Impact of increased visceral and cardiac fat on cardiometabolic risk and disease.

OBJECTIVE: Previous studies have highlighted the associations between abdominal, cardiac or total fat accumulation and cardiovascular disease. The aim of this study was to investigate the impact of different ectopic fat depots on measurements of metabolic dysfunction and cardiovascular disease risk. METHODS: Using magnetic resonance imaging in 113 subjects, we measured abdominal (visceral and subcutaneous) and cardiac (epicardial and extra-pericardial) fat depots and examined their association with overall (BMI) and abdominal obesity (waist circumference), dyslipidaemia (triglycerides, total and HDL cholesterol), glucose tolerance (by an oral glucose tolerance test) and insulin sensitivity, blood pressure and 10-year coronary heart disease risk by Framingham score. RESULTS: Fat accumulation was proportional to the degree of obesity, with body fat ranging from 14 to 33 kg, visceral fat from 0.8 to 1.8 kg and cardiac fat from 134 to 236 g. Most cardiac fat (70% on average) was extra-pericardial, with a wide variability for both cardiac depots (epicardial: 172-2008 mm(2); extra-pericardial: 100-5056 mm(2)). Only visceral and extra-pericardial fat, but not epicardial or subcutaneous fat, could discriminate between subjects with three or more factors of the metabolic syndrome or medium-to-high coronary heart disease risk score. Controlling for gender and BMI by multivariable analysis, the best marker of reduced insulin sensitivity was visceral fat (partial r = -0.35); extra-pericardial fat was the closest associate of increased blood pressure (partial r = 0.26) and both extra-pericardial and visceral fat clustered with hypertriglyceridaemia (partial r = 0.29 and 0.24; both P < 0.02). CONCLUSION: Increased epicardial fat per se does not necessarily translate into presence or prediction of disease. In contrast, increased deposition of visceral abdominal and extra-pericardial mediastinal fat are both associated with an enhanced cardiovascular disease risk profile.

Effect of pioglitazone on the metabolic and hormonal response to a mixed meal in type II diabetes.

We explored the mechanisms by which a 4-month, placebo-controlled pioglitazone treatment (45 mg/day) improves glycemic control in type II diabetic patients (T2D, n=27) using physiological testing (6-h mixed meal) and a triple tracer technique ([6,6-(2)H(2)]glucose infusion, (2)H(2)O and [6-(3)H]glucose ingestion) to measure endogenous glucose production (EGP), gluconeogenesis (GNG), insulin-mediated glucose clearance and beta-cell glucose sensitivity (by c-peptide modeling). Compared to sex/age/weight-matched non-diabetic controls, T2D patients showed inappropriately (for prevailing insulinemia) raised glucose production (1.05[0.53] vs 0.71[0.36]mmol min(-1) kg(ffm)(-1) pM, P=0.03) because of enhanced GNG (73.1+/-2.4 vs 59.5+/-3.6%, P<0.01) persisting throughout the meal, reduced insulin-mediated glucose clearance (6[5] vs 12[13]ml min(-1) kg(ffm)(-1) nM(-1), P<0.005), and impaired beta-cell glucose-sensitivity (27[38] vs 71[37]pmol min(-1) m(-2) mM(-1), P=0.002). Compared to placebo, pioglitazone improved glucose overproduction (P=0.0001), GNG and glucose underutilization (P=0.05) despite lower insulinemia. GNG improvement was quantitatively related to raised adiponectin. beta-cell glucose sensitivity was unchanged. In mild-to-moderate T2D, pioglitazone monotherapy decreased fasting and post-prandial glycemia, principally via inhibition of gluconeogenesis, improved hepatic and peripheral insulin resistance.

Visceral fat and beta cell function in non-diabetic humans.

AIMS/HYPOTHESIS: Preferential visceral adipose tissue (VAT) accumulation has been clearly associated with insulin resistance. In contrast, the impact of visceral obesity on beta cell function is controversial. METHODS: In 62 non-diabetic women and men (age 24-69 years, BMI 21-39 kg/m2), we measured VAT and subcutaneous adipose tissue (SAT) fat mass by magnetic resonance imaging. We also measured insulin secretion and beta cell function by C-peptide deconvolution and physiological modelling of data from a frequently sampled, 75-g, 3-h OGTT, respectively. RESULTS: VAT (range 0.1-3.1 kg) was strongly related to sex, age and BMI; SAT was related to sex and BMI. Controlling for sex, age, BMI and SAT by multivariate analysis, excess VAT was associated with a clinical phenotype comprising higher plasma glucose levels, BP, heart rate and serum transaminases. The corresponding metabolic phenotype consisted of insulin resistance (partial r=-0.38) and hyperinsulinaemia (partial r=0.29). The latter, however, was appropriate for the degree of insulin resistance regardless of obesity and abdominal fat distribution. Moreover, none of the model-derived parameters describing beta cell function (glucose sensitivity, rate sensitivity and potentiation) was independently associated with excess VAT. CONCLUSIONS/INTERPRETATION: In non-diabetic Caucasian adults of either sex, preferential visceral fat deposition in itself is part of an insulin-resistant phenotype. The insulin secretory response to a physiological challenge is increased to fully compensate for the insulin resistance, but the dynamics of beta cell function (glucose sensitivity, rate sensitivity and potentiation) are largely preserved.

Splanchnic and leg substrate exchange after ingestion of a natural mixed meal in humans.

The disposal of a mixed meal was examined in 11 male subjects by multiple (splanchnic and femoral) catheterization combined with double-isotope technique (intravenous [2-3H]glucose plus oral U-[14C]starch). Glucose kinetics and organ substrate balance were measured basally and for 5 h after eating pizza (600 kcal) containing carbohydrates 75 g as starch, proteins 37 g, and lipids 17 g. The portal appearance of ingested carbohydrate was maximal (1.0 mmol/min) between 30 and 60 min after the meal and gradually declined thereafter, but was still incomplete at 300 min (0.46+/-0.08 mmol/min). The total amount of glucose absorbed by the gut over the 5 h of the study was 247+/-26 mmol (45+/-6 g), corresponding to 60+/-6% of the ingested starch. Net splanchnic glucose balance (-6.7+/-0.5 micromol x kg(-1) x min(-1), basal) rose by 250-300% between 30 and 60 min and then returned to baseline. Hepatic glucose production (HGP) was suppressed slightly and only tardily in response to meal ingestion (approximately 30% between 120 and 300 min). Splanchnic glucose uptake (3.7+/-0.6 micromol x kg(-1) x min(-1), basal) peaked to 9.8+/-2.0 micromol x kg(-1) x min(-1) (P<0.001) at 120 min and then returned slowly to baseline. Leg glucose uptake (34+/-5 micromol x leg(-1) x min(-1), basal) rose to 151+/-29 micromol x leg(-1) x min(-1) at 30 min (P<0.001) and remained above baseline until the end of the study, despite no increase in leg blood flow. The total amount of glucose taken up by the splanchnic area and total muscle mass was 161+/-16 mmol (29+/-3 g) and 128 mmol (23 g), respectively, which represent 39 and 30% of the ingested starch. Arterial blood lactate increased by 30% after meal ingestion. Net splanchnic lactate balance switched from a basal net uptake (3.2+/-0.6 micromol kg(-1) x min(-1) to a net output between 60 and 120 min and tended to zero thereafter. Leg lactate release (25+/-11 micromol x leg(-1) x min(-1), basal) drastically decreased postprandially. Arterial concentration of both branched-chain amino acids (BCAA) and non-branched-chain amino acids (N-BCAA) increased significantly after meal ingestion (P<0.001). The splanchnic area switched from a basal net amino acid uptake (31+/-16 and 92+/-48 micromol/min for BCAA and N-BCAA, respectively) to a net amino acid release postprandially. The net splanchnic amino acid release over 5 h was 11.3+/-4.2 mmol for BCAA and 37.8+/-9.7 mmol for N-BCAA. Basally, the net leg balance of BCAA was neutral (-3+/-5 micromol x leg(-1) x min(-1)), whereas that of N-BCAA indicated a net release (54+/-14 micromol x leg(-1) x min(-1)). After meal ingestion, there was a net leg uptake of BCAA (20+/-6 micromol x leg(-1) x min(-1)), whereas leg release of N-BCAA decreased by 50%. It is concluded that in human subjects, 1) the absorption of a natural mixed meal is still incomplete at 5 h after ingestion; 2) HGP is only marginally and tardily inhibited; 3) splanchnic and peripheral tissues contribute to the disposal of meal carbohydrate to approximately the same extent; 4) the splanchnic area transfers >30% of the ingested proteins to the systemic circulation; and 5) after meal ingestion, skeletal muscle takes up BCAA to replenish muscle protein stores.

Effects of acute alpha 2-blockade on insulin action and secretion in humans.

We tested whether acute alpha 2-blockade affects insulin secretion, glucose and fat metabolism, thermogenesis, and hemodynamics in humans. During a 5-h epinephrine infusion (50 ng.min-1.kg-1) in five volunteers, deriglidole, a selective alpha 2-receptor inhibitor, led to a more sustained rise in plasma insulin and C-peptide levels (+59 +/- 14 vs. +28 +/- 6, and +273 +/- 18 vs. +53 +/- 14 pM, P < 0.01 vs. placebo) despite a smaller rise in plasma glucose (+0.90 +/- 0.4 vs. +1.5 +/- 0.3 mM, P < 0.01). Another 10 subjects were studied in the postabsorptive state and during a 4-h hyperglycemic (+4 mM) clamp, coupled with the ingestion of 75 g of glucose at 2 h. In the postabsorptive state, hepatic glucose production, resting energy expenditure, and plasma insulin, free fatty acid (FFA), and potassium concentrations were not affected by acute alpha 2-blockade. Hyperglycemia elicited a biphasic rise in plasma insulin (to a peak of 140 +/- 24 pM), C-peptide levels (1,520 +/- 344 pM), and insulin secretion (to 410 +/- 22 pmol/min); superimposed glucose ingestion elicited a further twofold rise in insulin and C-peptide levels, and insulin secretion. However, alpha 2-blockade failed to change these secretory responses. Fasting blood beta-hydroxybutyrate and glycerol and plasma FFA and potassium concentrations all declined with hyperglycemia; time course and extent of these changes were not affected by alpha 2-blockade. Resting energy expenditure (+25 vs. +16%, P < 0.01) and external cardiac work (+28% vs. +19%, P < 0.01) showed larger increments after alpha 2-blockade. We conclude that acute alpha 2-blockade in humans 1) prevents epinephrine-induced inhibition of insulin secretion, 2) does not potentiate basal or intravenous- or oral glucose-induced insulin release, 3) enhances thermogenesis, and 4) increases cardiac work.

Metabolic and cardiovascular assessment in moderate obesity: effect of weight loss.

Metabolic and hemodynamic abnormalities have been separately described in obesity, and weight reduction is known to lead to some improvement in each. Our aim was to simultaneously assess metabolic and cardiovascular function in normotensive, normotolerant patients with moderate obesity (body mass index = 32.6 +/- 1.1 kg/m2) before and after weight loss. The obese were insulin resistant [37.4 +/- 4.8 mumol/min.kg FFM; P < 0.02 vs. 12 lean controls (50.6 +/- 2.6), on a euglycemic insulin clamp], secreted more insulin both in the fasting state and after oral glucose (70 +/- 10 vs. 48 +/- 6 nmol/mmol.L plasma glucose; P < 0.05), and had higher resting energy expenditure (4.62 +/- 0.18 vs. 4.00 +/- 0.23 kJ/min), systolic and mean blood pressure, stroke volume (87 +/- 8 vs. 67 +/- 4 mL/min; P = 0.05), and cardiac output. There was, however, no relationship between the metabolic and hemodynamic abnormalities. After a weight loss of 11 +/- 1 kg (approximately 15%), insulin sensitivity improved in proportion to the weight reduction, whereas insulin hypersecretion and high energy expenditure persisted. In contrast, all hemodynamic changes reverted to normal. We conclude that in moderate obesity, the metabolic and cardiovascular abnormalities are largely independent of one another; accordingly, weight loss affects them differentially. Partial weight normalization may provide sufficient cardiovascular protection.

Effect of insulin on renal sodium and uric acid handling in essential hypertension.

In normal subjects, insulin decreases the urinary excretion of sodium, potassium, and uric acid. We tested whether these renal effects of insulin are altered in insulin resistant hypertension. In 37 patients with essential hypertension, we measured the changes in urinary excretion of sodium, potassium, and uric acid in response to physiological euglycemic hyperinsulinemia (by using the insulin clamp technique at an insulin infusion rate of 6 pmol/min/kg). Glucose disposal rate averaged 26.6 +/- 1.5 mumol/min/kg, i.e., 20% lower than in normotensive controls (33.1 +/- 2.1 mumol/min/kg, P = .015) In the basal state, fasting plasma uric acid concentrations were higher in men than women (P < .001), were positively related to body mass index (r = 0.38, P = .02), waist/hip ratio (r = 0.35, P < .05), and serum triglyceride levels (r = 0.59, P = .0001), and negatively related to HDL cholesterol concentrations (r = -0.59, P = .0001) and glucose disposal rate (r = 0.42, P < .01). Uric acid clearance, on the other hand, was inversely related to body mass index (r = 0.41, P = .01), plasma uric acid (r = 0.65, P < .0001) and triglyceride concentrations (r = 0.39, P < .02), and directly related to HDL cholesterol levels (r = 0.52, P < .001). During insulin infusion, blood pressure, plasma uric acid and sodium concentration, and creatinine clearance did not change. In contrast, hyperinsulinemia caused a significant decrease in the urinary excretion of uric acid (2.67 +/- 0.12 to 1.86 +/- .14 mumol/min/1.73 m2, P = .0001), sodium (184 +/- 12 to 137 +/- 14 mumol/min/1.73 m2, P = .0001), and potassium (81 +/- 7 to 48 +/- 4 mumol/ min/1.73 m2, P = .0001). Both in absolute terms (clearance and fractional excretion rates) and percentagewise, these changes were similar to those found in normotensive subjects. Insulin-induced changes in urate excretion were coupled (r = 0.55, P < .0001) to the respective changes in sodium excretion. In hypertensive patients, higher uric acid levels and lower renal urate clearance rates cluster with insulin resistance and dyslipidemia. Despite insulin resistance of glucose metabolism, acute physiological hyperinsulinemia causes normal antinatriuresis, antikaliuresis, and antiuricosuria in these patients.

Insulin sensitivity in cardiological syndrome X.

OBJECTIVES: To test whether cardiological syndrome X is an insulin-resistant state. SETTING, DESIGN AND SUBJECTS: The coronary care unit of a referral centre for angina pectoris in Pisa, Italy. A case-control study, involving 10 patients with unequivocal (angiographycally proven) cardiological syndrome X, but normal glucose tolerance, blood pressure and lipid levels, and 13 matched healthy subjects. MAIN OUTCOME MEASURES: Insulin sensitivity and pattern of substrate oxidation (assessed by the euglycaemic insulin clamp technique in combination with indirect calorimetry). RESULTS: Fasting plasma glucose and insulin levels were 5.05 +/- 0.11 versus 4.88 +/- 0.11 mmol l-1 and 68 +/- 10 versus 56 +/- 6 pmol l-1, respectively (controls versus patients, ns). During the insulin clamp, glucose disposal rate was nearly identical in patients and controls (25.9 +/- 1.8 and 27.2 +/- 1.8 mumol kg-1 min-1, respectively. P = 0.88). Non-oxidative glucose disposal accounted for similar proportions of total glucose uptake (59 versus 53%, patients versus controls, ns). Resting energy expenditure (13.7 +/- 0.6 versus 13.8 +/- 0.8 cal kg-1 min-1, ns) and insulin-induced thermogenesis were similar in the two groups. Fasting plasma NEFA concentrations (0.64 +/- 0.09 and 0.64 +/- 0.06 mmol l-1, patients and controls, ns) fell in a similar time-course and to virtually identical nadirs (0.13 +/- 0.02 and 0.14 +/- 0.02 mmol l-1) after insulin infusion. Fasting plasma potassium was similar in patients and controls (3.99 +/- 0.10 and 4.16 +/- 0.04 mmol l-1, ns), and insulin induced equivalent hypokalaemia (-14 versus -19%). CONCLUSIONS: None of the in vivo actions of insulin were impaired in patients with 'pure' syndrome X when compared to matched controls. Therefore, we conclude that cardiological syndrome X is not an insulin resistant state per se, and that any decrease in insulin sensitivity found in this condition is likely to be secondary.

Insulin resistance in microalbuminuric hypertension. Sites and mechanisms.

Microalbuminuria in patients with essential hypertension is a marker of incipient glomerular dysfunction and clusters with lipid and hemodynamic abnormalities. Recent evidence has shown that hypertensive patients with microalbuminuria have a hyperinsulinemic response to oral glucose, suggesting the presence of insulin resistance. To directly test this possibility we studied insulin action in two accurately matched groups (n = 10 each) of hypertensive patients with or without microalbuminuria (14 +/- 2 versus 52 +/- 7 mg/24 h-1, mean of three 24-hour collections). In response to glucose ingestion microalbuminuric patients showed slight hyperglycemia (area under the curve, 928 +/- 43 versus 784 +/-19 nmol/L-1/2h-1, P < .02) and a marked hyperinsulinemia (26.8 +/- 3.3 versus 49.8 +/- 3.7 nmol/L-1/2h-1, P < 0.01). Basal arterial blood pressure, heart rate, and forearm blood flow were similar in the two groups and did not change significantly during a 2-hour euglycemic insulin clamp. Insulin-stimulated wholebody glucose uptake was 25% lower in microalbuminuric patients (33.5 +/- 2.5 versus 25.2 +/- 2.1 mumol/min-1/kg-1, P < .02). This difference was entirely due to a 40% reduction in glycogen synthesis (12.9 +/- 1.8 versus 21.3 +/- 3.2 mumol/min-1/kg-1, P < .05) as glucose oxidation was similarly stimulated in the two groups. In contrast there was no difference in the ability of insulin to suppress hepatic glucose production (by approximately 100% at the end of the clamp), to decrease fractional sodium and potassium excretions (by 35%), to lower circulating free fatty acids (by 80%), and to reduce plasma potassium concentrations (by 10%).(ABSTRACT TRUNCATED AT 250 WORDS)

Effect of insulin on uric acid excretion in humans.

Although hyperuricemia is a frequent finding in insulin-resistant states, insulin's effect on renal uric acid (UA) handling is not known. In 20 healthy volunteers, diastolic blood pressure, body weight, and fasting plasma insulin were positively (and age was negatively) related to fasting plasma UA concentrations, together accounting for 53% of their variability. During an insulin clamp, urine flow was lower than during fasting conditions (1.01 +/- 0.12 vs. 1.56 +/- 0.32 ml/min, P = 0.04), whereas creatinine clearance was unchanged (129 +/- 7 and 131 +/- 9 ml/min, P = not significant). Hyperinsulinemia did not alter serum UA concentrations (303 +/- 13 vs. 304 +/- 12 microM) but caused a significant decrease in urinary UA excretion [whether expressed as absolute excretion rate (1.66 +/- 0.21 vs. 2.12 +/- 0.23 mumol/min, P = 0.03), clearance rate (5.6 +/- 0.8 vs. 7.3 +/- 0.8 ml/min, P = 0.03), or fractional excretion (4.48 +/- 0.80 ml/min vs. 6.06 +/- 0.64%, P < 0.03)]. Hyperinsulinemia was also associated with a 30% (P < 0.001) fall in urine Na excretion. Fractional UA excretion was related to Na fractional excretion under basal conditions (r = 0.59, P < 0.01) and during the insulin period (r = 0.53, P < 0.02). Furthermore, the insulin-induced changes in fractional UA and Na excretion correlated with one another (r = 0.66, P < 0.001). Physiological hyperinsulinemia acutely reduces urinary UA and Na excretion in a coupled fashion.

Effects of acute hypercarnitinemia during increased fatty substrate oxidation in man.

To test whether carnitine availability is rate-limiting for fat oxidation under conditions of augmented oxidative use of fatty substrates, two series of studies were performed. In study no. 1, L-carnitine (1 g + 0.5 g/h intravenously [i.v.]) or saline was given to eight volunteers during a 4-hour infusion of a 10% triglyceride emulsion, thereby increasing plasma free-carnitine levels from 38 +/- 4 to 415 +/- 55 mumol/L. Fat infusion increased plasma triglyceride levels (80%) and lipid oxidation (30%), and decreased (28%) carbohydrate oxidation (as measured by indirect calorimetry); hypercarnitinemia had no influence on these responses. In study no. 2 in 12 healthy subjects a bolus of L-carnitine (3 g) or saline was administered 40 minutes before aerobic exercise (bicycling for 40 minutes at 60 W), followed by 2 minutes of anaerobic exercise (250 W) and 50 minutes of recovery. Oxygen consumption (VO2), increased to 18.3 +/- 0.7 mL.min-1 x kg-1 during aerobic exercise, reached a maximum of 46.0 +/- 0.8 mL.min-1 x kg-1 during the anaerobic bout, and returned to baseline within a few minutes, with no difference between control and carnitine. At virtually identical mean energy expenditure rates (196 +/- 7 v 197 +/- 7 J.min-1 x kg-1, saline v carnitine), after carnitine administration the entire exercise protocol was sustained by a lower mean carbohydrate oxidation rate (42.1 +/- 3.6 v 36.5 +/- 2.3 mumol.min-1 x kg-1, P < .03) and a higher mean lipid oxidation rate (6.7 +/- 1.0 v 8.3 +/- 0.7 mumol.min-1 x kg-1, P < .05).(ABSTRACT TRUNCATED AT 250 WORDS)

Characterization of a reversed-phase high-performance liquid chromatographic system for the determination of blood amino acids.

High-performance liquid chromatography was used to separate physiological amino acids in perchloric acid supernatants of blood samples. Precolumn derivatization with phenyl isothiocyanate was carried out, starting with 20 microliters of supernatant; 2-10 microliters were injected into a 30-cm Pico Tag column, which was eluted with a gradient of two eluents in 64 min. Stock amino acid solutions prepared in water, hydrochloric acid or perchloric acid showed comparable recoveries on serial dilution (parallelism test). The recovery of crystalline amino acids added to blood in amounts ranging from normal to six times normal was generally satisfactory. The within-assay relative standard deviations were less than 5% for many amino acids. The performance of the system was less than satisfactory for cysteine and methionine. Glutamine and asparagine are interconverted into glutamate and aspartate, respectively, in a time-dependent fashion; a separate measurement of one member of the pair is therefore required in order to assay the other starting from the sum of both chromatographic peaks. The method is suitable for the relatively rapid, sensitive and accurate measurement of blood amino acids in perchloric acid supernatants (in which other relevant metabolites are customarily assayed) over a wide range of physiological concentrations, on very small amounts of sample.

Acute elevation of free fatty acid levels leads to hepatic insulin resistance in obese subjects.

Raised levels of free fatty acids (FFA) compete with glucose for utilization by insulin-sensitive tissues, and, therefore, they may induce insulin resistance in the normal subject. The influence of experimental elevations in FFA levels on glucose metabolism in native insulin-resistant states is not known. We studied seven women with moderate obesity (63% above their ideal body weight) but normal glucose tolerance with the use of the insulin clamp technique with or without an infusion of Intralipid + heparin. Upon raising plasma insulin levels to approximately 60 microU/mL while maintaining euglycemia, whole body glucose utilization (3H-3-glucose) rose similarly without (from 66 +/- 7 to 113 +/- 11 mg/min m2, P less than .02) or with (from 70 +/- 7 to 137 +/- 19 mg/min m2, P less than .02) concomitant lipid infusion. In contrast, endogenous glucose production was considerably (73%) suppressed (from 66 +/- 7 to 15 +/- 8 mg/min m2, P less than .001) during the clamp without lipid, but declined only marginally (from 70 +/- 7 to 48 +/- 7 mg/min m2, NS) with lipid administration. The difference between the control and the lipid study was highly significant (P less than .02), and amounted to an average of 3.8 g of relative glucose overproduction during the second hour of the clamp. Blood levels of lactate rose by 34 +/- 15% (.1 greater than P greater than .05) in the control study but only by 17 +/- 10% (NS) during lipid infusion. Blood pyruvate concentrations fell in both sets of experiments (by approximately 45% at the end of the study) with similar time courses.(ABSTRACT TRUNCATED AT 250 WORDS)

Plasma biguanide levels are correlated with metabolic effects in diabetic patients.

Metabolic abnormalities occur in biguanide-treated diabetic patients. We investigated the relationship between plasma metformin and phenformin concentrations and metabolic effects. Drug levels were measured in 37 type II diabetic patients by HPLC. The method was sensitive, specific, and linear over a wide range of drug concentrations. Metformin and phenformin values ranged from 236 to 718 ng/ml and from 28 to 114 ng/ml, respectively. The plasma metformin level was correlated with triglycerides (r = -0.55; P less than 0.05) but not with drug dosage, plasma glucose, HbA1, creatinine, creatinine clearance, lactate, pyruvate, lipid, and clinical parameters. Plasma phenformin concentrations correlated with lactate (r = 0.49; P less than 0.05) and HbA1 (r = 0.50; P less than 0.05) but not with drug dosage, parameters of diabetes control, creatinine, creatinine clearance, pyruvate, and clinical parameters. The clinical usefulness of this HPLC method, the evidence that the increase of lactate is related to the circulating phenformin levels, and the demonstration that the metformin effect on triglyceride metabolism is correlated to plasma drug levels are the positive findings of this work.