In people without diabetes, approximately 50% of daily insulin secretion is basal and the remainder is postprandial. Hence, it would be expected that insulin replacement therapy in a 50/50 ratio with each meal would mimic physiologic insulin secretion better than treatment with once-daily basal insulin in patients with diabetes mellitus. Using lispro mix (LM) 50/50 before meals may be a logical approach to achieving glycemic targets (glycosylated hemoglobin [HbA(lc)] and pre- and postprandial blood glucose [BG] concentrations) in these patients. The aim of this study was to test the hypothesis that treatment with a premixed insulin analogue containing 50/50 basal + prandial insulins administered before each meal would achieve lower overall and mealtime glycemic control than once-daily basal insulin analogue, both plus metformin (Met), in patients with type 2 diabetes mellitus. This 24-week, randomized, open-label, parallel-group trial was conducted at 38 sites across Australia, Greece, India, The Netherlands, Poland, Puerto Rico, and the United States. Male and female patients aged 35 to 75 years with type 2 diabetes mellitus and an HbA(1c) level of 6.5% to 11.0%, who were receiving metformin and/or a sulfonylurea with a stable dose of 0 to 2 daily insulin injections over the previous 3 months were eligible. Patients were randomly assigned to receive LM50/50 (50% insulin lispro protamine suspension [ILPS] and 50% lispro) TID plus metformin (to a maximally tolerated daily dosage of 500-1000 mg BID) (LM50/50 + Met) or insulin glargine QD at bedtime plus metformin (500-1000 mg BID) (G + Met) for 24 weeks. With LM50/50 + Met, the insulin dose was titrated to target a fasting BG (FBG) level of <6.7 mmol/L (<120 mg/dL) and a 2-hour post-prandial BG (PPBG) level of <8.0 mmol/L (<144 mg/dL); those who did not reach the FBG target would be switched from presupper LM50/50 to LM75/25 (75% ILPS, 25% lispro). A total of 315 patients were randomized and received treatment (158 women, 157 men; mean age, 57.7 years; mean body mass index, 32.1 kg/m2; LM50/50 + Met, 157 patients; G + Met, 158 patients). At 24 weeks, the mean (SD)HbA(1c) level was significantly lower in the LM50/50 + Met group than in the G + Met group (7.1% [0.9%] vs 7.5% [1.0%]; P<0.001), and the proportion who reached an HbA(1c) target of < or = 7.0% was greater (88 [56.1%] vs 63 [39.9%]; P = 0.005). The G + Met group had a lower mean (SD)FBG value (6.5 [1.6] vs 8.1 [1.8] mmol/L; P<0.001). The LM50/50 + Met group had lower mean preprandial BG levels prelunch (7.4 [1.9] vs 7.9 [2.1] mmol/L; P=0.03) and presupper (8.3 [2.0] vs 8.9 [2.8] mmol/L; P=0.04). The LM50/50 + Met group also had lower mean 2-hour PPBG values postbreakfast (8.7 [2.2] vs 9.2 [2.5] mmol/L; P=0.03), postlunch (8.4 [1.9] vs 9.8 [2.6], mmol/L; p<0.001), and postsupper (8.7 [2.2] vs 10.7 [3.2], mmol/L; P<0.001). The mean (SD) total insulin doses at study end point were 0.7 (0.3) U/kg in the LM50/50 + Met group and 0.6 (0.3) U/kg in the G + Met group (P<0.001). The mean (SD)M-value (an expression of mean glycemia and the effect of glucose swings) was statistically similar between the 2 groups at baseline but significantly lower in the LM50/50 + Met group at end point (17.3 [13.8] vs 25.1 [24.8] mmol/L; P<0.001). During the entire treatment period, mean (SD) overall and nocturnal hypoglycemia rates (episodes per patient for 30 days) were statistically similar between the 2 groups (overall, 0.8 [1.4] vs 0.5 [1.0]; nocturnal, 0.2 [0.7] vs 0.3 [0.6]). At end point, the mean (SD) nocturnal hypoglycemia rates were similar between the 2 groups (0.2 [0.9] vs 0.2 [0.6]), but the overall and non-nocturnal hypoglycemia rates were higher with LM50/50 + Met (overall, 0.7 [1.7] vs 0.3 [0.8]; P=0.02; non-nocturnal, 0.5 [1.2] vs 0.1 [0.4]; P=0.002). In these patients with type 2 diabetes, mealtime LM50/50 + Met was associated with lower overall (HbA(1c)) and preprandial BG and PPBG levels (except for FBG), with similar nocturnal hypoglycemia and less glycemic variability, compared with G + Met. Copyright (c) 2007 Excerpta Medica, Inc.