Thirty-one candidate peptides were available for PPV. All 31 peptides were cytotoxic T lymphocyte (CTL) epitopes restricted to the HLA-A2, -A24, or -A3 supertypes (A3, A11, A31, or A33), or HLA-A26 of HLA-class Ⅰmolecules (12.13). Twelve peptides were for the HLA-A2, 14 for the HLA-A24, and 9 for the HLA-A3 supertypes, and 4 were matched for HLA-A26 of cancer patients as reported previously (Table SI). The peptides were prepared under conditions of Good Manufacturing Practice using a Multiple Peptide System (San Diego, CA). Patients were vaccinated with 2 to 4 peptides based on human leukocyte antigen (HLA) type and pre-existing immunity by measuring peptide-specific immunoglobulin G (IgG) levels. Each of the selected peptides was mixed with incomplete Freund's adjuvant (Montanide ISA-51VG; Seppic, Paris, France) and injected subcutaneously into the inguinal, abdominal, or lateral thigh areas as 1.5 ml emulsion (3 mg/each peptide).

All protocols were approved by the ethical committee of Kurume University at first followed by the regional ethical committee (Fukuoka clinical research board <Number 718004>, and then registered in the UMIN Clinical Trials Registry of Japanese government. Detailed protocols are presented online only (https://upload.umin.ac.jp/cgi-open-bin/ctr/index.cgi?function=02) where the protocols were written by Japanese with brief translation to English. In brief, there were 3 different protocols with regard to the vaccination intervals. The patients who became resistant to the standard systemic therapies (n=180) received 6 injections of PPV at 1-week intervals (the 1st cycle) followed by 6 injections at 2-week intervals (the 2nd cycle) (protocol PRT1; UMIN registration numbers: 000001482, 000001881, 000006295, 000019390). Cancer patients in the early or any other stages except those who became resistant to the standard systemic therapies (n=109) received 4 injections of PPV at 1-week intervals and then 4 injections at 2-week intervals (the 1st cycle), followed by 4 injections at 2-week intervals and then 4 injections at 4-week intervals (the 2nd cycle) (protocol PRT2; UMIN registration numbers: 000006297, 000011593, 000029789). In these two protocols, patients received 1.5 ml emulsion (3 mg/each peptide) of 2 to 4 peptides at each visit for the vaccination. The third protocol was used for cancer patients who lived in the northern parts of Japan or in other countries (n=20) and thus both the patients and family needed to stay in Kurume at least two days; they received 4 injections of PPV at 4-week intervals (the 1st cycle) followed by the same schedule for the 2nd cycle (protocol PRT3; UMIN registration numbers: 000006927, 000011230). In this third protocol, patients received 3.0 ml emulsion (6 mg/each peptide) of 2 to 4 peptides at each visit for the vaccination, with half the dose injected into either side of the body. In all three protocols, patients who wished to continue the PPV after the 2nd cycle received the vaccination at 2- to 12-week intervals until withdrawal of consent or unacceptable toxicity. All these studies were in accordance with the Declaration of Helsinki and the International Conference on Harmonization of Good Clinical Practice guidelines, and the studies were conducted in an outpatient setting. Written-informed consent to participate in the clinical trial and to use their data for research and publication purposes was obtained from all individual participants before their inclusion in the study.

The phase 2 clinical trials of PPV were conducted from November 2008 to March 2018 at the Cancer Vaccine Center of Kurume University and Kurume University Hospital. A total of 309 patients with pancreatic cancer were enrolled, 301 patients with advanced-stage disease (258 with stage IV and 43 with recurrent disease) and 8 with non-advanced stage disease (stages I-III). The clinical outcomes for some of these patients have been partly reported previously (15,16). Eligibility criteria at the time of entry were a pathologically confirmed diagnosis of pancreatic cancer for those patients for whom a tumor sample was available (n=145) or clinically diagnosed pancreatic cancer for those without a tumor sample (n=160), positive IgG responses specific to at least 2 of the HLA-class-IA matched peptides in pre-vaccination plasma, positive status for the HLA-A2, -A24, or -A3 supertypes (A3, A11, A31, or A33) or positive status for HLA-A26, age of ≥20 years, Eastern Cooperative Oncology Group (ECOG) performance status (PS) of 0 or 1, life expectancy of at least 12 weeks, and adequate bone marrow function, hepatic function, and renal function. Exclusion criteria were acute infection, a history of severe allergic reactions, or other systemic diseases. The histological diagnoses for the 145 patients with available tumor samples consisted of adenocarcinoma (n=126), invasive ductal carcinoma (10), intra-ductal papillary mucinous carcinoma (IDMC) (5), mucinous cystadenocarcinoma (3), or neuroendocrine tumor (1). In all the protocols, concomitant chemotherapy regimens were permitted, including but not limited to gemcitabine (GEM), TS-1 (an oral fluoropyrimidine derivative), GEM+ nab-paclitaxel (nabPTX), GEM+TS-1, and leucovorin+5-fluorouracil+irinotecan+oxaliplatin (FOLFIRINOX). All patients provided written informed consent for study participation and data collection.

Peripheral blood (30 ml) was taken from patients before and after each cycle of vaccinations. Plasma was separated after centrifugation and frozen until use. The IgG responses specific to the vaccine peptides were measured in plasma by a Luminex system as reported previously (8-11).

The Kaplan-Meier method, log-rank test, Cox proportional hazards analysis, Student's t-test, chi-square test, and Fisher's exact test were used for the statistical analyses. OS was calculated as the time in months from the date of study enrollment to death or to the date of last contact. The data-cutoff for OS was June 2018. Time-to-event endpoints were analyzed using the Kaplan-Meier method, and between-group comparisons for OS were conducted using the log-rank test. All reported P-values were two-sided, and P-values of <0.05 were considered significant. JMP version 12 or SAS version 9.4 software (SAS Institute Inc.) was used to perform all analyses.

The baseline characteristics of the 309 pancreatic cancer patients are shown in Table I. There were 171 males and 138 females, and the median age was 64 years. HLA-status was HLA-A24 in 180, HLA-A2 in 130, and HLA-A3 supertypes (A3, A11, A31, A33) in 161 patients, and HLA-A26 in 58 patients. Performance status (PS) was 0 in 241 and 1 in 68 patients. Clinical stages were stage I in 3, II in 2, III in 3, and IV in 258 patients, and there were 43 recurrent cases. The number of chemotherapeutic courses prior to the vaccination was 0, 1, 2, 3, or 4 in 28, 136, 108, 29, or 8 patients, respectively. Chemotherapy regimens administered during the 1st to 2nd courses of PPV were none (n=59 patients), gemcitabine (GEM) (59), S-1(65), GEM+S-1(40), GEM+ nab-paclitaxel (14), and others (71), respectively. The median number of vaccinations was 9, with a range of 1 to 60. Informed consent was obtained from all patients prior to entering the study.

Eighty-one of 309 patients (26%) failed to complete the 1st cycle of PPV due to rapid disease progression. All 81 of these patients were in the advanced disease stages. Significant differences were seen between the 81 patients who did not complete the 1st cycle and the remaining 228 patients who did complete the 1st cycle in regard to PS (worse in the former group), number of chemotherapy courses prior to PPV (larger in the former), and the percentage of patients receiving chemotherapy combined with PPV (lower in the former) (Table I). In addition, the white blood cell and neutrophil numbers along with c-reactive protein (CRP), a typical inflammatory protein, prior to study entry were significantly higher in the 81 patients who did not complete the 1st cycle, whereas the red cell and lymphocyte numbers were significantly higher in the 228 patients who did complete the 1st cycle (Table II).

The numbers of grade 1, 2, 3, 4 or 5 AEs were 1,018, 555, 190, 14 and 86, respectively (Table SII). The most frequent adverse event was injection site reaction (grade 1: n=403; grade 2: n=55; grade 3: n=8). Frequently observed grade 3 AEs were elevation of GGT (n=24), anemia (21), lymphopenia (20), injection site reaction (8), ascites (7), glucose intolerance (7), and anorexia (7). Grade 4 AEs were anemia (n=4), hepatobiliary disorders (n=3), lymphopenia (2), leukocytopenia (1), neutropenia (1), colonic obstruction (1), duodenal stenosis (1), and hypercalcemia (1). Frequently observed grade 5 AEs were neoplasms (44), multi-organ failure (17), hepatic failure (4), and respiratory failures (3). According to the assessment by an independent safety evaluation committee in this trial, all of these severe AEs, except injection site reaction (8), were related to the cancer progression or the combination chemotherapies.

Pre-vaccination peptide-specific IgG titers to each of the 31 peptides were measured using a Luminex system with a cut-off level of 10 fluorescence intensity units (FIU) taken as a detectable level of IgG as reported previously (8-12). The positive rate of antibody in the patients' plasma (n=309) was largely different among the 31 peptides, ranging from 12 to 90% with a median positive rate of 56%; Table III lists the peptides in order from highest IgG positivity to lowest. The magnitudes of IgG titers for the 31 peptides among the patients showing detectable levels were also different from each other, with the median FIU being 35 (range: 23-132) among the 31 peptides (Table III).

Post-vaccination peptide-specific lgG levels were measured at the end of both the 1st cycle and 2nd cycle in plasma from the 228 patients who completed at least the 1st cycle of vaccination. It was considered to be a positive immune response when the post-vaccination IgG titer at the end of either the 1st or 2nd cycle was two times higher than the pre-vaccination titer (8-10). Six peptides (PSMA-624, Lck-208, EZH2-735, MRP3-503, UBE2V-85 and Lck-422) with <10 cases of evaluable patients were excluded in the following analysis to avoid a possible bias. Under these circumstances, the percentage of patients showing positive IgG responses among the 228 patients who completed at least the 1st cycle differed greatly for the 25 peptides, ranging from 25 to 79%, with a median rate of 55% (Table III). The magnitudes of IgG titers of the 25 peptides among the patients showing the positive responses were also very different, with a median FIU of 3,278 among the 25 peptides, which was 89-fold greater than the pre-vaccination level (37 FIU, as shown above).

We also investigated whether chemotherapy suppressed the vaccination-induced immune boosting by comparing the rate and magnitude of IgG boosting between the PPV patients with and those without chemotherapy (Fig. 1). IgG boosting was observed in post-vaccination plasma in 23 of 27 (85%) PPV patients without chemotherapy (10 of 10 patients (100%) who declined chemotherapy of their own will and 13 of 17 patients (76%) who could not tolerate chemotherapy), 42 of 45 (93%) with GEM, 5 of 7 (71%) with GEM + nabPTX, 27 of 28 (96%) with GEM+TS-1, 43 of 48 (90%) with TS-1, and 63 of 65 (97%) with the other chemotherapies (Fig. 1). We further investigated the n-fold increase of IgG boosting from the pre-vaccination to the post-vaccination IgG titers using the IgG levels shown in Fig. 1. The increase was 7.0-fold in 27 PPV patients without chemotherapy (13.5-fold in the 10 patients who declined chemotherapy of their own will, and 2.9-fold in the 17 patients who could not tolerate chemotherapy), 3.3-fold in the 45 patients with GEM, 27.9-fold in the 7 patients with GEM+nabPTX, 2.9-fold in the 29 patients with GEM+TS-1, 2.8-fold in the 48 patients with TS-1, and 11.6-fold in the 65 patients who received other chemotherapies. These results suggested that the combined chemotherapy did not suppress PPV-induced boosting of peptide-specific IgG from the viewpoint of either the positive rate or the magnitude of IgG boosting.

The median OS of 309 patients was 5.8 months (M) with a 95% confidence interval (CI) of 5.2-6.7 M from the 1st vaccination, while it was 17.6 M with a 95% CI of 16.0-19.5 M from the initial diagnosis (Table I). The median time from the 1st vaccination of the 81 patients who failed to complete the 1st cycle of PPV due to rapid disease progression was much shorter than that of the remaining 228 patients who completed the 1st cycle (2.1 M, 95% CI: 1.8-2.7 vs. 8.4 M, 95% CI: 8.4-9.9; P<0.01), although that from the 1st diagnosis was not different between the 81 patients who failed to complete the 1st cycle and the remaining 228 patients (11 vs. 11 M) (Table I).

The 309 patients were also divided into those with non-advanced (stages I-III) and those with advanced (stage IV or recurrence) stage disease to better understand the role of prophylactic or therapeutic PPV for pancreatic cancer, respectively. The median OS of the 8 patients with stages I to III disease from the 1st vaccination or from the initial diagnosis was 62.6 M (95% CI: 11.2-not reached) (Fig. 2A) or 72.3 M, (95% CI: 16.1-not reached) (data not shown), respectively. All 8 of these patients were over 60 years old (range: 61 to 83 years). Six and two patients were histologically diagnosed as having adenocarcinoma and IDMC, respectively. Seven patients and one patient received RO or R1 surgery, respectively. Recurrence was observed in 4 patients, and progression free survival and OS from the 1st vaccination were 42 and 63 M in the stage I adenocarcinoma case, 38 and >39 M in the stage III IDMC case, 6 and 53 M in the stage III adenocarcinoma case, and 6 and 11 M in the stage III adenocarcinoma case, respectively. The remaining 4 patients with PPV alone were free from recurrence.

The median OS of the 301 patients with advanced-stage disease (stage IV or recurrence) was 5.6 M (95% CI: 5.6-6.4) (Fig. 2A) from the 1st vaccination. Among them, no significant difference in the median OS was seen between the stage IV (n=258) and recurrent cases (n=43) (5.6 M, 95% CI: 5.0-6.3 vs. 7.2 M, 95% CI: 4.0-9.8; P=0.33) (data not shown). No significant difference in the median OS was found among the different types of HLA-class I (180A24+ patients: 5.7 M, 95% CI: 5.0-6.7 M; 130A2+ patients: 5.8 M, 95% CI: 5.1-7.7; 161A3 supertypes+ patients: 5.8 M, 95% CI: 4.8-6.7; 43 A26+ patients: 4.8 M, 95% CI: 3.9-10.0) (data not shown). In contrast, the median OS in the patients under PTR1 (n=180, 5.3 M, 95% CI: 4.6-6.1 M) was shorter than that of the patients under PRT2 (n=109, 7.7 M, 95% CI: 5.6-10.0) or PRT3 (n=20, 6.5 M, 95% CI: 3.7-12.1) (P<0.001), primarily since the vast majority of the patients under PRT1 failed to respond to all the available standard chemotherapies prior to entry into the PPV study (data not shown).

We investigated the effect of pre-vaccination chemotherapy status on OS. The median OS values of the 301 patients with advanced cancer were inversely correlated with the number of chemotherapy regimens conducted prior to the vaccination (Fig. 2B). The median OS values of the patients with 0 (n=20), 1(136), 2(108), 3(29), or 4(8) courses of chemotherapy prior to the vaccination were 9.4 M (95% CI: 4.9-33.1), 5.9 M (95% CI: 4.9-8.5), 5.6 M (95% CI: 4.6-6.5), 4.7 M (95% CI: 3.3-8.5), and 1.9 M (0.6-2.8), respectively (P<0.001).

We then investigated the effect of combination chemotherapies on the OS of the 301 patients with advanced cancer. The median OS values of patients who refused (n=19) or could not tolerate (n=33) combination chemotherapy were 5.4 M (95% CI: 1.9-9.6) and 2.9 M (95% CI: 1.8-3.4) (P=0.03), respectively. Those for the patients receiving the various chemotherapies were as follows: GEM (n=59) (6.1 M, 95% CI: 4.6-7.3), S-1 (n=65) (4.9 M, 95% CI: 4.0-7.8), GEM and TS-1 (n=40) (5.4 M, 95% CI: 4.6-9.4), GEM and nab-paclitaxel (n=14) (17.3 M, 95% CI: 3.4-24.1), and other chemotherapy regimens, including 6 cases of FORIFIRI (n=71) (7.9 M, 95% CI: 5.8-11.5) (data not shown). The median OS of the patients who received PPV alone for at least the 1st to 2nd cycles of PPV either because they refused the treatment of their own will or because they could not tolerate chemotherapy (n=52) (3.1 M, 95% CI: 2.0-4.1 M) was significantly shorter than that of the patients who received PPV in combination with chemotherapy (n=249) (6.2 M, 95% CI: 5.5-7.4 M) (P=0.01) (Fig. 2C). PPV combined with chemotherapy might be more appropriate for patients who had previously undergone a smaller number of chemotherapy regimens.

The median OS of the 208 patients who exhibited IgG boosting (9.2 M, 95% CI: 7.3-10.1) was significantly (P<0.01) longer than that of the 20 patients who did not show IgG boosting (4.9 M, 95% CI: 2.8-5.6) (Table II), confirming the results reported previously (8-14). We also examined whether this was also the case for each of the 31 peptides in the 309 patients with pancreatic cancer. Six (PSMA-624, Lck288, EZH2-735, MRP3-503, UBE2v-43, and Lck422) of the 31 peptides that were used in only a few cases (<10 tested cases) were excluded from this analysis to avoid a possible bias (Table III). The median OS of the patients that exhibited IgG boosting against each of the 17 of 25 peptides (65.4%) that were used for >10 cases was significantly longer (P<0.05) than that of the patients with no IgG boosting (Table III). No significant difference was found in the remaining 7 peptides.

Finally, we investigated the correlation between the OS and pre-vaccination CRP level or neutrophil numbers in each of the 309 patients (Fig. S1). The higher CRP levels or neutrophil numbers seemed to be associated with shorter OS. Indeed, the median OS of the patients having higher CRP levels or higher neutrophil numbers (median or higher than median value) was significantly shorter than that of the patients with lower values for these parameters (Fig. 3A and B), respectively. The opposite was true in the case of lymphocyte numbers (Fig. 3C). Similar results were obtained in the 228 patients who completed at least the 1st cycle of the vaccination (data not shown). In contrast, only a higher CRP level, but not either higher neutrophil numbers or lower lymphocyte numbers, was an unfavorable biomarker at the statistically significant level for the 81 patients who failed to complete the 1st cycle of the vaccination (data no shown).

We also investigated the correlation between the OS and either pre- or post-vaccination IgG levels against the vaccinated peptides in all 309 patients, but significant levels of correlation were not observed (data not shown). Then, we examined the correlation between the OS and the IgG levels for each of the vaccinated peptides in each of the 228 patients who completed at least the 1st cycle of the vaccination (Fig. S2). The higher levels of the increased IgG seemed to be associated with longer OS. Indeed, the median OS of the patients with higher increased IgG levels (median or higher than the median value) was significantly longer than that of their counterparts with lower IgG levels (10.3 M, 95% CI: 7.7-12.0 vs. 7.0 M, 95% CI: 5.6-8.6) (Fig. 4).