Measurement of metformin concentration in erythrocytes: clinical implications.

Although pharmacokinetics studies have long suggested a deep compartment for the antidiabetic drug metformin, there is still little information concerning metformin accumulation by individual tissues in man. In the present study, the erythrocyte was chosen to represent this putative deep compartment and metformin concentration in erythrocytes (EM) was compared with that in plasma (PM) to delineate clinical implications. A reference group of 58 patients with well-tolerated metformin treatment was studied to provide standard mean metformin concentrations in the fasting state. Secondly, to provide transverse data reflecting clinical practice, the authors reviewed an investigation group of 93 metformin-treated patients with available PM and EM, which had been requested either to adjust metformin dosage to renal function, or to screen for potential metformin accumulation following renal failure, metformin overdose or lactic acidosis. Thirdly, the case of an individual with major metformin accumulation was studied to provide information about metformin elimination. From the bulk of data, we performed three types of analyses: (1) PM and EM were compared. In the investigation group, this comparison was extended to subgroups separated according to low-to-normal, moderately increased or highly increased metformin concentration. (2) Correlative analyses of PM, EM and serum creatinine were performed. (3) A kinetic study of the spontaneous decline of PM and EM was conducted. PM and EM were, respectively, 0.5 +/- 0.4 mg/l and 0.8 +/- 0.4 mg/l in the reference group, and 11.7 +/- 17.8 mg/l (mean +/- SD, range 0.0-71.9 mg/l) and 7.5 +/- 9.4 mg/l (0.0-34 mg/l) in the investigation group, mean serum creatinine of which was 290 +/- 258 micro mol/l. In the low-to-normal PM subgroup (n = 28), PM and EM were, respectively, 0.39 +/- 0.38 mg/l and 0.84 +/- 0.68 mg/l (p < 0.001). In the moderately increased PM subgroup (from therapeutic concentrations +2 SD to 5 mg/l, n = 24), PM and EM were 2.82 +/- 1.13 mg/l and 2.72 +/- 2.03 mg/l (NS). In the sharply increased PM subgroup (> 5 mg/l, n = 41), PM and EM were 27.6 +/- 23.2 mg/l and 17.0 +/- 11.4 mg/l (p = < 0.001). PM and EM were tightly correlated (r = 0.72 in the reference group and r = 0.90 in the investigation group, p < 0.001 for both). Metformin concentrations were also correlated with those of serum creatinine, but more so in the investigation group; in subgroups, a positive correlation was found only at high metformin concentrations and in erythrocytes. The kinetic study performed in the patient with major metformin accumulation showed that PM and EM dropped within less than 3 days from a maximum concentration of 80.0 mg/l and 20.4 mg/l, respectively, to 0.67 mg/l and 6.52 mg/l. In conclusion, metformin appears to accumulate in erythrocytes and, consequently, may be part of a deep compartment for the drug. This evidence of slow decline in erythrocyte metformin concentration may contribute to retrospective diagnosis of metformin accumulation and to refinements in adjusting metformin dosage to renal function.