Colesevelam Hydrochloride Therapy in Patients With Type 2 Diabetes Mellitus Treated With Metformin:  Glucose and Lipid Effects FREE

Colesevelam hydrochloride is a bile acid sequestrant that was approved as a cholesterol-lowering agent in the United States in 2000 and approved in January 2008 to improve glycemic control in adults with type 2 diabetes mellitus (T2DM).¹ A pilot study suggested that colesevelam improved glycemic control in patients with T2DM, resulting in a mean reduction in hemoglobin A_1c (HbA_1c) level of 0.5% compared with placebo (P = .007).² Furthermore, in patients with T2DM with a higher baseline HbA_1c level (≥8.0%), colesevelam reduced mean HbA_1c level by 1.0% (P = .002 vs placebo).² The present report describes a larger trial to assess the efficacy and safety of colesevelam for improving glycemic control in patients with T2DM not adequately controlled by a stable regimen of metformin-based therapy.

This was a 26-week, randomized, double-blind, placebo-controlled, parallel-group study conducted at 54 sites in the United States and 2 sites in Mexico. The study protocol was conducted under the guidelines outlined by institutional review board regulations, the ethical principles of Good Clinical Practice Guidelines, and the fourth amendment of the Declaration of Helsinki. All patients documented undergoing the informed consent process by signing an informed consent document before participation.

This study randomized 316 patients with T2DM aged 18 to 75 years with inadequate glycemic control (HbA_1c level, 7.5%-9.5%, inclusive), taking a stable dose (for ≥90 days) of metformin monotherapy or metformin in combination with other oral anti–diabetes mellitus (DM) drugs (including 1 or more of the following: sulfonylureas, thiazolidinediones, α-glucosidase inhibitors, and/or meglitinides; glucagonlike peptide-1 analogues and dipeptidyl peptidase IV inhibitors were not included because these agents were not approved by the Food and Drug Administration prior to the development of the study protocol). Patients were to have been prescribed an appropriate T2DM diet, although no specific, protocol-directed dietary evaluation or dietary recommendations were made during the course of the trial (such as would occur through protocol-directed visits with a dietitian). Patients were excluded for any of the following reasons: body mass index greater than 45 (calculated as weight in kilograms divided by height in meters squared); low-density lipoprotein cholesterol (LDL-C) level lower than 60 mg/dL (to convert to millimoles per liter, multiply by 0.0259); triglyceride (TG) level greater than 500 mg/dL (to convert to millimoles per liter, multiply by 0.0113); uncontrolled hypertension (defined as systolic blood pressure >160 mm Hg and diastolic blood pressure >95 mm Hg); history of type 1 DM, ketoacidosis, dysphagia, swallowing disorders, or intestinal motility disorders; treatment with colesevelam within 8 weeks of the screening visit; long-term or recently initiated insulin therapy; or treatment with oral corticosteroids, cholestyramine, or colestipol. Patients were also excluded if they had a history of an acute coronary syndrome, coronary intervention and/or transient ischemic attack within 3 months of the screening visit, severe peripheral vascular disease, any serious disorder that might interfere with the study or affect interpretation of results, or any condition which, in the investigator's opinion, made it inappropriate for the patient to participate.

Oral contraceptives, hormone therapy, thyroid therapy, and other lipid-altering drugs (such as 3-hydroxy-3-methylglutaryl coenzyme A [HMG-CoA] reductase inhibitors, fibrates, niacin, fish oils, and cholesterol absorption inhibitors) were permitted, provided a stable dose had been maintained for 30 or more days prior to the initiation of the trial, and dosage changes were not anticipated. Women of child-bearing potential were allowed to participate as long as they engaged in an acceptable form of birth control.

Subjects were withdrawn from the study if HbA_1c level was 10.0% or greater at any study visit or if fasting plasma glucose (FPG) level from a routine visit was greater than 260 mg/dL (to convert to millimoles per liter, multiply by 0.0555) or lower than 60 mg/dL and was confirmed within 3 days. The management, reporting, and actions taken in response to hypoglycemia were left to the medical judgment of the blinded study investigators.

After agreeing to participate in the study, as documented by a signed informed consent, subjects underwent a 1-week screening period to determine eligibility and then entered a 2-week, single-blind, placebo run-in period. During the placebo run-in period, subjects took 6 colesevelam-matching placebo tablets daily, composed of magnesium stearate and microcrystalline cellulose, with a commercially supplied film-coating mixture. Subjects were instructed to take the study medication according to their preference (either 3 tablets with noon and evening meals or all 6 tablets with the evening meal) and were encouraged to follow the same dosing regimen throughout the trial.^1,3 After successful completion of the placebo run-in period, qualifying subjects were then randomized 1:1 to colesevelam hydrochloride, 3.75 g/d (6 tablets: 625 mg per tablet) or matching placebo for 26 weeks of double-blind treatment. Subjects continued taking their prestudy oral anti-DM drug(s) at the same dose and time(s) as before the start of the study.

The primary efficacy parameter was the mean change from baseline in HbA_1c level for the active drug compared with placebo at week 26 with subjects analyzed on an intent-to-treat (ITT) basis using a last-observation-carried-forward (LOCF) approach. Secondary efficacy parameters included the mean change in HbA_1c, FPG, and fructosamine levels from baseline to weeks 6, 12, 18, and 26. Mean change in HbA_1c level from baseline to weeks 6, 12, 18, and 26 was also analyzed for both the metformin monotherapy and metformin combination therapy cohorts. An additional secondary efficacy parameter included an assessment of subjects who experienced a predefined reduction in FPG level of 30 mg/dL or greater or in HbA_1c level of 0.7% or greater from baseline at week 26. Finally, other secondary end points included mean change in C-peptide, adiponectin, and insulin levels and homeostasis model assessment (HOMA) index from baseline to week 26; mean change and mean percentage change in concentrations of total cholesterol (TC), LDL-C, high-density lipoprotein cholesterol (HDL-C), non–HDL-C, apolipoprotein (apo) A-I, and apo B from baseline to week 26; mean change in TC/HDL-C, LDL-C/HDL-C, non–HDL-C/HDL-C, and apo B/apo A-I ratios from baseline to week 26; and median change and median percentage change in high-sensitivity C-reactive protein (hsCRP) and TG levels from baseline to week 26.

Laboratory panels were obtained under fasting conditions, and tests were performed by a certified laboratory (Medical Research Laboratories International, Highland Heights, Kentucky). Total cholesterol and TG levels were measured by enzyme assay. High-density lipoprotein cholesterol level was measured by cholesterol oxidase assay of the supernatant from the precipitate of non–high-density lipoproteins with heparin and manganese chloride. Apolipoprotein A-I, apo B, and hsCRP levels were quantitated by immunonephelometry. The method used to calculate LDL-C level was based on the TG concentration at screening; the Friedewald equation was used for subjects with a TG level of 400 mg/dL or lower, while the direct (β quantification) method was used for subjects with a TG level greater than 400 mg/dL and lower than 500 mg/dL. The method of LDL-C determination used at the screening visit was used throughout the study for an individual subject, regardless of changes in TG level. Therefore, analysis of mean LDL-C level was a combination of both calculated LDL-C and direct LDL-C levels.

Safety assessments included treatment-emergent adverse experiences (AEs), clinical laboratory blood test results, changes in vital signs, and findings on physical examinations. Compliance with the medication regimen was evaluated by counting unused tablets at each study visit.

The ITT population included all randomized subjects who took 1 or more doses of randomized study medication, had a baseline HbA_1c measurement, and 1 or more postbaseline HbA_1c measurements. The ITT population was the primary analysis population for the analysis of all efficacy parameters. Additional separate analyses were conducted for 2 cohorts defined in the protocol: subjects receiving a background therapy of metformin alone (metformin monotherapy) and those receiving a background therapy of metformin plus other oral anti-DM drugs (metformin in combination with other oral anti-DM drugs). The safety population included all randomized subjects who received 1 or more doses of randomized study medication.

This study required a total of 300 randomized subjects and had an 81% to a higher than 95% power to detect a difference of a 0.54% to 0.80% reduction in mean HbA_1c level from baseline between colesevelam and placebo (with a 2-sided type I error of .05), assuming a common standard deviation of 1.5% or lower and a maximum dropout rate of 15%.

The baseline parameters in the 2 treatment groups were compared by a 1-way analysis of variance (ANOVA) model with treatment as a factor for continuous variables and by the Fisher exact/Fisher-Freeman-Halton test for categorical variables. An analysis of covariance (ANCOVA) model with treatment and concomitant oral anti-DM drug status as fixed effects and baseline HbA_1c level as a covariate was used to evaluate the treatment effect. The normality assumption of the efficacy data was examined prior to fitting the ANCOVA models. When a significant departure from normality was observed, a nonparametric equivalent of ANCOVA (rank analysis of covariance) was applied.

The treatment effect in change in HbA_1c level from baseline to week 26 was estimated using least squares (LS) mean, standard error, 2-tailed 95% confidence intervals (CIs), and 2-sided P value. The LS means, standard errors, and the 2-tailed 95% CIs for each treatment group were also estimated. All statistical tests were considered significant at the .05 (2-sided) level. Secondary and other efficacy parameters were compared with the same statistical methodology unless otherwise noted.

Glycemic control response rates were tabulated and compared using Pearson χ² test. Median change and median percentage change in hsCRP and TG levels from baseline to week 26 were analyzed using a nonparametric ANCOVA model. The treatment difference was estimated by the Hodges-Lehmann estimator, and a 2-tailed 95% CI was obtained using the method of Moses. The statistical analysis plan was finalized prior to data unblinding. Statistical Analysis Software version 8 (SAS Institute Inc, Cary, North Carolina) was used for all statistical tests.

This study was conducted between August 2004 and July 2006. A total of 1009 subjects were screened, with 332 subjects entering the placebo run-in period (Figure 1). In total, 316 subjects were randomized (159 to colesevelam and 157 to placebo). Baseline demographic characteristics for the total population are summarized in Table 1, which revealed no significant differences between the colesevelam and placebo groups at randomization. When other oral anti-DM drugs were used in combination with background metformin, most were sulfonylureas and thiazolidinediones (Table 1).

Forty-three subjects in the colesevelam group withdrew prior to study completion compared with relative to 51 subjects in the placebo group (Figure 1). Seven subjects in the colesevelam group discontinued treatment because of any hyperglycemia-related issue compared with 20 subjects in the placebo group. Only 2 subjects in the colesevelam group compared with 10 in the placebo group were withdrawn from the study because of protocol-specified hyperglycemia (FPG level >260 mg/dL and/or HbA_1c level ≥10.0%). One subject in the colesevelam group and 4 in the placebo group were withdrawn from the study because of changes to their oral anti-DM drug regimen. The remaining subjects who discontinued treatment because of hyperglycemia did so because either they or the investigator assessed that control was not adequate to justify continued participation in a placebo-controlled study. No subject discontinued treatment because of hypoglycemia. Eight subjects in the colesevelam group discontinued treatment because of AEs compared with 4 in the placebo group. Finally, 28 subjects in the colesevelam group and 27 subjects in the placebo group discontinued treatment for other reasons, such as withdrawal of consent.

Compliance with study medication was similar between the groups during double-blind treatment (93.3% for the colesevelam group and 91.9% for the placebo group).

Regarding the primary end point, treatment with colesevelam for 26 weeks reduced mean HbA_1c level relative to placebo. In the total population, colesevelam reduced LS mean HbA_1c level by 0.39% by the week 26 LOCF end point (week 26 LOCF), while placebo was associated with an increase of 0.15%, resulting in a significant LS mean treatment difference of −0.54% (P < .001) (Figure 2 and Table 2). A significant LS mean treatment difference was observed as early as week 6 between the colesevelam and placebo groups (−0.46%; P < .001).

Mean reductions in HbA_1c level when colesevelam was added to either metformin monotherapy or metformin in combination with other oral anti-DM drugs were consistent with findings for the total population. At week 26 LOCF, colesevelam added to metformin monotherapy reduced LS mean HbA_1c level by 0.44% compared with an increase of 0.02% with added placebo, resulting in an LS mean treatment difference of −0.47% (P = .002) (Figure 2). Similarly, colesevelam added to metformin in combination with other oral anti-DM drugs reduced LS mean HbA_1c level by 0.35% compared with an increase of 0.27% with added placebo, resulting in an LS mean treatment difference −0.62% (P < .001) (Figure 2).

As a follow-up analysis within the total population, subjects with an HbA_1c level of 8.0% or lower at baseline and subjects with an HbA_1c level higher than 8.0% at baseline were analyzed separately. In the subgroup with an HbA_1c level of 8.0% or lower at baseline, HbA_1c level decreased in the colesevelam group by −0.24% compared with an increase of 0.24% in the placebo group, resulting in an LS mean treatment difference of −0.49% (P = .002). A greater effect was observed in the subgroup with an HbA_1c level higher than 8.0% at baseline; the mean change in HbA_1c level in the colesevelam group was −0.54% compared with an increase of 0.07% in the placebo group (LS mean treatment difference: −0.60%; P < .001).

Regarding secondary glycemic end points, colesevelam reduced FPG level compared with placebo at week 26 LOCF (−13.9 mg/dL; P = .01) (Table 2), with a significant LS mean treatment difference observed at week 6 (−20.8 mg/dL; P < .001). Consistent with effects on HbA_1c level, colesevelam reduced fructosamine level compared with placebo at week 26 LOCF (−23.2 μmol/L; P < .001; Table 2), with a significant LS mean treatment difference reported by 6 weeks (−25.5 μmol/L; P < .001).

In total, 71 subjects (47.7%) in the colesevelam group and 54 subjects (35.5%) in the placebo group experienced either a reduction in FPG level of 30 mg/dL or greater or in HbA_1c level of 0.7% or greater from baseline at week 26 LOCF (P = .03). A significantly greater percentage of subjects in the colesevelam group compared with placebo achieved an HbA_1c level reduction of 0.7% or greater (57 [38.3%] vs 31 [20.4%]; P < .001). In fact, 33 subjects (22.3%) achieved an HbA_1c level lower than 7.0% in the colesevelam group compared with only 9 subjects (5.9%) in the placebo group by study end.

Compared with placebo, colesevelam reduced LS mean and mean percentage LDL-C, TC, non–HDL-C, and apo B levels at week 26 LOCF (P < .001 for all), with no statistically significant effect on HDL-C or apo A-I levels compared with placebo at week 26 LOCF (Table 3 and Figure 3). Compared with placebo, colesevelam was not associated with a significant LS median treatment difference in TG level at week 26 LOCF (8.5 mg/dL; P = .24) (Figure 3). Consistent effects on the lipid profile were observed in those subjects who received concomitant statin treatment; mean percentage LDL-C, TC, and non–HDL-C concentration were significantly reduced, while HDL-C concentration was unchanged. No significant median increase in TG level was observed with colesevelam or placebo.

Finally, compared with placebo, colesevelam reduced LS mean TC/HDL-C, LDL-C/HDL-C, non–HDL-C/HDL-C, and apo B/apo A-I ratios at week 26 LOCF, as shown in Figure 3 (P < .003 for all).

Colesevelam did not produce a significant LS mean treatment difference for C-peptide level compared with the placebo group at week 26 LOCF (−0.1 ng/mL [to convert to nanomoles per liter, multiply by 0.331]; P = .54) (Table 3). Similarly, compared with placebo, colesevelam was not associated with a significant LS mean treatment difference in adiponectin (−0.3 μg/mL; P = .52), insulin (−0.9 μIU/mL [to convert to picomoles per liter, multiply by 6.945]; P = .51), or the HOMA index (−0.3; P = .68) at week 26 LOCF (Table 3).

Colesevelam treatment resulted in an LS median treatment difference in hsCRP level of −0.40 mg/L (to convert to nanomoles per liter, multiply by 9.524) (−14.4%; P = .02; Table 3) compared with placebo at week 26 LOCF.

Overall, colesevelam treatment was safe and generally well tolerated when added to metformin-based therapy in subjects with T2DM.

Adverse experiences were assessed by blinded investigators. Adverse experience intensity was described as mild, moderate, or severe, and AE causality was described as definitely, probably, possibly, probably not, and definitely not due to study drug. Adverse experiences judged as definitely, probably, or possibly due to study drug were designated as drug related. In this trial, no subject experienced a severe AE or serious AE that was thought to be drug related. No deaths occurred.

Overall, 8 subjects (5.0%) in the colesevelam group and 4 (2.5%) in the placebo group withdrew owing to an AE (Table 4). Six of the AEs in the colesevelam group were thought to be drug-related and included gastroesophageal reflux disease (1 subject), constipation (2 subjects), abdominal pain (2 subjects), and abdominal pain with vomiting and pyrexia (1 subject). Two AEs in the (blinded) placebo group were thought to be drug related and included abdominal distension (1 subject) and dyspepsia (1 subject).

Treatment-emergent AEs occurring in at least 5% of subjects are summarized in Table 4. Drug-related treatment-emergent AEs occurring in at least 5% of subjects included only constipation (6.9%) in the colesevelam group.

No clinically relevant changes in safety laboratory parameters or vital signs were noted in either group. No weight gain was noted in either group by study end; mean (SD) weight modestly decreased by 0.5 (2.89) kg and 0.3 (2.44) kg in the colesevelam and placebo groups, respectively, at week 26. One subject in the colesevelam group experienced mild hypoglycemia, defined as an FPG level lower than 60 mg/dL, which resolved, and the subject continued the trial.

In patients with T2DM with inadequately controlled glucose levels while taking a metformin-based treatment regimen, the addition of colesevelam reduced HbA_1c level by 0.54% at week 26 compared with placebo (P < .001). Therapeutic effects tended to be larger in subjects with higher baseline HbA_1c values but were consistent across subjects using metformin as background monotherapy and in combination with other oral anti-DM drugs. In addition, colesevelam significantly reduced LDL-C, TC, non–HDL-C, and apo B levels and improved various lipid ratios. No significant increase in TG level was observed with colesevelam compared with placebo. The significant reduction in hsCRP levels found in this trial is a finding consistent with prior studies of colesevelam administered either as monotherapy⁴ or in combination with statins.⁵

A strength of this study was that the number of subjects were substantially more than previous post hoc or pilot studies. A potential weakness was that this study was not designed to assess clinical outcomes, such as macrovascular or microvascular disease.⁶ Another potential weakness was that more subjects were withdrawn from the study because of hyperglycemia in the placebo group (20 subjects) than in the colesevelam group (7 subjects). This may have blunted further rises in HbA_1c level in the placebo group at week 26, thus attenuating the placebo-corrected HbA_1c-lowering effect of colesevelam. In addition, capping the entry HbA_1c level at 9.5% or lower (instead of ≥10.0%) may have also attenuated the between-group treatment difference in HbA_1c level. Allowing subjects with a higher baseline HbA_1c level to participate in the study would have likely increased the baseline HbA_1c level and thus resulted in greater potential for HbA_1c level reductions with colesevelam treatment.⁷ Otherwise, the glucose-lowering effects found with colesevelam in this study are similar to the degree of glucose lowering found with other bile acid sequestrants^2,8 and may be similar to other anti-DM agents when baseline HbA_1c level is taken into account.⁷

The predominant mechanism(s) accounting for the glucose-lowering effects of bile acid sequestrants, such as colesevelam, are unclear and have previously been reviewed.^9- 10 Proposed mechanisms have included effects related to altered intestinal bile acid composition, incretin effects (such as inhibition of cholecystokinin release), alterations in hepatocyte nuclear factor 4α expression, and both direct and indirect effects on farnesoid X receptors. The effect of colesevelam on incretins such as glucagonlike peptide-1 (GLP-1) has yet to be reported. However, because bile acids may increase GLP-1 secretion, the binding and increased fecal excretion of bile acids with colesevelam treatment may decrease the bile acid pool, decrease bile acid secretion, and theoretically reduce the secretion of GLP-1. Thus, until more information is known, any role of GLP-1 as a contributing mechanism for the glucose-lowering effect of colesevelam remains unknown.⁹

Overall, because bile acid sequestrants are not systemically absorbed, it is reasonable to hypothesize that any significant metabolic effects of these agents are related to their binding of bile acids¹¹ or alterations in bile acid metabolism. One of the more intriguing possibilities involves the interaction of bile acids with nuclear receptors. Bile acids are endogenous ligands for intestinal farnesoid X receptors. Binding of bile acids by a bile acid sequestrant would be expected to diminish or deactivate farnesoid X receptor activity. Farnesoid X receptor activation induces expression of the small heterodimer partner, which inhibits liver X receptor activity. When colesevelam binds bile acids, farnesoid X receptor and small heterodimer partner activity is decreased. Liver X receptor activity would then be expected to increase in response to reduced inhibition by small heterodimer partner.

Liver X receptor has been described as a glucose sensor,¹² with increasing liver X receptor activity resulting in reduced glucose levels.¹³ Thus, diminished farnesoid X receptor activity with bile acid sequestrants may indirectly increase liver X receptor activity, which in turn lowers glucose levels. This proposed mechanism has appeal in that liver X receptor agonist is known to reduce glucose and LDL-C levels and promote the seemingly paradoxical and simultaneous increase in HDL-C and TG levels, which are all metabolic effects found with bile acid sequestrants. However, it is interesting to note in this study that TG levels were not significantly increased with colesevelam treatment.

In conclusion, optimizing glucose and LDL-C levels are important treatment goals for patients with T2DM.^14- 15 This study provides support for colesevelam as a generally well tolerated and efficacious treatment when used in combination with an established metformin-based anti-DM treatment regimen in patients with T2DM with inadequate glycemic control.

Correspondence: Harold E. Bays, MD, Louisville Metabolic and Atherosclerosis Research Center Inc, 3288 Illinois Ave, Louisville, KY 40213 (hbaysmd@aol.com).

Accepted for Publication: April 14, 2008.

Author Contributions: Dr Bays had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis. Study concept and design: Truitt and Jones. Acquisition of data: Bays and Truitt. Analysis and interpretation of data: Bays, Goldberg, Truitt, and Jones. Drafting of the manuscript: Truitt and Jones. Critical revision of the manuscript for important intellectual content: Bays, Goldberg, Truitt, and Jones. Obtained funding: Truitt. Administrative, technical, and material support: Truitt and Jones. Study supervision: Bays, Truitt, and Jones.

Financial Disclosure: Dr Bays has served as a clinical investigator for and has received research grants from Abbott, Alteon, Arena, AstraZeneca, Aventis, Bayer, Boehringer Ingelheim, Boehringer Mannheim, Bristol-Myers Squibb, Ciba Geigy, Eli Lilly, Esperion, Fujisawa, GelTex, Genentech, GlaxoSmithKline, Hoechst Roussel, Hoffman LaRoche, InterMune, Kos Pharmaceuticals, Kowa, Lederle, Marion Merrell Dow, Merck, Merck/Schering Plough, Miles, Novartis, Parke Davis, Pfizer, Pliva, Purdue, Reliant, Roche, Rorer, Regeneron, Sandoz, Sankyo, Sanofi-Aventis, Searle, Shionogi, Schering Plough, SmithKline Beecham, Takeda, TAP Pharmaceutical Products, Upjohn, Upsher Smith, Warner Lambert, and Wyeth-Ayerst and has served as a consultant, speaker, and/or advisor to Arena, AstraZeneca, Aventis, Bayer, Bristol-Myers Squibb, Kos Pharmaceuticals, Merck, Merck/Schering Plough, Metabasis Therapeutics, Microbia, Novartis, Nicox, Ortho-McNeil, Parke Davis, Pfizer, Roche, Sandoz, Daiichi Sankyo Inc, Sanofi-Aventis, Schering Plough, SmithKline Beacham, Takeda, Upjohn, and Warner Lambert. Dr Goldberg has been a speaker and received honoraria from Eli Lilly, Takeda, Pfizer, Merck, Merck/Schering Plough, Kos Pharmaceuticals, Astra Zeneca, and Abbott; received research grants from Novo Nordisk, Pfizer, Merck, Kos Pharmaceuticals, Astra Zeneca, and Daiichi Sankyo Inc; and has been a consultant and received honoraria from Eli Lilly/Takeda, Pfizer, Merck, Merck/Schering Plough, Astra Zeneca, and Abbott. Drs Truitt and Jones are employees of Daiichi Sankyo, Inc.

Funding/Support: This study was funded by Daiichi Sankyo, Inc.

Role of the Sponsor: The sponsor designed and conducted the study and played a role in interpretation of the data and in the preparation and approval of the manuscript.

Additional Contributions: Editorial assistance was provided by Karen Stauffer, PhD.