Hi, everyone. My name is Jiaxi Joseph Chen, and I'm a second-year medical student at the Warren Alpert Medical School of Brown University. Today, I will be presenting findings from a project performed in the lab of Dr. Manish Aghi at UCSF that investigated the clinical significance of cytolytic activity in patients with medulloblastoma. This talk is titled Cytolytic Score Correlates with an Immunosuppressive Tumor Microenvironment and Reduced Survival in Medulloblastoma. I have no disclosures. The disease of interest in this study was medulloblastoma, which is the most common solid malignant tumor of children. These lesions are usually situated near the fourth ventricle and often metastasized through CSF pathways. Given its ability to metastasize to extraneural tissue, standard of care consists of maximal safe surgical resection, followed by chemotherapy and craniospinal radiotherapy. Five-year survival is encouraging, ranging from 60 to 80 percent. However, long-term morbidity is significant. Neurological, cognitive, and endocrine deficits are prevalent, largely due to the effects of radiation on the developing brain. As a result, immunotherapy is an attractive treatment modality because it optimizes immune cells to kill tumor cells while sparing the surrounding brain. However, there is an unmet need for immunotherapy largely because the contributions of the immune system to medulloblastoma tumor genesis, recurrence, treatment response, and survival remain unclear. Additionally, the advent of molecular diagnostics has led to four molecular subtypes, group 3, group 4, sonic hedgehog, and Wnt, each of which possess distinct transcriptional profiles, genetic abnormalities, treatment responses, and outcomes. Delivering effective immunotherapy is thus further complicated because each subtype likely responds differently, meaning the immune landscape must be studied and understood in a subgroup-specific context. Tumor-infiltrating immune cells, notably lymphocytes, can predict survival and response to chemotherapy and immunotherapy in various solid tumors. Characterizing the tumor microenvironment, notably the abundance and activity of tumor-infiltrating lymphocytes, is important for advancing the immunotherapy frontier. However, there is a paucity of studies that have investigated this in medulloblastoma. Given our current understanding of medulloblastoma, this project was performed with the following objectives in mind. First, we aimed to determine whether the abundance of tumor-infiltrating lymphocytes and degree of cytolytic activity was associated with survival in medulloblastoma. Next, we wanted to characterize the abundance of other relevant immune cells and the immunosuppressive milieu in medulloblastoma to understand how the rest of the immune system may be influencing the level and efficacy of cytolytic activity. Lastly, we wanted to see whether our findings were consistent across all molecular subtypes or only relevant in a subtype-specific manner. Previous studies on colorectal cancer, hepatocellular carcinoma, and pancreatic cancer have associated higher cytolytic score with better prognosis. However, studies associating cytolytic score with glioblastoma from our group and others have noticed an inverse relationship. Thus, while the natural hypothesis may be to assume that medulloblastoma prognosis improves with increased cytolytic score, the converse may be true given the unique relationship between central nervous system pathologies and the immune system. To accomplish these objectives, we undertook a bioinformatics approach to interrogate the transcriptional and clinical profiles of patients with medulloblastoma. The patient cohort in this study was a set of 92 medulloblastoma patients with RNA sequencing and clinical data, which was acquired from the Gabriella Miller Kids First Data Resource Center. With the RNA-seq data, we first quantified the cytolytic activity of tumor-infiltrating lymphocytes for each patient via a published technique by Rooney et al. This cytolytic activity score is the geometric mean of Gramzyme A and Perforin 1 transcript levels, which are two cytolytic effectors expressed specifically within cytotoxic T lymphocytes and natural killer cells. After calculating cytolytic activity scores, the cohort was dichotomized into two groups of high and low cytolytic activity using maximally selected rank statistics, which aims to find a clinically relevant threshold within a continuous variable. The molecular subtype of each patient was then derived using a published transcriptome-based classifier by Rothi et al. that is based upon gene expression ratios. Finally, the abundance of 22 different tumor-infiltrating immune cells was enumerated using the Cybersort deconvolution algorithm, which uses support vector regression and a validated gene signature to quantify immune cell infiltration from bulk RNA sequencing data. With this process data, a series of statistical techniques were used to analyze it. Specifically, Kaplan-Meier analysis and Cox regression to associate cytolytic activity with survival, Spearman correlation to correlate cytolytic activity with the expression of other immune genes, and the Wilcoxon rank sum test to compare immune cell abundance between cohorts. Finally, results were validated using an independent published cohort of 74 patients with microarray and clinical data from Robinson et al. The results from our investigation are as follows. After determining the cytolytic activity score of each patient and performing maximally selected rank statistics, we found a dichotomization threshold that separated the cohort into 27 patients with high cytolytic activity and 65 patients with low cytolytic activity. When comparing the demographics of the two subgroups, there were no significant differences in gender, age at diagnosis, or distribution of molecular subtypes. During Kaplan-Meier analysis, patients with high cytolytic activity had significantly worse survival. The proportion of five-year survivors in this group was 59%, which is lower than the 70% five-year survival rate of contemporary SEER population studies and the 73% five-year survival rate of low cytolytic activity patients. This was confirmed during multivariate Cox regression controlling for demographic variables, which found high cytolytic activity to be independently associated with worse survival. Molecular subtype was also associated with survival, with group 3 subtypes having the worst prognosis, especially relative to group 4 tumors, which is consistent with published studies. When comparing tumor infiltrating immune cell populations between patients with high and low cytolytic activity, there were several populations with greater abundance in the high cytolytic cohort, most notably CD8-positive cytotoxic T lymphocytes, as expected, and the myeloid-derived cell lineage of tumor-associated monocytes, M1 macrophages, and M2 macrophages. When characterizing the expression of notable immune genes, cytolytic activity was positively correlated with expression of several immune checkpoints, specifically PD1, CTLA4, ID01, and TIM3. However, cytolytic activity was also positively correlated with the expression of genes required to develop an effective T cell immune response, such as those involved in antigen presentation, T cell recruitment, and activation. However, when evaluating our findings within each of the molecular subtypes, only group 3 and group 4 tumors with high cytolytic activity had worse survival than their counterparts with low cytolytic activity. To further elucidate the subtype-specific survival, we examined the possible mechanisms of immunosuppression that were previously highlighted in the entire cohort. First, immunosuppressive gene expression, specifically CTLA4 and PD1, was positively correlated with cytolytic activity in group 3 and group 4 tumors. However, in group 3 and group 4 tumors, there was no significant difference according to the level of cytolytic activity in the infiltration of tumor-associated macrophages, which had been associated with worse prognosis and immunosuppression. On the contrary, when examining the sonic, hedgehog, and Wnt tumors, patients with high cytolytic activity had a significantly greater infiltration of tumor-associated macrophages. Lastly, we confirmed that medulloblastoma patients with high infiltration of CD8-positive cytotoxic T lymphocytes had worse survival than those with low infiltration in an independent cohort. During multivariate Koch regression, high cytolytic activity was again independently associated with worse survival while controlling for other demographic variables. Finally, the survival difference was again only significant in group 3 and group 4 medulloblastoma subtypes. While further investigation is required to validate these findings, several preliminary conclusions can be made. First, greater cytolytic activity and infiltration of cytotoxic T lymphocytes is associated with worse survival in group 3 and group 4 medulloblastoma subtypes. Second, immunosuppressive checkpoints, notably PD1 and CTLA4, are potentially preventing increased cytolytic activity in group 3 and group 4 tumors from conferring a potential survival benefit. Third, group 3 and group 4 tumors may benefit from immune checkpoint inhibitor immunotherapy, whereas sonic hedgehog and Wnt tumors may benefit from therapies intended to inhibit tumor-associated macrophages, given their respective immune landscapes. These findings and conclusions are consistent with those of prior investigations, notably Buchmeier et al., who reported that the two patterns of medulloblastoma immunosuppression exist in a subtype-specific fashion. Group 3 and group 4 tumors had stronger infiltration of cytotoxic CD8 positive T cells and a simultaneous activation of immunosuppressive cytokine and checkpoint pathways, while sonic hedgehog and Wnt tumors had a primarily monocytic macrophage-supported tumor environment. However, additional retrospective studies with larger transcriptomic databases are required to validate these findings, as well as assess the clinical relevance of delivering immune checkpoint inhibitor immunotherapy to patients who have high immune cell infiltration and high immunosuppressive gene expression. Thank you very much for your interest in this investigation, and I would like to thank Dr. Manish Aghi for his mentorship and guidance through this project. My collaborators, Dr. Alex Haddad, Dr. Tianmin Ou, Elena Wang, and Sud Perera for their contributions to this project, and the rest of the Aghi lab for their incredible support. Thank you all once again.

cytolytic activity

medulloblastoma

survival

molecular subtypes

immune checkpoint inhibitors