Treatment with lenalidomide induces immunoactivating and counter-regulatory immunosuppressive changes in myeloma patients



Lenalidomide activates the immune system, but the exact immunomodulatory mechanisms of lenalidomide in vivo are poorly defined. In an observational study we assessed the impact of lenalidomide on different populations of immune cells in multiple myeloma patients. Lenalidomide therapy was associated with increased amounts of a CD8+ T cell subset, phenotypically staged between classical central memory T cells (TCM) and effector memory T cells (TEM), consequently termed TCM/TEM. The moderate expression of perforin/granzyme and phenotypical profile of these cells identifies them as not yet terminally differentiated, which makes them promising candidates for the anti-tumour response. In addition, lenalidomide-treated patients showed higher abundance of CD14+ myeloid cells co-expressing CD15. This population was able to inhibit both CD4+ and CD8+ T cell proliferation in vitro and could thus be defined as a so far undescribed novel myeloid-derived suppressor cell (MDSC) subtype. We observed a striking correlation between levels of TCM/TEM, mature regulatory T cells (Tregs) and CD14+CD15+ MDSCs. In summary, lenalidomide induces both activating and inhibitory components of the immune system, indicating the existence of potential counter-regulatory mechanisms. These findings provide new insights into the immunomodulatory action of lenalidomide.


Multiple myeloma (MM) is a malignant plasma cell disease, affecting adults with a median age of 71 years. It is the second most common haematological malignancy, with an incidence of 4·3/100 000/year [1]. Like many malignancies, MM is associated with a marked immune dysfunction, which allows the tumour to escape and renders any immunotherapeutic approaches inherently difficult.

Thus far, studies investigating MM-induced immunosuppression have focused on T cell proliferation and function, which are believed to be directly impaired by myeloma cells [2, 3] or indirectly by MM-induced dendritic cell (DC) dysfunction [4, 5]. T cells in MM have been shown to exhibit multiple and profound signalling defects [6], resulting in intrinsic T cell dysfunction characterized by an altered activation and reduced anti-tumour reactivity. In addition, increased frequencies of functional regulatory T cells (Tregs) have been observed in MM patients [7, 8].

The innate immune response has been more or less neglected in the past, but recently we and others could show a defect in natural killer (NK) cell activity in MM [9, 10] and increased serum levels of soluble NKG2D ligand MICA [major histocompatibility complex (MHC) class I chain-related A]. In addition, myeloid-derived suppressor cells (MDSC) are another important immunoregulatory cell population also occurring in cancer patients and suppressing anti-tumor immunity [11-13]. In mice, MDSC are quite clearly defined by their characteristic phenotype CD11b+Gr-1+ and have been shown to exhibit suppressive properties in vitro and in vivo [11]. Reports on MDSC in human MM are scarce. To date, two publications report this cell type in MM patients, one describing an increased frequency of human leucocyte antigen D-related (HLA-DR)lo monocytes in patients with MM at various stages of their disease [14] and the other observing granulocytic MDSCs in the peripheral blood and bone marrow of MM patients [15].

A promising approach to counteract immunoinhibitory effects in MM is the implementation of immunotherapeutic agents such as lenalidomide (CC-5013, IMiD3, Revlimid), which is an effective drug in the treatment of newly diagnosed and relapsed MM. Furthermore, it has been employed successfully for maintenance therapy after high-dose chemotherapy of MM patients [16]. The combination of its direct anti-neoplastic properties on myeloma cells [17, 18] and its modulatory effect on the patient's immune system make lenalidomide a promising therapeutic drug, supported by its ability to confer an overall survival benefit in various clinical trials [19-21]. The high clinical activity of the compound depends at least in part on its potent immunoactivating properties. Lenalidomide has been reported to enhance activation and antigen-specific expansion of CD8+ T cells [22-24] and to increase the breadth of antigen-specific CD8+ T cell responses [22]. Although lenalidomide was reported to have no impact on NK cell frequencies, it was associated with enhanced NK cell activity [25-27]. Its immunoactivating properties are supported further by the observation that lenalidomide maintenance early after allogeneic stem cell transplantation in myeloma, MDS and AML patients induces severe graft-versus-host disease (GVHD) complications [28, 29].

The aim of this study was to identify potential changes in the immune system in MM patients compared to healthy controls and to investigate the role of various treatment regimens including lenalidomide. We therefore performed a comprehensive phenotypical and functional immunological screening. Here, we could identify changes in the immune system in MM patients exposed to lenalidomide compared to all other groups [untreated MM, monoclonal gammopathy of uncertain significance (MGUS) and other treatment regimens]. These changes comprise an increase of activating T cells as well as a possibly counter-regulatory increase of suppressing MDSCs.

Materials and methods


The study was approved by the local ethic committee and was conducted according to the Declaration of Helsinki. All patients and healthy donors gave written informed consent.

Diagnostic peripheral blood samples and clinical data from 68 MM patients, five patients with MGUS and 24 healthy controls were collected between May 2011 and October 2012 at the University Hospital in Bonn. Demographics and disease characteristics of patients and healthy controls are shown in Table 1. As this study was designed as an observational study, data from all MM patients were collected consecutively, irrespective of their therapy or disease status. All therapeutic decisions were entirely at the physician's discretion. They were categorized as untreated, lenalidomide-treated or treated with other therapies retrospectively, after collection of all clinical and experimental data.

Table 1. Demographic and clinical characteristics
 MM untreated (n = 36)MM lenalidomide (n = 17)MM other therapies (n = 15)MGUS (n = 5)Control (n = 24)P-value (all groups)P-value (MM only)
  1. Kruskal–Wallis. Chi-Quadrat (Pearson). Hb, haemoglobin; Ig, immunoglobulin; ISS, International staging system; MM, multiple myeloma; MGUS, monoclonal gammopathy of uncertain significance.
Age, median (range)65 (59–69)64 (54–69)63 (60–68)76 (70–76)59 (55–62)0·0450·885
Male21 (58%)10 (59%)8 (53%)2 (40%)4 (17%)0·0170·938
ISS stage      0·765
I4 (11%)0 (0%)0 (0%)  
II9 (25%)4 (24%)7 (47%)  
III23 (64%)13 (76%)8 (53%)  
Type of paraprotein     0·2260·779
IgA6 (17%)5 (29%)2 (13%)1 (20%)  
IgG19 (53%)9 (53%)7 (47%)3 (60%)  
IgM0 (0%)0 (0%)1 (7%)1 (20%)  
Free light chains11 (31%)3 (18%)5 (33%)0 (0%)  
Leucocytes (×103/μl), median (range)5·5 (4·2–7·0)3·1 (2·9–4·8)6·4 (5·1–7·3)5·5 (4·7–6·0)5·9 (5·2–7·7)0·0030·002
Hb (g/dl), median (range)11·8 (10·5–12·6)11·9 (11·7–12·9)11·4 (11·1–12·8)12·4 (11·9–14·6)14·2 (13·6–14·7)0·0000·557
Thrombocytes (×103/μl), median (range)208 (189–290)124 (107–210)250 (177–317)222 (141–266)233 (193–268)0·0100·003
Creatinine (mg/dl), median (range)1·02 (0·78–1·35)0·94 (0·78–1·22)1·09 (0·89–1·23)0·87 (0·79–0·90)0·8330·822
β2 microglobulin (mg/dl), median (range)3·5 (2·4–4·1)2·6 (2·3–2·8)2·9 (2·3–4·4)2·2 (1·9–2·5)0·3190·436

Sample processing and flow cytometry

Peripheral heparinized blood was obtained from patients and processed immediately. Peripheral blood mononuclear cells (PBMCs) were isolated by Ficoll density centrifugation and single-cell suspensions were stained with designated antibodies in magnetic-activated cell sorting (MACS) buffer [phosphate-buffered saline (PBS), 1·5% fetal calf serum (FCS), 2 mM ethylenediamine tetraacetic acid (EDTA)] for 15 min at room temperature, washed once with MACS buffer, and processed immediately for measurement. For intracellular forkhead box protein 3 (FoxP3) staining, cells were fixed and permeabilized using the FoxP3 Staining Buffer Set (eBioscience, San Diego, CA, USA), according to the manufacturer's protocol. For intracellular perforin/granzyme detection, cells were fixed with 2% (w/v) paraformaldehyde for 20 min at room temperature and permeabilized by washing twice with 0·2% saponin. Staining with appropriate antibodies was carried out in 0·2% saponin for 15 min; cells were washed in 0·2% saponin and stored in MACS buffer until measurement. Appropriate isotype controls were always included. Cells were measured on a Canto II fluorocytometer (BD Biosciences, San Jose, CA, USA) and data were analysed using FlowJo software (TreeStar, Ashland, OR, USA).

Intracellular cytokine detection

For intracellular cytokine detection, freshly isolated PBMCs were seeded at 2 × 105 per 96-well and cultured overnight. Thereafter, they were stimulated with 100 ng/ml phorbol 12-myristate 13-acetate (PMA; Sigma, St Louis, MO, USA) and 0·5 μM ionomycin (Sigma) in the presence of the protein transport inhibitor brefeldin-A (3 μg/ml; eBioscience) for 5 h. Intracellular staining for interferon (IFN)-γ and interleukin (IL)-10 was performed as for the perforin/granzyme detection (see above).

T cell proliferation assay

PBMCs from healthy donors were isolated and stained with 5-(and 6)-carboxyfluorescein diacetate succinimidyl ester (CFSE; eBioscience). Cells were then CD14-depleted using anti-CD14 microbeads (Miltenyi Biotec, Bergisch Gladbach, Germany) and seeded into round-bottomed 96-wells at 4 × 104 cells per well. Monocytes from MM patients were separated from PBMCs of MM patients by anti-CD14 microbeads (Miltenyi Biotec), stained with anti-CD14/CD15 antibodies and sorted into CD14+CD15 and CD14+CD15+ populations using a FACSVantage (BD Biosciences). As additional controls, monocytes from allogeneic healthy controls were bead-separated and sorted into CD14+CD15 cells. Sorted monocyte populations were then co-cultured with CFSE-stained responder cells at the indicated ratios. Proliferation was initiated by anti-CD2/CD3/CD28 beads (Miltenyi Biotec) at a cell : bead ratio of 16:1 and proliferation was assessed after 72 h by visualization of CFSE dye dilution using flow cytometry. Analysis and proliferation index calculation was performed using FlowJo software (Tree Star).

Statistical analysis

Results were analysed using spss (version 19·0.0). Graphic displays were created using spss (version 19·0.0) and GraphPad Prism (version 4·0a). Group comparisons were performed by non-parametric Kruskal–Wallis test or Mann–Whitney U-test as indicated. The correlation between continuous variables was assessed by non-parametric Spearman's rank correlation. A P-value <0·05 was considered statistically significant.


Patient characteristics

In order to examine the immunomodulatory impact of multiple myeloma itself and various therapeutic treatments, we analysed clinical and immunological parameters obtained from 36 untreated patients with MM, 32 MM patients receiving therapy, five patients with MGUS and 24 age-matched healthy controls (for patient characteristics see Table 1). The group of untreated MM patients included 16 newly diagnosed (47%), 12 relapsed (33%) and eight patients with stable remission after therapy not receiving maintenance therapy (22%). The group of lenalidomide-treated MM included 11 patients receiving lenalidomide as monotherapy (65%), four patients receiving bortezomib, dexamethasone, cyclophosphamide and lenalidomide (VDCR) (24%) and one patient receiving lenalidomide and dexamethasone (RD) (6%) and lenalidomide, dexamethasone and cyclophosphamide (RCD) therapy (6%), respectively. The group of MM patients treated with therapies excluding lenalidomide consisted of 10 patients receiving bortezomib, adriamycin and dexamethasone (PAD) (67%), two patients receiving bortezomib, cyclophosphamide and dexamethasone (VCD) (13%) and one patient treated with cyclophosphamide and dexamethasone (CD), bortezomib and dexamethasone (VD) and bortezomib, cyclophosphamide and dexamethasone (VCD), respectively (7%).

Lenalidomide is associated with the presence of a novel CD8+ T cell subset staged between central and effector memory cells

We set out to examine various T cell parameters such as cytokine profile, activation status and subset distribution in MM patients and controls. The activity of T lymphocytes can be assessed by their HLA-DR expression, IFN-γ production and the presence of cytolytic granules containing perforin or granzyme. In our study, CD3+CD4+ and CD3+CD8+ T cells from the peripheral blood of treated or untreated MM patients, MGUS patients or healthy donors were analysed initially with regard to their expression of HLA-DR. In lenalidomide-treated MM patients, HLA-DR expression in CD4+ and CD8+ T cells was increased significantly (CD4+ P = 0·0005, CD8+ P < 0·0001), implying the presence of activated T cells (Fig. 1a).

Figure 1.

Identification of a novel central memory CD8+ T cell type. (a) Human leucocyte antigen D-related (HLA-DR) expression in peripheral blood CD4+ and CD8+ T cells. Cells analysed by flow cytometry and gated as CD45+CD3+CD4+/CD8+. Box-plots represent untreated multiple myeloma (MM) patients (n = 36), lenalidomide-treated patients (n = 17), patients treated by therapies excluding lenalidomide (n = 15), monoclonal gammopathy of uncertain significance (MGUS) (n = 5) and age-matched healthy controls (n = 24). Lenalidomide treatment significantly increased the frequency of HLA-DR+ T cells in patients (median CD4+ = 12·4%, CD8+ = 11·7%) compared to untreated MM patients (median CD4+ = 3·6%, CD8+ = 4·5%; CD4+ P = 2·54E-04, CD8+ P = 4·54E-05). (b) Cytokine expression in CD4+ and CD8+ T cells after phorbol myristate acetate (PMA)/ionomycin stimulation for 5 h. Lenalidomide treatment significantly increases interleukin (IL)-10 production by CD4+ T cells as well as interferon (IFN)-γ and IL-2 production by CD8+ T cells (median CD4+/IL-10+ = 0·38%, median CD8+/IFN-γ = 59%, median CD8+/IL-2+ = 30%) compared to untreated patients (median CD4+/IL-10+ = 0·82%, median CD8+/IFN-γ = 80%, median CD8+/IL-2+ = 48%; CD4+/IL-10+ P = 0·017, CD8+/IFN-γ P = 0·030, CD8+/IL-2 P = 0·019). Lenalidomide treatment also results in an increase of IL-2 production within the CD8+IFN-γ+ subset (untreated median = 32%, Lena median = 51%, P = 0·019). (c) Commonly used gating strategies for CD8+ T cell populations by plotting either CD45RA versus CCR7 or CD45RA versus CD62L. CD8+ T cell subsets are defined as naive (CD45RA+CCR7+CD62L+), central memory [central memory T cells (TCM), CD45RACCR7+CD62L+], effector [effector memory T cells (TEM), CD45RACCR7CD62L] and late effector memory RA [(TEMRA), CD45RA+CCR7CD62L]. Compared to healthy controls (left) we observed a proportion of CCR7 cells within the CD45RACD62L+ gate in MM patients (right) leading to divergent results for TCM defined by gating strategies employing CD45RA/CCR7 and CD45RACD62L (TCM?). The level of CCR7 expression in this T cell population is associated strongly with the level of perforin/granzyme expression (right). (d) Extended gating strategies for CD8+ T cell populations including two subsets of CD45RACD62L+ central memory cells differing in their CCR7 expression. (e) Perforin/granzyme expression in the five CD8+ T cell subsets as defined in (c). Freshly isolated peripheral blood mononuclear cells (PBMCs) from patients and healthy controls (n = 36) were stained with extracellular markers as in (d) and intracellularly for perforin and granzyme after fixation/permeabilization. Boxes depict the interquartile range (IQR), the line the median and the whiskers the 95% confidence interval (CI).

Our screening panel comprised several cytokines to gain a first impression of possibly modulated T cell subsets. We included IL-2, IFN-γ, tumour necrosis factor (TNF)-α, IL-10 and IL-17 to identify effector functions and T helper (Th1, Th2, Th17) subsets. Lenalidomide treatment of MM patients resulted in higher frequencies of IL-10-positive cells among CD4+ T cells (P = 0·017) and elevated levels of IFN-γ-positive (P = 0·030) and IL-2-positive (P = 0·019) cells among CD8+ T cells (Fig. 1b). For the CD8+ subset, elevated IFN-γ points to enhanced effector differentiation to the target antigen; increased IL-2 implies augmented self-renewal for long-term survival and memory function. No significant differences in cytokine expression were observed between all other groups. We also did not observe any significant changes in the other cytokines screened (CD4+/IL-2+ P = 0·597, CD4+/IL17+ P = 0·170, CD4+/IFN-γ P = 0·150, CD8+/TNF-α+ P = 0·762). Notably, within the IFN-γ+ subset of CD8+ T cells, significantly more cells expressed IL-2 upon lenalidomide treatment (P = 0·019), which implies that activated CD8+ cells of lenalidomide-treated patients contain larger amounts of self-renewing memory-type cells. Based on these findings, we set out to perform further phenotyping of the CD8+ T cell population.

To determine which CD8+ T cell subsets are responsible for these elevated parameters, we analysed the CD8+ T cell pool in more detail. It has long been debated which combination of differentiation markers defines the CD8+ subsets most precisely [30, 31]. The most commonly used markers are CD45RA, CCR7, CD62L, CD127, CD27, CD28 and programmed cell death 1 (PD-1). We used the common definition of CD8+ T cell subsets by phenotyping CD45RA/CCR7 and CD45RA/CD62L for definition of naive, central memory (TCM), effector memory (TEM) and late effector memory (TEMRA) T cells (exemplary stainings are shown in Fig. 1c). Generally, CD45RA/CCR7 and CD45RA/CD62L phenotyping revealed equivalent results for all four CD8+ subsets, as observed by both gating strategies for healthy donors (left panel). However, disparate results were obtained when applying these gating strategies to samples from MM patients. CD45RACD62L+ cells, commonly referred to as TCM, contained a large proportion of CCR7 cells, expressing greater amounts of perforin and granzyme A than their CCR7+ counterparts. We therefore defined naive T cells and TEMRA classically by their CD45RA and CCR7 expression (naive: CD45RA+CCR7+; TEMRA: CD45RA+CCR7). TEM and TCM were differentiated by their CD45RA and CD62L expression (TEM: CD45RACD62L) and the CD45RACD62L+ subset was divided into TCM (CD45RACD62L+CCR7+) and T cells with a TCM/TEM phenotype (CD45RACD62L+CCR7, Fig. 1d). Perforin/granzyme expression of these CD8+ T cell subsets revealed low cytolytic enzyme expression by TCM, moderate expression by TCM/TEM, high expression by TEM and highest expression by TEMRA (Fig. 1e). These observations strengthen our hypothesis that the CD45RACD62L+CCR7 subset should be classified between TCM and TEM cells with respect to their phenotype and cytokine profile.

We then explored the frequency of these CD8+ T cell subsets in untreated MM patients, patients treated with lenalidomide, patients treated with regiments excluding lenalidomide, MGUS patients and age-matched healthy donors. We found a highly significant increase (P = 0·001) of the TCM/TEM subset among CD3+CD8+ cells in lenalidomide-treated compared to untreated patients at the expense of naive (P = 0·023) and TEMRA (P = 0·027) cells (Fig. 2a). These findings were also reflected by absolute cell numbers, as lenalidomide-exposed patients exhibited significantly higher numbers of TCM/TEM (P = 0·003) while showing decreased numbers of naive cells (P = 0·014), but a similar cell count of total CD8+ T cells (Fig. 2b).

Figure 2.

Central memory T cells (TCM)/ effector memory T cells (TEM) CD8+ T cells are elevated in multiple myeloma (MM) patients treated with lenalidomide. (a) Frequency of CD8+ T cell populations in the peripheral blood from untreated MM patients (n = 30), lenalidomide-treated MM (n = 14), MM treated by therapies excluding lenalidomide (n = 13), monoclonal gammopathy of uncertain significance (MGUS) (n = 4) and age-matched healthy controls (n = 15). Lenalidomide treatment significantly increased the frequency of CD8+CD45RACD62L+CCR7 TCM/TEM in MM patients (median untreated = 2·4%, median Lena = 9·0%; P = 0·001) at the expense of naive cells (median untreated = 18·8%, median Lena = 8·0%; P = 0·023) and late effector memory RA (TEMRA) (median untreated = 51%, median Lena = 25%; P = 0·027). (b) Significantly elevated absolute numbers of TCM/TEM in lenalidomide-treated compared to untreated MM patients (median untreated = 6·3/μl, median Lena = 25·6/μl; P = 0·003). Naive CD8+ populations were decreased in lenalidomide-treated patients compared to untreated patients (median untreated = 41·0/μl, median Lena = 20·7/μl; P = 0·014). (c) Human leucocyte antigen D-related (HLA-DR) expression on CD8+ T cells and TCM/TEM distribution among CD8+ T cells for all treatment plans. Box-plots for HLA-DR expression represent untreated MM patients (ut) including newly diagnosed (n = 16), relapsed (n = 12) and patients with stable disease (n = 10); lenalidomide-treated patients (Lena) including patients treated by lenalidomide monotherapy (n = 12), by bortezomib, dexamethasone, cyclophosphamide and lenalidomide (VDCR) (n = 4), by lenalidomide and dexamethasone (RD) (n = 1) and by lenalidomide, dexamethasone and cyclophosphamide (RCD) (n = 1); patients treated by other therapies excluding lenalidomide (other ther.) including bortezomib, adriamycin and dexamethasone (PAD) (n = 10), VCD (n = 3), cyclophosphamide and dexamethasone (CD) (n = 1), bortezomib and dexamethasone (VD) (n = 1) and bortezomib, cyclophosphamide and dexamethasone (VCD) for elderly patients (n = 1); MGUS patients (n = 4) and untreated healthy controls (n = 24). Box-plots for TCM/TEM frequency represent untreated MM patients (ut) including newly diagnosed (n = 12), relapsed (n = 9), and patients with stable disease (n = 6); lenalidomide-treated patients (Lena) including patients treated by lenalidomide monotherapy (n = 9), by VDCR (n = 3) and by RD (n = 1); patients treated by other therapies excluding lenalidomide (other ther.) including PAD (n = 6), VCD (n = 3), VD (n = 1) and VCD for elderly patients (n = 1); MGUS patients (n = 3) and untreated healthy controls (n = 15). Boxes depict the interquartile range (IQR), the line the median and the whiskers the 95% confidence interval (CI). (d) TCM/TEM frequencies during lenalidomide therapy: lenalidomide induction or therapy stop in single patients. Boxes depict the IQR, the line the median and the whiskers the 95% CI.

Among lenalidomide-treated patients, those receiving lenalidomide monotherapy showed the most prominent increase in TCM/TEM frequency compared to VDCR- or RD-treated patients, suggesting a potential counteracting effect on lenalidomide-induced immunoactivation by dexamethasone (Fig. 2c). Similarly, patients during lenalidomide monotherapy exhibit higher frequencies of HLA-DR+ CD8+ T cells compared to VDCR- or RCD-treated MM patients.

We next focused on the frequency of markers within the different CD8+ subsets known to be associated with negative immune regulation (PD-1), T cell exhaustion (CD127) and senescence (CD27/28). We found no difference in all CD8+ T cell subsets concerning their expression of these markers between untreated and lenalidomide-treated patients (Supporting information, Fig. S1). The low frequency of CD27/CD28 double-negative cells and the moderate expression of CD127 within the TCM/TEM population indicate that they are not yet fully differentiated or senescent as TEMRA or TEM.

To illustrate the impact of lenalidomide on the occurrence of TCM/TEM in individual patients, we depict the results of immunomonitoring of several patients over time, before induction of a lenalidomide-based therapy, during and after stopping the therapy. We observed an induction in four of four patients' (lines for patients 4 and 7 are overlapping) TCM/TEM by lenalidomide therapy and a decrease in TCM/TEM frequency in two of three patients after stopping the therapy (Fig. 2d). The reason why some patients showed only minor or no response to lenalidomide treatment by increasing TCM/TEM frequency is unclear. We compared clinical data for responders and non-responders and could not detect any association between the response to lenalidomide and clinical parameters, staging of the disease, status before treatment, duration of lenalidomide therapy and type of therapy (lenalidomide combination or monotherapy, untreated or treatment excluding lenalidomide) (Supporting information, Table S1).

In addition to the increase of effector T cells, we also detected a lenalidomide-induced increase in mature regulatory T cells (described in Supplementary Fig. S2). Again, no specific clinical parameter was associated with the ability to respond to lenalidomide by upregulation of this population (Supporting information, Fig. S2c and Table S1, patients are identical).

Lenalidomide-treated patients exhibit a novel monocytic myeloid suppressor cell type characterized by CD15 expression

In addition to describing well-characterized immune populations such as cytotoxic T cells and Tregs, we next aimed at identifying possible novel immunoregulatory cells. MDSCs consist of a heterogeneous population described by various marker combinations. Whereas the granulocytic population is very difficult to define by surface markers in the human setting and can probably only be identified by their suppressive activity, monocytic MDSCs appear to be better defined. CD14+ monocytes with reduced expression of HLA-DR have long been known as disturbed immunoreactive cells accumulating under immunosuppressive therapy or sepsis [32-35], and are now being referred to as MDSCs with potent immunosuppressive function in various cancer types [36, 37].

To screen various MDSC subsets, we gated on MDSCs with common markers such as lineage (CD3/CD19/CD56)-negative and myeloid (CD33, CD11b)-positive. The CD33hi CD11bhi population contained the monocytic (CD14+) fraction (Fig. 3a); the broader gate of CD33+CD11b+ cells also contained granulocytic cells (data not shown). A well-described monocytic MDSC population is the CD14+HLA-DRlo subset [14, 37-40] (Fig. 3a, lower left). Usually, the CD15 marker is not expressed by the monocytic lineage. However, in MM patients we detected a significant proportion of CD15+ cells within the CD14+ population (Fig. 3a, lower right), whereas we could not detect significantly increased frequencies of the recently described CD33+CD11b+HLA-DRCD14CD15+ MDSCs in our much larger cohort of MM patients (data not shown) [15].

Figure 3.

Identification of a novel myeloid-derived suppressor cell (MDSC) population. (a) Gating strategy for CD14+ MDSC populations. Examples are shown for cells from a multiple myeloma (MM) patient treated with lenalidomide and from an age-matched healthy control. (b) CD15 and human leucocyte antigen D-related (HLA-DR) expression in the CD14+ population of freshly isolated peripheral blood mononuclear cells (PBMCs). Box-plots represent untreated MM patients (n = 36), lenalidomide-treated patients (n = 17), patients treated by therapies excluding lenalidomide (n = 15), monoclonal gammopathy of uncertain significance (MGUS) (n = 5) and age-matched healthy controls (n = 24). Lenalidomide treatment significantly increased the frequency of the CD15+ population among CD14+ cells in patients (median = 33·2%) compared to untreated MM patients (median = 8·1%, P = 0·001). (c) Significantly elevated absolute numbers of CD14+CD15+ cells in lenalidomide-treated compared to untreated MM patients (median untreated = 11·0/μl, median Lena = 64·4/μl; P = 5·35E-05). (d) CD14+CD15+ cell frequency among CD14+ monocytic cells for all treatment plans. Box-plots represent untreated MM patients (ut) including newly diagnosed (n = 16), relapsed (n = 12), and patients with stable disease (n = 10); lenalidomide-treated patients (Lena) including patients treated by lenalidomide monotherapy (n = 12), by bortezomib, dexamethasone, cyclophosphamide and lenalidomide (VDCR) (n = 4), by lenalidomide and dexamethasone (RD) (n = 1) and by lenalidomide, dexamethasone and cyclophosphamide (RCD) (n = 1); patients treated by other therapies excluding lenalidomide (other ther.) including bortezomib, adriamycin and dexamethasone (PAD) (n = 10), bortezomib, cyclophosphamide and dexamethasone (VCD) (n = 3), cyclophosphamide and dexamethasone (CD) (n = 1), VD (n = 1) and VCD for elderly patients (n = 1); monoclonal gammopathy of uncertain significance (MGUS) patients (n = 4) and untreated healthy controls (n = 24). Boxes depict the interquartile range (IQR), the line the median and the whiskers the 95% CI. (e) CD14+CD15+ frequencies during lenalidomide therapy in single patients. Boxes depict the interquartile range (IQR), the line the median and the whiskers the 95% confidence interval (CI).

Monitoring patient groups on the occurrence of these MDSC populations revealed significantly elevated frequencies of CD14+CD15+ cells in lenalidomide-treated compared to untreated patients (P = 0·001) and to all other groups (Fig. 3b). In contrast, we did not observe any difference in the proportion of the HLA-DRlo subset within the CD14+ monocytic compartment between the different groups.

The increase in CD14+CD15+ cell frequency among monocytes was also reflected by a significant absolute increase (P < 0·0001) in lenalidomide-exposed patients (Fig. 3c).

Differential analysis of the data revealed the highest increase in CD15+ cells in the monocyte compartment in patients receiving lenalidomide monotherapy and a moderate increase in VDCR- and RCD-treated groups (Fig. 3d). As already observed for TCM/TEM, dexamethasone seems to alleviate the immunomodulatory effect of lenalidomide with regard to CD14+CD15+ induction.

As observed for TCM/TEM CD8+ T cells and Tregs, treatment with lenalidomide was also associated with CD14+CD15+ cell induction (Fig. 3e). In four of six patients, the start of lenalidomide therapy increased these cells and in four (of four) patients removal of lenalidomide was followed by a drastic decrease in CD14+CD15+ cell frequency to pre-therapy levels. This supports the concept that lenalidomide-mediated immune effects are transient and not long-lasting. Again, we did not see any association in patient clinical and demographic characteristics and response to lenalidomide with respect to CD14+CD15+ cell induction/decrease (Supporting information, Table S1).

CD14+CD15+ myeloid cells suppress CD4+ and CD8+ T cell proliferation in vitro

To gain deeper insight into phenotype and function of the CD14+CD15+ myeloid cells population we characterized this population phenotypically and functionally in comparison to their CD15 counterparts. A lower expression level of HLA-DR and higher CD62L expression on the surface of these cells compared to CD15 monocytes could be observed (Fig. 4a). However, not all HLA-DRlo cells were CD15+, indicating that CD14+HLA-DRlo and CD14+CD15+ are distinct cell populations. All other markers (CD16, CD86, B7-H1, CD4) were expressed equally on both CD14+CD15+ and CD14+CD15 cells, or were absent (CCR7, CD124).

Figure 4.

Phenotypical and functional characterization of CD15+CD14+ myeloid cells. (a) Expression of various surface antigens by CD14+CD15+ cells. Upper dot-plots are gated on CD14+ cells, lower histograms represent CD14+CD15+ and CD14+CD15 cells, respectively. Stainings were performed for four donors with similar results; shaded histograms depict indicated markers, solid lines depict matching isotype controls. Representative phenotypes are shown. (b) Inhibition of CD8+ (right) and CD4+ (left) T cell proliferation by CD14+CD15+ myeloid-derived suppressor cells (MDSCs). T cells were carboxyfluorescein diacetate succinimidyl ester (CFSE)-stained and stimulated with anti CD3/CD2/CD28 beads and co-cultured with the indicated CD14+ populations at the indicated ratios. Data from one experiment out of three with similar results are shown. (c) Histograms for the plots shown in (b).

Importantly, CD14+CD15+ myeloid cells from lenalidomide-treated patients suppressed the proliferation of both CD4+ and CD8+ T cells in vitro when co-cultured with CD14-depleted PBMCs stimulated by anti-CD2/CD3/CD28 beads (Fig. 4b,c). No such effect was seen when CD14+CD15 cells were used.

Thus, in this study we document a novel MDSC population for the first time in MM patients and demonstrate its functional properties.

Activating and regulatory immune cells in lenalidomide-treated MM patients are closely associated

The tumour immune response is a combination of the interplay of numerous players but, commonly, only the relevance of one or two immune cell populations in a pathological setting or in response to a certain treatment is studied. Here, we aimed at investigating the connections between various immune cells in MM patients compared to MGUS patients and controls. In this setting, we focused particularly on the populations we found to be elevated in lenalidomide-treated patients, namely TCM/TEM, mature Treg and MDSCs.

We also scanned various other immune cell populations such as B cells, NK cells, NK T cells, the α/β and γ/δ T cell repertoire, dendritic cell subsets and the expression of various surface antigens on these cells, monocyte subsets with surface antigens and granulocytic MDSC populations. Except for changes within the NK cell compartment, we did not observe any additional disparities in other immune populations other than the described cells (TCM/TEM, Treg and CD14+CD15+ MDSC) between lenalidomide-treated and untreated patients. Within the NK cell compartment, we detected a correlating elevated expression of the activating receptors NKp30 (P = 0·029) and NKp46 (P = 0·028) in lenalidomide-treated compared to untreated patients (Supporting information, Fig. S3). NKG2D was down-regulated in all MM patients compared to healthy controls, as described previously [9], but no effect of lenalidomide treatment could be detected regarding the expression pattern of this receptor (data not shown).

To uncover the potential connections between elevated immune cell populations in lenalidomide-treated patients we correlated TCM/TEM, mature Treg, CD14+CD15+ MDSCs and NKp46 (Fig. 5a). Using this approach, we detected strong positive correlations between TCM/TEM and MDSCs (P = 0·0001), TCM/TEM and Treg (P = 0·002), MDSCs and Treg (P = 0·0001) and between MDSCs and NKp46 (P = 0·001). These correlations are depicted graphically in Fig. 5b.

Figure 5.

Correlations between various cell populations induced by lenalidomide treatment. (a) Correlations between central memory T cells (TCM)/effector memory T cells (TEM), effector regulatory T cell (Treg), CD14+CD15+ myeloid-derived suppressor cell (MDSC) and NKp46 expression in natural killer (NK) cells were evaluated using the non-parametric Spearman's rank test; Spearman's correlation coefficients (rs) and P-values are indicated in each scatterplot. Linear regression is shown as a solid line. (b) Schematic display of populations induced by lenalidomide treatment and their relationship.


In this observational study, we describe a comprehensive analysis of the immune system in myeloma patients who received various therapies. The analysis included subsets of helper and cytotoxic T lymphocytes, Tregs, NK cells, dendritic cells, monocytes and MDSCs as well as the functional potential of T cell subsets, NK cells and MDSCs. To date, this has not been performed in such detail.

By examining various subsets of CD8+ T cells we discovered a novel subset, defined phenotypically as CD45RACD62L+CCR7, that was elevated in lenalidomide-treated MM patients. With respect to its potential effector function, determined by perforin/granzyme production, this subset was staged between TCM and TEM. The low expression of CD27/CD28 and CD127 among these cells indicated that they are not yet fully differentiated. Further analysis of these cells is subject to future studies.

Our observations that activated effector T cells are elevated in lenalidomide-treated MM patients are in line with other studies showing that lenalidomide is generally associated with enhanced T cell reactivity reflected by elevated HLA-DR expression [26] and increased cytokine levels, such as IFN-γ [23, 24, 41, 42]. A decreased expression of CD45RA has also been associated with lenalidomide treatment by Neuber et al. [23], who investigated the differentiation of T cells from MM and MGUS patients in vitro and in vivo. They also observed an increased antigen-specific activity within the T cell pool in the presence of lenalidomide but did not differentiate subsets further within the CD45RA population. The authors only observed a switch from a naive into a more mature state of T cells, but found divergent results for the expression of CCR7 and CD28 and did not focus on CD62L/CCR7. Two other groups documented elevated frequencies of TCM, characterized by CD45RO and CD62L expression [24, 42], similar to our findings. Their studies characterized this population in MM patients in response to a pneumococcal vaccine under lenalidomide therapy and also found elevated IFN-γ levels within the total CD8+ T cell population. However, the authors did not examine the CCR7 expression within this population nor their cytolytic enzyme profile or expression of other markers.

Our observation of TCM/TEM induction by lenalidomide and the detailed phenotypical characterization of these cells not only confirms these observations, but also adds a novel piece of information by defining this population using various novel markers, including perforin/granzyme, IFN-γ, PD-1, CD127 and CD27/CD28 expression. It will be crucial to analyse the functionality and tumour specificity of these cells in more detail in the future as well as their potential role in GVHD in lenalidomide-treated allotransplant recipients.

Our data on the elevated Treg frequencies in lenalidomide-treated MM patients are in line with the observations made by others [23, 43], and might represent a counter-regulatory reaction to the increase of activated T cells (TCM/TEM). If that were the case, the question is whether this counter-regulation is dominant. As there was no difference in the effector/Treg ratio between lenalidomide and untreated patients (data not shown), there seems to be no overall immunosuppression by Tregs. Others report the development of severe acute GvHD in lenalidomide-treated, allo-SC-transplanted patients, despite high levels of Tregs in the peripheral blood [43], indicating that Tregs may not always be capable of neutralizing the immunostimulatory effects of lenalidomide. Furthermore, lenalidomide has been shown to inhibit the proliferation and function of Tregs in vitro [44], possibly explaining the occasionally observed incapacity of Tregs to inhibit T cell-mediated immunity.

In addition to changes within the T cell population, we detected significant changes within the monocytic compartment. In this paper we document a novel MDSC population for the first time in MM patients and provide phenotypical and functional characterization of these cells. Only one publication to date has described this particular phenotype in patients with colorectal cancer, but the authors provided no functional data [45]. So far, the origin of CD14+CD15+ MDSCs remains unclear. They might derive from an immature precursor in the bone marrow or from CD14+ monocytes, as lenalidomide has been reported to modulate monocyte function [46]. However, we could not observe a direct effect of lenalidomide on the development of CD14+CD15+ MDSCs from monocytes, as we could not generate CD14+CD15+ MDSC by exposing isolated monocytes or unseparated PBMCs to lenalidomide in vitro (data not shown).

In our study, we could not confirm the induction of CD14+HLA-DRlo MDSCs in untreated MM patients that was described by others [14]. However, in the publication by Brimnes et al. the groups were substantially smaller than ours (eight patients, treatment not described, and 10 controls; as opposed to 26 untreated patients and 24 controls in our study) and the report is somewhat descriptive without providing functional data.

Conversely, we identified a novel monocytic (CD14+) MDSC subset expressing CD15 in MM patients receiving lenalidomide. This MDSC population was able to suppress CD4+ and CD8+ T cell proliferation, proving its functional relevance in immunoregulation.

As patients receiving combination therapy of lenalidomide and dexamethasone exhibit lower frequencies of TCM/TEM and CD14+CD15+ MDSCs, dexamethasone treatment seems to alleviate the immunoregulatory effect of lenalidomide. This observation is in line with observations made by others, reporting an antagonism of dexamethasone to the immunostimulatory effects of lenalidomide on T cell-derived IL-2 and NK-derived IFN-γ production [18].

In our study we could show that the immunomodulatory effects of lenalidomide on CD8+ effector T cells (TCM/TEM), mature Tregs, NK cells and CD14+CD15+ MDSCs were closely related. Induction of immunostimulatory cells, such as TCM/TEM and activated NK cells, correlated strongly with the induction of immunosuppressive cells, such as Tregs and monocytic MDSCs. Our findings support the hypothesis that upon activation of the immune system (such as T cell activation by antigen-presenting cells) endogenous counterbalancing takes place, e.g. by induction of immunoregulatory cells, such as Tregs or MDSCs. We propose that this is the explanation for divergent results regarding the exact mechanism and complexity of immunoregulation of lenalidomide.


We acknowledge the assistance of E. Endl, A. Dolf and P. Wurst from the Flow Cytometry Core Facility at the Institute of Molecular Medicine and Experimental Immunology, University of Bonn for cell sorting of MDSC subsets.

Author contributions

A. B. performed the study and wrote the paper. She designed and performed all experiments and delivered the resulting data to D. Z. and L. F. for transfer into the databank. She also performed the statistical analysis of the data. M. v. L.-T. designed the study and revised the paper. D. Z. and L. F. collected informed consent from patients and healthy controls and transferred all experimental data into the spss databank. The also documented all clinical data. V. J., D. W. and O. M. recruited patients for the study and gave informed consent to patients to be included in the study. They and L.-O. M. revised the paper. C. H.-A. is the study coordinator of the clinical department and coordinated patient recruitment. P. B. is the head of the department of Hematology, Oncology and Rheumatology of the University Hospital of Bonn.


M. v. L.-T. received honoraria from Celgene and Janssen-Cilag. D. W. received honoraria from Novartis, Amgen, Pfizer and BMS, obtained research support from Novartis, Amgen and Roche and is on the advisory board of Novartis, BMS, Pfizer and Amgen. L.-O. M. received honoraria from Celgene, Novartis and Jannsen-Cilag as well as research funding by Celgene. Apart from this, the authors report no financial relationships or conflicts of interest regarding the content herein.