Whole-Blood Counting Immunoassay as a Short-Turnaround Test for Detection of Hepatitis B Surface Antigen, Anti-Hepatitis C Virus Antibodies, and Anti-Treponema pallidum Antibodies

Abstract

Whole-blood samples were used for a counting immunoassay (CIA) with the aim of developing a short- turnaround test. After optimization of the CIA, hepatitis B surface antigen (HBsAg), anti-hepatitis C virus antibodies (anti-HCV), and anti-Treponema pallidum antibodies (anti-TP) were detected as efficiently as by an enzyme immunoassay (EIA) with serum samples. The correlations between whole-blood CIA and serum EIA were 99.8, 97.1, and 99.4% for HBsAg, anti-HCV, and anti-TP, respectively. Whole-blood CIA may be of value when rapid screening of many samples is required.

The counting immunoassay (CIA) is an application of particle counting technology to serological tests (6). Latex particles are agglutinated by antibodies or antigens of interest and are quantified by scattered laser light while passing through a controlled sheath flow, which is also used in flow cytometry. A similar method is the particle counting immunoassay, in which nonagglutinated particles are counted with the aid of instrumentation (1, 4, 5). Reported applications of these methods, most of which use serum samples, include the detection of hepatitis B virus (HBV) antigens or antibodies (1), anti-adult T-cell leukemia antibodies (6), antitoxoplasma antibodies (2), urinary cotinine (3), hormones (4), and serum acute-phase proteins.

The advantages of CIA in comparison with traditional enzymatic methods include a reaction time as short as 15 min, high throughput, and small sample volumes. Taking these advantages into account, we evaluated whole-blood assays by using CIA to develop a short-turnaround test. In general, the preferred sample for a short-turnaround test is whole blood because the preparation of serum samples, including centrifugation time, inevitably takes 30 to 40 min after the blood has been drawn. Since currently available CIA reagents are designed for serum samples, we decided to optimize CIA reagents for whole blood.

Whole-blood samples were tested for hepatitis B, hepatitis C, and syphilis, with the reagents, detector, and internal software all optimized. The results were compared with those obtained by an enzyme immunoassay (EIA) with paired serum samples.

MATERIALS AND METHODS

Principles of detection by CIA.

The principles of the CIA have been described elsewhere in detail (6). In brief, latex particles 0.75 μm in diameter (standard deviation, 0.015 μm) are coated with antigens or antibodies that interact with the substance of interest. Next, the particles are mixed with a whole-blood or serum sample in an automated detector. Fifteen minutes later, when the particles are agglutinated by an antigen-antibody reaction, they are sent to a flow cell mechanism for passage in a single line. Their sizes and frequencies are measured by scattered laser light while they pass through the flow cell, and the numbers of agglutinated multimeric and nonagglutinated monomeric particles are counted (Fig. 1). When a whole-blood sample is used, the number of agglutinated multimeric particles is automatically compensated for by the number of red blood cells (RBC) to obtain an equivalent value from a plasma sample of the same volume as the whole-blood sample.

FIG. 1.

Detection of an antigen-antibody complex (multimeric particles) by whole-blood CIA. Monomers and multimers of latex particles are shown on the screen of the detector. The sizes of the particles are determined by the intensity of the scattered laser light. RBC and platelets are also counted, and the number of RBC is used to compensate for multimer counts.

Detector and reagents for CIA.

A Pamia-40i (Sysmex Corp., Kobe, Japan) detector was used for evaluation. Ranream HBsAg, Ranream HCV II ex, and Ranream TP (Sysmex) were the reagent kits used for HBV surface antigen (HBsAg), anti-hepatitis C virus (HCV) antibodies (anti-HCV), and anti-Treponema pallidum antibodies (anti-TP), respectively. Ranream HBsAg uses gout polyclonal antisera after immunization against human HBsAgs of multiple subtypes, Ranream HCV II ex contains recombinant antigens and synthetic peptides, and Ranream TP uses native antigens of the pathogenic Nichols strain. Samples of 10 μl were used for CIA of either whole blood, serum, or plasma.

EIA.

Serum or plasma samples were evaluated for HBsAg by using AxSYM HBsAg (version 2) (Abbott Japan Corp., Tokyo, Japan), samples were evaluated for anti-HCV by using AxSYM HCV2.0 (Abbott Japan), and samples were evaluated for anti-TP by using the Lumipulse forte TPN2 system (Fujirebio Inc., Tokyo, Japan). All of the assays were performed and interpreted according to the manufacturers' instructions, except for the evaluation of anti-TP results obtained with the Lumipulse system. Since the T. pallidum particle agglutination test (Fujirebio) is among the most popular anti-TP tests, the Lumipulse system and the T. pallidum particle agglutination test were compared and showed significant discordance at our facility (unpublished data). In order to avoid a false-negative report, we introduced an indeterminate range around the cutoff value of 1.0, which was advised by the manufacturer. Our modified cutoff values for the EIA were as follows: 0 to 0.5, negative; 0.6 to 1.9, indeterminate; 2.0 or above, positive.

Ethical considerations.

All of the patients in this study were enrolled in accordance with the guidelines of the ethics committee of the Kyoto University Graduate School of Medicine and the Japanese Society of Laboratory Medicine. All of the samples and the results were numbered, and personal identification information was removed. Written informed consent was obtained from the patients, who were provided with written explanatory information approved by the ethics committee.

Development process.

The prototype detector and the reagents for CIA were designed for serum samples. In the first step of optimization, paired whole-blood and serum samples were compared by using CIA alone to determine the influence of whole-blood components. The anti-HCV test showed a false positivity rate of 1.6% (6 out of 381 negative samples), while the other two tests did not show false-positive results. There were no false-negative results. The software was modified to silence the background noise. A detailed comparison of whole-blood and plasma samples revealed that there appeared to be trailing noise around the peaks of dimeric and other multimeric latex particles in whole-blood samples. Since the sizes of the noise particles were different from those of monomers, dimers, or other multimers, each count of noise particles was deducted from the measured particle counts. This noise reduction program worked well; the false positivity rate for the anti-HCV test decreased to 0.4% (2 out of 509 negative samples) without an increase in false-negative results for all three tests.

In the second step, samples with an abnormal hemogram were used to evaluate the influence of unusual whole-blood components. When samples containing fragmented RBC were applied, 20 out of 42 negative samples had false-positive results in the anti-HCV test, while there were none in the other two tests. These false-positive results were ascribed to an interaction between the latex particles for anti-HCV detection and the RBC fragments, and so the buffer ingredients were empirically modified by the addition of a detergent. The optimized reagent worked well, and none of 21 negative samples with RBC fragments had false-positive results with the whole-blood-optimized anti-HCV reagent. Reevaluation of samples with a normal hemogram showed no change in sensitivity or specificity. The cutoff values for whole-blood CIA were set at the mean plus approximately 2.0 standard deviations of the results for normal samples. Finally, the whole-blood CIA was evaluated by the serum EIA as described below.

Specimens for final evaluation.

A total of 1,545 pairs of whole-blood and serum samples were tested both by whole-blood CIA and by tests commonly used for HBV, HCV, and T. pallidum, and the results were compared. Separately, 459 selected whole-blood samples with abnormal hemograms were compared with plasma samples from the same patients as a challenge test. Artificial samples with borderline positive results were prepared as follows. Twofold dilutions of HBsAg-, anti-HCV-, and anti-TP-positive controls were made with normal donor plasma, the blood types of which were identical to those of positive samples. Washed RBC from the donors, containing trace amounts of phosphate-buffered saline, were added to these diluted plasma samples to make the hematocrit 50%. Aliquots of these artificial whole-blood samples were used for whole-blood CIA, and the remaining aliquots were used to obtain plasma samples by centrifugation for plasma CIA and plasma EIA.

RESULTS AND DISCUSSION

To determine the reproducibility of the three CIAs, a low-titer sample was examined 10 times with each assay. The coefficients of variation for each whole-blood assay were 7.5% for HBsAg, 5.4% for anti-HCV, and 3.1% for anti-TP.

The final comparison of whole-blood CIA and serum EIA after optimization demonstrated that the CIA results were similar to those obtained with the EIA (Tables 1 and 2). Discordance between whole-blood CIA and serum EIA was evaluated with the same serum samples as those used for CIA (serum CIA). For HBsAg, all of the serum CIA and whole-blood CIA results for 518 samples were in agreement (100%). For anti-HCV, all of the results except for one were concordant for 522 samples (99.9%), and for anti-TP, all of the results except for two showed agreement for 505 samples (99.6%). Thus, the discordance between whole-blood CIA and serum EIA was ascribed to the difference between CIA and EIA, not to the whole-blood sample used.

TABLE 1.

Comparison of whole-blood CIA and serum EIA for detection of HBsAg, anti-HCV, and anti-TP

Whole-blood CIA resulta	No. of samples with the indicated serum EIA resultb for:
	HBsAg (n = 518)		Anti-HCV (n = 522)		Anti-TP (n = 505)
	Positive	Negative	Positive	Negative	Positive	Indeterminate	Negative
Positive	18	0	81	9	14	1	1
Indeterminate					1	2	0
Negative	1	499	6	426	0	0	486

^aThe cutoff values for CIA were as follows: for HBsAg, 0.5 U/ml; for anti-HCV, 1.0 of cutoff index (COI); and for anti-TP, <10 Sysmex units (SU)/ml (negative results), 10 to 30 SU/ml (indeterminate results), and >30 SU/ml (positive results).

^bThe cutoff values for EIA were as follows: for HBsAg, 2.0 signal-to-noise (S/N) ratio; for anti-HCV, 1.0 signal-to-cutoff ratio; and for anti-TP, <0.5 S/N (negative results), 0.6 to 1.9 S/N (indeterminate results), and <2.0 S/N (positive results).

TABLE 2.

Evaluation of whole-blood CIA and serum EIA

Antigen or antibody (n)	Correlation (%)	Sensitivity (%)	Specificity (%)	Predictive value (%)
				Positive	Negative
HBsAg (518)	99.8	95.0	100	100	99.8
Anti-HCV (522)	97.1	93.1	98.0	90.0	98.6
Anti-TP (505)a	99.4	93.3	99.9	93.3	100

^aIndeterminate results were excluded from calculations.

To elucidate the difference in the sensitivities of whole-blood CIA and serum EIA further, artificial whole-blood samples containing serial dilutions of positive plasma were prepared and evaluated. For HBsAg, CIA had a lower sensitivity than EIA, while CIA had a higher sensitivity for anti-HCV and equivalent results for anti-TP (Table 3).

TABLE 3.

Limiting dilution analysis for comparison of sensitivities of whole-blood CIA, plasma CIA, and plasma EIA with artificial samples

Antigen or antibody	Series	Highest dilution of plasma that tested positive (indeterminate) in:
		Whole-blood CIA	Plasma CIA	Plasma EIA
HBsAg	1	4	4	128
	2	8	8	128
	3	4	4	64
Anti-HCV	4	4	4	1
	5	4	4	1
	6	16	16	2
Anti-TP	7	1 (4)	1 (4)	2 (4)
	8	2 (8)	1 (4)	2 (4)
	9	2 (8)	2 (4)	2 (4)

In order to confirm the reliability of whole-blood CIA under unusual conditions, 459 selected whole-blood samples with abnormal hemograms were examined. Abnormal hemograms included microcytosis (mean corpuscular volume of less than 80 fl) with more than 4 × 10⁶ RBC/μl (n = 217), macrocytosis (mean corpuscular volume of more than 100 fl) with more than 4 × 10⁶ RBC/μl (n = 94), polycythemia (hematocrit of more than 55% [n = 8] or more than 5.5 × 10⁶ RBC/μl [n = 77]), and thrombocytosis (more than 60 × 10⁶ platelets/μl [n = 65]). More than 94% of the samples in each group were found to be HBsAg, anti-HCV, and anti-TP negative by EIA. None of the abnormal samples yielded more than one false-positive result, and there was no false-negative result.

In the present study, the whole-blood assay results seemed to be in agreement with those of EIA, which is commonly used in clinical laboratories. When serial dilutions of controls were examined, the three CIAs had different sensitivities. There could be two reasons for this result: (i) a difference in the principles for detection or (ii) a difference in the antigenicities (or the specificities of antibodies for HBsAg) for the target materials in the assays. Both of these differences may have been responsible in each assay.

A theoretical difficulty associated with the use of whole blood is false-positive results from noise produced by blood cells. However, the effectiveness of optimization was proved by the evaluation of abnormal whole-blood samples, which showed no increase in false-negative results.

Another possible disadvantage of whole-blood CIA is the small sample volume in comparison with that used for serum CIA or EIA. Since blood cells have a significant volume, the use of whole blood causes a decrease in the volume of the essential plasma used, a situation which may reduce the sensitivity of the assay. However, this effect was not observed in this series.

The costs per assay were similar for the CIA and EIA currently available in Japan. For example, anti-HCV detection costs $8.18 with CIA and $9.03 with EIA (Abbott Japan).

As expected with the use of whole-blood CIA, the time required to obtain results after drawing blood was reduced by 30 to 50 min, depending on the reference serum EIA. A Pamia-40i detector can examine 70 samples per hour at most. Since high throughput is the main advantage of this method, its application may be extended to include markers required in emergency rooms, quarantine, or some pandemic situations.

In summary, the whole-blood CIA results showed good concordance with the results of serum EIA. Even when challenge with abnormal whole-blood samples was used, the differences observed between the whole-blood and serum assays were minimal. The short time needed for a whole-blood assay may be helpful for clinical procedures.