Enblad P, Hesselager G, Bongcam-Rudloff E, Hallin I, Westermark B, Nistér M
Department of Neuroscience, Uppsala University, Uppsala, Sweden.
Int J Cancer. 2000 Mar 15;85(6):819-28. doi: 10.1002/(sici)1097-0215(20000315)85:6<819::aid-ijc15>3.0.co;2-q.
Three phenotypically different clonal human glioma cell lines were injected stereotactically into nude-rat brains, to determine their individual growth potential and to establish an in vivo system in which different therapeutic modalities could be tested. As assessed by serial sectioning, microscopic evaluation, and computer analysis, the mean approximate tumour volume after 3-7 weeks in vivo was 0.42 mm(3) for U-343 MG, 2.6 mm(3) for U-343 MGa Cl2:6, and 50.3 mm(3) for U-343 MGa 31L. When compared with the initial injected cell volume, only U-343 MGa 31L had increased in size, U-343 MGa Cl2:6 remained approximately the same but showed a certain proliferative potential, and U-343 MG regressed. Thus, only U-343 MGa 31L cells had high tumorigenic potential, invaded and replaced brain tissue in every direction, while U-343 MGa Cl2:6 cells grew in sheet-like tumour extensions along white-matter nerve-fibre tracts, in this respect mimicking foetal astrocytes. The tumorigenic potential of the U-343 MGa 31L cell clone was associated with a variable phenotype, as observed when the in vivo and in vitro characteristics were compared. The in vivo phenotype was characterized by the loss of GFAP immunoreactivity, the gain of heterogeneously distributed cellular tenascin, fibronectin, and laminin, but absence of extracellularly deposited material, and by the formation of irregular vessels. It appears that the intrinsic capacity of glioma cells to adapt to in vivo conditions is decisive for their tumorigenicity in the brain, rather than any single phenotypic property in itself. Moreover, the 2 glioma cell clones best suited for in vitro growth were no longer tumorigenic.