Multidimensional Immunophenotyping Identifies Hallmarks of Systemic Light-Chain Amyloidosis (AL) and Maps the Disease in the Crossroad between MGUS and Multiple Myeloma (MM)

Background: Since survival in AL mainly depends on the extent of organ involvement of patients at presentation, early diagnosis and risk stratification are key to improve patients' outcome. Therefore, together with surrogates of organ involvement, biomarkers identifying patients with MGUS or MM at greater risk of developing AL would be highly valuable to prevent organ damage, to maximize therapeutic efficacy and to improve outcomes in AL.

Aim: To investigate the value of multidimensional flow cytometry (MFC) for simultaneous fast diagnostic screening of plasma cell (PC) clonality and risk stratification, as well as to identify immunophenotypic markers useful for the selection of patients with monoclonal gammopathies candidates for monitoring of pre-symptomatic organ damage related to AL.

Methods: We used MFC to characterize a large series of patients with newly-diagnosed (ND) AL (N=94) vs MGUS (N=20) and NDMM (N=52), as well as age-matched healthy adults (HA, N=30). For each patient with AL, automated risk stratification was performed using principal component analysis (PCA) based on the relative frequency of bone marrow (BM) PCs, plus the percentage of clonal and normal PCs within the whole BM PC compartment, vs a database containing information on the same three parameters from a total of 1,774 patients, including 497 MGUS and 1,227 NDMM. In parallel, immunophenotypic protein expression profiles (iPEP) of AL patients were clustered using t-SNE, and the comparison between the iPEP of clonal PCs from patients with AL vs MGUS and MM cases was performed using canonical-correlation analysis (CCA). To identify additional immunophenotypic hallmarks of AL, the BM cellular composition in HA, MGUS, AL and MM patients was compared using 2-dimensional minimum spanning tree (MST) force-directed classification to determine the distance among individual cases.

Results: PC clonality was detected by MFC in 93/94 (99%) AL patients, whereas an M-component was detectable in 96% of cases by electrophoresis, immunofixation and sFLC. PCA as defined above, identified AL patients displaying an MM-like (n=6) and an MGUS-like (n=38) signature, as well as 49 cases with an intermediate signature between the MGUS and MM reference datasets. Multivariate analysis of baseline prognostic factors for survival, including patients' age, number of organs involved, Mayo staging, the percentage of BM PCs based on cytomorphology and eligibility for ASCT, showed that having an intermediate- or an MM-like profile had an independent adverse effect on patients' progression-free (PFS) and overall survival (OS) (HR:3.4; P≤.02).

t-SNE based on the iPEP of clonal PCs revealed two major clusters of AL patients with significantly different PFS, defined by opposite patterns of expression for CD45, CD56 and CD138 (P≤.02). CCA of tumor iPEP showed partial overlap between AL vs MGUS and MM, with progressively higher percentages of cases with a CD38^lo, CD45^-ve, CD81^-ve and CD138^lo iPEP being observed from MGUS to AL and MM. In contrast, AL patients displayed significantly lower reactivity for CD56 (P≤ .03).

Further characterization of the BM cellular composition allowed the systematic assessment of 16 cell populations and 18 phenotypic parameters that, by MST, mapped AL in between MGUS and MM. Of note, while AL patients displayed a predominantly-clonal PC compartment in the absence of an MM-like tumor PC expansion, the percentage of B-cell precursors was consistently lower in AL patients than in HA, MGUS and MM (P=.004). Thus, using optimal cut-off values to discriminate between AL vs MGUS and MM, we built a scoring model based on the presence of <100% CD56^+ve clonal PCs, <0.1% B-cell precursors, >80% clonal PCs within total BM PCs and <2% BM PCs. Overall, a significant (P<.001) association was found between a progressively higher score and the diagnosis of AL, with a 74% accurate classification based on ROC analysis (AUC of 0.74; 95% CI = 0.66 - 0.82; P < .001) of the performance of the scoring model.

Conclusions: We demonstrate the value of MFC for fast diagnostic screening of PC clonality in AL and simultaneous automated patient risk-stratification, based on the BM tumor burden and PC phenotype. In addition, our results also provide new immunophenotypic markers for the identification of patients with monoclonal gammopathies that are candidates for monitoring of pre-symptomatic organ damage related to AL.