PHARMACOKINETICS -THERA PEUTICS

Clin. Pharmacokinel. 21 (4): 308-317. 1991 0312-5963/91 /00 I0-0308/$05.00/0 © Adis International Limited. All rights reserved. CPK1068

Pharmacokinetic Optimisation of Oral Hypoglycaemic Therapy Piero Marchetti, Rosa Giannarelli, Alberto di Carlo and Renzo Navalesi Cattedra Malattie del Metabolismo, Istituto di Clinica Medica II, Universita di Pisa, Pisa, Italy

Contents 308 309 309 309 3/0

312 311 313

315 315 315 315 316 316

Summary

Summary I. Sulphonylureas 1.1 Chemistry 1.2 Mechanism of Action 1.3 Pharmacokinetics 1.4 Drug Interactions I.5 Clinical Use 1.6 Choice of a Particular Sulphonylurea 2. Biguanides 2.1 Chemistry 2.2 Mechanism of Action 2.3 Pharmacokinetics 2.4 Clinical Use 3. Conclusions

Two main classes of oral hypoglycaemic drugs, the sulphonylureas and the biguanides, are currently used in the therapy of type II, non-insulin-dependent diabetes mellitus (NIDDM). The basic pharmacokinetic properties of these agents are discussed with a view to efficient and safe treatment. Both first- and second-generation sulphonylureas are rapidly absorbed from the gastrointestinal tract. In the plasma compartment, these drugs are strongly bound to serum proteins. All sulphonylureas are metabolised in the liver, and the metabolites and the parent drugs are eliminated mainly in the urine, but also (second-generation derivatives) in the faeces. Rapid- and short-acting sulphonylureas may improve early insulin release and promote better postprandial glucose control. Long-acting derivatives may ensure better control of overnight glycaemia. The elderly are at risk of developing severe sulphonylurea-induced hypoglycaemia, and in this population the agent chosen should have a short or intermediate duration of action and no active metabolites. Caution is needed when prescribing any sulphonylurea in patients receiving drugs known to affect sulphonylurea action, and in those with impaired liver and/or kidney function. The bioavailability of the biguanides ranges from 40 to 60%. Binding to plasma proteins is absent or very low. Metformin and buformin are not metabolised and are excreted in the urine; phenformin undergoes hepatic hydroxylation and is excreted in both urine and faeces. Metformin is the only agent of this class currently recommended for clinical use. The main 'indications of metformin treatment are NIDDM associated with obesity and/or hyperlipidaemia, and in com-

Optimisation of Hypoglycaemic Therapy

309

bination with sulphonylurea both as primary treatment and when secondary failure occurs with sulphonylurea alone. Lactic acidosis may develop in patients receiving therapy with biguanides, especially in the presence of a preexisting contraindication to their use.

The result of optimal treatment for patients with diabetes mellitus should be the alleviation of symptoms and, more challenging, the prevention or minimisation of diabetic complications, with no detrimental effect on the patient. There is increasing evidence that diabetic microangiopathy and neuropathy are at least partially related to the degree and duration of the hyperglycaemia, and that the development of macroangiopathy results from the combined effects of various different risk factors, such as hyperglycaemia, hyperinsulinaemia, hyperlipidaemia, hypertension and cigarette smoking (Colwell et al. 1981; West et al. 1980). Thus, although the specific treatment goal for each patient should be defined by the patient and the physician (with consideration given to the age of the patient, other illnesses and treatments, and the potential danger of hypoglycaemia), nevertheless, any effective therapeutic intervention in patients with diabetes mellitus should ideally restore normoglycaemia, correct all other possible metabolic derangements and (when necessary) modify the lifestyle (Marchetti & Navalesi 1989). In type II, non-insulin-dependent diabetes mellitus (NIDDM), the most frequently encountered type of diabetes mellitus in clinical practice, some of these targets can initially be pursued by diet therapy and physical exercise. If good diabetic control (table I) is not achieved or maintained, then oral hypoglycaemic agents must be started. Two main classes are available for the treatment· of NIDDM: sulphonylureas and biguanides. Knowledge of the basic pharmacokinetic properties of each agent is the key to efficient and safe treatment.

this class are substituted arylsulphonylureas. Firstgeneration agents have a phenyl ring with simple substituent groups as R\ and usually an aliphatic side-chain as R2; in the second-generation derivatives, R\ is a more complex radical and R2 is a cyclic group. 1.2 Mechanism of Action

Sulphonylureas increase insulin secretion and enhance insulin action, thereby improving glucose tolerance. Boyd (1988) proposed the following model for sulphonylurea-induced insulin release from {3 cells: the drugs bind to plasma membrane receptors on the outside of the {3 cells; binding inhibits the ATP-sensitive K+ channels, with reduced K+ conductance causing membrane depolarisation, opening of the voltage-dependent Ca++ channels and influx of Ca++ into the cytoplasm; this latter event results in stimulation of the movement of insulin secretory granules to extrusion through the plasma membrane. In the past few years several studies have shown that sulphonylureas also have extrapancreatic effects. Indeed, during long term treatment plasma insulin levels may be unchanged, or even decreased, despite sustained reduction of plasma glu-

Table I. Definition of diabetic control. 70% of the values of individual blood tests must fit the criteria (after Lebovitz 1990; Marchetti & Navalesi 1989)

1. Sulphonylureas 1.1 Chemistry

The sulphonylureas are traditionally divided into 2 groups or generations, and their structural relationships are shown in figure I. All members of

Plasma glucose fasting after meal 1h 3h

Excellent

Good

Fair

Poor

(mg/L) 1500

1500

310

c/in. Pharmacokinet. 21 (4) 1991

cose levels. Extrapancreatic antidiabetic actions of sulphonylureas include potentiation of insulin action in muscle, liver and adipose tissue, and reduction of hepatic extraction of insulin (Kolterman & Olefsky 1984). The current consensus is that the action of insulin is potentiated through postreceptor mechanism, with any increases in insulin receptor number probably being a secondary effect. Both {3 cell function and insulin action are also improved as a consequence of correction of hyperglycaemia.

First generation

1.3 Pharmacokinetics

The major pharmacokinetic properties of the sulphonylureas are summarised in table II. Both first- and second-generation agents are ordinarily rapidly absorbed from the gastrointestinal tract.The time to peak plasma drug concentration (t max ) is shortest (2 to 3h) for glipizide, glibenclamide (glyburide) and tolbutamide, and longest (3 to 6h) for chlorpropamide and gliclazide. Absorption of

Rl

Tolbutamide

Chlorpropamide

Tolazamide

H3C-

-C4H9

CI-

-C3H7

H3C-

Acetohexamide

R2

H3CCO-

-NO

-0

Second generation

CI Glibenclamide (glyburide)

~ONH(CH"'OCH3

Glipizide

H3C~ONH(CH2)2N

Gliclazide

Fig.

1. Structural formulae of some sulphonylureas.

H3C-

-0 -0 -NCl)

Optimisation of Hypoglycaemic Therapy

311

Table II. Pharmacokinetic parameters of the principal oral hypoglycaemic agents

Drug

t1J2

Vd

(h)

(L/kg)

PB

Main mode of

Activity of

fe a

Duration of

metabolism

metabolites

(%)

action (h)

Hepatic hydroxylation

250% of parent

60

8-18

Hepatic hydroxylation

Partial activity

80-90

24-72

Hepatic hydroxylation

< 15% of

50

16-24

Sulphonylureas

Acetohexamide

0.8-2.4b

Ionic

4-8c Chlorpropamide

24-48

0.09-0.27

Ionic

or side-chain cleavage Nonionic

parent

1.4-5

0.3

Gliclazide

6-15

0.2

Nonionic

Hepatic hydroxylation

No activity

60-70

.,;; 24

Glipizide

1-7

0.16

Nonionic

Hepatic hydroxylation

No activity

70

12-24

Tolazamide

4-7

Ionic

Hepatic carboxylation

Very low activity

85

12-18

No activity

100

6-12

Glibenclamide (glyburide)

and hydroxylation Tolbutamide

3-5

0.1-0.25

Ionic

Hepatic hydroxylation and carboxylation

Biguanides

Buformin

2.6

3.1

?

Not metabolised

Metformin

1-4.5

1

0

Not metabolised

Phenformin

5-15

5-10

0-20

Hepatic hydroxylation

a Remainder excreted in the bile. b Parent drug. c Active metabolite. Abbreviations: tv, = elimination half-life; Vd into the urine.

No activity

90-100

6-10

90-100

8-12

50-60

8-14

= apparent volume of distribution; PB = protein binding; fe = fraction of drug

chlorpropamide is delayed by food, and that of glic1azide is decreased by age. On entering the plasma compartment, sulphonylureas are strongly bound to serum proteins, mostly albumin. First-generation agents bind to albumin through ionic forces, whereas second-generation sulphonylureas bind through a nonpolar (non ionic) site. The latter should be less susceptible to displacement by drug anions, such as phenylbutazone, warfarin and salicylate. Nevertheless (MacKichan 1989), pure displacement interactions are seldom clinically important. All the sulphonylureas are metabolised in the

excreted

liver; metabolites and parent compounds are eliminated mainly in the urine, but also (second-generation agents) in the faeces. The metabolism of chlorpropamide and gliclazide is incomplete and about 20% of each drug is excreted unchanged. Their metabolites are usually inactive or only weakly active; however, the activity of acetohexamide in vivo is mostly due to its metabolite, hydroxy-hexamide. Genetic control of drug disposition has been reported for tolbutamide (with a monogenic autosomal transmission of rapid and slow inactivation), and suggested for chlorpropamide and glibenclamide. Microsomal oxidation of the

C/in. Pharmacokinet. 21 (4) 1991

312

Table III. Drug interactions with the sulphonylureas Drug

Mechanism

Potentiation of sulphonylurea effects Pharmacokinetic

Antiepileptics

Displacement from plasma proteins 8

Clofibrate Indobufen Phenylbutazone Salicylates

drugs to the hydroxy derivative is the site of genetic control. Alkalinisation and acidification of the urine respectively increase and reduce the urinary excretion of chlorpropamide. Glipizide, glibenclamide and tolbutamide have the shortest half-lives (to;,), whereas chlorpropamide has by far the longest. There is no clear correlation between half-life and duration of hypoglycaemic action; the reason for this is still unclear.

Sulfaphenazole Sulfinpyrazone

1.4 Drug Interactions

Sulphonamides Chloramphenicol

Inhibition of hepatic metabolism

Dicumarol Halofenate Indobufen Phenylbutazone Propranolol Sulfafenazole Sulfinpyrazone Allopurinol

Reduced urinary excretion

Phenylbutazone Probenecid Salicylates

Table III represents an attempt to summarise the interactions of sulphonylureas with drugs known to potentiate or attenuate their hypoglycaemic action through pharmacokinetic and/or pharmacodynamic mechanisms (Marchetti & Navalesi 1989). The list will seem incomplete to some readers, and excessively detailed to others; both are probably right, and the message is that extreme caution and attention are needed when prescribing any oral hypoglycaemic agent in patients who are receiving other drugs, and vice versa.

Sulfinpyrazone Sulphonamides

1.5 Clinical Use

Pharmacodynamic

Monoamine oxidase inhibitors

Increased insulin secretion, decreased gluconeogenesis

Salicylates Alcohol (short term)

Decreased gluconeogenesis

Attenuation of sulphonylurea effects Pharmacokinetic

Alcohol (long term)

Induction of hepatic metabolism

Phenobarbital Rifampicin Pharmacodynamic

iJ-Blockers

Antagonism of insulin release or

Corticosteroids

action

Diuretics Estrogens Indomethacin Nicotinic acid Verapamil a

Clinically important with sulfaphenazole only (MacKichan 1989).

Although rates of primary failure as high as 36% have been reported for the sulphonylureas, careful initial selection of patients drastically reduces this figure (Marchetti & Navalesi 1989). The patients most likely to respond satisfactorily to these drugs are those with onset of diabetes after 40 years of age and duration of disease of less than 5 to 10 years, who are of normal weight or obese, have never taken insulin or have been well controlled with less than 30 units/day, do not have ketoacidosis and present with fasting plasma glucose levels of less than 2000 mg/L (200 mg/dl). Contraindications for sulphonylurea therapy are type 1 diabetes (inefficacy), pregnancy (risk of teratogenesis or hypoglycaemia of the neonates), history of severe adverse reactions to the drugs, severe liver or kidney disease (risk of hypoglycaemia), or the presence of severe stress such as infection, surgery or trauma (possible inefficacy). The suggested

Optimisation of HypogJycaemic Therapy

313

Table IV. Recommended dosages for the principal oral hypoglycaemic agents Drug

Initial

Maximum

dosage (mg/day)

dosage (mg/day)

No. of doses per day

Sulphonylureas Acetohexamide

250-500

1500

Chlorpropamide

100-150

500

Glibenclamide (glyburide) Gliclazide Glipizide

2.5-5.0

40-80 2.5-5.0

1-2 2

15

1-2

320

1-2

20

1-2

Tolazamide

100-250

1000

1-3

Tolbutamide

500-1000

2000

1-3

Buformin

50-100

300

Metiormin

1000-1500

2500

2-3

100

1-2

Biguanides

Phenformin

12.5-25

2

starting and maximum dosages are detailed in table IV. The dose can be increased weekly or bimonthly, until good control is achieved or the maximum dose is reached. If secondary failure occurs, and there is no evidence of dietary indiscretion or an intercurrent illness, then the addition of either a biguanide or insulin must be considered. 1.6 Choice of a Particular Sulphonylurea The choice of a particular sulphonylurea depends on the clinical characteristics of the patient and the properties of the specific agent. In particular, hypoglycaemic potency, therapeutic effectiveness, onset and duration of action, metabolism and excretion, and the risk of adverse effects must be carefully considered (Lebovitz 1990). Second-generation agents have 50 to 100 times the hypoglycaemic potency of first-generation sulphonylureas. Tolbutamide and chlorpropamide appear to be effective at a plasma concentration of 0.5 to I J,tmol/L, whereas glipizide and glibenclamide are active at concentrations of 50 to 100 nmol/ L. The higher molecular concentrations of first-

generation agents might explain the increased frequency of some adverse effects of these drugs, such as alcohol-induced flushing and water retention. In terms of therapeutic effectiveness, chlorpropamide and second-generation agents appear to have similar clinical efficacy (Melander et al. 1989). Tolbutamide has a lower hypoglycaemic effect, so it may be useful to switch a patient who is not well controlled on the maximum dosage of this drug to a second-generation agent. Onset and duration of action are 2 very important differences among the sulphonylurea agents, and depend on variations in the pharmacokinetic properties of each drug (rate and extent of absorption, distribution, metabolism and excretion). Glipizide and tolbutamide are the most rapid and short-acting sulphonylureas available at present. The rapid onset should reduce the delay in acute insulin release, but hypoglycaemia may occur soon after administration depending on the timing of the dosage in relation to meals. The short duration of action should reduce the risk of chronic hyperinsulinaemia, long-lasting hypoglycaemia and desensitisation to the drug. Chlorpropamide is the longest-acting sulphonylurea with the greatest delay in onset and glibenclamide has similar characteristics, this explains the fact that the greatest frequency of hypoglycaemia occurs in patients taking chlorpropamide (0.34/1000 treatment years) and glibenclam ide (0.38/1000 treatment years). So far as metabolism and excretion are concerned, an agent which is metabolised to inactive CH3-CH2-CH2-CH2-NH-~-NH-~-NH2

NH

NH

Buformin

Metiormin

": Q-c -

H2-CH2-NH-C-NH-C-NH2 II II NH NH

Phenformin Fig.

2. Structural formulae of the principal biguanides.

...,

""

Non-insulin-dependent diabetes mellitus

..

----~

GFR 30-100 ml/min

• Age

.

< SOy

Glibenclamide (glyburide) Ghpizide Gliclazide Tolbutamide Metformin

•

"

Age

Age > SOy

•

Ghpizide Gliclazide Tolbutamide Metformin

[

y

< SOy

.

Gliclazlde Glipizide

Age > SOy

J

.

Gllclazide Gllpizide Insulin

Q ~'

~ ~

~

l

5'

~

....""

~ Fig.

3. Indications for the use of oral hypoglycaemic drugs.

....'0 ....'0

Optimisation of HypogJycaemic Therapy

derivatives or can be excreted through alternative routes, such as the second-generation agents, should be preferred in patients with mild or modest impairment in renal function [glomerular filtration rate (GFR) 30 to 100 ml/min). None of the sulphonylureas should be used in patients with hepatic or renal failure (GFR < 20 ml/min). In conclusion, a patient with NIDDM, aged 30 to 60 years, otherwise healthy and well nourished, can be effectively treated with most sulphonylureas. Rapid- and short-acting agents may improve early insulin release, promote better control of postprandial blood glucose, and maintain the insulinotropic effect through discontinuous exposure. On the other hand, a long-acting derivative may ensure better control of overnight glycaemia. Older patients, especially ifless well nourished and chronically ill, are at risk of developing severe hypoglycaemia and should be treated with shortor intermediate-acting sulphonylurea that has no active metabolites. Second-generation agents offer some advantages in terms of fewer adverse effects and an alternative (biliary) route of excretion. The metabolism and excretion of each agent must be carefully taken into account in patients with impaired liver and kidney function.

2. Biguanides 2.1 Chemistry

Biguanides consist of 2 molecules of guanidine linked together by the removal of an ammonia group, and a hydrocarbon side-chain of various types and lengths. Of the more than 300 known biguanide derivatives, only 3 are of clinical interest (fig. 2): buformin, metformin and phenformin. At physiological pH values, biguanides are protonated. Due to the nonpolar hydrocarbon side-chains, they are able to bind to nonpolar hydrophilic structures such as the phospholipids of biological membranes. 2.2 Mechanism of Action

Biguanides should be considered as antihyperglycaemic rather than hypoglycaemic agents, since they do not have any significant effect on blood

315

glucose levels in healthy volunteers and are not known to cause hypoglycaemia. The suggested mechanisms of action of this class of drugs appear to differ among the various agents: for example, inhibition of gluconeogenesis plays a more important role with phenformin than with buformin and metformin. Metformin lowers blood glucose in the diabetic patient primarily by potentiating the action of insulin, most probably at a postreceptor site (Benzi et al. I 990). Increased muscle glucose uptake and decreased hepatic glucose production are the result of the improved insulin sensitivity. Although there is agreement that biguanides do not stimulate insulin release, recent experimental data have suggested that these drugs may have some pancreatic action (Giannarelli et al. 1988). 2.3 Pharmacokinetics

After oral administration, the bioavailability of biguanides is 40 to 60%. Metformin accumulates transiently in the oesophagus, stomach, duodenum and kidney, and phenformin and buformin in the liver, pancreas, kidney and muscle. Some of the pharmacokinetic properties of the biguanides are shown in table II: tmax and t'l2 are shorter for metformin than for the other agents, due to lack of hepatic metabolism. Metformin is excreted by the kidney, with tubular secretion probably being a major mechanism of urinary excretion (Sirtori et al. 1978). There is a clear correlation between the renal clearance of metformin and creatinine clearance. The binding of phenformin to plasma proteins ranges from insignificant to between 12 and 20%. Approximately 50% of the drug is metabolised in the liver, and biotransformation is characterised by genetic polymorphism; the metabolite is inactive, so that 'slow' metabolisers are probably at higher risk of developing lactic acidosis. Phenformin is excreted in both urine and bile. Information on the pharmacokinetics of buformin is scarce. Although in laboratory animals the agent is hydroxylated in the liver, this does not appear to be the case in humans. The lack of hepatic metabolism of metformin

Clin. Pharmacokinel. 21 (4) 1991

316

and buformin reduces the possibility of drug interactions through pharmacokinetic mechanisms. Nevertheless, it is possible that drugs which are able to potentiate or reduce the effects of sulphonylureas by pharmacodynamic mechanisms (table III) may affect the action of biguanides also. 2.4 Clinical Use Table IV shows the suggested dosage for all 3 biguanides. However, metformin is the only agent currently recommended for clinical use, phenformin and buformin having been withdrawn from the market in most countries because of the risk of lactic acidosis (see below). The main indication for metformin as primary therapy is NIDDM associated with obesity and/or hyperlipidaemia (Vigneri & Goldfine 1987). It can also be used in addition to a sulphonylurea when secondary failure occurs with the latter agents. Finally, since the adverse effects of oral agents may be dosage related, sulphonylurea and metformin as combination therapy for primary treatment may offer the possibility of reducing such adverse effects, if therapeutic action can be obtained with lower dosages than are required for monotherapy. When used as a primary drug, metformin appears to have a lower antihyperglycaemic potency than the sulphonylureas. The biguanide, however, is better tolerated than most sulphonylurea agents: in fact, the risk oflactic acidosis caused by metformin is less than the risk of severe hypoglycaemia induced by chlorpropamide and other sulphonylurea drugs (Campbell 1985). Although insulin is the most effective therapeutic option for treating patients not well controlled by maximum-dosage sulphonylurea therapy, the incidence of severe hypoglycaemia is nevertheless higher in patients receiving sulphonylurea plus insulin. The addition of metformin to sui phony urea in NIDDM patients with secondary failure is currently thought to be better tolerated and is usually, although not always, effective. Unfortunately, there is no study addressing the possible pharmacokinetic interactions between sulphonylureas and biguanides. Since the latter can reduce the intestinal

absorption of many substances, and both sulphonylureas and biguanides are eliminated mainly by the kidney, this issue requires investigation. Lactic acidosis (due to overproduction oflactate through inhibition of mitochondrial respiration and increased anaerobic glycolysis, and decreased lactate utilisation by inhibition of gluconeogenesis) is the most life-threatening adverse reaction to the biguanides. The relative risk is much lower for metformin (0.05 cases/1000 patients/year) than for phenformin (0.64 cases/1000 patients/year). The short half-life and duration of action of metformin make this drug less hazardous than the others of the same class. In almost all the cases oflactic acidosis associated with metformin, a preexisting contraindication to its use was present, such as a concomitant disease (uraemia, liver disease, cardiac and/or respiratory insufficiency, alcoholism) or the use of drugs able to reduce the cellular redox potential (barbiturate, salicylate, phenothiazine derivatives).

3. Conclusions The sulphonylureas and biguanides (i.e. metformin) represent useful tools in the treatment of patients with NIDDM, when diet therapy and physical exercise have failed. Figure 3 offers some guidelines for the proper clinical use of these oral agents. Consideration has been given to the halflife and duration of action of the drugs used alone. Possible interactions with other drugs have not been included in the figure, but must also be taken into account. In the western world, from 3.0 to 11.0% of the population is affected by NIDDM. Five to 10% of these patients still develop end-stage renal disease, 10 to 20% proliferative diabetic retinopathy, and approximately 40% macro vascular disease (peripheral and/or coronary artery disease). It is to be hoped that optimisation of oral hypoglycaemic treatment will lead to a permanent reduction in these dramatic figures in the future.

Optimisation of Hypoglycaemic Therapy

References Benzi L, Trischitta v, Ciccarone AM, Cecchetti P, Brunetti A, et al. Improvement with metformin in insulin internalization and processing in monocytes from NIDDM patients. Diabetes 39: 844-849, 1990 Boyd III AE. Sulfonylurea receptors, ion channels, and fruit flies. Diabetes 37: 847-850, 1988 CampbelllW. Metformin and the sulfonylureas: comparative risks. Hormone and Metabolic Research 15 (Suppl.): 105-111, 1985 Colwell JA, Lopes-Virella M, Halushka PV. Pathogenesis ofatheroslerosis in diabetes mellitus. Diabetes Care 4: 121-131, 1981 Giannarelli R, Marchetti P, Zappella A, Masoni A, di Carlo A, et al. Metformin stimulates insulin release from porcine islets of Langerhans. Diabetologia 31 : 493A, 1988 Kolterman OG, Olefsky JM . The impact of sulfonylurea treatment upon the mechanisms responsible for the insulin resistance in type II diabetes. Diabetes Care 7 (Suppl. I): 81-88 1984 ' Lebovitz HE. Oral hypoglycemic agents. In Ellenberg & Rifkins (Eds) Diabetes mellitus: theory and practice - 2nd ed., pp. 554574, 1990

317

MacKichan JJ. Protein binding drug displacement interactions: fact or fiction? Clinical Pharmacokinetics 16: 65-73, 1989 Marchetti P, Navalesi R. Pharmacokinetic-pharmacodynamic relationships of oral hypoglycaemic agents: an update. Clinical Pharmacokinetics 16: 100-128, 1989 Melander A, Bitzen PO, Faber 0 , Groop L. Sulfonylurea antidiabetic drugs: an update of their clinical pharmacology and rational therapeutic use. Drugs 37: 58-72, 1989 Sirtori CR, Franceschini G, Galli-Kienle M, Cighetti G, Galli G. Disposition of metformin (N,N-dimethylbiguanide) in man. Clinical Pharmacology and Therapeutics 24: 683-693, 1978 Vigneri R, Goldfine ID. Role of metformin in treatment of diabetes mellitus. Diabetes Care 10: 118-112, 1987 West KM, Erdreich U, Stober JA. A detailed study of risk factors for retinopathy and nephropathy in diabetes. Diabetes 29: 501508, 1980

Correspondence and reprints: Dr Piero Marchetti. Cattedra di Malattie del Metabolismo, Istituto di Clinica Medica II, via Roma 67, 56100 Pisa, Italy.