This was an open-label, phase I/II adaptive design study of LPV7±IFA and TLR agonists in participants with resected Stage IIB-IV melanoma. The protocol is provided as online supplemental material 1. The primary objective was to assess safety of each vaccine combination. A secondary objective was to estimate the immunogenicity of LPV7 in each of the seven adjuvant preparations, with an expectation that CD8⁺ T cell responses to the embedded short peptides would be enhanced compared with prior vaccines with short peptides themselves. A summary goal was to determine the optimal vaccine plus adjuvant combination based on safety and immunogenicity.

Patients were eligible if 18 years or older, with ECOG performance status of 0–1 and Stage IIB-IV melanoma (AJCC 7th edition, at original diagnosis or recurrence), rendered clinically free of disease by surgery or other therapy or by spontaneous remission within 6 months of study entry. Patients with brain metastases were eligible if they had had no more than three lesions, none >2 cm in diameter at the time of protocol entry, all completely removed or treated. Eligible patients must have had at least one intact axillary or inguinal node basin. The study was limited to patients expressing HLA-A1, A2, A3, B35, or B51. Ocular primary melanoma patients were not eligible.

Six peptides from gp100, tyrosinase, MAGE-A1, and MAGE-A10 proteins were selected based on high immunogenicity in prior clinical trials using MEPs for CD8⁺ T cells14 15 31 restricted by HLA-A1, A2, or A3: long peptides encompassing those MEPs were constructed. Also, the NY- ESO-1 _94-102 peptide, containing a MEP for CD8⁺ T cells restricted by HLA-B35 and B51, was selected based on immunogenicity in a prior trial and.12 The sequences and known immunogenic MEP for each peptide are listed in table 1.

Montanide ISA-51 was purchased from Seppic Inc (Fairfield, New Jersey, USA) as cGMP material in sterile vials. PolyICLC was provided as a clinical grade reagent (Hiltonol; Oncovir, Washington, DC, USA) by the Ludwig Institute for Cancer Research and its Cancer Vaccine Consortium. Resiquimod was provided by 3M Pharmaceuticals (St Paul, Minnesota, USA).

Participants were vaccinated with LPV7 (300 mcg long peptide/dose) plus Tet (200 mcg/dose) on the schedule shown in figure 1. Tet was included to provide a stimulus for helper T cell responses to support CD8⁺ T cell responses,14 15 32 because the extent of CD4⁺ T cell response to LPV7 was not known in advance. Participants were adaptively assigned to one of seven adjuvant combinations (figure 1 and online supplemental table 1) and were enrolled at two institutions. The adaptive assignment included equal randomization and allocation among allowable arms until a weighted allocation scheme or the modeling stage was triggered. Vaccine sites were rotated between upper arm and thigh, avoiding any extremity that had undergone lymph node removal. Each vaccine was administered in one skin site, with 1 mL volume injected into subcutaneous (sc) tissue and 1 mL volume injected intradermally (id), through the same skin puncture site. Each participant underwent three 4 mm punch biopsies of skin at the vaccine injection site at days 8 and 22. Control biopsies of normal skin were also collected at these time points for the first six participants enrolled.

Adverse events (AEs) were collected continuously from the time of first injection to study completion and the AEs were graded using NCI CTCAE V.4.03. Dose-limiting toxicities (DLTs) were defined as any unexpected, treatmenft-related AE that was ≥Grade 1 ocular AE, ≥Grade 2 allergic/autoimmune reaction, or ≥Grade 3 (any) except for Grade 3 injection site reaction (ISR) with ulceration ≤2 cm.

Peak CD4⁺ and CD8⁺ T cell responses in the peripheral blood were evaluated by direct (ex vivo) ELIspot assay for IFN-gamma, as reported.14 This was performed in a core laboratory dedicated to immune monitoring, with standard operating procedures and quality assurance measures.14 Participants were evaluable for immune response if at least one post-treatment sample was measurable for response. A minimum of 4 weeks of data was used to guide decisions about the range of optimal dose combinations. Durability of T cell responses was assessed at 26 weeks and for longer intervals when available, with the hypothesis that durability of T cell responses would be enhanced compared with prior experience with short peptides and would be improved with TLR agonists when compared with IFA alone in this study.

The short peptides (9–12 amino acids long) contained within each long peptide (table 1, underlined) represent MEPs for CD8⁺ T cells. A CD8⁺ T cell response to these peptides was defined as at least a twofold increase in IFN-gamma-secreting cells over background (maximum of 2 negative controls) and over pre-existing responses at baseline, an increase over background of at least 20 IFN-gamma secreting cells per 100,000 CD8⁺ T cells, and no overlap in SD with the negative control, as described.14 A CD4⁺ T cell response to the Tet was defined similarly but with at least 20 IFN-gamma secreting cells per 100,000 CD4⁺ T cells. A T cell response to the full-length LPV7 required an increase over background of at least 10 IFN-gamma secreting cells per 100,000 total peripheral blood mononuclear cells (PBMC). Immune response rate (IRR) is defined as the proportion of subjects with a T-cell immune response and is reported as a point estimate with 90% CIs.

Interassay coefficients of variation (CVs) were calculated for the ELIspot response of two normal donors to a pool of viral peptides (CEF peptide pool33): for the high and low responders, mean numbers of spots per 200,000 cells were 271 and 42, respectively, and CVs were 28% and 26%, respectively.

Sera were evaluated by ELISA for IgG Ab to the LPV7 peptide pool at weeks 1, 7, 13, and 26, using methods described.12 Briefly, 96-well half-area cluster plates (Corning Costar) were coated with 30 mcL of LPV7 peptides (pooled) diluted in carbonate/bicarbonate buffer (pH 9.4; Sigma-Aldrich) at 1.67 mcg/mL of each peptide. For quantitation of specific serum levels of anti-peptide Ab, purified IgG immunoglobulin (Fitzgerald Industries International) was prepared in coating buffer at 1 mcg/mL, serially diluted fourfold to 0.25 ng/mL, and 30 mcL of each dilution added to duplicate wells. After incubation overnight at 4°C, plates were washed with phosphate-buffered saline (PBS) with 0.1% Tween 20 (TPBS), then blocked 1 hour with 5% nonfat dry milk in TPBS (blocking 143 buffer). Beginning at 1:100, fourfold serial dilutions of participant and control sera were prepared in blocking buffer and added to individual wells. After 2 hours at room temperature (RT) and washing, secondary Ab (goat anti-human IgG AP conjugate, Southern Biotech) was added to all wells, incubated 1 hour at RT, then washed. Attophos substrate (Sigma) was added to each well for 30 min. 3N NaOH was added to stop the reaction, and fluorescence was recorded on a Molecular Devices SPECTRAmax Gemini EM Fluorescent plate reader, excitation 450 nm, emission 580 nm. The FORECAST function in Microsoft Excel was used to calculate the Ab titer of participants' sera,12 which was defined as the reciprocal of the serum dilution that yields a fluorescent intensity 10 times greater than the cut-off value. The cut-off value was defined as the average fluorescence obtained from the first four dilutions of serially diluted normal donor serum (negative control). Ab titers≥100 were considered positive. For participants with Ab titers that were >100 in prevaccine sera, a vaccine-associated response also required a titer increase of at least 10-fold (1 log).

The study was designed with two stages. The initial stage accrued participants in cohorts of one per arm (randomized within a zone) until a participant experienced a DLT. The escalation plan for the first stage was based on grouping treatment combinations into three zones (online supplemental table 1), beginning with Zone 1. With this dose-escalation design, participants could be accrued and assigned to other open combinations within a zone, but escalation would not occur outside the zone until the minimum follow-up period of 3 weeks was observed for the first participant accrued to a combination. A continual reassessment method (CRM)34 35 directed enrollment in the second stage. This method used a selected set of possible orderings of combinations for the DLT probabilities and a working model for the DLT probabilities under each ordering. The CRM model was used to fit the working model with the accumulated data. In the event of a tie between the likelihood values of two or more orderings, then the selected order of combinations was chosen at random from among the tied orderings. The DLT probabilities defined a set of acceptable combinations with a toxicity tolerance of 33%. Assuming at least one optimal combination existed, up to 52 evaluable participants could have been accrued to determine the optimal combination. Simulation results were run to display the performance of the design characteristics, which have been reported.36 Immune responses were assessed overall, but differences between enrolling institutions were also explored in the context of data on PBMC function.

The study was not designed to make statistical comparisons between arms. Frequency and magnitude of treatment-related adverse events (TRAEs) were summarized by arm. IRR for defined categories was estimated as point estimates with 90% exact CIs. Graphical representations were used to present study outcomes. Fisher’s exact test was used to assess associations of maximum immune response to maximum TRAE grade and other select AEs. Disease-free survival was defined from start of treatment to recurrence/progression or death from any cause, whichever occurred first. Participants who did not experience an event were censored at date of last contact. Overall survival was defined as the time from start of treatment to time of death from any cause. Disease-free survival and overall survival distributions were estimated by the product-limit method of Kaplan and Meier.

Total enrollment was 51 participants; however, one did not receive study treatment. Thus, demographic, safety, and immunologic data are reported for 50 participants who were enrolled and treated. These included 30 males (60%) and 20 females (40%), with 5, 7, 4, 6, 16, 6, and 6 treated on arms A-G, respectively (online supplemental file 1, CONSORT diagram). Most patients had Eastern Cooperative Oncology Group (ECOG) performance status (PS) of 0 (90%) and stage III disease at registration (82%). Details are provided in online supplemental table 2. Four patients experienced new metastatic disease on study (arms D (1), E (1), G (2)), but two of them had received all six vaccines (arms E and G): the others received three and four vaccines, respectively; so, all 50 were evaluable for toxicity, immune responses, and clinical outcomes (online supplemental file 1, CONSORT diagram).

There was one DLT across all groups: Group E (Grade 3 ISR, 6% in that group), after receiving all six vaccines. All other AEs were Grades 1–2; most common were ISR (90%), fatigue (62%), skin induration (56%), chills (48%), fever, myalgia, and headache (30%), and were similar across groups, but lower for Group C (online supplemental table 3). No study combinations crossed the statistical boundaries for unacceptable toxicity.

Each of the seven long peptides contained an MEP for CD8 T cells restricted by HLA-A1, A2, A3, B35, or B51 (table 1). T cell responses to these MEPs were assessed singly or as peptide pools by ex vivo ELIspot assay for each patient with the appropriate HLA expression. Example data for a T cell response to the MEP tyrosinase_240-251S (DAEK) are shown in figure 2D; however, the data in that plot represent raw values before adjusting for the percent of CD8⁺ T cells in the PBMC. After that adjustment, there were 56.2 IFN-gamma-secreting cells per 100,000 CD8⁺ T cells, with negative control values of 3.0 (not shown). Overall, CD8⁺ T cell responses were detected to MEPs in 9 patients (18%, 90% CI 10 to 29) with the best IRRs of 40% and 25% in arms A and E, respectively (table 2). The magnitude of those T cell responses to MEPs is shown by arm in figure 3. Among the nine participants with responses to MEPs, six had responses both to individual peptides and to the MEP pool, two had responses just to the pool, and one responded to an individual peptide. Responses to individual peptides were detected to tyrosinase _240-251S, NY-ESO-1 _94-102, gp100 _17-25, and MAGE-A1 _96-104 in four, two, one, and one patient, respectively, which represent 19%, 17%, 5%, and 5% of participants with the corresponding Class I MHC. In a prior trial using 12 MEPs+Tet+IFA in 41 participants (Mel44 trial, NCT00118274), CD8⁺ T cell responses were detected ex vivo to six of the MEPs using the same assay criteria. The IRRs to those six peptides in the prior study ranged from 4% to 78%, but IRRs to those same peptides in the present study ranged from 0% to 19%. Differences between IRR to these peptides between the studies ranged from 4% to 78% (online supplemental table 4).

Immune responses to the Tet peptide provide another measure of immunogenicity with each adjuvant preparation. Example data for responses in four patients are shown in figure 2. Overall, responses were detected in 20 subjects (40%, 90% CI 28 to 53), with the highest IRRs of 80% in Arm A, 50% in Arms E and G, and 43% in Arm B (table 2). Data for immune response magnitude are summarized in figure 3.

This study was not designed to make comparisons among arms, but an exploratory assessment shows that the overall proportion of patients with T cell responses to MEPs, LPV7, and Tet trended higher for adjuvants containing IFA (arms A, E, F, G, n=33) than for those without IFA (IFA (arms B-D, n=17, table 2 and figure 4A). Adjuvants containing polyICLC yielded IRRs trending slightly higher than those without polyICLC for LPV7 and tetanus, but not for the short peptides. There was a trend to lower IRRs with adjuvants containing resiquimod versus those without resiquimod (figure 4A). For participants receiving IFA-containing vaccines, T cell responses were detected at more time points during the study (figure 4B), to more peptides (figure 4C), and with higher fold-increase and absolute magnitude of T cell response (figure 4D–E).

Overall, 26 participants (52%) had immune responses to short peptides, LPV7, and/or Tet. Of the nine with T cell responses to short peptides, five (56%) also had responses to LPV7, and seven (78%) also had responses to Tet. Of 15 with responses to LPV7, 5 had responses to the short peptides and 10 had responses to Tet.

As shown in online supplemental figure 1A, participants with higher grade 2/3 TRAEs had greater magnitude T cell responses to LPV7 (p=0.014), Tet (p=0.022), and MEP (p=0.068) by Fisher’s exact test. Also, grade 2/3 ISR (vs grade 0/1) was associated with higher immune response to MEP, LPV7, and Tet (p=0.008, 0.002, 0.009, respectively; Online supplemental figure 1B). Grade 2 induration at injection sites (vs 0/1) was associated with higher magnitude T cell responses to LPV7 (p=0.001) and to Tet (p=0.002), but not to MEPs (p=0.4) (online supplemental figure 1C).

IgG Ab responses to LPV7 were evaluated in 44 of the 50 participants. The induced IgG titers were highest in arms E and G and lowest in B and C (figure 5A–C). The proportions of participants with serum IgG response by week 7 were 0%, 20%, 43%, 50%, 75%, 83%, and 100% for arms C, A, B, D, G, F, and E, respectively, and increased by week 26%–100% for arms D–G, 80% for arm A, 57% for arm B, 25% for arm C, and 84% overall (figure 5D). For IFA-containing adjuvants, Ab responses were observed in 79% by week 7 and 97% overall. On the other hand, for adjuvants without IFA, Ab responses were observed in 33% by week 7% and 60% overall. IgG responses to Tet were also observed for 12/44 evaluable subjects (27%), including 75% (3/4) on arm G, 50% (7/14) on arm E, and one each on arms D and F (data not shown). IgG responses to HIV gag were assessed as negative controls with 42/44 (95%) negative and only 2 having low level titers (max 808 and 162) that presumably represent false positives (data not shown).

The overall immune response evaluation was performed for the full dataset in accord with the study design. However, differences were noted in T cell immune response by institution, details for which are in online supplemental table 5 . All T cell responses to LPV7 and to MEP were observed for participants treated at Institution 1, the site where immune analyses were performed. Most responses to Tet were also observed at Institution 1, but some were observed at Institution 2. The proportions with T cell responses by institution are shown in online supplemental figures 2A, B. Blood samples were shipped overnight from Institution 2 to Institution 1; thus, the possibility that shipping conditions may have reduced T cell viability and function was explored. Controls for each analysis included response to two mitogens (phytohemagglutinin (PHA) and phorbol myristate acetate (PMA) plus ionomycin) as well as a pool of over 30 short peptides from viral proteins (CMV, EBV, influenza) restricted by Class I MHC molecules (CEF peptides). The number of cells producing IFN-gamma in response to CEF peptides varied widely among participants, which is expected, but ranges appear similar between institutions (online supplemental figure 2C, D). PMA responses also were similar between sites: p=0.93 (median 1534 vs 1427 per 100,000). However, the responses to PHA for Institution 2 were significantly lower than for Institution 1 (p=0.027; median 181 (95% CI 129 to 231)) vs 265 (214 to 326) (online supplemental figure 2C, D). Interestingly, the serum IgG (antibody) response to the LPV7 peptides was high for both sites, without appreciable differences in response frequency or magnitude (online supplemental figure 2E–G). Thus, participants at both institutions did receive vaccines and developed immune responses.

We had performed two prior multicenter clinical trials of melanoma vaccines incorporating 12) Class I MHC-restricted peptides (12MP) plus Class II MHC-restricted peptides (either Tet or six melanoma helper peptides, 6MHP):Mel4337 and Mel44.15 To explore whether this difference in IRR by institution had been observed previously, we reviewed the data on IRRs as a function of institution and found that in both of those trials, the IRRs were comparable across all sites, including the two institutions participating in the current trial (online supplemental figure 3).

With a median follow-up of 4.7 years, 43 (86%) of patients remain alive, with 32 (64%) without known disease recurrence. For arms with IFA-containing adjuvants (A, E, F, G), recurrences have occurred in 11/33 (33%, 90% CI 20 to 49), vs 7/17 (41%, 90% CI 21 to 64) in those without IFA. Deaths due to melanoma have occurred in 3/33 with IFA (9%, 90% CI 3 to 22) vs 4/17 (24%, 90% CI 8 to 46) without IFA (arms B, C, D). Kaplan-Meier plots for disease-free and overall survival in the whole study population are shown in online supplemental figure 4.