Venetoclax dose escalation rapidly activates a BAFF/BCL-2 survival axis in chronic lymphocytic leukemia

Abstract

Venetoclax, a first-in-class BH3 mimetic drug that targets B-cell lymphoma-2 (BCL-2), has improved the outcomes of patients with chronic lymphocytic leukemia (CLL). Early measurements of the depth of the venetoclax treatment response, assessed by minimal residual disease, are strong predictors of long-term clinical outcomes. However, there are limited data on the early changes induced by venetoclax treatment that might inform strategies to improve responses. To address this gap, we conducted longitudinal mass cytometric profiling of blood cells from patients with CLL during the first 5 weeks of venetoclax monotherapy. At baseline, we resolved CLL heterogeneity at the single-cell level to define multiple subpopulations in all patients based on proliferative, metabolic, and cell survival proteins. Venetoclax induced a significant reduction in all CLL subpopulations and caused rapid upregulation of the prosurvival BCL-2, BCL-extra large, and mantle cell lymphoma-1 proteins in surviving cells, which had reduced sensitivity to the drug. In mouse models, the venetoclax-induced elevation of survival proteins in B cells and CLL-like cells that persisted was recapitulated, and genetic models demonstrated that extensive apoptosis and access to the B-cell cytokine, B-cell activating factor (BAFF), were essential. Accordingly, in patients with CLL who were treated with venetoclax or the anti-CD20 antibody obinutuzumab there was marked elevation in BAFF and an increase in prosurvival proteins in leukemic cells that persisted. Overall, these data highlight the rapid adaptation of CLL cells to targeted therapies through homeostatic factors and support cotargeting of cytokine signals to achieve deeper and more durable long-term responses.

Graphical abstract

Figure 1.

Mass cytometric analysis resolves CLL heterogeneity. (A) Schematic representation of the experimental strategy with PB samples from 20 patients with CLL collected at screening and during weekly venetoclax dose escalation. (B) Lymphocyte counts (cells per liter) during venetoclax dose escalation in each patient. (C) UMAP projection of PB cells (subsampling of 2000 cells per each sample) from patients with CLL (n = 18 VEN patients plus n = 2 VEN+IBR patients, 6 timepoints) and healthy donors colored by the following clusters: B cells (C1), CLL cells (C2-5), T cells (C6-C12), NK cells (C14), and myeloid cells (C15:C30). (D) UMAP plots colored by expression levels of the indicated markers used to identify the major immune cell populations. (E) UMAP plots of a patient with CLL (CLL27) and healthy donor with B/CLL clusters circled. IBR, ibrutinib; NKT, natural killer T; Scr, screening.

Figure 2.

Intrapatient and interpatient heterogeneity of CLL cell subpopulations. (A) UMAP of reclustered healthy B/CLL cells (from Figure 1C) from patients and healthy controls. (B) UMAP plots colored by median protein expression of CD20, pS6, and MCL-1, distinguishing between healthy B cells (CD20^high pS6^high MCL-1^low) and CLL cells (CD20^lowpS6^lowMCL-1^high). (C) UMAP plots colored by protein expression level of IgM, distinguishing IgM^high and IgM^low B cells. (D) UMAP plots colored by amounts of pRB, TACI, BAX, and pH2AX, thereby distinguishing between 2 minor subpopulations of CLL cells, namely pRB^highTACI^highBAX^high (cluster 2.9) and pH2AX^high (cluster 2.10). (E) UMAP plots colored by protein expression of CD5, CXCR4, and BCL-2, thereby distinguishing 2 main subpopulations of CLL cells, namely CD5^highCXCR4^lowBCL-2^high (cluster 2.6) and CD5^lowCXCR4^highBCL-2^v.high (cluster 2.7). (F) Representative 2-dimensional plot of CXCR4 vs CD5 expression showing CD5^highCXCR4^low and CD5^lowCXCR4^high CLL cell populations from patient CLL20. (G) Violin plots showing the mean intensities of BCL-2, BCL-XL, MCL-1, pH3, and IgM of old and new CLL cells that have low, medium, and high levels of CXCR4 protein. (H) Frequency of each cluster in the PB of patients before venetoclax treatment. Student paired t test was used. ∗∗∗P < .005; ∗∗∗∗P < .001. BAX, BCL-2 associated protein X; HC, healthy control; pH2AX, phosporylated histone H2AX; pRB, retinoblastoma protein; TACI, transmembrane activator and CAML interactor.

Figure 3.

Venetoclax dose-dependent increase in BCL-2 protein detected in CLL cells. (A) Mean proportions of different CLL clusters from all patients in the cohort at the indicated timepoints during treatment. (B) Graph of the results of the LDA of data from the CLL clusters from patient CLL19 across increasing doses of venetoclax. The magnitude of the projection of each marker direction onto the plane of the first 2 LDA components is represented by blue dots. This has a maximum value of 1 if the marker direction lies in the planes of the first 2 LDA components. To reveal the markers that drive the changes, the markers are ordered on the x-axis by the magnitude of this projection. The green shaded area represents the distribution of ranking curves of randomly oriented planes (see “Methods”). (C) Summary of markers ranked by contributions to the first 2 LDA components in all patients for venetoclax dose escalation (seen in panel B for patient CLL19). The higher the ranking of a marker, the more it is affected by venetoclax dose changes with 20 being the highest ranked marker and 0 the lowest. (D) Representative histograms showing BCL-2 protein expression of curated pairs of patients with CLL at screening (blue) and after 200 mg VEN treatment (red) in patient CLL13, CLL19, and CLL22 as measured by flow cytometry. The distribution of BCL-2 levels in each sample is represented by the box and whisker plots in the lower panel. The box represents the 25th, 50th, and 75th percentiles of the population with x marking the mean and the whiskers representing the minimum and maximum values (excluding outliers). (E) Fold change in BCL-2, MCL-1, and BCL-XL protein expression in samples after 200 mg treatment normalized to screening in individual patients. P values were calculated using raw data and Student 2-tailed paired t test. (F) Representative results of venetoclax sensitivity in an in vitro assay of CLL cells from patient CLL19. (G) Summary of the LC50 values of CLL cells at screening and after 200 mg venetoclax treatment. (H) Representative results of venetoclax sensitivity in an in vitro assay in nontransformed CD4⁺ T cells from patient CLL19. (I) Summary of the LC50 values of CD4⁺ T cells at screening and after 200 mg venetoclax treatment. (J) Schematic representation of the hypothesis. CLL cells express a range of BCL-2 amounts, from BCL-2⁺ to BCL-2⁺⁺⁺. CLL cells with relatively lower levels of BCL-2 (BCL-2⁺) are more sensitive to venetoclax treatment, thereby enriching for CLL cells with higher BCL-2 levels (BCL-2⁺⁺⁺) after venetoclax monotherapy. (K) Distribution of BCL-2 in CLL cells before (Scr) and 1 week after 200 mg dose of venetoclax with a dashed line indicating the threshold distinguishing the top 20% of BCL-2 expressors (BCL-2^top20) and 80% lower BCL-2 expressors (BCL-2^lower80) at screening. This threshold was then applied to each paired profile after 200 mg treatment. (L) Estimated concentration of BCL-2^top20 and BCL-2^lower80 in CLL cells in the blood at screening and after 200 mg venetoclax treatment based on the BCL-2^top20/BCL-2^lower80 threshold assigned in the screening sample. P values calculated using ratio paired t test. For box and whisker plots in panel D, outliers were omitted from the plot with an outlier factor of 1.5. Outliers represented less than 2.5% of the total population in all samples. For panels G and I, Student 2-tailed paired t test was used (patients with missing values were excluded from analysis). For panels E and L, the mean ± standard error of the mean (SEM) is shown. ∗P < .05; ∗∗∗P < .005; ∗∗∗∗P < .001. BIM, BCL-2 interacting mediator of cell death; cMyc, myelocytomatosis oncogene protein; Ikba, NK-κ-B inhibitor α; LC50, lethal concentration 50; ns, not significant; pERK, phosphorylated extracellular signal-regulated kinase; pPLCg, phosphorylated phospholipase C γ; pRb, retinoblastoma protein.

Figure 4.

In vivo modeling demonstrates the elevation of BCL-2 in VEN^surv cells in mice. (A) Schematic representation of the experimental strategy. wt, Bak^–/–Bax^Δcd23, and vav-huBcl-2 mice were treated daily with 100 mg/kg body weight venetoclax or vehicle (a mixture of 60% Phosal 50 propylene glycol, 30% polyethylene glycol 400, and 10% ethanol) for 1 week. (B) Absolute numbers of splenic B cells before and after venetoclax treatment in wt, Bak^–/–Bax^Δcd23, and vav-huBcl-2 mice. (C) Histograms of BCL-2, MCL-1, and BCL-XL protein in CD19⁺CD21⁺IgM⁺ B cells from wt, Bak^–/–Bax^Δcd23, or vav-huBcl-2 mice measured by flow cytometry after vehicle or venetoclax treatment. (D) Geometric mean (± SEM) of mBCL-2, huBCL-2, MCL-1, and BCL-XL protein levels in vehicle- and venetoclax-treated mice of the indicated genotypes. (E) Schematic representation of the Eμ-TCL-1 transgenic mice modeling of CLL cell response to venetoclax. Cohorts of C57BL/6 mice received 5 × 10⁵Eμ-TCL-1 transgenic CLL cells from the same donor, and once they reached 80% leukemic burden in blood, they were treated with vehicle or venetoclax daily for 7 days. (F) Total splenic cell numbers and the proportions of Ki67⁺Eμ-TCL-1 transgenic CLL cells and wt B cells before and after treatment. (G) Histograms of BCL-2, MCL-1, and BCL-XL protein in CLL cells recovered from the spleen of mice treated with vehicle or venetoclax. (H) Quantification of BCL-2, MCL-1, and BCL-XL protein expression in CLL cells recovered from the spleen of mice treated with vehicle or venetoclax. Data from panels B-D are representative of 3 independent experiments with n = 2 to 5 mice per group. Data from panels F-H are representative of 3 independent experiments with n = 5 to 6 mice per group. For all bar graphs, the mean ± SEM are shown, and each symbol represents an individual mouse; Student 2-tailed t test was used. ∗P < .05; ∗∗P < .01; ∗∗∗P < .005; ∗∗∗∗P < .001. hBCL-2, human BCL-2; mBCL-2, murine BCL-2.

Figure 5.

BAFF controls the homeostatic survival response to venetoclax treatment in B cells. (A) Schematic representation of hematopoietic chimeras reconstituted with a mixture of CD45.2⁺, Bak^–/–Bax^Δcd23, and CD45.1⁺ wt BM progenitors, followed by venetoclax treatment. (B) Total numbers of splenic B cells of Bak^–/–Bax^Δcd23 or wt origin recovered from vehicle- and venetoclax-treated chimeras. (C) Histograms of BCL-2, MCL-1, and BCL-XL staining gated on splenic B cells of wt or Bak^–/–Bax^Δcd23 origin recovered from the chimeric recipient mice treated with vehicle or venetoclax. (D) Geometric mean of BCL-2, MCL-1, and BCL-XL protein expression levels in splenic B cells of Bak^–/–Bax^Δcd23 or wt origin recovered from the vehicle- or venetoclax-treated chimeric recipient mice. (E) Serum BAFF from wt and Bak^–/–Bax^Δcd23 mice after 7-day treatment of venetoclax. Data are from 2 experiments with n = 2 to 3. (F) Schematic representation of the experimental design. Purified CD45.2⁺ C57BL/6 B cells from Bak^–/–Bax^Δcd23 or Tnfrsf13c^–/–Bak^–/–Bax^Δcd23 mice were transferred into venetoclax-treated CD45.1⁺ C57BL/6 wt recipient mice, which were then maintained on daily venetoclax treatment before analysis. (G) Histograms of BCL-2, MCL-1, and BCL-XL protein expression in donor CD45.2⁺ B cells from unmanipulated control mice with the same genotype or from venetoclax treated recipients. (H) Geometric mean of BCL-2, MCL-1, and BCL-XL protein expression in Bak^–/–Bax^Δcd23 and Tnfrsf13c^–/–Bak^–/–Bax^Δcd23 B cells from control (untreated) recipient mice or from venetoclax-treated recipient mice as indicated in (F), measured by flow cytometry. (I) Fold change in BCL-2, MCL-1, and BCL-XL protein levels in Bak^–/–Bax^Δcd23 and Tnfrsf13c^–/–Bak^–/–Bax^Δcd23 B cells in venetoclax-treated mice. The data are expressed relative to the MFI in B cells recovered from unmanipulated control mice of the same genotype. Data from panels B-H are representative of 3 independent experiments with n = 3 to 6 mice per group. For all bar graphs, the mean ± SEM are shown and each symbol represents an individual mouse; Student 2-tailed t test was used. ∗P < .05; ∗∗P < .01; ∗∗∗P < .005; ∗∗∗∗P < .001. ELISA, enzyme-linked immunosorbent assay; MFI, mean fluorescence intensity.

Figure 6.

Healthy human B cells undergoing venetoclax treatment have increased BCL-2 expression. (A) Schematic representation of the study design. PB samples from 4 breast cancer patients were collected at screening and during venetoclax treatment (all patients received 800 mg venetoclax). (B) Change in lymphocyte concentration (cells per μL) during venetoclax treatment in each patient after 31 days of treatment. (C) Concentrations of BAFF and APRIL (pg/mL) detected in the serum samples of the breast cancer patients using a Luminex assay at screening and 15 days after venetoclax treatment. (D) Histograms of BCL-2 protein expression by flow cytometry in B cells in patients before and after venetoclax treatment (BC_01013 after 201.7 weeks, BC_01014 after 61.1 weeks, BC_01026 after 67.4 weeks, BC_01029 after 112.1 weeks). (E) Geometric mean of the BCL-2 levels of B cells in 4 patients. (F) KEGG enrichments analysis of differentially expressed genes in single-cell cellular indexing of transcriptomes and epitopes sequencing data from 5 paired samples under long-term venetoclax treatment. Student 2-tailed paired t test was used to calculate P values. ∗P < .05; ∗∗P < .01. APRIL, a proliferation-inducing ligand (also known as TNFSF13); KEGG, Kyoto Encyclopedia of Genes and Genomes; TNF, tumor necrosis factor.

Figure 7.

Increase in serum BAFF levels correlates with BCL-2 upregulation in CLL cells in patients undergoing targeted therapies. (A) Schematic representation of the patient groups sampled before or after venetoclax treatment. (B) CLL burden (proportion of lymphocytes) for each patient at collection. (C) Concentration (pg/mL) of serum BAFF. (D) MFI of BCL-2 in CLL cells from each patient. (E) Schematic representation of the study design. PB samples from 3 patients with CLL were collected at screening and during obinutuzumab and venetoclax dose escalation. Patients received IV infusion of obinutuzumab on days 1/2 (first 1000 mg dose divided over 2 days 100/900 mg), 8 (1000 mg), 15 (1000 mg), and 29 (1000 mg). Venetoclax was introduced at 20 mg daily on day 22 and increased weekly (20, 50, 100, 200, 400 mg). (F) Change in lymphocyte concentrations (cells per liter) during obinutuzumab and venetoclax treatment in each patient. (G) Concentration (pg/mL) of BAFF detected in the serum samples of the patients with CLL using a Luminex assay. (H) Histograms (top panel) and MFI (lower panel) of BCL-2 protein expression in CLL cells in each patient at the indicated time points. For panels B-D, the mean ± SEM are shown and each symbol represents a measurement from an individual sample; Student 2-tailed t test was used. ∗P < .05; ∗∗P < .01; ∗∗∗P < .005; ∗∗∗∗P < .001. Obin, obinutuzumab; PBMC, peripheral blood mononuclear cell.