New-onset diabetes provides ‘window of opportunity’ for early pancreatic cancer detection
Pancreatic cancer often is called the silent killer.
Most people with early disease show no symptoms. Consequently, 80% of cases are diagnosed in late stages when surgery is no longer possible and other treatment options are limited. Fewer than 10% of patients survive 5 years after diagnosis.
Early detection has been heralded as the holy grail to reduce mortality of pancreatic cancer, set to become the second leading cause of cancer death in the United States by 2040.
A deeper understanding of the link between new-onset diabetes and pancreatic cancer may play a key role in that effort.
One study showed individuals with new-onset diabetes are up to eight times more likely than the general population to develop pancreatic cancer. Another suggested the increased risk for pancreatic cancer among people with new-onset diabetes is comparable to the elevated risk for lung cancer observed among those who smoked a pack of cigarettes per day for 2 decades.
A history of pancreatitis also may be a key indicator, with a study showing patients with diabetes that developed secondary to a pancreatitis attack had a seven times higher risk for pancreatic cancer than individuals with type 2 diabetes.
Two major research collaborations are underway to explore the link between pancreatic cancer and new-onset diabetes.
The NCI is leading a multidisciplinary effort to determine which individuals newly diagnosed with diabetes are at elevated risk for pancreatic cancer. A second study will examine whether imaging at the time of new-onset diabetes results in earlier pancreatic cancer detection.
This potentially practice-informing research is crucial to improve the typically poor prognosis for patients with this aggressive malignancy, according to Brian M. Wolpin, MD, MPH, director of the gastrointestinal cancer center and Hale Family Center for Pancreatic Cancer Research at Dana-Farber Cancer Institute.
“The idea that there may be people out there who have some signal [to show] the cancer may be present but we aren’t identifying them in a way that [allows us] to diagnosis the cancer early is upsetting,” Wolpin told HemOnc Today. “It’s frustrating, but it’s also the challenge we want to tackle, so that we find pancreatic cancer earlier when it can be cured.”
HemOnc Today spoke with oncologists and endocrinologists about the link between new-onset diabetes and pancreatic cancer, the actionable insights that may emerge from current research efforts, the biologic factors that may predict which patients with new-onset diabetes may be most at risk for pancreatic cancer, and the need for increased awareness of this association across medical specialties.
It is difficult — and often impossible — to determine the exact temporal or causal relationship between diabetes and pancreatic cancer, according to Max Petrov, MD, PhD, MPH, professor of pancreatology at The University of Auckland School of Medicine in New Zealand and founder of the COSMOS group, which focuses on translational clinical and epidemiological research in pancreatic diseases.
“It is possible that, in some patients, diabetes leads to development of pancreatic cancer. It is also possible pancreatic cancer leads to development of diabetes,” Petrov told HemOnc Today.
Diabetes duration — specifically new-onset diabetes, meaning diagnosed within the prior 3 years, as opposed to long-standing diabetes — can be used as a guide, he added.
“The shorter the time since diabetes onset, the higher the chance the diabetes the primary care provider or endocrinologist sees is the first manifestation of cancer that has not been diagnosed,” Petrov said.
In these cases, the mechanism is “fairly straightforward,” Petrov said.
“It is the so-called paraneoplastic syndrome,” he said. “As a part of this syndrome, endocrine dysfunction in the beta cells develops. The key thing to note is this is a nonspecific syndrome, which will have implications for screening.”
Mark O. Goodarzi, MD, PhD, FACP, director of the division of endocrinology, diabetes and metabolism and Eris M. Field chair in diabetes research at Cedars-Sinai, suggested a “bidirectional relationship” is at play.
“Factors such as inflammation or altered immune markers might play a role,” he said. “As far as longstanding diabetes, obesity may be a key contributor. People who are obese and have diabetes often have high insulin levels, and insulin can stimulate cell division. Over time, high levels of insulin could promote tumor formation. This is not proven, but it is a theory.
“The concept with new-onset diabetes is that the pancreatic cancer cells themselves might be producing some factor that causes diabetes,” Goodarzi added.
An ‘astounding’ finding
Suresh T. Chari, MD, professor in the department of gastroenterology, hepatology and nutrition in the division of internal medicine at The University of Texas MD Anderson Cancer Center, has examined the link between new-onset diabetes and pancreatic cancer since the late 1990s.
During a fellowship in gastroenterology at Mayo Clinic, Chari and colleagues assessed a population-based cohort of 2,122 residents of Rochester, Minnesota, aged 50 years or older who met criteria for diabetes between 1950 and 1994. They identified those who developed pancreatic cancer within 3 years of meeting diabetes criteria, then compared incidence in that cohort with expected rates from a separate SEER registry.
Their results appeared in Gastroenterology in 2005.
“An astounding 0.85% [of the Rochester cohort] had pancreatic cancer,” Chari told HemOnc Today. “That is a six- to eightfold higher risk compared with the general population.”
For context, the observed rate was comparable to yields of colorectal cancer by colonoscopy (0.7%), lung cancer by CT scan (0.65%) and breast cancer by mammography, Chari added.
Chari tried for years to highlight the importance of this observation. The medical community finally began to take notice within the last decade.
Pancreatic Cancer Action Network (PanCAN) — a U.S.-based charity that provides research funding, as well as patient support and community outreach — formed the Deadliest Cancers Coalition in 2008 to address policy issues related to the most lethal cancers.
In 2012, Congress passed the Recalcitrant Cancer Research Act, which required NCI to develop a strategic plan to advance pancreatic cancer research. A committee met for the first time in 2014.
“Among the concepts presented was that diabetes is a marker for pancreatic cancer,” Chari said. “The director caught onto it and, when [a subsequent] white paper came out on directions for pancreatic cancer research, the No. 1 research priority was to understand the link between diabetes and pancreatic cancer.”
NCI statistics now show one in four people diagnosed with pancreatic cancer had a prior diabetes diagnosis, and about one in 100 people diagnosed with new-onset diabetes will be diagnosed with pancreatic cancer within 3 years.
Kenner and colleagues published a review earlier this year in Pancreas that further quantified the association.
The authors, who aimed to explore the potential role of artificial intelligence in early detection of pancreatic cancer, divided study participants into six risk groups.
The group at highest risk, which included individuals with new-onset diabetes, had a six- to 10-fold higher risk (0.67% to 1% absolute risk) for pancreatic cancer — comparable to the elevated risk for lung cancer observed among people with a 20 pack-year smoking history.
However — unlike for colorectal, lung or breast cancer — there is no widely used screening modality or protocol for pancreatic cancer. Testing every patient with new-onset diabetes is neither logistically nor financially feasible.
“It is certainly not going to be cost-effective to send every patient with new-onset diabetes for imaging for pancreatic cancer,” Goodarzi said. “More than 95% of them simply will have type 2 diabetes.”
In addition — unless a person has undergone regular blood tests — it is difficult to pinpoint when they became diabetic and, subsequently, identify those most likely to present short term with pancreatic cancer.
“Patients don’t have blood tests frequently enough — even at [practices] where their health providers actually pay for it,” Ziding Feng, PhD, professor in the public health sciences division and co-leader of the biostatistics program at Fred Hutchinson Cancer Research Center, told HemOnc Today.
“We want to be able to see [results from] the past 18 months to see if they had at least one normal blood sugar,” Feng added. “That requires at least two annual tests. If we don’t have that, we don’t know if the diabetes is new or old.”
Identifying those most at risk
The challenges that make pancreatic cancer detection and treatment so difficult are a daily reality for many oncologists.
“Based on the numbers, when 80% of my patients walk through the door for the first time, I know that we are highly unlikely to cure their cancer,” Wolpin said.
However, Wolpin is encouraged by advances in early detection over the past half-dozen years. Among them:
- understanding the inherited basis of pancreatic cancer, including uncovering genetic mutations that identify high-risk individuals for screening and that now indicate a role for genetic testing in all patients who present with pancreatic cancer;
- increased knowledge of cystic lesions of the pancreas, important precursors to invasive cancer that are more detectable via imaging than other lesions;
- advances in blood-based cancer detection assays; and
- improved understanding and awareness of the relationship of pancreatic cancer with hyperglycemia.
Each advance provides a potential strategy to detect pancreatic cancer — and identify which individuals with new-onset diabetes may be at greatest risk.
“If there is a test that can detect cancer before anything else is visible, the subset of patients with new diabetes or new prediabetes could undergo that test,” Chari said. “Now you have a dual approach to finding someone who is at risk.”
Lack of useful biomarkers, however, remains a challenge.
“We not only need blood collected from patients with new-onset diabetes and pancreatic cancer, we also need blood from patients with new-onset diabetes without pancreatic cancer at the time of diabetes onset,” Feng said. “That is the right specimen, but you cannot find such a specimen anywhere.”
Considerable investment, potentially big dividends
Oncologists and endocrinologists are optimistic that two major research initiatives may yield new insights, as well as the biological specimens necessary to understand the diabetes-pancreatic cancer link in a way that could lead to more effective detection strategies and, thus, improved patient outcomes.
The NCI-led New Onset Diabetes Study (NOD) — launched in 2018 — is designed to enroll 10,000 people aged 50 to 85 years across the United States newly diagnosed with diabetes or hyperglycemia.
Researchers will follow study participants for 3 years and calculate pancreatic cancer incidence rates.
The regular blood and tissue samples they collect during that “critical time window” could capture important insights, said Feng, the NOD study’s principal investigator.
Researchers hope the information they glean will help develop a blood test that can determine which individuals with new-onset diabetes are most likely to develop pancreatic cancer and could benefit from additional imaging or further workup.
“If we could leverage that new-onset diabetes to pick up these patients well before the tumors have grown to a large size, it would make a huge difference in their survival,” Goodarzi said. “That is why the NIH is investing so much in this NOD study. We will have blood samples before the cancer manifests. It will be a great resource to go back to and look for biomarkers in the patients who did develop cancer.”
The target completion date for the NOD study is December 2025. However, enrollment has been slower than expected, standing at about 1,500 patients — or 15% of the target.
PanCAN announced a randomized study — the Early Detection Initiative for Pancreatic Cancer (EDI) — this summer.
The $25 million initiative will enroll 12,500 participants based on first-time elevation in fasting blood glucose or HbA1c.
Researchers will randomly assign participants to an observational group or an intervention group. Those in the intervention group will have enriching new-onset diabetes for pancreatic cancer (ENDPAC) scores calculated based on age, body weight and glucose/HbA1c values. Those with ENDPAC scores greater than 0 will undergo abdominal imaging.
Participants also will provide blood samples at up to five time points.
Researchers hope to determine whether imaging at the time of new-onset diabetes helps detect pancreatic cancer earlier. Study participants’ blood samples will be added to an NCI collection so they can be analyzed for possible pancreatic cancer biomarkers, which could lead to development of a screening modality that allows for detection of pancreatic cancer when it is amenable to surgery.
The target completion date is 2030.
“These two big investments will pay dividends,” Feng said. “Of course, the study teams have to deliver.”
‘Awareness is critical’
Results of the NOD and EDI studies eventually may lead to earlier detection of some pancreatic cancer cases, potentially altering the treatment landscape.
Feng said he is optimistic technology also could play a “game-changing” role in detection.
However, until trial data yield new insights or technologic breakthroughs become mainstream, experts said physicians and researchers who treat pancreatic cancer must continue to advocate for their patients and the research that could help them.
It is essential to educate providers in a range of medical specialties — including primary care, hospital medicine and endocrinology — about the link between new-onset diabetes and pancreatic cancer, according to experts with whom HemOnc Today spoke.
“Don’t delay diagnosis of diabetes because you’ll lose that window of opportunity to intervene,” Chari said.
Health changes around the time of diabetes diagnosis may be indicative of an eventual pancreatic cancer diagnosis. If a patient presents with diabetes, physicians should run through a checklist, Wolpin said.
“Does the elevation in glucose seem much more rapid than [for the] average patient or is the blood sugar particularly difficult to control? Is the patient losing weight in a way that seems abnormal and isn’t due to a lifestyle change?” Wolpin said. “And what is their age? Are they fine and then, all of a sudden, at age 72 they get diabetes and it’s not clear why?
“These are all warning signs that physicians should be aware of,” Wolpin added. “These are not biomarkers in terms of advanced tests, but they all go together to help raise suspicion of whether [an underlying] pancreatic cancer is present.”
Family history of pancreatic cancer is a potential warning sign, Goodarzi said. So is lack of relatives with diabetes, given type 2 diabetes “almost always” runs in a family, he added. Other considerations include gastrointestinal symptoms, loss of appetite or jaundice.
“The ENDPAC model includes factors such as age of diabetes diagnosis, weight changes and blood glucose changes over 1 year,” Goodarzi said. “The sensitivity of this model seems to be around 55%. Down the road, we will need a combination of clinical features and biomarkers, but we don’t have that yet.
“For now, it is the ‘art of medicine,’” Goodarzi added. “If a patient’s clinical presentation of diabetes makes you think twice, and it does not look like garden-variety diabetes, you may want to screen for pancreatic cancer.”
Petrov, however, cautioned that weight loss and increases in blood glucose are “nonspecific symptoms” and it remains impossible to determine with certainty which patients will develop pancreatic cancer or any other malignancy.
However, he highlighted a study his COSMOS group conducted that suggested history of pancreatitis could be a key indicator.
His group performed a large, population-based study of 140,000 people in New Zealand with type 2 diabetes. They aimed to determine if a specific subpopulation may be at higher risk for pancreatic cancer.
The results, which included up to 18 years of follow-up, appeared in 2020 in Diabetes Care.
“What we found was quite spectacular,” Petrov said. “Patients who had diabetes that developed secondary to an attack of pancreatitis had a seven times higher risk for pancreatic cancer than people with type 2 diabetes. In absolute numbers, the frequency of pancreatic cancer increased from 0.7% [in the overall cohort] to 3.1% for those with diabetes and a history of pancreatitis. This didn’t require sophisticated blood tests or costly imaging — just asking about history. It is conceivable that taking into account history of pancreatitis will markedly enrich cohorts of people with new-onset diabetes for pancreatic cancer and ultimately enable cost-effective and achievement-appropriate screening for this disease.”
The results also suggested it is important to consider the temporal relationship between pancreatitis and diabetes, Petrov said.
“When new-onset diabetes after pancreatitis was compared with diabetes that developed before pancreatitis, the former was associated with more than two times higher risk for pancreatic cancer,” he said. “This is important not only from the perspective of screening, but also from the perspective of trying to figure out the intricate mechanisms underlying the link between diabetes and pancreatic cancer. Our findings indicate that new-onset diabetes in itself is a relatively minor, although certainly non-negligible, risk factor for pancreatic cancer. New-onset diabetes plays a role as a nonspecific amplifier of major pancreas-specific risk factors such as pancreatitis.”
Aside from physical predictors, awareness among the general population also would be valuable, Wolpin said.
A deeper understanding of smoking’s role in lung cancer led to large-scale trials that resulted in the development of national guidelines for screening high-risk populations. Increased awareness of the diabetes-pancreatic cancer relationship could be equally transformational, Wolpin said.
“We need to mimic what the scientific community did for smoking and lung cancer,” Wolpin said. “Ultimately, trials were conducted that showed low-dose CT scans could detect the disease early, and now this form of screening is considered a standard of quality care.
“If no one realizes the diabetes-pancreatic cancer link is present, they will be less likely to participate in studies like NO and EDI, which are greatly needed,” Wolpin added. “Awareness is a critical part of making a difference.”
- Chari ST, et al. Gastroenterology. 2005;doi:10.1016/j.gastro.2005.05.007.
- Johns Hopkins Medicine. Pancreatic cancer prognosis. Available at: www.hopkinsmedicine.org/health/conditions-and-diseases/pancreatic-cancer/pancreatic-cancer-prognosis. Accessed Oct. 1, 2021.
- Cho J, et al. Diabetes Care. 2020;doi:10.2337/dc20-0207.
- Kenner, B, et al. Pancreas. 2021:doi:10.1097/MPA.0000000000001762.
- NCI. Could a diabetes diagnosis help detect pancreatic cancer early? Available at: www.cancer.gov/news-events/cancer-currents-blog/2021/pancreatic-cancer-diabetes-early-detection. Accessed Oct. 1, 2021.
- Rahib L, et al. JAMA Netw Open. 2021;doi:10.1001/jamanetworkopen.2021.4708.
- Yuan C, et al. JAMA Oncol. 2020;doi:10.1001/jamaoncol.2020.2948.
- For more information:
- Suresh T. Chari, MD, can be reached at Department of Gastroenterology, Hepatology and Nutrition, The University of Texas MD Anderson Cancer Center, 1400 Pressler St., Houston, TX 77030; email: email@example.com.
- Ziding Feng, PhD, can be reached at Public Health Sciences Division, Fred Hutchinson Cancer Research Center, 1100 Fairview Ave. North, P.O. Box 19024, Seattle, WA 98109; email: firstname.lastname@example.org.
- Mark O. Goodarzi, MD, PhD, FACP, can be reached Cedars-Sinai, 8723 Alden Drive #SSB-250, Los Angeles, CA 90048; email: email@example.com.
- Max Petrov, MD, PhD, MPH, can be reached at School of Medicine, University of Auckland, 28 Park Ave., Auckland 1023, New Zealand; email: firstname.lastname@example.org.
- Brian M. Wolpin, MD, MPH, can be reached at Department of Medical Oncology, Dana-Farber Cancer Institute, 450 Brookline Ave., Boston, MA 02215; email: email@example.com.