SAN FRANCISCO – A new diagnosis of diabetes may help identify older adults who will develop pancreatic cancer while there is still time for screening and early detection, researchers reported at a meeting on gastrointestinal cancers sponsored by the American Society of Clinical Oncology.
In an observational study of more than 20,000 older adults with pancreatic cancer, 10 antecedent diagnoses were found to be significantly associated with the cancer diagnosis.
Of these, a diagnosis of new-onset diabetes preceded the cancer diagnosis by the greatest amount of time – more than 2 years, on average – or potentially enough time to catch the cancer early with targeted screening. A diagnosis of abdominal pain was second, at 1.5 years.
Late diagnosis is a major contributor to the generally "dismal" survival of pancreatic cancer, lead investigator Dr. Elizaveta Ragulin-Coyne said in an interview.
"Colonoscopy screening works great, mammography works great. But those cancers are really a lot more common, so it makes sense to screen the whole population," she commented.
By contrast, pancreatic cancer is relatively uncommon, so population-based screening with current tests would generate many false positives. At present, only individuals from families having hereditary pancreatic cancers associated with certain mutations are screened.
The goal of the study was therefore to identify "the factors that can precede the diagnosis of pancreatic cancer, that sort of can act as red flags to identify that population at risk," she explained. "So we are trying to identify the risk-rich population of individuals who can benefit from potential future screening."
The investigators analyzed data from the Surveillance, Epidemiology, and End Results (SEER) database for the years 1991-2005 and the linked Medicare database for the years 1991-2007 to identify older adults with a diagnosis of pancreatic cancer and diagnoses preceding the cancer.
They evaluated 30 possible antecedent diagnoses for their association with the pancreatic cancer diagnosis, and narrowed it down to 10 that were significantly associated (P less than .05) in a stepwise logistic regression analysis: acute pancreatitis, chronic pancreatitis, cyst-pseudocyst, other pancreatic disease, bile duct obstruction, diabetes, weight loss, jaundice, abdominal pain, and hepatomegaly.
The 22,493 study patients were 77 years old on average; 55% were women and 86% were white, according to results reported in a poster session at the meeting.
The 10 antecedent diagnoses ranged in prevalence in this population from a low of 4% for hepatomegaly to a high of 76% for abdominal pain. A diagnosis of diabetes was seen in 45%.
In most cases, the median time between the antecedent diagnosis and the pancreatic cancer diagnosis was less than 3 months. The exceptions were abdominal pain, diagnosed a median of 18 months before the cancer, and diabetes, diagnosed a median of 28 months before the cancer.
The latter intervals are long enough to provide a window of opportunity for intervention, according to Dr. Ragulin-Coyne, a surgical resident and research fellow at the University of Massachusetts Medical Center in Worcester.
"It doesn’t make sense if you have preceding diagnoses within a month before, it doesn’t really make a difference," she explained. "But if it’s over 6 months or over a year, it is actually clinically significant because you can hypothesize that those people are potentially at an early stage and could have more interventions that give you a possibility of cure."
The average number of antecedent diagnoses decreased with increasing stage of pancreatic cancer at diagnosis, from 3.91 among patients with stage 0 disease to 2.04 among patients with stage IV disease.
This finding initially seemed counterintuitive, Dr. Ragulin-Coyne said. But perhaps patients having more advanced cancer at diagnosis have had less contact with the health care system in general, and therefore have fewer diagnoses on record.
In a logistic regression model among just the patients with an antecedent diabetes diagnosis, the odds of the gap between that diagnosis and the pancreatic cancer diagnosis being greater than 24 months were higher for nonwhite versus white patients; for patients aged 75-84 years or aged 85 years or older, compared with those aged 65-74 years; and for patients in the Midwest versus the Northeast.
The reason pancreatic cancer is diagnosed earlier in some patients and later in others is not yet clear, but it is likely multifactorial, according to Dr. Ragulin-Coyne.
"We can make guesses, whether it is socioeconomic or cultural or there is something else in play." For example, some patients may "tell the doctor about all their symptoms and get worked up early and get their doctors concerned more," she said. "But if they never come to the physician or they never mention what’s going on, they get diagnosed late."