Identifying molecular features associated with overall survival outcomes in Acute myeloid leukemia patients over the age of 60; How Physician-Assisted Death Redefines the Role of Physicians

The Introduction
Cancer has been a consistent and debilitating disease from the beginning of time. The different nefarious ways it may present are difficult to manage and treat. One of the deadliest of cancers is acute myeloid leukemia (AML). This cancer has a survival rate of 30% over 5 years, and this drops to 15% for patients above the age of 60. Predicting the survival of patients is critical for informing treatment options. Currently, the tool to make the prognosis for AML patients was created with data from patients below the age of 60. However, the average age of diagnosis for AML is 68 years old, and intuitively, the predictions made for those above 60, or aged AML (aAML) patients, are often inaccurate. When the outcome for survival is bleak, which is often the case with aAML patients, patients turn to different types of end-of-life care. One type that is offered in various states and countries is physician-assisted death (PAD). Depending on the country, different patients are eligible for a procedure that allows for a lethal dose to be administered to induce death. Understanding the role physicians play in this procedure redefines the way we think of physicians since in this case, they are inducing death instead of prolonging life.
The Body
To tackle the issue of poor prognosis tools for aAML patients, we took molecular data from about 220 aAML patients and analyzed them to find features that are predictive of survival outcome. These features would then be used in tandem to predict survival outcomes for patients. We explored copy number aberrations (CNAs) data that determines which sections of the DNA are amplified or deleted for each patient. By splitting the chromosomes into regions of 6,400 kbp, we used cox regression to identify regions that were associated with patient survival, but these findings were inconclusive. Next, we used RNA sequencing (RNA-seq) data that tells us which genes are being expressed. A PLSR model was trained on filtered RNA-seq data. This model achieved an R2 value of 0.9 proving effective in predicting patient survival with only 167 genes. Lastly, we considered DNA methylation data which denotes which sections of the DNA are compacted and thus inaccessible. Cox regression isolated 411 methylation sites which were clustered into three groups via hierarchical clustering. These patient groups had significant differences in their survival outcomes. From this, we can say that the features from RNA-seq and DNA methylation data should be used in a multivariate model to address the issues of the current low accuracy prognosis tools.
For practicing physicians, there has recently been a change in the responsibility of the role. Due to PAD, physicians are moving away from the historical role as a healer and preserver of life and toward a new redefined role. Because of the nature of PAD, state and national governments legislate to allow and regulate or criminalize the practice. I explored the legislation from multiple countries with varying regulations surrounding PAD and detailed how the different choices made by lawmakers change the role of the physician. Specifically, I looked at the Netherlands, Canada, Switzerland, and the United States. The countries allowed for PAD in different circumstances like for patients who are terminally ill and in Canada and Switzerland, for patients who are not terminal. A terminal diagnosis within six months in the US allows physicians to act as changing small but important details of a patient's death like location and timing. PAD for patients without a reasonably foreseeable death creates implications that physicians are actively causing patient death. Additionally, PAD can be carried out either by self-administration or by physician-administration of a lethal dose. This difference shifts the role of a physician from an aid in death to administrator of death.
The Conclusion
Overall, I am satisfied with the work I was able to do this year. Although I made less progress in my technical project than I had hoped, I learned about the time it takes to effectively create appropriate methodology and then shift gears when hit with a roadblock. This is crucial for me as I plan on joining a doctoral program in the coming fall. For the technical project, the molecular features found from the aAML data sets are important for future projects to use and validate when creating a robust and effective multi omics model for aAML prognosis. For the STS project, I began to detail the changes that physicians are facing with the relatively new technology of assisted death. This is not the first technology to redefine the role of a physician and it will not be the last.

School of Engineering and Applied Science

Bachelor of Science in Biomedical Engineering

Technical Advisor: Francine Garrett-Bakelman

STS Advisor: Kent Wayland

Technical Team Members: Deepika Sahoo