| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Original Research |
1 Department of Radiology, University of Michigan, 1500 E Medical Center Dr.,
Ann Arbor, MI 48109-0030.
2 School of Public Health, University of Michigan, Ann Arbor, MI.
3 Department of Neurology, University of Michigan, Ann Arbor, MI.
Received June 14, 2004;
accepted after revision December 20, 2004.
Address correspondence to P. C. Sundgren.
Abstract
 Top Abstract IntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
MATERIALS AND METHODS. Two-dimensional CSI (point-resolved spectroscopy sequence [PRESS]; TR/TE, 1,500/144) was performed in 29 consecutive patients (4-54 years old; mean age, 34 years) who had a new contrast-enhancing lesion in the vicinity of a previously diagnosed and treated brain neoplasm. Clinical and imaging follow-up, and histopathology in 16 patients, were used as indicators of the identity of a lesion.
RESULTS. Diagnostic-quality spectra were obtained in 97% of the patients. The Cho/Cr (choline/creatine) and Cho/NAA (choline/N-acetyl aspartate) ratios were significantly higher, and the NAA/Cr ratios significantly lower, in tumor than in radiation injury (all three differences, p < 0.0001). The Cho/Cr and Cho/NAA ratios were significantly higher in radiation injury than in normal-appearing white matter (p < 0.0003 and p < 0.0001, respectively), whereas NAA/Cr ratios were not different (p = 0.075). Mean Cho/Cr ratios were 2.52 for tumor, 1.57 for radiation injury, and 1.14 for normal-appearing white matter. Mean Cho/NAA ratios were 3.48, 1.31, 0.79, and mean NAA/Cr ratios were 0.79, 1.22, and 1.38, respectively. When values greater than 1.8 for either Cho/Cr or Cho/NAA ratios were considered evidence of tumor, 27 of 28 patients could be correctly classified.
CONCLUSION. Two-dimensional CSI MR spectroscopy can differentiate tumor from radiation injury in patients with recurrent contrast-enhancing intracranial lesions. In these lesions, the Cho/NAA and Cho/Cr ratios may be the best numeric discriminators.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Single-voxel MR spectroscopy used in earlier investigations [7-10] resulted in interpretative difficulties, with overlapping metabolic ratios as a result of partial volume contamination in these histologically heterogeneous lesions. Multivoxel spectroscopic imaging—either 2D or 3D CSI—enables, during the same acquisition, coverage of a larger volume and investigation of multiple regions of the lesion and surrounding tissue. One could assume that the smaller the sampling unit, the more specific MR spectroscopy will be in these heterogeneous lesions. However, lesions in the posterior fossa and those supratentorial lesions that lie in close proximity to the ventricular system or skull are difficult to evaluate with 2D CSI because of bulk magnetic susceptibility variations. This problem can be largely overcome by using both outer volume suppression slices and in-field-of-view saturation bands for the suppression of osseous regions adjacent to the lesion of interest [11].
To our knowledge, neither the feasibility nor the utility of 2D CSI MR spectroscopy in the discrimination of recurrent CNS neoplasms from radiation injury has been shown in a statistically relevant patient population. The aim of this investigation was to explore the feasibility of performing 2D CSI MR spectroscopy and to identify its utility in the evaluation of these complex lesions in 29 consecutive patients who presented with new areas of contrast enhancement at the site of a previously treated neoplasm, irrespective of the lesion's histopathologic nature.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Each of these patients had a new contrast-enhancing lesion visualized on their scheduled follow-up MRI. A determination of recurrent tumor versus radiation injury was deemed difficult from the imaging characteristics of each lesion on conventional MRI. Therefore, MR spectroscopy was performed on these 29 patients at the request of the referring clinician in an attempt to distinguish recurrent tumor from radiation injury. Exemption by our institutional review board was obtained for this retrospective study.
In all but one patient, the contrast-enhancing lesions occurred at the site of the original treated tumor, while in one patient the contrast enhancement occurred in the white matter of the contralateral hemisphere. Six of the patients had a recurrent contrast-enhancing lesion at the site of the original lesion in the posterior fossa. Two patients had contrast-enhancing lesions in the brainstem, and in four patients the lesion was in the cerebellum. The remaining patients had recurrent lesions at the site of their original supratentorial intraaxial neoplasm. The mean interval between completion of radiation therapy and the development of a new contrast-enhancing lesion on scheduled follow-up MRI was 23.4 months (range, 2-108 months; median, 16.5 months) in the 16 patients who were classified as having recurrent tumor, and 13.9 months (range, 3-36 months; median, 13 months) in the 12 with lesions classified as radiation injury.
Clinical, neuroradiologic, and neuropathologic follow-up after MR spectroscopy was used to establish the identity of the lesion. Lesions were regarded as tumor progression if they met either of the following criteria: later verification of active tumor at this site by biopsy (n = 2), surgical resection (n = 5), or autopsy (n = 3); or continued progression on subsequent MRI in a manner consistent with tumor growth (n = 6). Lesions classified as radiation injury met either of the following criteria: later histopathologic evidence of radiation injury without tumor by biopsy (n = 4) or surgical resection (n = 2); or MRI follow-up showing prolonged stability or spontaneous regression unaccompanied by clinical worsening (n = 7). The follow-up time of the patients after the initial MR spectroscopy, which was performed after initial identification of the recurrent contrast-enhancing lesion, was a mean of 20.5 months (range, 9-34 months; median, 18 months) in patients classified as having tumor, and 21.5 months (range, 14-35 months; median, 19 months) in patients whose lesions were classified as radiation injury.
All examinations were performed on a 1.5-T scanner. Conventional MRI was performed in conjunction with all MR spectroscopy examinations and included unenhanced and gadolinium-enhanced T1-weighted images in the axial and sagittal projections, axial T2-weighted sequences with fat saturation, axial FLAIR and diffusion-weighted images, and gadolinium-enhanced T1-weighted images in the coronal projection. MR spectroscopy was always performed as the last sequence in the study so that voxel placement could be made over the region of interest, which was an area of contrast enhancement. Therefore, MR spectroscopy was necessarily always performed after the administration of gadolinium ([gadopentetate dimeglumine] Magnevist, Berlex).
Successful 2D CSI was performed in 28 of 29 patients. The following
parameters were used for all 2D CSI examinations: a point-resolved
spectroscopy sequence (PRESS); TR/TE, 1,500/144; field of view, 16 cm; matrix,
16 x 16; slice thickness, 10-20 mm; acquisition, 1 average; scanning
time, 4 min 20 sec. The volume of interest (VOI) was placed on either
nonangled contrast-enhanced axial T1-weighted or FLAIR images to ensure that
voxels were placed with certainty both over the contrast-enhancing area and
over white matter that appeared normal on T2 and FLAIR images. The volume of
the VOI varied depending on where in the brain the lesion was located, with
the smaller VOI in the posterior fossa and the larger in the supratentorial
compartment. In all patients, the VOI was placed to include both the lesion
and adjacent normal-appearing brain. Commonly, at least two MR spectroscopy
sequences were performed with the VOI placed to include the lesion and
adjacent normal-appearing white matter (defined as not displaying signal
alterations on T2-weighted imaging) at two different levels. In addition to
out-of-field-of-view saturation bands that are routinely placed in all MR
spectroscopy examinations, within-field-of-view saturation bands for
suppression of osseous structures and CSF-containing structures adjacent to
the tissue of interest were placed when necessary to obtain suitable 2D CSI
spectra. Automatic prescanning was performed twice before each spectroscopic
scan to ensure adequate water suppression. The full-width half-maximum (FWHM)
was kept under 10 (
16 is suggested by the manufacturer) with a flip angle
of approximately 125° and water saturation between 98-99% (
96% is
suggested by the manufacturer), allowing separation of the choline peak from
the creatine peak.
In all cases, the data were transferred to a work-station (Sun, GE Healthcare) for offline postprocessing using Functool 2000 software (GE Healthcare). Within the obtained VOI, separate 1 x 1 x 1 cm voxels (small regions of interest) were individually placed in the lesion; in the adjacent normal-appearing white matter; and, in 26 of 28 patients, in a corresponding site in the contralateral hemisphere. The signal intensity of various metabolite peaks was evaluated in every voxel, using integrals of each peak as a measure of its intensity. The integration limits of the respective peaks were manually defined by the same neuroradiologist before computerized calculations. For calculation of metabolite ratios, the integrals of metabolites in the same voxel were used. The spectra were analyzed for the signal intensity of NAA, choline, and creatine and for the presence of lactates and lipids. Ratios were manually calculated for NAA/Cr, Cho/Cr, and Cho/NAA. Metabolite ratios among the recurrent tumor population and radiation injury population were compared. The highest Cho/Cr, Cho/NAA, and NAA/Cr ratios in one voxel were used for comparison.
Two locations in the brain of 26 of 28 patients were measured, one in the contrast-enhancing lesion and one in the normal-appearing white matter. Measurements from these two locations in a subject are likely to be correlated. The generalized estimating equation technique was used to fit a linear regression model for each of the three ratios to take into account the within-subject correlation [12]. Robust SEs were used to compute Wald test p values. The level for statistical significance was set to a p value of less than 0.05.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
On the basis of the clinical and imaging follow-up data or histopathology results from biopsy, resection, or autopsy, the contrast-enhancing lesions of 16 patients were categorized as tumor recurrence and the lesions of 12 patients were categorized as radiation injury.
The following statistical evaluation was obtained using the generalized estimating equation technique. The lesions in the recurrent or residual tumor group had significantly higher Cho/Cr ratios than those in the radiation injury group (p < 0.0001; SE, 0.18), and these, in turn, had significantly higher Cho/Cr ratios than the normal-appearing white matter of 26 of the patients (p = 0.0003; SE, 0.09). The lesions in the tumor group also had significantly higher Cho/NAA ratios than those in the radiation injury group (p < 0.0001; SE, 0.32), and lesions in this group again had significantly higher Cho/NAA ratios than the normal-appearing white matter of 26 patients (p < 0.0001; SE, 0.08). Lesions in the tumor group had significantly lower NAA/Cr ratios (p < 0.0001; SE, 0.09) than those in the radiation injury group, which had insignificantly lower NAA/Cr ratios than the normal-appearing white matter of 26 patients (p = 0.075; SE, 0.09). The mean values (and ranges) of the Cho/Cr, Cho/NAA, and NAA/Cr ratios in respective lesions and normal-appearing white matter are summarized in Table 1.
|
Cases with ratio values in the borderline range between the two types of lesions are given in Table 2. None of the lesions classified as radiation injury reached the value of 1.8 in either Cho/Cr or Cho/NAA ratios. Two lesions classified as radiation injury had Cho/NAA ratios of 1.71 and 1.78, respectively. Both these patients had scattered lesions in the posterior fossa, with interval regression and no recurrent tumor identified at brain biopsy. Two patients with tumor who had Cho/Cr ratios lower than 1.8 had Cho/NAA ratios higher than 1.8 (3.07 and 2.00). One patient with tumor who had a Cho/NAA ratio less than 1.8 had a Cho/Cr ratio greater than 1.8 (1.96). Only one patient with tumor had both Cho/Cr and Cho/NAA ratios lower than 1.8. In this patient, only parts of the lesion, which was located adjacent to mastoid air cells and skull base, could be evaluated with 2D CSI, and within that area no spectra consistent with tumor recurrence were evident. The biopsy specimen that showed tumor was likely taken from another, more inferior portion of the lesion that was unsuitable for MR spectroscopy because of susceptibility artifacts from adjacent bone and air.
|
When values greater than 1.8 for either Cho/Cr or Cho/NAA ratios were considered evidence of tumor, then all but one (27/28) patient with technically satisfactory 2D CSI could be correctly categorized as radiation injury or recurrent tumor, within the limits of reliability of the diagnosis of radiation injury. In the remaining patient with recurrent tumor that would have been incorrectly categorized using the 1.8 threshold, incomplete MR spectroscopy sampling of parts of the enhancing lesion may have been a factor.
In 6 (38%) of 16 patients with recurrent or residual tumor, pathologic spectra consistent with the presence of tumor (i.e., markedly elevated choline and depressed NAA) were identified both in voxels placed inside and in voxels outside the contrast-enhancing lesion. In these six patients, the highest metabolic ratios were found in the contrast-enhancing lesion.
No significant differences were seen in Cho/Cr, Cho/NAA, or NAA/Cr ratios between males and females, regardless of diagnosis. The Cho/NAA ratios were significantly higher (p = 0.045) in the supratentorial lesions than in the posterior fossa lesions, and the NAA/Cr ratios were significantly lower (p = 0.0023) in supratentorial lesions than in the posterior fossa lesions. No differences were seen in the Cho/Cr ratios between those obtained in the posterior fossa and those in the supratentorial lesions. In no case were the signal intensities of lactates or lipids alone useful in differentiation of tumor from radiation injury. Typical examples of metabolic spectra obtained in areas consistent with radiation injury and in areas of recurrent tumor are given in Figures 1A, 1B, 2A, 2B, and 2C.
|
|
|
|
|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Statistically significant differences in Cho/Cr and Cho/NAA ratios were found between recurrent tumor, radiation injury, and normal-appearing brain. Using both ratios in a retrospective classification (cutoff value, 1.8 for either ratio), all but one patient (97%) would have received a correct diagnosis from the results of 2D CSI MR spectroscopy. In a similar population examined with single-voxel MR spectroscopy, similar significant differences were found, and using Cho/Cr and Cho/NAA ratios allowed a correct retrospective classification in more than 80% of the cases [8]. Cho/Cr and NAA/Cr ratios in the recurrent tumor and radiation injury populations in our study are in close agreement with those obtained using single-voxel MR spectroscopy in similar populations [8, 13], although less overlap between these two populations exists in our study using 2D CSI.
In this specific group of patients, the magnitude of the range of
metabolite ratios likely reflects primarily the heterogeneity of these lesions
with volume averaging of normal and abnormal tissues (tumor or radiation
injury or combination of both), although the type and grade of the possible
tumor or the stage of radiation may also contribute. A prior study
[14] of multivoxel MR
spectroscopy found that "spectral patterns do allow reliable
differential diagnostic statements to be made when the tissues are composed of
either pure tumor or pure necrosis, but the spectral patterns are less
definitive when tissues composed of varying degrees of mixed tumor and
necrosis are examined." In that study, a Cho/Cr ratio of greater than
1.79 had a sevenfold increased likelihood of being pure tumor
[14]. Interestingly, none of
our patients with lesions classified as radiation injury had Cho/Cr ratios
greater than 1.76, and only three (19%) of our patients with recurrent tumor
had ratios slightly less than 1.76 (Table
2). Thus, indications are that a threshold value of Cho/Cr or
Cho/NAA ratios (
1.8 in our study) could be used to indicate the presence
of tumor. This possibility testifies to the importance of minimizing volume
averaging in MR spectroscopy analyses by using 2D or 3D techniques in future
studies.
An increase in Cho/Cr ratios and a reduction in NAA/Cr ratios after radiation therapy, proportional to radiation dose, have been reported from studies using single-voxel MR spectroscopy [8, 15, 16]. In our study, the Cho/Cr ratios of lesions related to radiation injury were significantly higher than those of normal-appearing white matter, whereas the decrease of NAA/Cr ratios was not significant.
In the normal-appearing white matter of our patients, the Cho/Cr ratios (mean, 1.14; range, 0.86-1.59) are in excellent agreement with those (0.85-1.47) reported previously from 2D CSI [11] or those obtained by single-voxel MR spectroscopy (0.94 ± 0.17) [8]. Also, Cho/NAA ratios (mean, 0.79; range, 0.56-1.20) were similar to those in a prior single-voxel MR spectroscopy study (0.53 ± 0.17) [8]. Although the NAA/Cr ratios for normal-appearing white matter in our study (mean, 1.38; range, 0.64-2.0) are concordant with those from 2D CSI (1.20-2.33) [11] and from single-voxel MR spectroscopy (1.86 ± 0.46) [8], in four of our patients with tumor and in two with radiation injury the ratios were lower than 1. The normal-appearing white matter of our patients had unavoidably been subjected to a varying degree of radiation. Because a decrease in NAA or a reduction in NAA/Cr ratios after radiation therapy is a frequently reported finding [8-10, 15-18], it is possible that effects of radiation account in part for the low NAA values of these six patients.
A limitation of this study is that a histopathologic diagnosis was not available in 13 of 29 patients, and clinical and imaging follow-up data had to be used as surrogate markers of the identity of the lesion. The prolonged follow-up of these lesions after MR spectroscopy should minimize any possibility of misclassification. It is unlikely that a tumor, once it has shown new enhancement, will show prolonged stability or regression, or that the patient will improve clinically; thus, it is highly unlikely that the seven patients with lesions classified as radiation injury on this basis actually had tumor progression. Similarly, although radiation injury can progress, continued progression over a prolonged period is strong evidence of tumor. Among the six patients considered to have a tumor on the basis of clinical and imaging progression with no histopathologic diagnosis, the progression was monitored for 22-36 months in four patients. The remaining two patients, who had continued progression over 9 and 16 months, had Cho/Cr ratios (3 and 4.26) and Cho/NAA ratios (2.75 and 6.19) that were strongly indicative of tumor.
In accordance with other studies [6, 19, 20], abnormal metabolic spectra strongly suggestive of tumor in this study were seen not only in the contrast-enhancing regions but, even more important, outside this area in 38% of patients with recurrent tumor. This ability of multivoxel spectroscopic imaging to detect, in the same acquisition, tumor outside the area of contrast enhancement may not only increase the chance of discovering tumor recurrence but may also change the options for further treatment.
We conclude that 2D CSI MR spectroscopy has the potential to discriminate recurrent tumor from radiation injury in most patients with recurrent contrast-enhancing intracranial lesions, including those in the posterior fossa or adjacent to osseous or CSF-containing structures. Further multivoxel studies are needed to determine cutoff values of metabolite ratios, with attention to Cho/Cr and Cho/NAA ratios, which may be the best numeric discriminators.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
CiteULike 
Complore 
Connotea 
Del.icio.us 
Digg 
Reddit 
Technorati What's this?
This article has been cited by other articles:
|
|
 |
|
  E. L. Gasparetto, M. A. Pawlak, S. H. Patel, J. Huse, J. H. Woo, J. Krejza, M. R. Rosenfeld, D. M. O'Rourke, R. Lustig, E. R. Melhem, et al. Posttreatment Recurrence of Malignant Brain Neoplasm: Accuracy of Relative Cerebral Blood Volume Fraction in Discriminating Low from High Malignant Histologic Volume Fraction Radiology, March 1, 2009; 250(3): 887 - 896. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 E. A. Smith, R. C. Carlos, L. R. Junck, C. I. Tsien, A. Elias, and P. C. Sundgren Developing a Clinical Decision Model: MR Spectroscopy to Differentiate Between Recurrent Tumor and Radiation Change in Patients with New Contrast-Enhancing Lesions Am. J. Roentgenol., February 1, 2009; 192(2): W45 - W52. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 C. Plathow and W. A. Weber Tumor Cell Metabolism Imaging J. Nucl. Med., June 1, 2008; 49(Suppl_2): 43S - 63S. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 M. Petrou, R.E. Harris, B.R. Foerster, S.A. Mclean, A. Sen, D.J. Clauw, and P.C. Sundgren Proton MR Spectroscopy in the Evaluation of Cerebral Metabolism in Patients with Fibromyalgia: Comparison with Healthy Controls and Correlation with Symptom Severity AJNR Am. J. Neuroradiol., May 1, 2008; 29(5): 913 - 918. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 A. A. Brandes, A. Tosoni, F. Spagnolli, G. Frezza, M. Leonardi, F. Calbucci, and E. Franceschi Disease progression or pseudoprogression after concomitant radiochemotherapy treatment: Pitfalls in neurooncology Neuro-oncol, January 1, 2008; 10(3): 361 - 367. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 Y. Cao, P. C. Sundgren, C. I. Tsien, T. T. Chenevert, and L. Junck Physiologic and Metabolic Magnetic Resonance Imaging in Gliomas J. Clin. Oncol., March 10, 2006; 24(8): 1228 - 1235. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
