Bezafibrate for Dyslipidemic Patients With Diabetes
We enrolled 6,449 patients and collected 6,407 case report forms. Safety was evaluated in 5,978 patients. Patients were excluded from safety analysis for the following reasons, namely 1) protocol violation: some patients did not receive bezafibrate (n = 13), had a delay in enrollment (n = 292), and had a history of bezafibrate treatment (n = 20); and 2) insufficient data for safety analysis (n = 104). A total of 429 of 6407 patients were excluded. Patients were excluded from efficacy analysis for the following reasons: 1) protocol violation: patients had neither TG ≥ 1.7 mmol/l nor TC ≥ 5.7 mmol/l (n = 920), or patients had neither diabetes nor FBG ≥ 6.1 mmol/l (n = 653, overlapping); 2) insufficient data for efficacy analysis: dose or duration of bezafibrate unknown (n = 86); data incomplete for baseline or after administration for any one of TC, TG, and HbA1c (n = 2,387); pretreatment for hyperlipidemia unknown or pretreatment with diabetes drugs, or both (n = 783); unknown concomitant drug for treating hyperlipidemia or unknown diabetes drug (n = 6); or both (these cases overlapped, resulting in 2,662 cases being excluded from safety analyses).
Patients' backgrounds in the efficacy analysis group are shown in Table 1. The number of patients with diabetes was 3,312 (99.9%), of whom 3,235 (97.6%) were diagnosed with type 2 diabetes. Mean duration of diabetes was 6.6 ± 5.9 years. A total of 602 (18.2%) and 1,977 (59.6%) patients had been previously treated for dyslipidemia and diabetes, respectively. Medication characteristics of the efficacy analysis group showed that while most patients (3,149; 95.0%) were treated for dyslipidemia with bezafibrate alone, 167 patients (5.0%) took other lipid-lowering drugs concurrently (Table 2).
All patients were classified into four subgroups according to diabetes drug administration before and during the study as follows: diabetes drug not taken before or during the study (n = 818, 24.7%), taken only during the study (n = 521, 15.7%), taken only before the study (n = 106, 3.2%) and taken before and during the study (n = 1,871, 56.4%). Thus, a diabetes drug was taken by 2,392 patients (72.1%) during the study. The most frequently used diabetes drug regimen was single administration of sulfonylureas (22.1%) followed by combined use of sulfonylureas and an α-glucosidase inhibitor (6.9%) (Table 2). The number of patients receiving both monotherapy and combination use of diabetes drugs was as follows: sulfonylureas (n = 1,505, 45.4%), glinides (n = 341, 10.3%), α-glucosidase inhibitor (n = 685, 20.7%), biguanide (n = 550, 16.6%), thiazolidine (n = 287, 8.7%), and insulin (n = 284, 8.6%) (data not shown). All of 3,316 patients took bezafibrate at least 6 weeks, of whom 3,299 (99.5%) took the drug for at least 12 weeks and 17 (0.5%) for at least six weeks. The dose of bezafibrate was 400 mg/day in 2,728 (82.3%) patients and 200 mg/day in 472 (14.2%).
Table 3 presents the changes in lipid and glucose parameters at baseline and after bezafibrate administration. All lipid values except LDL-C improved significantly compared with baseline values. Significant decreases from baseline were observed for TC, TG, non-HDL-C, and TG/HDL-C ratio; in contrast, a significant increase was observed in HDL-C. Although LDL-C levels decreased significantly in patients with LDL-C ≥ 3.6 mmol/l at baseline (n = 737), they increased significantly in those with LDL-C < 3.6 mmol/l at baseline (n = 1,489). Thus, LDL-C levels increased significantly overall. Moreover, HbA1c, FBG, and HOMA-R levels decreased significantly.
Table 4 summarizes the changes in HbA1c concentrations after treatment, classified according to use of diabetes drug and baseline HbA1c levels. We analyzed 2,086 patients whose baseline HbA1c values were ≥7.0%. HbA1c levels decreased significantly by -0.76% from baseline in all patients. This change was greater in patients whose baseline HbA1c levels were higher. The trend analysis showed significant differences between the HbA1c-subgroups. Similar results were observed in all medication subgroups classified by diabetes drug administration before and during the study.
There were 1,464 patients for which we could determine the change in HbA1c levels as a function of duration of diabetes (Table 5). HbA1c levels significantly decreased in all diabetes-duration subgroups, most strongly in the subgroup with diabetes for < 1 year. Trend analysis showed significant differences between the diabetes-duration subgroups. Similar results were also observed in the medication subgroup of patients who were not administrated diabetes drug before or during the study.
We also analyzed HbA1c levels according to concomitant use of a single diabetes drug during the study. Changes in HbA1c levels for each drug were as follows: sulfonylureas, n = 536, −0.84%, p < 0.001; glinides, n = 125, −0.82%, p < 0.001; alpha glucosidase inhibitor, n = 70, −0.66%, p < 0.001; biguanide, n = 239, −0.73%, p < 0.001; thiazolidine, n = 28, −0.64%, p = 0.004; insulin: n = 130, −0.59%, p < 0.001. Each regimen showed a significant decrease in HbA1c. There were no significant differences between any of the regimens (p = 0.363) (data not shown).
We conducted univariate and multivariate regression analyses to evaluate the influence of bezafibrate on HbA1c levels. For initial univariate analysis of 1,854 patients, the difference in HbA1c levels was as an objective variable. Explanatory variables were patient age, sex, baseline BMI, baseline HbA1c level, diabetes-drug administration during the study, duration of diabetes, and rate of change in levels of each lipid (TC, LDL-C, TG, and HDL-C). Based on these results, baseline BMI, baseline HbA1c, diabetes-drug administration, rate of change in TG, HDL-C, and TC were found to be significant variables.
We next performed multivariate regression analysis using these variables and found that the estimated influence rate per unit change in each variable on HbA1c was as follows: baseline HbA1c, -0.489; baseline BMI, 0.016; diabetes-drug administration, 0.176; rate of change in TG, 0.004; and rate of change in TC, 0.005. The rate of change in HDL-C was not a significant variable (Table 6). Further, the average rates of change in TG and TC for all subjects at 24 weeks were -45.4% and -5.4%, respectively (Table 3). When the rate of change in each lipid was considered, their estimated influences on altering HbA1c levels were -0.182% and -0.027%, respectively. Next, we analyzed the relationship between rate of change in TG and amount of change in HbA1c in 3,316 patients (Figure 1). Irrespective of diabetes drug administration during the study, a strong positive relationship was observed between the rates of change in TG and amount of change in HbA1c levels.
(Enlarge Image)
Figure 1.
TG and HbA1c levels in patients depending on use of a diabetes drug. The graphs show the changes in HbA1c levels determined for all patients (•), patients taking a diabetes drug (▵), and patients not taking a diabetes drug (□).
Drug safety was assessed for 5,978 patients. ADRs were experienced by 306 patients (5.1%) (Table 7). The most common ADRs were increased blood creatine phosphokinase (0.8%), blood creatinine (0.8%), blood urea (0.5%), renal impairment (0.3%), and aspartate aminotransferase (0.3%). Severe ADRs were experienced by 17 patients (0.28%) and mainly included stroke (n = 2, 0.03%), renal impairment (n = 2, 0.03%), elevated blood urea nitrogen (n = 2, 0.03%), abnormal hepatic function (n = 1, 0.02%), pneumonia (n = 1, 0.02%), rhabdomyolysis (n = 1, 0.02%), hypoglycemia (n = 1, 0.02%), and others. ADR rate increased slightly in patients concurrently taking diabetes drugs (4.3% not taking vs. 5.6% taking), but no apparent clinical differences were observed. One hypoglycemia case occurred in a patient concurrently taking a diabetes drug. Rhabdomyolysis was observed in five patients, four of whom were treated with a diabetes drug. No ADRs were observed that could be unequivocally ascribed to concurrent diabetes drug use.
Results
Patients
We enrolled 6,449 patients and collected 6,407 case report forms. Safety was evaluated in 5,978 patients. Patients were excluded from safety analysis for the following reasons, namely 1) protocol violation: some patients did not receive bezafibrate (n = 13), had a delay in enrollment (n = 292), and had a history of bezafibrate treatment (n = 20); and 2) insufficient data for safety analysis (n = 104). A total of 429 of 6407 patients were excluded. Patients were excluded from efficacy analysis for the following reasons: 1) protocol violation: patients had neither TG ≥ 1.7 mmol/l nor TC ≥ 5.7 mmol/l (n = 920), or patients had neither diabetes nor FBG ≥ 6.1 mmol/l (n = 653, overlapping); 2) insufficient data for efficacy analysis: dose or duration of bezafibrate unknown (n = 86); data incomplete for baseline or after administration for any one of TC, TG, and HbA1c (n = 2,387); pretreatment for hyperlipidemia unknown or pretreatment with diabetes drugs, or both (n = 783); unknown concomitant drug for treating hyperlipidemia or unknown diabetes drug (n = 6); or both (these cases overlapped, resulting in 2,662 cases being excluded from safety analyses).
Patients' backgrounds in the efficacy analysis group are shown in Table 1. The number of patients with diabetes was 3,312 (99.9%), of whom 3,235 (97.6%) were diagnosed with type 2 diabetes. Mean duration of diabetes was 6.6 ± 5.9 years. A total of 602 (18.2%) and 1,977 (59.6%) patients had been previously treated for dyslipidemia and diabetes, respectively. Medication characteristics of the efficacy analysis group showed that while most patients (3,149; 95.0%) were treated for dyslipidemia with bezafibrate alone, 167 patients (5.0%) took other lipid-lowering drugs concurrently (Table 2).
All patients were classified into four subgroups according to diabetes drug administration before and during the study as follows: diabetes drug not taken before or during the study (n = 818, 24.7%), taken only during the study (n = 521, 15.7%), taken only before the study (n = 106, 3.2%) and taken before and during the study (n = 1,871, 56.4%). Thus, a diabetes drug was taken by 2,392 patients (72.1%) during the study. The most frequently used diabetes drug regimen was single administration of sulfonylureas (22.1%) followed by combined use of sulfonylureas and an α-glucosidase inhibitor (6.9%) (Table 2). The number of patients receiving both monotherapy and combination use of diabetes drugs was as follows: sulfonylureas (n = 1,505, 45.4%), glinides (n = 341, 10.3%), α-glucosidase inhibitor (n = 685, 20.7%), biguanide (n = 550, 16.6%), thiazolidine (n = 287, 8.7%), and insulin (n = 284, 8.6%) (data not shown). All of 3,316 patients took bezafibrate at least 6 weeks, of whom 3,299 (99.5%) took the drug for at least 12 weeks and 17 (0.5%) for at least six weeks. The dose of bezafibrate was 400 mg/day in 2,728 (82.3%) patients and 200 mg/day in 472 (14.2%).
Effect of Bezafibrate on Blood Lipid and Glucose Levels
Table 3 presents the changes in lipid and glucose parameters at baseline and after bezafibrate administration. All lipid values except LDL-C improved significantly compared with baseline values. Significant decreases from baseline were observed for TC, TG, non-HDL-C, and TG/HDL-C ratio; in contrast, a significant increase was observed in HDL-C. Although LDL-C levels decreased significantly in patients with LDL-C ≥ 3.6 mmol/l at baseline (n = 737), they increased significantly in those with LDL-C < 3.6 mmol/l at baseline (n = 1,489). Thus, LDL-C levels increased significantly overall. Moreover, HbA1c, FBG, and HOMA-R levels decreased significantly.
Effect of Bezafibrate on Blood Glucose Levels - Subgroup Analysis
Table 4 summarizes the changes in HbA1c concentrations after treatment, classified according to use of diabetes drug and baseline HbA1c levels. We analyzed 2,086 patients whose baseline HbA1c values were ≥7.0%. HbA1c levels decreased significantly by -0.76% from baseline in all patients. This change was greater in patients whose baseline HbA1c levels were higher. The trend analysis showed significant differences between the HbA1c-subgroups. Similar results were observed in all medication subgroups classified by diabetes drug administration before and during the study.
There were 1,464 patients for which we could determine the change in HbA1c levels as a function of duration of diabetes (Table 5). HbA1c levels significantly decreased in all diabetes-duration subgroups, most strongly in the subgroup with diabetes for < 1 year. Trend analysis showed significant differences between the diabetes-duration subgroups. Similar results were also observed in the medication subgroup of patients who were not administrated diabetes drug before or during the study.
We also analyzed HbA1c levels according to concomitant use of a single diabetes drug during the study. Changes in HbA1c levels for each drug were as follows: sulfonylureas, n = 536, −0.84%, p < 0.001; glinides, n = 125, −0.82%, p < 0.001; alpha glucosidase inhibitor, n = 70, −0.66%, p < 0.001; biguanide, n = 239, −0.73%, p < 0.001; thiazolidine, n = 28, −0.64%, p = 0.004; insulin: n = 130, −0.59%, p < 0.001. Each regimen showed a significant decrease in HbA1c. There were no significant differences between any of the regimens (p = 0.363) (data not shown).
Factors Influencing HbA1c Levels
We conducted univariate and multivariate regression analyses to evaluate the influence of bezafibrate on HbA1c levels. For initial univariate analysis of 1,854 patients, the difference in HbA1c levels was as an objective variable. Explanatory variables were patient age, sex, baseline BMI, baseline HbA1c level, diabetes-drug administration during the study, duration of diabetes, and rate of change in levels of each lipid (TC, LDL-C, TG, and HDL-C). Based on these results, baseline BMI, baseline HbA1c, diabetes-drug administration, rate of change in TG, HDL-C, and TC were found to be significant variables.
We next performed multivariate regression analysis using these variables and found that the estimated influence rate per unit change in each variable on HbA1c was as follows: baseline HbA1c, -0.489; baseline BMI, 0.016; diabetes-drug administration, 0.176; rate of change in TG, 0.004; and rate of change in TC, 0.005. The rate of change in HDL-C was not a significant variable (Table 6). Further, the average rates of change in TG and TC for all subjects at 24 weeks were -45.4% and -5.4%, respectively (Table 3). When the rate of change in each lipid was considered, their estimated influences on altering HbA1c levels were -0.182% and -0.027%, respectively. Next, we analyzed the relationship between rate of change in TG and amount of change in HbA1c in 3,316 patients (Figure 1). Irrespective of diabetes drug administration during the study, a strong positive relationship was observed between the rates of change in TG and amount of change in HbA1c levels.
(Enlarge Image)
Figure 1.
TG and HbA1c levels in patients depending on use of a diabetes drug. The graphs show the changes in HbA1c levels determined for all patients (•), patients taking a diabetes drug (▵), and patients not taking a diabetes drug (□).
Safety
Drug safety was assessed for 5,978 patients. ADRs were experienced by 306 patients (5.1%) (Table 7). The most common ADRs were increased blood creatine phosphokinase (0.8%), blood creatinine (0.8%), blood urea (0.5%), renal impairment (0.3%), and aspartate aminotransferase (0.3%). Severe ADRs were experienced by 17 patients (0.28%) and mainly included stroke (n = 2, 0.03%), renal impairment (n = 2, 0.03%), elevated blood urea nitrogen (n = 2, 0.03%), abnormal hepatic function (n = 1, 0.02%), pneumonia (n = 1, 0.02%), rhabdomyolysis (n = 1, 0.02%), hypoglycemia (n = 1, 0.02%), and others. ADR rate increased slightly in patients concurrently taking diabetes drugs (4.3% not taking vs. 5.6% taking), but no apparent clinical differences were observed. One hypoglycemia case occurred in a patient concurrently taking a diabetes drug. Rhabdomyolysis was observed in five patients, four of whom were treated with a diabetes drug. No ADRs were observed that could be unequivocally ascribed to concurrent diabetes drug use.