The Clinicopathological Features of BRG1-decient Non-small Cell Lung Cancer and Its Response to Immunotherapy: A Single-center Retrospective Study

BRG1-decient NSCLCs have been more intriguing recently for its highly aggressive clinical behavior and no effective therapies. This study characterized the clinical and pathological features of BRG1-decient NSCLCs and investigated their response to immunotherapy. Forty-seven cases with BRG1-decient NSCLC were included. Immunohistochemical markers such as CK7, TTF-1, NapsinA, P40, HepPar-1, Ki-67, BRM, ARID1A and ARID1B were stained. Meanwhile, the PD-L1 expression level, overall survival, progression-free survival and disease control rate of patients received immunotherapy were evaluated. This study revealed that: (1) Patients with BRG1-decient NSCLC have a male predominance(89.4%), smoker enrichment(76.6%) and poor prognosis(median OS: 7.0 months for advanced stage). (2) Histologically, BRG1-decient NSCLCs presented signicant morphological diversity and no lepidic pattern. Inammatory inltration and tumor necrosis was a prominent feature. Immunohistochemical analyses showed a distinctive uniform immunophenotype (TTF-1-/NapsinA-/CK7+) in 60.9% (28/46) of cases and HepPar-1 positive in 46.5% (20/43) of cases. BRM loss or signicant reduction coexisted in 11.8% (4/34) of cases. No case (0/37) showed loss of ARID1A or ARID1B. (3) For twenty-nine patients with advanced stage, eight patients had received immunotherapy and 4 cases achieved a sustainable clinical response with the disease control rate of 50%. ICIs treated patients had better OS than those who received non-ICIs treatment settings (median OS, 27.0m versus 6.0m, p=0.02). Moreover, patients received ICIs have a median PFS of 17.0 months, while, median PFS on platinum doublet chemotherapy were only 6.0 months(p=0.04). NSCLC showed diverse histopathological patterns a unique immunohistochemical phenotype. ICIs–based immunotherapy was a benecial therapy for BRG1- decient NSCLC. eosin (H&E) staining and immunohistochemistry (IHC). IHC staining for PD-L1(22C3) was performed on the Dako Autostainer Link 48 platform and for other antigens was performed on an automated immunohistochemistry system (Bond-III; Leica Biosystems, Buffalo Grove, Illinois, USA) using BOND Polymer Rene DS9800 Detection kit according to the manufacture’s specications. Total loss (TL) of BRG1 was dened as complete absence of nuclear expression in tumor cells compared with strong nuclear expression in the background stromal cells and/or non-neoplastic epithelial cells. Partial loss(PL) of BRG1 was dened as expression present in some tumor cell nuclei but absent in other tumor cell nuclei, while with strong staining in surrounding normal epithelial cells, mesenchymal cells and inammation cells. “Signicant reduction (SR)” of BRG1 was dened as uniformly-attenuated expression in tumor cell nuclear staining compared with surrounding normal epithelial cells, mesenchymal cells and inammatory cells. For PD-L1, the proportion of viable tumor cells exhibiting complete or partial membranous staining (at least 1+ intensity) was evaluated. PD-L1 scores were categorized as<1%(negative) or ≥ 1% tumor cells, and<50%(low) or ≥ 50% tumor cells(high) (Hirsch et al.2017). For several cases, clinically-relevant molecular testing for EGFR mutations, KRAS mutations, gene rearrangement of ALK/EML4 and ROS1 has been performed routinely and data were summarized in this study. and 75.6% BRG1-decient adenocarcinomas respectively. Interestingly, in BRG1-partially lost cases, 4 were positive for TTF-1 and 3 were positive for NapsinA only in the area with BRG1 expression intact. CK7 was positive in 92.9% of adenocarcinomas and 93.6% of all NSCLCs, which was consistent with previous studies. HepPar-1 was reported to be positive in 85% BRG1-decient lung adenocarcinomas by Agaimy et al(2017).However, in our study, HepPar-1 was only positive in 48.7% of adenocarcinomas and 46.5% of all NSCLCs. HepPar-1 staining showed strong granular cytoplasmic(mitochondrial) pattern as seen in normal liver and hepatoid and hepatocellular carcinomas. Therefore, when HepPar-1 is positive in biopsy sample, it is necessary to search for BRG1-decient lung cancer, hepatocellular carcinoma and hepatoid carcinoma. As a further test, we examined the expression of other three subunits of the SWI/SNF complex in BRG1-decient NSCLCs. Four cases (11.8%) had loss or reduction of BRM and no case showed ARID1A or ARID1B loss. Interestingly, in one case with partial loss of BRG1 (No.41), the BRM expression decreased exactly in BRG1 loss area, which further suggested the underlying relationship between BRG1 and BRM. In a study, ARID1A, ARID1B, BRG1 and BRM were lost in 1.3%, 0.3% 2.4% and 2.4% of NSCLCs respectively, suggesting that loss of ARID1A or ARID1B was not common, especially for ARID1B (Naito et al.2019).


Introduction
Mammalian switch/sucrose non-fermentable (mSWI/SNF) chromatin remodeling complex is a group of polyprotein complexes that hydrolyze ATP to break up the DNA-histone interaction, mobilize nucleosome, and subsequently change the structure of chromatin in a non-covalent way. This complex and its subunits participate in tissue genesis and differentiation through multiple cell signaling processes by altering transcription, duplication, repair and reconstruction of chromatin. The mutation rate of mSWI / SNF subunits has been up to 20% in all human cancers, and their tumor-suppressive role has been established by early studies (Versteege et al.1998;Wilson.2011;Masliah-Planchon et al.2015). SMARCA4, the essential ATPase subunit of mSWI/SNF complex, is located on chromosome 19p. It encodes protein BRG1, which interacts with multiple cell cycle regulatory proteins including Rb, p53 etc. Thus, loss of SMARCA4 subunit causes cell cycle defects and the following tumorigenesis (Alessio et al. 2010). Furthermore, the dysfunctional BRG1 upregulates the transcription of several key tumor-associated genes, which also contributes to tumorigenesis (Orvis et al.2014). Both gene mutations and protein de ciency in several subunits of mSWI/SNF (SMARCA4/BRG1, SMARCA2/BRM, ARID1A and ARIDIB) have been identi ed in a variable proportion of non-small cell lung cancer (NSCLC) (Huang et al.2015;Jones, Siân et al.2012). Previous studies have shown that SMARCA4/BRG1-de ciency presented in about 30% of NSCLC cell lines and 10-15% of clinical tissue samples (Girard et al. 2000;Reisman et al.2003;Oike et.al.2013). In recent years, BRG1-de cient non-small cell lung cancer (NSCLC) has been gradually classi ed as a special subtype of NSCLCs. However, most of these studies focused on NSCLC cell lines, while detailed clinicopathological and morphological studies on BRG1-de cient tumors are lacking.
The prognosis for patients with BRG1-de cient NSCLC is poor because therapeutic strategies for this phenotype of NSCLC have not been established. Existing evidences suggested BRG1-de cient NSCLC was incompatible with classic driver genes such as EGFR, ALK, ROS1, which may lead to its insensitivity to traditional targeted drugs (Araujo et al.2015). A previous study reported that low expression of SMARCA4/BRG1 was a novel predictive biomarker for increasing sensitivity to platinum-based chemotherapy in NSCLC (Bell et al.2016). However, this is challenged by another study, in which the median progression-free survival for patients with BRG1-de cient NSCLC was only 38 days after rst-line platinum doublet chemotherapy (Dagogo-Jack et al.2020). In addition, several studies have implicated that inactivation of the mSWI/SNF complex had enhanced tumor sensitivity to immune checkpoint inhibitors(ICIs) (Henon et al.2019;Li et al. 2019;Miao et al.2018). However, little is known about the response of BRG1-de cient NSCLCs to immunotherapy.
Herein, we described the clinical, pathomorphological and immunohistochemical characteristics of 47 NSCLC cases with total loss, partial loss or signi cant reduction of BRG1 protein (collectively referred to as BRG1-de cient) in detail. In addition, we evaluated the PD-L1 expression level in BRG1-de cient NSCLC.
For patients receiving immunotherapy, we collected clinical and imaging information of these patients. The results can provide more references for evaluating the response of BRG1-de cient NSCLCs to immunotherapy.

Data collection and Statistical analysis
Forty-seven cases of BRG1-de cient NSCLC (including biopsy and surgical cases) were diagnosed during daily practice in the Department of Pathology of Shanghai Ruijin Hospital from October 1st 2015 to May 31th 2020. The study was conducted under the approval by the Institutional Review Board. Ethical approval was granted by the clinical Research Ethics Committee of Shanghai Ruijin Hospital, Jiaotong University. Signed informed consents were obtained from all patients for the acquisition and use of tissue samples and anonymized clinical information. All cases were con rmed by repeating the BRG1 immunohistochemistry staining with whole sections. In addition, these tumors were stained for CK7, TTF-1, NapsinA, P40, HepPar-1, Ki-67, PD-L1(22C3), BRM, ARID1A and ARID1B. Other markers such as CK-pan, CD34, Claudin 4, SALL4, SOX2, CgA, SYN, CD56 and PAX-8 were stained to exclude thoracic BRG1de cient undifferentiated tumors and metastatic tumors. All sections were reviewed and evaluated by two pulmonary pathologists. The information about antibodies used in this study are shown in Supplemental Table 1.
The patients' medical records including gender, age, smoking history, clinical stage and pathological morphology were reviewed and summarized. The 8th edition of the TNM classi cation for lung cancer (International Association for the Study of Lung Cancer) was used for tumor staging. We acquired follow-up data from outpatient visits or by telephone. Overall survival (OS) was de ned as the days from pathological diagnosis to death or July 31th, 2021. For patients receiving immunotherapy, progression-free survival (PFS) was de ned as the days from the start of immunotherapy to disease progression detected by imaging or death, and disease control rate(DCR) was de ned as the proportion of patients who had a best response rating of complete response, partial response or stable disease. The cut off date for this analysis was July 31th, 2021. The level of signi cance was de ned as p < 0.05. Median OS and PFS was calculated using the Kaplan-Meier analysis. All statistical analyses were performed using GraphPad Software, Inc.(version 6.0). Categorical variables were presented as percentages.

Immunohistochemical and genetic evaluation
The specimens were xed with 10% formalin and embedded in para n. Serial 4-μm-thick sections were prepared and used for hematoxylin and eosin (H&E) staining and immunohistochemistry (IHC). IHC staining for PD-L1(22C3) was performed on the Dako Autostainer Link 48 platform and for other antigens was performed on an automated immunohistochemistry system (Bond-III; Leica Biosystems, Buffalo Grove, Illinois, USA) using BOND Polymer Re ne DS9800 Detection kit according to the manufacture's speci cations. Total loss (TL) of BRG1 was de ned as complete absence of nuclear expression in tumor cells compared with strong nuclear expression in the background stromal cells and/or non-neoplastic epithelial cells. Partial loss(PL) of BRG1 was de ned as expression present in some tumor cell nuclei but absent in other tumor cell nuclei, while with strong staining in surrounding normal epithelial cells, mesenchymal cells and in ammation cells. "Signi cant reduction (SR)" of BRG1 was de ned as uniformly-attenuated expression in tumor cell nuclear staining compared with surrounding normal epithelial cells, mesenchymal cells and in ammatory cells. For PD-L1, the proportion of viable tumor cells exhibiting complete or partial membranous staining (at least 1+ intensity) was evaluated. PD-L1 scores were categorized as<1%(negative) or ≥1% tumor cells, and<50%(low) or ≥50% tumor cells(high) (Hirsch et al.2017). For several cases, clinically-relevant molecular testing for EGFR mutations, KRAS mutations, gene rearrangement of ALK/EML4 and ROS1 has been performed routinely and data were summarized in this study.

Patient and public involvement statement
No patients or public were involved in setting the research question or the outcome measures, nor were they involved in the design and implementation of the study. There are no plans to involve patients in dissemination.

Clinical features of BRG1-de cient NSCLCs
The main clinical features of the 47 cases with BRG1-de cient (including total loss, partial loss and signi cant reduction of BRG1) NSCLC con rmed by immunohistochemistry are summarized in Table 1. This study included 42 (89.4%) male and 5(10.6%) female patients, whose ages ranged from 37 to 83 years with the mean age of 63.0 years old. Thirty-six (76.4%) patients had smoking history. Eighteen patients underwent surgical resection of the primary tumor: nine(50.0%) of them received preoperative or postoperative adjuvant chemotherapy and ve (27.8%) patients suffered recurrence within the rst year after surgery. The other patients (n=29) were in the advanced stage or can not tolerate surgery. Among twenty-seven patients whose therapeutic information were available, eleven(37.9%) patients received platinum doublet chemotherapy, eight(27.6%) patients were treated with ICIs-based immunotherapy. Four(13.8%) patients received chemotherapy combined other therapies include anti-angiogenesis, radiotherapy and Gamma Knife and the other four(13.8%) patients received palliative treatment(n=2), tyrosine kinase inhibitor (TKI) (n=1), and Gamma Knife(n=1). Up to July 31th, 2021, thirteen patients(27.7%) stay alive with disease free after complete tumor resection, ve patients(10.6%) still live with tumor and twenty-nine(61.7%) patients had died. Meanwhile, for twenty-nine patients who with advanced stage, 82.8%(24/29) patients died before the end of this study, with the median OS of 7.0 months.

The e cacy of immunotherapy of BRG1-de cent NSCLC
Eight patients received anti-PD-1/PD-L1 inhibitors treatment (Table 2). Among them, patient No.19 disease progression after two cycles of cisplatin, pemetrexed and avastin as the rst-line treatment, and then was treated with pembrolizumab combined with docetaxel as the second-line treatment. Patient No.30 received sintilimab as the second-line treatment after the rst-line treatment (cisplatin and pemetrexed) failed. Patient No.35 received pembrolizumab combined with anlotinib as the third-line treatment. The CT scans demonstrated a sustained partial response after 21 cycles of intravenous administration of pembrolizumab (Fig 1A-1C). Until July 31, 2021, four patients mentioned above continue to bene t from immunotherapy and their PFS were not reached (exceeding 37, 20, 14 and 31 months, respectively). Three patients (No.20,No.37 and No.43) showed no response to immunotherapy alone or combined with chemotherapy, and their PFS were 5, 3 and 2 months respectively. Patient No.44 who bene ted from nivolumab combined with chemotherapy as the rst-line treatment for 14 months but died of severe myelosuppression. Overall, for eight patients who received immunotherapy, 4 cases achieved a sustainable clinical bene t with the DCR of 50%. Furthermore, for patients who with advanced BRG1-de cient NSCLC, ICIs ever treated patients(n=8) had better median OS than those who received non-ICIs treatment settings(n=19) including chemotherapy, anti-angiogenesis therapy, TKI, radiotherapy, Gamma Knife and palliative therapy (median OS, 27.0m versus 6.0m, p=0.02). Moreover, patients received ICIs have a median PFS of 17.0 months, while median PFS on platinum doublet chemotherapy(n=11) were only 6.0 months (p=0.04) (Fig 2A-2B).

Pathological, immunohistochemical and molecular ndings in BRG1-de cient NSCLCs
The histological specimens were obtained from lobectomy resection (n = 18) or needle biopsies from primary lesions and metastatic lesions (n = 29). The histopathological pro les of 47 cases were: 42 adenocarcinomas, 1 non-keratinizing squamous cell carcinoma, 1 adenosquamous carcinoma, 2 non-small cell carcinomas, not otherwise speci ed, and 1 pleomorphic carcinoma (Table 3-1 and Table 3-2).Among the 42 adenocarcinomas, 31 were pure poorlydifferentiated solid adenocarcinoma with minimal abortive small lumina and scattered goblet-like cells (Fig.3A-3B), 6 were solid adenocarcinoma with a small amount of acinar-like component (additional foci of micropapillary component seen in one of them) , 2 were papillary adenocarcinoma with a small amount of solid and acinar-like components, 2 were pure acinar adenocarcinoma and 1 was mucinous adenocarcinoma. No lepidic growth pattern was seen in all adenocarcinomas. Most of the tumors were composed of large polygonal tumor cells arranged in compact solid nests and sheets (Fig.3A, Fig.3C) except for differentiated acinar-like, papillary and micropapillary areas. Tumor cells frequently showed clear cytoplasm, vesicular chromatin( Fig.3C-3D), distinct eosinophilic nucleoli and a high mitotic rate (13 mitoses per 2mm 2 ) (Fig.3E). Scattered pleomorphic giant cells and bizzare cells were seen in 5 cases (Fig.3B), and 1 case with extensive giant cells was diagnosed as pleomorphic carcinoma. Of the 47 cases, 32 cases(68.1%) showed more or less rhabdoid tumor cells including 2 cases with frankly rhabdoid tumor cells (Fig.3E) , 37 cases (78.7%) showed cytoplasm hyaline secretory globules of varying size (Fig.3F), 37 cases(78.7%) showed necrosis (Fig.3A). Extensive interstitial and intraepithelial in ltrate of plasma cells, lymphocytes and neutrophils was a frequent nding. (Fig.3B-3C,3E).

Discussion
BRG1-de cient NSCLCs have been more intriguing recently for its highly aggressive clinical behavior and no effective therapies. We summarized the clinical, pathomorphological and immunohistochemical characteristics of BRG1-de cient NSCLCs in detail. Patients with BRG1-de cient NSCLCs tend to be male with tobacco exposure, as a previous study reported (Decroix et al. 2020). About 27.8% of patients suffered recurrence in the rst year after lung surgery. For twenty-nine patients who were in the advanced stage, 82.8%(24/29) patients died before the end of this study, with the median OS of only 7.0 months. SMARCA4 mutation usually coexists with mutation of KRAS, LKB1, KEAP1, P16 and P53 genes, which suggests that BRG1-de cient NSCLC may be insensitive to traditional treatments such as chemotherapy, radiotherapy and targeted inhibitors (Dagogo-Jack et al.2020). The ICIs-based immunotherapy has achieved remarkable success in advanced NSCLCs. Existing evidences revealed that when compared to wild-type subgroups, the inactiviation of SMARCA4/BRG1 may predict poorer clinical response to ICIs (Zhou et al.2020). However, little is known about the ICIs e cacy of the BRG1-de cient NSCLC when compared to the traditional therapeutic settings. In our cohort, for advanced BRG1-de cient NSCLCs, ICIs ever treated patients had better OS than those who received non-ICIs treatment settings including chemotherapy, anti-angiogenesis therapy, TKI, radiotherapy, Gamma Knife and palliative therapy (median OS, 27.0m versus 6.0m, p=0.02). Moreover, patients received ICIs have an improved median PFS of 17.0 months, while median PFS on platinum doublet chemotherapy were only 6.0 months (p=0.04). Notably, three patients bene ted from ICIs therapy as second or third-line treatment even after failing from traditional settings. These data suggest that patients with BRG1-de cient NSCLC may be more sensitive to immunotherapy than traditional settings such as chemotherapy. However, as a retrospective study, the number, age, stage and the treatment lines among the groups have unbalanced differences, without setting the control or matching the patients, which may affect the conclusion. Therefore, randomized controlled trials are necessity to examine e cacy difference between ICIsbased immunotherapy and traditional therapies for BRG1-de cient NSCLCs. Recently, a case report indicated that BRG1-de cient NSCLC had higher level of PD-L1 and TMB, and therefore could generate sustained response to nivolumab, the inhibitor of PD-L1(Naito et al. 2019). However, two cases in our cohort showed sustainable response to ICIs even with low PD-L1 expression level. Furthermore, another study indicated that NSCLC with SMARCA4 mutation harbored high TMB but infrequently(15%) had high PD-L1(Dagogo-Jack et al.2020) (13). Therefore, despite the expression level of PD-L1, other effective predictors such as TMB may also account for the response of BRG1-de cient NSCLCs to immunotherapy.  .2013). Notably, a smaller area of micro-papillary pattern was observed in one solid predominant adenocarcinoma, which has not been reported previously. Signi cant morphological diversity makes it di cult to predict BRG1 de ciency only by histomorphometric analysis. Accordingly, we consider it necessary to detect BRG1 staining in all NSCLCs. Tumor cells of BRG1-de cient NSCLCs showed overt cytological atypia even in papillary or acinus predominant adenocarcinoma. More or less cytoplasmic hyaline droplets (their nature remains unknown) can be seen in most cases. Multifocal and irregular tumor necrosis accompanied by in ammatory in ltration were the prominent features. Frequent necrosis, overt cytological atypia, high mitotic rate and high proliferation index all predicted the aggressive behavior of BRG1-de cient NSCLCs. The in ammatory cells were seen in both interstitium and tumor nests, suggesting in ammation may play an important role in the establishment of tumor microenvironment. Partial loss and signi cant reduction of BRG1 in NSCLCs have only been occasionally described in literatures (Dagogo-Jack et al.2020; Nambirajan et al. 2021;Herpel et al. 2017). Partial loss of BRG1 were frequently seen in our study possibly due to the whole-section immunohistochemistry staining rather than tissue microarray. There was a relatively clear boundary between the BRG1-de cient and BRG1-intact areas in the 6 cases with partial loss. Among them, 5 showed no signi cant morphological difference between the two areas while one showed loss of BRG1 in solid growth pattern but intact BRG1 in acinus pattern. One study presumed that loss of BRG1 occurs at the early phase of carcinogenesis of lung adenocarcinoma (Matsubara et al.2013). However, ndings in our cases with partial loss of BRG1 suggested that this presumption needs to be reconsidered. BRG1-de cient lung adenocarcinomas were typically negative for TTF-1 and positive for CK7 and HepPar-1 (Hirsch et al.2017;Nambirajan et al.2020). In our study, TTF-1 and NapsinA was negative in 64.3% and 75.6% BRG1de cient adenocarcinomas respectively. Interestingly, in BRG1-partially lost cases, 4 were positive for TTF-1 and 3 were positive for NapsinA only in the area with BRG1 expression intact. CK7 was positive in 92.9% of adenocarcinomas and 93.6% of all NSCLCs, which was consistent with previous studies. HepPar-1 was reported to be positive in 85% BRG1-de cient lung adenocarcinomas by Agaimy et al(2017).However, in our study, HepPar-1 was only positive in 48.7% of adenocarcinomas and 46.5% of all NSCLCs. HepPar-1 staining showed strong granular cytoplasmic(mitochondrial) pattern as seen in normal liver and hepatoid and hepatocellular carcinomas. Therefore, when HepPar-1 is positive in biopsy sample, it is necessary to search for BRG1-de cient lung cancer, hepatocellular carcinoma and hepatoid carcinoma. As a further test, we examined the expression of other three subunits of the SWI/SNF complex in BRG1de cient NSCLCs. Four cases (11.8%) had loss or reduction of BRM and no case showed ARID1A or ARID1B loss. Interestingly, in one case with partial loss of BRG1 (No.41), the BRM expression decreased exactly in BRG1 loss area, which further suggested the underlying relationship between BRG1 and BRM. In a study, ARID1A, ARID1B, BRG1 and BRM were lost in 1.3%, 0.3% 2.4% and 2.4% of NSCLCs respectively, suggesting that loss of ARID1A or ARID1B was not common, especially for ARID1B (Naito et al.2019).
In our cohort, there was minimal overlap between BRG1 de ciency and actionable oncogenic driver alterations. KRAS mutations rather than EGFR mutations occurred frequently in BRG1-de cient NSCLC. As we all known that, KRAS gene mutations are the most prevalent molecular alteration detected in human lung adenocarcinoma, accounting for up to 25% of all genetic mutations. The prior efforts to develop speci c targeted inhibitors for the KRAS-mutant patients has been unsuccessful (Ferrer et al.2018). A recent study had revealed that SMARCA4 mutation was a risk factor for KRAS-mutant patients with lung adenocarcinoma (Liu et al.2021). On the other hand, IHC for BRG1 can capture SMARCA4-de cient tumor which is associated with SMARCA4 mutations (Dagogo-Jack et al.2020). Therefore, evaluation of BRG1 expression by IHC may further of KRAS-mutant patients with lung adenocarcinoma.

Conclusion
In summary, our study demonstrated that BRG1-de cient NSCLC is enriched in smokers, predominantly occurs in male and has aggressive clinical behavior. Given the varied histological types and signi cant morphological diversity in BRG1-de cient NSCLC, we recommend staining BRG1 in all NSCLCs in clinical practice. Importantly, ICIs-based immunotherapy may be a promising therapy for BRG1-de cient NSCLCs, underscoring the necessity of developing multicenter clinical trials to con rm the bene ts of immunotherapy in this subtype of NSCLC. Disclosure: The authors declared that they have no con ict of interest. Other therapies include anti-angiogenesis, radiotherapy and Gamma Knife, exclude ICIs.

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download. sup.table.1antibodyinformation.docx