Impact of radiotherapy and sequencing of systemic therapy on survival outcomes in melanoma patients with previously untreated brain metastasis: a multicenter DeCOG study on 450 patients from the prospective skin cancer registry ADOREG.

BACKGROUND: Despite of various therapeutic strategies, treatment of patients with melanoma brain metastasis (MBM) still is a major challenge. This study aimed at investigating the impact of type and sequence of immune checkpoint blockade (ICB) and targeted therapy (TT), radiotherapy, and surgery on the survival outcome of patients with MBM.
METHOD: We assessed data of 450 patients collected within the prospective multicenter real-world skin cancer registry ADOREG who were diagnosed with MBM before start of the first non-adjuvant systemic therapy. Study endpoints were progression-free survival (PFS) and overall survival (OS).
RESULTS: Of 450 MBM patients, 175 (38.9%) received CTLA-4+PD-1 ICB, 161 (35.8%) PD-1 ICB, and 114 (25.3%) BRAF+MEK TT as first-line treatment. Additional to systemic therapy, 67.3% of the patients received radiotherapy (stereotactic radiosurgery (SRS); conventional radiotherapy (CRT)) and 24.4% had surgery of MBM. 199 patients (42.2%) received a second-line systemic therapy. Multivariate Cox regression analysis revealed the application of radiotherapy (HR for SRS: 0.213, 95% CI 0.094 to 0.485, p<0.001; HR for CRT: 0.424, 95% CI 0.210 to 0.855, p=0.016), maximal size of brain metastases (HR for MBM >1 cm: 1.977, 95% CI 1.117 to 3.500, p=0.019), age (HR for age >65 years: 1.802, 95% CI 1.016 to 3.197, p=0.044), and ECOG performance status (HR for ECOG ≥2: HR: 2.615, 95% CI 1.024 to 6.676, p=0.044) as independent prognostic factors of OS on first-line therapy. The type of first-line therapy (ICB vs TT) was not independently prognostic. As second-line therapy BRAF+MEK showed the best survival outcome compared with ICB and other therapies (HR for CTLA-4+PD-1 compared with BRAF+MEK: 13.964, 95% CI 3.6 to 54.4, p<0.001; for PD-1 vs BRAF+MEK: 4.587 95% CI 1.3 to 16.8, p=0.022 for OS). Regarding therapy sequencing, patients treated with ICB as first-line therapy and BRAF+MEK as second-line therapy showed an improved OS (HR for CTLA-4+PD-1 followed by BRAF+MEK: 0.370, 95% CI 0.157 to 0.934, p=0.035; HR for PD-1 followed by BRAF+MEK: 0.290, 95% CI 0.092 to 0.918, p=0.035) compared with patients starting with BRAF+MEK in first-line therapy. There was no significant survival difference when comparing first-line therapy with CTLA-4+PD-1 ICB with PD-1 ICB.
CONCLUSIONS: In patients with MBM, the addition of radiotherapy resulted in a favorable OS on systemic therapy. In BRAF-mutated MBM patients, ICB as first-line therapy and BRAF+MEK as second-line therapy were associated with a significantly prolonged OS. © Author(s) (or their employer(s)) 2022. Re-use permitted under CC BY-NC. No commercial re-use. See rights and permissions. Published by BMJ.

Despite the advent of new systemic therapies, the prognosis of patients with melanoma brain metastases (MBM) remains poor.

Data from recent prospective trials showed effectiveness of targeted therapy and immune checkpoint blockade, but these studies did not include concomitant radiotherapy and while they have shown that combined CTLA-4 and PD-1 blockade is superior in MBM patients compared with PD-1 monotherapy, there have been no comparisons between the outcome with targeted therapy with BRAF/MEK inhibitors and immunotherapy and their optimal therapy sequence in MBM patients yet.

Our study in a large real-world patient cohort with 450 patients reveals stereotactic radiotherapy as an independent factor for better overall survival (OS) in MBM patients.

When comparing immune checkpoint blockade with combined CTLA-4 and PD-1 blockade, PD-1 monotherapy and targeted therapy with BRAF/MEK inhibitors, the type of first-line therapy did not lead to a difference in OS in the multivariate analysis of our patient cohort.

Our data show the importance of applying additional radiotherapy to systemic therapy in MBM patients.

Background

Historically, due to low intracerebral efficacy of cytotoxic chemotherapies,1 the treatment of melanoma brain metastases (MBM) was based on surgical excision, stereotactic radiosurgery (SRS), or whole brain radiotherapy (WBRT) being associated with poor survival outcomes of affected patients.2 3 Recently published results from randomized trials designed for patients with MBM have shown intracranial effectiveness of BRAF+MEK inhibition (BRAF+MEK)4 and immune checkpoint blockade (ICB), particularly with the combination of CTLA-4+PD-1 inhibitors (CTLA-4+PD-1).5–7 Nevertheless, there are no results of head-to-head trials yet, comparing the survival of ICB versus BRAF+MEK, and their optimal sequence as first-line and second-line therapies in patients with MBM. Regarding radiotherapy, SRS can effectively treat single MBM but cannot prevent the occurrence of new intracranial lesions. The optimal timing of SRS, before, during, or after systemic therapy is also still a matter of debate. In patients with multiple MBM, conventional WBRT is often applied, despite its high toxicity and its unclear benefit for the patients’ survival.

The aim of this study was to assess the outcome of different systemic treatments (ICB, BRAF+MEK) with or without locoregional treatments (SRS, conventional radiotherapy, and surgery) in first-line and second-line therapy of patients with MBM in a prospectively collected multicenter real-world patient cohort.

Patients and methods

Study design

Melanoma patients with MBM who received first-line non-adjuvant systemic treatment with inhibitors of CTLA-4+PD-1 (ipilimumab + nivolumab), PD-1 (nivolumab, pembrolizumab) or BRAF+MEK (dabrafenib + trametinib, vemurafenib + cobimetinib, encorafenib + binimetinib) between January 2013 and January 2021 were identified from the prospective multicenter skin cancer registry ADOREG of the German Dermatologic Cooperative Oncology Group. For study inclusion, MBM had to be diagnosed before start of the first non-adjuvant systemic treatment. Data on patient and tumor characteristics, as well as baseline parameters of the first and second non-adjuvant systemic treatment were collected. The number and maximal size of MBM, the presence of symptoms from MBM, and the intake and dose of dexamethasone for symptomatic MBM were additionally collected. For patients with radiotherapy of MBM, the type (SRS or conventional), and the timing (before or after start of systemic therapy) were determined. Since we could not distinguish between WBRT and postoperative radiotherapy of the tumor cavity, we assessed the effect of conventional radiotherapies (CRT) in general in this study. Best response as assessed by the investigators was categorized as complete response (CR), partial response (PR), stable disease (SD), and progressive disease (PD) according to RECIST V.1.1.8–10 Best overall response (BOR), best extracranial response (BER), and best intracranial (BIR) response to treatment were assessed retrospectively by the investigators. Study endpoints were progression-free survival (PFS) and overall survival (OS).

Statistical analysis

Univariate and multivariate Cox proportional hazards regression analyses were performed to assess the impact of baseline patient and tumor characteristics and therapeutic measures on PFS and OS. The following parameters were included into the univariate and multivariate analyses: sex, age, type of therapy, BRAF mutation status, ECOG performance status (ECOG-PS), serum LDH, number and maximal size of MBM, dexamethasone intake, application and type of radiotherapy for MBM (SRS and CRT), and surgery of MBM. OS was defined as time from start of systemic therapy until death or last patient contact (censored OS); PFS as time from start of systemic therapy until disease progression or last patient contact (censored PFS). Kaplan-Meier estimates were used for PFS and OS calculation; differences between groups were assessed by two-sided log-rank test. P values <0.05 were considered statistically significant. Patients with missing data were excluded from the respective analyses. Statistical analyses were performed with IBM SPSS Statistics V.27.

Results

Patient characteristics and treatment response

Data freeze was February 1st, 2021. From 458 patients identified, 450 met the study inclusion criteria. These 450 patients had a median follow-up time of 33.4 (range: 0–93.3) months after start of the first systemic therapy. One hundred and ninety-nine patients (44.2 %) received a second-line therapy. A detailed study flow is provided in figure 1. Of 450 patients eligible for analysis, 63.3% (n=285) were male and 54.9% (n=274) were ≤65 years old. An activating BRAF V600E/K mutation was detected in 48.7% (n=219). For detailed patient characteristics, see table 1. The median time from first diagnosis of melanoma until first diagnosis of brain metastasis was 20 (range: 0–391) months; the median time from first diagnosis of brain metastasis until start of first systemic therapy was 35 (range: 1–8594) days.

Figure 1

Study flow. Four hundred and fifty-eight patients from 35 skin cancer centers were identified in the prospective multicenter ADOREG registry. Of these, 450 patients were eligible for analysis.

Table 1

First-line treatment of patients with melanoma brain metastasis: baseline characteristics and therapy outcome

	All patientsn=450 (100%)	CTLA-4+PD-1n=175 (100%)	PD-1n=161 (100%)	BRAF+MEKn=114 (100%)
Age
≤65 years >65 years	274 (54.9)203 (45.1)	102 (58.3)73 (41.7)	73 (45.3)88 (54.7)	72 (63.2)42 (36.8)
Gender
Male Female	285 (63.3)165 (36.7)	109 (62.3)66 (37.7)	96 (59.6)65 (40.4)	80 (70.2)34 (29.8)
Site of primary
Cutaneous Mucosal Uveal Unspecified or unknown	357 (79.3)5 (1.1)4 (0.9)84 (18.6)	139 (79.4)2 (1.1)1 (0.6)33 (18.9)	127 (78.9)3 (1.9)3 (1.9)28 (17.4)	91 (79.8)0 (0.0)0 (0.0)23 (20.2)
BRAF status
V600 wildtype V600E/K mutation Unknown	191 (42.4)219 (48.7)40 (8.9)	85 (48.6)63 (36.0)27 (15.4)	106 (65.8)42 (26.1)1 (8.1)	0 (0.0)114 (100.0)0 (0.0)
Previous adjuvant therapy stage III
Yes No	78 (17.3)372 (82.7)	30 (17.1)145 (82.9)	32 (19.9)129 (80.1)	16 (14.0)98 (86.0)
Previous adjuvant therapy stage IV
Yes No	20 (4.4)430 (95.6)	6 (3.4)169 (96.6)	8 (5.0)153 (95.0)	6 (5.3)108 (94.7)
ECOG performance status
0 1 ≥2 Unknown	164 (36.4)84 (18.7)30 (6.7)172 (38.2)	83 (47.4)26 (14.9)8 (4.6)58 (33.1)	51 (31.7)38 (23.6)9 (5.6)63 (39.1)	30 (26.3)20 (17.5)13 (11.4)51 (44.7)
Serum LDH
Normal (≤ULN) Elevated (>ULN) >10× ULN Unknown	146 (32.4)199 (44.2)43 (9.6)105 (23.3)	57 (32.6)73 (41.7)17 (9.7)45 (25.7)	62 (38.5)72 (44.7)12 (7.5)27 (16.8)	27 (23.7)54 (47.4)14 (12.3)33 (28.9)
Extracranial affected organ sites
0 1–2 ≥3	81 (18.0)218 (48.4)151 (33.6)	31 (17.7)89 (50.9)55 (31.4)	32 (19.9)77 (47.8)52 (32.3)	18 (15.8)52 (45.6)44 (38.6)
Number of brain metastases
1 2–4 ≥5 Unknown	114 (25.3)111 (24.7)157 (34.9)68 (15.1)	47 (26.9)42 (24.0)56 (32.0)30 (17.1)	40 (24.8)41 (25.5)61 (37.9)19 (11.8)	27 (23.7)28 (24.6)40 (35.1)19 (16.7)
Maximal size of brain metastases
≤1 cm >1 cm Unknown	127 (28.2)136 (30.2)187 (41.6)	51 (29.1)47 (26.9)77 (44.0)	46 (28.6)47 (29.2)68 (42.2)	30 (26.3)42 (36.8)42 (36.8)
Maximal size of brain metastases
No Yes Unknown	291 (64.7)118 (26.2)41 (9.1)	118 (67.4)42 (24.0)15 (8.6)	106 (65.8)37 (23.0)18 (11.2)	67 (58.8)39 (34.2)8 (7.0)
Dexamethasone intake at therapy start
No Yes Unknown	314 (69.8)95 (21.1)41 (9.1)	127 (72.6)35 (20.0)13 (7.4)	114 (70.8)28 (17.4)19 (11.8)	73 (64.0)32 (28.1)9 (7.9)
Radiotherapy of brain metastases
None Stereotactic Conventional Stereotactic and conventional	147 (32.7)137 (30.4)135 (30.0)31 (6.9)	61 (34.9)46 (26.3)56 (32.0)12 (6.9)	47 (29.2)52 (32.2)50 (31.1)12 (7.5)	39 (34.2)39 (34.2)29 (25.4)7 (6.1)
Surgery of brain metastases
No Yes	340 (75.6)110 (24.4)	134 (76.6)41 (23.4)	122 (75.8)39 (24.2)	84 (73.7)30 (26.3)
Best overall response
CR PR SD PD Unknown	42 (9.3)120 (26.7)62 (13.8)189 (42.0)36 (8.0)	12 (6.9)42 (24.0)20 (11.4)82 (46.9)19 (10.9)	24 (14.9)38 (23.6)23 (14.3)66 (41.0)10 (6.2)	6 (5.3)40 (35.1)19 (16.7)41 (36.0)8 (7.0)
Best extracranial response
NED CR PR SD PD Unknown	35 (7.8)51 (11.3)105 (23.3)75 (16.7)106 (23.6)78 (18.3)	12 (6.9)14 (8.0)37 (21.1)28 (16.0)47 (26.9)37 (21.1)	15 (9.3)25 (15.5)33 (20.5)20 (12.4)39 (24.2)29 (18.0)	8 (7.0)12 (10.5)35 (30.7)27 (23.7)20 (17.5)12 (10.5)
Best intracranial response
CR PR SD PD Unknown	70 (15.6)104 (23.1)60 (13.3)135 (30.0)81 (18.0)	24 (13.8)35 (20.0)24 (13.7)58 (33.1)34 (19.4)	35 (22.7)31 (19.3)17 (10.6)46 (28.6)32 (19.9)	11 (9.7)38 (33.3)19 (16.7)31 (27.2)15 (13.2)
Therapy end reason
Planned stop Toxicity Disease progression Patient wish Other Ongoing Lost to follow-up	7 (1.6)81 (18.0)238 (52.9)16 (3.6)36 (8.0)65 (14.4)7 (1.5)	1 (0.6)54 (30.9)69 (39.4)3 (1.7)12 (6.9)33 (18.9)3 (1.7)	6 (3.7)19 (11.8)86 (53.4)9 (5.6)17 (10.6)20 (12.4)4 (2.5)	0 (0.0)7 (6.1)85 (74.6)5 (4.4)6 (5.3)11 (9.6)0 (0.0)
Progression
No Yes	120 (26.7)330 (73.3)	63 (36.0)112 (64.0)	40 (24.8)121 (75.2)	17 (14.9)97 (85.1)
Death
No Yes	236 (52.4)214 (47.6)	112 (64.0)63 (36.0)	82 (50.9)79 (49.1)	42 (36.8)72 (63.0)
Progression-free survival
Median in months (95% CI)	4.7 (3.9 to 5.6)	3.9 (2.3 to 5.5)	5.4 (2.0 to 8.8)	5.0 (4.3 to 5.8)
Overall survival
Median in months (95% CI)	21.5 (16.9 to 26.0)	Not reached	26.6 (12.0 to 40.5)	16.5 (8.9 to 24.1)

CR, complete response; LDH, lactate dehydrogenase; NED, no evidence of disease; PD, progressive disease; PR, partial response; SD, stable disease; ULN, upper level of norm.

For first-line non-adjuvant therapy, 175 (38.9%) patients received CTLA-4+PD-1 ICB, 161 (35.8%) received PD-1 ICB, and 114 (25.3%) were treated with BRAF+MEK (dabrafenib + trametinib: 82 patients; vemurafenib + cobimetinib: 19 patients; encorafenib + binimetinib: 13 patients). The baseline characteristics for BRAF-mutated patients were comparable with those of the overall cohort (online supplemental table S1). Sixty-three (28.8%) of BRAF-mutated patients received a first-line therapy with CTLA-4+PD-1, 42 (19.2%) PD-1 and 114 (52.1%) BRAF+MEK. One hundred and ninety-nine patients (44.2%) of the total cohort received a second-line therapy that consisted of CTLA-4+PD-1 in 56 (28.1%), PD-1 in 43 (21.6%), BRAF+MEK in 64 (32.2%), and other therapy types including chemotherapy in 36 (18.1%) patients (for details, see online supplemental table S2). The median duration of first-line systemic therapy was 3.2 (range: 0–70.9) months and of second-line systemic therapy 2.4 (range: 0–39.4) months.

At baseline of first systemic therapy, 18% (n=81) of patients had only MBM, in 48.4% (n=218) one to two extracranial organs, and in 33.6% (n=151) three or more extracranial organs were additionally affected. A percentage of 25.3% (n=114) of patients had a solitary brain metastasis, 24.7% (n=111) had oligometastatic disease with 2–4 MBM, and 34.9% (n=157) had multiple (≥5) MBM. In 15.1% (n=68) of patients, the number of MBM was unknown. A percentage of 26.2% (n=118) of patients had symptomatic MBM, 64.7% (n=291) were asymptomatic, and in 9.1% (n=41) this information was missing. Ninety-five (21.1%) patients received dexamethasone at therapy start, 126 (28%) during the first 3 months of systemic therapy, and 89 (19.8%) patients after the first 3 months of systemic therapy. The median dose of dexamethasone at therapy start was 5.5 (range: 0.5–32) mg, and the median maximal dose within the first 3 months was 8 (range: 0.5–32) mg.

BOR of all first-line therapies was 9.3% (n=42) CR, 26.7% (n=120) PR, 13.8% (n=62) SD, 42.0% (n=189) PD, and 8% (n=36) unknown (table 1). Intracranial and extracranial objective response rates were similar, with 30% (n=135) of patients showing PD as BIR compared with 23.6% (n=106) as BER. BOR was 30.6% (n=54) for CTLA-4+PD-1, 38.5% (n=62) for PD-1, and 40.4% (n=46) for BRAF+MEK. A percentage of 52.9% (n=238) of patients stopped first-line therapy due to disease progression, and 18% (n=81) because of side effects. Discontinuation rates due to toxicity were 30.9% (n=54) for CTLA-4+PD-1, 11.8% (n=19) for PD-1, and 6.7% (n=7) for BRAF+MEK.

With regard to radiotherapy, 30.4% (n=137) of patients received SRS, 30.0% (n=135) CRT, 6.9% SRS and CRT (n=31), and 32.7% (n=147) no radiotherapy (online supplemental table S3). Sixty-eight (15.1%) patients received SRS during first-line systemic therapy, and 37 (8.2%) within 1 month before therapy start. Forty-one (9.1%) patients received CRT during first-line systemic therapy and 52 (11.6%) within 1 month before therapy start. The median total dose was 25.3± (range: 7–58) gray for SRS and 34.6 (range: 6–60) gray for CRT. Surgical resection of MBM was performed in 110 patients (24.4%). Eigty-five of those had MBM surgery with consecutive radiotherapy, and 25 had surgery only; one patient received five consecutive surgical resections of MBM.

Survival on first-line systemic therapy in patients with MBM

At database closure, 330 of the total 450 patients (73.3%) had progressed on first-line therapy, and 236 patients (52.4%) had died. The median PFS was 4.7 (range: 0.0–74.8) months, and the median OS was 21.5 (range: 0.0–74.8) months. PFS and OS were strongly correlated with BOR (figure 2A, B), being best for patients who achieved a CR. Similar correlations were found for BIR and BER, with higher median OS times for BER than for BIR. Median PFS was 3.9 months for CTLA-4+PD-1, 5.4 months for PD-1, and 5.0 months for BRAF+MEK; median OS was not reached for CTLA-4+PD-1, 26.6 months for PD-1, and 16.5 months for BRAF+MEK (table 1; figure 2C, D). Patients with symptomatic MBM showed a decreased median OS (CTLA-4+PD-1: 11.7 months, PD-1: 8.6 months, BRAF+MEK inhibition: 14.3 months) compared with patients with asymptomatic MBM (CTLA-4+PD-1: not reached, PD-1: 36.9 months, BRAF+MEK inhibition: 20.1 months) with an objective intracranial response rate of 36.4% for symptomatic and 44% for asymptomatic patients. Additional survival rates of selected patient groups at 6, 12, and 24 months can be found in table 2. Survival outcomes were similar when BRAF-mutated patients were assessed separately (online supplemental figure S1).

Figure 2

Kaplan-Meier curves showing progression-free and overall survival for first-line and second-line systemic therapy in melanoma patients with brain metastases. (A and B) Best overall response to first-line therapy (CR, complete response; PR, partial response; SD, stable disease; PD, progressive disease); (C and D) type of first-line therapy; (E and F) type of second-line therapy. The log-rank test was used to compare between groups; p<0.05 was considered significant. The p value states that there is a difference between the groups calculated by the log-rank test.

Table 2

Selected overall survival (OS) rates in patients with melanoma brain metastasis

	6 months OS rate(deceased; at risk)	12 months OS rate(deceased; at risk)	24 months OS rate(deceased; at risk)	MedianOS (months)	P value(log-rank test)
Total patient cohort (n=450 pts)	76.0% (92; 292)	57.9% (150; 206)	37.0%(191; 112)
First-line systemic therapy
CTLA-4+PD-1 (n=175) PD-1 (n=161) BRAF+MEK (n=114)	74.3% (35; 101)75.9% (35; 110)78.6% (22; 81)	57.5% (51; 69)61.9% (53; 86)52.6% (46; 51)	33.7% (61; 31)44.3% (68; 54)29.2% (62; 27)	nr26.616.5	p=0.037
Symptomatic brain metastases
No (n=291) Yes (n=118)	80.5% (48; 198)65.1% (37; 69)	62.7% (84; 141)45.0% (55; 45)	43.1% (107; 81)22.7% (68; 20)	30.811.5	p<0.001
Radiotherapy of brain metastases
None (n=147) Stereotactic (n=137) Conventional (n=135)	65.5% (40; 76)89.0% (13; 105)69.7% (36; 83)	45.5% (55; 46)73.0% (30; 82)51.3% (55; 58)	25.6% (61; 21)52.1% (45; 49)33.3% (68; 34)	19.536.417.0	p<0.001
Surgery of brain metastases
No (n=340) Yes (n=110)	72.0% (80; 206)87.8% (12; 86)	54.7% (119; 144)66.3% (31; 62)	33.6% (150; 76)46.8% (41; 36)	20.136.4	p=0.031
Second-line systemic therapy
CTLA-4+PD-1 (n=56) PD-1 (n=43) BRAF+MEK (n=64) Others (n=36)	58.8% (21; 30)59.5% (15; 22)84.5% (9; 49)48.3% (15; 14)	39.1% (28; 18)41.2% (19; 15)63.3% (18; 31)28.6% (20; 8)	27.3% (31; 13)21.8% (25; 7)26.5% (25; 9)17.9% (23; 5)	10.410.519.77.3	p=0.010

P values from log-rank test comparing the median OS values; p<0.05 was considered significant.

NR, not reached.

In the univariate Cox regression analysis of survival on first-line systemic therapy, we found the following significant prognostic factors for OS: number of MBM, maximum size of MBM, ECOG-PS, serum LDH, symptoms of MBM, dexamethasone intake, dexamethasone dose, number of affected extracranial organs, presence and type of radiotherapy, surgery of MBM, type of systemic therapy, and type of second-line systemic therapy. For PFS, we found these significant prognostic factors: ECOG-PS, serum LDH, presence of symptomatic MBM, dexamethasone intake, number of affected extracranial organs, radiotherapy, timing of SRS, type of systemic first-line therapy, and type of second-line therapy. For details of the univariate analyses, see online supplemental table S4, figures 2A–F and 3A–F, and online supplemental figure S2A-F and figure S3A-F.

After adjusting for confounders using the multivariate Cox regression analysis, we found radiotherapy (HR 0.213 for SRS vs none, p<0.001; HR 0.424 for CRT versus none, p=0.016), maximal size of MBM (HR 1.977 for size >1 cm, p=0.019), age (HR 1.8 for age >65 years, p=0.044) and ECOG-PS (HR 2.615 for ECOG-PS ≥2, p=0.044) as independent prognostic factors for OS on first-line therapy in patients with MBM (table 3). We did not detect any independent prognostic factors for PFS.

Table 3

Multivariate Cox regression analysis for progression-free survival (PFS) and overall survival (OS) on first-line systemic therapy in melanoma patients with brain metastasis

Parameters included	PFSHR (95% CI) P value	OSHR (95% CI) P value
Gender (reference: male)
Male versus female	0.993 (0.627 to 1.572) 0.975	0.978 (0.554 to 1.724) 0.938
Age (reference: ≤65 years)
≤65 versus >65 years	0.945 (0.569 to 1.569) 0.828	1.802 (1.016 to 3.197) 0.044
BRAF status (reference: wildtype)
Wildtype versus BRAFV600E/K	0.756 (0.420 to 1.360) 0.350	0.839 (0.406 to 1.734) 0.636
ECOG-PS (reference: 0)
0 versus 1 0 versus ≥2	0.899 (0.518 to 1.563) 0.7071.446 (0.594 to 3.521) 0.416	1.404 (0.753 to 2.617) 0.2862.615 (1.024 to 6.676) 0.044
Serum LDH (reference: normal)
Normal versus elevated	1.415 (0.863 to 2.321) 0.168	1.291 (0.711 to 2.342) 0.401
Number of brain metastases (reference: 1)
1 versus 2–4 1 versus ≥5	1.361 (0.738 to 2.509) 0.3241.000 (0.559 to 1.786) 0.999	1.208 (0.557 to 2.620) 0.6330.940 (0.445 to 1.983) 0.870
Maximal size of brain metastases (reference: ≤1 cm)
≤1 cm versus >1 cm	0.837 (0.531 to 1.319) 0.443	1.977 (1.117 to 3.500) 0.019
Dexamethasone intake at therapy start (reference: no)
No versus yes	0.903 (0.470 to 1.737) 0.760	1.159 (0.547 to 2.457) 0.700
Number of affected extracranial organ sites (reference: 0)
0 versus 1–2 0 versus ≥3	0.968 (0.509 to 1.837) 0.9200.944 (0.487 to 1.837) 0.865	0.834 (0.385 to 1.844) 0.6690.839 (0.406 to 1.734) 0.636
Radiotherapy (reference: none)
None versus stereotactic None versus conventional	0.647 (0.334 to 1.251) 0.1951.033 (0.551 to 1.936) 0.920	0.213 (0.094 to 0.485)<0.001 0.424 (0.210 to 0.855) 0.016
Surgery of brain metastases (reference: no)
No versus yes	0.794 (0.429 to 1.471) 0.463	2.117 (0.947 to 4.731) 0.068
Type of systemic therapy (reference: BRAF+MEK besides*)
BRAF+MEK versus CTLA-4+PD-1 BRAF+MEK versus PD-1 CTLA-4+PD-1 versus PD-1*	1.259 (0.606 to 2.615) 0.5371.290 (0.587 to 2.831) 0.5261.024 (0.607 to 1.728) 0.928	0.995 (0.433 to 2.290) 0.9910.536 (0.221 to 1.300) 0.1680.538 (0.273 to 1.063) 0.074

P<0.05 was considered significant. Significant values in bold letters.

*Reference: CTLA-4+PD-1.

Radiotherapy and surgical resection of MBM

With regard to radiotherapy, patients who received additional SRS showed a significantly improved median OS (36.4 months) on first-line systemic therapy compared with patients without any radiotherapy of MBM (19.5 months) or to patients who received CRT (16.9 months) (figure 3A, B). While the median OS on CTLA-4+PD-1 was not reached in patients treated with SRS and with no radiotherapy, it was 20.2 (range: 0.5–46.0) months with CRT. In patients treated with PD-1, additional SRS led to a median OS of 38.7 (range: 1.1–75.1) months, whereas CRT revealed 15.1 (range: 0.0–55.5) and no radiotherapy 19.8 (range: 0.0–54.9) months. Patients treated with BRAF+MEK inhibition showed median OS of 24.7 (range: 2.3–63.2) months with SRS, 17.0 (range: 0.7–58.2) months with CRT, and 8.3 (range: 0.0–41.5) months without radiotherapy. Median PFS with SRS was 5.7 months for patients treated with CTLA-4+PD-1 and 5.5 months for patients treated with PD-1 or BRAF+MEK inhibition.

Figure 3

Kaplan-Meier curves showing progression-free and overall survival on first-line systemic therapy in melanoma patients with brain metastases with or without additional radiotherapy or surgical resection. (A and B) radiotherapy (RT) in addition to systemic therapy; (C and D) surgical resection of brain metastases in addition to systemic therapy; (E and F) timing of stereotactic radiotherapy in addition to systemic therapy (before: RT within 1 month before start of systemic therapy; during: RT while systemic therapy ongoing). The log-rank test was used to calculate differences between groups; p<0.05 was considered significant. The p value states that there is a difference between the groups calculated by the log-rank test.

The presence and type of radiotherapy of MBM were detected as the strongest independent prognostic parameters in the multivariate Cox proportional hazard model (table 3). Regarding the timing of radiotherapy, patients who received SRS during systemic treatment showed a significantly favorable PFS, but not OS, compared with patients receiving SRS within 1 month before systemic therapy (median PFS 5.5 vs 2.9 months, p=0.028, figure 3E, F). This difference was significant in the ICB cohort only (ICB: median PFS 8.8 (SRS during) vs 2.8 months (SRS before); BRAF+MEK inhibition: median PFS 4.5 (SRS during) versus 6.8 (SRS before) months. The number of SRS (one vs multiple), or whether SRS was applied as single exposure or fractionated did not show a relevant survival difference.

Regarding surgery of MBM, the excision of brain metastases was significantly associated with a favorable OS but not PFS as compared with patients with no surgery (median 36.4 vs 20.1 months, p=0.031, figure 3C, D).

Survival on second-line systemic therapy in patients with MBM

Next, we compared the baseline characteristics of patients who died after first-line therapy to those who received a second-line therapy. Patients who received second-line therapy were significantly younger, had a better ECOG-PS, less often elevated serum LDH, less symptomatic MBM, and fewer and smaller MBM (for details, see table 4). A larger fraction of patients who received second-line therapy also received radiotherapy (SRS 38.7% vs 15.2%). Interestingly, patients who died after first-line therapy had been more often treated with PD-1 and less often with BRAF+MEK (41.9% vs 24.8%) compared with patients who received a second-line therapy (29.1% vs 36.2%).

Table 4

Baseline and therapy characteristics of patients who received second-line therapy compared with patients who died before second-line therapy

	All patients n=450 (100%)	Patients who received second-line therapy n=199 (100%)	Patients who died before second-line therapy n=105 (100%)
Age
≤65 years >65 years	274 (54.9)203 (45.1)	131 (65.8)68 (34.2)	49 (46.7)56 (53.3)
Gender
Male Female	285 (63.3)165 (36.7)	122 (61.3)77 (38.7)	65 (61.9)40 (38.1)
Site of primary
Cutaneous Mucosal Uveal Unspecified or unknown	357 (79.3)5 (1.1)4 (0.9)84 (18.6)	162 (81.4)1 (0.5)2 (1.0)34 (17.1)	84 (80.0)2 (1.9)1 (1.0)18 (17.1)
BRAF status
V600 wildtype V600E/K mutation Unknown	193 (42.9)217 (48.2)40 (8.9)	59 (29.6)137 (68.8)3 (1.5)	57 (54.3)36 (34.3)12 (11.4)
Previous adjuvant therapy in stage III
Yes No	78 (17.3)372 (82.7)	41 (20.6)158 (79.4)	15 (14.3)90 (85.7)
Previous adjuvant therapy in stage IV
Yes No	20 (4.4)430 (95.6)	9 (4.5)190 (95.5)	5 (4.8)100 (95.2)
ECOG performance status
0 1 ≥2 Unknown	164 (36.4)84 (18.7)30 (6.7)172 (38.2)	77 (38.7)29 (14.6)15 (7.5)78 (39.2)	30 (28.6)27 (25.7)11 (10.5)37 (35.2)
Serum LDH
Normal (≤ULN) Elevated (>ULN) >10× ULN Unknown	146 (32.4)199 (44.2)43 (9.6)105 (23.3)	70 (35.2)78 (39.2)19 (9.5)51 (25.6)	28 (26.7)64 (61.0)24 (9.6)13 (12.4)
Extracranial affected organ sites
0 1–2 ≥3	81 (18.0)218 (48.4)151 (33.6)	30 (15.1)104 (52.3)65 (32.7)	23 (21.9)42 (40.0)40 (38.1)
Number of brain metastases
1 2–4 ≥5 Unknown	114 (25.3)111 (24.7)157 (34.9)68 (15.1)	63 (31.7)47 (23.6)56 (28.1)33 (16.6)	17 (16.2)32 (30.5)44 (41.9)12 (11.4)
Maximal size of brain metastases
≤1 cm >1 cm Unknown	127 (28.2)136 (30.2)187 (41.6)	62 (31.2)55 (27.6)82 (41.2)	21 (20.0)40 (38.1)44 (41.9)
Symptomatic brain metastases
No Yes Unknown	291 (64.7)118 (26.2)41 (9.1)	140 (70.4)43 (21.6)16 (8.0)	52 (49.5)45 (42.9)8 (7.6)
Radiotherapy of brain metastases
None Stereotactic Conventional Stereotactic and conventional	147 (32.7)137 (30.4)135 (30.0)31 (6.9)	34 (17.1)77 (38.7)68 (34.2)20 (10.1)	46 (43.8)16 (15.2)36 (34.3)7 (6.7)
Surgery of brain metastases
No Yes	340 (75.6)110 (24.4)	141 (70.9)58 (29.1)	84 (80.0)21 (20.0)
Type of first-line systemic therapy
CTLA-4+PD-1 PD-1 BRAF+MEK	175 (38.9)161 (35.8)114 (25.3)	69 (34.7)58 (29.1)72 (36.2)	35 (33.3)44 (41.9)26 (24.8)
Best overall response
CR PR SD PD Unknown	42 (9.3)120 (26.7)62 (13.8)189 (42.0)36 (8.0)	7 (3.5)59 (29.6)28 (14.1)97 (48.7)8 (4.0)	3 (2.9)11 (10.5)19 (12.4)63 (65.7)9 (8.6)
Best extracranial response
NED CR PR SD PD Unknown	35 (7.8)51 (11.3)105 (23.3)75 (16.7)106 (23.6)78 (18.3)	11 (5.5)14 (7.0)60 (30.2)40 (20.1)49 (24.6)25 (12.6)	12 (11.4)3 (2.9)8 (7.6)16 (15.2)44 (41.9)22 (21.0)
Best intracranial response
CR PR SD PD Unknown	70 (15.6)104 (23.1)60 (13.3)135 (30.0)81 (18.0)	17 (8.5)51 (25.6)32 (16.1)75 (37.7)24 (12.1)	10 (9.6)12 (11.4)11 (10.5)44 (41.9)28 (26.7)
Therapy end reason
Planned stop Toxicity Disease progression Patient wish Other Ongoing Lost to follow-up	7 (1.6)81 (18.0)238 (52.9)16 (3.6)36 (8.0)65 (14.4)7 (1.5)	1 (0.5)41 (20.6)147 (73.9)4 (2.0)6 (3.0)0 (0.0)0 (0.0)	0 (0.0)12 (11.4)79 (75.2)2 (1.9)10 (9.5)0 (0.0)2 (1.9)
Progression
No Yes	120 (26.7)330 (73.3)	14 (7.0)185 (93.0)	6 (5.7)99 (94.3)
Death
No Yes	236 (52.4)214 (47.6)	90 (45.2)109 (54.8)	0 (0.0)105 (100.0)
Progression-free survival
Median in months (95% CI)	4.7 (3.9 to 5.6)	3.8 (2.7 to 5.0)	2.1 (1.9 to 2.3)
Overall survival
Median in months (95% CI)	21.5 (16.9 to 26.0)	23.9 (19.8 to 28.0)	19.5 (6.6 to 32.4)

CR, complete response; PD, progressive disease; PR, partial response; SD, stable disease.

For second-line therapy in patients with MBM, we detected the following significant factors for OS using an univariate Cox regression analysis (online supplemental table S5): ECOG-PS, presence of symptomatic MBM, intake of dexamethasone within first 3 months of second therapy, dexamethasone dose at start of second systemic therapy and therapy with BRAF+MEK as second-line therapy compared with all other therapy regimens. ECOG-PS, presence of symptomatic MBM, dexamethasone within the first 3 months of systemic therapy and BRAF+MEK as second-line therapy were also significant parameters for PFS.

We then performed a multivariate Cox regression analysis (online supplemental table S6) of survival on second-line therapy, which showed an ECOG-PS of 0 at therapy start and treatment with BRAF+MEK as independent prognostic factors for longer OS. The same parameters were also independently prognostic of PFS on second-line therapy.

One hundred and thirty-seven of 219 (62.6%) patients with an activating BRAF mutation and 59 of 191 (30.9%) BRAF wildtype patients received a second-line therapy. Presuming that the therapy sequence influences OS, we performed univariate and multivariate Cox regression analyses for all possible therapy sequences in the entire patient cohort. When referred to BRAF+MEK as first-line therapy without subsequent therapies, the following sequences were favorable in the univariate analysis: CTLA-4+PD-1 followed by no further therapy (HR 0.578, p=0.034) or followed by BRAF+MEK (HR 0.322, p=0.001), and PD-1 followed by no therapy (0.519, p=0.009), by CTLA-4+PD-1 (HR 0.316, p=0.002) or BRAF+MEK hour 0.407, p=0.027). In the multivariate analysis (online supplemental table S7), including age, gender, ECOG-PS, LDH, radiotherapy and therapy sequence, an ECOG-PS of 0 (HR 1.536, p=0.039 for ECOG ≥1), SRS (HR 0.474, p=0.008 compared with no radiotherapy) and CTLA-4+PD-1 or PD-1 both followed by BRAF+MEK (HR 0.370, p=0.035; HR 0.290, p=0.035 respectively), as well as PD-1 followed by CTLA-4+PD-1 (0.333, p=0.046) were independent prognostic parameters for increased OS.

Discussion

The results of our present study reveal an impact of additional radiotherapy, maximal size of brain metastases, age and performance score on the outcome of systemic treatment of MBM. Interestingly, the favorable survival effect of radiotherapy is detectable for SRS as well as for CRT. Our data underline a particular importance of SRS additional to systemic treatment in patients with MBM, with no regard to the type of systemic treatment. Patients with SRS showed longer median survival and SRS was confirmed as independent prognostic factor for OS in our multivariate analysis. An improved survival on SRS had also been detected in the univariate analysis of a recent study by Amaral et al11 in MBM patients treated with CTLA-4+PD-1, but type of treatment and addition of radiotherapy were not included in that study’s multivariate analysis to adjust for confounders. Notably, the number of MBM in our cohort did not show an independent prognostically relevant effect (when comparing 1, 2–4 and ≥5 metastases), which might be explained by the fact that SRS is nowadays applied in patients with up to 10–15 brain metastases and not restricted to patients with ≤5 MBM anymore. This finding is in line with a study by Rauschenberg et al12 that included type of radiotherapy and number of brain metastasis in the multivariate analysis with only type of radiotherapy, but not the number of brain metastasis (when comparing 1, 2–3 and >3) being an independently prognostic factor.

Whether SRS before or after initiation of systemic therapy is more beneficial is still unclear though. SRS can eradicate inhibitory T cells in the tumor microenvironment that otherwise dampen the immune response.13 14 Most studies dealing with this question focused on CTLA-4 monotherapy and assessed intracranial response rate or time to cerebral progression. Several smaller studies and meta-analyses showed improved OS of patients treated with ICB and concomitant (±1 month before or after therapy start) SRS when compared with ICB and non-concomitant SRS.15–17 In our overall cohort, when focusing only on patients who received SRS up to 1 month before or after first-line systemic therapy, we detected a significantly higher median PFS, but not OS, for patients who received SRS during systemic therapy compared with those who received SRS up to 1 month before. This difference could only be detected in patients treated with ICB as first-line treatment.

Although SRS has been restricted to lesions ≤3 cm diameter in the past, a fractionated SRS approach allows treatment of bigger and critically located metastases now.18 19 SRS has largely replaced surgery and postsurgical conventional radiation, but the optimal timing of SRS is still a matter of debate.20 21 Compared with surgery and postsurgical radiation, SRS has many advantages, in particular reduction of intraoperative seeding of viable tumor cells, but does not allow histopathological analysis of the metastasis.22 23 Since locoregional control is as good if more than five metastases are treated by SRS compared with fewer,24 more and more centers are currently treating up to 15 MBM with SRS.25 Nevertheless, the occurrence of new distant intracranial metastases cannot be reduced by this way; therefore, hippocampal-avoidant conventional WBRT is often performed. Several randomized controlled trials showed no improved OS but reduced intracranial relapse rates with WBRT.26–28 Reduction of intracranial relapse can improve the patients’ health-related quality of life, while the side effects of WBRT can impair it. Therefore, the necessity of WBRT, especially in asymptomatic patients, has been highly debated. In our study, CRT showed a positive effect on OS in the multivariate analysis, which included also the size and number of MBM. It has to be kept in mind that in the CRT group we could not differentiate between WBRT and postoperative CRT of the tumor cavity. Nevertheless, we clearly distinguished between SRS and conventional radiotherapy and only included patients who received conventional radiotherapy in the CRT cohort. Since metastasectomy is usually followed by postoperative radiotherapy of the tumor cavity (in our study in 77.3%), but metastasectomy itself did not have a favorable effect on survival in the multivariate analysis, it is not likely that this beneficial effect can be attributed to postoperative radiotherapy of the tumor cavity. Although these data are preliminary, they suggest that there might be a survival benefit for patients with CRT additional to systemic therapy compared with systemic therapy alone, which may be explained by increased immunogenic cell death induced by radiotherapy.29–32 Therefore, the indication for CRT has to be carefully discussed individually with every patient.

Toward safety of SRS combined with ICB, four patients of our cohort reported with radionecrosis. There is controversial data about the increased risk of adverse radiation effects such as radionecrosis in patients treated concomitantly with ICB and SRS33–35 including a meta-analysis of the literature (mainly MBM patients), indicating that the risk of adverse effects is not increased.17 BRAF+MEK inhibition is known to increase the general risk of bleeding.36 However, as reported in previous studies12 37 38 concomitant treatment with BRAF+MEK inhibition and SRS was safe in our study.

With regard to systemic therapy, recently published treatment recommendations for MBM recommend CTLA-4+PD-1 as first-line treatment.39 In line with this recommendation, our univariate results revealed that patients receiving BRAF+MEK inhibition as first-line therapy showed a significantly shorter median OS than those receiving PD-1 based ICB. However, these effects could not be demonstrated in the multivariate model, suggesting that the univariately significant difference between treatment types could be caused by poorer prognostic baseline characteristics in patients treated with BRAF+MEK inhibition (ECOG ≥2: 11.4% for BRAF+MEK inhibition vs 5.6% for CTLA-4+PD-1% and 4.6% for PD-1; symptomatic MBM 34.2% for BRAF+MEK inhibition vs 23.0% for PD-1% and 24.0% for CTLA-4+PD-1).

In our investigated cohort, patients receiving CTLA-4+PD-1 showed a significantly shorter median PFS than patients receiving PD-1 (3.9 vs 5.4 months) in the univariate analysis. Herewith our data oppose the results for median PFS from the prospective randomized phase II ABC trial,5 which showed significantly improved PFS in patients treated with CTLA-4+PD-1 compared with PD-1 (median PFS 5.4 months compared with 2.5 months) for asymptomatic MBM with no prior radiotherapy and no steroid intake. In that study, median PFS for CTLA-4+PD-1 in treatment-naïve patients was not yet reached after 5 years compared with 2.5 months for PD-1. In our real world, not randomized patient cohort, CTLA-4+PD-1 treated patients were younger (patients ≤65 years 58.3% vs 45.3%) and in better performance state (ECOG-PS 0 47.4% vs 31.7%) than patients treated with PD-1, with otherwise equal distribution of prognostically relevant patient characteristics. Therefore, a better outcome in the CTLA-4+PD-1 group in our study would be expected. The difference in median PFS observed by us could be explained by confounding factors, especially by additional SRS. Interestingly, when only assessing the patients treated with SRS, the median PFS in our cohort was similar for systemic therapy with CTLA-4+PD-1, PD-1, and BRAF+MEK inhibition. Patients in arm A and B in the ABC trial did not receive concomitant radiotherapy; therefore, these patients would be better comparable with patients in our cohort who did not receive radiotherapy together with systemic therapy. In patients without radiotherapy for MBM, the median OS rates in our study were in line with the OS results of the ABC trial with median OS for CTLA-4+PD-1 not being reached after 5 years, and 26.1 months for nivolumab in asymptomatic patients. Nevertheless, although our study suggests that patients receiving PD-1 monotherapy and SRS can have survival outcomes comparable with CTLA-4+PD-1, considering the shortcomings of a retrospective analysis, this has to be addressed and assessed in a prospective randomized trial before drawing conclusions from it for daily clinical practice. So far, clinical trials with MBM patients (not addressing effects of concomitant radiotherapy though) show a clearly superior outcome with CTLA-4+PD-1 compared with PD-1 monotherapy.

Toward therapy sequencing, our results suggest that PD-1 based ICB in first-line therapy followed by BRAF+MEK inhibition as second-line therapy shows the best OS outcome in patients with MBM. We could also show that BRAF+MEK inhibition is the most favorable second-line therapy in patients with MBM compared with other therapies such as chemotherapy, but also to CTLA-4+PD-1 and PD-1. This could be due to the fact that in clinical practice BRAF+MEK is often applied to MBM patients with worse prognostic parameters who need a fast-acting therapy and that patients with better baseline characteristics receive ICB. In fact, several baseline characteristics in the group of patients who received targeted therapy as first-line therapy in our study were associated with a worse prognosis compared with the group who received ICB as first-line therapy. It also has to be noted that the patient numbers in the different systemic therapy sequences were not big enough to include them into the multivariate Cox regression analysis with all prognostic parameters of interest. We therefore evaluated the therapy sequences in a separate multivariate analysis focusing on the most important prognostic factors only. Nevertheless, the patient numbers in the single groups were small, limiting its validity. Additionally, it has to be kept in mind that a large number (23.3%) of patients died before receiving a second-line therapy. Interestingly, while the percentage of patients with CTLA-4+PD-1 and CRT was similar in patients who received a second-line therapy compared with those who died after the first-line therapy, a significantly higher fraction of the patients who died before second-line therapy did not show an activating BRAF-mutation and were treated with PD-1 without additional radiotherapy as first-line treatment, while a significantly larger fraction of patients who received a second-line therapy received BRAF+MEK inhibitors and also SRS as first-line treatment. Therefore, although the sequence of BRAF+MEK inhibition as second-line therapy seems to be more favorable, it has to be kept in mind that there is also a fraction of patients who died after first-line therapy comprising a large fraction of BRAF wildtype patients with unfavorable baseline characteristics who received PD-1 monotherapy without additional radiotherapy.

Patients with symptomatic MBM in our study cohort showed a decreased median OS compared with patients with asymptomatic MBM. Of note, in symptomatic patients, median OS on BRAF+MEK inhibition was slightly better compared with PD-1 and CTLA-4+PD-1, while for asymptomatic patients, it was significantly decreased compared with ICB. In the Checkmate 204 trial, a small cohort of 18 symptomatic patients showed a poor intracranial response rate to CTLA-4+PD-1 (22.2% CR, PR or SD >6 months),6 with two of six patients who received corticosteroids showing an objective intracranial response. Median PFS was 1.2 months and median OS 8.7 months in these patients. Similarly, symptomatic patients treated with PD-1 in the ABC trial (n=15) had a median PFS of 2.6 months and median OS of 5.1 months, but an intracranial response rate of only 6%. In comparison, the intracranial objective response rate for dabrafenib + trametinib in the Combi-MB trial was 59% with a short duration of response of 4.5 months and median OS of 11.5 months in 17 symptomatic MBM patients.4 Altogether, in our studied real-world patient cohort, symptomatic MBM patients showed higher intracranial response rates compared with symptomatic MBM patients reported in prospective clinical trials, which could be explained by small numbers of MBM patients in these trials, as well as the additional radiotherapy received in most cases of our cohort. Taken together, these data suggest that targeted therapy with BRAF+MEK inhibition is more favorable as first-line therapy in symptomatic than asymptomatic patients, most probably because of the rapid treatment effect and high response rate. Moreover, our data show that additional SRS is beneficial to achieve an intracranial response and prolonged survival in symptomatic patiens. Additionally, we could show that while the intake of dexamethasone generally impairs the treatment outcome of ICB, as also reported in previous studies,40 41 this negative effect is enhanced by the dose of dexamethasone, with doses of <4 mg resulting in better OS.

It has been reported, mainly in historical cohorts of patients with MBM that surgical resection of single MBM can be beneficial; however, in most studies, a clear impact of MBM surgery on OS could not be shown42 43 or only if compared with patients with no systemic treatment at all.44 In a historic analysis by Eigentler et al, for example, local treatment with SRS or surgical resection was an independently prognostic factor for survival compared with other treatments (WBRT and chemotherapy) in patients with a single brain metastasis. Compared with our study, that study did not separately assess the effects of SRS and metastasectomy.45 One retrospective study showed a statistically relevant survival benefit for patients with surgical resection before immunotherapy versus immunotherapy alone in a smaller heterogeneous cohort with different immunotherapy regimens.46 In our real-world cohort of patients with ICB and targeted therapy, surgical resection of MBM was also significantly associated with improved OS in the univariate analysis, however not in the multivariate analysis. Hence, improved survival may rather be attributable to selection bias, since surgery is often performed in patients with solitary or oligometastatic MBM. Indeed, in our study, solitary metastases were detected in 40.4% of patients with surgical resection versus 26.4% in patients who did not receive surgical resection.

Limitations of this study are its retrospective data evaluation and the high percentage of missing data for some important prognostic parameters of MBM. More comprehensive statistical analyses of subgroups would require a higher total patient number. Therefore, parameters like the timing of SRS before or during systemic therapy could not be included into the multivariate analysis, and an additional multivariate analysis including fewer prognostically relevant parameters had to be performed to assess the independent prognostic value of the therapy sequence. Additionally, the median OS for CTLA-4+PD-1, a type of treatment that had been approved for melanoma later than BRAF+MEK inhibition, has not been reached yet, hereby impairing comparability of groups. Nevertheless, compared with other retrospective studies, we have excluded patients who received previous systemic therapies, and therefore, we are able to show a clear effect of our analyzed parameters on first-line therapy.

In conclusion, this study shows that SRS combined with systemic therapy is beneficial in MBM and should be an integral part of the therapeutic management of patients with MBM. Our results favor the sequence of ICB as first-line and BRAF+MEK as second-line therapy. Also, we could show a possible survival benefit of CRT when combined with systemic therapy.

Since retrospectively analyzed data have always to be treated with caution, because the quality of the documented data is never as high as in a prospectively randomized study, more and larger prospectively conducted clinical studies with an adequately high number of patients with MBM taking into account all prognostically relevant parameters are warranted to further answer the question of the optimal therapy sequence of ICB and BRAF+MEK inhibition together with SRS. Generally, these trials should separately assess different patient scenarios (BRAF-mutated and BRAF-wildtype, oligometastastic MBM versus multiple metastases, asymptomatic versus symptomatic and patients without extracranial organ affection versus few or many affected extracranial organs) and should incorporate SRS with all systemic therapy regimens.