Diffusion tensor magnetic resonance imaging of glial brain tumors

Diffusion tensor magnetic resonance imaging of glial brain tumors

European Journal of Radiology 74 (2010) 428–436 Contents lists available at ScienceDirect European Journal of Radiology journal homepage: www.elsevi...

3MB Sizes 0 Downloads 135 Views

European Journal of Radiology 74 (2010) 428–436

Contents lists available at ScienceDirect

European Journal of Radiology journal homepage: www.elsevier.com/locate/ejrad

Diffusion tensor magnetic resonance imaging of glial brain tumors Jiˇrí Ferda a,∗ , Jan Kastner a , Petr Mukenˇsnabl b , Milan Choc c , Jana Horemuˇzová a , Eva Ferdová a , Boris Kreuzberg a a b c

Department of Radiology, Charles University Hospital Plzen, Medical Faculty Plzen, Alej Svobody 80, 304 60 Plzen, ˇ Czech Republic Sˇ ikl’s Institute of Pathological Anatomy, Charles University Hospital Plzen, Medical Faculty Plzen, Alej Svobody 80, 304 60 Plzen, ˇ Czech Republic Department of Neurosurgery, Charles University Hospital Plzen, Medical Faculty Plzen, Alej Svobody 80, 304 60 Plzen, ˇ Czech Republic

a r t i c l e

i n f o

Article history: Received 28 September 2008 Accepted 13 March 2009 Keywords: Diffusion tensor imaging (DTI) Tractography Glial tumors Intracranial neoplasm

a b s t r a c t Aim: To evaluate the author’s experience with the use of diffusion tensor magnetic resonance imaging (DTI) on patients with glial tumors. Methods: A retrospective evaluation of a group of 24 patients with glial tumors was performed. There were eight patients with Grade II, eight patients with Grade III and eight patients with Grade IV tumors with a histologically proven diagnosis. All the patients underwent routine imaging including T2 weighted images, multidirectional diffusion weighted imaging (measured in 60 non-collinear directions) and T1 weighted non-enhanced and contrast enhanced images. The imaging sequence and evaluation software were produced by Massachusetts General Hospital Corporation (Boston, MA, USA). Fractional anisotropy (FA) maps were calculated in all patients. The white matter FA changes were assessed within the tumorous tissue, on the tumorous borderline and in the normally appearing white matter adjacent to the tumor. A three-dimensional model of the white matter tract was created to demonstrate the space relationship of the tumor and the capsula interna or corpus callosum in each case using the following fiber tracing parameters: FA step 0.25 and a tensor declination angle of 45 gr. An additional assessment of the tumorous tissue enhancement was performed. Results: A uniform homogenous structure with sharp demargination of the Grade II tumors and the wide rim of the intermedial FA in all Grade III tumors respectively, were found during the evaluation of the FA maps. In Grade IV tumors a variable demargination was noted on the FA maps. The sensitivity and specificity for the discrimination of low- and high-grade glial tumors using FA maps was revealed to be 81% and 87% respectively. If the evaluation of the contrast enhancement was combined with the evaluation of the FA maps, both sensitivity and specificity were 100%. Conclusion: Although the evaluation of the fractional anisotropy maps is not sufficient for glioma grading, the combination of the contrast enhancement pattern and fractional anisotropy maps evaluation improves the possibility of distinguishing low- and high-grade glial tumors. Three-dimensional models of the white matter fibers in the corpus callosum and the internal capsule may be used in the presurgical planning. © 2009 Elsevier Ireland Ltd. All rights reserved.

1. Introduction In spite of the progress in surgical procedures, radiotherapy and chemotherapy, the prognosis of patients with glial tumors is still largely dependent on their degree of differentiation and remains poor for Grades III and IV. As it is rather important to determine the local extent of a tumor and its infiltration to important structures to be able to estimate the probability of therapeutic success, new imaging possibilities of the structure of healthy and tumor-infiltrated cerebral tissues using diffusion weighted

∗ Corresponding author. E-mail address: [email protected] (J. Ferda). 0720-048X/$ – see front matter © 2009 Elsevier Ireland Ltd. All rights reserved. doi:10.1016/j.ejrad.2009.03.030

magnetic resonance imaging (MRI) sequences have been studied extensively over the last decade. Magnetic resonance imaging employs signal changes in the tissues caused by the diffusion of water molecules in diffusion weighted imaging (DWI). DWI is currently a routine component of investigation protocols. These protocols play a fundamental role in the detection and evaluation of ischemic brain damage, especially of the time that has elapsed since its origin, and can also be used for imaging traumatic changes in the brain tissue, the assessment of demyelinating disorder activity and in evaluation of the cellularity of brain tumors. Diffusion weighted imaging employs the detection of changes of diffusivity of water molecules in individual pathological processes for their characterization, in differential diagnostics, or to quantify the degrees of the damage. A significant improvement

J. Ferda et al. / European Journal of Radiology 74 (2010) 428–436

in water diffusion imaging in tissues has been achieved through the introduction of a new, much more comprehensive imaging method that takes into account, in addition to scalar components, the vector component that manifests as heterogeneity of diffusion in space—diffusion anisotropy. Though discovered in the 1960s [1], the possibility of diffusion anisotropy imaging was only used in the 1990s for tensor imaging—diffusion tensor imaging (DTI) [2]. Early studies were concerned with the possibility of imaging nervous tissue microstructure and micro-architecture of the nervous tissues [3–5], including changes accompanying pathological processes affecting the brain tissue [6]. Tracing contiguous/continuing white matter fibers using vector imaging in DTI has been carried out since the late 1990s, with attempts to obtain three-dimensional images of white matter tracts [7–9]. The presented work is concerned with our experience obtained with DTI in brain imaging in patients with glial tumors. 2. Materials and methods Our analyzed sample included 24 patients with cerebral glial tumors with homogenous representation of Grades II–IV tumors diagnosed according to WHO criteria. Evaluation of the sample included images of fractional anisotropy (FA) maps, apparent diffusion coefficient (ADC) maps, and we performed three-dimensional reconstruction of the white matter tracts in cases where the tumor involved the region of the corpus callosum or that of the corticospinal tract. Simultaneously, images were obtained according to standard investigation protocol, and the analysis used evaluations of post-contrast T1-weighted spin echo images with magnetization transfer (SE MTC) and T1-weighted gradient echo images (3D FLASH) with an isotropic voxel 1 mm × 1 mm × 1 mm. All patients provided routine informed consent prior to the investigation. 2.1. Data acquisition The investigations were conducted with the Magnetom Avanto 1.5 T instrument (Siemens, Erlangen, Germany) in a twelve-channel coil allowing the use of parallel acquisition techniques. Data were acquired with multidirectional diffusion weighted echo planar sequence (Massachusetts General Hospital Corporation, Boston, MA, USA), with the following parameters: TR 7600 ms, TE 83 ms, band width 1630 MHz, field of view 287 mm × 287 mm, matrix 128 × 128, layer width 2.2 mm, number of layers 60, measurement in 60 non-collinear directions with b value = 700 s/mm2 and one measurement with b value = 0 s/mm2 . The integrated parallel acquisition technique (iPAT) was used, with acceleration factor 2 with the GRAPPA algorithm to reduce image distortion arising in echo planar sequences. 2.2. Data processing All DTI data were processed with DTI software (Massachusetts General Hospital Corporation, Boston, MA, USA). Fractional anisotropy and apparent diffusion coefficient maps were generated gradually, and we further used DWI images with b value = 700 s/mm2 and “low B” images, i.e. actually T2* images. Using 3D applications from the above software package, we also evaluated ADC and fractional anisotropy values in voxels in the homogenous tumor centre, the transition intermediary—zone, and in normal apparent white matter—NAWM. To distinguish central tumor regions, transition zones or NAWM, a comparison was made of the T2* images and FA maps in the orthogonal 3D system of the evaluation software. For tumors located close to the corpus callosum, or when infiltration of this structure by the tumor was very probable, reconstruction 3D tractography was used, the starting region of interest

429

(ROI) for the construction of fibers being in the sagittal plane, following a longitudinal section through the corpus callosum. In tumors in intimate relationship with the corticospinal tract the starting ROIs were chosen in axial image, at a site where it was possible to identify the corticospinal tract bundle, or at least the course of the posterior limb of the internal capsule. To search for contiguous vectors, we used the allowed values for a deviation angle between adjacent voxels of 45◦ and for a threshold for FA difference of 0.25, to determine the white matter voxels. 2.3. Evaluation In all groups, visual evaluation of the maps was used to assess the FA homogeneity of the focus (lesion), the presence of intermediary transition between the tumor mass and the surrounding normal apparent white matter. In cases where the width of the border of the intermediary FA was 5 mm or less (two voxels) the lesion was evaluated as sharply circumscribed, while it was considered without clear demarcation when the border was broader. In the region of interest chosen, we determined the FA value in the central part of the tumor, in its intermediary transition zone and in the NAWM. Simultaneously the presence and character of the contrast medium saturation was registered. Three-dimensional models of the corpus callosum or the corticospinal tract were used to evaluate the possibility of white matter visualization, changes of integrity of the tracts, or their replacement (dislocation). 3. Results The group of low-grade gliomas included two cases of Grade II oligodendrogliomas, two Grade II oligoastrocytomas, and four Grade II fibrillary astrocytomas. The group of Grade III tumors included one anaplastic oligoastrocytoma, two cases of anaplastic oligodendroglioma, and five anaplastic astrocytomas. The group of Grade IV tumors comprised tumors characterized histologically as glioblastoma multiforme. There were six cases of sharp transition between the tumor mass and the NAWM, with borders of intermediary FA 5 mm or less, i.e. more than two voxels in the Grade II group of tumors. One tumor in this group was recorded to invade the corpus callosum, manifesting as marked reduction of the FA with an observable structural disturbance in the three-dimensional model as well. In this one case, islet-like contrast medium saturation was found, but the tumor did not exhibit any observable intermediary FA border. One case had a transitional intermediary zone that was exactly 5 mm. There was no finding of corticospinal tract disturbance, with the exception of a shift in one of these cases. The mean FA value in the centre of the tumors was 0.012, and 0.215 and 0.617 in the intermediary region and the surroundings, respectively. In the Grade III tumor group no recording was made in any of the cases of a sharp transition between the tumor tissue and the NAWM, with the intermediary border widths being less than 5 mm, i.e. the border was broader in all eight cases. The mean FA value in the tumor centre was 0.015, and 0.155 and 0.530 in the intermediary region and surrounding tissue, respectively. All three cases exhibited marked changes that concerned the anatomical location of the corticospinal tract, and three cases were found with direct corpus callosum invasion. Regarding contrast medium saturation, this was found in a small tissue region only in three cases, and was surrounded by an extensive zone of reduced FA; there was also the finding of a lesion with annular saturation in another 5 cases. In the Grade IV tumor group there were three cases of sharp transition between the tumor tissue and the NAWM, with an intermediary FA border less than 5 mm, with five cases having borders broader than 5 mm. The mean FA value in the tumor centre was 0.008, 0.251 in the intermediary region, and 0.562 in the surrounding region.

430

J. Ferda et al. / European Journal of Radiology 74 (2010) 428–436

Table 1 A value of the FA and contrast enhancement evaluation in detection of the high-grade glioma.

TP TN FP FN SENS SPEC

FA

CE

FA + CE

13 7 1 3 0.8125 0.875

16 7 1 0 1 0.875

16 8 0 0 1 1

FA, wide rim of the lower fractional anisotropy; CE, contrast enhancement; FA + CE, conjunction of positivity of FA and CE; TP, true positivity; TN, true negativity; FP, false positivity; FN, false negativity; SENS, sensitivity; SPEC, specificity.

A marked deviation of the corticospinal tract was recorded in three cases, one of which also showed corpus callosum invasion, manifested as reduced anisotropy, contrast medium saturation and the presence of a defect observable in a three-dimensional model of the white matter pathways. Contrast medium saturation was recorded in all cases, having the character of a “geographic” lesion. The finding of isolated corpus callosum invasion was made only once. A summary of the results is given in Table 1. Evaluation with FA maps led to findings with 81% sensitivity and 87% specificity for the differentiation between low- and high-grade glial tumors. When an assessment was made in conjunction with the contrast medium saturation, FA maps brought values of sensitivity and specificity that were 100% equally for the whole sample. 4. Discussion 4.1. Diffusion of water molecules in the brain tissue The principle of diffusion is the constant disordered movement of water molecules called Brownian motion. Unhindered, the direction of this movement is entirely random. An environment in which diffusion occurs equally in all directions is called isotropic. Brain cerebrospinal fluid is a characteristic area of isotropic diffusion. In the presence of a restriction to movement in some directions the diffusion occurring in such environments is called anisotropic. The most commonly encountered barrier to diffusion in living tissues is the cellular wall [3]. The presence of obstacles to diffusion in the grey matter of the brain is comparable in all directions, and the diffusion, though restricted in comparison with the cerebrospinal fluid, is isotropic. The situation in the white matter is quite the opposite. Axon fibers are long tubular structures in which diffusion is much more intense in the direction of their course. Structures formed by fibers with a high degree of directional ordering channel the diffusion in one predominant direction [2–5]. The highest level of such ordering found in white matter fibers occurs in the corpus callosum and corticospinal tract [9,10]. 4.2. Parameters of diffusion Changes of position of water molecules occurring in time can be described with Gaussian distribution. Diffusivity D, a value that characterizes the diffusion in a particular material, is a vectorial quantity. In DWI, however, diffusivity is only described using scalar values. In a medium that is isotropic from the point of view of the diffusion of water molecules, such as the grey matter in the brain, it is sufficient to use a scalar quantity, the apparent diffusion coefficient. The ADC expresses the mobility of free water molecules in a tissue, located in both the intracellular and extracellular spaces, the importance of the molecules present in the vascular space being minimum [6,11].

White matter structures, on the other hand, that prevent the free diffusion of water molecules in some directions through their architecture, are anisotropic in terms of diffusivity, and the use of the simple scalar ADC does not necessarily result in accurate characterization of a possible change in water molecule diffusion in individual directions. The mathematical concept expressing the anisotropy of Gaussian diffusion is the diffusion tensor. All tensors can be analyzed into three mutually perpendicular non-zero components with scalar diffusivity values 1 , 2 , 3 , respectively, where the component with highest absolute value is designated 1 [2–4]. The values above are assigned a triplet of orthogonal vectors, ε1 , ε2 , ε3 . The largest ε1 vector is represented by the main direction of diffusion. The scalar value of diffusivity (D) in a given voxel, referred to by some authors as mean diffusivity (MD) is obtained as the arithmetic average of 1 , 2 , 3 . 4.3. Diffusion tensor imaging It is possible to use strong bipolar magnetic field gradients to obtain images expressing the differences in water molecule diffusion in the tissues [1]. Such images are independent of T1, T2 and proton density. The Brownian motion of molecules between the application of two diffusion gradients leads to irreversible dephasing of the MRI signal, resulting in signal amplitude in each individual voxel. ADC as measured in the brain is computed on the basis of the use of several sequences with varying sensitivity to diffusion imaging. It has already been mentioned that diffusivity is a vector quantity, which results in different images being obtained in different gradient orientations in the same tissue layer. This is the reason for using the computation of the diffusion tensor trace in routine DWI sequences where the trace is just the average of the diagonal elements of a matrix obtained through diffusion measurement in three orthogonal planes. The situation is different for DTI—differing diffusivity in different planes is used to compute the sizes of individual directional vectors. Sequences obtained with gradual switching on of bipolar gradients for diffusion imaging in multiple directions are called MDDWI—multidirectional diffusion weighted imaging [4,6]. Computation of the correct orientation of the direction of dominant diffusion requires values from at least 6 non-collinear directions. These values can only be obtained after there has been gradual switching on of bipolar gradients in at least six directions, and is thus obtained from at least six data sets of diffusion weighting. In practice 6, 12, and frequently even more, e.g. 60, 128, 256 directions of diffusion weighting are used. For these measurements the following b values are used: b = 700 s/mm2 , up to b = 1000s/mm2 . Data acquisition is completed by adding one data set with measurements using b = 0 s/mm2 [4,6,12–14]. In cases where DTI is used for three-dimensional reconstruction in tractography, it is best to choose the parameters of data acquisition so that the space is imaged with the isotropic cube voxel [13,14]. Currently, echo planar sequences (EPI) are used with motion correction—motion-corrected multi-shot echoplanar imaging. Parallel acquisition techniques (PAT) are used to reduce image distortion [14]. Problems accompanying imaging consist in artifacts that are most frequently caused by eddy currents, signal losses due to variability of susceptibility, and artifacts due to motion [14]. 4.4. Methods of imaging diffusion tensors The measurement of diffusion with MDDWI gives rise to a rather a voluminous data set. Anisotropy imaging can be accomplished using many rather complicated mathematical computations, all of

J. Ferda et al. / European Journal of Radiology 74 (2010) 428–436

which use the construction of orthogonal vectors as an intermediary step. The common product (output) of various algorithms currently contains computation of diffusivity (D), apparent diffusion coefficient, and fractional anisotropy. The latter is computed as part of the diffusion tensor that corresponds to anisotropy. The value of FA = 0 is typical for the isotropic medium, which in the brain is represented by the cerebrospinal fluid [4,5]. A high symmetrical anisotropy where diffusion occurs on one direction only is characterized by the value of FA = 1 [4,5]. The product of this computation of fractional anisotropy is then given as a greyscale fractional anisotropy map, and the computed image of the brain is analogous to maps of the apparent diffusion coefficient [4–6]. In addition to FA maps, images are created that take into account, besides the scalar value of fractional anisotropy, the directional component of the vector of predominant diffusion. The results are color maps of the brain called DEC FA (directionally encoded fraction anisotropy) maps [7,13]. Using a color spectrum, the predominant diffusion direction can be expressed in voxels; as a convention, blue has been chosen to show the superior–inferior direction, while the anterior–posterior and left-to-right directions are indicated by green and red, respectively. Modifications of DEC FA maps are representations where each voxel is depicted as a probabilistic diffusion relation that is graphically expressed as either an ellipsoid or a prism. The more elongated the shape of a graphical element, the higher the probability of directional ordering in a single direction.

431

Three-dimensional imaging of the brain’s white matter with diffusion tensors is based on the idea that the diffusion is restricted to the long-axis direction of the axonal fibers only, over the course of the white matter fibers, through their myelin sheaths, as well as on the analogy that white matter fibers passing from voxel to voxel have similar FA values and a directional deviation corresponding to the curvature of the fiber. The computation of fractional anisotropy and direction of predominant diffusion is then used to trace the direction of the white matter fibers, fiber tracking [8,9,15]. Using special software it is possible to look up contiguous vectors and FA values, after a range for the FA values and a range for angular deviations in the adjacent voxels have been permitted. According to contiguous vectors sought it is possible to construct probabilistic graphics expression of the spatial ordering of white matter fiber bundles. Due to the fact that some anatomical structures in the brain exhibit a high percentage values of ordering of axons to bundles – tracts – it is possible to use 3D tractography to obtain images of their spatial arrangement [8,9,15]. The simplest examples of structures with high degrees of spatial arrangement in the white matter, highly amenable to imaging with 3D tractography, are the corpus callosum and the corticospinal tract. The principal problems to be solved by the algorithms used to reconstruct three-dimensional courses of fibers include the problem of the crossing, kissing, and branching fibers [8,9,15,16]. As a result of the fact that some can touch each other, or cross or branch in their course, the algorithm searching for contiguous

Fig. 1. Oligoastrocytoma gr. II. in the left angular region (A) TSE T2; (B) echoplanar image b = 0 s/mm2 ; (C) echoplanar image b = 700 s/mm2 ; (D) ADC map, decreased values showing increased tumorous cellularity; (E) FA map, the tumor is sharply demarginated; (F) contrast enhanced SE T1 with magnetisation transfer, no contrast enhancement is present in the tumorous tissue.

432

J. Ferda et al. / European Journal of Radiology 74 (2010) 428–436

vectors can erroneously reconstruct adjacent vectors with angular deviation, still within the allowed declination range, without there in fact being such a morphological contiguity in the form of a contiguous/continuing white matter fiber. This, together with differing views regarding the threshold of FA differences between neighboring elements, explains why it is impossible to regard 3Dtractography as a standardized method. Early applications of diffusion magnetic resonance imaging of the brain published were concerned with the detection and differential diagnosis of cerebral ischemia, and this application is still predominant as an indication of DWI investigation. The possibilities to use ADC in differential diagnosis of intracranial tumors and to distinguish between peritumorous edema and infiltration were investigated at the turn of the millennium [17–20]. Most of the studies concluded unequivocally that isolated ADC values cannot be used to differentiate between peritumorous infiltrations, and it is also impossible to obtain information about the degree of differentiation of glial tumors. The only result obtained consisted in the finding that there is a reduction of ADC values in tumor tissues with high cellularity compared to low-cellularity tumors—the probability of higher grading is reduced in solid tumors with high ADC values. The occurrence of cystic formations in the glioblastoma multiforme, however, is associated with the relation between the ADC and grading being below the level of statistical significance [18,19,21]. DTI imaging can be used to record both ADC and FA values in the cystic and solid parts of tumors, and in tumor zones with contrast medium saturation, in the white matter surrounding a tumor, with normal apparent white matter [22,23]. Intracranial tumors have been observed to exhibit a changed organization of

the brain’s white matter [24]. Individual studies were concerned with differences in fractional anisotropy in the tumor tissue, and demonstrated an association between the character of the damage in the white matter fibers and the changed FA value. In tumors with high degree of tissue disintegrations reduced anisotropy has been observed as compared to the NAWM, and even values consistent with isotropy in cystic and necrotic formations. In homogeneous tumors without contrast medium enhancement, the presence of reduced anisotropy is regarded to be a sign of higher grading (Figs. 1–3). The detection of peritumorous invasion relies on changes in white matter anisotropy and diffusivity. Studies suggest [25] that an increasing water volume found in edematous white matter with intact fiber organization is associated with increased diffusivity and reduced anisotropy. Different findings are made in tissues with fiber disorganization and simultaneous but minor increase of water volume, where the marked reduction of FA without a significant change of diffusivity occurs. This finding could in fact be used to differentiate between secondary and primary brain tumors’ [26]. This method of assessment of the extent of peritumorous infiltration through tumor elements is problematic in Grades III and IV. These diffusely growing tumors – anaplastic astrocytoma and, for example gliomatosis cerebri – are characterized by tumor infiltration that follows the course of fibers in the white matter. This fact could also explain why tumors that, though having a higher growth potential but still respecting the organization of fibers in the white matter in the process of their local invasion, do not disrupt the white matter architecture, and there is no detectable reduction in FA [27–30]—similar results can be arrived at from the findings in the present study. The results of our study also sup-

Fig. 2. Oligodendroglioma gr. II. left frontal lobe (A) TSE T2; (B) echoplanar image b = 0 s/mm2 ; (C) echoplanar image b = 700 s/mm2 ; (D) ADC map; (E) FA map, tumor invaded the genu corporis callosi; (F) contrast enhanced SE T1 with magnetisation transfer, focal contrast enhancement is present in the tumorous tissue.

J. Ferda et al. / European Journal of Radiology 74 (2010) 428–436

433

Fig. 3. Diffusely growing anaplastic astrocytoma gr. III. (A) TSE T2; (B and C) contrast enhanced SE T1 with magnetisation transfer, an island-like enhancement pattern is showed; (D–F) ADC maps demonstrated diffusely increased ADC of the white matter through both hemispheres; (G–I) FA maps, enlarged areas of decreased FA.

port conclusions made in other studies that isolated FA values do not carry sufficient information for differential diagnosis to enable the differentiation of glioblastoma multiforme (Grade IV astrocytomas). The finding of a rather low mean FA value does not point here to disintegration of the white matter fibers, but is rather a sign of the early presence of fluid inside cystoid formations in the tumor. In some glioblastomas, the sharp transition from FA inside the tumor tissue to the surrounding NAWM is caused by the peritumorous NAWM being compressed through the expansive nature of the tumor, and the transition region having, at the same time, a low FA value approaching the low values from the tumor centre. The reason is a combination of extracellular edema and infiltration by tumor elements (this phenomenon is noticeable in the casuistic given in Fig. 4). The problem in obtaining an objective

evaluation of real FA values in tumor tissue, the intermediary zone and the surrounding NAWM is to be seen in the fact that there are large inter-individual differences in FA on the one hand and marked intra-individual differences in various brain tissue regions on the other, compounded further by the changes in white matter FA depending on age. Attempts at generalizing FA quantification are made on the basis of comparisons of the proportions of FA in the damaged area with contra-lateral white matter [14,22,23] (Fig. 5). To evaluate the white matter surrounding a tumor one can possibly use DE DTI imaging. White matter surrounding a tumor is characterized by four possible kinds of characteristic behavior [13]. Firstly, there is the tract with normal FA and ADC, with abnormal positions of tensors associated with expansive behavior of the

434

J. Ferda et al. / European Journal of Radiology 74 (2010) 428–436

Fig. 4. Glioblastoma multiform left frontal.lobe (A–C) contrast enhanced 3D-GRE T1 (FLASH); (D–F) ADC maps, the central part of the tumor is formed of a cystic structure, enhancing the rim with low ADC; (G–I) FA maps, the tumor is not creeping to the contralateral side, but the corpus callosum has been partially infiltrated, the enhancing rim shows decreased FA, the frontolateral white matter has increased its FA due to the space-occupying behavior of the tumor.

tumor tissue. Secondly, there are moderately decreasing FA values and moderately increasing ADC in a still normal localization of the tract. Cases of location of tensors that deviate from normal positions and changes of FA and ADC corresponding to the second type of behavior constitute the third kind of behavior. The fourth type of behavior consists in a change from anisotropic to isotropic tissue. Types 2 and 3 white matter changes in the neighborhood of a tumor increase the probability of infiltration of the surrounding white matter with tumor elements. There are observations of the zone of white matter being replaced by the expansive behavior of a tumor with a locally increased FA value. The differences in white matter behavior described above also explain the variable picture of three-dimensional reconstructions obtained with 3D tractography in glial tumors in our experience.

Our study made use of a combination of the known phenomenon that higher grade tumors are saturated by a contrast medium after intravenous application, with a new finding regarding the transition border of the FA. In our patient sample, we were able to achieve 100% sensitivity as well as specificity, when these two findings were used in conjunction, for differentiation of Grades III and IV tumors. These results support the effectiveness of the inclusion of DTI in the algorithm of MRI and FA maps as part of brain tumor evaluation. Limitations of the results obtained in our sample include the subjective selection of the regions for evaluations of diffusion parameters, heterogeneity of the sample regarding histological findings in Grades II and III tumors, with a proportion of oligodendroglial tumors, and the small sample size. Most of these

J. Ferda et al. / European Journal of Radiology 74 (2010) 428–436

435

Fig. 5. Three-dimensional tractography. (A) Oligodendroglioma gr. II – a partially destroyed genu corporis callosi – the same patient as shown in Fig. 2, right lateral view; (B) glioblastoma multiform, some fibers from the corpus callosum have been pushed around the tumor, some have been destroyed, the same patient as shown in Fig. 4, left lateral view; (C) anaplastic astrocytoma gr. III, infiltrative tumorous tissue growing between the white matter fibers of the occipital lobe, right lateral view; (D) anaplastic astrocytoma gr. III, lateral displacement of the white matter fibers of the corticospinal tract, the same patient as shown in Fig. 3, left lateral view.

limitations, however, are faced by the majority of both early and recent studies dealing with similar topics.

Acknowledgement The study was supported by the research project MSM0021620819 of the Czech Government.

5. Conclusion Diffusion tensor imaging offers significant additional information that might be of help in the differentiation of tumors with infiltrative growth from circumscribed tumors, and it is possible, in conjunction with the evaluation of ADC and contrast medium saturation assessment, to advance the accuracy of estimation of tumor grading prior to final determination of the morphological diagnosis. The finding of reduced fractional anisotropy in the white matter around a tumor in the direction increases the probability of peritumorous white matter infiltration by tumor elements. The evaluation of planar images and three-dimensional modeling of white matter structures helps us to obtain a clearer idea of the spatial arrangement of some clearly defined white matter structures, such as the course of the corticospinal tract and the arrangement of the corpus callosum, and to evaluate their spatial relationships to the tumor tissue. The results of our own observations point to the possibility of obtaining good quality estimates of the character of the growth and grading of glial tumors using a combination of conventional MRI imaging and DTI.

References [1] Stejskal EO. Use of spin echoes in a pulsed magnetic-field gradient to study naisotropic, restricted diffusion and flow. J Chem Phys 1965;43: 3597–603. [2] Pierpaoli C, Jezzard P, Baser PJ, Barnet A, Di Chiro G. Diffusion tensor imaging of the human brain. Radiology 1996;201:637–48. [3] Hajnal JV, Doran M, Hall AS, et al. MR imaging of anisotropically restricted diffusion of water in the nervous system: technical, anatomic and pathologic considerations. J Comput Assist Tomogr 1991;15:1–18. [4] Pierpaoli C, Basser PJ. Toward a guantitative assessment of diffusion anisotropy. Magn Reson Med 1996;36:893–906. [5] Basser PJ, Pierpaoli C. Microstructural and physiological features of tissues elucidated by quantitative-diffusion-tensor MRI. J Magn Reson B 1996;111: 209–19. [6] Sorensen A, Wu O, Copen W, et al. human acute cerebral ischemia: detection of changes in water diffusion anisotropy by using MR imaging. Radiology 1999;212:785–92. [7] Pajevic S, Pierpaoli C. Color schemes to represent the orientation of anisotropic tissues from diffusion tensor data: application to white matter fiber tract mapping in human brain. Magn Reson Med 1999;42:526–40. [8] Virta A, Barnet A, Pierpaoli C. Visualizing and characterizing white matter fiber structure and architecture in the human pyramidal tract using diffusion tensor MRI. Magn Reson Imaging 1999;17:1121–33.

436

J. Ferda et al. / European Journal of Radiology 74 (2010) 428–436

[9] Mori S, Crain BJ, Chacko VP, van Zijl PC. Three-dimensional tracking of axonal projections in the brain by magnetic resonance imaging. Ann Neurol 1999;45:265–9. [10] Stadlbauer A, Nimsky C, Gruber S, et al. Changes in fiber integrity, diffusivity, and metabolism of the pyramidal tract adjacent to gliomas: a quantitative diffusion tensor fiber tracking and MR spectroscopic imaging study. AJNR Am J Neuroradiol 2007;28(3):462–9. [11] Krabbe K, Gideon P, Wagn P, Hansen U, Thomsen C, Madsen F. MR diffusion imaging of human intracranial tumours. Neuroradiology 1997;39(7):483–9. [12] Beppu T, Inoue T, Shibata Y, et al. Fractional anisotropy value by diffusion tensor magnetic resonance imaging as a predictor of cell density and proliferation activity of glioblastomas. Surg Neurol 2005;63(1):56–61. [13] Field AS, Alexander AL, Wu YC, Hasan KM, Witwer B, Badie B. Diffusion tensor eigenvector directional color imaging patterns in the evaluation of cerebral white matter tracts altered by tumor. J Magn Reson Imaging 2004;20(4):555–62. [14] Papanikolaou N, Karampekios S, Papadaki E, Malamas M, Maris T, Gourtsoyiannis N. Fractional anisotropy and mean diffusivity measurements on normal human brain: comparison between low-and high-resolution diffusion tensor imaging sequences. Eur Radiol 2006;16(1):187–92. [15] Basser PJ, Pajevic S, Pierpaoli C, Duda J, Aldroubi A. In vivo fiber tractography using DT-MRI data. Magn Reson Med 2000;44:625–32. [16] Mori S, Frederiksen K, van Zijl PC, et al. Brain white matter anatomy of tumor patients evaluated with diffusion tensor imaging. Ann Neurol 2002;51:377–80. [17] Muti M, Aprile I, Principi M, et al. Study on the variations of the apparent diffusion coefficient in areas of solid tumor in high grade gliomas. Magn Reson Imaging 2002;20(9):635–41. [18] Castillo M, Smith JK, Kwock L, Wilber K. Apparent diffusion coefficients in the evaluation of high-grade cerebral gliomas. AJNR Am J Neuroradiol 2001;22(1):60–4. [19] Pauleit D, Langen KJ, Floeth F, et al. Can the apparent diffusion coefficient be used as a noninvasive parameter to distinguish tumor tissue from peritumoral tissue in cerebral gliomas? J Magn Reson Imaging 2004;20(5):758–64.

[20] Tien RD, Felsberg GJ, Friedman H, Brown M, MacFall J. MR imaging of high-grade cerebral gliomas: value of diffusion-weighted echoplanar pulse sequences. AJR Am J Roentgenol 1994;162(3):671–7. [21] Kono K, Inoue Y, Nakayama K, et al. The role of diffusion-weighted imaging in patients with brain tumors. AJNR Am J Neuroradiol 2001;22(6):1081–8. [22] Stadlbauer A, Ganslandt O, Buslei R, et al. Gliomas: histopathologic evaluation of changes in directionality and magnitude of water diffusion at diffusion-tensor MR imaging. Radiology 2006;240(3):803–10. [23] Goebell E, Paustenbach S, Vaeterlein O, et al. Low-grade and anaplastic gliomas: differences in architecture evaluated with diffusion-tensor MR imaging. Radiology 2006;239(1):217–22. [24] Sinha S, Bastin ME, Whittle IR, Wardlaw JM. Diffusion tensor MR imaging of high-grade cerebral gliomas. AJNR Am J Neuroradiol 2002;23(4):520–7. [25] Lu S, Ahn D, Johnson G, Law M, Zagzag D, Grossman RI. Diffusion-tensor MR imaging of intracranial neoplasia and associated peritumoral edema: introduction of the tumor infiltration index. Radiology 2004;232(1):221–8. [26] Provenzale JM, McGraw P, Mhatre P, Guo AC, Delong D. Peritumoral brain regions in gliomas and meningiomas: investigation with isotropic diffusion-weighted MR imaging and diffusion-tensor MR imaging. Radiology 2004;232(2):451–60. [27] Lu S, Ahn D, Johnson G, Cha S. Peritumoral diffusion tensor imaging of high-grade gliomas and metastatic brain tumors. AJNR Am J Neuroradiol 2003;24(5):937–41. [28] Morita K, Matsuzawa H, Fujii Y, Tanaka R, Kwee IL, Nakada T. Diffusion tensor analysis of peritumoral edema using lambda chart analysis indicative of the heterogeneity of the microstructure within edema. J Neurosurg 2005;102(2):336–41. [29] Tropine A, Vucurevic G, Delani P, et al. Contribution of diffusion tensor imaging to delineation of gliomas and glioblastomas. J Magn Reson Imaging 2004;20(6):905–12. [30] van Westen D, Latt J, Englund E, Brockstedt S, Larsson EM. Tumor extension in high-grade gliomas assessed with diffusion magnetic resonance imaging: values and lesion-to-brain ratios of apparent diffusion coefficient and fractional anisotropy. Acta Radiol 2006;47(3):311–9.