Accurate MS-Based Diagnostic Amyloid Typing Using Endogenously Normalized Protein Intensities in Formalin-Fixed Paraffin-Embedded Tissue.

Amyloidoses are a group of diseases characterized by the pathological deposition of non-degradable misfolded protein fibrils. These include plasma cell neoplasms, chronic inflammatory conditions, and age-related disorders, among others. Precise identification of the fibril-forming, and thereby amyloidosis-type defining protein is crucial for prognosis and correct therapeutic intervention. While immunohistochemistry (IHC) is widely used for amyloid typing, it requires extensive interpretation expertise and can be limited by inconclusive staining results. Thus, mass spectrometry (MS), if available, has been proposed as the preferred method for amyloid typing by international specialized centers (United States and United Kingdom) using primarily spectral counts for quantification. Here, we introduce an alternative method of relative quantification to further enhance the accuracy and reliability of proteomic amyloid typing. We analyzed 62 formalin-fixed, paraffin-embedded (FFPE) tissue samples, primarily endomyocardial biopsies, using liquid chromatography-tandem mass spectrometry (LC-MS/MS) and employed internal normalization of iBAQ values of amyloid-related proteins relative to serum amyloid P component (APCS) for amyloidosis typing. The APCS method demonstrated robust performance across multiple LC-MS/MS platforms and achieved complete concordance with clear cut IHC typed amyloidosis cases. More importantly, it resolved unclear amyloid cases with inconclusive staining results. Additionally, for samples without a distinct fibril-forming protein identified in the standard procedure, de novo sequencing uncovered immunoglobulin light chain components, enabling the diagnosis of rare AL-amyloidosis subtypes. Finally, we established machine learning approach (XGBoost) achieving 94% accuracy by using ∼160 amyloid-related proteins as input variables. In summary, the iBAQ APCS normalization method extended by de novo sequencing allows robust, accurate, and reliable diagnostic amyloid typing, and can be complemented by an AI-based classification. Careful reviewing of each histological sample and the clinical context, nevertheless, remains indispensable for accurate interpretation.