Matching Items (3)
Filtering by
- Genre: Doctoral Dissertation
Description
Cancer is a disease of multicellularity, with deep evolutionary origins. As such, the forces of both evolution and natural selection operate on multiple scales to govern tumor dynamics. As multicellular organisms increase in complexity, cellular-level fitness must be controlled in order to maintain organismal-level fitness. Mutations that might provide a benefit at the cellular level by allowing for rapid proliferation are subject to the same forces that function on the organismal level, wherein cancer suppression is a benefit – especially as organisms increase their body size and lifespan. In order to maintain these large cellular bodies and long lifespans, organisms must increase their means of cancer suppression, and it is likely that these two phenomena co-evolved together. On a smaller scale, the cooperative dynamics of circulating tumor cell (CTC) clusters engage in cooperation to form networks of connected single cells that provide protection, stability, and cooperative sharing of resources to enhance their survival as they detach from a primary tumor and metastasize at secondary sites. This work seeks to explore the phenomenon of multi-level selection in neoplastic disease by examining A) the mechanisms of cancer suppression at multiple scales, B) the ecological resilience and stability of cooperating cellular clusters and C) a large-scale dataset on cancer prevalence across mammals, sauropsids (birds and reptiles), and amphibians, illuminating the evolutionary life history characteristics that explain the tradeoffs between cancer suppression and overall organism fitness. By taking an ecological and evolutionary approach to understanding cancer, novel strategies of cancer treatment may be discovered alongside fundamental discoveries about the fundamental forces of selection that govern evolutionary dynamics from the cellular to the organismal scale.
ContributorsHarris, Valerie (Author) / Maley, Carlo C. (Thesis advisor) / Aktipis, Athena (Committee member) / Boddy, Amy M. (Committee member) / Compton, Carolyn (Committee member) / Arizona State University (Publisher)
Created2022
Description
Cancer researchers have traditionally used a handful of markers to understand the origin of tumors and to predict therapeutic response. Additionally, performing machine learning activities on disparate data sources of varying quality is fraught with inherent bias. The Caris Life Sciences Molecular Database (CMD) is an immense resource for discovery as it contains over 215,000 molecular profiles of tumors with consistently gathered clinical grade molecular data along with immense amounts of clinical outcomes data. This resource was leveraged to generate two artificial intelligence algorithms aiding in diagnosis and one for therapy selection.
The Molecular Disease Classifier (MDC) was trained on 34,352 cases and tested on 15,473 unambiguously diagnosed cases. The MDC predicted the correct tumor type out of thirteen possibilities in the labeled data set with sensitivity, specificity, PPV, and NPV of 90.5%, 99.2%, 90.5% and 99.2% respectively when considering up to 5 predictions for a case.
The availability of whole transcriptome data in the CMD prompted its inclusion into a new platform called MI GPSai (MI Genomic Prevalence Score). The algorithm trained on genomic data from 34,352 cases and genomic and transcriptomic data from 23,137 cases and was validated on 19,555 cases. MI GPSai can predict the correct tumor type out of 21 possibilities on 93% of cases with 94% accuracy. When considering the top two predictions for a case, the accuracy increases to 97%.
Finally, a 67 gene molecular signature predictive of efficacy of oxaliplatin-based chemotherapy in patients with metastatic colorectal cancer was developed - FOLFOXai. The signature was predictive of survival in an independent real-world evidence (RWE) dataset of 412 patients who had received FOLFOX/BV in 1st line and inversely predictive of survival in RWE data from 55 patients who had received 1st line FOLFIRI. Blinded analysis of TRIBE2 samples confirmed that FOLFOXai was predictive of OS in both oxaliplatin-containing arms (FOLFOX HR=0.629, p=0.04 and FOLFOXIRI HR=0.483, p=0.02).
The Molecular Disease Classifier (MDC) was trained on 34,352 cases and tested on 15,473 unambiguously diagnosed cases. The MDC predicted the correct tumor type out of thirteen possibilities in the labeled data set with sensitivity, specificity, PPV, and NPV of 90.5%, 99.2%, 90.5% and 99.2% respectively when considering up to 5 predictions for a case.
The availability of whole transcriptome data in the CMD prompted its inclusion into a new platform called MI GPSai (MI Genomic Prevalence Score). The algorithm trained on genomic data from 34,352 cases and genomic and transcriptomic data from 23,137 cases and was validated on 19,555 cases. MI GPSai can predict the correct tumor type out of 21 possibilities on 93% of cases with 94% accuracy. When considering the top two predictions for a case, the accuracy increases to 97%.
Finally, a 67 gene molecular signature predictive of efficacy of oxaliplatin-based chemotherapy in patients with metastatic colorectal cancer was developed - FOLFOXai. The signature was predictive of survival in an independent real-world evidence (RWE) dataset of 412 patients who had received FOLFOX/BV in 1st line and inversely predictive of survival in RWE data from 55 patients who had received 1st line FOLFIRI. Blinded analysis of TRIBE2 samples confirmed that FOLFOXai was predictive of OS in both oxaliplatin-containing arms (FOLFOX HR=0.629, p=0.04 and FOLFOXIRI HR=0.483, p=0.02).
ContributorsAbraham, Jim (Author) / Spetzler, David (Thesis advisor) / Frasch, Wayne (Thesis advisor) / Lake, Douglas (Committee member) / Compton, Carolyn (Committee member) / Arizona State University (Publisher)
Created2020
Description
Evolutionary theory provides a rich framework for understanding cancer dynamics across scales of biological organization. The field of cancer evolution has largely been divided into two domains, comparative oncology - the study of cancer across the tree of life, and tumor evolution. This work provides a theoretical framework to unify these subfields with the intent that an understanding of the evolutionary dynamics driving cancer risk at one scale can inform the understanding of the dynamics on another scale. The evolution of multicellular life and the unique vulnerabilities in the cellular mechanisms that underpin it explain the ubiquity of cancer prevalence across the tree of life. The breakdown in cellular cooperation and communication that were required for multicellular life define the hallmarks of cancer. As divergent life histories drove speciation events, it similarly drove divergences in fundamental cancer risk across species. An understanding of the impact that species’ life history theory has on the underlying network of multicellular cooperation and somatic evolution allows for robust predictions on cross-species cancer risk. A large-scale veterinary cancer database is utilized to validate many of the predictions on cancer risk made from life history evolution. Changing scales to the cellular level, it lays predictions on the fate of somatic mutations and the fitness benefits they confer to neoplastic cells compared to their healthy counterparts. The cancer hallmarks, far more than just a way to unify the many seemingly unique pathologies defined as cancer, is a powerful toolset to understand how specific mutations may change the fitness of somatic cells throughout carcinogenesis and tumor progression. Alongside highlighting the significant advances in evolutionary approaches to cancer across scales, this work provides a lucid confirmation that an understanding of both scales provides the most complete portrait of evolutionary cancer dynamics.
ContributorsCompton, Zachary Taylor (Author) / Maley, Carlo C. (Thesis advisor) / Aktipis, Athena (Committee member) / Buetow, Kenneth (Committee member) / Nedelcu, Aurora (Committee member) / Compton, Carolyn (Committee member) / Arizona State University (Publisher)
Created2023