Background Diffuse Large B-cell Lymphoma (DLBCL) is the most common hematologic malignancy, originating either from the germinal-center (GCB) or post-germinal cells (ABC). Despite a ~60% response rate to first-line therapy, DLBCL remains incurable for ~40% of patients. Currently, risk stratification includes the International Prognostic Index, PET-CT scans, Cell-Of-Origin (COO), and expression/translocation of BCL2/BCL6/c-MYC. Nevertheless, no molecular marker has been widely correlated with the risk of treatment failure. Objectives We aim to provide a study of the existing prognostic tools for DLBCL and to evaluate potentially key players of the genomic instability and molecular complexity related to prognosis. Methods The project comprises a well characterized DLBCL cohort (204 patients), uniformly treated and with clinical data. COO classification has been analyzed through immunohistochemistry (IHC - Hans’ algorithm) and Lymph2Cx assay (Nanostring). BCL2/MYC expression has been evaluated through IHC. RNAseq has been performed using a total RNA approach and sequenced on a NovaSeq 6000 (ILLUMINA). Expression analysis was based on “Ecotyper algorithms”. Progression free survival (PFS) has been estimated by the Kaplan–Meier method. Results Hans’ algorithm on 84 patients identified 41% GCB, and 59% ABC cases. Among 191 patients, Lymph2Cx discovered 47% GCB, 35% ABC, and 18% unclassified cases (concordance between methods=0.719 k-statistic). In keeping with literature data, GCB patients showed significantly longer PFS compared to ABC subset both using Hans’ algorithm (p=0.011) and Lymph2cx (p=0.027). Unclassified patients showed an intermediate PFS between GCB and ABC. Analysis of BCL2 and MYC expression in 201 cases showed 57% of BCL2+/c-MYC-, 2.6% BCL2-/c-MYC+, and 12% BCL2+/c-MYC+ cases. Cox regression analysis showed a 5-year PFS of 92% for BCL2-/c-MYC-, 62% for BCL2+/c-MYC+ and 64% for BCL2+/c-MYC-. These data support the idea that BCL2 overexpression could be responsible for poor prognosis of DLBCL patients. RNASeq on 186 cases correctly recapitulated the total gene expression profile showing superimposable results with Lymph2cx classification. We then applied EcoTyper, a machine-learning tool able to derive 44 cell states related to malignant B cells and other cell types in the DLBCL tumor microenvironment (TME). Among the most important cell states for DLBCL, B cells showed 5 cell states (S1-S5). In our cohort, S1 showed a significantly longer PFS respect to the other cell states, with S5 cell state displaying the worse prognosis. The difference in term of survival outcome appeared even more evident when B cell states were stratified for the COO, showing how S1 and S5 tend to be mutually exclusive with the S1 cell state displaying a GCB-like signature and S5 cell state an ABC-like signature. Ecotyper assembled the cell states into communities generating 9 different Lymphoma Ecotypes (LEs) subgroups. Again, PFS analysis showed significant differences between LEs (p=0.028) with the maximum distance between LE1 and LE9, the latter showing a better outcome. In keeping, ABC cases are more frequently associated with LE1 and LE2 due to the high B cell infiltration, whereas GCB cases are more related to Ecotypes with stronger TME activity. Conclusions Here we confirm in a real-life cohort of DLBCL patients the prognostic significance of COO classification and BCL2 expression. In addition, RNASeq data demonstrated to be independently informative for DLBCL patient’s outcome. These observation lead to the possibility to improve the risk stratification of DLBCL patients taking advantage of new bioinformatic tools selecting cases that might be potentially eligible for different therapy approaches.

Introduzione Il Linfoma diffuso a grandi cellule B (DLBCL) è il tipo di linfoma maggiormente diagnosticato. Sono noti due sottogruppi della Cell of Origin: germinal center-like (GCB) e activated-like (ABC). Il tasso di risposta alla terapia con R-CHOP è circa 60%. Attualmente, la stratificazione del rischio include l'International Prognostic Score, PET-CT, la classificazione COO e l'espressione/traslocazione di BCL2/BCL6/MYC. Nessun marcatore molecolare è stato correlato al rischio di fallimento del trattamento. Obiettivi Lo studio si propone di fornire un’analisi completa dei classici fattori prognostici e di valutare nuovi potenziali protagonisti della fitta complessità molecolare e instabilità genomica. Metodi Il progetto comprende una coorte di DLBCL ben caratterizzata a livello clinico di 204 pazienti trattati uniformemente con R-CHOP. La COO è stata valutata attraverso immunoistochimica (IHC - algoritmo di Hans) e il pannello Lymph2Cx (Nanostring). L'espressione di BCL2/MYC è stata valutata tramite IHC. RNAseq è stato eseguito utilizzando un approccio total-RNA e sequenziato su piattaforma NovaSeq 6000 (ILLUMINA). L'analisi dell'espressione è stata condotta utilizzando algoritmi Ecotyper pubblicati di recente. La Progression Free Survival (PFS) è stata stimata con il metodo Kaplan-Meier. Risultati Algoritmo di Hans: 41% dei casi come GCB e 59% come ABC. Lymph2Cx: 47% come GCB, 35% come ABC, 18% come Unclassified (concordanza=0,719). I casi GCB mostravano PFS significativamente più lunga rispetto al sottogruppo ABC sia utilizzando IHC (p=0,011) che Lymph2cx (p=0,027). I casi BCL2+ erano 67%, mentre quelli c-MYC+ 15%. I casi Double Expressor erano 12%. L'analisi di regressione Cox ha mostrato PFS a 5 anni dell'87% per BCL2-/c-MYC-, del 60% per BCL2+/c-MYC+ e del 45% per BCL2+/c-MYC-. 186 campioni sono stati amplificati e sequenziati con successo. RNASeq è stato in grado di ricapitolare correttamente il profilo di espressione genica totale mostrando risultati sovrapponibili con Lymph2cx. L'analisi attraverso Ecotyper ha consentito una deconvoluzione di 13 tipi cellulari (cellule B + 12 cellule TME) indicando diversi programmi trascrizionali, noti come cell states, per ciascun tipo di cellula. I linfociti B mostravano 5 stati cellulari (S1-S5), dove S1 rappresenta una firma del centro germinale e S5 una firma correlata alle plasmacellule. La PFS era significativamente differente tra gli stati delle cellule B (p=0,0059), con prognosi peggiore per i casi S5. I cell states di diversi tipi cellulari sembrano coesistere, pertanto, attraverso l'analisi di clustering è stato possibile notare come più cell states siano assemblati in comunità che massimizzano i modelli di co-associazione, formando 9 diversi ecotipi (LE). I casi ABC sono più frequentemente associati a LE1 e 2 dove l'infiltrazione dei linfociti B è più evidente, mentre i GCB sono correlati agli ecotipi con attività del microambiente più forte. La PFS è risultata significativamente differente tra i vari LE (p=0,028) con distanza massima tra LE1 e LE9, quest'ultimo avente esito migliore. Conclusioni Lo studio mostra quindi il significato prognostico della COO e dell'espressione BCL2 nel contesto real-life. La over espressione di BCL2, rispetto a c-MYC, può essere responsabile della prognosi infausta dei casi di Double Expressor. Inoltre, i dati di RNA-seq hanno dimostrato di essere estremamente informativi per l'esito dei pazienti con DLBCL. Queste osservazioni portano alla possibilità di migliorare la stratificazione del rischio dei pazienti con DLBCL sfruttando nuovi strumenti bioinformatici in grado di osservare nuovi pattern molecolari all'interno dei sottogruppi COO, indicando casi che potrebbero essere potenzialmente eleggibili per diversi approcci terapeutici.

Caratterizzazione clinica e molecolare del Linfoma Diffuso a Grandi cellule B: studio del trascrittoma e del microambiente tumorale per l’identificazione di nuovi sottotipi molecolari / Robel Papotti , 2023 May 19. 35. ciclo, Anno Accademico 2021/2022.

Caratterizzazione clinica e molecolare del Linfoma Diffuso a Grandi cellule B: studio del trascrittoma e del microambiente tumorale per l’identificazione di nuovi sottotipi molecolari

PAPOTTI, ROBEL
2023

Abstract

Background Diffuse Large B-cell Lymphoma (DLBCL) is the most common hematologic malignancy, originating either from the germinal-center (GCB) or post-germinal cells (ABC). Despite a ~60% response rate to first-line therapy, DLBCL remains incurable for ~40% of patients. Currently, risk stratification includes the International Prognostic Index, PET-CT scans, Cell-Of-Origin (COO), and expression/translocation of BCL2/BCL6/c-MYC. Nevertheless, no molecular marker has been widely correlated with the risk of treatment failure. Objectives We aim to provide a study of the existing prognostic tools for DLBCL and to evaluate potentially key players of the genomic instability and molecular complexity related to prognosis. Methods The project comprises a well characterized DLBCL cohort (204 patients), uniformly treated and with clinical data. COO classification has been analyzed through immunohistochemistry (IHC - Hans’ algorithm) and Lymph2Cx assay (Nanostring). BCL2/MYC expression has been evaluated through IHC. RNAseq has been performed using a total RNA approach and sequenced on a NovaSeq 6000 (ILLUMINA). Expression analysis was based on “Ecotyper algorithms”. Progression free survival (PFS) has been estimated by the Kaplan–Meier method. Results Hans’ algorithm on 84 patients identified 41% GCB, and 59% ABC cases. Among 191 patients, Lymph2Cx discovered 47% GCB, 35% ABC, and 18% unclassified cases (concordance between methods=0.719 k-statistic). In keeping with literature data, GCB patients showed significantly longer PFS compared to ABC subset both using Hans’ algorithm (p=0.011) and Lymph2cx (p=0.027). Unclassified patients showed an intermediate PFS between GCB and ABC. Analysis of BCL2 and MYC expression in 201 cases showed 57% of BCL2+/c-MYC-, 2.6% BCL2-/c-MYC+, and 12% BCL2+/c-MYC+ cases. Cox regression analysis showed a 5-year PFS of 92% for BCL2-/c-MYC-, 62% for BCL2+/c-MYC+ and 64% for BCL2+/c-MYC-. These data support the idea that BCL2 overexpression could be responsible for poor prognosis of DLBCL patients. RNASeq on 186 cases correctly recapitulated the total gene expression profile showing superimposable results with Lymph2cx classification. We then applied EcoTyper, a machine-learning tool able to derive 44 cell states related to malignant B cells and other cell types in the DLBCL tumor microenvironment (TME). Among the most important cell states for DLBCL, B cells showed 5 cell states (S1-S5). In our cohort, S1 showed a significantly longer PFS respect to the other cell states, with S5 cell state displaying the worse prognosis. The difference in term of survival outcome appeared even more evident when B cell states were stratified for the COO, showing how S1 and S5 tend to be mutually exclusive with the S1 cell state displaying a GCB-like signature and S5 cell state an ABC-like signature. Ecotyper assembled the cell states into communities generating 9 different Lymphoma Ecotypes (LEs) subgroups. Again, PFS analysis showed significant differences between LEs (p=0.028) with the maximum distance between LE1 and LE9, the latter showing a better outcome. In keeping, ABC cases are more frequently associated with LE1 and LE2 due to the high B cell infiltration, whereas GCB cases are more related to Ecotypes with stronger TME activity. Conclusions Here we confirm in a real-life cohort of DLBCL patients the prognostic significance of COO classification and BCL2 expression. In addition, RNASeq data demonstrated to be independently informative for DLBCL patient’s outcome. These observation lead to the possibility to improve the risk stratification of DLBCL patients taking advantage of new bioinformatic tools selecting cases that might be potentially eligible for different therapy approaches.
Clinical and molecular characterization of Diffuse Large B-cell Lymphoma: identification of novel molecular subtypes based on RNA transcriptome and its tumor microenvironment
19-mag-2023
POZZI, Samantha
File in questo prodotto:
File Dimensione Formato  
Thesis PhD Robel Papotti reviewed.pdf

embargo fino al 18/05/2026

Descrizione: Tesi definitiva Papotti Robel
Tipologia: Tesi di dottorato
Dimensione 2.84 MB
Formato Adobe PDF
2.84 MB Adobe PDF   Visualizza/Apri   Richiedi una copia
Pubblicazioni consigliate

Licenza Creative Commons
I metadati presenti in IRIS UNIMORE sono rilasciati con licenza Creative Commons CC0 1.0 Universal, mentre i file delle pubblicazioni sono rilasciati con licenza Attribuzione 4.0 Internazionale (CC BY 4.0), salvo diversa indicazione.
In caso di violazione di copyright, contattare Supporto Iris

Utilizza questo identificativo per citare o creare un link a questo documento: https://hdl.handle.net/11380/1305582
Citazioni
  • ???jsp.display-item.citation.pmc??? ND
  • Scopus ND
  • ???jsp.display-item.citation.isi??? ND
social impact