Nineteen RCTs (n=4,396) were included in the review.
Rosiglitazone (11 RCTs, n=2,701).
Most of the trials were of a poor quality: two scored 4 out of a possible 5, while the rest scored 1 or 2.
Glycaemic control.
Rosiglitazone monotherapy was associated with a statistically significant decrease in FPG (WMD -0.62, 95% CI: -1.07, -0.17, P=0.007), based on 507 patients in 2 RCTs (one versus glyburide and one versus repaglinide). There was no evidence of statistical heterogeneity (P=0.93). No significant difference in HbA1c was found between rosiglitazone monotherapy and control (WMD -0.08, 95% CI: -0.65, 0.49, P=0.8), based on 507 patients in the same 2 RCTs. There was evidence of statistical heterogeneity (P=0.01).
Add-on therapy with rosiglitazone was associated with a statistically significant decrease in FPG (WMD -2.82, 95% CI: -3.15, -2.48, P<0.00001), based on 1,890 patients in 7 RCTs. There was evidence of statistical heterogeneity (P=0.09). Add-on therapy with rosiglitazone was also associated with a statistically significant reduction in HbA1c (WMD -1.29, 95% CI: -1.37, -1.22, P<0.00001), based on 1,907 patients in 8 RCTs. There was no evidence of statistical heterogeneity (P=0.74).
Secondary outcomes.
Compared with other antidiabetic therapy, rosiglitazone was associated with a significant increase in total cholesterol, LDL-cholesterol and HDL-cholesterol levels. No significant difference was found in the levels of triglycerides. Most studies reported normal levels of liver enzymes and no serious adverse effects on liver function were reported. Rosiglitazone was associated with a small decrease in haemoglobin and haemocrit. Hypoglycaemic events were mild to moderate in severity, and were more frequent in patients given add-on therapy than rosiglitazone monotherapy. Hypoglycaemia was most common in patients given rosiglitazone combined with insulin. Rosiglitazone was associated with a weight gain of 0.7 to 5.3 kg, depending on dose and combination regimen used, and a mild hypotensive effect was reported. Oedema occurred in 2.5 to 3.5% of patients given rosiglitazone monotherapy, 10.8% of patients given rosiglitazone combined with gliclazide, and 13.1 to 16.2% of patients given rosiglitazone combined with insulin.
Pioglitazone (8 RCTs, n=1,695). Most of the trials were of a poor quality: one scored 3 out of 5, seven scored 2, and one scored 1.
Glycaemic control.
Pioglitazone was associated with a statistically significant increase in HbA1c (WMD 0.46, 95% CI: 0.03, 0.9, P=0.04) compared with glyburide or repaglinide, based on 141 patients in 2 RCTs. There was no evidence of statistical heterogeneity (P=0.59). No significant difference was found in FPG levels between pioglitazone monotherapy and repaglinide (WMD 0.89, 95% CI: -0.26, 2.04, P=0.13), based on 123 patients in 1 RCT.
Add-on therapy with pioglitazone was associated with a statistically significant reduction in HbA1c compared with continuing non-thiazolidinedione monotherapy (WMD -1.29, 95% CI: -1.6, -0.99, P<0.00001), based on 1,422 patients in 6 RCTs. There was evidence of statistical heterogeneity (P=0.0002). Add-on therapy with pioglitazone was also associated with a statistically significant decrease in FPG compared with monotherapy with other diabetic agents (WMD -2.87, 95% CI: -3.59, -2.15, P<0.00001), based on 1,052 patients in 5 RCTs. There was evidence of statistical heterogeneity (P=0.0006).
Secondary outcomes.
Compared with other antidiabetic therapy, pioglitazone was associated with a statistically significant greater increase in HDL-cholesterol levels and a decrease in triglyceride levels. No significant difference was found for levels of total cholesterol or LDL-cholesterol. Most studies reported normal levels of liver enzymes and no serious adverse effects on liver function were reported. Pioglitazone was associated with a small decrease in haemoglobin and haemocrit. Pioglitazone monotherapy was not frequently associated with hypoglycaemia; occurrence increased when used in combination with other antidiabetics, most notably when combined with insulin. Pioglitazone was associated with a weight gain ranging from 0.95 to 3.6 kg, depending on the dose and combination regimen used, and a small decrease in systolic blood-pressure was reported. Oedema was frequently reported and was highest in patients given pioglitazone combined with insulin.