transferred NK/LAK cells were thought to play a major role in mediating the 
antitumor effects of IL-2, CD56+ cells (the NK/LAK phenotype) were not found 
in biopsies of responding lesions from patients with metastatic melanoma (8). 
Selective lymphocyte depletion experiments in a variety of murine tumor models 
demonstrated that T lymphocytes (particularly CD8 cytotoxic T lymphocytes 
(CTL) ) , and not NK/LAK cells, were responsible for the tumor regression 
observed when IL-2 was administered to the tumor-bearing animals. Rosenberg 
et al subsequently demonstrated that tumor antigen-specific CTL could be 
derived from ex vivo cultures of tumor biopsies supplemented with IL-2 (also 
called tumor-infiltrating lymphocytes or TIL) (9). TIL had more potent 
antitumor activity than LAK in the animal models by a factor of approximately 
100 . 
On the basis of these data, and evidence that autologous tumor-specific TIL 
could be expanded from tumor biopsies of patients with metastatic melanoma, 
clinical trials of IL-2 in combination with TIL were initiated. To date, IL- 
2/TIL has been reported to produce response rates of 35-40% (3,10), although 
the contribution of TIL to the activity of IL-2 has not been demonstrated in 
randomized studies. One interesting observation has been that some patients 
failing IL-2 alone or with LAK will respond to IL-2/TIL. The importance of 
these antigen-specific CTL in producing tumor regression is further supported 
by evidence that response to IL-2/TIL is correlated to both specific 
cytotoxicity of the TIL cells for autologous tumor in vitro, and to the number 
of TIL cells administered to the patient (10,11). 
The clinical and preclinical studies of TIL indicate that induction or 
adoptive transfer of antigen-specific T cells can lead to tumor regression in 
vivo, and that antitumor effects are cell dose-related. In the clinical 
studies of TIL conducted to date, the ability to expand TIL in vitro has been 
limited. Furthermore, even when large cell doses are administered, many of 
the cells may not be specific for the relevant tumor antigen. It may be 
possible to more effectively induce or expand the antigen-specific response 
to melanoma, which may result in higher response rates or improved quality of 
tumor responses in patients. To accomplish this goal, we believe it may be 
necessary to first induce a strong CTL response in vivo, followed by isolation 
and expansion of the melanoma-specific CTL in vitro, followed by repeated 
adoptive transfer of large numbers of the CTL to patients. 
The goal of the first generation of our clinical studies is to induce a strong 
melanoma-specific CTL response in vivo, by immunizing with genetically 
modified lethally irradiated melanoma cell lines containing shared melanoma- 
associated antigens. The modifications are intended to increase the 
[722] 
Recombinant DNA Research, Volume 18 
