ity test for multiple sclerosis: Implications for pathogenesis and
handling of the disease. J Neumi 214, 113-127 (1977).
3. Tainblyn CH, Swank RL, SeamanGyP, Zukoski CF W. Redcell
electrophoretic mobility test for early diagnosis of multiple sclerosi8. Neurol Res 2, 69-83 (1980).
4. GuI S. Smith AD, Thompson RHS, et al. Fatty acid composition
of phospholipids from platelets and erythrocytes in multiple sclerosis. J Neurol Neurosurg
Psychiat,y 33, 506-510 (1970).
5. Boulton CH, Hampton JR, Phillipson OT. Platelet behaviour
and plasma phospholipids in multiplesclerosis.
Lancet ii, 99-104
(1968).
6. Homa ST, Belin J, Smith AD, et al. Levels of linoleate and
arachidonate in red blood cells of healthy individuals and patients
with multiple sclerosis. JNeurolNeurosurgPsychiatiy
43, 106-110
(1980).
7. Dore-Duffy P. Zurier RB. Lymphocyte adherence in multiple
sclerosis:Role of monocytes and increased sensitivity of MS lymphocytes to prostaglandin E. Gun Immunollmmunopathol
19,303-313
(1981).
ethers. Indust Eng Chem Prod Res And Dev, in press.
12. Walter H, Krob EJ, Brooks DE. Membrane surface properties
other than charge involved in cell separation by partition in
polymer aqueous two-phase systems. Biochemistry 15, 2959-2964
(1976).
13. Walter H, Krob EJ, Tung R. Hydrophobic affinity partition in
aqueous two-phase systems of erythrocytes of different species.
Systems containing polyethylene glycol-palmitate. Exp Cell Res
102, 14-25 (1976).
14. Eriksson E, Albertsson PA, Johansson G. Hydrophobic surface
properties of erythrocytes studied by affinity partition in aqueous
two-phase systems. Mol Cell Biochem 10, 123-128 (1976).
15. Harris JM, Yalpani M, Van Alstine JM, et al. Synthesis and
characterization of poly(ethylene glycol) derivatives. J Polymer Sci
Polymer Chem Ed, in press.
16. Mendenhall W, Scheaffer RL. Mathernoiica.l Statistics, with
Applications, Duxbury Press, North Scituate, MA, 1973, pp 539-
542.
Lancet i, 1139 (1979).
10. Van Alstine JM, Hon K, Frohlich J, Brooks DE. Lecithin:cholesterol acyltransferase activity in plasma of patients with multiple
sclerosis. Gun Chem 29, 1995 (1983). Letter.
11. Harris JM, Case MG, Hovanes BA, et al. Purification of
17. Love CW, Cashell A, Reynolds M, Callaghan N. Linoleate and
fatty acid patterns of serum lipids in multiple sclerosis and other
diseases.Br Med J iii, 18-21 (1974).
18. Butterfield DA, Markesbery WR. Specificity of biophysical and
biochemical alterations in erythrocyte membranes in neurological
disorders. J Neurol Sci 47, 261-271 (1980).
19. Pavone L, Klose G, Mollica F, et al. Increased serum phospholipids in epileptic children treated with phenobarbitone.
J Neural
Neurosurg Psychiatry 45, 81-82 (1982).
20. Sharp KA. Cell partition and surface free energies related
through contact angles in two polymer phase systems. Biophys J 41,
biomaterials by phase-partitioning with poly(ethylene glycol)-alkyl
353a (1983).
8. Tsang ‘NM, Belin J, Monro JA, et al. Relationship between
plasma and lymphocyte linoleate in multiple sclerosis. J Neurol
Neurosurg Psychiatry 39, 767-772 (1976).
9. Seaman GVF, Swank RL, Zukoski CF, IV. Red cell membrane
differences in multiple sclerosis are acquired from the plasma.
CLIN. CHEM. 30/3, 443-446 (1984)
Direct Analysis for Urinary Protein with Biuret Reagent, with Use of Urine
Ultrafiltrate Blanking: Comparison with a Manual Biuret Method Involving
Trichloroacetic Acid Precipitation
John H. Eckfeldt,”2 Marcia J. Kershaw,’ and Irven I. Dahi’
We describe a method for measuring urinary protein with a
centrifugal analyzer. Biuret reagent is used, and blanking
with an ultrafiltrate of urine eliminates interferences from the
nonprotein, biuret-positive chromogens in urine. We compare results by this new method with those by a manual
method in which trichloroacetic acid precipitation and biuret
reagent are used. The new method shows good precision
and excellent correlation (r = 0.997) with the manual method.
The ease and convenience of this assay should make this a
useful method for the routine clinical laboratory.
Additional Keyphrase: centrifugal analyzer
Not all hyperproteinuria
is clinically
significant,
but
persistently
high concentrations
of protein in the urine may
indicate renal or urinary-tract
disease. Thus, detection and
quantification
of urinary proteins are important laboratory
procedures (1), for which many methods have been devised.
‘Laboratory Service 113, Veterans Administration Medical Center, Minneapolis, MN 55417.
2Depastment of Laboratory Medicine and Pathology, University
of Minnesota, Minneapolis, MN.
Received September 19, 1983; accepted November 28, 1983.
data from interlaboratory
surveys show poor
comparability of results within and among various quantitative methods
as well as frequent instances of poor
intralaboratory
precision. The methods most often used by
clinical laboratories to quantify
urinary protein are acid
precipitation turbidimetry, protein dye-binding colorimetry,
and biuret reagent with prior acid precipitation of protein
Alkaline benzathomum chloride precipitation turbidimetry (9) is also likely to be widely used as a consequence
of its recent introduction
for use in the Du Pont aca discrete
analyzer. The major problems with all direct precipitation
turbidimetric methods are the difficulty in obtaining reproducible and stable suspensions of the protein precipitate and
the inaccuracies caused by interfering substances and variability in the composition of proteins appearing in urine (5,
6, 9-11). The Ponceau S and Coomassie Brillant Blue dyebinding methods are also subject to inaccuracies caused by
differences in urinary protein composition, although their
precision is somewhat better than that of the acid precipitation turbidimetric methods (5, 6).
Biuret-based methods, being generally accepted as more
nearly accurate, have been proposed as the basis for standardizing quantification of urinary protein (12). However,
However,
(2-4),
(2-8).
various
nonprotein,
biuret-positive
CLINICAL
CHEMISTRY,
interfering
substances
Vol.30, No. 3, 1984 443
are a problem. Two general approaches have been used for
their removal.
Sephadex gel-filtration,
to isolate the macromolecular components of urine, has been used in a Selected
Method (13). Alternatively,
the protein components of urine
are precipitated
with trichloroacetic or phosphotungstic
acid, after which the acid-soluble interferences are removed
in the decanted supernate (7, 8).
We have approached the problem of biuret-positive, low
molecular-mass components in urine somewhat differently.
Rather than trying to remove them before biuret colorimetry, we assay the urine with the interfering small molecules
present and, in parallel, assay an ultrafiltrate
of urine to
quantifr
the contribution
apparent urinary
protein.
of these small molecules to the
We describe here this new aphow results compare with those by a
proach and detail
manual biuret method
chioroacetic acid.
involving
precipitation
with
tri-
Materials and Methods
Materials
Reagents.
Human
albumin was from the American Red
Cross Blood Services, Washington, DC 20006; human gainma-globulin from Cutter Biological, Emeryville, CA 94608;
and bovine albumin from Sigma Chemical Co., St. Louis,
MO 63178. Piperacillin (Pipracil#{174})
was from Lederle Pipracillin, Inc., Carolina, PR 00630; nafcillin (Unipen#{174})
from
Wyeth Laboratories,
Inc., Philadelphia,
PA 19101; and
mezlocillin
(Mezlin#{174})
from Miles Laboratories,
Inc., West
Haven, CT 06516. All other reagents were Alt grade or
better.
Samples.
Samples were from 24-h urine specimens submitted for measurement
of urinary
protein. Samples were
initially
analyzed by the manual method (8), and then
stored at 4 #{176}C
for no longer than 48 h before analysis by the
present method.
Control.s. Collect urine from normal subjects and store it
refrigerated. Before analysis, filter through Whatman no.
42 filter paper to remove particulate material, add 1 g of
sodium aside per liter, add an appropriate amount of bovine
albumin for abnormal
controls, and store 3-mL aliquots
frozen at -20 #{176}C.
Such aliquots are stable for at least six
months.
Procedures
Ultrafilt ration procedure.
Add 1.5 mL of centrifuged (1000
x g, 10 mm) urine to a Centrifree#{174}
Micropartition
system
MPS-1 ultrafilter (prod. no. 4010; Amicon Corp., Danvers,
MA 01923) equipped with an anisotropic,
hydrophilic
YMT
ultrafiltration
membrane (prod. no. 40420). This apparatus
and membrane
were originally
designed for separation
of
low-molecular-mass compounds such as drugs from protein
in serum. Centrifuge the sample in the ultraffitration
apparatus for 10 mm at 1500 x g in a fixed-angle rotor to
produce about 0.5 mL of ultrafiltrate.
Some lots of the YMT membrane contain a water-soluble,
biuret-positive substance that elutes into the urine ultrafiltrate, causing spuriously high results for urinary interferents in the ultrafiltrate
and, consequently, low calculated
results for urinary protein. To resolve this problem, we soak
all new membranes in de-ionized water for 15 mm and blot
before use. Membranes can be stored in 0.15 mol/L sodium
chloride solution containing 2 g of sodium aside per liter and
reused, but care must be taken not to damage or puncture
them.
Biuret analysis procedure. Analyze the original urine and
the above ultraflltrate
by a serum biuret method, increasing
the sample volume by about 10- to 20-fold to compensate for
444
CLINICAL CHEMISTRY, Vol. 30, No. 3, 1984
the lower protein concentration in urine. We used a Multistat ifi Microcentrifugal
Analyzer (Instrumentation Laboratory, Lexington, MA 02193), with the following reagents
and procedure:
Biuret reagent: Instrumentation
Laboratory’s Biuret Reagent Kit (prod. no. 35173), which contains, per liter, 333
mmol of sodium hydroxide, 66 mmol of potassium sodium
tartrate, 50 mmol of potassium iodide, and 20 mmol of
cupric sulfate.
Working urinary protein standard, 4 g/L: Dilute the 80
g/L bovine serum albumin standard provided in the reagent
kit 20-fold with 0.15 mol/L sodium chloride solution containing 1 g of sodium aside per liter. This standard is stable for
at least six months at 4 #{176}C.
Load the Multistat III cuvette with samples and controls,
using the following settings:
Sample syringe:
Sample volume: 80 L (80%)
Sample plus wash volume: 80 ,aL (80%)
Reagent syringe:
Reagent volume: 125 iL (50%)
Reagent plus wash volume: 125 L (50%)
Reagentidiluent
switch: reagent
Second-reagent
switch: off
Place biuret reagent in the reagent boat. In the sample
ring, place 200 .uL of water in cup 1, 200 pL of urinary
protein working standard in cup 2, and controls or patients’
urines and urinary ultrafiltrates
in the remaining cups.
Place the loaded rotor into the analyzer. Parameters for
the bichromatic endpoint analysis are: substrate II tape,
program 42 (Total Protein); first filter, no. 7 (620 nm);
second filter, no. 6 (550 nm); delay to first (620 nm)
absorbance reading, 600 s; delay from first (620 nm) to
second (550 nm) absorbance reading, 20 5; number of data
points, 2; start mode 2 (no incubation to temperature before
mixing).
Calculate the urinary protein concentration by subtracting the “protein” result for the urinary interferents in the
ultrafiltrate
(usually between 0.2 and 2.0 g/L, depending on
the total 24-h urine volume) from the “protein” results for
the unfiltered urine.
Results and Discussion
As shown in Table 1, the ultrafiltrate-blanked
method for
urinary protein has precision
similar
to that of the widely
used quantitative
urinary protein methods. Upon addition
of 4 g of bovine albumin, human albumin, or human
gamma-globulin per liter to 16 randomly selected, hospitalized patients’ urines, mean analytical recoveries were 99%
(CV 6.1%), 98% (CV 7.0%), and 96% (CV 7.9%), respectively.
The response of the method is linear with concentration to
10 gIL. Urines with apparent protein exceeding 10 g/L are
very rarely encountered, but can be diluted with 0.15 mol/L
saline before analysis. Depending on the instrumentation
selected for protein analysis, precision in the low protein
range may be improved somewhat by increasing the volume
of urine and urinary ultrafiltrate
relative to reagent volume. However, the range of method linearity is thereby
truncated.
Using this method primarily to follow patients
with mild to moderate proteinuria, we chose assay conditions that would obviate frequent urine dilution, yet be
reasonably precise in this concentration range for urinary
protein.
As shown in Figure 1 and Table 1, the accuracy of the
proposed method, as assessed by comparison with the trichloroacetic acid precipitation biuret method, is better than
that for any of the dye-binding, acid precipitation turbidimetric, or alkaline benzathonium
chloride precipitation
Table 1. RepresentativeBetween-BatchPrecision
of Various Quantitative Urinary Protein Methods
Protsin, g/L
Apparent protein, g/Lb
cv,
Mean SD
%
n Ret.
Proposed ultrafiltrate-blanked biuret method:
Normal urine control
0.28 0.051 177 19 -#{149}
Medium-protein urine control
0.89 0.054 6.1 24 .3.71 0.098 2.7 24
High-protein urine control
Other widely used methods:
Tnchloroacetic acid turbidimetry
Sulfosalicylicacid turbidimetry
Benzethoniumchlorideturbidimetry
CoomassieBrilliantBlue
colorimetry
Ponceau S colorimetry
Trichloroacetic acid biuret
colorimetry
0.62
0.62
0.81
0.79
0.62
0.72
0.62
0.80
2.03
Table 2. interference with Urinary Protein
Analysis by Penicillin-Type Antibiotics
0.046
7.4
11.3
4.4
0.070
0.035
0.036
0.087
0.035
0.036
10
10
20
5b
Sc
4.6 20
14.0 10
4.9 20
5.8 10
0.044 5.5 20
3.9 25
0.079
#{149}
Datafrom ourown laboratory.
belation
with trichloroacetic acid precipitation biuret method:r
6
6
6
5b
6
5
a
= 0.98
(5).
Correlation withthetrichloroacetic
acidprecipitation-biuret
method:r
0.96(5).
C
=
8
Adult dose,
g/24 h
1-12
Penicillin
G
2-12
Ampicillin
At minimum
At maximum
dose
0 (neg.)
0 (neg.)
0 (neg.)
0 (neg.)
0.2(1+)
dose
0 (neg.)
0 (neg.)
1.1 (2+)
Nafcillin
2-12
30
3.3 (4+)
Mezlocilliri
6-18
60
Piperacillin
12-24
70
3.1 (4+)
#{149}
Normal parenteral dose arid approximate percent that is excreted unchangedin the urine (15, 16).
#{176}Apparent
protein concentrationsmeasured with the quantitative tnchloroacetic acid biuret method (and qualitative sulfosalicylic acid precipitation, in
parentheses),as describedin reference8. in protein-free,normalhumanurine
supplemented
withantibioticat a concentrationcalculatedfrom the renally
excreted fraction, the maximum
or minimum dose, and an assumed24-h
volumeof1 L.
line copper solutions. Although drug interference
can sometimes be recognized by abnormal colors of the protein
precipitate itself or after addition of biuret reagent, antibiotics have caused misdiagnosis of proteinuria in patients,
which may be more common than is generally recognized
(17, 18). With the proposed method, addition
of similar
concentrations
0
% renaily
excreted
70
70
of nafcillin,
piperacillin,
or mezlocillin
to
normal urine or urine supplemented
with 5 g of pooled
serum protein per liter merely increased the apparent
“protein” in the ultrafiltrate
blank and unfiltered urine
equally, leaving the amount of measured urine protein
C
6
unchanged.
I-
-.a.
In conclusion, the proposed ultrafiltrate-blanked
biuret
method for urinary protein is accurate, precise, and reasonably free of interferences; its ease and convenience should
make it a very useful method for use in the routine clinical
laboratory.
4
I-
C
I..
2
a
References
0
I-
0
0
0
2
Trichioroacetic
4
Acid
6
8
Precipitation,
Biuret Urinary
Protein,
gIL
Fig. 1. Comparison of the direct biuret, u(trafiftrate-blankedurinary
prote,n method with the manual trichloroacetic acid precipitation, biuret
urinary protein method (
Linearregressionanalysisof the data (n = 28) gives a slope of 0.990 andan
interceptof 0.05 g/L (r = 0.997). The inset expands the data for unnes with
proteincontent<1 g/L. For these data (n = 13),the slope is 0.809, the intercept
0.11 g/L (r = 0.957).The lines indicateperfectcorrelation,for companson
methods. The lack of correlation of these other
methods with the biuret-based method is even more obvious
if one examines actual scattergrams
(e.g., see
A further
concern with some dye-binding methods is serious lack of
linearity of urinary protein results with dilution (14).
Another advantage of the ultrafiltrate-blanked
method
over direct acid precipitation turbidimetric
and acid precipitation biuret methods is the absence of positive interference
from certain renally excreted drugs; e.g., most penicillintype antibiotics, which are excreted in large quantities. As
shown in Table 2, some of these can significantly interfere
with both qualitative and quantitative
acid precipitation
turbidimetric
5).
biuret-based
methods, owing to their insolubility
solution and their formation
of colored complexes
in acid
in alka-
1. Manuel Y, Revillard JP, Betuel H, Eds. Proteins in Normal and
Pathological Urine, S Karger, Basel, Switzerland, 1970.
2. Glenn CC. Evolution of the urine chemistry survey program of
the CAP. Am J Clin Pathol 72, 299-305 (1979).
3. Glenn CC. The results of analyte enhancement and use of
supplied urine protein standard in the CAP urine chemistry survey
program. Am J Gun Pathol 74, 531-534 (1980).
4. Shepard MDS, Penberthy LA, Fraser CG. A quality assurance
programme for quantitative urine analyses. Pathology
14, 327-331
(1982).
5. McElderry LA, Tarbit IF, Cassells-Smith AJ. Six methods for
urinary proteins compared. Clin Chem 28, 356-360 (1982).
6. Dilena BA, Penberthy LA, Fraser CG. Six methods for urinary
proteins compared. Clin Chem 29, 553-557 (1983).
7. Savory J, Pu PH, Sunderman FW. A biuret method for determination of protein in normal urine. Clin Chem 14, 1160-1171
(1968).
S. Tietz N. Fundamentals
of Clinical Chemistry, 2nd ed., Saunders,
Philadelphia, PA, 1976, pp 358-363.
9. Iwata J, Nishikaze 0. New micro-turbidimetric
method for
determination of protein in cerebrospinal fluid and urine. Ciin
Chem 25, 1317-1319 (1979).
10. Schriever H, Gambino SR. Protein turbidity produced by tnchloroacetic acid and sulfosalicylic acid at varying temperatures
and varying ratios of albumin and globulin. Am J Gun Pat/wi 44,
667-672 (1965).
11. Lizana J, Bnito M, Davis MR. Assessment of five quantitative
methods for determination of total proteins in urine. Gun Biochem
10, 89-93 (1977).
12. Peters T. Proposals for standardization of total protein assays.
Clin Chem 14, 1147-1159 (1968).
CLINICAL CHEMISTRY, Vol. 30, No. 3, 1984
445
13. Doetach K, Gadsden RH. Determination of urinary total protein by use of gel filtration and a modified biuret method. Selected
Methods Ciin Chem 8, 179-182 (1977).
14. Shahangian S, Brown P1, Ash KO. A Coomassie Blue dyebinding method for determining urinary protein is dilution-dependent. Clin Chem 29, 1452 (1983). Letter.
15. Physicians’ Desk Reference, 37th ed., Medical Economics Co.,
Inc., Oradell, NJ, 1983.
16. Mandell GL, Sande MA. Antimicrobial agents: The penicillins
and cephalosponins. In The Pharmacologic Basis of Therapeutics,
6th ed., LS Goodman, A Gilman, Eds., Macmillan, New York, NY
1980, pp 1126-1150.
17. Line DE, Adler S, Fraley DS, Burns FJ. Massive psuedoprotein-
uria causedby nafcillin. JAm Med Assoc 235, 1259 (1976).
18. Andrassy K, Ritz E, Kodensch J, et al. Psuedoproteinuria
patients taking penicillin. Lancet ii, 154 (1978). Letter.
in
CLIN. CHEM. 30/3, 446-449 (1984)
Quantification of Nonenzymically Glycated Albumin and Total Serum Protein
by Affinity Chromatography
Randall W. Yatscoff,1Gerald J. M. Tevaarwerk,2 and J. Courtney MacDonald2
We have evaluated
an affinity-chromatographic
procedure
for determination of glycated albumin (GA) and glycated total
serum protein (GSP). Recovery of these analytes was inversely related to free glucose concentration, thus necessitating removal of free glucose. For this we used molecularexclusion chromatography on G-25 Sephadex, or dialysis,
the latter procedure resulting in significantly (p < 0.05) lower
concentrations
of GSP and GA. Total protein concentration
and percent glycation
protein concentrations
are also inversely related, and so
must be standardized
before the
assay. Within- and between-run CVs for both GSP and GA
were <6.5 and 18%, respectively, the determination of GA
being generally the more precise of the two. Labile glycated
fractions, lipemia, icterus, hemolysis, and type of anticoagulant did not affect the results, but assay temperature did.
Diabetic subjects showed substantially higher concentrations
of GA and GSP than did normal subjects. Because of the life
span of these analytes in circulation, their measurement may
provide a short-term index of glycemic control.
AdditIonal Keyphrases: affinity chromatography
diabetes
glycemia
glycated hemoglobin
reference inte,val
dialysis
short-term index to control of
In nonenzymic “glycation,” the terminal amino group of
glucose reacts with free amino moieties of protein, forming
an aldimine-labile
Schiff’s base that subsequently undergoes an Amadori rearrangement
to form a stable ketoamine
adduct (1, 2). Of the many proteins that are susceptible to
nonenzymic glycation, glycated hemoglobin has been the
most widely studied (2, 3), primarily because of its usefulness as an index of long-term (two to three months) control
of glycemia (2-4). A short-term
(two to three weeks) indicator of glycemic control is also needed, for monitoring the
effects of changes in diet or with insulin therapy (2). For this
the determination
of nonenzymically
glycated albumin and
glycated total protein in serum has been suggested (2,5-17):
their biological half-lives in the circulation are 17 and 30
days, respectively, compared with 120 days for hemoglobin
(2).
‘Department of Clinical Chemistry, Health SciencesCentre, 700
William Ave., Winnipeg, Manitoba, Canada, R3E 0Z3.
2Department of Biochemistry and Medicine, St. Joseph’s Hospital, 268 Grosvenor St., London, Ontario, Canada, N6A 4V2.
Received August 15, 1983; accepted November 28, 1983.
446
CLINICAL CHEMISTRY, Vol. 30, No. 3, 1984
Glycated albumin and total serum protein have been
measured with ion-exchange and colorimetric procedures (2,
5-17). These methods are relatively imprecise, subject to
interference from free glucose and enzymically
glycated
protein, and too time-consuming for routine use in a clinical
chemistry
laboratory.
Recently, an affmity-chromatographic
method in which
glycated proteins are separated from nonglycated proteins
on the basis of interactions of glycosyl moieties with boronic
acid has been adapted for measuring glycated hemoglobin
(18, 19). Here we report our evaluation of a commercially
available affinity method as an alternative means of quantifying glycated albumin and glycated total serum protein.
Materials and Methods
Specimens
At random times during the day blood was collected into
tubes containing either EDTA or no anticoagulant,
from
laboratory personnel with no history of diabetes and from
diabetic subjects who were being treated by diet, oral
medication, or insulin. The samples were centrifuged at
1000 x g for 5 mm to separate cellsfrom serum or plasma.
Hemolysates for determination of glycated hemoglobin were
prepared as described below. All samples were stored at
-80 #{176}C
until analysis.
Procedures
Quantification
of glycated albumin
and glycated
total
serum protein. Nonenzymically
glycated albumin and total
serum protein were quantified with a kit from Isolab Inc.,
Akron, OH, as follows.
1. Protein standardization and glucose elimination. Standardize the concentrations of total protein and albumin to 50
and 30 g/L, respectively, by adding appropriate volumes of
isotonic saline. Then remove free glucose from the sample
by gel ifitration through a pre-packed Sephadex G-25 column (included in the kit) as follows. Add 200 iL of serum or
plasma to the column, followed by 100 pL of sample wash
buffer (per liter, 0.1 mol of MgCl2, 0.1 mol of glycine, Triton
X-100, and 0.1 g of sodium azide). Once the column drains,
add 500 p.L of wash buffer, and again allow to drain,
discarding both eluates. Add 400 tL of wash buffer to the
column, allowing the eluate top
directly onto the top disc
of the affinity column. Do not re-use these columns.
To eliminate free glucose by dialysis, use tubing having a