The role of radiotherapy in immunotherapy strategies in the central nervous system

Abstract

The clinical efficacy and relative tolerability of adverse effects of immune checkpoint immunotherapy have led to its increasingly routine use in the management of multiple advanced solid malignancies. Radiation therapy (RT) is well-known to have both local and distant immunomodulatory effects, which has led to extensive investigation into the synergism of these 2 therapies. While the central nervous system (CNS) has historically been thought to be a sanctuary site, well-protected by the blood–brain barrier from the effects of immunotherapy, over the last several years studies have shown the benefits of these drugs, particularly in metastatic disease involving the CNS. This review explores current progress and the future of combination therapy with immune checkpoint inhibitors and RT.

Approximately 60% of patients with solid tumors receive radiation therapy (RT) as part of the management of their disease, whether it be for curative or palliative intent.^1,2 Ionizing radiation induces damage to deoxyribonucleic acid (DNA), either via direct ionization of DNA or indirectly by generation of reactive oxygen/nitrogen species. When DNA damage becomes too significant for repair, cell death occurs via a multitude of mechanisms including mitotic catastrophe and apoptosis, among others. Radiation can also induce cellular senescence, a permanent cell cycle arrest. These mechanisms are largely responsible for the local control benefit of RT.^3,4

In the past several years, a greater understanding of the immunomodulatory effects of RT on the tumor microenvironment (TME) have led to new ideas on how these effects may enhance the efficacy of RT.⁵ The TME consists of tumor cells, stromal cells, endothelial cells, and immune cells that interact through complex signaling networks to foster a permissive niche for tumorigenesis and metastasis, and ultimately frustrating attempts at treatment with therapy.⁶ Knowledge of RT immunogenicity dates back to the 1950s when the abscopal effect was first described, where localized radiation-induced antitumor response throughout the body at sites that were not subjected to targeted RT.⁷ Within and around the irradiated tumor(s), RT can modulate immune signaling within the TME, promoting immune cell recruitment and activation, and triggering immune-mediated cell death.^8,9 RT-induced immunogenic cell death results in a cascade of events, one of which includes the release of danger signals in the form of specific molecules called damage-associated molecular patterns (DAMPs). DAMPs bind to pattern recognition receptors, such as Toll-like receptors, and drive downstream activation of the innate and adaptive immune response.¹⁰ Nevertheless, despite having a plausible mechanism and being reported clinically for nearly 70 years, the abscopal effect remains uncommonly seen in the clinic. It appears to be described most often—though not exclusively—in tumors for which immune checkpoint therapy also has a role to play, such as in melanoma.¹¹

Immune checkpoint inhibitors (ICIs) shift the TME from a state of immunosuppression to a state of immune activation.¹² Cytotoxic T-lymphocyte antigen-4 (CTLA-4) was discovered in 1987—however, its role as an immune checkpoint was not recognized until 1995. The first CTLA-4 inhibitory antibody was developed and tested in animals in 1996, and the first FDA approval was for ipilimumab in 2011. Programmed cell death protein 1 (PD-1) was identified in 1992, and nivolumab, the first PD-1 inhibitor, was approved in 2014. Preclinical data support the notion of combining ICIs with RT, with RT improving the immune response over immune checkpoint blockade alone. CTLA-4 blockade inhibits T-regulatory cells, PD-1 blockade reinvigorates exhausted T cells and drives T-cell expansion, and radiation increases both the number and diversity of the T-cell receptor (TCR) repertoire of tumor infiltrating lymphocytes.¹³ Several prospective trials have investigated the addition of ICIs to RT in solid malignancies. The landmark PACIFIC trial randomized patients with stage III nonsmall cell lung cancer (NSCLC) to adjuvant programmed death ligand 1 (PD-L1) durvalumab or placebo after definitive chemoradiation. Durvalumab resulted in an impressive increase in progression-free survival (PFS) and overall survival (OS).^14–16 Furthermore, in a post hoc analysis of the KEYNOTE-001 phase I trial of programmed cell death protein 1 (PD-1) inhibitor pembrolizumab in NSCLC, patients who previously received RT prior to receiving pembrolizumab experienced an increased median PFS and OS.¹⁷ Concurrent chemoradiation and pembrolizumab for unresectable locoregionally advanced NSCLC is being investigated in Phase 2 KEYNOTE-799 trial,¹⁸ while concurrent durvalumab and chemoradiation are under investigation in a phase III Eastern Cooperative Group Alliance for Clinical Trials in Oncology (ECOG-ACRIN) EA5181 study (with the standard of care adjuvant immunotherapy given in both studies).¹⁹ In CheckMate 577, patients with esophageal or gastroesophageal junction cancer with residual disease after chemoradiation and surgery were randomized to adjuvant PD-1 inhibitor nivolumab or placebo. Patients receiving adjuvant immunotherapy experienced prolonged disease-free survival.²⁰ Several other trials have been completed and remain ongoing to investigate the role of synergism between RT and immunotherapy in several solid cancers.²¹

Historically, patients with central nervous system (CNS) disease were excluded from immunotherapy-based clinical trials due to concerns about blood–brain barrier penetration as well as the thought that the CNS was considered an immune-privileged environment.²² Over the last few years, however, studies have shown favorable outcomes in patients with brain metastases receiving immunotherapy. To this end, we review the role of immunotherapy in CNS disease, with a focus on the role of multi-modality therapy with RT.

Brain Metastases

ICI, with and without additional systemic therapy, has shown promise in the management of brain metastases in recent years. For example, in a phase II trial in patients with untreated metastatic NSCLC or melanoma involving the brain, pembrolizumab demonstrated a 29.7% response rate in cancers with at least 1% PD-L1 expression.²³ Another phase II trial using a combination of ipilimumab (CTLA-4 inhibitor) and nivolumab (PD-1 inhibitor) in patients with melanoma brain metastases showed complete or partial response within the brain in over half of patients.²⁴ Despite these promising results, stereotactic radiosurgery (SRS) remains the accepted standard of care for patients with limited brain metastases over systemic therapy alone or whole brain RT (WBRT), with the benefit over WBRT supported by multiple phase III studies.^25–28 In considering the combination of SRS and immunotherapy, the potential synergism and toxicities of these 2 treatment modalities (considering the immunomodulatory effects of RT as above) is an area of ongoing investigation. Clinicians should consider both efficacy and risk of toxicity when considering multi-modality treatment with SRS and ICI.

Efficacy

The promising independent capabilities of ICI and SRS in the treatment of brain metastases have raised the question of efficacy in combining these 2 modalities. To date, several retrospective analyses have examined this treatment approach, and prospective studies are accruing (Table 1). In a retrospective study of 260 patients with NSCLC, melanoma, or renal cell carcinoma (RCC) brain metastases who were treated with concurrent ICI with SRS (defined as both therapies within a 2-week interval) or SRS alone, the median OS was 24.7 months in the concurrent group vs 12.9 months for patients not receiving ICI; additionally, ICI with SRS predicted for a decreased likelihood of developing 3 or more new brain metastases.²⁹ Several other retrospective analyses report increased intracranial control³⁰ and survival benefit³¹ with the addition of ICI to SRS in advanced lung cancer. Similar findings have been reported for metastatic melanoma as well.³⁰ One single-institution retrospective study of patients with brain metastases from melanoma receiving SRS with and without ipilimumab identified better tumor and edema volume reduction in the presence of concurrent ICI (defined as treatment within 4 weeks of SRS).³² Two other studies showed improved survival with ipilimumab and SRS in melanoma brain metastases.^33,34 A multi-institutional retrospective analysis showed patients who received SRS for new brain metastases within 5.5 months after ipilimumab therapy had better intracranial disease control than those who received SRS later.³⁵ While these data are encouraging, prospective studies on the use of ICI and RT in brain metastases remain ongoing as outlined in Table 1.

Table 1.

Table of Ongoing Prospective Trials With ICI and Concurrent SRS in Brain Metastases

Trial	Phase	NCT	Condition	Arm(s)	Primary Endpoint	Enrollment	Status
Durvalumab (MEDI4736) and radiosurgery (fSRT vs PULSAR) for the treatment of nonsmall cell lung cancer brain metastases	II	NCT04889066	NSCLC	Durvalumab + standard fSRS (three fractions every other day) vs durvalumab + PULSAR (3 fractions every 4 weeks)	Intracranial clinical benefit (CR, PR, or SD by brain-modified RECIST)	46 (estimated)	Active, not yet recruiting
Radiosurgery dose reduction for brain metastases on immunotherapy (RADREMI): a prospective pilot study (RADREMI)	I	NCT04047602	1-10 BMs with biopsy-confirmed primary	Dose reduced SRS based on tumor size concurrent with standard of care ICI	Symptomatic RN	42 (estimated)	Active, recruiting
Single vs multi-fraction SRS patients on immunotherapy (MIGRAINE)	II	NCT04427228	1-10 BMs with biopsy-confirmed primary	Three fraction SRS (27 Gy) vs single-fraction SRS (18–20 Gy)	Rate of RN	74 (estimated)	Active, recruiting
Anti-PD-1 brain collaboration + radiotherapy extension (ABC-X Study) (ABC-X)	II	NCT03340129	Melanoma	Nivolumab + ipilimumab vs nivolumab + ipilimumab with concurrent SRS	Neurological-specific cause of death	218 (estimated)	Active, recruiting
Comparing single vs multiple dose radiation for cancer patients with brain metastasis and receiving immunotherapy (HYPOGRYPHE)	III	NCT05703269	NSCLC, Renal cell carcinoma, breast Carcinoma, Melanoma, NSCLC	Single-fraction SRS vs fSRS over 3-5 treatments	Grade 2+ adverse events	244 (estimated)	Active, recruiting
Stereotactic radiosurgery and immune checkpoint inhibitors With NovoTTF-100M for the treatment of melanoma brain metastases	I	NCT05341349	Melanoma	SRS + pembrolizumab + TTFs vs SRS + nivolumab + ipilimumab + TTFs	Grade 3 + CNS toxicity	10 (estimated)	Active, recruiting

fSRS: fractionated stereotactic radiosurgery; PULSAR: personalized ultra-fractionated stereotactic adaptive radiotherapy; RN: radionecrosis; TTFs: tumor treatment fields.

Regarding the timing of immunotherapy relative to RT and efficacy, it appears that immunotherapy prior to RT may be superior to immunotherapy started after RT, at least as extrapolated from response rates to ipilimumab and SRS in melanoma brain metastases.³⁴ These data are not definitive, however, and in contrast, Knisely et al. noted no difference in survival outcomes between those who started ipilimumab before or after SRS (also in melanoma brain metastases).³³

Systemic corticosteroid use is oftentimes unavoidable during the clinical management of brain metastases, whether used for control of vasogenic cerebral edema, radiation necrosis (RN), seizures, or immune-related events. Corticosteroids suppress the immune system by upregulating T-regulatory cells and inhibiting T effector cell activation, thereby counteracting the effects of checkpoint blockade. Although the use of corticosteroids does not preclude response to immune checkpoint inhibition, with one study even noting response in melanoma brain metastases similar to that of historical controls,³⁶ some evidence suggests that the use of corticosteroids is associated with poorer survival outcomes (both PFS and OS),³⁷ including in brain metastases.³⁸

Toxicity

A concerning adverse effect of SRS is RN, which is particularly concerning when treating larger tumors or postoperative cavities. With SRS alone (ie, not in conjunction with systemic therapy), the risk of RN is dependent on both the treatment volume and volume of normal brain tissue receiving moderate to high doses of radiation. For single-fraction SRS, the volumes of tissue or volumes of the brain receiving at least 12 Gray (Gy) (V12 Gy) are routinely evaluated to assess risks of RN, which can range from much less than 10% to >25% with typically prescribed doses and target volumes.^39–41 When considering SRS in the presence of ICI, potential adverse events remain a significant concern. In addition to evaluating therapeutic efficacy, several of the retrospective analyses that were discussed above also report on toxicity outcomes with combined ICI and SRS. Some have reported an association between ICI and symptomatic RNs in a mixed cohort of patients with brain metastases.⁴² One retrospective study found overall CNS toxicity with combined treatment to be similar to historical controls treated with SRS monotherapy, with melanoma disproportionately represented in high-grade toxicity relative to RCC and NSCLC, possibly as a result of combination with ipilimumab.⁴³ Lehrer et al. conducted an international multicenter retrospective study of 50 patients with RCC brain metastases. Those investigators found any grade RN occurred in 17.4% of those receiving concurrent ICI and SRS, and 22.2% in those receiving nonconcurrent therapy (greater than a 4-week gap between each therapy). These differences were not statistically significant.⁴⁴ A more recent large multicenter cohort study of 657 patients from the same group included patients with NSCLC, melanoma, or RCC primaries; 1- and 2-year rates of any grade RN were 6.4% and 9.9%, respectively. When accounting for the timing of SRS and ICI (less than or greater than a 4-week gap between modalities), there was no difference in rates of RNs. The authors also used recursive partitioning analysis, identifying 3 risk groups based on V12 Gy as the dominant variable predictive of RN.⁴⁵ With regards to melanoma brain metastases specifically, a single institutional retrospective analysis of 151 patients receiving SRS within a 30-day window of ICI administration identified radiographic RN in 9.2% of patients, comparable to historical controls with SRS monotherapy.⁴⁶ Although concurrent SRS and ICI appear safe, clinicians should consider both efficacy and risk for toxicity as we await prospective validation of these findings (Table 1).

Glioblastoma

By the advent of clinically-available ICIs in the early 2010s, immunotherapy as a general treatment strategy for high-grade glioma (HGG) had already been explored for decades. Prior to the registration of the first study of ICI for glioblastoma (GBM) in late 2013 (CheckMate 143⁴⁷), there had already been 78 registered interventional clinical trials using cytokine treatments (such as IL-4, IL-13, and interferon-beta), cellular therapy (dendritic cells), and vaccine therapy (with recombinant viral therapy and conjugate vaccines). Given the initial success seen with ICI in other histologies—primarily in NSCLC and melanoma (with many other histologies to follow)—optimism that ICI would change management of HGG was high. Within the first 3 years of CheckMate 143’s registration in 2013, nearly 20 interventional studies were registered on ClinicalTrials.gov for HGG. Despite initial optimism, however, three major studies—a trio of prospective phase III studies investigating nivolumab—did not demonstrate a survival benefit in GBM (CheckMate 143,⁴⁷ CheckMate 498,⁴⁸ and CheckMate 548⁴⁹), and raised significant questions regarding the efficacy of ICI in HGG. Completed phase III immunotherapy studies for HGG are summarized in Table 2, and ongoing phase III studies are shown in Table 3. It is important to note that while trials are registered to ClinicalTrials.gov, the database does not actively track the status of individual trials and thus the provided information can be out of date. Regardless, it provides a reasonable overview of the trial landscape.

Table 2.

Table of Completed Trials With Immunotherapy in High-Grade Glioma

Trial	Phase	NCT	Condition	Arms	IO plus RT?	Primary Endpoint	N	Outcome
NRG-BN00773	III	NCT04396860	New glioblastoma, MGMT promoter unmethylated	Ipilimumab + Nivolumab + RT + TMZ vs RT + TMZ	Yes	PFS	159 (combination 79 / standard 80)	mPFS dual checkpoint combination 7.7 months vs RT plus TMZ alone 8.5 months
CheckMate 14347	III	NCT02017717	Recurrent glioblastoma	Nivolumab vs bevacizumab (Cohort 2)	No	OS	369 (nivolumab 184/ bevacizumab 185)	mOS nivolumab 9.8 months vs bevacizumab 10 months (P = 0.76)
CheckMate 49848	III	NCT02617589	New glioblastoma, MGMT promoter unmethylated	Nivolumab + RT vs TMZ + RT	Yes	OS	560 (nivolumab + RT 280/ TMZ + RT 280)	mOS nivolumab + RT 13.4 months vs TMZ + RT 14.9 months (HR 1.31, P = 0.0037)
CheckMate 54849	III	NCT02667587	New glioblastoma, MGMT promoter methylated	Nivolumab + RT + TMZ vs Placebo + RT + TMZ	Yes	PFS, OS	716 (nivolumab + RT + TMZ 358/ placebo + RT + TMZ 358)	mOS nivolumab + RT + TMZ 28.9 months vs placebo + RT + TMZ 32.1 months (HR 1.1; 95% CI 0.9–1.3)
PRECISE66	III	NCT00076986	Recurrent glioblastoma	Convection-enhanced delivery (CED) of cintredekin besudotox vs gliadel wafers	No	OS	296	mOS Cintredekin besudotox 9.1 months vs Gliadel wafer 8.8 months (P =.476)
Immunocell-LC67	III	NCT00807027	New glioblastoma	Autologous cytokine-induced killer cells (CIK) + RT + TMZ vs RT + TMZ	Yes	PFS	180 (91 treatment arm/ 89 control)	mPFS treatment 8.1 months vs control 5.4 months (P = 0.0401; HR 0.745) mOS (secondary) treatment 22.47 months vs 16.88 months (P = 0.5237)
Guo et al. IFN-alfa68	III	NCT01765088	New high-grade glioma (supratentorial glioblastoma, gliosarcoma, anaplastic astrocytoma, anaplastic oligodendroglioma, anaplastic oligoastrocytoma)	Interferon alfa + RT + TMZ vs RT + TMZ	Yes	OS	199 (100 treatment arm/99 control)	Grade 3 tumors: mOS treatment arm 39.6 months vs control 29.4 months (P = 0.04) Grade 4 tumors: mOS treatment arm 20.5 months vs control 17.7 months (P = 0.04)
DC-Vax69	III	NCT00045968	New glioblastoma	New glioblastoma: DCVax-L + RT + TMZ vs Placebo + RT + TMZ with crossover to DCVax-L at recurrence	Yes	OS	331 (232 treatment/99 control of which 64 crossed over to DCVax-L on recurrence)	mOS treatment arm 19.3 months
ACT-IV70	III	NCT01480479	New glioblastoma, EGFRvIII mutant	Rindopepimut vaccine with keyhole lymphet hemocyanin (KLH) vaccine adjuvant + adjuvant TMZ following chemoradiation vs Control KLH + adjuvant TMZ following chemoradiation	No	OS	745 (treatment arm 371/ control 374)	mOS treatment arm 20.1 months vs control 20 months (P = 0.93)
Toca 511/ Toca FC71	III	NCT02414165	Recurrent glioblastoma or anaplastic astrocytoma	Toca 511/FC and surgery followed by Toca FC vs surgery followed by standard of care with either lomustine, TMZ, or bevacizumab	No	OS	403 (treatment arm 201/control 202)	mOS treatment arm 11.1 months/ control 12.22 months (P = 0.62)
GLOBE72	III	NCT02511405	Recurrent glioblastoma	VB-111 + bevacizumab vs Bevacizumab monotherapy	No	OS	256	mOS treatment arm 6.8 months/control 7.9 months (P = 0.19)
Cloughesy et al.57	Pilot	NA (Ivy foundation early phase clinical trials consortium)	Recurrent glioblastoma, surgically resectable	Neoadjuvant pembrolizumab with surgery vs adjuvant pembrolizumab following surgery	No	–	35 (neoadjuvant 16/ adjuvant only 19)	mOS neoadjuvant 13.7 months vs adjuvant 7.5 months (P = 0.04)

Table 3.

Table of Ongoing Phase III Trials With Immunotherapy in High-Grade Glioma

Trial	Phase	NCT	Condition	Arms	IO plus RT?	Primary Endpoint	Enrollment	Status
DSP-7888 (synthetic peptide vaccine derived from Wilms’ tumor 1 protein)	III	NCT03149003	Recurrent glioblastoma	DSP-7888 + bevacizumab vs bevacizumab	No	OS	221 (actual)	Completed, pending results
DEN-STEM (trivalent dendritic cells transfected with mRNA from autologous stem cells, survivin, and hTERT)	III	NCT03548571	New glioblastoma, IDH wildtype, MGMT promoter methylated	DC vaccination + RT + TMZ vs RT + TMZ	Yes	PFS	60 (estimated)	Active, not recruiting
ADCTA (autologous dendritic cells derived from PBMNCs cocultured with autologous tumor cells)	III	NCT04277221	Recurrent glioblastoma	DC vaccination + bevacizumab vs bevacizumab	No	OS	118 (estimated)	Recruiting
AV-GBM-1 (autologous dendritic cells loaded with autologous tumor antigens)	III	NCT05100641	New glioblastoma	AV-GBM-1 + RT + TMZ vs autologous monocytes + RT + TMZ	Yes	OS	672 (estimated)	Not yet recruiting

CheckMate 498 investigated the concurrent delivery of nivolumab with RT against concurrent standard alkylator temozolomide (TMZ) with RT, the standard of care in newly diagnosed glioblastoma. The study failed to meet its primary endpoint of improved OS. In CheckMate 548, nivolumab similarly failed to demonstrate a survival benefit when added to standard concurrent and adjuvant TMZ and RT in MGMT promoter hypermethylated new GBMs.⁴⁹ This outcome was unchanged whether stratified by baseline corticosteroid use, tumor baseline PD-L1 expression ≥1%, or tumor baseline PD-L1 expression ≥5%. Taken together, these 2 prospective phase III studies suggest that the combination of monotherapy ICI and RT has limited benefit in GBM—and furthermore, that this is true whether ICI is combined with or used instead of alkylator chemotherapy as the partner to RT.

The results in glioblastoma stand in contrast to the successes seen in other tumor types. Given that immune therapy as a general strategy had not gained significant traction in glioblastoma prior to the advent of checkpoint immunotherapy, perhaps the high hopes for checkpoint blockade were unrealistic. There are several reasons to consider for the lack of response. One possible explanation is that immune checkpoint inhibition, at least against PD-1, is not an appropriate target in glioblastoma. Considering NSCLC and melanoma, both the lungs and the skin function as barrier organs that routinely mount physiologic immune responses throughout life. While immune responses can be mounted in the brain (such as in infectious meningitis), the brain clearly does not suffer external insults to the regularity that the brain and skin do. There is no reason to assume that the checkpoints governing activation of an immune response play the same roles and are balanced to the same thresholds in the brain as lung and skin—and given that the lungs and the skin are far more likely to be actually invaded by external pathogens, one would expect the immune system to have a relatively higher threshold for immune activation in the brain in comparison. Antigen-presenting cells (APCs) are critical for cross-presentation and activation of the adaptive immune response. APCs such as dendritic cells are abundant in the skin. In the lungs there reside alveolar and lung parenchymal APCs capable of activating antigen-primed T cells.⁵⁰ In contrast, microglia, the most abundant resident APC in the brain, is most well-known for actively driving tumor progression by playing an immunosuppressive role and secreting IL-1β.^51,52 CheckMate 143 cohort 2, wherein monotherapy nivolumab was compared against monotherapy bevacizumab in recurrent GBM, suggests PD-1 alone may not be a sufficient target in GBM.⁴⁷ Randomized data have shown that the addition of bevacizumab to the standard of care chemoradiation does not improve OS in GBM (RTOG 0825); failure of nivolumab to demonstrate improvement in OS compared to bevacizumab thus suggests a similar lack of efficacy. A randomized phase II/III NRG Oncology study (NRG-BN007) evaluating the combination of ipilimumab, nivolumab, and RT against standard RT and TMZ in newly diagnosed MGMT promoter unmethylated tumors found no improvement in PFS at the pre-planned phase II analysis (median 7.7 months for combination vs 8.5 months); the study permanently closed after phase II. At the time of writing, no other phase III studies for ICI are active or have been registered for GBM, either with or without RT.

Several efforts have focused on the use of hypofractionated stereotactic radiotherapy (HFSRT) for HGG, which is distinct from the conventional fractionated RT 60 Gy in 30 fractions or moderately hypofractionated regimens of 40 Gy in 15 fractions commonly used in upfront treatment for GBM. As the biological mechanisms of cellular injury and TME effects may be different with HFSRT than with conventional fractionation, there is a potential benefit of adding ICI that would not be realized with conventional fractionation.⁵³ A phase I/II study of HFSRT 24 Gy in three fractions to the 80% isodose line delivered every other day with durvalumab in recurrent GBM is underway in Toulouse, France (STERIMGLI; NCT02866747). Phase I results with 6 patients were reported in 2023,⁵⁴ with a median OS of 16.7 months, and 1 dose-limiting toxicity (vestibular neuritis, attributed to durvalumab); the randomized phase II trial is underway. A separate phase I study of pembrolizumab in combination with bevacizumab and HFSRT 30 Gy in five fractions (NCT02313272) treated 23 recurrent GBMs, with 53% of patients showing at least partial response for at least 6 months, and an OS of 64% at 12 months.⁵⁵ An ongoing phase I/II NRG Oncology study (NRG-BN010) is investigating the efficacy of a combination of atezolizumab (PD-L1 inhibitor), tocilizumab (IL-6R inhibitor), and HFSRT 24 Gy in 3 fractions, in recurrent GBM.⁵⁶

More promising data exist in the form of a pilot study in surgically-amenable recurrent GBM, wherein neoadjuvant PD-1 blockade (pembrolizumab prior to resection) enhanced antitumor immune response compared to PD-1 blockade initiated postresection.⁵⁷ The study randomized 35 patients to PD-1 blockade beginning preresection and continuing postresection (neoadjuvant arm), or PD-1 blockade beginning only following surgery (adjuvant arm). Though this was a pilot study and not powered for efficacy, the observed median OS was longer in the neoadjuvant group at 13.7 months (compared to 7.5 months in the adjuvant arm). This observation was held in the 30 patients with pathologically confirmed tumor progression, the median OS was 13.2 months with neoadjuvant therapy vs 6.3 months with adjuvant.

Despite being a smaller study and needing confirmation of clinical survival findings, this is an important study to highlight given the magnitude of the observed survival difference, the correlation with biologic observations, and the potential explanation it offers for past results. If these clinical outcomes are validated in a powered study, it would be the first study to demonstrate the effectiveness of checkpoint inhibition in HGG. It may also provide mechanistic insight into why ICIs in CheckMate 143, 498, and 548 failed to improve survival. There were additional observations that supported the notion of an immunologically driven response by neoadjuvant PD-1 blockade. The authors observed enhanced clonal expansion of T-cell upregulation of interferon-γ gene expression, and downregulation of cell cycle-related gene expression in the neoadjuvant patients. They also noted increased focal PD-L1 expression within the tumor, decreased PD-1 expression on T cells in the periphery, and a decreased monocyte population. The authors reasoned that a higher antigenic burden in the neoadjuvant arm at the time of PD-1 blockade may have led to a more productive immune response. Indeed, a separate phase II single-arm study also investigating presurgical PD-1 blockade (with nivolumab in a mixed group of 27 recurrent GBMs and 3 newly diagnosed GBMs) also noted increased TCR clonal diversity in tumor-infiltrating T lymphocytes from samples collected postnivolumab.⁵⁸

Sufficient antigenic burden at the time of PD-1 blockade is not the only prerequisite for a productive antitumor response. This concept is supported by the lack of survival benefit seen with monotherapy PD-1 blockade in CheckMate 143, where recurrent GBMs with ample burden of disease were directly treated with nivolumab without surgical resection. It is possible that surgical resection (after neoadjuvant PD-1 blockade) achieves an important second step—removing local immunosuppression. Immune activation may be therefore contingent on both maximizing T-cell activation with PD-1 blockade prior to resection and minimizing local immunosuppression by surgical removal of the immunosuppressive microenvironment afterwards. PD-1 blockade reinvigorates exhausted T cells that have lost effector function.⁵⁹ Reinvigoration may be the most effective when antigen load is highest (before the bulk of the tumor is removed). Surgery subsequently removes the tumor-promoting immunosuppressive compartment, comprising primarily of microglia and blood-derived macrophages. Optimizing the ratio of presurgical T cell activation to postsurgical residual local immunosuppression may then determine the degree to which there is a clinically relevant immune response. While the literature suggests that tumor mutation load is not related to checkpoint immunotherapy response in glioblastoma,⁶⁰ it is important to note that this mutational load as determined at the time of surgery is distinct from that load at the time of checkpoint immunotherapy delivery. In nearly all clinical contexts, the antigen burden is much lower in the latter, as checkpoint immunotherapy is typically delivered only once surgical resection has not only established the diagnosis but also removed a large portion of the tumor in the process. This may account for the apparent lack of correlation with antigenic load.

RT may be able to play a similar role to surgery. One could envision a study where RT was delivered following an adjuvant course of checkpoint immunotherapy. CheckMate 498 and CheckMate 548 both explored the combination of RT with checkpoint inhibition, but not with checkpoint inhibition delivered prior to RT—the key insight may be that the timing of checkpoint inhibition relative to RT is critical. RT holds some advantages over surgery when considered as a partner to neoadjuvant checkpoint immunotherapy. Neoadjuvant checkpoint blockade prior to surgery is well-suited to the recurrent setting, where the diagnosis is already known—in the newly diagnosed setting, committing to checkpoint blockade before a pathologic diagnosis may be problematic. In contrast, neoadjuvant checkpoint blockade followed by RT allows for checkpoint blockade to potentially be delivered after a biopsy. RT also has the potential to modulate the immune response by activating the innate and adaptive immune response.

RT remains, for the foreseeable future, an inseparable cornerstone of the primary management of HGG. The standard by which future immunotherapeutic approaches are judged will be in the setting of combination with RT, and the need for immunotherapy to demonstrate efficacy in this context is of even greater importance than chemotherapy. With a median survival of between 1 to 2 years for GBM with current chemoradiation-based regimens, the present treatment paradigm is one that no one is satisfied with. Checkpoint immunotherapy, at least in its present form, has yet to demonstrably move this needle.

Meningioma

Surgical resection remains the mainstay of treatment for meningioma, with definitive and/or adjuvant RT in cases of an unresectable tumor, suboptimal resection, recurrence after resection, or grade 3 disease (even after complete resection).⁶¹ SRS has emerged as an important primary modality for the treatment of low-grade meningiomas, with local control rates greater than 90%.⁶² Its role in higher grade or recurrent disease is not as well established given poorer local control rates in these scenarios. Furthermore, there remains a limited role for systemic therapy in meningioma, including immunotherapy-based regimens.⁶³ However, patients with meningiomatosis or those who have exhausted local treatment options (unrelenting local failure after multiple surgeries and courses of RT) may potentially benefit from systemic therapies. Excitingly, a recently published single-arm phase II study met its primary endpoint with a 6-month PFS of 48% in patients with recurrent/progressive grade 2/3 meningioma receiving pembrolizumab.⁶⁴ A separate phase II study enrolled 25 patients with grade 2/3 meningioma recurring after surgery/RT. They received nivolumab monotherapy, which was well tolerated, and patients had a 6-month PFS of 42%. Although numerically higher than historical controls, it did not pass their prespecified significance threshold of 51%.⁶⁵ Given the extensive body of work on RT with ICI for enhanced therapeutic efficacy in other malignancies, this technique is being actively investigated in 2 ongoing prospective trials as outlined in Table 4, to hopefully reveal therapeutic benefit in aggressive meningiomas.

Table 4.

Table of Ongoing Prospective Trials With ICI in Meningioma

Trial	Phase	NCT	Condition	Arm(s)	Primary Endpoint	Enrollment	Status
Stereotactic radiosurgery and immunotherapy (pembrolizumab) for the treatment of recurrent meningioma	II	NCT04659811	Recurrent/progressive grade II/III meningioma	SRS with concurrent pembrolizumab	12 month PFS	37 (estimated)	Active, recruiting
Nivolumab and multi-fraction stereotactic radiosurgery with or without ipilimumab in treating patients with recurrent grade II and III meningioma	I/II	NCT03604978	Recurrent/progressive grade II/III meningioma	SRS + nivolumab vs SRS + nivolumab + ipilimumab +	Adverse events, objective response rate	38 (estimated)	Active, recruiting

Conclusion

From a theoretical and preclinical standpoint, the synergistic role of ICI and RT is quite logical. The road to bring these ideas to the clinic for the treatment of CNS disease remains a long one and would benefit most from evidence-based approaches that progressively build understanding. Although ICI with RT appears safe and efficacious in the management of brain metastases of various histologies, larger prospective trials are pending to answer these questions more definitively. In particular, questions remain regarding relative efficacy with ICI delivered before or after SRS, the correlation of treatment-related toxicity and efficacy, and the dose of RT required to deliver optimal treatment benefit in the context of ICI. In primary CNS tumors, a clear time and place for ICI remain to be determined, particularly in its present form. RT will likely remain an integral part of primary CNS cancer care at least in the near term, and we expect there to be continued interest in exploring the intersection of RT with immunotherapy. While we believe that immunotherapy will likely be an important part of the treatment regimen for glioblastoma, further advances in understanding of the mechanisms regulating immune system activation and immunosuppression are required. Insights revealed by studies such as Cloughesy et al.,⁵⁷ showed that neoadjuvant administration of checkpoint immunotherapy may be important, and may help advance understanding. How and when to use ICI continues to be investigated, as studies continue to search for signals that may lead to even a small therapeutic benefit in these devastating diseases with historically poor prognoses.

Supplement sponsorship

This article appears as part of the supplement “Pushing the Boundaries of Radiation Technology for the Central Nervous System,” sponsored by Varian Medical Systems.

Conflict of interest statement

T.J.C.W. reports personal fees and nonfinancial support from AbbVie, personal fees from Cancer Panels, personal fees from Doximity, personal fees and nonfinancial support from Elekta, personal fees and nonfinancial support from Merck, personal fees and nonfinancial support from Novocure, personal fees and nonfinancial support from RTOG Foundation, personal fees from Rutgers, personal fees from the University of Iowa, personal fees from Wolters Kluwer, grants and nonfinancial support from Genentech, grants and nonfinancial support from Varian, personal fees from Iylon Precision Oncology, outside the submitted work. M.T.M. reports royalties from Wolters Kluwer outside the submitted work. M.D.C. reports consulting for Monteris, Inc. and is on the Data Safety Monitoring Board for Biomimetix. All other authors have no relevant conflicts of interest to disclose.