Page 149 - 2018_12-Haematologica-web

Trisomy 12 CLL
justed P=0.00091). Forty-three (5.9%) of the DE probes represent genes on 11q22.3 (e.g. ATM, NPAT, DDX10, CUL5, ACAT1);17 all are expressed at higher levels in +12 and at lower levels in del(11q) cases. In addition, 208 out of 736 (28%) of the DE probes are on chromosome 12; 206 out of 208 are over-expressed in +12 compared to del(11q). Comparing +12 and diploid cases, we identified 1229 DE probes representing 964 unique genes, 49 ncRNAs, and 18 ESTs (FDR=5%, unadjusted P=0.00164); 413 out of 1229 (34%) DE probes are on chromosome 12 (χ2=92.5, P<2e-16), and all 413 are over-expressed in +12 compared to diploid.
Consistent differences between +12 and other cytogenetic subtypes
Because the pair-wise DE comparisons included genes whose expression characterizes other cytogenetic sub- types (e.g. BCL2 and other miR-15-a/16-1 targets when comparing +12 to del(13q) cases, or ATM and other genes in the commonly deleted region when comparing +12 to del(11q) cases), we used t-tests to identify DE genes when comparing +12 cases to the union of all other cases in the study. We found 1226 DE probes representing 953 unique genes, 40 ncRNAs, and 17 ESTs (FDR=1%, unadjusted P=0.000347); 419 out of 1226 (34%) were on chromosome 12, and 418 out of 419 were over-expressed in +12 cases. We also observed that 1194 out of 1226 (97%) of these probes were DE in the direct comparison with del(13q); 892 out of 1226 (73%) were DE in the direct comparison to diploid cases, and 526 out of 1226 (43%) were DE in the direct comparison to del(11q). The full list of DE probes is presented in Online Supplementary Table S1.
Among cases with sole +12, microarray profiling data and NOTCH1 mutation status were available for 15 patients in the discovery set: n=9 wild-type (60%), n=6 mutated (40%). To identify DE genes between these sub- sets, we performed the following analysis. We removed low-expressing probes, and retained a probe only if its expression was >4 on the log2 scale in at least 3 out of 15 samples; 20,776 of the 47,231 probes satisfied this criteri- on. We performed probe-by-probe t-tests to compare expression between the NOTCH1 mutated versus wild- type samples. We found 389 DE genes with P<0.01 (FDR=45%; Online Supplementary Table S2).
Pathway analysis
We performed Ingenuity Pathway Analysis (IPA) to identify differentially regulated canonical pathways that distinguish +12 from other cytogenetic subtypes individu- ally. For del(13q), del(11q), and diploid subtypes we per- formed analyses using the 1181, the 736, and the 1229 DE probes described above. For each comparison, ten path- ways were identified as either activated or down-regulat- ed with the most statistical significance, based on the Ingenuity z-scores (Table 3). Complete data are listed in Online Supplementary Table S3.
Validation of potential targets
To validate potential mRNA targets identified by whole- genome transcriptional profiling performed on the discov- ery set, we assessed expression of a subset of these genes by MF-QRT-PCR assay on an independent validation set of 50 patient samples. We used the QRT-PCR assay because it is more reproducible and has a wider dynamic range than microarray profiling.13 Of 135 genes assayed,
Figure 1. Kaplan Meier plots stratified by cytogenetic subtype. (A) Time to treatment, and (B) progression-free survival.
64 (47%) were fully validated for all comparisons between +12 and other pure cytogenetic subsets, and 91 (67%) were validated for at least half of the comparisons. A sub- set of 31 genes assayed using the MF-QRT-PCR assay are included in the network diagram in Figure 3; of these, 19 (61%) were fully validated and all 31 (100%) were validat- ed in at last half of the comparisons of +12 with other cytogenetic groups. Complete data are listed in Online Supplementary Table S4.
A +12 specific network
Using the MF-QRT-PCR data, we constructed a gene network whose expression in +12 cases differed from other subtypes. We selected all DE genes with FDR=5% (unadjusted P=0.0016) and fold change (FC) ≥2 in compar- ison to both del(13q) and diploid cases. Intersecting these lists yielded 109 probes that represent 92 distinct genes. Although 17 probes did not satisfy the criteria when we compared +12 cases directly to del(11q), we chose to retain them because all 109 probes satisfied the selection
A
B
haematologica | 2018; 103(12)
2073