It is well known that lung cancer patients have abnormalities in their immune system and that the extent of these abnormalities may have prognostic relevance (1, 2). The promise of immune therapy for lung cancer has been suggested for many years, but real improvements in survival have remained elusive. An early report suggested that resected lung cancer patients who developed empyema had a superior outcome (3), perhaps on the basis of immune stimulation. This led to a number of trials of nonspecific immune stimulants, such as Calmette-Guerin bacillus (BCG) and nocardia rubra cell wall skeleton (4, 5). Although several small, uncontrolled trials showed promise to these approaches, larger randomized trials failed to show any survival benefit (6–9). Prior studies also evaluated the use of thymosin factor V, interferon, and other nonspecific immune stimulants. Once again, promising early study results could not be confirmed for thymosin factor V (10, 11), interferon (12–15), IL2 with or without adopted cells (16), or other means. However, immune approaches remain appealing because initial therapy with surgery, chemotherapy, and radiotherapy alone and in combination produce high response rates in all histologic types of lung cancer, but relapse is nearly inevitable. Over the past decade, a number of tumor antigens, which are expressed on the majority of lung cancer cells, were described. Some of these are expressed in nearly all small-cell lung cancers (SCLC) and are overexpressed on these tumor cells in comparison with normal lung tissues. These tumor antigens provide a new specific means for development of successful vaccine approaches. Tumor antigens that are being explored for potential vaccines include mutant oncogene proteins such as p53, ras, or erb B2, and also cell surface proteins such as CEA, MUC1, GD2, or HuD. At this point, the ideal antigen for immune therapy is unknown. There are also many ways in which these antigens may be developed as vaccines. One approach is to isolate the tumor protein, such as mutant p53, and administer the protein alone or with nonspecific immune stimulants. A group from the National Cancer Institute (NCI) evaluated the clinical relevance of human antitumor immune responses to tumor antigens, including p53 and HuD (17). They found serum antibodies against autologous tumor proteins in 50% of SCLC and 75% of non–small-cell lung cancer (NSCLC) patients, including two patients who had anti-p53 antibodies and two patients who had anti-HuD antibodies. The presence of antibodies was found to correlate with improved survival and limited stage. In subsequent studies, the NCI showed that missense p53 mutations give rise to new tumor-specific peptide sequences, which are efficiently processed and presented by human lung cancer cells. They then showed that these mutant p53 peptides were effectively targeted by cytotoxic T cells specific for the endogenous mutant epitopes (18). They are currently conducting a clinical trial using the patient-specific p53 peptides as the immunogen. However, this approach is very labor intensive because the mutant p53 protein must be identified and isolated from each patient and then mass produced. Another approach to using oncogene mutations evaluated activating mutations of the p21 ras proto-oncogene because these activating mutations are limited in number and occur in many common malignancies and in otherwise completely conserved regions. Triozzi and colleagues (19) engineered a chimeric ras immunogen incorporating a promiscuous T-cell epitope to enhance the immunogenicity of an oligopeptide corresponding to a weakly immunogenic substitution, and they are evaluating this immunogene as a vaccine. Many lung cancers overexpress erb B1 and its protein, epidermal growth factor receptor (EGFR), and/or erb B2 and its protein, Her2/neu. These oncoproteins have also been used in vaccine development. Gonzalez and associates (20) linked the human EGF protein to either tetanus toxoid or Neisseria meningitidis recombinant protein and immunized patients with intradermal injection using aluminum hydroxyde as an adjuvant. Anti-EGF antibodies developed in 60% of the immunized patients without major toxicities. Finally, Amici and coworkers developed a plasmid DNA encoding the rat neu NT oncogene (21). In a transgenic mouse model, intramuscular injections of neu NT plasmids drastically reduced the outgrowth of new and mammory neoplasms. Common tumor antigens such as CEA, MUC1, GD2, and HuD have been used as the basis for vaccine strategies. Human MUC1 is overexpressed in many human car( Received in original form May 6, 1999 )