Supplementary Material 1: Survival Analysis for Mirnas

Supplementary Material 1: Survival analysis for miRNAs

SignS2, a web-based tool for gene selection for survival data( used to investigate miRNAs related to outcome in SMZL using the FCMS method, a stepwise model selection for filtering and clustering genes. For significant miRNAs, survival distributions were estimated with SPSS (v.17.0) for a univariate (gene-by-gene) Cox analysis and Kaplan-Meier curves. In all comparisons, miRNAs with values of p<0.05 were considered significant.

To search for miRNAs related to outcome in SMZL, 28 of 31 cases (cases 2, 13 and 31 had no overall survival data) were subjected to the FCMS method using SignS2. The significant miRNAs (p<0.05) were used in a univariate Cox regression, identifying34 miRNAs that correlated positively or negatively with OS. Survival probabilities were estimated using the Kaplan-Meier method obtaining a set of eight miRNAs (Supp Table 1; Supp Figure 1). Four of them (miR-382, miR-768-3p,miR-28 and miR-141) were selected for qRT-PCR validation in an independent FFPE series.

The four miRNAs tested by RT-PCR were not significant in the Cox and Kaplan-Meier analysis possibly due to the clinical data of the FFPE independent series being relatively incomplete, as only 57 of 77 cases of the FFPE series had OS data.

Supp Table 1. Survival analysis. MiRNAs with significant p-valuesin SignS and Kaplan-Meier analyses. 28 of 31 SMZL cases were subjected to a univariate (gene-by-gene) Cox analysis after FCMS using SignS ( Thirty-four miRNAs were correlated positively or negatively with overall survival. The Cox regression analysis showed significant values for all 34 miRNAs(p<0.05; data not shown). Survival distributions were estimated using the Kaplan-Meier method obtaining a set of 8 miRNAs with p<0.05. NS, not significant.

Supplementary Table 2. SMZL gene expression signature. We used a t-test ( to obtain the SMZL gene signature when comparing its expression with that of the RLH. The comparison considered genes with FDR<0.01 to be significant.Genes with more than two-fold upregulation or downregulationwere also selected. The fold change (FC) corresponds to a log2 difference between the SMZL and RLH averages. FDR values are those from the t-test. To identify associations between differentially expressed miRNAs (FDR<0.05) and gene expression signatures (FDR<0.05) we followed the previously described procedure1.

Supplementary Table 3. Gene set enrichment analysis comparing GEP of SMZL with RLH. We carried out at-test in GSEA, significanceconcluded for values of FDR<0.25. p-values and FDRs(false discovery rates)correspond to t-tests.

Supplementary Table 4. MiRNAs with differential expression between SMZL B cells and non-tumoral B cells.We carried out at-test with B cell data obtained from isolated CD19+ B cells from peripheral blood of SMZL patients and healthy donors. FDR is the false discovery rate from t-tests; FC is the fold change (log2 scale) difference between the corresponding averages of classes.

Supplementary Table 5. MiRNAs with differential expression between SMZL and NMZL. FDRis the false discovery rate from a t-test comparing SMZL and NMZL. Fold change (FC) is the log2 difference between the SMZL and RLH averages and between the SMZL and NMZL means.

Supplementary Figure 1. Kaplan-Meier survival curves comparing SMZL patients with high- and low-level expression of significant miRNAs. Patients with high-level expression of miR-28, miR-768-5p and miR-141 had significantly shorter disease-related overall survival while miR-382 had significantly longer overall survival. Green and blue colors represent high- and low-level expression, respectively. p-values are those of the Kaplan-Meier analysis; * p<0.05, ** p<0.005.

REFERENCES

1.Di Lisio L, Gómez-López G, Sánchez-Beato M, et al. Mantle cell lymphoma: transcriptional regulation by microRNAs. Leukemia. 2010;24(7):1335-1342.

2.Diaz-Uriarte R. SignS: a parallelized, open-source, freely available, web-based tool for gene selection and molecular signatures for survival and censored data. BMC Bioinformatics. 2008;9:30.

1