Based on the results of surgical resection for a single metastasis, investigators explored the use of SRS for brain metastases. This interest was based on potential advantages of SRS, including outpatient delivery, no general anesthesia use leading to shorter recovery time, minimal risk for bleeding or infection, and lower cost. In addition, SRS is not limited by location, number of lesions, comorbidity, or coagulopathy. For these various reasons, SRS has become a viable alternative to surgery given the potential cost savings and quality-of-life benefits [43, 44].

A multi-institutional retrospective review from the University of Wisconsin, University of Florida, and Joint Center for Radiation Therapy included 122 patients with a newly diagnosed, potentially resectable single brain metastasis treated by SRS with WBRT [45]. The local control rate was 86 %. The median survival was 56 weeks with 25 % of deaths related to brain tumor progression. Based on results of this retrospective review, the investigators concluded that SRS could be delivered in place of surgery.

Another study from M.D. Anderson Cancer Center compared 13 patients treated with SRS and 62 patients treated by surgery [46]. These patients were retrospectively matched for important prognostic factors. The median survival was 7.5 month for the SRS group and 16.4 months for the surgery group. The local recurrence for SRS was 21 % vs. 8 % for surgery. They concluded surgery should be the preferred treatment for surgically accessible lesions. However, careful analysis of the SRS dosing strategy in this study indicates that relatively low tumor margin doses were used, well below the doses used in RTOG studies. A study from the Mayo Clinic compared outcomes of radiosurgery vs. surgery and found no difference in 1-year survival between the groups [47]. This retrospective review included patients with tumors who would have allowed for either type of treatment.

A retrospective study from Muacevic compared surgery + WBRT vs. Gamma Knife radiosurgery (GKRS) alone for patients with a single tumor ≤3.5 cm in diameter [48]. The 1-year survival rates (53 % vs. 43 %, p = 0.19), 1-year local control rates (75 % vs. 83 %, p = 0.49), and 1-year neurologic death rates (37 % vs. 39 %, p = 0.8) for the surgery + WBRT group and SRS group, respectively, were not statistically different.

Since all of the previously described studies are retrospective, patient selection bias was probably a key contributor to the mixed results and controversy regarding the benefits of surgery vs. SRS. The Joint Center for Radiation Therapy attempted to enroll patients onto a phase III trial comparing surgery to SRS in the 1990s. After 3 years, only six patients were enrolled due to patient or physician preference [49].

A prospective randomized trial from Germany compared GKRS alone vs. surgery + WBRT for patients with a single metastasis [50]. The study was closed early due to poor accrual, with 33 patients in the surgery + WBRT arm and 31 patients in the GKRS alone arm. Overall survival, neurologic death, and local recurrence were equivalent. Patients treated with GKRS developed more distant recurrences (p = 0.04); however, this difference was lost after adjusting for salvage GKRS. GKRS patients had shorter hospital stay, less steroid treatment, and lower CNS Grade 1 and 2 toxicities.

M.D. Anderson performed a prospective randomized trial of surgery vs. radiosurgery for patients with a single brain metastasis who are considered eligible for either treatment [51]. Fifty-nine patients were entered into the randomized arm (30 underwent surgery and 29 SRS), and 155 patients were entered on the nonrandomized arm (89 chose surgery and 66 chose SRS). Although the results demonstrating increased local recurrence with SRS alone have been presented in abstract form, the statistical manipulations used in this study are controversial and the results are yet to be published.

For patients with large resectable tumors or those without histologic confirmation of a primary tumor, most clinicians would agree that surgical resection should be considered rather than radiosurgery. Surgical removal of the tumor is more effective in alleviating mass effect, offers faster improvement in neurologic function, and minimizes chronic steroid therapy. Prolonged use of steroids can result in a number of problems including diabetes, proximal muscle weakness, peripheral edema, psychosis, susceptibility to infections, and gastrointestinal perforation [52]. Conversely, most would agree that small, deep lesions are best treated with radiosurgery. To date, no randomized trial pitting radiosurgery vs. surgery of comparable brain metastases has been completed.

SRS for brain metastases was initially implemented in the salvage setting. One of the first reports of SRS for brain metastases consisted of patients with recurrent brain metastases after WBRT or surgery. Loeffler et al. reported the Joint Center for Radiation Therapy retrospective results of 18 patients with recurrent metastases after previous surgery, WBRT, or both [53]. The majority of patients improved neurologically after SRS with local control being achieved in all patients. A follow-up report by Alexander et al. of patients mostly with recurrent brain metastases reported tumor control rates of 85 % at 1 year and 65 % at 2 years, although the control rates were lower for recurrent lesions [54]. Median survival was 9 months after SRS in that study.

The University of Cincinnati reported the results of 84 patients with 1–6 lesions who developed recurrence after WBRT [55]. Median survival was 43 weeks after SRS. Median time to local failure was 35 weeks for all lesions and 52 weeks for lesions treated with ≥18 Gy. A study by Noël et al. of 54 consecutively treated patients reported 1- and 2-year local control rates of 91.3 % and 84 % and 1- and 2-year brain control rates of 65 % and 57 %, respectively [56]. Cleveland Clinic published their experience of 111 patients who underwent initial WBRT followed by SRS as salvage treatment [57]. The median overall survival from the initial diagnosis of brain metastasis was 17.7 months. Median survival after salvage SRS was 9.9 months. Median survival after salvage SRS was 12.3 months in patients who had their first recurrence >6 months after WBRT vs. 6.8 months for those who developed disease recurrence ≤6 months after (p = 0.0061). Primary site did not affect survival. Twenty-eight patients (25 %) developed local recurrence after salvage SRS with a median time of 5.2 months. A dose <22 Gy and lesion size >2 cm were found to be predictive of local failure. In this study, patients who recurred after WBRT and were treated with salvage SRS were found to have good local control and survival after SRS. WBRT provided good initial control, as 45 % of these patients failed >6 months after WBRT. Those with a longer time to failure after WBRT had significantly longer survival after SRS.

Based on the early results of SRS for recurrent brain metastasis, SRS with whole brain radiation therapy was investigated for newly diagnosed patients with the hypothesis that the combination of WBRT with SRS would improve local and regional control of brain metastases. A number of studies have demonstrated high local control rates with the addition of WBRT to SRS [54, 58–66]. These local control rates ranged from 63 to 97 % with addition of WBRT vs. 42 to 87 % with SRS alone [36, 64, 66–68].

RTOG 9005 is an important phase I/II trial which determined the maximum tolerated radiosurgery dose for patients with recurrent primary brain tumors or brain metastases treated previously with fractionated radiation therapy [69]. In this trial, 80 % of patients were treated on a linear accelerator (LINAC) and 20 % were treated with Gamma Knife radiosurgery. The maximum tolerated doses were inversely correlated with the maximum tumor diameter. The doses were 24 Gy for a tumor <20 mm in diameter, 18 Gy for 21–30 mm, and 15 Gy for tumors 31–40 mm in diameter (Table 14.7). Of note, investigators were reluctant to escalate the dose for tumors <20 mm in diameter even though the maximum tolerated dose was not reached. The rates of radiation necrosis were 5 %, 8 %, 9 %, and 11 % at 6, 12, 18, and 24 months after SRS, respectively. Other factors that predicted for Grade 3–5 neurotoxicity were tumor dose and KPS. Table 14.8 lists the RTOG CNS toxicity criteria used for RTOG 9005. The study also reported on physics and quality control assessments: MD/PD ratio, a measure of dose homogeneity, and PIV/TV ratio, a measure of conformity of the treated volume relative to target volume [70]. The results of RTOG 9005 are the standard for determining dosing of SRS for brain metastases based on size and have been incorporated into subsequent studies.

Shehata et al. evaluated the optimal SRS dose and influence of WBRT on tumor control among 160 patients with 468 recurrent and newly diagnosed metastases <2 cm in diameter [66]. On multivariate analysis, the most important factor for local tumor control was the addition of WBRT to SRS (97 % vs. 87 % for SRS alone; p = 0.001). For patients undergoing WBRT, doses >20 Gy resulted in higher Grade 3 and 4 neurotoxicity (5.9 % vs. 1.9 %) and did not result in better local control. Thus, the authors recommended using 20 Gy for metastases <2 cm with combined WBRT.

The largest SRS dosing study is a single institution analysis reported by Valery et al. on 377 patients with 760 lesions treated with LINAC-based SRS. Seven patients had severe complications including nine patients who developed radiation necrosis. The median tumor volume was 4.9 cm³, and median prescribed tumor dose was 15.6 Gy. The only factor that influenced the risk for radiation necrosis was the conformality index [71]. Age, KPS, homogeneity index, tumor volume, and delivered dose were not significant. This study highlights the importance of achieving a highly conformal plan for LINAC-based radiosurgery.

A multi-institutional retrospective study from ten institutions analyzed outcomes for 502 patients with newly diagnosed brain metastases treated by SRS and WBRT. Patients were stratified by RTOG RPA class [72]. For all RPA classes, survival improved with the addition of SRS boost compared with WBRT alone. Median survival for RTOG RPA class I, II, and III were 16, 10, and 8 months for the SRS + WBRT patients vs. 7.1, 4.2, and 2.3 months for the WBRT alone patients, respectively. This study suggests a survival benefit to the addition of SRS for all patients.

Three prospective trials compared the use of SRS + WBRT vs. WBRT alone. The first trial from the University of Pittsburgh enrolled 27 patients with two to four brain metastases measuring <2.5 cm in maximum diameter [73]. The study was stopped at an interim analysis at 60 % since a significant difference in the 1-year local failure rate was noted (8 % in the SRS + WBRT arm vs. 100 % in the WBRT arm). The median time to local failure was 6 months for WBRT vs. 36 months for the SRS + WBRT arm. Median survival was not significantly different for the SRS + WBRT vs. WBRT arm (11 months vs. 7.5 months, respectively). The small sample size, as well as possible selection and statistical bias, may have affected the results.

Brown University enrolled 96 patients onto a 3-armed, prospective randomized trial of Gamma Knife radiosurgery, WBRT, or both [74]. Fifty-one patients underwent surgical resection of large symptomatic lesions prior to randomization, which was not evenly distributed among the treatment arms. No difference in overall survival was noted among the three study arms (7, 5, and 9 months for the SRS alone, WBRT + SRS, and WBRT arms, respectively). The local control rates were 87 %, 91 %, and 62 %, respectively. The risk for developing new brain lesions was higher for the patients who did not receive WBRT (43 % SRS alone vs. 19 % WBRT + SRS, vs. 23 % WBRT alone). However, the study had major treatment bias given the unequal distribution of patients undergoing surgical resection in the three arms.

The RTOG performed a large prospective trial (RTOG 9508) of 333 patients with 1–3 newly diagnosed brain metastases on MRI, each <4 cm in diameter, randomized to WBRT (3,750 cGy in 15 fractions) and SRS vs. WBRT alone [75]. Patients had a minimum KPS of 70 and were excluded if a metastasis was located in the brainstem or within 1 cm from the optic nerves or chiasm. Survival significantly improved for patients with a single brain metastasis treated with SRS + WBRT vs. WBRT (6.5 months vs. 4.9 months, p = 0.04) despite the fact that 19 % of patients failed to receive the intended treatment. On multivariate analysis, RPA class (1 vs. 2) and tumor type (squamous or non-small cell vs. other) had a statistically significant effect on survival. Overall survival did not improve for the patients with 2–3 lesions although KPS at 6 months had stabilized or improved with the addition of SRS + WBRT vs. WBRT alone (43 % vs. 27 %, respectively; p = 0.03). Steroid use was also lower in the SRS + WBRT arm. Local control was improved at 1 year (82 % vs. 71 %; p = 0.01). The risk of developing local recurrence was 43 % greater with WBRT alone (p = 0.0021). The type of SRS unit did not influence results.

A retrospective study from Cleveland Clinic of 202 patients that met study entry criteria of RTOG 9005 analyzed local control rates based on differences in prescription dose [76]. With a median follow-up duration of 10.7 months, a dose of 24 Gy to the tumor margin had significantly lower local failure than 18 Gy or 15 Gy (hazard ratio 0.277, p = 0.0005). The 18 and 15 Gy groups exhibited no statistical difference in local control. The 1-year local control rates by dose were 85 % for 24 Gy, 49 % for 18 Gy, and 45 % for 15 Gy. The improved control rates with 24 Gy suggests that small tumors ≤2 cm are better controlled than larger tumors, as one would expect.

Despite evidence supporting the use of WBRT for patients with single brain metastasis, SRS alone for treatment of brain metastases has gained popularity and become a contentious issue. The use of SRS alone has been mostly driven by concerns regarding quality-of-life issues and the potential side effects of WBRT, especially cognitive effects for long-term survivors [77, 78]. One of the early and commonly referenced papers regarding cognitive decline in WBRT is by DeAngelis et al., which reported complications in patients treated with fractionation schemes of 300–600 cGy/fraction to a total dose of 25–30 Gy [77]. Of these 12 patients identified, all developed progressive dementia, ataxia, and urinary incontinence, causing death in seven patients. Although the incidence of WBRT-induced dementia was 1.9–5.1 % overall, it is likely that this is an underestimate of true incidence because only severely affected patients were identified from chart review. Another argument to treating with SRS alone is that some physicians believe that patients can be effectively salvaged by repeat SRS rather than WBRT [79].

It is important to note that patients undergoing SRS alone may be at higher risk for morbidity associated with brain tumor recurrence. A retrospective study from the University of Kentucky reviewed 36 patients treated with SRS alone with newly diagnosed brain metastases of which 22 patients had a single lesion [80]. Seventeen of the 36 patients (44 %) developed recurrent brain metastases, with 12 patients (71 %) symptomatic and 10 (59 %) having neurologic deficits.

Several prospective trials have assessed whether patients with brain metastases have neurocognitive deficits prior to initiation of treatment. A recent prospective phase III trial evaluated patients with brain metastasis with monthly neurocognitive testing consisting of memory, fine motor speed, executive function, and global neurocognitive testing [81]. Table 14.9 lists the neurocognitive tests that were performed on this study. This trial demonstrated that 90.5 % of patients had impairment of one or more neurocognitive tests at baseline and 42.4 % of patients had impairment in four or more tests. Another study from the RTOG demonstrated that tumor progression resulted in a significant decline in the mini-mental status exam (MMSE) after WBRT [82].

Chang et al. reported a prospective trial of neurocognitive testing for patients with 1–3 newly diagnosed brain metastases to determine if neurocognitive function was spared with SRS alone [83]. Eighteen of the 27 enrolled patients had evaluable neurocognitive function data. At baseline, 66 % of patients had some degree of impairment that was related to total brain metastases volume. After SRS, acute and subacute improvements in executive function and processing speed occurred, but learning and memory skills declined. At 7–9 months after SRS, there was a suggestion of global decline of neurocognitive function.

The RTOG reported on the feasibility of performing five neurocognitive measures and administering a quality-of-life instrument in patients with brain metastases [84]. The overall compliance rate for administration and completion of these measures and instruments was >95 % prior to treatment, >84 % at the completion of WBRT, and >70 % at 1 month after WBRT. Based on the encouraging results of this study and others of similar design, future studies have more routinely incorporate baseline and follow-up neuropsychometric testing as an inherent part of study design and patient outcome evaluation. It is becoming increasingly evident that neurologic/neurocognitive decline seen in cancer patients is multifactorial and not exclusive to the effects of radiation therapy alone. This has been demonstrated in recent publications implicating the potential significant impact of other treatment modalities such as chemotherapy and/or surgery on neurologic/neurocognitive function [85–88].

A UCSF retrospective study by Sneed et al. reported outcomes of patients with one to four metastases treated by SRS alone compared to WBRT plus SRS [68]. Physician preference and referral patterns influenced the use of upfront WBRT. The patients had similar median survivals (11.1 months for SRS alone vs. 11.3 months for WBRT + SRS) and local tumor control rates (71 vs. 79 %). The incidence of distant brain metastasis, however, was significantly higher in the SRS only group vs. WBRT + SRS (72 vs. 31 %). Despite the higher rate of distant brain metastases, these lesions were controlled with salvage therapies including WBRT, partial brain radiotherapy, SRS, and surgery. The authors concluded that survival was not compromised by the omission of WBRT.

Another retrospective paper from Hasegawa et al. reviewed 172 patients with brain metastases (3.5 cm or less in diameter) managed by SRS alone [89]. One hundred twenty-one patients had follow-up imaging with 80 % of patients having solitary lesions. The median survival was 8 months, and the local tumor control rate was 87 %. At 2 years, the local control rate, distant brain control, and total intracranial control rates were 75 %, 41 %, and 27 %, respectively. Tumor volume significantly predicted for local control (p = 0.02) with tumor volumes of at least 4 cm³ having local control rates of 49 % at 1 and 2 years. Based on these results, the authors advocated withholding WBRT for select patients with one or two tumors with good control of their primary cancer, better KPS (90 or higher), and younger age (<60 years old).

One study prospectively evaluated the role of SRS alone for patients with up to three brain metastases measuring <3 cm in maximum diameter [90]. The study consisted of 101 patients with a total of 155 lesions. At 1 year, the local control, distant brain freedom from progression, and overall brain freedom from progression rates were 91 %, 53 %, and 51 %, respectively. The best predictor for improved overall brain freedom from progression was the presence of a single lesion and a greater than 2-year interval from primary diagnosis to diagnosis of brain metastases. The overall survivals stratified by RTOG RPA class were 13.4, 9.3, and 1.5 months for class I, II, and III, respectively (p < 0.0001). Median survival was 7.6 months. These results suggested that SRS alone can provide good local control but that distant failure can be expected.

Another study from Pirzkall et al. reviewed an experience of 236 patients with 311 brain metastases who underwent SRS alone (158 patients) or SRS with WBRT (73 patients) [67]. No difference was noted in terms of survival or local control. The SRS with WBRT group had a trend toward better survival compared with SRS alone (15.4 vs. 8.3 months; p = 0.08) for the patients with no evidence of extracranial disease. The local control and distant brain control rates were worse with SRS alone.

A multi-institutional retrospective review from ten institutions analyzed the results of 569 patients with newly diagnosed single or multiple brain metastases treated initially with SRS alone vs. SRS and WBRT [91]. For all RPA classes, survival was comparable between the two treatment groups (RPA class I: 14 vs. 15 months, RPA class II: 8 vs. 7 months, RPA class III: 5 vs. 6 months) suggesting that upfront WBRT does not improve survival over SRS alone.

The Eastern Cooperative Oncology Group (ECOG) phase II trial of SRS alone for patients with renal cell carcinoma, sarcoma, and melanoma (radioresistant tumors) with one to three brain metastases was conducted with a primary endpoint of intracranial progression [92]. The purpose of this trial was to evaluate the feasibility of SRS alone in management of brain metastases before pursuing a phase III study. This trial enrolled 36 patients; 31 were eligible by entry criteria. The SRS dose was based on tumor size (24, 18, or 15 Gy). The median survival was 8.3 months with a median follow-up of 32.7 months. The 3- and 6-month intracranial failures with SRS alone were 25.8 % and 48.3 %, respectively. Approximately 38 % of patients experienced death attributable to neurologic cause. The authors concluded that although survival was equivalent to published reports of SRS + WBRT, the high rate of intracranial failures suggests a judicial approach to omission of WBRT.

The Japanese Radiation Oncology Study Group (JROSG) conducted a phase III study of SRS vs. SRS plus WBRT (3,000 cGy/10 fractions) [93]. This trial randomized patients with one to four brain metastases <3 cm in diameter to SRS + WBRT (65 patients) or SRS alone (67 patients). The primary endpoint of the study was overall survival with the secondary points including brain recurrence, salvage brain treatment, functional preservation, toxic effects of radiation, and cause of death. At 12 months, WBRT decreased brain tumor recurrence from 76.4 to 46.8 % (p < 0.001). Freedom from new brain metastasis was also significantly better at 1 year for the WBRT and SRS group (63.7 %) compared with the SRS alone group (41.5 %) (p = 0.003). Consequently, salvage treatment was more frequently required in the SRS alone group. No significant difference in overall survival, toxic effects of radiation, death due to neurologic causes, or differences in systemic or neurologic functional preservation were seen.

A prospective randomized trial for patients with 1–3 lesions comparing SRS plus WBRT vs. SRS alone with the primary endpoint of neurocognitive function measured by the Hopkins Verbal Learning Test-Revised (HVLT-R) was conducted by Chang et al. at MD Anderson [94]. After accrual of 58 patients, the trial was stopped early based on the 96 % probability that the SRS + WBRT arm would show a statistically significant decline in learning and memory function (total recall) at 4 months. Congruous to the Japanese study, there were more CNS recurrences in the group receiving SRS alone; 73 % of patients in the SRS + WBRT group were free from CNS recurrence at 1 year, compared with 27 % of patients who received SRS alone (p = 0.0003). The authors conclude that SRS with close follow-up is the preferred strategy. However, this study did not provide information regarding long-term impact of WBRT or distant tumor progression on neurocognitive function.

The EORTC completed a prospective phase III trial comparing the addition of WBRT to initial surgery or SRS for patients with stable systemic disease and up to three brain metastases [95]. The primary endpoint of this study was time to WHO performance status >2. There were no differences in overall survival or time to performance status deterioration in either group. WBRT reduced the 2-year relapse rate at the initial tumor site (surgery: 59 % vs. 27 %, p < 0.001; SRS: 31 % vs. 19 %, p = 0.04) and reduced new brain mets (surgery: 42 % vs. 23 %, p = 0.008; SRS: 48 % vs. 33 %, p = 0.023) in both groups. Furthermore, neurologic deaths were higher in the arms not receiving WBRT (44 % vs. 28 %, p < 0.002). As expected, salvage WBRT was more frequent in the arms not receiving WBRT initially. An update to this trial assessed health related quality-of-life (QOL) measures, which was a secondary endpoint of the trial [96]. Patients in the observation arm reported higher QOL compared with the arm receiving WBRT. However, compliance was poor with 45 % of patients completing QOL assessment at 1 year.

The North Central Cancer Treatment Group (NCCTG) recently completed enrollment of a phase III trial (N0574) comparing SRS vs. SRS followed by WBRT delivered within 14 days for patients with 1–3 brain metastases. The primary endpoint is neurocognitive function.

With the success of SRS in treating patients with 1–4 brain metastases and efforts to eliminate the effects of WBRT on neurocognitive functioning, there has been increasing use of SRS alone for patients with multiple (≥5) lesions.

Previously, there had been concern regarding the cumulative dose to the whole brain when treating many lesions with SRS. However, Yang et al. reported that 50 % of the brain received less than 5 Gy when a maximum tumor dose of 40 Gy was used when treating 25 metastatic intracranial tumors [97]. A more recent study by Yamamoto et al. reviewed the median cumulative dose to the whole brain for patients with at least ten lesions who were treated by GKRS [98]. The median number of lesions treated with SRS was 17 (range: 10–43). The median volume for all tumors was 8.02 cm³ (range: 0.46–81.41). The median prescribed dose was 20 Gy (range: 12–25 Gy). The median cumulative dose to the whole brain was 4.71 Gy (range: 2.16–8.51 Gy). The median brain volumes receiving >10 Gy, 15 Gy, and 20 Gy were 64 cm³, 24 cm³, and 8 cm³, respectively.

With further assurance about the safety and efficacy of SRS to multiple lesions, there have been more recent studies. The local control and median survival of these studies are summarized in Table 14.10 [99–106]. The North American Gamma Knife Centers is currently proposing a trial of SRS alone for patients with five or more metastases.