Treatment of Large Brain Metastases With Stereotactic Radiosurgery

SRS Alone

The body of literature regarding the use of SRS for large tumors (>3 cm in maximum diameter, roughly equivalent to 14 cm³ in volume) is growing and is summarized in Table 3. The studies using multisession SRS dosing schedules for large tumors appear to report better survival rates than those using a single session, including our study, which may be related to the higher overall dose that can be achieved with multiple sessions. A phase II clinical trial of multisession SRS included patients with larger metastases, but the median tumor size and volume were still relatively small. In that study, tumor volume (less than 6 cm³) was a significant prognostic indicator, ¹² which is similar to the trend seen with single-session SRS studies. As expected, studies that include only patients with large metastases report worse outcomes.

OPEN IN VIEWER

Table 3.

Studies Examining SRS Alone for the Treatment of Large Brain Metastases.

Reference	Number of patients	Median tumor size, cm (max)	Median tumor volume, cm3 (max)	Median SRS dose, Gy (range)	Median fractions (range)	Median survival (months)	Survival, % at 6/12 months	Local control, % at 6/12 months	Definition of local recurrence
Ernst-Stecken et al 12	51	2.3 (5.2)	6 (65.6)	(30-35)	5	11	71/57	89/76	25% volume
Yomo et al 13	27		17.8 (53.3)	27 (20-31)	2	11.9	63/45	85/61	20% volume
Wegner et al 14	36		15.6 (82.7)	24 (12-27)	(2-5)	3	22/13	73/63	20% diameter
Jiang et al 15	40	4.1 (5.5)	17.5 (64.7)	40 (20-53)	10 (4-15)	15	NR/55	NR/94	25% diameter
Wiggenraad et al 16	41		NR	15	1	5.3	41/23	89/54	25% diameter
	51		NR	24 (12-27)	3			84/61
Han et al 17	80	3.9 (5.7)	22.4 (49.6)	14 (10-16)	1	7.9	64/39	92/85	25% volume
Present study	62	3.5 (5.8)		15 (10-18)	1	6.6	58/22	78/68	20% diameter

Abbreviations: NR, not reported; SRS, stereotactic radiosurgery.

Four studies have examined multisession SRS for large tumors alone. One was a prospective cohort study examining staged SRS in 2 sessions for patients with metastases larger than 10 cm³ in volume with a resultant median survival of 11.9 months and a 1-year local control of 61%. ¹³ The remaining studies were retrospective in nature. ^{14 -16} Overall survival in our current single-session SRS study was less than that reported in the above-mentioned studies that used multiple sessions schedules, but comparable to the single-session SRS studies. ^16,17 The other retrospective study of single-session SRS for patients with metastases larger than 3 cm in diameter, without a comparison group, had slightly better median survival (7.9 months) and 1-year local control, although this study used less strict criteria for local failure. ¹⁷ These studies are summarized in Table 3.

It is hard to make any specific recommendations of single versus multiple sessions for larger brain metastases as there has not been a randomized controlled trial (RCT) comparing the two. That said, given the limited dose that is typically given with single-session SRS, there is interest in multisession SRS at our institution.

Whole-Brain Radiation Therapy With SRS Boost

Whole-brain radiation therapy improves local and distant control of brain metastases but does not significantly improve OS. ¹⁸ Moreover, it is associated with an acute decline in various quality of life measures ¹⁹ as well as the potential for chronic neurocognitive changes. However, there is controversy regarding the relative effect of tumor progression versus WBRT itself on neurocognitive decline. ^20,21 Despite this, WBRT remains the standard treatment for patients with multiple (>5) brain metastases. ² Close imaging surveillance, allowing for possible salvage therapy, is recommended if WBRT is withheld. ¹⁹

As demonstrated in 2 RCTs, combination therapy with SRS boost to WBRT improves survival ²² and local control ^22,23 versus WBRT alone for patients with fewer than 5 metastases. ⁵ One of these studies, which included patients with metastases up to 4 cm, found significantly improved survival (median 6.5 vs 5.3 months) for patients with metastases >2 cm when treated with combined therapy versus WBRT alone. The study also reported better local control in the combined therapy group, although tumor size was not analyzed as a variable. ²²

A prospective cohort study comparing WBRT alone, SRS alone, and the two in combination for patients with a single metastasis from lung cancer included tumors up to 4.5 cm in diameter and found that both SRS alone and SRS boost to WBRT yielded significantly better median survival (9.3 and 10.6 months vs 5.7 months) and local control (6.9 and 8.6 months to local failure vs 4.0 months) than WBRT alone. Adding WBRT to SRS, however, did not confer a local control (P = .33) or survival (P = .71) benefit over SRS alone. ²⁴ Similarly, 1 RCT that excluded patients with metastases >3 cm found that adding WBRT to SRS yielded no significant survival advantage (median 7.5 vs 8.0 months) over SRS alone. However, the study reported improved local (P = .002) and distant control (P = .003) with combined therapy. ²⁵

Interestingly, our results show that adding WBRT to SRS provided some benefit for local or distant control versus SRS alone, but this benefit failed to reach statistical significance. Our results, keeping in mind that our WBRT with SRS cohort is small and limits the power of this comparison, suggest that for patients with large brain metastases undergoing SRS, WBRT may be withheld as long as regular imaging surveillance is performed and salvage therapy is available for potential intracranial failures.

Stereotactic Radiosurgery and Surgery

Surgery followed by WBRT yields comparable survival to SRS boost to WBRT. ⁵ There are, however, no phase 3 trials comparing the 2 modalities because of the difficulty in randomizing patients to an invasive therapy (ie, surgery) compared to SRS. ² Surgical resection followed by WBRT is often recommended for patients with tumors larger than 3 cm and causing a clinically significant mass effect. In addition, there is little specific evidence for the use of SRS for these large lesions ²⁶ and SRS’s effect on a lesion is less predictable. A retrospective study, which included patients with metastases up to 4 cm, compared surgery plus WBRT to SRS alone and found no significant difference in either survival rates (38% vs 54% at 1 year) or local control rates (56% vs 64% at 1 year). ²⁷ Several retrospective studies have compared surgery plus WBRT versus SRS boost to WBRT—all but one of them excluded patients with large brain metastases for SRS. One of these studies reported better survival and local control rates in the surgery group. ²⁸ The other 3 studies found no difference in survival between the groups, whereas the SRS group had better local control. ^{29 -31} The study that included patients with metastases up to 4 cm in diameter found significantly better survival rates (56% vs 47% at 1 year) and local control rates (82% vs 66% at 1 year) in the SRS group. ³²

In regard to SRS to the resection bed for large tumors, 1 prospective cohort study examined 112 patients with resected brain metastases with an average volume of 8.7 cm³ (up to 67 cm³) who were treated with a median SRS dose of 20 Gy (range 12-30 Gy) in a median 2 sessions (range 1-5 sessions). That study reported a median survival of 17 months with 1-year local and distant control rates of 91% and 46%, respectively. ³³ Our group of patients who were treated with SRS to the resection bed had a lower median survival (12.2 months) than this prospective study that included patients with smaller tumors, which is not surprising. While our 6-month local control rates are comparable at 92%, our 12-month local control rates are much worse (66%). This difference can perhaps be partially explained by the higher average dose received in the prospective study (20 vs 15 Gy), which is dictated by the size of the resection bed in this cohort.

There are 3 retrospective studies that examined the role of SRS to the resection cavity of metastases with a median diameter greater than 3 cm or a median volume greater than 14 cm³. These studies report median survival from 10.9 to 15.6 months, with 6- and 12-month local control rates of 87% to 95% and 80% to 91%, respectively. ^{34 -36} Our results are comparable in terms of survival and 6-month local control. However, our 12-month local control is again worse.

Two retrospective studies completely excluded patients with smaller tumors, including only those with brain metastases greater than 3 cm in maximum diameter. One of these studies treated the metastases with 27 Gy in 3 sessions and found a median survival of 17 months, with a 1-year local control rate of 93%. ³⁷ Both of these outcomes are better than our results, suggesting again that there is a dose–response relationship for SRS, even to a large resection bed. The other study used 24 Gy in 3 sessions, found a median survival of 5.5 months, and only reported a 6-month local control rate, which was 80%. The lower survival rates in this latter study may be partially accounted for by the larger tumor volumes included (average 29 cm³ and up to 81 cm³ vs an average 18 cm³ and up to 36 cm³ in the other study). ³⁸

Collectively, the above-mentioned studies of surgery with adjuvant SRS—including our current study—reported poor distant control rates (ranging from 35% to 57% at 12 months). There may be a risk of leptomeningeal seeding of the tumor during resection, which localized treatment such as SRS does not address, but further studies are needed. ³⁹ In our study, DBR occurred significantly more often in patients undergoing surgery with adjuvant SRS versus SRS alone (HR 2.13, 95% CI 1.05-4.31). Distant brain recurrence rates were higher in patients with more CNS metastases (HR 1.24, 95% CI 1.08-1.42) and radioresistant primary histology (HR 2.10, 95% CI 1.18-3.74), while patients with NSCLC primary tumors had lower rates of DBR. Despite the increased rates of DBR, our results indicate that surgery followed by SRS to the resection bed significantly improves survival versus SRS alone for these patients with large metastases (HR 0.52, 95% CI 0.30-0.88), although selection bias may be at play.

Prognostic Variables

There was a significant difference in patient age among the 4 groups in our study (62 vs 66 years for surgery vs SRS alone, respectively), and patients in the surgery group had fewer brain metastases on average. Other known prognostic factors, such as KPS (P = .64), GPA (P = .27), systemic metastases (P = .87), and primary disease control (P = .35), were comparable among the different groups. A prospective study that directly compares SRS alone to surgery with adjuvant SRS would help to clarify which variables are most important when recommending one treatment over the other. Our results found that only the number of metastases treated was predictive of survival on univariate and multivariate analyses.

In our study, none of the clinical variables, including treatment strategy, yielded a significant improvement in local control, which suggests that more studies are needed for optimization. The low rates of radiation necrosis in our study, however, suggest that dose escalation may be possible. A few strategies are being investigated at our institution, one of which includes staged SRS treatment in 2 to 3 sessions spaced within a month to intact metastases that are larger than 3 cm in diameter up to 5 cm in size. This technique has been previously described. ^13,40 Four patients have received this treatment, and none have developed local failure at a median follow-up time of 4.6 months. This treatment technique is being investigated retrospectively at our institution and will be reported separately. Prospective studies, with dose escalation or neoadjuvant SRS, may be helpful to further clarify the best treatment strategies for patients with large brain metastases. ⁴¹ Neoadjuvant SRS, where SRS is delivered prior to resection of the brain metastasis, has a number of advantages, which include reducing the risk of leptomeningeal spread at the time of surgery and treating a tumor with an intact blood supply, which may improve control. Also, the true resection bed may be difficult to define for SRS planning. Having a target one can see at the time of SRS may reduce the risk of missing disease. This technique is being investigated prospectively at our institution.

Limitations of our study include its retrospective nature as well as a short follow-up time due to the poor prognosis for these patients. Also, there is an inherent selection bias in which patients are chosen for surgery; these patients are typically in better health, have good functional status, have well-controlled primary disease, and have fewer brain metastases. Another limitation is that we did not study a cohort of patients with surgical resection followed by WBRT. This was intentional since the study was designed to include contemporary patients since 2007. Given the institutional shift to do SRS postresection in ideal candidates, this group was the focus of this study.