Prediction of Glioma Stemlike Cell Infiltration in the Non-Contrast-Enhancing Area by Quantitative Measurement of Lactate on Magnetic Resonance Spectroscopy in Glioblastoma.

Background We previously reported that glioma stemlike cells (GSCs) exist in the area of the tumor periphery showing no gadolinium enhancement on magnetic resonance imaging. In the present work, we analyzed glucose metabolism to investigate whether lactate could be predictive of tumor invasiveness and of use in detection of the tumor invasion area in glioblastoma multiforme (GBM). Methods The expression of lactate dehydrogenase A (LDH-A) and pyruvate dehydrogenase (PDH) was investigated in 20 patients. In GSC lines, LDH-A and PDH expression also was examined in parallel to assessments of mitochondrial respiration. We then investigated the relationship between lactate/creatine ratios in the tumor periphery measured by magnetic resonance spectroscopy, using learning-compression-model algorithms and phenotypes of GBMs. Results In 20 GBMs, high-invasive GBM expressed LDH-A at significantly higher expression than did low-invasive GBM, whereas low-invasive GBM showed significantly higher expression of PDH than did high-invasive GBM. The highly invasive GSC line showed higher expression of LDH-A and lower expression of PDH compared with low-invasive GSC lines. The highly invasive GSC line also showed the lowest consumption of oxygen and the lowest production of adenosine triphosphate. Lactate levels, as measured by magnetic resonance spectroscopy, showed a significant positive correlation with LDH-A transcript levels, permitting classification of the GBMs into high-invasive and low-invasive phenotypes based on a cutoff value of 0.66 in the lactate/creatine ratio. Conclusions In the tumor periphery area of the highly invasive GBM, aerobic glycolysis was the predominant pathway for glucose metabolism, resulting in the accumulation of lactate. The level of lactate may facilitate prediction of the tumor-infiltrating area on GBM.