We report the Phase I clinical evaluation of a peptide-based candidate HIV vaccine formulated as an emulsion with Montanide ISA-51. The 4 peptides were designed to elicit CD4 T-cell responses to conserved regions in Envelope, and antibody and CD8 T-cell responses to epitopes contained in gp120 V3. To evaluate the vaccine induction of CD8 T-cell responses, 50% of the enrolled subjects were HLA-B7+ which is known to be the restriction element for the V3 (RPNNNTRKSI) CTL epitope. This product concept was extensively characterized and in preclinical studies was shown to elicit antibody and CD4 and CD8 T-cell responses 
. Although the product was shown to be immunogenic after two doses in this study, immunogenicity was not fully evaluated because the trial was stopped early due to unacceptable local reactogenicity. Therefore, this analysis is focused primarily on the safety endpoints of the trial.
The primary differences between the immunotherapy studies done with this product formulated in Montanide ISA-51 and the current studies are: 1) the immunological health of the subjects, 2) the 0.5 ml instead of 0.25 ml dose volume, and 3) the strength of the emulsion created with the EmulsiFlex 1000™ mechanical device. In preclinical toxicity studies, the clinical product had been tested in rabbits and rhesus monkeys. In one of two monkeys given a series of three immunizations with an 8 mg dose of peptide, granulomatous inflammation was noted at necropsy at the injection site. Otherwise, there was nothing notable about local toxicity.
The distinct differences in local reactogenicity between the current study and the studies evaluating therapeutic vaccines in HIV-infected subjects suggest that underlying host immune competence may be a primary determinant of the intensity of the local response to the Montanide ISA-51 adjuvant. This is supported by the prolonged local reactions in the Pfs25/Pvs25 malaria vaccine study in healthy subjects 
. Even though sterile abscesses requiring drainage were not reported, there was induration lasting up to 6 months. It is also supported by the studies using this adjuvant with therapeutic vaccines for a variety of neoplasms in which the vaccines were characterized as well-tolerated 
. Two of the subjects in the present study with the delayed subcutaneous reactions and prolonged induration or abscess formation had reported arm soreness from overuse at the time of the initial vaccination (one from floor scrubbing and the other from skiing). The role of significant arm exercise in the subsequent reaction is unknown.
The assumption is that sterile abscess formation was related to the inflammatory effect induced by the adjuvant. Licensed aluminum-based adjuvants rarely are associated with sterile abscess formation, and the cause is thought to be related to aluminum hypersensitivity reactions 
. The high frequency of local reactions in this study suggests adjuvant-induced inflammation rather than hypersensitivity was the cause of abscess formation. The adjuvant effect of IFA is in part related to the Toll-like receptor (TLR) stimulation by products contained in the oil, and in part by a depot effect. The removal of the vaccine dose in a stable emulsion by professional antigen presenting cells can take months. Therefore, using a higher dose volume would not only enlarge the central nidus of a sterile abscess, but would prolong the length of the inflammatory stimulus underlying the genesis of the abscess. The volume of the vaccine dose in the HIV therapeutic trial using this product was 0.25 ml. The dose volume in the Pfs25/Pvs25 malaria study was 0.5 ml. Based on the experience in this study with a 0.5 ml dose, we suggest that the lowest possible volume needed to deliver an adequate antigen dose should be used with potent adjuvants that have the potential for sterile abscess formation.
The method of producing the emulsion was the major difference between the current study and the trial done in healthy adults with malaria antigens in which there were some prolonged local reactions and two cases of erythema nodosum 
. The emulsion was intentionally produced to be stable in the current study to reduce the potential for escape of the oils to organs like the spleen and liver, where they could potentially serve as a stimulus for granulomatous inflammation. One could speculate that the erythema nodosum seen in the malaria trial was related to a slightly less stable emulsion, and the sterile abscess formation in this study was related to extremely high stability of the emulsion. It is also possible that antigenic differences could explain the differences in systemic reactogenicity between this study and the one evaluating malaria antigens with IFA. In this study the only severe systemic reactions occurred after the second dose in the 4 mg recipients. The rapid onset of headache, chills, nausea, and myalgia within hours after vaccination in this study suggests the vaccine formulation activated innate responses, but the higher severity of reactions after the second vaccination suggests that tissue-resident antigen-specific effectors from the first vaccination may have also contributed to the exaggerated response. The reason why subjects with the most severe systemic reactions did not develop severe local reactions is not known. It is possible this was a coincidence because of the small trial size or may reflect something about the nature of the underlying inflammation responsible for the local and systemic reactions.
Relative to the prior experience with peptide-based vaccines, this product had good immunogenicity. Antibody responses were detectable in 100% of subjects after two immunizations, which is notable for subunit vaccines in general. T cell responses were detected in the majority of subjects despite the early termination and incomplete vaccination schedule and the fact that this study was performed and analyzed during the historical period when both cryopreservation techniques and assay methodology were in transition. Therefore, the frequency and magnitude of the ELISpot and ICS responses in this report may be lower than they would have been if processed and tested with current methodologies. This study was specifically designed to detect HLA-B7-restricted CD8 T-cell responses to the V3 peptide epitope (RPNNNTRKSI). The absence of a 51
Cr CTL response to this epitope in 4/6 HLA-B7+ subjects tested may be due to the relatively low sensitivity of the assay. In contrast, 5/7 HLA-B7+ subjects had vaccine-induced CD8 T-cell responses by ICS. The two nonresponders may have not had sufficient vaccine-induced immune activation, the assays may have been insensitive, or there may have been other influences related to epitope liberation, processing, or degradation, competing responses to other epitopes, or subject-specific response hierarchies that explain the lack of detectable T cell responses. Three of the 9 HLA-B7-negative subjects also had detectable CD8 T-cell responses. Two of these subjects were HLA-A2+ and could have recognized a variant of the RGPGRAFVTI epitope described in V3 
, although the exact sequence is not represented in this vaccine. The other subject was HLA-A1, B8, and B57 and the epitope in the vaccine that is restricted by these alleles is not known. These data suggest that peptides can elicit CD8 CTL when formulated with a potent adjuvant. However, it should be kept in mind that control of HIV may require broad responses to multiple epitopes, and eliciting potent epitope-specific CD8 T-cell responses requires more than MHC compatibility. Therefore, restricting the antigenic content of vaccine antigens may limit the host options for selecting and presenting effective T cell targets and will limit the breadth of the response.
This study, AIDS Vaccine Evaluation Group (AVEG) Protocol 020, was designed to enroll 28 subjects to evaluate two doses of the peptide-based vaccine in IFA, but was terminated early because of sterile abscess formation in 4 of the first 24 enrollees. While the peptide-based vaccine was immunogenic, the local reactogenicity was unacceptable for continued development. We show that achieving an immunogenic peptide formulation is feasible in humans, but it may be associated with a significant reactogenicity cost, which in this study was unacceptable. Considering the outcome of this study and the malaria vaccine study, future studies using Montanide ISA-51 as an adjuvant in healthy subjects should be carefully considered from the standpoint of both local and systemic reactogenicity. Overly stable emulsions that may mitigate systemic reactogenicity have the potential to cause unacceptable local reactogenicity including chronic sterile abscesses.