Assessing global invasion of newly diagnosed glial tumors with whole-brain proton MR spectroscopy

Abstract

Background and purpose: Because of their invasive nature, high-grade glial tumors are uniformly fatal. The purpose of this study was to quantify MR imaging-occult, glial tumor infiltration beyond its radiologic margin through its consequent neuronal cell damage, assessed by the global concentration decline of the neuronal marker N-acetylaspartate (NAA).

Methods: Seventeen patients (10 men; median age, 39 years; age range, 23-79 years) with radiologically suspected (later pathologically confirmed) supratentorial glial neoplasms, and 17 age- and sex-matched controls were studied. Their whole-brain NAA (WBNAA) amounts were obtained with proton MR spectroscopy: for patients on the day of surgery (n = 17), 1 day postsurgery (n = 15), and once for each control. To convert into concentrations, suitable for intersubject comparison, patients' global NAA amounts were divided by their brain volumes segmented from MR imaging. Least squares regression was used to analyze the data.

Results: Pre- and postoperative WBNAA (mean +/- SD) of 9.2 +/- 2.1 and 9.7 +/- 1.8 mmol/L, respectively, in patients were indistinguishable (P = .369) but significantly lower than in controls (12.5 +/- 1.4 mmol/L). Mean resected tumor size (n = 15) was approximately 3% of total brain volume.

Conclusion: The average 26% WBNAA deficit in the patients, which persisted following surgical resection, cannot be explained merely by depletion within the approximately 3% MR imaging-visible tumor volume or an age-dependent effect. Although there could be several possible causes of such widespread decline--perineuronal satellitosis, neuronal deafferentation, Wallerian and retrograde degeneration, vasogenic edema, functional diaschisis, secondary vascular changes--most are a direct or indirect reflection of extensive, MR imaging-occult, microscopic tumor cell infiltration, diffusely throughout the otherwise "normal-appearing" brain.

Fig 1.

Histopathology of the invasive edge of the glioblastoma multiforme from patient 12 in the Table. Top left, A, Low power, ×50, hematoxylin and eosin stain depicting tumor core (TC) and brain adjacent to tumor (BAT). Note the high cellularity of the core in the white matter, with invading cells spreading as far as the leptomeningeal space (LMS). Top right, B, High power, ×400, of the BAT emphasizing perineuronal satellitosis—glioma cells around neuronal cell bodies (arrows). Bottom left, C, Medium power, ×100, with a monoclonal antibody (NeuN) staining neurons. Bottom right, D, High power, ×400, view of C highlighting immunoreactive neurons (arrows) surrounded by infiltrating tumor cells. These vacuolated neurons appear unhealthy, and therefore, are likely to contain less NAA. A healthy neuron is indicated for comparison (arrowhead).

Fig 2.

Top, A and B, Segmented preoperative T1-weighted sagittal section for 2 patients, 15 and 5 in the Table, respectively. The red sulci and ventricle outline is the output of the tissue/CSF segmentation process. Patient A had a 41-cm³ grade III tumor, 3.0% of his brain volume, and 37% WBNAA deficit. Patient B had a 93-cm³ grade II tumor, 6.2% of his brain volume, but only half the WBNAA decline (∼20%). Bottom, A′ and B′, Corresponding whole-head ¹H-MR spectroscopy. The hatched regions indicate the peak-areas used to obtain Q_NAA of Equation (1). Note the overall similarity, indicating cross-sectional reproducibility, and good lipid suppression. Also note that, because localization relies on knowledge that NAA is found only in neuronal cells, it cannot quantify the brain’s contribution to metabolites also present in other tissue types (eg, choline, creatine, and so forth).

Fig 3.

First, 2nd (median), and 3rd quartiles (box) and ±95% (whiskers) of WBNAA in the 17 controls, patients day of surgery (n = 17) and 1 day postsurgery (n = 14). Note that pre- and postsurgical WBNAA are significantly different from the controls, but not each other.

Fig 4.

Scatter plot of WBNAA versus age for the patients before surgery and their age- and sex-matched controls. Note that the patients’ WBNAA deficit is significant (P < .0005), even after taking into account age (P = .007) and sex effects (P > .05).

Fig 5.

First, 2nd (median), and 3rd quartiles (box) and ±95% (whiskers) of preoperative WBNAA in the 9 patients with low-grade tumors compared with their matched controls (left) and 8 patients with high-grade tumors compared with their matched controls (right). Note preoperative WBNAA was lower in patients with high-grade (8.3 ± 1.8 mmol/L) than in those with low-grade (10.0 ± 2.1 mmol/L) tumors.