 |
|
||||
| Home | Login | |||||
| About | Editorial board | Articles
|
NSI Publications
|
Search | Instructions | Online Submission | Subscribe | Videos | Etcetera | Contact |
|
||||||||||||||||||||
|
|
|
Dose fractionated gamma knife radiosurgery for large arteriovenous malformations on daily or alternate day schedule outside the linear quadratic model: Proof of concept and early results. A substitute to volume fractionation
Correspondence Address: Source of Support: None, Conflict of Interest: None DOI: 10.4103/neuroindia.NI_220_17
Keywords: Common terminology criteria for adverse events, radiotherapy, bevacizumab, arteriovenous malformation, nidus, linear quadratic model, stereotactic radiosurgery
The rationale for gamma knife radiosurgery (GKRS) has always been strong. Traditionally, GKRS has been considered as a single session treatment but with the passage of time, this horizon has also expanded. A larger volume of a lesion is usually considered a contraindication for GKRS. Management of large volume arteriovenous malformations (AVMs) has been challenging and controversial. Conventional radiation treatment has been associated with a poor efficacy for such lesions. The key determinant for a successful outcome after GKRS is the marginal dose and the nidus volume. Due to inherent limitations, the marginal dose needs to be reduced while treating large volume lesions to prevent toxicity to the surrounding brain parenchyma. This translates into poor obliteration/control rates in comparison to that seen in smaller volume lesions.[1],[2],[3] To achieve a good obliteration rate with less toxicity, various multimodal treatments have been tried such as pre- or post- GKRS embolization and volume-staged stereotactic radiosurgery (SRS). Dose fractionated SRS has been infrequently described with Cyberknife but not with GKRS. The treatment protocol, radiation dosage and number of fractionations remain controversial because of paucity of literature. The tool most commonly used for quantitative prediction of dose fractionation dependencies in radiotherapy is the mechanistically based linear-quadratic (LQ) model.[4],[5] The LQ model is an appropriate methodology for determining the iso-effective doses delivered using the dose per fraction lesser than or equal to 5Gy and collapses at >8 Gy.[4],[5],[6] However, the same is not applicable when fractionated treatment is considered with GKRS using 50% marginal isodose prescription. We describe our experience with 14 patients harbouring an AVM managed with dose fractionated GKRS (DFGKRS). Our protocol, early outcome and complications are also discussed. Our proposed aim in this article are: (a) to test the feasibility of administering DFGKRS for intracranial AVMs; (b) to evaluate the parameters affecting the complication profile of DFGKRS; and, (c) to define the short-term efficacy of DFGKRS.
Study design This was a prospective study approved by the Institute Ethics Committee. From 2011-2015, a total of 51 patients of various intracranial pathologies were treated with DFGKRS using Leksell Gamma Knife Perfexion machine (Elekta Instruments, Norcross, GA). These included lesions that were too large for single fraction radiosurgery. In this paper, the authors are restricting the focus to their experience with AVMs which were not suitable for surgery or embolization. Treatment protocol DFGKRS was delivered in 2-3 fractions according to the volume of the AVM. All patients with an AVM volume more than 10 cc were offered DSFGKRS with explained consent. A team of neurosurgeons and neuroradiologists confirmed the target volume. All the patients received fractionated doses of radiation as per the dosimetry plan [Table 1]. The other confounding variable under consideration was the proximity of the AVM to critical structures. Our practice of fractionation can be divided into two time periods: 2011-12 and 2012-16. From 2011-12, the total accumulated prescribed marginal dose for DFGKRS was 85% of the marginal dose of single fraction Gamma Knife Radiosurgery (GKRS); while, from 2012-16, the total accumulated marginal dose of DFGKRS was 75% of the single fraction GKRS marginal dose. All patients received injectable dexamethasone, 4 mg four times a day, that was continued for the duration of the treatment. Tablet dexamethasone was also continued after the procedure for 3-4 weeks. The patients were discharged after a day of observation following the completion of the procedure. The first follow up was done at 3 months and subsequently at 3-6 months. The clinical and radiological outcomes of all patients were evaluated up to the last available follow up. The radiological end point was nidus obliteration assessed on the follow up angiograms. The response was considered good, if the nidus was completely obliterated; moderate, if 50-99% of the volume of AVM was obliterated; and suboptimal, if <50% of the AVM volume was obliterated.[6] Every adverse event was evaluated as per the criteria given by the Common Terminology Criteria for Adverse Events (CTCAE).[7] Radiation induced complications were suspected if the neurological deterioration could not be explained by the occurrence of haemorrhage, the administration of another form of treatment, or the presence of concurrent or any other intercurrent pathology.
Gamma knife radiosurgery system On the first day of the procedure, Leksell G frame was fixed to the patient's head under local anesthesia. The patients were treated in 3 sessions on a daily, or 2 sessions on an alternate day basis, respectively, in the same hospital admission with the frame remaining in situ, firmly fixed to the head of the patient throughout the treatment. The treatment was given daily for three days in 8 patients or on alternate days in 6 patients. All attempts were made to start the next day's schedule as close to the earlier daytime, attempting a gap of 22-24 and 46-48 hours (on alternate day schedule) respectively, between the beginning of the two schedules. In the first two cases, daily imaging was performed and the images were co-registered with the images of the previous day treatment plan in Gamma Plan ™ treatment planning system to check for the fidelity of the images. This step was abandoned later as we felt that the repeat imaging was not contributing in improving the accuracy of the treatment. However, the patient was instructed that the frame should not get loosened during the course of the treatment. Treatment planning T1 weighted (T1W1) and T2 weighted (T2W1) images as well as contrast enhanced MRI (CEMRI), time of flight (TOF) and digital substraction angiography (DSA) imaging were performed in all the patients. Skull measurements were performed to generate the skull contour for the dose calculation in Gamma Plan Treatment Planning System. At the start of every session, measurements of skull frame were taken every day for the congruency check of the skull radii. The variations of skull radii were less than 2mm. The dose variations required due to the change of skull radii of 2 mm were less than 1%. The treatment plan generated during the first day was used for the treatment of successive fractions. Irrespective of the dose schedule, the volume of the brain exposed to a cumulative dose of 8 Gy was limited to less than or around 200 cc Patient selection for DFGKRS Patients harbouring an AVM larger than 10 cc volume, and in whom we felt that marginal dose of 21-22 Gy may cause toxicity to the adjacent eloquent area, were enrolled for DFGKRS. The target dose was 23-25 Gy of a single fraction. Only one patient had the AVM volume less than 10 cc but fractionated dosage was administered in him due to the proximity of the AVM to the internal capsule. Patients with AVM who presented without bleed have not been selected after 2013, following the publication of the Randomized Trial of Unruptured Brain Arteriovenous Malformations (ARUBA) trial.[8] Calculation of the marginal dose for dose prescription. The marginal dose for dose prescription was calculated using the Biological Effective Dose (BED) according to LQ model [8] as shown below BED = D {1 + d/(α/β)} BEDDFGKRS = DDFGKRS {1 + dDFGKRS/(α/β)} BEDDFGKRS = 75% or 85% of DGKRS (Where d denotes dose per fraction; n denotes fraction number; D = n × d; α and β are constants of proportionality of linear and quadratic components of LQ model respectively). This equation assumes that the effects of two-fractionation schemes remain equal if the BED of any fractionation schemes are equal. The α/β ratio for the AVM remains variable in various studies and has been mentioned to range from 3.5-6.4, but was arbitrarily taken by us as '3'.[9],[10] In theoretical studies concerning the radiobiology of radiosurgery, it has often been assumed that AVMs behave like late responding tissues with an α/β ratio of about 3 Gy. This is probably due to the fact that it is virtually impossible to derive these values from the dose–response curves of single-dose radiosurgery or from small, incomplete series of fractionated radiotherapy. It has been observed that small AVMs (<3 cms) have a higher α/β ratio (4-6-6.4 Gy) while larger volume AVMs have a lower α/β ratio.[9] Since our patients had significantly large volume AVMs (>10 cc), the traditional value of 3 was taken in the analysis. Initially from 2011 to 2012, 85% of the marginal dose for dose prescription in a single fraction GKRS was used for marginal prescription dose calculation of multiple fraction DFGKRS. This was done because the efficacy of the hypofractionated dose calculation using LQ model is higher than that of the single fraction SRS dose. Later on, from 2012 to 2014, 75% of the single fraction marginal dose of GKRS was used for the calculation of dose per fraction of multiple fraction DFGKRS. This was due to the fact that the 50% isodose curve was used for marginal dose prescription. The calculated dose per fraction for different fractionation schemes according to this correction are given in [Table 1]. Statistical analysis Statistical analysis was performed using Statistical Package for the Social Sciences (SPSS; version 21.0, IBM Corp.). Continuous variables that were considered nonparametric were evaluated as per the Shapiro-Wilk test and were reported as median with interquartile range (IQR). Categorical data were reported as counts and proportions in each group. The bivariate relationship between 2 continuous variables was assessed using the Spearman's correlation coefficient. Univariate analyses of continuous variables across binary categories were compared using the Mann-Whitney U-test. Two-sided significance tests were used throughout, and the significance level was kept at P< 0.05. Multivariate analysis on ordinal outcomes were conducted using PLUM (Polytomous Universal Model) ordinal regression, with mandatory significance of the model coefficient set at <0.05, after adjusting for various known prognostic factors such as age, AVM volume, number of fractions and volume of brain receiving 8 Gy radiation.
DFGKRS was given to 14 patients in 2 (n = 6 patients) or in 3 (n = 8 patients) fractions. The cumulative dose to the AVM varied from 45.2-69 Gy [Table 2]. None of the selected patients had received any previous intervention (e.g., embolization or down stage surgery) for the lesions.
Demographics Out of the 14 patients, 9 (64.3%) were male and 5 (35.7%) female. The mean age of the patients was 30.2 years (range 15-60 years). 12 out of 14 (86%) patients had radiological evidence of a prior bleed. 6 (43%) patients also had a history of seizures that was controlled on antiepileptic medication. The volume of the AVM ranged from 5.5-41.8 cc (mean 22.9 cc, median 26.5cc). In the three fraction scheme, the marginal dose ranged from 8.9 Gy- 11.5 Gy at 50% per fraction; while, in the two fraction scheme, it ranged from 11.3 Gy- 15 Gy at 50% per fraction. Follow up evaluation The clinical outcome was evaluated up to the latest possible time. All patients were prospectively followed up for 8-57 months (mean 35.6 months). One patient did not report after 3 years of follow up (the patient has been included in the toxicity profile but excluded from the efficacy evaluation). In the present study, we have evaluated the acute and long term toxicity of DFGKRS [Table 2]. The most common early postoperative complication was headache (noted in 6/14 [43%] patients) attributable to perilesional edema. This edema was controlled with regular doses of steroids, which could be tapered over a course of a few weeks. In 2 patients, edema could not be controlled on steroids and these patients were managed with bevacizumab. One patient presented with worsened headache and contralateral hemiparesis (4+/5), 3 months posttreatment, while another patient presented with behavioral changes. The MR scan showed significant perilesional edema, which was steroid resistant and was successfully managed with bevacizumab [Figure 1]. Nine patients had no neurological deficits before treatment, while 5 had a preexisting motor deficit. Twelve patients did not experience any deterioration after the treatment. The time interval between the treatment and the first documented neurological deterioration ranged from 3 months to 4 years. There was no correlation between the pre-procedural neurological deficit and the post-treatment headache. The initial two patients, who received total fractionated marginal dose of 30 Gy (2 fractions) and 34.5 Gy (3 fractions) had significant perilesional edema. Subsequently, the marginal dose was reduced from 85% to 75% of the total calculated dose. There was no correlation between the irradiated volume and the post-procedural neurological deficits.
Angiographic Follow up Follow up DSA at or after 3 years demonstrated complete nidus obliteration in 3 (21.4%) [Figure 1],[Figure 2],[Figure 3],[Figure 4] and, 50-99% obliteration in 4 (28.6%) patients. The remaining 7 (50%) patients with a follow up of ≤3 years had <50% obliteration, observed on the early digital subtraction angiogram (DSA). Overall, there was 67.8% reduction in the AVM volume with the median initial AVM volume being 26.5 cc. In this proof of concept study, however, where the emphasis was on demonstration or realization of a method or an idea, an undue emphasis on the actual results may not be warranted.[11]
The marginal dose to responders was 29Gy. Four patients showed an initial volume reduction for 2 years and then remained static for the next 2 years. The occlusion process was sustained in 11 patients on the subsequent follow up angiography. Changes in normal adjoining vasculature (radiographic arteritic changes) were observed in 8 patients. Among the 14 patients, only one patient developed a minor bleed (at 5 months posttreatment), which could be managed conservatively. Univariate analyses Nidus obliteration at 3 years showed a significant Spearman's rank-order correlation with the cumulative prescription dose (p 0.95, P value 0.01). Near-total (more than 95%) obliteration rates were attained beyond 29 Gy of the cumulative prescription dose [Figure 5]. The severity of adverse reaction (CTCAE criteria) also showed significant rank order correlation with the cumulative prescription dose (p 0.66, P value 0.01) [Figure 6]. No patient receiving a cumulative prescription dose less than 31 Gy had any severe adverse reaction. However, those with severe adverse reactions had received a significantly higher median cumulative prescription dose of 32.75 (interquartile range [IQR]: 31-34.5) Gy compared with 27 (24.5-29) Gy given to others (Mann-Whitney P value 0.02) [Figure 7]. The three dimensional surface plot depicting the relationship among the cumulative prescription dose, the extent of nidus obliteration and the CTCAE adverse events, suggests that a cumulative prescription dose of 29-30 Gy results in a near-total nidus obliteration with least adverse events [Figure 8]. The volume of the surrounding brain that had received atleast 8 Gy of radiation was also significantly greater among those patients with a severe adverse reaction (CTCAE 3-4) compared to the group with mild-to-moderate side effects (CTCAE 1-2) [median 258.9 vs. 109 cc; P value 0.02] [Figure 9]. The 6,10 and 12 Gy volumes have also been depicted in [Table 2].
Multivariate analysis In co-variate adjusted Polytomous Universal Model (PLUM) ordinal regression, only the cumulative prescription dose had a significant correlation with the CTCAE severity (P value 0.04), independent of age, AVM volume, number of fractions and volume of brain receiving atleast 8 Gy of radiation.
Patients with large or SM grade 4,5 AVMs pose a therapeutic challenge especially when they present with bleed. Most of them are not operable and are even not suitable for embolization when en-passage vessels share their blood supply with the adjoining neural tissue.[12],[13],[14] This necessitates unconventional approaches in managing these lesions. By definition, 'stereotactic radiosurgery' is typically performed in a single session, but can be performed in up to a maximum of five sessions.[15] The strength of GKRS lies in its precision and sharp dose fall out, which makes it a better alternative than linear accelerator based radiosurgery or fractionated radiotherapy.[16],[17] The most important variables affecting the effectiveness of SRS are the marginal dose and prescription isodose volume (PIV). As the PIV increases, the exposure of normal brain to radiation also increases because the dose gradient is less steep when large volumes are covered. To maintain a steep dose fall off while using a focused technique, SRS has been traditionally limited to small focal targets of, approximately, less than 10 cc volume.[2],[18],[19],[20] This limitation has prevented the inclusion of several pathologies for GKRS treatment. Fractionated radiosurgery is considered a compromise offering the efficacy of SRS and the safety of radiotherapy. To overcome this limitation, several attempts have been made in the recent past with fractionated radiosurgery.[21],[22],[23],[24],[25] Few papers have addressed the issue of an ideal marginal dose (23-25 Gy) prescription with ≥5 Gy dose being administered to large volume lesions on a daily schedule instead of an interval treatment of 3-6 months.[26] Dose volume calculations Linear Quadratic (LQ) equation is the most commonly used model for quantitative prediction of dose/fractionation dependencies in radiotherapy. LQ model has useful properties for predicting the iso-effective doses, which supports the basic purpose of GKRS. LQ model precisely describes the cell killing, both for tumor control and for normal tissue complications. According to the LQ model, a lethal radiation dose produces a double stranded DNA breakdown at the same time as a misjoining to produce a lethal lesion (e.g., formation of a dicentric chromosome). However, the LQ model tends to fail beyond the 5Gy single fraction and collapses at >8 Gy.[4],[5],[27] Optimal dose and fractions The purpose of DFGKRS is to deliver an ideal dose to a larger volume than is permitted by the conventional means of radiation administration, while keeping the normal tissue damage to acceptable levels. As a general radiotherapy principle, to produce the same dose effect, using fractionation, a greater total dose is required than would be needed if all of the radiation dosage were given in a single session. Dose fractionated GKRS is a hybrid of neurosurgical and radiotherapeutic principles, that, in a practice similar to the conventional radiation treatments, involves radiation therapy given daily. In addition, in a practise similar to radiosurgery, it also permits relatively sharp dose gradients to be administered to smaller volumes of tissue. Fractionation may permit delivery of a higher dose of radiation that in turn encompasses a larger margin of normal tissue within the treatment volume than is characteristic of SRS. Fractionation can be achieved in either volume or dose parameters. In view of limited literature and no documentation of long-term follow up, there are no specific guidelines for the optimal doses and fractions to be administered in DFGKRS. Dose staging has been described either as a hypofractionated stereototactic radiotherapy (HSRT)[13] or as a repeat SRS.[21],[22],[23],[24],[25],[26],[27] In HSRT, several small radiation doses are administered over a period of a few weeks. However, in repeat radiosurgery, a higher initial dose followed by another dose is administered to the same or different volume after a variable period of several months to years. Fractionated SRS has been shown to be as effective as a single session SRS with reduced complication rates. Most of the dose fractionation has been performed with Cyberknife while volume staged fractionation is achieved by GKRS.[1],[21],[22],[23],[24],[25],[26],[27] As the 50% isodose curve is used for marginal dose prescription in GKRS, a large dose variation may be seen inside the prescription dose isovolume. The same dose fractionation and dose per fraction used in Cyberknife is also not applicable to DFGKRS. In our treatment protocol, the LQ model was not applicable for DFGKRS. In a study by Iwata et al., the efficacy of hypofractionation is compared with the conventional single session stereotactic radiosurgery. The actual efficacy of hypofractionation has been considered to be higher by about 15%. In vivo, a higher effect of hypofractionation has been expected, especially when reoxygenation of the hypoxic cells takes place.[28],[29] In view of this concept, we used the marginal dose of DFGKRS plan, calculated using the BED of 15-25% lesser marginal dose of a single session GKRS plan. The time interval between different sessions of DFGKRS is a debatable issue. Jee et al., have kept a 12 hour interval between the sessions for DFGKRS in the treatment of perioptic benign tumors.[30] One of the radiobiological principles, “repair” of a normal tissue, requires at least a gap of 12 hours in between treatment sessions for usual doses. We arbitrarily took an interval of 24-48 hours as we were using a higher dose schedule. In our experience, GKRS offers a lower morbidity and a shorter hospitalization time in addition to being non-invasive and cost effective (as the patient has to pay for the entire treatment only once). Complication rates The side effects of stereotactic radiosurgery can be summarized under three broad categories i.e., acute (during the course of treatment), early delayed (weeks to months after treatment) and late delayed (months to years after treatment). The acute complications are defined as new focal deficits, exacerbation of existing deficits, or the precipitation of a seizure or a decline in the mental status occurring within 7 days after treatment with GKRS. The early complications are usually due to radiation induced edema while delayed complications are due to radiation induced vascular injury leading to radiation necrosis. Theoretically, DFGKRS may cause fewer complications as compared to the conventional GKRS. [Table 2] shows the representative acute complication rates for DFGKRS. In our study, all the targeted volumes were 26.5cc (median volume) in size that would not be normally treated with a single session GKRS, or there was a limitation to the administration of a dosage because of a nearby eloquent structure. Despite these limitations, the main acute complication that has been encountered so far has been the occurrence of edema leading to headache and occasional vomiting, which could be easily controlled with peri-procedural steroids or bevacizumab. In their long term follow up of 28 patients having a large volume AVM, who had received a total of 41-50 Gy irradiation in a fractionated radiotherapy treatment plan, Karlsson et al., have found obliteration in only 8% patients but significant neurological deterioration in rest of the patients. Fractionated radiotherapy did not protect from rebleed in the intermediate period, until the nidus got completely obliterated.[6] In view of the significant collateral damage, fractionated radiotherapy is not recommended for large volume AVMs. Similar to the findings of this report, Panet al., reported an obliteration rate of only 25% at a 40 month follow up in patients with a large volume AVM of size >15 cc managed with a single session SRS with a dose of <17.5 Gy.[23] Volume staged radiosurgery has been an attractive, albeit, a non-standard treatment option for such large volume AVMs in which an AVM is divided into separate anatomic volumes for the administration of treatment in stages. The treatment duration is separated at intervals of 3-12 months with the overall angiographic occlusion rates in the range of 40-61%.[22] However, with the upgraded Gamma Plan 10 software and the feasibility of frameless radiosurgery with the ICON model, this approach can be easily practiced. It allows for an actual overlay of the previous stages of treatment on top of new radioimages for better comparison and nidus definition. Limitations Fractionated radiosurgery for benign intracranial lesions is a controversial subject, the benefits and shortcomings of which will require judgment over a long period of time. The benign nature of the lesions and the greater long-term survival of such patients may demonstrate a higher incidence of long-term complications, which is not an issue with those patients harbouring malignant lesions of the brain. Both the efficacy as well as the adverse events are dependent on the radiation dosage. This complex neurosurgical quandary mandates a balanced analysis to arrive at the optimal therapeutic window. The purpose of this paper is not to present the obliteration rate, as assessment of the latter requires a long term follow up. In a single session GKRS, we cannot treat larger volume lesions. However, with DFGKRS, we could treat larger volume lesions with comparable outcomes that were obtained following the administration of single session-GKRS for smaller volume lesions, with an acceptable toxicity rate. In a single hospital admission, a common argument against the use of DFGKRS is the inconvenience of stereotactic frame fixation. But, in our experience, the in situ frame for 3-4 days is well tolerated by the patients. By analogy, it can be compared to a halo fixation for various cervical spine trauma patients who tolerate the frames for 3-4 months on an average. With the rigid frame in situ, inter-fractionation displacement errors are minimized and higher cumulative doses can be delivered keeping the marginal dose at the 50% isodose line. Elekta had further proposed the frameless “Extend” machine to promote various fractionated treatments. This has been, however, withdrawn from the market in view of the discomfort encountered with vacuum-based jaw-block system. The ICON model has recently been launched to promote a frameless fractionated radiosurgery. Though, we believe that frame-based fractionation is likely to be more precise but the discomfort of pin fixation will reduce with the advent of the ICON model and it may broaden the applications for DFGKRS. Only time will tell whether a frame based fractionation or a frameless technology is better. In our series, another limitation was absence of any candidate with a history of prior embolization. Pre GKRS embolization may lead to a false localization, may obscure the targeting and may lead to a lower rate of successful nidus obliteration.
DFGKRS is feasible for large AVMs with a fair nidus obliteration rate and acceptable toxicity. Cumulative prescription dose seems to be the most significant independent predictor responsible for the efficacy of DFGKRS. Hence, we recommend that a cumulative prescription dose of 29-30 Gy facilitates a near-total nidus obliteration with the least adverse events. We reiterate that the results of this study are a preliminary proof of concept. Acute complications following DFGKRS seem to be acceptable but there is a definite risk of developing new neurological deficits, which is treatable with medications. The outcome at three years is also encouraging. The long-term outcome of these patients will require further follow up. Acknowledgement We would like to acknowledge Professor SC Sharma (Former Head of Dept. of Radiotherapy, PGIMER, Chandigarh) for his valuable support for the project. Financial support and sponsorship Nil. Conflicts of interest There are no conflicts of interest.
[Figure 1], [Figure 2], [Figure 3], [Figure 4], [Figure 5], [Figure 6], [Figure 7], [Figure 8], [Figure 9]
[Table 1], [Table 2]
|
|||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
|||||
