT population were analysed separately for HLA A2+ and HLA A2– patients (see Figure 4 and Table 2). Major protocol deviations were also listed for the ITT population (see Box 1).

Patient characteristics recorded at baseline, including the patient’s age, gender, ECHO result and WHO performance status, were summarised by HLA A2 status for the ITT population (see Table 3). In addition, baseline BCR–ABL and WT1 molecular data were analysed separately for the molecular analysis population and summarised by HLA A2 status (see Tables 4 and 5).

Dosing information on pre-trial imatinib treatment was summarised by HLA A2 status for the ITT population (see Table 6).

Treatment analyses

Vaccination administration and electroporation failure summarises were presented for the ITT population (HLA A2+ patients only), together with details of any reasons for premature withdrawals from treatment (see Tables 7–9).

Safety analyses

Serious adverse events (SAEs), including a SAE summary and listing of SAEs by system organ class and CTCAE term, were presented by HLA A2 status for the safety analysis population (Tables 10 and 11). In addition, a complete SAE listing was provided for the safety analysis population (see Table 12).

Information on all AEs was also presented by HLA A2 status for the safety analysis population, which comprised a summary by worst CTCAE grade recorded across the following type of AEs:

• all AEs (see Table 13)

• cardio-related AEs (see Table 14)

• renal-related AEs (see Table 15)

• bone marrow-related AEs (see Table 16)

• other-related AEs (see Table 17).

Fisher’s exact test was used to compare differences in rates of toxicity in CML patients with controls for the safety population. No adjustment was made for multiple testing and, therefore, borderline significance should not be overinterpreted.

A list of all AEs was also presented for the safety analysis population (see Table 18). In addition, any concomitant medications taken because of pain within 48 hours following vaccination were summarised for HLA A2+ patients only in the safety analysis population (see Table 19).

Primary analyses

The primary outcome is a BCR–ABL molecular response (major or minor response or CMR), which was summarised for the molecular analysis population by HLA A2 status (see Table 20).

Secondary analyses

Secondary outcomes carried out on the molecular analysis population included:

• BCR–ABL (major response or CMR), summarised by HLA A2 status (see Table 21)

• WT1 (major or minor response or CMR), summarised by HLA A2 status (see Table 22).

In addition, the following time-to-event secondary outcomes were also analysed by HLA A2 status for the ITT population:

• time to last follow-up (see Table 23 and Figure 5)

• time to progression or death (see Table 24 and Figure 6)

• time to death (see Table 25 and Figure 7)

• time to next treatment (see Table 26 and Figure 8).

In each case, the median time to event was presented along with time to event listings, a log-rank test, results from an unadjusted Cox proportional hazards model with alpha set at the 10% two-sided level and a Kaplan–Meier plot.

Time to a BCR–ABL response (major or minor or CMR) and time to a WT1 response (major or minor or CMR) from the beginning of inhibitor treatment and also from the date of informed consent were also analysed using a similar approach for the molecular analysis population, together with the duration of BCR–ABL and WT1 responses (see Table 27 and Figure 9, and Table 28 and Figure 10, respectively).

Pain assessment information recorded immediately after vaccination and recorded 48 hours post vaccination were summarised in terms of the median pain score recorded and the worst pain score recorded for HLA A2+ patients in the ITT population (see Tables 29 and 30, respectively).

Flow of participants through the trial

A total of 91 CML patients were assessed for eligibility for entry into the WIN trial and 23 CML patients were registered. Of these, 13 patients were HLA A2+ and 12 out of 13 began vaccination with a control group of nine HLA A2– patients.

Twelve HLA A2+ patients were vaccinated and 12 out of 12 as well as nine out of nine control patients were included in both the safety and molecular analyses. Four vaccinated and two control patients completed the study protocol treatment. The other patients either withdrew consent (n = 11) or underwent modification of their TKI treatment, thus requiring study termination.

Only one AML patient was assessed for eligibility for entry into the WIN trial, but decided not to enter into the trial.

Figure 4 shows the CONSORT diagram for the CML group only.

FIGURE 4.

Consolidated Standards of Reporting Trials flow chart (CML group only).

Serious adverse events

Details of SAEs are provided in Tables 10–12. No vaccine-related SAEs were observed. Three SAEs were observed in the control group and are not vaccine related.

Adverse events

In the vaccine group, 11 grade 1–3 AEs were observed, compared with nine grade 1–3 AEs in the control group (Table 13). Of these AEs, palpitations in one vaccinated patient were of special interest (Table 14). In one patient, 12 grade 1–3 renal AEs were documented (pre-existing grade 1 renal dysfunction), with no such AEs observed in the control group (Table 15). No haematological toxicities were observed (Table 16).

A range of mild to moderate AEs was observed (Table 17). No detectable significance was observed between the vaccination and observation arms.

Table 18 provides a complete list of all AEs for the safety analysis population.

Concomitant medications taken because of pain within 48 hours of vaccination

Two patients used painkillers after vaccination/electroporation, with good symptomatic relief (Table 19). No new clinical insights beyond the pre-trial data on electroporation were observed.

Breakpoint cluster region–Abelson response (major or complete molecular response)

The BCR–ABL major and CMR responses were evaluated, with no change in results to those seen in the primary end point (Table 21).

WT1 response

There were three WT1 responses observed, all of which were defined as CMR. Two responses were observed in the HLA A2+ group and one in the HLA A2– group. There were no differences observed between the groups (Table 22).

Follow-up

Median follow-up time was 18 months for HLA A2+ patients and 15 months for HLA A2– patients. No significant differences between the study arms were detected (Table 23).

A Kaplan–Meier plot for time to last follow-up in the ITT population is presented in Figure 5.

FIGURE 5.

Kaplan–Meier for time to last follow-up: ITT population.

Progression-free survival

There were no disease progression or deaths observed in this trial (Table 24).

A Kaplan–Meier plot for time to disease progression or death in the ITT population is presented in Figure 6.

FIGURE 6.

Kaplan–Meier for time to disease progression or death: ITT population.

Overall survival

There were no deaths observed in this trial (Table 25).

A Kaplan–Meier plot for time to death in the ITT population is presented in Figure 7.

FIGURE 7.

Kaplan–Meier for time to death: ITT population.

Time to next treatment

Time to next treatment is defined as time from date of consent to the date of next CML treatment or date of last follow-up (whichever occurs first). No additional CML treatment was administered to any patients within this trial (Table 26).

A Kaplan–Meier plot for time to next treatment in the ITT population is presented in Figure 8.

FIGURE 8.

Kaplan–Meier for time to next treatment: ITT population.

Time to response

Two BCR–ABL responses were observed. One patient in the HLA A2+ group had a response at week 8 and one patient in the HLA A2– group had a response at month 23. The data suggest a vaccine response in the HLA A2+ patient and a late imatinib response in the HLA A2– patient (Table 27).

A Kaplan–Meier plot for time to for time to BCR–ABL molecular response in the molecular analysis population is presented in Figure 9.

FIGURE 9.

Kaplan–Meier for time to BCR–ABL molecular response: molecular analysis population. (a) From beginning of TKI treatment (in years); and (b) from informed consent (in months).

There were three WT1 responses observed. HLA A2+ patients experienced responses at week 20 and week 32. The patient in the HLA A2– group had a response at week 8 (Table 28).

A Kaplan–Meier plot for time to for time to WT1 molecular response in the molecular analysis population is provided in Figure 10.

FIGURE 10.

Kaplan–Meier for time to WT1 molecular response: molecular analysis population. (a) From beginning of inhibitor treatment (in years); and (b) from informed consent (in months).

Pain assessment immediately after vaccination

Patient 2-C-021 did not receive any treatment, as outlined by the missing pain assessment data values described in Table 29.

Pain assessment 48 hours post vaccination

Pain and discomfort data were measured on a visual analogue scale ranging from 0 (no pain/discomfort) to 10 (worst ever pain/discomfort). Electroporation was well tolerated, with short-lived clinical effects on the patient. No new safety information emerged on electroporation.

Patient 2-C-021 did not receive any treatment, as outlined by the missing pain assessment data values described in Table 30.