Eur
J
Clin
Chcm
Clin
Biochem
1996;
34:257-264
©
1996
by
Walter
de
Gruyter
·
Berlin
· New
York
Ionised
and
Total
Magnesium
Serum
Concentrations
in
Renal
and
Hepatic
Diseases
WolfR.
K
lpntann
1
,
Jan R
ler
}
,
Reinhard
Brunkhorst
2
and
Andreas
Sch
ler
3
1
Institut
f r
Klinische Chemie
I
2
Abteilung Nephrologie
3
Abteilung Gastroenterologie
und
Hepatologie
Medizinische
Hochschule Hannover, Hannover, Germany
Dedicated
to
Prof.
Dr. Dr. J. B
ttner
on the
occasion
of
his
65th
birthday
Summary: Ionised
and
total magnesium concentrations were determined
in the
serum
of
different
groups
of
patients
suffering
from
renal
or
hepatic
diseases.
Ionised magnesium
was
measured
by
Microlyte
6
(KONE,
Espoo,
Finland)
and
total magnesium
by
atomic absorption
spectrometry.
In
renal
insufficiency
ionised
and
total magnesium concentrations were almost equally increased.
In
proteinuria
with
a
normal
glomerular
filtration rate,
"pseudohypomagnesaemia"
was
observed,
i. e.
decreased total magnesium
concentration
in
parallel with
a
decreased albumin concentration with
no
significant change
in the
concentration
of
ionised magnesium.
Hypermagnesaemia
occurred
in
liver
diseases
combined with renal
insufficiency,
whereas "pseudohypomagnesae-
mia"
was
most
often
found
in the
absence
of
renal
failure.
Also
treatment with
an
aldosterone antagonist
was
associated with
a
normal ionised magnesium concentration,
but the
total magnesium concentration
was
decreased;
when
additional magnesium
was
administered,
the
total magnesium concentration approached
a
normal
value,
while
ionised magnesium slightly exceeded reference values. Only during
cyclosporin
treatment
did
both ionised
and
total
magnesium
concentrations become lowered. However,
the
decrease
of
total magnesium exceeded that
of
ionised
magnesium
due to
concomitant hypoalbuminaemia with reduction
of the
protein-bound fraction.
It is
concluded
that especially
low
total magnesium concentrations should
be
investigated
by
measurement
of
ionised magnesium
to
exclude "pseudohypomagnesaemia".
Introduction circumstances
in
which
the
concentration
of
active
mag-
^
.
,.
,
.
nesium ions
is
decreased
or the
less
meaningful
pseudo-
Renal
as
well
as
hepatic
diseases
may
have
an
.mpact
h
esaemia
be
ted
.
on
magnesium balance.
As
long
as
only total magnesium
concentration could
be
determined
(1-7),
there were
al-
ways some doubts
as to
whether alterations reflected
changes
of the
ionised (unbound) magnesium concentra- Materials
and
Methods
tion
or the
protein-bound concentration.
As in the
case
,
*
,
-
ι
.
^
ι
· ·
j
Patients
of
calcium,
a
decrease
or an
increase
of the
ionised
mag-
nesium
concentration
may be
considered
as a
"true"
Renal
diseases
change, whereas
a
decrease
of
the
total
magnesium
con-
L
Group
(RD
,
):
21
patients
(14
male)
with
renal
diseascs
centration
due to a
reduction
of
protein-bound magne- (tab.
I)
and
hypoalbuminaemia
(albumin
in the
serum
< 37
g/l)
but
sium might
be
called
a
"pseudohypomagnesaemia".
On
"normal'*
creatinine
clearance
(after
adjustment
to
1.73
m
2
body
,
^
r
j
A
,
^
,
.
A
A
.
surface)
were
investigated
(tab.
2).
Patients
suffering
at
the
same
the
other hand,
the
total magnesium concentration
may
time
from
gastro
.
intestina
i
diseaseSj
hypo
.
or
hyperthyroidism
or
lie
within
the
reference interval, while
the
ionised con-
diabetes
mellitus
were
excluded,
as
well
as
patients
treated
with
centration,
reflecting
the
active magnesium ions,
is
low-
cycJosporin
(only
two
cases),
cisplatin
or
calcium
concentrations
,
,
-.
ι
v·
j·
·*
*
-
A
L
exceeding
2.60
mmol/1
serum,
ered, because
of
complex binding,
e. g. to
citrate
m the
event
of
massive
transfusion
of
blood during liver trans-
2.
Group
(RD 2): 29
patients
(19
male)
with
renal
diseases
plantation
(8).
In
this
study, ionised
and
total
magnesium
<
lab
·
]>·
hypoalbuminaemia
and
decreased
creatinine
clearance
(ad-
r
. .
,
.
·
j
· ·
justed
to
1.73
m
2
body
surface)
were
investigated
(tab.
2).
They
concentrations
m the
serum were determined
in
patients
were
under
Λί|οηβη1
with
furoscmidc
(mean
dosage
103
mg/d;
suffering
from
liver
or
kidney
diseases,
to
determine
the
range
40-250
mg/d).
For
exclusion
criteria
see RD
1.
258
Külpmann
et
al.:
Ionised magnesium
in
renal
and
hepatic
diseases
Tab.
1
Patients
with
renal
diseases
Diagnosis
Minimal
change
glomcruloncphritis
Membranous
glomerulonephritis
Focal-segmcntal
glomerulonephritis
Focal-scgmental
mesangioproliferative
glomerulonephritis
Membranoproliferative
glomerulonephritis
Type
1
Systemic
lupus
erythematosus
Mesangioproliferative
glomerulonephritis
(IgA)
Panarteriitis
nodosa
Amyloidosis
Proteinuria
(unknown
origin)
No. of
patients
10
16
5
2
2.
Group
(LD 2): 20
patients
(13
male)
with
hypoalbuminaemia
and
increased
creatinine concentration
in the
serum, treated with
furosemide
(mean dosage
76
mg/d;
range
20-240
mg/d)
(tab.
4).
3.
Group
(LD 3):
11
patients
(5
male)
after
orthotopic
liver trans-
plantation
with hypoalbuminaemia, treated with
cyclosporin
(cyclosporin
in
serum: mean
115
g/l;
range
59—161
g/l)
(tab.
4).
4.
Group
(LD 4): 26
patients
(17
male),
with
hypoalbuminaemia,
treated
with
furosemide
(mean
dosage
45
mg/d;
range
20-80
mg/d)
and
spironolactone (mean dosage
115
mg/d;
range
50-200
mg/d)
(tab.
4).
5.
Group
(LD 5): 9
patients
(4
male) with hypoalbuminaemia,
treated
with
spironolactone (mean dosage
83
mg/d;
range
50-200
mg/d)
(tab.
4).
6.
Group
(LD 6): 15
patients
(9
male) with hypoalbuminaemia,
treated
with
furosemide (mean dosage
50
mg/d;
range
20-120
mg/d),
spironolactone (mean dosage
96
mg/d;
range
50—200
mg/d)
and
magnesium
p. o.
(Magnesium
Verla™:
mean dosage
112
mg/d;
range
80-240
mg/d)
(tab.
4).
Tab.
2
Renal
diseases
Property
1.
Group
(RD
1)
Mean
Range
2.
Group
(RD 2)
No.
of
patients
male
female
21
14
7
29
19
10
Mean
Range
Age
(years)
Creatinine
clearance
(ml/min)
S-Protein
(g/1)
S-Albumin
(g/1)
S-Cholinesterase
(kU/1)
S-Cholesterol
(mmol/1)
S-Triacylglycerol
(mmol/1)
U-Protein
(g/d)
33.7
164.6
53.4
25.2
7.16
9.6
3.34
5.59
9-54
93
-288
36
- 80
14
-
36
3.22- 10.92
4.0
-
19.2
0.78- 7.26
0.03- 15.96
46.1
53.9
57.6
28.9
7.19
8.5
3.18
6.66
11
-73
10
-73
34
-80
10
-36
2.98-12.10
4.9
-12.2
1.22- 8.20
0.41-16.00
Tab.
3
Patients
with
diseases
of the
liver
Diagnosis
No.
of
patients
Chronic
active
hepatitis
B 16
Chronic
active hepatitis
C 23
Chronic
active
hepatitis
B and C 4
Alcohol
toxic
liver
cirrhosis
28
Toxic
liver
cirrhosis (apart
from
alcohol toxic)
5
Autoimmune
hepatitis
5
Primary
biliary
cirrhosis
9
Primary
sclerotic cholangitis
6
a
r
Proteinase
inhibitor
deficiency
1
M.
Mlson
2
Budd-Chiari
syndrome
2
Liver
cirrhosis (unknown origin)
9
Liver
transplantation
11
Carcinoma
of the
liver/liver
metastases
11
Liver
diseases
The
following
groups
with
liver diseases were examined (for
ex-
clusion
criteria
see RD
1):
1.
Group
(LD 1): 51
patients
(31
male)
with
diseases
of the
liver
(tab.
3) and
hypoalbuminaemia,
but
"normal"
creatinine concentra-
tion
in the
serum (tab.
4).
The
groups were compared with
a
group
of 60
blood donors,
who
were
selected according
to the
guidelines
(9) and
matched
for age
(mean
±
standard deviation (years):
46.5
±
12.2))
and sex (34
male,
26
female).
Sampling
Blood
was
obtained
from the
patients
and the
blood donors
be-
tween
7 a. m. and
9.30
a.
m.
by
venipuncture
with
Monovettes
(Sar-
stedt,
Nümbrecht).
The
serum
was
separated after
centrifugation
(25
°C,
1200
g) and
stored
in a
refrigerator (max.
5
days),
if
analy-
ses
could
not be
performed
at
once.
Sample preparation
Serum
for
ionised magnesium measurement
was
adjusted
to pH
7.40
±
0.05
by
bubbling
CO
2
gas
prior
to
analysis.
Methods
Ionised
magnesium
The
concentration
of
ionised magnesium
in the
serum
was
deter-
mined
by
Microlyte
6
(KONE,
Espoo, Finland).
The
ion-selective
electrode which
was
used
for the
measurement
had a
modified
ETrt
5220-containing
PVC
membrane.
An
Ag/AgCl-electrode
was
used
as a
reference electrode.
The
ionised calcium concentration
was
determined simultaneously, since
this'is
needed
to
account
for
K
lpmann
et
al.:
Ionised magnesium
in
renal
and
hepatic
diseases
259
Tab.
4
Liver diseases
No.
of
patients
male
female
Property
Age
(years)
S-DeRitis ratio
n
S-Cholinesterase
(kU/1)
S-Albumin
(g/1)
S-Biiirubin
(μπιοΐ/ΐ)
P-Prothrombin
time
(%)
S-Creatinine
(μπιοΐ/ΐ)
1.
Group
51
31
20
Mean
50.6
0.94
3.12
31.3
45.7
75.7
62.0
(LD1)
Range
20
- 73
0.18- 1.70
1.06- 6.33
23
- 36
4
-478
43
-100
30
- 92
2.
Group
20
13
7
Mean
55.1
1.21
2.37
28.1
73.3
65.7
179.8
(LD2)
Range
32
- 72
0.61- 2.06
0.54- 5.96
16
- 36
7
-234
39
-100
124
-484
3.
Group
11
5
6
Mean
53.2
0.60
3.72
32.9
20.4
85.0
77.5
(LD3)
Range
38
-69
0.41-
1.05
1.72- 7.20
24
-36
6
-58
73
-93
65
-92
Property
No.
of
patients
male
female
Age
(years)
S-DeRitis
ratio
0
S-Cholinesterase
(ku/1)
S-Albumin
(g/1)
S-Bilirubin
(μηιοΐ/ΐ)
P-Prothrombin time
(%)
S-Creatinine
(μτηοΐ/ΐ)
4.
Group
Mean
26
17
9
52.2
1.28
1.95
27.9
81.0
60.1
70.1
(LD4)
Range
30
- 70
0.59- 2.00
0.81- 4.31
18
- 35
4
-410
37
-100
40
- 92
5.
Group
Mean
9
4
5
42.8
1.45
1.80
28.1
108.8
65.1
53.8
(LD5)
Range
•
16
- 60
0.87- 2.67
0.81- 4.31
20
- 33
15
-371
44-99
44
- 91
6.
Group
Mean
15
9
6
52.7
1.22
1.80
29.4
65.6
62.7
63.2
(LD6)
Range
31
- 69
0.52- 1.90
0.82- 5.60
18
- 35
11
-183
28
- 93
40
- 91
S-Aspartate
aminotransferase/S-Alanine
aminotransferase
the
common calcium interference
of the
magnesium electrode.
As
the
degree
of
interference
may
change with time,
the
actual degree
at
the
time
of
measurement
was
determined
by
measurement
of
standards (without protein) containing
different
concentrations
of
calcium
and
magnesium
just
prior
to the
determination. Ionised
magnesium concentrations
are
reported
for the
actual
pH of the
sample
and for pH
7.40 after calculation
by a
built-in computer.
In
this
study ionised magnesium concentration
is
given
after
adjusting
to
pH
7.4,
to
make comparison with
the
reference interval easier.
Other
quantities
Ionised
calcium,
pH,
ionised sodium
and
ionised potassium
in se-
rum
were determined simultaneously with ionised magnesium
by
Microlyte
6
(KONE,
Espoo,
Finland).
Total magnesium
in
serum
and
urine
was
determined
by
atomic
absorption
spectrometry
(AAS
1100,
Perkin-Elmer,
Uberlingen),
at
285.2
nm
after addition
of
LaCl
3
.
Albumin concentration
in
serum
was
determined
by
immunochem-
istry
(Array,
Beckman
Instr.,
M
nchen).
Creatinine
concentration
in
serum
was
measured
enzymatically
(Hitachi 747,
Boehringer
Mannheim, Mannheim).
Total
calcium
was
determined after reaction
with
cresolphthalein
complexone
(Hitachi 747, Boehringer Mannheim, Mannheim).
Total
protein
in
serum
was
measured bichromatically
by the
biuret
reaction (Hitachi 747, Boehringer Mannheim, Mannheim).
Cholesterol
and
triacylglycerol
in
serum were determined enzymat-
ically
(Hitachi 747, Boehringer Mannheim, Mannheim).
Bilirubin
was
measured bichromatically
after
reaction
with
2.5-
dichlorphenyldiazonium
salt (Hitachi 747, Boehringer
Mannheim,
Mannheim).
Enzymes
(aspartate
aminotransferase,
alanine
aminotransferase,
cholinesterase) were determined according
to the
standard pro-
cedures
of the
German Society
for
Clinical Chemistry.
Creatinine concentration
in
urine
was
measured
by
Jqffe's
reaction
(Elan,
Eppendorf,
Hamburg).
Total
protein
in
urine
was
determined
with
the
biuret reaction
after
precipitation
with trichloroacetic acid
and
blank correction.
Quality assessment
Precision
Kontrollogen-L
(lot
623136)
(Behring,
Frankfurt/M.),
and
Qualitrol
Precision (Merck, Darmstadt) were used
for
assessment
of
precision.
All
measurements
of the
ionised magnesium concentra-
tions were performed
in
duplicates, which were accepted,
if
their
values
differed
by
less than
2%.
Statistics
Data
of the
different
groups were compared
by
analysis
of
covari-
ance
and
Scheffe's
test
(10).
Adequate
Gaussian
distribution
was
tested
by the
Kolmogorov-Smirnov-test.
Results
Quality
assessment
Precision
Two
control
sera
were
analysed
on 10
consecutive days
in
duplicate.
The
coefficient
of
variation
for
ionised
260
Külpmann
et
al.:
Ionised
magnesium
in
renal
and
hepatic
diseases
Tab.
5
Precision
within
days
Specimen
Number
Mean Relative
of
deter-
value
SD'>
(CV)
2)
minations
(mmol/1)
(%)
ionised
Mg
concentration
KontrollogenL
10
0.534
2.2
Qualitrol
Precision
10
0.547
2.0
Ionised
Ca
concentration
KontrollogenL
10
1.18
1.8
Qualitrol
Precision
10
1.30
1.5
H
+
ion
concentration
KontrollogenL
10 6.5 ·
10~
5
7.8
Qualitrol
Precision
10 7.0
·
10"
6
8.6
!)
Standart deviation
2)
Coefficient
of
variation
magnesium
based
on
mean values
was
2.2%
and
2.0%
(tab.
5).
Accuracy
Reference
method values were
not yet
available
to
test
accuracy.
The
following
may be
considered
as
indicators
of
accuracy.
1.
Magnesium concentration
in
ultrafiltrates
exceeded
ionised magnesium concentration
by
4.9%
(2).
2.
Reference values obtained
by an
indirect method,
by
the
same
or
a
different
ion-selective electrode were sim-
ilar (tab.
6).
Tab.
6
Reference intervals
Property Reference
interval
(mmol/1)
Ionised
Mg
0.43
-0.66
(<?)
0.45-0.66
(?)
0.53-0.67
0.54-0.74
0.51-0.66
Total
Mg
0.77-1.
03
(<?)
0.73-1.06(5)
0.70-0.96
0.70-0.98
0.69-0.92
Ionised
Mg
0.53
-0.69
(c?)
Total
Mg
0.55
-0.69
(?)
0.61-0.85
0.65-0.80
0.65-0.77
Method
UC-ER
UC-ER
ISE
(Nova)
ISE
(KONE)
ISE
(KONE)
AAS
AAS
AAS
AAS
AAS
UC-ER/
AAS
UC-ER/
AAS
ISE/AAS
ISE/AAS
ISE/AAS
Author
Speich
et
al.
(12)
Speich
et
al.
(12)
Altura
et
al. (4)
Maj-Zurawska
et
al.
(13)
this study
Speich
et
al.
(12)
\
/
Speich
et al.
(12)
Altura
et al. (4)
Maj-Zurawska
et
al.
(13)
this study
Speich
et
al.
(12)
Speich
et
al.
(12)
Altura
et
al, (4)
Maj-Zurawska
et
al.
(13)
this
study
AAS: atomic absorption
spectrometry
ISE:
ion-selective
electrode
Clinical
study
Reference
interval
The
reference intervals
for
total
and
ionised magnesium
were
calculated
from
data
of 60
healthy blood donors
and
include
the
range between
the
2.5th
and
97.5th per-
centile
(tab.
6).
There
was no
significant influence
of
age or sex on the
reference interval
of
ionised
and
total
magnesium
nor on the
pertinent ratio
as
proven
by
analysis
of
covariance.
The
same held true
for the
following
groups
of
patients.
Renal
diseases
Ionised
and
total magnesium concentrations
in the se-
rum
of
patients
suffering
from
renal
diseases
are
pre-
sented
in
table
7.
In
patients with
proteinuria
(RD
1)
and
decreased albu-
min
concentration
in the
serum,
the
total magnesium
concentration
was 9%
lower (statistically significant)
than
in the
reference
group,
whereas
the
ionised magne-
sium
concentration
was
almost identical
(-1%,
statistic-
ally
not
significant); therefore
the
percentage
of
ionised
magnesium
was
increased. Figure
1
indicates that total
magnesium
is
dependent
on
albumin concentration,
UC-ER:
difference
between magnesium
concentration
in
superna^·
tant after
ultracentrifugation
(determined
by
AAS)
and
complexed
magnesium concentration
as
obtained
from
exchange
resin
(ER)
(determined
by
AAS).
0.90
0.85
*
'-
3
0.80
ff
w
0.75
\
0.70
0.65
0.60
§
10
15
20 25 30
Albumin
[g/1]
35
40
Fig.
1
Fraction
of
ionised
Mg and
albumin
in
renal
diseases
(RD1)
Slope:
-0.002
Intercept
0.83
Coefficient
of
correlation:
-0.44
RD
1:
Renal
disease.
whereas
the
ionised magnesium concentration
is not in-
fluenced.
In
renal
insufficiency
(RD 2)
both,
jthe
ionised
and
total
magnesium concentration
in the
serum were equally
ele-
Külpmann
et
al.:
Ionised magnesium
in
renal
and
hepatic
diseases
261
Tab.
7
Renal diseases: Magnesium
in the
serum,
X
(± s)
Property
Control
1
)
(n
= 60)
1.
Group
(RD 1)
(n
= 21)
2.
Group
(RD2)
(n
= 29)
Total
Mg
0.80
(0.06)
0.73
(0.05)
[-']
0.93
(0.09)
[+]
(mmol/1)
Ionised
Mg
0.57
(0.04)
0.56
(0.04)
0.65
(0.05)
[+]
(mmol/1)
Ionised
Mg
—
0.71
(0.03)
0.77
(0.04)
[+]
0.71
(0.05)
Total
Mg
[ ]
Statistically significant negative
[—]
or
positive
[+]
deviation
as
compared
with
the
mean
of the
control group
!
)
Control: Reference group
of
healthy blood donors
RD
1:
Renal disease; creatinine concentration
in the
serum
within
reference
interval
RD
2:
Renal disease; creatinine concentration
in the
serum exceed-
ing
reference interval
0.90
0.85
0.80
0.75
0.70
0.65
0.60
10
15 20 25 30
Albumin
[g/1]
35
40
Fig.
2
Fraction
of
ionised
Mg and
albumin
in
renal diseases
(RD2)
Slope:
-0.002
Intercept:
0.78
Coefficient
of
correlation:
-0.35
RD 2:
Renal
insufficiency.
vated
(4-
15.6%
vs
+
16.2%)
(tab.
7) as
compared with
the
reference
group,
and the
fraction
of
ionised magne-
sium
was
unchanged. Nevertheless,
the
dependency
of
total magnesium
on
albumin concentration
in the
serum
was
still
obvious
(fig.
2).
Magnesium excretion
in the
urine
was
below
the
reference interval
(2.5—8.5
mmol/d)
and
lower than
in the
patients with
proteinuria
(RD
1),
but the
mean
fractional
excretion
(7.0%
±
3.9%
(standard
deviation))
exceeded
group
RD 1
(3.5%
±
1.2%
(standard
deviation)).
Liver diseases
Ionised
and
total magnesium concentrations
in the se-
rum
of
patients suffering
from
liver
diseases
are
pre-
sented
in
table
8.
In
group
LD
1
with
normal
renal function
but
with
hy-
poalbuminaemia,
the
total
magnesium
concentration
was
decreased
by
6.2%
as
compared
with
the
reference
in-
terval,
and the
ionised
magnesium concentration
was de-
creased
by
only
2.7%
(statistically
not
significant).
Hence,
the
ionised
magnesium
fraction was
elevated.
It
may
be
concluded that total magnesium
is
correlated
with
the
albumin concentration (fig.
3),
whereas ionised
magnesium
is not
significantly
affected.
In
liver diseases combined with renal
insufficiency
(and
therapy
with furosemide),
LD 2
results resembled
RD 2:
the
mean concentration
of
ionised
and
total magnesium
were increased similarly
(17%
vs
13.8%),
and the
ionised magnesium
fraction
was not
significantly
changed. Correlation
of
total magnesium
and
albumin
concentration
may be
supposed
from
figure 4.
In
patients treated with
cyclosporin
(LD 3), the
decrease
of
total magnesium concentration exceeded
the
decrease
of
ionised magnesium
(—17.5%
vs
—11.3%),
and the
fraction of
ionised magnesium
was
significantly
ele-
vated (7.0%).
Its
correlation with albumin
is
presented
in
figure 5.
In
cases
of
treatment with furosemide
and
spironolac-
tone
(LD 4)
only
the
total magnesium concentration
was
significantly
changed
(-7.5%)
and
hence
the fraction of
ionised magnesium
was
elevated
(+
5.6%).
It was
poorly correlated with
the
albumin concentration
(fig-
ure
6).
During
therapy
of 9
patients with spironolactone
as the
only diuretic
(LD 5),
ionised magnesium
was
slightly
decreased (1.9%),
but
total magnesium
was
clearly
(8.8%) lowered
(but
statistically
not
significant with
only
9
patients),
and the
fraction
of free
magnesium
was
significantly
elevated. Because
of the
small number
of
patients
the
correlation with albumin
may be
misleading
and is not
shown.
In
patients treated with diuretics
and
substituted with
magnesium
(LD
6),
ionised magnesium
was not
signifi-
cantly elevated,
total
magnesium
not
significantly
low-
ered,
and the fraction of
ionised magnesium
not
signifi-
cantly increased
but
correlated with
the
albumin concen-
tration (fig.
7).
Discussion
Before
starting
the
study
the
reliability
of the
measure-
ment
of
ionised magnesium
was
evaluated. According
to
the
guidelines
(11),
the
pertinent
coefficient
of
variation
should
not
exceed
2.1%.
In
fact
2.2%
and
2.0%
were
achieved. Accuracy could
not be
directly checked
as no
appropriate control sera
are
available.
But
there
was in-
direct evidence that determinations were accurate
(see
results). Precision
and
accuracy
of the
other methods
was
better than required
by the
guidelines
(11)
(data
not
shown).
Blood
samples
from all
individuals were taken between
7 a. m. and
9.30
a. m. to
exclude
any
influence
of
pos-
sible
circadian
rhythms.