Integrating intracranial hypofractionated radiotherapy with immunotherapy for melanoma brain metastases: therapeutic impact and immune modulation

Abstract

Background:

Combined radiotherapy and immunotherapy can enhance antitumor efficacy, but its effects on melanoma brain metastases (MBM) remain unclear.

Objectives:

To evaluate the intracranial efficacy, safety, and exploratory immune biomarkers of zimberelimab combined with intracranial hypofractionated radiotherapy in patients with MBM.

Methods:

Eligible patients were aged 18–75, had an Eastern Cooperative Oncology Group performance status of 0–2 and measurable brain metastases. Patients received 30–50 Gy/5–10 fractions intracranial hypofractionated radiotherapy and two cycles zimberelimab (240 mg every 3 weeks). The primary endpoint was the intracranial objective response rate (ORR). Secondary endpoints included the disease control rate, overall survival, progression-free survival (PFS), quality of life, safety and potential biomarkers. In addition, serum cytokines were profiled and linked to proteomics-based exploratory prognostic model constructed using the UK Biobank dataset.

Results:

Between November 2022 and February 2025, 10 patients, with median age 60.5 years, completed the treatment, and their data were analyzed. The median intracranial metastatic lesion size was 2.4 cm (range 1.0–5.9 cm). The intracranial ORR was 70%, and intracranial disease control rate was 90%. The median overall survival and intracranial PFS were both 12 months, while the overall PFS was 10.5 months. The patients’ quality of life improved, without grade ⩾3 treatment-related adverse events. Furthermore, post-treatment increases in selected immune-related biomarkers, such as CXCL9, were descriptively associated with treatment response, and baseline levels showed exploratory associations with survival outcomes.

Conclusion:

HFRT plus zimberelimab demonstrated promising intracranial activity, quality of life improvement, and acceptable early safety in patients with MBM and may potentially induce an abscopal antitumor effect. Exploratory immune analyses suggest that baseline immune readiness and therapy-induced immune activation may be associated with treatment benefit.

Design:

This was a prospective, single-arm, open-label phase II study.

Trial registration:

ChiCTR, ChiCTR2200057001. Registered 25 February 2022 (https://www.chictr.org.cn/showproj.html?proj=153144).

Keywords

brain metastases hypofractionated radiotherapy immunotherapy melanoma UK Biobank

Introduction

Melanoma, a type of skin cancer, is characterized by the malignant proliferation of melanocytes. Notably, up to 60% of patients with stage IV melanoma develop brain metastases.¹ Malignant melanoma brain metastases (MBM) pose a significant clinical challenge, as the median overall survival (OS) for patients with MBM is approximately 4 months.² The increasing incidence of MBM has driven the development of immune checkpoint inhibitors (ICIs), such as anti-CTLA-4 and anti-PD-1 antibodies, and targeted therapies, such as BRAF and MEK inhibitors, offering renewed hope for patients.

A phase II trial (NCT02320058) investigating nivolumab plus ipilimumab in patients with MBM reported a central clinical benefit rate of 57%, surpassing previous outcomes with ipilimumab alone.³ Furthermore, observational data from the Netherlands indicate a median OS of 24 months for patients with advanced melanoma receiving anti-PD-1 monotherapy.⁴ Overall, current studies show that immunotherapy can enhance OS, progression-free survival (PFS), and objective response rate (ORR) in MBM. However, further research is needed to optimize these therapies and achieve better outcomes compared to other malignancies.

Hypofractionated radiotherapy (HFRT) is the preferred treatment for patients with a limited number of small intracranial metastases. Combining radiotherapy with ICIs can enhance antitumor responses. Retrospective studies reveal high survival rates for MBM treated with dual immunotherapy (PD-1 and CTLA-4 inhibitors) and stereotactic body radiation therapy (SBRT) or with PD-1 inhibitors and SBRT.⁵ A French study found that among 50 patients with melanoma treated with SRS and anti-PD-1 therapy, only limited neurological symptoms were observed during follow-up, suggesting that the strategic integration of radiotherapy with ICIs may optimize therapeutic efficacy while maintaining a favorable safety profile.⁶

Zimberelimab is a novel, fully human anti-PD-1 monoclonal antibody that effectively blocks the PD-1/PD-L1 interaction, stimulating the immune system to target tumor cells. Early clinical studies suggest that zimberelimab holds promise in treating advanced refractory lymphomas, solid tumors, and recurrent or metastatic cervical cancer, with an acceptable safety profile and manageable adverse effects.^7–9 Its powerful blockade mechanism makes zimberelimab a compelling candidate for combination therapy, potentially enhancing efficacy in challenging cases, such as MBM. Here, we investigated the synergistic effects of combining intracranial HFRT with zimberelimab in treating patients with MBM and evaluated the safety profile of this combined therapy.

Materials and methods

Patients

This prospective, single-arm, phase II trial was conducted at Fujian Cancer Hospital. Eligible patients were aged 18–75 years, had pathologically confirmed stage IV malignant melanoma according to the 8th TNM staging system of the American Joint Committee on Cancer, had at least one measurable brain metastasis detectable through imaging, an Eastern Cooperative Oncology Group (ECOG) performance status score of 0–2, and estimated survival time exceeding 3 months. The primary exclusion criteria were: presence of untreated intracranial disease, history of other malignant tumors, use of immunosuppressants within 2 weeks prior to enrollment, presence of active autoimmune disease or history of autoimmune diseases, and existence of severe cardiovascular disease. See the trial protocol (Supplemental Data) for details regarding the qualifying criteria. Moreover, the participants with melanoma from the UK Biobank dataset who were enrolled in the UKB-PPP consortium were utilized as the sample for exploratory prognostic modeling.

This study was conducted in accordance with the ethical principles of the Declaration of Helsinki and was approved by the Ethics Committee of Fujian Cancer Hospital (Approval Number K2022-077-02). Written informed consent was obtained from all participants prior to study participation, including consent for the use of clinical data for research purposes. The clinical trial was registered with the Chinese Clinical Trial Registry under registration number ChiCTR2200057001. The UK Biobank study protocol was approved by the North West Multi-Center Research Ethics Committee (REC reference: 11/NW/0382), and data were accessed under application number 194370.

Procedures

Before treatment, a comprehensive medical history and assessment of patient quality of life were conducted. Essential evaluations to establish the tumor baseline included a thorough physical examination, hematologic tests, chest and abdominal computed tomography (CT), and brain magnetic resonance imaging (MRI).

Patients received intracranial hypofractionated radiotherapy directed at their brain metastatic lesions, with a prescribed dose of 30–50 Gy administered in 5–10 fractions. Dose and fractionation were selected based on lesion number, size, cumulative tumor volume, and anatomical location, with consideration of proximity to critical structures. Detailed radiotherapy planning procedures, target delineation principles, dose coverage requirements, organ-at-risk constraints, and image guidance protocols are described in the study protocol (Supplemental Data). Concurrently, they received two cycles of zimberelimab at 240 mg via intravenous infusion. The initial dose of zimberelimab was administered 2 days (± 1 day) after starting radiotherapy, and the second dose was administered 3 weeks later. Dose adjustments were allowed to manage drug-related toxicities during treatment. Detailed dosage alteration protocols can be found in Supplemental Data.

During therapy, patients underwent weekly laboratory evaluations, including complete blood cell counts, liver and kidney function tests, and electrolyte measurements. Quality-of-life assessments were conducted every 2 weeks utilizing the ECOG performance status, European Organization for Research and Treatment of Cancer (EORTC) Quality-of-Life Questionnaire (QLQ-C30), and comprehensive physical examinations.¹⁰ After completing treatment, follow-up imaging, including brain MRI and chest and abdominal CT, was scheduled every 12 weeks to assess tumor response. In cases of clinical instability, such as tumor progression requiring immediate intervention, additional imaging could be conducted at any time during the trial. Tumor response was independently evaluated by two experienced radiologists using the Response Evaluation Criteria in Solid Tumors (RECIST; version 1.1), based on brain MRI and enhanced CT of the chest and abdomen.¹¹ Safety was continuously monitored throughout the trial, with adverse events (AEs) evaluated using the Common Terminology Criteria for Adverse Events (CTCAE; version 5.0). Treatment-related AEs (TRAEs) included those attributed to the study treatment or with uncertain causality, and their incidence was reported in both numerical and percentage formats.

Whole blood samples were collected in pyrogen- and endotoxin-free tubes and allowed to clot for 20–30 min at room temperature. The samples were centrifuged at 4°C, 1000×g for 10 min, and the serum fractions were either used immediately or stored at −80°C for future analysis. Human Immune Monitoring 65-Plex ProcartaPlex Panel (Invitrogen, Thermo Fisher Scientific, USA) was used to measure 65 protein targets (APRIL; BAFF; BLC; CD30; CD40L; ENA- 78; eotaxin; eotaxin-2; eotaxin-3; FGF-2; fractalkine; G-CSF; GM-CSF; GRO-alpha; HGF; IFN-α; IFN-γ; IL-10; IL-12p70; IL-13; IL-15; IL-16; IL-17A; IL-18; IL-1α; IL-1β; IL-2; IL-20; IL-21; IL-22; IL-23; IL-27; IL-2R; IL-3; IL-31; IL-4; IL-5; IL-6; IL-7; IL-8; IL-9; IP-10; I-TAC; LIF; MCP-1; MCP-2; MCP-3; M-CSF; MDC; MIF; MIG; MIP-1α; MIP-1β; MIP-3α; MMP-1; NGF-β; SCF; SDF-1α; TNF-β; TNF-α; TNF-R2; TRAIL; TSLP; TWEAK; and VEGF-A). Protein measurements were performed using Luminex xMAP technology, following the manufacturer’s protocol.

In addition, baseline serum samples were analyzed for 65 cytokines based on UK Biobank database. Cytokines were filtered for prognostic relevance and subsequently used to construct a composite cytokine-based risk score for exploratory prognostic modeling.

Endpoints

The primary endpoint was the intracranial ORR assessed 12 weeks after treatment, defined as the proportion of patients achieving complete or partial response. Complete response (CR) was defined as the complete disappearance of all intracranial metastatic lesions on imaging, whereas partial response (PR) was defined as reduction of at least 30% in the sum of the longest diameters of all target lesions, using baseline measurements as a reference, with no new lesions detected.

Secondary endpoints included the intracranial disease control rate (DCR), encompassing CR, PR, and stable disease; OS, defined as the time from the start of treatment to death or last follow-up; intracranial PFS, defined as the time from treatment initiation to intracranial disease progression, death from any cause, or last follow-up; overall PFS, defined similarly but including disease progression; and quality of life and safety assessment. The EORTC QLQ-C30, utilized for quality-of-life assessment, includes five functional scales (physical, role, cognitive, emotional, and social), three symptom scales (fatigue, pain, and nausea/vomiting), six single-item measures (dyspnea, insomnia, appetite loss, constipation, diarrhea, and financial difficulties), and a global health status scale. Higher scores on the global and functional scales indicate better functioning, whereas higher scores on the symptom scales and individual items indicate more severe symptoms. Exploratory outcomes included the expression levels of serum cytokines in the clinical patients and the all-cause mortality among patients with melanoma in UK Biobank cohort, in which OS was defined as the time from enrollment to death from any cause or last follow-up (from May 19, 2007 to May 31, 2024).

Statistical analysis

The sample size was determined a priori using A’ Hern’s exact single-stage phase II design to test whether the intracranial ORR exceeded 20%, assuming a target response rate of 60%, a one-sided type I error of 0.05, and 80% power. Under this design, 10 evaluable patients were required, and ⩾5 intracranial responses (CR/PR) were needed to reject the null hypothesis of an ORR of 20% or lower. Considering a 10% dropout rate among patients who could not be assessed, we recruited 11 participants for this study. The Kaplan–Meier method was used to estimate the OS and PFS, whereas the reverse Kaplan–Meier method was used to estimate the median follow-up time. Group differences were analyzed using a two-tailed paired Student’s t-test, while the Benjamini–Hochberg false discovery rate (FDR) correction was utilized to control the expected proportion of false positives. Descriptive statistics were presented to facilitate interpretation of the results. To facilitate the comparison of scores across various domains in the QLQ-C30, a range normalization method was used for linear transformation, standardizing raw scores to a range of 0–100. Each patient’s maximum toxicity grade for AEs and treatment-relatedness was summarized in tabular form.

For exploratory prognostic modeling, baseline cytokines were initially screened for association with prognosis (favorable: PFS ⩾ 2 months and OS ⩾ 10 months; otherwise poor) based on the Wilcoxon rank-sum test and Cliff’s delta > 0.33.¹² Subsequently, LASSO–Cox regression was applied in the UK Biobank cohort to identify cytokines associated with melanoma-specific mortality and to construct a cytokine-based risk score. In order to ascertain the risk score for each melanoma patient in UK Biobank cohort, the following formula was utilized:

R i s k S c o r e_{i} = \sum_{j = 1}^{n} ({Exp}_{i} \cdot {Coei}_{i j})

The association of the cytokine-based risk score and sixteen clinical covariates with mortality was evaluated using multivariable Cox proportional hazards regression (covariate details are provided in Table S1). Missing clinical data were handled using multiple imputation by chained equations. Five imputed datasets were generated, and estimates were combined using Rubin’s rules. The proportional hazards assumption was examined within each imputed dataset using Schoenfeld residuals, with the cox.zph() function applied to each covariate and to the global model. No consistent evidence of violation was observed across imputations, supporting the validity of the model assumptions.

Given the limited number of events and the potential risk of overfitting, a ridge-penalized Cox model was additionally fitted as a sensitivity analysis to assess predictive stability. Model performance was primarily assessed using bootstrap internal validation with 1000 resamples. Discrimination was quantified by the optimism-corrected Harrell’s C-index. As no independent external dataset was available, model calibration and overall prediction error were assessed internally using optimism-corrected bootstrap calibration (B = 1000) and time-specific Brier scores calculated with inverse probability of censoring weighting (IPCW). A ridge-based nomogram was constructed to provide a visual tool for individualized risk estimation.

All statistical analyses were performed using R (version 4.2.0; R Foundation for Statistical Computing, Vienna, Austria) and GraphPad Prism (version 8.0; GraphPad Software Inc., San Diego, CA, USA). p < 0.05 was considered significant.

The reporting of this study adheres to the CONsolidated Standards of Reporting Trials (CONSORT) statement for clinical trials. A completed CONSORT checklist is provided as Supplemental File.

Results

From November 2, 2022, to February 7, 2025, 20 patients were screened for eligibility and 11 with MBM were ultimately enrolled. One patient withdrew consent prior to treatment, resulting in 10 patients included in the efficacy and safety analyses (Figure S1). All patients underwent baseline imaging to assess tumor characteristics. The median age was 60.5 years (IQR 51–70; 60% female, 40% male). The ECOG performance status was 0 or 1 in all patients. The median size of the intracranial metastatic lesions was 2.4 cm (range 1–5.9 cm). Additionally, all patients had extracranial metastases, involving the lungs, liver, gastrointestinal tract, and abdominal/pelvic lymph nodes. Historical treatment data revealed that 30% of the patients had not received prior immunotherapy. Table 1 summarizes baseline patient characteristics.

Table 1.

Baseline patient and tumor characteristics.

Characteristic	Patients (n = 10)
Median age, years (IQR)	60·5 (51–70)
Sex, female	6 (60%)
ECOG performance status
0	1 (10%)
1	9 (90%)
Primary site
Skin	3 (30%)
Limb	3 (30%)
Mucosa	3 (30%)
Unknown	1 (10%)
Molecular alteration
BRAF mutation	4 (40%)
PD-L1 (+)	3 (30%)
Sum of diameters of intracranial lesions (cm)
Mean (range)	2·4 (1–5·9)
Extracranial metastases
Yes	10 (100%)
No	0
Previous melanoma treatment
Surgery	7 (70%)
Medical therapy	9 (90%)
Radiotherapy	1 (10%)
Immunotherapy received
Yes	7 (70%)
No	3 (30%)

Data are n (%) or median (IQR).

ECOG, Eastern Cooperative Oncology Group.

All 10 patients (100%) completed the planned intracranial HFRT and two cycles of immunotherapy. Among them, nine (90%) received 50 Gy in 10 fractions, while one (10%) received 48 Gy in six fractions. No patient required dose reduction due to zimberelimab-related toxicity.

Clinical responses were evaluated in all patients. At the primary endpoint assessment, 12 weeks post-treatment, the intracranial ORR was 70% (95% CI, 34.8%–93.3%), with two patients achieving CRs and five PRs. Two patients exhibited stable disease, whereas one experienced progressive disease, resulting in an intracranial DCR of 90% (Figure 1(b)). Figure 1(c) shows the tumor response within the cranial irradiation field. After HFRT, all patients showed tumor size reduction: two achieved complete response, seven PR, and one stable disease. Consequently, the intracranial ORR was 90% (95% CI, 55.5%–99.8%), and the intracranial DCR was 100%. We selected imaging from three patients to illustrate these outcomes (Figure 1(d)–(g)). In one patient, brain metastases completely resolved post-treatment. Another patient with multiple intracranial metastases achieved complete disappearance of one lesion and 39% reduction in another. Notably, one patient exhibited an abscopal effect, with significant regression of bowel and mediastinal lymph node metastases following intracranial HFRT and systemic immunotherapy, resulting in partial response.

Figure 1.

Tumor responses of patients in the study cohort. (a) Swimmer plot illustrating the treatment duration for patients who received intracranial hypofractionated radiotherapy in combination with zimberelimab. (b) Maximum change in intracranial tumor size from baseline assessed according to the Response Evaluation Criteria in Solid Tumors (RECIST; version 1.1). (c) Maximum change in the cranial irradiation field from baseline. (d) Pre- and post-treatment cranial MRI of patients with CR. (e) Pre- and post-treatment chest CT of patients with PR. (f) Pre- and post-treatment cranial MRI of patients with PR. (g) Pre- and post-treatment abdominal CT of patients with PR.

The data analysis cutoff was February 7, 2025, with a median follow-up of 13 months (range: 2–23 months) for the cohort. The median OS and intracranial PFS were both 12 months (Figure 2(a) and (b)), whereas the median overall PFS was 10.5 months. During follow-up, four patients developed distant metastases; three subsequently received chemotherapy combined with immunotherapy, whereas one received supportive care. At the time of final analysis, two patients had died. Notably, three patients who had never received immunotherapy prior to enrollment exhibited an intracranial ORR of 100% following treatment, with imaging showing complete resolution of intracranial tumor lesions in two. Consequently, we categorized the patients based on their prior immunotherapy status and separately plotted survival curves for OS and PFS (Figure 2(c) and (d)).

Figure 2.

Kaplan–Meier estimates of survival. (a) Overall survival of the whole cohort. (b) Progression-free survival of the whole cohort. (c) Overall survival of patients classified according to their prior immunotherapy status. (d) Progression-free survival of patients classified according to their prior immunotherapy status.

All patients completed the EORTC QLQ-C30 questionnaire, with changes from baseline to weeks 2 and 4 shown in Figure S2. The mean global health score at baseline was 60/100, which increased to 62.50 following HFRT and further to 65 after the completion of all treatments. The five functional domains exhibited similar steady upward trends, with role functioning showing the most pronounced increase, rising from 71.67 to 83.33. In contrast, scores across the three symptom domains and six individual items generally declined, indicating reduction in symptom burden. Fatigue scores showed the most pronounced decrease, dropping from a baseline of 26.67 to 22.22 post-radiotherapy and subsequently to 15.56 after treatment. However, no significant changes were observed in dyspnea and constipation scores.

Throughout treatment and within a month post-treatment, no patient experienced grade 3 or higher TRAEs (Table 2). The most common grade 1–2 AEs included anemia (grade 1, four cases (40%); grade 2, one case (10%)). Other hematologic TRAEs included decreased lymphocyte count (grade 1: two cases (20%)), decreased platelet count (grade 1: one case (10%), grade 2: one case (10%)), and decreased white blood cell count (grade 1: one case (10%), grade 2: two cases (20%)). All AEs were clinically manageable. Moreover, due to the presence of intracranial metastases, patients often experience dizziness, nausea, vomiting, and peripheral neuropathy during radiotherapy. Notably, one patient experienced grade 2 pancreatic-related AE, evidenced by elevated lipase and amylase levels, potentially associated with the patient’s primary melanoma site. AEs were prospectively collected up to 1 month after treatment. Accordingly, late neurologic toxicities, including radiation necrosis and delayed cerebral edema, were not systematically assessed in this pilot study.

Table 2.

Treatment-related adverse events.

Adverse event	Grade 1	Grade 2
Fatigue	2 (20%)	1 (10%)
Fever	0	1 (10%)
Weight loss	1 (10%)	0
Dizziness	2 (20%)	2 (20%)
Peripheral neuropathy	1 (10%)	2 (20%)
Nausea and vomiting	2 (20%)	0
Constipation	0	2 (20%)
Lymphocyte count decrease	2 (20%)	0
Platelet count decrease	1 (10%)	1 (10%)
White blood cell decrease	1 (10%)	2 (20%)
Anemia	4 (40%)	1 (10%)
Alkaline phosphatase increase	2 (20%)	0
Amylase increase	0	1 (10%)
Lipase increase	0	1 (10%)
Cardiac troponin T increase	1 (10%)	0

Data are n (%).

As part of our exploratory analysis, we measured serum cytokine levels in each patient using a 65-plex cytokine array. We compared the data before and after treatment, revealing variations in the expression levels of different cytokines. Each cytokine is represented in a heatmap using different colors according to its relative expression level, with red indicating high expression and blue indicating low expression (Figure 3(a)). Post-treatment, several cytokines, predominantly proinflammatory and chemotactic factors, showed nominally significant increases. However, none remained significant after adjustment for multiple comparisons using a strict FDR threshold of p < 0.05. Fifteen cytokines met a more permissive threshold of FDR-adjusted p < 0.20 (Figure 3(b)). Among Th1-associated cytokines, IFN-γ demonstrated a mean paired increase of 0.177 ± 0.220 pg/mL (fold change 1.75; p = 0.031; FDR-adjusted p = 0.167). CXCL10 showed a mean increase of 0.341 ± 0.273 pg/mL (fold change 1.26; p = 0.003; FDR-adjusted p = 0.096), while CXCL9 increased by 0.324 ± 0.404 pg/mL (fold change 1.22; p = 0.032; FDR-adjusted p = 0.167). Interleukins, including IL-1α, IL-1β, IL-9, IL-15, IL-22, IL-4, and IL-17, consistently increased (fold change ranged from 1.22 to 10.51, FDR-adjusted p < 0.20, Figure 3(b)). Colony-stimulating factors, including GM-CSF and M-CSF, displayed similar upward trends (Paired differences were 2.153 ± 2.421 pg/mL and 4.781 ± 5.758 pg/mL, FDR-adjusted p < 0.20, Figure 3(b)). Furthermore, in patients with varying RECIST responses, M-CSF, IL-18, CXCL9, and CXCL10 expression levels showed a notable trend. Patients achieving CR tended to show greater increases in these cytokines compared with other response groups. Complete cytokine statistics are provided in Table S2.

Figure 3.

Expression differences of cytokines between pre- and post-treatment. (a) Heatmap representation of cytokine expression in patients at pre- and post- treatment. (b) Expression levels of various cytokines measured at baseline and post-treatment. p values were adjusted for multiple testing using the Benjamini–Hochberg FDR method.

In exploratory analyses within the UK Biobank cohort, 31 cytokines showed significant between-group differences according to the prespecified prognostic classification, indicating substantial heterogeneity in pretreatment immune states (The distribution of cytokine expression abundance was illustrated in Figure S3). From these, 16 cytokines retained non-zero coefficients in the LASSO–Cox model and were incorporated into a composite risk score derived from 197 UK Biobank participants, among whom 31 deaths occurred during a median follow-up of 14.4 years (range, 0.8–17.2 years). In the multivariable Cox model, the cytokine-based risk score remained significantly associated with OS. Each one-unit increase in the risk score was associated with a higher risk of death (pooled HR 5.84, 95% CI 3.10–10.98; p < 0.001). Among selected clinical covariates, body mass index was inversely associated with mortality (HR 0.87, 95% CI 0.78–0.96; p = 0.007), while other covariates, including albumin levels and neutrophil-to-lymphocyte ratio, were not significantly associated with mortality, as shown in Table 3.

Table 3.

Multivariable Cox proportional hazards analysis of overall survival in the UK Biobank melanoma cohort.

Variable	Category	Pooled HR (95% CI)	p-Value
Risk score	Per unit increase	5.84 (3.10–10.98)	<0.001
Age	Per year increase	1.02 (0.95–1.11)	0.580
Sex	Female	1.00 (reference)	–
Sex	Male	1.48 (0.54–4.02)	0.448
Employment status	Unemployed/retired	1.00 (reference)	–
Employment status	Employed	1.10 (0.37–3.29)	0.860
Education attainment	Per level increase	0.96 (0.87–1.07)	0.465
Townsend deprivation index	Per unit increase	1.12 (0.96–1.30)	0.144
Household income	Low	1.00 (reference)	–
	Middle	0.91 (0.23–3.56)	0.891
	High	0.64 (0.13–2.93)	0.561
Tobacco use	Never	1.00 (reference)	–
	Ever	1.48 (0.64–3.41)	0.358
Use of sun/UV protection	Low	1.00 (reference)	–
	High	0.88 (0.36–2.12)	0.770
Facial aging	Younger than you are	1.00 (reference)	–
	About your age	1.00 (0.39–2.56)	0.997
Outdoor light exposure time	Per unit increase	0.98 (0.76–1.25)	0.857
Body mass index	Per kg/m² increase	0.87 (0.78–0.96)	0.007
Albumin level	Per g/L increase	0.96 (0.80–1.15)	0.665
Neutrophil-lymphocyte ratio	Per unit increase	0.77 (0.52–1.12)	0.172

Hazard ratios (HRs) and 95% confidence intervals (CIs) were pooled across five multiply imputed datasets using Rubin’s rules.

CI, confidence interval; HR, hazard ratios.

In the ridge-penalized Cox model, bootstrap internal validation yielded an optimism-corrected Harrell’s C-index was 0.763 (95% CI 0.669–0.844). Prediction error increased over time, as reflected by time-specific IPCW Brier scores of 0.0085, 0.1658, and 0.1838 at 1, 3, and 5 years, respectively. As shown in Figure S4, bootstrap calibration suggested reasonable short-term agreement between predicted and observed risks, with mean absolute errors of 0.011, 0.033, and 0.06 at 12, 36, and 60 months, respectively. A ridge-based nomogram is provided in Figure S5 for visualization of model-derived risk estimate.

Discussion

Retrospective analyses have demonstrated the potent radio-sensitizing effects of combining radiotherapy with ICIs, highlighting their synergy and abscopal effects as an optimal therapeutic paradigm. However, prospective studies investigating the combination of immunotherapy and HFRT in MBM are lacking. We addressed this gap by assessing the efficacy and safety of HFRT combined with zimberelimab. Our results demonstrate that this combination therapy achieves notable ORR and DCR in intracranial metastases, with median survival outcomes showing potential superiority over current standard treatments. The TRAEs were grade 1–2 within the predefined safety monitoring window, reflecting favorable early tolerability, and patients reported improved quality of life following treatment. Furthermore, a cytokine risk score model was constructed using UK Biobank data, thereby providing a potential prognostic indicator for melanoma patients. These findings offer clinically relevant insight for the therapeutic management of MBM and generate hypotheses regarding the potential interplay between baseline immune status and treatment responsiveness.

In recent years, researchers have explored the potential of combining ICIs with radiotherapy. However, most studies have been limited to retrospective analyses or case studies. A retrospective analysis involving 46 patients with MBM found that patients receiving stereotactic radiation during or before ipilimumab treatment had better OS than those treated with radiation after ipilimumab (1-year OS: 65% vs 56% vs 40%; p = 0.008). Additionally, those receiving radiation during ipilimumab treatment had improved local control (1-year local recurrence rate 0%, p = 0.21) and reduced extracranial recurrences compared to other groups (1-year regional recurrence rate 69% vs 64% vs 92%, p = 0.03).¹³ Similarly, combining stereotactic radiation with nivolumab significantly improved tumor control with a 70% ORR and median survival of 12 months, alongside manageable side effects.¹⁴ Despite these promising outcomes, prospective evaluations remain limited. We aimed to fill this knowledge gap by prospectively evaluating the efficacy and safety of combining HFRT with zimberelimab in MBM. The patients had at least one radiographically visible intracranial metastasis. Our findings indicated that this combination therapy achieved a 70% intracranial ORR and 90% intracranial DCR. The median OS and intracranial PFS for all patients were both 12 months. Essentially, ongoing follow-up may reveal a longer median OS. Moreover, no grade 3 or higher AEs were reported, and quality of life improved post-treatment across several domains, including overall health and cognitive functioning, as measured with the EORTC QLQ-C30.

Therefore, identifying biomarkers predicting which patients are likely to derive the greatest benefit from immunotherapy is essential. A prospective, multicenter international clinical study analyzed 101 patients with melanoma and found that higher expression of IFN-γ was associated with a favorable treatment response. This research demonstrated that IFN-γ significantly upregulates the expression of various HLA class I and II genes, particularly HLA-E, and other genes involved in the antigen presentation pathway, including NLRC5 and CIITA.¹⁵ This enhancement of antigen presentation enables the immune system’s ability to more effectively recognize and target tumor cells. Additionally, exposure to IFN-γ leads to increased expression of chemokines, including CXCL9, CXCL10, and CXCL11, attracting more T-cells to the tumor site.^16,17 In our exploratory analysis, IFN-γ increased at the nominal level post-treatment, and patients with elevated IFN-γ appeared to have more favorable responses. Conversely, patients with progressive disease exhibited decreased IFN-γ levels after treatment. A similar trend was observed for CXCL9, CXCL10, and CXCL11, with most patients showing increases in expression following treatment compared to baseline at the nominal significance level (Figure 3). Notably, the magnitude of the increase was greater in patients with CR. For instance, the mean pre- and post-treatment differences in CXCL11 levels were 62.4 for CR, 32.6 for PR, and 12.3 for stable disease. Conversely, these markers decreased in patients with progressive disease after treatment, suggesting a potential link between treatment response and enhanced antigen presentation, along with increased T-cell recruitment. However, these findings are hypothesis-generating, and confirmation in independent prospective cohorts is required. Moreover, it has been reported that IFN-γ inhibits angiogenesis by affecting endothelial cell proliferation and survival, thereby inducing tumor stroma ischemia.^18,19 CXCL10 additionally inhibits tumor angiogenesis by suppressing the proliferation of endothelial cells.²⁰ Future studies in larger cohorts and with predefined hypotheses are warranted to validate these cytokine dynamics and clarify their clinical relevance in immunotherapy.

As immunomodulators, interleukins regulate leukocyte activity, promote cell proliferation, and play roles in antibody responses, hematopoiesis, and tumor surveillance. The therapeutic potential of interleukins has been a focal point in foundational and translational cancer research.²¹ IL-2 was the first interleukin approved for cancer treatment and can induce long-lasting complete responses in some patients with melanoma and renal cell carcinoma.²² Similarly, IL-15 is known for its ability to activate and proliferate T cells and natural killer cells. A phase Ib clinical study investigated the combination of ALT-803, an IL-15 agonist, and nivolumab in metastatic non-small cell lung cancer, revealing an ORR of 29% and DCR of 76%, even in patients resistant or recurrent to anti-PD-1 therapy.²³ More recently, the IL-9 signaling pathway has been implicated in tumor immunity. Studies have indicated a positive correlation between increased physiological Th9 cell counts and improved clinical responses in patients with metastatic melanoma undergoing nivolumab treatment. In patients treated with nivolumab, Th9 cells facilitated melanoma-specific CD8+ T cell-mediated antitumor functions.²⁴ Similarly, there was a significant increase in IL-9 expressing Th9 cells in peripheral blood and circulating CD4+ T cells of patients with breast cancer, enhancing CD8+ T cell-mediated cytotoxicity.²⁵ Furthermore, studies have shown that IL-9 serves as a potential predictive biomarker to evaluate and predict the efficacy of adoptive TIL therapy in melanoma, with elevated levels of IL-9 production being closely linked to optimal therapeutic outcomes.²⁶ Here, we found that the expression levels of several interleukins, including IL-1, IL-9, and IL-15, showed an upward trend after treatment. Immunotherapies targeting PD-1/PD-L1 have revolutionized the treatment landscape for melanoma. Identifying effective biomarkers for immunotherapy will help accurately predict and select patients who will benefit, thereby expanding the reach of immunotherapy to a wider range of patients with cancer.

Interestingly, several cytokines highlighted in our exploratory UK Biobank-aligned prognostic model (including M-CSF, CXCL10, BLC, CXCL9, IL-17A, IL-4) overlapped with those showing post-treatment increases in our cohort (Figure S6). This pattern aligns with previous reports suggesting that, in certain contexts, specific cytokines may demonstrate both treatment-induced immune activation and baseline prognostic information, rather than representing a fixed phenotype.²⁷ This interpretation is further supported by multi-cohort analyses reporting associations between pre-treatment inflammatory or interferon-related immune features and durable benefit from immunotherapy.^28–30 Taken together, these exploratory findings generate the hypothesis that both baseline immune readiness and treatment-induced immune activation may be associated with therapeutic benefit and require validation in larger cohorts.

Additionally, we observed a noteworthy instance of the abscopal effect, a phenomenon where tumors located outside the direct irradiation field shrink or regress. In 1953, British scientist Mole introduced the term “abscopal effect.³¹” In 2012, Postow et al. documented its first occurrence in melanoma treatment, where radiotherapy combined with ipilimumab induced systemic tumor regression, including in distant lesions.³² With the rise of immunological research, clinical studies focused on the abscopal effect have increasingly garnered attention. However, the mechanisms behind this phenomenon are not fully understood. A systematic review of 46 cases found higher incidences of the abscopal effect in immunogenic tumors, such as lymphoma (37%, 17/46), renal cell carcinoma (15%, 7/46), and melanoma (15%, 7/46).³³ This disparity highlights the differences in immunogenicity among tumor pathologies, which may influence the occurrence rate of the abscopal effect.

Here, one patient exhibited systemic disease progression, including brain, intestinal wall, and multiple lymph node metastases, after five cycles of toripalimab treatment. Subsequently, enrolled in our clinical trial, the patient demonstrated a classic abscopal effect 2 months after initiating treatment, as evidenced by the significant regression of intestinal wall metastases and enlarged mediastinal lymph nodes (Figure 1(f) and (g)). According to RECIST v1.1, this response was classified as partial. Although the abscopal effect remains rare, it provides a crucial basis for further exploration of how combined therapeutic strategies can more effectively activate systemic antitumor immunity.

From a translational perspective, intracranial HFRT combined with PD-1 blockade may provide a practical foundation for future combination strategies. Newer checkpoint targets such as LAG-3 inhibition, perioperative immune-priming approaches, and molecularly defined subsets with potential DNA-repair vulnerability may represent directions for further investigation.^34–36 Although these strategies remain exploratory in the setting of melanoma brain metastases, our results support continued evaluation of immunoradiotherapy within carefully designed combination and sequencing studies.

This study has some limitations. First, due to its single-arm design and single-institution setting, the generalizability of the results is limited. Second, the number of patients enrolled was relatively small, reflecting the low incidence of MBM. However, the enrolled patients demonstrated favorable prognoses, including extended survival and disease control. Thirdly, AEs were captured only within 1 month after treatment completion, which may underestimate late neurologic toxicity. Future studies could include longer follow-ups and dedicated neurotoxicity surveillance. Lastly, our exploratory study only investigated the differential expression of inflammatory mediators related to immune cell activation. Future studies could involve a more comprehensive exploration of the tumor microenvironment.

Conclusion

Our findings suggest that combining HFRT and zimberelimab was associated with encouraging intracranial control rates in patients with MBM, thereby enhancing their quality of life. This combination therapy demonstrated favorable early tolerability and may potentially induce an abscopal antitumor effect. Additionally, exploratory analyses revealed treatment-related changes in immune-related inflammatory mediators, supporting a working hypothesis that both baseline immune readiness and therapy-induced immune activation may be associated with treatment response.

Supplemental Material

sj-docx-1-tam-10.1177_17588359261444616 – Supplemental material for Integrating intracranial hypofractionated radiotherapy with immunotherapy for melanoma brain metastases: therapeutic impact and immune modulation

Supplemental material, sj-docx-1-tam-10.1177_17588359261444616 for Integrating intracranial hypofractionated radiotherapy with immunotherapy for melanoma brain metastases: therapeutic impact and immune modulation by Jing Lin, Ting Zhang, Yibin Zeng, Lirui Tang, Jieyuan Cai, Ling Chen, Jiazhen Fang, Yu Chen and Jinluan Li in Therapeutic Advances in Medical Oncology

Supplemental Material

sj-docx-2-tam-10.1177_17588359261444616 – Supplemental material for Integrating intracranial hypofractionated radiotherapy with immunotherapy for melanoma brain metastases: therapeutic impact and immune modulation

Supplemental material, sj-docx-2-tam-10.1177_17588359261444616 for Integrating intracranial hypofractionated radiotherapy with immunotherapy for melanoma brain metastases: therapeutic impact and immune modulation by Jing Lin, Ting Zhang, Yibin Zeng, Lirui Tang, Jieyuan Cai, Ling Chen, Jiazhen Fang, Yu Chen and Jinluan Li in Therapeutic Advances in Medical Oncology

Footnotes

Acknowledgements

We would like to thank Editage () for English language editing.

Declarations

ORCID iDs

Ting Zhang

Jieyuan Cai

Jinluan Li

Supplemental material

Supplemental material for this article is available online.

References

Lamba

Wen

Aizer

AA.

Epidemiology of brain metastases and leptomeningeal disease. Neuro-oncology 2021; 23(9): 1447–1456.

Eroglu

Topcu

, et al. How I treat brain metastases of melanoma. ESMO Open 2022; 7(6): 100598.

Tawbi

Forsyth

Algazi

, et al. Combined nivolumab and ipilimumab in melanoma metastatic to the brain. N Engl J Med 2018; 379(8): 722–730.

van Zeijl

MCT

Haanen

Wouters

, et al. Real-world outcomes of first-line anti-PD-1 therapy for advanced melanoma: a nationwide population-based study. J Immunother 2020; 43(8): 256–264.

Chen

Gao

Huang

, et al. SBRT combined with PD-1/PD-L1 inhibitors in NSCLC treatment: a focus on the mechanisms, advances, and future challenges. J Hematol Oncol 2020; 13(1): 105.

Carron

Gaudy-Marqueste

Amatore

, et al. Stereotactic radiosurgery combined with anti-PD1 for the management of melanoma brain metastases: a retrospective study of safety and efficacy. Eur J Cancer 2020; 135: 52–61.

Liu

, et al. Dose escalation and expansion (phase Ia/Ib) study of GLS-010, a recombinant fully human antiprogrammed death-1 monoclonal antibody for advanced solid tumors or lymphoma. Eur J Cancer 2021; 148: 1–13.

Xia

Zhou

, et al. Gls-010 (Zimberelimab), a novel fully human Anti-Pd-1 mab in chinese patients with recurrent/metastatic cervical cancer: results from a multicenter, open-label, single-arm phase Ii trial. Meeting abstract. Int J Gynecol Cancer 2020; 30: A147–A147.

Lin

Zhang

Bai

, et al. Efficacy and safety of GLS-010 (zimberelimab) in patients with relapsed or refractory classical Hodgkin lymphoma: a multicenter, single-arm, phase II study. Eur J Cancer 2022; 164: 117–126.

10.

Aaronson

Ahmedzai

Bergman

, et al. The European organization for research and treatment of cancer QLQ-C30: a quality-of-life instrument for use in international clinical trials in oncology. J Natl Cancer Inst 1993; 85(5): 365–376. doi:10.1093/jnci/85.5.365

11.

Eisenhauer

Therasse

Bogaerts

, et al. New response evaluation criteria in solid tumours: revised RECIST guideline (version 1.1). Eur J Cancer 2009; 45(2): 228–247.

12.

Yao

KME

. Using Cliff’s delta as a non-parametric effect size measure: an accessible web app and R tutorial. Pract Assess 2024; 29(2): 1–12. doi:10.7275/pare.1977

13.

Kiess

Wolchok

Barker

, et al. Stereotactic radiosurgery for melanoma brain metastases in patients receiving ipilimumab: safety profile and efficacy of combined treatment. Int J Radiat Oncol Biol Phys 2015; 92(2): 368–375.

14.

Ahmed

Stallworth

Kim

, et al. Clinical outcomes of melanoma brain metastases treated with stereotactic radiation and anti-PD-1 therapy. Ann Oncol 2016; 27(3): 434–441.

15.

Grasso

Tsoi

Onyshchenko

, et al. Conserved interferon-γ signaling drives clinical response to immune checkpoint blockade therapy in melanoma. Cancer Cell 2020; 38(4): 500–515.e3.

16.

Gocher

Workman

Vignali

DAA

. Interferon-γ: teammate or opponent in the tumour microenvironment? Nat Rev Immunol 2022; 22(3): 158–172.

17.

Hoch

Schulz

Eling

, et al. Multiplexed imaging mass cytometry of the chemokine milieus in melanoma characterizes features of the response to immunotherapy. Sci Immunol 2022; 7(70): eabk1692.

18.

Briesemeister

Sommermeyer

Loddenkemper

, et al. Tumor rejection by local interferon gamma induction in established tumors is associated with blood vessel destruction and necrosis. Int J Cancer 2011; 128(2): 371–378.

19.

Zhang

Hou

Wei

, et al. Lack of interferon-γ receptor results in a microenvironment favorable for intestinal tumorigenesis. Oncotarget 2016; 7(27): 42099–42109.

20.

Haabeth

Lorvik

Hammarström

, et al. Inflammation driven by tumour-specific Th1 cells protects against B-cell cancer. Nat Commun 2011; 2: 240.

21.

Briukhovetska

Dörr

Endres

, et al. Interleukins in cancer: from biology to therapy. Nat Rev Cancer 2021; 21(8): 481–499.

22.

Siefker-Radtke

Cho

Diab

, et al. Bempegaldesleukin plus nivolumab in first-line metastatic urothelial carcinoma: results from PIVOT-02. Eur Urol 2022; 82(4): 365–373.

23.

Wrangle

Velcheti

Patel

, et al. ALT-803, an IL-15 superagonist, in combination with nivolumab in patients with metastatic non-small cell lung cancer: a non-randomised, open-label, phase 1b trial. Lancet Oncol 2018; 19(5): 694–704.

24.

Nonomura

Otsuka

Nakashima

, et al. Peripheral blood Th9 cells are a possible pharmacodynamic biomarker of nivolumab treatment efficacy in metastatic melanoma patients. Oncoimmunology 2016; 5(12): e1248327.

25.

You

Zhang

Cui

, et al. Th9 cells promote antitumor immunity via IL-9 and IL-21 and demonstrate atypical cytokine expression in breast cancer. Int Immunopharmacol 2017; 52: 163–167.

26.

Forget

Haymaker

Hess

, et al. Prospective analysis of adoptive TIL therapy in patients with metastatic melanoma: response, impact of anti-CTLA4, and biomarkers to predict clinical outcome. Clin Cancer Res 2018; 24(18): 4416–4428.

27.

Pedersen

Sokac

Sørensen

, et al. Increased soluble PD-1 predicts response to nivolumab plus ipilimumab in melanoma. Cancers (Basel). 2022; 14(14): 3342.

28.

Wang

Zheng

Shao

, et al. Development of a prognostic composite cytokine signature based on the correlation with nivolumab clearance: translational PK/PD analysis in patients with renal cell carcinoma. J Immunother Cancer 2019; 7(1): 348.

29.

Virazels

Lusque

Brayer

, et al. TNF signature in advanced melanoma patients treated with immune checkpoint inhibitors: results from the MELANFα clinical study. Int J Cancer 2025; 157(3): 534–548.

30.

Ding

, et al. Serum cytokine analysis in a cohort of advanced non-small cell lung cancer treated with PD-1 inhibitors reveals predictive markers of CXCL12. Front Immunol 2023; 14: 1194123.

31.

Mole

RH.

Whole body irradiation; radiobiology or medicine?

Br J Radiol 1953; 26(305): 234–241.

32.

Postow

Callahan

Barker

, et al. Immunologic correlates of the abscopal effect in a patient with melanoma. N Engl J Med 2012; 366(10): 925–931.

33.

Abuodeh

Venkat

Kim

Systematic review of case reports on the abscopal effect. Curr Probl Cancer 2016; 40(1): 25–37.

34.

Tawbi

Schadendorf

Lipson

, et al. Relatlimab and nivolumab versus nivolumab in untreated advanced melanoma. N Engl J Med 2022; 386(1): 24–34.

35.

Patel

Othus

Chen

, et al. Neoadjuvant-adjuvant or adjuvant-only pembrolizumab in advanced melanoma. N Engl J Med 2023; 388(9): 813–823.

36.

Stylianakis

Benjamin

, et al. Comprehensive analysis of BRCA1/2 mutations, “BRCAness” and PARP inhibitors in melanoma. Crit Rev Oncol Hematol 2025; 215: 104962.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

1.23 MB

0.04 MB