Mantle cell lymphoma (MCL) is an aggressive B-cell neoplasm with a median survival of the patients of 3–5 years.1 This aggressive behavior has been related to its genetic and molecular pathogenesis that integrates the deregulation of cell proliferation due to the t(11;14)(q13;q32) and cyclin D1 overexpression, and the accumulation of a high number of chromosomal aberrations mainly targeting genes related to DNA damage response and the cell survival pathways.2 Thus, it is usually recommended that patients with MCL should be treated with chemotherapeutic regimens immediately following their diagnosis. However, this attitude is being reconsidered, due in part to the increasing recognition of subsets of patients that do not need therapy for a long period of time3, 4, 5, 6, 7 and the observation that deferral of treatment does not seem to impair their global outcome.3, 5 The identification of asymptomatic patients at diagnosis that may benefit from initial watch and wait approach is challenging because some studies have shown that intensive treatment may improve the survival of patients with MCL.8 Therefore, it is of paramount importance to develop clinical and biological criteria that may assist in the selection of the optimal individual management for patients with MCL.
In a recent study, we compared a subgroup of MCL patients with a very indolent clinical behavior that did not received chemotherapy for more than 2 years with a subset of patients that required treatment at diagnosis.4 The indolent MCL had a profile of clinical and biological features that differed from the conventional MCL (cMCL), suggesting that at least a subset of cases might correspond to a particular biological subtype of the disease. One of the major findings was the differential expression of a small signature of 13 genes, including SOX11, that was highly expressed in cMCL but negative or very low in indolent tumors.4
The aim of the present study was to determine whether the recognition of this gene expression signature in an independent and larger series of patients could identify subgroups of MCL with different biological and clinical features and may assist in devising management strategies more accordingly to the biology of the disease.
We designed and validated a simple quantitative PCR assay using three genes (SOX11, HDGFRP3 and DBN1) of the previously described gene signature differentially expressed between indolent and cMCL4 (Supplementary Table S1 and Supplementary Figures S1 and S2).
Because indolent MCL frequently presented with a leukemic non-nodal disease, we studied this simplified three-gene signature and CCND1 expression in blood samples from a series of 68 MCL and 42 leukemic non-MCL neoplasms. SOX11, HDGFRP3, DBN1 and CCND1 were highly expressed in MCL but were negative in the 42 non-MCL samples (Supplementary Table S1 and Supplementary Figure S2). CCND1 was overexpressed in all MCL but the expression of the other three genes was variable, with a subset of MCL showing low or negative expression. To determine the significance of the variable expression of the three-gene signature, we performed an unsupervised hierarchical clustering analysis of the tumors according to the expression of this three-gene signature. This analysis identified two clusters of MCL with relatively concordant high (n=46 cases) or low (n=22 cases) expression of the three genes (Supplementary Table S1 and Supplementary Figures S3 and S4).
These two subgroups of MCL had significant differences in clinical and biological features (Table 1). Palpable lymphadenopathy (>1 cm) was detected more frequently at diagnosis in patients carrying tumors with high than low expression signature (77% vs 12%; P<0.001). Patients with high signature were more frequently treated with chemotherapy during the evolution of the disease (93%) than patients with low expression (58%) (P=0.001). Anthracycline-containing regimens were administered in 78% and 60% of cases treated in the subgroup of MCL with high and low signature, respectively. The clinical outcome was significantly different in these two subgroups of patients with a 5-year overall survival (OS) rate of 32% (95% confidence interval (CI), 11–53) for the group with high expression signature and 75% (95% CI, 48–100) for patients with low expression signature (P=0.006)(Table 1 and Figure 1a).
The IGHV genes were mutated (97% identity) in 77% of the tumors with low signature but only in 17% with high signature (P<0.001) (Table 1). The genomic profile was analyzed in 36 tumors with high and 17 with low signature using SNP6.0 arrays (Supplementary Figure S5). Both subgroups had the characteristic profile of chromosomal imbalances of MCL9, 10 but tumors with high signature had significantly more gains of 3q and losses of 9p and 11q (P=0.016, 0.007 and 0.029, respectively). Genomic complexity (copy number alterations 5 per case) was significantly more frequent in MCL with high expression signature (29/36, 80%) than in tumors with low expression signature (6/17, 35%) (P=0.001). Interestingly, the six cases with low signature carrying complex genomes had 17p loss, and TP53 mutations were detected in the five cases studied. TP53 mutations were also detected in 6 of the 10 MCLs with high signature that had 17p alterations (Supplementary Table S2).
The variables predicting for poor OS were the high expression of the three-gene signature, nodal presentation and unmutated IGHV (P=0.01, 0.03 and 0.05, respectively)(Supplementary Table S3). The combined three-gene signature provided a better prediction of OS than any of the three genes individually (Supplementary Figure S6). To determine whether the three-gene signature added information to the nodal presentation, we compared the clinical impact of these two variables (Figure 1b). Patients with non-nodal presentation and low signature had a better outcome (5-year OS 86%, 95% CI, 67–100) than patients with non-nodal presentation and high signature (5-yr OS 42%, 95% CI, 1–83). Patients with nodal presentation had a poor outcome regardless of the expression levels of the signature. Although only two patients had lymphadenopathy and low signature, these two patients were treated at diagnosis. The association of non-nodal presentation with a better outcome was in agreement with previous clinical observations.7, 11 However, our study also indicates that patients with leukemic presentation without lymphadenopathy can be further stratified based on the three-gene signature. Non-nodal MCL with high expression signature had a more aggressive clinical behavior with significantly shorter OS than patients with low signature. These findings are particularly relevant because the presence of high white blood cell counts has been traditionally associated with a poor prognosis in MCL.12 However, our study supports the idea that MCL patients with a non-nodal disease, high leukocyte count and low signature may correspond to a different subtype of the disease with different clinical behavior. The use of the three-gene signature may assist to distinguish these two groups of patients. Further studies in patients with nodal disease are needed to evaluate the impact of the signature on this group of MCL.
To have a broader overview and increase the statistical power in the evaluation of the prognostic parameters, we combined this series with the 17 patients from our previous study.4 The variables significantly predicting for poor OS were also the high expression of the three-gene signature, nodal presentation and unmutated IGHV (P=0.001, 0.013 and 0.006, respectively). Interestingly, patients with high genomic complexity and 17p alterations had also a poor impact on survival calculated from the time of the genetic assessment (P<0.001). Of note, the detection of 17p alteration had a significant adverse prognosis in both subgroups of MCL with high (P=0.006) and low expression signature (P=0.045) (Figures 1c and d).
In spite of the global good prognosis of our patients with low signature, some of them died rapidly after diagnosis or developed progressive disease. Interestingly, the detection of a 17p deletion had an adverse impact on outcome in these patients, suggesting that similarly to cMCL and other lymphoid neoplasms, the acquisition of 17p/TP53 alterations may also be a mechanism of tumor progression in MCL with low expression signature impairing the outcome of the patients.2 This finding suggests that the evaluation of 17p/TP53 alterations may be also important to refine the prognosis of patients with MCL. Intriguingly, although the presence of 17p alterations had a poor prognostic impact on both subgroups of MCL, patients with low signature had a longer survival than patients with high signature carrying 17p alterations (5-yr OS, 48% vs 17%). This observation parallels the finding of a subset of CLL patients with Binet stage A and mutated IGHV that had a stable disease in spite of the presence of 17p/TP53 alterations.13
All the findings of the present study indicate that MCL presenting with a non-nodal disease and low expression of the three-gene signature may correspond to a particular clinical and biological subtype of the disease with a more indolent clinical behavior. The selection of a term to name this subtype of tumors is not easy. We suggest the term ‘non-nodal type of MCL’ because it reflects their major, although not entirely specific, clinical feature and, on the other hand, it may give credit to the initial clinical observations of this subgroup of tumors.7, 14 The presence of 17p/TP53 alterations in these patients impairs their outcome, suggesting that they may influence the progression of the disease. The evaluation of the three-gene signature in leukemic samples together with the study of 17p/TP53 may help to identify this particular subtype of MCL and to determine a parameter of higher risk in the evolution of the patients, respectively.
It may be arguable whether non-nodal type of MCL corresponds to a totally different disease. However, the similar global genomic profile observed in this study and the similar global gene expression profile of both types of tumors identified in our previous study4 would favor the idea that they are subtypes of the same disease. In addition, a recent study of ‘in situ MCL’ has observed that ‘in situ’ lesions with SOX11-positive or -negative MCL-like cells have a similar distribution in the mantle-zone area of reactive follicles, indicating that early lesions in these two types of tumors share the same topographic microenvironment.15
In conclusion, our findings support the idea that non-nodal type of MCL with a low signature of these three genes may correspond to a specific subtype of the disease. The study of the three-gene signature and 17p/TP53 status may assist in devising management strategies for patients with MCL more adjusted to the biology of the disease.
This study has been supported by ‘Instituto de Salud Carlos III, Fondo Investigaciones Sanitarias’ (FIS06/0150 and PI08/0077) (SB), the Spanish Ministry of Science and Innovation SAF 2008-03630 (EC), ‘Red Temática de Investigación Cooperativa de Cáncer’ (RTICC) RD06/0020/0039 (EC), RD06/0020/0051 (ALG) and RD06/0020/0014 (DC), Instituto de Salud Carlos III ‘Beca Predoctoral de Formación en Investigación en Salud’ (FI08/00437) (CR), Ministry of Science and Innovation ‘Formación de Personal Investigador’ (BES-2007-16330) (AN), European Regional Development Fund (ERDF) ‘Una manera de fer Europa’, European Mantle Cell Lymphoma Network, and Generalitat de Catalunya (2009-SGR-992) (EC).
About this article
Supplementary Information accompanies the paper on the Leukemia website (http://www.nature.com/leu)
Der Pathologe (2019)