Genomic Landscape of CXCR4 Mutations in Waldenström Macroglobulinemia

This article is featured in Highlights of This Issue, p. 1299

Whole-genome sequencing has revealed CXCR4 as the second most frequent somatic mutation, identified in approximately 30% of Waldenström macroglobulinemia, similar to germline mutations found in the WHIM (warts, hypogammaglobulinemia, infections, and myelokathexis) syndrome (1–3). MYD88 L265P mutations remain the most frequent mutation reported in Waldenström macroglobulinemia to date, in nearly 90% of Waldenström macroglobulinemia (4, 5). Interestingly, MYD88 L265P may act as a founder mutation because of its high frequency in Waldenström macroglobulinemia; and thus it is suspected that a second hit may accelerate development and progression of the malignant clones leading to full-blown in Waldenström macroglobulinemia. Although the mechanism of deregulation of CXCR4 axis is not fully understood in Waldenström macroglobulinemia, it is possible that CXCR4 mutation might represent one of these secondary events.

C-X-C chemokine receptor type 4 (CXCR4) is a G-protein–coupled receptor that plays an important role in lymphopoiesis and cell trafficking (2, 6), along with its ligand, the stromal cell–derived factor-1 (CXCL12/SDF-1). The SDF1/CXCR4 axis promotes activation of several pathways including RAS, Akt, and NF-κB and interplays with BCR pathway (6–8). The CXCR4 gene is located on the long arm of chromosome 2 at position 21 and codes for a chemokine receptor that promotes migration and survival of various B lymphoid malignancies (8–10).

CXCR4 mutations were identified using Sanger sequencing by Treon and colleagues in nearly 25% of Waldenström macroglobulinemia, and the CXCR4 C1013G mutation was described as the most frequent recurrent CXCR4 mutation in 7% (11). However, the occurrence of potential subclonal of CXCR4 mutation and the proportion of CXCR4 mutation over the other known mutations was not available; in other words, how this mutation lies in the architecture of the MYD88-mutated clone. On contrary, Roccaro and colleagues have reported 30% incidence rate of the CXCR4 C1013G mutation using qPCR; however, they have solely studied this mutation (12).

In this study, we characterized the genomic landscape of CXCR4 somatic mutation in a large cohort of 98 patients with Waldenström macroglobulinemia, combining deep sequencing of target genes allowing study of the clonal architecture in the Waldenström macroglobulinemia cells, including subclonal analysis, and to confirm the incidence rate of these mutations and single polymorphism nucleotide array (SNPa) to decipher the genomic landscape of CXCR4-mutated Waldenström macroglobulinemia.

Ninety-eight patients diagnosed with Waldenström macroglobulinemia were included in this study (63 males, 35 females). The diagnosis and treatment initiation criteria of Waldenström macroglobulinemia were as published (13, 14). Among the cohort with available clinical follow up (n = 93), 68 patients were treated. First-line treatment was initiated at diagnosis in 41% of the patients. Front-line therapy included chloraminophene alone in 12 cases, fludarabine alone in 3 cases, rituximab (R) alone in 3 cases or in association with chemotherapy in 51 cases (dexamethasone-R- cyclophosphamide in 19 cases, bortezomib-R-dexamethasone in 7, R-bendamustine in 5 cases). Patients were untreated at time of bone marrow collection and gave informed consent prior to research sampling. No familial form of Waldenström macroglobulinemia was included.

Research sampling consisted of bone marrow samples and blood cells for all Waldenström macroglobulinemia. All samples were enriched using immunomagnetic beads (B cell isolation kit for tumoral cells from bone marrow sample, and Pan T cell isolation kit, Miltenyi-Biotec), following Ficoll-paque gradient centrifugation. The purity of samples was confirmed by multiparametre flow cytometry using a marker combination (including CD19, κ, λ, CD2, CD38, CD138, and CD27). More than 90% a light-chain isotype–positive B cells was detected in all cases (mean, 95.8% of tumoral cells; ref. 15).

DNA was extracted from isolated cells using Qui Amp kit (Sigma-Aldrich Co.). MYD88 L265P, CD79A, and CD79B mutations were analyzed as previously described (16) in 98 patients. C-terminal CXCR4 gene was amplified from gDNA by PCR as previously described (n = 98 patients; ref. 5). All the exons of CXCR4 gene were analyzed using targeted next-generation sequencing (NGS) in a cohort of 53 patients. NGS was analyzed using the Ion Torrent PGM platform (Life Biotechnologies). Amplicons covering the regions of interest were designed with an amplicon length of 150 to 250 bp. Libraries were sequenced with 200-bp read length on a 318 chip. Mutation calling was performed using the torrent suite variant caller under the low stringency somatic settings (TS4.0). We have studied CXCR4 mutations along with MYD88 L265P (exon 5), CD79A (ITAM domain), CD79B (ITAM domain), CARD11 (exons 5–9), N-RAS (exons 2 and 3), K-RAS (exons 2 and 3), BRAF6 (exon 15), and PTEN (exon 5+7) using NGS (n = 53).

To quantify the mutated clone, we have analyzed the VAF (variant allele frequency) defined as the number of reads that mapped each studied to this position, cover the variant base and show the reference allele, divided by all fragments covering the site. VAF was corrected by the percentage of tumoral cells in B cell selected sample. Mutation calls were considered positive when called by at least 20 variants reads. Complete sequence data of the coding and splice site regions of CXCR4 were generated at a mean depth coverage of 2,000× per nucleotide. NGS assay also allowed a better assessment of the percentage of CXCR4 mutant allele whose sensitivity is of 1%.

Gene expression profiling (GEP) was performed using U133A arrays (Affymetrix) for seven of the patients with Waldenström macroglobulinemia. Total RNA was extracted from purified B tumoral cells population isolated from bone marrow using the TRIzol method. Expression data were normalized using the Robust Multi-array Average (RMA) algorithm. Differential gene expression was analyzed with the Bioconductor R package “limma,” which is well-known to improve variance modeling when the number of replicates is small. All probes with an adjusted P value under 0.05 were considered differentially expressed. Their normalized expression values were then represented in a heatmap, whose colors vary according to the standardized by row values of the input data. In parallel, a gene set enrichment analysis (GSEA) was performed for several gene sets among them the C2 collection of the MSigDB (Supplementary Tables S1 and S2). The PGSEA package (17) was used to calculate z scores for all gene sets, and P values were adjusted for multiple testing (18).

Genome-wide detection of copy number alteration (CNA) and LOH was performed using the Genome-Wide Human SNP Array 6.0 (Affymetrix). Conventional cytogenetic analysis was performed on DSP30 + IL2 stimulated bone marrow cells. FISH was performed to detect chromosomal aberrations: 6q23 deletion, 17p12, 11q22 deletions, 13q14 deletion, trisomy 12, trisomy 4 (n = 90; ref. 15). Immunophenotypic expression of CXCR4 (R et D systems, UK), CD49d, CD27, CD80, CD86, CD138 gating on CD19⁺ cells was performed along with CD38 (Immunotech) and determination of the Matutes's score by flow cytometry. The ratio of mean fluorescence intensity (MFIR) was calculated as the ratio of specific fluorescence with an isotype control.

The relationships between the clinical, biologic, and molecular parameters were determined using a nonparametric test (Mann–Whitney), a t test, a χ², or Fisher exact tests when appropriate. Correlations were tested through Spearman correlation coefficient. The differences between the results of the comparative tests were considered statistically significant if P < 0.05. All statistical analyses were done with the SPSS 15.0 software. Survival was studied using Kaplan–Meier test, comparisons were made with log-rank test

We have first used NGS (deep sequencing) to screen the entire sequence of CXCR4 (n = 53 Waldenström macroglobulinemia samples) with a greater sensitivity and to quantify the allelic frequency of the variant of CXCR4 mutations in B selected cells. We have identified 14 of 53 (26.4%) Waldenström macroglobulinemia with CXCR4 mutations. We then confirmed the observed mutations using NGS using Sanger sequencing (SaS) and extend to the entire cohort of 98 patients with Waldenström macroglobulinemia (including the 53 done by NGS). We identified CXCR4 mutations (CXCR4^mut) for a total of 24 of 98 (24.5%), and we have characterized several new mutations across the entire coding exons of CXCR4 (Fig. 1).

Overall, 17 different mutations were identified in CXCR4, including 12 that we have described for the first time in Waldenström macroglobulinemia. Interestingly, only one type of CXCR4 mutation was observed in either patient, and all mutations identified were located in the 45 amino acid intracytoplasmic carboxy-terminal tail. These mutations were never observed in the paired T lymphocytes and thus confirmed their somatic feature. All mutations were heterozygous, and no UPDa (acquired uniparental disomy, LOH without variation of copy number) was observed at CXCR4 locus nor variation of copy gene number (gain or deletion) in our cohort, using SNP array.

NGS also allowed the quantification the allelic frequency of the variant of CXCR4 mutations. This quantification allows to conclude for the percentage of mutation in the clonal population for a given patient and thus to differentiate between clonal (dominant clonal) and subclonal. The mutation load of CXCR4 varied from 13.5% to 47.58% (mean, 35.2%) using the VAF (n = 14, CXCR4^mut Waldenström macroglobulinemia in our series). Interestingly, for 12 patients with Waldenström macroglobulinemia, the high VAF of CXCR4 mutation was suggested to be present in the dominant clone. However, we were also able to identify the presence of CXCR4 mutations at the subclonal level in 4 patients using NGS, and for these 4 patients, the VAF varied from 13.5% to 26% (Fig. 2).

Among the CXCR4 mutations observed, 11 of 24 (45.8%) and 13 of 24 (54.2%) were nonsense (CXCR4^NS) and frameshift (CXCR4^FS) mutations, respectively. The most frequent mutation was the C1013G (S338X) mutation (5 of 98, 5.1%) followed by C1013A (S338X; 3 of 98, 3%). The modelization of C1013G/CXCR4 and C1013A/CXCR4 variants predicts a stop codon in place of a serine at amino acid position 338 (S338X). We also found several other mutations to S338, suggesting a hotspot locus in CXCR4 gene. The S338 hotspot mutation was targeted in 50% of mutated cases in Waldenström macroglobulinemia. Other CXCR4^NS mutations were here described. Interestingly, the exact same acquired R334X mutation in Waldenström macroglobulinemia was also previously described in the constitutional WHIM syndrome (19).

Because we observed that all mutations were located in the C-terminal tail of CXCR4 and that this C-terminal domain plays a critical role in the desensitization process that contributes to the regulation of the signal transduction and CXCR4 expression (2); we then sought to study the impact of CXCR4 mutation on the expression of CXCR4 on the membrane of Waldenström macroglobulinemia tumor cells using flow cytometry. CXCR4 was expressed in all cases (n = 53), including the 12 CXCR4^mut patients. The pattern varied widely using the MFIR to evaluate the CXCR4 expression (MFI varied from 4.6 to 148.7). However, there was a significant greater expression of CXCR4 protein in Waldenström macroglobulinemia with CXCR4^mut (P = 0.003; Fig. 3). This increased expression was observed independently of the type of mutation (data not shown).

CXCR4 directly interacts with CD49d in response to CXCL12 signaling in regulating migration and adhesion of Waldenström macroglobulinemia cells to the bone marrow microenvironment (10). CD49d favors lymphocyte homing by cooperating with CXCR4 in stromal cell adhesion and extracellular matrix (9). We then studied the impact of CXCR4 mutation on the expression of the integrin VLA4 (CD49d) using flow cytometry (n = 53). CD49d was expressed in all Waldenström macroglobulinemia cases; but no difference in CD49d expression was observed according to CXCR4 mutation. In contrast, a higher expression of CD49d was observed in Waldenström macroglobulinemia with MYD88 L265P mutation (P = 0.048; Fig. 3). Similarly, we found no difference in CD38 expression according to CXCR4 mutation, despite of the known functional link between CXCR4, CD49d, and CD38 (20). CD138 mediates cell adhesion to the extracellular matrix and is a marker of plasmacytic differentiation (21). Interestingly, we found a significant lower expression of CD138 (syndecan1) in the CXCR4^mut group (P = 0.034). We found no difference in the CD27, CD23, CD10, K/L expression and the Matutes' score between CXCR4^mut and CXCR4^wild subgroups.

We then sought to study whether a specific high-throughput SNPa signature was associated to CXCR4^mut Waldenström macroglobulinemia in a series of 53 Waldenström macroglobulinemia including 12 CXCR4^mut. For this analysis, we have considered “presence of one” as any genomic abnormality including gain, loss, and/or copy number without LOH (CN-LOH), identified using SNPa. In this first analysis, we have compared presence versus absence of genomic abnormality according to presence of CXCR4 mutation. We found a relationship between CXCR4^mut and a greater frequency of genomic aberrations in Waldenström macroglobulinemia compared with CXCR4^wild (91% vs. 68%, P = 0.010).

We then evaluated the number of genomic abnormality as a reflection of the genomic complexity on the basis of the number of SNP abnormalities. We found that CXCR4-mutated Waldenström macroglobulinemia had a greater mean number of abnormality (5.8 vs. 2.8 per patient, P = 0.046). No significant difference was observed between Waldenström macroglobulinemia with clonal or subclonal CXCR4 mutation.

Furthermore, we analyzed the relationship between CXCR4 mutation and certain type of SNP aberration. We have observed that Waldenström macroglobulinemia with CXCR4^mut had a greater incidence rate of trisomy 4 (complete or partial), 58% versus 12% respectively (P = 0.002); a greater frequency of gain of Xq, including MYD88 pathway–based IRAK1 gene, and of deletion 8p (P = 0.002 and P = 0.007, respectively; Fig. 4). Finally, we also found a greater frequency of deletion 6q, including the NF-κB pathway–based TNFAIP3 gene (P = 0.038). We found no relationship to deletion 7q, deletion 11q, deletion 17p, deletion 13q14, trisomy 18 in our study. No difference of SNPa feature was observed according to the type of CXCR4 mutation (nonsense (CXCR4^NS) versus frameshift (CXCR4^FS) mutations.

We next attempted to delineate the mutational spectrum and intratumoral heterogeneity related to presence of CXCR4^mut for genes involved in the RAS (this pathway may be activated by CXCR4/protein G proteins), BCR and TLR pathways (MYD88 exon 5), known to be interconnected with CXCR4 pathway, using targeted NGS (6).

When looking at the incidence rate of the CXCR4 mutations with presence of MYD88 L265P mutation in the cohort of 98 patients, it appeared that 23 of 24 (95.8%) of the CXCR4^mut were MYD88 L265P mutated, except one Waldenström macroglobulinemia with a CXCR4^FS mutation. The high correlation between MYD88 L265P and CXCR4 mutational loads tends to suggest the coexistence of the 2 mutations in the same clone (Fig. 2). However, in four patients with Waldenström macroglobulinemia, a lower VAF of CXCR4 in comparison to MYD88 L265P suggested a subclonal CXCR4 mutation in the dominant clone harboring MYD88 mutation.

Regarding the presence of mutations in the RAS pathway in Waldenström macroglobulinemia, including PTEN, K-Ras, and N-Ras, crucial regulators of RAS pathway, we have not identified any mutation in our cohort (n = 53).

BCR pathway can be constitutively activated, either by signals from the microenvironment or by genetic aberration (22). Several mutations of key signaling molecules and regulators of the BCR pathway were reported overtime, particularly in the activated B-cell–like (ABC) diffuse large B-cell lymphoma (DLBCL) and in Waldenström macroglobulinemia (15, 23), including the immunoreceptor tyrosine–based activation motif (ITAM) signaling modules of the CD79A and CD79B BCR coreceptors and the coiled-coil domain of CARD11 (22). Mutations of CD79B and CD79A were observed in 12 of 98 (12.2%) of Waldenström macroglobulinemia. Interestingly, we found no coexistence of CXCR4^mut and CD79A/CD79B mutations, but in one patient who had CXCR4^mut at the subclonal level. No mutation of CARD11 was observed in our studied group (n = 53). One might conclude that there is no unique somatic mutation in the BCR pathway, but instead a wide variety of molecules that could be altered through molecular alterations in the BCR pathway, likely mutually exclusive in the same clone, such as CXCR4^mut and CD79A/CD79B mutations in the present situation in our study.

We finally studied the gene expression profile of Waldenström macroglobulinemia on 7 patients with Waldenström macroglobulinemia to understand whether the CXCR4 mutation impacted the gene expression and to identify a potential gene signature of Waldenström macroglobulinemia with CXCR4^mut. The differential analysis allowed to identify a gene signature attached to the CXCR4^mut molecular profile, corresponding to 32 probes and 27 genes. We found several genes among these, such as MMP8, CRISP3, or BRCC3 involved in cell death and survival or cell movement (ARG1, LYST). These probes were all underexpressed in the five CXCR4^wild Waldenström macroglobulinemia compared with the two CXCR4^mut Waldenström macroglobulinemia (Fig. 5 and Supplementary Tables S1 and S2). We then sought to identify deregulated pathways using Ingenuity analysis software. We found CXCR4^mut to be related to several functional network networks such as cell cycle and DNA replication or cell-to-cell signaling and interaction, cellular growth, and proliferation. Using GSEA for targeted gene sets previously associated with CXCR4 mutation (12), we also identified deregulation of proteasome pathway and BCR/B lymphocyte pathway in CXCR4^mut Waldenström macroglobulinemia (P = 0.01 and P = 0.04, respectively; Supplementary Table S1).

We thought to identify clinical–biologic characteristics of Waldenström macroglobulinemia according to CXCR4^mut features. The median age was 67 years (range, 36–92 years) in our series. We found that CXCR4^mut was associated to a higher IgM monoclonal component (r = 0.309; P = 0.006) and thrombocytopenia (r = −0.226; P = 0.048), markers of adverse prognosis in Waldenström macroglobulinemia. Indeed, 35% and 12% of patients had serum IgM level greater than 30 g/L (P = 0.023) and 27% and 6.5% of patients had low platelet count lower than 100 g/L (P = 0.018), respectively, in patients with CXCR4^mut versus CXCR4^wild. There was a trend for CXCR4^mut patients to have more often IPSS Waldenström macroglobulinemia stage 3 (P = 0.086). No association was found with regard to other markers of tumoral syndrome, such as adenopathy and splenomegaly, extramedullary localization, or lymphocytosis (one would consider a marker of egression of tumoral cells from the bone marrow; Supplementary Table S3). With a median follow-up of 6 years, 21 (22%) patients in our series had died with a median OS (95% confidence interval) not reached and estimated at 63% at 10 years. No significant difference in type of treatment or number of treatment line was observed (data not shown). The estimated overall survival (OS) at 10 years according to CXCR4^mut status was lower for CXCR4^mut patients, 50% versus 65.5% for patients with CXCR4^wild (P = ns, Fig. 6). Interestingly, the difference in OS between CXCR4^mut versus CXCR4^wild became significant in patients with indolent Waldenström macroglobulinemia (estimated 5-year OS rate at 50% and 93%, respectively, P = 0.019); the CXCR4^mut patients required therapy more often and much earlier than the CXCR4^wild.

We have described 12 new CXCR4 mutations for a total of 17 patients with Waldenström macroglobulinemia so far, observed in 25% of patients with Waldenström macroglobulinemia, using targeted deep sequencing. These data suggest a highly heterogeneous pattern of CXCR4 mutations in Waldenström macroglobulinemia and confirms using a more sensitive technique that was reported by Treon and colleagues using Sanger (11). We have also confirmed CXCR4 1013C>G, followed by 1013C>A, mutations to be the most frequent CXCR4 mutations in Waldenström macroglobulinemia, although in a lesser frequency as previously reported, 12.5% (for the 2 mutations using Sanger sequencing in CD19 selected cells) and 30% (only 1013C>A using allele-specific PCR in unselected tumoral samples) in Treon and colleagues and Roccaro and colleagues, respectively (11–12). Overall the high frequency of CXCR4 Whim-like mutation is a new hallmark of Waldenström macroglobulinemia (12, 24).

Interestingly, all mutations, either nonsense or frameshift, occurred in the C-terminus of CXCR4. It was suggested that the truncation of the distal amino acid region of the C-terminus known to regulate signaling of CXCR4 by CXCL12, might deregulate the CXCR4/SDF1 axis signaling pathway, and thus participate in the pathogenesis of Waldenström macroglobulinemia (2, 11, 12, 25). One might propose that the type of mutation does not matter much in Waldenström macroglobulinemia, on the contrary to the loss of C-terminus domain of CXCR4 protein which modified function might play a role in Waldenström macroglobulinemia pathogenesis (2). Functional studies may confirm this hypothesis in the near future.

We found a greater CXCR4 expression on tumor cells of Waldenström macroglobulinemia carrying CXCR4 mutation, independently of the type of mutation, similar to that was described in WHIM syndrome. It is suggested that the increase of CXCR4 receptor expression would induce an altered migration profile in response to SDF1 (25), resulting in an increased egression of tumor cells as demonstrated in mouse model (12). In our study, we also found a decreased expression of CD138, a cell surface syndecan-1 that mediates cancer cell adhesion to the extracellular matrix, and alteration of expression of several genes involved in matrix degradation such as MMP8 or cell migration on gene expression profiling. Taking together, CXCR4 mutation might alter the ability of the tumor cells to interact with the bone marrow microenvironment in Waldenström macroglobulinemia, as shown in CXCR4 S338X cells (12).

Nearly all patients with Waldenström macroglobulinemia with CXCR4 mutations harbored the MYD88 L265P mutation, as initially described by Treon and colleagues suggesting a potential cooperation between these pathways (11). An important dataset obtained in our series came from the quantification of the allelic frequency of the variant (VAF) of CXCR4 mutations into tumoral cells using NGS that may identify minor subclones of less than 1% in our study. Targeted deep sequencing allows to conclude for the percentage of mutations in the clonal population for a given patient and thus to differentiate between clonal (dominant clonal) and subclonal mutations. Combining analysis of VAF of several genes, CXCR4 mutation was expressed in the same clone as MYD88 L265P, as we have identified a high allelic frequency of CXCR4 mutation MYD88 L265P in Waldenström macroglobulinemia tumor cells; but for 4 patients who had CXCR4 mutation present only at the subclonal level. Overall, these data may suggest the following hypothesis, where MYD88 L265P mutation is a founder mutation in Waldenström macroglobulinemia, a first genetic hit that would promote NF-κB and JAK/STAT3 signaling (4, 26). Direct inhibition of MYD88 L265P signaling overcomes CXCL12-trigered survival effects in CXCR4-mutated cells, supporting a primary role for MYD88 signaling in Waldenström macroglobulinemia (25). CXCR4 could then be one of the secondary events in some patients who showed subclonal mutation of CXCR4 compared with MYD88 L265P mutation present in the main clone. However, longitudinal analysis studies are needed to explore the dynamic of clonal architecture to identify driver or additional mutations in Waldenström macroglobulinemia that could contribute to clinical progression or chemoresistance. Indeed, it was shown that a diminished clinical activity of ibrutinib was observed in patients with Waldenström macroglobulinemia with CXCR4 mutations (25;30).

In addition, we also found absence of co-occurrence of CD79A/CD79B and CXCR4 mutations in Waldenström macroglobulinemia. These data may suggest that these types of activating mutations of BCR and CXCR4 pathway were mutually exclusive in MYD88 L265P Waldenström macroglobulinemia (6). A cooperation between CXCR4 and TLR signaling activated by MYD88L265P mutation was described in B cells (6). The functional role of mutations in the BCR pathway is not fully described in Waldenström macroglobulinemia, but CD79B mutation was shown to alter the BCR response in diffuse large B lymphoma (15, 23). We might thus suggest the existence of two subgroups of MYD88 L265P Waldenström macroglobulinemia, one with CXCR4 mutations and one with BCR pathway mutations that may involve in cooperation with the aforementioned driver mutation which primarily affect TLR signaling. These data further may advance the concept that a complex rather than a unique deregulation of the TLR/NF-κB pathway characterizes MYD88 L265P Waldenström macroglobulinemia with others interconnected pathways.

We have observed that CXCR4 mutation was more frequently associated to a complex genomic signature, including gain of chromosome 4, deletion 6q, and gain of chromosome X using SNPa. High-throughput genomic studies have identified multiple mechanisms of genetic changes in Waldenström macroglobulinemia including several recurrent CNA such as deletion 6q, deletion 13q, or gain of chromosome 4 and CN-LOH (15, 27). This genomic pattern segregates further Waldenström macroglobulinemia among the others B- cell tumors. The prognostic role of these genomic alterations is not fully understood, but the presence of more than 3 SNPa abnormalities was associated with symptomatic status (15, 27, 28). Taking together, this finding suggests an inter- and intraclonal heterogeneity and the potential selective advantage of specific combinations of genetic lesions in Waldenström macroglobulinemia, in particular in the subset of CXCR4^mut Waldenström macroglobulinemia. The understanding of these genomic abnormalities, including gene mutations and CNA, described herein might help deciphering the pathogenesis of Waldenström macroglobulinemia to determine associated risk of progression, relapse, and drug resistance.

Overall, Waldenström macroglobulinemia with CXCR4 mutation had a specific clinicobiologic and genomic signature associated to features characterized with adverse prognosis in Waldenström macroglobulinemia, including high IgM M-spike and thrombocytopenia (part of the IPSS Waldenström macroglobulinemia score adverse features) and greater frequency of complex genomic aberrations pattern using SNP array. Our data therefore may suggest a worse prognosis for CXCR4 mutation subgroup of asymptomatic Waldenström macroglobulinemia, but we did not observe significant impact on overall survival in our cohort as previously described (11). Further studies are needed to explore the prognosis value of CXCR4 mutation in Waldenström macroglobulinemia in clinical trials. Interestingly, the potential clinical impact of Waldenström macroglobulinemia with CXCR4 mutation has already been suggested by Roccaro and colleagues and Treon and colleagues on the basis of their in vitro studies showing drug resistance in CXCR4-mutated Waldenström macroglobulinemia cells exposed to BTK, mTOR, and PI3K inhibitors but not proteasome inhibitors (29, 30). This study, along with other, thus led us to propose that it is primetime for a systematic evaluation of CXCR4 mutation as part as the pretreatment package of Waldenström macroglobulinemia, to identify patients with CXCR4 mutation that might benefit more from proteasome inhibitors as compared with other options, such a BTK inhibitors.

In conclusion, approximately 25% of Waldenström macroglobulinemia harbor CXCR4 mutation, leading to altered CXCR4 protein at the C-terminal end, irrespective of the pleiotropic pattern of mutations. The study of CXCR4 mutations showed existence of intraclonal (variation in co-expression of MYD88 and CXCR4 mutations) and interclonal (BCR and CXCR4 mutations in MYD88L265P Waldenström macroglobulinemia) heterogeneity when analyzing the molecular landscape of CXCR4 mutation in Waldenström macroglobulinemia. Patients with CXCR4 mutation displayed a specific clinical–biologic–genomic and transcriptomic signature, possibly related to specific physiopathologic mechanism related to CXCR4 mutation. This dataset further suggests the genomic heterogeneity of Waldenström macroglobulinemia, beyond the current indolent versus symptomatic classification.

No potential conflicts of interest were disclosed.

Conception and design: S. Poulain, M. Figeac, F. Lepretre, X. Leleu

Development of methodology: S. Poulain, A. Venet-Caillault, M. Figeac, S. Geffroy, X. Leleu

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): S. Poulain, C. Roumier, M. Figeac, C. Herbaux, G. Marot, E. Bertrand, A. Decambron, E.M. Boyle, A. Renneville, S. Tricot, B. Quesnel, X. Leleu

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): S. Poulain, C. Roumier, M. Figeac, G. Marot, E. Doye, B. Quesnel, X. Leleu

Writing, review, and/or revision of the manuscript: S. Poulain, C. Roumier, C. Preudhomme, X. Leleu

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): S. Poulain, C. Roumier, E. Doye, F. Lepretre, O. Nibourel, B. Quesnel, X. Leleu

Study supervision: S. Poulain, E. Doye, P. Duthilleul, X. Leleu

The authors thank Pauline Vandycke, Valérie Grandières, Claudine Delsaut, Axelle Séghir, and Sabine Ranwez for their excellent technical assistance.

This work was supported by the Comité Septentrion de la Ligue contre le Cancer et la Fondation Française pour la Recherche contre le Myélome et les Gammapathies (FFRMG).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.