Human epididymis protein 4 is a new biomarker to predict the prognosis of progressive fibrosing interstitial lung disease

BACKGROUND The clinical course and prognosis of progressive fibrosing interstitial lung diseases ( PF-ILDs) vary from individual to individual. Predictive serum biomarkers for the management of the disease are needed. Serum human epididymis protein 4 (HE4) has been reported to be elevated in patients with IPF, yet its clinical utility has not been elucidated. We evaluated whether serum HE4 could be a biomarker for patients with PF-ILD. METHODS Serum HE4 was measured in a retrospective study that consisted of 34 patients with PF-ILD and 40 healthy volunteers. The relationship between serum HE4 levels and clinical parameters or prognosis was investigated. To validate the significance of the results, a prospective observational study that consisted of 37 patients with PF-ILD and 40 control patients without PF-ILD was performed. RESULTS Serum HE4 was higher in patients with PF-ILD than in health volunteers ( P < 0.01). A correlation of the serum HE4 levels with the extent of honeycombing on chest high-resolution computed tomography was identified (r = 0.41, P = 0.015). In multivariate analysis by the Cox proportional hazard model, higher HE4 levels (> 238 pmol/l) were associated with elevated mortality risk (HR 7.27, 95% CI 1.56-34.0, P = 0.01 in the derivation cohort; HR 44.3, 95% CI 4.19-468, P < 0.01 in validation cohort). Values or median (25th and 75th percentile). A comparison between PF-ILDs groups was performed using test, chi-square(χ2) test, and Mann-Whitney's U test appropriate.


Introduction
Interstitial lung diseases (ILDs) encompass a large variety of diseases that are caused by a known etiology (such as autoimmune disease, vasculitis, drugs, tumors and occupational or environmental exposure) or unknown etiology (idiopathic). Among ILDs, idiopathic pulmonary fibrosis (IPF), which is a representative chronic fibrosing lung disease, has a poor prognosis with a median survival of 2-3 years [1]. Some types of ILDs other than IPF (such as chronic hypersensitivity pneumonitis (CHP), connective tissue disease-associated ILD (CTD-ILD), idiopathic nonspecific interstitial pneumonia (iNSIP), unclassifiable ILD and asbestosis) are also known to have progressive disease behaviors similar to IPF [2,3]. Although ILDs with a progressive phenotype are termed progressive fibrosing-ILD (PF-ILD), these disease behaviors are unpredictable and widely variable. Therefore, clinical biomarkers to predict progression are required.
Human epididymis protein 4 (HE4), also known as whey-acidic-protein (WAP) four-disulfide core domain protein 2 (WFDC2) containing two WAP domains, was first identified as a secretory protein in the human epididymis [4,5]. Several studies have revealed that HE4 is expressed in multiple tissues, such as the respiratory tract, kidney, prostate, and ovarian cancer [6][7][8][9]. Although its biological role remains largely unknown, it has been suggested that HE4 has a functional role in the formation of fibrosis. The correlation between renal fibrosis and serum HE4 levels in patients with kidney disease and the function of HE4 in inhibiting matrix metalloproteinase 2 and 9 (MMP2, MMP9), inducing the degradation of type I collagen, have been reported [10]. For the lung, serum HE4 levels have been reported to be higher in patients with IPF than in patients without respiratory disease [11]. However, the clinical association between serum HE4 levels and pulmonary fibrosis has not been elucidated.
Therefore, we performed this study to evaluate whether serum levels of HE4 are elevated in patients with PF-ILD or can be used as a biomarker. We retrospectively measured the serum levels of HE4 and analyzed survival data, high-resolution computed tomography (HRCT) findings, and other clinical variables in correlation with HE4. The usefulness of HE4 as a biomarker was confirmed in a prospective cohort study.

Study Design and Population
This study was approved by the Institutional Review Board at the Tokyo Medical and Dental University (TMDU) in accordance with the Declaration of Helsinki (approval number M2019-010). The study consisted of two cohorts. First, a retrospective medical record-based observational cohort study, which served as a derivation cohort, was performed. One hundred two patients with ILD who were admitted to TMDU between June 2010 and July 2011 were included. Among the 102 patients, serum samples from 47 patients were stored. A total of 34 patients who were diagnosed with PF-ILD were analyzed ( Figure 1A). Forty staff members from our hospital and department were evaluated as healthy volunteers (HVs). Second, a prospective observational cohort study, which served as a validation cohort, was performed. Informed consent from all of the patients was obtained. In addition to the patients recruited at the time of diagnosis, patients attending our outpatient department on a routine basis were recruited from December 2012 to July 2014. Patients with bronchial asthma, chronic obstructive pulmonary disease, and bronchiectasis were also recruited as disease controls during the same study period. Serum samples were collected from 39 patients with PF-ILD and 40 control subjects during the recruitment period (Figure1B). Patients with cancer at baseline and an estimated glomerular filtration rate (eGFR) of less than 50 ml/min/1.73m 2 were excluded in the two cohorts. None of the patients experienced any episode of acute exacerbation of ILD prior to recruitment or overlapped in the two cohorts.
Clinical data of patients were included age, gender, smoking history, laboratory findings, physiological test results, radiological findings, and the etiology of ILD. The outcome of patients with PF-ILD was followed from baseline to 3 years.
The diagnosis for IPF and non-IPF IIPs (NSIP and unclassified) was based on the criteria from the American Thoracic Society and European Respiratory Society [12,13]. CTD was diagnosed based on the diagnostic criteria commonly used for each disease [14][15][16]. Chronic HP was diagnosed based on combinations of clinical, radiological, and histological criteria [17,18]. Acute exacerbation of ILD was diagnosed based on diagnostic criteria for acute exacerbation of IPF [19].
Blood samples for measuring HE4 were taken using a routine procedure, centrifuged, aliquoted, and stored at -20°C or below until analysis. The concentration of serum HE4 was measured by chemiluminescent microparticle immunoassay on an ARCHITECT ® i2000 analyzer (Abbott Japan LLC).

Evaluation of high-resolution computed tomography
HRCT scans were analyzed at the level of the aortic arch, carina trachea, and right pulmonary vein in each lung. The following 7 categories that were classified by parenchymal abnormalities were evaluated by one expert chest radiologist (M.K.) and one pulmonary specialist (Y.N.) with no knowledge of the clinical information: reticulation, honeycombing, centrilobular nodules, ground-glass opacity, consolidation, emphysema, and traction bronchiectasis (TBE). The extent of each radiologic abnormality, excluding TBE, was quantified as a percentage of lung parenchyma affected, to the nearest 5%, in each of the six lung zones. TBE was graded on a scale of 0-3 as follows: 0 = none; 1 = mild; 2 = moderate; and 3 = severe [20]. The average total score for each variable was calculated as the mean score of the six zones. The readings of the two observers on the extent of each abnormality were combined by calculating the averages. Interobserver agreement on the extent of each abnormality was evaluated with the Spearman rank correlation coefficient.

Evaluation of histological findings
Lung tissues were obtained from 8 patients with PF-ILD who underwent SLB from 2005 to 2013 in our institution. The lung tissues were retrospectively selected and analyzed separately from this observational study. Control lung tissues were obtained from normal parts of lungs excised in lung cancer surgeries.
The lung tissue sections were subjected to antigen retrieval by autoclaving for 20 minutes in citrate buffer after deparaffinization. Endogenous peroxidase was quenched with 3% H 2 O 2 , and 5% goat serum and 1% BSA were used for blocking. After the addition of rabbit monoclonal antibody to HE4 (Abcam, Japan, Cat. No. ab200828) overnight at 4°C, the sections were incubated with a biotinylated secondary antibody (Vector Laboratories) for 60 minutes followed by ABC reagent (VECTASATAIN) for 30 minutes. Next, the sections were incubated in 3,3'-diaminobenzidine tetrahydrochloride (DAB; VECTOR) solution and counterstained with Meyers' hematoxylin solution. Negative controls were incubated without primary antibody.

Statistical analysis
Differences between the groups were analyzed by Student's t-test, Welch's t test, Mann-Whitney U test, or Pearson's chi-square test. Correlations were determined by the Spearman rank test. The most suitable cutoff level for serum HE4 levels was determined using a receiver operating characteristic (ROC) curve analysis with the highest Youden index (i.e., sensitivity specificity-1). The Kaplan-Meier method was used to assess survival curves with GraphPad Prism version 7 (Graph Pad Software Inc., La Jolla, CA, USA). The log-rank test was used to evaluate the statistical significance of differences between the higher HE4 and lower HE4 groups. Cox proportional hazards models were used to examine the influences of serum HE4 levels on the prognosis of patients with PF-ILD. Statistical analyses were performed using EZR software version 1.37 (Saitama Medical Center, Jichi Medical University, Saitama, Japan) [21]. A P value < 0.05 was considered statistically significant.

Patient characteristics
The patient characteristics in the two cohorts are shown in Table 1. Patients with PF-ILD were significantly older than healthy volunteers (P < 0.01) in the derivation cohort. A greater proportion of patients with PF-ILD in the two cohorts were male (55.8% in the derivation cohort and 64.1% in the validation cohort). In the validation cohort, patients with PF-ILD had a significantly lower smoking history than control patients. The control subjects in the validation cohort consisted of 11 patients with bronchial asthma, 15 with chronic obstructive pulmonary disease (COPD), and 14 with bronchiectasis. The etiology of PF-ILDs mainly consisted of CHP (16/34, 47%) in the derivation cohort and IIPs (28/39, 71.7%) in the validation cohort. Krebs von den Lugen-6 (KL-6) and the estimated glomerular filtration rate (eGFR) were significantly higher in patients with PF-ILD in the derivation cohort than in those in the validation cohort (KL-6, P = 0.035; SP-D, P = 0.011). Diffusing capacity divided by the alveolar volume (DLCO/VA) was lower in the derivation cohort than that in the validation cohort (P = 0.002). Eighteen patients (steroid, n = 17; tacrolimus, n = 1) were undergoing immunosuppressive treatment at baseline in the derivation cohort. None of the patients had taken any treatment for PF-ILD at baseline in the validation cohort.
In the derivation cohort, serum HE4 levels in patients with PF-ILD were significantly correlated with age (r = 0.49, P < 0.01) and with the extent of honeycombing on HRCT (r = 0.41, P = 0.015) but not with serum KL-6 or SP-D levels ( Table 2). The assessments of the various abnormalities on HRCT showed good to excellent interobserver agreement (r = 0.73-0.81, P < 0.01, Supplementary Table   S1).

Identification of prognostic biomarkers
During the follow-up period, 9 patients (9/34, 26.4%) in the derivation cohort died. The causes of death were acute exacerbation (n = 5), chronic respiratory failure (n = 3), and malignant neoplasm (n = 1). First, to identify factors associated with the occurrence of acute exacerbation, characteristics were compared between patients who developed acute exacerbation and those who did not, but no factors related to acute exacerbation were identified (data not shown). Second, to identify factors associated with survival, characteristics were compared between survivors and nonsurvivors (Table   3). A serum HE4 level was the only factor significantly associated with the prognosis of PF-ILD. ROC analysis was conducted to determine the best cutoff value of serum HE4 level between the survivors and nonsurvivors (cutoff 238 pmol/l, AUC 0.76 (95% confidence interval (CI): 0.576 -0.944)). The patients were divided into a higher HE4 group (n = 9, HE4 > 238 pmol/l) and a lower HE4 group (n = 25, HE4 ≤ 238 pmol/l) to analyze the survival using the Kaplan-Meier method. The log-rank test showed a significant difference in survival between the two groups (P < 0.01) ( Figure 3A). The hazard ratio (HR) for mortality was calculated using the Cox proportional hazards model (Table 4). An association with an increased risk of mortality was observed in patients with serum HE4 levels above 238 pmol/l in the univariate model (HR 6.54, 95% CI: 1.71-25.0, P < 0.01), while no association with mortality risk was found in KL-6 and SP-D levels. Variables with a P value less than or equal to 0.20 in the univariate analysis were included in the multivariate analysis. Older age ( ≥ 65, HR 12.0, 95% CI: 1.31-109, P = 0.027) and higher serum HE4 levels (> 238 pmol/l, HR 7.27, 95% CI: 1.56-34.0, P = 0.012) showed statistical significance in the multivariate analysis.
Furthermore, the cutoff value of serum HE4 was validated for the prediction of survival in the validation cohort. Ten patients (10/39, 25.6%) died from recruitment to 3 years. The causes of death were acute exacerbation (n = 7), and chronic respiratory failure (n = 3). The log-rank test analyzed by the Kaplan-Meier method for survival between the higher (n = 4, HE4 > 238 pmol/l) and lower HE4 (n = 35, HE4 ≤ 238 pmol/l) groups showed a significant difference between the two groups (P < 0.01) ( Figure 3B). Multivariate analysis by the Cox proportional hazard model for mortality, using the same variables included in the analysis of the derivation cohort, identified higher HE4 as an independent predictor for mortality (HR 44.3, 95% CI: 4.19-468, P < 0.01).

Immunohistochemical analysis
The characteristics of the patients whose lung tissues were obtained are shown in e- Table 2

Discussion
Investigators who are researching PF-ILD have suggested that we should figure out biomarkers that can be easily obtained through simple, noninvasive examinations. In this study, we showed that serum HE4 levels were higher in patients with PF-ILD than in control subjects and that higher HE4 levels were associated with worsened prognosis of patients with PF-ILD. This is the first report to indicate that HE4 is valuable as a diagnostic and prognostic biomarker of PF-ILD.
PF-ILD is a progressive fibrosing phenotype with poor prognosis, but the individual disease course is heterogeneous [3]. The duration from the time when the progressive fibrosing phenotype was detected by physicians to patient death was reported as 30-45 months [22]. Therefore, biomarkers to predict early disease progression or death could be useful for making decisions during treatment. In the clinical setting, the methods of assessing progression are pulmonary function tests, laboratory data, evaluation of symptoms, the 6-min walk test and chest imaging [23]. Lower forced vital capacity (FVC) and a decline in FVC > 10% predicted are established predictors of mortality in patients with IPF [24]. Among several serum biomarkers, higher KL-6 levels (> 1000 U/ml) are associated with increased mortality in patients with IPF [25, 26]. However, KL-6 was not identified as a prognostic biomarker in the present study. Since patients with CHP who have been reported to have seasonal variation in the elevation of KL-6 levels [27] were predominantly included in our study, the difference in the etiology of ILD might have led to the result.
Elevation of serum HE4 levels has been reported in patients with lung cancer [28], ovarian cancer [29], and chronic kidney disease [30], and has already been applied to the diagnosis of ovarian cancer in clinical practice. Therefore, these patients were excluded from this study to avoid cofounding factors. Another previous study has showed a positive correlation between serum HE4 levels and age [31], which was similar to our result. In our derivation cohort, the serum HE4 levels were significantly higher in the patients with PF-ILD than in the healthy volunteers, but a significant difference in age between the two groups might affect this result. Therefore, in our validation cohort, control subjects who were close to the age of patients with PF-ILD were enrolled. Although the proportion of subjects with past smoking history was higher in control subjects than in patients with PF-ILD, a positive correlation between past smoking history and serum HE4 levels was identified [31].
Taking into account these factors, patients with PF-ILD had significantly higher serum HE4 levels than control subjects .
The reason for the association between HE4 levels and prognosis is currently unclear. The expression of HE4 in the lung has been identified in epithelial cells of the airway but not in peripheral lung [7]. In our immunohistochemistry analysis of lung specimens, HE4 was not observed in the bronchiolar epithelium or lung parenchyma of normal lungs. However, we found that HE4 was expressed in metaplastic bronchiolized epithelium lining regions of honeycombing and/or bronchiolar epithelium in PF-ILD lungs. Since HE4 is a secretory protein [32], this result might be associated with elevation of serum HE4 concentration in patients with PF-ILD. HE4 suppresses the activity of multiple proteases and especially inhibits MMP2/9 mediated degradation of type I collagen in fibrotic kidneys [10].
MMP2/9 have also been shown to be upregulated in IPF [33], and HE4 might be associated with lung fibrosis by the a mechanism similar to the profibrotic process identified in the fibrotic kidney.
Although the role of HE4 in the formation of pulmonary fibrosis does not go beyond the range of speculation, we found a correlation between serum HE4 levels and the extent of honeycombing on a HRCT findings. Honeycombing is one of the pivotal CT characteristics for the diagnosis of IPF and is defined as clustered cystic airspaces of the same diameters (approximately 3-10 mm) with thick, well-defined walls in the subpleural area [34]. The clinical significance has been suggested that pathological honeycombing reflects the end-stage of fibrotic changes [35]. The correlation between the serum HE4 levels and the HRCT findings implies that HE4 may play an important role in ongoing fibrosis and disease progression in PF-ILD.
The present study had several limitations. First, this study was a single-center study with a small number of analyses. Second, several types of PF-ILD were examined altogether in our analysis. We did not examine potential differences in the elevation of HE4 according to the etiology of PF-ILD.

Ethics approval and consent to participate
This study was approved by the ethics committee of Tokyo Dental and Medical University (approval number M2019-010). Informed consent from subjects in the retrospective cohort was obtained in the form of opt-out. Written informed consent was obtained from subjects in the prospective cohort before recruitment.

Consent for publication
Not applicable.

Availability of data and materials
All the data are available from the corresponding author on reasonable request.

Competing interests
Abbott Japan LLC covered the cost for serum HE4 measurement.
The authors declare that they have no competing interests.

Supplementary Files
This is a list of supplementary files associated with this preprint. Click to download. HE4_Supplementary_Table.docx