Pancreatic cancer biology and genetics from an evolutionary perspective

Abstract

Cancer is an evolutionary disease, containing the hallmarks of an asexually reproducing unicellular organism subject to evolutionary paradigms. Pancreatic ductal adenocarcinoma (hereafter referred to as pancreatic cancer) is a particularly robust example of this phenomenon. Genomic features indicate that pancreatic cancer cells are selected for fitness advantages when encountering the geographic and resource-depleted constraints of the microenvironment. Phenotypic adaptations to these pressures help disseminated cells to survive in secondary sites, a major clinical problem for patients with this disease. In this Review we gather the wide-ranging aspects of pancreatic cancer research into a single concept rooted in Darwinian evolution, with the goal of identifying novel insights and opportunities for study.

Figure 1. Stages of pancreatic cancer evolution

a | Stage 1: initiation. A normal cell of the pancreas acquires an initiating driver gene mutation as a result of environmental exposure or a lapse in DNA repair. In most instances, this initiating mutation causes the cell to undergo apoptosis or senescence, or to be lost owing to immune surveillance or during a bottleneck event or tissue turnover (genetic drift). If these mechanisms fail, the cell carrying the initiating mutation (red) escapes from senescence and immunosuppression, and continues to fixation because of a survival or growth advantage. b | Stage 2: clonal expansion. The cell carrying the initiating mutation and its progeny continue to divide, creating a clonal population defined by the presence of the driver gene mutation. In the stepwise progression model, as the population grows in both cell number and geographic space, the descendants gradually acquire additional driver gene mutations (dark beige, blue and green cells in the left panel) and passenger mutations that increase clonal heterogeneity of the neoplasm. In the punctuated evolution model, a catastrophic genome-wide event occurs in a single cell cycle that results in widespread structural damage and acquisition of multiple driver gene alterations simultaneously (green cells in the right panel). c | Stage 3: introduction to foreign microenvironments. Ongoing clonal expansion may lead to a population of cells (green cells) that break through the basement membrane into the surrounding stroma. This event represents a genetic bottleneck that leads to a reduction in genetic diversity. Additional genetic events, signals provided by the stroma, deposition of dense extracellular matrix (ECM) and immune infiltrates all provide selective forces that shape the adaption of these cells into subclonal populations that differ with respect to their overall fitness (represented by cells coloured different shades of green). Dissemination is probably an ongoing process during tumour development; however, the extent to which cells from the entire neoplasm uniformly enter the circulation and/or whether dissemination is restricted to a subpopulation is unknown. Nonetheless, those disseminated cells that achieved high fitness in the primary site may have the greatest chance of colonizing new microenvironments, such as the liver, lung or peritoneum, common sites of metastasis in pancreatic cancer. Colonization of secondary sites represents yet another genetic bottleneck that may further reduce genetic heterogeneity. CDKN2A, cyclin-dependent kinase inhibitor 2A; MDSC, myeloid-derived suppressor cell; WT, wild type.

Figure 2. Geographic heterogeneity in pancreatic cancer

Geographic heterogeneity refers to the spatial variation within a single patient’s tumour with respect to genotypes and phenotypes. a | Spatial concordance. In spatial concordance, genetic subclonal populations and regions of distinct stromal biology, immune and/or metabolic phenotypes are geographically linked, as shown in the serial sections of a hypothetical primary tumour. This pattern would be consistent with genotypic heterogeneity driving phenotypic heterogeneity within a neoplasm. A representative example of a genotype that directly drives phenotypic features in pancreatic cancer are KRAS mutations that lead to immunological and metabolic alterations^,. In this scenario, targeting of subclones or the dominant subclone that drives tumour progression may be most efficacious. b | Spatial discordance. In spatial discordance, genotypic and phenotypic variations are unrelated to each other, implying that phenotypic variations in distinct regions of a tumour are unrelated to genotype and are more influenced by epigenomic or polygenic models of tumour behaviour. Unlike targeting of subclones, in this situation methods to modulate the epigenome to reduce cellular plasticity may have greater value. Spatial discordance has not been formally shown in human tumours because so far all global analyses have relied on single tumour samples, and thus it is of theoretical concern only until proved.

Figure 3. The three forms of intratumoural heterogeneity within a patient

a | Intratumoural heterogeneity within a primary tumour. The founder clone (indicated by the grey cell) is the ancestral cell population whose lineage contains all mutations acquired post-fertilization by the most recent common ancestor in the primary tumour. Thus, each mutation that was present in the founder cell is present in every descendant subclone and is inferred by the trunk of the phylogeny that contains n mutations. The founder cell itself no longer exists, as once it divides and accumulates a new mutation or mutations it has evolved. Subclone 1 (red cells) is composed of cells that have acquired mutation a. Subclone 2 (blue cells) and subclone 3 (beige cells) are also descendants of the founding cell that have acquired mutations d and b, respectively. Subclone 4 (green cells) has mutations b and c, indicating that it shares a common ancestor with subclone 3. b | Intratumoural heterogeneity within a metastasis. The metastasis-initiating cell (dark green cell) contains the initial, distinct set of mutations common to all cells of the metastasis (genotype n plus mutations b and c from panel a). The metastasis-initiating cell itself no longer exists, as once it divides and accumulates a new mutation or mutations it has evolved. Subclone 1 (light green cell) represents direct descendants of the metastasis-initiating cell that acquired mutation x, and subclone 2 (pink cell) represents the direct descendants that acquired mutation y. Subclone 3 (dark beige cell) has mutations y and z, indicating that it shares a common ancestor with subclone 2. c | Intratumoural heterogeneity of metastasis-initiating cells within a primary tumour. The metastasis-initiating cells share a common ancestor, yet, nonetheless, have distinct mutations that distinguish one from the other (that is, blue versus green genotypes). As each initiating cell is the ancestral cell for its respective metastasis, every descendant cell will inherit this founding set of mutations.