| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Technical Briefs |
1
Institute of Clinical Chemistry,
2
Department of Internal Medicine II, Klinikum Grosshadern, Ludwig-Maximilians-Universität Munich, D-81377 Munich, Germany
a author for correspondence: fax 49-7095-3240, e-mail
Michael.Vogeser{at}klch.med.uni-muenchen.de
The measurement of adrenocorticotropic hormone (ACTH 1-39; Mr 4500) plays a key role in the evaluation of hyper- and hypocortisolism (1)(2)(3), but low ACTH concentrations make accurate measurement challenging. Two automated, nonisotopic ACTH assays have become available in recent years. Because it is usual for rather small series to be assayed for ACTH in a clinical setting and because nonisotopic methods with stable calibration are of great practical interest, we compared these assays.
Both assays investigated were solid-phase, sandwich immunoassays that use chemiluminescence for signal generation and are implemented on benchtop, multichannel, random-access analyzers with ready-to-use reagents.
The Nichols Advantage ACTH assay (Nichols Institute Diagnostics) uses
one acridinium ester-labeled mouse monoclonal antibody that
specifically binds to the C-terminal region of ACTH and one
biotin-labeled goat polyclonal antibody that binds to the N-terminal
region. The sample is incubated for 21 min at 37 °C with both
antibodies simultaneously, leading to the formation of a soluble
sandwich complex in the presence of ACTH molecules. Streptavidin-coated
magnetic particles are then added, and the reaction mixture is
incubated for 10 min, during which the sandwich complex binds to the
particles by biotin-avidin interaction. The particles are fixed to the
wall of the single-use reaction cell magnetically, and unbound, labeled
antibody is separated by aspiration and subsequent washing. Finally,
two trigger solutions are injected into the reaction cells to initiate
the chemiluminescent reaction, and the light emission is quantified
over 2 s by a luminometer. A lot-specific master calibration curve
is loaded by barcode and adjusted weekly with two calibrators by the
user. The reported measuring range extends to 1500 ng/L.
Barcode-labeled sample tubes are processed directly. The sample
throughput is 
90/h, and the time to first result is 37 min with a
start-up time of 15 min.
In the DPC IMMULITE ACTH assay (Diagnostic Products Corporation), a
polystyrene bead within a test unit serves as the solid phase; the bead
is coated with a monoclonal murine anti-ACTH antibody. The patient
sample and a buffer are introduced and incubated for 30 min at 37 °C
with intermittent agitation of the test unit. Unbound sample is then
removed by a centrifugal wash procedure. Subsequently, an alkaline
phosphatase-labeled polyclonal rabbit anti-ACTH antibody is introduced
for a second 30-min cycle, after which unbound enzyme conjugate is
removed by another centrifugal wash. Finally, the chemiluminescent
substrate, a phosphate ester of adamantyl dioxetane, is added. This
compound undergoes hydrolysis in the presence of alkaline phosphatase
to yield an unstable intermediate, producing a sustained emission of
light that is photomultiplied and quantified for 12 s. Adjustment
of a predefined, lot-specific master calibration curve is performed
with two calibrators at recommended intervals of 2 weeks. The reported
measuring range is 10–1250 ng/L. The system reads sample barcodes
after an aliquot of the sample has been pipetted into an assay cup and
the test units have been placed with the cups on a load chain. The
sample throughput is 90 determinations per hour; the time to first
result is 70 min with a start-up time of 10 min.
Both assays were handled according to the manufacturers’ instructions in all respects by experienced technicians in a routine laboratory setting. Quality control was performed for all analytical runs using lyophilized control materials provided by the respective manufacturers (two concentrations for the DPC assay and three for the Nichols assay) according to the German Medical Association guidelines. The study protocol was approved by the institutional review board.
Linearity of the assays was investigated by serial dilution of a high-ACTH patient sample (>1250 ng/L in the DPC assay and 1110 ng/L in the Nichols assay) with a low-ACTH pool in nine steps up to a dilution of 1:512. In both assays, results were linear (r2 >0.99).
To study the interassay imprecision of the assays, four plasma pools
were prepared from residual patient samples, aliquoted after
equilibration, and then stored at -70 °C. ACTH was measured in 12
analytical runs using two different lots of reagents and five
calibration cycles in both assays over a 3-month period. After thawing
at 4 °C, the samples were centrifuged and split for analysis by both
assays simultaneously; each sample was analyzed once in each assay. The
results of the imprecision study are given in Table 1
.
|
For the method comparison study, 444 clinical EDTA-plasma samples were used. In 42 of the samples, ACTH values determined with the DPC and Nichols assays were <10 ng/L; in 17 samples, ACTH was >10 ng/L with the DPC assay and <10 ng/L with the Nichols assay; in 37 samples, ACTH was <10 ng/L with the DPC assay and >10 ng/L with the Nichols assay. Results for the remaining 348 samples were compared by unbiased regression analysis according to Deming: Nichols = 0.81(DPC) + 13.3 ng/L; 95% confidence interval for slope, 0.79–0.82; 95% confidence interval for intercept, 11.9–14.6 ng/L (Pearson r = 0.982).
Mean concentrations of all analyzed samples were 80.3 and 83.1 ng/L with the Nichols and DPC assays, respectively (medians, 33.9 and 30.4 ng/L, respectively). For all paired results with a mean concentration between 10 and 50 ng/L by both assays (n = 223), regression analysis yielded: Nichols = 1.10(DPC) + 0.1 ng/L; 95% confidence interval for slope, 1.00–1.20; 95% confidence interval for intercept, -2.3 to 2.5 ng/L (r = 0.71).
Further comparisons of all results >10 ng/L using the method described
by Bland and Altman (4) demonstrated a mean difference of
-7% (DPC minus Nichols) with a SD of 29% (Fig. 1
).
|
Of the samples used for the method comparison, 356 were obtained from 89 patients during corticotropin-releasing hormone (CRH) stimulation testing. For this test, 100 µg of recombinant CRH (CRF Ferring®; Ferring) was given intravenously, and samples were drawn at -15, 0, 30, and 60 min. Individual absolute increments of ACTH concentrations during CRH stimulation tests were compared between the two assays by Deming regression analysis [baseline concentration subtracted from stimulated concentration at 30 and 60 min, respectively (In30 and In60)]: In30 Nichols = 0.99 x In30 DPC + 1.5 ng/L; 95% confidence interval for slope, 0.94–1.04; 95% confidence interval for intercept, -0.8 to 3.7 ng/L; r = 0.970; and In60 Nichols = 1.04 x In60 DPC - 0.5 ng/L; 95% confidence interval for slope, 0.97–1.10; 95% confidence interval for intercept, -2.3 to 1.3 ng/L; r = 0.954.
Apparently discordant results between the two assays were found in two patients.
Patient 1 was a 19-year-old woman with septo-optical dysplasia, diabetes insipidus, and amaurosis. Her admission was for clinical signs of hypocortisolism. In the CRH stimulation test, her ACTH results at -15, 0, 30, and 60 min were 40.3, 35.6, 96.8, and 111 ng/L, respectively, for the Nichols assay, and <10, <10, 59.5, and 64.0 ng/L for the DPC assay. Serum cortisol concentrations (DPC IMMULITE Cortisol assay) at the same times were 52, 44, 63, and 72 nmol/L.
Patient 2 was a 51-year-old woman with adrenomyeloneuropathy. Her admission was for exclusion of adrenocortical insufficiency, which may occur in adrenomyeloneuropathy. Her unstimulated basal ACTH was 15.0 ng/L for the Nichols assay and 140 ng/L for the DPC assay; her serum cortisol concentrations were 522 and 834 nmol/L before and after stimulation with exogenous ACTH.
The explanation for these discrepant results is not known. All determinations were repeated from the original tubes to exclude benchwork and clerical errors. Heterophilic antibodies may cause such discrepancies.
To assess the manufacturers’ stated reference intervals, basal morning ACTH was determined in 20 healthy volunteers. The central 95% interval was 13–57 ng/L (mean, 27 ng/L; median, 21 ng/L) for the DPC assay (95% reference interval given by DPC <10–46 ng/L; median, 24 ng/L; n = 59) and 15–46 ng/L (mean, 29 ng/L; median, 28 ng/L) for the Nichols assay (95% reference interval given by Nichols, 9–52 ng/L; mean, 22 ng/L; n = 99).
The possible influence of typical interferents on ACTH determination was studied by dilution and recovery experiments. Pooled plasma from uremic patients (creatinine, 681 µmol/L; urea, 36.7 mmol/L) was diluted (1 volume + 1 volume and 1 volume + 3 volumes) with a low-ACTH pool (creatinine, 62 µmol/L; urea, 5.3 mmol/L). ACTH as measured in the two pools and in the diluted samples in duplicate was subjected to linear regression analysis. In the same way, the possible influence of hyperbilirubinemia was studied (bilirubin, 457 and 6.8 µmol/L in the high and low pools, respectively). ACTH results exhibited linearity in both experiments (r >0.95), making relevant interference unlikely. To study the influence of lipemia, we added a lipid-rich solution for parenteral nutrition (Lipofundin N 20%; B. Braun) to a plasma pool with a triglyceride concentration of 20 g/L. ACTH concentrations were determined in duplicate in the native and treated samples; after correction for volume dilution (70 µL of solution plus 1000 µL of sample), 94% of the original concentration was found by the DPC assay and 82% by the Nichols assay. Addition of hemolyzed erythrocyte suspension (10 µL plus 1000 µL of sample; final hemoglobin concentration, 1.88 g/L) led to a lower mean ACTH concentration with the DPC assay (83%) and a higher concentration with the Nichols assay (143%) compared with the native sample.
No carryover effect could be demonstrated in either assay using a sample with a low ACTH concentration (<10 ng/L by both assays) analyzed in quadruplicate, in each case pipetted immediately after a sample with a mean ACTH of 1326 ng/L.
In routine use, the mechanical components of the IMMULITE analyzer (with only one rotating pipetting arm) were rugged. The Nichols Advantage analyzer allowed convenient workflow, with host-data download and no manual pipetting, but several mechanical malfunctions occurred during the study.
Reproducibility of these automated, short-incubation tests was slightly inferior to that reported for an IRMA with a CV of 9.3% at a concentration of 8.7 ng/L (n = 20) (5). This probably reflects the distinctly shorter incubation times of the automated assays (at a temperature of 37 °C vs ambient temperatures in manual assays, which may allow analyte degradation). Also using the Nichols assay reporting results <10 ng/L precise determination of subnormal ACTH concentrations is possible only in a narrow range below the lower limit of the reference range for unstimulated morning ACTH.
Comparability of the ACTH results obtained by the two automated assays under investigation was found to be limited for individual samples within the reference range for basal ACTH. ACTH increments during CRH simulation testing, however, showed close correlation, yielding identical diagnostic information in most cases and similar reference ranges. Nevertheless, discrepancies of results must be expected in a relevant number of tests when these different ACTH assays are used, possibly because of heterophilic antibodies. We speculate that the limited overall comparability may be attributable to different cross-reactivities of the assays with related molecules with longer half-lives, e.g., proopiomelanocortin. In the DPC ACTH assay, the analyte-antibody sandwich is formed in two separate incubations with a wash step, compared with a one-step incubation with both antibodies in the Nichols assay. The latter approach may produce less specific binding and might explain the constant and proportional bias between the two assays, with lower concentrations found with the DPC assay.
We conclude that the use of automated ACTH assays represents progress in the clinical laboratory setting. Either of the assays investigated here can allow cost-efficient analysis near the site of patient care without the need for sample shipment on dry ice to larger laboratories. This will allow straightforward evaluation without the delays in obtaining laboratory reports that might increase the overall cost of treatment.
References
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
