- Email: [email protected]
Predictors of response to liraglutide in Japanese type 2 diabetes
diabetes research and clinical practice 106 (2014) 451–457
Contents available at ScienceDirect
Diabetes Research and Clinical Practice journ al h ome pa ge : www .elsevier.co m/lo cate/diabres
Predictors of response to liraglutide in Japanese type 2 diabetes Masao Toyoda a,*, Hiroki Yokoyama b, Katsushige Abe c, Shuji Nakamura d, Daisuke Suzuki e a
Division of Nephrology, Endocrinology and Metabolism, Department of Internal Medicine, Tokai University School of Medicine, Isehara 259-1193, Kanagawa, Japan b Jiyugaoka Medical Clinic Internal Medicine, Obihiro 080-0016, Hokkaido, Japan c Abe Diabetes Clinic, Oita 870-0039, Oita, Japan d Internal Medicine, Heiwadai Hospital, Miyazaki 880-0034, Miyazaki, Japan e Suzuki Diabetes Clinic, Atsugi 243-0035, Kanagawa, Japan
article info
abstract
Article history:
Aim: In Japan, liraglutide is approved for use alone or in combination with sulfonylureas,
Received 29 March 2014
and the approved maximum dosage is 0.9 mg/day. This restriction could limit the glucose-
Received in revised form
lowering effect of liraglutide in Japanese patients with type 2 diabetes mellitus (T2DM). This
8 August 2014
study was designed to identify predictors of response to liraglutide therapy at the approved
Accepted 15 September 2014
dosage.
Available online 7 October 2014
Methods: This observational retrospective study included 380 patients with T2DM who were treated with liraglutide alone or in combination with sulfonylureas at Diabetes Centers
Keywords:
located in four geographically different areas of Japan. Binary logistic regression analysis
Human GLP-1 analog
was used to identify patient characteristics associated with discontinuation of liraglutide,
Type 2 diabetes
while multiple regression and decision tree analyses were used to identify predictors of response to liraglutide therapy.
BMI
Results: Factors associated with discontinuation of liraglutide included high BMI, long duration of diabetes, and prior insulin therapy. Predictors of response to liraglutide therapy in patients who did not use insulin previously included previous use of few oral glucoselowering agents and high baseline HbA1c level. Conclusion: The results suggest greater efficacy of liraglutide monotherapy or liraglutidesulfonylurea combination therapy in patients with short duration of diabetes, non-insulin therapy, and low BMI and high HbA1c level at baseline. # 2014 Elsevier Ireland Ltd. All rights reserved.
1.
Introduction
Glucagon-like peptide-1 (GLP-1), a gastrointestinal hormone secreted from L-cells in the lower small intestine, is composed of 29–30 amino acids [1]. Various experimental and clinical
studies have reported that, among its diverse physiological effects [2], GLP-1 lowers blood glucose by promoting glucosedependent insulin secretion [3] and lowers body weight by reducing appetite and slowing of gastric emptying. GLP-1 is reportedly useful for the treatment of type 2 diabetes (T2DM) [4], but the clinical application of GLP-1 necessitates continuous
* Corresponding author. Tel.:+81 463 93 1121Ex2490; fax: +81 463 91 3350. E-mail address: [email protected] (M. Toyoda). http://dx.doi.org/10.1016/j.diabres.2014.09.052 0168-8227/# 2014 Elsevier Ireland Ltd. All rights reserved.
452
diabetes research and clinical practice 106 (2014) 451–457
subcutaneous infusion because it is rapidly degraded by dipeptidyl peptidase-4 [5]. Liraglutide (brand name Victoza1; 18-mg subcutaneous injection) is a human GLP-1 analogue in which a fatty acid addition to human GLP-1 increases its affinity to blood albumin and sustains GLP-1-mediated effects, thus making daily subcutaneous injection possible [6]. Liraglutide was approved for clinical use in Japan in 2010. Two Japanese clinical trials on liraglutide (monotherapy trial and trial of liraglutide combined with sulfonylurea (SU)) confirmed its glucose-lowering effects. Since the inclusion of anti-diabetic drugs under the Japanese Government-based Healthcare scheme depends on demonstration of efficacy and safety in Japanese Phase III clinical trials, liraglutide is used in Japan alone or in combination with SU. The approved maximum dose of liraglutide is 0.9 mg/day alone or in combination with SU, which is much smaller than that used in the US and Europe. The restrictions imposed on the dose and combination use seem to have reduced the glucose-lowering effects of liraglutide in Japanese patients with T2DM. Since 2010, liraglutide has been used in Japanese patients with a wide clinical spectrum of T2DM, spanning from drug naı¨ve patients to those requiring intensive insulin therapy. According to a recent Novo Nordisk drug safety report, of the patients treated with liraglutide in Japan between June 11 and October 7, 2010, 4 patients developed diabetic ketoacidosis (2 of these died later) and 16 developed hyperglycemia. Of these 20 patients, 17 experienced hyperglycemia after discontinuation of insulin therapy and switching to liraglutide. Because liraglutide is an injectable drug, many patients were switched from insulin to liraglutide therapy. These findings suggest that the clinical indications for liraglutide use are not clear and that errors in prescribing this drug could result in serious adverse events. It is therefore important to elucidate the clinical factors related to the efficacy and side effects of liraglutide. Previous studies identified several clinical parameters related to insulin secretion capacity, such as fasting and postprandial plasma C-peptide (CPR), C-peptide index (CPI) and increment in plasma CPR during glucagon tolerance test, which are useful for the prediction of efficacy of liraglutide [7– 10]. Usui et al. [7] reported that increments of CPR after glucagon stimulation test (GST-DCPR) and CPI are potentially useful predictors for a successful insulin-to-liraglutide switch. They reported that GST-DCPR of 1.34 ng/ml and CPI of 0.93 were cutoff points for good glycemic control [7]. Furthermore, another study showed that the cut-off values of insulinogenic index (0.14), CPI (1.1), fasting C-peptide (ng/ml), duration (19.5 years), urine C-peptide (33.3 mg/day) were useful predictors [8]. However, these tests are often difficult to measure in daily clinical practice. Therefore, there is a need for simple clinical parameters that can predict the effect of liraglutide. The aim of this study was to identify simple clinical and laboratory parameters that can predict the response to liraglutide based on decision tree analysis [11–13].
2.
Materials and methods
This multi-center collaborative observational retrospective study involved four medical facilities in different regions of Japan and included all 380 patients with T2DM who started
treatment with liraglutide between June 2010 and June 2011. The study analyzed a variety of patient characteristics and their relationship to both liraglutide efficacy and the continuation/discontinuation of the drug. All patients were managed clinically at one of the following Diabetes Clinics: Jiyugaoka Medical Clinic, Internal Medicine (Obihiro, Hokkaido), Tokai University School of Medicine (Isehara, Kanagawa), Abe Diabetes Clinic (Oita, Oita), and Heiwadai Hospital, Internal Medicine (Miyazaki, Miyazaki). Commencement of treatment with liraglutide required discontinuation of all other glucose-lowering agents, with the exception of SU drugs. Liraglutide was administered at 0.3 mg/day for 1 week or longer. Then, the dose was increased to 0.6 mg/day for 1 week or longer, and finally 0.9 mg/day was administered for the remaining part of the 6-month study period, unless the patient experienced severe gastrointestinal adverse events. Only SU drugs were approved as the combination therapy with liraglutide, and SU were administered according to the current Japanese approved dose and procedure of administration, which were selected by the attending physicians. The decision to continue or discontinue liraglutide was made by the attending physician. Body weight, body mass index (BMI), hemoglobin A1c (HbA1c) level, systolic blood pressure, and diastolic blood pressure were recorded at baseline (immediately before start of liraglutide therapy), and changes in these variables were monitored during the treatment. We also recorded age, sex, and duration of diabetes from the medical records. The participating patients were divided into two groups based on the use of insulin; (1) patients treated previously with insulin formed the insulin group (n = 266), and data on total daily insulin dose (total ID), basal insulin dose (basal ID), and bolus insulin dose (bolus ID) were recorded. (2) Patients who had been treated with oral glucose-lowering agents and those on diet therapy alone formed the non-insulin group (n = 114). The number and type of oral medications administered previously were recorded. HbA1c levels were measured according to the NGSP values and determined by highperformance liquid chromatography. The results are presented as mean standard deviation. The Mann–Whitney U test was used to examine the effects of liraglutide on each clinical and laboratory parameters during the study in both groups. The relationships between continuation/discontinuation of liraglutide and treatment group (insulin and non-insulin), age, sex, diabetes duration, baseline BMI, systolic blood pressure, diastolic blood pressure, and HbA1c level were tested using binary logistic regression analysis. Other variables used in the analysis included the number and type of oral medications used in prior treatment by the non-insulin group and the total ID, basal ID, and bolus ID for the insulin group. To identify factors associated with the response to liraglutide therapy, we examined the correlation between changes in HbA1c level after 6 months of treatment (dependent variable), and age, sex, diabetes duration, initial BMI, systolic blood pressure, diastolic blood pressure, and initial HbA1c level (independent variables). Multiple regression analysis was implemented using total ID, basal ID, and bolus ID as independent variables for the insulin group, and the number and types of oral glucose-lowering drugs
453
diabetes research and clinical practice 106 (2014) 451–457
administered in prior treatment as independent variables for the non-insulin group. A decision-tree analysis was also implemented using the extent of HbA1c change after 6 months of liraglutide treatment as the dependent variable and the above independent variables. The delta change in HbA1c (DHbA1c) served as a marker of response to liraglutide treatment. Separately, the Pearson product-moment correlation coefficient was used to assess the number and type of oral glucose-lowering drugs and DHbA1c. Statistical significance was determined at a p value of <5%. The study protocol was approved by the Human Ethics Review Committees of the participating institutions and a signed consent form was obtained from each subject.
3.
Results
3.1.
Clinical background
reason for the discontinuation was hyperglycemia in 79 patients (59.9%), gastrointestinal side effects in 17 patients (12.9%), and other reasons [drop-out and patient desire to stop the treatment (n = 6, 4.5%), reactions at injection site (n = 2, 1.5%), other reasons (n = 13, 9.8%), and unknown reason (n = 15, 11.4%)]. Patients who discontinued liraglutide had significantly longer T2DM disease duration and higher baseline HbA1c levels than those who continued the treatment (Table 1). Further analysis showed that large proportion of patients of the non-insulin/liraglutide discontinuation used large numbers of oral glucose-lowering agents, while patients of the insulin treatment/liraglutide discontinuation group had significantly longer duration of T2DM, higher HbA1c levels, and used insulin at high doses (total ID, basal ID, and bolus ID; Table 1).
3.2. Clinical parameters related to discontinuation of liraglutide
The mean age of the 380 patients was 59.1 11.8 years, and the mean baseline HbA1c level was 8.0 1.6%. Liraglutide was discontinued in 132 patients during the 6-month study period [insulin group, n = 100 (38%), non-insulin, n = 32 (28%)]. The
Logistic regression analysis identified prior insulin therapy (OR 1.734; 95%CI: 1.012–2.969; p = 0.045), high baseline BMI (OR 1.063; 95%CI: 1.014–1.115; p = 0.011), and long duration of T2DM (OR 1.024; 95%CI: 1.006–1.042; p = 0.009) as significant factors
Table 1 – Patient demographics and disease characteristics at baseline. Continued
Discontinued
Entire group n Pre-treatment non-insulin/insulin treatment Sex (male/female) Age Duration of diabetes (year) Body mass index (kg/m2) Systolic blood pressure (mmHg) Diastolic blood pressure (mmHg) HbA1c (%)
248 82/166 147/101 59.3 11.1 13.8 12.8 27.2 5.1 128.4 15.2 74.5 12.1 7.9 1.6
132 32/100 84/48 58.6 13.2 17.6 13.6 28.6 6.5 126.8 14.5 72.9 12.6 8.2 1.5
0.074 0.408 0.659 0.001 0.079 0.421 0.289 0.013
Non-insulin group n Pre-treatment (naı¨ve/1/2/3/4 GLA) Sex (male/female) Age Duration of diabetes (year) Body mass index (kg/m2) Systolic blood pressure (mmHg) Diastolic blood pressure (mmHg) HbA1c (%)
82 8/24/19/14/17 53/29 55.8 11.3 10.4 8.4 29.1 5.2 129.4 14.3 75.5 11.8 8.4 1.7
32 1/6/3/5/17 20/12 53.3 13.2 13.9 14.2 32.1 8.4 128.1 13.0 76.2 12.0 8.0 1.4
0.002 0.832 0.510 0.497 0.074 0.639 0.747 0.379
Insulin group n Sex (male/female) Age Duration of diabetes (year) Body mass index (kg/m2) Systolic blood pressure (mmHg) Diastolic blood pressure (mmHg) HbA1c (%) Total daily insulin dosage (U/day) Bolus daily insulin dosage (U/day) Basal daily insulin dosage (U/day)
166 94/72 61.1 10.6 15.6 14.2 26.2 4.7 127.9 15.7 74.1 12.2 7.6 1.5 19.4 17.2 12.7 13.2 7.8 9.3
100 64/36 60.3 12.8 18.8 13.2 27.4 5.3 126.3 15.0 71.8 12.6 8.2 1.6 31.8 21.2 20.3 17.4 14.0 11.2
0.236 0.589 0.007 0.054 0.557 0.209 <0.001 <0.001 <0.001 <0.001
Values are mean SD; GLA: oral glucose-lowering agents.
p Value
454
diabetes research and clinical practice 106 (2014) 451–457
Table 2 – Logistic regression analysis of baseline characteristics for discontinuation of liraglutide in the entire group, treatment-naı¨ve patients, patients treated previously with glucose-lowering agents (GLA), and insulin-treated patients. Logistic regression analysis was performed by the force entry method. Odds ratio (95% CI)
Table 3 – Multiple regression analyses of baseline characteristics for the response to liraglutide in the treatment-naı¨ve patients, patients treated previously with glucose-lowering agents (GLA), and insulin-treated patients. Multiple regression analysis was performed by the force entry method. Partial regression coefficient (95% CI)
p Value
p Value
Entire group (n = 380) Pre-treatmenta Sexb Age Duration of diabetes Baseline BMI Baseline SBP Baseline DBP Baseline HbA1c
1.734 0.716 0.991 1.024 1.063 0.995 0.987 1.125
(1.012–2.969) (0.443–1.156) (0.968–1.015) (1.006–1.042) (1.014–1.115) (0.975–1.015) (0.963–1.012) (0.966–1.310)
0.045 0.172 0.454 0.009 0.011 0.596 0.297 0.129
Non-insulin group (n = 82) Sexb Age Duration of diabetes Baseline BMI Baseline SBP Baseline DBP Baseline HbA1c Number of GLAa
0.063 ( 0.636–0.509) 0.001 ( 0.03–0.031) 0.011 ( 0.043–0.022) 0.028 ( 0.027–0.083) 0.018 ( 0.006–0.042) 0.019 ( 0.053–0.015) 0.645 ( 0.821 to 0.468) 0.411 (0.183–0.639)
0.825 0.955 0.513 0.308 0.133 0.266 <0.001 0.001
Non-insulin group (n = 114) Sexb Age Duration of diabetes Baseline BMI Baseline SBP Baseline DBP Baseline HbA1c Number of GLAc
1.102 0.985 1.061 1.093 0.989 1.008 1.042 1.771
(0.413–2.942) (0.936–1.038) (1.011–1.112) (1.000–1.196) (0.945–1.035) (0.951–1.068) (0.732–1.485) (1.188–2.642)
0.846 0.578 0.015 0.051 0.631 0.795 0.819 0.005
Insulin group (n = 266) Sexb Age Duration of diabetes Baseline BMI Baseline SBP Baseline DBP Baseline HbA1c Bolus daily insulin dosage Basal daily insulin dosage
Insulin group (n = 166) Sexb Age Duration of diabetes Baseline BMI Baseline SBP Baseline DBP Baseline HbA1c Bolus daily insulin dosage Basal daily insulin dosage
0.192 ( 0.466–0.082) 0.003 ( 0.011–0.018) 0.013 (0.004–0.022) 0.082 (0.050–0.114) 0.001 ( 0.011–0.009) 0.007 ( 0.020–0.006) 0.480 ( 0.603 to 0.357) 0.003 ( 0.014–0.008) 0.002 ( 0.018–0.013)
0.169 0.643 0.007 <0.001 0.833 0.311 <0.001 0.602 0.770
0.639 0.991 1.014 1.005 0.994 0.985 1.128 1.017 1.043
(0.355–1.149) (0.962–1.021) (0.992–1.036) (0.939–1.076) (0.970–1.018) (0.956–1.014) (0.919–1.384) (0.995–1.038) (1.008–1.079)
0.135 0.548 0.204 0.879 0.600 0.305 0.248 0.126 0.015
a
a Dummy indicator variables were set with treatment-naı¨ve patients and patients treated previously with glucose-lowering agents as 0 and insulin-treated patients as 1. Therefore, the odds ratio of 1.734 means that discontinuation of liraglutide was higher in insulin-treated patients than non-insulin-treated patients. b Dummy indicator variables were set with male as 1 and female as 2. Accordingly, the odds ratio of 0.716 means that discontinuation of liraglutide tended to be lower in females, albeit statistically insignificant. c Number of different glucose-lowering agents as pre-treatment (sulfonylureas, thiazolidinediones, a-glucosidase inhibitors, metformins, glinides). Thus, the odds ratio of 1.771 means discontinuation of liraglutide was higher in patients using several glucose-lowering agents compared with those on a single glucose-lowering agent.
for discontinuation of liraglutide therapy (Table 2). For the non-insulin group, longer duration of T2DM (OR 1.061; 95%CI: 1.011–1.112; p = 0.015) and use of several oral glucose-lowering agents (OR 1.771; 95%CI: 1.188–2.642; p = 0.005) correlated with the discontinuation of liraglutide (Table 2). On the other hand, for the insulin group, discontinuation of liraglutide correlated significantly with basal ID (OR 1.043; 95%CI: 1.008–1.079; p = 0.015).
3.3.
Clinical parameters related to efficacy of liraglutide
We also analyzed the factors involved in the response to liraglutide in 248 patients who continued treatment (Table 3).
Number of different glucose-lowering agents as pre-treatment (sulfonylureas, thiazolidinediones, a-glucosidase inhibitors, metformins, glinides). Thus, the partial regression coefficient of 0.411 means that reduction in HbA1c correlated significantly with lower number of glucose-lowering agents used before study enrolment. b Dummy indicator variables were set with male as 1 and female as 2. Thus, the partial regression coefficients of 0.063 in the noninsulin group and 0.192 in the insulin group means that reduction in HbA1c did not correlate with sex.
Multiple regression analysis was used to identify patient characteristics that correlated with DHbA1c after 6-month treatment. In the non-insulin group, HbA1c level at baseline correlated significantly with HbA1c reduction ( 0.645; 95%CI: 0.821 to 0.468; p < 0.001). On the other hand, the use of several oral glucose-lowering agents in prior treatment correlated negatively and significantly with DHbA1c, i.e. the use of large number of oral medications before the present clinical trial correlated with deterioration of glucose control after 6-month liraglutide treatment (0.411; 95%CI: 0.183 to 0.639; p = 0.001). Conversely, a higher baseline HbA1c level correlated with higher reduction in HbA1c after 6 months of liraglutide treatment, reflecting improvement in glycemic control (Table 3). The mean DHbA1c after 6-month treatment was calculated for each type of prior glucose-lowering agents. The results showed 3% decrease in HbA1c level in patients naı¨ve to oral glucose-lowering agents, although there were only a few patients in this category. Further analysis showed that the larger the number of glucose-lowering agents used before study enrollment the smaller the reduction in HbA1c level (Fig. 1; Pearson product-moment correlation coefficient r = 0.536, p < 0.01).
diabetes research and clinical practice 106 (2014) 451–457
455
Fig. 1 – Reduction in HbA1c after 6-month treatment with liraglutide according to the number of different glucoselowering agents used previously (n = 66. HbA1c values at baseline and 6 months were available in 82 patients). Values are mean W SD, p < 0.01, Pearson’s correlation coefficient: 0.536.
Similarly, in the insulin group, the response to liraglutide correlated significantly with duration of T2DM, BMI, and HbA1c level at baseline. Liraglutide was less efficacious in patients with long T2DM duration and high BMI at baseline; however, the response to liraglutide correlated with high HbA1c level at baseline (Table 3). The same independent variables were assessed in the decision tree analysis. In the non-insulin group, the initial HbA1c level branched at 11.5%, and equal or greater values were associated with a significantly greater reduction in HbA1c; the mean reduction in HbA1c was 5.6% irrespective of patient characteristics (Fig. 2). When the baseline HbA1c level was <11.5%, the baseline HbA1c level branched at 8.35%. When the baseline HbA1c level was 8.35%, the mean reduction in HbA1c level was 2.19% when two or fewer oral glucose-lowering agents had been used, and the mean reduction in HbA1c level was 0.49% when three or more oral glucose-lowering agents had been used. When the baseline HbA1c level was <8.35%, the mean reduction in HbA1c level was 0.64% in patients who used one type of oral glucose-lowering agents, but no HbA1c reduction was seen in patients who used two or more types oral glucose-lowering agents (Fig. 2). Similar to the non-insulin group, the baseline HbA1c level in the insulin group was also a significant independent factor that correlated with the reduction in HbA1c after 6 months of liraglutide treatment. In patients with baseline HbA1c level of 9.8%, the mean HbA1c level decreased by 2.97%., whereas patients with baseline HbA1c level of <9.8% showed branching of baseline HbA1c level at 8.25%. Baseline HbA1c level of 8.25% resulted in a mean reduction in HbA1c level of 0.9% after 6 months of liraglutide treatment. In patients with baseline HbA1c level of <8.25%, branching with BMI was noted; BMI branched at 31.1 and <31.1. Patients with BMI of <31.1 kg/m2 showed no reduction in HbA1c. Deterioration of HbA1c was noted in patients with HbA1c <8.25% and BMI 31.1 kg/m2 (Fig. 2).
Fig. 2 – Decision tree analysis of patients background for prediction of the response to 6-month liraglutide therapy based on reduction in HbA1c level in (A) patients treated previously with glucose-lowering agents (n = 66, HbA1c values at baseline and 6 months were available in 82 patients), and (B) patients treated previously with insulin (n = 127, HbA1c values at baseline and 6 months were available in 166 patients). In these box-and-whisker plots, lines within the boxes represent median values; the upper and lower lines of the boxes represent the 25th and 75th percentiles, respectively; and the upper and lower bars outside the boxes represent the 90th and 10th percentiles, respectively.
4.
Discussion
Based on logistic regression analysis, multi regression analysis and decision tree analysis, the present study demonstrated that the therapeutic response to liraglutide is influenced by treatment regimens and number of oral glucose-lowering agents, and that the use of liraglutide at an early stage is more efficacious when used alone or in combination with SU. In Japan, liraglutide has been used in a variety of patients with T2DM, including patients who switch from insulin therapy. However, justification for such use of liraglutide
456
diabetes research and clinical practice 106 (2014) 451–457
requires further analysis. In fact, the only Japanese phase III clinical trials that tested the efficacy of liraglutide included only patients who were on diet/exercise therapy and/or treated with oral glucose-lowering agents [14–17]. Liraglutide is contraindicated for insulin-dependent type 1 or type 2 diabetes mellitus because it enhances insulin secretion. However, the largest group of patients in the present study was patients who switched from insulin therapy to liraglutide. Based on our results, it is anticipated that switching from insulin therapy to liraglutide could improve the quality of life and treatment compliance by reducing episodes of hypoglycemia, lowering body weight, and/or decreasing the number of drug injections. Further increase in the number of patients who switch from insulin to liraglutide could allow further analysis of the pathophysiology and prognosis of T2DM. Based on the finding that total and basal ID contributed to the discontinuation of treatment in the insulin group, we suggest that insulin dosage and endogenous insulin secretion capacity and insulin resistance should be considered before the introduction of liraglutide for the treatment of T2DM. But at this point in time, further intervention studies are needed to clarify the relationship between efficacy of liraglutide and insulin secretion capacity and insulin resistance. The present study has certain limitations. First, we did not evaluate insulin secretion capacity by using glucagon tolerance test or CPI. Several studies have suggested the usefulness of these two parameters as predictors for the efficacy of liraglutide [7–9,11]. However, our study examined patient characteristics linked to the practicing physician’s decision to continue or discontinue liraglutide since patients characteristics are easy tools for evaluation of efficacy of liraglutide in daily clinical practice. However, taking into consideration the results of this study, application of tests of endogenous insulin secretion capacity would be highly useful to ensure treatment efficacy and safety of patients switching from insulin to liraglutide. Such tests would be especially useful in patients with a high basal ID. Second, this study was retrospective in nature. Accordingly, discontinuation was based on the decision of the individual investigator. In other words, the criteria used for discontinuation, especially hyperglycemia, varied among the participating medical institutions. This limitation probably introduced some bias and thus a prospective study with predefined criteria of discontinuation based on glucose control is needed. Finally this study was multi-center analysis but included a small number of institutions, with limited sample size. Accordingly, prospective studies that include larger sample size and several institutions are needed to confirm the present results. Because the mean baseline BMI of patients of the noninsulin group was within the obesity range (27–28 kg/m2), the main reasons for the introduction of liraglutide were improvement of blood glucose level and weight loss. While the mean BMI of patients of the Japanese phase III clinical trial of liraglutide was 25 kg/m2, some patients in actual clinical practice are grossly obese and the dose of liraglutide per unit body weight is lower than the recommended clinical dose. This fact suggests that the recommended liraglutide dose could be less than ideal at least in some patients. In this regard; the approved dose in Japan is half the dose approved in Europe
and United States. Based on these discrepancies and the two actions of liraglutide on blood glucose regulation and body weight reduction, the recommended treatment regimen and dosage used should be reconsidered. Analysis of the correlation between patient characteristics and response to liraglutide demonstrated that, in the non-insulin group, liraglutide resulted in successful regulation of blood glucose levels, as measured by HbA1c levels, in both treatment-naı¨ve patients and patients who switched to liraglutide from other oral glucose-lowering drugs. In this study, which followed the treatment regimen approved for use in Japan, all oral glucose-lowering drugs other than SU were discontinued at study entry. Further studies are needed to analyze changes in efficacy-related factors when the indications for concurrent liraglutide therapy are expanded. Irrespective of disease duration, age, and sex, good response to liraglutide (reduction of HbA1c by 2%) was based on HbA1c level at baseline of 11.5% or 8.35% with the use of 2 oral glucose-lowering drugs in prior treatment (Fig. 2). However, in some cases, liraglutide treatment was effective even when insulin therapy was used in prior treatment. However, 20 insulin units/day used in these patients was a relatively low dosage, and in virtually all cases, liraglutide was administered with the aim of withdrawal from intensive insulin therapy. The present study identified duration of diabetes, baseline BMI, and baseline HbA1c levels as significant determinants of the response to liraglutide therapy in patients with long duration of diabetes. The results also demonstrated that the best response to liraglutide was achieved by patients with high HbA1c level at baseline. It is noteworthy that glucotoxicity itself is reported to have a major impact on the ability of GLP-1 to promote insulin secretion [18]. Therefore, careful attention should be exercised when introducing liraglutide in patients with HbA1c levels. Introduction of liraglutide therapy to patients with T2DM with near-normalization of blood glucose seems safe and potentially most efficacious. The present study demonstrated no improvement in patients with baseline HbA1c levels of <8.25% and in patients with baseline BMI of >31.1% who had been previously treated with insulin, even after 6 months of liraglutide therapy. This finding indicates that for patients to be considered for liraglutide treatment with the current regimen and dosage ranges in Japan, they should be examined comprehensively, and the number of different oral glucose-lowering drugs, insulin dose, and baseline HbA1c and BMI values should be taken into consideration in the decision making. With regard to co-treatment with liraglutide, three different statistical tests identified large number of glucoselowering agents as a significant determinant of discontinuation and efficacy of liraglutide. The present study was designed to comply with the Government-approved guidelines for use of liraglutide (monotherapy and combination with SU only). The indications for liraglutide could change in the future, and additional oral anti-diabetic medications may be approved for use in combination with liraglutide. In particular, combination with metformin and other incretinrelated drugs could theoretically enhance weight loss and increase active GLP-1 levels.
diabetes research and clinical practice 106 (2014) 451–457
Conflict of interest statement [9]
The authors declare no conflict of interest in relation to this article. [10]
Acknowledgment This study was sponsored by the Division of Nephrology, Endocrinology and Metabolism, Department of Internal Medicine (Grant No. 12F188139), Tokai University School of Medicine, Japan.
references
[1] Bell GI, Santerre RF, Mullenbach GT. Hamster preproglucagon contains the sequence of glucagon and two related peptides. Nature 1983;302:716–8. [2] Drucker DJ, Nauck MA. The incretin system: glucagon-like peptide-1 receptor agonists and dipeptidyl peptidase-4 inhibitors in type 2 diabetes. Lancet 2006;368:1696–705. [3] Ritzel R, Orskov C, Holst JJ, Nauck MA. Pharmacokinetic, insulinotropic, and glucagonostatic properties of GLP-1 [736 amide] after subcutaneous injection in healthy volunteers. Dose–response relationships. Diabetologia 1995;38:720–5. [4] Gutniak M, Orskov C, Holst JJ, Ahre´n B, Efendic S. Antidiabetogenic effect of glucagon-like peptide-1 (736)amide in normal subjects and patients with diabetes mellitus. N Engl J Med 1992;326:1316–22. [5] Larsen J, Hylleberg B, Ng K, Damsbo P. Glucagon-like peptide-1 infusion must be maintained for 24 h/day to obtain acceptable glycemia in type 2 diabetic patients who are poorly controlled on sulphonylurea treatment. Diabetes Care 2001;24:1416–21. [6] Knudsen LB, Nielsen PF, Huusfeldt PO, Johansen NL, Madsen K, Pedersen FZ, et al. Potent derivatives of glucagon-like peptide-1 with pharmacokinetic properties suitable for once daily administration. J Med Chem 2000;43:1664–9. [7] Usui R, Yabe D, Kuwata H, Fujiwara S, Watanabe K, Hyo T, et al. Retrospective analysis of safety and efficacy of insulin-to-liraglutide switch in Japanese type 2 diabetes: a caution against inappropriate use in patients with reduced b-cell function. J Diabetes Invest 2013;4:585–94. [8] Kozawa J, Inoue K, Iwamoto R, Kurashiki Y, Okauchi Y, Kashine S, et al. Liraglutide is effective in type 2 diabetic
[11]
[12]
[13]
[14]
[15]
[16]
[17]
[18]
457
patients with sustained endogenous insulin-secreting capacity. J Diabetes Invest 2012;3:294–7. Iwao T, Sakai K, Sata M. Postprandial serum C-peptide is a useful parameter in the prediction of successful switching to liraglutide monotherapy from complex insulin therapy in Japanese patients with type 2 diabetes. J Diabetes Complications 2013;27:87–91. Nambu T, Matsuda Y, Matsuo K, Kanai Y, Yonemitsu S, Muro S, et al. Liraglutide administration in type 2 diabetic patients who either received no previous treatment or were treated with an oral hypoglycemic agent showed greater efficacy than that in patients switching from insulin. J Diabetes Invest 2013;4:69–77. Worachartcheewan A, Nantasenamat C, Isarankura-NaAyudhya C, Pidetcha P, Prachayasittikul V. Identification of metabolic syndrome using decision tree analysis. Diabetes Res Clin Pract 2010;90:e15–8. Hische M, Luis-Dominguez O, Pfeiffer AF, Schwarz PE, Selbig J, Spranger J. Decision trees as a simple-to-use and reliable tool to identify individuals with impaired glucose metabolism or type 2 diabetes mellitus. Eur J Endocrinol 2010;163:565–71. Mo¨hlig M, Flo¨ter A, Spranger J, Weickert MO, Schill T, Schlo¨sser HW, et al. Predicting impaired glucose metabolism in women with polycystic ovary syndrome by decision tree modelling. Diabetologia 2006;49:2572–9. Seino Y, Rasmussen MF, Nishida T, Kaku K. Efficacy and safety of the once-daily human GLP-1 analogue, liraglutide, vs glibenclamide monotherapy in Japanese patients with type 2 diabetes. Curr Med Res Opin 2010;26:1013–22. Kaku K, Rasmussen MF, Nishida T, Seino Y. Fifty-two-week, randomized, multicenter trial to compare the safety and efficacy of the novel glucagon-like peptide-1 analog liraglutide vs glibenclamide in patients with type 2 diabetes. J Diabetes Invest 2011;2:441–7. Kaku K, Rasmussen MF, Clauson P, Seino Y. Improved glycaemic control with minimal hypoglycaemia and no weight change with the once-daily human glucagon-like peptide-1 analogue liraglutide as add-on to sulphonylurea in Japanese patients with type 2 diabetes. Diabetes Obes Metab 2010;12:341–7. Seino Y, Rasmussen MF, Nishida T, Kaku K. Glucagon-like peptide-1 analog liraglutide in combination with sulfonylurea safely improves blood glucose measures vs sulfonylurea monotherapy in Japanese patients with type 2 diabetes: results of a 52-week, randomized, multicenter trial. J Diabetes Invest 2011;2:280–6. Højberg PV, Vilsbøll T, Rabøl R, Knop FK, Bache M, Krarup T, et al. Four weeks of near-normalisation of blood glucose improves the insulin response to glucagon-like peptide-1 and glucose-dependent insulinotropic polypeptide in patients with type 2 diabetes. Diabetologia 2009;52:199–207.
Contents available at ScienceDirect
Diabetes Research and Clinical Practice journ al h ome pa ge : www .elsevier.co m/lo cate/diabres
Predictors of response to liraglutide in Japanese type 2 diabetes Masao Toyoda a,*, Hiroki Yokoyama b, Katsushige Abe c, Shuji Nakamura d, Daisuke Suzuki e a
Division of Nephrology, Endocrinology and Metabolism, Department of Internal Medicine, Tokai University School of Medicine, Isehara 259-1193, Kanagawa, Japan b Jiyugaoka Medical Clinic Internal Medicine, Obihiro 080-0016, Hokkaido, Japan c Abe Diabetes Clinic, Oita 870-0039, Oita, Japan d Internal Medicine, Heiwadai Hospital, Miyazaki 880-0034, Miyazaki, Japan e Suzuki Diabetes Clinic, Atsugi 243-0035, Kanagawa, Japan
article info
abstract
Article history:
Aim: In Japan, liraglutide is approved for use alone or in combination with sulfonylureas,
Received 29 March 2014
and the approved maximum dosage is 0.9 mg/day. This restriction could limit the glucose-
Received in revised form
lowering effect of liraglutide in Japanese patients with type 2 diabetes mellitus (T2DM). This
8 August 2014
study was designed to identify predictors of response to liraglutide therapy at the approved
Accepted 15 September 2014
dosage.
Available online 7 October 2014
Methods: This observational retrospective study included 380 patients with T2DM who were treated with liraglutide alone or in combination with sulfonylureas at Diabetes Centers
Keywords:
located in four geographically different areas of Japan. Binary logistic regression analysis
Human GLP-1 analog
was used to identify patient characteristics associated with discontinuation of liraglutide,
Type 2 diabetes
while multiple regression and decision tree analyses were used to identify predictors of response to liraglutide therapy.
BMI
Results: Factors associated with discontinuation of liraglutide included high BMI, long duration of diabetes, and prior insulin therapy. Predictors of response to liraglutide therapy in patients who did not use insulin previously included previous use of few oral glucoselowering agents and high baseline HbA1c level. Conclusion: The results suggest greater efficacy of liraglutide monotherapy or liraglutidesulfonylurea combination therapy in patients with short duration of diabetes, non-insulin therapy, and low BMI and high HbA1c level at baseline. # 2014 Elsevier Ireland Ltd. All rights reserved.
1.
Introduction
Glucagon-like peptide-1 (GLP-1), a gastrointestinal hormone secreted from L-cells in the lower small intestine, is composed of 29–30 amino acids [1]. Various experimental and clinical
studies have reported that, among its diverse physiological effects [2], GLP-1 lowers blood glucose by promoting glucosedependent insulin secretion [3] and lowers body weight by reducing appetite and slowing of gastric emptying. GLP-1 is reportedly useful for the treatment of type 2 diabetes (T2DM) [4], but the clinical application of GLP-1 necessitates continuous
* Corresponding author. Tel.:+81 463 93 1121Ex2490; fax: +81 463 91 3350. E-mail address: [email protected] (M. Toyoda). http://dx.doi.org/10.1016/j.diabres.2014.09.052 0168-8227/# 2014 Elsevier Ireland Ltd. All rights reserved.
452
diabetes research and clinical practice 106 (2014) 451–457
subcutaneous infusion because it is rapidly degraded by dipeptidyl peptidase-4 [5]. Liraglutide (brand name Victoza1; 18-mg subcutaneous injection) is a human GLP-1 analogue in which a fatty acid addition to human GLP-1 increases its affinity to blood albumin and sustains GLP-1-mediated effects, thus making daily subcutaneous injection possible [6]. Liraglutide was approved for clinical use in Japan in 2010. Two Japanese clinical trials on liraglutide (monotherapy trial and trial of liraglutide combined with sulfonylurea (SU)) confirmed its glucose-lowering effects. Since the inclusion of anti-diabetic drugs under the Japanese Government-based Healthcare scheme depends on demonstration of efficacy and safety in Japanese Phase III clinical trials, liraglutide is used in Japan alone or in combination with SU. The approved maximum dose of liraglutide is 0.9 mg/day alone or in combination with SU, which is much smaller than that used in the US and Europe. The restrictions imposed on the dose and combination use seem to have reduced the glucose-lowering effects of liraglutide in Japanese patients with T2DM. Since 2010, liraglutide has been used in Japanese patients with a wide clinical spectrum of T2DM, spanning from drug naı¨ve patients to those requiring intensive insulin therapy. According to a recent Novo Nordisk drug safety report, of the patients treated with liraglutide in Japan between June 11 and October 7, 2010, 4 patients developed diabetic ketoacidosis (2 of these died later) and 16 developed hyperglycemia. Of these 20 patients, 17 experienced hyperglycemia after discontinuation of insulin therapy and switching to liraglutide. Because liraglutide is an injectable drug, many patients were switched from insulin to liraglutide therapy. These findings suggest that the clinical indications for liraglutide use are not clear and that errors in prescribing this drug could result in serious adverse events. It is therefore important to elucidate the clinical factors related to the efficacy and side effects of liraglutide. Previous studies identified several clinical parameters related to insulin secretion capacity, such as fasting and postprandial plasma C-peptide (CPR), C-peptide index (CPI) and increment in plasma CPR during glucagon tolerance test, which are useful for the prediction of efficacy of liraglutide [7– 10]. Usui et al. [7] reported that increments of CPR after glucagon stimulation test (GST-DCPR) and CPI are potentially useful predictors for a successful insulin-to-liraglutide switch. They reported that GST-DCPR of 1.34 ng/ml and CPI of 0.93 were cutoff points for good glycemic control [7]. Furthermore, another study showed that the cut-off values of insulinogenic index (0.14), CPI (1.1), fasting C-peptide (ng/ml), duration (19.5 years), urine C-peptide (33.3 mg/day) were useful predictors [8]. However, these tests are often difficult to measure in daily clinical practice. Therefore, there is a need for simple clinical parameters that can predict the effect of liraglutide. The aim of this study was to identify simple clinical and laboratory parameters that can predict the response to liraglutide based on decision tree analysis [11–13].
2.
Materials and methods
This multi-center collaborative observational retrospective study involved four medical facilities in different regions of Japan and included all 380 patients with T2DM who started
treatment with liraglutide between June 2010 and June 2011. The study analyzed a variety of patient characteristics and their relationship to both liraglutide efficacy and the continuation/discontinuation of the drug. All patients were managed clinically at one of the following Diabetes Clinics: Jiyugaoka Medical Clinic, Internal Medicine (Obihiro, Hokkaido), Tokai University School of Medicine (Isehara, Kanagawa), Abe Diabetes Clinic (Oita, Oita), and Heiwadai Hospital, Internal Medicine (Miyazaki, Miyazaki). Commencement of treatment with liraglutide required discontinuation of all other glucose-lowering agents, with the exception of SU drugs. Liraglutide was administered at 0.3 mg/day for 1 week or longer. Then, the dose was increased to 0.6 mg/day for 1 week or longer, and finally 0.9 mg/day was administered for the remaining part of the 6-month study period, unless the patient experienced severe gastrointestinal adverse events. Only SU drugs were approved as the combination therapy with liraglutide, and SU were administered according to the current Japanese approved dose and procedure of administration, which were selected by the attending physicians. The decision to continue or discontinue liraglutide was made by the attending physician. Body weight, body mass index (BMI), hemoglobin A1c (HbA1c) level, systolic blood pressure, and diastolic blood pressure were recorded at baseline (immediately before start of liraglutide therapy), and changes in these variables were monitored during the treatment. We also recorded age, sex, and duration of diabetes from the medical records. The participating patients were divided into two groups based on the use of insulin; (1) patients treated previously with insulin formed the insulin group (n = 266), and data on total daily insulin dose (total ID), basal insulin dose (basal ID), and bolus insulin dose (bolus ID) were recorded. (2) Patients who had been treated with oral glucose-lowering agents and those on diet therapy alone formed the non-insulin group (n = 114). The number and type of oral medications administered previously were recorded. HbA1c levels were measured according to the NGSP values and determined by highperformance liquid chromatography. The results are presented as mean standard deviation. The Mann–Whitney U test was used to examine the effects of liraglutide on each clinical and laboratory parameters during the study in both groups. The relationships between continuation/discontinuation of liraglutide and treatment group (insulin and non-insulin), age, sex, diabetes duration, baseline BMI, systolic blood pressure, diastolic blood pressure, and HbA1c level were tested using binary logistic regression analysis. Other variables used in the analysis included the number and type of oral medications used in prior treatment by the non-insulin group and the total ID, basal ID, and bolus ID for the insulin group. To identify factors associated with the response to liraglutide therapy, we examined the correlation between changes in HbA1c level after 6 months of treatment (dependent variable), and age, sex, diabetes duration, initial BMI, systolic blood pressure, diastolic blood pressure, and initial HbA1c level (independent variables). Multiple regression analysis was implemented using total ID, basal ID, and bolus ID as independent variables for the insulin group, and the number and types of oral glucose-lowering drugs
453
diabetes research and clinical practice 106 (2014) 451–457
administered in prior treatment as independent variables for the non-insulin group. A decision-tree analysis was also implemented using the extent of HbA1c change after 6 months of liraglutide treatment as the dependent variable and the above independent variables. The delta change in HbA1c (DHbA1c) served as a marker of response to liraglutide treatment. Separately, the Pearson product-moment correlation coefficient was used to assess the number and type of oral glucose-lowering drugs and DHbA1c. Statistical significance was determined at a p value of <5%. The study protocol was approved by the Human Ethics Review Committees of the participating institutions and a signed consent form was obtained from each subject.
3.
Results
3.1.
Clinical background
reason for the discontinuation was hyperglycemia in 79 patients (59.9%), gastrointestinal side effects in 17 patients (12.9%), and other reasons [drop-out and patient desire to stop the treatment (n = 6, 4.5%), reactions at injection site (n = 2, 1.5%), other reasons (n = 13, 9.8%), and unknown reason (n = 15, 11.4%)]. Patients who discontinued liraglutide had significantly longer T2DM disease duration and higher baseline HbA1c levels than those who continued the treatment (Table 1). Further analysis showed that large proportion of patients of the non-insulin/liraglutide discontinuation used large numbers of oral glucose-lowering agents, while patients of the insulin treatment/liraglutide discontinuation group had significantly longer duration of T2DM, higher HbA1c levels, and used insulin at high doses (total ID, basal ID, and bolus ID; Table 1).
3.2. Clinical parameters related to discontinuation of liraglutide
The mean age of the 380 patients was 59.1 11.8 years, and the mean baseline HbA1c level was 8.0 1.6%. Liraglutide was discontinued in 132 patients during the 6-month study period [insulin group, n = 100 (38%), non-insulin, n = 32 (28%)]. The
Logistic regression analysis identified prior insulin therapy (OR 1.734; 95%CI: 1.012–2.969; p = 0.045), high baseline BMI (OR 1.063; 95%CI: 1.014–1.115; p = 0.011), and long duration of T2DM (OR 1.024; 95%CI: 1.006–1.042; p = 0.009) as significant factors
Table 1 – Patient demographics and disease characteristics at baseline. Continued
Discontinued
Entire group n Pre-treatment non-insulin/insulin treatment Sex (male/female) Age Duration of diabetes (year) Body mass index (kg/m2) Systolic blood pressure (mmHg) Diastolic blood pressure (mmHg) HbA1c (%)
248 82/166 147/101 59.3 11.1 13.8 12.8 27.2 5.1 128.4 15.2 74.5 12.1 7.9 1.6
132 32/100 84/48 58.6 13.2 17.6 13.6 28.6 6.5 126.8 14.5 72.9 12.6 8.2 1.5
0.074 0.408 0.659 0.001 0.079 0.421 0.289 0.013
Non-insulin group n Pre-treatment (naı¨ve/1/2/3/4 GLA) Sex (male/female) Age Duration of diabetes (year) Body mass index (kg/m2) Systolic blood pressure (mmHg) Diastolic blood pressure (mmHg) HbA1c (%)
82 8/24/19/14/17 53/29 55.8 11.3 10.4 8.4 29.1 5.2 129.4 14.3 75.5 11.8 8.4 1.7
32 1/6/3/5/17 20/12 53.3 13.2 13.9 14.2 32.1 8.4 128.1 13.0 76.2 12.0 8.0 1.4
0.002 0.832 0.510 0.497 0.074 0.639 0.747 0.379
Insulin group n Sex (male/female) Age Duration of diabetes (year) Body mass index (kg/m2) Systolic blood pressure (mmHg) Diastolic blood pressure (mmHg) HbA1c (%) Total daily insulin dosage (U/day) Bolus daily insulin dosage (U/day) Basal daily insulin dosage (U/day)
166 94/72 61.1 10.6 15.6 14.2 26.2 4.7 127.9 15.7 74.1 12.2 7.6 1.5 19.4 17.2 12.7 13.2 7.8 9.3
100 64/36 60.3 12.8 18.8 13.2 27.4 5.3 126.3 15.0 71.8 12.6 8.2 1.6 31.8 21.2 20.3 17.4 14.0 11.2
0.236 0.589 0.007 0.054 0.557 0.209 <0.001 <0.001 <0.001 <0.001
Values are mean SD; GLA: oral glucose-lowering agents.
p Value
454
diabetes research and clinical practice 106 (2014) 451–457
Table 2 – Logistic regression analysis of baseline characteristics for discontinuation of liraglutide in the entire group, treatment-naı¨ve patients, patients treated previously with glucose-lowering agents (GLA), and insulin-treated patients. Logistic regression analysis was performed by the force entry method. Odds ratio (95% CI)
Table 3 – Multiple regression analyses of baseline characteristics for the response to liraglutide in the treatment-naı¨ve patients, patients treated previously with glucose-lowering agents (GLA), and insulin-treated patients. Multiple regression analysis was performed by the force entry method. Partial regression coefficient (95% CI)
p Value
p Value
Entire group (n = 380) Pre-treatmenta Sexb Age Duration of diabetes Baseline BMI Baseline SBP Baseline DBP Baseline HbA1c
1.734 0.716 0.991 1.024 1.063 0.995 0.987 1.125
(1.012–2.969) (0.443–1.156) (0.968–1.015) (1.006–1.042) (1.014–1.115) (0.975–1.015) (0.963–1.012) (0.966–1.310)
0.045 0.172 0.454 0.009 0.011 0.596 0.297 0.129
Non-insulin group (n = 82) Sexb Age Duration of diabetes Baseline BMI Baseline SBP Baseline DBP Baseline HbA1c Number of GLAa
0.063 ( 0.636–0.509) 0.001 ( 0.03–0.031) 0.011 ( 0.043–0.022) 0.028 ( 0.027–0.083) 0.018 ( 0.006–0.042) 0.019 ( 0.053–0.015) 0.645 ( 0.821 to 0.468) 0.411 (0.183–0.639)
0.825 0.955 0.513 0.308 0.133 0.266 <0.001 0.001
Non-insulin group (n = 114) Sexb Age Duration of diabetes Baseline BMI Baseline SBP Baseline DBP Baseline HbA1c Number of GLAc
1.102 0.985 1.061 1.093 0.989 1.008 1.042 1.771
(0.413–2.942) (0.936–1.038) (1.011–1.112) (1.000–1.196) (0.945–1.035) (0.951–1.068) (0.732–1.485) (1.188–2.642)
0.846 0.578 0.015 0.051 0.631 0.795 0.819 0.005
Insulin group (n = 266) Sexb Age Duration of diabetes Baseline BMI Baseline SBP Baseline DBP Baseline HbA1c Bolus daily insulin dosage Basal daily insulin dosage
Insulin group (n = 166) Sexb Age Duration of diabetes Baseline BMI Baseline SBP Baseline DBP Baseline HbA1c Bolus daily insulin dosage Basal daily insulin dosage
0.192 ( 0.466–0.082) 0.003 ( 0.011–0.018) 0.013 (0.004–0.022) 0.082 (0.050–0.114) 0.001 ( 0.011–0.009) 0.007 ( 0.020–0.006) 0.480 ( 0.603 to 0.357) 0.003 ( 0.014–0.008) 0.002 ( 0.018–0.013)
0.169 0.643 0.007 <0.001 0.833 0.311 <0.001 0.602 0.770
0.639 0.991 1.014 1.005 0.994 0.985 1.128 1.017 1.043
(0.355–1.149) (0.962–1.021) (0.992–1.036) (0.939–1.076) (0.970–1.018) (0.956–1.014) (0.919–1.384) (0.995–1.038) (1.008–1.079)
0.135 0.548 0.204 0.879 0.600 0.305 0.248 0.126 0.015
a
a Dummy indicator variables were set with treatment-naı¨ve patients and patients treated previously with glucose-lowering agents as 0 and insulin-treated patients as 1. Therefore, the odds ratio of 1.734 means that discontinuation of liraglutide was higher in insulin-treated patients than non-insulin-treated patients. b Dummy indicator variables were set with male as 1 and female as 2. Accordingly, the odds ratio of 0.716 means that discontinuation of liraglutide tended to be lower in females, albeit statistically insignificant. c Number of different glucose-lowering agents as pre-treatment (sulfonylureas, thiazolidinediones, a-glucosidase inhibitors, metformins, glinides). Thus, the odds ratio of 1.771 means discontinuation of liraglutide was higher in patients using several glucose-lowering agents compared with those on a single glucose-lowering agent.
for discontinuation of liraglutide therapy (Table 2). For the non-insulin group, longer duration of T2DM (OR 1.061; 95%CI: 1.011–1.112; p = 0.015) and use of several oral glucose-lowering agents (OR 1.771; 95%CI: 1.188–2.642; p = 0.005) correlated with the discontinuation of liraglutide (Table 2). On the other hand, for the insulin group, discontinuation of liraglutide correlated significantly with basal ID (OR 1.043; 95%CI: 1.008–1.079; p = 0.015).
3.3.
Clinical parameters related to efficacy of liraglutide
We also analyzed the factors involved in the response to liraglutide in 248 patients who continued treatment (Table 3).
Number of different glucose-lowering agents as pre-treatment (sulfonylureas, thiazolidinediones, a-glucosidase inhibitors, metformins, glinides). Thus, the partial regression coefficient of 0.411 means that reduction in HbA1c correlated significantly with lower number of glucose-lowering agents used before study enrolment. b Dummy indicator variables were set with male as 1 and female as 2. Thus, the partial regression coefficients of 0.063 in the noninsulin group and 0.192 in the insulin group means that reduction in HbA1c did not correlate with sex.
Multiple regression analysis was used to identify patient characteristics that correlated with DHbA1c after 6-month treatment. In the non-insulin group, HbA1c level at baseline correlated significantly with HbA1c reduction ( 0.645; 95%CI: 0.821 to 0.468; p < 0.001). On the other hand, the use of several oral glucose-lowering agents in prior treatment correlated negatively and significantly with DHbA1c, i.e. the use of large number of oral medications before the present clinical trial correlated with deterioration of glucose control after 6-month liraglutide treatment (0.411; 95%CI: 0.183 to 0.639; p = 0.001). Conversely, a higher baseline HbA1c level correlated with higher reduction in HbA1c after 6 months of liraglutide treatment, reflecting improvement in glycemic control (Table 3). The mean DHbA1c after 6-month treatment was calculated for each type of prior glucose-lowering agents. The results showed 3% decrease in HbA1c level in patients naı¨ve to oral glucose-lowering agents, although there were only a few patients in this category. Further analysis showed that the larger the number of glucose-lowering agents used before study enrollment the smaller the reduction in HbA1c level (Fig. 1; Pearson product-moment correlation coefficient r = 0.536, p < 0.01).
diabetes research and clinical practice 106 (2014) 451–457
455
Fig. 1 – Reduction in HbA1c after 6-month treatment with liraglutide according to the number of different glucoselowering agents used previously (n = 66. HbA1c values at baseline and 6 months were available in 82 patients). Values are mean W SD, p < 0.01, Pearson’s correlation coefficient: 0.536.
Similarly, in the insulin group, the response to liraglutide correlated significantly with duration of T2DM, BMI, and HbA1c level at baseline. Liraglutide was less efficacious in patients with long T2DM duration and high BMI at baseline; however, the response to liraglutide correlated with high HbA1c level at baseline (Table 3). The same independent variables were assessed in the decision tree analysis. In the non-insulin group, the initial HbA1c level branched at 11.5%, and equal or greater values were associated with a significantly greater reduction in HbA1c; the mean reduction in HbA1c was 5.6% irrespective of patient characteristics (Fig. 2). When the baseline HbA1c level was <11.5%, the baseline HbA1c level branched at 8.35%. When the baseline HbA1c level was 8.35%, the mean reduction in HbA1c level was 2.19% when two or fewer oral glucose-lowering agents had been used, and the mean reduction in HbA1c level was 0.49% when three or more oral glucose-lowering agents had been used. When the baseline HbA1c level was <8.35%, the mean reduction in HbA1c level was 0.64% in patients who used one type of oral glucose-lowering agents, but no HbA1c reduction was seen in patients who used two or more types oral glucose-lowering agents (Fig. 2). Similar to the non-insulin group, the baseline HbA1c level in the insulin group was also a significant independent factor that correlated with the reduction in HbA1c after 6 months of liraglutide treatment. In patients with baseline HbA1c level of 9.8%, the mean HbA1c level decreased by 2.97%., whereas patients with baseline HbA1c level of <9.8% showed branching of baseline HbA1c level at 8.25%. Baseline HbA1c level of 8.25% resulted in a mean reduction in HbA1c level of 0.9% after 6 months of liraglutide treatment. In patients with baseline HbA1c level of <8.25%, branching with BMI was noted; BMI branched at 31.1 and <31.1. Patients with BMI of <31.1 kg/m2 showed no reduction in HbA1c. Deterioration of HbA1c was noted in patients with HbA1c <8.25% and BMI 31.1 kg/m2 (Fig. 2).
Fig. 2 – Decision tree analysis of patients background for prediction of the response to 6-month liraglutide therapy based on reduction in HbA1c level in (A) patients treated previously with glucose-lowering agents (n = 66, HbA1c values at baseline and 6 months were available in 82 patients), and (B) patients treated previously with insulin (n = 127, HbA1c values at baseline and 6 months were available in 166 patients). In these box-and-whisker plots, lines within the boxes represent median values; the upper and lower lines of the boxes represent the 25th and 75th percentiles, respectively; and the upper and lower bars outside the boxes represent the 90th and 10th percentiles, respectively.
4.
Discussion
Based on logistic regression analysis, multi regression analysis and decision tree analysis, the present study demonstrated that the therapeutic response to liraglutide is influenced by treatment regimens and number of oral glucose-lowering agents, and that the use of liraglutide at an early stage is more efficacious when used alone or in combination with SU. In Japan, liraglutide has been used in a variety of patients with T2DM, including patients who switch from insulin therapy. However, justification for such use of liraglutide
456
diabetes research and clinical practice 106 (2014) 451–457
requires further analysis. In fact, the only Japanese phase III clinical trials that tested the efficacy of liraglutide included only patients who were on diet/exercise therapy and/or treated with oral glucose-lowering agents [14–17]. Liraglutide is contraindicated for insulin-dependent type 1 or type 2 diabetes mellitus because it enhances insulin secretion. However, the largest group of patients in the present study was patients who switched from insulin therapy to liraglutide. Based on our results, it is anticipated that switching from insulin therapy to liraglutide could improve the quality of life and treatment compliance by reducing episodes of hypoglycemia, lowering body weight, and/or decreasing the number of drug injections. Further increase in the number of patients who switch from insulin to liraglutide could allow further analysis of the pathophysiology and prognosis of T2DM. Based on the finding that total and basal ID contributed to the discontinuation of treatment in the insulin group, we suggest that insulin dosage and endogenous insulin secretion capacity and insulin resistance should be considered before the introduction of liraglutide for the treatment of T2DM. But at this point in time, further intervention studies are needed to clarify the relationship between efficacy of liraglutide and insulin secretion capacity and insulin resistance. The present study has certain limitations. First, we did not evaluate insulin secretion capacity by using glucagon tolerance test or CPI. Several studies have suggested the usefulness of these two parameters as predictors for the efficacy of liraglutide [7–9,11]. However, our study examined patient characteristics linked to the practicing physician’s decision to continue or discontinue liraglutide since patients characteristics are easy tools for evaluation of efficacy of liraglutide in daily clinical practice. However, taking into consideration the results of this study, application of tests of endogenous insulin secretion capacity would be highly useful to ensure treatment efficacy and safety of patients switching from insulin to liraglutide. Such tests would be especially useful in patients with a high basal ID. Second, this study was retrospective in nature. Accordingly, discontinuation was based on the decision of the individual investigator. In other words, the criteria used for discontinuation, especially hyperglycemia, varied among the participating medical institutions. This limitation probably introduced some bias and thus a prospective study with predefined criteria of discontinuation based on glucose control is needed. Finally this study was multi-center analysis but included a small number of institutions, with limited sample size. Accordingly, prospective studies that include larger sample size and several institutions are needed to confirm the present results. Because the mean baseline BMI of patients of the noninsulin group was within the obesity range (27–28 kg/m2), the main reasons for the introduction of liraglutide were improvement of blood glucose level and weight loss. While the mean BMI of patients of the Japanese phase III clinical trial of liraglutide was 25 kg/m2, some patients in actual clinical practice are grossly obese and the dose of liraglutide per unit body weight is lower than the recommended clinical dose. This fact suggests that the recommended liraglutide dose could be less than ideal at least in some patients. In this regard; the approved dose in Japan is half the dose approved in Europe
and United States. Based on these discrepancies and the two actions of liraglutide on blood glucose regulation and body weight reduction, the recommended treatment regimen and dosage used should be reconsidered. Analysis of the correlation between patient characteristics and response to liraglutide demonstrated that, in the non-insulin group, liraglutide resulted in successful regulation of blood glucose levels, as measured by HbA1c levels, in both treatment-naı¨ve patients and patients who switched to liraglutide from other oral glucose-lowering drugs. In this study, which followed the treatment regimen approved for use in Japan, all oral glucose-lowering drugs other than SU were discontinued at study entry. Further studies are needed to analyze changes in efficacy-related factors when the indications for concurrent liraglutide therapy are expanded. Irrespective of disease duration, age, and sex, good response to liraglutide (reduction of HbA1c by 2%) was based on HbA1c level at baseline of 11.5% or 8.35% with the use of 2 oral glucose-lowering drugs in prior treatment (Fig. 2). However, in some cases, liraglutide treatment was effective even when insulin therapy was used in prior treatment. However, 20 insulin units/day used in these patients was a relatively low dosage, and in virtually all cases, liraglutide was administered with the aim of withdrawal from intensive insulin therapy. The present study identified duration of diabetes, baseline BMI, and baseline HbA1c levels as significant determinants of the response to liraglutide therapy in patients with long duration of diabetes. The results also demonstrated that the best response to liraglutide was achieved by patients with high HbA1c level at baseline. It is noteworthy that glucotoxicity itself is reported to have a major impact on the ability of GLP-1 to promote insulin secretion [18]. Therefore, careful attention should be exercised when introducing liraglutide in patients with HbA1c levels. Introduction of liraglutide therapy to patients with T2DM with near-normalization of blood glucose seems safe and potentially most efficacious. The present study demonstrated no improvement in patients with baseline HbA1c levels of <8.25% and in patients with baseline BMI of >31.1% who had been previously treated with insulin, even after 6 months of liraglutide therapy. This finding indicates that for patients to be considered for liraglutide treatment with the current regimen and dosage ranges in Japan, they should be examined comprehensively, and the number of different oral glucose-lowering drugs, insulin dose, and baseline HbA1c and BMI values should be taken into consideration in the decision making. With regard to co-treatment with liraglutide, three different statistical tests identified large number of glucoselowering agents as a significant determinant of discontinuation and efficacy of liraglutide. The present study was designed to comply with the Government-approved guidelines for use of liraglutide (monotherapy and combination with SU only). The indications for liraglutide could change in the future, and additional oral anti-diabetic medications may be approved for use in combination with liraglutide. In particular, combination with metformin and other incretinrelated drugs could theoretically enhance weight loss and increase active GLP-1 levels.
diabetes research and clinical practice 106 (2014) 451–457
Conflict of interest statement [9]
The authors declare no conflict of interest in relation to this article. [10]
Acknowledgment This study was sponsored by the Division of Nephrology, Endocrinology and Metabolism, Department of Internal Medicine (Grant No. 12F188139), Tokai University School of Medicine, Japan.
references
[1] Bell GI, Santerre RF, Mullenbach GT. Hamster preproglucagon contains the sequence of glucagon and two related peptides. Nature 1983;302:716–8. [2] Drucker DJ, Nauck MA. The incretin system: glucagon-like peptide-1 receptor agonists and dipeptidyl peptidase-4 inhibitors in type 2 diabetes. Lancet 2006;368:1696–705. [3] Ritzel R, Orskov C, Holst JJ, Nauck MA. Pharmacokinetic, insulinotropic, and glucagonostatic properties of GLP-1 [736 amide] after subcutaneous injection in healthy volunteers. Dose–response relationships. Diabetologia 1995;38:720–5. [4] Gutniak M, Orskov C, Holst JJ, Ahre´n B, Efendic S. Antidiabetogenic effect of glucagon-like peptide-1 (736)amide in normal subjects and patients with diabetes mellitus. N Engl J Med 1992;326:1316–22. [5] Larsen J, Hylleberg B, Ng K, Damsbo P. Glucagon-like peptide-1 infusion must be maintained for 24 h/day to obtain acceptable glycemia in type 2 diabetic patients who are poorly controlled on sulphonylurea treatment. Diabetes Care 2001;24:1416–21. [6] Knudsen LB, Nielsen PF, Huusfeldt PO, Johansen NL, Madsen K, Pedersen FZ, et al. Potent derivatives of glucagon-like peptide-1 with pharmacokinetic properties suitable for once daily administration. J Med Chem 2000;43:1664–9. [7] Usui R, Yabe D, Kuwata H, Fujiwara S, Watanabe K, Hyo T, et al. Retrospective analysis of safety and efficacy of insulin-to-liraglutide switch in Japanese type 2 diabetes: a caution against inappropriate use in patients with reduced b-cell function. J Diabetes Invest 2013;4:585–94. [8] Kozawa J, Inoue K, Iwamoto R, Kurashiki Y, Okauchi Y, Kashine S, et al. Liraglutide is effective in type 2 diabetic
[11]
[12]
[13]
[14]
[15]
[16]
[17]
[18]
457
patients with sustained endogenous insulin-secreting capacity. J Diabetes Invest 2012;3:294–7. Iwao T, Sakai K, Sata M. Postprandial serum C-peptide is a useful parameter in the prediction of successful switching to liraglutide monotherapy from complex insulin therapy in Japanese patients with type 2 diabetes. J Diabetes Complications 2013;27:87–91. Nambu T, Matsuda Y, Matsuo K, Kanai Y, Yonemitsu S, Muro S, et al. Liraglutide administration in type 2 diabetic patients who either received no previous treatment or were treated with an oral hypoglycemic agent showed greater efficacy than that in patients switching from insulin. J Diabetes Invest 2013;4:69–77. Worachartcheewan A, Nantasenamat C, Isarankura-NaAyudhya C, Pidetcha P, Prachayasittikul V. Identification of metabolic syndrome using decision tree analysis. Diabetes Res Clin Pract 2010;90:e15–8. Hische M, Luis-Dominguez O, Pfeiffer AF, Schwarz PE, Selbig J, Spranger J. Decision trees as a simple-to-use and reliable tool to identify individuals with impaired glucose metabolism or type 2 diabetes mellitus. Eur J Endocrinol 2010;163:565–71. Mo¨hlig M, Flo¨ter A, Spranger J, Weickert MO, Schill T, Schlo¨sser HW, et al. Predicting impaired glucose metabolism in women with polycystic ovary syndrome by decision tree modelling. Diabetologia 2006;49:2572–9. Seino Y, Rasmussen MF, Nishida T, Kaku K. Efficacy and safety of the once-daily human GLP-1 analogue, liraglutide, vs glibenclamide monotherapy in Japanese patients with type 2 diabetes. Curr Med Res Opin 2010;26:1013–22. Kaku K, Rasmussen MF, Nishida T, Seino Y. Fifty-two-week, randomized, multicenter trial to compare the safety and efficacy of the novel glucagon-like peptide-1 analog liraglutide vs glibenclamide in patients with type 2 diabetes. J Diabetes Invest 2011;2:441–7. Kaku K, Rasmussen MF, Clauson P, Seino Y. Improved glycaemic control with minimal hypoglycaemia and no weight change with the once-daily human glucagon-like peptide-1 analogue liraglutide as add-on to sulphonylurea in Japanese patients with type 2 diabetes. Diabetes Obes Metab 2010;12:341–7. Seino Y, Rasmussen MF, Nishida T, Kaku K. Glucagon-like peptide-1 analog liraglutide in combination with sulfonylurea safely improves blood glucose measures vs sulfonylurea monotherapy in Japanese patients with type 2 diabetes: results of a 52-week, randomized, multicenter trial. J Diabetes Invest 2011;2:280–6. Højberg PV, Vilsbøll T, Rabøl R, Knop FK, Bache M, Krarup T, et al. Four weeks of near-normalisation of blood glucose improves the insulin response to glucagon-like peptide-1 and glucose-dependent insulinotropic polypeptide in patients with type 2 diabetes. Diabetologia 2009;52:199–207.