Fasting C‐peptide at type 2 diabetes diagnosis is an independent risk factor for total and cancer mortality

We assessed the association between insulin resistance and blood glucose concentrations at type 2 diabetes diagnosis and future development of diabetes‐related complications and mortality.


| INTRODUCTION
The prevalence of diabetes is increasing worldwide, and mortality is approximately doubled among individuals with type 2 diabetes compared to those without it. 1,2 Moreover, type 2 diabetes leads to a number of comorbidities, such as coronary heart disease, stroke, retinopathy, nephropathy, and neuropathies. Type 2 diabetes is often clustered with other cardiovascular risk factors, including hypertension and dyslipidemia. 3 Therefore, treatment for this condition aims not only at reducing blood glucose but also at mitigating these risk factors. Treating hypertension and dyslipidemia undoubtedly decreases the risk for cardiovascular events and cardiovascular death in individuals with type 2 diabetes. [4][5][6] However, for blood glucose-lowering therapy, a positive treatment effect in terms of cardiovascular event and mortality prevention is not guaranteed. In those with newly diagnosed type 2 diabetes, glucose-lowering therapy may decrease myocardial infarction and death risk after 12 years of follow-up at the earliest, and many more years may be required to yield a positive effect. 7 In contrast, with a median diabetes duration of 10 years, strict glucose lowering is linked to increased mortality after a mean follow-up of 3.5 years. 8 In the past decade, new drug classes for the treatment of type 2 diabetes have emerged. In large placebo-controlled randomised trials, sodium glucose cotransporter 2 inhibitors and glucagon-like peptide 1 analogues have been associated with decreased total and cardiovascular mortality. 9,10 However, the mean HbA1c improvement in those trials was less than 0.5% (5 mmol/mol). Thus, lowering blood glucose may not be the most promising point of attack for decreasing mortality in type 2 diabetes. C-peptide is usually interpreted as a measure of beta cell function. However, fasting C-peptide is not a good estimate for this endpoint and should ideally be measured after a carbohydrate load. 11 In contrast, fasting C-peptide and fasting insulin are more suitable measurements of insulin resistance. 12 For these reasons, we interpret fasting C-peptide as a measure of insulin resistance.
In the pathophysiology of type 2 diabetes, blood glucose rises because of insulin resistance and insufficient beta cell function. 13 Insulin resistance per se has been proposed as a marker of risk for diabetes complications and mortality. Ahlqvist and colleagues proposed a classification of adult-onset diabetes into five subgroups based on age at diagnosis, BMI, and insulin resistance estimated with the homoeostatic model assessment (HOMA). 14 After a follow-up of 11 years, they found that those with insulin resistance had the highest risk for diabetic kidney disease. Moreover, those with higher levels of insulin resistance had a higher risk of coronary events and stroke compared to those with lower levels of insulin resistance, although not in the fully adjusted model. Similarly, Pikkemaat et al.
found that insulin resistance measured as fasting C-peptide levels at diabetes diagnosis was an independent risk factor for all-cause and cardiovascular mortality. 15 Moreover, in a study of people with insulin-treated type 2 diabetes, people with diabetic kidney disease had higher fasting C-peptide levels than those without this diabetes complication. 16 However, the results of the United Kingdom Prospective Diabetes Study (UKPDS) showed no association between insulin resistance at diabetes diagnosis and later cardiovascular disease. 17 A possible explanation for these mixed findings is that in UKPDS, blood samples for measuring insulin resistance were obtained after 3 months of dietary intervention, which may have affected the values. Overall, evidence is conflicting regarding the impact of insulin resistance at diabetes diagnosis on later diabetes complications and mortality.
To clarify the association, we conducted this retrospective population-based study, which to our knowledge is the largest such study addressing this question and measuring fasting C-peptide levels at type 2 diabetes diagnosis. We hypothesised that fasting C-peptide levels at type 2 diabetes diagnosis would be associated with total, cancer-related, and cardiovascular-related mortality and coronary heart disease independently of other risk factors.

| Study population
The diabetes registry DiabNorth includes inhabitants of the counties of Västerbotten and Norrbotten in Sweden who are diagnosed with diabetes and participate in either the Västerbotten Intervention Programme (VIP) or in the multinational MONitoring of trends and determinants in CArdiovascular disease (MONICA) study. 18 We carried out a retrospective cohort study of DiabNorth participants diagnosed with type 2 diabetes who had C-peptide measured at the time of diagnosis or up to 5 years prior. All identified patients with diabetes who were still alive were asked to participate in the DiabNorth registry, and 74% provided written informed consent. 18  HbA1c measurements were available from the year of diabetes diagnosis for 727 patients.

| Measurements
Systolic and diastolic blood pressure were measured with the patient either in the sitting position or lying down. Measurements obtained with the patient lying down were adjusted using sex-and age-specific formulae. 19 Hypertension was defined as systolic blood pressure ≥140 mmHg and/or diastolic blood pressure ≥90 mmHg and/or the use of antihypertensive medication.
In the VIP, capillary plasma glucose was analysed using a HbA1c was analysed (TOSOH G5, Tosoh, Tokyo, Japan) at the nearest hospital.   H350, H352, H356, H359, H360, E103, E113, E143; ICD9: 362A). For end-stage renal disease, we had only seven events in total during our follow-up and could not conduct a survival analysis. We created a composite of major cardiovascular events (MACE), which was defined as the first coronary heart disease event, stroke, or death from cardiovascular causes.

| Statistical analysis
Data are presented as medians with IQRs. Hazard ratios (HRs) with 95% confidence intervals (CIs) were calculated by Cox regression. Time from diabetes diagnosis was used as the survival time. Values for Cpeptide, HbA1c, and fasting glucose were transformed into Z-scores before analysis. In model 1, the analysis was adjusted for sex and age at diabetes diagnosis as continuous variables and for C-peptide laboratory measurements as two categories. In model 2, the analysis was

| RESULTS
A total of 864 individuals had fasting C-peptide available at their type 2 diabetes diagnosis and were included in this study (Figure 1). Their baseline characteristics are presented in In the univariate Cox regression, an increase by one standard deviation in fasting C-peptide at diabetes diagnosis was associated with all-cause mortality (HR 1.16; 95% CI 1.02-1.32; p = 0.022). In the fully adjusted model, this association was also significant ( Table 3). The association between C-peptide levels and all-cause mortality was independent of HbA1c levels at diabetes diagnosis (model 3, Table 3).
Moreover, HbA1c levels were associated with all-cause mortality.
Fasting C-peptide levels at diabetes diagnosis also were associated with cancer mortality in the fully adjusted model and independent of HbA1c levels (Table 3). However, in the univariate model, the association between fasting C-peptide and cancer mortality was not significant (HR 1.20; 95% CI 0.96-1.49; p = 0.110). HbA1c levels at diabetes diagnosis were not associated with cancer mortality (Table 3).
We found no association between fasting C-peptide levels at diabetes diagnosis and cardiovascular mortality, MACE, coronary heart disease events, stroke, or retinopathy (Table 3). However, HbA1c at diabetes diagnosis was associated with later development of retinopathy, stroke, and MACE in the fully adjusted model (Table 3).
We measured C-peptide at diabetes diagnosis and up to 5 years prior to this diagnosis. We therefore conducted a sensitivity analysis including only individuals who had C-peptide analysed no longer than one year before their diabetes diagnosis (n = 539). We still found an independent association between fasting C-peptide and all-cause

| DISCUSSION
In this large retrospective cohort study, a high fasting C-peptide concentration at the time of type 2 diabetes diagnosis was associated with total and cancer-related mortality independently of HbA1c and fasting blood glucose levels.
We conclude that insulin resistance at diabetes diagnosis is important for later mortality risk independent of HbA1c levels at diabetes diagnosis. Our results are in line with a recent mediation analysis showing that insulin resistance is responsible for 15% of the increased mortality risk in type 2 diabetes. 20 In other studies, insulin resistance had no impact on mortality or even was associated with decreased mortality. 21,22 In those studies, insulin resistance was measured as fasting C-peptide or HOMA at 9 and 4 years after diabetes diagnosis, respectively. Long-standing diabetes may lead to beta cell failure and lower fasting insulin levels, resulting in lower fasting C-peptide and HOMA measures despite or regardless of high insulin resistance. Thus, our study makes an important contribution because C-peptide levels were measured at diabetes diagnosis and before initiation of any diabetes medication. In line with our results, another smaller study found that individuals with high insulin resistance, based on fasting C-peptide at diabetes diagnosis, had higher total mortality than those with lower C-peptide levels. 15 Insulin resistance has previously been described as an independent risk factor for cancer incidence and cancer mortality. 23,24 The associated increase in plasma insulin levels may promote carcinogenesis by inducing tumour cell proliferation and through antiapoptotic and angiogenetic effects. 25 However, other epidemiological studies have suggested that the hyperglycemia itself increases cancer risk and cancer mortality. 26,27 In these study cohorts, no measurements for insulin resistance were available, precluding investigation of the influence of hyperglycemia independent of insulin resistance.
The effect attributed to hyperglycemia in these studies may trace in part to insulin resistance. When we controlled for insulin resistance in our study, HbA1c level at diabetes diagnosis was not an independent risk factor for cancer mortality. Based on our results, insulin resistance at diabetes diagnosis, not hyperglycemia, seems to be linked to the increased risk for cancer death.
In our study, fasting C-peptide at diabetes diagnosis was not associated with cardiovascular mortality. In line with these results, Bo and colleagues did not find increased cardiovascular mortality with higher fasting C-peptide in patients with type 2 diabetes after adjustment for hypertension and other risk factors. 21  The present study has some limitations. First, C-peptide was not estimated exactly at the time of diabetes diagnosis for all individuals, but up to 5 years before diagnosis. However, according to our sensitivity analysis, this variability did not seem to influence the results. Second, only study participants who were still alive could give informed consent, which led to a proportionally greater inclusion of deceased individuals. This imbalance may have introduced a selection bias towards patients with more cardiovascular complications. Third, we could not control for kidney function in our analysis which is important since it is known that impaired kidney function might increase in C-peptide levels.
This study also has some important strengths. To the best of our knowledge, it is the largest retrospective population-based study measuring C-peptide at or before diabetes diagnosis. Furthermore, we adjusted for important confounders, such as sex, age, smoking, hypertension, BMI, and total cholesterol. The long follow-up of 16 years and the complete follow-up of all participants and high participation rates (∼70%) in both the VIP and MONICA studies are also notable. 30 In conclusion, insulin resistance measured as fasting C-peptide at type 2 diabetes diagnosis was found to be an independent risk factor for total mortality and cancer mortality. Thus, treatment of type 2 diabetes should focus not only on normalising blood glucose levels but also on mitigating insulin resistance.