Traditionally, insulin resistance is thought to be the precursor to many metabolic diseases. It is now believed that compensatory hyperinsulinaemia, previously thought to be a symptom of insulin resistance, may independently associated with metabolic disease and have its own pathological implications. Further understanding of compensatory hyperinsulinaemia may offer new insights into the aetiology of metabolic disease.

This thesis provides novel work in hyperinsulinaemia and is broadly divided into four parts. Part 1 comprises a collation of the literature to show the aetiology and pathologies of hyperinsulinaemia, and to critically review the current diagnostic methods. The aetiology of hyperinsulinaemia is not yet fully elucidated, but is likely to include excessive carbohydrate ingestion, excessive cortisol or uric acid production, and/or medications. Subsequent pathologies include: cardio-, cerebro-, and peripheralvascular disorders; type 2 diabetes; inflammation; and certain cancers or dementias. This is the first review to comprehensively link hyperinsulinaemia to such a wide range of metabolic disorders. Except for fasting insulin levels being considered unreliable, there was no consensus regarding diagnostic criteria. This means that diagnostic criteria needs to be determined prior to further research.

Part 2 examined the prevalence of hyperinsulinaemia in the Kraft database. This important database comprises a large sample of oral glucose tolerance tests with insulin assays collected over 20 years in Chicago, USA. From the 15 000 available tests, those involving men aged ≥ 20 years and women ≥ 45 years, with a BMI > 18kg/m² were included (n = 7750). Participants were stratified according to glucose response (WHO criteria) and insulin response according to the Kraft (2014) response patterns. The results showed that > 90% of people with diabetes or impaired glucose tolerance were hyperinsulinaemic. Of those with normal glucose tolerance (n = 4030), approximately 75% were hyperinsulinaemic. This had a limited association with obesity. As this is the first time a cohort of people with normal glucose tolerance have had their insulin response patterns analysed, these results show that there may be high prevalence of hyperinsulinaemia in the wider community. As this was not associated with obesity, this implies that hyperinsulinaemia is a silent disease. Together with the implications of the potential pathologies resulting from hyperinsulinaemia, there is a need for a robust diagnostic test. These results are important because it is the first time the potential impact of hyperinsulinaemia in the wider community has been investigated.

Part 3 investigates which test(s) could best be used for diagnosing hyperinsulinaemia. Chapter 4 investigates whether the existing insulin resistance tests, which include the homeostasis model assessment variants (HOMA2 %B, %S, and IR), and the oral glucose insulin sensitivity (OGIS) have sufficient test-retest reliability to be considered as a potential diagnostic test. Using the methods of Bland and Altman, the test-retest reliability was calculated as Test 1≈ Test 2 ± repeatability coefficient, while the repeatability coefficients were derived from the square root of the residual mean square errors from one-way analyses of variance. This is the first time repeatability coefficients have been calculated for these variables with potentially higher practical utility compared to coefficient of variation. The results showed that the repeatability coefficients for the HOMA2 %B, %S, and IR variants were 72.91, 189.75, and 0.9, which equated to 89%, 135%, and 89% of their respective grand means. OGIS had a repeatability coefficient of 87.13 which equated to 21% of the grand mean. These finding are important as they demonstrate that dynamic measures should be preferred to fasting measures when assessing either insulin resistance or hyperinsulinaemia. These results also question the validity of the widespread use of HOMA. There was no test-retest repeatability data for either the Kraft or Hayashi insulin response patterns and a limited amount for the McAuley Index, another measure of insulin resistance. Therefore, Chapter 5 reports on three-hour, 100 g, oral glucose tolerance tests with insulin assays that were conducted four times on six healthy individuals at weekly intervals. Test-retest repeatability assessments were conducted as according to the methods previously described for measures of insulin resistance (HOMA2 variants, OGIS, and McAuley Index), while Fleiss’ kappa was applied to Kraft and Hayashi dynamic insulin response patterns. The results showed that Kraft patterns had a higher repeatability compared to Hayashi patterns based on a combination of Fleiss’ kappa (0.290 vs 0.186,) p-value (0.15 vs 0.798) and 95% confidence intervals. OGIS and McAuley index recorded a lower CV compared to HOMA2 variables. However, the McAuley index was unable to distinguish between people with normal or a hyperinsulinaemic response, suggesting a low-overall sensitivity.

These results show that a dynamic insulin response following an oral glucose load is needed to effectively diagnose hyperinsulinaemia. However, this requires a minimum of three blood tests over a minimum of three hours test duration. In Chapter 6, the Kraft database was re-examined to determine whether a simplified diagnostic algorithm could be derived. In people with normal glucose tolerance and fasting plasma insulin < 30 µU/mL, sensitivity and specificity calculations showed that hyperinsulinaemia can be diagnosed by 2-hr plasma insulin (> 30 µU/mL sensitivity/specificity = 0.98/0.62; > 50 µU/mL sensitivity/specificity = 0.79/0.99). Given that first-line treatment for hyperinsulinaemia is lifestyle management, the lower level of > 30 µU/mL was recommended as the new diagnostic criteria for hyperinsulinaemia.

The fourth part of this thesis reviewed potential treatment options for hyperinsulinaemia including pharmacotherapy, physical activity and diet. Hyperinsulinaemia cannot be managed without concurrent management of glycaemia; thus limiting pharmaceutical agents. Physical activity, especially high intensity interval training in combination with resistance training, and dietary management, especially carbohydrate restriction, appear to offer the most promise, and may even work synergistically.

Overall, this thesis represents new knowledge in examining hyperinsulinaemia; from aetiology to management. Hyperinsulinaemia contributes to a significant number of metabolic diseases, including cancer and dementia. Hyperinsulinaemia affects almost every person with a glucose tolerance disorder and many people with normal glucose tolerance; but is not associated with obesity. Repeatability coefficient testing determined that dynamic measures should be preferred over fasting for both insulin resistance and hyperinsulinaemia. Following a 100 g, oral glucose tolerance test, 2-hr insulin > 30 µU/mL is diagnostic for hyperinsulinaemia. Public health clinicians and researchers can build on these foundations to determine further means of stemming the tide against metabolic disease.