1. Sensitivity of human osteosarcomas to chemotherapy Skjalg Bruheim, Knut Breistøl, Birgitte Smith-Sørensen, Øyvind S Bruland 2, Gunhild M Maelandsmo, Øystein Fodstad 1 Department of Tumor Biology, Institute for Cancer Research and Department of Radiation Oncology 2, The Norwegian Radium Hospital, Oslo, Norway. 1 Present address: MSB 2332, Cancer Research Institute, University of South Florida, Mobile, AL 36688-0002 Introduction Osteosarcoma is a relatively rare tumor, comprising around 0.2 % of all malignancies and most commonly affecting children and young adults. In contrast to the situation for most solid tumors, in osteosarcoma patients chemotherapy has a curative potential. In the absence of reliable markers that could predict the response of the individual drugs used, patients are treated with multimodal regimens essentially including high-dose methotrexate, doxorubicin, cisplatin and ifosfamide. Due to the aggressive treatment of osteosarcoma patients, long-term toxicity and morbidity is not uncommon. It is likely that some patients could be cured with a simpler and less toxic treatment regimen, whereas one third of the patients succumb to their disease without beneficial effect from current treatment options. Hence, for major improvements in the management of osteosarcoma, a more individualised treatment as well as novel treatment alternatives are warranted. Immunodeficient mice and rats have contributed substantially in preclinical evaluation of anti-cancer drugs. The xenograft models can also be in used in the search for molecular markers of drug sensitivity/resistance, and drugs of interest can be studied individually. We have established a panel of eleven human osteosarcoma xenograft lines, grown sc in nude mice. The anti-tumor activities of doxorubicin, cisplatin, ifosfamide methotrexate, and lomustine, as well as the expression of some genes known to encode drug resistance, have been assessed. Microarray analysis on human osteosarcoma xenograft tissue before and after therapy is ongoing. Materials and methods Establishment of tumour xenografts and assessment of antitumor activity Xenografts were established by sc transplantation of fragments of tissue from biopsies or surgically removed tumours into the flanks of nude mice. For therapy experiments, tumour fragments of about 2x2x2 mm were implanted sc, and the animals were randomised for treatment according to tumor size when the average tumor diameter was about 6 mm. Maximal tolerable doses (MTD), allowing a median bodyweight loss up to 15 %, were found to be 8 mg/kg of doxorubicin, 5 mg/kg of cisplatin, 150 mg/kg of methotrexate, 240 mg/kg of ifosfamide and 20 mg/kg of lomustine All drugs were administered weekly x 2, lomustine ip, whereas the other drugs were given iv. Tumor volume was calculated according to the formula 0.5 x length x width 2. Antitumor effects were assessed from calculations of specific growth delay (SGD) and maximal growth inhibition (T/C%). Based on median relative tumor volumes (RTV), SGD and T/C% was calculated according to the formulas: SGD = (TD treated – TD control ) / TD control (the time for one (TD 200 ) or two (TD 400 ) median RTV doubling times was used to calculate values for SGD 200 or SGD 400 ) and T/C% = (RTV treated / RTV control ) * 100% The antitumor activity was defined as: (+) SGD >1.0 or T/C% 1.0 and T/C% 1.5 and T/C% 2.0 and T/C% 3.0 and T/C% < 10 %. Assessment of mRNA expression The mRNA expression of the drug resistance-associated genes O 6 -Methylguanine DNA methyltransferase (MGMT), DNA topoisomerase II (Topo II), Glutathione-S-transferase (GST)-, Multidrug-resistance related protein (MRP-1) and p-Glycoprotein (MDR1) were determined by Northern blot analysis. The levels of specific RNA were calculated relative to the amount of 18s rRNA and subsequently classified as follows: -/+, undetectable/low expression, ++ and +++, high or very high expression. Statistical analysis To estimate correlation between treatment efficacy and the mRNA expression of resistance associated genes, sample correlation coefficients were calculated according to the formula: r = (X i - X mean )(Y i - Y mean ) / [ (X i - X mean ) 2 ] [ (Y i - Y mean ) 2 ], where X i represented the sensitivity (1-T/C) of xenograft i to drug X and Y i represented the expression of gene Y in xenograft i. X mean represented the mean sensitivity to the drug X and Y mean represented the mean expression of gene Y. The test statistic t = ( n-2) r / ( 1-r 2 ), where d.f. = n-2 for t, was used to test for the level of significance at which the null hypothesis H 0 : r = 0 could be rejected in favour of H 1 : r 0. Results Around 20% of xenotransplanted osteosarcoma specimens showed take in nude mice, and gave rise to 11 established xenografts, from which parental tumor tissue was received from ten different patients. Histology and origin of the xenografts along with the main clinical characteristics of the patients are displayed in Table 1. Seven xenografts originate from primary tumors, whereas four were established from lung metastases. The median tumor volume doubling time varied between 3 and 20 days. Growth curves for the individual xenografts are shown in Figure 1. Table 2 summarises the antitumor activity of the various drugs and the mRNA expression of MGMT, Topo II, GST-, MRP-1 and MDR1. Statistical analysis revealed relatively low, but significant negative correlations between antitumor activity of lomustine and MGMT expression, and for doxorubicin and GST- whereas the efficacy of cisplatin somehow surprisingly correlated positively with MRP-1 expression (Table 3). The patients, from whom the xenografted tissue originated, responded poorly or moderately to chemotherapy and all ten succumbed to their disease. This indicates that growth in nude mice is associated with poor clinical outcome, as others previously have shown for soft tissue sarcoma. When defining drug efficacy as SGD>1 and T/C%<50, ifosfamide, lomustine, doxorubicine, cisplatin and methotrexate were effective in 5/11, 4/11, 3/11, 3/11 and 1/10 of the xenografts, respectively. Five of the 11 tumours were resistant to all compounds tested. For ifosfamide, cisplatin and doxorubicin, these results corroborates with response rates seen with these drugs in the clinic. Because of the low dose-intesity achieved without lecovourin rescue it is not supprising that methotrexate only was weakly effective in just one of the xenografts. Because osteosarcoma patients are treated with multimodal chemotherapy, it was not possible to compare directly responses in the individual patients with the respective xenografts. Conclusion We have established a well-characterised panel of human osteosarcoma xenograft lines, for use in preclinical evaluation of new anticancer therapies. This panel gives a relatively conservative estimate of the antitumor activity of the conventional drugs tested. With the exception of MTX, the effects of commonly used drugs in the clinic were reflected in the xenograft models. The responsiveness of human osteosarcoma xenografts to doxorubicin was conversely associated with GST- expression, whereas the sensitivity to cisplatin surprisingly showed positive correlation to MRP-1 expression