August 2009
Feature Story

Karen Titus

In diabetes testing, the equivalent of the reporter’s imperative “Follow the money” could very well be “Follow the glucose.” Follow it closely and keep it tight.

But is it possible to unravel the story sooner, before someone is diagnosed with diabetes and requires glucose monitoring? Quite possibly. Laboratories are likely to have a few more items to track in coming years, among them autoantibody markers and genetic markers, thanks to researchers who are taking a good, hard look at new ways to predict, prevent, and treat diabetes.

They like what they’re seeing.

With type 1 diabetes, “The ability to predict disease onset has dramatically improved over the years,” says Gerald T. Nepom, MD, PhD, director, Benaroya Research Institute at Virginia Mason Medical Center, Seattle. Prospective clinical trials, he continues, have advanced to the point where physicians can predict and provide information to subjects about their risk for diabetes. The academic and research communities have developed tests to screen and select subjects for major, large-scale clinical trials for intervention and therapy. The next step would be to bring these tests into clinical practice a matter that appears to hinge on “when” rather than “if.”

In fact, some tests have entered into practice already. For all the excitement that accompanies their work, however, most researchers urge restraint. “None of it, in my opinion at least, is ready for prime time until we have a therapy that works,” says Dr. Nepom.

Ross J. Molinaro, PhD, MT (ASCP), is of the same opinion, noting that current practice guidelines do not recommend autoantibodies for routine diagnosis or screening of diabetes. Until type 1 diabetes can be prevented, islet cell autoantibody measurements should be limited to research protocols, says Dr. Molinaro, assistant professor, pathology and lab medicine, and medical director, core laboratories, Emory University Hospital Midtown, Atlanta.

The protocols offer an enticing peek at the future of diabetes testing, however.

In one of the more striking findings, researchers have demonstrated that people who have high-risk genotypes and two or more of four specific autoantibody markers have between a 70 and 90 percent risk of progression to type 1 diabetes over a three- to five-year period, says Dr. Nepom. “That’s an extremely high predictive risk ratio,” he says. It’s now being used routinely to screen subjects for entry into prevention clinical trials and early intervention clinical trials being conducted by several large consortia, including the National Institutes of Health’s TrialNet.

In one study, known as TEDDY (The Environmental Determinants of Diabetes in the Young), researchers are doing HLA typing on more than 400,000 newborns. George S. Eisenbarth, MD, PhD, sees a day when physicians will have genetic information on newborns for a series of immunologic disorders, enabling them to define, at birth, risk for type 1 diabetes, celiac disease, Addison’s disease, and perhaps others, says Dr. Eisenbarth, executive director, Barbara Davis Center for Childhood Diabetes, and professor of pediatrics, medicine, and immunology, University of Colorado Denver School of Medicine.

A study known as DAISY (Diabetes Autoimmunity Study in the Young) strongly suggests that even relatives of patients with diabetes present with extremely high levels of glucose, with ketoacidosis and risk of death, says Dr. Eisenbarth, one of the study’s authors (Barker JM, et al. Diabetes Care. 2004;27:1399–1404). “Thus, we believe screening of relatives for autoantibodies has benefit clinically now,” says Dr. Eisenbarth. The study concluded that childhood type 1 diabetes diagnosed through such a program has less severe onset and a milder clinical course in the first year after diagnosis.

The trials are also dropping strong clues about potential therapies. Some 14 clinical trials of immunomodulators are underway involving high-risk populations or those newly diagnosed with type 1 disease. “Patients with newly diagnosed disease are just at a later stage of disease progression,” Dr. Nepom says. “They still have a little bit of residual insulin secretory function. So the therapy’s the same at the different stages—it’s just designed to rescue remaining beta cells, either early or late.”

The more promising approaches include two different anti-CD3 monoclonal antibodies, teplizumab and otelixizumab. Both are anti-T cell therapies. There’s also a similar, anti-B cell therapy, called rituximab (anti-CD20), currently used to treat certain cancers and rheumatoid arthritis. Trials for CTLA-4 Ig (abatacept), originally targeted at rheumatoid arthritis, have just been completed. In addition, GAD-alum is being looked at as part of a vaccination protocol, says Dr. Nepom.

At the heart of many of these efforts lies a two-step testing protocol. The current standard of practice in the research community combines a genotype assay and an antibody screening assay.

The genotyping is done once, as the first step. This is a simple, PCR-based test for the high-risk HLA genes, which are found on both the HLA-DR4 and HLA-DR3 haplotypes. On the former haplotype, the genes are DQB1*0302 and DRB1*0401. On the DR3 haplotype, the target is DRB1*0301.

“For any clinical lab that does PCR, this is straightforward,” says Dr. Nepom. His lab uses a high-throughput mode with RT-PCR because he and his colleagues want, for research purposes, to distinguish homozygotes from heterozygotes. “I wouldn’t imagine a clinical lab would want to get that fancy.”

The most important thing to keep in mind, says Dr. Nepom, is that these are common genes. They confer risk but don’t cause disease.

Nor do they necessarily represent the complete story. Dr. Nepom says a lot of non-Caucasoid populations drop out after the first screen —the high-risk HLA genes are much more common in Caucasian populations. “So overall, the type 1 screening is heavily biased toward European and American Caucasians. And that is an issue. It hasn’t been exhaustively studied in the other major ethnic groups, largely because the incidence is lower, and the amount of resources that would need to go into that screening is quite large,” Dr. Nepom says. “It’s not lack of interest. The NIDDK [National Institute of Diabetes and Digestive and Kidney Diseases] is explicitly interested in that question. It’s an unmet need in research right now.”

Might there be more genes worth looking at? Certainly more genes are popping up. A recent article in Nature Genetics (published May 11 online) lists 42 genes involved in type 1 diabetes (including 25 new discoveries), at various stages of the disease.

Those who are genotype-positive then have their serum drawn for the antibody screening.

The antibody screening indicates whether the immune response has already been activated against the islets. Those who are genetically at risk, but don’t have the autoantibodies, actually have low risk of developing diabetes unless there is a triggering event, says Dr. Nepom. Fully 16 percent of the population carries one or more of the genes that confer risk. “It’s a big number. It doesn’t rule out tons of people,” he concedes. But only 1.6 percent are genotype- and antibody-positive.

The antibody panel consists of an antibody to insulin itself (IAA); an antibody to GAD65 (glutamic acid decarboxylase); an antibody to IA-2, which is a tyrosine phosphatase-like molecule; and an antibody to zinc transporter, called ZnT8. Having any two of the four is considered positive, says Dr. Nepom. Other commonly used antibodies include islet cell cytoplasmic autoantibodies (ICA) and IA-2β (also known as phogrin, and, like IA-2, also a tyrosine phosphatase-like molecule).

“We have data to suggest that approximately one in 300 to one in 400 individuals in the United States carry multiple islet autoantibodies,” says Dr. Eisenbarth.

ZnT8 is the newest kid on the block. A study published in the Proceedings of the National Academy of Sciences (Wenzlau JM, et al. 2007;104:17040–17045) showed ZnT8 was targeted by autoantibodies in 60 to 80 percent of new onset type 1 diabetes, compared with less than two percent of controls. The combined measurement of ZnT8 along with the other known autoantibodies—GAD65, IA-2, and IAA—raised autoimmunity detection rates to 98 percent at disease onset in this study, yet another promising finding that researchers hope to validate.

“The study is intriguing,” says Dr. Molinaro, noting his own interest in the potential of ZnT8 before adding, “We really don’t want to rush into anything before we know the validity of the finding.”

The four antibodies are not the final say in the matter—other targets are seen by the immune response as well, says Dr. Nepom. “But there’s a sort of diminishing returns,” he says. ZnT8 was discovered approximately four years ago; when it was put into use, sensitivity increased, but not specificity. “That’s probably going to be true if there’s a fifth or sixth marker that comes along as well. So the clinical benefit gets smaller and smaller as we go forward.”

The goal now is to distinguish between the 1.6 percent of people who have these markers, and the 0.4 percent (a conservative estimate, Dr. Nepom says) of those who develop type 1 diabetes. T cell response may be the key. In those who do not progress (“The ones who surprise the researchers,” as Dr. Nepom puts it ), the T cells are able to counterregulatean ability missing in the T cells of those who do progress to diabetes.

No related biomarkers are ready for use in the clinical lab setting, Dr. Nepom says. “You pick up any immunology of diabetes journal, or go to the immunology of diabetes meetings, and there’s a good chunk of work focused exactly on that question. But there’s nothing quite ready yet,” he says.

Immune markers have primarily centered on indirect measurements of T cell responses, by detecting the presence of circulating serum islet autoantibodies, says Dr. Molinaro. This creates some critical goals for type 1 diabetes researchers: identifying, quantifying, and characterizing T cells reactive within the islet autoantigens, GAD65, and IA-2.

“In the Immunology of Diabetes Society’s first workshop on autoreactive T cells, the quality of recombinant preparations of these autoantigens was identified as a significant weakness,” says Dr. Molinaro. That, he says, was likely the reason for the inconsistency in published studies of peripheral blood T cell reactivity to islet autoantigens. “Poor antigen quality has also hampered the development of novel technologies for the detection of islet reactive T cells.” However, he continues, T cell assays recently have been published to measure direct, reactive T cell responses in diabetic patients, though more studies will be needed to draw reliable conclusions.

Dr. Nepom’s lab has developed biomarkers called Class II tetramers, human soluble MHC peptides, to track antigen-specific T cells. “We work on detecting and phenotyping the T cell response by flow cytometry, using the tetramers to see if we can make that distinction between those who progress and those who don’t,” he says. Though he and his colleagues have published several papers on this approach, he’s not giving it the hard sell. “It’s not ready to be a clinical test,” he says.

That’s true for any of these tests, pending further investigation and new recommendations. That doesn’t mean clinicians aren’t already using them.

At Dr. Molinaro’s institution, diabetologists will order autoantibody tests (which the lab sends out) to identify a subset of adults initially thought to have type 2 diabetes, but who have islet cell autoantibody markers of type 1 and progress to insulin dependency. “They can screen nondiabetic family members who wish to donate a kidney for transplantation,” Dr. Molinaro says. Clinicians there also use the autoantibodies to distinguish type 1 from type 2 in children to institute insulin therapy at the time of diagnosis. He then repeats his cautionary words: “Now, these are not recommended for screening.” But, he adds, these reasons parallel the arguments for screening in the “Guidelines and Recommendations for Laboratory Analysis in the Diagnosis and Management of Diabetes Mellitus: Update” from NACB—the Academy of AACC.

In Dr. Molinaro’s view, it might be more useful to screen patients who are at increased risk for type 2, because specific lifestyle changes could be targeted to delay the disease.

William Winter, MD, professor, Department of Pathology, Immunology and Laboratory Medicine, and professor, Department of Pediatrics, University of Florida College of Medicine, Gaines­ville, offers another window into how autoantibodies might fit into the clinical picture. (The University of Florida’s ICA core lab plays two major roles in TrialNet: First, it accessions and aliquots samples, which are then sent to Dr. Eisenbarth’s lab in Denver for the biochemical autoantibody testing. It also provides the testing for islet cell cytoplasmic autoantibodies, done by indirect immunofluorescence.) The university has a commercial laboratory enterprise, the Diagnostic Reference Laboratory, which offers to practicing physicians patient testing for the islet cell cytoplasmic autoantibodies, GAD autoantibodies, IA-2 autoantibodies, and insulin autoantibodies, in addition to other endocrine autoantibodies.

If a patient doesn’t clearly fit into the type 1 or type 2 categories, and a physician were to ask about additional diagnostic testing, “probably the most widely available and cheapest would be to run an islet autoantibody panel,” says Dr. Winter. If the panel were to come back positive, even in someone with mild diabetes, that would be considered to be pathognomonic for autoimmune type 1 diabetes.

That’s important, Dr. Winter says, because some studies suggest that aggressive treatment of type 1 diabetes will preserve whatever beta cell function that individual has, making the diabetes easier to manage for a longer period of time. “The payoff is, tight control from the beginning can improve at least your short-term outcome.” He argues that it might also be possible to extrapolate further: that longer-term diabetic control (resulting from improved short-term control) would mean fewer microvascular and possibly macrovascular complications.

Negative antibody tests can point physicians in other helpful directions. If an autoantibody panel is negative in a person with undifferentiated diabetes, then physicians need to consider other forms of the disease, such as MODY (maturity onset diabetes of the young), although Dr. Winter does emphasize that most cases of diabetes are type 1 or type 2, and that the majority of cases usually can be adequately distinguished through the tools physicians have had for years: history, physical examination, and basic laboratory testing.

About 15 percent of children with new-onset diabetes exhibit none of the four antibodies used in the research protocols, a subset of whom have monogenic forms of diabetes, says Dr. Eisenbarth. Some are better treated with oral agents rather than insulin. Other monogenic forms, such as Wolframs/DIDMOAD syndrome, have a terrible prognosis. “It’s a multisystem disease, with blindness and deafness and many other things that develop over time,” Dr. Eisenbarth says.

Physicians need to bear in mind an important caveat with antibody testing. Dr. Molinaro explains: “IAA testing, as a marker of autoimmune type 1 diabetes, must be performed before insulin injections are begun.” Injections of human- or animal-derived insulin can stimulate autoantibodies to insulin, giving a positive result; unfortunately, he says, the assays cannot distinguish between autoantibodies that form in response to insulin injections versus endogenous insulin. Dr. Molinaro, who also serves as co-director of Emory’s Clinical Translational Research Laboratory, says he is taking advantage of multiple reaction monitoring (MRM) capabilities to develop an LC/MS-MS assay that can specifically identify and quantitate insulin of various recombinant formulations, as well as in vivo origin, which may prove useful in some clinical situations.

Jerry P. Palmer, MD, of the VA Puget Sound Health Care System, Seattle, injects several other words of caution. Physicians must be careful not to confuse autoimmunity, which is part of normal biology, with autoimmune disease. “It’s not as simple as positive and negative for antibodies, which is how it’s sometimes portrayed,” says Dr. Palmer, director, Division of Endocrinology, Metabolism, and Nutrition, VA Puget Sound Health Care System; director, Diabetes Endocrinology Research Center, and professor of medicine, University of Washington. That’s why high-sensitivity assays will be crucial in any future testing efforts. “One has to be sophisticated to look at the markers to distinguish this.”

He draws attention to a recent paper (Oak S, et al. Proc. Natl Acad Sci. 2008;105:5471–5476) that says with GAD, it’s the lack of antiidiotypic antibodies (anti-Ids), and not the presence of GAD antibodies, that’s crucial to defining type 1 diabetes. This idea appears to be gaining traction, Dr. Palmer says. “Everybody thinks it’s just a matter of GAD antibodies. But it’s probably nowhere near as simple as that,” Dr. Palmer says. “And right now it looks like anti-Id is key.

“But we get fooled once in a while,” he adds.

As researchers explore the role of biomarkers and learn more about the underlying nature of diabetes, seemingly firm ground occasionally gives way. Though it’s hardly the Berlin Wall coming down, the divide between type 1 and type 2 diabetes appears to be crumbling a bit.

Take insulin resistance, which (along with inadequate insulin secretion) is essential for standard type 2 diabetes. In type 1 diabetes, it’s also clear that insulin resistance can accelerate the progression of the autoimmune attack on the islets, possibly through a feedback loop that asks the beta cells to produce more insulin. When the beta cell gets that signal, and it’s already under immune attack, it perhaps becomes sensitive to cell death and becomes a better target, or is less able to heal itself, says Dr. Nepom “So it’s quite likely that insulin resistance is an overlap area between type 1 and type 2, but for different reasons.”

Inflammation is another area of overlap. Dr. Nepom uses the word generically to mean the immune response, which is clearly key to type 1 diabetes but is also a component of type 2. A dramatic bit of evidence for this was published in the New England Journal of Medicine (Larsen CM, et al. 2007;356:1517–1526), with a study showing that an anti-inflammatory drug called anakinra (a recombinant human interleukin-1–receptor antagonist) was of benefit in type 2 diabetes, improving beta cell function. This has been seen as a clear sign that inflammation is a component of type 2 diabetes—probably, Dr. Nepom suggests, in the context of adipose cells that make the cytokine IL-6, which is an accelerant for systemic inflammatory response.

“That’s an important overlap,” says Dr. Nepom. “You can even find about five percent of type 2 diabetics who actually have the antibody markers that are indicators of type 1.”

These carrefours have drawn the interest of Dr. Palmer. He and his colleagues use a cellular immunoblotting assay they developed, originally to identify key antigens in type 1 diabetes, to test patients with phenotypic type 2 diabetes. They initially studied people who were antibody positive, a population known as LADA (latent autoimmune diabetes in adults). The vast majority of these folks also had T cell responses, says Dr. Palmer. He and his colleagues are now looking at whether T cell positivity is related to a more rapid decline in beta cell function. “It looks like it is,” he says.

Dr. Palmer and his colleagues hope to be funded for a longitudinal study to explore further this T cell reactivity and address a number of questions. Once it’s positive, does it remain positive, until all the beta cells are destroyed? Or does it come and go? How is it related to beta cell function?

“We’re moving away from classic type 1 diabetes into what looks like phenotypic type 2 diabetes, and we think autoimmunity explains part of it,” says Dr. Palmer.

Interestingly, there’s almost no overlap in the susceptibility genes, says Dr. Nepom. “That’s giving everybody some pause. Because the field was really moving along toward these overlapping areas of pathogenic interest. But if the genes don’t overlap, then maybe what they are is some common pathway involving beta cell death, or something, rather than some mechanistic commonality.”

As compelling as their research is, experts are just as keen to see how their work will translate to clinical practice.

One issue, says Dr. Nepom, is to decide the format of the screening tests. Will this be done primarily in reference labs, or will it be point of use?

“You could make an argument that it could be cost-effective either way,” says Dr. Nepom. “If it’s point of use, the question is, Can we educate all the practitioners and labs into this fairly complex combination of genetics and antibodies?”

Dr. Eisenbarth sees the need for a point-of-service assay if this rolls out as a fairly wide-scale screening. In this scenario, a POC assay for diabetes, celiac disease, and related diseases could be done in pediatricians’ offices, for example; followup testing would use more refined assays.

A second big issue, says Dr. Nepom, involves the obligations that accompany the testing. Do we need a formal counseling system? he asks. Is it sufficient to have guidelines that can be translated for patients? Or do test results need to come with a 16-page explanation of risks and statistics? These questions aren’t unique to diabetes—they jog alongside all other kinds of genetic and predictive testing. “But in heart disease, they get their lipid profile and their internist or GP talks to them about cardiovascular risk without any of that. Could we do the same in diabetes, or do we have to go to the other extreme, like for BRCA1 and 2 testing?”

What’s his preference? “We want to give them accurate information at a basic level, but not go overboard. The practitioner can communicate that—if the people doing the test provide the practitioner with accurate, simple information.”

Labs will need to do what they’ve always needed to do, says Dr. Winter: interpret results and talk to referring physicians about their meaning, significance, and implications. If, for example, the lab is dealing primarily with nonendocrinologists, and a test is positive for, say, islet cell cytoplasmic autoantibodies in someone who appears to have type 2 diabetes, “you have to have somebody with enough expertise to counsel the physician that they might be dealing with LADA.” Or (to take another example), if an autoantibody panel is positive only for IAA, well, that clearly is a marker of autoimmune diabetes. But there are other conditions associated with the presence of insulin autoantibodies besides type 1 diabetes, such as autoimmune thyroiditis—a point labs may need to explain to clinical colleagues.

“It’s not just about giving a number,” says Dr. Winter. “It’s about providing information to help the clinician manage the patient.”

More broadly, will this be a public health measure? “Sort of like PKU testing at birth?” Dr. Nepom asks. “Is this really for everybody? Or should it be done in stages, where we first deal with people who have a first-degree relative with diabetes?” Some 10 percent of the newly diagnosed children report a family history of a first-degree relative with diabetes. “So you can see it would be a big cost savings to do this initially in people who do have a relative. But you’d miss a lot of kids with diabetes who could probably be helped,” Dr. Nepom says.

Or, he continues, “We could try to roll this out in some massive, universal way. This is a huge discussion that’s occurring right now,” he says. In some states, public health departments are considering adding the genetic component of diabetes testing to their Guthrie cards.

Would the two-step testing used now in research protocols make its way to clinical testing? It’s an interesting question, says Dr. Molinaro. NACB practice guidelines say routine measurement of genetic markers is not of value for diagnosing or managing patients with type 1 diabetes. But evidence is being considered in the updated draft, he says. He adds that genetic markers may eventually prove useful for classifying diabetes in neonates, as well as in young patients with a dominant family history of diabetes.

“If you’re going to do newborn screening, you have to do genetics,” concurs Dr. Eisenbarth, because the antibodies emerge over time and are generally not present at birth. “But if we were screening a population and we find antibodies,” he continues, “then we often do a genetic test to look for HLA alleles, because they change the probability of progressing.”

Dr. Molinaro offers a few ideas about where clinical testing might take place, noting that it will depend primarily, of course, on the hospital practice and clinical need. “However, I would predict that reference labs would be performing these tests if they were to become routine in the near future, because of the specialized nature of some of these genetic and autoantibody testing methodologies.” Some autoantibody methods are radioisotopic, not an attractive option for hospital laboratories. And the ELISAs that have been developed, he says, are of limited use in hospital settings. Any autoantibody assay that is used, he says, should have a specificity greater than 99 percent.

Dr. Eisenbarth’s lab primarily uses fluid phase radioassays. “There’s at least one excellent GAD ELISA, but it’s not a standard ELISA. Simply putting the target on a solid surface and looking for antibody binding is not sufficient to get the specificity we need.”

He echoes Dr. Molinaro’s call for high specificity. “We can get the kinds of specificities we need—99 percent—with competition assays. But a lot of the kit assays for autoimmune diseases that give five or 10 percent false-positive would be just about useless. They’ve settled for convenience,” he says. “For disease prediction, we just can’t use those assays, and I don’t think we should accept them,” says Dr. Eisenbarth, who’s written on the subject. (See Liu E, Eisenbarth GS. Accepting clocks that tell time poorly: fluid phase versus standard ELISA autoantibody assays. Clin Immunol. 2007;125:120–126.)

Standardization methods are underway, put forward by the Diabetes Autoantibody Standardization Program, a proficiency testing program organized by the CDC under the support of the Immunology of Diabetes Society. Dr. Molinaro also reports that commercially available autoantibody methods exist for GAD65 and IA2A, which use WHO standards; those vendors, he says, also participate in the standardization program.

Says Dr. Winter: “If you’re going to do this testing, you should participate in some sort of international proficiency or international assessment,” such as DASP.

All of this, should it come to pass, will not come cheaply. As Dr. Molinaro points out, “The economic cost of diabetes is astounding. So how is treatment or insurance going to be affected based on the results of these screenings?” Answering such questions may have health care providers following the money—as well as autoantibodies and genetic markers—after all.