| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Articles |
1
University of Brescia, Brescia, Italy.
2
University of Kansas Medical Center, Kansas City, KS,
USA.
3
University of Wales College of Medicine, Cardiff, UK.
a Author for correspondence: Division of Hematology, NIBSC, Blanche Lane, South Mimms, Potters Bar, Herts EN6 3QG, UK. Fax 44-(0)1707-646730; e-mail sthorpe{at}nibsc.ac.uk
 |
Abstract |
|---|
 Top Abstract IntroductionMaterials and MethodsResultsDiscussionAppendix 1References |
|---|
|
Introduction |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionAppendix 1References |
|---|
Because of the difficulties in obtaining suitable human tissue for ferritin purification and to eliminate potential differences between separate preparations of purified human ferritin, a recombinant ferritin preparation of L subunits has been assessed as a potential replacement for the IS. Our preliminary investigations indicated the recombinant ferritin was immunologically similar to the IS, and pilot studies were carried out to determine the optimum conditions for lyophilization to ensure minimum destruction on lyophilization and prolonged storage. A recombinant L ferritin preparation was subsequently ampouled as a candidate IS. The results of a large international collaborative study to evaluate the material are presented in this report.
The first aim of the study was to compare the recombinant ferritin with the 2nd IS for ferritin (spleen) in a wide range of immunoassays and assign a ferritin content to the ampouled recombinant material. The second was to estimate the stability of the recombinant ferritin on storage at -20 °C.
|
Materials and Methods |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionAppendix 1References |
|---|
Distribution into ampoules.
The recombinant ferritin was
diluted to ~5.6 µg/mL in human plasma (pooled individual
cryoprecipitate-deleted donations that had been tested and found
negative for HBsAg, anti-HIV, and anti-HCV, kindly provided by North
East Thames Regional Transfusion Centre) and dispensed into ampoules
(~1 mL/ampoule). The recombinant ferritin solution was kept at
4 °C throughout the procedure. The ampoule contents were lyophilized
and sealed under dry nitrogen with heat fusion of the ampoule glass.
Secondary desiccation to remove residual moisture was not carried out
because pilot studies indicated that this procedure resulted in loss of
immunoreactivity. The ampoules were stored in the dark at -20 °C
except for a small number that were stored at -70 °C, 4 °C,
20 °C, 37 °C, 45 °C, and 56 °C for accelerated degradation
studies. The material was coded 94/572. The mean weight of the
dispensed solution in 60 ampoules was 1.0082 g. The imprecision of the
filling (CV) was 0.3%, and the residual moisture content was 1.2%.
2nd is for ferritin, spleen
The 2nd IS for ferritin (80/578), prepared from human spleen
ferritin, was established in 1992 (6). Details of its
purification and characterization have been described (5).
The ferritin concentration of the reconstituted ampoule contents (with
1 mL of H2O) is 9.1 µg/mL.
participants
The 18 laboratories that participated in the study are listed in
the Appendix in alphabetical order by country. Each is
referred to in this report by an arbitrarily assigned number (1–18),
not necessarily in order of listing. When a laboratory performed more
than one method, each method is treated as if performed by separate
laboratories. For example, laboratory 14 carried out three different
methods, which are referred to as 14A, 14B, and 14C.
assays contributed to the study
The assay methods used by the participating laboratories are shown
in the Appendix. A total of 20 different assay systems were
used and included different commercial kits and automated analyzers
from several manufacturers, as well as in-house assays. Only one assay
system was duplicated (laboratories 2 and 14B). Laboratory 14 used
three methods, laboratory 15 used two methods, and the remaining 16
laboratories used one method.
study design
Participants were instructed to reconstitute ampoule contents with
1 mL of distilled water. They were requested to assay a series of
dilutions of the recombinant ferritin preparation 94/572 together with
dilutions of the 2nd IS for ferritin 80/578 such that similar ranges of
response resulted for at least four dilutions of each falling in the
linear portion of the dose–response curve. Participants were asked to
assay replicate dilutions (i.e., two independent sets from undiluted
samples) in duplicate. Three independent assays were requested, on
separate days, with dilutions of freshly reconstituted ampoules of each
of the recombinant ferritin preparation 94/572 and the 2nd IS for
ferritin 80/578 per assay.
Participants were asked to supply raw data (i.e., dilutions tested and the actual responses) in a standard format for each assay for analysis at the National Institute for Biological Standards and Control (NIBSC) as described below. Participants were also requested to give their own calculated estimate of the ferritin concentration of the reconstituted recombinant ferritin preparation 94/572 relative to the 2nd IS 80/578 for each assay and their laboratory mean potency for 94/572.
statistical analysis
Raw data submitted to NIBSC were analyzed to give the potency of
94/572 relative to 80/578 for each assay, and a laboratory mean potency
for 94/572 (NIBSC calculations). These values and the participating
laboratories' own potency estimates (laboratories' own calculations)
were analyzed separately to give an overall mean potency for 94/572 and
intra- and interlaboratory variability.
The results of both sets of analysis are reported. Details are as follows: dose–response curves were constructed from the raw data supplied to NIBSC, from which the potency of the recombinant ferritin preparation relative to the 2nd IS was determined by parallel line bioassay methodology (7).
Because parallel linear response/transformed response lines are essential for this method of analysis, it was necessary to determine the appropriate treatment of raw data generated by each laboratory to meet these requirements. Where the majority of responses in an assay fell in the linear portion of the log dose–response or log dose–log response curve, standard parallel line analysis was performed. Responses from doses outside the linear portion were omitted. For the assays in which the responses fell over the full sigmoid dose–response curve, a logistic transformation was used. This was done by the WRANL program, which transforms data to percentages of the estimated upper and lower limits of curves for each assay (8). The statistical validity of parallelism and linearity of the assays was assessed by analysis of variance tests.
The potency of the recombinant ferritin preparation 94/572 relative to the 2nd IS for ferritin 80/578 was calculated in µg/mL (corresponding to µg/ampoule) for each assay. For each laboratory, a combined potency estimate was obtained by taking the geometric mean of results from all their assays. The overall potency estimate of 94/572 relative to 80/578 was calculated as a geometric mean of the laboratories' means.
Variability between assays within each laboratory and between laboratories was measured by calculating geometric coefficients of variation (GCV x%) (9).
|
Results |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionAppendix 1References |
|---|
Deviations from the study protocol and other anomalies were as follows. (a) The recombinant ferritin was not assayed in replicate by laboratory 8. For the purpose of the analysis of variance tests, the duplicates were treated as replicates. After transformation, the response lines proved to be nonparallel. Therefore, this analysis was restricted to areas where the response range was common to both preparations. The same restriction was applied for laboratory 15A, leaving only two dose amounts for each preparation so that the assumption of linearity could not be tested. (b) Laboratories 12 and 14 reported concentration readings for each dilution tested that had been read from an internal standard curve and not the actual responses. However, for laboratory 12, the readings had not been corrected for dose amounts so they were treated as responses and analyzed as described. This laboratory carried out 6 assays in duplicate, but assays carried out on the same day were treated as the same assay to give replicates. (c) Laboratory 18 returned raw assay data but did not provide their own potency calculations. In addition, no replicates or duplicates had been included so the assumptions of linearity and parallelism could not be tested.
assay validity
In the majority of assay systems (nine laboratories), log
responses gave an approximately linear relationship to log dose. For
two laboratories, untransformed responses were used; in the remaining
seven laboratories, the logistic transformation was used. The
assumptions of linearity and parallelism each held separately in 88%
of the total assays. Both linearity and parallelism held in 77%.
However, from comparison of the slopes across all assays,
nonparallelism was not detected for the study as a whole. Therefore,
all assays were included in subsequent analyses.
intra- and interlaboratory variability
The variability within each laboratory (i.e., between the assays
carried out by each laboratory), expressed by a percentage as a GCV, is
given in Table 1
. With the NIBSC calculations, this ranged from 0.3% to 8.6%
(representing good repeatability) except for laboratories 8 and 10,
which had GCVs of 11.8% and 10.7%, respectively. However, on the
basis of the laboratories' own potency calculations, laboratory 9
appeared to have by far the highest variability with GCV of 20.4%.
|
Interlaboratory variabilities for the potency estimates of the recombinant ferritin preparation relative to the 2nd IS were found on the basis of NIBSC calculations (GCVs of 16.37% when laboratory 6 was included and 11.56% when laboratory 6 was omitted; see below) and the laboratories' own potency calculations (GCVs of 15.19% when laboratory 6 was included and 10.61% when laboratory 6 was omitted; see below).
On the basis of the NIBSC calculations and excluding laboratory 6, the interlaboratory variation was 2.5 times that of the average intralaboratory variability, although equivalent to that of two individual laboratories.
estimates of ferritin content of preparation 94/572
The individual laboratory mean potencies of the recombinant
ferritin preparation relative to the 2nd IS are listed in Table 2
together with 95% confidence limits for the NIBSC calculations
only. The results are also shown in histogram form (for the NIBSC
calculations only) in Fig. 1
.
|
|
There was reasonable agreement on the potency of 94/572 between 17 laboratories, with potency estimates ranging from 5.2 to 7.3 µg/mL (i.e., 5.2–7.3 µg/ampoule). Potency values from laboratories 2 and 14B, which used the same method, were 6.7 and 6.2 µg/ampoule, respectively. The estimates from laboratory 6 were exceptionally higher than those from the other laboratories.
In general, each laboratory's own estimated potency calculation lay within the 95% limits calculated by NIBSC. Exceptions are the estimated potencies reported by laboratories 7, 9, and 13. Laboratory 9 reported estimates from assays 2 and 3 that agreed well with each other but not with their estimate from assay 1, which explains the high variability noted above. However, the NIBSC calculations showed all their assays to be in agreement. Similarly, the internal quality control of laboratory 13 suggested that the third assay was slightly higher than the previous assays. Again, this was not supported by the NIBSC calculations. The cause of the discrepancy for laboratory 7 is not clear.
The overall mean potency values of the recombinant ferritin preparation
94/572 relative to the 2nd IS for ferritin 80/578, on the basis of the
NIBSC calculations and the laboratories' own calculations are shown in
Table 2
. The figures were also calculated excluding laboratory 6 and,
on the basis of the NIBSC calculations, give a mean ferritin content of
6.3 µg/ampoule.
stability
Three of the laboratories participated in the accelerated
degradation study to estimate the long-term stability of the
recombinant ferritin preparation 94/572. Coded ampoules of the
recombinant ferritin preparation 94/572, which had been stored at
various temperatures for 13 months, were compared with ampoules stored
continuously at -70 °C. Each laboratory performed three assays with
separate ampoules, and the results were analyzed as multiple parallel
line bioassays comparing log response to log dose. The assumptions of
linearity and parallelism held separately in 98% and 89% of the
assays, respectively. However, for one of the preparations in assay 1
and two of the preparations in assay 2 of laboratory 2, only two dose
concentrations were reported. Hence, the assumption of linearity could
not be tested for these preparations. To gain parallelism in assay 1 of
laboratory 2, the preparation stored at 37 °C was omitted.
The estimated mean ratios of the concentration of the recombinant
ferritin stored at higher temperatures relative to the concentration of
material stored at -70 °C for each laboratory are given in Table 3
. Each mean is based on 3 estimates except for the mean
estimates at -20 °C, which are each based on 6 estimates because
ampoules stored at this temperature were replicated. All three
laboratories reported difficulty in reconstituting the material after
storage at 45 °C and 56 °C so these data were excluded.
|
The long-term stability of the recombinant ferritin preparation was
predicted with the Arrhenius equation (10). The estimated
ratios were homogeneous between laboratories and were pooled. The
analysis was weighted depending on the variability of assay results.
The predicted percent losses per month and per year at various
temperatures are shown in Table 4
.
|
These results indicate that the recombinant ferritin preparation (94/572) is stable when stored at -20 °C.
|
Discussion |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionAppendix 1References |
|---|
A recombinant L form is, in theory, an ideal standard for assaying serum ferritin, which is ordinarily composed almost entirely of L subunits and has a relatively low iron content (13). Furthermore, polyclonal antibodies raised by injection of ferritin from liver or spleen, which contain ~15% H subunit, or heart ferritin, which may contain up to 60% H subunit (14), show specificity for the L-rich forms of ferritin. Production of specific, polyclonal antibodies to the acidic isoferritins found in heart and erythrocytes requires both the injection of acidic isoferritins (prepared by fractionation of heart ferritin) and absorption of the antiserum with spleen or liver ferritin to remove the antibodies to L subunits (15). That there is little immunological difference between the recombinant L-ferritin, serum ferritin, and the 1st and 2nd IS, which contain ~85% L subunits (5), is therefore not surprising. An additional advantage is that the in vitro production of recombinant ferritin allows for a theoretically unlimited supply of a consistent preparation, whereas lyophilization in cryosupernatant plasma provides a matrix similar in constitution to clinical samples of serum.
The clustering of potency values calculated from the participating laboratories' raw data allowed a consensus value of 6.3 µg/ampoule, relative to the potency of the 2nd IS, to be assigned to the recombinant ferritin preparation. This value was close to the amount of ferritin protein ampouled (~5.6 µg/ampoule), except in laboratory 6, which appeared to have an antibody with unusual specificity. Although, like the 2nd IS, the lyophilized recombinant ferritin preparation was not secondary-desiccated, it showed adequate stability.
On the basis of the results of this collaborative study and with the overall agreement of the participants, the WHO Expert Committee on Biological Standardization established preparation 94/572 as the 3rd IS for ferritin, recombinant, with an assigned ferritin content of 6.3 µg/ampoule (16). Ampoules are available for distribution from the NIBSC.
|
Appendix 1 |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionAppendix 1References |
|---|
ferritin assays used by participating laboratories
1, in-house ELISA; 2, chemiluminometric immunoassay
(automated analyzer no. 1); 3, enzyme immunoassay (automated
analyzer no. 2); 4, enzyme immunoassay (automated analyzer no. 3); 5,
IRMA (kit no. 1); 6, enzyme immunoassay (automated analyzer no. 4); 7,
in-house ELISA; 8, in-house ELISA; 9, in-house ELISA; 10,
chemiluminometric immunoassay (automated analyzer no. 5); 11, in-house
fluorogenic ELISA; 12, RIA (kit no. 2); 13, immunoturbidimetric
analyzer (automated analyzer no. 6); 14A, nephelometric assay (kit no.
3); 14B and 14C, two different chemiluminometric immunoassays
[automated analyzers no. 1 (also used by laboratory 2) and no. 7];
15A and 15B, automated enzyme immunoassay analyzers (automated
analyzers nos. 8 and 9); 16, latex photometric immunoassay (automated
analyzer no. 10); 17, in-house ELISA; 18, IRMA (kit no. 4).
|
Acknowledgments |
|---|
|
Footnotes |
|---|
1 Nonstandard abbreviations: IS, International Standard; NIBSC, National Institute for Biological Standards and Control; GCV, geometric coefficient of variation. 
|
References |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionAppendix 1References |
|---|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
