Abstract
Metabolic modeling of ^13^C NMR spectroscopy (^13^C MRS) data using two‐compartment neuronal‐glial models enabled non‐invasive measurements of the glutamate‐glutamine cycle rate (V~NT~) in the brain in vivo. However, the reliability of such two‐compartment metabolic modeling has not been examined thoroughly. This study uses Monte‐Carlo simulations to investigate the reliability of metabolic modeling of ^13^C positional enrichment time courses measured in brain amino acids such as glutamate and glutamine during [1‐^13^C]‐ or [1,6‐^13^C~2~]glucose infusion. Results show that the determination of V~NT~ is not very precise under experimental conditions typical of in vivo NMR studies, whereas the neuronal TCA cycle rate V~TCA(N)~ is determined with a much higher precision. Consistent with these results, simulated ^13^C positional enrichment curves for glutamate and glutamine are much more sensitive to the value of V~TCA(N)~ than to the value of V~NT~. We conclude that the determination of the glutamate‐glutamine cycle rate V~NT~ using ^13^C MRS is relatively unreliable when fitting ^13^C positional enrichment curves obtained during [1‐^13^C] or [1,6‐^13^C~2~]glucose infusion. Further developments are needed to improve the determination of V~NT~, for example using additional information from ^13^C‐^13^C isotopomers and/or using glial specific substrates such as [2‐^13^C]acetate. © 2007 Wiley‐Liss, Inc.