Glial Neoplasms without Elevated Choline-Creatine Ratios

The role of proton MR spectroscopy (MRS) in the differential diagnosis of a brain lesion often centers on cautiously assessing the choline (Cho) and total creatine (Cr) resonances and the choline-creatine ratio (Cho/Cr) in spectra from the lesion. In this issue of the AJNR, Londono et al and Saraf-Lavi et al underscore the pitfall of excluding neoplasm from the differential diagnosis when there is no elevation of Cho or Cho/Cr. The results from both groups of investigators indicate that some glial neoplasms, which present with variable mass effect and a lack of postcontrast enhancement on MR images, may not show elevation of Cho or Cho/Cr relative to values from normal-appearing brain, yet do show a marked increase in the 3.5–3.6 ppm resonance assigned to myo-Inositol (m-Ins) or glycine (Gly) or both. Evidence favoring m-Ins as the metabolite responsible for the increase may be gleaned from evaluation of both short TE and long TE spectra, and possibly from changes in the ∼3.35 ppm scyllo-Inositol (s-Ins) resonance. Thus, other spectral measures should be sought before eliminating neoplasm from the differential diagnosis for a brain lesion with certain imaging features and without elevation of Cho or Cho/Cr.

For intracerebral lesions with variable mass effect and postcontrast enhancement, short TE (∼ 20–30 ms) and longer TE (∼ 135 ms) spectra showing marked elevation of Cho or Cho/Cr, and diminished N-acetyl (NA) or NA/Cr, favor a diagnosis of neoplasm over infection, inflammation, ischemia, or infarction. (Note: NA represents the combined resonances of N-acetyl aspartate [NAA] and N-acetyl aspartylglutamate [NAAG]. NAA and NAAG are often unresolved, and since NAA is the dominant metabolite, NAA is used interchangeably with NA for practical purposes.) Spectra showing only mild elevation of Cho or Cho/Cr are less helpful, because non-neoplastic conditions, including post-radiation therapy necrosis, can produce similar changes. In such cases, physiologic MR imaging that displays diffusion or perfusion properties can aid in narrowing the differential diagnosis. In general, the larger the increase in Cho or Cho/Cr, the more likely that a solitary focal or infiltrating lesion is a glial neoplasm when there is no history or other evidence of metastatic disease or hematologic malignancy. The converse of this statement may also be true; a Cho level or Cho/Cr ratio that is not elevated is unlikely to be a neoplasm. Nonetheless, as illustrated by Londono et al and Saraf-Lavi et al, the pitfall of excluding neoplasm from the differential diagnosis can and should be avoided. The key is to pay attention to the metabolite concentrations (Cho, Cr, and NA) and the ratios (low Cho/Cr values may be due to elevated Cr), and to evaluate other resonances, especially those at 3.5–3.6 (or 3.5 –3.7) ppm and 3.35 ppm.

The report by Londono and colleagues is a sequel to an earlier study by the same group (1) in which patients with previously treated astrocytic neoplasms (low-grade astrocytoma, anaplastic astrocytoma, and glioblastoma multiforme [GBM]) were examined by short TE (stimulated echo acquisition mode) single voxel spectroscopy. In that study, Cho/Cr was elevated in the tumor spectra and showed a trend toward higher values in GBMs compared with those in low-grade astrocytomas, as has been generally reported; however, the intriguing finding was that values for the ratio of the 3.5–3.6 ppm resonance (presumed m-Ins) relative to Cr showed an inverse trend, being higher in low-grade astrocytomas than in GBMs. In the current case report, the finding of an elevated 3.5–3.6 ppm resonance in the multi-voxel spectra (spectroscopic imaging, using point resolved spectroscopy technique with a TE of 30 ms) from a pathologically proven low-grade astrocytoma (grade II according to the World Health Organization classification system) corroborates the earlier results, yet is accompanied by a new finding—no significant increase in Cho or in Cho/Cr. Interestingly, this unusual combination of MRS findings corresponds to the findings in the case of gliomatosis cerebri (GC), reported by Saraf-Lavi et al in this issue. Thus, the reader may conclude that low-grade astrocytoma can have spectral findings similar to those of GC (a diffusely infiltrative lesion involving at least two lobes of the brain by definition), just as the former may resemble GC on routine MR imaging and stereotactic biopsy results. This conclusion, however, seems to contradict the recent hypothesis by Galanaud et al (2) that MRS can differentiate GC from low-grade glioma. Is Galanaud’s hypothesis oversimplified? Perhaps. An alternative interpretation would be that the grade II astrocytoma reported by Londono et al is actually a case of GC! The spectral evidence necessary to confirm this latter interpretation is inconclusive, however, since Londono et al used different spectral analysis techniques from those used by Galanaud and colleagues to establish their criteria.

The relationships between spectral results and the pathophysiology of the tumors are discussed in both case reports. Regarding Cho and Cho/Cr, both groups of investigators attribute the relatively normal values to diminished (or disrupted) membrane lipid turnover and a low proportion of rapidly dividing cells (“lack of cellular proliferation”) occurring in low-grade as compared with high-grade tumors. Tumor grade alone, however, seems unlikely to account for the Cho/Cr results because 1) the GC lesion was not strictly low grade and 2) most low-grade astrocytomas evaluated previously by Castillo and colleagues were found to have elevated Cho/Cr values (1). Similar issues have been raised by Galanaud et al. Londono and colleagues have suggested an alternative possibility: the presence of mixed oligodendroglial components in the tumor could have contributed to the unusual spectral results.

Regarding the 3.5–3.6 ppm resonance, its identity and the implications for pathophysiology are approached somewhat differently in the two reports. Two metabolites, m-Ins and Gly, have resonances detectable on short TE spectra in the 3.5–3.6 ppm region, and in healthy volunteers a single unresolved peak at approximately 3.56 ppm is usually assigned to m-Ins with a small contribution (∼10%) from Gly. Londono and colleagues speculate that increased glycine, due to the presence of oligodendroglial cells, may account for the increase in the 3.5–3.6 ppm resonance. While there is some evidence in the literature to support this, there is also compelling in vitro MRS data suggesting that increased Gly occurs in high-grade (eg, GBM) rather than low-grade tumors (3). As discussed by Saraf-Lavi and colleagues, and by others (2, 4), a rough estimate of the Gly contribution to the 3.5–3.6 ppm resonance at short TE can be made by examining the same chemical shift region on long TE spectra. A strong resonance in this region would favor abundant Gly, since its -CH₂ group has a relatively long T2 and no J-coupling effects. m-Ins, on the other hand, produces a weak or absent resonance at TE 135 ms due to complex coupling of its -CH groups. This latter condition was observed by Saraf-Lavi and colleagues, who concluded that m-Ins primarily accounts for the increased 3.5–3.6 ppm resonance on short TE spectra.

Additional support for this conclusion may come from the prominent singlet resonance that was observed at ∼3.35 ppm in both the short TE and longer TE GC spectra, consistent with elevated levels of s-Ins. m-Ins and s-Ins are the two most abundant isomers of the five naturally occurring stereoisomers of inositol, an amino alcohol. The concentration of s-Ins seems to be tightly coupled to the m-Ins concentration at a ratio of {12 m-Ins:1 s-Ins} (5), so that the elevated s-Ins might provide indirect evidence of elevated m-Ins. Unfortunately, this reasoning is oversimplified and possibly misleading because 1) the previously reported tight coupling was based on spectra from normal brain and not low-grade astrocytoma or GC and 2) even for normal brain the tight coupling has been questioned (6). Clearly the observations by Saraf-Lavi and colleagues on the prominent s-Ins peak deserve further study before their clinical significance can be assessed.

How does one explain the elevated m-Ins in low-grade astrocytomas? Two explanations, based on evidence that m-Ins is a “glial marker,” are discussed by the authors of the case reports in this issue. In the first explanation, the elevation is attributed to changes in the phospholipid composition or abundance of glial cell membranes or both. Castillo and colleagues (1) earlier proposed a more specific mechanism in which mitogen-influenced metabolism of phophatidyl inositol (PI) results in 1) increased PI synthesis and corresponding depletion of the MR-visible m-Ins pool in high-grade astrocytomas, and conversely, 2) decreased PI synthesis and corresponding elevation of MR-visible m-Ins in low-grade astrocytomas. Galanaud et al (2) have proposed that the elevated m-Ins in GC is related to “proliferation of glial elements or, more probably, activation of normal glia.” In the second explanation, elevation of m-Ins is attributed to its action as an organic osmolyte, playing a major role in the volume and osmoregulation of astrocytes, although the nature of this role in low-grade astrocytomas (or GC) versus that in high-grade astrocytomas remains to be elucidated.

In summary, what have we learned from these two case reports? First, for hyperintense lesions without specific morphologic findings on fluid-attenuated inversion recovery images and minimal or no enhancement on postcontrast T1-weighted images, the lack of significant Cho or Cho/Cr elevation does not exclude the diagnosis of a primary glial neoplasm. Second, a prominent resonance at 3.5–3.6 ppm likely signifies increased m-Ins, and both short TE and long TE spectra should be acquired in order to better characterize this and other resonances. Third, if m-Ins, m-Ins/Cr, and/or m-Ins/NA are shown to be elevated, include low-grade astrocytoma and gliomatosis cerebri in the differential diagnosis. Differentiation between these two lesions may become possible in the future with a better understanding of their spectral properties and underlying pathophysiology.

• Copyright © American Society of Neuroradiology