rejection in the HS-tk system in mice is the observation that not 
all the tumor's cells must contain the inserted gene in order to 
be killed by GCV. In mice given a subcutaneous tumor in which 100% 
of the cells carry the HS-tk gene, complete tumor regressions were 
seen in 13 of 15 animals following GCV treatment. Interestingly, 
when tumors established from cell mixtures containing 50% HS-tk- 
gene-modif ied cells mixed with 50% wild-type unmodified tumor cells 
were treated with GCV, almost all tumors regressed. Even in 
situations where the mixed tumor contained 90% unmodified, wild- 
type tumor cells mixed with only 10% HS-tk modified tumor cells, 
complete regression of the cancer was observed with GCV treatment 
in 9 of 15 animals (Table 2, Appendix A: Reprint 1). 
The mechanism of this "bystander tumor kill" is not yet 
completely understood. This may involve the production of toxic 
triphosphates produced by the interaction of thymidine kinase and 
GCV within the tumor leading to inhibition of DNA synthesis and 
death of replicating cells. This does not seem to involve 
generalized non-specific cellular toxicity since the overlying skin 
and other tissues surrounding these HS-tk treated tumors are not 
injured while the tumors expressing the genes and the admixed wild- 
type tumor cells are completely destroyed. 
The 9L rat gliosarcoma and the human glioblastoma cell lines 
U251 and A172 were also sensitive to this "bystander" effect in an 
in vitro mixing experiment. GlTkSvNa.29 transduced and non- 
transduced tumor cells were mixed at different ratios in 96 well 
microtiter dishes. GCV was added to the wells and 24 or 48 hours 
later, the cultures were pulsed with tritiated thymidine. These 
bar graphs depict a greater decrease in proliferation than would 
be expected at a GCV level of 5.1 ug/ml in the medium (the numbers 
over the bars represent the percent of transduced tumor in the 
culture) (Appendix B: Figure 1). The 5.1 ug/ml concentration is 
easily within the therapeutic range established in humans with a 
dose of 5 mg/kg/dose. Since it is unlikely that 100% of the tumor 
cells in the brains of our patients will be successfully gene- 
modified, this "bystander" effect is very important for the 
successful elimination of the tumor using this approach (See 
Appendix B: Figure 1) . 
F. Effects of Ommaya Reservoir on VPC. 
In Vitro : In order to determine if the Ommaya reservoir would bind 
significant numbers of VPC or induce cell clumping, 5 mis of 
PA317/GlTkSvNa VPC's were passed through an Ommaya reservoir with 
a 5% loss of cells without visual evidence of cell clumping. 
In Vivo: In order to determine if the Ommaya reservoir would 
inhibit in vivo dene transfer, GlTkSvNa.29 VPC's which were passed 
through an Ommaya, mixed (3:1) with the MCA 205 fibrosarcoma and 
injected SQ into C57BL/6 syngeneic mice. Control animals were 
injected with mixtures of GlTkSvNa.29 VPC's and MCA 205 that were 
taken directly from culture or injected with mixtures of 
GlBgSvNa.29 VPC's and MCA 205 tumor cells. GCV treatment was 
initiated 9 days after injection of the cells and continued twice 
daily for 14 doses of 150mg/kg/dose . 
All tumor /VPC mixtures grew well without GCV treatment. The 
growth of those tumors injected with mixtures of GlBgSvNa.29 VPC 
and MCA 205 tumors were not affected by GCV treatment. All five 
mice in each GlTkSvNa.29 VPC/ 2 05 tumor mixture group whether passed 
[62] 
Recombinant DNA Research, Volume 18 
