Normand Allaire

Convergent Random Forest predictor: methodology for predicting drug response from genome-scale data applied to anti-TNF response.

Biomarker development for prediction of patient response to therapy is one of the goals of molecular profiling of human tissues. Due to the large number of transcripts, relatively limited number of samples, and high variability of data, identification of predictive biomarkers is a challenge for data analysis. Furthermore, many genes may be responsible for drug response differences, but often only a few are sufficient for accurate prediction. Here we present an analysis approach, the Convergent Random Forest (CRF) method, for the identification of highly predictive biomarkers. The aim is to select from genome-wide expression data a small number of non-redundant biomarkers that could be developed into a simple and robust diagnostic tool. Our method combines the Random Forest classifier and gene expression clustering to rank and select a small number of predictive genes. We evaluated the CRF approach by analyzing four different data sets. The first set contains transcript profiles of whole blood from rheumatoid arthritis patients, collected before anti-TNF treatment, and their subsequent response to the therapy. In this set, CRF identified 8 transcripts predicting response to therapy with 89% accuracy. We also applied the CRF to the analysis of three previously published expression data sets. For all sets, we have compared the CRF and recursive support vector machines (RSVM) approaches to feature selection and classification. In all cases the CRF selects much smaller number of features, five to eight genes, while achieving similar or better performance on both training and independent testing sets of data. For both methods performance estimates using cross-validation is similar to performance on independent samples. The method has been implemented in R and is available from the authors upon request: Jadwiga.Bienkowska@biogenidec.com.

A6.44 Only BAFF mRNA, not BAFF protein level in blood, is associated with SLE activity over one year

Background B-cell–activating factor (BAFF; also known as B lymphocyte stimulator or BLyS) is a prominent factor in the selection and survival of B cells. BAFF has been demonstrated to be elevated in the blood of systemic lupus erythematosus (SLE) patients and is implicated in the pathogenesis of the disease. We have shown that BAFF gene expression level (mRNA) in whole blood associates with same day disease activity and predicts future activity in SLE patients. The concentration of the BAFF protein in serum has also been used as a marker of disease activity. In this study, we investigated the utility of BAFF mRNA versus protein level as a predictor of future global disease activity in SLE patients. Methods 292 patients (59% Caucasian, 34% African-American, 92% female, mean age 46 ± 12 years) were enrolled in a prospective observational study. At baseline, BAFF gene expression level was measured in peripheral blood RNA (PAXgene) using quantitative PCR. Serum BAFF (protein) levels were measured using the Rules Based Medicine platform. The number of visits per patient over the following year ranged from 1–9. Six patients had 1 visit, 46 patients had 2–3 visits, 159 patients had 4 visits, and 81 patients had more than 4 visits. P-values were calculated using generalised estimating equations as implemented in SAS 9.2. P-values were then adjusted for ethnicity. Results By two separate measures, PGA (physician global assessment) and SLEDAI, higher levels of measured BAFF mRNA were associated with higher levels of disease activity. The association with SLEDAI was stronger. Multiple measures of disease activity including proteinuria, anti-dsDNA antibodies and hypocomplementemia were also associated with higher levels of BAFF mRNA. This same association was NOT seen with the serum BAFF protein. Higher levels of BAFF protein were only associated with elevated anti-dsDNA antibodies, not global measures of disease activity. Conclusion BAFF mRNA at the baseline visit was strongly associated with global disease activity, urine protein/creatinine ≥0.5, serologies, and ESR over the next year. In contrast, BAFF protein level in the blood at baseline only correlated with anti-dsDNA over the next year. This study supports the use of BAFF mRNA level in peripheral blood rather than protein as a predictive biomarker of disease activity in SLE patients.

Experimental comparison and cross-validation of Affymetrix HT plate and cartridge array gene expression platforms

The successful use of gene expression microarrays in basic research studies has spawned interest in the use of this technology for clinical trial and population-based studies, but cost, complexity of sample processing and tracking, and limitations of sample throughput have restricted their use for these very large-scale investigations. The Affymetrix GeneChip Plate Array System addresses these concerns and could facilitate larger studies if the data prove to be comparable to industry-standard cartridge arrays. Here we present a comparative evaluation of performance between Affymetrix GeneChip Human 133A cartridge and plate arrays with an emphasis on the assessment of systematic variation and its impact on log ratio data. This study utilized two standardized control RNAs on four independent lots of plate and cartridge arrays. We found that HT plate arrays showed improved specificity and were more reproducible over a wide intensity range, but cartridge arrays exhibit better sensitivity. Not surprisingly, artifactual changes due to positional effects were detectable on plate arrays, but were generally small in number and magnitude and in practice may be removed using standard fold-change and p-value thresholds. Overall, log ratio data between cartridges and plate arrays were remarkably concordant. We conclude that HT arrays offer significant improvements over cartridge arrays for large-scale studies.

Pristane-Accelerated Autoimmune Disease in (SWR X NZB) F1 Mice Leads to Prominent Tubulointerstitial Inflammation and Human Lupus Nephritis-Like Fibrosis.

Mouse models lupus nephritis (LN) have provided important insights into disease pathogenesis, although none have been able to recapitulate all features of the human disease. Using comprehensive longitudinal analyses, we characterized a novel accelerated mouse model of lupus using pristane treatment in SNF1 (SWR X NZB F1) lupus prone mice (pristane-SNF1 mice). Pristane treatment in SNF1 mice accelerated the onset and progression of proteinuria, autoantibody production, immune complex deposition and development of renal lesions. At week 14, the pristane-SNF1 model recapitulated kidney disease parameters and molecular signatures seen in spontaneous disease in 36 week-old SNF1 mice and in a traditional IFNα-accelerated NZB X NZW F1 (BWF1) model. Blood transcriptome analysis revealed interferon, plasma cell, neutrophil, T-cell and protein synthesis signatures in the pristane-SNF1 model, all known to be present in the human disease. The pristane-SNF1 model appears to be particularly useful for preclinical research, robustly exhibiting many characteristics reminiscent of human disease. These include i) a stronger upregulation of the cytosolic nucleic acid sensing pathway, which is thought to be key component of the pathogenesis of the human disease, and ii) more prominent kidney interstitial inflammation and fibrosis, which have been both associated with poor prognosis in human LN. To our knowledge, this is the only accelerated model of LN that exhibits a robust tubulointerstitial inflammatory and fibrosis response. Taken together our data show that the pristane-SNF1 model is a novel accelerated model of LN with key features similar to human disease.