were made acidic before aliquots of all specimens were
stored frozen at -20 #{176}C.
Magnesium
was measured
with a Model SP9 atomic
absorption spectrometer (Pye Unicam
Ltd., Cambridge,
U.K.). Creatinine
was assayed by alkaline picrate colorim-
Technical Briefs (-300 words text) summarize findings
that are of interest to a relatively limited audience. Readers desiring fuller details may obtain them by writing
directly
to the author(s) at the address
ing with procedure
or
instrumentation
given. Briefs dealinteroomparisons,
evaluations, or improvements (including kit applications)
should be sent to Clinical Chemistry News, 2029K Street,
Washington,
etry with a Roche Urn-Kit II and a Cobas-Fara parallel
centriflgual
analyzer
(both from Roche, Welwyn Garden
City, Herts., U.K.). All specimens
were assayed in dupli-
DC 20006.
cate.
Biological Variation of Urinary Magnesium, G. W.
Nicoll,’ A. D. Sti-uthers,2 and C. G. Fraser’ (Depth. of
1Biem.
Med. and2 Clin. Pharmacol., Ninewells
Hospital and Med. School, Dundee DD1 9SY, Scotland)
The diagnosis of magnesium
deficiency and assessment
of magnesium
status is not straightforward,
and assays of
intracellular magnesium concentration are of value (1).
However, these are not easy to do, and simple excretion
tests have become increasingly popular (2). Moreover, the
renal excretion of magnesium is of value for assessing
magnesium
wasting caused by therapy or physiological
states (1).
To assay urinary magnesium
appropriately and interpret the results objectivelyrequires knowledge ondesirable
performance
standards, the utility of conventional population-based
reference intervals, and the significance
of
changes in serial results. Information
on these and other
aspects can be obtained by generation and application of
data regarding
analytical,
within-subject,
subject components of variation (3).
and between-
Eight specimens
of first morning urine, randomly collected (untimed) urine, and 24-h collections
of urine were
obtained from each of 16, 12, and 12 subjects, respectively.
The composition of the respective groups was nine men and
seven women, seven men and five women, and nine men
and three women. All subjects were apparently healthy,
ages 20 to 45 years, and maintained their usual life-styles
throughout. The volumes of the 24-h urine specimens,
which were collected into 50 mL of 3 mol/L HC1, were
noted. The pH of first morning and random urine specimens was checked soon after collection and the samples
Table
1. Components
Analytical, within-subject, and between-subject components of variation were calculated by using nested analysis
of variance and, from the overall means, the coefficients of
variation (CVA, CV1, and CV0, respectively) were calculated. The analytical goals (0.5 CV,), indicesof individuality (CVI/CVG), critical differences required for significant
(P 0.05) changes in serial results [2.77 (CVA2 + CV,2)],
and indices of heterogeneity of within-subject variationthe CV of(CVA2 + CV12)”2 divided by [21(n - 1)]”, where
n is the number of specimens collected per subject-were
calculated as described previously (3), and are listed in
Table 1.
The within-subject
variations of urinary analytes are
greater than those in serum or plasma (3); thus, in theory,
less stringent analytical
goals are applicable.However,
because serum, plasma, and urinary specimens may be
assayed by use of the same technique and even in the same
run, the goal foi’ serum magnesium analyses (CV <1.1%)
should be adopted (4).
For all specimen types, the indices of individuality are
>0.6, implying that conventional reference values will be
of use (3), although the utility of the test as a diagnostic
procedure will depend on the overlap of distributions between healthy and diseased subjects. Interestingly, the
magnesium/creatinine
ratios had the least individuality.
To interpret changes in results from an individual, e.g.,
to assess the effect of magnesium
supplementation on
magnesium
status, one must know the significance of
changes
in results, which can be calculated as 1.414Z
(CVA2 + CVI2)112, where Z depends on the probability
required for significance (4); for P 0.05, Z
1.96. For
urinary magnesium, the critical difference is large compared with that for serum or plasma. Calculation of the
=
of Variation and Derived Indices for Magnesium and Magneslum/Creatinine
Specimens Collected Three Ways
Ratio In Urine
24-h wins
First morning
Mg1
Mean
3.71
3.3
CVA, %
CV,,%
CV0, %
Analyticalgoal,%
Indexof Indlvidualfty
Indexof heterogeneIty
Criticaldifference,%
mmot/L.
58.4
30.3
29.2
1.9
2.57
162
Mg/cru.tlnlnsb
0.29
6.5
85.6
31.0
42.8
2.7
4.03
238
Cmd
1794 CLINICALCHEMISTRY,Vol.37, No. 10, 1991
Random urine
Mg
Mg
Mg1
Mg/crs.IJnlneb
3.34
0.31
6.4
70.9
3.6
46.8
40.7
24.0
42.7
47.2
35.4
23.4
20.3
3.0
1.1
0.9
1.56
1.99
130
1.77
113
3.3
57.4
33.0
28.7
1.7
1.15
159
197
sxcrstlon’
2.68
output#{176}
3.65
3.0
Mgrs.linIn.5
0.28
7.5
75.6
32.6
37.8
2.3
2.58
210
magnesium/creatinine
ratios makes the difference even
greater.
In part because of circadian rhythms in glomerular filtration rate and magnesium excretion, the traditional approach
to the use of urine assays to assess some aspects of magnesium homeostasis is to collect 24-h specimens and report
results as magnesium output. Collection of first morning and
random urine specimens is much easier, and correction for
flow by using creatinine is widespread. However, the ideal
specimen has the following characteristics: objective analytical goals should be easily met; the index of individuality
should be high, making traditional
referencevalues useful;
the critical difference should below, allowing the significance
of changesin results to be of higher probability, and withinsubject variation should have little heterogeneity, so that
data generated in experiments such as those described here
may be ubiquitously applicable. The collection of 24-h urine
specimens
and reporting
of results as magnesium output
most closely approaches this ideal, and this study supports
current practice. However,we point out that the single result
has wide inherent dispersion (±80% forP 0.05), and large
changesin serialresults(±113% forP 0.05) are required for
significance;this make the interpretation of resultsof these
assaysdifficult.
We thank Zoe Baichin and Paul Moss for organizing
tion of urine specimens.
the collec-
References
1. Elm
RJ. Assessment
of magnesium
status.
Clin Chem
1987;33:1965-70.
2 Ryan MF. The role of magnesium in clinical biochemistry: an
overview. Ann Chin Biochem 1991;28:19-26.
3. Fraser CG, Harris EK. Generation and application of data on
biological
variation
in clinical chemistry.
Crit Rev Clin Lab Sci
1989;27:409-37.
4. Fraser
biological
CG. The application of theoretical goals based upon
variation in proficiency testing. Arch Pathol Lab Med
1988;112:405-15.
Serum Laminin P.,In Metastatlc Colon Carcinoma,
Franck Felden,’ Pascal Renkes,2 Sophie Fremont,’
Jean-Francois
Chambre,2 Bruno Champigneulle,2
Pierre
Gaucher,2 and Jean-Louis Gueant’ (1 issari U308,
Equipe de Biochim.-Immunol., Faculte de Medecine, B.P.
184, 54505 Vandoeuvre lea Nancy Cedex; 2 Service des
Maladies de l’Appareil Digestif, CHU de Nancy Brabois,
Rue du Morvan, 54500 Vandoeuvre lea Nancy, France)
Laminin is a high-Mr glycoprotein,
first isolated from a
(1). A protein of basementmembranes,it is
synthesizedby normal cells such as endothelial cells, epimouse sarcoma
thelial cells, muscle cells, and fibroblasts (2) and also by
tumor cells such as HT 29 cells (human colon carcinoma)
(3). The serum concentration of isminin
P1 (a fragment of
laminin
obtained by proteolysis) is increased in various
malignancies
such as digestive carcinoma (esophagus,
stomach, and colon), breast cancer, osteosarcoma, lymphoma, and leukemia (4). It is also increased in chronic
liver diseases, including primary biliary cirrhosis, alcoholic
cirrhosis, and alcoholic hepatitis (5,6). The serum concentration of larninin P, has not been studied in liver tumors.
We have quantified laminin P1 in serum of five groups of
subjects: apparentlyhealthy subjects(n = 69; inclusioncriteria: absence of alcohol intake and no liver or intestinal
disease)and subjects with coloncarcinoma (n = 32; inclusion
criteria: absence of alcoholintake and absence of metastasis),
metastasis (n = 20; inclusion criteria colon carcinoma +
positive hepatic ultrasonography and a high concentration of
carcinoembryonic antigen), alcoholiccirrhosis (n = 50, classified according the Child-Pugh scale: group A = 15 cases,
groupB = 20, and group C 15), and hepatocarcinoma (n =
18; inclusion criteria: positive ultrasonography
+ high a-fetoprotein concentration and histological diagnosis). We measured laminin P1 in serum by using the Bebring (Marburg,
F.R.G.) radioimmunoassay (7). For statistical evaluation, we
used the Mann-Whitney
test for comparison of means.
The concentration of laminin P1 in blood in each group is
=
presented in Table 1. The concentration was significantly
higher in patients with colon carcinoma without metastasis
than in healthy subjects (P <0.001). This increase could
correspond to a synthesis of laminin by the tumor: laminin
synthesis has been demonstrated in primary epithelial
cultures
of colorectal carcinoma (3).
In addition, the patients with colon carcinoma and liver
metastasis
had a higher lsin,inin P1 blood concentration
than did patients without metastasis (Table 1). In our
study, the highest concentration
of lRminin P1 observedin
the cancer patients without metastasis was 1.8 kilounitsfL, but 12 of the 20 patients with metastasis
had a
laminin P1 higher than this. Several mechanisms
might
explain the increased concentration of laminin P1 in patients with metastasis: a synthesis of the protein by the
metastatic cells, or destruction of basement membranes
during the metastatic process. A fibrotic reaction in the
liver might also result in increased laminin P1 in blood
becausehigh concentrations were alsoobserved in alcoholic
liver cirrhosis (Table 1) and in chronic hepatitis (6).
In conclusion,the concentration of serum laminin P1
may potentially be useful for detecting a metastatic extension of a colon carcinoma.
References
1. Orkin RW, Gehron P, Mal Goobwine B, Martin GR, Valentine
T, Swarm
R. A murine
tumor producing
a matrix
of basement
Table 1. DIstrIbutIon of Laminln P., Blood Concentration In Several Groups of Patients
Control
No.otcases
69
LamininP1, kilo-unitslL
Range
0.6-1.5
Mean ± SD
0.98 ± 0.17
Statistical
comparison(Mann-WhItneytest), P
Patientsvs control
Patientsvscoloncarcinoma
-
Colon
carcinoma
Uver
metastasIs
Hepatoma
Cirrhosis
20
18
50
32
0.7-1.9
± 0.25
1.31
<0.001
-
1.2-5.0
2.30
±
1.00
1.5-7.1
3.00
±
1.40
1.1-6.2
3.25 ± 1.30
<0.002
<0.001
<0.001
<0.002
<0.001
<0.001
CLINICALCHEMISTRY,Vol.37, No.10, 1991 1795