Genomic and Immunologic Characterization of INI1-Deficient Pediatric Cancers

Author Manuscript Published OnlineFirst on March 2, 2020; DOI: 10.1158/1078-0432.CCR-19-3089 Author manuscripts have been peer reviewed and accepted for publication but have not yet been edited.

Genomic and immunologic characterization of INI1-deficient pediatric cancers

Suzanne J. Forrest1*, Alyaa Al-Ibraheemi2, Duong Doan1, Abigail Ward1, Catherine M. Clinton1, Juan Putra2, R. Seth Pinches2, Cigall Kadoch3, Susan N. Chi1, Steven G. DuBois1, Patrick J. Leavey4, Nicole R. LeBoeuf5, Elizabeth A. Mullen1, Natalie Collins1, Alanna J. Church2 Katherine A. Janeway1*

1 Department of Pediatric Hematology/Oncology, Dana-Farber / Boston Children’s Cancer and Blood Disorders Center and Harvard Medical School, Boston, MA 2 Department of Pathology, Boston Children’s Hospital, Boston, MA 3 Department of Pediatric Oncology, Dana-Farber Cancer Institute and Harvard Medical School, Boston, MA 4 Department of Pediatric Hematology-Oncology, University of Texas Southwestern Medical Center, Dallas, TX 5 Department of Dermatology, Center for Cutaneous Oncology, Dana-Farber/Brigham and Women’s Cancer Center and Harvard Medical School, Boston, MA

Running title: Characterization of INI1-deficient pediatric cancers

Keywords: SMARCB1, INI1, rhabdoid, epithelioid, PD-L1

*Corresponding Authors:

Suzanne J. Forrest, Dana-Farber Cancer Institute, 450 Brookline Avenue, Boston, MA, 02215, Phone (617) 632-6443, Fax (617) 632-4410. Email: [email protected]

Katherine A. Janeway, Dana-Farber Cancer Institute, 450 Brookline Avenue, Boston, MA, 02215. Email: [email protected]

Disclosure of Potential Conflicts of Interest: K.A. Janeway has received consulting fees from Bayer. S.G. Dubois reports travel expenses from Loxo Oncology and Roche and consulting fee from Loxo Oncology. N. R. LeBoeuf reports speaking fee from Bayer and consulting fee from Seattle Genetics. C. Kadoch reports that she is the Scientific Founder, fiduciary Board of Directors member, Scientific Advisory Board member, Shareholder, and Consultant for Foghorn Therapeutics, Inc. (Cambridge, MA). The remaining authors declare no potential conflicts of interest.

Downloaded from clincancerres.aacrjournals.org on September 23, 2021. © 2020 American Association for Cancer Research. Author Manuscript Published OnlineFirst on March 2, 2020; DOI: 10.1158/1078-0432.CCR-19-3089 Author manuscripts have been peer reviewed and accepted for publication but have not yet been edited.

Statement of Translational Relevance:

The prognosis for patients with INI1-deficient pediatric cancers is poor and therapeutic options, particularly in the relapsed or refractory setting, are limited. We sought to characterize the association between SMARCB1 genetic variants identified by next-generation sequencing and INI1 protein expression and to evaluate PD-L1 and CD8 expression in INI1-deficient tumors. Available evidence suggests a low rate of PD-L1 expression and response to immune checkpoint blockade in pediatric cancers. INI1-negative pediatric cancers may represent an important exception to this despite remarkably low tumor mutational burden. Our findings provide evidence supporting a novel strategy for treating INI1-deficient cancers. This study contributes to emerging data that immune checkpoint blockade either as monotherapy or in combination with other agents may be an effective treatment approach for these aggressive cancers.

Abstract

Purpose: Several aggressive pediatric cancers harbor alterations in SMARCB1, including rhabdoid tumors, epithelioid sarcoma and chordoma. As tumor profiling has become more routine in clinical care, we investigated the relationship between SMARCB1 genetic variants identified by next-generation sequencing (NGS) and INI1 protein expression. Therapeutic approaches for INI1-deficient tumors are limited. Early reports suggest a potential role for immune checkpoint inhibition in these patients. Thus, we also investigated PD-L1 and CD8 expression in INI1-negative pediatric brain and solid tumors.

Experimental Design: We performed immunohistochemistry (IHC) for INI1 and immune markers (PD-L1, CD8, and CD163) and NGS on tumor samples from 43 pediatric patients who had tumors with INI1 loss on previous IHC or SMARCB1 genomic alterations on prior somatic sequencing.

Results: SMARCB1 two-copy deletions and inactivating mutations on NGS were associated with loss of INI1 protein expression. Single-copy deletion of SMARCB1 was not predictive of INI1 loss in tumor histologies not known to be INI1-deficient. In the 27 cases with INI1 loss and successful tumor sequencing, 24 (89%) had a SMARCB1 alteration detected. Additionally, 47% (14/30) of the patients with INI1-negative tumors had a tumor specimen that was PD-L1 positive and 60% (18/30) had positive or rare CD8 staining. We report on 3 patients with INI1-negative tumors with evidence of disease control on immune checkpoint inhibitors.

Conclusions: A significant proportion of the INI1-negative tumors express PD-L1, and PD-L1 positivity was associated with extracranial tumor site. These results suggest that clinical trials of immune checkpoint inhibitors are warranted in INI1-negative pediatric cancers.

Introduction

Multiple aggressive pediatric cancers are characterized by alterations in SMARCB1, which is a core subunit of the SWItch/Sucrose Non-Fermentable (SWI/SNF) chromatin remodeling complex (1). Inactivation of the SMARCB1 gene and subsequent loss of INI1 protein expression is well described in several tumor types including rhabdoid tumors, epithelioid sarcoma, and chordoma (2-8). These rare tumor diagnoses are associated with a dismal prognosis, particularly in the relapsed or refractory setting, and few novel treatment options are available (9-14). Aside from SMARCB1 alterations, next-generation sequencing (NGS) of INI1- deficient tumors has not shown other recurrent genetic abnormalities (15,16). Malignant rhabdoid tumors can arise in the brain, kidney, liver, and other soft tissues. These tumors, most often diagnosed in infants and young children, are highly resistant to treatment and are often fatal, with historical 5-year overall survival between 17% and 33% despite intensive therapy (10,12). Epithelioid sarcoma is a rare tumor that typically occurs in the extremities of adolescents and young adults. In the setting of relapse or recurrence there is no standard of care and prognosis is poor (6,17,18). Poorly differentiated chordomas with loss of INI1 expression are rare pediatric tumors which have been reported more recently and are also associated with a dismal prognosis (4,19). EZH2 is essential for all cancers with alterations of subunits of the SWI/SNF complex (20). EZH2 targeted inhibitors are currently in clinical trials that are available to children and adults [NCT02601950, NCT02601937]. Genetic variants in SWI/SNF subunits and the results of IHC protein expression are among the clinical trial enrollment criteria. Partial responses to EZH2 inhibitors have been observed in patients with INI1-negative tumors, and thus, in specific diagnoses, loss of INI1 has been identified as a potential biomarker of response to this class of targeted therapy (21). As more pediatric cancers are subjected to tumor sequencing, genomic alterations in the SWI/SNF subunits will be identified. The association between these genetic variants and the IHC findings is not known. Conversely, the ability to detect the genomic alterations responsible for loss of INI1 expression with targeted NGS panels in common use has not been evaluated previously. In the present study, we analyzed the relationship between SMARCB1 genetic variants identified on a NGS panel (OncoPanel) and INI1 protein expression (22). Beyond EZH2

inhibitors, therapeutic approaches for INI1-deficient tumors are limited. Thus, we also investigated the potential role for other targeted therapies by evaluating for other genomic events and for PD-L1 protein expression in a cohort of INI1-deficient pediatric brain and solid malignancies.

Materials and Methods

Patients and samples This study was approved by the Dana-Farber Cancer Institute (DFCI) Institutional Review Board with a waiver of informed consent and was conducted in accordance with the U.S. Common Rule. Patients were identified for inclusion in the study by investigating two institutional databases in parallel. The Boston Children’s Hospital Pathology database was used to identify patients with tumor specimens collected between the years of 2000 and 2017 that had negative nuclear INI1 staining by immunohistochemistry (IHC). A database of somatic sequencing results from 280 pediatric patients with solid malignancies from the PROFILE Cancer Research Study was used to identify patients with SMARCB1 genomic alterations (single nucleotide variants, 1-copy deletions or 2-copy deletions) on panel sequencing (23). Patients were included in the study if sufficient archival formalin-fixed paraffin- embedded (FFPE) tissue was available for both OncoPanel sequencing and IHC. A total of 43 patients were included in the study. One tumor block from each of the 43 patients was cut to obtain material for IHC for INI1, PD-L1, CD8 and CD163 and next-generation sequencing. A second tumor specimen was available for 10 of the 43 patients from a different time point during the patient’s disease course. These additional specimens were cut to obtain material for PD-L1 IHC. Medical record review of the patients in the cohort was conducted to obtain clinical characteristics including diagnosis, sample timing, and lesion treatment. Sample timing was defined as either at initial diagnosis (which included local control surgery during upfront treatment) or at relapse/progression. Lesion treatment was defined as pre-treatment if the procedure to obtain the sample was performed at initial diagnosis prior to initiation of treatment or at relapse/recurrence before treatment of that specific relapse/recurrence. Lesion treatment was defined as post-treatment if the procedure was obtained during treatment with chemotherapy

or radiation (including local control surgery). The three patients treated with immune checkpoint blockade were identified through medical care provided in the pediatric oncology clinics of study authors (KAJ, EM and PL) following institutional guidelines for case reports. The clinical cases were selected from the case records of the Dana-Farber Cancer Institute and University of Texas Southwestern based on observed clinical response to immune checkpoint inhibitor and so these clinical cases are not entirely overlapping with the study cohort. Only one of these patients (P56) was included in the study cohort. The other two patients had IHC and NGS results available from clinical testing.

IHC analysis Immunohistochemical stains for INI-1, PD-L1, CD8, and CD163 were performed on all cases using the Leica Bond according to standardized procedures on automated platforms. Antibody PD-L1, clone 9A11 was run at 1:100, using the Leica Biosystems Refine Detection Kit with citrate antigen retrieval (24), antibody CD163: clone MRQ-26, and antibody CD8: clone 4B11. Stains were interpreted by two pathologists (AA, JP) who were blinded to the sequencing results. INI1 stain was considered “retained” when all tumor cells had nuclear expression, while it was considered to be “totally lost/negative” when all nuclei lacked nuclear positivity and “partially lost” when some cells had retained nuclear expression. The extent of staining for PD- L1 was recorded as percentage in both tumor cells and tumor infiltrating lymphocytes (TILs) and was considered “positive” if there was 1% PD-L1 staining of the tumor cells or TILs. PD-L1 was considered negative if there was no staining. The cut-off point used for PD-L1 of 1% tumor cells or TILs was used for many clinical trials and is an FDA indication for immune checkpoint inhibitors in multiple diagnoses (25,26). Staining for CD8 and CD163 was assessed within the tumor and in the peri-tumor tissue. The extent of staining for CD8 and CD163 was recorded as “absent/negative” if there was no staining, “rare” if 1-20 cells per 20x field stained positive and “positive” if greater than 20 cells per 20x field stained positive.

Next-generation sequencing Molecular profiling of tumor DNA was achieved via massively parallel sequencing (OncoPanel), as previously described (22,27). Briefly, DNA was extracted from FFPE specimens using standard extraction methods (Qiagen, Valencia, California) and quantified using

PicoGreen dsDNA detection (Life Technologies, Carlsbad, California). Targeted next-generation sequencing assessed 447 genes, using the TruSeq LT library preparation kit (Illumina, San Diego, California), a custom RNA bait set (Agilent SureSelect), and the Illumina HiSeq2500. Read analysis was conducted using Picard tools, the Genome Analysis Toolkit, Riker, MuTect and Oncotator. Sequence variants were filtered to exclude those that occur at a population frequency of greater than 0.1% in the gnomAD genome database (Broad Institute). Copy number variants were identified using RobustCNV, an algorithm in development at the Center for Cancer Genome Diagnostics at the Dana-Farber Cancer Institute. RobustCNV relies on localized changes in the mapping depth of sequenced reads including normalization algorithms. Structural variants were detected using the custom Breakmer algorithm, as previously described (28). Mutational burden, microsatellite stability, and mutational signatures were calculated, as previously described (23). Each alteration was reviewed manually by a molecular genetic pathologist. Clinical interpretation by a molecular pathologist includes tiered reporting of identified genetic alterations. As previously described, Tier 1 alterations are those with well- established published evidence of clinical utility in the tumor type sequenced. Tier 2 alterations are those that may have clinical utility. Tier 3 alterations are those with possible but uncertain utility and tier 4 alterations are those with unknown significance (23).

Gene Expression Analysis Correlation of SMARCB1 and CD8A or PD-L1 mRNA levels was performed using Spearman correlation of mRNA levels from publicly available RNA-sequencing data across The Cancer Genome Atlas (TCGA) cancer types calculated by TIMER (Tumor Immune Estimation Resource) and adjusted for tumor purity (29). Correlations were considered significant at a p- value < 0.01.

Statistical analysis Fisher exact test and unpaired t-test was used to compare clinical variables according to categories defined by PD-L1 status. P-values were considered significant at a p-value < 0.05. Data was analyzed using GraphPad Prism Version 8 (San Diego, CA, USA).

Results

Determination of study population From the pathology database, 85 patients with INI1-negative tumors were identified. Of the 85 patients, 29 had tumor tissue available from previous clinical procedures for further studies. From the sequencing database, 23 patients with SMARCB1 alterations were identified, 17 of whom had tumor specimen available. The SMARCB1 alterations included single-copy deletion (n=14), two-copy deletion (n=2), and single nonsense mutation (n=1). Three patients with available tumor tissue were present in both databases. Thus, a total of 43 patients (26 from the pathology database, 14 from the sequencing database, and 3 from both) were included in the study cohort.

Clinical characteristics The clinical characteristics of the 43 patients in the study cohort are summarized in Table 1. Within the pathology cohort, 6 tumor types were represented, of which the most common histologies were extrarenal rhabdoid tumor, atypical teratoid/rhabdoid (ATRT) tumor and malignant rhabdoid tumor of the kidney, comprising 79% of the cohort. One patient with each of those diagnoses was identified in both databases. In the sequencing cohort, a broader range of diagnoses were identified, with 1-2 patients per diagnosis. The majority of samples in the pathology cohort were obtained at initial diagnosis (69%) and prior to treatment (65%). In the sequencing cohort, 53% of samples were obtained at initial diagnosis and 76% of samples were obtained prior to treatment.

Correlation of SMARCB1 alterations and INI1 protein expression Upon prospective IHC staining for INI1, 69% (n=30/43) of patients were found to have INI1-negative tumors (26 from the pathology cohort, 1 from the sequencing cohort, and 3 from both cohorts). Of the patients in the pathology cohort, all patients were confirmed to be INI1- negative. Panel NGS was successful in 91% (n=39/43) of the patients in the study cohort. In the sequencing cohort (n=17), two-copy deletion of SMARCB1 (n=2) as well as single nonsense mutations (n=1) were predictive of INI1 loss on IHC (Figure 1). In contrast, single-copy deletion of SMARCB1 on NGS was not predictive of loss of INI1 expression with only 1/10 (10%) of

SMARCB1 single-copy deletion cases from the sequencing cohort having loss of INI1 on IHC (Figure 1). In the 27 cases with INI1 loss by IHC and successful tumor sequencing, 24 (89%) had a genomic alteration in SMARCB1 on NGS and 3 (11%) had no SMARCB1 alteration detected. Of the 24 cases with a SMARCB1 alteration on sequencing, twelve cases (50%) had two-copy deletion, 5 (21%) had single-copy deletion, 5 (21%) had two inactivating alterations, and 2 (8%) had single nonsense mutations (Figure 2).

Additional genomic alterations in INI1-negative tumors Very few additional genomic events were identified in the INI1-negative tumors, and all the Tier 1-3 alterations excluding low copy number amplifications are displayed in Supplementary Figure 1. The most common alterations in genes other than SMARCB1 were single-copy deletions in MAPK1, CHEK2, NF2, EWSR1, ZNRF3, EP2300, and CRKL. These genes are adjacent to SMARCB1 and upon review there was a combination of arm level and segmental deletions in the area of SMARCB1. Other genes containing variants tiered 1-3 in more than one of the confirmed INI1-negative cases were ARID1B, APC and TP53. On further review, the ARID1B variants are of uncertain significance being likely benign with respect to impact on protein function. On the other hand, the APC variants while also of uncertain significance are more likely pathogenic. Of note, 1 of the APC variants with a population frequency of 3.6% occurs in 2 cases.

PD-L1 status in INI1-negative tumors Of the 30 patients with INI1-negative tumors, 47% (n=14/30) had at least one tumor specimen that was PD-L1 positive with  1% PD-L1 staining of the tumor cells or TILs (Figure 2). In the PD-L1 positive samples, 12 had PD-L1 staining only in tumor cells, 2 had PD-L1 staining only in TILs, and 1 had PD-L1 staining in both tumor cells and TILs (Supplementary Table 1). Only one of the nine ATRT patients had a PD-L1 positive tumor specimen. There was an increased frequency of PD-L1 positive tumor samples in the extracranial tumors as compared to the intracranial tumors (p=0.017; Table 2). The median tumor mutational burden (TMB) was 4.1 mutations per megabase of DNA in the PD-L1 positive tumors, and 3.7 in the PD-L1 negative tumors. There was no association between PD-L1 status and timing of tumor sampling, prior treatment, or TMB (Table 2). Ten patients had specimens available at more than one time

point relative to diagnosis/relapse and pre/post-treatment status for determination of PD-L1 status. There was no consistent pattern of PD-L1 status by sample time point (Supplementary Figure 2).

Immune cell infiltration in INI1-negative tumors INI1-negative tumors were also assessed for immune cell infiltration via CD8 and CD163 IHC within the tumor and in the peri-tumoral tissue (Supplementary Table 1). Of the 30 patients with INI-negative tumors, 16% (5/30) had a tumor specimen with positive CD8 staining (greater than 20 cells per 20x field) and 43% (13/30) with rare CD8 staining (1-20 cells per 20x field) (Figure 2). Notably, all 5 of tumor specimens that were CD8 positive were in tumors that were also PD-L1 positive. CD163 staining was positive (greater than 20 cells per 20x field) in 83% (25/30) of the INI1-negative tumor specimens and was rare (1-20 cells per 20x field) in 10% (3/30). There was no association between CD163 staining and PD-L1 status.

TCGA Analysis of SMARCB1 and Immune Microenvironment To evaluate the relationship between SMARCB1 gene expression and the immune microenvironment across human cancer, we examined the relationship between SMARCB1 and CD8A or CD274 (PD-L1) gene expression. Using TCGA RNA Sequencing and TIMER, we found an inverse correlation between mRNA levels for SMARCB1 and both CD8A and PD-L1 in human cancer types (Figure 3) (29). This is consistent with our finding of increased PD-L1 expression in INI1-deficient tumors.

Patients Treated with Immune Checkpoint Inhibition An 18-year-old male (P56) presented with a forearm soft tissue mass and upon biopsy was diagnosed with epithelioid sarcoma with the tumor having loss of nuclear INI1 by IHC (Figure 4A). Initial staging work up revealed Stage III (T2b, N1, M0) diagnosis with PET avid clinically involved axillary lymph nodes and four of five lymph nodes involved by epithelioid sarcoma on axillary lymph node dissection. He was treated with neoadjuvant pazopanib and concurrent radiation therapy to the primary tumor and regional lymph nodes (48 Gray) followed by two large surgical excisions of the primary tumor both with microscopic positive margins. IHC staining for PD-L1 was negative on the tumor specimen from his initial diagnosis prior to

treatment and was positive (40% of tumor cells and 30% of peritumor lymphocytes) on the tumor specimen from his surgical resection following neo-adjuvant pazopanib and radiation (Figure 4B). Approximately 7 months after surgery, he experienced local disease recurrence and had multiple enlarging sub-centimeter cystic lung nodules concerning for distant metastatic disease. He was treated with an immune checkpoint inhibitor (pembrolizumab 200mg IV every 3 weeks) for 12 months (18 cycles) with stable disease. He then developed palpable lymph nodes and was started on intralesional talimogene laherparepvec (T-VEC) injections into his right arm and axillary nodules every 3 weeks combined with continued systemic pembrolizumab. He remains on this treatment regimen after 8 months (10 cycles) with stable disease and has had a decrease in the measured size of the treated lesions with an 18.5% reduction in the largest 1.8cm tumor after 2 cycles, which was no longer palpable after 4 cycles (Figure 4C-D). A 10-year-old female presented with neck pain and swelling and upon biopsy was diagnosed with anaplastic chordoma of the spine with nuclear loss of INI1 expression by IHC (Figure 4E). Further evaluation revealed metastatic pulmonary disease. Clinical next-generation sequencing (Foundation One) revealed SMARCB1 loss, with stable microsatellite status and low tumor mutational burden (4 Mut/mb). She was initially treated with an aggressive regimen of combination chemotherapy including ifosfamide, doxorubicin, cyclophosphamide, carboplatin and etoposide and local control with proton beam radiation. Approximately 6 months after completing therapy, she experienced metastatic relapse in the lung. She was treated on a clinical trial with an EZH2 inhibitor which was stopped due to disease progression with a new metastatic lesion in her thoracic spine. The thoracic lesion was treated with resection followed by proton beam radiation with good control locally but with progressive disease in her lungs. IHC staining for PD-L1 was positive (5% of tumor cells) on the relapsed tumor specimen (Figure 4F). She was then treated with an immune checkpoint inhibitor (nivolumab 3mg/kg IV every 2 weeks) with a partial response in some lesions (Figure 4G-H) and overall disease control for 9 months. She then received stereotactic body radiation therapy (SBRT) to two growing lung lesions and one cervical spine lesion combined with continued nivolumab with subsequent shrinkage of the radiated as well as multiple non-radiated lesions in the lung. Treatment with nivolumab was discontinued after a total of 14 months due to pneumonitis which responded to systemic steroids. She continued to have stable disease off of nivolumab for 12 months after the occurrence of pneumonitis, and then had progressive disease in the lung. She was re-challenged with

nivolumab which was stopped after 3 doses due to recurrent pneumonitis, which was again responsive to steroids. She has had stable disease off of nivolumab for 6 months after the second occurrence of pneumonitis. A 5-month-old male presented with asymmetric soft tissue masses on his back and was diagnosed with malignant rhabdoid tumor with the tumor having INI1 loss on IHC (Figure 4I). Clinical next-generation sequencing (Foundation One) revealed SMARCB1 loss, with stable microsatellite status and low tumor mutational burden (2 Mut/mb). Staging work up revealed advanced metastatic disease with multiple soft tissue and bony sites of disease involvement. The family declined conventional cytotoxic chemotherapy and he was started on treatment with pembrolizumab (2mg/kg IV every 3 weeks). He was treated with pembrolizumab monotherapy for 15 weeks (5 cycles) with stable soft tissue sites of disease (Figure 4J-L). After 5 cycles he had progressive disease in his cervical spine for which he was treated with radiation and continued pembrolizumab for 2 additional cycles. Treatment was stopped after Cycle 7 of pembrolizumab due to continued progression of the cervical lesion. He died from disease progression 6 weeks after his last dose of pembrolizumab at approximately 11 months of age.

Discussion In summary, we have shown that SMARCB1 2-copy deletions and inactivating mutations are associated with loss of INI1 protein expression, however single-copy SMARCB1 deletion on NGS is not predictive of loss of protein expression. Panel NGS was able to identify alterations in SMARCB1 in the vast majority of tumors with INI1 loss by IHC. Alterations in SMARCB1 were identified in 89% (24/27) of those with successful sequencing. We have further demonstrated that 47% (14/30) of patients with INI1-negative tumors of multiple histologies had positive PD-L1 staining despite low TMB. Interestingly, PD-L1 expression was associated with extracranial tumor site. Additionally, 60% (18/30) had positive or rare CD8 staining suggesting some degree of immune infiltration in the majority of INI1- negative tumors. Our results are in keeping with a recently published report of 16 cases of malignant rhabdoid tumor where 50% (8/16) showed membranous expression of PD-L1 in 10- 70% of tumor cells (30). Previous studies have also shown immune infiltrates in ATRT and extracranial rhabdoid tumor samples as well as response to immune checkpoint blockade in a genetically engineered mouse model of rhabdoid tumor (31). Previous reports suggest a variable

rate of PD-L1 and CTLA-4 expression in pediatric cancers and a low rate of response to immune checkpoint blockade (32). The presence of PD-L1/PD-1 and CTLA-4 expression and response to immune checkpoint inhibitor therapy has, in lung cancer and melanoma, been correlated with high TMB (33,34). INI1-deficient cancers in children have, of all cancers, amongst the lowest mutational burden (30,35,36). Importantly, patients with PD-L1 expressing tumors of multiple histologies have improved responses and survival with immune checkpoint inhibition compared to patients with PD-L1 negative tumors (37). The three cases reported in this study add to an expanding literature of patients of INI1- negative tumors benefiting from immune checkpoint inhibitor therapy. A pediatric patient with INI1-negative malignant rhabdoid tumor enrolled on a clinical trial of atezolizumab was reported to have a response (38,39). There is a published report of a 24-year-old young adult with metastatic INI1-negative epithelioid sarcoma progressing on pazopanib who had a partial response to nivolumab given with pazopanib (40). Another young adult patient with recurrent INI1-negative renal medullary carcinoma had a complete response to nivolumab lasting greater than nine months despite low TMB (41). Two subsequent adult patients with INI1-negative renal medullary carcinoma treated with nivolumab did not have durable responses (42). A single-arm phase II trial of nivolumab plus ipilimumab in adults with INI1-negative renal medullary carcinoma in adults is now open [NCT03274258] (43). PD-L1 expression and responses to immune checkpoint blockade have also been seen in cancers harboring loss of SMARCA4, another subunit of the SWI/SNF complex, despite low TMB (44,45). Further studies are needed to elucidate the mechanism potentially linking loss of INI1 expression to PD-L1 expression and response to immune checkpoint inhibition. Previously, it was shown that inactivation of other members of the SWI/SNF complex (Pbrm1, ARID2, and BRD7) sensitized tumor cells to T cell killing most likely from role of the SWI/SNF complex in regulating chromatin accessibility of interferon-gamma inducible genes (46). Loss of SMARCB1 may lead to PD-L1 expression and sensitivity to immunotherapy from a similar mechanism. Along with recently published reports, our study supports the development of clinical trials of immune checkpoint inhibitors in pediatric patients with INI1-negative malignancies which portend a poor prognosis and have limited treatments available. EZH2 inhibition has emerged as an attractive treatment to combine with immune checkpoint blockade based on the immunologic effects seen in both regulatory T cells and tumors, including increased PD-L1 expression, in the

setting of EZH2 inhibition (47-49). Furthermore, there was a recently published report of a patient with an aggressive INI1-negative chordoma who had a remarkable and durable response to sequential treatment with an EZH2 inhibitor, radiation, and immune checkpoint inhibition (47). Given the dependence of INI1-deficient tumors on EZH2, the combination of immune checkpoint inhibition and EZH2 inhibition may be a promising combination strategy for these difficult to treat cancers (50,51).

Acknowledgements We thank Dana-Farber/Harvard Cancer Center in Boston, MA, for the use of the Specialized Histopathology Core, which provided histology and immunohistochemistry service. Dana- Farber/Harvard Cancer Center is supported in part by an NCI Cancer Center Support Grant (NIH 5P30CA006516-55). We would like to acknowledge the PROFILE Study at Dana- Farber/Brigham and Women’s Cancer Center and Dana-Farber/Boston Children’s Cancer and Blood Disorders Center for generating the sequencing data. The authors would like to acknowledge the DFCI Oncology Data Retrieval System (OncDRS) for the aggregation, management, and delivery of the clinical and operational research data used in this project. We also thank Rebecca Collins, MD at University of Texas Southwestern for the photomicrographs for third case of a patient treated with immune checkpoint inhibition. This work was supported in part by the Dana-Farber/Harvard Cancer Center Support Grant (NIH 5P30CA006516-55), and by philanthropy funding (Gilmore Fund, Shobokshi Solid Tumor Fund, and the Lamb Family Fund).

References:

1. Wang X, Sansam CG, Thom CS, Metzger D, Evans JA, Nguyen PT, et al. Oncogenesis caused by loss of the SNF5 tumor suppressor is dependent on activity of BRG1, the ATPase of the SWI/SNF chromatin remodeling complex. Cancer Res 2009;69(20):8094- 101 doi 10.1158/0008-5472.CAN-09-0733. 2. Margol AS, Judkins AR. Pathology and diagnosis of SMARCB1-deficient tumors. Cancer Genet 2014;207(9):358-64 doi 10.1016/j.cancergen.2014.07.004. 3. Hollmann TJ, Hornick JL. INI1-deficient tumors: diagnostic features and molecular genetics. Am J Surg Pathol 2011;35(10):e47-63 doi 10.1097/PAS.0b013e31822b325b. 4. Mobley BC, McKenney JK, Bangs CD, Callahan K, Yeom KW, Schneppenheim R, et al. Loss of SMARCB1/INI1 expression in poorly differentiated chordomas. Acta Neuropathol 2010;120(6):745-53 doi 10.1007/s00401-010-0767-x. 5. Trobaugh-Lotrario AD, Tomlinson GE, Finegold MJ, Gore L, Feusner JH. Small cell undifferentiated variant of hepatoblastoma: adverse clinical and molecular features similar to rhabdoid tumors. Pediatr Blood Cancer 2009;52(3):328-34 doi 10.1002/pbc.21834. 6. Chbani L, Guillou L, Terrier P, Decouvelaere AV, Gregoire F, Terrier-Lacombe MJ, et al. Epithelioid sarcoma: a clinicopathologic and immunohistochemical analysis of 106 cases from the French sarcoma group. Am J Clin Pathol 2009;131(2):222-7 doi 10.1309/AJCPU98ABIPVJAIV. 7. Calderaro J, Moroch J, Pierron G, Pedeutour F, Grison C, Maille P, et al. SMARCB1/INI1 inactivation in renal medullary carcinoma. Histopathology 2012;61(3):428-35 doi 10.1111/j.1365-2559.2012.04228.x. 8. Modena P, Lualdi E, Facchinetti F, Galli L, Teixeira MR, Pilotti S, et al. SMARCB1/INI1 tumor suppressor gene is frequently inactivated in epithelioid sarcomas. Cancer Res 2005;65(10):4012-9 doi 10.1158/0008-5472.CAN-04-3050. 9. Chi SN, Zimmerman MA, Yao X, Cohen KJ, Burger P, Biegel JA, et al. Intensive multimodality treatment for children with newly diagnosed CNS atypical teratoid rhabdoid tumor. J Clin Oncol 2009;27(3):385-9 doi 10.1200/JCO.2008.18.7724. 10. Ginn KF, Gajjar A. Atypical teratoid rhabdoid tumor: current therapy and future directions. Front Oncol 2012;2:114 doi 10.3389/fonc.2012.00114. 11. Morgenstern DA, Gibson S, Brown T, Sebire NJ, Anderson J. Clinical and pathological features of paediatric malignant rhabdoid tumours. Pediatr Blood Cancer 2010;54(1):29- 34 doi 10.1002/pbc.22231. 12. Sultan I, Qaddoumi I, Rodriguez-Galindo C, Nassan AA, Ghandour K, Al-Hussaini M. Age, stage, and radiotherapy, but not primary tumor site, affects the outcome of patients with malignant rhabdoid tumors. Pediatr Blood Cancer 2010;54(1):35-40 doi 10.1002/pbc.22285. 13. Reinhard H, Reinert J, Beier R, Furtwangler R, Alkasser M, Rutkowski S, et al. Rhabdoid tumors in children: prognostic factors in 70 patients diagnosed in Germany. Oncol Rep 2008;19(3):819-23. 14. Bourdeaut F, Chi SN, Fruhwald MC. Rhabdoid tumors: integrating biological insights with clinical success: a report from the SMARCB1 and Rhabdoid Tumor Symposium, Paris, December 12-14, 2013. Cancer Genet 2014;207(9):346-51 doi 10.1016/j.cancergen.2014.10.004.

15. Lee RS, Stewart C, Carter SL, Ambrogio L, Cibulskis K, Sougnez C, et al. A remarkably simple genome underlies highly malignant pediatric rhabdoid cancers. J Clin Invest 2012;122(8):2983-8 doi 10.1172/JCI64400. 16. Kieran MW, Roberts CW, Chi SN, Ligon KL, Rich BE, Macconaill LE, et al. Absence of oncogenic canonical pathway mutations in aggressive pediatric rhabdoid tumors. Pediatr Blood Cancer 2012;59(7):1155-7 doi 10.1002/pbc.24315. 17. Jawad MU, Extein J, Min ES, Scully SP. Prognostic factors for survival in patients with epithelioid sarcoma: 441 cases from the SEER database. Clin Orthop Relat Res 2009;467(11):2939-48 doi 10.1007/s11999-009-0749-2. 18. Baratti D, Pennacchioli E, Casali PG, Bertulli R, Lozza L, Olmi P, et al. Epithelioid sarcoma: prognostic factors and survival in a series of patients treated at a single institution. Ann Surg Oncol 2007;14(12):3542-51 doi 10.1245/s10434-007-9628-9. 19. Hasselblatt M, Thomas C, Hovestadt V, Schrimpf D, Johann P, Bens S, et al. Poorly differentiated chordoma with SMARCB1/INI1 loss: a distinct molecular entity with dismal prognosis. Acta Neuropathol 2016;132(1):149-51 doi 10.1007/s00401-016-1574- 9. 20. Kim KH, Kim W, Howard TP, Vazquez F, Tsherniak A, Wu JN, et al. SWI/SNF-mutant cancers depend on catalytic and non-catalytic activity of EZH2. Nature medicine 2015;21(12):1491-6 doi 10.1038/nm.3968. 21. Italiano A, Keilhack H, Toulmonde M, Coindre JM, Michot JM, Massard C, et al. 302 A phase 1 study of EPZ-6438 (E7438), an Enhancer of Zeste-Homolog 2 (EZH2) inhibitor: Preliminary activity in INI1-negative tumors. European Journal of Cancer 2015;51:S54- S5 doi 10.1016/S0959-8049(16)30168-X. 22. Garcia EP, Minkovsky A, Jia Y, Ducar MD, Shivdasani P, Gong X, et al. Validation of OncoPanel: A Targeted Next-Generation Sequencing Assay for the Detection of Somatic Variants in Cancer. Arch Pathol Lab Med 2017;141(6):751-8 doi 10.5858/arpa.2016- 0527-OA. 23. Sholl LM, Do K, Shivdasani P, Cerami E, Dubuc AM, Kuo FC, et al. Institutional implementation of clinical tumor profiling on an unselected cancer population. JCI Insight 2016;1(19):e87062 doi 10.1172/jci.insight.87062. 24. Gaule P, Smithy JW, Toki M, Rehman J, Patell-Socha F, Cougot D, et al. A Quantitative Comparison of Antibodies to Programmed Cell Death 1 Ligand 1. JAMA Oncol 2017;3(2):256-9 doi 10.1001/jamaoncol.2016.3015. 25. Kazandjian D, Suzman DL, Blumenthal G, Mushti S, He K, Libeg M, et al. FDA Approval Summary: Nivolumab for the Treatment of Metastatic Non-Small Cell Lung Cancer With Progression On or After Platinum-Based Chemotherapy. Oncologist 2016;21(5):634-42 doi 10.1634/theoncologist.2015-0507. 26. Fashoyin-Aje L, Donoghue M, Chen H, He K, Veeraraghavan J, Goldberg KB, et al. FDA Approval Summary: Pembrolizumab for Recurrent Locally Advanced or Metastatic Gastric or Gastroesophageal Junction Adenocarcinoma Expressing PD-L1. Oncologist 2019;24(1):103-9 doi 10.1634/theoncologist.2018-0221. 27. Nowak JA, Yurgelun MB, Bruce JL, Rojas-Rudilla V, Hall DL, Shivdasani P, et al. Detection of Mismatch Repair Deficiency and Microsatellite Instability in Colorectal Adenocarcinoma by Targeted Next-Generation Sequencing. J Mol Diagn 2017;19(1):84- 91 doi 10.1016/j.jmoldx.2016.07.010.

28. Abo RP, Ducar M, Garcia EP, Thorner AR, Rojas-Rudilla V, Lin L, et al. BreaKmer: detection of structural variation in targeted massively parallel sequencing data using kmers. Nucleic Acids Res 2015;43(3):e19 doi 10.1093/nar/gku1211. 29. Li T, Fan J, Wang B, Traugh N, Chen Q, Liu JS, et al. TIMER: A Web Server for Comprehensive Analysis of Tumor-Infiltrating Immune Cells. Cancer Res 2017;77(21):e108-e10 doi 10.1158/0008-5472.CAN-17-0307. 30. Abro B, Kaushal M, Chen L, Wu R, Dehner LP, Pfeifer JD, et al. Tumor mutation burden, DNA mismatch repair status and checkpoint immunotherapy markers in primary and relapsed malignant rhabdoid tumors. Pathol Res Pract 2019;215(6):152395 doi 10.1016/j.prp.2019.03.023. 31. Lu JQ, Wilson BA, Yong VW, Pugh J, Mehta V. Immune cell infiltrates in atypical teratoid/rhabdoid tumors. Can J Neurol Sci 2012;39(5):605-12 doi 10.1017/s031716710001533x. 32. Ring EK, Markert JM, Gillespie GY, Friedman GK. Checkpoint Proteins in Pediatric Brain and Extracranial Solid Tumors: Opportunities for Immunotherapy. Clin Cancer Res 2017;23(2):342-50 doi 10.1158/1078-0432.CCR-16-1829. 33. Rizvi NA, Hellmann MD, Snyder A, Kvistborg P, Makarov V, Havel JJ, et al. Cancer immunology. Mutational landscape determines sensitivity to PD-1 blockade in non-small cell lung cancer. Science 2015;348(6230):124-8 doi 10.1126/science.aaa1348. 34. Van Allen EM, Miao D, Schilling B, Shukla SA, Blank C, Zimmer L, et al. Genomic correlates of response to CTLA-4 blockade in metastatic melanoma. Science 2015;350(6257):207-11 doi 10.1126/science.aad0095. 35. Grobner SN, Worst BC, Weischenfeldt J, Buchhalter I, Kleinheinz K, Rudneva VA, et al. The landscape of genomic alterations across childhood cancers. Nature 2018;555(7696):321-7 doi 10.1038/nature25480. 36. Vogelstein B, Papadopoulos N, Velculescu VE, Zhou S, Diaz LA, Jr., Kinzler KW. Cancer genome landscapes. Science 2013;339(6127):1546-58 doi 10.1126/science.1235122. 37. Patel SP, Kurzrock R. PD-L1 Expression as a Predictive Biomarker in Cancer Immunotherapy. Mol Cancer Ther 2015;14(4):847-56 doi 10.1158/1535-7163.MCT-14- 0983. 38. Geoerger B, Karski EE, Zwaan M, Casanova M, Marshall LV, DuBois SG, et al. A phase I/II study of atezolizumab in pediatric and young adult patients with refractory/relapsed solid tumors (iMATRIX-Atezolizumab). Journal of Clinical Oncology 2017;35(15_suppl):10524- doi 10.1200/JCO.2017.35.15_suppl.10524. 39. Bourdeaut F, Thaku MD, Bergthold G, Karski E. ATRT-11. Marked Response to Atezolizumab in a Patient with Rhabdoid tumor: A Case Study from the IMATRIX- Atezolizumab Trial. Neuro-Oncology 2017;19(suppl_4):iv3-iv doi 10.1093/neuonc/nox083.010. 40. Paoluzzi L, Cacavio A, Ghesani M, Karambelkar A, Rapkiewicz A, Weber J, et al. Response to anti-PD1 therapy with nivolumab in metastatic sarcomas. Clin Sarcoma Res 2016;6:24 doi 10.1186/s13569-016-0064-0. 41. Beckermann KE, Jolly PC, Kim JY, Bordeaux J, Puzanov I, Rathmell WK, et al. Clinical and immunologic correlates of response to PD-1 blockade in a patient with metastatic renal medullary carcinoma. J Immunother Cancer 2017;5:1 doi 10.1186/s40425-016- 0206-1.

42. Sodji Q, Klein K, Sravan K, Parikh J. Predictive role of PD-L1 expression in the response of renal Medullary carcinoma to PD-1 inhibition. J Immunother Cancer 2017;5(1):62 doi 10.1186/s40425-017-0267-9. 43. Msaouel P, Slack-Tidwell R, Tannir NM. Phase II trial of nivolumab (nivo) plus ipilimumab (ipi) in patients with SMARCB1-deficient kidney malignancies. Journal of Clinical Oncology 2019;37(7_suppl):TPS677-TPS doi 10.1200/JCO.2019.37.7_suppl.TPS677. 44. Henon C, Blay J-Y, Massard C, Mir O, Bahleda R, Dumont S, et al. Long lasting major response to pembrolizumab in a thoracic malignant rhabdoid-like SMARCA4-deficient tumor. Annals of Oncology 2019;30(8):1401-3 doi 10.1093/annonc/mdz160. 45. Jelinic P, Ricca J, Van Oudenhove E, Olvera N, Merghoub T, Levine DA, et al. Immune- Active Microenvironment in Small Cell Carcinoma of the Ovary, Hypercalcemic Type: Rationale for Immune Checkpoint Blockade. J Natl Cancer Inst 2018;110(7):787-90 doi 10.1093/jnci/djx277. 46. Pan D, Kobayashi A, Jiang P, Ferrari de Andrade L, Tay RE, Luoma AM, et al. A major chromatin regulator determines resistance of tumor cells to T cell-mediated killing. Science 2018;359(6377):770-5 doi 10.1126/science.aao1710. 47. Gounder MM, Zhu G, Roshal L, Lis E, Daigle SR, Blakemore SJ, et al. Immunologic Correlates of the Abscopal Effect in a SMARCB1/INI1-negative Poorly Differentiated Chordoma after EZH2 Inhibition and Radiotherapy. Clin Cancer Res 2019;25(7):2064-71 doi 10.1158/1078-0432.CCR-18-3133. 48. Wang D, Quiros J, Mahuron K, Pai CC, Ranzani V, Young A, et al. Targeting EZH2 Reprograms Intratumoral Regulatory T Cells to Enhance Cancer Immunity. Cell Rep 2018;23(11):3262-74 doi 10.1016/j.celrep.2018.05.050. 49. Emran AA, Chatterjee A, Rodger EJ, Tiffen JC, Gallagher SJ, Eccles MR, et al. Targeting DNA Methylation and EZH2 Activity to Overcome Melanoma Resistance to Immunotherapy. Trends Immunol 2019;40(4):328-44 doi 10.1016/j.it.2019.02.004. 50. Yamagishi M, Uchimaru K. Targeting EZH2 in cancer therapy. Curr Opin Oncol 2017;29(5):375-81 doi 10.1097/CCO.0000000000000390. 51. Knutson SK, Warholic NM, Wigle TJ, Klaus CR, Allain CJ, Raimondi A, et al. Durable tumor regression in genetically altered malignant rhabdoid tumors by inhibition of methyltransferase EZH2. Proc Natl Acad Sci U S A 2013;110(19):7922-7 doi 10.1073/pnas.1303800110.

Tables

Table 1. Clinical characteristics of the study cohort Clinical characteristics Pathology Sequencing Demographic, n (%) Cohort (n=29) Cohort (n=17) Gender Male 11 (38) 5 (29) Female 18 (62) 12 (71) Age at diagnosis in years Range 1.54 4.53 Median 0.11-18.51 0.51-17.35 Diagnosis Extrarenal rhabdoid tumor 9 (31) 1* (5.8) Atypical teratoid/rhabdoid tumor 9 (31) 1* (5.8) Malignant rhabdoid tumor of the kidney 5 (17) 1* (5.8) Epithelioid sarcoma 4 (14) Anaplastic chordoma 1 (3.5) Renal medullary carcinoma 1 (3.5) Osteosarcoma 2 (11.8) Wilms tumor 2 (11.8) Hepatocellular carcinoma 1 (5.8) Pilomyxoid astrocytoma 1 (5.8) Ependymoma 1 (5.8) Astrocytoma 1 (5.8) Adrenal cortical carcinoma 1 (5.8) Paraganglioma 1 (5.8) Glioblastoma 1 (5.8) Glioma 1 (5.8) Meningioma 1 (5.8) Myofibroblastic tumor 1 (5.8) Sample type Initial diagnosis 20 (69) 9 (53) Relapse/progression 9 (31) 8 (47) Treatment timing Pre-treatment 19 (65.5) 13 (76) Post-treatment 9 (31) 4 (24) Unknown 1 (3.5)

*Three patients identified in both pathology and sequencing cohorts.

Table 2. Characteristics associated with PD-L1 Status in INI1-Negative Tumors PD-L1 Positive PD-L1 Negative p value (n=14) (n=16) Tumor Site, n (%) 0.017 Extracranial tumors 13 (93) 8 (50) Intracranial tumors 1 (7) 8 (50) Sample timing, n (%) 1 Initial diagnosis 10 (71) 12 (75) Recurrence/progression 4 (29) 4 (25) Treatment, n (%) 0.23 Pre-treatment 7 (50) 13 (81) Post-treatment 6 (43) 3 (19) Unknown 1 (7) TMB 0.61 Median 3.04 2.28 Range 2.3-9.1 0.8-6.8

Figure Legends

Figure 1: Confirmatory SMARCB1 Sequencing and INI1 Expression in Sequencing Cohort. A. Two cases with 2-copy gene deletion and one case with a nonsense mutation were identified. Repeat sequencing was successful and identified SMARCB1 alterations for 2 of the 3 cases and INI1 IHC was negative in all 3 cases. B. Fourteen cases with 1-copy gene deletion were identified. Repeat sequencing was successful for 13 cases and SMARCB1 alterations were confirmed in 10 cases. INI1 IHC was positive for 13 cases and negative for 1 case.

Figure 2. Clinical and Genomic Features of INI1-Negative Tumors. OncoPrint of SMARCB1 alterations, tumor mutational burden (TMB), PD-L1 and CD8 staining and clinical features including diagnosis, sample timing and lesion treatment in 30 INI1-negative tumors.

Figure 3. Expression of SMARCB1 is Inversely Correlated with CD8A and PD-L1 mRNA levels in The Cancer Genome Atlas (TCGA) data. Volcano plot showing the Spearman’s correlation and estimated significance of SMARCB1 with CD8A (left) and PD-L1 (right) mRNA levels from RNA sequencing data across TCGA cancer types calculated by TIMER. Each dot indicates a cancer type in TCGA and red dots denote significant correlations (p < 0.05).

Figure 4. Patients Treated with Immune Checkpoint Inhibition. A-D Epithelioid sarcoma case. A. INI1 immunohistochemistry (IHC) from initial diagnosis showing negative nuclear INI1 in tumor cells at 40x (400 x magnification). B. PD-L1 IHC from resection after neoadjuvant pazopanib and radiation therapy showing PD-L1 staining in tumor cells and peritumor lymphocytes. C. Cutaneous lesions prior to talimogene laherparepvec (T-VEC) injections. D. Cutaneous lesions decreased in size after 3 cycles of T-VEC and continued pembrolizumab treatment. E-H Chordoma case. E. INI1 IHC from initial diagnosis showing negative INI1 staining in tumor cells. F. PD-L1 IHC from resection after relapse showing PD-L1 staining in 5% of tumor cells. G. Computerized Tomography (CT) scan showing pulmonary nodule (arrow) measuring 1.2cm before treatment with nivolumab. H. CT scan showing persistent response of the same pulmonary nodule after 28 cycles of treatment with nivolumab. I-L Malignant Rhabdoid tumor. I. IHC from initial diagnosis showing negative INI1 staining in tumor cells. J. Magnetic resonance imaging (MRI) of soft tissue lesion (arrow) measuring 14.3mm by 8.9mm at the time of diagnosis. K. MRI of same lesion (arrow) measuring 12.2mm by 8.0mm after 6 weeks of treatment with pembrolizumab. L. MRI of same lesion (arrow) measuring 9.4mm by 6.9mm after 14 weeks of pembrolizumab therapy.

Genomic and immunologic characterization of INI1-deficient pediatric cancers

Suzanne J. Forrest, Alyaa Al-Ibraheemi, Duong Doan, et al.

Clin Cancer Res Published OnlineFirst March 2, 2020.

Updated version Access the most recent version of this article at: doi:10.1158/1078-0432.CCR-19-3089

Supplementary Access the most recent supplemental material at: Material http://clincancerres.aacrjournals.org/content/suppl/2020/02/29/1078-0432.CCR-19-3089.DC1

Author Author manuscripts have been peer reviewed and accepted for publication but have not yet been Manuscript edited.

E-mail alerts Sign up to receive free email-alerts related to this article or journal.

Reprints and To order reprints of this article or to subscribe to the journal, contact the AACR Publications Subscriptions Department at [email protected].

Permissions To request permission to re-use all or part of this article, use this link http://clincancerres.aacrjournals.org/content/early/2020/02/29/1078-0432.CCR-19-3089. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.