Peroxisome proliferator activated receptor 3' (PPAR3') agonists for treatment of neoplastic diseases in vitro data and early clinical observations in chronic myelomnocyfie leukemia (CMML) and in fiposareoma. C. Den71inser, A. Mdhle, F. Machicao, R. M61de, P. Brossart, H.-U. Hiring, L. Kanz, K. Dittmann. University Clinic Tfibingnn, D-72076 Tfibingen, Germany. PPAR 7 is a nuclear hormone receptor which modulates gene expression upon agonist binding and heterodimelzation with activated retinoid X receptor. PPAR 7 agonists comprise both natural (cyclopentenone prostanoids) and pharmaceutical compounds (e.g. thiamlidinediones) and have been shown to exert antiproliferative effects in several cell lines in vitro. We investigated whether the thiazolidinediones troglitazone (TgO) or rosiglitazone (ROS) have an effect on proliferation ofmononuclenr ceils from healthy individuals and form patients with CMML in vitro. At a concemration of 50pM, TRO reduced DH]thymidine incorporation in mononudear celh from healthy individuals (n=5) or from CMML patients (n=10) during 24 h of incubation to 57_+18% (mean_+gD) or to 9-+7% of control, g o s had similar but less potent antiproliferative effects in CMML ceils in vitro whereas ceils from healthy individuals were not significantly affected. Our in vitro data prompted us to offer TRO to 3 selected CMML patients. Another 4 CMML pationts were treated with ROS. Furthermore, 3 patients with metastasized liposarcoma were treated with TRO or ROS. Treatment with either drug was very well tolerated. Control of monocyte counts was transiently improved in 5 of the 7 CMML patients. In 2 of the 3 liposarcoma patients minor responses and tran~emt disease stabilizations were noted. Our results suggest that PPAR 7 agonists may have therapeutic potentiel in CMML and in liposarcoma justifying further study of this non-genotoxic approach. Multicellular gasta'ic eaneer spheroids represent a valid model for pre-dinical testing of new therapeutic strategies B, Mayer, G. Klement, M. Kuneko, S. Jothy and R.S. Kerbel Gastric carcinoma is largely unresponsive to current chemotherapy and better pro~linical models to study therapeutic targets are urgently needed. An in vitro multicellular gastric cancer spheroid model was established, and compared to the subcutaneous xenogra~s in nu/nu mice as to: i) morphological and functional differentiation and ii) expres~on of cell adhesion molecules. Twelve out of 17 (71%) gastric cancer cell lines grown in 3D reflected growth characteristics of then parental gastric carcinomas. Cell lines derived from peritoneal and pleural carcinomatous grew as single cells (HSC-39, KATO-II, KATO-III) or cell aggregates (SNU-5, SNU-16). Adenosquamous (MKN-1) and tubularly differentiated (MKN-28, MKN-74, N87) cell lines formed partly compact multicellular spheroids and also reflected their parental architecture. Most importantly, difl'ereatiation was lost afier s.c. implantation of these preformed spheroids. Levels ofnmcin and constitutive E-cadherin expression reflected the degree of morphological differentiation, but the changes ot other adhesion molecules (EpCAM, c~2131, CD44s, Le', sLe') were heterogeneous. In contrast, cell lines derived from poorly differanfiated gastric carcinomas (Hs-746T, RF-1, RF-48) formed fully compact spheroids mimicking the poorly differentiated phenotype, were E-cadherin negative, and showed only CD44s upregulation. In summary, rm~Mcellular gastric cancer spheroid recapitulate the complexity of their in wvo counterparts, and represent a superior model for studying the biology of this cancer than s.c. xenogrefls. They embody a physiologically relevant model to test new therapeutic strategies such as differentiating and antiadhesion agents.