Skip to main content

Table 1 Preclinical and patient data necessary to model drug combination effects across the tumor microenvironmenta

From: Looking beyond the cancer cell for effective drug combinations

  Data type Advantages Limitations Recommendations
Pre-clinical Cancer cell-line drug screens - Cost-effective route to generate a significant amount of data
- Detailed data from many cell lines is available, for example, GDSC [93] and CCLE [94]
- Limited to only modeling intracellular effects
- Cell culture process affects important biology
- Limited to testing existing drug targets
- Obtain more data for drug combinations
- Use mixed cell assays to model the roles of other cells in the tumor and microenvironment
- Develop better models of diverse mechanisms (for example, longer assays and different endpoints)
Functional genomic screens (using siRNA, CRISPR, and mutagenesis) - No limit to the number of combinations of targets testable
- Synthetic lethalities can be readily identified
- Limited to intracellular mechanisms
- Focus on loss of function
- Only work in a limited number of cell backgrounds
- Use a broader range of cell contexts (including non-cancer cells)
- Develop gain-of-function screens
- Set up repositories that enable data to be shared and publicly accessible
Drug or target perturbation screens (post- treatment functional data) - Provide information about a drug or target’s mechanistic impact and provide pharmacodynamics maps that are likely to be relevant across cell types
- Large data sets are publicly available, for example, connectivity map and LINCS [63, 95]
- Typically focus on a few cancer cell types and/or global disease processes
- Only a few small in vivo screens are available
- Typically provide data on monotherapy only
- Obtain more data from non-tumor cell types involved in tumor biology
- Carry out larger in vivo screens and/or meta-analyses
- Acquire more data for drug combinations
Organoids (three-dimensional buds) or ex vivo screens Can be used to obtain data about cell–cell interactions (for example, interactions between tumor cells and cells in the microenvironment) and about environmental plasticity Few established and/or reproducible models Develop standards to identify non-typical phenotypic parameters that are relevant to the effects of a drug on the tumor microenvironments, for example, cell-type-specific effects and cell–cell communication
Patient-derived tumor xenograft screens Can model the effects of drugs on components of the tumor microenvironment - Do not model immune interactions
- Cost and ethical considerations need to be take into account when using as a discovery (versus test) tool
In vivo screens in GEM, syngeneic, or humanized models Can model immune interactions Cost and ethical considerations need to be taken into account when used as a discovery (versus test) tool
Patient Electronic health records Provide information about environmental exposures, immunological and metabolic measures, diagnostic assays, comorbidities and wellness, and longitudinal follow-up data - Key data are split across isolated records in primary care and specialist hospitals, claims systems, assay providers, and others
- There are currently no curation or digitization standards
- Address data confidentiality (for example, use honest brokers) and connect disparate records for patients
- Improve curation and standardization
Deconvoluting failed trials Necessary to follow up from failed drug trials that may overlook responding populations that are mutually exclusive Investment is rarely available to generate and mine data from failed trials - Use a retrospective approval route in which the responding population is shown to be distinct from the comparator or standard of care
Profiling of cell types from healthy individuals Projects are large and well-funded, for example, GTEx [96] and the Human Protein Atlas [97] - Public references may not capture interpatient (or disease-influenced) variability
- Findings are often assessed separately from efficacy data
- Improve integrative analyses across different types of patient data
- Agree on critical measures that should be assessed for the tumor microenvironment, drug toxicity, and patient comorbidities and wellness
Comprehensive profiling of tumor genetics and heterogeneity Projects are large, well-funded, and include tens of thousands of tumors, for example, projects by TCGA [98] and the ICGC [99] - Exploratory NGS is not routine for patients
- Public efforts are typically diagnostic, focus on the primary tumor, use purified tumor cell content, and have low sequencing depth
- Perform more and deeper spatial single-cell profiling of longitudinal and metastatic samples
- Increase multi-omic profiling of samples
Longitudinal and metastatic tumor genomic profiles Obtaining information about genetic shift after therapy could dramatically change our understanding of tumor drivers and heterogeneity - Currently only a limited amount of such data is available
- Ethical and practical considerations regarding the necessary invasive sampling procedures need to be taken into account
Continue to advance non-invasive monitoring approaches
Single-cell sequencing Provide unprecedented high-resolution information about genotypic and phenotypic heterogeneity of tumors and the tumor microenvironment, including information about cell differentiation and the effects of drugs - Current technologies require fresh tissue biopsies and obtaining these is often impractical
- Mature approaches are limited to RNA profiling
- Set up central repositories for single-cell omics data from patients and models
- Advance technology that enables exploration of molecules other than RNA and of non-fresh tissue samples
Germline genetic variation Provide information about a patient’s inherent immunological and metabolic competencies, susceptibility to adverse events, and other aspects of wellness Information is rarely available for patients with cancer as it is removed to avoid the risk of patients being identified and confidentiality being breached Add functional germline variant information to public databases of tumor genetics
Biosensors and smart wearables Enable real-time reactive adaptation of therapy to manage response, health, and adverse events A limited number of devices are currently available and few molecular measures are currently possible Develop technologies to identify the most important and relevant measures
Know- ledge Pathway and interaction networks Able to link drug target to biology - Typically focus only on intracellular pathways and interactions
- Typically limited to protein–protein interactions
Acquire more information about the cell-context specificity of interactions and about cell–cell communication
Regulomes Provide omics information that is indicative of active processes - Often miss cell-context specificity of regulomes
- Focus on intracellular processes
- Public resources focus on the transcriptome
- Obtain more data on cell-type-specific regulomes (as in, for example, ENCODE [100]) and on extracellular communication regulomes
  1. CCLE Cancer Cell Line Encyclopedia, CRISPR clustered regularly interspaced short palindromic repeats, ENCODE The Encyclopedia of DNA Elements, GDSC Genomics of Drug Sensitivity in Cancer, GEM genetically engineered mouse, GTEx Genotype-Tissue Expression Project, ICGC International Cancer Genome Consortium, LINCS Library of Network-Based Cellular Signatures, NGS next-generation sequencing, siRNA small interfering RNA, TCGA The Cancer Genome Atlas
  2. aKey pieces of preclinical and patient data that need to be generated, collected, and shared to achieve the computational ambition of modeling/predicting multimodal combination effects encompassing ECM, immune, angiogenic, and stromal components of tumor biology