Eosinophil protein X and childhood asthma: A systematic review and meta‐analysis

Abstract Background: There are no reference guidelines for health care providers regarding appropriate use and interpretation of urine eosinophil protein X (u‐EPX) in clinical practice. Currently, there are no clear‐cut clinical or laboratory parameters to diagnose asthma in young children. Objective: In this study, we (1) systematically reviewed and qualitatively appraised the epidemiological evidence to determine diagnostic u‐EPX cut points for pediatric asthma, and (2) performed a meta‐analysis to provide u‐EPX estimates for diagnosing pediatric asthma. Methods: Research articles in literature were identified from PubMed/Medline and Web of Science databases from 1966 to August 2015. Children <18 years of age were included. Both serum and urine EPX were included. Twenty‐seven studies met the inclusion criteria for the systematic review and nine studies for the meta‐analysis. Details regarding EPX analyses, treatment efficacy, and outcomes were assessed. For meta‐analyses, effect estimates were abstracted using standardized means. Results: Over 70% of studies found a significant relationship between u‐EPX and childhood asthma. There was 1.94 times higher standardized means of u‐EPX among acute asthmatics compared to healthy controls (confidence interval [CI]: 1.67–2.22). Similarly, the difference in standardized means between asymptomatic asthmatics and healthy controls was 1.58 times higher (CI: 1.27–1.88). Conclusions and Clinical Relevance: Despite differences in sample sizes, EPX processing and measurement, and ages of children, a consistent trend of higher EPX levels with childhood asthma was revealed.

EPX may be the most valuable biomarker of eosinophil activation in childhood asthma [11] because it is sensitive, non-invasive, and easily measured in urine. Urinary eosinophil protein X (u-EPX) correlates inversely with nocturnal peak expiratory flow rates (PEFRs) and 1-second forced expiratory volume, reflecting its role as a marker in asthma activity [11]. u-EPX does not undergo significant in vivo degradation and is not influenced by perennial allergy and polysensitization [12] and can differentiate symptomatic from asymptomatic patients.
There is a much higher release of EPX than serum eosinophil cationic protein (s-ECP) in asthmatics, perhaps, indicating that eosinophils selectively release their granule proteins depending upon the type of stimulus to which the cells are exposed [13]. Thus, the determination of EPX in urine gives the same, if not better, information on disease activity and impact of treatment on inflammation as s-ECP.
Currently, normal reference ranges for u-EPX in children with asthma and healthy controls are lacking. Hence, the objective of this meta-analysis is to determine whether u-EPX is a valuable marker in diagnosing childhood asthma. Additionally, preliminary cut points will be proposed for EPX levels in asthmatics and healthy children. Both of these aims will be useful in guiding clinical practice for pediatric asthma.

Methods
Data sources, search strategy, and study selection A systematic literature search of PubMed/Medline and Web of Science core collection databases were performed for studies published until August 2015, using the search terms, childhood asthma, pediatric asthma, serum EPX, urine EPX, Eosinophil protein X, Eosinophil-derived neurotoxin, EDN, and inflammatory markers. The search strategies are reported in detail in the flowchart in Figure 1.
Using the title and abstract, the literature search was reviewed independently in quadruplicate (H. Klonoff-Cohen, J. Lee, O. Refugio, M. Polvarapu) and screened against the study selection criteria to identify potentially relevant studies for full review. After the removal of duplicates, a total of 1191 articles were found from electronic databases, and from reference mining. Searches of biographies were also conducted to ascertain additional studies. Relevant studies were selected, assessed and data were extracted independently by four people (H. Klonoff-Cohen, J. Lee, O. Refugio, and M. Polvarapu). Disagreement was resolved by discussion and mutual agreement among the reviewers. Of the 1191 articles identified, 1128 articles were excluded based on the article's title and abstract (not relevant), and 63 articles were retrieved for full text review. After complete assessment, only 27 articles met the established inclusion criteria. Authors of these 27 articles were contacted at least three times to obtain missing information.

Eligibility criteria
The following inclusion criteria were used for the systematic review: (1) Peer reviewed articles written in English; (2) cohort studies, cross-sectional studies, and case-control studies; (3) asthma as the index disease; (4) study sample: Children up to 17 years of age; (5) urinary and/or serum EPX used as a biological marker; (6) assessment of EPX in asthmatics and healthy control children. Review studies were excluded. Additionally, one randomized clinical trial was excluded because of the design. In this study, asthma patients were randomly assigned deliberate doses of montelukast in order to understand the variation in eosinophilic protein markers with medication. In all other studies, patients received treatments based on physician prescribed medications rather than being randomly allocated to pre-defined treatment groups. Hence, a total of 27 studies that complied with the pre-stated inclusion criteria were identified.
Among the 27 selected studies, for quantitative analysis (meta-analysis), the criteria used for inclusion were (1) presence of control group in the study consisting of healthy non-asthmatic children; (2) study group consisting of acute asthmatics OR asymptomatic asthmatics during the collection of urine sample; (3) reported values of u-EPX among asthmatics and healthy controls with measures of center and spread (mean/median; standard deviation/interquartile range); and (4) for a valid comparison, all the values of u-EPX reported in units mg/mmol Creatinine.

Data extraction
The results from studies that met the inclusion criteria and contained the outcomes of interest were included in the meta-analyses. All data were double entered and triple checked.
Summary outcome measures consisting of urinary EPX levels for childhood asthmatics/non-asthmatics, symptomatic/asymptomatic asthmatics, those on medication/without medications, atopic asthmatics/non-atopic asthmatics, as well as children experiencing acute exacerbations were extracted from studies and tabulated. Following data were extracted from the studies identified for this manuscript: author, publication year, study hypothesis, study design, sample size and characteristics of study and comparison groups, EPX measures, conclusion, and disadvantages of the studies. We also assessed the methodology used to measure u-EPX levels with the data, including test kit, detection limit/sensitivity of the kit, time and rate of centrifuging, freezer temperature, duration before freezing the sample, and any additional unique information provided pertaining to the process.

Assessing the quality of studies
The quality of each of study included in the quantitative analysis (meta-analysis) was assessed using the dichotomous criteria (Table 1) which was designed based on the ''Quality Assessment Tool for observational cohort and crosssectional studies,'' ''QUADAS 2,'' and ''Newcastle-Ottawa scale.''

Statistical analyses
The meta-analysis was performed using Stata software. The effect size representing the difference in u-EPX levels between asthmatics and control group children were calculated using standardized mean differences (SMD), also referred to as Cohen's d. For studies that did not report the measures as means and standard deviations, attempts to contact the authors for the actual data were unsuccessful. For uniform measures of u-EPX to compare, evidence based conversion formulas [14][15][16] were employed to calculate estimated means and standard deviations from the reported values (e.g., median values, quartiles, and ranges).
Treatment protocols affected urinary EPX levels differently in studies. The change in level of u-EPX with medications (anti-inflammatory) for asthma was demonstrated in nine studies [10, 20, 25, 27-30, 32, 37], but a clear pattern could not be established. For instance, in a study by Nuijsink et al. [31], u-EPX levels among asthmatics receiving different doses of fluticasone (200 and 500 mg/day) demonstrated varying u-EPX levels of Median 189 (Range 2-2828) and median 180 (range 10-3114) mg/mmol. This wide range of U-EPX levels makes it difficult to interpret the effect of different fluticasone doses on the u-EPX levels.
A total of nine studies used serum EPX levels as a measure of airway inflammation in asthma [10-12, 17-21, 4]. Rao et al. [11] did not report any numerical results and Kim [17] compared mild, moderate, and severe asthmatics but with no comparison group. Remes's asthmatic group (estimated mean 64.23 mg/mmol) [19] was fairly similar to Zimmerman's atopic asthmatic group (Mean 69 mg/ mmol) [20]. Among the healthy control group, the central measures of serum EPX are comparable in the studies by Remes (estimated mean ¼ 28.5 mg/mmol) [19], Kim (estimated mean ¼ 24.03 mg/mmol) [18], Koller (estimated mean ¼ 25.26 mg/mmol) [12], Kristjansson (estimated mean ¼ 30.8 mg/mmol) [10], and Zimmerman (mean ¼ 31.2 mg/mmol) [20].    Based on the qualitative results, we determined preliminary u-EPX cut-points for asthmatics and healthy controls as 134-140 mg/mmol for acute asthmatics; 65-75 mg/mmol Cr for acute asthmatics after treatment; and 56-61 mg/mmol Cr for healthy controls (based on five studies). However, these data were extremely limited and consisted of medians and quartiles, rather than means and standard deviations. Therefore, estimates of means (and SD) were calculated from reported medians (and quartiles).

Quantitative analysis: diagnosing asthma with EPX
Due to the difficulty of diagnosing pediatric asthma in young children, several studies used terms such as airway inflammation, wheezing, or upper respiratory infections to describe asthmatics, making it difficult to determine if there were consistent trends of u-EPX across these studies. Nine studies [25-30, 33, 35, 40] reporting the levels of u-EPX in uniformly defined asthmatic subgroups (i.e., acute asthmatics, asymptomatic asthmatics) and healthy controls were selected for meta-analysis. Seven [25-29, 33, 35] out of nine studies reported u-EPX levels for acute asthmatics and healthy controls, and five [26,[28][29][30]40] out of nine studies reported u-EPX levels for asymptomatic asthmatics and healthy controls with three studies [26,28,29] reporting u-EPX levels for acute asthmatics, asymptomatic asthmatics, and healthy controls. One study, Severien et al. [37], which met the inclusion criteria for the quantitative analysis was not included in the meta-analysis because of a high mismatch between reported (in the article) and the estimated standard deviations.
The comparison of SMD between acute asthmatics and healthy controls, and asymptomatic (stable) asthma patients and healthy controls were performed and illustrated in Figures 2 and 3. Both forest plots indicate that net u-EPX levels among asthmatics (irrespective of symptoms) are higher than those of non-asthmatics. Comparing acute asthmatics and controls, the forest plot in Figure 2 suggests that, on average, the mean u-EPX level among asthmatics is 1.94 times higher than those in the healthy control group with a confidence interval of 1.67-2.22. Similarly, the difference in standardized means between asymptomatic asthmatics and controls (Fig. 3) showed 1.58 times higher levels among asymptomatics than in healthy controls (CI: 1.27-1.88). varying endpoints, 19 out of 27 studies [10, 12, 17-21, 24-30, 33, 35-37, 40] reported a statistically significant relationships between EPX and childhood asthma.
A meta-analysis was performed that reported u-EPX values for both asthmatic cases and healthy controls. From the metaanalysis of seven studies [25-29, 33, 35], it was determined that the overall SMD between acute asthmatics and nonasthmatic healthy controls was 1.94 (CI: 1.67-2.22). Additionally, the meta-analysis results of five studies [26,[28][29][30]40] revealed an overall SMD of 1.58 (CI: 1.27-1.88) between asymptomatic asthmatic cases and healthy controls. Hence, Urinary EPX is elevated in children with either symptomatic or asymptomatic asthma compared to controls.
We attempted to establish preliminary u-EPX cut-points. The estimated cut-points for acute asthmatics after treatment range from 65 to 75 mg/mmol Cr [10,25,26], with large standard deviations. For healthy controls, based on five studies, the u-EPX range was from 56 to 61 mg/mmol Cr [27,33,35,37,40]. However, there were three additional studies with ranges from 35 to 43 mg/mmol Cr [26,27,29]. There were three studies that had extremely close results for acute asthmatics-from 134 to 140 mg/mmol Cr [26,27,35]. However, there were four studies that did not reflect these values and had huge ranges and standard deviations: 103-233 mg/mmol Cr [25,27,28,33]. Hence, these cut points are not definitive, but serve as a first step towards further research in the diagnosis of asthma over time.
Translating the proposed EPX cut-points and values for diagnosing pediatric asthma into the clinical arena is not appropriate at this time. Accurate determination of cut points requires obtaining sensitivity and specificity measures for every study. At this time, there is no data in existence to assess sensitivity and specificity of u-EPX measurement for diagnosis of pediatric asthma.

Limitations
Although the present meta-analysis establishes an association between u-EPX and asthma and makes a case for clinical managing asthma, there remain a few limitations. Studies were conducted over 2 decades (from 1993 to 2015). This affected both the diagnostic criteria of childhood asthma as well as the methodologies used for estimating urine or serum EPX. Regardless, the diagnosis of asthma is exceedingly difficult to confirm in young children and infants, and usually requires years of follow-up. Different criteria were used among the different studies (e.g., American Thoracic Society). Furthermore, the classification of asthma into mild, moderate, or severe was determined through a variety of clinical tests including asthma scores, chest X-rays, measures of oxygen saturation, NO, and PEFR values.
Several studies were cross-sectional and did not assess asthma over time. There were differences in methodologies including equipment and testing for EPX (which can significantly alter the values of u-EPX and serum EPX) [41], and these measurements were not repeated posttreatment. Circadian rhythm (lowest levels at 7 PM and highest at 7 AM) may have introduced scatter of u-EPX [42], weakening possible correlations. There were also variations in performing PFTs (i.e., type, equipment) from one study to another.
Some studies included patients who were well-controlled with treatment, thereby underestimating EPX levels during exacerbations. A few studies used comparison groups such as population-based controls. There were also differences in sample selection, patient characteristics, and inclusion/ exclusion criteria. Most studies were conducted in Europe on a small sample. All studies were based on whites, with no consideration of other racial/ethnic backgrounds.
Although u-EPX is affected by age, symptomatic versus asymptomatic asthma, atopy, inhaled steroids, and infections specific for bronchial asthma [18], none of the studies adjusted for these effects. Additional confounding factors, including emotional triggers and environmental tobacco smoke, were also never considered.

Future directions
To date, this is the first systematic review to implicate a useful role for urine EPX in childhood asthma, as well as suggest age-specific cut-points for urine EPX for asthmatics and healthy young children. Urinary EPX, a simple and noninvasive technique, could be invaluable for diagnosing future asthma in children. Despite the limitations of the studies, a consistent trend of higher EPX levels with asthma was revealed. These results are too premature to draw any firm or generalizable conclusions. The utility of u-EPX as a clinically relevant diagnostic metric in childhood asthma has yet to be established. Further research is required to replicate these findings and to validate this biomarker in a methodologically sound population-based prospective cohort study consisting of measuring u-EPX over several years in age-specific (e.g., 0-4, 5-11, >12 years [43]) children with newly diagnosed probable persistent asthma, stratified by severity (i.e., mild, moderate, and severe), prior to and after initiation of longterm medications as well as during exacerbations, until a definitive diagnosis of pediatric asthma is confirmed. This will delineate the relationship between u-EPX and the progression of asthma; thereby determining whether this biomarker is potentially useful in the clinical arena.