Central nervous systemic efficacy of immune checkpoint inhibitors and concordance between intra/extracranial response in non-small cell lung cancer patients with brain metastasis

Immune checkpoint inhibitors (ICIs) markedly improve the clinical outcomes of advanced non-small-cell lung cancer (NSCLC). However, the intracranial efficacy of ICI is not well elucidated, and previous studies showed discordant outcomes of ICI between intracranial and extracranial diseases. We aimed to evaluate the clinical outcomes and the intracranial and extracranial response of patients with NSCLC and brain metastasis who were treated with ICI in the real-world setting. A total of 55 patients (median age, 63 years [range 42–80]; male, 78%) who had NSCLC with brain metastasis and treated with ICI monotherapy were retrospectively analyzed. We separately assessed the response rates of brain lesions and systemic lesions, and estimated the overall survival (OS) and progression-free survival (PFS). The median OS and overall PFS were 17.0 months (95% CI 10.3–25.6) and 3.19 months (95% CI 2.24–5.03), respectively. The intracranial objective response rate and disease control rate of ICI were 36 and 54%, respectively. Among the 44 patients who showed disease progression, only 32% (n = 14) showed concordant outcomes and 9 patients (20%) showed opposing discordant outcomes. Eight patients continued ICI with local brain therapy after intracranial progression, and their median extracranial PFS and OS were 15 months (95% CI 5.0—not assessed [NA]) and 23.8 months (95% CI 14.7—NA), respectively. ICI monotherapy had a clinically meaningful intracranial efficacy in NSCLC patients with brain metastasis. Watchful waiting and close monitoring without local radiotherapy might be feasible in NSCLC patients with asymptomatic active brain metastasis.


Introduction
Discovery of immune checkpoint inhibitors (ICIs) markedly improved the survival outcomes of patients with metastatic non-small cell lung cancer (NSCLC) and substantially changed the treatment paradigm for NSCLC. As of today, ICI is firmly established as a standard of care for patients with metastatic NSCLC (National Comprehensive Cancer Network 2021). As the life expectancy gradually increases due to improvements in systemic treatment options, nearly half of NSCLC patients eventually develop brain metastasis during the course of the disease (Hanibuchi et al. 2014). Thus, optimal treatment for intracranial disease became an essential issue in managing patients with NSCLC. However, as most of the patients with untreated brain metastasis were excluded in the majority of clinical trials on ICI, (Borghaei Sora Kang and Hyehyun Jeong contributed equally as co-first authors. 1 3 Carbone et al. 2017;Herbst et al. 2016;Langer et al. 2016;Rittmeyer et al. 2017) the clinical efficacy of ICI in metastatic brain lesions remained uncertain.
Several studies were conducted to evaluate the intracranial response of ICI; however, most of the studies were conducted in patients with malignant melanoma and the real-world data of NSCLC patients are limited (Di Giacomo et al. 2019;Goldberg et al. 2020;Nieblas-Bedolla et al. 2021;Teixeira Loiola de Alencar et al. 2021). In addition, a recent study reported the notable evolution and difference of genomic characterization and tumor microenvironment between primary diseases and brain metastatic lesions, (Brastianos et al. 2015;Kim et al. 2019;Shih et al. 2020) suggesting that primary lesions and brain metastatic lesions may respond differently to the same treatment. Yet, extracranial and intracranial responses to ICI and their mutual association were rarely analyzed in previous studies (Goldberg et al. 2020).
In this study, we aimed to provide comprehensive realworld clinical and radiologic response data in patients with NSCLC and brain metastasis who were treated with ICI. In addition, we hypothesized that there would be discordance between systemic response and intracranial response to ICI in patients with NSCLC and brain metastasis. Thus, we assessed both systemic response and intracranial responses in each patient and analyzed the association between them.

Patients and study design
We identified NSCLC patients treated with ICI monotherapy (i.e., pembrolizumab, nivolumab, atezolizumab) at Asan Medical Center (Seoul, South Korea) between January 2014 and April 2021 who had brain metastasis at the time of initiation of index ICI therapy regardless of the status of brain metastasis at the time of initial diagnosis. The exclusion criteria were as follows: patients who did not have adequate follow-up brain images for response evaluation; patients who had concomitant leptomeningeal seeding; patients who received any prior immunotherapy (e.g., anti-PD-1 antibody, anti-PD-L1 antibody, anti-cytotoxic T-lymphocyte associated protein 4 [CTLA-4] antibody) before the administration of index ICI; patients who received only one dose of ICI; patients without active (untreated or progressive after local treatment) brain lesions; and patients who were treated with whole-brain radiotherapy (WBRT) within 3 months prior to ICI initiation. Patients who received surgical resection, local radiotherapy, or WBRT before 3 months prior to ICI initiation were included. Brain magnetic resonance imaging (MRI) obtained between 8 weeks prior and 3 weeks after ICI initiation was considered as baseline brain imaging.
Intracranial radiologic response assessment for brain metastasis was performed by a consensus between two expert neuroradiologists (H.S.K. and J.E.P., with 22 and 7 years of experience in neuro-oncologic imaging, respectively) after a complete review of the MRI images. When assessing disease response, brain lesions that had been treated with local therapy prior to the index ICI initiation were not considered as target lesions and were not included as tumor measurements unless they had clear evidence of disease progression after local treatment at the time of ICI initiation. We primarily used the modified Response Evaluation Criteria in Solid Tumors criteria (mRECIST), version 1.1 which allowed for CNS target lesions to be 5 mm or larger in maximum diameter (Eisenhauer et al. 2009). To compare the efficacy of the response criteria, we additionally evaluated the images using the Response Assessment in Neuro-Oncology Brain Metastases (RANO-BM) criteria  at each timepoint of response evaluation. When using the RANO-BM criteria, lesions with maximum diameter ≥ 10 mm were defined as measurable lesions, and the clinical status and use of steroids were incorporated when determining between response and progression. The detailed information of each response criteria is summarized in Supplementary Table 1. If the patients received local brain therapy to target lesions (e.g., WBRT, gamma knife stereotactic radiosurgery [GKRS], cyberknife stereotactic radiosurgery), they were censored at the date of local treatment. The information of extracranial disease response was extracted from the clinical notes of the attending physician, which was based on the standard RECIST criteria, version 1.1 (Eisenhauer et al. 2009).
The medical records of patients were obtained from abstracted electronic health records, which included sex, age, Eastern Cooperative Oncology Group (ECOG) performance status, smoking status, tumor characteristics, molecular alteration (e.g., EGFR, ALK, BRAF, ROS) assessed by real-time polymerase chain reaction (PCR) and/or nextgeneration sequencing (NGS), PD-L1 expression status of the tumor assessed using 22C3, SP142, and SP263 assay, previous systemic chemotherapy, number of brain metastasis lesions and previous treatment for brain lesions.

Outcomes
The primary endpoint was overall response rate (defined as the percentage of patients who showed complete response [CR] or partial response [PR]), best overall response (BOR) of brain lesions, and intracranial progression-free survival (PFS) according to the mRECIST and RANO-BM criteria. The secondary endpoints were overall survival (OS), overall PFS, and intracranial/extracranial PFS. The BOR of brain lesions was defined as the single best brain response status across all time-point response evaluations until intracranial disease progression. Overall PFS was calculated from the date of ICI initiation to the date of CNS or systemic progression, whichever occurred first, or death from any cause. Intracranial PFS was calculated from the date of ICI initiation to the date of intracranial progression or death from any cause, whichever occurred first. OS was calculated from the date of ICI initiation to the date of death from any cause using the information obtained from the national health care data linked to our hospital.

Statistical analysis
Descriptive statistics were used to analyze categorical and continuous variables. To assess statistical heterogeneity between categorical variables, chi-squared test or Fisher's exact test was used. The Kaplan-Meier method was used to estimate the median values of OS, PFS, and intracranial PFS with 95% confidence interval (CI). The log-rank test was used to compare the survival outcomes of subgroups according to the PD-L1 expression status. To assess the level of agreement between mRECIST and RANO-BM in determining the BOR, a Kappa analysis was used. The agreement between the two response criteria was defined as follows: poor (≤ 0.2), mild (0.2-0.4), moderate (0.4-0.6), substantial (0.6-0.8), and almost perfect (> 0.8). Because of the differences in the response criteria for measurable lesions, 16 patients whose brain lesions were measurable according to the mRECIST but not according to the RANO-BM (maximal diameter between 5 and 10 mm) were excluded from the Kappa analysis. A paired Wilcoxon signed-rank test was used to analyze the differences in the PFS between mRECIST and RANO-BM. For all statistical analyses, twosided P values less than 0.05 were considered as statistically significant. All statistical analysis was conducted using statistical software R (version 4.0.5, Vienna, Austria).

Patient selection and baseline characteristics
Between January 2014 and April 2021, 88 patients with NSCLC and brain metastasis were treated with ICI monotherapy. Among them, we excluded patients who had leptomeningeal seeding (n = 6), who had been treated with immunotherapy before the administration of the index ICI (n = 4), who received only one dose of ICI due to death (n = 1), pneumonia (n = 3), or disease progression (n = 2), who did not have an active brain lesion (n = 15), and whose brain imaging was not adequate for follow-up (n = 2). Finally, a total of 55 patients were included in the analysis.

Treatment pattern of prior chemotherapy and ICI
Before the initiation of ICI, 87% (n = 48) of the patients had been treated with cytotoxic chemotherapy and 22% (n = 12) of patients had received targeted therapy. As for previous local treatment for brain metastasis, local radiotherapy, tumor resection, and WBRT were used in 39 (71%), 9 (16%), and 4 (7.3%) patients, respectively. Fifteen (26.3%) patients did not receive any previous local treatment for brain metastasis.

Survival outcomes
Of the 30 patients whose control of intracranial disease was assessed by mRECIST, the median intracranial PFS was not reached (95% CI NA-NA) and the 1-year intracranial PFS was 86.7% (95% CI 71.1-100%). In the subgroup analysis, no significant differences in intracranial PFS were noted among the subgroups classified by the PD-L1 expression status ( Supplementary Fig. 1).

Patterns of disease progression and concordance between systemic and CNS disease progression
The patterns of failure of patients who showed disease progression are shown in Table 2. Among the patients who showed intracranial and/or extracranial disease progression (n = 44 according to mRECIST), only 32% (n = 14) showed concordant outcomes, taking into account both intracranial and extracranial progression. A total of 9 (20%) patients showed opposing discordant outcomes; among them, 3 (7%) patients showed progression only in the brain lesion and showed a response in the extracranial disease, while 6 Fig. 1 Best response for target brain lesions in patients with non-small cell lung cancer with brain metastasis treated with immune check point inhibitors (by mRECIST criteria). mRE-CIST modified response evaluation criteria for solid tumor Fig. 2 Comparison of mRECIST and RANO-BM criteria. a Population of best overall response of brain lesion. b Intracranial progression-free survival, mRECIST modified response evaluation criteria for solid tumor, RANO-BM response assessment in neuro-oncology brain metastases Overall survival, mRECIST modified response evaluation criteria for solid tumor patients (14%) showed progression in the extracranial disease and showed a response in the brain lesion.
Among the 12 patients who showed only intracranial progression and responsive or stable systemic disease, 2 patient was transferred to a hospice center due to poor performance and disease extent, whereas the remaining 10 patients had received local brain therapy (6 GKRS and 4 WBRT), of whom 8 patients continued ICI beyond progression. The median extracranial PFS and OS of the 8 patients who continued ICI after intracranial progression was 15 months (95% CI 5.0-NA) and 23.8 months (95% CI 14.7-NA), respectively.

Discussion
As the role of ICI monotherapy in active brain metastasis from NSCLC is yet to be fully elucidated, we conducted this analysis to evaluate the clinical and radiologic intracranial responses of active brain metastasis to ICI monotherapy. We found that ICI monotherapy achieved an intracranial ORR (icORR) of 36% and an intracranial DCR (icDCR) of 54%, which are comparable with the results of the previous reports (Goldberg et al. 2016(Goldberg et al. , 2020Hendriks et al. 2019;Teixeira Loiola de Alencar et al. 2021) that reported icORRs of 16.4% (Teixeira Loiola de Alencar et al. 2021) and 27.3% (Hendriks et al. 2019). Notably, the intracranial PFS and OS of our patients were better than the historically documented outcomes of patients with brain metastasis treated with cytotoxic chemotherapy (median OS, 6 months) (Waqar et al. 2018). Considering that nearly 90% of patients included in this study were treated with second or subsequent lines of treatment, the rates of intracranial PFS and OS were clinically meaningful. We believe that our results provide crucial real-world evidence of meaningful intracerebral activity of ICI monotherapy even in patients with active brain metastasis.
Several recent studies reported that there were no significant differences in the OS and PFS between patients with active brain metastasis and those without when treated with ICI, (Wakuda et al. 2021) and that ICI showed benefits in OS and PFS compared with cytotoxic chemotherapy regardless of brain metastasis (Li et al. 2021). Considering the efficacy of ICI in active brain metastasis, close monitoring without local radiotherapy might be feasible in patients with asymptomatic active brain metastasis.
Another important finding of our study was that only 32% of patients showed concordant outcomes between intracranial and extracranial responses. The discordant rate was 20%, which was similar to those reported in previous studies (22% (Goldberg et al. 2020) and 24% (Skribek et al. 2020)). One possible explanation of the discordance is the genetic distinction between the primary disease and brain metastatic lesions. Brastianos et al. conducted whole-exome sequencing (WES) of matched primary tumors and brain metastasis lesions and reported that the clinically actionable alterations shown in brain metastasis lesions were not detected in 53% of primary diseases, whereas separated brain metastasis lesions showed homogenous genetic characteristics (Brastianos et al. 2015). Recently, Shih et al. reported that amplification of MYC, YAP1, and MMP13 and deletion of CDKN2A/B in WES were more commonly identified in brain metastatic lesions than in primary lung adenocarcinoma lesions (Shih et al. 2020). These findings suggest that tumor biology is different between primary lesions and brain metastatic lesions.
The heterogeneity of tumor microenvironment between primary extracranial lesions and brain lesions may be another reason for the disagreements between intracranial and extracranial outcomes. Kudo et al. conducted an NGS of matched primary tumor tissues and metastatic brain tumor tissues and reported that relative increases in tumor-associated macrophages and reductions in CD8 + T cell infiltration were observed in brain tumor tissues compared with primary lesions (Kudo et al. 2019). Other studies also reported that the density of tumor-infiltrating lymphocytes and PD-L1 expression were different between metastatic brain lesions and primary tumor tissues (Berghoff et al. 2014;Kim et al. 2019;Mansfield et al. 2016Mansfield et al. , 2018Zhou et al. 2018). Taken together, these heterogeneities of genetic alteration and tumor microenvironment partially explain our findings, but further molecular exploration with matched primary and brain tumor tissues from patients treated with ICI is needed to provide a firm conclusion on this issue.
When patients show discordant outcomes, deciding whether to maintain or change the treatment regimen may be Brain PD/extracranial SD 9 (20%) 7 (17%) Brain CR + PR/extracranial PD 6 (14%) 5 (12%) Brain SD/extracranial PD 7 (16%) 10 (24%) Not evaluable for brain lesions at time of systemic progression 5 (11%) 5 (12%) challenging. Furthermore, there is little data for the efficacy of local treatment such as radiotherapy or surgery combined with systemic ICI treatment in patients with a discordant response between intracranial and extracranial disease. In our study cohort, among the 12 patients who showed disease progression in the brain and sustained response in the extracranial disease, 8 patients continued ICI with local brain therapy beyond CNS progression. The median extracranial PFS and OS of those 8 patients was 15 and 23.8 months, respectively, and 1 patient showed durable extracranial response at 35 months with stable brain lesion after local radiotherapy. Therefore, when patients show a sustained response in the extracranial disease but progression in the brain lesion, continuing ICI with local treatment for brain progression may be a feasible treatment strategy. We believe that our results provide real-world evidence for implementing local as well as systemic approach in NSCLC patients who experience oligo-CNS progression. The major difference between mRECIST and RANO-BM is the minimum size of measurable lesions (5 mm in mRECIST vs. 10 mm in RANO-BM). In the present study, 16 patients with brain lesions between 5 and 10 mm were not eligible for assessment according to the RANO-BM criteria. Among the 39 patients who were eligible in both criteria, mRECIST and RANO-BM had a high concordance rate of 82% with a Kappa value of 0.75 for assessing BOR, and there were no significant differences in the median PFS assessed by the two criteria. Based on these results, we suggest that both criteria are useful, while mRECIST criteria may allow more patients into clinical trials due to its lower cut-off for a measurable lesion (Qian et al. 2017).
There were several limitations to our study. First, it was a retrospective study based on single-center data and the sample size was limited. However, with this specially targeted review of intracranial response with mRECIST as well as RANO-BM in patients with ICI monotherapy, we could evaluate the intracranial efficacy of ICI monotherapy and analyze the concordance between intracranial and extracranial responses. Another limitation was the lack of data on the reason for using a local treatment for asymptomatic brain lesions during ICI. It is considered that some patients may be treated with local therapy for asymptomatic brain lesions due to the physician's concern about disease progression and lack of confidence in the intracranial efficacy of ICI. Finally, we could not evaluate the adverse events of ICI. However, considering that ICI has been established as the standard of care for NSCLC and that its immune-related adverse events are well-known, we believe that immune-related adverse events in our cohort would not have significantly affected the clinical course of our patients.
In conclusion, ICI monotherapy showed clinically meaningful efficacy in NSCLC patients with brain metastasis. Watchful waiting and close monitoring without local radiotherapy might be feasible in patients with asymptomatic active brain metastasis. Considering the low concordant response rate between primary diseases and brain lesions during ICI monotherapy, close monitoring of both lesions is needed and tailored local therapy should be considered for each metastatic site.
Author contributions All authors contributed to the clinical data resources and curation. JEP and SY contributed to study conceptualization, project administration, and supervision. SK and HJ contributed to data analysis and interpretation, and written first draft of the manuscript. All authors were reviewed and edited on previous versions of the manuscript. All authors read and approved the final manuscript.

Data availability
The data sets generated during and/or analysed during the current study are available from the corresponding author on reasonable request.
Ethical approval This study was reviewed and approved by the Institutional Review Board of Asan Medical center (approval number: 2021-1300, date of approval: August 25, 2021). The requirement of informed consent was waived due to the retrospective nature of the study.