Table 3

Influence of immunophenotyping cell composition on differentially expressed (DE) genes and pathways

DEDiscoveryValidation
NoYesNoYes
RNA transcript adjR2<0.1 n720818525768
adjR2≥0.1 n7917819258641
OR (CI95)4.09 (2.46 to 6.83)5.35 (4.16 to 6.89)
Median (IQR)0.202 (0.146–0.261)0.217 (0.153–0.298)0.222 (0.165–0.291)0.304 (0.219–0.396)
0.202 (0.146–0.262)0.224 (0.167–0.296)
FAIME
Reactome pathway
adjR2<0.1 n4381723147
adjR2≥0.1 n614208370629
OR (CI95)8.73 (5.24 to 14.53)8.35 (5.95 to 11.73)
Median (IQR)0.201 (0.149–0.268)0.217 (0.161–0.294)0.188 (0.146–0.25)0.271 (0.206–0.344)
0.207 (0.153–0.276)0.24 (0.171–0.316)
  • Correlation between absolute cell counts and RNA transcripts or FAIME scores in the discovery and validation sets after regression analysis and correction for concurrent therapy. The adjusted coefficient of determination (adjR2) of elastic net models is shown. ORs and 95% confidence intervals (CI95) are calculated for categorised values; the 10% of explained variance is chosen as threshold to categorise regression models (meaningful≥10%, not meaningful<10%). Significant DE genes or pathways are more likely to be discovered when the transcriptome or functional data correlate with immunophenotyping data.

  • DE, differentially expressed; FAIME, Functional Analysis of Individual Microarray Expression.