Developing breath volatiles fingerprinting for Alzheimer's and Parkinson's diseases

Abstract Background We tested the feasibility that an AD‐breath volatiles (BVOC) fingerprint will be useable to predict, track disease progression in combination with more traditional modalities, and therapeutic interventions responsiveness for AD resulting in an inexpensive, non‐invasive point‐of‐care tool. Additionally, we also hypothesized that the AD BVOC fingerprint will be functional for early differentiation of AD from other neurodegenerative diseases. Method Alveolar breath was collected by breathing into a sampling device and mass spectrometric analysis. Discriminatory statistical modeling was applied to the identified BVOCs, classifying MCI versus controls and identifying BVOCs with greatest discriminatory impact. Differentiation between AD and PD was examined by mixing our MCI and previously collected PD data. Result MCI from controls were classified with 97.8% accuracy with an ROC Curve AUC of 0.998. PD or MCI samples were differentiated with 92.5% correctly controls versus MCI or PD and 100% as either MCI or PD. Investigation of probable mechanistic BVOC formation pathways in MCI/AD resulted in two primary pathways: neuroinflammation and e stress and dysregulated glucose metabolism with differences between inflammatory BVOCs in AD versus PD. Conclusion We demonstrated feasibility of MCI/AD BVOC fingerprinting for early stage identification of disease via a 97.8% correct classification and the ability to differentiate between AD and PD as example of two neurodegenerative diseases. We showed the probable dysregulated glucose metabolism and inflammatory pathways via BVOCs.