Abstract
Preclinical studies suggest SYK and JAK contribute to tumor-intrinsic and microenvironment-derived survival signals. The pharmacodynamics of cerdulatinib, a dual SYK/JAK inhibitor, and associations with tumor response were investigated.
In a phase I dose-escalation study in adults with relapsed/refractory B-cell malignancies, cerdulatinib was administered orally to sequential dose-escalation cohorts using once-daily or twice-daily schedules. The study enrolled 8 patients with chronic lymphocytic leukemia (CLL)/small lymphocytic lymphoma (SLL), 13 with follicular lymphoma, 16 with diffuse large B-cell lymphoma (DLBCL), and 6 with mantle cell lymphoma. Correlation of tumor response with pharmacodynamic markers was determined in patients with meaningful clinical responses.
Following cerdulatinib administration, complete SYK and JAK pathway inhibition was achieved in whole blood of patients at tolerated exposures. Target inhibition correlated with serum cerdulatinib concentration, and IC50 values against B-cell antigen receptor (BCR), IL2, IL4, and IL6 signaling pathways were 0.27 to 1.11 μmol/L, depending on the phosphorylation event. Significant correlations were observed between SYK and JAK pathway inhibition and tumor response. Serum inflammation markers were reduced by cerdulatinib, and several significantly correlated with tumor response. Diminished expression of CD69 and CD86 (B-cell activation markers), CD5 (negative regulator of BCR signaling), and enhanced expression of CXCR4 were observed in 2 patients with CLL, consistent with BCR and IL4 suppression and loss of proliferative capacity.
Cerdulatinib potently and selectively inhibited SYK/JAK signaling at tolerated exposures in patients with relapsed/refractory B-cell malignancies. The extent of target inhibition in whole-blood assays and suppression of inflammation correlated with tumor response. (ClinicalTrials.gov ID:NCT01994382).
The spleen tyrosine kinase (SYK) and Janus kinase (JAK) family members can contribute to both tumor-intrinsic and microenvironment-derived survival signals, promoting cancer cell growth in certain B-cell malignancies. The dual inhibitor of SYK and JAK, cerdulatinib, was therefore developed and investigated in a phase I dose-escalation study. To further explore the mechanism of action of cerdulatinib antitumor activity in patients with relapsed/refractory B-cell malignancies, multiple pharmacodynamic measures of SYK/JAK inhibition were correlated with tumor response. These data are presented herein. A phase IIa study is ongoing to confirm safety and antitumor activity of cerdulatinib in additional patients with B- or T-cell malignancies.
Introduction
Normal B-cell development depends on signals originating from the B-cell antigen receptor (BCR) and various costimulatory cytokines; importantly, these cell signaling networks may also cooperate to support the growth and survival of subsets of B-cell malignancies. This was originally observed with diffuse large B-cell lymphoma (DLBCL) cell lines, which demonstrate a reliance on spleen tyrosine kinase (SYK; refs. 1, 2) and Janus kinase (JAK; ref. 3) signaling for survival. Selective inhibition of SYK additionally antagonizes BCR survival signals in primary chronic lymphocytic leukemia (CLL) and blocks tumor cell secretion of chemokine ligands 3 and 4, which facilitate recruitment of accessory cells to the tumor microenvironment (4, 5). Proinflammatory cytokines are elevated in patients with CLL (6) and are of predictive value for disease outcome (7–9). IL2, IL4, and interferon alpha enhance CLL survival in vitro by upregulation of BCL-2 family members (10–13). By this mechanism, IL4 protects cultured CLL cells from death induced by fludarabine and chlorambucil (14). Moreover, IL4 promotes surface IgM expression and restores BCR signaling capacity in CLL cells, which is associated with resistance to idelalisib and ibrutinib in vitro (15–17).
Study of the tumor microenvironment of follicular lymphoma (FL) also suggests an important IL4 signaling axis that is critical for survival. In contrast to unaffected nodes, lymph nodes from patients with FL have greater numbers of follicular helper T cells that express high levels of IL4 (18). A subpopulation of patients with FL present with activating mutations to STAT6; these mutations prolong the transcription factor's engagement with its promoter and overall retention in the nucleus, again suggesting FL tumors rely on IL4 signaling (19). Finally, subsets of DLBCL cell lines, particularly of the activated B-cell–like subtype, appear to generate an autocrine cytokine survival signal by secreting IL6 and/or IL10, resulting in activation of the JAK/STAT3 signaling pathways (20). IL6 has been demonstrated in multiple hematologic malignancies to promote STAT3 activation and BCL-2 family upregulation as a means of promoting survival (21).
Cerdulatinib is a small-molecule ATP-competitive kinase reversible inhibitor that dually targets SYK and JAK family members, sparing inhibition of JAK2 (22). Cerdulatinib distinguishes itself from selective BCR pathway inhibitors by this dual mechanism of action, simultaneously suppressing survival signals originating from the BCR and cytokine receptors. The proposed mechanism of action of cerdulatinib is illustrated in Fig. 1. In preclinical whole-blood assays, the agent was found to inhibit BCR/SYK, IL2, IL4, and IL6 signaling pathways with submicromolar half-maximal inhibitory concentrations (IC50s; ref. 22). In a mouse model of chronic BCR stimulation, cerdulatinib blocked B-cell activation and splenomegaly following oral dosing and achieved statistically significant anti-inflammatory effects at lower plasma concentrations compared with SYK-selective inhibition in the rat collagen-induced arthritis model (22, 23). Cerdulatinib additionally demonstrated potent antitumor activity in DLBCL cell lines (24) and induced apoptosis in primary CLL cells in the presence of stromal cell, BCR, CD40L, or IL4 survival support (16, 24, 25). Antitumor activity was additionally observed in primary CLL cells that carry Bruton tyrosine kinase (BTK) mutations resulting in ibrutinib resistance (25). MCL-1 and BCL-XL were downregulated in CLL cells following treatment with cerdulatinib, which likely contributed to its antitumor activity and observed synergy with the BCL-2 inhibitor venetoclax (16, 25).
Proposed mechanism of action of cerdulatinib. The illustration represents a concept for BCR- and cytokine-mediated survival signals in malignant B cells. Selective BCR pathway inhibitors (entospletinib, ibrutinib, and idelalisib) have demonstrated clinical activity in relapsed/refractory patients with B-cell malignancies, helping to validate BCR signaling as a therapeutic target. SYK is upstream of BTK and PI3K on the BCR signaling pathway, suggesting its inhibition may lead to broader BCR pathway suppression. Cytokine signaling, represented here as IL4, promotes survival in part by upregulation of BCL-2 family members but also by amplification of the BCR signaling pathway. Cerdulatinib distinguishes itself from selective BCR pathway inhibitors by dually suppressing BCR and cytokine signaling pathways.
Proposed mechanism of action of cerdulatinib. The illustration represents a concept for BCR- and cytokine-mediated survival signals in malignant B cells. Selective BCR pathway inhibitors (entospletinib, ibrutinib, and idelalisib) have demonstrated clinical activity in relapsed/refractory patients with B-cell malignancies, helping to validate BCR signaling as a therapeutic target. SYK is upstream of BTK and PI3K on the BCR signaling pathway, suggesting its inhibition may lead to broader BCR pathway suppression. Cytokine signaling, represented here as IL4, promotes survival in part by upregulation of BCL-2 family members but also by amplification of the BCR signaling pathway. Cerdulatinib distinguishes itself from selective BCR pathway inhibitors by dually suppressing BCR and cytokine signaling pathways.
A phase I dose-escalation study with cerdulatinib in patients with relapsed/refractory B-cell malignancies was recently completed, and a dose was selected for the ongoing phase II trial. The data described herein focus on the pharmacodynamic evaluations performed as part of the phase I study. The potency and selectivity of SYK and JAK inhibition by cerdulatinib in whole-blood assays and the impact of cerdulatinib inhibition on serum markers of inflammation following dosing were evaluated. Tumor response to therapy was then correlated with cerdulatinib-mediated inhibition of SYK/JAK signaling and serum markers of inflammation.
Patients and Methods
Study design
This was a two-part, phase I/IIa, multicenter, open-label study. Described herein is the part 1 dose-escalation study of cerdulatinib in adults ages 18 years or older with relapsed or refractory CLL/SLL or B-cell non-Hodgkin lymphoma and failure of at least one prior established treatment regimen. One of the study objectives was to perform correlative analyses between pharmacodynamic parameters and tumor response upon treatment with cerdulatinib. Analyses included determination of the potency and selectivity of cerdulatinib to inhibit SYK/JAK in circulating blood leukocytes, the impact of cerdulatinib on serum markers of inflammation, and the correlation of these effects with cerdulatinib-mediated tumor response.
The study was conducted according to the provisions of the Declaration of Helsinki and the International Conference on Harmonisation–Good Clinical Practice. The study protocol was approved by the institutional review board at each study site, and all patients provided written informed consent before enrollment. The study was registered at ClinicalTrials.gov (NCT01994382).
Study treatment
Sequential dose cohorts received oral cerdulatinib at increasing dose levels until the maximum tolerated dose was identified. The starting dose level was 15 mg orally once daily for 28 days (1 cycle), except on days 2 and 3 of cycle 1, when single-dose pharmacokinetic assessments were performed. If cerdulatinib was well tolerated, patients continued to receive treatment at the discretion of the investigator until discontinuation criteria were met.
Reagents for pharmacodynamic assays
For induction of cell signaling events, the following reagents were procured: goat anti-human IgD (IgG fraction; Bethyl Laboratories Inc.); donkey anti-human IgM F(ab)′2 (Jackson ImmunoResearch); and recombinant human IL2, IL4, IL6, and granulocyte–macrophage colony-stimulating factor (GM-CSF; R&D Systems). Lyse/Fix buffer and BD FACS/Lyse buffer (BD Biosciences) were used to prepare whole blood for intracellular and surface antibody staining, respectively. Cell lineages were identified by flow cytometry by using the following antibodies: mouse anti-human CD3 APC-Cy7 and Alexafluor PE-CF594 conjugates, CD5 Alexafluor 700, CD14 APC, CD16 APC-Cy7, CD19 FITC and PerCP conjugates, CD20 PE-Cy7, and CD56 FITC (BD Biosciences). Intracellular phosphorylation events were detected using rabbit anti-human pSYK Y525/526 PE and pERK Y204 APC (Cell Signaling Technology) and mouse anti-human pAKT S473 PE-CF594, pSTAT3 Y705 PE, pSTAT5 Y695 PE, and pSTAT6 Y641 PE conjugates (BD Biosciences). The CLL surface phenotype was monitored using cell lineage markers combined with mouse anti-human CD69 PE, CD86 PE-CF594, CD5 Alexafluor 700, and CXCR4 PerCP (BD Biosciences).
Bioanalysis, pharmacokinetics, and pharmacodynamics
Blood samples were collected on K2EDTA for the determination of total cerdulatinib plasma concentrations on day 1 prior to dosing and at 0.5, 1, 2, 3, 4, 6, 8, and 12 hours after dose; on day 8 predose and at 2 hours after dose; on day 15 predose; and on day 28 (end of the first treatment cycle) predose and at 0.5, 1, 2, 3, 4, 6, 8, and 12 hours after dose. A liquid chromatography tandem–mass spectrometry (LC-MS/MS) method was developed and validated by Alturas Analytics, Inc. for the determination of cerdulatinib concentration in human plasma. Plasma pharmacokinetic analytical methods are published elsewhere (22). Chromatographic separations were performed over a Phenomenex Synergi Polar-RP column (50 × 2.0 mm, 4 μm; Phenomenex). MS/MS analysis was performed using a Sciex API-4000 triple quadrupole mass spectrometer with a TurboSpray ion source (Applied Biosystems). The peak area of the m/z 394→360 cerdulatinib product ion was measured against the peak area of the m/z 397→363 internal standard product ion. Intra-assay precision (% coefficient of variation [%CV]) and accuracy (% bias) were within 0.8% to 4.5% and −10.1% to 7.8%, respectively, and interassay precision (%CV) and accuracy (% bias) were 2.0% to 3.8% and −7.3% to 6.0%, respectively.
For pharmacodynamic assessments, serial blood samples were drawn into lithium-heparin vacutainer tubes on day 1 predose and again at 0.5, 1, 2, and 4 hours after dose; on day 8 predose and at 2 hours after dose; and day 28 predose. Multiple assays were performed using the day 1 and day 8 blood samples. SYK-mediated BCR signaling in whole blood was measured predose and postdose by stimulating 100 μL whole blood with 2 μL of anti-human IgD (IgG fraction) and 10 μg anti-human IgM for 10 minutes at 37°C, measuring the induction of pSYK Y525/526, pAKT S473, and pERK Y204. Similarly, whole blood was stimulated with 10 ng/mL of IL2 (JAK1/3-dependent), IL4 (JAK1/3-dependent), IL6 (JAK1/TYK2-dependent), or GM-CSF (JAK2-dependent) for 20 minutes, measuring the induction of pSTAT5 Y694 in T cells and natural killer (NK) cells; the induction of pSTAT6 Y641 in B cells, T cells, NK cells, and monocytes; the induction of pSTAT3 Y705 in monocytes, B cells, and T cells; and the induction of pSTAT5 Y694 in monocytes, respectively. The technical details for these assays were previously published (22). With blood samples from the CLL patients only, tumor cell-surface expression of CD5, CD69, CD86, and CXCR4 predose on days 1 and 28 was monitored. The recommended volumes of antibodies were applied directly to 100 μL whole blood and incubated for 1 hour at room temperature. Afterward, 4 mL of BD Lyse/Fix reagent was added to the blood to lyse red blood cells and fix the remaining leukocytes, followed by washing and fluorescence-activated cell sorting (FACS) analysis. For each assay, data were collected using an LSR II instrument (BD Biosciences) and analyzed using FlowJo software (FlowJo LLC). Data were normalized to the induction of each parameter prior to dosing on day 1 to generate the percentage of postdose inhibition.
Whole blood for isolation of serum was collected predose on days 1, 8, and 28 to measure changes in protein markers of inflammation and immune function. Serum samples were snap frozen on dry ice immediately following separation and stored at −80°C. Samples were analyzed using a multiplexed Luminex-based technology (Myriad RBM) with the ImmunoMap (40 analytes) and InflammationMap (45 analytes) platforms. Serum from healthy donors was used as a control.
Peripheral blood B cells were isolated at baseline from CLL patients using the RosetteSep B-cell isolation kit (Stem Cell Technologies) following the manufacturer's protocol. Cells were washed twice in phosphate-buffered saline and snap frozen as a pellet on dry ice. Cell pellets, along with formalin-fixed, paraffin-embedded archival tumor sections, were delivered to the Department of Genomic and Molecular Pathology at the University of Chicago Medical Center, where DNA was isolated using standard methods and subjected to next-generation sequencing using UCM-OncoPlus on Hi-Seq 2500 (26). The UCM-OncoPlus panel consisted of 1,213 tumor-related genes, of which 150 genes are specifically associated with lymphomas and were the focus for this analysis.
Statistical analysis
The data were analyzed using the statistical language R and the accessory packages ggplot2 (27) and drc (28). For cell signaling assays, the percentage of inhibition was determined by normalizing the receptor-induced phosphorylation events mean fluorescent intensity to that of predose receptor-induced mean fluorescent intensity. The percentage of inhibition of cell signaling was related to serum cerdulatinib concentrations to generate pharmacokinetic/pharmacodynamic relationships, or to tumor response using nonlinear regression to a three-parameter log-logistic function with an upper and lower limit set at 100% and 0%. This was a global fit of all available patient data. For the analysis of serum cytokines and protein markers of inflammation, the varying concentration units were converted to a common scale in pg/mL. Values below and in excess of the detection limit were replaced by half the detection limit and the upper limit, respectively. Statistically significant differences between the healthy and patient serum marker at all cycles (cycle 1 day 1, cycle 1 day 8, and cycle 2 day 1) were detected by a paired t test. Logarithms (base of 2) of the ratios of the median concentrations in patient versus healthy normal control serum were used for statistical comparisons. Dimension reduction of expression of the biomarker table in healthy and patient serum was performed using linear discriminant analysis, as is implemented in R. To relate treatment-related changes in serum proteins to tumor response, serum protein concentrations after treatment were normalized to that before treatment and correlated to the maximal tumor response (growth or reduction), representing the ratio of minimal tumor area following treatment and tumor area before treatment. The significance of the relation between maximum tumor response and change in a specific biomarker was evaluated using Spearman rank correlation and P value.
Results
Patient population
Fifty-two patients were enrolled in the phase I dose-escalation portion of the study, and 43 patients were dosed with cerdulatinib. Eight of the dosed patients had CLL/SLL, 13 had FL (including one transformed FL grade 3B), 16 had DLBCL, and 6 had mantle cell lymphoma (MCL). The median age was 67 years (range, 23–85 years). All patients had received prior rituximab, and they were generally heavily pretreated, with a median of four prior regimens (range, 1–10).
Tumor response
Of 43 patients dosed, 37 were evaluated for tumor response by computed tomography scans: 6 patients with CLL/SLL, 13 with FL, 12 with DLBCL, and 6 with MCL. Tumor responses of >50% were observed in patients with CLL/SLL (n = 3 of 6) and FL (n = 4 of 13; Fig. 2A). Limited clinical benefit was observed in DLBCL patients, and no benefit was observed in MCL patients. To investigate pharmacokinetic/pharmacodynamic and tumor response correlates, further evaluations focused on the 19 patients with CLL/SLL and FL.
Tumor response to cerdulatinib and association with exposure. A, Waterfall plots depicting maximum percent change in the sum of tumor volumes as measured by CT scans (y-axis) for each patient on study are shown for each disease subgroup. B, Maximum percent change in tumor volume as it relates to plasma SSCtrough cerdulatinib concentration, expressed as μmol/L (x-axis). Data for CLL/SLL (gray circles) and FL (black circles) are shown. The Spearman correlation coefficient (R) and P values are shown.
Tumor response to cerdulatinib and association with exposure. A, Waterfall plots depicting maximum percent change in the sum of tumor volumes as measured by CT scans (y-axis) for each patient on study are shown for each disease subgroup. B, Maximum percent change in tumor volume as it relates to plasma SSCtrough cerdulatinib concentration, expressed as μmol/L (x-axis). Data for CLL/SLL (gray circles) and FL (black circles) are shown. The Spearman correlation coefficient (R) and P values are shown.
In an effort to understand the relationship between exposure and tumor response, it was noted that steady-state maximal plasma concentration (SSCmax) and the area under the curve did not clearly relate, but preliminary analysis suggested that increased steady-state minimal plasma concentration (SSCtrough) did relate to greater antitumor responses, although the relationship was not statistically significant (Fig. 2B). For the most part, patients with CLL/SLL (gray circles in Fig. 2B) were in lower dose cohorts, achieving SSCtrough of 0.004 to 0.325 μmol/L. At this exposure, 3 of 5 evaluated patients achieved >50% nodal reductions. The 1 CLL patient who achieved higher exposure came on study following an aggressive relapse on ibrutinib and did not respond to cerdulatinib. The FL patients (black circles) appeared to have a different response to cerdulatinib, and tumor responses were more apparent at SSCtrough exposures in the range of 0.73 to 1.3 μmol/L. At this higher exposure range, 2 patients achieved a partial response (including 1 who was a transformed FL patient, grade 3B) and 2 patients achieved a complete response. Lower SSCtrough resulted in two stable diseases and one progressive disease.
Relationship between tumor response and SYK/JAK inhibition
The potency and selectivity for target inhibition following oral dosing in patients using a variety of whole-blood assays was measured. Select pharmacokinetic/pharmacodynamic relationships from the phase I study are presented in Fig. 3A. High-level inhibition of BCR-induced SYK autophosphorylation (pSYK Y525/526) and downstream signaling to ERK (pERK Y204) and AKT (pAKT S473) were observed at tolerated exposures. Similarly, IL2, IL4, and IL6 signaling (JAK1-, JAK3-, and TYK2-dependent) were potently inhibited in a concentration-dependent manner. To demonstrate specificity within the JAK family, GM-CSF stimulations, which induce a JAK2-dependent STAT5 phosphorylation, were performed on patient samples. Consistent with preclinical data, cerdulatinib demonstrated potent inhibition of SYK and JAK family members, sparing JAK2. No inhibition of phorbol myristate acetate–mediated B-cell pERK Y204 was observed, again demonstrating specificity of action. The significance of the curve fit for the pharmacokinetic/pharmacodynamic relationships is represented at the top left of each graph.
Cerdulatinib potently inhibits SYK and JAK in peripheral whole-blood assays. A, The scatter plots depict concentration–response profiles for patient whole-blood SYK and JAK signaling assays, as indicated. Percent inhibition relative to predose cell signaling capacity is shown on the y-axis. Cerdulatinib total plasma concentration (μmol/L) is on the x-axis. The cell signaling pathways activated are indicated within each figure. Data were analyzed via a nonlinear regression to a three-parameter log-logistic function with an upper and lower limit set at 100% and 0% using the R software package, representing a global fit of all available patient data. The significance of the concentration–response relationship was assessed by ANOVA and shown within each plot. B, IC50 values in μmol/L (and 95% CIs) for whole-blood cell signaling assays are shown as a forest plot. The total number of data points used for these calculations, as well as number of patients from which data were obtained, are indicated. The Hill slope and 95% CIs are also shown. The nature of the assay is indicated to the left of the plot; total plasma cerdulatinib concentration is shown in μmol/L on the x-axis. The dashed vertical line at 1 μmol/L approximates the SSCtrough to be achieved with the phase II dose. C, The scatter plots demonstrate correlations observed between maximum percent inhibition (y-axis) of cell signaling pathways and maximum percent change in sum of tumor volume (x-axis). Negative values on the x-axis represent tumor shrinkage. Data were fit to a linear regression. The Spearman correlation coefficient (R) and P values are shown.
Cerdulatinib potently inhibits SYK and JAK in peripheral whole-blood assays. A, The scatter plots depict concentration–response profiles for patient whole-blood SYK and JAK signaling assays, as indicated. Percent inhibition relative to predose cell signaling capacity is shown on the y-axis. Cerdulatinib total plasma concentration (μmol/L) is on the x-axis. The cell signaling pathways activated are indicated within each figure. Data were analyzed via a nonlinear regression to a three-parameter log-logistic function with an upper and lower limit set at 100% and 0% using the R software package, representing a global fit of all available patient data. The significance of the concentration–response relationship was assessed by ANOVA and shown within each plot. B, IC50 values in μmol/L (and 95% CIs) for whole-blood cell signaling assays are shown as a forest plot. The total number of data points used for these calculations, as well as number of patients from which data were obtained, are indicated. The Hill slope and 95% CIs are also shown. The nature of the assay is indicated to the left of the plot; total plasma cerdulatinib concentration is shown in μmol/L on the x-axis. The dashed vertical line at 1 μmol/L approximates the SSCtrough to be achieved with the phase II dose. C, The scatter plots demonstrate correlations observed between maximum percent inhibition (y-axis) of cell signaling pathways and maximum percent change in sum of tumor volume (x-axis). Negative values on the x-axis represent tumor shrinkage. Data were fit to a linear regression. The Spearman correlation coefficient (R) and P values are shown.
Pharmacokinetic/pharmacodynamic relationships were evaluated to estimate IC50 values, which are depicted as a forest plot in Fig. 3B. The vertical dotted line intersecting with 1 μmol/L approximates the SSCtrough expected to be achieved with the phase II dose of 35 mg b.i.d. Measures of BCR signaling were inhibited with IC50 values in the 0.39 to 0.73 μmol/L range. Depending on the cell lineage, measures of JAK/STAT signaling were inhibited with IC50 values in the 0.27 to 1.11 μmol/L range. The slope of each curve fit is also depicted along with the 95% confidence interval (CI; Fig. 3B). The relationship between maximum percent change in tumor volume and percent inhibition of SYK and JAK signaling pathways in the whole-blood assays is presented in Fig. 3C. Inhibition of BCR-induced SYK autophosphorylation (pSYK Y525/526) significantly correlated with tumor response, with an R of −0.80 (P = 0.02). Four of the 5 patients in whom pSYK Y525/526 was inhibited by >50% achieved a meaningful clinical response. Inhibition of B-cell and monocyte (data combined) IL4 also significantly correlated with tumor response (R of −0.59; P = 0.006), although several patients with high-level IL4 inhibition had marginal reductions in tumor size. There was no significant relationship between inhibition of the remaining cell signaling pathways and tumor response (Fig. 3C).
Relationship between markers of inflammation and tumor response
Cancer patients often present with underlying inflammation that can be detected by serum protein profiling. To evaluate this and determine the effect of cerdulatinib on systemic inflammation, serial serum samples collected from patients were analyzed for serum proteins associated with inflammation and general immune function. Serum concentrations of 90 proteins were determined, of which 31 were consistently below limits of detection. Figure 4 represents an analysis of the remaining 59 proteins for which measurements were possible. At baseline (cycle 1 day 1), the inflammatory profiles of the different patient cohorts were quite divergent and could be distinguished from each other and healthy controls by cluster analysis (Supplementary Fig. S1), lending confidence to the validity of the data.
Significant inhibition of serum markers of inflammation in patients following treatment with cerdulatinib. Serum protein concentrations from patients were normalized to healthy control average (n = 6), then averaged and plotted as Log2 (y-axis). Red bars indicate significant differences from patient versus healthy control, and black bars indicate lack of significance, as determined by t test. Data for predose cycle 1 day 1 (C1D1), postdose cycle 1 day 8 (C1D8), and postdose cycle 2 day 1 (C2D1) are presented in descending rows. The serum protein measured is indicated on the x-axis for CLL/SLL (A) and FL (B) patients.
Significant inhibition of serum markers of inflammation in patients following treatment with cerdulatinib. Serum protein concentrations from patients were normalized to healthy control average (n = 6), then averaged and plotted as Log2 (y-axis). Red bars indicate significant differences from patient versus healthy control, and black bars indicate lack of significance, as determined by t test. Data for predose cycle 1 day 1 (C1D1), postdose cycle 1 day 8 (C1D8), and postdose cycle 2 day 1 (C2D1) are presented in descending rows. The serum protein measured is indicated on the x-axis for CLL/SLL (A) and FL (B) patients.
In both CLL/SLL and FL patients, the common serum markers that were significantly elevated at baseline relative to healthy normal control serum were von Willebrand factor (vWF), MIP3β, C-reactive protein (CRP), HCC4, β2M, VCAM1, IL18, TNFR2, IP10, and MIG (Fig. 4). By cycle 1 day 8 and cycle 2 day 1, several of these markers lost significance relative to healthy controls, indicating a normalization of inflammation with cerdulatinib treatment. For the most part, reductions in serum markers of inflammation occurred within the first 8 days of therapy with cerdulatinib, by cycle 1 day 8. Cerdulatinib significantly reduced MIP3β, CRP, and VCAM1 in both CLL/SLL and FL patients. Serum markers that were elevated at baseline but unaffected by cerdulatinib in both patient groups were HCC4, IL18, and MIG. Additionally, for CLL/SLL patients only, thrombospondin, BDNF, DKK1, myeloperoxidase, CD40, RANTES, MMP9, and ENA78 were significantly reduced at baseline when compared with healthy controls (Fig. 4A). Of these, myeloperoxidase and CD40 serum levels were normalized with cerdulatinib treatment. For FL patients only, CD40 and BDNF were reduced at baseline relative to healthy controls, the latter being normalized with cerdulatinib treatment (Fig. 4B).
The serum protein data are also presented in Supplementary Fig. S2 for CLL/SLL, FL, and DLBCL/MCL patients to represent changes in serum concentration of these proteins over time with treatment as log2 pmol/L. Consistent with the antitumor responses across these malignancies, limited inhibition of serum inflammation markers was observed in the aggressive lymphomas (DLBCL/MCL; Supplementary Fig. S2C).
Next, tumor response was related to percent inhibition of serum markers of inflammation. Significant correlations between tumor response in CLL/SLL patients and reductions in serum CRP and IP10 were observed (Fig. 5A). In FL patients, significant correlations existed between tumor response and inhibition of MIP3β, MDC, IP10, β2M, and APRIL (Fig. 5B). Baseline serum concentrations of these proteins did not predict tumor response to treatment. These data demonstrate that cerdulatinib can modulate systemic inflammation, which for several proteins was associated with tumor response.
Inhibition of serum markers of inflammation significantly correlates with tumor response in FL and CLL/SLL patients. Percent inhibition of serum protein concentration on (normalized to predose) is presented on the y-axis. Maximum percent change in the sum of tumor volume is plotted on the x-axis. Data depict the significant correlations observed for CLL/SLL (A) and FL (B) patients. The Spearman correlation coefficient (R) and P values are shown.
Inhibition of serum markers of inflammation significantly correlates with tumor response in FL and CLL/SLL patients. Percent inhibition of serum protein concentration on (normalized to predose) is presented on the y-axis. Maximum percent change in the sum of tumor volume is plotted on the x-axis. Data depict the significant correlations observed for CLL/SLL (A) and FL (B) patients. The Spearman correlation coefficient (R) and P values are shown.
Cerdulatinib-mediated lymphocytosis
Treatment-related increases in blood absolute lymphocyte counts (ALC) occurred in both CLL/SLL and FL patients (Fig. 6A). Five patients with CLL/SLL remained on treatment long enough to monitor ALC over time, which was elevated 0.3- to 10-fold relative to pretreatment. Treatment-related changes in tumor cell-surface activation and homing markers were additionally evaluated in 4 of these patients (Fig. 6B). FACS analysis of these cells prior to treatment cycle 1 day 1 and again at cycle 2 day 1 revealed decreased expression of the surface activation markers CD69 and CD86 (P = 0.006; Supplementary Fig. S3), as well as reduced CD5 expression (a negative regulator of BCR signaling; ref. 30) and enhanced CXCR4 expression (responsible for cell homing to lymphoid tissues; ref. 30) in some, but not all, patients.
Cerdulatinib induces lymphocytosis and alters CLL tumor cell-surface phenotype. A, ALC normalized to predose is presented on the y-axis for CLL (top) and FL (bottom) patients. The cycle and day numbers are presented on the x-axis. B, Cell-surface phenotype changes observed from predose cycle 1 day 1 (C1D1) to cycle 2 day 1 (C2D1) for CLL patients. The FACS scatter plots shown are from a CD3-negative CD19-positive B-cell gate. Cerdulatinib-mediated changes in CD69 and CD86 cell-surface activation markers and changes in CXCR4 and CD5 are shown. Maximum nodal reduction for each patient is shown at the top of the plots.
Cerdulatinib induces lymphocytosis and alters CLL tumor cell-surface phenotype. A, ALC normalized to predose is presented on the y-axis for CLL (top) and FL (bottom) patients. The cycle and day numbers are presented on the x-axis. B, Cell-surface phenotype changes observed from predose cycle 1 day 1 (C1D1) to cycle 2 day 1 (C2D1) for CLL patients. The FACS scatter plots shown are from a CD3-negative CD19-positive B-cell gate. Cerdulatinib-mediated changes in CD69 and CD86 cell-surface activation markers and changes in CXCR4 and CD5 are shown. Maximum nodal reduction for each patient is shown at the top of the plots.
Next-generation sequencing
Genetic abnormalities were monitored by next-generation DNA sequencing using UCM-OncoPlus, a panel of 1,213 cancer-related genes (27). Fresh tumor samples were obtained from the peripheral blood of 7 CLL patients prior to dosing with cerdulatinib, as well as archival tumor biopsies obtained from 4 FL patients and one MCL patient. A list of the mutations observed is detailed in Supplementary Table S1. Clinical activity was observed in CLL patients bearing mutations to NOTCH1, ATM, TP53, and KRAS, which are among the most frequently observed mutations in this disease (31). One of the responding patients bore a 17p deletion encompassing the TP53 locus. Importantly, 2 patients with ibrutinib-relapsed CLL who progressed within the first cycle of therapy with cerdulatinib uniquely shared 3 mutations in common: TP53, EP300, and BTKC481S. Genetic correlates in FL were more limited, with data on 4 patients, 3 of whom had stable disease as best response to therapy and 1 who had progressed. The 2 best responding patients for whom genetic information was available bore mutations to ZMYM3, KMT2D, FAT4, BCL-2, BCL-6, and STAT6. Lastly, clinical activity was observed in a transformed FL 3B patient who presented with increased MYC, BCL-2, and BCL-6 expression by IHC (“triple-hit” lymphoma).
Discussion
The primary focuses of the work presented herein were to determine if the clinical data were consistent with the proposed mechanism of action of cerdulatinib as a dual SYK/JAK inhibitor and whether there was any evidence that SYK/JAK inhibition was related to tumor response. SYK and JAK signaling pathways are critical mediators of inflammatory responses. Therefore, the impact of cerdulatinib on these signaling pathways was directly measured via pFlow and indirectly measured by assessing serum markers of inflammation in patients. The 2 measures were then correlated with tumor response. We demonstrated significant and positive correlations between cerdulatinib exposure and extent of SYK and JAK pathway inhibition, as well as showed that cerdulatinib reduced systemic inflammation, as would be expected of an SYK/JAK inhibitor. Hence, the drug behaved as expected from a mechanistic point of view. Importantly, inhibition of the SYK/JAK signaling pathways, particularly in circulating B cells, significantly correlated with antitumor response, as did suppression of various markers of systemic inflammation. These data lend support to the conclusion that, not only does cerdulatinib inhibit SYK and JAK, but that inhibition of these pathways is at a minimum related to tumor regression in patients with relapsed/refractory B-cell malignancies.
The potency and specificity of SYK/JAK inhibition in whole blood following oral dosing of cerdulatinib in patients largely recapitulated the ex vivo experience using healthy normal whole blood (22). Phosphorylation events more distal to the BCR, namely, pERK Y204 and pAKT S473, appeared to be more potently inhibited relative to the SYK Y525/526 autophosphorylation site, possibly reflecting a threshold for SYK inhibition at which the signaling pathway is shut off. Therefore, the IC50 of cerdulatinib was estimated against BCR signaling to be in the range of 0.25 to 0.53 μmol/L following oral dosing, reflecting the lower and upper limits of the CIs for ERK and AKT. The two assays that were performed to monitor JAK/STAT pathway activation in B cells were IL4-induced and IL6-induced pSTAT6 Y641 and pSTAT3 Y705, respectively. Inhibition of IL4 signaling was quite variable among patients, with an average IC50 of 1.11 μmol/L (CI, 0.63–1.58). IL6 signaling was inhibited with more consistent potency across patients, with an average IC50 of 0.33 μmol/L (CI, 0.25–0.42). Inhibition of BCR-mediated SYK Y525/526 and IL4-mediated pSTAT6 Y641 both significantly correlated with tumor response.
To put these exposures and inhibitory potencies into perspective from preclinical models, cerdulatinib significantly inhibited autoimmune arthritis and autoantibody production at 0.3 μmol/L average steady-state concentration in rats and induced apoptosis and/or cell-cycle arrest in a variety of tumor B-cell lines and primary tumors in the 0.2 to 2 μmol/L range (22, 24, 32). In the dose-escalation study, it was observed that complete inhibition of most SYK and JAK signaling assays was achievable at tolerated exposures. The selected phase II dose of 35 mg twice daily targets an SSCtrough of approximately 0.8 to 1 μmol/L, with a low peak-to-trough ratio. Hence, SYK and JAK signaling pathways are expected to be inhibited by >50% to 90% throughout the day. Exposure and extent of target inhibition in the actual tumor microenvironment is unknown, and may therefore be higher or lower than what was estimated from the whole-blood assays.
There is a well-established link between inflammation and cancer, as reviewed elsewhere (33). As expected, the patients enrolled in this trial presented with varying degrees of inflammation. Cerdulatinib rapidly (within first week of therapy) and significantly reduced the serum concentrations of several protein markers of inflammation. Moreover, reductions of several of these serum proteins with time on therapy significantly correlated with tumor response. In FL, where a larger number of patient samples were available, these reductions in serum proteins included APRIL, β2M, IP10, MDC, and MIP3β. APRIL is involved in normal B-cell development, but has also been demonstrated to promote B-cell tumorigenesis in mouse models, presumably by preventing apoptosis (34). Elevated serum β2M was significantly associated with poor overall survival in patients with NK and T-cell lymphoma (35) and B-cell non-Hodgkin lymphoma (36). IP10 (CXCL10) is a chemoattractant for multiple leukocyte lineages, directing tissue homing of macrophages, T cells, and dendritic cells among others. This chemokine ligand has also demonstrated prognostic value in various cancer types (37). Of note, cerdulatinib-mediated reduction in serum IP10 levels correlated with tumor response in both FL and CLL patients. MDC and MIP3β (CCL19) also serve as chemoattractants for leukocytes (38–40). CCL19 binds to CCR7, a key chemokine receptor controlling T-cell homing to secondary lymphoid organs. These data suggest that one mechanism by which cerdulatinib may exert antitumor activity is by disruption of key signals responsible for the organization of the tumor microenvironment.
As part of this study, we additionally attempted to gain some insight into the relationship between tumor genetic aberrations and tumor response by cerdulatinib. Fresh tumor biopsies were requested on a voluntary basis from all patients; however, none were collected. However, formalin-fixed archival tumor sections were received from 4 patients with FL and 1 patient with MCL, in addition to pretreatment isolation of circulating CLL tumors from whole blood. DNA was prepared from these materials and subjected to next-generation sequencing. This limited data set only offered hints into a pharmacogenomics relationship. The 2 CLL patients who did not respond to cerdulatinib had relapsed on ibrutinib prior to study entry and presented with an aggressive disease. Both of these patients carried missense mutations to EP300 (Ser697Arg in one and Cys1247Phe in the other), TP53 (Glu285Lys in one and Arg273Cys in the other), and BTK (Cys481Ser in both). In preclinical studies, cerdulatinib demonstrated potent antitumor activity against CLL cells isolated from ibrutinib-relapsed patients bearing the Cys481Ser and Thr316Arg mutations (25), which are known to lead to ibrutinib resistance (41–43). Additional patients with similar genetics will be required to draw any conclusions regarding cerdulatinib clinical activity in this unique group. Interestingly, a patient with FL who relapsed following five prior therapies, including progressive disease on bendamustine/rituximab, progressive disease on ibrutinib, and a less than 4-month response to R-CHOP as his/her last three therapies, contained a novel mutation to STAT6 (Ser86Ala), which is contained within the STAT dimerization domain. This patient achieved SSCtrough to SSCmax cerdulatinib serum concentrations of 0.32 to 0.38 μmol/L, which is considerably lower than our phase II exposure, and yet demonstrated a 20% reduction in tumor bulk with >6-month durability of response. The functional impact of this mutation has not yet been verified, but mutations to other domains of this transcription factor were found to result in hyperresponsiveness to the IL4 signaling pathway in FL patients (19). Additional mutations associated with the 3 CLL patients with greatest nodal reductions were REL (Ile354Thr), a component of NFkB2; TET2 (Met66Leu), a dioxygenase that regulates DNA methylation status; A20 (Gln150Arg), an inhibitor of NFkB; and HIST1H1E (Ala47Val). We are continuing these efforts in the phase II study to bring more clarity to these data.
In summary, data from the phase I dose-escalation study identified a phase II dose that was well tolerated, achieved drug levels that resulted in high-level inhibition of SYK and JAK signaling pathways, and demonstrated evidence of antitumor activity. Inhibition of relevant signaling pathways and serum markers of inflammation both correlated with tumor response. It is important to note the key limitation of this study, however. All correlates of tumor response were evaluated only in the CLL/SLL and FL patients, totaling 19 patients in all. Moreover, within each of these analyses, we did not have data on all patients, hence any outliers that may be present in the data could have a disproportionate impact on the interpretation. Although the appropriate statistical analyses were performed, we cannot be certain that the observations reported here will be exactly reproduced in a larger study. We will continue to build upon these observations as the clinical development of cerdulatinib continues. The phase II study is ongoing to establish safety and efficacy of cerdulatinib in an additional 20 to 40 CLL/SLL and FL patients. A third cohort of relapsed/refractory peripheral T-cell lymphoma patients was recently opened, based on the observation of SYK expression and its potential role as an oncogenic driver in this disease (44–47).
Disclosure of Potential Conflicts of Interest
G.P. Coffey, J. Feng, M. Birrell, J.M. Leeds, and K. Der have ownership interests (including patents) at Portola Pharmaceuticals. S. Kadri is a consultant/advisory board member for GLG consultants. J. Segal reports receiving speakers bureau honoraria from Bristol-Myers Squibb and reports receiving commercial research support from AbbVie. Y.L. Wang reports receiving commercial research support from Portola. No potential conflicts of interest were disclosed by the other authors.
Authors' Contributions
Conception and design: G.P. Coffey, A. Pandey, G. Michelson, J.T. Curnutte, P.B. Conley
Development of methodology: G.P. Coffey, J. Segal, Y.L. Wang, G. Michelson
Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): G.P. Coffey, J. Segal, Y.L. Wang, G. Michelson
Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): G.P. Coffey, J. Feng, A. Betz, M. Birrell, J.M. Leeds, S. Kadri, J. Segal, Y.L. Wang, G. Michelson, J.T. Curnutte, P.B. Conley
Writing, review, and/or revision of the manuscript: G.P. Coffey, A. Pandey, M. Birrell, J.M. Leeds, P. Lu, Y.L. Wang, G. Michelson, J.T. Curnutte, P.B. Conley
Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): G.P. Coffey, J. Feng, M. Birrell, K. Der
Study supervision: G.P. Coffey, J.T. Curnutte
Other (discovered cerdulatinib): A. Pandey
Acknowledgments
This study was sponsored and financially supported by Portola Pharmaceuticals, Inc. We thank the patients who participated in this study, as well as their families and the research staff at each site. Preparation of the manuscript was supported by Iwona Bucior, PhD, of Portola Pharmaceuticals, Inc. Editorial support was provided by Kimberly Brooks, PhD, CMPP, of SciFluent Communications.
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.