Recombinant adenoviruses have several advantages for use in the active immunization of patients with cancer [reviewed in (12
)]. Genes encoding tumor antigens can be readily inserted into replication-incompetent adenoviruses that can be safely administered in large doses to humans. Adenoviruses have a broad range of infectivity of various cell types and have been popular for use in gene therapy (16
). Vaccines based on recombinant viruses have been widely used in multiple animal species (12
). In murine tumor models, immunization with recombinant adenovirus-encoding model tumor-associated antigens has resulted in successful prevention as well as treatment of metastatic disease. Significant reduction of established pulmonary metastases occurred when 108
pfu of recombinant adenovirus expressing the model β-galactosidase tumor antigen was administered to mice bearing pulmonary micrometastases (10
). The administration of exogenous IL-2 significantly augmented this antitumor effect.
In recent years, multiple tumor-associated antigens present in human melanomas have been identified and characterized [reviewed in (1
)]. TILs obtained from HLA-A2-positive individuals predominantly recognize the MART-1/MelanA and gp100 nonmutated, melanoma–melanocyte differentiation antigens expressed by the great majority of melanomas and these melanoma–melanocyte differentiation antigens are also expressed on melanocytes but not other normal tissues or other tumors (3
). The adoptive transfer of TILs that recognize MART-1/MelanA and gp100 has been associated with regression of cancer in patients with metastatic melanoma (34
To evaluate the therapeutic potential of immunization with recombinant adenoviruses expressing tumor antigens, replication-incompetent adenoviruses were constructed that encoded either the MART-1 or gp100 antigens in deleted E1A and E1B portions of the viral genome (28
). The current study has evaluated the safety, immunologic, and potential therapeutic aspects of the immunization of patients with metastatic melanoma using these recombinant adenoviruses.
The patients in this study all had metastatic melanoma, many had been heavily pretreated, and all had progressive disease at the time of entrance in this phase I study. The administration of recombinant adenoviruses at doses up to 1011
pfu per injection was well tolerated, with no significant side effects associated with up to four sequential doses given at monthly intervals. One of 16 patients who received recombinant adenovirus MART-1 alone achieved an objective response (; ), which is ongoing at 31 months. The objective response rate in patients who received recombinant adenovirus in conjunction with high-dose IL-2 was 16% and was similar to the 17% objective response rate seen in 134 consecutive patients with metastatic melanoma treated by us with this regimen of high-dose bolus IL-2 alone in prior studies (36
). Thus, except for the single responding patient to adenovirus MART-1 alone, there was no evidence that the adenoviral immunizations enhanced the antitumor effects seen with IL-2 administration.
In previous studies using the same in vitro
assay used here, we had demonstrated that the administration of an immunodominant MART-1 peptide (MART-1:27–35) in IFA or the administration of several gp100 peptides (gp100:209–217 or gp100:280–288) in IFA could increase the immune precursors present in immunized patients reactive against peptide and tumor (7
) and that clinical responses could be enhanced by IL-2 administration (9
). Because patients with melanoma have pre-existing immune reactivity against the MART-1 and gp100 antigens, it was necessary to compare preimmunization and postimmunization samples (7
). In patients receiving multiple injections of 0.1–10 mg of the immunodominant MART-1 peptide in IFA, 15 of 16 postvaccination PBMCs demonstrated at least a threefold increase in immune reactivity compared with prevaccination PBMCs (7
). Similarly, immunization with the two gp100-immunodominant peptides (gp100:209–217 and gp100:280–288) could successfully immunize patients as well (8
). Modification of the gp100:209–217 peptide by substituting anchor amino acids to increase binding to HLA-A2 substantially increased its immunogenicity. Administration of this modified peptide plus IL-2 resulted in a 42% objective response rate in patients with metastatic melanoma (9
In the present study, patients receiving adenovirus MART-1 or adenovirus gp100 showed only sporadic evidence of HLA-A2-restricted reactivity to either the MART-1:27–35 peptide or the gp100:154–162, 209–217, or 280–288 peptides. Given that the adenoviral vectors encode the entire MART-1 or gp100 proteins, one cannot rule out the possibility that reactivity was generated to peptides other than the immunodominant peptides examined here, that peptide reactivity was manifested via a histocompatibility restriction other than HLA-A2, or that assays measuring lymphocyte properties other than specific γ-interferon secretion may have given other results. Thus, only five of the 22 patients tested before and after immunization exhibited some evidence of immunization as a result of the adenoviral administration. It was not possible to assess the antipeptide reactivity in PBMCs derived from 11 of the 34 non-HLA-A2 patients who completed treatment with the Ad2/MART-1 virus, two of whom showed objective clinical response. Therefore, there are insufficient data to permit one to conclude that there was antipeptide T-cell reactivity in PBMCs in those patients who showed objective clinical responses following administration of Ad2/MART-1 virus either with or without IL-2. Similarly, there was little evidence of immunization against the gp100 epitopes in patients receiving adenovirus gp100.
To understand the relative lack of immunizing and therapeutic activity of the recombinant adenoviruses, studies were performed of the antiadenoviral IgG and neutralizing antibodies circulating in the sera of patients prior to immunization. Fifty-two of 53 patients tested had a preimmunization titer of antiadenoviral IgG antibody of 1000 or greater and 48 of 54 pretreatment sera had neutralizing titers of 100 or greater. These high titers of IgG and neutralizing antibodies might be expected to significantly impair the immunizing potential of recombinant adenovirus, although little is known about the level or impact of neutralizing antibodies at a subcutaneous or intramuscular injection site. Neutralizing antibodies can diminish but often not eliminate the efficiency of gene transfer (38
). Furthermore, it should be emphasized that the patient exhibiting a complete tumor regression following administration of adenovirus-MART-1 alone had a serum-neutralizing titer of 400.
More than 75% of patients showed a significant increase in IgG or neutralizing antibody titers after undergoing two immunizations with adenoviral vectors. Thus, the ability to give repeated injections of adenovirus is limited. Reactivity to adenoviral proteins has been seen by others. In a study (42
) of potential gene therapy patients, 57% of adult humans exhibited neutralizing antibodies to adenovirus type 5. In several models, both humoral and cellular responses against adenoviral antigens prevented successful immunization with recombinant adenovirus and could, in fact, lead to destruction of cells expressing the adenoviral genes (43
). Thus, the immune response to adenoviral proteins may be a significant limitation of the ability of recombinant adenoviruses to successfully immunize experimental animals as well as humans. These studies emphasize the need to develop recombinant adenoviruses that are not susceptible to pre-existing immune reactions in patients and a variety of approaches are being explored (47