Absence of driver mutations in persistent polyclonal B-cell lymphocytosis with binucleated lymphocytes

To the editor:

Most hematologic malignancies are preceded by a premalignant condition. In myeloid neoplasms, genetically defined premalignant conditions such as clonal hematopoiesis of indeterminate potential¹  reflect the clonal dominance of hematopoietic stem cells harboring oncogenic mutations (such as TET2 or DNMT3A).²  In contrast, non–genetically driven premalignant conditions have been described in lymphoid malignancies. For example, autoimmunity or chronic infections have been causally associated with marginal zone lymphomas^3-5  and can be seen as phenotypically defined premalignant conditions.

Persistent polyclonal B-cell lymphocytosis (PPBL) with binucleated lymphocytes is a rare and indolent condition that evolves to overt lymphoma in around 10% of patients.⁶  Although the B cells are polytypic, they harbor recurrent clonal abnormalities at the cytogenetic level (+isochromosome 3q and/or +3).⁷  Moreover, as these cytogenetic clonal markers have been observed in both the κ- and λ-expressing B lymphocytes in several individual patients,⁸  it raises the hypothesis that PPBL results from an acquired genetic driver event in pre-B cells before the rearrangement of the immunoglobulin locus. Recently, congenital polytypic expansion of B lymphocytes has been linked to a hereditary mutation of CARD11,^9,10  suggesting that mutations in oncogenes involved in lymphomagenesis might recapitulate some aspects of PPBL.

To investigate whether acquired mutations can drive PPBL pathogenesis, we performed whole exome sequencing (WES) of sorted B lymphocytes from 15 PPBL patients with available dimethyl sulfoxide frozen blood samples. The diagnosis of PPBL required the persistence for at least 3 months of typical binucleated lymphocytes on blood smear analysis, and the confirmation of their polytypic phenotype by flow cytometry. Cytogenetic analysis was performed in all patients (supplemental Table 3 available on the Blood Web site). The demographic characteristics were as expected (mainly female, mean age of 44 years, tobacco exposure for all patients), as were the biological results (moderate hyperlymphocytosis with a small proportion of binucleated lymphocytes, increased polyclonal immunoglobulin M serum levels, and overrepresentation of HLA-DR7; Table 1). Of note, the rate of evolution to overt lymphoma during follow-up was within the previously described range (1 [7%] of 15).

B lymphocytes were purified by CD20 cell sorting (RoboSep, STEMCELL Technologies, Grenoble, France), yielding very high purity (86% to 98%). The CD20^– fraction was used as constitutional material for all patients except for 2 with nonhematologic genetic material available (Unique Patient Number 3 [UPN3] and UPN18; fibroblasts and hair follicles, respectively). Exome sequences were analyzed by comparing the CD20⁺ PPBL fractions with their normal counterpart. For the PPBL fractions, the mean depth of coverage was more than 200×, with 98% to 99% of the targets covered by more than 30 independent sequencing reads. A detailed description of the exome sequencing method and computational analysis is provided in the supplemental Data.

First we compared copy number variation (CNV) analysis performed by conventional karyotyping of cultured cells to CNV analysis derived from WES read depth ratios (supplemental Data; supplemental Table 1). Half (9 of 18) of the abnormalities identified by karyotyping were also called using WES data, whereas the other half consisted of more subclonal events that did not reach depth ratio calling thresholds (supplemental Figure 1). Conversely, 1 monosomy of chromosome 21 was observed only in the WES profile, which suggests possible negative selection during prekaryotyping cell culture. Of note, we did not identify cryptic copy number alterations, which is coherent with a recent report of single nucleotide polymorphism array analysis of 10 PPBL patients.¹¹  More specifically, no cryptic amplification of the MECOM locus (3q26.2) was observed in this cohort. Estimation of the clonality of the CNVs showed that these were present in up to 53% of the sequenced cells (supplemental Table 1), demonstrating that cell sorting efficiently purified pathological cells. Thus, if early mutational events were driving PPBL, we postulate that they would be present in a majority of the pathological cells (ie, with large variant allele frequencies [VAFs]) in our experiment.

Variants were called by using parameters that allowed the detection of very low VAFs followed by rigorous filtering (supplemental Data). As shown in Table 2, the overall number of somatically acquired coding mutations was very low across the 15 patients, and none were recurrent. Twelve patients had no coding mutation, and the mutational burden was very low in the 3 remaining patients (2 patients had 1 mutation, and 1 patient had 2 mutations). The 4 mutations were single nucleotide variants with very low VAFs (2.2% to 5.1%) affecting exons of OR4C11, RNF40, and MYD88, and a splice site of TET2. The mutation of MYD88 is of interest in the setting of B-lymphocyte malignancies. However, we can rule out a driver function in the lymphoproliferation of this patient given its very low VAF suggesting a subclone of less than 5% mutated cells, whereas CNV analysis showed that 23% of the cells harbored the isochromosome 3q. Regarding the TET2 mutation identified in UPN9, it was absent from the myeloid blood cells used as control, which rules out a clonal hematopoiesis of indeterminate potential. Again, the low VAF of this mutation is a strong argument against its role as a driver of PPBL pathophysiology. Interestingly, another study of 4 patients failed to identify exonic mutations in a panel of 12 nuclear factor kappa B pathway genes.¹²  We conclude that PPBL is not driven by protein-coding mutations. However, we cannot exclude a role for mutations outside coding sequences or modifications of the epigenetic state, but these hypothesis were not evaluated in this study.

We then focused our attention on UPN3 who developed a diffuse large B-cell lymphoma (DLBCL) 11 years after initial diagnosis of PPBL. CNV analyses were coherent with the cytogenetic results, showing gain of 3q chromosome and loss of chromosome 21 in the PPBL sample, and gain of the whole chromosomes 3 and X in the DLBCL sample (supplemental Figure 2). The DLBCL sample harbored 11 coding somatic mutations, including a stopgain single nucleotide variant in the well-known lymphoma driver KMT2D (Table 2). The VAFs of these mutations were indicative of high clonality (mean VAF, 24.7%; range, 7.7% to 29.8%; Table 2). Of note, none of these variants were called in the PPBL sample, although careful examination of the raw data found individual supporting reads for 3 of these (KMT2D, LNX1, and LAMA2) at very low VAFs (<1%) making them indistinguishable from sequencing noise (Table 2). Previous studies of the immunoglobulin heavy chain locus rearrangement have suggested that a founding clone was shared between the 2 lymphoproliferations.⁶  The discrepancies between the DLBCL and PPBL samples at the cytogenetic and molecular levels indicate that if such a clonal relationship does exist between the 2, the lymphoma probably originates from a minor subclone predating PPBL genomic alterations. This might be further supported by the finding that traces of some of the variants called in the DLBCL could be seen in the PPBL sample obtained 11 years earlier, although the numbers of supporting reads do not allow us to draw strong conclusions on this point.

Taken together, our results suggest that PPBL is not primarily driven by coding mutations, but is rather a phenotypically defined premalignant condition. The t(14;18) translocation has been identified by polymerase chain reaction in some patients with PPBL.^13,14  Although the presence of a BCL-2 fusion gene is suggestive, it does not formally demonstrate that an oncologic pathway drives the pathogenesis of PPBL, given the occasional presence of the translocation in healthy individuals. Alternatively, the association with HLA-DR7 expression and tobacco exposure⁶  suggest that PPBL might correspond to an abnormal immune response that paves the way to lymphoma. The B-cell proliferation pool associated with the disease could favor the emergence of subclones carrying oncogenic mutations, some of which may develop years later into subsequent lymphomas.