Predictors of HbA1c reduction and hypoglycemia in type 2 diabetes mellitus individuals switching from premixed to basal insulin: an exploratory analysis of optimization study

Abstract

Objective

To identify population with type 2 diabetes mellitus (T2DM) who are more likely to benefit from switching to basal insulin (BI) treatment from premixed insulin.

Methods

This secondary analysis included data from a previously published study (Optimization: NCT00693771) which was a single-arm, multicenter, 16 weeks, phase IV study. The analysis included participants with T2DM inadequately controlled with premixed insulin plus oral hypoglycemic drugs (OADs) who switched to BI plus OADs.

Results

Among the 297 participants included for analysis, subjects with fasting C-peptide (FCP)>1.2 nmol/L group showed a trend for greater reduction in HbA1c [Least square mean difference (LSMD), −0.59; 95% confidence interval (CI), −0.98 to −0.21; p = 0.003] and FPG (LSMD, −1.36; 95% CI, −2.20 to −0.53; p = 0.002) than those with FCP ≤ 0.4 nmol/L. The baseline insulin glargine 100 U/mL (Gla-100) dose increased significantly in 0.4 to ≤ 1.2 nmol/L group with LS mean difference (SE) of 0.16 (0.01) U/kg/day (p = 0.008) compared to FCP ≤ 0.4 nmol/L group. When combined with Diabetes Treatment Satisfaction Questionnaire (DTSQ) score, participants with a C-peptide level of 0.4 to ≤1.2 nmol/L (OR, 4.05; 95% CI, 1.08 to 15.22; p = 0.039) had significantly higher odds of achieving HbA1c <7%. The number of participants experiencing documented symptomatic hypoglycaemia (≤3.9 mmol/L) was higher in the FCP ≤0.4 nmol/L group compared to those in 0.4 to ≤1.2 nmol/L FCP group at any time of the day (31.6 vs. 17.1%) and during night (00:00 ∼ 05:59) (17.1 vs. 7.5%).

Conclusion

The findings from this study proposed that FCP is an important biomarker to identify T2DM participants who experience improved glucose control without compromising on hypoglycemia levels during switch from premixed insulin to BI. Participants especially with FCP levels >1.2 nmol/L may respond better in terms of HbA1c reduction without increased hypoglycemia risk compared to those with lower FCP values.

Keywords:

• Basal insulin
• premixed insulin
• C-peptide
• type 2 diabetes
• hypoglycemia

Introduction

Diabetes is a progressive disease with high incidences of mortality and morbidity especially in South East Asia and Western Pacific region. In 2019, the regional prevalence was 9.6% in Western Pacific and 8.8% in South East Asia¹. China and Japan are among the top five global countries with an increasing number of people with diabetes (116.4 million and 7.4 million respectively)¹. It is well documented that most individuals with long-standing type 2 diabetes mellitus (T2DM) will need to incorporate insulin into their treatment regimen due to its progressive nature².

As per the recommendations of several international guidelines, including the Chinese treatment guidelines, insulin is the preferred treatment modality when other oral antidiabetic drugs (OADs) and lifestyle modifications have failed to achieve and maintain target glycemic levels^3–6. As proof-of-concept, to overcome hyperglycaemia, earlier and more intensive insulin initiation in individuals with T2DM has demonstrated efficient results not only in maintaining glycemic control, but also has aided in restoration and preservation of pancreatic β-cell function^7,8. Thus, a few guidelines recommend treatment initiation with basal insulin (BI)^3,4,6, a long-acting insulin analogue that maintains basal glycemic levels throughout the day, whereas others recommend initiation with basal or premixed insulin^9,10.

Premixed insulin is known to mimic physiological endogenous insulin secretion and is a combination of rapid-acting bolus insulin and intermediate-acting basal insulin^11,12. In China, approximately 77% of individuals with T2DM receive premixed insulin, while only about 11% receive basal insulin¹³. However, premixed insulin might not be conveniently adjusted to meet individualized clinical need because of its fixed-ratio combination nature¹⁴. Apart from inadequate glycemic control, premixed insulin is also associated with an increased risk of hypoglycemia^15,16. Various studies have demonstrated that switching to insulin glargine-based therapy can lead to a significant reduction in glycated hemoglobin (HbA1c) and an improvement in treatment satisfaction, along with a low incidence of hypoglycemic events in individuals who were inadequately controlled on premixed insulin^17–20. Switching to BI decreased the incidence of hypoglycemia from 49.5% to 5.2% after 6 months of treatment²¹. In ATLANTIC study, participants switching from premixed insulin to BI experienced significant improvement in treatment satisfaction (mean change [SD]: −0.63[1.45]; p < 0.001) without concomitant increase in hypoglycemic events and weight²². Recent studies demonstrated that switching of treatment from premixed insulin to BI in Chinese participants demonstrated significant HbA1c and fasting blood glucose (FBG) improvements after 12–20 weeks of treatment regimen^12,14,23,24.

C-peptide is a widely used marker for assessing pancreatic β-cell function^25,26. Measurement of C-peptide level in insulin-treated individuals with diabetes also avoids the drawbacks of cross-reaction of assay between exogenous and endogenous insulin²⁷. Further, low baseline levels of C-peptide is associated with increased need for insulin as well as for shorter time to insulin treatment²⁷. Considering that limited evidence is available on switch from premixed insulin to BI^{14,19,21,28–30}, only two of these studies explored the characteristics of individuals who might benefit from switching therapy. In these studies, 2-hour postprandial C-peptide (2 h CP)²⁹ and 2 h-CP/Fasting C-peptide (FCP)²¹ were used as a marker to assess β-cell function in T2DM participants. FCP, a biomarker of β-cell function, can predict hypoglycemia risk in insulin naïve people with T2DM commencing insulin glargine 100 U/mL (Gla-100) as add-on therapy to OADs including sulfonylurea³¹.

It is therefore necessary to identify population with T2DM who are more likely to benefit in switching from premixed insulin to BI treatment. This secondary analysis of Optimisation study¹⁴ aimed to assess efficacy and safety of participants with T2DM stratified by baseline characteristics utilizing predefined baseline FCP groups and prior premixed insulin dose.

Materials and methods

Study design and plan of treatment

A post hoc study was conducted retrospectively using the data of a previously published study (Optimization study: NCT00693771) which was a single-arm, multicentre, 16 weeks, phase IV study¹⁴. The study population was regrouped by baseline characteristics according to predefined baseline C-peptide levels and prior premixed insulin dose. The selection of the study population, sample size calculation, inclusion, and exclusion has been described previously in Optimization study¹⁴. In brief, participants were eligible for enrolment if they were 35–75 years of age, diagnosed with T2DM for not more than 10 years and inadequately controlled for at least 3 months on premixed insulin plus one or two OADs. Additionally, participants were required to have an HbA1c level of 7.5–9.5%, an FPG level ≥ 6.7 mmol/L and a current premixed insulin dose of ≤50 IU per day. Eligible participants were switched from their previous premixed insulin therapy plus OADs to receive once-daily Gla-100 at bedtime using the OptiSet injection device (Sanofi-Aventis, Paris, France), while their OAD regimen remain unchanged. Insulin glargine was initiated at a dose equivalent to 80% of the total dose of intermediate insulin in the participants’ former premixed dose, and was up-titrated by hypoglycemia without the need for external assistance, and with or without a blood glucose level ≤3.3 mmol/L.

Ethical considerations

All the procedures in the original studies involved human participants. The studies were approved by the institutional review boards or ethics committees of the participating centers’ and were in accordance with the Declaration of Helsinki and Good Clinical Practice guidelines. All participant provided informed consent and the study had full ethical approval from the Ethics Committees at each participating site.

Assessment of clinical outcomes

All efficacy outcomes were stratified according to baseline C-peptide (FCP ≤0.4/0.4 to ≤1.2/>1.2 nmol/L) and prior premixed insulin dose (≤34/>34 U/day) in T2DM participants. We used FCP cut-offs as reported in the previous literature³¹, but combined the two highest groups as the >2.00 nmol/L group was small and similar to the group below it. For prior premixed insulin dose, median value of 34 U/day was used as the cut-off point for analysing the efficacy outcomes. The primary objective of this study was to identify the predictors of treatment outcomes (percentage of participants with HbA1c reduction >0.3% from baseline to week 16) in participants with T2DM who switched from premixed insulin to BI regimen. The baseline predictors of glycemic control explored were: age, gender, diabetes duration, weight, baseline BMI, baseline HbA1c, baseline fasting plasma glucose (FPG), baseline 2 h-post prandial glucose (2 h-PPG), baseline PPG excursion, baseline FCP, baseline 2hCP, baseline 2hCP/FCP, prior premixed insulin dose, insulin glargine starting dose, any diabetic complications and number of OADs. Among all the predictors, FCP and prior premixed insulin dose both were found to have statistical and clinical significance and was worth further exploration. The secondary objective was to characterize the associations of FCP, prior premixed insulin dose with efficacy and safety outcomes. The efficacy outcomes included percentage of participants achieving target HbA1c <7.0% and <7.5%, percentage of participants with HbA1c < 7.0% and <7.5% combined with change in Diabetes Treatment Satisfaction Questionnaire (DTSQ) score, percentage of participants achieving target HbA1c <7.0% and <7.5% when combined with hypoglycemia over 16 weeks of treatment, changes in HbA1c, FPG and insulin glargine dose from baseline to week 16 of study period. The safety outcomes were hypoglycemic events, serious adverse events (SAE), other adverse events (AE) and change in body weight.

Data abstraction and assessment

Information including the insulin dose, injection time, fasting, pre-prandial, and 2 h-PPG concentrations, as well as hypoglycemic episodes, were recorded in each individuals’ diary. The study included data for 16 weeks treatment period from all the participants. Other extracted data included laboratory data for hematology, urine analysis and biochemistry (HbA1c, FCP level, and 2 h-CP level that were collected at baseline and endpoint.

DTSQ (8 questions, 0-6 points) was used to assess the participants’ satisfaction to treatment at baseline and week 16 endpoint³². Safety was evaluated using adverse event (AE) reporting during the study. All possible hypoglycemic events were recorded and categorized according to the definitions recommended by the ADA 2005 Working Group on Hypoglycemia³³. ‘Documented symptomatic hypoglycemia’ was defined as an event during which typical symptoms of hypoglycemia were confirmed by a measured plasma glucose concentration of ≤3.9 mmol/l (≤70mg/dl)] and ‘severe hypoglycemia’ as events requiring assistance by another person to administer carbohydrate, glucagon or other therapy. Severe hypoglycemia was defined as an episode requiring the assistance of another person owing to severe impairment of consciousness and behavior with a PG level ≤3.0 mmol/L, or without PG measurement but the recovery being attributed to the restoration of normal FPG levels. Mild-to-moderate hypoglycemia was defined as having symptoms of hypoglycemia without the need for external assistance, and with or without a blood glucose level ≤3.3 mmol/L.

Statistical analysis

Multivariable stepwise logistic regression was performed to identify the major predictors for better glycemic control, that is, HbA1c reduction >0.3% from baseline to week 16 of treatment period. Analysis of covariance (ANCOVA) model was used to analyze associations between baseline FCP subgroups (≤0.4/0.4 to ≤1.2/>1.2 nmol/L), prior premixed insulin dose subgroups (≤34/>34, U/day) and HbA1c reduction, FPG reduction, body weight change and insulin glargine dose (U/day and U/kg/day) from baseline to week 16. Least square (LS) mean difference between groups and corresponding 95% confidence interval (CI) were demonstrated. Similarly, associations between baseline FCP subgroups, prior premixed insulin dose and HbA1c reduction (>0.3%/≤0.3%) and target HbA1c <7% or <7.5%, with or without hypoglycemia or DTSQ change ≥50% and hypoglycemia incidence were analyzed using logistic regression. Odds ratio (OR) and 95% CI were obtained. The confounding variables adjusted were baseline HbA1C, FPG, 2hPPG, body weight, age, gender, prior premixed insulin dose, baseline insulin glargine dose, diabetes duration, any diabetic complications and number of oral antidiabetic drug (OAD). All the analyses were 2-sided with significance level of 0.05 and performed using SAS^® Version 9.4 (SAS Institute Inc., Cary, NC).

Results

A total of 313 participants from 21 study sites in China were initially enrolled in this study, among which 297 participants comprised intent to treat (ITT) population who completed clinical results and FCP testing results. The baseline characteristics of the participants as per FCP and insulin dose subgroups have been described in Table 1 and Supplementary Table 1.¹⁴

Table 1. Participant characteristics and HbA1c reduction >0.3% from baseline to week 16.

	ORcrude (95%CI)	p Value	ORadj (95%CI) ^a	p Value
Age (years)	0.95 (0.93, 0.98)	0.002	0.95 (0.92, 0.98)	.001
Male (vs. female)	2.05 (1.27, 3.33)	0.004	2.00 (1.21, 3.31)	.007
Diabetes duration (years)	0.88 (0.80, 0.98)	0.015
Weight (kg)	1.02 (1.00, 1.04)	0.073
Baseline BMI (kg/m^2)	0.99 (0.92, 1.07)	0.024
Baseline HbA1c (%)	1.34 (0.88, 2.05)	0.176
Baseline FPG (mmol/L)	1.04 (0.93, 1.17)	0.465
Baseline 2hPPG (mmol/L)	0.99 (0.93, 1.05)	0.713
Baseline 2hPPG excursion (mmol/L)	0.97 (0.90, 1.04)	0.368
Baseline FCP (nmol/L)	2.63 (1.38, 5.00)	0.003	3.14 (1.59, 6.18)	.001
Baseline 2hCP (nmol/L)	1.60 (1.07, 2.38)	0.022
Baseline 2hCP/FCP (nmol/L)	0.94 (0.83, 1.06)	0.312
Prior premixed insulin dose (U/day)	0.98 (0.96, 1.01)	0.126	0.97 (0.95, 1.00)	.032
Insulin glargine dose (U/day)	0.98 (0.94, 1.02)	0.226
Insulin glargine dose (U/kg/day)	0.04 (0.00, 0.74)	0.031
Any diabetic complications	0.59 (0.35, 0.99)	0.044
No. of OAD (2 vs. 0–1)	1.29 (0.66, 2.50)	0.459

a: Adjusted ORs were obtained from multivariable logistic regression, stepwise method was used to select major predictors for patients with HbA1c reduction >0.3%.

After stepwise model selection in regression, age [OR, 0.95; 95% CI, 0.92 to 0.98 p = 0.001], baseline FCP [OR, 3.14; 9% CI, 1.59 to 6.18; p = 0.001] and prior premixed insulin dose [OR, 0.97; 9% CI, 0.95 to 1.00; p = 0.032] were found to be significant predictors of HbA1c reduction >0.3% from baseline to week 16 (Table 2).

Table 2. Baseline characteristics by FCP subgroups (ITT population).

	FCP ≤ 0.4 (nmol/L) N = 76	0.4 < FCP ≤ 1.2 (nmol/L) N = 187	FCP > 1.2 (nmol/L) N = 34	All N = 297	p Value
Age (years)	57.35 (8.35)	57.04 (8.28)	55.99 (8.54)	57.00 (8.31)	.729
Male	35 (46.1)	94 (50.3)	16 (47.1)	145 (48.82)	.806
Diabetes duration (years)	7.31 (2.73)	7.66 (2.61)	8.32 (2.05)	7.64 (2.59)	.171
Weight (kg)	67.72 (11.18)	70.56 (11.99)	69.53 (8.04)	69.72 (11.43)	.188
Baseline BMI (kg/m^2)	25.04 (3.11)	25.85 (3.05)	25.23 (2.55)	25.57 (3.03)	.109
Baseline HbA1c (%)	8.38 (0.65)	8.35 (0.53)	8.38 (0.58)	8.36 (0.57)	.967
Baseline FPG (mmol/L)	8.92 (1.94)	9.68 (2.18)	9.52 (1.77)	9.47 (2.10)	.030
Baseline 2hPPG (mmol/L)	14.79 (4.39)	15.46 (3.80)	13.55 (3.15)	15.07 (3.93)	.025
Baseline 2hPPG-FPG (mmol/L)	5.87 (3.73)	5.78 (3.09)	4.03 (3.17)	5.60 (3.31)	.012
Baseline FCP (nmol/L)	0.26 (0.09)	0.70 (0.22)	1.55 (0.22)	0.68 (0.41)	<.001
Baseline 2hCP (nmol/L)	0.76 (0.38)	1.38 (0.50)	2.21 (0.52)	1.32 (0.63)	<.001
Baseline 2hCP/FCP (nmol/L)	3.70 (2.51)	3.48 (2.11)	1.74 (0.75)	3.34 (2.19)	<.001
Prior premixed insulin dose (U/day)	33.47 (9.84)	33.19 (9.83)	36.15 (8.26)	33.60 (9.68)	.261
Insulin glargine dose (U/day)	19.38 (5.53)	19.05 (5.86)	20.36 (5.28)	19.28 (5.71)	.469
Insulin glargine dose (U/kg/day)	0.29 (0.09)	0.27 (0.08)	0.29 (0.08)	0.28 (0.08)	.128
Any diabetic complications	22 (29.0)	53 (28.3)	8 (23.5)	83 (27.95)	.826
No. of OAD
0–1	67 (88.2)	158 (84.5)	25 (73.5)	250 (84.18)	.149
2	9 (11.8)	29 (15.5)	9 (26.5)	47 (15.82)

The participants were stratified into various FCP subgroups as follows: FCP ≤ 0.4 (nmol/L) (n = 76), 0.4 to ≤1.2 (nmol/L) (n = 187) and FCP > 1.2 nmol/L (n = 34). After adjusting confounders, a statistically significant reduction in the mean change in HbA1c was observed from baseline to week 16 in participants with FCP > 1.2 nmol/L group than those with FCP ≤ 0.4 (nmol/L) group [Least square mean difference (LSMD), −0.59; 95% CI, −0.98 to −0.21; p = 0.003]. Similarly, a significant reduction in FPG was observed from baseline to week 16 in participants belonging to FCP > 1.2 nmol/L group (LSMD, −1.36; 95% CI, −2.20 to −0.53; p = 0.002) than those with FCP ≤ 0.4 (nmol/L) group. The baseline insulin glargine dose increased in both FCP > 1.2 nmol/L group and 0.4 to ≤1.2 (nmol/L) group of participants, however there was significant increase in 0.4 to ≤1.2 (nmol/L) group with LS mean difference (SE) of 0.16 (0.01) U/kg/day; p = 0.008 when compared with FCP ≤ 0.4 (nmol/L). Body weight change was comparable between FCP groups with a tendency of slight weight loss in the highest FCP group. (Table 3).

Table 3. Analysis of mean change in HbA1c, FPG, weight and insulin dose from baseline to week 16 by baseline C-peptide subgroups (ITT population).

	FCP ≤ 0.4 (nmol/L) N = 76	0.4 to ≤ 1.2 (nmol/L) N = 187	FCP > 1.2 (nmol/L) N = 34
HbA1c (%)
Baseline	8.38 (0.65)	8.35 (0.53)	8.38 (0.58)
Week 16	8.02 (1.06)	7.86 (1.06)	7.43 (0.77)
Change from baseline to week 16	−0.36 (0.98)	−0.49 (0.99)	−0.95 (0.83)
LS Mean (SE)^a	−0.25 (0.12)	−0.37 (0.09)	−0.84 (0.17)
LS Mean difference (SE) vs. Gla-100^a	Reference	−0.12 (0.13)	−0.59 (0.20)
95% CI		(−0.37 to 0.14)	(−0.98 to −0.21)
P-value		0.365	0.003
FPG (mmol/L)
Baseline	8.92 (1.94)	9.68 (2.18)	9.52 (1.77)
Week 16	6.77 (2.14)	6.68 (2.09)	5.63 (1.26)
Change from baseline to week 16	−2.15 (2.36)	−3.00 (2.98)	−3.89 (2.33)
LS Mean (SE)^a	−2.35 (0.26)	−2.54 (0.19)	−3.71 (0.37)
LS Mean difference (SE) vs. Gla-100^a	Reference	−0.19 (0.28)	−1.36 (0.42)
95% CI		(−0.73 to 0.36)	(−2.20 to −0.53)
P-value		0.498	0.002
Body Weight (kg)
Baseline	67.72 (11.18)	70.56 (11.99)	69.53 (8.04)
Week 16	67.88 (10.99)	70.68 (12.09)	69.32 (7.90)
Change from baseline to week 16	0.16 (1.89)	0.11 (2.49)	−0.21 (1.08)
LS Mean (SE)^a	−0.09 (0.30)	−0.07 (0.22)	−0.44 (0.41)
LS Mean difference (SE) vs. Gla-100^a	Reference	0.01 (0.31)	−0.35 (0.47)
95%CI		(−0.60 to 0.62)	(−1.28 to 0.58)
P-value		0.968	0.460
Insulin glargine dose (U/day)
Baseline	19.38 (5.53)	19.05 (5.86)	20.36 (5.28)
Week 16	25.92 (9.77)	29.49 (10.79)	29.71 (8.59)
Change from baseline to week 16	6.39 (7.59)	10.29 (8.68)	8.91 (7.67)
LS Mean (SE)^a	7.95 (1.11)	10.76 (0.79)	9.40 (1.48)
LS Mean difference (SE) vs. Gla-100^a	Reference	2.81 (1.14)	1.46 (1.72)
95%CI		(0.56 to 5.06)	(−1.94 to 4.85)
P-value		0.015	0.399
Insulin glargine dose (U/kg/day)
Baseline	0.29 (0.09)	0.27 (0.08)	0.29 (0.08)
Week 16	0.39 (0.15)	0.42 (0.16)	0.43 (0.13)
Change from baseline to week 16	0.09 (0.11)	0.15 (0.13)	0.13 (0.11)
LS Mean (SE)^a	0.11 (0.02)	0.16 (0.01)	0.14 (0.02)
LS Mean difference (SE) vs. Gla-100^a	Reference	0.04 (0.02)	0.03 (0.02)
95%CI		(0.01 to 0.08)	(−0.02 to 0.07)
p Value		0.008	0.312

^aAnalysis of covariance (ANCOVA) model adjusted for baseline HbA1C, FPG, 2hPPG, body weight, age, gender, prior premixed insulin dose, baseline insulin glargine dose, diabetes duration, any diabetic complication, No. of OAD.

Bold values denote statistical significance.

About 18.7% participants in 0.4 to ≤1.2 (nmol/L) and 23.5% participants in FCP > 1.2 nmol/L achieved target HbA1c <7%, but the ORs were not statistically significant. However, when combined with DTSQ score change ≥50%, participants in 0.4 to ≤1.2 (nmol/L) (OR, 4.05; 95% CI, 1.08 to 15.22; p = 0.039) had significantly higher odds of achieving target HbA1c <7%. Similarly, significant findings were achieved in participants with FCP 0.4 to ≤1.2 (nmol/L) (OR, 2.82, 95% CI, 1.25 to 6.39; p = 0.013) and FCP > 1.2 nmol/L (OR, 5.23, 95% CI, 1.77 to 15.42; p = 0.003) with target HbA1c <7.5% in combination with DTSQ score change ≥50%. The odds of achieving target HbA1c <7% or <7.5% in participants without hypoglycemia at week 16 was comparable between FCP subgroups with no statistical difference between them (Table 4).

Table 4. Analysis of target HbA1c < 7% or <7.5% at week 16 by baseline C-peptide subgroups (ITT population).

	FCp ≤ 0.4 (nmol/L) N = 76	0.4 to ≤ 1.2 (nmol/L) N = 187	FCp > 1.2 (nmol/L) N = 34
Target HbA1c < 7%
N (%)	11 (14.5)	35 (18.7)	8 (23.5)
OR (95% CI)	Reference	1.46 (0.64 to 3.35)	2.33 (0.72 to 7.50)
P-value		0.372	0.156
Target HbA1c < 7.5%
N (%)	28 (36.8)	70 (37.4)	19 (55.9)
OR (95% CI)	Reference	1.01 (0.54 to 1.91)	2.54 (0.98 to 6.59)
P-value		0.965	0.055
Target HbA1c < 7% and DTSQ change ≥ 50%^b
N (%)	3 (4.0)	24 (12.8)	4 (11.8)
OR (95% CI)	Reference	4.05 (1.08 to 15.22)	5.10 (0.97 to 26.72)
P-value		0.039	0.054
Target HbA1c < 7.5% and DTSQ change ≥ 50%^b
N (%)	9 (11.8)	48 (25.7)	11 (32.4)
OR (95% CI)	Reference	2.82 (1.25 to 6.39)	5.23 (1.77 to 15.42)
P-value		0.013	0.003
Target HbA1c < 7.0% and without hypoglycemia
N (%)	10 (13.2)	27 (14.4)	5 (14.7)
OR (95% CI)	Reference	1.14 (0.46 to 2.83)	1.28 (0.32 to 5.03)
P-value		0.774	0.727
Target HbA1c < 7.5% and without hypoglycemia
N (%)	17 (22.4)	52 (27.8)	10 (29.4)
OR (95% CI)	Reference	1.39 (0.68 to 2.85)	1.61 (0.56 to 4.62)
P-value		0.369	0.375

^aLogistic regression model adjusted for baseline HbA1C, FPG, 2hPPG, body weight, age, gender, prior premixed insulin dose, baseline insulin glargine dose, diabetes duration, any diabetic complication, No. of OAD ^bMedian of DTSQ change from baseline, i.e. 12.

Over the 16 weeks treatment period, the number of participants experiencing documented symptomatic hypoglycaemia (≤3.9 mmol/L) was significantly higher in those with FCP ≤0.4 nmol/L (nmol/L) at any time of the day (24 hrs.) (31.6 vs 17.1%; OR, 0.41, 95% CI, 0.21 to 0.81; p = 0.010) and at night (00:00 ∼ 05:59) (17.1% vs 7.5%; OR, 0.34, 95% CI, 0.14 to 0.82; p = 0.016) compared with FCP 0.4 to ≤1.2 (nmol/L) group (Table 5).

Table 5. Hypoglycemia by baseline C-peptide subgroups (ITT population).

	Anytime			Nocturnal (00:00 ∼ 05:59)
	FCp ≤ 0.4 (nmol/L) N = 76	0.4 to ≤ 1.2 (nmol/L) N = 187	FCp > 1.2 (nmol/L) N = 34	FCp ≤ 0.4 (nmol/L) N = 76	0.4 to ≤ 1.2 (nmol/L) N = 187	FCp > 1.2 (nmol/L) N = 34
Any hypoglycemia
N (%)	30 (39.5)	57 (30.5)	14 (41.2)	14 (18.4)	24 (12.8)	4 (11.8)
No. of events	105	119	29	34	32	7
OR (95% CI)^a	Reference	0.67 (0.36 to 1.24)	1.08 (0.43 to 2.68)	Reference	0.54 (0.25 to 1.18)	0.41 (0.11 to 1.48)
P-value		0.204	0.870		0.123	0.173
Severe hypoglycemia
N (%)	0 (0)	2 (1.1)	0 (0)	0 (0)	1 (0.5)	0 (0)
No. of events	0	2	0	0	1	0
OR (95% CI)^a	–	–	–	–	–	–
P-value		–	–		–	–
Documented symptomatic hypoglycaemia
N (%)	28 (36.8)	52 (27.8)	14 (41.2)	14 (18.4)	23 (12.3)	4 (11.8)
No. of events	92	107	27	34	30	7
OR (95% CI)^a	Reference	0.65 (0.35 to 1.21)	1.23 (0.50 to 3.07)	Reference	0.51 (0.23 to 1.12)	0.39 (0.11 to 1.43)
P-value		0.174	0.653		0.094	0.155
Documented symptomatic hypoglycaemia (≤3.9 mmol/L)
N (%)	24 (31.6)	32 (17.1)	11 (32.4)	13 (17.1)	14 (7.5)	3 (8.8)
No. of events	74	62	19	30	17	6
OR (95% CI)^a	Reference	0.41 (0.21 to 0.81)	0.97 (0.37 to 2.53)	Reference	0.34 (0.14 to 0.82)	0.34 (0.08 to 1.42)
P-value		0.010	0.943		0.016	0.140
Documented symptomatic hypoglycaemia (<3.0 mmol/L)
N (%)	6 (7.9)	9 (4.8)	2 (5.9)	2 (2.6)	4 (2.1)	0 (0)
No. of events	6	10	4	2	4	0
OR (95% CI)^a	Reference	0.51 (0.15 to 1.78)	0.94 (0.16 to 5.65)	–	–	–
P-value		0.294	0.942		–	–

^aLogistic regression model adjusted for baseline HbA1C, FPG, 2hPPG, body weight, age, gender, prior premixed insulin dose, baseline insulin glargine dose, diabetes duration, any diabetic complication, No. of OAD.

Bold values denote statistical significance.

When participants were regrouped according to prior premixed insulin dose ≤34 U/day (n = 155) and prior premixed insulin dose >34 U/day (n = 142), analysis of mean change in HbA1c, FPG, weight and insulin dose from baseline to week 16 revealed no statistically significant difference between them (Supplementary table 2). Similarly, no statistical significance was observed when prior premixed insulin dose was analyzed for target HbA1c < 7% or <7.5% combined with DTSQ score and without hypoglycemia (Supplementary table 3). The number of participants experiencing any hypoglycemia (19.7% vs 9.0%; p = 0.026) and documented hypoglycemia (19.7% vs 8.4%; p = 0.006) at night (00:00 ∼ 05:59) was higher in participants having prior premixed insulin dose >34 U/day compared to those with dose ≤34 U/day (Supplementary table 4).

Discussion

This post-hoc analyses of previously published Optimization study aimed to distinguish individuals stratified by FCP levels and prior premixed insulin dose who may achieve greatest benefit in terms of HbA1c reduction during treatment, by switching from premixed to BI. Results from the current analysis of participants with T2DM suggest that FCP levels >1.2 nmol/L provide significant reduction in levels of HbA1c without increasing hypoglycemic events.

C-peptide is an emerging valuable and widely used method to assess pancreatic β-cell function^25,26. After its formation following the cleavage of proinsulin, the 31-amino-acid C-peptide^34,35 is degraded slower than that of insulin (half-life: 20–30 min, vs. 3–5 min, respectively), allowing a more stable test window of fluctuating β-cell response. Hitherto, very limited evidence exists in the literature highlighting whether C-peptide can effectively predict need of insulin in individuals with diabetes^36,37. Since C-peptide is inversely associated with glycemic variability and post meal glucose in T2DM individuals, very low C-peptide levels (<0.2 nmol/L) is predictive of insulin requirement in these individuals²⁵. A pooled analysis of individuals with T2DM and on treatment with BI showed that low FCP levels (≤4 nmol/L) is useful in identifying individuals with insulin deficiency, enhanced insulin sensitivity and higher risk of hypoglycemia in contrast to those with higher FCP levels (>0.40 nmol/L)³¹. A recent review suggested that the current clinical role of C-peptide is to assist classification and management of insulin-treated individuals. The utility is greatest after 3–5 years from diagnosis in differentiating the type of diabetes^25,27. However, further substantial studies are needed to prove the effectiveness of this biomarker.

Nonetheless, in the current analysis, we observed that age [OR, 0.95; 95% CI, 0.92 to 0.98 p = 0.001], baseline FCP [OR, 3.14; 9% CI, 1.59 to 6.18; p = 0.001] and prior premixed insulin dose [OR, 0.97; 9% CI, 0.95 to 1.00; p = 0.032] were significant predictors of greater HbA1c reduction >0.3% at the end of treatment period. Participants with FCP > 1.2 nmol/L resulted in a statistically significant reduction in HbA1c (LSMD, −0.12; 95% CI, −0.37 to 0.14; p = 0.365) and FPG (LSMD, −1.36; 95% CI, −2.20 to −0.53; p = 0.002) levels from baseline to week 16. Similar findings were observed in a study where the higher standard-reaching rate of HbA1c in 2hCP/FCP > 3 group was significant (p < 0.05) compared with 2hCP/FCP ≤ 3 group²⁸. These present findings supports the value of measuring FCP when switching to BI as part of a β-cell-centric approach to identify individual subgroups with different treatment requirements as described by others^38,39. Furthermore, we speculate that T2DM individuals with better pancreatic β-cell functioning would achieve better glycemic control when switching from premixed to BI.

Relationship between FCP and DTSQ scores showed a significant proportion of individuals with FCP > 1.2 nmol/L (OR, 5.23, 95% CI, 1.77 to 15.42; p = 0.003) achieved target HbA1c <7.5% when combined with DTSQ scores. This is in accordance to participants switching from premixed insulin to BI where treatment satisfaction score showed significant increase from baseline to endpoint (27.12 ± 4.29 to 32.07 ± 5.16, p < 0.05)¹². The risk of hypoglycemia (19.7% vs 9.0%; p = 0.026) and documented hypoglycemia (19.7% vs 8.4%; p = 0.006) at night (00:00 ∼ 05:59) was significantly lower in individuals with prior premixed insulin dose ≤34 U/day. In a study by Wang et al. it was noted that insulin glargine combined with OADs was a suitable treatment option in individuals who received premixed insulin or insulin analogues in case of FCP ≥ 0.8 ng/ml, which is similar to our study that reported significance in participants with FCP > 1.2 (nmol/L)²⁸.

Few studies reported that there were no differences between premixed–prandial and BI in severe hypoglycemic events, and only minor hypoglycemic events were observed^40–42. Some researchers report a higher risk of minor nocturnal hypoglycemia with premixed insulin than with BI regimen⁴³. In our study, prior premixed insulin dose was not a significant predictor of HbA1c reduction. We hypothesized that this may be due to possible interaction with body weight or BMI. However, on stratification of participants by prior premixed insulin dose, a dose ≤34 U/day was beneficial in significantly reducing the incidence of hypoglycemia especially at night.

There should be wide awareness of FCP levels which will also enable clinicians to better decide on insulin needs, switch and titration schemes by taking into account the predominant defect, whether it is insulin deficiency or insulin resistance. The more highly insulin-sensitive person may benefit from a cautious and slow initiation and titration of BI, eventually requiring prandial supplementation, whereas the insulin-resistant individual will benefit from a more aggressive and forced titration of BI³¹. Moreover, clinicians can also use FCP measurements to monitor β-cell function in routine clinical practice in participants with T2DM treated with OADs for several years³¹.

Our study findings hold particular relevance in China, wherein clinicians usually prefer treatment initiation with premixed insulin whilst most of the T2DM individuals have functional β-cells⁴⁴. This is one of the first study to showcase the clinical importance of FCP to distinguish individuals with T2DM who could achieve maximum benefit during a switch from premixed to BI. Additionally, it is noteworthy to mention that the concept of switching from premixed to BI in poorly controlled individuals is relatively new, especially in Western countries. Thus, characterising individuals who could achieve better results using FCP proposes a simple, yet, a novel method.

Limitations

Our study is not without limitations. Since the Optimization study was a single-arm and treatment was at discretion of the physicians, an element of bias to our study results cannot be ruled out. Besides not being a pre-specified post-hoc analysis, another major limitation of this study is the lack of information on the measurement of C-peptide levels and its correction for influencing factors such as concomitant blood glucose values and previous night injected insulin dose.

Conclusion

In summary, this post hoc analysis of patient-level data from the Optimization study aimed to identify participants based on FCP levels and prior premixed insulin dose who may achieve greatest benefits in terms of HbA1c reduction during treatment after switching from premixed to basal insulin. The results suggest that T2DM participants with FCP levels >1.2 nmol/L may respond better in terms of HbA1c reduction without increased hypoglycemia risk compared to those T2DM participants with lower FCP values representing a group being more insulin sensitive and more insulin deficient. This FCP levels could be used as standard to carefully select individuals to minimise the risk of hypoglycemia, insulin dose requirements, and different treatment responses. Additionally, prospective studies should attempt in elucidating the further role of FCP on treatment outcomes and also to examine a possible causal relationship to hypoglycemia during the switch in insulin regimens.

Transparency

Declaration of financial/other relationships

Wenying Yang has received honoraria for speakers’ bureau and advisory board participation from Sanofi Aventis, Novo Nordisk, AstraZeneca, Bayer, Boehringer Ingelheim, Eli Lily, Merck Sharp & Dohme, Merck, and Servier and has also received investigator-initiated trial research grants from AstraZeneca outside of the submitted work. Minlu Zhang, Xia Zhang, and Nan Cui are employees of Sanofi China. Juan Du and Jing Hou were former employees of Sanofi. The authors thank the participants included in the study. Peer reviewers on this manuscript have no relevant financial or other relationships to disclose.

Author contributions

The study was designed by Wenying Yang. Juan Du coordinated the manuscript development process, reviewed and proofread the contents, and adjusted the structure. All authors contributed to the writing and reviewing the manuscript before submission. Wenying Yang is the guarantor for this manuscript.

Ethics statement

The study was conducted in accordance with the principles stated in the Declaration of Helsinki and in line with the International Conference on Harmonization Guidelines for Good Clinical Practice. An institutional review board at each site approved the study, and all participants gave written informed consent to participate.

Acknowledgements

The authors would like to acknowledge Anwesha Mandal of Indegene Pvt. Ltd., Bangalore, India for their medical writing support founded by Sanofi. And the study was funded by Sanofi.

Data availability statement

Qualified researchers may request access to patient-level data and related study documents including the clinical study report, study protocol with any amendments, blank case report form, statistical analysis plan, and dataset specifications. Patient-level data will be anonymized, and study documents will be redacted to protect the privacy of trial participants. Further details on Sanofi’s data sharing criteria, eligible studies, and process for requesting access can be found at: https://www.clinicalstudydatarequest.com

Additional information

Funding

This study was funded by Sanofi.