Substrate Oxidation Errors Calorimetry-Hyperinsulinemic
During Combined Indirect Glucose Clamp Studies
Anne W. Thorburn, Barry Gumbiner, Therese Flynn, and Robert R. Henry The procedure of indirect calorimetry is often combined with the hyperinsulinemic, euglycemic clamp technique so that intracellular rates of glucose oxidation (G,), fat oxidation (F,,), and energy expenditure (EE) can be determined at different insulin concentrations and rates of whole-body glucose uptake. In order to perform these calculations, rates of protein oxidation (PJ must be known and are usually estimated from urinary nitrogen (N) excretion. The use of urinary N assumes that this measurement accurately reflects P.. and is unaltered by the glucose clamp technique. To examine these assumptions and determine potential errors in rates of G,,, F,,, and EE with this method, eight healthy subjects each had basal urinary N excretion determined on 4 different days and during a 300 pmol/m*/min hyperinsulinemic, euglycemic clamp. Mean basal urinary N excretion was 6.4 f 1.6 mg/min. Within individuals, basal urinary N was highly variable on the 4 different days with a mean coefficient of variation (CV) of 36% f 16%. Over the range of basal respiratory quotient (RQ) values in this study (0.76 to 0.65) the day-to-day variation in basal urinary N resulted in potential errors of 11% to 23% for G,, 16% to 24% for F,,, but minimal effects ( 5 1%) on EE. During the hyperinsulinemic, euglycemic clamp, RQ increased to 0.95 or greater, while urinary N excretion, rather than decreasing as expected, increased by 47% (6.4 f 1.6 to 9.4 2 2.6 mg/min) due in part to increases in urea clearance from 37.5 it 6.7 to 75.2 2 12.4 mL/min (P < .025). This increased urinary N excretion had minimal influence on F,, and EE, but underestimated G,, by up to 5% at RG less than 0.95. A more accurate estimate of urinary N excretion during hyperinsulinemic clamps may be obtained by correcting for changes in urea clearance. These results indicate that basal urinary N excretion is highly variable and influenced by hyperinsulinemic glucose clamps. Thus, urinary N excretion, particularly during the basal state, may not accurately reflect changes in P,, and can lead to substantial errors in G, and F,,. Copyright 0 199 1 by W.B. Saunders Company
0 been widely used in combination with hyperinsuhnemic glucose clamping to partition glucose uptake into VER THE PAST 10 years, indirect calorimetry
has
oxidative and nonoxidative components while simultaneously assessing fat oxidation (F,) and energy expenditure (EE).‘.’ Measurement of these intracellular events is assumed not to be significantly affected by protein oxidation (P,,,),‘-6 which is calculated during most of these studies from a single urinary nitrogen (N) excretion measurement. However, urinary N may not accurately reflect P, under basal conditions and during hyperinsulinemic glucose clamping, and an inaccurate measurement could significantly affect the calculation of glucose oxidation (G,), F,,, and EE. Therefore, the aims of the current study were (1) to determine day-to-day individual variability in basal urinary N excretion, (2) to quantitate changes in urinary N excretion during hyperinsulinemic, euglycemic clamping, and (3) to determine the impact of these changes in urinary N excretion on substrate oxidation rates and EE. To do this, studies were performed in a manner analogous to those reported in the literature during combined calorimetry/ hyperinsulinemic, euglycemic clamps. Urinary N measurements were compared in the fasting (basal) state on 4 different days and on one of these days basal values were compared with those determined under hyperinsulinemic (300 pmol/m*/min), euglycemic clamp conditions in eight healthy subjects. The effect of any variations in urinary Non the calculation of G,, F,, and EE was calculated over a range of respiratory quotients (RQs). MATERIALS
AND METHODS
Subjects
Eight healthy male subjects with normal glucose tolerance, an average age of 33 years (range, 24 to 43) and weight 75.0 kg (range, 61.7 to 86.6), volunteered for the study. Written informed consent Mefabolism, Vol40,
No
4
(April), 1991: pp 391-398
was obtained from each subject and the experimental protocol was
approved by the UCSD Committee on Investigations Involving Human Subjects. No subject was ingesting any drug known to affect glucose or insulin metabolism during the course of the study. Subjects were hospitalized and studied over a 2-week period. During this time they consumed a standardized weight-maintenance diet containing55% carbohydrate, 30% fat, and 15% protein given in three feedings of l/5,2/5, and 2/5 of the total daily calories at 8:00 AM, 12:OOPM, and 590 PM, respectively. Mean energy intake was 2,380 kcalid (range, 2,100 to 3,000). Mean N intake was 14 g/d (range, 12 to 16) and was the same from day to day for each subject. Experimental
Protocol
Hyperinsulinemic, euglycemic clamp technique. The protocol used for the hyperinsulinemic, euglycemic clamp study is summarized in Fig 1 and was similar to those reported in the literature that have used this technique combined with indirect calorimetry.” All studies were performed after a lo- to 12-hour overnight fast. On the morning of the test, the subjects were woken at 6:00 AM and, after voiding, two peripheral venous catheters were inserted. One was placed in an antecubital vein for infusates, the other was inserted in a retrograde fashion into a hand vein for blood
From the Department of Medicine, University of California, San Diego; and the Veterans Administration Medical Center, San Diego, CA. Supported in part by the American Diabetes Association (ADA), the Medical Research Service of the Veterans Administration Medical Center Grants No. DK 38949from the National Institute of Diabetes and Digestive and Kidney Diseases, and Grant No. MO1 RR-00827 from the General Clinical Research Branch, Division of Research Resources, National Institutes of Health. B.G. is the recipient of an ADA California Afiliate Fellowship Grant. A.T. is a Neil Hamilton Fairley Fellow funded by National Health and Medical Research Council OfAustralia. Address reprint requests to Robert R. Henry, MO, VA Medical Center (V-l I I G), 3350 La Jolla Village Dr, San Diego. CA 92161. Copytight 0 1991 by W.B. Saunders Company 0026-0495/91/4004-0010$03.00/0 391
392
THORBURN ET AL
INSULIN GLUCOSE SOMATOSTATIN
CALORIMETRY URINE COLLECTION l
t -90
.
I
I
0
60
BASAL
SERUM UREA N COLLECTION
I
1
120
180
240
minutes
CLAMP Fig 1.
Summary of the study design.
sampling. The hand was kept in a warming device at 60 to 70°C to achieve arterialization of venous blood. After a 90-minute basal period, an intravenous infusion of insulin was begun with a bolus of 6000 pmol/m*, followed by a continuous insulin infusion of 300 pmol/m’/min for 240 minutes. Somatostatin (Bachem, Torrence, CA) was infused at a rate of 0.08 pmol/kg/min to ensure complete suppression of endogenous insulin secretion. Glucose was clamped at 5 mmol/L by a variable infusion of 20% glucose, at a rate determined in a feedback fashion by the measurement of serum glucose at 5-minute intervals. Serum insulin levels were measured twice during the basal period and at 120, 180, and 240 minutes during the clamp. Urea N was measured in serum once during the basal period and after 120 and 240 minutes of clamping to determine whether urea clearance was changing during the study. Indirectcalorimetrytechnique. Rates of G,, F,,, and EE under steady-state basal and clamp conditions were determined by open-circuit indirect calorimetry. A transparent plastic ventilated hood was placed over the subject’s head. To avoid air loss, an airtight jacket attached to the hood was snugly fastened around the upper arms and chest and a slight negative pressure maintained in the hood using a suction fan. The subject lay in a recumbent position and was not permitted to watch television, read, or listen to the radio so that he remained calm while in the hood. The subject was checked every 5 to 10 minutes to ensure he remained awake. Air from a nearby empty room was drawn through the hood at a rate of 35 to 50 L/min. Ventilation at the outlet was measured by a digital turbine volume transducer (Ventilation Measurement Module-l, SensorMedics, Anaheim, CA). A constant fraction of the air flowing out of the hood was dried by condensers (at WC), filtered for dust, continuously withdrawn by pumps, and analyzed = 50 times per second for oxygen and carbon dioxide by analyzers (0, by Magnos 4G and CO, by Uras 3G, Hartmann and Braun, Frankfurt, Germany), which measure differences in O2 and CO, concentration between outflowing and fresh incoming air. These analyzers are sensitive, with an accuracy of 0.02% vol for 0, and 0.004% vol for CO,. Standard gas mixtures of 20.00% 0, and 1.00% CO, were used to calibrate the analyzers before and after each study. Acetone, which has an RQ of 0.75, was burnt in a container inside the hood before and after each study to check the complete system of indirect calorimetry. The measurements of EE were verified from the weight of acetone burned and the stoichiometric equation of acetone oxidation. Measured EE corresponded within 2% of that calculated from acetone stoichiometry. The analogue voltage output of the analyzers and flowmeter were connected to an IBM PC AT via an analogue to digital conversion board (Data Translation, Marlbora. MA). The mean rates of CO2 produced
(VCO,, in Umin) and 0, consumed (Vo,, in Umin) were continuously measured, displayed every minute, and printed every 5 minutes. From these data, the rates of G,,, FOX,and EE were calculated over 20-minute intervals (see Data Analysis) during the basal and clamp periods. Subjects were removed from the hood for 1 hour between 60 and 120 minutes during the clamp. Urinnry N measurements. After completion of the basal and clamp calorimetry measurements and halfway through the clamp at 120 minutes, subjects voided and these urine samples were analyzed for creatinine, urea, and uric acid (see Fig 1). The nitrogen in these compounds was combined to give total urinary N excretion. On half the specimens, total N was also measured on an aliquot of urine by the Kjeldahl method, since this is the method most commonly used by other investigators. The two methods gave comparable results (within 7%) and showed comparable trends. As we were interested to see whether changes occurred in the concentration of these three major components of urinary N, we have reported these results rather than those obtained from Kjeldahl analysis. The rate of urinary N excretion over a 90-minute basal period was also measured on three additional mornings under identical conditions in the subjects to determine the daily variation in urinary N excretion over 4 days. Studies in each subject were completed within a 2-week period. Analytical Methods Blood drawn for glucose was separated by microcentrifuge and serum glucose determined by the glucose oxidase method (YSI 23A, Yellow Springs, OH). Blood for serum insulin was collected in untreated tubes and allowed to clot at room temperature. Serum insulin levels were measured by a specific double-antibody radioimmunoassay.’ Total urinary N was determined by Kjeldahl technique.’ Blood and urinary urea N, creatinine, and uric acid were measured using routine, semiautomated methods (Hitachi 737, Boehringer Mannheim Diagnostic, Indianapolis, IN). The intraassay and interassay coefficients of variation (CV) for urine urea N, uric acid, and creatinine were 1.5% and 1.5%, 0.6% and 1.2%, and 1.4% and 2.3%, respectively. Calculation of Substrate Oxidation Rates and EE The following equations can be derived from the number of liters of 0, consumed and CO2 produced per gram of glucose, fat, or protein.9,‘” VoZ = 0.746 G,, + 2.029 F,, + 0.966 P,,,
(equation 1)
VcoZ = 0.746 G,, + 1.430 F,, + 0.782 P,,,
(equation 2)
where G,, F,,, and P,, represent the oxidation rates of glucose, fat and protein respectively (in g/min), and Vo, and VcoZ are the rates of O2 consumed and CO, produced (in Wmin). The RQ was calculated by dividing Vco, by Voz. If P,, = 6.25 N,
(equation 3)
where N is the urinary N excretion rate (in g/min), then the following equations can be derived when equation 3 is substituted into equations 1 and 2 and the equations solved simultaneously: F,, = 1.67 VoL - 1.67 Vco: - 1.92 N,
(equation 4)
G,, = 4.54 Vco, - 3.20 Vo: - 2.87 N,
(equation 5)
In addition, since EE (in kcalimin) is equal to the following equation: EE = (G,, x 3.74) + (F,, x 9.50) + (P,, x 4.10),
(equation 6)
SUBSTRATE OXIDATION
393
ERRORS AND GLUCOSE CLAMPS
(from 6.4 to 1.6 mg/min), G, and F,, increase by constant amounts (0.014 and 0.010 gimin, respectively) over the range of basal RQ values (0.78 to 0.85). To illustrate these points, observe the dashed lines in Fig 2. At a basal RQ of 0.82, which was a typical value found during our studies (Table 2), a 4.8 mg/min variation in urinary N produces a 15% change in G, (0.091 to 0.105 g/min) (Fig 2A) and an 18% change in F,, (from 0.054 to 0.064 gimin) (Fig 2B). Over the range of basal RQ values (0.78 to 0.85) the variation in basal urinary N resulted in potential errors of 11% to 23% for G, and 16% to 24% for F,,. In contrast, the effect of a 4.8 mg/min variation in urinary N on EE is negligible (I 1%) over the range of basal RQ values (not illustrated). The reason that changes in urinary N excretion have minimal influence on EE is apparent from equation 7; urinary N is a small number that is subtracted from the sum of two much larger numbers. Even a threefold increase in urinary N would decrease EE by only 5% over the range of basal RQ values (0.78 to 0.85).
where 3.74, 9.50, and 4.10 are the calorific values of glucose, fat, and protein in kcal/g, respectively.‘” then substituting equations 3 to 5 into equation 6, EE = 1.11 Vcoz + 3.90 Vo, - 3.34 N.
(equation
7)
Correction of Urinary N for Changes in Urea Nitrogen Clearance If urea excretion increased under hyperinsulinemic clamp conditions, urea N in the blood (BUN) would decrease and total urinary N excretion would increase. This situation would result in an overestimate of the P,, rate and could underestimate G, and F,,. To examine this, P,, was compared with or without correcting the urinary N excretion rate for changes in urea excretion during the glucose clamp. This correction was made with the following equation: Urinary
N, (gimin)
=
-
Change
N during clamp (gimin)
in BUN (mg/L) change
x 0.60 x body weight
(kg)
in time (min) x 1,000 (equation
6)
This equation assumes that the urea pool is 60% of body weight” and does not change over time. NL represents the corrected urinary N excretion rate. All results are expressed as mean + SEM unless otherwise specified. The paired Student’s 1 test was used to make statistical comparisons.
Variation in Basal Hypetinsulinemic Clamp Urinary N Excretion
After 240 minutes of hyperinsulinemic clamping, there was an increase in RQ from 0.82 to 0.97 (Table 2). During this period, insulin levels increased from 29 k 7 to 653 2 50 pmol/L, and a 47% increase in urinary N excretion (6.4 2 2.3 to 9.4 + 2.8 mg/min, P < .05, Table 1) and hence calculated P,, (0.040 to 0.059 gimin, Table 2) occurred. Figure 3 illustrates the effect of this 47% increase in urinary N on G,, and F,,. When urinary N increased from 6.4 to 9.4 mg/min, G,, and F,, decreased by constant amounts (0.009 and 0.006 g/min, respectively) over the range of clamp RQ values (0.95 to LOO). This effect can be observed from the dashed lines illustrated in Fig 3. At a typical clamp RQ value of 0.97 (see Table 2) G,, decreased from 0.259 to 0.250 g/min (Fig 3A), corresponding to a 4% variation in G,, (Fig 3C) when urinary N increased from 6.4 to 9.4 mgimin. Over the range of clamp RQ values, the variation in G,, resulting from a 47% increase in urinary N ranged from 5% (at an RQ of 0.95) to 3% (at an RQ of 1.00).
RESULTS
Variation in Basal Urinary N Excretion Urinary N excretion in the basal state on 4 different days in each subject is listed in Table 1. The mean value for all studies was 6.4 +- 1.6 mg/min (mean t SD). On different days the mean range in the subjects’ urinary N excretion was 4.8 mg/min, with a CV of 36% ? 18%. Figures 2A and 2B illustrate that this day-to-day variation in basal urinary N can produce significant errors in the calculation of G, and F,, rates. Urinary N values corresponding to the mean (6.4 mg/min) and the mean minus the intraindividual range (6.4 - 4.8 = 1.6 mgimin) were chosen to illustrate this point. From Table 1 it is apparent that the values of 6.4 and 1.6 mg/min are realistic basal urinary N excretion rates. When urinary N decreases by 4.8 mg/min
Table 1. Urinary N (mg/min) Under Basal Conditions [Over a 90-Minute Period) on 4 Different Days and During a 300-pmol/m’/min
Clamp (From 120 to 240 Minutes)
Basal .~ Subject
DW
Day
1
2
3
W
cv
Dw 4
Clamp
Meall
WI
Dw 1
1
8.0
4.0
6.1
4.0
5.5
35
10.1
2
6.1
9.0
8.0
7.0
7.5
17
14.1
3
8.0
4.0
5.0
10.1
6.8
41
7.0
4
7.0
6.1
8.0
10.1
7.8
22
5.0
5
8.0
6.1
7.0
10.1
7.8
12.0
6
7.0
10.1
6.1
6.1
7.3
22 26
7
1.0
6.1
3.0
3.0
3.3
64
9.0
8
6.1
1.0
6.1
9.0
5.5
60
9.0
6.4 + 1.6
36 k 18
9.0
Mean k SD
6.4 _f 2.3
9.4 r 2.8
THORBURN ET AL
0.16 1
UN-l.6 UN=6.4
clamp RQ values (0.95 to 1.00). The effect of a 47% increase in urinary N on EE was also negligible ( < 2%) over the range of clamp RQ values (not illustrated). Therefore, increases in urinary N excretion during the clamp have a modest effect on G,, and much less effect on F,, and EE. Correcting Urinary N Excretion for Changes in Urea N Clearance During the Hyperindinemic Clamp
0.06 0.06 0.04 I 0.77
0.76
0.79
0.60
0.61
0.62
0.63
0.64
0.65
28 0.06 1
UN=1.6 UN=6.4 0.77
25
0.76
0.79
0.60
0.81
0.82
0.83
0.84
1
0.85
n
Fox
Gox 10: 0.77
I
1
I
0.78
0.79
0.80
I
0.81
1
0.82
I
0.83
I
0.84
I
0.85
In this study, urea constituted approximately 93% of the total N in urine. In the basal state, urea N excretion was 6.04 mg/min, uric acid was 0.13 mg of N/min, and creatinine was 0.27 mg of Nimin. During the clamp, the excretion of urea N increased to 8.85 mg/min, uric acid to 0.19 mg/min, and creatinine to 0.40 mg/min. Therefore, most of the 3 mg/min increase in urinary N excretion during the clamp (Table 1) was due to an increase in the excretion rate of urea (94%) with a very small increase in uric acid (2%) and creatinine (4%). Although creatinine and uric acid are not end products of protein oxidation, we included them in our measurement of urinary N-derived protein oxidation since total urinary N is what most investigators use to calculate protein oxidation and we wished to mimic the methods used by others. An increase in urea N clearance (37.5 2 6.7 to 75.2 2 12.4 mL/min,P < .025) (Fig 4) was evident from the decrease in serum urea N (10.1 -C 0.8 to 8.0 +_0.7 mmol/L, P < .05) and increase in urine urea N excretion during the clamp (0.36 k 0.06 to 0.57 + 0.08 mmol/min, P < .05). Urine output increased twofold to threefold from the basal to the end of the clamp period (1.4 2 0.2 to 3.7 2 0.8 ml/mitt, P < .025). To determine if changes in the rate of urine formation during the glucose clamp influence urea clearance and explain some of the variability in urinary N excretion, the rate of urinary N excretion in each basal and clamp study was compared with its corresponding urine volume formation and is shown in Fig 5. A significant correlation (r = .49, P < .Ol) existed between these parameters. To account for changes in urea clearance and provide a better estimate of protein oxidation from urinary N measurements, urinary N values were corrected for changes in serum urea N concentrations (equation 8). Correcting urinary N excretion for changes in urea clearance over the
Basal RQ Fig 2. Effect of a 4.8 mg/min variation in urinary nitrogen (UN) excretion (6.4 Y 1.6 mg/min) on (A) glucose oxidation and (B) fat oxidation over a range of basal RQ values in a normal subject. (C) Variation in glucose (G,,) and fat (F,,) oxidation as a percentage. Dashed lines highlight the results obtained at a typical basal FIG value of 0.82.
F,, values during the clamp were almost completely suppressed up to RQ values of 0.97, above which F,, rates were calculated as negative values. This meant that any error in F, led to a minimal change in F,,, but corresponded to a large percent variation in F,,. For example, at a typical clamp RQ value of 0.97, F, decreased from 0.001 to -0.005 g/min (Fig 3B), equivalent to a 100% variation in F, (Fig 3C) when urinary N increased from 6.4 to 9.4 mg/min. The variation in F, ranged from 75% to 100% over the range of
Table 2. Calorimetry Measurements
Over ZO-Minute Intervals During
Basal and Clamp Periods RQ
vco,
vo,
PO.*
Limin
Vco,No,
G,,
Fox
glmin
Basal (min) -60
to -40
0.82
0.200
0.243
0.040
0.112
0.060
-40
to -20
0.83
0.209
0.251
0.040
0.127
0.058
-20
to 0
0.82
0.219
0.266
0.040
0.125
0.067
Clamp
(min)
140 to 160
0.95
0.242
0.254
0.059
0.259
0.002
160 to 180
0.95
0.238
0.250
0.059
0.254
0.002
180 to 200
0.96
0.244
0.255
0.059
0.265
0.000
200 to 220
0.96
0.240
0.251
0.059
0.259
0.000
0.97
0.254
0.259
0.059
0.297
220 to 240 NOTE. *Protein
Results
are means
oxidation
of eight subjects.
= 6.25 x urinary
N excretion
rate (g/min).
-0.010
SUBSTRATE
OXIDATION
ERRORS
AND
GLUCOSE
CLAMPS
395
3A z
1
0.30
-iE
s
0.28
H
-we+/
m ,,zfj-t( _----
$j
E UN=6.4 _
P
0.24-)
UN=9.4
DISCUSSION
In most combined calorimetry-clamp studies, P,, determined from the urinary N excretion rate, is assumed to be constant and this assumption is thought not to influence the However, these assumptions are based calculation of G0X,2-5 on little evidence, since information on the way different experimental manipulations affect P,, is scanty.‘” The current study supports the earlier results of Tappy et al,” who clearly demonstrated that hyperinsulinemic glucose clamps alter not only urine flow, urea pool size, and clearance, but urinary N excretion and substrate oxidation rates as well. Our study now adds the new finding that large day-to-day variability in basal urinary N excretion occurs and can lead
/i
_ 0.95
0.94
0.96
0.97
in urea clearance during the basal period had little effect on P,, and G,, values. However, during hyperinsulinemic clamp conditions, this correction eliminated the increase observed in P,, determined from urinary N measurements and increased clamp G, values by 0.013 g/min (4% to 5%).
0.98
0.99
1.00
38
16 -
UN=6.4 14 -
“N-9.4
12 10 -
_I.“”
.
I
8-
0.94 0.95 0.96 0.97 0.98 0.99 1.00
1.0
-
0.8
-
0.6
-
0.4
-
0.2
-
3c --yyJ--
Fox
I I E
40
5 k P
20 __ 0.94
I
1 0.95
c
I 0.96
0.97
0.98
0.99
Gox
1.00
Clamp RQ
O160 -
Fig 3. Effect of a 47% increase in urinary nitrogen (UN) excretion (6.4 Y 9.4 mg/min) on (A) glucose oxidation and (B) fat oxidation over a range of clamp Ml values in a normal subject. (C) Variation in glucose (G..j and fat (FJ oxidation as a percentage. Dashed lines highlight the results obtained in a typical clamp RQ value of 0.97.
basal period had negligible effect on P, values and did not reduce the large intraindividual variability observed in the basal state. Figure 6 illustrates that during the clamp, correcting P,, for the change in serum urea N reduced P,, by 46% (P < .OOS).The mean corrected clamp P,, value was 20% less than P, values obtained during the basal period. The effect of correcting for changes in urea clearance on rates of P,, and G, in both the basal state and under hyperinsulinemic clamp conditions are illustrated in Fig 7. From this Figure, it can be seen that correcting for changes
140
-
120
-
100
-
80
-
60
-
40
-
20
-
O-
BASAL
CLAMP
Fig 4. Serum urea nitrogen, urine urea nitrogen formation, and urea nitrogen clearance in eight normal subjects under basal and hyperinsulinamic clamp conditions. Urea ckarance = (urine urea nitrogen formation/serum urea nitrogen) x 1,000.
THOREURN
396
-40
-20
160
0
BASAL
0
(./A1”“‘11”1”“11~
0
0002
0.004
0.006 URINARY
0.006
0010
NITROGEN
0.012
0014
0.016
0.016
to substantial errors in basal substrate oxidation rates. We found that basal urinary N excretion is not a constant; it varies on average by 4.8 mg/min from day to day (CV 36%), which is in contrast to more direct measures of P, using labeled leucine turnover studies, which have shown basal P,, to be constant.13 This variation would be of little concern if it did not influence the calculation of G, and F,,, but we have shown that it does, causing changes of 11% to 23%
Pox
240
(min)
Fig 7. Rates of G,, and P,, calculated from urinary N excretion with and without correcting urinary N for changes in urea clearance during the basal state and hyperinsulinemic clamp conditions in eight healthy subjects.
(g/min)
Fig 5. The rate of urinary nitrogen excretion and urine volume formation in eight subjects under four basal conditions (0) and a hyperinsulinemic, euglycemic clamp (0).
BASAL
AL
CLAMP Time
.I;
200
ET
CLAMP Pox
CLAMP Pox WITH BUN CORRECTION
Fig 6. Protein oxidation (g/min) in eight normal subjects calculated in the basal state and during the hyperinsulinemic, euglycemic clamp using the urinary N excretion rate (x6.25) with and wkhout correcting for changes in serum urea N concentration.
and 16% to 24% in basal G, and F,, rates, respectively. As reported previously,” we also found that P, varied during the hyperinsulinemic, euglycemic clamp as calculated from urinary N measurements, increasing by 47%. These urinary N-derived values for protein oxidation contrast with measures of P,, based on Y- or ‘H-leucine turnover, which have demonstrated that P, decreases by 28% to 60% under hyperinsulinemic clamp conditions identical to ours.‘3.‘4 If one assumes that these radiolabeled leucine studies accurately estimate P,,, which they probably do given the anabolic effect insulin has on protein metabolism,‘5 then urinary N-derived P,, values as determined in this study will underestimate G,, rates by up to 0.02 g/min (-10%). However, if urinary N measurements are corrected for changes in urea clearance during hyperinsulinemic glucose clamps, the recalculated P, value tends toward those reported using labeled leucine, reducing the error to less than 5% and providing a better estimate of G,, under ciamp conditions. Urinary N may be a poor index of P, in the basal state, since not all protein results in the liberation of equivalent amounts of N. Although the total amount of N derived from dietary protein intake did not vary from day to day for each subject, it is feasible that variations in the type of antecedent protein intake from day to day were responsible for some of the variation in the amount of urinary N excreted. For example, Livesay and Elial have recently shown that ingestion of different dietary proteins can theoretically result in a +14% to -38% variation from the accepted general value of 6.0 L/g of urinary N (or 0.966 L/g of protein used in equation 1). From this information it would be prudent to recommend that the type of protein, as well as the total amount of protein consumed before studies, be kept consistent during comparative study periods. Increased urinary N excretion rates during hyperinsulinemic clamps is not a widely appreciated phenomena and could be the result of a number of factors. Since urea forms 80% to 90% of all N in urine, the most likely explanation for the increase in urinary N excretion during the clamp is an
SUBSTRATE OXIDATION
397
ERRORS AND GLUCOSE CLAMPS
increase in urea clearance. In support of this, our results clearly show that urea clearance increases during the clamp. Although increased insulin levels might play a role in altering sodium retention, and hence changes in volume and urea clearance, Thiebaud et al” have shown no difference in urinary N during hyperinsulinemic glucose clamps over a range of serum insulin concentration from 455 to 8,128 pmol/L. Increased urea clearance is more likely to be due to an increase in urine output, since augmented urine flow is known to result in an increase in the fractional extraction rate of urea.‘8.‘9 This occurs because urea reabsorption is a passive process completely dependent on the rate of water reabsorption, which establishes the diffusion gradient within the kidney tubules.” This effect would also explain the positive correlation we observed between urine output and urinary N excretion. The increase in urine output during the clamp was largely a result of the large volume of fluid (-2 L), which had to be infused intravenously into the subjects in order to administer the various infusates during the hyperinsulinemic clamp. Therefore, it appears that at least some of the increase in urinary N under hyperinsulinemic clamp conditions can be explained by variations in tluid intake that cause changes in urea clearance. In the basal state, changes in urea clearance were smaller and did not explain the large intraindividual variation in urinary N under these conditions. The short urine collection period (90 minutes for basal and 120 minutes for clamp) would also contribute to the variability in urinary N. These periods were chosen to mimic urine collection times typically used by investigators during hyperinsulinemic clamps. Although we did not set out to determine the etiology of variations in urinary N, we have shown that this variability causes substantial errors in G,, and F,, in the basal state and errors in G,, under hyperinsulinemic clamp conditions when an erroneous urinary N value is used to calculate substrate oxidation rates. Our findings strongly support a recent report by Tappy et al,” which showed that without correcting for changes in the urea N pool, urinary N excretion may substantially overestimate P,. In contrast, variations in urinary N do not significantly affect EE. This has been previously demonstrated by Bursztein et al’ and, more recently, by Ferrannini.‘” Bursztein’s report is commonly used to substantiate the incorrect claim that G, is unaffected by P,. However, this report only illustrates that EE is unaffected by P,,. It needs to be emphasized that the present study was not designed to determine the most accurate method of quantitating P,,. Rather, the aim was to determine whether urinary N excretion is variable when measured in a manner common to most investigators combining indirect calorimetry with hyperinsulinemic clamps and to quantitate what effect this variability would have on substrate oxidation rates and EE. This study does not indicate that P, per se is variable or influenced by the hyperinsulinemic glucose clamp procedure. It simply demonstrates that urinary N is a highly variable index of P, under both basal and glucose clamp conditions. Furthermore, the methods most com-
monly used to quantitate urinary N content during glucose clamps tend to be inaccurate, since they either include the measurement of N in substances not associated with P, (Kjeldahl technique) or fail to measure some of them that are (eg, ammonia). Further studies are needed to determine methods by which basal P, may be more accurately determined from urinary N measurements. However, during glucose clamp studies, errors that result from calculating P, directly from urinary N are smaller and at least partially correctable. The first precaution is to keep intravenous fluid intake, and therefore urinary volume, relatively constant during comparative study periods to keep variations in urea clearance to a minimum. Second, corrections can be made for changes in urea clearance by measuring BUN (equation 8). It should be noted that correction of urinary N excretion only for changes in urea N clearance assumes that the decrease in BUN that occurred in all subjects in this study was due solely to increased urea N clearance during glucose clamping. Since both decreased urea production and increased urea volume of distribution could also account for some of the decrease in BUN that occurred, this correction formula would tend to overestimate the contribution of increased urea N clearance. When we made this correction, urinary N during the clamp decreased by 4 mg/min. Others have also found that BUN decreases under hyperinsulinemic clamp conditions’3.20.21or after a mea1*2~23 and have often corrected urinary N values accordingly.‘2~“~22”In our study, correcting clamp urinary N for these changes in urea clearance reduced P,, by approximately 50% and resulted in a 20% lower P,, value than basal. Therefore, correcting urinary N for changes in urea clearance provides a better estimate of P,, which, although not perfect, results in smaller errors in the calculation of G,, during clamping. This study has shown that the measurement of basal urinary N is highly variable and can result in significant errors in G,, and F,,, but not EE. During hyperinsulinemic, euglycemic clamp conditions, urinary N increases by 47%, rather than decreasing as expected, due in large part to increases in urea clearance. This change has minimal effect on F,, and EE, but underestimates G,, by up to approximately 5% compared with values calculated from basal urinary N values and by approximately 10% compared with values derived from leucine turnover studies. This study should alert investigators that urinary N measurements can be highly variable, both in the basal state and during hyperinsulinemic glucose clamping, leading to substantial errors in substrate oxidation rates as determined by indirect calorimetry. It is clear that improved methods to accurately quantitate in vivo P,, during glucose clamp-calorimetry studies need to be developed.
ACKNOWLEDGMENT We would like to thank the nursing staff of the Special Diagnostic and Treatment Unit, San Diego Veterans Administration Medical Center, John Murphy and Ken Winders for excellent technical assistance, and Cleon Tate for typing the manuscript.
398
THORBURN ET AL
REFERENCES
1. Golay A, Schutz Y, Felber JP, et al: Lack of thermogenic response to glucose/insulin infusion in diabetic obese subjects. Int J Obes l&107-116,1986 2. DeFronzo RA, Jacot E, Jequier E, et al: The effect of insulin on the disposal of intravenous glucose. Results from indirect calorimetry and hepatic and femoralvenous catheterization. Diabetes 30:1000-1007,198l 3. Felber JP, Thiebaud D, Maeder E, et al: Effect of somatostatin-induced insulinopenia on glucose oxidation in man. Diabetologia 25:325-330,1983 4. Jacot E, DeFronzo RA, Jequier E, et al: The effect of hyperglycemia, hyperinsulinemia, and route of glucose administration on glucose oxidation and glucose storage. Metabolism 31:922930,1982 5. Thiebaud D, Jacot E, DeFronzo RA, et al: The effect of graded doses of insulin on total glucose uptake, glucose oxidation, and glucose storage in man. Diabetes 31:957-963,1982 6. Bursztein S, Saphar P, Glaser P, et al: Determination of energy metabolism from respiratory functions alone. J Appl Physiol42:117-119, 1977 7. Desbuquois B, Aurbach GA: Use of polyethylene glycol to separate free and antibody bound peptide hormones in radioimmunoassay. J Clin Endocrinol Metab 33:732-738,197l 8. DiGiorgio J: Nonprotein nitrogenous constituents, in Henry RJ, Connor DC, Winkelman JW (eds): Clinical Chemistry: Principles and Technics. Hagerstown, MD, Harper & Row, 1974, pp 503-563 9. Consolazio CF, Johnson RE, Pecora LJ: Physiological Measurements of Metabolic Functions in Man. New York, NY, McGraw-Hill, 1963, pp 313-339 10. Ferrannini E: The theoretical bases of indirect calorimetry: A review. Metabolism 37:287-301,1988 11. Jequier E, Acheson K, Schutz Y: Assessment of energy expenditure and fuel utilization in man. Ann Rev Nutr 7:187-208, 1987 12. Tappy L. Owen OE, Boden G: Effect of hyperinsulinemia on
urea pool size and substrate oxidation rates. Diabetes 37:12121216,1988 13. Castelhno P, Luzi L, Simonson DC, et al: Effect of insulin and plasma amino acid concentrations on leucine metabolism in man. Role of substrate availability on estimates of whole body protein synthesis. J Clin Invest 80:1784-1793,1987 14. Tessari P, Nosadini R, Trevisan R, et al: Defective suppression by insulin of leucine-carbon appearance and oxidation in type I, insulin-dependent diabetes mellitus. Evidence for insulin resistance involving glucose and amino acid metabolism. J Clin Invest 77:1797-1804,1986 15. Jefferson LS: Role of insulin in the regulation of protein synthesis. Diabetes 29:487-496,198O 16. Livesay G, Elia M: Estimation of energy expenditure, net carbohydrate utilization, and net fat oxidation and synthesis by indirect calorimetry: Evaluation of errors with special reference to the detailed composition of fuels. Am J Clin Nutr 47:608-628,1988 17. Thiebaud D, Jacot E, DeFronzo RA, et al: The effect of graded doses of insulin on total glucose uptake, glucose oxidation, and glucose storage in man. Diabetes 31:957-963, 1982 18. Sullivan LP, Grantham JJ: Physiology of the Kidney. Philadelphia, PA, Lea & Febiger, 1982, pp 109-111 19. Vander AJ: Renal Physiology. New York, NY, McGrawHill, 1980, pp 44-46 20. Acheson KJ, Ravussin E, Wahren J, et al: Thermic effect of glucose in man. Obligatory and facultative thermogenesis. J Chn Invest 74:1572-1580,1984 21. Owen OE, Trapp VE, Reichard GA, et al: Effects of therapy on the nature and quantity of fuels oxidised during diabetic ketoacidosis. Diabetes 29:365-372, 1980 22. Flatt JP, Ravussin E, Acheson KJ, et al: Effects of dietary fat on postprandial substrate oxidation and on carbohydrate and fat balances. J Clin Invest 76:1019-1024, 1985 23. Acheson KJ, Flatt JP, Jequier E: Glycogen synthesis versus lipogenesis after a 500 gram carbohydrate meal in man. Metabolism 31:1234-1240, 1982
During Combined Indirect Glucose Clamp Studies
Anne W. Thorburn, Barry Gumbiner, Therese Flynn, and Robert R. Henry The procedure of indirect calorimetry is often combined with the hyperinsulinemic, euglycemic clamp technique so that intracellular rates of glucose oxidation (G,), fat oxidation (F,,), and energy expenditure (EE) can be determined at different insulin concentrations and rates of whole-body glucose uptake. In order to perform these calculations, rates of protein oxidation (PJ must be known and are usually estimated from urinary nitrogen (N) excretion. The use of urinary N assumes that this measurement accurately reflects P.. and is unaltered by the glucose clamp technique. To examine these assumptions and determine potential errors in rates of G,,, F,,, and EE with this method, eight healthy subjects each had basal urinary N excretion determined on 4 different days and during a 300 pmol/m*/min hyperinsulinemic, euglycemic clamp. Mean basal urinary N excretion was 6.4 f 1.6 mg/min. Within individuals, basal urinary N was highly variable on the 4 different days with a mean coefficient of variation (CV) of 36% f 16%. Over the range of basal respiratory quotient (RQ) values in this study (0.76 to 0.65) the day-to-day variation in basal urinary N resulted in potential errors of 11% to 23% for G,, 16% to 24% for F,,, but minimal effects ( 5 1%) on EE. During the hyperinsulinemic, euglycemic clamp, RQ increased to 0.95 or greater, while urinary N excretion, rather than decreasing as expected, increased by 47% (6.4 f 1.6 to 9.4 2 2.6 mg/min) due in part to increases in urea clearance from 37.5 it 6.7 to 75.2 2 12.4 mL/min (P < .025). This increased urinary N excretion had minimal influence on F,, and EE, but underestimated G,, by up to 5% at RG less than 0.95. A more accurate estimate of urinary N excretion during hyperinsulinemic clamps may be obtained by correcting for changes in urea clearance. These results indicate that basal urinary N excretion is highly variable and influenced by hyperinsulinemic glucose clamps. Thus, urinary N excretion, particularly during the basal state, may not accurately reflect changes in P,, and can lead to substantial errors in G, and F,,. Copyright 0 199 1 by W.B. Saunders Company
0 been widely used in combination with hyperinsuhnemic glucose clamping to partition glucose uptake into VER THE PAST 10 years, indirect calorimetry
has
oxidative and nonoxidative components while simultaneously assessing fat oxidation (F,) and energy expenditure (EE).‘.’ Measurement of these intracellular events is assumed not to be significantly affected by protein oxidation (P,,,),‘-6 which is calculated during most of these studies from a single urinary nitrogen (N) excretion measurement. However, urinary N may not accurately reflect P, under basal conditions and during hyperinsulinemic glucose clamping, and an inaccurate measurement could significantly affect the calculation of glucose oxidation (G,), F,,, and EE. Therefore, the aims of the current study were (1) to determine day-to-day individual variability in basal urinary N excretion, (2) to quantitate changes in urinary N excretion during hyperinsulinemic, euglycemic clamping, and (3) to determine the impact of these changes in urinary N excretion on substrate oxidation rates and EE. To do this, studies were performed in a manner analogous to those reported in the literature during combined calorimetry/ hyperinsulinemic, euglycemic clamps. Urinary N measurements were compared in the fasting (basal) state on 4 different days and on one of these days basal values were compared with those determined under hyperinsulinemic (300 pmol/m*/min), euglycemic clamp conditions in eight healthy subjects. The effect of any variations in urinary Non the calculation of G,, F,, and EE was calculated over a range of respiratory quotients (RQs). MATERIALS
AND METHODS
Subjects
Eight healthy male subjects with normal glucose tolerance, an average age of 33 years (range, 24 to 43) and weight 75.0 kg (range, 61.7 to 86.6), volunteered for the study. Written informed consent Mefabolism, Vol40,
No
4
(April), 1991: pp 391-398
was obtained from each subject and the experimental protocol was
approved by the UCSD Committee on Investigations Involving Human Subjects. No subject was ingesting any drug known to affect glucose or insulin metabolism during the course of the study. Subjects were hospitalized and studied over a 2-week period. During this time they consumed a standardized weight-maintenance diet containing55% carbohydrate, 30% fat, and 15% protein given in three feedings of l/5,2/5, and 2/5 of the total daily calories at 8:00 AM, 12:OOPM, and 590 PM, respectively. Mean energy intake was 2,380 kcalid (range, 2,100 to 3,000). Mean N intake was 14 g/d (range, 12 to 16) and was the same from day to day for each subject. Experimental
Protocol
Hyperinsulinemic, euglycemic clamp technique. The protocol used for the hyperinsulinemic, euglycemic clamp study is summarized in Fig 1 and was similar to those reported in the literature that have used this technique combined with indirect calorimetry.” All studies were performed after a lo- to 12-hour overnight fast. On the morning of the test, the subjects were woken at 6:00 AM and, after voiding, two peripheral venous catheters were inserted. One was placed in an antecubital vein for infusates, the other was inserted in a retrograde fashion into a hand vein for blood
From the Department of Medicine, University of California, San Diego; and the Veterans Administration Medical Center, San Diego, CA. Supported in part by the American Diabetes Association (ADA), the Medical Research Service of the Veterans Administration Medical Center Grants No. DK 38949from the National Institute of Diabetes and Digestive and Kidney Diseases, and Grant No. MO1 RR-00827 from the General Clinical Research Branch, Division of Research Resources, National Institutes of Health. B.G. is the recipient of an ADA California Afiliate Fellowship Grant. A.T. is a Neil Hamilton Fairley Fellow funded by National Health and Medical Research Council OfAustralia. Address reprint requests to Robert R. Henry, MO, VA Medical Center (V-l I I G), 3350 La Jolla Village Dr, San Diego. CA 92161. Copytight 0 1991 by W.B. Saunders Company 0026-0495/91/4004-0010$03.00/0 391
392
THORBURN ET AL
INSULIN GLUCOSE SOMATOSTATIN
CALORIMETRY URINE COLLECTION l
t -90
.
I
I
0
60
BASAL
SERUM UREA N COLLECTION
I
1
120
180
240
minutes
CLAMP Fig 1.
Summary of the study design.
sampling. The hand was kept in a warming device at 60 to 70°C to achieve arterialization of venous blood. After a 90-minute basal period, an intravenous infusion of insulin was begun with a bolus of 6000 pmol/m*, followed by a continuous insulin infusion of 300 pmol/m’/min for 240 minutes. Somatostatin (Bachem, Torrence, CA) was infused at a rate of 0.08 pmol/kg/min to ensure complete suppression of endogenous insulin secretion. Glucose was clamped at 5 mmol/L by a variable infusion of 20% glucose, at a rate determined in a feedback fashion by the measurement of serum glucose at 5-minute intervals. Serum insulin levels were measured twice during the basal period and at 120, 180, and 240 minutes during the clamp. Urea N was measured in serum once during the basal period and after 120 and 240 minutes of clamping to determine whether urea clearance was changing during the study. Indirectcalorimetrytechnique. Rates of G,, F,,, and EE under steady-state basal and clamp conditions were determined by open-circuit indirect calorimetry. A transparent plastic ventilated hood was placed over the subject’s head. To avoid air loss, an airtight jacket attached to the hood was snugly fastened around the upper arms and chest and a slight negative pressure maintained in the hood using a suction fan. The subject lay in a recumbent position and was not permitted to watch television, read, or listen to the radio so that he remained calm while in the hood. The subject was checked every 5 to 10 minutes to ensure he remained awake. Air from a nearby empty room was drawn through the hood at a rate of 35 to 50 L/min. Ventilation at the outlet was measured by a digital turbine volume transducer (Ventilation Measurement Module-l, SensorMedics, Anaheim, CA). A constant fraction of the air flowing out of the hood was dried by condensers (at WC), filtered for dust, continuously withdrawn by pumps, and analyzed = 50 times per second for oxygen and carbon dioxide by analyzers (0, by Magnos 4G and CO, by Uras 3G, Hartmann and Braun, Frankfurt, Germany), which measure differences in O2 and CO, concentration between outflowing and fresh incoming air. These analyzers are sensitive, with an accuracy of 0.02% vol for 0, and 0.004% vol for CO,. Standard gas mixtures of 20.00% 0, and 1.00% CO, were used to calibrate the analyzers before and after each study. Acetone, which has an RQ of 0.75, was burnt in a container inside the hood before and after each study to check the complete system of indirect calorimetry. The measurements of EE were verified from the weight of acetone burned and the stoichiometric equation of acetone oxidation. Measured EE corresponded within 2% of that calculated from acetone stoichiometry. The analogue voltage output of the analyzers and flowmeter were connected to an IBM PC AT via an analogue to digital conversion board (Data Translation, Marlbora. MA). The mean rates of CO2 produced
(VCO,, in Umin) and 0, consumed (Vo,, in Umin) were continuously measured, displayed every minute, and printed every 5 minutes. From these data, the rates of G,,, FOX,and EE were calculated over 20-minute intervals (see Data Analysis) during the basal and clamp periods. Subjects were removed from the hood for 1 hour between 60 and 120 minutes during the clamp. Urinnry N measurements. After completion of the basal and clamp calorimetry measurements and halfway through the clamp at 120 minutes, subjects voided and these urine samples were analyzed for creatinine, urea, and uric acid (see Fig 1). The nitrogen in these compounds was combined to give total urinary N excretion. On half the specimens, total N was also measured on an aliquot of urine by the Kjeldahl method, since this is the method most commonly used by other investigators. The two methods gave comparable results (within 7%) and showed comparable trends. As we were interested to see whether changes occurred in the concentration of these three major components of urinary N, we have reported these results rather than those obtained from Kjeldahl analysis. The rate of urinary N excretion over a 90-minute basal period was also measured on three additional mornings under identical conditions in the subjects to determine the daily variation in urinary N excretion over 4 days. Studies in each subject were completed within a 2-week period. Analytical Methods Blood drawn for glucose was separated by microcentrifuge and serum glucose determined by the glucose oxidase method (YSI 23A, Yellow Springs, OH). Blood for serum insulin was collected in untreated tubes and allowed to clot at room temperature. Serum insulin levels were measured by a specific double-antibody radioimmunoassay.’ Total urinary N was determined by Kjeldahl technique.’ Blood and urinary urea N, creatinine, and uric acid were measured using routine, semiautomated methods (Hitachi 737, Boehringer Mannheim Diagnostic, Indianapolis, IN). The intraassay and interassay coefficients of variation (CV) for urine urea N, uric acid, and creatinine were 1.5% and 1.5%, 0.6% and 1.2%, and 1.4% and 2.3%, respectively. Calculation of Substrate Oxidation Rates and EE The following equations can be derived from the number of liters of 0, consumed and CO2 produced per gram of glucose, fat, or protein.9,‘” VoZ = 0.746 G,, + 2.029 F,, + 0.966 P,,,
(equation 1)
VcoZ = 0.746 G,, + 1.430 F,, + 0.782 P,,,
(equation 2)
where G,, F,,, and P,, represent the oxidation rates of glucose, fat and protein respectively (in g/min), and Vo, and VcoZ are the rates of O2 consumed and CO, produced (in Wmin). The RQ was calculated by dividing Vco, by Voz. If P,, = 6.25 N,
(equation 3)
where N is the urinary N excretion rate (in g/min), then the following equations can be derived when equation 3 is substituted into equations 1 and 2 and the equations solved simultaneously: F,, = 1.67 VoL - 1.67 Vco: - 1.92 N,
(equation 4)
G,, = 4.54 Vco, - 3.20 Vo: - 2.87 N,
(equation 5)
In addition, since EE (in kcalimin) is equal to the following equation: EE = (G,, x 3.74) + (F,, x 9.50) + (P,, x 4.10),
(equation 6)
SUBSTRATE OXIDATION
393
ERRORS AND GLUCOSE CLAMPS
(from 6.4 to 1.6 mg/min), G, and F,, increase by constant amounts (0.014 and 0.010 gimin, respectively) over the range of basal RQ values (0.78 to 0.85). To illustrate these points, observe the dashed lines in Fig 2. At a basal RQ of 0.82, which was a typical value found during our studies (Table 2), a 4.8 mg/min variation in urinary N produces a 15% change in G, (0.091 to 0.105 g/min) (Fig 2A) and an 18% change in F,, (from 0.054 to 0.064 gimin) (Fig 2B). Over the range of basal RQ values (0.78 to 0.85) the variation in basal urinary N resulted in potential errors of 11% to 23% for G, and 16% to 24% for F,,. In contrast, the effect of a 4.8 mg/min variation in urinary N on EE is negligible (I 1%) over the range of basal RQ values (not illustrated). The reason that changes in urinary N excretion have minimal influence on EE is apparent from equation 7; urinary N is a small number that is subtracted from the sum of two much larger numbers. Even a threefold increase in urinary N would decrease EE by only 5% over the range of basal RQ values (0.78 to 0.85).
where 3.74, 9.50, and 4.10 are the calorific values of glucose, fat, and protein in kcal/g, respectively.‘” then substituting equations 3 to 5 into equation 6, EE = 1.11 Vcoz + 3.90 Vo, - 3.34 N.
(equation
7)
Correction of Urinary N for Changes in Urea Nitrogen Clearance If urea excretion increased under hyperinsulinemic clamp conditions, urea N in the blood (BUN) would decrease and total urinary N excretion would increase. This situation would result in an overestimate of the P,, rate and could underestimate G, and F,,. To examine this, P,, was compared with or without correcting the urinary N excretion rate for changes in urea excretion during the glucose clamp. This correction was made with the following equation: Urinary
N, (gimin)
=
-
Change
N during clamp (gimin)
in BUN (mg/L) change
x 0.60 x body weight
(kg)
in time (min) x 1,000 (equation
6)
This equation assumes that the urea pool is 60% of body weight” and does not change over time. NL represents the corrected urinary N excretion rate. All results are expressed as mean + SEM unless otherwise specified. The paired Student’s 1 test was used to make statistical comparisons.
Variation in Basal Hypetinsulinemic Clamp Urinary N Excretion
After 240 minutes of hyperinsulinemic clamping, there was an increase in RQ from 0.82 to 0.97 (Table 2). During this period, insulin levels increased from 29 k 7 to 653 2 50 pmol/L, and a 47% increase in urinary N excretion (6.4 2 2.3 to 9.4 + 2.8 mg/min, P < .05, Table 1) and hence calculated P,, (0.040 to 0.059 gimin, Table 2) occurred. Figure 3 illustrates the effect of this 47% increase in urinary N on G,, and F,,. When urinary N increased from 6.4 to 9.4 mg/min, G,, and F,, decreased by constant amounts (0.009 and 0.006 g/min, respectively) over the range of clamp RQ values (0.95 to LOO). This effect can be observed from the dashed lines illustrated in Fig 3. At a typical clamp RQ value of 0.97 (see Table 2) G,, decreased from 0.259 to 0.250 g/min (Fig 3A), corresponding to a 4% variation in G,, (Fig 3C) when urinary N increased from 6.4 to 9.4 mgimin. Over the range of clamp RQ values, the variation in G,, resulting from a 47% increase in urinary N ranged from 5% (at an RQ of 0.95) to 3% (at an RQ of 1.00).
RESULTS
Variation in Basal Urinary N Excretion Urinary N excretion in the basal state on 4 different days in each subject is listed in Table 1. The mean value for all studies was 6.4 +- 1.6 mg/min (mean t SD). On different days the mean range in the subjects’ urinary N excretion was 4.8 mg/min, with a CV of 36% ? 18%. Figures 2A and 2B illustrate that this day-to-day variation in basal urinary N can produce significant errors in the calculation of G, and F,, rates. Urinary N values corresponding to the mean (6.4 mg/min) and the mean minus the intraindividual range (6.4 - 4.8 = 1.6 mgimin) were chosen to illustrate this point. From Table 1 it is apparent that the values of 6.4 and 1.6 mg/min are realistic basal urinary N excretion rates. When urinary N decreases by 4.8 mg/min
Table 1. Urinary N (mg/min) Under Basal Conditions [Over a 90-Minute Period) on 4 Different Days and During a 300-pmol/m’/min
Clamp (From 120 to 240 Minutes)
Basal .~ Subject
DW
Day
1
2
3
W
cv
Dw 4
Clamp
Meall
WI
Dw 1
1
8.0
4.0
6.1
4.0
5.5
35
10.1
2
6.1
9.0
8.0
7.0
7.5
17
14.1
3
8.0
4.0
5.0
10.1
6.8
41
7.0
4
7.0
6.1
8.0
10.1
7.8
22
5.0
5
8.0
6.1
7.0
10.1
7.8
12.0
6
7.0
10.1
6.1
6.1
7.3
22 26
7
1.0
6.1
3.0
3.0
3.3
64
9.0
8
6.1
1.0
6.1
9.0
5.5
60
9.0
6.4 + 1.6
36 k 18
9.0
Mean k SD
6.4 _f 2.3
9.4 r 2.8
THORBURN ET AL
0.16 1
UN-l.6 UN=6.4
clamp RQ values (0.95 to 1.00). The effect of a 47% increase in urinary N on EE was also negligible ( < 2%) over the range of clamp RQ values (not illustrated). Therefore, increases in urinary N excretion during the clamp have a modest effect on G,, and much less effect on F,, and EE. Correcting Urinary N Excretion for Changes in Urea N Clearance During the Hyperindinemic Clamp
0.06 0.06 0.04 I 0.77
0.76
0.79
0.60
0.61
0.62
0.63
0.64
0.65
28 0.06 1
UN=1.6 UN=6.4 0.77
25
0.76
0.79
0.60
0.81
0.82
0.83
0.84
1
0.85
n
Fox
Gox 10: 0.77
I
1
I
0.78
0.79
0.80
I
0.81
1
0.82
I
0.83
I
0.84
I
0.85
In this study, urea constituted approximately 93% of the total N in urine. In the basal state, urea N excretion was 6.04 mg/min, uric acid was 0.13 mg of N/min, and creatinine was 0.27 mg of Nimin. During the clamp, the excretion of urea N increased to 8.85 mg/min, uric acid to 0.19 mg/min, and creatinine to 0.40 mg/min. Therefore, most of the 3 mg/min increase in urinary N excretion during the clamp (Table 1) was due to an increase in the excretion rate of urea (94%) with a very small increase in uric acid (2%) and creatinine (4%). Although creatinine and uric acid are not end products of protein oxidation, we included them in our measurement of urinary N-derived protein oxidation since total urinary N is what most investigators use to calculate protein oxidation and we wished to mimic the methods used by others. An increase in urea N clearance (37.5 2 6.7 to 75.2 2 12.4 mL/min,P < .025) (Fig 4) was evident from the decrease in serum urea N (10.1 -C 0.8 to 8.0 +_0.7 mmol/L, P < .05) and increase in urine urea N excretion during the clamp (0.36 k 0.06 to 0.57 + 0.08 mmol/min, P < .05). Urine output increased twofold to threefold from the basal to the end of the clamp period (1.4 2 0.2 to 3.7 2 0.8 ml/mitt, P < .025). To determine if changes in the rate of urine formation during the glucose clamp influence urea clearance and explain some of the variability in urinary N excretion, the rate of urinary N excretion in each basal and clamp study was compared with its corresponding urine volume formation and is shown in Fig 5. A significant correlation (r = .49, P < .Ol) existed between these parameters. To account for changes in urea clearance and provide a better estimate of protein oxidation from urinary N measurements, urinary N values were corrected for changes in serum urea N concentrations (equation 8). Correcting urinary N excretion for changes in urea clearance over the
Basal RQ Fig 2. Effect of a 4.8 mg/min variation in urinary nitrogen (UN) excretion (6.4 Y 1.6 mg/min) on (A) glucose oxidation and (B) fat oxidation over a range of basal RQ values in a normal subject. (C) Variation in glucose (G,,) and fat (F,,) oxidation as a percentage. Dashed lines highlight the results obtained at a typical basal FIG value of 0.82.
F,, values during the clamp were almost completely suppressed up to RQ values of 0.97, above which F,, rates were calculated as negative values. This meant that any error in F, led to a minimal change in F,,, but corresponded to a large percent variation in F,,. For example, at a typical clamp RQ value of 0.97, F, decreased from 0.001 to -0.005 g/min (Fig 3B), equivalent to a 100% variation in F, (Fig 3C) when urinary N increased from 6.4 to 9.4 mg/min. The variation in F, ranged from 75% to 100% over the range of
Table 2. Calorimetry Measurements
Over ZO-Minute Intervals During
Basal and Clamp Periods RQ
vco,
vo,
PO.*
Limin
Vco,No,
G,,
Fox
glmin
Basal (min) -60
to -40
0.82
0.200
0.243
0.040
0.112
0.060
-40
to -20
0.83
0.209
0.251
0.040
0.127
0.058
-20
to 0
0.82
0.219
0.266
0.040
0.125
0.067
Clamp
(min)
140 to 160
0.95
0.242
0.254
0.059
0.259
0.002
160 to 180
0.95
0.238
0.250
0.059
0.254
0.002
180 to 200
0.96
0.244
0.255
0.059
0.265
0.000
200 to 220
0.96
0.240
0.251
0.059
0.259
0.000
0.97
0.254
0.259
0.059
0.297
220 to 240 NOTE. *Protein
Results
are means
oxidation
of eight subjects.
= 6.25 x urinary
N excretion
rate (g/min).
-0.010
SUBSTRATE
OXIDATION
ERRORS
AND
GLUCOSE
CLAMPS
395
3A z
1
0.30
-iE
s
0.28
H
-we+/
m ,,zfj-t( _----
$j
E UN=6.4 _
P
0.24-)
UN=9.4
DISCUSSION
In most combined calorimetry-clamp studies, P,, determined from the urinary N excretion rate, is assumed to be constant and this assumption is thought not to influence the However, these assumptions are based calculation of G0X,2-5 on little evidence, since information on the way different experimental manipulations affect P,, is scanty.‘” The current study supports the earlier results of Tappy et al,” who clearly demonstrated that hyperinsulinemic glucose clamps alter not only urine flow, urea pool size, and clearance, but urinary N excretion and substrate oxidation rates as well. Our study now adds the new finding that large day-to-day variability in basal urinary N excretion occurs and can lead
/i
_ 0.95
0.94
0.96
0.97
in urea clearance during the basal period had little effect on P,, and G,, values. However, during hyperinsulinemic clamp conditions, this correction eliminated the increase observed in P,, determined from urinary N measurements and increased clamp G, values by 0.013 g/min (4% to 5%).
0.98
0.99
1.00
38
16 -
UN=6.4 14 -
“N-9.4
12 10 -
_I.“”
.
I
8-
0.94 0.95 0.96 0.97 0.98 0.99 1.00
1.0
-
0.8
-
0.6
-
0.4
-
0.2
-
3c --yyJ--
Fox
I I E
40
5 k P
20 __ 0.94
I
1 0.95
c
I 0.96
0.97
0.98
0.99
Gox
1.00
Clamp RQ
O160 -
Fig 3. Effect of a 47% increase in urinary nitrogen (UN) excretion (6.4 Y 9.4 mg/min) on (A) glucose oxidation and (B) fat oxidation over a range of clamp Ml values in a normal subject. (C) Variation in glucose (G..j and fat (FJ oxidation as a percentage. Dashed lines highlight the results obtained in a typical clamp RQ value of 0.97.
basal period had negligible effect on P, values and did not reduce the large intraindividual variability observed in the basal state. Figure 6 illustrates that during the clamp, correcting P,, for the change in serum urea N reduced P,, by 46% (P < .OOS).The mean corrected clamp P,, value was 20% less than P, values obtained during the basal period. The effect of correcting for changes in urea clearance on rates of P,, and G, in both the basal state and under hyperinsulinemic clamp conditions are illustrated in Fig 7. From this Figure, it can be seen that correcting for changes
140
-
120
-
100
-
80
-
60
-
40
-
20
-
O-
BASAL
CLAMP
Fig 4. Serum urea nitrogen, urine urea nitrogen formation, and urea nitrogen clearance in eight normal subjects under basal and hyperinsulinamic clamp conditions. Urea ckarance = (urine urea nitrogen formation/serum urea nitrogen) x 1,000.
THOREURN
396
-40
-20
160
0
BASAL
0
(./A1”“‘11”1”“11~
0
0002
0.004
0.006 URINARY
0.006
0010
NITROGEN
0.012
0014
0.016
0.016
to substantial errors in basal substrate oxidation rates. We found that basal urinary N excretion is not a constant; it varies on average by 4.8 mg/min from day to day (CV 36%), which is in contrast to more direct measures of P, using labeled leucine turnover studies, which have shown basal P,, to be constant.13 This variation would be of little concern if it did not influence the calculation of G, and F,,, but we have shown that it does, causing changes of 11% to 23%
Pox
240
(min)
Fig 7. Rates of G,, and P,, calculated from urinary N excretion with and without correcting urinary N for changes in urea clearance during the basal state and hyperinsulinemic clamp conditions in eight healthy subjects.
(g/min)
Fig 5. The rate of urinary nitrogen excretion and urine volume formation in eight subjects under four basal conditions (0) and a hyperinsulinemic, euglycemic clamp (0).
BASAL
AL
CLAMP Time
.I;
200
ET
CLAMP Pox
CLAMP Pox WITH BUN CORRECTION
Fig 6. Protein oxidation (g/min) in eight normal subjects calculated in the basal state and during the hyperinsulinemic, euglycemic clamp using the urinary N excretion rate (x6.25) with and wkhout correcting for changes in serum urea N concentration.
and 16% to 24% in basal G, and F,, rates, respectively. As reported previously,” we also found that P, varied during the hyperinsulinemic, euglycemic clamp as calculated from urinary N measurements, increasing by 47%. These urinary N-derived values for protein oxidation contrast with measures of P,, based on Y- or ‘H-leucine turnover, which have demonstrated that P, decreases by 28% to 60% under hyperinsulinemic clamp conditions identical to ours.‘3.‘4 If one assumes that these radiolabeled leucine studies accurately estimate P,,, which they probably do given the anabolic effect insulin has on protein metabolism,‘5 then urinary N-derived P,, values as determined in this study will underestimate G,, rates by up to 0.02 g/min (-10%). However, if urinary N measurements are corrected for changes in urea clearance during hyperinsulinemic glucose clamps, the recalculated P, value tends toward those reported using labeled leucine, reducing the error to less than 5% and providing a better estimate of G,, under ciamp conditions. Urinary N may be a poor index of P, in the basal state, since not all protein results in the liberation of equivalent amounts of N. Although the total amount of N derived from dietary protein intake did not vary from day to day for each subject, it is feasible that variations in the type of antecedent protein intake from day to day were responsible for some of the variation in the amount of urinary N excreted. For example, Livesay and Elial have recently shown that ingestion of different dietary proteins can theoretically result in a +14% to -38% variation from the accepted general value of 6.0 L/g of urinary N (or 0.966 L/g of protein used in equation 1). From this information it would be prudent to recommend that the type of protein, as well as the total amount of protein consumed before studies, be kept consistent during comparative study periods. Increased urinary N excretion rates during hyperinsulinemic clamps is not a widely appreciated phenomena and could be the result of a number of factors. Since urea forms 80% to 90% of all N in urine, the most likely explanation for the increase in urinary N excretion during the clamp is an
SUBSTRATE OXIDATION
397
ERRORS AND GLUCOSE CLAMPS
increase in urea clearance. In support of this, our results clearly show that urea clearance increases during the clamp. Although increased insulin levels might play a role in altering sodium retention, and hence changes in volume and urea clearance, Thiebaud et al” have shown no difference in urinary N during hyperinsulinemic glucose clamps over a range of serum insulin concentration from 455 to 8,128 pmol/L. Increased urea clearance is more likely to be due to an increase in urine output, since augmented urine flow is known to result in an increase in the fractional extraction rate of urea.‘8.‘9 This occurs because urea reabsorption is a passive process completely dependent on the rate of water reabsorption, which establishes the diffusion gradient within the kidney tubules.” This effect would also explain the positive correlation we observed between urine output and urinary N excretion. The increase in urine output during the clamp was largely a result of the large volume of fluid (-2 L), which had to be infused intravenously into the subjects in order to administer the various infusates during the hyperinsulinemic clamp. Therefore, it appears that at least some of the increase in urinary N under hyperinsulinemic clamp conditions can be explained by variations in tluid intake that cause changes in urea clearance. In the basal state, changes in urea clearance were smaller and did not explain the large intraindividual variation in urinary N under these conditions. The short urine collection period (90 minutes for basal and 120 minutes for clamp) would also contribute to the variability in urinary N. These periods were chosen to mimic urine collection times typically used by investigators during hyperinsulinemic clamps. Although we did not set out to determine the etiology of variations in urinary N, we have shown that this variability causes substantial errors in G,, and F,, in the basal state and errors in G,, under hyperinsulinemic clamp conditions when an erroneous urinary N value is used to calculate substrate oxidation rates. Our findings strongly support a recent report by Tappy et al,” which showed that without correcting for changes in the urea N pool, urinary N excretion may substantially overestimate P,. In contrast, variations in urinary N do not significantly affect EE. This has been previously demonstrated by Bursztein et al’ and, more recently, by Ferrannini.‘” Bursztein’s report is commonly used to substantiate the incorrect claim that G, is unaffected by P,. However, this report only illustrates that EE is unaffected by P,,. It needs to be emphasized that the present study was not designed to determine the most accurate method of quantitating P,,. Rather, the aim was to determine whether urinary N excretion is variable when measured in a manner common to most investigators combining indirect calorimetry with hyperinsulinemic clamps and to quantitate what effect this variability would have on substrate oxidation rates and EE. This study does not indicate that P, per se is variable or influenced by the hyperinsulinemic glucose clamp procedure. It simply demonstrates that urinary N is a highly variable index of P, under both basal and glucose clamp conditions. Furthermore, the methods most com-
monly used to quantitate urinary N content during glucose clamps tend to be inaccurate, since they either include the measurement of N in substances not associated with P, (Kjeldahl technique) or fail to measure some of them that are (eg, ammonia). Further studies are needed to determine methods by which basal P, may be more accurately determined from urinary N measurements. However, during glucose clamp studies, errors that result from calculating P, directly from urinary N are smaller and at least partially correctable. The first precaution is to keep intravenous fluid intake, and therefore urinary volume, relatively constant during comparative study periods to keep variations in urea clearance to a minimum. Second, corrections can be made for changes in urea clearance by measuring BUN (equation 8). It should be noted that correction of urinary N excretion only for changes in urea N clearance assumes that the decrease in BUN that occurred in all subjects in this study was due solely to increased urea N clearance during glucose clamping. Since both decreased urea production and increased urea volume of distribution could also account for some of the decrease in BUN that occurred, this correction formula would tend to overestimate the contribution of increased urea N clearance. When we made this correction, urinary N during the clamp decreased by 4 mg/min. Others have also found that BUN decreases under hyperinsulinemic clamp conditions’3.20.21or after a mea1*2~23 and have often corrected urinary N values accordingly.‘2~“~22”In our study, correcting clamp urinary N for these changes in urea clearance reduced P,, by approximately 50% and resulted in a 20% lower P,, value than basal. Therefore, correcting urinary N for changes in urea clearance provides a better estimate of P,, which, although not perfect, results in smaller errors in the calculation of G,, during clamping. This study has shown that the measurement of basal urinary N is highly variable and can result in significant errors in G,, and F,,, but not EE. During hyperinsulinemic, euglycemic clamp conditions, urinary N increases by 47%, rather than decreasing as expected, due in large part to increases in urea clearance. This change has minimal effect on F,, and EE, but underestimates G,, by up to approximately 5% compared with values calculated from basal urinary N values and by approximately 10% compared with values derived from leucine turnover studies. This study should alert investigators that urinary N measurements can be highly variable, both in the basal state and during hyperinsulinemic glucose clamping, leading to substantial errors in substrate oxidation rates as determined by indirect calorimetry. It is clear that improved methods to accurately quantitate in vivo P,, during glucose clamp-calorimetry studies need to be developed.
ACKNOWLEDGMENT We would like to thank the nursing staff of the Special Diagnostic and Treatment Unit, San Diego Veterans Administration Medical Center, John Murphy and Ken Winders for excellent technical assistance, and Cleon Tate for typing the manuscript.
398
THORBURN ET AL
REFERENCES
1. Golay A, Schutz Y, Felber JP, et al: Lack of thermogenic response to glucose/insulin infusion in diabetic obese subjects. Int J Obes l&107-116,1986 2. DeFronzo RA, Jacot E, Jequier E, et al: The effect of insulin on the disposal of intravenous glucose. Results from indirect calorimetry and hepatic and femoralvenous catheterization. Diabetes 30:1000-1007,198l 3. Felber JP, Thiebaud D, Maeder E, et al: Effect of somatostatin-induced insulinopenia on glucose oxidation in man. Diabetologia 25:325-330,1983 4. Jacot E, DeFronzo RA, Jequier E, et al: The effect of hyperglycemia, hyperinsulinemia, and route of glucose administration on glucose oxidation and glucose storage. Metabolism 31:922930,1982 5. Thiebaud D, Jacot E, DeFronzo RA, et al: The effect of graded doses of insulin on total glucose uptake, glucose oxidation, and glucose storage in man. Diabetes 31:957-963,1982 6. Bursztein S, Saphar P, Glaser P, et al: Determination of energy metabolism from respiratory functions alone. J Appl Physiol42:117-119, 1977 7. Desbuquois B, Aurbach GA: Use of polyethylene glycol to separate free and antibody bound peptide hormones in radioimmunoassay. J Clin Endocrinol Metab 33:732-738,197l 8. DiGiorgio J: Nonprotein nitrogenous constituents, in Henry RJ, Connor DC, Winkelman JW (eds): Clinical Chemistry: Principles and Technics. Hagerstown, MD, Harper & Row, 1974, pp 503-563 9. Consolazio CF, Johnson RE, Pecora LJ: Physiological Measurements of Metabolic Functions in Man. New York, NY, McGraw-Hill, 1963, pp 313-339 10. Ferrannini E: The theoretical bases of indirect calorimetry: A review. Metabolism 37:287-301,1988 11. Jequier E, Acheson K, Schutz Y: Assessment of energy expenditure and fuel utilization in man. Ann Rev Nutr 7:187-208, 1987 12. Tappy L. Owen OE, Boden G: Effect of hyperinsulinemia on
urea pool size and substrate oxidation rates. Diabetes 37:12121216,1988 13. Castelhno P, Luzi L, Simonson DC, et al: Effect of insulin and plasma amino acid concentrations on leucine metabolism in man. Role of substrate availability on estimates of whole body protein synthesis. J Clin Invest 80:1784-1793,1987 14. Tessari P, Nosadini R, Trevisan R, et al: Defective suppression by insulin of leucine-carbon appearance and oxidation in type I, insulin-dependent diabetes mellitus. Evidence for insulin resistance involving glucose and amino acid metabolism. J Clin Invest 77:1797-1804,1986 15. Jefferson LS: Role of insulin in the regulation of protein synthesis. Diabetes 29:487-496,198O 16. Livesay G, Elia M: Estimation of energy expenditure, net carbohydrate utilization, and net fat oxidation and synthesis by indirect calorimetry: Evaluation of errors with special reference to the detailed composition of fuels. Am J Clin Nutr 47:608-628,1988 17. Thiebaud D, Jacot E, DeFronzo RA, et al: The effect of graded doses of insulin on total glucose uptake, glucose oxidation, and glucose storage in man. Diabetes 31:957-963, 1982 18. Sullivan LP, Grantham JJ: Physiology of the Kidney. Philadelphia, PA, Lea & Febiger, 1982, pp 109-111 19. Vander AJ: Renal Physiology. New York, NY, McGrawHill, 1980, pp 44-46 20. Acheson KJ, Ravussin E, Wahren J, et al: Thermic effect of glucose in man. Obligatory and facultative thermogenesis. J Chn Invest 74:1572-1580,1984 21. Owen OE, Trapp VE, Reichard GA, et al: Effects of therapy on the nature and quantity of fuels oxidised during diabetic ketoacidosis. Diabetes 29:365-372, 1980 22. Flatt JP, Ravussin E, Acheson KJ, et al: Effects of dietary fat on postprandial substrate oxidation and on carbohydrate and fat balances. J Clin Invest 76:1019-1024, 1985 23. Acheson KJ, Flatt JP, Jequier E: Glycogen synthesis versus lipogenesis after a 500 gram carbohydrate meal in man. Metabolism 31:1234-1240, 1982
