Skip to main content
NCBI home page
Radiology logoLink to Radiology
. 2009 Nov;253(2):486–496. doi: 10.1148/radiol.2532090007

Differentiation of Recurrent Glioblastoma Multiforme from Radiation Necrosis after External Beam Radiation Therapy with Dynamic Susceptibility-weighted Contrast-enhanced Perfusion MR Imaging1

Ramon F Barajas Jr, Jamie S Chang, Mark R Segal, Andrew T Parsa, Michael W McDermott, Mitchel S Berger, Soonmee Cha ✉
PMCID: PMC2770116  PMID: 19789240

In this study, we showed that quantitative measurement of hemodynamic values derived from T2*-weighted dynamic susceptibility-weighted contrast material–enhanced MR imaging results can be used to distinguish recurrent glioblastoma multiforme from external beam radiation therapy–induced necrosis.

Abstract

Purpose:

To investigate whether cerebral blood volume (CBV), peak height (PH), and percentage of signal intensity recovery (PSR) measurements derived from the results of T2*-weighted dynamic susceptibility-weighted contrast material–enhanced (DSC) magnetic resonance (MR) imaging performed after external beam radiation therapy (EBRT) can be used to distinguish recurrent glioblastoma multiforme (GBM) from radiation necrosis.

Materials and Methods:

Fifty-seven patients were enrolled in this HIPAA-compliant institutional review board–approved retrospective study after they received a diagnosis of GBM, underwent EBRT, and were examined with DSC MR imaging, which revealed progressive contrast enhancement within the radiation field. A definitive diagnosis was established at subsequent surgical resection or clinicoradiologic follow-up. Regions of interest were retrospectively drawn around the entire contrast-enhanced region. This created T2*-weighted signal intensity–time curves that produced three cerebral hemodynamic MR imaging measurements: CBV, PH, and PSR. Welch t tests were used to compare measurements between groups.

Results:

Mean, maximum, and minimum relative PH and relative CBV were significantly higher (P < .01) in patients with recurrent GBM than in patients with radiation necrosis. Mean, maximum, and minimum relative PSR values were significantly lower (P < .05) in patients with recurrent GBM than in patients with radiation necrosis.

Conclusion:

These findings suggest that DSC perfusion MR imaging may be used to differentiate recurrent GBM from EBRT-induced radiation necrosis.

© RSNA, 2009

Introduction

External beam radiation therapy (EBRT) has become an important therapeutic adjunct to surgical resection in patients with newly diagnosed glioblastoma multiforme (GBM) (1–3). This technique delivers high doses of radiation that kill tumor cells within a target region. However, this therapy can lead to delayed radiation necrosis that can manifest as progressive contrast enhancement on follow-up magnetic resonance (MR) images (1,4–6). As a result, it is often difficult to determine whether a progressively enhancing lesion after EBRT is due to recurrent GBM or radiation necrosis.

Physiologic imaging techniques, such as T2*-weighted dynamic susceptibility-weighted contrast material–enhanced (DSC) perfusion MR imaging, have substantially advanced the clinical use of MR imaging. DSC perfusion MR imaging enables the following three hemodynamic measurements to be obtained within the brain: cerebral blood volume (CBV), peak height (PH), and percentage of signal intensity recovery (PSR) (7–16). These easy-to-obtain hemodynamic measurements complement the anatomic information obtained with conventional contrast-enhanced MR imaging. In several studies, researchers have used DSC perfusion MR imaging to accurately grade newly diagnosed primary glioma, differentiate between a single brain metastasis and GBM, and distinguish recurrent brain metastases from gamma knife–induced radiation necrosis (9–13).

The fact that DSC perfusion MR imaging can be used to successfully characterize these causes of progressive contrast enhancement may be due to quantifiable differences in tissue vasculature. At histologic analysis, recurrent brain tumors—when compared with radiation necrosis—have significantly higher levels of vasculature expression within contrast-enhanced areas (7–13). Relatively recent comparisons of DSC perfusion MR images with histologic slices of malignant gliomas obtained with stereotactic image-guided biopsy have shown a significant positive correlation between tumor vascularity and CBV measurements (14–16).

The ability to distinguish recurrent GBM from radiation necrosis with a noninvasive imaging technique that can be used to reliably measure regions of elevated tumor vasculature carries obvious clinical implications. Thus, the purpose of our study was to investigate whether CBV, PH, and PSR measurements derived from the results of T2*-weighted DSC perfusion MR imaging performed after EBRT can be used to distinguish recurrent GBM from radiation necrosis.

Materials and Methods

Patient Population

In total, 57 patients (33 men, 24 women; mean age, 54.2 years ± 10.2 [standard deviation]) who underwent EBRT at the University of California, San Francisco, from January 2003 to January 2007 were retrospectively selected for this Health Insurance Portability and Accountability Act–compliant institutional review board–approved study if they met the following criteria: (a) They received a new diagnosis of GBM on the basis of MR findings in the brain; (b) they underwent image-guided gross total tumor resection of the enhancing component, and subsequent histologic findings enabled us to confirm the diagnosis of GBM; (c) they underwent treatment with postsurgical conventional EBRT; (d) they subsequently developed a progressively enlarging region of contrast enhancement within the radiation field, as indicated by two or more consecutive MR examinations with findings that were suggestive of GBM recurrence or radiation necrosis; (e) they underwent T2*-weighted DSC perfusion MR imaging of the progressively enhancing region; and (f) they underwent surgical resection of enhancing tissue or adequate clinicoradiologic follow-up, which enabled us to definitively diagnose GBM recurrence or EBRT-induced radiation necrosis.

All patients included in this study underwent fractionated EBRT with proscribed dosages of either 180 cGy per day for a total dose of 5940 cGy or 200 cGy per day for a total dose of 6000 cGy over approximately 6 weeks. (This is the standard EBRT protocol at our institution.) All patients initially received a standard regimen of temozolomide (Temodar; Schering-Plough, Kenilworth, NJ). When clinically indicated, other adjuvant chemotherapeutic regimens were administered. MR imaging follow-up was performed at 2-month intervals or sooner.

In two of the 17 patients with radiation necrosis, the clinical diagnosis was made when the size of the enhancing region decreased definitively or remained unchanged at serial follow-up MR imaging for at least 22 months. In the remaining 15 patients with radiation necrosis and the 40 patients with recurrent GBM, the diagnosis was made on the basis of histologic findings in contrast-enhanced tissue obtained with image-guided biopsy or surgical resection. Patients with more than 5% viable tumor, as estimated during the histologic examination, were classified as having a recurrence.

Imaging Protocol

All MR images were obtained with a 1.5-T imager (Signa Horizon; GE Medical Systems, Milwaukee, Wis). Conventional anatomic MR images were acquired with the following MR protocols: (a) three-plane localizer imaging (repetition time msec/echo time msec, 8.5/1.6), (b) sagittal T1-weighted spin echo imaging (600/17), (c) axial three-dimensional T2-weighted fast spin-echo imaging (3000/102), (d) axial fluid-attenuated inversion recovery imaging (repetition time msec/echo time msec/inversion time msec, 10 000/148/2200), (e) DSC perfusion gradient-echo echo-planar imaging (1250/54, 35° flip angle, acquisition with the array spatial sensitivity encoding technique), (f) contrast-enhanced three-dimensional spoiled gradient-recalled acquisition in the steady state T1-weighted imaging (34/8), and (g) T1-weighted postcontrast spin-echo imaging (1000/54).

We used the standard DSC perfusion MR imaging protocol used at our institution to acquire a series of gradient-echo echo-planar images immediately before, during, and after bolus injection of gadopentetate dimeglumine (Omniscan; GE Medical Systems). Eight axial 5-mm-thick sections were used to cover the entire tumor volume, as determined with T2-weighted fluid-attenuated inversion recovery and fast spin-echo MR imaging. The first 10 echo-planar acquisitions were performed before injection of gadopentetate dimeglumine to establish a precontrast baseline. At the 10th acquisition, gadopentetate dimeglumine (0.1 mmol per kilogram of body weight) was injected intravenously with an MR-compatible power injector (Spectris Solaris; Medrad, Indianola, Pa) at a rate of 5 mL/sec through a 20-gauge angiocatheter and followed immediately by a 20-mL continuous saline flush. A multisection image set was obtained every 1.25 seconds before, during, and after the first pass of contrast agent until images were obtained at 60 time points.

We did not administer contrast agent prior to DSC perfusion MR imaging. Extravasation of contrast media into the extravascular extracellular space can result in increased T1-weighted signal competition with decreased T2*-weighted susceptibility-induced signal loss, confounding relative CBV (rCBV) estimates within GBM (17–19). To counter this effect, some investigators have advocated the administration of a small dose of contrast agent prior to DSC perfusion MR imaging; however, on the basis of our past experience with optimizing DSC perfusion sequences, we use echo time and flip angle to counter possible T1- and T2*-weighted changes caused by contrast material extravasation (7,8,12,13,16–22). An echo time of 54 msec was chosen for the gradient-echo technique to maximize the effect of susceptibility changes. A flip angle of 35° was selected to minimize the T1 effect during the first pass of contrast agent. Altering echo time or flip angle would most likely alter the signal intensity–time curve characteristic during the first pass of contrast agent. By increasing echo time and flip angle, we would expect the PH to decrease and the PSR to increase, respectively (12).

It has been suggested that use of this echo time and flip angle can result in overestimation of rCBV (17). To counter this possible effect, we used the negative enhancement integral to calculate rCBV. Finally, we chose not to administer contrast agent prior to DSC perfusion MR imaging because previous study findings have indicated that reducing the T1-weighted effects caused by administering contrast agent prior to DSC perfusion MR imaging causes increased T2*-weighted susceptibility effects, leading to a reduction in the postcontrast bolus baseline signal intensity (23–25). If contrast agent were to be administered prior to DSC perfusion MR imaging, we would expect there to be a reduction in PSR estimations.

Image Processing

The postcontrast three-dimensional spoiled gradient-recalled acquisition in the steady state and T2*-weighted DSC perfusion MR images were transferred to a commercially available perfusion image processing workstation (Advantage; GE Medical Systems). Image processing was performed in a blinded manner by one author (R.F.B., more than 3 years experience) using commercially available software (Functool; GE Medical Systems). The T2*-weighted DSC perfusion MR images were used in the production of CBV maps and corresponding T2*-weighted susceptibility signal intensity–time curves on a voxel-by-voxel basis. Regions of signal intensity threshold values less than 100 or greater than 4200 were excluded from T2*-weighted signal intensity–time curve analysis.

Postcontrast three-dimensional spoiled gradient-recalled acquisition in the steady state MR images were aligned with the T2*-weighted DSC perfusion MR images and CBV maps by using the aforementioned image processing software. For each transaxial section, the aligned imaging data set was used to manually draw a single region of interest (ROI) around the entire contrast-enhanced region. In addition, one 50-mm2 ROI was manually drawn around the enhancing region with the highest CBV for each transaxial section. A 50-mm2 ROI manually drawn around the contralateral normal-appearing white matter for each transaxial section was used to standardize the three cerebral hemodynamic measurements. All ROIs were approved by a blinded neuroradiologist (S.C., more than 20 years experience) certified by the American Board of Radiology with a certificate-added qualification in neuroradiology.

T2*-weighted Signal Intensity–Time Curve Analysis

The T2*-weighted signal intensity–time curves generated from each ROI for all 24 transaxial sections were used in the quantification of CBV, PH, and PSR in a blinded manner by one author (R.F.B., more than 3 years experience) (Fig 1). CBV was obtained by integrating the negative enhancement portion of the T2*-weighted signal intensity–time curve. rCBV was calculated as the ratio of the CBV within the contrast-enhanced region to the CBV within the contralateral normal-appearing white matter. The calculation of PH and PSR imaging values is described in the legend of Figure 1 (13). PH and PSR values were then normalized to the contralateral normal-appearing white matter (NAWM) values through a proportion of ratios, resulting in relative PH (rPH) and relative PSR (rPSR). rPH was calculated with the following equation: rPH = [S0(ROI) − Smin(ROI)]/[S0(NAWM) − Smin(NAWM)], where S0(ROI) is the precontrast T2*-weighted signal intensity of the ROI, Smin(ROI) is the minimum T2*-weighted signal intensity of the ROI, S0(NAWM) is the precontrast T2*-weighted signal intensity of the NAWM, and Smin(NAWM) is the minimum T2*-weighted signal intensity of the NAWM. rPSR was calculated with the following equation: rPSR = {[S1(ROI) − Smin(ROI)]/[S0(ROI) − Smin(ROI)]}/{[S1(NAWM) − Smin(NAWM)]/[S0(NAWM) − Smin(NAWM)]}, where S1(ROI) is postcontrast T2*-weighted signal intensity of the ROI and S1(NAWM) is postcontrast T2*-weighted signal intensity of the NAWM (13). Histogram-based signal intensity normalization, model curve fitting, and threshold fitting with data acquired at baseline, as first proposed in the calculation of rPH and rPSR by Lupo et al (7) and Cha et al (12), was not performed in the present study. Mean and minimum rCBV, rPH, and rPSR were calculated from each ROI drawn around the entire contrast-enhanced region for each transaxial section. Maximum rCBV, rPH, and rPSR was calculated from each 50-mm2 ROI drawn around the contrast-enhanced region with the highest CBV for each transaxial section. The average of the mean, minimum, and maximum rCBV, rPH, and rPSR for all ROIs was then calculated for each patient (Fig 2).

Figure 1:

Figure 1:

Open in a new tab

Left: Contrast-enhanced T1-weighted axial MR image with one ROI (outlined in pink) surrounding a contrast-enhanced region that was surgically diagnosed as recurrent GBM. Right: Representative T2*-weighted signal intensity–time curve. PH was calculated with the following equation: PH = S0 − Smin, where S0 is precontrast T2*-weighted signal intensity and Smin is minimum T2*-weighted signal intensity. PSR was calculated with the following equation: PSR = (S1 − Smin)/(S0 − Smin), where S1 is postcontrast T2*- weighted signal intensity.

Figure 2a:

Figure 2a:

Open in a new tab

(a, b) Left: Transverse contrast-enhanced T1-weighted MR images. Right: T2*-weighted signal intensity–time curves. Images were obtained in patients with histologically diagnosed (a) recurrent GBM and (b) radiation necrosis and show a significant difference between cerebral hemodynamic imaging values. Areas outlined in pink are the ROIs in which the DSC MR imaging hemodynamic values were obtained. (a) Images obtained in patient 22, a 49-year-old woman with a right frontal contrast-enhanced lesion, 8.2 months after GBM resection and EBRT reveal markedly elevated rPH and rCBV associated with 92% rPSR. (b) Images obtained in patient 53, a 67-year-old man with a left temporoparietal contrast-enhanced lesion, 13.2 months after EBRT show a significant difference in cerebral hemodynamic imaging values and only slightly elevated rPH associated with baseline rCBV and rPSR.

Figure 2b:

Figure 2b:

Open in a new tab

(a, b) Left: Transverse contrast-enhanced T1-weighted MR images. Right: T2*-weighted signal intensity–time curves. Images were obtained in patients with histologically diagnosed (a) recurrent GBM and (b) radiation necrosis and show a significant difference between cerebral hemodynamic imaging values. Areas outlined in pink are the ROIs in which the DSC MR imaging hemodynamic values were obtained. (a) Images obtained in patient 22, a 49-year-old woman with a right frontal contrast-enhanced lesion, 8.2 months after GBM resection and EBRT reveal markedly elevated rPH and rCBV associated with 92% rPSR. (b) Images obtained in patient 53, a 67-year-old man with a left temporoparietal contrast-enhanced lesion, 13.2 months after EBRT show a significant difference in cerebral hemodynamic imaging values and only slightly elevated rPH associated with baseline rCBV and rPSR.

Statistical Analysis

Univariate analyses comparing patient age at diagnosis, prescribed dosage of EBRT, elapsed time between EBRT and DSC MR imaging and final diagnosis, time to progression, time to death, time to clinical follow-up, volume of the contrast-enhanced lesion, and hemodynamic MR imaging measurements (mean, minimum, and maximum rCBV, rPH, and rPSR) between the recurrent GBM and radiation necrosis groups were conducted with a two-sample Welch t test. A binomial test was used to assess equality of sex proportions between the recurrent GBM and radiation necrosis groups. A P value of less than .05 was considered to indicate a significant difference between the two groups. Thresholds for dichotomizing the individual markers were obtained by using classification trees, with attendant predictive performance assessed with the cross-validation technique. Analyses were conducted by one author (M.R.S., more than 20 years experience) using the R statistical package (version 2.6.1; R Foundation, Vienna, Austria) (26,27).

Results

Patient Population

A total of 57 patients with a history of GBM previously treated with EBRT underwent DSC perfusion MR imaging to study 57 progressively enhancing lesions. Seven of the 57 patients were evaluated on two or more occasions because of separate instances in which they were suspected of having lesion progression. This resulted in a total of 66 perfusion MR examinations. Of the 66 MR examinations, 46 were performed in lesions that were surgically determined to be recurrent GBM and 20 were performed in lesions surgically or clinically determined to be radiation necrosis. Patient and lesion characteristics are provided in Table E1 (online).

Twelve patients in whom radiation necrosis was diagnosed at surgery eventually developed newly enhancing satellite lesions that were histologically confirmed to be recurrent GBMs and were separate from the progressively enhancing lesions examined with DSC perfusion MR imaging. Six patients were lost to follow-up 4.5–113.1 months (mean, 31.2 months ± 40.6) after EBRT, eight were alive 10.9–119.0 months (mean, 52.3 months ± 37.2) after EBRT, and 43 patients died 9.7–63.0 months (mean, 29.1 months ± 14.0) after EBRT. The minimum clinical follow-up time for any patient in whom radiation necrosis was diagnosed was 11.6 months (mean, 36.5 months ± 27.7). The minimum clinical follow-up time for any patient in whom recurrent GBM was diagnosed was 4.5 months (mean, 31.1 months ± 21.1). Clinical outcomes of the 57 patients included in this study are summarized in Table E1 (online). There was no significant difference in the distribution of sex (P = .17) or mean patient age (54.1 years ± 10.5 vs 54.7 years ± 9.7, P = .74) between the recurrent GBM and radiation necrosis groups (Table E1 [online]).

Radiation Dose, Volume of Contrast Enhancement, and Time to Final Diagnosis

The EBRT dose prescribed to patients with recurrent GBM (5966 Gy ± 30.1) was not significantly different from that prescribed to patients with radiation necrosis (5960 Gy ± 29.2) (P = .51). The mean volume of contrast enhancement on DSC perfusion MR images was significantly higher in subjects with recurrent GBM (mean, 270 mL ± 162) than in patients with radiation necrosis (mean, 180 mL ± 96; P = .01). There was no significant difference in the time from EBRT to DSC perfusion MR imaging (range, 1.7–50.2 months), final diagnosis (range, 1.8–50.2 months), progression (range, 5.1–62.6 months), or death (range, 9.7–63 months) between the groups (P > .05, Table 1). Patients in whom radiation necrosis was diagnosed tended to experience progression with newly enhancing satellite lesions later than did patients with local recurrence; however, this difference was not significant (P = .07, Table 1). Time to last clinical follow-up was not significantly different between the two groups (P = .48, Table 1).

Table 1.

Patient Population Descriptive Statistics

graphic file with name r09au23t01x.jpg

Open in a new tab

Note.—Data are means ± standard deviations. Time from EBRT to progression represents the time elapsed between initial EBRT and diagnosis of recurrent GBM (if recurrent GBM was diagnosed). Time from EBRT to death represents the time elapsed between initial EBRT and death (if death occurred).

* P = .14.

† P = .16.

‡ P = .07.

§ P = .29.

∥ P = .48.

rPH and rCBV

Table E1 (online) summarizes the hemodynamic measurements obtained in the 66 DSC perfusion MR examinations performed in this study. Mean, minimum, and maximum rPH and rCBV were significantly higher in the recurrent GBM group than in the radiation necrosis group (P < .01, Table 2). Of these measurements, rPH was most reliable in the differentiation of patients with recurrent GBM from those with radiation necrosis. An rPH cutoff value of 1.38 yielded a sensitivity of 89.32% and a specificity of 81.38%. An rCBV cutoff value of 1.75 was not reliable for differentiation between patients with recurrent GBM and those with radiation necrosis (sensitivity, 78.92%; specificity, 71.58%). rPH values greater than 2.17 were observed only in contrast-enhanced regions consistent with recurrent tumor.

Table 2.

DSC Perfusion MR Imaging Measurements

graphic file with name r09au23t02x.jpg

Open in a new tab

Note.—Data are means ± standard deviations. Data in parentheses are 95% confidence intervals. P values were calculated with the Welch t test.

* P < .01.

† P = .01.

‡ P = .05.

§ P = .04.

rPSR

Subjects in the recurrent GBM group had significantly lower mean, minimum, and maximum rPSR values within the entire contrast-enhanced region than did patients in the radiation necrosis group (P < .05, Table 2). An rPSR cutoff value of 87.3% yielded a sensitivity of 78.26% and a specificity of 76.19%. Figures 2 and 3 show cases in which DSC perfusion values varied between patients with recurrent GBM and those with radiation necrosis. The level of significance for T2*-weighted signal intensity–time curve–derived hemodynamic measurements was unchanged when we compared ROIs placed around the entire contrast-enhanced region with ROIs placed in regions with the highest CBV measurements.

Figure 3a:

Figure 3a:

Open in a new tab

(a, b) Left: Transverse contrast-enhanced T1-weighted MR images. Right: T2*-weighted signal intensity–time curves. Images were obtained in patients with histologically proved (a) recurrent GBM and (b) radiation necrosis and show significant differences in rPH, rCBV, and rPSR. Areas outlined in pink are the ROIs in which the DSC MR imaging hemodynamic values were obtained. (a) Images obtained in patient 33, a 41-year-old man with a right parietal contrast-enhanced lesion, 50.2 months after tumor resection and EBRT reveals elevated rPH and rCBV associated with 70% rPSR. (b) Images obtained in patient 50, a 56-year-old woman with a right frontal contrast-enhanced lesion, 16.4 months after EBRT reveals the ROI surrounding the lesion has slightly increased rCBV and rPH associated with 89% rPSR.

Figure 3b:

Figure 3b:

Open in a new tab

(a, b) Left: Transverse contrast-enhanced T1-weighted MR images. Right: T2*-weighted signal intensity–time curves. Images were obtained in patients with histologically proved (a) recurrent GBM and (b) radiation necrosis and show significant differences in rPH, rCBV, and rPSR. Areas outlined in pink are the ROIs in which the DSC MR imaging hemodynamic values were obtained. (a) Images obtained in patient 33, a 41-year-old man with a right parietal contrast-enhanced lesion, 50.2 months after tumor resection and EBRT reveals elevated rPH and rCBV associated with 70% rPSR. (b) Images obtained in patient 50, a 56-year-old woman with a right frontal contrast-enhanced lesion, 16.4 months after EBRT reveals the ROI surrounding the lesion has slightly increased rCBV and rPH associated with 89% rPSR.

Discussion

In this study, we retrospectively used DSC perfusion MR imaging–derived rPH to determine whether a progressively enhancing lesion was caused by recurrent GBM or EBRT-induced necrosis. In addition, we found rCBV to be significantly higher and rPSR to be significantly lower within contrast-enhanced regions of recurrent GBM when compared with these parameters in patients with radiation necrosis.

Radiation necrosis is typically indistinguishable from recurrent GBM at conventional contrast-enhanced MR imaging. Both entities often manifest as a mass lesion with varying degrees of surrounding edema and progressive enhancement on serial MR images (5,28,29). However, at histopathologic examination, one finds that radiation necrosis and recurrent GBM are markedly dissimilar. Radiation necrosis is characterized by extensive fibrinoid necrosis, vascular dilation, and endothelial injury of surrounding normal cerebral vasculature (30). Recurrent GBM is characterized by vascular proliferation manifested by elevated tumor vasculature density (31–33).

Several publications have shown that DSC MR imaging measurements strongly correlate with primary glioma tumor vascular density and overall histopathologic grade (7,16,33–42). PH has been shown to be a quantifiable measure of tumor vasculature and to strongly correlate with rCBV (7,8,13–16). From these findings, it has been suggested that PH may also reflect total vascular volume within an ROI (7,8).

Our findings of elevated rPH and rCBV within contrast-enhanced regions of recurrent GBM are in agreement with the findings of previous studies in which researchers examined the use of DSC perfusion MR imaging to differentiate (a) newly diagnosed GBM from a brain metastasis and (b) recurrent cerebral metastasis from gamma knife–induced radiation necrosis (9,12,13). As indicated by the findings of these studies, the ability of DSC perfusion values to be used to distinguish between these various entities of contrast enhancement is due to inherent differences in their hemodynamic characteristics. Our findings are also supported by the findings of several histologic studies that showed that tumor vasculature was markedly elevated in tissue specimens obtained from contrast-enhanced portions of GBM (14,15,31,32).

The marked difference in the expression of tumor vasculature between the recurrent GBM and radiation necrosis groups is likely responsible for the differences in hemodynamic measurements observed in our study. An rPH cutoff of 1.38 could be used to predict which patients would receive a diagnosis of recurrent GBM, indicating that rPH may prove useful in complementing conventional contrast-enhanced MR imaging in differentiating recurrent GBM from radiation necrosis. The ability to noninvasively distinguish recurrent GBM from radiation necrosis carries obvious therapeutic implications. Currently, tissue sampling with biopsy or surgical resection is the only definitive way to differentiate recurrent GBM from radiation necrosis. Patients with recurrent GBM often benefit from additional surgery; however, invasive surgical resection of radiation necrosis in deep-seated or eloquent cortical locations can lead to further damage to adjacent cerebral parenchyma and result in a worse outcome.

In our study, rCBV was found to be significantly higher within the recurrent GBM group than within the radiation necrosis group; however, a large degree of rCBV overlap observed between the two groups resulted in this imaging value being a less reliable predictor than rPH. CBV overlap in recurrent primary glial brain tumors and radiation necrosis has been noted by Sugahara et al (37). We speculate that the overlap of rCBV in our study may be caused by tumor heterogeneity and previously described inherent shortcomings associated with CBV measurements obtained from vasculature with a disrupted blood-brain barrier (1–16,32,39,40).

Regional tumor heterogeneity results from the coexistence of viable tumor vasculature with areas of radiation-induced dilated vasculature and tissue necrosis, as described by Spiegelmann et al (38). Regional DSC perfusion MR imaging measurements within contrast-enhanced areas of recurrent GBM were noted to be highly heterogeneous, with regions of elevated rCBV interspersed with areas of low rCBV. In our study, histologic analysis of enhancing tissue from patients with recurrent GBM revealed more than 65% tissue necrosis in all but several patients. In addition, rCBV measurements may have been insufficient to differentiate elevated microvascular density occurring within residual tumors from radiation-induced hyperplastic dilated vasculature (31,38). This tumor heterogeneity likely contributed to the rCBV overlap observed in our study (38).

In our study, rPSR was an additional hemodynamic imaging variable that was significantly different between the groups. Lower rPSR values within the recurrent GBM group were likely caused by the presence of a disrupted neoplastic blood-brain barrier that was permeable to macromolecular contrast agents (7,12,32). The degree of capillary leakiness in both groups, while quantifiably different at DSC imaging, remained similar enough that a large degree of rPSR value overlap between the two groups was observed, making this value a less robust predictor of recurrent GBM. We used T2*-weighted echo-planar DSC MR imaging–derived rPSR as an estimate of capillary leakiness in this cohort of patients because rPSR has been shown to be clinically useful in differentiating a single brain metastasis from GBM and distinguishing recurrent brain metastasis from gamma knife–induced radiation necrosis (12,13,33). This method exploits the susceptibility effects of contrast media extravasated into the extravascular-extracellular space, thereby causing a reduction in T2*-weighted signal intensity from the prebolus baseline level during the postbolus phase of the signal intensity–time curve. Other effective methods with which to noninvasively measure capillary leakiness exist (7,12,13,41).

Hazle et al (42) previously described the use of T1-weighted fast spin-echo dynamic contrast-enhanced MR imaging–derived vascular endothelial contrast transfer constant (Ktrans) measurements to differentiate recurrent brain tumors from radiation necrosis. This approach to hemodynamic MR measurements is based on pharmacokinetic modeling, as proposed by Tofts and colleagues (43,44). This method assumes that contrast agent is distributed among a number of separate communicating tissue compartments. Measurement of concentration time curves then permits one to estimate Ktrans, which has been suggested to be a measure of capillary leakiness.

The use of rPSR has several advantages over the use of Ktrans. First, the measurement of Ktrans considers changes in T1-weighted relaxation that occur during the infusion of contrast media or during the washout phase from the enhancing lesion. As a result, signal acquisition is much slower with the T1-weighted sequences than with the T2*-weighted echo-planar imaging sequences. Temporal resolution and region of lesion coverage are limited in the use of Ktrans because of the short acquisition time. Second, researchers have recently suggested that the underlying assumption that T1-weighted signal is linearly proportional to contrast concentration in vivo is invalid (45–48). As a result, without prior calibration, Ktrans measurements are subject to inaccuracy. The results of this and other studies suggest that such calibrations may not be required in the use of PSR as a possible measure of capillary leakiness (13). Finally, the most important limitation in the use of T1-weighted DCE image acquisition methods is that they do not permit one to simultaneously estimate CBV and PH, two variables that are most directly related to elevated microvascular density, which is a histopathologic marker of brain tumor malignancy, with the use of a single dose of contrast media.

Metabolic and spectroscopic imaging modalities are two techniques that have been used to distinguish recurrent GBM from radiation-induced necrosis. Studies in which researchers evaluated the usefulness of fluorine 18 fluorodeoxyglucose positron emission tomography and single-photon emission computed tomography in distinguishing between these two entities have yielded low sensitivity and specificity (49,50). MR spectroscopic imaging has also been studied as a noninvasive method with which to evaluate metabolic changes associated with recurrent GBM and radiation-induced necrosis. Since recurrent GBM and radiation necrosis have different metabolic characteristics, the results of several studies in which MR spectroscopic imaging was used have shown promise in distinguishing between these two groups. In patients with radiation necrosis, MR spectroscopic imaging can reveal a decrease in all major metabolite levels, with a slight increase in the choline-to-creatine ratio compared with that in normal-appearing white matter with or without the presence of lactate. Recurrent GBM has been shown to result in marked increases in the choline-to-creatine ratio and the choline-to-N-acetylasparate ratio when compared with normal-appearing white matter (51–55).

MR spectroscopic imaging undoubtedly provides additional physiologic information that positively contributes to the differentiation of recurrent GBM from radiation-induced necrosis; however, this MR imaging sequence has known technical limitations that can limit its clinical utility. Lesions in close proximity to the ventricular system or skull are often difficult to reliably evaluate because of the magnetic susceptibility artifact induced by these structures. In addition, studies have shown that MR spectroscopic imaging can be used to differentiate between tissues composed of either pure recurrent GBM or pure radiation necrosis. However, currently, MR spectroscopic imaging cannot be used to assess heterogeneous tissue mixtures composed of recurrent GBM and radiation necrosis, despite the fact that this is the tissue composition observed most often in clinical practice (56,57).

Previous studies in which researchers investigated rPH and rPSR used software to perform histogram-based signal intensity normalization and model curve threshold fitting with data acquired at baseline (7,12). In the present study, image processing did not involve these timely and sophisticated analysis techniques (13). We purposely used this simplified method to obtain a clinically meaningful estimate of hemodynamic parameters that could be easily obtained in a timely manner within a clinical setting by using nonproprietary software that could easily be implemented in the clinical interpretation of images.

Our study had several limitations, including its retrospective nature and small patient population. Partial volume averaging within ROIs may have affected the three hemodynamic imaging values used in this study. We attempted to control for this potential volume averaging by obtaining imaging values from a second standardized ROI; however, the results were not significantly different. The use of a smaller standardized ROI may have negated some of the rCBV overlap observed between the two groups; however, this change in method may have yielded a less robust clinical measurement. In addition, our results were specific to the imaging parameters (field strength, echo time, flip angle, not administering contrast agent prior to DSC perfusion MR imaging, and type of contrast agent) and image processing methods (software, CBV calculation with the negative enhancement integral technique, signal threshold values used, and standardization of hemodynamic imaging values to the contralateral normal-appearing white matter) used in this study. We recommend that further prospective evaluation be undertaken with a larger sample size and image-guided histopathologic correlation to confirm the efficacy of the techniques described in this article.

In this study, we showed that quantitative measurement of hemodynamic values derived from T2*-weighted DSC MR imaging results can be used to distinguish recurrent GBM from EBRT-induced necrosis. This additional physiologic information provides insight into how tumor cellular biology affects MR imaging and how, if prospectively validated, it may prove useful in complementing conventional contrast-enhanced MR imaging in the differentiation of recurrent GBM from radiation necrosis.

Advance in Knowledge.

  • The quantitative measurement of cerebral blood volume, peak height, and percentage of signal intensity recovery hemodynamic values derived from T2*-weighted dynamic susceptibility-weighted contrast material–enhanced perfusion MR imaging can be used to distinguish recurrent glioblastoma multiforme (GBM) from external beam radiation therapy (EBRT)-induced necrosis.

Implication for Patient Care.

  • The results of this study provide additional insight into how GBM cellular biology affects MR imaging and may prove useful in complementing conventional contrast-enhanced MR imaging in the differentiation of recurrent GBM from EBRT-induced necrosis.

Supplementary Material

Supplemental Table
supp_253_2_486__index.html (1.3KB, html)

Acknowledgments

The first author thanks Bethany J. Barajas, RN, MSN, CNL, for her helpful comments regarding this article.

Received January 2, 2009; revision requested March 4; revision received April 30; accepted May 12; final version accepted May 14.

Funding: This research was supported by the National Institutes of Health (grants NS045013 and TL1 RR024129-01).

Authors stated no financial relationship to disclose.

Abbreviations:

CBV
cerebral blood volume
DSC
dynamic susceptibility-weighted contrast material–enhanced
EBRT
external beam radiation therapy
GBM
glioblastoma multiforme
Ktrans
transfer constant
PH
peak height
PSR
percentage of signal intensity recovery
rCBV
relative CBV
ROI
region of interest
rPH
relative PH
rPSR
relative PSR

References

  • 1.Kumar AJ, Leeds NE, Fuller GN, et al. Malignant gliomas: MR imaging spectrum of radiation therapy–and chemotherapy-induced necrosis of the brain after treatment. Radiology 2000;217:377–384 [DOI] [PubMed] [Google Scholar]
  • 2.Chang JE, Khuntia D, Robins HI, Mehta MP. Radiotherapy and radiosensitizers in the treatment of glioblastoma multiforme. Clin Adv Hematol Oncol 2007;5:894–915 [PubMed] [Google Scholar]
  • 3.Fiveash JB, Spencer SA. Role of radiation therapy and radiosurgery in glioblastoma multiforme. Cancer J 2003;9:222–229 [DOI] [PubMed] [Google Scholar]
  • 4.Curnes JT, Laster DW, Ball MR, Moody DM, Witcofski RL. MRI of radiation injury to the brain. AJR Am J Roentgenol 1986;147:119–124 [DOI] [PubMed] [Google Scholar]
  • 5.Dooms GC, Hecht S, Brant-Zawadzki M, Berthiaume Y, Norman D, Newton TH. Brain radiation lesions: MR imaging. Radiology 1986;158:149–155 [DOI] [PubMed] [Google Scholar]
  • 6.Hohwieler ML, Lo TC, Silverman ML, Freidberg SR. Brain necrosis after radiotherapy for primary intracerebral tumor. Neurosurgery 1986;18:67–74 [DOI] [PubMed] [Google Scholar]
  • 7.Lupo JM, Cha S, Chang SM, Nelson SJ. Dynamic susceptibility-weighted perfusion imaging of high-grade gliomas: characterization of spatial heterogeneity. AJNR Am J Neuroradiol 2005;26:1446–1454 [PMC free article] [PubMed] [Google Scholar]
  • 8.Cha S, Lu S, Johnson G, Knopp EA. Dynamic susceptibility contrast MR imaging: correlation of signal intensity changes with cerebral blood volume measurements. J Magn Reson Imaging 2000;11:114–119 [DOI] [PubMed] [Google Scholar]
  • 9.Hu LS, Baxter LC, Smith KA, et al. Relative cerebral blood volume values to differentiate high-grade glioma recurrence from posttreatment radiation effect: direct correlation between image-guided tissue histopathology and localized dynamic susceptibility-weighted contrast-enhanced perfusion MR imaging measurements. AJNR Am J Neuroradiol 2009;30:552–558 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 10.Aronen HJ, Gazit IE, Louis DN, et al. Cerebral blood volume maps of gliomas: comparison with tumor grade and histologic findings. Radiology 1994;191:41–51 [DOI] [PubMed] [Google Scholar]
  • 11.Knopp EA, Cha S, Johnson G, et al. Glial neoplasms: dynamic contrast-enhanced T2*-weighted MR imaging. Radiology 1999;211:791–798 [DOI] [PubMed] [Google Scholar]
  • 12.Cha S, Lupo JM, Chen MH, et al. Differentiation of glioblastoma multiforme and single brain metastasis by peak height and percentage of signal intensity recovery derived from dynamic susceptibility-weighted contrast-enhanced perfusion MR imaging. AJNR Am J Neuroradiol 2007;28:1078–1084 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 13.Barajas RF, Chang JS, Sneed PK, Segal MR, McDermott MW, Cha S. Distinguishing recurrent intra-axial metastatic tumor from radiation necrosis following gamma knife radiosurgery using dynamic susceptibility-weighted contrast-enhanced perfusion MR imaging. AJNR Am J Neuroradiol 2009;30:367–372 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 14.Maia AC, Jr, Malheiros SM, da Rocha AJ, et al. MR cerebral blood volume maps correlated with vascular endothelial growth factor expression and tumor grade in nonenhancing gliomas. AJNR Am J Neuroradiol 2005;26:777–783 [PMC free article] [PubMed] [Google Scholar]
  • 15.Sadeghi N, D'Haene N, Decaestecker C, et al. Apparent diffusion coefficient and cerebral blood volume in brain gliomas: relation to tumor cell density and tumor microvessel density based on stereotactic biopsies. AJNR Am J Neuroradiol 2008;29:476–482 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 16.Barajas R, Chang J, Hodgson G, et al. Characterization of GBM cellular and molecular biology with DSC perfusion MR imaging [abstr]. In: American Society of Neuroradiology annual meeting program Oak Brook, Ill: American Society of Neuroradiology, 2009 [Google Scholar]
  • 17.Paulson ES, Schmainda KM. Comparison of dynamic susceptibility-weighted contrast-enhanced MR methods: recommendations for measuring relative cerebral blood volume in brain tumors. Radiology 2008;249:601–613 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 18.Rosen BR, Belliveau JW, Vevea JM, Brady TJ. Perfusion imaging with NMR contrast agents. Magn Reson Med 1990;14:249–265 [DOI] [PubMed] [Google Scholar]
  • 19.Belliveau JW, Rosen BR, Kantor HL, et al. Functional cerebral imaging by susceptibility-contrast NMR. Magn Reson Med 1990;14:538–546 [DOI] [PubMed] [Google Scholar]
  • 20.Maeda M, Itoh S, Kimura H, et al. Tumor vascularity in the brain: evaluation with dynamic susceptibility-contrast MR imaging. Radiology 1993;189:233–238 [DOI] [PubMed] [Google Scholar]
  • 21.Cha S. Perfusion imaging of brain tumors. Top Magn Reson Imaging 2004;15:279–289 [DOI] [PubMed] [Google Scholar]
  • 22.Bruening R, Kwong KK, Vevea MJ, et al. Echo-planar MR determination of relative cerebral blood volume in human brain tumors: T1 versus T2 weighting. AJNR Am J Neuroradiol 1996;17:831–840 [PMC free article] [PubMed] [Google Scholar]
  • 23.Levin JM, Wald LL, Kaufman MJ, Ross MH, Maas LC, Renshaw PF. T1 effects in sequential dynamic susceptibility contrast experiments. J Magn Reson 1998;130:292–295 [DOI] [PubMed] [Google Scholar]
  • 24.Runge VM, Kirsch JE, Wells JW, Dunworth JN, Hilaire L, Woolfolk CE. Repeat cerebral blood volume assessment with first-pass MR imaging. J Magn Reson Imaging 1994;4:457–461 [DOI] [PubMed] [Google Scholar]
  • 25.Levin JM, Wald LL, Kaufman MJ, Ross MH, Maas LC, Renshaw PF. Sequential dynamic susceptibility contrast MR experiments in human brain: residual contrast agent effect, steady state, and hemodynamic perturbation. Magn Reson Med 1995;34:655–663 [DOI] [PubMed] [Google Scholar]
  • 26.Breiman L, Friedman JH, Olshen RA, Stone CJ. Classification and regression trees Belmont, Calif: Wadsworth, 1984; [Google Scholar]
  • 27.R Development Core Team R: a language and environment for statistical computing. Vienna, Austria: R Foundation for Statistical Computing, 2006 [Google Scholar]
  • 28.Babu R, Huang PP, Epstein F, Budzilovich GN. Late radiation necrosis of the brain: case report. J Neurooncol 1993;17:37–42 [DOI] [PubMed] [Google Scholar]
  • 29.Morris JG, Grattan-Smith P, Panegyres PK, O'Neill P, Soo YS, Langlands AO. Delayed cerebral radiation necrosis. Q J Med 1994; 119–129 [PubMed] [Google Scholar]
  • 30.Hopewell JW, Calvo W, Jaenke R, Reinhold HS, Robbins ME, Whitehouse EM. Microvasculature and radiation damage. Recent Results Cancer Res 1993;130:1–16 [DOI] [PubMed] [Google Scholar]
  • 31.Wesseling P, Ruiter DJ, Burger PC. Angiogenesis in brain tumors: pathobiological and clinical aspects. J Neurooncol 1997;32:253–265 [DOI] [PubMed] [Google Scholar]
  • 32.Oh BC, Pagnini PG, Wang MY, et al. Stereotactic radiosurgery: adjacent tissue injury and response after high-dose single fraction radiation. I. Histology, imaging, and molecular events.. Neurosurgery 2007;60:31–45 [DOI] [PubMed] [Google Scholar]
  • 33.Sugahara T, Korogi Y, Kochi M, et al. Correlation of MR imaging-determined cerebral blood volume maps with histologic and angiographic determination of vascularity of gliomas. AJR Am J Roentgenol 1998;171:1479–1486 [DOI] [PubMed] [Google Scholar]
  • 34.Earnest F, 4th, Kelly PJ, Scheithauer BW, et al. Cerebral astroycytomas: histopahtologic correlation of MR and CT contrast enhancement with stereotactic biopsy. Radiology 1988;166:823–827 [DOI] [PubMed] [Google Scholar]
  • 35.Le Bihan D, Douek P, Argyropoulou M, Turner R, Patronas N, Fulham M. Diffusion and perfusion magnetic resonance imaging in brain tumors. Top Magn Reson Imaging 1993;5:25–31 [PubMed] [Google Scholar]
  • 36.Cha S, Knopp EA, Johnson G, et al. Dynamic contrast-enhanced T2-weighted MR imaging of recurrent malignant gliomas treated with thalidomide and carboplatin. AJNR Am J Neuroradiol 2000;21:881–890 [PMC free article] [PubMed] [Google Scholar]
  • 37.Sugahara T, Korogi Y, Tomiguchi S, et al. Posttherapeutic intraaxial brain tumor: the value of perfusion-sensitive contrast-enhanced MR imaging for differentiating tumor recurrence from nonneoplastic contrast-enhancing tissue. AJNR Am J Neuroradiol 2000;21:901–909 [PMC free article] [PubMed] [Google Scholar]
  • 38.Spiegelmann R, Friedman WA, Bova FJ, Theele DP, Mickle JP. LINAC radiosurgery: an animal model. J Neurosurg 1993;78:638–644 [DOI] [PubMed] [Google Scholar]
  • 39.Heiland S, Benner T, Debus J, Rempp K, Reith W, Sartor K. Simultaneous assessment of cerebral hemodynamics and contrast agent uptake in lesions with disrupted blood-brain-barrier. Magn Reson Imaging 1999;17:21–27 [DOI] [PubMed] [Google Scholar]
  • 40.Boxerman JL, Schmainda KM, Weisskoff RM. Relative cerebral blood volume maps corrected for contrast agent extravasation significantly correlate with glioma tumor grade, whereas uncorrected maps do not. AJNR Am J Neuroradiol 2006;27:859–867 [PMC free article] [PubMed] [Google Scholar]
  • 41.Uematsu H, Maeda M, Sadato N, et al. Measurement of the vascularity and vascular leakage of gliomas by double-echo dynamic magnetic resonance imaging: a preliminary study. Invest Radiol 2002;37:571–576 [DOI] [PubMed] [Google Scholar]
  • 42.Hazle JD, Jackson EF, Schomer DF, Leeds NE. Dynamic imaging of intracranial lesions using fast spin-echo imaging: differentiation of brain tumors and treatment effects. J Magn Reson Imaging 1997;7:1084–1093 [DOI] [PubMed] [Google Scholar]
  • 43.Tofts PS, Kermode AG. Measurement of the blood-brain barrier permeability and leakage space using dynamic MR imaging. Magn Reson Med 1991;17:357–367 [DOI] [PubMed] [Google Scholar]
  • 44.Tofts PS, Brix G, Buckley DL, et al. Estimating kinetic parameters from dynamic contrast-enhanced T1-weighted MRI of diffusable tracer: standardized quantities and symbols. J Magn Reson Imaging 1999;10:223–232 [DOI] [PubMed] [Google Scholar]
  • 45.Buckley DL. Uncertainty in the analysis of tracer kinetics using dynamic contrast-enhanced T1-weighted MRI. Magn Reson Med 2002;47:601–606 [DOI] [PubMed] [Google Scholar]
  • 46.Choyke PL, Dwyer AJ, Knopp MV. Functional tumor imaging with dynamic contrast-enhanced magnetic resonance imaging. J Magn Reson Imaging 2003;17:509–520 [DOI] [PubMed] [Google Scholar]
  • 47.Evelhoch JL. Key factors in the acquisition of contrast kinetic data for oncology. J Magn Reson Imaging 1999;10:254–259 [DOI] [PubMed] [Google Scholar]
  • 48.Taylor JS, Tofts PS, Port R, et al. MR imaging of tumor microcirculation: promise for the new millennium. J Magn Reson Imaging 1999;10:903–907 [DOI] [PubMed] [Google Scholar]
  • 49.Di Chiro G, Oldfield E, Wright DC, et al. Cerebral necrosis after radiotherapy and/or intraarterial chemotherapy for brain tumors: PET and neuropathologic studies. AJR Am J Roentgenol 1988;150:189–197 [DOI] [PubMed] [Google Scholar]
  • 50.Olivero WC, Dulebohn SC, Lister JR. The use of PET in evaluating patients with primary brain tumours: is it useful? J Neurol Neurosurg Psychiatry 1995;58:250–252 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 51.Chuang CF, Chan AA, Larson D, et al. Potential value of MR spectroscopic imaging for the radiosurgical management of patients with recurrent high-grade gliomas. Technol Cancer Res Treat 2007;6:375–382 [DOI] [PubMed] [Google Scholar]
  • 52.Wald LL, Nelson SJ, Day MR, et al. Serial proton magnetic resonance spectroscopy imaging of glioblastoma multiforme after brachytherapy. J Neurosurg 1997;87:525–534 [DOI] [PubMed] [Google Scholar]
  • 53.Weybright P, Sundgren P, Maly P, et al. Differentiation between brain tumor recurrence and radiation injury using MR spectroscopy. AJR Am J Roentgenol 2005;185:1471–1476 [DOI] [PubMed] [Google Scholar]
  • 54.Schlemmer HP, Bachert P, Herfarth KK, Zuna I, Debus J, van Kaick G. Proton MR spectroscopic evaluation of suspicious brain lesions after stereotactic radiotherapy. AJNR Am J Neuroradiol 2001;22:1316–1324 [PMC free article] [PubMed] [Google Scholar]
  • 55.Kamada K, Houkin K, Abe H, Sawamura Y, Kashiwaba T. Differentiation of cerebral radiation necrosis from tumor recurrence by proton magnetic resonance spectroscopy. Neurol Med Chir (Tokyo) 1997;37:250–256 [DOI] [PubMed] [Google Scholar]
  • 56.Rock JP, Hearshen D, Scarpace L, et al. Correlations between magnetic resonance spectroscopy and image-guided histopathology, with special attention to radiation necrosis. Neurosurgery 2002;51:912–919 [DOI] [PubMed] [Google Scholar]
  • 57.Dowling C, Bollen AW, Noworolski SM, et al. Preoperative proton MR spectroscopic imaging of brain tumors: correlation with histopathologic analysis of resection specimens. AJNR Am J Neuroradiol 2001;22:604–612 [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Supplemental Table
supp_253_2_486__index.html (1.3KB, html)
supp_radiol.2532090007_tableE1.gif (313.3KB, gif)
supp_radiol.2532090007_tableE2.gif (111.7KB, gif)

Articles from Radiology are provided here courtesy of Radiological Society of North America

RESOURCES