High IGHV Intraclonal Diversification Identifies CLL with Good Prognosis, Features of Anergy and Specific T-Cell Profiles

Introduction

The immunoglobulin heavy chain variable (IGHV) mutational status, identifying mutated (M) vs. unmutated (U) cases, is a relevant prognostic biomarker in chronic lymphocytic leukemia (CLL). In this context, CLL may exhibit ongoing clonal evolution in IGHV genes, (a.k.a. intraclonal diversification, ID), although a rigorous workflow for ID assessment and quantification in the NGS era is still missing.

Aim

1. i) To develop a bioinformatic workflow for ID quantification in CLL; ii) To evaluate the clinical impact of ID in CLL; iii) to investigate the biology of CLL cells with ID.

Methods

The study included 983 CLL patients (759 with Time-to-first treatment, TTFT) analyzed at first presentation. A UMI-independent Lymphotrack-based procedure was used for IGHV repertoire libraries exploiting an original bioinformatic pipeline generated to correct systematic sequencing errors (patent request #102022000027138, 12/29/2022). The inverse Simpson Index (iSI) was used as diversity measure to evaluate ID (PMID:25100741). Maximally selected log-rank statistics, used along with TTFT data, identifies the 1.2 iSI cutoff to discriminate ID-high (≥1.2 iSI) vs. ID-low (<1.2 iSI) samples. ID cutoff was validated on 127 cases affected by lymphoproliferative disorders (LPD) with different cell of origin. The majority of LPD characterized by frank ID features, i.e. HCL, DLBCL, and FL, were classified ID-high while MCL usually lacking somatic hypermutation features, turned out ID-low. Tissue-AdaPtive autoEncoder (TAPE; PMID:36347853) was applied to extract cell-specific RNASeq profiles from bulk RNASeq from 18 unsorted M-CLL samples. Deconvoluted RNASeq from CLL and T cell fractions were evaluated for differential gene expression and pathways enrichment with gsea/gsva/mlm functions of decoupler package. Enrichment in T cell phenotypes was assessed using pan-cancer T cell expression profiles (PMID:34914499). Multivariate analyses (MVA) were performed with the MedCalc software.

Results

Among 983 CLL (508 M-CLL, 475 U-CLL), we identified 144 ID-high (≥1.2 iSI) and 839 ID-low CLL (<1.2 iSI), with a significant overrepresentation of ID-high cases in M- vs. U-CLL (P=0.002). Focusing on M-CLL only, ID-low patients had significant shorter TTFT than ID-high patients (P=0.015). Consistently, a MVA analysis on 299 demonstrated that the ID-low phenotype was an independent predictor of shorter TTFT (HR=2.12, 95%CI 1.13-3.99, P=0.019) in M-CLL along with Rai Stage, CD49d and del11p/del17p. Bulk RNASeq on M-CLL (9 ID-low, and 9 ID-high) were deconvoluted with the TAPE autoencoder using a custom signature matrix extracted from published single-cell(sc)RNA data (GSE161610) and validated by bulk cell-type specific marker genes; estimated cell-type fraction percentages values estimated from deconvoluted RNAseq overlapped flow cytometry data. Distinct gene expression profiles between ID-high and ID-low were observed both in the CLL and T-cell components. ID-high M-CLL cells displayed features of CLL anergy, as defined by: i) significant up-regulation of the ERK, STAT3, NFAT, Ca++ signaling pathways; ii) significantly lower IGHV RNA levels; down-regulation of genesets associated to metabolism and survival was in keeping with the indolent clinical behavior of ID-high M-CLL. Analyses of the CD8 and CD4 T-cell components identified as up-regulated in ID-high M-CLL the RAC1, RHOA, CDC42 pathways associated with immunologic synapse activity. Using a multivariate linear model evaluating the phenotypic enrichment of deconvoluted T cell fractions based on pan-cancer T cell expression profiles, we observed, in ID-high M-CLL, CD8/CD4 expression profiles compatible with a significant expansion of specific memory and effector cells, including IL17-producing or Granzyme K-producing T cells, while other T cell subsets (e.g. NME1-expressing) were down-regulated. Consistently, analysis of TCRB showed higher heterogeneity in the T cell repertoire of ID-high respect to ID-low M-CLL (P=0.0037).

Conclusions

Here we were able to develop and validate a solid NGS protocol to quantitatively evaluate ID in CLL, demonstrating that: i) analysis of ID is feasible by a large-scale UMI-independent approach; ii) high degree of ID is clinically relevant in M-CLL identifying a subset with better outcome; iii) the ID-high M-CLL phenotype is associated with anergy markers and a functionally efficient T cell repertoire.