Multiclass cancer diagnosis using tumor gene expression signatures

Abstract

The optimal treatment of patients with cancer depends on establishing accurate diagnoses by using a complex combination of clinical and histopathological data. In some instances, this task is difficult or impossible because of atypical clinical presentation or histopathology. To determine whether the diagnosis of multiple common adult malignancies could be achieved purely by molecular classification, we subjected 218 tumor samples, spanning 14 common tumor types, and 90 normal tissue samples to oligonucleotide microarray gene expression analysis. The expression levels of 16,063 genes and expressed sequence tags were used to evaluate the accuracy of a multiclass classifier based on a support vector machine algorithm. Overall classification accuracy was 78%, far exceeding the accuracy of random classification (9%). Poorly differentiated cancers resulted in low-confidence predictions and could not be accurately classified according to their tissue of origin, indicating that they are molecularly distinct entities with dramatically different gene expression patterns compared with their well differentiated counterparts. Taken together, these results demonstrate the feasibility of accurate, multiclass molecular cancer classification and suggest a strategy for future clinical implementation of molecular cancer diagnostics.

Comparative Study of QSAR/QSPR Correlations Using Support Vector Machines, Radial Basis Function Neural Networks, and Multiple Linear Regression

“…6 As a new and powerful modeling tool, support vector machine (SVM) has gained much interest in pattern recognition and function approximation applications recently. In bioinformatics, SVMs have been successfully used to solve classification and correlation problems, such as cancer diagnosis, [7][8][9][10] identification of HIV protease cleavage sites, 11 protein class prediction, 12 etc. SVMs have also been applied in chemistry, for example, the prediction of retention index of protein, and other QSAR studies.…”

Gene expression profiling of primary cultures of ovarian epithelial cells identifies novel molecular classifiers of ovarian cancer

“…Expression-based classification schemes have been applied to distinguish ovarian cancers from other tumour types Ross et al, 2000;Ramaswamy et al, 2001), but rarely to distinguish normal samples from ovarian cancers. A recent study classified pooled normal epithelial samples from ovarian tumours based on the gene expression of CLDN3 and VEGF (Lu et al, 2004).…”

Approaches to working in high-dimensional data spaces: gene expression microarrays

“…Another approach is to apply support vector machines (SVMs), which attempt to avoid overfitting by finding a linear discriminant function (or generalised linear discriminant) that maximises the margin (the minimum distance of any sample point to the decision boundary) (Vapnik, 1998). The number of free parameters in SVMs is not a function of the dimensionality, but instead is upper-bounded by the number of samples, which for microarrays is much smaller (Ramaswamy et al, 2001). However, whether using linear or nonlinear kernels, SVMs are not immune to the curse of dimensionality.…”

“…In microarray data analysis, unsupervised clustering must be cautiously applied and may be unnecessary when samples come with appropriate and reliable supervising labels (Ramaswamy et al, 2001;Clarke et al, 2008). However, unsupervised clustering constitutes an important tool for discovering underlying cancer subtypes or gene modules (Frey and Dueck, 2007;Miller et al, 2008).…”

“…Gene expression microarrays provide a wealth of information on gene expression patterns and cancer pathways with potential for (1) cancer diagnosis, prognosis, and prediction of therapeutic responsiveness (Ramaswamy et al, 2001;Dupuy and Simon, 2007); (2) discovering new cancer subtypes (Golub et al, 1999;Lange et al, 2004); and (3) identifying cancer-associated (signalling) molecular markers and their complex interactions (Shedden et al, 2003;Ransohoff, 2004). However, achieving these biological/clinical objectives requires comprehensive analysis of microarray gene expression profiles that exist in high-dimensional data spaces, and relies critically on the functional capabilities and accuracy of the relevant analytical techniques (Allison et al, 2006).…”