Evaluation of a Chemiluminescent Enzyme Immunoassay for the Detection of Prostaglandin E‐Major Urinary Metabolite (PGE‐MUM)

ABSTRACT

Background

We developed a fully automated quantitative immunoassay for the detection of prostaglandin E‐major urinary metabolite (PGE‐MUM). In this study, we evaluated the analytical performance of this assay.

Methods

Sensitivity, within‐run reproducibility, correlation with radioimmunoassay (RIA), cross‐reactivity, dilution linearity, spike recovery performance, analyte stability, and effects of coexisting substances were evaluated. The assay was also used to measure PGE‐MUM in 211 healthy people.

Results

The limit of detection and quantification were 1.0 and 1.3 ng/mL, respectively. When the assay was performed six times in a single run, the coefficient of variation ranged from 1.4% to 2.2%. The coefficient of correlation with a preceding RIA method was 0.970 with a correlation slope of 0.88. There was no cross‐reactivity with PGE‐MUM analogs. Linearity of dilution was confirmed at up to 16‐fold dilution with assay results within 100 ± 20% of the theoretical values calculated based on the undiluted sample. Spike recovery was good and ranged from 94% to 101%. Analyte stability was tested by storing samples at 25°C for 6 days, 10°C for 1 month, and by performing up to five freeze–thaw cycles. Assay results were all within 100 ± 10%, the values measured before storage and before the freeze–thaw process. Assay results in healthy people ranged from 3.1 to 162.7 ng/mL (mean: 35.8 ng/mL). After correction for creatinine, the 95% confidence interval was 8.68–42.25 μg/g creatinine.

Conclusion

The assay precisely detects PGE‐MUM.

We developed a chemiluminescent enzyme immunoassay (CLEIA), which can detect prostaglandin E‐major urinary metabolite (PGE‐MUM) with automated procedure in ~30 min. Here, we evaluated its analytical performance. This assay can precisely detect PGE‐MUM with good reproducibility and showed a good correlation with preceding radioimmunoassay (RIA).

1. Introduction

Prostaglandin E2 (PGE2) is a key mediator involved in the promotion and control of inflammation. At sites of inflammation, phospholipase A2 catalyzes the hydrolysis of membrane phospholipids to generate arachidonic acid, which is converted to PGE2 as catalyzed by cyclooxygenase and PGE2 synthase. Although this establishes blood PGE2 as a potential biomarker of inflammation level in the body, PGE2 generated at sites of inflammation and released into the blood undergoes metabolic degradation rapidly with a half‐life in the tens of seconds. Accordingly, blood PGE2 levels are difficult to measure in routine testing.

PGE2 undergoes an enzymic reaction in the lungs catalyzed by 15‐hydroxyprostaglandin dehydrogenase and oxidation in the liver and kidneys and is mainly excreted as a prostaglandin E‐major urinary metabolite (PGE‐MUM). The main component of PGE‐MUM is 9,15‐dioxo−11α hydroxy‐13,14‐dihydro‐2,3,4,5‐tetranor‐prostan‐1,20‐dioic acid (tetranor PGEM). The clinical utility of tetranor PGEM measurement has been demonstrated using liquid chromatography–tandem mass spectrometry (LC–MS/MS) to determine tetranor PGEM levels [1, 2]. However, LC–MS/MS analysis is time consuming and involves complex procedures, which is important since, while being more stable than blood PGE2, tetranor PGEM is unstable during long‐term storage, even at −20°C [3]. PGE‐MUM can be converted to a very stable bicyclic form (4‐(1‐(2‐carboxyethyl)‐2,5‐dioxooctahydro‐1H‐inden‐4‐yl)butanoic acid; bicyclic PGE‐MUM) by alkali treatment, a radioimmunoassay (RIA) has been developed that measures bicyclic PGE‐MUM using anti‐bicyclic PGE‐MUM polyclonal antibodies [4, 5]. This RIA method has been used to show the clinical utility of PGE‐MUM as a marker of ulcerative colitis (UC) activity based on endoscopic and pathological findings [6, 7, 8, 9, 10, 11]. Treatment of UC is typically based on reducing inflammation via drug therapy. Crucial aspects of the choice of the therapeutic method are the severity and extent of involvement and the induction and maintenance of remission. For this reason, UC requires regular, accurate, and sensitive monitoring, particularly during remission and while it is active. PGE‐MUM levels can be monitored in urine samples, which offers the advantages of uniform samples, ease of quantification, minimal burden on the patient, and the ability to perform frequent measurements.

In this paper, we report the development of a chemiluminescent enzyme immunoassay (CLEIA) with sufficient performance for clinical application. RIA methods require a facility equipped to manage radiological materials and manual operations, posing challenges in terms of convenience and speed. However, our developed assay method addresses these issues by performing measurements fully automatically on a device. Especially, our updated approach automates specimen alkali preprocessing on the device, reducing the time from the 30 min manual process to just 6.5 min. This enhancement contributes to a more efficient measurement process and a reduction in overall analysis time. Furthermore, we have adopted monoclonal antibodies over the previously used polyclonal antibodies, which is preferable from an animal welfare perspective due to the use of cultured cells for antibody production, and is expected to enhance product lot‐to‐lot reproducibility due to reduced variability compared to polyclonal antibodies.

2. Methods

2.1. Assay Samples

Urine specimens were (1) collected from adult patients with UC who visited The Jikei University School of Medicine and provided informed consent; (2) from people who visited The Center for Preventive Medicine, The Jikei University Hospital, for medical checkup and provided informed consent; (3) from healthy volunteers without subjective symptoms who provided informed consent; and (4) purchased from NOVA Biologics (Oceanside, CA, USA). Urine samples with high PGE‐MUM levels or urine samples spiked with tetranor PGEM (Cayman Chemical, Ann Arbor, MI, USA) were used as reference specimens L, M, and H.

2.2. Reagents and Analytical Instruments

PGE‐MUM was measured by the CLEIA method using a LUMIPULSE PrestoII (Fujirebio, Tokyo, Japan) or a LUMIPULSE L2400 (Fujirebio). In the analytical instrument, 15 μL of sample was dispensed into a sample pretreatment solution containing sodium hydroxide to convert the PGE‐MUM in the sample to bicyclic PGE‐MUM. Ferrite particles with bicyclic PGE‐MUM chemically bound to their surface and antibodies labeled with alkaline phosphatase were added to the mixture. Bicyclic PGE‐MUM bound to the particles' surface and bicyclic PGE‐MUM in the sample each reacted competitively with the anti‐bicyclic PGE‐MUM antibodies, forming immune complexes, respectively. After washing the particles, substrate solution was added and luminescence was measured. PGE‐MUM concentration was determined based on a calibration curve prepared by measuring the luminescence of standard PGE‐MUM solutions prepared from a bicyclic PGE‐MUM solution standardized based on tetranor PGEM. The assay range was 2.0–200.0 ng/mL.

RIA measurements of PGE‐MUM were performed using the Bicyclic PGE‐MUM RIA kit (Institute of Isotopes Co., Ltd., Budapest, Hungary) at SRL, Inc. (Tokyo, Japan). The RIA method involved adding NaOH solution to 50 μL of sample and reacting for 30 min at room temperature to convert PGE‐MUM to the bicyclic form, then neutralizing the mixture and assaying by the competitive method.

Creatinine was measured by enzymatic method at SRL, Inc.

Coexisting substances were purchased as follows: Interference check A plus containing bilirubin C, bilirubin F, and hemoglobin from Sysmex (Kobe, Japan); glucose, ascorbic acid, acetone, creatinine, ethanol, sodium chloride, and riboflavin from FUJIFILM Wako Pure Chemical Corporation (Osaka, Japan); urea from Nacalai Tesque Inc. (Kyoto, Japan); and human albumin from Sigma‐Aldrich (St. Louis, MO, USA).

3. Results

3.1. Sensitivity

Bicyclic PGE‐MUM was diluted by weight with standard solution diluent to prepare a dilution series of seven low‐concentration samples. Each was assayed 20 times. The limit of detection was calculated as the concentration at which the mean +2 standard deviations (SDs) of the measured count fell below the mean −2 SDs of the measured count at 0 ng/mL, introducing a nuanced adjustment to the methodologies typically applied [12]. The limit of quantification was defined as the concentration at which the coefficient of variability (CV) of the measurement value becomes the predetermined CV [12], and this predetermined CV was set at 15%. The resulting limit of detection was 1.0 ng/mL, and limit of quantification was 1.3 ng/mL; both limits were below the lower limit of the assay range (2.0 ng/mL) (Figure 1).

FIGURE 1.

Sensitivity test results. (A) Limit of detection (mean RLU ± 2SD); RLU, relative light unit. (B) Limit of quantification (CV of assay result).

3.2. Within‐Run Reproducibility

The coefficient of variability for each reference specimen (L, M, and H) was determined by dividing the standard deviation of the assay results by the mean of these results. This calculation was performed for each set of six assays conducted on the reference specimens. The calculated coefficients of variability ranged from 1.4% to 2.2%, demonstrating good within‐run reproducibility, as indicated in Table 1.

TABLE 1.

Within‐run reproducibility test results.

Reference sample	L	M	H
Assay result (ng/mL)	15.3	44.4	165.6
	15.7	44.8	161.1
	15.5	44.3	164.9
	15.1	45.1	162.8
	15.5	47.0	167.5
	15.6	44.9	167.7
Mean	15.5	45.1	164.9
Standard deviation	0.22	0.99	2.61
Coefficient of variability	1.4%	2.2%	1.6%

3.3. Correlation

Correlation with the RIA kit was examined by assaying samples from 112 adult patients with UC (N = 1). The coefficient of correlation of 0.970 was calculated as the Pearson product–moment correlation coefficient. The correlation slope was calculated as 0.88 using the Passing–Bablok regression. At or below 50 ng/mL, the coefficient of correlation was 0.921 and the correlation slope was 0.99 (Figure 2).

FIGURE 2.

Correlation between the CLEIA method and the RIA kit (the solid line represents a linear regression of the data).

3.4. Cross‐Reactivity

Each of the PGE‐MUM analogs tetranor PGFM, bicyclic PGE2, bicyclic PGE1, PGE1, PGE2, PGF2β, 13,14‐dihydro‐15‐keto‐PGE1, and 13,14‐dihydro‐15‐keto‐PGE2 (Cayman Chemical) was added to the urine samples in a 1:19 ratio to a final concentration of 10 μg/mL. Each analog‐added sample (test sample) and a sample with no analog (control sample) were assayed four times. Cross‐reactivity was calculated as:

$$\text{Cross}-\text{reactivity}\left(\%\right)=\left[\left(\text{Mean assayed level in test sample}-\text{Mean assayed level in control sample}\right)/\text{Final analog concentration}\right]\times 100.$$

The resulting cross‐reactivity was low (<0.1% for all analogs) (Table 2).

TABLE 2.

Cross‐reactivity test results.

Test analog	Cross‐reactivity rate
Tetranor PGFM	0.089%
Bicyclic PGE2	0.020%
Bicyclic PGE1	0.012%
PGE1	0.012%
PGE2	Not detected
PGF2β	Not detected
13, 14‐Dihydro‐15‐keto‐PGE1	0.005%
13, 14‐Dihydro‐15‐keto‐PGE2	0.003%

3.5. Dilution Linearity

Six samples spiked with tetranor PGEM were diluted by weight with sample diluent to create a dilution series (2‐, 4‐, 8‐, and 16‐fold) for assaying (N = 1). All dilutions, including the 16‐fold dilution, were within 100 ± 20% of theoretical levels calculated based on the undiluted sample, confirming the dilution linearity of the assay (Figure 3).

FIGURE 3.

Dilution linearity test results.

3.6. Spike Recovery

Tetranor PGEM was added to sample diluent and each of three urine samples from patients with renal impairment in a 1:9 ratio by volume. The antigen‐spiked (test) and non‐spiked (control) samples were assayed (N = 1). Spike recovery was calculated as:

$$\text{Spike recovery}\left(\%\right)=\left[\left(\text{Assayed level in test sample}-\text{Assayed level in control sample}\right)/\left(\text{Final tetranor PGEM concentration}\right)\right]\times 100.$$

Spike recovery was good for all three urine samples with results that ranged from 94% to 101% (Table 3).

TABLE 3.

Spike recovery test results.

Sample	Assay result (ng/mL)			Spike recovery rate (%)
	Spiked antigen	Control sample	Test sample
A	34.1	6.8	39.2	95
B	86.7	6.7	88.6	94
C	140.7	4.0	145.6	101

3.7. Analyte Stability

The stability of analytes was evaluated over a 6‐day period and a 1‐month period at two different temperature settings (Figure 4A,B). The three reference samples were stored at 25°C for 6 days or at 10°C for 1 month. The results indicate that the analytes maintained their integrity within an acceptable range of the initial measured values (100 ± 10%). After 6 days of storage at 25°C, the measured values remained within 92%–103% of the baseline measurements. Similarly, after 1 month of storage at 10°C, the measured values remained within 90%–107% of the original measured values. On the other hand, when the storage period at 25°C was extended to 2 months, the measured values slightly decreased, ranging from 79% to 90% of the initial measurements. These findings suggest that the analytes exhibit robust stability under both room temperature and refrigerated conditions during the period anticipated for clinical use, which is critical for ensuring the reliability of subsequent analyses.

FIGURE 4.

Analyte stability test results. (A) Storage at 25°C, (B) storage at 10°C, and (C) impact of freeze–thaw cycle.

The three reference samples were also put through five freeze–thaw cycles to examine the impact of freezing and thawing (each reference sample was assayed four times). Assay results for up to the fifth freeze–thaw cycle were within 100 ± 10% of the assayed level before the first freeze–thaw cycle in each reference sample, verifying the stability of analytes during freeze–thaw operations (Figure 4C).

3.8. Impact of Coexisting Substances

Bilirubin C, bilirubin F, hemoglobin, glucose, ascorbic acid, acetone, creatinine, ethanol, sodium chloride, riboflavin, urea, and human albumin were each added to the three reference samples in a 1:9 ratio by volume to the final concentrations listed in Tables 4 and 5. Each spiked sample (test sample) and nonspiked samples (control samples) were assayed four times, and the percentage variance of the assayed level in each spiked sample was calculated relative to the nonspiked sample. The percentage assay variance ranged from −8% to 13%, showing that coexisting substances had little impact at the concentrations examined in this test (Tables 4 and 5).

TABLE 4.

Impact of coexisting substances (Bilirubins and Hemoglobin).

Coexisting substance and amount added	Reference sample	Assay result (ng/mL)		Percentage variation (%)
		Control sample	Test sample
Bilirubin C 20.4 mg/dL	L	9.0	8.9	−1
	M	59.0	57.4	−3
	H	148.3	148.8	0
Bilirubin F 19.0 mg/dL	L	8.7	8.8	2
	M	56.3	57.2	2
	H	148.4	148.2	0
Hemoglobin 490.0 mg/dL	L	8.0	9.1	13
	M	52.3	55.6	6
	H	134.6	139.4	4

TABLE 5.

Impact of coexisting substances (other substances).

Coexisting substance and concentration	Reference sample	Assay result (ng/mL)		Percentage variation (%)
		Control sample	Test sample
Glucose 1000 mg/dL	L	8.2	7.9	−5
	M	53.8	51.1	−5
	H	137.1	128.0	−7
Ascorbic acid 500 mg/dL	L	8.0	7.9	−2
	M	53.5	49.1	−8
	H	133.2	121.9	−8
Acetone 100 mg/dL	L	8.2	8.1	−2
	M	53.8	50.7	−6
	H	137.1	133.9	−2
Creatinine 1000 mg/dL	L	8.2	8.4	2
	M	53.8	51.4	−4
	H	137.1	131.3	−4
Ethanol 1000 mg/dL	L	8.2	8.1	−2
	M	53.8	52.0	−3
	H	137.1	133.1	−3
Albumin 1 g/dL	L	8.2	9.2	12
	M	53.8	55.5	3
	H	137.1	140.0	2
Sodium chloride 2 g/dL	L	8.2	9.1	10
	M	53.8	51.6	−4
	H	137.1	132.4	−3
Urea 12 g/dL	L	8.2	8.3	1
	M	53.8	52.5	−2
	H	137.1	135.6	−1
Riboflavin 10 mg/dL	L	8.3	8.0	−4
	M	52.6	51.5	−2
	H	134.2	133.9	0

3.9. Assaying Samples From Healthy People

Samples from 211 healthy people were assayed (N = 1). Assay results ranged from 3.1 to 162.7 ng/mL (mean: 35.8 ng/mL). After correction for creatinine, the 95% confidence interval was 8.68–42.25 μg/g creatinine (Figure 5).

FIGURE 5.

Assay results in healthy people.

4. Discussion

The results from the sensitivity and within‐run reproducibility tests performed as part of the performance evaluation verified that the CLEIA method can accurately measure PGE‐MUM, including in the low‐concentration range. Based on the results from the cross‐reactivity, dilution linearity, and spike recovery tests, the CLEIA method was also capable of accurately measuring analyte levels in individual samples.

Stability testing showed no substantial variation in assay results when samples were stored at 25°C for 6 days or at 10°C for 1 month, confirming stability under these storage conditions.

The clinical utility of PGE‐MUM measurement has so far only been verified with RIA methods; correlation with an RIA method is an important performance requirement for the CLEIA method. The coefficient of correlation between the CLEIA method and the RIA method was high at 0.970. Correlation in the low‐concentration range below 50 ng/mL was also good. Until now, the cutoff value for PGE‐MUM in the diagnosis of UC patients has been set by the RIA method [6, 8, 10, 11]. Although the measurement method is different, the CLEIA method correlates well with the RIA method. Therefore, it is expected that similar cutoff values will be established for the CLEIA method in the future.

The RIA kit used in this study has complex procedures and requires 2 days to produce results. In contrast, the CLEIA method can produce results in approximately 30 min on an automated analytical instrument, with sample pretreatment included in this time. The CLEIA method also uses a smaller sample volume of 15 μL compared to 50 μL used in the RIA kit. The CLEIA method eliminates various issues with RIA kits, such as restrictions on reagent storage and assay use in radiation‐controlled areas. These features of the CLEIA method should allow urine samples to be collected during a hospital visit and medical care to be provided based on PGE‐MUM assay data within the same day.

Reports have shown the usefulness of PGE‐MUM measurements in relation to not only UC but also pediatric necrotizing enterocolitis, interstitial pneumonia, and lung adenocarcinoma [13, 14, 15]. Furthermore, a disease termed chronic nonspecific multiple ulcers of the small intestine (CNSU), the cause of which has long been unknown, has been attributed to mutations in SLCO2A1, a gene related to prostaglandin transporters. PGE‐MUM levels are reported to be significantly higher in patients with CNSU than in patients with Crohn's disease, which is a differential diagnosis of CNSU [16]. We look forward to the benefits of the CLEIA method being utilized in demonstrating its clinical utility in relation to these diseases and in its adoption in clinical settings.

Conflicts of Interest

J. Takagi, K. Moriyama, M. Wakabayashi, H. Nasu, and S. Yagi are employees of FUJIREBIO Inc. N. Katagiri and S. Yagi are employees of Advanced Life Science Institute, Inc. M. Saruta: Scholarship/research grants from EPS Corporation, PPD‐SNBL K.K., EA Pharma Co., Ltd., Kissei Pharmaceutical Co., Ltd., CMIC Holdings Co., LTD., Mochida Pharmaceutical Co., Ltd., Zeria Pharmaceutical Co., Ltd., and Abbvie GK; honoraria for lecture fee from Abbvie GK, Mitsubishi Tanabe Pharma., EA Pharma Co., Ltd., Jassen Pharma K.K., Takeda Pharmaceutical Co., Ltd., and Gilead Sciences K.K. N. Katagiri and S. Yagi, patent application PCT/JP2017/009963.

Permission to Reproduce Material from Other Sources

This is the secondary English version of the original Japanese manuscript for “Basic performance evaluation of assay reagent for PGE‐MUM based on CLEIA” published in the Japanese Journal of Medicine and Pharmaceutical Science 77 (3): 393–401, 2020. The authors have obtained permission for secondary publication from the Editor of the journal.

Data Availability Statement

The data presented in this study are available on request from the corresponding author.