0021-972X/92/7405-1020$03.

00/0 Journal of Clinical Endocrinology Copyright 0 1992 by The Endocrine

and Metabolism
Society

Vol. 74, No. 5 Printed in U.S.A.

Combined Metformin-Sulfonylurea Patients with Noninsulin-Dependent Poor Glycemic Control*

GERALD M. REAVEN, ROMAN SKOWRONSKI,
Department and Clinical

Treatment Diabetes

of in Fair to

PETER JIN-C

JOHNSTON, CLARIE B. HOLLENBECK, ZHANG, IRA D. GOLDFINE, AND Y.-D. IDA CHEN

of Medicine, Stanford University School of Medicine, and the Geriatric Research, Education, Center, Department of Veterans Affairs Medical Center, Palo Alto, California 94304

ABSTRACT. The effect of metformin treatment was studied in I3 patients with noninsulin-dependent diabetes mellitus (NIDDM), whose fasting plasma glucose concentration was greater than 10 mmol/L with maximal sulfonylurea doses. Patients were studied before and 3 months after receiving 2.5 g/ day metformin. The fasting plasma glucose concentration (12.4 + 0.8 vs. 8.8 + ,0.7 mmol/L), mean hourly postprandial plasma glucose concentration from 0800-1600 h (14.0 f 1 vs. 9.4 + 0.9 mmol/L), and glycosylated hemoglobin level (12.3 + 0.6% vs. 9.0 + 0.6%) were all significantly (P < 0.005-0.001) lower after the administration of metformin. The improvement in glycemic control was associated with a 24% increase (P < 0.05) in insulinstimulated glucose uptake during glucose clamp studies and a 16% decrease in basal hepatic glucose production (P c 0.05). Mean hourly concentrations of plasma insulin (411 Lf 73 vs. 364 + 73 pmol/L) and FFA concentrations (440 + 31 vs. 390 -L 40 amol/L) were also lower after 3 months of metformin treatment.

However, neither insulin binding nor insulin internalization by isolated monocytes changed in response to metformin. Finally, plasma triglyceride, very low density lipoprotein triglyceride, and very low density lipoprotein cholesterol were significantly decreased (P < O.Ol-0.001). and high density lipoprotein cholesterol was significantly increased (P < 0.001) after metformin treatment. Thus, the addition of metformin to sulfonylureatreated patients with NIDDM not in good glycemic control significantly lowered fasting and postprandial plasma glucose concentrations, presumably due to the combination of enhanced glucose uptake and decreased hepatic glucose production. Since the dyslipidemia present in these patients also improved, the results suggest that metformin may be of significant clinical utility in patients with NIDDM not well controlled with sulfonylurea compounds. (J Clin Endocrirwl Metab 74: 1020-1026, 1992)

ETFORMIN, a biguanide derivative, is widely used as an oral antidiabetic agent throughout the majority of the world (1, 2). More recently, studies of metformin have been initiated in the United States, and we have shown (3) that metformin significantly lowers plasma glucose concentrations in untreated patients with noninsulin-dependent diabetes mellitus (NIDDM). Furthermore, the improvement in glycemic control in metformin-treated patients with NIDDM was associated with a decrease in fasting and postprandial plasma insulin and triglyceride (TG) concentrations and an increase in plasma high density lipoprotein (HDL) cholesterol concentrations (3). In addition to being an effective antidiabetic agent by itself, metformin has been widely used in sulfonylurea-treated patients with NIDDM not
Received March 12, 1991. Address all correspondence and requests for reprints to: G. M. Reaven, M.D., Geriatric Research, Education, and Clinical Center (182. B), Veterans Administration Medical Center, 3801 Miranda Avenue, Palo Alto, California 94304. * This work was supported by grants from the NIH (RR-00070 and DK-30732), the Nora Eccles Treadwell Foundation, and Lipha Chemicals, Inc. 1020

in good glycemic control (2). In such patients, the only therapeutic alternative currently available in the United States to achieve glycemic control is insulin. Although insulin treatment is capable of achieving excellent glycemic control in these patients, the amounts necessary are often quite large (4, 5), and evidence has been presented indicating that macroangiopathy in NIDDM may be related to insulin dose (6). In light of these considerations, metformin seems to provide an alternative therapeutic option. Consequently, we decided to initiate the current study, which had three major goals. The first was a pragmatic one and represented an attempt to quantitate the clinical utility of adding metformin to the treatment program of patients with NIDDM who had responded with a fall in plasma glucose concentration during sulfonylurea treatment, but who were still not in good glycemic control. The second aim was to determine whether the improvement in glycemic control associated with the addition of metformin to sulfonylurea treatment led to any change in lipoprotein metabolism. The third and final aim was to conduct experiments aimed at

Downloaded from jcem.endojournals.org at Indonesia:Endo Jnls Sponsored on May 6, 2007

COMBINED

METFORMIN associto the metabefforts

TREATMENT

IN NIDDM

providing insight into the physiological changes ated with the addition of metformin that led improvement of glycemic control and lipoprotein olism. The results presented here represent our to respond to these three goals. Materials and Methods

Thirteen patients (10 males and 3 females) with NIDDM were studied. They were 57 f 2 yr old (mean + SEM) and had a body mass index of 28.7 + 1.2 kg/m. The treatment program in all 13 patients had been initiated with a diabetic diet. However, all 13 patientsremained significantly hyperglycemic. In 11 of the 13 subjects, the persistence of hyperglycemia had led to the introduction of a sulfonylurea agent before enrollment in our study. They all had demonstrated an initial fall in fasting plasma glucose concentration greater than 2.5 mmol/L, but had continued to have a fasting plasma glucose concentration above 10.0 mmol/L during maximal oral therapy. To avoid the confounding variable of different sulfonylurea compounds, patients receiving other drugs were swit,ched to maximal doses of glipizide (40 mg/day) and followed for an additional 6-8 weeks. In 2 instances, patients had never been treated with sylfonylurea compounds. Since their fasting plasma glucose level was above 10.0 mmol/L on the diet alone, they were started on glipizide as well. Metformin was started when the fasting plasma glucose concentration remained above 10 mmol/ L after 6-8 weeks of treatment with 40 mg glipizide/day. Patients were not taking any drugs other than glipizide or metformin known to affect carbohydrate or lipid metabolism. The subjects were admitted to the Stanford General Clinical Research Center (GCRC) and fed a weight maintenance diet containing (as percentage of total calories) 17% protein, 40% fat, and 43% carbohydrate. Each meal had this same relative content of nutrients. Meals were eaten at 0800,1200, and 1800 h and contained 20%, 40%, and 40%, respectively, of the days total caloric intake. Measurements were made of plasma glucose (7), FFA (8), insulin (9), and TG (10) concentrations, both in the fasting state and at hourly intervals from 0800-1600 h. Plasma was separated immediately and stored frozn until analyzed. All samples from each individual were analyzed in the same assay. Measurement of the glycosylated hemoglobin concentration (11) was also made during the baseline period. In addition, blood was obtained after an overnight fast on two occasions before metformin treatment for measurement of fasting plasma cholesterol (12) and lipoprotein TG and cholesterol concentrations. These samples were subjected to sequential density ultracentrifugation (13) to separate very low density lipoprotein (VLDL), intermediate low density lipoprotein (IDL), low density lipoprotein (LDL), and HDL fractions at densities below 1.006 and at 1.019 and 1.063 g/mL, respectively, and the triglyceride and cholesterol contents of various fractions were determined as described previously (3, 14). Glucose clamp studies were performed after an overnight fast before metformin treatment was initiated. Because this method was previously described in detail (15, 16), only the general procedure is outlined here. Arterialized venous blood samples were obtained from an indwelling catheter inserted

retrograde into a hand vein, with the hand being placed in a radiant warmer maintained at 70 C. Plasma was immediately separated in a Beckman microfuge (Fullerton, CA), and glucose was determined with a Beckman Glucose Analyzer II. After determining the baseline plasma glucose concentration, a primed continuous infusion of insulin (40 mu/m. min) and somatostatin (350 pg/h) was started and continued for 240 min. Plasma glucose was determined every 5 min thereafter. A variable infusion of glucose was started 5 min after the start of the insulin infusion and adjusted to maintain plasma glucose within 10% of the baseline value during the 240-min infusion period. To quantify total glucose turnover during the clamp studies, [3-HIglucose (25 tiCi) was injected as an iv bolus dose before the start of the clamp study, followed by a constant infusion of 0.25 &i/min for 4 h. Aliquots of plasma were collected at 30-min intervals for 210 min and at lo-min intervals during the last half-hour, precipitated with Ba(OH), and ZnSOI, then centrifuged, and the protein-free supernatant was evaporated in a scintillation vial. The plasma glucose concentration and radioactivity were determined, and glucose specific activity was calculated. The glucose appearance rate (Ra) and disappearance rate (Rd) were calculated during the last 60 min of the clamp study, with the nonsteady state equation of Steele (17). Urinary glucose loss was also quantified and subtracted from Rd to provide an estimate of total tissue glucose uptake during this period. The difference between Ra and glucose infusion rate during the clamp study provided an estimate of hepatic glucose production (HGP). To compare the glucose Rd of patients with different plasma glucose concentrations, the glucose MCR was calculated by dividing the value of total tissue glucose uptake by the plasma glucose concentration during the study (16). HGP was also assessedunder basal conditions (18). In this situation, HGP is equal to glucose Ra, since no exogenous glucose is administered. Patients received a primed (25 &i) continuous (0.25 &i/min) infusion of [3-HIglucose for 300 min (18); blood samples were obtained at 30-min interval for 270 min, then every 10 min for the last 30 min. The plasma glucose concentration and radioactivity were determined, and Ra was calculated as described above. To assess the impact of metformin treatment on insulin receptor activity, insulin binding and internalization were measured using freshly isolated mononuclear cells from blood samples. Mononuclear cells were isolated by using SepracellMN (19), and monocyte content was quantified by counting cells in a hemocytometer and determining the percentage of monocytes by esterase staining (20). To determine insulinbinding activity (3, al), mononuclear cells (2 X lo7 cells/mL) were incubated in 1.2 mL Krebs-Ringer and 10 mM HEPES buffer, pH 7.8, containing 0.6 ng/mL A-14 monoiodinated human insulin, 5 mM n-glucose, and 2% BSA in the absence or presence of 100 pg/mL unlabeled insulin. After l-h incubation at 37 C in a shaking water bath, aliquots of cell suspension were washed twice in ice-cold 0.15 M saline containing 3.5% BSA, and the radioactivity associated with the cell pellet was determined in a -,-counter. Specific binding was determined by subtracting the amount of [Ijinsulin bound in the presence of excess (100 Fg/mL) unlabeled insulin. To determine the

Downloaded from jcem.endojournals.org at Indonesia:Endo Jnls Sponsored on May 6, 2007

1022

REAVEN ET AL.

JCE&M.19!32 Vol74.No6

amount of [261]insulin internalized, surface-bound [SI]insulin was extracted by mild acid (22). Briefly, at the end of incubation for [*I]insulin binding, aliquots of cell suspension were incubated with a barbital acetate buffer, pH 3.0 (27 mM acetic acid and 20 mM barbital), at 0 C for 10 min and washed in ice-cold 0.15 M saline-3.5% BSA, and the residual (internalized) radioactivity was quantitated in a T-spectrophotometer. Specific internalization was determined by subtracting the amount of [*I]insulin internalized in the presence of an excess of unlabeled insulin, as described above for binding studies. After the baseline measurement, patients were started on metformin at a dose of 500 mg at the beginning of breakfast and discharged from the GCRC. They were seen at weekly intervals, when the plasma lactate concentration was determined (23), and the dose of metformin was gradually increased to 2.5 g/day. Metformin was given in two divided doses at the beginning of the meal (1.5 and 1.0 g before breakfast and dinner, respectively). A weeks supply of medication was given at each visit, and patients were questioned as to their compliance. Within the limitation of this approach, there was essentially 100% compliance throughout the study. After 3 months of metformin treatment at the maximum dose, patients were again admitted to the GCRC, and all baseline measurements were repeated, as described previously, with one exception. During the first glucose clamp study, glucose was given at the rate necessary to maintain basal plasma glucose concentration. When glucose clamp studies were performed after metformin treatment, each patient was infused with glucose at the same rate as during the premetformin study. This was to eliminate the possibility of unintentional operator bias (16) as well as provide a second approach to assess the change in insulinstimulated glucose disposal. More specifically, the plasma insulin concentration and the glucose infusion rate were the same in each patient before and after metformin treatment. Consequently, the difference in the steady state plasma glucose concentration during the clamp study provides another estimate of insulin action. All patients finished the study, with the only symptomatic complaints being occasional feelings of abdominal bloating, cramping, and fullness. Weight did not change significantly (84.9 + 3.6 kg before and 85.7 it 3.7 kg after) with metformin treatment. Data are expressed as the mean t SEM, and statistical evaluation was performed with the Statistical Analysis System program (SAS Institute, Cary, NC), using the general linear models procedure. To evaluate the effect of metformin treatment, values before and after treatment were compared by either Students paired t test or three-way analysis of variance (24, 25).

treated patients after the midday meal. The decrease in

ambient plasma glucose concentration

before and after

initiation of metformin treatmentwas highly significant (P < 0.001, by three-way analysis of variance), as was

the fall in glycosylated

0.6 vs. 12.3 5 0.6%).

hemoglobin

concentration

(9.0 +

Fasting and postprandial

FFA (B), TG (C), and insulin

(D) concentrations from 0800-1600 h before and after metformin treatment are also shown in Fig. 1. These

data indicate that the plasma concentrations

of all three

of these variables were lower through the day after the addition of metformin (P < 0.001). As was the case with

the changes in plasma glucose concentration, the effect of metformin to lower plasma FFA, TG, and insulin levels tended to be of greater magnitude in the afternoon. The combination of lower ambient plasma glucose and insulin concentrations after metformin treatment is consistent with the view that metformin administration was

associated with enhanced insulin-mediated glucose disposal. The results of glucose clamp studies to test this hypothesis are shown in Table 1. As indicated in Muterials and Methods, the rates at which glucose and insulin were infused were identical before and after metformin treatment. As a result, the steady state plasma insulin concentrations were identical (861 & 70 vs. 834 + 50 pmol/L) during the two studies. However, despiie the same steady state plasma insulin concentrations and identical glucose infusion rates, steady state plasma glucose concentrations were significantly lower (P < 0.05)
after metformin treatment (8.8 + 0.7 vs. 10.5 & 0.5 mmol/

L). Evidence that metformin

ated with enhanced

administration

was associ-

glucose uptake was also supported by the higher values of glucose MCR (P C 0.05) after

metformin treatment. During the clamp studies, hepatic glucose output was indistinguishable from zero both before and after metformin treatment. Although insulinmediated glucose uptake appeared to be enhanced after
metformin treatment, there was no significant increase in monocyte insulin binding (5.3 + 0.9 vs. 6.4 f 1.6%) or insulin internalization (71 + 4 vs. 76 + 3%) associated with metformin treatment.

Measurements
min treatment

of basal HGP before and after metforare also seen in Table 1. Although basal

Results The effect of adding metformin on overall glycemic control is illustrated in Fig. 1A. Although the fasting plasma glucose concentration was lower in metformintreated patients (8.8 + 0.7 vs. 12.4 + 0.8 mmol/L; P < O.OOl), the metformin effect seems to be greatest from 1200-1600 h. Indeed, there was essentially no increase in the plasma glucose concentration in metformin-

plasma insulin concentrations were similar before (134 + 26 pmol/L) and after (121 & 23 pmol/L) metformin treatment, basal HGP was significantly lower after treatment (P < 0.05).

Fasting plasma and lipoprotein

concentrations

cholesterol

and TG
are

before and after metformin

treatment

listed in Table 2. The plasma cholesterol concentration was significantly (P C 0.001) lower after metformin treatment, due entirely to a decrease in VLDL cholesterol
(P < 0.001). Although IDL and LDL cholesterol concen-

trations

did not change in association

with metformin

Downloaded from jcem.endojournals.org at Indonesia:Endo Jnls Sponsored on May 6, 2007

COMBINED 20 r
16 -

METFORMIN

TREATMENT

IN NIDDM
FFA

102:1

A. &core
p-2000~

r
. .. . "'9. . . . . .p' .... .F . . . . . y. . . . . . f$ . . . . .p

B.

f 8 J

10

-P-J-++ * Y"'

. ..L.p

O! 0
0 r C.

a
ltiglycwide

10

11'

12

FIG. 1. Mean (+SEM) plasma glucose (A), FFA (B), TG (C), and insulin (D) concentrations of metformin treatment. TABLE 1. Estimates of insulin-mediated glucose clearance rate (MCR) and HGP before and after initiation of metformin treatment MCR (mL/m. min) 112 + 15 139 f 24 <0.05 HGP (pmol/m min) 400 + 29 334 + 12 <0.05

from 0800-1600 h before (0) and after (0) 3 months

LDL TG concentrations were also significantly lower (P < O.OOl), this change contributed quantitatively less to the decrease in plasma TG. Discussion The present study had three goals. The first was to ascertain the degree to which plasma glucose concentrations decreased when metformin was administered to patients with NIDDM, not ideally controlled with glipizide. The data in Fig. 1 indicate that glycemic control improved after the addition of metformin; the integrated plasma glucose response from 0800-1600 h was 34 f 0.03% lower after treatment, and the fasting plasma glucose concentration fell 28 f 0.3%. It should be noted that the average decline in the fasting plasma glucose concentration of 3.5 + 0.3 mmol/L was achieved in patients who already had responded to glipizide treat(millimoles per L) before and after initiation LDL 2.64 f 0.21 2.79 -c 0.17 NS 0.23 zk 0.02 0.20 + 0.02 -co.00 1 HDL 1.00 f 0.05 1.13 k 0.04 <O.OOl 0.12 f 0.01 0.11 f 0.01 NS of metformin treatment Plasma/HDL 5.3 k 0.29 4.3 f 0.20 <0.002

Before After P

administration, the HDL cholesterol concentration was significantly (P < 0.001) higher after metformin treatment. As a result, the ratio of total plasma cholesterol to HDL cholesterol decreased from 5.3 + 0.29 to 4.3 k 0.20 (P < 0.002). Metformin treatment was also associated with a significant decrease (P < 0.005) in total plasma TG concentration. The lowering of the total plasma TG concentration was primarily due to a decrease in the plasma VLDL TG concentration (P < 0.001). Although
TABLE 2. Plasma lipid and lipoprotein Plasma Cholesterol Before After 5.17 + 0.22 4.85 + 0.12 <O.OOl 2.74 + 0.46 2.14 k 0.43 <0.005 cholesterol and TG concentrations VLDL 1.03 * 0.18 0.61 + 0.16 <O.OOl 2.19 + 0.44 1.64 + 0.42 <O.OOl IDL

0.29 f 0.06 0.32 -c 0.05 NS 0.19 +- 0.01 0.19 * 0.01 NS

P
TG Before After P

Downloaded from jcem.endojournals.org at Indonesia:Endo Jnls Sponsored on May 6, 2007

1024

REAVEN

-__ ._-. E7 Al,

JCEBrM.1992
Vol74.NoS

ment with at least a 2.5 mmol/L fall in the fasting plasma glucose concentration. Thus, the glucose-lowering effect of metformin was additive to that of glipizide, providing an additional therapeutic approach to patients with NIDDM, who, although responding to sulfonylurea treatment, were still not in good glycemic control. Our second goal was to determine whether lipoprotein metabolism improved after the the addition of metformin to glipizide-treated patients. The most common lipid abnormalities in patients with NIDDM are a high plasma TG concentration and a low HDL cholesterol concentration (26, 27), and the current results show that the plasma TG concentration decreased and the HDL cholesterol concentration increased after metformin treatment. Since the total plasma cholesterol concentration also decreased in association with metformin treatment, the ratio of total cholesterol to HDL cholesterol was lower (from 5.3 to 4.3). The third goal of our study was to explore possible explanations for changes in glucose and lipid metabolism associated with the addition of metformin to glipizidetreated patients. The observation that the improvement in glycemic control with metformin was associated with lower day-long plasma insulin concentrations suggested that treatment led to an improvement in insulin action. This conclusion is supported by the results of the glucose clamp studies, which demonstrated that glucose uptake rates during the hyperinsulinemic glucose clamp study were significantly increased after metformin treatment. The improvements noted in insulin action in the present study are in contrast to our previous experience with metformin treatment (3). However, the patients in the present study were receiving concomitant treatment with glipizide, while in our previous report they were treated with metformin alone. Thus, the improvement noted in this patient population may be a result of the combination of therapies, rather than a direct effect of metformin per se; a possibility consistent with the results of previous data (28, 29) in which metformin-treated patients were receiving concomitant treatment with a sulfonylurea compound. Basal HGP was also lower after metformin treatment, a phenomenon described previously (30). The lower production rate of glucose in the basal state along with improved glucose uptake provide ample reasons for significant improvement in glycemic control associated with metformin treatment. In this context, it is worth noting that plasma FFA concentrations were also lower after metformin treatment. An increase in the plasma FFA concentration has been shown to decrease glucose uptake during glucose clamp studies (31), and acute lowering of the plasma FFA concentration has been shown to decrease HGP in a rat model of NIDDM (32). Thus, it is possible that an effect of metformin on FFA metabolism contributed to the enhanced glucose uptake

and the lower basal HGP after treatment. The current data also provide support for the view that metformin may act to lower glucose by decreasing the rate at which substrate is absorbed from the gastrointestinal tract. Metformin has been shown to decrease vitamin B12 and D-XyiOSe absorption (33), and it is possible that it could also decrease intestinal glucose absorption. The results in Fig. 1, A and C, showing that postprandial plasma glucose concentrations and TG concentrations were markedly lower after the noon meal are also consistent with an effect of metformin at the intestinal level. The experimental data also provide an explanation for the treatment-associated changes in lipoprotein metabolism. There is substantial evidence that the higher the circulating levels of insulin and FFA, the greater the hepatic TG secretion, and the higher the plasma TG concentration (34, 35). Since both plasma FFA and insulin concentration were lower after metformin treatment, the decrease in the plasma TG concentration was not unexpected. Furthermore, since plasma TG and HDL cholesterol concentrations are inversely related (36, 37), it is not surprising to see an increase in plasma HDL cholesterol concentration in association with a decrease in the plasma TG concentration. Plasma HDL cholesterol concentrations are also inversely related to the plasma insulin concentration (37), and the increase in plasma HDL cholesterol concentration may also be partly due to the observed decrease in ambient insulin concentration after metformin treatment. Clinical studies with metformin are now being performed in the United States as part of the process of developing a clinical data base to support a New Drug Application for the therapeutic use of metformin (personal communication, Goodman, A. M., Vice-President for Medical Affairs, Lipha Pharmaceuticals, Inc., New York, NY), and it seems reasonable in this context to also discuss the potential clinical utility of adding metformin to the treatment of patients not well controlled by sulfonylurea compounds alone. The results presented indicated that the use of metformin in this patient population was associated with a significant improvement in both glycemic control and lipoprotein metabolism in patients with NIDDM. Thus, the ability to add metformin to the treatment programs of the kind of patients we studied provides an alternative to the initiation of insulin treatment in these individuals. In general, patients prefer taking pills to giving themselves injections, but there are also theoretical advantages as to why combined sulfonylurea-metformin treatment may be preferable to insulin treatment. There are data from both normal subjects (38, 39) and patients with NIDDM (40) suggesting that hyperinsulinemia may increase the risk of coronary heart disease. Furthermore, it has been suggested that insulin dose is an independent risk factor for

Downloaded from jcem.endojournals.org at Indonesia:Endo Jnls Sponsored on May 6, 2007

COMBINED

METFORMIN

TREATMENT

IN NIDDM

coronary heart disease in patients with NIDDM (6). Since the addition of metformin to sulfonylurea treatment of patients was associated with a decline in circulating insulin concentrations, and in a significant number of individuals may remove the need to initiate insulin treatment, the putative clinical benefits of this approach in terms of coronary heart disease are self-evident. Furthermore, in a comparison of the effects of combined sulfonylurea metformin us. sulfonylurea-insulin treatment, weight gain was shown to be significantly greater in the group treated with sulfonylurea and insulin (29). Given the crucial role of obesity in the treatment of NIDDM, there are obvious advantages to a treatment approach that minimizes weight gain. Finally, the improvement in plasma lipoprotein metabolism associated with metformin also supports the view that the availability of metformin in the United States may increase treatment options for patients with NIDDM in a useful manner. In conclusion, the addition of metformin to the treatment program of patients not well controlled by sulfonylurea compounds was associated with improvements in both glycemic control and lipoprotein metabolism. The drug was well tolerated, body weight did not. change, and all patients enrolled were able to finish the study without difficulty. The fasting plasma lactate concentration, measured before and at weekly intervals during the metformin treatment period, did not change significantly in association with metformin treatment (1.42 + 0.12 to 1.39 f 0.09 mmol/L). On the other hand, it is important that our results be placed in perspective and not be the basis of excessive generalizations. In the first place, the effect of the intervention was only observed over a relatively short time interval in a very specific group of patients. Secondly, we did not perform a double blind placebo-controlled study. This decision was based upon the following considerations. The study was not performed to determine whether metformin was capable of lowering plasma glucose in NIDDM in light of previous evidence that this was the case (1, 2). Our patient population was in poor glycemic control and candidates for insulin therapy, and we did not believe it appropriate to withhold effective therapy for the several months needed to complete the study. Furthermore, although the ahdominal symptoms associated with metformin are relatively mild in magnitude, they occur commonly enough to make a blind study essentially impossible. A major goal of this study was to gain insight into the physiological changes associated with the addition of metformin in a carefully selected patient population. Given these considerations, we do believe that the data are useful despite the fact that it was not a double blind study. Finally, the results of the study of Groop et al. (29), which was quite similar to ours, did not show any improvement in plasma lipid

and lipoprotein concentrations. There was no ohvious difference in experimental design, and we can only point out that metformin treatment per se has been associated with a fall in the plasma TG concentration in a variety of different patient populations (41-43). Obviously, there is a need for further study of these and other pathophysiological changes associated with metformin treatment, both by itself and when added to other pharmacological agents used to improve glycemic control. Furthermore, it is necessary to again emphasize that this study was relatively short, and we do not know what would happen with continued treatment with combined metforminsulfonylurea. On the other hand, the data presented support the view that the use of metformin may both improve glycemic control and reduce the risk of coronary heart disease in patients not well controlled by glipizide alone. As such, it will he a drug of potential benefit if and when it is made available in the United St,ates.
Acknowledgments

These studies could not have been completed without the help of our Research Assistant Benjamin Varasteh and nurse practitioner B. Tighe.
References
1. Hermann 2. 3.

4.

5.

6.

7.

8. 9. 10.

11.

1%.

13.

LS. Metformin: a review of its pharmacological properties and therapeutic use. Diabetes Metab. 1979;5:233-45. Vigneri R, Goldfine I. Role of metformin in treatment of diabetes mellitus. Diabetes Care. 1987;10:118-22. Wu M-S, Johnston I, Sheu WH-H, Hollenbeck CB, Jeng C-Y. Effect of metformin on carbohydrate and lipoprotein metabolism in NIDDM patients. Diabetes Care. 1990;13:1-8. Reaven GM, Chen Y-DI, Coulston AM, et al. Insulin secretion and action in noninsulin-dependent diabetes mellitus. Is insulin resistance secondary to hypoinsulinemia? Am J Med. 1983;75:85-93. Hollenbeck CB, Chen Y-DI, Greenfield MS, Lardinois CK, Reaven GM. Reduced plasma high density lipoprotein-cholesterol concentrations need not increase when hyperglycemia is controlled with insulin in noninsulin-dependent diabetes mellitus. ,J Clin Endocrino1 Metab. 1986;62:605-8. Janka HU, Ziegler AG, Stand1 E, Mehnert H. Daily insulin dose as a predictor of macrovascular disease in insulin treated noninsulin-dependent diabetics. Diabetes Metab. 1987;13:359-64. Kadish AH, Litle RI,, Sternberg JC. A new, rapid method for determination of glucose by measurement of rate of oxygen consumption. Clin Chem. 1968;14:116-31. Noma A, Okabe H, Kita M. A new calorimetric micro-determination of free fatty acids in serum. Clin Chim Acta. 1973;43:317-20. Hales CN, Randle P,J. Immunoassay of insulin with insulin antibody precipitate. Biochem J. 1968;88:137-9. Wahlefeld AW. Triglycerides. Determination after enzymatic hydrolysis. In: Bergmeyer HV, ed. Methods of enzymatic analysis, 2nd ed. New York: Academic Press; 1974:1831-5. Abraham EC, Perry RE, Stallings M. Applications of affinity chromatography for separation and quantiiation of glycosylated hemoalobins. .J Lab Clin Med. 1983;102:187-97, AllaiiCC, Poon LS, Chan CSG, Richmond W, Fu PC. Enzymatic determination of total serum cholesterol. Clin Chem. 1975;20:4700 Have1 RJ, Eder HA, Bragdon JH. The distribution of ultracentrifugally separated lipoprnteins in human serum. J Clin Invest. 1955;34:1:145-53.

Downloaded from jcem.endojournals.org at Indonesia:Endo Jnls Sponsored on May 6, 2007

REAVEN 14. Swislocki ALM, Hoffman BB, Sheu WH-H, Chen Y-DI, Reaven GM. Effect of prazosin treatment on carbohydrate and lipoprotein metabolism in patients with hypertension. Am J Med. 1989;86:148. 15. Donner CC, Fraze E, Chen Y-DI, Reaven GM. Quantitation of insulin-stimulated glucose disposal in patients with non-insulindenendent diabetes mellitus. Diabetes. 1985;34:831-5. 16. Gr&mtield MS, Doberne L, Kraemer F, Tobey T, Reaven G. Assessment of insulin resistance with the insulin suppression test and the euglvcemic clamp. Diabetes. 1981:30:387-92. 17. Steele R. IInfluences of glucose loading and of injected insulin on hepatic glucose output. Ann NY Acad Sci. 1959;82:420-30. 18. Chen Y-DI, Swislocki ALM, Jeng C-Y, Juang J-H, Reaven GM. Effect of time on measurement of hepatic glucose production, J Clin Endocrinol Metab. 198&67:1084-8. 19. Vissers MCM, Jester SA, Fantone JC. Rapid purification of human peripheral blood monocytes by centrifugation through Ficoll-HyPaqie and Sepracell-MN. J Immunol Methods. 1988:110:203-7. 20. Moloney WC, McPherson K, Fliegelman L. E&erase activity in leukocytes demonstrated by the use of naphtol AS-D diformacetate substrata. J Histochem Cytochem. 1960;8:200-7. 21. Olefskv JM. Reaven GM. Insulin binding to monocvtes and total mono&clear leukocytes from normal anddiabetic patients. J Clin Endocrinol Metab. 197%43:226-a. 22. Trischitta V, Gullo D, Squatrito S, Pezzino V, Goldfine ID, Vigneri R. Insulin internalization into monocytes is decreased in patients with type II diabetes mellitus. J Clin Endocrinol Metab. 1986;62:522-8. 23. Henry RJ. Clinical chemistry: principles and technics. New York: Harper and Row; 1968;664-6. 24. Winer BJ. Statistical principles in experimental design. New York: McGraw-Hill; 1971;514-603. 26. Godfrey K. Statistics in practice. Comparing the means of several erou~s. N End J Med. 1985:313:1450-6. 26. Goldberg RI% Lipid disorders in diabetes. Diabetes Care. 1981;4:561-72. 27. Greenfield MS, Doberne L, Rosenthal M, Vreman JH, Reaven GM. Lipid metabolism in non-insulin-dependent diabetes mellitus: effect of glipizide therapy. Arch Intern Med. 1982;142:1498-500. 28. Prager R, Schemthaner G, Graf H. Effect of metformin on peripheral insulin sensitivity in non-insulin dependent diabetes mellitus. Diabetes Metab. 1985;12:346-50. 29. Groop L, Widen E, Franssila-Kallunki A, et al. Different effects of insulin and oral antidiabetic agents on glucose and energy metab-

El AL.

JCE&M*1992 Vol74.No5

30. 31. 32. 33. 34. 35. 36. 37.

38.

39. 40. 41. 42. 43.

olism in type 2 (non-insulin-dependent) diabetes mellitus. Diabetologia. 1989;32:599-605, Jackson RA, Hawa MI, Jaspan JB, et al. Mechanism of metformin action in non-insulin-dependent diabetes. Diahetes. 1987;36:63240. Ferrannini E, Barrett EJ, Bevilaqua S, DeFronzo RA. Effect of fattv acids on glucose production and utilization in man. J Clin Invest. 1983;72:1737-47. Reaven GM, Chang H, Ho H, Jeng C-Y, Hoffman BB. Lowering of plasma glucose in diabetic rats by antilipolvtic agents. Am J __ ._ Physiol. 1988;254:E23-30. Berchtold P, Bolli P, Arbenz U, Keiser G. Intestinale absorptionsstorung infolge metforminbehandlung. Diabetologia. 1969;5:405-12. Weiland 0, Mondon C, Reaven BM. Evidence for multiple causality in the development of diabetic hypertriglyceridemia. Diabetologia. 1980;18:335-40. Reaven GM, Greenfield MS. Diabetic hypertriglyceridemia. Evidence for three clinical syndromes. Diabetes. 1981:30366-75. Miller GJ. Miller NE. Plasma-high-densitv-lipoprotein concentration and development of ischaemic hea-rt disease. Lancet. 1975;1:16-9. Zavaroni I, DallAglio E, Alpi 0, et al. Evidence for an independent relatonship between plasma insulin concentration and high density lipoprotein cholesterol and triglyceride. Artheroeclerosis. 1985;55:259-66. Ducimetiere P, Eschwege E, Papoz L, Richard JL, Claude JR, Rosselin G. Relationships of plasma insulin levels to the incidence of myocardial infarction and coronary heart disease mortality in the middle-aged population. Diabetologia. 1980;19:205-10. Pyorala K. Relationship of glucose tolerance and plasma insulin to the incidence of coronary heart disease: results from two population studies in Finland. Diabetes Care. 1979;2:131-41. Pyorala K. Hyperinsulinaemia as predictor of atherosclerotic vascular disease: epidemiological evidence. Diabetes Metab. 1991;17:87-92. Gustafson A, Bjbrntorp P, Fahlen M. Metformin administration in hyperlipidemic states. Acta Med Stand. 1971;190:491-3. Sirtori CR, Tremoli E, Sirtori M, Conti F, Paoletti R. Treatment of hypertriglyceridemia with metformin. Atherosclerosis. 1977;26:583-92. Vague PH, Juhan-Vague I, Alessi MC, Badier C, Valadier J. Metformin decreases the high plasminogen activator inhibition capacity and plasma insulin and triglyceride levels in non-diabetic obese subjects. Thromb Haemost. 1987;57:326-8.

Downloaded from jcem.endojournals.org at Indonesia:Endo Jnls Sponsored on May 6, 2007