Activation of Th1 Immunity within the Tumor Microenvironment Is Associated with Clinical Response to Lenalidomide in Chronic Lymphocytic Leukemia

This article is featured in In This Issue, p.1811

Evading immune destruction is a hallmark of tumor progression (1). Immune cells not only fail to control tumor growth but may in fact sustain proliferation and survival of tumor cells (2). In patients with chronic lymphocytic leukemia (CLL), global gene expression profiling of CD4⁺ and CD8⁺ cells revealed defects involving cell differentiation, cytotoxicity, and cytoskeletal pathways (3). Thus, restoration of T cell antitumor immunity represents an attractive treatment strategy to restore immune surveillance (2, 4).

The immunomodulatory drug lenalidomide upregulates costimulatory molecules on tumor cells (5, 6) and repairs impaired immunologic synapse formation between T cells and CLL cells (7). Lenalidomide promotes NK cell–mediated killing of tumor cells in vitro (8) and stimulates the production of Igs by normal B cells (6). The proliferation of CLL cells is also directly inhibited by lenalidomide in culture via a cereblon-dependent induction of the cell-cycle inhibitor p21 (9). Two recent clinical trials showed that maintenance therapy with lenalidomide delayed disease progression without deepening responses (10, 11). In the absence of tumor eradication, the in vivo mechanisms by which lenalidomide exerts activity against CLL are poorly understood.

In this study, we comprehensively evaluated changes in the T cell compartment in patients with relapsed or refractory CLL treated with lenalidomide. Our data link IFN-γ production, T cell proliferation, and Th1 polarization in the lymph node (LN) microenvironment to clinical response.

Samples were collected from patients with relapsed CLL or small lymphocytic lymphoma treated with lenalidomide under a phase 2 investigator-initiated study (identifier: NCT00465127). Between May 2007 and February 2010, 33 patients received lenalidomide at 10 or 20 mg daily, cycled 3 wk on, 3 wk off for up to eight cycles (5, 6). The study was approved by the institutional review board at the National Heart, Lung, and Blood Institute and conducted in accordance with the Declaration of Helsinki. All patients provided written informed consent. The primary endpoint was overall response after four cycles as assessed by modified International Workshop on Chronic Lymphocytic Leukemia criteria (12). Lymphadenopathy was assessed by the sum of the product of the greatest diameters of representative LNs with computed tomography. Samples for in vitro studies were collected from patients with treatment of naive CLL after obtaining written informed consent (identifier: NCT00923507). PBMCs and LN core biopsy specimens were collected prior to and on day 8 of therapy and stored as previously described (5).

Total RNA was isolated from CD19 positively selected PBMCs and LN core biopsy specimens. Microarray analysis was performed on Affymetrix Human Genome U133 Plus 2.0 Array chips (Santa Clara, CA) as described (13). Biotin-labeled RNA (20 μg) was fragmented to ∼200 bp and hybridized to U133 Plus 2.0 chips for 16 h, washed, and stained on a fluidics station. Affymetrix Expression Console software was used to calculate signal intensities and present calls on the hybridized chips. The signal intensity values of the probe sets were normalized by Robust Multi-Array Average across the chips (14). Only probe sets with a present signal on >5 arrays were selected for analysis. The expression of multiple probe sets corresponding to a gene was averaged. Two-way ANOVA was applied to evaluate patient and lenalidomide treatment effects on day 8 relative to day 0. The Benjamini–Hochberg (15) method was used to correct for multiple testing. Cluster and Tree View (Eisen Laboratory, Stanford University, Palo Alto, CA) and Ingenuity Pathway Analysis (Ingenuity Systems, Redwood, CA, accessed June 8, 2018) were used for gene expression analysis. The microarray data set is available on the National Center for Biotechnology Information Gene Expression Omnibus Web site (https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE112953) under accession number GSE112953.

Previously described CD4⁺ and CD8⁺ T cell gene signatures were used for T cell subsets analysis (16–18).

Enumeration of CD3⁺ cells and intracellular staining for IFN was performed as previously described (19, 20). IFN-γ in the serum was measured using Meso Scale Diagnostics (Gaithersburg, MD).

LN core biopsy specimens were stained with CD3, CD4, and CD8 (Dako, Carpinteria, CA). The number of CD3⁺ cells was scored in five representative high-power fields by a trained pathologist blinded to the samples. Images were captured at original magnification ×400 on an Olympus Bx41 microscope (Center Valley, PA).

TCRα- and β-chain deep sequencing was performed to assess lenalidomide-induced clonal expansion of T cells in LNs as previously described (21). In brief, 1 μg of total RNA (only 0.75 μg of total RNA available at pretreatment from subject L2) was used for PCR-based amplification of TRA or TRB gene products with adapter-conjugated primer sets. The template library was amplified by Nextera XT DNA Library Preparation Kit (Illumina, San Diego, CA). Subsequently, the prepared library was analyzed using MiSeq Reagent 600-cycle Kit v3 and MiSeq System (Illumina). After deep sequencing, each V, (D), J, and C segment of TCRα-chains and β-chains was mapped to reference sequences in the International ImMunoGeneTics/GENE-DB (22) and assigned for determination of the complement-determining region 3 (CDR3) amino acid sequence as previously described (21). The diversity index (inverse Simpson index) of CDR3 sequences was calculated to assess overall diversity and clonality in the TCRα and β clonotypes.

PBMCs (5 × 10⁵ cells/ml) were cultured for 3 wk in RPMI 1640 supplemented with penicillin, streptomycin, and glutamine (all Life Technologies, Grand Island, NY); FCS (10%; Sigma-Aldrich, St Louis, MO); IL-2 (100 U/ml), IL-7 (50 IU/ml), and IL-15 (5 IU/ml; all PeproTech, Rocky Hill, NJ); and in presence (2 μM dissolved in DMSO [0.1%]) or absence (DMSO 0.1%) of lenalidomide (Sequoia Research Products, Berkshire, U.K.). Cells were once stimulated at a 1:1 ratio with irradiated PBMCs (25 Gy), then CD3 negative selection (RoboSep; STEMCELL Technologies, Vancouver, BC, Canada) was performed. CD19⁺ cells were obtained from autologous PBMCs, cultured with lenalidomide (2 μM) or vehicle (DMSO) for 48 h (CD19 positive selection; Miltenyi Biotec, Auburn CA). CD3⁺ cells were stained with 0.5 μM CFSE (Life Technologies) as described (23). CD3⁺ cells and CD19⁺ cells were cocultured at a 1:1 ratio at 1 × 10⁶ cells in RPMI 1640 alone (no cytokines, no lenalidomide, no FCS) on a 96-well plate. Flow cytometric analysis was performed at 72 h poststimulation (BD LSRFortessa; BD Biosciences, San Jose, CA) following manufacturer’s instructions. The following experiments were set up: CD3, CD3 (lenalidomide treated), CD3/CD19, CD3 (lenalidomide treated)/CD19, CD3/CD19 (lenalidomide treated), and CD3 (lenalidomide treated)/CD19 (lenalidomide treated). Intracellular staining was described previously (BD Cytofix/Cytoperm Plus, Fixation/Permeabilization Solution Kit with BD GolgiPlug; BD Biosciences). A total of 5 × 10⁵ cells were stained with the following mouse anti-human Abs: Vivid and CD14 Pacific Blue, CD19 APC, CD4 V500, CD8 H7APC, IFN-γ, PE-Cy7 (all BD Biosciences Pharmingen, San Jose, CA), and CD3 eFluor 605 (Thermo Fisher Scientific, eBioscience, San Diego, CA).

A paired t test was used to compare pre- and on-treatment samples, and an unpaired t test was used to compare responders and nonresponders. A p value <0.05 was considered statistically significant. Statistical analyses were performed using JMP 13 software (SAS Institute, Cary, NC).

Thirty-three patients with relapsed CLL were enrolled in a phase 2 study of lenalidomide at 10 or 20 mg daily for 3 wk, followed by 3 wk off for up to eight cycles. Patients received a median of two prior lines of therapy (range 1–5), including purine analog in 81% and anti-CD20 mAb in 100% of patients. All patients had progressive disease requiring treatment at the time of enrollment. On an intention-to-treat basis, five (15%) patients achieved partial remission, 20 patients (60%) had stable disease, and eight patients (25%) had progressive disease. Twenty-three patients completed four cycles of therapy and were evaluated by absolute lymphocyte count and computed tomagraphy. Thirteen patients showed a ≥ 10% reduction in absolute lymphocyte count (Supplemental Fig. 1A). Nine patients showed a ≥10% reduction in lymphadenopathy (Supplemental Fig. 1B) and were considered responders for the correlative analyses presented in this article.

To understand the antitumor effects of lenalidomide, we performed gene expression profiling on circulating CD19 positively selected cells from 11 patients treated with lenalidomide. We identified 79 lenalidomide-responsive genes (fold change ≥2, false discovery rate [FDR] <0.2 between pretreatment and cycle 1, day 8 samples, Supplemental Table I), of which 67 were upregulated and 12 were downregulated (Fig. 1A). Upregulated genes encoded chemokines (CCL3, CCL4), cytokines or cytokine receptors (IL13RA1, TNF, and TNFSF13B), signal transduction molecules (STK3, RCAN1, and KSR2), and molecules involved in the regulation of apoptosis (DDIT4, PAPRP9, and CFLAR). Ingenuity Pathway Analysis identified the IFN-γ signaling pathway as the most significantly overrepresented pathway (p = 6.5 × 10⁻¹⁰). Specifically, among 36 known target genes of IFN-γ, six were upregulated and none were downregulated in response to lenalidomide, suggesting that tumor cells respond to IFN-γ. Indeed, serial measurements of serum IFN-γ in these patients showed a significant increase as early as day 4, which more than doubled by day 8, and remained elevated during the first 3 wk on lenalidomide (Fig. 1B). After 3 wk off lenalidomide, serum IFN-γ returned to baseline levels before increasing again with the start of cycle 2.

IFN-γ is the canonical cytokine of Th1-type tumor immune surveillance and eradication (24, 25). Using flow cytometry, we found an increased proportion of circulating IFN-γ⁺ CD4⁺ and CD8⁺ T cells in patients treated with lenalidomide (p = 0.003 and 0.04, respectively, Fig. 1C). Although the relative increase was more pronounced in CD4⁺ compared with CD8⁺ T cells, the frequency of IFN-γ⁺ cells on day 8 was comparable between the two T cell subsets, suggesting that both contributed to IFN-γ secretion.

A classic observation in CLL patients starting lenalidomide is the tumor flare reaction (TFR), an often rapid and painful swelling of LNs. TFR is thought to be due to increased T cell infiltration of the tumor (5, 6). To further dissect the T cell response induced by lenalidomide, we performed gene expression profiling of pretreatment and cycle 1, day 8 LN samples from the 11 patients described above. Overall, 56 genes were differentially expressed (fold change ≥2, FDR <0.2, Supplemental Table I) between pre- and on-treatment LN samples and suggested a T cell response induced by lenalidomide. Next, we asked if and how gene expression could differ based on clinical response. We identified 119 differentially expressed genes among seven responders (Fig. 2A, Supplemental Table II) and none among four nonresponders. This discrepancy between responders and nonresponders suggested that changes in gene expression within the LN could help predict clinical response to lenalidomide.

To investigate the relationship between changes in the tumor microenvironment (TME) and clinical response, we were particularly interested in the group of 98 genes that were upregulated by lenalidomide in LN samples (Fig. 2A). These genes comprised important immune regulatory molecules, including IFNG (Fig. 2B), cytotoxic effector molecules (GZMA, GZMB), lymphocyte activation markers (CD38), and multiple chemokines (e.g., CXCL11). We note that IFNG was the only member of the IFN family consistently expressed across samples (Fig. 2D). Notably, 42 of 98 genes (43%) that were significantly upregulated by lenalidomide in responders compared with nonresponders were IFN-γ–regulated genes (p = 0.02, Fig. 2C).

Tumor-infiltrating lymphocytes have been associated with improved survival and response to treatment across multiple cancers (26). We previously reported that the number of T cells in LN biopsy specimens increased in some but not all patients treated with lenalidomide (5). In patients with available pre- and on-treatment LN biopsy specimens, we compared the degree of T cell infiltration between responders and nonresponders. Lenalidomide appeared to induce a more prominent T cell infiltrate in the LN biopsy specimens of responders than nonresponders but not quite meeting statistical significance, likely owing to a small sample size (p = 0.056, Fig. 3A). Additional immunohistochemistry suggested that this T cell infiltrate was composed of more CD4⁺ T cells than CD8⁺ T cells (Fig. 3A). Therefore, we compared the expression of CD4⁺ and CD8⁺ T cell–specific gene signatures (16–18, 27–29) between pre- and on-treatment biopsy specimens. The CD4⁺ T cell–specific gene signature was significantly upregulated by lenalidomide, whereas expression of the CD8⁺ T cell gene signature remained unchanged (Fig. 3B).

The differentiation of CD4⁺ T cells into Th1 or Th2 cells is determined by the opposing transcription factors T-bet and GATA-3 (30–32). Th1 cells mediate antitumor immunity by producing IFN-γ; recruiting CD8⁺ T cells, NK cells, and macrophages; and inhibiting angiogenesis (33). To investigate the effect of lenalidomide on Th1/Th2 balance, we analyzed the expression of TBX21 encoding T-bet and GATA3 (31, 34). Lenalidomide induced TBX21 expression in the LN biopsy specimens of responders compared with nonresponders (p = 0.008, Fig. 3C), supporting a shift toward a Th1-type immune response. In contrast, GATA3 expression was not different between pre- and on-treatment biopsy specimens or between responders and nonresponders (Fig. 3C).

Skewing of the TCR repertoire and the presence of shared clonotypes between patients suggest common Ag selection in CLL (35). In solid tumors, response to immunotherapy with checkpoint inhibitors has been associated with the oligoclonal expansion, and resultant decreased diversity, of tumor-infiltrating T cells (21). To explore the shifts in T cell diversity on lenalidomide, we performed TCRα and TCRβ deep sequencing on pre- and on-treatment LN biopsy specimens of responders. In the three patients analyzed, the diversity of both TCRα and TCRβ repertoire decreased following treatment with lenalidomide, suggesting expansion of select clonotypes (Fig. 3D, Supplemental Fig. 2).

To dissect the lenalidomide-induced immune response, we performed a set of in vitro assays (Fig. 4). CD3⁺ T cells from CLL patients were stimulated with autologous irradiated CD19⁺ CLL cells and exposed to lenalidomide in vitro or left untreated. Exposure to lenalidomide increased proliferation of both CD4⁺ and CD8⁺ T cells compared with controls (Fig. 4A, 4B, 4D). The strongest responses were seen when lenalidomide- prestimulated CD4⁺ T cells were cocultured with CLL cells (Fig. 4B). Whether the CLL cells were or were not also treated with lenalidomide did not significantly change the rate of CD4⁺ T cell proliferation. In contrast, exposing CD8⁺ T cells or CLL cells to lenalidomide increased the rate of proliferation of CD8⁺ T cells.

We measured the frequency of IFN-γ–producing cells in the CD4⁺ and CD8⁺ T cell subsets by flow cytometry and found that the addition of lenalidomide to CLL PBMCs induced production of IFN-γ from both CD4⁺ and CD8⁺ cells compared with untreated controls (Fig. 4C, 4E), consistent with our observations in lenalidomide-treated patients (Fig. 1C).

TME interactions support the development and progression of CLL (36). Prior studies have examined the effect of lenalidomide on circulating T cell subsets in CLL patients (37, 38). However, in vivo analysis of the TME, where the effects of lenalidomide arguably matter most, are needed. In this study, our data link immunomodulation of the TME and clinical response to lenalidomide. By gene expression profiling of paired peripheral blood and LN samples, we show that transcriptional changes induced by lenalidomide are tissue specific. Specifically, lenalidomide activated a Th1-type immune response within the TME that was associated with LN regression.

Lenalidomide has been shown to reverse several aspects of immune evasion. First, lenalidomide upregulates costimulatory molecules on tumor cells and enhances their immunogenicity (39–41). Second, lenalidomide repairs defective interactions between tumor and T cells (42). Third, lenalidomide induces T cell secretion of IFN-γ and IL-2, which promotes Th1 differentiation (43–45). Last, lenalidomide improves cytotoxic effector function against tumor cells (43, 44). In this study, we provide a valuable extension of these prior in vitro observations by characterizing the in vivo immune responses induced by lenalidomide within the TME.

Better T cell function and more CD4⁺ T cells before treatment initiation have been associated with improved clinical response to lenalidomide (46). Consistent with these findings, we identified an association between the rapid onset of Th1-type immune activation within the TME and treatment response. In responders, we also observed an expansion of certain T cell clonotypes by TCR repertoire analysis. Because costimulatory molecules on CLL cells are upregulated by lenalidomide (5), we propose that antigenic stimulation may contribute to the clonal expansion of antitumor T cells.

How and if lenalidomide fits into the current treatment paradigm for CLL remains unclear. Lenalidomide has single-agent activity in treatment-naive (47, 48) and relapsed or refractory (49, 50) CLL. Overall response rates ranged between 12 and 72%, and complete responses were seen in less than 20% of patients (47, 50–52). Side effects, including neutropenia and TFR, which have been associated with T and NK cell activation against CLL cells, were often dose limiting (5, 39, 40). The early termination of a randomized phase 3 trial comparing lenalidomide to chlorambucil highlighted the significant morbidity and mortality associated with lenalidomide use (53). However, there are positive aspects: lenalidomide improves Ig levels (47), induces long-term responses in a subset of patients (52), and prolongs response when given as maintenance therapy (10, 11). In addition, lenalidomide enhances the activity of anti-CD20 Abs, and the combination has become an important treatment regimen for patients with certain B cell lymphomas (54–56). Thus, judicious incorporation of lenalidomide into treatment regimens may be beneficial and should be weighed against the safety and efficacy of novel immunotherapies.

Activation of antitumor immunity has emerged as one of the most promising therapeutic strategies against cancer. In addition to conventional immunotherapies, small molecules, particularly inhibitors of BCR signaling, also appear to modulate the immune system (57–59). As combinations of immunotherapy and small molecules are being investigated for CLL, it is important to understand their impact on the immune system. We have shown that immunological changes are tissue specific and that clinically relevant effects may occur primarily within the TME. Characterization of tissue biopsy specimens may, therefore, be required to identify meaningful biomarkers in ongoing immunotherapy clinical trials.

We thank our patients for their participation in this research.

The authors have no financial conflicts of interest.