Aspirin hypersensitivity diagnostic index (AHDI): In vitro test for diagnosing of N‐ERD based on urinary 15‐oxo‐ETE and LTE₄ excretion

Abstract

Background

15‐oxo‐eicosatetraenoic acid (15‐oxo‐ETE), is a product of arachidonic acid (AA) metabolism in the 15‐lipoxygenase‐1 (15‐LOX‐1) pathway. 15‐oxo‐ETE was overproduced in the nasal polyps of patients with nonsteroidal anti‐inflammatory drug–exacerbated respiratory disease (N‐ERD). In this study we investigated the systemic biosynthesis of 15‐oxo‐ETE and leukotriene E₄ (LTE₄) and assessed their diagnostic value to identify patients with N‐ERD.

Methods

The study included 64 patients with N‐ERD, 59 asthmatics who tolerated aspirin well (ATA), and 51 healthy controls. A thorough clinical characteristics of asthmatics included computed tomography of paranasal sinuses. Plasma and urinary 15‐oxo‐ETE levels, and urinary LTE₄ excretion were measured using high‐performance liquid chromatography and tandem mass spectrometry. Repeatability and precision of the measurements were tested.

Results

Plasma 15‐oxo‐ETE levels were the highest in N‐ERD (p < .001). A receiver operator characteristic (ROC) revealed that 15‐oxo‐ETE had certain sensitivity (64.06% in plasma, or 88.24% in urine) for N‐ERD discrimination, while the specificity was rather limited. Modeling of variables allowed to construct the Aspirin Hypersensitivity Diagnostic Index (AHDI) based on urinary LTE₄‐to‐15‐oxo‐ETE excretion corrected for sex and the Lund‐Mackay score of chronic rhinosinusitis. AHDI outperformed single measurements in discrimination of N‐ERD among asthmatics with an area under ROC curve of 0.889, sensitivity of 81.97%, specificity of 87.23%, and accuracy of 86.87%.

Conclusions

We confirmed 15‐oxo‐ETE as a second to cysteinyl leukotrienes biomarker of N‐ERD. An index based on these eicosanoids corrected for sex and Lund‐Mackay score has a similar diagnostic value as gold standard oral aspirin challenge in the studied group of patients with asthma.

Plasma 15‐oxo‐ETE is higher in N‐ERD than in aspirin‐tolerant asthmatics or healthy subjects. Urinary LTE₄ positively correlates with the Lund‐Mackay score in N‐ERD. Urinary 15‐oxo‐ETE is higher in women than in men. An index (AHDI) based on urinary LTE₄ to 15‐oxo‐ETE ratio corrected for sex and Lund‐Mackay score predicts N‐ERD with sensitivity 82% and a specificity of 87.2%.

Abbreviations: AHDI, Aspirin Hypersensitivity Diagnostic Index; ATA, asthmatic patients who tolerated aspirin well; L‐M_score, Lund‐Mackay score; LTE₄, leukotriene E₄; N‐ERD, nonsteroidal anti‐inflammatory drug‐exacerbated respiratory disease; 15‐oxo‐ETE, 15‐oxo‐eicosatetraenoic acid.

Abbreviations

15‐HETE

15‐hydroxyeicosatetraenoic acid

15‐LOX‐1

15‐lipoxygenase‐1

15‐oxo‐ETE

15‐oxo‐eicosatetraenoic acid

AA

arachidonic acid

AHDI

Aspirin Hypersensitivity Diagnostic Index

ALOX‐15‐1

arachidonate 15‐lipoxygenase‐1 coding gene

CRSwNP

chronic rhinosinusitis with nasal polyps

HPGD

15‐hydroxyprostaglandin dehydrogenase

LTE₄

leukotriene E₄

NSAIDs

nonsteroidal anti‐inflammatory drugs

1. INTRODUCTION

Nonsteroidal anti‐inflammatory drug (NSAID)–exacerbated respiratory disease (N‐ERD) is a particular asthma phenotype distinguished by acute respiratory reactions following ingestion of nonselective cyclooxygenases (COXs) inhibitors such as aspirin or other NSAIDs. ¹ , ² , ³ Inflammatory cellular pattern within the airways of N‐ERD asthmatics is variable. In common with all asthma can be eosinophilic or noneosinophilic. ⁴ , ⁵ , ⁶ , ⁷ N‐ERD is usually difficult to treat and accompanied by a chronic rhinosinusitis with recurrent nasal polyps (CRSwNP). An unique metabolic feature of N‐ERD is activation of arachidonic acid (AA) pathways producing pro‐inflammatory leukotrienes and prostaglandins ⁴ , ⁸ , ⁹ , ¹⁰ , ¹¹ and decreased anti‐inflammatory prostaglandin E₂ (PGE₂) or lipoxins. ¹² , ¹³ 15‐oxo‐eicosatetraenoic acid (15‐oxo‐ETE), a product of 15‐lipoxygenase (15‐LO‐1) expressed in the airways epithelium, was recently suggested as a mediator of CRSwNP in N‐ERD. ¹⁴ Alterations of other AA‐derived mediators are detectable in blood, ¹⁵ urine, ¹⁶ nasal lavage fluid, ¹⁷ or induced sputum supernatant of N‐ERD asthmatics, ⁴ , ⁶ , ⁸ but a systemic production of 15‐oxo‐ETE was not investigated. Only a few tests have been proposed to diagnose N‐ERD in vitro ⁶ , ¹⁷ , ¹⁸ , ¹⁹ , ²⁰ , ²¹ , ²² , ²³ but performed poorly. Of note, 15‐hydroxyeicosatetraenoic acid (15‐HETE) produced in vitro by peripheral blood leukocytes was also proposed. ²⁴ The most reliable diagnostic method for N‐ERD remains an oral aspirin provocation test. ³ , ²⁵ , ²⁶ Expression of arachidonate 15‐lipoxygenase‐1 coding gene (ALOX‐15‐1) encoding 15‐lipoxygenase‐1 (15‐LOX‐1) was significantly elevated in the nasal polyp tissue of patients with N‐ERD. Within the polyps 15‐LOX‐1 was predominantly expressed by epithelial cells and correlated with a lower forced expiratory volume in the first second (FEV₁) or radiographic severity of sinus disease in N‐ERD. ¹⁴ , ²⁷ Hydroxyprostaglandin dehydrogenase (HPGD) enzyme, converting 15‐HETE to 15‐oxo‐ETE, has a high expression in macrophages and mast cells. Therefore, a transcellular metabolism in the polyp tissues by neighbor cells using LOX‐15‐1 epithelial activity was suggested as a route of 15‐oxo‐ETE synthesis. ¹⁴

We assessed a diagnostic value of 15‐oxo‐ETE measured in plasma or urine to diagnose N‐ERD. Results were compared to urinary excretion of leukotriene E₄ (LTE₄). The technical and time‐dependent biological reproducibility of these biomarkers was studied. Combined urinary 15‐oxo‐ETE and LTE₄ levels were used to construct an index for N‐ERD diagnosis. Urinary 15‐oxo‐ETE combined with LTE₄ outperformed any other proposed in vitro tests for N‐ERD.

2. METHODS

2.1. Patients

In a prospective study 64 N‐ERD asthmatics, 59 aspirin‐tolerant asthmatics (ATA), and 51 healthy subjects (HC) were enrolled. All patients were diagnosed and treated at the 2nd Department of Internal Medicine, Jagiellonian University Medical College, Kraków, Poland, according to the 2022 Global Initiative for Asthma update. ²⁸ Diagnosis of N‐ERD was confirmed by oral aspirin provocation test as described previously. ²⁵ Patients with N‐ERD and ATA received nasal and inhaled corticosteroids as well as long‐acting β₂‐agonists, except three patients with N‐ERD treated with small (2–8 mg methylprednisone per day) oral corticosteroids, however, all asthmatics were in stable condition. None of patients received antileukotrienes or biologic drugs during the preceding year. The controls (ATA or HC) did not use aspirin or other NSAIDs during 6 preceding weeks. On the day of blood and urine collection, each patient had a baseline FEV₁ of 70% or higher. The characteristics of the study subjects is presented in Table 1. The study was approved by the Jagiellonian University Ethics Committee. All the subjects of the study provided written informed consent to participate. The study protocol complied with the Declaration of Helsinki.

TABLE 1.

Characteristics of the study groups.

Variable	N‐ERD (n = 64)	ATA (n = 59)	HC (n = 51)	p‐Value
Age, years, median (25%–75%)	51.0 (36.0–56.0)	52.0 (40.0–62.0)	48.0 (31.0–54.0)	.184
Male sex, n (%)	15 (23.4)	17 (28.8)	9 (17.6)	.388
BMI, kg/m2, median (25%–75%)	25.9 (23.9–30.6)	25.7 (23.5–29.1)	24.7 (22.4–28.0)	.113
Age of asthma onset, years, median (25%–75%)	33.0 (25.5–44.0)	36.0 (24.0–45.0)	—	.524
Asthma duration, years, median (25%–75%)	12.5 (7.5–20.5)	16.0 (6.0–24.0)	—	.709
Mini‐AQLQ score, median (25%–75%)	5.27 (4.12–6.24)	5.73 (4.35–6.47)	—	.114
ACQ‐7 score, median (25%–75%)	1.0 (0.43–2.14)	0.71 (0.14–1.86)	—	.215
ACT score, median (25%–75%)	22.0 (17.0–24.0)	21.0 (16.0–25.0)	—	.616
Asthma severity based on GINA 2022, n (%)
Mild	7 (11.0)	8 (13.6)	—	.459
Moderate	26 (40.6)	29 (49.1)
Severe	31 (48.4)	22 (37.3)
ICS, n (%)	57 (89.1)	57 (96.6)	—	.108
ICS dose (fluticasone propionate or equivalent), μg, median (25%–75%)	500 (400–1000)	500 (250–1000)	—	.093
Baseline FEV1, % predicted, median (25%–75%)	94 (82–102)	97 (90–104)	—	.095
CRSwNP, n (%)	64 (100)	32 (58.2)	—	<.001
History of sinonasal surgery (yes/no), n (%)	55/9 (85.9)	17/42 (28.8)	—	<.001
Number of sinonasal surgeries, median (25%–75%)	2 (1–3)	0 (0–1)	—	<.001
SNOT‐22 score, median (25%–75%)	48 (33–63)	31 (16–53)	—	.002
Total Lund‐Mackay score ascertained in all asthmatic patients, median (25%–75%)	15 (11–20)	2 (0–12)	—	<.001
Nasal corticosteroids (yes/no), n (%)	50/14 (78.1)	20/39 (33.9)	—	<.001
Immunoglobulin E, IU/mL, median (25%–75%)	96.8 (54.5–162.0)	58.8 (29.3–226.0)	18.4 (17.7–34.3)	<.001a,c
Positive skin prick test, n (%)	26 (41.9)	40 (71.4)	—	<.001
Peripheral blood eosinophil count, cells/μL, median (25%–75%)	340 (215–575)	190 (110–370)	120 (80–150)	<.001a,b .002c
Blood eosinophil count, %, median (25%–75%)	4.9 (3.4–8.7)	2.9 (1.6–5.0)	1.9 (1.2–2.6)	<.001a,b .008c
Urinary LTE4, pg/mg creatinine, median (25%–75%)	425 (225–791)	139 (107–203)	109 (71–154)	<.001a,b .004c
Urinary 15‐oxo‐ETE, pg/mg creatinine, median (25%–75%)	8.1 (5.2–15.9)	11.7 (5.6–21.2)	9.2 (5.9–19.8)	.125
Plasma 15‐oxo‐ETE, pg/mL, median (25%–75%)	13.4 (8.8–18.2)	8.4 (5.8–11.8)	6.4 (3.9–11.6)	<.001a,b

Note: p‐value for the analysis of variance or the Freeman–Halton extension of the Fisher exact test, as appropriate. Post hoc tests: ^a p‐value for N‐ERD versus HC; ^b p‐value for N‐ERD versus ATA; ^c p‐value for ATA versus HC.

Abbreviations: ACT, Asthma Control Test; ACQ‐7, 7‐item Asthma Control Questionnaire; BMI, body mass index; GINA, Global Initiative for Asthma; ICS, inhaled corticosteroids; mini‐AQLQ, Mini Asthma Quality of Life Questionnaire; N‐ERD, SNOT‐22, 22item SinoNasal Outcome Test.

2.2. Study design

Demographic, clinical, and biochemical data, computed tomography (CT) scans of nasal sinuses in all asthmatic patients, CRSwNP severity evaluation, skin prick test, and spirometry were assessed in patients with asthma. Plasma and urinary 15‐oxo‐ETE levels as well as urinary LTE₄ levels were measured in all participants. The metabolism of AA with reference to 15‐oxo‐ETE biosynthesis is presented in Supplementary Material (Figure E1). The detailed design of the study is presented in Figure 1.

FIGURE 1.

An overview of the trial design. In patients with asthma, all the procedures were performed. The procedures carried out in the healthy control group are underlined. ACQ‐7, 7‐item Asthma Control Questionnaire; ACT, Asthma Control Test; CT, computed tomography; FEV1, forced expiratory volume in the first second; GINA, Global Initiative for Asthma; ICS, inhaled corticosteroids; IgE, total serum immunoglobulin E; mini‐AQLQ, Mini Asthma Quality of Life Questionnaire; SNOT‐22, 22‐item Sino‐Nasal Outcome Test.

2.3. Clinical and radiological evaluation

Details on the Asthma Control Test, 7‐item Asthma Control Questionnaire, Mini Asthma Quality of Life Questionnaire, severity of asthma according to GINA guidelines, 22‐item Sino‐Nasal Outcome Test (SNOT‐22), spirometry, CT, and the Lund‐Mackay score are presented in Supplementary Material.

2.4. Measurements

Measurement of AA metabolites were accomplished using a high‐performance liquid chromatography–tandem mass spectrometry (HPLC‐MS/MS), as described previously in our study ²⁹ and by other investigators. ¹⁴ , ³⁰ All samples were measured in triplicates and results were averaged. Details of analytical methods are included in Supplementary Material.

2.5. Technical repeatability

Technical repeatability was assessed using triplicate measurements (intraassay). To assess the intermediate precision, each sample was duplicated and both aliquots were measured (interassay, Figure E2, Supplementary Material). ³¹ , ³²

2.6. Time‐related variations in 15‐oxo‐ETE and urinary LTE₄ levels

To assess individual variation of assessed plasma and urinary biomarkers over time, the test was repeated in 31 participants: 14 patients with N‐ERD, 8 patients with ATA, and 9 HCs. Blood and urine samples were collected approximately 10 months after initial sampling. Characteristics of study subjects participating in the follow‐up and individual measurements of eicosanoids were presented in the Supplementary Material.

2.7. Urine collection and reporting of biomarkers content

Morning urine samples were collected after a 2‐h interval from a previous micturition. Urinary 15‐oxo‐ETE and LTE₄ excretion as assessed by HPLC‐MS/MS were recalculated in picograms per mg of creatinine.

2.8. Aspirin Hypersensitivity Diagnostic Index (AHDI)

The Aspirin Hypersensitivity Diagnostic Index (AHDI) for diagnosing aspirin hypersensitivity was constructed using a machine learning algorithm (Linear Discriminant Analysis) to provide the best distinction between N‐ERD and ATA on the premises of: (1) measurements of urinary biomarkers, (2) correction for CRSwNP using the Lund‐Mackay score, (3) assumption of a sex‐related differences in eicosanoid metabolism. The final formula of AHDI = $\frac{{\text{uLTE}}_4}{\mathrm{u}15-\mathrm{oxo}-\mathrm{ETE}\mathrm{x}\beta }\mathrm{x}L-M_{\text{score}}$ (for women, β = 1; for men, β = 2) was applied to results of all asthmatic participants. For all study participants the ratio of urinary LTE₄ to urinary 15‐oxo‐ETE was approximately twice as much in males than in females, which was compensated by β factor. In asthmatics without paranasal sinuses opacities Lund‐Mackay score was assumed L‐M = 1. Details of are presented in Supplementary Material.

2.9. Statistical analysis

Statistical analysis was performed with Statistica 13.3 (TIBCO Software Inc.). A type I statistical error p ≤ .05 was considered statistically significant. Details of the analysis are presented in Supplementary Material.

3. RESULTS

3.1. Characteristics of the study groups

Demographic variables (age, sex, body mass index), clinical, biochemical, and radiological characteristics of N‐ERD patients and control groups are presented in Table 1, all the study participants were Caucasians. There were no significant differences in age, sex, and body mass index between patients with N‐ERD versus ATA or HC. Peripheral blood eosinophilia differed between the groups (p < .001), the highest count was in N‐ERD group (340 cells/μl). Positive skin prick tests were more common in ATA than in N‐ERD or HC groups (p = .001). N‐ERD asthmatics had a higher incidence of CRSwNP and a higher number of sinonasal surgeries (both p < .001) than ATA. Moreover, patients with N‐ERD reported higher SNOT‐22 (p = .002) and had higher Lund‐Mackay scores than ATA (p < .001). The urinary LTE₄ levels were also higher in the N‐ERD group than in ATA (p < .001).

3.2. Technical validation of eicosanoid measurements by HPLC/MS/MS

The intraclass correlation coefficient was 0.988 (95% confidence interval [CI], 0.985–0.991) for plasma 15‐oxo‐ETE levels, 0.997 (95% CI, 0.996–0.998) for urinary 15‐oxo‐ETE levels, and 0.998 (95% CI, 0.998–0.999) for urinary LTE₄ levels. Therefore, technical repeatability was acceptable. ³¹ , ³² As for intermediate precision, at 95% probability relative error of the 15‐oxo‐ETE measurement was less than 12.6% in plasma and 11.9% in urine. Error of LTE₄ measurement at 95% probability was less than 12.4%. The intermediate precision of measurements is presented in Supplementary Material along with the standard error of means for replicated samples (Figure 2).

FIGURE 2.

The Bland–Altman plots of the intermediate precision of 15‐oxo‐ETE measurement in plasma (A), 15‐oxo‐ETE measurement in urine (B), and LTE₄ measurement in urine (C) in the first and second series. The vertical axis shows the extent of the differences; the horizontal axis presents the mean values; red lines indicate a standard deviation. CI, confidence interval; SD, standard deviation.

3.3. Plasma and urinary 15‐oxo‐ETE levels

Plasma 15‐oxo‐ETE levels differed significantly between N‐ERD, ATA, and HC groups. The highest median levels were noted in patients with N‐ERD, followed by ATA patients and HCs (13.4, 8.4, and 6.4 pg/mL, respectively; p < .001). No differences were noted for urinary 15‐oxo‐ETE levels in the N‐ERD group, as compared with ATA patients or HCs (8.1 vs. 11.7 vs. 9.2 pg/mg creatinine; p = .125) (Figure 3).

FIGURE 3.

Between the study groups comparison of 15‐oxo‐ETE levels in plasma (A) and in urine (B). Comparison of urinary LTE₄ levels between the study groups (C). AHDI calculated on urinary LTE₄ and urinary 15‐oxo‐ETE to distinguish N‐ERD from ATA (D). ***p‐value < .001.

3.4. Urinary LTE₄ levels

Patients with N‐ERD had much higher median urinary LTE₄ levels than patients with ATA and HCs (425, 138, and 109 pg/mg creatinine; p < .001).

3.5. Correlations between 15‐oxo‐ETE and clinical variables

In the N‐ERD and ATA groups, no correlations were noted between plasma or urinary 15‐oxo‐ETE levels and other variables assessed in this study listed in Table 1.

3.6. Correlations between urinary LTE₄ levels and clinical variables

There was a positive correlation between the Lund‐Mackay score and urinary LTE₄ levels in patients with N‐ERD (r = .407, p = .033). Moreover, there was a positive correlation between blood eosinophil count and urinary LTE₄ levels in these patients (r = .473, p = .004). No other correlations between urinary LTE₄ levels and variables assessed in this study were noted.

3.7. Time‐related variations in the levels of assessed parameters

There were time‐related variations in plasma and urinary 15‐oxo‐ETE levels as well as urinary LTE₄ levels (Supplementary Material, Figure E3). Despite these differences, both the first and the second measurement assigned patients to the same groups: either ATA or N‐ERD using AHDI (Figure 4).

FIGURE 4.

Receiver operating characteristic curve demonstrating the sensitivity and specificity of AHDI to predict aspirin hypersensitivity in (A) first measurement, (B) second measurement. AHDI, Aspirin Hypersensitivity Diagnostic Index.

3.8. Urinary 15‐oxo‐ETE and LTE₄ in patients with N‐ERD and severe asthma

Patients with severe asthma according to GINA updated 2022 report, that is severe asthma if it remained uncontrolled despite adherence to the maximal optimized high‐dose ICS‐LABA treatment and management of contributory factors, or if it worsened when the ICS dose was reduced, had aspirin hypersensitivity (n = 31) There was no straightforward association between these urinary eicosanoids in N‐ERD asthmatics. Details are presented in Figure  5 and in a scatterplot (Supplementary Materials, Figure E4).

FIGURE 5.

Validation of the association between urinary 15‐oxo‐ETE levels and urinary LTE₄ levels in 31 N‐ERD patients with severe asthma according to GINA updated 2022 report. The red tones mean stronger local positive association and the green tones mean local negative association between urinary 15‐oxo‐ETE levels and urinary LTE₄ levels. 15‐oxo‐ETE, 15‐oxo‐eicosatetraenoic acid; LTE₄, leukotriene E₄.

3.9. Plasma 15‐oxo‐ETE as a diagnostic marker for N‐ERD

Plasma 15‐oxo‐ETE levels were higher in the N‐ERD than in the ATA and HC groups (p < .001) (Table 1). Based on these findings, a diagnostic value of the plasma 15‐oxo‐ETE level alone to diagnose N‐ERD was estimated. For the cutoff point of 12.87 pg/mL, the area under the curve (AUC) was 0.733 (95% CI, 0.645–0.822), sensitivity was 64.06% (95% CI, 51.1%‐ 75.68%), specificity was 72.88% (95% CI, 59.73%–83.64%), and accuracy was 72.26% (95%CI, 63.47%–79.95%) (Supplementary Materials, Figure E5).

3.10. Urinary 15‐oxo‐ETE or LTE₄ alone as a diagnostic marker for N‐ERD

Urinary 15‐oxo‐ETE levels tended to be lower in the N‐ERD group versus ATA and HC groups, but the difference was not significant (Table 1). For the cutoff point of 20.72 pg/mg creatinine, the AUC was 0.602 (95% CI, 0.5–0.704), sensitivity was 88.24% (95% CI, 78.13%–94.78%), specificity was 25.9% (95% CI, 15.26%–39.04%), and accuracy was 30.23% (95% CI, 22.37%–39.04%). For of urinary LTE₄, the cutoff point for predicting aspirin hypersensitivity was 281.53 pg/mg creatinine, with an AUC of 0.834 (95% CI, 0.759–0.909), sensitivity of 66.67% (95% CI, 53.66%–78.05%), specificity of 98.1% (95% CI, 83.27%–98.09%), and accuracy of 91.25% (95% CI, 84.73%–95.62%) (Supplementary Materials, Figure E5).

3.11. Aspirin Hypersensitivity Diagnostic Index (AHDI)

In patients with asthma, median urinary 15‐oxo‐ETE levels were higher in women than in men (11.65 vs. 5.31 pg/mg creatinine; p = .003). Based on these differences, the AHDI was corrected for sex (β = 2 for men and β = 1 for women). There was a positive correlation between urinary LTE₄ levels and the Lund‐Mackay score in the N‐ERD group (r = .407, p = .033). At the cutoff point of uLTE₄ = 160.38, AHDI had AUC of 0.889 (95% CI, 0.882–0.956), a sensitivity of 81.97% (95% CI, 70.02%–90.64%), specificity of 87.23% (95% CI, 74.26%–95.17%), and an accuracy of 86.87% (95% CI, 79.01%‐92.%) to predict N‐ERD. The positive predictive value was 32.58% and the negative predictive value was 98.47%. Receiver operating characteristic presenting sensitivity and specificity of AHDI to N‐ERD is shown in Figure 5. Using a second sampling of urine, AHDI also assigned patients to the same groups: either ATA or N‐ERD.

When only asthmatic patients with CRSwNP were selected (N‐ERD n = 64, ATA = 33), AHDI index still discriminated properly between N‐ERD and ATA. In details, plasma 15‐oxo‐ETE was significantly increased in N‐ERD compared with ATA with CRSwNP (median 13.35 pg/mL vs. 9.04 pg/mL; p < .01). Moreover, N‐ERD patients had significantly higher urinary LTE₄ and Lund‐Mackay scores when compared to ATA with CRSwNP (uLTE₄: median 424.85 pg/mg creatinine vs. 176.97, p < .01; L‐M score: 15 vs. 10, p = .002). Consequently, AHDI was statistically higher in N‐ERD as compared to ATA with CRSwNP (median 667.05 vs. 68.06; p < .01). This was despite lack of difference in urinary 15‐oxo‐ETE between the N‐ERD and ATA with CRSwNP (p = .084). Area under ROC curve was 0.901 (95% CI 0.835–0.971), p < .0001 for discrimination between N‐ERD and ATA with CRSwNP.

3.12. Comparison of AHDI performance versus standard oral or bronchial provocation tests according to EAACI/GA2LEN guideline 2007

At the significance level of p = .05, there were no differences in sensitivity and specificity between AHDI and oral or bronchial provocation test according to EAACI/ GA2LEN guideline, 2007. Reported sensitivity for oral and bronchial challenge was 89% and 77%, whereas specificity was 93% or 93%. Sensitivity of the AHDI test (81.97%) was significantly higher than that of the bronchial provocation test (77%). However, as compared with the sensitivity of oral provocation test, lower AHDI sensitivity was not different from oral drug challenge (probability of γ = .51). Specificity of AHDI (87.23%) was lower than that of oral or bronchial provocation test with a probability of γ = .09. Details of these comparisons are provided in Supplementary Material.

4. DISCUSSION

We assessed a systemic production of AA chemotactic metabolite 15‐oxo‐ETE in the plasma and urine. The parent molecule of 15‐HETE is produced in high amounts at respiratory epithelium by 15‐LOX‐1. The systemic biosynthesis of 15‐oxo‐ETE was determined using HPLC‐MS/MS. ¹⁴ , ³⁰ Measurement of plasma and urinary 15‐oxo‐ETE levels, as well as urinary LTE₄, were reproducible. Despite some biological variation of individual results within a time, these urinary measurements provided an efficient index to diagnose N‐ERD.

In patients with N‐ERD, plasma or urinary 15‐oxo‐ETE levels were not related to any demographic, clinical, biochemical, and radiological variables assessed in the study, such as asthma severity, blood eosinophilia, or the Lund‐Mackay score. Urinary LTE₄ levels were positively correlated with the Lund‐Mackay score in the N‐ERD group. In patients with asthma, urinary 15‐oxo‐ETE levels were dependent on sex, irrespective of aspirin hypersensitivity. Plasma or urinary 15‐oxo‐ETE levels were not sufficient alone for an in vitro diagnostic of N‐ERD. Therefore, we proposed a combined urinary LTE₄‐to‐15‐oxo‐ETE ratio with Lund‐Mackay score, and sex to construct the diagnostic index. Previously, expression of ALOX‐15‐1 was reported correlated with the radiographic severity of sinus disease in patients with N‐ERD. ¹⁴ , ²⁷ This was a rationale for inclusion of Lund‐Mackay score correction in the proposed AHDI formula.

Our study provided evidence to support the use of AHDI in a relatively large group of patients with N‐ERD. Notably, no difference was found between the performance of AHDI and oral or bronchial provocation test in the diagnosis of N‐ERD. ²⁵ Therefore, we suggest that a single collection of urine sample to examine urinary 15‐oxo‐ETE and LTE₄ levels, together with the CT scan of the paranasal sinuses, can distinguish N‐ERD in a patient with asthma. The oral provocation tests with aspirin should be used for research purposes, aspirin desensitization or in some difficult diagnostic cases, whereas, AHDI seems feasible in a daily clinical practice.

To our knowledge, this is the first study to show significantly higher plasma 15‐oxo‐ETE levels in patients with N‐ERD compared with controls. Surprisingly, N‐ERD patients tended to have lower urinary 15‐oxo‐ETE levels compared with ATA. But in another our study, N‐ERD patients who responded to aspirin therapy had a high baseline genetic expression of the HPGD in sputum cells, an enzyme required 15‐oxo‐ETE synthesis. ³³

Based on the recent findings that 15‐oxo‐ETE contributed to tissue inflammation, it may serve as a useful biomarker of the underlying inflammatory process. ¹⁴ , ²⁷ Many cells involved in the asthmatic inflammation, including epithelial, eosinophils, monocytes, macrophages, and dendritic ones, express 15‐LOX‐1. ³⁴ , ³⁵ , ³⁶ These cells contribute to the pathogenesis of N‐ERD. ⁷ , ³⁵ , ³⁷ , ³⁸ , ³⁹ Importantly, when comparing the transcriptional profile of the nasal polyps in patients with N‐ERD and CRSwNP using RNA sequencing, ALOX‐15‐1 expression was higher in the nasal polyp tissue of patients with N‐ERD. ¹⁴ Moreover, interleukins IL‐4 and IL‐13 induce expression of 15‐LOX‐1 in human airway epithelial cells. ⁴⁰ Increased 15‐LOX‐1 levels induced by IL‐13 in nasal polyp epithelial cells contribute to eotaxin3/CCL26 production through the activation of the extracellular signal‐regulated pathway. ⁴¹ This chemokine selectively binds to a specific receptor (CCR3) highly expressed on eosinophils, mast cells, and basophils. ³⁴ , ⁴¹ , ⁴² , ⁴³ , ⁴⁴

A research on a local levels of 15‐oxo‐ETE in induced sputum, bronchoalveolar lavage fluid or nasal washings ought to be correlated with a cellular patterns of inflammatory cells. N‐ERD characteristics for a cellular inflammatory patterns of induced sputum was already described. It seems necessary to compare the blood and sputum cells for a profile of ALOX‐15‐1, ALOX‐15‐2, and HPGD expression, as well as the encoded enzymes in patients with N‐ERD and ATA. Previously, we investigated the transcriptional profile of sputum cells for AA metabolic pathways before and after 52‐week of aspirin therapy following desensitization. All AA‐related genes except LTC4S and ALOX15 were grouped together, indicating the ability of aspirin to regulate the expression of these genes in N‐ERD. ⁴⁵

Our study has several limitations. First, we notice fluctuations in blood and urinary 15‐oxo‐ETE levels over time. However, study subjects resampled after several weeks interval, although not all, ensured a stability of AHDI performance. 15‐oxo‐ETE synthesis can change over time along with the cellular pattern of the airway inflammation. ⁴⁶ In addition to a free 15‐oxo‐ETE in urine, conjugates of 15‐oxo‐ETE or a kidney break down products ought to be studied. ⁴⁷ , ⁴⁸ Another limitation is lack of activity markers for eosinophils and mast cells, which could identify the cellular source of the investigated eicosanoids. Eventually, a standardized multicenter study on a larger group of patients can only confirm a robustness of AHDI. Currently, many tertiary hospitals are equipped with HPLC‐MS/MS instruments or 15‐oxo‐ETE immunochemistry ELISA assay can be developed like for LTE₄ to apply AHDI in asthma centers.

In conclusion, a ratio of urinary LTE₄ to 15‐oxo‐ETE, corrected for sex and severity of CRSwNP, can be used as an in vitro test of N‐ERD. Plasma levels of 15‐oxo‐ETE can differentiate asthmatics with N‐ERD form ATA in a similar way as urinary LTE₄ levels. Apparently, AA metabolism via the 15‐LOX pathway plays a role in the pathomechanism of N‐ERD. Targeting of AA metabolic pathways remains a promising therapeutic strategy in patients with N‐ERD. Future research should explore the pathways mediated by 15‐LOX‐1 and 15‐LOX‐2.

AUTHOR CONTRIBUTIONS

GT, RK, and JS collected the clinical data, LM designed and overviewed the study and prepared initial draft of the manuscript. PS and AG processed the samples and did HPLC/MS/MS measurements, AĆ did statistical interpretation of results, HP and BJ participated in sample collection, MS wrote the final version of the manuscript and prepared its revised revision. All authors provided individual input through a discussion and preparation of the manuscript.

FUNDING INFORMATION

This work was supported by the National Science Centre, Poland (Narodowe Centrum Nauki), grant number UMO‐2021/41/B/NZ5/00305.

CONFLICT OF INTEREST STATEMENT

None declared.

Supporting information

Data S1.

Mastalerz L, Trąd G, Szatkowski P, et al. Aspirin hypersensitivity diagnostic index (AHDI): In vitro test for diagnosing of N‐ERD based on urinary 15‐oxo‐ETE and LTE₄ excretion. Allergy. 2025;80:534‐544. doi: 10.1111/all.16281

DATA AVAILABILITY STATEMENT

The data that support the findings of this study are available from the corresponding author upon reasonable request.