Ultrafractionation in human malignant glioma xenografts

Beauchesne et al.1 have published very interesting data on low-dose hyper-radiosensitivity (HRS) of different human malignant glioma cell lines in vitro and on the effect of repeated low-dose irradiation (ultrafractionation) in G152 human malignant glioma xenografts in nude mice. In vitro G152 cells showed HRS, i.e., a greater effect on clonogenic cell survival per unit of dose at low compared to higher doses, which was most marked at 0.8 Gy. Ultrafractionation in G152 tumors was performed with 3 fractions of 0.8 Gy per day over 2 weeks and compared to one daily fraction of 2.0 or 2.4 Gy applied within the same overall treatment time. Average tumor volume 12 weeks after irradiation was significantly smaller after ultrafractionation than after daily 2.0 Gy fractions. No statistical comparison was provided between ultrafractionation and daily 2.4 Gy fractions. From their data, the authors concluded that repeated low-dose irradiation could greatly improve the effectiveness of radiotherapy in human malignant gliomas. Using the same overall dose as conventional radiotherapy, this regime was considered to be twice as effective at killing cells in glioma xenografts and to allow safe treatment of patients with cumulative doses >60 Gy. We agree with the authors that the HRS effect is a biologically very interesting phenomenon and that both the mechanisms underlying this response and the possibility to explore HRS for radiotherapy should be carefully explored.2 However, the conclusion of greatly improved effectiveness and a 2-fold higher kill of glioma cells and xenografts is not justified from the data presented. It is well recognized that tumor regression and tumor volume are not valid end points to measure the survival of clonogenic tumor cells, i.e., cells than can cause recurrences.3, 4 Besides the number of surviving clonogenic cells, these end points are significantly influenced by many other parameters, including the number of nonclonogenic doomed cells, resorption of dead cells, cell edema, cell-cycle effects, etc. For these reasons, there is longstanding consensus among radiobiologists and experimental radiotherapists that the most reliable end point for preclinical studies is tumor control dose 50% (TCD50), which is influenced only by clonogenic cells.4 As an alternative, evaluation of tumor growth delay at different dose levels with correction for differences in growth rate in the different experimental arms may be as reliable as a tumor control assay.4 Such studies were not performed by Beauchesne et al.1 We have performed ultrafractionation experiments using another human glioma model, A7. In this cell line, a significant HRS effect was shown in vitro, with most marked expression at 0.4 Gy.5 Despite this HRS in vitro, the efficacy of ultrafractionation (126 fractions of 0.4 Gy over 6 weeks) in A7 tumor xenografts in nude mice was significantly less than the effect of conventional treatment (30 fractions of 1.68 Gy over 6 weeks) using both tumor growth delay and top-up TCD50 as experimental end points.6 It is our view that the data of Beauchesne et al.1 do not demonstrate an improvement in the effectiveness of radiotherapy in human malignant gliomas by ultrafractionation and do not provide a preclinical rationale for or against clinical trials. In contrast, the data on A7 glioma clearly show a need for further preclinical investigations with appropriate end points to explore whether clinical trials on ultrafractionation are indicated. Yours sincerely, Mechthild Krause, Michael Joiner, Michael Baumann