- Email: [email protected]
Validation of a practical in vivo insulin dose-response curve in man
Validation of a Practical In Vivo Insulin Dose-Response Curve in Man J. Proietto,
M. Harewood,
P. Aitken,
A. Nankervis,
G. Caruso, and F. Alford
To adequately investigate the state of insulin resistance, an insulin dose-response curve should be constructed so that insulin sensitivity (right shift of dose-response curve) and insulin responsiveness (maximal response) can be determined. This paper describes and validates in man a practical in vivo insulin dose-response curve technique, using a modification of the euglycaemic clamp described by De Fronro et al. Insulin action at steady state was expressed as metabolic clearance rate of glucose (MC&) rather than overall rate of glucose disappearance (M or Rd). MCR, was chosen because at plasma insulin concentrations >25~U/ml it was shown (n = 5) not to be altered by changes in blood glucose concentration (MCR, 379 + 23 and 408 + 19 ml/mz/min: at plasma glucose concentrations 5.4 + 0.3 and 10.2 f 0.7 respectively). whereas Rd was critically dependent on the prevailing blood glucose concentration (Rd 2007 +- 128 and 4124 + 219~mol m*/min respectively). MCR, was demonstrated to be stable over a 8 hr period (n = 7) and to be reproducible In = 4). Insulin dose-response curves (MCR, Versus insulin concentration) were performed on two obese and seven normal weight individuals. The insulin dose-response curves were linearized, allowing accurate prediction of the maximal MCR,, as compared to the experimentally determined maximal response (r = 0.953 p < 0.01). The use of this transformation obviates the need to employ very high insulin infusion rates to determine the maximal insulin response. In conclusion, the technique permits, in a single 6 hr study, a precise insulin dose-response curve to be constructed for accurately determining insulin sensitivity and responsiveness.
I
NSULIN RESISTANCE, defined as a state in which normal insulin levels elicit a less than normal biological response,’ is thought to be responsible for the glucose intolerance of many disorders including obesity,* acromegaly,3 cirrhosis,4 uraemia,’ and noninsulin dependent diabetes.6 The resistance can be of three types. In the first type, greater than normal hormone levels are needed to elicit the maximal response; in the second, the maximal response cannot be achieved despite the presence of supraphysiological or pharmacological insulin levels. Thirdly, a combination of both types is possible. Kahn’ has suggested that the first type be called insulin insensitivity and the second, insulin unresponsiveness. The distinction between these types of resistance is important because each may have a different underlying mechanism. For example, the spare receptor theory suggests that moderately reduced receptor binding could result in insensitivity. A post receptor defect or very severely reduced receptor binding may cause unresponsiveness. To distinguish between insulin insensitivity, unre-
From the Endocrine Unit and Department of Medicine, University of Melbourne, St. Vincent’s Hospital, Fitzroy, Victoria, Australia. Received for publication January 9. 1981. J.P. is a recipient of a National Health & Medical Research Council Medical Postgraduate Research Scholarship. This work was performed in partial fulfillment of the thesis requirement. Presented in part to the Annual Scientific Meeting of the Ausiraban Diabetes Society, Armidale. New South Wales. August 1980. Address requests for reprints to: Dr. J. Proietto, Endocrine Unit, St. Vincent’s Hospital, Victoria Parade, Fitzroy, VIC. 3065. AUStralia. 0 1982 by Grune & Stratton. Inc. 0026-0495/82/3104-0007$01.00/0
354
sponsiveness or a combination of the two, it is essential that adequate insulin dose response curves be obtained. The development of the euglycaemic glucose clamp technique by Andres, R. et al7 as described by De Fronzo, R. et al* made it possible to construct such curves. Initial reports of insulin dose-response curves using this technique required several studies over two to three days,’ although recently a one day insulin dose-response curve has been described.” The aim of this study was to devise and validate in man a practical in vivo insulin dose-response curve technique using the metabolic clearance rate of glucose (MCRo) as the measured parameter of insulin action. The method proved simple to perform without the use of complex equipment. A mathematical approach to the analysis of insulin dose-response curves is also presented which allows the derivation of the maximal response without the need for very high insulin infusions. MATERIALS
AND
METHODS
Subjects All subjects were healthy, ambulant volunteers on no medication and with no family history of diabetes. Sixteen normal weight subjects, 7 males and 9 females aged between 21 and 39 yr took part in the three validation experiments. Nine additional subjects, 7 males and 2 females aged between 29 and 50 yr had insulin dose-response curves performed (Table 1). Seven of these subjects had ponderal indices in the normal range (PI 18-30) and two were obese (Table 1). Written informed consent was obtained from all subjects and the protocols were approved by the Ethics and Research Committee of St. Vincent’s Hospital, Melbourne. All subjects were studied after an overnight fast and at least a 30 min period of recumbency. All solutions were infused, using 2 Gilson Minipuls II pumps (Gilson, Villiers le Bel. France) via a 20 cm catheter placed in an antecubital vein. Blood samples were collected at appropriate intervals from an indwelling butterfly needle placed
Metabolism, Vol. 3 1, No. 4 (April), 1982
INSULIN DOSE RESPONSE CURVE IN MAN
355
Table 1. Clinical Data and Results in the 9 Subjects who had Insulin Dose-Response sex Subject
Age (Yrs.)
17
Maximal MCR,
Fasting III&X
M
WJ/mll
Method
I#
Method
IRI Levels at* MCR,,,,
X Max MCR,
(ml/m’/min)
IRI
P0ildaW.l
Curves
IRI Levels at* Experimental’
II#
ml/m’/man
(ED.,) W/ml)
(ED,.,) W/ml)
20.9
8
686
641
650
35
35
21.3
8
639
624
630
43
44
23.5
7
666
764
41
43
24.2
6
597
613
600
46
53
24.8
3
602
525
-
49
66
26.3
10
576
647
54
68
26.5
6
464
459
43
76
35.9
16
557
593
210
360
43.9
27
492
509
165
360
29 18
M 31
19
M 35
20
M 39
21
M 43
22
F 41
23
M
440
36 24
M 50
25
F 34
1
1 #
Method I: Vs [IRI] ~ [IRI],, MCR,
-
IIRII
’ Method‘I: MCR,
Vs [IRI]
*Data determined from raw dose-response curve data contained in Fig. 3. in a contralateral arm vein. To overcome the problem of hormones adsorbing to the IV apparatus, insulin was placed in normal saline and pre-run for one hour prior to use. Somatostatin (Bachem, Torrance, USA) was infused in a solution containing 0.2% human serum albumin (Beringwerke, Germany) and 1000 KIU Trasylol (Aprotinin, Bayer Pharmaceuticals, Germany). Blood for hormone estimations was collected in 10% Trasylol, promptly centrifuged and plasma stored at ~ 20°C until the time of assay.
Oral Glucose Tolerance
3;
‘8
2 18
-.-.
---
--/
4 1
2e
Test
These were performed, using a 75 gram glucose subjects who had insulin dose-response curves.
Dose-Response
12
load, on the 9
Curve Technique
Nine subjects had euglycaemic clamps performed after the method of De Fronzo et al,* as illustrated in Fig. 1. When it became possible to estimate the approximate glucose infusion rate needed to maintain euglycaemia (usually after 60 min), the infusion rate of glucose was fixed (Fig. 1). On rare occasions, unexplained changes in blood glucose during the initial period of the euglycaemic clamp necessitated the clamp procedure being continued beyond the usual 60 min. In these cases, the glucose infusion rate was fixed by 90 min. Thirty minutes after fixing the glucose infusion rate, four blood samples were taken at 10 min intervals for estimation of plasma glucose and insulin. This process was then repeated at higher insulin infusion rates in order to construct three or more point dose-response curves. Three point curves were performed in a single day and four or more point curves over two sessions. At the end of each experiment three, 5 min samples of the fluid delivered by the glucose and insulin pumps were collected into weighed tubes. From these were calculated the exact rates of infusion of glucose and insulin. Insulin infusion rates ranged from 30mU/Kg/hr to 600mU/Kg/hr which resulted in plasma insulin levels of from 33 to 2190rU/ml. In view of our previous studies” hepatic glucose production (HGP) was assumed to be suppressed to insignificant values during all the euglycaemic clamps performed. This assumption could not be made in the two very obese individuals? and in them, glucose turnover was
765 432-
‘1 AA
.
1, .
.
.
.
.
.
.
.
.
.
AAAA ,
-10 0
30
60
90
120
T 1 ME (mins) Fig. 1. Plasma glucose, plasma insulin and insulin (shaded area) and glucose infusion rates during a representative euglycaemic clamp. The arrows indicate sampling times. Note that despite plasma insulin reaching a plateau level by 10 min. the glucose infusion rate needed to maintain euglycaemia continued to rise for 45 min.
356
estimated nique.‘*
PROIE-ITO ET AL.
at steady
state,
using
the H3-3
glucose
tracer
tech-
Validation Studies Validation of MCRG as a Measure of Insulin Action. It has been suggested from experiments in conscious dogs that MCRo is a measure of insulin action and is independent of the plasma glucose level.” To validate this in man, we studied 5 normal weight subjects during combined 5 hr cyclic Somatostatin (5rg/min) (Bachem Inc.. California, USA) glucose clamp procedures. An initial eugylycaemic clamp, as described above using insulin infusion rates of between 30 to 75mU/Kg/hour was followed by a hyperglycaemic c1amp.s The insulin infusion rate remained unchanged during both euglycaemic and hyperglycaemic study periods. During the last 30 min of each period, serial blood samples were taken for estimation of plasma glucose, insulin and C-peptide. Urine was collected over the last hour of each hyperglycaemic clamp for measurement of urinary glucose loss. Stability ofMCRG. To determine if steady state conditions and therefore MCRo are stable over a prolonged period, 7 subjects had euglycaemic clamps performed as described above using 75mU/ Kg/hr of insulin. Once steady state insulin and glucose infusion rates were achieved, they were fixed for the subsequent 5 hrs. Four blood samples were taken at IO min intervals between 90 and I20 min. 210 and 240 min and 330 and 360 min, for estimation of IRI and glucose levels and calculation of MCRo. Hepatic glucose production was again assumed to be suppressed to insignificant values” during these studies. Reproductibility of MCR,. To prove that this technique is reproducible, 4 normal weight subjects were studied I wk apart. On each occasion an euglycaemic clamp was performed using 75mU/ Kg/hr of insulin.
Analytical Procedures During the euglycaemic and hyperglycaemic clamps, blood glucose was estimated at the bedside, using a Boehringer reflomat (Boehringer Mannheim GmBh-Mannheim, W. Germany) and the glucose or Hypoglycaemie strips’” which give blood glucose readings in 2 min. Calculations for the clamps were rapidly performed with the help of a desk calculator using the formulae of De Fronzo et a1.s Plasma from samples taken at steady state, was analysed for glucose using a Union Carbide Centrifichem C400 analyser (Union Carbide Corp., New York, N.Y.) employing the hexokinase method. Only the latter hexokinase glucose estimations were used in subsequent calculations of MCRc. Plasma immunoreactive insulin (IRI) was estimated by radioimmunoassay employing dextran charcoal separation of the bound and free factions.” C-peptide was measured using the Novo C-peptide assay kit, employing synthetic human C-peptide and guinea pig anti human C-peptide antiserum.16 The relevant plasma samples from the tracer studies were processed as previously described.”
Calculations As stated above, during these studies hepatic glucose production is assumed in the non-obese normal subjects to be insignificant.” Therefore the rate of appearance of glucose (Ra) = the rate of infusion of glucose. In the obese individuals hepatic contribution to total Ra was measured at steady state using H’-3 glucose. When plasma glucose is stable (defined as a coefficient of variation of plasma glucose concentration at plateau of
plotting MCRo against the steady state plasma insulin levels. The curvilinear insulin dose-response curves were linearised by two transformations (i) I/MCR Vs I/[IRI]-[IRI] (basal); and (ii) [IRI]/MCR, Vs [IRI].” The straight lines so obtained by linear regression analyses were used to derive the maximal response. Linear regression analysis, paired and non-paired t tests were performed by standard statistical techniques.
RESULTS
Fig. 1 shows plasma glucose, plasma insulin, insulin infusion and glucose infusion rates, during a representative euglycaemic clamp. Glucose concentration remained at fasting levels throughout the 2 hr period. Plasma IRI rose to peak values within 5 min of the bolus insulin infusion and reached its steady state value by 10 min. It is important to note that despite the early rise of insulin to the plateau level, the amount of IV glucose needed to maintain euglycaemia rose steadily over the initial 45 min and only commenced to oscillate about its mean after 45 min. Similar patterns of glucose and insulin dynamics were observed in all the glucose clamps. Validation of MCR, as a Measure of Insulin Action Table 2 shows plasma glucose, IRI, C-peptide, Rd and MCR, values in the five subjects who underwent the Somatostatin/glucose clamp experiments. During the hyperglycaemic period, plasma insulin remained constant and C-peptide did not rise above basal values indicating adequate suppression of endogenous insulin production by Somatostatin during the hyperglycaemic clamp. In each case, Rd was shown to be critically dependent on the plasma glucose concentration with a rise in plasma glucose (5.4 * 0.3 to 10.2 f 0.7mmolL) resulting in a dramatic rise in Rd (2007 + 128 to 4124 2 219~mol/m2/min). In contrast, MCR, changed minimally (379 k 23 to 408 + 19ml/m’/min) (Table 2). Thus, a rise in blood glucose from 5.4 to 10.2mmolL resulted in a 105% rise in Rd whereas MCR, changed little (rose by 7%). Stability of MCRG Fig. 2 shows the mean + SEM of the plasma glucose and plasma IRI, and the mean and individual MCRo values measured at 2 hr intervals during the 6 hr euglycaemic clamp studies. Plasma glucose and insulin levels remained stable throughout the 6 hrs and there was no significant change in the mean MCRo’s (by paired t test p > 0.20). The mean coefficient of variation of the plasma glucose levels over the 6 hr was 9.9 k 0.9%, the variation for the plasma IRI values was 12.9 k 2.0%.
357
INSULIN DOSE RESPONSE CURVE IN MAN
Table 2. Effect of Changes of Plasma Glucose Level on the Rate of Disappearance (Rd) and Metabolic Clearance Rate (MCR,) of Glucose Plasma
Plasma GIUCOS?
1.
Fasting
2.
.05
1
5.2
26
,005
1,953
372
Permd
2
9.9
21
.005
3,868
387
5.0
3
Penod
1
5.0
27
.014
1,810
369
Period
2
8.0
29
,025
3,436
443
Fastmg Penod
4.
(ml/m’/min)
Period
Fasting
3.
5
MCR,
(~mollm’lmin~
(pmol/Ll
NJ/mli
4.0
Rd’
C-Peptide
IRI
(mmol/Ll
Subject
4.4
.09
7
.26
1
6.3
55
.04
1,904
302
Pernod 2
10.7
54
.18
4,311
402
Fastmg
4.5
5
.28
Period
1
5.8
30
.03
2,511
435
Penod
2
12.3
25
.03
4,282
351
4.7
9
.47
-
5.
Fasting Period
1
4.5
107
.08
1,858
418
Period
2
10.3
98
.13
4,722
457
Mean
c SEM
Fasting
4.7
+ 0.1
Period
1
5.4
k 0.3
49k
Period
2
10.2
+ 0.7
45
*When
calculating
Rd during
5.8
the hyperglycaemic
periods
k
k
1.0
.23
k .08
15
.03
k .Ol
2,007
+
128
379
f
23
14
.07
*
4,124
i
219
408
+
19
urinary
glucose
loss was
.03
subtracted
from
the total
rate of drsappearance
of glucose.
Reproducibility
Table 3 shows the results of the studies. Mean MCR,, IRI and glucose no significant difference between the (p > 0.50 for IRI levels; p > 0.95 for p > 0.95 for MCRG values). Dose- Response 80. T
60. 40. 20.
T,TT
r_____ .~~_-________~i-------zr~~ k I ! : !’
reproducibility levels showed paired studies glucose levels;
Curves
All seven normal weight subjects who had doseresponse curves performed had normal glucose tolerance.” Of the two obese subjects one had normal glucose tolerance (No. 25-Table 1) and the other had glucose intolerance (No. 24-Table 1). Fig. 3 is a graph of the individual dose-response curves (MCR, Vs plasma IRI level). The enclosed area in the upper panel represents the range of plasma insulin concentraTable 3. Reproducibility of Metabolic Clearance Rate of Glucose (MCR,) Metabolic Clearance Rate
T I ME
(mins)
Study II
13
400
471
Subject
14
447
458
Subject
15
361
336
Subject
16
428
380
Mean
MCR,
ml/m2/min
Fig. 2. Mean + SEM plasma glucose, plasma insulin and individual and mean metabolic clearance rates of glucose IMCR,) in seven subjects during a constant six hr insulin and glucose infusion. There were no significant changes in mean plasma glucose, plasma insulin or MCR,.
Study I
Subject
Mean
NJ/ml Mean
409
+
19
411
+ 32
IRI 71
+ 4
68
+ 2
Glucose
mmol/L
4.6
+ 0.1
4.6
+ 0.2
358
PROW-TO ET AL
I 10
1Q
lob0
tion and corresponding MCRo of 8 normal subjects in whom glucose turnover was measured previously, using H3-3 glucose.” The lower panel which depicts the normal range as a shaded area, shows the curves for the two obese individuals. These curves show that both obese individuals (Subjects 24 and 25-Table 1) have reduced insulin sensitivity but normal insulin responsiveness. When the dose-response curves presented in Fig. 3 were transformed (see Methods) a linear relationship between MCR, and plasma IRI was obtained for all individuals (group mean and range of the individual correlation coefficients for linearisation of data points were r = 0.948 + 0.014; range r = 0.887 to 0.999 for method I and r = .9983 f .0008; range r = .9936 to .9998 for method II). These transformations allow the derivation of the
Insulin dose-response curves Fig. 3. (MCR, Vs Plasma IRI) in the two obese and 7 normal weight individuals. In the upper panel the enclosed area represents the range of MC& and insulin levels obtained in a previous study in 8 normal individuals using H’-3 glucose to measure glucose turnover. The lower panel shows the normal range as a shaded area and the individual curves for the two obese individuals.
maximal MCRo. When a comparison is made between the experimentally determined and the derived maximal MCRo in the 4 subjects in whom maximal MCRo was reached experimentally, (Table 1, subjects 17, 18, 20, and 23), excellent agreement is observed (r = 0.953 p < 0.01). The accuracy of the predicted MCRo is further emphasized by the fact that the experimentally determined 600mU/Kg/hr point was deliberately omitted from the calculations for the purpose of this comparison. Table 1 gives the details of the 9 subjects studied, listed in order of increasing ponderal index, and shows the maximal MC&, the insulin concentration giving half the maximal response (ED,,,) and the insulin concentration giving an MCRo of 350ml/m2/min determined maximal (ED Mc~sSO)_The experimentally
INSULIN DOSE RESPONSE CURVE IN MAN
MCR,, the ED,, and the EDMcb_so data were read directly from the unsmoothed curves shown in Fig. 3. There was a significant correlation between ponderal index and insulin sensitivity for the normal weight group (subjects 17-23), expressed as ED,crt;jso ml/ m*/min (r = 0.930 p < 0.001) and as EDso (r = 0.713 p < 0.05). There was also a significant negative correlation between the ponderal index and the maximal MCR, (r = -0.820 p -c 0.01) for normal weight subjects. These relationships between ponderal index and insulin sensitivity and insulin responsiveness were strengthened when the obese subjects were included in the analysis. In the obese subjects the contribution of hepatic glucose production (HGP) to total Ra was measured at steady state using a primed constant infusion of H33 glucose. Subject 24 (Table 1) at a plasma insulin of 122pU/ml, had total suppression of HGP. In subject 25 the contribution of HGP to the total Ra during the lowest insulin infusion (plasma IRI 97rUml) was 9% (HGP 86~mol/m’/min; total Ra 920~mol/m2/min). If it is assumed that no further suppression occurred in this subject with the subsequent higher insulin infusion rates, the maximum percent contribution of HGP to total Ra in the subsequent 2 periods would be 5.4 and 3.4% respectively. The mean of the coefficients of variation of the steady state plasma glucose levels at plateau (n = 36) of all the dose-response curves was 7.0 + 0.6%. The mean coefficient of variation for the steady state plasma lR1 levels was 9.8 t 1.6%. DISCUSSION
This paper outlines and validates in man a practical method of constructing in vivo insulin dose-response curves using a modification of the euglycaemic clamp technique of De Fronzo et al.’ Two main modifications are described. The glucose infusion rate isJixed over the last 60 min of each 2 hr study and the data are presented as the metabolic clearance rate of glucose (MCR,), rather than the rate of disappearance of glucose (Rd or M). This modified technique gives a stable plateau of plasma glucose and MCR, over prolonged periods (6 hrs) (Fig. 2), and is shown to be highly reproducible (Table 2). There were several advantages from fixing the rate of the glucose infusion over the last 60 min and of sampling only over the last 30 mitt, of each euglycaemic clamp. First, true steady state conditions of glucose delivery apply, rather than the alternative approach of mathematical averaging of a variable glucose infusion rate.9.‘9 The second point is that, despite achieving plateau plasma insulin levels by 10 min, glucose infusion rates will continue to rise steadily over the initial 45 to 60 min of each clamp (Fig. 1).
359
This phenomenon was first described by Sherwin et al*’ who demonstrated that there is a consistent lag time of approximately 60 min from the time plasma insulin concentrations reach steady state to the time that the hormone has its maximal effect on glucose uptake. Therefore, only at steady state will the measured rate of glucose uptake reflect the effect of the prevailing IRI concentration. This means that the practice of measuring Rd by averaging the glucose infused from 20 to 120 rnin8,19 will underestimate the true Rd. Thus, we deliberately ensured that steady state insulin and glucose levels were present for at least 30 min before sampling commenced. In this study metabolic clearance rate of glucose (MCR,) rather than the disappearance rate of glucose (M or Rd) was used as a measure of insulin action. Previous experiments in dogs have suggested that MCR, is independent of plasma glucose level.13 The data presented in this paper confirms this fact in normal man over a range of plasma glucose of 5.4 f 0.3 to 10.2 t 0.7mmol/L and at plasma insulin concentrations above 25pI_J/ml. However, recently Best et al” found that at low plasma insulin levels (< 18 pU/ml) MCR, falls with increasing plasma glucose concentration. The current data shows that in normal man, MCR, is a valid measure of insulin mediated glucose uptake provided that plasma insulin levels are >25 pU/ml. In contrast, Rd (or M) is dependent on both the prevailing plasma glucose and plasma insulin concentrations and is significantly altered by variations in either of these parameters. In particular, changes in blood glucose dramatically affect Rd (Table 2). Therefore, to construct an insulin dose-response curve using Rd or M, it is necessary to ensure that plasma glucose remains identical at all times and in all individuals. In practice, identical plasma glucose concentrations may be difficult to achieve. The use of MCRo removes the need to have identical plasma glucose concentrations in all subjects, and therefore makes the technique more practical. Nonetheless, an attempt should still be made to keep blood glucose within narrow limits in all cases even when employing MCR,. Hypoglycaemia should be avoided because of the effects of the consequent hormone-induced counter regulation on glucose metabolism.‘* Significant hyperglycaemia should also be avoided because, while it is true that in normal man at insulin concentrations >25 pU/ml, MCR, is independent of plasma glucose concentration, as shown in this study for the range 4.5 12.3 mmolL, this fact has not been proven in disease states such as diabetes. An additional point has to be made concerning the measurement of MCR, or M for insulin dose-response curves. In order to accurately measure these parame-
360
ters, it is essential to have a true value for the rate of appearance of glucose (Ra) (see Methods). We have previously shown in normal subjects that hepatic glucose production (HGP) is insignificant at the insulin levels used in the present studies.” Therefore at plateau, Ra can be assumed to equal the rate of glucose infused. This assumption cannot be made in obesity’ or in other disease states, and in these situations, the contribution of HGP to total Ra must be determined. In our obese subjects HGP was determined at steady state during infusion of the lower doses of insulin, using H3-3 glucose and was shown to be adequately suppressed at these insulin concentrations. A measure of insulin sensitivity (degree of right shift of the curve) and of responsiveness (maximal rate of glucose clearance) can only be obtained from an insulin dose-response curve.’ Because steady state conditions are required, the minimum time to gather data for a single point on a dose-response curve is 2 hrs. This limits the number of points that can be obtained for any one individual. The use of very high insulin infusion rates (giving plasma IRI levels of about 10,000 pU/ml) to ensure that the maximal response has been reached, has the potential danger of producing significant hypokalaemia. This problem can be resolved by adding an infusion of KC1 to the experimental protocols.’ An additional practical problem with such high doses of insulin is the need to infuse the subjects with glucose for several hrs at the end of the experiment so as to prevent hypoglycaemia. For these reasons, linearisation of the data to predict a maximal MCR, from the data points obtained with the lower insulin infusion rates would be of benefit. When the in vivo insulin dose-response curve is plotted on a linear insulin scale it describes an hyperbola which can be linearised in several ways, in order to derive the maximal response.17 The two most appropriate methods are plots of (I/MCR, Vs l/{IRI}-{IRI}basal) or {IRI}/MCR, Vs {IRI}.17 In the former method, the intercept of the straight line with the vertical axis gives each point is 1/Max MCRo. With this method, weighted equally and therefore depends on the experimental accuracy of every point. In the latter method 1/Max MCRG is given by the slope of the line and the points at the highest insulin concentrations are most weighted. For this reason it is unnecessary to subtract basal insulin levels prior to linearising the data. Excellent linearisation was obtained using either method, with a good correlation between the two methods of predicting the maximal response (r = 0.81, p < 0.01). Moreover, when the data from the four subjects in whom maximal MCRc was achieved experimentally were linearised, omitting in each case the experimen-
PROIElTO
ET AL.
tally derived data point at the highest insulin level excellent correlation was observed between the predicted and the experimentally derived MCRo (r = 0.953,~ < 0.01) (Table 1). Therefore, in practice, to define insulin sensitivity and responsiveness using a single 6 hr study, 2 points at IRI infusion rates between 30 and 100 mU/Kg/hr are needed to determine insulin sensitivity, and 1 point (at 600mU/Kg/hr) is needed to measure maximal response. If insulin resistance is likely, the insulin infusion rates for the 2 lower points should be raised to between 50 and 300 mU/Kg/hr. In such situations if the maximal MCR, is not reached experimentally, then this point can be confidently predicted from the transformed data. In our normal subjects (Fig. 3) the insulin level at which maximal response was obtained was between 200-250 pU/ml which is similar to other reports.” Before a 6 hr, 3 point dose-response curve technique can be accepted, the plasma glucose and MCRG must be shown to be stable over such a prolonged period. Previous studiesz3 have shown a reduction of insulin binding 5 hr after the start of a constant insulin infusion, and theoretically this could lead to a reduction in insulin action and thus MCR,. The data presented in Fig. 2, indicates that in this study the MCR, did not change significantly over a 6 hr period, despite the ongoing physiological hyperinsulinaemia. Although Doberne, L. et alz4 have recently reported a significant increase in MCRF 5 hrs after combined glucose/insulin infusion, Insel, J.R. et alz3 also found no significant change in the plasma glucose concentrations between the third and fifth hrs of a combined infusion of adrenaline, propanolol, insulin and glucose. Traditionally insulin sensitivity has been defined in terms of the Km or ED,, (insulin concentration giving half the maximal response). For this parameter to be measured the maximal response needs to be known. An alternative estimate of sensitivity is the insulin concentration at which an arbitrarily chosen insulin mediated response is achieved, ie. the degree of right shift of the curve. Table 1 shows a comparison between the ED50 and the insulin concentrations giving an MCR, of clearly 350ml/m*/min (ED M,-RG3S,,).Both methods separate the group of normal subjects from the two significantly obese individuals. No firm conclusions can be drawn from our data regarding the insulin response in obesity as only two of our subjects were significantly obese. Nonetheless, the obvious insulin insensitivity in these two individuals confirms a similar finding in a large group of patients studied using multiple glucose clamps over several days.’ An addi-
INSULIN DOSE RESPONSE CURVE IN MAN
361
tional important observation was the excellent correlation between insulin sensitivity and ponderal index in the 7 nonobese subjects. In this study blood glucose measured from the venous blood drawn from a cubital vein was used in the performance of the euglycaemic clamp and for estimation of MCRd. There is a recognised arterio-venous difference in glucose level due to the removal of glucose by muscle. In normal resting individuals at steady state this difference is insignificant, but the difference may become important when glucose uptake is stimulated by high insulin levels.26 This could be a potential problem in comparing insulin sensitivity in normal and resistant states. For this reason it would be preferable to use arterialised venous blood, when making comparisons between groups. In conclusion, this paper describes a practical
method of constructing in vivo insulin dose-response curves. The method is precise and allows in a single 6 hr study the construction of a dose-response curve from which insulin sensitivity and insulin responsiveness can be estimated.
ACKNOWLEDGMENTS We wish to thank the staff of the St. Vincent’s Hospital Biochemistry Department for their excellent technical assistance. We also wish to thank Novo Laboratories (Australia) Pty. Ltd. for financial assistance with the purchase of Somatostatin. We are grateful to Ms. Karen Dawson for her assistance in the preparation of the manuscript. This work was supported in part by the Rhoden White Foundation, The Sheppard Lowe Scholarship and the John Claude Kellion Foundation and was performed whilst the authors were members of the staff of the Endocrine Unit of St. Vincent’s Hospital, Melbourne. We are indebted to the Hospital authorities for permission to publish.
REFERENCES 1. Kahn RC: Insulin resistance, insulin sensitivity and insulin unresponsiveness: A necessary distinction. Metabolism 27: 18931902, 1978 2. Rabinowitz D, Zierler KL: Forearm metabolism in obesity and its response to intra arterial insulin. J Clin Invest 41:2173-2181, 1962 3. Beck P, Schalch DS, Parker ML: Correlative studies of growth hormone and insulin plasma concentrations with metabolic abnormalities in acromegaly. J Lab Clin Med 66:366-376, 1965 4. Megyesi C, Samols E, Marks V: Glucose tolerance and diabetes in chronic liver disease. Lancet 2:1051-1056, 1967 5. De Fronzo RA: Pathogenesis of glucose intolerance in uremia. Metabolism 27:1866-1880, 1978 6. Reaven GM, Olefsky JM: The role of insulin resistance in the pathogenesis of diabetes mellitus. Adv Metab Disord. 9:313-331, 1978 7. Andres R, Swerdolff R, Pozefsky T et al: Manual feedback technique for the control of blood glucose concentration. In Skeggs, L. ed. Automation in analytical chemistry. White Plains, N.Y.. Mediad 1966, ~486-49 I 8. De Fronzo RA, Tobin JD, Andres R: Glucose clamp technique: A method for quantifying insulin secretion and resistance. Am J Physiol237(3) E214-E223, 1979 9. Kolterman OG, lnsel J, Saekow M et al: Mechanisms of insulin resistance in human obesity. J Clin Invest 65:1272-1284, 1980 IO. Rizza R, Miles J, Verdonk C et al: Insulin sensitivity in man: A method for single day dose-response assessment using the euglycaemic glucose-insulin clamp technique. Clin Res 28:409, 1980 1 I. Proietto J, Alford FP, Dudley FJ: The mechanism of the carbohydrate intolerance of cirrhosis. J. Clin Endocrinol Metab 51:1030-1036,198O 12. Steele R, Wall JS, de Bodo RC et al: Measurement of size and turnover rate of body glucose pool by the isotope dilution method. Am J Physiol 187: 15524, 1956 13. Cherrington AD, Williams PE, Harris MS: Relationship
between the plasma glucose level and glucose uptake in the conscious dog. Metabolism 27:787-791, 1978 14. Walford S, Gale EAM, Allison SP et al: Self monitoring of blood glucose, Lancet 1 732-735, 1978 15. Albano JDM, Ekins RP, Maritz G et al: A sensitive precise radioimmunoassay of serum insulin relying on charcoal separation of bound and free hormone moieties. Acta Endocrinol (Copenh) 70:487-509, 1972 16. Heding LG: Radioimmunological determination of human c-peptide in serum. Diabetologia 11:541-548, 1975 17. Riggs DS: The mathematical approach to physiological problems. Baltimore, Williams and Wilkins, 1963, ~272-3 I 1 18. National Diabetes Data Group: Classification and diagnosis of diabetes mellitus and other categories of glucose intolerance. Diabetes 28:1039-1057, 1979 19. De Fronzo RA: Glucose intolerance and aging. Evidence for tissue insensitivity to insulin. Diabetes 28:1095-l 101, 1979 20. Sherwin RS, Kramer KJ, Tobin JD, et al: A model of the kinetics of insulin in man. J Clin Invest 53:1481-1492, 1974 21. Best JD, Taborsky GJ Jr, Halter JB, et al: Is metabolic clearance rate of glucose a valid measure of hormone-dependent glucose disposal when insulin levels are low? Clin Res 29:93, 1981 22. Sacca L, Sherwin R, Hendler R, et al: Influence of continuous physiologic hyperinsulinaemia on glucose kinetics and counterregulatory hormones in normal and diabetic humans. J. Clin Invest 63:849-857, 1979 23. lnsel JR, Kolterman regulation of insulin receptor 1981
OG, Saekow M, et al: Short-term affinity in man. Diabetes 29: 132-l 39,
24. Doberne L, Greenfield physiologic hyperinsulinaemia 1981
MS, Reaven GM: Acute influence of of insulin sensitivity. Clin Res 53A,
25. McGuire EAH, Helderman JH, Tobin JD, et al: Effects of arterial versus venous sampling on analysis of glucose kinetics in man. J Appl Physiol41 (4) 565-573, 1976
M. Harewood,
P. Aitken,
A. Nankervis,
G. Caruso, and F. Alford
To adequately investigate the state of insulin resistance, an insulin dose-response curve should be constructed so that insulin sensitivity (right shift of dose-response curve) and insulin responsiveness (maximal response) can be determined. This paper describes and validates in man a practical in vivo insulin dose-response curve technique, using a modification of the euglycaemic clamp described by De Fronro et al. Insulin action at steady state was expressed as metabolic clearance rate of glucose (MC&) rather than overall rate of glucose disappearance (M or Rd). MCR, was chosen because at plasma insulin concentrations >25~U/ml it was shown (n = 5) not to be altered by changes in blood glucose concentration (MCR, 379 + 23 and 408 + 19 ml/mz/min: at plasma glucose concentrations 5.4 + 0.3 and 10.2 f 0.7 respectively). whereas Rd was critically dependent on the prevailing blood glucose concentration (Rd 2007 +- 128 and 4124 + 219~mol m*/min respectively). MCR, was demonstrated to be stable over a 8 hr period (n = 7) and to be reproducible In = 4). Insulin dose-response curves (MCR, Versus insulin concentration) were performed on two obese and seven normal weight individuals. The insulin dose-response curves were linearized, allowing accurate prediction of the maximal MCR,, as compared to the experimentally determined maximal response (r = 0.953 p < 0.01). The use of this transformation obviates the need to employ very high insulin infusion rates to determine the maximal insulin response. In conclusion, the technique permits, in a single 6 hr study, a precise insulin dose-response curve to be constructed for accurately determining insulin sensitivity and responsiveness.
I
NSULIN RESISTANCE, defined as a state in which normal insulin levels elicit a less than normal biological response,’ is thought to be responsible for the glucose intolerance of many disorders including obesity,* acromegaly,3 cirrhosis,4 uraemia,’ and noninsulin dependent diabetes.6 The resistance can be of three types. In the first type, greater than normal hormone levels are needed to elicit the maximal response; in the second, the maximal response cannot be achieved despite the presence of supraphysiological or pharmacological insulin levels. Thirdly, a combination of both types is possible. Kahn’ has suggested that the first type be called insulin insensitivity and the second, insulin unresponsiveness. The distinction between these types of resistance is important because each may have a different underlying mechanism. For example, the spare receptor theory suggests that moderately reduced receptor binding could result in insensitivity. A post receptor defect or very severely reduced receptor binding may cause unresponsiveness. To distinguish between insulin insensitivity, unre-
From the Endocrine Unit and Department of Medicine, University of Melbourne, St. Vincent’s Hospital, Fitzroy, Victoria, Australia. Received for publication January 9. 1981. J.P. is a recipient of a National Health & Medical Research Council Medical Postgraduate Research Scholarship. This work was performed in partial fulfillment of the thesis requirement. Presented in part to the Annual Scientific Meeting of the Ausiraban Diabetes Society, Armidale. New South Wales. August 1980. Address requests for reprints to: Dr. J. Proietto, Endocrine Unit, St. Vincent’s Hospital, Victoria Parade, Fitzroy, VIC. 3065. AUStralia. 0 1982 by Grune & Stratton. Inc. 0026-0495/82/3104-0007$01.00/0
354
sponsiveness or a combination of the two, it is essential that adequate insulin dose response curves be obtained. The development of the euglycaemic glucose clamp technique by Andres, R. et al7 as described by De Fronzo, R. et al* made it possible to construct such curves. Initial reports of insulin dose-response curves using this technique required several studies over two to three days,’ although recently a one day insulin dose-response curve has been described.” The aim of this study was to devise and validate in man a practical in vivo insulin dose-response curve technique using the metabolic clearance rate of glucose (MCRo) as the measured parameter of insulin action. The method proved simple to perform without the use of complex equipment. A mathematical approach to the analysis of insulin dose-response curves is also presented which allows the derivation of the maximal response without the need for very high insulin infusions. MATERIALS
AND
METHODS
Subjects All subjects were healthy, ambulant volunteers on no medication and with no family history of diabetes. Sixteen normal weight subjects, 7 males and 9 females aged between 21 and 39 yr took part in the three validation experiments. Nine additional subjects, 7 males and 2 females aged between 29 and 50 yr had insulin dose-response curves performed (Table 1). Seven of these subjects had ponderal indices in the normal range (PI 18-30) and two were obese (Table 1). Written informed consent was obtained from all subjects and the protocols were approved by the Ethics and Research Committee of St. Vincent’s Hospital, Melbourne. All subjects were studied after an overnight fast and at least a 30 min period of recumbency. All solutions were infused, using 2 Gilson Minipuls II pumps (Gilson, Villiers le Bel. France) via a 20 cm catheter placed in an antecubital vein. Blood samples were collected at appropriate intervals from an indwelling butterfly needle placed
Metabolism, Vol. 3 1, No. 4 (April), 1982
INSULIN DOSE RESPONSE CURVE IN MAN
355
Table 1. Clinical Data and Results in the 9 Subjects who had Insulin Dose-Response sex Subject
Age (Yrs.)
17
Maximal MCR,
Fasting III&X
M
WJ/mll
Method
I#
Method
IRI Levels at* MCR,,,,
X Max MCR,
(ml/m’/min)
IRI
P0ildaW.l
Curves
IRI Levels at* Experimental’
II#
ml/m’/man
(ED.,) W/ml)
(ED,.,) W/ml)
20.9
8
686
641
650
35
35
21.3
8
639
624
630
43
44
23.5
7
666
764
41
43
24.2
6
597
613
600
46
53
24.8
3
602
525
-
49
66
26.3
10
576
647
54
68
26.5
6
464
459
43
76
35.9
16
557
593
210
360
43.9
27
492
509
165
360
29 18
M 31
19
M 35
20
M 39
21
M 43
22
F 41
23
M
440
36 24
M 50
25
F 34
1
1 #
Method I: Vs [IRI] ~ [IRI],, MCR,
-
IIRII
’ Method‘I: MCR,
Vs [IRI]
*Data determined from raw dose-response curve data contained in Fig. 3. in a contralateral arm vein. To overcome the problem of hormones adsorbing to the IV apparatus, insulin was placed in normal saline and pre-run for one hour prior to use. Somatostatin (Bachem, Torrance, USA) was infused in a solution containing 0.2% human serum albumin (Beringwerke, Germany) and 1000 KIU Trasylol (Aprotinin, Bayer Pharmaceuticals, Germany). Blood for hormone estimations was collected in 10% Trasylol, promptly centrifuged and plasma stored at ~ 20°C until the time of assay.
Oral Glucose Tolerance
3;
‘8
2 18
-.-.
---
--/
4 1
2e
Test
These were performed, using a 75 gram glucose subjects who had insulin dose-response curves.
Dose-Response
12
load, on the 9
Curve Technique
Nine subjects had euglycaemic clamps performed after the method of De Fronzo et al,* as illustrated in Fig. 1. When it became possible to estimate the approximate glucose infusion rate needed to maintain euglycaemia (usually after 60 min), the infusion rate of glucose was fixed (Fig. 1). On rare occasions, unexplained changes in blood glucose during the initial period of the euglycaemic clamp necessitated the clamp procedure being continued beyond the usual 60 min. In these cases, the glucose infusion rate was fixed by 90 min. Thirty minutes after fixing the glucose infusion rate, four blood samples were taken at 10 min intervals for estimation of plasma glucose and insulin. This process was then repeated at higher insulin infusion rates in order to construct three or more point dose-response curves. Three point curves were performed in a single day and four or more point curves over two sessions. At the end of each experiment three, 5 min samples of the fluid delivered by the glucose and insulin pumps were collected into weighed tubes. From these were calculated the exact rates of infusion of glucose and insulin. Insulin infusion rates ranged from 30mU/Kg/hr to 600mU/Kg/hr which resulted in plasma insulin levels of from 33 to 2190rU/ml. In view of our previous studies” hepatic glucose production (HGP) was assumed to be suppressed to insignificant values during all the euglycaemic clamps performed. This assumption could not be made in the two very obese individuals? and in them, glucose turnover was
765 432-
‘1 AA
.
1, .
.
.
.
.
.
.
.
.
.
AAAA ,
-10 0
30
60
90
120
T 1 ME (mins) Fig. 1. Plasma glucose, plasma insulin and insulin (shaded area) and glucose infusion rates during a representative euglycaemic clamp. The arrows indicate sampling times. Note that despite plasma insulin reaching a plateau level by 10 min. the glucose infusion rate needed to maintain euglycaemia continued to rise for 45 min.
356
estimated nique.‘*
PROIE-ITO ET AL.
at steady
state,
using
the H3-3
glucose
tracer
tech-
Validation Studies Validation of MCRG as a Measure of Insulin Action. It has been suggested from experiments in conscious dogs that MCRo is a measure of insulin action and is independent of the plasma glucose level.” To validate this in man, we studied 5 normal weight subjects during combined 5 hr cyclic Somatostatin (5rg/min) (Bachem Inc.. California, USA) glucose clamp procedures. An initial eugylycaemic clamp, as described above using insulin infusion rates of between 30 to 75mU/Kg/hour was followed by a hyperglycaemic c1amp.s The insulin infusion rate remained unchanged during both euglycaemic and hyperglycaemic study periods. During the last 30 min of each period, serial blood samples were taken for estimation of plasma glucose, insulin and C-peptide. Urine was collected over the last hour of each hyperglycaemic clamp for measurement of urinary glucose loss. Stability ofMCRG. To determine if steady state conditions and therefore MCRo are stable over a prolonged period, 7 subjects had euglycaemic clamps performed as described above using 75mU/ Kg/hr of insulin. Once steady state insulin and glucose infusion rates were achieved, they were fixed for the subsequent 5 hrs. Four blood samples were taken at IO min intervals between 90 and I20 min. 210 and 240 min and 330 and 360 min, for estimation of IRI and glucose levels and calculation of MCRo. Hepatic glucose production was again assumed to be suppressed to insignificant values” during these studies. Reproductibility of MCR,. To prove that this technique is reproducible, 4 normal weight subjects were studied I wk apart. On each occasion an euglycaemic clamp was performed using 75mU/ Kg/hr of insulin.
Analytical Procedures During the euglycaemic and hyperglycaemic clamps, blood glucose was estimated at the bedside, using a Boehringer reflomat (Boehringer Mannheim GmBh-Mannheim, W. Germany) and the glucose or Hypoglycaemie strips’” which give blood glucose readings in 2 min. Calculations for the clamps were rapidly performed with the help of a desk calculator using the formulae of De Fronzo et a1.s Plasma from samples taken at steady state, was analysed for glucose using a Union Carbide Centrifichem C400 analyser (Union Carbide Corp., New York, N.Y.) employing the hexokinase method. Only the latter hexokinase glucose estimations were used in subsequent calculations of MCRc. Plasma immunoreactive insulin (IRI) was estimated by radioimmunoassay employing dextran charcoal separation of the bound and free factions.” C-peptide was measured using the Novo C-peptide assay kit, employing synthetic human C-peptide and guinea pig anti human C-peptide antiserum.16 The relevant plasma samples from the tracer studies were processed as previously described.”
Calculations As stated above, during these studies hepatic glucose production is assumed in the non-obese normal subjects to be insignificant.” Therefore the rate of appearance of glucose (Ra) = the rate of infusion of glucose. In the obese individuals hepatic contribution to total Ra was measured at steady state using H’-3 glucose. When plasma glucose is stable (defined as a coefficient of variation of plasma glucose concentration at plateau of
plotting MCRo against the steady state plasma insulin levels. The curvilinear insulin dose-response curves were linearised by two transformations (i) I/MCR Vs I/[IRI]-[IRI] (basal); and (ii) [IRI]/MCR, Vs [IRI].” The straight lines so obtained by linear regression analyses were used to derive the maximal response. Linear regression analysis, paired and non-paired t tests were performed by standard statistical techniques.
RESULTS
Fig. 1 shows plasma glucose, plasma insulin, insulin infusion and glucose infusion rates, during a representative euglycaemic clamp. Glucose concentration remained at fasting levels throughout the 2 hr period. Plasma IRI rose to peak values within 5 min of the bolus insulin infusion and reached its steady state value by 10 min. It is important to note that despite the early rise of insulin to the plateau level, the amount of IV glucose needed to maintain euglycaemia rose steadily over the initial 45 min and only commenced to oscillate about its mean after 45 min. Similar patterns of glucose and insulin dynamics were observed in all the glucose clamps. Validation of MCR, as a Measure of Insulin Action Table 2 shows plasma glucose, IRI, C-peptide, Rd and MCR, values in the five subjects who underwent the Somatostatin/glucose clamp experiments. During the hyperglycaemic period, plasma insulin remained constant and C-peptide did not rise above basal values indicating adequate suppression of endogenous insulin production by Somatostatin during the hyperglycaemic clamp. In each case, Rd was shown to be critically dependent on the plasma glucose concentration with a rise in plasma glucose (5.4 * 0.3 to 10.2 f 0.7mmolL) resulting in a dramatic rise in Rd (2007 + 128 to 4124 2 219~mol/m2/min). In contrast, MCR, changed minimally (379 k 23 to 408 + 19ml/m’/min) (Table 2). Thus, a rise in blood glucose from 5.4 to 10.2mmolL resulted in a 105% rise in Rd whereas MCR, changed little (rose by 7%). Stability of MCRG Fig. 2 shows the mean + SEM of the plasma glucose and plasma IRI, and the mean and individual MCRo values measured at 2 hr intervals during the 6 hr euglycaemic clamp studies. Plasma glucose and insulin levels remained stable throughout the 6 hrs and there was no significant change in the mean MCRo’s (by paired t test p > 0.20). The mean coefficient of variation of the plasma glucose levels over the 6 hr was 9.9 k 0.9%, the variation for the plasma IRI values was 12.9 k 2.0%.
357
INSULIN DOSE RESPONSE CURVE IN MAN
Table 2. Effect of Changes of Plasma Glucose Level on the Rate of Disappearance (Rd) and Metabolic Clearance Rate (MCR,) of Glucose Plasma
Plasma GIUCOS?
1.
Fasting
2.
.05
1
5.2
26
,005
1,953
372
Permd
2
9.9
21
.005
3,868
387
5.0
3
Penod
1
5.0
27
.014
1,810
369
Period
2
8.0
29
,025
3,436
443
Fastmg Penod
4.
(ml/m’/min)
Period
Fasting
3.
5
MCR,
(~mollm’lmin~
(pmol/Ll
NJ/mli
4.0
Rd’
C-Peptide
IRI
(mmol/Ll
Subject
4.4
.09
7
.26
1
6.3
55
.04
1,904
302
Pernod 2
10.7
54
.18
4,311
402
Fastmg
4.5
5
.28
Period
1
5.8
30
.03
2,511
435
Penod
2
12.3
25
.03
4,282
351
4.7
9
.47
-
5.
Fasting Period
1
4.5
107
.08
1,858
418
Period
2
10.3
98
.13
4,722
457
Mean
c SEM
Fasting
4.7
+ 0.1
Period
1
5.4
k 0.3
49k
Period
2
10.2
+ 0.7
45
*When
calculating
Rd during
5.8
the hyperglycaemic
periods
k
k
1.0
.23
k .08
15
.03
k .Ol
2,007
+
128
379
f
23
14
.07
*
4,124
i
219
408
+
19
urinary
glucose
loss was
.03
subtracted
from
the total
rate of drsappearance
of glucose.
Reproducibility
Table 3 shows the results of the studies. Mean MCR,, IRI and glucose no significant difference between the (p > 0.50 for IRI levels; p > 0.95 for p > 0.95 for MCRG values). Dose- Response 80. T
60. 40. 20.
T,TT
r_____ .~~_-________~i-------zr~~ k I ! : !’
reproducibility levels showed paired studies glucose levels;
Curves
All seven normal weight subjects who had doseresponse curves performed had normal glucose tolerance.” Of the two obese subjects one had normal glucose tolerance (No. 25-Table 1) and the other had glucose intolerance (No. 24-Table 1). Fig. 3 is a graph of the individual dose-response curves (MCR, Vs plasma IRI level). The enclosed area in the upper panel represents the range of plasma insulin concentraTable 3. Reproducibility of Metabolic Clearance Rate of Glucose (MCR,) Metabolic Clearance Rate
T I ME
(mins)
Study II
13
400
471
Subject
14
447
458
Subject
15
361
336
Subject
16
428
380
Mean
MCR,
ml/m2/min
Fig. 2. Mean + SEM plasma glucose, plasma insulin and individual and mean metabolic clearance rates of glucose IMCR,) in seven subjects during a constant six hr insulin and glucose infusion. There were no significant changes in mean plasma glucose, plasma insulin or MCR,.
Study I
Subject
Mean
NJ/ml Mean
409
+
19
411
+ 32
IRI 71
+ 4
68
+ 2
Glucose
mmol/L
4.6
+ 0.1
4.6
+ 0.2
358
PROW-TO ET AL
I 10
1Q
lob0
tion and corresponding MCRo of 8 normal subjects in whom glucose turnover was measured previously, using H3-3 glucose.” The lower panel which depicts the normal range as a shaded area, shows the curves for the two obese individuals. These curves show that both obese individuals (Subjects 24 and 25-Table 1) have reduced insulin sensitivity but normal insulin responsiveness. When the dose-response curves presented in Fig. 3 were transformed (see Methods) a linear relationship between MCR, and plasma IRI was obtained for all individuals (group mean and range of the individual correlation coefficients for linearisation of data points were r = 0.948 + 0.014; range r = 0.887 to 0.999 for method I and r = .9983 f .0008; range r = .9936 to .9998 for method II). These transformations allow the derivation of the
Insulin dose-response curves Fig. 3. (MCR, Vs Plasma IRI) in the two obese and 7 normal weight individuals. In the upper panel the enclosed area represents the range of MC& and insulin levels obtained in a previous study in 8 normal individuals using H’-3 glucose to measure glucose turnover. The lower panel shows the normal range as a shaded area and the individual curves for the two obese individuals.
maximal MCRo. When a comparison is made between the experimentally determined and the derived maximal MCRo in the 4 subjects in whom maximal MCRo was reached experimentally, (Table 1, subjects 17, 18, 20, and 23), excellent agreement is observed (r = 0.953 p < 0.01). The accuracy of the predicted MCRo is further emphasized by the fact that the experimentally determined 600mU/Kg/hr point was deliberately omitted from the calculations for the purpose of this comparison. Table 1 gives the details of the 9 subjects studied, listed in order of increasing ponderal index, and shows the maximal MC&, the insulin concentration giving half the maximal response (ED,,,) and the insulin concentration giving an MCRo of 350ml/m2/min determined maximal (ED Mc~sSO)_The experimentally
INSULIN DOSE RESPONSE CURVE IN MAN
MCR,, the ED,, and the EDMcb_so data were read directly from the unsmoothed curves shown in Fig. 3. There was a significant correlation between ponderal index and insulin sensitivity for the normal weight group (subjects 17-23), expressed as ED,crt;jso ml/ m*/min (r = 0.930 p < 0.001) and as EDso (r = 0.713 p < 0.05). There was also a significant negative correlation between the ponderal index and the maximal MCR, (r = -0.820 p -c 0.01) for normal weight subjects. These relationships between ponderal index and insulin sensitivity and insulin responsiveness were strengthened when the obese subjects were included in the analysis. In the obese subjects the contribution of hepatic glucose production (HGP) to total Ra was measured at steady state using a primed constant infusion of H33 glucose. Subject 24 (Table 1) at a plasma insulin of 122pU/ml, had total suppression of HGP. In subject 25 the contribution of HGP to the total Ra during the lowest insulin infusion (plasma IRI 97rUml) was 9% (HGP 86~mol/m’/min; total Ra 920~mol/m2/min). If it is assumed that no further suppression occurred in this subject with the subsequent higher insulin infusion rates, the maximum percent contribution of HGP to total Ra in the subsequent 2 periods would be 5.4 and 3.4% respectively. The mean of the coefficients of variation of the steady state plasma glucose levels at plateau (n = 36) of all the dose-response curves was 7.0 + 0.6%. The mean coefficient of variation for the steady state plasma lR1 levels was 9.8 t 1.6%. DISCUSSION
This paper outlines and validates in man a practical method of constructing in vivo insulin dose-response curves using a modification of the euglycaemic clamp technique of De Fronzo et al.’ Two main modifications are described. The glucose infusion rate isJixed over the last 60 min of each 2 hr study and the data are presented as the metabolic clearance rate of glucose (MCR,), rather than the rate of disappearance of glucose (Rd or M). This modified technique gives a stable plateau of plasma glucose and MCR, over prolonged periods (6 hrs) (Fig. 2), and is shown to be highly reproducible (Table 2). There were several advantages from fixing the rate of the glucose infusion over the last 60 min and of sampling only over the last 30 mitt, of each euglycaemic clamp. First, true steady state conditions of glucose delivery apply, rather than the alternative approach of mathematical averaging of a variable glucose infusion rate.9.‘9 The second point is that, despite achieving plateau plasma insulin levels by 10 min, glucose infusion rates will continue to rise steadily over the initial 45 to 60 min of each clamp (Fig. 1).
359
This phenomenon was first described by Sherwin et al*’ who demonstrated that there is a consistent lag time of approximately 60 min from the time plasma insulin concentrations reach steady state to the time that the hormone has its maximal effect on glucose uptake. Therefore, only at steady state will the measured rate of glucose uptake reflect the effect of the prevailing IRI concentration. This means that the practice of measuring Rd by averaging the glucose infused from 20 to 120 rnin8,19 will underestimate the true Rd. Thus, we deliberately ensured that steady state insulin and glucose levels were present for at least 30 min before sampling commenced. In this study metabolic clearance rate of glucose (MCR,) rather than the disappearance rate of glucose (M or Rd) was used as a measure of insulin action. Previous experiments in dogs have suggested that MCR, is independent of plasma glucose level.13 The data presented in this paper confirms this fact in normal man over a range of plasma glucose of 5.4 f 0.3 to 10.2 t 0.7mmol/L and at plasma insulin concentrations above 25pI_J/ml. However, recently Best et al” found that at low plasma insulin levels (< 18 pU/ml) MCR, falls with increasing plasma glucose concentration. The current data shows that in normal man, MCR, is a valid measure of insulin mediated glucose uptake provided that plasma insulin levels are >25 pU/ml. In contrast, Rd (or M) is dependent on both the prevailing plasma glucose and plasma insulin concentrations and is significantly altered by variations in either of these parameters. In particular, changes in blood glucose dramatically affect Rd (Table 2). Therefore, to construct an insulin dose-response curve using Rd or M, it is necessary to ensure that plasma glucose remains identical at all times and in all individuals. In practice, identical plasma glucose concentrations may be difficult to achieve. The use of MCRo removes the need to have identical plasma glucose concentrations in all subjects, and therefore makes the technique more practical. Nonetheless, an attempt should still be made to keep blood glucose within narrow limits in all cases even when employing MCR,. Hypoglycaemia should be avoided because of the effects of the consequent hormone-induced counter regulation on glucose metabolism.‘* Significant hyperglycaemia should also be avoided because, while it is true that in normal man at insulin concentrations >25 pU/ml, MCR, is independent of plasma glucose concentration, as shown in this study for the range 4.5 12.3 mmolL, this fact has not been proven in disease states such as diabetes. An additional point has to be made concerning the measurement of MCR, or M for insulin dose-response curves. In order to accurately measure these parame-
360
ters, it is essential to have a true value for the rate of appearance of glucose (Ra) (see Methods). We have previously shown in normal subjects that hepatic glucose production (HGP) is insignificant at the insulin levels used in the present studies.” Therefore at plateau, Ra can be assumed to equal the rate of glucose infused. This assumption cannot be made in obesity’ or in other disease states, and in these situations, the contribution of HGP to total Ra must be determined. In our obese subjects HGP was determined at steady state during infusion of the lower doses of insulin, using H3-3 glucose and was shown to be adequately suppressed at these insulin concentrations. A measure of insulin sensitivity (degree of right shift of the curve) and of responsiveness (maximal rate of glucose clearance) can only be obtained from an insulin dose-response curve.’ Because steady state conditions are required, the minimum time to gather data for a single point on a dose-response curve is 2 hrs. This limits the number of points that can be obtained for any one individual. The use of very high insulin infusion rates (giving plasma IRI levels of about 10,000 pU/ml) to ensure that the maximal response has been reached, has the potential danger of producing significant hypokalaemia. This problem can be resolved by adding an infusion of KC1 to the experimental protocols.’ An additional practical problem with such high doses of insulin is the need to infuse the subjects with glucose for several hrs at the end of the experiment so as to prevent hypoglycaemia. For these reasons, linearisation of the data to predict a maximal MCR, from the data points obtained with the lower insulin infusion rates would be of benefit. When the in vivo insulin dose-response curve is plotted on a linear insulin scale it describes an hyperbola which can be linearised in several ways, in order to derive the maximal response.17 The two most appropriate methods are plots of (I/MCR, Vs l/{IRI}-{IRI}basal) or {IRI}/MCR, Vs {IRI}.17 In the former method, the intercept of the straight line with the vertical axis gives each point is 1/Max MCRo. With this method, weighted equally and therefore depends on the experimental accuracy of every point. In the latter method 1/Max MCRG is given by the slope of the line and the points at the highest insulin concentrations are most weighted. For this reason it is unnecessary to subtract basal insulin levels prior to linearising the data. Excellent linearisation was obtained using either method, with a good correlation between the two methods of predicting the maximal response (r = 0.81, p < 0.01). Moreover, when the data from the four subjects in whom maximal MCRc was achieved experimentally were linearised, omitting in each case the experimen-
PROIElTO
ET AL.
tally derived data point at the highest insulin level excellent correlation was observed between the predicted and the experimentally derived MCRo (r = 0.953,~ < 0.01) (Table 1). Therefore, in practice, to define insulin sensitivity and responsiveness using a single 6 hr study, 2 points at IRI infusion rates between 30 and 100 mU/Kg/hr are needed to determine insulin sensitivity, and 1 point (at 600mU/Kg/hr) is needed to measure maximal response. If insulin resistance is likely, the insulin infusion rates for the 2 lower points should be raised to between 50 and 300 mU/Kg/hr. In such situations if the maximal MCR, is not reached experimentally, then this point can be confidently predicted from the transformed data. In our normal subjects (Fig. 3) the insulin level at which maximal response was obtained was between 200-250 pU/ml which is similar to other reports.” Before a 6 hr, 3 point dose-response curve technique can be accepted, the plasma glucose and MCRG must be shown to be stable over such a prolonged period. Previous studiesz3 have shown a reduction of insulin binding 5 hr after the start of a constant insulin infusion, and theoretically this could lead to a reduction in insulin action and thus MCR,. The data presented in Fig. 2, indicates that in this study the MCR, did not change significantly over a 6 hr period, despite the ongoing physiological hyperinsulinaemia. Although Doberne, L. et alz4 have recently reported a significant increase in MCRF 5 hrs after combined glucose/insulin infusion, Insel, J.R. et alz3 also found no significant change in the plasma glucose concentrations between the third and fifth hrs of a combined infusion of adrenaline, propanolol, insulin and glucose. Traditionally insulin sensitivity has been defined in terms of the Km or ED,, (insulin concentration giving half the maximal response). For this parameter to be measured the maximal response needs to be known. An alternative estimate of sensitivity is the insulin concentration at which an arbitrarily chosen insulin mediated response is achieved, ie. the degree of right shift of the curve. Table 1 shows a comparison between the ED50 and the insulin concentrations giving an MCR, of clearly 350ml/m*/min (ED M,-RG3S,,).Both methods separate the group of normal subjects from the two significantly obese individuals. No firm conclusions can be drawn from our data regarding the insulin response in obesity as only two of our subjects were significantly obese. Nonetheless, the obvious insulin insensitivity in these two individuals confirms a similar finding in a large group of patients studied using multiple glucose clamps over several days.’ An addi-
INSULIN DOSE RESPONSE CURVE IN MAN
361
tional important observation was the excellent correlation between insulin sensitivity and ponderal index in the 7 nonobese subjects. In this study blood glucose measured from the venous blood drawn from a cubital vein was used in the performance of the euglycaemic clamp and for estimation of MCRd. There is a recognised arterio-venous difference in glucose level due to the removal of glucose by muscle. In normal resting individuals at steady state this difference is insignificant, but the difference may become important when glucose uptake is stimulated by high insulin levels.26 This could be a potential problem in comparing insulin sensitivity in normal and resistant states. For this reason it would be preferable to use arterialised venous blood, when making comparisons between groups. In conclusion, this paper describes a practical
method of constructing in vivo insulin dose-response curves. The method is precise and allows in a single 6 hr study the construction of a dose-response curve from which insulin sensitivity and insulin responsiveness can be estimated.
ACKNOWLEDGMENTS We wish to thank the staff of the St. Vincent’s Hospital Biochemistry Department for their excellent technical assistance. We also wish to thank Novo Laboratories (Australia) Pty. Ltd. for financial assistance with the purchase of Somatostatin. We are grateful to Ms. Karen Dawson for her assistance in the preparation of the manuscript. This work was supported in part by the Rhoden White Foundation, The Sheppard Lowe Scholarship and the John Claude Kellion Foundation and was performed whilst the authors were members of the staff of the Endocrine Unit of St. Vincent’s Hospital, Melbourne. We are indebted to the Hospital authorities for permission to publish.
REFERENCES 1. Kahn RC: Insulin resistance, insulin sensitivity and insulin unresponsiveness: A necessary distinction. Metabolism 27: 18931902, 1978 2. Rabinowitz D, Zierler KL: Forearm metabolism in obesity and its response to intra arterial insulin. J Clin Invest 41:2173-2181, 1962 3. Beck P, Schalch DS, Parker ML: Correlative studies of growth hormone and insulin plasma concentrations with metabolic abnormalities in acromegaly. J Lab Clin Med 66:366-376, 1965 4. Megyesi C, Samols E, Marks V: Glucose tolerance and diabetes in chronic liver disease. Lancet 2:1051-1056, 1967 5. De Fronzo RA: Pathogenesis of glucose intolerance in uremia. Metabolism 27:1866-1880, 1978 6. Reaven GM, Olefsky JM: The role of insulin resistance in the pathogenesis of diabetes mellitus. Adv Metab Disord. 9:313-331, 1978 7. Andres R, Swerdolff R, Pozefsky T et al: Manual feedback technique for the control of blood glucose concentration. In Skeggs, L. ed. Automation in analytical chemistry. White Plains, N.Y.. Mediad 1966, ~486-49 I 8. De Fronzo RA, Tobin JD, Andres R: Glucose clamp technique: A method for quantifying insulin secretion and resistance. Am J Physiol237(3) E214-E223, 1979 9. Kolterman OG, lnsel J, Saekow M et al: Mechanisms of insulin resistance in human obesity. J Clin Invest 65:1272-1284, 1980 IO. Rizza R, Miles J, Verdonk C et al: Insulin sensitivity in man: A method for single day dose-response assessment using the euglycaemic glucose-insulin clamp technique. Clin Res 28:409, 1980 1 I. Proietto J, Alford FP, Dudley FJ: The mechanism of the carbohydrate intolerance of cirrhosis. J. Clin Endocrinol Metab 51:1030-1036,198O 12. Steele R, Wall JS, de Bodo RC et al: Measurement of size and turnover rate of body glucose pool by the isotope dilution method. Am J Physiol 187: 15524, 1956 13. Cherrington AD, Williams PE, Harris MS: Relationship
between the plasma glucose level and glucose uptake in the conscious dog. Metabolism 27:787-791, 1978 14. Walford S, Gale EAM, Allison SP et al: Self monitoring of blood glucose, Lancet 1 732-735, 1978 15. Albano JDM, Ekins RP, Maritz G et al: A sensitive precise radioimmunoassay of serum insulin relying on charcoal separation of bound and free hormone moieties. Acta Endocrinol (Copenh) 70:487-509, 1972 16. Heding LG: Radioimmunological determination of human c-peptide in serum. Diabetologia 11:541-548, 1975 17. Riggs DS: The mathematical approach to physiological problems. Baltimore, Williams and Wilkins, 1963, ~272-3 I 1 18. National Diabetes Data Group: Classification and diagnosis of diabetes mellitus and other categories of glucose intolerance. Diabetes 28:1039-1057, 1979 19. De Fronzo RA: Glucose intolerance and aging. Evidence for tissue insensitivity to insulin. Diabetes 28:1095-l 101, 1979 20. Sherwin RS, Kramer KJ, Tobin JD, et al: A model of the kinetics of insulin in man. J Clin Invest 53:1481-1492, 1974 21. Best JD, Taborsky GJ Jr, Halter JB, et al: Is metabolic clearance rate of glucose a valid measure of hormone-dependent glucose disposal when insulin levels are low? Clin Res 29:93, 1981 22. Sacca L, Sherwin R, Hendler R, et al: Influence of continuous physiologic hyperinsulinaemia on glucose kinetics and counterregulatory hormones in normal and diabetic humans. J. Clin Invest 63:849-857, 1979 23. lnsel JR, Kolterman regulation of insulin receptor 1981
OG, Saekow M, et al: Short-term affinity in man. Diabetes 29: 132-l 39,
24. Doberne L, Greenfield physiologic hyperinsulinaemia 1981
MS, Reaven GM: Acute influence of of insulin sensitivity. Clin Res 53A,
25. McGuire EAH, Helderman JH, Tobin JD, et al: Effects of arterial versus venous sampling on analysis of glucose kinetics in man. J Appl Physiol41 (4) 565-573, 1976