Concentrations of HMGB1 and Hsp70 of healthy subjects in upper and lower airway: Literature Review and Meta-analysis

Although high-mobility group box 1 and heat-shock protein 70 are implicated in airway diseases and suggested as relevant diagnostic biomarkers, their control concentrations in the airways have not yet been determined. This study aimed to evaluate concentration of healthy subjects for both these proteins in the upper and lower airways via meta-analysis. We searched MEDLINE, EMBASE, and Google Scholar for articles describing concentration of healthy subjects for these proteins. Data from healthy populations were combined using a random-effects model, and subgroup and sensitivity analyses were performed to determine between-study heterogeneity. We analyzed 22 studies involving 485 patients. Concentration of healthy subjects of high-mobility group box 1 and heat-shock protein 70 varied from “not detected” to 326.13 ng/mL and from 0.20 pg/mL to 9240.00 pg/mL, respectively, with the values showing significant heterogeneity. Subgroup analysis for high-mobility group box 1 revealed 13.63 ng/mL (95% CI 12.13-15.14), 100.31 ng/mL (95% CI -31.28-231.91), 9.54 ng/mL (95% CI 8.91-10.17), and 65.82 ng/mL (95% CI 55.51-76.14) for the lower airway, upper airway, pediatric populations, and adults, respectively, whereas that for heat-shock protein 70 revealed 20.58 pg/mL (95% CI 7.87-33.29) for the lower airway and 9240.00 ±11820 pg/mL for the upper airway. Although concentrations of healthy subjects of these proteins varied in the upper and lower airways, the levels of both these proteins were higher in the upper airway than in the lower airway, and these concentrations differed according to the age and sampling procedure. Our findings support the further evaluation of these proteins as biomarkers for airway-related diseases.


Introduction
The airway mucosa is important as it serves as the first line of host defense in the innate immune system [1,2]. Several structurally unrelated damage-associated molecular patterns (DAMPs) have been identified since the introduction of "the danger model" [1], with these molecules being released from host cells in response to cellular stress or tissue injury. DAMPS are capable of inducing inflammatory responses by activating the innate immune system as important mediators of the innate immune response [2].
High-mobility group box 1 (HMGB1) is an actively studied DAMP molecule. HMGB1 functions as a cytokine following its extracellular release from injured cells, activated sensory neurons [3], or cells associated with the immune system (macrophages and dendritic cells) [4]. It is a nuclear protein that regulates gene expression by binding to DNA in a non-sequence-specific manner, with its affinity for DNA being regulated by phosphorylation and Ivyspring International Publisher acetylation [4]. In the extracellular compartment, HMGB1 binds to receptors, including toll-like receptor (TLR) 2, TLR4, and receptor for advanced glycation end products (RAGE) [5,6], which triggers the activation of proinflammatory signaling pathways and exerts pleiotropic effects depending on the cell type [5].
Recently, heat-shock protein (Hsp) 70 was suggested as another DAMP molecule. Unlike HMGB1, Hsp70 is present in the cytoplasm and functions as a highly conserved molecular chaperone by facilitating intracellular protein folding and the degradation of misfolded proteins. Upon its extracellular release during cell stress or injury, Hsp70 activates both TLR2 and TLR4 signaling [2,7].
We previously reported the expression of both HMGB1 and Hsp70 in upper airway epithelial cells and their release into the extracellular area [5,[8][9][10]. Additionally, previous studies have reported HMGB1 and Hsp70 expression and their roles as DAMPs in airways [11][12][13]. Although multiple studies have demonstrated that HMGB1 and Hsp70 function as DAMPs and suggest their importance as diagnostic markers in upper and lower airway diseases, their concentration of healthy subjects under physiological conditions remain unknown. Although the upper and lower airways are generally considered to be united, their immunological conditions are suggested to be different [14]. This suggests that the concentrations of HMGB1 and Hsp70 and their potential immunological roles in the upper and lower airways may also differ. In this study, we performed a metaanalysis to determine the concentration of healthy subjects of HMGB1 and Hsp70 in the upper and lower airways and analyzed their differences through a literature review.

Methods
We developed and prospectively registered the study protocol using PROSPERO (registration number: CRD42018105478; www.crd.york.ac.uk/ PROSPERO). This study was performed according to the Meta-analysis of Observational Studies in Epidemiology guidelines [15] and reported according to the Preferred Reporting Items for Systematic reviews and Meta-Analysis guidelines [16].

Inclusion and Exclusion Criteria
All randomized controlled studies, cohort studies, case-control studies, including nested case-control studies, and cross-sectional studies reporting normal values of HMGB1 and Hsp70 in the upper and lower airways of healthy populations were included. We included only studies that performed enzyme-linked immunosorbent assay (ELISA) to objectively quantify the HMGB1 and Hsp70 levels in nasal lavage (NAL) or bronchoalveolar (BAL) fluids or induced sputum.
The exclusion criteria were: 1) review articles, case reports, case series, letters to the editor, commentaries, proceedings, abstracts without complete published articles, laboratory studies, and any other non-relevant studies; 2) studies reporting HMGB1 and Hsp70 levels in patients with a disease; 3) studies reporting HMGB1 and Hsp70 levels in animals; 4) studies reporting HMGB1 and Hsp70 levels in regions other than the airways; and 5) studies performing a relative comparison of HMGB1 and Hsp70 concentrations using western blot, real-time polymerase chain reaction, or immunohistochemistry. No language limitations or date restrictions were imposed.

Information Sources and Search Strategy
Two authors (HJM and HK) independently performed a database search using OVID-MEDLINE, EMBASE, and Google Scholar in August 2018 and updated the search results in March 2019. Additionally, the reference lists of the identified studies and eligible articles were searched manually. The search strategy, which included a combination of free text, medical subject headings, and EMBASE subject headings, is described in Appendix S1 in the Supplementary Information.

Study Selection
Two authors (HJM and KSK) independently scanned the titles and abstracts of the reports identified via the search strategies and retrieved the full paper upon determining the eligibility of the study. Potentially relevant studies selected by at least one author were retrieved, and the full-text versions were evaluated. Two authors (HJM and KSK) discussed the relevance of each study to arrive at a consensus regarding study inclusion or exclusion. Disagreements were settled via discussion with a third investigator (HK). None of the authors had any objection to the final results.

Data Extraction
All interrelated data from the included studies were independently extracted, entered in a standardized form by two authors (HJM and KSK), and crosschecked. When an agreement could not be reached, the dispute was resolved with the aid of a third investigator (HK). The standardized form included the title, name of the first author, name of the journal, year of publication, study design, the country where the study was performed, sex, age, number of patients, history of smoking, the location where the sample was drawn, sampling method, analysis method, normal range of Hsp70, and normal range of HMGB1. The data were initially extracted from the tables or text. In cases of missing or incomplete data, an attempt was made to contact the authors to obtain the relevant information. Whenever the reported data were in the form of a median (P25-P75), median (range), or mean (standard error of the mean), the mean and standard deviation were calculated from these values [17,18]. For data presented as figures rather than numbers [19][20][21][22][23], the open-source software Plot Digitizer (v.2.6.8; http://plotdigitizer. sourceforge.net) was used to extract the numbers.

Statistical Analysis
Data were combined using a random-effects model, and subgroup and sensitivity analyses were performed to determine between-study heterogeneity. We performed all analyses using Stata SE (v.15.0; StataCorp, College Station, TX, USA) and comprehensive meta-analysis software (v.2.0; Biostat, Englewood, NJ, USA). The methods used for statistical analysis are described in detail in Appendix S1 in the Supplementary Information.
The Hsp70 concentration in the upper airway of healthy populations was reported in only one study (9240 ± 11,820 pg/mL) [8].

Discussion
DAMPs, such as HMGB1 and Hsp70, are conducive to inflammatory damage associated with inflammatory diseases of the airways [2,49]. Because the concentrations of HMGB1 and Hsp70 can be elevated and have been suggested as potential biomarkers in various diseases, verifying the normal control concentrations of these molecules in the airways is essential to assess their efficacy as biomarkers.
This study revealed that HMGB1 and Hsp70 are present in substantial concentrations in the extracellular areas in the upper and lower airways of healthy controls. Analysis of the primary outcome showed that the mean concentration of HMGB1 was 13.63 ng/mL in the lower airway [20,21,25,30,34,36,[39][40][41][42][43][44][45][46] and 100.31 ng/mL in the upper airway [5,19,22,37] and that the mean concentration of Hsp70 was 20.58 pg/mL in the lower airway [22,47,48] and 9240.00 pg/mL in the upper airway [8]. However, a limited number of studies evaluating the status of these molecules under normal circumstances, especially their status in the upper airway, have been published, making it difficult to accurately determine their concentration of healthy subjects.
The mean concentrations of HMGB1 and Hsp70 were higher in the upper airway than in the lower airway. The accumulated data suggest that inflammatory diseases of the upper and lower airways share a common background, which is sometimes described as "one airway, one disease" [14,50]. However, immunological responses in the upper and lower airways are not always identical. Compared with the upper airway mucosa, which is exposed to various microorganisms and toxic materials, the lower airway mucosa is less frequently associated with pathologic microbes [14]. Additionally, the epithelial-barrier cells in the upper airway epithelium differ from those in the lower airway [14]. Therefore, we hypothesized that the concentrations of DAMPs might differ between the upper and lower airways. In this analysis, the mean concentrations of both HMGB1 and Hsp70 were higher in the upper airway than in the lower airway, although the difference was not statistically significant, which could be attributed to the limited number of studies evaluating these levels. Nevertheless, our findings support the existence of different immunological responses between the upper and lower airways. This meta-analysis demonstrated different mean concentrations of HMGB1 in induced sputum (17.62 ng/mL) [39,[42][43][44]48] and BAL fluid (6.19 ng/mL) [20,21,23,25,34,38,40,41,[45][46][47].
Sputum analysis represents a non-invasive and cost-effective method potentially useful in clinical practice as a diseasescreening procedure [49]. However, because the mean concentration of HMGB1 was higher in sputum than in BAL fluid, a known control level for HMGB1 is necessary before its consideration as a diagnostic biomarker. Additionally, NAL fluid is more safely and easily collected from patients (i.e., via nasal irrigation) relative to procedures involving tissue harvesting or blood sampling. A review described previous studies evaluating the importance of HMGB1 in upper airway inflammatory diseases, such as rhinosinusitis and allergic rhinitis; however, the analyses were performed using nasal tissue [51]. In this analysis, we found only a limited number of studies using NAL fluid [5,8,19,22]. Therefore, clinical studies using NAL fluid and a larger cohort should be performed to analyze the potential significance of HMGB1 and Hsp70 concentrations as non-invasive and efficacious biomarkers for diagnosing upper airway diseases. Furthermore, standardized sampling method for airway fluid collection should be set to get control reference range for these potential biomarkers in regional area.
The HMGB1 concentration differed between pediatric patients and adults. The HMGB1 concentration in the upper airway was 9.27 ng/mL (95% CI: -8.47-10.07) in the pediatric population [37] and 160.20 ng/mL (95% CI: -156.58-163.83) in adults [5,22], whereas the HMGB1 concentration in the lower airway was 9.99 ng/mL (95% CI: -8.96-11.01) in pediatric patients [42,45] [20,21,25,30,34,[38][39][40][41]43,44,46]. A previous study identified HMGB1 as a candidate gene involved in aging and age-related diseases in humans [52], with another study reporting changes in its expression during aging in the mouse brain [53]. This suggests that the HMGB1 concentration in the upper and lower airways could depend on patient age. Given that NAL fluid and sputum collection represents a safe and simple procedure in pediatric subjects, further evaluation in children is required to compare changes in HMGB1 concentrations according to age. Additionally, the concentration of healthy subjects of extracellular HMGB1 (ng/mL) were higher than those of Hsp70 (pg/mL) in both airways. Both HMGB1 and Hsp70 are ubiquitous proteins; therefore, the determination of their DAMP-specific functions as inflammatory modulators is potentially important. Interestingly, a study by Tasaka et al. [34] was the only one to report failure to detect HMGB1 in BAL fluid from healthy controls, though they detected HMGB1 in other groups of patients, such as those with autoimmune disease, malignancy, and acquired immunodeficiency syndrome. Moreover, sputum analysis by another study revealed HMGB1 concentrations ranging from 0 ng/mL to 45.2 ng/mL [38]. This large concentration range could be the result of the excessive dilation of target samples or instability of the sampling method; therefore, standardized sampling methods are required.
There are limitations to this meta-analysis. First, compared with studies involving pathological conditions, studies evaluating the control concentration of HMGB1 and Hsp70 in regional airways are very limited, especially, in case of Hsp70 in upper airway. Therefore, the relatively small sample size of the included studies might have influenced the heterogeneity of the results. As interaction between HMGB1 and Hsp70 has been reported in previous study [54], large population study considering HMGB1 and Hsp70 at the same time might reveal true control concentrations. Second, in some studies, the authors did not provide original data regarding the actual concentrations. Although we attempted to contact the authors to obtain this original data, we were unsuccessful in some instances. In those cases, we extracted the data regarding the HMGB1 and Hsp70 concentrations using Plot Digitizer. Thrid, we did not separate Hsp70 into subtypes owing to the small sample size of the enrolled studies [47]; however, we note that Hsp70 is among the most commonly studied members of the Hsp family and can be sub-grouped accordingly [55]. Furthermore, the levels of other Hsps might be worth evaluating in the context of this analysis. Finally, we did not evaluate the effect of smoking history as only a few studies supplied this information [20,30,34,[38][39][40]42,43,48]. Taylor et al. [56] suggested that smoke exposure could influence the concentration of HMGB1. Also, specific information of enrolled population such as racial differences or obesity was not considered in this study. Therefore, the effect of specific characteristics such as smoking history, racial difference, and obesity on healthy subjects' of HMGB1 and/or Hsp70 concentration in regional airways should be evaluated further.

Conclusion
Overall, in this study, although the concentration of healthy subjects of these proteins varied in the upper and lower airways, the levels of both were higher in the upper airway than in lower airway, and the concentration of healthy subjects differed according to the age and sampling procedure. These findings support the further evaluation of these proteins as biomarkers for airway-related diseases.