Prediction of high-risk cancer patients using clinical factors and expression profile of apoptosis regulators

Abstract

With about 19 million occurrences and 10 million mortalities in 2020, “Cancer” is the second leading source of mortality worldwide (WHO GLOBOCAN). The top three continents burdened with cancer deaths are Asia (58.3 percent), Europe (19.6 percent) and Latin America (7.2 percent). As of today, patient management in cancer care involves three broad steps: (a) screening and diagnosis, (b) risk assessment and prognosis, and (c) therapy. Since therapeutic intervention follows the risk assessment step, it is known to be the most critical phase in the cancer care and treatment. Risk estimation is done by means of multiple staging schemes for most cancers. The 'TNM system' for which the staging directives are issued by the “American Joint Committee on Cancer” (AJCC) and the “Union for International Cancer Control” (UICC), is the most extensively used system. The overall stage in the TNM system is determined when a letter (often with a number) is allocated to the cancer to describe the stages of T: tumour, N: node and M: metastasis , in which T specifies the size and location of the initial tumour, N indicates cancer spread to the adjacent lymph nodes, and M shows the cancer spread to distant body parts. The traditional TNM staging only involved anatomical considerations, but the modern staging system is continuously revised to provide details on other characteristics such as cancer biomarkers that include the profile/status of certain molecules that are altered in cancer tissues and clinical characteristics such as the location of tumour or age. These insights are integrated into the staging processes for various kinds of cancer, which makes it more reliable and useful to both doctors and patients. For example the recent inclusion of HER2 status was a result of a new Neo-Bioscore staging system, thereby allowing more precise prognostic stratification of all breast cancer subtypes . The addition of ‘Age’ in Thyroid cancer staging has also been reported to improve risk assessment. The heterogeneity associated with cancer is a major hurdle in the formulation of “cancer biomarkers”, as each cancer is comprised of multiple phenotypes and frequently responds differently to the same therapeutic intervention. This heterogeneity exists because of the aberrant behaviour of cancer cells, not just in different types of cancer, but even in same cancer type. In order to resolve this and persuade a “personalized medicine” approach, modern oncologists are actively seeking to develop a thorough understanding of the molecular processes that drive cancer. Biomarker development using genomic and proteomic data is now considered to be a superior means of carefully approaching the problem of cancer heterogeneity. This is largely achieved by a detailed study of data obtained from subcellular processes that drive oncogenesis. In this study, we focused on a prominent cellular pathway, Apoptosis, which has a strong and proven background in the growth and development of cancer. In the framework of genomic data, for the particular case of thyroid cancer, we demonstrate that certain genes belonging to the apoptotic pathway are associated with patient survival. The elevation and suppression of mRNA levels of these genes may be responsible for an aggressive or a mild phenotype of thyroid cancer thereby affecting patient outcome. The proposed signature in a further analysis was shown to perform better than AJCC staging, for risk stratification purposes,. The identified genes also exhibit a differential expression between normal and cancerous tissue, suggesting their ability to distinguish between individuals with and without cancer. Further, it was shown that the application of a similar approach to a pan-cancer analysis revealed universal gene signatures that have prognostic significance across various cancer types. This is in contrast with the conventional cancer-specific biomarker development process. The study centred at identification of prognostic biomarker and devised a 11 gene panel that is applicable across 27 cancer types. Although, the panel’s efficiency is seen to differ among cancer types, a substantial stratification is achieved in all cases. In addition to this, the study provides a new cross cancer biomarker development approach and sheds light on a new gene signature that can be used in patients with brain or kidney cancer. In the area of cancer treatment and rehabilitation, the practical realisation of such versatile biomarkers poses enormous benefits. Gene expression profiling is a very accurate strategy for the understanding of cancer and its prognosis, but, in the context of signalling networks, the activity of these genes depends on their translation into functional proteins. Because fundamental protein families controlling the apoptotic pathway together with their roles are commonly known, an in-depth study of the proteomic profiles of different tissues retrieved from cancer diagnosed individuals is anticipated to enhance our comprehension of tumour pathogenesis, prognosis, and recognition of therapeutic targets. To this end, the analysis included a proteomic dataset with the expression profile of Bcl2 family proteins in the scope of colo-rectal cancer. Information from previous apoptotic pathway studies has been used to establish a predictive biomarker for the estimation of response to treatment in colorectal cancer patients. This research illuminated the synergistic function of proteins in conferring therapeutic “resistance” to colorectal cancer and the critical role of apoptosis. The prognostic power of the biomarker was compared to different clinical features and methods. The method was released into public domain by the means of a web-server, thereby enforcing its practical utility to both researchers and clinicians. However, a major problem with biomarkers focused on "omics" is that inclusion of these biomarkers makes staging processes more complicated, rendering them difficult for people to understand. Thus, considering their outstanding success in cohort trials, most biomarkers have not yet been applied to the staging schemes. Therefore, our current research also examines the importance of numerous 'clinical factors/features' that collectively include pathological features, demographic characteristics, lifestyle-related features, anatomical characteristics, blood protein status (such as ER) in evaluating cancer patients' survival outcomes. Apart from the comprehensive assessment of clinical factors and their integration to the gene/protein signatures proposed above, we explicitly studied the case of "Melanoma" and looked at the prognostic power of genomic information pertaining to many cancer-associated pathways as well as clinical factors. We demonstrate that a prognostic model that incorportates only clinical factors is superior to the model focused on gene expression. This research also illustrates the significance of clinical factors for risk assessment. It shows how the schematic incorporation of existing clinical features into the staging process can be more successful. It also indicates that while omics-based biomarkers could be desirable due to their inherent biological correlation, clinical factors should not be undermined. On the basis of this pretext, the study is further expanded to the pan-cancer framework of designing risk prediction models by using only clinical factors. The clinical factors concerned include a wide variety of characteristics, ranging from inherent or heritable factors, different extrinsic risk factors, physiological features and surgical or therapeutic procedures used. The study established risk prediction models that are easy to apply and understand. Models were also evalauted against staging systems in various cancer cohorts. Overall, the study discussed in this thesis suggests some novel prognostic biomarkers and approaches for improved risk management in cancer patients. On the one side, the pipeline used in the analysis exploited a key cellular mechanism by using recent "omics"-based information. On the other hand, different clinical factors were examined both independently and in conjunction with proposed biomarker genes/proteins in regard to patient survival. The study discussed here can be useful for the development of better therapeutic modalities and thereby aid in the advancement of cancer research.