miR-29 levels are downregulated in CXCR4^dimCD5^bright intraclonal subpopulation and associate with prognosis in CLL. (Ai) Representative example of sorting CXCR4^dimCD5^bright and CXCR4^brightCD5^dim CLL cell intraclonal subpopulations using flow cytometry. (Aii) Heatmap of differentially expressed miRNAs (fold-change > 1.5, adjusted P < .0005) in 7 pairs of CXCR4/CD5 sorted subpopulations (purity > 99%; for sample characteristics, see supplemental Table 1). Heatmap was generated from counts per million reads (rows centered to the median of the row). For details on individual miRNA expression see supplemental Table 3. (B) miR-29a/b/c levels analyzed using qRT-PCR in the CXCR4/CD5 sorted subpopulations from primary CLL samples (miR-29a/c, n = 10; miR-29b, n = 9; statistical differences were compared by Wilcoxon matched pairs test). (C) Overall survival is depicted using Kaplan-Meier curves in the CLL cohort (n = 107) divided by terciles of miR-29a/b/c expression (with log-rank test for comparison of low vs high miR-29s levels; miR-29a: HR, 3.1; 95% CI, 1.3-7.4; miR-29b: HR, 3.6; 95% CI, 1.3-10.20; miR-29c: HR, 3.4; 95% CI, 1.4-8.3). (D-F) Expression of miR-29a/b/c and its relationship with clinico-biological features. miR-29a, b, and c expressions were quantified in a cohort of 107 CLL patients (cohort characteristics in Table 1) and correlated to the clinico-biological characteristics such as (D) IGHV mutation status, (E) ZAP-70 expression, and (F) Rai stage (at sampling). The expression level of miR-29c was not available for 3 samples because of technical issues with its quantification. The statistical differences were tested using unpaired t test.