Seeing pancreatic cancer one cell at a time

The RAS Dialogue Blog posts are written by RAS experts sharing the latest research, updates, and scientific RAS news. The content is curated by the RAS Initiative.

Image

Pancreatic ductal adenocarcinoma (PDAC) is often described as a “KRAS-driven” disease. Activating mutations in KRAS are present in the >90% cases and are widely regarded as early, foundational events in tumorigenesis. However, it has been recognized that (1) many tumors become less dependent on their initiating mutations over time [1, 2]; (2) in preclinical studies, PDAC showed less addiction to KRAS through various bypass mechanisms [3, 4].

As RAS-targeting therapies become more widely-used in PDAC, responses are heterogeneous, likely induced by the tumors’ genetic adaptation. It becomes imperative to understand at high-resolution how the genomes of KRAS-mutant PDACs evolve.

In our recent Nature Genetics study, “Genomic evolution of pancreatic cancer at single-cell resolution,” [5] we sought to address this question by moving beyond bulk sequencing and into the cellular scale.

By profiling genomic alterations in 137,491 individual tumor cells from 24 pancreatic cancer patients without RAS-targeting therapies, we aimed to reconstruct evolutionary trajectories with unprecedented resolution — and, in doing so, better understand how PDACs clonally evolve in their natural history, prior to therapeutic selection.

From bulk averages to cellular evolution

Traditional genomic analyses of PDAC have relied largely on bulk tumor sequencing, which aggregates signals across millions of cells. While these approaches have been invaluable for identifying recurrent driver alterations such as KRAS, TP53, CDKN2A, and SMAD4, they inherently obscure intratumoral heterogeneity. 

In other words, they tell us what mutations are present, but not how those mutations are distributed across individual clones or how different subpopulations evolve under selective pressure.

This distinction is particularly important for RAS-driven cancers. If some PDACs indeed develop reduced functional dependence on KRAS signaling over time, we would expect not necessarily a loss of KRAS mutations themselves, but rather diversification of genomic backgrounds that reshape pathway output, fitness, and clonal competitiveness. Single-cell DNA sequencing provides the necessary resolution to directly observe how KRAS-mutant lineages expand, diversify, and coexist within the same tumor ecosystem.

KRAS as a clonal driver — but not a uniform dependency

Consistent with prior models of pancreatic tumorigenesis, our single-cell data confirmed that KRAS mutations are overwhelmingly clonal events. The vast majority of malignant cells within each tumor shared the same KRAS-mutant genotype, reinforcing the concept that oncogenic KRAS activation is an early initiating event in PDAC evolution.

However, a key insight from our study is that early occurrence of KRAS mutations does not equate to uniform KRAS dependence later in tumor progression.

At single-cell resolution, we observed that oncogenic KRAS dosage is not constant across tumors or even across subclones within the same tumor. Many tumors exhibited allelic imbalance favoring the mutant KRAS allele, while a substantial subset of cases contained minor populations of cells where the mutant KRAS allele appeared reduced or lost relative to the dominant clone. Rather than a binary KRAS-mutant versus KRAS-wildtype state, our data support a spectrum of KRAS genomic configurations within PDAC.

This finding reframes how we interpret “KRAS-driven” disease: the driver mutation may have initiated the tumor, yet its effective dosage—and in turn, its functional presence—can vary across evolutionary lineages.

Image

Convergent activation of the MAPK–ERK axis beyond KRAS

Another key RAS-relevant insight from our study is the observation that some PDACs acquire additional alterations that can converge on MAPK–ERK pathway activation downstream or parallel to KRAS. We identified cases harboring clonal or subclonal mutations in genes such as PIK3CA or MAP2K4 that coexisted with KRAS-mutant lineages.

This pattern suggests a model of evolutionary redundancy: Once oncogenic KRAS establishes a malignant lineage, subsequent genomic evolution may reinforce or diversify MAPK pathway activation through alternative mechanisms. In such contexts, tumor cells may remain genetically KRAS-mutant yet become less exclusively dependent on KRAS signaling itself, as pathway output is buffered by additional oncogenic alterations.

An alternative path to PDAC tumorigenesis without canonical KRAS mutation

Image

What happens in the <10% PDACs without KRAS mutation? Our single-cell analyses suggest  that some of these tumors harbored germline BRCA2 alterations, and followed vastly different evolutionary paths. In some of these cases, biallelic loss of BRCA2 occurred early and was associated with widespread genomic instability, indicating that genome remodeling itself may act as a key early driver rather than canonical KRAS mutation.

Yet even in these alternative trajectories, tumors still converge on activating growth signaling, not necessarily through KRAS mutation but other mechanisms, such as FGFR mutation or BRAF and NRG fusions. This suggests that pancreatic cancer is less defined by a single initiating mutation than by a shared evolutionary objective: sustained activation of core growth-promoting pathways. Thus in these cases, targeting KRAS would likely be inefficacious.

A refined model of RAS-driven tumor evolution and its clinical implications

Taken together, our single-cell analyses support a refined view of KRAS in pancreatic cancer. In most cases, KRAS mutations function as stable initiating events that anchor tumorigenesis. 

However, subsequent evolution generates a diverse landscape of subclones with variable KRAS dosage, distinct genomic architectures, and potentially differing degrees of dependence on relevant pathways. In some tumors without canonical KRAS mutation, genome instability likely initiates the tumors which later develop alternative alterations that converge on similar growth-promoting pathways.

By resolving tumor evolution at single-cell resolution in untreated disease, our study provides a baseline map of how PDAC lineages diversify naturally—offering a framework for interpreting therapeutic response, resistance, and the future design of combination strategies in RAS-driven cancers. Understanding this heterogeneity of KRAS dependence will be essential as RAS-targeted therapies advance in pancreatic cancer. 

If you're interested in participating in the dialogue, consider joining RASLab. RAS Lab is an online discussion forum to facilitate the exchange of scientific information among researchers in cancers related to RAS. If you would like to join RAS Lab, send an email to SolveRAS@nih.gov with the subject line "I would like to join RAS Lab."