|
|
Blood, Vol. 113, Issue 18, 4331-4340, April 30, 2009
Inhibition of aurora kinases for tailored risk-adapted treatment of multiple myeloma
Blood Hose et al.
113: 4331
Supplemental materials for: Hose et al
Calculation of the gene expression based proliferation index
The gene expression based proliferation index is calculated as follows. In brief, genes are selected based on genes overexpressed in proliferating cells (malignant: human myeloma cell lines HMCL, benign: polyclonal plasmablastic cells PPC) compared to non-proliferating cells (normal bone marrow plasma cells BMPC and memory B cells MBC). Here, four comparisons between the groups are made (i) HMCL vs. MBC, ii) HMCL vs. BMPC, iii) PPC vs. BMPC and iv) PPC vs. MBC) by a onesided t-test, with the alternative hypothesis being that expression values of HMCL and PPC values are greater compared to BMPC and MBC in each comparison. P-values are permutation-adjusted regarding a family wise error rate with an α level of 0.025. To adjust for comparing each group twice, the α level is halved to 0.0125. Only genes being statistically significant in each of the 4 comparisons are retained for the index. To select biologically (in terms of proliferation) relevant genes, only genes matching with the gene-ontology term “cell proliferation” or “cell cycle” were retained. Thus, 50 genes (57 probesets) represent the final index. For genes with more than one probeset per gene, the probeset with the highest variance within the TG is selected. The index is calculated as follows: as proliferation genes determined as stated above are overexpressed by definition, the individual gene expression based proliferation index for each sample is calculated as the sum of expression values of each of the 50 genes in an individual sample. For genes not expressed as judged by PANP, the expression level of the respective gene is defined as 0.
Files in this Data Supplement:
- Table S1. Clinical patient data for age, serum-β2–microglobulin, and plasma cell infiltration (PDF, 20.4 KB) -
In the training group (TG), the validation group (VG) and the Arkansas data. Median value and range are given.
- Table S2. Overview over patient populations, subpopulations and samples used (PDF, 19.5 KB)
- Table S3. Differential gene expression (PDF, 41.6 KB) -
(Δ) as determined by EB statistics using Benjamini-Hochberg correction for multiple testing for the comparison of the respective populations with normal bone marrow plasma cells (BMPC; upper row), memory B cells (MBC; middle row), as well as between early- (MGUS+MMI) vs. late- (MMII+MMIII) stage myeloma (lower row) for (A) training group and (B) validation group. Significant values for the respective comparisons are depicted in red.
- Figure S1. The centrosome index calculated as published by Chng et al.49 (JPG, 116 KB)
-
Shown are (A) event-free (EFS) and (B) overall survival (OAS). Low (black curve) vs. high (red curve) centrosome index in CD138-purified myeloma cells. The centrosome index is significantly predictive for EFS and OAS in the Arkansas group, on which it has been derived, but not our series.
|
|
