Lung cancer has the characteristics of high incidence, high mortality, and low detection rate (1, 2). Metastasis is considered a leading reason for lung cancer patients’ death, especially brain metastases (BMs) (3). Based on the pathological type, lung cancer consists of small cell lung cancer (SCLC) and non-small cell lung cancer (NSCLC) (4). The mean survival time of untreated NSCLC patients with BM is as short as 1-2 months (5).

Although advances have been achieved in BM patients’ treatment recently, the survival rate is still unsatisfactory, possibly because the blood-brain barrier (BBB) hinders drug entry into the brain, such as chemotherapy (CT), of which the median overall survival was only 4-8 months (6–9). Surgery, whole-brain radiation therapy (WBRT), and stereotactic radiation surgery (SRS) are often referred to as conventional local treatments for BM (10). However, WBRT and SRS have certain limitations, such as radiation neurotoxicity, cognitive deterioration, etc. (11–13). In recent years, the emergence of tyrosine kinase inhibitors (TKIs) changed the treatments for BM, particularly in those patients with positive driver genes like epidermal growth factor receptor (EGFR), anaplastic lymphoma kinase (ALK), and c-ros oncogene 1 (ROS1) (14). Compared with first- and second-generation EGFR-TKIs, third-generation EGFR-TKIs demonstrated truly higher BBB permeability and better efficacy in patients with BM (15). Still, approximately 26% of patients with BM have no driver gene mutations (16).

Therefore, the emergence of immune checkpoint inhibitors (ICIs) that target PD-1, PD-L1, or CTLA-4 offers hope to advanced NSCLC patients with negative driver genes (17). Cohen JV et al. proposed that ICIs and active T cells can penetrate BBB (18), which is necessary for ICIs to work (19). Keynote-024 established immunotherapy as a first-line treatment for advanced NSCLC patients with positive PD-L1 (20). However, for those with unknown PD-L1 expression levels, the response rate was only 17%-19% (21–23). In addition, the results of previous clinical trials also showed that immunotherapy in combination with radiotherapy (RT) or CT might improve the survival of NSCLC patients (24–26).

However, the efficacy of immunotherapy in BM remains controversial, depending on various immunotherapy regimens (27, 28). Besides, most studies included in Alencar’s analysis (29) were retrospective with limited sample size and long-term efficacy. Although Yin et al. and Vivianedid et al. did some analyses about the efficacy of BMs immunotherapy, but the number of studies they included was limited and their analyses lacked efficacy comparison between diversified immunotherapy approaches (30, 31). Therefore, we designed and conducted this meta-analysis to evaluate the efficacy of immunotherapy more comprehensively in NSCLC patients with BM.

NSCLC patients complicated with BMs, especially those whose driver genes mutations are negative or TKI resistant, remain a challenge to treat. However, the roles of ICI regimens for those patients are not completely established. We, therefore, summarized the ICI regimens in NSCLC patients with BMs and performed a meta-analysis to provide a theoretical basis for future treatment strategies. In total, we included 46 articles with 3160 NSCLC patients with BMs. Twenty-two articles were prospective studies, and 24 articles were retrospective studies. Three findings were yielded in our study.

Firstly, immunotherapy may be preferable over non-immunotherapy for NSCLC patients with BMs, with longer PFS (HR = 0.48) and OS (HR = 0.64). The advantage may be caused by the synergy between immunotherapy and chemotherapy/radiotherapy (81–86). For example, immunotherapy enhanced the radiotherapy-induced abscopal effect and reversed the immunosuppressive effect of radiation, by blocking the immune checkpoint between antigen-presenting cells and lymphocytes in regional lymph nodes and other organs. Meanwhile, radiation destroyed the endothelial junctions of the BBB, promoted tumor antigen release, and up-regulated T-cell mediated immune response and PD-L1 expression, which in turn prompted the efficacy of immunotherapy. Simultaneously, chemotherapy promoted tumor immunity by inducing immunogenic cell death as part of its intended therapeutic effect and disrupting strategies that tumors use to evade immune recognition. Moreover, CTLA-4 can impair the critical signal transmission of T-cell activation by competitively inhibiting the CD28 receptor’s binding to B-7 ligands, and PD-1 blocks T-cell activation directly. ICI could inhibit the two pathways so that active peripheral immune cells could penetrate BBB. Our findings were consistent with the conclusions of real-world clinical studies, which showed that immunotherapy had a therapeutic benefit in NSCLC patients with BM.

Secondly, no significant difference in PFS (HR = 0.97, 95%CI: 0.40-2.35), OS (HR = 0.69, 95%CI: 0.23-1.15), EORR (OR = 0.75, 95%CI: 0.28-2.01), iORR (OR = 1.27, 95%CI: 0.65-2.47), iDCR (OR = 1.52, 95%CI: 0.80-2.91), and EDCR (EORR (OR = 0.99, 95%CI: 0.26-3.81) was observed between ICI combined RT and ICI monotherapy, which is consistent with Alencar’s study (30). The difference in PD-L1 expression or lymphocyte tumor infiltration might be the biological mechanism of variable response rates on ICIs for BMs (87, 88). However, there were insufficient data to assess survival outcomes and stratify response based on other important factors like the time of BM diagnosis (newly diagnosed or recurrent), the number, size, and location of metastases, the presence of extracranial disease, the presence of actionable driver gene mutations, and PD-L1 expression.

Interestingly, the sequencing of immunotherapy and RT in the treatment of BM from NSCLC might influence the efficacy of ICI combined RT. Concurrent ICI combined RT might have a lower recurrence rate than sequential ICI combined RT, especially RT after ICI, with a DBF of 0.15. However, no significant improvement in efficacy was found in concurrent ICI combined RT, like iORR, iDCR, EORR, or EDCR, which is inconsistent with previous studies. The interval of ICI and cranial RT might be the core of inconsistency. The largest published retrospective study of patients receiving concurrent immuno-combined RT, the interval of which was shorter than 2 weeks, had a significantly longer OS, reduced incidence of new BM lesions, and an acceptable safety profile (89). The latest prospective study by Wang (unpublished, NCT02978404), a phase II study, in which the interval between nivolumab and SRS was also 2 weeks, achieved a median intracranial PFS of 6.5 months, and an OS of 21.4 months. Coincidentally, in the Emory trial of Khan (unpublished, NCT02858869), the interval of which between pembrolizumab and SRS was 2 to 3 days, also reached a median intracranial PFS of 7.5 months, and an OS of 32.8 months. In addition, a prospective study (unpublished, NCT02696993) that adopted concurrent SRS with dual immunotherapy (Nivolumab and Ipilimumab) with a 7-day interval, achieved a median intracranial PFS of 9.7 months, and a 4-month intracranial PFS rate of 75%. In combination with our analysis and the positive results of ongoing prospective trials, it provides strong support for the efficacy of concurrent ICI combined with RT in 4 weeks (90, 91), which is the drug wash-out period and the destruction of the BBB endothelial cell generated by radiotherapy, for NSCLC patients with BM.

Moreover, when evaluating the efficacy and tolerability of ICI combined RT for BM, it’s important to consider the variety of radiation treatment modalities and dose fractionation prescriptions used. Initial analysis suggested SRS achieved an obvious increase in iORR and EORR, compared to WBRT, which were 75% versus 0 and 73% versus 0, respectively. It was possibly connected to dosage distribution. WBRT delivers the same modest palliative radiation dosage to healthy brain tissue (non-ablative). While SRS delivers a strong ablative dosage solely to metastatic tissue (92). In our included research, the average number of BM was two, the diameter of which ranged from 0.5 to 2 cm, which was defined as small oligometastases. However, WBRT might be suitable for diffuse BM, and it had more acute toxicities than SRS, increases fatigue, lowers the quality of life, and impairs cognitive function (93). Our results correspond to multiple clinical studies, which indicated SRS might eventually replace WBRT for patients with localized (1-3) minor lesions (less than 4 cm in size).

Thirdly, dual ICI combined CT and ICI combined CT provided a better PFS and OS. Compared to dual ICI, the HRs of dual ICI combined CT were 0.61 and 0.67. The HRs of ICI combined CT versus ICI were 0.58 and 0.52, respectively. Surprisingly, there was no difference between dual ICI and ICI combined CT. The mechanisms and efficacy of ICI combined CT varied by different CT agents. Such as platinum-based CT, on the one hand, its favorable immunomodulation effects may boost tumor cells’ susceptibility to PD-1/PD-L1 inhibitors. On the other hand, down-regulation of intracellular PD-L1 expression by PD-1/PD-L1 inhibitors could make patients sensitive to platinum-based CT. However, the mechanisms and efficacy of ICI combined with CT varied by different chemotherapy agents. A dual ICI regimen was defined as a combined blockade of PD-1/L1 and CTLA-4, which offers a number of advantages over a single PD-1 inhibitor without the limitation of BBB. Firstly, PD-1/L1 inhibition is linked to CTLA-4 overexpression, thus anti-CTLA-4 inhibitors might prevent further immune escape directly. Secondly, myeloid-derived suppressor cells can severely limit T cell function inside the tumor microenvironment, but dual ICI can raise the fraction of CD8+ effector T cells relative to MDSCs synergistically. Thirdly, dual ICI could raise inflammatory cytokine production, such as TNF-α and IFN-γ, while lowering T cell anergy. Finally, dual ICI could lead to the growth of memory T cells, which facilitates longer-term anti-tumor immunity (94). But the brain was an immune-specialized environment, in which immune responses against tumors were restricted. Taken together, these confounding factors might weaken the difference of efficacy of ICI combined CT and dual ICI. After combining the results of dual ICI combined CT and ICI combined CT in our study, the ranking of efficacy and recommendation based on the five treatment groups were as follows: dual ICI combined CT, ICI combined CT, dual ICI, ICI monotherapy, and CT.

Another aspect that determines the effectiveness of ICIs is the choice of PD-1, PD-L1, or CTLA-4 inhibitors. Even though Duan et al. (95) found PD-1 inhibitors had better OS and PFS than PD-L1 inhibitors in various cancers, it’s uncertain if the three forms of ICIs, PD-1, PD-L1, and CLTA-4 inhibitors in NSCLC patients with BMs have different intracranial activity. Our analysis found no statistical difference in PD-1, PD-1 combined CLTA-4, and PD-L1 inhibitors for the first time. Still, further exploration was warranted to elucidate the specific mechanism.

Our strengths involved a comprehensive, systematic review of studies by a multidisciplinary team including specialists in NSCLC with BM and epidemiological methods. A broad search strategy was employed to catch all relevant studies. Therefore, our analysis was more comprehensive than other literature with similar topics. Furthermore, there was no proof of publication bias. Importantly, our study not only was the first meta-analysis concerning different ICI regimens, but also the first to compare the intracranial efficacy of WBRT and SRS, and three different ICI types for NSCLC patients with BM.

Our limitations deserve comments. In the analysis of ICI monotherapy versus ICI combined RT, most studies were retrospective, which makes pairwise analysis unavailable, and more prone to selection bias. Despite that, the results that ICI monotherapy might be effective as a single treatment for patients with BM were consistent with, and replenish, Alencar’s therapeutic perspective.

In conclusion, ICI should be considered for selected individuals lacking actionable driver gene mutations. Furthermore, concurrent ICI combined RT in 4 weeks demonstrated improved DBF, and SRS was superior to WBRT for localized and tiny BMs. Our findings revealed that dual ICI combined CT and ICI combined CT had better OS and PFS, giving possible efficacy speculations for clinical decisions. More prospective clinical studies evaluating the benefit of PD-1, PD-L1, and CTLA-4 inhibitors are required in the future, to elucidate why no significant difference in the efficacy of three distinct types of ICI was identified.

The original contributions presented in the study are included in the article/ Supplementary Material . Further inquiries can be directed to the corresponding authors.

All authors were involved in the design of the study, organization of the study, manuscript writing, and approval of final version of the manuscript. XC and LN generated the search strategy, performed the search, independently performed the article selection procedure, data extraction, and summarized the data. ZZL performed the data analyses, and RZ checked the data analysis and validity. XC and LN wrote the manuscript, had contributed equally to the manuscript. RZ and ZZL had contributed equally to the manuscript. All authors contributed to the article and approved the submitted version.

This study was supported by Beijing Xisike Clinical Oncology and Research Foundation (Y-HR2019-0185) and National Multidisciplinary Cooperative Diagnosis and Treatment Capacity Building Project for Major Diseases of China (z027002).

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.