Biological and Clinical Relevance of miRNA Expression Signatures in Primary Plasma Cell Leukemia Free

Plasma cell leukemia (PCL) is a rare variant of multiple myeloma, characterized by poor prognosis and presence of malignant plasma cells (PCs) circulating in the peripheral blood (1). Diagnostic criteria defined by the International Myeloma Working Group consist of an absolute PC count of more than 2.0 × 10⁹/L or a relative proportion of greater than 20% of the peripheral blood leukocyte count, with concomitant evidence of monoclonal gammopathy (1, 2). PCL can be subdivided clinically into primary and secondary types: primary PCL (pPCL) presents de novo in the leukemic phase, while secondary PCL (sPCL) arises in the context of a preexisting multiple myeloma (1). Compared with multiple myeloma, patients with pPCL show some clinical and biological peculiarities, in that they are younger and at a more advanced clinical stage, have more frequent tumor-related organ or tissue impairment, although with less bone involvement, and higher proliferative activity of neoplastic cells (3, 4). Furthermore, patients with pPCL show a higher prevalence of TP53 inactivation and rearrangements involving MYC, both relatively rare and late genetic events in multiple myeloma that are correlated with disease progression (5, 6). All of these adverse characteristics result in short survival, and no standard effective treatments are currently available for these patients. Several studies have investigated PCL in terms of immunophenotype and molecular cytogenetics (5–10). In contrast, to date, analyses of other molecular aspects of the disease, such as transcriptomic and microRNA (miRNA) profiling, are almost completely lacking (11).

MiRNAs are small, evolutionarily conserved noncoding RNAs that bind to the 3′-untranslated region (UTR) of target mRNAs, resulting in translation repression or mRNA degradation, and play important roles in cellular processes such as proliferation, development, differentiation, and apoptosis. Aberrant miRNA expression in multiple myeloma has been reported by us and other groups, showing that miRNA deregulation characterizes the progression of the disease, affects important pathways involved in multiple myeloma cell survival, and reflects the different multiple myeloma genetic subtypes (12).

Here, we describe global miRNA expression profiling in the context of a prospective series of patients with pPCL compared with a representative cohort of patients with intramedullary multiple myeloma. Our data may have implications for the biological and clinical features of this very aggressive form of plasma cell dyscrasia.

Bone marrow (BM) aspirates from patients with pPCL were obtained during standard diagnostic procedures. Patients with PPCL had been diagnosed on the basis of previously described criteria (1, 2) and belonged to a national, multicenter, pilot clinical trial including 23 patients (12 males and 11 females; median age at diagnosis: 60 years, range: 44–80 years). This study aimed to evaluate safety and antitumor activity of the lenalidomide and dexamethasone combination in previously untreated pPCL (RV-PCL-PI-350, registered at www.clinicaltrials.gov as NCT01553357; ref. 13). The primary endpoint was the response rate, defined according to the International Uniform Criteria (4, 14); secondary endpoints were safety, progression-free survival (PFS), and overall survival (OS; see Supplementary Methods for details). All patients gave their informed consent for molecular analyses. Normal control PCs (NCs) were obtained from 4 healthy donors.

PCs were purified using CD138 immunomagnetic microbeads (MidiMACS, Miltenyi Biotec). The purity of the positively selected PCs (≥90%) was assessed using flow cytometry. All pPCL cases were investigated by fluorescence in-situ hybridization (FISH) for the major immunoglobulin heavy chain locus (IGH@) translocations and genetic lesions: results are reported in Table 1 and described extensively in ref. 15.

MiRNA expression profiling was carried out in 18 out of 23 patients with pPCL for whom material was available (molecular and clinical data of the 18 patients with pPCL are reported in Table 2). A series of 39 multiple myelomas, most of which were included in our previous report, were also profiled for global miRNA expression (16); patients were selected on the basis of their representativeness with respect to molecular characteristics. Overall, 9 patients with multiple myeloma had t(11;14), 7 had t(4;14), 2 had t(14;16), and 1 had t(14;20). The frequencies of the major genetic lesions in patients with multiple myeloma were globally consistent with those reported in the literature (Table 1); only del(17p) occurred in a slightly lower percentage (17).

Total RNA samples were profiled on the Agilent Human miRNA Microarray V2 (Agilent Technologies) as previously described (16) and detailed in Supplementary Methods. The raw and normalized miRNA data are available through GEO accession number GSE37053. Hierarchical agglomerative clustering of the samples was carried out using Pearson's correlation coefficient and average linkage as distance and linkage metrics, respectively, on those probes whose average change in expression levels varied at least 2-fold from the mean across the dataset; P-value threshold for sample enrichment was set at 0.0005. Supervised analyses were carried out using samr package for Bioconductor (Supplementary Methods). Enrichment analysis of miRNA categories was done using TAM (Tool for Annotations of MicroRNAs, V2; http://cmbi.bjmu.edu.cn/tam).

The differentially expressed miRNAs were validated using an independent multiple myeloma cohort (GEO series GSE17306; ref. 18) generated on Agilent Human microRNA Microarray V1, including 52 patients with multiple myeloma and 2 healthy donors (Supplementary Methods). Supervised analysis was carried out using the same approach (SAM, q-value = 0) applied in the proprietary series.

Selected mature miRNAs underwent quantitative real-time PCR (qRT-PCR) using TaqMan microRNA assays (Applied Biosystems) in accordance with the manufacturer's protocol. All of the RNA samples were normalized on the basis of the RNU48 or RNU44 TaqMan miRNA Assays-Control. MiRNA expression was relatively quantified using the 2^−ΔCt method (Applied Biosystems, User Bulletin No.2).

The transcriptional profiles of the patients were generated using GeneChip Gene 1.0 ST Array (Affymetrix Inc.; Supplementary Methods). The functional annotation analysis on the selected lists was conducted using the Database for Annotation, Visualization and Integrated Discovery 6.7 tool (DAVID, http://david.abcc.ncifcrf.gov/).

Sixteen of the 18 pPCL samples profiled for miRNA expression underwent genome-wide DNA analysis using Affymetrix GeneChip Human Mapping 250K NspI microarrays as recently described (15).

To evaluate whether gene dosage effects could be identified for a specific miRNA, the relationship between miRNA expression and the inferred copy number (CN) of the corresponding miRNA gene (according to NCBI36/hg18) was measured by using the Kendall tau test. The analyses were conducted using appropriate R software packages.

The integrated analysis of target predictions and miRNA and gene expression data was conducted using scripts for R software developed in our laboratory. In particular, we assessed the correlation between the expression data of all possible miRNA–gene pairs in 18 cases of pPCL and 36 cases of multiple myeloma, and for each significantly anticorrelated pair (using Pearson correlation coefficient and setting the threshold for Benjamini–Hochberg P-adjusted at 0.05), the existence of a predicted targeting relationship was verified by 9 computational algorithms. Specifically, queries were submitted to miRecords (http://mirecords.biolead.org) and results were retrieved for the following databases: DIANA-microT, RNA22, NBmiRTar, MirTarget2, PicTar, miRanda, and RNAhybrid; in addition, updated precompiled datasets were downloaded from PITA (http://genie.weizmann.ac.il/pubs/mir07/mir07_data.html) and TargetScan (http://www.targetscan.org/). Then, we selected those genes predicted as targets by at least 5 miRNA–target-prediction programs.

The 293T cells were plated at 600,000 cells per well in a 6-well plate. The next day, cells were transfected with 30 nmol/L of specific mirVana miRNA mimics (Ambion) or pre-miR negative control (Ambion), 3.0 μg of pMir target firefly luciferase reporter plasmid containing 3′-UTR sequences from the putative target gene of interest (Origene), and 0.1 μg Renilla luciferase expression plasmid pRL-SV40 (Promega) using Lipofectamine 2000 (Invitrogen). At 24 hours post-transfection, cells were analyzed using the Dual Luciferase Assay Kit (Promega) according to the manufacturer's instructions. Each sample was prepared in triplicate and the entire experiment was repeated twice.

Synthetic pre-miRNAs or miRNA inhibitors were purchased from Ambion (Applied Biosystems). Multiple myeloma cells RPMI-8226 and NCI-H929 (authenticated by Short Tandem Repeats DNA typing) were purchased from DSMZ (Germany) in October 2012. Cells were transfected with 100 nmol/L miR-22 or miR-126 mimics, miR-330, or miR-146 inhibitors, or scrambled oligonucleotide (NC) as control, by Neon Transfection System (Life Technologies) as previously reported (19). The transfection efficiency of these cell lines was evaluated by flow cytometric analysis of FAM-dye-labeled synthetic miRNA inhibitor transfection as previously reported (19). Cells were collected at different time points after electroporation (24, 48, and 72 hours) and processed for qRT-PCR and viability assays. In particular, for cell proliferation analysis, 2 × 10⁵ cells were seeded after electroporation in 96-well plates and then tested with the Cell Counting Kit-8 (CCK-8, Dojindo Molecular Technologies) colorimetric assay according to the manufacturer's instructions. At the same time, cell viability was estimated by ATP quantification with Cell Titer Glo Luminescent cell viability assay and GloMax-multi detection system (Promega). All assays were conducted in triplicate and repeated twice.

The correlation between the expression levels of each tested miRNA and OS or PFS was tested using the Cox proportional hazards model in the globaltest function of R software. Patients were stratified into 2 groups using sliding thresholds on the expression levels of the most significant miRNAs identified by globaltest, and the groups identified by this approach were then tested for association with survival. Survival analysis was conducted with survcomp package in R software, using the Kaplan–Meier estimator and log-rank test, and P values were calculated according to the standard normal asymptotic distribution.

MiRNA expression profiling was carried out in 18 pPCLs, detecting 269 miRNAs that were expressed above background levels in at least 1 sample. The hierarchical clustering of the samples based on the 46 most variable miRNAs across the dataset is reported in Supplementary Fig. S1A: the groupings appeared unrelated to known molecular characteristics as indicated above the heatmap.

The supervised 3-class comparison of miRNA expression between pPCL cases carrying different IGH@ chromosomal translocations identified 7 differentially expressed miRNAs, of which 3 were associated with t(4;14) [let-7e, miR-135a, and miR-148a], 3 with MAF translocations (miR-7, miR-7-1*, and miR-454) and 1 with t(11;14) [miR-342-3p; Supplementary Fig. S1B]. Notably, some of these miRNAs had already been found in association with specific molecular subgroups in a proprietary dataset of patients with multiple myeloma (16).

The influence on miRNA expression of the allelic imbalances detected at DNA genome-wide level was evaluated by conducting an integrative analysis of mature miRNA expression and the inferred DNA CN values of the corresponding miRNA gene/s available for 16 of 18 pPCL cases. In particular, the most recurrent somatic CN alterations identified were represented by gain of all or part of the long arm of chromosome 1 (8/16, 50%), and losses involving all or part of chromosomes 13q (12/16, 75%), 16q (9/16, 56%), 1p (7/16, 44%), 14q (7/16, 44%), 17p (6/16, 38%), 8p (6/16, 38%), and 6q (4/16, 25%; data not shown). Taking into consideration all miRNA genes except those located on chromosome X, the expression of 23 miRNAs resulted to be significantly correlated with DNA CN on different chromosomes, with chromosome 13 (22%) and the short arm of chromosome 1 being the most involved (22%; Supplementary Table S1). Specifically, we identified miR-15a at 13q14.3 and 4 members of the cluster mir-17–92 (i.e., miR-19a, miR-20a, miR-20a*, and miR-92a) at 13q31.3, and 5 mature miRNAs (miR-30e, miR-30e*, miR-30c, miR-186, and miR-197) at 1p. In addition, we highlighted 2 miRNAs (miR-210 and miR-483-5p) encoded by genes mapped at 11p15.5; 2 miRNAs (miR-140-5p and miR-1225-5p) encoded by genes mapped to chromosome 16, respectively to 16q22.1 and 16p13.3; and mir-22, mapped to 17p13.3.

To determine whether the natural grouping of miRNA expression profiles might distinguish PCs from NCs, multiple myeloma, and pPCL, we conducted an unsupervised analysis of our dataset including 4 NCs, 39 multiple myelomas, and 18 pPCLs. On the basis of the 76 most variable miRNAs across the dataset, healthy donors represented a clearly distinct transcriptional entity, and all pPCL samples were grouped together, along with 8 multiple myeloma samples (Fig. 1A); interestingly, all the multiple myeloma cases carrying t(14;16) or t(14;20) were included in the pPCL cluster, whereas all but 2 of the t(4;14) and 2 of the t(11;14) multiple myeloma cases were grouped in the multiple myeloma main cluster. Prompted by these observations, we evaluated in pPCL cases the expression of miRNAs previously identified as distinctive of multiple myeloma translocation/cyclin D (TC) groups (16); notably, the overall pattern of miRNA expression in pPCL was globally similar to that of the TC5 multiple myeloma molecular subgroup, regardless of the presence of MAF translocations (Fig. 1B), with the only exception of those miRNAs that strongly discriminated multiple myeloma TC4 cases [miR-125a-5p, miR-99b, and let-7e, which were found to be overexpressed also in t(4;14) pPCLs].

The direct comparison of miRNA expression profiles between pPCL and multiple myeloma samples by supervised analysis identified 42 upregulated and 41 downregulated miRNAs in the pPCL group (Table 3; Supplementary Fig. S2A); the upregulation of miR-155, miR-21, miR-142-3p, miR-142-5p, and miR-103 in pPCL versus multiple myeloma patients was confirmed by conducting qRT-PCR in all samples for which RNA was available (11 pPCL and 30 multiple myeloma patients; Pearson correlation coefficient = 0.97, 0.68, 0.65, 0.66, and 0.68, respectively). The TAM enrichment analysis of miRNA categories of the list of 42 upregulated miRNAs identified as most significant (FDR < 15%) the over-representations of 4 classes of miRNAs, namely miRNAs defined as onco-miRNAs or involved in immune response, immune system, or hematopoiesis (Supplementary Fig. S2B). No enriched functional category was found in the list of the 41 downregulated miRNAs. Notably, when comparing our pPCL cases with multiple myeloma cases of an independent dataset (GSE17306), we found that 41 (80%) of the 51 miRNAs represented on Agilent microRNA Microarray V1 showed the same trend of differential expression between pPCL and multiple myeloma (Table 3). In addition, on the basis of the hypothesis that the deregulation of these miRNAs might be compatible with the “strength” of neoplastic transformation, we investigated whether, among the 83 differentially expressed miRNAs between multiple myeloma and PCL, a trend might be identified in relationship with normal donors. Interestingly, for 56 of 83 (approximately 70%) miRNAs differentially expressed between pPCL and multiple myeloma, the trend of expression (increasing or decreasing) from healthy controls, through multiple myeloma, to pPCL was significantly maintained (Jonckheere–Terpstra test, P value < 0.005; Table 3; Supplementary Fig. S2C).

To gain further insights into the function of differentially expressed miRNAs between pPCL and multiple myeloma cases, we conducted an integrative analysis of miRNA and gene expression data combined with miRNA target prediction. This approach can only identify putative target genes whose regulation involves mRNA degradation rather than translation repression. The analysis was carried out on 36 out of 39 multiple myeloma samples and all 18 pPCL cases, for which both miRNA and gene expression profiling (GEP) profiling were available. Seventy-four statistically significant negative correlations involving differentially expressed miRNAs and putative target genes were computed (Supplementary Table S2A). These targeting relationships were exerted by 24 distinct miRNAs (11 down- and 13 upregulated) on 69 different genes and, in 5 cases, 1 gene was anticorrelated with more than 1 miRNA. Functional annotations were generated according to DAVID tool (Supplementary Table S2B). Interestingly, most of these anticorrelated genes were differentially expressed between pPCL and multiple myeloma samples (as highlighted in Supplementary Table S2A). We then conducted a target validation assay for some of the anticorrelations found in our analysis, focusing on genes that, based on available literature data, might have potential involvement in plasma cell dyscrasia, such as ACVR1/miR-301a and miR-301b (Supplementary Fig. S3A and S3B) as well as SULF2/miR-330-3p and TNFAIP3/let-7a. Interestingly, luciferase expression from the 3′UTR sequence of ACVR1 was suppressed by almost 20% (1-sided Student t test, P = 0.002) in 293T cells cotransfected with pACVR1 3′UTR, miR-301a, and miR-301b (Supplementary Fig. S3C) while the same approach did not confirm the putative SULF2/miR-330-3p and TNFAIP3/let-7a anticorrelations (data not shown), suggesting that such identified anticorrelations might have occurred by chance or due to mechanisms other than direct or exclusive targeting. ACVR1 codes for the receptor of bone morphogenetic protein 4 (BMP4), and was recently reported to be increased in multiple myeloma samples compared with normal controls and correlated with the level of plasma cell infiltration (20). These data, although preliminary, provide an important basis for further validation experiments, necessary to determine the actual occurrence of putative targeting relationships and, thus, the final impact of miRNA deregulation on gene expression.

Furthermore, we evaluated miRNA expression in the context of clinical outcome of our pPCL series, including all cases except 1, lost at follow-up. Specifically, we aimed to assess whether the occurrence of a miRNA signature at diagnosis might be associated with response to treatment, PFS, or OS. Considering the primary endpoint of our study, we investigated the dataset seeking differentially expressed miRNAs in patients who failed to respond to frontline therapy. The analysis led to the identification of 4 mature miRNAs (miR-106b, miR-497, miR-181b, and miR-181a*) upregulated in non-responder patients compared with responder ones, including complete response (CR), very good partial response (VGPR), and partial response (PR; Supplementary Fig. S4). No specific differentially expressed miRNA could be evidenced when PR, VGPR, and CR were considered as different classes of response. This finding will helpfully integrate results on efficacy and side effects of the first-line treatment of lenalidomide/dexamethasone in pPCL (Musto and colleagues, manuscript in preparation).

Furthermore, we assessed the relationship between each of the most variable miRNAs across the dataset (i.e., those whose average ratio of the expression values on the mean was >1.5) and either PFS or OS, representing the secondary endpoints of the prospective protocol. Of the 114 most variable miRNAs, 2 reached a significant correlation (P < 0.01) with PFS (miR-22 and miR-146a; Fig. 2A and B), allowing the division of samples into 2 groups with different outcome. The expression of the 2 miRNAs retained independency from all the molecular characteristics available, as well as from age, sex, LDH levels, renal function, and hematologic parameters (Supplementary Fig. S5A and S5B). As regards OS, 2 miRNAs reached a significant correlation with the clinical endpoint (miR-92a and miR-330-3p;Fig. 2C and D). In multivariate analyses, none of the 2 miRNAs were independent of patients being subjected to autologous stem cell transplantation, indicating that this therapeutic approach points definitively toward a more favorable outcome. In addition, the model based on miR-330-3p expression lost its independency when combined with the occurrence of del(8p) (Supplementary Fig. S5C and S5D). It is worth mentioning that none of the cytogenetic aberrations was associated per se with PFS and OS (15).

Finally, on the basis of in silico evidences, the biological role of some miRNAs potentially related to the aggressiveness of PC dyscrasia was studied in vitro by testing the effect of their manipulation on viability of multiple myeloma cell lines. In particular, miR-146a and miR-330-3p (the first associated with reduced PFS in pPCL cases and the latter upregulated in pPCL and involved in the outcome of patients with pPCL) were inhibited in the RPMI-8226 cell line, whereas miR-22 and miR-126 (the first associated with better PFS in pPCL cases and the second downregulated in pPCL) were forcedly expressed in NCI-H929 cell line. The transfection efficiency was more than 70% in all cases, as determined by FAM-dye-labeled oligonucleotide transfection with subsequent flow cytometric analysis. As reported in Fig. 3, we found significantly decreased cell survival as assessed by MTT assay after transfection of all molecules, particularly in response to miR-22 and miR-126 enforced expression; these data were confirmed by a cell viability assay that was based on ATP quantification (data not shown).

In the present study, we provided information concerning the pPCL miRNA profiles associated with the most recurrent chromosomal abnormalities, compared with a representative series of patients with multiple myeloma, or putatively involved in clinical outcome. To the best of our knowledge, this is the first report investigating global miRNA expression patterns in a prospective representative cohort of newly diagnosed patients with pPCL.

Concerning the impact of chromosomal alterations on miRNA expression, IGH@ translocations seemed to be less associated with distinct miRNAs profiles compared with what we observed in patients with multiple myeloma, although some miRNAs deregulated in a similar manner have emerged. An interesting similarity emerged between miRNA expression profiles of pPCL and multiple myeloma TC5 patients, suggesting further investigations on whether pPCL might share some clinicobiological features with MAF-translocated multiple myeloma.

With regard to numerical chromosomal alterations, several deregulated miRNAs mapped to hotspot altered regions, such as chromosomes 13 and 1. Specific CN/miRNA expression correlations found in the present series of pPCLs were previously identified by us in patients with multiple myeloma and/or human myeloma cell lines, that is, miR-342-3p, miR-22, miR-19a, miR-20a, miR-20a*, miR-140-5p, miR-210, miR-15a, miR-30e, and miR-30e* (16, 21). In particular, several of the identified downregulated miRNAs in association with the loss of the corresponding genomic loci have already been linked to cancer as having a tumor suppressor role: this is the case of miR-22 (22, 23), miR-30e and miR-30c (24), miR-331-3p (25), and miR-342-3p (26). The downregulation of miR-15a in patients carrying chromosome 13 deletion is of particular importance given the frequency of this genomic lesion in pPCL and the experimental evidence that it may act as a tumor suppressor in malignant plasma cells (27, 28). As regards miR-17-92 cluster, copy number of chromosome 13 was unlikely to be the only factor affecting its expression in our pPCL cohort; in fact we found that some miRNAs belonging to the cluster were significantly overexpressed in pPCL compared with multiple myeloma cases independently of their CN status. This appears to be in line with the reported evidence of the oncogenic potential of this miRNA cluster, as also specifically shown in the context of myeloma cells (29, 30).

The differences in miRNA expression profiles of multiple myeloma and pPCL highlighted by the unsupervised analysis were confirmed by direct comparison of these clinical entities: patients with pPCL and multiple myeloma were found to differentially express 83 miRNAs, roughly half of which were overexpressed in 1 group with respect to the other. It is worth noting that the expression of approximately 70% of these miRNAs maintained a statistically significant trend when considering normal donors. Furthermore, upregulated miRNAs in pPCL were enriched in “onco-miRNAs”, such as miR-21, miR-155 (31, 32), and miRNAs belonging to the paralogous clusters miR-17–92 and miR-106a∼363 (miR-18a, miR-19a, miR-18b and miR-20b). In particular, miR-21 was found to be overexpressed in the majority of human cancers, acting as a cancer biomarker. Its relevance in multiple myeloma was first suggested by Loffler and colleagues, who showed that miR-21 transcription is controlled by IL-6 through a mechanism involving STAT3, and that its ectopic expression gives independence from the IL-6-growth stimulus (33). More recently, we provided evidence that antagonism of miR-21 exerts anti-multiple myeloma activity in vitro and in vivo (34). Importantly, miR-21 expression in multiple myeloma cells was significantly enhanced by the adherence of cells to human BM stromal cells (hBMSC), and anti-multiple myeloma activity of miR-21 inhibitors was exerted also in the context of BM milieu, antagonizing the protective role of BMSCs on multiple myeloma cells (34). Globally, these findings suggest that the independence of the leukemic cells from the BM microenvironment might be closely related to the increased expression of miR-21 in pPCL compared with patients with multiple myeloma. MiR-17–92 functions pleiotropically during both normal development and malignant transformation to promote proliferation, inhibit differentiation, increase angiogenesis, and sustain cell survival, and its overexpression has been observed in multiple tumor types (35). In addition to the initial evidence of its important role in multiple myeloma (29), a recent study showed that MYC may inhibit multiple myeloma cell apoptosis by activating the miR-17–92 cluster, leading to the down-modulation of the proapoptotic protein BIM. Interestingly, patients with multiple myeloma with higher expression of miR-17, miR-20, or miR-92 had shorter PFS (30). On the contrary, several miRNAs with lower expression in pPCL compared with multiple myeloma are thought to play a tumor suppressor role in various tumor types, such as miR-663 (36, 37), miR-193b (38–40), and miR-126 (41), as also confirmed by our preliminary functional studies in HMCLs, and let-7a (42). Interestingly, let-7a has been reported to be repressed by, and in turn repress, MYC (42), and to be involved in the IL-6 pathway (43, 44). Notably, high expression levels of miR-142-5p, miR-21, miR-125b, miR-103, miR-99a, miR-26b, or members of the cluster miR-17–92, all of them upregulated in pPCL, were found to be associated with GEP-defined high-risk score in patients with multiple myeloma (18).

Several overexpressed miRNAs in pPCL also seem to be related to immune response, particularly the abovementioned miR-17–92 cluster (35), miR-155 (31, 32), and miR-21 (45, 46). Specifically, miR-155 and miR-21, known to share an important role in tumorigenesis, may represent important links between cancer and inflammation, likely through their relationship with the NF-κB pathway, known to promote their transcription (47). Furthermore, miR-301a, another upregulated miRNA in pPCL, has been shown to be activated by NF-κB (47). Both miR-21, miR-155, and miR-301a in turn affect the NF-κB pathway, leading to its activation or inhibition, in part dependent on the cellular context (47, 48). The repression of a member of NF-κB pathway, IKK-α, has been reported also for miR-223 (47), a hematopoietic specific miRNA involved in several types of leukemia and solid tumors. Mir-223 was downregulated in our pPCL series and, interestingly, its absence has recently been observed in extramedullary plasmacytoma (49); it was also found to suppress cell proliferation by targeting IGF-1R (50), involved in motility and invasiveness control in multiple myeloma cells.

The patients investigated in this study were included in a prospective clinical trial specifically designed for the initial therapy of pPCL, in which a combination of lenalidomide and low-dose dexamethasone was assessed (Musto and colleagues, manuscript in preparation). In this context, we investigated whether specific miRNA signatures could be associated with the response rate after 4 cycles of therapy, which represented the primary endpoint of the study. Interestingly, a 4-miRNA signature (miR-106b, miR-181a*, miR-181b, and miR-497) was found to be significantly upregulated in non-responding cases. Notably, both miR-106b and miR-181b were reported to be upregulated in multiple myeloma cells compared with healthy PCs (27, 29) and associated with increased multiple myeloma GEP-defined risk score (18), and, interestingly, miR-181a* was 1 of the upregulated miRNAs in pPCL versus multiple myeloma cases; this finding makes this miRNA a potential therapeutic target in pPCL. Moreover, a further contribution to clinical prognostication was represented by identification of 4 additional miRNAs, the expression of which was associated with pPCL cases with a shorter PFS (i.e., downregulation of miR-22 and upregulation miR-146a) or OS (downregulation of miR-330-3p and upregulation of miR-92a). As regards miR-146a, its deregulation has been associated with the pathogenesis of several human diseases, including solid tumors and hematopoietic malignancies; it is thought to play an important role in the regulation of innate immune and inflammatory responses through a negative feedback pathway involving NF-κB (47). This finding, together with previously described interactions between deregulated miRNAs and NF-κB pathway, warrants further investigations to clarify whether the miRNAs/NF-κB network may have a role in pPCL outcome. Notably, miR-22 was found significantly downregulated in our series in accordance with allelic loss of its residing locus at 17p13.3; however, TP53 deletion per se was not associated with PFS (data not shown), thus suggesting that factors other than deletion of 17p may affect miR-22 expression. Mir-22 has already been linked to cancer through its putative tumor suppressor role; it is activated by TP53, suppresses NF-κB activity and is thought to repress the MYC-binding proteins MAX and MYCBP and to be inhibited by MYC itself (22, 23). Our data were also corroborated by experiments conducted in multiple myeloma cell lines, which showed a pro- and antisurvival effect exerted respectively by miR-146a and miR-22. As regards miR-92a, it is a member of the abovementioned cluster miR-17–92, and its upregulation in multiple myeloma has been reported in comparison with normal PCs (29), in relation to increased multiple myeloma GEP-defined risk score (18), and in association with shorter PFS (30). Considering that pPCL represents a high-risk clinical entity per se, biological information at diagnosis could be helpful to guide clinicians in therapeutic decisions; therefore, larger prospective series of patients would be required to better elucidate the clinical relevance of miRNAs in pPCLs.

Overall, our study represents the first attempt to investigate the involvement of miRNAs in pPCL, the most aggressive form of plasma cell dyscrasia, and may contribute toward the development of functional approaches to analyze the activity of deregulated miRNAs and their possible role as novel therapeutic targets.

No potential conflicts of interest were disclosed.

Conception and design: M. Lionetti, P. Musto, S. Fabris, A. Neri

Development of methodology: P. Omedè

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): M. Lionetti, P. Musto, M.T. Di Martino, S. Fabris, K. Todoerti, M.E. Gallo Cantafio, V. Grieco, G. Bianchino, F. D'Auria, T. Statuto, C. Mazzoccoli, L. De Luca, M.T. Petrucci, F. Di Raimondo, A. Falcone, T. Caravita, A. Palumbo

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): M. Lionetti, M.T. Di Martino, S. Fabris, L. Agnelli, K. Todoerti, G. Tuana, L. Mosca, F. Morabito, M. Boccadoro

Writing, review, and/or revision of the manuscript: M. Lionetti, P. Musto, L. Agnelli, M. Offidani, F. Morabito, P. Tassone, A. Palumbo, A. Neri

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): V. Grieco

Study supervision: P. Musto, A. Neri

The authors thank Mrs. Gabriella Ciceri for the expert technical assistance.

This work was financially supported by grants from: the Associazione Italiana Ricerca sul Cancro (AIRC) IG10136 (to A. Neri); AIRC “Special Program Molecular Clinical Oncology- 5 per mille” n. 9980, 2010/15 (to P. Tassone, A. Neri, and F. Morabito); the Ministero Italiano dell'Istruzione, Università e Ricerca (MIUR; 2009PKMYA2 to A. Neri); the Fondazione Matarelli (Milan, Italy). M. Lionetti is supported by a fellowship from the Fondazione Italiana Ricerca sul Cancro (FIRC). P. Musto is supported by a Research fund from Celgene. K. Todoerti is supported in part by the Italian Health Minister, Finalized Research for Young Researchers, CUP Project E66110000230001.

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.