Harvard University scientists have found a suspicious and alarming association between coffee drinking and cancer of the pancreas, one of the fastest-killing of all cancers.

Though the link is not yet proved and more studies are needed, the Harvard study's main author, epidemiologist Dr. Brian MacMahon, said yesterday he gave up coffee a month ago.

"I'm fairly convinced this is a real thing," he said, but "I would not give any advice to others yet. I'm in the business of collecting facts."

The data, reported in today's New England Journal of Medicine, show that:

About 24,000 Americans a year get cancer of the pancreas, a large abdominal organ that secretes digestive enzymes and insulin. Only 1 or 2 percent will survive as long as five years. This dismal record means pancreatic cancer ranks only after lung, colon and breast cancers as a cause of death.

MacMahon's Harvard School of Public Health team studied 369 patients with this cancer and compared them with a "control group" of 644 patients with other disorders. There were only 20 noncoffee drinkers among the pancreas cancer patients, but 88, or 13 percent, among the controls, enough to make a reliable statistical base. The inquiry showed drinking up to two cups of coffee daily had apparently increased the risk of pancreas cancer by 1.8 times or nearly twofold, and drinking three or more cups increased it by 2.7 times or nearly three-fold.

If the coffee-pancreas cancer link is confirmed, just over 50 percent of such cancers may be caused or partly caused by coffee, the Harvard group believes.

The chance of developing such cancer, like most cancers, is generally small in younger years and increases with age. Many kinds of cancer take 20 to 30 years to develop.

If coffee is indeed implicated in half of all pancreas cancer, said MacMahon, a noncoffee drinker aged 50 to 54 might have 7 chances in 100,000 of getting pancreas cancer in any single year; one aged 60 to 64, 19 chances in 100,000; one aged 65 to 69, 27 chances -- with coffee drinkers' chances about doubled or tripled.

The Harvard group also found a "weak association" between cigarette use and pancreas cancer, but none with cigars, pipes, alcohol or tea. Since both coffee and tea contain caffeine -- the average cup of tea, one not too strong, has about half as much caffeine as the average cup of coffee -- "we don't believe" caffeine is the guilty substance, "though it's not yet ruled out," MacMahon reported.

Coffee has been under investigation in recent years in several disorders. There is no strong evidence that moderate coffee use increases the risk of heart disease, despite some allegations.

The National Coffee Association in New York said yesterday that the Harvard study has many serious defects and that "extensive animal research" by the coffee industry has found no connection between coffee and cancer.

Last year the Food and Drug Administration warned pregnant women they should stop or "minimize" coffee, tea and cola drinking because the caffeine in all three may cause birth defects. The evidence here, too, is not yet conclusive, said FDA officials, but there is enough evidence to warn the "prudent."