Lymphocytes extracted from freshly resected melanomas can be expanded in vitro and can often mediate specific lysis of autologous tumor cells but not allogeneic tumor or autologous normal cells. We treated 20 patients with metastatic melanoma by means of adoptive transfer of these tumor-infiltrating lymphocytes and interleukin-2, after the patients had received a single intravenous dose of cyclophosphamide. Objective regression of the cancer was observed in 9 of 15 patients (60 percent) who had not previously been treated with interleukin-2 and in 2 of 5 patients (40 percent) in whom previous therapy with interleukin-2 had failed. Regression of cancer occurred in the lungs, liver, bone, skin, and subcutaneous sites and lasted from 2 to more than 13 months. Toxic effects of interleukin-2 occurred, although the treatment course was short (five days); these side effects were reversible. It appears that in patients with metastatic melanoma, this experimental treatment regimen can produce higher response rates than those achieved with interleukin-2 administered alone or with lymphokine-activated killer cells. It is too early to determine whether this new form of immunotherapy can improve survival, but further trials seem warranted.

These studies demonstrate the feasibility and safety of using retroviral gene transduction for human gene therapy and have implications for the design of TIL with improved antitumor potency, as well as for the possible use of lymphocytes for the gene therapy of other diseases.

We have administered 1039 courses of high-dose interleukin-2 (IL-2) to 652 cancer patients. Five hundred ninety-six patients had metastatic cancer that either had failed standard effective therapies or had disease for which no standard effective therapy existed, and 56 patients were treated in the absence of evaluable disease in the adjuvant setting. IL-2 was administered either alone (155 patients) or in conjunction with activated immune cells such as lymphokine activated killer (LAK) cells (214 patients) or tumor infiltrating lymphocytes (TIL) (66 patients), with other cytokines such as alpha interferon (a-IFN)(128 patients) or tumor necrosis factor (TNF)(38 patients), with monoclonal antibodies (32 patients), or with the chemotherapeutic agent cyclophosphamide (19 patients). Initial results with the treatment of high-dose IL-2 alone or in conjunction with LAK cells have indicated that objective regressions of cancer can be achieved in 20% to 35% of patients with selected advanced metastatic cancers. Although most responses have been seen in patients with metastatic renal cell cancer, melanoma, colorectal cancer, and non-Hodgkin's lymphoma, many histologic types of cancer have not been treated in significant numbers. These regressions can be durable; of 18 patients achieving a complete response, ten have not experienced recurrence at intervals from 18 to 52 months. Although combinations of IL-2 with TNF do not appear to result in increased responses, there is a suggestion in our initial phase I studies that the combination of a-IFN and IL-2 is more effective than the administration of cytokine alone and this combination deserves further study. Similarly the adoptive transfer of TIL in conjunction with IL-2 also appears to be more effective than the use of IL-2 alone. The toxic side effects in patients treated with high-dose IL-2 are presented and include malaise, nausea and vomiting, hypotension, fluid retention, and organ dysfunction. Treatment-related deaths were seen in 1% of all treatment courses and in 1.5% of patients. These studies demonstrate that a purely immunologic manipulation can mediate the regression of advanced cancers in selected patients and may provide a base for the development of practical, effective biologic treatments for some cancer patients.

As these studies continue, efforts are underway to develop improved immunotherapies using tumor-infiltrating lymphocytes (TIL) and gene-modified TIL.

Patients with metastatic melanoma undergoing therapy with cyclophosphamide (CPM), tumor-infiltrating lymphocytes (TIL), and interleukin-2 (IL-2) were studied for the ability of their 111In-labeled TIL or peripheral blood lymphocytes (PBL) to localize in sites of tumor using gamma camera imaging and biopsies. Nineteen infusions of radiolabeled TIL were given to 18 patients, while five patients received radiolabeled autologous PBL during TIL therapy. Clear tumor localization was seen on 13 of 18 nuclear scan series performed on 111In-TIL recipients, while tumor was imaged in only one of four scan sequences on patients given 111In-PBL. Nineteen paired biopsies of tumor and normal skin were completed on 10 patients receiving 111In-TIL, while eight biopsies were done on three PBL patients receiving 111In-PBL. The mean percentage of total injectate activity localizing per gram of tumor tissue was 0.0049% in the TIL group and 0.0010% in the PBL group (P2 = .0004). The mean of the tumor to normal skin ratios of the 111In-TIL group was three times that for 111In-PBL (P2 = .0072). One patient was studied by nuclear scanning on three consecutive treatment courses of CPM, TIL, and IL-2. He initially demonstrated clear tumor localization by 111In-TIL at several sites, then faint localization with 111In-PBL at a single site, and subsequently positive tumor imaging on repeat 111In-TIL infusion at multiple sites. These results confirm and expand our initial data demonstrating that human TIL transferred with CPM pretreatment and followed by IL-2 preferentially localize to tumor sites and indicate that this localization is greater for TIL than PBL.

Tumor-infiltrating lymphocytes (TILs) can be grown in vitro in medium containing interleukin-2 (IL-2). In clinical trials at the Surgery Branch of the National Cancer Institute, patients with metastatic malignant melanomas were treated with IL-2 plus the adoptive transfer of autologous TILs. At the time of treatment, TILs were assayed for in vitro lysis of fresh autologous and allogeneic melanoma cells and Daudi cells. Patients were evaluated for clinical response 4-8 weeks later. Lysis of autologous tumor cells by TILs was significantly higher for responding than for nonresponding patients. Tumor cells from responding and nonresponding patients were equally sensitive to lysis by allogeneic lymphokine-activated killer (LAK) cells. There was no difference between TILs from responding and nonresponding patients for lysis of LAK-sensitive Daudi cells, which was low in most cases and demonstrated that TIL lysis of autologous tumor cells was not due to LAK cells. The observed association of autologous tumor cell lysis by TILs with clinical response suggests that the development of culture methods to optimize lysis of autologous tumors may lead to increased response rates using this TIL treatment regimen.

The application of bone marrow gene therapy has been stalled by the inability to achieve stable high-level gene transfer and expression in the totipotent stem cells. We show that retroviral vectors can stably introduce genes into antigenspecific murine and human T lymphocytes in culture. Murine helper T cells were transduced with the retroviral vector SAX to express both neomycin-resistance and human adenosine deaminase genes. These cells were expanded in culture and selected for expression of neomycin resistance with G418.

Tumor-infiltrating lymphocytes (TILs) are cells generated from tumor suspensions cultured in interleukin 2 that can mediate cancer regression when adoptively transferred into mice or humans. Since TILs proliferate rapidly in vitro, recirculate, and preferentially localize at the tumor site in vivo, they provide an attractive model for delivery of exogenous genetic material into man. To determine whether efficient gene transfer into TILs is feasible, we transduced human TILs with the bacterial gene for neomycin-resistance (NeoR) using the retroviral vector N2. The transduced TIL populations were stable and polyclonal with respect to the intact NeoR gene integration and expressed high levels of neomycin phosphotransferase activity. The NeoR gene insertion did not alter the in vitro growth pattern and interleukin 2 dependence of the transduced TILs. Analyses of T-cell receptor gene rearrangement for 13-and y-chain genes revealed the oligoclonal nature of the TIL populations with no major change in the DNA rearrangement patterns or the levels of mRNA expression of the (3 and y chains following transduction and selection of TILs in the neomycin analog G418. Human TILs expressed mRNA for tumor necrosis factors (a and (3) and interleukin 2 receptor P55 but not for interleukin 1/3, granulocyte/macrophage colony-stimulating factor, interleukin 6, and interferon 'y when grown with high-dose interleukin 2 without subsequent activation with mitogen or specific antigen. This pattern of cytokine-mRNA expression was not significantly altered following the transduction of TILs. The NeoR gene-transduced TILs could thus be used to follow the trafficking and survival of TILs in vivo, and clinical protocols using these transduced TILs in cancer patients have begun. The studies demonstrate the feasibility of TILs as suitable cellular vehicles for the introduction of therapeutic genes into patients receiving autologous TILs.Tumor-infiltrating lymphocytes (TILs) are lymphoid populations found associated with solid tumors and can be selectively grown in the presence of interleukin 2 (IL-2) (1, 2). Recent clinical trials have shown that the adoptive transfer of TILs in conjunction with IL-2 can result in cancer regression in some patients with advanced, otherwise refractory, cancers (1). Studies of the administration of human TILs labeled with "'In have shown that TILs recirculate and preferentially localize at tumor sites by several days after infusion (3). There has been no method available, however, to study long-term distribution and survival of autologous human TILs in vivo. We have proposed to use an inserted neomycinresistance (NeoR) gene (4, 5) into human TILs as a cell label to answer these questions. It has recently been shown that the NeoR gene can be inserted into murine lymphocytes and human TILs (K. Culver, K. Cornetta, S.M., P.A., S.F., A.K., R. Morgan, M.T.L., S.A.R., W.F.A., and R.M.B., unpublished results). This report represents results of studies to transduce and characterize the NeoR-transduced TIL populatio...