Table 4

Multivariate best subsets modeling of quantitative gene expression patterns in brushed buccal cells: among all subjects and current smokers only

A. All 42 subjects included
Quantitative expression correlates (notable patterns)
TranscriptFactorR2 of modelDirection of relationship*Coefficient (β)P
CYP1A1na (expressed in 2/42 subjects)nananana
GSTM1na (not expressed, any subject)nananana
GSTM3na (not expressed, any subject)nananana
GSTP1Smoke history0.621Negative4.2510.007
B. 19 current smokers only
TranscriptFactorR2 of modelDirection of relationshipCoefficient (β)P
CYP1A1na (expressed in 2/19 smokers)nananana
CYP1B1Plasma nicotine levels0.922Positive−0.889<0.001
Female gender0.922Positive6.7440.008
GSTM1na (not expressed, any smoker)nananana
GSTM3na (not expressed, any smoker)nananana
GSTP1Lung CA diagnosis0.011Positive−6.7570.011
  • NOTE: Quantitative mRNA expression assessed by RNA-specific real-time quantitative RT-PCR in buccal cells was regressed for several known clinical and demographic factors relevant to lung carcinogenesis, as described in the Materials and Methods section. Top portion of table is for all 42 subjects and bottom is for the 19 smokers. The complete accounting of interindividual variability (R2) of the model; direction of relationship; and β-coefficient.

  • Abbreviation: AhR, aromatic hydrocarbon receptor.

  • * “Positive” means the levels of gene transcript expression increased directly with the noted factor.

  • Negative coefficients yield positive relationships between covariates, because gene expression is quantitated as a CRO difference between target gene and referent gene transcripts.

  • As coding convention is set up for gender (male = 1 and female = 0), a positive coefficient implies a positive correlation between transcript quantity and female gender. Significance values for the individual factor in the model (P) are listed.