A key component of the pancreatic tumour microenvironment (TME) is its dense and fibrotic stroma. This ecosystem surrounding the tumour is a characteristic feature of the disease and can make up over 80% of the tumour mass. Indeed, many describe PDAC as “the poster child of the tumour microenvironment”. Cancer-associated fibroblasts (CAFs) are key cellular players within this stroma and characterising their function, heterogeneity, and spatial relationships with tumour epithelium have been key priorities for PDAC biologists in the past decade.

In spite of these efforts, one aspect of CAF biology which remains less understood is their shape.

The shape of things to come

Morphology has been a defining feature of fibroblasts since their description as spindle-shaped cells by Virchow in 1858. So, the question naturally arises; if the shape of a CAF reflects its underlying function, could morphology be used as a proxy for biology?

To explore this, my lab recently profiled 201 resected PDAC specimens using multiplex immunofluorescence (mIF). In addition to assessing CAF morphology, we also wanted to explore how this morphology relates to wider tumour architecture.

Initially the main task was to segment out the cells into epithelium and FAP (Fibroblast activation protein) -positive and -negative stromal fractions. This allowed an exploration of how epithelial and stromal compartments of the TME spatially relate to each other, and how these spatial relationships aggregate into broader architectural patterns. This revealed a number of local cellular neighbourhoods, which were then used to define architectural subtypes of PDAC. Interestingly, these were strongly correlated with outcome. Prolonged survival is seen among tumours with cohesive, organised architecture enriched for spatial interactions between epithelium and FAP-negative stroma. Conversely, tumours with disorganised, fragmented architecture were enriched for spatial interactions between epithelium and FAP-positive stroma and demonstrated much shorter survival.

These architectural subtypes were subsequently named according to their key features: Cohesive, Organised, Desmoplastic and Fragmented.

Our results also suggest that CAF morphology encodes biology, and highlights CAF morphological states associated with adverse prognosis. Profiling the morphology of a cell with ‘low-plex’ methods could therefore be a more accessible means of inferring its underlying phenotype, compared to more costly or time-intensive sequencing or highly multiplexed techniques.

More broadly however, these results could support a model of epithelial-stromal co-evolution in which both tumour and stroma converge towards an adverse state, with corresponding changes in CAF morphology, spatial architecture, and clinical outcome. Although further mechanistic studies are required to explore this hypothesis, it raises an important therapeutic question. If epithelium and stroma co-evolve in tandem, is this process bidirectional and therefore potentially targetable? In other words, could reversing these adverse morphological and architectural processes in turn reverse adverse tumour biology?

Unlocking these secrets could provide further opportunities for more effective treatment to improve outcomes for patients with pancreatic cancer.