- Email: [email protected]
The response to intravenous glucose of patients on maintenance hemodialysis: Effects of dialysis
The Response to Intravenous Glucose of Patients on Maintenance Hemodialysis: Effects of Dialysis Eleuterio Ferrannini, with the technical
assistance
Alessandro
To learn whether a single dialysis can acutely improve the intravenous glucose tolerance (i.v.GTT) of chronically dialyzed patients, a standard i.v.GTT was performed on 10 nonobese uremic subjects on maintenance hemodialysis for 27 * 9 (mean ? SEM) mo, and on a control group of 13 normal subjects. The uremic patients were tested first 0.2-17 (range) hr, and then 65-109 hr, from last dialysis. In the uremic sera, plasma glucose was analyzed by 4 methods; 2 reducing (neocuproine and ferricyanide) and 2 enzymatic (hexokinase and glucose oxidase). The reducing methods markedly overestimated plasma glucose concentration because of the presence of nonglucose, reducing substances (notably, creatinine). This interference was significantly cut down by dialysis. A single dialysis, on the other hand, failed to improve the glucose fractional decay rate (KG) computed from the glucose oxidase data (1.69 + 0.2%/min before and 1.35 + 0.1 after dialysis, versus 1.47 + 0.1 of the normal subjects). The same conclusion was derived from the data measured by the other 3 methods of glucose assay. Fasting plasma insulin concentrations were, on average, above normal (5.5 + 0.6 NU/ml) both before (12.3 k 2.7, p < 0.05) and after (17.2 k 3.5, p < 0.01) a single dialysis. Likewise, the area under the glucoseinduced plasma insulin curve was significantly greater than normal (1.46 ? 0.21 mu/ml. min) both before (2.26 f 0.34, p < 0.05), and after (2.86 + 0.43, p < 0.01) dialysis. A single dialysis had little effect on either basal or glucose-stimulated insulin release, and no significant difference in the insulinogenic index (insulin area/glucose area) was found between the control and the uremic group in either test. Insulin response was not correlated with KG, whereas it was significantly associated with higher triglyceride levels. Creatinine, urea or methylguanidfne did not appear to have any influence on KG, but lower serum potassium levels were significantly associated with poorer i.v.GTT’s. Plasma calcium bore a reciprocal relation to the insulinogenic index. Chronically dialyzed subjects show some degree of tissue insulin resistance, which a single dialysis fails to correct. Electrolyte disturbances may play a role in this metabolic derangement.
R
ENAL FAILURE is frequently associated with a glucose intolerance that is distinct from that of diabetes mellitus.’ Although hardly ever requiring any treatment, this abnormality is believed to shorten the lifespan of uremic subjects on prolonged hemodialysis through the sequence hyperglycemia-hyperinsulinemiahyperlipidemia-accelerated atherosclerosisfatal vascular accidents.* Mefabolism, Vol. 28, No. 2 (February), 1979
Pilo, Mario Tuoni,
of Giuseppe Buzzigoli and Claudia Boni
Whether hemodialysis can improve glucose tolerance is, at present, controversial. Early studies,‘-” in which glucose was measured by nonspecific, reducing methods, have shown that hemodialysis corrects carbohydrate intolerance to some extent. Reducing methods, however, yield falsely raised glucose values for uremic sera, in which reducing substances other than glucose (e.g., creatinine) are present at concentrations much higher than normal.” In this way, hemodialysis might appear to lead to an improvement of glucose tolerance by removing these nonglucose, reducing substances. In a recent study,13 using a glucose-specific assay method, the oral glucose tolerance of uremic patients, either undialyzed or on maintenance hemodialysis, was not found to be improved by dialytic treatment to any relevant degree. Davidson and coworkers,lJ on the other hand, also using a glucose-specific assay, have reported that glucose KG values are significantly higher after dialysis. We thought it would be of interest to investigate the acute effects of dialysis on the response to intravenous glucose in uremic subjects on maintenance hemodialysis with the use of 4 methods of glucose assay, 2 reducing (neocuproine and ferricyanide), and 2 enzymatic (hexokinase and glucose oxidase). In this paper, we report that, in our group of patients, a single dialysis failed to improve the i.v.GTT whichever method of glucose measurement was used. The influence on plasma glucose disappearance of factors such as insulin, toxic metabolites. and electrolytes is also discussed. From the C.N.R. Clinical Physiolog_v Laboratory and the 2nd Medical Clinic, University of Pisa. Ita1.v. Received for publication February 3, 1978. Supported in part by USPHS Grant Nol-AM-8-0707. bt, lhe Chronic Uremia-Artificial Kidney Program of the National Institutes of’ Arthritis. Metabolism. and Digestive Diseases, National Institutes of Health, Bethesda. Md. Address reprint requests to Dr. E. Ferrannini, C.N. R. Clinical Physiology Laboratory, Via Savi 8, 56100 Piss. Italy. @ I979 by Grune & Stratton, I~C. 00260495/78/2802-0005$02.00/#
125
126
FERRANNINI
ET AL.
Table 1. Clinical Data of the Study Subjects
Case Number
Sax
Age (vr)
Weight’ (kg)
Height (cm1
Percent ldealt BodyWt
Underlying+ Disease
Durationof Dlalytlc Treatment bno)
Uremic Subjects 1
M
40
68
164
100
2
F
38
45
166
72
3
M
49
80
176
104
CPN
36
4
M
31
62
170
a7
CGN
15
5
M
33
75
173
102
CGN
1
6
M
28
47
170
68
CGN
7
F
37
58
156
103
CGN
8
M
56
63
158
97
CPN
90
9
M
66
77
165
113
CGN
28
10
M
44
74
177
95
CGN
42
42
65
168
94
27
4
4
2
5
9
99
SEM
CGN CGN
4 2
54
Healthy Subjects 1
F
30
59
162
2
F
54
43
158
69
3
M
36
78
173
107
4
F
35
56
168
89
5
F
35
58
167
93
6
F
16
41
155
84
7
F
35
63
157
111
8
F
57
48
158
77
9
M
32
75
170
105 104
10
F
55
59
147
11
F
40
63
168
94
12
F
27
52
159
95
13
F
46
62
159
101
38
58
162
94
3
3
2
3
Mean SEM
*In the uremic subjects. mean of pre- and postdialysis weight. The average weight loss after a standard 5-hr hemodialysis was about 2 kg. t Society of Actuaries Ed: Build and Blood Pressure Study. vol I. Chicago, 1959. p 16. SCGN = chronic glomerulonephntis; CPN = chronic pyelonephritis.
MATERIALS AND METHODS Study subjects. Ten uremic subjects on maintenance hemodialysis were studied; their pertinent clinical data are given in Table I. Criteria for inclusion in the study were that patients were of normal body weight, had negative family histories of diabetes mellitus, showed no evidence of gastrointestinal or endocrine disorders, had no clinical, laboratory or radiologic signs of hyperparathyroidism. that there was no known glucose intolerance before the onset of renal failure, no intercurrent illness, anorexia, nausea, vomiting, or recent change in body weight. All patients were in good general condition, their hematocrit was about 25% on average, and they were taking no medication such as antihypertensive, diuretic, steroid, or oral contraceptive drugs. They were on a free diet, with at least 200 g of carbohydrate daily (the average Italian diet consists of 43%-48% carbohydrates, 12%-l?‘% proteins, and 35%-40% fat). Salt intake was unrestricted. All patients had been on a thrice-weekly, 5-hr hemodialysis schedule for an average of 27 mo. Hemodialysis was performed using Kiil hollow-fiber dialyzers. The dialysate
was free of glucose, and contained (as mEq/ 1) sodium (I 32), potassium (2.2), calcium (3.4), magnesium (I .O), chloride (102). and acetate (38.8). Thirteen age-matched, healthy volunteers (Table I), having normal body weight and rlegative family histories of diabetes mellitus, served as the cohtrol group. These controls were not sex-matched; this, however, was not regarded as an important source of bias, because good evidence indicates that sex per se has no effect on gilucose tolerance.” All the subjects studied gave their informed consent to the investigation. Experimental protocol. Each normal subject received one. i.v.GTT, whereas each uremic patient was tested first a short time (range 0.2-17 hr. “post-dialysis”) from last dialysis, and then after a longer time (65-109 hr, “predialysis”), the order of the two studies being randomized. All studies were performed in the fasted (8-14 hr) state. A plastic cannula was inserted in an antecubital vein and kept patent by intermittent irrigation with isotonic saline. Fasting blood was obtained for glucose and immunoreactive insulin (IRI) determination. In thb uremic subjects, blood for measuring creatinine, nonproteic hitrogen, methylguanidine,
INTRAVENOUS
total
GLUCOSE EFFECT ON DIALYSIS
proteins,
calcium
triglyceride,
and chloride
cholesterol,
was also taken.
PATIENTS
sodium, After
127
potassium,
minus-basal
a stabilization
period of IO-1 5 mitt, 0.33 g of glucose per kg of body weight (as a 40% solution) period.
was injected
intravenously
were computed
and blood samples were taken 2.5, 5, 7.5, 10, 15. 20,
IRI determination. drawn
For these latter
into tubes containing
plasma
was promptly
separated
and sodium
The insulinogenic
blood was fluoride;
of IA to GA.
the
and stored at -20°C
until
procedures.
(azotemia),
total
plasma
terol were measured was analyzed et al.‘”
Creatinine, proteins,
triglyceride.
by standard
methods;
by the chromatographic
Electrolytes
spectrometry.
were
Plasma
measured
glucose methods:
mated on a Technicon
Autoanalyzer,”
also automated (hk).
Glucose was
on a Technicon
automated
Analyzer,19
and
(D)
Analyzer.
measured
normals.
(A)
on glucose
by
the
in addition,
glucose
plasma
method.
following
5.6% (nc), 2.8% (fe),
errors:
(gx). Plasma
concentration
Using
charcoal
centrifugation
bound from free insulin. coethcients the
at
auto-
(B) ferricyanide
(fe),
(gx),
basal
so that
(3-25
coefficient
Rate
In
Data
analysis.
Glucose
fractional
by monoexponential
least squares fit) of either
the absolute
(IRI)
variation
in the range of samples
in the above
disappearance (KG)
glucose
between
analysis
the
intercept
and the standard
by the
the means of unpaired variances
was performed, and
error
were
that
of
by the the stan-
regression
of the estimate
The comparison
a regression
line
glucose in each subject.
(SEE)
between two regression
line and the identity
line. was
by the t test.
Dialysis effectively reduced the plasma concentration of creatinine, nonproteic nitrogen, methylguanidine, and potassium in nearly all the uremic patients (Table 2). Total plasma protein concentration was found to be significantly raised at the shorter distance from dialysis, as a result, most probably, of the plasma volume reduction that occurs with dialysis. No patient had raised serum cholesterol, and dialysis was without effect on this. In cases 1, 3,
separating
interpolation
the
RESULTS
range. was computed
regression
by
and 3.0%
Plasma
to
eight
insulin
fell
system
of the
of the respon-
mean values were compared
we had the
for
area inclusive
by the t test, or where
of
performed
glucose
values.
as the ratio
F) at the 5% level of significance.
were also computed.
using dextran-coated
readings
deviation
lines. or between
5.8% (hk).
pU/ml).
all insulin
linear
plasma
The inter- and intra-assay
concentrations
test.*’
When
on a Beckman
method.
3,000Xg
Behrens’ dard
(C) hexokiReaction
was also measured
were 14% and 7%, respectively,*’
were diluted
by
(nc).
of immunoreactive
by radioimmunoassay,
and
in duplicate
these methods.
was measured
Pre- and postdialysis
(Fisher’s
base-line
estimate
secretory
upon plasma
different
oxidase
the neocuproine
insulin
area-under-curve
as the slope (S) of the regression
paired t test. The differences
absorption
IO to
of the experimen-
was then obtained*’ Another
insulin
data were analyzed
subjects,
glucose
the
system of Giovannetti
8600
oxidase
lo In the normal
was obtained
by atomic neocuproine
LKB
of
stimulus of plasma
taken from
of the respective
The value of the insulin
siveness
and choles-
Autoanalyzer.‘*
a
nitrogen
methylguanidine
was assayed
each of the following
nase
nonproteic
(IA)
integration
index (IG)
basal was also calculated.
assayed. Anal~~tical
and the insulin
by trapezoidal
tal curves after subtraction
for glucose and
measurements,
EDTA
plasma glucose readings
injection.
The glucose (GA)
over a 2-min
25, 30, 35, 40, SO, and 60 min after injection
(KG,)
60 min after
rate’” (using
a
or the absolute-
Table 2. Blood Chemistry in 10 Uremic Patients Before (bJand After la) Dialysis
1 --_---
2
3
4
5
6
7 --____
8
9
10
Mean + SEM
babababababababababa Hours
afterdlalysls
b
15
72
17
109
13
109
13
70
1
70
1
68 02 68 02 68 02 68 02
128
59
67
62
203
100
146
66
156
82
139
.75
144
51
65
serum cremmne img/dll Plasma
72
196
82
170
59
guan,d,ne
l,,gidlI (mg/dU
hst,ng
71-06
64
60
70
63
168
92
67
45
72
72
72
46
120
44
124
64
136
64
88
136
221
94
142
$5
220
125
189
69
161
91
205
100
17,
50
216
87
186
82
176
56
191
98t
11.
7 3
112
60
59
72
165
42
35
93
107
134
120
254
206
257
240
202
240
39
416
123
73
74
70
62
64
7.9
60
76
90
105
70
106
$6
79
72
8,
82
75
72
106
120
130
200
130
200
200
180
200
115
145
130
115
166
190
120
115
190
200
120
145
154
+ I,
157;
i
11,133
*s*
69
f
851
9 7
lAJ/rnli
12,
172
135
plasma
protems Serum
(g/dlk
73104’
T35*o!i
cholesterol
img/dll Serum
12
tnglycer&des
lmg/dll Sod,““? Potasswn
h&,/II ImEq/l)
125
195
100
115
200
170
90
100
150
50
100
25
150
25
136
50
345
360
120
80
132
135
130
146
131
143
135
142
140
125
137
140
136
132
130
141
136
135
134
143
136
138
43
40
5 8
5
52
102
51
82
40
65
48
50
37
86
41
62
43
50
35
64+06t 42
1 53
Ca,c,um
hEq/l,
44
45
39
46
43
26
25
44
40
73
50
70
55
67
45
69
33
2,
60
6,
Chlorode
ImEq/ll
95
94
106
96
107
$6
103
97
93
99
104
95
93
91
98
94
102
9,
92
$5
005
+!J ,’ 001
tp
114
plasma
ll,S”ll”
‘P /
1561
61~-2
methyl-
Azotem~a
Toia,
188
a
77: 51
0001
$9
27 t2
13611 441-02
tO2’ *
2
53106 96
+
,
128
FERRANNINI
ET AL.
400_
10 UREMIC PATIENTS BEFORE DIALYSIS
!I
t_ 0 0 .--.
neocuproine 0 ferricyanide D hexokinase glucose oxidase
Fig. 1. Plasma glucose concentrations (mean ? SEY) after intravenous glucose (0.33 g/kg) in 10 uremic subjects on chronic hemodialysis 66-109 hr after last dialysis (“before dialysis”), as obtained by 4 methods of glucose assay.
100,
and 9, on the other hand, elevated serum triglyceride was found on either or both occasion(s). The very low triglyceride concentration measured postdialysis in cases 5-8 was probably the effect of antecedent heparinization; this is a common finding in the early postdialytic period.25 Dialysis, however, did not change mean triglyceride concentration significantly. Minor changes were seen in the sodium and chloride levels, whereas total plasma calcium was significantly higher postdialysis, presumably owing to the concomitant increase in plasma protein concentration. In fact, when the pre- and
postdiaiysis values were pooled, it was found that calcium and proteins changed consensually (y = 0.73 + 0.7x, r = 0.7, p < 0.01). Figure 1 shows the plasma glucose concentrations after the i.v. glucose load in the uremic group, as measured by the 4 methods of glucose assay: the 2 reducing methods, of which the ferricyanide gave the higher readings, overestimated the glucose concentration as read by the enzymatic assays all through the predialysis curve. This systematic error, or bias, of the reducing versus the specific methods was also evident in the postdialysis test (Fig. 2), but was
lo UREMIC PATIENTS AFTER DIALYSIS 0 0 -
Plasma glucose conFig. 2. centrations (mean 2 SEM) after intravenous glucose (0.33 g/kg) in 10 chronically dialyzed subjects 0.2-17 hr after last dialysis (“after dialysis”), as obtained by 4 methods of gluCOSe assay.
6 2.5 j
1.5 10
1
20 1
I
30
minutes
I
neocuproine o ferricyanide a hexokinase glucose oxidase
40 I
50
60 1
INTRAVENOUS
GLUCOSE EFFECT ON DIALYSIS
129
PATIENTS
60
1
0
20 54erwn80di2nine
24
‘t&Al
Plot of the mean difference Fig. 3. nide and the glucose oxidase reading lated for each uremic subject, against creatinine value. Pre- and postdialysis The equation of the regression line is: 0.74. p i 0.001.
between the ferricya(glucose bias), calcuthe respective serum figures were pooled. y = 11 + 1.68x, r =
lower; when calculated as the difference between the ferricyanide and the glucose oxidase value in each sample, the bias was found to be of approximately constant magnitude over the range of glucose concentrations measured after i.v. glucose, both pre- and postdialysis. In addition, the bias was significantly reduced by dialysis at most sample times, its average value being 28 mg/dl before, and 18 after dialysis (p < 0.001). When the individual mean values for the bias of the ferricyanide method were plotted against the
creatinine concentration (Fig. 3), a highly significant, positive correlation was found. Also of interest is that the patients’ serum creatinine and this glucose bias both fell after dialysis by about 10 mg/dl. Figure 4 shows the mean plasma glucose curves of the eight normal subjects in whom the plasma glucose values by both the neocuproine and the glucose-oxidase method were available. The reducing method appeared to overestimate plasma glucose also in the normal sera. Linear regression analysis of all the glucose values for each method against those of the glucose oxidase system (considered as the reference analysis) was performed. The equations of the regression lines are given in the legend to Table 3. A quick comparison of these regression lines was obtained (Table 3) by computing the means ( ? 99% confidence limits) of the readings by the ferricyanide, the neocuproine, or the hexokinase method corresponding to three nominal plasma glucose values as read by the glucose oxidase assay. It can be seen that neocuproine was less precise than either ferricyanide or hexokinase over the range of concentrations chosen. Both neocuproine and ferricyanide significantly overestimated the glucose oxidase measures over the range explored. Furthermore, this overestimation was nearly always signihcantly reduced in the uremic patients after dialysis, whereas hexokinase did not appear to “read” dialyzable interferences. The individual figures of the glucose parameters computed from the glucose oxidase data are
normal subjects
-I? -h;
c
L“t
i”
glucose oxidase
l
o---a
neocuproine
loo_
Plasma glucose conFig. 4. centrations (mean f SEM) after intravenous glucose (0.33 g/kg) in 8 healthy subjects, as measured by the neocuproine and the glucose oxidase method.
“E : z
9 si
0,
1 I I I0
5
K)
I
20
30 minutes
I
40
I
50
1
66
FERRANNINI
130
Table 3.
Comparison
ET
AL.
of Four Methods of Glucose Assay in Uremic Patients, Before (b) and After la) Dialysis and in Normal Subjects Glucose Oxldase
Assay
Method
100
Uremics
mg/dl
200
b
Ferricyanide
Neocuproine Hexokinase
a
mg/dl
300
b
mg/dl
b
a
a
mg/dl
127 + 5’t
119i4’
228 f 4.t
217 f 3’
330*
11.t
315*4’
mg/dl
119+8’
119&9’
229 f
218 i
339 f
12.t
316 f
10’
104 f 4
198zt4
287 i
5’
mg/dl
1oozt
5
6-t
6’
195 f 3’
295 f 8
Normals Neocuproine
mg/dl
113i7’
226 i
The values in this table are derived from linear regression
4‘
340 f
9’
analysis of all the plasma glucose data of the uremic subjects for each
method against the glucose oxidase data, both before and after dialysis. The equations
of the regresston lines are:
fe = 25(+3)
+ 1.017(~0.016)gx,SEE
nc =
8&6)
+ 1.104(+0.032)gx,
SEE = 25, r = 0.95
= 12.r
= 0.99
hk =
2(&3)
+ 0.976(rtO.O19)gx,
SEE = 15.r
= 0.98
= 0.99
before dialysis, and fe = 21(&3)
+ 0.980(+0.013)gx,
SEE = 1l.r
nc = 21 f&6)
+ 0.983(&0.029)gx.
SEE = 25.r
= 0.95
hk = 12b3)
+ 0.915(+0.014)gx.SEE
= 12.r
= 0.99
after dialysis. In the normal subjects, the equation of the neocuproine nc = -0.6&5) The abbreviations
are: fe = ferricyanide,
+ 1.134(&0.028)gx.
nc = neocuproine,
at the 1% (at least) level of significance
*Different
t The difference
data against the glucose oxidase data is: SEE = 22.r
hk = hexokinase,
from the value heading the column.
between the mean value before (b) and that after (al dialysis is srgnificant at the 1% level
reported in Tables 4 and 5. The means of these parameters did not differ significantly from those obtained with the other three assay systems in the uremic group, although the KGs calculated from the ferricyanide data tended to be the lowest ones. Concerning the effect of dialysis on i.v.GTT, Table 4.
Glucose and Insulin Parameters KG
Number
significantly (p < 0.05 or less) higher mean plasma glucose levels were found after dialysis with all the glucose assay systems. Although a slight decrease in both the KG and the KG, values was apparent postdialysis by all the methods, none of the paired differences reached statistical significance. GA, however, was signif-
in 10 Uremic Subjects Before (b) and After (a) a Single Dialysis’ GA
KG, %/rmn
%/min
Case
g/dl.
IA
ml”
S
mU/ml~mm
mU/mg
b
a
b
a
b
a
b
a
1
0.91
087
1 63
1 59
61
66
1.24
2 35
2
1.47
1 50
440
4 57
31
31
091
3
1 66
2 08
3.28
5.03
50
4.2
2.67
4
1 40
1.42
4.07
2.95
3.1
49
0.68
a
b 0.420
0 528
1 01
0 460
0 746
4 54
0.736
1.726
1 03
0 238
0.291
5
2.39
1 55
6.85
2 80
32
70
0.97
1.60
0.082
0.133
6
1.80
1.89
491
3.99
43
56
2.02
1.51
0 387
0 145
7
1 68
1 24
4 74
1.98
5.1
68
1 60
2 21
0.235
0.505
8
1 79
0.85
6.36
1 96
4.3
6.3
1 65
0 68
0.199
0.043
9
241
0 79
7.82
1.42
39
7.9
2 74
2 85
0 187
0.211
1 37
1.33
3.53
8.92
37
3.2
081
0 53
0 352
0 355
10 Mean+
SEM
1.69
ztO.2
1.35ztO.l
476
i0.6
3.52
ztO.7
4.2
i
0.3
56*05
Ferricyanide
1.34
f
0.1
1.20
+ 0.1
448
f
0.6
3.43
*
4.1
*
0.3
5.5
*
0.4
Neocuprome
1.52
zt 0.1
1.30
f
0.1
4.71
zt 0.5
3.32
+ 0.5
44
f
0.4
5.8
I
0.4
Hexokinase
1.63
+ 0 1
1.31
f
0.2
5.17
i
3.45
f
3.9
i
0.4
5.3
+ 0 5
NS
P ‘KG
= 0.95
and gx = glucose oxidase.
= fractional
the glucose
decay
and the msuhn
wdual data are computed NS. not sigmficant.
0.7
respectively.
glucose
values:
subtracted
153
i0.24
1005
NS
rate of absolute curve.
0 6
0 7
KG,
= fractional
of the basal
from the glucose oxldase measurements.
value
decay
1.83
+ 038
0.33OztOOti
NS
NS rate of glucose
S = slope
mcremenfs
of the regrewon
0.468*0.16
above
fastmg
line of plasma
Insulin
level.
GA and IA
= area
under
on plasma glucose. The indi-
INTRAVENOUS
GLUCOSE
EFFECT
ON
DIALYSIS
131
PATIENTS
Table 5. Glucose and Insulin Parameters in 13 Normal Subjects’ CC%?
KG
Number
%/mm
Fasting IRI
GA
KG, %/mm
g/dl.m,n
mu/ml,
1.64
3.62
4.0
4.4
067
1.08
2 24
5.8
3.2
0.20
0.092
3
2.30
11.78
1.6
4.8
1.54
0.650
4
1.49
4.26
3.3
6.9
1.72
0.718
5
1 .oo
1 75
6.7
4.2
071
0210
6
1.01
1.87
62
4.8
0.60
0 270
7
1.40
3.21
5.2
3.3
1.38
0.303
0.274
8
1.45
2.67
6.7
4.3
0.50
0 129
9
1.85
4.73
4.4
6.7
2.26
0.324
10
1.38
2.97
6.3
7.2
0.86
0 083
1.29
2.35
62
3.2
0.52
0.252
12
1.61
3.28
71
10.2
2.01
0 205
13
1.59
3.52
6.1
9.1
1.73
1.47
* 0.1
2.71
f 0.7
5.0 * 0.4
Uremics
b
1.69
f 0.2
4.76
f 0.6
4.2 + 0.3
12.3
Uremics
a
1.35
zt 0.1
3.52
f 0.7
5.6 i 0.5
17.2
tp < $p <
mU/mg
2
SEM
*The
5 mln
11
Mean
purposes.
IA
IrU/ml
i
symbols
are as in Table
4. The mean
The data are by glucose
values
for the uremic
9roup
5.6 i
before
0.6
0 329
1.13
* 0.19
0.295
f 2.77
1.53
zt 0 24
0.330
i
0.06
+ 3.54
1.83
f 0.38
0.468
I
0.16
(b) and after (a) dialysis
are given here for comparison
oxidase.
0.05 0.01.
icantly higher postdialysis (Table 4) and, again, this difference was detected by all the methods of glucose assay. The KG and GA values obtained by glucose oxidase did not differ significantly between the uremic subjects, either before or after dialysis and their controls (Table 5). In these, the glucose parameters by neocuproine (KG = 1.44 + O.l4%/min, KG, = 3.30 -t 0.67, and GA = 6.3 f 0.8 g/dl.min) were not statistically different from those by glucose oxidase. The mean fasting plasma IRI level was significantly higher in the uremic than in the reference
group, both before (p < 0.05) and after (p < 0.01) dialysis, but was not appreciably changed by the dialytic treatment. Significantly higher mean IRI concentrations were also found in the uremic patients in comparison with the controls at nearly all the time points during the postdialysis, but not the predialysis, test (Fig. 5). The mean values of the insulin area-under-curve of the uremic subjects differed from those of the healthy subjects (1.46 -t 0.21 mu/ml. min), before (2.26 t- 0.34, p < 0.05)as well as after (2.86 +- 0.43, p < 0.01)dialysis, but were not significantly changed by a single dialysis. When
mean t SEM
normals (shaded) uremics before uremics after o------(1 l
Plasma insulin conFig. 5. centrations (mean ? SEMI after i.v. glucose (0.33 g/kg) in 10 uremic patients on maintenance hemodialysis. before and after dialysis. The shaded area is + SEM of the mean values in 13 normal subjects. The stars indicate the mean postdialysis IRI values that are significantly different from those of the normal subjects (‘p < 0.05. lp < 0.01). l
f 0.05
l
FERRANNINI
Ol
0
1
2
3 serum
4 calcium
5
6
7
6
ET AL.
similar to the mean value of the controls (Tables 4 and 5). The existence of correlations between the clinical variables and the glucose and insulin parameters in the uremic group was systematically investigated by linear regression analysis of the pooled pre- and postdialysis data. The glucose oxidase measurements were chosen as the reference for glucose values. Both KG (y = 0.84 + 0.13x, r = 0.5,~< 0.05) and KG, (y = -0.39 + 0.76x, r = 0.7, p < 0.01) were directly correlated with serum potassium concentrations. Higher triglyceride levels were highly significantly associated with higher insulin areas (IA) (y = 36 i- 53x, r = 0.6, p < 0.01). The insulinogenic index changed inversely with plasma calcium (Fig. 6). Lastly, a strong direct relationship existed between IA and methylguanidine levels before, but not after, dialysis (Fig. 7).
mEq/l
Fig. 6. Relationship between plasma calcium and insulinogenic index in 10 uremic subjects. Pre- and postdialysis values were pooled.
the insulin response was evaluated as the insulin area above the fasting level (IA), the uremic group still exhibited a larger response than the control group, even though the differences between the means did not reach statistical significance in either test (Table 5). In this connection, however, we note that the entire insulin area is a better estimate of the insulin response26-2sand more closely reflects the actual insulin concentration at the sites of action. It must also be observed that a large scatter in the insulin values was found in the uremic subjects, some of whom (e.g., cases 3 and 9) were definitely “hyperresponders,” whereas others were low-normal responders. The insulinogenic index (IG) was not affected by dialysis (3.64 + 0.51 mU/mg before, and 3.43 + 0.86 after dialysis), and did not prove to be different from the average normal value (2.72 k 0.72) in either test. The slope (S) of the regression line of plasma insulin upon plasma glucose was significantly (p < 0.01 or less) steeper postdialysis in patients 3 and 7, less steep in cases 6 and 8, essentially unchanged in the others. Hence, the mean pre- and postdialysis S values did not differ from each other, and were
DISCUSSION
Our data confirm that the plasma of uremic patients contains reducing substances that interfere with the glucose measurements obtained by nonspecific methods (especially the ferricyanide method). This interference is roughly constant throughout the range of glucose concentrations found after i.v. loading and is significantly cut down in parallel with the removal of toxic metabolites by dialysis. Creatinine appears to contribute in vivo a great share of this interference, in close analogy with what is seen when creatinine is added in vitro.”
Fig. 7. + 0.016x,
The equation of this regression line is: y = -0.068 I = 0.79, p < 0.01.
INTRAVENOUS
GLUCOSE EFFECT ON DIALYSIS
PATIENTS
A bias in the glucose assay system may explain why the prevalence of glucose intolerance in our patients, estimated as a KG value < 1, in either test was greater (35%) with the ferricyanide than with the other methods (20%). When the fasting level is subtracted from the postload glucose readings (i.e., a KG, value is calculated), this between-method discrepancy is almost canceled (7-8 KG, values less than 3, in either test, by all the methods), presumably because most of the interference is eliminated. Thus, the KG,s computed from the 4 curves in Figs. I and 2 all decreased postdialysis by a similar amount (24%-33%). This means that if i.v.GTT’s performed with the use of different assay systems are to be compared, it is better to choose the KG, parameter. Surprisingly, however, the nonspecific methods were in good agreement with the specific ones also with respect to the KG parameter, which showed after dialysis quite comparable changes in the two sets of reading (- 13% versus -2O%, respectively). In fact, subtracting a constant factor (i.e., the reading due to interfering substances) from a monoexponentially decaying function increases per se the slope, and, consequently, higher KGs are expected after dialysis by the reducing methods. Since this was not the case in the present series of patients, it seems logical to conclude that the effect of the glucose bias on the computation of KG was small. In different patients, or on different occasions,29 however, the weight of a bias in the system of glucose assay on the evaluation of an i.v.GTT may be greater. In the oral glucose tolerance test. it would lead to erroneous ranking of the results. It could be concluded that the carbohydrate tolerance of patients on maintenance hemodialysis, as judged from the i.v.GTT, is not acutely improved after a single dialysis and that this can be shown by virtually all the methods of glucose assay. Nevertheless, some caveat to this conclusion seems in order. In fact, i.v.GTT is a rather crude estimate of glucose tolerance. First, when a KG (or KG,) parameter is computed, the assumption is made that the glucose system is monocompartmental; this is known not to be true.30 Second, glucose changes after i.v. administration are brought about both by peripheral glucose uptake and by inhibition of endogenous
133
glucose production (and, in the normal subject, by glucose spill-over in the urine). Clearly, a single parameter can describe these simultaneous processes only imperfectly. Additional uncertainties, then, stem from the glucose dose given and the time interval over which the experimental points are interpolated.3’ Similarly, the glucose area-under-curve, though obviously related to the rate of glucose change, is no better approach to an i.v.GTT, since it also depends on the volume into which the administered glucose is distributed. The higher GA values that we found in the uremic subjects postdialysis (Table 4) may be, at least in part, an effect of the fluid loss occurring with dialysis, as suggested by the inverse relationship between GA and the number of hours after the end of the dialytic course (y = 5.6 - 0.02x, r = 0.6, p < 0.01). This overall uncertainty of the i.v.GTT is greatest in the individual case: in patients 6 and 10, for instance, the three parameters “read” differently the same changes in glucose concentration (Table 4). Therefore, if small differences in intravenous glucose tolerance are to be detected, it is probably wise to increase the number of observations.29 These methodological limitations, adding to the great variability of the uremic population (age, obesity, clinical condition, etc.) may account for the discordant results of i.v. glucose testing in chronic uremia.1,‘4 The problem remains of why a single dialysis, though effectively detoxicating our uremic patients did not improve their carbohydrate tolerance. Insulin is the first factor to be considered. Since the metabolic clearance rate of insulin is not very different in the chronically dialyzed subject from that of the normal person,32 plasma insulin concentrations can be taken as reflecting insulin secretion, given the assumption that the hepatic extraction of the hormone is the same in both.33 As judged, then, from their plasma insulin levels, our patients secreted more insulin than normal both in the fasting state and following glucose stimulation in both tests, in agreement with previous findings. 5V7V9~‘0534*35 Postdialysis, this hyperresponsiveness was, at least in part, the effect of higher plasma glucose levels. In fact, when the insulin response was normalized by its stimulus (GA), the insulinogenic index was not significantly different from that of the controls. The same was
FERRANNINI
134
true of the slope of the regression line of plasma insulin upon plasma glucose, indicating that, on the whole, the ability of the P-cell to cope with rising glucose levels was maintained. This enhanced insulin secretory activity was associated with fasting glucose levels (84 + 5 mg/dl before, and 88 + 9 after dialysis versus 76 -t 3 of the controls), and KG values that were not significantly different (or, if anything, worse) from those of the normal subjects. Thus, a reduced sensitivity of the tissues to insulin action was present in our uremic patients36*37and a single dialysis failed to correct this insulin resistance. It must be stressed that the behavior of uremic subjects is not uniform and individual patients may behave contrary to the common trend. This variable responsiveness is very likely to reflect the complex influence of numerous other factors. In our patients, for instance, hypokalemia was associated with lower KG values. Though serum potassium may not mirror intracellular potassium, it is nevertheless probable that potassium depletion of some degree contributed to impair glucose tolerance. This interpretation is in line with available clinical38 and experimental39 evidence. In addition, the insulinogenic index bore a reciprocal relation to plasma calcium ions and a similar, even if weaker (r = -0.39, 0.1 > p > O.OS), relationship existed between KG and calcium. This finding agrees well with the results of Amend and coworkerq4’ who reported that in uremic subjects with severe secondary hyperparathyroidism, hypercalcemia induced by calcium infusion was associated with slower glucose decay rates but unchanged insulin areas on i.v.GTT. Though these authors did not compute an insulinogenic index, it is to be expected that, because of the lower KG values, it would have been reduced. This observation suggests that, on the one hand, hypercalcemia causes diminished tissue insulin sensitivity; on the other hand, the specific secretory capacity of the P-cell in response to acute glucose stimulation is impaired in the presence of excessive extracellular calcium concentrations. However, Linda11 et al.” reported increased insulin respanses in dialyzed patients with severe
ET AL.
secondary hyperparathyroidism. Furthermore, it is well known that glucose-induced insulin release is inhibited in the absence of a threshold concentration of extracellular calcium,4’ though a rise of extracellular calcium above the optimal secretagogic level is capable of depressing insulin secretion.42 Thus, the effects of calcium on glucose-insulin interrelationships in uremia remain to be elucidated. An unusual finding, which we have already reported for another series of uremic subjects,43 was the relation between methylguanidine and insulin response (Fig.7). A possible explanation is that methylguanidine, when chronically retained, may stimulate insulin secretion by directly acting on the pancreas, in analogy with what has been described to happen with other guanidine derivatives in the perfused pancreas.44 Finally, the usual markers of uremic intoxication (creatinine and urea) did not show any relation to either glucose tolerance or insulin secretion. This has been reported by others.5 However, it is conceivable that a longer withdrawal of the dialytic treatment would have brought up some link between the accumulation of toxic metabolites and the glucose-insulin system.35 In conclusion, in chronically dialyzed patients, a single dialysis does not improve glucose tolerance and fails to correct the insulin resistance. Thus, the beneficial effect of the removal of dialyzable toxins can be counterbalanced by negative influences such as electrolyte disturbances, blood flow and volume changes, fluid loss and redistribution, all occurring with the dialytic procedure and extending through the early postdialytic period. In the long run, the insulin resistance may lead to other metabolic derangements such as hyperlipemia.45 ACKNOWLEDGMENT We are indebted
to the patients who participated in study for their generous collaboration. Thanks are due to R. Giordani, Dr. Ester Morelli, P. Cecchetti, A. Masoni, V. Bartolini, for their help, and to Carol Ann Hacon editing the manuscript. We would also like to thank Ralph A. DeFronzo for his helpful criticism.
this Dr. and for Dr.
REFERENCES 1. DeFronzo RA, Andres R, Edgar P, et al: Carbohydrate metabolism in uremia: A review. Medicine 52:469-481, 1973
2. Lindner A, Charra B, Sherrard DJ, et al: Accelerated atherosclerosis in prolonged maintenance hemodialysis. N Engl J Med 290:697-701, 1974
INTRAVENOUS
GLUCOSE EFFECT ON DIALYSIS
135
PATIENTS
3. Westervelt FB Jr, Schreiner GE: The carbohydrate intolerance of uremic patients. Ann Intern Med 57:266-274, 1962 4. Sagild U: Glucose tolerance in acute ischemic renal failure. Acta Med Stand 172:405-411, 1962 5. Hampers CL, Soeldner JS, Doak PB, et al: Effect of chronic renal failure and hemodialysis on carbohydrate metabolism. J Clin Invest 45:1719-1731, 1966. 6. Alfrey AC, Sussman KE. Holmes JH: Changes in glucose and insulin metabolism induced by dialysis in patients with chronic uremia. Metabolism 16:733-740, 1961 7. Hampers CL, Soeldner JS, Gleason RE, et al: Insulinglucose relationships in uremia. Am J Clin Nutr 21:414421, 1968 8. Spitz I. Rubenstein AH, Bershon AI, et al: The effect of dialysis on the carbohydrate intolerance of chronic renal failure. Horm Metab Res 2:86-93, 1970 9. Lowrie EC, Soeldner JS, Hampers CL, et al: Glucose metabolism and insulin secretion in uremic, prediabetic, and normal subjects. J Lab Clin Med 76:603-615, 1970 IO. Linda]] A, Carmena R, Cohen S, et al: Insulin hypersecretion in patients on chronic hemodialysis. Role of parathyroids. J Clin Endocrinol Metab 32:653-658, 1971 I I. Kokot F, Kuska J: Influence of extracorporeal dialysis on glucose utilization and insulin secretion in patients with
acute renal failure. Eur J Clin Invest 3:105-l 11, 1973 12. Powell JB. Djuh Y-Y: A comparison of automated methods for glucose analysis in patients with uremia before and after dialysis. Am J Clin Pathol 56%16, 1971 13. Swenson RS, Weisinger J, Reaven GM: Evidence that hemodialysis does not improve the glucose tolerance of patients with chronic renal failure. Metabolism 23:929-936, 1974 14. Davidson MB, Lowrie EC, Hampers CL: Lack of dialyzable insulin antagonist in uremia. Metabolism 18:387394. 1969 15. de Nobel E, van’t Laar A: The size of the loading dose as an important determinant of the results of the oral glucose tolerance test. A study in subjects with slightly impaired glucose tolerance. Diabetes 27:4248, 1978 16. Menichini CC. Gonella M, Barsotti G, et al: Determination of methylguanidine in serum and urine from normal and uremic subjects. Experientia 27:l 157-l 158, 1971 17. Bittner DL. McCleary ML: The cupric-phenanthroline chelate in the determination of monosaccharides in whole blood. Am J Clin Pathol 40:423. 1963 18. Hoffmann WS: A rapid method for the determination of glucose in blood and urine. J Biol Chem 120:51-55, 1937 19. Schmidt FH: Die enzymatische Bestimmung von glucose und fructose nebeneinender. Klin Wochenschr 39:1244-1247, 1961 20. Robin M, Saifer A: Determination of glucose in biologic fluids with an automated enzymatic procedure, Clin Chem I l:840-845. 1965 21. Pilo A, Zucchelli CC: Automatic treatment of radioimmunoassay data: An experimental validation of the results. Clin Chim Acta 64:1-9, 1975 22. Lundbaek K: Intravenous glucose tolerance as a tool in defnition and diagnosis of diabetes mellitus. Br Med J 1:1507--1513, 1962
23. Seltzer
HS, Allen EW, Herron
AL Jr, et al: Insulin
secretion in response to glycemic stimulus: Relation of delayed initial release to carbohydrate intolerance in mild diabetes mellitus. J Clin Invest 46:323-335. 1967 24. Dixon WJ, Massey FJ: Introduction to statistical analysis (ed 3). New York, McGraw-Hill, 1969 25. Gutman RA, Uy A, Schaloub glyceridemia in chronic non-nephrotic Clin Nutr 26:165-171, 1973
RJ, et al: Hypertrirenal failure. Am J
26. Reaven GM, Olefsky JM: Relationship between insulin response during the intravenous glucose tolerance test, rate of fractional glucose removal and the degree of insulin resistance in normal adults. Diabetes 23:454-459. 1959 27. Brunzell JD. Robertson RP, Lerner RL, et al: Relationship between fasting plasma glucose level and insulin secretion during intravenous glucose tolerance tests. J Clin Endocrinol Metab 42:222-229, 1976 28. Pilo A, Ferrannini E, Navalesi R: Measurement of glucose-induced insulin delivery rate in man by deconvolution analysis. Am J Physiol 233:E500-E508, 1977 29. Richards CJ, Freeman RM, Samaan NA: Glucose tolerance tests and their insulin and growth hormone responses in hemodialysis patients: Are they reproducible? Clin Nephrol 8:384-394. 1977 30. lnsel PA, Liljenquist JE. Tobin JD. et al: Insulin control of glucose metabolism in man. A new kinetic analysis. J Clin Invest 55:1057-1066, 1975 31. Lafferty HH, Giddings AEB, Mangnall D: The analysis of decay curves. Clin Sci Mol Med 52:97-101, 1977 32. Navalesi R, Pilo A, Lenzi S, et al: Insulin metabolism in chronic uremia and in the anephric state: Effect of the dialytic treatment. J Clin Endocrinol Metab 40:70-80, 1975 33. Pilo A, N avalesi R, Ferrannini E: Insulin kinetics after portal and peripheral injection of ‘*51-insulin. I. Data analysis
and modeling.
Am J Physiol 230: 1626-l 630, I976
34. Classman AB: Serum insulin levels of “diabetic” and “non-diabetic” patients on chronic hemodialysis. Am J Clin Pathol 57:320-323, 1972 35. Hutchings RH, Hegstrom RM, Scribner BH: Glucose tolerance in patients on long-term intermittent dialysis. Ann Intern Med 65:275-285, I966 36. Horton ES, Johnson C, Lebovitz HE: Carbohydrate metabolism in uremia. Ann Intern Med 68:63-74. 1968 37. Luke RG, Dinwoodie AJ, Linton AL, et al: Fructose and glucose intolerance in uremia. J Lab Clin Med 64:731740, 1964 38. Corm JW: Hypertension. impaired carbohydrate tolerance. 1141, 1965
the potassium ion and N Engl J Med 273: I 135-
39. Sagild U, Andersen V. Andreasen PB: Glucose tolerance and insulin response in experimental potassium depletion. Acta Med Stand 169:243-251, 1961 40. Amend WJC Jr, Steinberg SM. Lowrie EC, et al: The influence of serum calcium and parathyroid hormone upon glucose metabolism in uremia. J Lab Clin Med 86:435-444, 1975 41. Grodsky GM, Bennett LL: Cation requirements of insulin secretion in the isolated perfused pancreas. Diabetes 15:9 I O-920. I966
136
42. Hellman B: The significance of calcium for glucose stimulation of insulin release. Endocrinology 97:392-398, 1975 43. Ferrannini E, Navalesi R, Pilo A, et al: Methylguanidine and insulin response in patients on maintenance haemodialysis. Lancet 2: 1028-1029, 1976
FERRANNINI
ET AL.
44. Alsever RN, Georg RH, Sussman KE: Stimulation of insulin secretion by guanidinoacetic acid and other guanidine derivatives. Endocrinology 86:332-336, 1970 45. Feldman HA, Singer I: Endocrinology and metabolism in uremia and dialysis: A clinical review. Medicine 541345-376, 1974
assistance
Alessandro
To learn whether a single dialysis can acutely improve the intravenous glucose tolerance (i.v.GTT) of chronically dialyzed patients, a standard i.v.GTT was performed on 10 nonobese uremic subjects on maintenance hemodialysis for 27 * 9 (mean ? SEM) mo, and on a control group of 13 normal subjects. The uremic patients were tested first 0.2-17 (range) hr, and then 65-109 hr, from last dialysis. In the uremic sera, plasma glucose was analyzed by 4 methods; 2 reducing (neocuproine and ferricyanide) and 2 enzymatic (hexokinase and glucose oxidase). The reducing methods markedly overestimated plasma glucose concentration because of the presence of nonglucose, reducing substances (notably, creatinine). This interference was significantly cut down by dialysis. A single dialysis, on the other hand, failed to improve the glucose fractional decay rate (KG) computed from the glucose oxidase data (1.69 + 0.2%/min before and 1.35 + 0.1 after dialysis, versus 1.47 + 0.1 of the normal subjects). The same conclusion was derived from the data measured by the other 3 methods of glucose assay. Fasting plasma insulin concentrations were, on average, above normal (5.5 + 0.6 NU/ml) both before (12.3 k 2.7, p < 0.05) and after (17.2 k 3.5, p < 0.01) a single dialysis. Likewise, the area under the glucoseinduced plasma insulin curve was significantly greater than normal (1.46 ? 0.21 mu/ml. min) both before (2.26 f 0.34, p < 0.05), and after (2.86 + 0.43, p < 0.01) dialysis. A single dialysis had little effect on either basal or glucose-stimulated insulin release, and no significant difference in the insulinogenic index (insulin area/glucose area) was found between the control and the uremic group in either test. Insulin response was not correlated with KG, whereas it was significantly associated with higher triglyceride levels. Creatinine, urea or methylguanidfne did not appear to have any influence on KG, but lower serum potassium levels were significantly associated with poorer i.v.GTT’s. Plasma calcium bore a reciprocal relation to the insulinogenic index. Chronically dialyzed subjects show some degree of tissue insulin resistance, which a single dialysis fails to correct. Electrolyte disturbances may play a role in this metabolic derangement.
R
ENAL FAILURE is frequently associated with a glucose intolerance that is distinct from that of diabetes mellitus.’ Although hardly ever requiring any treatment, this abnormality is believed to shorten the lifespan of uremic subjects on prolonged hemodialysis through the sequence hyperglycemia-hyperinsulinemiahyperlipidemia-accelerated atherosclerosisfatal vascular accidents.* Mefabolism, Vol. 28, No. 2 (February), 1979
Pilo, Mario Tuoni,
of Giuseppe Buzzigoli and Claudia Boni
Whether hemodialysis can improve glucose tolerance is, at present, controversial. Early studies,‘-” in which glucose was measured by nonspecific, reducing methods, have shown that hemodialysis corrects carbohydrate intolerance to some extent. Reducing methods, however, yield falsely raised glucose values for uremic sera, in which reducing substances other than glucose (e.g., creatinine) are present at concentrations much higher than normal.” In this way, hemodialysis might appear to lead to an improvement of glucose tolerance by removing these nonglucose, reducing substances. In a recent study,13 using a glucose-specific assay method, the oral glucose tolerance of uremic patients, either undialyzed or on maintenance hemodialysis, was not found to be improved by dialytic treatment to any relevant degree. Davidson and coworkers,lJ on the other hand, also using a glucose-specific assay, have reported that glucose KG values are significantly higher after dialysis. We thought it would be of interest to investigate the acute effects of dialysis on the response to intravenous glucose in uremic subjects on maintenance hemodialysis with the use of 4 methods of glucose assay, 2 reducing (neocuproine and ferricyanide), and 2 enzymatic (hexokinase and glucose oxidase). In this paper, we report that, in our group of patients, a single dialysis failed to improve the i.v.GTT whichever method of glucose measurement was used. The influence on plasma glucose disappearance of factors such as insulin, toxic metabolites. and electrolytes is also discussed. From the C.N.R. Clinical Physiolog_v Laboratory and the 2nd Medical Clinic, University of Pisa. Ita1.v. Received for publication February 3, 1978. Supported in part by USPHS Grant Nol-AM-8-0707. bt, lhe Chronic Uremia-Artificial Kidney Program of the National Institutes of’ Arthritis. Metabolism. and Digestive Diseases, National Institutes of Health, Bethesda. Md. Address reprint requests to Dr. E. Ferrannini, C.N. R. Clinical Physiology Laboratory, Via Savi 8, 56100 Piss. Italy. @ I979 by Grune & Stratton, I~C. 00260495/78/2802-0005$02.00/#
125
126
FERRANNINI
ET AL.
Table 1. Clinical Data of the Study Subjects
Case Number
Sax
Age (vr)
Weight’ (kg)
Height (cm1
Percent ldealt BodyWt
Underlying+ Disease
Durationof Dlalytlc Treatment bno)
Uremic Subjects 1
M
40
68
164
100
2
F
38
45
166
72
3
M
49
80
176
104
CPN
36
4
M
31
62
170
a7
CGN
15
5
M
33
75
173
102
CGN
1
6
M
28
47
170
68
CGN
7
F
37
58
156
103
CGN
8
M
56
63
158
97
CPN
90
9
M
66
77
165
113
CGN
28
10
M
44
74
177
95
CGN
42
42
65
168
94
27
4
4
2
5
9
99
SEM
CGN CGN
4 2
54
Healthy Subjects 1
F
30
59
162
2
F
54
43
158
69
3
M
36
78
173
107
4
F
35
56
168
89
5
F
35
58
167
93
6
F
16
41
155
84
7
F
35
63
157
111
8
F
57
48
158
77
9
M
32
75
170
105 104
10
F
55
59
147
11
F
40
63
168
94
12
F
27
52
159
95
13
F
46
62
159
101
38
58
162
94
3
3
2
3
Mean SEM
*In the uremic subjects. mean of pre- and postdialysis weight. The average weight loss after a standard 5-hr hemodialysis was about 2 kg. t Society of Actuaries Ed: Build and Blood Pressure Study. vol I. Chicago, 1959. p 16. SCGN = chronic glomerulonephntis; CPN = chronic pyelonephritis.
MATERIALS AND METHODS Study subjects. Ten uremic subjects on maintenance hemodialysis were studied; their pertinent clinical data are given in Table I. Criteria for inclusion in the study were that patients were of normal body weight, had negative family histories of diabetes mellitus, showed no evidence of gastrointestinal or endocrine disorders, had no clinical, laboratory or radiologic signs of hyperparathyroidism. that there was no known glucose intolerance before the onset of renal failure, no intercurrent illness, anorexia, nausea, vomiting, or recent change in body weight. All patients were in good general condition, their hematocrit was about 25% on average, and they were taking no medication such as antihypertensive, diuretic, steroid, or oral contraceptive drugs. They were on a free diet, with at least 200 g of carbohydrate daily (the average Italian diet consists of 43%-48% carbohydrates, 12%-l?‘% proteins, and 35%-40% fat). Salt intake was unrestricted. All patients had been on a thrice-weekly, 5-hr hemodialysis schedule for an average of 27 mo. Hemodialysis was performed using Kiil hollow-fiber dialyzers. The dialysate
was free of glucose, and contained (as mEq/ 1) sodium (I 32), potassium (2.2), calcium (3.4), magnesium (I .O), chloride (102). and acetate (38.8). Thirteen age-matched, healthy volunteers (Table I), having normal body weight and rlegative family histories of diabetes mellitus, served as the cohtrol group. These controls were not sex-matched; this, however, was not regarded as an important source of bias, because good evidence indicates that sex per se has no effect on gilucose tolerance.” All the subjects studied gave their informed consent to the investigation. Experimental protocol. Each normal subject received one. i.v.GTT, whereas each uremic patient was tested first a short time (range 0.2-17 hr. “post-dialysis”) from last dialysis, and then after a longer time (65-109 hr, “predialysis”), the order of the two studies being randomized. All studies were performed in the fasted (8-14 hr) state. A plastic cannula was inserted in an antecubital vein and kept patent by intermittent irrigation with isotonic saline. Fasting blood was obtained for glucose and immunoreactive insulin (IRI) determination. In thb uremic subjects, blood for measuring creatinine, nonproteic hitrogen, methylguanidine,
INTRAVENOUS
total
GLUCOSE EFFECT ON DIALYSIS
proteins,
calcium
triglyceride,
and chloride
cholesterol,
was also taken.
PATIENTS
sodium, After
127
potassium,
minus-basal
a stabilization
period of IO-1 5 mitt, 0.33 g of glucose per kg of body weight (as a 40% solution) period.
was injected
intravenously
were computed
and blood samples were taken 2.5, 5, 7.5, 10, 15. 20,
IRI determination. drawn
For these latter
into tubes containing
plasma
was promptly
separated
and sodium
The insulinogenic
blood was fluoride;
of IA to GA.
the
and stored at -20°C
until
procedures.
(azotemia),
total
plasma
terol were measured was analyzed et al.‘”
Creatinine, proteins,
triglyceride.
by standard
methods;
by the chromatographic
Electrolytes
spectrometry.
were
Plasma
measured
glucose methods:
mated on a Technicon
Autoanalyzer,”
also automated (hk).
Glucose was
on a Technicon
automated
Analyzer,19
and
(D)
Analyzer.
measured
normals.
(A)
on glucose
by
the
in addition,
glucose
plasma
method.
following
5.6% (nc), 2.8% (fe),
errors:
(gx). Plasma
concentration
Using
charcoal
centrifugation
bound from free insulin. coethcients the
at
auto-
(B) ferricyanide
(fe),
(gx),
basal
so that
(3-25
coefficient
Rate
In
Data
analysis.
Glucose
fractional
by monoexponential
least squares fit) of either
the absolute
(IRI)
variation
in the range of samples
in the above
disappearance (KG)
glucose
between
analysis
the
intercept
and the standard
by the
the means of unpaired variances
was performed, and
error
were
that
of
by the the stan-
regression
of the estimate
The comparison
a regression
line
glucose in each subject.
(SEE)
between two regression
line and the identity
line. was
by the t test.
Dialysis effectively reduced the plasma concentration of creatinine, nonproteic nitrogen, methylguanidine, and potassium in nearly all the uremic patients (Table 2). Total plasma protein concentration was found to be significantly raised at the shorter distance from dialysis, as a result, most probably, of the plasma volume reduction that occurs with dialysis. No patient had raised serum cholesterol, and dialysis was without effect on this. In cases 1, 3,
separating
interpolation
the
RESULTS
range. was computed
regression
by
and 3.0%
Plasma
to
eight
insulin
fell
system
of the
of the respon-
mean values were compared
we had the
for
area inclusive
by the t test, or where
of
performed
glucose
values.
as the ratio
F) at the 5% level of significance.
were also computed.
using dextran-coated
readings
deviation
lines. or between
5.8% (hk).
pU/ml).
all insulin
linear
plasma
The inter- and intra-assay
concentrations
test.*’
When
on a Beckman
method.
3,000Xg
Behrens’ dard
(C) hexokiReaction
was also measured
were 14% and 7%, respectively,*’
were diluted
by
(nc).
of immunoreactive
by radioimmunoassay,
and
in duplicate
these methods.
was measured
Pre- and postdialysis
(Fisher’s
base-line
estimate
secretory
upon plasma
different
oxidase
the neocuproine
insulin
area-under-curve
as the slope (S) of the regression
paired t test. The differences
absorption
IO to
of the experimen-
was then obtained*’ Another
insulin
data were analyzed
subjects,
glucose
the
system of Giovannetti
8600
oxidase
lo In the normal
was obtained
by atomic neocuproine
LKB
of
stimulus of plasma
taken from
of the respective
The value of the insulin
siveness
and choles-
Autoanalyzer.‘*
a
nitrogen
methylguanidine
was assayed
each of the following
nase
nonproteic
(IA)
integration
index (IG)
basal was also calculated.
assayed. Anal~~tical
and the insulin
by trapezoidal
tal curves after subtraction
for glucose and
measurements,
EDTA
plasma glucose readings
injection.
The glucose (GA)
over a 2-min
25, 30, 35, 40, SO, and 60 min after injection
(KG,)
60 min after
rate’” (using
a
or the absolute-
Table 2. Blood Chemistry in 10 Uremic Patients Before (bJand After la) Dialysis
1 --_---
2
3
4
5
6
7 --____
8
9
10
Mean + SEM
babababababababababa Hours
afterdlalysls
b
15
72
17
109
13
109
13
70
1
70
1
68 02 68 02 68 02 68 02
128
59
67
62
203
100
146
66
156
82
139
.75
144
51
65
serum cremmne img/dll Plasma
72
196
82
170
59
guan,d,ne
l,,gidlI (mg/dU
hst,ng
71-06
64
60
70
63
168
92
67
45
72
72
72
46
120
44
124
64
136
64
88
136
221
94
142
$5
220
125
189
69
161
91
205
100
17,
50
216
87
186
82
176
56
191
98t
11.
7 3
112
60
59
72
165
42
35
93
107
134
120
254
206
257
240
202
240
39
416
123
73
74
70
62
64
7.9
60
76
90
105
70
106
$6
79
72
8,
82
75
72
106
120
130
200
130
200
200
180
200
115
145
130
115
166
190
120
115
190
200
120
145
154
+ I,
157;
i
11,133
*s*
69
f
851
9 7
lAJ/rnli
12,
172
135
plasma
protems Serum
(g/dlk
73104’
T35*o!i
cholesterol
img/dll Serum
12
tnglycer&des
lmg/dll Sod,““? Potasswn
h&,/II ImEq/l)
125
195
100
115
200
170
90
100
150
50
100
25
150
25
136
50
345
360
120
80
132
135
130
146
131
143
135
142
140
125
137
140
136
132
130
141
136
135
134
143
136
138
43
40
5 8
5
52
102
51
82
40
65
48
50
37
86
41
62
43
50
35
64+06t 42
1 53
Ca,c,um
hEq/l,
44
45
39
46
43
26
25
44
40
73
50
70
55
67
45
69
33
2,
60
6,
Chlorode
ImEq/ll
95
94
106
96
107
$6
103
97
93
99
104
95
93
91
98
94
102
9,
92
$5
005
+!J ,’ 001
tp
114
plasma
ll,S”ll”
‘P /
1561
61~-2
methyl-
Azotem~a
Toia,
188
a
77: 51
0001
$9
27 t2
13611 441-02
tO2’ *
2
53106 96
+
,
128
FERRANNINI
ET AL.
400_
10 UREMIC PATIENTS BEFORE DIALYSIS
!I
t_ 0 0 .--.
neocuproine 0 ferricyanide D hexokinase glucose oxidase
Fig. 1. Plasma glucose concentrations (mean ? SEY) after intravenous glucose (0.33 g/kg) in 10 uremic subjects on chronic hemodialysis 66-109 hr after last dialysis (“before dialysis”), as obtained by 4 methods of glucose assay.
100,
and 9, on the other hand, elevated serum triglyceride was found on either or both occasion(s). The very low triglyceride concentration measured postdialysis in cases 5-8 was probably the effect of antecedent heparinization; this is a common finding in the early postdialytic period.25 Dialysis, however, did not change mean triglyceride concentration significantly. Minor changes were seen in the sodium and chloride levels, whereas total plasma calcium was significantly higher postdialysis, presumably owing to the concomitant increase in plasma protein concentration. In fact, when the pre- and
postdiaiysis values were pooled, it was found that calcium and proteins changed consensually (y = 0.73 + 0.7x, r = 0.7, p < 0.01). Figure 1 shows the plasma glucose concentrations after the i.v. glucose load in the uremic group, as measured by the 4 methods of glucose assay: the 2 reducing methods, of which the ferricyanide gave the higher readings, overestimated the glucose concentration as read by the enzymatic assays all through the predialysis curve. This systematic error, or bias, of the reducing versus the specific methods was also evident in the postdialysis test (Fig. 2), but was
lo UREMIC PATIENTS AFTER DIALYSIS 0 0 -
Plasma glucose conFig. 2. centrations (mean 2 SEM) after intravenous glucose (0.33 g/kg) in 10 chronically dialyzed subjects 0.2-17 hr after last dialysis (“after dialysis”), as obtained by 4 methods of gluCOSe assay.
6 2.5 j
1.5 10
1
20 1
I
30
minutes
I
neocuproine o ferricyanide a hexokinase glucose oxidase
40 I
50
60 1
INTRAVENOUS
GLUCOSE EFFECT ON DIALYSIS
129
PATIENTS
60
1
0
20 54erwn80di2nine
24
‘t&Al
Plot of the mean difference Fig. 3. nide and the glucose oxidase reading lated for each uremic subject, against creatinine value. Pre- and postdialysis The equation of the regression line is: 0.74. p i 0.001.
between the ferricya(glucose bias), calcuthe respective serum figures were pooled. y = 11 + 1.68x, r =
lower; when calculated as the difference between the ferricyanide and the glucose oxidase value in each sample, the bias was found to be of approximately constant magnitude over the range of glucose concentrations measured after i.v. glucose, both pre- and postdialysis. In addition, the bias was significantly reduced by dialysis at most sample times, its average value being 28 mg/dl before, and 18 after dialysis (p < 0.001). When the individual mean values for the bias of the ferricyanide method were plotted against the
creatinine concentration (Fig. 3), a highly significant, positive correlation was found. Also of interest is that the patients’ serum creatinine and this glucose bias both fell after dialysis by about 10 mg/dl. Figure 4 shows the mean plasma glucose curves of the eight normal subjects in whom the plasma glucose values by both the neocuproine and the glucose-oxidase method were available. The reducing method appeared to overestimate plasma glucose also in the normal sera. Linear regression analysis of all the glucose values for each method against those of the glucose oxidase system (considered as the reference analysis) was performed. The equations of the regression lines are given in the legend to Table 3. A quick comparison of these regression lines was obtained (Table 3) by computing the means ( ? 99% confidence limits) of the readings by the ferricyanide, the neocuproine, or the hexokinase method corresponding to three nominal plasma glucose values as read by the glucose oxidase assay. It can be seen that neocuproine was less precise than either ferricyanide or hexokinase over the range of concentrations chosen. Both neocuproine and ferricyanide significantly overestimated the glucose oxidase measures over the range explored. Furthermore, this overestimation was nearly always signihcantly reduced in the uremic patients after dialysis, whereas hexokinase did not appear to “read” dialyzable interferences. The individual figures of the glucose parameters computed from the glucose oxidase data are
normal subjects
-I? -h;
c
L“t
i”
glucose oxidase
l
o---a
neocuproine
loo_
Plasma glucose conFig. 4. centrations (mean f SEM) after intravenous glucose (0.33 g/kg) in 8 healthy subjects, as measured by the neocuproine and the glucose oxidase method.
“E : z
9 si
0,
1 I I I0
5
K)
I
20
30 minutes
I
40
I
50
1
66
FERRANNINI
130
Table 3.
Comparison
ET
AL.
of Four Methods of Glucose Assay in Uremic Patients, Before (b) and After la) Dialysis and in Normal Subjects Glucose Oxldase
Assay
Method
100
Uremics
mg/dl
200
b
Ferricyanide
Neocuproine Hexokinase
a
mg/dl
300
b
mg/dl
b
a
a
mg/dl
127 + 5’t
119i4’
228 f 4.t
217 f 3’
330*
11.t
315*4’
mg/dl
119+8’
119&9’
229 f
218 i
339 f
12.t
316 f
10’
104 f 4
198zt4
287 i
5’
mg/dl
1oozt
5
6-t
6’
195 f 3’
295 f 8
Normals Neocuproine
mg/dl
113i7’
226 i
The values in this table are derived from linear regression
4‘
340 f
9’
analysis of all the plasma glucose data of the uremic subjects for each
method against the glucose oxidase data, both before and after dialysis. The equations
of the regresston lines are:
fe = 25(+3)
+ 1.017(~0.016)gx,SEE
nc =
8&6)
+ 1.104(+0.032)gx,
SEE = 25, r = 0.95
= 12.r
= 0.99
hk =
2(&3)
+ 0.976(rtO.O19)gx,
SEE = 15.r
= 0.98
= 0.99
before dialysis, and fe = 21(&3)
+ 0.980(+0.013)gx,
SEE = 1l.r
nc = 21 f&6)
+ 0.983(&0.029)gx.
SEE = 25.r
= 0.95
hk = 12b3)
+ 0.915(+0.014)gx.SEE
= 12.r
= 0.99
after dialysis. In the normal subjects, the equation of the neocuproine nc = -0.6&5) The abbreviations
are: fe = ferricyanide,
+ 1.134(&0.028)gx.
nc = neocuproine,
at the 1% (at least) level of significance
*Different
t The difference
data against the glucose oxidase data is: SEE = 22.r
hk = hexokinase,
from the value heading the column.
between the mean value before (b) and that after (al dialysis is srgnificant at the 1% level
reported in Tables 4 and 5. The means of these parameters did not differ significantly from those obtained with the other three assay systems in the uremic group, although the KGs calculated from the ferricyanide data tended to be the lowest ones. Concerning the effect of dialysis on i.v.GTT, Table 4.
Glucose and Insulin Parameters KG
Number
significantly (p < 0.05 or less) higher mean plasma glucose levels were found after dialysis with all the glucose assay systems. Although a slight decrease in both the KG and the KG, values was apparent postdialysis by all the methods, none of the paired differences reached statistical significance. GA, however, was signif-
in 10 Uremic Subjects Before (b) and After (a) a Single Dialysis’ GA
KG, %/rmn
%/min
Case
g/dl.
IA
ml”
S
mU/ml~mm
mU/mg
b
a
b
a
b
a
b
a
1
0.91
087
1 63
1 59
61
66
1.24
2 35
2
1.47
1 50
440
4 57
31
31
091
3
1 66
2 08
3.28
5.03
50
4.2
2.67
4
1 40
1.42
4.07
2.95
3.1
49
0.68
a
b 0.420
0 528
1 01
0 460
0 746
4 54
0.736
1.726
1 03
0 238
0.291
5
2.39
1 55
6.85
2 80
32
70
0.97
1.60
0.082
0.133
6
1.80
1.89
491
3.99
43
56
2.02
1.51
0 387
0 145
7
1 68
1 24
4 74
1.98
5.1
68
1 60
2 21
0.235
0.505
8
1 79
0.85
6.36
1 96
4.3
6.3
1 65
0 68
0.199
0.043
9
241
0 79
7.82
1.42
39
7.9
2 74
2 85
0 187
0.211
1 37
1.33
3.53
8.92
37
3.2
081
0 53
0 352
0 355
10 Mean+
SEM
1.69
ztO.2
1.35ztO.l
476
i0.6
3.52
ztO.7
4.2
i
0.3
56*05
Ferricyanide
1.34
f
0.1
1.20
+ 0.1
448
f
0.6
3.43
*
4.1
*
0.3
5.5
*
0.4
Neocuprome
1.52
zt 0.1
1.30
f
0.1
4.71
zt 0.5
3.32
+ 0.5
44
f
0.4
5.8
I
0.4
Hexokinase
1.63
+ 0 1
1.31
f
0.2
5.17
i
3.45
f
3.9
i
0.4
5.3
+ 0 5
NS
P ‘KG
= 0.95
and gx = glucose oxidase.
= fractional
the glucose
decay
and the msuhn
wdual data are computed NS. not sigmficant.
0.7
respectively.
glucose
values:
subtracted
153
i0.24
1005
NS
rate of absolute curve.
0 6
0 7
KG,
= fractional
of the basal
from the glucose oxldase measurements.
value
decay
1.83
+ 038
0.33OztOOti
NS
NS rate of glucose
S = slope
mcremenfs
of the regrewon
0.468*0.16
above
fastmg
line of plasma
Insulin
level.
GA and IA
= area
under
on plasma glucose. The indi-
INTRAVENOUS
GLUCOSE
EFFECT
ON
DIALYSIS
131
PATIENTS
Table 5. Glucose and Insulin Parameters in 13 Normal Subjects’ CC%?
KG
Number
%/mm
Fasting IRI
GA
KG, %/mm
g/dl.m,n
mu/ml,
1.64
3.62
4.0
4.4
067
1.08
2 24
5.8
3.2
0.20
0.092
3
2.30
11.78
1.6
4.8
1.54
0.650
4
1.49
4.26
3.3
6.9
1.72
0.718
5
1 .oo
1 75
6.7
4.2
071
0210
6
1.01
1.87
62
4.8
0.60
0 270
7
1.40
3.21
5.2
3.3
1.38
0.303
0.274
8
1.45
2.67
6.7
4.3
0.50
0 129
9
1.85
4.73
4.4
6.7
2.26
0.324
10
1.38
2.97
6.3
7.2
0.86
0 083
1.29
2.35
62
3.2
0.52
0.252
12
1.61
3.28
71
10.2
2.01
0 205
13
1.59
3.52
6.1
9.1
1.73
1.47
* 0.1
2.71
f 0.7
5.0 * 0.4
Uremics
b
1.69
f 0.2
4.76
f 0.6
4.2 + 0.3
12.3
Uremics
a
1.35
zt 0.1
3.52
f 0.7
5.6 i 0.5
17.2
tp < $p <
mU/mg
2
SEM
*The
5 mln
11
Mean
purposes.
IA
IrU/ml
i
symbols
are as in Table
4. The mean
The data are by glucose
values
for the uremic
9roup
5.6 i
before
0.6
0 329
1.13
* 0.19
0.295
f 2.77
1.53
zt 0 24
0.330
i
0.06
+ 3.54
1.83
f 0.38
0.468
I
0.16
(b) and after (a) dialysis
are given here for comparison
oxidase.
0.05 0.01.
icantly higher postdialysis (Table 4) and, again, this difference was detected by all the methods of glucose assay. The KG and GA values obtained by glucose oxidase did not differ significantly between the uremic subjects, either before or after dialysis and their controls (Table 5). In these, the glucose parameters by neocuproine (KG = 1.44 + O.l4%/min, KG, = 3.30 -t 0.67, and GA = 6.3 f 0.8 g/dl.min) were not statistically different from those by glucose oxidase. The mean fasting plasma IRI level was significantly higher in the uremic than in the reference
group, both before (p < 0.05) and after (p < 0.01) dialysis, but was not appreciably changed by the dialytic treatment. Significantly higher mean IRI concentrations were also found in the uremic patients in comparison with the controls at nearly all the time points during the postdialysis, but not the predialysis, test (Fig. 5). The mean values of the insulin area-under-curve of the uremic subjects differed from those of the healthy subjects (1.46 -t 0.21 mu/ml. min), before (2.26 t- 0.34, p < 0.05)as well as after (2.86 +- 0.43, p < 0.01)dialysis, but were not significantly changed by a single dialysis. When
mean t SEM
normals (shaded) uremics before uremics after o------(1 l
Plasma insulin conFig. 5. centrations (mean ? SEMI after i.v. glucose (0.33 g/kg) in 10 uremic patients on maintenance hemodialysis. before and after dialysis. The shaded area is + SEM of the mean values in 13 normal subjects. The stars indicate the mean postdialysis IRI values that are significantly different from those of the normal subjects (‘p < 0.05. lp < 0.01). l
f 0.05
l
FERRANNINI
Ol
0
1
2
3 serum
4 calcium
5
6
7
6
ET AL.
similar to the mean value of the controls (Tables 4 and 5). The existence of correlations between the clinical variables and the glucose and insulin parameters in the uremic group was systematically investigated by linear regression analysis of the pooled pre- and postdialysis data. The glucose oxidase measurements were chosen as the reference for glucose values. Both KG (y = 0.84 + 0.13x, r = 0.5,~< 0.05) and KG, (y = -0.39 + 0.76x, r = 0.7, p < 0.01) were directly correlated with serum potassium concentrations. Higher triglyceride levels were highly significantly associated with higher insulin areas (IA) (y = 36 i- 53x, r = 0.6, p < 0.01). The insulinogenic index changed inversely with plasma calcium (Fig. 6). Lastly, a strong direct relationship existed between IA and methylguanidine levels before, but not after, dialysis (Fig. 7).
mEq/l
Fig. 6. Relationship between plasma calcium and insulinogenic index in 10 uremic subjects. Pre- and postdialysis values were pooled.
the insulin response was evaluated as the insulin area above the fasting level (IA), the uremic group still exhibited a larger response than the control group, even though the differences between the means did not reach statistical significance in either test (Table 5). In this connection, however, we note that the entire insulin area is a better estimate of the insulin response26-2sand more closely reflects the actual insulin concentration at the sites of action. It must also be observed that a large scatter in the insulin values was found in the uremic subjects, some of whom (e.g., cases 3 and 9) were definitely “hyperresponders,” whereas others were low-normal responders. The insulinogenic index (IG) was not affected by dialysis (3.64 + 0.51 mU/mg before, and 3.43 + 0.86 after dialysis), and did not prove to be different from the average normal value (2.72 k 0.72) in either test. The slope (S) of the regression line of plasma insulin upon plasma glucose was significantly (p < 0.01 or less) steeper postdialysis in patients 3 and 7, less steep in cases 6 and 8, essentially unchanged in the others. Hence, the mean pre- and postdialysis S values did not differ from each other, and were
DISCUSSION
Our data confirm that the plasma of uremic patients contains reducing substances that interfere with the glucose measurements obtained by nonspecific methods (especially the ferricyanide method). This interference is roughly constant throughout the range of glucose concentrations found after i.v. loading and is significantly cut down in parallel with the removal of toxic metabolites by dialysis. Creatinine appears to contribute in vivo a great share of this interference, in close analogy with what is seen when creatinine is added in vitro.”
Fig. 7. + 0.016x,
The equation of this regression line is: y = -0.068 I = 0.79, p < 0.01.
INTRAVENOUS
GLUCOSE EFFECT ON DIALYSIS
PATIENTS
A bias in the glucose assay system may explain why the prevalence of glucose intolerance in our patients, estimated as a KG value < 1, in either test was greater (35%) with the ferricyanide than with the other methods (20%). When the fasting level is subtracted from the postload glucose readings (i.e., a KG, value is calculated), this between-method discrepancy is almost canceled (7-8 KG, values less than 3, in either test, by all the methods), presumably because most of the interference is eliminated. Thus, the KG,s computed from the 4 curves in Figs. I and 2 all decreased postdialysis by a similar amount (24%-33%). This means that if i.v.GTT’s performed with the use of different assay systems are to be compared, it is better to choose the KG, parameter. Surprisingly, however, the nonspecific methods were in good agreement with the specific ones also with respect to the KG parameter, which showed after dialysis quite comparable changes in the two sets of reading (- 13% versus -2O%, respectively). In fact, subtracting a constant factor (i.e., the reading due to interfering substances) from a monoexponentially decaying function increases per se the slope, and, consequently, higher KGs are expected after dialysis by the reducing methods. Since this was not the case in the present series of patients, it seems logical to conclude that the effect of the glucose bias on the computation of KG was small. In different patients, or on different occasions,29 however, the weight of a bias in the system of glucose assay on the evaluation of an i.v.GTT may be greater. In the oral glucose tolerance test. it would lead to erroneous ranking of the results. It could be concluded that the carbohydrate tolerance of patients on maintenance hemodialysis, as judged from the i.v.GTT, is not acutely improved after a single dialysis and that this can be shown by virtually all the methods of glucose assay. Nevertheless, some caveat to this conclusion seems in order. In fact, i.v.GTT is a rather crude estimate of glucose tolerance. First, when a KG (or KG,) parameter is computed, the assumption is made that the glucose system is monocompartmental; this is known not to be true.30 Second, glucose changes after i.v. administration are brought about both by peripheral glucose uptake and by inhibition of endogenous
133
glucose production (and, in the normal subject, by glucose spill-over in the urine). Clearly, a single parameter can describe these simultaneous processes only imperfectly. Additional uncertainties, then, stem from the glucose dose given and the time interval over which the experimental points are interpolated.3’ Similarly, the glucose area-under-curve, though obviously related to the rate of glucose change, is no better approach to an i.v.GTT, since it also depends on the volume into which the administered glucose is distributed. The higher GA values that we found in the uremic subjects postdialysis (Table 4) may be, at least in part, an effect of the fluid loss occurring with dialysis, as suggested by the inverse relationship between GA and the number of hours after the end of the dialytic course (y = 5.6 - 0.02x, r = 0.6, p < 0.01). This overall uncertainty of the i.v.GTT is greatest in the individual case: in patients 6 and 10, for instance, the three parameters “read” differently the same changes in glucose concentration (Table 4). Therefore, if small differences in intravenous glucose tolerance are to be detected, it is probably wise to increase the number of observations.29 These methodological limitations, adding to the great variability of the uremic population (age, obesity, clinical condition, etc.) may account for the discordant results of i.v. glucose testing in chronic uremia.1,‘4 The problem remains of why a single dialysis, though effectively detoxicating our uremic patients did not improve their carbohydrate tolerance. Insulin is the first factor to be considered. Since the metabolic clearance rate of insulin is not very different in the chronically dialyzed subject from that of the normal person,32 plasma insulin concentrations can be taken as reflecting insulin secretion, given the assumption that the hepatic extraction of the hormone is the same in both.33 As judged, then, from their plasma insulin levels, our patients secreted more insulin than normal both in the fasting state and following glucose stimulation in both tests, in agreement with previous findings. 5V7V9~‘0534*35 Postdialysis, this hyperresponsiveness was, at least in part, the effect of higher plasma glucose levels. In fact, when the insulin response was normalized by its stimulus (GA), the insulinogenic index was not significantly different from that of the controls. The same was
FERRANNINI
134
true of the slope of the regression line of plasma insulin upon plasma glucose, indicating that, on the whole, the ability of the P-cell to cope with rising glucose levels was maintained. This enhanced insulin secretory activity was associated with fasting glucose levels (84 + 5 mg/dl before, and 88 + 9 after dialysis versus 76 -t 3 of the controls), and KG values that were not significantly different (or, if anything, worse) from those of the normal subjects. Thus, a reduced sensitivity of the tissues to insulin action was present in our uremic patients36*37and a single dialysis failed to correct this insulin resistance. It must be stressed that the behavior of uremic subjects is not uniform and individual patients may behave contrary to the common trend. This variable responsiveness is very likely to reflect the complex influence of numerous other factors. In our patients, for instance, hypokalemia was associated with lower KG values. Though serum potassium may not mirror intracellular potassium, it is nevertheless probable that potassium depletion of some degree contributed to impair glucose tolerance. This interpretation is in line with available clinical38 and experimental39 evidence. In addition, the insulinogenic index bore a reciprocal relation to plasma calcium ions and a similar, even if weaker (r = -0.39, 0.1 > p > O.OS), relationship existed between KG and calcium. This finding agrees well with the results of Amend and coworkerq4’ who reported that in uremic subjects with severe secondary hyperparathyroidism, hypercalcemia induced by calcium infusion was associated with slower glucose decay rates but unchanged insulin areas on i.v.GTT. Though these authors did not compute an insulinogenic index, it is to be expected that, because of the lower KG values, it would have been reduced. This observation suggests that, on the one hand, hypercalcemia causes diminished tissue insulin sensitivity; on the other hand, the specific secretory capacity of the P-cell in response to acute glucose stimulation is impaired in the presence of excessive extracellular calcium concentrations. However, Linda11 et al.” reported increased insulin respanses in dialyzed patients with severe
ET AL.
secondary hyperparathyroidism. Furthermore, it is well known that glucose-induced insulin release is inhibited in the absence of a threshold concentration of extracellular calcium,4’ though a rise of extracellular calcium above the optimal secretagogic level is capable of depressing insulin secretion.42 Thus, the effects of calcium on glucose-insulin interrelationships in uremia remain to be elucidated. An unusual finding, which we have already reported for another series of uremic subjects,43 was the relation between methylguanidine and insulin response (Fig.7). A possible explanation is that methylguanidine, when chronically retained, may stimulate insulin secretion by directly acting on the pancreas, in analogy with what has been described to happen with other guanidine derivatives in the perfused pancreas.44 Finally, the usual markers of uremic intoxication (creatinine and urea) did not show any relation to either glucose tolerance or insulin secretion. This has been reported by others.5 However, it is conceivable that a longer withdrawal of the dialytic treatment would have brought up some link between the accumulation of toxic metabolites and the glucose-insulin system.35 In conclusion, in chronically dialyzed patients, a single dialysis does not improve glucose tolerance and fails to correct the insulin resistance. Thus, the beneficial effect of the removal of dialyzable toxins can be counterbalanced by negative influences such as electrolyte disturbances, blood flow and volume changes, fluid loss and redistribution, all occurring with the dialytic procedure and extending through the early postdialytic period. In the long run, the insulin resistance may lead to other metabolic derangements such as hyperlipemia.45 ACKNOWLEDGMENT We are indebted
to the patients who participated in study for their generous collaboration. Thanks are due to R. Giordani, Dr. Ester Morelli, P. Cecchetti, A. Masoni, V. Bartolini, for their help, and to Carol Ann Hacon editing the manuscript. We would also like to thank Ralph A. DeFronzo for his helpful criticism.
this Dr. and for Dr.
REFERENCES 1. DeFronzo RA, Andres R, Edgar P, et al: Carbohydrate metabolism in uremia: A review. Medicine 52:469-481, 1973
2. Lindner A, Charra B, Sherrard DJ, et al: Accelerated atherosclerosis in prolonged maintenance hemodialysis. N Engl J Med 290:697-701, 1974
INTRAVENOUS
GLUCOSE EFFECT ON DIALYSIS
135
PATIENTS
3. Westervelt FB Jr, Schreiner GE: The carbohydrate intolerance of uremic patients. Ann Intern Med 57:266-274, 1962 4. Sagild U: Glucose tolerance in acute ischemic renal failure. Acta Med Stand 172:405-411, 1962 5. Hampers CL, Soeldner JS, Doak PB, et al: Effect of chronic renal failure and hemodialysis on carbohydrate metabolism. J Clin Invest 45:1719-1731, 1966. 6. Alfrey AC, Sussman KE. Holmes JH: Changes in glucose and insulin metabolism induced by dialysis in patients with chronic uremia. Metabolism 16:733-740, 1961 7. Hampers CL, Soeldner JS, Gleason RE, et al: Insulinglucose relationships in uremia. Am J Clin Nutr 21:414421, 1968 8. Spitz I. Rubenstein AH, Bershon AI, et al: The effect of dialysis on the carbohydrate intolerance of chronic renal failure. Horm Metab Res 2:86-93, 1970 9. Lowrie EC, Soeldner JS, Hampers CL, et al: Glucose metabolism and insulin secretion in uremic, prediabetic, and normal subjects. J Lab Clin Med 76:603-615, 1970 IO. Linda]] A, Carmena R, Cohen S, et al: Insulin hypersecretion in patients on chronic hemodialysis. Role of parathyroids. J Clin Endocrinol Metab 32:653-658, 1971 I I. Kokot F, Kuska J: Influence of extracorporeal dialysis on glucose utilization and insulin secretion in patients with
acute renal failure. Eur J Clin Invest 3:105-l 11, 1973 12. Powell JB. Djuh Y-Y: A comparison of automated methods for glucose analysis in patients with uremia before and after dialysis. Am J Clin Pathol 56%16, 1971 13. Swenson RS, Weisinger J, Reaven GM: Evidence that hemodialysis does not improve the glucose tolerance of patients with chronic renal failure. Metabolism 23:929-936, 1974 14. Davidson MB, Lowrie EC, Hampers CL: Lack of dialyzable insulin antagonist in uremia. Metabolism 18:387394. 1969 15. de Nobel E, van’t Laar A: The size of the loading dose as an important determinant of the results of the oral glucose tolerance test. A study in subjects with slightly impaired glucose tolerance. Diabetes 27:4248, 1978 16. Menichini CC. Gonella M, Barsotti G, et al: Determination of methylguanidine in serum and urine from normal and uremic subjects. Experientia 27:l 157-l 158, 1971 17. Bittner DL. McCleary ML: The cupric-phenanthroline chelate in the determination of monosaccharides in whole blood. Am J Clin Pathol 40:423. 1963 18. Hoffmann WS: A rapid method for the determination of glucose in blood and urine. J Biol Chem 120:51-55, 1937 19. Schmidt FH: Die enzymatische Bestimmung von glucose und fructose nebeneinender. Klin Wochenschr 39:1244-1247, 1961 20. Robin M, Saifer A: Determination of glucose in biologic fluids with an automated enzymatic procedure, Clin Chem I l:840-845. 1965 21. Pilo A, Zucchelli CC: Automatic treatment of radioimmunoassay data: An experimental validation of the results. Clin Chim Acta 64:1-9, 1975 22. Lundbaek K: Intravenous glucose tolerance as a tool in defnition and diagnosis of diabetes mellitus. Br Med J 1:1507--1513, 1962
23. Seltzer
HS, Allen EW, Herron
AL Jr, et al: Insulin
secretion in response to glycemic stimulus: Relation of delayed initial release to carbohydrate intolerance in mild diabetes mellitus. J Clin Invest 46:323-335. 1967 24. Dixon WJ, Massey FJ: Introduction to statistical analysis (ed 3). New York, McGraw-Hill, 1969 25. Gutman RA, Uy A, Schaloub glyceridemia in chronic non-nephrotic Clin Nutr 26:165-171, 1973
RJ, et al: Hypertrirenal failure. Am J
26. Reaven GM, Olefsky JM: Relationship between insulin response during the intravenous glucose tolerance test, rate of fractional glucose removal and the degree of insulin resistance in normal adults. Diabetes 23:454-459. 1959 27. Brunzell JD. Robertson RP, Lerner RL, et al: Relationship between fasting plasma glucose level and insulin secretion during intravenous glucose tolerance tests. J Clin Endocrinol Metab 42:222-229, 1976 28. Pilo A, Ferrannini E, Navalesi R: Measurement of glucose-induced insulin delivery rate in man by deconvolution analysis. Am J Physiol 233:E500-E508, 1977 29. Richards CJ, Freeman RM, Samaan NA: Glucose tolerance tests and their insulin and growth hormone responses in hemodialysis patients: Are they reproducible? Clin Nephrol 8:384-394. 1977 30. lnsel PA, Liljenquist JE. Tobin JD. et al: Insulin control of glucose metabolism in man. A new kinetic analysis. J Clin Invest 55:1057-1066, 1975 31. Lafferty HH, Giddings AEB, Mangnall D: The analysis of decay curves. Clin Sci Mol Med 52:97-101, 1977 32. Navalesi R, Pilo A, Lenzi S, et al: Insulin metabolism in chronic uremia and in the anephric state: Effect of the dialytic treatment. J Clin Endocrinol Metab 40:70-80, 1975 33. Pilo A, N avalesi R, Ferrannini E: Insulin kinetics after portal and peripheral injection of ‘*51-insulin. I. Data analysis
and modeling.
Am J Physiol 230: 1626-l 630, I976
34. Classman AB: Serum insulin levels of “diabetic” and “non-diabetic” patients on chronic hemodialysis. Am J Clin Pathol 57:320-323, 1972 35. Hutchings RH, Hegstrom RM, Scribner BH: Glucose tolerance in patients on long-term intermittent dialysis. Ann Intern Med 65:275-285, I966 36. Horton ES, Johnson C, Lebovitz HE: Carbohydrate metabolism in uremia. Ann Intern Med 68:63-74. 1968 37. Luke RG, Dinwoodie AJ, Linton AL, et al: Fructose and glucose intolerance in uremia. J Lab Clin Med 64:731740, 1964 38. Corm JW: Hypertension. impaired carbohydrate tolerance. 1141, 1965
the potassium ion and N Engl J Med 273: I 135-
39. Sagild U, Andersen V. Andreasen PB: Glucose tolerance and insulin response in experimental potassium depletion. Acta Med Stand 169:243-251, 1961 40. Amend WJC Jr, Steinberg SM. Lowrie EC, et al: The influence of serum calcium and parathyroid hormone upon glucose metabolism in uremia. J Lab Clin Med 86:435-444, 1975 41. Grodsky GM, Bennett LL: Cation requirements of insulin secretion in the isolated perfused pancreas. Diabetes 15:9 I O-920. I966
136
42. Hellman B: The significance of calcium for glucose stimulation of insulin release. Endocrinology 97:392-398, 1975 43. Ferrannini E, Navalesi R, Pilo A, et al: Methylguanidine and insulin response in patients on maintenance haemodialysis. Lancet 2: 1028-1029, 1976
FERRANNINI
ET AL.
44. Alsever RN, Georg RH, Sussman KE: Stimulation of insulin secretion by guanidinoacetic acid and other guanidine derivatives. Endocrinology 86:332-336, 1970 45. Feldman HA, Singer I: Endocrinology and metabolism in uremia and dialysis: A clinical review. Medicine 541345-376, 1974