Pancreatic tumours are among the most lethal of human cancers and cause ∼30,000 deaths each year in the United States alone. New progress in understanding the genetic and cellular origins of this type of cancer has now been made in two studies using transgenic mouse models. In one paper, Ronald DePinho, Nabeel Bardeesy and colleagues describe cooperative roles for lesions at the Kras and Cdkn2a loci in the development of pancreatic ductal adenocarcinoma, and in another, work from Harold Varmus's laboratory indicates that different oncogenic mutations give rise to distinct types of pancreatic tumour.

Although several genetic lesions are associated with pancreatic ductal adenocarcinoma, their contributions to disease progression are poorly understood. Activating mutations in KRAS are found in most cases of this type of cancer, as are loss-of-function mutations in CDKN2A — which encodes the tumour suppressors INK4A and ARF. To investigate the relative roles of these genes in tumour progression, DePinho, Bardeesy and colleagues generated mice that combined activation of an oncogenic Kras transgene and deletion of Cdkn2a specifically in the pancreas. All of the animals developed invasive tumours, which were very similar in terms of their histology and metastatic behaviour to human pancreatic ductal adenocarcinoma. In mice that carried only the activated mutant Kras transgene — with an intact Cdkn2a locus — pre-malignant lesions developed, but did not become invasive. No pre-malignant lesions or invasive tumours were seen in mice with the Cdkn2a deletion only. So, neither Kras nor Cdkn2a mutations are sufficient to give rise to pancreatic ductal adenocarcinoma. Instead, it seems that Kras activation is required to initiate tumorigenesis, but additional homozygous mutation of Cdkn2a is essential for progression to advanced malignancy.

But are mutations in Kras and Cdkn2a sufficient for full progression to malignancy? As invasive tumours developed rapidly without the known disruption of other loci, activated Kras in combination with loss of function of Ink4a and/or Arf might be sufficient for the development of metastatic tumours. Alternatively, other mutations could be required, but might occur at such a high frequency in this model that they do not limit the rate of progression to invasiveness. This possibility will need to be tested by genomic analysis of the tumours in these mice.

In a second study by Harold Varmus and colleagues, another mouse model was used to investigate how different types of pancreatic tumour arise. The pancreas contains acinar, ductal and endocrine cells, and different tumours develop that resemble each of these cell types. Whether these tumours arise from different cells or result from distinct effects of different genetic lesions in the same cells is unknown.

Varmus and co-workers used a retrovirus-based method to express oncogenes in the developing pancreas in a Cdkn2a-null background. Expression of the polyoma virus middle T antigen gene — which stimulates the Ras and Pi3k pathways that are activated in pancreatic cancers — led to the formation of ductal and acinar tumours. By contrast, expression of the human c-MYC oncogene gave rise exclusively to endocrine tumours, indicating that specific oncogenes induce the formation of different types of pancreatic tumour.

Whether different oncogenes target distinct groups of cells to produce the various pancreatic tumour types or exert their effects in the same set of cells, which are then guided towards particular fates, remains to be determined.