Tracing malignant precursor mutations and their cellular origins in blood and bone marrow of primary central nervous system lymphomas by ultrasensitive genomic profiling

Abstract Introduction Primary central nervous system lymphomas (PCNSL) are aggressive extranodal non-Hodgkin lymphomas confined to the CNS, accounting for 5–7% of primary brain tumors. PCNSL are biologically unique among brain malignancies due to their extracerebral origin and their specific migration and proliferation in the CNS compartment. Somatic mutations in genes involved in the B-cell receptor and Toll-like receptor signaling pathways are frequently found in PCNSL, suggesting a common genetic mechanism underlying CNS tropism. However, the developmental stage at which these genetic aberrations arise and their phylogenetic hierarchies remain unknown. Here, we applied ultrasensitive targeted capture sequencing to directly characterize tumor-specific mutational profiles in circulating and bone marrow- (BM-) resident cells to map the cellular hierarchies of lymphoma-specific genetic alterations and elucidate potential processes involved in PCNSL tropism. Methods We collected peripheral blood (PB) and BM mononuclear cell (PBMC/BMMC) samples at baseline from nine patients with PCNSL, all of whom showed no evidence of systemic lymphoma involvement by conventional imaging and PB/BM assessment. PBMCs and BMMCs were FACS-sorted into purified B-cell and T-cell populations. We then performed targeted capture next-generation sequencing (CAPP-Seq), covering immunoglobulin (Ig) regions and 100 additional genes recurrently mutated in PCNSL, to profile the mutational landscape in all nine tumor samples. These somatic alterations were tracked in bulk PBMC and BMMC samples (n=18) as well as the corresponding sorted B-cell and T-cell subsets (n=36). To control for sequencing errors and to evaluate the specificity of the monitoring approach, the same sorting and sequencing strategy was applied to PBMC/BMMC samples from four patients in complete remission one year after allogeneic stem cell transplantation for myeloid neoplasia. Results We detected a median of 256 somatic mutations in PCNSL tumor samples (range: 24-327), with PIM1 (78%), MYD88 (67%), and CD79B (67%)being the most frequently mutated genes. At a predefined specificity threshold of 95%, we identified tumor-specific alterations in 7 of 9 PCNSL patients (78%) by ultrasensitive tracking of mutations in PB or BM samples. While tumor mutations were detectable in 2 of 9 bulk PBMC and 4 of 9 bulk BMMC samples (median allele frequency [AF]: 0.003%), detection rates and allelic fractions were substantially higher in purified B-cell subsets, with 7 of 9 PB B-cell (78%) and 6 of 9 BM B-cell samples (67%) yielding positive monitoring results with a median AF of 0.03%. Notably, no PCNSL-specific somatic variants were found in any T-cell subset from either compartment. Furthermore, tumor-specific clonal VDJ rearrangements were observed in most PB B-cell samples (5/9, 56%), but only in 2 of 9 BM B-cell samples and in none of the bulk specimens or T-cell populations. Next, we focused on the PB B-cell compartment to delineate PCNSL-specific genetic patterns in circulating cells. The majority of detected variants were located in immunoglobulin regions (median of 81%) and corresponded to subclonal events in matched tumor biopsies. However, we also identified various pathogenic somatic mutations in key driver genes, including MYD88 (n=2), CD79B (n=1), and TBL1XR1 (n=1), all of which represented truncal variants in corresponding tumors. Of note, the most frequently observed aberrations involved the 5′UTR region of DTX1 (5/9 cases, 56%), a gene implicated in extranodal dissemination of germinal center (GC) B-cells. Conclusion Our findings reveal the presence of circulating and BM-resident malignant or premalignant cells in the majority of PCNSL patients, and the detection of mutations in driver genes commonly associated with PCNS lymphomagenesis. Tumor-specific mutations were exclusively identified in the B-cell compartment, while T-cell subsets remained unaffected. This absence of somatic alterations in T-cell populations together with an accumulation of Ig mutations and the presence of clonal VDJ rearrangements in B cells suggests that (pre-)malignant clones originate outside the CNS at the GC-experienced B-cell stage, without involvement of earlier hematopoietic precursors.