Imaging After GliaSite Brachytherapy
Purpose
In this study, we analyzed the magnetic resonance imaging (MRI) changes in patients after GliaSite treatment and characterized the prognostic MRI indicators in these patients.
Methods and Materials
A total of 25 patients with recurrent glioblastoma multiforme were treated with the GliaSite Radiation Therapy System. Patients at the Johns Hopkins Hospital with recurrent glioblastoma multiforme underwent surgical resection followed by GliaSite balloon implantation. Available MRI scans for 20 patients were obtained throughout the post-GliaSite treatment course. These were reviewed and analyzed for prognostic significance.
Results
After GliaSite treatment, all patients developed some degree of T1-weighted contrast and T2-weighted hyperintensity around the resection cavity. The development of enhancement on T1-weighted contrast-enhanced imaging and the size of these lesions, in the absence of increasing T2-weighted hyperintensity, before clinical progression was not associated with decreased survival. Patients with T1-weighted enhancement >1 cm had a median survival of 13.6 months and those with T1-weighted lesions ≤1 cm had a median survival of 8.5 months (p = .014). In contrast, the development of larger areas of T2-weighted hyperintensity surrounding the resection cavity was significantly associated with poorer survival (p = .027).
Conclusion
After GliaSite treatment, characteristic T1- and T2-weighted changes are seen on MRI. Greater T1-weighted changes in the absence of increasing edema appears not to indicate disease progression; however, greater T2-weighted changes were associated with decreased survival. These findings suggest that T1-weighted enhancement in the absence of concomitant edema after GliaSite treatment might represent pseudoprogression. Conversely, increasing T2-weighted hyperintensity might reflect infiltrative disease progression. These results provide a framework for the analysis of disease control in future prospective studies of GliaSite treatment.
Glioblastoma multiforme, Recurrent glioma, GliaSite, Brachytherapy, Imaging
Article Outline
• Abstract
• Introduction
• Methods and Materials
• Results
• Discussion
• Conclusion
• References
• Copyright
Introduction
In the United States, glioblastoma multiforme (GBM) is nearly universally fatal and occurs at a frequency of approximately 8,500 cases annually (1). The disease constitutes about 80% of all malignant gliomas (2). The infiltrative nature of GBM prohibits cure with surgery alone. Despite surgical resection and adjuvant external beam radiotherapy, GBMs typically recur within 1 year. Local failure is the predominant mode of failure, and most relapses occur within 2 cm of the margin of surgical resection (3). In nearly all cases, disease progression is followed shortly by death.
In randomized trials, adjuvant external beam radiotherapy has been shown to result in a modest increase in survival in patients with newly diagnosed GBM 4, 5. The use of systemic chemotherapy such as carmustine provides an additional marginal benefit in survival for selected patients, resulting in an absolute increase in the 1-year survival rate from 40% to 46% (6). Adjuvant therapy with oral temozolomide, an oral alkylating agent with some activity against glioma, during and after RT increases the median survival to approximately 15 months (7). Likewise, the use of local chemotherapy regimens such as biodegradable carmustine wafers (Gliadel) has been shown to increase survival by about 2 months, with fewer toxicities than systemic chemotherapy (8).
The GliaSite device allows the delivery of a radiation dose to areas most at risk of recurrence while addressing some of the potential limiting factors of interstitial brachytherapy. The GliaSite Radiation Therapy System (RTS) makes use of a silicone balloon catheter and an aqueous iodinated radiation source (Iotrex [sodium 3-(125I)-iodo-4-hydroxybenzenesulfonate]). Immediately after surgery, the distal balloon portion of the GliaSite device is placed inside the resection cavity. After placement of the balloon, the injection port is brought to the surface of the skull to allow subcutaneous access to the balloon. Intracavitary brachytherapy is then performed by filling the balloon with Iotrex using the subcutaneous port. The balloon is left filled for a predetermined dwell time to deliver the prescribed radiation dose. After treatment, the Iotrex is taken out of the balloon, and the implanted device is surgically removed. Tatter et al. (9) described the safety and feasibility of the GliaSite RTS.
The GliaSite RTS has now been used to treat recurrent high-grade glioma, primary high-grade glioma (as a boost), and brain metastases in the adjuvant setting 10, 11, 12, 13. The treatment of recurrent glioblastoma after repeat resection has resulted in survival that is favorable compared with that after surgery alone. Proper patient selection is important because favorable results are seen in patients with a Karnosfky performance status of ≥70 at GliaSite treatment but not for patients with a lower Karnosfky performance status (10). The rate of symptomatic radionecrosis was low—3–4%—and the treatment was well tolerated 10, 11. Prospective trials have been conducted to determine the efficacy of GliaSite treatment for patients with recurrent GBM, as well as for patients newly diagnosed with glioblastoma. However, the results from these trials have proved difficult to interpret, because the magnetic resonance imaging (MRI) characteristics during the post-treatment period were difficult to interpret. In the weeks after GliaSite treatment, it was common for symmetric contrast enhancement to be seen on T1-weighted images. This enhancement was at times accompanied by symptoms, and it was unclear whether these findings indicated disease progression or post-RT change. As such, the duration of disease control afforded by GliaSite treatment has been difficult to determine accurately owing to these uncertainties, and the patients might be subject to unnecessary neurosurgical procedures. In the present study, we analyzed the post-GliaSite MRI studies for patients with recurrent GBM who were treated with repeat resection followed by GliaSite brachytherapy at Johns Hopkins Hospital. This analysis was performed to determine the prognostic value of MRI findings for patients in the post-treatment period.
Methods and Materials
We retrospectively analyzed the survival and MRI imaging data for 20 patients with recurrent GBM (World Health Organization Grade IV malignant glioma). Between February 2000 and April 2004, patients at Johns Hopkins Hospital with recurrent GBM, who had previously undergone surgery and external beam radiotherapy, underwent surgery for a gross total resection followed by GliaSite balloon implantation. The primary tumor site for all patients was supratentorial. All primary and recurrent tumors were histologically proven to be GBM. All patients had previously been treated for newly diagnosed GBM with surgical resection and adjuvant external beam radiotherapy. The patient characteristics are listed in Table 1. The median patient age was 48 years (range, 25–69). The mean Karnofsky performance status at initial diagnosis was 80 (range, 50–100). Patients had undergone gross total resection for their initial surgical resection. At repeat resection, 18 patients (90%) underwent gross total resection and 2 (10%) underwent subtotal resection (90% tumor removal).
Table 1. Patient characteristics
Patients (n) 20
Gender (n)
Male 6 (30)
Female 14 (70)
Age (y)
Median 48
Range 25–69
Age (n)
<50 y 11 (55)
≥50 y 9 (45)
KPS at diagnosis
Median 80
Range 50–100
KPS at GliaSite
Median 80
Range 60–100
KPS ≥70 (n) 18 (90)
KPS <70 (n) 2 (10)
RPA class at diagnosis (n)
3 5 (25)
4 8 (40)
5 6 (30)
6 1 (5)
Surgical resection extent (n)
Biopsy 0 (0)
Subtotal 2 (10)
Gross total 18 (90)
Postoperative external beam radiotherapy (cGy)
Mean dose 5,972
Mean dose per fraction 190.7
Chemotherapy (n)
Yes 13 (65)
No 7 (35)
Abbreviations: KPS = Karnofsky performance status; RPA = recursive partitioning analysis.
Data in parentheses are percentages.
All patients who underwent repeat resection and placement of the GliaSite balloon after recurrence of their GBM were found to have recurrence as determined by both clinical and radiographic (MRI) progression. The GliaSite balloon implantation technique and criteria for treatment have been previously described (10). The appropriate candidates were patients with tumors ≤30 cm3 that did not extend across the midline. After resection of the GBM, a GliaSite balloon catheter with a diameter of 2, 3, or 4 cm was placed in the resection cavity. During the post-GliaSite period, all patients were given a standard taper of dexamethasone for 30 days. No patients received antiangiogenic therapy before GliaSite treatment.
Follow-up was performed with serial MRI scans and clinical assessments. Post-GliaSite MRI scans were obtained at 6 and 12 weeks after treatment and at clinical progression. We obtained all MRI scans for the 20 patients with GBM after GliaSite brachytherapy. All MRI scans were performed using 4-mm-thick, T1-weighted imaging after administration of 0.1 mmol/kg Magnevist (gadopentetate dimeglumine), T2-weighted imaging, and 4-mm-thick, axial fast fluid-attenuated inversion recovery (FLAIR) imaging. The imaging studies were analyzed with particular attention to changes in contrast enhancement and signal intensity surrounding the resection site. The assessment of T1-weighted contrast-enhanced images included measurement of the lesion size by measuring the maximal thickness of T1 enhancement around the surgical cavity. The size of the lesions seen on the T2-weighted images was determined by measuring the elliptical area of the T2 lesion at the image slice, with the largest area of T2 size. The measurements were performed using the Merge eFilm Workstation software. The MRI scans were read and interpreted by a neuroradiologist.
Four patients underwent fluorodeoxyglucose-positron emission tomography (FDG-PET) of the brain to evaluate for disease progression after GliaSite therapy. The agent administered during this test was intravenous 10.06 MCI F-18 fluoro-deoxy-d-glucose. A neuroradiologist analyzed the PET images for abnormal activity and interpreted the results. Deaths were confirmed using the U.S. Social Security Death Index and patient records.
Survival estimates were obtained using the Kaplan-Meier method. Statistical analysis was performed using the Stata software suite. Survival estimates were calculated using the Kaplan-Meier method and the log–rank statistic for univariate analysis. Spearman's test was used for correlation analysis.
Results
A total of 20 patients with recurrent GBM underwent repeat resection and subsequent GliaSite brachytherapy. At the analysis, all 20 patients had died. We defined overall survival as the interval from the initial tissue diagnosis to death. The Kaplan-Meier curve for overall survival is shown in Fig. 1. Survival after GliaSite was defined as the interval from the end of GliaSite treatment to death. The Kaplan-Meier curve for survival after GliaSite is shown in Fig. 2.
Fig. 1. Kaplan-Meier survival curve for all patients treated with GliaSite and included in our imaging analysis (n = 20). Survival defined as interval from pathologic diagnosis to death.
Fig. 2. Kaplan-Meier curve for survival after GliaSite therapy for patients in our imaging analysis.
Follow-up was performed with clinical assessments and serial MRI scans after GliaSite therapy. We defined clinical progression as recurrent disease initially determined by clinical assessment (e.g., increasing headache, steroid dependence, and/or new neurologic symptoms) and then confirmed by MRI findings (e.g., development of mass effect and/or dramatic increase in T1-weighted positive contrast/T2-weighted hyperintensity compared with the previous scan). All 20 patients in the study had had clinical progression using these criteria before death. The median interval from GliaSite brachytherapy to clinical progression was 4 months. The median survival after clinical progression was 5.8 months.
We analyzed a total of 79 MRI studies for 20 patients after GliaSite treatment for recurrent GBM. The MRI characteristics of the patients are summarized in Table 2.
Table 2. MRI characteristics of patients with GBM after treatment with GliaSite
Interval from GliaSite to first post-GliaSite MRI (mo)
Median 1.5
Range 0.1–17.6
Interval between consecutive MRI scans (mo)
Median 1.6
Range 0.2–10.4
Changes observed on MRI
Maximal thickness of T1-weighted contrast enhancement (cm) 1.2 ± 0.4
Interval to maximal thickness of T1 contrast enhancement (mo)
Median 2.8
Range 0.3–21.7
Maximal elliptical area of T2/FLAIR hyperintensity (cm2) 240 ± 104
Interval to maximal elliptical area of T2/FLAIR hyperintensity (mo)
Median 3.8
Range 0.2–20.7
Interval from GliaSite to clinical progression (mo)
Median 4.0
Range 0.6–17.6
Changes observed on MRI before clinical progression
Maximal thickness of T1 contrast enhancement (cm) 0.86 ± 0.40
Interval to maximal thickness of T1 contrast enhancement (mo)
Median 2.6
Range 0.3–11.3
Maximal elliptical area of T2/FLAIR hyperintensity (cm2) 196 ± 88.5
Interval to maximal elliptical area of T2/FLAIR hyperintensity (mo)
Median 2.6
Range 0.3–11.3
Abbreviations: MRI = magnetic resonance imaging; GBM = glioblastoma multiforme; FLAIR = fast fluid-attenuated inversion recovery.
Data presented as mean ± standard deviation.
For all 20 patients, T1-weighted contrast-enhanced and T2-weighted/FLAIR MRI scans were analyzed for prognostic significance. After GliaSite therapy, all patients developed some degree of enhancement around the resection cavity on T1 and T2/FLAIR imaging. The mean maximal thickness of the T1-weighted contrast enhancement was 1.2 ± 0.4 cm, with a median time to maximal thickness of 2.8 months after GliaSite therapy. For the T2-weighted/FLAIR images, the mean maximal elliptical area of hyperintensity was 240 ± 104 cm2. The MRI studies were also analyzed in relation to the time to clinical progression. For T1-weighted images, the mean maximal thickness of enhancement attained before clinical progression was 0.86 ± 0.40 cm. The median interval between GliaSite therapy and maximal thickness before clinical progression was 2.6 months. For the T2-weighted/FLAIR images, the mean maximal elliptical area attained before clinical progression was 196 ± 88.5 cm2, with a median time of 2.6 months after GliaSite therapy. At clinical progression, the common findings on MRI included increased T2 signal hyperintensity, vasogenic edema, and mass effect. Figure 3 shows a representative patient's T1-weighted and T2-weighted/FLAIR changes after GliaSite therapy.
Fig. 3. Magnetic resonance imaging scans from representative patient after GliaSite therapy showing concomitant increases in lesion size on both T1- and T2-weighted magnetic resonance imaging. (a) Axial contrast-enhanced T1-weighted image showing resection cavity and surrounding enhancement. Image taken 36 days after GliaSite therapy. (b) Corresponding axial T2/fast fluid-attenuated inversion recovery image showing signal hyperintensity. Image obtained 36 days after GliaSite therapy. (c) Axial contrast-enhanced T1-weighted image showing increased thickness of rim of enhancement surrounding resection cavity at clinical progression of glioblastoma multiforme, 78 days after GliaSite therapy. (d) Corresponding axial T2/fast fluid-attenuated inversion recovery image showing increased signal hyperintensity at clinical progression of glioblastoma multiforme, 78 days after GliaSite therapy.
In general, FDG-PET or MRI spectroscopy results confirmed progressive disease at clinical progression. At clinical progression, 4 of the 20 patients with GBM underwent FDG-PET imaging after GliaSite therapy. These patients underwent FDG-PET imaging after they had clinical progression of their GBM. The median interval from clinical progression to FDG-PET imaging was 16 days. In 3 patients, the FDG-PET studies confirmed tumor recurrence in the region of their resection site; the fourth patient had a FDG-PET result that was indeterminate but progression could not be ruled out. Figure 4 shows FDG-PET images from a representative patient with tumor recurrence. MRI spectroscopy imaging was performed in 1 patient 20 days after the date of clinical progression. The patient's spectroscopy results showed reduced N-acetylaspartate/choline ratios in the region surrounding the resection cavity suggestive of tumor recurrence.
Fig. 4. Fluorodeoxyglucose-positron emission tomography (FDG-PET) cerebral glucose scans for representative patient at clinical progression of glioblastoma multiforme (GBM) after GliaSite therapy. (a) Enhancement on axial FDG-PET scan surrounding resection cavity. (b) Enhancement on coronal FDG-PET scan in corresponding region.
The MRI data were analyzed for prognostic significance on the basis of the thickness of the maximal T1 enhancement and the size of the T2/FLAIR hyperintensity. Patients with T1 enhancement >1 cm had a median survival after GliaSite treatment of 13.5 months and those with T1 enhancement of ≤1 cm had a median survival of 8.4 months (p = .014). We also analyzed the changes before clinical progression for all patients. Patients with T1-weighted enhancement >1 cm before clinical progression had a median survival of 19.3 months, but those with T1 enhancement of <1 cm before clinical progression had a median survival of 8.4 months (p = .004). In contrast, the development of larger T2-weighted lesions was associated with poorer survival, with Spearman's correlation test yielding r = −0.45 (p = .027). Table 3 summarizes the results for the MRI findings and prognostic factors. The measurements before clinical progression were unlikely to have been related to differences in steroid use because no difference was found in the amount of steroids taken by the patients throughout the range of measurements during this period.
Table 3. Prognostic MRI findings after treatment with GliaSite
Prognostic factor Survival (mo) p
Maximal T1 enhancement (cm)
>1 13.5 (5.6–27.3) .014
≤1 8.4 (4.5–12.9)
Maximal T1 enhancement before clinical progression
>1 19.3 (13.8–23.3 .004
≤1 8.4 (4.5–12.9)
Maximal T2 enhancement elliptical area r = −0.45 .027†
Abbreviation: MRI = magnetic resonance imaging.
Data in parentheses are ranges.
For T2 area as continuous variable in proportional hazards model, p value likelihood ratio statistic) was < .05 after elimination of other variables (i.e., dose, duration of steroid use), none of which were significant.
Log–rank test.
†
Spearman's test for correlation, significant as continuous variable.
We next attempted to evaluate post-GliaSite T2 changes in relationship to the T1 changes. Although the numbers of patients available for investigation did not allow for meaningful multivariate analysis, we did perform a subgroup analysis of the survival of patients with or without significant increases in T2 hyperintensity during the period before clinical progression. Significant increases in T2 hyperintensity were defined as an increase of >30% in the maximal diameter of the area of T2 hyperintensity surrounding the resection cavity during the period of interest. Patients with T1 enhancement >1 cm during the period before clinical progression had a median survival of 19.3 months. In contrast, those with T1 enhancement <1 cm before clinical progression had a median survival of 8.4 months (p = .004). However, the subset of patients with T1 enhancement >1 cm and a concomitant significant increase in T2 hyperintensity before clinical progression had a median survival of only 10.1 months (19.3 vs. 10.1 months, p = .04). Thus, it appears that the development of T1-positive contrast enhancement alone at the treatment site does not necessarily indicate disease progression. Large areas of T1-positive contrast enhancement in the absence of T2 changes were associated with longer progression-free survival. Thus, T1-positive contrast lesions might often correspond to areas of radiation change and not proliferating tumor. In contrast, progressive increases in T2 hyperintensity likely signifies infiltrative disease and disease recurrence.
Discussion
The GliaSite RTS enables the delivery of a high, homogeneous radiation dose to the tissues most at risk after surgical resection of brain lesions. This treatment is currently in use by a number of institutions for recurrent glioma and as a boost for the initial adjuvant treatment of GBM 11, 12. The survival of patients treated with GliaSite has appeared favorable, but data from well-controlled, prospective studies are lacking. Successful completion of such studies have been hampered by difficulty in interpreting post-GliaSite MRI data, which is required for accurate determination of the clinical endpoints required to gauge the efficacy of the treatment.
We conducted the present study to gain a better understanding of the MRI changes that occur after patients with recurrent GBM are treated with the GliaSite. Our analysis has indicated that distinct imaging changes exist, evident on MRI scans, that are associated with a better or worse outcome in the post-GliaSite treatment period. The development of a contrast-enhancing area around the treatment site is not, in and of itself, indicative of recurrence in the absence of significantly increased vasogenic edema. These lesions might represent post-RT changes in the area of the brain that received the greatest radiation doses or, alternatively, postoperative scarring (14). The cycling endothelium is a sensitive target for radiation damage that promotes blood–brain barrier breakdown (15). This breakdown can present as an area of contrast enhancement on T1-weighted images. Thus, it is not surprising that patients who developed large T1-positive contrast changes after GliaSite treatment might not all have had disease progression.
The presence of T1-positive contrast lesions at the GliaSite treatment location does not necessarily indicate disease progress; however, in some patients, it can. Our analysis has indicated that in patients with progression in this setting, the MRI scans from these patients will often demonstrate a concomitant increase in T2 signal surrounding the T1-weighted contrast-enhancing lesion. It appears that T2 lesion increases, together with T1-weighted contrast enhancement, are more indicative of disease progression. The increased T2 signal is consistent with the development of progressive infiltrative disease into the surrounding brain parenchyma. This imaging finding was also described in a smaller study reported by Matheus et al. (16) to be associated with poorer progression-free survival. They also concluded that progressive hyperintense abnormalities on T2-weighted images were due to tumor recurrence, although the low numbers analyzed in their study did not allow for statistical evaluation. Patients with the largest T1-positive contrast lesions without increasing vasogenic edema appeared to have a better survival than those with progressively increasing edema.
Currently, distinguishing between residual tumor and radiation necrosis is not feasible with MRI alone. In our experience, the incidence of symptomatic radionecrosis in patients with recurrent GBM treated with GliaSite is 8% (2 of 25) (10). The 2 patients with symptomatic radionecrosis at our institution were not included in the present analysis, because they had both undergone repeat resection and extended imaging was not available. Of the patients analyzed in our study, there were no cases of symptomatic radionecrosis. Nevertheless, it is possible that some of our patients did experience some level of subclinical radionecrosis after GliaSite treatment. Definitive documentation of this would require pathologic verification, which was not feasible at the time. Additional study is required to determine how the presence or absence of radionecrosis affects the prognostic MRI findings we have identified.
The use of PET and/or MRI spectroscopy might aid in differentiating between progressive disease and radionecrosis. Five patients underwent either PET or MRI spectroscopy at suspected clinical progression. On the MRI studies, all 5 patients had developed contrast-enhancing lesions on T1-weighted imaging accompanied by a ≥30% increase in the size of the T2-FLAIR enhancement. Of the 5 patients, 4 had PET or spectroscopy results indicative of disease progression, and 1 patient had indeterminate study findings. This suggests that PET and/or MRI spectroscopy, although not definitive, might be useful in detecting disease progression. It should be noted, however, that FDG-PET can yield false-positive results in this setting. Thus, although helpful, PET alone cannot reliably distinguish radionecrosis from tumor recurrence (17).
The finding that T1-weighted contrast-enhancing lesions might not herald local recurrence is increasingly being documented in GBM patients treated with external beam radiotherapy. This so called “pseudoprogression” usually manifests itself as the development of contrast enhancement in the treatment site shortly after adjuvant radiotherapy (18). This phenomenon is due to treatment-associated scarring or regional necrosis, resulting in the disruption of the hematoencephalic barrier and passage of contrast medium (19). Given that GliaSite treatment often treats the adjoining resection cavity surface to a high radiation dose, our findings are consistent with the T1-enhancing lesions we observed as being a type of pseudoprogression that does not signify disease recurrence. The time course in which we see these imaging effects was similar to that seen for central nervous system tumors treated with radiosurgery or interstitial brachytherapy 20, 21, 22. For example, Moringlane et al. (22) noted that a ring-like signal on T1-positive contrast images developed as early as 1 month in most of their glioma patients treated with interstitial 125I. This timing is very consistent with our own observations after GliaSite treatment.
Conclusion
We have identified several useful MRI factors that can aid in the interpretation of the imaging findings from patients with recurrent GBM who have undergone GliaSite radiotherapy. Post-GliaSite patients often present with complex MRI findings, often with a large compendium of changes that occur over time. Because of the difficulty in interpreting these images, the determination of whether a patient has developed recurrence or not is often empirical. Although not definitive, the prognostic imaging factors we have identified can aid the clinician with interpretation of the MRI changes that occur during the post-GliaSite period. These factors should be considered in any future prospective studies of the GliaSite RTS.
In this study, we analyzed the magnetic resonance imaging (MRI) changes in patients after GliaSite treatment and characterized the prognostic MRI indicators in these patients.
Methods and Materials
A total of 25 patients with recurrent glioblastoma multiforme were treated with the GliaSite Radiation Therapy System. Patients at the Johns Hopkins Hospital with recurrent glioblastoma multiforme underwent surgical resection followed by GliaSite balloon implantation. Available MRI scans for 20 patients were obtained throughout the post-GliaSite treatment course. These were reviewed and analyzed for prognostic significance.
Results
After GliaSite treatment, all patients developed some degree of T1-weighted contrast and T2-weighted hyperintensity around the resection cavity. The development of enhancement on T1-weighted contrast-enhanced imaging and the size of these lesions, in the absence of increasing T2-weighted hyperintensity, before clinical progression was not associated with decreased survival. Patients with T1-weighted enhancement >1 cm had a median survival of 13.6 months and those with T1-weighted lesions ≤1 cm had a median survival of 8.5 months (p = .014). In contrast, the development of larger areas of T2-weighted hyperintensity surrounding the resection cavity was significantly associated with poorer survival (p = .027).
Conclusion
After GliaSite treatment, characteristic T1- and T2-weighted changes are seen on MRI. Greater T1-weighted changes in the absence of increasing edema appears not to indicate disease progression; however, greater T2-weighted changes were associated with decreased survival. These findings suggest that T1-weighted enhancement in the absence of concomitant edema after GliaSite treatment might represent pseudoprogression. Conversely, increasing T2-weighted hyperintensity might reflect infiltrative disease progression. These results provide a framework for the analysis of disease control in future prospective studies of GliaSite treatment.
Glioblastoma multiforme, Recurrent glioma, GliaSite, Brachytherapy, Imaging
Article Outline
• Abstract
• Introduction
• Methods and Materials
• Results
• Discussion
• Conclusion
• References
• Copyright
Introduction
In the United States, glioblastoma multiforme (GBM) is nearly universally fatal and occurs at a frequency of approximately 8,500 cases annually (1). The disease constitutes about 80% of all malignant gliomas (2). The infiltrative nature of GBM prohibits cure with surgery alone. Despite surgical resection and adjuvant external beam radiotherapy, GBMs typically recur within 1 year. Local failure is the predominant mode of failure, and most relapses occur within 2 cm of the margin of surgical resection (3). In nearly all cases, disease progression is followed shortly by death.
In randomized trials, adjuvant external beam radiotherapy has been shown to result in a modest increase in survival in patients with newly diagnosed GBM 4, 5. The use of systemic chemotherapy such as carmustine provides an additional marginal benefit in survival for selected patients, resulting in an absolute increase in the 1-year survival rate from 40% to 46% (6). Adjuvant therapy with oral temozolomide, an oral alkylating agent with some activity against glioma, during and after RT increases the median survival to approximately 15 months (7). Likewise, the use of local chemotherapy regimens such as biodegradable carmustine wafers (Gliadel) has been shown to increase survival by about 2 months, with fewer toxicities than systemic chemotherapy (8).
The GliaSite device allows the delivery of a radiation dose to areas most at risk of recurrence while addressing some of the potential limiting factors of interstitial brachytherapy. The GliaSite Radiation Therapy System (RTS) makes use of a silicone balloon catheter and an aqueous iodinated radiation source (Iotrex [sodium 3-(125I)-iodo-4-hydroxybenzenesulfonate]). Immediately after surgery, the distal balloon portion of the GliaSite device is placed inside the resection cavity. After placement of the balloon, the injection port is brought to the surface of the skull to allow subcutaneous access to the balloon. Intracavitary brachytherapy is then performed by filling the balloon with Iotrex using the subcutaneous port. The balloon is left filled for a predetermined dwell time to deliver the prescribed radiation dose. After treatment, the Iotrex is taken out of the balloon, and the implanted device is surgically removed. Tatter et al. (9) described the safety and feasibility of the GliaSite RTS.
The GliaSite RTS has now been used to treat recurrent high-grade glioma, primary high-grade glioma (as a boost), and brain metastases in the adjuvant setting 10, 11, 12, 13. The treatment of recurrent glioblastoma after repeat resection has resulted in survival that is favorable compared with that after surgery alone. Proper patient selection is important because favorable results are seen in patients with a Karnosfky performance status of ≥70 at GliaSite treatment but not for patients with a lower Karnosfky performance status (10). The rate of symptomatic radionecrosis was low—3–4%—and the treatment was well tolerated 10, 11. Prospective trials have been conducted to determine the efficacy of GliaSite treatment for patients with recurrent GBM, as well as for patients newly diagnosed with glioblastoma. However, the results from these trials have proved difficult to interpret, because the magnetic resonance imaging (MRI) characteristics during the post-treatment period were difficult to interpret. In the weeks after GliaSite treatment, it was common for symmetric contrast enhancement to be seen on T1-weighted images. This enhancement was at times accompanied by symptoms, and it was unclear whether these findings indicated disease progression or post-RT change. As such, the duration of disease control afforded by GliaSite treatment has been difficult to determine accurately owing to these uncertainties, and the patients might be subject to unnecessary neurosurgical procedures. In the present study, we analyzed the post-GliaSite MRI studies for patients with recurrent GBM who were treated with repeat resection followed by GliaSite brachytherapy at Johns Hopkins Hospital. This analysis was performed to determine the prognostic value of MRI findings for patients in the post-treatment period.
Methods and Materials
We retrospectively analyzed the survival and MRI imaging data for 20 patients with recurrent GBM (World Health Organization Grade IV malignant glioma). Between February 2000 and April 2004, patients at Johns Hopkins Hospital with recurrent GBM, who had previously undergone surgery and external beam radiotherapy, underwent surgery for a gross total resection followed by GliaSite balloon implantation. The primary tumor site for all patients was supratentorial. All primary and recurrent tumors were histologically proven to be GBM. All patients had previously been treated for newly diagnosed GBM with surgical resection and adjuvant external beam radiotherapy. The patient characteristics are listed in Table 1. The median patient age was 48 years (range, 25–69). The mean Karnofsky performance status at initial diagnosis was 80 (range, 50–100). Patients had undergone gross total resection for their initial surgical resection. At repeat resection, 18 patients (90%) underwent gross total resection and 2 (10%) underwent subtotal resection (90% tumor removal).
Table 1. Patient characteristics
Patients (n) 20
Gender (n)
Male 6 (30)
Female 14 (70)
Age (y)
Median 48
Range 25–69
Age (n)
<50 y 11 (55)
≥50 y 9 (45)
KPS at diagnosis
Median 80
Range 50–100
KPS at GliaSite
Median 80
Range 60–100
KPS ≥70 (n) 18 (90)
KPS <70 (n) 2 (10)
RPA class at diagnosis (n)
3 5 (25)
4 8 (40)
5 6 (30)
6 1 (5)
Surgical resection extent (n)
Biopsy 0 (0)
Subtotal 2 (10)
Gross total 18 (90)
Postoperative external beam radiotherapy (cGy)
Mean dose 5,972
Mean dose per fraction 190.7
Chemotherapy (n)
Yes 13 (65)
No 7 (35)
Abbreviations: KPS = Karnofsky performance status; RPA = recursive partitioning analysis.
Data in parentheses are percentages.
All patients who underwent repeat resection and placement of the GliaSite balloon after recurrence of their GBM were found to have recurrence as determined by both clinical and radiographic (MRI) progression. The GliaSite balloon implantation technique and criteria for treatment have been previously described (10). The appropriate candidates were patients with tumors ≤30 cm3 that did not extend across the midline. After resection of the GBM, a GliaSite balloon catheter with a diameter of 2, 3, or 4 cm was placed in the resection cavity. During the post-GliaSite period, all patients were given a standard taper of dexamethasone for 30 days. No patients received antiangiogenic therapy before GliaSite treatment.
Follow-up was performed with serial MRI scans and clinical assessments. Post-GliaSite MRI scans were obtained at 6 and 12 weeks after treatment and at clinical progression. We obtained all MRI scans for the 20 patients with GBM after GliaSite brachytherapy. All MRI scans were performed using 4-mm-thick, T1-weighted imaging after administration of 0.1 mmol/kg Magnevist (gadopentetate dimeglumine), T2-weighted imaging, and 4-mm-thick, axial fast fluid-attenuated inversion recovery (FLAIR) imaging. The imaging studies were analyzed with particular attention to changes in contrast enhancement and signal intensity surrounding the resection site. The assessment of T1-weighted contrast-enhanced images included measurement of the lesion size by measuring the maximal thickness of T1 enhancement around the surgical cavity. The size of the lesions seen on the T2-weighted images was determined by measuring the elliptical area of the T2 lesion at the image slice, with the largest area of T2 size. The measurements were performed using the Merge eFilm Workstation software. The MRI scans were read and interpreted by a neuroradiologist.
Four patients underwent fluorodeoxyglucose-positron emission tomography (FDG-PET) of the brain to evaluate for disease progression after GliaSite therapy. The agent administered during this test was intravenous 10.06 MCI F-18 fluoro-deoxy-d-glucose. A neuroradiologist analyzed the PET images for abnormal activity and interpreted the results. Deaths were confirmed using the U.S. Social Security Death Index and patient records.
Survival estimates were obtained using the Kaplan-Meier method. Statistical analysis was performed using the Stata software suite. Survival estimates were calculated using the Kaplan-Meier method and the log–rank statistic for univariate analysis. Spearman's test was used for correlation analysis.
Results
A total of 20 patients with recurrent GBM underwent repeat resection and subsequent GliaSite brachytherapy. At the analysis, all 20 patients had died. We defined overall survival as the interval from the initial tissue diagnosis to death. The Kaplan-Meier curve for overall survival is shown in Fig. 1. Survival after GliaSite was defined as the interval from the end of GliaSite treatment to death. The Kaplan-Meier curve for survival after GliaSite is shown in Fig. 2.
Fig. 1. Kaplan-Meier survival curve for all patients treated with GliaSite and included in our imaging analysis (n = 20). Survival defined as interval from pathologic diagnosis to death.
Fig. 2. Kaplan-Meier curve for survival after GliaSite therapy for patients in our imaging analysis.
Follow-up was performed with clinical assessments and serial MRI scans after GliaSite therapy. We defined clinical progression as recurrent disease initially determined by clinical assessment (e.g., increasing headache, steroid dependence, and/or new neurologic symptoms) and then confirmed by MRI findings (e.g., development of mass effect and/or dramatic increase in T1-weighted positive contrast/T2-weighted hyperintensity compared with the previous scan). All 20 patients in the study had had clinical progression using these criteria before death. The median interval from GliaSite brachytherapy to clinical progression was 4 months. The median survival after clinical progression was 5.8 months.
We analyzed a total of 79 MRI studies for 20 patients after GliaSite treatment for recurrent GBM. The MRI characteristics of the patients are summarized in Table 2.
Table 2. MRI characteristics of patients with GBM after treatment with GliaSite
Interval from GliaSite to first post-GliaSite MRI (mo)
Median 1.5
Range 0.1–17.6
Interval between consecutive MRI scans (mo)
Median 1.6
Range 0.2–10.4
Changes observed on MRI
Maximal thickness of T1-weighted contrast enhancement (cm) 1.2 ± 0.4
Interval to maximal thickness of T1 contrast enhancement (mo)
Median 2.8
Range 0.3–21.7
Maximal elliptical area of T2/FLAIR hyperintensity (cm2) 240 ± 104
Interval to maximal elliptical area of T2/FLAIR hyperintensity (mo)
Median 3.8
Range 0.2–20.7
Interval from GliaSite to clinical progression (mo)
Median 4.0
Range 0.6–17.6
Changes observed on MRI before clinical progression
Maximal thickness of T1 contrast enhancement (cm) 0.86 ± 0.40
Interval to maximal thickness of T1 contrast enhancement (mo)
Median 2.6
Range 0.3–11.3
Maximal elliptical area of T2/FLAIR hyperintensity (cm2) 196 ± 88.5
Interval to maximal elliptical area of T2/FLAIR hyperintensity (mo)
Median 2.6
Range 0.3–11.3
Abbreviations: MRI = magnetic resonance imaging; GBM = glioblastoma multiforme; FLAIR = fast fluid-attenuated inversion recovery.
Data presented as mean ± standard deviation.
For all 20 patients, T1-weighted contrast-enhanced and T2-weighted/FLAIR MRI scans were analyzed for prognostic significance. After GliaSite therapy, all patients developed some degree of enhancement around the resection cavity on T1 and T2/FLAIR imaging. The mean maximal thickness of the T1-weighted contrast enhancement was 1.2 ± 0.4 cm, with a median time to maximal thickness of 2.8 months after GliaSite therapy. For the T2-weighted/FLAIR images, the mean maximal elliptical area of hyperintensity was 240 ± 104 cm2. The MRI studies were also analyzed in relation to the time to clinical progression. For T1-weighted images, the mean maximal thickness of enhancement attained before clinical progression was 0.86 ± 0.40 cm. The median interval between GliaSite therapy and maximal thickness before clinical progression was 2.6 months. For the T2-weighted/FLAIR images, the mean maximal elliptical area attained before clinical progression was 196 ± 88.5 cm2, with a median time of 2.6 months after GliaSite therapy. At clinical progression, the common findings on MRI included increased T2 signal hyperintensity, vasogenic edema, and mass effect. Figure 3 shows a representative patient's T1-weighted and T2-weighted/FLAIR changes after GliaSite therapy.
Fig. 3. Magnetic resonance imaging scans from representative patient after GliaSite therapy showing concomitant increases in lesion size on both T1- and T2-weighted magnetic resonance imaging. (a) Axial contrast-enhanced T1-weighted image showing resection cavity and surrounding enhancement. Image taken 36 days after GliaSite therapy. (b) Corresponding axial T2/fast fluid-attenuated inversion recovery image showing signal hyperintensity. Image obtained 36 days after GliaSite therapy. (c) Axial contrast-enhanced T1-weighted image showing increased thickness of rim of enhancement surrounding resection cavity at clinical progression of glioblastoma multiforme, 78 days after GliaSite therapy. (d) Corresponding axial T2/fast fluid-attenuated inversion recovery image showing increased signal hyperintensity at clinical progression of glioblastoma multiforme, 78 days after GliaSite therapy.
In general, FDG-PET or MRI spectroscopy results confirmed progressive disease at clinical progression. At clinical progression, 4 of the 20 patients with GBM underwent FDG-PET imaging after GliaSite therapy. These patients underwent FDG-PET imaging after they had clinical progression of their GBM. The median interval from clinical progression to FDG-PET imaging was 16 days. In 3 patients, the FDG-PET studies confirmed tumor recurrence in the region of their resection site; the fourth patient had a FDG-PET result that was indeterminate but progression could not be ruled out. Figure 4 shows FDG-PET images from a representative patient with tumor recurrence. MRI spectroscopy imaging was performed in 1 patient 20 days after the date of clinical progression. The patient's spectroscopy results showed reduced N-acetylaspartate/choline ratios in the region surrounding the resection cavity suggestive of tumor recurrence.
Fig. 4. Fluorodeoxyglucose-positron emission tomography (FDG-PET) cerebral glucose scans for representative patient at clinical progression of glioblastoma multiforme (GBM) after GliaSite therapy. (a) Enhancement on axial FDG-PET scan surrounding resection cavity. (b) Enhancement on coronal FDG-PET scan in corresponding region.
The MRI data were analyzed for prognostic significance on the basis of the thickness of the maximal T1 enhancement and the size of the T2/FLAIR hyperintensity. Patients with T1 enhancement >1 cm had a median survival after GliaSite treatment of 13.5 months and those with T1 enhancement of ≤1 cm had a median survival of 8.4 months (p = .014). We also analyzed the changes before clinical progression for all patients. Patients with T1-weighted enhancement >1 cm before clinical progression had a median survival of 19.3 months, but those with T1 enhancement of <1 cm before clinical progression had a median survival of 8.4 months (p = .004). In contrast, the development of larger T2-weighted lesions was associated with poorer survival, with Spearman's correlation test yielding r = −0.45 (p = .027). Table 3 summarizes the results for the MRI findings and prognostic factors. The measurements before clinical progression were unlikely to have been related to differences in steroid use because no difference was found in the amount of steroids taken by the patients throughout the range of measurements during this period.
Table 3. Prognostic MRI findings after treatment with GliaSite
Prognostic factor Survival (mo) p
Maximal T1 enhancement (cm)
>1 13.5 (5.6–27.3) .014
≤1 8.4 (4.5–12.9)
Maximal T1 enhancement before clinical progression
>1 19.3 (13.8–23.3 .004
≤1 8.4 (4.5–12.9)
Maximal T2 enhancement elliptical area r = −0.45 .027†
Abbreviation: MRI = magnetic resonance imaging.
Data in parentheses are ranges.
For T2 area as continuous variable in proportional hazards model, p value likelihood ratio statistic) was < .05 after elimination of other variables (i.e., dose, duration of steroid use), none of which were significant.
Log–rank test.
†
Spearman's test for correlation, significant as continuous variable.
We next attempted to evaluate post-GliaSite T2 changes in relationship to the T1 changes. Although the numbers of patients available for investigation did not allow for meaningful multivariate analysis, we did perform a subgroup analysis of the survival of patients with or without significant increases in T2 hyperintensity during the period before clinical progression. Significant increases in T2 hyperintensity were defined as an increase of >30% in the maximal diameter of the area of T2 hyperintensity surrounding the resection cavity during the period of interest. Patients with T1 enhancement >1 cm during the period before clinical progression had a median survival of 19.3 months. In contrast, those with T1 enhancement <1 cm before clinical progression had a median survival of 8.4 months (p = .004). However, the subset of patients with T1 enhancement >1 cm and a concomitant significant increase in T2 hyperintensity before clinical progression had a median survival of only 10.1 months (19.3 vs. 10.1 months, p = .04). Thus, it appears that the development of T1-positive contrast enhancement alone at the treatment site does not necessarily indicate disease progression. Large areas of T1-positive contrast enhancement in the absence of T2 changes were associated with longer progression-free survival. Thus, T1-positive contrast lesions might often correspond to areas of radiation change and not proliferating tumor. In contrast, progressive increases in T2 hyperintensity likely signifies infiltrative disease and disease recurrence.
Discussion
The GliaSite RTS enables the delivery of a high, homogeneous radiation dose to the tissues most at risk after surgical resection of brain lesions. This treatment is currently in use by a number of institutions for recurrent glioma and as a boost for the initial adjuvant treatment of GBM 11, 12. The survival of patients treated with GliaSite has appeared favorable, but data from well-controlled, prospective studies are lacking. Successful completion of such studies have been hampered by difficulty in interpreting post-GliaSite MRI data, which is required for accurate determination of the clinical endpoints required to gauge the efficacy of the treatment.
We conducted the present study to gain a better understanding of the MRI changes that occur after patients with recurrent GBM are treated with the GliaSite. Our analysis has indicated that distinct imaging changes exist, evident on MRI scans, that are associated with a better or worse outcome in the post-GliaSite treatment period. The development of a contrast-enhancing area around the treatment site is not, in and of itself, indicative of recurrence in the absence of significantly increased vasogenic edema. These lesions might represent post-RT changes in the area of the brain that received the greatest radiation doses or, alternatively, postoperative scarring (14). The cycling endothelium is a sensitive target for radiation damage that promotes blood–brain barrier breakdown (15). This breakdown can present as an area of contrast enhancement on T1-weighted images. Thus, it is not surprising that patients who developed large T1-positive contrast changes after GliaSite treatment might not all have had disease progression.
The presence of T1-positive contrast lesions at the GliaSite treatment location does not necessarily indicate disease progress; however, in some patients, it can. Our analysis has indicated that in patients with progression in this setting, the MRI scans from these patients will often demonstrate a concomitant increase in T2 signal surrounding the T1-weighted contrast-enhancing lesion. It appears that T2 lesion increases, together with T1-weighted contrast enhancement, are more indicative of disease progression. The increased T2 signal is consistent with the development of progressive infiltrative disease into the surrounding brain parenchyma. This imaging finding was also described in a smaller study reported by Matheus et al. (16) to be associated with poorer progression-free survival. They also concluded that progressive hyperintense abnormalities on T2-weighted images were due to tumor recurrence, although the low numbers analyzed in their study did not allow for statistical evaluation. Patients with the largest T1-positive contrast lesions without increasing vasogenic edema appeared to have a better survival than those with progressively increasing edema.
Currently, distinguishing between residual tumor and radiation necrosis is not feasible with MRI alone. In our experience, the incidence of symptomatic radionecrosis in patients with recurrent GBM treated with GliaSite is 8% (2 of 25) (10). The 2 patients with symptomatic radionecrosis at our institution were not included in the present analysis, because they had both undergone repeat resection and extended imaging was not available. Of the patients analyzed in our study, there were no cases of symptomatic radionecrosis. Nevertheless, it is possible that some of our patients did experience some level of subclinical radionecrosis after GliaSite treatment. Definitive documentation of this would require pathologic verification, which was not feasible at the time. Additional study is required to determine how the presence or absence of radionecrosis affects the prognostic MRI findings we have identified.
The use of PET and/or MRI spectroscopy might aid in differentiating between progressive disease and radionecrosis. Five patients underwent either PET or MRI spectroscopy at suspected clinical progression. On the MRI studies, all 5 patients had developed contrast-enhancing lesions on T1-weighted imaging accompanied by a ≥30% increase in the size of the T2-FLAIR enhancement. Of the 5 patients, 4 had PET or spectroscopy results indicative of disease progression, and 1 patient had indeterminate study findings. This suggests that PET and/or MRI spectroscopy, although not definitive, might be useful in detecting disease progression. It should be noted, however, that FDG-PET can yield false-positive results in this setting. Thus, although helpful, PET alone cannot reliably distinguish radionecrosis from tumor recurrence (17).
The finding that T1-weighted contrast-enhancing lesions might not herald local recurrence is increasingly being documented in GBM patients treated with external beam radiotherapy. This so called “pseudoprogression” usually manifests itself as the development of contrast enhancement in the treatment site shortly after adjuvant radiotherapy (18). This phenomenon is due to treatment-associated scarring or regional necrosis, resulting in the disruption of the hematoencephalic barrier and passage of contrast medium (19). Given that GliaSite treatment often treats the adjoining resection cavity surface to a high radiation dose, our findings are consistent with the T1-enhancing lesions we observed as being a type of pseudoprogression that does not signify disease recurrence. The time course in which we see these imaging effects was similar to that seen for central nervous system tumors treated with radiosurgery or interstitial brachytherapy 20, 21, 22. For example, Moringlane et al. (22) noted that a ring-like signal on T1-positive contrast images developed as early as 1 month in most of their glioma patients treated with interstitial 125I. This timing is very consistent with our own observations after GliaSite treatment.
Conclusion
We have identified several useful MRI factors that can aid in the interpretation of the imaging findings from patients with recurrent GBM who have undergone GliaSite radiotherapy. Post-GliaSite patients often present with complex MRI findings, often with a large compendium of changes that occur over time. Because of the difficulty in interpreting these images, the determination of whether a patient has developed recurrence or not is often empirical. Although not definitive, the prognostic imaging factors we have identified can aid the clinician with interpretation of the MRI changes that occur during the post-GliaSite period. These factors should be considered in any future prospective studies of the GliaSite RTS.
posted by stefanoxx at 4:15 AM
0 Comments:
Post a Comment
<< Home