An assessment of artefacts in localized and non-localized 31P MRS studies of phosphate metabolites and pH in rat tumours.

UA hepatomas, GH3 prolactinomas and N-methyl-N-nitrosourea-induced mammary tumours, which were subcutaneously grown in rats, have been studied by 31P MRS using non-localized pulse-acquire, image selected in vivo spectroscopy (ISIS) and one-dimensional chemical shift imaging (1-D CSI) techniques. Comparisons have been made with measurements from acid extracts of these tumour types and surrounding tissues (i.e., muscle and skin). Since muscle containing high concentrations of phosphocreatine (PCr) is often found adjacent to the tumour, we have compared the ratio of the PCr to gamma-NTP peaks in the spectra with the same ratio calculated from the acid extract data, and have used deviations between the two sets of data to assess the discrimination of the MRS localization technique to signals from the tissue surrounding the tumour. Extract data showed an average NTP content of 1.25 mumol/g wet wt for all three tumour types. PCr (at 0.42 mumol/g wet wt), was significant only in the GH3 prolactinoma whereas it was negligible in the other tumour types (< 0.1 mumol/g wet wt). There was good agreement between the ISIS PCr/gamma-NTP ratio and the extract data for all tumours. However, the 1-D CSI data showed an unexpectedly large contamination of the tumour spectrum with PCr signals from the skin which was shown by subsequent phantom experiments to be due to the curved geometry of tumour and skin rather than Fourier bleed. In pH measurements by MRS it was found that biological variability was greater than the effects of artefacts (due to either the chemical shift artefact in the ISIS technique or partial volume effects) in the localization technique. An average pH of 7.2 was observed for all tumours. By initially comparing data from different localization schemes with that from chemical extracts potential sources of error have been highlighted and show that phantom studies alone are not sufficient to fully assess the accuracy of localized MRS data.