Predicting resistance or response to chemotherapy by proton magnetic resonance spectroscopy in neuroblastoma

Background: We previously showed that proton magnetic resonance spectroscopy (H-1-MRS) enables estimation of neuroblastoma tumor viability. Here we investigated if H-1-MRS can predict response or resistance to chemotherapy in neuroblastoma. Methods: Neuroblastoma cell lines with various drug sensitivities were treated with cytotoxic drugs (cisplatin, etoposide, and irinotecan) and examined by H-1-MRS. Viability was assessed by trypan blue staining and the 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyl tetrazolium bromide assay. Nude rats carrying drug-sensitive or drug-resistant neuroblastoma xenografts were treated for 4 days with irinotecan (n = 11) or saline (n = 11) and were examined with H-1-MRS at 4.7 T before and during treatment. The Wilcoxon matched-pairs test was used to test statistical significance of difference within treatment groups. Independent groups were compared using the Mann-Whitney U test. Correlation was assessed with Spearman's rank correlation. All statistical tests were two-sided. Results: Cytotoxic drug treatment of drug-sensitive SH-SY5Y neuroblastoma cells resulted in increased methylene and polyunsaturated fatty acid resonances and decreased choline resonance. The methylene/choline ratio correlated with cell death (r(s) =.94, P<.001) and was increased in cisplatin-treated drug-sensitive (SH-SY5Y, IMR-32) but not drug-resistant [SK-N-BE(2), SK-N-FI, SK-N-AS] cell lines. No changes were observed in SK-N-BE(2) cells treated with irinotecan or cisplatin, whereas circumvention of the resistance by arsenic trioxide treatment led to lipid accumulation and choline depletion. Irinotecan therapy of rats carrying drug-sensitive xenografts caused the methylene/choline ratio of tumors to increase eightfold after 3 days (95% confidence interval [CI] = fivefold to 12-fold; P =.005 compared with pretreatment spectra at day 0) and caused tumors to regress statistically significantly on day 10 compared with pretreatment volume on day 0 (difference = -60%, 95% CI = -12% to -100%, n = 6; P =.012). The methylene/choline ratio of nonregressing drug-resistant xenografts was unaffected. No differences were observed after saline treatment. Conclusions: Response or resistance to chemotherapy is accurately predicted by H-1-MRS in experimental neuroblastoma models in vivo.