MEDICAL STUDENT SUMMER FELLOWSHIPS

Yekaterina Shpanskaya*

Radiomic Discovery of Novel Prognostic Imaging Biomarkers in Medulloblastoma

Medulloblastoma is the most common malignant brain tumor in children. There are four known molecular subtypes (SHH, WNT, group 3, and group 4) each with different patient prognosis. Currently, strategies for identifying which subtype patients have are costly, invasive and require complex technologies that are not readily available to most hospitals.

Magnetic Resonance Imaging (MRI) is conducted for all medulloblastoma patients as a primary method for tumor diagnosis and surveillance. In this project, we identified accurate quantitative MRI markers that relate to the different medulloblastoma molecular subtypes. To do this, we used advanced computer-based imaging techniques combined with machine-learning algorithms.

In our large set of 109 medulloblastoma patients from three institutions in the United States and Canada, we extracted 590 MRI features that describe tumor characteristics. Our algorithms were able to successfully predict medulloblastoma subtypes that typically have moderate prognosis (SHH and group 4). Importantly, the algorithms were able to predict these from images taken within and across the three institutions, indicating that the technology can be used at other institutions. Tumors with favorable prognoses (WNT) and those with very poor prognosis (group 3) were more challenging to predict, likely because there were too few cases of these tumors in our set to adequately develop the algorithms.

This work demonstrated for the first time that quantitative MRI features can successfully predict molecular markers of medulloblastoma. Therefore, it is possible that computer software programmed with our algorithms could provide for a rapid, non-invasive, and low-cost alternative to determining medulloblastoma subtypes and {tooltip}prognosis{end-texte}Prediction of the course of a disease.{end-tooltip}.

*Rubinstein Award Winner


Neil Almeida

Development of mass cytometry probes to assess function and phenotype of T-cells

The Okada lab recently implemented a novel vaccine trial in patients with grade II gliomas. Genetic differences among the cells that make up a tumor pose a major challenge when trying to effectively treat a tumor.

  • Antigen: A molecule that can activate an immune response to fight tumor cells.

Selecting a single antigen as the target of immunotherapy often leads to failure because the tumor cells that are not weakened by that specific antigen will continue to grow. Therefore, in the current trial we selected 10 antigens that are found at high levels in glioma cells.

To monitor patients’ immune responses against each of these 10 target antigens, we utilized a new technology called CyTOF mass cytometry. CyTOF can detect and measure various characteristics of many different cells at the same time. Specific probes, called tetramers, were developed to detect immune responses against the antigens. We also created a special probe to validate our tetramer production program, which will be used for generating the tetramers to analyze tumor and blood samples derived from patients on the clinical trial.


Joshua Bernstock, PhD

Immune checkpoint blockade in combination with oHSVs for pediatric brain tumors

Oncolytic herpes simplex virus-1 (oHSV; more commonly known as herpes or a cold-sore virus) is a promising approach to target brain tumors. Genetically modified forms of oHSV, called G207 and M032, are currently in phase I clinical trials in both children and adults respectively.

While these viruses target and kill tumor cells, they also help activate the body’s immune system to kill the remaining tumor cells. However, boosting and then sustaining this anti-tumor immune response is the primary challenge of this type of therapy.

In this project, we compared the sensitivity of patient-derived pediatric and adult tumors to these viruses in mouse models. To boost the immune response, we combined oHSVs with an {tooltip}immune checkpoint inhibitor{end-texte}A type of immunotherapy that blocks the molecules that cancer cells use to hide from the immune system.{end-tooltip}.

In doing so, we found:

  1. Pediatric brain tumors were more sensitive to the viruses than adult brain tumors. Our results suggest that pediatric brain tumors are ideal targets for oHSV.
  2. The pediatric tumors also had high levels of a protein called nectin-1, which may be a useful biomarker to predict patient response to oHSV.

Pediatric brain tumors express targetable {tooltip}immune checkpoint molecules{end-texte}Molecules that cancer cells use to hide from the immune system.{end-tooltip}, particularly IDO, which we are continuing to investigate. Our preliminary data suggests that blocking IDO in combination with oHSV accelerates the killing of tumor cells in a pediatric group 3 medulloblastoma mouse model.


Saksham Gupta

Defining the Immunophenotype of Meningioma

Traditionally, tumor behavior and response to therapy has been predicted by identifying a few key genetic changes or biomarkers in the tumor. Typically, these predictions do not take into account the microenvironment, which we now know plays a key role in tumor development and response to treatments.

  • Tumor microenvironment: Everything that is in and around a tumor, including blood vessels, immune cells, other normal cells, hormones and other signaling molecules, and the proteins that give tissues structure.

The goal of this project was to characterize immune biomarkers in the tumor microenvironment for meningiomas.

We found that:

  1. More aggressive meningiomas have biomarkers that indicate that the tumor microenvironment suppresses the immune.
  2. Compared to other tumors, meningiomas appear to have more macrophages (a type of immune cell); however, additional studies will need to be done to fully understand how this impacts the clinical course of the disease.

These preliminary findings are the first step in understanding the microenvironment of each tumor and understanding the clinical significance of each part of that microenvironment.


Casey Jarvis

Effects of Host Pericyte Deficiency on Angiogenesis in Glioblastoma

Glioblastoma (GBM) is the most common and deadly type of primary brain cancer, with 12,500 new cases diagnosed each year. Despite aggressive therapy, most patients do not survive more than a year after diagnosis.

  • Angiogenesis: The growth of new blood vessels. Angiogenesis is essential for tumor growth and progression.
  • Pericytes: {tooltip}Contractile{end-texte}Capable of contracting or causing contraction.{end-tooltip} cells that wrap around the cells that form blood vessels to control the expanding and contracting of capillaries.

For this project, we proposed that if pericytes were absent in and around GBM tumors, then angiogenesis would be blocked, preventing GBM growth.

To study this, we compared GBM tumors in normal mice and in mice that lack pericytes. In mice that lacked pericytes, we observed reduced tumor growth and poorly formed blood vessels. These tumors also had an increase in necrosis (cell death).

In order to better understand how pericytes affect brain tumors, we have begun further studies focusing on the changes in {tooltip}gene expression{end-texte}Whether genes are turned on or off.{end-tooltip} of GBM cells and the compounds that they release into their environment when exposed to pericytes. We are continuing to work towards our goal of identifying new targets for developing new treatments that we hope will improve outcomes and quality of life for patients with glioblastoma in the future.


Adam Lauko

Effects of a Glioblastoma Secreted Cytokine on Myeloid-Derived Suppressor Cells

Glioblastoma (GBM) is the most common malignant brain tumor and continues to have a poor prognosis, with survival averaging between 12 and 15 months. While advancements in chemotherapy and radiation have been effective in other cancers, improvement of these treatments has done little to increase glioblastoma survival rates. One explanation is that glioblastoma does an excellent job at shutting down the immune system at the tumor site, allowing the tumor to hide.

Normally, after fighting an infection, cells known as myeloid-derived suppressor cells (MDSCs) deploy a full arsenal of proteins that shut down the immune system, preventing damage to healthy cells. We recently discovered that glioblastoma has figured out a way to hack this regulation, causing MDSCs to shut down the immune system at the tumor site.

  • Macrophage migration inhibitory factor (MIF): The molecule that glioblastoma uses to suppress immune function.

For this project, I studied two ways that MIF suppresses the immune system:

  1. I looked at the signaling that occurs inside of MDSCs after MIF stimulates the cells and found that a MIF-blocking drug decreased the growth signals in MDSCs. This indicates that MIF can turn on this immune suppressor function, and it can be turned off with MIF-blocking drugs.
  2. I also examined the effects of MIF on another type of immune cell called Natural Killer (NK) cells. I found that when NK cells were exposed to MIF, they were less effective in killing GBM cells in culture.

These findings, along with additional studies on how MIF acts on these immune cells, will bring us closer to identifying novel treatments that use a person’s own immune system to target and fight cancer.


Gina Rhee

The contribution of stromal cell senescence to sex differences in glioblastoma

Men have a greater chance of developing brain tumors and poorer survival rates compared to women. Additionally, the risk for brain tumors increases with age more rapidly for men than women. These patterns show that patient sex likely influences brain tumor formation and growth, particularly in age-related processes.

As cells acquire damage over time, there is an increased risk that that they can become cancerous. As a defense mechanism, damaged cells can become senescent or “retire” and permanently halt growth. However, senescent cells are not completely inactive. They can release compounds that can impact the cells around them, and can even impact tumor formation and growth.

There is little known about the role of senescent cells in the brain or whether there are differences between male and female senescent brain cells that may contribute to the higher male risk for brain tumors.

In this study, we compared the senescence process in male and female brain cells and found that:

  1. Female brain cells become senescent more readily than male brain cells.
  2. Male senescent brain cells release compounds that encourage tumor cell growth while female senescent brain cells release compounds that inhibit tumor cell growth.

These results support that male senescent brain cells seem to promote tumors, while female senescent brain cells seem to suppress tumors. Continued research in this area will be insightful in uncovering age- and sex-related risk factors for brain tumors.


Johnathan Zeng

Optimizing a Radiomic Risk Score to Characterize Brain Tumor Progression in MRI

Radiation therapy is a common treatment that can be effective for patients with metastatic brain cancer. Unfortunately, radiation necrosis, or dead tissue, is a common side effect of radiation therapy.

When viewing an MRI of the brain, tumors (both new and recurrent) and dead tissue look very similar so it is difficult to distinguish one from another. Many times, more invasive methods, such as surgery, are used to confirm the diagnosis in order to determine the next steps in treating the patient. Therefore, our lab set out to find an effective, less invasive method to distinguish new tumors from dead tissue.

In this project, we tested a software called CoLIAGe, which uses advanced mathematics and machine learning to conduct a special imaging analysis on MRI scans to accurately predict whether or not a patient has a new or recurrent brain tumor. We first had to modify our CoLlAGe software so that we could test it on existing patient records in the hospital. Using CoLIAGe, we have begun to identify features on MRI that are different in patients with recurrent tumors than in patients with radiation necrosis. Going forward we plan to further develop and optimize this technology by testing it on additional patient records.