gested as a guide to therapy with iron, in view of the difficulties in estimating. Table 1. Comparison of 10. Serum Ferritin Kits for Sera from Chronic Hemodialysis.
The addition of ammonium hydroxide to serum, as recommended by Gambino and Schreiber, can falsely increase the value for potassium when measured
with
the Astra
8. We corn-
pared potassium concentrations in 10 sera before and after the addition of ammonium hydroxide; the mean (and the range) was 4.5 (4.1-5.2) and 4.8 (4.3-5.4) mmol/L, respectively. However, we find that lithium hydroxide, at a final concentration of 2.5 mmolIL (5 L of 0.75 rnolIL lithium to 1.5 mL of serum),
hydroxide added will increase the
serum pH to about 8.2 and stabilize carbon dioxide
for at least 4 h at room
temperature. Thus lithium hydroxide can be used instead of ammonium bydroxide
to stabilize
carbon
dioxide
in
serum. Because lithium ions in excess of 2.5 mmolIL will falsely increase the value
for sodium concentration as measured with the Astra 8, one should be aware that the addition of lithium hydroxide to a serum
from a patient relithium salts may the value for sodium
#{149}:
Kitb
Becton Dickinson
0.95
Clinical
0.96
-7
0.99
0.94 0.95
3 10
1.08 1.12
Assays
NEN
DPC Pharmacia
4
0.85
0.96
16
1.03
Hybritech
0.94
Nordiclab
0.88
-9 31 6 19 36
0.91 2.70 0.75 0.51 0.78
Ramco 0.92 CIS 0.87 Amersham 0.89 a Sara from 36 chronic-hemodialysis patients assayedaccordingto kit instructions.Only patients with values for serum ferritin within the calibration curv!s of all 10 kits were taken into account. L
and b are the means of the nine linear correlation coeflicients, intercepts, and slopes, respectively, for results obtained with the kit under consideration (y) and resultsobtained with each of the other kits. S Kit enumerated = y.
sample
ceiving therapeutic falsely
North Billerica, MA 01862; Becton Dickinson and Co., Orangeburg, NY 10962; Nordiclab, P.O. Box 3590121 Oulu 12, Finland; and Diagnostic Products Corp (DPC), Los Angeles, CA 90045.
Table 1. Comparison of 10 Serum Ferritin Kits for Sera from Chronic Hemodialysis Patientsa
increase
concentration. References 1. Gambino SR, Schreiber H. Measurement of COz content with the AutoAnalyzer. A comparison of a new method (alkalinization) for preventing loss of CO2 from open cups. Am J Clin Pat/wi 45, 406-411 (1966). 2. Beckman Astra 8 Sodium-Potassium Operating and Service Instruction Manual. Beckman Instruments, Inc., Clinical Instruments Division, Brea, CA 92621, 1979.
George R. Gotelli Pokar M. Kabra Div. of Gun. Chem.
Dept. ofLab. Med. School of Medicine Univ. of California, San Francisco San Francisco, CA 94143
Table 2. Comparison of Six Serum Ferritin Kits Showing Good Linear Correlations for Sera from Chronic Hemodialysis Patientsa Kftb f A B Becton Dickinson Clinical Assays NEN DPC
0.98 0.98 0.97 0.97
1.8 7.5 -6.0 -10.4
1.16 0.88 0.84 0.95
Pharmacia
0.98
20.5
1.06
Hybritech
0.97
6.5
1.00
Sera from 36 chronic hemodialysis patients were assayed for serum ferritin according to the instructions of the manufacturers. Only patients with serum ferritin levels within the calibration curvesof all 6 kits were taken into account. f: mean of the five linear correlation coefficients between resultsobtainedwith the kit under consideration (y) andresults obtained with each of the other kits; A: mean of the five intercepts of the linear relationships between the kit under consideration (y) and each of the other kits (cg/L); : mean of the five slopes of the linear relationships between the kit
underconsiderationty) and each of the other kits. b Kit enumerated = y.
Ten CommercialKits for the Determination of Serum Ferritin Compared by Use in Chronic
HemodialysisPatients To the Editor: Van Oost et al. compared four serum ferritin radioimmunoassay (lilA) kits with sera from 35 healthy subjects (1). They observed a good correlation between results with the four, but the slopes of the regression lines differed widely. Using 10 different commercial serum ferritin kits, we obtained similar results on dialysis patients. Indeed, ferritin determinations in sera from dialysis patients have often been suggested as a guide to therapy with iron, in view of the difficulties in estimating 994
Only patients
whose
ferritin
concen-
trations in serum fell within the limits of the calibration curves of all 10 kits were taken into consideration. The serum ferritin kits were compared two by two. In all instances the linear correlation coefficient, r, and the equation y = a + hr, of the best-fitting line were calculated (Table 1). As compared with the other six, one kit (Nordiclab) yielded obviously higher values and three kits (Ramco, CIS, Amersham) produced lower values. After these four kits
were
excluded,
the
relationship
between the others was expressed as the mean value ofr, a, and b (Table 2). The correlation coefficients of this group of six kits, appeared to be higher UP= 0.97) than between all 10 kits (f = 0.92). The fact that in five of these six kits liver is used as the source of antigen might
be responsible
for their bet-
ter linear correlations. Van Oost et al. suggest that in pathological sera with a very high ferritin content the linear correlation between measurements with different serum ferritin kits might be disturbed. The data we collected in our study are not appropriate for testing this hypothesis, because we only took into account serum ferritin values that were within the limits of the calibration curves of the serum ferritin kits. In our experience, difficulties may arise with some kits in determining serum ferritin values that initially fall above the calibration curve by diluting and reassaying the samples. Although
serum
sured with these sera from uremic correlations, the fered markedly. We thank the
ferritin
as mea-
10 commercial kits in patients showed good absolute values difmanufacturers
of the
commercial serum ferritin kits for making their products available with-
out charge. the iron needs of such patients means.
by other
We prepared sera from blood, collected before dialysis, from 36 chronic hemodialysis patients. Commercial rum ferritin kits were obtained
se-
from Clinical Assays, Division of Travenol Laboratories, Inc., Cambridge, MA 02139; Ramco Laboratories Inc., Houston, TX 77098; Hybritech Inc., San Diego, CA 92121; Pharmacia
Diagnos-
tics AB, Box 17, 5-75103, Uppsala 1, Sweden; CIS, Laboratoire des Produits Biomedicaux, ORIS, Commissariat a l’Energie Atomique, B.P. 21-91190 Gifstir-Yvette, France; The Radiochemical Centre,
UK;
CLINICAL CHEMISTRY, Vol. 29, No. 5, 1983
Amersham,
New
England
Buckinghamshire,
Nuclear
(NEN),
References 1. van Goat BA, Willekens FLA, van Neerboa BR, van den Bald B. Implications of using different tissue ferritins as antigens for ferritin in serum: four radioimmunoassay kits compared. Clin Chem 28, 2429-2433 (1982). 2. Hussein S, Prieto J, O’Shea M, et al. Serum ferritin assay and iron status in chronic renal failure and haemodialysis. Br Med J I, 546-548 (1975). 3. Mirahinadi KS, Paul WL, Winer RI, at al. Serum ferritin level. Determinant of iron requirement in hemodialysis patients. J Am Med Assoc 238, 601-603 (1977). 4. Eschbach JW, Cook JD, Seribner BH, Finch CA. Iron balance in hemodialysis
patients.
Ann
Intern
Med 87, 710-713
(1977). 5. Aljama P, Ward MK, Pierides AM, et al. Serum ferritin concentration: A reliable guide to iron overload in uremic and hemodialyzed patients. Clin Nephrol 10, 101-104 (1978). 6. Ellis D. Serum ferritin compared with other indices of iron status in children and teenagers undergoing maintenance hemodialysis. Clin Chem 25, 741-744 (1979). 7. Milman N, Christensen TE, Pedersen NS, Visfeldt J. Serum ferritin and bone
marrow iron
in nondialysis,
peritoneal
dial-
ysis and hemodialysis patients with chronic renal failure. Acta Med Scand 207, 20 1-205 (1980). 8. Bell JD, Kincaid WR, Morgan RG, at al. Serum ferritin assay and bone-marrow iron stores in patients on maintenance hemodialysis. Kidney ml 17, 237-241 (1981). 9. BirgegArd G, Nilsson P, Wide L. Regulation of iron therapy by S-ferritin estima-
mg/L in a Du Pont aca, in which an endpoint hexokinase procedure with bichromatic photometry at 340/383 nm
curve has drawbacks for accuracy and ease of handling. Recently, Huang and Tietz (2) re-
isused.
ported
The ultraviolet spectrum for metronidazole shows an absorption peak at 325 nm, with a molar absorptivity of 9.5 x lO3Lmol’cm’.At340nmthe
combination with other amylase reagents (BMD Single-Vial Amylase Reagent Kit of Boehringer Mannheim and the Beckman Enzymatic AmylaseDS Reagent Kit). They not only claimed to use about 12% of the inhibitor concentration proposed by O’Donnell et al. (1), but they also showed a linear standard curve for the calcula-
molar
absorptivity
is
6.6 x iO
L
mol’ cm’, slightly higher than the value offi.3 x i0 for NADH, which is the measured end product of hexokinase methodology. At 383 nm the molar absorptivity is 0.3 x i0 L mol’ cm1. Interference by metronidazole is probably
ascribable
to its absorption
at
340 am. This mechanism would account for an increase in apparent glucose concentration of 480 mgIL in a pool containing 450 mg of metronidazole per liter. In the case of the aca the interference was 590 mgIL, presumtions in patients on chronic hemodialysis. ably because the interference filters in Scand J Urol Nephrol 15, 69-72 (1981). this instrument have a wide enough 10. Muller HAG, Schneider H, H#{246}velbom bandpass that the effective absorptivU, Streicher E. Ferritin: A reliable indicator ity of metronidazole is increased. The of iron supplementation in patients on much greater interference in the case chronic hemodialysis/hemofiltration treatof src may be the result of metroniment? ArtifOrgans 5, 168-174 (1981). dazole’s passing the dialyzer memFrank L. Van de Vyver brane more quickly than glucose. Pierre P. Biockx Our patient was on a standard dosWaiter Bleys age of metronidazole, which, according Ludo Muyile to the package insert provided by Searle, should produce peak concentraDept. ofNuclear Medicine tions in plasma of about 25 mg/L, a Univ. Hosp. of Antwerpen concentration that would cause glucose Wilrijkstraat 10 measurement errors of minor conseB2520 Edegem, Belgium quence (60 mg/L or less). From our data, we can estimate that one specimen had a drug concentration of about 200 mg/L. This high value may have been because the patient was in hepatic failure, which decreases metronidaMetronidazole Interference in zole clearance, or because the plasma HexokinaseGlucose was contaminated with insufficiently Determinations mixed intravenous medication. To the Editor: Recently, our laboratory happened to measure glucose in a patient’s specimen with two different instruments: a Beckman Astra-8, in which glucose oxidase and an oxygen-sensitive electrode are used, and a Technicon SMAC,
in which a bound-hexokinase
William
V. Dorwart
The B,yn Mawr Hosp.
inhibitor
and salivary
in
isoamy-
lase from the remaining activity after inhibition. This result is then cornmented on as follows: “The demonstrated
linear
relation
contrasts
the results of O’Donnell
with
et al., who
found a non-linear relationship when they used the ‘Phadebas’ method and
the same inhibitor. This may have been the result of either a different affinity of inhibitor to substrate, or the longer reaction period, which would allow a greater
degree ofdissociation
of
the enzyme-inhibitor complex, as discussed above.” I would like to comment on both statements.
In the publication of O’Donnell et al. (1) no evidence can be found of an inhibitor concentration about eightfold that used by Huang and Tietz (2), either during the incubation period or after addition of reagent. Admittedly the absolute amount of inhibitor used by Huang and Tietz is only 12% of that added in the original method. But this amount is in a volume of 10 L added to 50 L of sample for the preincubation in their case, while O’Donnell et al. perform the incubation with 0.3 mL
of buffer, 0.2 mL of sample, and 10 L of inhibitor solution. Therefore the inhibitor concentration used by Huang and Tietz is in fact 102% ofthe original concentration during the preincubation. Because the final volume after addition of amylase reagent is 1.06 mL instead of the original 4.1 mL, the inhibitor concentration used by Huang and Tietz is only 46.4% ofthe original concentration during the reaction with substrate.
A LinearStandardCurvefor
presented
A comparison of the standard curve by Huang and Tietz (2 ) and the one presented by O’Donnell at al.
with
the srsuc was 510 mgfL higher. In our hands, results by these two techniques generally agree within 50 mg/L. For a second specimen, obtained just before the patient’s death, the glucose result from the SMAC was 200 mgfL higher. One of the drugs the patient was receiving was metronidazole (Flagyl, Searle Pharmaceuticals). We added metronidazole to a serum pool to give a
tion ofpancreatic
same
Chem. Lab. Bryn Mawr Ave. Bryn Mawr, PA 19010
method-
ology is used. The value obtained
use of the
Determining Amylase lsoenzymes
(1 ) is hindered by the fact that different ratios are given on the abscissa, IPancreatic isoamylase (P)/Total amylase
by SelectiveInhibition To the Editor:
A method of selective inhibition
for
(T)1 x 100 in the former and IPancreatic isoamylase (P)/Salivary isoamylase
measurement of amylase isoenzymes originally was published by O’Donnell
(5)] on a logarithmic scale in the latter. Therefore I recalculated the figures
et al. (1) in combination
obtained from the graphical representation of the standard curve in the O’Donnell publication to the same units as used by Huang and Tietz. The results are shown in Figure 1 . Linear regression analysis of the 18 values
with the Pha-
concentration of 450 mgfL. The added drug had no effect on results with the Astra methodology, but the src re-
debas method (Pharmacia Diagnostics AB, Uppsala, Sweden) for amylase. In
suit was higher by 1090 mg/L. Moreover, the 450 mgtL metronidazole addition increased apparent glucose by 590
reported
their study a sigmoid standard curve is for the calculation
of the two
(pancreatic and salivary) amylase isoenzymes. Such a nonlinear standard
gives a perfectly
straight
line with y
CLINICAL CHEMISTRY, Vol. 29, No. 5, 1983
=
995
