High surface IgM levels associate with shorter response to ibrutinib and BTK bypass in patients with CLL

The B-cell receptor (BCR) can define B-cell fate via antigen-induced and antigen-independent (constitutive) signals.¹  BCR signaling is also key to survival and proliferation of chronic lymphocytic leukemia (CLL) cells,^2-4  in which BCR-associated kinase inhibitors appear very effective.⁵  In CLL, clinical use is skewed toward Bruton's tyrosine kinase (BTK) pathway inhibitors (BTKi), among which ibrutinib has been the first-in-kind BTKi to rapidly shift medical algorithms away from chemotherapy.^6-8 

CLL cells have variably low surface immunoglobulin (sIgM) levels and recover expression in vitro, indicating that chronic antigen engagement is occurring at tissue sites.^9-11  The variable levels associate with signal strength, with a balance toward anergy when sIgM is low or proliferation when sIgM is high.^2,3,4  Levels and signaling capacity are typically higher in the subset with unmutated (U) Ig gene heavy-chain variable regions (IGHV), of pregerminal center CD5⁺ B-cell origin and poor prognosis, than in the subset with mutated (M) IGHV, of postgerminal center CD5⁺ B-cell origin and generally good prognosis.^12-16  However, sIgM levels/signaling vary within U-CLL and M-CLL, and this variability further informs natural propensity for disease progression.¹⁷ 

By irreversibly binding BTK at C481 and causing inhibition of downstream events,^18,19  ibrutinib leads to rapid redistribution of CLL cells from lymph nodes into the peripheral blood. While the redistribution is likely due to inhibition of migration more than adhesion,^20,21  the circulating CLL cells have a selective recovery of sIgM levels and proximal function, consequent to release from tissue-based antigen drive.^10,11,24  Suspension of ibrutinib can reverse the redistribution of CLL cells back to tissue sites, often leading to nodal disease progression and more aggressive disease, particularly if discontinuation occurs early, suggesting specific environmental requirements for disease resistance.^25-28 

Occurrence of genetic lesions, including mutations of the BCR-associated kinases BTK and PLCγ2, have been frequently documented in the CLL cells of patients who develop resistance to ibrutinib and rapidly progress.^29,30  However, these mutations are subclonal,^30-38  and appear insufficient to keep the signaling pathway active unless there is also sIgM engagement,²⁹  suggesting that environmental stimuli through sIgM are also necessary for evasion from ibrutinib. Conceivably, we have hypothesized that levels of sIgM would affect signaling strength and that this would ultimately have clinical consequences on response to ibrutinib.²⁴ 

In this study, we observed that high pretherapy sIgM levels associate with short duration of response to ibrutinib and with incomplete inhibition of intracellular calcium mobilization (iCa²⁺) and ERK1/2 phosphorylation (pERK) downstream to fully inhibited BTK. At this stage, BTK/PLCγ2 lesions were not evident. The surviving cells with high sIgM recovered constitutive AKT phosphorylation, potentially keeping the cells ready to become resistant to ibrutinib.

Seventy patients participating in the “real-world” observational study at the University of Southampton (NIHR/UKCRN ID: 31076) were investigated. Diagnosis of CLL was according to the 2008 International Workshop on Chronic Lymphocytic Leukemia (IWCLL2008)/National Cancer Institute (NCI) criteria.³⁹  Diagnoses were confirmed by a flow cytometry “Matutes score” > 3 in all cases.

Phenotypic (CD38, CD49d, and ZAP70 expression), immunogenetic (tumor IGHV usage and mutational status), fluorescence in situ hybridization (FISH) characteristics according to Dohner classification, and TP53 mutational status were determined pretreatment (baseline), as previously described.^17,40  -⁴³  Patients received ibrutinib once daily (starting at 420 mg dose) until toxicity and/or disease progression. Peripheral blood mononuclear cells (PBMCs) from patients with CLL were isolated prior to and during ibrutinib therapy by density gradient centrifugation and cryopreserved as previously described.^10,24  Blood was taken no later than 2 hours after dosing.

The study was approved by the Institutional Review Boards at the University of Southampton (REC: H228/02/t). All patients provided written informed consent.

After thawing, PBMCs were allowed to recover for 1 hour at 37°C in RPMI-1640 medium (Sigma Aldrich) supplemented with 10% fetal bovine serum, 2 mM glutamine, and 1% penicillin/streptomycin, as described.²⁴  For clinical correlation studies, viable PBMCs were stained in fluorescence-activated cell sorting (FACS) buffer (1% bovine serum albumin, 4mM ethylenediaminetetraacetic acid, and 0.15mM NaN₃ in phosphate-buffered saline) for 30 minutes on ice with the following antibodies: PE-conjugated anti-IgM rabbit F(ab′)₂ (Dako, Agilent Technologies), PE-conjugated or PerCP/Cy5.5-conjugated anti-CD5 (UCHT2), APC-conjugated or Pacific Blue-conjugated anti-CD19 (HIB19) (all from Biolegend), and corresponding controls. Cells were then washed, and data were acquired using a FACSCalibur (BD).^10,17  For functional assays, PE-conjugated anti-IgM (MHM-88, Biolegend) and a FACSCanto (BD) were used. The analysis of marker expression, reported as mean fluorescence intensity (MFI), was performed on the CD19+/CD5+ tumor population using FlowJo software.

Anti-IgM–induced signaling capacity was measured by intracellular calcium (iCa²⁺) mobilization and immunoblotting. The iCa²⁺ mobilization assay was performed using our previously described methods.¹⁷  Briefly, cells were incubated with 4 µM of Fluo3-AM (Life Technologies) and 0.02% (vol/vol) Pluronic F-127 (Sigma Aldrich) for 30 minutes at 37°C. Cells were then washed in medium, and data were acquired by flow cytometry. The signal from unstimulated cells was recorded for 30 seconds before addition of polyclonal goat F(ab′)₂ anti-human IgM (20 µg/mL, Southern Biotech). The percentage of IgM-mediated iCa²⁺ mobilization was calculated as peak percentage of cells responding to the stimulation (increase in fluorescent intensity above the set threshold measured on unstimulated cells) and normalized to the percentage of B cells in the lymophocyte population. In the text and figures, the term “sIgM iCa²⁺” has been used to indicate the percentage of responding cells to anti-IgM induced iCa²⁺ mobilization.

For immunoblotting, viable PBMCs were stimulated with polyclonal goat F(ab′)₂ anti-human IgM or soluble polyclonal control (20 µg/mL; Southern Biotech) at 37°C for 10 minutes or left untreated. Cells were then lysed in RIPA buffer, and equal amounts of proteins were resolved by SDS-PAGE and transferred to nitrocellulose membranes, as previously described.²⁴  Membrane staining with the primary antibodies anti-phosphoERK1/2 (Thr202/Tyr204), anti-ERK1/2, anti-phosphoAKT (Ser473), anti-AKT, anti-phosphoBTK (Y223), anti-BTK, anti-phosphoPLCγ2 (Y1217), anti-PLCγ2 (all from Cell Signaling Technology), and anti-GAPDH (Ambion) was followed by incubation with the appropriate anti-mouse or anti-rabbit horseradish peroxidase-conjugated secondary antibody (Dako). Imaging was performed using chemiluminescence reagents (Pierce ECL and SuperSignal West Femto, Thermo Scientific) and the ChemiDoc-It imaging system (UVP). The optical density of the bands was quantified using the ImageJ software. Protein phosphorylation of each sample was measured as the ratio between phospho-protein optical density and total-protein optical density (phospho/total) for each condition. “Inducibility” was defined as the ratio of induced/basal phosphoprotein levels at each time point, as previously described.²⁴  When specified, CLL cells were treated in vitro with 10 µM of ibrutinib (Selleckchem) for 1 hour at 37°C before signaling analysis.

BTK and PLCγ2 mutations were determined using custom assays with the NexteraXT kit (Illumina) for sequencing on the MiSeq platform or for long-read sequencing on the GridION (Oxford Nanopore Technologies) at the WISH laboratory, University of Southampton and validated at the Karolinska Institutet, Stockholm. DNA was extracted using the Qiagen miniprep kit (Qiagen) and diluted to 20 ng/µL. A multiplex polymerase chain reaction consisting of 4 assays with primers designed to amplify the mutational hotspots in BTK codon 481 and PLCγ2 codons 665, 707, and 848 was performed. Primers were designed using the Primer3 software (4.1.0) (supplemental Table 1). Amplified DNA was diluted and processed using either the NexteraXT kit (Illumina) or Amplicon barcoding kit SQK-AMB111.12 (Oxford Nanopore Technologies), according to manufacturer’s instructions. Libraries were quantified using Qubit High Sensitivity reagents (Thermo Fisher) and Agilent Bioanalyser 2100 high sensitivity DNA chips (Agilent). For Illumina sequencing, an equimolar pool of 57 samples at 4 nm was prepared and loaded into a 500-cycle Miseq Reagent kit v2 (Illumina). The fastq files generated by the MiSeq were manipulated using the Illumina BaseSpace Labs FASTQ Toolkit to remove adapter sequences and small-read sequences (<32 bp). The BaseSpace Labs BWA aligner was used to create a BAM file. For Nanopore sequencing data, samples were split into 2 pools at 120 ng to be prepared, barcoded, and loaded onto an R9.4.1 flow cell. The raw fast5 data were basecalled and demultiplexed via guppy, and minimap2 was used for alignment to a reference genome to create the BAM files. The BAM files from both sequencing pathways were manually curated in the Integrative Genomics Viewer ⁴⁴  for the presence/absence of variants.

Survival curves were analyzed by Kaplan-Meier log-rank statistics. Time to progression requiring a new treatment from ibrutinib start (TTNT) was defined according to the 2018 iwCLL guidelines,³⁹  and was used as the primary endpoint. Fisher’s exact test was used for categorical variables, Wilcoxon-signed rank, or Mann-Whitney nonparametric tests were used for continuous variables. The statistical correlations were performed by nonparametric Spearman’s rank test. All statistical tests were 2-sided. Statistical significance was defined as P < .05. Analyses were performed using the Statistical Package for the Social Sciences (SPSS) software or Graphpad Prism 8 software.

Patients undergoing treatment with ibrutinib single agent (n = 70) were studied for clinical, phenotypic, and functional outcome. The general characteristics of the patients are described in supplemental Table 2. The median age was 71 years (range, 51-84), and median number of prior therapies was 1 (range, 0-6). Eighteen patients progressed, 9 while on ibrutinib and 9 after ibrutinib discontinuation. SIgM levels and signaling capacity were measured by flow cytometry before start of ibrutinib. Median sIgM levels before ibrutinib start were 65 (5th-95th percentile, 9-610) and median sIgM iCa²⁺ was 47% (5th-95th percentile, 5-99). The majority of these patients had U-CLL (69%). The sIgM levels and iCa²⁺ distinguished this cohort from a cohort of U-CLL and M-CLL at diagnosis because they showed features associated with aggressive disease.¹⁷  In fact, while sIgM levels and iCa²⁺ are higher in U-CLL than M-CLL at diagnosis (P < .001),¹⁷  the pre-ibrutinib sIgM levels and iCa²⁺ analyzed in this cohort at CLL progression were high and not different between U-CLL and M-CLL and significantly higher than at diagnosis in M-CLL (supplemental Figure 1A-B). These data are consistent with features of the aggressive subsets within U-CLL and M-CLL, which have progressed to require treatment.¹⁷ 

Pre-ibrutinib sIgM levels correlated with sIgM iCa²⁺ (r=0.70; P < .0001) (supplemental Figure 1C), reinforcing our previous findings.¹⁰  We used our previously established MFI cutoff of 50 to define CLL with high sIgM (CLL^high) or low sIgM (CLL^low)¹⁷  to perform correlations with the clinical characteristics and outcome of these patients. The MFI of 50 also appeared the best cutoff using receiver operating characteristic (ROC) and Youden’s t test when a new treatment following ibrutinib was used as a state variable. Supplemental Table 2 describes the clinical characteristics of CLL^high and CLL^low. CLL^high had higher sIgM iCa²⁺ than CLL^low (P = .0001). All patients developed a lymphocytosis and had at least a partial response. No significantly different kinetics of the lymphocytosis were observed in the peripheral blood between CLL^high and CLL^low during ibrutinib (supplemental Figure 2).

Univariate analysis of serum biochemistry, clinical, phenotypic, FISH, and immunogenetic characteristics for TTNT is represented in Table 1. We found that high sIgM levels associated with a shorter TTNT (P = .008; Figure 1A). With a median follow-up of 66 months (95% CI, 59-71), only 2/26 (8%) CLL^low had progressed (median TTNT not reached) compared with 16/44 (36%) CLL^high. Also, sIgM iCa²⁺ was associated with a shorter time to progression (P = .022; Figure 1B). However, sIgM levels were a more significant factor than signaling, and the cutoff that best discriminated CLL that received a new treatment from CLL that did not was particularly high (39% iCa²⁺ using ROC and Youden’s t test). This suggested that a strong signaling capacity is required to circumvent BTK blockade in patients. Conversely, U-IGHV, B2M, LDH, CD38, ZAP-70, CD49d, or TP53 status did not associate with significantly different TTNT in this series.

These data revealed a critical link between pretherapy sIgM levels or sIgM iCa²⁺ and the duration of response to ibrutinib.

At the doses of ibrutinib given to patients with CLL, there was complete BTK inhibition in both CLL^high and CLL^low samples.²⁴  Hence, an in vitro dose (10 µM) that completely inhibits BTK and potentially other anti-IgM–induced kinases⁴⁵  was used to investigate the effect of ibrutinib on sIgM iCa²⁺ (downstream of BTK) in 13 samples collected before therapy. SIgM iCa²⁺ was significantly reduced in the presence of ibrutinib (P = .0002; Figure 2A). However, the reduction was variable, ranging from 3% to 100%. The percent iCa²⁺ reduction was inversely correlated with pretreatment sIgM levels (r = −0.68; P = .01) and iCa²⁺ (r = −0.71; P = .009; Figure 2B-C). These data indicated that sIgM levels and signaling capacity influenced the effects of ibrutinib in vitro and prompted us to investigate this also in patients during therapy.

We had previously observed that levels of functional sIgM were selectively maintained or increased during ibrutinib therapy in patients.²⁴  In this study, we assessed the effect of anti-IgM induced iCa²⁺ and pERK during ibrutinib therapy in 18 patients, including 12 previously investigated.²⁴  This cohort included 10 CLL^high and 8 CLL^low. All lacked BTK or PLCγ2 mutations (supplemental Figure 3) and had complete pBTK (Y223) inhibition by ibrutinib at the time points investigated (supplemental Figure 4).²⁴  PLCγ2 phosphorylation at site Y1217 also appeared inhibited in 5/5 cases investigated (supplemental Figure 4).

SIgM iCa²⁺ and pERK reduced significantly during ibrutinib therapy compared with pre-ibrutinib (Figure 3; supplemental Figure 5). However, neither iCa²⁺ nor pERK were completely abolished, and the degree of inhibition was variable between individuals. Hence, we investigated if pre-ibrutinib sIgM levels of CLL^high vs CLL^low informed on different residual activity downstream of BTK during ibrutinib. Anti-IgM–induced iCa²⁺ and pERK were inhibited less in CLL^high than CLL^low at all time-points (Figure 4; P < .05 in all cases). When sIgM levels were measured at the individual time points investigated during therapy, the association with the induced iCa²⁺ and pERK remained significant (supplemental Figure 6), confirming that the expressed sIgM remained functional during therapy.

Phosphorylation of AKT at S473 (pAKT) was investigated for changes on the PI3K/AKT pathway in the circulating cells of 13 CLL patients (6 CLL^high and 7 CLL^low). Basal pAKT levels, although similar to pre-ibrutinib in the first week of therapy (P = .3), gradually increased at subsequent time points (P = .03 and P = .04; Figure 5A). Conversely, anti-IgM–induced pAKT was not reduced by ibrutinib therapy (Figure 5B), confirming little influence of ibrutinib on the induction of the PI3K/AKT pathway.

The increase of basal pAKT associated with the sIgM levels. In fact, the increase in pAKT inducibility was observed specifically in the CLL^high and not in the CLL^low patients (Figure 5C). These data indicated that constitutive AKT activity increased with therapy duration, suggesting that the AKT pathway provided a signal, which was rescued by the tumor cells with high sIgM, to survive in the absence of antigen engagement.⁴⁶ 

We have previously shown that sIgM levels increase in patients with CLL during ibrutinib therapy and that they are selectively maintained higher than pretherapy levels during therapy.²⁴  Here, we looked at the dynamics of sIgM expression in 13 patients with CLL before and after discontinuation of ibrutinib therapy. Seven of the 13 patients had discontinued because of CLL progression, and 6 had discontinued ibrutinib because of cardiovascular complications (3 patients), second cancer, cognitive deterioration or infection, respectively.

Following ibrutinib discontinuation, sIgM levels increased significantly compared with pretherapy levels in patients with progressive disease. Instead, sIgM levels remained stable or dropped below pretherapy levels in patients who did not progress (Figure 6A). Only 1 out of 7 patients who progressed converted from CLL^low to CLL^high, whereas the remaining 6 were already CLL^high (Figure 6B). Conversely, none of 6 CLL^low converted to CLL^high. These results suggest that high sIgM levels are selectively maintained or increased in patients who will have progressive disease.

In this study, we found that high sIgM levels are associated with more rapid progression following ibrutinib and confirmed that sIgM signaling strength depends on sIgM expression. SIgM levels and signaling capacity appear critical for CLL behavior, with a balance toward anergy and an indolent clinical course when sIgM is low or toward proliferation/growth and a more aggressive course when sIgM is high.¹⁷  We reinforce this concept in the specific context of ibrutinib therapy, demonstrating that high sIgM levels measured before treatment can inform the duration of response to therapy. Our data highlight the need of a high signaling threshold required to overcome BTK inhibition.

We found that BTK blockade by ibrutinib is bypassed with ERK activation and iCa²⁺ mobilization when sIgM-mediated signaling is high. Previous studies documented that BTK inhibition can be overcome when either mutations of BTK prevent ligation of ibrutinib to the binding pocket³⁰  or when a hypertrophic signal is generated by anti-IgM in the presence of PLCγ2 mutations.²⁹  Our cases had no evidence of BTK or PLCγ2 mutations at the time of the functional analyses, and the complete inhibition of BTK activity suggested full occupancy by ibrutinib. The only association remained with high sIgM-mediated signaling.

During ibrutinib therapy, sIgM function (measured by its ability to activate SYK, which is upstream to BTK) recovers with the level of expression.²⁴  The accompanying iCa²⁺ mobilization and ERK phosphorylation indicate that the pathways downstream to BTK are also still operative. Residual iCa²⁺ mobilization has been observed in BTK-deficient X-linked hypogammaglobulinemia,⁴⁷  and knocking out BTK in B cells does not abolish sIgM-mediated mobilization iCa²⁺.⁴⁸  In these models, sIgM is not anergized, suggesting that a strong signal is required to overcome BTK deficiency or arrest.

Studies in vitro document that sIgM-mediated signaling to AKT and PKCδ operates via SYK in CLL, and this is unlikely to be affected by BTKi.^49-51  Our data in patients show that the AKT pathway remains open for stimulation during ibrutinib therapy. However, they also document a gradual increase in basal pAKT. This observation is in line with studies in which constitutive AKT activity recovered in the tumor cells of patients receiving therapy.⁵²  The PI3K/AKT pathway is central for the maintenance of tonic signals in resting tumor and normal B cells.^51,53,54  In CLL, AKT activation promotes cell survival by induction of the anti-apoptotic Mcl-1, whereas basal AKT activation is sustained by SYK.^51,55  AKT has also shown to be a regulator of cell cycle progression, and the intensity of AKT activation has been related to the proliferative response capacity in CLL.⁵⁶  Here, we identify the AKT pathway as one of the possible routes for CLL cells counteracting BTK blockade during ibrutinib. In lymphoma, in which sIgM is high, inactivating lesions of BTK are accompanied by increased AKT activity,⁵⁷  suggesting that AKT is a rescue pathway when BTK is not operative. Therefore, a combinatorial therapeutic approach with ibrutinib and AKT/PI3Ki might be required when sIgM is high.

While expression of sIgM was relatively low and stabilized in patients who did not progress, there was a further sIgM increase in patients who progressed and after ibrutinib therapy discontinuation. In the absence of ibrutinib, CLL cells will be exposed to tissue-based antigen again and subject to sIgM downmodulation by endocytosis.²⁴  Thus, the sIgM increase identifies a heavily selected subpopulation with high sIgM expression that will incorporate those cells with high levels of CXCR4.⁵⁸  This subgroup of cells is expected to be particularly dangerous, being therefore equipped to migrate to tissue and receive proliferative stimuli.⁵⁸ 

These data provide the phenotypic and functional characteristics of the tumor CLL cells surviving ibrutinib therapy in patients. They set a rationale for the need for combinatorial therapeutic approaches to target these cells, which have been heavily selected and appear to carry more dangerous potentials than the tumor population existing prior to the start of single-agent ibrutinib.

The authors are grateful to the Faculty of Medicine Tissue Bank (Cancer Sciences, University of Southampton) for the processing and storage of the primary CLL specimens.

This work was funded by Blood Cancer UK (grant 18009), the Eyles Cancer Immunology Fellowship, Cancer Research UK (ECRIN-M3 accelerator award C42023/A29370, Southampton Experimental Cancer Medicine Centre grant C24563/A15581, Cancer Research UK Southampton Centre grant C34999/A18087, program C2750/A23669, and BTERP project C36811/A29101), and the Czech Science Foundation (20-02566S).

Contribution: G.C. performed research, analyzed and interpreted data, and wrote the manuscript; S.D., L.O., I.S., L.d.R., I.T., and A.D. collected samples and acquired and analyzed data; E.A.M., O.B., and B.S. collected and analyzed clinical data; J.F., H.P., S.B., L.S., and L.-A.S. performed and interpreted the mutational status analysis; M.M., J.C.S., S.L., K.N.P., and P.G. contributed to the interpretation and discussion of the data; F.K.S. supervised research and interpreted data; F.F. designed the study, supervised research, analyzed and interpreted data and wrote the manuscript; and all authors reviewed and approved the manuscript.

Conflict-of-interest disclosure: The authors declare no competing financial interests.

Correspondence: Francesco Forconi, Cancer Sciences, University of Southampton, Cancer Research UK Centre, Somers Cancer Research Building, MP824, Southampton General Hospital, Southampton, SO16 6YD, UK; e-mail: f.forconi@soton.ac.uk.