Tumor evolution and immune microenvironment dynamics in primary and relapsed mantle cell lymphoma

Abstract

Mantle cell lymphoma (MCL) is a rare but often aggressive type of B cell lymphoma with a high risk of relapse. To explore intratumoral clonal diversity and tumor evolution related to disease relapse, we integrate single-cell RNA and B cell receptor sequencing with whole-genome sequencing in 20 diagnosed/untreated and/or relapsed samples from 11 MCL patients. Our results reveal significant intratumor heterogeneity in MCL already at diagnosis. We further show that the evolutionary paths during disease progression for each patient are unique, where minor clones present at diagnosis may acquire different mutations and copy-number variations and/or migrate to various microenvironments. Despite significant interpatient heterogeneity, recurrent genetic and transcriptomic changes in tumor cells affecting key signaling pathways, along with alterations involved in the tumor microenvironment, are also observed during disease progression. Taken together, our findings elucidate the diverse and dynamic tumor-immune evolution processes associated with disease progression and relapse in MCL.

Keywords: clonal evolution; immunotherapy; mantle cell lymphoma; relapse; scRNA-seq; single-cell RNA sequencing; tumor heterogeneity; tumor microenvironment.

Graphical abstract

Figure 1

Single-cell atlas of MCL (A) Overview of the patient cohort with a schematic of treatment and time for sample collection for scRNA-seq. (B) UMAPs of sequenced single cells. The cells are colored by diagnosis, donor, sampling time, tissue, and major cell type. (C) Malignant cells were inferred based on clonal BCR, a consistent IGK/IGL ratio, and CCND1 expression. (D) Heatmap of the inferred copy-number variations (CNVs) from patients’ B cells (middle) across 22 chromosomes compared with B cells from normal controls (top). The color represents the CNV gain (red) and loss (blue). (E) Fractions of cells from major cell types for each sample. The arrows represent the sampling timeline for the patients. See also Figure S1 and Tables S1 and S4.

Figure 2

Inter- and intrapatient heterogeneity of MCL (A) Uniform manifold approximation and projection (UMAPs) of B cells sequenced from all MCL patients, colored by malignant cell type, subcluster, patient, and sample. Tumor cells in the same cluster but from different tumor samples of the same patient are highlighted by circles. (B) Highly variable genes of malignant cells between tumor samples. The selected lymphoma-related genes are labeled. (C) Identification of common cellular programs in malignant cells. The rows and columns are mirrored and represent the expression programs. The color represents the number of top genes shared between programs. The annotation of meta-programs identified from hierarchical clustering is shown on the left. (D) Changes in the average gene expression levels in the meta-programs between paired samples from the same patient. The p value was calculated from Wilcoxon signed-rank test. (E) Kaplan-Meier survival analysis of NF-κB and MYC pathway activity in a public MCL cohort (GEO: GSE93291). The NF-κB pathway gene set was derived from BioCarta, and the MYC pathway gene set was sourced from the Hallmark collection. The p value was calculated via the log rank test. (F) Nonsilent somatic gene mutations detected by whole-genome sequencing were either present in at least two patients or previously reported in other lymphoma studies. The color indicates the mutation type. Structural variants in CCND1 represent IGH-CCND1 translocations. Meta-P, meta-program. See also Figure S2 and Tables S2, S3, S4, and S5.

Figure 3

Tumor microenvironment in MCL (A) UMAPs of T and NK cell subtypes. (B) UMAPs of myeloid cell subtypes. (C) TME composition of each sample. The color is the scaled proportion in each sample. (D) Ligand (L)-receptor (R) interactions between malignant cells and TME cells in MCL patients. L-R pairs are labeled on the left and colored according to their costimulation or coinhibition features. (E) Interactions between malignant cells and TME cells at different sampling times. p values for inferred L-R interactions (D and E) were calculated by CellPhoneDB using permutation testing. See also Figures S3 and S4.

Figure 4

Intratumor heterogeneity and tumor evolution during disease progression (A) UMAPs of tumor cells in each patient with available paired samples (upper) and trajectory analysis between different sampling times (lower). The similarity index (SI) between paired samples is shown in the subtitle. (B) Heatmap showing the scaled average expression of MCL-related key genes in malignant cells at the intrasample subcluster level. Clusters are first grouped by donor, followed by samples. (C) CNV scores of the transcriptional clusters in each sample. CNV scores were calculated by quantifying the number of genes with CNVs inferred from the scRNA-seq data, using a threshold of less than 0.95 or greater than 1.05. Each violin represents the distribution of CNV scores for the respective groups, with a line indicating the median. The color represents different samples.

Figure 5

Intratumor heterogeneity and tumor evolution from primary to relapsed tumors in MC3 (A) Alignment of IGHV nucleotide sequences identified from the clonal BCR. The unique clone sequences were named by the SHM number compared with the germline sequence IGHV4-34 derived from the international ImMunoGeneTics information system (IMGT) database. (B) The cell numbers of three unique clonotypes in each sample according to unfiltered scBCR-seq data. The cell number is indicated by the percentage of cells in each sample. (C) UMAPs of malignant cells colored by sample, transcriptional subclusters, and BCR clonotypes. (D) The composition and CNV features of transcriptional subclusters were identified in each sample. The left pie chart shows the composition of subclusters, and the background of the subcluster is colored according to the BCR clonotype. The right panel shows inferred CNVs (middle) and matched bulk WGS CNVs (top). Red represents copy-number gain, and blue represents copy-number loss. The distinct CNV features are highlighted for each subcluster. (E) Fish plot showing patterns of tumor clone evolution inferred based on somatic mutations according to WGS data. (F) Inferred clonal evolution in MC3. One dot represents one subcluster, its size reflects the relative proportion in each stage, and its color represents the BCR clonotype. See also Figures S5–S9 for clonal evolution analysis in other patients.

Figure 6

High CD70 expression and strong CD70-CD27 interaction in MC3 relapse tumors (A) CD70-CD27 interaction between malignant subclusters and T subsets. (B) CD70 expression levels across MC3 malignant subclusters. Each box represents the interquartile range (IQR) with a line indicating the median. Whiskers extend to the minimum and maximum values within 1.5 times the IQR, excluding outliers. (C) UMAP of sequenced single cells in MC3 samples colored by cell type and CD70 and CD27 expression. (D) Immunohistochemistry images of MC3 primary and relapsed tumors stained with an anti-CD70 antibody. Scale bars, 100 μm (top left micrograph), 2 mm (top right micrograph), 100 μm (bottom zoomed-in micrographs). (E) Multiplex immunofluorescence staining of the MC3Ri sample showing high CD70 expression in malignant cells (row 1: CD20, CCND1, and CD70) and the presence of CD4⁺FOXP3⁺ and CD8⁺PD-1⁺ cells in the tumor microenvironment. Scale bars, 2 mm (overview micrographs) and 50 μm (right zoomed-in micrographs). See also Figure S4.