Using a novel loading technique, IL-12 is reported here to be efficiently encapsulated within large multilamellar liposomes. The preclinical efficacy of the cytokine loaded liposomes to deliver IL-12 into human tumors and to reactive tumor-associated T cells in situ is tested using a human tumor xenograft model. IL-12 is released in vivo from these liposomes in a biologically active form when injected into tumor xenografts that are established by the subcutaneous implantation of non-disrupted pieces of human lung, breast or ovarian tumors into immunodeficient mice. The histological architecture of the original tumor tissue, including tumor-associated leukocytes, tumor cells and stromal cells is preserved anatomically and the cells remain functionally responsive to cytokines in these xenografts. The local and sustained release of IL-12 into the tumor microenvironment reactivates tumor-associated quiescent effector memory T cells to proliferate, produce and release IFN-γ resulting in the killing of tumor cells in situ. Very little IL-12 is detected in the serum of mice for up to 5 days after an intratumoral injection of the IL-12 liposomes. We conclude that IL-12 loaded large multilamellar liposomes provide a safe method for the local and sustained delivery of IL-12 to tumors and a therapeutically effective way of reactivating existing tumor-associated T cells in human solid tumor microenvironments. The potential of this local in situ T cell re-stimulation to induce a systemic anti-tumor immunity is discussed.

Abstract.A very simple speech coder is investigated based on modified discrete-cosine transform (MDCT), run-length coding, and entropy code. The speech signal is first passed through a MDCT, quantized by a uniform quantizer, and run-length coded to achieve good compression ratio. Not suitable for wireless communication systems, the speech coder is perfect for Internet and speech storage applications. With a performance far better than most speech coders for wireless applications, low data-rate speech of 3 kbit/s can be achieved with a signal-to-noise ratio (S/N) of about 12 dB.

A bstructA very simple speech lcoder is investigated based on discrete-cosine transform (DCT) and run-length coding. The speech signal is first passed through a DCT, quantized by a uniform quantizer, and runlength coded to achieve good compression ratio. Not suitable for wireless communication systems, the speech coder is perfect for Internet and speech storage applications. With a performance far better than most speech coding for wireless applications, low data-rate speech of 4 kbit/s can be achieved with a signal-tonoise ratio (S/N) of about 11 dB, comparable to 8 kbit/s VSELP and CS-KELP speech coder. 0-7803-5284-X/99/$10.00 0 1999 IEEE.