Stereotactic Radiosurgery of the Postoperative Resection Cavity for Brain Metastases

 

Scott G. Soltys, M.D.   Department of Radiation Oncology, Stanford University Medical Center, Stanford, CA  IJROBP Volume 70, Issue 1, Pages 187-193 (1 January 2008)

The purpose of this study was to analyze results of adjuvant stereotactic radiosurgery (SRS) targeted at resection cavities of brain metastases without whole-brain irradiation (WBI). Patients who underwent SRS to the tumor bed, deferring WBI after resection of a brain metastasis, were retrospectively identified.

Seventy-two patients with 76 cavities treated from 1998 to 2006 met inclusion criteria. The SRS was delivered to a median marginal dose of 18.6 Gy (range, 15–30 Gy) targeting an average tumor volume of 9.8 cm3 (range, 0.1–66.8 cm3). With a median follow-up of 8.1 months (range, 0.1–80.5 months), 65 patients had follow-up imaging assessable for control analyses. Actuarial local control rates at 6 and 12 months were 88% and 79%, respectively. On univariate analysis, increasing values of conformality indices were the only treatment variables that correlated significantly with improved local control; local control was 100% for the least conformal quartile compared with 63% for the remaining quartiles. Target volume, dose, and number of sessions were not statistically significant.

In this retrospective series, SRS administered to the resection cavity of brain metastases resulted in a 79% local control rate at 12 months. This value compares favorably with historic results with observation alone (54%) and postoperative WBI (80–90%). Given the improved local control seen with less conformal plans, we recommend inclusion of a 2-mm margin around the resection cavity when using this technique.

Brain metastases are the most common form of brain tumor. Approximately 25% of patients with cancer develop intracranial metastases. The incidence appears to have increased during the past decade, perhaps because of increasing survival in patients with better systemic treatment. Improved survival of patients with intracranial metastatic disease may lead to a greater incidence of neurocognitive deficits for long-term survivors.

Treatment options for patients with brain metastases include medical management, surgical resection, whole-brain irradiation (WBI), and stereotactic radiosurgery (SRS). After resection of a brain metastasis, adjuvant radiotherapy in the form of WBI decreased the rate of local recurrence from 46% to 10% and elsewhere in the brain from 37% to 14%. Should immediate WBI be deferred, the option is retained to manage distant intracranial relapses with such repeated local modalities as SRS and surgery or with salvage WBI.

To minimize the potential late effects of WBI investigators explored the use of SRS alone, deferring the use of WBI for salvage treatment if needed. Both retrospective analyses  and prospective trials reported no apparent survival benefit in patients with brain metastases when WBI was combined with SRS as opposed to using SRS alone.

Given the lack of a survival benefit for using immediate WBI after resection or radiosurgery and the potential late effects of WBI, we generally deferred WBI in patients locally treated with limited number of brain metastases. However, recognizing the high local failure rate of surgical removal alone, we used adjuvant SRS to the resection cavity. The purpose of this retrospective review was to analyze the outcome in this patient population.

Radiosurgical technique

The CyberKnife Robotic Radiosurgical System (Accuray) was used to deliver radiosurgical treatments. In this frameless method, patients were immobilized supine on the CyberKnife treatment table with an Aquaplast mask (WFR/Aquaplast Corp., Wyckoff, NJ). A high-resolution thin-slice (1.25-mm) computed tomogram was obtained by using a GE Light Speed 8i or 16i Scanner (Milwaukee, WI) after administration of 125 ml of Omnipaque intravenous contrast (iohexol, 350 mg I/ml; Nycomed, Inc., Princeton, NJ). In selected patients, when resection cavity borders or other intact metastases could not be well visualized on computed tomographic (CT) imaging, a postcontrast stereotactic magnetic resonance imaging (MRI) scan also was obtained and fused to the stereotactic CT scan to improve target identification. The acquired scans were transferred by network to the CyberKnife treatment planning workstation.

The neurosurgeon, radiation oncologist, and radiation physicist performed tumor delineation, dose selection, and planning. Given the limitations of a retrospective analysis with multiple treating physicians, final target definition and dose selection were variable. Target volume was delineated as the edge of the resection cavity, including any residual tumor in cases of subtotal resection. In these cases, the surrounding intact brain parenchyma received a fall-off dose outside the prescription isodose line. As we gained additional experience, in a minority of cases (n = 10), the resection cavity volume was expanded with a 2-mm margin to define the final target volume. Prescribed dose was physician dependent, but based roughly on guidelines from Radiation Therapy Oncology Group 90-05, with multisession treatments primarily used for treatment volumes larger than 3-cm diameter targets.

Local and distant control

Overall, local failure was observed in 10 of 69 cavities. Of note, all failures were adjacent to the resection cavity and none occurred along the unirradiated surgical corridor leading to the resection cavity. Crude overall local tumor control rate was 86%. Kaplan-Meier local control rates at 6 and 12 months were 88% and 79%, respectively. Of 10 patients who experienced local failure, median time to progression was 5.7 months (range, 3.2–14.2 months). Two patients experienced local failure at an intact metastasis treated using primary SRS.

Using univariate analysis, conformality index and modified conformality index (by quartile) were the only significant predictors of local failure. Increases in absolute values of both conformality and modified conformality indices, which correspond to less conformal plans, resulted in improved local control. The rate of local control in the least conformal quartile was 100%, significantly different than the 64% rate of control observed with the 75% most conformal plans (p < 0.05). Although smaller tumor volume generally was associated with larger conformality indices, target volume was not associated with local control (p = 0.29).

Through analysis of the conformality index (prescription isodose volume/target isodose volume [PIV/TIV]), one can roughly estimate the additional volume of brain irradiated compared with the outlined target volume. If one assumes that the TIV and PIV correspond to idealized spheres (volume = 4/3 π radius3), the additional margin provided by the PIV covering the TIV can be calculated. Patients with a 2.4-mm margin of brain tissue included in the prescribed volume (the least conformal quartile) had 100% local control compared with 1.7-mm (78% local control), 1.5-mm (52% local control), and 0.8-mm (43% local control) margins in the remaining quartiles

The addition of a planned 2-mm margin on the resection cavity (n = 10) yielded local control of 100% compared with 70% (n = 59) for those treated without the planned addition of a margin. However, this difference was not significant (p = 0.15). No other factors, including tumor volume, dose, single-session equivalent dose, number of sessions, fusion of MRI for planning, histologic characteristics, extent of tumor resection, and number of metastases, proved significant on univariate analysis.

Distant failure occurred in 32 of 65 patients (49%). Kaplan-Meier 6- and 12-month distant control rates were 70% and 47%, respectively. Median time to distant failure was 10.2 months.

Discussion

In patients with limited brain metastases, WBI is associated with an acute detriment in quality-of-life measures potential delayed neurocognitive deficits, and lack of overall survival benefit. Consequently, an alternative to conventional cranial irradiation for patients who have undergone surgical resection of brain metastases remains desirable. Based on this rationale, with patients in whom surgical resection is preferred to obtain histologic confirmation or alleviate mass effect, we generally deferred immediate WBI for those with a limited number of brain metastases and a favorable KPS. In place of WBI, we used SRS as the primary treatment for unresected metastases or as adjuvant therapy to the postoperative resection cavity. To the best of our knowledge, this retrospective analysis is the first published report describing the use of postoperative radiosurgery to the surgical resection bed of brain metastases, thereby reserving WBI.

To justify the use of SRS alone to the resection cavity, local control rates should be similar to those reported for postoperative WBI. In this regard, Patchell  reported outcomes in 95 patients with a gross totally resected isolated metastasis who were randomly assigned to either no additional therapy or postoperative WBI. The WBI decreased the rate of local failure at the original tumor site from 46% to 10%. Of note, no difference in overall survival or functionally independent survival was seen with the addition of WBI; observed median survival was approximately 11 months. Moreover, an additional study by Patchell , in which patients with single metastases were randomly assigned to either WBI or surgery and WBI, reported a 20% rate of local failure in the postoperative WBI arm.

A similar retrospective series of postoperative radiosurgical boost to the resection cavity, reported in only abstract form, analyzed 61 patients treated to a median marginal dose of 16 Gy to a median target volume of 8.7 cm3. Local failure occurred in 30% of patients, with 1-year probability of 39%. Median survival time was 14.9 months.

Results of a Phase II trial investigating the use of GliaSite RTS (Cytyc Surgical Products II, Mountain View, CA) intracavitary low-dose rate brachytherapy for treatment of the postsurgical cavity for single metastases, deferring immediate WBI, recently were reported. Sixty-two patients underwent brachytherapy balloon catheter system placement at the time of surgical resection. During a median dwell time of 114 hours, 60 Gy was delivered at a 1-cm depth. Assessed by using MRI, the crude local control rate was 83%, with a predicted 1-year rate of 79%.

The local control rate in our series was 86% overall, with actuarial rates of 88% and 79% at 6 and 12 months, respectively. The 14% rate of local failure we observed when using SRS to treat the resection cavity compares favorably with the 46% recurrence rate described after surgical resection alone and 10%  to 20% in patients who underwent surgical resection and subsequent WBI. Moreover, the local relapse rate we observed is similar to that reported for resected brain metastases treated with GliaSite brachytherapy (17%)

When the present series was analyzed by means of univariate analysis, conformality indices were the only factors associated with local failure; a higher conformality index (i.e., a less conformal isodose volume) was associated with a decreased likelihood of local relapse. Although smaller target volumes tend to be associated with higher conformality indices, in the present series, no significant correlation could be discerned between the size of the resection cavity treated and rate of local control. Two possible explanations for this phenomenon are: (1) the difficulties encountered in accurately delineating the margins of a recent resection cavity on MRI and CT scans, or (2) in some cases, the possibility of an infiltrating radiographically invisible tumor margin. The end result is that an overly conformal margin in such cases appears to be counterproductive. In lieu of using less conformal plans, we recommend the addition of a margin that encompasses the cavity. Given the approximately 2 mm suggested through calculation of the idealized spherical margin of the PIV on the TIV based on conformality index, as well as a trend toward improved local control seen with the planned addition of a 2-mm margin, we suggest and now routinely add a 2-mm margin with this technique. Similar recommendations for the addition of a 1-mm margin for SRS treatment of intact metastases were made. Although inclusion of a margin may improve local control rates, it is notable that failure rates of 17% were seen with brachytherapy, even when the median balloon surface dose was 314 Gy, a dose expected to be adequate for treating malignant disease.

The actuarial distant intracranial failure rate in this series was 53% at 1 year, similar to other series: 50% for brachytherapy alone, 43–63% for SRS alone , and 50% for surgery alone (estimated from survival curve). A concern with the omission of immediate WBI in the treatment of patients with brain metastases is that a greater distant brain failure rate may lead to more deaths from neurologic progression of disease; Patchell  reported that the addition of WBI decreased the rate of death from neurologic causes from 44% to 14%. However, a recent randomized trial involving patients with unresected brain metastases showed no differences in neurologic function when combined WBI and SRS treatment was compared with SRS alone. At 1 year, neurologic death rates were 19% and 23%, respectively; these are similar to the 25% seen in our series. Meanwhile, other series reported lower neurologic death rates, with 13% (1-year actuarial) for SRS alone and 11% (crude) for brachytherapy. Given the limitations inherent to retrospective studies, it was not possible with the current series to formally measure the impact of tumor recurrence on quality of life.

As with any approach in which WBI is omitted, frequent surveillance imaging is required to detect distant occurrence of new metastases at a time when they are small and amenable to treatment. Studies reported that of 21% of patients with breast cancer who required repeated SRS, 12% and 50% experienced failure within 2 and 6 months, respectively. Therefore, in patients for whom such surveillance imaging is not possible, postoperative WBI is recommended.

Historically, the use of WBI after surgical resection of a brain metastasis was considered the standard of care. Although interpretation of available data regarding the omission of WBI is controversial, we routinely defer the use of immediate postsurgical WBI in patients with limited number of metastases and favorable KPS to minimize the acute toxicities and potential long-term neurocognitive dysfunction in these patients. In our series, SRS treatment of the postoperative resection cavity yields local control rates similar to previous reports investigating postoperative WBI. Less conformal treatment plans appear to produce greater rates of local control. As a surrogate, we recommend the addition of a 2-mm margin around the resection cavity when using the current technique.