Solid tumors are not genetically uniform. A biopsy from a single tumor lesion will contain multiple genetically distinct subclones, each with its own alteration profile, each occupying a different niche within the tumor microenvironment, each contributing to the tumor's behavior in potentially different ways. This is the reality of intratumoral heterogeneity, and it has direct consequences for how we think about target identification.
The challenge is not that heterogeneity makes targets impossible to find. It is that heterogeneity changes what a valid target looks like and raises the bar for what kind of evidence you need before committing resources to a candidate. Here is how we think about it.
Clonal Architecture and Target Confidence
Target nominations derived from somatic alteration data require an additional interpretive step when heterogeneity is accounted for: is the alteration present in the dominant clonal population, or is it a subclonal event present in a fraction of the tumor? A dominant clonal alteration, one that is present in the vast majority of cells and is therefore likely an early event in tumor evolution, represents a different kind of target than a subclonal alteration present in only a minority of cells.
The distinction matters for several reasons. A dominant clonal alteration that is functional in driving tumor growth is present in essentially all cells that need to be eliminated for a durable response. Inhibiting it affects the whole tumor. A subclonal alteration, even one that is functional in the subpopulation that carries it, will leave the majority of the tumor untouched if it is the only target. This consideration does not disqualify subclonal targets, but it substantially changes the strategic logic for pursuing them.
Much of the large-scale tumor genomic data that feeds into target nomination does not resolve clonal architecture at the level of individual samples. Bulk sequencing provides variant allele frequencies that are population-level averages across all cells in the sequenced sample. High variant allele frequency is consistent with dominant clonal presence, but is not definitive evidence for it when stromal and immune cell contamination varies across samples. Target nominations that rest primarily on bulk sequencing alteration frequency need to be contextualized by what is known about clonal architecture in the relevant tumor type.
Phenotypic Heterogeneity and Functional Evidence
Beyond genetic heterogeneity, solid tumors express phenotypic diversity across cell populations that shares some features with developmental hierarchies. Cells at different stages of differentiation, cells in hypoxic versus well-oxygenated regions, and cells in direct contact with stromal components versus cells deeper in the tumor mass can show substantially different behavior and gene expression profiles. This phenotypic diversity compounds the genetic heterogeneity in ways that matter for target identification.
The practical implication is that functional evidence obtained in a single cell-line system may not represent the full range of biological contexts in which the target will need to operate. A kinase inhibitor that robustly kills a well-characterized cell line carrying the target alteration may fail to address cells in the hypoxic tumor core, or may fail to kill a rare subpopulation that has adopted a mesenchymal phenotype and upregulated compensatory survival pathways. This is a fundamental limitation of standard in-vitro validation models that the field has long recognized without fully solving.
Higher-quality functional evidence in the heterogeneity context comes from studies that account for multiple tumor cell states: in-vivo studies in patient-derived models that better preserve the original tumor's cellular diversity, multi-cell-type co-culture systems that include stromal interactions, or single-cell analysis of target expression and dependency across the tumor cell population. These study types are more resource-intensive, which is why they appear less frequently in the literature. But their presence in the evidence base for a target is a meaningful quality indicator.
Population-Level Heterogeneity: The Patient Stratification Problem
Intratumoral heterogeneity is one dimension. Inter-tumor heterogeneity, the genetic and phenotypic diversity across different patients' tumors of the same histologic type, is equally important for target identification.
A target that is highly relevant in a genomically-defined subgroup of a tumor type may be nominated from a broadly-typed tumor study without adequate consideration of whether it is relevant across the population or only within a specific molecular subtype. Targets that look like population-level signals in bulk analysis can sometimes be subtype-specific signals that require patient stratification to be useful therapeutically.
Evidence synthesis that does not account for this conflation will score a target higher than its true population-level relevance warrants. Conversely, a target that looks modest at the population level may have very strong evidence in a specific molecular subtype, making it an excellent candidate for a stratified development program. The evidence needs to be read with subtype context in mind.
How the Avenzo Evidence Framework Addresses Heterogeneity
When we built the evidence classification and scoring model, heterogeneity-relevant evidence types were built in as considerations, not afterthoughts. Specifically, the quality tier for functional evidence weights in-vivo and multi-cellular-system evidence higher than single-cell-line evidence, partly because of its greater relevance to the heterogeneous in-vivo tumor environment. Evidence from patient-derived xenograft models, which preserve more of the original tumor's cellular and genetic diversity than standard cell-line models, is classified separately within the tier structure.
For the clinical context evidence dimension, we flag evidence that is subtype-specific or population-stratified as distinct from evidence that comes from unselected patient populations. A target with strong evidence in a molecularly-defined subtype may score lower on total population-level clinical context signal, but the subtype-specific signal is recorded and visible in the output. We do not want to penalize subtype-specific targets by comparing them to targets with broad population-level signals.
We are not claiming this fully solves the heterogeneity problem in target identification. The problem is biological and partially irreducible. What the evidence framework does is make visible where the evidence is and is not responsive to heterogeneity considerations. A target with functional evidence from only a single homogeneous cell-line system scores differently than one with multi-model functional evidence. That distinction carries information the discovery team can use.
The Practical Takeaway for Target Selection
Tumor heterogeneity should raise the evidence quality bar for target advancement, not just add complexity to the selection rationale. A target with strong evidence from diverse experimental systems, including contexts that capture more of the tumor's cellular diversity, is more likely to hold up across the patient population than one with equivalent total evidence concentrated in a single, narrow experimental context.
This does not mean that targets with limited evidence breadth cannot be pursued. The science may simply not have gotten to them yet, and a thin evidence base in diverse experimental contexts can be an opportunity for a discovery program to do work that establishes confidence. But it does mean that the evidence profile should be transparent before commitment, not discovered after assay resources are allocated.
If your program is working through target selection in a tumor type with known high heterogeneity, and you want to understand how the evidence landscape for your candidates looks when heterogeneity-relevant evidence types are explicitly classified and weighted, reach out to discuss what the Avenzo evidence analysis would look like for your list.