Effect of Dulaglutide Versus Liraglutide on Glucose Variability, Oxidative Stress, and Endothelial Function in Type 2 Diabetes: A Prospective Study

Twenty-two patients with T2DM who received liraglutide for at least 12 weeks were randomized to either continue liraglutide (n = 11) or dulaglutide (n = 11) for 24 weeks. The study was conducted from September 2016 to March 2018. The inclusion criteria were as follows: (1) a diagnosis of T2DM, (2) age ≥ 20 years, (3) HbA1c ≥ 6.5%, and (4) treatment with liraglutide for ≥ 12 weeks. The exclusion criteria were as follows: (1) the use of steroids or anti-inflammatory drugs, (2) diabetic ketosis and coma within 3 months before the start of the study, (3) severe infection, (4) severe trauma or malignancy, (5) an estimated glomerular filtration rate of < 30 ml/min/1.73 m², according to the Cockcroft-Gault formula, (6) recent surgery, (7) severe liver dysfunction, and (8) pregnancy.

Supplementary Figure S1 summarizes the protocol of this open-label, prospective, and randomized study. Patients were randomly divided into two groups: one that continued treatment with liraglutide and a second that was switched to treatment with dulaglutide. The number of patients withdrawing because of adverse events was similar between both treatment groups (Supplementary Figure S2). At baseline and 24 weeks after continuation of liraglutide or switching to dulaglutide, the following clinical and laboratory parameters were measured for all the patients in both groups before breakfast on day 1 of CGM: body weight (BW), low-density lipoprotein cholesterol (LDL-C), high-density lipoprotein cholesterol (HDL-C), triglycerides (TG), estimated glomerular filtration rate, blood pressure, fasting plasma glucose (FPG), and HbA1c content. Plasma oxidant capacity against N,N-diethylparaphenylenediamine was measured using the d-ROMs test on day 1. CGM was performed for 72 h. The parameters of glucose variability, such as mean glucose level (MGL), percentage coefficient of variation for glucose (%CV), mean amplitude of glycemic excursion (MAGE), and the mean of daily differences (MODD) in glucose levels were measured on days 2 and 3. Peripheral vascular arterial tonometry (PAT), as measured with the EndoPAT 2000^® system, provides noninvasive assessment of the RHI, a marker of vascular endothelial function and vascular health [13]. PAT was measured at baseline and 24 weeks after the start of the study. Because the RHI has a heteroscedastic error structure, L_RHI (a natural logarithm transformation) was used in all the analyses. All the clinical data (i.e., age, sex, smoking habit, and duration of diabetes in years) were retrieved from medical records. No changes were made to the type and dosage of oral hypoglycemic agents, insulin, angiotensin II receptor blocker, angiotensin-converting enzyme inhibitor, and statins during the study period to avoid possible influences on oxidative stress and endothelial function. All the drugs had been prescribed for at least 12 weeks before starting the study.

The Ethics Committee of the Showa University School of Medicine approved the study protocol. The study protocol is in compliance with the Declaration of Helsinki and current legal regulations in Japan. All procedures are in accordance with the ethical standards of the institutional and national committees responsible for human experimentation and with the Helsinki Declaration of 1964, as revised in 2013. Informed consent was obtained from all patients before being included in the study.

For each patient, a MiniMed iPro2 CGM system (Medtronic Inc., Northridge, CA, USA) was inserted subcutaneously on day 1 and removed on day 4. To avoid inaccurate results, glucose variability was calculated only on days 2 and 3. Venous blood samples were drawn for laboratory measurements on day 1 before breakfast. All the patients received a weight-maintaining diet (25–30 kcal/kg ideal BW). Glucose variability was calculated using the CGM data. The MGL and incidences of hypoglycemia (percent of h under 70 mg/dl in 24 h) were measured from the data recorded during CGM and adjusted for self-monitored blood glucose. The MAGE [14] was calculated to assess glucose variability. The MODD [15] was calculated as the mean of the absolute difference between corresponding glucose values on days 2 and 3. The percentage of CV was calculated using the coefficient of variation obtained by dividing the standard deviation by the MGL and multiplying by 100 [16].

Oxidative stress was measured using the d-ROMs test [17] and a dedicated photometer (F.R.E.E. System; imported by LTD Tokyo from Diacron International s.r.l., Grosseto, Italy), as reported previously [18]. According to the Wismerll kinetic procedure, the change in absorbance/min was expressed as arbitrary units after correction (U.CARR, where 1 U.CARR = the oxidant capacity of a 0.08 mg/dl of H₂O₂ solution; normal range, 250–300 U.CARR). The intra- and inter-assay coefficients of variation were 2.1% and 3.1%, respectively. An automated analyzer (BM6070; Japan Electron Optics Laboratory Co., Ltd., Tokyo, Japan) was used to measure serum total cholesterol, LDL-C, HDL-C, TG, and creatinine levels. Plasma glucose was measured by the glucose oxidase method, and HbA1c was measured by high-performance liquid chromatography [19].

Treatment satisfaction was assessed using a validated self-administered questionnaire, the diabetes treatment satisfaction questionnaire (DTSQ), at baseline and 24 weeks after the start of the study [20, 21]. The DTSQ is used to assess treatment satisfaction according to eight items that are scored on a 7-point Likert scale from 0 (very dissatisfied) to 6 (very satisfied). The total score was calculated as the sum of scores for the following six satisfaction items: “current treatment,” “convenience,” “flexibility,” “understanding,” “recommend,” and “continue,” yielding a total score from 0 to 36. The other two items of “perceived frequency of hyperglycemia” and “perceived frequency of hypoglycemia” were treated individually.

Eating behavior was assessed using the questionnaire of The Guideline for Obesity, issued by the Japan Society for the Study of Obesity, at baseline and 24 weeks after the start of the study. As reported previously [22, 23], this questionnaire consists of 55 questions of seven major scales, as follows: (1) recognition of weight and constitution (e.g., “Do you think it is easier for you to gain weight than others?” and “Do you think you gain weight because of less exercise?”), (2) external eating behavior (e.g., “If food smells and looks good, do you eat more than usual?,” “If you walk past a supermarket, do you have a desire to buy something delicious?,” and “If you see others eating, do you also have the desire to eat?”), (3) emotional eating behavior (e.g., “Do you have a desire to eat when you are irritated?” and “Do you have a desire to eat when you have nothing to do?”), (4) sense of hunger (e.g., “Do you get irritated when you feel hungry?” and “Do you often regret that you have eaten a lot of food?”), (5) eating style (e.g., “Do you eat quickly?” and “Are you known to eat a lot of food?”), (6) food preference (e.g., “Do you often snack on bread?,” “Do you like meat?,” and “Do you like noodles?”), and (7) regularity of eating habits (e.g., “Is your dinner time too late at night?” and “Do you gain body weight during holidays?”). All the items were rated on a four-point scale ranging from 1 (seldom) to 4 (very often).

The primary end points were d-ROMs and L_RHI change from baseline after 24 weeks of treatment. The secondary end points were changes in BW, glucose variability, lipid metabolism, DTSQ score, and eating behavior score.

All normally distributed continuous data sets, as determined by the Shapiro-Wilk test, are expressed as the mean ± standard deviation. Differences in continuous variables between the liraglutide and dulaglutide groups at baseline and after treatment were evaluated using the independent sample t test or Mann-Whitney U test. The chi-square test was used to compare categorical variables. Pearson’s simple linear regression was used to calculate the correlation coefficients. A probability p value < 0.05 was considered statistically significant. IBM SPSS, version 22, for Windows (IBM Corp., Armonk, NY, USA) was used to perform the analyses.

Of the 27 participants who were enrolled, 22 (81.5%) had completed the protocol. Table 1 shows the clinical and laboratory characteristics of the 22 participants (11 participants in the dulaglutide and liraglutide groups each). For the dulaglutide group, the mean age, body mass index, duration of diabetes, and HbA1c content were 69.2 ± 9.1 years, 26.7 ± 4.4 kg/m², 14.5 ± 8.9 years, and 7.1% ± 0.5%, respectively. The baseline characteristics of the patients in the liraglutide group were similar to those in the dulaglutide group. The values of FPG, MGL, MAGE, MODD, %CV, and hypoglycemia were similar in both groups. Oxidative stress, as measured by the d-ROMs test, and endothelial function, as measured by L_RHI, were similar between groups.

As Fig. 1 and Table 2 show, there were no significant differences in d-ROMs and L-RHI from baseline to 24 weeks or in the degree of change in d-ROMs and L-RHI between the dulaglutide and liraglutide groups.

Table 2 summarizes the general clinical and biochemical parameters. After 24 weeks of treatment, there were no significant differences in the changes from pre- to post-treatment of clinical and biochemical parameters, with the exception of glucose metabolism, between the dulaglutide and liraglutide groups.

Figure 2 shows the 24-h blood glucose profiles by CGM. There were no significant differences in the change of MODD and  %CV between the dulaglutide and liraglutide groups (5.50 ± 10.40 vs. 11.03 ± 22.91, p = 0.562; − 0.73 ± 3.33 vs. 2.07 ± 8.04, p = 0.300, respectively). The dulaglutide group was superior to the liraglutide group in terms of the changes in MGL and MAGE. In particular, regarding the changes in MGL and MAGE, the dulaglutide group remained unchanged, whereas the liraglutide group showed aggravation. (− 3.41 ± 21.66 vs. 27.45 ± 35.34, p = 0.023; –7.84 ± 26.10 vs. 27.00 ± 48.28, p = 0.034, respectively). There was no significant difference in the frequency of hypoglycemia between the dulaglutide and liraglutide groups.

Table 3 summarizes the self-reported patient treatment satisfaction, as measured by the DTSQ, and eating behavior at baseline at 24 weeks. After 24 weeks of treatment, the sum of the satisfaction-related DTSQ scores for the items of “current treatment,” “perceived frequency of hypoglycemia,” “flexibility,” “understanding,” “recommend,” and “continue” was unchanged in the dulaglutide and liraglutide groups (from 28.45 ± 3.93 to 30.00 ± 5.60 and from 26.27 ± 7.07 to 26.36 ± 3.98, respectively). In the dulaglutide group, the DTSQ score for “convenience” was significantly improved (from 4.55 ± 1.04 to 5.45 ± 0.93, p = 0.040), while the DTSQ score for “perceived frequency of hyperglycemia” was significantly poorer in the liraglutide group (from 2.45 ± 1.92 to 4.18 ± 1.47, p = 0.040). In addition, the DTSQ score for “perceived frequency of hyperglycemia” significantly improved in the dulaglutide group compared with the other group (− 0.45 ± 2.25 vs. 1.73 ± 1.74, p = 0.034).

After 24 weeks of treatment, the eating behavior scores were unchanged in the dulaglutide and liraglutide groups (from 36.31 ± 19.13 to 37.11 ± 19.49 and from 39.16 ± 20.67 to 39.45 ± 21.05, respectively).