|
|
Blood, Vol. 113, Issue 21, 5237-5245, May 21, 2009
microRNA-29c and microRNA-223 down-regulation has in vivo significance in chronic lymphocytic leukemia and improves disease risk stratification
Blood Stamatopoulos et al.
113: 5237
Supplemental materials for: Stamatopoulos et al
Files in this Data Supplement:
- Figure S1. CD38 assessment by flow cytometry (FC) (JPG, 29.3 KB)
-
We used 25 ng of cDNA (produced by a standard reverse transcription) in a qPCR reaction with SYBR® Green PCR Master Mix (Applied Biosystems) and 0.32 mol/L of gene-specific forward and reverse primers (Invitrogen). We standardized all results using cyclophilin A (PPI) gene expression. The primer sequences used to amplify ZAP70, PPI and LPL are listed in the table below. Standard real-time PCR was performed on an ABI Prism 7900 HT (Applied Biosystems). A calibrator sample (cDNA from the Namalwa cell line, a human B-lymphoid leukemia cell line that expresses ZAP70 at a low level; ATCC) was included as a control in each experiment. In all cases, we created dissociation curves to confirm PCR specificity. Data were analyzed using the comparative ΔΔCt method. We evaluated the cell surface expression of CD38 by FC in a CD19+ gate with a panel of fluorochrome-labeled monoclonal antibodies (phycoerythrin-conjugated CD38, Cyanine-5 CD19, Immunotech). CD38 expression was deemed positive if 7% of the cells stained positive in a standard 3-color FC analysis. This cut-off was calculated using ROC curve analysis maximizing the concordance with IgVH mutational status.1 sCD23 and β2-microglobulin serum levels were determined using commercial immunoassay kits. Standards were used to fully quantify the sCD23 or β2-microglobulin level, and the provided controls were loaded in each experiment to monitor the assay performance and the inter-assay variability. Lymphocyte doubling time was determined as described by Montserrat et al.2 and is defined as the time needed to double the peripheral lymphocyte count. For conventional cytogenetic analysis, culture conditions, harvesting, slide preparation, and G-banding were carried out as described previously.3 Additional cytogenetic abnormalities were investigated with the Chromoprobe Multiprobe® CLL System. Fresh or frozen CLL cells were washed twice with PBS and incubated in KCl (0.075M − pH7) for 10min. Cells were then fixed with Carnoy’s Fixative (3:1 methanol: glacial acetic acid). Hybridization was performed according to the manufacturer’s recommendation. The cells (100 to 200) were counted to generate representative results. CLL FISH panel allows for the detection of trisomy of 12, deletions in 13q14 ATM (11q22.3), TP53 (17p13.1) and MYB (6q23.3) and also translocation involving IGH Fission (14q32), IGH/CCND1 (14q32/11q13.3) and IGH/BCL2 (14q32/18q21.3). IgVH gene mutational analysis was performed as previously described,4 and the sequences were aligned with those in the international ImMunoGeneTics information system database (http://imgt.cines.fr). Sequences with ≤2% deviation from any germ line IgVH sequence were considered unmutated.5
- Figure S2. qPCR-score 10-fold cross validation (JPG, 110 KB)
-
Cross-validation is the statistical practice of partitioning a sample of data into subsets such that the analysis is initially performed on a single subset, while the other subsets are retained for subsequent use in confirming and validating the initial analysis. The initial subset of data is called the training set; the other subsets are called the validation sets. In a 10-fold cross-validation, the original sample is partitioned into 10 subsamples (10 subsamples of 11 patients in our study). Of the 10 subsamples, a single subsample is retained as the validation data for testing the model, and the remaining 9 subsamples are used as training data. The cross-validation process is then repeated 10 times (see figure below), with each of the 10 subsamples used exactly once as the validation data. The 10 results from the folds are then combined to produce a single estimation and correlated to survival data (TFS and OS in our study). The advantage of this method over repeated random sub-sampling is that all observations are used for both training and validation, and each observation is used for validation exactly once. 10-fold cross-validation is commonly used to estimate the prediction accuracy of a classification model.6–11 The first table states the variation of the cut-off in the ten subsamples of the 10-fold cross-validation. ZAP70 and LPL cut-offs remain quite stable, but the miR-29c and miR-223 cut-offs are more influenced by the loss of 10% of the population. The second table is a confusion table. Finally, the score computed by the cross-validation model remains significant to predict TFS and OS.
- Figure S3. TFS and OS distribution according to classical prognostic factors (JPG, 170 KB)
-
(A and J), indicated are TFS and OS curves for Binet stage A vs B–C (n=110); (B and K) IgVH mutational status (n=110); (C and L), ZAP70 by qPCR (n=110); (D and M), LPL by qPCR (n=110); (E and N), CD38 (n=104); (F and O), β2-M (n=78); (G and P), sCD23 (n=91); (H and Q), LDT (n=93); and (I and R), cytogenetic abnormalities detected by classical karyotype analysis or by FISH (normal/del(13q)/other) vs (del(17p)/(11q)/(6q)/+12/complex) (n=81). ROC curves were used to determine the ZAP70, LPL, CD38, miR-29c, miR-223, sCD23, and β2-M expression cut-off values that best distinguished mutated and unmutated cases. IgHV mutational status is based on a 98% cut-off value.
- Figure S4. ROC curve analysis for miR-29c and miR-223 vs IgVH mutational status (JPG, 32.4 KB)
-
When evaluating a diagnostic test, it is often difficult to determine the threshold laboratory value that separates a clinical diagnosis of “normal” from one of “abnormal.” A receiver operating characteristic (ROC), or simply ROC curve, is a graphical plot of the sensitivity vs. (1 − specificity) for a binary classifier system as its discrimination threshold is varied. (A) ROC curve analysis for miR-29c. (B) ROC curve analysis for miR-223; chosen cut-off are indicated by the black arrow. AUC, Area Under the Curve.
- Figure S5. Univariate (A) and multivariate (B) Cox analysis for TFS and/or OS prediction (JPG, 82.2 KB)
-
- Figure S6. Cross-comparison between each qPCR-score for TFS prediction among all Binet stages (A), Binet stage A (B), for OS prediction among all Binet stages (C), and Binet stage A (D) (JPG, 79.5 KB)
-
For each comparison, hazard ratio calculated by univariate Cox regression and confidence level at 95% are provided.
- Figure S7. Median TFS and OS according to prognostic factors and qPCR-score (JPG, 121 KB)
-
- Figure S8. Multivariate Cox analysis including classical prognostic factors and our qPCR-score (JPG, 51 KB)
-
- Figure S9. Serial measurements of miR-29c and miR-223 at different times after diagnosis (JPG, 56.9 KB)
-
MicroRNAs were measured on RNA extracted from CD19 purified cells of 7 patients for who we had RNA at diagnosis and 3 different samples collected at different times. The level of these 2 microRNAs remains quite stable along time for the majority of the patients. Even if microRNA expression is not totally stable, the “positive” or “negative” microRNA status does not change, and our qPCR score was thus not affected (patients 2 to 7). We can conclude that miR status determined at diagnosis remains unchanged for patients who present a non evolutive disease. However, we also showed that miR expression is statistically linked to tumor burden, particularly to lymphocyte doubling time. These results were confirmed in our serial measurement of miR: indeed, for patient 1, we observed a switch of microRNA status. There was a clear decrease of miR-expression along time from diagnosis till the end of 2008. This decrease is associated with a drastic increase of the lymphocytosis. We thus conclude that our score could be influenced during disease course, particularly after lymphocytosis increase. Nevertheless, the role of a prognostic factor is to prognosticate patient evolution before disease acceleration. Our score remains thus an excellent prognostic factor at diagnosis.
REFERENCES
1. Stamatopoulos B, Meuleman N, Haibe-Kains B, et al. Quantification of ZAP70 mRNA in B cells by real-time PCR is a powerful prognostic factor in chronic lymphocytic leukemia. Clin Chem. 2007;53:1757–1766.
2. Montserrat E, Sanchez-Bisono J, Vinolas N, Rozman C. Lymphocyte doubling time in chronic lymphocytic leukaemia: analysis of its prognostic significance. Br J Haematol. 1986;62:567–575.
3. Heimann P, Devalck C, Debusscher C, Sariban E, Vamos E. Alveolar soft-part sarcoma: further evidence by FISH for the involvement of chromosome band 17q25. Genes Chromosomes Cancer. 1998;23:194–197.
4. Campbell MJ, Zelenetz AD, Levy S, Levy R. Use of family specific leader region primers for PCR amplification of the human heavy chain variable region gene repertoire. Mol Immunol. 1992;29:193–203.
5. Hamblin TJ, Davis Z, Gardiner A, Oscier DG, Stevenson FK. Unmutated Ig V(H) genes are associated with a more aggressive form of chronic lymphocytic leukemia. Blood. 1999;94:1848–1854.
6. Cassidy A, Myles JP, van Tongeren M, et al. The LLP risk model: an individual risk prediction model for lung cancer. Br J Cancer. 2008;98:270–276.
7. Kim TJ, Choi JJ, Kim WY, et al. Gene expression profiling for the prediction of lymph node metastasis in patients with cervical cancer. Cancer Sci. 2008;99:31–38.
8. Keller A, Ludwig N, Comtesse N, Hildebrandt A, Meese E, Lenhof HP. A minimally invasive multiple marker approach allows highly efficient detection of meningioma tumors. BMC Bioinformatics. 2006;7:539.
9. Heuser M, Wingen LU, Steinemann D, et al. Gene-expression profiles and their association with drug resistance in adult acute myeloid leukemia. Haematologica. 2005;90:1484–1492.
10. Kattan MW, Giri D, Panageas KS, et al. A tool for predicting breast carcinoma mortality in women who do not receive adjuvant therapy. Cancer. 2004;101:2509–2515.
11. Pass HI, Liu Z, Wali A, et al. Gene expression profiles predict survival and progression of pleural mesothelioma. Clin Cancer Res. 2004;10:849–859.
|
|
