Multivoxel Proton MR Spectroscopy in Malformations of Cortical Development

Materials and Methods

MR Spectroscopy

MR spectroscopy protocol included a multivoxel acquisition by using a point-resolved spectroscopy sequence technique with the following parameters: section thickness = 10 mm, TR = 1500 ms, TE = 135 ms, FOV = 24 cm, NEX = 1, and 16 × 16 phase-encoding steps. The MCD were identified and classified on the basis of conventional images of the MR imaging study, and the image that best represented the MCD in the axial T2-weighted sequence was identified. Within this image, a large volume of interest was defined for field homogeneity optimization. The size of this volume of interest varied, depending on the MCD, from 64 to 112 mm in the right-left direction and 35–103 mm in the anteroposterior direction. This larger volume of interest included MCD, surrounding tissues, and the NACS in cases of unilateral involvement. Water-suppressed multivoxel MR spectroscopy of this volume of interest was performed, lasting approximately 6.5 minutes.

As mentioned previously, we also studied a control group with the same MR spectroscopy protocol. In every volunteer, MR spectroscopy was performed from 2 axial sections, 1 at the level of the atrium of the lateral ventricles (labeled as inferior) and the other at the level of centrum semiovale bilaterally (labeled as superior). The size of the large inferior volume of interest varied, depending on the brain size, from 79 to 107 mm in the right-left direction and 50–120 mm in the anteroposterior direction. The superior volume of interest varied from 60 to 87 mm in the right-left direction and 73–118 mm in the anteroposterior direction.

MR spectroscopy raw data were transferred to a workstation (UltraSPARC 60, Sun Microsystems, Santa Clara, Calif) and processed by using the spectroscopy analysis software SAGE (GE Healthcare, Milwaukee, Wis). Postprocessing steps included line broadening of 3 Hz, zero-filling to 1024 points, and automatic baseline and phase correction. Every volume of interest was divided into smaller ones (15 × 15 × 10 mm³), and only some of these smaller volumes of interest were selected for further metabolite quantification. In patients, the smaller volume of interest was centered in the MCD to minimize contamination with normal-appearing tissues. In cases of unilateral involvement, a smaller voxel was also studied in NACS (Fig 1A, -B). Metabolite quantification was performed after fitting the spectral curve, by using the Marquardt-Levenberg method, to 3 Lorentzian lines: N-acetylaspartate (NAA) at 2.02 ppm, creatine (Cr) at 3.03 ppm, and choline (Cho) at 3.23 ppm. NAA/Cr was calculated from the obtained amplitudes values.

• Download figure
• Open in new tab
• Download powerpoint

Fig 1.

Axial inversion-recovery image (A) shows a subcortical heterotopia on the right side. Axial T2-weighted image (B) demonstrates the selected volumes of interest located at the MCD (1) and at the normal-appearing contralateral side (2). MR spectroscopy obtained from the volume of interest centered at the MCD (C) shows a decreased NAA/Cr. D, MR spectroscopy obtained from the NACS. E, MR spectroscopy obtained from the white matter in a control subject.

In the control group, in the 2 large volumes of interest selected for MR spectroscopy (inferior and superior), NAA/Cr was calculated for up to 8 smaller volumes of interest: 4 in the inferior large volume of interest (composed predominantly of periventricular white matter or of gray matter on the left and right side, respectively) (Fig 2A) and 4 in the superior large volume of interest (composed predominantly of white matter or of gray matter on the left and right side, respectively) (Fig 2B). At the level of the lateral ventricles, we studied the parieto-occipital gray and white matter. At the level of the centrum semiovale, we sampled the frontoparietal gray and white matter bilaterally. Hence, the small volumes of interest were classified according to their location as inferior or as superior, according to the predominant composition as gray or white matter, and according to the hemisphere as right or left.

• Download figure
• Open in new tab
• Download powerpoint

Fig 2.

Axial T2-weighted images demonstrate the 4 selected volumes of interest in the inferior level (A) and also in the superior level (B).

Results

We studied the spectra from 16 volumes of interest including MCD (9 gray matter heterotopias and 7 cortical dysplasias) and also 10 from the NACS.

There was no lactate peak present in the spectra, neither in the patients nor in the control group. NAA/Cr obtained in patients and in the control subjects is summarized in the Table.

Discussion

MCD are disruptions of the cortical formation process that occur at different stages of the embryonic life. These interruptions can affect proliferation, migration, and/or organization of neuronal cells. A varied spectrum of neuropathologic changes can be observed in MCD.¹⁰ In MCD, epileptogenic activity could be explained by abnormalities in cortical architecture, with consequent abnormal synaptic circuitry and unstable membrane excitability.¹¹ Widespread use of MR imaging allowed recognition of these abnormalities in vivo and their further categorization from a radiologic standpoint.^{2–4,12–14}

MR spectroscopy has been used in patients with intractable temporal lobe epilepsy and demonstrated a reduction in NAA, a marker of neuronal loss or dysfunction, ipsilateral to the side of the seizure focus identified clinically or by using paraclinical studies such as EEG, positron-emission tomography, or single-photon emission CT (SPECT).^15–18

NAA is present in high concentration in neurons, but its exact role in the brain is unknown. NAA is considered a marker of neuronal attenuation and viability, and its concentration decreases in response to various types of insults to the brain.⁵

Previous published studies about MR spectroscopy and MCD are very different in the number of cases and methodologies used.^9,19–31 Of these studies, 8 used single-voxel MR spectroscopy and 6 used multivoxel MR spectroscopy. Six are case reports, and in the remaining 8 articles, the number of patients studied varied from 8 to 31. In 11 of these articles, the metabolite ratios obtained in the lesion were compared with those in a control group, and in 9 articles, the NACS was also studied.^{9,19,20,24–27,29}

The abnormality most frequently found in MCD was decreased NAA or NAA/Cr in the lesion compared with a that in the control group,^{9,20,22–24,27,28,31} in agreement with our results, but some articles reported cases presenting a normal NAA peak in MCD compared with that in a control group.^21,29,30 This observation might be explained by the fact that these authors used single-voxel MR spectroscopy with a larger voxel volume size (3–8 cm³) than the one used in our study (2.25 cm³). This larger voxel may be influenced more intensely by partial volume averaging, and that could potentially affect the sensitivity of the method to subtle abnormalities. Mueller et al,³¹ by using spectroscopic imaging, showed that it is possible to identify metabolically abnormal clusters interspaced in metabolically normal regions in some cases of cortical malformations.

One of the studies²⁹ found reduced NAA/Cr only in cortical dysplasia. An in vitro MR spectroscopy study also showed decreased NAA in surgical biopsies of cortical dysplasias.³² Kuzniecky et al²³ studied patients with MCD by using MR spectroscopy with 4.1T equipment and hypothesized that the abnormal metabolic ratios could indicate an immature cellular component because in some cases of focal cortical dysplasia, the number of neurons and glia were not different from that of controls.

Previous MR spectroscopy studies have already demonstrated that a structurally normal-appearing perilesional zone is characterized by metabolic abnormalities, similar to what is observed in MCD.^24,28,31 Another study using diffusion anisotropy and diffusivity also demonstrated abnormality beyond the limits of the visualized lesion.³³

For focal lesions, a valid strategy is to perform comparisons between the metabolic findings of the lesion spectrum and the contralateral side,⁸ with the NACS used as a control to detect biochemical abnormalities in the lesion. To verify if we could compare the NAA/Cr of NACS with MCD, we also evaluated NACS in patients presenting unilateral involvement. MR spectroscopy of NACS showed decreased NAA/Cr in comparison with that of the control group and slightly increased NAA/Cr in comparison with that of MCD, but NAA/Cr from NACS and MCD did not show significant statistical differences. In the control group, we evaluated the gray as well as the white matter, whereas the contralateral volumes of interest were composed mainly of white matter. Nine previous studies evaluated, in different ways, the NACS.^{9,19,20,24–29} However, in only 4 was it compared directly with a control group, and the reported results were not in agreement. Hanefeld et al²⁰ described MR spectroscopy in 2 children with hemimegalencephaly and evaluated the white matter of the contralateral side, which showed reduced NAA in comparison with that of a control group. Preul et al²⁷ described the case of 1 patient with unilateral heterotopia and cortical dysplasia and did not find reduced NAA in NACS compared with that in a control group. Simone et al⁹ evaluated 11 patients by using single-voxel MR spectroscopy and found no differences for NAA/Cr between NACS and the white matter or gray matter of the control group. Woermann et al²⁸ studied the contralateral white matter of 6 patients with unilateral MCD by using multivoxel spectroscopy and detected reduced NAA/Cr in 4 cases.

As shown previously, literature reports are not in agreement as to whether the NACS shows altered NAA/Cr. We were able to study the NACS in 10 patients and compare it with that of a large group of controls (n = 30), in whom gray and white matter were studied separately. With our data, we contribute to reinforcing the idea that NAA/Cr is in fact decreased in the NACS of MCD. The lack of a statistically significant difference between the lesions and the NACS found in our results suggests that MCD lesions should not be compared with the contralateral side of the same patient because the comparison might prevent the detection of metabolic alterations.

The idea that in MCD, the visible lesion is only the “tip of the iceberg” has been extensively discussed, and indirect evidence from SPECT and EEG studies in unilaterally affected patients supports the hypothesis that the NACS could also be abnormal.^13,34 This could explain why surgical resection does not yield good results in the control of epilepsy in some patients with focal MCD, suggesting that the visualized MCD does not constitute the entire epileptogenic zone and that abnormalities can be found also in the perilesional as well as in the contralateral side.^35–37 Wieshmann et al,³⁸ by using diffusion tensor imaging, found reduced anisotropy affecting both hemispheres in a patient with unilateral MCD. They hypothesized that there was loss of directional organization in the contralateral white matter. Because our NACS was composed mainly of white matter, our spectroscopic observations of reduced NAA/Cr are in agreement with those of Wieshmann et al, suggesting that there is really some dysfunction at this level. Our findings reinforce the idea that a more widespread metabolic abnormality extends beyond the visualized lesion, suggesting neuronal loss and/or dysfunction in the NACS, corroborating the previous concept.

The abnormality in the contralateral side can also be explained by the fact that because the development of gray and white matter is closely linked, malformations can affect both cortex and the white matter in both hemispheres.

The reliability of MR spectroscopy for revealing abnormalities in normal-appearing brain has been previously demonstrated in other neurologic diseases, such as multiple sclerosis and X-linked adrenoleukodystrophy.^39–41 In these progressive neurologic diseases, biochemical alterations may anticipate the presence of structural damage.

A limitation of this study was that we did not perform absolute quantification of metabolite concentrations, which would require too long acquisition times and possibly sedation of patients. Because we chose to use long TE spectroscopy to improve the spectral resolution of peaks, it was not possible to study other short T2 metabolites such as glutamate or myo-inositol, which may provide other insights into the metabolic disturbances of MCD. We did not include ratios with Cho because in an analysis of the control group data (not shown here), we found important differences in the Cho/Cr ratios between the superior and inferior areas. Hence, for comparison purposes of Cho/Cr data between patients and controls, we would need to divide the patient group (n = 16) also into 2 subgroups (inferior and superior), reducing considerably the number of patients for each group, which would also reduce the power of our statistical analysis.

The limited number of each type of MCD examined in this study precluded the detection of possible differences between MCD subtypes. Further studies should be performed to elucidate possible differences between different types of MCD, by using MR spectroscopy with short and long TEs.

Conclusion

Despite the fact that MR imaging has revolutionized the detection of the anatomic abnormalities found in MCD, MR spectroscopy could be an important key to understanding its pathophysiologic role, demonstrating that damage extends beyond the identified lesions on MR imaging, including NACS in cases of unilateral MCD. Although in the literature a commonly used strategy is to compare focal lesions with the normal contralateral side,⁸ our results indicate that in MCD, the NACS should not be used as a control.