The response of the PPTP cell lines to GSK923295A is reminiscent of the response that was previously observed for ispinesib [20
]. The PPTP previously demonstrated that the response of the PPTP cell lines for ispinesib and vincristine are highly related in terms of the minimum T/C% (Ymin
) values for each agent [30
]. To analyze GSK923295A and ispinesib response patterns in vitro, the Ymin
values for the two agents were compared. There was a highly significant correlation between the Ymin
values for GSK923295A and ispinesib (R2
=0.83). Although the contribution to this correlation by the cell line Rh18, which has Ymin
above 50% for both GSK923295A and ispinesib, is considerable, when Rh18 is omitted from the analysis there remains a significant correlation between the Ymin
values for the two agents (R2
The strong correlation between Ymin
responses to ispinesib and GSK923295A suggests that the PPTP cell lines have a shared pattern of response to these two agents in terms of whether the cell lines undergo a cytotoxic response versus a cytostatic/partial cytotoxic response. The pattern of response does not show complete histotype dependence, although the ALL cell lines have a high percentage of near complete cytotoxic responders. The similar activity patterns for GSK923295A, ispinesib and vincristine against the PPTP’s in vitro
panel is consistent with NCI-60 cell line panel testing results, for which similar activity profiles for antimitotic agents such as vincristine and the kinesin spindle protein inhibitor S-trityl-L-cysteine (STLC) were also observed [31
]. The two different patterns of response observed for the PPTP cell lines to these diverse anti-mitotic agents suggest that the cellular decision-making process for a cytotoxic response (e.g., apoptosis) versus a cytostatic/partial cytotoxic response lies downstream of the primary mitotic effect of the compounds. However, the biological basis for the similar responses of the PPTP cell lines to these anti-mitotic inhibitors is not clear.
Objective responses were noted in 13 of 35 solid tumor xenografts, including 9 with MCRs, 3 with CRs, and 1 with PR. Three of 5 Ewing sarcoma xenografts achieved MCR or CR, as did 2 of 3 rhabdoid tumor models, and 2 of 5 rhabdomyosarcoma models. For the neuroblastoma panel, the best response was progressive disease with growth delay (PD2), which was observed for 5 of 6 xenografts. The determinants of this differential sensitivity of diverse tumor histotypes to GSK923295A currently remain unresolved. However, similar to the cell line panels, the xenograft responses to GSK923295A were remarkably concordant with those to ispinesib. Out of 14 xenografts common to both studies, 9 achieved identical median group responses, while the median group responses of 3 out of the remaining 5 xenografts differed by only a single category. The median group responses of these 14 xenografts showed a significant correlation between the two drugs (r=0.78; p=0.001). Again, these data indicate a shared in vivo response to two drugs targeting different components of the mitotic machinery.
The ALL xenograft panel was inevaluable because of toxicity. There is no obvious reason that would explain why the NOD/SCID mice did not tolerate GSK923295A as well as the SCID mice used for most of the solid tumor testing. GSK923295A is relatively impotent towards murine CENP-E [18
], and so drug-related toxicity is not expected for mice at the dose selected for testing. Treatment for the ALL xenografts was administered only on Days 1–3 and Days 8–10, and most of the toxicity was observed at Day 21 or thereafter. If responses at Day 21 were evaluated, then 6 of 8 ALL xenografts had sufficient numbers of evaluable treated animals to assess response. Among these, 5 of 6 xenografts met criteria for CR at Day 21, whereas 1 model (ALL-4) met criteria for PR as best response. The two xenografts with excessive toxicity by Day 21 were ALL-3 and ALL-19, and a repeat experiment with ALL-3 at 50% dose achieved a PR. Thus, GSK923295A, like other anti-mitotic agents such as vincristine and ispinesib [19
], demonstrated substantial remission-inducing anti-leukemia activity against the PPTP ALL xenografts, although for unexplained reasons the dose used for testing exceeded that which can be tolerated by leukemia-bearing NOD/SCID mice.
Our studies highlight similarities between GSK923295A and ispinesib. The dose limiting toxicities for ispinesib have been largely hematologic but this agent has not demonstrated clinical efficacy in phase 2 trials [33
]. As these two drugs demonstrated similar activity in the PPTP preclinical models it may suggest that GSK923295A may also have little human efficacy. However, GSK923295A demonstrates a significantly different pattern of tumor responses compared to vincristine, previously evaluated against the same in vivo
panel of solid tumors [19
]. Comparison of these data sets shows that the response to GSK923295A and vincristine differed by at least two response categories (e.g. PD2 vs
PR) in 16 of 30 solid tumor models. In 10 models GSK923295A was superior to vincristine, whereas it was inferior in 6 models. Superiority was particularly noticeable in 3 of 5 Ewing sarcoma models.
In conclusion, GSK923295A exhibited substantial in vitro and in vivo activity against the PPTP’s preclinical models. These data further indicate the relevance of targeting the process of mitosis for treatment of childhood malignancies. Although the similarity between ispinesib and GSK293295A is of concern, as the former drug has not demonstrated efficacy against adult malignancies, the novel spectrum of activity compared to vincristine in these solid tumor models suggests that GSK293295A may have a role in treatment of childhood cancers. A key issue in relating the observed high level of preclinical activity to the clinical setting will be the relationship between the drug exposures achieved in the SCID mice used for solid tumor testing compared to those achieved in humans at tolerable doses.