Close X

Personalising prevention for type 2 diabetes

12th October 2022

As rates of type 2 diabetes continue to rise around the world, prevention of the condition has never been more important. A review in a recent issue of Diabetologia looks at the opportunities and challenges offered by a personalised approach to type 2 diabetes prevention. Dr Susan Aldridge reports.

It is 25 years since randomised controlled trials began to report findings that showed that type 2 diabetes is a preventable condition. This caught the attention of healthcare professionals, researchers, policy makers and, most importantly, the general public. So, over the past two decades, many efforts have been made to translate these research findings into real-world strategies that will actually prevent or delay the onset of type 2 diabetes and so improve public health.

The challenges involve both the delivery of effective prevention programmes and the identification of those individuals who have most to gain from them. In most of the original diabetes prevention studies, people were recruited following a finding of impaired glucose tolerance on an oral glucose tolerance test (OGTT). But this approach is hard to translate into real-world settings, as the OGTT test isn’t really suited to large-scale screening. The HbA1c test is easier to apply and is used to identify both those with undiagnosed diabetes and those with prediabetes.

Personalised prevention

In a review article, Professor Nick Wareham of the Medical Research Council Epidemiology Unit in Cambridge discusses how we might integrate a more personalised prevention approach into current screening programmes. Looking at this in a global context, he notes that many countries are already implementing large-scale screening for prediabetes and diagnosis of type 2 diabetes itself, while others have yet to make a start on this kind of work. Where a screening programme does exist, it’s important that any moves to develop personalisation are used to augment, rather than replace, what is already happening.

One place to start when looking at personalisation is at the nature of the intervention and how this impacts the individual. In prevention programmes that build on the classical trials, like the US Diabetes Prevention Program (DPP), the focus is usually on weight reduction, increasing physical activity and dietary change. Yet there is heterogeneity among the target population for these programmes. Early data from the NHS Diabetes Prevention Programme shows that 44% of those referred were obese, 33% were overweight, but 15% had a body mass index (BMI) under 25kg/m2. How relevant and effective is a programme focused on weight loss to this last group? How does it affect their motivation to participate?

There is also evidence for heterogeneity in pharmacological approaches to prevention. With metformin, people with higher BMI were more likely to reduce their risk of type 2 diabetes than those with lower BMI. Specifically, those with BMI >35kg/m2 had a 53% risk reduction, while those with BMI between 22 and 30kg/m2 had only a 3% reduction in risk. This opens up the possibility of personalising pharmacological prevention – at least with metformin – on the basis of BMI. Metformin is currently the only drug approved for prevention but, if others are approved – as seems likely – they too should be considered for this kind of personalised approach.

It is also well known that South Asian populations tend to develop type 2 diabetes at an earlier age and lower BMI than those of European origin. However, there is currently little evidence for differential responses to lifestyle intervention on the basis of ethnicity in the trials. For instance, the US DPP had 46% of participants of non-white origin, yet found no heterogeneity in response. However, the authors of the classic paper from the trial say the study did not have the power to assess significance of effects between subgroups. In other words, failure to demonstrate any differences in response by ethnicity, age, BMI or sex in a trial does not mean these differences do not exist. Trials could be designed to bring out these individual differences.

Towards more personalised prevention

In the last 15 years, there has been a significant expansion in the number of genetic loci known to be associated with type 2 diabetes, thanks to the application of technical innovation on a large scale. While these have given us important insights into the pathophysiology of type 2 diabetes, their role in risk prediction has been limited. So, the prospect of cheap genetic screening to identify people at high risk seems some way off, at least for the whole population.

However, it might be worthwhile researching whether genetic risk scores enhance prediction in various subgroups. It is also possible that genetic factors could influence the response to a particular preventive intervention. This was looked at in the US DPP – and no difference was found in response to lifestyle intervention by genetic risk assessed by one specific risk score. But other genetic risk scores might show differential responses. Again, more research is needed, particularly given the wealth of genetic information now available for type 2 diabetes.

There is also a lot of new information on ‘omic data – epigenetic markers, proteins and metabolites. Like genetic markers, they have added to our understanding of type 2 diabetes, although at present they do not enhance the predictive ability of risk tools. Professor Wareham warns that the way forward in prevention is not to keep adding more information in the hope of improve prediction of risk, but focus on successful implementation of the basics of diagnosis, screening and prevention.

Type 2 diabetes subgroups

It is widely accepted now that type 2 diabetes is not just one condition. Targeting subgroups certainly has some application in personalising treatment and potentially in prevention too. For example, Swedish researchers have identified five subgroups of type 2 diabetes – namely – and put simply, autoimmune, insulin deficient, insulin resistant, obese and age-related. Personalised therapy might be possible within these subgroups, which are also associated with different patterns of complications.

Does this subgrouping also have value in prevention? This is challenging, for the clinical features that characterise these subgroups are only identified at diagnosis, when it is too late for preventive efforts. This limitation could be overcome if each subgroup could somehow be linked to a distinct pathophysiological pathway that could be detected at the prediabetes stage. In other words, more research is needed into linking what we know of the subtypes of type 2 and its prediabetes counterparts.      

Individualising targets for preventive action

Preventive programmes might be more effective if they were tailored more precisely towards the behavioural factors that an individual needed to address and was willing to change, rather than just focusing on weight loss and exercise.

One promising approach to personalised prevention is personalised nutrition, which could appeal to many potential participants. A landmark paper showed that an algorithm that brings together data from various sources, including blood biomarkers, diets, physical activity and the gut microbiome could predict glycaemic responses to meals. This led to a small randomised controlled trial where participants consumed a diet recommended by either the algorithm or a dietitian, based upon postprandial glucose response to meals. They were prescribed one diet designed to be beneficial in this respect and one designed to be disadvantageous. The former did seem to produce a better postprandial glucose response, although there wasn’t any significant difference in whether the recommended diet came from the algorithm or the dietitian.

In a follow-up trial involving 225 people with prediabetes, a so-called personalised postprandial targeted (PPT) diet was compared with the Mediterranean diet for six months. As before, the PPT diet was tailored to an individual’s postprandial glucose responses using a machine-learning algorithm. The PPT group had a significantly greater reduction in HbA1c and a reduction in time spent above glucose level of 7.8 mmol/l than those on the Mediterranean diet, though both groups achieved reductions in these clinically relevant measures.

The field of personalised prediction is rapidly becoming commercialised, with the PPT diet being made available from companies linked to the groups that carried out the above research. The growing interest in the field is evidenced by a recent award from the National Institutes of Health for a precision nutrition study of 10,000 people to develop algorithms that predict responses to dietary regimes. The role this technology might play in population screening and prevention programmes remains to be seen. It will be partly dependent on how the regulatory agencies judge the claims made and the research evidence that can be put forward.

Personalised versus whole-population approaches

Professor Wareham suggests that type 2 diabetes can be seen as a “public health manifestation of a societal problem”. Unhealthy diets and low levels of physical activity are not just matters of personal choice – they also arise from our food system and built environment. Interventions that target these broader, societal issues can influence behaviour in the whole population and lead to small changes in a large number of individuals. But this is not an alternative to a more personalised approach to prevention. It should be implemented alongside it.

Whatever the future of more personalised approaches to type 2 prevention, ongoing effort and investment should ensure their integration with the roll-out of established prevention programmes, Professor Wareham concludes.

To read this paper, go to: Wareham N. Personalised prevention of type 2 diabetes. Diabetologia 2 August 2022.

For more on this topic, view the following EASD e-Learning modules:

Any opinions expressed in this article are the responsibility of the EASD e-Learning Programme Director, Dr Eleanor D Kennedy.