Gene expression signatures for colorectal cancer microsatellite status and HNPCC - PubMed (original) (raw)
. 2005 Jun 20;92(12):2240-8.
doi: 10.1038/sj.bjc.6602621.
J L Jensen, P Laiho, L Dyrskjøt, R Salovaara, D Arango, K Birkenkamp-Demtroder, F B Sørensen, L L Christensen, L Buhl, J-P Mecklin, H Järvinen, T Thykjaer, F P Wikman, F Bech-Knudsen, M Juhola, N N Nupponen, S Laurberg, C L Andersen, L A Aaltonen, T F Ørntoft
Affiliations
- PMID: 15956967
- PMCID: PMC2361815
- DOI: 10.1038/sj.bjc.6602621
Gene expression signatures for colorectal cancer microsatellite status and HNPCC
M Kruhøffer et al. Br J Cancer. 2005.
Abstract
The majority of microsatellite instable (MSI) colorectal cancers are sporadic, but a subset belongs to the syndrome hereditary non-polyposis colorectal cancer (HNPCC). Microsatellite instability is caused by dysfunction of the mismatch repair (MMR) system that leads to a mutator phenotype, and MSI is correlated to prognosis and response to chemotherapy. Gene expression signatures as predictive markers are being developed for many cancers, and the identification of a signature for MMR deficiency would be of interest both clinically and biologically. To address this issue, we profiled the gene expression of 101 stage II and III colorectal cancers (34 MSI, 67 microsatellite stable (MSS)) using high-density oligonucleotide microarrays. From these data, we constructed a nine-gene signature capable of separating the mismatch repair proficient and deficient tumours. Subsequently, we demonstrated the robustness of the signature by transferring it to a real-time RT-PCR platform. Using this platform, the signature was validated on an independent test set consisting of 47 tumours (10 MSI, 37 MSS), of which 45 were correctly classified. In a second step, we constructed a signature capable of separating MMR-deficient tumours into sporadic MSI and HNPCC cases, and validated this by a mathematical cross-validation approach. The demonstration that this two-step classification approach can identify MSI as well as HNPCC cases merits further gene expression studies to identify prognostic signatures.
Figures
Figure 1
Unsupervised hierarchical clustering of 101 colorectal tumours. The phylogenetic tree shows the spontaneous clustering of 101 tumour and 17 normal biopsies into four clusters mainly consisting of normal biopsies, MSI or MSS tumours, respectively. In the left column, the microsatellite status is indicated as MSS (S) or MSI (I). Hereditary nonpolyposis colorectal cancer tumours are indicated by (H) and normal biopsies by (N). In the second column the tumour location is indicated as right-sided (R) or left-sided (L) colon, or rectum (Rt).
Figure 2
Performance of the MI classifier in the training set. The bars indicate the relative distance of every single tumour to the centres of the microsatellite unstable and microsatellite stable groups. The distances are log2 and defined through the cross-validation steps. A value of +2 indicates that the distance of a tumour to the microsatellite unstable group is four times the distance to the microsatellite stable group. The upper 34 tumours (open bars) are MSI tumours and the solid bars are MSS tumours. (*) Indicate samples that are always misclassified, and (+) indicate samples that are almost equally close to both groups.
Figure 3
Classification of MI status based on real-time PCR. Panel A shows a cluster analysis of a subset (18 samples) of the 101 tumour samples using the nine signature genes, based on either the microarray data or the real-time PCR data. Blue colours indicate relative low expression and yellow colour high expression. Panel B shows the classification result of 47 new independent samples based on PCR data using seven of the nine genes. Relative distances are explained in the legend to Figure 2. The two misclassified tumours are indicated with an asterisk. For PCR primers and hybridisation probes, see supplementary data 2.
Figure 4
Kaplan–Meier estimates of crude survival among patient with Stage II and Stage III colorectal cancer, according to microsatellite status of the tumour determined by a nine-gene expression signature. Open triangles indicate censored samples. The patients left at risk are denoted in brackets. The _P_-values were calculated with use of the log-rank test. (A) Patients with Stage II colon cancer (not adjuvant chemotherapy). (B) Patients with Stage III colon cancer (adjuvant chemotherapy).
Figure 5
Classification of MSI tumours as hereditary or sporadic. Panel A shows the number of classification errors in cross-validation as a function of the number of genes used. The minimum number of errors found was one using two genes, and adding more genes increased the number of errors. Panel B shows log2 of the ratio of the distance between a tumour to the centres of the sporadic microsatellite unstable group and the hereditary microsatellite unstable group. Panel C shows microarray signal values for MLH1 and PIWIL1 genes for all tumours. Asterisk indicates the misclassified tumour.
References
- Andersen CL, Jensen JL, Orntoft TF (2004) Normalization of real-time quantitative reverse transcription-PCR data: a model-based variance estimation approach to identify genes suited for normalization, applied to bladder and colon cancer data sets. Cancer Res 64: 5245–5250 - PubMed
- Birkenkamp-Demtroder K, Christensen LL, Olesen SH, Frederiksen CM, Laiho P, Aaltonen LA, Laurberg S, Sorensen FB, Hagemann R, Orntoft TF (2002) Gene expression in colorectal cancer. Cancer Res 62: 4352–4363 - PubMed
- Bolstad BM, Irizarry RA, Astrand M, Speed TP (2003) A comparison of normalization methods for high density oligonucleotide array data based on variance and bias. Bioinformatics 19: 185–193 - PubMed
Publication types
MeSH terms
LinkOut - more resources
Full Text Sources
Other Literature Sources