Metformin acts primarily by decreasing hepatic glucose output, largely by inhibiting gluconeogenesis. It also seems to induce weight loss, preferentially involving adipose tissue.

Excessive hepatic glucose output is an important factor in the fasting hyperglycemia of non-insulin-dependent diabetes mellitus (NIDDM). To determine the relative contributions of gluconeogenesis and glycogenolysis in a quantitative manner, we applied a new isotopic approach, using infusions of [6-3H]glucose and [2-14C]acetate to trace overall hepatic glucose output and phosphoenolpyruvate gluconeogenesis in 14 postabsorptive NIDDM subjects and in 9 nondiabetic volunteers of similar age and weight. Overall hepatic glucose output was increased nearly twofold in the NIDDM subjects (22.7 +/- 1.0 vs. 12.0 +/- 0.6 mumol.kg-1.min-1 in the nondiabetic volunteers, P less than .001); phosphoenolpyruvate gluconeogenesis was increased more than threefold in the NIDDM subjects (12.7 +/- 1.4 vs. 3.6 +/- 0.4 mumol.kg-1.min-1 in the nondiabetic subjects, P less than .001) and was accompanied by increased plasma lactate, alanine, and glucagon concentrations (all P less than .05). The increased phosphoenolpyruvate gluconeogenesis accounted for 89 +/- 6% of the increase in overall hepatic glucose output in the NIDDM subjects and was significantly correlated with the fasting plasma glucose concentrations (r = .67, P less than .01). Glycogenolysis, calculated as the difference between overall hepatic glucose output and phosphoenolpyruvate gluconeogenesis, was not significantly different in the NIDDM subjects (9.9 +/- 0.06 mumol.kg-1.min-1) and the nondiabetic volunteers (8.4 +/- 0.3 mumol.kg-1.min-1). We conclude that increased gluconeogenesis is the predominant mechanism responsible for increased hepatic glucose output in NIDDM.

The present studies were undertaken to determine whether Iipolysis was increased in non-insulin-dependent diabetes mellitus (NIDDM) and, if so, to assess the influence of increased glycerol availability on its conversion to glucose and its contribution to the increased gluconeogenesis found in this condition. For this purpose, we infused nine subjects with NIDDM and 16 age-, weight-matched nondiabetic volunteers with 12-3H1 glucose and IU-'4C1 glycerol and measured their rates of glucose and glycerol appearance in plasma and their rates of glycerol incorporation into plasma glucose. The rate of glycerol appearance, an index of lipolysis, was increased 1.5-fold in NIDDM subjects (2.85±0.16 vs. 1.62±0.08 gmol/kg per min, P < 0.001). Glycerol incorporation into plasma glucose was increased threefold in NIDDM subjects (1.13±1.10 vs.036±0.02 ,mol/kg per min, P < 0.01) and accounted for twice as much of hepatic glucose output (6.0±0.5 vs. 3.0±0.2%, P < 0.001). Moreover, the percent of glycerol turnover used for gluconeogenesis (77±6 vs. 44±2, P < 0.001) was increased in NIDDM subjects and, for a given plasma glycerol concentration, glycerol gluconeogenesis was increased more than twofold. The only experimental variable significantly correlated with the increased glycerol gluconeogenesis after taking glycerol availability into consideration was the plasma free fatty acid concentration (r = 0.80, P < 0.01). We, therefore, conclude that lipolysis is increased in NIDDM and, although more glycerol is thus available, increased activity of the intrahepatic pathway for conversion of glycerol into glucose, due at least in part to increased plasma free fatty acids, is the predominant mechanism responsible for enhanced glycerol gluconeogenesis. Finally, although gluconeogenesis from glycerol in NIDDM is comparable to that of alanine and about one-fourth that of lactate in terms of overall flux into glucose, glycerol is probably the most important gluconeogenic precursor in NIDDM in terms of adding new carbons to the glucose pool. (J. Clin. Invest. 1992. 89:169-175.)

To compare glutamine and alanine as gluconeogenic precursors, we simultaneously measured their systemic turnovers, clearances, and incorporation into plasma glucose, their skeletal muscle uptake and release, and the proportion of their appearance in plasma directly due to their release from protein in postabsorptive normal volunteers. We infused the volunteers with [U-14C] glutamine, [3-`3C] alanine, [2H5] phenylalanine, and [6-3H] glucose to isotopic steady state and used the forearm balance technique. We found that glutamine appearance in plasma exceeded that of alanine (5.76+0.26 vs. 4.40+0.33 ,Amol kg-l mind, P < 0.001), while alanine clearance exceeded glutamine clearance (14.7±1.3 vs. 9.3+0.8 ml kg-1 min', P < 0.001). Glutamine appearance in plasma directly due to its release from protein was more than double that of alanine (2.45±0.25 vs.1.16+0.12 jAmol kg-'lmin', P < 0.001). Although overall carbon transfer to glucose from glutamine and alanine was comparable (3.53+0.24 vs 3.47+0.32 atoms kg-'mind'), nearly twice as much glucose carbon came from protein derived glutamine than alanine (1.48±0.15 vs 0.88+0.09 atoms-kg-l mind1, P < 0.01). Finally, forearm muscle released more glutamine than alanine (0.88±0.05 vs 0.48±0.05Amol 100 ml`-min -1, P < 0.01). We conclude that in postabsorptive humans glutamine is quantitatively more important than alanine for transporting protein-derived carbon through plasma and adding these carbons to the glucose pool. (J. Clin. Invest. 1995. 95:272-277.)

To quantitate alanine and lactate gluconeogenesis in postabsorptive humans and to test the hypothesis that muscle is the principal source of these precursors, we infused normal volunteers with [3-14C]lactate, [3-13C]alanine, and [6-3H]glucose and calculated alanine and lactate incorporation into plasma glucose corrected for tricarboxylic acid cycle carbon exchange, the systemic appearance of these substrates, and their forearm fractional extraction, uptake, and release. Forearm alanine and lactate fractional extraction averaged 37 +/- 3 and 27 +/- 2%, respectively; muscle alanine release (2.94 +/- 0.27 mumol.kg body wt-1.min-1) accounted for approximately 70% of its systemic appearance (4.18 +/- 0.31 mumol.kg body wt-1.min-1); muscle lactate release (5.51 +/- 0.42 mumol.kg body wt-1.min-1) accounted for approximately 40% of its systemic appearance (12.66 +/- 0.77 mumol.kg body wt-1.min-1); muscle alanine and lactate uptake (1.60 +/- 0.7 and 3.29 +/- 0.36 mumol.kg body wt-1.min-1, respectively) accounted for approximately 30% of their overall disappearance from plasma, whereas alanine and lactate incorporation into plasma glucose (1.83 +/- 0.20 and 4.24 +/- 0.44 mumol.kg body wt-1.min-1, respectively) accounted for approximately 50% of their disappearance from plasma. We therefore conclude that muscle is the major source of plasma alanine and lactate in postabsorptive humans and that factors regulating their release from muscle may thus exert an important influence on hepatic gluconeogenesis.