Dm

ORIGINAL ARTICLES

The Short Insulin Tolerance Test for Determination of Insulin Sensitivity: A Comparison with the Euglycaemic Clamp A. Akinmokun,

P. 1. Selby, K. Ramaiya and K. C . M. M. Alberti

Department of Medicine, University of Newcastle upon Tyne, UK

The glucose clamp technique is currently regarded as the standard test for measuring insulin sensitivity against which other methods are compared but is unsuitable for routine screening of patients outside a hospital base. There i s thus a need for a simpler test to measure insulin sensitivity. We have therefore compared the glucose disappearance rate KITTin the first 15 min of the insulin tolerance test (ITT) with the M and MI values derived from the standard euglycaemic clamp in nine normal subjects and eight subjects with Type 2 (non-insulin dependent) diabetes mellitus and coexisting obesity. All subjects underwent the I T 1 and euglycaemic clamp in random order. Nine subjects later had a repeat ITT to determine the reproducibility of the test. In the ITT, 0.1 U kg-' body weight, human Actrapid insulin was given as an IV bolus and simultaneous arterialized and venous blood samples were obtained every minute for 15 min. The first order rate constant for the disappearance of glucose KbTT over the period 3-15 min was taken as a measure of insulin sensitivity. The euglycaemic clamp was performed with an insulin infusion of 50 mU kg-' h-' for 120 min and a variable rate glucose infusion to maintain blood glucose concentration at 0.5 mmol I-' below fasting level to minimize the effect of endogenous insulin secretion. The ratio of the mean rate of glucose infused (M, pmol kg-' min-') t o the plasma insulin over the last 30 min of the clamp was taken as a measure of tissue sensitivity to insulin (MI)assuming endogenous glucose output was suppressed. A close correlation was found between the glucose disappearance rate KITT and MI/ values derived from the ITT and euglycaemic clamp, respectively, in both the normal ( r = 0.86, p < 0.01) and the diabetic subjects ( r = 0.81; p < 0.01) when arterialized blood was used but correlations were not significant with the venous blood sampling. The ITT was reproducible with a CV of 6%. The results suggest that the I l l is a suitable method of assessing insulin sensitivity and will be particularly useful in epidemiological studies, although the requirement for arterialized blood adds a measure of complexity. KEY WORDS

Insulin tolerance test diabetes

Introduction

Insulin sensitivity

Euglycaemic clamp

Type 2

The ITT has been used to assess insulin sensitivity, but has two major disadvantages. First, insulininduced hypoglycaemia is unpleasant and potentially dangerous. Second, counter-regulatory hormones are secreted in response to hypoglycaemia blunting the fall in plasma glucose.13-16 However, following intravenous insulin injection, blood glucose reaches its nadir by 20 min and counter-regulatory hormone response occurs only after this It is reasonable to assume, therefore, that the rate of glucose disappearance Klrr in the first 15-20 min following intravenous insulin injection represents insuIi n-mediated glucose uptake by the tissues. Recently, Bonora et a1.19 have suggested that the ITT is safe and shows close correlation with the euglycaemic clamp when used as a measure of insulin sensitivity. We have cowpared the values for glucose disappearance rate K l T T with the M values (glucose infused per unit time to maintain euglycaemia) derived from the standard euglycaemic clamp in normal and diabetic subjects. 13-16

Insulin resistance has been shown by many studies as an important entity contributing to glucose intolerance in It has now been implicated in the pathogenesis of glucose intolerance observed i n human diseases other than d i a b e t e ~ . ~It- is ~ said to exist when a given amount of insulin produces a subnormal biological The euglycaemic hyperinsulinaemic clamp techniques is currently regarded as the standard for measuring insulin sensitivity in vivo, against which other methods are compared,9-12 but its use is presently restricted to the laboratory because it requires sophisticated equipment and highly trained personnel, and is expensive to perform. There is thus a need for a simple test to measure insulin sensitivity outside the laboratory. Correspondence to: Professor K. G. M. M. Alberti, Department of Medicine, University of Newcastle upon Tyne, The Medical School, Framlington Place, Newcastle upon Tyne, NE2 4 H H , UK.

4 32

0742-307 119210504 3 2-06$05.00

@ 1992 by John Wiley & Sons, Ltd.

Accepted 14 January 1992 DIABETIC MEDICINE, 1992; 9: 432-437

DTT7

ORIGINAL ARTICLES

Subjects and Methods Subjects Nine healthy subjects (all male) aged 30-49 years with body mass indices between 20 and 30 kg m-* and eight obese subjects with Type 2 diabetes aged 41-65 years with body mass indices between 26 and 42 kg m-2 (Table 1) were recruited. They were all instructed to remain on their usual diet for at least 3 days prior to study and to fast for 10-1 4 h before study. None of the normal subjects had a family history of diabetes mellitus, none was known to have diabetes or impaired glucose tolerance, and no subject was on any medication capable of affecting glucose tolerance. Each subject was studied on two separate occasions with an interval of at least 1 week between tests. The order of the test was randomized. In nine subjects a second ITT was performed at a later stage to study reproducibility of the test. For all comparisons results of the first test have been used. Permission for the study was obtained from Newcastle Joint Ethics Committee and informed written consent was obtained from all subjects.

Protocol insulin Tolerance Test (ITT) The ITT consisted of a bolus intravenous injection of Human Actrapid insulin (Novo Laboratories, Basingstoke, UK) 0.1 U kg-' body weight. Blood samples were obtained every minute for 15 min, simultaneously from an arterialized hand vein and contralateral antecubital vein for blood glucose determination. Blood samples Table 1 . Clinical and metabolic characteristics of subjects Normal

Age Sex (M:F) Body mass index (kg

m-2) Fasting blood glucose (rnmol 1-l) HbAl (%) Serum C-peptide (nmol

*7 910 24.1 * 1.6 36

4.5 i 0.4 6.1 0.44

1-11

* 0.7 -t

0.1

*

Type 2 diabetes

p value

* 9 0.03 513 32.9 * 4.6 0.01 52

6.8 i 1.8

0.05

7.8 ? 1.5 1.3 -+ 0.6

0.04 0.02

Fasting insulin (mu I-') 6.2 3 . 0 29.2 -t 15.3 0.001 Glucose uptake ( M : 29.9 5 8.9 14.5 2.9 0.004 yrnol kg-' min-') KITT (mmol 1-l min-') 0.042 5 0.01 0.020 0.01 0.001 Reference range: Fasting blood glucose 4.2-5.5 (mrnol I-') HbAl (%) 5.0-7.5 Fasting C-peptide (nmol 0.1 5-0.52

* *

1-11

Fasting insulin

(mu I-')

Results are expressed as (mean

2.8-13.5

* SD).

THE SHORT INSULIN TOLERANCE TEST FOR INSULIN SENSITIVITY

were taken basally and at 15 min for cortisol, glucagon, growth hormone, and catecholamines. Subjects were fed at 15 min. Insulin sensitivity was indicated by the first order rate constant for disappearance rate of glucose Klrr estimated from the slope of the regression line of the logarithm of blood glucose against time during the first 3-15 min using the statworks/Cricket graph programmes on an Apple Macintosh computer. The Euglycaemic-Hyperinsulinaemic Clamp The euglycaemic clamp was performed using the technique of DeFronzo et a/.8 On the morning of the test, two teflon cannulae were inserted under local anaesthesia; one placed retrogradely in a hand vein was used for sampling. The second cannula was inserted in a contralateral antecubital vein and was used for all infusions. Venous blood was arterialized by maintaining the hand at approximately 55 "C in a thermoregulated plexiglass box.20 A period of 30 min following cannulation was allowed for baseline stabilization. Thereafter, subjects were infused for 120 min with Human Actrapid diluted in polygeline (Haemaccel, Hoechst UK, Hounslow, UK) at 0.05 U kg-' h-' using a Harvard pump (Harvard Apparatus, South Nattick, MA, USA). The infusion was primed with 0.01 U kg-' insulin over 10 min. During the insulin infusion, blood glucose was maintained at 0.5 mmol I-' below the fasting level (to minimize the influence of endogenous insulin secretion) by a variable infusion of 20 % glucose to which 20 mmol I - l of potassium chloride was added. The coefficient of variation (CV) of blood glucose during the last 60 min was less than 3 % in all cases. The rate of glucose infusion was adjusted according to blood glucose levels measured every 5 min. Blood samples were taken basally and during the last 30 min of the clamp for serum insulin measurement. The average rate of glucose infused during the last 30 min of each clamp was used for calculating insulin sensitivity ( M ) assuming endogenous glucose production was fully suppressed. The measure of tissue sensitivity to insulin was derived from the MI/ ratio. Metabolic clearance rate (MCR) of glucose was also calculated.

Analytical Methods Blood glucose was measured using the Yellow Springs glucose analyser (Yellow Springs Instruments, Yellow Springs, OH, USA); with intra- and interassay CVs of 1 % and 3 %, respectively. Serum insulin was measured by radioimmunoassay (interassay CV of 7.5 %) using the method of Soeldner and Slone.21 Serum cortisol, growth hormone, and glucagon were measured by r a d i o i m m u n ~ a s s a y with ~~-~~ interassay CVs of 6.4 %, 11 %, and 9.5 %, respectively. Plasma catecholamines were measured by high performance liquid ~hromatography~~ with interassay CV of 9.2 % for noradrenaline and 7.5 % for adrenaline. 433

Dm

ORIGINAL ARTICLES Statistical Analysis Results were analysed using Student's t-test (for paired and unpaired data as appropriate) and by Pearson's correlation test: p < 0.05 was taken as significant. Results are expressed as mean f SD unless otherwise stated.

Results Following intravenous injection of insulin, blood glucose level started to fall after 3 min. The glucose disappearance rate K l T T was therefore estimated from the slope of glucose disappearance from 3-1 5 min. In normal subjects, the mean glucose disappearance rate was 0.042 f 0.016 mmol I-' min-' (range 0.020-0.070) using arterialized samples and 0.050 k 0.018 mmol I-' min-' (range 0.026-0.085) using venous samples (p < 0.05). In the obese subjects, mean KIT, was 0.022 f 0.011 mmol I-' min-' (range 0.01-0.041) using arterial samples and 0.024 f 0.01 1 (range 0.01 2-0.01 7) using venous samples (NS) (Figure 1). K l T T values were reproducible in nine subjects (normal and obese subjects selected randomly) who had a repeat insulin tolerance test performed: 0.042 f 0.01 2 and 0.051 f 0.014 (mean f SE) with a CV of 6 f 2 %. In the euglycaemic clamp, plasma insulin was raised from the basal level of 6.2 ? 3.0 mU I-' to and 29.2 f 15.3 mU I-' to 66.6 f 26.7 mU I-' 97.6 ? 22.8 m U I-' in the normal and diabetic subjects,

respectively. The amount of glucose required to maintain euglycaemia during the last 30 min of the clamp (MI ranged between 13.6 and 43.4 kmol kg-' min-' (mean 29.9 f 8.9) in the normal subjects and between 9.9 and 19.4 Fmol kg-' min-' (mean 14.5 f 2.9) in the obese diabetic subjects (p < 0.001). KlTT, M values, and MI/ ratio were, as expected, significantly lower in the obese diabetics compared to normal subjects (p < 0.004, 0.001, and 0.001, respectively). There was a highly significant correlation between the K l T T derived from the slope of the arterial glucose disappearance rate and the MI/ ratio derived from the euglycaemic clamp in both the normal ( r = 0.86, p < 0.003; Figure 2(a)) and the obese diabetic subjects ( r = 0.81, p < 0.01 ; Figure 2(b)). Using venous blood samples, the correlations between KITT and MI/ were non-significant in both the normals and the diabetic subjects ( r = 0.59, p = 0.09 vs r = 0.56, p = 0.06, respectively). There was no significant change in the levels of the counter-regulatory hormones during the first 15 min following the injection of insulin (Table 2). No subject developed symptoms of hypoglycaemia or a whole blood glucose level < 2 mmol I-' before 15 min. When all subjects were analysed together there were significant correlations between KITT and the MI/ values ( r = 0.80; Figure 3(a)) and the M value ( r = 0.72; Figure 3(b)) but the correlation against MCR glucose failed to reach significance ( r = 0.53; Figure 3(c)).

Discussion

f 0.7

0

E E

v

(3

m 0 0

0.5

Time t m i n l Figure 1. Glucose disappearance rate in normal and obese diabetic subjects (showing A-V difference). Normal (arterial) --El--, normal (venous)-C,obese (arterial)4-,obese (venous)

4 434

Carbohydrate intolerance usually occurs as a result of both relative or absolute insulin deficiency and insulin resistance. In the last two decades, diminished sensitivity to insulin has become established as an important pathogenic mechanism in the aetiology and perpetuation of glucose intolerance. Indeed, most clinical conditions associated with glucose intolerance are characterized by variable degree of insulin resistan~e.*-~ Currently there is a variety of therapeutic interventions which aim at improving the action of insulin at the level of target organs. This has led to a search for simple tests of insulin sensitivity in man in vivo. The response of blood glucose to exogenously administered insulin, the insulin tolerance test (ITT), has long been used to estimate insulin sensitivity. The fall in plasma glucose concentrations in the test is, however, dependent on both the inhibition of endogenous glucose production (principally from the liver) and the stimulation of glucose uptake by tissues, principally muscle; the faster the decline in plasma glucose concentration, the more sensitive the individual to insulin. Usually the dose of insulin used (0.1 U kg-') is supraphysiological, causing suppression of hepatic glucose production. Hence, the rate of glucose disappearance is a measure of glucose uptake by the tissues. However, hypoglycaemia often results, and this will trigger a pronounced counterA. AKINMOKUN ET Al.

Dm

ORIGINAL ARTICLES Table 2. Counter-regulatory hormone level (basal and at 15 min) during the insulin tolerance test

c

Basal Adrenaline (nmol I-') Noradrenaline (nmol 1-1)

Glucagon (pg m l -') Cortisol (nmol 1-l) Growth hormone ( m u

15 rnin

* 0.01 0.30 -C 0.05 * 0.1 1 2.14 * 0.15 53 * 4 48 * 4 325 * 20 299 2 1 7 0.6 * 0.1 0.6 * 0.2

0.21 1.84

p value 0.11 0.11 0.18 0.09 0.66

1-11

Results expressed as (mean

0.0 0.01

0.05

0.03 KITT :mmol I

-1

min

-1

1.2-

0.8

0.4-

0.01

0.03

K~~~

0.05

0.07

: mmol I - 'min-

Figure 2. Correlation of insulin sensitivity as measured by euglycaemic hyperinsulinaemic clamp with that measured by insulin tolerance test. Horizontal bars show confidence intervals of individual KiTT slopes. (a) Regression of M against KlTT in the obese diabetic subjects: MI/ = 1.26 KITT+l.83; r = 0.81, p < 0.01. (b) Regression of M / / against Klrr in normal subjects: MI/ = 3.04 KITT-0.03 r = 0.86, p < 0.003

regulatory response which in turn impairs the decline in plasma glucose concentration, rendering interpretation of the latter difficult. Following intravenous insulin injection blood glucose reaches its nadir by about 20 min and counter-regulation occurs only after this time. The THE SHORT INSULIN TOLERANCE TEST FOR INSULIN SENSITIVITY

?

SE).

glucose disappearance rate in the first 15 min should therefore measure only the action of the injected insulin. The euglycaemic clamp was developed to overcome the above problems and is now regarded as the standard test of insulin sensitivity. Its use is restricted to the laboratory because of the need for feedback control of glucose levels. A simple test for measuring insulin sensitivity for widespread clinical use is therefore desirable. One possible alternative is the glucose insulin infusion testz6which correlates well with the euglycaemic clamp and merely requires measurement of four blood glucose levels at steady state. It takes 3 h to perform, however, and is thus costly in time. The ITT provides a short useful alternative providing it is terminated with food after 15 min. Using the ITT we, like Bonora et a/., have found a good correlation between the glucose disappearance rate KITT during the first 15 min of the insulin tolerance test and the MI/ ratio and the M values derived from the standard euglycaemic clamp. The purpose of the study was not to compare normal and diabetic subjects so much as to study subjects with a wide range of insulin sensitivity. It is worth noting that overall there was a seven-fold range in sensitivity with regard to K,, similar to the range shown for clamp measurements. Measuring arterio-venous (A-V) differences provided additional information. There were wide variations in A-V differences in blood glucose among individuals, but in normal subjects the arterial blood glucose was higher or at least equal to the venous glucose in the basal state and this difference became greater as blood glucose fell. In the obese diabetic patients, however, there was no difference between the arterial and venous blood glucose concentrations both basally and in the first 15 min following insulin injection. Indeed, negative A-V differences were observed in some of the most resistant patients as blood glucose fell. The reason for this is not clear, but certainly reflects the overall failure of muscle to remove glucose from the blood in these subjects. The negative A-V difference, although physiologically unlikely, was commented upon as long ago as 1938 by Griffith~.~~ Our study confirms the observation of Bonora et a/., in which he showed the possible efficacy of the ITT as 435

ORIGINAL ARTICLES

0.00

0.02

0' 0.00

0.02

0.04

I

0.08

0.06

I

I

0.04

0.06

I

I

0.08

la4 c

.-c

E

/

m

/

/

n no

Acknowledgements

0

E

2 a 2

0.04

We are grateful to L. Ashworth, L. Brigham, M. Brown, and D. B. Cook for assistance and to the British Diabetic Association and Hoechst (UK) Ltd for financial support.

0.02

0.00' 0.oo

1

I

I

I

0.04

0.02 K

ITT

I

I

0.ow

I

J

0.08

(rnrnol I - ' min-'1

Figure 3. Correlation of insulin sensitivity as measured by euglycaemic hyperinsulinaemic clamp with that measured by insulin tolerance test in normal and diabetic subjects using various indices of insulin sensitivity. (a) MI/ ratio: MI/ = 2.88 x -0.028; r = 0.80. (b) M value: M = 521.5 x +5.49; r = 0.72 (c) Metabolic clearance rate (MCR) = 1.24 K,,+0.0084; r = 0.53. Horizontal bars show confidence intervals of individual K ,, slopes: -0- normal subjects; 4- obese Type 2 diabetes

a measure of insulin sensitivity. The measurement of AV difference in our study, however, provided additional evidence to support this claim. It should be noted that the short ITT only allows relative ranking of individuals with respect to insulin sensitivity. The test cannot determine which tissue (liver or muscle) is responsible

436

for the defect in insulin action and the data are difficult to convert into physiologically relevant clearance states. It i s worth emphasizing that the correlation with the euglycaemic clamp results was better when arterialized blood samples were used to calculate Kin. Correlation was insignificant using venous samples, suggesting that a reliable index of insulin sensitivity of ITT is only obtained when arterial blood glucose rather than venous is used to calculate the glucose disappearance rate. This serves to emphasize the importance of considering the nature of blood samples used when interpreting glucose kinetics. Interestingly, correlations were poorest with the metabolic clearance rate for glucose which should improve comparisons when starting glucose values are different. This did not occur but glucose-induced changes in glucose metabolism may confound interpretation. Because this test is simple, rapid to perform, and reasonably accurate it will be a useful tool for screening for insulin resistance especially in field studies where it is not practicable to use the standard but complex euglycaemic clamp technique. Two workers can comfortably do six such tests in a 2 h period with no need for a glucose analyser or computer at the place of testing. Indeed blood glucose could be measured every 2 min with little loss of accuracy rather simplifying the procedure. By contrast the same two workers would be hard pressed to carry out two clamps and a longer time would be required. The disadvantage is that precise quantitation of glucose uptake in response to a physiological increment in insulin is not possible. In the end the choice of test will depend on the objectives sought.

References Reaven GM. Role of insulin resistance in human disease. Diabetes 1988; 37: 1595-1 607. 2. Olefsky JM. Insulin receptor: Its role in insulin resistance of obesity and diabetes. Diabetes 1976; 25: 1 154-1 165. 3. DeFronzo RA. Glucose intolerance and aging, evidence of tissue insensitivity to insulin. ) C/in lnvest 1979; 28: 1.

1095-1 101. Bar RS, Levis WR, Rechler MM, Harrison LC, Siebert C, Podskalny J, et a/. Extreme insulin resistance in ataxia telangiectasia. N €ng/ ) Med 1978; 298: 1164-1 171. 5. Kahn RC, Flier JS, Barr RS, Archer JA, Gordon P, Malcolm MM. Syndrome of insulin resistance and acanthosis nigricans. N Engl) Med 1976; 294: 739-745. 6. Himsworth HP, Kerr RB. Insulin-sensitive and insulininsensitive types of diabetes mellitus. Clin Sci 1939; 4: 119-1 52. 7. Kahn CR. Insulin resistance, insulin insensitivity and insulin unresponsiveness:a necessary distinction. Metabolism 1978; 27 (suppl 2): 1893-1 902.

4.

A. A K I N M O K U N ET AL.

Dm 8. 9.

10. 11.

12.

13. 14.

15. 16. 17.

DeFronzo RA, Tobin ID, Andres R. The glucose clamp technique: a method of quantifying insulin secretion and resistance. Am / Physiol 1979; 237: E214-E223. Greenfield MS, Doberne L, Kraemer F, Tobey T, Reaven G. Assessment of insulin resistance with the insulin suppression test and the euglycaemic clamp. Diabetes 1981; 30: 387-392. Bergman RN, lder YZ, Bowden CR, Cobelli C. Quantitative estimation of insulin sensitivity. Am / Physiol 1979; 236: E667-E677. Donner CC, Fraze E, Chen YDL, Hollenbeck CB, Foley JE, Reaven GM. Presentation of a new method for specific measurements of in vivo insulin-stimulated glucose disposal in humans. Comparison of this new approach with insulin clamp and minimal model techniques. / Clin Endocrinol Metab 1985; 60: 723-726. Ponchner M, Heine RJ, Pernet A, Hanning I, Francis AJ, Cook, D, et a/. A comparison of the artificial pancreas (glucose controlled infusion system) and a manual technique for assessing insulin sensitivity during euglycaemic clamping. Diabetologia 1984; 26: 420-425. Alford FP, Martin FIR, Pearson MI. The significance and interpretation of mildly abnormal glucose tolerance. Diabetologia 1971; 7: 173-1 80. Beck-Nielsen H, Pedersen 0. Insulin receptors on monocytes of young healthy persons correlated with glucose intolerance and insulin sensitivity. Diabetologia 1978; 14: 159-163. Harrison LC, King-Roach AP. Insulin sensitivity of adipose tissues in vitro and response of exogenous insulin in obese human subjects. Metabolism 1974; 25: 1095-1 101. Yeung RTT, Wang CCL. A study of carbohydrate metabolism in postnecrotic cirrhosis of liver. Cut 1974; 15: 907-91 2. Gerich J, Cryer P, Rizza R. Hormonal mechanisms in acute glucose counterregulation: the relative role of glucagon, epinephrine, norepinephrine, growth hormone and cortisol. Metabolism 1980; 29 (suppl 1): 1164-1 175.

THE SHORT INSULIN TOLERANCE TEST FOR INSULIN SENSITIVITY

ORIGINAL ARTICLES 18.

19.

20.

21. 22. 23. 24. 25.

26.

27.

Rizza RA, Cryer PE, Gerich JE. The role of glucagon, catecholamines and growth hormone in human glucose counterregulation. Effects of somatostatin and combined alpha and B-adrenergic blockade on plasma glucose recovery and glucose flux rates after insulin-induced hypoglycaemia. / Clin lnvest 1979; 64: 62-71. Bonora E, Moghetti P, ZanCanaro C, Cigotini M, Querena M, Cacciatori V, et a/. Estimates of in vivo insulin action in man: Comparison of insulin tolerance tests with euglycaemic clamp and hyperinsulinaemic clamp studies. / Clin Endocrinol Metab 1989; 8 : 374-378. Abumrad NN, Rabin ND, Diamond MP, Lacy WW. Use of heated superficial hand vein as an alternative site for the measurement of amino acid concentrations and for the study of glucose kinetics in man. Metabolism 1981; 30: 936-940. Soeldner J, Slone D. Critical variable in the radioimmunoassay of serum insulin using double antibody technic. Diabetes 1965; 14: 771-779. Brack P, Eldred EW, Woiszillo JE, Doran M, Schomaker JH. Direct solid-phase insulin radioimmunoassay of serum cortisol. Clin Chem 1978; 24: 1595-1 598. Boden G, Soeldner IS. A sensitive antibody radioimmunoassay for HGH. Levels of serum HGH following rapid tolbutamide infusion. Diabetologia 1967; 3: 41 3-421. Orskov H, Thomsen HG, Yde H. Wick chromatography for rapid and reliable immunoassay of insulin, glucagon and growth hormone. Nature 1968; 219: 193-195. Smedes F, Kraak JK, Poppe H. Simple and fast solvent extraction system for selective and quantitative isolation of adrenaline, noradrenaline and dopamine from plasma and urine. / Chromatogr 1982; 231: 25-29. Harano Y, Hidaka H, Takatsuki K, Ohgaka S, Haneda M, Motoi S, et a/. Glucose, insulin and somatostatin infusion for the determination of insulin sensitivity in vivo. Metabolism 1978; 27: 1149-1 452. Griffiths WJ. Insulin resistance and the arterio-venous blood sugar difference. Clin Sci 1938; 3: 91-104.

437