Exhaled biomarkers in adults with non-productive cough

Non-productive cough was defined as a positive reply to the question “Have you during the last 12 months had dry cough, i.e. cough without sputum?” Altogether 24 participants answered “Yes” and 83 answered “No”. Of the 24 participants with non-productive cough, 10 were current smokers, and therefore 14 were eligible for the main analysis.

Of the 83 participants without cough, 28 were current smokers, and therefore 55 were eligible for the main analysis.

The participants performed spirometry using the Spirare spirometer (Spirare, Stockholm, Sweden). Reversibility test was performed using 400 µg of salbutamol in accordance with the ATS/ERS criteria [17]. Forced vital capacity (FVC), forced expired volume in one second (FEV1) an FEV1/FVC ratio were expressed as a percentage of the reference value (% pred) according to Brisman et al. (note the corrigendum) [18].

The participants also performed IOS using a Jaeger Masterscreen system (CareFusion, Würzburg, Germany) before and after bronchodilation with 400 µg of salbutamol in accordance with the ERS task force criteria [19]. The mean values of resistance at 5 Hz and 20 Hz (R5 and R20), frequency dependence of resistance (R5-R20) and area under the reactance curve (AX), were calculated and expressed as % predicted according to Kjellberg et al. [20]

Asthma was defined as a positive reply to the question “Do you have a physician-diagnosed asthma?”.

Exhaled particles (PEx) were collected using the PExA instrument version 1.0 (PExA AB, Gothenburg, Sweden), as previously described [21, 22]. The measured particle sizes cover diameters between 0.41 and 2.98 µm. Subjects inhaled HEPA-filtered air for a minimum of three breaths before the sampling in order to remove particles from ambient air. All participants wore a nose clip throughout the procedure. A standardized breathing manoeuvre was used [23, 24], starting with an exhalation at normal flow rate to residual volume, breath holding for 5 s, followed by a maximal inhalation to total lung capacity, immediately followed by a normal exhalation to functional residual capacity. Between breathing manoeuvres, the subject breathed particle-free air tidally for 30 to 60 s. Each sampling session continued until 120 ng of exhaled particles were collected. After collection the sample holder was transferred to a clean air room and the substrate was cut out with a scalpel from the sample holder area and placed in Millipore Ultrafree-MC LH Centrifugal Filter insert (FC30LH25) and stored at – 80 °C for subsequent extraction and SOMAscan analysis. True blank samples were generated by applying the same sample handling procedure as for real samples but without collecting PEx sample from the study subjects.

The SOMAscan (SomaLogic Inc, Boulder, USA) proteomics platform that uses slow off-rate modified DNA aptamers can simultaneously quantify more than 1300 human proteins in blood samples. SOMAscan analysis of PEx samples and processing of data has been previously described in detail [16]. In short, prior to SOMAscan analysis, the volume of sample buffer was adjusted to reach the same concentration of PEx in all samples in order to normalize the samples for the differences in the collected amount of PEx. Intra-plate and inter-plate normalization were performed by SomaLogic according to their SOMAscan assay good laboratory practice (GLP) data quality-control procedures. Limit of detection (LOD) was calculated as the relative fluorescent unit (RFU) mean plus 3 standard deviations based on two blank samples. For the primary analysis, proteins with (RFU) values > LOD in more than 50% of the samples were considered for further analyses. For a secondary analysis, proteins above the LOD among more than 50% of those with non-productive cough, but among less than 50% of those without cough, were identified.

As mentioned above, active smoking can have a significant effect on the small airway protein profile, and therefore current smokers were excluded from the primary analysis.

Statistical analyses of the protein data were performed using general linear model-based statistics (Qlucore Omics Explorer 3.8 software, Qlucore AB, Lund, Sweden). SOMAscan data was log₂ transformed before the analysis to achieve normal distribution. General linear model, with each variable normalized to mean 0 and variance of 1, was used to determine differences in protein abundance between subjects with and without non-productive cough. The analysis was adjusted for the investigator performing the PExA measurements, as well as the sex of the subjects.

To identify which biological pathways the identified proteins were mostly involved in through protein–protein interactions, a STRING (Search Tool for the Retrieval of INteracting Genes/proteins) analysis was performed using an online available tool [25].

In keeping with the exploratory nature of this study, protein differences between groups with a p-value < 0.05 were considered to be of interest. A q-value was also calculated using the Benjamini–Hochberg correction for multiple testing adjustment. Statistical analysis of clinical and demographic data was performed using IBM SPSS Statistics for Windows, version 28 (IBM Corp., Armonk, N.Y., USA) with the significance level set to p < 0.05.

The participants’ baseline characteristics are presented in Table 1. In summary, participants with non-productive cough had a similar age, BMI, CRP, blood eosinophils, IOS, and FeNO, as participants without cough. Also, spirometry results were largely similar, except for a slightly higher FVC percent predicted among participants with non-productive cough, compared with participants without cough, however the p-value did not reach the level of statistical significance (Table 1). Asthma was present in one of 14 participants with non-productive cough.

All in all, 203 proteins were detected in at least 50% of the 107 samples. Results from the analysis of the proteomics profile found 75 proteins significantly altered among participants with non-productive cough compared to controls, after adjusting for sex and investigator performing the PExA measurement (all with p-value < 0.05 and q-value ≤ 0.13, Table 2, Figs. 1 and 2). Thereof, 21 proteins had a q-value < 0.05. The five proteins with the most significant difference (p ≤ 0.0008, q ≤ 0.028) between participants with and without non-productive cough were adhesion G protein-coupled receptor E2 (fold change 0.64, p < 0.0001, q = 0.005), endothelial cell-selective adhesion molecule (ESAM) (fold change 0.79, p = 0.0002, q = 0.015), complement factor H (fold change 2.35, p = 0.0002, q = 0.015), polymeric immunoglobulin receptor (fold change 0.51, p = 0.0008, q = 0.028), and complement factor B (fold change 1.64, p = 0.0008, q = 0.028) (Table 2). The protein kallikrein was also significantly more common among participants with non-productive cough (fold change 1.81, p = 0.002, q = 0.04). Further details are seen in Table 2 and Fig. 1.

Additional 6 proteins were identified above the LOD among more than half of those with non-productive cough, but in fewer subjects without cough. The proteins Tenascin-C, immunoglobulin M, and Neurogenic locus notch homolog protein 3 (NOTCH3) had a difference of more than 20 percentage points. Further details are given in Table 3.

A sensitivity analysis including current smokers, with adjustment for smoking status, found no significant differences in the findings described above (results not shown).

The STRING analysis identified mostly proteins associated with pathways regulating the immune and inflammatory responses, as well as with complement activation and coagulation (Fig. 3).

Some of the novelties and main strengths of this study are the combination of a non-invasive collection of non-diluted biosamples from the small airways, and the unbiased proteomic approach. However, some methodological issues need to be discussed. First of all, the sample size is small and an independent validation cohort is needed to confirm our findings. We also observed variability in the protein-profile depending on the investigator performing the PEx sampling, possibly explained by differences in how the breathing maneuver was performed, and as a result the statistical analysis had to be adjusted accordingly. Choosing to include proteins in the analysis with RFU values > LOD in more than 50% of samples could be considered another possible limitation. However, due to the exploratory nature of the study, we chose to select 50% as the limit instead of other commonly used percentages such as 70–80%, to open the analytical window and explore proteins that could potentially be missed otherwise. For that same reason and also due to the small sample size, a more inclusive approach was chosen and differences in protein abundance were considered significant at p-value < 0.05.

Some potentially interesting biomarkers, such as the neuroinflammatory marker Substance P which is implicated in cough [12], were not among the proteins analysed by the SOMAscan platform. Also, in part because of the relatively low sample volume, some proteins were not easily identified, such as Cadherin-1 which was only detected above LOD in 6% of all samples.

Because of the exploratory nature of this study, strong conclusions cannot be made, even though the results show interesting findings. The clear differences in abundance of many proteins between participants with non-productive cough and those without cough, merit larger studies to validate these findings, preferably including clinical patients with chronic cough.