BOSTON-Diabetologists are getting all churned up over cow’s milk.   Five years and numerous studies after Canadian and Finnish teams suggested bovine serum albumin might trigger the immune attack on pancreatic beta cells that initiates Type 1 disease, there’s still no definitive proof one way or the other.   But the conflicting evidence that’s piled up in the interim, thoroughly aired at the American Diabetes Association meeting here, seems only to have amplified the faith of believers and the scorn of skeptics.   University of Toronto researcher Hans-Michael Dosch helped launch the debate in 1992 when he discovered that all of the 142 Finnish children with new-onset Type 1 disease he and his colleagues studied had elevated levels of anti-BSA antibodies, and that these could cross-react with a pancreatic islet-cell antigen.   Today Dr. Dosch argues that 44 of 51 studies point to a causal role for one or more bovine protein antigens. Some epidemiologic studies have found that breast milk is protective, others that protection appears to be lost if cow’s milk is introduced before age 3 to 4 months. Skim milk powder has been shown to be diabetogenic in animal models, and 92% of Western babies encounter their first foreign protein in cow’s milk.The known effects of wheat in celiac disease prove that food can induce autoimmunity. And if cow’s milk is the culprit in diabetes, Dr. Dosch says, then exclusive early feeding with mother’s milk or a nonantigenic hydrolyzed casein formula could be a simple way to prevent onset in genetically susceptible children.   Voicing strong opposition is Dr. Jill Norris, a University of Colorado epidemiologist whose meta-analysis of 17 case-control studies last year found only a weak association between infant diet and diabetes-one she says can’t support a causal role. She cites methodologic flaws, such as faulty maternal recall, that may have biased many studies. Her own ongoing cross-sectional study of siblings and offspring of Type 1 patients (DAISY, for Diabetes Autoimmunity Study in the Young) has turned up no differences in early exposure to cow’s milk or other dietary protein between 16 children who’ve developed beta-cell autoimmunity, including three with diabetes, and 360 who haven’t.   Moreover, adds Dr. Norris, no one has duplicated the 100% anti-BSA antibody results of Dr. Dosch. Indeed, Dr. Mark Atkinson’s team at the University of Florida in Gainesville weighed in with a study that found antibody levels up in only 10% of newly diagnosed diabetics, and an equal frequency among patients with other autoimmune diseases.   His coauthor, Dr. Noel Maclaren, now at LSU, rejoined the fray again last year in the wake of a 1994 stance by the American Academy of Pediatrics that infants of families with a Type 1 history avoid cow’s milk protein in the first year. Alarmed by reports of infants undernourished on alternative diets, and armed with early DAISY results, he declared cow’s milk innocent until proven guilty.   That may be okay in a court of law, Dr. Dosch responds, but not in the pursuit of causes of disease. He says the “dismaying” variability in BSA immune-response results smells more of technicalities than biological variation, and he predicts that more accurate tests will soon clarify them.   Meanwhile, he points to the marked protective effect of a hydrolyzed casein-based diet seen in NOD mice in his lab. After one year, 60% of them fed standard mouse food with a variety of animal and plant proteins had developed diabetes, but only 9% of animals on the test diet had. And in test-diet mice crossed over to standard diets at three months, the diabetes rate didn’t go up, suggesting diabetes resistance is conferred early in life and retained.   But Dr. Norris, who doesn’t deny the probability of dietary triggers, thinks other substances removed from the mouse test diet warrant a broader look.Wheat and soy have been more potent and consistent in other animal studies, she says.   That view got a boost this summer when Dr. Aldo Rossini’s University of Massachusetts group reported that adding BSA to hydrolyzed casein chow fed to BB/Wor rats didn’t increase spontaneous diabetes.   Dr. Dosch concedes that cow’s milk is probably not the only thing in the diet that’s diabetogenic, and no one thinks any dietary cause is responsible for even most autoimmunity.   Dr. Gisela Dahlquist of the University of Umeå in Sweden, whose own studies have confirmed a link between BSA antibodies and increased risk, has also identified a host of other potential environmental triggers in efforts to explain why Sweden has had a 50% rise in Type 1 diabetes since 1978. Among them are cold temperature, growth rate, some pre- and postnatal infections, blood group incompatibility, and nitrosamines.   Her “arbitrary calculations” are that cow’s milk could account for only 12.6% of Type 1 in Sweden, vs. 30% for enterovirus infection in utero.   Both sides are awaiting prospective natural history studies, now under way in the U.S. and Europe.   The world’s only intervention trial with high-risk children-a Finnish study comparing nine months of a hydrolyzed casein formula with a cow’s-milk-based formula-isn’t likely to yield answers soon, and has already been tossed a monkey wrench. One baby who’d been in the intervention group developed diabetes at age 14 months, says Dr. Hans Åkerblom of the University of Helsinki, but there was strong evidence of enterovirus infection. -Judy Ismach