Type 2 diabetes – biomarker models promise new means to predict risk
Considerable rewards could be obtained from early identification of Type 2 diabetes mellitus (T2DM). One of the most obvious, as suggested in a recent report on diabetes’ global burden, would be better disease management. The report, by the University of East Anglia in the UK, concludes that “early investments into prevention and disease management may therefore be particularly worthwhile.”
Such perspectives are strengthened by evidence that the onset of T2DM can be delayed by behaviour modification. A study in the ‘British Medical Journal’ in 2007 noted that lifestyle changes could be “at least as effective as drug treatment” in slowing the onset of diabetes. It concluded that the only barrier to the effectiveness of such a strategy was to identify diabetes quickly enough.
Much is now known about the risk factors associated with T2DM such as parental history, age, body mass index and elevated blood glucose levels. Combining these with measurable indicators of metabolic syndrome – high blood pressure, LDL and HDL cholesterol and excess triglyceride – can result in a credible degree of prediction. However, there are several barriers to the process.
Fasting glucose and oral glucose tolerance
The typical method for assessing T2DM risk is to measure fasting plasma glucose (FPG). However, the test’s specificity is poor. Two decades ago, the so-called Hoorn study at Amsterdam warned about significant levels of variation in blood glucose levels. Although many individuals are identified as having impaired fasting glucose (IFG), their absolute risk of conversion to diabetes is a mere 5 to 10% per year.
Over this period, differences have also emerged about how best to measure glucose. In the year 2000, while some experts (including the American Diabetes Association) recommended the use of fasting plasma glucose (FPG) alone, others noted that many diabetic subjects would have been classified as non-diabetic on the FPG test. As a result, they recommended use of the two-hour oral glucose tolerance test (OGTT). Nevertheless, in spite of its greater accuracy, OGTT is rarely used since it requires two hours to perform and is an unpleasant experience for the patient.
Glucose tolerance only one risk indicator
The above factors have provoked a search for new approaches to predict T2DM. Some beliefs about OGTT have been brought into question, too. In 2002, clinical epidemiologists at the University of Texas Health Center in San Antonio published the results of a prospective cohort study to identify people at high risk of T2DM.
The results were unequivocal. Impaired glucose tolerance was only one indicator of risk. Persons at high risk for T2DM, the study concluded, were “better identified by using a simple prediction model than by relying exclusively on the results of a 2-hour oral glucose tolerance test.”
Subsequent years have been witness to significant efforts to develop and refine predictive models for T2DM. However, five years after the San Antonio study, the choices are still less than wholly clear.
In 2007, the Framingham Offspring study in the US estimated seven-year T2DM risk based on a pyramid of metrics consisting – at the base – of age, sex, parental history and body mass index. This was followed by the inclusion of simple clinical measurements on metabolic syndrome traits, and thereafter, the 2-hour post-oral glucose tolerance test, fasting insulin and C-reactive protein levels. At its most complex, the model used the Gutt insulin sensitivity index or a homoeostasis model of insulin resistance.
For proponents of new alternatives to impaired glucose tolerance, the conclusions of the Framingham study were stark. Complex clinical models, it stated, were not superior to the simple one, and in spite of the definite existence of T2DM prediction rules, “we lack consensus for the most effective approach.”
The limitations of biotech
More recently, investigations at the frontiers of biotech have also faced challenges to clear-cut answers. Although it is clear that multiple genetic loci are associated with the risk of T2DM, researchers have not managed to connect the genetics underlying a family history of diabetes with predictability.
In 2008, researchers at Harvard/Massachusetts General and Emory University published results of a study on 18 single-nucleotide polymorphisms (SNPs) known to have associations with the risk of T2DM, to predict new cases in a large, prospectively examined, community-based cohort. However, the outcome, in terms of risk prediction, was less than encouraging. In reality, it proved to be only slightly better at making a prediction than did traditional risk factors on their own. The authors concluded: “Our findings underscore the view that identification of adverse phenotypic characteristics remains the cornerstone of approaches to predicting the risk of type 2 diabetes.”
Adiponectin and ferritin
Meanwhile, the effort to identify and validate alternate biomarkers for prediction and screening continue. Two especially promising ones appear to be adiponectin, an adipocyte-derived, insulin-sensitizing peptide, and ferritin, a protein that binds to iron and accounts for most of the iron stored in the body.
Studies in the early 2000s in the US and Germany confirmed that adiponectin was independently associated with a reduced risk of type 2 diabetes.
Interest in this area goes back a long time, to a cross-sectional and longitudinal study of Arizona’s Pima Indians, who have the world’s highest reported prevalence and incidence of non-insulin-dependent diabetes mellitus (NIDDM). The study dates to the early 1980s when it sought to document the sequence of metabolic events occurring with “the transition from normal to impaired glucose tolerance and then to diabetes.”
In 2004, a prospective study within the US Nurses’ Health Study investigated iron storage, given a belief that T2DM was a manifestation of hemochromatosis, due to iron overload. Researchers have established that higher iron store (reflected by an elevated ferritin concentration and a lower ratio of transferrin receptors to ferritin) is associated with increased T2DM risk in healthy women, independent of known diabetes risk factors.
However, there still are reasons for caution. In July 2014, or more than a decade after the US Nurses’ Health Study, a meta-analysis of T2DM risk and ferritin in the journal ‘Diabetes/Metabolism Research and Reviews’ warned that though evidence suggested a causal link, “publication bias and unmeasured confounding cannot be excluded.”
Nevertheless, ferritin and adiponectin do appear to play a key role in predicting T2DM when combined with other selected biomarkers.
The Danish model
One predictive model that has emerged in Denmark selected a panel of six biomarkers out of a total of 64, to assess T2DM risk. The selected biomarkers include adiponectin and ferritin, as well as four of their more common counterparts: glucose and insulin, as well as the inflammation markers C-reactive protein (CRP) and interleukin-2 receptor A (IL2RA).
The model was developed by a research team from Copenhagen’s Glostrup Hospital and Steno Diabetes Centre, along with the Copenhagen and Aarhus universities, and Tethys Bioscience of the US.
The researchers used the so-called Inter99 cohort, a study of about 6,600 Danes with the primary outcome of 5-year conversion to T2DM, to select 160 individuals who developed T2DM and 472 who did not. They carefully measured several clinical variables and candidate biomarkers from a multitude of diabetes-associated pathways, using an ultrasensitive immunoassay microsample molecular counting technology.
Their effort ultimately led to six biomarkers that gave a Diabetes Risk Score. This, they concluded in a July 2009 issue of ‘Diabetes Care’, provided “an objective and quantitative estimate of the 5-year risk of developing type 2 diabetes, performs better than single risk indicators and a noninvasive clinical model, and provides better stratification than fasting plasma glucose alone.”
The researchers who developed the Danish Diabetes Risk Score are modest in their claims. In an appendix to their report in ‘Diabetes Care’, they point out that their selection process for biomarkers may not have identified the best possible model, but do state that they identified a ‘good’ model.
Some outside observers are however less circumspect, given what many acknowledge to be one of the most exhaustive and profound selection efforts to date. James Meigs of Harvard Medical School calls the Danish Diabetes Risk Score “the most robust multimarker prediction model possible.”
Beyond Europeans to Chinese
One of the only major caveats in the Danish effort consisted of demographics. The report on the Danish model in ‘Diabetes Care’ noted that it “may only apply to white Northern Europeans enrolled in a lifestyle intervention trial” and that it was an open question whether the model “would produce the same biomarkers or discriminate well in race/ethnicity populations that are differentially affected by diabetes.”
Answers to these are still emerging. In 2013, a study on 2,198 community-living Chinese by the Shanghai Institutes for Biological Sciences endorsed the use of ferritin as a biomarker. Though the focus of the research was on iron storage, two of three other biomarkers used in the effort were the same as those in the Danish study, namely adiponectin and CRP (the fourth was γ-glutamyltransferase).
Biomarker search continues
Meanwhile, the search for TD2M biomarkers continues.
Two endothelial dysfunction biomarkers being investigated for T2DM risks consist of E-selectin and ICAM-1. The US Nurses Health Study mentioned above also found that significantly elevated levels of the latter predicted incident diabetes in women independent of traditional risk factors such as BMI, family history, diet and activity. In addition, adjustment for baseline levels of C-reactive protein, fasting insulin, and hemoglobin A (1c) did not alter these associations.
Incretins and melatonin
Incretins, metabolic hormones which lower blood glucose by causing an increase in insulin after eating, are another potentially significant biomarker. An ‘incretin effect’ is associated with the fact that oral glucose elicits a higher insulin response than does intravenous glucose. There are two hormones responsible for the incretin effect: glucose-dependent insulinotropic hormone (GIP) and glucagon-like peptide-1 (GLP-1).
In patients with type 2 diabetes, the incretin effect is reduced. In addition, about half first-degree relatives of patients with T2DM show reduced responses toward GIP, without any significant change in GIP or GLP-1 secretion after oral glucose. To some researchers, this opens the possibility that a reduced responsiveness to GIP is an early step in the pathogenesis of type 2 diabetes.
Variation in the Circadian system has also drawn a great deal of attention.
Reverse transcription polymerase chain reaction (RT-PCR) analyses, led by a team at the University of Lille in France, investigated melatonin receptor 2 (MT2 transcripts) in neural tissues and MT2 expression in human pancreatic islets and beta cells. Their findings suggest a link between circadian rhythm regulation and glucose homoeostasis through the melatonin signalling pathway.