top of page
  • Writer's pictureDr. Robert A. Nagourney, MD


Pancreatic cancer is among the most lethal of all human malignancies, with 60,430 new diagnoses and 48,220 deaths in 2021, it is the 3rd leading cause of cancer death in the US. Most pancreatic cancer patients are diagnosed with advanced disease for whom the five-year survival is only 3%; a survival rate that has changed little in several decades

Genetic profiles have shown that over 90% of pancreatic cancers carry mutations in the KRAS gene and over 50% have mutations in the TP53 gene.

To learn more about pancreatic cancer, researchers developed a mouse model that expresses the KRAS and P53 genes in genetically engineered KPC mice. KPC mice have been used around the world in hundreds of studies but when scientists took their mouse data and put it into people the results were profoundly disappointing.

One study used KPC mouse data to design a trial that targeted the hedgehog pathway (SHH), a pathway that is essential for human embryonic development and is expressed in pancreatic cancer. The drug Vismedogib clearly inhibited the SHH in KPC mice so Vismedogib was selected as a treatment.

Although in the KPC mouse model the drug Vismedogib performed beautifully, when it was given with Gemcitabine in a study of 104 patients there was no benefit. Vismedogib proved to be a dud.

Where did this trial go wrong?

First, it used a highly artificial mouse model that generated a conditional expression by manipulating the mouse genome. This could not re-create the complexity of human cancer that arises spontaneously in humans. Despite good mouse data, this treatment did not translate into clinical benefit for patients.

The second scientific misadventure occurred when investigators decided to attack the dense fibrous stroma that surrounds pancreatic cancer cells. It was reasoned that this dense protein matrix prevented drugs from penetrating into the center of the pancreas tumors, rendering them drug-resistant.

The HALO-301 clinical trial accrued 494 patients with advanced pancreatic cancer who were well-balanced for clinical characteristics. One half got chemotherapy alone and the other half received hyaluronidase combined with the same chemotherapy.

The results revealed no benefit whatsoever. In fact, the median overall survival was actually numerically longer for the patient that received placebo (11.5 months), over the people who received hyaluronidase with chemotherapy (11.2 months) (Van Cutsem E J Clin Oncology, 2020). Another dud.

What went wrong?

Among the most obvious possibilities is that the KPC mouse model isn’t representative of the human cancer microenvironment. It is unclear if the cancer-associated fibroblasts in mice are the same in mice and humans.

Another concern is that the stroma may serve as a barrier to tumor progression, such that removing the barrier allowed tumors to grow and invade. More importantly, the reductionist thinking that focused exclusively on a single feature of the cancer, fibrosis failed to take into consideration low mutational burden, immune dysregulation, metabolic changes, and the epithelial to mesenchymal transition (EMT), so characteristic of tumors in their natural microenvironment. The HALO-301 is an example of a good idea going bad.

So where has the science gone wrong?

We have relied upon artificial models like the KPC mouse with mouse stroma, a mouse immune system, and the highly artificial expression of two specific genes in a biological vacuum.

It is well-known that cancer is polygenic and that numerous contributing factors lead to the ultimate tumor phenotype.

Alexander Pope (1688-1744) said, “The proper study of mankind is man” To wit, the proper study of human cancer is human cancer. We know that human tumors are complex ecologic systems that provide interaction between stroma, tumor cell to tumor cell communication, interactions with vasculature, inflammatory cells, inflammatory cytokines, and cancer-associated fibroblasts that cannot be easily reproduced in an animal model.

Our laboratory pioneered the use of pancreatic cancer studies using portions of tumor removed directly from patients. We do not destruct the integrity of the microenvironment, but instead utilize that micro-environmental integrity, to study the response to drugs, targeted agents, and other interactions.

We have previously shown that this improves response and time to progression in a variety of cancers and have achieved durable, complete remissions in patients with advanced pancreatic cancer. Each patient we treat receives their own specific therapy crafted from their own tumor biology and not protocol or guideline-driven therapy.

For patients diagnosed with advanced pancreatic cancer, it is critical to receive the right treatment the first time.

Like the KPC mouse model that engineers a target and then tries to hit it, today’s oncologists rely upon gene profiles to select actionable targets and then try to hit them. I am not sure whether they need better aim or better bullets, but it is obvious that it is not working.

To re-create the complexity of human tumors one needs to study human tumors. At the Nagourney Cancer Institute that is exactly what we do.


bottom of page