Clinical and Translational Applications
From research to clinical questions
While their impact is mainly on research, single-cell and spatial transcriptomics are starting to be applied directly to clinical samples, patient cohorts, and translational programs where the goal is answering specific questions about clincial response. The analytical and experimental requirements are substantially different from basic research applications.
Clinical sample types and their constraints
Clinical material is inherently more variable and more constrained than cell lines or mouse tissue. The sample type determines which experimental approaches are feasible.
Fine needle aspirates and core biopsies provide small amounts of material from accessible tumor sites. They are sufficient for single-cell RNA-seq but require efficient dissociation protocols and careful attention to cell viability. The small cell numbers rule out assays that require high input.
Surgical resection specimens provide larger material but introduce cold ischemia time between resection and processing. RNA quality degrades rapidly after tissue devascularization, and processing within two hours of resection is the standard target. When same-day processing is not feasible, tissue fixation at the time of collection followed by single-nucleus RNA-seq from fixed frozen material preserves data quality.
FFPE archival material from biobanks enables retrospective studies with long clinical follow-up but requires probe-based single-nucleus RNA-seq (snRNA-seq) rather than poly-A capture, and RNA quality varies considerably across blocks depending on fixation conditions and storage history. Despite these limitations, access to archived clinical material with linked outcomes data makes FFPE snRNA-seq one of the most powerful tools for translational biomarker discovery.
Peripheral blood provides a minimally invasive window into systemic immune responses. PBMCs isolated from blood are well-suited for single-cell profiling of circulating immune cell compositions and states, and serial sampling enables longitudinal monitoring of treatment response. TCR sequencing from blood can track clonal expansion of tumor-reactive T cells over the course of therapy. For many clinical programs, however, PBMCs are not the relevant tissue and cells, and are limited in clinical information.
Experimental design for clinical studies
Clinical single-cell studies require careful experimental design before the first sample is collected. Key considerations that differ from standard research applications include:
Sample size and statistical power. Unlike cell line experiments where technical replicates are straightforward, clinical studies are powered by the number of patients, not the number of cells. The pseudobulk DE framework treats each patient as one observation, so studies with fewer than 6 patients per group have limited statistical power regardless of how many cells are profiled per sample.
Matched versus unmatched designs. Longitudinal studies comparing pre- and post-treatment samples from the same patients are more powerful than cross-sectional comparisons and should be the default design when treatment response is the primary question. Sample collection timing relative to treatment administration must be specified in advance and adhered to consistently.
Confounding variables. Patient age, sex, prior treatment history, tissue site, and sample processing site all introduce variation that can confound biological signals. Experimental designs that randomize these factors across conditions, or that collect sufficient metadata to model them as covariates, are essential for interpretable results.
Cell type rarity and sampling depth. If the analysis requires characterizing a rare cell population, such as circulating tumor cells or antigen-specific T cells, the required sequencing depth and cell numbers must be estimated before study initiation rather than discovered after the data is collected.
Patient stratification and endotype discovery
One of the most powerful applications of single-cell transcriptomics in translational research is the unsupervised discovery of patient subgroups, endotypes, defined by their cellular and molecular biology rather than by clinical criteria alone. Diseases that appear homogeneous by clinical presentation often contain multiple distinct endotypes with different underlying mechanisms, different prognoses, and different responses to therapy.
In oncology, endotype discovery from single-cell data has identified TME configurations associated with response or resistance to checkpoint blockade, subclonal structures associated with metastatic potential, and transcriptional states in malignant cells associated with drug sensitivity or resistance. In autoimmune and inflammatory diseases, single-cell profiling of target tissues and blood has revealed patient subgroups with distinct immune cell compositions and activation states that predict disease course and treatment response.
Translating endotype definitions into clinically usable classifiers requires training machine learning models on single-cell data from one patient cohort and validating them in independent cohorts. This remains technically challenging due to batch effects across collection sites, the high dimensionality of single-cell data, and the need for prospective validation before clinical deployment.
Biomarker discovery and development
Single-cell transcriptomics identifies candidate biomarkers at the level of cell types, cell states, and gene expression programs rather than single analytes. Converting these candidates into actionable biomarkers, measurements that can be made reproducibly at clinical scale, requires a translation step that is often underestimated.
The most tractable path is identifying single-cell-defined cell type or state signatures that can be detected by simpler, cheaper, and more scalable methods: flow cytometry, immunohistochemistry, or targeted gene expression panels. The single-cell data provides the discovery layer; the validation assay provides the clinical deployment layer. 3DG supports the discovery layer and can advise on which findings are most likely to be translatable to downstream assay formats.
Spatial biomarkers, or features of tissue organization such as immune infiltration pattern, TLS density, or tumor-immune boundary topology, are increasingly recognized as clinically relevant and are uniquely accessible through spatial transcriptomics. These spatial features are in principle measurable by digital pathology methods, making spatial transcriptomics data a productive source of biomarker hypotheses for downstream validation.
Drug mechanism and target validation
Single-cell transcriptomics is increasingly used to characterize drug mechanism of action at cellular resolution. By comparing TME composition and cell state distributions before and after treatment, or in treated versus untreated patient-derived samples, it is possible to determine which cell types are directly affected by a drug, what transcriptional changes are induced, whether off-target effects are detectable at the cellular level, and how resistant cell populations emerge under drug pressure.
For early-stage programs, ex vivo treatment of patient-derived material such as tissue slices, organoids, or disaggregated primary cells, combined with single-cell readout provides mechanistic data that informs dose selection and combination strategies before clinical trials. 3DG has experience designing and executing these studies for pharmaceutical clients across oncology, dermatology, and immunology.
Data governance and compliance
Clinical single-cell studies involving human samples require careful attention to regulatory and ethical requirements. All studies involving human tissue at 3DG are conducted under appropriate material transfer agreements, in compliance with IRB or ethics board approvals held by the providing institution, and with full de-identification of patient-identifying information before data transfer. Data handling, storage, and transfer practices comply with applicable privacy regulations.