科研之友
微信
新浪微博
Facebook
分享链接
摘要
Primary brain tumors frequently return after radiation therapy (RT). In addition, RT can provoke changes in brain tissue that are difficult to distinguish from tumor recurrence using magnetic resonance imaging. Magnetic resonance spectroscopy (MRS) and spectroscopic imaging (MRSI) have been applied quite extensively with the aim of better differentiating recurrent brain tumors from radiation necrosis. To the best of our knowledge, however, there have not been any published papers within this context in which the results of MRS and MRSI have been analyzed in a comprehensive, integrative manner. Through meta-analysis, the present paper aims to determine which metabolite concentration ratios could potentially be the most reliable for differentiating post-RT recurrent brain tumor from radiation necrosis. We systematically reviewed the literature to find empirical studies of patients treated with RT for primary malignant brain tumors, and who developed a new lesion post-RT, which was assessed using MRS or MRSI. Reported data from individual patients were entered into a single data set for detailed statistical analysis. Six articles were identified that fulfilled the criteria for inclusion in the present study. From these, quantitative MRS/MRSI data were provided for sixty-three patients with recurrent brain tumors and for thirty-eight patients with radiation necrosis. Higher choline to creatine and choline to N-Acetyl Aspartate (NAA) concentration ratios were associated with a significantly greater likelihood of recurrent tumor, as opposed to radiation necrosis. This was found with and without statistical adjustment for high grade of the primary tumor, echo time (TE) and static magnetic field strength (B-0). Higher NAA to creatine and lactate to choline concentration ratios were associated with a significantly greater likelihood of radiation necrosis as opposed to recurrent tumor, both with and without adjustment for tumor grade, TE and B-0. Only the lactate to choline concentration ratio yielded 100 % correct prediction of all the cases for which data were available. However, data on lactate to choline concentration ratios were available for less than one-third of the patients; these were all recorded at long TE, at which most of the major signal components for the other metabolites have decayed to the level of background noise. No cutpoint values for the choline to creatine or choline to NAA concentration ratios could be identified that optimally distinguished recurrent tumor from radiation necrosis. We conclude that metabolite concentration ratios assessed within MRS/MRSI could potentially be helpful in distinguishing tumor recurrence from radiation necrosis. However, optimal distinction of these two entities has not yet been provided, mainly because of the nearly exclusive reliance upon the conventional estimations by post-processing of Fourier spectral envelopes with various fitting computations that are all equivocal. In order to more accurately identify recurrent brain tumor post-RT as opposed to radiation-induced changes, a more advanced and unequivocal mathematical approach for quantification of MRS/MRSI signals is needed. The answer of choice to this quest is offered by the universally applicable method of rational functions from the mathematical theory of approximations.
The most salient feature of rational functions is their capability to accurately describe the essential behavior of generic systems by means of the least number of quantifying parameters, as an indispensable prerequisite for a parsimonious mathematical modeling. The most established rational response function, which has passed the test of time across interdisciplinary research, is the Pad, approximant. Adapted to signal processing and accordingly called the fast Pad, transform, this versatile method is the optimal spectral analyzer for MRS/MRSI data. It is within such a strategy that the questions addressed in this study could be answered in an adequate manner.
- 出版日期2012-10
