Table 7Reproducibility, MammaPrint®

Study, yearContext and MethodsResultsConclusions
Ach,200757MammaPrint assay intra- and inter-laboratory reproducibilityReplicate hybridizations Pearson's correlation at the same site:The authors showed very low influence on sample-to-reference ratios based on averaged triplicate measurements in the two-color experiments
Variation in RNA amplification and labeling, hybridization and wash, and slide scanning was measured on 4 tumors, dye-swap design, 24 slides (8 per site) For 1 tumor in 1 sub-array = 0.983RNA labeling was the largest contributor to inter-laboratory variation
Methods: To assess reproducibility in this study, ANOVA P values and Pearson's correlation were used For 2 tumor in 2 sub-arrays = 0.988Overall, despite this variation, measurement of 70-gene signature in three different laboratories was found to be highly robust
 For all the other technical replicates > 0.993
Scanning reproducibility across sites:
 Cy3: Pearson correlation >0.995, slope = 0.97
 Cy5: less reproducible (data not shown)
70-gene signature reproducibility:
 No differences between hybridization sites
 No differences between hybridization days (regardless of site)
Statistically significant difference (P value <0.05) between labeling sites for two tumors
Glas, 200658Pearson's correlation to assess correlation with original data and reproducibilityComparison to original 70-gene signature data, Pearson's correlations and in repeated measurements:Microarray technology can be used as a reliable diagnostic tool
ANOVA analysis to model variability in repeated experiments using the 70 genes of the signature 78 van't Veer21 patients, r = 0.92, p value < 0.0001The MammaPrint assay performed similarly to the original 70-genes signature
Reproducibility in time was assessed by repeated measurements of RNA aliquots: 145 (84+61) van de Vijver25 LN-negative patients: r = 0.88 p value < .0001
 1 patient with cosine correlation to Good profile = 0.61, for 12 months 49 patients analyzed twice, r = 0.995
 1 patient with cosine correlation to Good profile = -0.44, for 12 monthsReproducibility results from ANOVA analysis:
 1 border-line with cosine correlation to Good profile = 0.43, for 4 months No variation within individuals (p value = 0.96)
 Significant variation between individuals and genes
Reproducibility in time analysis results:
 For both patients assessed over a period of 12 months measurements SD was 0.028 of the cosine correlation
For the 1 border-line sample assessed over a period of 4 months measurements SD was 0.027 of the cosine correlation
This latter sample was misclassified 6 times (15%)

ANOVA = analysis of variance; Cy3: the green fluorescent dyes commonly used in two colors design microarray hybridization; Cy5 the red fluorescent dyes commonly used in two colors design microarray hybridization; LN = lymph node; SD = standard deviation

From: 3, Results

Cover of Impact of Gene Expression Profiling Tests on Breast Cancer Outcomes
Impact of Gene Expression Profiling Tests on Breast Cancer Outcomes.
Evidence Reports/Technology Assessments, No. 160.
Marchionni L, Wilson RF, Marinopoulos SS, et al.

NCBI Bookshelf. A service of the National Library of Medicine, National Institutes of Health.