The disease model of type 1 diabetes as a chronic, progressive autoimmune disorder that generally arises in childhood and renders individuals insulin-dependent for life has been settled for decades. However, a peek under the hood of this serious condition with far-reaching implications for health and quality of life reveals a vibrant research effort that is redefining conventional notions about its natural history, calling into question current definitions of type 1 diabetes, and posing a series of what-ifs about screening, testing for, and preventing the disease. A recent milestone in this parade of evidence substantiated that children with multiple insulin autoantibodies are virtually certain to develop diabetes within 20 years, and that those who have positive autoantibodies before age 3 years generally progress faster to diabetes than those who test positive after age 3.
"We've proven not only that approved existing markers predict diabetes—almost 100 percent in some cases—years before clinical diagnosis but also that these markers operate with similar parameters in different populations," explained a senior author of the analysis, Marian Rewers, MD, PhD. "It's a universal finding that's not dependent on ethnicity, access to care, or other factors that often bias early versus late diagnosis of clinical disease. We also showed that the rates of progression are very predictable despite the differences in the study populations." Rewers is a professor of pediatrics and medicine and interim executive director of the Barbara Davis Center for Diabetes at the University of Colorado School of Medicine in Aurora.
Quantifying Long-Term Risk
The authors pooled data from three prospective studies involving 13,377 children genetically at risk of type 1 diabetes in Colorado, Germany, and Finland (JAMA 2013;309:2473–9). Recruitment criteria for the trials—DAISY in Colorado, BABYDIAB/BABYDIET in Germany, and DIPP in Finland—were slightly different for each one, but the subjects—all newborns or infants at enrollment—were considered at risk of type 1 diabetes based on HLA genotyping or by having a first-degree relative with the disease.
The subjects all had repeat autoantibody testing at short intervals early in life which continued at longer intervals (ranging from 3 months to 3 years) as the children grew up. All the trials measured the same three autoantibodies, glutamic acid decarboxylase 65 (GAD 65), insulinoma antigen 2 (IA2), and insulin autoantibodies (IAA); two also included zinc transporter 8 (ZnT 8) autoantibodies.
Overall, slightly fewer than 8% of subjects (1,059) seroconverted to positive autoantibodies, and of these, 55% had multiple autoantibodies. In all, 428 children eventually developed type 1 diabetes based on symptomology or standard blood glucose definitions for the disease, including 25 who were not autoantibody positive prior to diagnosis. The hazard ratio for developing diabetes in children with multiple positive autoantibodies was an impressive 395.6 in comparison to those without autoantibodies. Overall, 43.5% of the former progressed to diabetes within 5 years of seroconversion, 69.7% did so within 10 years, and 84.2% at 15 years.
"The real significance of these finding is they show that nearly 100 percent of people who have two or more autoantibodies go on to develop type 1 diabetes if you follow them long enough. What we've known heretofore is that people with at least two autoantibodies are at high risk. We felt pretty sure about predicting five-year risk, but we didn't know and couldn't state what happened to the other 50 percent who didn't develop it in five years. Did they not develop it at all, or does it just take more time?" said Jay Skyler, MD. "This study, by looking at people over a long period of time, is profoundly important because it says that this is a disease process that doesn't stop, and we really need to find new interventions." Skyler is a professor of medicine, pediatrics, and psychology at the University of Miami Leonard M. Miller School of Medicine. He also chairs the National Institute of Diabetes and Digestive and Kidney Diseases'
(NIDDK) Type 1 Diabetes TrialNet, an international consortium of 18 clinical centers studying prevention and early treatment of the disease. His editorial about Rewer's study was published in the same issue of the Journal of the American Medical Association (JAMA 2013;309:2491–2).
A New Definition of Diabetes?
Based on the study's conclusive findings along with other evidence, Skyler is among a chorus of experts who have begun to call for a new definition of type 1 diabetes that would include not only existing criteria—symptoms and glycemia thresholds like a fasting plasma glucose concentration >126 mg/dL—but also autoantibody positivity. He and others insist that such a change, even in the absence of prevention or cure, would at least prepare families for the inevitable, no small consideration given that as many as 45% of American type 1 diabetics learn they have the disease when they present at the emergency department in full-blown, potentially lethal, diabetic keto-acidosis.
"Kids and families who know the child is antibody-positive—even if they go on to diabetes—experience a very different clinical course," said Rewers. "They rarely need to go to the hospital or emergency department for diagnosis. That's because from the moment we know the child is antibody positive, we teach the family how to test blood sugar, what the symptoms of hypoglycemia are, and whom to call when they experience symptoms."
He and Skyler also contend that keeping the diagnostic criteria where it is now is tantamount to diagnosing cardiovascular disease when a patient arrives at the emergency department in cardiac arrest. Both are serving on a working group of the American Diabetes Association, Juvenile Diabetes Research Foundation, and the National Institutes of Health exploring the possibility of changing the diagnostic criteria for type 1 diabetes.
While Rewers and Skyler insist the time has come to update the diagnostic criteria, others are not so certain. "The big problem is, even though years of research have been going on, there's nothing that can be done to prevent or even delay the onset of the disease. So despite many large therapeutic trials, people who are going to develop it, develop it. This has resulted in a big controversy," explained David Sacks, MB ChB, chief of clinical chemistry at the National Institutes of Health Clinical Center in Bethesda, Md., and chair of the National Association of Clinical Biochemistry's Laboratory Medicine Practice Guideline for laboratory analysis in diagnosing and managing diabetes. "Some people say it's absurd while others say it's worth doing, because you'll prevent people from developing diabetic ketoacidosis. This whole debate is very polarized."
Prevention Trials Fall Flat
The researchers agree that a long history of tertiary prevention trials started after the subjects already had type 1 diabetes has failed to make a dent in arresting the disease. Ditto so far for more recent and arguably more rigorous secondary or primary prevention efforts, which enrolled subjects with positive autoantibodies or no evidence of autoimmunity, respectively. However, more such trials are in process.
A pilot primary prevention study of 230 infants with HLA subtypes and first-degree relatives with type 1 diabetes showed a decreased risk of autoimmunity when babies received infant formula without cow's milk protein. TRIGR, a multinational trial exploring the same intervention and powered enough to provide statistically significant conclusions is underway, with results expected in 2017.
In addition, a post-hoc analysis from the DPT-1 trial showed oral insulin administration was associated with a projected 4.5–5 year delay in diabetes development for certain patients, including those with high autoantibody titers. NIDDK's TrialNet has another oral insulin trial underway, enrolling patients with high IAA titers to test again, prospectively, whether this intervention might be successful.
Skyler is particularly intrigued by the work of Dennis Karounos, MD, using an inactive insulin analog which does not lower glucose levels. This has enabled Karounos to study in mice the preventive effects of long-term high-dose insulin therapy (OmicsOnline 2012 doi:10.4172/scientificreports.161). Based on this and other research, Rewers also believes that preventive trials to date might have missed the mark by starting too late in the disease process. He explained that investigators are in a catch-22 because the U.S. Food and Drug Administration wants to see efficacy of new interventions in newly diagnosed diabetics before giving the green light to test these preventive therapies in people who are merely at risk.
Is Screening Realistic?
Rewers' work and that of his collaborators in Germany and Finland also raise questions about how any type 1 diabetes screening schema could be implemented. If multiple autoantibodies are practically a sure bet for developing the disease, the case is not so clear for individuals with one positive autoantibody; only about 10–20% progress to diabetes. Further complicating matters, a distinct minority of subjects who developed diabetes—just 0.4% in the combined DIPP, DAISY, and BABYDIAB/BABYDIET analysis—lacked autoantibodies. What about genetic or genomic analysis? Would it help or hinder any screening schema, considering that approximately 85% of newly diagnosed patients do not have a first degree relative with type 1 diabetes and that HLA genotyping accounts for only about 50% of risk associated with diabetes? All of this in the context of a relatively rare disease, which, though rising in prevalence, still only affects about 0.4% of the population worldwide.
Sacks looks at all these factors and sees a murky picture at best. "If you were to screen selective populations, you'd only identify a small percent of the people with type 1 diabetes. Also, because of its low prevalence, if you screened the general population you'd get more false positives than true positives, which means a very low positive-predictive value," he explained.
Sacks also pointed out challenges with knowing the best timing for screening. "How long would you screen to see when patients become positive? What would you do once they seroconvert, and how long would you keep testing them for diabetes? Remember, some in the study developed diabetes weeks after they seroconverted, whereas others took years," he said. "My personal opinion is that in the absence of therapy, this debate will never be resolved because some people will say screen, and others will say don't screen."
Even as medicine embraces genetic and genomic testing, experts agree that in the case of type 1 diabetes, these methodologies will not simplify any potential screening program. "Genetics would reduce the number of individuals who would need to be tested for autoantibodies, but we would then miss a number of people due to their genetics. It's a very useful tool for research but it would be hard to justify as a means of selecting up to four percent of the population," contended Rewer's coauthor Ezio Bonifacio, PhD, a professor of pre-clinical stem cell therapies and diabetes at the Technical University of Dresden in Germany. "If we get to a situation where we could prevent diabetes and we knew we had an agent that would reduce the risk substantially, we'd want to find everyone who has diabetes. In that case, it might make sense to test everybody for autoantibodies."
He and his colleagues recently talked through a what-if screening scenario for the German state of Bavaria. "We thought we could just screen for GAD65 and IA2 as anyone with multiple positive antibodies is likely to have those two. Both assays work well, and there are even very good commercial tests," he explained. "We could identify the GAD65 or IA2 positives, then process additional specimens to confirm that they remain positive." Rewers also thinks autoantibody testing would be the best screening approach. He and his University of Colorado colleagues have thought about a model of screening for antibodies two or perhaps three times during childhood.
Meanwhile, Skyler looks at the challenges of any type 1 diabetes screening strategy and believes universal vaccination would be a better approach. "What we actually need, in my view, is development of a vaccine that is so safe that we could give it to the general population like we do other vaccines and not worry that we're immunizing against diabetes in people who'll never get it," he explained.
Improving Autoantibody Measurement
A landmark study in 1974 first identified pancreatic islet cell auto-antibodies in type 1 diabetes and established the disease as an autoimmune disorder. Autoantibody testing evolved over time, but the method some laboratorians still swear by is indirect immunofluorescence. However, this labor-intensive, subjectively interpreted test largely has been replaced by biochemical autoantibody assays for glutamic acid decarboxylase 65 (GAD 65), insulinoma antigen 2 (IA2), insulin autoantibodies (IAA), and zinc transporter 8 (ZnT 8) autoantibodies.
Many labs measure these antigens using fluid phase radiobinding assays, but this method also has fallen out of favor as many labs no longer work with radioactive compounds. Commercial enzyme-linked immunosorbent assays subsequently entered the picture, but as is typical with immunoassays, achieving comparable results between the tests has proven challenging.
Diabetes researchers and public health organizations have launched several initiatives over the years to ensure consistent, reliable autoantibody measurements. The World Health Organization adopted a serum reference standard for GAD 65 and IA2 and reportedly is still working on standards for IAA and ZnT8.
In 2000, the U.S. Centers for Disease Control and Prevention in collaboration with the Immunology of Diabetes Society established the Diabetes Antibody Standardization Program to improve assay comparability and evaluate new autoantigens and test methodologies. This very active initiative, now known as the Islet Autoantibody Standardization Program (IASP), has conducted numerous international workshops in which at least 50 labs from 20 countries have tested blinded samples from 50 new-onset type 1 diabetics and up to 100 controls. According to one of the IASP leaders, Peter Achenbach, MD, the IASP currently is conducting another workshop and expects new data within the next year.
Separately, the National Institute of Diabetes and Digestive and Kidney Diseases led a harmonization effort and set up an islet autoantibody harmonization committee among four labs participating in its multicenter studies of type 1 diabetes. Among other efforts, this group adopted common working calibrators, units, and methods, and achieved high concordance among participants in measuring GAD and IA2.
- Bonifacio E, Yu L, Williams A, et al. Harmonization of glutamic acid decarboxylase and islet antigen-2 autoantibody assays for National Institute of Diabetes and Digestive and Kidney Diseases Consortia. J Clin Endocrinol Metab 2010;95:3360–7.
- Lampasona V, Schlosser M, Mueller P, et al. Diabetes antibody standardization program: First proficiency evaluation of assays for autoantibodies to zinc transporter 8. Clin Chem 2011;57:1693–1702.
- Mueller P, Achenbach P, Lampasona V, et al. Type 1 diabetes autoantibodies. Clinical Laboratory News. October 2010.
The Role of Clinical Labs
As more evidence comes out about the relationship between autoantibody positivity and development of type 1 diabetes, and as diabetes prevention trials continue, clinicians as well as researchers might turn to antibody testing in some sort of screening or diagnostic capacity. In a forthcoming perspectives article in Clinical Chemistry, Sacks raises several questions about how labs might respond to this ramped-up interest. Among them, he ponders how many clinical labs will be prepared to use the quantitative radiobinding assays that have been used to detect autoantibodies. While noting the great progress that has been made in standardizing autoantibody assays, he also wonders, given the limited amount of standardization material available, whether autoantibody testing should be conducted only in specialized labs.
Rewers and Bonifacio acknowledged that some challenges with existing assay methods remain, despite very considerable standardization and harmonization efforts, the latter of which Bonifacio co-chairs (see Improving Autoantibody Measurement, above). For example, Rewers in his clinical practice sees patients who have had autoantibody testing performed by labs which have not been active in standardization efforts and that use commercial kits. "Their results are definitively inferior to what the leading labs can offer. It may matter less for clinical care, but for research where you're relying on absolute precision you have to work with one of the four labs that have been part of the harmonization initiative," he explained. "If the lab is not part of that effort the results are going to be disappointing."
Rewers and Bonifacio maintain that measurement of islet autoantibodies is moving beyond the current generation of assays. Rewers' lab has been on the vanguard of developing a robust electrochemiluminescent assay that detects autoantibody positivity earlier than other prototype assays (Diabetes Care 2013;36:2266–70). "I'm very optimistic about the acceptance of this new platform for IAA. It's a precarious position to be first in the field when people are trying to replicate what you are doing, but this assay has been verified in four patient populations and the results were outstanding," said Rewers.
If neither the prototype chemiluminescent assay nor routine autoantibody screening are ready today to implement clinically, experts still encourage clinical labs to keep abreast of developments in the science involving type 1 diabetes and be prepared to change lab practices for this field perhaps sooner rather than later. As Skyler put it, "Know that the definitions are changing, and that we'll be switching more and more to relying on autoantibodies and HLA genotype as indexes of the disease process. That will cause a profound shift in what labs will need to be measuring."