Introduction
Brain metastasis (BrM), one of the most common distant metastases in non-small-cell lung cancer (NSCLC), still remains an awkward disease with unsatisfactory overall prognosis [
1‐
5]. BrMs occur in approximately 40% of patients with NSCLC during his/her lifetime [
6,
7]. Radiotherapy remains to be the mainstay of local treatment in patients with NSCLC and BrMs. Recently, several systemic treatments, especially molecularly targeted therapy, have shown promising activity on BrMs with driver genes alterations due to the satisfactory central nervous system penetration [
4,
5,
8]. However, most patients with BrMs but without driver gene alterations have very limited therapeutic options together with dismal long-term survival.
Immune checkpoint inhibitors (ICIs) including programmed cell death 1 (PD-1) and its ligand (PD-L1) inhibitors have significantly improved the prognosis and shifted the treatment paradigm in advanced NSCLC. However, patients with BrMs were often excluded from clinical trials, or only high-selected patients could be included [
3,
4,
9]. Although several recent publications reported that anti-PD-1/PD-L1 monotherapy showed good activity for controlling BrMs, the efficacy was barely satisfactory [
10]. Even some studies indicated that the presence of BrMs was correlated with inferior survival in patients with NSCLC received anti-PD-1/PD-L1 monotherapy [
11,
12], suggesting an unmet treatment need for these populations. More recently, several clinical trials demonstrated that PD-1/PD-L1 inhibitors plus chemotherapy and/or antiangiogenic therapy could significantly prolong progression-free and overall survival (PFS and OS) in patients with advanced or metastatic NSCLC [
13‐
15]. Subgroup analyses in those with BrMs suggested that anti-PD-1/PD-L1 based combination therapy could result in prolonged OS than chemotherapy [
16]. However, these trials only included high-selected patients with BrMs (for example, untreated or asymptomatic BrMs) and their control group is often traditional chemotherapy, which was not well consistent with currently clinical practice. Thus, whether anti-PD-1/PD-L1 based combination treatments could show better efficacy than anti-PD-1/PD-L1 monotherapy in NSCLC patients with less-selected BrMs remains unknown.
Here, we performed this multicenter retrospective study to investigate the impact of BrMs on the efficacy of ICI based treatments in NSCLC, and the outcome differences between PD-1/PD-L1 inhibitor-based monotherapy and combination therapies in patients with NSCLC and BrMs. To dissect the specific tumor immune microenvironment (TIME) of BrMs and investigate the potential explanations for different treatment outcomes of anti-PD-1/PD-L1 based therapies, we conducted a transcriptomic analysis on paired samples from primary lung cancers and BrMs.
Discussion
To date, PD-1/PD-L1 inhibitors monotherapy or plus chemotherapy have become the new standard of care for patients with NSCLC without driver gene alterations in first-line setting. Nonetheless, the optimal treatment strategy for patients with NSCLC and BrMs remains undetermined mainly due to lack of clear immunophenotyping of BrMs. Previous publications on these populations reported heterogeneous results. Here, the current study enrolled 308 patients and reported that BrMs presence was correlated with significantly shorter PFS and OS in anti-PD-1/PD-L1 monotherapy group, while it was only associated with inferior PFS but similar OS in anti-PD-1/PD-L1 based combination treatment group. Of patients with BrMs, anti-PD-1/PD-L1 plus antiangiogenic therapy was associated with the longest PFS and OS when compared with anti-PD-1/PD-L1 monotherapy or plus chemotherapy. Multivariate analyses showed that anti-PD-1/PD-L1 based combination treatment was independently correlated with significantly longer PFS and OS in patients with BrMs. Furthermore, transcriptomic analysis of paired primary lung tumors and BrMs showed a suppressive TIME in BrMs with decreased CD4+ T cells and M1 macrophages but increased M2 macrophages infiltration. Collectively, these findings indicate that BrMs of NSCLC possessed an immunosuppressive TIME; anti-PD-1/PD-L1 monotherapy showed limited antitumor efficacy in patients with NSCLC and BrMs; the anti-PD-1/PD-L1 based combination treatment, especially anti-PD-1/PD-L1 plus anti-angiogenic treatment, could be an alternative and effective treatment option for patients with NSCLC and BrMs.
In this study, 308 patients with anti-PD-1/PD-L1 treated metastatic or advanced NSCLC were included and 83 (26.9%) of them had BrMs. Although the percentage is higher than that reported in previous studies [
16,
22,
23], it was similar to that reported in a series of retrospective studies [
12,
24,
25] and is expected in this population. Of note, the rate of patients with corticosteroid usage history (33.1%) was much higher than that in a recent study [
12] mainly due to the high percentage of patients with symptomatic BrMs (39.8%) in our study.
BrM is one of the negative prognostic factors in patients with NSCLC [
26]. The efficacy of anti-PD-1/PD-L1 based treatment in patients with BrMs remained undetermined. In the current study, despite much more cases with histology of adenocarcinoma and received ICIs-based combination therapy, the overall ORR was significantly lower in patients with BrMs than those without BrMs (18.1% vs. 29.3%;
P = 0.0464). Moreover, we observed that the presence of BrMs was correlated with significantly inferior PFS and OS in anti-PD-1/PD-L1 monotherapy group. These results were consistent with the findings of the French expanded access program series [
11]. However, a recent retrospective study included 1025 patients showed that the ORR was comparable between patients with and without BrMs (20.6% vs. 22.7%) and the presence of BrMs was not correlated with inferior survival with ICI monotherapy in multivariate analysis [
12]. The potential reason for this discordance may include the different baseline features and ethnicity of research populations, high percentage of patients with symptomatic BrMs, and high rate of patients with liver metastasis in this study. Collectively, BrMs presence showed limited impact on the efficacy of anti-PD-1/PD-L1based combination therapy but the impact of BrMs presence on efficacy of anti-PD-1/PD-L1monotherapy in NSCLC still remains undetermined. Future prospective studies with larger populations and strict design are warranted.
Several clinical trials suggested that PD-1/PD-L1 inhibitor plus chemotherapy and/or antiangiogenic therapy could dramatically prolong both PFS and OS in patients with NSCLC [
13‐
15]. Furthermore, the updated analysis from KEYNOTE-189 showed that pembrolizumab plus chemotherapy could also benefit patients with BrMs, which reported that HRs for OS and PFS were comparable regardless of BrMs [
16]. Consistently, our study found that patients received anti-PD-1/PD-L1 based combination therapy had substantially longer PFS and OS than those received anti-PD-1/PD-L1 monotherapy. Subgroup analysis showed that PD-1/PD-L1 inhibitor plus antiangiogenic therapy was correlated with the longest PFS and OS when compared with PD-1/PD-L1 inhibitor plus chemotherapy and anti-PD-1/PD-L1 monotherapy. Recently, Kudo et al. reported that TIME of BrMs from NSCLC is immunosuppressed with less T cell infiltration and increased immunosuppressive tumor-associated macrophages [
27]. Li et al. also reported that BrMs of lung cancers had reduced tumor infiltrating lymphocytes, decreased scores of immune-related signatures, and a lower proportion of tumors with high PD-L1/high CD8A [
28]. Our current results were consistent with these two studies. In addition, the current findings showed that BrMs had an increased infiltration of M2 macrophages. Our previous study reported that reasonable dose of antiangiogenic agents could induce the polarization of M2 tumor-associated macrophages to M1 tumor-associated macrophages in TIME of lung cancer, subsequently potentiating the antitumor effect of PD-1/PD-L1 inhibitors [
29]. Thus, we could hypothesize that the addition of antiangiogenic drug increased the antitumor effect of PD-1/PD-L1 inhibitor via shaping the phenotype of tumor-associated macrophage in BrMs. Collectively, these findings indicate that PD-1/PD-L1 inhibitor plus antiangiogenic therapy may be one of the promising therapeutic options for patients with NSCLC and BrMs. Notably, the innovative clinical trial to investigate the safety and efficacy of bevacizumab plus pembrolizumab in patients with melanoma and NSCLC BrMs is also ongoing (NCT02681549) and the results are anticipated.
Previous publications revealed that common driver gene alterations and low or negative PD-L1 expression level could impair the efficacy of ICIs treatment in advanced NSCLC [
30‐
32]. Hence, we conducted the subgroup analysis based on the driver gene status and PD-L1 expression in patients with BrMs. The results showed that BrM patients with one of common driver gene alterations received ICI based combination therapy had significantly longer OS to those received ICI monotherapy, indicating that patients with driver gene alterations and BrMs could be treated with ICI based combination therapy. Interestingly, BrM patients with PD-L1 expression > 50% had both numerically better PFS and OS compared with those with PD-L1 expression ≤ 50%, suggesting PD-L1 expression may have potential impact on the efficacy of ICIs treatment in advanced NSCLC with BrMs. This is reminiscent of the updated analysis of a phase II trial that reported intracranial ORR of 27.3% in PD-L1-positive patients with NSCLC and untreated BrMs received pembrolizumab, while no BrM response was observed in PD-L1-negative cohort [
10], suggesting the importance of appropriate biomarker selection in treatment decision for this population.
The current study had several limitations that should be acknowledged. First, the small sample size and the retrospective nature will inevitably have several biases such as selection bias. Second, the cranial radiotherapy history was not recorded in details, resulting in the bias of outcomes assessment. Given the radiotherapy could not only alter the immune microenvironment of BrMs but also enhance the efficacy of ICI through synergy effect or abscopal effect, we should emphasize that these results should be interpreted with caution and large-scale prospective study is still warranted. Intriguingly, Lizza et al. reported that cranial radiotherapy before start of ICIs treatment was not correlated with OS in NSCLC with BrMs [
12], suggesting its uncertain effect on ICIs treatment for BrMs disease. Third, we did not record the intracranial tumor response and number and disease status of BrMs (active or stable), making further subgroup analysis difficult. Fourth, although previous studies revealed the different efficacy of PD-1 and PD-L1 inhibitors in solid tumors [
33], we did not investigate its potential impact on the combination therapy due to limited sample size. Last but not least, we did not evaluate several potential biomarkers including tumor mutational burden, pro-inflammatory gene signatures, etc. due to limited tissue samples. Further investigation of these biomarkers is required to investigate their predictive or prognostic value in patients with NSCLC and BrMs.
In summary, the current study suggests that NSCLC with BrMs could obtain barely satisfactory treatment benefit from anti-PD-1/PD-L1 monotherapy, partly due to the specific immunosuppressive TIME of BrMs. Anti-PD-1/PD-L1 based combination therapy could significantly improve the clinical outcomes of patients with NSCLC and BrMs. Particularly, anti-PD-1/PD-L1 plus anti-angiogenic treatment was correlated with the longest PFS and OS, indicating that this combination strategy might be one of the promising therapeutic options for these populations. Given the retrospective nature and small sample size of this study, future large-scale prospective study is warranted to validate the current findings.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.