Neurology India Medknow Publications on behalf of the Neurological Society of India ISSN: 0028-3886 EISSN: 1998-4022 Vol. 58, Num. 6, 2010, pp. 914-921

Neurology India, Vol. 58, No. 6, November-December, 2010, pp. 914-921

Topic of the Issue: Original Article

White matter tract alterations assessed with diffusion tensor imaging and tractography in patients with solid posterior fossa tumors

Mario N Carvi y Nievas¹, Hans-Georg Hoellerhage¹, Christian Drahten²

¹ Department of Neurosurgical Clinic and Radiological, Clinic Frankfurt-Höchst, Frankfort, Germany
² Academic Teaching Hospital, J-W Goethe University, Frankfort, Germany
Correspondence Address: Mario N Carvi y Nievas, Oberarzt der Neurochirurgische Klinik, Klinikum Frankfurt Höchst, Academic Teaching Hospital of the Johann-Wolfgang-Goethe-Universität 65929 Goten street 6-8 , Germany
mcnievas@t-online.de

Date of Acceptance: 08-Jul-2010

Code Number: ni10257

PMID: 21150059
DOI: 10.4103/0028-3886.73746

Abstract

Background: This study assesses the tract alterations observed before and after resection of solid posterior fossa tumors (PFTs) using diffusion tensor imaging (DTI) and white matter tractography (WMT).

Materials and Methods: Pre- and post-surgical DTI and WMT data were acquired in eight patients undergoing surgical resections of PFT. A tensor deflection algorithm was used to reconstruct the tracts adjacent to the lesions. The tracts were evaluated regarding their spatial orientation and integrity. A software additionally assessed the luminosity and pixel values from specific regions of interest (ROIs) quantifying the magnitude of tract alterations. The individual neurological condition was additionally evaluated.

Results: In eight tumors (metastases (4), neurinomas (2) and meningiomas (2), the preoperative WMT revealed different tract alterations, including deviation (all cases), deformation (7 cases), thinning (8 cases) and apparent tract interruption (4 cases). The ROI histograms from noncompromised tracts showed a tendency for value concentration and peak formation, while affected tracts showed different dispersion patterns. After tumor resection, the compromised white matter tracts showed a resolution (3 cases) or reduction (5 cases) of the deviation. Postoperatively, 7 cases of tract thinning and 3 cases with tract interruptions showed an improvement. The comparison between ROI histograms from preoperative and postoperative compromised tracts mostly revealed a postoperative accentuated reduction of the luminosity with a simultaneous increase of pixels and improved histogram definition (homogeneous values distribution). At this time (5 weeks post-surgery), several neurological functions had improved to different levels.

Conclusions: Preoperative and postoperative tract alterations in patients with solid PFT can be accurately assessed with WMT. The magnitude of tract changes can be quantitatively analyzed by assessing color, signal brightness in compromised bundles.

Keywords: Posterior fossa tumor, white matter tractography

Introduction

The capability of diffusion tensor imaging (DTI) to display secondary alterations of white matter tracts caused by different tumoral lesions and brain diseases, as well as its potential utility for the treatment, planning and follow-up, has been already discussed in previous reports. ^{[1],[2],[3],[4],[5],[6],[7],[8],[9],[10],[11]} Most of these reports have analyzed patients with supratentorial gliomas compromising eloquent areas. ^{[1],[2],[3],[4],[8]} In such patients, the tumor-related infiltration of neural fibers and the associated tract alterations are difficult to be clearly interpreted. Moreover, at present, the magnitude of the post-interventional tract changes has not been quantified.

As is well known, the white matter tract anatomy of the posterior fossa follows characteristics and well-defined distribution patterns running through the peduncles. Due to the reduced anatomic space, even small tumors can produce measurable tract alterations in the posterior fossa. In this study, white matter tractography (WMT) using a tensor deflection algorithm was specifically used to reconstruct WM tracts adjacent to solid posterior fossa tumors (PFTs). In all tractography images, the assessed luminosity and pixel value from specific regions of interest (ROIs) were additionally evaluated, while trying to quantify the magnitude of the tract alterations. All tumors were resected by employing a debulking technique. This should even allow an improved preservation of the surrounding neural tracts and consequently a more accurate assessment of possible tract alterations. The early post-interventional clinical evolution of the patients was additionally evaluated in order to examine its possible association with the magnitude of the assessed tract alterations.

Materials and Methods

Patients

The pre- and post-surgical (between 1 and 5 weeks) DTI and WMT data of eight patients who underwent total surgical resections of their solid PFTs were analyzed. Clinical and anatomical data of the patients are given in [Table - 1]. All these patients were clinically examined by a neurologist blinded to the DTI findings in both pre- and post-surgical phases. The neurological evaluation was performed by the same physician in all patients.

Imaging - Diffusion tensor imaging

Diffusion tensor imaging was performed on a 1.5-T Magneton Symphony scanner (Siemens Erlangen, Germany). A single-shot spin-echo echo-planar imaging pulse sequence with diffusion-weighting gradients was used to obtain DTI images. The total image-acquisition time varied between 10 and 20 minutes, depending on the number of sections acquired. The diffusion tensor was calculated for each voxel by using single-value decomposition.

Image processing - Method of white matter tractography

Mean diffusivity (MD) and fractional anisotropy (FA) values were measured in the cerebella and brainstem white matter regions adjacent to the tumors. A tensor deflection algorithm was used to reconstruct WM tracts adjacent to these lesions. The spatial relationship between the tumors and the surrounding fibers was analyzed with FA, color-coded directional map and three-dimensional WM tractography.

Fiber trajectories were generated from regions enclosing a cross section of the tracts of interest using 3D software (Siemens, Erlangen, Germany). A continuous tracking algorithm was used in which the path follows the principal eigenvector of the diffusion tensor on a subvoxel level until the voxel edge is met, at which point the direction abruptly changes to that of the new voxel. Tracking terminated when the relative anisotropy of the voxel decreased to below 0.15, which indicated that the directionality of the vector had become unreliable. ^[12] The region of interest (ROI) in the surrounding areas of each tumor included several ascendant and descendant tracts in at least 3 projected cut surfaces (coronal, sagittal and transversal planes).

Brainstem-assessed fibers included the corticospinal tracts, the transverse pontine fibers as well as the lateral and medial lemniscus. [Figure - 1] shows the preoperative (a) and postoperative (b) magnetic resonance images (MRI) of a patient with a large neurinoma at the craniocervical junction compressing the low portion of the brainstem and medulla. The tumor is seen to displace posteriorly the assessed neural structures.

In order to simplify imaging analysis, WMT assessments of the cerebella were performed for areas where the bundles of the cerebellar peduncles could be easily differentiated from each other. Due to the volume of the analyzed tumors, a bundle of fibers rather than the isolated tracts was assessed. This included the following:

• The inferior cerebellar peduncle (containing afferent fibers from the spinocerebellar, cuneocerebellar, olivo-, reticulo- and vestibulo-cerebellar tracts; vermis efferents such as the cerebellovestibulospinal and reticulospinal tracts, as well as vestibulo-cerebellar efferents arising from the flocculonodular lobe)
• The middle cerebellar peduncle with afferent fibers via the pontocerebellar tract
• The superior cerebellar peduncle (containing axons of the cerebellothalamic tract or dentatothalamic tract) passing through and around the red nucleus to the thalamus

Additional tract connections between cerebellar cortex and nuclei were only assessed if these were satisfactorily reproduced on preoperative and postoperative images.

The tracts were evaluated regarding their spatial orientation (deviation, deformation) and integrity (thinning and interruption). A tract was considered "deviated" if it was displayed by abnormal location and/ or direction as a result of the lesion mass effect. A tract deformation was defined as the change in shape of the involved tract. Tract deviation and deformation were established by comparison with the unaffected homologous contralateral tract or after repeated comparison with known normal anatomy. A tract-thinning was characterized by reduced size and/ or anisotropy by at least 30% in comparison with the unaffected contralateral tract. A tract was considered to be "apparently interrupted" if any portion of the tract was visibly discontinuous on anisotropy-weighted directional color maps and fiber-tracking terminated at the discontinuity despite reasonable relaxation of stopping criteria.

Additionally, the magnitude of preoperative and postoperative tract alterations was compared by assessing the color signal intensity of the examined tracts (luminosity and pixel values) within specific ROIs with a commercially available software program (Adobe Photoshop 7.0, USA). Each of the pixels that represent an image stored inside a computer has a pixel value which describes how bright that pixel is. The color in an image is made up of three components. One of them, the luminosity, assesses the brightness of the color. The obtained pixel and luminosity values as well as their preoperative and postoperative histograms were individually analyzed. The purpose of this examination was to obtain additional information quantifying the magnitude of the afferent and efferent tract alterations. The obtained histograms displayed on the x-axis the scale of color's luminosity from the darkest (0) to the brightest (255) values; and on the y-axis, the amount of assessed pixels.

Results

The histological examination of the lesions revealed 4 metastatic tumors, 2 neurinomas and 2 meningiomas. White matter tracts of the posterior fossa outside of the area where the tumors developed their mass-occupying effect did not reveal any considerable changes. In these tracts, no spatial, morphological or histogram-related alterations could be found between their respective preoperative and postoperative examinations. However, at least three afferent or efferent reconstructed tracts in contact with the tumors demonstrated different kind of alterations in each patient. Tractography displayed the complex spatial relationships between WM tracts and lesions much better in all 3D reconstructed maps. Comparing the compromised tracts of the ipsilateral and contralateral sides, the preoperative WMT revealed a series of tract alteration patterns on the compromised side, including deviation (all cases), deformation (7 cases), thinning (8 cases) and apparent tract interruption (4 cases). After tumor resection, the organization of compromised tracts appeared more similar to normal anatomy, with either disappearance or reduction of the deviation (3 cases and 5 cases, respectively). Postoperatively, cases of tract thinning (7) and those of tract interruptions (3 cases) improved. Tract deformations improved in all patients who were preoperatively found to be affected. [Table - 1] summarizes the preoperative and postoperative spatial tract alterations, as well as some clinical and anatomical data of patients.

The magnitudes of the assessed preoperative and postoperative tract alterations and the postoperative neurological conditions are given in [Table - 2]. The comparison between the preoperative and postoperative compromised tracts using automatically calculated ROI histograms mostly revealed a postoperative reduction of the luminosity with simultaneous increase of the pixel values. The reconstructed histograms from these affected bundles showed an improved postoperative homogeneous visualization from color distribution values of the assessed tracts. While the histograms of preoperatively compromised tracts did not show any defined pattern of value distribution, those from postoperatively released tracts showed a clear tendency towards value concentration and individual peak formation. All these changes increased image definition and were not strongly influenced by the date of postoperative MRI examination.

[Figure - 2] shows three preoperatively obtained histograms, two from the affected corticospinal tracts (over and below the lesion) and one from the less compromised lateral lemniscus, of the patient with a neurinoma at the craniocervical junction. The two histograms assessing corticospinal tracts show middle-to-high luminosity with associated low pixel values. There is no defined pattern of value distribution in the histograms of these affected fibers; while in the histogram of less compromised lateral lemniscus, a tendency towards value concentration and peak formation can be clearly observed.

[Figure - 3] shows the postoperative evaluation of the same patient. The spatial deviation of the compromised fibers became less pronounced, and the two histograms evaluating the corticospinal tracts showed at this time a clear tendency towards value concentration and peak formation. The preoperatively assessed peak-building of the less affected lateral lemniscus tract showed only scarcely measurable changes.

In our study, findings such as those differentiating between preoperatively compromised and noncompromised tracts were observed in all patients. [Figure - 4] shows three preoperatively obtained histograms, one from compromised tracts running into the inferior cerebellar peduncle, one from the less affected corticospinal tract and one from the noncompromised transverse pontine fibers, of the patient with a metastatic lesion in the right tonsil. The spatial tract alterations and the lack of a defined pattern of value distribution in the histogram of the inferior cerebellar peduncle, as well as a less-pronounced value disaggregation for the corticospinal tract, are evident. In the same image, a tendency towards value concentration and peak formation in the unaffected transverse pontine fibers can be clearly observed.

The absence of a defined pattern of value distribution, as well as a pronounced histogram disaggregation tendency, can also be observed in the preoperative histograms of the affected superior and middle cerebellar peduncles of the patient with a metastatic lesion in the cerebellar central lobule and culmen [Figure - 5].

Several postoperatively examined neurological functions had improved 5 weeks after surgery to different levels depending on the location of the process. [Table - 2] shows some associations between the location of the lesion and the magnitude of the assessed radiological and neurological changes.

Discussion

White matter tractography based on diffusion tensor imaging has become a well-accepted noninvasive tool for exploring the white matter architecture of the human brain in vivo.^{[12],[13],[14]}

Due to the fact that the white matter tracts anatomy in the posterior fossa follows characteristics and well-defined spatial distribution patterns running through the peduncles or ascending throughout the brainstem; the reconstruction, identification and assessment of such complex fascicular pathways could be easily performed in this study. Taking into account that this series only included patients with solid PFTs removed by employing a debulking technique, an improved postoperative preservation of most of these surrounding neural tracts was possible, and consequently a more detailed assessment of all postoperative tract alterations. This study also found that inside the compromised tracts surrounding the lesions, a postoperative reduction of color luminosity with simultaneous increase of pixel values can be frequently observed. This is the first report in the literature where the magnitude of the tract alterations in the affected nerve bundles was quantitatively assessed. The graphic illustration of these findings (reconstructed histograms) showed an improved and homogeneous postoperative visualization of the color values in all compromised tracts. While the histograms of preoperatively affected tracts did not show any defined pattern of value distribution, those from unaffected or postoperatively released tracts showed a clear tendency towards building up of individual peaks or value concentration.

At present, some authors have reported that WMT revealed a series of tract alteration patterns, including deviation, deformation, infiltration and apparent tract interruption, in patients who underwent resection of their tumoral lesions. ^[5],[8],[15] However, these reports included different types of tumors. Smits et al. incorporated functional magnetic resonance imaging (fMRI) into WMT in the preoperative assessment of patients with different types of brain tumors. ^[16]

At this point, it is important to remark that, in patients suffering from infiltrative space-occupying lesions with distortion of anatomical landmarks, WMT usually has interpretation difficulties. While in meningiomas, DTT illustrated a bulk displacement of the corticospinal tract in the affected hemisphere as well as preservation of the deviated axons, in anaplastic astrocytoma the fiber-tracking analysis demonstrated a disruption of WM tracts at the tumor origin as well as intact axons through areas of tumor infiltration. ^[17]

Other authors investigated the advantage of WMT in patients treated with radiosurgery. ^[18] In such cases, the obtained information was useful in reducing irradiation to surrounding normal structures. Gulati et al. reported that, in their surgically treated patients, the combination of blood oxygenation level-dependent fMRI, tractography and 3D ultrasound facilitated maximal tumor resection with minimal deficits. ^[19] Krishnan et al. even postulate that WMT could also predict the natural history of regional progression in irradiated patients with primary brain tumors. ^[4] These authors found a strong relationship between routes of elevated water diffusion from the primary tumor and the location of tumor progression.

Recently, some authors reported that, compared with the information provided by conventional MRI, DTI and WMT provide superior visualization of lesions located in brainstem. ^[15],[20]

Regarding the specific employment of WMT in posterior fossa lesions, Taoka et al. reported that, tractography enabled the identification of those fibers considered to represent the facial nerves in a patient with a vestibular schwannoma. ^[21] We confirmed such observations in a case examined by us (number 8 in [Table - 1]).

In a previous report, Lui et al. determined whether or not abnormalities of DTI and WMT correlate with corticospinal tract lesions and clinical status in patients with primary PFT. ^[6] In their study, patients with well-circumscribed lesions and weakness had higher mean diffusivity and lower FA in the contralateral corticospinal tract. No such associations were seen in patients with infiltrating tumors. In addition to these observations and other in previous studies, the present report allowed a quantitative assessment of the magnitude of tract alterations by analyzing the individual changes in each compromised bundle. We believe that the postoperative decompression improved the optical definition of preoperatively compressed tracts as well as reduced their spatial and morphological alterations, contributing to the homogeneous distribution in histogram values.

There were no problems or disadvantages with the employment of this technique. However, setting and assessment of ROIs must be performed by a physician with good anatomical knowledge of involved areas. Unfortunately, the small number of cases in this study prevents a more elaborate conclusive statistical analysis of the data. However, we think that these results should be kept in mind and reconsidered when planning future clinical studies investigating the relationship between tumors and white matter tract alterations.

References

Copyright 2010 - Neurology India

The following images related to this document are available:

Photo images