|Home | About | Journals | Submit | Contact Us | Français|
CXCR4WHIM somatic mutations are distinctive to Waldenstrom Macroglobulinaemia (WM), and impact disease presentation and treatment outcome. The clonal architecture of CXCR4WHIM mutations remains to be delineated. We developed highly sensitive allele-specific polymerase chain reaction(AS-PCR) assays for detecting the most common CXCR4WHIM mutations (CXCR4S338X C>A and C>G) in WM. The AS-PCR assays detected CXCR4S338X mutations in WM and IgM monoclonal gammopathy of unknown significance (MGUS) patients not revealed by Sanger sequencing. By combined AS-PCR and Sanger sequencing, CXCR4WHIM mutations were identified in 44/102 (43%), 21/62 (34%), 2/12 (17%) and 1/20 (5%)untreated WM, previously treated WM, IgM MGUS and marginal zonelymphoma patients, respectively, but no chronic lymphocytic leukaemia, multiple myeloma, non-IgM MGUS patients or healthy donors. Cancer cellfraction analysis in WM and IgM MGUS patients showed CXCR4S338X mutations were primarily subclonal, with highly variable clonal distribution(median 35·1%, range 1·2–97·5%). Combined AS-PCR and Sangersequencing revealed multiple CXCR4WHIM mutations in many individual WM patients, including homozygous and compound heterozygous mutations validated by deep RNA sequencing. The findings show thatCXCR4WHIM mutations are more common in WM than previously revealed, and are primarily subclonal, supporting their acquisition after MYD88L265P in WM oncogenesis. The presence of multiple CXCR4WHIM mutations within individual WM patients may be indicative of targeted CXCR4 genomic instability.
We recently described the existence of somatic mutations in the C-terminal domain of CXCR4 in patients with Waldenström Macroglobulinaemia (WM) (Treon et al, 2012; Hunter et al, 2014). Mutations in CXCR4 were the second most common somatic variant identified after MYD88L265P, and are distinctive for WM. By whole genome sequencing, somatic mutations in CXCR4 were detected in 27% of WM patients, and confirmed by Sanger sequencing (Hunter et al, 2014). The location of somatic mutations in the C-terminal domain in WM patients is similar to that observed in the germline of patients with WHIM (Warts, Hypogammaglobulinaemia, Infections and Myelokathexis) syndrome, a congenital immunodeficiency disorder characterized by chronic noncyclic neutropenia (Dotta et al, 2011).
CXCR4 plays a critical role in WM cell trafficking to the bone marrow (BM) stroma, wherein protection is afforded to tumour cells against many anti-neoplastic agents (Ngo et al, 2008). Over 30 different types of somatic C-terminal mutations in CXCR4 have been described in WM patients, including frameshift and nonsense variants (Hunter et al, 2014; Treon et al, 2014). The most common variant is CXCR4S338X, which represents over 50% of CXCR4 mutations found in WM patients. A C>A or C>G transversion at nucleotide position 1013 in the CXCR4 gene results in the generation of a stop codon which leads to truncation of protein at amino acid position 338, and loss of the terminal 15 amino acids of the C-regulatory domain. Both CXCR4S338X C>A and C>G nonsense mutations are associated with more aggressive disease presentation at diagnosis, including higher BM involvement and serum IgM levels, as well as symptomatic hyperviscosity (Treon et al, 2014; Schmidt et al, 2015). In vitro modelling has shown that, in response to CXCL12 (SDF-1a), CXCR4S338X engineered cells show muted CXCR4 receptor internalization, which results in enhanced and prolonged AKT and ERK activation and confers resistance to many anti-neoplastic agents used to treat WM patients, including BTK, PI3Kd, BCL2 and proteasome inhibitors (Roccaro et al, 2014; Cao et al, 2015). CXCR4WHIM mutations are also associated with clinical resistance to the BTK inhibitor ibrutinib in patients with previously treated WM (Treon et al, 2015). The detection of CXCR4WHIM mutations therefore has both diagnostic and therapeutic implications in WM.
The Sanger technique is commonly used to sequence genesalthough is not very sensitive and often requires a minimummutation burden of 15–20%. In contrast, allele-specific polymerase chain reaction (AS-PCR) is considerably more sensitive, with a range of detection down to 0·1%, and is easier toadopt and provide interpretive results in a clinical diagnosticsetting. We therefore developed quantitative AS-PCR assaysto detect the most common CXCR4WHIM mutation, S338X, which results from C>G and C>A nucleotide substitutions. We then investigated its use in a separate cohort of patients with untreated and previously treated WM, IgM monoclonal gammopathy of unknown significance (MGUS), marginal zone lymphoma (MZL), chronic lymphocytic leukaemia (CLL), multiple myeloma (MM), and non-IgM MGUS and compared these findings against those derived from Sanger sequencing. Finally, we determined the cancer cell fraction (CCF) of CXCR4 mutations in WM by synchronous quantitative AS-PCR analysis against MYD88L265P. The findings from these studies detail a highly complex clonal architecture for CXCR4 mutations, and are indicative of targeted genomic instability for the C-terminal region of this gene in WM patients.
CD19-selected cells derived from BM aspirates of 13 WMpatients, and peripheral blood of 12 healthy donors was usedin the development of CXCR4S338X AS-PCR assays. Tumour samples from a separate cohort of WM, IgM and non-IgM MGUS, MZL, CLL, MM (including IgM MM), and healthy donors were then assayed. Consensus criteria were used for WM and IgM MGUS determinations, and their baseline characteristics are depicted in Table I (Owen et al, 2003). CD19-selected cells from BM aspirates were isolated, and both DNA and RNA extracted as previously described (Xu et al, 2013, 2014). MYD88 mutation status was determined by quantitative AS-PCR and Sanger sequencing (Xu et al, 2013). Subject participation was approved by the Harvard Cancer Center/Dana-Farber Cancer Institute Institutional Review Board. All participants provided written consent.
Since CXCR4S338X mutations occur due to either C>G or C>A transversions at nucleotide position 1013, we developed two AS-PCR assays to permit their detection. The details for the development of these assays are presented in the online supplement.
The forward PCR primer 5′ - ATG GGG AGG AGA GTT GTA GGA TTC TAC -3′ and reverse PCR primer 5′- TTG GCC ACA GGT CCT GCC TAG ACA-3′ were designed to amplify the CXCR4 open reading frame. Amplified PCR products were isolated by QIA quick gel extraction kit (Qiagen, Valencia, CA) and sequenced using both forward and reverse PCR primers and an additional sequencing primer 5′-GCTGCCTTACTACATTGGGATCAGC-3′.
The CXCR4 forward PCR primer 5′- ATG GGG AGG AGA GTT GTA GGA TTC TAC -3′ and reverse PCR primer 5′-TTG GCC ACA GGT CCT GCC TAG ACA-3′ were used to amplify the gene fragment. TOPO Cloning Kits were used as per manufacturer's protocol (Thermo Fisher Scientific Inc., Grand Island, NY).
Deep RNA sequencing of CD19-selected cells from BM aspirates was performed to validate findings in five patients with multiple CXCR4WHIM mutations. RNA sequencing data was generated from 50 cycle HiSeq paired-end sequences (Illumina Inc., San Diego, CA) and aligned to HG19/GRCh37 ensemble genome reference using STAR (https://github.com/alexdobin/STAR/). Reads supporting each call were calculated using Integrative Genomics Viewer (IGV; Broad Institute, Cambridge, MA, USA). The median number of reads mapping to the CXCR4 C-terminal domain per patient was 8208 (range 5316–12 235).
Cancer cell fraction analysis for CXCR4S338X mutations was performed using synchronous, parallel quantitative AS-PCR analyses for MYD88L265P and CXCR4S338X C>A or C>G in sorted CD19-cells derived from BM aspirates from 21untreated WM and IgM MGUS patients expressingMYD88L265P and CXCR4S338X C>A and/or C>G mutations. Ahighly sensitive quantitative MYD88L265P AS-PCR was used (Xu et al, 2013). Standard curves were established for each AS-PCR assay that was run on the same plate for each sample, and cell count expressive of these mutations was determined by standard curves and ΔCT. The CCF was determined as the ratio of cells expressing CXCR4S338X C>A or C>G/MYD88L265P. MYD88 and CXCR4 copy number was determined using TaqMan Copy Number Assays (Applied Biosystems, Grand Island, NY, USA).
Estimates of sensitivity, specificity and predictive values were performed using VassarStats (Poughkeepsie, NY, USA).
Given that CXCR4S338X mutations can result from either C>A or C>G transversions at nucleotide position 1013, and are the most common WHIM-like mutations in WM, we developed and validated sensitive real-time AS-PCR assays for detecting these mutations in samples obtained from WM patients and healthy donors. The details for the development of these assays are presented in the online supplement. In a separate cohort of 102 untreated WM patients whose mutation status was established by Sanger sequencing, the sensitivity for the CXCR4S338X C>A AS-PCR assay was 100% (95% confidence interval [CI] 39·5–100%) and specificity was 100% (95% CI 95·2–100%). For the CXCR4S338X C>G AS-PCR assay, the sensitivity was 100% (95% CI 67·8–100%) and specificity was 100% (95% CI 94·9–100%).
We next applied the CXCR4S338X C>A and CXCR4S338X C>GAS-PCR assays using the standard curve established intumour-containing samples from a separate cohort of 164WM (102 untreated, 62 previously treated), 12 IgM MGUS,20 MZL, 32 CLL, 14 MM (including 2 IgM MM), and 7non-IgM MGUS patients, as well as 32 healthy donors. Allsamples underwent Sanger sequencing. Sanger sequencing ofBM CD19-selected samples from WM patients revealedCXCR4WHIM mutations that included nonsense or frameshift mutations in 37/102 (36·3%) of untreated and 17/62 (27·4%)previously treated WM patients (Table II). Sanger tracingsshowed the presence of multiple CXCR4WHIM mutations within each of three untreated patients (WM1-WM3) (Fig 1A). Cloning and sequencing studies indicated that the CXCR4WHIM mutations were compound heterozygous for all three of these patients, with 3/44 and 12/44 clones expressing CXCR4G332 fs and CXCR4S338X, respectively for WM1; 12/43 and 6/43 clones expressing CXCR4K333X and CXCR4S338X, respectively, for WM2; and 2/40 and 8/40 clones expressing CXCR4S338X and CXCR4S338Xfs, respectively, for WM3 (Fig 1B). Deep RNA sequencing performed in one of these patients (WM3) also showed map reads consistent with a compound heterozygous mutation (Table III). Sanger tracings in three other untreated patients (WM4-WM6) showed an allele burden for CXCR4S338X that exceeded that of the wild-type CXCR4 allele. The MYD88L265P allele burden for these individuals was less than wild-type MYD88, a finding highly suggestive of homozygous CXCR4S338X expression. Deep RNA sequencing performed in one of these patients (WM4) also showed map readings supporting the existence of a homozygous CXCR4S338X subclone (Table III). No copy number variant was detected in the CXCR4 region for any of the above patients (data not shown).
Among the 102 untreated WM patients, the AS-PCR assay for CXCR4S338X C>A identified 11 patients with this mutation (Fig 2). This included 4 patients previously identified as having this variant by Sanger sequencing, as well as 7 patients who were wild-type by Sanger sequencing (Fig 3). Among the same cohort of 102 untreated patients, the AS-PCR assay for CXCR4S338X C>G identified 20 patients with this mutation (Fig 2). This included 11 patients previously identified as having this variant by Sanger sequencing, as well as 9 patients who were wild-type by Sanger sequencing (Fig 3). Five of these patients had both CXCR4S338X C>A and CXCR4S338X C>G mutations detected by the AS-PCR assays. In total, 7 individuals among the 65 untreated patients whose Sanger sequencing showed wild-type CXCR4 were identified as having a CXCR4WHIM mutation by AS-PCR. Taken together, the results of AS-PCR and Sanger sequencing revealed that 44/102 (43%) of untreated patients had CXCR4WHIM mutated disease (Table IV), with multiple mutations within 13/44 (31%) of these patients. Deep RNA sequencing confirmed the presence of compound heterozygous mutations in 3 patients (WM7-WM9) whose samples were used for validation (Table III). Forty-three of the 44 (97·7%) untreated CXCR4WHIM WM patientswere also positive for the MYD88L265P mutation by AS-PCR. Sanger sequencing did not reveal any other MYD88 mutations in the sole patient who did not express the MYD88L265Pmutation by AS-PCR.
Among the 62 previously treated WM patients, the AS-PCR assay for CXCR4S338X C>A identified seven patients with this mutation (Fig 2). This included five patients who had this variant by Sanger sequencing, as well as two patients who were wild-type by Sanger sequencing (Fig 3). Among the same cohort of 62 previously treated WM patients, the AS-PCR assay for CXCR4S338X C>G identified 9 patients with this mutation (Fig 2). This included seven patients who had this variant by Sanger sequencing, as well as two patients who were wild-type by Sanger sequencing (Fig 3). In total, four additional individuals were identified as having a CXCR4WHIM mutation among 45 previously treated WM patients who were wild-type by Sanger sequencing. Combining the results of AS-PCR and Sanger sequencing revealed that 21/62 (34%) of untreated patients had CXCR4WHIM mutated disease (Table IV), with multiple mutations present in 4/21 (19%) of these patients. All 21 of the previously treated CXCR4WHIM mutated patients were also positive for the MYD88L265P mutation by AS-PCR.
In 12 patients with IgM MGUS, the CXCR4S338X C>A AS-PCR assay did not detect any mutations (Fig 2). However, the CXCR4S338X C>G AS-PCR assay detected this mutation in 2 of 12 (17%) patients with IgM MGUS (Fig 2). All 12 of these IgM MGUS patients were negative for CXCR4WHIM mutations by Sanger sequencing. Both IgM MGUS patients who were found to have the CXCR4S338X C>G mutation by AS-PCR were also positive for MYD88L265P by AS-PCR.
Sample analysis using both AS-PCR assays failed to detect the CXCR4S338X C>A or C>G variants in any of the tumour samples taken from the 20 MZL, 32 CLL and 14 MM patients, nor in any of 32 healthy donors (Fig 2). Sanger sequencing was also performed in these samples, and was remarkable for one (5%) MZL patient having a CXCR4S344 fs mutation that resulted from r.1030_1031insT (Table IV). This patient was wild-type for MYD88 by both AS-PCR assay for MYD88L265P, and also by Sanger sequencing of the entire MYD88 gene.
MYD88L265P expression levels, as determined by AS-PCR analysis, show strong correlation with tumour cell content in BM specimens from WM patients (Xu et al, 2013, 2014). To clarify the CCF of CXCR4 mutations in WM, parallel quantitative AS-PCR analyses for MYD88L265P and CXCR4S338X C>A or C>G were performed using tumour samples from 21 untreated WM and 2 IgM MGUS patients who were known to express the MYD88L265P and CXCR4S338X C>A and/or C>G mutations. The cell fraction expressive of these mutations was determined by ΔCT and standard curves for MYD88L265P and CXCR4S338X C>A or C>G that were run on the same plate for each sample. The CCF was determined as the ratio of cells expressing CXCR4S338X C>A or C>G/MYD88L265P. No copy number alterations for MYD88 and CXCR4 were found in this cohort by TaqMan Copy Number Assays.
Cancer cell fraction analysis for all (WM and IgM MGUS) patients showed CXCR4S338X mutations were primarily subclonal, with highly variable clonal distribution (median 35·1%, range 1·2–97·5%) (Fig 4). For the 13 WM patients who expressed CXCR4S338X C>G only, the fraction of cells expressing CXCR4S338X C>G relative to MYD88L265P was 44·5% (range 1·2–97·5%). For 7 WM patients who expressed both CXCR4S338X C>G and CXCR4S338X C>A, the fraction of cells expressing CXCR4S338X C>G relative to MYD88L265P was 4·2% (range 2·4–37·2%) and for CXCR4S338X C>A to MYD88L265P, it was 4·1% (range 1·1–55·8%). In one patient who expressed only CXCR4S338X C>A, the fraction of cells expressing CXCR4S338X C>A relative to MYD88L265P was 1·2%. For the two IgM MGUS patients who expressed CXCR4S338X C>G, the fraction of cells expressing CXCR4S338XC>G relative to MYD88L265P was 21·8% and 1·2%.
Despite the common dysregulation of CXCR4 in many solidand haematological malignancies, somatic mutations in thisgene remain largely distinctive of WM. In WM, CXCR4WHIMsomatic mutations are important determinants of WM disease presentation, as well as treatment outcome (Treon et al, 2014, 2015; Schmidt et al, 2015). We therefore investigated the clonal architecture of CXCR4WHIM mutations in WM patients using Sanger sequencing and quantitative AS-PCRassays, which detect the most common CXCR4WHIM mutation variants (CXCR4S338X C>A and C>G), as well as by deep RNA sequencing.
The AS-PCR assays developed for these studies showed highlevels of specificity and sensitivity for the CXCR4S338X C>A and CXCR4S338X C>G detection, and discriminated samples bearing their target variants from those samples with CXCR4WT, CXCR4WHIM frameshift and other nonsense mutations. Importantly, the AS-PCR assays identified CXCR4WHIM mutations in many WM and IgM MGUS patients who were identified as CXCR4WT by Sanger sequencing. At first glance, thefindings might suggest that the tumour burden for thesepatients was under the detection limit for CXCR4S338X by Sanger sequencing. However, the BM burden for these 11 individuals with WM who were identified as having CXCR4S338X byAS-PCR (but not Sanger sequencing) was 40% (range 5–70%), and the median CD19+ cell clonality by light chain restrictionanalysis was 89·8% (range 4·2–100%). These findings wouldsuggest that CXCR4S338X mutations in these patients are likely to be subclonal. To affirm these findings, a CCF analysis wasperformed by synchronous AS-PCR analysis of MYD88L265Pand CXCR4S338X that included untreated WM and IgM MGUS patients. The results from this study showed that for mostuntreated WM, as well as IgM MGUS patients, CXCR4S338Xmutations are subclonal with highly variable clonal distribution. The subclonal existence of CXCR4S338X, as well as otherCXCR4WHIM mutations identified by Sanger, as well as deep RNA sequencing studies in WM and IgM MGUS patients supports that CXCR4WHIM mutations are probably an early oncogenic event that follows acquisition of MYD88L265P. Thefinding of CXCR4S338X mutations in IgM MGUS has also beenshown by another group, though MYD88 mutation status forthese patients was not reported (Roccaro et al, 2014). In ourseries, both MYD88L265P and CXCR4WHIM mutation status wasdetermined for all patients. Between 50% and 80% of IgMMGUS patients harbour the MYD88L265P somatic mutation, and the presence of MYD88L265P is associated with a higherrate of evolution to malignancy, including WM and MZL(Landgren & Staudt, 2012; Jiménez et al, 2013; Varettoni et al, 2013; Xu et al, 2013). Both IgM MGUS patients with theCXCR4S338X mutation in our series also expressed theMYD88L265P mutation, and co-expression of bothCXCR4WHIM and MYD88L265P mutations is nearly universal inWM (Hunter et al, 2014; Treon et al, 2014). It is thereforeplausible that co-expression of CXCR4WHIM and MYD88L265Pmutations may facilitate progression of IgM MGUS to WM ina subset of patients. Further studies are needed to help expandon these findings and hypotheses. It is interesting that the oneMZL patient had a CXCR4S344 fs mutation but was MYD88WTSimilarly, Martínez et al (2014) identified a CXCR4R334X nonsense mutation in one of 15 MZL patients, who was also wild-type for MYD88. MYD88L265P is uncommon in MZL patientswith an estimated frequency of 6–10% (Ngo et al, 2011; Trøen et al, 2013). Additionally, determination of both MYD88 andCXCR4 mutation status may help in further discriminatingWM from MZL and other overlapping B-cell malignancies, which often share similar morphological, immunophenotypic, cytogenetic and clinical findings (Swerdlow et al, 2008; Arcaini et al, 2011; Kyrtsonis et al, 2011).
A remarkable but surprising finding when using the AS-PCR assays with Sanger sequencing was the common occurrence of multiple CXCR4 mutation types within individualpatients. A third of WM patients harboured multiple mutations at initial presentation that included both nonsense andframeshift mutations, findings that were supported by nextgeneration sequencing. The clinical significance for thesefindings remains to be clarified, and larger studies with longitudinal follow-up will invariably be required to delineatetheir importance. However, the common presence of multiple CXCR4WHIM mutation types in many WM patients issuggestive of targeted genomic instability within the C-terminal regulatory domain and warrants further investigation.
The findings of this study may also have implications for the management of WM patients given the importance of CXCR4 mutations in disease presentation and treatment response. The use of targeted deep next generation sequencing may provide the most comprehensive assessment of nonsense and frameshift mutations, including the presence of multiple CXCR4 mutations in WM patients. Longitudinal studies to address clonal evolution in patients with single and multiple CXCR4 mutations will also be interesting, particularly with targeted and highly selective agents, such as ibrutinib. It is interesting that fewer previously treated versus untreated patients harboured CXCR4WHIM mutations in our study, a finding that will invariably need further validation in a larger study cohort. However, these findings may allude to therapies that differentially impact CXCR4WHIM expressing clones, and longitudinal studies examining evolution of CXCR4WHIM clones across various WM therapies may be illuminating. The use of CXCR4 antagonists has also been proposed for CXCR4WHIM WM patients following encouraging preclinical data, and a clinical trial combining the anti-CXCR4 antibody ulocuplomab with ibrutinib is being initiated in WM patients (Roccaro et al, 2014; Cao et al, 2015). The development of assays that comprehensively evaluate CXCR4WHIM mutations could also help identify candidates for CXCR4 antagonist therapy.
In summary, our findings show that CXCR4WHIM mutations are more common in WM patients than previously revealed by whole genome sequencing or Sanger sequencing. Moreover, CXCR4 mutations are primarily subclonal, with highly variable clonal distribution among CXCR4WHIM mutated WM patients. The subclonal existence of CXCR4WHIM mutations in WM, as well as IgM MGUS patients, supports that the acquisition of CXCR4WHIM mutations is likely to be an early oncogenic event, but follows acquisition of MYD88L265P. Lastly, the presence of multiple CXCR4WHIM mutations in many WM patients may be indicative of targeted CXCR4 genomic instability and warrant further study.
Data S1. Development of quantitative AS-PCR assays for CXCRs338x mutations.
Supported by Peter S. Bing M.D., a translational research grant from the Leukemia and Lymphoma Society, a research grant from the International Waldenström Macroglobulinaemia Foundation, the Linda and Edward Nelson Fund for WM Research, the D'Amato Fund for Genomic Discovery, the Coyote Fund, and an NIH SPORE post-doctoral fellowship award to ZRH (P50CA100707).
Author contributions: SPT, LX and ZRH designed the study. MLP, RA, SK, JJC, AT, AT, AG, EM, MV, LA, JB, YT, JS, RC, NM, KCA and SPT collected study samples and data. LX, NT, YC, GY, JC, XL, JC, JZ and KA processed tumour samples and performed CXCR4 genotyping studies. LX, ZRH and SPT analysed the study data. SPT, LX and ZRH wrote the manuscript.
Conflicts of interest: No conflicts of interest are identified by the investigators for this work.