TL;DR
This paper introduces DBPOM, a novel, comprehensive database that integrates pharmaco-omics data to predict both the therapeutic and adverse effects of drugs on cancer patients. By bridging the gap between cell line studies and patient-level analysis, DBPOM aims to accelerate drug development and enable more personalized cancer treatments.
What Problem Does This Paper Address?
Current approaches to understanding drug response in cancer often focus on efficacy while neglecting adverse effects, or they are limited to cell line studies. This creates a significant challenge for clinicians and researchers trying to implement precision medicine, as the complex interplay of drug effects at the patient level remains poorly understood. Integrating data from various sources and performing extensive calculations is time-consuming and difficult. The authors aim to address this by creating a unified, user-friendly platform that consolidates pharmaco-omics data and provides a more holistic view of drug impact.
Methods
The DBPOM database is built upon a multi-step process:
- Data Collection: Genomics and transcriptomics data from The Cancer Genome Atlas (TCGA) and the Cancer Cell Line Encyclopedia (CCLE) were collected, alongside compound-perturbed gene expression profiles from CMap and IC50 data from GDSC.
- Data Restructure and Analysis:
- Differential Gene Expression (E1): Genes associated with carcinogenesis were identified by comparing cancer samples to normal samples from TCGA using DESeq2, with thresholds for significance set at p-value=0.001 and fold-change=2.
- Differential Gene Expression (E2): The effect of compounds on cell lines was assessed by comparing compound-treated cell lines to DMSO-treated cell lines from CMap. Genes were considered significantly differentially expressed if their fold-change was above 15.
- Reversed and Adverse Effect Estimation: A tool was developed to distinguish between "reversed" genes (where expression is opposite in cancer vs. normal and in treated vs. untreated cell lines) and "adverse" genes (where expression changes are not oppositional).
- RADE Score: A "Reversed and Adverse of Drug Effect" (RADE) score was defined to quantify the overall safety and effectiveness of a compound, with lower scores indicating higher effectiveness and safety.
- Mutation Mapping: Genomic mutation information from CCLE and TCGA was used to estimate similarity scores between cancer cell lines and patient tissues, enabling the mapping of individual mutation phenotypes to cell line responses. Three mapping options are provided: TCGA vs. CMap, subTCGA-mutation vs. CMap, and subTCGA-specific vs. CMap.
- Pathway Enrichment Analysis: To understand the biological pathways targeted by drugs, function analysis was performed on reversed and adverse genes using the hypergeometric test.
Datasets
The DBPOM database integrates data from several major sources:
- The Cancer Genome Atlas (TCGA): Over 3,000 cancer and normal samples across five cancer types.
- Cancer Cell Line Encyclopedia (CCLE): 28 cancer cell lines with genomic mutation data.
- CMap (Connectivity Map): Over 14,000 compound-perturbed gene expression profiles.
- GDSC (Genomics of Drug Sensitivity in Cancer): Nearly 4,000 IC50 profiles.
- Literature: Over 500 examples of drug combination information.
The database itself contains: 19,406 small molecular compounds and drugs, 28 cell lines, 5 cancer types, 27 dosages, 3,078 TCGA cancer samples and normal samples, 2,598 specific mutations with high mutation frequency, 397 FDA-approved cancer drug combination pairs, and 112 literature-published cancer drug combination pairs, totaling 376,486 files.
Key Insights and Ideas
- Bridging the Cell Line-Patient Gap: A major contribution is the explicit effort to connect cell line drug response data to patient-level outcomes by incorporating mutation similarity mapping. This is crucial for translating findings from in vitro studies to clinical relevance.
- Holistic Drug Assessment: The DBPOM's focus on both therapeutic (reversed) and adverse effects provides a more comprehensive picture of drug behavior than many existing resources.
- RADE Score as a Predictive Metric: The development of the RADE score offers a quantitative way to assess drug safety and efficacy, which the authors demonstrate can be reasonably successful in evaluating drug efficiency.
- Personalized Medicine Focus: The database is designed with a patient-centric approach, offering modules for personalized medicine that leverage mutation profiles to guide treatment recommendations.
Key Results
- The DBPOM database integrates a substantial amount of diverse data, including genomics, transcriptomics, and drug response profiles.
- The RADE score was applied to FDA-approved drugs and those in clinical trials, showing that drugs with low RADE values were generally effective. For example, vemurafenib and curcumin, known for their therapeutic roles, showed high RADE values in the database, indicating their effectiveness.
- The database provides detailed information on reversed and adverse genes, as well as pathway enrichment analyses for specific compounds.
- The personalized medicine module allows users to explore drug effects based on specific mutations or upload their own patient data.
Limitations and Future Work
- Data Collection Cutoff: All data was collected before September 2019, meaning more recent findings may not be included.
- Adverse Effects Definition: The paper notes that the "adverse effects" defined in the study are distinct from pharmacological adverse reactions observed in clinical trials, suggesting a need for further validation and integration of clinical data.
- Future Directions: The authors plan to continue collecting data for more cancer types, cell lines, compounds, and drug combinations to enrich the database. They also intend to develop and integrate more computational tools for effect estimation and enhance precision medicine recommendation capabilities. The ultimate goal is for DBPOM to become a vital resource for drug development, mechanism studies, and the discovery of new therapies.
Summary updated