|Home | About | Journals | Submit | Contact Us | Français|
The aim of this study was to reveal the volumetric changes in tumor necrosis, reactive zone, and edema referred to as the “triple ring” appearing after low-dose-rate iodine-125 (I-125) interstitial irradiation of 20 inoperable low-grade gliomas. To enable prediction of these volumetric changes, we provide mathematical expressions that describe the dynamics of the triple ring. Volumes of the three regions on image-fused control CT/MR images were measured for a 24-month period. The delivered dose on the tumor surface was 50–60 Gy. Dose planning and image fusion were performed with Brain-Lab Target 1.19 software; mathematical and statistical computations were carried out with Matlab numeric computation and visualization software. To determine the volumes, control images with the triple rings were fused with the planning images. Relative volumes normalized with respect to the volume of reference dose were calculated and plotted in the time domain. First, the mean values of volumes were determined from the patients’ measured data; then, polynomials were fitted to the mean values using the polynomial curve-fitting method. The accuracy of our results was verified by correlating the predicted data with the measured ones. The polynomial prediction approach proposed here reveals the dynamics of the triple ring. These polynomials will assist with (1) designing the best treatment, (2) following the patient’s condition, and (3) planning reirradiation if necessary.
Interstitial low-dose-rate (LDR) iodine-125 (I-125) brachytherapy based on image fusion is considered to be a multimodality treatment of inoperable low-grade gliomas. After LDR I-125 brachytherapy, a so-called “triple ring” appears on the control CT and MR images around the implanted radionuclides.1 The three regions observed are (1) necrotized tumor, (2) a surrounding layer referred to as the reactive ring, and (3) edema of the white matter.
Here we present the dynamics of the triple ring, the pathological consequence of glioma I-125 brachytherapy. Although the changes in tumor volume, tumor necrosis, reactive zone, and edema can be monitored by imaging, few data are available in the literature on this topic. Postirradiation volumetric changes have not yet been studied; that is, the dynamics of the triple ring determined from a series of control CT/MR images taken after irradiation has not yet been evaluated.
Our goal was to determine whether it is possible to fit polynomials to the measured data in the time domain using the polynomial curve-fitting method in order to describe the dynamics of the triple ring.
Triple rings that developed in 20 patients with low-grade glioma were analyzed for 24 months following LDR I-125 interstitial irradiation.
Brachytherapy treatments were done with photon radiation of I-125 seeds (model 6702, Medi-Physics Inc., Arlington Heights, IL, USA) placed in Pastyr catheters (Pastyr GmbH, Cologne, Germany) under local anesthesia with stereotactic control. The placement of the catheters was temporary; they were removed at the end of treatment, usually after 1–5 weeks, depending on the activity of the sources.
Dose planning and image fusion were performed with Target 1.19 software (BrainLab AG, Kirchheim-Heimstetten, Germany) to determine the critical brain areas and the target tumor volume, and for verification of catheter locations.
In a previous study, we showed the importance of choosing the most appropriate seed placing configuration that gives the most homogeneous irradiation of tumor by the reference dose.2 Optimal seed placing assures the best homogeneity index while simultaneously minimizing radiotoxicity of the normal tissue.
Volumetric analysis was performed for 24 months after irradiation. Following the fusion of planning and control CT-CT/CT-MR images, we traced out the areas of necrosis, reactive ring, and edema in each slice and then calculated volumes of the selected regions.
Image fusion was performed between the CT-CT and CT-MR images. The following imaging devices were used: Elscint Elite 2400 CT scanner (Elscint, Haifa, Israel), with Philips Tomoscan AP (Kashiwa, Japan) and GE Brightspeed (Brightspeed Excel, Hangwei, China); and Siemens Magnetom Vision Plus 1.5-T MRI scanner. The contrast tracers were 100 ml Ultravist at CT and 10 ml Magnevist at MRI examinations (Bayer Schering Pharma AG, Leverkusen, Germany).
For planning purposes, we used contiguous CT slices with 1.5 mm thickness and MR slices with 1–3 mm thickness. The field of view and the in-plane resolution were 300 mm and 0.59 mm for CT and 240 mm and 0.47 mm for MRI.
The average applied dose on the tumor surface was 50–60 Gy. The reference volume (Vref) was defined as the brain volume irradiated with 50–60 Gy. According to the literature, these doses are accepted prescribed doses for gliomas; we applied 60 Gy for smaller tumor volumes and 50 Gy for larger ones. To make the measured data of different patients comparable, each measured volume was normalized with respect to Vref. For example, the normalized necrosis volume is given by V = Vnecr/Vref × 100%, where Vnecr is a volume of necrosis of one patient on a control MRI. Consequently, data are presented here as percentage of Vref. Mathematical and statistical computations were carried out with Matlab numeric computation and visualization software.
The dynamics of the triple ring is described by polynomials that are suitable for the prediction of the volumetric changes in tumor necrosis, reactive zone, and edema as a function of time. To get the three polynomials, we used the polynomial fitting technique, which is a multiple regression tool that fits an nth-order polynomial to a measured data sequence. The derived polynomial fits the measured data in the least-squares sense; that is, the average deviation of the measured data from the ones predicted by the polynomial is minimized.
The volumetric data of the triple ring were measured for the 20 patients. The measured data before and after normalization with respect to Vref are shown in Tables 1 and and2,2, respectively. The mean tumor volume was 10.3 cm3 (range, 2.3–24.3 cm3); the mean volume irradiated with the reference dose was 17.3 cm3 (range, 2.5–36.6 cm3).
Following irradiation, the volumetric measurements were carried out for 24 months. Only two patients died (after 18 and 22 months) during this time. The mean values of necrosis, reactive zone, and edema related to Vref were 19.9%, 30.4%, and 293.2%, respectively, after 6 months; 8.3%, 34.8%, and 220.5% after 14 months; and 7.0%, 16.7%, and 107.0% after 24 months. Our goal was to express the dynamics of tumor necrosis, reactive ring, and edema by means of three polynomials in the time domain. Fig. 1 shows the measured and normalized volumes of tumor necrosis, reactive zone, and edema for the observed 20 patients as a function of time elapsed after irradiation. We then determined the mean values of normalized volumes (Fig. 1, circles) and fitted polynomials to the means to derive a polynomial prediction (Fig. 1, curves).
Fig. 1A shows the dynamics of tumor necrosis modeled by the tumor necrosis polynomial. We applied an average dose of 50–60 Gy to the tumor surface. Tumor necrosis, observed in all of our patients, developed in the area where the radiation dose exceeded 79.2 Gy, on average. Tumor necrosis reached its maximum at an average of 6 months after irradiation. We observed shrinkage in necrosis between 9 and 15 months; after 16 months, the necrosis achieved its steady-state value (5.4%, on average).
Fig. 1B shows the dynamics of the reactive zone modeled by the reactive zone polynomial. The curve is very similar to that for necrosis, but the reactive ring seems to have developed earlier and reached its maximum 2 months after irradiation. After 12 months, its shrinkage was more pronounced.
In contrast, the edema polynomial, shown in Fig. 1C, describes a dynamics of greater diversity. Some of our patients had developed edema before irradiation, which we modeled as edema appearing at month 0. The volume of edema is about 17 times and 7 times greater than that of necrosis and the reactive zone, respectively. Edema peaked at 6 months, and most was reabsorbed by 16 months after irradiation, although we measured a considerable amount of edema 2–4 years after irradiation (data not shown).
We found that a sixth-order polynomial is required to model the dynamics of the triple ring over the 24-month time period:
where t denotes the time in months passed after irradiation, and V is the percentage volume normalized with respect to Vref. This polynomial is equally valid for necrosis, reactive zone, and edema; the constants C0–C6 are given in Table 3 for the three regions. Fig. 2 shows the dynamics of the triple ring modeled by the tumor necrosis, reactive zone, and edema polynomials.
Although the variances of the measured data seem to be too large to perform an accurate and detailed analysis of the dynamics of the triple ring, the high correlations obtained between each patient’s measured data and the fitted polynomials verify the applicability of our approach. On average, the correlations are 0.85, 0.89, and 0.84 for necrosis, reactive ring, and edema, respectively. This correlation and our long-term experience suggest that the polynomial prediction approach provides useful information in human radiotherapy.
Because the vast majority of space-occupying processes during brachytherapy are caused by edema, our most important result is its polynomial predictability. As shown in Fig. 1C, the edema polynomial predicts the volume occupied by edema as a function of time. Maximum edema is to be expected between 4 and 9 months after irradiation. The relative volumes are 408% (after 4 months) and 384% (after 9 months) compared with Vref.
Postirradiation morphological changes have been studied by several authors.3–6 However, their occurrence and volumetric change as a function of time have not yet been analyzed. We measured the quantitative changes of tumor necrosis, reactive zone, and edema for a 24-month time period after irradiation to reveal the dynamics of the triple ring and the relationships among its regions.
After LDR I-125 interstitial brachytherapy, a so-called “triple ring” appears on the control CT and MR images around the implanted radionuclide. The necrotized tumor is surrounded by a region referred to as the reactive ring. Histologically, it consists of dilated and proliferating vessels, reactive astrocytes, and macrophages and forms a histological reaction to remove the necrotized tumor.4 Edema, and later demyelinization, occurs around the reactive ring, especially in the white matter. Fig. 3 shows CT-MR fused images where the inner dark gray areas of necrosis are surrounded by the white reactive ring and light gray edema.
The time interval between irradiation and occurrence of radionecrosis depends on the dose applied: higher doses cause earlier development of necrosis. According to Ishikawa et al.,3 necrosis occurred within 2 weeks after Gamma Knife irradiation of a rat’s brain with a dose of 200 Gy. With irradiation of 100 Gy, necrosis was observed within 10 weeks, and with 75 Gy, at 1 year after irradiation. Irradiation with a dose of 50 Gy or less did not produce necrosis within 1 year. In our study, tumor necrosis developed in the area where the radiation dose exceeded 79.2 Gy, on average. It reached the maximum at an average of 6 months after irradiation.
The dynamics of the reactive ring is almost identical to that of tumor necrosis. In a previous study, we examined the changing activity of the microglia-macrophage system in the reactive zone.4 Kumar et al.5 described a contrast-enhancing CT ring lesion at 2–6 months after high-dose-rate Co-60 treatment of glioblastomas (21 of 41 patients). Harisiadis et al.6 also observed an “enhancing rim” on CT images of irradiated gliomas.
The frequency,7 intracranial location,8 mechanism,9,10 and time evolution of the occurrence of edema after cerebral irradiation11,12 have been discussed by numerous authors. Helenowski et al.7 investigated 600 patients after linear accelerator irradiation (eight and nine isocenters were used; 65% of the patients had >13.5 cm3 tumor volume); edema, which could be balanced by steroids, was observed in 30% of the patients. In most cases, edema can be reduced by treatment with antiedema drugs (mannitol, dexamethasone) in our clinical practice in accordance with experimental data on nonhuman primates.13
Ganz et al.14 monitored 35 meningeomas treated with Gamma Knife surgery of 12 Gy and found that edema developed more frequently in patients with tumors adjacent to parasagittal regions than in those with tumors at the base of the skull.14 This observation has been confirmed by several other authors (for details, see Kalapurakal et al.15 and Singh et al.16).
Groothius et al.11 investigated the effect of 5–7 mCi I-125 seeds on dog brains over a 2-year period after irradiation. Damage to the blood-brain barrier was examined with C-14–labeled α-aminoisobutyric acid. Edema caused by lesions of blood-brain barrier increased 7–80 days after implantation, then equilibrated, and disappeared after 2 years. Nakata et al.12 examined the damage to the blood-brain barrier of rats after 20–40 Gy doses of irradiation with albumin extravasation. Edema reached its maximum on day 3 after irradiation and then began to decrease; by day 30, it was totally reabsorbed.
Edema is caused by damage to the blood-brain barrier, increased permeability of capillaries, and decreased regional cerebral blood flow. Thalacker et al.17 found the change of Hounsfield unit values in CT images to be a good indicator of brain edema.
In the case of brachytherapy of low-grade gliomas, recurrences occur generally after 2 years. Increases in volumes in the 6- to 8-month period after treatment are not likely to be the result of recurrence in our study. We are examining this issue with 11C-methionine PET and will present our results in the near future.
Image fusion after LDR I-125 brachytherapy is a new and important method for postoperative follow-up of patients. The predictability of the volumetric change in the regions of the triple ring, especially focused on edema, enables a more effective and accurate plan for patient treatment. Knowing the dynamics of the triple ring, physicians can adjust mannitol and steroid therapy to accommodate the development of edema. Consequently, patients can be effectively treated with smaller drug doses, thus reducing heavy side effects. The polynomial prediction approach we present here suggests that the best time for control CT-MR images is between 4 and 9 months after irradiation, and that reirradiation should not be carried out during this time period.
Our research is continuing, and after collecting more patient data, we will calculate and publish new prediction polynomials valid for longer time periods and with greater accuracy.
Although in this study we examined I-125 LDR brachytherapy, we propose the application of this method for Gamma Knife and linear accelerator treatments, as well.
This study was supported by the Hungarian Ministry of Health and Welfare (ETT grant 01/059).