Safety and efficacy of immune checkpoint blockade in patients with advanced nonsmall cell lung cancer and brain metastasis

The presence of brain metastases (BM) is a negative prognostic factor for patients with advanced nonsmall cell lung cancer (NSCLC). Their incidence seems to be higher in patients with oncogene‐driven tumours, especially those with EGFR‐mutated or ALK‐rearranged tumours. Although targeted treatments demonstrate significant efficacy regarding BM, they only apply to a minority of NSCLC patients. On the other hand, systemic therapies for nononcogenic‐driven NSCLC with BM have shown limited clinical benefit. In recent years, immunotherapy alone or combined with chemotherapy has been adopted as a new standard of care in first‐line therapy. This approach seems to be beneficial to patients with BM in terms of efficacy and toxicity. Combined immune checkpoint inhibition as well as the combination of immunotherapy and radiation therapy show promising results with significant, but overall acceptable toxicity. A pragmatic approach of allowing enrolment of patients with untreated or symptomatic BM in randomised trials evaluating immune checkpoint inhibitors strategies, possibly coupled with central nervous system‐related endpoints may be needed to generate data to refine treatment for this patient population.

system (CNS) metastases throughout the disease course. 1,3-7 After radical treatment for stage III NSCLC, about a third of patients will subsequently develop BM. 8 The increase in the incidence and prevalence of BM is associated with advancements in diagnostic imaging and the rise in overall survival (OS) rates. 9 Brain dissemination in NSCLC patients is correlated with a dismal prognosis, especially for patients that are not candidates for systemic therapy. 10,11 The use of targeted therapies for oncogenic-driven tumours, results in higher intracranial (IC) response rates and longer OS compared to tumours that lack mutations of clinical interest. 4,[12][13][14][15][16][17][18][19][20][21][22][23] Immune checkpoint inhibitors (ICI) have altered the treatment landscape of nononcogene-driven NSCLC. In the second-line setting, single-agent antiprogrammed cell death (ligand) 1 [anti-PD-(L)1] antibody therapy results in longer OS vs standard docetaxel. 5,24,25 Several first-line studies have further established the role of therapy with single ICI or combined with conventional chemotherapy as the new standard of care. [26][27][28][29][30][31][32] Registration studies only allowed patients with treated and stable BM and, for this reason, prospective data evaluating the efficacy and safety of ICI in patients with BM is scant. The adoption of ICI in NSCLC patients with active or untreated BM remains a matter of debate.
With few available life-prolonging therapies, the treatment of nononcogene-addicted brain metastasized NSCLC is complex and challenging. Besides surgery for single BM, other local CNS treatments have yet to demonstrate a profound survival benefit. 33 Patients with BM should be carefully selected for stereotactic radiosurgery (SRS) and whole brain radiation therapy (WBRT) because of toxicity issues, radiotherapy dose constraints and limited efficacy in patients with poor performance status (PS), multiple BM, age older than 65 years and uncontrolled extracranial disease. 34,35 However, an individual patient data meta-analysis from three phase III randomised controlled trials has shown a statistically significant survival benefit for patients younger than 50 years who received SRS. 36 Another recent meta-analysis reported that patients with four or fewer BM ≤4 cm in maximal diameter reaped a greater benefit in terms of local tumour control, when they were solely treated with SRS in comparison with surgery and subsequent radiation therapy (RT). 37 The preferred systemic treatment for nononcogene-driven NSCLC with BM is platinum-doublet-based chemotherapy leading to IC response rates that range from 20% to 45%. [38][39][40] After local CNS treatment, salvage chemotherapy only offers a very small therapeutic benefit. 41 This narrative review aims to summarize the existing safety and efficacy data on ICI for the management of BM in NSCLC.

| IMMUNE CHECKPOINT INHIBITION AND THE CNS-SPECIAL CONSIDERATIONS
Not only does the brain microenvironment play a crucial role in the development of BM, but also it is vital for the proliferation, migration and survival of tumour cells. The blood-brain barrier (BBB) is the most important part of this microenvironment as it is metabolically regulated and promotes the proliferation of tumour cells after extravasation in the perivascular space or brain parenchyma. The BBB is a significant limitation, acting as a natural barrier regarding the delivery of systemic treatment to the CNS and might be relevant as a mechanism of resistance to ICI. 42 Historically, the CNS has been considered an immune-privileged site. The histological absence of observable CNS lymphatics has led to the assumption that there was a lack of functional immunity. The paucity of dendritic cells in the brain parenchyma, the lack of major histocompatibility complex (MHC) class I on glial and neuronal tissue together with the paucity of MHC class II expression in BM cells have been considered as reasons for treatment failure with ICI in the brain. 43 All the above-mentioned observations have been challenged through the years, especially after the discovery of lymphatic vessels in the CNS. 44 The CNS is now considered immune-competent and seems to actively interact with the peripheral immune system, whereas microglial cells can serve as functional antigen-presenting cells, expressing MHC class II molecules. Moreover, the BBB is a relative rather than an absolute barrier, thus, allowing the trafficking of immune system cells and the delivery of antineoplastic therapy to the brain. 43 The response rate seen in BM from NSCLC with cytotoxic drugs constitutes clinical proof of the BBB's relative permeability. [38][39][40] A study by Kudo et al. used comparative immune gene profiling and sequencing analyses to identify 161 differentially expressed genes, which are associated with the inhibition of dendritic cell maturation, T helper 1 cells and leukocyte extravasation signalling pathways in BM in comparison with primary tumours (P < .01). In general, the immune microenvironment of BM seems to be immunosuppressed, partially via increased infiltration of tumour-associated macrophages and less expression of T cells in comparison with matched primary NSCLC tumours (P < .001). 45  patients who were younger than 70 years (P = .019) with histologically confirmed adenocarcinoma (P = .038), with history of cranial irradiation (P = .022), stable BM at baseline (P = .001) and primary brain metastatic disease (P = .011) were related to a higher IC disease control rate (DCR). Symptomatic brain metastatic disease (P = .373) did not affect IC control rates (P = .373). Based on these findings, an immunotherapy CNS score was proposed dividing patients into two different predictive groups based on age, BM at presentation and previous CNS RT. 47 Corticosteroid use in the BM setting is a major clinical issue in relation to the clinical activity of ICI in the CNS. A retrospective study on the impact of baseline steroid use on the efficacy of PD-1 and PD-L1 inhibitors reported worse OS, progression-free survival (PFS) and objective response rate (ORR) for patients (90 out of 640) with baseline daily corticosteroid dose of ≥10 mg of prednisone or equivalent.
One of the most common indications for the use of corticosteroids was symptomatic BM. Baseline corticosteroid use remained significantly associated with poorer PFS after adjustment for performance status, smoking history and presence of active BM. 48 However, some retrospective data indicate that the use of steroids does not affect OS (P = .38) when needed to manage immunotherapy-related adverse events (irAEs). On the contrary, their administration in the palliative setting was documented to be an independent predictive factor for shorter OS [hazard ratio (HR) = 2.7; 95% confidence interval (CI), 1.5-4.9]. 49 More specifically, steroid administration of more than 10 mg of prednisolone has not been found to significantly affect the IC outcome of ICI (P = .221). 47 The maximum acceptable dose of corticosteroids that can be administered during immunotherapy needs to be further investigated.
Another issue with the clinical utility of immunotherapy in patients with BM is the assessment of IC response. There may be a higher proportion of BM patients with pseudo-progression and/or the timing of radiological follow-up may be suboptimal in some patients.
The response assessment in the neuro-oncology (RANO) working team has suggested a novel treatment algorithm to confirm the progressive disease in the brain for patients treated with ICI, namely immunotherapy RANO (iRANO). IRANO proposes to repeat a magnetic resonance imaging (MRI) of the brain after ≥3 months if radiological progression occurs ≤6 months after starting immunotherapy, without significant clinical decline. Repeat brain MRI can help clinicians to distinguish between pseudo-progression and progressive disease in the CNS. 50

| CLINICAL TRIALS EVALUATING ICI IN NSCLC PATIENTS WITH BM
The anti-PD-(L)1 inhibitors atezolizumab, nivolumab and pembrolizumab are approved as a first-choice treatment option for pretreated advanced NSCLC. 5,[25][26][27]51 Currently, there are several treatment options for first-line treatment of advanced NSCLC. Single-agent pembrolizumab, with or without platinum-based chemotherapy, is one of the first-line treatment options for advanced NSCLC patients without EGFR/ALK molecular aberrations. 26,[52][53][54][55] Based on the Impower110 trial, single-agent atezolizumab has also been approved as a treatment option in NSCLC patients with high expression of PD-L1. 56,57 As far as nonsquamous NSCLC is concerned, the quadruplet carboplatin/paclitaxel/bevacizumab/atezolizumab is another firstline approved regimen, as well as the combination of atezolizumab plus chemotherapy with carboplatin and nab-paclitaxel for EGFR/ALK wild-type NSCLC based on the results of Impower150 and 130, respectively. 28,58 Recently, the combination of nivolumab and ipilimumab with or without chemotherapy led to a significant OS benefit independently of the PD-L1 status and also gained approval as a first-line treatment, whereas nivolumab plus chemotherapy and bevacizumab (TASUKI-52 regimen) or atezolizumab plus chemotherapy with cisplatin or carboplatin and pemetrexed (IMpower132 trial) may consist reasonable alternatives in the future [59][60][61][62][63] Finally, cemiplimab with or without chemotherapy has also been approved for treatmentnaïve patients with metastatic NSCLC 64,65 (Table 1).
CNS metastasized NSCLC patients were underrepresented in the majority of the aforementioned first-and second-line registration studies, which only included asymptomatic cases. When ICI was compared to single-agent docetaxel in subgroup analyses done in the second-line setting, no clinical advantage was observed. 5 The OAK and the POPLAR trials were the two landmark trials investigating the PDL-1 inhibitor atezolizumab in previously treated metastatic NSCLC. In the subgroup of patients (85 out of 850) with stable BM included in the OAK trial, atezolizumab showed a longer OS in comparison with docetaxel (HR = 0.54; 95% CI: 0.31-0.94). 5,24,25,51,66 The similarly designed POPLAR trial randomised 287 patients with treated and stable CNS metastases and demonstrated a longer OS benefit in favour of the atezolizumab arm, but a subgroup analysis of the patients with BM has not been conducted. 66,67 As far as the IMpower 132 and Impower150 trials are concerned, patients who had been diagnosed with active or untreated BMs were excluded from these trials. 59,68 The final exploratory analyses of the aforementioned Impower150 study have demonstrated that the combination of atezolizumab with chemotherapy and antiangiogenic agents is associated with beneficial outcomes compared to the controlled arm regarding various subgroups. One of these subgroup analyses has shown OS benefit for the EGFR-mutant patients who were pretreated with EGFR-tyrosine kinase inhibitors (TKI). 68     According to the updated analysis from the atezo-brain trial, no grade 5 trAEs occurred and the combination treatment with atezolizumab plus chemotherapy was well tolerated. 71 It is also worth noting that approximately 96% of patients in the BM subgroup, who received the atezolizumab-based quadruplet in the Impower150 trial, experienced at least one irAE regardless of the received treatment. MPFS and mOS were 2.1 (95% CI: 1.9-2.5) vs 1.7 (95% CI: 1.5-2.1) months and 13 (95% CI: 9-16) months vs 9 (95% CI: 7-13) months for patients without and with BM, respectively. The variables found to be linked to worse OS and PFS in the multivariate analysis were the use of corticosteroids, PS ≥ 2 and more than two metastatic sites. 53 Three hundred seventy-two squamous NSCLC patients were  Abbreviations: BM, brain metastases; CNS, central nervous system; CR, complete responses; CT, chemotherapy; EAP, expanded access programme; EC, extracranial; IC, intracranial; mo, months; NR, not reported; ORR, objective response rate; OS, overall survival; pat, patients; PFS, progression-free survival; RT, radiation therapy; vs, versus. a IC if not otherwise specified. treatment of CNS metastatic disease is the right approach. Practical issues, such as the concomitant use of corticosteroids and improved radiological assessments, should also be addressed in a randomised setting, to optimize the treatment of BM in NSCLC in the future.