Using Online Cancer Genomics Databases to Provide Teaching Resources for Pharmacy Education

DISCUSSION

Targeted oncology drugs inhibit specific molecular drivers of cancer progression, including somatic mutations. Clinical trials of targeted drugs that include -omics analyses generally share datasets through publicly accessible cancer genomics databases. Datasets may include tumor genomics, such as differential gene analyses, somatic mutations, polymorphisms, gene amplifications, and chromosomal rearrangements, as well as pharmacotherapy and clinical outcomes data. One example of a cancer genomics database is the cBioPortal database (https://www.cbioportal.org/), which contains more than 300 datasets representing over 30 distinct tumor types.^9,10 The data available in each dataset vary and may include genomics, proteomics, tumor mutation count, gene amplification events, race, sex, age, tumor histopathology, treatment regimens, and clinical outcomes. Because of the breadth of genomic and genetic data available from trials of targeted cancer therapies, cBioPortal represents a valuable resource for oncology pharmacogenomics education. Another cancer genomics database, Oncomine (https://www.oncomine.org), contains data from more than 4000 gene microarray experiments.¹¹ Oncomine supports differential gene expression analysis by tumor type, facilitating classroom discussions about pharmacogenomics and cancer precision medicine. Another resource, KM Plotter (https://kmplot.com), is designed to identify tumor markers associated with survival outcomes across 21 tumor types.¹² Data retrieved from KM Plotter can be used to support learning activities focused on understanding how genetic variants impact disease course and clinical outcomes.

To illustrate the application of a cancer genomics database in pharmacy education, we provide three examples using datasets from the cBioPortal database. The first two examples target CAPE core competencies to understand how genetic mutations inform pharmacotherapy selection based on an approved pharmacogenetics test. The third example targets a learning objective to understand immunotherapy and tumor genomics profiling. We present a case-based discussion with each example to illustrate how the data support active learning.

For our first example of using a cancer genomics database, we discuss how a somatic mutation in PIK3CA can be used to guide treatment of breast cancer (Appendix 1A). We started the data retrieval process by entering a specific tumor type and drug name. Following the steps shown in the flowchart, we mined cBioPortal to obtain data for a single patient case from a clinical trial of alpelisib in breast cancer.¹³ Case details, including tumor subtype, genetic mutation, treatment regimen and response, were downloaded from cBioPortal as shown in the table. Using these data, we created a case-based discussion (shown below) focused on treatment selection guided by molecular and genetic information, pharmacogenetic testing, and drug-drug interactions. In addition to this case, we downloaded data for a larger cohort of patients (shown in the graph) to introduce concepts related to clinical response and resistance.

In this case, a postmenopausal female patient presented with advanced breast cancer. The patient’s breast tumor biopsy was positive for estrogen receptor (ER) and negative for human epidermal growth factor receptor 2 (HER2). The patient was started on exemestane. However, disease progression eventually occurred, and pharmacogenetics testing of tumor tissue was then performed for the PIK3CA gene. The patient was found to have an H1047R PIK3CA somatic mutation and was started on alpelisib plus fulvestrant.

Case study questions should include: What are exemestane and alpelisib? What information guided selection of these treatments? Why does a positive pharmacogenetics test not guarantee response to a targeted treatment? What drug-drug interactions need to be considered with alpelisib? Questions in the case-based discussion guide should focus on targeted therapies, including those targeting somatic mutations and companion pharmacogenetics testing. Discussion should also note variability in patient responses and could include discussions about intra- and inter-tumoral heterogeneity, patient-specific factors influencing response, and acquired drug resistance. Finally, drug-drug interactions and the effects of cytochrome p450 (CYP) enzymes on drug activity should be discussed.

Our second example describes how a somatic mutation in BRAF can be used to guide treatment of melanoma (Appendix 1B). For this case, we started by entering a specific cancer type and gene name. Similar to the process described above, we mined cBioPortal following the steps in the flowchart to identify a single patient case of BRAF-mutant melanoma.¹⁴ In addition to the single case, we downloaded a graph showing responses in a larger cohort of patients with BRAF-mutant melanoma treated with dabrafenib or vemurafenib, which provided us with the opportunity to discuss other compounds in this class of drugs. We also introduced the concept of pharmacogenetics testing to assess a marker associated with adverse events.

In this case, a 44-year old female patient with metastatic melanoma was found to have a somatic BRAF^V600E mutation. The patient was also found to have wild-type alleles for the glucose-6-phosphate dehydrogenase (G6PD) gene. The patient was treated with dabrafenib for 30 weeks with minimal toxicity and exhibited complete response to therapy.

Case study questions included: What information guided the selection of dabrafenib as a treatment in this patient? What is the mechanism of action of dabrafenib and vemurafenib? Why is pharmacogenetics testing for the G6PD gene important when considering dabrafenib as a potential treatment? Questions in the case-based discussion guide should address BRAF mutations, pharmacogenetics testing for somatic mutations, and mechanisms of action of kinase-targeted therapy. The question about G6PD testing should guide discussion about patient-specific factors that may influence adverse drug effects.

For our third example, we discussed PD-1-directed immunotherapy and tumor genome alterations (Appendix 2). For this case, we started by entering a drug name and mined cBioPortal to display tumor gene analysis from a translational study in patients with melanoma.¹⁵ We downloaded differential gene expression data for pre-treatment tumor biopsies of responders and non-responders and a case profile to discuss concepts related to immunotherapy and tumor genomics.

In this case, an 84-year-old male patient with metastatic melanoma showed disease progression while on prior therapy with a mitogen-activated protein kinase (MAPK) inhibitor. The patient was started on treatment with pembrolizumab as part of a trial evaluating pembrolizumab or nivolumab in patients with melanoma. Immune cell profiling and pre-treatment tumor genomics profiling were performed as part of this trial.

Case study questions included: What is the mechanism of action of pembrolizumab and nivolumab? Why was immune cell profiling performed? What do the data in the tumor genomics table show? The case questions should guide discussion about cancer immunotherapy, immune cell types and functions, and the role of genomics data in biomarker and drug discovery research.

These examples are provided to illustrate the potential utility of genomics databases for case study development in cancer pharmacogenomics and pharmacogenetics. Although student feedback, perceptions, and knowledge gains remain to be assessed, we believe there are several advantages to this approach. First, integrating clinical datasets into active learning may assist students in applying foundational concepts to solve practical problems,¹⁶ such as the use of genetics data to create informed treatment plans. Second, faculty who lack clinical context may benefit from the ability to retrieve real-life cases from genomics databases, rather than creating hypothetical scenarios. Thus, clinical genomics databases may support pre-course preparation and potentially reduce interdisciplinary workload.¹⁷ Third, students can be introduced to emerging pharmacogenomics research, which may enhance their awareness of academic research and related career opportunities.^18,19 Fourth, the functions we used to retrieve data were freely accessible, making these cost-effective options.¹⁹ Pharmacy programs with limited resources can integrate genomics databases into mentored research experiences to foster critical thinking and problem-solving.^18,19 Finally, online databases can be used in synchronous, asynchronous, in-person, blended, and virtual learning environments, making them versatile resources.

Despite the benefits of publicly accessible cancer genomics databases, there are limitations. Because of a wide range of user interfaces, databases vary in complexity and in the amount of time and skill required to retrieve and analyze data. Additionally, although there are many publicly accessible databases suitable for oncology pharmacy education, faculty must be made aware of these resources and know where to find datasets appropriate for their specific learning objectives. We found online tutorials, manuals, and published research articles helpful as we learned to use these databases.