Cancer immunotherapy in patients with brain metastases

Abstract

The exclusion of “real-world” patients from registration clinical trials of cancer immunotherapy represents a significant emerging issue. For instance, a large fraction of cancer patients develops brain metastases during the course of the disease, but results from large prospective clinical trials investigating this considerable proportion of the cancer patient population are currently lacking. To provide a useful tool for the clinician in a “real-world” setting, we have reviewed the available literature regarding the safety and efficacy of immune check-point inhibitors in patients with cancer metastatic to the brain. Overall, these data provide encouraging evidence that these therapeutic agents can induce intracranial objective responses, particularly in patients with asymptomatic and previously untreated brain metastases. Larger prospective studies are needed to confirm these initial results.

Introduction

Targeting inhibitory signals which constrain the host immune response may hold the key for the long-term survival of patients with widely metastatic cancers [1]. Blockade of these “immune checkpoints”, such as PD-1 and CTLA-4, has demonstrated impressive clinical activity in several types of solid tumors, in particular melanoma, lung cancer, and renal cancer [2–5]. However, the first registration clinical trials, including CheckMate 057 [2], CheckMate 017 [6], and OAK [7] in non-squamous and squamous non-small cell lung cancer (NSCLC); CheckMate 067 [3] and KEYNOTE-006 [8] in melanoma; and IMvigor210 [9] in bladder cancer, excluded patients with active brain metastases from enrollment due to the associated poor prognosis [10, 11]. Therefore, this substantial group of patients has gone largely unstudied.

Characteristics of patients with brain metastases

In “real-world” clinical practice, patients with cancer are often diagnosed with disease metastatic to the brain. It is estimated that over 25% of patients have already developed brain metastases when diagnosed with stage IV melanoma [12], and that 50% of patients develop brain metastases during the course of this disease [13]. The incidence of brain metastases in unselected populations of patients affected by various types of cancer (breast, melanoma, lung, renal, and colorectal) ranges from approximately 8 to 10% [14]. The most common histology found in metastatic brain tumors is lung cancer (either squamous or non-squamous), which alone accounts for approximately 20% of all cases. However, common primary sources of brain metastases also include melanoma, renal, breast, and colorectal cancers [15]. Brain metastases are most commonly localized to the cerebral hemispheres, followed by the cerebellum and brainstem [16].

Patients may present with neurological symptoms, depending on the number, anatomical location, presence and extension of perilesional edema, and size of lesions. In addition to the aforementioned factors, prior local radiotherapy and the need for steroid treatment may further affect the prognosis and the response to treatment of brain metastases [17–19]. The use of immunosuppressive doses of steroids (> 10 mg prednisone/equivalent per day) is currently a mainstay of symptomatic treatment for brain metastases [20].

The blood–brain barrier represents a complicated hurdle to overcome for many systemic therapies, including several forms of chemotherapy and targeted therapies [21, 22]. Thus, most patients are treated with surgery and radiotherapy as systemic treatments have limited efficacy. Historically, the median OS of unselected patients with metastatic brain tumors has been very poor, typically in the range of 2–4 months [23], but these figures may be changing due to the recent introduction of new and more effective therapeutic strategies.

Anti-CTLA-4 cancer immunotherapy in patients with brain metastases

Two major clinical programs evaluating anti-CTLA-4 antibodies, ipilimumab (ipi) and tremelimumab, have been reported. Ipi, a fully human IgG1 monoclonal antibody, was the first immune check-point inhibitor to be approved by major regulatory agencies. It is currently approved by the US Food and Drug Administration (FDA) for the treatment of unresectable or metastatic melanoma in monotherapy or in combination with nivolumab (nivo), and as an adjuvant treatment for completely resected high-risk stage III melanoma. Therefore, clinical results documenting the responses of patients with brain metastases to anti-CTLA-4 antibodies are only available from melanoma (case reports are not reviewed here).

Weber et al. [24] described a retrospective analysis of a phase II trial, wherein 12 of the 115 patients with unresectable or metastatic melanoma had stable brain metastases at baseline and were treated with ipi at 10 mg/kg (dose approximately 3 times higher than approved by FDA for metastatic melanoma, given every 3 weeks for up to 4 doses). A partial response (PR) was achieved in 2 of the 12 patients, and 3 patients achieved stable disease (SD) (objective response rate (ORR): 16.7% by modified World Health Organization criteria, not reported whether intra or extracranial). The median OS of this cohort was 14 months (range 2.7–56.4 +). Margolin et al. [25] completed a phase II study that enrolled 72 patients with melanoma metastatic to the brain from 10 US centers. This study utilized a treatment dose and schedule identical to the aforementioned study (10 mg/kg every 3 weeks for up to 4 doses). Neurologically asymptomatic patients not receiving corticosteroid treatment at study entry were placed into cohort A (n = 51), whereas symptomatic patients on a stable dose of corticosteroids formed cohort B (n = 21). In cohort A, the intracranial ORR by immune-related response criteria (irRC) was 16% and the brain disease control rate (DCR—defined as CR (complete response), PR, or SD 12 weeks after treatment initiation) was 24%. The systemic ORR by irRC of cohort A was 10%. In cohort B, the brain DCR was 10%, and 1 patient achieved an intracranial objective response (OR) (ORR 5%). It is currently unknown whether the low rate of benefit in cohort B was due to the immunosuppressive activity of baseline steroids or the extension of central nervous system metastases. However, despite the low rate of benefit in this small cohort, these results showed that the use of steroids does not completely abrogate the possibility of a response to check-point inhibitor therapy. In the Italian Expanded Access Program (EAP), Queirolo et al. [26] enrolled 146 patients with metastatic melanoma and asymptomatic brain metastases and evaluated the feasibility of ipi treatment at 3 mg/kg given every 3 weeks for up to four doses. The global DCR, defined as CR, PR, or SD for a period of at least 3 months by irRC, was 27%, including 4 patients with a CR and 13 patients with a PR (ORR 12%—not reported whether intra or extracranial). The median progression-free survival (PFS) and OS were 2.8 and 4.3 months, respectively. Interestingly, the frequency of adverse effects reported in the overall population was lower than in the previous studies, as grade 3/4 immune-related adverse events were observed in only 6% of patients. Importantly, treatment with ipi was not associated with unexpected or higher rates of serious toxicity in any of the studies reported above.

Overall, these results indicate that both the safety of the treatment and the rates of tumor regression induced by ipi in patients with brain metastases were comparable to those observed in larger trials of patients with extracranial disease only. This was particularly evident when brain lesions were stable, asymptomatic, and did not require corticosteroid treatment. The intracranial ORs observed in these studies are summarized in Table 1.

Table 1.

Objective response rates observed in studies utilizing immune check-point inhibitors to treat patients with brain metastases

Author	Phase	Tumor type	Agent	No. of patients	Objective response rate (%)
Weber (2011) [24]	II (Retrospective)	Melanoma	ipi	12	17
Margolin (2012) cohort A [25]	II	Melanoma	ipi	51	16*
Margolin (2012) cohort B [25]	II	Melanoma	ipi	21	5*
Queirolo (2014) [26]	EAP (Retrospective)	Melanoma	ipi	146	12
Parakh (2017) [44]	Real-world (Retrospective)	Melanoma	nivo or pembro	66	21*
Goldberg (2016) [45]	II	Melanoma	pembro	18	22*
Goldberg (2016) [45]	II	NSCLC	pembro	18	33*
Long (2017) Cohort B [46]	II	Melanoma	nivo	25	20*
Long (2017) Cohort C [46]	II	Melanoma	nivo	16	6*
Haanen (2016) [47]	I	Melanoma	nivo	10	50
Bidoli (2016) [50]	EAP (Retrospective)	NSCLC	nivo	37	19*
Goldman (2016) [51]	I	NSCLC	nivo	12	16*
Haanen (2016) [47]	I	Melanoma	nivo + ipi	10	50
Tawbi (2017) [53]	II	Melanoma	nivo + ipi	75	55*
Long (2017) Cohort A [46]	II	Melanoma	nivo + ipi	26	42*

EAP expanded access program

*Intracranial ORR

Tremelimumab is another fully human antibody designed to block CTLA-4; however, unlike ipi, it is an IgG2 isotype [27]. Several clinical trials utilizing tremelimumab monotherapy demonstrated that the treatment was able to induce ORs in some patients. However, this therapy did not result in a significant increase of OS in patients with metastatic melanoma when compared to standard chemotherapy [28]. In a phase I/II trial, Camacho et al. [29] enrolled three patients with stable or previously treated melanoma metastatic to the brain; however, in all 3 cases, rapid tumor progression was observed. Tremelimumab has been tested in other clinical trials enrolling patients with different types of solid tumors [30, 31], but not on patients with brain metastases.

Anti-PD1/PD-L1 cancer immunotherapy in patients with brain metastases

Two major clinical programs evaluating the safety and efficacy of anti-PD-1 antibodies, nivolumab and pembrolizumab (pembro), have been reported. Nivo, a fully human IgG4 monoclonal antibody, and pembro, a human monoclonal IgG4-kappa antibody, selectively block the interaction of the PD-1 receptor with its ligands PD-L1 and PD-L2, thereby disrupting a signal that suppresses T-cell activation and proliferation [32]. Based on the efficacy data from phase III trials, these PD-1 antibodies were approved by the FDA for the treatment of melanoma (pembro and nivo), NSCLC (pembro and nivo), renal cell carcinoma (nivo), and head–neck cancer (pembro and nivo) [2, 4, 6, 8, 33–37]. Three new human monoclonal antibodies directed against PD-L1 (durvalumab, atezolizumab, and avelumab) are currently undergoing evaluation in several phase III trials. In addition, these agents have recently been approved by the FDA to treat selected tumor types, including urothelial carcinomas and Merkel cell carcinoma [38, 39].

Despite a number of large studies enrolling patients with stable brain metastases [33, 40–42], specific subgroup analyses have not been reported. Here, we have reviewed the limited data available in the literature regarding the safety and efficacy of anti-PD1/PD-L1 agents in patients with brain metastases.

Melanoma

Ahmed et al. [43] retrospectively evaluated the clinical outcomes of 26 patients with melanoma metastatic to the brain who were treated with stereotactic radiation therapy within 6 months of receiving nivo. The primary endpoint of the study was neurotoxicity and, importantly, no unexpected or increased rates of toxicity were reported. The OS rates at 6 and 12 months post-radiation treatment were 78 and 55%, respectively. In a larger retrospective study, Parakh et al. [44] evaluated the efficacy of anti-PD-1 antibodies in 66 patients with melanoma metastatic to the brain, including patients with an ECOG performance status ≥ 2 (n = 21), symptomatic brain metastases (n = 20), and ongoing steroid treatment (n = 15, 0.5–12 mg of dexamethasone or equivalent). Patients were treated with either nivo or pembro, resulting in an overall intracranial ORR of 21% (n = 14) and 8% (n = 5) CRs. Three patients with ORs had symptomatic brain metastases. The ORR of patients receiving steroid treatment was not reported in this study; however, a previous analysis of the same cohort determined that intracranial ORs were achieved in 2 of the 13 patients treated with steroids at baseline [44]. The calculated intracranial DCR was 56% (n = 37). Out of the 66 patients, 48 could be evaluated for extracranial outcomes, resulting in an extracranial ORR of 38% (n = 18), of which 14% (n = 9) demonstrated both intracranial and extracranial ORs. Thus, this study demonstrates that both symptomatic and asymptomatic patients, including those receiving steroid treatment, can obtain intracranial ORs from anti-PD-1 monotherapy.

Recently, results from three prospective studies testing anti-PD-1 antibodies in patients with brain metastases were reported. Goldberg et al. [45] enrolled 36 patients diagnosed with brain metastases with either melanoma or NSCLC in a prospective phase II trial. Patients had 1 or more brain metastases, measuring between 5 and 20 mm, which were either untreated or progressing after radiation treatment. Intracranial ORs were observed in 4 of the 18 (22%) patients with melanoma and 6 of the 18 (33%) patients with NSCLC. A strong correlation between intracranial and systemic ORs was observed, as 8 of the 9 patients with a confirmed systemic OR also achieved an intracranial OR. Of note, none of the patients with continued growth of brain lesions, despite previous irradiation, exhibited a response to anti-PD-1. Long et al. [46] reported the initial results of the Australian ABC study, a phase II trial that enrolled 3 cohorts of patients with melanoma metastatic to the brain. In two of these cohorts, B (asymptomatic brain metastases without prior local therapy) and C (symptomatic, and/or that failed local therapy, and/or with leptomeningeal disease), patients were treated with nivo monotherapy. In cohort B (n = 25), the reported intracranial ORR was 20% (n = 5), including 12% (n = 3) CRs. The extracranial ORR was 30% and the 6-month intracranial PFS was 28%. In cohort C (n = 16), the outcome was very poor, achieving an intracranial ORR of only 6% (n = 1, PR), an extracranial ORR of 25%, and a 6-month intracranial PFS of 13%. Finally, the initial results from a phase I clinical trial (CA209-038) investigating the effects of nivo monotherapy in ten patients with untreated melanoma brain metastases have been reported. Here, the reported ORR was 50% (n = 5, all PR) [47].

Overall, these data suggest that anti-PD-1 monotherapy demonstrates some clinical activity against untreated or progressive melanoma brain metastases. No unexpected or higher rates of toxicity were reported in any of the studies discussed above, indicating an acceptable safety profile. Despite intracranial responses were reported in patients symptomatic and/or undergoing treatment with steroids, it appears that asymptomatic patients with no prior failure of local therapy can derive the most benefit from treatment with anti-PD-1 monotherapy. Table 1 provides a summary of the major results.

Non-melanoma cancers

Only a few available studies have investigated the efficacy and safety of anti-PD-1 monotherapy in patients with non-melanoma tumors and brain metastases. The data from the NSCLC phase II prospective trial by Goldberg et al. [45] are reported above. Kanai et al. [48] administered nivo to ten patients with NSCLC metastatic to the brain; however, treatment was discontinued in 8 of the 10 patients with brain metastases after only 1 dose of nivo. The major reason for discontinuation (7 of the 8 patients) was the exacerbation of neurologic symptoms. The authors of this study concluded that, regardless of whether the neurologic exacerbations represented true disease progression or initial and temporary aggravation of symptoms, it did not appear that this patient population benefited from nivo. Dudnik et al. [49] retrospectively reviewed the safety and efficacy of nivo in 5 patients with metastatic NSCLC and untreated or progressive brain metastases. Two intracranial responses were observed, including 1 CR (parenchymal brain metastases) and 1 PR (leptomeningeal carcinomatosis). Bidoli et al. [50] evaluated the outcome of treatment with nivo in patients with advanced squamous NSCLC metastatic to the brain, in an expanded access program. Thirty-seven patients had asymptomatic pre-treated brain metastases. The observed intracranial ORR was 19% (n = 7), including 1 CR and 6 PRs. Goldman et al. [51] reported the efficacy and safety of nivo in patients with advanced NSCLC and previously treated (n = 46 treated with nivo and n = 42 treated with docetaxel, representing pooled data from 3 pivotal nivo studies) or untreated (n = 12, phase I CheckMate 012) asymptomatic brain metastases. The observed OS in patients with previously treated brain metastases was greater in the nivo group (8.4 versus 6.2 months); however, intracranial ORs were not reported. In CheckMate 012, a phase I trial investigating 12 patients with untreated NSCLC brain metastases, the observed intracranial ORR was 16% (n = 1 with CR and n = 1 with PR).

Importantly, no unexpected or unexpectedly high rates of toxicity were observed, with the notable exception of the study by Kanai, wherein most patients discontinued nivo shortly after beginning treatment [48].

Overall, these results provide evidence that anti-PD-1 monotherapy can induce intracranial ORs in patients with NSCLC and brain metastases, especially in cases of asymptomatic disease. The safety profile of this therapy appears acceptable. However, larger prospective studies are needed to confirm these promising the initial results.

Combination anti-CTLA-4 and anti PD1/PD-L1 immunotherapy

Simultaneous blockade of multiple immunosuppressive pathways may increase the response rates of check-point immunotherapy. In melanoma, a combination of ipi and nivo led to increased response rates in extracranial disease, but with a significantly worse toxicity profile [3, 52]. This regimen is now approved by major regulatory agencies solely for the treatment of metastatic melanoma, as there are no data available for other tumor types.

Haanen et al. [47] presented the initial results of a phase I study, in which patients with melanoma metastatic to the brain were treated with ipi and nivo, followed by nivo maintenance (n = 10). The best ORR was 50% (n = 5, all PR). Tawbi et al. [53] reported the preliminary results of CheckMate 204, a phase II study in which patients with melanoma metastatic to the brain (asymptomatic brain metastases measuring 0.5–3.0 cm) were treated with a combination of ipi and nivo for 4 cycles, followed by nivo maintenance. Prior systemic treatment or stereotactic radiation therapy was permitted. Of the 75 patients treated before database lock, 55% (n = 41) achieved an intracranial OR, including 21% (n = 16) CRs. 52% of patients (n = 39) reported treatment-related grade 3/4 adverse events, including 8% (n = 6) who reported neurologic events. Finally, the results from 26 patients treated in cohort A of the ABC Australian study, wherein patients with melanoma and asymptomatic untreated brain metastases were treated with ipi plus nivo for 4 cycles followed by nivo maintenance, were recently reported. Here, the observed intracranial ORR was 42% (n = 11), including 15% (n = 4) CRs. The extracranial ORR was 48% and the 6-month intracranial PFS was 46%. Importantly, the observed intracranial ORR of patients with treatment-naive disease was 50%, whereas the patient group previously treated with BRAF and MEK inhibitors only achieved an ORR of 16%. The rate of treatment-related grade 3/4 toxicities was 46% [46]. Importantly, in these studies, nivo plus ipi did not appear to induce unexpected or higher rates of toxicity when compared to data previously obtained in patients with melanoma without brain metastases.

Overall, the data from these melanoma cohorts are encouraging and provide a rationale for upfront treatment of melanoma metastatic to the brain using combination immunotherapy. The major results from these studies are summarized in Table 1. Unfortunately, there are currently no available data concerning the effect of this therapy in non-melanoma tumor types with brain metastases.

Anti-CTLA-4 and/or anti-PD1 immunotherapy in combination or in sequence with radiotherapy

As the role of immunotherapy in patients with brain metastases is still evolving, the available data on check-point inhibitors combined with brain radiotherapy are mostly derived from a small number of retrospective studies analyzing limited numbers of patients with melanoma [43, 54–56]. The short-term safety profile of immune check-point inhibitors, either combined or sequenced with stereotactic radiosurgery or whole brain radiotherapy, appeared acceptable. However, because of the retrospective nature and heterogeneity of these studies, it is very difficult to draw any conclusion on whether the addition of radiotherapy improved tumor control over check-point inhibitor immunotherapy alone. In addition, radiation necrosis (RN) following immunotherapy in combination or in sequence with radiotherapy is emerging as a significant issue [57]. In a recent study, the reported overall rate of RN following immunotherapy in combination or sequence with stereotactic radiosurgery was 27%. In addition, a median time of 6 months to RN was reported [56]. Although check-point immunotherapy did not appear to increase the rate of RN [58, 59], the long-term effects of this combination treatment still need to be established. This is especially true for potential long-term survivors, where the incidence of RN might be higher.

Conclusions

Patients with active or untreated brain metastases are conventionally excluded from registration clinical trials. However a large fraction, accounting for over 25% of patients with aggressive cancers such as melanoma, may already present with brain metastases when diagnosed with metastatic disease [12]. In addition, many more patients will develop brain metastases during the course of their disease [13].

While treatment with immune check-point inhibitors significantly improved the clinical outcomes of patients with various forms of metastatic cancers enrolled in phase III registration trials, high-quality evidence supporting the safety and efficacy of these agents in patients with untreated/active brain metastases is lacking. The limited available data, from relatively small studies, indicate that these agents often demonstrate acceptable safety profiles and may induce intracranial responses in melanoma and lung cancer, especially when brain metastases are asymptomatic, stable, and/or previously untreated. Additional prospective trials are already ongoing, but more efforts are needed in order to collect data on other cancer types.

In addition, despite most of the available data being specific for melanoma metastatic to the brain, there are still several questions concerning the treatment of this tumor type which remain unanswered:

• Should we treat patients with check-point inhibitors while on immunosuppressive doses of steroids? Patients from cohort B in the study from Margolin et al. [25] appeared not to obtain any clinical benefit from treatment with ipi, whereas only 2 of the 13 patients in the study by Parakh et al. obtained an intracranial OR after anti-PD-1 [44]. Whether treatment with steroids affects the therapeutic efficacy of anti-PD-1 blockade or the combination therapy with anti-CTLA-4 is still unknown.

• Given the sparse clinical data available, the optimal treatment sequences involving radiotherapy (stereotactic radiosurgery or whole brain radiotherapy) and immunotherapy need to be established. This is particularly important for potential long-term survivors who might experience significant chronic neurotoxicity, as recently reported by Silva et al. [57]

• What is the duration of intracranial responses? It is still unclear whether patients with brain metastases can achieve prolonged responses such as those documented in cases of extracranial disease.

These data highlight the need for further studies targeting these difficult patient populations to offer them additional therapeutic options. Based on current knowledge, combination immunotherapy with nivo plus ipi is a viable treatment option for patients with melanoma and asymptomatic brain metastases.

Abbreviations

CR

Complete response

DCR

Disease control rate

FDA

US Food and Drug Administration

Ipi

Ipilimumab

irRC

Immune-related response criteria

nivo

Nivolumab

NSCLC

Non-small cell lung cancer

OR

Objective response

ORR

Objective response rate

pembro

Pembrolizumab

PFS

Progression-free survival

PR

Partial response

RN

Radiation necrosis

SD

Stable disease

Compliance with ethical standards

Conflict of interest

Marco Donia has received honoraria for lectures from Bristol-Myers Squibb, Merck, Astra Zeneca, and Genzyme; and financial support for attending symposia from Bristol-Myers Squibb, Merck, Novartis, Pfizer, and Roche. Inge Marie Svane has received honoraria for consultancies and lectures from Novartis, Roche, Merck, and Bristol-Myers Squibb; a restricted research grant from Novartis; and financial support for attending symposia from Bristol-Myers Squibb, Merck, Novartis, Pfizer, and Roche. All other authors declare that they have no conflict of interest.