Exhaled breath condensate proteomics using amphipols improves protein detection but reveals statistical challenges in respiratory disease biomarker discovery
BackgroundInvasive diagnostic techniques for pleural mesothelioma (PM) delay early detection, underscoring the need for non-invasive biomarkers. Exhaled breath condensate (EBC) is a promising matrix but has extremely low protein concentration. This study applies an amphipol (APol)-based concentration method to enhance detection of potential PM biomarkers in the EBC proteome. MethodsFour replicate EBC samples from a healthy individual were processed using APols or lyophilization. Subsequently, 145 EBC samples from PM patients (n=25), lung cancer patients (n=28), asymptomatic asbestos-exposed individuals (n=60) or healthy controls (n=32) were concentrated with APols and analysed by liquid chromatography coupled with tandem mass spectrometry. Technical variability from batch effects and collection devices was addressed using omicsGMF, for simultaneous imputation and covariates correction. Statistical analysis employed linear models with technical and clinical covariates. ResultsAPols outperformed lyophilization, yielding 1.7-fold higher proteome coverage and improved reproducibility. Although high data missingness was observed, proteome analysis across 145 samples identified 248 unique proteins. OmicsGMF corrected the substantial variability, while maintaining biological signal. Whereas unsupervised clustering did not reveal distinct disease-specific patterns, supervised statistical analysis detected overall subtle differences between clinical groups (PERMANOVA p-value = 0.008), though no individual proteins reached statistical significance after adjustment for technical and clinical covariates and correction for multiple testing. ConclusionAPols outperforms lyophilization for concentrating proteins in low-abundant EBC samples, enabling enhanced proteome coverage for biomarker discovery studies. However, the inherent high variability of EBC proteomes has a strong effect on detecting possible biomarkers across clinical groups. Key messageO_ST_ABSWhat is already known on this topicC_ST_ABSExhaled breath condensate (EBC) is an attractive non-invasive biofluid for respiratory disease biomarker discovery, but no protein biomarkers have reached clinical application despite extensive investigation. What this study addsUsing superior APols-based protein concentration and rigorous analytical and statistical methods in 145 EBC samples, we demonstrate that fundamental limitations - 96% peptide missingness, 80% skin-protein contamination, high variability, low protein identifications and low fold-changes- make EBC unsuitable for reliable protein biomarker discovery, when proper covariate adjustment is applied. How this study might affect research, practice or policyResearchers pursuing EBC proteomics should be cautious, as inadequate handling of missingness and variability, particularly without proper covariate adjustment, may yield findings that do not reflect true biological differences.