Model Gauges Risk for Pancreatic Cancer at Onset of Diabetes

Nick Mulcahy

July 06, 2018

A tiny percentage of new-onset diabetes cases are due to pancreatic cancer, but who are these patients? A new model based on three easily accessed patient characteristics provides a way of zooming in on this small group by stratifying the risk for this cancer in the larger group of patients, according to American researchers.

The predictive model, known as END-PAC (Enriching New-Onset Diabetes for Pancreatic Ductal Adenocarcinoma), is based on change in weight, change in blood glucose, and age at onset of diabetes (as determined by glycemia). It was developed and validated by a multi-institutional team from the United States. The researchers describe the model in an article published online May 15 in Gastroenterology.

The new model "readily risk stratifies subjects with glycemically-defined new-onset diabetes, clearly defining those who warrant work-up for pancreatic cancer," write Suresh Chari, MD, professor of medicine at the Mayo Clinic, Rochester, Minnesota, and colleagues.

Diabetes patients who are identified as being at high risk via the model's score would be candidates for workup. "This will greatly help in targeting a small group of individuals with new-onset diabetes for early detection," Chari told Medscape Medical News.

The model needs independent, prospective validation before it can be put into practice, the authors say.

The team explains that the vast majority of patients (about 99%) who develop diabetes after age 50 have type 2 diabetes. However, about 1% of cases of new-onset diabetes are induced by pancreatic cancer.

Currently, there is no enriched method for selecting and testing diabetes patients who are at an elevated risk for pancreatic cancer. Detecting pancreatic cancer at a potentially earlier stage is highly desirable, because most of these cancers (85%) are diagnosed at an advanced stage and are associated with a very poor prognosis.

Being able to screen some diabetes patients for pancreatic cancer would be a significant development in screening for the malignancy, suggested Chari.

"So far, the only folks who have been targeted for screening for pancreatic cancer are those who have a strong family history of pancreatic cancer. It has been felt there are no high-risk groups for nonfamilial pancreatic cancer," he said.

In 2005, Chari and other Mayo Clinic researchers showed, for the first time, that roughly 1% of patients with new-onset diabetes are diagnosed with pancreatic cancer within 3 years.

Since then, other investigators have also shown that recent-onset diabetes is a risk factor for pancreatic cancer, including among minorities, as reported by Medscape Medical News.

Model Performance

In their new article, Chari and colleagues describe a cohort of 1561 patients older than 50 years who met diagnostic criteria for new-onset diabetes. Of those patients, 16 (1.0%) developed pancreatic cancer within 3 years of meeting those criteria.

In this discovery cohort of new-onset diabetes patients, the END-PAC model retrospectively identified those patients who developed pancreatic cancer within 3 years of onset of diabetes as having an area under the receiver operating characteristic curve value of 0.87. A score >3 identified patients who developed pancreatic cancer with 80% sensitivity and specificity.

In the validation cohort of diabetes patients (n = 1096), a score >3 retrospetively identified seven of the nine patients with pancreatic cancer (78%), with 85% specificity.

The prevalence of pancreatic cancer in patients with score >3 (3.6% of the cohort) was 4.4-fold higher than in patients with new-onset diabetes.

On the other hand, an END-PAC score <0 (49% of patients) in the validation cohort meant that patients had an "extremely low-risk" for pancreatic cancer, said the Mayo Clinic team.

The END-PAC score classifies new-onset diabetes patients into high-, intermediate-, and low-risk groups for pancreatic cancer.

The high-risk group (score ≥3) would be candidates for clinical workup for pancreatic cancer, the researchers comment. The 3-year incidence of pancreatic cancer in patients with these high-risk scores was more than triple that of the lower-risk groups (3.6% vs 1%) in the validation cohort.

Features That Suggest Pancreatic Cancer

The authors observe that the END-PAC model is based on three previously noted features that distinguish new-onset type 2 diabetes (T2-NOD) from new-onset diabetes in pancreatic cancer (PC-NOD).

"While T2-NOD is often accompanied by weight gain, PC-NOD paradoxically occurs in the face of weight loss," the authors write.

"Progression from normal fasting glucose to T2-NOD is a slow process occurring over ~8 years, while PC-NOD progresses rapidly, over 2-3 years," they also say.

The investigators observe that patients with T2-NOD are younger at diabetes diagnosis than patients with PC-NOD; the mean age at T2-NOD diagnosis (~52 years) is much lower than the mean age at PC-NOD diagnosis (~71 years).

Referring to other research, Mayo Clinic investigators say that about 20% of pancreatic cancers meet criteria for new-onset diabetes. Of that group, about 70% could potentially be diagnosed more than 3 months prior to clinical diagnosis if the END-PAC score was applied at the onset of diabetes. The team believes that this earlier detection would "lead to improved resectability and consequent better survival."

Currently, the average survival of patients with newly diagnosed pancreatic cancer is 6 months.

The study was supported by the NIH Consortium for Study of Chronic

Pancreatitis, Diabetes and Pancreatic Cancer, the Pancreatic Cancer

Action Network, the Kenner Family Research Fund, and the Prokopanko Gift to Mayo Foundation. The authors have disclosed no relevant financial relationships.

Gastroenterol. Published online May 15, 2018. Abstract

Follow Medscape senior journalist Nick Mulcahy on Twitter: @MulcahyNick

For more from Medscape Oncology, follow us on Twitter: @MedscapeOnc


Comments on Medscape are moderated and should be professional in tone and on topic. You must declare any conflicts of interest related to your comments and responses. Please see our Commenting Guide for further information. We reserve the right to remove posts at our sole discretion.