Treating Precancerous Pancreatic Lesions With KRAS Inhibitors Nearly Doubled Survival in Mice

Pancreatic cancer is diagnosed late, treated with limited success, and kills most patients within a year. The five-year survival rate hovers around 12%. For decades, researchers have tried to improve outcomes by developing better treatments for established tumors. A study published in Science asks a different question: what if you treat the cancer before it becomes cancer?

Interception versus prevention

Cancer interception is distinct from cancer prevention. Prevention aims to stop cancer from ever initiating - think HPV vaccines or smoking cessation. Interception targets cells that have already started down the path toward malignancy but have not yet become tumors. The closest existing example is colonoscopy, where precancerous polyps are physically removed before they can develop into colorectal cancer.

For pancreatic cancer, no equivalent exists. There is no screening test, no established prevention strategy, and by the time symptoms appear, the disease is usually advanced. But the biology offers a potential opening.

PanINs: the precancerous starting point

Most pancreatic ductal adenocarcinomas (PDAC) - the most common form of pancreatic cancer - arise from microscopic lesions called PanINs (pancreatic intraepithelial neoplasias). PanINs are common in adult pancreases and nearly all carry mutations in the KRAS gene, the most frequently mutated cancer-causing gene across all cancer types. But only a tiny minority of PanINs ever progress to cancer. What triggers that rare malignant switch remains poorly understood.

PanINs are also invisible on standard imaging. They are too small to detect with CT scans or MRI. This makes them impossible to monitor and, until now, impossible to target therapeutically.

The experiment: KRAS inhibitors before versus after cancer

The Penn Medicine team, led by co-corresponding authors Robert Vonderheide, MD, DPhil, and Ben Stanger, MD, PhD, used two experimental KRAS inhibitors developed by Revolution Medicines. RMC-9945 selectively targets the KRAS G12D mutation (the most common form in pancreatic cancer), while RMC-7977 targets multiple active RAS variants.

Using a Penn-developed immunocompetent mouse model - considered the gold standard for preclinical PDAC research - the team first established baseline timelines of PanIN-to-PDAC progression in untreated controls. They then treated an intervention group with either inhibitor after PanINs had developed but before tumors appeared.

After 10 days of treatment, precancerous lesions were visibly reduced. At 28 days, the reduction was more pronounced. Tumors were slower to develop, and survival increased compared with untreated mice. Long-term treatment with RMC-7977 tripled median overall survival compared with untreated controls. Most strikingly, mice that received treatment before tumor development survived nearly twice as long as mice treated only after cancer had been diagnosed.

From mice to the right human population

The team plans to translate this work into a clinical trial, but the path is not straightforward. PanINs cannot be seen on imaging exams, and the proposed treatment would be given to people who do not have cancer. That raises the bar for safety and requires careful patient selection.

The initial clinical focus will be on high-risk patients already being monitored for pancreatic cysts - larger growths than PanINs with a low but real cancer risk that are sometimes surgically removed. Beyond cyst surveillance, the approach could eventually apply to individuals with genetic predispositions including BRCA1, BRCA2, or PALB2 mutations, hereditary pancreatitis, or other strong risk factors.

Mouse models and human realities

The study's limitations are significant and worth stating plainly. This is preclinical work in mice. Mouse models of pancreatic cancer, even immunocompetent ones, do not fully replicate human disease biology, drug metabolism, or immune responses. The KRAS inhibitors used are preclinical tool compounds, not the exact drugs that would be tested in patients (though they represent the same drug classes as investigational compounds currently in clinical trials).

The approach treats all PanINs as potential threats, even though the vast majority will never become cancer. Whether the benefits of eliminating benign lesions alongside the rare dangerous ones outweigh the risks and costs of treatment in healthy individuals is an open question that only human trials can answer.

There is also a detection problem. Without the ability to see PanINs on imaging, there is no way to confirm treatment is working in patients short of waiting years to see whether cancer develops. Developing biomarkers or imaging techniques that can track precancerous pancreatic changes will be critical for clinical translation.

Still, the core proof-of-concept is clear: in mice, treating the precancerous state works better than treating the cancer. For a disease with so few treatment options, that principle - even in its preclinical form - is worth pursuing.