Introduction
Individuals with type 1 diabetes usually develop severe endogenous insulin deficiency, which results in high glucose variability and hypoglycaemia risk [
1‐
4]. Treatment guidelines for type 1 diabetes, therefore, incorporate early intensive strategies to minimise hypoglycaemia, including multiple daily insulin injections, carbohydrate counting and insulin pumps [
5].
Endogenous insulin deficiency is best assessed using C-peptide [
6]. In the DCCT study, intensively treated participants with type 1 diabetes with mixed meal test-stimulated C-peptide < 200 pmol/l had three times as many episodes of severe hypoglycaemia than other participants, despite having higher HbA
1c [
2,
3]. This threshold of 200 pmol/l is commonly described as identifying absolute insulin deficiency, although modern assays can measure below this range [
7,
8] and a relationship between hypoglycaemia and lower levels of C-peptide has been described [
9,
10].
We have recently demonstrated that a random non-fasting plasma C-peptide (rCP) sample is a sensitive and specific measure for mixed meal tolerance test-defined absolute insulin deficiency (stimulated C-peptide < 200 pmol/l), with the same C-peptide threshold having a sensitivity and specificity of 100% and 93%, respectively [
11]. As C-peptide is stable in whole blood collected into EDTA for > 24 h [
12], this means severe insulin deficiency can be identified from a routine non-fasting blood sample when a patient is seen in the clinic.
Severe insulin deficiency can occur in insulin-treated patients who have clinical features consistent with type 2 diabetes, but the deficiency is usually not recognised [
13‐
16]. This is likely to be an increasing problem as obesity rates increase, making clinical classification more difficult. We hypothesise that individuals with severe insulin deficiency will have high glycaemic variability and high hypoglycaemia risk whatever the underlying diabetes aetiology or classification. Identifying individuals with apparent type 2 diabetes who have developed severe insulin deficiency could assist appropriate management, including consideration of the intensive strategies to minimise hypoglycaemia traditionally used in type 1 diabetes.
We therefore aimed to determine whether rCP measurement can be used to assess risk of hypoglycaemia in insulin-treated patients with a clinical diagnosis of type 2 diabetes.
Methods
We assessed whether severe insulin deficiency defined by a low rCP is associated with high rates of hypoglycaemia and glucose variability, as assessed by continuous glucose monitoring (CGM), in patients with insulin-treated type 2 diabetes. We replicated our findings using questionnaire-based assessment of hypoglycaemia in a large population cohort of patients with insulin-treated type 2 diabetes and a cohort with type 1 diabetes as a comparison group.
CGM assessment of glucose variability and hypoglycaemia in insulin-treated patients with type 2 diabetes with low or preserved endogenous insulin secretion
Assessment of the relationship between rCP and self-reported hypoglycaemia in patients with insulin-treated diabetes
Results
Assessment of glucose variability and hypoglycaemia in insulin-treated patients with type 2 diabetes with low or preserved endogenous insulin secretion
Replication using self-reported hypoglycaemia in a population cohort
Discussion
Our results demonstrate that patients with insulin-treated type 2 diabetes but low C-peptide levels have markedly increased incidence of hypoglycaemia in comparison to those with retained C-peptide, whether measured by CGM or self-reported. On CGM, glycaemic variability (when defined by SD, the most robust measure [
19,
23‐
25]), frequency and duration of hypoglycaemia were markedly increased in those with severe insulin deficiency. More significant daytime hypoglycaemia (< 3 mmol/l, [
26]) was entirely confined to these participants. These differences occurred despite similar glycaemic control in those with and without preserved endogenous insulin secretion. Severe insulin deficiency was not uncommon in participants with insulin-treated type 2 diabetes in our cohort, occurring in 13% of participants even when strict clinical criteria for classification were applied. While many of these participants may have diabetes of autoimmune aetiology, this was not clinically recognised or apparent, with low-C-peptide participants having late-onset diabetes, raised BMI and time from diagnosis to insulin treatment of several years.
Comparison with other studies
Our findings of a strong association between C-peptide and hypoglycaemia are consistent with previous findings in type 1 diabetes using self-reported hypoglycaemia [
1,
3,
9,
10,
19] and CGM studies in type 1 diabetes demonstrating higher SD or CV (SD/glucose), defined glucose variability and hypoglycaemia in individuals with lower C-peptide, including after islet transplant [
27‐
29]. In these studies, C-peptide was measured using a mixed meal tolerance test, which is not suitable for routine use in clinical practice. One study examined the correlation between fasting C-peptide and glucose variability (CGM SD) and found inverse associations in both type 1 diabetes and insulin-treated type 2 diabetes, although hypoglycaemia was not examined [
30].
The relationship between C-peptide levels and frequency of hypoglycaemia in those with a clinical diagnosis of type 2 diabetes is less widely recognised. The frequency of hypoglycaemia in type 2 diabetes is known to be associated with diabetes duration [
31], and is markedly higher in individuals treated with insulin for a long period than in those in the initial stages of treatment [
32]. It has also recently been shown that individuals with type 2 diabetes who were unable to achieve the ACCORD study’s treatment target of HbA
1c < 6.0% (42 mmol/mol) due to severe hypoglycaemia had lower C-peptide, with an OR of 23 for low C-peptide (baseline fasting C-peptide < 0.15 nmol/l) [
33].
While the potential utility of rCP as a biomarker for hypoglycaemia in individuals with diabetes has not been previously examined, previous research has shown that this test is highly correlated with mixed-meal-test-measured C-peptide [
11] and has similar or superior utility to glucagon-stimulated C-peptide assessment for differentiating between type 1 and type 2 diabetes [
34].
Strengths and limitations
This is to our knowledge the first study to examine the relationship between C-peptide and CGM-assessed hypoglycaemia in type 2 diabetes and the first to assess the potential utility of rCP for stratification of hypoglycaemia risk. This is an inexpensive test that can be measured at the point of clinical contact and could therefore easily be incorporated into clinical practice. Our additional study in a larger cohort using self-reported hypoglycaemia as an outcome strengthens our findings. We have assessed an older adult population in our study, an age group wherein hypoglycaemia is often not recognised, and consequences are more severe [
35] and where risk biomarker-based stratification would therefore be particularly helpful.
A weakness of our study is that C-peptide assessment in our replication study was in the most part performed on routine non-fasting plasma–EDTA samples received by our laboratory, therefore did not assess concurrent glucose or timing of samples in relation to meals. We have previously shown concurrent hypoglycaemia may result in reduced rCP levels and recommended that when assessing rCP, concurrent hypoglycaemia is excluded and a sample taken 1–5 h post meal [
11]. An additional weakness is that we cannot fully account for potential confounders which may alter the relationship between C-peptide and hypoglycaemia. In our CGM study, although most clinical characteristics were very similar between groups, prandial insulin use was more common in those with low C-peptide. This may reflect a causal association (those with low C-peptide and high glucose variability are likely to need prandial insulin for glycaemic control), although insulin regimen may directly influence hypoglycaemia [
20,
21]. Importantly, adjusting for insulin regime in our analysis did not alter our findings. In our questionnaire study, rates of self-reported hypoglycaemia were likely to be very dependent on rates of self-monitoring of blood glucose levels, which was not assessed.
Clinical implications
Our results suggest that an rCP sample can identify patients with insulin-treated type 2 diabetes who have a markedly increased risk of hypoglycaemia. These patients could not be identified by their clinical characteristics and only 59% would be identified by islet autoantibody testing, supporting the potential utility of rCP testing, which is practical, stable and inexpensive and therefore ideal for clinical use.
Identifying patients with insulin-treated type 2 diabetes at high risk of hypoglycaemia would be clinically helpful in guiding management (e.g. in setting appropriate glycaemic targets and levels of glucose monitoring) and would potentially allow consideration of treatment strategies for high glucose variability traditionally used in type 1 diabetes, such as carbohydrate counting and use of subcutaneous insulin pumps. An additional area where a robust, easily measurable biomarker for hypoglycaemia risk would be useful is in stratification of hypoglycaemia risk in relation to driving. Our CGM data showing no daytime episodes ≤ 3 mmol/mol in those with high rCP and an OR of 9.5 for self-reported severe hypoglycaemia (unconscious or fit, adjusted analysis) in the last year in those with low rCP are of interest in this regard and warrant further investigation.
Unanswered questions and future research
In this study, we used a previously defined threshold for hypoglycaemia risk based on previous studies in type 1 diabetes, and in our CGM cohort we excluded participants with C-peptide between 200 and 600 pmol/l (23% and 7% of participants in our type 2 diabetes and type 1 diabetes questionnaire cohorts, respectively). The relationship between C-peptide and glucose variability in previous literature is continuous but non-linear, with a very strong association at low but not high C-peptide levels [
28,
29,
36]. This is similar to the relationship observed with self-reported hypoglycaemia in our cohort (ESM Fig.
1). Further studies are needed to define optimal clinical cut-offs in obese populations with type 2 diabetes or alternatively to examine more complex approaches to risk prediction which might account for C-peptide as a continuous variable alongside other predictive features. Larger studies are needed to assess rCP against robust clinically important hypoglycaemia outcomes such as severe hypoglycaemia and to formally assess effectiveness and cost effectiveness of biomarker-based stratification of management. While it is likely that our results will be equally applicable to individuals diagnosed with type 1 diabetes, we did not directly assess this in our CGM study, which is thus an area of future work.
Conclusions
Low rCP is associated with increased glucose variability and hypoglycaemia in patients with insulin-treated type 2 diabetes and represents a practical, stable and inexpensive biomarker for assessment of hypoglycaemia risk.
Acknowledgements
The authors thank staff of the National Institute for Health Research Exeter Clinical Research Facility (University of Exeter and Royal Devon and Exeter Hospital) and the Blood Sciences Department, Royal Devon and Exeter Hospital for assistance with conducting the study. In particular, T. Libretto, R. Bolt, D. McGill, B. Knight, R. Kelland and A. Tregarthen (National Institute for Health Research Exeter Clinical Research Facility) are thanked for help with data collection. We thank A. Hattersley (University of Exeter) for helpful discussion.
Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made.