Alpha fetoprotein DNA prime and adenovirus boost immunization of two hepatocellular cancer patients

Enrollment inclusion criteria included patients 18 years and older, with a history of AFP⁺ HCC, stage II-IVa, after locoregional therapy (resection, RFA, cryoablation, ethanol injection, chemoembolization and radioembolization), with Karnofsky performance status ≥70%, and conserved liver function (Child-Pugh class A or B). Exclusion criteria included acute infection (other than HBV or HCV), HIV infection, organ allografts, or chemotherapy or steroid therapy in the last 30 days.

The first patients screened were excluded for: lack of HLA-A2 positivity (the first screened patient; a criterion later eliminated), very high levels of circulating anti-AdV neutralizing antibodies (patients 2 and 3), early progression (patients 2 and 4), or other non-HCC-related conditions. Patient 7, the first vaccinated, was male, 80 years, Caucasian, HBV^-/HCV^-, with stage II HCC and was vaccinated between 11/2010-2/2011. Patient 8, the second vaccinated, was female, 81 years, Caucasian, was stage II HCC with cryptogenic cirrhosis. She was vaccinated between 12/2010-3/2011. Further clinical details are presented in the Results.

The vials of plasmid DNA (pAFP and pGM-CSF) and the AdV (AdVhAFP) were prepared to GMP by the NCI BRB RAID program (Stephen Creekmore, Chief), and underwent stability testing at months 3, 6, 12 and then annually at the SAIC Frederick, Inc. Vials were stored at -80°C in a temperature controlled and monitored freezer in the University of Pittsburgh Immunologic Monitoring and Cellular Products Laboratory (IMCPL). For each vaccination, vials were transported to the clinic on wet ice, thawed, and drawn into syringes for intramuscular deltoid injections. Before vaccination, patients received 650 mg acetaminophen and 50 mg diphenhydramine. For each of the three monthly plasmid injections, 2.5 mg of pAFP and 2.5 mg of pGM-CSF were mixed together in the syringes before injection. For the dose of 10⁹ pfu of AdVhAFP, the virus was diluted in sterile saline in the IMCPL before preparing the syringe for injection.

All patients were presented at a weekly multi-disciplinary Liver Tumor Board at the University of Pittsburgh Medical Center and deemed eligible for the clinical trial. All patients consented to participate in the AFP vaccine trial which was IRB-approved by the University of Pittsburgh Medical Center. Patients underwent routine history and physical examination, laboratory work, and serum AFP level. Radiologic evaluation consisted of triphasic abdominal CT scan or liver MRI with contrast. Serial imaging was done every 3–4 months after treatment. HCC recurrence was documented by rise in AFP level and CT/MRI imaging showing a recurrent tumor meeting HCC criteria.

Blood samples were transported to the IMCPL on the date of draw immediately were processed according to laboratory SOPs. Serum was isolated from red top vacutainers (BD) by centrifugation after clot formation and frozen in aliquots at -80°C. After removal of whole blood for fresh flow cytometry, PBMC were isolated from the green top vacutainer tubes (BD) containing heparin via Ficoll gradients, and PBMC were cryopreserved in aliquots and stored in liquid nitrogen vapor. Viability was 92-98% upon thaw.

To determine the immune status of the patients and the effects of vaccination, peripheral blood was tested as follows in the IMCPL. All procedures followed SOPs and included healthy donor (HD) controls.

F16 Maxisorp Nunc Immuno plates were blocked with 3% BSA/PBS, then coated with AdVLacZ at 4∙10⁸ pfu/ml for 18–24 hours at 4˚C. Sera were thawed, serially diluted, and plated in duplicate. Plates were incubated for 2 hours at RT, washed 4 times, developed with rabbit anti-goat IgG-peroxidase, and read on a Dynex MRX plate reader. Negative control was HD serum with known low anti-AdV antibody levels, the positive control was HD serum with known high levels of anti-AdV antibodies, and the standard was goat anti-human AdV antibody.

Anti-AdV neutralizing antibodies were measured using serial dilutions of serum (1:4 to 1:512) cultured with an indicator cell line, A549 (ATCC), followed by transduction with an AdV-encoding enhanced Green Flourescent Protein (AdVeGFP), as described [26]. The cells were tested for the MFI of eGFP by flow cytometry. Controls included no AdVeGFP (negative) and no serum (positive).

The IFNγ ELISPOT assay to detect AFP and AdV-specific T cell responses was standardized with PBMC from three HD. Responder T cell subsets were purified before plating (CD8⁺, CD4⁺, Miltenyi Biotec). There were no detectable baseline AFP-specific CD8⁺ or CD4⁺ T cells, and 1 of 3 HD had detectable AdV-specific CD8⁺ T cells, while 2 of 3 HD had AdV-specific CD4⁺ T cells. Backgrounds were 0, and PMA/ionomycin positive controls were >1,000 spots/10⁵ CD8⁺ or CD4⁺ T cells.

Multi-screen HA plates (Millipore, MAHAS4510) were coated with 4–10 ug/mL of monoclonal capture Ab anti-human IFNγ (1-D1K, MabTech), in PBS overnight at 4°C. After blocking the plates with RPMI/10% AB (1 h, 4°C), CD8⁺ or CD4⁺ T cells were plated at 10⁵ cells/well in duplicate or triplicate wells. Autologous DC (0.5-1 × 10⁵ well) were pulsed with different peptides (AFP₁₅₈, AFP₁₃₇, AFP₃₂₅, AFP₅₄₂ (University of Pittsburgh Peptide Synthesis Facility)) loaded with protein (cord blood-derived AFP) or transduced with AdV, rinsed and plated. Control wells contained T cells (alone or with 1 ng/mL PMA + 1 μM ionomycin), and T cells with unloaded DC. Cells were removed and captured cytokine was detected by corresponding biotinylated mAb (MabTech) at 2 μg/mL in PBS/0.5% BSA. After washing, Avidin Peroxidase Complex (Vectastain Elite Kit) was added for 60 min. After rinsing, peroxidase staining was performed with 3-amino-9-ethyl-carbazole (AEC, Sigma) and stopped by rinsing the plates under tap water. Spot numbers were automatically determined with ImmunoSpot imaging system from Cellular Technology, Ltd (Immunospot analyzer software version 5.0). To calculate the number of responding T cells, the mean number of spots detected with DC alone were subtracted from mean spot numbers induced by antigen-loaded DC. A “positive” response was considered ≥ 2x over baseline.

Sera were thawed and simultaneously analyzed with the multiplex Luminex assay (30-plex, Invitrogen) per manufacturer’s protocol in a BioRad reader (IMCPL). The following analytes were tested: GM-CSF, IFNγ, IP-10, MCP-1, TNFα, IL-1β, IL-2, IL-4, IL-5, IL-6, IL-8, and IL-10, IL-15,IL-13, IL-17, IL-1Rα, MIP-1β, IL-2r,IL-7 Exotaxin, MCP-1, MIG, MIP-1alpha and RANTES, in a kit pre-tested for any potential cross-reactivity by the manufacturer. Controls included the kit standard curve and Multiplex QC standards (R&D Systems).

Whole blood was utilized for flow cytometry to determine the frequencies and phenotype of T cells, NK cells, NK/T cells, Treg and MDSC. Blood was processed and stained automatically per manufacturer’s instructions on a TQ-PrepPlus2 (Beckman Coulter), and stained with CD3, CD16, CD56, CD69, CCR7, CD4, CD25, CD39, lineage, HLA-DR, CD14, CD11b, CD33 (Beckman Coulter) and/or intracellular FoxP3 (eBioscience), and analyzed on an FC500 flow cytometer using CXP v2.1 (Beckman Coulter) software.

At baseline, patient 7 was free of active disease (CT scan), had normal AFP (1 ng/ml), Karnofsky performance status of 100%, was Child-Pugh A, was HLA-A2⁺ and had undetectable levels of anti-AdV neutralizing antibodies. At baseline, patient 8 was free active disease but positive for cirrhosis, had normal AFP (5 ng/ml), had Karnofsky performance status 80%, was Child-Pugh A, HLA-A2⁺, and had intermediate detectable levels of anti-AdV neutralizing antibodies which were in the acceptable range for trial enrollment.

Vaccinated patients had no clinically significant adverse events. Patient 7 is an 83 year old male who underwent laparoscopic partial right hepatic lobectomy for HCC on 1/18/07. His background liver was normal, and pathology confirmed a 4.5 cm moderate to poorly differentiated HCC with negative margins. His pre-operative serum AFP level was 2,168 ng/ml, and it normalized to < 2 ng/ml post-resection. He developed a 2 cm HCC recurrence in his liver and underwent laparoscopic radiofrequency ablation (RFA) on 2/19/10 which was 37 months after his initial HCC surgical resection. Liver biopsy at that time confirmed well-differentiated HCC just prior to the RFA. His AFP increased to 14 ng/ml at the time of recurrence, and normalized to 1 ng/ml post-RFA. He received his first AFP vaccine injection on 11/17/10 which was 46 months after his initial liver resection and 9 months after his RFA. He recurred on 8/11/11 when liver MRI showed a new enhancing liver tumor. This was 56 months after his initial liver resection, 19 months after his RFA, and 9 months after his first AFP vaccine injection. His AFP level was in the normal range at 5 ng/ml on 7/25/11. He was started on Sorafenib therapy. His AFP remained in the normal range until 6/7/12 when his serum AFP level increased to 98 ng/ml. It peaked at 508 ng/ml on 11/12/12, and he was treated with yttrium⁹⁰ Theraspheres on 11/12/2012. His AFP normalized to 2.5 ng/ml on 5/7/13, and remained normal at 1.9 ng/ml on 8/20/13. He is still alive as of 10-10-13.

Patient 8 is an 84 year old female who underwent laparoscopic partial right hepatic lobectomy for HCC on 11/6/09. Her background liver was grossly cirrhotic, and surgical pathology of the resected tumor confirmed a 2 cm moderately differentiated HCC with negative margins. Her pre-operative serum AFP level was 61 ng/ml, and it normalized to 4 ng/ml post-resection. She received her first AFP injection on 12/22/10. She had HCC recurrence in her liver documented by CT scan on 6/6/12 which was 31 months after her liver resection, and 18 months after the first AFP vaccine injection. Her AFP level remained in the normal range. She was treated with a brief course of Sorafenib, followed by observation. Her AFP increased to 13 ng/ml on 9/25/13. She is still alive as of 10-10-13, which is 47 months from her resection, and 34 months after the first AFP vaccine injection.

CD8⁺ and CD4⁺ T cell responses to the AFP antigen and AdV vector were tested by direct IFNγ ELISPOT at every time point, and assays were batched to reduce assay variability. HD did not have detectable IFNγ-producing AFP-specific T cells, and 2 of 3 of HD had detectable AdV-specific T cells (>10 spots/10⁵ cells) (Additional file 1: Figure S1). Patient 7 had AFP-specific CD4⁺ T cells at baseline, but no AFP-specific CD8⁺ T cells (Figure 1A). The AFP-specific CD4⁺ T cell frequency fluctuated through the vaccination period, and declined during follow up (d168) but did not increase with vaccination. A low frequency of AFP-specific CD8⁺ T cells were only detected at d168. A very low frequency response to the HLA-A2-restricted AFP₃₂₅ peptide was detected after the first and third pAFP injections, and responses to other characterized AFP epitopes were not detected (not shown). Overall, the AFP-specific immune response to vaccination was very weak in patient 7.

Patient 7 also had detectable baseline CD4⁺ T cells specific to AdV, at almost 2/10³ T cell frequency. These levels were maintained through the plasmid DNA injections, but increased after the AdV boost. A strong induction of AdV-specific CD8⁺ T cells was detected at d140 and d168. These data indicate that the intramuscularly-delivered (i.m.) AdVhAFP was immunogenic and capable of broadly activating a polyclonal T cell response to AdV antigens.

Patient 8 had a very different cellular response to the vaccination regimen (Figure 1B). Her AFP-specific CD8⁺ and CD4⁺ T cell response predominated, and her minimal baseline AdV-specific response decreased and was not increased with the AdVhAFP boost. Her AFP-specific CD4⁺ T cell response was low at baseline, but strongly increased after the third pAFP injection, and remained above baseline through d168. The AFP-specific CD8⁺ T cells were minimally induced by the first two pAFP injections, but highly positive after the AdVhAFP boost, and still high at d168. Responses to individual HLA-A2-restricted AFP peptides were not detected (data not shown). These data demonstrate that the prime-boost regimen could induce high frequencies of circulating AFP-specific T cells, even with a relatively low dose of 10⁹ AdVhAFP viral particles.

These different cellular responses may have been impacted by the levels of circulating anti-AdV neutralizing antibodies in each patient, which might influence the ability to respond to the i.m. injected AdVhAFP boost. Therefore, we examined both anti-AdV neutralizing antibodies and total AdV-specific antibodies in each patient.

Patients were initially screened for trial eligibility to exclude those with very high levels of circulating anti-AdV neutralizing antibodies. Ten HD were tested to standardize the assay, and to identify the range of antibodies likely to be detected (Additional file 2: Figure S2). Most donors had detectable antibodies, 1 of 10 had high levels (pink squares).

Interestingly, patient 7 did not have detectable anti-AdV neutralizing antibodies (Figure 2A), and there was no induction of an antibody response after AdVhAFP injection. In contrast, patient 8 did have a detectable level of neutralizing antibodies (Figure 2B). The level was stable through pAFP injections, was boosted after the AdVhAFP injection, and then stably higher through d168. These data suggest that the lack of anti-AdV neutralizing antibodies in patient 7 does not explain the limited activation of AFP-specific T cells, and that a higher level of anti-AdV neutralizing antibodies in no way inhibited development of an AFP-specific cellular response in patient 8.

Total AdV-specific antibodies were also tested by ELISA. Patients 7 and 8 had levels that were in a similar range with other screened HCC patients and HD (Figure 3). Patient 7 had lower levels and patient 8 had higher levels (similar to their relative neutralizing antibody levels, Figure 2). These total AdV antibodies were not detectably modulated by i.m. AdVhAFP injection.

Fresh flow cytometry was performed at each time point to detect changes in activated T cells and NK cells, as well as frequencies of suppressive cells MDSC and Treg. We hypothesized that high baseline levels of Treg or MDSC might inhibit antigen-specific immune responses, and that the vaccine might activate a detectable percentage of circulating T or NK cells. Differences in several cellular subsets were detected, however it is impossible to draw firm conclusions from the two different patients (Additional file 3: Figure S3). Of note, patient 8 did have a lower frequency of monocytic CD11b⁺/CD33⁺ MDSC, and higher frequencies of activated NK cells (CD56⁺/CD69⁺), which could be examined in future trials to determine whether these subsets are potential immune biomarkers.

Patient sera was tested for a variety of cytokines, chemokines and growth factors by Multiplex Luminex to identify serum biomarkers of response (Figure 4). Patient 8 had higher levels of G-CSF and CXCL10/IP-10, and lower CXCL8/IL-8, MIP-1β, Eotaxin, RANTES, MCP-1, EGF and HGF, relative to patient 7. Patient 8 also had a greater increase in IL-15 than patient 7.

There was no detectable IL-2, IL-5, IL-6, IL-7, IL-13, IL-17, TNFα, IFNγ, GM-CSF or VEGF over background (LLD). IL-1β, IL-4, IL-10, MIP-1α and bFGF were detectable but at very low levels that did not change with vaccination. IL-12p40/p70, IL-1RA and IL-2R were detected at somewhat higher levels, but also showed no change with vaccination. These data serve to suggest specific analytes which may be worthwhile to examine in a larger trial.