Extracellular Vesicles in Diffuse Large B Cell Lymphoma: Characterization and Diagnostic Potential

Abstract

Diffuse large B cell lymphoma (DLBCL) is an aggressive B cell lymphoma characterized by a heterogeneous behavior and in need of more accurate biological characterization monitoring and prognostic tools. Extracellular vesicles are secreted by all cell types and are currently established to some extent as representatives of the cell of origin. The present study characterized and evaluated the diagnostic and prognostic potential of plasma extracellular vesicles (EVs) proteome in DLBCL by using state-of-the-art mass spectrometry. The EV proteome is strongly affected by DLBCL status, with multiple proteins uniquely identified in the plasma of DLBCL. A proof-of-concept classifier resulted in highly accurate classification with a sensitivity and specificity of 1 when tested on the holdout test data set. On the other hand, no proteins were identified to correlate with non-germinal center B-cell like (non-GCB) or GCB subtypes to a significant degree after correction for multiple testing. However, functional analysis suggested that antigen binding is regulated when comparing non-GCB and GCB. Survival analysis based on protein quantitative values and clinical parameters identified multiple EV proteins as significantly correlated to survival. In conclusion, the plasma extracellular vesicle proteome identifies DLBCL cancer patients from healthy donors and contains potential EV protein markers for prediction of survival.

Keywords: cancer; diagnostic; diffuse large B cell lymphoma; extracellular vesicles; liquid biopsy; plasma.

Figure A1

Summary of functional categories that are either enriched in HD or DLBCL based on functional enrichment in all identified protein from DLBCL and HD when compared with the cellular component.

Figure A2

Summary of functional categories that are either enriched in HD or DLBCL based on functional enrichment in all identified protein from DLBCL and HD when compared with molecular function.

Figure A3

Summary of functional categories that are either enriched in HD or DLBCL based on functional enrichment in all identified protein from DLBCL and HD when compared with the biological process.

Figure A4

Boxplot of total number of gene collapsed proteins identified from 50 random subsamples of 10 DLBCL and 10 HD samples.

Figure A5

Principal component analysis of all quantitative values obtained from (a) batch 1 and (b) batch 2.

Figure A6

Analysis of overlapping proteins between enriched KEGG pathways for DLBCL. The color coding of the nodes indicates the number of proteins regulated for each category. The size of the nodes reflects −log₁₀ of enrichment p-value. The thickness of the edges reflects the number of shared regulated proteins between the KEGG pathways (ranging from one to 31 proteins).

Figure 1

Biophysical characterization by nanoparticle tracking analysis of plasma EV samples from HD and DLBCL groups. Particle size distribution in representative samples of HD and DLBCL groups ((a,b), respectively). Distribution of estimated particle concentration and particle size mode between DLBCL patients and HDs ((c,d), respectively). * means significance.

Figure 2

MS-based quantitative comparison between enriched plasma EVs, BAL EVs and cell line isolated EVs of frequently reported exosome protein markers and the ten most abundant exosome markers from a publicly available EV database (ExoCarta). Blue indicates generally accepted exosome markers. Green indicates micro vesicle markers. Red-labeled proteins indicate non-EV proteins.

Figure 3

Overview of identifications in DLBCL patients and HDs. (a) Total number of identifications from DLBCL patients and HDs. (b) Venn diagram comparing identifications in DLBCL patients and HDs. (c) Summary of functional KEGG pathway analysis that are either enriched in HDs or DLBCL patients based on all identified proteins from DLBCL and HD groups.

Figure 4

Protein richness in DLBCL and HD groups. (a) Protein richness curves for DLBCL (red boxes) and HD group (green boxes) when all protein isoforms are used. (b) Protein richness curves for DLBCL (red boxes) and HD group (green boxes) when gene-collapsed proteins were used. The lower tables indicate the estimated number of proteins in the two sample groups using different estimation methods and the corresponding standard error.

Figure 5

Volcano plot depicting deregulated proteins when comparing (a) DLBCL versus HD and (b) non-GCB versus GCB. A p-value threshold of 0.05 is indicated by a horizontal red dotted line. The log₂ fold up or down regulation is represented by the red vertical line. Potential contaminants from erythrocyte, platelet and coagulation are indicated by red, orange and blue dots, respectively.

Figure 6

Functional enrichment analysis of differentially expressed proteins between DLBCL and HD and non-GCB and GCB. Analysis of (a) KEGG pathway, (b) biological process, (c) molecular function and (d) cellular component gene categories.

Figure 7

Performance of partial least squares regression analysis. (a) A summary of the training and test data sets used. (b) Receiver operating characteristics for the PLS regression model. The blue dot indicates the threshold based on Youden’s J statistic. (c) The top ten protein features in terms of importance. (d) The confusion matrix and the performance parameters associated with it.

Figure 8

Kaplan–Meier plots based on dividing high and low expression of (a) IGLC1, (b) IGLL5, (c) PSMB2 and (d) CORO1a.