Type 2 diabetes (T2D) results from a combination of insulin resistance and impaired ability of the pancreatic beta-cell to secrete sufficient insulin (Kahn ²⁰⁰³). Diet composition, especially the amount and type of dietary fat, is a recognised environmental risk factor for T2D, which particularly affects peripheral insulin sensitivity (Parillo and Riccardi ²⁰⁰⁴). The quality of dietary carbohydrate is also important with a higher glycaemic index (GI) diet being associated with an increased risk of T2D (Barclay et al. ²⁰⁰⁸). Genome-wide association studies (GWAS) have identified loci that show robust association with increased risk of developing T2D (Saxena et al. ²⁰⁰⁷; Sladek et al. ²⁰⁰⁷; Voight et al. ²⁰¹⁰; Zeggini et al. ²⁰⁰⁷) and with glycaemic traits (Dupuis et al. ²⁰⁰⁹). Most of these SNPs reside in or near genes that have a presumed role in pancreatic beta-cell dysfunction (Billings and Florez ²⁰¹⁰). These loci have mostly been identified using overt measures of impaired glucose tolerance or frank diabetes; however, the impact on insulin sensitivity and beta-cell function has only been assessed using proxy measures. More detailed phenotyping has the advantage of identifying early-stage defects that exist before impaired fasting glucose becomes apparent.

In this study, we examined the effect of a T2D genetic predisposition score (GPS) on beta-cell function and peripheral insulin sensitivity, assessed by intravenous glucose tolerance test (IVGTT) in a cohort of overweight non-diabetic participants at increased cardiometabolic risk (Jebb et al. ²⁰¹⁰). This was in order to determine the effect of SNPs that have been most robustly associated with T2D in GWAS on early-stage defects that can precede T2D. Furthermore, we examined whether this genetic predisposition to T2D influenced changes in insulin sensitivity and beta-cell function in response to changes in dietary fat and carbohydrate intake. This was tested in participants in a 24-week dietary intervention study to lower saturated fat intake through iso-energetic replacement with low or high GI carbohydrate or monounsaturated fat (MUFA). The objective of this analysis was to assess whether the T2D-GPS can act to modify dietary associated changes in peripheral insulin sensitivity and beta-cell function.

At baseline, we observed no association of T2D-GPS with peripheral insulin sensitivity (Si) (Fig. 1a). However, the T2D-GPS was associated with lower acute insulin secretion (4% per risk allele, P = 0.006, Fig. 1b) and with a lower disposition index (5% per risk allele, P = 0.002, Fig. 1c).

Of all the SNPs tested, only the TCF7L2 SNP rs7901695 (Table 2) showed a convincing association with both AIRg (13% lower per risk allele, P = 0.00005) and DI (11% lower per risk allele, P = 0.001), but not with Si. To examine whether the associations observed for the GPS with AIRg and DI were driven by the effect of the TCF7L2 SNP, we excluded this SNP from the GPS and tested for association again. The associations remained for AIRg (3% lower per risk allele, P < 0.05) and DI (4% lower per risk allele, P = 0.02), but the effect was diminished, and they were no longer significant when accounting for multiple testing.

In response to the 24 week dietary interventions to lower saturated fat intake, there were no effects of T2D-GPS to modify the change in Si, AIRg or DI, accounting for baseline values, in response to any of the dietary interventions (Table 3).

Of all the SNPs tested, no convincing effect was observed on the change in Si, AIRg or DI in response to any of the dietary interventions (Supplementary Table 1).

This study used detailed phenotyping from the IVGTT to characterise the effects of genetic predisposition to T2D on insulin secretion and sensitivity in non-diabetic overweight participants at increased cardiometabolic risk. Our results show that at baseline, a genetic predisposition to T2D was associated with an impairment of beta-cell function, but not with insulin sensitivity in participants on a run-in diet high in saturated fat, designed to reflect a ‘Western diet’. However, genetic predisposition to T2D did not modify changes in insulin sensitivity or beta-cell function in response to a dietary intervention to lower saturated fat intake by isoenergetic replacement with MUFA or carbohydrate.

Previous GWAS have identified genetic loci associated with T2D, the majority of which have a presumed role in beta-cell dysfunction (Billings and Florez ²⁰¹⁰). Our findings are supportive of a mechanism contributing to impaired insulin secretion but not insulin sensitivity. The DI is a more informative measure of beta-cell function than acute insulin secretion; by taking into account the insulin sensitivity it is a measure of the ability of the beta-cell to compensate for the degree of insulin sensitivity (Bergman et al. ²⁰⁰²). This measure can also be more sensitive as differences in DI can be detected before changes in acute insulin secretion are apparent (Bergman et al. ²⁰⁰²). At baseline, genetic predisposition to T2D was associated with a lower acute insulin response, but also with a lower DI, which confirmed an impaired acute insulin response for the level of insulin resistance.

The dietary modifications in the RISCK study were designed to be achievable through simple dietary modification and implementable on a population scale, but produced no overall significant effect on insulin sensitivity (Jebb et al. ²⁰¹⁰). Here we demonstrate that genetic predisposition to T2D did not moderate any effects of the dietary intervention on changes in insulin secretion or sensitivity. An effect of genetic predisposition on changes in these measures may have been evident with a more extreme intervention, particularly one that also achieved weight loss. For example, in response to a lifestyle intervention to decrease caloric intake and increase energy expenditure (TULIP, n = 1576), carriers of the rs7903146 (TCF7L2) risk allele, who had impaired glucose tolerance prior to the intervention, showed an increase in post-glucose-load insulin secretion (adjusted for change in BMI), which was not shown in non-carriers (Heni et al. ²⁰¹⁰). The same locus had also previously been associated with increased risk of progression to T2D, which was lessened by metformin therapy in the diabetes prevention programme (DPP), a larger multi-ethnic cohort at high risk of developing T2D (n = 3548) (Florez et al. ²⁰⁰⁶). Several other SNPs were also identified to have nominal interactions with metformin on incidence of T2D (n = 2994) (Jablonski et al. ²⁰¹⁰) or in treatment response (n = 3920) (Zhou et al. ²⁰¹¹). However, also in participants of the DPP (n = 2843), insulin sensitivity indices were studied using a genetic risk score (analogous to our GPS and with 23 of 34 SNPs covered in our study). Similar to the current study, they found a trend for a lower estimated insulin secretion (insulinogenic index) and oral DI (from a 2 h post-oral glucose load value) at baseline, but there was also no overt effect of GPS on change in insulin sensitivity or secretion indices following 1 year of intensive lifestyle modification or metformin treatment (Hivert et al. ²⁰¹¹). It appears that SNPs that are more strongly associated with a given trait in cross-sectional data may not be the most important SNPs in terms of change in trait, especially when improvements are seen in the trait that underlies the association.

A meta-analysis conducted by MAGIC (Meta-Analysis of Glucose- and Insulin-related traits Consortium) identified SNPs associated with increased glycaemia and insulin resistance in non-diabetic participants, using surrogate measures of beta-cell function (HOMA-B) and insulin sensitivity (HOMA-IR) in >35,000 participants (Dupuis et al. ²⁰⁰⁹). MAGIC identified more than twelve robust associations with fasting plasma glucose and beta-cell function, but only two with HOMA-IR or fasting insulin as a measure of insulin sensitivity. However, the studies included in this meta-analysis used proxy measures of insulin sensitivity and secretion, whilst in the current study, we used detailed phenotyping from the IVGTT to fully characterise insulin secretion and sensitivity by genetic predisposition to T2D, albeit in a small number of participants. Furthermore, examining these parameters before and in response to a dietary intervention showed that the effect of this genetic predisposition did not change the ability to respond to environmental changes. Only a small number of studies have examined the cumulative effect of loci most strongly associated with risk of T2D in response to an intervention or over time. The effects of a combined GPS in the DPP are discussed above (Hivert et al. ²⁰¹¹). An increased T2D risk score, composed of four alleles, has been associated with an accelerated age-related decline in beta-cell function in a longitudinal study (Haupt et al. ²⁰⁰⁹). In this case, a combined GPS from SNPs associated with 2 h post-load glucose was shown to contribute to a steeper age-related decline in glucose tolerance (Jensen et al. ²⁰¹¹).

Despite the small sample size and potential insufficient statistical power to examine effects of individual SNPs, in an exploratory analysis, we found significant associations of the TCF7L2 SNP with lower acute insulin secretion and DI, but no association between this SNP and the change in insulin secretion or DI in response to the intervention. The sensitivity analysis, in which we excluded the TCF7L2 SNP from the GPS, indicated that the accumulation of risk alleles, other than TCF7L2, still contributed to the impairment of beta-cell function, although the effect was not as strong. In participants from the DPP (n = 3548), the T2D risk-conferring allele of TCF7L2, which was associated with proxy measures of insulin secretion, but not sensitivity at baseline, also found no association with change in these measures in response to metformin or lifestyle interventions (Florez et al. ²⁰⁰⁶).

In conclusion, using detailed phenotyping from IVGTT, we demonstrated that genetic predisposition to T2D is associated with impaired beta-cell function in non-diabetic, overweight participants on a high saturated fat diet, which reflects an average ‘Western diet’. This genetic predisposition did not moderate effects of a reduction in dietary saturated fat by replacement with MUFA or carbohydrates on changes in insulin sensitivity or beta-cell function.

On behalf of the RISCK study group.

The RISCK dietary intervention trial was funded by the Food Standards Agency project number NO2031. Genetic analyses were funded by the participating centres including UK Medical Research Council (grant codes U105960389, U1052.00.014). The RISCK Study Group comprised of: Carmel Moore, Mark Chatfield, Christine Williams, Hannah Farrant, Claire Lawrence, Edel Magee, Kit Tsoi, Darren Cole, Steve Austin, Hanneke Mfuni, Kate Guberg, Anna Gent, Celia Greenberg, Caroline Stokes, Mario Siervo, Rosemary Hall, Louise Goff, Claire Howard, Namrata Dhopatkar, Bushra Siddiqui, Anne Dornhurst, Fiona Lewis, Samantha Bowen, L Chen, Robert Gray, Nuala Booth, Gary Moore, Roy Sherwood, Anthony Leeds, A Shah, G Saran, J Niehuser-Saran, JA Cockburn, Rachel Gitau, Katie Newens, Sean Lovegrove, Ana Rodriguez-Mateos, John Wright, Margaret Griffin and Nicola Harman. Special thanks to Aseel Al Saleh (Kings College London) for DNA extraction of samples. Thanks to Ashley Olson for statistical advice.